INTERNATIONAL MIGRATIOt\r ^ REVIEW ••• THE AMNESTY AFTERMATH: 1986 IMMIGRATION REFORM AND CONTROL ACT ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS INDIRECT IMMIGRATION TO CANADA MIGRANT REMITTANCES: THE CASE OF GREECE UNDOCUMENTED LATINA IMMIGRANTS IN ORANGE COUNTY, CALIFORNIA VIETNAMESE DOUBLE MARRIAGE SQUEEZE INTERETHNIC MARRIAGE IN AUSTRALIA NETWORK MIGRATION OF ETHNIC GERMANS JOB DISPLACEMENT EFFECTS OF CANADIAN IMMIGRANTS IMPACT OF REMIGRATION ON CASTE MOBILITY AUSTRALIA: FROM MIGRANT COUNTRY TO MULTICULTURAL NATION ol. 31, Spring 1997 J^ f 'CMTCD Cr\D AAI/^DATI^M CTI II-MCC I I • Editor Emeritus: SILVANO M. TOMASI Editor: LYDIO F. TOMASI Managing Editor: MARK J. MILLER Book Review Editor: ELEANOR M. ROGG Review of Reviews Editors: ElLEEN REITER, NORMAN KANTER Editorial Boar^T^OMAs]. ARCHDEACON • ROBERT L. BACH • GEORGE J. BORJAS • MONICA BOYD » ROGERS BRUBAKER • DALE S. DE HAAN • AUSTIN T. FRAGOMEN, JR. • LAWRENCE H. FUCHS • DONALD F. HEISEL • CHARLES B. KEELY • MARY M. KRITZ • C. MICHAEL LANPHIER • JOHN J. MACISCO, JR. • MARK J. MILLER • VICTOR NEE • ELEANOR M. ROGG • ALAN B. SIMMONS • JOHN SWENSON • MARTA TIENDA • LYDIO F. TOMASI • SILVANO M. TOMASI • ARISTIDE R. ZOLBERG Advisory Editors: ADERANTI ADEPOJU • R. T. APPLEYARD • JORGE BALAN • W. R. BOEHNING • Ivo BAUCIC • JOSEPH CHAMIE • BARRY CHISWICK • ANDREW M. GREELEY • GRAEME HUGO • GILBERT JAEGER • LASZEK A. KOSINSKI • APRODICIO A. LAQUIAN • IVAN LIGHT • LIN LEAN LIM • PHILIP L. MARTIN • URSULA MEHRLANDER • CHRISTOPHER D. MITCHELL • CHARLES A. PRICE • GlANFAUSTO ROSOLI • SETENEY SHAMI • LASZLO SZOKE • GEORGE TAPINOS • RUDOLPH J. VECOLI • JONAS WIDGREN The INTERNATIONAL MIGRATION REVIEW is published quarterly by the Center for Migration Studies of New York, Inc. (ISSN-0197-9183 ;LC-85-645414), a member of the Federation of Centers for Migration Studies J.B. Scalabrini, in cooperation with Migration and Refugee Services, U.S.C.C., Washington, D.C. The Center for Migration Studies is an educational, nonprofit institute founded in New York in 1964 to encourage and facilitate the study of sociological, demographic, historical, economic, legislative, and pastoral aspects of human migration and refugee movements. The opinions expressed in this Review are those of the contributors. U.S. Subscription Rates: Institutions: 1 year, $72.00 2 years, $140.00 3 years, $210.00 Individuals: 1 year, 35.50 2 years, 65.50 3 years, 95.00 Foreign Subscription Rates: Institutions: 1 year, 84.00 2 years, 164.00 3 years, 248.00 Individuals: 1 year, 48.00 2 years, 84.00 3 years, 129.00 Single Copy Rates: Institutions, $15; Individuals, $9; postage and handling, $5 each. Special Issues Rates: $19.50 each; postage and handling, $5 each. Subscriptions are payable in advance. Please make all remittances payable to the INTERNATIONAL MIGRATIONREVIEW) in U.S. currency or its equivalent in postal or express money orders or bank drafts or through Visa or Mastercard. Change of address: four weeks advance notice to the Center for Migration Studies — an old address as well as the new are necessary for change of subscriber's address. Printed in the United States. Periodicals postage paid at Staten Island, NY and Hanover, PA. POSTMASTER: Send address changes to IMR, 209 pT agg Place, Staten Island, NY 10304-1199 USA. Microfilms: University Microfilms International, 300 North Zeeb Road, Ann Arbor, MI 48106. Microfilm back issues are available only to regular subscribers to INTERNATIONAL MIGRATION REVIEW. MANUSCRIPTS: (in quadruplicate) should be addressed to the Editor of the INTERNATIONAL MIGRATION REVIEW, Center for Migration Studies, 209 Flagg Place, Staten Island, NY 10304-1199. Business correspondence should be addressed to the Center for Migration Studies. ADVERTISING space in the INTERNATIONAL MIGRATION REVIEW is available. For information and rates, please contact the CMS Marketing Department, 209 Flagg Place, Staten Island, NY 10304-1199. Tel: (718) 351-8800; fax: (718) 667-4598; e-mail: cmslft@aol.com; website: http://www.cmsny.org IMR is indexed/abstracted in the following sources: Abstracts in Anthropology! Academic Abstracts; America: History and Life; Applied Social Science Index e^ Abstracts, ASSIA; Bulletin Analytique de Documentation Politique, Economi se et Sociale Contemporaine; Bulletin Signaletique Part 521: Sociologie-Ethnologic; Chicano Database; Currents Contents; Current Index to Journals in Education, ERICCUE; DOCPAL; Resumenes Sobre Problacion en America Latino; Ethnizat ^Migration; Geo Abstracts, GEOBASE," Hispanic American Periodical Index; Historical Abstracts; Human Rights Internet Reporter; Monthly Bibliography: Part II; PAIS Bulletin, Public Affairs Information Service, Inc. ; Population Index; Refugee Survey Quarterly; R.E.M.I.S.I.S.; Research Alert; Review of Population Reviews, C.I.C.R.E.D.; SAGE Human Resources; Social and Behavioral Sciences; Social Science Index; Social Sciences Abstracts; Sociological Abstracts; Social Sciences Citation Index; United States Political Science Documents and Re fe rence Systems. Copyright © 1997 by the CENTER FOR MIGRATION STUDIES OF NEW YORK, INC. 209 Flagg Place, Staten Island, NY 10304-1199, USA Phone: (718) 351-8800 Fax: (718) 667-4598 INTERNATIONAL MIGRATION REVIEW VOLUME XXXI NUMBER 1 SPRING 1997 5 The "Amnesty " Aftermath: Current Policy Issues Stemming from the Legalization Programs of the 1986 Immigration Reform and Control Act SUSAN GONZALEZ BAKER 28 Determinants of English Proficiency among Mexican Migrants to the United States KRISTIN E. ESPINOSAAND DOUGLAS S. MASSEY 51 Geographically Indirect Immigration to Canada: Descri p tion and Analysis MICHAEL J. GREENWOOD AND PAULA. YOUNG 72 Factors Determining Migrant Remittances: The Case of Greece THEODORE P LlANOS 88 Undocumented Latina Immigrants in Orange County, California: A Comparative Analysis LEO R. CHAVEZ, E ALLAN HUBBELL, SHIRAZ I. MISHRAAND R. BURCIAGA VALDEZ 108 The Vietnamese Double Marriage Squeeze DANIEL M. GOODKIND 128 Interethnic Marriage: Identifying the Second Generation in Australia PARIMAL ROY AND IAN HAMILTON RESEARCH NOTES 143 Network Migration of Ethnic Germans THOMAS BAUERAND KLAUS E ZIMMERMANN 150 Job Displacement Effects of Canadian Immigrants by Country of Ori gin and Occupation ARUN S. ROY 162 Remigration: The Return of the Prodigals, An Analysis of the Impact of Cycles of Migration and Remigration on Cast Mobility VIBHA PURI CHANDRA 2 INTERNATIONAL MIGRATION REVIEW DOCUMENTATION NOTE 171 Australia: From "Migrant Country " to Multicultural Nation J. J. SMOLICZ BOOK REVIEWS 187 REVIEW ESSAY DANIEL C. TURACK Laws Harsh As Tigers LucyE. Salyer, Mass Expulsion in Modem International Law and Practice Jean-Made Henckaerts Seeking Asylum: Comparative Law and Practice in Selected European Countries Hetene Lambert 189 The Face of the Nation: Immigration, the State and the National Identity Keith Fitzgerald KEVIN R. JOHNSON 190 Fresh Blood: The New American Immigrants SanfovdJ. Ungar PETER I. ROSE 191 American Dreaming: Immigrant Life on the Margins Sarah J. Mahler ELIZABETH}. MUELLER 191 Waiting on Washington Terry A. Repak AUDREY SINGER 192 Special Sorrows: The Diasporic Imagination of Irish, Polish, and Jewish Immigrants in the United States Matthew FryeJacobson JOHN J. BUKOWCZYK 193 FDR and the Holocaust Verne W Newton, Ed, MARKJ. MILLER Peri pheral Migrants, Haitians and Dominican Republic Sugar Plantations SamuelMartinez ROBERT D. MANNING CONTENTS A Contest of Faiths: Missionary Women and Pluralism in the American Southwest Susan M. Yohn MARY ELIZABETH BROWN Farmers ^ and Farm Workers ^ Movements Social Protest in American Agriculture Patrick H. Mooney and TheoJ. Majka JACK EGAN China ^ s Minority Cultures: Identities and Integration since 1912 Colin Mackerras ANTHONY B. CHAN Monsoon Season Dr. William Q,Wu BETTY LEE SUNG "For the Sake of Our Japanese Brethren " : Assimilation, Nationalism, and Protestantism among the Japanese of Los Angeles, 1895—1942 Brian Masaru Hayashi FRANKLIN ODO Sojourners and Settlers: The Macedonian Community in Toronto to 1940 Lillian Petroff GERALD TULCHINSKY REVIEW OF REVIEWS INTERNATIONAL NEWSLETTER ON MIGRATION BOOKS RECEIVED The "Amnesty" Aftermath: Current Policy Issues Stemming from the Legalization Programs of the 1986 Immigration Reform and Control Act The 1986 Immigration Reform and Control Act (IRCA) created two one-time only legalization programs affecting nearly 3 million undocumented immigrants. Legalization has produced important changes among immigrants and in immigration policy. These changes include new patterns of immigrant social and economic adaptation to the United States and new immigrant flows through family ties to IRCA-legalized aliens. The heightened salience of immigration, produced in part by legalization, has also generated a wave of "backlash" policymaking at the state and local levels in high-immigration sites. This article combines data from a longitudinal survey of the IRCA-legalized population with qualitative field data on current immigration issues from key informants in eight hig h-immigration metropolitan areas. It reviews the political evolution and early implementation of legalization, the current socioeconomic position of legalized aliens, and changes in the immigration " policy space " resulting from legalization. Although restrictive policies have again captured public attention, legalization has also sparked renewed efforts at immigration advocacy, particularly where immigrants who adjust to U.S. citizenship hold the potential for influencing local politics. In 1986, after more than a decade of contentious debate, the U.S. Congress passed an immigration reform bill with an ambitious goal — to restore control over burgeoning undocumented immigration. The 1986 Immigration Reform and Control Act (IRCA) offered two new policy tools. First, IRCA created civil and criminal penalties for U.S. employers who knowingly hired undocumented immigrants. Second, IRCA authorized a set of temporary, one-timeonly immigration benefits programs to "legalize " certain undocumented immigrants already living in the United States - programs that came to be known in the immigrant community as " amnesty. " This reform produced some dramatic short-term changes in migration dynamics. In the year following IRCA' s passage, some undocumented immigrants went home (Hagan, 1994); some delayed coming (Donato, Durand and Massey, 1992); and many came forward to adjust to legal status (Baker, 1990). ©1997 by the Center for Migration Studies of New York. All rights reserved. 0197-9183/97/3101.0117 Susan Gonzalez Baker University of Texas at Austin 6 INTERNATIONAL MIGRATION REVIEW However, contemporary evidence suggests that, over the long haul, undocumented immigration persists virtually undaunted. For instance, U.S. Border Patrol apprehensions dropped nearly 50 percent in the three years after IRCAs passage, but are now back at pre-IRCA levels (Bean, Edmonston and Passel, 1990; Donato, Durand and Massey, 1992; Donato, 1993). 1 Similarly, although IRCAs legalization options reached nearly 3 million undocumented immigrants, recent INS estimates of the current undocumented stock — some 3.2 million — imply that IRCA did litde to reduce undocumented setdement (INS, 1994). Post-IRCA scholarly attention has focused on INS law enforcement efforts to control immigration (Lowell and Jing, 1994; Bean et ai ^ 1994; Donato, Durand and Massey, 1992). However, only a few projects have assessed the U.S. legalization programs. Shortly after IRCAs passage, policy analysts evaluated the amnesty efforts as they unfolded in the field, focusing on emerging challenges to street-level bureaucracies implementing the program (North, 1988; North and Portz, 1989; Meissner and Papademetriou, 1988). Subsequent efforts extracted some general themes about immigration reform from the legalization experience (Hagan and Baker, 1993; Baker, 1990) and examined how legalization affected setdement experiences in particular immigrant communities (Hagan, 1994; Hondagneu-Sotelo, 1995). Finally, preliminary results are emerging that assess the labor market experiences of the newly legalized (Borjas and Tienda, 1993; U.S. Department of Labor, 1995) and the implications of IRCA legalization for specific sectors of the U.S. industrial structure (Martin, 1994; North, 1991). Still, a comprehensive assessment has simply had to wait until most applications have been adjudicated and data could be gathered — both quantitative data on the nations " amnestied" cohort and qualitative data on the dynamics unfolding in communities where that cohort is settling. This analysis uses both data types to examine the design, implementation, and outcomes of " amnesty, " some ten years after IRCA' s passage into law, to determine how well the expectations of immigrants, scholars, and the policy community have been met. It begins with an overview of legalization as a policy tool: its features, strengths, and weaknesses. It then examines the social and economic characteristics of the U.S. legalized population. Next, I draw upon new field data in several key sites to assess the current policy concerns emerging around immigration in general and the IRCA-legalized population in particular. In conclu-^ n 1986, INS apprehensions of undocumented immigrants reached an all-time high of 1.7 million. By 1989, that figure had dropped to 900,000. Demographic analyses suggest that slightly over half the drop is owed not to increased deterrent effects but to the fact that IRCA legalization programs pulled roughly 3 million undocumented immigrants from the pool of eligibles at risk for apprehension by legalizing their status. In 1993, apprehensions had risen to 1.3 million, a figure equal to that of 1985, the year preceding IRCA's passage. (Bean, Edmonston and Passel, 1990; INS, 1994). IRCA LEGALIZATION PROGRAMS 7 sion, the policy issues emerging as most salient in the " amnesty " aftermath are summarized and I suggest how these issues mi ght affect the current immigration policy climate. TOUGH CHOICES: THE "AMNESTY 9 POLICY OPTION Ambrose Bierce (1925) wryly defined " amnesty" as " the state ' s magnanimity toward those offenders whom it would be too expensive to punish. " Truly, amnesty presents the state with tough choices. It offers an expedient way out of the heavy costs involved in fully prosecuting systematic rule violation - be it tax evasion, the committing of war crimes, widespread avoidance of military service obligations, or the circumventing of immigration law. But amnesty also represents the state admitting that its policies have been honored in the breach. Not surprisingly, it is rarely a popular option among policymakers. To counter this image of vulnerability, policymakers usually introduce amnesty proposals alongside either new policies or redouble efforts at enforcing old policies aimed at the undesirable actions the " amnesty " targets have already taken (Freeman, 1994). Policymakers may fear, understandably, that these new enforcement threats ring hollow alongside an expansive, well-publicized amnesty for the very behavior being sanctioned. Such has been the case throughout the world when amnesty was directed toward undocumented or irregular migration. Amnesty; it is argued, implies capitulation to those migrants who, collectively, have taken the migration and settlement process into their own hands, often as a result of changes in or cessation of " temporary " guestworker programs that encouraged their entry into the host country but sought to constrain the possibility of settlement (Snowden, 1990). Case studies in Spain (Cornelius, 1994; Baker, 1995), France, (Hollifield, 1994), Japan (Cornelius, 1994), Italy (Calavita, 1994), Bel gium, Canada (Garcia y Griego, 1994), and Venezuela (Meissner, Papademetriou and North, 1986) cohere into a portrait of governments adopting the amnesty option reluctantly, if at all, and only on the heels of renewed efforts to limit irregular immigration. In every case, the primary concern is not fear of the numbers who might come forward, but fear of the precedent amnesty sets regarding the legitimacy of the institutions controlling entry and settlement into the host country. Although turnout never exceeded a few hundred thousand in these cases (and more typically leveled off at less than 100,000 applicants), policymakers consistently voiced fears that amnesty created a perverse incentive for increased undocumented immigration through its seeming reward for bypassing legal routes to foreign residence (Cornelius, Martin and Hollifield, 1994). As bad as this dilemma has been for countries with small to moderate size undocumented populations, it has proved all the more nettlesome for the United States, where these themes intersect with a numerically large population of potential amnesty-eligibles. The interaction of the ever-present " underminine of 8 INTERNATIONAL MIGRATION REVIEW the stated authority " theme and the real possibility of big numbers served to stall U.S. amnesty proposals for more than a decade. In the mid-1970s, the INS began issuing estimates of the resident illegal immigrant population on the order of 12 million. By the time 1980 census figures substituted these estimates with more reasonable alternatives of 2 million to 4 million, the social construction of a nation losing control of its borders was well entrenched in public sentiment, making border control a popular policy theme and amnesty a very hard political sell. But the invasion metaphor, with its implication that new immigration control efforts were necessary, shed little light on the problem of what should be done about those undocumented immigrants already living in the United States. Several alternatives existed: wholesale efforts at deportation; a broad, " slate-cleaning " amnesty that would wave a "legalization " wand over the U.S. resident undocumented stock and let the United States concentrate resources on discouraging and interdicting new undocumented flows; or a more targeted pathway to legal residence and full citizenship open only to those undocumented immigrants with "firm equities " built up during their years of U.S. residence, coupled with the hope that inelig ibles would find the new immigration law enforcement climate so inhospitable they would leave. In the end, IRCAs legalization programs represented an amalgam of the "firm-equities " and " slate-cleaning " options in its two major legalization programs. The main amnesty program offered a chance at legal status to those undocumented immigrants who could prove they had been in the United States continuously since January 1,1982, and who could surmount several intermediate obstacles to permanent resident status, including a " temporary" residence period and fulfillment of English language and civics education requirements. A separate, much more liberally-structured program for undocumented farmworkers with recent experience harvesting U.S. crops (the Special Agricultural Worker or SAW program) made its way into IRCA in the closing weeks of Congressional debate. 2 HARD BARGAINS: A BRIEF REVIEW OF AMNESTY IN THE UNITED STATES The IRCA amnesty programs emerged from a byzantine series of cleavages and coalitions wheeling and dealing their way through Congress. The eleventh-^n addition, two much smaller programs granted legal status to very specific target groups. First, an extant " registry" program offering legal status to undocumented immigrants living in the United States since 1948 was updated to include those who had been resident since 1972. Roughly 58,000 people would apply under this category, nearly three-quarters of whom were of Mexican origin. Second, a Cuban-Haitian Adjustment provision targeted Mariel boatlift era "special entrants," largely of Haitian origin, who had been omitted from other refogee/asylee adjustment programs. IRCA LEGALIZATION PROGRAMS 9 hour introduction of SAW legalization, for example, both saved the IRCA package and condemned it to its worst criticism. Well documented elsewhere (Mar tin, 1994), the SAW amnesty bears special mention here as an example of the hard bargains characterizing the IRCA debate. All observers recognized that IRCA would not see daylight without support from California, home to and employer of most U.S. undocumented immigrants, including a large share of undocumented farmworkers. Throughout IRCAs debate, California growers and their advocates sought agricultural " guestworker " provisions and exemptions from employer sanctions to ease IRCA' s impact. Farmworkers and their advocates wanted, instead, a generous amnesty that would not condemn them to second-class " guestworker " status. Immigration restrictionists throughout Congress insisted on strong sanctions enforcement across the board in exchange for their votes supporting amnesty provisions. Immigration advocates mobilized along ethnic lines to insist on generous amnesty in exchange for their support of sanctions against the major employers of their ethnic compatriots. Congressional powerbrokers attempted to balance both the restrictive and inclusionary impulses of all the players with increasingly complex tinkering with sanctions and legalization parameters. These machinations produced a fairly complex set of statutory requirements for longer-term undocumented immigrants seeking legalization, including continuous residence since 1982, significant burden of proof imposed on their petitions, an application process that took p lace in multiple stages, educational requirements, and bans on access to social services following successful completion of the process. In contrast, as a final compromise intended primarily to shut out a deal-breaking agricultural " guestworker " provision being promoted by then-Senator Pete Wilson of California, SAW legalization offered full-fledged legal resident status to farmworkers whose migration history was as limited as 90 days employment in the previous growing season, promising growers uninterrupted access to their traditional workforce. Gritting their teeth, employer sanctions supporters accepted these legalization programs as par t of the IRCA deal (though amnesty survived the final House battle by only four votes) and the package reached President Ronald Reagans desk on November 6, 1986. EVALUATING AMNESTY: EARLY DATA ON APPLICANTS AND IMPLEMENTATION To monitor the IRCA implementation effort, the Ford Foundation sponsored a nationwide research project codirected by the Rand Corporation and the Urban Institute. During 1988 and 1989, researchers conducted field interviews in eight sites where legalization was expected to draw a significant response: Los Angeles, San Jose, El Paso, San Antonio, Houston, Chicago, 10 INTERNATIONAL MIGRATION REVIEW New York, and Miami. Respondents included local INS officials, advocates, trade and union representatives, public officials, and employers. In addition, interviews took place with regional and national-level respondents. Approximately 600 interviews were conducted. My analysis of early legalization implementation draws from this database (Baker, 1990). INS statistical analysts estimated that up to 2.1 million persons might be eligible for pre-1982 legalization. Demographers estimated another 250,000 farmworkers might be eligible for SAW status. Ultimately, the 107 INS legalization offices around the country processed approximately 1.7 million pre-1982 applications. SAW applications, however, dwarfed all reasonable estimates, with 1.3 million applicants coming forward claiming to have worked in U.S. agriculture in as little as a single growing season. Table 1 presents a comparison of legalization turnout to pre-implementation estimates from various sources. Census Bureau estimates of the undocumented population resident in the selected SMSAs (standard metropolitan statistical areas) in 1980 should be treated with some caution. First, the estimates do not include the 1981 additions to the undocumented population who were also eligible for legalization. Furthermore, the estimates may vary in quality across sites. Nor can we account for post-1980 migration within the United States, which may have affected the population at risk for applying in each site. Nonetheless, the estimates serve as a baseline for program turnout comparisons. Los Angeles hosted the largest legalization program, and its turnout was in close keeping with pre-implementation estimates. Each Texas SMSA took in far more applications than predicted. Chicago and San Jose corresponded closely to pre-implementation estimates. New York City and Miami are notable in their relatively low legalization showings. Statewide totals in Table 1, Panel B mirror the profiles for the individual SMSAs, with high turnout in Texas; California and Illinois falling between the "hig h"and "low " range; and New York and Florida demonstrating relatively modest turnout. Across the board, in every site, SAW turnout "bursts through the roof of INS planning estimates. Miami s low pre-1982 numbers are obviated somewhat by the 50,500 additional immigrants who applied under the SAW program, some of whom may also have been eligible for the pre-1982 program. In New York City, however, all indications are that pre-1982 legalization turnout was modest. Field interviews with New York respondents identify several reasons. First, the diversity of the immigrant community in New York made legalization outreach particularly difficult. No single national-origin group constitutes more than ten percent of the foreign-born population in New York City. Second, INS district leadership expressed less enthusiasm for legalization than for other district priorities, particularly in li g ht of the modest resources afforded to the district. Only two legalization offices operated inside the SMSA boundaries, versus 16 Los Angeles offices. IRCA LEGALIZATION PROGRAMS 11 TABLE 1 COMPARISONS OF LEGALIZATION TURNOUT WITH PRE-IMPLEMENTATION ESTIMATES OF ELIGIBLE UNDOCUMENTED IMMIGRANTS Panel A 1980 Census Prc-1982as Undocumented Prc-1982 SAW Percent of SMSAEstimateTurnoutTurnoutCensus Estimate Los Angeles 657,900 596,650 172,808 90.7 Houston 51,956 113,870 23,880 219.2 NewYork 211,658 109,951 29,201 51.9 Chicago 127,113 118,815 29,613 93.5 ElPaso 15,696 30,374 17,796 193.5 San Antonio 13,041 22,423 6,468 171.9 San Jose 26,477 24,865 16,365 93.9 Miami 49,672 28,369 50,509 57.1 Total1,153,513 1,044,957346,64090.6 Panel B Pre-1982 SAW INS Planning INS Planning INS Planning StateEstimate: High Estimate: LowTurnoutEstimateTurnout California 1,374,000 685,000 948,200 123,000 699,100 Texas 310,000 130,000 313,200 17,000 132,500 Illinois 174,000 88,000 120,400 6,000 32,600 NewYork 230,000 128,000 119,900 8,000 47,400 Florida 105,000 53,000 48,400 56,000 120,000 Total2,193,000 1,084,000 1,550,100210,000 1,031,600 Sources, Panel A: Census Estimates; U.S. Census Bureau unpublished estimates, see Warren and Passel (1987) for explanation of technique. Legalization Turnout: Applications filed at Legalization offices within 1980 SMSA boundaries. INS, Statistical Analysis Branch, unpublished data, J anuary 27, 1989. Sources, Panel B: Planning Estimates: Unpublished data, INS Statistical Analysis Branch, June 23, 1988. Turnout: INS, Statistical Analysis Branch, "Provisional Legalization Application Statistics," May 12, 1989, under "State of Residence," pp. 3-4. In contrast, Houston boasted one of the most organized, proactive advocacy communities in the country, working in concert with a highly-motivated INS district. Drawing together the private immigration bar, refugee resettlement programs, reli gious organizations, and human rights groups, this community mounted its own publicity campaign and took advantage of the INS districts interest in rehabilitating its public image by meeting regularly with district leadership under the auspices of an immigration task force. The result was a turnout twice the expected total in the SMSAs single legalization office. In addition, the Texas undocumented population was much more homogeneous than that of New York; and this largely-Mexican population was exactly who the framers of IRCA had in mind as they designed and implemented the program. Evidence for IRCA' s tilt toward Mexicans (and, to some extent, other Latino immigrants) can be seen in several key INS implementation decisions. For 12 INTERNATIONAL MIGRATION REVIEW example, throughout legalization, the INS battled in the courts to exclude applicants who did not fit the typical profile of the Mexican undocumented immigrant ' entering without inspection" (INS parlance for crossing the border surreptitiously). Visa overstayers, for example, were much more likely to hail from a wide range of countries. Court challenges to INS regulations proved necessary before many of these undocumented immigrants were allowed to apply for legalization. In addition, INS respondents in several key sites - Houston and Los Angeles, for example — reported making a special effort to hire temporary employees for legalization office duty who either were Hispanic or who spoke Spanish. No mention was made of such effort on behalf of other ethnic groups. Immigration attorneys complained that this misled non-Hispanic eligibles, with one Houston attorney reporting, for example, that a Zambian client failed to apply because he " thought that program was only for Mexicans. " Given these observations, how closely did the demographic, social and economic characteristics of legalized aliens match the expectations of policymakers and researchers? Answering this question requires a closer look at the composition of the legalized population and a comparison of those features to policy assumptions. Some seven years after the legalization window closed, questions have emerged regarding the fate of the legalized and the lessons of the program for policymakers. These questions include the following themes. 1) Who was brought in and who was left out? As we shall see, some sociodemographic groups were more successful than others in availing themselves of the amnesty opportunity. 2) What did the " amnestied" gain? Policymakers assumed that legal status would promote economic and civic integration of its beneficiaries and that the residual undocumented population would leave the United States. Field evidence indicates that some assumptions were more plausible than others. 3) What did U.S. communities gain? Legalization produced new economic and political constituencies in some of the largest labor markets in the United States. I examine how these new constituencies have produced new social and economic dynamics, and new conflict points, in the communities where they now reside. 4) What does legalization offer for the future of U.S. communities in general and their labor market dynamics in par ticular? Legalization imposed a new set of economic rights and responsibilities upon an extant group of workers. Preliminary evidence on the legalized population ' s position in U.S. labor markets can offer some guidance on how this cohort will IRCA LEGALIZATION PROGRAMS 13 operate in the future and what that might mean for its U.S.-born counterpart. 5) What lesson does legalization offer for the making of immigration policy? As a policy tool for coping with illegal immigration, legalization was virtually unprecedented in the United States. Evaluating whether legalization met its goals can allow us to assess whether it merits consideration in other immigration contexts. This report employs two data sources to address these research questions. First, I use the results of federally sponsored data collection projects: administrative data on the entire legalized alien population known as LAPS data (from the Legalized Alien Processing System) and a more detailed survey of 6,193 persons drawn from that sampling frame. The survey, conducted by Westat, Inc., in two waves — one in 1989 under the direction of the INS and a second in 1992 under the Department of Labor - identifies key sociodemographic and economic characteristics of the legalized population. Because these data are restricted to the pre-1982 cohort, evaluations of SAW characteristics are not possible. Therefore, the demographic overview is restricted to the 1.7 million recipients of legal status through the main amnesty program. Second, I employ the results of updated field interviews conducted in late 1993 and early 1994 in the same implementation sites that served as the sample for an earlier review of legalization (Baker, 1990). Table 2 summarizes some key sociodemographic features of the legalized population. Disaggregating total amnesty turnout by region of origin and implementation site reveals useful information. First, it is clear that Mexicans were the overwhelming winners. Pre-implementation estimates placed the Mexican share at roughly half the undocumented total (Warren and Passel, 1987), yet Mexicans accounted for 70 percent of all legalization applicants and greater shares in California, Texas, and Illinois. Interestingly, the Mexican share was more modest in Houston, where a particularly active advocacy community concentrating on Central Americans conducted extensive outreach among the Salvadoran and Guatemalan constituencies and brought thousands into the program (Hagan and Baker, 1993). Active legalization outreach among Polish immigrants accounts for much of the Eastern Hemisphere showing in Chicago; while Caribbean migrants — Dominicans, Haitians, Jamaicans — were significantly represented in New York and Florida. If this national-origin composition reflects implementation bias, it may be the case that legalization cleaned some parts of " the slate " better than others. Legalization may well have altered the immigration-status composition of the Mexican-origin community in the United States while leaving other nationalorigin subgroups relatively untouched. Indeed, recent estimates of the resident 14 INTERNATIONAL MIGRATION REVIEW TABLE 2 LEGALIZATION TURNOUT BY REGION OF ORIGIN IN KEY IMPLEMENTATION SITES Region of Origin (%)  Total Central Other Western Eastern State/MSATurnoutMexicoAmerica Hemisphere Hemisphere California 955,875 77 16 2 5 Los Angeles 610,726 74 21 2 4 Texas 308,478 81 13 2 4 Houston 106,608 66 25 4 5 Iltmois 120,983 81 5 3 12 Chicago 106,783 79 5 3 14 New York 118,210 10 16 50 25 New York 105,905 10 12 53 25 Florida 50,114 6 16 71 7 Miami 27,774 2 26 69 3 Total United States 1,759,705701389 Sources: U.S. Immigration and Naturalization Service, Statistical Yearbook of the Immigration awl Naturalization Service ^ 1991. Washington, DC; U.S. Government Printing Office, 1992. TaMe 23, p. 73; U.S. Immigration and Naturalization Service, Immigration Reform and Control Act: Report on the Legalized Alien Population, Washington, DC: U.S. Government Printing Office, 1992. Table 11, p. 20. undocumented population show a dramatic shift in national-ori gin composition. Whereas pre-IRCA estimates placed the Mexican share at roughly half, new INS estimates place it at one-third (INS, 1994). To the extent that non-Mexican subgroups are concentrated in particular geographic areas - New York, for example — the labor market dynamics of undocumented immigration for those local areas are unlikely to evince much change as a result of amnesty An official from the New York Association for New Americans, for example, noted that little labor market change seemed to be taking place among legalized aliens or their undocumented counterparts, with both " greatly underutilizing their skills . . . although (the legalized) do have the psychological benefit now of not having to hide. ... " An INS respondent noted that " they re still the low-paid, hard-working individuals . . . with jobs in landscaping, restaurants, garages . . . you know, the unskilled work. " Table 3 summarizes selected sociodemographic characteristics of the legalized population. As expected, the legalized population is largely a young adult cohort in the early stages of family formation. Median age was 30 years in 1989, younger than the total U.S. population (32 years) and older than the U.S. Hispanic population (26 years). Although the trend in legal migration has been toward an increasing share of women, men predominate among the legalized. Rather than assuming that the gender patterns in legalization accurately reflect population composition, it is important to consider how implementation affected the gender distribution. In each site visited during early implementation, advocates and INS officials noted great difficulty facing IRCA LEGALIZATION PROGRAMS 15 TABLE 3 SELECTED SOCIODEMOGRAPHIC CHARACTERISTICS OF LEGALIZED POPULATION IN KEY IMPLEMENTATION SITES Median Percent Percent Median Years Percent State/MSAAge-*MaleMarried 1 * ofEducadon Speaking En^ish California 31 56 63 7 13 Los Angeles 32 56 62 7 13 Texas 30 59 62 6 8 Houston 30 62 64 6 11 Illinois 31 61 61 6 17 Chicago 32 61 62 6 17 New York 34 61 49 12 32 New York 35 61 47 12 33 Florida 35 51 61 12 42 Miami 35 47 61 12 35 Total United States325861715 Sources: U.S. Immigration and Naturalization Service, Immigration. Refenn and Control Act: Report on the Legalized Mien Population. Washington, DC: U.S. Government Printing Office, 1992. Tables 12, 13, pp. 22-23. 'Median age at time of reported arrival in United States. Percent reporting currently married or living with partner as if married. undocumented women trying to adjust to legal status. Legalization required a paper trail demonstrating U.S. residence since 1982 — e.g., rent receipts, paycheck stubs, utility bills. Women were less likely to have such evidence, par ticularly when 1) they were partnered with men whose names appeared alone on leases, bank accounts, etc. and 2) when they were employed in the underground economy of domestic service (Hagan, 1994; Baker, 1990; Hagan and Baker, 1993). Although IRCA recognized affidavits from neighbors, coworkers and employers as evidence of continuous residence, these ' affidavits-only " cases faced higher levels of adjudicator scrutiny and became the basis for litigation on behalf of petitioners relying on such sworn testimony (Loe u Thomburgk 1989). In sum, the concentration of males among the legalized likely reflects both migration dynamics and program implementation decisions. Further evidence that women were present in the networks of legalized aliens comes from the high proportion of amnesty recipients who reported being married or living with a partner as married in the 1989 Westat survey. While many of these spouses/par tners may not have been eli g ible for legalization, it is clear that a portrayal of the undocumented immigrant who legalized through IRCA as a solitary target-earner is incomplete. Along with basic demographic characteristics, the Legalized Person Survey provides human capital profiles that are useful in anticipating labor market incorporation. Clearly, the legalized population lags far behind other groups of U.S. workers in education. Median education level is seven years for the entire leg^ized population, with lower educational attainment in Texas. Higher educa- 16 INTERNATIONAL MIGRATION REVIEW tion levels emerge in New York and Florida, in keeping with the greater diversity of national-origin backgrounds evident in these sites (Borjas and Tienda, 1993). English-language skills vary across sites. Overall, fifteen percent of legalized adults reported speaking English "well" in the Westat survey. That proportion was much lower in Texas (8%), and considerably hig her in New York and Florida. Thus, it appears that the most numerous amnesty beneficiaries did not rank highly on orthodox indicators of human capital. That is, the largely low-educated, non-English-speaking undocumented immigrants in the southwestern United States were the most numerous applicants, and it is in that region, and among the job and industry categories where those workers are found, that the effects of legalization are likely to be greatest. Again, implementation decisions facilitated the ability of these modestly educated immigrants to complete the legalization process successfully. One important decision was the creation of alternatives to the IRCA-mandated "English/Civics " test for permanent residence. IRCA also provided for adjustment to permanent residence if an applicant could demonstrate attendance of an English/Civics course of study approved by the INS, for a minimum of 40 hours, even if proficiency levels were insufficient to pass the test. Roughly three-quarters of the legalizing immigrants surmounted the English/Civics barrier through these alternate routes (INS, 1992). These amnesty "Phase II" classes, as they were called in most sites, brought thousands of legalizing immigrants into the classrooms of adult education providers, even prompting one Houston respondent to call the program " the largest Hispanic adult education program ever mounted in this city." Still, 40 hours of instruction could not promote true English literacy among a population starting with so little educational background. Although IRCA originally authorized $ 1 billion a year for four years to be returned to states and localities for the expenses incurred in delivering such services as English/Civics classes, follow-up field interviews in 1993 indicated that the monies actually disbursed dropped so dramatically over time that many course providers in the nonprofit community simply closed up shop. In Los Angeles, where respondents from the L.A. Unified School District reported IRCA adding approximately 270,000 students to the adult education caseloads, federal reimbursement was limited to those first 40 hours of instruction necessary to meet the permanent residence requirement. The " amnesty coordinator " for the school district noted that IRCA students were then forced into the general student population and had to compete with our non-IRCA students for slots. "Most of them require at least 90 to 150 hours of English training for even the most limited skills. All 40 hours does is keep them in their present low-wage job and social position. " Thus, IRCAs goal of strengthening the connection of legalized aliens to their host society may have been undercut by the inability of program directors to implement "English/Civics " as part of a meaningful educational opportunity. IRCA LEGALIZATION PROGRAMS 17 Table 4 presents selected family/household composition variables characterizing the legalized population. Legalized aliens were much more likely to be living in households with family members than living alone. This was particularly true of legalized aliens living in California, where over 80 percent of the applicants reported family members sharing their U.S. households. Respondents to the Westat survey identified persons living in their households at the time of their application and whether each person was either a U.S. citizen or a legal permanent resident. The residual household members in the legalized aliens household who were neither U.S. citizens nor permanent residents may serve as a proxy for part of the post-IRCA residual undocumented population. Table 4 indicates that as many as 710,000 presumably undocumented persons were living in the United States with legalized aliens in 1989. This phenomenon of mixed-immigration status within families proved to be one of the thorniest implementation issues facing legalization. Early in the program, it became apparent that one of the factors " chilling " application rates was the immigrants ' concern that their ineligible family members would come to the attention of the INS as a function of their own applications. As a result, the INS issued implementing regulations instructing district offices to grant administrative relief through such mechanisms as ' voluntary departure " status to the undocumented minor children and, ultimately, the spouses, of legalizing aliens. In addition, the Immigration Act of 1990 created a temporary allotment of 55,000 extra visa slots in fiscal years 1993 and 1994 to be applied to spouses and minor children of legalized aliens. In sum, both regulatory and statutory relief sought to cope with the problem of mixed-immigration status without resorting to the deportation of undocumented family members attached to legalized aliens. Still, by 1989, for every 100 legalized aliens, 41 family members remained in the United States without either citizenship or permanent resident status having been secured through relationship to their " amnestied" relative. Table 5 summarizes some key economic characteristics of the legalized population, both confirming and calling into question many assumptions about this population. Considerable variation exists across implementation sites in the labor market position of the legalized population. In Florida, one in three legalized aliens worked in a white-collar occupation. Texas legalized aliens, in contrast, were highly concentrated in blue-collar work and were more dispersed across industries, including a much higher share in construction work. Legalized aliens were more concentrated in the manufacturing and service sectors in California and Newark. Wage rates and individual earnings confirm the assumption that legalized aliens were a low-wage group. Median wages for legalized aliens at the time of application stood at roughly 60 percent of that for the total U.S. workforce, and median earnings were less than half (INS, 1992). These dollars earned 18 INTERNATIONAL MIGRATION REVIEW TABLE 4 SELECTED FAMILY/HOUSEHOLD CHARACTERISTICS OF LEGALIZED POPULATION BY KEY IMPLEMENTATION SITES Family/Household Size (%) Number of Probable Undocumented in State/MSA12-34+Legalized Family/Household California 17 30 53 436,075 Los Angeles 16 30 54 263,578 Texas 30 31 39 91,569 Houston 27 32 41 38,910 Illinois 28 28 44 45,307 Chicago 29 30 41 37,061 New York 39 35 26 47,481 New York City 41 35 24 41,413 Florida 24 40 36 17,515 Miami 21 38 41 9,236 Total United States243046710,399 Source: U.S. Immigration and Naturalization Service, Immigration Reform and Control Act: Report on the Legalized Alien Population, Washington, DC: U.S. Government Printing Office, 1992. Tables 14, 16, pp. 24-27. were also being stretched beyond the boundaries of the U.S. households. Fully two-thirds of the 1989 Westat survey respondents reported sending an average of seven percent of their family income back to their country of ori gin through remittances. These figures were higher in Texas, with greater shares of the sample (over 70%) sending greater amounts (roughly 10% of family income) and lower in Florida. Thus, although legalized aliens had demonstrated firm ties to the United States through their extended residence, they remained connected to their countries of origin. This observation yields one of the more important themes in IRCA legalization. Field interviews with advocates indicate that one of the strongest incentives for legalizing, from the immigrant perspective, was the ri ght to travel internationally (Hagan and Baker, 1993). Second was the right to petition for relatives to immigrate legally. In sum, rather than fully integrating legalizing aliens into their identities as U.S. residents, legalization was away to keep their ties alive with the country of ori g in. With this profile, we can reevaluate the research questions outlined above. Who won? Clearly, the undocumented Mexican and Central American populations of the southwestern United States availed themselves most thoroughly of the legalization opportunity. The legalized population is also heavily male and concentrated in the ages of young adulthood. These results are due to both the composition of the eligible population and the implementation decisions made throughout the program. From program regulations to the nature of legalization publicity, the program targeted the undocumented border-crosser from a Spanish-speaking country. IRCA LEGALIZATION PROGRAMS IS TABLE 5 SELECTED ECONOMIC CHARACTERISTICS OF LEGALIZED POPULATION BY KEY IMPLEMENTATION SITES Panel A Occupation (%) Industry (%) State/MSA White Collar Blue Collar Agriculture Construction Manufacturing Service California 14 86 8 6 36 24 Los Angeles 16 84 3 6 42 24 Texas 8 92 7 19 16 29 Houston 8 92 3 21 14 25 Illinois 12 88 1 6 43 18 Chicago 11 89 2 6 39 20 New York 21 79 0 7 24 28 New York City 21 79 0 7 23 27 Florida 31 69 3 10 15 39 Miami 32 69 1 2 20 41 Total United States 1486371932 Panel B Median Median Family V^geat Individual Median Income Time of Earnings Family Remitted State/MSAApplication (1987) Income(%) California 5.50 9,913 16,703 6 Los Angeles 5.40 9,206 16,349 6 Texas 4.70 7,283 11,668 10 Houston 4.85 7,939 12,492 10 Illinois 5.56 10,045 19,117 7 Chicago 5.56 10,061 18,241 7 New York 6.00 10,633 14,295 7 New York City 6.00 10,643 13,978 7 Florida 6.25 8,902 17,971 3 Miami 6.00 7,857 14,863 4 Total United States5^58.982 15.3647 Source, Panel A: U.S. I mmigration and Naturalization Service, ImmigMtion Reform and Control Act: Report on the Legalized Alien Population. Washington, DC: U.S. Government Printing Office, 1992. Tables 21, 23, pp.33-36. Source, Panel B; U.S. I mmigration and Naturalization Service, Immigration Reform and Control Act: Report on the Legalized Alien Population. Washington, DC: U.S. Government Printing Office, 1992. Tables 25, 26, pp. 38-39. With these demographic characteristics come human capital repertoires reflecting low levels of education and English proficiency; and these characteristics translate into an unsurprising occupational, industrial and income profile. As would be expected, legalized aliens work in lower-skill, lower-pay jobs in the U.S. labor markets where they settle. 20 INTERNATIONAL MIGRATION REVIEW The legalized cohort is tied significantly to both U.S.-resident kin and to family members abroad. Not only do legalized immigrants demonstrate this through continued remittances, but they also continue to share U.S. households with both legally resident and undocumented family members. The next question becomes, what did the immigrants win? Legalization brought work authorization, travel authorization, and authorization to petition for relatives to immigrate. Although the transition to legal status did not translate immediately into dramatic economic gains for the legalized cohort, both INS data and field data indicate that legalized aliens took advantage of the opportunity to immigrate their relatives and to maintain ties to their countries of orig in. Thus, the relationship between the legalized alien and identity construction as a U.S. resident may be more complex than policymakers may have anticipated. While legal U.S. residence is the first step toward full citizenship and par ticipation in U.S. civic culture, legalized aliens appear to be using this resource, in the short term, as a bridge between home and host countries. In the words of one New York INS official, "The most valued card is the Employment Authorization Card. The 'Green Card' is no longer impor tant. " The oppor tunity to work, regardless of formal immigration status, is paramount. Additionally, little evidence exists to support the claim that the residual undocumented population has been "chilled" from staying in the United States, nor that new flows have been halted (Bean, Edmonston and Passel, 1990; Koussoudji, 1992; Donato, 1993). Rather, legalization has introduced a wider variety of possibilities in the building of international immigrant social networks and the strategies of work and household formation. NEW PRESSURE POINTS AND NEW OPPORTUNITIES: THE EFFECTS OF LEGALIZATION ON HOST COMMUNITIES The legalization program extends beyond the boundaries of the legalized. It also affects the social, economic and political landscape upon which immigrant and U.S.-born groups come together. Field interviews in key legalization sites yield several themes that allow us to assess how effectively legalization accomp lished its policy goals. A decade after IRCAs passage, field respondents report very little concrete labor market change. However, two trends did emerge in California: 1) an increase in immigrant day labor; and 2) a gradual transition from farm labor to low-wage manufacturing and service work. INS officials and advocates in Los Angeles noted that day-labor pools had expanded and often included IRCA-legalized aliens whose petitions for permanent resident status were still pending. INS respondents attributed this day-labor increase to their enforcement of employer sanctions reducing opportunities in the formal sector, although sanctions enforcement should not have affected the job search process IRCA LEGALIZATION PROGRAMS 21 for legalized aliens, who are authorized to work wherever they manage to get a job offer. Advocates pegged the trend, instead, to the stagnant California economy. In San Jose, advocates and employment training program directors noted that clients were beginning to come in from agricultural work to seek assistance in finding manufacturing jobs. "But, " as a training director noted, " they ' re still low-wage, unskilled jobs ... as long as their education level is so low, they won ' t move too far up the ladder. " Across the country, respondents noted that, in lieu of job changes, IRCAlegalized aliens evinced changes in awareness regarding their rights in the workplace. An immigration program director in San Jose noted that legalized aliens were making inquiries about union jobs. This kind of awareness fulfilled the expectations of a Houston carpenter ' s union local that assisted in processing amnesty applicants during early implementation in an effort to reach undocumented construction workers. As the union director noted, "We knew they were there, but we didnt have a way to reach them. ... By helping with legalization, we didn ' t have to preach unionism to them ... we had them at a time when they were going to be receptive. " Although union officials did not identify dramatic increases in their membership as a function of IRCA (and indeed very little change was noted in such key union strongholds as the garment workers in New York), they did identify legalization as a force reducing the barriers to successful organizing. As one respondent noted, "They re no longer looking over their shoulders for the INS van. " Nonetheless, the removal of " undocumented" status altered other constraints very little, like the dour state of the economy, particularly in California, during the post-implementation period. Respondents repeatedly noted that the full effect of legalization upon the labor market could not be seen separately from the lack of job growth characterizing the entire economy. In Los Angeles, an immigrant advocate noted that the office had seen the highest levels of unemployment among its immigrant constituency in recent memory. Again, this perspective coincided with the observations of INS officials who saw an upsurge in informal, irregular employment. With nearly 1 million legalized people residing in California, little economic mobility was anticipated by respondents for the population as a whole. Another constraint upon economic mobility cited frequently was the low education level among IRCA-legalized aliens. Here, educational service providers were particularly adamant about the disjuncture between the amnesty promise and its realization. The IRCA program for state reimbursement of social service costs - State Legalization Impact Assistance Grants (SLIAG) - came under particular fire. In Houston, a nonprofit service organization reported its frustration at the low level of reimbursement for its IRCA clients. Its SLIAG-funded services shut down in 1992 because, according to the director, "it just wasn ' t cost-effective. " What had been touted in 1988 by a 22 INTERNATIONAL MIGRATION REVIEW Houston respondent as the biggest Hispanic adult education program in the city' s history had become, by 1993, a program whose funding limitations removed providers with strong ties to the immigrant community from the pool of those to whom immigrants could go for educational assistance. In sum, the legalization program has not produced significant labor market change. Rather, information is the resource identified in the short term as accruing to legalized aliens. That is, linkages to social service providers and advocates have improved, so that the IRCA-legalized possess better information about their rights as p ermanent legal residents. But this integration into the host community infrastructure was clouded by a simultaneous theme emerging in the field - the anti-immigration backlash. In August 1993, California Governor Pete Wilson (who had been instrumental in the design and passage of IRCA as a U.S. Senator) mobilized immigration restrictionists by placing an " open letter " advertisement directed at President Bill Clinton in several leading newspapers calling for an end to citizenship rights for the U.S.-born children of undocumented immigrants and an end to federal mandates providing health and education services to undocumented immigrants. The letter touched off a firestorm of controversy surrounding both legal and undocumented immigration, and its effects are evident in abundance in the field interviews conducted for this project. Without exception, every respondent reported the "backlash" as a key immigration issue facing the local community. From the INS in Los Angeles, whose respondents noted that legal and illegal immigrants were being "lumped together unfairly," to a reporter in New York, who portrayed the community as consumed by " the politics of hysteria, " respondents noted that anti-immigrant sentiment was commanding most of the media attention on immigration. Although most observers agreed that anti-immigrant sentiment was strongest toward the currently undocumented, many noted an IRCA-related issue contributing to the anti-immigration milieu — the secondary migration of legalized alien family members. In site after site, respondents noted the increasing presence of whole families among their client populations. "The pre-1982 applicants started petitioning for family members as soon as they could. We estimate that 60 percent of our clients have done so, and this appears among all nationalities, " noted a San Jose educator. A Houston advocate explained the same dynamic in different terms: We see, particularly among Central Americans, the tendency to leave the kids at home, or even to send them back home when they ^ re bom here. . . . Now that the parents are legalized, they 've got Grandma calling them up saying, "Hey, when are you coming to get the kids?" The large cohort of IRCA beneficiaries, overwhelmingly from Mexico and Central America, all becoming eligible at the same time to petition for their IRCA LEGALIZATION PROGRAMS 23 immediate relatives, has produced an upswing in the salience of immigrant families in key sites and thus has produced an upswing in concerns about their use of such services as healthcare and education. In response to the " anti-immigration/anti-immigrant " sentiment, a variety of policy actors have now come together to promote another immigration benefit for the newly-legalized: naturalization. Both government and private advocacy group respondents identified naturalization as a top priority in immigration policy in the coming years - par ticularly for the IRCA-legalized cohort, who one advocate described as a ' special client community, " given its unique entry into legal status. The IRCA-legalized are seen as likely candidates for a break from the traditionally low naturalization rates characteristic of Mexican immigrants, for reasons outlined by a private nonprofit group attorney: They can bring their relatives more easily if they 're citizens. But, just as importantly, in this climate, the lines are being drawn in ways that further incentivize naturalization. The ante is upped now. If you're not a citizen, you're going to have to be concerned about changes in criminal ri ghts that are coming up in the Senate crime bill, you're going to have to worry about your eligibility under welfare reform, under health reform. Citizenship is self-protection. I hate to make it a defensive tactic, but there's a real sense among the advocates that we don't know what's coming next. .. . INS officials echoed this sentiment. Los Angeles officials noted that the next step should be to ' close the door on illegal immigration, but bring legal immigrants in completely. Help us help them become citizens. " In sum, the resources of the advocacy community and the INS are in alignment and directed toward naturalization, and respondents were optimistic about their chances of persuading the IRCA cohor t to naturalize. As legalization itself demonstrated, varied organizations marshalled toward a common goal in the implementation of immigration policy can have a profound impact on immigrant behavior. Yet, this element — immigrant behavior — remains as the unknown factor. Will IRCA immigrants avail themselves of the naturalization benefit? Evidence from the field is mixed. During early implementation, one immigration researcher noted that the immigrant community, particularly the Latino immigrant community, "has the least possible contact necessary with the government. . . they tend to want to deal with the government only as much as they have to in order to survive. " Today, respondents identify the backlash itself as producing a perverse incentive for IRCA immigrants to align themselves more deeply with their U.S. identity by naturalizing as a survival strategy, one that is facilitated by the economic crisis in Mexico and proposed changes in Mexican policy that might allow for dual-citizenshi p status. Should IRCA beneficiaries naturalize in great numbers, we are left to ponder economic and political implications. Ironically, some evidence suggests that the transformation to legal status may exacerbate competition effects with 24 INTERNATIONAL MIGRATION REVIEW native-born workers. That is, the presence of legal immigrants has demonstrated more consistent, albeit mild, depressing effects on wage rates and earnings in local labor markets than has the presence of undocumented workers (Bean, Lowell and Taylor, 1988). As the IRCA cohort moves up the job ladder, it may set off new competition dynamics with the native born, and, par ticularly, with other legal immigrants (Sorensen and Bean, 1994; Borjas, 1990). But will these dynamics become more pronounced with the acquisition of U.S. citizenship ? Here, again, we return to the issue of immigrant policy rather than immigration policy. The legacy of legalization, economic and political, will depend on both the human capital repertoires of the IRCA cohort and the fit between those repertoires and the labor market. With such modest skill levels, and few oppor tunities for education and training, amnestied citizens are likely to remain largely where they were as legal resident aliens - among the ranks of the working poor. Thus, concern about public assistance entitlements for poor people becomes the new immigrant policy battleground. The field interviews conducted in this project hi ghlig ht a growing political tendency for the distinction among immigration-status groups to be drawn in broader and broader terms. Whereas the political rhetoric surrounding IRCA focused on undocumented migration, the political debates today are more wide-ranging, with concerns being voiced about the overall level of immigration to the United States. With renewed emphasis on domestic reform in the areas of health, education and welfare, immigrants are finding themselves caught in a protracted battle over benefits eligibility and state/federal division of the cost for delivering those services. The impact of naturalization on this dynamic is uncertain. While facilitating access to public services may exacerbate the backlash, it can also facilitate the incorporation of the IRCA cohort by affording it the full array of social services intended to promote economic mobility. This is certainly the appeal being made to permanent resident aliens by naturalization advocates. If, instead, the legalized cohor t remains outside the boundaries of human-capital enhancing programs, it will likely continue on its current economic path - occupying lower-wage, lower-skill niches, subject to dislocation in times of economic downturn, and incurring the wrath of local political actors when its demands on the public coffers appear to outstrip its contributions. On its own terms, legalization worked well. It reached and successfully adjusted the status of millions of undocumented immigrants who are now more connected to the institutions of the host society and are the focus of an INS-sponsored campaign encouraging their further incorporation through naturalization. But as immigrant-focused policy promoting social and economic adaptation, IRCA left much to be desired. Field respondents strongly critique the arrangements to assist states in financing immigrant incorporation. The legacy is a new battlefield, pitting immigrants against natives once again. IRCA LEGALIZATION PROGRAMS 25 At present, the negative incentives of increased border enforcement, employer sanctions, and a reduction in funds for immigrant-serving social programs appear to have done little to stem immigration (Massey, 1995). At the same time, political energy is now being marshalled toward a package of positive incentives - citizenship and increased adult education opportunities, for example - that can channel legally-resident immigrants more rationally into their U.S. communities. The legalization experience illustrates that public agencies, partnered with trusted private organizations, can attract immigrants to use services. Although such efforts will not combat the continued flow of undocumented immigrants seeking to improve their economic life-chances, they improve the chances that legally resident immigrants, including the IRCA cohort, will succeed in their efforts to settle successfully in the United States. REFERENCES Baker, S. G. 1995 "California and Catalunya: The Politics of Immigration Control in the United States of America and the Kingdom of Spain." Unpublished manuscript. 1993 "Immigration Reform: The Empowerment of a New Constituency." In Public Policy for Democracy. Ed. H. Ingram and S. Rathgeb-Smith. Washington, DC: The Brookings Institution. Pp. 136-158. 1990 The Cautious Welcome: The Legalization Programs of the Immigration Reform and Control Act. Washington, DC: The Urban Institute Press. Bean, ED., et al. 1994 Illegal Mexican Migration and the United States/Mexico Border: The Effects of Operation Hold-the-Line on El Paso andJuarez. Washington, DC: U.S. Commission on Immigration Reform. Bean, E D., B. Edmonston andJ. Passel 1990 Undocumented Migration to the United States: IRCA and the Experience of the 1980s. Washington, DC: The Urban Institute Press. Bean, E D., B. L. Lowell and L. J. Taylor 1988 "Undocumented Mexican Immigrants and the Earnings of Other Workers in the United States," Demography, 25:35-52. Bierce,A. 1925 The Devil ' s Dictionary. New York: A. & C. Boni. Borjas, G. 1990 Friends or Strangers: The Impact of Immigrants on the U.S. Economy. New York: Basic Books, Inc. BorJas, G. and M. Tienda 1993 " The Employment and Wages of Legalized Immigrants," International Migration Review, 27(4): 712-747. Calavita, K. 1994 "Italy and the New Immigration." In Controlling Immigration: A Global Perspective. Ed. W Cornelius, PJL. Martin andJ. E Hollifield. Stanford: Stanford University Press. Pp.303-326. 26 INTERNATIONAL MIGRATION REVIEW Cornelius, W A. 1994 "Spain: The Uneasy Transition from Labor Exporter to Labor Importer." In Controlling Immigration: A Global Perspective. Ed. W Cornelius, R L. Martin and J. E Hollifield. Stanford: Stanford University Press. Pp. 331-370. Cornelius, W A., P L. Martin andJ. F. Hollifield 1994 "Introduction: The Ambivalent Quest for Immigration Control." In Controlling Immigration: A Global Perspective. Ed. W Cornelius, P.L. Martin andJ. F. Hollifield. Stanford: Stanford University Press. Pp. 3-42. Department of Labor 1995 "Effects of the Immigration Reform and Control Act of 1986: Characteristics and Labor Market Behavior of the Legalized for Five Years Following Legalization." U.S. Department of Labor. Washington, DC. J uly 27. Donate, K. 1993 "Current Trends and Patterns of Female Migration: Evidence from Mexico," International Migration Review, 27 (4): 748-771. Donate, K., J. Durand and D. Massey 1992 "Stemming the Tide? Assessing the Deterrent Effects of the Immigration Reform and Control Act," Demography 29(2) ; 139-158. Fix, M. and P.Hill 1990 Enforcing Employer Sanctions: Challenges and Strategies. Washington, DC: Urban Institute Press. Freeman, G. 1994 "Can Liberal States Control Unwanted Migration?" The Annals of the American Academy of Political and Social Science, 534. July. Garcia y Griego, M. 1994 "Canada: Flexibility and Control in Immigration and Refugee Policy." In Controlling Immigration: A Global Perspective. Ed. W Cornelius, P.L. Martin andJ. F. Hollifield. Stanford: Stanford University Press. Pp. 119-140. Hagan, J. M. 1994 Deciding to be Legal. Philadelphia, PA: Temple University Press. Hagan, J. M. and S. G. Baker 1994 "Implementing the U.S. Legalization Program: The Influence of Immigrant Communities and Local Organizations on Immigration Policy Reform," International Migration Review, 27(3):513-536. Hollifield, J. F. 1994 "Immigration and Republicanism in France: The Hidden Consensus." In Controlling Immigration: A Global Perspective. Ed. W Cornelius, P.L. Martin andJ. F. Hollifield. Stanford: Stanford University Press. Pp. 143-176. Hondagneu-Sotelo, P. 1995 Gendered Transitions: Mexican Experiences of Immigration^ Berkeley: University of California Press. Immigration and Naturalization Service, U.S. (INS) 1994 "INS Estimates of Nonimmigrant Overstays, Resident Illegal Alien Population," Unpublished memorandum. 1992 The Immigration Reform and Control Act: Report on the Legalized Alien Population. Washington, DC: Government Printing Office. March. Koussoudji, S. 1992 "Playing Cat and Mouse at the U.S.-Mexican Border," Demography, 29(2): 159-181. IRCA LEGALIZATION PROGRAMS 27 Loev. Thomburgh 1989 No. 88 Civ. 7363 (PKL) (S.D. New York). In Litigation Update: Legalization/Special Agricultural V^brkers. Washington, DC: American Immigration Lawyers Association. June 16.2. Lowell, B. L. and Z. Jing 1994 "Unauthorized Workers and Immigration Reform: What Can We Ascertain from Employers?" International Migration Review, 28(3):427-448. Martin, P.L. 1994 "Good Intentions Gone Awry: IRCA and U.S. Agriculture," The Annals of the American Academy of Political and Social Science, 534:44-57. July. Massey, D. 1995 "The New Immigration and Ethnicity in the United States," Population and Development Review, 21(3):631-652. Massey, D., K. Donate and Z. Liang 1991 "Effects of Immigration Reform and Control Act of 1986: Preliminary Data from Mexico. In Undocumented Migration to the United States: IRCA and the Experience of the 1980s. Ed. F. Bean etaL Washington, DC: The Urban Institute Press. Meissner, D. and D. Papademetriou 1988 The Legalization Countdown: A Third-Quarter Assessment. Washington, DC: The Carnegie Endowment for International Peace. Meissner, D., D. Papademetriou and D. North 1986 Legalization of Undocumented Aliens: Lessons from Other Countries. Washington, DC: Carnegie Endowment for International Peace. North, D. 1991 "IRCA Did Not Do Much to the Labor Market: A Los Angeles County Case Study, " Immigration Policy and Research Working Paper 10. Washington, DC: Department of Labor. 1988 Decision Factories. Washington, DC: Transcentury Development Associates. North, D. and A. M. Portz 1989 The U.S. Legalization Program. Washington, DC: Transcentury Development Associates. Sorensen, E. and E Bean 1994 "Immigration Reform and the Wages of Mexican Origin Workers: Evidence from Current Population Surveys," Social Science Quarterly, 75 (I): 1-17. March. Snowden, L. 1990 "Collective versus Mass Behavior: A Conceptual Framework for Temporary and Permanent Migration in Western Europe and the United States," International Migration Review, 24:577-590. Warren, R. and J. Passel 1987 'A Count of the Uncountable: Estimates of Undocumented Aliens Counted in the 1980 U.S. Census," Demography, 24(3):375-393. Determinants of English Proficiency among Mexican Migrants to the United States We replicate prior research into the determinants of English language proficiency among immigrants using a dataset that controls for potential biases stemming from selective emigration, omitted variables, and the mismeasurement of key constructs. In general, we reproduce the results of earlier work, leading us to conclude that despite inherent methodological problems, research based on cross-sectional censuses and surveys yields fundamentally accurate conclusions. In particular, we find unambiguous evidence that English proficiency rises with exposure to U.S. society, and we reaffirm earlier work showing a clear pattern of language assimilation among Mexican migrants to the United States. The recent surge in immigration to the United States has revived a longstanding debate about the extent and rate of immigrant assimilation, and one facet of assimilation — language — has drawn unusual attention. Social scientists have examined this issue from two viewpoints: the degree of English proficiency and the extent of English usage. These two dimensions of linguistic assimilation are quite different from one another. Whereas English dominance suggests underlying assimilation into U.S. culture (see Veltman, 1983, 1988; Mirowsky and Ross, 1984; Grenier, 1984), English proficiency by itself implies nothing about the degree of identification with U.S. culture or the extent to which English is used in practice. We focus on English language proficiency because studies have shown it to be a crucial step in the broader process of economic mobility within the United States. Tienda and Neidert (1984) demonstrate that English proficiency increases socioeconomic status among male Hispanic immigrants, whether or not they retain Spanish. Similarly, Stolzenberg (1990) shows that if Hispanic men speak English very well and have completed at least twelve years of schooling, their occupational status is nearly identical to that of non-Hispanic whites. McManus, Gold and Walsh (1983) demonstrate that English proficiency is strongly and positively related to earnings, a finding that has been replicated many times (see Chiswick, 1984,1986,1991; Chiswick and Miller, 1988,1993; Massey, 1987a; Tainer, 1988). Whether or not one favors policies to encourage the retention of immigrant languages, proficiency in English is crucial to broader economic success. Because of the close connection between socioeconomic status and the demand ©1997 by the Center for Migration Studies of New York. All rights reserved. 0197-9183/97/3101.0117 Kristin E. Espmosa Douglas S. Massey University of Pennsylvania ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS 29 for public social services, moreover, the degree and rate of English language acquisition also carry important fiscal implications. Studies show that the likelihood of using public social services increases as socioeconomic status falls (Tienda and Jensen, 1986; Jensen, 1988; Donato and Massey, 1993). For several reasons, therefore, understanding how and why immigrants acquire proficiency in English is an important policy issue. The standard social science approach to analyzing the process of language attainment among immigrants is based on the human capital model, in which acquiring English is viewed as an investment made in order to improve economic opportunities (Grenier, 1984; Chiswick, 1986, 1991; Jasso and Rosenzweig, 1990). In making a decision about whether or not to learn English, an immigrant weighs the expected benefits against the anticipated costs, which include the effort required for study and practice, the expenses of language instruction, the psychic costs involved in embracing a new linguistic culture, and the opportunity costs of income foregone while learning the new language. In its most basic form, the human capital model looks only at individual characteristics; but since the costs and benefits of English acquisition cannot be measured directly, traits such as age, duration of U.S. residence, education, and gender are assumed to be indirect indicators of the net returns to fluency (^ Vel tm an, 1983; Grenier, 1984; Chiswick, 1986). Studies have shown that people who enter the country at younger ages attain greater proficiency in English than those who arrive when they are older (Grenier, 1984). If immigrants arrive as children, they are especially likely to learn English because of their exposure to the language within U.S. schools (Veltman, 1983, 1988). Controlling for age of arrival, the odds of speaking English rise with years of U.S. residence (Grenier, 1984; Veltman, 1983,1988; Chiswick, 1991; Stevens, 1992, 1994) and education (Stevens, 1994; Lopez, 1978; Jasso and Rosenzweig, 1990). The relationship between English proficiency and education is probably reciprocal, however (Mirowsky and Ross, 1984; Tienda and Neidert, 1984). Women are generally less likely to learn English than men (Jasso and Rosenzweig, 1990); but once they do learn English, they tend to speak it more often (Stevens, 1986). The human capital model has been expanded to incorporate household-level variables. Grenier (1984) found that Spanish speakers married to non-Hispanics were less likely to continue using the language, and Stevens (1992) found that members of linguistic minorities who married native English speakers were more likely to use English than those who married someone of their own group. Again, however, the relationship between exogamy and English proficiency is probably reciprocal, since not only do those who marry native English speakers have stronger incentives to learn English, but immigrants who speak English are more likely to marry natives (Stevens, 1985; Stevens and Swicegood, 1987; Stevens and Schoen, 1988). 30 INTERNATIONAL MIGRATION REVIEW Work by Grenier (1984) suggests that another household factor, the presence of children, also affects the likelihood of English usage, at least in Hispanic origin groups. Having young children in the family is generally associated with hig h levels of Spanish language retention, while older children increase the likelihood that parents use English. The effect of young children is particularly significant for women, supporting Stevens 5 (1986) view that mothers, in an effort to transmit immigrant culture, speak foreign languages to their children. The human capital model has also been broadened to include communitylevel factors such as the size of the immigrant community and its level of segregation. Jasso and Rosenzweig (1990) argued that the costs of not acquiring English were significantly lower for Hispanics in areas with greater concentrations of Spanish speakers; and they showed that Latino immigrants living in urban areas with large Hispanic populations were, in fact, less likely to learn English. Stevens (1992) replicated this finding and showed that an individuals English proficiency also depends on the degree of segregation experienced by group members, with segregation operating to lower proficiency. She also found that group size affected English usage indirectly through marriage patterns, as well as by directly determining exposure to English. Despite the consistency of these results, prior studies of English language acquisition suffer from several drawbacks. First, most rely on rather crude measures of time spent in the United States that do not reflect the reality of modem migration processes. Studies typically measure duration using a question that asks respondents when they " came to the United States to stay. " For groups in which circular migration is common, notably Mexicans, this measure is likely to understate the total amount of U.S. experience. A seasonal migrant who happens to be in the United States at the time of the census or a survey may answer the question by saying " six months, " indicating the time elapsed since most recent entry, even though he or she may have accumulated 60 months across ten earlier trips. Likewise, a settled migrant who answers " three years " may actually have accumulated a total often years across several trips that were taken prior to setdement. A second problem with earlier studies is that virtually all of them employ cross-sectional samples of immigrants interviewed in the United States. Such samples confound the process of language acquisition with that of selective emigration, yielding potentially biased estimates of model coefficients (Lindstrom and Massey, 1994). If immigrants who have difficulty learning English tend to return home over time, then people at advanced experience levels are not representative of all cohort members who originally entered the United States. On the contrary, they are the survivors of a process of selective emigration in which the "linguistic failures " have steadily returned home. The longer the duration, the greater the selection, and the more skewed toward linguistic success the survivors will be. ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS 31 Since return migration is quite common in many immigrant groups (see]^sso and Rosenzweig, 1982,1990; Warren and Kraly, 1985), the typical curve of rising English proficiency with growing U.S. experience could par tly or totally reflect the effect of selective emigration rather than exposure to U.S. society. Moreover, if emigration rates are linked to individual characteristics such as age, sex and education — as evidence presented byJasso and Rosenzweig (1990) and Stevens (1994) suggests - then estimates of the effect of human capital variables on English language proficiency will also be biased (Lindstrom and Massey, 1994). Finally, prior research generally has not controlled for legal status, despite the fact that undocumented migrants are enumerated in several data sources commonly used to study English proficiency, such as the U.S. census and the Current Population Survey (^^ Passel and Woodrow, 1984, 1987; Warren and Passel, 1987). The problem, of course, is that legal status is not available in these sources; but a lack of documents represents a clear disincentive to learning English. If one ' s stay in the United States can end at a moment s notice, then it makes little sense to invest heavily in English language training. Unless documentation is controlled, therefore, model coefficients may be biased downward, and intergroup differences in proficiency will be confounded with intergroup differences in the relative proportion of undocumented migrants. This ar ticle examines the process of English language acquisition among Mexican migrants to the United States, using a unique dataset that contains a fuller and more complete measure of U.S. experience, provides an accurate indication of legal status, and includes migrants surveyed in their home communities as well as those located in U.S. destination areas. Mexican migrants provide a particularly oppor tune case for systematic study since they include a large number of undocumented migrants, display a high degree of circular migration, and are characterized by a high rate of emigration (see Massey et aL, 1987). Thus, they maximize the potential for bias and error in studies using normal procedures and data. Focusing on Mexicans is also advantageous because they are the largest and most visible immigrant group in the United States. In 1991, more than 947,000 Mexicans were admitted as permanent legal immigrants (U.S. Immigration and Naturalization Service, 1992) and the total Mexican origin population in that year reached 13.4 million persons (U.S. Bureau of the Census, 1991). During the 1980s, the annual level of undocumented Mexican migration ran into the hundreds of thousands (Woodrow, 1992). Worries about linguistic assimilation have been exacerbated for Mexican immigrants, in par ticular, because so many are undocumented, because they concentrate in several large urban communities, and because they tend to be poorly educated, all factors that impede the process of language acquisition. A detailed study of Mexican immigrants thus provides a strong test of the linguistic assimilation hypothesis. 32 INTERNATIONAL MIGRATION REVIEW DATA AND METHODS Our data come from samples of returned U.S. migrants surveyed in 25 Mexican communities during the winter months of 1987 through 1994, and from nonrandom samples of outmigrants from those same communities interviewed in U.S. destination areas during the subsequent summers. The communities were selected to provide an array of different sizes, economic bases, political categories, and ethnic compositions; they are located in the states ofJalisco, Michoacan, Guanajuato, and Nayarit, which together constitute a region that has consistently sent a majority of migrants to the United States (see Gamio, 1930; Dagodag, 1975; North and Houstoun, 1976; CONAPO, 1986; Jones, 1988). Basic characteristics of the sample are summarized in Table 1. For each Mexican community, 100 to 200 households were selected and interviewed using random sampling methods (in one community 364 households were sampled), and in most cases ten to twenty outmigrant households were located and surveyed in the United States using snowball sampling techniques. Since the U.S. samples are not strictly representative of settled outmigrant communities, we use a set of weights to account for their relative contribution to the total sample. (5' ^ Massey and Parrado, 1993, for details.) These weights are applied to each case in our analyses. After wei ghting, persons interviewed in the United States comprise just over 18 percent of our sample, meaning that almost one in five members of 25 binational communities lives north of the border. Each household was administered an ethnosurvey questionnaire (^^Massey, 1987b) that collected basic information on family members ' demographic, social and economic characteristics. Complete life histories were compiled for all household heads; those who had been to the United States also received a detailed battery of questions about the most recent U.S. trip. Because the question on English language ability was asked within this battery, our analysis focuses on household heads who reported making at least one trip to the United States. English language proficiency was measured at the rime of the respondent s last tri p using a four-point ordered scale where 0 indicated that the respondent did not speak or understand English; 1 meant that he or she did not speak English but understood it somewhat or well; 2 signified that he or she spoke and understood some English; and 3 indicated that he or she spoke and understood English well. This variable was analyzed as an ordered probit. The distribution of cases across the categories of this variable is shown in Table 2. While more than one-third (37%) of the migrants in the sample reported neither speaking nor understanding English, more than one-fifth understood some English, and a quarter spoke as well as understood some. Nearly 18 percent reported that they spoke and understood English well. ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS TABLE 1 CHARACTERISTICS OF MEXICAN COMMUNITIES SAMPLED FOR STUDY OF MIGRATION TO THE UNITED STATES Households on Mexican Size of U.S. 1990 Political Year of Sampling Sample Sampling Refusal U.S. Sampling State Population Category Survey Frame Size Fraction Rate Sample Fraction Metropolitan Area' 1 2.Guanajuato 867,920 b Mun. Seat198786l c 200 19.Michocan 492,90l b State Cap.19913,578 C 200 15.Guanajuato 362,915 b Mun. Seat19912,009'200 29.Michocan 217,068bMun. Seat1992IW200 28.Jalisco 74,068 b Mun. Seat1992\W201 Smaller Urban Areas' 1 I.Guanajuato 52,291Mun. Seat19877S(f200 26.Guanajuato 33,123Mun. Seat19922,76^200 9.Michocan 32,474Mun. Seat19896 ,776(l 200 17.Jalisco 30,882Mun. Seat19911,77^200 27.Guanajuato 23,726Mun. Seat19921.774 0 200 13.Guanajuato 20,614Mun. Seat1990 3,771(l200 II.Nayarit 19,645Mun. Seat19904,472(l 200 4.Guanajuato 16 ^ Mun. Seat19882,723 d 200 Towns B 12.Nayarit ll '54l Tenencia19902,693 d 200 18.Zacatecas 7,750Mun. Seat19911.71711364 14.Michocan 7,025Mun. Seat19901,44^200 8.Michocan W^ Tenencia1989l,395 d 200 6.Jalisco 4»760Mun. Seat1988 79911200 3.JaHsco 3,516Mun. Seat1988510 d 200 Ranchos a 7.Jalisco 3,098Tenencia1988 53411 200 10.Michocan 2 ' 240Tenencia1989 44811 150 5.Guanajuato 1»737Tenencia1988248 d 150 3.Zacatecas 5,785Tenencia1991233 d 116 16.Guanajuato WOTenencia1991143 d 100 25.Talisco 894Tenencia19922l4 d 100 "Community names have been suppressed to ensure confidentiality of respondents copulation of metropolitan area. '^nsus of neighborhood within community. complete census of all households in community. 232.1190.000 056.08320.098 100.05720.999 184.08313.065 103.07420.052 256.03420.121 072.09515.023 030.03720.009 113.04420.038 113.12715.217 053.04720.047 045.02920.012 073.05720.816 074.01020 212.12720 139.05720 143.05020 250.11520 392.03820 375.010 335.152 605.085 498.025 699.029 467.029 34 INTERNATIONAL MIGRATION REVIEW TABLE 2 DISTRIBUTION OF RESPONDENTS ACROSS LEVELS OF ENGLISH LANGUAGE ABILITY: HOUSEHOLD HEADS FROM 25 MEXICAN COMMUNITIES Ability Category Score Percentage Does Not Speak or Understand English ^ ^7.4 Does Not Speak but Understands English Some or Well ^ ^ -^ Speaks and Understands Some English 2 23.6 Speaks and Understands English Well 3 ^ Average/A^ 1 .22 2,208 These outcomes were expressed as a function of selected independent variables defined at the individual, household and community levels, which are listed in Table 3. They are grouped into two basic categories: factors that are typically found in census or survey data and those that are uniquely available in our data. Individual characteristics available in census data include personal traits such as sex and education, while occupation in Mexico is an atypical individual-level factor available only from our special survey. We examine several other aspects of an individuals migration history that are generally unavailable in census or survey sources, including the amount of U.S. experience acquired before the most recent trip, age at the time of the first trip, and the number of years that have elapsed since legal documents were obtained. Since English proficiency is more likely to be acquired over the course of a long, continuous stay in the United States than over the course of a series of short trips, we control for the duration of the most recent trip, the average number of trips taken per year between the first and most recent visit, and we also measure the proportion of total time the respondent has spent in the United States since the first entry. Although respondents were asked to report their English language ability as of the time of the last U.S. trip, this trip may have occurred some time ago. In this event, former migrants might understate their ability by reporting current English proficiency rather than the level they had achieved when they were last in the United States. If language ability deteriorates without practice, then the gap between English proficiency at the time of the last trip and English ability at the time of the survey will tend to increase the longer the migrant has been back in Mexico. To control for potential biases arising from this fact, we measure the number of months elapsed since the migrant returned to Mexico from his or her last trip to the United States (which was set to 0 if the migrant was in the United States at the time of the survey). English proficiency is also likely to depend on characteristics and conditions surrounding entry. Most of these are generally available in standard data sources. From census or survey data, the investigator can usually determine the age at time of arrival, whether or not the migrant is accompanied by spouse or ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS TABLE 3 MEANS AND STANDARD DEVIATIONS OF VARIABLES USED IN ANALYSIS OF ENGLISH LANGUAGE PROFICIENCY: 2,208 HOUSEHOLD HEADS FROM 25 MEXICAN COMMUNITIES Standard Explanatory Variables Mean Deviation Minimum Maximum Typical Census Variables Personal characteristics Female 0.05 0.22 0.00 Years of Schooling 5.38 4.31 0.00 Most recent trip characteristics Age 33.18 11.73 2.00 Period of last entry 1940-1950 0.03 0.17 0.00 1950-1959 0.07 0.25 0.00 1960-1964 0.05 0.22 0.00 1965-1969 0.04 0.20 0.00 1970-1974 0.09 0.29 0.00 1975-1979 0.12 0.32 0.00 1980-1981 0.05 0.23 0.00 1982-1984 0.07 0.26 0.00 1985-1986 0.09 0.28 0.00 1987-1990 0.32 0.47 0.00 1991-1994 0.07 0.26 0.00 Spouse on trip 0.38 0.49 0.00 Number of children on trip 1.19 1.82 0.00 Children in U.S. schools 0.31 0.46 0.00 Household characteristics Number of children < 18 2.25 2.24 0.00 U.S.-bom children 0.33 0.47 0.00 U.S. receiving community Los Angeles area 0.36 0.48 0.00 San Francisco Bay Area 0.08 0.28 0.00 Chicago area 0.06 0.23 0.00 Other 0.44 0.50 0.00 Additional Variables Migrant background Months of prior U.S. experience 56.01 79.55 0.00 Documented 0.33 0.47 0.00 Years since documents obtained 3.92 8.91 0.00 Average number of trips per year 0.68 0.46 0.00 Proportion of time in U.S. 0.33 0.33 0.00 Duration (months) of most recent trip 53.05 93.51 1.00 Months in Mexico since return 85.08 138.21 0.00 Age on first U.S. trip 23.99 8.31 0.00 35 l.OG 22.00 85.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 14.00 1.00 13.00 1.00 1.00 1.00 1.00 1 00 612.00 1.00 62.00 7.50 1.00 534.00 598.00 81.00 INTERNATIONAL MIGRATION REVIEW TABLE 3 (CONTINUED) MEANS AND STANDARD DEVIATIONS OF VARIABLES USED IN ANALYSIS OF ENGLISH LANGUAGE PROFICIENCY: 2,208 HOUSEHOLD HEADS FROM 25 MEXICAN COMMUNITIES Standard Explanatory Variables Mean Deviation Minimum Maximum Period of first U.S. trip 1940-1950 1950-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1981 1982-1984 1985-1986 1987-1990 1991-1994 Occupation on first U.S. trip Out of labor force Professional, technical, administrative, service Skilled Unskilled Agricultural Migrant status of family members Spouse a U.S. migrant Child a U.S. migrant Parent a U.S. migrant Grandparent a U.S. migrant Number of siblings U.S. migrants Other U.S. social connections Member of sports/social club Has Chicano/Latino friends Has Anglo friends Has Afro-American friends In touch with townspeople Lodged with relative or friend Household characteristics Land owned Business owned Personal characteristics Mexican occupation pior to last tri(. Out of labor force 0.04 0.20 0.00 1.00 Professional, technical, administrative, service 0.25 0.43 0.00 1.00 Skilled 0.12 0.32 Unskilled 0.25 0.43 0.00 1.00 Agricultural 0.34 0.47 0.00 1.00 Community characteristics Population of municipio Proportion of males U.S. mierants 0.100.300.001.00 0.130.340.001.00 0.070.250.001.00 0.080.280.001.00 0.160.370.001.00 0.200.400.001.00 0.060.240.001.00 0.070.250.001.00 0.060.240.001.00 0.060.240.001.00 0.010.080.001.00 0.040.200.001.00 0.170.370.001.00 0.080.260.001.00 0.230.420.001.00 0.470.500.001.00 0.260.440.001.00 0.560.500.001.00 0.370.480.001.00 0.090.290.001.00 1.611.810.006.00 0.180.390.001.00 0.650.48 0.500.500.001.00 0.280.450.001.00 0.670.470.001.00 0.670.470.001.00 0.100.290.001.00 0.150.350.001.00 65,975 80,994 5,175 804,287 0.24 0.09 0-01 0.62 ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS 37 children, and whether or not the migrant has children enrolled in U.S. schools. One facet of the last trip is typically not measured in census data, however - whether or not the migrant possessed legal documents. Unlike prior studies of English language acquisition, we are able to introduce detailed controls for the amount of social capital accessible to migrants in the United States. Within the family we measure whether or not a parent or grandparent had ever been to the United States and count the number of siblings with U.S. migration experience. As of the time of the last entry, we also determine whether or not the migrant s spouse or child had been to the United States; whether or not the individual belonged to a U.S. sports association or club; whether or not he or she knewAnglos, blacks, Chicanos, or Latinos in the United States; and whether or not the respondent was in contact with fellow townspeople while abroad. Finally, we measure the extent to which social ties are used instrumentally by asking whether or not the respondent lodged with a friend, relative, or townsperson upon arrival in the United States. Household-level characteristics employed in the analysis include a dichotomous variable indicating whether or not the household owned land and another dummy variable indicating whether or not it owned a business. These variables are typically unavailable from censuses or surveys administered in the United States, but we also include two other household-level indicators generally available from such sources: whether the household contains children born in the United States and, as a measure of household dependency, the number of children under age 18. In order to complete the analysis, we examine several community-level factors in Mexico that are, by definition, unavailable from U.S. sources. To assess the degree of contact with U.S. culture within the respondents community, we include the relative prevalence of U.S. migratory experience among men in the community and the population of the surrounding municipio (roughly equivalent to a U.S. county). The prevalence index is the ratio of men aged 15 and over with U.S. migratory experience divided by the total number of men aged 15 and over, measured in the year when the last trip was taken (.^Massey, Goldring and Durand, 1994). Although the diversity of U.S. destinations makes it difficult to assess receiving community effects, we have included a set of dummy variables to indicate whether or not the respondent went to one of the principal Mexican settlement areas of the United States: Los Angeles, San Francisco, or Chicago. This information, of course, is readily available in census-based studies of language assimilation. The mean values presented in Table 3 provide a snapshot of migrant household heads captured by this binational sample. Nearly all were males (95%) of labor force age at the time they undertook their last trip (mean age 33). On average they had completed just over five years of schooling and most 38 INTERNATIONAL MIGRATION REVIEW were employed either in agricultural or unskilled occupations in Mexico (34% and 25%, respectively). The average age at which they made their first trip to the United States was 24, and almost a half (47%) worked in U.S. agriculture on their first trip. By the time of their last U.S. trip, the migrants in our sample had accumulated more than four and a half years of migratory experience (about 56 months), underscoring the problematic nature of duration measures defined from the date of last entry. Most migrants had made multiple trips, averaging about two trips every three years (0.7 per year), and the typical respondent spent just about one-third of his or her time since the first trip (33%) in the United States. One-third of the respondents had documents at the time they made their most recent trip. The average length of that trip was almost four and a half years (53 months), and the typical migrant had returned to Mexico about seven years prior to being surveyed (85 months). Respondents had access to considerable social capital in the United States. More than a third (37%) reported a migrant parent, 9 percent had a migrant grandparent, and the average number of siblings with U.S. experience was 1.6. (This average is probably biased downward slightly because migration information was only collected for six siblings.) While in the United States on their last trip, nearly two-thirds (65%) of the respondents reported having a Chicano or Latino friend and half said they had Anglo friends; only 28 percent reported Afro-American friends, however. Two-thirds (67%) said they were in touch with other community members while abroad, and the same percentage said they had lodged with a friend, relative, or townsperson when they arrived in the United States. Some 18 percent stated that they belonged to a U.S. sports club (usually a soccer club) or some other recreational association. Respondents generally headed households that were landless (90%) and only 15 percent reported owning a business. The level of family dependency was relatively high, with the typical household containing 2.25 members under the age of 18. In about a quarter (26%) of the households the respondents spouse had migrated to the United States at some point before the respondent ' s most recent trip, and fully 56 percent had children who had migrated by that time. One-third of the migrants reported U.S.-born children. The average migrant came from a municipio with around 66,000 residents, but this belies a huge range in size from just over 5,000 to more than 800,000 inhabitants. Within the respondents home community, the percentage of men with migrant experience ranged from 1 percent to 62 percent, with a mean value of 24 percent. On their most recent trip, fully 36 percent of the sample went to the Los Angeles metropolitan area, by far the largest Mexican community in the United States. Only 8 percent went to the San Francisco Bay Area; only 6 percent went to Chicago, currently the second-largest settlement of ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS 39 Mexicans in the United States. The remaining 44 percent of respondents were scattered over a wide variety of different U.S. destinations. THE DETERMINANTS OF ENGLISH PROFICIENCY We estimate four successive ordered probit models to examine the effect of different sets of factors on English language proficiency. The first model provides baseline estimates similar to those that would be obtained from a typical analysis of census data. We then add in successive sets of variables unavailable from the census. The second model adds in detailed indicators of the respondent ' s migrant background; the third model includes measures of social capital; and the final model adds in selected factors pertaining to the migrant ' s background in Mexico. The results of the exercise are presented in Table 4. Levels of significance are indicated with asterisks, and the associated standard errors are reported in the Appendix. The estimates for Model 1 replicate many of the findings established by earlier studies of English language assimilation. The odds of speaking proficient English generally rise as education increases and are boosted substantially by having U.S.-born children and offspring enrolled in U.S. schools. In addition, although the effect is not statistically significant, women are generally less likely to speak English well. The lack of significance probably reflects the small number of female migrants, which produces rather large standard errors (in-depth interviews also suggest women are more likely than men to understate their English proficiency). Compared to migrants going to Los Angeles, those going to Chicago were considerably more likely to speak proficient English, while those going to other places (largely agricultural destinations) were less likely to do so. The odds of speaking English well were not significantly different for migrants going to San Francisco and Los Angeles. Although English language proficiency was unrelated to age or the presence of children, it was related (positively) to the presence of a spouse. English language ability was also significantly influenced by period of entry. This variable is usually interpreted to indicate the length of exposure to U.S. society, and the typical finding is that the earlier the period of entry, the hi g her the reported proficiency. Our estimates, however, display precisely opposite this pattern: as the period becomes more recent, the odds of speaking proficient English rise. This apparent anomaly does not indicate that recent immigrant cohorts have higher aptitudes for English, however, or that proficiency somehow decays with exposure to U.S. society. Rather it stems from the flawed nature of period of last entry as an indicator of U.S. experience in an immigrant population characterized by considerable movement back and forth, such as that sampled here. Mexican migrants make multiple trips and accumulate 40 INTERNATIONAL MIGRATION REVIEW TABLE 4 ORDERED PROBIT REGRESSION OF SELECTED EXPLANATORY VARIABLES ON ENGLISH LANGUAGE PROFICIENCY: HOUSEHOLD HEADS FROM 25 MEXICAN COMMUNITIES (STANDARD ERRORS IN APPENDIX) Explanatory VariablesModel 1Model 2Model 3Model 4 Typical Census Variables Personal characteristics Female -0.128 Years of schooling 0.122** Characteristics of most recent trip Age -0.001 Age 2 0.000003 Period of entry 1940-1950  1950-1959 0.459 1960-1964 0.824** 1965-1969 0.743** 1970-1974 0724** 1975-1979 0.554* 1980-1981 0^678** 1982-1984 0.854** 1985-1986 0.743** 1987-1990 0.930** 1991-1994 0^629** Spouse on trip 0.459** Number of children on trip 0.033 Children in U.S. schools 0.446** Household characteristics Number of children < 18 -0.007 U.S.-bom children 0.880** Receiving community in U.S. Los Angeles area — San Francisco Bay Area -0.122 Chicago area 0.396** Other -0.123* Additional Variables Migrant background Months of pior U.S. experience Documented Years since documents obtained Average number of trips per year Proportion of time in U.S. Duration (months) of most recent trip Months in Mexico since return Age on first U.S. trip 0.046-0.060-0.143 0.133**0.121**0.120*' 0.079**0.065**0.049* 0.0002-0.00010.0001 0.838*-0.648-0.697 0.799-0.362-0.363 1.466**-0.937-1.038* •1.737**-1.115-1.193* 2.015**-1.343*-1.511* 2.067**-1.345-1.554* 1.942**-1.192-1.455 2.266**-1.465-1.760* 2.199**-1.368-1.662 2.672**-1.733-1.994* 0.316**0.200*0.174 0.043-0.040-0.028 0.290**0.226*0.250** 0.0270.0220.029 0.190*0.1720.124 0.345**0.381**0.398** 0.2380.2140.212 0.012-0.044-0.028 0.002*0.002**0.002* 0.086-0.244**-0.220* 0.0010.0040.0002 0.295**0.278**0.296** 0.697**0.505**0.553** 0.006**0.007**0.007** 0.002-0.001-0.001 0.134**-0.123**-0.117** 0.001*0.001*0.001* ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS TABLE 4 (CONTINUED) ORDERED PROBIT REGRESSION OF SELECTED EXPLANATORY VARIABLES ON ENGLISH LANGUAGE PROFICIENCY: HOUSEHOLD HEADS FROM 25 MEXICAN COMMUNITIES (STANDARD ERRORS IN APPENDIX) Explanatory Variables Model 1 Model 2 Model 3 Model 4 Period ofBrst U.S. trip 1940-1950 1950-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1981 1982-1984 1985-1986 1987-1990 1991-1994 Occupation on first U.S. trip Agicultural Out of labor force Professional, technical, administrative, service Skilled Unskilled Migrant status of family members Spouse a U.S. migrant Child a U.S. migrant Parent a U.S. migrant Grandparent a U.S. migrant Number of siblings U.S. migrants Other U.S. social connections Member of sports/social club Has Chicano/Latino friends Has Anglo Friends Has Afro-American friends In touch with townspeople Lodged with relative or friend Household characteristics Land owned Business owned Personal characteristics Most recent Mexican occupation Agricultural Out of labor force Professional, technical, administrative, service Skilled Unskilled Community characteristics Log of population ofmunicipio Proportion of males U.S. migrants Intercept 1 Intercept 2 Intercept 3 Log likelihood N * ^<.05 ** /»<.01 0.416* 0.434* 0.338 1.025** 0.919** 0.826** 1.593** .418** 1.329** 1.672** .621** 1.439** 1.755** .721** 1.535** 1.846** .703** 1.524** 1.809** .808** 1.557** 2.267** 2.103** 1.864** 1.929** 1.930** 1.679** 1.722* 1.687* 1.324 0.182 0.244 -0.043 0.306** 0.311** 0.172 0.260* 0.232* 0.107 0.379** 0.371** 0.329** 0.050 0.031 0.057 0.092 -0.010 0.013 0.018 -0.013 0.106** 0.096** 0.354** 0.342** 0.269** 0.276** 0.361** 0.377** 0.047 0.008 -0.129* -0.128* -0.186** -0.195** -0.084 0.277** 0.547** 0.406** 0.340** 0.068 0.093* 1.373** 0.339 -1.157 -1.980* 1.090 -1.165 -1.181 2.439 -2.579 -2.630 1,859.23 1,748.49 1,723.46 2,182 2,107 2,107 42 INTERNATIONAL MIGRATION REVIEW considerable time in the United States before the last entry. As the estimates of Model 2 reveal, once migrant background and experience are appropriately controlled, period-of-entry effects are reversed and the explanatory power of the model improves significantly (seethe shift in likelihood ratios, p < .05). The pattern of rising coefficients across periods observed in Model 1 occurred because Mexicans migrating during the late 1980s were increasingly those who had accumulated large amounts of U.S. experience. Another change from Model 1 is that age is now significant and positive, indicating that migrants who made their most recent trip at older ages had more English skills at the time of the interview than their younger counterparts, even after controlling for amount of U.S. experience. The effects of the variables indicating receiving community also shifted. In Model 2, migrants who were in San Francisco on their most recent trip had greater English ability than those in Los Angeles, while the coefficients for Chicago and other places were no longer significant. Most indicators of migrant background are highly significant predictors of English proficiency. Consistent with hypotheses derived from human capital theory and the assimilation model, the odds of speaking English rise significantly as migrants' total experience accumulates across U.S. trips, and they increase substantially as the total propor tion of time spent in the United States grows. They also rise significantly with tri p duration. The greater the average number of trips the migrant took per year since the first trip is related to higher English proficiency at the time of the interview, but documentation and time spent back in Mexico do not have significant effects. The odds of speaking English with greater proficiency are also lowered by having taken a first trip at an older age, although the significant and positive squared term for age at first arrival suggests that the effect diminishes as age rises. This result is again consistent with human capital theory since immigrants who arrive at older ages have less time to reap whatever material and psychological rewards stem from English language acquisition, but they must pay all of the costs. The period of entry effects indicate that, after controlling for age, migrants who made their first trip more recently had greater English skills at the time of the survey. Finally, the occupational niche that migrants occupy upon initial entry into the United States also plays a role in determining the odds of learning English. Compared to migrants who worked initially in agriculture, those who worked in other occupations upon first arrival were more likely to learn English proficiently. This effect may indicate the influence of the workplace on English language ability, but it probably also reflects the fact that Mexicans who had some English ability upon arrival were more likely to be able to obtain nonagricultural jobs upon their initial arrival in the United States. ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS 43 Model 3 adds measures of the migrant s social capital to the prediction equation, which again boosts the explanatory power significantly (seethe shift in likelihood ratios, p < .05). The addition of these variables causes a loss of significance among the period of most recent entry variables. Having U.S.-born children also becomes statistically insignificant, but now documented migrants are significantly less likely to speak and understand English. In general, the greater a migrants array of social connections to the United States, the higher the likelihood of speaking English well. The odds of English proficiency are significantly raised for migrants who belong to a U.S. sports or social club, and migrants who have Anglo or Chicano and Latino friends tend to report much hi g her levels of language proficiency on their most recent trip. Learning English has less to do with immediate family ties within the United States, beyond having an accompanying spouse or offspring in U.S. schools. Among the family connections we considered, only having siblings with prior migrant experience seems to affect odds of acquiring English: as the number of siblings with U.S. migrant experience increases, the likelihood of speaking English well increases. Finally, in Model 4, we add variables covering different aspects of the migrant ' s background in Mexico. Including these variables changes very few of the other coefficients in the model, although it once again boosts the explanatory power significantly (seethe shift in likelihood ratios, p < .05). Once we control for the most recent job in Mexico, indicators of occupation on first U.S. trip are no longer significant, except that those who worked as unskilled laborers still evince greater English ability compared with those working in agriculture. The most recent occupation in Mexico displays a contrasting pattern of effects: those who had done unskilled work before leaving were not significantly different from those who had worked in agriculture, but other occupational groups displayed greater English proficiency. Although owning land had no effect on a migrants ability to speak and understand English, business ownership increased those skills. Community characteristics also prove meaningful. The greater the population of the migrant s municipio of origin, the better he or she is able to understand and communicate in English. Finally, the more involved a community is in the migration process, the greater the likelihood of English proficiency. Thus, as the share of men with U.S. experience rises in a respondent s community, so do the odds that he or she will speak and understand English proficiently. DISCUSSION In order to shed new light on the process of English language assimilation in the United States, we employed a unique, binational source of data on Mexico-U.S. migrants that overcomes some of the weaknesses of earlier 44 INTERNATIONAL MIGRATION REVIEW studies. Prior work has been largely based on cross-sectional censuses or surveys carried out among immigrants located in the receiving country, and they generally have not included adequate measures of legal status, total migrant experience, and other variables relevant to the process of language acquisition that are usually unmeasured in cross-sectional data sources. Our analysis reaffirms most of the findings established by prior studies of linguistic assimilation, lending some confidence to the belief that they are not simply ar tifacts of sample selection or omitted variable bias. We thus concur with earlier research in showing that the odds of acquiring English language ability rise with education and are increased by having children in U.S. schools. We also replicate earlier findings that women are less likely to speak and understand English well, although the significance of this result is undermined by the relatively small number of women in our sample. Our results for period of entry are also similar to those in census-based models, demonstrating that, net of migrant background, social capital and community effects, migrants whose latest trips took place recendy are less likely to achieve English proficiency. Like most prior research, moreover, we find clear and unambiguous evidence that English proficiency rises sharply with exposure to U.S. society. The greater the experience built up over multiple trips, and the greater the share of time Mexican migrants spend in the United States, the more likely they are to speak English well. Although the odds of achieving proficiency are conditioned on the age at first arrival, our results strongly reaffirm earlier work showing a clear pattern of language assimilation among Mexican migrants to the United States. Thus, our replication generally suggests that prior analyses carried out on cross-sectional data yield fundamentally accurate and unbiased estimates for effects for most variables. The principal discrepancies we uncovered concerned the effects for age and the presence ofU.S.-born children. Once we controlled for social capital and migrant background, we found that having U.S.-born children did not significantly influence reported English language ability and that English proficiency tended to rise, rather than fall, as age increased. Even though we confirm earlier results, however, our analyses suggest certain limitations in prior analyses and in par ticular point to the relevance of factors that are usually not measured in censuses and surveys. The possession of legal documents plays a key role in conditioning the odds of English language acquisition. Surprisingly, however, the effect of documents is opposite to that predicted. Migrants who lack documents actually spoke and understood significantly more English at the time of their most recent tri p than did those who already had received legal U.S. residence. A migrant ' s access to social capital also plays an impor tant role in structuring the process of language acquisition. The odds of English proficiency are much higher for migrants with siblings who are also U.S. migrants, and they are also hi gher for those who have more extensive contacts with members of U.S. racial and ethnic groups. ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS 45 We also find that migrants ' background in Mexico has significant consequences for their odds of being able to speak and understand English at the time of their most recent trip. At the individual level, those who owned businesses and those in skilled, professional or service occupations had greater capabilities in English. Community characteristics also matter. Both community size and the proportion of adult men with U.S. migrant experience had significant positive effects on the odds of English language acquisition. At the broadest level, therefore, our results question the alarm frequently expressed about the threat of immigration to the status of English as the language of government, commerce and public affairs in the United States. Our study of people from the largest contemporary source of immigrants to the United States, a population whose connection to the United States and level of sociocultural integration are often regarded as problematic, we find very clear evidence of an ongoing process of linguistic assimilation. Among Mexican immigrants to the United States, English proficiency is acquired in a highly structured fashion as part of a well-defined social process. As migrants become more integrated and acquire more ties north of the border, the costs of learning English drop, the benefits increase, and the odds of speaking English will grow commensurately. If Spanish is increasingly spoken in the United States, it is because a large number of Spanish-speaking immigrants arrived in a short time, not because there is any perceptible resistance on the part of Mexican migrants to learning English per se. Furthermore, judging from our models, recent initiatives to refuse public schooling to undocumented children may end up being counterproductive. As we show, schooling strongly boost the odds of English language proficiency among adults. If the ultimate goal of these policies is to reduce the costs of undocumented migration, then they may have the opposite effect, for in lowering English proficiency, they will also lower immigrants ' wages and tax receipts and increase rates of social service usage. A more successful strategy might be to increase the ties between Mexican immigrants and U.S. society, thereby raising their English proficiency and socioeconomic status and decreasing the potential burden on states such as California, Texas, and Illinois. 46INTERNATIONAL MIGRATION REVIEW APPENDIX STANDARD ERRORS FOR ORDERED PROBIT REGRESSION OF SELECTED EXPLANATORY VARIABLES ON ENGLISH LANGUAGE PROFICIENCY acolanatorv VariablesModel 1Model 2Model 3Model 4 Typical Census Variables Personal characteristics Female 0.118 0.139 Years of schooling 0.007 0.008 Characteristics of most recent trip Age 0.014 0.023 Age 2 0.0002 0.0002 Period of entry 1940-1950 — — 1950-1959 0.241 0.348 1960-1964 0.243 0.432 1965-1969 0.251 0.507 1970-1974 0.231 0.569 1975-1979 0.228 0.658 1980-1981 0.242 0.710 1982-1984 0.234 0.752 1985-1986 0.232 0.794 1987-1990 0.222 0.843 1991-1994 0.237 0.877 Spouse on trip 0.087 0.091 Number of children on trip 0.021 0.023 Children in U.S. schools 0.082 0.090 Household characteristics Number of children < 18 0.015 0.017 U.S.-bom children 0.082 0.095 Receiving community in U.S. Los Angeles area — — San Francisco Bay Area 0.093 0.103 Chicago area 0.124 0.136 Other 0.057 0.064 Additional Variables Migrant background Months of prior U.S. experience 0.001 Documented 0.095 Years since documents obtained 0.005 Average number of trips per year 0.067 Proportion of time in U.S. 0.167 Duration (months) of most recent trip 0.001 Months in Mexico since return 0.001 Age on first U.S. trip 0.023 Aee 2 0.0003 0.1430.144 0.0080.008 0.0240.024 0.00020.0002 0.3600.367 0.4490.455 0.5230.531 0.5880.597 0.6780.689 0.7320.745 0.7760.789 0.8170.831 0.8680.882 0.9040.919 0.0970.097 0.0230.024 0.0930.094 0.0170.018 0.1170.118 0.1080.111 0.1410.143 0.0660.069 0.0010.001 0.0990.100 0.0060.006 0.0680.069 0.1730.175 0.0010.001 0.0010.001 0.0230.024 0.00030.0003 ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS APPENDIX (CONTINUED) STANDARD ERRORS FOR ORDERED PROBIT REGRESSION OF SELECTED EXPLANATORY VARIABLES ON ENGLISH LANGUAGE PROFICIENCY ixolanatorv Variables Model 1 Model 2 Model 3 Model 4 Period of first U.S. trip 1940-1950 1950-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1981 1982-1984 1985-1986 1987-1990 1991-1994 Occupation on first U.S. trip Agricultural Out of labor force Professional, technical, administrative, service Skilled Unskilled Migrant status of family members Spouse a U.S. migrant Child a U.S. migrant Parent a U.S. migrant Grandparent a U.S. migrant Number of siblings U.S. migrants Other U.S. social connections Member of sports/social club Has Chicano/Latino friends Has Anglo friends Has Afro-American friends In touch with townspeople Lodged with relative or friend Household characteristics Land owned Business owned Personal characteristics Most recent Mexican occupation Agricultural Out of labor force Professional, technical, administrative, service Skilled Unskilled Community characteristics Log of population ofmunicipio Proportion of males U.S. migrants Intercept 1 Intercept 2 Intercept 3 0.193 0.1990.202 0.282 0.2890.291 0.340 0.3500.354 0.395 0.4040.409 0.455 0.4650.471 0.508 0.5200.526 0.539 0.5510.558 0.568 0.5820.589 0.610 0.6250.634 0.753 0.7740.777 0.169 0.1730.216 0.082 0.0850.093 0.110 0.1150.127 0.074 0.0770.084 0.0850.086 0.0970.098 0.0630.063 0.0950.097 0.0170.018 0.0760.078 0.0710.072 0.0700.071 0.0710.072 0.0630.064 0.0700.071 0.101 0.082 0.210 0.098 0.117 0.091 0.040 0.458 0.833 0.860 0.951 0.045 0.049 0.049 0.065 0.070 0.071 48 INTERNATIONAL MIGRATION REVIEW REFERENCES Chiswick, B. R. 1991 "Speaking, Reading, and Earnings among Low-skilled Immigrants," Journal of Labor Economics, 9:149-170. 1986 "Is the New Immigration Less Skilled than the Old?" Journal of Labor Economics, 4:168-192. 1984 "Illegal Aliens in the United States Labor Market: Analysis of Occupational Attainment and Earnings," International Migration Review, 18:714—732. Chiswick, B. R. and P.W Miller 1993 "The Endogeneity between Language and Earnings: An International Analysis." Unpublished Paper, Department of Economics, University of Illinois at Chicago. 1988 "Earnings in Canada: The Roles of Immigrant Generation, French Ethnicity, and Language," Research in Population Economics, 6:183—224. Consejo Nacional de Poblacion (CONAPO) 1986 Encuesta en la Frontera a Trabajadores Indocumentados Devueltospor las Autoridades de losEstados Unidos de America: Diciembrede 1984. Mexico, D.E: Consejo Nacional de Poblacion. Dagodag, "W. T. 1975 "Source Regions and Composition of Illegal Mexican Immigration to California," International Migration Review, 9:499-511. Donato, K. M. and D. S. Massey 1993 "Legal Status and Social Service Use: Mexican Migrants to the United States." Paper presented at the Annual Meetings of the Population Association of America, April 1-3, Cincinnati, Ohio. Gamio, M. 1930 Mexican Immigration to the United States. Chicago: University of Chicago Press. Grenier, G. 1984 "Shirts to English as Usual Language by Americans of Spanish Mother Tongue," Social Science Quarterly, 65:537-550. Jasso, G. and M. R. Rosenzweig 1990 The New Chosen People: Immigrants in the U.S. New^brk ; Russell Sage Foundation. 1982 "Estimating the Emigration Rates of Legal Immigrants Using Administrative and Survey Data," Demography, 19:279-290. Jensen, L. I. 1988 "Patterns of Immigration and Public Assistance Utilization, 1970-80," International Migration Review, 22:51-83. Jones, R. C. 1988 "Micro Source Regions of Mexican Undocumented Migration," National Geographic Research, 4:11-22. Lindstrom, D. P.and D. S. Massey 1994 "Selective Emigration, Cohort Quality, and Models of Immigrant Assimilation," Social Science Research, 23:315-349. Lopez, D. E. 1978 "Chicano Language Loyalty in an Urban Setting," Sociology and Social Research, 62:267-278. ENGLISH PROFICIENCY AMONG MEXICAN MIGRANTS 49 Massey, D. S. 1987a "Do Undocumented Migrants Earn Lower Wages than Legal Immigrants? New Evidence from Mexico," International Migration Review, 21:236—274. 1987b "The Ethnosurvey in Theory and Practice," International Migration Review, 21:1498-1522. Massey, D. S., R. Alarcon, J. Durand and H. Gonzalez 1987 Return to Aztlan: The Social Process of International Migration from Western Mexico. Berkeley and Los Angeles: University of California Press. Massey, D. S., L. P.GoldringandJ. Durand 1994 "Continuities in Transnational Migration: An Analysis of 19 Mexican Communities," American Journal of Sociology, 99:1492-1534. Massey, D. S. and E. A. Parrado 1993 "Migradollars: The Remittances and Savings of Mexican Migrants to the USA," Population Research and Policy Review, 13:3-30. McManus, W, W Gould and F. Welch 1983 "Earnings of Hispanic Men: The Role of English Language Proficiency," Journal of Labor Economics, 1:101-103. Mirowsky, J. and C. E. Ross 1984 "Language Networks and Social Status among Mexican Americans," Social Science Quarterly 65:551-564. North, D. S. and M. E Houstoun 1976 The Characteristics and Role of Illegal Aliens in the U. S. Labor Market: An Exploratory Study Washington, DC: Linton. Passel, J. S. and K. A. Woodrow 1987 "Change in the Undocumented Alien Population in the United States, 1979-83, " International Migration Review, 21:1304-1323. 1984 "Geographic Distribution of Undocumented Immigrants: Estimates of Undocumented Aliens Counted in the 1980 Census by State," International Migration Review, 18:642-671. Stevens, G. 1994 "The English Language Proficiency of Immigrants in the U.S." In Immigration and Ethnicity: The Integration ofAmericas Newest Arrivals. Ed. B. Edmonston and J. S. Passel. Washington, DC: Urban Institute Press. Pp. 163-185. 1992 "The Social and Demographic Context of Language Use in the United States," American Sociological Review, 57:171—185. 1986 "Sex Differences in Language Shift in the United States," Sociology and Social Research, 71:31-35. 1985 "Nativity, Intermarriage, and Mother-Tongue Shift," American Sociological Review, 50:74-83. Stevens, G. and R. Schoen 1988 "Linguistic Intermarriage in the United States," Journal of Marriage and the Family, 50:267-279. Stevens, G. and G. Swicegood 1987 "The Linguistic Context of Ethnic Endogamy," American Sociological Review, 52:73-82. 50 INTERNATIONAL MIGRATION REVIEW Stolzenberg, R. M. 1990 ethnicity Geography, and Occupational Achievement of Hispanic Men in the United States," American Sociological Review, 55:143-154. Tainer, E. 1988 "English Language Proficiency and Earnings among Foreign Bom Men," Journal oj Human Resources, 23:108-122. Tienda, M. and L. Jensen 1986 "Immigration and Social Program Participation: Dispelling the Myth of Dependency," SocialScience Research, 15:372-400. Tienda, M. and L. Neidert 1984 "Language, Education, and the Socioeconomic Achievement of Hispanic Origin Men" Social Science Quarterly, 65:519-536. U.S. Bureau of the Census 1991 The Hispanic Population of the United States. Current Population Reports, Series P-20, No. 455. Washington, DC: U.S. Government Printing Office. U.S. Immigration and Naturalization Service 1992 1991 Statistical Yearbook of the Immigration and Naturalization Service. Washington, DC: U.S. Government Printing Office. Veltman, C. 1988 The Future of the Spanish Language in the United States. New York: Hispanic Policy Development Project. 1983 "Modelling the Language Shift Process of Hispanic Immigrants," International Migration Review, 22:545-562. Warren, R. and E. P Kraly 1985 The Elusive Exodus: Emigration from the United States. Population Reference Bureau Occasional Paper No. 8. Washington, DC: Population Reference Bureau. Warren, R. andJ. S. Passel 1987 "A Count of the Uncountable: Estimates of Undocumented Aliens Counted in the 1980 United States Census," Demography, 24:375-393. Woodrow, K. A. 1992 *A Consideration of the Effect of Immigration Reform on the Number of Undocumented Residents in the United States," Population Research and Policy Review, 11(2): 117-144. Geographicall y Indirect Immigration to Canada: Description and Analysis This ar ticle is concerned with geographically indirect immigration to Canada over the period 1968—1988. A geographically indirect immigrant is an individual legally admitted to Canada whose country of last permanent residence differs from country of birth. Records maintained by Employment and Immigration Canada on every immigrant legally admitted over the period were used in the study. Relative to geographically direct immigrants, geographically indirect immigrants tend to be older, more educated, and more highly skilled. Moreover, if they were not born in an English or French speaking country, indirect immigrants are more likely to speak English and/or French capably than direct migrants born in such countries. The study also contains bivariate logit estimates of a model of geographically indirect Canadian immigration. This model suggests that indirect migrants tend to be influenced by personal characteristics (age, sex, marital status, occupation, language ability), as well as by various characteristics of the country of birth (distance from Canada, income level, political conditions). For many years the United States and Canada have been among the world's major countries of immigration, almost every year ranking among the top five such countries in terms of annual flows (United Nations, 1979). A great deal of descriptive and formal analytical research has focused on immigration to and from these North American neighbors, as well as on migration between the two countries. For example, Boyd (1976) compares immigration to the United States and Canada between the late 1950s and the early 1970s, par ticularly stressing differences in the admittance policies of the two countries. Michalowski (1991) studies the emigration of foreign-born Canadians and shows that a large fraction of Canada ' s immigrant population subsequently depar ts. Recently, the U.S. Bureau of the Census and Statistics Canada (1990) jointly prepared a detailed descri ptive study of migration between the countries. Moreover, Greenwood and McDowell (1991) model the determinants of immigration to the United States and Canada as a system. One aspect of North American immigration that has not been studied extensively is geographically indirect migration, which occurs when an immigrant has a country of last permanent residence different from his or her ©1997 by the Center for Migration Studies of New York. All rights reserved. 0197-9183/97/3101.0117 Michael J. Greenwood Paul A. Young University of Colorado at Boulder 52 INTERNATIONAL MIGRATION REVIEW country of birth. 1 On a worldwide scale this phenomenon has been studied far less than called for based on the frequency of its occurrence. This point is illustrated by the fact that various influential compendiums (Shryock, Siegel et aL ^ 1976; Ross, 1982) make no mention of geographically indirect immigration, and one of the most comprehensive studies of international migration (United Nations, 1979) only briefly mentions specific indirect flows, such as those that have occurred through Hong Kong. Barrett (1976) and Gould (1979), apparently believing the p henomenon to be important, discuss geographically indirect international migration, but they provide no substantive data that relate to the relative importance of this type of movement. Greenwood and Trabka (1991) suggest that the reason for the lack of quantitative detail is that published immigration statistics fail to report immigrants crossclassified by country of birth and country of last permanent residence. Using the U.S. Immigration and Naturalization Service Micro Data files, Greenwood and Trabka (1991) cross-classify immigration by country of birth and country of last permanent residence. They show that for the United States, at least, geographically indirect immigration is important. In 1987, 76,032 persons, or 12.6 percent of all legal U.S. immigrants, migrated indirectly to the United States. Not only do Greenwood and Trabka provide the first detailed numerical accounting of geographically indirect U.S. immigration, they also show that indirect migrants differ in many respects from geographically direct migrants. For example, the indirect group is generally older and more skilled. The present study uses microdata records on every legal immigrant to Canada from 1968 to 1988 to provide a detailed descriptive account of geographically indirect migration to Canada. This account parallels that of Greenwood and Trabka to yield comparable information on the United States and Canada. However, the present study goes well beyond that of Greenwood and Trabka in two respects. First, the Canadian data provide more detail than the U.S. data, specifically with respect to schooling and Canadian language (t.e.y English and/or French) capabilities. This additional information allows a more detailed description of Canadian indirect immigration. Second, whereas Greenwood and Trabka discuss how indirect migration might be modeled without actually developing a formal model, in the present study we estimate a model of geographically indirect migration to Canada. We believe that this is the first time that a model has been estimated for this important international migration phenomenon. The dataset used for this paper was compiled by Employment and Immigration Canada. As noted above, it contains a record on every immigrant legally Geographically indirect migration has been studied in the context of internal migration especially for the United States and Canada. The major distinctions in this literature are between primary (or "new" or first time) migrants and repeat migrants, who are separated into return and nonreturn (sometimes called "onward") migrants (.^DaVanzo, 1978; Kau and Sirmans, 1976;Vanderkamp, 1972). GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA 53 admitted to Canada from 1968 to 1988. Among other information, each record contains gender, marital status, immigration class, Canadian language capabilities, years of schooling, and intended occupation. Similar to the Immigration and Naturalization Service files on U.S. immigrants, both the country of birth and the country of last permanent residence are given, allowing the researcher to distinguish indirect from direct migrants. This article has two main parts, the first descriptive and the second analytical. In the descriptive part, we provide various details on indirect migration to Canada including: the overall importance of indirect migration, by immigrant class of entry; specific patterns (or " paths") taken by indirect immigrants to Canada and changes over time in these patterns; and personal characteristics of indirect relative to direct migrants, including age, skills, schooling, and language abilities. We develop a number of descriptive measures because Canadian geographically indirect immigration, to our knowledge, has not been previously studied in such detail. The stylized facts that come out then lead to our bivariate logit model of geographically indirect immigration, including specification and estimation for three years - 1980, 1984, and 1988. INDIRECT CANADIAN IMMIGRATION, 1968-1988 For the period 1968-1988, Table 1 shows annual immigration to Canada distinguished by migrant type (/.^., direct versus indirect). During this period, 458,640 persons, or 16.0 percent of all immigrants, moved to Canada from a country different than their country of birth. The highest annual indirect total occurred in the first year of the period (41,683 in 1968), whereas in 1988 21,147 persons migrated indirectly, the hi ghest total since 1976. While the first two years shown in Table 1 had the hi g hest percentages of indirect migrants, with 22.7 percent of all migrants going indirectly to Canada in 1968 and 19.5 percent in 1969, the most noteworthy fact is that prior to 1979 the percentage of indirect migrants remained relatively stable at an average of approximately 18 percent. However, beginning in 1979, this percentage fell to about 14 percent and, after this decline, remained approximately stable once again. One explanation for this occurrence is that in 1979 immigrants for the first time entered primarily under the Immigration Act of 1976. 2 The new law of 1976 apparently made indirect migration to Canada more difficult. It placed more emphasis on practical training and labor market needs and less on formal education. This shift seems to indicate that the law could work either for or against indirect migrants since (as shown below) they have more years of schooling, but are also older and more experienced. However, indirect immigrants are more likely to qualify under the independent class than 2 The Immigration Act of 1976 went into effect in 1978. The data show, however, that in 1978, 84.2% of all immigrants fell under the old legislation. In 1979, this percentage dropped to 10.3%. 54 INTERNATIONAL MIGRATION REVIEW TABLE 1 COUNTRY OF BIRTH THE SAME OR DIFFERENT FROM COUNTRY OF LAST PERMANENT RESIDENCE: CANADIAN IMMIGRANTS, 1968-1988 Total Same Different Different 'Year Immigrants Country Country Country (%) 1968 183,984 142,301 41,683 22.7 1969 162,882 131,116 31,766 19.5 1970 148,303 122,161 26,142 17.6 1971 120,976 100,237 20,739 17.1 1972 124,118 102,962 21,156 17.1 1973 185,761 154,454 31,307 16.9 1974 217,702 181,488 36,214 16.6 1975 189,600 154,233 35,367 18.7 1976 145,779 117,885 27,894 19.1 1977 114,053 93,155 20,898 18.3 1978 85,645 70,058 15,587 18.2 1979 3 112,282 96,544 15,738 14.0 1980 143,287 125,386 17,901 12.5 1981 128,711 111,290 17,421 13.5 1982 121,253 104,803 16,450 13.6 1983 89,291 78,037 11,254 12.6 1984 88,431 77,524 10,907 12.3 1985 84,496 73,822 10,674 12.6 1986 99,587 88,480 11,107 11.2 1987 152,758 135,470 17,288 11.3 1988161,166140,01921,14713.1 Source: Public Use Tapes from Employment and Immigration Canada. Beginning in 1979, the new immigration law went into effect. direct immigrants. The independent class requires qualifying under a point system for which the number of points for educational attainment dropped from 20 to 12, whereas vocational training could account for up to 15 points. The maximum number of points attainable from all sources was unchanged at 100. This shift in emphasis away from formal education appears to have reduced the flow of indirect migration to Canada. Table 2 divides indirect immigrants into the four separate immigration classes: family class, refugee, independent class, and assisted relative. Assisted relative is actually a subgroup of the independent class, but is separated for greater clarity. The class into which an immigrant falls refers to the specific circumstances that entitle the person to become a permanent resident of Canada. To enter under the assisted relative and independent classes, the immigrant (except for retirees) must qualify under a point system based upon Canadian demographic and labor market needs, whereas for the family class GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA 55 TABLE 2 CANADIAN GEOGRAPHICALLY INDIRECT IMMIGRATION BY IMMIGRANT CLASS:  SELECTED YEARS, 1968-1988  Immigration Class 1968 1972 1976 1980 1984 1988 Independent 20,259 11,161 9,337 7,412 3,256 13,099 Family Class 5,993 4,677 9,080 6,872 5,839 4,931 Assisted Relative 5,853 4,876 7,719 1,704 584 1,546 Refugees 3 9,364 327 1,758 1,911 1,228 1,571 Unknown 214 115 0 2 0 0 Total:41,683 21,156 27,894 17,901 10,907 21,147 Source: Public Use Tapes from Employment and Immigration Canada. refugees include all immigrants who entered under humanitarian concern, including those who entered under the Designated Class and the Refugee Class. and for those entering under humanitarian concerns, the formal selection criteria do not apply. 3 As noted above, indirect migrants are more likely to qualify under the independent class than direct migrants. For example, in 1976 33.5 percent of those who moved to Canada indirectly were in the independent class, compared to 30.4 percent of the direct migrants. Comparable respective figures are 41.4 percent and 24.6 percent in 1980,29.9 percent and 22.8 percent in 1984, and 61.9 percent and 39.2 percent in 1988. Until 1988, a general decline occurred in the relative importance of the independent and assisted relative classes as means of entry for indirect immigrants, whereas the family class accounted for a larger percentage of these immigrants. This trend reversed for the independent and family classes in 1988 with 61.9 percent of indirect immigrants falling under the independent class and the family class percentage falling from 53.5 percent in 1984 to 23.3 percent. In recent years, refugees have accounted for a surprisingly small percentage of indirect migrants, even when compared to the percentages of direct migrants who are refugees. For example, in 1980 30.6 percent of the direct immigrants were refugees, whereas only 11.4 percent of the indirect immigrants were refugees. Comparable respective figures are 11.3 percent and 18.4 percent in 1984, and 7.4 percent and 17.9 percent in 1988. Greenwood and Trabka (1991) find that the number of indirect migrants to the United States increased during the 1980s due to increased admittance of refugees and that U.S. refugees make up a relatively large ^he independent class includes retirees, investors, entrepreneurs, self-employed persons, and other immigrants applying on their own initiative. Assisted relatives are individuals who receive additional points for having a relative offer support for up to five years. To qualify under the family class, an applicant must be sponsored by a close relative who resides in Canada. These sponsored relatives often include dependents. The Immigration Act of 1976 (effective 1979) created three designated classes which allowed immigrants to enter Canada under humanitarian concerns without meeting the strict definition of the United Nations Convention. Both government-sponsored Convention refugees and members of a designated class are deleted from the analytical work. For the purposes of the present study, the terms "refugee " and "humanitarian migrant " are used to describe both types of migrants and are used synonymously. 56 INTERNATIONAL MIGRATION REVIEW fraction of U.S. indirect immigrants. The same does not hold for Canada, For example, Greenwood and Trabka find that in 1984, out of the 80,467 indirect moves to the United States, 32,541 (40.4%) were refugees. For the same year, Canada had 10,907 indirect moves, of which 1,228 (11.3%) were refugees. COUNTRIES OF BIRTH AND LAST PERMANENT RESIDENCE For geographically indirect migrants to Canada, Table 3 provides information on specific countries of birth for the years 1968, 1976, 1980, and 1988. The ten most frequent countries of birth for indirect migrants are identified for each year, as well as those countries of birth of a significant number of refugees. Certain countries are particularly consistent in sending a large number of indirect migrants to Canada, such as China, the United Kingdom, and India, which are ranked in the top five for all four years. The rankings for other countries are more variable. For example, Yugoslavia, Italy, and Portugal dropped in relative importance, with lower totals for 1988 than previously. In contrast, the number of indirect migrants from Iran and Vietnam increased over the same period and peaked in 1988. The percentage of those moving indirectly to Canada varies considerably by country of birth. Several countries have a large fraction who moved indirectly, such as China. Indirect movement consistently accounts for a large percentage of migrants from Malaysia, Thailand, Kenya, and Tanzania. In contrast, the United Kingdoms percentage never exceeds 8.4 percent, and comparable figures are 6.2 percent for the United States, 8.3 percent for Hong Kong, and 5.5 percent for the Philippines. Many of the countries of birth listed in Table 3 have declining percentages of migrants entering indirectly {e.g., India, Yugoslavia, Czechoslovakia, Poland). For Poland and India this phenomenon is due to an increased number of direct migrants, whereas Yugoslavia and Czechoslovakia have had fewer migrants pass through Germany and Austria. Thailand is one of the few countries of birth that has had increasingly larger shares of indirect migrants, which is due to the large flow of refugees through Kampuchea on the way to Canada (only 23 in 1980, but 303 in 1984 and 198 in 1988). Table 4 is similar to Table 3 except that it provides information on country of last permanent residence rather than country of birth. Hong Kong, the United Kingdom, and the United States rank in the top five in each of the four years for which data are reported, but Hong Kongs impor tance increased significantly and accounted for a larger percentage of all indirect immigrants in 1988 than previously. Whereas Hong Kong is an important intermediate country, it has become even more important as a country of birth for direct migrants. In 1988, 23,187 persons moved to Canada from Hong Kong, 76.6 percent of whom were born in Hong Kong. GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA TABLE 3 TOTAL AND GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA BY COUNTRY OF BIRTH:  SELECTED YEARS, 1968-1988  Country 1968 1976 1980 1988 of Birth Indirect Total Indirect Total Indirect Total Indirect Total Czechoslovakia 8,969s1 ' 9,854267335 China 4,889s1 5,0765, 121s1 5,907 Yugoslavia 2,387s1 6,8414912,185 U.K. 2,110s1 33,8131,578 s 19,011 Italy 1,942 s 20,878556 s 3,853 India 1,625s1 4,6732,045s1 8,448 West Germany 1,100s1 6,153 509s1 2,255 Portugal 1,075s1 3,7671,285 s 3,877 Canada 1,052 s 1,052868 s 868 Jamaica 852 s 3,5993297,093 Poland 828 1,8564781,318 U.S.A. 403 17,07934013,691 U.S.S.R. 357 491995^1,238 Hong Kong 199 3,3523176,374 Kenya 134 3374581,275 Phillipines 125 2,7613006,056 Malaysia 119 241195597 Lebanon 108 1,1082206,139 Iran 98 24252379 Tanzania 66 150550 s 1,644 Thailand 16 402385 Chile 8 145764 a • c 2,675 Vietnam 8 9197 11 229 Kampuchea45156'^254 Source: Public Use Tapes from Employment and Immigration Canada indicates country made one of top ten largest totals for year. Indicates at least 25% were refugees. indicates at least 75% were refugees. In terms of the relative importance of indirect immigration from specific countries of last permanent residence, cer tain countries clearly are preferred to others, including France, Israel, and Saudi Arabia. For each year repor ted in Table 4, as countries of last permanent residence, more than 50 percent of all immigrants from Israel, over 33 percent from France, and over 75 percent from Saudi Arabia were born in a different country. On the other hand, China and Vietnam have relatively lower percentages of indirect migrants. When economic and political conditions in China and Vietnam are taken into consideration, this observation is not surprising. These types of considerations will be taken into account in the model we speci fy and estimate below. 1081,20977921 4,093 s 8,9665,261 s 7,871 140781841,397 1,209s1 16,449659 s 7,893 2571,873198958 1,250 s 9,5341,641 s 11,889 380 s 1,7452281,556 2122,023803,978 576 s 576338338 1953,2441223,999 2601,39531411 9,311 1978,1063475,558 424 s 2,438555 s 1,147 321^3,874526 s ' 11 18,281 2614592401,202 1646,149418 s 8,637 252874588 s 2,125 1341,251879 s 3,695 2161,145557 s 4,088 333 s 713251654 86°461379°528 1021,257681,001 397 s ' 11 24,694537 s 6,159 399 s ' 0 3,58683 11 1,359 58 INTERNATIONAL MIGRATION REVIEW TABLE 4 TOTAL AND GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA BY COUNTRY OF LAST PERMANENT RESIDENCE: SELECTED YEARS, 1968-1988 Country of1968 1976 1980 1988 Last Permanent Residence Indirect Total Indirect Total Indirect Total Indirect Total Austria7,548^8,12754374068 U.K.6,186 s 37,8893,852 s 21,2853,010" Hong Kong4,440s1 7,5934,561s110,6182,756s1 West Germany3,916s18,969 885s12,631278 U.S.A.3,750s120,4263,279s116,6302,025s1 France3,224 s 8,1851,094s13,102643 s Switzerland1,670^3,5293071,162137 Australia1,2253 3,7126453 1,372387 s Israel837 s 1,496647 s 1,155 840s1 Italy835 a • b 19,7711,031^4,328124 Lebanon6791,6791 ,300s17,219 289s1 Taiwan3015574631,165355s1 Pakistan2306275852,162185 Argentina225467 765s1 ' 11 1,419135 South Africa1631,0824151,595 430s1 China445,0763482070 Saudi Arabia4248283372 Uganda40625251 Singapore28130195415104 Brunei7284811687 U.A.E.00181840 Vietnam017391,546^ KampucheaO57^10583° Source; Public Use Tapes from Employment and Immigration Canada. ^Indic ates country made one of top ten largest totals for year. indicates at least 75% were refugees. Indicates at least 25% were refugees. The absence of South and Central American countries from Tables 3 and 4 is noteworthy. With the exception of 1976 when Chilean refugees went to Argentina before moving to Canada, no country in this region appears among the top ten. Whereas large numbers of migrants go from Central and South America to the United States, few continue north to Canada. Moreover, few emigrate to other Latin American countries and then continue on to Canada. A g^in , we attempt to account for this phenomenon in the model described below. PATTERNS OF INDIRECT MOVEMENT The number of potential paths an indirect migrant may follow to Canada is very large. Over 50,000 different combinations of country of birth and country 24053218 18,2501,924s1 9,158 6,3095,432 s 23,187 1,6433641,692 9,9341,309s1 6,520 1,901964 s 2,582 857125596 885230518 1,498740 s 1,426 1,74097857 1,4063053,121 8273782,176 8812111,235 44987428 1,3703281,668 4,9431562,766 81649 s 858 833150 290 440s1 1,140 336456 3 1,305 47685 s 913 25,643 518sl• c 6,140 3,270262°1,538 GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA 59 of last p ermanent residence may potentially be distinguished. Of course, when a country' s proximity to Canada, political climate, economic attractiveness, and other factors are taken into consideration, few of these paths are actually realized. In 1988, indirect migrants were born in 184 different countries, moved through 165 different countries of last permanent residence, and followed 2,174 different paths to Canada. Fewer countries served as intermediate countries than countries of birth for indirect migrants for every year during the period studied here. This observation is not surprising. It would be unlikely, for example, for an individual to immigrate to Canada using North Korea as an intermediate country, but it would not be surprising to find migrants who were born in North Korea and who moved indirectly to Canada. Certain countries appear repeatedly as places of birth, but not places of last permanent residence, including: Estonia, Latvia, Lithuania, Albania, North Korea, Republic of Guinea, Mozambique, Cape Verde, Mali Republic, Togo Republic, Yemen, Falkland Islands, St. Helena, and Benin People s Republic. The largest groups of indirect migrants falling into this category were from Mozambique and Burma in 1988, Estonia, Latvia, and Lithuania in 1973, North Korea in 1973, Tibet in 1972, and Tunisia in 1985. For several of these countries the political climate or its remoteness may be the motivating factor behind moving indirectly to Canada. We will test these hypotheses more rigorously below. In 1968 Austria ranked first among the countries of last permanent residence with 7,548 indirect migrants, of whom 6,617 were refugees from Czechoslovakia. Other large blocks of refugees on their way to Canada moved from Chile to Argentina in 1976 (512) and from Kampuchea to Vietnam in 1980 (327). These examples give evidence of the importance of initial short moves, but the most compelling example may be that of China and Hong Kong. Chinese emigrants to Hong Kong account for at least 80 percent (4,362 in 1988) of the indirect migrants from Hong Kong. In 1968, this portion was 93.1 percent. Other patterns of indirect movement are significant. Persons born in Turkey, Egypt, Israel, and Syria consistently migrated through Lebanon. For France, migrants were predominantly from Morocco, Portugal, and Algeria. In 1988, Lebanon was the country of birth of the most indirect migrants who moved through France. Along with migrants from Hong Kong, those born in China consistently moved to Canada through Taiwan. For those migrating through Israel, the top countries of birth were Poland, U.S.S.R., Romania, Morocco, and Iraq. The number born in the U.S.S.R. and using Israel as an intermediate country on the way to Canada has increased over the years with 64 in 1968, but 400 in 1988. The United Arab Emirates (U.A.E.) and Saudi Arabia appear in the top ten for the first time in 1988. For Saudi Arabia, this is largely due to Lebanon and Syria, and for United Arab Emirates to Lebanon, India, and Iraq. 60 INTERNATIONAL MIGRATION REVIEW Cer tain indirect migrants use England and the United States as the intermediate country before moving to Canada. For England, the importance of individuals from Commonwealth countries is readily apparent. Migrants from India often move through England. Despite the relative proximity of the United States to Canada, in 1976 only 52 persons born in Jamaica moved to Canada through the United States, whereas 228 moved through England. Other important countries sending migrants through England were Guyana, Tanzania, Kenya, and Uganda. Those who migrated through the United States before continuing on to Canada were largely European. In 1988, only 13 persons born in Mexico migrated from the United States to Canada. AGE, SKILLS AND SCHOOLING Greenwood and Trabka (1991) show that indirect migrants to the United States are both older and more hi ghly skilled than direct migrants. The same is true of Canadian indirect migrants compared to direct migrants. A number of possible explanations seem plausible for this observation. First, if the initial move from the country of birth is made at about the same age for indirect and direct migrants, the indirect migrants are aging in their first host country. At the same time, they acquire more education, experience, training, and skills. Second, indirect migrants have cleared the entry requirements of not one but two countries and were deemed "desirable" by both. In this sense, indirect migrants may be a select group. Third, costs may play a role in determining indirect migration. In many cases, the first move of the indirect migrants is over a relatively shor t distance compared to the second move. Migrating longer distances is more costly, both in terms of the acquisition of information and of the actual costs of making the move. Certain individuals may be unable to finance an initial long-distance move so they move over shorter distances and then work to accumulate the assets to finance the next move. Fourth, in many instances, the initial move may be "forced" in the sense that an individual wishes to escape some undesirable aspect of a country of birth, such as a political regime. After settling for a time in a nearby country, the individual considers more fully his/her options and moves again. These types of factors are taken into account in the model developed below. For the years 1980, 1984, and 1988, the average age of direct immigrants was 26.9, 29.6, and 27.5 years, respectively. The corresponding numbers for indirect migrants are 38.6,42.8, and 36.7 years. When we separate those immigrants who declare an occupation (which removes dependents and the retired), indirect migrants are still considerably older, with a difference in average age of about five years (31.6 versus 36.3 years in 1980, 32.3 versus 38.5 years in 1984, and 33.4 versus 38.4 years in 1988). When specific countries are considered, even greater disparities between the ages of the two types of migrants are revealed. For China, GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA 61 the average age of indirect immigrants in 1980 was 53.1 years compared to 31.1 years for direct migrants. Comparable figures for the U.S.S.R. are 40.8 and 29.9 years. For certain other countries the differences are more modest. Comparable respective figures for average age are 35.6 and 31.9 years for West Germany, 32.1 and 26.0 years for France, 28.8 and 23.2 years for Malaysia, 41.0 and 36.7 years for India, and 33.1 and 27.8 years for England. Occasionally, the average age of direct migrants exceeds that of the indirect. For example, in 1988 Frances direct migrants were on average 1.1 years older. Similar figures show that indirect migrants are more likely to be professionals, i.e ^ professional, technical, and kindred (PTK) and managers and administrators. For 1980, the percentage of all direct migrants who were PTKs was 8.3 percent versus 13.4 percent of all indirect migrants. For 1988, the respective figures are 10.0 percent versus 17.9 percent. Indirect migrants are also more likely to be managers or administrators (1.4% versus 2.7% in 1980, 2.6% versus 5.2% in 1988). In 1988, among those who declared an occupation, 42.6 percent of indirect migrants were PTKs, whereas for direct migrants this figure was only 29.7 percent. For specific countries in 1988, PTKs were somewhat more likely to come indirectly. For example, those falling in the PTK category made up 10.7 percent of indirect migrants born in India, but only 2.2 percent of direct migrants born there. Comparable figures are 19.6 percent and 14.3 percent for Malaysia, and 15.5 percent and 13.0 percent for England. For the U.S.S.R., despite indirect migrants being older, PTK migrants were more likely to move directly, with comparable figures of 9.5 percent and 20.9 percent in 1988. One important advantage of the Canadian microdata over those for the United States is the availability of information on schooling. As indicated in Table 5, for almost every year indirect immigrants had more years of schooling than direct immigrants. The only exceptions are 1984 and 1985, when direct immigrants had more schooling, and 1986, when each group had the same number of years. In 1988, indirect immigrants had an average of 10.1 years of schooling, whereas direct immigrants had 8.9 years. Differences in schooling between direct and indirect immigrants are frequendy somewhat more extreme for specific countries. For example, persons born in Israel who moved directly to Canada had on average 6.7 years of schooling, whereas those bom in Israel and moving indirectly had 13.1 years of schooling. Comparable respective figures for direct and indirect immigrants are 9.3 years and 10.4 years for England, 7.6 and 10.2 years for India, and 10.1 and 11.3 years for the U.S.S.R. In contrast, direct migrants from France (10.2 versus 8.8) and Germany (9.6 versus 9.1) had more schooling in 1988 than indirect immigrants bom in these countries. Whereas indirect migrants are often older, more skilled and better educated, the data reveal that during the mid-1980s the skill and schooling gaps between the two groups closed somewhat. From 1984 to 1986, an occupation was declared by a smaller percentage of indirect migrants compared to direct. Furthermore, the 62 INTERNATIONAL MIGRATION REVIEW TABLE 5 MEAN NUMBER OF YEARS OF SCHOOLING OF CANADIAN DIRECT AND INDIRECT IMMIGRANTS, 1968-1988 Year Direct Indirect Year Direct Indirect 1968 5.8 6.6 1979 7.9 9.0 1969 6.5 7.3 1980 7.9 9.3 1970 6.7 7.6 1981 8.5 10.1 1971 6.8 7.6 1982 8.9 10.3 1972 6.5 7.8 1983 8.9 9.2 1973 7.6 8.8 1984 9.0 8.7 1974 7.2 8.8 1985 9.2 8.5 1975 6.7 8.5 1986 9.3 9.3 1976 6.9 8.4 1987 9.2 10.0 1977 7.2 8.4 1988 8.9 10.1 19787 ^ 58 ^ 6 difference between the two groups (indirect percentage minus direct percentage declaring an occupation) for those with at least 13 years of schooling went from 10.2 percentage points in 1980, to 0.2 percentage points in 1985, and then up to 7.0 percentage points in 1988. For the percent who were professionals, the difference was 7.1 percentage points in 1980, 4.4 percentage points in 1985, and 9.6 percentage points in 1988. Moreover, as we saw above, direct immigrants had more schooling in 1984 and 1985 than indirect immigrants. One potential explanation for the convergence in skills and education between the two groups during the mid-1980s is that these were years of high unemployment in Canada. As shown in Table 1, the total number of immigrants fell significantly in 1983, remaining at a relatively low level until 1987. With fewer admittances and the family class having the hig hest priority, a relatively low-skilled base existed that was not subject to the greater restrictions. This resulted in a smaller percentage of both groups being professionals, but whereas the percentage of direct migrants who were professionals fell 4.0 percentage points in 1983 (14.2% to 10.2%), it fell 8.0 percentage points for indirect migrants (23.5% to 15.5%). LANGUAGE CAPABILITIES A second important advantage of Canadian data relative to those for the United States is that the former report Canadian language skills. The record for each immigrant indicates whether the person is capable in English, capable in French, capable in both English and French, or capable in neither. For 1980 and 1988, Table 6 reveals that indirect migrants were more likely to speak English and/or French. In 1980, 61.7 percent of indirect migrants could speak either English and/or French, whereas the corresponding figure for direct migrants was 46.1 percent. In 1988, the corresponding numbers are 64.2 percent for indirect GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA 63 TABLE 6 CANADIAN LANGUAGE ABILITIES OF DIRECT AND INDIRECT MIGRANTS: 1980,1984,1988 (PERCENTAGE  1980   1984   1988  Language Direct Indirect Direct Indirect Direct Indirect English 39.8 48.9 43.9 37.8 45.0 52.1 French 3.9 5.4 4.7 4.6 3.0 4.2 English/French 2.4 7.5 3.3 6.9 3.2 8.0 No Language  53.9 3^2 48J  507 4^8 35.8 immigrants and 51.2 percent for direct migrants. Again, as we saw with respect to skills and schooling, 1984 is an exception. In this year direct migrants were more likely to speak English or French than indirect migrants, but as in other years, indirect migrants were more likely to speak both languages. Clear differences exist between countries in reported language capabilities. Table 7 reports language abilities of direct and indirect immigrants for 1988 for all countries listed in Table 3 (except Canada). Among all 1988 immigrants born in the United Kingdom, 96.8 percent could speak capable English but not capable French. The corresponding percentage is 1.5 percent for France, 16.6 percent for Germany, 45.1 percent for Israel, and 77.3 percent for Malaysia. More revealing, however, is a comparison between direct and indirect migrants that shows that indirect migrants are less likely, in several cases, to achieve efficiency in the country of births native language. While 98.2 percent of direct migrants born in England were capable in only English, for those indirect migrants having England as their place of birth only 86.2 percent had this trait. Comparable numbers for two other countries have significandy greater differences - India (23.9% versus 80.5%) and the United States (93.5% versus 36.0%). These examples lead to the conclusion that direct migrants are often more likely to speak the native language of the country of birth than indirect migrants who leave and spend time in an intermediate country. However, for non-English and non-French speaking countries of birth, the evidence suggests that indirect migrants are more likely to speak English and/or French. For example, those migrants who speak either French and/or English capably made up 64.5 percent of these indirect migrants born in West Germany, but only 59.1 percent of those who moved directly could speak either language. The same type of pattern holds for Italy (40.9% versus 21.2%), Poland (74.8% versus 13.3%), Spain (71.4% versus 50.8%), and the U.S.S.R. (30.8% versus 15.3%). Presumably, this observation is due both to indirect migrants acquiring language skills in the intermediate country and to their having fewer dependents, who often at the time of immigration are capable in no language. 4 4 ln 1980, 1984, and 1988, the 17.4%, 15.1% and 21.7%, respectively, of indirect migrants were dependents. Comparable numbers for direct migrants were 35.6%, 26.9%, and 32.7%. 64 INTERNATIONAL MIGRATION REVIEW TABLE 7 CANADIAN LANGUAGE ABILITIES OF DIRECT AND INDIRECT MIGRANTS FROM SPECIFIC COUNTRIES OF BIRTH, 1988 ^ Direct Migrants Indirect Migrants of Birth English French Both Neither English French Both Neither China15.10.20.384.437.50.3 U.K.98.20.01.70.186.20.6 India23.90.10.175.980.50.1 West Germany 51.40.57.240.952.25.3 Yugoslavia25.90.72.271.256.00.0 U.S.S.R.13.21.01.084.827.91.4 Vietnam7.10.70.591.717.59.7 U.S.A.93.50.13.72.636.00.9 Lebanon17.013.524.644.939.112.1 Malaysia80.10.00.519.470.10.3 Iran29.71.62.366.449.610.8 Hong Kong60.40.20.339.052.70.0 Philippines77.00.00.222.786.10.5 Czechoslovakia 10.41.41.386.863.66.5 Italy15.32.53.478.821.711.6 Portugal6.34.15.983.740.015.0 Thailand28.90.00.770.55.00.5 Poland11.90.80.687.551.917.5 Kenya86.50.01.911.690.40.0 Tanzania84.90.01.014.188.40.0 Chile13.07.42.677.141.27.4 Kampuchea6.21.20.592.115.718.1 Tamaica99.80.00.20.097.50.0 These results suggest that indirect migration plays a useful role in the immigrant ' s assimilation in Canada. English language ability has repeatedly been shown to be a critical determinant of economic assimilation in the United States (Greenwood and McDowell, 1986). The presumption that English-language skills are critical to assimilation has caused Australia to adopt an immigrant policy that centers around instruction in the English language. MODELING INDIRECT MIGRATION Greenwood and Trabka (1991) discuss a number of possible reasons for indirect as opposed to direct migration. Barrett (1976) also provides such a 0.461.7 7.35.9 1.518.0 7.035.5 9.534.5 1.469.2 9.763.1 7.255.9 35.613.2 0.529.1 16.523.2 1.146.2 1.212.2 5.224.7 7.659.1 12.532.5 0.593.9 5.425.2 4.25.4 6.45.2 14.736.8 19.347.0 2.50.0 GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA 65 discussion. In this section we specify and estimate a bivariate logit model that distinguishes between geographically indirect and direct migrants to Canada. Because the model is estimated with microdata, we are able to take into consideration both personal characteristics of the migrants and country characteristics (such as political conditions, geographic location, and other factors) that could influence the decision to make an indirect move as opposed to a direct move. We believe that this is the first attempt to actually estimate a model of indirect international migration. The model specified below contains four vectors of variables: one to reflect (nonoccupational) personal characteristics, one to reflect the occupational characteristics of the migrant, one to reflect the immigration class of the migrant, and a fourth to reflect conditions in the country of birth. The general form of the model is: M . -" O +^Wh- + l^/y 5 . + ^ 6 ^. + ^ 5 e rQri+^ where M i = is equal to one if migrant /moved indirectly to Canada and is equal to zero if the move was direct; P^ = a vector of k variables that relate to perso nal characteristics of migrant z; S ji = a vector ofj variables that relate to the occupation of migrant z; C zi = a vector of z variables that reflect the immigration class of migrant z; Q^ = a vector ofr variables that reflect the political and economic conditions, as well as location of migrant fs country of birth; and C{ = the error term. Six variables are included in the personal characteristics vector and each is defined for the time of zs entry into Canada: AGE ^, the age of immigrant z; SCHLp the number of years of schooling obtained by z; ENGFR f equals one if the migrant can speak either English or French capably; MALE equals one if immigrant z is a male and zero if female; MARR^ equals one if immigrant z is married and zero otherwise; and DEPEND ^ equals one if migrant z is a dependent and zero otherwise. We saw previously that indirect migrants are on average older, and we provided several hypotheses to explain this observation. We expect AGE ? SCHLp ENGFRp and MARR f to have positive signs. While indirect migrants are older and are likely to have more children, a child born in the intermediate country to an indirect migrant will be a direct migrant to Canada. Moreover, indirect migrants are more likely to have adult children. Therefore the sign on DEPEND ^ is ambiguous. We specify no sign on MALE ^. 66 INTERNATIONAL MIGRATION REVIEW Four variables are included in the occupation vector S. The occupational dummy variables are: PROF/, equals one if individual z is in a PTK, managerial, or administrative occupation and otherwise zero; SEMISK/, equals one if the individual is in a semiskilled occupation (^., clerical, processing, or machining occupation) and zero otherwise; LAB / equals one if the individual is a laborer and zero otherwise; and RETIRED/ equals one if individual / is retired and zero otherwise. Because the expectation is that indirect migrants are older and more skilled, PROF/, SEMISK/, and RETIRED / should have positive coefficients, whereas the coefficient on LAB/ should be negative. The benchmark group is those individuals who declare no occupation (excluding retired and dependents). Two variables are included in the immigration class vector C; they are INDEP/, equals one if individual qualified under the independent class and zero otherwise; and FAMCL/, equals one if migrant ^ ' qualified under the family class and otherwise zero. The expectation is that indirect migrants are more likely to qualify under the independent class and less likely to qualify under the family class. Therefore, INDEP/ should have a positive coefficient and FAMCL/ should have a negative coefficient. The benchmark group is those migrants who qualified under the assisted relative class. The vector of country characteristics has five explanatory variables: MILES, miles to the nearest major Canadian city from the major city of migrant zs country of birth; POLIT, political conditions of migrant fs country of birth; COMWLTH, a dummy variable taking the value of one if the country of birth is a member of the British Commonwealth; SACAC, a dummy variable taking the value of one if the country is in South or Central America or the Caribbean, and RELINC, the country of birth's gross national product (GNP) per capita relative to Canada' s GNP per capita. Presumably, one of the reasons for migrating indirectly is that the country of birth is distant from Canada. Moreover, information about Canada may be more costly to acquire than information available about a nearby potential intermediate country. Hence, MILES should be positive, which would indicate that the greater the distance a person is born from Canada, the greater the likelihood of an indirect move to Canada. Greenwood and Trabka (1991) point out that indirect migrants have a tendency to be born in politically repressive countries. POLIT, which is drawn from Gastil (1986), is indexed from one to seven, where one represents the least politically repressed and seven the most. Thus, we expect POLIT to take a positive sign. Institutions may also affect indirect migration. One such institution is immigration policies of possible countries through which an indirect migrant may move, as well as Canada ' s own policies. For many years, migration from British Commonwealth countries to the United Kingdom was essentially unconstrained. Moreover, Canadian immigration policy specifically favored persons from Britain. Through the family ties that could result from such GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA 67 linkages, British Commonwealth countries could be indirectly favored in terms of more recent Canadian policy, even if no direct Commonwealth connection existed. Thus, we expect the Commonwealth dummy to take a positive sign. As noted previously, countries geographically below the United States do not typically serve either as countries of birth or countries of last permanent residence of indirect migrants to Canada. The United States provides a substantial alternative opportunity that lies more or less directly in the path between these countries and Canada. Hence, SACAC, which takes a value of one for South American, Central American, and Caribbean countries, should have a negative coefficient. The sign on RELINC is not obvious a priori. Studies of indirect migration within the United States indicate that better-educated, higher-income persons tend to move on more rapidly, whereas those who are less well educated and subject to hig her unemployment tend to return. If this were also true in the international migration context, the "intermediate " countries would be left with a pool of potentially more successful persons who would be at risk to make another move, such as to Canada. If the success of this pool is partly a function of the countries of birth having higher incomes, then higher-income countries of birth would send more migrants indirectly. However, countering this tendency is the idea that persons from higher-income countries of birth are better able to afford longer-distance direct moves to begin with. Consequently, we do not specify a sign on RELINC. EMPIRICAL RESULTS The model discussed above has been estimated for three years, 1980, 1984, and 1988. Migrants included in the estimation are all those who were bom in 65 countries for which appropriate data on the independent variables (country characteristics) were available. Additionally, all refugees were deleted from the data, as were those migrants bom in Canada. Refugees were deleted because they have the potential to cause spikes in the data due to their irregular admittance from certain source countries. Moreover, the movement of refugees is probably motivated for reasons different than other migrants, although they probably seek better employment opportunities and higher wages. To a larger extent than other Canadian immigrants, however, refugee admittances are discretionary on the part of Canadian authorities, who must decide which potential migrants qualify as refugees, how many to accept in a given year, and from where to accept them. It is precisely this discretion and the resulting irregular admittances of refugees from different countries that has the potential to overwhelm any effort to empirically estimate a model with them in it. As a practical matter, comparable data on source-country characteristics do not exist for important countries of birth of refugees. Migrants born in the 65 countries used in the estimation account for 53.4 percent, 34.5 percent, and 39.8 percent of indirect, nonrefugee migrants for 68 INTERNATIONAL MIGRATION REVIEW TABLE 8 VARIABLE MEANS AND STANDARD DEVIATIONS 1980 1984 1988 Standard Standard Standard Mean Deviation Mean Deviation Mean Deviation M 0.1060 0.3078 0.0773 0.2670 0.1017 0.3022 AGE 30.34 18.91 32.50 19.00 29.68 18.38 DEPEND 0.2942 0.4557 0.2126 0.4092 0.3017 0.4590 MALE 0.4767 0.4995 0.4199 0.4936 0.4628 0.4986 MARR 0.4933 0.5000 0.4924 0.4999 0.4708 0.4991 RETIRED 0.0139 0.1172 0.0225 0.1482 0.0155 0.1235 PROF 0.1319 0.3384 0.1138 0.3176 0.1442 0.3513 SEMISK 0.1849 0.3882 0.1068 0.3088 0.1512 0.3583 LAB 0.0503 0.2185 0.1032 0.3042 0.0574 0.2326 SCHL 8.9464 5.4530 9.6170 5.4159 9.3039 5.6233 ENGFR 0.7425 0.4373 0.7309 0.4435 0.6774 0.4675 SACAC 0.1235 0.3290 0.1782 0.3827 0.1668 0.3728 MILES 4,301 2,411 4,379 2,651 5,002 2,543 POLIT 2.2450 1.6886 2.3801 1.7580 2.2368 1.3877 COMWLTH 0.5011 0.5000 0.4369 0.4960 0.4892 0.4999 RELINC 0.5349 0.4495 0.3991 0.4197 0.3444 0.4198 INDEP 0.4235 0.4941 0.3012 0.4589 0.4575 0.4982 FAMCL0.4994 0.50000.6601 0.47370.4481 0.4973 the years 1980, 1984, and 1988, respectively. The corresponding shares for direct migrants are 71.6 percent, 58.1 percent, and 56.6 percent. The total numbers of observations used in estimating the three years are 72,853 in 1988, 40,119 in 1984, and 70,698 in 1980. Table 8 provides means and standard deviations for each variable included in the model and for each year for which the model has been estimated. Table 9 repor ts the coefficient estimates and the significance levels for the three years. Most explanatory variables take the expected sign and are highly significant. Seven variables are significant at the one one-hundredth of a percent for all three years (AGE, DEPEND, MALE, MARR, PROF, ENGFR, MILES), whercas three variables take the expected sign and arc significant at least at the ten percent level for all three years (RETIRED, SEMISK, LAB). Moreover, four variables arc significant for two of the three years (POLIT, RELINC, INDEI? FAMCL). Considering these strong empirical results, we can conclude that ceterus paribns indirect migrants, relative to direct migrants, arc more likely to be: 1) older, 2) dependents, 3) female, 4) married, 5) retired, 6) in a professional or semiskilled occupation, 7) capable in English and/or French, 8) born in countries farther away f rom Canada, 9) born in politically repressive countries, 10) born in countries with hi g her per capita GNPs, and 11) more likely to qualify under the independent class. Direct migrants, however, GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA 69 TABLE 9 BlVARIATE LOGIT ESTIMATES OF MODEL OF CANADIAN INDIRECT IMMIGRATION:  1980, 1984 AND 1988  Independent Variable198019841988 INTCPT -5.4657B -5.5068 a -5.4359 a AGE 0.0213'1 0.0206a 0.0248a DEPEND 0.2529B 0.4416s 0.4933a MALE -0.1694B -0.194l a -0.1105- 1 MARR 0.2788s 0.3225-1 0.4399s RETIRED 0.1769C 0.6061s 0.l474c PROF 0.3240s 0.6984s 0.2163s SEMISK 0.1196a 0.5741s 0.1153a LAB -0.1l40 c -0.7961B -0.6370 s SCHL 0.0152 s -0.0052 -0.01 l6 a ENGFR 1.0159s 0.8379a 0.8239B CASAC 0.7650 s -0.1016 -0.0302 MILES 0.0004s 0.0002s 0.0002s POLIT 0.0322 b 0.0179 0.1105 s COMWLTH -0.1171 s 0.0817° 0.070l b RELINC 0.2895 s -0.0988 0.4950 s INDEP 0.0483 0.7961s 0.3879a FAMCLA-1.0962 a 0.0877-0.7625 a  Statistically significant at better than 1 percent. Statistically significant between 1 and 5 percent. Statistically significant between 5 and 10 percent. arc significantly more likely to be laborers and qualify under the family class. Generally, the conclusions with respect to years of schooling (SCHL), the Commonwealth dummy (COMWITH), and the Central and South American and Caribbean dummy (SACAC) arc not as expected. That is, the empirical results do not support the hypotheses that indirect migrants have more years of schooling (except in 1980) or arc bom in Commonwealth (except in 1988) or Central and South American countries (except in 1980). The positive and significant coefficient on DEPEND in 1984 and 1988 indicates that because indirect migrants arc older, they have more children, and this outweighs those other factors discussed. CONCLUSIONS Indirect geographic migration has received little attention, probably in large par t because appropriate data to study the phenomenon have not been available. However, Greenwood and Trabka (1991), using microdata from the U.S. Immigration and Naturalization Service, provide a fairly detailed description of geographically indirect migration to the United States. We have used roughly comparable microdata for Canada in this study to provide a parallel treatment of 70 INTERNATIONAL MIGRATION REVIEW indirect migration to Canada. Together, the two studies constitute a reasonably comprehensive examination of geographically indirect migration to North America, because no other country in North America is a significant nation for immigration. In two important respects, the present study goes beyond Greenwood and Trabka. First, the Canadian microdata records contain important information not contained in the U.S. records. For the purposes of this study, two data items particularly stand out, namely, schooling and language ability. We show here that geographically indirect migrants generally differ from geographically direct migrants in terms of each measure. Indirect migrants to Canada are better educated and more likely to speak English or French than direct migrants. However, the educational gap between the two groups narrows when Canada restricts admittances. The bivariate logit model estimated in this study indicates that, other factors held constant, geographically indirect compared to direct Canadian immigrants are more likely to be older, female, married, a dependent, retired, in a professional or semiskilled occupation, capable in English and/or French, born in a country more distant from Canada, born in a politically repressive country born in a higher income country, and more likely to qualify for admittance under the independent class. Direct immigrants are more likely to be laborers and to qualify under the family class. Future work on modeling indirect migration is needed. Because at least three countries are involved in such migration, data requirements and a comparison of benefits and costs of choosing between direct and indirect migration are complicated. Ideally, such a model should consider the relationship and attributes of the country of birth and the country of last permanent residence, along with the characteristics of the destination country. Unfortunately, neither the Canadian nor U.S. microdata has information on the duration of residence in the country of last permanent residence or the year of depar ture from the country of birth. Lack of data on the year of departure from the country of birth prevents investigators from identifying the age at departure from that country, which would be valuable information for many purposes. Another shortcoming which appears to be characteristic of all such international migration data is that multiple moves that may have occurred between the country of birth and country of last permanent residence are not recorded. REFERENCES Barrett, R 1976 "A Schema for Indirect International Migration," International Migration Review, 10(1):3-1L GEOGRAPHICALLY INDIRECT IMMIGRATION TO CANADA 71 Boyd,M. 1976 "Immigration Policies and Trends: A Comparison of Canada and the United States," Demography, 13(1):83-104. Da Vanzo, J. S. 1978 "Repeat Migration in the U.S.: Who Moves Back and Who Moves On?" The RAND paper series P-5961. Santa Monica, CA: The RAND Corporation. Davies, J. B. 1992 "Tax Incidence: Annual and Lifetime Perspectives in the United States and Canada." In Canada-U.S. Tax Comparisons. Ed. J. B. Shoven andJ. Whalley. Chicago: University of Chicago Press. Pp. 151-188. Gastil, R. D. 1986 Freedom in the World New York: Greenwood Press. Gould.J.D. 1979 "European Inter-Continental Emigration 1815-1914: Patterns and Causes," Journal of European Economic History, 8(3):593-640. Greenwood, M. J. andJ. M. McDowell 1991 "Differential Economic Opportunity, Transferability of Skills, and Immigration to the United States and Canada," Review of Economics and Statistics, 73(4): 612-623. 1986 "The Factor Market Consequences of U.S. Immigration," Journal of Economic Literature , 24(4): 1738-1772. Greenwood, M. J. and E. Trabka 1991 "Temporal and Spatial Patterns of Geographically Indirect Immigration to the United States," International Miration Review, 25(1):93-112. Kau, J. B. and C. F. Sirmans 1976 "New, Repeat, and Return Migration: A Study of Migrant Types," Southern Economic Journal, 43(2): 1144-1148. Michalowski, M. 1991 "Foreign-born Canadian Emigrants and Their Characteristics (1981-1986)," International Migration Review, 25(1):28-59. Ross, J. A., ed. 1982 International Encyclopedia of Population. 2 volumes. New York: The Free Press. Shryock, H. S., J. S. Siegel, and Associates 1976 The Methods and Materials of Demography. New York: Academic Press. United Nations 1979 Trends and Characteristics of International Migration since 1950. Demographic Studies No. 64. New York: U.N. Department of Economic and Social Affairs. U.S. Bureau of the Census 1990 "Migration between the United States and Canada. " Current Population Reports, Series P-23, No. 161. Washington, DC: U.S. Government Printing Office, Vanderkamp, J. 1972 "Return Migration: Its Significance and Behavior," Western Economic Journal, 10(4):460-465. Factors Determining Migrant Remittances: The Case of Greece This article examines the flow of migrant remittances from Germany, Belgium, and Sweden to Greece. The statistical analysis uses the following determining factors: migrant ' s income, family income, the rate of interest, the rate of inflation, the exchange rate, the rate of unemployment, and the number of migrants. The results of this study reaffirm some earlier results of other studies, but they differ in some respects. An attempt is also made to detect cohort effects on the propensity to remit, and the evidence is that such effects do exist. In the labor-sending countries, migration has been seen as resulting in a mixture of benefits and costs. The costs include the loss of labor supply in which substantial amounts of human capital have been invested, the distortion of the age structure of the population (in the instance of substantial volumes of emigration), depopulation of rural areas, etc. The benefits include the reduction of social tensions from unemployment and/or underemployment, the acquisition of skills in the foreign countries by the returning migrants, and the money transfers from migrants to their families back home, /. e., remittances, etc. These were the issues debated among Greek policymakers, politicians, analysts, and academics during the period of late 1950s and particularly in the 1960s when Greek labor migration to Western Europe took on dramatic dimensions. In more recent literature, the emphasis was placed on remittances, the magnitude and use of which is central to the benefits from migration (Martin, 1983). With respect to the magnitude, the available evidence from various countries is that remittances are important for the receiving countries and, in some cases, they represent a very impressive percentage of the gross domestic product (GDP) - for example in 1982, 5 percent in Bangladesh and 8.4 percent in Pakistan (Russell, 1986; Keely and Tran, 1989; Appleyard, 1989; Choucri, 1986). As to the effects of remittances on the receiving economy, apar t from being substantial income for consumption and small investment for the recipients, there seems to be no consensus (Russell, 1986,1992; Chandarvarkar, 1980; Serageldin etaL, 1981; Ecevit and Zachariah, 1978; Keely and Tran, 1989). Another branch of the literature deals with the theory and factors that determine the magnitude of remittances. What is surprising here is the small number of published studies. Although remittances are directly related to migration, about which the literature is immense, theoretical and empirical ©1997 by the Center for Migration Studies of New York. All rights reserved. 0197-9183/97/3101.0117 Theodore P. Lianos Athens University of Economics and Business, Athens, Greece MIGRANT REMITTANCES: THE CASE OF GREECE 73 studies on the determinants and consequences of remittances are few. One mig ht think that it is exactly the direct connection of remittances with migration that makes remittances less interesting to investigate. However, the pursuance of this literature will soon find that the subject is much more complex than it seems to be, and the multip licity of factors affecting remittances is much wider. In this study we review the literature regarding the theory and the empirical evidence regarding migrants ' remittance behavior, and we examine the flow and determinants of remittances from Greek migrants for the period 1961 to 1991. The main body of data is for remittances to Greece from Germany, but for some years data are available for remittances from Belgium and Sweden as well. The objective of this study is to test the significance of certain factors in terms of their effects on remittances to Greece. These factors are the migrant ' s income, the migrant ' s family income, the rate of unemployment, the rate of interest, the exchange rate, and the rate of inflation. The study attempts to see if these factors have had any effect on the volume of remittances. It also attempts to see if there are any structural changes during this period that affect migrants ' propensity to remit. The period of migration extends over 30 years, and it is interesting to examine whether the remittance behavior of migrants has changed over the years. REVIEW OF THE LITERATURE Theory There is only one paper, known to this author, in the English-language literature that attempts to give answers to the basic question regarding the motivation to remit (Lucas and Stark, 1985). The central idea in the theory proposed by these authors is that the remittances can be seen as an implicit contractual arrangement between migrant and home, which is mutually beneficial and intertemporal. Two of the elements of this implicit agreement are investment and risk. Investment involves the creation of human capital through the education of the migrant, with the costs (direct cost and subsistence support while not earning) borne by the migrant ' s immediate family. Remittances can thus be viewed as the return of this investment to the family. The element of risk appears in understanding the act of migration as diversifying behavior in the presence of risk. The existence of many risks in the environment of the family (for example, in a rural-urban setting, crop failures, livestock diseases, price falls) may force the family to spread its risks by allocating a member of the family, the migrant, to (urban) migration. The family supports the migrant in covering search costs, moving expenses, and periods of unemployment in the urban area, as well as remittances flow to the family at times of need, as when crops fail. 74 INTERNATIONAL MIGRATION REVIEW These informal arrangements are voluntary and self-enforcing. Mutual altruism and self-interest are the two forces working against delinquency Altruism refers to the mutual altruism among the members of the family Self-interest refers not only to the benefits from the implicit arrangement itself but to other benefits, such as aspiration to inherit part of the family wealth, convenience in investing at home, and the intention to return home. The theory we have briefly summarized leads to certain testable hypotheses regarding the factors that affect the size of remittance flows. In fact, Lucas and Stark test some of these hypotheses on the basis of statistical data from Botswana. Before we review these and other empirical findings, a discussion of the theory sketched above is in order. There are several difficulties with this theory that should be pointed out. First, the notion of altruism used by Lucas and Stark is the same, conceptually and formally, as that ofBecker (1981). According to this notion, an individual transfers part of his income to someone else because in this way he increases his own utility In the case of migration, the migrant maximizes his own utility by remitting part of his income to the members of his family back home. It is questionable if this behavior can be considered as altruistic. It is, rather, moral egoism even though others may benefit from it (Nowell-Smith, 1959; Norman, 1983). Second, it is difficult to accept that when a child goes to school and invests in human capital, he or she is consciously entering into an informal agreement voluntarily, i.e., that the child could refuse the agreement or modify the terms. It is also difficult to conceive of a father or mother who would refuse to send to school the child that mi ght refuse to agree to migrate at some future time. Third, the informal contractual arrangement hypothesis appears as a post hoc rationalization of the observable events. Of course, many aspects of economic theory can be taken as simple rationalizations rather than explanations, but it is hard to shake the feeling that it is more so in this case. In addition, even if we can speak of contractual arrangements, informal or not, in a genuine sense, we cannot ignore the social environment in which migration takes place. It is exactly the neglect of social environment and institutions that makes the informal contractual arrangement hypothesis questionable or, at best, incomplete. We have expressed doubts about the explanations of remittances as a result of intertemporal contractual arrangement. Now, we can discuss briefly remittances as a result of a different decisionmaking framework. Migration can be seen as an act of a rational individual in pursuit of his economic welfare. At the moment of migration the individual has plans regarding his main as well as his secondary objectives, i.e., to stay in the country of destination for a given period of time, or to accumulate a certain amount of money, or to stay in the country of destination for as long as possible and bring his family to his place of residence, or to move to a different country after some time, etc. Of course, as the circumstances change the migrant can change his plans. The MIGRANT REMITTANCES: THE CASE OF GREECE 75 initial plan may never materialize, and the migrant may find himself in a much better or much worse situation than he expected. Therefore, in the act of migration there is an inherent dynamic element that has to do with insufficient information, changing circumstances, unpredictable psychic costs, etc. Remittances are strictly connected to the circumstances of the migrant, economic and otherwise. The plans of the migrant and his decisions will determine the flow of remittances. Thus, whatever uncertainty exists in the plans of the migrant and whatever changes in his decisions, all might be reflected in the flow of remittances. The multiplicity of elements that affect the life of a migrant is so great that it makes it difficult to speak of a representative migrant and to make useful statements with regard to his objectives. A migrant is not (simply) a consumer living in a stable and known environment in which he attempts to maximize his utility. Neither is he simply an investor who reinvests his profits in an attempt to maximize his return. Of course, a migrant does not behave randomly, and various objectives can be assumed as descriptive of his motivation. For example, three reasonable but different objectives may be assumed. First, a migrants objective is to remain permanently in the receiving country. Second, a migrants objective is to minimize, under certain conditions, the time spent in the receiving country. Third, a migrant s objective may be to maximize the amount of his wealth within a certain time horizon. Each of these objectives implies a different saving behavior and different flow of remittances, both in terms of size and in terms of timing. In addition, the effect of a factor on remittances may differ depending on the assumption we make about a migrants objective. Of course, one could group migrants in certain categories on the basis of various criteria and then assume objective functions for each group. There does not seem to be much usefulness in such an exercise, except perhaps in identifying factors that would lead to the same type of migrant response within the different groups. Instead, it is preferable to use a common sense approach to model specification of remittances on the assumption that a migrant is a rational economic agent in pursuit of happiness under constraints. In understanding migrants ' behavior with respect to remittances, one basic element must be taken into account. This is the existence of family members (parents, spouse, children) remaining back home. How is the existence of family related to remittances? One answer could simply be love. This is certainly true but from the point of view of theory is not adequate because it does not yield testable hypotheses. We have already discussed the view of mutually beneficial informal arrangements self-enforced by altruism and interest and expressed the opinion that this is a rationalization of mtrafamily relationships. We believe that the motivation to remit part of ones income is based on commitment and loyalty to ones family. The family, as a social institution, creates within itself roles and allocates responsibilities. Within this framework of responsibilities and duties, a migrant feels committed to the welfare of his family and remits part of his income. From this 76 INTERNATIONAL MIGRATION REVIEW consideration one cannot immediately derive testable hypotheses. For this purpose, a formalization of the procedures according to which the family makes decisions is required. There are various alternatives that can be considered. One is to assume that the head of the family makes all the relevant decisions according to the traditionally established rules and to his own judgment. In this case, remittances are determined culturally and on the basis of an economic calculus that serves the interest of the family as seen by the head. Another alternative is to assume that the decisionmaking in the family follows an explicit or implicit negotiation process on the basis of responsibilities and demands from all members of the family. In this case, remittances are determined by the short-run as well as the long-run plans of the family regarding the p lace of residence, consumption versus investments, present consumption versus future consumption, allocation of the financial resources to the members of the family, etc. However, the purpose of this study is not to develop a model of remittances on the basis of microdata, but rather to investigate remittances of Greek migrants from existing time series data. Evidence There are two types of studies that investigate the factors affecting the flow of remittances: studies based on individual data obtained from surveys (i.e., cross-sectional data) and studies based on time series macro data. Of the many survey-type studies (for a review, .^Rempel and Lodbell, 1978; Russell, 1986), only four report regression results and tests of hypotheses of possible factors that may determine the flow of remittances. Two of these studies refer to Kenya (Rempel, Harris and Todaro, presented in Rempel and Lodbell, 1978; Johnson and Whitelaw, 1974), one to Pakistan (Mohammad, Butcher and Gotsh, 1973), and one to Botswana (Lucas and Stark, 1985). A summary of the results of these studies is presented in Table 1. In all four studies, the level of income of the migrant has a positive effect on the level of remittances, the presence of a wife in the host country has a negative effect, and the number of children back home has a positive effect. For the other variables the results are mixed, but it is important to notice that when they are significant they have the same sign. No such uniformity of results seems to be evident in the studies using time series macro data. The three studies that examine the flow of remittances to Greece - Glytsos (1988), Katseli and Glytsos (1989), and Swamy (1981) - agree as to the effect of some variables but disagree as to the effect of others. Thus in all three studies, per capita income of the migrants has a positive and significant effect on remittances and so does the rate of interest in Greece but not in all model specifications. Other variables — for example per capita income in Greece, the rate of inflation, and the exchange rate - do not seem to affect MIGRANT REMITTANCES: THE CASE OF GREECE 77 TABLE 1 SUMMARY OF RESEARCH FINDINGS FROM CROSS-SECTIONAL STUDIES Mohammad, Johnson and Butcher and Lucas and AuthorsRempel, Harris, Tbdaro WhitelowGotshStark Independent Dependent VariableVariableRWWRK Income of Migrant + + + + + Home Income n.a. n.a. n.a. nonsignif. nonsignif. - nonsienif. + + n.a Education +++n.a. Married (Head of Household)nonsignif.nonsignif.nonsignif.+ Wife with Migrant----Number of Children Back Homen.a.n.a.++ Duration of Migration-nonsignif.-n.a Possesion of Prooertvnonsienif.nonsienif.+n.a n.a. Notes: + indicates positive and significant coefficient. - indicates negative and significant coefficient. nonsignif. means statistically nonsignificant coefficient. n.a. means observations not available. R = Remittances. Y = Income. the flow of remittances in a uniform way in these studies and in the various regression estimations. In contrast, in a recent study, ^ihba (1991) reports significant negative effects on remittances to Egypt of exchange rate (official and parallel market) differentials and positive effects of devaluation. He also discusses the positive effects of domestic interest rates and the availability of financial intermediation in attracting remittances to the country of origin through the official channels. In a study of migrant remittances using data from Algeria, Morocco, Portugal, Tunisia, Turkey, and Yugoslavia ofElbadawi and Rocha (1992) found that the level of income in the host country, the exchange rate (defined as the percent difference between the black market exchange rate and the official exchange rate), the rate of domestic inflation, and the length of stay abroad are significant factors in explaining variations in remittances. The level of income at the country of origin and the interest rate differential did not produce significant results. On the basis of the above results, one may conclude that: 1) the level of income in the country of destination, the exchange rate, the length of stay, and the devaluation of the sending country ' s currency are significant factors in explaining the variation in migrant remittances; 2) the rate of inflation and the rate of interest do not seem to give consistent results among studies; and 3) the level of income in the country of origin does not seem to have a significant effect on migrants remittances. It should be observed that the number of studies examined above is small and some results are mixed. Thus the existing studies leave a number of open questions regarding the effects of the factors discussed. Also, none of the above studies has used the rate of unemployment in the host country as an explanatory variable. This is probably due to the assumption that the effect of unemployment is reflected in the level of income of migrants that lose their jobs 78 INTERNATIONAL MIGRATION REVIEW as unemployment rises. Even if this is true, the rate of employment should be included as an additional explanatory variable because, as the rate of unemployment increases, migrants may become uncertain about their future incomes and thus may reduce remittances. The present study contributes some evidence on the above questions from the Greek experience. MIGRANT REMITTANCES TO GREECE Specification of the Model Migration from Greece to Western Europe and mainly to (West) Germany began in the late 1950s and took on dramatic dimensions in the 1960s. In about twelve years, more than 1 million Greeks (10% of the Greek population) left Greece for Germany as the main destination. Although many returned after a short stay, the migrant Greek population in Germany in the peak year (1975) was 406,000. Between that year and 1991, it fluctuated around 300,000. The flow of migrant remittances from Germany in its peak year, 1982, was US$382 million and represented more than one percent ofGNP and about four percent of imports. For the period 1961-1991, the average annual flow of remittances from Germany was about US$204 million. The corresponding magnitudes for Belgium and Sweden are much smaller. In Belgium the annual average migrant population was about 20,000 with an average annual flow of remittances of US$18 million, whereas the respective figures for Sweden are US$9,600 and US$13 million. Flow of remittances (the dependent variable) in every time period is hypothesized to be a function of the following factors (independent variables): migrants income in the host country, family income in the country of origin, the rate of interest in the countries of ori g in and destination, the rate of inflation, the exchange rate, the rate of unemployment, and the number of migrants. A brief explanation of the expected effect of each independent variable on the flow of remittances follows. Migrant Income. It seems obvious that the effect of the migrant ' s income on the amount of remittances should be positive. That is, higher incomes in the host country should generate higher remittances. Family Income. The level of family income at the country of ori g in may affect the level of remittances in two opposite ways. A relatively high family income may necessitate large amounts of remittances so that the family can sustain a relatively high level of consumption. On the other hand, a hi gh family income may mean that larger amounts of the migrant ' s income can be used for other objectives (^., accumulation abroad) instead of being remitted. Also the place of residence of the family may be important as this may mean, for example, MIGRANT REMITTANCES: THE CASE OF GREECE 79 different consumption patterns (e.g., rural versus urban life). Thus the direction of the effect of family income on the level of remittances is uncertain. The Rate of Interest The rates of interest prevailing at origin and destination may affect the amount of remittances depending on the relation (differential) between the two interest rates. A relatively hig h interest rate in the country of origin may increase the amount of remittances because it is preferable, ceteris paribus, to shift savings to that country. However there is an additional effect having to do with absolute rather than relative interest rates. Higher interest rates may increase the propensity to save at each level of income and thus the quantity of funds that can be remitted. Thus, a hi g her interest rate in the country of ori gin provides both a motive for saving more and a motive for remitting more. The Rate of Inflation. Inflationary pressures in the country of origin reduce real income and thus may make necessary increased remittances. On the other hand, inflation increases the risk of reducing the real purchasing power of remitted funds and thus may postpone remittances until safe placements of funds are found. It appears that the effect of inflation would depend par tly on the nature of remittances. Part of the effect of inflation would appear through the exchange rate to the extent that the exchange rate is determined by relative price levels in the two countries. Exchange Rate. The exchange rate affects the quantity of goods and services that can be bought with a certain amount of remittances, and the effect can be in either direction. For example, depreciation of the drachma means, ceteris paribus, that less foreign currency and thus a lesser volume of remittances is necessary for the same quantity of goods. On the other hand, purchases in drachmas are now to be preferred, and thus remittances (to be changed into drachmas) may increase. The exchange rate can change gradually as a result of internal and external economic forces, or it can change at a point distant in time by government decisions to devaluate the national currency. Information about gradual changes is available to individuals and may affect remittances positively or negatively, as indicated above. A devaluation is likely to have only positive effects on remittances because purchases in the national currency are preferable now and for some time before the level of prices begins to increase as a result of the devaluation. Recent studies (Elbadawi and Rocha, 1992; Wahba, 1991) have shown that parallel or black market exchange rates are very impor tant factors affecting remittances, but there is no evidence of extensive illegal or other undocumented foreign exchange transactions in the Greek market. The Rate of Unemployment. The rate of unemployment in the country of destination may have negative effects on the amount of remittances for two reasons. First, some migrants may become, par tly or wholly, unemployed and their incomes may be reduced. Second, increased unemployment may create 80 INTERNATIONAL MIGRATION REVIEW uncertainties for the immediate future and thus reduce remittances at least temporarily. The Number of Migrants. Naturally, the number of migrants in a country affects directly the volume of remittances from that country. Of course, one may choose to examine remittances per migrant and thus eliminate the number of migrants as an explanatory variable. Statistical Data and Regression Results The statistical data used for the estimated regressions come from diverse sources; all are published except for remittances. All monetary values are expressed in current values of U.S. dollars. Remittances (REM) were collected from unpublished data of the Bank of Greece. Migrant ' s income (Y) in the country of destination (Germany, Belgium, Sweden) is measured by per capita income in that country, namely the gross national product divided by the population. Family income (Y g ) in Greece is measured in the same way. The source of these data are various issues o f International Financial Statistics (IMF). Instead of per capita income, the industrial hourly wage rate (W for the country of destination and W g for Greece) has been used in some regressions. These data are obtained from the Yearbook of Labour Statistics (\LO). From this source, the rate of unemployment (UN) has also been obtained. The rate of inflation (INF) in Greece is measured by the GDP deflator rather than the consumer price index, because remittances are not used solely for consumption. The data come from the Monthl y Statistical Bulletin of the Bank of Greece. The exchange rate (EX), measured in drachmas per unit of foreign currency, is also obtained from this source. The rate of interest (R for the country of destination; R g for Greece) is measured by the discount rate for Germany and by the deposit rate for Bel gium and Sweden. In Greece, a different interest rate was applicable, for a period of time, for deposits in foreign exchange. For that period we have used this rate of interest and for the other years the deposit rate, both in real terms. These data are available in International Financial Statistics (IMF). The numbers of Greek migrants (MP) are given in Trends in International Migration for 1961-1991 for Germany, 1981-1991 for Belgium, and 1980-1991 for Sweden. The results for various regression equations for Germany are presented in Table 2. The t-values are given under each coefficient within parenthesis without sign and are compared to the critical values of the t-statistic at 5 percent level of significance. From these results, it appears that the level of income in Germany (migrants income), measured either by the per capita income or by the wage rate, in all cases has a positive and significant effect on remittances. MIGRANT REMITTANCES: THE CASE OF GREECE 81 TABLE 2 REGRESSION EQUATIONS FOR REMITTANCES FROM GERMANY (1961-1991) Independent Regression  Elasticity Variable1234Regression 1 Regression 3 Constant -196,245 -191,050 -210,384 -210,942 (3.66) (3.44) (3.84) (3.82) Y 29.9 27.6 (6.87) (5.93) 1.40 W 63,955 65,630 (4.00) (4.33) 1.61 W^ -47,836 -43,289 (0.97) (0.88) MP 596 590 (3.94) (3.57) 0.89 MP(-l) b 654464 (4.61)(2.68) JNF 24,78810,48232,102 14,872 (2.10)(1.81)(2.49) (1.85) 1.57 EX -6,995-5,814-6,600 -6,064 (5.60)(4.78)(5.11) (5.37) -1.09 UN -12,685-17,370 (1.24)(1.80) R -14,5491,042-9,448 382 (1.55)(0.20)(1.01) (0.07) R^ 26,37013,64434,956 17,595 (2.12)(1.78)(2.58) (1.83) 0.20 R- 2 0.890.880.89 0.89 D-W 2.192.042.16 1.83 0.70 o^s Note: Diagnostic tests for auto regressive conditional heteroskedasticity show no evidence ofheteroskedasticity. White heteroskedasticity tests show no evidence ofheteroskedasticity except for Regression 2. Subscript g Indicates that the variable refers to Greece. MP lagged one period. The level of income in the country of origin (family income) is statistically insignificant in all cases and therefore those regressions are not reported in Table 2. The exchange rate has a negative effect because of continuing devaluation of the Greek drachma in relation to the German mark, as a result of which migrants tended to postpone remittances as long as possible. The rate of inflation in Greece is significant and has a positive sign in two regressions, but it is insignificant in the other two. The rate of unemployment in Germany has a negative coefficient but is statistically insignificant except in one case in which it is almost significant (equation 3). The rates of interest in Germany and in Greece have coefficients with opposite signs, negative for Germany and positive for Greece, and with different statistical significance. The rate of interest in Greece shows a high value of students t for two regressions whereas for Germany it is insignificant. Finally, the number of migrants in Germany has a positive and very significant coefficient in all cases. 82 INTERNATIONAL MIGRATION REVIEW In general, the regression results reported in Table 2 support the findings summarized in this study for time series data. Of the variables discussed earlier, migrant income in the country of destination, the exchange rate, the rate of inflation, the rate of interest in the country of origin, and the number of migrants seem to have statistically significant effects on the volume of remittances. The level of income in the country of origin and the rate of interest in the country of destination seem to have no effect on the amount of remittances. The evidence about the effect of the rate of unemployment in the country of destination is not conclusive. Experimentation with time lags of some variable did not produce different results. Table 2 also contains the elasticity of remittances with respect to the significant variables estimated at mean values. The elasticity of remittances with respect to income, measured by per capita income or by the wage rate, is greater than one (1.40 and 1.61, respectively), much higher than those found, e.g., by Elbadawi and Rocha (1992); it indicates a hi g h propensity to save and remit. The elasticity with respect to the stock of migrants is 0.89 (and 0.70 for the lagged value), and it falls within the range of values found by Elbadawi and Rocha (1992) and Swamy (1981). 1 The elasticity with respect to inflation is positive and greater than unity, but it varies considerably among regressions as is evident from the sizes of the coefficient. The elasticity with respect to the exchange rate is negative and equal to unity. This is an interesting result because it indicates that Greek migrants adjust their remittances to exchange rate changes so that the same value in terms of drachmas is sent back home. In other words, the changing value of the drachma does not make purchasing in Greece more or less attractive. The elasticity with respect to unemployment and the Greek rate of interest are rather small in absolute values. The regression results for Belgium and Sweden together are presented in Table 3. Here, the statistically significant variables are the level of income in the country of destination (measured by the industrial wage), the number of migrants and the level of unemployment, all with the expected signs. The rate of inflation in Greece is very close to significance and with the proper sign. However, interest rates seem to have no effect on remittances. The exchange rate (not reported in Table 3) was also statistically insignificant. A dummy variable (not reported) used to capture host country effects was statistically insignificant. The differences between the results of the two tables are not easy to explain. However, some suggestions can be offered. With respect to the effect of unemployment rates, it is reasonable to suggest that migrants in Germany felt more secure than in Belgium and Sweden, 1 Swamy argues, and Elbadawi and Rocha repeat, that the unitary elasticity with respect to the stock of migrants is theoretically attractive because a failure to uphold the proportionality between remittances and number of migrants suggests an underspecified regression equation. It should be pointed out that this is so only if all migrants, new and old, have the same propensity to remit. MIGRANT REMITTANCES: THE CASE OF GREECE 83 not only because of their different perceptions of the dynamism of the corresponding economies but also because of the legal framework under which they worked (i.e., contract of certain duration, etc). With respect to the effect of the interest rates and the exchange rate, the differences may be due to the much hig her number of migrants in Germany which made easier and more profitable the dissemination of relevant information through branches of Greek banks in German cities with large numbers of Greeks. It was mentioned earlier thatWahba (1991) attributes considerable significance to the availability of financial intermediation in attracting remittances. It is likely that the presence of this intermediation in Germany and the lack of it in Belg ium and Sweden makes the difference. Table 3 also presents some elasticity estimates. Relative to those presented in Table 2 for Germany, the differences are substantial. The income elasticity is lower, while the elasticities for stock of migrants and for unemployment are higher in absolute value. Cohort Effects For migrations extending over a period of many years, it is reasonable to assume that migrants of various periods may have different remittance behavior. This difference may reflect the impact of factors other than those captured in the variables used in the analysis. For example, the first Greek migrants came from urban areas, and migrants from the rural areas followed. It is possible that these two groups of migrants may have different remittance behavior for reasons related to cultural characteristics, ownership of wealth, attitudes, etc., and which cannot be reflected in the other explanatory variables. Assume that the group of migrants of the first five years (or from a certain region) has a propensity to remit equal to d\, i.e., REM = a^ 0 < ^ < 1 The group of migrants of the next five years has a propensity to remit equal to ^ 2, i.e., REM = a^ 0 < a ^ < 1 If in the second five-year period both groups of migrants remit part of their incomes, it can be shown that the propensity to remit for all migrants is a, i.e., REM=aY Vi ^2 where ^ = ^1 ————— + ^———-——. where ^d-Y ^ Y I +Y ^ In other words, the coefficient of income, which expresses the propensity to remit, may not be a parameter of constant value but a parameter with changing value as different groups of migrants are added to their total number. 84 INTERNATIONAL MIGRATION REVIEW TABLE 3 REGRESSION EQUATIONS FOR REMITTANCES FROM BELGIUM (1981-1991) AND SWEDEN (1980-1991) Independent Regression Elasticity Variable12Regression 1Regression 2 Constant -16,045 -22,356 (0.70) (0.94) Y 0.94 1.21 (2.09) (2.73) 0.91 1.81 MP 1,956 2,210 (4.35) (4.95) 1.82 R -1,754 -245 (1.08) (0.17) UN -2,277 -1,795 (2.35) (1.84) -0.86 INF 937 (1.70) R- 2 0.65 0.61 D-W•U 2^ Note: Diagnostic tests for white heteroskedasticity and for autoregressive conditional heteroskedasticity show no evidence of heteroskedasticity. The data for Germany that were used in the previous regression analysis can also be used to explore this hypothesis. This time the same regression equation of per capita remittances was used for the period of the first thirteen years, and then the following five years were added to the set of data, and so on. The results of these regressions are presented in Table 4. The point of interest in this table is the changing coefficient of the wage rate (W) in the first four regressions. It is clearly shown that these coefficients, which are estimates of the propensity to remit, are all positive and statistically significant. As data of more recent years until 1983 are added, the coefficient of W declines by approximately 50 percent, until it stabilizes in the fourth equation. The coefficient of income reduces from 950 in the first period to 448 in the second and 248 in the third, and it is highly significant in all equations. These results support the idea expressed earlier that there may be cohort effects in remittances behavior. However, the same result, that of a declining propensity to remit as the duration of migrations increases, can be explained by the hypothesis that the same group of migrants would reduce remittances as time passes and they established themselves in the country of destination. This alternative explanation is supported by the significance of the length of stay variable in the regression equations presented byElbadawi and Rocha (1992). The time series data we have do not allow discrimination between the two hypotheses. SUMMARY AND CONCLUSIONS The purpose of this study is twofold - first to review the theory and the evidence of migrant remittance behavior and second to present the evidence MIGRANT REMITTANCES: THE CASE OF GREECE 85 TABLE 4 REGRESSION EQUATIONS FOR REMITTANCES FROM GERMANY, VARIOUS PERIODS Regression Indepedent 123456 Variable 1961-1973 1961-1978 1961-1983 1961-1988 1961-1991 1974-1991 Constant 431 -223 -392 -390 -416 -254 (1.33) (0.57) (1.99) (2.47) (2.78) (0.50) W 950 448 248 251 252 208 (3.07) (2.34) (4.09) (5.88) (6.89) (2.35) W^ -2,007 -606 -297 -238 -251 -92 (1.71) (1.47) (1.22) (1.65) (2.08) (0.48) UN -38 -43 -2 -11 -13 -15 (1.33) (1.21) (0.05) (0.43) (0.84) (0.58) JNF 64 35 83 83 88 85 (1.15) (0.98) (2.63) (2.92) (3.30) (1.72) EX -131 -16 -20 -23 -23 -26 (-2.08) (0.35) (1.14) (6.07) (7.37) (3.59) iy 84 51 92 95 101 120 (1.39) (1.37) (2.53) (2.85) (3.21) (2.09) R' 2 0.97 0.94 0.90 0.90 0.90 0.54 D-W2^L951.661.631.721.78 Note: Values of t are given in parentheses. Subscript g indicates that variable refers to Greece. from Greece. This article has attempted a critique of the explanation of remittances that is based on the idea of an informal mutually beneficial contract between the migrant and his family. It is suggested that remittances should be understood as the result of a rational agent ' s behavior acting freely but constrained by his loyalty and commitment to his family. In terms of the objective of the migrant, it is suggested that there is a wide variety of plans that may apply to different circumstances. The study also presents the results of an econometric investigation that utilizes the data on remittances from Germany, Belgium, and Sweden. From these results it appears that the volume of remittances is affected by the level of income of migrants, the rate of inflation, the exchange rate, the rate of interest, and the number of migrants. The rate of unemployment seems to be important but not in all cases. The results also suggest that there are important cohorts effects that may be attributed to different attitudes of the corresponding groups of migrants and/or to the effect that the duration of migration may have on remittances even if migrants of different cohorts have identical behavior. Our results reaffirm the significance of the level of income in the host country and the exchange rate and also the lack of significance of income levels in the country of origin. They provide support for the rate of inflation and the interest rate differentials as important explanatory variables of migrant remit- 86 INTERNATIONAL MIGRATION REVIEW tance behavior, and they also suggest that the rate of unemployment should be considered as a factor affecting remittances. Finally, our results suggest that there are important cohort effects. REFERENCES Appleyard, R. T. 1989 " Migration and Development: Myths and Reality," International Migration Review, 23(3):486-499. Bank of Greece Various Monthly Statistical Bulletin. Athens: Bank of Greece. Years Becker, G. 1981 A Treatise on the Family Cambridge, MA: Harvard University Press. Chandarvarkar, A. B. 1980 "Use of Migrants' Remittances in Labor-Exporting Countries," Finance and Development, June:36—39. Choucri, N. 1986 "The Hidden Economy: A New View of Remittances in the Arab World," World Development, l4(6):697-712. Ecevit, Z. and K. C. Zachariah 1978 "International Labor Migration," Finance and Development December:32-37. Elbadawi, I. A. and R. Rocha 1992 Determinants ofExpartiate Workers Remittances in North Africa and Europe. World Bank Working Paper 1038. Washington, DC: The World Bank. Glytsos, N. 1988 "Remittances in Temporary Migration: A Theoretical Model and Its Testing with the Greek-German Experience," Weltwirtschaftliches Archiv, 124:524-549. International Labor Office (ILO) Various Yearbook of Labor Statistics. Geneva: ILO. Years International Monetary Fund (IMF) Various International Financial Statistics. Washington, DC: IMF. Years Johnson, G. E. and W E. Whitelaw 1974 "Urban-Rural Income Transfers in Kenya: An Estimated Remittances Function," Economic Development and Cultural Change, 22(3):473-479. Katseli, L. and N. Glytsos 1989 "Theoretical and Empirical Determinants of International Labour Mobility: A Greek-German Perspective." In European Factor Mobility ^ Trends and Consequences. Ed. I. Gordon and A. P.Thirlwall. London: MacMillan. Keely, C. and B. N. Tran 1989 "Remittances from Labor Migration: Evaluations, Performance and Implications," International Migration Review, 23(3): 261—287. Lucas, R. E. B. and 0. Stark 1985 "Motivation to Remit: Evidence from Botswana," Journal of Political Economy, 93:901-918. MIGRANT REMITTANCES: THE CASE OF GREECE 87 Martin, P.L. 1983 "The Economic Effects of Temporary Worker Emigration." East-West Population Institute Conference, Asian Labor Migration to Middle East. Honolulu: East-West Center. Mohammad, A., W A. Butcher and C. H. Gotsch 1973 Temporary Migration of Workers and Return Flow of Remittances in Pakistan. Economic Development Reports No. 234. Cambridge, MA: Center for International Affairs, Harvard University. Norman, R. 1983 The Moral Philosophers. Oxford: Clarendon Press. Nowell-Smith, P.H. 1959 Ethics. Middlesex: Penguin. Rempel, H. and R. Lodbell 1978 "The Role of Urban-to-Rural Remittances in Rural Development," Journal of Development Studies, 14:324-341. April. Russell, S. S. 1992 "Migrant Remittances and Development," International Migration, 30:267-287. 1986 "Remittances from International Migration: A Review in Perspective," World Development, 14:677-696. Serageldin, G. etal. 1981 "Manpower and International Labor Migration in the Middle East and North Africa." Final Report, Research Project on International Labor Migration and Manpower in the Middle East and North Africa. Washington, DC: The World Bank. Swamy, G. 1981 International Migrant Workers Remittances: Issues and Prospects. World Bank Staff Working Paper No. 481. Washington, DC: The World Bank. Wahba, S. 1991 "What Determines Workers' Remittances?" Finance and Development, December^ 1-44. UndocumentedLatino Immigrants in Orange County, California: A Comparative Anal y sis 1 F. Allan Hubbell University of California, Irvine Shiraz I. Mishra University of California ^ Irvine R. Burciaga Valdez University of California) Los Angeles This article examines a unique data set randomly collected from Latinas (including 160 undocumented immigrants) and non-Hispanic white women in Orange County, California, including undocumented and documented Latina immigrants, Latina citizens, and non-Hispanic white women. Our survey suggests that undocumented Latinas are younger than documented Latinas, and immigrant Latinas are generally younger than U.S.-citizen Latinas andAn g lo women. Undocumented and documented Latinas work in menial service sector jobs, often in domestic services. Most do not have job-related benefits such as medical insurance. Despite low incomes and likelihood of having children under age 18 living with them, their use of public assistance was low. Undocumented and documented Latina immigrants lived in households that often contained extended family members; they were more likely than other women in the study to lack a regular source of health care, to utilize health clinics, public health centers, and hospital emergency rooms rather than private physicians or HMOs, and to underutilize preventative cancer screening services. Despite their immigration status, undocumented Latina immigrants often viewed themselves as part of a community in the United States, which significantly influenced their intentions to stay in the United States. Contrary to much of the recent public policy debate over immigration, we did not find that social services influenced Latina immigrants ' intentions to stay in the United States. ^Supported by a grant from the National Cancer Institute (5 R01 CA 52931). The contents of the manuscript are solely the responsibility of the authors and do not necessarily represent the views of the funding agency. The authors are indebted to Juliet McMullin and the anonymous reviewers for their generous comments. ©1997 by the Center for Migration Studies of New York. All rights reserved. 0197-9183/97/3101.0117 Leo R. Chavez University of California ^ Irvine COMPARATIVE ANALYSIS OF UNDOCUMENTED LATINA IMMIGRANTS 89 Orange County is the third most populous county in California, with the 1990 population estimated at 2,410,556 (U.S. Census Bureau, 1991). It covers an area of 786 square miles, is largely urban, and contains 31 cities and numerous unincorporated communities. Approximately 23 percent of Orange County ' s population is Latino. Most Latinos are of Mexican heritage. Latino immigrants from other nations in Latin America, par ticularly Central America, also live in the county (Hispanic Development Council, 1989). Latinos are found in greater concentrations in the northern half of the county, which includes Santa Ana, where two out of three residents were Latino in 1990 (U.S. Census Bureau, 1991). The southern half of the county has been an area of rapid growth in new middle, upper-middle, and exclusive residential communities. Latino immigrants often work in southern county communities but find less expensive housing in the many working-class communities in the northern part of the county. Orange County was also one of the areas where Proposition 187 found substantial support. Ronald Prince, one of the cofounders of the Save Our State (SOS) initiative (which later became Proposition 187), is based in Orange County and regularly spoke at pro-Proposition 187 rallies in the county (McDonnell, 1994). U.S. Representative Dana Rohrabacher (R-Huntington Beach, Calif.) was also a vociferous proponent of Proposition 187 (Martinez and McDonnell, 1994). In sum, the demographics of the county and the local concern for the public policy issues surrounding immigration reform make Orange County a particularly apt place to examine questions about immigrant behavior and intentions. This article examines a unique data set, one that was randomly collected from Latinas and non-Hispanic ("Anglo") women in Orange County as part of a large study on Latinas ' knowledge, attitudes, beliefs, and practices related to breast and cervical cancer. Researchers have made substantial contributions toward understanding the role of women in migration, settlement, the maintenance of transnational linkages, and their labor market experiences (Basch, Glick Schiller and Szanton Blanc, 1994; Hagan, 1994; Chavez, 1994, 1991, 1988; Hondagneu-Sotelo, 1994; Borjas and Tienda, 1993; Donato, 1993; Lamphere, 1992; Chavez, Flores and Lopez-Garza, 1990; Rouse, 1991; Marmora, 1988; Massey et aL, 1987; Alvarez, 1987; Simon and Brettell, 1986; Simon and DeLey, 1984; Ranney and Kossoudji, 1984; Reichart and Massey, 1979). However, it is often difficult to generalize from studies on undocumented immigrants (Valdez et aL, 1993; National Research Council, 1985). The problem is that a lack of immigration authorization and the living and working conditions of undocumented immigrants often makes them wary, uncooperative, and difficult to locate, all of which pose problems for applying a random sampling frame (Cornelius, 1982). Therefore, studies of undocumented immigrants often rely on apprehension statistics provided by the Immigration and Naturalization Service, snowball sampling techniques, data 90 INTERNATIONAL MIGRATION REVIEW supplied by return migrants in their place of origin, census data which relies on assumptions to extrapolate immigration status, or data provided by legalized, formerly undocumented, immigrants (Valdez et aL, 1993). In contrast, the random sample of Latinas that we examine here includes undocumented immigrants. Trained bilingual women interviewers from the Field Research Corporation in San Francisco conducted our telephone survey from September 1992 to March 1993. Eligible participants were English or Spanish speaking women 18 years of age or older who were not institutionalized and who identified themselves as White (Anglo, Caucasian, non-Hispanic White) or Latino (Hispanic or more specific ethnic identifiers such as Mexican or Mexican-American). We sought a larger subsample of Latino respondents in order to examine variation within the population. The telephone survey used a crosssectional sample of random digit telephone listings to identify eligible subjects. Both listed and unlisted numbers appeared in the listings, avoiding potential bias due to exclusion of households with unlisted numbers (Survey Sampling, 1990). Telephone survey findings may not be generalizable to families without telephones. In Orange County, however, approximately 94 percent of Latinos and 99 percent of Anglos have telephones (California State Census Data Center, 1995). Another potential limitation of the study would be that it would not find hard to reach members of the population - the homeless and those engaged in street corner employment and migrant agricultural labor. This may be more of a bias, however, for male rather than female Latinas, who are less likely to be homeless or seek day work by standing on street corners. Our survey randomly selected both households and respondents within households - the woman 18 years or older who had the most recent birthday The cooperation rate was 78.5% (Hubbell et aL, 1995). 2 Latin respondents could choose to answer the questions in Spanish or English. We pilot-tested the questionnaire, tested its content validity and translated it from English to Spanish to English. The final questionnaire included inquiries about demographic characteristics and medical care access as well as breast and cervical cancer related knowledge, attitudes, and practices. It also included a previously validated 5-point ' acculturation " scale (Marin et aL, 1987) that measured acculturation primarily on the use of Spanish (e.g., to read with, speak with, think with, used as child, and speak with friends). The question about immigration status came at the end of the interview. Perhaps because this was a study focusing on health issues, many respondents gave us information about their immigration status. We recorded their self-reported immigration status, which is presented below, and then classified ^he cooperation rate, defined as the number of completed interviews divided by the sum of the completed interviews and refusals by eligible women [1,225/(1,225+336)], was 78.5%. COMPARATIVE ANALYSIS OF UNDOCUMENTED LATINA IMMIGRANTS 91 the respondents into undocumented immigrants, documented immigrants and U.S. citizens. Out of 803 Latina respondents, 160 were undocumented immigrants; this accounts for about 20 percent of all the Latinas sampled and 30 percent of Latina immigrants sampled. It is reasonable to assume that the actual number of undocumented women in the sample is actually higher given the reluctance of some to admit to such a status. SAMPLE CHARACTERISTICS We interviewed 803 Latinas (270 U.S.-born, 428 Mexican immigrants, and 105 immigrants from other Latin American countries) and 422 Anglo women. About a third of Latinas in the sample were born in the United States or Puerto Rico (Table 1). Most Latinas were born in Mexico (53.3%), with Central American countries (7.6%) and South American countries (4%) accounting for smaller proportions. Latina immigrants self-reported their immigration status as permanent legal resident (60.1%); without immigration papers or with false documents (17.7%); without immigration papers but having requested a work permit from the INS (3.3%), without immigration papers but having requested permanent residence (7.6%); without immigration papers but having requested permanent asylum (0.4%); naturalized U.S. citizen (8.4%); Temporary Protective Status (TPS) (2.1%); and political asylee (0.4%). 3 For analytical purposes, we classified all the respondents who self-reported as" without papers " at the time of the interview as undocumented immigrants. 4 We also included TPS holders in this category since this was a temporary status for otherwise undocumented immigrants. Our four comparative subsamples are: a) 160 undocumented Latina immigrants; b) 311 documented Latina immigrants; c) 313 Latina citizens (270 U.S.-born and 43 naturalized); and d) 422 Anglo women. Undocumented immigrants comprised 19.9% of our total sample of Latinas. Assuming this is a minimum proportion of the county ' s Latinas that are undocumented, that would equal about 33,430 of the total Latinas 18 years and older counted in the 1990 census. Looked at another way, undocumented Latinas in our sample represent about 3.7% of the total females age 18 and older in the county. Table 1 shows the proportion of undocumented Latinas from specific countries of ori gin. ^Frequencies based on a subsample of Latina immigrants that has nineteen missing cases due to no response (533-19=514). Determination of an immigrant's status is very complicated and is best left up to an immigration judge. 92 INTERNATIONAL MIGRATION REVIEW TABLE 1 LATINAS COUNTRY OF BIRTH AND PERCENTAGE OF UNDOCUMENTED IMMIGRANTS FROM EACH COUNTRY All (%) Undocumented Latina BornN= 803Immigrants per Country (%) U.S. Except Puerto Rico 32.5 0 Puerto Rico 1.1 0 Cuba 0.9 0 Mexico 53.3 33.2 El Salvador 3.0 25.0 Guatemala 2.7 35.0 Nicaragua 0.4 33.3 Other Central American Country 1.5 16.7 South American Country4.06.3 Sociodemographic Characteristics The respondents ' demographic characteristics appear in Table 2. The median age for all Latinas in our sample was 31, which is comparable to the 30—34 median age category for Latinas age 18 and older found by the 1990 census (State of California Demographic Research Unit, 1990). Anglo women had a median age of 41, which was also comparable to the 40—44 median age category for "Non-Hispanic white " women age 18 and older in the 1990 census of the county. Latinas ' ages varied by immigration/citizenshi p status. Undocumented Latinas, with a median age of about 27, were significantly (p < 0.001) younger than documented Latina immigrants (median = 33), and Latina citizens (median = 34). Along these lines, few undocumented (1.2%) and documented immigrant (8.1%) Latinas were 50 years old or older, compared to about 17 percent of Latina citizens and more than a quarter of Anglo women. Undocumented Latinas had been in the United States for a median of four years, about a third of the time for documented Latina immigrants, a significant difference (p < 0.001). Interestingly, the median time for both undocumented and documented Latina immigrants in Orange County came close to the amount of time they had been in the United States, indicating a relatively stable population. Undocumented and documented Latina immigrants generally scored much lower on the language/acculturated scale than citizen Latinas. Undocumented immigrants had a median score of 1 and documented Latina immigrants a median score of 1.2 on a 5-point scale (Marin et aL, 1987). Citizen Latinas ranged had a median score of 4.2, indicating a highly " acculturated, " or English-language dominant, population. Undocumented and documented Latina immigrants had a median of nine years of education, well below the thirteen and fourteen years of education for COMPARATIVE ANALYSIS OF UNDOCUMENTED LATINA IMMIGRANTS 93 TABLE 2 SOCIODEMOGRAPHIC CHARACTERISTICS Latina Latina Undocumented Legal Ladna White Immigrants Immigrants Citizens "Wbmen N=160 A^=311 N=W N=422 Demographic Characteristics Median Age 27 Married (%) 64 Median Years of Schooling 9 Median Years in United States 4 Median Years in Orange County 3 Median Language/Acculturation Score (5-Point Scale) 1.0 Family Composition Median Age at First Birth 20 Median Children Born 2 Children < 18 Living with Respondent (%) 81 Median Immediate Family Members Living Together 4 Median Persons in Household 5 >1 Married Couple in Household (%) Current Work Status Employed Full-Time (%) Employed Part-Time (%) Homemaker (%) Unemployed - Seeking Work (%) Unemployed - Not Seeking Work (%) Retired (%) Spouse's Work Status Employed Full-Time (%) Employed Part-Time (%) Unemployed - Seeking Work (%) Unemployed - Not Seeking Work (%) Retired (%) Household Income <$15,000 (%) $15,000-$24,999 (%) $25,000-$34,999 (%) $35,000+ (%) Latina citizens and Anglo women, respectively. According to the 1990 census, 28.2 percent of all (both male and female) Latinos and 63.5 percent of all Anglos in the county had some education beyond high school. We found that 35.1 percent ofLatinas and 74.5 percent of Anglo women had thirteen or more years of schooling. 333441 676159 91314 13NANA 111818 1.24.2NA 212124 322 826249 44 0 251151 24425351 14101212 44311718 10653 91096 01411 67807980 15622 16763 3442 021013 76461410 17312312 481210 1155168 94 INTERNATIONAL MIGRATION REVIEW Famil y and Household Characteristics About the same proportions of all respondents were currently married (Table 2). Undocumented Latinas (15%), however, were more likely to be living with (not formally married to) their partner than Latina documented immigrants (7%) and citizens (4%), as well as Anglo women (4%). The median age at which all women in the sample had their first child was in their 20s, although all Latinas were generally a few years younger than Anglo women. Because they were relatively young, undocumented Latinas had given birth to fewer children (median = 2) than documented Latina immigrants (median = 3 children), and the same number as U.S. citizen Latinas and Anglo women. Also, because they were younger and in the reproductive stages of the life cycle, immigrants had more children under 18 years of age living with them than citizens. Of the undocumented women, 81 percent lived in households with children under 18 years of age, as did 82 percent of Latina documented immigrants. Fewer Latina citizens (62%) and Anglo women (49%) had children under age 18 living with them. Latina immigrants, both documented and undocumented, co-resided with a median of four immediate family members (spouse, if any, and children), one more than Latina citizens or Anglo women. But Latina immigrants in general had more extended family and nonrelated individuals in their households, which was reflected in higher median household sizes. Latina citizens and Anglo women, in contrast, had the same median number of immediate family members and total household members. The number of married couples in the household varied by immigration/citizenship status. A quarter of undocumented Latina immigrants lived in households that contained two or more married couples, about twice as often as documented Latina immigrants (11%), five times as often as Latina citizens (5%), and 25 times as often as Anglo women (1%). Labor Market Participation Table 2 also presents the occupational profile of the sample. Most of the Latinas (61%) and Anglo women (65%) in our sample were to some extent in the labor force; a participation rate comparable to that for Latinas (64%) and Anglo women (63%) found in the 1990 census of the county (State of California Demographic Research Unit, 1990). Labor market participation varied by immigration/citizenship status. Less than a quarter of undocumented Latina immigrants were employed full-time, compared to over 40 percent of documented Latina immigrants and over 50 percent of Latina citizens and Anglo women (Table 2). Undocumented and documented Latina immigrants were much more likely than Latina citizens or Anglo women to be homemakers. Fourteen percent of undocumented Latinas were employed par t-time, which was only sli ghtly more than the other women COMPARATIVE ANALYSIS OF UNDOCUMENTED LATINA IMMIGRANTS 95 in the study. The ten percent rate of unemployed (but seeking employment) for undocumented Latinas was about twice that of documented immigrant and citizen Latinas, and over three times that ofAnglo women. About two-thirds of undocumented Latinas had a husband who was employed full-time, compared to well over three-quarters of documented immigrant and citizen Latinas, and 80 percent ofAnglo women. Undocumented (15%) and documented immigrant (6%) Latinas were much more likely than citizen Latinas (2%) and Anglo women (2%) to have a husband employed par t-time. In addition, 16 percent of undocumented Latinas had a husband who was unemployed and looking for work, about twice that of documented immigrant women (7%), almost three times that of citizen Latinas, and about five times that ofAnglo women. Table 3 summarizes the occupations for the employed respondents. Undocumented Latinas were employed in only eight job categories, much fewer than documented immigrant (17 categories) and citizen Latinas (17 categories), as well as Anglo women (22 categories). The majority of undocumented Latinas (58%) worked in service jobs, which included mostly housecleaning and child care but also work as waitresses, hotel maids, and kitchen workers. Many undocumented Latinas also worked in sales (10%), clerical work (9%), and precision production work (9%) which included dressmakers. About seven percent of undocumented Latinas worked in jobs categorized as laborers, helpers, and equipment cleaners, which included warehouse workers, construction trade workers, and handpickers and packaging. Some undocumented Latinas (5%) had jobs categorized as machine operators and tenders, which were non-precision jobs, such as metal and plastic processing, packaging, textile and apparel machine operators, and par ts assemblers and sorters. Documented Latina immigrants also concentrated in household and other services, which accounted for 34 percent of their occupations. One out of five documented Latina immigrants were machine operators and tenders. In addition, clerical work (14%) was also a substantial occupational category for the documented Latina immigrants. The most common occupation among Latina citizens was clerical work (36%). Clerical work was also the most common occupation among Anglo women (25%) as well, but proportionately less so than among Latina citizens. Sales, managerial positions, and education workers (i.e., teachers, professors, vocational counselors, librarians) were also frequent occupations of Latina citizens and Anglo women. The propor tion of Latinas in such occupations, however, was consistently less than that ofAnglo women, especially in the managerial and other occupations requiring college or greater educational levels for entry. Earnings also varied by immigration/citizenship status. As Table 2 indicates, the yearly income (which includes spouses income and any government-supplied income) of more than three-fourths (76%) of the undocumented Latinas fell 96 INTERNATIONAL MIGRATION REVIEW TABLE 3 OCCUPATIONS OF WOMEN IN STUDY Ladna Women Undocumented Legal Immigrants Immigrants N= 57 N= 159  (%) (%) Professional Scientific-engineering 0 0 Medical-related work 0 1 Education-related work 2 1 Social science 0 1 Law (judge-lawyer) 0 0 Arts and athletics (actors, musicians, 0 0 wr iters, editors, artists, athletes) Managerial Executive, coporate, military officer 0 0 Public administrators 0 1 Other manager 0 6 Technical, Sales, Administrative Support Techinicians, support 0 4 Sales 10 4 Administrative support - clerical 9 14 Services Private household 14 6 Protective services 0 0 Other service jobs 44 28 Precision Production, Crafts, Repairs Mechanics, repairers 0 1 9 4 Precision production (baker, butcher, cabinetmaker, dental lab technician, dressmaker, tailor, etc.) Operators, Fabricators, Laborers Machine operators and tenders 5 21 Transportation, material moving 0 1 Laborers, helpers, equipment cleaners 7 6 Farming, Forestry Agriculture-related 0 1 Forestry 0 0 Totals100100 actual percentage was 0.4, which was rounded down to 0. COMPARATIVE ANALYSIS OF UNDOCUMENTED LATINA IMMIGRANTS 97 predominately into the under $15,000 category. Less than one percent of undocumented Latinas had a yearly income above $35,000. Most documented Latina immigrants (77%) had a yearly income under $25,000. In contrast, the majority of citizen Latinas (51%) and Anglo women (68%) had a yearly income above $35,000. More than a third (37%) of citizen Latinas, however, were clustered in the under $25,000 income categories compared to only 22 percent of Anglo women. Public Assistance, Medical Insurance, and the Use of Health Services The jobs held by undocumented Latinas and their spouses have implications for their use of public benefits and their forms of medical insurance coverage. Respondents were asked if they were receiving any form of public assistance or welfare at the time of the interview. Aid to Families with Dependent Children (AFDC) is the principal government program available to low-income undocumented women with U.S. citizen children; AFDC is provided for the citizen child and not the undocumented mother. Although undocumented Latinas had lower incomes and were more likely to have children under 18 living with them, their rate of use of public assistance was equal to that of documented immigrants and Latina citizens, and not significantly different from Anglo women. Four (3%) undocumented Latinas, eight (3%) documented Latina immigrants, eight (3%) citizen Latinas, and four (1%) Anglo women indicated they were receiving assistance at the time of the interview. We should note that we did not measure the use of public assistance at other times, the frequency of use, nor the duration of use, all of which could have different outcomes than this variable. Table 4 presents medical insurance coverage and source of health care for the Latinas and Anglo women. Medical insurance can be either private, typically provided on a voluntary basis through the workplace, or governmentsponsored, which includes a number of programs such as Medicare (or Medi-Cal as it is called in California) and county indigent medical services. We found that only 21 percent of undocumented women had private medical insurance coverage, significantly below all other groups of women in our study. In addition, fewer documented Latina immigrants (52%) compared with Latina citizens (77%) and Anglo women (86%) had private medical insurance. Undocumented immigrants are ineli g ible for most government-sponsored health programs, the principal exception being for emergency services and pregnancy-related services for U.S.-born children. About 18 percent of undocumented Latinas turned to government-sponsored programs for assistance with medical care. Private and government medical insurance together meant that only 39 percent of undocumented women had some type of medical insurance, well below the coverage held by the other women in the study. 98 INTERNATIONAL MIGRATION REVIEW TABLE 4 MEDICAL INSURANCE COVERAGE AND SOURCE OF HEALTHCARE Ladna Women Anglo Women Undocumented Legal Immigrants Immigrants Citizens N=160 N=311 N=313 N=422 (%) (%)  (%)  (%) Medical Insurance Private insurance 21 52 77 85 Government insurance (Medicare, Medi-Cal, IMS, etc.) 18 13 13 14 Medically uninsured 61 35 10 1 Sources of Health Care No regular source of healthcare 41 16 4 2 Healthcare type for interviewees with regular source of healthcare Private physician 21 44 66 77 HMO 1 6 16 16 Hospital outpatient clinic 25 18 8 3 Public community clinic or health center 45 30 9 3 Hospital emergency room 410 1 Other411O 8 actual percentage was 0.2. Variation in the level of insurance coverage influences patterns of medical service utilization. Undocumented Latinas (41%) were significantly more likely than all other women to lack a regular source of medical care (Table 4). Documented Latina immigrants (16%) were much less likely to lack a regular source of medical care than undocumented Latinas, but still about four times less likely than Latina citizens and almost eight times less likely than Anglo women. Table 4 shows the types of medical care providers regularly used by respondents. Undocumented and documented Latina immigrants rely on health department medical clinics, community clinics, and hospital outpatient clinics. A few undocumented women used hospital emergency rooms, the most costly form of primary care, as their source of primary medical care. Insured Anglo women, citizen Latinas, and to a lesser extent documented Latina immigrants relied on private physicians and HMOs. Logistic regression analyses on the use of cancer screening services underscore the disadvantage of being an undocumented immigrant. Table 5 presents results of a logistic regression analysis with the dependent variable is the acquisition of a Pap smear, a test for detecting cervical cancer. The dependent variable is coded: 0 = never had a Pap smear test or had the test more than three years before the interview; 1 = had the test within the last three years. The COMPARATIVE ANALYSIS OF UNDOCUMENTED LATINA IMMIGRANTS 99 TABLE 5 LOGISTIC REGRESSION: USE OF PAP SMEAR CANCER SCREENING TESTS BY LATINAS IN ORANGE COUNTY, CALIFORNIA Pap Smear Test Within Last Three Years Betas^ORp= Documented Immigrant/Citizen .60 .26 1.82 <0.03 Medical Insurance .97 .23 2.64 <0.0001 Family Income >$20,000/Year -.01 .27 .99 n.s. >12 Years of Education .44 .24 1.55 n.s. Married .77 .24 2.17 <0.01 Employed .19 .23 1.21 n.s. Language/Assmilation Scale.89.292.44<0.01 independent variables were inmiigradon status (undocumented = 0, documented immigrant or U.S. citizen = 1); public or private medical insurance (none = 0, yes = 1); marital status (not married = 0, married or living together =1); years of schooling (0-12 = 0, >12 = 1); employment status (unemployed = 0; employed = 1); yearly family income (defined as interviewees income plus spouses income, above or below the median income for Latinas) (<$20,000 = 0, >$20,000 = 1), and the language/assimilation 5-point scale (<3 = 0; 3-5 = 1). According to the Odds Ratio (OR), even when controlling for the other independent variables, documented and citizen Latinas were over 80 percent more likely than undocumented Latinas to have recently had a Pap smear. Latinas with medical insurance, also controlling for the other variables, were over two-and-half times more likely than the uninsured to have had a Pap smear recently. Being married and having a higher score on the language/assimilation scale also significantly influenced the use of Pap smear tests. The Imagined Community and Future U.S. Residence 5 Most undocumented (84%) and documented immigrant Latinas (89%) indicated that they felt part of a community in the United States. However, only 60 percent of undocumented Latinas who had been in the United States less than three years indicated they felt a sense of community in the United States. The proportion rises to 87 percent among undocumented Latinas in the United States for more than five years. Most undocumented Latinas (63.6%) and documented immigrants (93%) desired to stay in the United States. Only about half (51.7%) of the undocu-5 Benedict Anderson suggested that communities are "imagined." Members of modern nations cannot possibly know their fellow members and yet "in the minds of each lives the image of their communion. ... It is imagined as a community because regardless of the actual inequality and exploitation that may prevail in each, the nation is always conceived as a deep, horizontal comradeship " (Anderson, 1983: 15-16). 100 INTERNATIONAL MIGRATION REVIEW mented Latinas who had been in the United States for five years or less intended to stay, compared to 78.6 percent of documented immigrants with similar time in the United States. Over three-quarters (79.5%) of undocumented Latinas who had been in the United States over five years intended to stay, as did 93.9 percent of documented immigrants. Latinas living in the northern half of the county — which is where Latinos are concentrated — were also significantly more likely than Latinas living in the southern part of the county to feel part of a community (X 2 < .05). Mexican immigrant women in the northern part of the county were also significantly more likely to feel part of a community compared to Mexican immigrants in the southern part of the county (X 2 < .05). The influences on Latina immigrants ' desire to settle in the United States were examined using logistic regression analysis. The dependent variable was the respondents intention to stay in the United States (coded 1) or not (coded 0). The independent variables, length of time in the United States, immigration status, employment status, marital status, years of schooling, yearly family income, were coded as in the previous logistic regression analysis. This analysis also includes as independent variables private medical insurance (no = 0, yes = 1); a regular source of health care (no = 0, yes = 1); using public assistance at the time of the interview or having relied on public assistance with medical care costs (no = 0, yes = 1); length of time in the U.S. (<5 years = 0, >5 years = 1); child under 18 years old in household (no = 0, yes =1); and feels part of a community in the United States (no = 0, yes = I). 6 Table 6 shows the results of two logistic analyses, the first of which includes all Latina immigrants (Model 1) and the second of which includes only undocumented Latina immigrants (Model 2). Four variables were significant for explaining Latina immigrants ' desire to continue residing in the United States. First of all, controlling for all the other independent variables, Latina immigrants who had been in the United States more than five years were, according to the Odds Ratio, over six times more likely than those with less time to intend to stay. In addition, Latina immigrants who felt part of a community in the United States were over six-and-a-half times more likely than Latina immigrants without such sentiments to express intentions to stay in the United States. Documented immigrants were about four-and-a-half times as likely as undocumented immigrants to intend to settle. Finally, having a regular source of medical care was also a significant predictor of U.S. residency intentions, but in the opposite direction of that suggested by anti-immigrant discourse. Latina immigrants with a regular source of health care were 66 percent less likely than those without it to intend to stay in the United States. ^he language/acculturation variable was left out of the logistic regression analyses because of a high correlation with time in the United States (.49). COMPARATIVE ANALYSIS OF UNDOCUMENTED LATINA IMMIGRANTS 101 TABLE 6 LOGISTIC REGRESSION: INFLUENCES ON LATINA IMMIGRANT INTENTIONS TO STAY IN THE UNITED STATES Model2 Model 1 Undocumented Ladna All Latina Immigrants Immigrants Beta s.e. OR p= Beta s.e. OR p= Private Medical Insurance -.39 .51 .68 n.s. -.34 .56 .72 n.s. More than 5 Years in United States 1.86 .48 6.40 <0.001 1.12 .48 3.07 <0.02 Married -.33 .55 .72 n.s. -.38 .59 .69 n.s. Children <18 in Household .50 .44 1.66 n.s. .75 .53 2.11 n.s. Employed .37 .46 1.44 n.s. .13 .48 1.13 n.s. Regular Source of Health Care -1.09 .48 .34 <0.03 -1.05 .48 .35 <0.05 Public Assistance -.11 .55 .90 n.s. -.18 .57 .84 n.s. Feels Part of Community in United States 1.89 .45 6.59 <0.0001 2.3 .65 10.02 <0.001 > 12 Years of Schooling -.17 .48 .84 n.s. Family Income > $20,000/year .71 .60 2.03 n.s. a Legal Immigration Status1.60 .51 4.95 <0.01 a  "Variable left out of analysis because of the smaller A^ofthe undocumented immigrant subsample. Model 2 examines the intention to stay in the United States for undocumented Latina immigrants only/Three variables were significant predictors of undocumented Latina immigrants desire to settle in the United States. Controlling for the other variables, undocumented Latinas who expressed a sentiment of community membership were ten times as likely as those without such sentiments to intend to stay in the United States. Undocumented Latina immigrants who had been in the United States for more than five years were three times as likely as those with less time to intend to stay. Having a regular source of medical care was significant. It was, however, in a negative direction; undocumented Latinas with a regular source of health care were 65 percent less likely than those without it to intend to stay in the United States. DISCUSSION This study used a randomly selected group of Latinas and Anglo women to answer five questions related to undocumented immigrants and their behavior. The questions concerned sociodemographic characteristics, family and household composition, labor market participation, medical service utilization, and settlement in the United States. The portrait of undocumented Latinas drawn from these data is one of young women in the family formation stage of their lives. More than other women in the study, undocumented Latinas were more likely to be "living ^eft out of Model 2 were education and income because of the smaller size of the undocumented Latina subsample. 102 INTERNATIONAL MIGRATION REVIEW together " rather than formally married. Hagan (1994) suggests that such informal living arrangements may be due to the lessened social control over young couples by parents and other older members of the family who are usually back in the country of origin. They are also likely to live in households that have young children and that have extended family members, including more than one married couple. Such complex households among undocumented Latinas are the product of their tenuous existence as undocumented immigrants, their low incomes in relation to housing costs, and the use of social networks during the migration process (Chavez, 1985; 1990). Documented Latina immigrants, who also provide lodging to newly arrived relatives, are more likely than Latina citizens and Anglo women to live in complex households, but much less so than undocumented Latinas. The labor market participation of undocumented Latinas was also influenced by their age and stage in the reproductive cycle of the family. Undocumented and documented Latina immigrants were much more likely than Latina citizens or Anglo women to be homemakers. Undocumented Latinas in the labor force worked in low paying, menial service sector jobs, often in domestic services. They were also clustered in fewer types of jobs than other women in the study. Over three-quarters of undocumented women had annual family incomes under $15,000, which was much lower than the other groups of women in the study Most did not have job-related benefits such as medical insurance. Despite their low incomes and likelihood of having young children living with them, undocumented Latinas ' use of public assistance was expectedlylow. Some undocumented women did turn to government programs for assistance for medical care costs. The evidence here, however, suggests that undocumented Latinas and many documented Latina immigrants are among the working poor and that they primarily rely on their own and their familys resources for their financial support. Undocumented Latinas do not appear to be enmeshed in the government-sponsored social safety net. Undocumented, and to some extent documented, immigrant Latinas ' use of medical services follows a pattern characteristic of low-income, uninsured people: relying on public health centers, community clinics, and hospital outpatient clinics, or, for some, hospital emergency rooms for primary medical care. This profile of the use of health services does not correspond to their image as abusers of the healthcare system. Many undocumented Latinas have no regular source of health care, and preventative care is a low priority or difficult to obtain. Indeed, their use of cancer screening tests were significantly below that of all other women in our study. Even when controlling for medical insurance, undocumented Latinas faced obstacles to their use of preventive cancer services. Undocumented and documented Latinas, like other uninsured Americans, have a pattern of medical service utilization that differs radically from medically insured citizens and legal residents, who seek health care COMPARATIVE ANALYSIS OF UNDOCUMENTED LATINA IMMIGRANTS 103 principally from private physicians and HMOs. It should be noted, however, that although these findings suggest the extent of structural segmentation in Orange Countys medical care delivery system, the data do not address issues related to the relative quality of care. Undocumented Latinas develop feelings of belonging to a community in the United States. Why this occurs most likely has to do with their migration experiences and motivations. They are working toward their own, and their family s, economic goals; they are searching for a "better life " and opportunity In addition, Latina immigrants often migrate into established households and into areas which have other family and friends from the same region, even village, in the country of ori g in. They often move into nei ghborhoods consisting of other Spanish-speaking Latinos (Chavez, 1992). Importantly, Latinas in general, and Mexican immigrants in particular, living in the northern part of the county (where Latinos are concentrated) were significantly more likely to feel a sense of community than those living in the southern part of the county These findings underscore the observations of other researchers who stress the ability of undocumented Latinas to develop a sense of community even under harsh social and economic conditions (Hondagneu-Sotelo, 1994; Hagan, 1994; Chavez, 1994). What motivates undocumented Latinas to stay in the United States? We found that feeling part of a community in the United States and acquiring more time in the United States independently influenced their desire to stay. Documented immigrants were also significantly more likely than undocumented Latina immigrants to intend stay in the United States. Some of the key variables in the public debate over immigrants turned out not to be significant influences on settlement. Neither undocumented nor documented Latina immigrants were influenced in their intentions to stay in the United States by having a regular source of medical care. In fact, they were less likely to intend to stay, perhaps, as other studies have suggested, because of a dissatisfaction with the quality of care they were receiving, language differences, difficulty of translating culturally specific health syndromes (e.g., emphacho, susto, molliera de caida) ^ and the high cost of care for those without medical insurance or whose insurance does not cover the total cost of medical care (Chavez and Torres, 1994; Chavez, 1984). In addition, some Latino immigrants with a regular source of medical care may desire to return to their native country, perhaps to retire. Latina immigrants were also not significantly influenced to stay by receiving public assistance. In addition, Latinas with children under 18 years of age in their households would, theoretically, be concerned with public education. And yet, they were not significantly more likely to desire to stay in the United States than respondents without similaraged children in the household. These findings are contrary to the assumption that these social services attract immigrants and encourage them to stay, a basic 104 INTERNATIONAL MIGRATION REVIEW premise in much of the contemporary debate over immigration reform (Chavez, 1996). Limitations of the study are those inherent in phone survey research. There is a bias toward nontransient respondents or, in this case, toward more settled immigrants. Even with this bias, however, we found significant differences between undocumented Latinas and other respondents. In addition, the study was not a study of immigration per se. Consequently, we were not able to collect a wide range of variables on the social conditions, social service use, and the migration and labor market experiences of the respondents. CONCLUSION Our findings generally confirm the characterizations of Latina immigrant behavior and intentions suggested by other studies, including those relying on nonrandom samples. For example, we found that undocumented Latinas in Orange County work at the margins of the labor market. They are more likely than other Latinas and Anglo women to be unemployed and looking for work. Their family incomes are the lowest among the respondents in this study. Their lack of medical insurance makes it difficult to establish a regular source of medical care. They face a number of obstacles to establishing a foothold in this relatively affluent county. And yet despite all these formidable obstacles, undocumented Latina immigrants often want to stay in the United States. The experiences of wage work, family and friends in the area, large concentrations of Latinos, a weakening of traditional patriarchy, social and economic opportunities, a hig her standard of living, and personal freedoms have all been suggested as possible reasons why immigrant Latinas desire to settle in the United States rather than return to their country of ori g in. All of these factors, perhaps, contributed to our finding that many Latina immigrants, even undocumented immigrants, feel a sense of community in the United States. It is important to emphasize that our findings underscore the diversity of Latinas, which often goes unrecognized by the public and policymakers. In particular, there are important differences between Latina immigrants and Latina citizens, most of whom were born in the United States. Latina citizens are much closer to Anglo women than to Latina immigrants based on socioeconomic characteristics such as education, income, and utilization of medical services. When discussing Latinas, it is best not to paint with too broad a brush, especially one that ascribes the behaviors and characteristics of immigrant Latinas to all Latinas. Finally, the debate over immigration and welfare reform has embedded in it a number of assumptions about immigrant behavior. One assumption receiving wide currency of late is that immigrants come to the United States and then stay because of social services. This study suggests, however, that COMPARATIVE ANALYSIS OF UNDOCUMENTED LATINA IMMIGRANTS 105 undocumented Latina immigrants in Orange County come to the United States and then stay because of work, family, and community. Moreover, our findings are consistent with the literature that suggests undocumented immigrants are not a drain on society. The undocumented immigrants' contributions as low-income workers in Orange County s economy has yet to be determined. But that they work and raise families with the lowest incomes among the participants in this study is clear. Much of the public policy debate would have us believe that these undocumented Latinas and their families are the ones benefiting from living under these conditions and that the rest of us, particularly U.S.-born citizens, are being hurt by their presence. And yet, U.S. citizens, especially Anglo women and their families, had the highest incomes, best jobs, least unemployment, and greatest access to health care in the county. They also benefit from lowered costs for grocery prices, child care, and household services subsidized by undocumented workers wages, and increased time opportunities to pursue other interests and careers. Why is this so if undocumented immigrants are taking so much from citizens, as the anti-immigrant rhetoric contends? Is it possible that the presence of undocumented immigrants contributes positively to the middle-class and affluent lifestyles of citizens in Orange County? It is important that we sort out the relative strength of competing explanations for immigrant behavior and intentions, especially when they serve as the basis for public policies toward immigrants. Every indication is that immigration will continue to be an important phenomenon well into the twenty-first century We must be very careful that the public policies we make today lead to improving the integration of immigrants into society today and tomorrow. REFERENCES Alvarez, R. R. 1987 Familia: Migration and Adaptation in Baja andAlta California, 1800-1975. Berkeley: University of California Press. Anderson, B. 1983 Imagined Communities. London: Verso. Basch, L., N. Glick Schiller and C. Szanton Blanc 1994 Nations Unbound: TransnationalProjects, Postcolonial Predicaments, and Deterritorialized Nation-States. Amsterdam: Gordon and Breach. Borjas, G. J. and M. Tienda 1993 " The Employment and Wages of Legalized Immigrants," International Migration Review, 27:712-747. California State Census Data Center 1995 " 1990 Census of the Population and Housing," Summary Tape File 4 [Source: U.S. Bureau of the Censusl. Sacramento: California State Census Data Center. 106 INTERNATIONAL MIGRATION REVIEW Chavez, L. R. 1997 "Immigration Reform and Nativism: The Nationalist Resonse to the Transnationalist Challenge. " In Immigrants Out!: The New Nativism and the Anti-Immigrant Impulse in the United States. Ed. Juan Perea. New^brk: New ^brk University Press. 1994 "The Power of the Imagined Community: The Settlement of Undocumented Mexicans and Central Americans in the United States,"American Anthropologist, 96:52—73. 1992 Shadowed Lives: Undocumented Immigrants in American Society. Fort Worth, TX: Harcourt, Brace and Jovanovich College Publishers. 1991 "Outside the Imagined Community; Undocumented Settlers and Experiences of Incorporation," American Ethnologist, 18:257-278. 1990 "Coresidence and Resistance: Strategies for Survival among Undocumented Mexicans and Central Americans in the United States," Urban Anthropology, 19:31—61. 1988 "Settlers and Sojourners: The Case of Mexicans in the United States," Human Organization, 47:95-108. 1985 "Households, Migration, and Labor Market Participation: The Adaptation of Mexicans to Life in the United States," Urban Anthropology 14:301-346. 1984 "Doctors, Curanderos, and Brujas: Health Care Delivery and Mexican Immigration in San Diego," Medical Anthropology Quarterly, 15:31-37. Chavez, L. R., and V. M. Torres 1994 "The Political Economy of Latino Health." In The Anthropology of Hispanic Groups in the United States. Ed. Tomas Weaver. Houston, TX: Arte Publico Press 1994. Chavez, L. R., E. T. Flores and M. Lopez-Garza 1990 "Here Today, Gone Tomorrow? Undocumented Settlers and Immigration Reform," Human Organization, 49:193-205. Cornelius, W A. 1982 "Interviewing Undocumented Immigrants: Methodological Reflections Based on Fieldwork in Mexico and the United States," International Migration Review, 16:378-411. Donato, K. M. 1993 "Current Trends and Patterns of Female Migration: Evidence from Mexico," International Migration Review, 27:748-771. Hagan, J. M. 1994 Deciding to Be Legal: A Maya Community in Houston. Philadelphia, PA: Temple University Press. Hispanic Development Council 1989 "Impact of IRCA: Immigration Reform and Control Act, the Orange County Experience." Report on Puolic Hearings co-sponsored by the Hispanic Development Council of the United Way and the Orange County Human Relations Commission. Santa Ana, CA: Orange County Human Relations Commission. Hondagneu-Sotelo 1994 Gendered Transitions: Mexican Experiences of Immigration. Berkeley: University of California Press. COMPARATIVE ANALYSIS OF UNDOCUMENTED LATINA IMMIGRANTS 107 Hubbell, F. A., L. R. Chavez, S. I. Mishra, J. R. Magana and R. Burciaga Valdez 1995 "From Ethnography to Intervention: Developing a Breast Cancer Control Program for Latinas," Journal of the National Cancer Institute Monographs ^ No. 18:109-115 (refereed). Lamphere, L., ed. 1992 Structuring Diversity: Ethnographic Perspectives on the New Immigration, Chicago, IL: University of Chicago Press. Marin, G., R Sabogal, B. V. Marin, R. Otero-Sabogal and E. J. Perez-Stable 1987 "Development of a Short Acculturation Scale for Hispanics," Hispanic Journal oj Behavioral Science, 9:183-205. Marmora, L. 1988 "Social Integration and Employment of Central American Refugees," In When Borders Dont Divide. Ed. Patricia Pessar. New York: Center for Migration Studies. Pp. 142-155. Martinez, G. and P.J. McDonnell 1994 "Prop. 187 Forces Rely on Message - Not Strategy," Los Angeles Times, October 30, Al. Massey, D., S. R. Alarcon, J. Durand and H. Gonzalez 1987 Return to Aztlan, Berkeley. University of California Press. McDonnell, P J. 1994 "Prop. 187 Heats Up Debate Over Immigration," LosAngeles Times, August 10, Al. National Research Council 1985 Immigration Statistics: A Story of Neglect. Washington, DC: National Academy Press. Ranney, S. and S. A. Kossoudji 1984 "The Labor Experience of Female Migrants: The Case of Temporary Mexican Migration to the U.S," International Migration Review, 18:1120-1143. Reichert, J. S. and D. S. Massey 1979 "Patterns of U.S. Migration from a Mexican Sending Community: A Comparison of Legal and Illegal Migrants," International Migration Review, 13:559—623. Rouse, Roger 1991 "Mexican Migration and the Social Space of Postmodernism," Diaspora 1:8-23. Simon, R. J. and C. B. Brettell 1986 International Migration: The Female Experience. Tbtowa, NJ: Rowman and Allanheld. Simon, R. J. and M. DeLey 1984 "Undocumented Mexican Women in Los Angeles," International Migration Review, 18:1212-1229. Survey Sampling, Inc. 1990 Statistical Analysis of Sample. Fail-field, CT: Survey Sampling, Inc. U.S. Bureau of the Census 1991 Race and Hispanic Origin, 1990 Census Profile. Washington, DC: U.S. Department of Commerce. Valdez, R.B., J. DaVanzo, G. Vemez and M. Wade 1993 "Immigration: Getting the Facts." Rand Issue Paper, Program for Research on Immigration Policy. Santa Monica, CA: RAND. The Vietnamese Double Marriage Squeeze 1 According to Guttentag and Secord (1983), the relative balance of power between the sexes is determined by dynamic interactions of dyadic and structural power. Dyadic power accrues to whichever sex is relatively rare owing to the larger proportion of potential relationships available to it, the demographic dimensions of which are commonly known as a marriage squeeze. Structural power refers to control over economic, political or legal resources: men or women, according to the thesis, attempt to overcome existing deficits in dyadic power by gaining such resources. The unique dual case study presented here illuminates the state political institutions and other contextual conditions under which both dyadically disadvantaged men and women have been unable to garner such structural resources. Young women in Vietnam during the 1970s and 1980s faced a severe deficit of male partners due to population growth, war, and excess male migration. At the other end of the Vietnamese diaspora, overseas Vietnamese men during the 1980s and 1990s have faced an even greater shortage of Vietnamese women. In each area, the sex in surplus has not only been forced to delay or forego marriage, but has also lost structural power. Women ' s advocates in Vietnam have been weakened in the postreunification era due in part to the implementation of free market reforms in a nondemocratic political context. Overseas men have been disadvantaged due to a more equitable Western social and legal climate that has eroded their former advantage. Gender imbalances in the number of young adults have implications for marriage markets as well as the relative power of men and women. Guttentag and Secord (1983), in an influential work, posited that when one sex is in relative surplus, the rarer sex has an advantage in what they call dyadic power due to the larger potential pool of partners available to it. The common result in Libertina and Eros, the mythical lands described by Guttentag and Secord where either men or women l This research grew out of the author's field experiences in Vietnam throughout 1993 and several return visits since then. That research was undertaken during a fellowship in the Department of Demography at The Australian National University and was graciously facilitated by Tuong Lai and Pham Bich San at the Institute of Sociology (National Center for Social Sciences and the Humanities, Hanoi). Gavin Jones, Terence Hull and David Marr provided logistical support, andWendyCosford provided editorial assistance. This manuscript has benefited from comments by Gordon Carmichael, Mariah Evans, Fr ank Jones, Tuan Mclntyre, Ronald Rindfuss, Mandy Thomas and Cristabel Young. More recently, three anonymous reviewers provided useful suggestions. This research has been supported by the Andrew W. Mellon Foundation. ©1997 by the Center for Migration Studies of New York. All rights reserved. 0197-9183/97/3101.0117 Daniel Goodkind University of Michigan THE VIETNAMESE DOUBLE MARRIAGE SQUEEZE 109 are in short supply, is that the surplus group must either many later or remain single. The demographic dimension of this gender imbalance is commonly known as a marriage squeeze (e.g., Akers, 1967; Heer and Grossbard-Schechtman, 1981; Schoen, 1983). However, imbalances in sex ratios among young adults may not automatically determine the overall distribution of gender power. In addition to dyadic power, Guttentag and Secord describe what they call structural power. This latter kind of power stems from economic, political and legal resources, and it can be employed to bend social customs bearing on gender roles and relations. Unlike dyadic power, which in theory should be determined solely by sex ratios in the available mate pool, structural power is presumed to be enjoyed by men in most societies. A critical question concerns how the two kinds of power interact. Guttentag and Secord posit that when the dyadic advantage of one sex increases, the other will attempt to counteract that through the garnering of structural resources. For instance, when womens dyadic advantage increases, men may devise laws or customs that sequester women into traditional roles, lest the latter become more selective about which husband to marry or which lifecourse option to pursue. Conversely, when mens dyadic power increases, womens position may be initially perilous owing to their lesser structural power; however, womens dyadic disadvantage can precipitate other social changes that eventually result in greater (even if not total) equality. For example, the feminist movement in the United States during the 1960s and 1970s may have resulted in part from womens dyadic disadvantage, a factor which fomented their greater participation in the labor force and clout for demanding other legislative changes (Heer and Grossbard-Schectman, 1981; Guttentag and Secord, 1983). Empirical evidence suggests that unbalanced sex ratios may indeed affect family formation behavior and gender roles. South and Trent (1988), for instance, in the most comprehensive cross-national test to date, found that women in societies with a surplus of women had less traditional roles, as Guttentag and Secord predicted; the women were more likely to be literate, marry late, have fewer children, and remain unmarried. However, the exact causal mechanisms responsible for these relationships were not completely elucidated. For instance, broader issues of political economy and state politics were conspicuously neglected (and were also absent from Guttentag and Secord's original framework). The only variable in South and Trent (1988) bearing any relation to these issues was a development index. It suggested that sex ratio imbalances had more significant social consequences in developed nations than developing ones. The authors speculated that this latter finding, rather than having substantive significance, might simply be attributable to greater intranational variation in sex ratios in developing countries. But a potential alternative explanation is that dyadically disadvantaged women can better garner structural resources in more developed societies because those 110 INTERNATIONAL MIGRATION REVIEW societies tend to have more liberal democratic institutions throughwhich women can better wield political clout. This alternative theory opens out upon an issue that remains as grand as it is unresolved; that is, whether the institutions that tend toaccompanyliberaldemocraciesorrevolutionarysocialistsocietiesofferwomen better opportunities for equality over the course of development (e.g., Boserup, 1970; Kruks, Rapp and Young, 1989; Molyneux, 1990; Rueschemeyer, 1994). The relevance of political economy for the issues raised by Guttentag and Secord is suggested by the dual case study presented here of Vietnamese at home and abroad. It is unique because inverse gender imbalances exist simultaneously (and can be measured) among the same ethnic group across national boundaries, a factor that inhibits rebalancing of the marriage market. Such simultaneity is important; those studies claiming a causal role of changing sex ratios on social outcomes within a single area or ethnic group may confound the influence of dyadic imbalances with other historical or structural factors (Carmichael, 1988). The configurations of this dual case study permits systematic and rigorous comparative measurements across space and time while allowing for some breadth of contextual detail to be explored. I will call this phenomenon the Vietnamese double marriage squeeze, a manifestation of inverse losses in dyadic power at both ends of the Vietnamese diaspora. In the 1970s and 1980s, because of rapid population growth, war, and excess male emigration since the war of reunification, women in Vietnam have faced one of the strongest national deficits of male partners in the world. Conversely, because of emigration patterns, young male overseas Vietnamese (VietfGeu) have faced an even greater shortage of Vietnamese women. Kibria (1993) was among the first to bring this gender power inversion to li ght through interviews with Vietnamese Americans in South Philadelphia (see also Gordon, 1989; Rumbaut, 1989; Coughlan, 1990; Gold, 1992). This paper documents evidence of the Vietnamese double marriage squeeze more formally on the basis of censuses and vital registration. Limitations in Vietnamese data preclude the calculation of the most sophisticated measures of mate availability over rime (Muhsam, 1974; Schoen, 1983; Goldman, Westoffand Hammers, 1984; Fossett and Kiecolt, 1991; Lichter, LeClere and McLaughlin, 1991), but simple measures of gender imbalances among younger people will nevertheless reveal the cohorts most severely affected by marriage squeeze. The authors observations in Vietnam throughout 1993 (and during several return visits since then) further corroborate the social consequences of gender imbalances. The twist in the Vietnamese case is that the dyadically disadvantaged group in each area, in contrast to the expectations of Guttentag and Secord, has been blocked from gaining structural power as a result of broad political and other contextual constraints. Women in Vietnam have been disadvantaged in a postwar environment characterized first by severe resources constraints and then by the state' s recent embrace of free market economics. The absence of a THE VIETNAMESE DOUBLE MARRIAGE SQUEEZE 111 similar liberalization of political par ticipation has weakened the Vietnam Women ' s Union, the major organization allowed to speak for women ' s concerns under Vietnam ' s single-party rule. Conversely, Vietnamese males emigrating to the West have left this relatively favorable environment, vis a vis women, for one in which they have experienced pronounced downward mobility. Men ' s new-found high rates of unemployment, relative poverty, and loss of patriarchal advantages in the Western political and legal context have eroded, although not completely eliminated, their ability to garner resources. Such changes have reduced men ' s attractiveness as mates and limited alternate options of outmarriage with a non-Vietnamese. CAUSES OF GENDER IMBALANCE AND MARRIAGE SQUEEZE A marriage squeeze against women is common wherever population growth combines with norms requiring women to marry earlier than men: as each successive birth cohort of females grows, they must seek a mate from the smaller pool of older males born just before them. In Vietnam, men indeed marry later than women, and rapid population growth accelerated sharply in the North during the late 1950s and early 1960s (Goodkind, 1995b). Thus, the birth cohorts contributing to that growth reached peak marital ages in the 1970s and 1980s. In Vietnam, however, two other unique factors further reduced the ratio of men to women. The first is excess male mortality during the war of reunification (1965—1975)? which was the sixth most severe of all wars since 1816 in terms of per capita battle deaths to the home country (Small and Singer, 1982:102, 238). Among the 1.2 million Vietnamese battle deaths, the vast majority were males. The effects of excess male mortality have lingered in the marriage market; an 18-20 year old male casualty at the time of Vietnam ' s reunification in 1975 would have been 32-34 years old at the time of its 1989 census. A second explanation for the gender imbalance concerns the excess of male emigrants in the Vietnamese diaspora after the war of reunification. The reasons for the sex differentials in migration include the greater mobility of males in the wake of Vietnam ' s social and economic turmoil during the late 1970s (Freeman, 1989; Haines, 1989), as well as Vietnamese teenage males attempting to avoid being drafted (Kibria, 1993). In the decade and a half since reunification, well over a million and a half Vietnamese left the country, and currently some 2 million live abroad. The resulting inverse gender imbalances at home and abroad around 1990 are depicted in Figure 1, which shows the ratio of all men to all women within five-year age groups in Vietnam and among Vietnamese residing in the United States and Australia. Between the ages of 20 and 34 years, women in Vietnam exceeded their male counterparts by about 10 percent; Vietnamese men residing abroad exceeded their female counterparts by 20 to 30 percent. 112 INTERNATIONAL MIGRATION REVIEW Figure 1. The Vietnamese Double Marriage Squeeze: Sex Ratios (M/F) in Vietnam and among Vietnamese Abroad 140 ^————————————————————————-----------———.................................................................................. 1.30 1 20 1.10 1 00 0.90 15-19 20-24 25-29 30-34 35-39 40-44 45-49 Sources: Census of Vietnam, 1989; United States, 1990; Australia, 1991. Women in Vietnam: The Marriage Squeeze Against Them The severity of the gender imbalance in Vietnam is best appreciated when placed in an international perspective. Table 1 depicts sex ratios at peak marrying ages for the fifteen most populous countries in the world; in 1989 Vietnam had more than 64 million people and was the thirteenth most populous. In 1979, Vietnamese sex ratios among the young were far lower than anywhere else in the world. There was a 10-12 percent shortfall of men relative to women in the age group 20-34 years. Men in this cohort were born between 1945 and 1959 and were of primary draft age during the war of reunification. By 1989, this shortfall of men had declined somewhat, which augurs an eventual end to marriage squeeze conditions, although this deficit was still stronger than in any other country. Table 1 also indicates gender ratios between the number of men aged 25-29 and women aged 20-24, a measure which provides a better idea of marriage squeeze conditions in countries with wider gender differentials in marital ages. This measure also implies severe marriage squeeze conditions, particularly in 1979. THE VIETNAMESE DOUBLE MARRIAGE SQUEEZE 113 TABLE 1 SEX RATIOS AT PEAK MARRYING AGES IN 15 MOST POPULOUS NATIONS: VIETNAM, 1979 AND 1989; ALL OTHERS, 1985 S€xRado(M/F) Populadol^ M25-29/ Country(Millions)20-2425-2930-34F20-24 Vietnam 79 53 0.88 0.88 0.90 0.67 Vietnam'89 64 0.92 0.91 0.91 0.86 Philippines 62 0.95 0.93 0.99 0.81 Nigeria 108 0.99 0.97 0.97 0.80 Indonesia 184 1.00 0.99 0.98 0.88 Brazil 150 1.01 1.00 0.99 0.86 Mexico 89 1.01 1.00 0.98 0.82 USA 249 1.02 1.01 1.00 1.05 Japan 123 1.03 1.02 1.01 0.98 USSR-x 289 1.05 1.04 0.99 1.08 Bangladesh 116 1.05 1.06 1.05 0.88 Germany 77 1.06 1.07 1.04 0.94 Korea (ROK) 61 1.05 1.00 1.06 0.97 China 1,139 1.08 1.08 1.06 0.86 India 853 1.09 1.10 1.10 0.96 Pakistan123UOUlU10.93 Sources: Dan So VietNam, 1-10-1979 (Population of Vietnam, 1-10-1979), 1983. Hanoi: Census Steering Committee {Ban Chi Dao TongDieu Tm Dan So Trung Ifong); Vietnam Population Census, 1989, 1991. Completed Census Results: Volume 1. Hanoi: Central Census Steering Committee; United Nations, 1990. Patterns of First Marriage. ST/ESA/SER.R/ 111. New York. ^A\[ populations as of 1992, except for Vietnam. Vietnam ' s gender imbalance has affected relative marital opportunities for men and women. According to the 1989 census, the singulate mean age at marriage (SMAM) was 24.5 for men and 23.2 for women. (SMAM is a census-derived life table measure indicating the number of years one would expect to remain single (United Nations, 1990). )The 1989 gender differential in SMAM of 1.3 years was exceptionally small. Only three countries in the world exhibited a lower SMAM differential than Vietnam (Ireland, Brunei, and Jamaica) (United Nations, 1990). Unfortunately, Vietnam ' s 1979 census did not include questions about marital status, so we cannot determine the extent to which the small SMAM differential was due to a rising SMAM for women rather than a falling SMAM for men. But a variety of cross-national comparisons (not shown) suggest that the small SMAM differential was due primarily to a late SMAM for women. In Thailand, for instance, a nearby country which has not experienced a pronounced gender imbalance, SMAM for men in 1980 was almost identical to that for Vietnam in 1990, yet the SMAM differential there was 1.8 years. Thus, if Thailand is used as a standard, the gender imbalance in Vietnam likely reduced the SMAM differential by about half a year. 114 INTERNATIONAL MIGRATION REVIEW The marriage squeeze was strongest against the female cohort aged 30—44 in 1989 (^., those aged 15-29 in 1974, the year before the fall of Saigon). Table 2 shows the sex ratios in Vietnam and Thailand by five-year age groups and the corresponding propor tions of men and women never-married. The comparison in the bottom section indicates large differences in proportions never-married between the two countries. The age pattern of the ratio of never-married men in Vietnam to never-married men in Thailand reveals an increasing marital advantage among Vietnamese men at progressively older ages (from 1.010 to 0.349). However, while the proportions of never-married women between Vietnam and Thailand also fall in similar fashion between ages 15—19 and 25—29, they then increase for the 30-44 year old age group, indicating a severe relative disadvantage for Vietnamese women. Of course, since Table 2 depicts only the never-married population, it underestimates the extent of the marriage squeeze on women, which is far worse among the currently single population (including the divorced, separated, and widowed). The marriage squeeze is par ticularly problematic for women in Vietnam due to its strong Confiician heritage (Keyes, 1977), under which nonmarriage has traditionally been both rare and stigmatized. Vietnamese social scientists are well aware of the social consequences resulting from the gender imbalance. Field studies indicate that some single women in their 30s who see little prospect of finding a husband are deciding to have children, especially those recruited during the war of reunification to work in female brigades of the forestry industry (Nguyen, 1992; Nguyen, 1993). Unmarried women are said to " ask for a child" from men, with arrangements made to pay a fee depending on the sex of the resulting child (Le, 1994) - yet lingering discriminatory treatment occurs against women who make " the mistake of having a child without having a husband" (Hiebert, 1994). In addition, the surplus of women has facilitated mens traditional option of remarriage after divorce or widowhood (an option that women generally have not had) and perhaps even rekindled traditions of polygamy and concubinage. (For a case study of this tradition, see Luong, 1992.) Men are also better able to recycle themselves through the less formal par tner market. The authors own field observations confirm these and other associated consequences ofVietnarns marriage squeeze. Appearances of the stereotypical romancestarved war widow in popular films cause murmurs of recognition among Vietnamese theatergoers. Vietnamese males praise the virtues ofVietnamese wives, who are inevitably described as mt chung thuy (very faithful), a characteristic that may be reinforced by the gpnder imbalance. Informal matchmaking has become for many a par t-time profession because singlehood amongwomen over 30 is still a matter for public concern and disapproval; when young men or women arc introduced to strangers, polite questions about their marital status often result in offers to make a match for those who arc unattached. THE VIETNAMESE DOUBLE MARRIAGE SQUEEZE 115 TABLE 2 COMPARISON OF SEX RATIOS AND NEVER-MARRIED PROPORTIONS IN THAILAND (1980) AND VIETNAM (1989), BY AGE AND SEX Sex RatioThailandVietnam 15-19 ^994 ^973 20-24 .981 .920 25-29 .962 .911 30-34 .976 .914 35-39 .981 .876 40-44 .965 .865 45-49.959.815 Proportion Never Married MaleFemaleMaleFemale 15-19 .947 .825 .953 .885 20-24 .657 .432 .623 .424 25-29 .269 .208 .231 .177 30-34 .114 .117 .075 .111 35-39 .061 .073 .033 .085 40-44 .043 .053 .019 .059 45-49 .034 .041 .013 .033 15-49 3 .426.333.401.339 Intercountry Ratio (Vietnam/Thailand) of Proportions Never Married Male Female 15-19 L0101.070 20-24 .948 .981 25-29 .859 .850 30-34 .657 .942 35-39 .542 1.163 40-44 .449 1.107 45-49.349.807 Sources: Thailand Census, 1980; Vietnam Census, 1989 (see Table 1). "Standardized to age distribution of Vietnam, 1989. Women in Vietnam: Concurrent Loss of Structural Power Contrary to the Guttentag and Secord model, Vietnamese women ' s loss of dyadic power in the 1970s and 1980s was not counterbalanced by an increase in structural power, such as that posited to have occurred among U.S. females during the 1960s and 1970s (Heer and Grossbard-Schectman, 1981). The reasons include the postwar aftermath which was characterized by a shortage of resources and then a subsequent transition away from socialism towards the free market. This section demonstrates that these factors, on balance, have weakened groups advocating women' s concerns and fostered traditional roles for women. To understand current changes in gender equality in Vietnam requires a brief review of recent Vietnamese history. Following Vietnams political par tition into northern and southern halves in 1954, the revolutionary socialist regime in the nor th attempted to raise the status of women. There were several reasons for this 116 INTERNATIONAL MIGRATION REVIEW policy. First, in striving for social equity, the regime attempted to fulfill its revolutionary socialist vision; for instance, the Marriage and Family Law of I960 called for a ban on forced marriage, child betrothal, concubinage, wife beating, and other customs and practices deemed to be "feudal" (Mat and Le, 1978). Second, the emancipation of women from household roles fit into the socialist strategy of industrial development, which required the extraction of a surplus from a collectivized agricultural sector. Lastly, by encouraging the activities of the Women ' s Union, a mass organization officially established in the 1940s to rally womens support for government decrees, the Communist party strengthened and extended its own base of political power. The mobilization of women took on a special urgency in the mid-1960s at the onset of the war of reunification, when there was near universal draft of men. The Womens Union called upon women to fulfill " three responsibilities " (Mat and Le, 1978) which included encouraging their husbands, sons and brothers to go to war, taking responsibility for households affairs, and assuming a greater role in industrial and agricultural production. Despite such exhortations, womens contributions to the war effort were still considered insufficient at the time of major American troop buildups in DaNang during 1965. The Central Committee of the Communist Party responded in 1967 by calling for formal quotas in employment (V^mer, 1981); women were to fill at least 35 percent of all jobs, 50—70 percent in education, medicine, and light industry (Mai and Le, 1978). Official statistics suggest the attainment of these goals by the early 1970s. Womens participation increased not only in politics and industry but also in agriculture, where women in certain areas constituted 80 percent of the workforce (Wemer, 1981). However, shor tly after Vietnams political reunification in 1975, the welfare of Vietnamese women deteriorated rapidly, with the " promissory notes" implicitly offered to women remaining unpaid (Pelzer, 1989). With the elimination of formal quotas, womens representation in the National Assembly fell from a peak of 32 percentin 1971, to 27 percent in 1981, and to 18percentin 1987 and 1992 (Alien, 1992, 1993), and this defeminization of politics was mirrored by a rise in occupational segregation along more traditional gender lines (Goodkind, 1995a). One reason for increasing gender segregation was the postwar return of male soldiers who reclaimed their former jobs. Postreunification poverty amidst continuing military conflicts with Kampuchea and China further mitigated womens position, as scarce government resources were diverted away from the Womens Union, the activities of which were no longer considered so crucial to national survival. Equally significant was the governments official shift away from socialist economic principles and ideals beginning in the mid 1980s, its own version of perestroika. The new set of free market-oriented policies, collectively known as Doi Moi, have loosened price controls, ended state subsidies to local cooperatives, and otherwise allowed households to become the primary organizers of agricultural production (Turley and Seldon, 1993). Although these reforms have contributed THE VIETNAMESE DOUBLE MARRIAGE SQUEEZE 117 to a boom in agricultural production, they have also contributed to a vulnerability in the womens position not unlike those described in other postsocialist societies (Molyneux, 1990; Rueschemeyer, 1994). For example, £WAf<whas disadvantaged rural female-headed households in gaining access to new forms of credit (Tran, 1992); ethnographic observations suggest that female-headed households may have been more likely to descend into poverty since the free market reforms (Ba, 1992) (For a recent contrasting view, ^ UNDI? 1995.) Of course, the Womens Union has recently scored some well-publicized successes. The 1986 Family Law increased maternity leave with pay from three to six months for wage-earning workers and addressed discrimination treatment against unwed mothers and their children (Le, 1993). The 1988 Council of Ministers ' Decision No. 163 g^ve the Womens Union the rig ht to be involved in any discussions or p lans that have ramifications for the welfare of women or children (Alien, 1992). However, although such decrees have increased the clout of the Womens Union in theory, their position has been compromised due to the emphasis on economic efficiency under the new free market calculus. For instance, maternity leave was rolled back from six months to "four to six " in 1995 Ar ticle 114 (Ministry of Labor, Invalids and Social Affairs, 1995:195) because some employers found womens labor more expensive. In addition, aside from the Womens Union, there are no other organizations ready and permitted to serve as women ' s advocates, because Vietnamese perestroika has not been accompanied by an associated g lasnost. Since Chinas Tian An Men incident in 1989, Vietnam has been especially cautious about permitting social liberalization and the growth of civil organizations outside the governments control. Like its counterpart organization in China (Rosen, 1995), the Vietnamese Womens Union has been caught in the role of loyal opposition. Since its inception in the 1940s, its role has been to mobilize womens support for the Communist party agenda, and the current free market reforms arc par t of that agenda. But these reforms have partially eroded equalities of opportunity Thus, although Vietnamese womens loss in dyadic power, in theory, could have g iven an impetus to feminism, women ' s ability to gain structural power has been thwarted in a postwar milieu of severe poverty as well as more recent free market reforms. Furthermore, unlike in the U.S. context following World War II (Heer and Grossbard-Schechtman, 1981), the authoritarian single-party political system remaining in postsocialist Vietnam has precluded the germination of other nongovernmental organizations outside the Womens Union that might substitute as womens advocates. Overseas Vietnamese Men: An Even Worse Marriage Squeeze At the other end of the Vietnamese diaspora, in contrast to women' s palpable loss of dyadic and structural power in Vietnam, there has been an even more dramatic loss in such power among the excess of Vietnamese men migrating overseas. As 118 INTERNATIONAL MIGRATION REVIEW Figure 1 demonstrated, Vietnamese males living overseas around 1990 exceeded their female counterparts by 20-30 percent or more between the ages of 20 and 34 years, which is more than twice the surplus of females within Vietnam. This section focuses upon Vietnamese in Australia, rather than in the United States, for a variety of reasons. First, Australian censuses are conducted every five years, which provides a tighter interval for examining the most recent trends in marital patterns. Second, Australia ' s marriage registration records are more versatile and cover the entire nation. Australia ' s Vietnamese population is also more geographically concentrated than in the United States. Almost 75 percent of Vietnamese in Australia live in either the Sydney or Melbourne metropolitan areas and socioeconomic characteristics of Vietnamese in these and other areas are almost identical (BIPR, 1994a). Thus, Australia ' s national data on marriage patterns are more likely than that in the United States to encompass a marriage market that is national in scope. In most other respects, however, social conditions and problems of Vietnamese in Australia and the United States have been fairly congruent (see Viviani, 1984; Krupinski and Borrows, 1986; Tran and Holton, 1991; Coughlan, 1991, 1992). Furthermore, temporal patterns of emigration from Vietnam to Australia and the United States have been similar since 1976, with Vietnamese populations more than tripling in both countries since the early 1980s. In Australia, sharp gender imbalances at peak marital ages have been associated with large and increasing proportions ofVietnamese males never marrying. According to Table 3, the proportion of males at ages 15—49 who were never married increased from .558 to .642 between 1986 and 1991. (The age-specific proportions on which these figures arc based have been standardized according to the age distribution of Vietnam ' s 1989 population to facilitate comparison with Table 2.) This sharp increase in never-married males is understandable when one considers the sex ratio among the never married in 1986 (bottom, Table 3), when single males exceeded females by a ratio of more than 2 to 1 at ages 20-24 and 2.5 to 1 at ages 25—29. Vietnamese men ' s never-married propor tions were significantly higher than forAustralian-bom men, and the differential in married propor tions between men and women was apparently hig her than for all other immigrant groups in Australia (not shown). But Vietnamese women, despite their favorable position in the marriage market, have also married later than Australian-bom women; their never-married propor tions increased from .442 to .484 between 1986 and 1991, and the proportions of never-married women increased the most at the primary marital ages between 20 and 29. These increases in singleness cannot be due to selection processes in which unmarried migrants have become more prominent over time. In fact, the age-standardized never-married proportions of both Vietnamese men and women in 1991 were marginally higher among the group arriving before 1986 than those arriving after 1986 (not shown). Of course, that is not to deny the presence of heterogeneity and selection processes. For instance, the gender imbalance against overseas Vietnamese men of non-Chinese origin has been far THE VIETNAMESE DOUBLE MARRIAGE SQUEEZE 119 TABLE 3 COMPARISON OF NEVER-MARRIED PROPORTIONS IN AUSTRALIA AMONG VIETNAM-BORN AND NATIVE AUSTRALIAN-BORN, 1986 AND 1991, AND CORRESPONDING SEX RATIOS IN AUSTRALIA, 1986 AND VIETNAM, 1989 Australian-BornVietnam-Born 1986 Proportion Never MarriedMaleFemaleMaleFemale 15-19 .995 .980 .993 .969 20-24 .839 .652 .840 .639 25-29 .451 .283 .551 .284 30-34 .221 .131 .258 .175 35-39 .127 .073 .127 .136 40-44 .095 .056 .076 .116 45-49 .087 .050 .066 .071 15-49' 1 .537.427.558.442 1991 Proportion Never Married Male Female Male Female 15-19388^976LOOO ^952 20-24 .876 .754 .910 .742 25-29 .546 .360 .730 .479 30-34 .290 .185 .380 .176 35-39 .170 .107 .182 .082 40-44 .099 .064 .224 .119 45-49 .073 .045 .071 •000 11 15-49' 1 .575.473.642.484 Australian-Born Vietnam-Born Vietnam Sex Ratio of Number Never Married198619861989 15-19 1.058 1.482 1.048 20-24 1.318 2.066 1.352 25-29 1.595 2.600 1.189 30-34 1.689 1.612 .632 35-39 1.739 1.084 .340 40-44 1.701 .900 .286 45-491.6911.195.337 Sources: Australian Census, 1986; Australian Census, 1991 (1 percent sample); Vietnam Census, 1989 (see Table 1). 'All standardized to Vietnam's 1989 age distribution. Based on less than 25 cases. more severe than that against Vietnamese men of Chinese ancestry; the former exceeded their female counterparts by 50 percent at ages 15—29, compared to only 19 percent for the latter. (Chinese Vietnamese constituted about 27% of Vietnamese in Australia in 1991, compared to less than 2% in Vietnam; for economic perspectives on differences between Vietnamese of Chinese and non-Chinese origin in Australia, seeEIR, 1990; for the United States, see Gold, 1994.) Overseas Vietnamese Men: Concurrent Loss of Structural Power The Guttentag and Secord model predicts that the strong dyadic advantage among overseas Vietnamese women should facilitate their early marriage and otherwise encourage men to constrain women to traditional roles. However, Table 3 reveals 120 INTERNATIONAL MIGRATION REVIEW that Vietnamese women, like men, were also less likely between 1986 and 1991 to have ever married. One explanation for this finding is that Vietnamese women, realizing that they face a large surplus of men (Coughlan, 1990:21) and a social context in which late marriage is acceptable, can afford to wait and be selective. Despite an overwhelming surplus of young unmarried Vietnamese males (see Table 3), there are several reasons why Vietnamese women do not always perceive these men as attractive potential marriage partners. First, hig h unemployment or lower-status employment among overseas Vietnamese males, many of whom held high-status jobs or came from prestigious families in preunification South Vietnam (Starr and Roberts, 1982;LewinsandLy, 1985; Freeman, 1989; Dunning, 1989:61; Strand and Jones, 1985), reduces their attractiveness as mates as well as their ability to counteract womens newfound dyadic power through the garnering of other resources. Second, overseas Vietnamese language radio broadcasts and periodicals (reviewed by this author) broach the topic of " womens rights " (quyenphu nu), a notion quite distinct feom the " responsibilities " to society that the Women ' s Union in Vietnam continues to stress. Naturalization procedures normally require immigrants to be familiar with the historical background of such rights, which, at the very least, acquaint them with laws which may discourage wife beating and otherpractices more common in Vietnam (Kibria, 1993). Complaints by frustrated Vietnamese male suitors that "in Vietnam, the girls were more simple and honest " compared to the West where they " act royal" attest to this inversion of both dyadic and structural power (Kibria, 1993:112-121), which was unanticipated by most men and women at the time of emigration. Thus, ironically, many overseas Vietnamese men wait longer to marry because they do not have any other choice and Vietnamese women wait longer because they do. (For related perspectives on why womens marital rates may decline in response to the quality, as well as quantity, of the available mate pool, see Wilson, 1987 and Uchter, LeClerc and McLaughlin, 1991.) Given that Vietnamese Australians live in a multiethnic society, an expected outcome for the surplus of men would be a greater likelihood ofoutmarriage. Vital statistics reveal that this has indeed occurred. The proportion of Vietnamese men outmarrying has increased from 14.4 percent in 1984—1986 to over 20 percent in 1990-1992. This finding cannot be attributable to melting-pot theory, because outmarriage rates for Vietnamese women have decreased marginally over the same period, from 18.7 to 17.2 percent. Such divergent trends in outmarriage rates for men and women have been very rare among Australia ' s immigrant groups in the twentieth century (Price, 1989). Despite the surplus of Vietnamese men in Australia, inmarriage rates for Vietnamese men and women have remained higher than for other ethnic groups in Australia (ABS, 1990-1992). One explanation for such high rates of inmarriage is the recency of Vietnamese immigration and the problem of English language proficiency. According to the 1991 census, only 10.4 percent of Vietnamese men and 7.5 percent of Vietnamese women (aged 25-^4) said that they spoke English THE VIETNAMESE DOUBLE MARRIAGE SQUEEZE 121 " very well"; 44.3 percent of men and 33.5 percent said they spoke English " well" (BIPR, 1994a). Levels of English proficiency were even lower among Vietnamese above age 45 and for that reason may have increased the pressure on children to marry someone with whom their parents could communicate easily. The aforementioned issue of unemployment further limits outmarriage prospects. The Vietnamese unemployment rate was 36.1 percent for males and 45.1 percent for females in 1991, more than three times higher than that for native Australians and much hig her than that for other ethnic groups (BIPR, 1994a; seeaIsoTrm and Nguyen, 1994). Of course, Vietnamese mens slightly better employment conditions relative to women should not be construed as an improvement in their economic and social position since their arrival in Australia — since men were the primary breadwinners in Vietnam, their newfound high rates of unemployment (even if lower than for women) likely represent a growing dependency on women ' s economic contributions. Nevertheless, the above figures for women also suggest that their high rates of inmarriage may reflect vulnerability as well as power - women still face substantial challenges in garnering their economic resources. These constraints likely explain Kibrias (1993) observation that, despite increases in dyadic and other aspects of structural power, Vietnamese women still have a strong stake in keeping the traditional family unit intact. Because of the shortage of overseas Vietnamese females, their male counterparts use various strategies to maximize contacts with potential mates. One overseas Vietnamese informant notes that some Vietnamese men visit both Catholic churches and Buddhist temples on weekends to maximize potential social contacts. Another Viet Kieu reports that the parents of a recent groom were not very enthusiastic about his fiancee, but that they consented to the marriage because " there just wasn ' t any choice. " Given the dismal marriage market for overseas men, a potential solution is to return to Vietnam to seek a bride. This trans-Pacific marriage market, to the extent that it exists, 2 serendipitously also relieves the surplus of female partners at the other end of the Vietnamese double marriage squeeze. The overseas Vietnamese press regularly carries adver tisements for husbands seeking brides from Vietnam, some of which stress the " quality " {chet luong) of their services and offer guarantees that their brides have never married before, reflecting the concerns and suspicions of potential grooms. There are similar concerns among women in Vietnam about the quality and reliability of Viet Kieu men. Although a trans-Pacific marriage could represent an opportunity for upward social mo-^egrettably, the prevalence of trans-Pacific marriages cannot be estimated with Australian census data because they do not indicate year of marriage; similarly, vital registrations lack information on year of immigration. Even if such data were available, the extent to which trans-Pacific marriages represent arrangements by outside brokers, the family, or the couple themselves would remain unclear. Extensive in-depth interviews would likely yield more reliable insights into this issue. 122 INTERNATIONAL MIGRATION REVIEW bilityfor some women and their families, others complain that overseas men are too picky and often just want to ' shop around. " 3 CONCLUSIONS, LIMITATIONS AND FUTURE IMPLICATIONS This ar ticle has examined a particular social dynamic which links both sides of the Vietnamese diaspora. Large and inverse gender surpluses of women at home and men abroad have resulted in substantial marriage squeezes, a manifestation of the loss of dyadic power of each surplus group. The marriage squeeze in Vietnam against women was most severe in the 1970s and 1980s. Its counterpart against overseas Vietnamese males has been peaking in the 1980s and 1990s due to the excess of young male emigrants from Vietnam since the 1970s. In contrast to the expectations of Guttentag and Secord, each dyadically disadvantaged group has been unable to counter that disadvantage through the garnering of structural resources. These unexpected findings become comprehensible once we consider a school of thought neglected in the orig inal model -- political economy. Among the broader political and contextual factors that have influenced Vietnamese women' s (and men ' s) ability to access structural resources in the face of dyadic imbalances, the presence or absence of liberal democratic institutions is one common denominator. The potential salience of this hypothesis has been established by other research exposing the special bottlenecks to gender equality under socialism (Kruks, Rapp and Young, 1989; Rueschemeyer, 1994). Of course, the dual case study presented here is limited in at least three respects, all of which suggest avenues for future research. First, due in par t to space limitations, it has not determined the extent to which dyadic gpnder imbalances have affected the full range of structural power dimensions, which encompass a variety of attitudes, household work, and other social institutions (Chafetz, 1981). Second, although these par ticular cases suggest the role of democratic institutions in women ' s ability to access structural resources (in the case of a diaspora as well as more generally), that theory cannot be globalized without more stringent cross-national tests employing multivariate methods. Methodological challenges in performing these tests will begin with identifying comparable and reliable measures of democratic institutions {e.g., singlevs. multi-party systems, legal codes, and overall effectiveness of institutions advocating women ' s concerns) and structural power dimensions across a variety of social setting?. Once constructed, such variables may be added in to models to test their salience in determining gender roles and well-being. Third, questions Vietnam has heretofore made little attempt to counter this image. Government-approved movies typically portray Viet JGeu males as beer swillers, thugs, gamblers, and woman-pinchers, although slick Viet JGeu entertainment videos arriving from Paris have recently offered a less menacing image. For their part, the Viet Kieu community has mocked claims by Vietnam's government to have elevated women's position. See, e.g., the scathing short stories by Hoang» ^o Serve the Cause of Women's Liberation" (Hyunh, 1988:43-77) andNguyen, "In the Line of Socialist Duty" (Hyunh, 1988:78-94). THE VIETNAMESE DOUBLE MARRIAGE SQUEEZE 123 may be raised concerning causal ordering; this paper has treated dyadic imbalances (and other contextual factors, such as war and political economy) as largely exogenous, with the dynamics of structural power as an endogenous outcome. However, in certain circumstances, dyadic imbalances may themselves be determined to some extent by structural factors (Guttentag and Secord, 1983). Despite such challenges, the issue of sex ratio imbalances and their effect on gender well-being is likely to remain relevant, par ticularly in the Asian context. In the coming decades, for instance, China, Taiwan, Korea and eventually Vietnam will experience progressively stronger marriage squeezes against men. This phenomenon will be caused less by continued sex-selective migration than rapid declines in fertility, which will reduce available cohorts of women to slig htly older men. In addition, the persistence of male-child preferences interacting with prenatal sex-testing technologies in these low fertility societies has contributed to sex-selective abortion of females: sex ratios at birth, which normally amount to 104—106 male babies per 100 female babies, have exceeded 110 over the past several years in China, Korea and Taiwan, and are above 150 for later birth orders (Zeng et aL ^ 1993). Sex ratios at birth in Vietnam should also rise as these technologies become available there. In these societies, persistent structural power imbalances favoring men (manifested by male-child preferences) will, ironically, contribute to a future dyadic power to shift in favor of women (Park and Cho, 1995). It is unclear whether men, as part of a collective enterprise, will attempt to counteract this increase in womens dyadic power to constrain them to traditional roles. If the implications drawn from this dual case study are correct, men will be least able to do so in more democratic societies where social interests more openly contend. REFERENCES Akers, D. S. 1967 "On Measuring the Marriage Squeeze," Demograph y^ 4:907-924. Alien, S. 1993 "Health -The Current Situation and Recent Changes. " In Vietnam and the Rule of Law. Ed. C. Thayer and D. Marr. Canberra: Department of Political and Social Change, The Australian National University. 1992 Country Gender Analysis: Vietnam. Hanoi: Swedish International Development Authority (2505a). Australian Bureau of Statistics (ABS) 1984- Marriages Australia. Canberra. 1992 Ba, R V. 1992 Socio-economic Renovation in North Vietnam audits Effect on Fertility and Development in the Rural Plain. Report prepared for United Nations Fund for Population Activities. Hanoi: Institute of Sociology. 124 INTERNATIONAL MIGRATION REVIEW Bosemp, E. 1970 Womens Role in Economic Development. London: Alien and Unwin. Bureau of Immigration Research (BIR) 1990 Community Profiles: Vietnam Born, Canberra: Australian Government Publishing Service. Bureau of Immigration and Population Research (BIPR) 1994a Community Profiles, 1991 Census: Vietnam Born. Canberra: Australian Government Publishing Service. 1994b BIPR Bulletin No. 12. Australia. Carmichael, G. A. 1988 With This King: First Marriage Patterns, Trends and Prospects in Australia, Canberra: Australian National University. Chafetz, J. S. 1981 Sex and Advantage: A Comparative, Macro-structural Theory of Sexual Stratification. Totowa, NJ: Rowman and Allanheld. Coughlan, J. E. 1992 "Patterns of Settlement in Australia of Indochinese Refugees." In Asians in Australia The Dynamics of Migration and Settlement. Ed. C. Inglis, S. Gunasekaran, G. Sullivan, and C. T. Wu. Singapore: Institute of Southeast Asian Studies. Pp. 73-115. 1991 "International Factors Influencing Australian Government Response to the Indochinese Refugee Problem," The Vietnam Generation, 3(2):84-97. 1990 The Fertility and Living Arrangements ofAustralias Three Indochinese-Bom Communities: 1976—86. Australia—Asia Paper No. 55. Nathan: Griffith University. Dunning, B. 1989 "Vietnamese in America: The Adaptation of the 1975-1979 Arrivals." In Refugees as Immigrants: Cambodians, Laotians, and Vietnamese in America. Ed. D. W Haines. Totowa, NJ: Rowman and Littlefield Publishers. Pp. 55-85. Fossett, M. A. and K. J. Kiecolt 1991 "A Methodological Review of the Sex Ratio: Alternatives for Comparative Research," Journal of Marriage and the Family, 53:941-957. Freeman, J. 1989 Hearts of Sorrow: Vietnamese-American Lives. Stanford, CA: Stanford University Press. Gold, S. 1994 "Chinese-Vietnamese Entrepreneurs in California." In The New Asian Immigration in Los Angeles and Global Restructuring. Ed. P Ong, E. Bonacich, and L. Cheng. Philadelphia, PA: Temple University Press. Pp. 196-226. Goldman, N., C. Westoffand C. Hammerslough 1984 "Demography of the Marriage Market in the United States," Population Index, 50:5-25. Goodkind, D. M. 1995a " Rising Gender Inequality in Vietnam since Reunification," Pacific Affairs, 68:342-359. 1995b "Vietnam's One-or-Two-Child Policy in Action," Population and Development Review " 22:85-115. THE VIETNAMESE DOUBLE MARRIAGE SQUEEZE 125 1994 "Vietnamese Marriage Rituals as Political History: 1948-1993. " Paper presented at the Annual Meetings of the American Sociological Association, Los Angeles. Gordon, L. 1989 "National Survey of Southeast Asian Refugees ; Methods, Findings, Issues." In Refugees as Immigrants: Cambodians ^ Laotians ^ and Vietnamese in America. Ed. D. W Haines. Totowa, NJ: Rowman and Littlefield Publishers. Pp. 24-39. Guttentag, M. and P.F. Secord 1983 Too Many Women? The Sex Ratio Question. Beverly Hills: Sage Publications. Haines, D. W, ed. 1989 Refugees as Immigrants: Cambodians, Laotians, and Vietnamese inAmerica. Totowa, NJ Rowman and Littlefield Publishers. Heer, D. M. and A. Grossbar-Schechtman 1981 "The Impact of the Female Marriage Squeeze and the Contraceptive Revolution in Sex Roles and theWomens Liberation Movement in the United States, 1960-1975," Journal of Marriage and the Family, 43:49-65. Hiebert, M. 1994 "Single Mothers: Women in Men-Short Vietnam Are Having Children Out of Wedlock," Far Eastern Economic Review, 24:60-61. February. Huynh, S. T, ed. 1988 To Be Made Over: Tales of Socialist Reeducation in Vietnam. New Haven, CT: Yale Center for International and Area Studies (co-published with the William Joiner Center, UMass/Boston). Keyes, C. F. 1977 The Golden Peninsula: Culture and Adaptation in Mainland Southeast Asia, New York: MacMillan. Kibria, N. 1993 Family Tightrope: The Changing Lives of Vietnamese Americans. Princeton, NJ: Princeton University Press. Kruks, S., R. Rapp and M. B. Young, eds. 1989 Promissory Notes: Women in the Transition to Socialism. New York: Monthly Review Press. Pp. 172-192. Krupinski, J. and G. Burrows, eds. 1986 The Price of Freedom: Young Indochinese Refugees inAustralia. Sydney: Pergamon Press. Le, T. N. T. 1993 "Demand for Children in An Hiep Village," Vietnamese Studies, 3(109):60-66. Le,N. 1994 "Asking for a Child Practice at An Hiep Commune," Vietnam Social Sciences, 1 (39). Hanoi: National Center for Social Sciences and the Humanities. Lewins, F. andJ. Ly 1985 The First Wave: The Settlement of Vietnam ' s First Refugees. Sydney: Alien and Unwin. Lichter, D. T, F. B. LeClere and D. K. McLaughlin 1991 "Local Marriage Market Conditions and the Marital Behavior of Black and White Women," American Journal ofSociology^ 96:843-867. Luong, V. H. 1992 Revolution in the Village: Tradition and Transformation in North Vietnam, 1925-1988. Honolulu: University of Hawaii Press. Mai . T.T.andT.N.T.Le 1978 Women in Vietnam. Hanoi: Foreign Languages Publishing House. 126 INTERNATIONAL MIGRATION REVIEW Ministry of Labor, Invalids and Social Affairs 1995 Labor Code of the Socialist Republic of Vietnam. Hanoi: The Gioi Publishers. Molyneux, M. 1990 "The Woman Question in the Age ofPerestroika," New Left Review, 183:23-49. Muhsam, H. V. 1974 "The Marriage Squeeze," Demography, 11:291-300. Nguyen, T. K. 1993 " The Single Mother," Vietnamese Studies, 3(109):44-59. Nguyen, T. T. 1992 "Remarks on the Women Living without Husband," Vietnam Social Sciences (Special Issue: Women in Vietnam) 4:76-80. Hanoi: National Center for Social Sciences. Park, C. B. and N. H. Cho 1995 "Consequences of Son Preference in Allow-Fertility Society. Implications of the Sex Ratio at Birth in Korea," Population and Development Review, 21:59-84. Pelzer, C. 1989 "Vietnam: War, Socialism, and the Politics of Gender Relations." In Promissory Notes' . Women in the Transition to Socialism. Ed. S. Kruks, R. Rapp, and M. B. Young. New York: Monthly Review Press. Pp. 172-192. Price, C. A. 1989 Ethnic Groups in Australia, Canberra: Office of Multicultural Affairs, Department of the Prime Minister and Cabinet. Rosen, S. 1995 "Women and Political Participation in China," Pacific Affairs, 68:315-341. Rueschemeyer, M., ed. 1994 Women in the Politics of Postcommunist Eastern Europe. Armonk, NY: M. E. Sharpe. Rumbaut, R. G. 1989 "Portraits, Patterns, and Predictors of the Refugee Adaptation Process: Results and Reflections from the IHARP Panel Study. " In Refugees as Immigrants: Cambodians, Laotians, and Vietnamese in America. Ed. D. W Haines. Totowa, NJ: Rowman and Littlefield Publishers. Pp. 24-39. Schoen, R. 1983 "Measuring the Tightness of a Marriage Squeeze," Demography, 20:61-78. Small, M. andJ. D. Singer 1982 Resort to Arms: International and Civil Wars, 1816-1980. Beverly Hills: Sage Publications. South, S. and K. Trent 1988 "Sex Ratios and Women s Roles: A Cross-national Analysis," American Journal of Sociology, 93:1096-1115. Starr, P.D. and A. E. Roberts 1982 "Community Structure and Vietnamese Refugee Adaptation: The Significance of Context," InternationalMigration Review ^ 16:595—618. Strand, PJ. and W Jones, J r. 1985 Indochinese Refugees in America: Problems of Adaptation and Assimilation. Durham, NC: Duke University Press. Tran .T.V.A. 1992 "The Direct Loan of Capital from the Bank to Develop Production and Gender Equality," Vietnam Social Sciences ^ 34:25-36. THE VIETNAMESE DOUBLE MARRIAGE SQUEEZE 127 Tran, M. V. and R. Holton 1991 Sadness Is Losing Our Country: Happiness Is Knowing Peace: Vietnamese Social Mobility in Australia, 1975-1990. Working Papers on Multiculturalism No. 12. Wollongong: Center for Multicultural Studies, University ofWollongong. Tran, T. V. and T. D. Nguyen 1994 "Gender and Satisfaction with the Host Society among Indo-Chinese Refugees," International Migration Review 28:323-337. Turley, W S. and M. Seldon, eds. 1993 Reinventing Vietnamese Socialism: Doi Moi in Comparative Perspective. Boulder, CO: Westview Press. United Nations 1990 Patterns of a First Marriage. ST/ESA/SER.R/111. New York. United Nations Development Program (UNDP) 1995 Vietnam through the Lens of Gender: An Empirical Analysis Using Household Survey Data — The Abstract. Hanoi. Viviani, N. 1984 The Long Journey: Vietnamese Migration and Settlement in Australia. Carlton: Mel bourne University Press. Werner, J. 1981 "Women, Socialism, and the Economy of Wartime North Vietnam, 1960-1975, Studies in Comparative Communism ^ 14:165-190. Wilson .WJ. 1987 The Truly Disadvanta ged: The Inner City) the Underclass, and Public Policy. Chicago IL: University of Chicago Press. Zeng, Y, P.Tu, B. C. Gu, Y. Xu, B. H. Li and Y P Li 1993 "Causes and Implications of the Recent Increase in the Reported Sex Ratio at Birth in China," Population and Development Review, 19:283-303. Interethnic Marriage: Identifying the Second Generation in Australia 1 Studies in Australia show that an increasing proportion of the population have ancestors from more than one country. Evidence regarding differences in the marriage patterns of first and second generation migrants has been restricted in scope as published marriage registration data includes only birthplace of partners. Marriage registration records include information about the birthplace of parents of partners, but it is available only through specially produced tabulations. Changes in the census for 1986 and 1991 make it possible to identify the second generation in households, and this article examines the use of census data as an alternative to marriage registration records in tracing changes in intermarriage patterns and differences between urban and rural areas. High levels of migration have transformed the world into one in which the populations of many countries include groups drawn from other societies which have quite different cultures. The relationship of diverse cultural groups to each other and the rate at which their cultures change are matters that have been of interest to sociologists, psychologists, and other social scientists for a considerable period. The social processes involved in cultural change are complex and affected by both major social structures and the informal relationships between different cultural groups. One way in which informal relationships may be affected is by intermarriage between members of different groups. In the early stages of migration to a country, the cultural heritage of most of the immigrant population will be that of their country of origin or birthplace. However, the close relationship between birthplace and cultural heritage may not necessarily hold for those people who have migrated more than once (Price, 1968; Hugo, 1994:81). The close relationship between cultural heritage and birthplace is likely to weaken after a period of continuing migration and settlement in the host society, as second and subsequent generations of immigrants acquire a cultural identity which is shaped by the culture of the society in which they have grown up, as well as the cultural background of their family (Young, 1991; Hout and Goldstein, 1994). While members of some ethnic ^his paper was prepared in 1993-94 and is based on research supported by a grant from the Monash University Gippsland Research Initiative Fund 1991-92. The authors would like to acknowledge the assistance received from Monash University. ©1997 by the Center for Migration Studies of New York. All rights reserved. 0197-9183/97/3101.0117 Parimal Roy lan Hamilton Monash University Gippsland Campus INTERETHNIC MARRIAGE IN AUSTRALIA 129 groups display a preference for a partner drawn from the same group, many select a partner from a different group, and therefore the cultural identification of both first and second generation immigrants may be affected by out-marriage (Taft, 1965:50; Price, 1968). McDonald (1989:41-42) has discussed the customs tending to preserve endogamy among migrant families in Australia. The importance of intermarriage in creating links between groups with different social and cultural backgrounds, and the changes that may flow from this, have long been recognized; a number of studies have explored the extent to which this occurs and its social significance. The ethnic identity of partners is only one element in a matrix of social background factors that affect the process of negotiating a relationship leading to marriage, and a decision to marry may be influenced by one or more of several important factors including class, education, and social background (Goode, 1964; Adams, 1986). An early study by Merton (1941) focused on differences in class and social background, while later studies have concentrated on religious and ethnic differences (Kennedy 1944, 1952; Price, 1963, 1968, 1993; Gordon, 1964; Alba, 1976, 1981; Alba and Chamblin, 1983; Alba and Golden, 1986; Lieberson and Waters, 1985; Judd 1990; Muhsam, 1990; Gariano, 1994). Jones (1987), Mare (1991) and Qian and Preston (1993) have traced changes in differences in educational achievement between partners over a number of decades in Australia and the United States. Muhsam (1990) has examined the different rates of " marrying up" and " marrying down " for males and females and argues that, in spite of the considerable "interest in intermarriage, one of its most significant aspects is being widely neglected: the comparison between two groups, namely marriages where the groom belongs to one group and the bride to the other and those where the bride belongs to the ' one' and the groom to the ' other '" (1990:307). Marriage registration data cover the whole population and are readily available, and this has resulted in their use as an indicator of cultural change through mapping intermarriage rates. Despite widespread use, discussion regarding the validity of birthplace as a measure of ethnic identity has been limited. One of the first arguments about this issue was presented by Price and Zubrzycki (1962b:58), who concluded that intermarriage statistics could be used as an indicator of assimilation or cultural change. Gordon (1964:80) proposed a theory of assimilation in which intermarriage was an important stage in the acceptance of a new group into a host society. This interpretation has been questioned by Newman (1973:162-163), who suggested that high rates of intermarriage may not necessarily be a valid indicator of assimilation to the culture of the host society. He proposed an alternative explanation which focused upon the role of " social contact and shared social space " rather than shared culture and argued that an increase in intermarriage rates may not lead to a change in culture and that ethnic intermarriage may be explained by " a trade-off theory " in which groups with 130 INTERNATIONAL MIGRATION REVIEW significant resources are assumed to be in a better position to attract potential marriage partners than those without (Newman, 1973:162—163). This view does not appear to have received support from other researchers. While marriage registration data record changes in Intermarriage patterns, they do not provide direct evidence regarding cultural change. Their interpretation involves inferences based upon theoretical considerations which are not directly assessed by the data. In the case of marriage registration records, an important limitation is concerned with the ethnic identity of the second generation. SOME PROBLEMS OF MEASUREMENT: MARRIAGE REGISTRATION AND THE CENSUS This article explores the task of measuring the ethnic identity of the children of migrants, the second generation, and discusses some of the important issues involved in attempts to track shifts in cultural change using intermarriage data and, on the basis of the authors ' recent research on intermarriage patterns in Australia (Roy and Hamilton, 1993,1994), makes some suggestions regarding the use of Australian census data. The investigation of ethnic intermarriage patterns involves several complications that appear to be unavoidable. One important difficulty is conceptual and arises from the notion of ethnicity and the issue of how a persons ethnic identity may be determined. This question has been discussed by a number of researchers, including Price and Zubrzycki (1962a, 1962b), Gordon (1964:23-30), Bottomley (1976:119), Bostock (1981:1-4), Lieberson and Waters (1985:43), Price (1985:51), and Young (1991:86), and is recognized in policy documents (Office of Multicultural Affairs, 1988; Jones and McAllister, 1992). Price and Zubrzycki (1962a:6l) identified some of the limitations involved in using birth place or nationality as a measure of ethnic identity due to the geographical mobility of many people and the ethnic mix in some societies, and they described the position as follows: Birthplace and nationality categories can be helpful in providing some approximation of ethnic origin and affiliation, but at times they are most misleading; especially where a country of origin . . . contains more than one major ethnic grouping, or when one ethnic grouping, the Greeks for example, are distributed over several countries of origin. Displaced persons and other refugee groups present similar difficulties. . . . About nine-tenths of Polish children coming to Australia were born in German displaced persons camps and their marriages with Poles born in Poland will appear in the birthplace statistics as mixed marriages between Germans and Poles... neither birthplace nor nationality categories reveal ethnic groups that are religious in basis. Bottomley (1976:119) made a similar point when she argued that "identity is formed by social processes; crystallized during the early years of life, it is maintained and modified by social relations, " and hence the influence of birthplace must be interpreted in conjunction with these social experiences. INTERETHNIC MARRIAGE IN AUSTRALIA 131 Price has discussed alternative measures of ethnic identity which may be used in cases of mixed ancestry (1985:51). In a recent paper on the collection of ethnicity data, Jones and McAllister (1992:172) acknowledged that the range of measures used, such as "birthplace, first language, language spoken at home, language proficiency and length of residence, " are only " surrogate measures" and take no account of the subjective dimension of " ethnic self-identification. " They conclude that birthplace has " considerable advantages " with regard to simplicity and interpretation even if it is not an "ideal measure. " If outmarriage continues over a period, an increasing proportion of the population will have mixed ancestry, and the percentage may become quite hi gh. Price (1988-89:34) has calculated that in Australia only 10 percent of third and subsequent generations have ancestors who share a single ethnic origin; some 30 percent have ancestors belonging to two ethnic groups. About 40 percent have ancestors from three or four ethnic groups; the remaining 20 percent of the population have ancestors from five or more ethnic groups. If only second generation Australians are considered, around 60 percent are children of mixed marriages. The incidence of intermarriage varies by both gender and ethnic group; using special tabulations from Australian marriage registration records, Price (1993) computed the incidence of interethnic marriage for the second generation, showing that rates of inmarriage were hig her for brides than for grooms, and that for brides, rates ranged from over 60 percent for second generation descendants of Italian, Greek and Turkish immigrants to under 10 percent for the descendants of Dutch and German immigrants. Another difficulty in the study of interethnic marriage patterns has its origin in the limited scope of the two major sources of data available for analysis — marriage registration records and the census. Price and Zubrzycki (1962a) recognized the limitation of marriage registration data, which requires the production of special tabulations in order to identify the second generation, and made a case for using census or survey data which provide more information about ethnic identity but may be more difficult to obtain. Choice of data source, therefore, involves a significant degree of judgement in balancing factors such as data quality, comprehensiveness and cost (Price and Zubrzycki, 1962a; Besanceney, 1965; Lieberson, 1966). Lieberson and Waters (1985:44) distinguished between two alternative measures of intermarriage which they describe as prevalence and incidence. "Prevalence refers to the frequency that a given characteristic is found at a given time in the population without reference to when it occurred. Incidence refers to the occurrence of a given event within a specified time period to the population at risk to the event within that period. " These measures may be derived from two different data sources. Briefly, incidence measures are computed from marriage registration data and provide an estimate of intermarriage rates for marriages occurring in a country during a defined time period. Prevalence measures are 132 INTERNATIONAL MIGRATION REVIEW obtained from survey or census data and provide an estimate of the total percentage of intact interethnic marriages in a population regardless of when or where the couple were married. In certain very rare circumstances incidence and prevalence rates may be the same, as Lieberson (1966:93) indicated: the two measures of intermarriage will be identical if all five of the following conditions occur: either all marriages are in the country under study or the proportion intermarried among those marrying elsewhere is the same as in the country of settlement; the proportion of outmarriages among members of the group entering marriage is constant through time; there is a stationary population; the dissolution rates for in- and out-group marriages do not differ; and the propensity to outmarry is the same for those remarrying as for the population making their first marriage. For most situations, however, the two measures will differ and a choice must be made between them. In the 1960s, it was not possible to calculate " prevalence rates55 from either Australian or United States census data, due to lack of information regarding time of marriage and household composition (Price and Zubrzycki, 1962a; Lieberson, 1966:99). By the 1980s, however, the availability of " new and superior data55 from the United States census, which included information on the ancestry of several generations, made it possible and desirable to investigate the intermarriage patterns of persons with mixed ancestry (Lieberson and Waters, 1985:5). In a very large-scale study based on ancestry data from the U.S. census, Lieberson and Waters (1988:22) considered two categories of people with mixed ancestry: included in the first are people in cross-cultural marriages, that is, partners whose parents were born in different countries; the second and larger category includes persons for whom both parents and grandparents did not share a common country of birth. In Australia there are three sources of data which allow for the identification of the second generation and hence the study of ethnic intermarriage patterns. These are marriage registration records, sample surveys, and the census. In 1986, the Australian census included information which enabled households containing second generation marriages to be easily and reliably identified and opened the way for the analysis of the intermarriage patterns of the Australian born whose parents were born overseas. Marriage Registration Records Intermarriage rates computed on the basis of annual marriage registration data are known as "incidence rates " or "T ype B indexes " (Price and Zubrzycki, 1962a:63), and these have been used in a number of studies. In Australia this data is available for marriages celebrated under the Marriage Act (1961) as amended. Marriages are registered on standard forms, provided by the Commonwealth, which record the birth place of brides and bridegrooms and also the birthplaces of their parents. This information is forwarded to the Australian INTERETHNIC MARRIAGE IN AUSTRALIA 133 Bureau of Statistics for processing and publication. The data regarding the birthplace of both members of the couple is coded and published annually Unfortunately for researchers and policy analysts the Bureau does not issue data on the second generation in readily usable form. The potential magnitude of second generation intermarriage may be illustrated by a brief examination of marriages registered in 1990: in about 8 percent of marriages both partners were born in the same overseas country; some 17 percent of marriages were between partners who were born in different overseas countries; and in 13 percent of marriages one partner was born in Australia and one partner in an overseas country. The remaining 60 percent of marriages were between partners who were born in Australia. This category includes many " second generation " Australians whose ethnic identity may not be accurately reflected when using birthplace as a measure (ABS, 1991). The first Australian investigations of intermarriage patterns were small-scale studies which used marriage registration data for particular ethnic groups and regions (Price, 1955, citedin Price and Zubrzycki, 1962b; Price, 1968; Cronin, 1970; Huber, 1977). Price (1968) provided estimates of differences in the marriage patterns of first generation immigrants and their Australian-born offspring. More extensive analyses of intermarriage patterns using marriage registration data have been undertaken by Gray (1987), Dyer (1988), McCaa (1989), Jones (1991), and Price (1993). In discussing intermarriage, Gray (1987:366) acknowledged the importance of membership in a group " with common social and psychological characteristics " in the choice of a partner and identified the limitations of census data. However, he did not include in his analysis an assessment of second generation influence on group membership as he used the birthplace of partners only Dyer (1988), McCaa (1989) and Jones (1991) limited their analyses to marriages between "Australians " and "immigrants, " indicated by place of birth. While these studies have provided evidence regarding intermarriage patterns for the first generation, they have not adequately addressed questions regarding group membership and the ethnic identity of the " second generation. " For marriage partners born outside Australia, " place of birth" may, for the most part, be used as a reliable indicator of ethnicity; there are, however, difficulties in attempting to determine the ethnicity of marriage partners born in Australia (Price and Zubrzycki, 1962a). For those people whose families have been resident in Australia for two or more generations, the question arises as to the determination of the appropriate ethnic identity for the second and subsequent generations. Lieberson and Waters (1985:43) referred to people in this category as " persons of mixed ethnic ancestry. " Many ethnic groups have been resident in Australia for long enough to have children of marriageable age who were born in Australia and who therefore may be involved in second generation marriages. These will appear in the marriage registration records as 134 INTERNATIONAL MIGRATION REVIEW Australian-Australian marriages or Australia-other birthplace group marriages. While this limitation was pointed out in an early paper by Price and Zubrzycki (1962b), some subsequent analyses have relied upon birthplace of couple only For example, as indicated previously, Gray (1987), Dyer (1988), McCaa (1989), and Jones (1991) have computed intermarriage rates using birthplace data for the couple only, without providing any indication of how the results might be affected by the reclassification of second generation partners of the Australian born. In the early period of large-scale migration to Australia, Price (1955, cited in Price and Zubrzycki, 1962b:62-64) estimated that about 50 percent of the apparent " out-marriages " of southern Europeans could be reclassified as "in-marriages, " as they involved a second generation partner with at least one parent from the same ethnic group. Evidence from other societies indicates that ethnicity is important in the choice of a partner (Judd, 1990; Muhsam, 1990; Hout and Goldstein, 1994). Price (1968:110-111) suggested that the percentage of second generation inmarriages may decline with longer duration of residence. In a later study, after the computerization of marriage registration records by the Australian Bureau of Statistics had facilitated the analysis of second generation marriage patterns using special tabulations, Price (1993) compared second generation intermarriage rates in three periods between 1965 and 1990. He found that while there had been a reduction in the inmarriage rate to around 23 percent overall during the period 1987-1990, for southern European groups inmarriage rates had remained remarkably stable at about 50 percent (Price, 1993: Table 1, p. 7). In the United States, marriage registration records are not as comprehensive as in Australia, and this has led Hwang and Saenz (1990:564) to argue that " although the theoretical advantages associated with the incidence-based measure of intermarriage seem clear, limited information regarding race and ethnicity in the registration data have prevented researchers from doing more detailed analysis concerning interracial and interethnic marriages. " Sample Surveys Lieberson (1966) indicated that one source of more detailed intermarriage data was to be found in specially designed surveys. In Australia a number of large sample surveys incorporating over 2,000 respondents have collected information about married couples, including their place of birth and the birthplace of their parents. Two large datasets containing this information and available for secondary analysis come from the study of family formation in Australia. The first is the Australian Family Formation Project, a survey carried out in Melbourne in 1971 by the Department of Demography at the Australian National University, and the second is the Family Formation Study carried out in 1981—82 by the Australian Institute of Family Studies. These sample INTERETHNIC MARRIAGE IN AUSTRALIA 135 surveys provided data which enable the second generation to be identified and collected information about values regarding marriage partners. However, the relatively small number of interethnic marriages they identified means that estimates of second generation intermarriage rates derived from them are subject to a relatively large sampling error (ABS, 1977). Even with a sample size of about 2,500, the number of interethnic marriages is relatively small. In the 1981-82 Australian Institute of Family Studies survey, 1,356 out of 2,544 respondents reported that they were either married or living with a partner. Of these, 1,032 had partners from the same birthplace group, and only 324 had a partner from a different birthplace group. The largest birthplace group for males was the United Kingdom with 169 males; the smallest was northeast Europe with35 males. Price ' s early studies (1955 citedin Price and Zubrzycki, 1962a:62-64) of Italians in Griffith and Greeks in Sydney provided data for the period 1947 to the mid-1950s at the beginning of the era of large-scale immigration. Cronin (1970) provided some material in relation to the expectations of marriage partners for Sicilians in Sydney during the 1950s and 1960s, and Huber (1977:89) reported intermarriage rates for marriages registered in four parishes of the Catholic Church in the Murrumbidgee Irrigation Area during the period 1950 to 1969. Each documented the strong preference for partners from within the same ethnic group. A much larger data source, which will reduce some of the inaccuracy due to sample size, has recently become available. The survey, Families in Australia 1992, which was conducted by the Australian Bureau of Statistics, and released late in 1994, obtained detailed information on a wide range of family relationships and activities from a national sample of some 12,400 families, which included about 35,000 subjects (Dean, 1993). Census Data Census data can be used to calculate the prevalence of intermarriage provided household identification is possible. Prior to 1986, the Australian census was not a suitable source for the analysis of intermarriage patterns due to the difficulties in reconstructing households and the tracking of persons absent from their regular place of residence on census night. From the point of view of the present study, the most significant changes introduced by the Australian Bureau of Statistics for the 1986 census were concerned with the identification of households and the structure of relationships within households. These changes have been highlighted by Maher and Burke (1991:55), who described them as follows: The 1986 census no longer used the concept of head of household or head of family. Respondents were instructed to complete the household form with the householder or any adult. . . member as "person I," and the spouse/partner then as " person 2." 136 INTERNATIONAL MIGRATION REVIEW Family structures were then determined on the basis of each household member's relationship with persons one and two. This then differed from the previous practice of identifying family structure on the basis of family and household head. In addition, a change in the question on relationships within a household in 1986 made it possible to identify de facto living arrangements and the existence of blended families. Further improvements to the coding of household and family structures were achieved through the inclusion of a question relating to details of usual residents of a dwelling who were temporarily absent on census nig ht. This enabled a more accurate representation of family structures, and minimized the problem of an overcounting of single parent families created by temporarily absent spouses on previous censuses. Because of these changes households can now be identified clearly and reconstituted accurately and estimates of the prevalence of intermarriage may be computed. When information regarding the birthplaces of the parents of partners, date of marriage, language usage and period of residence in Australia are combined, a more reliable measure of ethnic identity may be made. This has significantly enhanced the value of census data as a source for the analysis of ethnic intermarriage patterns. However, it must be acknowledged that census data do not provide any information regarding former partners in households containing widowed, divorced or separated persons. The availability of census data in electronic form has enabled researchers to undertake analyses that would have been impossible earlier. In the study of intermarriage, Jones (1987) has used the one percent sample tape from the 1981 census to analyze trends in educational homogamy in Australia, and Gariano (1994:41) has used " customized matrix data " from the 1991 census to examine the religious identification of marriage partners. REGIONAL VARIATION IN THE PREVALENCE OF INTERMARRIAGE IN AUSTRALIA: AN ANALYSIS OF THE 1986 CENSUS The ethnic composition of the Australian population has undergone significant change since the 1950s, and this has been accompanied by increased interaction between groups which has led to an increase in the rate of intermarriage. An exploratory study (Roy and Hamilton, 1993; 1994) was designed to compute the prevalence of intermarriage using data from the 1986 census and to examine differences between metropolitan and rural areas, as well as changes over time. A special data matrix was obtained for all households in which two adult "heads " were identified, thus including couples in first or subsequent marriages and de facto couples. The matrix contained 3,377,082 households from all states and territories. It is estimated that 94 percent of the households identified by this method contained married couples and the remaining 6 percent cohabiting couples (ABS, 1992:167). Budgetary constraints limited the measurement of group membership to birthplace of INTERETHNIC MARRIAGE IN AUSTRALIA 137 partners only, and hence this study shares some of the limitations of the studies mentioned above. However, use of the census as a source of birthplace data facilitated the computation of intermarriage rates for different geographical areas, a task which cannot be undertaken with data from marriage registration records except for interstate comparisons. Changes over time were estimated using age as a surrogate measure for length of residence in Australia. The study identified significant variations between geographical areas and found that immigrants who settled in rural areas had a different pattern of intermarriage from those living in the major urban centers. The study also corroborated the pattern of inmarriage declining with duration of residence found by Price (1968; 1993) and Young (1991:89). Some of the variation in intermarriage rates between metropolitan and rural areas may be explained by the "lack of opportunity " (Gray, 1987) which results when there is a shortage, in rural areas, of single females of the appropriate age and in the same ethnic group. Low levels of geographic mobility in some rural areas may also contribute to differences in rates of intermarriage. THE PREVALENCE AND INCIDENCE OF INTERMARRIAGE IN AUSTRALIA: A COMPARISON OF CENSUS AND MARRIAGE REGISTRATION DATA This study has been concerned with the relationship between ethnic identity and birthplace for second generation migrants. Price and Zubrzycki (1962a), Price (1963, 1968, 1993), Gray (1987), McCaa (1989), Jones (1991), and Young (1991) have analyzed changes in intermarriage patterns and discussed the question of how the relationship between different birthplace groups may be explained. Young (1991:89) identified the main factors contributing to the changes in intermarriage as follows: While the extent of in-marriage generally decreases with duration of residence in Australia, the pace of change varies between different birthplace groups, being influenced by cultural differences, differential retention of the customs and traditions of the source countty, the availability of spouses, and the size and rate of increase of the ethnic group in Australia. The ethnic identification of second generation migrants has been acknowledged by these researchers, but in the studies reported outmarriage rates have been computed using marriage registration data based solely on the birth places of the partners. The debate between Gray (1987), McCaa (1989), and Jones (1991) is concerned with the appropriate measures to be used in analyzing changes in outmarriage rates. There are a number of difficulties involved in using marriage statistics in interpreting the incidence of intermarriage rates. While the incidence of intermarriage will be related to the availability of partners, that cannot be measured simply by the number of unmarried people 138 INTERNATIONAL MIGRATION REVIEW in a certain age range but must be related to cultural prescriptions regarding the qualities of desirable spouses. This has been stressed by Gray (1987:365), who stated that a marriage market is not necessarily measurable in terms of numbers of unmarried persons within those age ranges in which marriage is culturally acceptable. . . . [W]ithin the category of unmarried persons in some cultures, there may be people actively seeking marriage partners, but with varying degrees of strong selectivity . . . which may not be reciprocated. In other cultures, it may be parents who are most actively involved in seeking marriage partners for their children approaching the age at which they should be married, but again the search is not random but strongly selective. In either type of culture, there is also the possibility of what might be described as a passive search for marriage partners, and if the right person comes along marriage would be a possible outcome. Price and Zubrzycki (1962a) acknowledged the limitation involved in using marriage registration data to measure interethnic marriage rates. Subsequently Price (1993) has overcome this by using special tabulations to provide data about the rate of second generation inmarriages. Inmarriage rates do not simply reflect preferences but, as Lieberson (1966), Alba and Golden (1986), and Lieberson and Waters (1988) suggested, are affected by group size, demographic composition, and opportunities for contact. As Alba and Golden (1986:203) argued: in-marriage rates do not simply reflect preferences for in-group marriage, but are influenced by a variety of other factors, including most importantly group size. . . . [A] smaller group provides fewer possibilities for a compatible spouse than a larger one, and thus its in-marriage rate ought to be lower, ceteris paribw. Until the confounding influence of group size is controlled, no one can be sure how far current intermarriage rates fall short of the rates that would prevail in the absence of any tendency to in-marry. In discussing preferences for inmarriage, Lieberson and Waters (1988:206) were explicit in arguing that: Contact and experience with members of the ingroup and outgroup will play a role in the disposition young adults develop towards potential mates from different groups. Moreover, there is the distinct possibility that persons who live in an area with relatively few ethnic compatriots on the average have an initially weaker disposition toward ingroup maintenance. Many ethnic groups tend to congregate in the same geographic location and localized employment sector. Through these factors, ethnic identification and crystallization is encouraged, and hence contact between ethnic groups becomes selective, and previously loose-knit networks become solidified and crystallized. The drive towards their own network is aided by ascription, the identification of individuals "imposed upon them by government, the political machine, the Church, the school, the natives and even by other immigrants" (Sarna, 1978:376-377). INTERETHNIC MARRIAGE IN AUSTRALIA 139 CONCLUSION Marriages between members of different cultural groups are an important indicator of the relationship between those groups, and the level of intermarriage is related to the extent to which cultural patterns are shared by partners. A number of studies have shown that the proportion of cultural groups with mixed ancestry increases with continuing migration and longer residence (Alba and Chamblin, 1983; Lieberson and Waters, 1985; Alba and Golden, 1986; Price, 1988-89; Young, 1991). Research suggests that for most ethnic groups the intermarriage patterns of the second generation are different from that of the first. Until the recent analysis of special marriage registration tabulations by Price (1993), the Australian data to test this finding was limited to a few small-scale studies which reported changes in marriage patterns (Price and Zubrzycki, 1962a, 1962b; Price 1968; Cronin, 1970; Huber, 1977). While special marriage registration tabulations enable the second generation to be identified, they do not provide information about the subjective dimensions of a person ' s ethnic identity (Young, 1991:67, 86). Another limitation is that they do not allow for comparison between regions as the data record only the state in which the marriage was registered. This article has argued that census data relating to households provides an alternative source for analyzing the prevalence of interethnic marriage in Australia and can complement data based on marriage registration records. The census includes information on birth place, parental birthplace, date of first marriage, length of residence, main language spoken other than English, and place of residence. The information regarding birthplace and parent ' s birthplace allows the second generation to be identified. When combined with data regarding language usage, it provides for a more comprehensive assessment of ethnic identity, and the inclusion of place of residence enables comparisons to be made between different regions and between urban and rural areas. REFERENCES Adams, B. N. 1986 The Family: A Sociological Interpretation. 4th ed. New York: Harcourt Brace Javanovich. Alba, R. D. 1981 " The Twilight of Ethnicity among American Catholics of European Ancestry," The Annals, 454:86-97. 1976 "Social Assimilation among American Catholic National Origin Groups, n American Sociological Review, 41(4): 1040-1044. Alba, R. D. and M. B. Chamblin 1983 "A Preliminary Examination of Ethnic Identification among Whites, n American Sociological Review, 48(1):240-247. 140 INTERNATIONAL MIGRATION REVIEW Alba, R. D. and R. M. Golden 1986 "Patterns of Ethnic Marriage in the United States," Social Forces, 65 (I): 202-223. Australian Bureau of Statistics (ABS) 1994 Families in Australia 1992. Canberra: Australian Government Publishing Service. 1992 Year Book Australia , 1992. Canberra: Australian Government Publishing Service. 1991 Marriages, Australia: 1990. Canberra: Australian Government Publishing Service. 1977 Household Expenditure Survey 1974-75 Bulletin 3: Standard Errors. Canberra: Australian Government Publishing Service. Besanceney, P.H. 1965 "On Reporting Rates of Intermarriage," American Journal of Sociology 70(6):717-721. Bostock, W. W. 1981 Alternatives of Ethnicity: Immigrants and Aborigines in Anglo-Saxon Australia, 2nd ed. Melbourne: Corvus Publishers. Bottomley, G. 1976 "Ethnic Identity among Greek Australians," Australian and New Poland Journal oj Sociology 12(2): 118-125. Cronin, C. 1970 The Sting of Change: Sicilians in Sicily and Australia. Chicago, IL: University of Chicago Press. Dean, J. 1993 "Examining the Nature of Family Support in the 1992 Survey of Families in Australia. " Paper presented at the 4th Australian Family Research Conference, Sydney. February 17-19. Belconnen, ACT: Australian Bureau of Statistics. Dyer, K. F. 1988 "Changing Patterns of Marriage and Mating in Australia," Australian Journal of Sex, Marriage and Family 9(2): 107-119. Gariano, A. C. 1994 "Religious Identification and Marriage," People and Place, 2(l):4l-47. Goode, W. ]. 1964 The Family. Englewood Cliffs, NJ: Prentice-Hall. Gordon, M. M. 1964 Assimilation in American Life. New York: Oxford University Press. Gray, A. 1989 "Measuring Preference for In-marriage: A Response to McCaa," Population Studies, 43(1):163-166. 1987 "Intermarriage: Opportunity and Preference," Population Studies, 4l(3):365-379. Hout, M. and J. R. Goldstein 1994 "How 4.5 Million Irish Immigrants Became 40 Million Irish-Americans: Demographic and Subjective Aspects of the Ethnic Composition of White Americans," American Sociological Review, 59(l):64-82. Huber, R. 1977 From Pasta to Pavlova: A Comparative Study of Italian Settlers in Sydney and Griffith. St. Lucia: University of Queensland Press. INTERETHNIC MARRIAGE IN AUSTRALIA 141 Hugo, G. 1994 "Demographic and Spatial Aspects of Immigration. In Australian Immigration: A Survey of the Issues. Ed. M. Woden et al. Bureau of Immigration and Population Research, Canberra: Australian Government Publishing Service. Pp. 30-110. Hwang, S.-S. and R. Saenz 1990 "The Problem Posed by Immigrants Married Abroad in Intermarriage Research: The Case of Asian Americans," International Migration Review, 24(3): 563-576. Jones, F. L. 1991 "Ethnic Intermarriage in Australia, 1950-52 to 1980-82," Population Studies, 45(1):27-42. 1987 "Marriage Patterns and the Stratification System: Trends in Educational Homogamy since the 1930s," Australian and New Poland Journal of Sociology 23(2): 185-198. Jones, R. and I. McAUister 1992 "An Overview of Data Collection. " In Access and Equity Evaluation Research. Ed. J. Jupp and A. McRobbie. Office of Multicultural Affairs, Canberra: Australian Government Publishing Service. Pp. 171-185. Judd, E. P 1990 "Intermarriage and the Maintenance of Religio-Ethnic Identity. A Case Study: The Denver Jewish Community," Journal ofComparative Family Studies, 21 (2):251-268. Jupp, J. and A. McRobbie, eds. 1992 Access and Equity Evaluation Research. Office of Multicultural Affairs, Canberra; Australian Government Publishing Service. Kennedy, R. J. R. 1952 "Single or Triple Melting Pot? Intermarriage Trends in New Haven, 1870-1950," American Journal of Sociology 58(1):56—59. 1944 "Single or Triple Melting Pot? Intermarriage Trends in New Haven, 1870-1940," American Journal of Sociology 49(2) :331-339. Lieberson, S. 1966 "The Price-Zubrzycki Measure of Ethnic Intermarriage," Eugenics Quarterly 13(2):92-100. Lieberson, S. and M. Waters 1988 From Many Strands: Ethnic and Racial Groups in Contemporary America. New York: Russell Sage Foundation. 1985 "Ethnic Mixtures in the United States," Sociology and Social Research, 70(1):43-52. Maher, C. and T. Burke 1991 Informed Decision-making: The Use of Secondary Sources in Policy Studies. Melbourne: Longman Cheshire. Mare, R. D. 1991 "Five Decades of Educational Assertive Mating," American Sociological Review, 56(1):15-32. McCaa, M. 1989 "Isolation or Assimilation? A Log Linear Interpretation of Australian Marriages, 1947-1960, 1975, 1986," Population Studies, 43(1):155-162. McDonald, P. 1989 "Ethnic Family Structure," Family Matters, 23:38-45. Merton, R. K. 1941 "Intermarriage and Social Structure: Fact and Theory," Psychiatry^ 4:361-374. 142 INTERNATIONAL MIGRATION REVIEW Muhsam, H. 1990 "Social Distance and Asymmetry in Intermarriage Patterns," Journal of Comparative Family Studies, 21(3) .'307-324. Newman, W M. 1973 American Pluralism: A Study of Minority Groups and Social Theory. New York: Harper and Row. Office of Multicultural Affairs 1988 National Guidelines for the Collection of Ethnicity Data. Canberra: Australian Government Publishing Service. Price, C. A. 1993 "Ethnic Intermixture in Australia," People and Place, 1(1):6-8. 1988- "The 'Melting Pot ' is Working, " IPA Review, 43(3):34-35. 89 1988 " The Peoples of Australia." In Australia: 1988. Ed. J. Jupp. Melbourne: Longman Cheshire 1985 "The Ethnic Composition of the Australian Population. " In Immigration and Ethnicity in the 1980s. Ed. I Bumley, S. Encel and G. McCall. Melbourne: Longman Cheshire. Pp. 43-57. 1968 "Intermarriage and the Second Generation. " Migrants in Australian Society. Background Paper 6, H.R.H. The Duke of Edinburgh' s Third Commonwealth Study Conference, Sydney. Published as Anatomy of Australia, Melbourne: Sun Books. 1963 Southern Europeans in Australia. Melbourne: Oxford University Press. Price, C. A. and J. Zubrzycki 1962a "The Use of Inter-marriage Statistics as an Index of Assimilation," Population Studies, 16(1):58-69. 1962b "Immigrant Marriage Patterns in Australia," Population Studies, 16(2): 123-133. Qian, Z. and S. Preston 1993 "Changes in American Marriage, 1972 to 1987: Availability and Forces of Attraction by Age and Education," American Sociological Review, 58(4):482-495. Roy, P.K. and I. V. Hamilton 1994 "Patterns of Intermarriage in Australia: Regional Variations in Intermarriage Rates." Australian Journal of Marriage and Family 15(3): 124—136. 1993 Intermarriage in Australia: A Comparison of Intermarriage Rates in Gippsland and the Rest of Australia, Research Report, Monash University Gippsland Campus. Sama, J. D. 1978 "From Immigrants to Ethnics: Towards a New Theory of 'Ethnicization'," Ethnicity, 4(4):370-378. Taft.R. 1965 From Stranger to Citizen: A Survey of Immigrant Assimilation in Western Australia, Perth: University of Western Australia Press. Young, C. M. 1991 "Changes in the Demographic Behaviour of Migrants in Australia and the Transition between Generations," Population Studies, 45(1):67-89. RESEARCHNOTE Network Mi g ration of Ethnic Germans An exploration of the real world's data suggests that international migration occurs largely in ethnic networks (^^Zimmermann, 1995), a fact most theories and applications, at least in economics, currently fail to account for. Interpersonal and ethnic connections link the new migrant with previous migrants and natives in the receiving country by ties of ethnicity, kinship, and friendship. Co-movements of migrants might be affected by preferences to live in the neighborhood of people that share the same traditions, experiences, and values, but also by advantages in receiving information and help in order to achieve easy integration in the economic, cultural and political systems of the destination country. Another neglected area is migration of ethnic Germans (see, however, Heller, 1993; Goddecke-Stellmann, 1994; Schmidt, 1994; Veith, 1994) that has played a very significant role in the post-World War II experience of Germany. According to the German Basic Law, ethnic Germans have a right to receive a German passport immediately, even if they are coming from an environment with significantly different culture such as the former Soviet Union, Rumania, and Poland. This group (called Aussiedler) has often no proper knowledge of German and is different from those Germans that came from the former East Germany (the Ubersiedler) to the west. The first wave of a new migration survey attached to the German Socioeconomic Panel that was recently made available allows for a proper analysis of both types of ethnic migrants. (See Bauer and Zimmermann, 1996, for an analysis of earnings and unemployment using this source.) Here, we will use this innovative new data source to study network behavior of ethnic Germans while comparing differences in the attachments to relatives and friends across sending regions. The study then deepens the economic framework and provides more information about the data source, and outlines estimation procedure and results. ECONOMICFRAMEWORKAND DATA According to human capital theory, migration occurs if the discounted expected returns in a potential destination country net of the migration costs are larger than the discounted expected returns in the home country. For a time horizon t = [0, T], the decision rule of an individual can be modeled by: ©1997 by the Center for Migration Studies of New York. All rights reserved. 0197-9183/97/3101.0117 Thomas Bauer Klaus F. Zimmermann SELAPO, University of Munich 144 INTERNATIONAL MIGRATION REVIEW R(t = 0) = J[^ (X/N)YtD(X)-^(X)Y ^ (X)] e -rt ^-C(X,N)e -rt (1) where R(t = 0) denotes the expected return to migration before departure at t =0, ptD and ptH are the probabilities of being employed, and YfD and Y(H are the incomes the individuals expect in the destination country D and the country of ori g in H, respectively. The employment probabilities, as well as the expected incomes, are functions of the socioeconomic characteristics X of the individual, like age, schooling, and labor market experience. A subjective constant discount rate is denoted by r. The costs of movement C not only include money and opportunity costs like travel expenses, differences in the costs of living, search costs, and forgone earnings while moving, but also include psychological costs arising from separation from family and friends and habitual environment. Recently the importance of migrant networks for the migration decisions has been stressed (Boyd, 1989; Massey, 1990; Massey etai, 1993; Gurak and Caces, 1992). In this literature, migration networks are defined as interpersonal connections that link migrants, former migrants, and nonmigrants in the destination country and the country of origin by ties of kinshi p and friendship. These personal connections to individuals in a destination country, denoted by N in equation (1), lower all mentioned costs of movement and raise the employment probability in the country of destination because potential migrants can expect the support of the network. Therefore, existence of migrant networks increases the expected net returns of migration and migration becomes more likely. This paper examines whether migration networks are used by migrants when they reach West Germany. Furthermore, we analyze the determinants of integration of migrants in an existing network during the first years after immigration. The sample used for the analysis stems from a new German data set collected in 1994 that complements the German Socioeconomic Panel (see Schuiz etal ^ 1993). The dataset includes only Ubersiedler and Aussiedler who migrated to West Germany not longer than ten years ago. An extensive comparison with official data revealed that this data of immigrated ethnic Germans could be treated as representative (see Bauer and Zimmermann, 1996). Originally the dataset included 471 interviewed persons. Because of missing values, some observations had to be dropped. Therefore, the final sample used for our estimations contains 328 observations. Ubersiedler are Germans from the former German Democratic Republic (GDR) who were leaving the GDR to live permanendy in West Germany. Aussiedler are ethnic Germans who move from countries in East Europe to NETWORK MIGRATION OF ETHNIC GERMANS 145 West Germany. Furthermore, we are able to differentiate Aussiedler into migrants from Poland, Rumania, and the former Soviet Union. To study the use of networks by migrants, we construct a dependent variable that takes the value 1 if a person has settled in a city where some relatives or friends of the migrant are living and 0 otherwise. The integration of migrants in an existing network is analyzed by using a binary variable which takes the value 1 if at least one of the individual's three best friends is from the same country of origin and 0 otherwise. Both equations are estimated using the probit model. According to the human capital theory of migration, the probability of migration should decrease with age and increase with education, reflecting the smaller expected lifetime gain from moving for older people and the hi g her ability to collect and process information about the destination country by better educated individuals. Because of this relative disadvantage of older and less educated people, we expect that the probability of using existing networks just after the immigration to increase with age at immigration and decrease with the years of schooling in the country of origin. For the same reason, older and less educated individuals should be better integrated in an existing network than younger and more educated people. Furthermore, we expect that Aussiedler depend significantly more on the support of relatives and friends than Ubersiedler because the former often do not speak German well and grew up in a different cultural environment. When ethnic Germans arrive in West Germany they have to live in reception camps until their status as Aussiedler or Ubersiedler is confirmed. Among other things, the time an individual spends in a reception camp depends on the availability of accommodations and can be shortened if relatives or friends hel p to find a housing space. Therefore, we use the time a migrant has spent in a reception camp as a proxy which reflects ability to participate in an existing network. This variable is supposed to have a negative influence on the probability to live close to relatives or friends and on the integration in an existing network. We further want to test the hypothesis that the probability to participate in ethnic networks is higher in German states with a high population density relative to regions with low population density because in the former, networks should be better organized. Living in rural areas should have a negative influence on the integration in existing networks because of a weaker presence of these networks there. To control for the competing role of government help to network support we include per capita government expenditures in the states in which the migrants have settled the year before immigration. A negative effect on the process of integration is also expected from the duration of stay in the country. Several other factors may influence the participation in migration networks. We examine differences generated by gender, marital status, and the presence of children below age 16. 146 INTERNATIONAL MIGRATION REVIEW ESTIMATION RESULTS The descriptive statistics of the used variables are given in the first column of Table 1. Note that 73 percent of the immigrants live close to relatives and friends and 77 percent have friends of the same ethnicity, indicating the impor tant role of networks for newly arrived immigrants. From the observed individuals, 34 percent were born in the former GDR, 11 percent in Rumania, 27 percent in Poland, and 28 percent in the former U.S.S.R. On average, the migrants are 38 years old and have spent 5 years in Germany, including about 3 months in a reception camp. Both Ubersiedler and Aussiedler had ten years of schooling in their countries of origin and about two years of schooling in West Germany. Only 14 percent of the migrants live in a rural area. Half of the sample is female, 77 percent of the sample is married, and 55 percent repor ted having children under the age of 16. The second and third columns of Table 1 show the estimated impact of the independent variables on the probability to settle in a city where relatives or friends are living (column 2) and that at least one of the three best friends is from the same country of origin, respectively. The likelihood-ratio test statistic LRT and a pseudo R 2 measure that is discussed in Veall and Zimmermann (1992) are used for model evaluation. R^z ls calculated by using individual model predictions of the latent endogenous variable and is known to mimic the true OLS-R 2 . The overall fit of the model is quite acceptable. The estimates show a significantly positive relationship between the age at immigration and the probability to settle close to relatives or friends, whereas age has no significant influence on the use of an existing network after immigration. The probability to have friends of the same country of origin significantly decreases with the years since migration. This result indicates that with duration of stay in the receiving country, the migrant assimilates into the new society and becomes more independent from ethnic networks. Years of schooling in East Germany significantly decreases the probability to live near relatives or friends, which confirms the hypothesis that more-educated people are less dependent on the help of networks. However, years of schooling in West Germany and years of schooling in East Germany significantly increase the probability to have close friends from the same ethnic origin, while years of schooling in East Europe have no effect at all, which is not in line with the theoretical considerations outlined above. The time an individual spends in a reception camp after immigration has a significantly negative influence on the probability to live close to relatives and friends but has no significant effect on the integration in the network, reflecting the great importance of connections immediately after immigration. Also, living in rural areas significantly decreases the probability to have friends from the same country of origin. As expected, a high population density significantly increases NETWORK MIGRATION OF ETHNIC GERMANS TABLE 1 DESCRIPTIVE STATISTICS AND ESTIMATION RESULTS Constant Age Age at Immigration Female Married Children under 16 Years since Migration Ubersiedler Rumanian Pole Former USSR Years of Schooling in West Germany Years of Schooling in East Germany Years of Schooling in East Europe Months in Reception Camp Rural Population Density 3 ' 15 Per Capita Government Expenditures in DM 11 Friends of Same Country of Origin Relatives Close By LRT(x 2 ) R^z Note; Standard deviations and absolute t-values g iven in parentheses respectively. LRT(^ 2 ) is the likelihood-ratio test statistic on the hypotheses that all coefficients except the constant are zero. Pseudo-J? 2 is explained in the text. Tlie number of observations is 328. *Significance at least at 5% level (one-side-test). copulation density defined as population per Ion 2 . Source; several issues of the Statistical Yearbook of Germany, Descriptive Relatives or Friends of Same Statistics Friends Close By Country of Orion 37.878 (14.06) 33.198 (14.14) 0.497 (0.50) 0.765 (0.42) 0.549 (0.50) 4.680 (1.57) 0.341 (0.47) 0.110 (0.31) 0.265 (0.44) 0.284 (0.45) 1.741 (3.61) 10.496 (4.44) 9.833 (4.85) 3.439 (7.65) 0.140 (0.35) 437.28 (721.30) 4308.90 (1405.50) 0.771 (0.42) 0.729 (0.45) 2.948* 0.173 (3.93) (0.33) 0.009 ~~ (1.19) 0.011* (1.83) ~~ 0.163 0.160 (1.02) (0.92) -0.303 -0.341 (1.44) (1.34) 0.194 ~~ (0.83) -0.289* ~~ (4.77) -0.811 2.518* (1.48) (4.48) -0.714 1.805* (1.32) (3.86) 0.080 1.791* (0.15) (3.92) 0.056* — (1.80) -0.107* 0.105* (2.75) (3.02) -0.035 0.018 (1.50) (0.66) -0.033* -0.017 (2.94) (1.57) -0.550* ~ (2.34) 0.001* 0.0003 (3.42) (1.85) -0.0005* (3.04) ~ 50.28 67.48 0.25 0.34 148 INTERNATIONAL MIGRATION REVIEW the use of migration networks and the settlement close to relatives and friends. However, note that average population densities in regions can conceal vast contrasts in the types of settlements contained within their boundaries. The significantly negative coefficient of per capita government expenditures in the state a migrant has settled indicates that the provision of public benefits lowers the dependence of migrants on the support of relatives and friends. This result is in line with recent theoretical and empirical work that stresses the importance of government expenditures in the receiving relative to the sending country for the migration decisions of individuals. Neither migrant^ sex, nor marital status, nor the existence of young children has a significant effect on the probability to move to a city where relatives or friends are living or to have friends of the same country of origin. How different are ethnic Germans according to their source country? Table 1 compares the effects with reference to East Germans. The insignificant coefficient on the country of ori g in dummies indicates that ethnic migrants do not differ with respect to their settlement decisions. However, East Europeans are more likely to link themselves with people from the same ori g in than East Germans, which holds strongest for the Rumanians. This is consistent with the finding of Bauer and Zimmermann (1996) that Rumanians also perform well in the labor market. CONCLUSIONS This study has argued that network migration is central to actual migratory movements, which suggests that its determinants should be carefully investigated. The analysis demonstrates that the network behavior of ethnic Germans can be explained, to a large degree, by observable characteristics. Two crucial variables were investigated, namely the connections to friends from the country of origin and the settlement close to relatives and friends. East Europeans, especially from Rumania, are significantly more connected to friends from the same country of origin than are East Germans. These connections strongly decline with duration of residence and presence in rural areas. Older people are more likely to settle close to friends or relatives at the time of immigration. Population density increases, education, length of stay in a reception camp, and per capita government expenditures decreases the likelihood of such a settlement. REFERENCES Bauer, T. and K. F. Zimmermann 1995 "Arbeitslosigkeit und Lohne von Aus- und Obersiedlem." In Mikrookonomik des ArbeitsmarketS) BeitragezurArbeitsmarket- undBerujsforschung^ No. 192. Ed. V. Steiner and L. Bellmann, Niimberg: Institut fur Arbeitsmarket- undBerufsforschimg. NETWORK MIGRATION OF ETHNIC GERMANS 149 Boyd,M. 1989 "Family and Personal Networks in International Migration: Recent Developments and New Agendas," International Migration Review ^ 23:638-670. Goddecke-Stellmann, J. 1994 "Raumliche Implikationen der Zuwanderungen von Aussiedlern und Auslandern: Riickkehr zu alten Mustern oder Zeitenwende?" Informationen zur Raumentwicklung, 5/6:373-386. Gurak, D. T. and F. Caces 1992 "Migration Networks and the Shaping of Migration Systems." In International Migration Systems: A GlobalApproach. Ed. M. Kritz, L. L. Lim and H. Ziotnik. Oxford: Clarendon Press. Heller, W. 1993 Integration von Aussiedlem aufdem deutschen Wohnungsmarkt, Studio im Auftrag des Bundesministeriums fur Raumordnung, Bauwesen und Stadtebau. Geographisches Institut, University of Gottingen. Massey, D. S. 1990 "Social Structure, Household Strategies, and the Cumulative Causation of Migration," Population Index, 56:3-26. Massey, D. S., J. Arango, G. Hugo, A. Kouaouci, A. Pellegrino andJ. E. Taylor 1993 "Theories of International Migration: A Review and Appraisal," Population and Development Review, 19:431-466. Schmidt, C. M. 1994 "The Economic Performance of Germany s East European Immigrants." CEPR Discussion Paper, No. 963. London: CEPR. (Forthcoming in Quarterly Review of Economics and Finance^ Schuiz, E., U. Rendtel, J. Schupp and G. Wagner 1993 "Das Zuwanderer-Problem in Wiederholungsbefragungen am Beispiel des Sozio-oekonomischen Panels (SOEP)." DIW Diskussionspapiere, Nr. 71. Veall, M. R. and K. F. Zimmermann 1992 "Pseudo-m's in the Ordinal Probit Model," Journal of Mathematical Sociology, 16:333-342. Veith, K. 1994 "Uberlegungen zur Zuwandenmg am Beispiel Aussiedler," Informationen zur Raumentwicklung^ 5/6:363-371. Zimmermann, K. F. 1995 "European Migration: Push and Pull," Supplement to The World Bank Economic Review and The World Bank Research Observer - Proceedings of the World Bank Annual Conference on Development Economics 1994 ^ 313-342. RESEARCHNOTE J ob Displacement Effects of Canadian Arun S. Roy Human Resources Development Canada Since the 1970s, significant shifts occurred by country of origin in Canada ' s immigration flow. While the importance of Europe as a source region declined, that of Asia and Africa increased substantially. Large numbers of immigrants were also admitted from the Caribbean and Latin America (.^Marr, 1992; Hawkins, 1989). The major reason for this shift was a new immigration policy adopted since 1967 which based immigration on labor market criteria rather than the country of birth or racial origin. In 1978, another new immigration act was implemented which was more restrictive than the 1967 act. The restrictions were reflected in specific annual immigration target levels to be approved by the Parliament each year, coupled with a closer scrutiny of the immigrant ' s potential labor market impact. Entry under the independent class was linked solely to labor market requirements. Closely related to the inflow of immigrants by source was their occupational background. For instance, a relatively large number of immigrants from the United States were in the professional, technical and managerial category. Immigrants from Asian countries were concentrated in the service sector and clerical and related occupations, but large numbers were also found in the professional, technical and managerial category. European immigrants, on the other hand, more commonly were found in construction trades and in machining and product fabricating. Another development which is of some importance in terms of the possible labor market effects of immigration is a shift in the distribution by class/category of immigrants. The Independent class, which was admitted mainly on the basis of labor market criteria, dominated Canada ' s inflow until 1974. Since 1975, the importance of the Family class and the Refugee class increased l An earlier version of the study was presented at the Conference on the Economics of Immigration, University of Nebraska, Lincoln, May 1996. The author is grateful to GingWong for encouragement and comments. Views expressed in this paper are those of the author and do not necessarily represent those of the Human Resources Development Canada (formerly Employment and Immigration Canada) or the Government of Canada. ©1997 by the Center for Migration Studies of New York. All rights reserved. 0197-9183/97/3101.0117 Immigrants b y Country of Ori g in and Occupation 1 JOB DISPLACEMENT EFFECTS OF CANADIAN IMMIGRANTS 151 considerably. As the Family class and Refugee class immigrants were not selected on the basis of labor market requirements in the Canadian economy, they had a larger potential for job displacement effects. Thus, in the post-1978 period, two contradictory forces appeared to have changed the potential labor market effects of the new immigration flows. The emphasis on labor market requirements reduced the possibility of job displacement effects due to immigration. On the other hand, accommodation under family reunification partially circumvented the labor market criterion and opened the possibility of competition for jobs (DeVoretz, 1992:176). In recent years, a variety of Canadian research studies have become available on different impacts and effects of immigration (for macroeconomic impacts of immigration, .^Nakamura, Nakamura and Percy, 1992; Robertson and Roy, 1982;RaoandKapsalis, 1982; for earning differentials of immigrants and native-born, ^ deSilva, 1992; Abbot and Beach, 1993; Meng, 1987; Bloom and Gunderson, 1991; DeVoretz and Fagnan, 1990; for decline in the quality of recent immigration flows, see DeVoretz and Fagan, 1990; for demographic impact of immigrants, ^^ Beaujo t, 1992; for a comparative analysis of unemployment experience of immigrants, see Robertson and Roy, 1986; Samuel and Woloski, 1985; for costs and benefits of immigrants, .^Akbari, 1989; Simon, 1992; Swan, 1991). Empirical evidence on job displacement effects of the immigration flows for Canada has, however, been very limited. Two studies by Roy (1987; 1988), using slightly different sample selection criteria and methodologies in each, analyzed the substitutability of immigrants and Canadianborn workers. At the aggregate labor market level neither of the two studies found any substitutability between immigrants and Canadian-born workers. More recently, a study byAkbari and DeVoretz (1992) estimates the impact of immigrant workers on the employment of Canadian-born workers in selected Canadian industries. A translog production function is used and cross-elasticities are estimated. It confirmed the absence of any significant job displacement effects for the Canadian born. The present study is an extension of the production function approach to analyze the job displacement effects of immigrants or foreign born disaggregated by country of origin and occupation. THEORETICAL FRAMEWORK Immigration flows can be viewed as exogenous shifts in the supply of labor to geographically defined labor markets. An increase in labor supply through increased immigration in a given labor market would initially tend to increase competition for jobs among the immigrants themselves to the extent they are not integrated into the host-country labor market. This would reduce the market wages for immigrants. Also, immigrants usually have lower reservation 152 INTERNATIONAL MIGRATION REVIEW wages than the native-born workforce as their access to transfer payments is more limited. Depending upon their skill requirements, employers are likely to substitute immigrant labor for the native-born workers since the former is now cheaper. Thus competition for jobs in the local labor market between the native-born and immigrant workers would adversely affect the earnings of the native-born. If variations in the number of immigrants relative to the nativeborn workers across the selected local labor markets demonstrate that a hi gher ratio of foreign-born to native-born workers is associated with a lower wage rate for the native born, then the immigrants and the native born are substitutable labor inputs in production. In this case, foreign-born workers can be considered to have an adverse impact upon the earnings and employment opportunities for the native-born workforce. As one study puts it ^ an increase in new immigrants to a locale will reduce the relative earnings of new immigrants and to a lesser extent the earnings of others for whom immigrants are substitutes " (Lalonde and Topel, 1991:174). Thus, at least initially, immigrants and native-born workers operate in segmented labor markets (Lalonde and Topel, 1991:175). On the other hand, immigration flows could lead to increased wages for the native-born workers. If there are skill shortages in the host country and immigrants relieve these bottlenecks, it would expand job opportunities in general, resulting in an increased demand for labor and eventually leading to higher wages for native-born workers. In this case, immigrants and the domestic workers are complementary inputs in production. Also, the demand curve for labor could shift upward through income and market size factors. Whether foreign-born workers act as a complement or substitute for the native born and what the magnitudes of such effects are is an empirical issue. Essentially it involves the measurement of how a change in the relative quantities of two types of labor interacts on one another and affects their relative prices. MODEL SPECIFICATION The model specification for the present study is based on a production function which is linearly homogeneous and possesses the usual neoclassical properties. Specifically, we employ the generalized Leontieff production function: F(4, L,, L,, L,) = ^^ (LiLjf2 (1) (f,j = c, u, e, t) where arguments refer to the labor force of Canadians, United States, European and Third World immigrants respectively in local labor markets in Canada and where the technology parameters are restricted so that Y» = Y». JOB DISPLACEMENT EFFECTS OF CANADIAN IMMIGRANTS 153 This specification is a second order approximation to a concave neoclassical production function exhibiting constant returns to scale. Equating wage rates and marginal products for the four labor force groups leads to the following empirical specification: ^c = ^ +^(P„\Pc) l\2 +^ ce(Pe\Pcf 2 +^ cMPcf 2 +^P n^ +£ , 2(a) ^ = V^P^ 2 +^(PeW \2 +^(P^ PJ l\2 +^" x « +£ " ^ ^e=^e+^c(Pc\Pef 2 +^(Pu\Pef 2 + ^ (PtW 2 +^ Pn^ +^ 2(c) W , =Y , + Y^\^) 1\2 +Y^\p,) l\2 +Y^\p,) l\2 +^^X ,+e, 2(d) where W = a measure of earnings such as weekly wages and the subscripts c, u, e, and t refer to Canadian born, U.S., European and Third World immigrants, respectively. p^ = proportion of the U.S.-born immigrants in the working age population in the fcth labor market, and similarly pc = proportion of Canadians; pc + pu + pe + pt =1. X = a vector of usual socioeconomic and human capital variables (education, age, sex, skill characteristics, etc.), and n is the number of such variables Y//, Yjj = technology parameters which capture the own-substitution effects for i andy respectively (i = j '). Yy, Yji = technology parameters which capture cross-wage effects (i^j) and £i, £/ = error terms. Equations (2) have the interpretation that each type of labor input is paid the value of its marginal productivity. From these profit maximization conditions it is evident that the productivity of the Zth group depends not only on the relative size of this group, but also on the relative size ofthe ^th group. (For a more detailed and technical discussion of the properties of the production function used here and their interpretation, see Diewert, 1971, and more specifically Borj as, 1983.) The parameters that are of main interest in the present study with symmetry constraints imposed are Y^^^iic)/ ^ ce( = ^ ec)/ ^ct^^ tc) 9 If? f01* instance, Y cu is greater than zero, it means that an increase in the share of the U.S. immigrants relative to the native-born Canadians would increase the earnings of the Canadian born, and thus the U.S. immigrants and Canadians are complementary inputs in the labor market. This would lend support to the view that either because the skill characteristics of immigrants are different from those of the 154 INTERNATIONAL MIGRATION REVIEW domestic workforce and/or that U.S. immigrants take jobs that native-born workers arc unwilling to accept, they act as complementary rather than competing groups. On the other hand, a negative coefficient implies that they are substitutes in the labor market and the U.S. immigrants have an adverse impact upon the earnings and employment opportunities for the Canadian workers. The elasticities of complemetarity, <3>y, implied by equations (2) and its empirical counterpart in the present context have been shown to be {see Borjas, 1983) ^ = ((^W) /(2^.^.(p,^.) l\2 ), ^7 (3) and where W = (piWi + pj W j), Wi and W j are the earnings of the Zth and jth groups, respectively. It would be noted that the model specification abstracts from the role of capital. It implies that capital and various labor inputs are assumed to be separable. Ideally, it would be desirable to include the interaction between capital and inputs of labor. Difficulties in constructing a series for capital even at the aggregate level are well known. These problems are many times greater in the context of geographic labor markets. Grossman (1982) reported that capital appears to be separable from labor (in another study, Borjas, 1983, found that the assumption of strong separability between capital and labor is not rejected by the data). This means that " when researchers are estimating demand functions for immigrants, the price of capital or capital stock need not be included" (Grossman, 1982:602). DATA AND SAMPLE SELECTION The empirical analysis presented in this study is based upon the individual file of the 1981 Census of Canada (Public Use Sample Tape). A random sample of men was selected from the census file subject to a set of selection criteria which included at least 30 weeks of work during the reference year 1980 and 30 hours or more of employment during the reference week. The sample thus consisted of men who had a strong attachment to the labor market. "Immigrants " were defined as those born outside of Canada. In addition to the selection criteria enumerated above, a ten year cut-off point was used for the foreign born which meant that the sample consisted of individuals who had immigrated into Canada not longer than ten years ago. The implicit assumption here is that immigrants living in the host country for longer than ten years are more or less fully assimilated into the Canadian labor market. A summary of some relevant statistics for the Canadian born and for foreign-born groups of immigrants are presented in Table 1. The total sample consisted of 94,798 individual records, of which 912 were immigrants from JOB DISPLACEMENT EFFECTS OF CANADIAN IMMIGRANTS 155 TABLE 1 SUMMARY STATISTICS OF CANADIAN-BORN AND FOREIGN-BORN SAMPLES Third Canadian U.S. European World Observations (N) 84,488 912 3,369 2,029 Average Weekly Wage (C$) 347 487 415 323 Average School Education (Years) 10.7 12.1 10.3 11.2 Average University Education (Years) 0.5 3.1 0.9 1.6 Average Age (Years) 39 38 42 34 Residence in CMA (Percent) 46.1 63.0 79.2 89.6 Years Since Immigration (Percent) 5 years or less — 34 39 44 6-10 years — 66 61 56 Occupations (Percent) Professional, technical, and managerial 22.0 45.1 24.2 24.5 Clerical and related 10.9 6.9 6.2 11.6 Sales 11.8 12.2 6.3 9.3 Services 10.7 6.9 9.7 17.2 Primary 5.2 1.9 2.9 0.9 Processing, machining, product fabricating, assembling, 22.9 14.6 31.2 29.5 and repairing Construction 9.3 6.5 15.4 3.2 Transportation equipment operating7.25 ^ 94J3.8 Note: There are 15 occupations groups reported in census data. For the sake of brevity, some occupations have been grouped together. For Instance, professional, technical (first grouping) includes "managerial, administrative, and related," "natural sciences, engineering and mathematics," and "social sciences and related fields," "teaching and related," and "medicine and health." the United States, 3,369 were European born, and 2,029 were born in the Third World countries. The average weekly wages of the U.S.-born and European-born were hi gher and of the Third World immigrants lower than those of the Canadian-born workers. Part of these wage differentials can be attributed to observable heterogeneity such as skill, education, occupation, and geographic location. The level of education was higher among the U.S. and Third World immigrants than among the Canadian and European born. A much larger propor tion of immigrants from the Third World countries were immigrants of recent vintage (5 years or less). The occupational profiles of immigrants by country of origin varied widely There was a relatively large concentration of U.S. immigrants in the professional and managerial category (45%). These occupations, on the other hand, accounted for only 22 percent among the Canadian born. A relatively large concentration of the Europeans was found in construction trades and as production workers. Relatively large numbers of immigrants from Third World countries were found in the service sector, clerical and related occupations, and in machining and fabricating. As compared with the Canadians, there was a 156 INTERNATIONAL MIGRATION REVIEW heavy concentration of all immigrant groups in a Census Metropolitan Area (CMA, population of 100,000 or more). While 46 percent of Canadians lived in a CMA, close to 80 percent of European immigrants and 90 percent of immigrants from the Third World countries lived in a CMA. More than 70 p ercent of all immigrants were concentrated in three provinces - Ontario, British Columbia, and Quebec. MAJOR FINDINGS Impacts of Immigrants by Source Country The system of equations in (2) was estimated using ordinary least squares (OLS) method. By estimating both the unconstrained and symmetry constrained equations, likelihood ratio tests were made on the validity of the symmetry restrictions. Symmetry restrictions were valid in four cases. Table 2 presents the symmetry constrained wage equations for the four labor force groups. The regression results were generally significant and had a high explanatory power. The results for the socioeconomic variables were in the expected direction. All three human capital variables used - namely, years of school education, years of university education, and experience — had large positive effects on wages for all labor force groups (except in the case of the school variable for two immigrant groups). For instance, an additional year of university education increased the average weekly wage by $29.9 for Canadians, $35.8 for U.S. immigrants, $29.6 for European immigrants, and $15.6 for immigrants from the Third World countries, other things being the same (see Table 2). It is noted that in the case of the Third World immigrants, the increase in earnings due to an additional year of university education is much smaller than in the case of native-born Canadians. The result is not surprising. Immigrants coming from the Third World countries where the education system is somewhat different often have difficulty in getting full credit and recognition for their educational qualifications and professional credentials. Residence in an urban area (CMA) had a positive effect upon the wage rate of the Canadians. Borjas (1994:1699) observes in the li ght of similar empirical evidence in the United States that this may be masking the macro effect of immigration. In the case of immigrants, residence in an urban area, on the contrary, had a negative impact on their earnings (especially in the case of immigrants from the Third World countries). Immigrants tend to concentrate in and around large cities, and the relatively greater competition for jobs among themselves possibly exerts a downward pressure on their wages. Similar pieces of empirical evidence have been reported for the American labor markets {see Borjas, 1994:1697). JOB DISPLACEMENT EFFECTS OF CANADIAN IMMIGRANTS 157 TABLE 2 SYMMETRY CONSTRAINED WAGE EQUATIONS SHOWING JOB DISPLACEMENT EFFECTS OF IMMIGRANTS BY REGION OF ORIGIN ON CANADIAN BORN AND PARTIAL REGRESSION COEFFICIENTS FOR SOME SELECTED VARIABLES CanadiansUnited StatesEuropeThird ^X^)rld SQT(pu\pc) -40.8* -40.8* — — (-.09)** (-.09)** SQT(pe\pc) -16.2 — -16.2 — SQT(pt\pc) -17.1 — — -17.1 SQT(pe\pu) — 4.2 4.2 — SQT(pt\pu) — 9.8 9.8 SQT(pt\pe) — — -3.7 -3.7 SCHOOL 26.1* -16.2 29.2* 23.7 UNIV 29.9* 35.8* 29.6* 15.6* MARST 5.2* 8.7* 4.8* 7.2 CMA 20.0* -14.7 18.2 -20.8* YSM5 — -8.3 -18.6 -19.2* LUR -9.2* -4.6 -15.4* -29.5* EXPER 6.3* 12.7* 10.3* 4.2* EXPERT -.12* -.18* -.08 -.07* Occupation (control dummies) Constant term 94.2 126.8 -70.9 285.7 Adjusted R 2 3^7.23.21 *Regression coefficients statistically significant at 95% level of confidence. ^Elasticity estimates. An elasticity of-.09 means that a 10% increase in ratio ofU.S.-born immigrants to Canadians lowers the earning of Canadians by .9% Notes: Dependent variables for columns 1 to 4 are We, W«, Wg, and W( respectively. W = calculated weekly wage; subscripts C, U, e, and t refer to native-born Canadians, U.S. born, European born, and Third World born respectively. SQT = square root; (pu\f}c) = ratio of U.S. born to Canadians in given local labor market; similarly (pe\pc), (pt\pc), etc. as defined above. MARST = 1 if married, 0 otherwise. SCHOOL = numbers of years of school education. UNIV = number of years of university education. CMA = residence in a CMA (population of 100,000 or larger), a dummy variable used as proxy for urban labor market (equal to 1 if urban residence, 0 otherwise). YSM5 = dummy variable with value of 1 if residence in Canada for less than 5 years. LUR = local employment rate in 1980. Each individual assigned to UI economic region. EXPER = AGE - SCHOOL - UNIV - 5 (years). For estimates by occupation categories, similar regression format for each occupation category was employed. Immigrants who had been in Canada for a relatively short period of time, less than five years, tend to earn a lower wage than those who have been in the country longer. Immigrants from the Third World countries appear to be par ticularly disadvantaged on account of this. The length of residence probably captures the Canadian experience or the lack of it. The result is understandable as it takes time for immigrants to learn new trades and acquire new kinds of skills that are more acceptable in the labor market of the host country. The process of being integrated into the institutional structure of the labor market of a foreign country is necessarily time-consuming. 158 INTERNATIONAL MIGRATION REVIEW A hig her local unemployment rate has a significant negative impact, as might be expected, on the earnings for all groups (except for U.S.-born immigrants). 2 The impact is especially large in the case of immigrants from the Third World An increase in the unemployment rate by one percentage point reduces the earnings of immigrants from the Third World by almost twice that of European immigrants and almost three times that of the native-born Canadians. Results would tend to indicate that immigrants in general bear a heavier brunt of a recession than others in terms of job losses and lower earnings. When immigrants are disaggregated by country of origin, the U.S. immigrants and the Canadians are found to be substitutes or competing groups in the labor market. A 10 percent increase in immigrants from the United States would tend to reduce the wage rate of the native-born Canadians by 0.9 percent. The result may not appear to be surprising in view of the relatively large concentration of immigrants in the professional and managerial occupations. Also, the Canadian and American labor markets are similar in many respects, including the technology in the work place, their institutional structure, and language that is spoken. When considered as aggregate groups, neither the European-born nor the Third World-born immigrants had any job displacement effects on the Canadians. However, neither were they complementary to the native-born Canadians. The impact of these two immigrant groups on the labor market was statistically insignificant. Also, between any two immigrant groups, there was no evidence of substitutability or complementarity. Substitutability b y Occupation Separate regression equations were estimated by country of origin of immigrants disaggregated by major occupation groups (two digit level). In some cases, the number of observations was too small to carry out the statistical tests and, therefore, no inferences were drawn. Wherever sample size permitted, the results showed that Canadians and European-born immigrants were substitutes in clerical occupations, services, and processing occupations. On the other hand, they had skills that were complementary with those of the Canadian born in " natural sciences, engineering, and mathematics " and transportation equipment operating occupations. ^ocal unemployment rates used consisted of those generated by Statistics Canada from the Labor Force Surveys for the Department of Employment and Immigration for purposes of administration of the Unemployment Insurance (UI) program. A UI economic region consists of several Census Divisions within a province. Individuals were assigned to a UI economic region depending upon their geographic location. For construction of the series on local unemployment rates, see Statistics Canada (1992). JOB DISPLACEMENT EFFECTS OF CANADIAN IMMIGRANTS 159 U.S.-born immigrants had large and statistically significant job displacement effects on Canadians in ' natural sciences, engineering, and mathematics " and ' managerial, administrative, and related occupations. " They had complementary skills in teaching and related occupations. Canadians and immigrants from the Third World countries were found to be substitutes in ' machining, product fabricating, assembling, and repairing " and to a smaller extent in transportation occupations. 3 For all other occupation categories, immigrants were neither substitutes nor were they complementary with the Canadian-born workforce. SUMMARY AND CONCLUDING REMARKS A variety of different complex factors must be considered in analyzing and understanding whether immigrants are in competition with the native born in the job market. The size of the immigrant group relative to the native born, occupations and industries the immigrants move into, their geographic location in the host country, time of their arrival, the period of their stay in the host country, and their skills relative to those of the native born are some of the crucial factors that determine whether the foreign born would tend to displace the native-born workers from the labor market. Some previous Canadian studies have shown that considering the labor market as a whole and also pooling all immigrants as a group, immigrants do not have any job displacement effects on the Canadian born. This study presents some new evidence. It disaggregates immigrants by country of origin and by occupation groups and provides an analysis of job displacement effects of immigrants on the native-born Canadians by these dimensions. The study finds that 1) U.S. immigrants and the Canadians are substitutes or competing groups in the labor market and the effect is quite significant; 2) Canadians and Europeans are competing groups in certain occupations, while they have complementary skills in others; and 3) immigrants from the Third World and the Canadians are slightly competing groups in certain occupations. Thus when the analysis is carried out at an aggregative level, there could be some concealed job displacement effects. This study, therefore, underscores the importance of disaggregated analysis of job displacement effects of immigrants by country of origin and their occupational skills and background. The present analysis is a cross-sectional one and is thus incapable of capturing the time dimension of job displacement effects of immigrants. It focuses only on whether stocks of immigrants or immigrant groups and Canadian workers are substitutes or complements at one point in time. Immigrants who are substitutes with the Canadian workforce at one point in retailed results by occupation are available upon request. 160 INTERNATIONAL MIGRATION REVIEW time may not be so at other times and vice versa. It is thus important to update this kind of analysis periodically. Also, the analysis focuses on the micro aspect to the extent an inflow of immigrant workers affects levels of earnings of the native-born workers in the local labor markets. It does not capture the various macroeconomic effects that immigration flows have on the consumption expenditure, investment, employment and income, and the longer-run population growth and demographic structure of the economy that have been documented elsewhere. REFERENCES Abbott, M. G. and C. M. Beach 1993 "Immigrant Earning Differentials and Birth-Year Effects for Men in Canada," Canadian Journal of Economics, 26(3):506-524. Abowd, J. M. and R. B. Freeman, eds. 1991 Immigration, Trade, and the Labor Market. Chicago, IL: University of Chicago Press. Akbari, A. 1989 "The Benefits of Immigrants to Canada: Evidence on Tax and Public Services," Canadian Public Policy, 15(4):424-435. Akbari, A. and D. J. DeVoretz 1992 "The Substitutability of Foreign-Bom Labor in Canadian Production circa 1980," Canadian Journal of Economics, 25(3):604—6l4. Bartel, A. P and M. J. Koch 1991 "Internal Migration of U.S. Immigrants." In Immigration, Trade, and the Labor Market. Ed. J. M. Abowd and R. B. Freeman. Chicago, IL: University of Chicago Press. Beaujot, R. 1992 "The Socio-demographic Impact of Immigration." In The Immigration Dilemma, Ed. S. Globerman. Vancouver, B.C.: The Fraser Institute. Bloom, D. E. and M. Gunderson 1991 "An Analysis of the Earnings of Canadian Immigrants." In Immigration, Trade, and the Labor Market. Ed. J. M. Abowd and R. B. Freeman. Chicago, IL: University of Chicago Press. Borjas, G. 1994 "The Economics of Immigration," Journal of Economic Literature , 22:1667-1717. 1983 "The Substitutability of Black, Hispanic, and White Labor," Economic Inquiry 21:93-106. Chiswick, B., C. Chiswick and P.Miller 1985 "Are Immigrants and Natives Perfect Substitutes in Production?" International Migration Review, 19 (4); 674-68 5. deSilva, A. 1992 Earnings of Immigrants: A Comparative Analysis. Study prepared for the Economic Council of Canada, Ottawa. DeVoretz, D. J. 1992 "Immigration and the Canadian Labor Market." In The Immigration Dilemma. Ed. S. Globerman. Vancouver, B.C.: The Fraser Institute. JOB DISPLACEMENT EFFECTS OF CANADIAN IMMIGRANTS 161 DeVoretz, D. J. and S. Fagnan 1990 "Some Evidence on Canadian Immigrant Quality Decline." Vancouver, B.C.: Simon Fraser University. Discussion paper. Mimeo. Diewert, E. 1971 "An Application of the Shephard Duality Theorem: A Generalized Leontieff Production Function, " Journal of Political Economy, 79:481-507. May/June. Globerman S., ed. 1992 The Immigration Dilemma, Vancouver, B.C.: The Fraser Institute. Grossman, J. B. 1982 "The Substitutability of Natives and Immigrants in Production," Review of Economics and Statistics, 64:596-603. Hawkins, F. 1989 Critical Years in Immigration, Montreal: McGill-Queens University Press. Lalonde, R. J. and R. H. Topel 1991 "Labor Market Adjustment to Increased Immigration. " In Immigration, Trade, and the Labor Market. Ed. J. M. Abowd and R, B. Freeman. Chicago, IL: University of Chicago. Marr, W L. 1992 "Post-War Canadian Immigration Patterns." In The Immigration Dilemma. Ed. S. Globerman. Vancouver, B.C.: The Fraser Institute. Meng, R. 1987 "The Earnings of Canadian Immigrants and Native-Bom Males," Applied Economics, 10:1107-1119. Nakamura, A., M. Nakamura and M. Percy 1992 "Macroeconomic Impacts of Immigration." In The Immigration Dilemma. Ed. S. Globerman. Vancouver, B.C.: The Fraser Institute. Rao, S. and C. Kapsalis 1982 "Labor Shortages and Immigration Policy, " Canadian Public Policy, 8(3):377-383. Robertson, M. and A. Roy 1986 "Immigration Flows to Canada: An Analysis of Comparative Unemployment Behaviour and Labor Market Complementarity." Paper presented at the Fifth World Congress of the Econometric Society, MIT, Harvard University. 1982 "Labor Shortages and Immigration Policy. A Reply, " Canadian Public Policy 8(3):383-387. Roy, A. S. 1988 "Arc Immigrants and Canadian-Bom Workers Substitutes?" Paper presented at the Annual Meeting of the Southwestern Society of Economists, San Antonio, Texas. March. 1987 "An Analysis of Substitutability and Complementarity of Immigrants and Canadian-Born Workforce, Labor Market Bulletin, 4(9):5-11. Samuel, T. J. and B. Woloski 1985 "The Labor Market Experiences of Canadian Immigrants." Paper presented at the Canadian Economic Association meeting, Guelph. May. Simon, J. L. 1992 "The Economic Effects of Immigration: Theory and Evidence." In The Immigration Dilemma, Ed. S. Globerman. Vancouver, B.C.: The Fraser Institute. Statistics Canada 1992 Guide to Labour Force Survey Data. Catalog No. 71-528. Ottawa: Statistics Canada. Swan, N. 1991 Economic and Social Impacts of Immigration. Ottawa: Economic Council of Canada. RESEARCHNOTE Remigration: Return of the Prodigals — An Anal y sis of the Impact of the Cycles of Mig ration and Remigration on Caste Mobility The focus of this research note is the migration of the Patidar community to East Africa — and remigration to Gujarat, India. The primary motive for migration of the immigrant Patidars was to work, accumulate money and return to India, claiming a hi gher caste status. By 1931, a sufficient number of the community had become economically affluent and were given a higher caste status by the census enumerators. This study illustrates the transient nature of Indian migration to East Africa and its impact on caste mobility. Migration and remigration made it possible for the Patidars to shift social positions, i.e ^ between the modern and traditional systems of stratification. The move between the two systems is critical to the understanding of mobility in the Indian context. Caste mobility, unlike class mobility, is collective and in this instance codependent on outmigration. For caste mobility to be meaningful, the entire caste has to move vertically in order to establish a hi gher caste claim. DATA SOURCES This study analyzes census and migration data from Kenya and India from 1911 to 1939. The Kenya Reports on migration show the total volume of migrants entering and leaving the East African territories through Mombasa, the only port of entry. The migration and remigration pattern of Indians into and out of East Africa was recorded by the Kenya Department of Statistical Research in Annual Reports of Migration Statistics in East Africa, which were compiled from shipping company s schedules, submitted by passengers from 1924 to 1930. However, in 1930 detailed information was sought by the Governors Conference and Immigration Authorities, which led to new forms being issued soliciting detailed information. Reports subsequent to 1930 make distinctions regarding the places to which migrants were headed. However, the Kenya Colony remained targeted by immigrant groups because it was the most developed. An annual report on migration was published every year after 1930 except in the year 1933. ©1997 by the Center for Migration Studies of New York. All rights reserved. 0197-9183/97/3101.0117 Vibha Purl Chandra California State University^ Hayward IMPACT OF MIGRATION AND REMIGRATION ON CASTE MOBILITY 163 This data was co-related with the data available in the Indian census for Baroda, Gujarat, thus enabling discernment of the regular pattern of movement. In 1921, the Indian census recorded the movement of Indians overseas. Simultaneously, Kenya recorded a higher number of Indian migrants in the colony, during the period 1924-1926. In 1931, the Indian Baroda census noted the greatest surge in population and attributed it to overseas return migration. In the same year, Kenya recorded the highest number of Indian emigrants leaving the colony. The Indian census attempted to maintain records of the movement of populations overseas but readily admits that their numbers underrepresent the true nature and degree of this movement. Nonetheless, the Indian census is emphatic that the Patidar caste has the hi ghest number of participants in the movement of population from India to East Africa. HISTORICAL BACKGROUND At the turn of the present century, Indian attitude towards migration depended on caste status. For the upper castes, migration was forbidden because it was a polluting enterprise. Migration held no stigma for the lower castes because they were born contaminated. To bolster their claim to a higher caste status, the Patidars formulated Brahmanical caste rules to ban supposedly 'impure ' customs such as widow remarriage, bride price, meat eating, and group participation in the worship of a mother goddess. They substituted these practices with pure ' Brahmanical customs like dowries, vegetarianism, and the worship of Lord Krishna (Hardiman, 1981:41-42). In tracingthehistoryofthe Patidars, Pocock (1972:71) attributed the severe famines from 1899—1902 as the leading cause of migration to East Africa. Within 50 years, the industrious Patidars in East Africa established themselves as an affluent trading community, reinvesting their wealth in land and buildings back home in Gujarat. It was not the affluent Patidars who went to East Africa; rather it was the poor members of the community who migrated in the early 1920s. These migrants sent large remittances home by the mid-1950s. The capital resulted in improving the land by increasing the implementation of facilities like boring new tube wells. There was an increase in tobacco cultivation (a cash crop), aided by good monsoons, in the 1930s. These combined with a " rise in prices of agricultural produce during the Second World War, heralding the return to prosperity " (Hardiman, 1977:70). Thus it was an opportune moment to legitimize the claim to a hi gher ritual status. This remigration coincided with other variables such as the numerical strength and political presence of the Patidar caste, their reputed wealth and 164 INTERNATIONAL MIGRATION REVIEW the fact that they had adopted a Brahmanical life style by Sanskritizing 1 their rituals and traditions. These factors conspired to remove the last barriers and doubts the census authorities may have had regarding their claim to a higher caste status. The identity transformation was symbolized in a new name, Patidar, as opposed to their former function/tribe name, Kanbi. 2 THE INDIAN CENSUS RECORDS AND IMMIGRATION Unfortunately, detailed records of early Indian migration are not available. The Baroda branch of the Census of India constantly bemoans the difficulty in accounting for immigrants from the region (United Kingdom, 1921:105, 19 31 a: 9 6). Since 1911, census authorities attempted to acquire the exact number of people residing outside India. They reported that figures available underestimated the overseas movement considerably. Nonetheless, the census provided the numbers shown in Table 1. The numbers show an increase in immigration by 295 percent in twenty years from all divisions except northern Gujarat. Immigration to East Africa also became important during 1921-1931. In 1921, 317 immigrants had gone to East Africa, whereas in 1931, 1,528 had migrated from the region (Table 2). The caste with the highest number of immigrants was the Patidar caste, with 2,532 migrants (United Kingdom, 1931a:96). A comparison of Table 1 with Table 3 shows that the Census of India underestimated the duration of stay as well as the number of Indians overseas. The estimated number of Indians in the East African territories, as allocated in the Indian census (Table 4) shows higher numbers. These figures accounted for all the migrants from India. The data is frequently contradictory, and is ambiguously labeled by the Census of India. KENYA REPORTS That a large number of these immigrants went to East Africa is evidenced from the Kenya figures, where 7,369 Indians entered the territory in 1931 (United Kingdom, 1932:9). This brought the Indian population to 50,281. However, the Kenya figures for 1920—1930 show an inward movement of 70,615. If the ^A process through which the lower caste moved up in the caste system. The term was coined by M.N. Srinivas (1978). Sanskritization and Westernization were "twin " processes; as the lower castes emulated the upper caste rituals and lifestyies, the upper castes were increasingly Westernizing. ^he Kanbis of Gujarat have three divisions namely, Lewas, Kadwa, and Anjana. Lewas, at the top of the hierarchy, led in numerical strength (United Kingdom, 189 la: 159). Their inclusion into the Hindu fold is attributed to following a peaceful occupation like agriculture and being completely subservient to the Brahman caste (United Kingdom, 1891b:122). IMPACT OF MIGRATION AND REMIGRATION ON CASTE MOBILITY 165 TABLE 1 BARODA SUB J ECTS RESIDING OUTSIDE INDIA Natural Division 1931 1921 1911 Central Gujarat 2,471 844 94 Northern Gujarat 325 387 810 Southern Gujarat 6,687 3,802 2,499 Kathiawad 1,007 377 152 Total10,4905,4103,555 Source; United Kingdom. 1931. Census of India, Report Vol. XIX, Part I, Baroda. London, New Delhi: India Office Library; National Archives. TABLE 2 NUMBERS OF EMIGRANTS FROM BARODA, GUJARAT Country19311921 East Africa1,528317 Source: United Kingdom. 1931. Census of India, Report Vol. XDC, Part I, Baroda. London, New Delhi: India Office Library; National Archives. P.96. TABLE 3 ESTIMATED YEARS OF RESIDENCE OVERSEAS Period of ResidenceNumber Below 1 Year 996 Below 1 & 3 Years 2,406 Below 3 & 5 Years 2,038 Below 10 Years8,465 Source: United Kingdom. 1931. Census of India. Report Vol. XIX, Part I, Baroda. London, New Delhi: India Office Library, National Archives. P 97. TABLE 4 CENSUS ESTIMATES FOR INDIANS IN EAST AFRICA Country 8 Population Tanganyika 23,422 Zanzibar 15,246 Uganda11,613 Source: United Kingdom. 1931. Census of India, Report Vol. XDC, Part I, Baroda. London, New Delhi: India Office Library; National Archives. P 71 "East African territories. outward movement, /.^., 49,884, is subtracted it still leaves an inward excess of 20,731 (United Kingdom, 1930:4). These numbers reveal the presence of a constant Indian population base capable of supporting newcomers from the home country. There were several restrictions imposed upon Indian immigrants in East Africa. Incorporated in a large anonymous ethnic hierarchy, the Indians were discriminated against through legislation, impeding their full participation in East Africa. The fear of Indian hordes inundating the colony led to proposals restricting Indian migration into the colony. The British voice in championing the cause of the downtrodden African native further frightened Indians with 166 INTERNATIONAL MIGRATION REVIEW capital at risk. The British had always complained about the lack of Indian commitment to the colony. A greater charge rested on the fact that the Indians did not reinvest their profits in the colony, but preferred to remit them to India. Evidence of this is found in the Kenya migration statistics, where not only Indians but all migrants, including European, are classified as temporary Most immigrants remigrated to the home country with their savings. The Indian population in the Kenya colony had increased by 272 percent since 1911, with 99.9 percent of Indians in the colony born British subjects (United Kingdom, 1931b:7). This means the immigrants were younger, i.e ^ born after 1858 when India came directly under the rule of the Crown. In 1938, it was observed that Indians came in large numbers when economic conditions were conducive and left when the economy and employment opportunities were low (United Kingdom, 1938:6). Indian migration in this sense was almost self-regulatory. The 1939 Kenya Report observed the same phenomenon and noted that the trend either way was always set by Indians (United Kingdom, 1938:6). Table 5, Indian migration from East African territories via Mombasa, shows an inward excess of 3,593 Indians in 1924. By 1926, the inward excess peaked at 5»224, then began to decline again. In 1931, emigration exceeded immigration with the record number of 13,153 Indians leaving the colony. The 1929 worldwide depression (which actualized late in East Africa) and the outbreak of war in 1938 caused a higher number of Indians to emigrate from the colony. According to the Kenya Annual Report, Indian exodus from the colony is the highest as compared to the Arabs, Goans, Europeans, and others (United Kingdom, 1934:35? 1939:2) because the Indians had the largest immigrant pool in Kenya. Between 1924 and 1930, European migration into the Extra Kenya Territories was 15 percent of the total in-migrating population as compared to Indian migration, which was 23 percent (United Kingdom, 1930:5—6). The 1930 Kenya Report also mentions that 47 percent of the Indians who entered the colony were entering for the first time. Only 31 percent had resided there over five years (United Kingdom, 1930:17). In Kenya the authorities were concerned about the decreasing population. A population gain over a period of seven years by 1931 translated to an equally large loss with 34 percent of the Indians leaving the colony (United Kingdom, 1931:2, 8). Of the Indian emigrants in 1931, 80 percent were primarily in the productive age group of 20-44, and these adults had resided in the colony less than five years (United Kingdom, 1931c:9). The Annual Report on Statistics of Migration for 1933 shows that 98 percent of Indian migration was to and from India directly (United Kingdom, 1934:34). In July 1931, a statistician in the colony of Kenya observed that 33 percent of Indian emigrants had resided less than a year. He attributed that to tourists IMPACT OF MIGRATION AND REMIGRATION ON CASTE MOBILITY 167 TABLE 5 INDIAN MIGRATION VIA MOMBASA^ 1924-1939 ^earImmigrantsEmigrantsInward Excess 1924 7,973 4,380 3,593 1925 9,540 4,865 4,675 1926 9,753 6,529 5,224 1927 10,156 7,045 3,111 1928 9,580 7,227 2,353 1929 11,768 7,793 3,975 1930 11,845 12,045 -200 1931 9,629 13,153 -3,524 1932 7,799 10,654 -2,855 1933 7,776 9,644 -1,868 1934 10,364 8,857 1,507 1935 10,745 9,479 1,266 1936 12,118 10,228 1,890 1937 15,959 10,889 5,070 1938 14,111 11,664 2,447 193911,18514,284-3,099 Source: United Kingdom. 1939. Annual Report on Migration for the Colony and Protectorate of Kenya, Native Population and Statistical Section. P 4. Migration via the port ofMombasa to and from the East African Territories (Kenya, Uganda, Tanganyika, and Belgian Congo), excluding African natives. and temporary visitors leaving the colony to return home (United Kingdom, 1930:19). In 1939, the Annual Report noted that Indians, unlike Europeans, had an alarmingly high rate of emigration from the colony. It elaborated further by stating it was not only economic considerations that were responsible for the Indian exodus but also psychological fears of the spread of hostilities that were driving the Indians away (United Kingdom, 1939:4). This was one of the primary reasons for Indian emigration from East Africa. After 1931, the largest outward excess was -3,099 in 1939 (United Kingdom, 1939:6), which was also the period of the outbreak of World War II. Even as late as 1948, the Migration Report is unable to explain the high percentage of emigrants leaving the colony after residing for only a year — 62 percent of the Indians and Goans left the colony within a year of arrival (United Kingdom, 1948:10). The Kenya migration statistics merely list Indians, not specific castes or groups involved in these movements. The census also does not provide an adequate sense of who these Indians were; all Indians are put into one category The Kenya census, however, made an attempt to differentiate Indians according to their religious beliefs: there were Hindus, Mohammedans, Zorastrians, Jains, and Sikhs. The Hindus were further subdivided into various subcategories as given in Table 6. These categories are vaguely defined based on religious beliefs. The 168 INTERNATIONAL MIGRATION REVIEW TABLE 6 CENSUS ESTIMATES FOR INDIANS PER RELIGION ReligionPopulation Hindu 15,908 Sanatan Dharmist 1,233 Arya Samaj 922 Brahmin 119 Rajput 76 Vaishya 32 Sweepers49 Source: United Kingdom. 1931. Report on the Census Enumeration of Non-Native Population Made in Colony and Protectorate o/Keny on the Night of March 6, 1931. P.47. last four groups are part of the Hindu caste system yet they are enumerated separately. No satisfactory explanation is provided with the table, only the emphasis that caste and reli g ion was the same for Indians (United Kingdom, 1931:47). The census also provided a table compiling the years of residence for different ethnic groups. The same dismal picture emerged, where 52 percent of the Indians left the colony within one to five years of residence. Although, 53.1 percent of the Goans also left within the same time frame, Indians still constituted the higher numbers because Goan totals were 3,979 as opposed to 39,644 for the Indians. In terms of duration of residential stay, Indians were second only to the Arabs in length of residence, 51-55 years (United Kingdom, 1931b:48-49). THE INDIAN CENSUS RECORDS The outward movement of Indian migrants recorded in the Kenya Statistics were complimented by the Indian census as inward movement. The Census of India recorded part of the inward movement of the Patidar community from East Africa. However, the figures were underestimated. The return migrants formed a substantial flow into Gujarat, warranting some explanation by the census authorities. In 1931, Charotar, the Patidar ' s home in central Gujarat, recorded the highest population increase. Three-quarters of the 10.6 percent increase was attributed to gain through return migration (United Kingdom, 1931a:39-40). The census reiterated that the remigration figures underestimated the return flow of the volume of the overseas migration. The report concluded by noting the excess of 53,937 returned migrants. The leading Hindu caste among the return migrants was the Patidars (193 la: 89-96). The report admits that the data is unsatisfactory because it is based on guess work (United Kingdom, 1931a:97). The numbers are not a true picture of the extent of the movement; however, they are significant because they observe the overall movement to East Africa and back. IMPACT OF MIGRATION AND REMIGRATION ON CASTE MOBILITY 169 By 1931, the Patidars were not only numerically the strongest group, they were also a politically dominant community in the region. They had participated in the national movement to overthrow the British and had come into political prominence (Hardiman, 1981:40—42). Sanskritization of their rituals and lifestyles, made possible through an accumulation of wealth, transformed them into an economic and political elite. They had left no stone unturned in their bid to stake a higher caste status, including remigration with accumulated wealth from East Africa. The 1929 worldwide depression signaled the Patidars return home with their savings by 1931. When a high percentage of the community succeeded economically, they joined the existing Patidar elites in India. Their numbers and their prosperity was quite spectacular. No longer could they be denied a higher caste status. In 1931, the census recorded the transformation with a nonchalant explanation, <( ... the great agricultural communities, who seemed to dislike their old name of ' Kanbi/ were thereupon shown under Patidar.... " (United Kingdom, 193 la:394). The Kanbis were transformed into the respectable Patidars. CONCLUSION The migration of the Patidar community to East Africa advanced the relatively lower caste Kanbi to the middle of the ethnic hierarchy in Africa. In India, the process of transforming an identity from the lower caste to an upper caste status was difficult because the Kanbis confronted the barrier ofBrahmanical rituals, which entailed tremendous financial expense. In Africa, another identity awaited formation - that of ethnicity predetermined by skin color, which is itself intersected by economic class. The Patidars rejected the ethnic hierarchy and remigrated to India to celebrate their collective movement to a higher caste status. Through the mechanism of migration and remigration, the Patidars literally brought themselves out of the oppression of both caste (in India) and ethnicity (East Africa) by ascending to a higher caste status on their return to India. The Indian census dutifully provided the seal of approval where the identity transformation is recorded in a new name, Patidar. REFERENCES Hardiman, D. 1981 Peasant Nationalists ofGujarat. Delhi: Oxford University Press. 1977 "The Crisis of the Lesser Patidars: Peasant Agitations in Gujarat, 1917-1934. " In Congress and the Raj. Ed. D. A. Low. Columbia; South Asia Books. Pocock, D. F. 1972 Kanbi and Patidar. Oxford: Clarendon Press. 170 INTERNATIONAL MIGRATION REVIEW Srinivas, M. N. 1978 Caste in Modem India and Other Essays. Bombay: Asia Publishing House. United Kingdom 1948 Report on Migration for 1947. Prepared under direction ofC. J. Martin. Director, East African Statistical Department. February. London: Commonwealth Library. 1939 Annual Report on Migration for the Colony and Protectorate of Kenya, Native Registration and Statistical Section. Colony and Protectorate of Kenya. London: Commonwealth Library. 1938 Annual Report on Migration for the Colony and Protectorate of Kenya. Native Registration and Statistical Section. Colony and Protectorate of Kenya. London: Commonwealth Library. 1934 Annual Report on Statistics of Migration Through the Port ofMombassa, Statistical Section. Colony and Protectorate of Kenya. Native Registration and Statistical Department. London: Commonwealth Library. 1932 Statistics of Migration Through the Port ofMombasa. Department of Statistical Research. British East Africa. Memoir No. 8. Prepared under the Direction of A. Walter. London: Commonwealth Library. 1931a Census of India, Report Vol. XDC, Part I, Baroda. London; New Delhi: India Office Library; National Archives. 1931b Report on the Census Enumeration of the Non-Native Population Made in the Colony and Protectorate of Kenya on the Night of the 6th March 1931. London: Commonwealth Library. 1931c Statistics of Migration Through the Port ofMombasa. Department of Statistical Research. British East Africa. Memoir No. 7. Prepared Under the Direction of A. Walter. London: Commonwealth Library. 1930 Statistics of Migration Through the Port ofMombasa, Department of Statistical Research. British East Africa. Memoir No. 3. Prepared Under the Direction of A. Walter. 1924-1930. London: Commonwealth Library. 1921 Census of India. The Imperial Tables. Vol. XVII, Part II, Baroda. London; New Delhi: India Office Library; National Archives. 1911 Census of India, The Imperial Tables. Vol. XVI, A, Part II. London; New Delhi: India Office Library; National Archives. Census of India, Vol. XXIII, Part II, Baroda. London; New Delhi: India Office Library; National Archives. Census of India. Vol. XXIV, Part I, Baroda. London; New Delhi: India Office; National Archives. DOCUMENTA TION Australia: From Migrant Country to Multicultural Nation 1 Most countries exhibit some specific " ecology of diversity " (Muhlhausler, 1992), but Australia, as the only continent with a multi-ethnic population to be governed as a single political entity, merits particular attention. It may even be regarded as a laboratory for multiculturalism, because of the way it has " managed" its internal pluralism, especially in relation to the settlement of ethnically diverse immigrant groups. From a global perspective it is, therefore, important to examine to what extent Australia is able to make a contribution to our understanding of how diverse groups can maintain and develop their cultures while coexisting and interacting within the framework of a single multi-ethnic state. The proposition that is advanced here is that to succeed, more is required than political democracy to maintain the stability of multi-ethnic states. There is, in addition, the need for fully fledged cultural democracy (South Australian Task Force to Investigate Multiculturalism and Education, 1984). In the case of Australia, there is little doubt that any success in establishing both political and cultural democracy within the country will extend its ability to build intercultural bridges beyond its boundaries and help to forge transnational links within the Asian-Pacific region and globally. While much attention has been given to the political aspects of such international cooperation, it is the intercultural barriers (both within a country and with its neighbors) that are most intractable and need our closest attention (Price, 1993). THE POLITICAL DEMOCRACY DIMENSION The extent of Australia ' s success in bringing together people from well over 100 ethnic backgrounds has undoubtedly been greatest in the sphere of political democracy. Virtually all newcomers who arrive on resident visas, have no criminal record and wish to be naturalized, are accepted into citizenship. What is more, citizenship has automatically been g iven to Australian-born children of foreign-born residents, irrespective of whether the parents have actually been naturalized. excerpted from an address to the preconference to the 1995 Global Diversity Conference held at The University of Adelaide, April 21, 1995. ©1997 by the Center for Migration Studies of New York. All rights reserved. 0197-9183/97/3101.0117 J. J Smolicz University of Adelaide 172 INTERNATIONAL MIGRATION REVIEW This achievement has failed to be replicated in a number of the other countries that have enormous accumulations of resident aliens and their descendants. The issue of citizenship for guestworkers and their families is already a political time bomb in Germany, which harbors the ideal of a monoethnic state (Smolicz, 1990). Japan is bound to face similar problems in the future (Horio and Platzer, 1990) unless the Australian example of political democracy is followed. Here is an opportunity for Australia to engage in the export of the theory and practice of multiculturalism, a commodity vastly different from the mineral resources which it currently ships to Japan, but ultimately of possibly greater significance to the future of Japan and relationships between both countries. Australia ' s success at the political level means that it has managed to create a multi-ethnic state with no ethnic political parties, with very little real political friction between the various communities, and with political equality for all citizens, whatever their ethnocultural background (Jupp, 1989; Jupp and Kabala, 1993). The current frissons among some of the ethnic communities are relatively minor in comparison with the deep rift that existed between the Anglo- and Irish-Australian communities at the time of the World War I - now virtually a forgotten part of the country 's history (Kiernan, 1988). Australia ' s multi-ethnic state does not confer any legal status upon the existing ethnic groups, with the exception of Aboriginal people who are accorded specific land rights, as well as being able to invoke certain legal and other provisions acknowledged as due them as the original occupants of the land (Bourke, Bourke and Edwards, 1994; Bayne etaL, 1988). Other ethnic groups have not received any specific constitutional recognition, in contrast to those countries with "legislated" ethnic groups, such as Malaysia where there is legal distinction by " race " or Israel where differentiation is related to " religion. " In Switzerland, distinctions among citizens by language are territorially legislated through the naming of French, Italian, or German (and in special areas Romansch) as the official languages of specific cantons or communes within cantons (Watts, 1991). Although legislation for citizenship by ethnicity does not apply in Australia, the ethnic presence has been acknowledged through the establishment of government departments or commissions of ethnic affairs. In fact, ethnic groups are assumed to exist for as long as par ticular individuals choose to identify themselves with a particular ancestry and/or preserve a par ticular culture in conformity with other group members (Smolicz, 1991a; Wooden etaL, 1994). In terms of popular usage, the occurrence of ethnic variation is normally associated with linguistic, cultural, reli gious, and racial or ' visible " minorities. The limitation of the term " ethnic " to minorities is, however, misleading, since it is equally applicable to the Anglo-Celtic majority (with its numerous subgroups, such as English, Scottish, ^?(felsh, Irish, Cornish, etc.). There is an increasing FROM MIGRANT COUNTRY TO MULTICULTURAL NATION 173 appreciation that the application of the word ' ethnic " to all groups makes for the equity of all Australians (Smolicz, 1992). Former designations, such as "New Australian " or " migrant, " tended to single out the new arrivals from non-English-speaking backgrounds as somehow less than ' real Australians, " who were deemed to be descended from the dominant group. Once ethnicity is understood as a universal phenomenon that embraces everyone, whatever their background, the issue arises of how successful Australia has been in giving cultural consideration to all ethnic groups, both majority and minority, and the extent to which they can all contribute to the Australian nation. THE CULTURAL DEMOCRACY DIMENSION It is important to examine, therefore, the degree to which Australia has been able to parallel its success in the political field with the development of cultural democracy. The dimension of culture pertains to the idea of a nation and national identity, rather than to a state. Such considerations throw into sharp focus the distinctions between a state, as a political and territorial entity, and a nation, which evokes much deeper sentiments related to peoples 5 perceptions of their uniqueness and place in history. In Australia over the past two decades, these ideas have undergone some dramatic transformations. The assumption of the Australian nation-state, in which the state could be identified almost completely with the Anglo-Saxon (and subsequently Anglo-Celtic) nation has been superseded by the ideology of multiculturalism. In this way the unfortunate, largely nineteenth century Western European coupling of nation and state into " nation-state " like a pair of Siamese twins who could never be separated from each other, is at last being shown as misleading in an ethnically plural society. The unity of the state, as manifested by its common citizenship and its institutional structures, is being complemented by the recognition, toleration, and acceptance of cultural diversity at a national level. This diversity is not, however, some kind of random mish-mash of beliefs and styles of life but is intended to operate within the framework of a consensus on fundamental values that is applicable to all Australians. The implication here is that we could have a nation built on multicultural principles. Ever since the prime ministership of Malcolm Fraser (1981), multiculturalism as a national ideal has sought to uphold and develop an overarching framework of Australian values in which the rig ht of individuals from minority backgrounds to maintain their ethnic identity was assured. Within such a context, the development of ethnic identity presupposes support for cultures contributing to the overarching framework, while maintaining the central elements or core values (Smolicz, 1981, 1988, 1991b) that are essential for their integrity and creative force. 174 INTERNATIONAL MIGRATION REVIEW The balance between sharing and diversity rests on the degree of consensus which has developed on a number of cultural and political issues in areas such as democratic forms of government, the economy and the law, as well as exclusion from the framework of any requirement for uniformity in relation either to race and ancestry or to religion. This idealized model to which Australia aspires is therefore free from the divisions that are most difficult to bridge, as when one par ticular religion is made mandatory or when racial or ancestral characteristics are regarded as exclusion markers that set the limits of nationhood. However, the Australian model has had different interpretations, according to the degree ofmulticulturalism that people have been prepared to accept. Some have perceived the shared cultural framework to be essentially dynamic in its capacity to adjust to the existing, as well as the future, complexity of the population. They point to the fact that the framework has already proved its flexibility through the belated political incorporation of Abori ginal-Australians within the arch as voters since 1967, and the abolition of the White Australian Policy, sealed by the acceptance of the Vietnamese refugees during the 1970s (Stretton and Fennimore, 1988), even if much remains to be achieved in the sphere of " reconciliation " between the descendants of immigrants and the indigenous population (Roberts, 1994:212—236). Others, afraid of fragmentation, have much more limited notions of plurality and prefer the framework to be grounded mainly in Anglo-Saxon or Anglo-Celtic core values (Bullivant, 1981). Other cultures are then expected to contribute only peripherally, chiefly in relation to food and the celebration of colorful customs and festivals. In the extreme residualized form, where minority ethnic core values are largely dissipated, this model virtually ceases to be multicultural (Smolicz, 1988, 1994). One par ticular point of contention in striking the multicultural balance has centered on languages. This was manifested by the persistent use by a former Australian Minister of Education, John Dawkins, of the term "foreign" rather than " community" for languages other than English which are spoken by some 17 percent of Australian citizens, many of whom regard their ancestral language as the core of their groups culture (Clyne, 1991; Smolicz, 1994). This ministerial stance was best exemplified in the name of the report issued by his office - The Language of Australia , finally renamed, under protest, Australia ' s Language (DEET, 1990). These ministerial documents contradicted national language policy recommendations accepted by the same government (Lo Bianco, 1987), as well as the experience of states, such as South Australia, where the acceptance of English as the principal language of communication has not precluded the teaching and examining of some 30 languages other than English at ^fear-12 level and their recognition as fully fledged subjects that count towards university entry in the same way as, say, English, Mathematics, Latin, or History. The state of South Australia has one of the best records in the country for teaching languages other than Endish (LOTE) at primary school level, with FROM "MIGRANT COUNTRY TO MULTICULTURAL NATION 175 every child receiving at least some instruction in LOTE by 1995. This development is part of a comprehensive plan to develop multicultural education mapped out by the report of the South Australian Task Force to Investigate Multiculturalism and Education (1984), entitled Education/or a Cultural Democracy, The far-reaching recommendations of the report related to language and cultural education at all levels, including the creation of university centers for intercultural studies and multicultural education and the establishment of the South Australian Secondary School of Languages (SASSL). When SASSL was established a decade ago it began with the teaching of two languages to some 100 pupils. It now teaches a dozen languages to over 1,000 students and has proved of indispensable value for the teaching of those languages for which it would be difficult to cater in the mainstream schools, as well as providing seeding funds for the introduction of new languages which can subsequently be incorporated in to the regular day school curriculum (Secombe, 1995). SASSL is open to students of all three education systems - state, Catholic, and other nongovernment — free of charge. In its operation of language classes after normal school hours in a number of selected locations, its very economical use of resources, the full accreditation of its subjects by school authorities, and the high level of achievement of its students in the final year-12 examinations, SASSL provides an example of a structure which is already proving of value at the international level, especially in countries like South Africa, where at least nine African indigenous languages must be taken into consideration (Alexander, 1989). While concern is being expressed at the grossly inadequate proportion of students studying languages at secondary school and university levels, it must be acknowledged that progress in acceptance of the multicultural orientation in schools, and lately at university level, has been enormous when compared with post-World War II days. At that time, ethnic part-time community schools received no state support and students were discouraged from studying there by their regular school teachers, while those speaking in languages other than English in public often received reprimands for not behaving in an "Australian " way (Clyne, 1982, 1991). The special feature of the Australian overarching framework is that it includes English as a shared language for all people but not to the exclusion of other languages that are spoken in the community. Despite the absence of any religious requirements in the shared value system, some reli g ious communities, such as fundamentalist Islamic, cause occasional disquiet among sections of the Australian population because of their political implications and the way they are perceived to impinge on other areas of culture — particularly the family - and hence on the overarching framework. However, in the absence of any significant rift, even during the Kuwait-Iraq war, the question of religion has seemed much less pressing and mainly of historic interest. Despite the growing 176 INTERNATIONAL MIGRATION REVIEW but still relatively small propor tion (approximately 5%) of Asian-Australians, the issue of " visible migrant minorities " has been of very much less significance than in other countries with heterogeneous populations. The perception of racial unease varies, however, from community to community. For example, our study at the University of Adelaide revealed a greater degree of discomfort on this subject among Indian than among Chinese origin students (Smolicz et ai ^ 1990). It is essential for the preservation of Australian multiculturalism and its developing nationhood that race and religion remain outside the overarching framework and hence do not count as criteria by which people are judged to be Australian. WHO IS AN AUSTRALIAN? It was in the hope of achieving this dynamic and flexible interpretation of the shared framework that during an address entitled, '"Who is an Australian?" to the parliament of Queensland in 1988, and later during a symposium on National Identity at the Academy of Social Sciences in Canberra, I posed the question of whether it was already possible to be an Australian in an English way; an Australian in an Aboriginal way; an Australian in a Jewish, Polish, German, Italian, Greek and many other ways - corresponding to the complex ethnic origins of modern Australians. (Smolicz, 1991a:51) The Governor-General of Australia, the Hon. Bill Hayden, provided an answer in the affirmative to this question in an "Australia Day Message " delivered on January 26, 1994. He explained how, during the ceremony for the Unknown Australian Soldier, he "became very much aware of a distinctly changed mood. " He perceived that People were both affirming and seeking reaffirmation from their fellows about a great and important truth. I believe they were restating their commitment to the core values of what it means to be an Australian. Not necessarily a traditional Australian, Anglo-Saxon Australian or Celtic Australian. Every day, for instance, we see people who are Australians in an Italian way, in a Jewish way, a Vietnamese way. " (Hayden, 1994:3) The Governor-General's words confirm an expansion of the framework of overarching cultural values within the multi-ethnic Australian state as it develops towards a multicultural nation. But how does this optimistic vision match the popular image ofAustralianness? An interesting illustration of the popular conception of what is taken for Australian is provided in a recent news item about one of the most isolated townships in South Australia - Coober Pedy — whose fame rests on its opal fields. In describing the mining population there, the writer emphasized its multi-ethnic character and marveled at the mix which he said included Croats, Greeks, Italians, Poles, Aboriginals - and even FROM "MIGRANT COUNTRY"TO MULTICULTURAL NATION 177 Australians! The obvious inference is that only Australians ofAnglo or Celtic origin are to be counted as Australians, while all those Australian citizens of minority ethnic background are viewed as " other " or " ethnics " ^., not "fair dinkum Aussies, " and therefore in some way are less than Australian. They may all have political rights as citizens, but because of their cultural and ancestral distinction, they are not identified by the majority as being part of the Australian nation. Somehow, in the popular mind, that majority identifies itself with the whole of the nation so that it is only possible to be Australian in an Anglo-Celtic way. There are, however, some signs of change, especially when people from Australia encounter each other overseas and then discover their common Australianness. For example, a well-known Australian journalist, reporting from Japan on changes in the Russian seaport ofVladivostok, devotes pride of place to an "Australian " who runs an "Australian restaurant, " was "born in China [and] migrated to Australia with his parents - one an Austrian, the other a Russian " (McGregor, 1994:50-51). Concerns about Australian identity surfaced recently at the Ideas for Australia Program, which was focused on the issue of citizenship. The view was expressed that "Australia has been a single State - and a highly successful one - since Federation. But it has not been in the past, nor is it now, a nation in the classical sense of that word" (Reynolds, 1994:11). This evaluation reinforces the need for highlighting the distinction between Nation and State, and a possible redefinition of the meaning of " nation " in the context of a multiethnic state. CONCEPTS OF NATION AND STATE Concern for clarification of the concept of nation has been voiced in the past. Znaniecki (1952, 1990), in particular, was anxious to free the concept of any biological overtones brought about by the myth of common ancestry, as well as territorial concerns which, in so far as they were political in nature, pertained to the state rather than the nation. As early as 1921, he anticipated the horrors of World War II and its " purity of the blood" syndrome in his book, The Fate of Western Civilization ^ when he wrote, A nation is a product of civilization, and not a natural entity, such as a hive of bees or an ant heap. Its basis is neither biological nor geographical; it is based neither on the racial cohesion of individuals who claim descent from common ancestors and the possession of common biological traits, nor on external bonds imposed upon the inhabitants of a particular territory. (Znaniecki, 1921) From this perspective, state is an entirely different entity from that of the nation, in that it refers to the political and legal dimensions of managing the resources of a country. This includes issues pertaining to citizenshi p and political governance, which in a democracy signifies the right to vote as well 178 INTERNATIONAL MIGRATION REVIEW as the obligation to pay taxes. Citizenship may also be construed to include the ri g ht to work and to receive social insurance, unemployment benefits, and superannuation (Bauman, 1990). Nation, on the other hand, maybe observed as a collectivity actively identified with a range of cultural values that are perceived as reflective of its past and an influence on both its present existence and future development. While a state is grounded in a particular territory, the nation can extend over many parts of the g lobe and assert its cultural identity across continents. At times, however, native land becomes more than a territorial unit and assumes a cultural symbolic meaning pertaining to a nation, as can be seen in the instance of Israel or Ireland. The link between the nation and the state is not indispensable for a nations survival. A nation may be torn away from the state, placed under foreign rule or par titioned, yet persist for decades or even centuries until it can eventually reformulate its political organization. This situation can be illustrated by the fluctuations in size and degree of independence to be found in both the Polish and Hungarian states, and the resilience shown by the corresponding nations (Davies, 1983; Zamoyski, 1987). Another distinguishing characteristic of a nation that was pinpointed by Znaniecki (1990), is that it possesses a social organization which differs from, and cannot be reduced to, political structures. Such social organization is based on cultural, spiritual and moral leadership and solidarity, rather than political power. The example of the Armenians and the Jews, who for centuries preserved their identity and social organization without the underpinning of a state, gives support to this view. Ultimately, both these peoples became eager to seek what they saw as the reestablishment of their ancient states, although many of their members remained as ethnic minorities in other countries and were not all necessarily enthusiastic about procedures adopted to create a modern statehood (Paine, 1989). The trend to nation-state building is generally viewed as the consequence of the modernization that occurred in Western Europe, stimulated by mass public education and industrialization. In fact, such developments were built upon the age-long significance of cultures in shaping distinct communities, each with the perception of its own particular uniqueness and role in history (Smith, 1986). Modern nations arose through the imaginative reconstruction of this heritage (Szacki, 1969; Smolicz, 1974,1988), with some of them giving rise to states to support their cultural and social needs. Often this occurred without regard for other cultural groups in their midst, which remained as " stateless nations " (Williams, 1991). Few states can claim with justification the tide of nation-state in which there is a near perfect fit between a single nation and the state. According to Connor (1993:374), this occurs when " the borders of an ethnonational group closely coincide with the borders of the state.)) The number of such nation-states is FROM "MIGRANT COUNTRY"TO MULTICULTURAL NATION 179 very small; it is estimated that not more than a dozen out of some 180 contemporary states qualify — including Japan, Iceland, Korea, and Portugal. But even among the multi-ethnic countries, which clearly are the main type of state in the world today, the politically and culturally dominant group (usually labeled from the German as Staatvolk) tends to perceive the state as an extension of its own nation. In the words of Connor (1993:375), " there is a tendency on the part of such people to equate the entire country with their own homeland and, as in the case of a people in an ethnically homogeneous state, to perceive the state as a political expression of their particular ethnic group. " He illustrates this contention by reference to the way English people often use the term England to describe the whole country (The United Kingdom of Great Britain and Northern Ireland) without regard to the sensitivities of the Scots and Welsh, to say nothing of the Irish. The same point has been made by Grant (1994) in drawing attention to the varying perceptions of the " sociological map " of Britain from English and Scottish perspectives, and the way that terms such as England or Britain are often used at random by the English without regard for the Celtic peoples of the British Isles. While Connor and Znaniecki are in agreement on the need to draw a sharp distinction between the concepts of nation and state, their perceptions of the essential nature of the nation differs. Connor (1993) insists on seeing nation as basically a kinship group, even though 'most nations do in fact contain several genetic strands. " In his view, " authorities, " and Znaniecki would fit this description, "ignore the dictum that it is not what is but what people perceive as is which influences attitude and behavior. " Znaniecki (1990) regards belief in " archaic bloodlines " as erroneous and detrimental both to our perception of social reality and our understanding of its consequences. It is his firm view that their acceptance as part of every group s " national psychology " helps to perpetuate racial myths — or even evoke them where they do not exist — in a way that holds fearsome prospects for the modern world. A good example of the cultural basis of the contemporary Greek nation is provided by Hill (1993:45) when he argues that it is grounded on " a cultural ideology called ellinismos or' Greek culture. ' Ellinismos means allegiance to the ancient Greek and Byzantine heritage and the Greek Orthodox faith. " It was on the basis of ellinismos that "in the eighteenth, nineteenth and twentieth centuries, Greek nationalists could claim most of the European territories of the Ottoman Empire and parts ofAnatolia as being Greek. " With its demographic continuity disrupted on numerous occasions by massive influxes of people from different backgrounds, the culturalist perspective of the modern Greek nation is further supported by the way those Orthodox who, in the nineteenth century, remained loyal to the patriarch of Constantinople also adopted Greek language and culture and regarded themselves as Greek. This attachment to ellinismos was carried by Greek immigrants into the diaspora, 180 INTERNATIONAL MIGRATION REVIEW including Australia, where they have shown one of the strongest tenacities among all ethnic minorities in the maintenance of their cultural heritage, par ticularly language (Clyne, 1991; Smolicz, 1985). It must be acknowledged that in postulating his kinship concept of nationhood, Connor (1993:324) specifically stated that his comments did not refer to "diasporas, nor to immigrant societies, such as those within Australia, the United States and non-Quebec Canada. " Znaniecki (1990), on the other hand, while attempting to " purify " the concept of nation from all nations of common biological descent and to clarify the distinction between the cultural and political domains, did not sufficiently develop his ideas to include nations that are in the process of crystallization from more than one migrant cultural group. His concept of a distinctive culture and its corresponding social organization that a nation needs for its survival is still applicable in multicultural settings, but must be elaborated to ensure that cultural diversity can persist within the shared cultural framework of national consensus. A MULTICULTURAL NATION? The distinctive concepts of nation and state are intimately connected with the notion of national identity, which is of increasing impor tance in Australia as it approaches the celebration for the centenary of Federation in 2001. In this connection, writers such as Donald Home (1994) have rejected the concept of an Australian national identity and have argued for the transfer of attention to what they have labeled as a " new civic identity " as a much more satisfactory notion for Australian society at the threshold of the twenty-first century. As defined by the Ideas for Australia Program, this would involve " commitment to the rule of law, to the Constitution as amended by the Australian people, to the principles of parliamentary democracy and to the maintenance of Australia as an actively democratic community " (Reynolds, 1994). It is doubtful, however, whether the basically legal concept of civic identity can ever be completely disentangled from its cultural context. "Active democratic community" implies more than a mechanistic configuration of individuals held together by a democratic constitution, but rather a group of people that espouses a democratic culture. The cultural element stands out even more clearly when " civic identity " is defined as including a commitment to " equality under the law of all Australians irrespective of race, sex, ethnicity and creed. " This involves much more than political and legal concepts, but represents a powerful affirmation of the cultural values of equity that Australians (whatever their ethnic background) are expected to share. Hence what Home and other authors in the Ideas for Australia Program are really advocating, without naming it as such, is what has previously been described as an overarching framework of shared values that would embrace a number of ethnic cultural FROM "MIGRANT COUNTRY" TO MULTICULTURAL NATION 181 variants. Members of the contributing cultures would be expected to fit in with, and not violate, the overarching values. At the same time, however, the framework is assumed not to be a rigid structure like a Roman bridge' set in stone, " but to be flexible enough to respond to changes, both in patterns of cultural diversity and in the needs of the nation as a whole. The term " nation " refers here to a multicultural nation that shares a whole gamut of cultural values, is not monoethnic or monocultural, and does not harbor beliefs about descent from a common ancestry - either biological or cultural. The very existence of Aborig inal people makes any notion that Australians are just of Anglo-Saxon, ofAnglo-Celtic or even just of European descent, an absurd concept (Bourke, Bourke and Edwards, 1994), as does the growing number of Asian-Australians (Price, 1993). The same holds for cultural values. The Australian nation cannot be squeezed back into the Anglo-Celtic box, which it outgrew long ago and which, from an Aboriginal perspective, was a misfit from the start. Since this is the reality of the Australian situation, why the desire to limit the country ' s aspiration to" civic " identity and the fear of the concept of'national identity " with its clear cultural basis? The answer seems to be that the " nation " is still being perceived in monoethnic and monocultural terms — neither of which obviously fits contemporary Australia. THE PERILS OF A MONOCULTURAL NATION The danger of such a misconception is nevertheless not wholly imaginary. There is even a risk, however remote, that the espousal of a republic could rouse up the spirit of nationalism of the " new-born " nation variety. Minority ethnic Australians would then be increasingly required to conform to the whole spectrum of Anglo-Celtic values by ceasing to activate their particular languages and cultures. Those who were unwilling or unable to conform would remain outside the " mainstream, " conceived in almost exclusively Anglo-Celtic terms. This " mainstream" would then be equated with the "Australian nation; " those left out, even if citizens, would be excluded much more formallly from being Australian than happens at present, as in the case of Coober Pedy people ofnon-Anglo-Celtic background. The Anglo-assimilationist policies of republican America in the first half of the twentieth century provide a preview of the scenario that could develop in a nationalistic Australian republic dedicated to a homogenized single-culture policy. These misgivings might seem unnecessarily alarmist and premature, but the recent insistence of Australian federal authorities on an almost exclusive focus upon teaching foreign and trade languages from Asia and the downgrading of minority community languages arouse disquiet about the future of multiculturalism in this country. This lack of balance is best exemplified by the Council of Australian Governments in singling out Japanese, Chinese, Korean, and 182 INTERNATIONAL MIGRATION REVIEW Indonesian as the principal languages other than English to be promoted in schools. In contrast, Vietnamese, along with Italian, Greek, and other European languages spoken widely within the Australian community, was deliberately disregarded. This approach left the distinct impression that at stake was not Vietnam ' s still relative insignificance in trade during the 1980s and early 1990s, but the fact that the language was spoken by a large and growing number of Vietnamese-Australians. Reservations about the emphasis on Asian trade languages were recently expressed in a report to the Bureau of Immigration and Population Research compiled by Iredale and Fox. The report challenged the priority placed on the learning of the four " super " languages " as likely to increase racial tension and possibly cause a backlash against multiculturalism " (Jones, 1994:4). The authors expressed concern at the ' setting up of some groups to be more important than others " in a way that meant that "difference was not being valued. " Even if such measures could be explained in terms of trade, they ignored the importance of having a languages policy to fit the country ' s immigration profile, by undervaluing the reality of Australia n s unique cultural diversity. While acknowledging the need for all Australians to master English as part of the country s overarching values, any return to the early postwar policy of insularity towards the non-British world outside and assimilation inside the country would involve constricting a multiethnic state into the strait-jacket of a monocultural nation. This time, however, it would no longer be under the old British imperial or Commonwealth cloak, as happened during the two World Wars (Selleck, 1980), but in the name of young asser tive Australian nationalism. Such moves would encounter the opposition of minorities which are much more empowered both in knowledge and resources than were the Aboriginal peoples, or the Displaced Persons who arrived in Australia during 1949-1951 (Mar tin, 1965), or the economic migrants of the 1950s and 1960s from the then " poor " countries of Southern Europe (Price, 1963). Even those apparently helpless minorities from the past did not melt but retained aspects of their otherness (Jupp, 1989). Past experience shows how the expectation of uniformity can lead to conflict when monocultural dominant groups are confronted with the inevitable otherness of individuals from different cultural backgrounds. Fear of otherness disappears when the acceptance of cultural diversity becomes par t of the overarching framework of shared values in a multicultural nation. Our perception of otherness is the opposite side of the coin of the way we perceive ourselves and the extent to which our own identity, and that of our group, is exclusive or inclusive. Much depends on how a group views the manifestations of its own distinctiveness and the way this is perceived in terms of cultural markers or core values (Smolicz, 1981,1988,1994; Conversi, 1990, 1995; Grant, 1992, 1994; Smolicz and Secombe, 1989). These markers may be regarded as the building blocks of the group ' s cultural identity and may be FROM MIGRANT COUNTRY TO MULTICULTURAL NATION 183 based upon an ethnospedfic language but can also be linked to such factors as reli gion, family structure, or perceived descent. In spite of the theorists of cultural convergence, it can be taken that some form of national or ethnic identity is a persistent and universal phenomenon (Smith, 1986,1994). It may find negative expression in the form of ethnocentric nationalism that excludes and denigrates others - because of their cultural, racial or historical characteristics — or as a positive sign of patriotism, which is concerned to enjoy and protect the virtues of ones own culture, while remaining open to sharing it with others in a process of mutual interaction and exchange. While monocentric nationalism still plagues Europe through the inability to see beyond the horizon of sectional interests and avoidance of any responsibility for itself as a whole, Australia ' s recognition of the creative force of cultural interaction within a framework of shared beliefs has enabled it to develop beyond a " migrant country " to become a laboratory for a multicultural nation. REFERENCES Alexander, N. 1989 Language Policy and National Unity in South Africa/Azan ia, Cape Town: Buchu Books. Bauman, Z. 1990 Thinking Sociologically. Oxford: Blackwell. Pp. 161-177. Bayne, R, B. Dexter, K. Maddock, C. Jennett and C. Perkins 1988 "Politics, the Law and Aborigines." In The Australian People: An Encyclopedia of the Nation ^ Its People and Their Origins. Ed. J. Jupp. Sydney: Angus and Robertson. Pp. 212-239. Bourke, C., E. Bourke and B. Edwards, eds. 1994 Aboriginal Australia, St. Lucia: University of Queensland Press. Bullivant, B. M. 1981 The Pluralist Dilemma in Education. Sydney: Alien and Unwin. Clyne, M. 1991 Community Languages: The Australian Experience. Cambridge: Cambridge University Press. 1982 Multilingual Australia, Melbourne: River Seine. Connor, W 1993 "Beyond Reason, the Nature of the Ethnonational Bond," Ethnic and Racial Studies, 16(3):373-389. Conversi, D. 1995 "Reassessing Current Theories of Nationalism: Nationalism as Boundary Maintenance and Creation," Nationalism and Ethnic Politics, 1(1):73-85. 1990 "Language or Race? The Choice of Core Values in the Development of Catalan and Basque Nationalism," Ethnic and Racial Studies, 13(1):50-70. Davies, N. 1983 God ' s Playground: A History of Poland Oxford: Oxford University Press. 184 INTERNATIONAL MIGRATION REVIEW Department of Employment, Education and Training (DEET) 1990 The Language of Australia: Discussion Paper on an Australian Literacy and Language Policy for the 1990s. Canberra: Australian Government Publishing Service (AGPS). Fraser, M. 1981 "Multiculturalism." Inaugural address delivered to the Australian Institute of Multicultural Affairs, Melbourne. Grant, N. 1994 "Education and Cultural Identities. " Paper delivered at the 16th conference of the Comparative Education Society in Europe (CESE), Copenhagen. June. 1992 "Language, Culture and Curriculum among European Minority and Peripheral People." Paper delivered at the 15th Conference of the Comparative Education Society in Europe (CESE), Dijon, France. Hayden, B., The Hon. 1994 "Australia Day Message," Polyphony, 9:3. Canberra. Hill, P. 1993 "National Minorities in Emo^e , " Journal of Intercultural Studies, l4(l):33-98. Horio, T. and S. Platzer 1990 Educational Thought and Ideology in Modem Japan: State Authority and Intellectual Freedom. Tokyo: University of Tokyo Press. Home, D. 1994 Public Culture: An Argument with the Future. Sydney: Pluto Press. Jones, C. 1994 "Language Learning Policy Challenged," The Australian , 9369:4. September 30. J"PP> J-1989 The Challenge of Diversity: Policy Options for a Multicultural Australia, Canberra: AGPS. Jupp, J. and M. Kabala, eds. 1993 The Politics of Australian Immigration, Canberra: AGPS. Kiernan, C. 1988 "Irish in Australian Politics." In The Australian People: An Encyclopedia of the Nation ^ Its People andTheir Origins. Ed. J. Jupp. Sydney: Angus and Robertston. Pp. 586-589. Lo Bianco, J. 1987 National Policy on Languages. Commonwealth Department of Education report. Canberra: AGPS. Martin, J. I. 1965 Refugee Settlers. Canberra: The National Australian University Press. McGregor, R. 1994 "Eric and the Goodfellas of Vladivostok," The Australian Magazine, August 20-21, 50-54. Miihihausler, P. 1992 "Preserving Languages or Language Ecologies?" Oceanic Linguistics, 31 (2): 163-186. Paine, R. 1989 "Israel: Jewish Identity and Competition over Tradition." In History and Ethnicity. Ed. M. MacDonald Tonkin and M. Chapman. London and New York: Routledge. Pp. 121-136. Price, C. 1993 "Australia as an Intermediary with Asia," Journal of Intercultural Studies, l4(l):19-32. 1963 Southern Europeans in Australia, Melbourne: Melbourne University Press. FROM MIGRANT COUNTRY TO MULTICULTURAL NATION 185 Reynolds, H. 1994 "Australia Viewed in Optimistic Light," The Australian , February 21, 11. Roberts, D. 1994 "Self-determination and the Struggle for Aboriginal Equality. " In Aboriginal Australia. Ed. C. Bourke, E. Bourke and B. Edwards. St. Lucia: University of Queensland Press. Secombe, M. J. 1995 Cultural Interaction among Some Anglo-Australian Tertiary Graduates. Adelaide: The University of Adelaide. Unpublished monograph. Selleck. R.J.W 1980 "The Trouble with My Looking Glass: A Study of the Attitude of Australians to Germans during the Great War,' Journal of Australian Studies, 6:1-25. Smith, A. D. 1994 "The Problem of National Identity: Ancient, Medieval and Modem?" Ethnic and Racial Studies, 17(3):375-400. 1986 The Ethnic Origin of Nations. Louden: Basil Blackwell. Smolicz, J. J. 1994 "Australia n s Language Policies and Minority Rights: A Core Value Perspective." In Linguistic Human Rights: Overcoming Linguistic Discrimination. Ed. T. Skutnabb-Kangas, R. Phillipson and M. Rannut. Berlin and New York: Mouton de Gruyter. Pp. 235-252. 1992 Australian Diversity. Adelaide: Centre for Intercultural Studies and Multicultural Education (CISME). 199 la "Who is an Australian? Identity, Core Values and the Resilience of Culture." In Australian National Identity. Ed. C. Price. Canberra: Academy of the Social Sciences in Australia. Pp. 211-232. 1991 b "Language Core Values in a Multicultural Setting," International Review of Education, 37(1):35-52. 1990 "The Monoethnic Tradition and the Education of Minority Youth in Germany from an Australian Multicultural Perspective," Comparative Education, 26(1):27-44. 1988 "Tradition Core Values and Intercultural Development," Ethnic and Racial Studies, 11(4):382-410. 1985 "Greek Australians: A Question of Survival in a Multicultural Australia," Journal of Multilingual and Multicultural Development, 6( 1):7-29. 1981 "Core Values and Cultural Identity," Ethnic and Racial Studies, 4(1):75-90. 1974 "The Concept of Tradition: A Humanistic Interpretation," The Australian and New Zealand Journal of Sociology, 10(2) :75-83. Smolicz, J. J., L. Lee, M. Murugaian and M. J. Secombe 1990 "Language as a Core Value of Culture among Tertiary Students of Chinese and Indian Origin in Australia., " Journal of Asian Pacific Communications, 1(1):229-246. 186 INTERNATIONAL MIGRATION REVIEW Smolicz, J. J. and M. J. Secombe 1989 "Types of Language Activation and Evaluation in an Ethnically Plural Society." In Status and Function of Languages and Language Varieties. Ed. U. Ammon. Berlin and New York: Mouton de Gruyter. Pp. 478-514. South Australian Task Force to Investigate Multiculturalism and Education 1984 Education/or a Cultural Democracy. Adelaide: South Australian Government Printers. Stretton, E and C. Fennimore 1988 "Votes for Aborigines," The Adelaide Review, 46:10-11. Szacki, J. 1969 "Three Concepts of Tradition," The Polish Sociological Bulletin, 2:29. Watts, R.J. 1991 Power in Family Discourse. Berlin and New York: Mouton de Gruyter. Williams, C. H., ed. 1991 Linguistic Minorities, Society and Territory. Clevedon, England: Multilingual Matters. Wooden, M., R. Holton, G. Hugo andJ. Sloan 1994 Australian Immigration. Canberra: AGPS. Zamoyski, A. 1987 The Polish Way. London: John Murray Znaniecki, F. 1990 Wspolczesne Narody. Warsaw: PWN. 1952 Modem Nationalities. A Sociological Study of How Nationalities Evolve. Urbana: University of Illinois Press. 1921 Upadek Cywilizacji Zachodniej(The Fall of Western Civilization). Poznan, quoted by J. Szacki in F. Znaniecki, 1990. Wspolczesne Narody Warsaw: PWN, p. XIV. Book Reviews DANIEL C.TURACK Capital University Law School Laws Harsh As Tigers. By Lucy E. Salyer. Chapel Hill and London: University of North Carolina Press. 1995. Pp. xix, 338. ISBN 0-8078-2218-3 (c); 0-8078-4530-2 (p) Mass Expulsion in Modem International Law and Practice. ByJean-Marie Henckaerts. The Hague, Boston, London: Martinus Nijhoff Publishers. 1995. Pp. xviii, 257. ISBN 90-411-0072-5. $105. Seeking Asylum: Comparative Law and Practice in Selected European Countries. By Helene Lambert. Dordrecht, Boston, London: Martinus Nijhoff Publishers. 1995. Pp. xvi, 220. ISBN 0-7923-3152-4. These three books consider the individual^ desire to enter a country other than the country of their origin or nationality, either for a temporary purpose or to begin a "new life." While the first title is basically a historical study, the other two books are current technical treatises. In the case of the first title, the subtitle more readily identifies its contents - Chinese Immigrants and the Shaping of Modem Immigration Law in the United States. I think that readers might want to read the short appendix (p. 253) after the introduction to understand the'methodology used by Professor Salyer in her study. She explores the economic, social, and legally restrictive policies and their enforcement in the United States between 1891 and 1924 that impacted the prospective Chinese immigrant. One encounters the initial successful litigation of challenges to Chinese exclusion specifically concentrated in the Federal District Court for the Northern District of California. Salyer traces the growth of administrative power through case law, and is adroit in showing how the jurisdiction of the federal courts to hear these cases involving Chinese immigration becomes sharply curtailed and eventually replaced by an immigration administrative agency. ©1997 by the Center for Migration Studies of New York. All rights reserved 0197-9183/97/3101.0117 After 1905, immigrants challenged the Bureau of Immigration^ summary administrative procedures as a denial of due process. Professor Salyer's delves into the substantive features of the "new " immigration legislation through a principal focus on the institutional structure that became operational. Under the heading of "Judicial Justice," one finds the Chinese use of the courts to combat administrative decisions to exclude or deport them. There is an excellent analysis of the procedures and approaches utilized. The reader is led through the campaign and techniques adopted to remove jurisdiction from the courts in 1905. In the second part of her study. Professor Salyer deals with "Executive Justice 1905-1924." Once the U.S. Supreme Court endorsed restrictive justicial review, and accepted the expanded role of immigration official s decision as the final authority on the issue of whether the Chinese could legally enter or stay in the United States, new strategies were necessary for the potential immigrant. Between 1905 and 1924, the author traces immigration policy impact not only on the Chinese but on other Asians and Europeans. What began as a summary administrative procedure against the Chinese is aptly shown to have become the norm in general immigration administration. The author discusses the attempts to reverse this administrative power, but it is clearly evident that the rights of aliens occupied a secondary position to the objectives behind the restrictive U.S. immigration laws and practice. Dr. Henckaerts concentrates on the mass expulsions that have occurred in the post-World War II era. To assist the reader and better understand the intricacies of mass expulsion, I suggest turning to the summary of the 48 findings set out by the author (pp. 199-203) as conclusions. The substantive scope of this study is limited to transfrontier expulsions rather than internal displacements or forced relocations. Interestingly,"mass" is not defined by size. Rather, in international RF.VTPWFSSAY ROOK RF.VTFW.S 188 law, a qualitative, not quantitative, standard is used, and is more adequately reflected by the adjective "collective. " The parameters of this study are clearly stated, and the legal concept of expulsion is distinguished from other concepts such as exclusion, deportation, and extradition. As there is no specific prohibition against mass expulsion in customary international law, Henckaerts accounts for the subjects omission from the Universal Declaration of Human Rights (1948) and the first regional instrument on human rights, that is, the European Convention on Human Rights (1950). This situation was rectified for the latter in 1963 and was subsequently adopted in other universal and regional documents: the American Convention on Human Rights (1969); the African Charter on Human and Peoples Rights (1981). Professor Henckaerts links two concepts in analysis: an expulsion of a group of people on the basis that the expellees belong to that specific group; and the absence of an individual review for each person expelled. The former represents a form of "discrimination" while the latter reflects " arbitrariness." The author asserts that both practices are illegal, and the reader is made aware of additional conditions which must be met to avoid the brand of illegality. Mass expulsion of migrant workers differs from the general regime of mass expulsion. The author analyzes this type of state conduct based on situations that occurred in Nigeria in 1983 and 1985, and in the Dominican Republic in 1991. Sometimes nationals, resident aliens, refugees, stateless persons, or the denationalized are expelled, and the author ably describes these occurrences. His recommendations cover the presence of Jewish settlers on the West Bank and the Gaza Strip, a situation still awaiting resolution. Readers can become familiar with various refugee group expulsions that have taken place in recent years in Africa, Asia, and Europe. Root causes of expulsions arc identified that include national economic policies, population exchanges, and voluntary repatriation programs. W^ir gives rise to expulsion of enemy aliens, and the broader subject of deportation of civilians by an occupying force during belligerent occupation arc subjects skillfully explored. Dr. Henckaerts underscores the recent developments that took place during Iraqs invasion of Kuwait, the forced movement of peoples in the former Yugoslavia, and Iraqs treatment of its Kurdish population. One finds a comprehensive commentary on pragmatic measures of emergency relief, available remedies, enforcement mechanisms and most importantly, methods of prevention. The reader finds research that mirrors new developments and trends in the practice of mass expulsions. Nine relevant appendices, not easily accessible, such as the 1991 Dominican mass expulsion of Haitians and Dominico-Haitians, enhance the value of this treatise. Two caveats to my overall favorable view of this book. First, the author should have provided more of a historical base of the problem of mass expulsion, as for example the practices following World War I. Second, the author s injection of his political opinions at times might appear to affect his analysis (p. 177, involving Israeli settlements). Of the three books under review, Dr. Lambert has written the one which has a current and ongoing human and political concern for refugees and other asylum seekers. There is in place a body of refugee law that comprises an amalgam of national laws and procedures devised to implement regional and universal treaties. However, it is national law and practice - particularly with respect to immigration - that is the key to the individual s right to asylum. Dr. Lambert's book is a practical study of the current immigration laws in Belgium, France, Germany, Sweden, Switzerland, and the United Kingdom. Her analysis appraises the level of protection offered in each of the six states as these immigrants seek to qualify for asylum. Coverage is arranged on the basis of six separate issues. She initially looks at the admission process that involves first instance proceedings then sets out the right of appeal for those whose application was rejected. She concentrates on the issue of burden of proof 189 INTERNATIONAL MIGRATION REVIEW and the rules of evidence that are employed in the process. Throughout her in-depth analysis, she measures the states modus operandi with the states international obligations under the United Nations Convention relating to the State of Refugees (1951), and the 1967 Protocol (which removed the time and geographical limitations in the 1951 Convention). Dr. Lambert addresses the concept and implementation of protection against refoulement as well as the living conditions of these asylum seekers during the pre-asylum period. Here one gains an insight into the degree of the individuals freedom of movement and social/economic rights, e.g., right of family reunion, accommodations, education, and social assistance. This study also considers asylum seekers who are designated as defacto refugees or refugees for humanitarian reasons, that is, persons who are not de jure refugees according to international instruments. In completing her comparative review of national practices concerning this type of refugee, Dr. Lambert advocates an agreement on the harmonization of humanitarian status within Europe. Finally, the author considers the minimum standards set out in the 1951 Conventions, and how far each state under review has gone to meet minimal observance or exceed the standard. This work is not an overview. Of the three books under review, Dr. Lambert is a path blazer in terms of focus and range. Her work is creative. All three works contribute to a greater understanding than existed before their appearance. Each of the three authors have contributed a bibliography to assist those who would delve further into the topic. The Face of the Nation: Immigration) the State and the National Identity. By Keith Fitzgerald. Stanfotd, CA: Stanford University Press, 1996. Pp. 285. KEVIN R. JOHNSON University of California at Davis In light of the current political debate in the United States over immigration reform, The Face of the Nation is indeed timely. The book contends that pluralism, class conflict, and realism theories fail to offer a complete understanding of the development of three important aspects of immigration policy — "front-gate immigration," refugee policy and "back-door immigration" (with a near-exclusive focus on undocumented migration from Mexico). The thesis is that each theory fails to acknowledge the importance of political institutions in shaping policy initiatives in each of these areas. To remedy this theoretical failing, Fitzgerald articulates a theory of "improvisational institutionalism" that recognizes the states influence on policy development. Fitzgerald thoroughly documents the slow evolution of a national immigration bureaucracy, which in turn generated political forces that defy explanation by simple economic determinism. The state began articulating its own interests in debates over immigration policy and greatly shaped the ultimate policy judgments. Moreover, the burgeoning national immigration bureaucracy transformed the politics of immigration. Namely, national interest groups emerged in the nationalization of immigration policy. Fitzgerald offers many examples of how the state has shaped immigration policy. Consider a few. Despite the dire economic circumstances of the Great Depression, Congress somewhat surprisingly did not respond by passing restrictive immigration legislation. Instead, the administrative apparatus of the state decreased the number of people immigrating through heightened enforcement, thereby defusing restrictionist demands and, at the same time, promoted foreign policy interests by denying visas to Jews fleeing Germany. Similarly, the McCarren-Walter Act of 1952 reflects the states conclusion that immigration policy implicated national security concerns that were part and parcel of cold war foreign policy. Finally, even in acting in a benevolent way, Congress was influenced by the demands of the state that foreign policy interests required elimination of the national origin quotas system in 1965. One wonders why the role of the state is, as Fitzgerald suggests, greater in the formulation of immigration policy than in other policy areas. One unexplored possibility is that the persons most directly affected by immigration policy - BOOK REVIEWS 190 noncitizens - cannot vote and are represented through groups with limited political power The book also studies how the evolution of immigration policy reflects the ongoing debate about immigration's impact on the " national identity." Importantly, both racial exclusions in the immigration laws and the discriminatory national origin quotas system were designed to promote an Anglo-Saxon national identity. The Face of the Nation is a nuanced analysis of immigration law and politics in the United States. It demonstrates how the development of the national immigration bureaucracy affected U.S. immigration policy. However, the book' s analysis of immigration law and its implementation would have gained much by considering the wealth of legal commentary in the area. Legal scholars have, for example, criticized the discretionary power of the State Department to issue visas, which the book repeatedly discusses. Similarly, legal scholars offer valuable insights about the influence of foreign policy on refugee admissions, which also are important to Fitzgerald's analysis. Fresh BloocL The New American Immigrants. By Sanford J. Ungar. New Yorki Simon and Schuster, 1995. Pp. 399. $25. PETER I. ROSE Smith College Sanford J. Ungar, current Dean of the School of Communication at American University in Washington, has spent years covering political and social changes and writing about them. He has reported on the uprisings in France in 1968, the release of the Pentagon Papers, and the paradoxes of independence movements in Africa. Several years ago, he decided to investigate the new American immigration and its impact on our economy, culture, and polity. Fresh Blood is the result. Throughout its pages, Ungar, a skilled interviewer, keen observer, and consummate storyteller, focuses on the newcomers themselves. He takes his readers on an odyssey, crisscrossing the length and breadth and borderlands of the United States, visiting small towns and big city neighborhoods. Along the way we meet recent refugees and immigrants - legal and "undocumented" — from Cuba, Ethiopia, Haiti, Iran, Ireland, Korea, and Mexico, from various countries of Eastern Europe and Indochina, and from many other places. We not only leam who they are, but how they are faring - and how they are feeling. In a number of cases, the richly personalized accounts are examined in broader social contexts. We "hear" why people left, how they made it to this country, how they were received, their varied modes of adaptation and acculturation. This book is not an academic treatise, yet it is clearly a contribution to the sociology and politics of immigration. In case after case, Ungar and his subjects address issues of survival, the determination to overcome economic and social challenges, and the sorts of activities used to assist kin and countrymen in doing so. In this, without using the terms as such, he makes quite clear the critical differences between " primary " and "secondary" relationships, those with family and close friends and those with others with whom they must interact at school, on the job, in the marketplace. Throughout his collection of commentaries about people and places, Ungar constantly shows the interplay of general concerns, of, say, Congress or the INS or " public opinion," and of those concerns that are more situational or regionally specific or have to do with a particular racial and ethnic mix, or with economic opportunities, or with the resentment of those who feel threatened by the competitiveness of newcomers who still believe in the promise of America and often seem to find ways of turning their hopes into self-fulfilling prophesies. Ungar addresses other controversies, such as the magnet effect of current family provisions in immigration laws, the challenge of illegals who are seen by many as undercutting the labor market, and the problems of those dependent on public assistance. He cites polls indicating that there is a general tendency to want to limit immigration and discusses the resurgence ofnativism - and scapegoating - that has marked recent political campaigns. Yet, most of the book is devoted to what is good about the post-1965 immigration: for the newcomers, for the society at large. Time 191INTERNATIONAL MIGRATION REVIEW and again, Ungar gives examples of the energy new immigrants bring to America and how much they contribute - economically and culturally — to a still-evolving nation of nations. "Like those who came before them," he writes, " todays immigrants are occasionally controversial, but they are transforming- and constantly improving - America." Fresh Blood is a potent antidote to Peter Brimelow's Alien Nation. American Dreaming: Immigrant Life on the Margins. By Sarah J. Mahler. Princeton, NJ: Princeton University Press, 1995. Pp. 268. $15.95. ELIZABETH J. MUELLER New School for Social Research American Dreaming is an innovative analysis of the journey and settlement of undocumented migrants who leave Central and South America for a new life in suburban Long Island, New "Vbrk. Relying often on the voices of the immigrants themselves, it oners a fresh, critical perspective on the relations between immigrants, challenging received notions of ethnic solidarity. Instead, Mahler paints a portrait of a community fraught with internal tensions and with ample opportunities for migrants to exploit each other economically, and identifies the roots of these relations in the migration process itself - the false expectations migrants hold about their journey, likely earnings and living expenses in the United States, the resources spent during the journey itself, and migrants' subsequent impoverished (and/or indebted) state upon arrival. To complete the story, Mahler documents the various economic activities immigrants have developed to serve the needs of other, more recent undocumented immigrants while generating the income they need to cover their own living expenses and obligations back home. An anthropologist, Mahler initially set out through a range of qualitative research methods, to understand the competition and jealousy she noted among immigrants. After first documenting the size and distinctive character of Salvadoran migration to Long Island (and challenging much lower census estimates with her results) and surveying 350 ESL students to test her initial hypotheses and identify additional concerns, Mahler focused her more intensive interviewing and ethnographic work on two groups of recent, undocumented migrants in one Long Island community: Salvadorans of mostly rural peasant origins and South Americans from several countries, of mostly middle- and working-class origins. The contrasts between the two groups prove useful. Mahler adds to our understanding of settlement by focusing on a suburban settlement and by linking relations among migrants in communities of recent, undocumented arrivals to the migration process rather than only the conditions immigrants find at the end of it. Building on previous research on transnational communities, she develops the concept of "immigrant capital" to capture the process through which immigrants leam how to increase their income and decrease their expenses - often necessitated by the debt incurred during their journey as well as the need to meet higher than anticipated living costs and obligations back home. She argues that while these activities provide this group with their greatest chances for economic mobility at this point, they also undermine the bases for ethnic solidarity and mutual aid within the group. American Dreaming advances our understanding of intraethnic relations in the communities studied. It will be useful to those interested in understanding the barriers as well as the paths to economic mobility among undocumented immigrants and the potential to alter these paths. Most importantly, American Dreaming challenges us to rethink the prevalence of "ethnic solidarity." And, questions are raised for additional research: Would similar activities be carried out in urban areas and lead to the same undermining of ethnic solidarity? How will these relationships change over time as migrants become more "settled?" Waiting on Washington. By Terry A. Repak. Philadelphia, PA: Temple University Press, 1995. AUDREY SINGER Georgetown University Repak's ethnography, Waiting on Washington ^ joins the ranks of a growing literature that brings gender to the forefront of immigrant BOOK REVIEWS 192 adaptation and settlement studies (see also Hondagneu-Sotelo's GenderedTransitionsand Hagans Deciding to Be Legal) . This is an engaging and multifaceted analysis of labor migration from Central America to Washington, D.C., Contemporary Central American migration to the nation's capital was distinguished initially by an idiosyncratic form of informal labor recruitment. During the 1960s and 1970s, immigrants, predominantly women, were recruited for jobs as housekeepers and childcare workers by international diplomats and U.S. government employees. The 1980s brought many more migrants (both men and women) to Washington, D.C., and other U.S. locations when political and economic problems worsened as war pummeled the countries of Central America. These factors contribute to the fairly stable and permanent settlement of Central AJnericans in the Washington, D.C., area. This book constructively examines the origins of this migration, describes the conditions in both sending and receiving areas, analyzes the role of gender in labor market trends and the mobility of female and male migrants, and identifies the ensuing transformations in gender roles that living in the United States promotes. The strength of the book is its ethnographic account of the labor market opportunities and obstacles which present themselves to a distinct group of people during the migration and settlement process. W^ gain access to this group through rich life histories of some 50 Central American men and women, as well as interviews with 30 representatives of social service agencies, and with 75 employers of migrants in the city. This ethnographic data is backed by a supplementary survey of 100 respondents. Repak skillfully illustrates both the specific roots of this migration as well as the complexity of the networks that draw and place migrants in gender-segregated job markets in the city. Women migrants, even those who are educated and possess skills, have a difficult time moving out of domestic service jobs. For the majority of women, upward mobility results from moving from a live-in domestic position to day work. Repaks analysis suggests that this is due to limited job alternatives in the Washington, D.C., labor market for immigrant women who lack a command of the English language, rather than to legal status or deficiencies in standard human capital measures. In spite of the fact that, by most standards, the women hold distinct advantages over the men, it is the men who manage to experience upward occupational and wage mobility. The analysis is limited by the representativeness of this sample of Washington, D.C., Central Americans. How are the life history interviewees (from a snowball sample) different from the "randomized" supplementary survey sample that is included for the analysis of labor market mobility? The result is a somewhat muddled picture of the characteristics of the Central American population in Washington, D.C. This causes problems for the reader who tries to understand the basis and the context from which Repak draws her conclusions. Regardless of this shortcoming, Repaks analysis supports her argument that women were responsible for altering the course of events for their families by migrating, and the course of events also changed both women and men through migration. This makes the book an important contribution to the field of migration and settlement and a potentially valuable resource for the classroom. Special Sorrows: The Diasporic Imagination of Irish) Polish ^ and Jewish Immigrants in the United States. By Matthew Frye Jacobson. Cambridge, MA, and London: Harvard University Press, 1995. Pp. 321. $45. JOHN J. BUKOWCZYK Wayne State University Matthew Frye Jacobson leaves behind the questions and methodologies of the new social history that animated the study of immigrants and immigration during the past three decades and rejoins a subject which preoccupied older group historiography, namely, the contested terrain of immigrant nationalism. Jacobson imaginatively has drawn together the nationalist impulse of three ethnic groups - Poles, Irish, and Eastern European Jews - whom previous scholars noted more often for their cultural differences than similarities. All three immigrant diasporas, J a- 193 INTERNATIONAL MIGRATION REVIEW cobson argues, built a national consciousness around the "cherished injurpes]" (Jacobson's phrase) and "special sorrows'" (Rosa Luxemburg 's) endured by each as an oppressed people (p. 10). Jacobson demonstrates the gendered nature of ethnic nationalism as expressed in popular calls to arms and realized in armed struggle to attain state power; By turn, he also observes how Irish, Polish, and Yiddish (his term) nationalist sensibilities shaped immigrant attitudes in the 1890s as the United States turned to relieve the sufferings of a kindred colonized people closer to home, declaring war on Spain in tacit support of the Cuban insurrection, only later to see American imperialists transform the Philippines into what immigrant anti-imperialists began to call "America's Ireland." Along the way, however, these same immigrant groups discovered the role of gender and race in defining and articulating American nationhood, as their "whiteness" also inclined some to sympathize with American imperial ambitions abroad and thereby wedded them to putatively superior white Anglo-Saxon American civilization: "'becoming American' depended upon 'becoming Caucasian " (p. 182). It is in these immigrants' emergent American identity that Jacobson takes most interest. The translation of their respective diaspora political cultures "into some understanding of America as a political idea," J acobson argues, was founded upon the ideologies of "[nationalism, gender, and race" (p. 215). While Jacobson's work brings the perspective of scholarly distance to a vital subject typically draped in