CENTER FOR AMERICAN POLITICS AND PUBLIC POLICY OCCASIONAL PAPER SERIES (92-2) Circular Migration and Employment Among Puerto Rican Women Vilma Ortiz July 15, 1992 This paper was originally presented at the conference, Puerto Rican Poverty and Migration, on May 1, 1992 in New York City. This research was supported by grants from the Center for American Politics and Public Policy, the Chicano Studies Research Center, and the Faculty Senate at UCLA. I would like to acknowledge comments from Edward Telles, Saskia Sassen, Edwin Melendez, and Antonio Serrata and acknowledge research analysis assistance of Estela Garcia. Center for American Politics and Public Policy University of California, Los Angeles 310 GSLIS Building, 405 Hilgard Avenue Los Angeles, California 90024 (310) 206-3109 1300 19th Street, NW, Suite 300 Washington, D.C. 20036 (202) 296-8226 CIRCULAR MIGRATION AND EMPLOYMENT AMONG PUERTO RICAN WOMEN Economic status and migration among Puerto Ricans have long been thought to be associated although theorizing regarding this causal relationship has shifted. While earlier migration waves were considered strongly influenced by economic push factors on the island and economic pull factors in New York, current speculation is that migration, particularly circular migration, is a primary factor in explaining Puerto Ricans' recent economic decline. Thus the causal reasoning has shifted from that of economic forces impacting migration to migration impacting economic position. This paper attempts to evaluate the relationship between employment and migration among Puerto Rican women in the 1980s. Background Puerto Rican migration is a pervasive phenomenon, that is understudied and not well understood. Puerto Rican migration did not begin on a large-scale until the late 1940s and reached its peak during the 1950s and early 1960s. It has fluctuated systematically according to employment conditions in the U.S. relative to those in Puerto Rico (Maldonado, 1976). The earliest wave of migrants in the 1940s and 1950s came to the U.S. as a result of the limited employment opportunities in Puerto Rico and the lure of plentiful, low skilled, manufacturing jobs in New York City. Subsequently, migration has continued to be greater during periods of poor economic conditions in Puerto Rico, and return migration to the island greater when economic conditions in the U.S. declined (Centro de Estudios Puertorriquenos, 1979). Puerto Rican migrants have typically been of low socioeconomic status, particularly in comparison to nonmigrants remaining in Puerto Rico (Ortiz, 1986; Gurak, et al., 1987). Low education and few occupational skills has meant that Puerto Ricans were employed mostly in low-skilled jobs in New York City. As the structure of the economy shifted from a production base to a service one, particularly in high-tech sectors, Puerto Ricans lost existing job opportunities in the garment industry (Cooney, 1979; Cooney and Colon-Warren, 1979; Karsada, 1985, 1987; Ortiz, 1992; Sassen-Koob, 1985; Waldinger, 1985). Since Puerto Ricans have not secured an alternative niche in the labor market, the labor force participation of Puerto Ricans has declined over time, particularly among men (Cooney and Colon-Warren, 1979; Ortiz, forthcoming in Baca Zinn; Tienda, 1984). Coupled with the declines in employment have been dramatic changes in the family structure and economic well-being of Puerto Ricans (Ortiz, forthcoming in Baca Zinn). The rate of female-headed households has increased dramatically in the last two decades, from 15 percent in 1960 to 45 percent in 1985. These figures are similar to those among African- American families and higher than among any other racial/ethnic group in the U.S. Also, the rate of marriage has declined and the rate of premarital births has increased. Economic well-being, as indicated by median family income and poverty level, is declining and is the lowest of all racial/ethnic groups. Moreover, Puerto Ricans have shown an increase in, and high rate of, welfare utilization (Bean and Tienda, 1987; Tienda and Jensen, 1986). One explanation for the worsening conditions among Puerto Ricans has been their migration experiences. More specifically, it has been argued that circular migration between Puerto Rico and the mainland has not allowed Puerto Ricans to establish roots and move up economically (Tienda and Diaz, 1987). A number of articles in the popular media have argued that circular migration hurts the educational progress of children, threatens the stability and support system of the family, and produces an identity crisis (Los Angeles Times, 3/16/86, Section IV; New York Times, 6/5/87, p. 81). Despite these assumptions regarding the effect of circular migration, there are no empirical studies on the extent of circular migration experiences among Puerto Ricans, and the effect of this on their economic standing. The causal reasoning underlying this argument is that Puerto Ricans are not progressing economically because they are moving back-and-forth between the U.S. and Puerto Rico. While it may be true that migration will impact negatively on future employment and income, it may also be true that Puerto Ricans migrate because they lose or can not find a job (Rodriguez, 1988-89). In fact, Puerto Rico's Planning Board has documented that a majority of Puerto Rican migrants were without a job before migrating to the U.S. (Junta de Planificacion, 1986). In addition, migration from Puerto Rico has fluctuated systematically according to employment conditions in the U.S. and Puerto Rico (Maldonado, 1976). The purpose of this analysis is to evaluate the circular migration hypothesis among Puerto Rican women. This is done by (1) examining the effect of employment on subsequent migration and (2) comparing various migrant categories with respect to their current employment status. In this study, circular migration is defined as a series of sequential migration outcomes-first out-migration from Puerto Rico, return migration to Puerto Rico, and second instances of out and return migration. The analysis is based on two surveys of Puerto Rican women conducted in the mid-80s that collected detail migration and employment histories. Data The data for this analysis are from two surveys. One is the Puerto Rico Fertility and Family Planning Assessment (PRFFPA), an island-wide social, demographic, and maternal/child health study.[1] The population surveyed were women 15 to 49 years of age living in Puerto Rico in 1982. Retrospective event history data was collected on fertility, marriage, migration, education, employment, and contraceptive use since 1978. There were 3,175 women with complete interviews in the sample. The second survey is the New York Fertility, Employment and Migration (NYFEM) survey conducted in 1985.[2] The population surveyed were women aged 15-49 who were born in Puerto Rico or who had at least one parent born in Puerto Rico and who resided in one of ten counties of the New York City metropolitan area. The population from which the sample was selected accounts for 60 percent of Puerto Rican women aged 15-49 on the mainland. The survey yielded a total of 2,032 interviews with detailed event histories of pregnancies, live births, marital unions, employment, migration, education, and family planning/contraceptive behavior. In addition, the survey obtained information on parental and spouse's characteristics. One limitation of these surveys for the present analysis is that the sample of Puerto Ricans residing in the U.S. is restricted to New York. The Puerto Rico survey includes women who may migrate to (or have returned from) areas in the U.S. other than New York; these women, however, are not in the New York sample. This limitation can not be ignored since Puerto Rican migration to areas other than New York have increased significantly in recent years. However, a fairly high percentage of Puerto Ricans do continue to migrate to the New York area. Moreover, these are the only survey data currently available for addressing questions about migration in any detail. Variables Migration. The data available, because they are rich in detail, are complicated to analyze. Through the detailed migration histories, it was possible to identify instances of migration defined as a change in residence between Puerto Rico and the U.S. of greater than three months duration. Migration (or out-migration) then refers to a move from Puerto Rico to the U.S. while return migration refers to a move from the U.S. to Puerto Rico. We identified four key migration variables: (1) whether respondents migrated from Puerto Rico; (2) whether they returned to Puerto Rico; (3) whether they migrated a second time out of Puerto Rico; and (4) whether they returned to Puerto Rico a second time. Each subsequent migration experience is based on the subsample who experienced the prior migration experience. This analysis was restricted to women born in Puerto Rico and residing in Puerto Rico at age 15. The overwhelming majority of the Puerto Rico survey was born in Puerto Rico (89 percent) while approximately 64 percent of the New York survey was born in Puerto Rico. Also, most of the Puerto Rico sample resided there at age 15 (92 percent) while 39 percent of the Puerto Rican-born in the New York sample resided in Puerto Rico at age 15. In sum, 84 percent of the Puerto Rico sample, and 36 percent of the New York sample, were used in the analysis (for a total sample size of 3405).[3] Employment. The employment histories obtained information about eight jobs in the Puerto Rico survey and six jobs in the New York survey, including occupation, industry, start date, end date, etc. In order to examine the effect of employment on migration, the employment variables were defined relative to the migration experiences. Thus the key employment variables are: (1) employed prior to the first out-migration (or the interview date if never migrated); (2) employed between the first out- migration and first return-migration (or interview date if did not return to Puerto Rico); (3) employed between the first return-migration and the second out-migration (or interview date); and (4) employed between the second out- migration and second return-migration (or interview date if did not return to Puerto Rico).[4] Current employment status was also used in the analysis as a dependent variable, in order to examine the effect of prior migration experiences on subsequent employment. Education. Education was defined in categories because the effect on migration was found to be curvilinear. The categories used were 8 or less years of schooling; 9 to 11 years of schooling; 12 years of schooling; and more than 12 years of schooling. Age. Age was held constant in the analysis since older women have a greater opportunity to migrate. In the analysis examining migration subsequent to the first out-migration, age at the previous migration was held constant. While migration occurs at a fairly young age (as will be documented below), few women migrate between the ages of 15 and 19. Therefore, the analysis will be restricted to women age 20 and older.[5] After careful defining of the variables to make them consistent across surveys, the two data sets were merged in order to carry out multivariate analysis. Logistic regressions, in which the probability of migration was the dependent variable and age, education, and prior employment were the independent variables, were carried out. Descriptive Findings Extent of migration. Figure 1 describes the migration histories of Puerto Rican women. We see that the majority (60 percent) never migrate. Among women who migrated to the U.S. once, 50 percent are likely to return to Puerto Rico. Among those who return to Puerto Rico, 38 percent are likely to migrate again to the U.S.; and among those migrated a second time to the U.S., 45 percent are likely to return a second time to the island. Fully one-third to one-half of women who migrated are likely to have another migration. However, as can be seen from Table 1, the overall rate of migration is actually low since a significant number of women never migrate and among migrants, most women have migrated only once-from Puerto Rico to the U.S., or twice--from Puerto Rico to the U.S. and then back to Puerto Rico. The multivariate analysis will examine each of the migration outcomes presented in Figure 1: (1) migrate from Puerto Rico; (2) return to Puerto Rico; (3) migrate again; and (4) return again. The first outcome-migrate from Puerto Rico-is based on the women 20 years of age and older who were born in Puerto Rico and resided there at age 15. For each subsequent migration outcome, the sample population is restricted to those who experienced the previous migration; in other words, whether women return to Puerto Rico or not is examined among those who initially migrated from Puerto Rico. Description of age, education, and employment. Table 2 presents a description of women included in this analysis with respect to age, education, and employment separately for the four samples used in the multivariate analysis. While the average age of all women was 34 years, the migrants were slightly older-about 36 to 38 years old. Women's first migration tended to occur at a fairly young age (about 20 years old). Not surprisingly, subsequent migrations tended to occur at an older age. Approximately 30 percent of all women had low levels of education-eight years or less; and approximately a similar percentage had some college. The variation in the educational distribution for the three samples of migrants suggest that education has a significant effect on who migrates; this is documented in the multivariate analysis. Almost half of all women had jobs in Puerto Rico before migrating (or the interview date for non-migrants). A similar percentage had jobs between their first migration and first return (or interview date for those who never return). Approximately a third have jobs between returning to Puerto Rico and migrating again, and between migrating a second time and returning to Puerto Rico a second time. About 30 percent of all women were employed at the time of the interview. The differences among the migration groups with respect to current employment status are evaluated in the multivariate analysis. Multivariate Analysis Effect of age. Table 3 presents logistic regressions of migration regressed on age, education, and employment. Age has a consistent and positive effect on each migration experience-older women are more likely to have migrated to the U.S. and more likely to have returned to Puerto Rico. This is primarily the result of greater exposure to the "risk" of migration.[6] Effect of education. Education has a strong and curvilinear effect on migration (see Table 3). For the first out-migration, women with some high school (9 to 11 years of schooling) or who completed high school were significantly more likely to migrate than those with the lowest education and the highest education. This effect is similar for the second out- migration experience, where women with 9 to 11 years of schooling were the most the likely to migrate while the women with less than 8 years or with 12 or more years of education were less likely to migrate. The effect of education on return-migration (for both instances of return migration) was opposite to the effect of education on out-migration. Those with 9 to 11 years of education were less likely to return to Puerto Rico while those with the lowest education and those with a high school or greater education were more likely to return. Effect of employment. For each migration experience, we see that not having a job prior to the migration has a large negative effect on the subsequent migration (see Table 3). Not having a job in Puerto Rico meant that Puerto Rican women were more likely to migrate from Puerto Rico than if they had a job. Moreover, having a job in Puerto Rico prior to the first migration not only affects that migration decision but it also continues to affect subsequent decisions-those that previously held jobs in Puerto Rico were more likely to return to Puerto Rico after their migration and even after a second out-migration. In a similar fashion, those who did not have a job in the U.S. after their first migration were more likely to return to Puerto Rico. Moreover, having a job after their first migration, while not predicting the second out-migration from Puerto Rico, did predict the second return to Puerto Rico. Not having a job in Puerto Rico after the first return, affects a second out-migration from Puerto Rico and not having a job in the U.S. after this migration impacts on a second return migration to Puerto Rico. Thus employment status between migration experiences greatly affects subsequent migrations among Puerto Rican women. Current employment. Table 4 present a logistic regression in which current employment is regressed on age, education, having a job in Puerto Rico prior to the age of 20 [7], and migration status. Age has a significant effect on employment such that older women are more likely to be employed. More educated women are more likely to be employed than less educated ones. Having had a job in Puerto Rico strongly impacts current employment status. Most importantly for the purposes of testing the circular migration hypothesis is the comparison among migrant groups. Those who migrated once from Puerto Rico and returned once to Puerto Rico are less likely to be employed than nonmigrants who remained in Puerto Rico (the reference group). However, those with a second instance of out-migration and return-migration do not differ significantly from non-migrants in their current employment. Moreover, the four migrant groups do not differ significantly among each other with respect to their current employment status. Thus while migration status impacts subsequent employment for those who have migrated once, it does not appear to affect the employment of those who have migrated more than once. Summary and Conclusions While migration is a significant phenomenon among Puerto Rican women, we do not find that large numbers are migrating back and forth between the island and the mainland. This study indicates that the majority of Puerto Rican women had not migrated from Puerto Rico at the time of these surveys, while approximately 40 percent of Puerto Rican women had migrated. Among these, a minority have had more than one migration experience. This counters common perceptions that circular migration is prevalent and widespread among Puerto Ricans and consequently the primary reason for their economic status. The results support the findings of previous studies that Puerto Rican migration is not highly selective in that higher educated persons were not as likely to migrate than the less educated. On the other hand, the least educated were also not the most likely to migrate. Instead women with some high school education were the most likely to migrate from Puerto Rico and the least likely to return. This could be a result of greater job opportunities for this educational group in the New York area relative to Puerto Rico; in contrast, women with higher education should have better employment opportunities in Puerto Rico while women with the lowest educational level may have few opportunities in the New York region and face more difficulties in being able to migrate. This study evaluated the circular migration thesis by examining whether employment prior to migration was a critical factor in influencing migration. The finding that migration is strongly affected by lack of employment opportunities prior to migration supports the opposite explanation that employment impacts migration. Moreover, the fact that those who migrated back-and-forth between Puerto Rico and the U.S. did not have significantly lower employment at the time of the interview suggests that the circular migration hypothesis is not an adequate explanation for understanding the continued economic position of Puerto Ricans today. Better explanations might be found in research showing that Puerto Ricans are being negatively impacted by economic restructuring in the Northeast region of the U.S. (Ortiz, 1991). It appears then that Puerto Ricans use migration between Puerto Rico and the U.S. in a strategic manner to optimize employment opportunities. FOOTNOTES [1] This survey was conducted by the Center for Disease Control. [2] This project was directed by Mary Powers, Douglas Gurak, and John Macisco at Fordham University and the data collected by the staff of the Institute of Survey Research at Temple University. Prior analysis with these data have focused on the factors affecting migration and female headship (Falcon, 1988; Gurak, et al., 1987, 1988). [3] In the detailed migration history, information was obtained for eleven residences of least three months duration in the Puerto Rico survey; and eight residences in the New York survey. Since these moves included migration out of Puerto Rico as well as migration internal to Puerto Rico or to the U.S., only moves that involve leaving or returning to Puerto Rico were counted for the present analysis. Because questions were asked in different ways in the two samples, care was taken to make the variables comparable. For instance, in the Puerto Rico survey, the detailed migration history was obtained from the age of 15 on, while in the New York sample, this history was obtained from the age of 10 on. In order to make the two surveys comparable, information from the New York survey was limited to migration experiences occurring after the age of 15. In a similar manner, the Puerto Rico survey asked about dates of residence while the New York survey asked about dates of moves; this difference was made comparable by making the move date as the end date of residence at a particular location. [4] Because the calculation of these variables involves the comparison of dates (i.e., migration dates to beginning and end of employment dates), care was taken to identify logical errors in the dates. [5] Although the "risk" of migration increases significantly with age, the substantive findings from the multivariate analysis do not change by limiting the analysis to older women (e.g., those over 30 years old or those over 40 years old). Thus the analysis for women between 20 and 49 are presented. [6] While it is possible that these differences may reflect period effects during which women were more likely to migrate, descriptive statistics do not support this. Comparing migration (while controlling for age at migration) by age at interview (in 5- year categories) reveal small differences by age cohort. [7] For the purposes of this analysis, having a job in Puerto Rico prior to the first migration is defined slightly differently than in the previous analysis of migration. Only jobs held prior to the age of 20 are counted in this indicator for this analysis. The reason being that for many nonmigrants (the reference group in the comparison of migration status), current employment and employment in Puerto Rico refer to the same job. Since the analysis is conducted among women who are 20 years old or older, limiting this measure to jobs prior to the age 20 means that employment is examined for two different age periods. Approximately 30 percent of all women had a job in Puerto Rico prior to the age of 20. REFERENCES Bean, Frank and Marta Tienda. 1987. The Hispanic Population of the United States. New York: Russell Sage Foundation. Centro de Estudios Puertorriquenos. 1979. Labor Migration Under Capitalism. New York: Monthly Review Press. Cooney, Rosemary. 1979. Intercity variations in Puerto Rican Female Participation. JournaI of Human Resources 14: 222-235. Cooney, Rosemary and Alice Colon-Warren. 1979. Declining Female Participation among Puerto Rican New Yorkers: A Comparison with Native White Nonspanish New Yorkers. Ethnicity 6:281-297. Falcon, Luis, Douglas Gurak, and Mary Powers. 1988. Factors Associated with Female-Headship Among Puerto Rican on the Mainland. Presentation at the American Sociological Association meetings in Atlanta. Gurak, Douglas, et al. 1987. The Effects of Marriage and Family Characteristics of Puerto Rican Women Residing on the Island on their Probability of Migrating to the Mainland. Presentation at the American Sociological Association meetings in Chicago. Gurak, Douglas, Mary Powers, and Luis Falcon. 1988. The Effects of Marriage and Family Characteristics of Puerto Rican Women Residing on the Mainland on the Probability of Migrating to the Island. Presentation at the Population Association of America meetings in New Orleans. Junta de Planificacion. 1986. Kasarda, John. 1985. Urban Change and Minority Opportunities. In Paul Peterson (Ed.), The New Urban Reality. Washington, DC: Brookings Institution. Kasarda, John. 1987. Dual Cities: The Structure of Urban Poverty. New Perspectives Quarterly 4(1): 8-11. Los Angeles Times. 1986 (3/16). Latinos Enter the Mainstream Quickly. By Lester Thurow. Section IV. Maldonado, Rita. 1976. Why Puerto Ricans Migrated to the U.S. in 1947-1973. Monthly Labor Review 99(9): 7-18. New York Times. 1986 (6/5). Striving But Still Lagging: Puerto Ricans Wonder Why. By Linda Chavez. Page B1. Ortiz, Vilma. 1986. Changes in the Characteristics of Puerto Rican Migrants from 1955 to 1980. International Migration Review 20(3): 612-628. Ortiz, Vilma. 1991. Latinos and Industrial Change in New York and Los Angeles. In Edwin Melendez and Clara Rodriguez (Eds.), Hispanics in the Labor Force. New York: Plenum Press. Ortiz, Vilma. Forthcoming. Women of Color: A Demographic Overview. In Maxine Baca Zinn and Bonnie Thornten Dill (Eds.), Women of Color in American Society. Forthcoming. Rodriguez, Clara. 1988-89. Puerto Ricans and the Circular Migration Thesis. Journal of Hispanic Policy 3: 5-9. Sassen-Koob, Saskia. 1985. Changing Composition and Labor Market Location of Hispanic Immigrants in New York City, 1960-1980. In George Borjas and Marta Tienda (Eds.), Hispanics in the U.S. Economy. New York Academic Press. Tienda, Marta. 1984. The Puerto Rican Worker: Current Labor Market Status and Future Prospects. In Puerto Ricans in the Mid '80s: An American Challenge. Washington, DC: National Puerto Rican Coalition. Tienda, Marta and William Diaz. 1987 (8/28). Puerto Rican Circular Migration. New York Times. Op Ed page A31. Tienda, Marta and Leif Jensen. 1988. Poverty and Minorities: A Quarter- Century Profile of Color and Socioeconomic Disadvantage. In Gary Sandefur and Marta Tienda (Eds.), Divided Opportunities: Minorities, Poverty, and Social Policy. New York: Plenum Press. Waldinger, Roger. 1985. Immigration and Industrial Change in the New York Apparel Industry. In George Borjas and Marta Tienda (Eds.), Hispanics in the U.S. Economy. New York Academic Press. Figure 1 [unable to reproduce] Table 1 Migration History among Puerto Rican Women[a] Percent Percent Number of Migrations 0 60.0 -- 1 20.1 50.2 2 12.4 31.0 3 4.1 10.2 4 2.2 5.5 5 or more 1.2 3.0 N [2931] [1173] [a] Women, age 20 to 49, born in Puerto Rico and residing in Puerto Rico at age 15 Table 2 Age, Education, and Employment History among Puerto Rican Women[a] All First-time Return Migrants Women Migrants Migrants Again Mean Age At interview 34.2 36.6 36.9 37.9 At first migration 21.0 20.4 19.3 At first return 25.4 22.9 At second migration 26.3 Years of Schooling % 0 to 8 years 29.1% 30.8% 36.6% 37.6% % 9 to 11 years 15.6 28.6 15.4 27.1 % 12 years 23.2 24.3 26.7 21.7 % 13 or more years 32.0 16.2 21.4 13.6 [100.0] [100.0] [100.0] [100.0] Employment % with job prior to first migration 47.8% 19.8% 22.7% 17.6% % with job between first migration and first return 46.5% 38.6% 29.4% % with job between first return and second migration 33.8% 16.3% % with job between second migration and second return 34.8% % Currently Employed 29.2% 20.7% 23.8% 23.1% [N] [2931] [1173] [585] [221] [a] Women, age 20 to 49, born in Puerto Rico and residing in Puerto Rico at age 15 Table 3 First-Order Partial Derivatives' from Logistic Regression of Migration on Age, Education, and Employment History Migrate Return Migrate Return again again Age at interview 0.017*** 0.008*** 0.023*** 0.023*** Education 9 to 11 years 0.397*** -0.313*** 0.338*** -0.280* Education 12 years 0.200*** 0.017 -0.018 0.006 Education > 12 years 0.018 0.145*** -0.042 0.242 Employed prior to first -0.519*** 0.113*** -0.025 0.398** migration Age at first migration -0.024*** 0.002 -0.005 Employed between first -0.222*** -0.092 0.188* mig. & first return Age at first return -0.041*** -0.011 Employed between first -0.369*** -0.146 return & second mig. Age at second migration -0.042** Employed between second -0.316*** mig. & second return Average .400 .499 .378 .452 [N] [2930] [1171] [584] [220] *p<.05; **p<.01; ***p<.001 [a] Logistic regression is used for this analysis in which an iterative maximum-likelihood solution predicts the logarithm of the probability of migration. First-order partial derivatives are computed as bP'(1-P') where b1 is the coefficient of the relevant independent variable and P' is the proportion that migrated. These can be interpreted as the increment to the average probabiliy of migration associated with a one-unit increase in the independent variable. Table 4 First-Order Partial Derivatives[a] from Logistic Regression of Current Employment on Age, Education, Employment and Migration Status Current Employment Age at interview .011*** Education 9 to 11 years .107*** Education 12 years .258*** Education > 12 years .410*** Employed prior to first migration .126*** Migrated Once -.110** Returned Once -.097*** Migrated Twice -.045 Returned Twice -.086 Average .292 [N] [2930] *p <.05; **p <.01; ***p <.001 [a] Logistic regression is used for this analysis in which an iterative maximum-likelihood solution predicts the logarithm of the probability of migration. First-order partial derivatives are computed as b,P'(1-P') where b is the coefficient of the relevant independent variable and P' is the proportion that migrated. These can be interpreted as the increment to the average probabiliy of migration associated with a one-unit increase in the independent variable. Appendix Table A.1 Unstandardized Coefficients from Logistic Regression of Migration on Age, Education, and Employment History Migrate Return Migrate Return again again Age at interview .0706** .0326*** .0979*** .0922*** (.0057) (.0087) (.0152) (.256) Education 9 to 11 years 1.6548*** -1.2525*** 1.4385*** -1.1311* (.1456) (.1703) (.3159) (.4445) Education 12 years .8332*** .0698 -.0768 .0244 (.1259) (.1725) (.2604) (.4441) Education > 12 years .0749 .5804** -.1784 .9768 (.1276) (.2044) (.2996) (.5727) Employed prior to first -2.1602*** .4538** -.1070 1.6082** migration (.1058) (.1699) (.2736) (.5451) Age at first migration -.0973*** .0069 -.0188 (.0149) (.0351) (.0698) Employed between first -.8868*** -.3622 .7606* mig. & first return (.1356) (.2232) (.3847) Age at first return -.1752*** -.0452 (.0254) (.0741) Employed between first -1.5701*** -.5892 return & second mig. (.2455) (.5202) Age at second migration -.1685** (.0555) Employed between second -1.2787*** mig. & second return (.3823) Constant -2.4192 1.3974 .5604 2.2189 Average .400 .499 .378 .452 [N] [2930] [1171] [584] [220] *p<.05; **p<.01; ***p <.001 Appendix Table A.2 Unstandardized Coefficients from Logistic Regression of Current Employment on Age, Education, Employment and Migration Status Current Employment Age at interview .0535*** (.0057) Education 9 to 11 years .5197** (.1729) Education 12 years 1.2458*** (.1386) Education > 12 years 1.9808*** (.1321) Employed prior to first migration .6108*** (.0939) Migrated Once -.5326** (.1367) Returned Once -.4703*** (.1446) Migrated Twice -.2184 (.2316) Returned Twice -.4156 (.2819) Constant -3.8958 Average .292 [N] [2930] *p