Changes in the Relative Earnings Gap between Natives and Immigrants Along the U.S.-Mexico Border Alberto Dávila Professor and V.F. “Doc” and Gertrude Neuhaus Chair for Entrepreneurship Department of Economics and Finance College of Business Administration University of Texas—Pan American Edinburg, TX 78541 USA Fax: 011-956-384-5020; [email protected] and Marie T. Mora Associate Professor Department of Economics and Finance College of Business Administration University of Texas—Pan American Edinburg, TX 78541 USA Fax: 011-956-384-5020; [email protected] March 2006 Preliminary Draft: Please do not quote without permission from the authors. Changes in the Relative Earnings Gap between Natives and Immigrants Along the U.S.-Mexico Border Abstract Using 1990 and 2000 U.S. census data, this study investigates changes in the immigrant/native earnings disparities for workers in U.S.-cities along the Mexican border vis-à-vis the U.S.-interior during the 1990s. Our findings—based on estimating earnings functions and employing the Juhn-Murphy-Pierce (1993 Journal of Political Economy) wage decomposition technique—indicate that the average earnings of Mexican immigrants along the U.S.-Mexico border improved relative to those accrued by their U.S.born Mexican American counterparts during the 1990s. However, when comparing Mexican-born workers to U.S.-born non-Hispanic whites, the immigrant border-earnings “penalty” remained statistically unchanged between 1990 and 2000. Changes in the Relative Earnings Gap between Natives and Immigrants Along the U.S.-Mexico Border Introduction Workers along the U.S.-Mexico border earn less on average than those in the U.S. interior. This stylized fact has been well researched, and by most accounts it has been at least attributed to the relatively low levels of human capital along the border region.1 It has also been suggested that immigration serves to depress wages in the region because of an assumed highly elastic labor supply of Mexican labor. In pursuing this latter type of explanation for depressed earnings along the border, more recent studies by Mora (forthcoming) and Mora and Dávila (2006a) have pointed to the importance of distinguishing between U.S.-natives and Mexican immigrants, and have provided supporting evidence of significant differences in the border/earnings disparity according to immigration status. Extant recent literature, however, has not fully accounted for potential changes in regional earnings gaps over time. This research void is particularly important to address owing to the socioeconomic changes that have occurred along the U.S.-Mexico border over the last fifteen years. For one, the U.S.-Mexico border region has flourished as a result of increasing trade agreements between Mexico and the U.S., particularly as a result of the North America Free Trade Agreement (NAFTA). This enhanced trade might have had, and should continue to have, important labor market implications regarding the relative demand for immigrant and native labor along the U.S.-Mexico border. With regards to relative supply issues, immigration reform, starting most notably with the Immigration Reform and Control Act of 1986 (IRCA), has also brought about an increase of border enforcement during this time that might have impacted the labor flows between these two countries (e.g., Dávila, Pagán, and Soydemir, 2002). 1 See, for example, Mora and Dávila (2006a); Mora (forthcoming); Flota and Mora (2001); Fullerton (2001); Dávila and Mora (2000); Sharp (1998); Peach (1997); Dávila and Mattila (1985); and Dávila (1982). 1 This paper extends the foregoing border literature by empirically investigating how immigrant/native earnings disparities changed for residents in U.S.-cities along the Mexican border vis-àvis the U.S.-interior during the 1990s. In addition to estimating standard earnings functions, we employ the Juhn-Murphy-Pierce (1993) wage decomposition technique to analyze changes in immigrant/native earnings differentials for border and non-border residents during this time period. The findings from these analyses are important to a variety of policy-relevant issues. First, to what extent is the aforementioned border-interior earnings gap a native or immigrant phenomenon? Second, to the extent that rising trade opportunities with Mexico disproportionately favor U.S. cities located near Mexico, do they similarly affect the earnings of Mexican immigrants as U.S.-born workers? Answers to these questions are useful in designing a more comprehensive economic policy plan for the border region that distinguishes between the labor-market experiences of natives and immigrants. Moreover, such information is useful with respect to some of the intended (and unintended) consequences of U.S. immigration policy. Background and Conceptual Issues According to neoclassical theory, regional earnings differentials should disappear over the long run for at least three reasons. First, labor migration from low wage areas to high wage areas should reduce the supply of labor in the former region and increase the supply of labor in the latter region. The geographic changes in labor supply would lead to higher wages in those areas with depressed wages and vice-versa. The second reason for the eventual regional-wage conversion is that entrepreneurs and capital would gravitate to those areas with relatively lower labor costs. The lower costs would in turn lead to higher profits and, as a result, a stronger competitive position for capitalists. Capital migration to the relatively low-wage areas would increase labor demand there at the expense of the demand for labor in the high-wage region. As such, earnings would increase in the low wage areas vis-à-vis their high wage counterpart. 2 Finally, those regions with lower earnings would be able to produce goods and services at lower prices and would therefore attract consumer demand for its products. As in the case above, this consumer-demand effect would drive labor demand in the region with lower earnings up relative to the region with high earnings, leading to eventual wage equality between the two regions. This theory assumes, of course, perfect mobility of human and capital resources and perfect information. With regards to labor migration for the purposes of our study, there might be some conceptual reasons to believe that immigrant and native labor would have different tendencies to migrate as this theory predicts. On the one hand, it is possible that immigrant Mexican labor along the U.S.Mexico border is more mobile given that it has more migratory experience than native labor of the region. On the other hand, the immigrant labor might be more attached to living close to Mexico than natives. The neoclassical view presented here also holds if workers are homogeneous and are valued equally by entrepreneurs and capital. This assumption is problematic when comparing Mexican immigrants to U.S.-born workers in an era of the North American Free Trade Agreement (NAFTA). As argued by Mora and Dávila (2006a), expanding trade opportunities with Mexico might have favored bilingual and bicultural workers. In particular, Mexican-born workers on the U.S. side of the border potentially face more employment and business opportunities than their U.S. born counterparts because of the latters’ need to hire cultural “translators”, such as for language, insight into Mexican business practices, and understanding the Mexican consumer. In what follows, we employ this conceptual framework to discuss our results. Of particular interest in this discussion is the identification of changes in the border-interior earnings gaps between natives and Mexican immigrants over the 1990s as the results of increasing trade between the border region and Mexico and changes in immigration reform. Data and Empirical Results We employ data from the 1990 and 2000 U.S. censuses available from Ruggles and Sobek, et al. (2003) in the Integrated Public Use Microdata Series (IPUMS) to analyze the relative earnings of Mexican immigrants along the US-Mexico border. Our sample includes U.S.- and foreign-born Mexican 3 Americans between the ages of 25 and 64 who worked for wages and salaries for at least 20 hours per week for 32 weeks in the previous year. For the sake of comparison, we also consider the conventional base-group of US-born non-Hispanic whites who only speak English at home. We exclude the selfemployed here because recent work illustrates that factors related to entrepreneurial activities significantly differ between U.S.-cities located on the Mexican border versus the U.S.-interior (Mora and Dávila 2006a; Mora, forthcoming; Flota and Mora 2001). In this study, we define the border region as the eight identifiable metropolitan statistical areas (MSAs) in counties located directly on the Mexican border (Brownsville, El Paso, Laredo, McAllen, in Texas; Las Cruces in New Mexico; Tucson and Yuma in Arizona, and San Diego in California). Using MSAs to define the border is consistent with a host of extant work (e.g., Mora and Dávila 2006a; Robles 2002; Flota and Mora 2001; Dávila and Mora 2000; Dávila and Mattila 1985; Smith and Newmann, 1977), although alternative definitions exist.2 Because municipalities with less than 100,000 individuals cannot be identified in the IPUMS, we restrict the sample to workers in MSAs to avoid including those who reside along the border—but outside of an MSA—in the U.S.-interior sample. To ensure an adequate sample size for the border region, we use the 5% IPUMS for both years when constructing the border MSA sample. The U.S.-interior sample is from the 1% IPUMS in 1990, and a random 20% sample drawn from the 5% IPUMS in 2000;3 all analyses employ appropriate sampling weights. Table 1 provides the average characteristics of Mexican immigrants, U.S.-born Mexican Americans, and U.S.-born non-Hispanic whites in cities along the US-Mexico border and those in the 2 For example, some studies follow a “county-approach” to identify the border (e.g., Mora 2005; Fullerton 2001; Peach and Adkisson 2000; Peach 1997), but such an approach is not possible when using the 1990 and 2000 IPUMS. However, most of the recent growth in trade and labor market opportunities along the border has occurred in cities, such that we do not consider the lack of county information to be problematic. Also, while we are aware that some researchers occasionally use interior cities in their border definitions, such as San Antonio, Texas, or Los Angeles, California, because of their large Mexican populations, we believe that our border definition is superior because cities physically located near the border possess specific cultural and economic interactions between the US and Mexico that cannot be observed in the U.S.-interior. 3 We had initially planned to use the 1% IPUMS for 2000 to identify workers in the U.S.-interior. Unfortunately, metropolitan areas with fewer than 400,000 individuals cannot be identified in the 1% sample that year; as such, we use a randomly-drawn one-in-five sample from the larger 5% IPUMS (which identifies MSAs with at least 100,000 residents) for 2000. 4 interior. Despite higher average education levels in U.S. MSAs near Mexico, a border/interior earnings gap exists for the three groups. Earnings are defined as the annual wage and salary income divided by usual weekly work-hours times weeks worked. This finding indicates that low average wages along the U.S.-Mexico border are not unique to immigrants, although on the surface, they had the largest borderearnings “penalty”. Moreover, while the immigrant/native earnings differential is larger in U.S.-border cities than in the rest of the country when focusing on U.S.-born non-Hispanic whites, this was not the case when considering U.S.-born Mexican Americans. (TABLE 1 ABOUT HERE) Of specific interest to this study are changes in these wage differentials between 1990 and 2000. The average earnings gap between Mexican immigrants and U.S.-born Mexican immigrants narrowed slightly along the border (from 28.7 percent to 26 percent), but increased in the U.S.-interior (from 28.7 percent to over 31 percent). Moreover, compared to non-Hispanic whites, the immigrant/native wage gap increased in both border and non-border cities, but the increase in border cities was smaller. These results suggest that, compared to their counterparts in the U.S.-interior, Mexican immigrants in areas close to Mexico gained ground with respect to earnings relative to their U.S.-born peers. Some of the earnings improvement for foreign-born versus U.S.-born Mexican Americans in cities near Mexico presumably stems from the larger increase in the education levels of immigrants between 1990 and 2000. The average schooling of Mexican-born workers residing close to their home country rose by an entire year (from 8.9 to 9.9 years), compared to 0.7 years (from 8.3 to 9 years) for those in the U.S.-interior. The increase in the average education of U.S.-born Mexican Americans and non-Hispanic whites was similar between the border and interior regions (0.5 years for U.S.-born Mexican Americans, and 0.3 years for non-Hispanic whites). Additional observations in Table 1 are consistent with extant work (e.g., Mora and Dávila 2006a; Dávila and Mora 2000) in that, compared to those in the U.S.-interior, Mexican immigrants in border MSAs tend to be more “established” in the U.S. and possess greater levels of human capital. On average, Mexican workers in U.S.-border cities have more education, more years of potential work experience 5 (measured by the convention of age – education – 5), higher English-language fluency (defined as the ability to speak the English language “well”), longer U.S.-tenure, and a higher representation of female workers than their counterparts in the rest of the U.S. Note that these differences intensified between 1990 and 2000, which might explain the larger rise in the immigrant/native earnings differential in nonborder versus border cities. Earnings Regressions. For more insight, we now consider the extent to which these observable differences explain changes in the immigrant/native earnings differentials between 1990 and 2000. In particular, we first estimate a standard earnings function: (1) ln(W) = (Border) α1 + (Border x 2000IPUMS) α2 + X β1 + (X x 2000IPUMS) β2 + e , where ln(W) represents the natural logarithm of hourly earnings. The vector Border includes a binary variable equal to one for workers in U.S.-border cities (= 0 otherwise), as well as an interaction term between border residence and Mexican immigrants (= 1 for Mexican immigrants residing along the border; = 0 otherwise). (Border x 2000IPUMS) denotes a vector of the variables in Border interacted with a binary 2000IPUMS variable (=1 for individuals in the 2000 IPUMS; = 0 for 1990). As such, the coefficient vector α1 reflects earnings “penalties” associated with residing along the U.S.-Mexico border, while α2 measures changes in these penalties between 1990 and 2000.4 X is a vector of additional variables that affect earnings [including being a Mexican immigrant (= 1 for immigrants; = 0 otherwise), education, experience, experience-squared, limited-English-proficiency, immigrants’ time in the U.S., gender, and a constant term], while X x 2000IPUMS represents a vector of the variables in X interacted with 2000IPUMS. Finally, e is the normally distributed error term. Table 2 contains the results from estimating Equation (1) three times, using: (a) only Mexican immigrants in the sample, (b) both Mexican immigrants and U.S.-born Mexican Americans, and (c ) Mexican immigrants and non-Hispanic whites. In all three cases, consistent with the border literature, residing in cities along the U.S.-Mexico border relates to significantly lower earnings even when 4 For interpretative ease, we will discuss the estimated α’s as earnings “penalties” or “premiums”, but more refined interpretations of estimated coefficients for binary variables in semi-logarithmic equations exist (see Kennedy 1981). 6 controlling for other observable characteristics; this border earnings penalty is particularly pronounced among Mexican immigrants. However, when focusing on the first two columns, Mexican immigrants along the border experienced an earnings improvement between 1990 and 2000 compared to their counterparts in the rest of the U.S. (approximately 3.5 percent) as well as to U.S.-born Mexican Americans (over four percent). These earnings gains suggest that something happened to favor immigrants residing close to their home country in the 1990s beyond changes in their observable skill levels. As we noted in the conceptual section above, the expansion in trade opportunities with Mexico following NAFTA represents one explanation, and the (potentially) more mobile Mexican immigrant population represents another (but not mutually exclusive) explanation. (TABLE 2 ABOUT HERE) At the same time, when including non-Hispanic white men as the base group of comparison, Table 2 shows that the earnings of workers in U.S.-border cities did not change on average relative to those in the rest of the country during the 1990s. That is, relative to non-Hispanic whites, Mexican immigrants along the border continued to earn the same border “penalty” in 2000 as in 1990. Perhaps this finding reflects changes (or the lack thereof) in the relative immigrant/native skill-differentials between border and non-border metropolitan areas during the 1990s. We will discuss this issue in more detail below. Beforehand, other findings in Table 2 should be noted. First, Mexican Americans receive lower returns to human capital than non-Hispanic white men; this observation presumably reflects the fact that some of the Mexican immigrants’ skills were acquired abroad (e.g., McManus, Gould and Welch 1983). Second, when focusing exclusively on foreign- and U.S.-born Mexican Americans, the returns to schooling, experience, and U.S.-tenure fell slightly during the 1990s, suggesting that low-skilled workers gained ground with respect to labor market income compared to their skilled counterparts; similar findings have been noted elsewhere (e.g., Mora and Dávila 2006c). The inclusion of non-Hispanic whites provides the expected increase in the returns to education and experience, as discussed by Welch (2000) 7 among others.5 While going beyond the scope of this study, this finding raises the question as to whether the increasing returns to skills observed in recent years were primarily a non-Hispanic phenomenon; future research should address this issue. Regional Border Effects. One potential problem with the analyses thus far is the underlying assumption regarding the homogeneity of the U.S.-Mexico border region—a region that stretches nearly 2,000 miles. Other studies have observed striking differences among U.S. cities located near Mexico (e.g., Dávila, Schmidt and Ziegler 1984). As such, perhaps the observed relative improvement in the earnings of Mexican immigrants in U.S.-border MSAs only occurred in one particular region. To address this possibility, we re-estimate Equation (1) for the three samples while using four distinct border variables: (1) El Paso, Texas and Las Cruces, NM; (2) South Texas (Brownsville, Laredo, and McAllen); (3) Arizona (Tucson and Yuma), and (4) California (San Diego). To conserve space, Table 3 reports the regression results related to these border variables; the remaining results can be obtained from the authors. Concentrating on the Mexican immigrant sample, with the exception of San Diego, California, those who resided in border metropolitan areas experienced a significant earnings improvement of about 6-8 percent during the 1990s compared to similar workers in other U.S. cities. Moreover, relative to U.S.-born Mexican Americans, Mexican immigrants in three of the four border regions (this time, with South Texas as the exception) gained significant ground with respect to labor market earnings between 1990 and 2000. It appears that combining the border metropolitan areas into one general region does not distort the overall observation that the average wages of Mexican immigrants in the region improved vis-à-vis those earned by their counterparts in the U.S.-interior and by U.S.-born Mexican Americans, although the extent of this improvement varied between specific border vicinities. Similar to the results in Table 2, however, Table 3 shows that the average earnings of Mexican immigrants in U.S.-MSAs located close to 5 When including non-Hispanic whites in the sample, at first glance the positive coefficient on the limited-Englishproficient (LEP) variable is surprising. However, further robustness tests (available from the authors) indicate that this appears to be a female-effect; excluding women from the sample provides the expected negative sign, although its interaction with 2000IPUMS is positive and statistically significant. See Mora and Dávila (2006c) for a discussion of recent changes in the effects of being LEP on the earnings of Hispanic men versus women. 8 Mexico did not significantly improve or decline during the 1990s when comparing them to non-Hispanic whites. (TABLE 3 ABOUT HERE) Juhn-Murphy-Pierce Wage Decomposition. In short, Tables 2 and 3 indicate that when comparing Mexican immigrants in U.S.-Mexico border cities to those in the rest of the U.S. and to U.S.born Mexican Americans, the average earnings of Mexican-born workers along the border increased, ceteris paribus during the 1990s. However, when comparing their wages to those earned by nonHispanic whites, the border-earnings “penalty” remained statistically unchanged between 1990 and 2000. As noted above, this finding might reflect relatively stable immigrant/native skill-differentials between border and non-border MSAs during this time. An empirical method to address this issue was developed by Juhn, Murphy, and Pierce (1993)— henceforth JMP. The JMP technique allows for the analysis of wage differentials over time, not only in terms of changes with respect to observable characteristics, but also in underlying wage structures. The observable sources of earnings differentials are differences in the levels of human capital and the returns to these factors between two groups. The unobservable factors include the levels of, and the returns to, unmeasured characteristics such as innate ability. Consider the following earnings function for U.S.-native workers (N) in year t: (2) ln(W)tN = XtN βtN + σtN θtN , where (W)tN equals the hourly earnings of this group, and XtN and βtN represent vectors of observable characteristics and the returns to those characteristics. θtN denotes a normal standardized residual, while σtN is the standard deviation of non-border residual earnings in year t. It follows that the wage differential between natives and Mexican immigrants (M) in t can be expressed as: (3) ∆Wt ≡ ln(W)tN - ln(W)tM = ∆Xt βtN + σtN ∆θt , where ∆ denotes the differences in the average variables in X and θ between the two groups of workers. Extending Equation (3) to explore shifts in the immigrant/native wage differential between two time periods (t-1 and t) yields: 9 (4) ∆Wt - ∆Wt-1 = (∆Xt - ∆Xt-1 ) βtN + ∆Xt-1 (βtN - βt-1N ) + (∆θt - ∆θt-1N )σtN + ∆θt-1 (σtN - σt-1N ) . The four right-hand side terms measure the X-effect, the B-effect, the θ-effect, and the σ-effect. The X-effect reflects how changes in average observable characteristics between two time periods contributed to shifts in the immigrant/native wage differential, while the B-effect accounts for changes in the returns to these characteristics. The θ-effect estimates shifts in unobservable skills and characteristics over time, and the σ-effect captures changes in the returns to these unobservable characteristics between Mexican immigrants and U.S.-born workers.6 Table 4 provides the results from estimating Equation (4) when partitioning the sample between workers in U.S.-Mexico border MSAs and those in the U.S.-interior. We first define the native sample as U.S.-born Mexican Americans (see the first two columns), and then as non-Hispanic whites (the last two columns). The regression results for the native samples can be obtained from the authors. (TABLE 4 ABOUT HERE) The first two columns in Table 4 indicate that Mexican immigrants in U.S. cities on the Mexican border gained significant ground with respect to earnings relative to U.S.-born Mexican Americans during the 1990s: the immigrant/native wage gap fell by 2.7 percentage points in the region. In contrast, Mexican-born workers in the U.S.-interior lost ground on average relative to their U.S.-born counterparts between 1990 and 2000. Explanations behind these differences are seen in the four effects from the JMP. While the B-effects indicate that the returns to observable skills in the 1990s should have increased the immigrant/native wage differentials by around 0.065 log-points in both regions, Mexican immigrants residing in U.S.-border MSAs offset this tendency because of relative gains in their skill levels—both observed (the X-effect) and unobserved (the θ-effect). Turning to the final two columns in Table 4, the earnings differential between Mexican immigrants and non-Hispanic whites changed by a small 0.8 percentage-point for those along the U.S.Mexico border, but it widened by 4.3 percentage point for workers in other U.S. cities. The B-effect predicts that the returns to observable skills in the 1990s should have increased these earnings 6 For further details on these effects, see Juhn, Murphy, and Pierce (1993) and Blau and Kahn (1997, 1994). 10 differentials between 0.05 (border) and 0.07 (non-border) log-points. But, Mexican workers in U.S.-cities close to Mexico countered these skill-return-induced differential-increases because of their average skilllevel improvement vis-à-vis non-Hispanic whites (see the X-effect). Mexican-born workers residing in the U.S.-interior did not fare as well; their relative skill improvements were not enough to dissipate the increasing skill returns accrued by non-Hispanic whites. Concluding Remarks Our findings point to a marked difference in the earnings experience between Mexican immigrants and Mexican Americans during the 1990s along the U.S.-Mexico border relative to their interior counterparts. The conceptual issues section of our paper points to some potential explanations for this result. In general, U.S. metropolitan areas near Mexico have higher immigrant/U.S.-born-MexicanAmerican wage differentials than cities in the U.S.-interior; perhaps over the 1990s there were signs of wage convergence between these two groups as suggested by neo-classical theory. The results show, however, that such convergence did not occur between Mexican Americans (both U.S.- and foreign-born) and non-Hispanic whites. The mechanics of the convergence of Mexican immigrant and Mexican American earnings along the U.S.-Mexico border raises several conceptual possibilities. The first of these is that Mexican immigrants are more mobile than Mexican Americans. This could be, as we noted above, because Mexican immigrants who first move to the U.S.-Mexico border have a relative advantage, by virtue of having been through the migratory process before, and are more predisposed to move to areas that offer more attractive earnings. A second explanation pertains to the potential relative attractiveness for the Mexican immigrant among entrepreneurs and capital because this population has the familiarity and the skill set to more efficiently engage in trade with Mexican consumers and businesses. Outside of these explanations follow others that might be more difficult to test. One possibility is that the level of unmeasured factor endowments did not keep pace for Mexican Americans along the U.S.Mexico border relative to their interior counterparts during the 1990s, perhaps because of a declining quality of education along this region relative to the interior. Another possibility is that the Mexican 11 American in this region has not received, in a relative sense, the on-the-job training necessary to compete in labor markets along the U.S.-Mexico border. Finding which of these explanations best explains our results is important for social science and policy. For example, if the root cause of this finding depends on the relative immobility of resources, perhaps programs that foster national labor-market information as well as more information about business opportunities in the border region might help promote wage equality for Mexican Americans in the U.S. However, if the primary source of the relatively depressed wages for Mexican Americans along the U.S.-Mexico border stems from this population losing ground in terms of both the quantity and the quality of its human capital, then programs designed to remedy this region’s educational and labor force training infrastructures is warranted. 12 References Blau, Francine D., and Lawrence Kahn. 1997. “Swimming Upstream: Trends in the Gender Wage Differential in the 1980s.” Journal of Labor Economics 15(1): 1-42. Blau, Francine D., and Lawrence Kahn. 1994. “Rising Wage Inequality and the US Gender Gap.” American Economic Review, AEA Papers and Procedings 84(2): 23-28. Dávila, Alberto, and Marie T. Mora. 2000. “English Skills, Earnings, and the Occupational Sorting of Mexican Americans Working along the US-Mexico Border.” International Migration Review 34:133-57. Dávila, Alberto, and J. Peter Mattila, 1985. “Do Workers Earn Less along the U.S.-Mexico Border?” Social Science Quarterly 66:310-18. Dávila, Alberto, José A. Pagán, and Gökçe Soydemir. 2002. “The Short-Term and Long-Term Deterrence Effects of INS Border and Interior Enforcement on Undocumented Immigration.” Journal of Economic Behavior and Organization 49(4): 459-72. Dávila, Alberto, Ronald H. Schmidt, and Gary M. Ziegler. 1984. “Industrial Diversification, Exchange Rate Shocks, and the Texas-Mexico Border.” Federal Reserve Bank of Dallas Economic Review, May 1984, pp. 1-9. Dávila, Alberto. 1982. “Sources of Depressed Earnings Along the Texas-Mexico Border.” Federal Reserve Bank of Dallas Economic Review, November 1982. Flota, Chrystell, and Marie T. Mora. 2001. “The Earnings of Self-Employed Mexican Americans along the US-Mexico Border.” The Annals of Regional Science 35:483-99. Fullerton, Thomas M. 2001. “Educational Attainment and Border Income Performance.” Federal Reserve Bank of Dallas Economic and Financial Review, Third Quarter, pp. 2-10. Hansen, Niles. 1982. “Location Preferences and Opportunity Cost: A South Texas Perspective.” Social Science Quarterly 63:506-16. Juhn, Chinhui, Kevin M. Murphy, and Brooks Pierce. 1993. “Wage Inequality and the Rise in the Returns to Skill.” Journal of Political Economy 101(3): 410-42. McManus, Walter S., Gould, William, and Finis Welch. 1983. “Earnings of Hispanic Men: The Role of English Language Proficiency.” Journal of Labor Economics 1(2):101-130. Mora, Marie T., and Alberto Dávila. 2006a. “Mexican Immigrant Self-Employment along the U.S. Mexico Border: An Analysis of 2000 Census Data.” Social Science Quarterly 87(1): 91-109. Mora, Marie T., and Alberto Dávila. 2006b. “A Note on the Changes in the Relative Wages of LEP Hispanic Men between 1980 and 2000.” Industrial Relations 45(2): 169-72. Mora, Marie T., and Alberto Dávila. 2006c. “Hispanic Ethnicity, Gender, and the Change in the LEPEarnings Penalty in the US during the 1990s.” Social Science Quarterly (Special Issue on Ethnicity and Social Change), forthcoming December 2006. 13 Mora, Marie T. “Self-Employed Mexican Immigrants Residing along the U.S.-Mexico Border: The Earnings Effect of Working in the U.S. versus Mexico.” International Migration Review, forthcoming. Mora, Marie T. 2005. “Changes in Occupational Earnings along the US-Mexico Border between 1900 and 1920.” Journal of Economic Issues 39(4): 1043-59 Peach, James T., and Richard V. Adkisson. 2000. “NAFTA and Economic Activity along the U.S.Mexico Border.” Journal of Economic Issues 34: 481-89. Peach, James T., 1997. “Income Distribution along the United States Border with Mexico.” Journal of Borderlands Studies 12:1-15. Robles, Barbara J. 2002. “Latina Microenterprise and the US-Mexico Border Economy.” The Estey Centre Journal of International Law and Trade Policy 3(2): 307-27. Ruggles, Steven, and Sobek, Matthew, et al. 2003. Integrated Public Use Microdata Series, Version 3.0, University of Minnesota, Minneapolis: Historical Census Projects; http://www.ipums.org . Sharp, John. 1998. Bordering the Future: Challenge and Opportunity in the Texas Border Region, Office of the Texas Comptroller of Public Accounts. Smith, B., and R. Newman. 1977. “Depressed Wages along the US-Mexico Border.” Economic Inquiry 15: 51-66. Welch, Finis. 2000. “Growth in Women’s Relative Wages and in Inequality Among Men: One Phenomenon or Two?” American Economic Review: AEA Papers and Proceedings 90(2): 444-49. 14 Table 1: Average Characteristics of Mexican Immigrants, Mexican Americans, and Non-Hispanic Whites along the U.S.-Mexico Border and in the U.S.-Interior in 1990 and 2000 U.S.-Born Mexican Americans Mexican Immigrants Characteristic 1990 2000 U.S.-Born Non-Hispanic Whites 1990 2000 1990 2000 Residents of U.S. MSA along Mexican border: Natural log. hourly Earnings 1.810 (0.588) 2.154 (0.591) 2.097 (0.583) 2.414 (0.576) 2.427 (0.611) 2.779 (0.633) Education 8.922 (4.581) 9.926 (4.200) 12.424 (2.898) 12.856 (2.591) 13.981 (2.288) 14.285 (2.312) Experience 24.497 (11.864) 25.045 (11.156) 20.231 (10.799) 20.282 (10.019) 20.157 (10.296) 22.440 (10.132) Years in U.S. 17.934 (10.657) 19.460 (10.696) ----- ----- ----- ----- LEP 0.406 0.394 0.034 0.027 ----- ----- Female 0.385 0.389 0.442 0.460 0.428 0.441 N (unweighted) 7,522 12,336 10,814 12,593 36,427 35,186 159,206 264,569 226,017 279,039 763,778 760,533 N (weighted) Residents of U.S. Interior MSA: Natural log. hourly Earnings 1.955 (0.583) 2.257 (0.567) 2.242 (0.600) 2.570 (0.580) 2.454 (0.603) 2.798 (0.624) Education 8.321 (4.458) 8.990 (4.283) 12.261 (2.738) 12.701 (2.557) 13.669 (2.450) 14.001 (2.364) Experience 22.737 (10.786) 23.113 (10.556) 20.553 (10.783) 20.616 (10.166) 21.308 (10.793) 22.933 (10.342) Years in U.S. 15.114 (9.230) 16.228 (9.848) ----- ----- ----- ----- LEP 0.457 0.463 0.028 0.023 ----- ----- Female 0.297 0.293 0.442 0.462 0.441 0.459 N (unweighted) 10,925 23,629 12,507 15,812 404,439 427,569 1,183,871 2,527,710 1,360,004 1,746,295 42,795,422 46,200,810 N (weighted) Notes: The parentheses contain the standard deviations for the continuous variables. These statistics use the IPUMS-provided sampling weights to maintain the national representation of the sample. The samples include workers employed for at least 20 hours per week for a minimum of 32 weeks in the previous year, and who resided in a metropolitan statistical area. 15 Table 2: Earnings Regression Results for Mexican Immigrants, U.S.-Born Mexican Americans, and U.S.Born Non-Hispanic Whites along the U.S.-Mexico Border and in the U.S.-Interior (Dependent Variable = Natural Logarithm of Hourly Earnings) Mexican Immigrants Mexican Immigrants & U.S.-Born Mexican & U.S.-Born NonMexican Characteristic Americans Hispanic Whites Immigrants Only U.S.-Mexico Border MSA -0.186*** (0.009) -0.148*** (0.008) -0.049*** (0.003) Border x 2000IPUMS 0.035*** (0.012) -0.012 (0.011) 0.0004 (0.005) Mexican immigrant along border ----- -0.050*** (0.013) -0.162*** (0.011) Mex. imm. along border x 2000IPUMS ----- 0.043*** (0.016) 0.002 (0.013) Mexican immigrant ----- -0.294*** (0.015) -0.249*** (0.014) Mex. imm. x 2000IPUMS ----- -0.018 (0.018) -0.026 (0.017) 0.026*** (0.002) 0.047*** (0.001) 0.093*** (0.0004) Education x 2000IPUMS -0.001 (0.002) -0.005*** (0.002) 0.006*** (0.001) Experience 0.016*** (0.002) 0.023*** (0.001) 0.030*** (0.0004) Experience x 2000IPUMS -0.007*** (0.003) -0.007*** (0.002) 0.001* (0.001) Experience2/100 -0.026*** (0.003) -0.031*** (0.003) -0.042*** (0.001) Experience2/100 x 2000IPUMS 0.012*** (0.004) 0.011*** (0.003) 0.003* (0.001) LEP -0.137*** (0.011) -0.104*** (0.010) 0.024** (0.012) 0.005 (0.013) -0.007 (0.013) 0.020 (0.014) Years in U.S. 0.015*** (0.001) 0.012*** (0.001) 0.011*** (0.001) Years in U.S. x 2000IPUMS -0.003*** (0.001) -0.003*** (0.001) -0.001* (0.001) Female -0.269*** (0.011) -0.254*** (0.007) -0.337*** (0.002) Female x 2000IPUMS 0.028** (0.013) 0.025*** (0.009) 0.051*** (0.002) 2000 IPUMS 0.406*** (0.040) 0.440*** (0.031) 0.191*** (0.010) Education LEP x 2000IPUMS 16 Table 2—Continued. Characteristic Mexican Immigrants Only Mexican Immigrants & U.S.-Born Mexican Americans Mexican Immigrants & U.S.-Born NonHispanic Whites 1.462 (0.033) 1.465*** (0.025) 0.926*** (0.007) Constant R2 .208 .237 .297 N 54,412 106,138 957,033 ***, **, * Statistically significant at the one, five, or ten percent level. Notes: The parentheses contain robust standard errors. These regressions use the IPUMS-provided sampling weights to maintain the national representation of the sample. The samples include Mexican immigrants, U.S.-born Mexican Americans, and U.S.born monolingual-English non-Hispanic white workers employed for at least 20 hours per week for a minimum of 32 weeks in the previous year, and who resided in a metropolitan statistical area. 17 Table 3: Selected Earnings Regression Results for Specific Border MSAs (Dependent Variable = Natural Logarithm of Hourly Earnings) Mexican Immigrants Mexican Immigrants & U.S.-Born & U.S.-Born NonMexican Mexican Americans Hispanic Whites Immigrants Only Characteristic El Paso, TX and Las Cruces, NM -0.283*** (0.014) -0.157*** (0.011) -0.190*** (0.010) South Texas border MSA -0.332*** (0.019) -0.200*** (0.013) -0.198*** (0.022) Arizona border MSA -0.191*** (0.023) -0.072*** (0.016) -0.184*** (0.007) California border MSA (San Diego) -0.023*** (0.012) 0.103*** (0.013) 0.017*** (0.004) El Paso, Las Cruces x 2000IPUMS 0.081*** (0.018) 0.009 (0.015) 0.007 (0.016) South Texas border x 2000IPUMS 0.056** (0.022) 0.020 (0.016) 0.022 (0.028) Arizona border x 2000IPUMS 0.067** (0.028) -0.009 (0.021) 0.017* (0.010) California border x 2000IPUMS -0.011 (0.015) -0.031* (0.013) 0.002 (0.005) Mexican immigrant x El Paso, Las Cruces ----- -0.165*** (0.017) -0.146*** (0.018) Mexican immigrant x South Texas border ----- -0.148*** (0.022) -0.116*** (0.030) Mexican immigrant x Arizona border ----- -0.174*** (0.029) -0.091*** (0.029) Mexican immigrant x California border ----- -0.194*** (0.017) -0.136*** (0.013) Mex. imm. x El Paso, Las Cruces x 2000IPUMS ----- 0.055** (0.023) 0.026 (0.026) Mex. imm. x S. Texas border x 2000IPUMS ----- 0.011 (0.027) -0.038 (0.037) Mex. imm. x Arizona border x 2000IPUMS ----- 0.065* (0.035) 0.001 (0.033) Mex. imm. x California border x 2000IPUMS ----- 0.038* (0.023) -0.010 (0.017) ***, **, * Statistically significant at the one, five, or ten percent level. Notes: The parentheses contain robust standard errors. These regressions use the IPUMS-provided sampling weights to maintain the national representation of the sample. Other variables in these regressions include those listed in Table 2; the results for these variables can be obtained from the authors. F-tests for the interactions between Mexican immigrants, the four border regions, and 2000IPUMS indicate that, as a group, these four terms are statistically significant at the five percent level when combining Mexican immigrants and U.S.-born Mexican Americans, but are not statistically significant at conventional levels when combining immigrants and U.S.-born non-Hispanic whites. 18 Table 4: Juhn-Murphy-Pierce Decomposition of Changes in the Mexican Immigrant/Non-Hispanic White Wage Differential along the U.S.-Mexico Border and in the U.S.-Interior between 1990 and 2000 Change in the Wage Differential between Mexican Immigrants & U.S.-Born Mexican Americans Change in the Wage Differential between Mexican Immigrants & U.S.-Born Non-Hispanic Whites U.S.-Mexico Border MSAs U.S. Interior MSAs U.S.-Mexico Border MSAs U.S. Interior MSAs ∆W2000 ≡ ln(W)N2000 - ln(W)M2000 0.260 0.313 0.625 0.541 ∆W1990 ≡ ln(W)N2000 - ln(W)M2000 0.287 0.287 0.617 0.498 ∆W2000 – ∆W1990 -0.027* 0.026* 0.008 0.043* X-effect -0.068 -0.026 -0.059 -0.027 B-effect 0.064 0.065 0.053 0.073 θ-effect -0.024 -0.011 0.009 -0.005 σ-effect -0.001 -0.001 0.005 0.002 Effects * The change in the immigrant/native wage differential is statistically significant at the five-percent level based on t-tests while employing the IPUMS-provided statistical weights. Notes: Some of the total effects might not add up to the change in the wage differential because of rounding. The control variables in the native earnings functions for each year (results available from the authors) include education, potential experience, experience2, gender, limited-English-proficiency (for the Mexican American sample), and a constant term 19
© Copyright 2026 Paperzz