https://dh.scu.edu/exhibits/files/original/90/2941/30032389.pdf 8f58cbe5cb77750a0b261b2358124e31 PDF Text Text Illegal Migration from Mexico to the United States Author(s): Gordon H. Hanson Source: Journal of Economic Literature , Dec., 2006, Vol. 44, No. 4 (Dec., 2006), pp. 869924 Published by: American Economic Association Stable URL: https://www.jstor.org/stable/30032389 JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms is collaborating with JSTOR to digitize, preserve and extend access to Journal of Economic Literature This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Journal of Economic Literature Vol. XLIV (December 2006), pp. 869-924 Illegal Migration from Mexico to the United States GORDON H. HANSON* In this paper, I selectively review recent literature on illegal migration from Mexico to the United States. I begin by discussing methods for estimating stocks and flows of illegal migrants. While there is uncertainty about the size of the unauthorized population, new data sources make it possible to examine the composition of legal and illegal populations and the time-series covariates of illegal labor flows. I then consider the supply of and demand for illegal migrants. Wage differentials between the United States and Mexico are hardly a new phenomenon, yet illegal migration from Mexico did not reach high levels until recently. An increase in the relative size of Mexico's working-age population, greater volatility in U.S.-Mexico relative wages, and changes in U.S. immigration policies are all candidate explanations for increasing labor flows from Mexico. Finally, I consider policies that regulate the cross-border flow of illegal migrants. While U.S. laws mandate that authorities prevent illegal entry and punish firms that hire unauthorized immigrants, these laws are imperfectly enforced. Lax enforcement may reflect political pressure by employers and other interests that favor open borders. 1. Introduction percent of the U.S. foreign-born population There is increasing interest by academ- ics and policymakers in Mexican and equivalent to 10 percent of the total population of Mexico (see figure 1).1 The United States has not experienced a concen- migration to the United States. Mexico is trated immigration wave of this magnitude the most important source country for since the influxes from Germany and Ireland U.S. immigration, accounting for 34 perin the mid-nineteenth century.2 For Mexcent of total immigrant arrivals since 1990. In 2004, the 10.5 million Mexican immi- ico, the continuing population outflow is grants living in the United States were 31 * University of California, San Diego, and National Bureau of Economic Research. I thank Roger Gordon and seminar participants at the 2006 AEA meetings, Boston University, MIT, Syracuse University, and Tufts University for helpful comments and Jeff Lin and Maribel Pichardo for excellent research assistance. 1 Figure 1 reports the total number of Mexican immigrants in the United States (legal and illegal) as a share of Mexico's population, the total U.S. population, and the foreign-born U.S. population. 2 Over the period 1841 to 1860, Ireland accounted for 39 percent of U.S. immigration and over the period 1841 to 1890, Germany accounted for 30 percent of U.S. immigration (U.S. Department of Homeland Security 2004). 869 This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �870 Journal of Economic Literature, Vol. XLIV (December 2006) Figure 1. Mexican Immigrants in the U.S. unprecedented.3 In both countries, the cross-border flow of labor appears to have affected the structure of wages, the intrana- tional distribution of population, and the pattern of industrial specialization. Beyond its scale, the distinguishing feature of Mexican immigration is that most new arrivals enter the United States illegally. In 2004, there were an estimated 5.9 mil- lion unauthorized Mexican immigrants in the United States, among a total unauthorized population of 10.3 million (Jeffrey S. immigration in the United States is a rela- tively new phenomenon. It has provoked political debate about whether to provide public services to illegal immigrants, grant them status as legal residents, or militarize U.S. borders to prevent further illegal inflows. In Mexico, migration abroad has helped ease the country's adjustment to rapid growth in its working-age population and to macroeconomic shocks, though not without disrupting the families and communities whose members have moved to the Passel 2005). Thus, 56 percent of Mexican United States. immigrants appear to lack permission to be in the country, compared to 17 percent of There is an emerging body of economic research on illegal migration from Mexico to the United States. This literature has been all other immigrants. Large-scale illegal 3 The one episode that approaches the current outflow is the Mexican Revolution (1911-20). Between 1911 and 1925, 680,000 legal immigrants from Mexico (or 5 percent of Mexico's 1910 population) entered the United States (and were joined by a large number of illegal immigrants). By the late 1920s, many of these individuals had returned to Mexico. See Lawrence Cardoso (1980) and Fernando Alanis Enciso (1999). made possible by the recent availability of data sources on the cross-border move- ments of legal and illegal Mexican migrants. Also prompting attention is the realization that, with unauthorized entrants accounting for over half of Mexican immigrants and over three tenths of all U.S. immigrants, any This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 871 discussion of international migration in the United States or Mexico ends up confronting the issue of illegality either explicitly or implicitly. Much of the initial research on unautho- how prospective migrants respond to these policies. There is a large literature on U.S. immigration, which tends to focus on the labormarket consequences of immigrant inflows rized labor flows was done by nonecono- and the economic performance of immi- mists.4 The principal themes of this body of work resemble those in the economics liter- grants.6 This body of work examines, among ature on internal migration in developing other questions, whether immigration reduces wages for U.S. native workers; Hillel Rapoport and Fr6d6ric Docquier whether immigrants make relatively greater use of means-tested entitlement programs; countries (see Robert E. B. Lucas 1997; forthcoming). Early waves of illegal migration from Mexico appear to have originated in rural areas of the country (Wayne A. Cornelius 1992; Jorge Durand, Massey, and and whether earnings, education, fertility, or other outcomes for immigrants converge to native levels over time. Largely taken for granted in the U.S. liter- Rene Zenteno 2001), involved households financing the migration of one or more ature is why foreign residents migrate to the United States. One obvious reason is that members in return for remittances from the U.S. real wages far exceed those in many migrants (Durand, Emilio A. Parrado, and other countries. Large wage differentials, coupled with binding and slowly changing depended on family and community networks that helped migrants enter and find employment in the United States (Massey et quotas on U.S. legal immigration, create queues to enter the United States. Given extended delays in clearing such queues, al. 1994; Massey and Kristin E. Espinosa 1997). gration appears to be more or less insensitive Massey 1996; Durand et al. 1996), and Yet, internal migration and illegal interna- annual variation in the level of legal immi- to contemporaneous annual fluctuations in tional migration differ in important respects. U.S. or foreign econ-omies. Perhaps as a While policy barriers that restrict withincountry regional labor flows are rare, countries actively regulate the inflow of labor result, the quantity of literature on the con- from abroad. The United States determines sequences of U.S. legal immigration vastly exceeds that on its causes.7 With illegal immigration, the determi- the level of legal immigration through quo- nants of migrant flows and the hightas on entry visas, which change infrequent- frequency variation of these flows have ly over time.5 The country implicitly sets the level of illegal immigration through selecting the intensity with which it enforces bor- 6 For surveys of this literature, see James P. Smith and Barry Edmonston (1998) and George J. Borjas (1999a, ders against illegal entry. Key issues for the 1999b) and for recent work in the area see Borjas (2003) and David Card (2005). For analysis of these issues in the study of illegal migration are how countries context of Mexican immigration, see Smith (2003), choose their border regulation policies and Stephen J. Trejo (2003), Francise D. Blau and Lawrence M. Kahn (forthcoming), Borjas and Lawrence F. Katz (forthcoming), Card and Ethan Lewis (forthcoming), and 4 For reviews of this literature, see Douglas S. Massey Brian Duncan and Trejo (forthcoming). There is a smaller et al. (1994) and Thomas J. Espenshade (1995). literature on the consequences of emigration for Mexico. 5 The current U.S. immigration quota system was See Prachi Mishra (forthcoming) and Gordon H. Hanson established by the Hart-Cellar Act of 1965 and amended (forthcoming). by the Immigration Act of 1990. The 1990 law set a flexi7 Exceptions include Timothy J. Hatton and Jeffrey G. ble cap for U.S. legal admissions at 675,000 individuals of Williamson (2004) and Anna Maria Mayda (2005), who which 480,000 are to be family-based, 140,000 are to be examine the correlates of international migration flows. In employment-based, and 55,000 are to be "diversity immi- research on internal migration, there is considerable work grants." Immediate relatives of U.S. citizens are not subject on the incentive to migrate. See Michael J. Greenwood (1997) and Lucas (1997) for reviews of the literature. to immigration quotas. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �872 Journal of Economic Literature, Vol. XLIV (December 2006) attracted more attention. Geographic proximity allows unauthorized migrants from Mexico to move to the United States than tourism or foreign direct investment (Inter-American Development Bank 2004). In this paper, I selectively review recent relatively quickly. The existence ofliterature well- on illegal migration from Mexico to the United States. In section 2, I discuss established migration networks enables U.S. employers to communicate changes in methods their for estimating stocks and flows of demand for labor to prospective migrants illegalinmigrants. While there is uncertainty Mexico. Migrants use these same networks about the size of the unauthorized populato find jobs and housing in the United tion, new data sources make it possible to States. Shocks to either the Mexican or U.S. examine the composition of legal and illegal economies may be transmitted into changes populations and the time-series covariates of in cross-border labor flows with relatively illegal labor flows. In section 3, I consider the short time lags, making illegal migration supply of and demand for illegal migrants. potentially quite responsive to changes in Wage differentials between the United States and Mexico are hardly a new phenomenon, yet illegal migration from Mexico did not binational business-cycle conditions. Another feature that distinguishes legal and illegal migrant flows is their composition. While legal migrants face entry costs reach high levels until recently. An increase in the relative size of Mexico's working-age associated with queues in obtaining visas, population, greater volatility in U.S.-Mexico relative wages, and changes in U.S. immigraillegal migrants face costs associated with evading immigration authorities. Once in the tion policies are all candidate explanations for receiving country, the risk of detection may increasing labor flows from Mexico. In secmake some employers unwilling to hire ille- tion 4, I consider policies that regulate the gal migrants, limiting their occupational cross-border flow of illegal migrants. While prospects and reducing the returns to skill U.S. laws mandate that authorities prevent they perceive. Variation in migration costs illegal entry and punish firms that hire unauand in receiving-country wage profiles thorized immigrants, these laws are imperbetween legal and illegal migration suggest fectly enforced. Lax enforcement may reflect the characteristics of illegal migrants may political pressure by employers and other differ from those of legal migrants. interests that favor open borders. In section Observed changes in the composition of 5, I discuss directions for further research. U.S. immigrants, which has received much My goal in this paper is not to conduct an attention in research on consequences of exhaustive survey of work on illegal migraU.S. immigration, could be partly a by- tion but rather to highlight major findings in product of the rising share of unauthorized the literature, assess the state of important entrants in immigrant inflows. For Mexico, debates, and identify unresolved issues, with an eye toward advancing questions to help the composition of migrant outflows matters not just for the labor-market effects of emi- guide future work. As much of the literature gration but also for the ties that migrants retain with the country. Illegal migrants appear to be more likely than legal migrants to send remittances to family members in is empirical, I will focus on this strand of research, with occasional forays into theory. One topic I will not discuss at much length is the economic consequences of illegal migra- Mexico. Rising illegal migration from tion, in part because literature on the conse- Mexico may be partly responsible for the recent surge in remittances in the country, which rose from 0.1 percent of GDP in 1990 to 2.2 percent of GDP in 2004 and now generate more foreign exchange in the country quences of immigration has been subject to several recent surveys (see note 6) and in part because there is little research on the specific aspects of these consequences that are attributable to illegal immigration. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 873 Though my focus is on the United States and Mexico, insights from the literature are relevant for other regions, as well. Unauthorized migration has become a glob- Knowing the overall share of immigrants who are unauthorized is not the same as knowing which specific immigrants in a given data source are unauthorized. al phenomenon. In the last two decades, there have been sizable flows of illegal migrants from North Africa and Eastern Fortunately, there are now several microlevel surveys that provide information on Europe to Western Europe, from Indonesia design or default, migrants from Mexico to elsewhere in Southeast Asia, and from neighboring countries to South Africa. The U.S.-Mexico experience may be instructive for these and other cases regarding how to measure unauthorized migration, estimate the causes and consequences of migration flows, and gauge the potential impacts of policy interventions. individual migration status. Either by account for a large fraction of those represented in these data sources. While each survey has limitations, their use in conjunc- tion with large public data sets from Mexico and the United States provides considerable detail on the population of legal and illegal migrants from Mexico who are living or have lived in the United States. 2. Stocks and Flows of Illegal Migrants Illegal immigrants account for a large and growing fraction of the U.S. foreign-born population.8 One may imagine that, as part of the underground economy, unauthorized migrants are not easily subject to measurement. However, there is now abundant evidence that illegal immigrants are represented in official household surveys, including the U.S. Census of Population and Housing and the U.S. Current Population Survey. Given known levels of U.S. legal immigration, the number of foreign-born individuals enumerated in these sources is far too large for them all to be legal. The most common method to estimate A third data source on unauthorized migration is a by-product of U.S. immigration policy. To prevent illegal immigration, the U.S. Border Patrol polices U.S. borders and ports of entry, attempting to apprehend those seeking to enter the country illegally. The Border Patrol compiles high-frequency data on apprehensions and enforcement, the vast majority of which occur along the U.S.-Mexico border. Data on border apprehensions and enforcement allow one to examine how attempts at illegal entry vary with economic conditions in the United States and Mexico and to see which factors are associated with the intensity of U.S. enforcement activities. the number of illegal immigrants is to take 2.1 The Residual Foreign-Born Population the difference between the measured immi- grant population and the sum of past legal immigrant inflows. Estimates using this residual approach suggest that stocks of illegal immigrants have risen sharply over time. However, there is considerable variability in In the United States, there are two classes of legal immigrants who appear in official data sources (i.e., are surveyed by the U.S. Census Bureau, the U.S. Bureau of Labor Statistics (BLS), or other official entities). the estimates, associated with differing One is permanent legal immigrants, who assumptions about the magnitude of errors in enumerating legal and illegal immigrants in official data sources. 8 This also appears to be the case in Western Europe (Tito Boeri, Barry McCormick, and Hanson 2002). have the right to reside in the country indef- initely. Another is temporary legal immigrants, who have the right to reside in the country for a defined time period, as specified by an entry visa (such as those for stu- dents, specific skill classes of workers, diplomats, and family members of temporary This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �874 Journal of Economic Literature, Vol. XLIV (December 2006) legal immigrants).9 Since government surveys do not screen individuals based on their immigration status, illegal immigrants also appear in official data, to the extent they make themselves available to be surveyed. The standard method to estimate the num- ber of illegal immigrants is to assume it is equal to the residualforeign-born population, which is given by (1) Ut,= mortality and emigration rates to each entry cohort and determine the fraction of tempo- rary immigrants admitted in previous years that are still in the country. In practice, most discrepancies evident in table 1 appear unrelated to differences in these assumptions. Of greater importance are assumptions about measurement error in F,. To simplify matters, I reexpress equation (1) in contemporaneous values as Ft = Lt + Ut, where Lt is the total legal foreign-born pop- ulation in year t. The U.S. Census Bureau (and other entities that conduct household where Ut is the unauthorized foreign-born population in year t, Ft is the total foreignsurveys) tends to undercount the total population (with undercount rates for low-income permanent legal immigrants that entered in year s < t, m, is the mortality rate between born population in year t, Ls is the number of year s and year t for legal immigrants enter- ing in year s, es is the emigration rate between year s and year t for legal immigrants entering in year s, and T, is temporary legal immigrants present in year t. Table 1 reports estimates of the unauthorized foreign-born population by the U.S. Bureau of the Census (Joe Costanzo et al. 2001), Bean et al. (2001a, 2001b), Passel (2005), and the U.S. Immigration and Naturalization Service (INS) (2001).10 In the most recent projection, Passel (2005) esti- mates that between 2002 and 2004 the ille- gal immigrant population rose from 9.3 to 10.3 million, for an average annual net inflow of 500,000 migrants, with 57 percent of these individuals coming from Mexico. This compares to an average annual net illegal inflow during the 1990-2000 period of households, which would include many Mexican immigrants, believed to be relatively high), in which case the measured foreign- born population, F,, is less than the actual foreign-born population, Ft. The total legal- immigrant population, Lt, in contrast, appears to be measured with greater accura- cy, since immigration authorities have records on how many entry visas they award. Suppose the legal-immigrant population that is enumerated in the census is, (2) Lt where A, is the fr that go uncounted immigrant census (3) is, popul U, where p, is the 581,000 migrants (with 58 percent of net immigrants that new immigrants coming from Mexico), are mean zero iid based on U.S. Census Bureau estimates observed individ (Costanzo et al. 2001), and 350,000 migrants terms of their Census Bureau estimates that it under(with 79 percent coming from Mexico), based on INS (2001) estimates. counts the Hispanic population by 5 perObviously, there are a host of assumptions cent, Bean et al. (2001b) put the undercount involved in estimating (1). One must assign rate for legal Mexican immigrants at 2 percent to 4 percent. Similarly, while the U.S. 9 The Census Bureau does not survey individuals on Census Bureau estimates that it under- tourist or business visas or other short-term visitors. counts 10 See Bean et al. (1998) for a survey of the literature on estimating illegal immigrant populations. the illegal immigrant population by 15 percent, the INS assumes an undercount This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 875 TABLE 1 ESTIMATES OF THE U.S. ILLEGAL-IMMIGRANT POPULATION, 1990-2004 Total Population of Illegal Immigrants in the United States (millions) Costanzo et al. (2003) Bean et al. (2001a, 2001b) INS Undercount Rate Undercount Rate Passel Year (2001) 10% 15% 20% 15% Median 25% (2005) 1990 3.500 1991 3.766 4.025 4.430 -- -- -- 4.707 -- -- -- -- -- -- -- -- -- -- -- -- -- 1992 4,204 1993 4.492 -- -- -- -- -- -- -- -- -- 1994 4.750 -- -- -- -- -- -- -- 1995 5.146 -- -- -- -- -- -- -- 1996 5.581 -- -- -- -- -- -- -- 1997 5.862 -- -- -- -- -- -- -- 1998 6.098 -- -- -- -- -- -- -- 1999 6.488 -- -- -- -- -- -- -- 2000 7.000 2001 -- 2002 -- -- 2003 2004 8.705 -- -- 10.242 -- -- -- - -- -- 5.918 -- -- -- 10.882 7.751 -- -- -- --- -- 1996 2.040 -- 2000 2001 -- -- 4.808 -- 2002 2003 2004 -- --- -- 1.008 --- -- -- 1.524 3.872 --- -- -- 3.462 -- -- -- -- -- -- -- -- 9.300 -- -- 10.300 Illegal -- -- -- 2.543 3.706 -- -- -- -- 4.510 -- -- -- 5.765 --- -- -- -- Immig -- -- -- -- -- -- of -- -- 9.864 -- -- Population 1990 -- -- 5.300 -- -- 5.900 Notes: Costanzo et al. (2003) rep undercount rates for illegal imm undercount rate for illegal imm mates are based on undercount r grants or (b) 25 percent for ille estimates are based on undercou legal immigrants (2 percent for for all legal immigrants (4 perc 2001b) are for the full set of rep of 10 percent, and Bean representativ (see Bean et al. 1998). The U.S. Census undercount rates of 15 p cent. It is primarily the Bureau justifies its undercount assumptions assumptions based that account on results from its own postenumeraestimates in tion table 1. surveys, from which standard errors for fo Assumptions under populationabout estimates are derived (Costanzo based on comparisons et al. 2001). The INS (2001) justifies a 10 of population with the percent undercount rate based popul on a case postenumeration surveys study of Mexican immigrants in Los Angeles cific localities. requi County. Bean This et al. (2001a, 2001b) justify that underenumeration their range of undercount rates based on in This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �876 Journal of Economic Literature, Vol. XLIV (December 2006) results in Bean and Jennifer Van Hook (1998). Given observed values of F, and L, and percent Mexico. to gauge of th Missing how th assumed values for k, or ,t, which I indicate tions by a tilde, the estimated value of the illegal- time. The poste conducted by o immigrant population in year t can be about Bureau approximated as or BL undercount (4) 1-ftI u small number of communities around the as time of the main survey. Since samples for If true undercount rates fall over time and postenumeration surveys differ across survey we fail to account for this, estimatesperiods, of the there is little basis for making timeillegal-migration population will be series biased comparisons in undercount rates. Until public surveys ask about individupwards (U, > U).11 Consider values large-scale for F, and L, in 1996 and 2001 in Bean al. ualet migration status directly, estimating the (2001a, 2001b). If we change the assumption size of the illegal-immigrant population will remain a speculative enterprise. for the 1996-2001 period from a constant illegal undercount rate of 25 percent (which 2.2 Churning in Legal and Illegal is at the upper end of the Bean et al. rates) Immigrant Populations to a reduction in the illegal undercount rate of the stock of illegal immifrom 25 percent to 15 percent (to the Estimates lower end of the Bean et al. rates),12 the estimated grants give little indication of how long unauthorized migrants are likely to remain annual illegal net inflow from Mexico would without fall by 112,000 migrants (from 432,000 to a legal resident visa. Each year, appears to be a large flow of individu320,000), which is 30 percent of the there average from the pool of illegal migrants to the annual estimated illegal inflow over thealsperiod. Since different authors tend to use difpool of permanent legal immigrants. Many ferent postenumeration surveys as the basis immigrants who obtain visas for legal permanent residence (green cards) are at the time for selecting undercount rates, there is little consensus in the literature about what has they obtain their visas residing in the United happened to the undercount of illegal immi- States illegally. Figure 2 shows the number grants over time, other than it exhibiting a of Mexican immigrants awarded legal permanent residence and the fraction of these downward trend. In table 1, the span of estimates for the individuals who are adjusting status. Over illegal-immigrant population in the United the period 1992-2002, status adjusters States is wide. Between 1990 and 2004, the accounted for 56 percent of new legal permanent immigrants from Mexico. Some of estimated average annual net inflow of unauthose adjusting their visa status are tempothorized immigrants ranges from 350,000 to 580,000 individuals, with 55 percent to 80 rary legal immigrants who have succeeded in obtaining permanent entry visas. However, 11 Ignoring interactions in undercount rates, for Mexican immigrants, the majority of those adjusting status to permanent legal residence appear to have been living in the U,-t When the assumed undercount rate United States as illegalwill immigrantsbe (U.S. n is low, this expression upward bias in Department the of estimated Homeland Security 2004). nu grants). This effect will be exagg Further evidence of churning in the pop- undercount 12 In legal this rate for legal ulation of legal exercise, I migrants immigran and illegal immigrants is assume the isavailable held constant at 2 in the New Immigrant Survey This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 877 Figure 2. U.S. Legal Immigration from Mexico (NIS), which in its pilot form includes a ran- of Mexican immigrants who obtained dom sample of 1,134 immigrants who U.S. green cards qualified under family- received U.S. legal permanent resident visas in 1996 (Guillermina Jasso, Massey, Mark R. Rosenzweig, and Smith 2000). Based on the reunification provisions of U.S. immigration law. Since 1965, the United States has granted unrestricted legal entry to the immediate relatives of U.S. citizens and restricted legal entry, subject to annual immigration quotas, NIS data, Massey and Nolan J. Malone (2002) find that 54 percent of Mexican nationals who obtained a green card in 1996 reported having entered the United States illegally at an earlier date in time, either by crossing the U.S. border (41 percent) or overstaying a temporary entry visa (13 percent). Overall, 21 percent of U.S. green-card recipients in 1996 reported having crossed the U.S. border illegally and 11 percent reported having overstayed a temporary to more distant relatives of U.S. citizens and relatives of U.S. permanent legal residents. Most applicants take several years or more to clear the queue for a green card.13 While the United States periodically attempts to limit the granting of green cards to those applicants who either have valid temporary 13 Though immediate relatives of U.S. citizens are not entry visa. Transitions from illegal to legal residence status indicate that many individuals queuing for U.S. green cards choose to do so as illegal immigrants, rather than waiting out the process as residents of their home coun- tries. Between 1992 and 2004, 90 percent subject to immigration quotas, to obtain a green card they still must screened by immigration authorities, a process that can take as long as two years. The screening process is more protracted for individuals who meet the qualifications for a green card but whose preference category is subject to immigration quotas (e.g., more distant relatives of U.S. citizens and relatives of U.S. legal residents) and can take five years or more (David A. Martin 2005). This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �878 Journal of Economic Literature, Vol. XLIV (December 2006) entry visas or are residing abroad, the sheer volume of applications has made this provision difficult to enforce (Jessica Vaughan 2003). Consequently, the U.S. government routinely grants green cards to individuals who currently are residing in the country illegally. Massey and Malone (2002) report that prior illegality is more common among those who receive green cards under family- some unauthorized migrants (Hanson 2005). The prospect of a future amnesty is another factor that helps diminish distinctions between legal and illegal migrants. 2.3 Composition of the Legal and Illegal Immigrant Populations A longstanding conception of Mexico-toU.S. migration is that it is driven by the based immigration provisions than under needs and rhythms of agriculture. employment-based immigration provisions. The latter category requires sponsorship by a U.S. employer and applies mostly to highly skilled individuals. Churning in the illegal immigrant population suggests some unauthorized migrants According to this view, most migrants from Mexico are from the countryside, come to the United States to work as farm laborers may view their visa status as mutable. In rural in origin, relatively uneducated, and residing in the United States on an itinerant making the decision to migrate to the United States, individuals in Mexico who have rela- tives that are U.S. legal residents may internalize the prospect of obtaining a U.S. green card in the future. They may consider being an illegal immigrant as simply an intermediate step in becoming a legal permanent resident. The possibility of transitioning from illegal to legal status may blur differences in during peak agricultural months, and return to their families in Mexico for the winter off- season. Migrants would tend to be male, basis. While there is little doubt that at one time this view of Mexican migration was accurate, the Mexican immigrant population in the United States has since become more heterogeneous and more permanent. Large-scale emigration from Mexico began in the early twentieth century. Railroad construction in the late 1800s the expectations and behavior of legal and linked interior Mexico to the U.S. border, illegal migrants. giving U.S. employers improved access to Mexican labor (Cardoso 1980). In the early A second means by which prospective migrants might expect to obtain a U.S. green card is through a future amnesty for illegal aliens. In 1986, the U.S. Immigration Reform and Control Act (IRCA) awarded permanent legal residence visas to illegal immigrants who could demonstrate either (1) continuous U.S. residence since 1982 or (2) sixty days of employment in U.S. agriculture since 1985. Over the next eight years, 1.6 million illegal immigrants received green cards under the first provision and 1.1 million illegal agricul- tural workers received green cards under the second provision, with Mexican nationals accounting for 2 million of the 2.7 million IRCA legalizations (U.S. Immigration and Naturalization Service 2001). While there is political opposition in the United States to another amnesty, there have been numerous recent legislative proposals to legalize at least 1900s, Texas farmers began to recruit laborers in Mexico. They followed the main rail line into the country, which ran southwest through agricultural states in Mexico's central and western regions. Early migrants from Mexico came primarily from nine Mexican states in this area (Durand, Massey, and Zenteno 2001).14 Migration expanded further in the 1940s, after the U.S. Congress enacted the Bracero Program (1942-64), which allowed U.S. employers to bring in workers from Mexico (and the Caribbean) to fulfill short-term labor contracts (of less than 14 These nine states are Aguascalientes, Colima, Durango, Guanajuato, Jalisco, Michoacin, Nayarit, San Luis Potosi, and Zacatecas. During the period 1944 to 1964, this group of states accounted for 55 percent of migration from Mexico to the United States (Durand, Massey, and Zenteno 2001). This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 879 a year in length). At the end of their contracts, workers were required to return to their home countries. The vast majority of Perhaps as a result of migration networks, current generations of Mexican immigrants in the United States tend to live near indi- braceros worked on U.S. farms (Kitty viduals from their home regions in Mexico. Calavita 1992). At its peak, from 1954 to For instance, Kaivan Munshi (2003) finds 1960, 300,000 to 450,000 migrant workers from Mexico entered the United States that immigrants from the state of Jalisco are much more likely to live in Los Angeles or annually. The end of the Bracero ProgramSan Diego than immigrants from the state marked the beginning of large-scale illegalof Guanajuato, who prefer Chicago or Dallas. Migrants reinforce networks by creimmigration from Mexico.15 ating home-town associations that help After working in the United States, many members of their communities in Mexico braceros returned to Mexico where they assisted later generations in migratingmake the transition to living north of the abroad. They helped establish informal net-border. Of 218 home-town associations works through which earlier migrants helpformed by Mexican immigrants enumerated new migrants enter the United States, findin a 2002 survey of such organizations in housing in U.S. cities, and obtain jobs withsouthern California, 87 percent were associU.S. employers. Networks are often embed-ated with one of the nine west-central states ded in relationships involving family, kin, or that dominated migration to the United community of birth, which gives them States a under the Bracero Program (Gustavo Cano 2004). regional component. Partly as a result, there is strong historical persistence in migration While migration networks are a consistent rates to the United States across Mexican feature of cross-border labor flows, the com- regions. Figure 3 plots emigration rates in position of these flows is not. Since the 1960s, the 1950s against those in the 1990s acrossMexico has urbanized, become a more eduMexico's thirty-two states. Data for the cated nation, and incorporated women into 1950s are from Woodruff and Zenteno the labor force in greater numbers. Cornelius (2001), who calculate the fraction of each (1992) and Cornelius and Enrico A. Marcelli Mexican state's population that migrated(2001) to suggest these changes have shifted the the United States between 1955 and 1959 composition of Mexican migrants in the under the U.S. Bracero Program. Data for United States from sojourners, who follow the 1990s, taken from the 2000 Mexico the harvest season through the rural United States and then return to Mexico at the end Census of Population and Housing, report the fraction of households in a state having aof the year, to settlers, who have a permanent member migrate to the United States presence in U.S. communities. Resisting this between 1995 and 2000. The correlation notion, Durand, Massey, and Zenteno (2001) between state migration rates in the suggest instead that migration to the United 1995-2000 and the 1955-59 periods is 0.73. States remains dominated by men from traFigure 4 shows that most high-migration ditionally agricultural states in Western states are located in central Mexico, which is Mexico. While migrants have become better neither close to the United States nor homeeducated and more urban, they retain strong to Mexico's poorest households. States on ties to Mexico, returning often and tending to the U.S.-Mexico border tend to have low avoid permanent U.S. settlement. emigration rates, as do states in low-income Until recently, it would have been difficult southern Mexico. to muster much more than case-study evidence to evaluate these claims. There are 15 On illegal immigration and guest-worker programs, see Gil S. Epstein, Arye L. Hillman, and Avi Weiss (1999). now several data sources on migrants from Mexico that give details on an individual's This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �880 Journal of Economic Literature, Vol. XLIV (December 2006) Figure 3. Migration Rates to the U.S. by Mexican States, 1950s and 1990s Figure 4. Migration Rates to the U.S. by Mexican State and Distance to the U.S. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 881 TABLE 2 MIGRANT AND NONMIGRANT MEXICAN NATIONALS, MALES Mexico Migration Project, 1989-1991 In Mexico at Time of Survey 1990 1990 Currently In U.S. at U.S. Mexico All on Last Time of 1989 Census Census Respondents Migration Survey LPS Percent Male 56.2 48.0 48.7 64.7 55.9 57.2 Age 18 to 27 38.5 37.6 34.1 38.4 43.5 33.0 28 to 37 31.9 25.8 29.7 35.7 35.7 42.5 38 to 47 17.1 17.4 19.0 18.1 11.7 16.5 48 to 57 8.2 11.7 10.7 6.7 5.9 5.8 58 Males Years to 67 Aged of 4.4 18 None 7.5 to 6.4 47 9.6 9.2 5.2 1.1 3.8 3.2 2.3 2.1 4.0 Schooling 1 to 4 10.3 17.3 23.4 24.3 15.3 19.7 5 9 to to 12 8 28.0 11 to 17.2 15 30.1 32.5 24.7 31.7 41.4 18.5 11.1 34.3 19.4 12.9 9.3 39.1 23.8 18.4 19.1 18.3 16 plus 3.2 7.7 7.5 1.8 5.6 0.5 Live in Urban Area 91.9 74.8 80.6 74.5 89.2 -In Labor Force 91.0 85.2 95.8 98.6 94.1 95.2 Work in Agriculture 15.5 23.9 28.9 31.2 9.1 11.9 Has Migrated to U.S. -- -- 50.3 100.0 100.0 -Migrate U.S. Last Year -- -- 2.5 0.1 0.1 -Years in 0 U.S. to 6 11 to to 5 28.8 10 20 -- 23.4 35.7 65.6 45.2 36.9 -- 14.8 17.3 -- 15.2 28.2 13.9 20.2 36.3 19.4 6.7 20 plus 12.2 -- 4.5 9.3 6.6 59.9 Sample size, 18-67 96,487 196,729 5,370 722 375 1,670 Sample size, 18-47 83,703 158,917 4,448 666 341 1,535 Migration Project (MMP).17 The MMP is a migration status.16 Perhaps the best known and most utilized source is the Mexican household survey conducted in winter months (when seasonal migrants tend to to Mexico) in 1982 and over the 16A similar source is the National Surveyreturn of Demographic Dynamics (ENADID), conducted by the period 1987 to 1997 in several dozen rural government of Mexico in 1992 and 1997. The ENADID communities in western Mexico, chosen for asks households in Mexico whether any of its members have ever worked in or looked for work in the United States (and the year in which this occurred). In the 1997 ENADID, 9 percent of individuals report having been to the United States and 21 percent of households report having a member in the United States, up from 8 percent and 17 percent in 1992. As with the MMP, the ENADID having high rates of migration to the United States (Massey et al. 1994; Durand et al. 1996). In each community and in each year, the MMP surveyed a random sample of only includes households with at least one member remaining in Mexico. See Durand, Massey, and Zenteno 17 Recent papers using the MMP include Massey and (2001) and David McKenzie and Rapoport (2004) for work Espinosa (1997), Belinda I. Reyes (1997), Munshi (2003), using the ENADID and Cornelius and Marcelli (2001) Christina Gathmann (2004), Pia M. Orrenius and and Durand, Massey, and Zenteno (2001) for discussions of other surveys. Madeline Zavodny (2005), and McKenzie and Rapoport (2004). This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �882 Journal of Economic Literature, Vol. XLIV (December 2006) several hundred households, collecting infor- mation on past migration behavior of each household member.18 An advantage of the MMP is that it allows one to construct retro- spective migration histories on a reasonably large sample of individuals.19 Among male household heads, 23 percent report having migrated to the United States within three years of being surveyed (during the period all respondents; (b) those who report residing in the United States but who are present in Mexico at the time of the MMP survey (seasonal migrants); and (c) those in the United States at the time of the MMP sur- vey (permanent migrants), whose responses are provided by other members of their household in Mexico. 1984 to 1996). Of those reporting hav- While Mexican immigrants in the United States (census immigrants) and MMP per- ing migrated over the period 1970 to 1990, 89 manent migrants have relatively similar percent state that on their first trip to the United States they entered without documents (McKenzie and Rapoport 2004). The MMP is subject to several potential problems associated with how migrants and communities are selected into the sample. Since communities included in the MMP characteristics, they differ considerably from MMP seasonal migrants. Males account for 65 percent of MMP seasonal migrants, but only 56 percent of census immigrants and MMP permanent immigrants. And, while age profiles are similar among the three groups, educational attainment is not. Males are chosen on the basis of being rural and with nine or more years of schooling account having residents with high migration for 52 percent of census immigrants and 48 propensities, they are unlikely to be reprepercent of MMP permanent migrants, but sentative of Mexico as a whole (McKenzie only 31 percent of MMP seasonal migrants. and Rapoport 2004). Within communities,Employment patterns also differ across the households surveyed are those with at groups. Among males, 16 percent of census immigrants and 9 percent of MMP permaleast one member remaining in Mexico, thus excluding households that have migrated to nent migrants work in agriculture, compared the United States in their entirety. And, to 31 percent of MMP seasonal migrants. within households, the migrants surveyed Seasonal migrants also appear to be less estab- directly are those who have returned to lished in the United States. For males, 55 perMexico, for at least part of the year. There cent of MMP seasonal migrants have spent more than five years in the United States, compared to 71 percent of census immigrants residing in the United States. One way to evaluate the issue of sampleand 63 percent of MMP permanent migrants. selection in the MMP is to compare individFor each of these comparisons, results are are no direct observations on individuals similar for females. uals in the survey with individuals in Mexico's Census of Population and Housing and with Over time, Mexican immigrants have Mexican immigrants in the U.S. Census ofshifted out of agriculture as a main industry Population and Housing. Tables 2 and 3 proof U.S. employment. Using data from the U.S. census, Card and Lewis (forthcoming) vide summary statistics for working-age adults in the 1990 U.S. and Mexico censuses show that between 1990 and 2000, among and in the 1989, 1990, and 1991 MMP sur-recent Mexican immigrants (0-5 years in the United States), the share working in agriculveys. I consider three MMP subsamples: (a) ture fell from 23 percent to 15 percent for men and from 13 percent to 7 percent for 18 Different households are surveyed in different years, such that the MMP is a repeated cross-sections of women. Among men, construction accounthouseholds and not a true panel. ed for the largest growth in employment 19 On measurement error in retrospective data, see shares, Megan Beckett et al. (2001) and Smith and Duncan Thomas (2003). while among women retail trade showed the largest increase. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 883 TABLE 3 MIGRANT AND NONMIGRANT MEXICAN NATIONALS, FEMALES Mexico Migration Project, 1989-1991 In Mexico at Time of Survey 1990 U.S. Census 1990 Mexico Census Respondents 43.8 52.0 51.3 35.3 44.1 42.8 Percent Female Age 18 to 27 All Currently In U.S. at Migration Survey on Last Time of 1989 LPS 33.0 38.0 35.0 35.5 43.2 31.9 28 to 37 31.4 26.1 29.3 37.1 33.5 42.0 38 to 47 18.9 17.0 18.6 21.3 16.6 16.8 48 to 57 10.4 11.5 10.7 5.6 4.4 6.6 58 to 67 6.3 7.6 6.5 0.5 2.4 2.7 Years of None 8.5 12.7 6.6 3.0 5.1 4.6 Schooling 1 to 4 9.8 18.8 25.9 18.7 13.8 18.7 Females Aged 18 to 47 5 to 8 28.3 31.4 35.4 56.2 35.5 43.0 9 to 11 16.2 24.8 16.8 12.4 23.8 18.4 12 to 15 34.0 8.0 11.9 7.6 28.3 15.9 3.3 4.4 3.5 2.2 2.2 0.6 92.7 75.8 80.6 70.8 88.8 56.7 26.6 28.8 35.7 47.5 61.3 7.9 2.3 2.4 2.2 6.6 4.1 2.6 3.3 1.7 1.6 2.1 16 plus Live in Urban Area In Labor Force Work in Agriculture Children Ever Born Own Children in HH Has Migrated to US Migrated U.S. Last Year Years in U.S. 0 to 5 _-_ -__ --_ 1.8 1.0 2.2 18.6 100.0 100.0 0.8 0.0 0.0 26.2 58.1 44.3 41.5 10.7 6 to 10 20.9 17.5 17.6 16.5 11.1 11 to 20 37.3 18.7 28.9 29.2 4.7 20 plus 15.6 5.7 9.2 12.7 73.5 Sample size, 18-67 Sample size, 18-47 76,518 212,912 5,658 63,278 172,458 4,688 394 296 1,248 370 276 1,132 Note: Table 2 gives summary statistics on working-age adult residents of Mexico (1990 Mexico Census of Population and States (1990 U.S. Census of Population and Housing); respond Survey (all respondents, those residing in United States but in in United States at the time of the survey whose responses illegal immigrants from Mexico in the United States who qu Reform and Control Act (1989 Legalized Persons Survey). Th than 2,500 inhabitants. Other surveys of illegal immigrants from legal granted permanent Mexico also suggest their characteristics arethe a United States under more similar to permanent migrants (whether of IRCA (eligibility for wh in the U.S. census or the than seasonof MMP) U.S. residence from 19 al migrants. The Legalized Persons Survey LPS consisted of an initial s (LPS) covered illegal immigrants who werein 199 a follow-up survey This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �884 Journal of Economic Literature, Vol. XLIV (December 2006) qualifying for legalization.20 Among working-age adults in the LPS, shown in tables 2 and 3, 57 percent of Mexican immigrants are appears unrepresentative of Mexican immigrants in the United States. When examining results using the MMP, one should be mind- ful that they may apply only to seasonal migrants who return to Mexico with high ture. frequency and not to the general population of Mexican migrants. In sum, data sources that include illegal male, 37 percent had nine or more years of schooling, and 12 percent worked in agricul- immigrants are almost by definition subject to 2.4 Attempted Illegal Immigration sample-selection problems. Official sources, such as the population census, are likely to The majority of unauthorized immigrants undersample illegal immigrants, given theirfrom Mexico enter the United States by tendency to undercount low-income house-crossing the U.S.-Mexico border illegally.21 holds. Surveys that specifically target illegalThe U.S. government devotes vast resources immigrants, such as the MMP or the LPS, to policing U.S. borders, airports, and ports of explicitly select respondents on the basis ofentry. Between 1980 and 2004, real expendiobserved characteristics (e.g., residence in a ture on border enforcement increased by high-migration community, eligibility for an over six times and in 2005 will equal $2.2 bilamnesty). Yet, despite conflicting selectionlion (Hanson 2005). U.S. border-enforcement criteria, available data sources paint a con- activities provide a glimpse into the highsistent picture of Mexican immigrants in the frequency properties of illegal immigration. United States, suggesting they include aThe first line of defense against unauthorized high proportion of women, are overwhelm-entry is the U.S. Border Patrol. Border Patrol ingly employed outside agriculture, haveofficers on "linewatch" duty patrol the bor- high education levels (relative to nonmi-der, maintain electronic surveillance of major grants in Mexico), and have established acrossing points along the border, and staff long-term U.S. presence. traffic checkpoints along major highways this conclusion is the MMP sample of sea- Border Patrol officer hours devoted to One data source that is inconsistent with near the border. Figure 5 shows annual sonal migrants-individuals who reside in linewatch duty from 1964 to 2003. Officer the United States but return to Mexico in the winter months. Since MMP seasonal hours increased dramatically in the 1990s, rising from 2.5 million in 1994 to 9.8 million in 2001. This increase was due primarily to migrants are selected on the basis of having stepped up enforcement efforts at urban returned to Mexico, it is hardly surprising crossing points in California and Texas. that they fit the profile of itinerant migrant Concurrent with increased enforcement, laborers. Relative to other samples of legal apprehensions of those attempting illegal and illegal immigrants, MMP seasonal entry have increased, rising from 280,000 a migrants are disproportionately male, unedyear ucated, and agricultural. While much of the during the 1970s to 930,000 a year durliterature based on the MMP uses the inforing the 1990s. Individuals apprehended by mation on seasonal migrants to exam- Border Patrol officers on linewatch duty are ine Mexico-to-U.S. migration, this sample typically captured while trying to enter the United States or just after entering the country. Linewatch apprehensions are thus 20 The initial survey was of 6,193 respondents, of whom 5,691 had received amnesty by 1992 (with most of the rest awaiting decision). Of those granted amnesty, 82 percent were located for the follow-up interview. The LPS sample immigrants but more common among those from other Agricultural Worker provision of IRCA. countries, is to enter the United States on a temporary visa and then remain in the country after it expires. excluded those granted amnesty under the Special 21 A second strategy, less common among Mexican This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 885 Figure 5. Linewatch Apprehensions and Enforcement by the U.S. Border Patrol correlated with the contemporaneous level To gauge how apprehensions might be of attempted illegal immigration.22 However, related to illegal immigration, consider the apprehensions are likely to be a poor indica- level of apprehensions as a function of the average probability of apprehension tor of the actual level of illegal inflows (Espenshade 1995). Within a single month, one individual may be apprehended multiple times. Those apprehended who agree to be deported voluntarily are not processed by the U.S. justice system. For Mexican nationals, voluntary deportation often involves little more than a bus ride across the border, leav- ing them in position to attempt illegal entry again in the near future.23,24 22 Individuals apprehended in the U.S. interior, in contrast, could have crossed the border at a much earlier date, making interior apprehensions less strongly correlated with current attempts at illegal entry. 23 Between 1990 and 2003, 95 percent of those the Border Patrol apprehended agreed to depart voluntarily. 24 A further issue is that the majority of those attempt- ing illegal entry do not appear to be apprehended on any given attempt. Using MMP data, Massey and Audrey Singer (1995) find that for trips to the United States in the 1970s and 1980s the average probability of apprehension was 35 percent. and the number of attempts at illegal entry. Extending Wilfred J. Ethier's (1986) model, let (5) s where At is the number th o P(Ht, Mt) is th vidual is appr to cross the probability with is which a a and the numb enforcement sion probabili H,. For a give total attempts ability any sin (since This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms enforc �886 Journal of Economic Literature, Vol. XLIV (December 2006) United States. In late 2001, the U.S. Border more thinly across those attempting entry), making P(.) decreasing in M,. 25 Patrol increased its vigilance at border cross- Suppose P(H,,M,) = cHa' M-t,, where c, a, and a2 are positive constants, in which case log apprehensions can be expressed as migrants from crossing as frequently as they had in the past.27 (6) InA, = ao + alnH, + 1 - a2)ln4M Hanson and Antonio Spilimbergo (1999) use monthly data over the period 1968-96 to esti- mate equation (6), modeling attempted illegal entry as a reduced-form function of real wages in Mexico, real wages in the United States, other indicators of economic conditions in the two countries, a time trend, and monthly dummy variables.26 With an estimate of a,, we can solve for a function that is an affine transformation of InMt, given by ings, which may have dissuaded some The trend in figure 6 is roughly consistent with table 1 and results from previous estimates of the U.S. unauthorized population. Illegal immigration appears to have risen steadily after the end of the Bracero Program in 1965 and has been relatively stable at high levels for the past two decades.28 I will return to data on border apprehensions and enforcement when evaluating fac- tors that affect the level of illegal immigration and the political economy of U.S. policy on illegal immigration. (7) a0 + (1 - a2)lnM, = InA, - allnH, 2.5 Summary The expression on the right of (7) will posi- Currently, no data source gives precise estimates of the size of the U.S. illegal- tively covary with InMt as long as a2 < 1. Approximated attempts at illegal entry in (7) do not give an estimate of the level of illegal immigration. However, they may indicate the variation across time and the magnitude of log changes in attempted illegal entry. Figure 6 shows estimates of (7), based on instrumental-variables estimates of equation (6) (see note 26). Approximated attempts at illegal entry rise from the 1960s to the mid1980s, are stable from the mid-1980s to the mid-1990s, and then decline somewhat in 2000 and 2001. Part of the 2001 decline may reflect a change in border-crossing activity after the events of September 11th in the 25 For earlier work using apprehensions data, see Bean et al. (1990) and Borjas, Richard B. Freeman, and Kevin Lang (1991). 26 To deal with the possible correlation between enforcement and unobserved shocks to apprehensions, Hanson and Spilimbergo (1999) instrument for enforce- ment using U.S. government spending on national defense and indicators for whether there is an upcoming U.S. presidential, congressional, or border-state gubernatorial election. Border enforcement tends to follow a polit- ical cycle, dropping during election years (Hanson, Raymond Robertson, and Spilimbergo 2002). The reported coefficient estimate for a, is the long-run elasticity of apprehensions with respect to enforcement. immigrant population over medium or long time spans. Government data-gathering agencies have been wary of asking questions about an individual's immigration status, perhaps out of fear of dissuading illegal migrants from participating in surveys. The result is gaps in our knowledge about unauthorized migrants, which the literature has been able to partially fill in through other data sources. The perspective that emerges from the data that are available is that Mexico-to-U.S. illegal migration increased in the 1970s and 1980s and averaged around 200,000 to 300,000 net unauthorized entries per year in the 1990s and early 2000s. The population of illegal immigrants from Mexico in the 27 See Mary Jordan, "Mexicans Caught at Border in Falling Numbers," Washington Post, May 24, 2002. 28 One limitation of this exercise is that I assume border-crossing technology and border-apprehensions technology have been stable over time. There is anecdotal evidence that both may have changed considerably, especially since September, 2001. However, Hanson and Spilimnbergo (1999) find no evidence of a structural break in the apprehensions function for the 1968 to 1996 period. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 887 Figure 6. Estimated Attempts at Illegal Entry along the U.S.-Mexico Border United States includes a substantial fraction of women, is predominantly employed in nonagricultural jobs, and has schooling levels that are comparable to or higher than nonmigrating individuals in Mexico. Though many migrants maintain ties with family members in their origin communities, a majority appear to have settled in the United States on a medium or long-term basis. 3. The Supply of and Demand for Mexican Immigrants of future earnings, discount rates, and perceived cost of migrating. The cost of unau- thorized migration includes transport to the border, the physical risks and monetary charges incurred in crossing the border illegally, the psychic penalty from leaving one's friends and family behind, and the time and monetary expense of settling in another country. To uncover sources of variation in the demand for and the supply of illegal migrants, recent work estimates the sensitivity of migrant outflows from Mexico to variation in U.S. and Mexican wages, border-crossing costs, and access to Beginning with Larry A. Sjaastad (1962), economists have viewed migration as an investment decision. An individual migrates if the expected discounted difference in the migration networks. Also beginning with Sjaastad, economists tend to model the migration decision as irreversible (Greenwood 1997; Lucas 1997). stream of income between the new and old In many contexts, this assumption may be location exceeds moving costs. The incentive reasonable. U.S. legal immigrants, if they to migrate will vary across individuals wish to keep their green cards valid, must according to differences in their expectations make the United States their permanent This content downloaded from ff:ffff:ffff:ffff:ffff:ffff:ffff:ffff on Thu, 01 Jan 1976 12:34:56 UTC All use subject to https://about.jstor.org/terms �888 Journal of Economic Literature, Vol. XLIV (December 2006) residence.29 Those migrating to the United States legally would thus tend to be individuals expecting to stay in the country for an extended period of time. In the context of illegal migration from Mexico, individuals may operate on shorter time horizons. The substantial round-trip migration documented by the MMP indicates that for at least some Mexican nationals the cost of moving First, consider the components of illegal migration costs. Of these, transport costs to the border are likely to be small and psychic costs difficult to evaluate. Border-crossing costs and settlement costs are in principle measurable, though it is only the former that has been subject to much in the way of formal research. To avoid capture by the U.S. Border Patrol, migrants often pur- back and forth across the border is suffi- chase the services of a smuggler, known as a coyote.30 Coyotes offer a range of services, ciently low to warrant making the trip annufrom simply guiding migrants across the ally. Of interest to both Mexico and the United border to more complete packages that States is not just the volume of migrant flows include transport to an interior U.S. city, but their composition. Widening differences in earnings between immigrants and natives Phoenix.31 The MMP is one of the few data in the United States is cited as evidence that sources that asks migrants from Mexico recent U.S. immigrants are negatively select- about their border-crossing behavior. Using ed in terms of skill (Borjas 1999a). Following this line of thought, one might expect negative selection to be especially strong among illegal immigrants. Recent work examines migrant selection by comparing the charac- teristics of Mexican immigrants in the United States with those of nonmigrating individuals in Mexico. 3.1 The Incentive to Migrate from Mexico to the United States A long line of research applies the general framework in Sjaastad (1962) to examine the sensitivity of migration flows to the observed costs and benefits of migrating. For prospec- tive illegal migrants in Mexico, the costs include the four components identified above and the benefits include gains in real income associated with moving from low- wage Mexico to the high-wage United States. such as Houston, Los Angeles, or MMP data, Orrenius (2001) documents that, during the period 1978 to 1996, 69 percent of migrants reported hiring a coyote, as shown in figure 7.32 During this peri- od, the average price paid for coyote services varied between $385 and $715 ( in 2000 dollars). Since 2001, when the Border Patrol became more vigilant in monitoring U.S. border crossing points, coyote prices have risen. Based on a 2005 survey of return migrants in rural areas of two Mexican 30 Coyotes also help migrants navigate unfamiliar terrain. Recent changes in U.S. enforcement strategy have made it more difficult to cross the border in urban areas, encouraging migrants to enter the United States through the mountainous desert regions of Arizona and Eastern California, where temperature extremes expose migrants to physical risks. Annual deaths of border crossers have increased from an average of 100 during the mid 1990s to 410 during the period 2000-2004 (Cornelius 2005). With net annual illegal immigration from Mexico since 2000 at approximately 300,000 individuals (see table 1), there appear to be around 1.4 deaths per 1,000 successful net unauthorized migrants. 31 For the more expensive complete package of smuggling services, the coyote typically receives a portion of the fee up front and the remaining portion once the migrant is 29 A green-card holder may lose U.S. permanent resident status by taking permanent residence abroad, remaining abroad without obtaining a reentry visa, or by filing a foreign tax return as a nonimmigrant. Once a legal immigrant becomes U.S. citizens, he or she is free to enter and leave the country at will. safely delivered to friends or family members in the United States. See Charlie LeDuff, "The Crossing: A Special Report; A Perilous 4,000 Mile Passage to Work," New York Times, May 29, 2001, p. 1. 32 To reduce recall bias, I only show data for years within four of years of an MMP survey. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 889 Figure 7. Use of Smugglers by Migrants in the Mexican Migration Project states (that are also in the MMP), Cornelius (2005) finds that between the 1996-98 and 2002-04 periods average coyote prices rose by 37 percent from $1180 to $1680.33 One might be concerned that coyote prices based on MMP data are biased downwards. As mentioned in the last section, the hiring a smuggler in 2004 report using friends or family to find a coyote. Results using the MMP itself are consistent with concerns about bias in observed coyote prices. Gathmann (2004) finds that migrants with family members in the United States are less likely to use coyote services and, MMP surveys households in communities among migrants who do hire a smuggler, with high rates of migration to the United those with family members abroad pay lower prices. Yet, even if one accepts the high coy- States. Individuals in these communities ote prices quoted in the press,34 bordermay have relatively good access to migration networks, making them less dependent on crossing costs since 2001 appear to be no coyote services or better able to negotiate more than $2,000, which is 35 percent of lower prices from smugglers. Cornelius Mexico's 2003 per capita GDP. (2005) reports that 65 percent of migrants 34 Recent articles in the popular press give a price range for coyote services of $1,500-$2,000 ("Man Accused 33 The increase in prices is based on respondents recolof Smuggling Immigrants," The Oregonian, September 17, lections in 2005 of prices they paid in previous years and 2005, p. B2; "US Immigration," The Economist, May 19, thus may be subject to recall bias. In 2005, 90 percent2005; of "Between Here and There," The Economist, July 5, respondents in Cornelius's data (all of whom are from 2001; the "Illegal Immigration; Border Agents Understand a states of Jalisco or Zacatecas) report using a coyote on their Complex Issue," The San Diego Union-Tribune, May 26, previous trip to the United States. 2005, p. B12). This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �890 Journal of Economic Literature, Vol. XLIV (December 2006) Next, consider the benefits to migration. To calculate the gross return to migration, one would need to account for the expected in the migration incentive. Yet, since their specification is a reduced form, the coeffi- cient estimates do not allow one to recover length of stay in the United States, the expect- the elasticity of migration with respect to ed path of future earnings, and the discount rate applied to these earnings. In the absence of data on these components, much of the research on the decision to migrate takes a reduced-form approach to modeling how the returns to migration affect behavior.35 Using retrospective data from the MMP, Orrenius and Zavodny (2005) estimate the hazard that young males migrate to the United States as a function of individual and household characteristics and economic conditions in the two wage differentials between the United States and Mexico. Further, since the MMP is restricted to communities with historically high propensities to migrate to the United States, the results may not be informative about how prospective migrants in other regions of Mexico respond to changes in binational economic conditions. To examine how the gross level of attempted illegal migration responds to changes in U.S.-Mexican wages, Hanson countries.36 In the MMP, most migrants and Spilimbergo (1999) estimate an appre- appear to be entering the United States illegally. The migration hazard is nonmonotonic in schooling, increasing at low and medium hensions function, similar to that in equation (6). Using monthly data, they regress appre- hensions at the U.S.-Mexico border on schooling levels (primary and secondary lagged apprehensions, current and lagged school) and decreasing at higher schooling linewatch enforcement hours, the real peso levels (preparatory school and beyond).37 The wage for production workers in Mexican hazard is decreasing in per capita GDP in industry, real peso and real dollar U.S. wages (measured as the weighted average of wages Mexico and increasing in U.S. average wages. These results suggest that migrants tend to be drawn from the middle of the schooling dis- tribution and that migration is more likely during periods when U.S. income is expanding relative to income in Mexico. The migra- tion hazard is also higher for individuals whose fathers have migrated or whose siblings have migrated. This result could indicate the presence of family migration networks, or in U.S. industries that employ recent Mexican immigrants in large numbers), and other controls. They instrument for enforcement using U.S. government spending on national defense and indicators for whether there is an upcoming U.S. presidential, congressional, or border-state gubernatorial election (see note 26). Figure 8 shows a partial regression plot of log apprehensions on log Mexican real it could indicate the presence of unobserved household characteristics (e.g., unmeasured average hourly earnings, based on Hanson and Spilimbergo's estimates. Border appre- wealth) that influence migration behavior. hensions appear to be very responsive to changes in Mexican wages. A 10 percent By treating migration as a function of age, education, and other individual characteris- decline in Mexican real wages is associated tics, Orrenius and Zavodny pick up variation across individuals in the incentive to with a 6-8 percent percent increase in border apprehensions. Moreover, this effect is almost fully realized within three months following a migrate. By also including macroeconomic wage change, suggesting that shocks to the conditions, they pick up time-series variation Mexican economy are rapidly transmitted to changes in attempted illegal migration. 35 In earlier work, J. Edward Taylor (1987) examines migration behavior in a single rural community in Mexico.Over the past three decades, Mexico has 36 See also Oded Stark and Taylor (1989, 1991) and experienced wide variation in real income, as Massey and Espinosa (1997). periodic devaluations of the peso have lead 37 McKenzie and Rapoport (2004), who also use MMP data, obtain similar results. to bursts of inflation, which have caused This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 891 Figure 8. Border Apprehensions and Average Wages in Mexico Figure 9. Border Apprehensions and U.S.-Mexico Relative Per Capita Income This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �892 Journal of Economic Literature, Vol. XLIV (December 2006) incomes to fall sharply. Figure 9, which plots relative per capita GDP in the United States how long a migrant from Mexico would have to work in the United States in order to and Mexico, gives evidence of this volatility. recoup border-crossing costs, as approximat- During three separate episodes in the last ed by the price of coyote services. I focus on males, since, as tables 2 and 3 show, there are declined by five log points or more (relative large differences in labor-force participation rates between women in Mexico and thirty years, Mexico's per capita GDP to the United States) within the space of three years. Each of these episodes was followed by an increase in border apprehensions. Much of the research on interregional migration in the United States and other countries finds that it is labor earnings in the receiving region, and not the sending Mexican immigrant women in the United States, which complicates comparing female wage outcomes across national borders. By limiting the analysis to current wage differ- ences and a single component of migration costs, this exercise falls well short of a com- region, that appear to drive migration flows plete cost-benefit accounting of the migra- appear to be the case in Mexico, where tion decision. Still, given large back and forth flows of labor across the U.S.-Mexico bor- (Greenwood 1997). However, this does not income volatility appears to be a strong push der, the current U.S.-Mexico wage differen- factor for illegal migration. tial is likely to be the relevant gross return to While apprehensions are the only avail- migration for at least some prospective able high-frequency measure of gross migrants. have important limitations. Since individual Table 4 reports average hourly earnings by age and schooling categories for males in attempts at illegal migration, these data migrants may be apprehended multiple times in a given time period, the number of apprehensions may far exceed the gross number of migrants (Espenshade 1995). Controlling for border enforcement addresses this problem, but only if one has valid instruments for enforcement (since enforcement is likely to be endogenous to shocks to attempted illegal migration) and if the impact of enforcement on apprehen- sions (controlling for the incentive to migrate) is stable over time (see notes 26 and 28). 3.2 U.S.-Mexico Wage Differences Mexico (based on the 2000 Mexico Census of Population and Housing) and for immigrant males from Mexico in the United States (based on the 2000 U.S. Census of Population and Housing).38 To increase the share of illegal immigrants among Mexican immigrant men, I limit the sample to very recent immigrants (individuals residing in the United States for 0-3 years). To adjust for cost of living differences between the countries, I scale up Mexican hourly wages to achieve purchasing power parity with the United States, using the 2000 PPP adjustment factor for Mexico in the Penn World Tables.39 The reduced-form results of Orrenius and Zavodny (2005) and Hanson and tude of the returns themselves. As a crude 38 For Mexico, average hourly wages are calculated as monthly labor income/(4.5 x hours worked last week); for the United States, average hourly wages are calculated as annual labor income/(weeks worked last year x usual hours worked per week). For Mexico, I need to assume individuals work all weeks of a month, which could bias wage estimates downwards. To avoid measurement error associated approximation of the short-run gross return to migration, I examine differences in hourly wages for men in Mexico and Mexican immigrant men in the United States in 2000. I ask with implausibly low wage values or with top coding of earnings, I drop the largest and smallest 0.5 percent of wage values. 39 In 2000, Mexico's PPP-adjusted price level was 61 percent of the U.S. price level. Spilimbergo (1999) suggest that illegal migration flows are highly responsive to changes in the return to migration. However, these results give no indication of the magni- This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 893 TABLE 4 AVERAGE HOURLY WAGES FOR MEXICAN MALES, 2000 Years of Schooling Completed Age 4 5 to 8 9 to 11 12 13 to 15 16+ Mexican 18 to 22 7.83 7.60 7.45 8.07 8.76 8.44 Immigrants 23 to 27 8.44 8.19 8.21 9.06 9.53 13.02 in U.S. 28 to 32 8.27 8.56 8.70 9.66 9.56 15.69 33 to 37 9.46 9.25 9.34 10.07 11.36 16.84 38 to 42 9.19 9.39 9.33 11.01 12.11 16.26 43 to 47 9.75 8.90 9.35 10.68 12.80 15.88 48 to 52 9.57 9.37 9.42 9.31 11.65 17.78 Residents 18 to 22 1.36 1.56 1.76 2.06 2.61 3.91 of Mexico 23 to 27 1.43 1.80 2.10 2.79 3.77 5.20 28 to 32 1.56 1.93 2.42 3.22 4.80 33 to 37 1.65 2.08 2.56 3.45 5.25 6.63 7.07 38 to 42 1.64 2.14 2.88 3.74 5.62 7.42 43 to 47 1.69 2.30 3.00 4.40 5.86 8.05 48 to 52 1.66 2.30 3.15 4.21 6.11 8.71 Note: Table shows average hourly wages in 2000 or male residents of Mexico who report workin 0.5 percent of wage values are excluded). Data f microsample of the XIII Censo General de Pobla percent U.S. PUMS in 2000. Mexican immigrants States for three years or less. See the text on the Not surprisingly,high-migration wages are substanti states Mexico as a whole. higher among Mexican immigrants in United States than In among table 5, resident the log b Mexico. For 23-27 ence year-old males, the declines with sc adjusted hourly wage differential varies f with empirical resea $7.01 for those with 0-4 yearsreturns of schoolin estimated to Mexico. In the $5.76 for those with 13-15 years of 1990s schoo to $7.82 for those with 16 or year more of y an additional of schooling. Given that migration pro with an increase in w ties vary widely across regions grant men of ofMexico 2.5 t might think that the average hourly wag 1996; Trejo 1997; Jef 2002).may In the i the country as a whole not1990s be the and vant alternative wage for mostyear prosp an additional of migrants. Table 5 reports wage differe with an increase in w between the United States and high-m 9.7 log points (Dan tion states in Mexico, defined to be st Returns to experie with above-average 2000.40 emigration rat 41wage On the returns to ed U.S.-Mexico differentials Michael Ian Cragg and Mar Airola and Chinhui Juhn (2 40 These states are Aguascalientes, Colima, Dur schooling are correlated, es Guanajuato, Guerrero, Hidalgo, Jalisco, Mich may be biased. Also, self-se Morelos, Nayarit, Oaxaca, San Luis Potosi, Zac into migration may and introdu This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �894 Journal of Economic Literature, Vol. XLIV (December 2006) TABLE 5 LOG U.S.-MEXICO HOURLY WAGE DIFFERENTIAL, MALES IN 2000 Years of Schooling Completed Age 4 5 to 8 9 to 11 12 13 to 15 16+ Log of U.S.-Mexico Wage Differential, All Males in 18 to 22 1.263 1.098 0.985 0.895 0.750 0.361 (0.022) (0.013) (0.015) (0.018) (0.042) (0.099) 23 to 27 1.281 1.043 0.904 0.746 0.507 0.387 (0.024) (0.015) (0.018) (0.018) (0.038) (0.048) 28 to 32 1.225 1.036 0.821 0.652 0.306 0.357 (0.029) (0.018) (0.024) (0.023) (0.047) (0.052) 33 to 37 1.258 1.025 0.823 0.613 0.339 0.294 (0.034) (0.023) (0.034) (0.031) (0.057) (0.059) 38 to 42 1.287 1.019 0.775 0.575 0.288 0.216 (0.035) (0.029) (0.045) (0.043) (0.068) (0.075) 43 to 47 1.265 0.905 0.708 0.498 0.275 0.118 (0.040) (0.037) (0.063) (0.063) (0.103) (0.116) 48 to 52 1.264 0.987 0.715 0.418 0.164 0.122 (0.050) (0.046) (0.096) (0.076) (0.122) (0.142) Log of U.S.-Mexico Wage Differential, Males in High-Migration States 18 to 22 1.188 1.061 0.989 0.912 0.676 0.381 (0.026) (0.016) (0.017) (0.025) (0.052) (0.128) 23 to 27 1.211 1.013 0.895 0.722 0.481 0.425 (0.027) (0.017) (0.019) (0.024) (0.050) (0.052) 28 to 32 1.144 1.011 0.824 0.657 0.361 0.415 (0.031) (0.020) (0.025) (0.028) (0.057) (0.053) 33 to 37 1.151 0.988 0.821 0.652 0.317 0.411 (0.035) (0.025) (0.034) (0.034) (0.061) (0.060) 38 to 42 1.206 0.966 0.792 0.601 0.185 0.291 (0.036) (0.030) (0.044) (0.046) (0.070) (0.073) 43 to 47 1.194 0.867 0.671 0.528 0.237 0.181 (0.041) (0.037) (0.061) (0.067) (0.101) (0.112) 48 to 52 1.193 0.961 0.727 0.292 0.284 0.227 (0.049) (0.045) (0.092) (0.080) (0.125) (0.137) Notes: Table shows the log difference in average hourly wages (and standard err Mexican immigrant males and male residents of Mexico. The top half of the tab includes all regions of Mexico; the bottom half includes only high-migration sta table 4 and the text for details. higher for Mexican residents than for Mexican immigrants. Higher returns to education in Mexico are consistent with the In 2000, a 23-27 year-old recent Mexican migrant with 5 to 8 years of schooling (the category just below the national mean level of country having a low relative supply ofschooling for Mexico) would recoup borderskilled labor compared to the United States. crossing costs of $2,000 in 313 hours, or 7.8 Higher average wages in the United States weeks based on a forty-hour work week. The (within schooling groups) are consistent with speed with which migrants would seem to the country having a higher relative level ofrecover border-crossing costs suggests that TFP and higher relative supplies of physical other costs to migration (including psychic costs and financing costs associated with capital. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 895 credit constraints) are large. In a simple stat- ic model, the equilibrium wage differential between two countries is fixed by migration costs. To reconcile persistent U.S.-Mexico wage differences with small border-crossing costs, there would need to be positive unobserved migration costs (otherwise migration flows would be larger) and heterogeneity across individuals in these costs (otherwise migration flows would be lumpier, with either everyone or no one in a skill group willing to migrate). There are reasons to be skeptical about how well the observed U.S.-Mexico wage differ- What is surprising, perhaps, is that the estimated wage premium for legalization is so small. One might expect that being confined to the underground economy would substantially limit workers' employment prospects. Once legalized, they would enjoy strong wage growth. By this reasoning, a 6log-point gain over three years is unimpressive. One possibility is that the three years covered by the LPS is not long enough for individuals to realize the wage benefits associated with gaining legal status. Legalization may open up new opportunities to move between occupations or between regions, ential captures the incentive to migrate. This differential reflects the binational difference which migrants need more time to exploit. in returns to both observed and unobserved Cobb-Clark's estimate is biased downwards. characteristics. If migrants are positively selected in terms of unobserved skill, observed wage differences will tend to overstate true wage differences. In the next section, we address the issue of which types of individuals in Mexico select into migration. Another issue is that, since average wages for Mexican immigrants in the United States are a composite of wages for legal and illegal immigrants, they may overstate wages an illegal immigrant would expect to earn. Sherrie A. Kossoudji and Deborah A. CobbClark (2002) use the LPS to compare wages for illegal immigrants before and after they obtained green cards under the amnesty pro- vision of the 1986 IRCA. Between 1989 and Another possibility is that Kossoudji and Their control group includes legal and illegal immigrants from Latin America, as well as second and later generation Latin American immigrants. Suppose illegal immigrants are more negatively selected in terms of unobserved skill than legal immigrants (where overall legal immigrants may be positively or negatively selected in terms of skill). In an economy where the return to skill is rising (as in the United States during the 1990s), stronger negative selection of illegal immigrants in terms of unobserved skill would tend to make their unobserved wage growth relatively small and lead one to underestimate the wage premium due to legalization. The differential in U.S. and Mexican 1992, average hourly earnings for newly legalized immigrant men (64 percent of wages in table 5 may also miss important whom were Mexican nationals) rose by 6 log points relative to earnings for Latino men in the NLSY, controlling for observable characteristics. Also, prior to legalization, illegal- incentive to migrate. For instance, average wages for Mexican immigrants in the United States may hide heterogeneity in expecta- immigrant men had relatively slow wage growth. Not surprisingly, illegal status is asso- ciated with lower wages and less opportunity for occupational advancement (Kossoudji and Cobb-Clark 1996, 2000).42 sources of variation across individuals in the tions about U.S. wages among prospective migrants. Individuals with previous U.S. labor-market experience or who speak English well might have relatively strong wage expectations due to the ease with illegal immigrants. For other work on the wage consequences which they expect to find a U.S. job or the high productivity they anticipate having on the job. Though not specific to Mexico, Hoyt Bleakley and Aimee Chin (2004) find that, of IRCA, see Cynthia Bansak and Steven Raphael (2001). for immigrants from non-English-speaking 42 Francisco L. Rivera-Batiz (1999), who also uses the LPS, estimates larger wage differences between legal and This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �896 Journal of Economic Literature, Vol. XLIV (December 2006) countries that arrived in the United States as wage differentials are large today and have children, wages are higher for those with stronger English-language skills.43 Another source of heterogeneity in wage expectations among prospective migrants is variation in access to migration networks. Individuals with friends or family that have migrated abroad may have better information about how to find a job in the United States. Munshi (2003), using data from the MMP, finds that Mexican migrants in the been large for over a century (Durand, Massey, and Zenteno 2001). The emerging United States are more likely to be employed and more likely to be employed in a (higher-paying) nonagricultural job the larger is the U.S. population of residents from their origin community in Mexico.44 He instruments for the time-varying size of the U.S. population from a migrant's origin community in Mexico using lagged rainfall in the migrant's origin community (which literature on migration networks suggests that networks arise in response to hidden migration costs associated with finding employment and getting settled abroad. William J. Carrington, Enrica Detragiache, and Tara Vishwanath (1996) develop a model of regional labor movements in which migration networks lead migrant flows to be sluggish initially, when the migrant population is small, and then to accelerate over time, as migration costs endogenously fall in response to past migration. Imperfect credit markets, which we discuss in the next sec- tion, are another explanation for sluggish migration. 3.3 The Selection of Migrants from Mexico presumably affects the marginal productivity In an important body of work, Borjas of labor in Mexican agriculture and so the (1987, 1991) argues that who migrates to the United States from a particular country will depend on that country's wage distribution. In a country with high returns to skill and incentive to migrate abroad). His results suggest that having a larger network improves a migrant's ability to assimilate economically in the United States. Migration high wage dispersion, as in Mexico, there networks appear to be organized around will be negative selection of migrants. Those families. Among nonagricultural (agricultur- al) workers, 78 percent (74 percent) with the greatest incentive to migrate to the United States will be individuals with below- received assistance in finding a U.S. job and, among this group, 47 percent (43 percent) received help from a relative and 47 percent (43 percent) received help from a friend or paisano (someone from their home region in average skill levels in their home countries. In support of this idea, Borjas (1987, 1995) finds that as sources for U.S. immigration have shifted from Europe to Latin America and Asia, the economic performance of new Mexico). The small remaining fraction of immigrants has deteriorated. Relative to ear- those receiving assistance relied on an wage expectations among migrants, the volume of Mexico-to-U.S. migration is smaller lier cohorts, recent immigrants earn lower wages compared to natives at time of arrival and take longer for their earnings to converge to native levels. These findings counter an earlier belief that immigrants tend to than one might expect. U.S.-Mexico real have high potential for earnings growth employer, labor contractor, or other source. Even accounting for heterogeneity in (Barry R. Chiswick 1978). 43 Bleakley and Chin (2004) instrument for English-language ability using immigrant age at arrival, exploiting the fact that younger children appear to learn new languages more easily than older children. 44 For other work on migration networks in Mexico, see Paul Winters, Alain de Janvry, and Elisabeth Sadoulet (2001). A simple test of the negative selection hypothesis is to compare the observable skills of migrants from Mexico with individ- uals in Mexico who choose not to migrate abroad. While selection on observables does not necessarily reflect selection on This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 897 TABLE 6 EDUCATIONAL ATTAINMENT FOR THE MEXICO BORN POPULATION, 2000 21-65 Year Olds 28-37 Year Olds Residents Mex. Imm. Residents Mex. Immigrants in U.S. of Mexico in U.S. of Mexico 0-3 Yrs 4+ Yrs Males 0 0.069 0.127 0.036 0.076 0.078 1-4 0.166 0.080 0.103 0.045 0.043 5-8 0.270 0.307 0.263 0.303 0.275 Highest 9 0.189 0.087 0.246 0.114 0.116 Grade of 0-9 0.694 0.601 0.648 0.538 0.512 Schooling (%) 10-11 0.045 0.055 0.059 0.064 0.067 12 0.101 0.212 0.129 0.260 0.278 13-15 0.047 0.083 0.044 0.075 0.098 10-15 0.193 0.350 0.232 0.399 0.443 16+ 0.113 0.050 0.121 0.062 0.046 N 215,804 80,453 68,206 9,358 15,839 Females 0 0.092 0.133 0.047 0.070 0.078 1-4 0.179 0.087 0.116 0.041 0.043 5-8 0.280 0.315 0.278 0.269 0.294 Highest 9 0.174 0.085 0.219 0.112 0.125 Grade of 0-9 0.725 0.620 0.660 0.492 0.540 Schooling (%) 10-11 0.040 0.049 0.054 0.056 0.062 12 0.112 0.204 0.145 0.276 0.253 13-15 0.042 0.079 0.039 0.083 0.090 10-15 0.194 0.332 0.238 0.415 0.405 16+ 0.080 0.048 0.103 0.093 0.055 N 235,086 72,967 75,625 6,575 16,173 Notes: The sample is individuals 21-65 or 28-37 years old excluding group quarters; in Mexico, excluding those not b Residents of Mexico in 2000 are a 10 percent random sam microsample of the XIII Censo General de Poblacion y Viv grants are from the 2000 5 percent U.S. PUMS and are re years or older at time of entry into the country who have States for 0-3 years or 4 or more years. Schooling variabl individuals whose high grade completed is that indicated. unobservables, one might expect individuUnited States). Section 2 suggests a relaals' observable and unobservable skills torecent be immitively high fraction of very positively correlated. Table 6 likely shows educagrants are to be illegal immigrants. To tional attainment for Mexican immigrants control for age, I limit the sample to 28-37 in the United States and for residents of year olds, which in tables 2 and 3 is the age Mexico in 2000, based on census data from cohort with the highest likelihood of migrat- the two countries. To help isolate the popu- ing abroad. For comparison, the table also lation of illegal immigrants, the table shows describes educational outcomes for the full results separately for very recent immi- sample of working-age residents of Mexico grants (0-3 years in the United States) and and immigrants from Mexico in the United for longer-term immigrants (4+ years in the States. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �898 Journal of Economic Literature, Vol. XLIV (December 2006) countries in which there appears to be an It is well known that Mexican immigrants in the United States are much less educated inverted U-shaped relationship between rural household income and the likelihood that a than U.S. natives (Borjas and Katz forthcoming). However, Mexican immigrants, whether household finances urban migration for one or recent or longer term, compare favorably more of its members (Lucas 1997). when we examine residents of Mexico. In Chiquiar and Hanson (2005) develop a more formal approach to evaluate the selec2000, 65 percent of male residents of Mexico had nine or fewer years of schooling, com-tion of Mexican immigrants in terms of pared to 54 percent of recent immigrants andobservables. Let the observed density of wages for individuals working in Mexico be 51 percent of longer-term immigrants. Beyond nine years of education, Mexican immigrants outperform Mexican residents in (8) g(wli= MX)= JfMX(wIx)h(xi i= MX)dx, every category except college graduates. and the observed density of wages for Relative to male residents of Mexico, recent Mexicans working in the United States be Mexican immigrant men are more likely to have ten-fifteen years of education (40 per- (9) g(wli = US) =ffus(wIx)h(x|i = US)dx, cent versus 23 percent) and less likely to have sixteen plus years of education (5 percent where fi(wIx) is the density of wages w in country i, conditional on a set of observed versus 12 percent). A similar pattern holds for men in 1990 and for women in either characteristics x, and h(xli) is the density of year. It appears that, in Mexico, individuals observed characteristics in i. Consider the with moderate to high education levels havedensity of wages that would prevail for the highest likelihood of migrating abroad, Mexican immigrants in the United States if which is inconsistent with negative selectionthey were paid according to the price of skills in Mexico: of migrants in terms of observable skills. Other data on Mexican migration are also inconsistent with negative selection. Using his- (10) gM(w) = JfMX(wIx)h(xli = US)dx. torical data from U.S. and Mexico population John DiNardo, Nicole M. Fortin, and Thomas censuses, Zadia M. Feliciano (2001) finds that Lemieux (1996) show that, under the assumpaverage schooling of Mexican immigrants has tion that the distribution of unobservables been higher than for residents of Mexico since (conditional on the distribution of observat least 1940. Based on MMP data, in which ables) is the same in the two countries, a counthe vast majority of individuals who migrate to terfactual density as in (10) can be written as the United States do so illegally (at least on their first attempt), Orrenius and Zavodny (11) (2005) estimate that the probability a young ggs(w)= f OfMx(wlx)h(xli =MX)dx, adult male migrates to the United Stateswhere increases as schooling rises from low levels to (12) levels around the national mean (eight years) and then declines as schooling rises to levels above the national mean. McKenzie and Rapoport (2004) obtain similar results, also theta = h(xli = US) h(xli = MX) = Pr(i = US Ix) Pr(i = MX) 46 Pr(i = MX Ix) Pr(i = US) using the MMP.45 These findings are similar to 46 Chiquiar and Hanson (2005) show how to derive a literature on internal migration in Botswana, weighting function similar to (12) that controls for differ- India, the Philippines, and other developingences in labor-force participation among workers in the two countries. To simplify the exposition, I leave out the analytics behind this extension. The results in figure 10 control for differences in labor-force participation between 45 On migrant selection in Mexico, see also Pablo Ibarraran and Darren Lubotsky (forthcoming). Mexico and the United States. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 899 Thus, the counterfactual density in (11) can be estimated by taking the observed density for wage earners in Mexico and reweighting weights in (12) has as regressors dummy vari- grant workers in the United States. The wage individuals who exhibit a stronger ten- it to reflect characteristics of Mexican immi- weight in (12) can be estimated parametrically by running a logit on the probability of a Mexico-born adult being in the United ables for age, schooling, and marital status, and interactions of these variables. Consistent with table 6, it is not the lowest- dency to migrate to the United States. In either year, for Mexican immigrant males there is greater mass in the middle of the States, conditional on observed characteris- wage density and less mass in either tail tics, using a sample that combines Mexican immigrants and Mexican residents. By comparing actual and counterfactual when compared with the actual wage density of Mexico residents. The immigrant-resident wage densities, we can nonparametrically tail to just below zero, positive for middle wage values, and again close to zero in the summarize the nature of migrant selection in Mexico. Consider the difference between density difference is close to zero in the left right tail. This suggests that immigrant males are drawn disproportionately from the midthe actual wage density for residents of dle of Mexico's wage distribution, rather than Mexico and the counterfactual wage density that would obtain were Mexican immigrants from the bottom half. Low-wage and highwage individuals appear to be relatively less paid according to skill prices in Mexico: likely to migrate to the United States. These (13) gs(w) - gMX(w) counterfactual wage densities support inter- = f [OM - 1]fMx(wlx)h(x i = MX)dx. If there is negative selection of migrants in terms of observable skills, the difference in (13) would show positive mass in the lower part of the wage distribution-indicating migrants are overrepresented among Mexico-born individuals with below-average skills-and negative mass in the upper part-indicating migrants are underrepresented among the Mexico-born with above- average skills. In contrast, with positive selection there would be negative mass for low wages and positive mass for high wages. Figure 10 shows estimates of the density difference in (13) for men and women in 1990 and 2000, based on results in Chiquiar and Hanson (2005). The sample is workingage adults (21 to 65 years of age) who either reside in Mexico or are very recent Mexican immigrants in the United States. Immigrants are individuals twenty-one years or older at time of U.S. entry and who have been in the United States for three years or less. (Again, very recent immigrants are likely to include a relatively high fraction of illegal immigrants.) The logit regression used to estimate the mediate selection of immigrant men in terms of observable skills. The results for females contain even less support for negative selection. The immigrant-resident density difference is negative for low wage values, strongly positive for upper-middle wage values, and zero for high wage values. For women, there appears to be moderate positive selection of immigrants. Table 6 and figure 10 do not explicitly separate legal and illegal Mexican im-migrants, leaving it unclear how the two groups compare in terms of skill. What does theory suggest about the self-selection of legal versus illegal migrants? Consider a simple extension of Borjas (1991), who adapts the A. D. Roy (1951) model.47 Let individuals from Mexico, indexed by 0, choose whether or not to migrate to the United States, indexed by 1, where migration is a one-time decision (or, equivalently, a one-period decision). Residents of Mexico face a wage equation given by 47 For expositional simplicity, I consider a nonstochastic version of Borjas (1991), in which there is no unobserved component of skill. The extension to a stochastic setting is straightforward. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �900 Journal of Economic Literature, Vol. XLIV (December 2006) Figure 10. Immigrant (counterfactual)-Resident (actual) Wage Densities, 1990 and 2000 (counterfactual density for Mexican immigrants minus actual density for Mexican residents) This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 901 (14) In(wo) = g0 + o0s, agents are capable of migrating abroad, as in where for Mexico wo is the wage, p0 is the base wage, s is the level of schooling, and 60 is the return to schooling. If the population of Mexicans were to migrate legally to the (2004), and Orrenius and Zavodny (2005).48 To capture the relationship between credit constraints and skill, suppose that individual wealth in Mexico, Yo, is a linear function United States, they would face the wage equation (15) ln(wL) = pL + 6Ls, Rapoport (2002), McKenzie and Rapoport of skill, such that (18) Y0o = Po+ Cos. Low-skill individuals will be unable to while if they were to migrate illegally they would face wage equation, finance migration because they have insufficient collateral to secure a loan. The low- (16) ln(wI) = pL{ + 5s, feasible is est skill level for migration strategy i to be (19) = where wl is the wage, yp is the base wage, and S8 is the return to skill for Mexican migrants with status i. Higher levels of TFP and larger which is higher the larger ar relative supplies of capital in the United costs. Whether the least s States suggest the base wage is lower inmigrant is more or less skilled th Mexico (pi > p0, for i = I, L); higher U.S. rel- skilled illegal migrants depends ative supplies of human capital suggest that tive magnitude of legal and ill the return to schooling is higher in Mexico tion costs. In Cornelius (2005), recent (i8 > do, for i = I, L). Tables 4 and 5 supportborder-crossing costs for illegal migrants both assumptions. Kossoudji and Cobb-Clarkrange from $1,200 to $1,700 (in 2000 dol(2002) find that in the United States the base lars). The current costs of legal migration wage and the return to schooling are higherinclude fees paid to the U.S. government to for legal than for illegal immigrants, whichprocess a visa application, which range suggests p1> Ž4 p and 31 2 S~. from $700 and $1,000 for family-sponsored Migrant selection also depends on theimmigration visas, and the expense of hirstructure and financing of migration costs.ing a lawyer or immigration specialist to Consider a setup similar to Borjas (1987), inhandle the application process, which typi- which the migration cost for migration status i,cally ranges from $400 to $1,000.49 C', expressed in time-equivalent units,Ignoring the time costs involved in crossing 7' = C*/wo, is assumed to be constant acrossthe border illegally or in completing the individuals. Combining equations (14)-(16), a bureaucratic steps needed to obtain a green resident of Mexico will be willing to migrate tocard, entry costs for legal migrants from the United States under migration status i if Mexico appear roughly similar in value to (17) In(wi) - In(w0 + C') =- ln(wi) - In(wo) - ri > 0. those for illegal migrants, in which case s, and sL would also be similar. Based on (17), high-skill individuals will choose not to migrate because higher Suppose that migrants finance migration costs by borrowing and that to secure a loan 48 I limit the analysis to partial equilibrium. Rapoport of amount C' a migrant must have collateral (2002) develops a general-equilibrium model of credit conof amount yC', y > 0. Collateral requirements straints and migration, based on the occupation-choice model in Abhijit V Banerjee and Andrew F. Newman (1993). may reflect imperfect credit markets, which 49 On fees for immigration visas, see http://uscis.gov/ place initial wealth restrictions on whichand http://www.usavisanow.com/. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �902 Journal of Economic Literature, Vol. XLIV (December 2006) returns to skill in Mexico make it more U.S. green card in the future. As of yet, attract to stay at home. The highest skill level there is little empirical research on which for which migration strategy i is attractive is types of households in Mexico appear to (20) s have better access to legal channels of obtaining a U.S. green card. Combining (19) and (20), it is clear that as migration costs rise both legal and illegal which is higher the larger migrants the U.S.-Mexico become more positively selected in skillsmaller (i.e., the interval (s for i = I, difference in the base wage terms and of the the U.S.-Mexico difference in the to return L, shifts the right).to In the MMP, Orrenius skill. Thus, larger differences and in Zavodny the return to that as the level (2005) find of negative U.S. border enforcement rises (U.S. skill between countries make seleclinewatch enforcement tion of migrants in terms of skill more pro-hours increase) the nounced. I assume parameter values are probability a young adult male migrates to such that Si > si for i = I, L, which yields posthe United States falls, with the effect being itive levels of both legal and illegal migra-with lower schooling stronger for individuals tion. As long as the lower support the levels. This for suggests that higher migration distribution of skill is less costs than min (s_,sI)select lower-skilled disproportionately and the upper support of the distribution individuals out of of the migrant pool. Similarly, and Rapoport (2004) skill is greater than max (ssi), both McKenzie legal and in MMP communities with illegal migrants will tend tofind be that, drawn from individuals with intermediate skill U.S.-migration levels. If stronger networks, the legal and illegal migration costs are similar, inverted-U-shaped relationship between Kossoudji and Cobb-Clark's (2002) results migration and schooling is weaker. As migra- on base wages and returns to tion skill costs fall for (access legal to migration networks and illegal migrants would improves), suggest positive that selection of migrants becomes less pronounced. 9L >S or that the highest skilled legal There abundant evidence that Mexican migrants are at least as skilled as is the highest skilled illegal migrants. immigrants in the United States are disproThe simple theoretical model together portionately drawn from the middle of the distribution of observable skills in Mexico. with available empirical evidence suggests that legal migrants would beWhile more there positively are no data that explicitly differselected than illegal migrants in terms of entiate between the selection pattern of legal skill. However, this outcome and illegal implicitly migrants, there is intermediate selection among migrants depends on all prospective migrants in in the MMP, in Mexico having an equal probability which a very of high being fraction of migrants are illegal. Intermediate selection eligible for a U.S. legal immigration visa. is consistent with migrants credit constraints in financing There is no reason to expect thisfacing to be the case. If more-skilled individuals are more the cost of migration, such that low-skill, low- (less) likely to have family members who are income individuals are disproportionately U.S. citizens or U.S. legal permanent resiselected out of the migrant pool. Other factors that could contribute to intermediate dents, they will be more (less) able to obtain selection include higher discount rates, a U.S. green card, in which case there would be stronger (weaker) positive selection of greater risk aversion, or higher psychic migralegal migrants relative to illegal migrants.tion costs among low-skilled, low-income individuals, such that these individuals are Also, the discussion ignores dynamic considerations in which individuals may migrateless willing to migrate to the United States for any given binational wage differential. Credit illegally today in expectation of obtaining a This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 903 constraints or some other factor is needed to integration, as are cross-border trade and reconcile the facts that (a) the incentive to investment flows. migrate appears to be strongest for individuals with low schooling (as seen in tables 4 and labor flows from Mexico to the United 5), and (b) individuals with low schooling have a relatively low probability of migrating abroad (as seen in table 6). To examine further the determinants of States, consider a simple two-country model in which the aggregate production function for Mexico (indexed by 0) at time t is given by (21) Qo 3.4 Migration and the Supply of Labor in Mexico and the United States where A, is total factor produ the supply of capital, Loht is labor of skill type h, v < 1 d focuses primarily on the correlation elasticity of substitution betw between migration and U.S.-Mexico earnlabor (v= 1 - 1/KL), and < ings differences. There is little work on the the elasticity of substitution underlying causes on these wage differences skill types (K = 1- 1/o-hh, all h or whether migration is related to variation the wage to the marginal produ in these causal factors. The work that perhaps most closely addresses these issues is (22) InwOht = (1 - v)lnQ0 Robertson (2000), who examines the corre- As we have seen, the literature on labor flows from Mexico to the United States lation between U.S. and Mexican wages over time. Using synthetic cohorts con- structed from household data in the two + (v - K')lnL0, + (K - l)lnLoht where Lot is the labor aggreg Pot is the price of Mexico's out countries over 1987-97, he regresses the gate production function f quarterly change in Mexican wages for a given age-education-region cell on quarter-States (indexed by 1) has an an ture, with the added dimens ly changes in U.S. wages for the same skill type h, Llht, is an agg age-education cell and on the lagged difference in U.S. and Mexican wages for the cell.employment of native and im ers, who are imperfect substitu A shock that raises U.S. wages by 10 percent is associated with an increase in wages in (23) lht Mexican interior cities by 1.8 percent and wages in Mexican border cities by 2.5 per-where Llht is the supply of n cent.50 Positive comovements in U.S. and labor of skill type h, Iht is the s Mexican wages are consistent with the two grant labor of skill type h, an countries' labor markets being at least parmines the elasticity of substit tially integrated. Migration flows are one native and foreign labor (77 = factor that may contribute to labor-market marginal product of immigra United States (assumed to be 50 Robertson (2000) also finds that wage changes inlegal and illegal immigrants) i Mexico are negatively correlated with the lagged U.S.-Mexico wage difference, which suggests that over(24) time wages in the two economies tend to converge. The estimated convergence rates are very rapid, with equilibri- Inwlht = (1- v)lnQ1t + (v + (K - ?7)lnLlht + (hq - l)lnIht um U.S.-Mexico wage differentials being reached within one to two quarters. Rapid convergence seems at odds where with rising levels of trade, investment, and migration L,, is the U.S. labor agg is the price of U.S. output. between the two countries, which suggests that integration of U.S. and Mexican markets is incomplete and that wage If migration equalizes wages convergence between the two countries is more gradual. United States and Mexico, th This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �904 Journal of Economic Literature, Vol. XLIV (December 2006) (25) Inw0ht-lnP0t = Inwth, - InP, - Incht, where Pkt is the consumer price index in country k (expressed in terms of a common currency) and cht = 1- Cht/(wIht/Plht) is the Mexico-U.S. relative real GDP per worker, the Mexico-U.S. terms of trade, and the Mexico-U.S. real exchange rate over the 1960-to-2000 period (based on the Penn World Tables). Relative real GDP per work- migration cost for labor type h expressed as a markup over the wage rate. Cht may be a function of U.S. border enforcement or, if er is a crude proxy for a term that combines the first, second, fourth, and fifth terms in migration networks exist, of the existing U.S. stock of immigrants from Mexico. of labor for skill type h and migration costs). Using (22)-(25) to solve for the level of immigration of skill type h, (26) InIht= + L1, (26) (i.e., all terms except the relative supply Relative income declines in the 1980s and is flat in the 1990s. Slow growth in Mexico's GDP and rapid growth in Mexico's labor force have combined to make the U.S.-Mexico gap in income per worker la er in 2000 than it was in 1980. Changes the terms of trade appear to have matte little for relative income changes. Over last two decades, the Mexico-U.S. terms trade have been stable, consistent with two countries exporting similar type Immigration of skill type h worker manufactured goods.51 Mexico is (a) decreasing in Mexico-U Changes in aggregate income may und state the contribution of binational ative GDP (which, in turn, is income decrea Mexico-U.S. relative TFP and the differences to Mexico-U.S. migration. Mexico-U.S. relative supply of capital), Relative (b) average aggregate income may hide increasing in the Mexico-U.S. relative variation labor in income across sectors or regions aggregate (as long as different labor in Mexico skill that affect migrant outflows. For negative income shocks to Mexican types are more substitutable than instance, are aggreagriculture or to Mexico's high-migration gate labor and capital, such that K - v a 0), (c) increasing in the Mexico-U.S. states relative would likely increase migration supply of skill type h (as long as immigrant abroad, even if positive income shocks to and native workers of the same skillother typesectors are or regions helped smooth more substitutable than are labor ofincome different at the national level. Also, the rela- tive variability of income may affect migraskill types, such that 7 - c > 0), (d) decreastion, independent of changes in relative ing in the Mexico-U.S. terms of trade, (e) mean income. increasing in the Mexico-U.S. relative cost As movements in Mexico's relative price level in figure 11 indicate, of living, and (f) decreasing in migration Mexico's economy has recently been subject costs. All else equal, immigration rises when to a high degree of price volatility. Mexico has slower productivity growth, slow- er capital accumulation, faster labor-force growth, or negative terms-of-trade shifts 51 Terms-of-trade relative to the United States. What is the relative contribution of the changes in the Penn World Tables may understate true terms-of-trade changes. Within indus- tries, Mexico and the United States tend to specialize at different ends of the production chain, with the United variables on the right of (26) to the increase States focused more on capital-intensive component proin Mexican migration to the United States?duction and Mexico focused more on labor-intensive product assembly. Such within-industry specialization may not Unfortunately, existing literature offers few be adequately reflected in conventional measures of the answers to such a question. For a prelimi- terms of trade, which fail to account for the fragmentation production. nary take on the data, figure 11 plotsof This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 905 Figure 11. Mexico-U.S. Relative Incomes, Prices, and Terms of Trade To characterize changes in relative labor supplies in Mexico and the United States, figures 12 and 13 show the relative size of the working-age population in the two countries over the period 1970 to 2000. I take the population of Mexican nationals to be the sum of individuals born in Mexico residing in either Mexico or the United States, which I then divide into age cohorts by gender. To examine a population of U.S. natives who are likely to be substitutable in production with Mexican labor, I restrict U.S. age and gender cohorts to be individuals with a high-school education or less. Over the 1980-to-1990 and 1990-to-2000 olds. By 2000, Mexico's supply was larger than the U.S. supply. The change in relative supply occurred in part because a large cohort of young Mexicans entered the labor force in the 1980s and 1990s and in part because more U.S. natives now continue their education beyond high school. The increase in relative labor supply is especially strong among women, for whom skill upgrading in the United States has been most pronounced (Katz and David H. Autor 1999). As further evidence of the contribution of relative labor-supply changes to Mexico-U.S. migration, figure 14 plots the stock of periods, there is a dramatic increase in theMexican immigrants in the United States supply of young Mexican nationals relativeagainst the Mexico-U.S. relative supply of to the supply of young less-educated U.S.labor by gender and age cohort over the natives. In 1980, Mexico's supply of 20-291970-to-2000 period (where I use five-year year olds was just over half the size of theage cohorts to increase the sample size). I U.S. population of less-educated 20-29 yearrestrict U.S. natives to be either those with a This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �906 Journal of Economic Literature, Vol. XLIV (December 2006) Figure 12. Mexico-U.S. Relative Male Population, 1970-2000 Figure 13. Mexico-U.S. Relative Female Population, 1970-2000 This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 907 Figure 14. Mexico-U.S. Relative Labor Supply, 1970-2000 This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �908 Journal of Economic Literature, Vol. XLIV (December 2006) high-school education or less or those with considerable micro-level evidence that less than a high-school education, which migration networks affect individual migra- roughly approximates a plot of ln(Iht) against ln(LOht/L1ht) in (26).52 Across cohorts and time, tion decisions, there are few estimates of the there is a strong positive correlation between impact of these networks on aggregate labor flows. Thus, the literature provides little Mexico-U.S. migration and the relative supply of Mexican labor. Figure 14 gives sug- growth, population growth, skill upgrading, gestive evidence that increases in the relative supply of Mexican labor contribute to migration abroad, presumably by pushing down Mexican wages. Surprisingly, the determinants of wage differences between the United States and guidance for thinking about how GDP or other shocks to national economies affect aggregate labor flows from Mexico to the United States. 3.5 Summary Differences in earnings between the Mexico and their link to Mexico-U.S. migra- United States and Mexico are one factor that tion flows are largely unexplored topics. A contributes to Mexico-to-U.S. migration. cursory glance at the data suggests that a Consistent with previous migration research combination of slow growth in Mexico's in many other contexts, labor outflows from economy and rapid growth in Mexico's labor Mexico increase when U.S.-Mexico income force have contributed to rising labor out- or wage differences increase. Illegal migra- flows in recent years. However, many details tion flows, measured either using survey data about this story are unknown. In the last two on migrant-sending communities in Mexico decades, Mexico has undertaken major eco- or apprehensions at the U.S.-Mexico border, nomic reforms, which may have changed the are quite responsive to relative income distribution of factor rewards or the distri- changes. Access to migration networks bution of regional incomes in a manner that also appears to facilitate cross-border labor increased the incentive to emigrate.53 The flows. migration function in (26) suggests the link The literature has made little progress in between relative factor supplies and migrantidentifying how the underlying determinants flows depends on the substitutability ofU.S.-Mexico wage differences affect binaMexican and U.S. labor, which we know lit- tional migration. Another unresolved issue is tle about (Borjas 2003).54 And, while there iswhy, given large income differences between the United States and Mexico and small 52 To keep this approximation as literal as possible, I observed costs in crossing the border illegally, define the supply of Mexican labor to be residents of migration flows are not larger. There appear Mexico and the supply of U.S. labor to be the sum of Mexican immigrants in the United States and less-educated U.S. natives (where both variables are measured by age and to be unobserved sources of migration costs (or heterogeneity in the perceived benefits of gender cohort). migrating abroad) that are important enough 53 These reforms include a liberalization of foreign trade and investment (unilaterally over the period 1985 to 1989 to impede many individuals from leaving and trilaterally with Canada and the United States under Mexico. These unobserved costs appear to be the North American Free Trade Agreement in 1994), the especially large for less-educated individuals, privatization of state-run enterprises and the deregulation of industry (in the late 1980s and early 1990s), and changes who have relatively low migration propensiin the land-tenure system (in the early 1990s, which priva- ties despite having relatively large apparent tized land previously held by rural cooperatives). Ernesto Aguayo Tellez (2005) finds that in the 1990s individuals from communities more exposed land reform were more likely to migrate internally in Mexico. 54 Patricia Cortes (2005) and Gianmarco I. P. Ottaviano returns to migration. 4. Policies to Control Illegal Immigration and Giovanni Peri (2005) provide recent evidence that in Many government policies the United States low-skilled native and foreign labor are less than perfect substitutes. determining the level of illegal This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms play a role in immigration. �Hanson: Illegal Migration from Mexico to the United States 909 to enforcement, seen in figure 5, suggests Some policies, including the enforcement of international borders and the monitoring of hiring practices by employers, affect illegal that the lack of success is not for want of effort. immigration directly. By changing the intensity with which they enforce borders or mon- ture, rising illegal immigration tends to be itor employers, authorities effectively interpreted as a result of policy failure regulate the inflow of illegal immigrants from abroad. In the United States, enforce- (Peter Andreas 2000; Massey, Durand, and ment policies are under the control of feder- al government agencies and can be changed frequently, even on a day-to-day basis. Other policies, including quotas for permanent or temporary legal immigration, the rights of immigrants to draw on public assistance, and minimum-wage requirements, affect illegal inflows indirectly through their impact on In political science and sociology litera- Malone 2002; Cornelius et al. 2004; Cornelius 2005). Due to political con- straints, the United States has focused on border enforcement, rather than monitoring U.S. employers, which appears ill-suit- ed to curtail unauthorized entry in a country that shares a 2,000-mile long land border with a poor neighbor. Further, the United States has chosen to concentrate migration. In the United States, these and enforcement resources in border cities, leaving less populated corridors largely Congress and tend to change slowly over unpoliced through which illegal immigrants continue to enter the country in large num- time, which has made their impact on illegal bers. Are U.S. enforcement policies inef- the expected reward from unauthorized other such indirect policies are set by immigration difficult to gauge. In this section, I examine U.S. policies to fective? And, if so, why have U.S. enforcement policies, as these have been the immigration authorities chosen these policies? I deal with research on the first question in this section and on the second Theoretical literature considers the incen- subject of most academic research. question in the next. Each year, the U.S. Congress appropriates tives of countries to restrict illegal immigra- funds to the Department of Homeland Security (DHS) for enforcement of U.S. control illegal immigration. I focus on tion and the political economy of impediments to labor inflows from abroad; empirical literature examines the impact of borders, which falls under U.S. Customs and Border Protection, and for enforcement of enforcement policies on migrant flows. immigration laws in the U.S. interior, which falls under U.S. Immigration and Customs Enforcement. (Prior to 2002, both border There is little work on policies that restrict labor outflows, as such policies do not exist in Mexico and are uncommon except in highly authoritarian regimes. 4.1 U.S. Border and Interior Enforcement Policies There is considerable academic and policy interest in the economic impact of U.S. actions to prevent illegal immigration. The increase in the stock of illegal immigrants in the United States, evident in table 1, indicates that U.S. enforcement efforts have and interior enforcement belonged to the now-defunct INS in the U.S. Department of Justice.) The U.S. Border Patrol, the primary agency responsible for border enforcement, is part of U.S. Customs and Border Protection. Border Patrol officers may be deployed to linewatch duty, in which they attempt to apprehend unauthorized immi- grants at the U.S.-Mexico border; to entry points along the U.S.-Mexico and not succeeded in stopping illegal entry from U.S.-Canada borders, at which they monitor Mexico or other countries. The concomi- tant increase in U.S. resources devoted pedestrian and vehicular traffic entering the United States; or to traffic checkpoints along This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �910 Journal of Economic Literature, Vol. XLIV (December 2006) major highways inland from border cross- ings, at which they conduct inspections. Some of these activities (linewatch duty) are And, of apprehensions of Mexican illegal aliens, less than 1 percent occurred at U.S. worksites. The vast majority of Mexican more oriented toward preventing illegal nationals apprehended were "seeking immigration, while others (traffic checks) are more oriented toward preventing the smug- employment," which generally means they were caught trying to cross the U.S.-Mexico gling of contraband. The broad scope of border illegally. Border Patrol activities suggests that DHS officials (and INS officials before them) have policy output. How might one measure sufficient discretion in allocating resources to allow them to vary the intensity with which enforcement, the primary inputs are officer they enforce borders against illegal immigra- tion. Discretion creates an opportunity for political pressure to influence enforcement activities over short time horizons (as well as through the more-protracted congressional appropriations process). The detection of illegal immigrants in the U.S. interior falls under U.S. Immigration and Customs Enforcement (ICE). The activities of ICE agents (and of INS agents before Apprehensions of illegal immigrants are a inputs into enforcement policy? For border hours the Border Patrol spends policing the border and capital expenditure on enforcement infrastructure. In figure 5, between 1990 and 2003 officer hours devoted to bor- der enforcement increased by 3.8 times, from 2.5 million to 9.5 million. For interior enforcement, measures of policy inputs are more difficult to obtain. Those that are avail- able suggest immigration authorities devote a relatively small share of their resources to the creation of DHS) include attempts to the U.S. interior. Between 1999 and 2003, apprehend illegal immigrants at U.S. worksites, investigations of international smug- the number of man hours ICE agents devot- gling operations, and prosecutions and ed to worksite inspections declined from 480,000 (or 9 percent of total INS agent deportations of noncitizens who have been hours) to 180,000 hours (or 4 percent of total convicted of a felony in the United States. As ICE agent hours) (GAO 2005). Thus, in discretion in how ICE agents are deployed, 2003, U.S. immigration authorities devoted fifty-three times as many officer hours to with the Border Patrol, DHS officials have allowing them to vary the intensity of interi- linewatch enforcement as to worksite or enforcement against illegal immigration at the local, regional, and national level. worksite enforcement is that few U.S. Boeri, McCormick, and Hanson (2002) enforcement. One consequence of low employers who hire illegal immigrants are detected or prosecuted. The number of U.S. document that U.S. immigration authorities apprehend far more illegal immigrants at U.S. borders than in the U.S. interior. Table hiring unauthorized workers was only fifteen 7, which shows "deportable aliens" located by U.S. immigration authorities from 1992 two in 1998 and to zero in 2004. A recent to 2004, updates their data. (Deportable aliens include, primarily, apprehended illegal immigrants, and, secondarily, legal immi- grants subject to deportation because they have committed a criminal offense.) Over the period, 93 percent of deportable aliens were located by the Border Patrol, rather than by ICE or INS agents in the U.S. interior. Of those apprehended by the Border Patrol, 97 percent were Mexican nationals. employers paying fines of at least $5,000 for in 1990, which then fell to twelve in 1994 to U.S. General Accounting Office study concludes, "The worksite enforcement program has been a low priority under both the INS and ICE" (GAO 2005).55 55 Further, since September 2001 and the shift in government priorities toward preventing terrorist attacks, the majority of time ICE agents spend on worksite enforcement is devoted to monitoring "critical infrastructure sites," such as airports and power plants (GAO 2005). This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 911 This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �912 Journal of Economic Literature, Vol. XLIV (December 2006) The emphasis of border enforcement over interior enforcement does not appear to be due to illegal immigrants being difficult to locate once they are inside the United prospective migrants appear to have shifted their attempts to enter the United States to the more remote desert regions of the California-Mexico and Arizona-Mexico bor- States. Several U.S. industries, including ders. Border Patrol operations have suc- agriculture, construction, and restaurants and hotels, appear to employ large numbers ceeded in reducing illegal border crossings of unauthorized workers. The U.S. at targeted locations, but not, as the increase in illegal immigration during the 1990s reveals, at other locations along the border. Department of Labor (2005) reports that, Thereof is active debate in policy circles over the period 1999 to 2002, 54 percent the U.S. farm laborers it surveyed about werethe in effectiveness with which U.S. immigration authorities deploy the resources the United States illegally. At harvest time, have in the late summer and early fall, they many of available. In their own defense, authorities these workers are plainly visible at farms in claim that they are overwhelmed California, Texas, Washington, and by current else- levels of attempted illegal immi- where in the western United States. U.S. gration (U.S. General Accounting Office immigration authorities simply choose not to2001). Even with the increase in enforcement resources, authorities suggest they have conduct large-scale raids on U.S. farms, conbeen unable to staunch in the inflow of unaustruction sites, or other places of business where illegal immigrants tend to work. thorized migrants because the number of those attempting to enter the United States Beyond concentrating on border enforceillegally has risen too fast. An opposing line of ment, U.S. immigration authorities target argument is that immigration authorities their efforts at specific locations along the deploy their enforcement resources ineffecborder. Figure 15 plots annual hours the tively for strategic motives (Cornelius et al. U.S. Border Patrol officers spent on 2004). In response to political pressure from linewatch duty by region along the border employers and other groups that benefit over the period 1977 to 2003. Officer hours from illegal immigration, goes the reasoning, increase sharply in the early 1990s in Texas, in the mid-1990s in Western California (San U.S. immigration authorities choose not to enforce U.S. laws against hiring illegal immiDiego), and in the late 1990s in Arizona. These increases reflect successive Border grants and to deploy the Border Patrol in a Patrol operations near specific U.S. border manner that allows large numbers of illegal cities, including El Paso, San Diego, El immigrants to enter the country. Centro, and McAllen (Reyes, Hans P. Has increased U.S. border enforcement in Johnson, and Richard Van Swearingen fact made illegal immigration in the Un2002). Operations involve increased patrols, ited States more costly?56 Hanson and constructing walls and barricades, and Spilimbergo (1999) find that, controlling for mounting electronic surveillance equip- the endogeneity of enforcement, border ment. Figure 16 plots apprehensions by U.S. apprehensions increase as border enforceBorder Patrol officers on linewatch duty by ment increases. This finding suggests that border region. Following the increase in expanded enforcement makes crossing the Border Patrol activities in Western border more difficult (since more of those attempting illegal immigration are being California, apprehensions declined in the region but shortly thereafter increased in nearby Eastern California and Arizona. 56 InFor earlier work on this question, see Espenshade (1994, 1995), Kossoudji (1992), Katharine M. Donato, response to greater border security in San Durand, and Massey (1992), and Massey and Singer Diego, which in the 1970s and 1980s was theFor other recent work, see Manuela Angelucci (1995). (2003). primary illegal entry point along the border, This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 913 Figure 15. Linewatch Enforcement by U.S. Border Patrol Region Figure 16. Linewatch Apprehension by U.S. Border Patrol Region This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �914 Journal of Economic Literature, Vol. XLIV (December 2006) caught at the border), but it does not reveal whether greater difficulty in border crossing deters prospective migrants from attempting to enter the United States illegally. Gathmann (2004) provides more direct evidence of the consequences of expanded border enforcement for migration. She uses MMP data to examine the correlates of coyote prices paid by migrants from Mexico to smuggler prices. Gathmann documents that, as the United States carried out its 1990s bor- der buildup, many migrants shifted from higher-enforcement to lower-enforcement crossing points. We do not know how coyote prices would respond to a borderwide increase in enforcement as the United States has yet to carry out that experiment. Another issue is that MMP data may lead one to underestimate the impact of enforceimpact of coyote prices on migrantment de- on smuggler prices. MMP communities ahave sent migrants to the United States mand for smuggler services.57 The price forin decades. Many families in these commumigrant pays to a smuggler is higher nities years when border enforcement is higher. have long-term relationships with coythe United States and to estimate the otes. Prices coyotes charge long-term But the elasticity of coyote prices with may be less responsive to shocks respect to enforcement is small, incustomers the than prices they charge on the smuggling range of 0.2 to 0.5. During the sample period, a one-standard-deviation increase spotinmarket used by migrants in the rest of enforcement would have lead to an increase Mexico. Hence, in the rest of the country, attempted illegal migration could be more in coyote prices of less than $40; in the mid1990s average coyote prices were $410. The responsive to border enforcement than estimated demand for smuggler services Gathmann's results suggest. The United States has undertaken a masand the individual probability of choosing to migrate to the United States are both sive increase in the resources that it devotes quite responsive to changes in coyote to border enforcement. Yet, the apparent prices. However, given the small enforce- impact of this increase has been modest. observed increase in border enforcement reduced attempted illegal entry at what ment elasticity of coyote prices, the While expanded border enforcement has over 1986 to 1998 appeared to reduce the used to be major crossing points in average migration probability among MMP California and Texas border cities, it appears to have had a small effect on deterrespondents by only 10 percent. Gathmann's results, of course, are condi- ring illegal immigration overall (measured tional on the pattern of border enforcement either in terms of changes in smuggler that was realized over the sample period. By prices or the average probability a Mexican increasing enforcement in some border loca- national migrates to the United States). tions but not others, U.S. immigration author- One possibility is that there are important ities may have (intentionally or not) mitigated nonconvexities in enforcement, such that it the impact of expanded enforcement on only becomes an effective deterrent to ille57 In the estimation of coyote prices, Gathmann (2004) instruments for border enforcement using the drug budg- et of the U.S. Drug Enforcement Agency (DEA). In the estimation of the demand for coyote services, she includes both the smuggler price and the level of border enforcement as regressors, instrumenting for the former with the average U.S. prison term for smugglers (which rises over the sample period) and for the latter again with the DEA gal entry at high levels of resource commitment. This is perhaps the implicit argument of those calling for further expansion of U.S. enforcement efforts.58 Another possibility is that U.S. enforcement strategies are ineffective by design, due to the political economy of immigration control. drug budget. Under the assumption of normality in the 58 See Steven A. Camarota, "Use Enforcement to Ease errors, she is able to control for selection into migration Situation," Arizona Republic, October 23, 2005. (by exploiting MMP data on nonmigrants). This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 915 4.2 Determination of U.S. Enforcement Policies In the absence of economic distortions, the optimal immigration policy would be to have open borders. All else equal, immigration raises national income by allowing countries to use fixed factors more productively, making free immigration welfare maximizing. In practice, countries may choose to restrict immigration because existing distortions, such as the existence of social-insurance pro- grams financed by non-lump-sum taxes, make a departure from free immigration the constrained optimum (Ulrich Scholten and Marcel Thum 1996; Dietmar Wellisch and Uwe Walz 1998; Assaf Razin and Efraim Sadka 1999).59 Or, governments may choose to restrict immigration because they weight the welfare of different individuals unequal- ly, for whatever reason favoring those opposed to immigration (James ForemanPeck 1992). For instance, if the median voter is a worker whose wages would be reduced by immigration, politicians may choose to restrict immigration in order to enhance their future electoral prospects (Jess Benhabib 1996). Policies to address illegal immigration enter a country's choice set when legal immi- grant admissions are subject to binding restrictions and the enforcement of borders against illegal entry is costly. Enforcement costs may be due to the expense of policing the border or to agency costs associated with giving immigration authorities an incentive to implement laws against unauthorized entry. Using Gary S. Becker's (1968) crime- theoretic framework, Ethier (1986a, 1986b) 59 See Hanson (2005) and Hanson, Kenneth F. Scheve, and Matthew J. Slaughter (2006) for evidence on how the public-finance consequences of immigration affect indi- vidual preferences toward immigration in the United States. See Hans-Werner Sinn, Gebhard Flaig, Martin Werding, Sonja Munz and Herbert Hofmann (2003) and Razin and Sadka (2004) for analysis of the public-finance consequences of immigration in welfare states similar to the European context. derives conditions under which border enforcement raises national income. In the absence of migration, wages in a home coun- try exceed wages in a foreign country. Restrictions on legal migration prevent legal labor flows from equalizing international fac- tor prices, creating an incentive for illegal migration. The home country interdicts illegal migrants through costly border enforcement. Let the probability of apprehension, 0 g < 1, be an increasing, convex function of expenditure on enforcement, E, which is financed by taxes on skilled labor. The home-country wage for unskilled natives, w, is then related to the alternative foreign-country wage of illegal labor, w*, through the equalization of expected wages that results from illegal migration: (27) w* = g(E)w* + [1 - g(E)]w - k, where k is the cost of migration (which, dis- tinct from Ethier (1986a), I suppose is incurred whether or not an individual is apprehended). Consider the impact of an increase in enforcement on the home-foreign wage gap. Totally differentiating (27) (28) dw The expression in (28) is posi greater enforcement increa foreign wage difference, a apprehension probability is enforcement (g' 2 0), the init positive (absent which ther incentive to migrate), and m (e.g., smuggler prices) are we in enforcement (k' 2 0). In (2 that changes in enforcement af for unskilled workers in the through three channels: by l in the foreign country (great increases foreign labor supply foreign wage), by lowering t that those attempting migrat entering the country (greater means a given level of atte immigration has a smaller This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �916 Journal of Economic Literature, Vol. XLIV (December 2006) home-country's labor supply), and by raising the cost of crossing the border (greater enforcement raises coyote prices). Ethier (1986a) identifies two environ- ments in which interdicting illegal migrants raises national income: the exis- tence of downward wage rigidity in the home-country market for unskilled labor, such that border enforcement lowers use quarterly data for 1980 to 1997 on wages in immigrant-labor-intensive industries in U.S. border states and among less-skilled workers in Mexican border cities. For highimmigrant industries (apparel, textiles, food products, furniture) in California and Texas, they find zero correlation between wages and enforcement of the Mexico-U.S. border in that state. They also find no evidence of a positive effect of border enforcement on the unemployment by reducing illegal immiwages of workers with low-education levels gration; and the presence of market power in the home-country demand for foreign (high-school dropouts or high-school graduates) in border regions of California or labor, such that greater border enforceTexas.61 For Mexico, the impact of U.S. borment lowers the wage the home country enforcement appears to be larger. In has to pay immigrant workers. Inder the absence of these or other distortions, Tijuana, which is the most active crossing enforcement is welfare reducing.60 One implication of Ethier's work is that for border enforcement to be more than a point for illegal immigrants during the sam- ple period, greater U.S. enforcement at the city's border with San Diego is associated with lower wages for less-skilled workers (up resource drain on the home country it must to six years of education). affect outcomes in sending and receiving labor markets. We have already seen evi- There are two quite different inter- dence in Gathmann (2004) that, in Mexico,pretations of Hanson, Robertson, and higher U.S. border enforcement is associat-Spilimbergo's results. One is that that border ed with modestly higher smuggler prices andenforcement deters illegal immigrants, but illegal immigration has a minimal impact on modestly lower migrant outflows. Are there labor markets in U.S. border regions. U.S. measurable impacts of border enforcement in the United States? Hanson, Robertson, border regions may adjust to influxes of illeand Spilimbergo (2002) examine the effectgal immigrants without large changes in wages either through U.S. native workers of border enforcement on wages in U.S. and exiting these regions (Borjas, Freeman, and Mexican border regions. If enforcement impedes illegal immigration, and if illegalKatz 1997) or through border economies shifting toward industries that are relatively immigrants depress wages in the regions in intensive in the use of immigrant labor which they settle, then wages in receiving (Card and Lewis forthcoming). A second (sending) border regions will tend to rise (fall) after an increase in enforcement. They interpretation of their results is that border enforcement has a minimal impact on illegal immigration, consistent with Espenshade 60 Extending Ethier's analysis, Eric W Bond and Tain- (1994) and Massey and Singer (1995). It Jy Chen (1987) introduce international capital mobility, Slobodan Djajic (1987, 1999) puts migration in a dynamic would still be conceivable that illegal immisetting, Subhayu Bandyopadhyay and Sudeshna Champati gration puts downward pressure on wages in Bandyopadhyay (1998) and Helena Gaytin-Fregoso and Sajal Lahiri (2000) consider interactions of trade policy or U.S. border regions but, since border foreign aid and illegal migration, and Alan Woodland and Chisato Yoshida (forthcoming) examine border enforcement where illegal migrants are risk averse. Another dis- 61 Both sets of results hold with or without instrumenting for border enforcement (to control for the INS setting tortion that could potentially justify positive border enforcement in response to economic conditions in U.S. or enforcement is the presence of welfare policies that result in net fiscal subsidies to illegal immigrants (through theirMexican border areas) using data on U.S. political cycles (see note 26) and entry activity at other U.S. international use of public schools, emergency health care, and other boundaries (ports and Canadian border crossings). public services). This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 917 enforcement does not impede illegal immi- labor. Their second assumption has more gration, there would be zero correlation between enforcement and wages in U.S. border regions. Gathmann's (2004) results suggest border enforcement does affect empirical support. Periodic attempts by the INS or ICE to increase interior enforce- migration costs. And it is the case that wages Association, a business lobby representing farmers in the western United States, issued in Tijuana decline following increases in border enforcement, even if wages don't change in California or Texas. The framework in Ethier (1986a, 1986b) is a normative analysis of border enforcement. Turning to positive questions, how does the U.S. Congress decide on the level of funding for immigration control and how do U.S. immigration authorities allocate budgeted resources between enforcement and other activities for which they are responsible? In the context of trade policy, Grossman and Elhanan Helpman (1994) develop a framework in which endogenously determined industry campaign contributions affect import tariffs on foreign goods. Giovanni Facchini and Gerald Willmann ment are met with political opposition. For instance, in 2005 the Western Growers a statement complaining that excessive enforcement was preventing farmers in Arizona from hiring sufficient immigrant labor to harvest their winter lettuce crop.62 In 1998, INS raids of onion fields at harvest time in the state of Georgia prompted the U.S. Attorney General, both Georgia U.S. senators, and three Georgia congressional representatives to criticize the INS for injuring Georgia farmers.63 There is also historical evidence of antienforcement efforts by business groups. In the 1940s and 1950s, the district commissioner of the U.S. Border Patrol in El Paso would routinely issue orders to stop apprehending illegal immigrants during the agricultural harvest season (Calavita 1992). On occasions when (2005) extend the Grossman-Helpman the Border Patrol did increase enforcement model to consider policies on international factor mobility. In their setup, governments activities, Texas farmers often complained restrict factor inflows from abroad through toa their congressional representatives, who per-factor unit tax or quota. Facchini and pressured the INS (through formal written Willmann make two important assumptions communication) to be less aggressive. about the structure of immigration policy:Lax worksite enforcement by the U.S. (1) the receiving-country government capgovernment is indirect evidence that polititures factor tax revenues or quota rents, cal factors influence the intensity with which which it rebates to citizens, and (2) individthe country enforces against illegal immigra- uals are organized according to their factor tion. Hanson and Spilimbergo (2001) search type and lobby the government on immigrafor systematic evidence of such effects. They tion policy. In equilibrium, each factor estimate the sensitivity of border enforcelobby offers the government campaign conment to relative price changes in industries tributions to support stronger (weaker) that use unauthorized immigrant labor restrictions on inflows of factors for which intensively (apparel, perishable fruits and its members substitute (complement) vegetables, in slaughtered livestock, construcproduction. tion). In theory, higher relative prices for Facchini and Willmann's first assumption immigrant-intensive industries would appears to be counterfactual. In the context of illegal immigration, the U.S. and other 62 See Miriam Jordan, "As Border Tightens, Growers See Threat to 'Winter Salad Bowl,'" The Wall Street governments do not collect payments assoJournal, March 11, 2005, p. 1. ciated with factor inflows. On the contrary, the government spends resources on enforcement to impede the immigration of 63 See Mark Krikorian, "Lured by Jobs, Illegal Immigrants Risk Death at Border Crossings," Santa Barbara News-Press, April 25, 1999. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �918 Journal of Economic Literature, Vol. XLIV (December 2006) increase the returns to lobbying for weaker border enforcement. Controlling for macroeconomic conditions in the United States 5. Concluding Discussion In the United States, unauthorized immi- and Mexico, they find that increases in the gration accounts for one-third to one-half of relative product price for an immigrantnew immigrant inflows. Mexico is by far and intensive industry today is associated with away a the largest source country for those decrease in border enforcement six to ten entering the United States illegally. An months in the future. This finding suggestsemerging body of academic research on illeauthorities relax enforcement when the gal migration from Mexico to the United demand for undocumented workers increas- States has made progress on many fronts. es. Enforcement also rises when overall Yet, the literature is at an early stage. labor-market conditions in the United StatesHeightened policy interest in illegal immitighten, which suggests that the U.S. govern- gration, both in the United States and ment raises enforcement when attempted Mexico, suggests the many unanswered illegal immigration is expected to be high. questions, some of which I now highlight, It appears, then, that enforcement soft-will receive attention in the time to come. ens when the specific sectors that use illegal Available measures of the stock and flow aliens intensively expand but not when the of illegal migrants are imprecise, hampered overall demand for labor is high. This is by the unwillingness of official government suggestive of a free-rider problem among agencies to question individuals about their immigration status. With the U.S. populaspecial-interest groups, in which sectors that benefit greatly from lower bordertion of illegal immigrants now exceeding 10 enforcement, such as apparel and agricul- million individuals, one would expect that, at ture, lobby politicians on the issue (and lobby the very least, U.S. government household surveys would conduct postenumeration more strongly when the gains to higher immigration are greater), while sectors thatsurveys that explicitly ask individuals about their immigration status. This would allow benefit modestly are less politically active. researchers to estimate the stock of illegal 4.3 Summary immigrants with much more precision. Of Restricting unauthorized immigration only greater benefit would be incorporating questions about immigration status directly into makes sense for a country that is subject to distortions that would be exacerbated by ille- the U.S. Census of Population or U.S. Current Population Survey. There is precegal labor inflows. By selecting the intensity with which it enforces U.S. borders againstdent for U.S. government surveys including questions about whether a respondent has a unauthorized entry and monitors the employment practices of U.S. business, U.S.legal immigration visa. For over a decade, immigration authorities implicitly determine the U.S. Department of Labor has conductthe level of illegal immigration. The Uniteded surveys of U.S. farm workers in which it States makes stark choices in its enforcement explicitly asks whether an individual is in the United States legally or illegally (e.g., U.S. polices. It heavily polices specific U.S. border cities, maintains a lighter presence in less-Department of Labor 2005). populated areas, and weakly enforces U.S. A handful of data sources provide inforworksites. There has been little research bymation about illegal immigrants from economists on the political economy of poli-Mexico in the United States. Immigrants cies specifically related to illegal immigra-from Mexico, whether legal or illegal, are drawn disproportionately from the middle of tion. The work that does exist suggests that the country's schooling distribution. Over special interest groups are active in attempting to influence U.S. enforcement practices.time, illegal migrants appear to have become This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 919 more likely to be female, to work outside of in consumption, and credit constraints in agriculture, and to settle in the United States financing migration are all potential explana- on a long-term basis. Largely absent in the literature is analysis of the life-cycle behav- tions for the volume and composition of Mexican immigration. Research on these ior of migrants. Many individuals from issues is just beginning to emerge. Mexico first enter the United States as illegal immigrants and over time gain a legal permanent residence visa through sponsorship by a U.S. family member. One would expect that how a prospective migrant responds to changes in U.S. or Mexico economic condi- tions, or the extent to which a migrant already in the United States assimilates into U.S. society, would depend on whether the individual expects to obtain a U.S. green card in the future. Family sponsorship in the granting of entry visas may thus create a direct link between receiving-country policies on legal immigration and the incentive for illegal immigration. Individuals in Mexico with family members in the United States do appear to be more likely to emigrate. However, we do not know whether this represents the effect of migration networks, which lower the cost of migration, or prospects for obtaining a green card, which Over the last two decades, the United States has greatly increased the resources it devotes to controlling illegal immigration. The government has, in particular, beefed up enforcement at specific U.S. border cities. While the United States has criminalized the hiring of illegal immigrants, the government devotes few resources to monitoring U.S. worksites for the emplo ment of unauthorized workers. The net effect of changes in enforcement policy (coupled with changes in U.S. and Mexico economic conditions) has been increasing levels of illegal immigration. There is no formal political economy theory of immigration control that would explain why the United States chooses border over interior enforce- ment.64 The United States appears to be on the verge of granting an amnesty to at least some of the illegal immigrants residing in the country, which would come two decades raise the benefit to migration. after an earlier legalization under the Consistent with research in many other contexts, Mexico-to-U.S. migration flows are correlated with changes in relative incomes Immigration Reform and Control Act. There is also no formal theory that would explain why a country would choose to enact imperfect and costly enforcement against illegal in the two countries. Attempted illegal immigration appears to be particularly responsive to shocks to the Mexican econo- my, with surges in apprehensions at the U.S.-Mexico border coming shortly after downturns in Mexico. Yet, given the large magnitude of U.S.-Mexico wage differences and the small apparent cost of crossing the border illegally, the volume of migration flows from Mexico to the United States is surprisingly low. Also, given the high relative return to education in Mexico, it is puzzling that Mexican immigrants exhibit intermediate selection in terms of their observable immigration today and later grant an amnesty to those that entered illegally. Given the importance of illegal immigration for the U.S. labor market and for U.S. public finances, policies to control labor inflows have been the subject of surprisingly little research. REFERENCES Aguayo Tellez, Ernesto. 2005. "Rural-Urban Migration in the 1990s Mexico: Switching the 'Ejido' for the 'Maquiladora."'" Unpublished. Airola, James, and Chinhui Juhn. 2003. "Wage Inequality in Post-Reform Mexico." Unpublished. skills. One would expect less-skilled Mexican 64 Ethier (1986a) includes a normative analysis of borimmigrants to have the strongest incentiveder versus interior enforcement but takes the government preference to deviate from zero enforcement as given. See and Weiss (2001) for an extension of this analysis and consideration of an amnesty for illegal immigrants. bias to migrate abroad. Unobserved sources of Epstein heterogeneity in migration costs, home This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �920 Journal of Economic Literature, Vol. XLIV (December 2006) Constantijn Panis, and Christine Peterson. 2001. Alanis Enciso, Fernando. El primer programa bracero y el gobierno de Mexico, 1917-1918. San Louis "The Quality of Retrospective Data: An Examination Potosi, SLP: El Colegio de San Luis, 1999. Andreas, Peter. 2000. Border Games: Policing the of Long-Term Recall in a Developing Country." Journal of Human Resources, 36(3): 593-625. U.S.-Mexico Divide. Cornell Studies in Political Benhabib, Jess. 1996. "On the Political Economy of Economy. Ithaca and London: Cornell UniversityImmigration." European Economic Review, 40(9): Press. 1737-43. Angelucci, Manuela. 2003. "U.S. Border Enforcement Blau, Francine D. 1992. "The Fertility of Immigrant and the Net Inflow of Mexican Illegal Migration." Women: Evidence from High-Fertility Source Unpublished. Countries." In Immigration and the Work Force: Bandyopadhyay, Subhayu, and Sudeshna Champati Economic Consequences for the United States and Bandyopadhyay. 1998. "Illegal Immigration: A Source Areas, ed. George J. Borjas and Richard B. Supply Side Analysis." Journal of Development Economics, 57(2): 343-60. Banerjee, Abhijit V., and Andrew F. Newman. 1993. "Occupational Choice and the Process of Development." Journal of Political Economy, 101(2): 274-98. Freeman. A National Bureau of Economic Research Project Report. Chicago and London: University of Chicago Press, 93-133. Blau, Francise D., and Lawrence M. Kahn. Forthcoming. "Gender and Assimilation among Mexican Americans." In Mexican Immigration, ed. Bansak, Cynthia, and Steven Raphael. 2001.George J. Borjas. Chicago: University of Chicago Press and National Bureau of Economic Research. "Immigration Reform and the Earnings of Latino Workers: Do Employer Sanctions Cause Discrimination?" Industrial and Labor Relations Bleakley, Hoyt, and Aimee Chin. 2004. "Language Skills and Earnings: Evidence from Childhood Review, 54(2): 275-95. Immigrants." Review of Economics and Statistics, Bean, Frank D., Rodolfo Corona, Rodolfo Tuirdn, and 86(2): 481-96. Karen A. Woodrow-Lafield. 1998. "The QuantificaBoeri, Tito, Barry McCormick, and Gordon H. tion of Migration between Mexico and the United Hanson. 2002. Immigration Policy and the Welfare States." In Migration between Mexico and the United State. Oxford: Oxford University Press. States, Binational Study, vol. 1, ed. Mexican MinistryBond, Eric W., and Tain-Jy Chen. 1987. "The Welfare of Foreign Affairs and U.S. Commission of Effects of Illegal Immigration." Journal of Immigration Reform. Austin: Morgan Printing, 1-90. International 23(3-4): 31.5-28. Borjas, George J.Economics, 1987. "Self-Selection and the Bean, Frank D., Rodolfo Corona, Rodolfo Tuirin, Karen A. Woodrow-Lafield, and Jennifer Van Hook. Earnings of Immigrants." American Economic 2001. "Circular, Invisible, and Ambiguous Migrants: Review, 77(4): 531-53. Borjas, George J. 1991. "Immigration and SelfComponents of Difference in Estimates of the Number of Unauthorized Mexican Migrants in the Selection." In Immigration, Trade, and the Labor United States." Demography, 38(3): 411-22. Market, ed. John M. Abowd and Richard B. Bean, Frank D., Thomas J. Espenshade, Michael J.Freeman. A National Bureau of Economic Research White, and Robert F. Dymowski. 1990. "Post-IRCA Project Report. Chicago and London: University of Changes in the Volume and Composition ofChicago Press, 29-76. Undocumented Migration to the United States: AnBorjas, George J. 1995. "Assimilation and Changes in Assessment Based on Apprehensions Data." In Cohort Quality Revisited: What Happened to Undocumented Migration to the United States: IRCA Immigrant Earnings in the 1980s?" Journal of Labor and the Experience of the 1980s, ed. Frank D. Bean, Economics, 13(2): 201-45. Barry Edmonston, and Jeffrey S. Passel. Santa Borjas, George J. 1996. "The Earnings of Mexican Monica: Rand Corporation; Washington, D.C.: Immigrants in the United States." Journal of Urban Institute, 111-58. Development Economics, 51(1): 69-98. Bean, Frank D., and Jennifer Van Hook. 1998.Borjas, George J. 1999a. "The Economic Analysis of "Estimating Unauthorized Migration to the United Immigration." In Handbook of Labor Economics States: Issues and Results." In Migration between Volume 3A, ed. Orley C. Ashenfelter and David Mexico and the United States, Binational Study, vol. Card. Handbooks in Economics, vol. 5. Amsterdam, 2, ed. Mexican Ministry of Foreign Affairs and U.S. New York, and Oxford: Elsevier Science, North- Commission of Immigration Reform. Austin: Holland, 1697-1760. Borjas, George J. 1999b. Heaven's Door: Immigration Bean, Frank D., Jennifer Van Hook, and Karen A.Policy and the American Economy. Princeton: Morgan Printing, 511-50. Woodrow-Lafield. 2002. "Estimates of Numbers of Princeton University Press. Unauthorized Migrants Residing in the United Borjas, George J. 2003. "The Labor Demand Curve Is States: The Total, Mexican, and Non-Mexican Downward Sloping: Reexamining the Impact of Central American Unauthorized Populations in MidImmigration on the Labor Market." Quarterly 2001." Pew Hispanic Center Report. Becker, Gary S. 1968. "Crime and Punishment: An Economic Approach." Journal of Political Economy, 76(2): 169-217. Beckett, Megan, Julie Da Vanzo, Narayan Sastry, Journal of Economics, 118(4): 1335-74. Borjas, George J., Richard B. Freeman, and Lawrence F. Katz. 1997. "How Much Do Immigration and Trade Affect Labor Market Outcomes?" Brookings Papers on Economic Activity, 1: 1-67. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 921 Borjas, George J., Richard B. Freeman, and Kevin Lang. 1991. "Undocumented Mexican-Born Cornelius, Wayne A., and Enrico A. Marcelli. 2001. "The Changing Profile of Mexican Migrants to the Workers in the United States: How Many, How United States: New Evidence from California and Market, ed. John M. Abowd and Richard B. 105-31. Permanent?" In Immigration, Trade, and the Labor Freeman. A National Bureau of Economic Research Mexico." Latin American Research Review, 36(3): Cornelius, Wayne A., Takeyushi Tsuda, Philip L. Project Report. Chicago and London: University of Martin, and James F. Hollifield, eds. 2004. Chicago Press, 77-100. Controlling Immigration: A Global Perspective, 2nd ed. Stanford: Stanford University Press. Borjas, George J., and Lawrence F. Katz. Forthcoming. "The Evolution of Mexican Immigrants in the U.S. Cortes, Patricia. 2005. "The Effect of Low-Skilled Workforece, 1900-2000." In Mexican Immigration, Immigration on U.S. Prices: Evidence from CPI ed. George J. Borjas. Chicago: University of Chicago Data." Unpublished. Press and National Bureau of Economic Research. Costanzo, Joe, Cynthia Davis, Caribert Irazi, Daniel Calavita, Kitty. 1992. Inside the State: The Bracero Goodking, and Roberto Ramirez. 2001. "Evaluating Program, Immigration, and the I.N.S. New York: Components of International Migration: The Routledge. Residual Foreign Born Population." U.S. Bureau of the Census Working Paper no. 61. California Health and Welfare Agency. 1989. A Survey Cragg, Michael Ian, and Mario Epelbaum. 1996. "Why of Newly Legalized Persons in California. San Diego: Comprehensive Adult Student Assessment System. Has Wage Dispersion Grown in Mexico? Is It the Cano, Gustavo. 2004. "Organizing Immigrant Incidence of Reforms or the Growing Demand for Communities in American Cities: Is This Skills?" Journal of Development Economics, 51(1): 99-116. Transnationalism, or What?" Center for Comparative Immigration Studies Working Paper 103. DiNardo, John, Nicole M. Fortin, and Thomas 1996. "Labor Market Institutions and the Card, David. 2005. "Is the New ImmigrationLemieux. Really So Bad?" NBER Working Papers, no. 11547. Distribution of Wages, 1973-1992: A Semiparametric Card, David, and Ethan Lewis. Forthcoming. Approach." "TheEconometrica, 64(5): 1001-44. Diffusion of Mexican Immigrants in Djajic, the 1990s: Slobodan. 1987. "Illegal Aliens, Unemployment, Patterns and Impacts." In Mexican Immigration, ed. and Immigration Policy." Journal of Development George J. Borjas. Chicago: University ofEconomics, Chicago 25(1): 235-49. Djajic, Slobodan. 1999. "Dynamics of Immigration Cardoso, Lawrence. 1980.Mexican Emigration to the Control." Journal of Population Economics, 12(1): 45-61. United States, 1897-1931. Tuscon: University of Arizona Press. Donato, Katharine M., Jorge Durand, and Douglas S. Carrington, William J., Enrica Detragiache, and Tara Massey. 1992. "Stemming the Tide? Assessing the Vishwanath. 1996. "Migration with Endogenous Deterrent Effects of the Immigration Reform and Moving Costs." American Economic Review, 86(4): Control Act." Demography, 29(2): 139-57. 909-30. Duncan, Brian, and Stephen J. Trejo. Forthcoming. Press and National Bureau of Economic Research. Chiquiar, Daniel. 2003. Essays on the Regional "Ethnic Identification, Intermarriage, and Implications of Globalization: The Case of Mexico. Unmeasured Progress by Mexican Americans." In Ph.D. Dissertation, University of California, SanMexican Immigration, ed. George J. Borjas. Chicago: Diego. University of Chicago Press and National Bureau of Chiquiar, Daniel, and Gordon H. Hanson. 2005. "International Migration, Self-Selection, and the Distribution of Wages: Evidence from Mexico and the United States." Journal of Political Economy, 113(2): 239-81. Chiswick, Barry R. 1978. "The Effect of Economic Research. Durand, Jorge, William Kandel, Emilio A. Parrado, and Douglas S. Massey. 1996. "International Migration and Development in Mexican Communities." Demography, 33(2): 249-64. Durand, Jorge, and Douglas S. Massey. 1992. "Mexican Americanization on the Earnings of Foreign-Born Migration to the United States: A Critical Review." Latin American Research Review, 27(2): 3-43. Men." Journal of Political Economy, 86(5): 897-921. Cornelius, Wayne A. 1992. "From Sojourners to Durand, Jorge, Douglas S. Massey, and Rene Zenteno. Settlers: The Changing Profile of Mexican 2001. "Mexican Immigration in the United States: Immigration to the United States." In U.S.-Mexico Continuities and Changes." Latin American Research Review, 36(1): 107-27. Relations: Labor Market Interdependence, ed. J. A. Bustamante, C. W Reynolds and R. A. Hinojosa Ojeda. Stanford: Stanford University Press, 155-95. Cornelius, Wayne A. 2001. "Death at the Border: Efficacy and Unintended Consequences of U.S. Immigration Control Policy." Population and Development Review, 27(4): 661-85. Durand, Jorge, Emilio A. Parrado, and Douglas S. Massey. 1996. "Migradollars and Development: A Reconsideration of the Mexican Case." International Migration Review, 30(2): 423-45. Epstein, Gil S., Arye L. Hillman, and Avi Weiss. 1999. "Creating Illegal Immigrants." Journal of Population Economics, 12(1): 3-21. Cornelius, Wayne A. 2005. "Impacts of U.S. Immigration Control Policies on Migratory Epstein, Gil S., and Avi Weiss. 2001. "A Theory of Behavior: The View from Mexican Sending Immigration Amnesties." IZA Discussion Paper no. Communities." Unpublished. 302. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �922 Journal of Economic Literature, Vol. XLIV (December 2006) Review, 89(5): 1337-57. Hanson, Gordon H., and Antonio Spilimbergo. 2001. "Political Economy, Sectoral Shocks, and Border Espenshade, Thomas J. 1994. "Does the Threat of Border Apprehension Deter Undocumented U.S. Immigration?" Population and Development Review, 20(4): 871-92. Enforcement." Canadian Journal of Economics, 34(3): 612-38. Espenshade, Thomas J. 1995. "Unauthorized Immigration to the United States." Annual Review of Hatton, Timothy J., and Jeffrey G. Williamson. 2004. "International Migration in the Long-Run: Positive Sociology, 21: 195-216. Ethier, Wilfred J. 1986a. "Illegal Immigration." American Economic Review, 76(2): 258-62. Ethier, Wilfred J. 1986b. "Illegal Immigration: The Host-Country Problem." American Economic Selection, Negative Selection, and Policy." NBER Working Papers, no. 10529. Ibarraran, Pablo, and Darren Lubotsky. Forthcoming. "The Socioeconomic Status of Mexican Migrant Review, 76(1): 56-71. Facchini, Giovanni, and Gerald Willmann. 2005. "The Families: New Evidence from the 2000 Mexican Census." In Mexican Immigration, ed. George J. Borjas. Chicago: University of Chicago Press and Political Economy of International Factor Mobility." Journal of International Economics, 67(1): 201-19. Feliciano, Zadia M. 2001. "The Skill and Economic Performance of Mexican Immigrants from 1910 to 1990." Explorations in Economic History, 38(3): 386-409. National Bureau of Economic Research. Inter-American Development Bank. 2004. "Sending Money Home: Remittance to Latin America and the Caribbean." IADB Report. Jasso, Guillermina, Douglas S. Massey, Mark R. Foreman-Peck, James. 1992. "A Political Economy Rosenzweig, of and James P. Smith. 2000. "The New International Migration, 1815-1914." Manchester Immigrant Survey Pilot (NIS-P): Overview and New School of Economic and Social Studies, 60(4): 359-76. Findings about U.S. Legal Immigrants at Gathmann, Christina. 2004. "The Effects of Admission." Demography, 37(1): 127-38. Enforcement on Illegal Markets: Evidence from Katz, Lawrence F., and David H. Autor. 1999. "Changes Migrant Smuggling on the Southwestern Border." in the Wage Structure and Earnings Inequality." In Handbook of Labor Economics, Volume 3A, ed. IZA Discussion Paper no. 1004. Gaytain-Fregoso, Helena, and Sajal Lahiri. 2000. Orley C. Ashenfelter and David Card. Handbooks in "Foreign Aid and Illegal Immigration." Journal of Economics, vol. 5. Amsterdam, New York, and Oxford: Elsevier Science, North-Holland, 1463-1555. Development Economics, 63(2): 515-27. Greenwood, Michael J. 1997. "Internal Migration in Kossoudji, Sherrie A. 1992. "Playing Cat and Mouse at Developed Countries." In Handbook of Population the U.S.-Mexican Border." Demography, 29(2): 159-80. and Family Economics Volume IB, ed. Mark R. Rosenzweig and Oded Stark. Handbooks in Kossoudji, Sherrie A., and Deborah A. Cobb-Clark. Economics, vol. 14. Amsterdam, New York, and 1996. "Finding Good Opportunities within Oxford: Elsevier Science, North-Holland, 647-720. Unauthorized Markets." International Migration Review, 30(4): 901-24. Behind or Moving Up? The Intergenerational Kossoudji, Sherrie A., and Deborah A. Cobb-Clark. Grogger, Jeffrey, and Stephen J. Trejo. 2002. Falling Progress of Mexican Americans. San Francisco: 2000. "IRCA's Impact on the Occupational Concentration and Mobility of Newly-Legalized Grossman, Gene M., and Elhanan Helpman. 1994. Mexican Men." Journal of Population Economics, "Protection for Sale." American Economic Review, 13(1): 81-98. 84(4): 833-50. Kossoudji, Sherrie A., and Deborah A. Cobb-Clark. Hanson, Gordon H. 2005. Why Does Immigration 2002. "Coming out of the Shadows: Learning about Public Policy Institute of California. Divide America? Public Finance and Political Legal Status and Wages from the Legalized Opposition to Open Borders. Washington, D.C.: Population." Journal of Labor Economics, 20(3): Institute for International Economics. 598-628. Hanson, Gordon H. Forthcoming. "Emigration, Labor Lucas, Robert E. B. 1997. "Internal Migration in Supply, and Earnings in Mexico." In Mexican Developing Countries." In Handbook of Population Immigration, ed. George J. Borjas. Chicago: and Family Economics Volume 1B, ed. Mark R. and Oded Stark. Handbooks in University of Chicago Press and National BureauRosenzweig of Economics, vol. 14. Amsterdam, New York, and Economic Research. Hanson, Gordon H., Raymond Robertson, and Antonio Oxford: Elsevier Science, North-Holland, 721-98. Spilimbergo. 2002. "Does Border Enforcement Martin, David A. 2005. "Twilight Statuses: A Closer Examination of the Unauthorized Population." Protect U.S. Workers from Illegal Immigration?" Review of Economics and Statistics, 84(1): 73-92. Hanson, Gordon H., Kenneth F. Scheve, and Matthew J. Slaughter. 2006. "Local Public Finance and Individual Preferences over Globalization Migration Policy Institute Policy Brief no. 2. Massey, Douglas S., Joaquin Arango, Graeme Hugo, Ali Kouaouci, Adela Pelegrino, and J. Edward Taylor. 1994. "An Evaluation of International Migration Strategies." NBER Working Papers, no. 11028.Theory: The North American Case." Population and Development Review, 20(4): 699-751. Hanson, Gordon H., and Antonio Spilimbergo. 1999. Massey, "Illegal Immigration, Border Enforcement, andDouglas S., Jorge Durand, and Nolan J. Malone. Relative Wages: Evidence from Apprehensions at 2002. Beyond Smoke and Mirrors: Mexican the U.S.-Mexico Border." American Economic Immigration in an Era of Economic Integration. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �Hanson: Illegal Migration from Mexico to the United States 923 Born Population?" NBER Working Papers, no. New York: Russell Sage Foundation. Massey, Douglas S., and Kristin E. Espinosa. 1997. 10682. "What's Driving Mexico-U.S. Migration? A Reyes, Belinda I. 1997. Dynamics of Immigration: American Journal of Sociology, 102(4): 939-99. Massey, Douglas S., Luin Goldring, and Jorge Durand. 1994. "Continuities in Transnational Migration: An Reyes, Belinda I., Hans P Johnson, and Richard Van Swearingen. 2002. Holding the Line? The Effect of Theoretical, Empirical, and Policy Analysis." Analysis of Nineteen Mexican Communities." American Journal of Sociology, 99(6): 1492-1533. Massey, Douglas S., and Nolan J. Malone. 2002. "Pathways to Legal Immigration." Population Return Migration to Western Mexico. San Francisco: Public Policy Institute of California. Recent Border Build-Up on Unauthorized Imigration. San Francisco: Public Policy Institute of California. Rivera-Batiz, Francisco L. 1999. "Undocumented Workers in the Labor Market: An Analysis of the Earnings of Legal and Illegal Mexican Immigrants in Research and Policy Review, 21(6): 473-504. Massey, Douglas S., and Audrey Singer. 1995. "New Estimates of Undocumented Mexican Migration and the Probability of Apprehension." Demography, Robertson, Raymond. 2000. "Wage Shocks and North Mayda, Anna Maria. 2005. "International Migration: A Panel Data Analysis of Economic and Non-economic Determinants." IZA Discussion Paper no. 1590. Roy, A. D., 1951. "Some Thoughts on the Distribution of Earnings." Oxford Economic Papers, 3(2): 135-46. 32(2): 203-13. McKenzie, David, and Hillel Rapoport. 2004. the United States." Journal of Population Economics, 12(1): 91-116. American Labor-Market Integration." American Economic Review, 90(4): 742-64. "Network Effects and the Dynamics of Migration and Inequality: Theory and Evidence from Mexico." BREAD Working Paper no. 063. Mishra, Prachi. Forthcoming. "Emigration and Wages in Source Countries: Evidence from Mexico." Journal of Development Economics. Scholten, Ulrich, and Marcel Thum. 1996. "Public Pensions and Immigration Policy in a Democracy." Public Choice, 87(3-4): 347-61. Sinn, Hans-Werner, Gebhard Flaig, Martin Werding, Sonja Munz, and Herbert Hofmann. 2003. "EU Enlargement and Labour Mobility: Consequences for Labour Markets and Redistribution by the State Munshi, Kaivan. 2003. "Networks in the Modern of Germany." CESifo Research Report. Sjaastad, Larry A., 1962. "The Costs and Returns of Economy: Mexican Migrants in the U. S. Labor Market." Quarterly Journal of Economics, 118(2):Human Migration." Journal of Political Economy, 70(1): 80-93. Orrenius, Pia M. 1999. "Family Networks, CoyoteSmith, James P 2003. "Assimilation across the Latino Prices, and the Rural Economy in Migration from Generations." American Economic Review, 93(2): 315-19. Western Mexico: 1965-1994." Unpublished. 549-99. Orrenius, Pia M. 2001. "Illegal Immigration and Smith, James P., and Barry Edmonston, eds. 1998. The Enforcement along the U.S.-Mexico Border." Immigration Debate: Studies on the Economic, Federal Reserve Bank of Dallas Economic and Demographic, and Fiscal Effects of Immigration. Financial Review, 1st Quarter: 2-11. Washington, D.C.: National Academies Press. Orrenius, Pia M., and Madeline Zavodny. 2005. "Self-Smith, James P., and Duncan Thomas. 2003. Selection among Undocumented Immigrants from "Remembrances of Things Past: Test-Retest Mexico." Journal of Development Economics, 78(1): Reliability of Retrospective Migration Histories." 215-40. Journal of the Royal Statistical Society: Series A Ottaviano, Gianmarco I. P., and Giovanni Peri. 2005. (Statistics in Society), 166(1): 23-49. "Rethinking the Gains from Immigration: TheoryStark, Oded, and J. Edward Taylor. 1989. "Relative and Evidence from the U.S." NBER Working Deprivation and International Migration." Demography, 26(1): 1-14. Passel, Jeffrey S. 2005. "Estimates of the Size and Stark, Oded, and J. Edward Taylor. 1991. "Migration Characteristics of the Undocumented Population." Incentives, Migration Types: The Role of Relative Pew Hispanic Center Report. Deprivation." Economic Journal, 101(408): 1163-78. Rapoport, Hillel. 2002. "Migration, Credit ConstraintsTaylor, J. Edward. 1987. "Undocumented Mexico-U.S. and Self-Employment: A Simple Model of Migration and the Returns to Households in Rural Papers, no. 11672. Occupational Choice, Inequality and Growth." Mexico." American Journal of Agricultural Economics Bulletin, 15(7): 1-5. Economics, 69(3): 626-38. Rapoport, Hillel, and Fre6dric Docquier. Forthcoming.Trejo, Stephen J. 1997. "Why Do Mexican Americans "The Economics of Migrants' Remittances." In Earn Low Wages?" Journal of Political Economy, Handbook on the Economics of Reciprocity, Giving, 105(6): 1235-68. and Altruism, ed. L. A. Gerard Varet, S. C. Kolm,Trejo, Stephen J. 2003. "Intergenerational Progress of and J. Mercier Ythier. Amsterdam: North-Holland. Mexican-Origin Workers in the U.S. Labor Market." Razin, Assaf, and Efraim Sadka. 1999. "Migration and Journal of Human Resources, 38(3): 467-89. Pension with International Capital Mobility."JournalU.S. Department of Homeland Security. 2004. 2003 of Public Economics, 74(1): 141-50. Yearbook of Immigration Statistics. Washington, Razin, Assaf, and Efraim Sadka. 2004. "Welfare D.C.: Government Printing Office. Migration: Is the Net Fiscal Burden a Good MeasureU.S. Department of Labor. 2005. "Findings from the of Its Economic Impact on the Welfare of the Native National Agricultural Workers Survey (NAWS) This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms �924 Journal of Economic Literature, Vol. XLIV (December 2006) 2001-2002: A Demographic and Employment Profile of United States Farmworkers." Office of the 3/10-Year Ban." Center for Immigration Studies. Wellisch, Dietmar, and Uwe Walz. 1998. "Why Do Programmatic Policy, Research Report No. 9. Rich Countries Prefer Free Trade over Free Southwest Border Strategy: Resource and Impact Migration? The Role of the Modern Welfare State." European Economic Review, 42(8): 1595-1612. U.S. General Accounting Office. 2001. "INS' Issues Remain after Seven Years." GAO-01-842. U.S. General Accounting Office. 2005. "Immigration Enforcement: Preliminary Observations on Employment Verification and Worksite Enforcement Efforts." GAO-05-822T. U.S. Immigration and Naturalization Service. 2001. "Estimates of the Unauthorized Immigrant Winters, Paul, Alain de Janvry, and Elisabeth Sadoulet. 2001. "Family and Community Networks in Mexico-U.S. Migration." Journal of Human Resources, 36(1): 159-84. Woodland, Alan, and Chisato Yoshida. Forthcoming. "Risk Preference, Immigration Policy and Illegal Immigration." Journal of Development Economics. Population Residing in the United States: 1990 to Woodruff, Christopher, and Rene M. Zenteno. 2001. 2000." "Remittances and Microenterprises in Mexico." Vaughan, Jessica. 2003. "Bar None: An Evaluation of Unpublished. This content downloaded from 129.210.115.230 on Fri, 04 Mar 2022 01:54:59 UTC All use subject to https://about.jstor.org/terms � Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Illegal Migration from Mexico to the United States Creator An entity primarily responsible for making the resource Hanson, Gordon, H Subject The topic of the resource Immigration Illegal Immigration United States Immigration Policy Description An account of the resource In this paper, I selectively review recent literature on illegal migration from Mexico to the United States. I begin by discussing methods for estimating stocks and flows of illegal migrants. While there is uncertainty about the size of the unauthorized population, new data sources make it possible to examine the composition of legal and illegal populations and the time-series covariates of illegal labor flows. I then consider the supply of and demand for illegal migrants. Wage differentials between the United States and Mexico are hardly a new phenomenon, yet illegal migration from Mexico did not reach high levels until recently. An increase in the relative size of Mexico's working-age population, greater volatility in U.S.-Mexico relative wages, and changes in U.S. immigration policies are all candidate explanations for increasing labor flows from Mexico. Finally, I consider policies that regulate the cross-border flow of illegal migrants. While U.S. laws mandate that authorities prevent illegal entry and punish firms that hire unauthorized immigrants, these laws are imperfectly enforced. Lax enforcement may reflect political pressure by employers and other interests that favor open borders. Publisher An entity responsible for making the resource available The Journal of Economic Literature Date A point or period of time associated with an event in the lifecycle of the resource 2006–12 Type The nature or genre of the resource Text Journal Article Format The file format, physical medium, or dimensions of the resource PDF Identifier An unambiguous reference to the resource within a given context https://www.jstor.org/stable/30032389 Language A language of the resource English