Economic Outcomes among Latino Migrants to Spain and the United States: Differences by Source Region and Legal Status

Abstract

Using representative national surveys, this paper compares economic outcomes among Latin American migrants to Spain and the United States in the first cross-national comparison using quantitative data. Considering the geographic location and social proximity of each country with respect to Latin America, we detect a critical selection effect whereby the majority of Latin American migrants to Spain originate in South America from middle class backgrounds, whereas most migrants to the United States are Central Americans of lower class origins. This selection effect accounts for cross-national differences in the probability of employment, occupational attainment, and wages earned. Despite differences in the origins and characteristics of Latino immigrants to each country, demographic and human and social capital factors appear to operate similarly in both places; and when models are estimated separately by legal status, we find that effects are more accentuated for undocumented compared with documented migrants, especially in the United States.

For most of the 20th century, out-migration from Latin America was directed overwhelmingly to the United States. Migration from Mexico to the United States began early in the 20th century, and with the exception of a short break during the Great Depression, it has continued ever since. Puerto Ricans joined the northward flow during the 1940s and 1950s, followed by Cubans and Dominicans in the 1960s and 1970s, and Central and South Americans in the 1980s and 1990s (Durand and Massey, 2010). Until the late 1970s Spain was primarily a nation of emigrants, sending workers to Northern Europe and Latin America (Massey, 2008a). With Spain’s entry into the European Union in 1986, however, the country quickly switched from exporting to importing labor, and during the 1990s it emerged as an important immigrant-receiving society, taking in many migrants from former colonies in Latin America as well as North Africa.

Although both Spain and the United States offer Latino immigrants broad opportunities for employment at good wages, they differ in several key respects. Owing to its location in the Western Hemisphere, the United States offers geographic proximity and lower costs of movement. Despite its geographic location, however, the United States is more socially distant than Spain, which – owing to its history of colonization in the Americas – offers Latino immigrants cultural proximity and lower costs of integration. In deciding whether to migrate to Spain versus the United States, migrants therefore face a tradeoff between physical and social distance.

As a result, Latino immigrants from regions that are close to the United States – Mexico, Central America, and the Caribbean – typically respond to the lower costs of movement and move northward, whereas those in more distant regions – countries in South America – respond to the lower social costs of entry into Spain and head across the Atlantic. As a result, we expect Spain to receive comparatively more immigrants from South America than the United States, and given the high costs of migration to both destinations from South America, we expect migrants from South America generally to be more positively selected on the basis of socioeconomic status.

When one mentions “Latino immigrants” to citizens of the United States versus Spain, the term connotes two very different sets of people. To people in the United States, a Latino immigrant is typically a working class person from Mexico, the Caribbean, or Central America; however, for a Spanish audience, a “Latino immigrant” is more likely to be a middle class person from South America or perhaps more recently, a laborer from Ecuador. In addition to differences in nationality and class origins, the two countries also display significant differences in the legal auspices of migration. Although the large majority of Latino immigrants to both countries in recent years entered or stayed without authorization, in Spain a succession of legalization programs has converted most such people to documented status (Lopez, 2007), whereas in the United States the absence of an amnesty since 1986 leaves the vast majority in undocumented status.

However, at the same time, the destination countries have a number of similarities. Anti-immigrant hostilities among the native-born continue to be known within the public sphere (Calavita, 2005; Massey and Sánchez, 2010), the land and sea borders of each country have become aggressively militarized (Massey, Durand, and Malone, 2003; Calavita, 2005), and the immigrant continues to be regarded as both an economic and racial “other” from mainstream society (Calavita, 2005; Massey, 2008b, 2009b). Therefore, although Spain has had a series of amnesties, this may not result in any more integration of immigrants in practice as compared to the United States.

Differences in legal status, national origins, and class background portend possible differences in processes of assimilation in the two countries. To date, however, no study has systematically compared the two migratory flows. Until recently, such a comparison would have been impossible owing to a lack of data, but here we take advantage of two new data sets to compare the characteristics and experiences of Latin American migrants to Spain and the United States. As the vast majority of immigrants in both nations moved for economic reasons, we focus on the process of labor market insertion as our indicator of integration. Specifically, we draw on data from the Encuesta Nacional de Inmigrantes (ENI) of Spain and the New Immigrant Survey (NIS) of the United States to study how the characteristics of Latin American immigrants vary by destination. We then seek to determine how these differences affect employment, occupational attainment, and wage determination in each country. Because this study represents the first known quantitative attempt to compare Latino migrants in both Spain and the United States, these findings are exploratory rather than based on a deductive, hypothesis testing; therefore, we move immediately to the results once the data sources are explained.

SOURCES OF DATA

Our data come from two nationally representative surveys of immigrants conducted in 2003 (in the United States) and 2007 (in Spain). Although both surveys are representative probability samples, the target populations are quite different. The ENI was conducted by Spain’s National Statistical Institute and was designed to represent all foreign born persons aged 16 or more who lived in family households during the period from November 2006 to February 2007. Respondents had to have been in Spain for at least 1 year and intend to remain there for at least 1 year more. The sampling frame was the municipal household registry, a nation-wide registration system that includes all non-institutional dwellings in the country and registers all persons living within those dwellings for purposes of providing basic services. Since registration is required and tied to services, and because the registry has not heretofore been used for immigration enforcement, the ENI sample is generally thought to offer relatively complete coverage of the immigrant population, both documented and undocumented. In the end, the survey carried out in-person interviews with some 15,465 persons, of whom some 4,725 were from Latin America.

In contrast, the United States’ NIS was conducted by a team of researchers based at three private universities (Princeton, Yale, and New York University) and the RAND Corporation, with the project being funded by a consortium of U.S. federal agencies (including the National Institute of Child Health and Human Development, the National Institute on Aging, the National Science Foundation, the Immigration and Naturalization Service, the Bureau of Citizenship and Immigration Services, and the Department of Health and Human Services) along with Pew Charitable Trusts. The NIS is a stratified probability sample of the entering cohort of immigrants who received permanent residence documents during the period from May to November of 2003 (Jasso et al., forthcoming). They were randomly selected from monthly lists of immigrants who had recently received residence visas by the Bureau of Citizen and Immigrant Services. Either a face-to-face or a telephone interview was conducted within 6 months of visa issuance, yielding a sample of 8,573 households surveyed with a response rate of 68%, 2,490 of whom were from Latin America. Respondents were interviewed in whichever language they preferred, Spanish or English.

Thus, whereas the ENI is a national sample of the entire immigrant population residing in Spain at the time of the survey, the NIS is a national sample of a single cohort of immigrants who became legal residents during the study period. Given that the NIS consists of newly documented legal immigrants, whereas the ENI includes all current residents, one might expect the U.S. sample to be skewed toward less experienced immigrants compared with the Spanish sample, but actual differences are not that great because most “new” legal immigrants to the United States have already spent time in the country in one status or another.

A common prior status is that of an unauthorized migrant. Although by definition the NIS is composed exclusively of legal immigrants, it includes many people with prior undocumented experience. According to estimates by Jasso et al. (2008), around a third of new legal immigrants to the United States have prior experience as unauthorized migrants. This figure is much higher for Latin Americans, however, with 43% among immigrants from South America and 45% among those from the Dominican Republic, and figures as great as 65% among immigrants from El Salvador and 74% among those from Mexico. These figures correspond to estimates of the share of undocumented migrants in the overall foreign born population of the United States, which range from about 33% among all immigrants to 50% among those from El Salvador and 60% among those from Mexico (Massey, 2009a).

The two samples are thus quite different but nonetheless similar in many respects. Although the ENI surveys Spain’s entire immigrant population in 2007 and the NIS surveys the cohort of legal immigrants to the United States in 2003, both offer accurate representations of their target populations based on robust sampling designs. Each survey is multi-cohort in the sense that some respondents have lived in the destination country for many years while others have only come recently. The sample size is sufficiently large in both cases to enable detailed statistical estimation, and both surveys include undocumented as well as documented migrants. In the ensuing analyses we explicitly classify migrants by legal status and investigate its role as a factor in labor force insertion, but readers should bear in mind that among immigrants to Spain, the designation refers to current legal status whereas among those to the United States it refers to prior legal status.¹

GEOGRAPHIC ORIGINS OF LATINO IMMIGRANTS

In Table 1, we present the distribution of Latino immigrants to Spain and the United States by country of birth. As can be seen, the share of Latino immigrants coming from the Caribbean is similar in both countries, being 10% in Spain and 12% in the United States; but the shares originating in South and Central America are dramatically different. As expected from our earlier observations, immigrants to Spain are overwhelmingly from South America, whereas those to the United States are overwhelmingly from Central America.² Thus, while 86% of Latino immigrants in Spain are from South America and only 4% from Central America, in the United States 72% of Latino immigrants are from North or Central America and only 16% are from South America.

TABLE 1.

Origin of Latino Immigrants to the United States and Spain by Legal Status in Percent

Country and region	All migrants		Documented migrants		Undocumented migrants
	Spain	USA	Spain	USA	Spain	USA
South America
Argentina	13.88	0.84	12.89	1.32	8.40	0.24
Bolivia	8.04	0.64	4.71	0.97	26.47	0.40
Chile	2.86	0.44	2.46	0.88	4.06	0.08
Colombia	18.33	5.02	19.48	9.04	16.67	0.87
Ecuador	25.02	2.93	26.99	4.83	21.71	1.11
Paraguay	1.48	0.08	0.71	0.18	5.74	0.00
Peru	6.77	4.30	7.83	8.08	2.38	0.87
Uruguay	4.51	0.24	4.42	0.18	3.64	0.16
Venezuela	5.42	1.45	5.82	3.07	2.66	0.08
Total	86.31	15.94	85.31	28.55	91.73	3.81
Central America
Costa Rica	0.06	0.20	0.03	0.26	0.28	0.08
El Salvador	0.25	18.80	0.21	4.92	0.42	31.64
Guatemala	0.17	6.99	0.11	2.46	0.42	11.29
Honduras	0.32	1.53	0.19	1.67	0.98	1.51
Mexico	2.33	42.69	2.51	40.30	1.40	45.31
Nicaragua	0.32	1.85	0.21	1.49	0.98	2.15
Panama	0.21	0.28	0.24	0.61	0.14	0.00
Total	3.66	72.34	3.50	51.71	4.62	91.98
Caribbean
Cuba	5.35	5.70	6.06	7.64	1.54	3.58
Dominican Republic	4.68	5.98	5.13	12.03	2.10	0.64
Total	10.03	11.68	11.19	19.67	3.64	4.22
Total	100.00	100.00	100.00	100.00	100.00	100.00
N	4725	2490	3779	1139	714	1258

Source: New Immigrant Survey (NIS) 2003; Encuesta Nacional de Inmigrantes (ENI) 2007.

The bulk of this huge discrepancy stems from the dominance of Mexicans among immigrants to the United States and their near absence in Spain. Whereas 43% of all Latino immigrants to the United States are from Mexico, only 2% of the Latinos going to Spain are from Mexico. In general, the distribution of Latino immigrant by nationality is greater in Spain than in the United States. The top source country among Latino immigrants to Spain is Ecuador at 25% of the total, followed by 18% from Colombia, 14% from Argentina, 8% from Bolivia, and 7% from Peru. In contrast, for the United States, after Mexico at 43%, the next closest country is El Salvador at 19%, followed by Guatemala at 7%, and the Dominican Republic and Cuba at 6% each. The distribution of Latin American immigrants by national origins is thus more skewed in the United States than in Spain.

The relative dominance of South America in Spain and of Central America in the United States is even more pronounced among undocumented migrants (i.e., those currently undocumented in the ENI and formerly undocumented in the NIS). Some 92% of all undocumented migrants to the United States originate in Central America while 92% of all undocumented migrants to Spain originate in South America. In the ENI, some 15% of the respondents were present without authorization whereas in the NIS 50% reported prior experience in undocumented status.

Among those present in Spain without authorization, the top nationalities are Bolivians (26%), followed by Ecuadorans (22%), Colombians (17%), Argentines (8%), and Paraguayans (6%), a ranking that is somewhat different for documented migrants, who were led by Ecuadorans (27%) followed by Colombians (19%), Argentines (13%), Peruvians (8%), and Venezuelans (6%). Thus, Bolivians and Paraguayans in Spain are disproportionately composed of undocumented migrants compared with other groups. Indeed, the numbers shown in the table imply that 51% of Bolivians and 60% of Paraguayans in Spain are present without authorization.

In the United States, Mexico leads other countries among both documented migrants (40%) and undocumented migrants (45%), but the other countries in the top five differ sharply by legal status. Among undocumented migrants, Salvadorans come after Mexicans (32%), followed by Guatemalans (11%), Cubans (4%), and Nicaraguans (2%). But among documented migrants the ranking after Mexicans is Dominicans (12%), Colombians (9%), Peruvians (8%), and Salvadorans (5%). Thus, South Americans are more widely represented among the ranks of documented Latin American immigrants to the United States and the distribution is closer to that of Spain, though in both cases the distribution of national origins is still dominated by Mexicans.

SOCIAL ORIGINS OF LATINO IMMIGRANTS

Well-developed migratory networks connecting Mexican origin communities with destinations in the United States help mitigate the social costs of migrating northward (Massey and Espinosa, 1997; Massey and Phillips, 1999), and the financial costs of migration are markedly lower for Mexicans than for those traveling from South America The financial costs are likewise relatively low for aspiring migrants from the Caribbean who are separated from U.S. territory by narrow sea straits, and for Central Americans who gain access to a land border by migrating through Mexico. Although networks connecting these regions to the United States are less developed than those emanating from Mexico, they nonetheless exist and are important in reducing the costs of emigration (Massey and Aysa, 2005).

Among aspiring migrants from South America, however, the financial costs of migration to both Spain and the United States are quite high. But the social costs of migration (i.e., commonalities in culture, language, religion, and history) are much lower to Spain (Pellegrino, 2004). As a result, Spain is more likely to be the destination of choice among South Americans with the financial resources to get a visitor’s visa and buy an airline ticket, especially given recent economic dislocations affecting the middle and professional classes across South America (Pellegrino, 2004; Massey and Capoferro, 2006).³ In contrast, we do not expect to observe much class selectivity among migrants to the United States from Central America, the Caribbean, and especially Mexico (Massey and Aysa, 2005). Given the dominance of South Americans to Spain and of Central Americans and Caribbeans to the United States, we therefore hypothesize that Latin American immigrants to Spain will generally be more class-selected than their counterparts to the United States.

This hypothesis is explored in Table 2, which presents selected social and economic characteristics of Latino immigrants surveyed by the ENI and the NIS. In general, our hypothesis is borne out by the data. Although the demographic profile of Latino immigrants to both countries was similar, the class profile was quite different. The typical Latino migrant was 37 years old in Spain and 39 years old in the United States, and roughly the same share were women (57% in the former and 56% in the latter). In terms of education, however, 66% of Latinos in Spain had completed a secondary education compared with only 44% in the United States. Likewise 81% of those in Spain were employed compared with 75% in the United States, and 28% worked in skilled occupations compared with only 21% in the United States.⁴

TABLE 2.

Mean Characteristics of Latino Immigrants to Spain and the United States

Variable	Description	All migrants		Documented migrants		Undocumented migrants
		Spain	USA	Spain	USA	Spain	USA
Employment outcomes
Employed	Currently employed	0.81	0.75*	0.81	0.68*	0.83	0.81
Wage in PPP dollars	Current, average hourly wage for main job (PPP 2007)	5.65	10.37*	6.16	10.86*	4.76	10.70*
Skilled occupations	Including educators, scientists, entertainers, government	0.28	0.21*	0.30	0.30	0.12	0.15
Demographic background
Age	Age at time of interview	37.03	39.44*	37.65	41.95*	32.70	37.30*
Female	Female	0.57	0.56	0.57	0.62*	0.60	0.51*
Married	Married or in civil union/common-law relationship	0.47	0.69*	0.49	0.66*	0.36	0.72*
Children	Number of minor children in residence	0.66	1.00*	0.69	0.74	0.54	1.25*
Human and social capital
Secondary school	Completed secondary II (ENI) or 12 years (NIS)	0.66	0.44*	0.66	0.55*	0.62	0.33*
Experience years	Total number of years in destination country	9.56	8.71*	10.49	4.71*	3.80	12.82*
Parent	Mother or father in Spain or U.S. at time of interview	0.14	0.18*	0.15	0.15	0.06	0.21*
Spouse	Spouse in Spain or U.S. at time of interview	0.42	0.59*	0.44	0.57*	0.29	0.61*
N		4725	2490	3779	1139	714	1258

Source: New Immigrant Survey (NIS) 2003; Encuesta Nacional de Inmigrantes (ENI) 2007.

Notes: PPP, purchasing power parity.

^*Two countries statistically different at p < 0.05.

The main demographic contrasts between the two groups were that migrants to Spain were less likely to be married (47% versus 69%) and had fewer minor children in the household (0.7 versus 1.0). Not surprisingly, since the ENI surveys a resident population while the NIS surveys a single cohort, total years of host-county experience were greater among Latino immigrants to Spain than among those in the United States. However, the difference was not as great as one might expect given the divergent sampling designs. The average respondent in the ENI reported 9.6 years of experience in Spain at the time of the survey while the average respondent in the NIS reported 8.7 years of experience in the United States. In the end, the two samples are thus comparable as the NIS includes many long-term U.S. residents who were not arriving for the first time but were simply adjusting status from a non-resident visa (such as a student or temporary worker) or converting from undocumented to documented status. According to Massey and Malone (2003), two-thirds of all “new” immigrants to the United States had already been to the country at some point or were living here at the time of admission to permanent residence.

Although Latino immigrants in the United States may be “poorer” than those in Spain in terms of human capital, they are “richer” in terms of social capital, which is expected given their longer history of migration as well as their geographic proximity to the United States. Social capital derives from interpersonal connections that link aspiring migrants to people who have experience or currently live in the destination country, and these ties may be mobilized to reduce the costs and risks of international movement (Massey, 1990; Massey and Espinosa, 1997). Whereas 14% of ENI respondents report having a parent present with them in Spain, the comparable figure is 18% for NIS respondents in the United States, and only 42% of ENI respondents had a spouse present compared to 59% among those in the NIS.⁵

This pattern of relatively scarce human capital and abundant social capital among Latinos in the United States compared to those in Spain is accentuated when one considers undocumented migrants. While 63% of undocumented migrants in Spain had completed secondary school, only 33% of those in the United States had done so; but 21% of undocumented migrants in the United States had a parent present and 61% had a spouse in the country with the respective figures only 6% and 29% among undocumented migrants in Spain.

Although undocumented migrants in the United States may exhibit a deficit in general human capital (education) they had an abundance of migration-specific human capital (host-country experience), again consistent with their longer history of migration. Whereas the average undocumented Latino in the United States claimed 12.8 years of U.S. experience, the typical undocumented Latino in Spain only claimed 3.8 years of Spanish experience. The greater access of undocumented migrants in the United States to social and migration-specific human capital seems to offset their lack of access to education, for the share of undocumented migrants working in skilled services in the United States (15%) was slightly greater than in Spain (12%).

Wages were also higher among migrants to the United States compared with migrants in Spain. Each survey measured financial compensation according to the respondent’s preferred unit (i.e., hourly, weekly, monthly, or annually). The total salary was divided by total hours for each pay unit to estimate the hourly wage. The ENI hourly wage was then converted to U.S. dollars based on the exchange rate on January 2, 2007, and then further converted to Purchasing Power Parity over GDP (PPP).⁶ As shown in Table 2, the average Latino in Spain earned $5.65 per hour, and the average Latino in the United States earned $10.37. These differences do not indicate such drastic difference in living standards, however, for immigrants in Spain generally have access to more social benefits than in the United States, particularly health services. Additionally, these estimates do not take into account sub-national variation at a local level.

Among documented migrants, differences in access to social and human capital are diminished and Latino migrants to the two countries are substantially more similar in terms of class origins and wages. Whereas 66% of legal immigrants to Spain had completed secondary education, the share was a substantial 55% among legal immigrants to the United States, and the share working in skilled services was identical at 30% each. Likewise, 15% of legal immigrants in both nations had a parent present, and while the share of migrants with a spouse present was greater in the United States (57%) it was also substantial among those in Spain (44%). In terms of migration-specific human capital, however, the advantage of Spanish immigrants was decisive, with the average legal immigrant in Spain reporting 10.5 years in-country compared with only 4.7 years among those in the United States. In general, the Spain–U.S. wage gap was lower among documented than undocumented migrants, being $4.70 in 2007 PPP dollars for the former but $5.94 among the latter.

DETERMINANTS OF EMPLOYMENT

The foregoing paragraphs reveal clear differences in social and geographic selectivity between Latin American migrants to Spain and the United States. Latino immigrants in Spain come mainly from South America, most have completed a secondary education, work in the service sector, and they either enter with or quickly achieve legal status. In contrast, Latino migrants to the United States come mainly from Mexico, Central America, and the Caribbean, most have not completed a secondary education, work as manual laborers and unskilled service workers, and enter disproportionately in undocumented status. These cross-national differences in selectivity carry significant potential to yield differences in the process of labor market insertion.

Obviously, linguistic issues must be considered in any comparison of labor market outcomes, as lack of ability in a host-country’s language inevitably constitutes a hindrance to economic success. Language mismatch is not a problem for Latino immigrants in Spain, of course, as most are native Spanish-speakers (though a few may be native speakers of an indigenous language). In the United States, in contrast, Latino immigrants are non-native English speakers in an English-language economy, putting them at a distinct disadvantage. Not only will Latino immigrants in Spain experience an easier process of labor market insertion because of their linguistic compatibility, they will also find it easier to convert skills and education received in the sending country into achievements in the host-country economy, given that the language of instruction and learning is the same in both nations.

Undocumented status is also far more stigmatized in the United States than in Spain, given recent legislation that has increased penalties for violating U.S. immigration law (Durand, Massey, and Pren, 2009; Massey, 2011). Opposition to enacting any sort of legalization or amnesty is also greater in the United States than in Spain. The last amnesty in the United States occurred in 1986 while Spain has offered successive rounds of amnesty to its undocumented migrants, with the last being in February through May of 2005 (Lopez, 2007). In general, then, we expect undocumented migrants to be subject to greater levels of exclusion and discrimination in the United States than in Spain. Even though all migrants in the NIS are currently documented, those who are formerly undocumented accumulated much of their prior labor market experience in unauthorized status and have only recently adjusted to legality to enjoy full labor rights in the United States.

In order to consider differences in the process of labor market participation, we estimated logistic regressions predicting the likelihood of employment in Spain and the United States given the demographic and human capital variables shown in Table 2, along with controls for region of origin (sample sizes are insufficient to include country-level controls). Employment is defined by the respondent when asked their employment status.⁷ In order to account for possible clustering of responses owing to unobserved heterogeneity at the regional level, we used STATA’s cluster command at the state level in the United States and at the provincial level in Spain.⁸ Each analysis in both data sets is replicated for all migrants, as well as those with and without documents. Coefficients with bold-face font denote statistically different differences (weighted t-test) among statistically significant estimates between data sets.⁹

The left-hand columns of Table 3 show models of employment estimated for all Latino immigrants to Spain and the United States. Other things equal, Central Americans in both countries are less likely to be employed than South Americans, but the negative effect is greater and more significant in the United States than in Spain. In contrast, Caribbean migrants are equally less likely to be employed than South Americans in Spain and the United States. With respect to demographic characteristics, the odds of working are significantly lower for women in both countries, but the negative effect of being female is greater in the United States than in Spain, a pattern that is different, however, when undocumented migrants to Spain are separately assessed. Here, women are at a greater advantage in procuring employment compared to undocumented men. Age has the expected curvilinear effect in both countries and across both legal statuses, as indicated by the positive effect of age and the negative effect of its square. As is well known, the odds of employment rise through the early labor force ages, reach a peak, and then decline in the older ages.

TABLE 3.

Effect of Selected Variables on the Probability Employment

	All migrants		Documented migrants		Undocumented migrants
	Spain	USA	Spain	USA	Spain	USA
Region (REF = South America)
Central America	−0.046 (0.367)	−0.433 (0.131) **	0.583 (0.445)	−0.325 (0.209)	−0.733 (0.367) *	−1.121 (0.587)
Caribbean	−0.490 (0.196)*	−0.487 (0.199) *	−0.328 (0.187)	−0.537 (0.209) *	−2.270 (0.529) ***	−0.835 (0.645)
Demographic background
Age	0.153 (0.031) ***	0.100 (0.030) ***	0.161 (0.037) ***	0.131 (0.055) *	0.134 (0.080) ***	0.077 (0.033) ***
Age squared	−0.002 (0.000) ***	−0.002 (0.000) ***	−0.002 (0.000) ***	−0.002 (0.001) **	−0.002 (0.001) ***	−0.001 (0.000) **
Female	−0.480 (0.091) ***	−0.672 (0.123) ***	−0.744 (0.112) ***	−0.711 (0.196) ***	0.520 (0.225) *	−0.692 (0.167)
Married	0.836 (0.278) **	0.074 (0.191)	0.869 (0.333) *	−0.210 (0.367)	1.000 (0.556)	0.336 (0.272)
Children	−0.081 (0.069)	−0.040 (0.055)	−0.056 (0.080)	−0.063 (0.086)	−0.194 (0.150)	0.008 (0.063)
Human and social capital
Secondary school	0.027 (0.010)	0.131 (0.118)	0.016 (0.104)	0.100 (0.196)	0.184 (0.205)	0.015 (0.148)
Experience years	0.033 (0.009) ***	0.053 (0.017) **	0.029 (0.008) **	0.100 (0.029) **	0.144 (0.056) *	0.034 (0.031)
Parent	0.138 (0.169)	−0.064 (0.164)	0.095 (0.187)	0.095 (0.168)	0.560 (0.357)	−0.161 (0.185)
Spouse	−0.857 (0.282) *	0.199 (0.223)	−0.958 (0.340) *	0.401 (0.385)	−0.623 (0.514)	0.007 (0.331)
Illegal	−0.207 (0.150)	0.388 (0.118) *
Intercept	−0.985 (0.592)	0.176 (0.475)	−0.943 (0.645)	−0.216 (0.910)	−1.914 (1.415)	1.513 (0.603)
Pseudo R2	0.04	0.10	0.04	0.13	0.08	0.04
N	3451	1765	2890	764	561	1001

Source: New Immigrant Survey (NIS) 2003; Encuesta Nacional de Inmigrantes (ENI) 2007.

Notes:

^*p < 0.05,

^**p < 0.01,

^***p < 0.001, two-tailed.

Standard errors in parentheses.

Bold face type indicates significant difference in coefficients between Spain and U.S. at p < 0.05.

In terms of human capital, we generally expect that having more host-country experience will raise the odds of employment by increasing the potential returns to labor. These expectations are amply confirmed among Latino immigrants to the United States and Spain, among whom each year of experience increases the odds by 3% in Spain [exp(0.033) = 1.03] and 5% in the United States [exp(0.053) = 1.05]. In terms of social capital, the effect of having a social tie to another migrant is expected to be positive to the extent that he or she is able to assist in finding work and to the extent that he or she is a complement rather than a substitute for the migrant’s own employment. Consistent with this reasoning, having a migrant parent neither increases nor decreases the odds of employment among Latino immigrants in Spain or in the U.S., whereas having a migrant spouse decreases the likelihood of working in Spain.

Turning to the effect of documentation itself, we see that undocumented status raises the odds of employment among immigrants in the United States, but has no such effect in Spain. According to our estimates, previously undocumented immigrants in the United States are 47% more likely to be employed [exp(0.388) = 1.47] than those without prior undocumented experience. One way of achieving legal status in the United States, of course, is through employer sponsorship, possibly selecting employed migrants disproportionately into the population of those eligible to be included in the NIS. However, most Latin Americans achieve legal entry through family rather than occupational ties and perhaps more important is the fact that undocumented migrants in the United States have virtually no access to social services or material support in the absence of a job, compelling undocumented migrants either to work or go home.

The middle and right-hand columns show models estimated separately for documented and undocumented migrants in each country. In terms of demographic background and human/social capital, this contrast does not yield any notable changes compared with the overall equation, but the effects of region are quite different. The real story of regional origins between Spain and the United States becomes visible once undocumented migrants are considered separately. Among those without documents, we find no statistically significant differences by region in the United States, whereas in Spain Central Americans have a substantially lower probability of employment [1−exp(−0.733) = 52%] compared with South Americans, and Caribbean migrants display an even greater employment penalty, with the odds of working being 90% lower [exp(−2.270) = 0.10]. These differences most likely reflect the recent upsurge in manual and unskilled undocumented migration and the fact that Central Americans and Caribbeans cannot as readily access South American networks.

OCCUPATIONAL ATTAINMENT IN SPAIN AND THE UNITED STATES

Contingent on employment, we turn now to consider the process of occupational attainment by estimating a logit model to predict the likelihood of holding a skilled job versus an unskilled job. In Table 4, once again we estimate separate models for Latino immigrants in Spain and the United States and for documented and undocumented migrants in each country.

TABLE 4.

Effect of Selected Variables on the Probability of Being in a Skilled Occupation

	All migrants		Documented migrants		Undocumented migrants
	Spain	USA	Spain	USA	Spain	USA
Region (REF = South America)
Central America	1.531 (0.278)***	−0.039 (0.167)	1.575 (0.277) ***	0.351 (0.211)	1.481 (0.527) **	−0.595 (0.428)
Caribbean	−0.020 (0.205)	−0.213 (0.415)	0.008 (0.203)	−0.433 (0.346)		0.854 (0.408) *
Demographic background
Age	0.044 (0.027)	−0.087 (0.021) ***	0.048 (0.029)	−0.041 (0.059)	0.023 (0.156)	−0.093 (0.032) **
Age squared	−0.001 (0.000)	0.001 (0.000) **	−0.001 (0.000)	0.000 (0.001)	−0.001 (0.002)	0.001 (0.000) *
Female	−0.566 (0.082) ***	0.420 (0.145) **	−0.527 (0.097) ***	0.338 (0.157) *	−0.955 (0.297) **	0.501 (0.239) *
Married	−0.601 (0.194) **	0.193 (0.240)	−0.690 (0.228) **	0.240 (0.427)	0.109 (0.387)	−0.265 (0.364)
Children	−0.036 (0.055)	−0.098 (0.046) *	−0.046 (0.060)	−0.258 (0.123) *	0.069 (0.132)	0.054 (0.083)
Human and social capital
Secondary school	1.101 (0.138) ***	1.698 (0.194) ***	1.161 (0.143) ***	2.155 (0.330) ***	0.599 (0.352)	1.355 (0.168) ***
Experience years	0.070 (0.005) ***	0.035 (0.013) **	0.069 (0.006) ***	0.025 (0.011) *	0.110 (0.075)	0.060 (0.019) **
Parent	0.107 (0.136)	0.184 (0.152)	0.152 (0.141)	0.169 (0.216)	−0.752 (0.614)	0.272 (0.188)
Spouse	0.879 (0.190) ***	0.070 (0.316)	0.928 (0.232) ***	0.641 (0.479)	0.620 (0.425)	−0.305 (0.273)
Illegal	−0.655 (0.230) **	−0.654 (0.303) *
Intercept	−2.827 (0.491) ***	−0.858 (0.578)	−2.989 (0.562) ***	−2.424 (0.939) *	−2.464 (2.856)	−0.743 (0.925)
Pseudo R2	0.15	0.14	0.14	0.16	0.09	0.13
N	2981	1457	2520	581	454	876

Source: New Immigrant Survey (NIS) 2003; Encuesta Nacional de Inmigrantes 2007 (ENI) 2007.

Notes:

^*p < 0.05,

^**p < 0.01,

^***p < 0.001, two-tailed.

Standard errors in parentheses.

Bold face type indicates significant difference in coefficients between Spain and U.S. at p < 0.05.

Turning first to our demographic controls we see that women generally experience higher odds of working in skilled jobs than in unskilled jobs in the United States; however, the reverse is true in Spain. Likewise, rising age lowers the likelihood of working in skilled jobs in the United States. Whereas being married and having children has limited effects on occupational attainment in the United States, in Spain being married reduces the odds of holding a skilled job.

With respect to regional origins, in Spain immigrants from Central America are more likely to work in skilled occupations than South Americans. According to our estimates, Central Americans in Spain are nearly four and a half times [exp(1.53) = 4.62] more likely to be in a skilled job. In contrast, and in keeping with our expectation of less skilled workers from Central America migrating to the United States, there is no difference among source regions for the odds of skilled occupation among Latinos in the United States.

In general, this pattern of effects is consistent with the selection scenario we outlined earlier, in which relatively low skilled Caribbean and Central American migrants go to the United States, whereas those with skills and education head to Spain. However, education is strongly and positively related to occupational attainment both in Spain and in the United States with secondary education raising the odds threefold [exp(1.10) = 3.00] for holding a skilled job in Spain and increasing the odds by an even greater amount in the United States [exp(1.70) = 5.46).

In both countries, greater host-country experience increases the likelihood of moving out of unskilled occupations and into skilled occupations. Thus, each additional year of experience raises the odds of working in skilled jobs by 7% in Spain [exp(0.07) = 1.07] and 3% in the United States [exp(0.03) = 1.03]. This higher payoff for years of experience in Spain is significantly higher than in the United States suggesting that greater occupational mobility is available in Spain for Latino migrants. Given the social similarities presented earlier, we would expect this to be the case.

In terms of social capital, having a spouse present in Spain raises the odds of holding a skilled occupation by a factor of 2.4 [exp(0.88) = 2.40], contingent on the fact of being married (which itself reduces the odds of working in skilled jobs). Having a spouse present has no effect on occupational attainment in the United States.

Finally, undocumented status has a sizeable influence on the relative likelihood of working in skilled occupations in both Spain and the United States. In fact, it is an equally strong negative influence, reducing the odds by around 50% in both cases [exp(−0.65) = 0.52]. Despite the negative main effect of legal status, neither the effects of the different variables nor the between-country contrasts change much when the equations are estimated separately by legal status. The only minor shift is that the positive effect of secondary school is enlarged for both receiving countries among documented migrants, but the effect of a secondary education remains significant for undocumented migrants only in the United States.

WAGE DETERMINATION IN BINATIONAL PERSPECTIVE

In the last phase of our analysis, we consider how wages are affected by regional origins, demographic background, human capital, and social capital while holding constant the effect of occupational attainment, which we undertake by using OLS to regress the log of wages (expressed in constant 2007 PPP dollars) on these variables contingent on employment. The results of the exercise are presented in Table 5. Since we are working with logged wages, the coefficients can be interpreted as percentage changes in wage rates attributable to a unit change in the variable under consideration.

TABLE 5.

Effect of Selected Variables on Average Hourly Log Wage

	All migrants		Documented migrants		Undocumented migrants
	Spain	USA	Spain	USA	Spain	USA
Region (REF = South America)
Central America	0.109 (0.057)	0.008 (0.072)	0.159 (0.052) **	−0.028 (0.049)	−0.115 (0.128)	0.081 (0.129)
Caribbean	0.030 (0.035)	−0.098 (0.052)	0.032 (0.037)	−0.108 (0.031) **	−0.179 (0.096)	0.076 (0.100)
Demographic background
Age	0.004 (0.005)	0.022 (0.010) *	0.002 (0.006)	0.018 (0.009)	0.022 (0.017)	0.027 (0.012) *
Age squared	0.000 (0.000)	0.000 (0.000) ***	0.000 (0.000)	0.000 (0.000)	0.000 (0.000)	0.000 (0.000) **
Female	−0.219 (0.017) ***	−0.265 (0.024) ***	−0.223 (0.018) ***	−0.222 (0.041) ***	−0.193 (0.038) ***	−0.293 (0.022) ***
Married	−0.008 (0.038)	0.009 (0.036)	−0.003 (0.049)	−0.067 (0.122)	−0.008 (0.047)	−0.001 (0.036)
Children	0.008 (0.009)	0.025 (0.010) *	0.004 (0.009)	0.058 (0.018) **	0.029 (0.025)	0.018 (0.009)
Human and social capital
Secondary school	0.087 (0.021) ***	0.159 (0.039) ***	0.089 (0.021) ***	0.166 (0.041) ***	0.087 (0.049)	0.134 (0.058) *
Experience years	0.013 (0.002) ***	0.012 (0.002) ***	0.012 (0.002) ***	0.018 (0.003) ***	0.037 (0.008) ***	0.008 (0.002) ***
Parent	−0.015 (0.025)	−0.053 (0.021) *	−0.014 (0.025)	−0.104 (0.052)	0.005 (0.089)	−0.006 (0.027)
Spouse	0.054 (0.049)	0.035 (0.031)	0.043 (0.060)	0.136 (0.123)	0.096 (0.056)	−0.003 (0.031)
Skilled occupation	0.169 (0.026) ***	0.208 (0.045) ***	0.174 (0.028) ***	0.315 (0.057) ***	0.112 (0.069)	0.074 (0.040)
Illegal	−0.116 (0.017) ***	−0.003 (0.025)
Intercept	1.508 (0.104) ***	1.757 (0.224) ***	1.547 (0.105) ***	1.665 (0.190) ***	0.993 (0.288) ***	1.780 (0.313) ***
Pseudo R2	0.20	0.23	0.18	0.30	0.14	0.22
N	2900	1228	2452	492	448	736

Source: New Immigrant Survey (NIS) 2003; Encuesta Nacional de Inmigrantes (ENI) 2007.

Notes:

^*p < 0.05,

^**p < 0.01,

^***p < 0.001, two-tailed.

Standard errors in parentheses.

Bold face type indicates significant difference in coefficients between Spain and U.S. at p < 0.05.

In both countries female immigrants earn less than their male counterparts, about 22% less in Spain and 27% less in the United States, a significant difference that possibly indicates greater gender discrimination in the United States. The classic curvilinear pattern of wages with respect to age is observed among Latino migrants in the United States but not in Spain, where wages are unaffected by age. Although wages earned in the United States rise by 2.5% with each child present, the effect of children is not significant in Spain (though the point estimate of 1% is not statistically different from the United States). Having a parent present lowers wages in the United States by 5% whereas in Spain the effect is negative but not significant (though again the two countries are not statistically different). In terms of region, no significant differences exist when all the factors within the model are taken into account.

Other things equal, occupational attainments bring positive wage returns in both countries, although the size of the effect is significantly larger in the United States than in Spain. Likewise, a secondary education is associated with wage gains in both nations; however, the wage premium is significantly greater in the United States than in Spain, which is curious given the greater transferability to Spain expected for schooling received in Latin America. In the United States, a secondary education is associated with a 16% wage increase, whereas in Spain the gain is only 9%. With respect to host-country experience, we find a similar pattern in that wages increase as years of experience increase. After taking into account the variables in the model, the intercepts remain significantly different, but substantially not that great, meaning that the Spanish–U.S. wage gap observed in Table 2 is partially attributable to differences in the characteristics of migrants rather than differences in the way those characteristics are rewarded.

The main effect of legal status is significantly negative in Spain but not in the United States. Being currently unauthorized reduces wages by 12% in Spain while being formerly undocumented has no effect on the wages that Latino immigrants earn in the United States, a cross-national difference that is quite significant statistically. This difference may simply reflect the different definition of legal status used in each country – currently versus formerly undocumented – but the equations shown in the middle and right-hand side of the table reveal some interesting differences by legal status.

For example, the regional pattern of indifference observed earlier becomes significantly different for Central Americans in Spain among legal migrants. The negative effect of being female emerges across all legal-country categories, although the penalty is significantly greater in the United States for women who were undocumented compared to undocumented Latinos in Spain. We also find an accentuation of positive effects with respect to occupational status. The positive returns to a skilled occupation among documented migrants in the United States are 32% compared to 17% in Spain (a significant difference). Likewise, secondary school education is rewarded across most national and legal categories, and is surprisingly similar in each case as compared to the full sample. In general, undocumented status seems to exacerbate negative effects and mitigate positive effects in the model.¹⁰

Returning to the difference in intercepts between the two host countries, we find that the difference is insignificant among documented migrants, further supporting earlier results that the difference in wages is due more to the variation within the covariates in the model as opposed to fixed country effects. However, curiously, the difference in intercepts balloons among undocumented migrants where Latinos in the United States make almost twice as much as Latinos in Spain. Again, this could be due to a number of factors including the differing sampling frames for the two countries or differences in welfare services that on the surface would depress wages in Spain compared to the United States.

SUMMARY AND CONCLUSION

Although Latin America has a long history of emigration to the United States, international migration from Latin America to Spain is more recent. Indeed, through the 1970s, Spain was still sending migrants to Latin America rather than vice versa. This pattern reversed itself after Spain’s entry into the European Union in 1986, ushering in an economic boom that created a sustained demand for labor that increasingly was met through immigration, including new movements from former Spanish colonies in the Americas. In Spain, in-migration from throughout Latin America began at about the same time, whereas in the United States, Mexico began sending migrants first, followed by Cuba and the Dominican Republic, and then Central and South America.

We argue that owing to the realities of history and geography, for Latin Americans the choice of migrating to Spain versus the United States involves a tradeoff between geographic and social proximity. Nations in Central America and the Caribbean are geographically close but socially distant while those in South America are geographically distant from both nations but socially much closer to Spain. Among all countries in Latin America, Mexico’s situation is unique. Its 3,000 kilometer border with the United States and a century-long history of migration have given rise to well-developed networks connecting sending and receiving communities on either side of the border.

These circumstances set up different patterns of migrant selectivity from different countries in Latin America. Migration from Mexico to the United States is least selective given the low financial costs (owing to geography) and low social costs (owing to networks) involved in making the trip. Migration from other countries in Central America and the Caribbean is more selective, but these nations are still close enough that travel costs are low and migrant networks, though not as extensive as among Mexicans, have now been in place for several decades. Migration from South America to both nations is expected to be most selective in terms of class background given the financial resources needed to qualify for a visa and purchase an airline ticket, but socially the costs are much less for Spain versus the United States as there are fewer cultural barriers and no language barrier.

As a result, we argue Latin American migration to Spain should be predominantly South American and middle class, whereas migration to the United States should be predominantly Central American and Caribbean and from the lower middle and working classes. We confirmed this line of reasoning by undertaking the first-ever comparative analysis – at least to our knowledge – of Latin American emigration to Spain and the United States, drawing on two new sources of representative data. Our study relied on the ENI, a representative survey of foreigners living in Spain in 2007, and the NIS, a representative sample of legal immigrants who received permanent resident status in the United States during 2003.

A cohort sample of arriving Latino immigrants in the United States and a population sample of Latino residents living in Spain would not seem to have much in common, but they are in practice more comparable than one might expect. Both are well-executed cross-sectional surveys that yield representative data for a large sample of Latin American immigrants, and both sources include undocumented as well as documented migrants. In the ENI, undocumented migrants are people currently in unauthorized status, and in the NIS they are people with prior experience as undocumented migrants.

Our systematic comparison of characteristics across the two samples reveals clear differences in social and geographic selectivity among Latin American migrants to Spain and the United States. As expected, Latino immigrants in Spain come mainly from South America, have completed a secondary education, and either enter in or quickly adjust to legal status, whereas Latino migrants in the United States come mainly from Mexico, Central America, and the Caribbean and have not finished secondary school. They work as unskilled workers and enter predominantly via undocumented status.

These differences in socioeconomic and geographic selectivity, combined with very different contexts of reception, raise the possibility of significant differences in patterns and processes of assimilation, which we studied by focusing on the process of labor market insertion. Although Caribbean migrants were less likely to be employed in both countries, they were much less likely to work in Spain as undocumented migrants than similar immigrants without documents in the United States. Likewise, women were less likely to work in both countries, but the penalty for being female was greater in the United States than in Spain. Indeed, in Spain being female was actually associated with a greater likelihood of employment among undocumented migrants. Moreover, whereas the odds of employment varied in curvilinear fashion with age in each setting, the rise and fall of employment with age was steeper in Spain than in the United States. Finally, while host-country experience increased the odds of employment in both countries, it more dramatically increased the odds of working in the United States.

After considering the determinants of employment, we examined the pattern of occupational attainment in each setting among employed migrants. Consistent with our scenario of differential selectivity by region, we found that Central Americans were positively associated with mobility out of unskilled jobs in Spain. In both nations, education and host-country experience increased the odds of having a skilled job, but the effect of schooling was stronger in the United States while host-country experience was generally stronger in Spain. In addition, the effect of education on the odds of holding a skilled service job was greater for documented than undocumented migrants in both countries, indicating the barriers to occupational mobility imposed by unauthorized status.

The effect of legal status was also apparent in the process of wage determination. Among Latino immigrants in Spain, undocumented status reduced wages by around 12%. Although the main effect of legal status was not significant among Latino immigrants in the United States, we did find the negative effects of certain variables were accentuated and the positive effects were diminished among undocumented migrants. Although there are differences in the size of individual effects, the overall pattern of results is quite similar, and differences in wages that were observed in the descriptive data disappeared once background characteristics were controlled.

Thus, although our analyses reveal very different processes of geographic and social selectivity among Latin American migrants to Spain and the United States, the patterns of labor market insertion, occupational attainment, and wage determination are basically similar. In both nations, there are advantages associated with variables such as experience and education and disadvantages associated with being female and undocumented. Generally the factors that work to the advantage of immigrants are less beneficial for those in unauthorized status. In sum, although geographic and historical realities combine to produce Latin American immigrant populations with very different characteristics in Spain and the United States, the way those characteristics are treated in Spanish and American labor markets does not differ much, and controlling for these characteristics, Latino immigrants in both places can expect to achieve similar outcomes.

Naturally, our analysis has important shortcomings. Economic adaptation and social mobility are long-term processes that unfold over time and we are forced here to rely on cross-sectional samples and synthetic cohorts based on time spent in the host country. Moreover, although our findings hint at similar barriers associated with legal status, documentation is measured very differently in the two samples, referring to current status in Spain and prior status in the United States. If legal status in the United States sample were defined on the basis of current rather than past status, one might find sharper differentials.

Finally there is the issue of physical appearance and discrimination. Evidence does seem to suggest that immigrants who are racially different experience an economic disadvantage in both Spain (Sole and Parella, 2003) and the United States (Hirsch, 2008). Although the NIS included an independent measure of skin tone, the ENI did not, making it impossible to investigate this factor comparatively. Immigrants from places such as Argentina, Chile, and Uruguay tend to be very European, whereas immigrants from Mexico and Central America tend to be mestizo, and immigrants from the Caribbean are often of mixed African origins. Differing geographic selectivity between Spain and the United States also implies differing racial selectivity, so that some of the observed disadvantage of being from Central America or the Caribbean could reflect differences in treatment based on skin color or physical appearance, and not regional background per se.

Hence, this analysis should be considered a first step in the comparative study of Latin American immigration to two distinct and very different receiving societies. Despite the foregoing shortcomings, we nonetheless hope this study provides the basis for future comparative studies, not only of wages but of other salient outcomes. Economic integration is but one dimension in a much broader process of assimilation that incorporates social, cultural, and political, as well as economic outcomes.