New Evidence on the Healthy Immigrant Effect

This paper provides new empirical evidence on the contribution of selective migration to the health advantage of immigrants upon arrival to the new destination (i.e. the Healthy Immigrant Effect). It analyses a very interesting episode in international migration, namely the exodus of Ecuadorians in the aftermath of the economic collapse in the late 1990s. Between 1999 and 2005, more than 600,000 Ecuadorians left the country and most of them headed towards Spain. Using administrative data from the Vital Statistics, it compares the health distribution (in terms of birth outcomes) of immigrant children born in Spain to that of non-immigrants in Ecuador and immigrants from other nationalities, and not only to that of natives at destination. These comparisons suggest that positive selection is partly responsible for the health advantage of recent immigrants.


I. Introduction
Questions about the characteristics of those who migrate remain fundamental in immigration research. To evaluate the costs and bene…ts of population movements, immigrants are compared to non-immigrant in the source country and the native population at destination in many di¤erent dimensions (e.g. education, age, risk and entrepreneurial attitudes or health).
The health of immigrants is an issue of concern. Some critical voices argue that migration may represent a burden to the public health system at destination …nanced mainly by natives. The health of immigrants may also be a relevant factor for their integration and assimilation process. For the sending country, the characteristics of those who leave may as well have implications at the aggregate level in terms of, for instance, health and inequality.
A well established regularity is that new immigrants to developed countries such as the US, Canada, and Australia enjoy signi…cant health advantages relative to comparable native-born individuals in these countries. 1 This is known in the literature as the healthy immigrant e¤ect (HIE). The HIE is present among most immigrant groups, even though a large majority come from developing countries with worse life expectancy indicators. There is also evidence that the gap does not respond to socioeconomic di¤erences in terms of education and income as most recent immigrants fall behind the native population in these dimensions.
The literature has highlighted at least three explanations to account for the health advantage of recent immigrants. First, the better health of immigrants may respond to healthier diets, habits and behaviors inherited in the country of origin (i.e. the cultural hypothesis). Second, the migration episode may have a direct impact on health as a result of income shocks or other changes in life style directly related to the movement (i.e. the causal or direct e¤ect of migration). Finally, it may be that only healthy individuals are ready to make their way to a remote and unfamiliar labor market. As a result, individuals in the upper tail of the health distribution are more likely to migrate (i.e. the selective migration hypothesis).
The aim of this paper is to better understand the channels driving the healthy immigrant e¤ect. I will focus on a very interesting episode in international migration, namely the Ecuadorian Exodus in the aftermath of the economic collapse of the late 1990s. Between 1999 and 2005, more than 600,000 Ecuadorians left the country and most of them headed towards Spain rather than the US, a traditional destination for Ecuadorian migration (Bertoli et al. 2011). Taking advantage of some interesting features of this episode, I …nd a health advantage in terms of birth weight and other birth outcomes (i.e. gestational age, incidence of low birth, and death before 24 hours) among the children of new Ecuadorian immigrants in Spain. The comparison to children of non-immigrants in Ecuador and to those of another recent minority group in Spain (i.e. Romanians) suggests that positive selection in health is partly responsible for the health advantage of immigrants from Ecuador.
Health economists argue that birth weights are strongly correlated with a mother's habits during pregnancy and her health. For example, recent research shows that fasting during pregnancy results in lower birth weight and reduces gestational length (Almond and Mzaumder, 2013). There is also evidence that economic crisis have a negative impact on the weight at birth, that is particularly strong among low-income women (Bozzoli and Quintana, 2013). Moreover, the health literature indicates that birth weights are an important marker of an infant's health at birth and as an adult. Using administrative data for Norway, Black et al (2007) show that a 10 percentage points increase in birth weight increases height by 0.57 centimeters at the age of 18; increases the probability of overweight and decreases the probability of under weight by 1 percentage point each; increases the probability of high school completition by 1 percentage point and full-time earnings by a 1 percent. These e¤ects are constant across the birth weight distribution, rejecting the existence of non-linearlities. 2 In this paper I explore the existence and the nature of a health advantage in favor of recent Ecuadorian immigrants in Spain, using administrative data on birth outcomes (i.e. Vital Statistics) for Spain and Ecuador. The paper is structured as follows: the next section provides a brief overview of the literature; section III highlights the main features of the migration episodes analyzed in the paper; section IV describes the data and section V the empirical methodology; section VI discusses the results and section VII presents some …nal remarks.

II. Literature on the Healthy Immigrant E¤ect
Researchers from a wide array of disciplines have studied health di¤erences between immigrants and native-born individuals, mainly in the US, Canada and Australia. Alternative explanations have been proposed to account for the health advantage among recent immigrants. First, it has been suggested that the advantage could respond to the health screening that is part of the migration process in some countries. However, some evidence indicates that admission policies are not the principal determinant of the health gap. For example, Laroche (2000) reports that the percentage of applicants to Canada that are rejected on health grounds is very low and Uitenbroek and Verhoe¤ (2002) argue that the selection by authorities based on health can not explain the lower mortality of Mediterranean immigrants in Amsterdam.
The second explanation is that healthy diets, habits and behaviors in the home country lead to potential immigrants who are relatively healthier than the average person in the recipient country. The hypothesis based on cultural di¤erences is put forward in Abraido-Lanza et al. (1999) who argue that the lower mortality of Latinos in the US results from their more favorable health habits (i.e. less alcohol and cigarette consumption which are the major risk factors for cancer and heart diseases, the most common causes of death for both Latinos and non-Latino Whites).
A third possibility is that the migration episode has a direct impact on an individual's health due to the associated income shocks or environmental changes. Evidence on the causal impact of migration is rather scarce due to the methodological di¢culties involved in estimation. To identify the direct e¤ect of migration the health of immigrants upon arrival to the country has to be compared to what their health would have been had they stayed in their home country. This counterfactual is typically unobserved. An exception is the work by Stillman et al. (2012) where using data from a unique survey compare the health of migrant children who enter New Zealand through a random ballot with children in the home country of Tonga whose families were unsuccessful participants in the same ballots. Their …ndings indicate that migration increases height and reduces stunning of infants and toddler, but also increases BMI and obesity among 3 to 5 years old. The authors argue that changes in dietary habits (i.e. larger intakes of meat, fat and milk) rather than the income gains associated to migration explain the …ndings.
Finally, the better health of recent immigrants could respond to selective migration. There are reasons (and evidence) to suspect that immigrants are di¤erent from those who do not migrate. The literature on selection based on labor market outcomes (wages) and education tend to …nd evidence of positive selection (Chiquiar and Hanson 2005;Rapoport 2007, 2010;Orrenius and Zavodny 2005;Chiswick 1978Chiswick , 1999Chiswick , 2007Belot and Hatton 2008;Grogger and Hanson 2008), though some evidence of negative selection has also been reported for Mexico (Borjas 1987; Fernández-huertas Moraga 2011). If positive selection in productive skills dominates migration movements, given the strong correlation between income and health, positive selection in health should also be observed. Indeed if immigrants are selected from the high end of the income distribution in their home countries, they are likely to have access to better diets, better access to clean water and sanitation, less exposure to environmental risks and better child/maternal health care. Even in the absence of selective migration in skills, positive selection in health is also expected if immigrants are forward looking (i.e. make current behavioral choices that emphasize future health at the expenses of current time/e¤ort) or if sick individuals are more 4 reluctant to leave the origin to make his or her way in an unfamiliar labor market. 3 A major drawback in previous studies is that most of the conclusions are based on comparisons between immigrants (generally legal) and natives at destination. Such a comparison does not allow disentangling the contribution of selection from that of healthy habits or any direct e¤ect of migration on health. There are a couple of recent exceptions that compare the health of non-immigrants and immigrants before the movement occurs. The study by Rubalcava et al. (2008) employs longitudinal data from the Mexican Family Life Survey to compare emigrants from Mexico to the US to similar non-emigrants. The results suggest some evidence of positive selection in terms of physical health outcomes. In contrast, Stillman et al (2009) using data from Tongan potential immigrants and non-immigrants …nd that individuals in poor mental health are more likely to apply to migrate.
Data to compare immigrants and non-immigrants in the sending country are rather scarce since most migrants originate from developing countries without tradition on data collection. In this paper, I employ the Vital Statistics in Ecuador and Spain to compare the birth outcomes of immigrant children in Spain to that of non-immigrants in Ecuador and natives in Spain. Due to con…dentiality issues, the same individual cannot be identi…ed in the Vital Statistics of the two countries, and therefore the comparison before the movement occurs is not possible. I then propose an alternative strategy that relies on empirically testing an important prediction of the migration model in Borjas (1987). According to this model, individuals only migrate if the ben-e…t of the move is larger than its associated cost. Since the cost of migrating increase with distance, immigrants from more remote areas should be more positively selected than those from neighboring ones. The large and diverse migration wave to Spain during the last decade o¤ers an excellent scenario to investigate this conjecture. Since the early 2000s immigrants from di¤erent origins arrived to Spain attracted by the growing economy and the many job opportunities, in particular in the construction sector. The similarity between Ecuadorian and Romanian immigrants in many dimensions (e.g. socioeconomic characteristics and linguistic barriers) but geographical origin allows me to test whether selection is inversely proportional to geographical distance. For robustness, the comparison is also extended to children of Colombian and Bulgarian immigrants born in Spain.
The focus is on immigrants from Ecuador as the migration episode between the Latin American country and Spain has some other interesting features that make it an interesting case study to better understand the healthy immigrant e¤ect. First, since the bulk of Ecuadorian immigrants moved to Spain between 1999 and 2003, the sorting of immigrants across di¤erent countries are not likely to distort the results. Second, since 2001 the Vital Statistics in Spain contain information on immigrants irrespective of the legal status. A change in the law granted all registered individuals access to the public health and education system. This provided incentives to both legal and illegal immigrants to register their newborns and to appear in the Vitals Statistics. 4 Finally, immigration to Spain is a recent phenomenon, and most of the foreign-born in the early 2000s were likely to be recent immigrants. Hence, the e¤ect of acculturation or assimilation on the health gap in the early 2000s (if any) is likely to be small.

III. Two large migration episodes
Between 2000 and 2007 Spain received an impressive in ‡ow of immigrants -approximately 500,000 per year. The share of the foreign born population shifted from about 3% in the late 1990s to more than 16% by 2007. Table 1 display the stock of immigrants in Spain during the 2000s recorded in the Local Population Registry.
The composition of migrants changed over time. While in the 1990s migrants originated mainly from the EU-15 countries, they were rapidly overtaken by South Americans and migrants from the EU enlargement member states. The largest minority groups in Spain during the last decade were: Moroccans, Romanians and Ecuadorians. While the …rst group had a large tradition in the country, Romanians and Ecuadorians massively arrived in the early 2000s (see Table A1 in the Appendix).

The Ecuadorian Exodus
As a result of the economic and …nancial crisis Ecuador collapsed in 1999. This represented an important push factor for abut 600,000 individuals who over a period of a few years (1999)(2000)(2001)(2002)(2003)(2004)(2005) left from a country with a population of 12.7 millions. A unique feature of this migration episode is that the US and Spain received about 80 to 90% of all Ecuadorians. Moreover, the number of Ecuadorians that migrated to Spain was roughly 3 times larger than the corresponding ‡ow to the US. Bertoli et al (2011Bertoli et al ( , 2013 argue that the lower cost of migrating explains the huge exodus towards the low income country. The migration policy in Spain was particularly attractive for Ecuadorians. Since 1963 a visa waiver program allow them to enter as a tourist for a period of up to three months. Those who wished to migrate could simply overstayed the threemonths period, became undocumented workers, and wait for one of the frequent amnesties in the early 2000s to legalize their status. 5 The lax Spanish immigration policy substantially in ‡uence the location choices of immigrants. According to the calculations in Bertoli et al (2011) the Ecuadorian population in Spain increased from 76,000 individuals before 2000 to 457,000 in 2005, and represented 12 percent of immigration ‡ows to Spain between 1999 and 2005. 6 The visa waiver program was terminated in August 2003. After this date, Ecuadorian migrants needed a visa to enter any EU member state. The in ‡ows of Ecuadorians to Spain dropped sharply immediately after the requirement, and the United States became again the main destination (Bertoli et al 2011). Table 1 also shows the stabilization in the stock of immigrants from Ecuador during the second half of the 2000s.
A salient feature of the Ecuadorian exodus is that most of those who moved in the aftermath of the crisis headed towards Spain. Thus the analysis of the birth outcomes of immigrants in the early 2000s in Spain should be weakly a¤ected by sorting across countries. Bertoli et al (2011) investigate the selection and sorting of Ecuadorian immigrants in terms of productive skills (education and wages) during this period. They …nd that immigration to Spain is gender balanced and some evidence of negative selection in education (particularly among men).

The Romanian Experience
Before the collapse of the Romanian communist regime in December 1989, the of-…cial statistics reported very low numbers of emigrants, who were mostly political refugees and/or relatively highly educated Romanians of another ethnicity (Jews, Germans and Hungarians). By the mid 1990s a new pattern of labor migration emerged against the background of a slow pace of economic restructuring which resulted in a large decline in GDP, high in ‡ation, mass layo¤s, decreasing real wages and rising unemployment. 7 The migration out ‡ows sharply increased in 2001 when Schengen visa restrictions were lifted, and Romanian citizens gained right to free circulation within the Schengen area. By 2010, Romanian immigrants were the most represented foreign group in both Spain and Italy. These two countries each hosted around 40% of Romanian immigrants in Europe, followed by Germany (5.72%), the UK (3.78%), Austria (2.23%), France (2.3%), Portugal (1.52%), Greece (1.73%) and Belgium (1.24%) (Andrén and Roman, 2013). 2001 and 2005). 6 The same authors estimate that the Ecuadorian population in the US increased from 272,000 individuals before 2000 to 394,000 in 2005, and represented 1.3 percent of immigration ‡ows in the US during this period. 7 De-industrialization led to a decrease of industrial employment by almost 3 million jobs and particularly a¤ected younger and older workers, who were less likely to …nd new employment opportunities (Voicu 2005

7
The movements of Romanians towards European countries have been characterized by strong selection patterns (Ambrosini et al., 2012). A …rst group of strictly positive selected immigrants has characterized migration ‡ows to traditional immigration countries (US, Canada and Australia). These ‡ows are rather small but persistent and include a signi…cant share of young people who migrate for educational purposes. A second group of neutral average selected immigrants moved to the continental European countries over the 1990s: Germany, Austria and France. Finally, towards the end of the 1990s and the early 2000s, large ‡ows of Romanian migrants arrived in Mediterranean countries, mainly Spain and Italy. These ‡ows were characterized by negative selection in terms of productive skills.

IV. Data
This study employs birth outcomes, in particular weights, as a measure of an individual's health. The birth weight is the body weight of a baby measured at most one hour after birth. While it may su¤er from measurement error, it is not a¤ected by the biases inherent to self-reported health questions employed in other studies. A main problem with reported assessments of one's own health is that it depends on the reference group. If the group is not stated, comparisons across individuals become di¢ cult (King et al. 2004). This is particular relevant for immigrants whose comparison group may change with the process of assimilation. The use of the prevalence rate of some diseases (i.e. diabetes, heart diseases, asthma or diseases of the lung) is also subject to criticism as the lower incidence of chronic diseases reported by foreigners may simply result from their less frequent contact with western medial diagnostics.
This paper employs the information in the Vital Statistics of Ecuador and Spain. The Vital Statistics for Romania are not available, and the information regarding the birth outcomes of Romanian immigrants is collected from the Vital Statistics in Spain. The information in the Spanish and Ecuadorian Vital Statistics corresponds to all births in the Local Population Registry. In both countries, registration is the administrative procedure to legalize a vital event. 8 Hence, the Statistics give coverage to all legalized births occurred in both countries. 9 The analysis is restricted to the early 2000s, and in particular to 2001-2003. There are several reasons to justify this time constraint. First, the Vital Statistics do not contain information on years since arrival and it is therefore not possible to account 8 In order to register a birth, the parents or the legal representative of the child have to present a document with statistical information on the birth outcome (Informe Estadístico del Nacido Vivo in Ecuador, …gure 1A or Boletín Estadístico del Parto in Spain, …gure 2A). 9 As discussed, immigrants in Spain since 2001, independently of their legal status, have strong incentives to appear in the Local Population Register to have access to the public health and education system and to prove residence in Spain for future amnesties.  The table also shows the increase in the birth rate to Romanian immigrants, the largest minority group in Spain in the late 2000s. Table 3 displays the mean weight in grams for the period 2000-2005 by nationality in Spain. For a 5% of the births the information on weight is not recorded, and these observations are excluded. Following previous work on the determinants of birth weight, I focus on mothers aged 15-49, exclude multiple births and those newborns whose weight was either under 500 grams or above 9,000 grams. The table indicates that newborns to foreign mothers are about 50 to 80 grams heavier than those born to natives (in 2001, 3,292 grams for immigrants and 3,237 for natives). By foreign nationality, the heaviest babies are born to Ecuadorians (3,273 grams) and then Romanians (3,219 grams). This ranking is not consistent with the aggregate health statistics reported by the World Bank in the origin countries (see Table 4). Accordingly, babies born in Romania are, on average, heavier than those born in Ecuador (3,186 grams in Romania and 3,102 in Ecuador).
The second data source employed is the Vital Statistics for Ecuador, from the Instituto Nacional de Estadística y Censos. Table 5 compares the mean birth weight of non-immigrants in Ecuador to that of immigrants in Spain in the early 2000s. The comparison indicates an important health advantage in favor of immigrants: babies born in Ecuador are about 170-150 grams lighter than babies born in Spain to Ecuadorian mothers.
The incidence of undereport birth weight in the Ecuadorian data is substantial in the early 2000s. However, the rate was unevenly distributed across di¤erent groups. According to Table A2 underreport in 2001 was less than 30% among mothers with more than primary education and among births that happened in hospitals. This rate was also much lower in urban than in rural areas. By 2002, the underreport rate had decreased to 32% in urban areas, to 20% in hospitals and to 24% among mothers with more than primary education. Section VI investigates the implications of underport for the results. Due to the incidence of underreport, the information on birth weights collected in the Vital Statistics is not likely to be representative of the whole Ecuadorian population: mothers with more than primary education and middle/high-income groups living in urban areas are likely to be overrepresented. While this may be seen as a limitation, the validity of the study is reassured when looking at the characteristics of the migrants. Bertoli (2010) documents that the wave of Ecuadorian migration who moved in the aftermath of the crisis came mostly from the urban areas, which were more severely hit by the crisis (suspension of the wage payment to public employees and slash in real wages due to devaluation). Its has also been argued that in the early stage of the migration process is the middle class of the wealth distribution who has the means and incentives to migrate (McKenzie and Rapoport 2007). Hence, the group of non-immigrants in Ecuador with valid information on birth weights in the early 2000s is likely to be closer to immigrants to Spain than the Ecuadorian population as a whole. This will limit the magnitude of the bias due to di¤erent composition of the comparison group. Section VI discusses the implications of underreport for the results.
The paper employs two additional data sets to investigate the fertility patterns and socioeconomic characteristics of di¤erent ethnic groups. The Spanish Labor Force Survey for the years 2000-2004 (Encuesta de Población Activa, EPA) and the Ecuadorian Labor Force Survey for the year 2001 (Encuesta Nacional de empleo, desempleo y subempleo, EMENDU). Both surveys include household level information on the socioeconomic characteristics of their family members, with particular attention to their labor market status.
Finally, I also employ the National Immigrant Survey conducted in 2007 by the Statistical O¢ ce in Spain (Encuesta Nacional de Inmigrantes, ENI 2007). This survey analyzes the characteristics of the large in ‡ow of immigrants to Spain. It covers the entire national territory and all immigrant groups. It is aimed at studying the demographic and social characteristics of immigrants as well as their migration itineraries, work and residential histories.

V. Empirical Methodology
The …rst step in the empirical strategy is to asses the magnitude of the healthy immigrant e¤ect. Accordingly, I estimate the following model: where the dependent variable, birthweight i , is the weight at birth of the child born to individual i. I 1i is an indicator variable that equals 1 if individual i is an immigrant and 0 otherwise. The OLS estimate of HIE in equation (1) is obtained from the comparison of birth weights between children born to natives and those born to Ecuadorian immigrants in Spain. It can be interpreted as the healthy immigrant e¤ect. As discussed, di¤erences in birth weight may result from the healthier habits and behaviors of immigrants ( habits ), the existence of a causal or direct e¤ect of migration on health ( migration ) or from selective migration ( selection ). That is: Since healthy habits are common to individuals originating from the same country, the comparison of birth outcomes between immigrants at destination and nonimmigrants in the source country produces a joint estimate of the e¤ect of selection and of any causal e¤ect of migration. The OLS estimate of this e¤ect can be obtained from: where = ( selection + migration ) and I 2i is equal 1 if i is an immigrant in Spain and 0 if i is a non-immigrant in Ecuador.
To disentangle the contribution of selective migration from that of any direct or causal impact of migration, one would need to compare the health distribution of recent immigrants to their distribution had they not migrated. Experimental data to assess the presence of selective migration is rather scarce (see ). This paper takes an alternative approach and employs administrative data to test the prediction that the selection of immigrants increases with distance to destination. That is, immigrants from more remote areas should be healthier than those originating from neighboring areas to compensate the higher cost of the movement. The empirical exercise in the next section will compare the birth outcomes of two of the largest ethnic minorities that have recently arrived to Spain from very di¤erent geographical regions: Ecuador and Romania. Table 6 summarizes the economic costs of moving to Spain from di¤erent destinations over the period 1999-2003 and 1999-2007, when the largest in ‡ow of immigrants arrived to Spain. This information is collected from the National Immigrant Survey conducted in Spain in 2007. The survey interviews immigrants in many dimensions, and, in particular, they are asked to provide all the costs associated to the movement from their country of origin. These costs are not only restricted to transport expenditures such as air fares or train tickets, but also all types of travel allowances (food, accommodation, etc...), the cost of obtaining a visa or other legal document, and any other expenses incurred before or during the migration episode. Table 6 shows that the cost of migrating from Ecuador is 3,5 time larger than that of moving from Romania (i.e. 1,609.72 Euros from Ecuador and 464.95 Euros from Romania). Thus immigrants from Ecuador are expected to be drawn from a higher end of the health distribution.
To formalize this idea I use the notation in the program evaluation literature and de…ne a health equation (in terms of birth outcomes) for immigrants: and non-immigrants: In the presence of selection, the e¤ect of migration on health is heterogeneous in the population. In particular: where [ (1) (0)] = Ef i g is the average gain from migration in the population, namely the direct or causal e¤ect of migration ( migration ). The second part of equation (5), [U i (1) U i (0)]; represents the idiosyncratic gain from migration. Note that the model in equation (3) and (4) can be rewritten as: It can be shown that the OLS estimate of i in equation (6) is: Under perfect randomization (i.e. migration is the result of a lottery): and the OLS estimate of i in equation (6) is: In the presence of selective migration, the decision to migrate can be modeled as: . An individual will migrate only if the latent variable I 2i is above a certain threshold and will not otherwise. That is: When Cov(U i ; V i ) = 6 = 0 by the properties of the normal distribution: In this framework the contribution of selective migration on health can be assessed by rewriting the model in equation (6) as: Let us now consider two groups of immigrants from two di¤erent geographical origins, Ecuador (EC) and Romania (ROM ): Assume that: 1) The direct or causal e¤ect of migration is the same for Ecuadorians and Romanians: 2) The distribution of unobservable characteristics has mean equal zero in the two populations: 3) Migration is not due to health reasons (i.e. those in worse health are not more likely to migrate): Under these assumptions the di¤erence between the two equations in (7) is: The …rst term in equation (8), , is negative as, according to the health statistics reported in Table 4, the average birth weight in Ecuador (3,102 grams) is lower than the average birth weight in Romania (3,186 grams).
Under the assumption that migration costs increase with distance, one would expect immigrants from Ecuador to be more positively selected than those from Romania. Thus, the second term in equation (8), , should be positive. Thus, the hypothesis that selection increases with geographical distance can be tested by estimating the following model: where I 3i equals 1 if i is an immigrant from Ecuador and 0 if i is an immigrant from Romania. In this context, a > 0 in equation (9) is evidence of positive selection in favor of Ecuadorian immigrants.

VI. Results
The Healthy Immigrant E¤ ect The estimates of the healthy immigrant e¤ect are reported in Table 7. The estimated coe¢ cients correspond to the model in equation (1), where the birth weight of children born to Ecuadorian mothers in Spain is compared to that of children born to native women in Spain. The model in equation (1) is extended by including as additional controls an indicator for the gender of the child (male), a set of dummies for the age of the mother when the birth occurs, an indicator for the month of birth, and indicators for the province of residence in Spain. 10 Each column correspond to a di¤erent regression estimated between 2001 and 2003. The estimates indicate an advantage in favor of immigrants between 85 and 91 grams. Since the majority of Ecuadorians in the early 2000s were recent immigrants, these estimates are not likely to be a¤ected by the process of assimilation or acculturation. Figure 1 plots the kernel estimates of the birth weight distribution of immigrants (solid line) and natives (dashed line) in 2001 and Figure 2 the di¤erence between the two distributions. The …gures suggest that the health advantage in terms of birth weight is not only concentrated in the mean of the distribution, but it also present in other parts, in particular the upper tail. Natives and immigrants may di¤er in many dimensions, some of them having a direct impact on birth outcomes. First, immigrants tend to be positively selected in terms of education and productive skills. The health economic literature has established a strong relationship between parental education and a child's health (Currier 2009). Hence, positive selection in education could lead to higher birth weight among immigrants. Unfortunately the Spanish Vital Statistic does not contain information on maternal education until 2007. For the years in our analysis we can only control for di¤erences in productive skills by including in the regression the mother's labor market status and an indicator for being employed in a high skilled occupation. Since these variables are not perfect proxies for educational achievement, the estimate of the health gap could still be biased. However, Bertoli et al (2011) …nd some evidence of negative selection in term of the education of Ecuadorian immigrants to Spain. Thus, the omission of maternal education from equation (1) should, if any, produce a negative bias on the health gap.
Di¤erences in family size may also be relevant for health outcomes. The child quality investment model (Becker 1981 andChiswick 1988) predicts that, at any given level of family resources, more children imply smaller levels of investment per children and thus lower quality. Accordingly I control for the presence and number of previous children, and a variable that captures the e¤ect of birth spacing. As a robustness test at the end of this section I further investigate the implications of di¤erences in fertility behavior for the results. It has also been documented that parental income a¤ects child health (Currie and Moretti 2007). The Vital Statistics do not contain information on family income or wealth. To proxy for the level of economic resources I include as additional regressors in equation (1) an indicator for the marital status of the mother and another for being born at a hospital, in addition to the labor market indicators previously discussed.
The results of the extended model are presented in Table 8. For all years, the weight advantage in favor of immigrants increases by about 10 grams and remains highly signi…cant. The variables capturing the economic situation of the family (being born in a hospital, married, mother's work and mother working in a high skilled occupation) have all a positive e¤ect. The coe¢ cients on the variables related to family size are also positive. There is also evidence of a negative e¤ect from birth spacing.
It has also been highlighted that the process of cultural assimilation is faster among interethnic couples (Meng andGregoy 2005 andChiswick et al 1997). In an attempt to investigate the e¤ect of acculturation on birth outcomes, I estimate the e¤ect of intermarriage on birth outcomes. The results presented in Table A3 indicate that intermarriage does not have any e¤ect on birth weights. This is likely to be due to the high degree of sorting in the data. In 2001, a 0.31 percent of the births were to interethnic couples and this percentage increased to only 0.71 percent in 2005. Table 9 examines the presence of the healthy immigrant e¤ect in alternative birth outcomes that are popular in the literature. The table shows the estimates for the model in equation (1) where the dependent variable has been replaced by a low birth weight indicator (column 1), the number of gestational weeks (column 2), an indicator for being born between week 38 and 42 (i.e. normal term) (column 3), one for preterm birth (column 4), and one for death in the …rst 24 hours after birth (column 5). The estimates indicate a health advantage in favor of Ecuadorian immigrants in terms of the incidence of low birth weight (i.e. 2 percentage points lower probability), gestational age (i.e. 0.043 additional weeks of gestation), the probability of being born between week 38 and 42 (i.e. 1 percentage point higher) and the probability of pre-term birth (i.e. 1 percentage point lower). No di¤erences are observed for the probability of dying 24 hours after birth. The message from the estimates in Table 7 to 9 is clear: upon arrival to Spain, children born to Ecuadorian mothers are heavier and thus healthier than those born to native women.
As argued in the health economic literature, this health advantage in terms of birth outcomes may have implications on future outcomes and maybe compensate part of the negative e¤ect associated, for instance, to the presence of discrimination (Bosch et al. 2010). The …ndings are also consistent with the extensive evidence on the health immigrant e¤ect documented for Mexican immigrants in the US and other minority groups in Canada and Australia.
Next, I compare the weight of babies born to Ecuadorian immigrants in Spain to that of non-immigrants in Ecuador. Table 10 shows the estimates of the model in equation (2) including as additional controls the gender of the child, the age of the mother and the month of birth. The estimates indicate that newborns to immigrants are between 168-148 grams heavier than those born to non-immigrants. Health indicators are in general better in Spain than in Ecuador (see Table 4), and this may partly re ‡ect better health care systems or some other environmental factors (i.e. less pollution 11 ). Table 11 removes from the previous estimate the e¤ect of being born in Spain (common to both natives and immigrants). The net birth weight di¤erence is reduced to 60-65 grams, and remains highly signi…cant at any conventional level. Figure 2 plots the birth weight distribution of immigrants in Spain (solid line) and non-immigrants in Ecuador (dashed line) in 2001. The distribution for migrants lies clearly to the right of that for non-immigrants, reassuring that the health advantage estimated for the mean of the distribution by OLS is present along all the domain of the distribution, in particular the middle/upper part. 12 This result is also evident from the plot of the di¤erence between the native and immigrant distribution (see …gure 2b). Table 12 investigates the implications of di¤erences in observable characteristics between immigrants and non-immigrants. The set of additional controls included in estimation is limited to the variables that are common to the Vital Statistics of the two countries, namely those related to fertility histories (i.e. the presence and number of previous children) and whether the child was born at a hospital. The impact of these controls is small, as the birth weight gap is only reduced by 10 grams.
Two additional considerations should be taken into account when analyzing the results in Table 11 to 12. First, Ecuador was immerse in a major economic recession in the early 2000s, which may have had a negative e¤ect of birth outcomes. Indeed, Bozzoli and Quintana (2013) documents the existence of procyclicality in birth weights for Argentina. Second, a non-negligible fraction of the observations in the Ecuadorian Vital Statistics do not report information on birth weights in the early 2000s. To further investigate the implications of these two concerns, Table A5 13 In all three speci…cations, the weight advantage in favor of immigrants remains statistically signi…cant and of similar magnitude, suggesting that the previous concerns do not have implications for the results.
The estimated health advantage for Ecuadorian immigrants relative to their native counterpart may result from a direct causal e¤ect of migration and/or the presence of selective migration. To the best of my knowledge, no paper has been able to identify the causal e¤ect of migration on birth outcomes. The closest evidence is the paper 12 Table A4 replicates the results in Table 11 but replacing the birth weight dependent variable by a low birth indicator. While there is a statistically negative e¤ect on the immigrants'low birth probability, its magnitude is very small (i.e. the likelihood of low birth is 0.3 percentage points lower among immigrants than natives). This reinforces the result that most of the action occurs in the middle/upper part of the distribution. 13 The incidence of underreport birth weight substantially decreases over this period. To gain a better understanding of the factors behind the health gap I take advantage of the large and diverse in ‡ow of immigrants to Spain in the 2000s. I will focus the next comparison on immigrants from Ecuador and Romania. These two minority groups are comparable in many dimensions. First, the cultural and linguistic barriers are low for both groups (i.e. Spanish is the language of Ecuador, and Romanian is a Romance language very close to Spanish 14 ). Second, the bulk of Ecuadorians and Romanians arrived between 2000 and 2003. 15 Third, the two groups moved to Spain for economic reasons. Ecuadorians came escaping from the economic and …nancial collapse in 1999, while immigrants from Romania arrived looking for jobs, as a result of the high unemployment rates after the massive restructuring of state enterprises in the late 1990s. Finally, table A6 in the Appendix shows that Romanian and Ecuadorian immigrants in Spain are similar in terms of age, education and work status. The main di¤erence between the two is observed in terms of fertility outcomes: a 49 percent of the Romanians have children, as opposed to 76 percent of the Ecuadorians and the average number of kids is 1.42 and 1.79 respectively. The implications of these di¤erent fertility behaviors are investigated at the end of this section. The standard migration model predicts that immigrants from more remote areas should be more selected in terms of health. Table 13 investigates this hypothesis by estimating the model in equation (3) where the birth outcomes of Ecuadorians are compared to those of Romanians. Most of the birth outcomes indicate a health advantage in favor of Ecuadorians: newborns to Ecuadorian mothers are 52 grams heavier, have a smaller probability of low-birth weight (2.8 percentage points lower), longer gestational age (0.15 weeks), a higher probability of being born between week 38 and 42 (3.2 percentage points higher), a lower incidence of pre-term birth (3 percentage points lower) and a lower probability of death during the …rst 24 hours (0.2 percentage points). 16 All these e¤ects are statistically signi…cant at any conventional level.
Overall, the results from the previous comparisons reveal a health advantage for children born to new Ecuadorian immigrants in Spain. Upon arrival, newborn babies are 100-85 grams heavier than those born to natives, and 64-48 grams heavier than those left behind. While I cannot precisely estimate the contribution of selection to these results, the comparison in terms of several birth outcomes between newly arrived immigrants from Romania and Ecuador supports the idea that positive selection in health increases with distance to destination.

A) Comparison to other ethnic groups
To further investigate the hypothesis that immigrants from more remote areas are more likely to be positively selected I extend the comparison to other ethnic groups also popular in Spain, namely Colombians and Bulgarians. During the early 2000s Colombian immigrants were the third largest group after Moroccans and Ecuadorians (see Table A7). Bulgarians are a smaller group, but represented the 8 th most popular immigrant-sending country in 2004 and the second largest group among Eastern European immigrants in Spain. 17 Colombian emigration began in the 1960s, mainly in search of better economic opportunities. During the …rst migration wave of the sixties and the seventies, emigration was mostly to the United States, and mainly for economic reasons -the possibility of …nding work and augmenting income -and, to a certain extent, for political reasons, such as the threat of illegal armed groups. Typical of this migration was the level of education of the migrants, their good knowledge of the English, a larger proportion of women than men and a considerable presence of middle and upper income class migrants. In the case of emigration to Spain, it increased considerably as of 1998, which was largely the result of Colombia's economic crisis in the years 1998-1999 and because of Spain's attraction as a place in which to join the collective immigrant workforce, as well as the advantage of the language. A large percentage of Colombian migrants in Spain are women, who have a medium educational level (see Table A6).
Ecuador and Colombia are neighboring countries and Table 6 indicates that the economic cost of migrating from any of the two countries to Spain is similar. Accordingly, large di¤erences in birth outcomes are not expected among the children of those immigrants. Table 14 reveals a small, though statistically signi…cant, disadvantage for children born to Colombian immigrants in terms of gestational age. However, this disadvantage does not translate into other birth outcomes. Indeed, di¤erences in birth weight, the probability of low birth weight and that of dying before 24 hours are not statistically signi…cant. 18 Large-scale immigration from Romania and Bulgaria coincided with their inclusion in the list of countries exempted from the general visa requirements in early 2002 as a …rst step towards their membership of the EU. As for the economic aspects, both countries were by far the poorest of the 27 countries that would be part of the enlarged EU after January 2007, making emigration an attractive mean of improving the prospects for both the emigrants themselves and of those staying behind. Table  6A indicates that immigrants from Bulgaria and Rumania are comparable in terms of socioeconomic characteristics.
Bulgaria shares its northern border with Romania and Table 6 indicates that the economic cost of migrating from any of the two countries to Spain is similar and much lower than the cost of migrating from Ecuador. Table 15 compares the birth outcomes of all the four di¤erent immigrant groups considered here. The excluded category are Ecuadorian immigrants. Consistent with the hypothesis that selection is inversely proportional to distance, children born to immigrants from Romania and Bulgaria are lighter than those born to Ecuadorian immigrants (i.e. 53.20 grams and 34.45 grams respectively) and these di¤erences are statistically signi…cant at any level of signi…cance. Di¤erences in terms of other birth outcomes between Bulgarian and Ecuadorian immigrants are not statistically signi…cant.
From the previous results we can conclude that immigrants originating from more remote areas are healthier (in terms of birth weights) than those from neighboring ones. This evidence indicates that selection plays a role in determining who migrates and who does not, and can partly explain the large health advantage observed among recent Ecuadorian immigrants in Spain.

B) Di¤ erences in fertility
Di¤erences in the fertility behavior of immigrants and natives may have implications for the previous results. Migration may a¤ect the fertility pattern of families through several channels (Blau 1992). The assimilation mechanisms predicts that di¤erent tastes or family size preferences formed in the origin country can explain initial differences in fertility between natives and immigrants. Over time, immigrants are likely to alter their reproductive behavior to conform to childbearing practices in the host country. In the case of high-fertility source countries, the fertility of immigrant women is expected to exceed that of their native-born counterparts initially but approach to native fertility over time (see Kahn 1994 andBen-Porath 1973).
Migration however may have disruptive e¤ects on fertility. The postponement of fertility can arise from, at least, two reasons. First, the economic resources of the household can temporary decrease, and fertility will fall as a results of a negative income e¤ect. Second, fertility may also decrease due to demographic factors such as delayed marriages or temporary separation of couples. If disruption occurs, the fertility of recent immigrants will be low, and progressively increase to achieve the desired level (see Ford 1990 and Adserà and Ferrer 2013).
Finally, di¤erences in fertility may also result from selection. Immigrant women may be a self-selected group whose fertility is low relative to others in the source country due to either tastes or to characteristics associated to labor market success. These women may also have a stronger preferences to invest in child quality and reduce quantity (Schultz 1984).
I …rst explore di¤erences between natives and immigrants in the probability of having a children upon arrival to the country. To this aim, I estimate the following fertility model: where the dependent variable inf ant i is an indicator for the presence of an infant in the household (i.e. a child younger than 1 year old). I 1i is the foreign-born indicator, ysm i captures years since arrival in the country, X i is a vector of mother's controls that include education, marital status, fertility history, and labor supply. The model also includes a set of dummies for the age of the mother, and province and year indicators. To further explore di¤erences in the fertility the model is also estimated by replacing the presence of an infant by the total number of children younger than 17 (children i ) at the time of the interview. Equation (10) is estimated using data from the Spanish Labor Force Survey (Encuesta de Población Activa, EPA) for the year 2001 to 2003 and Ecuador (Encuesta Nacional de Empleo, Desempleo y Subempleo, EMENDU) for the year 2001. Most of the analysis is conducted on a sample of recent immigrants (with less than 4 years of residence in the country). 19 Table 16 presents the …rst set of results, where the fertility behavior of Ecuadorian immigrants is compared to that of natives in Spain. The …rst column displays the raw estimated di¤erence in the propensity of having an infant between 2001 and 2003, and thus being part of sample employed in the estimation of the healthy immigrant e¤ect. The estimate reveals a 4 percentage points higher probability for Ecuadorian immigrants. This positive gap in fertility remains after controlling for years since migration (column 2) and socioeconomic characteristics (column 3). 20 In column (4) the time period is extended to include the year 2000 and 2004 and column (5) adds to the analysis immigrants with longer experience in the country. All the speci…cations indicate a statistically signi…cant and positive e¤ect of being an Ecuadorian immigrant on the probability of having an infant, with a magnitude that oscillates between 3.3 and 4.2 percentage points. Column (6) and (7) explore di¤erences in total fertility (i.e. number of children at the time of the interview). The columns also reveal a higher fertility among Ecuadorian immigrants in terms of total number of children. In addition, there is a positive e¤ect on the total number of children related to the number of years in the country.
The higher fertility of Ecuadorian immigrants upon arrival is likely to respond to cultural di¤erences. Ecuador is a high-fertility country (the average number of children per women was 5.  21 The high-fertility context in which Ecuadorian immigrants were reared may have shaped their preferences for large families. Moreover the positive e¤ect of years since migration on the total number of children indicates that upon arrival the fertility of Ecuadorian immigrants may be below desired levels due to the presence of some disruptive e¤ects associated to the migration episode. On the whole, these results suggest that the stronger preferences for larger families observed among immigrants should produce, if any, a negative bias in the estimated health gap reported in the previous section. Moreover, the existence of disruptive e¤ects, if arising from economic di¢ culties, would reinforce the negative sign of the bias. Given the high levels of fertility in Ecuador, the results in Table 16 could also be consistent with the presence of immigrant selection. That is, immigrants have a higher fertility rate than natives in Spain but still lower that those in Ecuador. Table 17 compares the fertility behavior of immigrants to that of non-immigrants in Ecuador. Column (1) to (3) presents the results for the probability of having an infant, while column (4) presents the estimates for total fertility. In all the speci…cations, the fertility of immigrants is lower than that of natives. Immigrants to Spain between 2001 and 2003 have a probability 7 to 8 percentage points lower of having an infant and the total number of children is also smaller (i.e. on average, immigrant women have 0.5 less children than non-immigrants). Moreover these di¤erences do not change 20 The Labor Force Survey allows us to include in estimation as controls the education of the mother, her marital status, the presence of previous children and an indicator for whether the woman is employed. 21 World Bank indicators.

22
with years in the country. These results reinforce the view that immigrants to Ecuador are selected on the basis of characteristics that lead to lower fertility and probably to higher child quality. Finally, Table 18 explores di¤erences in the fertility between recent immigrants from Ecuador and Romania. First note that while Ecuador should be classi…ed as a high-fertiltiy country, Romania belongs to the low-fertility group (i.e. 2.4 in 1980, 1.8 in 1990, 1.3 in 2000 and 1.3 in 2010). The …rst two columns in Table 18 show the estimated di¤erence in the probability of having an infant during the period 2001-2003 among recent immigrants. The estimates reveal that the probability of having a child upon arrival is signi…cantly higher among immigrants from Ecuador. Regarding total fertility (column (3) and (4)) the positive initial e¤ect in favor of Ecuadorian immigrants remains, though the number of children increases with years in the country for the two groups at a similar rate.
The initial disadvantage in terms of fertility for Romanians is consistent with both the presence of disruptive e¤ects and with the behavior of immigrants originating from low-fertility countries. While the presence of disruptive e¤ects can not be rule out, the estimates in Table 18 suggest that, if any, they should equally a¤ect the fertility of both ethnic groups (i.e. the groups speci…c coe¢ cient on years since migration is never signi…cant). Accordingly, while the initial fertility of both immigrant groups may be below the desired level, the lower fertility of Romanians persists over time. Hence the fertility pattern of Romanians is then consistent with the assimilation mechanism of immigrants originating from a low-fertility countries. As a result, the positive health advantage in favor of Ecuadorian immigrants with respect to immigrants from Romania will be, if any, underestimated as a result of the stronger preferences of Romanians for less children (more quality).

VII. Conclusions
This paper presents new empirical evidence on the determinants of the health advantage observed among recent immigrants. It employs evidence from a large migration in ‡ow of Ecuadorian to Spain in the early 2000s. Using the Vital Statistics in both countries, I document an important health advantage for immigrants in terms of birth weights and other birth outcomes. The comparison to other recent minority groups in Spain suggests that this advantage is likely to be driven by the positive selection of Ecuadorian immigrants in terms of health.
The …ndings in this paper have at least two important policy implications. First, the health advantage of immigrant children at birth may translate into an advantage in terms of education and earnings that may compensate some of negative e¤ects associated to migration (i.e. discrimination, lower economic resources or poorer net-work quality). Second, immigration is not likely to represent a …nancial burden on the public health system, as long as the health advantage of recent immigrants does not deteriorate over time.   Note: Mean and standard deviation of birth weights to mothers 15 to 49, excluding multiple births and newborns whose weight was either under 500 grams or above 9,000 grams.     (1), included in estimation are also the gender of the child, the set of age dummies for the mother, and monthly and province dummies.        Source: Spanish and Ecuadorian Labor Force Survey Note: Estimates from the linear probability model in equation (2). Infant is an indicator for the presence of an infant (1 year old or less) at the time of the interview. Total fertility is the total number of children at the time of the interview. The sample includes non-immigrants in Ecuador and immigrants in Spain. Source: Spanish Labor Force Survey Note: Estimates from the linear probability model in equation (2). Infant is an indicator for the presence of an infant (1 year old or less) at the time of the interview. Children is the total number of children at the time of the interview. The sample includes immigrants from Ecuador and from Romania in Spain.   (1) with the additional controls in Table 7