چگونه اقتصادهای بسیار باز جریان مهاجرت های بزرگ را تنظیم کردند؟ مدارک و شواهد از مناطق اسپانیایی

We study the labor market effects of the large immigration wave in Spain between 2001 and 2006. In this period the foreign-born share increased from 6% to 13%, with a total inflow exceeding three million immigrants. Our analysis exploits the large variation in the size of immigration flows across Spain's regions. To identify causal effects, we take advantage of the fact that immigrants' location choices were strongly driven by early migrant settlements that arrived during the 1980s. We find that the relatively unskilled migration inflows did not affect the wages or employment rates of unskilled workers in the receiving regions. The growth of the unskilled labor force was absorbed mostly through increases in total employment. This increase did not originate in changes in the composition of regional output, but was instead driven by changes in skill intensity at the industry level. Regions that received a large inflow of unskilled immigrants increased the intensity of use of the now more abundant (unskilled) labor, relative to other regions. The key industries responsible for this absorption were retail, construction, hotels and restaurants and domestic services. These results are inconsistent with standard open economy models but are in line with recent empirical studies for the United States and Germany.

In recent years Spain received a massive wave of immigration, with the foreign-born share of total population jumping from 6% in 2001 to 13% in 2006.1 This paper studies how Spanish regional economies responded to the large changes to the size and skill composition of their labor force caused by immigration. Specifically, we adopt a spatial correlations approach and employ instrumental variables to provide causal estimates of the effects of immigration on employment rates, wages, and the structure of production for Spanish provinces in the period 2001–2006. Rising international migration flows over the last decade have revived interest on the economic effects of immigration, particularly in Europe.2 The recent eastward enlargement of the European Union has sharply increased migration flows across its member states. Moreover, for countries such as Spain or Ireland, large-scale immigration is a completely new phenomenon in modern times, with important macroeconomic implications.3 The long history of immigration in the U.S. gave rise to a vast literature on the economics of immigration.4 In contrast, relatively little is known about the effects of immigration in Europe and, in particular, regarding the new immigration countries. Given the large institutional differences between most European countries and the U.S. it is unclear how well the findings for the U.S. extrapolate to these countries.5 The Spanish immigration experience since year 2000 is particularly interesting for a number of reasons. First, the size of the inflows in absolute terms and relative to population has been spectacular. Except for Israel in the 1990s, no other OECD country has experienced such massive immigration flows in the postwar period. As noted earlier, the fraction of foreign-born individuals in the working-age population more than doubled in just 5 years, rising from 6% to 13% between 2001 and 2006 (see Fig. 1). During the same period, the foreign-born population in the U.S. went from 11 to 12.1%.6 Full-size image (24 K) Fig. 1. Share of the foreign-born population relative to total population in Spain. Source: Registry data at January 1st of each year (“Padrón”). Figure options Secondly, until recently Spain was a country of emigration. In modern times it is only during this period that immigrants started arriving in sizeable numbers. As a result, Spain's recent immigration surge was largely unexpected by economic agents. To the extent that the capital stock did not anticipate the immigration wave, we expect to observe negative short-run effects of immigration on wages. By the same argument, we also think it plausible to assume that the educational attainment of natives was not affected by the influx of immigrants in the short term. Another feature of the Spanish experience is that a large fraction of recent immigrants are native Spanish speakers from Latin America. These special features make the Spanish immigration episode particularly interesting. Some researchers have already recognized this.7 We conduct a spatial correlations analysis focusing on regional economies.8 Relative to countries, regions are very open economies, tightly interconnected by flows of factors, goods, and ideas. Consequently, absorption of immigration flows can operate through a variety of channels. In addition, the size of immigration flows relative to population is often orders of magnitude larger than for national economies. This methodological approach seems well suited to the Spanish case. First, there is very large regional variation regarding the size of immigration flows. Fig. 2 reports the foreign-born share in 2006 (age group 25–45) for the 52 Spanish regions. While the provinces in the South and West of Spain are mostly below 6%, those around Madrid and on the Mediterranean display foreign-born shares around 20% and higher. Secondly, despite their low numbers, there is a relatively long history of migration to Spain from Morocco and several South American countries. As we shall show, the location choices of early arrivals partially determined the geographical distribution of recent immigrants. This provides us with a valuable source of exogenous variation in the size of immigration flows by region, which allows us to construct a credible instrument for the identification of the effects of interest. Full-size image (51 K) Fig. 2. Foreign-born share in 2006 (age bracket 25–45) in Spanish provinces. Source: 2006 Spanish Labor Force Survey (EPA). Figure options The exercise we carry out in this paper is challenging in terms of data requirements. Our period of interest (2001–2006, roughly the period of the immigration surge) lies after the most recent Census year (2001), and thus we are restricted to the smaller samples available from the Labor Force Survey. In many countries these data are too sparse to accurately quantify changes in the foreign-born population at the regional level. However, this problem is much less serious in the case of Spain. The reason is that high-quality registry data exist that accurately track changes in the (both native and foreign-born) population at the local level. These data are an important input into the sampling design of the Spanish Labor Force Survey.9 Our main results are the following. First, we document that immigration flows were relatively unskilled and analyze their effect on aggregate labor market outcomes. We find that immigration did not have any significant impact on the structure of wages or on employment rates in Spanish regional labor markets. This finding is consistent with several prior studies using data for other countries.10 The recurrent finding of insensitivity of wages to immigration flows has led researchers to explore alternative mechanisms by which economies can absorb immigration flows. Recognizing that regional and local economies are highly interconnected by trade, empirical work has focused on the adjustment mechanism described by the Rybczynski theorem.11 According to this celebrated result, in response to an inflow of a factor of production, a small open economy may not suffer any changes to equilibrium factor prices and absorb the inflow simply by changing its structure of production. Specifically, production (and employment) would expand in sectors that use that factor intensively. The pioneering empirical explorations of this result are Hanson and Slaughter, 2002 and Gandal et al., 2005, who carry out accounting decompositions. We follow the more recent approach developed by Lewis (2003), which uses the spatial correlations methodology to provide a more formal econometric test of the Rybczynski theorem based on a between-within industry decomposition. In a study contemporaneous to ours, Dustmann and Glitz (2008) apply Lewis' approach using German data. We find that immigration did not significantly change regional output mix (between-industry absorption). Instead, the main channel of absorption of immigration-driven increases in labor supply was within-industry. In other words, following a relatively unskilled labor inflow, the typical industry in the receiving regions increased the intensity of use of this type of labor, relative to regions without immigration. Lastly, we analyze the role played by individual industries in the absorption process. We find that the industries that played the leading role are largely non-tradable: retail, hotels and restaurants, construction and domestic services. Our findings imply that the adjustment of Spanish regional economies to immigration shocks is very similar to the patterns found in the U.S. and in Germany. Moreover, our results reinforce the view that standard open-economy models are not able to account for the response of local and regional economies to factor supply shocks. In particular, we do not find the strong connection between relative factor intensities and relative factor prices implied by the theory. Finally, our results also contribute to the literature on the effects of the recent wave of immigration in Spain. Unlike previous work, our analysis uses the recently available new wage data based on Social Security records. In addition, we are the first to show that a Card-type instrument is useful also for Spain to identify the causal labor market effects of immigration. We are not the first to analyze the economic effects of immigration in Spain. Carrasco et al. (2008) use data for the period 1991–2001 from a variety of sources. Methodologically they follow the skill correlations approach introduced in Borjas (2003), adapted to the data availability for Spain. Their main finding is that growth in the foreign-born share across skill cells is negatively correlated with growth in employment rates and wages. However, the magnitudes are small and the effects not robust. The authors conclude that there is no robust evidence of negative labor market effects of immigration. In comparison, our period of interest is 2001–2006, corresponding to much larger inflows, and we use different data sources as well as a different methodology (spatial correlations). Also, in addition to studying wage and employment effects, we focus on the effects of immigration on industrial composition at the regional level. Amuedo-Dorantes and De la Rica (2007) estimate the immigration surplus, both at the national and regional levels in Spain. Amuedo-Dorantes and De la Rica (2008) show that immigration led Spain-born workers to shift occupations, toward less exposed, more communication-intensive occupations. Also related to this study, Blanes et al. (2008) analyze the effects of immigration on the industry structure of Spanish regions. Their data is for the period 1995–2002, prior to the largest inflows, and their methodology is an accounting decomposition as in Hanson and Slaughter (2002). The remainder of the paper is organized as follows. Section 2 presents the theoretical framework. Section 3 describes the data sources and introduces the empirical strategy. Section 4 presents the results of the empirical analysis, starting with the effects of immigration on wages and employment and moving on to the between and within industry absorption. Section 5 concludes.

We study the effects of recent migration flows on Spanish regional labor markets. The Spanish case is particularly suitable for this type of analysis given the large magnitude of the inflows in a very short time frame and from very low initial levels. Moreover, the inflows affected some regions much more than others, providing large cross-sectional variation. In terms of identification, we take advantage of the fact that immigrants' location choices were strongly driven by earlier migrant settlements for some of the main countries of origin. We find that the relatively unskilled migration inflows did not affect the wages or employment rates of unskilled workers in the receiving regions. Our finding that immigration did not seem to affect wages in Spain confirms the findings in Carrasco et al (2008) for a more recent period. This is reassuring, given the differences in the data and methodology between the two studies.45 Our results suggest that the increase in the unskilled labor force was absorbed mostly through increases in total employment. This increase did not originate from changes in output mix, but was instead driven by changes in skill intensities. The average industry responded to the increase in the supply of unskilled workers by using the more abundant type of labor more intensively. In particular, the industries that played the main role were retail, construction, hotels and restaurants and domestic services, as well as the public sector. All these industries produce non-traded goods. Overall, the response of Spanish regions to immigration shocks is remarkably similar to the response found by Lewis, 2003 and Dustmann and Glitz, 2008 for the US and Germany, respectively. By implication, the large differences in labor market institutions among these three countries do not appear to shape the channels through which local economies absorb immigration flows. Moreover, the pattern of adjustment documented in these studies is inconsistent with standard open economy models: the labor supply shocks induced by immigration substantially alter the skill composition of employment at the industry level without having an effect on the wage structure or on the regional industry composition. Hence, a new theory of local and regional economies is needed that can account for this robust set of facts. Currently, immigration economists are busy searching for such a theory. A promising venue builds on the idea that immigration shocks induce changes in production technologies at the industry level.46 Another potential explanation is that natives and immigrants are imperfect substitutes in production even controlling by education (Peri and Sparber, 2009). In fact, our finding that immigration did not depress wages and employment rates in Spain is consistent with the results in Amuedo-Dorantes and De la Rica (2008), who show that Spain-born workers have shifted toward occupations less exposed to immigration (more communication-intensive).47 In our view, future work should also focus on the role of physical capital. At the local or regional level capital flows face no impediments and thus are potentially very large. If the degree of substitution between capital and the different skill groups differ, it may be possible to build an alternative explanation for the evidence found in this paper. More broadly, it will be interesting to follow the assimilation experience of Spain's recent immigrants as well as the potential consequences for Spain's domestic policy.48