نتایج بازار کار، انباشت پس انداز و مهاجرت معکوس

In this paper, I examine the determinants of return migration from Germany for immigrants from four different source countries, and test the savings accumulation conjecture that is used to rationalize return migration decisions using both cross-country and time variation in purchasing power parity. The empirical results confirm the savings accumulation conjecture. Therefore, return migration can be seen as part of optimal life-cycle location choices in this context. I also examine how labor market outcomes influence return decisions. A key finding here is that unlike previous studies, which find a positive impact of unemployment on return migration, I find that the direction of the impact of unemployment changes by the spell length.

The level of return migration has been high both in North America and in Europe. Jasso and Rosenzweig (1982) report that of the 1971 cohort of immigrants in the U.S., the fraction that returned by 1979 could be as high as 50%.2 According to the German Federal Statistics Office, while almost eight hundred thousand immigrants entered Germany on average annually between 1962 and 2005, more than 560 thousand left each year on average.3 Moreover, many of these immigrants return to countries where wages are lower than those in the host countries. The question, then, is why so many immigrants return despite higher earnings in the host country. Borjas (1994) explains return migration as part of optimal life-cycle location decisions. At the time they immigrate, immigrants realize that after they acquire physical or human capital in the host country, it may be optimal for them to return because the returns to that type of capital are higher in the home country. The savings that immigrants accumulate in Germany have higher purchasing power in their home country due to the lower prices there. Djajic (1989), Dustmann, 1997 and Dustmann, 2003, and Stark et al. (1997) use this fact as a motivation for return migration. Mesnard (2004) examines a model of joint optimal migration duration and occupational choice after return where migrants accumulate savings in the host country to overcome liquidity constraints in their home country. In optimal location decisions over the life-cycle, it may also be the higher returns in the home country to the human capital acquired in the host country that rationalize the return migration decision. Return migration may also be the result of unexpected events, either in the host or home country (Berninghaus and Siefer-Vogt, 1993 and Tunalı, 2000). Even when it is optimal to immigrate ex-ante, it may be optimal to return after the realization of negative shocks in the host country like unemployment. Another explanation used in a number of studies (Hill, 1987 and Djajic and Milbourne, 1988) is that migrants prefer to live in their home country. According to this, in a model where immigrants choose lifetime consumption and time in the home country given the wages in the home and host countries, the remaining worklife and the cost of migration, it may be optimal to immigrate and then return depending on the preferences of the potential immigrant. The findings of the empirical literature on immigrants in Germany, in fact, suggest a savings accumulation motivation for immigrants in this country. Using a survey of Turkish emigrants from Germany in Turkey, Dustmann and Kirchkamp (2002) report that only 6% worked as salaried workers after return whereas 51% of the returners were self-employed. The other 43% were retired. The facts that half of these migrants engaged in entrepreneurial activities after return and that most of the rest lived as rentiers suggest a savings motive for immigrating to Germany. If the goal of immigrants was to accumulate savings, we would expect their saving rates to be high. Based on an empirical investigation of Turkish immigrants—the largest immigrant group in Germany—Kumcu (1989), in fact, finds evidence for very high savings rates. On the other hand, McLean Petras and Kousis (1986) find that labor market opportunities are very limited for Greek immigrants after they return from Germany. Most are forced to choose from unemployment, informal sector, and scatter jobs. This makes it unlikely that their return was motivated by the higher returns to the human capital they acquired in Germany. Therefore, I focus on the savings accumulation conjecture as the potential explanation for return migration from Germany and test this conjecture in this paper. In order to test the savings accumulation conjecture, we need a source of variation in the purchasing power of immigrants' savings. One such source of variation across immigrants would be their earnings. In fact, a number of studies examine the relationship between immigrants' earnings and their return propensity. Borjas (1989) finds a positive association between earnings and return migration for immigrants in the U.S. On the other hand, Klinthall (1999) finds a negative relationship between income and return migration for Greek migrants in Sweden. For immigrants in Germany, Dustmann (2003) reports a negative relationship whereas Constant and Massey (2002) find no statistically significant relationship between these two variables. However, the problem with this approach is that earnings are likely to be jointly determined with the return migration decision because immigrants who are more willing to return would work more to save faster. (Dustmann (2003) addresses this endogeneity issue, though, by instrumenting earnings using parental education.) Yang (2006) uses an exchange rate shock for immigrants from the Philippines as a source of exogenous variation in the purchasing power of their savings in order to distinguish between target-earnings motivation, where immigrants stay in the home country until their accumulated savings reach a threshold, and savings accumulation conjecture (called life-cycle considerations in his paper), where immigrants stay in the host country as long as the marginal benefit of higher savings there exceeds the marginal cost of staying. He concludes that overall the evidence supports the savings accumulation conjecture. In a similar approach, I use the exogenous variation in purchasing power parity between various source countries and Germany to test the savings accumulation conjecture. The quasi-experiment that Yang utilizes is the exchange rate crisis in the Philippines in 1997; as a result, there is only cross-country variation in the exchange rate shock whereas in my paper I use the variation in ppp over a long time period (16 years) as well as a cross-section of source countries. The immigrant pool in my sample is also different from that of Yang in the way that while the immigrants in Yang' analysis are mostly temporary migrants on short-term work contracts, the immigrants in my sample are longer-term immigrants.4 Another key difference in my analysis is that, unlike Yang, I allow the impact of ppp on the return rate to vary according to immigrants' age, which turns out to be very important empirically. Both micro and macro data are used in the estimation. The micro level data I use come from the German Socioeconomic Panel, which contains rich information on demographic as well as labor market outcomes of immigrants in Germany from various source countries. The sample is restricted to first-generation male immigrants. In addition, I employ macro data pertaining to immigrants' return decision. These macro data display variation both at the country level and over time. A key issue with the micro level data set utilized in this study is that it is a stock-sample of immigrants in Germany in 1984 from certain source countries. Like all stock-sample data, they are more likely to include immigrants with longer duration of residence. Not accounting for this would clearly result in biased estimates. However, my estimation strategy, which is based on duration analysis, accounts for the fact that the immigrants in the sample are those who survived long enough to be in the sample in 1984. The estimation results indicate that purchasing power parity is, in fact, an important factor that influences return migration behavior. A higher purchasing power parity decreases the return propensity of younger immigrants while increasing it for the older ones. When ppp increases by 10% from an initial level of one, the odds of returning for an 18-year-old immigrants fall by 37% while the odds of returning for a 50-year-old immigrants increase by 7%. These findings confirm the savings accumulation framework. Therefore, return migration in this context can be seen part of optimal life-cycle decisions, in which immigrants are in Germany to save. Yang (2006) finds that an increase in the purchasing power of immigrants' savings decreases their return rate. He does not distinguish among different age groups with regard to the impact of ppp on the return rate. However, his sample includes immigrants who are on average young: the mean age of the overseas workers in his sample is 34.4. Therefore, my finding that the impact of an increase in ppp on younger immigrants is negative is consistent with his finding. I also find that labor market outcomes are important determinants of return migration behavior. Bellemare (2003) as well as Constant and Massey (2003) report negative selection in terms of employment outcomes in return migration from Germany. However, my findings indicate that selection in return migration in terms of employment outcomes cannot be characterized independent of the length of unemployment spells. For immigrants who have been unemployed for less than 3 years, unemployment increases the return propensity. On the other hand, older working-age immigrants with longer unemployment spells are more likely to stay in Germany. That long-term unemployed immigrants are more likely to stay suggests that return policies targeting this group of immigrants such as financial bonuses conditional on return could be less of a burden on the German unemployment insurance system than the unemployment benefits that will be paid for many coming years in the generous German benefit system with relatively high replacement rates and long durations of entitlement. In fact, the estimation results indicate that a similar return policy implemented by the German government in 1984 brought about a major increase in the return rates of Turkish immigrants at that year. Next section discusses the conceptual framework. The data are described in section 3 and the estimation method is described in section 4. Section 5 presents the results and section 6 concludes.

In this paper, I test the savings accumulation conjecture proposed in rationalizing return migration decisions in the context of immigrants in Germany using exogenous variation in purchasing power parity, and examine how labor market outcomes like unemployment and retirement influence the return migration decision. In addition, I characterize the level and timing of return migration behavior as well as the selection in it, and derive a number of implications of these regarding the impact of immigrants. The empirical analysis uses a rich micro level data set (German Socioeconomic Panel) as well as macro level data that display both cross-country and time variation. In the estimation, I use a flexible duration analysis method. I find that purchasing power parity is in fact an important determinant of return migration behavior from Germany. A higher ppp decreases the return propensity of younger immigrants while increasing it for older ones. In other words, while the substitution effect, the opportunity cost of the purchasing power of foregone savings, dominates for younger immigrants; the income effect, the higher purchasing power of already accumulated savings, dominates for older immigrants. Quantitatively, a 25% increase in ppp from an initial level of 1 decreases the odds of returning by 64% for an 18-year-old immigrant while increasing it by 16% for a 50-year-old immigrant. This confirms the conjecture that an important motivation of these immigrants' residence in Germany is to accumulate savings. Therefore, return migration, in this context, can be seen as part of optimal life-cycle location decisions, where immigrants return when the marginal benefit of higher savings falls below the marginal cost of staying. Moreover, the fact that these immigrants migrate to Germany primarily to accumulate savings implies that the return of the accumulated savings in Germany along with the returning migrants could make an important contribution to the source country economies. Another contribution of the paper is with regard to the impact of unemployment on return migration decisions. The previous empirical literature uncovered negative selection in terms of employment status in return migration from Germany. However, what I find is that the answer to this question depends on the length of the unemployment spell. Immigrants whose unemployment spells are shorter than 3 years are in fact more likely to return. On the other hand, older working-age immigrants with longer unemployment spells are more likely to stay. Given the generosity of the German unemployment insurance system in terms of the duration of the benefits, it is no surprise that older long-term unemployed immigrants with a small chance of finding employment in either country prefer to stay in Germany whereas younger short-term unemployed immigrants are more likely to return once they realize that they will not be able to attain their savings accumulation goal due to poor employment prospects. Examining the distribution of unemployment spells of immigrants in Germany, I find that the proportion of unemployed immigrants for whom unemployment decreases return migration probability is much higher. That the long-term unemployed immigrants are more likely to stay and that there is very strong state dependence in unemployment suggest that return policies targeted toward this group, such as financial bonuses conditional on return, could be beneficial for the German unemployment insurance system. The amount of financial bonuses which would encourage these immigrants to return could be less than the total amount of unemployment benefits that are going to be paid in many years. In fact, the results of this study also indicate that a similar policy that was implemented in 1984 to encourage immigrants to return to their home countries increased the return rates of Turkish immigrants significantly at that year. Retirement has a strong impact on return migration of immigrants from Germany. The return probability within the first year after qualification for retirement increases more than sevenfold. This implies that return migration significantly alleviates the potential burden an aging immigrant population would exert on the health insurance system in Germany. Even when the German government pays the health expenses of an immigrant after return, it will be relatively less costly as health-care expenses are lower in the source countries than in Germany. This paper also characterizes the level and timing of return behavior. The level of return migration is quite high and exhibits significant variation according to country of origin. Immigrants from wealthier countries are more likely to return. However, the evidence for differences across the country of origin groups vanishes once the variations in the purchasing power parities and wage ratios with Germany as well as in individual characteristics are accounted for. The timing of return exhibits interesting differences by country of origin. The hazard function over age for Turkish immigrants has a hump-shaped profile that reaches its peak between the ages of 45 and 54. On the other hand, the hazard function of immigrants from EU countries is parabolic. Their hazard rates are higher at early ages and after retirement. Unemployment rates of Turkish immigrants in their fifties are very high. That many choose to return between the ages of 45 and 54 implies that return migration behavior brings about a much less take-up of unemployment benefits.23 Return migration at these ages also increases the net contributions to the pension insurance system for Turkish immigrants as periods of unemployment count toward the pension qualifying period while immigrants make no contributions at these periods.