Instituto I N S T Ide T Economía U T O D E E C O N O M Í A TESIS de MAGÍSTER DOCUMENTO DE TRABAJO 2012 Who Comes and Why? Determinants of Immigrants Skill Level in Early XXth Century US Matías Covarrubias. www.economia.puc.cl
Who comes and Why? Determinants of Immigrants Skill Level in early XXth century US Matías Covarrubias Second Draft Abstract This paper estimates the effect of changes in the determinants of migration on the average skill level measured by the quality of the occupations that immigrants to the United States performed in their origin country. For this purpose, I construct a panel of 40 countries with available data from 1899 to 1932. The results indicate that an increase in GDP per capita on the country of origin decrease the average skill level of immigrants and an increase in mobility costs has a positive effect on average skill level. The empirical findings are consistent with a model of intermediate selection where potential migrates must have both the resources to finance the migration cost (liquidity constraint restriction) and an expected income gain of migrating (economic incentives restriction), preventing workers in both tails of the skill distribution to migrate. Furthermore, this selection does not appear to impact the quality of occupations of migrants once in the US, suggesting great losses in terms of human capital for the host economy. This thesis was prepared for the Seminario de Tesis de Magister del EH Clio Lab (Conicyt PIA SOC1102). JEL codes: F22, H56, J61, O15 Keywords: immigration selection criteria, high skilled immigration, economic incentives for migration, political migration. May, 2012. Thesis written as a Master student at the Economic History and Cliometrics Lab (EH Clio Lab, Pontificia Universidad Católica de Chile, Department of Economics). This investigation has been supported through Conicyt Asociative Investigation Program SOC 1102. I would like to thank Jeanne Lafortune, Jose Tessada and all EH CLIO Lab members for their continuous guidance and support. The usual disclaimer applies. Any comment please send to the author s email address mcovarr1@uc.cl 1
1 Introduction What is the impact of immigration on domestic economies? Borjas and Friedberg (2009) argues that the skill level of immigrants is crucial in understanding this relationship for three reasons. First, who wins and who looses from immigration depends on the skill level of immigrants. Second, the assimilation process is different for each skill level as more skilled immigrants may assimilate faster. Finally, the skill level of immigrants may determine whether there are economic benefits of immigration or not. Thus, a better understanding of the determinants of the average skill level of immigrants is a valuable tool for interpreting the historical evidence on immigration inflows and their impact, for forecasting future trends in migratory movements, and for designing immigration policy. This paper attempts to identify and estimate the impact of the main theoretical determinants of the immigrants skill level by using a new set of administrative data from the Comission of Immigration previously digitalized by Lafortune and Tessada (2012) which includes a measure of skills of immigrants to the US from 1899 to 1932 based on their occupations in their country of origin 1. With this measure, I construct a panel data that allows us to test how variation over time and over country in characteristics of the country of origin affects the skill composition of immigrants inflows. This period is particularly useful to study the economic determinants of the migration decision because it was characterized by large and diverse immigration inflows and important restrictions were imposed over a previously unrestricted immigration process for many countries. This give us the opportunity to identify the determinants of the skill level of immigrants in a context of open gates and compare it to the restricted situation. Overview. I first setup a modified Roy selection model as in Borjas (1987) but with a fixed mobility cost component as in Chiquiar and Hanson (2005). Also, following Orrenius and Zavodny (2005), the model consider the fact that self-selected migrants should be able to finance the mobility cost in order to migrate.the model provides three main empirical predictions. If the liquidity constraint is binding, an increase in output on the origin country has an ambiguous effect on total flow of migrants and decreases the average skill level of migrants because it increases the amount of skilled workers that can afford the migration cost and reduces the amount of skilled workers that have economic incentives to depart. On the other hand, an increase in mobility costs reduces the total flow of migrants and has an ambiguous effect on average skill level of migrants because it prevents unskilled workers that cannot afford the migration costs to migrate and disincentives skilled workers to leave 1 Preliminary data covering 1873 to 1898 is also analyzed. 2
their origin country. Finally, an increase in inequality, everything else constant, has an ambiguous effect on both total flow of migrants and on the average skill level because even though it reduces the amount of unskilled workers that have enough savings to cover the migration costs, it can reduce or increase the economic incentives for skilled and unskilled workers to migrate. In order to test these empirical predictions I construct a panel of 40 countries with measures of average skill level, origin s country level of output, mobility costs and political instability. Average skill level is calculated using the mean of occupational scores associated to self-reported occupations that immigrants had in their origin country. Output in the origin country is measured by Purchasing Power Parity GDP per capita. To estimate mobility costs I use the product of the freight rate (cost of delivering a cargo from one point to another) for each year with the distance to the US for each country. Political Instability is measured by international wars and intranational wars involvement. With this panel I estimate OLS regressions of the average skill level against the mentioned explanatory variables adding year and country fixed effects to control for unobservables. As auxiliary regressions, I use the total flow, the flow of professionals, skilled and unskilled as well as the shares of professional, skilled and unskilled as the dependent variable. These auxiliary regressions allow us to identify the changes in quantities of immigrants from each skill category that drives the impact of the explanatory variables on average skill level. In a second stage, in consideration that data on inequality for the relevant period is not available, I regress the country fixed effects estimated in the main regressions against a proxy of country level inequality. The proxy consist in using the oldest gini data available for each country. The results confirm the theoretical finding that an increase in GDP has a negative effect over average skill level by altering the composition of the migrants towards a larger share of unskilled workers. Also, an increase in migration cost reduces the amount of migrants from all skill levels. For the period that we are studying the reduction of immigrants is proportionally more intensive for unskilled immigrants so the average skill level increases. Both findings are consistent with the micro data evidence presented by Orrenius and Zavodny (2005). Nevertheless, Borjas (1987) finds that bigger GDP per capita on the origin country implies bigger average wage in the US and that distance to the US, which is the time invariant component of our mobility costs, has no significant effect. 2 Finally, a cross-country analysis indicates, contrary to the finding of Borjas (1987), that an increase in inequality 2 These results are presented in table 5 of his paper. 3
has a negative effect on total amount of migrants and a positive effect on average skill level 3. These determinants of the average skill level, while having a significant effect over the quality of occupations that the immigrants had before migrating, does not have a significant effect over the quality of occupations that the immigrants had in the US, suggesting that the US labour market does not assimilate perfectly the skills of immigrants upon arrival. Related Literature. The first generation of studies regarding this subject (Chiswick, 1978; Carliner, 1980) found that after a relatively short adaptation period, earning of immigrants get to be bigger than earnings of comparable natives. The framework used by these studies to explain this result is that earnings of immigrants grow faster because they have more incentives to invest in human capital and they get to be even bigger because of positive self-selection: the foreigners that migrate from their origin countries are more able or motivated than the standard foreigners and also that the standard natives. As a reaction to this positive self-selection assumption, Borjas (1987) constructs a version of the Roy selection model (Roy, 1951) to analyze the migration choice of income maximizing agents with perfect information of earnings distributions on both the origin and destination country. His conclusion is that higher GDP and low political instability in the origin country result in more skilled immigrants. Also, positive self-selection will occur only if inequality on the origin country is smaller than in the US and correlation between wages in the origin country and the US are big. If the reverse is true, then we would have negative self-selection. Empirically, he finds that Eastern European countries, that have low inequality, provide immigrants that are succesful in the US. In contrast, less developed countries, that have higher inequality, provide immigrants that are unsuccessful in the US. Thus, his empirical study (weakly) supports his theoretical findings. Despite the results of Borjas (1987), controversy has arise because of critiques to the empirical design of Borjas paper (Jasso and Rosenzweig, 1990) and also because new studies have shown evidence against negative self-selection even in less developed countries (Chiquiar and Hanson, 2005; Mishra, 2007; Mckenzie and Rapoport, 2007; Orrenius and Zavodny, 2005). These papers found that migrants from Mexico come from the upper part of the distribution of skills of that country, but they still do worse than natives in the US. This result is also known as intermediate selection, in the sense that migrants actually come from the upper middle of the distribution 4. As a result, Orrenius and Zavodny (2005) and 3 A more exhaustive analysis of the differences between Borjas (1987) s and our results is presented on section 5. 4 The upper middle class selection hypothesis, supported by these authors, had been challenged with evidence from 2000 Census by Ibarraran and Lubotsky (2005) 4
Chiquiar and Hanson (2005) have incorporated in their theoretical models the fact that if the mobility cost is fixed then poor individuals will not migrate if they do not have enough resources to finance the mobility cost or if the cost is bigger than the differential of potential earnings. Also, Jasso and Rosenzweig (2008) developed a new framework making the point that self-selection according to economic incentives are not as relevant as distance to the US and immigration systems in determining the skill composition of immigrants. In this context, this paper contributes to the literature by using a panel data strategy that allows us to control for country specific and year specific factors, providing a test at the macro data level of the selection models that have been tested using census and micro data for more recent periods. Also, the mass immigration process that took place in the time period covered by this study has been empirically described in terms of assimilation of immigrants Abramitzky et al. (2012a) and broad self-selection (Abramitzky et al., 2012b) but the determinants of the skill composition has not been studied in this context. Furthermore, the early XXth century is a particularly interesting historical period for this purpose because it provides data before and after the several restrictions that were imposed predominately after the World War I. Finally, in contrast with other macro data analysis performed on this subject, my measure of skill is not based on the occupations that immigrants have in the US or their wages but instead I use data on the occupations immigrants performed in their origin country, which reflects better the skills composition if the US labour market take some time to detect and use immigrants skills, as suggested by Borjas (1987) and Lafortune and Tessada (2012) among others. Layout. The remainder of the paper is organized as follows. In Section 2 I present the theoretical model and his empirical predictions. Section 3 describes the empirical specification to be estimated, section 4 explains the data and section 5 presents the results. Section 6 summarize the results and conclude. 5
2 Theoretical Model I use a Roy selection model as in Borjas (1987) but with a fixed mobility cost component as in Chiquiar and Hanson (2005). Also, following Orrenius and Zavodny (2005), the model consider the fact that self-selected migrants should be able to finance the mobility cost in order to migrate. Thus, the model identify and analyze two main factors that determines wether a worker will migrate or not. First, the worker will migrate only if he has an expected income gain for doing so, after considering the relative skill prices between the countries and the mobility cost. I will refer to this factor as the economic incentive restriction. Second, if there is an economic incentive to migrate the individual must have enough savings to pay for the mobility cost. I will refer to this factor as the liquidity constraint restriction. Let i = 0, 1 denote the country, where 0 is the origin country and 1 is the destination country. Then, w i is the present value of the future earnings the potential immigrant can obtain in country i. µ i is the present value of the future base earnings that the potential immigrants would obtain in country i. x represents the skills of the immigrant and δ i is the skill price in country i. Therefore: ln w 0 = µ 0 + δ 0 x (2.1) ln w 1 = µ 1 + δ 1 x (2.2) The migration cost is M and I assume that is fixed and equal for all migrants. Economic Incentive Restriction. An individual with skills x will migrate only if he expects to have an income gain 5 : ln(w 0 ) < ln(w 1 M) < ln(w 1 ) + ln(1 + M w 1 ) < ln(w 1 ) M w 1 (2.3) Where we simplified the expression by approximating ln(1 + M w 1 ) M w 1. This approximation is accurate because the starting inequality is w 1 > M + w 0, so if the starting inequality holds then: 5 By focusing on expected income gain instead on utility gain I am assuming that individuals are risk neutral. If individuals are risk adverse the total flow of migrants is lower than the one predicted in this model but the comparative statics are the same 6
1 > M w 1 + w 0 w 1 >> M w 1 (2.4) In contrast, Borjas (1987), Chiquiar and Hanson (2005) and Orrenius and Zavodny (2005) assume that ln(1 + M w 0 ) M w 0, which is a less accurate approximation ( M w 0 > M w 1 and there is no certainty that M w 0 << 1). This distinction is relevant because the comparative statics resulting from both assumptions differ. Replacing the earnings equations 2.1 and 2.2, we get 6 : µ 0 + δ 0 x < µ 1 + δ 1 x Me µ 1 δ 1 x (2.5) The left hand side of this economic incentive restriction (henceforth LHS) represents the present value of earnings if he stay and the right hand side (henceforth RHS) represents the present value of earnings if he migrate. In Figures 1 and 2 these two expressions are graphed as a function of skills. If δ 0 > δ 1 the model suggest that individuals who have economic incentives to migrate come from the middle part of the skill distribution, because skilled individuals receive a better skill price in the origin country and highly unskilled individuals have a low earnings differential relative to the mobility cost 7. The two cut-offs that delimit the skills of individuals with economic incentives to departure are going to be referred as xmin EIR and xmax EIR. If δ 0 < δ 1, skilled individuals receive a bigger skill premium in the destination country so they have stronger economic incentives to departure than less skilled individuals. This case is referred as positive self-selection. Mathematically, we can corroborate this analysis by computing the slope of both curves: LHS x = δ 0 RHS x = δ 1 (1 + Me µ 1 δ 1 x RHS ) lim x x = δ 1 (2.6) 6 Another relevant feature of the model is that we do not make any further assumptions about migration costs. In contrast, Chiquiar and Hanson (2005) assume that M w 0 = µ δx in order to obtain a negative relation between this time equivalent expression of the mobility cost and skills, and Borjas (1987) ascertain that M w 0 is constant across individuals, thus assuming that richer migrants have bigger monetary migration cost, which has been widely criticized (Chiquiar and Hanson, 2005; Orrenius and Zavodny, 2005; Jasso and Rosenzweig, 2008) because there is no intuitive explanation or empirical evidence to make that claim and even though it simplifies the analysis it is not neutral and affects the conclusions of the model. 7 If µ 0 < µ 1 Me µ 1, then there is no unskilled individuals without economic incentives to departure. This case of negative self-selection is not analytically different from the intermediate self-selection case in the presence of liquidity constraints so I will not make the distinction in the analysis. 7
Figure 1: Economic Incentive Restriction (LHS<RHS) if δ 0 > δ 1. Figure 2: Economic Incentive Restriction (LHS<RHS) if δ 0 < δ 1. 8
Liquidity Constraint Restriction. Once the economic incentives restriction is satisfied, the worker must be able to finance the migration cost in order to migrate. Let S + sw 0 = S + se µ 0+δ 0 x be the savings of a potential immigrant with skill level x. Thus, he can only migrate if: S + se µ 0+δ 0 x > M x > ln ( ) M S s µ0 (2.7) δ 0 I will call this cut-off xmin LCR. The difference between this liquidity constraint restriction and the one presented in Orrenius and Zavodny (2005) is that this specification consider the fact that the contemporaneous component of savings depends on current wage and so an increase in the output 8 of the origin country helps the potential migrant to afford the migration cost. In Figure 3 we can observe that once the two restrictions are taken into account this model suggest that if skill price is bigger in the origin country migrants come from the middle part of the distribution, supporting the intermediate selection evidence found in (Chiquiar and Hanson, 2005; Mishra, 2007; Mckenzie and Rapoport, 2007; Orrenius and Zavodny, 2005). For countries with low skill price, positive selection may be observed. For the static comparative analysis, I will focus on the intermediate selection case and refer to the positive selection case only when its consideration affects the conclusions. 8 Changes in the output of the country are going to be represented as changes in the base wage parameters µ 0 and µ 1. 9
Figure 3: Distribution of skills in origin country. Individuals with skill levels between the two cutt-offs (colored) migrate. 10
Empirical Predictions. This model provide us with three main empirical predictions that our data will allow us to test. An increase in GDP per capita in the origin country reduce the average skill level of migrants unambiguously if the liquidity constraint restriction is binding (xmin LCR > xmin EIR ). As we can see graphically in 4 and mathematically in equation 2.8, an increase in GDP per capita reduce the amount of potential skilled migrants that has economic incentives to departure and at the same time it gives extra resources to low skilled workers that have economic incentives to depart but have not enough saving to cover the migration costs. In this case. The effect of an increase in GDP over the total quantity of migrants depends on wether the increase in low skilled workers is bigger or not than the decrease in skilled workers, which in turn depends on the exact location of the cut-off according to the parameters. Nevertheless, if the liquidity constraint is not binding (xmin LCR < xmin EIR ), less skilled workers with enough resources to migrate will have economic incentives to departure, so the aggregate response of average skill level will depend on wether the decrease in skilled migrants is more important or not than the decrease in unskilled migrants 9 and the quantity of migrants will decrease unambiguously. xmin LCR µ 0 = 1 δ 0 < 0 LHS µ 0 = 1 > 0 (2.8) Figure 4: Effect of an increase in the output of the origin country (µ 0 ) 9 If the liquidity constraint is not binding and the skill price in the origin country is smaller than in the destination country (positive self-selection case) the average skill level increase unambiguously 11
An increase in mobility costs has an ambiguous effect on the average skill level of migrants. In Figure 5 and equation 2.9 we can see that an increase in mobility cost affect the two cut-offs. Bigger mobility costs make the liquidity constraint more binding, which implies that less of the unskilled population will be able to migrate. Also, bigger mobility costs reduce the amount of potential skilled workers with economic incentives to migrate. Thus, the impact over the average skill level depends on which of these two effects is bigger 10. In contrast, an increase in mobility costs implies an unambiguous reduction in the total amount of migrants from all skill levels. Both results holds even if the liquidity constraint is not binding, because the economic incentive restriction for unskilled workers is also more restrictive when the mobility cost increase. xmin LCR M = 1 δ 0 1 M S < 0 RHS M = e µ 1 δ 1 x < 0 (2.9) Figure 5: Effect of an increase in mobility cost (M) 10 In the positive self-selection case there is only an impact on unskilled workers so the average skill level increase 12
The effect of an increase in inequality over average skill level of migrants is ambiguous. In order to analyze the effect of a mean preserving spread, we can write the earnings equations as a function of the standardized skill level: µ i + δ i x = µ i + δ i x + δ i σ x x x σ x = α i + δ i σ x z. (2.10) Where x is the average skill level, σ x is the standard deviation of the skills distribution z is the standardized skill level and α i = µ i + δ i x is a reparametrization. Thus, the new economic incentive restriction is: α 0 + δ 0 σ x z < α 1 + δ 1 σ x z Me α 1 δ 1 σ xz (2.11) And zmin LCR = ln ( ) M S s α0 (2.12) δ 0 σ x In our model, inequality is represented by the variance of the skills distribution (σ x ). A bigger inequality in the country of origin, for the same average GDP, implies a more restrictive liquidity constraint restriction because unskilled workers have less income. In contrast, the effect of a bigger skills inequality over the economic incentives cut-off is ambiguous. zmin LCR σ x = α ( 0 ln M S ) s δ 0 σx 2 > 0 LHS σ x = δ 0 z > 0 RHS σ x = δ 1 z+mδ 1 ze α 1 δ 1 σ xz > 0 (2.13) 13
Figure 6: Effect of an increase in inequality in the origin country (σ x ) As we can see in Figure 6, if the liquidity constriaint is binding the aggregate response of the average skill level will depend on whether the reduction in unskilled individuals because of more restrictive liquidity constraint is more important or not than the reduction in skilled individuals due to less economic incentives to departure. If the liquidity constraint is not binding, the response of the average skill level depends on wether economic incentives for the unskilled workers at the margin increase or decrease. If zmin EIR increase (economic incentives for unskilled workers at the margin decrease), then there are less skilled migrants an less unskilled migrants as well, so the impact on average skill level is ambiguous. zmin EIR decrease, then the average skill level decrease unambigously (Mathematically, if LHS(zmin EIR) σ x = δ 0 zmin EIR is larger (smaller) than RHS(zmin EIR) σ x = δ 1 zmin EIR + Mδ 1 ze α 1 δ 1 σ xzmin EIR, then zmin EIR decrease (increase)). If 14
3 Empirical Model In order to evaluate the empirical predictions, we use the regression equation: ln x ct = β 0 + β 1 ln u 0ct + β 2 ln M ct + β 3 W ct + α 2 φ c + α 3 θ t + ɛ ct (3.1) Where c represents the country of origin, t a given year and x represents the average skill level of immigrants. u 0ct denote the output level in the origin country and M ct the migration cost. W is a control variable representing noneconomic migration factors as political instability, φ c and θ t are country and time dummies and their coefficients α 2 and α 3 are the time and country fixed-effects. ɛ ct is the error term. This specification is commonly used in panel data literature. For example, Mayda (2010) performs the largest panel data analysis on determinants of migration using the same specification. The logarithmic assumption on the explanatory variables is made to control for changes in percentage as opposed to levels in GDP so the scale of the country is taken into account and does not distort the analysis. The logarithmic assumption on the dependent variable allows us to estimate elasticities and semielasticities and thus compare the coefficients associated to the different skill level measures that I will use. The country and year fixed effects, which represents the main gain from using a panel data, control for idiosyncratic and unobservables characteristics of each country and year. The country fixed effects capture the time invariant cross country variation that affect the skill level of immigrants as it can be the culture of the country or the persistent component of inequality. On the other hand, the year fixed effects controls for every shock to skill level of immigrants in a particular year that is common to all the countries. Very importantly, this kind of shocks include the economic conditions on the US, that are very relevant for the economic incentives component of the migration decision as our model suggest. In order to test the cross country effect of persistent inequality I will use the country fixedeffects estimated in the regression 3.1: ˆα 2 = β 4 + β 5 σc 2 + e c (3.2) This methodology of identification in two stages is based on the fact that country fixedeffects capture the cross-country differences in skill level that are not explained by the independent variables on the first stage 11. This empirical setting resembles the more informal analysis performed by Borjas (1987) to test his empirical prediction about the effect of inequality over immigrants skill. 11 This also implies that there is no point in including those controls in the second stage regression 15
In an ideal empirical model, we would like to test the effect of inequality on immigrants skills using not only cross country variation but also time variation, allowing us to use the fixed effects in order to control for non observables related to the specific periods or origin country. That strategy would require flow inequality data as skill prices or alternatively a inequality by cohort measure taken out of a age-cohort-period decomposition, because the stock inequality as measured by GINI is very persistent over time and does not capture much underlying variation on inequality over time. Provided that there is no inequality data (either flow or stock ) for such an old time span, and in consideration that inequality is very persistent, we can only use modern cross country inequality data to perform this empirical exercise. 4 Description of the Data Having described the empirical strategy, I now turn to the construction of the variables of interest. The regression presented above requires annual GDP variation at the country level. This reduce our sample to 40 countries for which such data is available. Table 1 shows the mean of all the relevant variables for each of these countries. Even though many countries are left out of the sample, the 40 countries included represent a 84% of the total immigrant flow, as they included all the large European countries with the exception of Russia (whose GDP data is not available for this period) and the largest American and Asian countries (including Australia). Amount and skill level of immigrants (x). Quantity and skill level of immigrants for each country were taken from the Report of the Comissioner of Immigration (henceforth RCI), which correspond to administrative data presented as summary tables based on questionnaires that every immigrant had to answer at departure from the origin country or at their arrival to the US 12. The RCI was published annually from 1899 to 1932 (except for 1931) and provides data on the ethnic origin and on self-reported occupation that each immigrant performed before migrating. This data is useful to evaluate our empirical question because in contrast with other macro data used in the literature, that measure skill level through the occupations or the wages the immigrants had in the US, the RCI allows us to measure the quality of the occupations immigrants had before they arrived to the US, which is a better measure of the actual skills if, as suggested by the literature (Borjas, 1987; Lafortune and Tessada, 2012), the immigrants take some time to assimilate to the US 12 This database was previously digitalized by Lafortune and Tessada (2012) 16
Table 1: Descriptive Statistics by Country country Avge. flow Avge. flow Avge. flow Avge. flow GDP cost Sum Inter. Sum Intra. Professionals Skilled Unskilled Wars Wars Argentina 110 16 34 41 3510 10617 0 3 Australia 1044 145 399 223 4507 20166 0 0 Austria 10262 243 2987 6710 2740 9011 5 0 Belgium 1203 75 356 720 3739 7862 5 0 Brazil 103 13 33 38 815 8591 0 5 Bulgaria 7707 73 713 6810 1245 10009 6 1 Canada 15896 1457 4968 8883 3184 927 0 0 Chile 90 16 21 35 2286 10210 5 1 China 4197 58 33 4101 560 14096 8 19 Colombia 117 14 35 49 1225 4839 0 9 Costa Rica 91 15 15 36 1623 4170 0 0 Czechoslovakia 22125 91 1621 20217 2366 8729 1 0 Denmark 3891 103 995 2708 3278 8237 0 0 Finland 3764 23 325 3399 1699 8768 3 2 France 4284 331 1067 2782 2931 7795 11 4 Germany 28197 879 8242 18754 2838 8095 6 1 Greece 8741 48 666 7976 1748 10439 9 0 Guatemala 97 15 18 40 1490 3930 3 0 Honduras 77 12 18 27 1370 3710 2 1 Hungary 8706 69 662 7906 2093 9281 6 2 India 175 15 11 149 643 15230 0 0 Italy 60640 466 8816 51040 2128 9127 6 0 Japan 4232 215 239 3752 1277 13787 12 7 Korea 4601 167 1831 2429 1071 14121 0 1 Mexico 8717 213 984 7402 1665 3840 0 20 Netherlands 2254 102 568 1505 3773 7832 0 0 Nicaragua 92 18 20 30 1349 3936 1 0 Norway 7544 163 1731 5566 2111 7883 0 0 Peru 86 17 24 24 1129 7166 5 4 Poland 23324 77 1421 21669 1730 9077 2 0 Portugal 2649 22 257 2361 1209 7257 3 0 Romania 4467 59 854 3459 1321 10090 4 1 Spain 3081 117 811 2020 1916 7704 3 4 Sweden 14090 179 2212 11614 2464 8393 0 0 Switzerland 2834 156 713 1643 3825 8349 0 0 Turkey 1687 26 275 1369 1072 11037 15 16 United Kingdom 33592 1650 9195 21694 4376 7461 7 3 Uruguay 81 16 16 24 2700 10726 0 1 Venezuela 74 13 16 26 1442 4191 0 4 Yugoslavia 13875 82 1071 12611 1136 9601 7 0 Average 8163 192 1447 6389 2386 8657 3 3 17
labour market and also the US labour market takes some time to identify or use the skills of immigrants. Furthermore, the time span covered by the data allow us to identify with more precision the self-selection issues involved in the migration decision, because before the restrictions on immigration imposed in the 1920 s, immigrants from most countries had no major obstacles entering the US 13, so the only factors determining the characteristics of immigrants where wether they have incentives to migrate and the means to afford it. This feature of the period also validates the self-reported information as there were no incentives to lie on past occupation. Despite of the advantages of the RCI data, there were also challenges on the adaptation of the data to our empirical design. First of all, occupational data must be matched with some quality indicator. For this purpose I use the occupational standing variables from the United States Census as presented in the Public Use Micro Sample (henceforth IPUMS). These variables are quality measures associated to the occupational classification of the 1950 United States Census. The basis for each quality score and the source data are presented in Table 2. In table 2, we can see the correlation between these variables for individuals included in the 1900, 1910, 1920 and 1930 US census. The first three variables, associated more strongly to income and prestige, and the last three variables, associated to education and earnings, have big correlation between them and less correlation with the other three variables. This distinction will be useful in the empirical results analysis. Table 2: Occupational Standing Variables Description Variable Label Basis of score Source data occscor Occupational Income Score Income 1950 census sei Duncan Socioeconomic Income, Education, 1950 census Index Prestige 1947 surveys presgl Siegel Prestige Score Prestige 1960s surveys erscor50 Occupational Earnings Score Earnings 1950 census edscor50 Occupational Education Score Education 1950 census npboss50 Nam-Powers-Boyd Earnings, Education 1950 census Occupational Status Score 13 in section 5 I identify those unrestricted countries and provide a more detailed description of the restrictions on immigration and their effect on skills composition of immigrants 18
Table 3: Correlation Between Occupational Standing Variables Corr occscore sei presgl erscor50 edscor50 npboss50 occscore 1 sei 0.9244 1 presgl 0.9374 0.9556 1 erscor50 0.8555 0.8203 0.8695 1 edscor50 0.766 0.8047 0.8411 0.9533 1 npboss50 0.8586 0.8746 0.8992 0.9877 0.9696 1 Before applying these quality measures to our occupational information, we must match the occupations from the RCI to the occupation classification of 1950 census. To accomplish that I used the matching constructed by Lafortune and Tessada (2012) 14. Another challenge is presented by the fact that the information taken from RCI data is aggregated at ethnic group level and the GDP, cost and wars variables are measured at a country level. The approach I took is to transform ethnic group level occupational data to country level data. In order to do that I used again the matching between countries and ethnic group done by Lafortune and Tessada (2012). Provided that there is more than one country matched to each ethnic group, I divided the flow of the ethnic groups between its corresponding countries calculating, for each occupation and year, the share of the total inflow of an ethnic group that correspond to a particular country according to the IPUMS data. That is, I took IPUMS data on amount of immigrants from each country and occupation that arrived in a particular year, added them into ethnic groups, and then calculated the shares of each country. Then, I used the shares for each year and occupation to divide the inflow of an ethnic group that appear on the RCI data on countries. This methodology is assuming that immigrants that stayed in the US, and thus appeared in the decennial census, are randomly selected from each country inside an ethnic group, so the shares that appear in the IPUMS are a good approximation of the actual shares at the time of entrance to the country. As an alternative methodology, I used the same shares calculated for each country-year-occupation to aggregate the independent variables at ethnic group level. The differences between these two methodologies are explained in the empirical results section. As complementary data, I use the Foreign Commerce and Navigation of the United States Report, which presents data on the flow of immigrants from each country classified in Professional, Skilled and Miscellaneous from 1873 to 1898 (except from 1892 to 1896) 15. The problem with this data is that there is many missing GDP information for that period, so 14 For some of the RCI occupations there were more than one 1950 census occupation. In those cases, I take the average of the quality measures associated to the different 1950 census occupations 15 the data actually have detailed information on occupations but the digital tables are in a bad condition so most of the detailed data is incomplete, which prevents us from using it. 19
we do not have enough data to perform a full analysis. Despite this drawback, we present the results of the regressions adding this data in appendix 2. Figure 7 presents the total flow divided by skill level of all the immigrants for each year. This aggregated data give us a broad understanding of the immigration process that is the subject of this study. The first important feature of this data that is worth mentioning is the big fall in total flow that went down on 1890 until 1898. Even though we have missing data between 1893 and 1897, the low levels of immigration flows in 1891, 1892 and 1898 confirm the historical finding that the US bad economic conditions of the 1890s 16 made migration to the US undesirable (O Rourke and Williamson, 1999). After that slum, from 1899 to 1914 we observe a big wave of immigration that is dramatically interrupted by World War I (henceforth WWI). Another historical event that had evident impact on immigration is the Johnson-Reed Act of 1924, that imposed restrictive quotas on immigrants from all countries, specially from eastern Europe and Asia. These quotas had an effect on the flow of immigrants of the three skill levels, not only on unskilled immigrants. Figure 7: Total, Professional, Skilled and Unskilled Flow for each year Finally, I also use the IPUMS data on the number of immigrants, their occupation and the year they immigrated from the decennial observation of the United States Census. This data give us the opportunity to test the US labour market assessment of the quality of immigrants and contrast that assessment with the ex ante occupational data of the RCI. 16 In particular, the recession of 1890-1891 and the panic episodes of 1893 and 1896. 20
This exercise provides new evidence on the hypothesis that US labour market does not immediately identify or use the skills of immigrants but only after an adaptation time. Migration Cost (M). This variable is constructed using the distance to the US multiplied by the freight rate (cost of delivering a cargo from one point to another) of transporting grain from Europe to East Latin America for each period taken from Mohammed and Williamson (2004). Even though there are freight rates available for other routes and commodities, the freight rates between routes and commodities are very correlated and the route I selected have complete information with no missing data for the time span of the study. Figure 6 shows the freight rate for each period. Coherently with the fall on immigration flow during the WWI observed in Figure 7, freight rates during that period had a huge spike reaching a peak in 1917, followed by a normalization. Figure 8: Freight Rates per capita for each year GDP per capita (µ o ). I took Purchasing Power Parity GDP per capita data available from Maddison s Historical Statistics of the World Economy database. The sixth column of Table 1 provides evidence of the big heterogeneity in GDP level between countries on the 1873-1932 period, ranging from an average GDP per capita of 560 for China to 51040 for Italy. Wars (W ). This variable is constructed using two war dummies available from the Correlate of Wars project. The first dummy indicates for each country and year the participation 21
of that country in any international war. I will call this dummy interwars. The second dummy does the same but for intranational wars. I will call this second dummy intrawars. Is important to test the effect of these dummies separately, instead of adding them into a wars indicator, because in the case of intranational wars the migration possibilities are bigger, as a worker can migrate to neighbor or closer countries that are not involved in wars, option that is typically not available in international wars as the WWI. The last two columns of Table 1 show the sum of the international wars and intranational wars where each country was involved. Inequality (σ 2 ). This parameter will be estimated using the inequality data from Deininger and Squire (1996) database. This dataset encompass all the inequality data that is available from different sources. As it was stated in the presentation of the empirical model, there is no inequality data for the time span of our study so I will proxy gini of each country the first available gini data. In Appendix 1 the gini of each country and the year it was taken from is presented. This methodology relies on the persistency of the stock inequality measures to assume that gini data from mid XXth century is a good approximation of gini on early XXth century. This measurement error should attenuate the coefficients estimates. 5 Empirical Results In Table 4 I present the results of estimating our empirical model using the six occupational standing variables. The time period covered by the data included in these regressions goes from 1899 to 1932 17, which is the largest period with detailed occupational information available. For the three first occupational standing variables, associated with income and prestige, the results show that an increase in GDP per capita decreases significantly the average skill level of migrants. For the other three variables, associated with education and earnings, the result is the same in sign but it is not significant (due mostly to larger standard deviation). This is consistent with the theoretical finding that, as explained by Figure 4, an increase in origin s country output decreases the economic incentives for highskill individuals to migrate and relax the liquidity constraint, resulting in more low-skill individuals migrating if the liquidity constraint is binding. The magnitude of the estimated income elasticities range between a 0.11& to a 0.25% decrease in the occupational standing scores as a response to a 1% increase in GDP per capita. Qualitatively, this empirical finding is coherent with micro data evidence from Orrenius and Zavodny (2005) who find the same relation between output and skill level using variation between regions in Mexico. In contrast, Borjas (1987) using a cross country analysis find that countries with bigger GDP 17 For 1931 the occupational information is missing. 22
per capita have better wages in the US. The differences between these two results may be explained by the fact that wages in the US can correlate better with skill for richer countries with similar production structures 18 or by unobservables that a cross-country design is not able to capture. On the other hand, the coefficients for migration cost in Table 4 show that an increase in migration costs increases the average skill level of immigrants. This result suggest, in terms of the analysis presented in Figure 5, that the increase on average skill level due to a more restricting liquidity constraint for unskilled individuals is more important than the reduction on average skill level produced by the economic disincentive for professionals and skilled individuals to migrate 19. Again, this result is equal to the one found in Orrenius and Zavodny (2005) but differ from Borjas (1987). In particular, Borjas (1987) find that distance to the US, which is the time invariant component of our mobility cost has no significant effect over the wages on the US of the immigrants. The fact that our results about the effect of mobility cost over skill level of migrants differ from the ones found by Borjas (1987) does not necessarily implies that one of the results is empirically wrong. Both results can be reconcilliated if we consider that theoretically both a negative or a positive effect can be found depending on the relative relevance of the liquidity constraint restrictions and the economic incentives restriction. Borjas result is obtained analyzing the 1970 and 1980 US Census, covering a period where the countries of origin of immigrants were much richer than in the period analyzed in this paper. For richer countries, the liquidity constraint restriction is binding for a smaller proportion of the population and the economic incentive restriction is active for more people (in Figure 3, both cut-offs are more at the left). Hence, it is plausible than for more recent migration movements as the one analyzed by Borjas (1987) the effect of an increase in mobility costs over the economic incentives restriction dominates his effect over the liquidity constraint restriction, decreasing the average skill level, while the contrary being true for the 1899-1932 period. Finally, the involvement in intranational wars and international wars does not appear to have a significant effect on the skill level of migrants. 18 This reasoning is made by Borjas (1987) to explain the cross country variation in the correlation between origin and destination wages. 19 if the liquidity constraint is not binding, then the theoretical interpretation is that economic disincentive is persuading more unskilled individuals than skilled one to stay in the origin country. 23
Table 4: Skill Level Regressions: 1899-1932 (1) (2) (3) (4) (5) (6) VARIABLES logoccscore logsei logpresgl logerscor50 logedscor50 lognpboss50 loggdp -0.11*** -0.18* -0.10** -0.11-0.25-0.13 (0.04) (0.10) (0.04) (0.09) (0.17) (0.08) logcost 0.09*** 0.10*** 0.07*** 0.13*** 0.20*** 0.11*** (0.00) (0.01) (0.00) (0.01) (0.01) (0.00) interwars -0.03-0.03-0.02-0.02-0.04-0.03 (0.02) (0.03) (0.02) (0.03) (0.06) (0.04) intrawars -0.00-0.04-0.03-0.00-0.07-0.01 (0.04) (0.11) (0.05) (0.07) (0.17) (0.09) Observations 1,026 1,026 1,026 1,026 1,026 1,026 R-squared 0.73 0.76 0.70 0.70 0.78 0.75 *** p<0.01, ** p<0.05, * p<0.1 Notes: Clustered standard errors at a country level. Year and country fixed-effects included. The dependent variables are the average score in occupational standing variables (see Table 2). An observation correspond to a country in a year. Forty countries are included in this regression. In order to corroborate that the mechanisms presented in the model are actually leading the results of Table 4, Table 5 shows the effect of our explanatory variables over the quantities and shares of professionals, skilled and unskilled individuals that immigrate from each country. Coherently with our model, an increase in GDP decrease the amount of professional and skilled migrants and increase unskilled migrants, with a bigger impact on unskilled migrants. This results, while suggestive of a binding liquidity constraint and consistent with our empirical predictions about both the economic incentive restriction and the liquidity constraint restriction, are not statistically significant. Nonetheless, these underlying trends in data are driving the results of columns (2), (3) and (4), from which we can state that the share of unskilled immigrants increase with statistical significance. Also, an increase in GDP per capita of a country increases nonsignificantly the total flow of immigrants from that country. Theoretically, an increase in GDP can either result in a larger or a smaller total flow, depending on wether the extra flow resulting from the liquidity constraint relaxation is bigger or not than the reduction in total flow due to the economic disincentive to migrate, which in turn depends on where those two margins are positioned on the distribution of skills according to their income, mobility cost and skill price parameters, so it is not surprising that the estimation of GDP coefficient on total flow and on professional, skilled and unskilled flow have big standard deviations. On the second row, we can note that an increase in mobility cost significantly reduces the flow of professionals, skilled and unskilled individuals, which is consistent with the theoret- 24
ical finding that an increase in mobility cost reduces economic incentives for high skilled individuals to migrate and increase the liquidity requirements for low skilled individuals. Also, as can be predicted from Table 4, the reduction in flow is more important for unskilled individuals so the share of unskilled immigrants decreases, the share of skilled immigrants does not change significantly and the share of professionals immigrants increases. In terms of magnitude, it is important to notice that a 1% increase in mobility costs have an estimated impact over total flow of unskilled workers of a -1.64%. This huge estimated impact may be explained by both the liquidity constraint restriction and by the fact that an increase in mobility cost also disincentive unskilled individuals to migrate as it can be seen in Figure 5. As for the intrawars coefficient estimation, we can conclude that the involvement on international wars does not affect significantly either the quantity of migrants or the skill level composition of the flow. In contrast, the point estimate of the impact of intranational wars involvement over the total flow is even bigger in magnitude than the income elasticity, but we can see that the extra flow is more evenly distributed among skill levels so the skill composition of the total flow does not change. Table 5: Quantities and Shares Regressions: 1899-1932 (1) (2) (3) (4) (5) (6) (7) VARIABLES log total flow share share share log total flow log total flow log total flow professionals skilled unskilled professionals skilled unskilled loggdp -0.28-0.02-0.09* 0.10* -0.04-0.08 0.49 (0.67) (0.05) (0.05) (0.05) (0.31) (0.50) (0.62) logcost -1.36** 0.02*** 0.13*** -0.15*** -0.70*** -0.60*** -1.64*** (0.52) (0.00) (0.00) (0.00) (0.02) (0.03) (0.04) interwars -0.13-0.01 0.01-0.01-0.12-0.07-0.10 (0.37) (0.01) (0.02) (0.02) (0.18) (0.21) (0.26) intrawars 0.33-0.01 0.02-0.01 0.22 0.47 0.27 (0.50) (0.02) (0.06) (0.07) (0.33) (0.34) (0.69) Observations 1,319 1,026 1,026 1,026 1,026 1,026 1,026 R-squared 0.81 0.55 0.58 0.67 0.87 0.87 0.83 *** p<0.01, ** p<0.05, * p<0.1 Notes: Clustered standard errors at a country level. Year and country fixed-effects included. The dependent variables are quantities and shares of immigrants for each skill category. An observation correspond to a country in a year. Forty countries are included in this regression. The Quantities and Shares Regressions that incorporate the pre 1989 data are presented in Table A2.1 on Appendix 2. 25
Second Stage: Gini Regressions. Table 6 shows the regression of the country fixed effects estimated by the Skill level and Quantities Regressions for 1899-1932 (Table 4 and 5) against a proxy of the gini of each country 20. The results indicate that more unequal countries provide immigrants with a higher skill level, In terms of our empirical predictions, this would be the case if the reduction in low skilled migrants is more important in terms of the average skill level than the reduction on high skilled migrants. Also, as presented in column (1), gini appears to have a strong negative effect over the total flow of migrants which is what the model predicted if the liquidity constraint is binding and the economic incentives to migrate for skilled individuals decrease as representd in Figure 6. We should be careful with interpreting these results as conclusive evidence, because we are only using cross country data that does not allow us to control for country unobservables with fixed effects as we have done in the previous regressions. Nevertheless, this kind of simple exercise is analogous to the one that Borjas (1987) uses to confirm his hypothesis that bigger inequality implies less skilled immigrants. Table 6: Fixed Effects Regressions Against Gini: 1899-1932 (1) (2) (3) (4) (5) (6) (7) VARIABLES log Total logoccscore logsei logpresgl logerscor50 logedscor50 lognpboss50 Flow FE FE FE FE FE FE FE loggini -4.40*** 0.40*** 0.65*** 0.30*** 0.68*** 1.25*** 0.72*** (0.79) (0.08) (0.12) (0.06) (0.14) (0.25) (0.14) Constant 15.15*** -1.80*** -2.78*** -1.36*** -2.85*** -5.12*** -2.97*** (2.85) (0.27) (0.44) (0.22) (0.49) (0.90) (0.51) Observations 38 38 38 38 38 38 38 R-squared 0.46 0.46 0.37 0.37 0.43 0.38 0.40 *** p<0.01, ** p<0.05, * p<0.1 Notes: Robust Standard Errors in parenthesis. Year and country fixed-effecs included. The dependent variables are the country fixed effects estimated in the regressions presented in Table 4 and Table 5. The regressions are weighted by the inverse of the standard deviation estimated for each fixed-effect to account for the fact that the dependent variable is an estimation (this increase the standard deviation but does not affect the coefficients estimation). An observation correspond to a country. The same exact argument that was made to explain the differences in the estimated effect of mobility cost between Borjas (1987) and these results can be made to explain why both Borjas estimation of the negative impact of inequality and our estimation of a positive impact can both be correct. Again, it is possible that for the period analyzed by Borjas the 20 are weighted by the inverse of the standard deviation estimated for each fixed-effect to account for the fact that the dependent variable is an estimation (this increase the standard deviation but does not affect the coefficients estimation). 26
liquidity constraint restriction was less important than for the 1899 to 1932 period. Skill Level Regressions including total flow as a control. In Table 7 the total flow of immigrants is added as a control in the skill level regressions. The purpose of these tables is to test the relation between quantity and quality and thus investigate if the effect of the theoretical determinants of average skill levels operates only through changes in quantities or also through changes in the composition of immigrants. Moreover, with this regression, we can test if changes in quantities are usually more intensive in proportion of professional workers (as a theory of negative selection would suggest) or in unskilled workers. The results indicate that even after controlling by quantities the main determinants of skill composition have a significant effect. In the case of GDP the sign and magnitude of the coefficient is similar than the ones previously found. This is not a surprise since theoretically changes in the origin country GDP affect strongly the composition of immigrants while their effect on quantity is ambiguous. On the contrary, the coefficient associated with mobility cost has not a unique sign between regressions as in previous regressions. This is also consistent with the empirical prediction that an increase in mobility cost decrease the quantity of immigrants but his effect on skill composition is ambiguous. Finally, quantity is found to be negatively correlated with average skill level, which indicate that changes in quantities are more intensive in proportion of unskilled workers. Table 7: Skill Level Regressions Controlling by Total Flow: 1899-1932 (1) (2) (3) (4) (5) (6) VARIABLES logoccscore logsei logpresgl logerscor50 logedscor50 lognpboss50 logtotflow -0.06*** -0.14*** -0.08*** -0.11*** -0.28*** -0.13*** (0.01) (0.01) (0.01) (0.01) (0.03) (0.01) loggdp -0.10** -0.14*** -0.08** -0.08-0.18* -0.10 (0.04) (0.05) (0.03) (0.11) (0.10) (0.08) logcost 0.04*** -0.02** 0.00 0.04*** -0.04* -0.00 (0.01) (0.01) (0.01) (0.01) (0.02) (0.01) interwars -0.03-0.03* -0.02-0.02-0.04-0.03 (0.02) (0.02) (0.01) (0.03) (0.04) (0.03) intrawars 0.02 0.01-0.00 0.03 0.03 0.03 (0.02) (0.04) (0.02) (0.03) (0.06) (0.03) *** p<0.01, ** p<0.05, * p<0.1 Notes: Clustered standard errors at a country level. Year and country fixed-effecs included. The dependent variables are the average score in occupational standing variables (see Table 2). An observation correspond to a country in a year. Forty countries are included in this regression. Restrictions Impact Assesment. The first important restriction imposed in the period under scope is the Chinese Exclusion Act of 1882 that prohibited all immigration of Chinese unskilled workers (O Rourke and Williamson, 1999). The prohibition was a reaction 27
to the strength in competition of the supply of low skilled workers in California. In Figure A3.9 we can observe the huge reduction in immigration from china produced by this restriction. The next big restriction imposed was the 1917 Immigration Act 21. This act exclude from entry immigrants from all the countries inside the Asian Barred Zone that included all the major countries of Asia with the exception of Japan and the Philippines. For example, in Table A3.9 we can confirm that the year after 1917 the inflow from Korea decreased enormously. Later, in 1921, a new Immigration Act aimed at reducing the inflow from eastern and south europe by the imposition of quotas that were calculated as a 3% of the foreign born population from each country that appeared in the 1910 census. These quotas were more restrictive for the late waves of immigration that came from southern and eastern Europe and that in 1910 had not yet settled massively in the US. Finally, in 1924 the Johnson-Reed Act of Immigration recalculated the quotas as a 2% of the population with foreign origin 22 from each country that appeared in the 1890 census and prohibited the entry of Japanese unskilled workers. This final Immigration Act, that was not revised until 1952, restricted immigration from all the countries in Europe. The effect of this restriction over European immigrations can be corroborated in Figures A3.4 to A3.8. In contrast, any of these immigration acts were imposed over Latin American countries. Using this historical data I construct a dummy that indicates for each country the years after binding restriction were imposed on immigrations. Thus, China start being restricted after 1882, Asian countries excluding Japan start being restricted after 1917, Eastern and Southern European countries start being restricted after 1921and finally the rest of Europe and Japan start being restricted after 1924. In Table 8 the effects of the restrictions can be clearly seen. As expected, restrictions decrease strongly the total flow of unskilled and skilled while the decrease on flow of professionals is not as significant. Table 8: Descriptive Statistics for restricted and unrestricted observations in 1899-1932 Total flow Total flow Total flow Total flow Ocupational Professionals Skilled Unskiled Score Not Restricted 6185.184 141.6814 1445.837 4597.666 23.88785 Restricted 3956.675 221.7314 1112.023 2622.92 21.45595 Total 5992.964 148.5862 1417.044 4427.333 23.55017 In Table 9, we use the dummy to estimate the impact of regulation over the selection process and over the effect of the determinants. In order to do that, interactions with the logarithm of GDP and the logarithm of mobility cost are included. Interestingly, the interactions 21 A good account on the history of restrictions in this period can be found at the US government official web page http://history.state.gov/milestones/1921-1936/immigrationact. 22 The new calculus included US natives with foreign origins, not only foreign born immigrants 28
suggest that the restrictions reinforce the selection process, making the average skill level more sensible to changes in GDP and in mobility costs. This result is puzzling, since the priorities on quotas distribution was family reunification motives and skilled immigrants. Further research is required to rationalize this empirically strong result. Table 9: Skill Level Regressions with Dummy for Restricted Countries (1) (2) (3) (4) (5) (6) VARIABLES logoccscore logsei logpresgl logerscor50 logedscor50 lognpboss50 loggdp -0.08*** -0.12** -0.06** -0.03-0.09-0.05 (0.03) (0.05) (0.03) (0.05) (0.10) (0.05) logcost 0.09*** 0.09*** 0.07*** 0.13*** 0.19*** 0.10*** (0.01) (0.01) (0.01) (0.01) (0.02) (0.01) loggdp*drestricted -0.06*** -0.11*** -0.08*** -0.19*** -0.34*** -0.20*** (0.02) (0.03) (0.02) (0.04) (0.06) (0.04) logcost*drestricted 0.07*** 0.13*** 0.08*** 0.18*** 0.35*** 0.20*** (0.02) (0.03) (0.02) (0.03) (0.06) (0.03) interwars -0.02-0.01-0.01-0.01-0.00-0.01 (0.01) (0.02) (0.01) (0.03) (0.04) (0.03) intrawars -0.01-0.04-0.03-0.01-0.08-0.02 (0.02) (0.05) (0.02) (0.04) (0.08) (0.04) Observations 1,026 1,026 1,026 1,026 1,026 1,026 R-squared 0.75 0.79 0.72 0.72 0.80 0.78 *** p<0.01, ** p<0.05, * p<0.1 Notes: Clustered standard errors at a country level. Year and country fixed-effects included. The dependent variables are the average score in occupational standing variables (see Table 2). An observation correspond to a country in a year. Forty countries are included in this regression. Regressions Using IPUMS Data. There are two main differences between using IPUMS occupational data and using, as in the previous regressions, the RCI administrative data. First, immigrants that appear on the decennial US census are the one that stayed in the country while RCI data capture all the immigrants that arrived to the US. Hence, IPUMS data reflects in part the return migration selection process. Second, occupational data from IPUMS describe the occupations that immigrants had on the US, while the RCI data contains information on the occupations immigrants performed in their origin country. In Table 10 it can be appreciated that when IPUMS data is used to construct the dependent variables our explanatory variables do not have an effect over the average occupational scores. This important difference with our previous findings cannot be entirely attributed to return migration, because the return migration decision, even if is determined by similar variables, is made in a different timing than the original migration decision. Thus, it is unlikely that the return migration selection reverse the original selection process in such a way that we no longer can identify the effects of the determinants of the migration decision. Furthermore, 29
the empirical literature about return migration (Coulon and Piracha, 2005; Ambrosini and Peri, 2012; Aguilar Esteva, 2013) found mixing evidence on wether the return migrants are the less or more skilled, but never make any links between the determinants of the original migration decision and this issue. Another hypothesis is that the determinants of the migration decision have no effect on the average occupational scores because the occupations that immigrants had on the US were not reflecting their skills. This would be the case if the US labour market is not able to recognize or use the immigrants skill once they arrived but only after an assimilation process. This hypothesis was proposed and empirically confirmed by Borjas (1987) and more recently by Lafortune and Tessada (2012). Thus, the results presented in Table 10 can be interpreted as suggestive evidence of the existence of an assimilation process in which the US labour market is not able to take advantage of the immigrants skills. Table 10: Skill Level Regressions using IPUMS data: (1) (2) (3) (4) (5) (6) VARIABLES logoccscore logsei logpresgl logerscor50 logedscor50 lognpboss50 loggdp -0.07-0.13 0.09-0.19-0.18-0.18 (0.10) (0.11) (0.16) (0.15) (0.31) (0.13) logcost 0.06 0.06 0.12-0.05-0.24-0.08 (0.05) (0.06) (0.08) (0.08) (0.17) (0.07) interwars -0.02 0.01 0.01-0.05 0.05-0.02 (0.03) (0.05) (0.04) (0.07) (0.11) (0.07) intrawars 0.03 0.12** 0.06* 0.02 0.27** 0.07 (0.04) (0.05) (0.03) (0.06) (0.10) (0.06) Observations 912 912 912 912 912 912 R-squared 0.35 0.41 0.29 0.29 0.30 0.33 *** p<0.01, ** p<0.05, * p<0.1 Notes: Clustered standard errors at a country level. Year and country fixed-effects included. The dependent variables are the average score in occupational standing variables (see Table 2). An observation correspond to a country in a year. Forty countries are included in this regression. Robustness Checks. In previous tables, the regressions are presented at a country level. In order to do that, I decomposed the dependent variables, that originally were available at ethnic group level, into a country level using intra-ethnic group shares for each country calculated from the IPUMS data. An alternative is, instead of decomposing the dependent variable, aggregating the independent variables from a country level to an ethnic group level using the same shares. There is two main differences between these methodologies that made the country level data preferable. First, at a country level data we have more observations and if GDP data is missing, we only miss that individual country observation. 30
At an ethnic group level data, if one of the countries corresponding to that ethnic group have a missing GDP data, we loose the whole ethnic group observation, because we cannot compare across time between weighted sums of different countries. Second, considering that the shares are constructed imperfectly using information only about the immigrants who stayed (IPUMS data), if we have measurement error in the dependent variables, as we do at a country level, we only loose statistical significance as the dependent variable has more error but we do not introduce bias in the coefficient estimation. If we use the data at ethnic group level, we have bigger measurement error in the dependent variable and that leads to attenuation bias in the estimation of the coefficient, so the point estimation should be closer to 0. The skill level regressions at ethnic group level are presented in Table A4.1 on the Appendix 4. To compare, in Table A4.2 I reestimate the skill level regressions at a country level using only the observations that are included in the ethnic group level regressions. Qualitatively the results are the same but the regressions at an ethnic group level show less statistical significance in all the coefficients. This can be explained by the aforementioned attenuation bias that is observed when comparing most coefficients and also because the ethnic group level regression have less observations. Another robustness check consist in excluding the WWI period from our data. This is important because for those year you have a combination of extremely high mobility costs as evidenced by Figure 8, low GDP and International war participation for many countries. Thus, if the results are driven in a significant way by this exceptional period data, then we cannot be confident that the coefficients estimation represents adequately the entire immigration process under scope. Table A5.1 and Table A5.2 in Appendix 5 exhibit the results, confirming our previous finding. In particular, most coefficients appear to be even larger once we exclude WWI from our sample. 6 Conclusions The last two centuries has been characterized by unprecedented international migratory movements. In the 20th century, new policies and restrictions arise in the popular destinations as the US in order to control the immigration inflows and thus protect the local community among others objectives. This inflow of workers, as well as the restrictions over this inflow, have had huge economic consequences that are the subject of a big and still growing literature. One of the relevant factors that are key to understand the practical implications of migration and restrictions on migration is whether immigrants are skilled or unskilled and what determines differences on occupational quality between immigrants from diverse nationalities. In this context, this paper contributes to the literature by identifying 31
the determinants of the skill level of immigrants and their effect using a panel data setting covering the mass migration period at the beginning of the twentieth century. The results support the theoretical finding that an increase in GDP has a negative effect over average skill level because by relaxing the liquidity constraint restriction imposed over low skilled workers and reducing the expected income gain for skilled workers the composition of the migrants change towards a larger share of unskilled workers. Also, an increase in migration costs reduce migration of individuals from all skill levels and for the period that we are studying the decrease is proportionally more intensive on unskilled immigrant so the average skill level increase. Other interesting findings that require further research are the fact that restrictions on immigrations appear to reinforce the self-selection process and that the determinants of the average skill level, while having a significant effect over the quality of occupation that the immigrants had before migrating, does not have a significant effect over the quality of occupations that the immigrants had in the US, suggesting that the US labour market does not assimilate perfectly the skills of immigrants upon arrival. 32
References Abramitzky, R., L. P. Boustan, and K. Eriksson (2012a, August). Europe s tired, poor, huddled masses: Self-selection and economic outcomes in the age of mass migration. American Economic Review 102 (5), 1832 56. Abramitzky, R., L. P. Boustan, and K. Eriksson (2012b). A nation of immigrants: Assimilation and economic outcomes in the age of mass migration. Working Paper 18011, National Bureau of Economic Research. Aguilar Esteva, A. A. (2013). Stayers and returners: Educational self-selection among u.s. immigrants and returning migrants. IZA Discussion Papers 7222, Institute for the Study of Labor (IZA). Ambrosini, J. W. and G. Peri (2012). The determinants and the selection of mexico us migrants. The World Economy 35 (2), 111 151. Borjas, G. J. (1987). Self-selection and the earnings of immigrants. American Economic Review 77 (4), 531 53. Borjas, G. J. and R. M. Friedberg (2009). Recent trends in the earnings of new immigrants to the united states. NBER Working Papers 15406, National Bureau of Economic Research, Inc. Carliner, G. (1980). Wages, earnings and hours of first, second, and third generation american males. Economic Inquiry 18 (1), 87 102. Carrington, W. J., E. Detragiache, and T. Vishwanath (1996). Migration with endogenous moving costs. American Economic Review 86 (4), 909 30. Chiquiar, D. and G. H. Hanson (2005). International migration, self-selection, and the distribution of wages: Evidence from mexico and the united states. Journal of Political Economy 113 (2), 239 281. Chiswick, B. R. (1978). The effect of americanization on the earnings of foreign-born men. Journal of Political Economy 86 (5), 897 921. Chiswick, B. R. (2000). Are immigrants favorably self-selected? an economic analysis. IZA Discussion Papers 131, Institute for the Study of Labor (IZA). Coulon, A. and M. Piracha (2005). Self-selection and the performance of return migrants: the source country perspective. Journal of Population Economics 18 (4), 779 807. 33
Deininger, K. and L. Squire (1996). A new data set measuring income inequality. World Bank Economic Review 10 (3), 565 91. Ibarraran, P. and D. Lubotsky (2005). Mexican immigration and self-selection: New evidence from the 2000 mexican census. NBER Working Papers 11456, National Bureau of Economic Research, Inc. Jasso, G. and M. R. Rosenzweig (1990). Self-selection and the earnings of immigrants: Comment. American Economic Review 80 (1), 298 304. Jasso, G. and M. R. Rosenzweig (2008). Selection criteria and the skill composition of immigrants: A comparative analysis of australian and u.s. employment immigration. IZA Discussion Papers 3564, Institute for the Study of Labor (IZA). Lafortune, J. and J. Tessada (2012). Smooth(er) landing. the dynamic role of networks in the location and occupational choice of immigrants. Working Papers ClioLab 14, EH Clio Lab. Instituto de Economia. Pontificia Universidad Catolica de Chile. Mayda, A. (2010). International migration: a panel data analysis of the determinants of bilateral flows. Journal of Population Economics 23 (4), 1249 1274. Mckenzie, D. and H. Rapoport (2007). Network effects and the dynamics of migration and inequality: Theory and evidence from mexico. Journal of Development Economics 84 (1), 1 24. Mishra, P. (2007). Emigration and wages in source countries: Evidence from mexico. Journal of Development Economics 82 (1), 180 199. Mohammed, S. I. and J. G. Williamson (2004). Freight rates and productivity gains in british tramp shipping 1869â1950. Explorations in Economic History 41 (2), 172 203. Munshi, k. (2003). Networks in the modern economy: Mexican migrants in the u.s. labor market. The Quarterly Journal of Economics 118 (2), 549 599. Neuman, K. E. and D. S. Massey (1997). Undocumented migration and the quantity and quality of social capital. University of chicago - population research center, Chicago - Population Research Center. O Rourke, K. and J. Williamson (1999). Globalization and History: The Evolution of a Nineteenth-Century Atlantic Economy. University Press Group Limited. Orrenius, P. M. and M. Zavodny (2005). Self-selection among undocumented immigrants from mexico. Journal of Development Economics 78 (1), 215 240. 34
Roy, A. D. (1951). Some thoughts on the distribution of earnings. Oxford Economic Papers 3 (2), 135 146. 35
1 Appendix 1: Gini and year it was taken from for each country Country Year Gini Argentina 1953 40 Australia 1969 32.02 Austria 1987 28.94 Belgium 1979 28.25 Brazil 1960 53 Bulgaria 1963 22.5 Canada 1951 32.56 Chile 1968 45.64 China 1980 32 Colombia 1970 52.02 Costa Rica 1961 50 Czechoslovakia 1958 27.19 Denmark 1976 31 Finland 1966 31.8 France 1956 49 Germany 1963 28.13 Greece 1974 35.11 Guatemala 1979 49.72 Honduras 1968 61.88 Hungary 1962 25.93 India 1951 35.56 Italy 1974 41 Japan 1962 37.2 Korea 1966 26 Mexico 1950 52.6 Netherlands 1975 28.6 Nicaragua 1993 50.32 Norway 1962 37.52 Peru 1971 55 Poland 1976 25.81 Portugal 1973 40.58 Romania 1989 23.38 Spain 1965 31.99 Sweden 1967 33.41 Switzerland 1982 37.37 Turkey 1968 56 United Kingdom 1961 25.3 Uruguay 1961 36.61 Venezuela 1971 47.65 Yugoslavia 1963 31.18 36
2 Appendix 2: Quantities and Shares Regressions for 1873 to 1932 The results that include the information of the period 1873-1899, presented in Table A2.1 differ qualitatively only respect to the coefficient on migration cost. Once we include the new information (293 observations) and we have the full data for the Quantities and Shares Regression, the analysis of the shares suggest that an increase in the mobility cost increase the share of unskilled. From column (6) we can see that the increase share of unskilled is due to the fact that the total flow of skilled and professionals decrease very strongly as a response to mobility costs increases. Compared to the results of Table 5, we have that now the total flow, the flow of unskilled and the flow of professional are more responsive to the mobility cost. Based on the model, we would observe these changes in elasticities if the mobility cost of the new observations were more prohibitive, so the two cutt-offs are more centered where more population mass is concentrated. Table A2.1: Quantities and Shares Regressions: 1873-1932 (1) (2) (3) (4) (5) (6) (7) VARIABLES log total flow share share share log total flow log total flow log total flow professionals skilled unskilled professionals skilled unskilled loggdp -0.28-0.03-0.10* 0.13** -0.23-0.52 0.30 (0.67) (0.04) (0.06) (0.06) (0.45) (0.55) (0.83) logcost -1.36** 0.02-0.17*** 0.15*** -1.22*** -2.52*** -0.89 (0.52) (0.03) (0.04) (0.05) (0.33) (0.41) (0.64) interwars -0.13 0.00-0.02 0.02-0.19-0.18-0.09 (0.37) (0.02) (0.03) (0.02) (0.27) (0.30) (0.36) intrawars 0.33-0.01-0.00 0.01 0.22 0.31 0.30 (0.50) (0.02) (0.05) (0.06) (0.30) (0.37) (0.60) Observations 1,319 1,319 1,319 1,319 1,319 1,319 1,319 R-squared 0.81 0.55 0.52 0.63 0.84 0.84 0.79 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Notes: Clustered standard errors at a country level. Year and country fixed-effects included. The dependent variables are quantities and shares of immigrants for each skill category. An observation correspond to a country in a year. Forty countries are included in this regression. 3 Appendix 3: Descriptive Graph of flows for each country The Graphs presented in this appendix describe the immigration process for each country. Part of the added value of these tables is to appreciate with detail the effect that restrictions had over the immigrants flow. The dashed black line in 1921 represents the 1921 Immigration Act. The solid black vertical line represents the 1924 Johnso-Reed Immigration Act. 37
Also, we can distinguished the involvement on international and intranational war for each country. 38
39 Figure A3.1: Total flow of Immigrants for each skill level by Countries
40 Figure A3.2: Total flow of Immigrants for each skill level by Countries
41 Figure A3.3: Total flow of Immigrants for each skill level by Countries
42 Figure A3.4: Total flow of Immigrants for each skill level by Countries
43 Figure A3.5: Total flow of Immigrants for each skill level by Countries
44 Figure A3.6: Total flow of Immigrants for each skill level by Countries
45 Figure A3.7: Total flow of Immigrants for each skill level by Countries
46 Figure A3.8: Total flow of Immigrants for each skill level by Countries
47 Figure A3.9: Total flow of Immigrants for each skill level by Countries
48 Figure A3.10: Total flow of Immigrants for each skill level by Countries