Preliminary Draft May 21, 2001 A Longitudinal Analysis of Post-Migration Education Jorgen Hansen Concordia University Magnus Lofstrom University of California at Irvine Kirk Scott Lund University Abstract The success of immigrants in the host country labor market depends greatly on the level of human capital they hold and how it changes with time spent in the new country. This paper uses a unique longitudinal database to examine the acquisition of Swedish educational qualifications by immigrants. The Swedish Longitudinal Immigrant database is essentially a panel containing both register information on natives and immigrants in Sweden, as well as certain information regarding pre-migration experience gleaned from visa applications. Using this information, we examine the post-migration educational attainment of immigrants from 8 countries compared to natives during the period 1986 1996. Of special interest here are the influences of visa status, i.e. refugee, tied mover, or labor migrant, and pre-migration educational level on educational investment in Sweden. JEL Classification: I21, J24, J60, J61 Key words: Immigration, education Notes: This paper is written within the research project "Labour Demand, Education and the Dynamics of Social Exclusion" funded by European Commission under the TSER network (contract number SOE2-CT97-3052) and the Swedish Council for Research in Humanities and Social Science.
Introduction The success of immigrants in the host country labor market depends greatly on the level of human capital they hold and how it changes with time spent in the new country. One crucial aspect of the labor market experience of immigrants is post-migration investment in education. Several studies have analyzed the relationship between entry skill levels and earnings growth (e.g. Duleep and Regets 1999, Borjas 2000) and returns to pre-migration and post-migration education (Betts and Lofstrom 2000). However, due to limited availability of appropriate data, very little research has been devoted to the relationship between pre- and post-migration investment in education (exceptions are Chiswick and Miller 1994, Khan 1997 and Rooth 1999). In this paper we give a detailed description of how immigrants education changes with time spent in Sweden. The unique longitudinal data used allows for identification of both the level of educational attainment at the time of entry as well as any new schooling obtained since arrival. Specifically, access to such a unique data allows us to address three questions that are central to the debate on immigrants investment behavior. First, to what extent do immigrants in Sweden invest in Swedish education? The incentives for investing in a Swedish education will most likely differ across different types of immigrants. For example, assuming a difference in opportunity costs between refugee immigrants and labor migrants, we may expect the former group to invest more than the latter. Moreover, the willingness to invest in a Swedish education may also depend on the economic conditions in Sweden at the time of arrival. As the data we use includes information on entry visa (i.e. refugee, tied mover, economic migrant or student) as well 1
as time of entry, we hope to gain more insight into the determinants immigrants investment behavior. The second question deals with the important relationship between pre-migration education and post-migration education. It is well established that the labor market experience of immigrants in the host country to a large extent depends on their endowment of, predominantly host-country specific, human capital. If pre- and postmigration education is positively correlated, then using a policy that implies selecting highly educated immigrants may be favorable, as these immigrants are more likely to perform well in the host-country s labor market. Moreover, the value of the pre-migration education depends to a large extent on the source country. It is therefore important in the empirical analysis to allow the effects of pre-migration education on post-migration investment to differ for different immigrant groups. The last question we will attempt to answer in this paper is how post-migration education changes with time spent in Sweden. It is expected that most of the investment activities take place during the first few years after arrival in the host country, as this gives the investor a longer time horizon in which he can obtain returns on his investment. 2
Background The history of Swedish immigration did not begin in earnest until the 1950s. The 1950s and 1960s were decades of economic expansion in Sweden, with immigrants being actively recruited to relieve production bottlenecks, and allow Swedish industry to meet international demand. During this period, immigrants were fairly easily integrated into the Swedish economy, which was characterized by full employment and high demand for labor. The 1970s, 1980s, and 1990s saw a hardening international market for Swedish goods and a slowdown in GDP growth. The Swedish response to this new situation was a restructuring of the economy, with a switch from traditional industrial activities to more modern industrial and service production. 1 This switch to new production techniques was accompanied during the 1980s by widespread, innovative changes in workplace organization (Rosholm et al 2000, Snower 1999, OECD 1999). Evidence from studies of changes in hiring practices among Swedish companies (Broomé, Carlson, Ohlsson 2001) indicates that these workplace changes led to increasing importance being placed on host-country-specific human capital, an asset which immigrants are most likely less endowed with than natives. This change has brought about a situation where immigrants in Sweden appear to face increasing difficulties obtaining and keeping employment, even during cyclical upturns. While the immigrant population is changing in composition, which accounts for a portion of the problem, many established immigrant groups in Sweden have been facing these changing hiring processes as well. 1 The solidaristic wage policy implemented in Sweden was a unique attempt to force up workers' wages, thereby making traditional, labor-intensive industrial activities less competitive on the international market. The idea was that the Swedish economy would be modernized through the elimination of these companies. 3
If the increased demand for country-specific human capital is a factor in declining immigrant economic performance, then we may expect an increase in the incentive for investment in Swedish formal education after immigration, and more satisfactory labor market outcomes for those immigrants who do make this investment. Specifically, more recent immigrants should have higher incentives to invest in Swedish education than immigrants who arrived in the 60s and 70s, due to greater perceived returns. Studies of post-migration educational investment in Sweden are limited. The notable exception is Rooth (1999). Rooth examines refugee migrants entering Sweden between 1987 and 1992. Rooth finds four factors to be important in determining postmigration educational investment: age at migration, pre-migration schooling, year of immigration, and country of origin. There exists a positive relation between pre- and post-migration educational attainment, while there is a negative relation between age at migration and education in the destination country. International studies of educational investment include Chiswick & Miller (1994), Khan (1997) and Cobb-Clark et al (2001). Chiswick and Miller find results similar to Rooth using Australian data. They examined all visa categories, however, and found that tied movers had the lowest propensity to educate themselves, while migrants sponsored by an employer high the highest probability of investing in destination country education. Using additional information on occupation in the home country, Chiswick and Miller found a positive relationship between pre-migration occupational skill level and postmigration educational investment. Khan examines US data and also finds a negative relationship between age and post-migration education. An interesting aspect of this study is a reported negative 4
relationship between pre and post migration education. 2 The argument here is that preimmigration education acts as a substitute for post-immigration education, and not as a complement, which has otherwise been the understanding. Cobb-Clark et al have recently performed another study using Australian data. Looking at enrollment status of immigrants only, they find that pre-migration education is positively related to enrollment for migrants living in traditional families. Unlike Chiswick and Miller, Cobb-Clark et al find that migrants entering Australia for employment purposes have the lowest probability of being enrolled of all visa categories. Data This paper will utilize the Swedish Longitudinal Immigrant Database (SLI). The SLI is essentially a register-based panel, which contains information on approximately 110,000 immigrants between the years 1968 1996. From this larger database, a smaller subsample of approximately 25,000 individuals have had extended information added to their records from the initial visa application investigation kept on file at the Swedish Immigration Board. This information included education obtained prior to entrance into Sweden and the actual visa type granted. This information makes this data unique, at least regarding Swedish studies, since all previous studies have either exclusively examined refugees (Rooth 1999), or made assumptions as to visa category based on country of origin and date of arrival. This study has exact classifications, and can thus be used to differentiate between individuals from the same country with different visa categories. 2 Khan finds a positive relationship for lower education levels in the 1980 Census, but not in the 1976 Survey of Income and Education. 5
Beyond the information from the home country, the database also contains all relevant information on income, social transfers, human capital, etc. The time period 1986 1996 has been chosen for this study because it is only after 1986 that reasonably reliable educational information exists. Prior to the establishment of a national education register in 1986, the most recent information source is the census of 1970, making the period before 1986 difficult to study. The data has been arranged in the form of a balanced panel, restricting participation to male immigrants aged 16-65 who had entered Sweden in 1986 or earlier, and male natives who had turned 16 by 1986. In addition the oldest individuals included in the restricted panel were 65 in 1996. Thus the panel gives us individuals of working age throughout the period, with zero attrition. Given the restrictions placed on the data, we are left with slightly more them 3,500 individuals, and almost 40,000 observations. The restricted dataset has information on natives and immigrants from the following countries: Germany, USA, Poland, Yugoslavia, Greece, Turkey, Iran, and Chile. Together, these eight nationalities account for approximately 30 percent of the total working age immigrant population in Sweden, and slightly over 50 percent of the non-nordic immigrant population. Descriptive Statistics As pointed out above, immigrants post-migration schooling investment decision is likely to partially depend on the educational attainment level at the time of arrival, i.e. pre-migration schooling. Table 1 shows the highest reported foreign education at the time of arrival in Sweden, broken down by arrival cohort and visa category. The information 6
here is self-reported, and retrieved from the interviews conducted with each immigrant by the Swedish Immigration Board during the visa application process. Note that Nordic immigrants are not included in these figures, since Nordic citizens do not require special permission to establish residence or take up employment in Sweden. Table 1 shows that refugee immigrants are the most highly educated group of immigrants upon arrival in Sweden. More than 1/3 of the refugee immigrants arrive with university education whereas slightly more than 1/5 of the labor migrants, 23 percent, and tied movers, 22 percent, enter with post-secondary education. Furthermore, more than one half of labor migrant and tied movers arrive with only primary education, 59 and 50 percent respectively. The corresponding percentage among refugee immigrants is 37 percent. Overall, it appears that labor migrants in Sweden display the lowest upon arrival schooling levels. However, there appears to be a decline in the average educational level of recently arriving refugee immigrants, compared to the pre-1970 arrival cohort. There are two plausible reasons for this trend. The first is that refugee migration to Sweden from 1945 through 1970 was almost exclusively composed of individuals from Eastern Europe, while from the 1970s onwards, individuals from developing countries, i.e. countries with lower average educational attainment levels, have dominated refugee migration. The second explanation lies in the very nature of refugee migration, with the earliest refugees from a given country often being more highly educated than those who follow. This may possibly be due to both being most likely to be persecuted and also having the ability and means to leave. The schooling trend seems to be the reverse among labor migrants. Those who came recently are, on average, better educated than those who arrived during the peak of 7
labor migration. This is may possibly be a result of increasing obstacles to non-eu labor migration due to more selective admission within this category. 3 Table 1 suggests that there are, not surprisingly, differences in pre-migration schooling levels across visa categories, but overall, there are smaller differences in the entry educational attainment across entry visa categories among the most recent arrival cohort. One of the main issues we would like to address in this paper is the extent to which immigrants invest in post-migration education. Table 2 shows the highest schooling level immigrants obtained in Sweden by entry visa category and arrival cohort in 1996. As can be seen in Table 2, after entering Sweden, a considerable share of immigrants chooses to invest in Swedish education. The overall percentage of immigrants who obtained some post-migration education for each of the entry categories is greater than 80 percent. The table shows that refugee immigrants have the highest proportion who further their education after arriving in Sweden, 87 percent do so. Among labor migrants, 82 percent invested in Swedish schooling while among tied movers the proportion was 83 percent. Tied movers appear to have invested more heavily in lower educational levels, without continuing on to university, at least among the earlier cohorts. The labor migrants invested slightly more in secondary and university education, while refugees, especially those from the earliest cohorts, invested quite heavily in university education. One possible reason we observe these differences across entry categories may be that the groups arrive with different levels of pre-migration education, as shown in Table 1. We next turn to the issue of the role of entry educational attainment levels on post-migration investment in schooling. 3 See Lundh & Ohlsson (1996) for a discussion of changing immigration policies in Sweden. 8
The difference in post-migration educational attainment among categories is at least partially a result of differences in initial educational levels on arrival to Sweden. Table 3 shows the difference between pre- and post-migration educational levels as of 1996. In other words, Table 3 shows the unadjusted probabilities of investing in a certain level of Swedish schooling, conditional upon a given level of pre-migration educational attainment. The table suggests that the majority of the immigrants who do choose to study in Sweden complete degrees at the level they had in the home country or one level above. For example, among refugees who arrived in Sweden with only primary schooling, 42 percent invest in Swedish primary education, 39 percent increase their schooling level to secondary education and 5 percent obtain a university degree. The investment levels are quite similar for tied movers and labor migrants who also enter Sweden with only primary schooling. At the other end of the educational distribution, immigrants who arrive with a university degree, we observe that a sizeable proportion, between 18 and 36 percent, invest in an education level below the entry level. This is more prevalent among refugee immigrants than among the other two visa categories and may be a reflection of the lack of transferability of education obtained in less developed countries. Language may also be a determinant in which the enrolment in lower education levels helps immigrants improving their language ability and hence increase transferability of existing foreign acquired human capital. Overall, Table 3 shows that the vast majority of immigrants do choose to make some human capital investment after arrival in Sweden. The discussion so far has only considered differences across immigrant groups, i.e. arrival cohorts and visa entry category. Clearly, it is of interest to compare how 9
immigrants educational attainment level changes over time relative to natives. To do so, we compare increases in schooling level from the first year we observe an individual, 1986, to the last year we observe our sample, 1996. Table 4 shows the probability of moving from one level of Swedish education in 1986 to another level in 1996, i.e. the conditional probability of a specific schooling level in 1996, given a particular educational attainment level in 1986. The table shows that 13 percent of refugee immigrants with Swedish secondary education in 1986 had obtained a Swedish university degree by 1996. The corresponding transition probabilities for labor migrants and tied movers are 3 and 13 percent respectively. Given secondary schooling in 1986, refugee immigrants and tied movers are more likely to invest in university than native born Swedes, whose equivalent transition probability is 9 percent. However, all immigrant groups are less likely to move from primary education to either secondary or university education. Table 4 indicates that immigrants who had no Swedish education in 1986 are the most likely immigrants to have increased their schooling levels to secondary and university degrees. For example, among refugee immigrants and tied movers with no Swedish education in 1986, 20 percent had acquired a university degree by 1996. This points to the fact that care must be taken in interpreting this table. One reason for this observation is that immigrants with no Swedish education in 1986 are likely to be very recent immigrants. This fact, in conjunction with the observation that post-migration investment in education is likely to take place soon after arrival in Sweden, explains the somewhat surprising finding. Support for the claim that post-migration schooling takes places rather sooner than later in the stay can be seen in Figure 1. 10
Figure 1 shows the highest Swedish educational level per year for the cohort arriving in Sweden in 1986. Here it is evident that the majority of the educational investments occurred very soon after arrival in Sweden. The figure shows that investment take place mostly within the first three years since migrating, with university education experiencing another increase after about six years. To address some of the concerns raised in analyzing the descriptive statistics, we next turn to the empirical results. Empirical Results Since our measure of post-migration education is discrete, measuring the highest degree obtained, we use ordered probit models to estimate the determinants of postmigration education. To be specific, the estimated models can be described as follows. Let * yit = Xit β + εit i = 1, 2,...,n and t = 1, 2,...,T and y µ * it = k + 1 if µ k < yit k+ 1 where k=-1,0,1,2, and µ k < µ k + 1 ensures that the probabilities are positive. The µ s are unknown parameters to be estimated jointly with β, and reflect threshold values for moving through the schooling participation decision. X it is a vector of observable characteristics. It includes both immigrant specific characteristics (visa category, premigration educational attainment, country of origin, time of arrival and unemployment rate at the time of entry to Sweden) and traditional socio-economic characteristics (age, 11
marital status, number of children and area of living). The results are presented in Tables 5 to 7. Because of the non-linear nature of the model, the magnitudes of the coefficient estimates provide little information about the size of the effects of the observable characteristics. Therefore, instead of discussing the coefficient estimates, which are reported in Table A1 in the Appendix, we present differences in predicted educational degrees. The predicted degrees are evaluated for a representative individual using the estimates reported in Table A1. In Table 5, we report the differences in predicted schooling attainment between natives and immigrants. All the entries in the table are conditioned on pre-migration education, and we report results for three different model specifications. In the first, we include only controls for pre-migration education. Model 2 adds country fixed-effects while Model 3 also includes observable individual characteristics. The table reveals several interesting results. First, there are only small differences in post-migration education investment between the three immigrant groups (labor migrants, refugees and tied movers). This is true regardless of model specification. Second, according to models 1 and 2, it appears as if immigrants with primary pre-migration education have substantially lower Swedish education than natives have. For example, we find that the difference in the predicted probability that a native and a refugee immigrant will have a Swedish University degree is 0.176. Part of this large difference can be explained by country of origin, as the difference decreases to 0.14 when we include controls for country-fixed effect in the model. However, most (perhaps all) of this difference can be attributed to differences in observable individual characteristics as the predicted 12
difference between a native and a refugee immigrant with a foreign primary education is 0.048. It is obvious from Table 5 that there is a strong relationship between pre- and post-migration education. For example, according to our results, there is virtually no difference in the proportion of university-educated natives and immigrants, when we condition on immigrants with a primary foreign education. However, if we compare natives with immigrants with a secondary or a university education, obtained before arrival in Sweden, it appears that natives are less likely to have a Swedish University degree. For all three immigrant groups, the difference in the predicted probability that a native and an immigrant will have a Swedish University degree is around 0.4. To further illustrate the importance of pre-migration education on post-migration educational investments, we calculated differences in predicted schooling attainments between immigrants conditioning on their foreign education. The results are shown in Table 6. In the first three columns, we present differences in predicted (post-migration) schooling between immigrants with primary and secondary foreign education. In columns 4 to 6, we present differences between immigrants with a primary education and a university education. Similar to the results shown in Table 5, we find only small differences in post-migration education investment between the three immigrant groups (labor migrants, refugees and tied movers). Again, this is true regardless of model specification. It appears as if about the same proportion of immigrants with a secondary and a primary foreign education choose not to invest in any post-migration education, as the difference reported in column three is close to zero. Moreover, compared to immigrants 13
with primary pre-migration education, we find that immigrants with a secondary education are less likely to invest in a Swedish primary or secondary education and more likely to invest in a university degree. When we compare immigrants with a foreign university degree with immigrants whose highest pre-migration education is primary school, we find a similar pattern even if the differences are larger in absolute terms. For example, the difference in the predicted probability of having a Swedish University degree between a refugee with a pre-migration university degree and a refugee with a primary (foreign) degree is 0.418. If we compare immigrants with secondary and primary degrees, the difference is 0.19. Overall, our results indicate a strong positive relationship between pre- and postmigration education. There may be two possible reasons for this result. First, the positive correlation between the two types of education indicates that ability or taste for schooling plays an important role. If unobserved ability is the main reason for educational investment, then immigrants who choose to invest in a pre-migration education are more able and they will then also invest more in post-migration education. On the other hand, if other factors, such as borrowing constraints, determine educational investment, then we would expect a negative (or a non-positive) correlation between pre- and post-migration education. The reason for this is that education is virtually free in Sweden, and immigrants who are prevented from investing in their home countries due to financial constraints should invest upon arrival in Sweden. In this case, we would find more investment taking place among low-educated immigrants and consequently a nonpositive relationship between pre- and post-migration. A second reason for the positive 14
relationship between the two types of education may simply be that higher levels of education are less transferable than lower degrees (see Friedberg 2000). Our results indicate that there are differences in schooling investment between immigrants and natives, and that this difference varies across pre-migration educational attainment among immigrants. Our next step is to analyze how immigrants postmigration investment behavior changes with time spent in the host country. The results can be found in the first three columns of Table 7. The entries show differences in predicted schooling between natives and immigrants. The predictions were obtained using the results from our most general model (Model 3) and the probabilities are evaluated at the same values of the observable characteristics as in Tables 5 and 6. In addition, for immigrants we condition on primary pre-migration education. The results indicate that immigrants, regardless of visa category, assimilate in terms of host-country specific educational attainment with time spent in Sweden. For example, the difference in predicted probabilities of having a university degree between natives and refugee immigrants who have spent only one year in Sweden is 0.082. The same figure, comparing natives with refugee immigrants with 20 years spent in Sweden is 0.015. Overall, for all visa categories and all post-migration educational levels, we find that the difference in predicted schooling attainments between natives and immigrants decrease with time spent in Sweden. One potentially important factor in determining post-migration educational investment is the labor market conditions in the host country at time of arrival. Immigrants who arrive during an economic boom may, everything else held constant, be more likely to find employment upon arrival than immigrants who arrive during an 15
economic slowdown. Assuming that positive relationship between education and employment, this would suggests that immigrants who arrive during times with higher than normal unemployment rate would be more likely to invest in post-migration education than other immigrants. This is a testable hypothesis, and in the last three columns of Table 7, we show differences in predicted schooling probabilities between natives and immigrants, for different unemployment rates at arrival in Sweden. In column 4, we present differences in predictions under the assumption that the unemployment rate at arrival is 4 percent. Note that this column is simply a replication of column 3 in Table 5. The entries in column 5 are based on an increase in the unemployment rate at arrival with 10 percent (implying an unemployment rate of 4.4 percent). For all visa categories and all post-migration educational levels, we find that the increase in unemployment implies more investment among immigrants compared to natives. To illustrate, the difference in the predictions for a university degree for refugees increase, the difference change from 0.048 to 0.058. This corresponds to a 20 percent increase in the difference, or an elasticity of about 2 percent. To further illustrate the effect of unemployment at arrival in Sweden, the last column shows differences in predicted schooling attainments when the unemployment rate is set to 8 percent. It is clear that such a large increase in the unemployment rate (which is still smaller than the increase in unemployment rates in Sweden between 1991 and 1996) has substantial effects on immigrants post-migration investment behavior. To summarize, our results indicate that, conditioning on observable individual characteristics, immigrants who arrive in periods with an economic slowdown and high unemployment rates are substantially more likely to invest in post-migration education than other immigrants. 16
Conclusions This paper uses a unique longitudinal database to examine the acquisition of Swedish educational qualifications by immigrants. The Swedish Longitudinal Immigrant database is essentially a panel containing both register information on natives and immigrants in Sweden, as well as certain information regarding pre-migration experience gleaned from visa applications. Using this information, we examine the post-migration educational attainment of immigrants from 8 countries compared to natives during the period 1986 1996. Of special interest here are the influences of visa status, i.e. refugee, tied mover, or labor migrant, and pre-migration educational level on educational investment in Sweden. Looking first at the effects of visa status on Swedish educational level, we can see a few differences between the categories. In an unrestricted model, controlling for only pre-migration education and visa status, we see very few differences between the visa categories, but these differences begin to appear when individual characteristics are controlled for, with refugees investing more in university education than other visa categories, but less in the lower educations. Moving to an examination of the role of pre-migration educational level, we see a clear positive effect. This strong effect is likely to be the result of several factors. The first is that pre-migration education may be a revealed preference for education itself, and as such would result in investment in the destination country. Another reason for this relationship may well be that there is a negative relationship between skills-transferability and higher educational levels. 17
The next purpose of this study was to examine the effects of time in the host country on educational attainment. Here we can say that there is an unambiguous evening of educational attainment levels between natives and all immigrant visa categories as time in the country increases. Finally, the effects of macroeconomic conditions at labor market entry are studied. Here we find that increased unemployment at labor market entry increase postmigration schooling for all visa categries relative natives. In all, it can be said that visa status matters somewhat in terms of post-migration human capital investment, but that pre-migration education is a more reliable indicator of post-migration behavior. In addition, the economic conditions at arrival also play a significant role in determining educational outcomes. 18
References Betts, Julian R. and Lofstrom, Magnus. (2000) The Educational Attainment of Immigrants: Trends and Implications, in George J. Borjas (ed) Issues in the Economics of Immigration. University of Chicago Press. Borjas, George J. (2000) The Economic Progress of Immigrants, in George J. Borjas (ed) Issues in the Economics of Immigration. University of Chicago Press. Broomé, Per, Carlson, Benny and Ohlsson, Rolf (2001) Ethnic Diversity and Labour Shortage. Rhetoric and Realism in the Swedish Context, SNS Occasional Paper No. 86, SNS, Stockholm. Chiswick, Barry and Miller, Paul (1994) The Determinants of Post-immigration Investments in Education, Economics of Education Review, 13(2):163-177. Cobb-Clark, Deborah, Connolly, Marie and Worswick, Christopher (2001) The Job Search and Education Investments of Immigrant Families, IZA Discussion Paper no. 290. IZA, Bonn. Duleep, Harriett Orcutt and Regets, Mark C. (1999) Immigrants and Human-Capital Investments, American Economic Review, 82:186-191. Friedberg, Rachel M. (2000) You Can t Take It with You? Immigrant Assimilation and the Portability of Human Capital, Journal of Labor Economics, 18(2): 221-251. Jasso, Guillermina, Massey, Douglas S., Rosenzweig, Mark R. and Smith, James P. (2000) The New Immigrants Survey Pilot (NIS-P): Overview and New Findings About U.S. Legal Immigrants at Admission, Demography, 37:127-138. Khan, Aliya H. (1997) Post-Migration Investment in Education by Immigrants in the United States, Quarterly Review of Economics and Finance, 37:285-313. OECD (1999) New Enterprise Work Practices and Their Labour Market Implications, in Employment Outlook, June 1999, OECD, Paris. Rooth, Dan-Olof (1999) Refugee Immigrants in Sweden: Educational Investments and Labour Market Integration, Doctoral Dissertation, Lund University, Sweden. Rosholm, Michael, Scott, Kirk and Husted, Leif (2001) The Times They Are A- Changin IZA Discussion Paper no. 258, IZA, Bonn. Snower, Dennis (1999) The Organizational Revolution and its Labor Market Implications, Keynote address to the EALE annual conference, 1999. 19
Tables: Table 1. Entry Educational Attainment, by Visa Category and Arrival Cohort. Males. <- 1970 1971-1980 1981-1986 Total REFUGEE Primary 10% 41% 36% 37% Secondary 40% 24% 27% 26% University 50% 35% 37% 36% LABOR MIGRANT Primary 60% 64% 51% 59% Secondary 24% 16% 19% 19% University 16% 20% 30% 22% TIED MOVER Primary 43% 56% 46% 50% Secondary 36% 22% 30% 27% University 21% 22% 25% 23% Table 2. Highest Swedish Education, 1996. Males. <- 1970 1971-1980 1981-1986 Total Natives REFUGEE No Investment 0% 11% 14% 13% 1% Primary 10% 24% 19% 20% 28% Secondary 30% 40% 40% 40% 46% University 60% 25% 27% 27% 24% LABOR MIGRANT No Investment 8% 18% 30% 18% 1% Primary 33% 28% 26% 29% 28% Secondary 42% 33% 24% 33% 46% University 17% 21% 21% 20% 24% TIED MOVER No Investment 11% 15% 19% 17% 1% Primary 14% 27% 21% 23% 28% Secondary 43% 35% 41% 38% 46% University 32% 22% 20% 22% 24% 20
Table 3. Transitions from foreign to Swedish education. Males. Swedish Education, 1996 Foreign Education REFUGEES No Education Primary Secondary University Primary 15% 42% 39% 5% Secondary 11% 11% 53% 24% University 12% 5% 31% 52% Total 13% 20% 40% 27% TIED MOVERS No Education Primary Secondary University Primary 16% 39% 41% 4% Secondary 17% 11% 46% 26% University 18% 3% 24% 55% Total 17% 23% 38% 22% LABOR MIGRANTS No Education Primary Secondary University Primary 19% 41% 36% 3% Secondary 15% 21% 42% 23% University 19% 1% 17% 63% Total 18% 29% 33% 20% 21
Table 4. Transitions: Change in educational level between Swedish education in 1986 and highest attained Swedish education by 1996, for those who invested in Swedish education. Males. Swedish Education, 1996 REFUGEES Swedish Education, 1986 No Education Primary Secondary University No Education 29% 18% 33% 20% Primary 0% 94% 6% 0% Secondary 0% 0% 87% 13% University 0% 0% 0% 100% Total 12% 24% 38% 25% TIED MOVERS Swedish Education, 1986 No Education Primary Secondary University No Education 29% 18% 33% 20% Primary 0% 94% 6% 0% Secondary 0% 0% 87% 13% University 0% 0% 0% 100% Total 12% 24% 38% 25% LABOR MIGRANTS Swedish Education, 1986 No Education Primary Secondary University No Education 73% 6% 12% 9% Primary 0% 98% 2% 0% Secondary 0% 0% 97% 3% University 0% 0% 0% 100% Total 19% 28% 33% 20% NATIVES Swedish Education, 1986 No Education Primary Secondary University Primary - 83% 11% 6% Secondary - 0% 91% 9% University - 0% 0% 100% Total - 28% 46% 24% 22
Table 5. Differences in Predicted Schooling Attainment between Natives and Immigrants, by Immigrant Group Conditioned on Foreign Primary Education Conditioned on Foreign Secondary Education Conditioned on Foreign University Education Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 LABOR MIGRANTS Primary -0.117-0.096-0.014-0.050-0.014 0.071 0.059 0.094 0.133 Secondary 0.119 0.069-0.005 0.020 0.003 0.067 0.009 0.031 0.240 University 0.173 0.136 0.024 0.073 0.022-0.159-0.102-0.173-0.403 REFUGEES Primary -0.119-0.099 0.025-0.054-0.019 0.096 0.055 0.090 0.142 Secondary 0.125 0.074 0.015 0.023 0.005 0.116 0.007 0.028 0.292 University 0.176 0.140-0.048 0.079 0.028-0.238-0.094-0.165-0.465 TIED MOVERS Primary -0.117-0.094-0.002-0.049-0.011 0.079 0.060 0.097 0.136 Secondary 0.118 0.066-0.001 0.020 0.002 0.080 0.009 0.034 0.255 University 0.172 0.133 0.004 0.072 0.017-0.181-0.104-0.180-0.422 Note: Model 1 includes only controls for foreign education. In addition to foreign education, Model 2 also includes country fixed-effects. Finally, Model 3 adds time-fixed effects as well as arrival cohort effects in addition to controls for area of living, age, marital status, number of children, years since migration and unemployment rate at entry in Sweden. 23
Table 6. Illustrating the Effects of Foreign Education on Investment in Post-Migration Schooling, by Immigrant Group. Foreign Secondary Education as opposed to Foreign Primary Education Foreign University Education as opposed to Foreign Primary Education Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 LABOR MIGRANTS No Investment -0.131-0.099-0.026-0.209-0.157-0.036 Primary -0.067-0.082-0.085-0.176-0.190-0.147 Secondary 0.099 0.066-0.072 0.111 0.038-0.245 University 0.099 0.114 0.183 0.275 0.309 0.427 REFUGEES No Investment -0.134-0.101-0.017-0.214-0.162-0.023 Primary -0.065-0.080-0.072-0.173-0.189-0.117 Secondary 0.102 0.069-0.101 0.118 0.046-0.277 University 0.097 0.112 0.190 0.270 0.305 0.418 TIED MOVERS No Investment -0.130-0.097-0.023-0.208-0.154-0.032 Primary -0.068-0.083-0.081-0.177-0.191-0.138 Secondary 0.098 0.064-0.081 0.109 0.032-0.256 University 0.100 0.116 0.186 0.276 0.313 0.426 Note: Model 1 includes only controls for foreign education. In addition to foreign education, Model 2 also includes country fixed-effects. Finally, Model 3 adds time-fixed effects as well as arrival cohort effects in addition to controls for area of living, age, marital status, number of children, years since migration and unemployment rate at entry in Sweden. 24
Table 7. Illustrating the Effects of Years since Migration and Unemployment at Arrival in Sweden on Investment in Post-Migration Schooling, by Immigrant Group. Years Since Migration Unemployment at Arrival in Sweden 1 Year 10 Years 20 Years 4% 4.40% 8% LABOR MIGRANTS No Investment -0.048-0.030-0.013-0.005-0.003 0.012 Primary -0.094-0.065-0.032-0.014-0.008 0.038 Secondary 0.003-0.007-0.009-0.005-0.003 0.026 University 0.139 0.101 0.054 0.024 0.015-0.076 REFUGEES No Investment -0.022-0.009 0.003 0.008 0.010 0.020 Primary -0.051-0.022 0.008 0.025 0.030 0.069 Secondary -0.009-0.007 0.004 0.015 0.019 0.064 University 0.082 0.038-0.015-0.048-0.058-0.154 TIED MOVERS No Investment -0.040-0.023-0.008-0.001 0.001 0.015 Primary -0.082-0.053-0.021-0.002 0.003 0.047 Secondary -0.002-0.009-0.007-0.001 0.001 0.036 University 0.124 0.084 0.036 0.004-0.006-0.098 Note: The entries in the table represent differences in predicted education between natives and immigrants. The table is based on the results for Model 3 (which includes time-fixed effects as well as arrival cohort effects in addition to controls for area of living, age, marital status, number of children, years since migration and unemployment rate at entry in Sweden. 25
Diagram 1. Investment in Swedish Education, given no Swedish education in 1986. Cohort arriving in 1986. 35% 30% 25% 20% 15% Primary Secondary University 10% 5% 0% 86 87 88 89 90 91 92 93 94 95 96 Year 26
Table A1. Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 forsec 0.502 0.468 0.483 age2-0.036 (0.018) (0.019) (0.019) (0.005) foruni 1.054 1.022 1.129 marr 0.153 (0.018) (0.020) (0.021) (0.016) labor -0.757-0.539-0.935 nkid 0.049 (0.018) (0.025) (0.036) (0.006) refugee -0.780-0.561-0.737 d70 0.529 (0.018) (0.031) (0.038) (0.055) tied -0.752-0.523-0.879 d75 0.317 (0.015) (0.024) (0.034) (0.039) Iran -0.056-0.040 d80 0.205 (0.034) (0.035) (0.028) Chile -0.225-0.327 ysm 0.015 (0.033) (0.034) (0.003) USA -0.299-0.303 eunemp 0.068 (0.034) (0.034) (0.014) Poland -0.105-0.176 (0.029) (0.030) _cut1-1.427-1.437-0.862 Yugo -0.191-0.255 (0.011) (0.011) (0.084) (0.027) (0.027) _cut2-0.531-0.535 0.103 Greece -0.424-0.450 (0.009) (0.009) (0.084) (0.032) (0.032) _cut3 0.682 0.686 1.447 Turkey -0.380-0.568 (0.010) (0.010) (0.084) (0.028) (0.029) yr1987 0.104 (0.027) yr1988 0.191 (0.027) yr1989 0.240 (0.028) yr1990 0.387 (0.028) yr1991 0.406 (0.028) yr1992 0.434 (0.029) yr1993 0.458 (0.029) yr1994 0.520 (0.030) yr1995 0.536 (0.031) yr1996 0.552 (0.032) metro 0.095 (0.012) age 0.015 (0.004) 27