DETERMINANTS OF INTERNATIONAL MIGRATION: A SURVEY ON TRANSITION ECONOMIES AND TURKEY Pınar Narin Emirhan 1 Preliminary Draft (ETSG 2008-Warsaw) Abstract This paper aims to test the determinants of international migration from 10 transition countries and Turkey to 23 OECD destination countries for the 1995-2005 period. The preliminary findings reveal that during the first years of the period non-economic factors were more effective on migration flows, however in consecutive years the importance of economic factors increased. The results also show that the economic characteristics of the destination countries are more effective than origin country characteristics. Keywords: International migration, transition economies. JEL Classification: F22 1 Dokuz Eylul University, Faculty of Business, Department of Economics, Turkey. pinar.emirhan@deu.edu.tr
1. Introduction International migration flows is a very important issue which is frequently analyzed by economists. However in recent years international migration flows became more important for the European Union (EU) countries and debated extensively in both political areas and economic literature. 2004 was a very important date for EU enlargement process. At this year ten transition economies became members of EU, and in 2007 two others followed. The relatively poor quality of the economic indicators of these new members caused a fear of mass migration flows within the EU 2. To cope with this problem EU proposed a gradual integration calendar for labor markets during the enlargement process, which went into effort with the 2003 Accession Treaty. According to this regulation, during the first two years of the accession the EU-15 was free to apply national rules; during the following three years EU-15 can continue to apply national rules but at the end of the period, they will be asked to open their labor markets. If EU-15 countries show any serious disturbances, they will be allowed for two additional years to apply their national rules. By 2011 all members will have to obey the European Commission Rules and free their labor markets. Besides these regulations, some of the EU-15 countries 3 never restricted labor movements, and some others 4 freed labor movements after the first two year period. In this sense, analysis of migration flows from ten new members of the EU becomes important. Svaton and Warin (2007) tested the determinants of migration flows into EU-15 countries from 72 countries from all over the world, by using an extended gravity model. The transition economies are among these 72 destination countries, but they are not analyzed separately. Gallardo-Sejas, Pareja and Llorca-Vivero (2006) also analyzed the determinants of European immigration from 139 origin countries. In one part of his paper Fertig (2001) performed a simulation to analyze the potential migration from ten transition countries to Germany. Similarly, Bijak, Kupiszewski and Kicinger (2004) worked with different scenarios to forecast migration flows for 27 EU countries in the 2002-2052 period. Drinkwater (2003) analyzed the subject from a different aspect and examined the willingness to move from several Central and Eastern European Countries. 2 International migration flows is not a new concern for EU and it came into question during all enlargement processes. Empirical studies followed these discussions. For example, Straubhaar (1986) studied the determinants of labor migrations from Italy, Spain, Greece, Portugal and Turkey to EU countries, just before Spain s and Portugal s accession. But the difference of the 2004 and post-2004 enlargements is the larger populations of the new accession countries and their relatively poor economic indicators. 3 Ireland, Sweden, United Kingdom. 4 Greece, Finland, Spain, Portugal, Italy.
This study aims to analyze empirically the determinants of international migration from ten transition economies to 23 OECD countries 5. These ten transition economies are Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovak Republic and Slovenia. Turkey, as a candidate country, is also included in the study as the eleventh country. The study covers the 1995-2005 period and data availability determined the time content of the study. This paper differs from the previous studies from two aspects: First, the country content of this study is different. Second, the data set includes migration flows from each origin country to every 23 destination countries separately. Therefore, the determinants of migration for every origin and destination countries are tested separately. This paper is organized as follows: Section 2 sets out the model and describes the data set. Empirical findings are reported in Section 3. Section 4 provides a brief summary of the findings and presents our concluding remarks. 2. Econometric Specification and Data The literature on international migration flows displayed that both origin country characteristics and destination country characteristics are effective on the size and direction of migration flows. Therefore in this paper both of these factors are considered. The model estimated is as follows: ln (M i,j,t ) = β 0 + β 1 ln(y i,t ) + β 2 ln(y j,t ) + β 3 ln(u i,t ) + β 4 ln(u j,t ) + β 5 (MS j,t ) + β 6 ln(d i,j ) + β 7 ln(p i,t ) + β 8 (L i,j ) + ε t (1) Where the indices i, j and t denote the origin country, destination country and time respectively. ε t represents the error term. All the variables except MS and L are logarithmic. The dependent variable M i,j,t represents the migration flows and it is the number of people migrated from country i to country j in year t 6. Migration data are obtained from OECD StatExtracts Database. The limits of this database also determined the content of this study. The database covers the legal migration flows so undocumented migration is not covered in 5 Destination countries are Austria, Australia, Belgium, Canada, Czech Republic, Denmark, Hungary, Finland, France, Germany, Greece, Italy, Netherlands, Norway, Poland, Portugal, Slovak Republic, Spain, Sweden, Switzerland, Turkey, United Kingdom and United States of America. 6 In the empirical literature (E.g. Hatton, 1995; Karras and Chiswick, 1999; Fertig, 2001; Bijak, Kupiszewski and Kicinger, 2004; Zoubanov, 2004; Svaton and Warin, 2007) a second specification of the dependent variable have been used. In these studies instead of migration flows, migration rates are taken as dependent variable. Migration rate is calculated by dividing the migration flows by the origin country s population or total labor force. In this study also a similar methodology is followed but we could not obtain significant results. So this results are not presented in the paper.
the paper. One important shortcoming of the database is that data for Russia is not available, which was an important route for migrants from transition economies. The explanations of the independent variables of the models and data sources are presented in Appendix 1. Y i,t and Y j,t are the per capita GDP levels in origin and destination countries. Since people will move from low income countries to high income countries, the theoretically expected sign of Y i,t is negative and Y j,t is positive. On the other hand, employment problems are the other major economic reason behind migration flows. Therefore the expected sign of the unemployment rate in the origin country, represented with U i,t, is positive; and the expected sign of the unemployment rate in the destination country, U j,t, is negative. 7 Migrant Stock (MS) in the destination country is another important factor. A large stock of migrants already resided in the destination country might show that the country in question has laws that favor migrants. Even though we do not have the data so cannot be sure, it increases the possibility that the potential labor has some relatives or friends in the destination country, which eases migration. Therefore the expected sign of this variable is positive. The distance between the origin and destination countries (D) increases the cost of moving, so its expected sign is negative. Final two independent variables are the population of the origin country (P i,t ) and the dummy variable for common language. Both variables are expected to carry positive signs. In related literature, some authors used per capita GDP differences and unemployment rate differences instead of using their levels separately in the models. In order to check whether these relative differences are important or not, a second model is also specified. The model is specified as follows: ln(m i,j,t )= β 0 + β 1 ln(rely i,j,t )+ β 2 ln(relu i,j,t )+ β 5 (MS j,t )+ β 6 ln(d i,j )+ β 7 ln(p i,t )+ β 8 (L i,j )+ ε t (2) In Model (2), RELY is the per capita GDP in country i less the per capita GDP in j. Similarly RELU is the unemployment rate in country i less the unemployment rate in country j. The expected signs of these variables are negative and positive respectively. In order to estimate the above models, migration flows from 11 countries to 23 destination countries for the period 1995-2005 are pooled. So an unbalanced panel of 876 observations is obtained. The models 7 Inflation rates in the origin and destination countries, gini coefficient in the destination country are the economic; average schooling, sharing a common border and civil liberties are the other factors that are stated as effective on international migration flows in the related literature. However in our estimations these variables were not significant and their inclusion did not change the coefficients of the other variables, so they were excluded from the models.
are first estimated by simply pooling the data and using OLS, then panel data regressions are run when necessary. In order to determine whether to use fixed or random effect models with panel data, Hausman test is performed and test results are presented in the related tables. All the tests are performed by using STATA. 3. Empirical Results Model (1) and Model (2) are estimated using annual data for the period from 1995 to 2005. At the first stage, whole database is included in the regressions. Table 1 presents these results. In the World Development Indicators database the stock of migrants data (MS) is available for the years 1995, 2000 and 2005 only. Therefore inclusion of this variable in the models causes a decline in degrees of freedom. In order to obtain more reliable results, models are estimated twice, with and without MS. The models with MS are marked with a, models without MS are marked with b in the tables. [Table 1] In Table 1 regression results for Model (1a) and Model (1b) are very similar: In both models the variables Y i,t and Y j,t,j are significant and the signs of the estimated coefficients are compatible with theoretical expectations. On the other hand, variable U i,t carry the correct sign but it is insignificant and variable U j,t is significant but with the wrong sign it in all models. These results show that income levels in the origin and destination countries affect the migration decisions. However, unemployment rates are not effective in contrast to economic expectations. The results of Model (2a) and Model (2b) also support these findings. The relative income levels of the countries (RELY) is significant with the correct sign indicating that people migrate from countries with low income levels to countries with high income levels. RELU is significant in only a few models and also generally carries the wrong sign. The other variables MS j,t, D i,j, P i,t and L i, are all significant. The economic variables sometimes affect migration flows in the following years. So determinants of migration are also regressed against one year lagged independent variables. The results of these regressions are presented in Table 2. The results obtained for Model (1) are exactly the same that were presented in Table 1: Income levels of origin and destination countries are important; however unemployment rates do not play a significant role. But in the case of Model (2) almost all variables are insignificant, which means that the lagged values of per capita income and unemployment rate differences do not explain migration flows.
[Table 2] In this paper our focus is on migration flows from transition economies to EU countries. So at this stage we run regressions for these countries only. Table 3 presents these results. Again results from Model (1) and Model (2) support each other. With regard to per capita income levels, our findings suggest that this variable explains migration flows. However unemployment rates do not play a role in migration flows within EU. All other control variables are significant with expected signs. [Table 3] In order to analyze whether the determinants of international migration changed within time or not, we also run yearly regressions for the years 1996, 2000 and 2005. In Table 4, we present these results. In 1995, the only significant variable was the MS, economic factors being unimportant. However, in 2000 and partly in 2005 economic variables became important. Also destination country characteristics are more important than origin country characteristics in 2000 and 2005. [Table 4] The determinants of international migration are analyzed in the above tables but the motivation of emigration might differ for origin countries. Therefore separate regressions are run for each origin country and the results are presented in Table 5 8. For the all countries per capita income of the destination country (Y j ) is found as a very important determinant of migration except Slovak Republic. And for Slovak Republic the only economic factor effective is RELU, which is not significant with the correct sign for any other country. So, where Y j is important for all countries, RELU is important for Slovak Republic. People migrate from Slovak Republic to search for a job. [Table 5] Also determinants of immigration might be different for destination countries. In order to explore cross-country differences, separate regressions are run. The results are presented in Table 6 9. For all nations Y j seems more important than Y i. Only for Spain, both Y i and Y j are important. For Spain and Poland, RELY; for Czech Republic, Hungary, Netherlands and Spain RELU is important. 8 Number of observations (N) is less than 40 for Estonia, Latvia and Lithuania therefore these results must be analyzed cautiously. 9 Estimation results for Belgium, Finland, France, Greece, Italy, Portugal, Slovak Republic, Sweden and UK are not presented because N is less than 40 for these countries.
[Table 6] 4. Concluding Remarks The purpose of this study was to test empirically the determinants of international migration. For this aim, migration flows from ten transition economies and Turkey to 23 OECD destination countries are analyzed for the 1995-2005 period. In the analyses the focus was on the migration flows to the EU countries. Also different regressions are run for each origin and destination countries, to explore the cross-country differences in determinants. Empirical findings reveal that per capita income of the destination countries are important when all countries are analyzed together. On the other hand, unemployment rates are not found as factors affecting migration flows. Generally it can be said that destination country characteristics determine the direction and size of migration flows. When models are estimated for countries separately, the results differ from the above findings. In these regressions unemployment rates became important. For example, relative unemployment rate is the factor that affects migration flows from Slovak Republic to EU countries. And also Czech Republic, Hungary, Netherlands and Spain attract migrants for their relatively lower unemployment rates. Another important finding of the paper is that, determinants of migration change in time: In 1995, after the fall of the Soviet-block, non-economic factors were important, but in 2000 economic factors became important.
References Bijak, J., M. Kupiszewski and A. Kicinger (2004), International Migration Scenarios for 27 European Countries, 2002-2052, CEFMR Working Paper, 4. Drinkwater, S. (2003), Go West? Assessing the Willingness to Move from Central and Eastern European Countries, University of Surrey Discussion Papers in Economics, DP 05/03. Fertig, M. (2001), The Economic Impact of EU-Enlargement: Assessing the Migration Potential, Empirical Economics, 26, 707-720. Gallardo-Sejas, H., S. Pareja and R. Llorca-Vivero (2006), Determinants of European Immigration: A Cross Country Amalysis, Applied Economics Letters, 13, 769-773. Hatton, T.J. (1995), A Model of U.K. Emigration, 1870-1913, The Review of Economics and Statistics, 77(3), 407-415. Karras, G. and C.U. Chiswick (1999), Macroeconomic Determinants of Migration: The Case of Germany 1964-1988, International Migration, 37(4), 657-677. Straubhaar, T. (1986), The Causes of International Labor Migrations- A Demand- Determined Approach, International Migration Review, 20 (4), 835-855. Svaton, P. and T. Warin (2007), European Migration: Welfare Migration or Economic Migration, International Trade and Finance Association Working Papers 2007, Paper: 9. Zoubanov, N. (2004), Assessing General and Country-Specific Determinants of Migration in the European Union A Panel Data Approach, paper presented at the IZA Summer School.
Appendix 1: Variable Explanations and Data Sources Variable Explanation Source Y Per capita GDP, PPP (constant 2000 World Development Indicators international $) Database U Unemployment Rate World Development Indicators Database MS Migrant Stock (% of population) World Development Indicators Database D Distance between the capital cities of i www.indo.com/distance and j measured in km. P Population of origin country World Development Indicators L Language dummy, takes the value of 1 if two countries use the same language, or if a large share of population use the same language Database
Table 1: Determinants of Migration, whole data set. Variables Model (1a) Model (1b) Model (2a) Model (2b) Yi Yj RELY Ui Uj RELU -1,13 (0,32)*** 2,58 (0,58)*** 0,309 (0,256) 1,309 (0,334)*** Di,j -1,64 (0,32)*** Li,j 2,86 (0,79)*** Pi i 0,54 (0,08)*** MSj 0,171 (0,02)*** Constant -17,60 (7,56) Panel -1,14 (0,65)* 2,80 (0,47)*** -0,65 (0,165)*** 3,50 (0,27)*** -0,35 (0,37) 3,63 (0,22)*** -2,14 (0,47)*** -2,57 (0,60)*** -2,57 (0,25)*** -3,72 (0,34)*** -0,02 (0,62) 0,319 (0,139)** 0,36 (0,27) 1,33 0,924 0,97 (0,284)*** (0,170)*** (0,14)*** -0,02-0,05 0,02-0,00001 (0,02) (0,02)** (0,01)** (0,01) -1,52-1,07-1,12-1,39-1,37-0,78-0,703 (0,315)*** (0,142)*** (0,15)*** (0,309)*** (0,306)*** (0,13)*** (0,15)*** 2,78 2,72 2,62 2,24 2,25 1,58 1,97 (0,54)*** (0,41)*** (0,29)*** (0,69)*** (0,52)*** (0,34)*** (0,305)*** -3,05 0,603 0,64 0,51 0,58 0,502-1,17 (3,58) (0,05)*** (0,19)*** (0,09)*** (0,15)*** (0,05)*** (0,68)* 0,153 0,16 0,163 (0,02)*** (0,02)*** (0,02)*** 40,27-31,74-36,56 2,20 1,29 1,93 29,89 (58,97) (3,39) (5,50) (1,61) (2,69) (0,92) (11,24) FIXED RANDOM RANDOM FIXED Hausman 50,72 8,79 (0,26) 4,85 (0,56) 12,30 (0,03) F 27,25 14,32 73,73 310,99 c 33,02 116,52 c 82,48 39,70 R 2 0,44 0,35 0,72 0,04 0,34 0,25 0,72 0,34 0,38 0,29 0,77 0,37 0,27 0,18 0,63 0,00 N 230 230 876 876 230 230 876 876 *** significant at 1 percent, ** significant at 5 percent, * significant at 10 percent. Numbers in parenthesis are standard errors. c: Wald test result
Table 2: Determinants of migration, whole data set, one-year lagged variables. Variables Model (1a) Model (1b) Model (2a) Model (2b) Y i, t-1-0,91 (0,45)** Y j, t-1 1,40 (0,69)** RELY t-1 U i, t-1 0,019 (0,32) U j t-1 1,00 (0,39)** RELU t-1 D i,j, t-1-1,37 (0,41)*** L i,j, t-1 2,55 (0,97)*** P i,t-1 0,48 (0,13)*** MS j, t-1 1,28 (0,33)*** Constant -7,22 (9,56) -0,77 (0,85) 1,67 (0,61)*** -0,18 (0,71) 1,07 (0,33)*** -1,37 (0,39)*** 2,57 (0,73)*** 0,47 (0,26)* 1,17 (0,31)*** -10,53 (12,15) -0,63 (0,17)*** 3,32 (0,25)*** 0,35 (0,14)** 0,79 (0,16)*** -1,09 (0,14)*** 2,72 (0,42)*** 0,62 (0,05)*** -30,29 (3,29) -0,21 (0,41) 3,46 (0,22)*** 0,56 (0,28)** 0,86 (0,13)*** -1,16 (0,15)*** 2,62 (0,30)*** 0,67 (0,20)*** -36,82 (5,79) -0,24 (0,51) -0,04 (0,16) -0,43 (0,48) 1,94 (1,08)* 0,44 (0,17)** 0,84 (0,40)** -1,08 (2,76) -0,47 (0,61) -0,14 (0,17) -0,519 (0,44) 1,99 (0,83)** 0,41 (0,24)* 0,84 (0,36)** -0,52 (3,97) -0,76 (0,208) 0,01 (0,07) -0,39 (0,16) 1,55 (0,48) 0,47 (0,06) -0,61 (1,00) -1,54 (0,26) -0,18 (0,08) -0,47 (0,18) 1,97 (0,37) -1,18 (0,65) 26,66 (10,90) Panel RANDOM RANDOM RANDOM FIXED Hausman 5,55 (0,69) 9,42 (0,22) 4,74 (0,57) 11,70 (0,03) F 13,08 51,13 c 78,84 312,21 c 3,93 14,32 c 19,38 9,58 c R 2 0,34 0,25 0,74 0,33 0,36 0,26 0,71 0,35 0,17 0,08 0,54 0,17 0,14 0,08 0,69 0,04 N 143 143 832 832 101 101 542 542 *** significant at 1 percent, ** significant at 5 percent, * significant at 10 percent. Numbers in parenthesis are standard errors. c: Wald test result
Table 3: Determinants of migration, migration flows to EU countries Variables Model (1) Total Migrant Flows to EU Model (2) Total Migrant Flows to EU Yi -0,65 (0,179)*** -0,45 (0,34) Yj 4,05 (0,219)*** 4,15 (0,23)*** RELY Ui Uj RELU 0,027 (0,145) 1,044 (0,155)*** Di,j -1,89 (0,194)*** Li,j 1,61 (0,31)*** Pi i 0,694 (0,055)*** Constant -35,76 (3,03) Panel 0,281 (0,26) 1,102 (0,147)*** -1,88 (0,217)*** 1,53 (0,308)*** 0,746 (0,14)*** -40,31 (4,90) RANDOM -3,17 (0,259)*** -0,004 (0,01) -1,86 (0,19)*** 0,463 (0,223)** 0,57 (0,06)*** 4,39 (1,00) -4,51 (0,35) -0,02 (0,014) -1,39 (0,23) 1,01 (0,32) -1,08 (0,67) 30,87 (11,05) FIXED Hausman 8,41 (0,298) 22,38 F 97,42 c 418,0 84,23 51,90 R 2 0,43 0,35 0,75 0,43 0,32 0,27 0,72 0,00 N 717 717 717 717 *** significant at 1 percent, ** significant at 5 percent, * significant at 10 percent. Numbers in parenthesis are standard errors. c: Wald test result
Table 4: Determinants of migration, yearly regressions Variables 1995 2000 2005 Yi 0.231 (1.158) Yj 1.817 (0.910)* Ui -0,459 (0,761) Uj 0,229 (0,587) MSj 0,193 (0,078)** Di,j -1,151 (1,103) Li,j 1,433 (1,498) Pi i 0,198 (0,293) Constant -12,714 (17,79) F 5,94 R 2 0,321-0,989 (0,466)** -1,717 (0,687)** 3,437 2,751 (0,929)*** (1,538)* 0,101 0,891 (0,359) (0,502)* 1,818 1,479 (0,589)*** (0,841)* 0,150 0,178 (0,041)*** (0,055)*** -1,593-1,699 (0,509)*** (0,501) 4,342 2,646 (1,544)*** (1,155)** 0,652 0,411 (0,128)*** (0,158)** -29,80-13,17 (11,53) (20,52) 16,24 12,29 0,492 0,503 N 44 94 91 *** significant at 1 percent, ** significant at 5 percent, * significant at 10 percent. Numbers in parenthesis are standard errors.
Table 5: Determinants of migration, regression results by origin countries Variables Bulgaria Czech Republic Estonia Hungary Latvia Lithuania Yi Yj RELY 0,177 (1,697) 3,93 (0,547)*** -8,93 (1,49)*** 1,44 (7,36) 4,68 (0,529)*** -4,34 (0,723)*** -0,039 (3,84) 3,63 (0,61)*** -5,27 (0,94)*** -2,38 (1,96) 6,51 (0,94)*** -6,16 (1,65)*** -6,95 (11,8) 3,74 (1,19)*** -4,28 (2,54) 4,42 (2,71) 6,81 (1,56)*** Ui 0,788 (0,710) -0,404 (0,941) 0,302 (1,24) 0,08 (1,505) -1,405 (6,00) 2,43 (1,92) Uj 1,477 (0,373)*** 2,238 (0,397)*** 0,814 (0,362)** 2,05 (0,715)*** 0,838 (0,688) 1,04 (0,96) RELU -0,07 (0,034)** -0,189 (0,056)*** -0,07 (0,038)* -0,098 (0,104) 0,008 (0,08) 0,02 (0,128) Di,j -0,380 (0,447) 0,04 (0,410) -3,568 (0,472)*** -3,56 (0,496)*** -2,15 (0,235)*** -2,28 (0,208)*** -3,01 (0,553)*** -2,69 (0,57)*** -2,99 (1,29)** -2,31 (1,41) 3,02 (1,30)** 3,05 (1,51)* Li,j 8,24 10,165-1,58-1,14 6,54 4,92 (0,523)*** (0,92)*** (0,513)*** (0,622)* (0,73)*** (0,99)*** Pi i 2,587 (7,71) -11,89 (5,21)** 68,87 (201,6) -80,12 (34,15)** -11,56 (55,12) -51,62 (13,08)*** -0,646 (0,303)** -0,797 (0,113)*** -42,63 (76,57) -12,33 (11,98) -3,00 (17,60) -32,97 (13,8)** Constant -78,79 (135,65) 198,77 (82,76) -1160,15 (3324,70) 1313,15 (551,30) 135,39 (815,57) 743,03 (185,02) -20,84 (16,43) 31,34 (1,68) 667,29 (1228,60) 194,55 (176,34) -77,02 (264,90) 495,85 (207,32) F 86,92 92,29 80,96 45,99 71,97 82,28 612,34 280,97 4,60 4,90 20,71 11,10 R 2 0,606 0,554 0,826 0,734 0,89 0,88 0,69 0,601 0,390 0,360 0,610 0,405-5,82 (3,98) N 90 90 45 45 38 38 51 51 32 32 37 37 *** significant at 1 percent, ** significant at 5 percent, * significant at 10 percent. Numbers in parenthesis are standard errors.
Table 5: Continues Variables Poland Romania Slovak Republic Slovenia Turkey Yi -1,14 (1,907) -0,76 (1,91) 1,94 (1,59) -5,25 (2,44)** -0,85 (2,60) Yj 4,44 (0,49)*** 3,39 (0,67)*** -0,033 (0,80) 5,35 (1,03)*** 3,74 (0,501)*** RELY -5,88 (0,936) -8,25 (1,55)*** 0,402 (1,11) -4,27 (1,11)*** -8,81 (1,33) Ui 0,92 (0,82) 2,12 (1,59) 1,22 (1,54) -2,39 (2,51) 0,77 (1,06) Uj 1,95 (0,34)*** 1,89 (0,42)*** -0,65 (0,44) 1,59 (0,79)* 0,481 (0,36) RELU -0,025 (0,02) -0,153 (0,44)*** 0,119 (0,04)*** -0,079 (0,09) -0,008 (0,044) Di,j -0,77-0,457 1,11 1,32-1,37-1,26-3,01-2,87-1,43-1,006 (0,29)*** (0,307) (0,47)** (0,42)*** (0,31) (0,27)*** (0,34)*** (0,32)*** (0,42)*** (0,374) Li,j 4,56 (0,35)*** 4,49 (0,35)*** -1,36 (0,49) -1,56 (0,45)*** Pi i -30,514 (38,95) -117,59 (33,6) -6,41 (17,7) -26,87 (11,23)** -196,45 (116,4) -166,8 (64,5)** -5,29 (16,8) -6,72 (12,9) -3,75 (5,64) 1,67 (2,74) Constant 502,10 (690,1) 2065,7 (587,3) 76,5 (313,2) 459,85 (190,25) 3035,5 (1799,09) 2595,2 (1000,6) 89,56 (227,8) 114,3 (188,2) 47,01 (87,16) -17,05 (49,29) F 19,91 14,84 27,30 38,88 10,76 19,22 27,75 30,49 20,06 26,12 R 2 0,35 0,21 0,32 0,27 0,38 0,42 0,71 0,67 0,27 0,26 N 159 159 152 152 59 59 46 46 167 167 *** significant at 1 percent, ** significant at 5 percent, * significant at 10 percent. Numbers in parenthesis are standard errors.
Table 6: Determinants of migration, regression results by destination countries Variables Austria Canada Czech Republic Denmark Germany Hungary Yi -0,65 (0,254)** -2,42 (0,69)*** 0,541 (0,25)** -1,86 (1,78) 0,07 (1,71) 0,968 (0,68) Yj 3,81 (1,72)** 12,77 (1,82)*** -1,64 (1,25) 0,919 (1,70) 9,39 (2,46)*** -7,51 (3,35)** RELY -1,27 (0,74)* -6,06 (3,73) 0,134 (0,95) -3,88 (2,60) 10,15 (4,37)** 0,79 (2,01) Ui -0,06 (1,66) -1,68 (0,197)*** 0,751 (0,137)*** -0,32 (0,613) -0,23 (0,32) 1,30 (0,35)*** Uj 1,003 (0,397)** 3,93 (0,95)*** -0,507 (0,45) 0,174 (0,82) -0,92 (0,38)** -0,83 (0,85) RELU -0,006 (0,01) -0,07 (0,021)*** 0,06 (0,01)*** -0,06 (0,04) -0,01 (0,04) 0,069 (0,02)*** Di,j -0,86 (0,205)*** -0,796 (0,21)*** -27,48 (4,06)*** -19,15 (6,67)*** 0,08 (0,288) -0,35 (0,35) -9,72 (3,10)*** -11,17 (3,22)*** -4,29 (1,98)** 2,06 (2,61) 2,86 (0,79)*** 1,43 (0,61)** Li,j 6,31 (0,801)*** 5,82 (0,90)*** Pi i 0,733 (0,09)*** 0,753 (0,09)*** 0,124 (0,117)*** 0,14 (0,16) 0,979 (0,09)*** 0,922 (0,127)*** 4,94 (0,79)*** 4,98 (0,76)*** -0,56 (0,208)*** -0,956 (0,23)*** 0,24 (0,05)*** 0,287 (0,044)*** Constant -36,34 (17,88) -1,97 (1,54) -1,21 (18,51) 81,02 (25,00) 4,49 (13,9) -4,52 (3,37) -45,95 (16,28) -47,97 (11,26) -61,50 (11,37) 10,65 (8,42) 59,57 (31,70) -4,01 (2,26) F 55,32 119,41 35,28 8,11 76,86 138,25 77,18 125,22 34,12 18,70 9,71 15,29 R 2 0,86 0,82 0,76 0,33 0,87 0,86 0,91 0,91 0,86 0,65 0,59 0,52 N 55 55 44 44 76 76 44 44 43 43 47 47 *** significant at 1 percent, ** significant at 5 percent, * significant at 10 percent. Numbers in parenthesis are standard errors.
Table 6: Continues Variables Netherlands Poland Spain USA Yi 1,907 (0,336)*** -0,002 (0,281) -4,51 (0,547)*** -0.49 (0.513) Yj -1,35 (1,505) 2,96 (1,02) 53,93 (12,49)*** 2.40 (1.06)** RELY 6,26 (0,952) -0,963 (0,401)** -6,20 (1,90)*** 0.177 (2.44) Ui 0,543 (0,139)*** 1,26 (0,166) 0,339 (0,373) -0.132 (0.136) Uj 0,403 (0,216)* 2,13 (0,453) 4,82 (2,88)* 0.04 (0.34) RELU 0,049 (0,012)*** -0,049 (0,02)** 0,17 (0,02)*** -0.007 (0.01) Di,j 4,182 (0,645)*** 4,89 (0,67)*** -1,15 (0,59) -2,14 (0,90)** -11,27 (2,45)*** -9,81 (3,98)** -22.85 (3.83)*** -18.90 (3.72)*** Li,j 0,845 (0,204) 0,856 (0,334)** Pi i 1,123 (0,048)*** 1,11 (0,04)*** 0,888 (0,082) 0,708 (0,119)*** 0,514 (0,173)*** 0,781 (0,186)*** 0.34 (0.09)*** 0.267 (0.12)** Constant -30,59 (15,17) -29,83 (2,75) -42,64 (8,00) 0,05 (2,58) -474,39 (128,70) 30,03 (14,94) 71.42 (11.39) 77.71 (13.25) F 201,88 228,03 36,70 16,79 43,67 29,29 60.47 (0.00) R 2 0,94 0,93 0,83 0,50 0,72 0,47 0.85 0.81 N 69 69 80 80 76 76 44 44 *** significant at 1 percent, ** significant at 5 percent, * significant at 10 percent. Numbers in parenthesis are standard errors.