Familiarity Breeds Investment August 2010 TABLE 4.Educated Migrants Portfolio FDI Log(migrant stock from 0.389* 0.344* 0.394* 0.439 0.376 0.439* dins) (0.0986) (0.0835) (0.0996) (0.130) (0.126) (0.131) %with tertiary education 1.376 1.414 2.087体 2.096* (0.901) (0.909) (0.840) (0.846) in management/FIRE 3.121* 3.696* -0.413 0.462 (1.485) (1.496) (1.458) (1.499) Log(product of GDPs) 0.0882 0.0949 0.0875 0.140* 0.155# 0.140* (0.0578) (0.0583) (0.0577) (0.0539) (0.0570) (0.0538) Log(distance) -0.636* -0.607* -0.616* -0.904 -0.899* -0.901* (0.310) (0.309) (0.310) (0.384) (0.393) (0.386) Common border -0.636 -0.675 -0.629 0.140 0.0562 0.141 (0.574) (0.576) (0.574) (0.877) (0.885) (0.876) Official common language 0.0106 0.0847 0.00951 1.545* 1.621* 1.544* (0.397) (0.394) (0.399) (0.686) (0.682) (0.685) Correlation of growth rates -0.156 -0.143 -0.153 0.176 0.182 0.177 (0.273) (0.276) (0.274) (0.373) (0.370) (0.374) Common exchange 1.017* 1.046* 1.006* -0.0837 -0.0210 -0.0855 rate peg (0.602) (0.615) (0.599) (0.847 (0.848) (0.844) Dual taxation treaty 0.754* 0.748* 0.750 0.895* 0.877* 0.895* (0.354) (0.361) (0.355) (0.467) (0.466) (0.467 Preferential trade 0.000456 0.106 0.0551 0.301 0.263 0.309 agreement (0.619) (0.632) (0.627) (0.648) (0.681) (0.656) Common legal heritage -0.0787 -0.0869 -0.0788 0.643 0.606 0.644 (0.219) (0.225) (0.221) (0.409) (0.413) (0.409) Common dominant religion 0.8054 0.821* 0.8064 0.275 0.295 0.275 (0.306) (0.312) (0.311) (0.448) (0.454) (0.449) Genetic distance 0.000173 0.0000791 0.000137 0.0000412 -0.000134 0.0000360 (0.000593) (0.000548) (0.000583) (0.000702) (0.000736) (0.000711) Log(bilateral telephone 0.160 0.171* 0.163 0.220 0.231 0.220 volume) (0.102) (0.0970) (0.101) (0.189) (0.195) (0.189) Constant -63.70 -68.74 -63.33 -116.2* -127.4* -116.0* (51.86) (52.23) (51.76) (47.85) (50.57 (47.76) Observations 1,993 1,993 1,993 1,365 1,365 1,365 Adjusted R2 0.840 0.839 0.840 0.651 0.649 0.651 FDI,foreign direct investment:GDP,gross domestic product. Robust standard errors in parentheses. Dependent variable:log(portfolio investment)in columns 1-3;log(FDI)in columns 4-6. Robust standard errors clustered by both origin and destination country. All models include both origin and destination dummy variables. *p<.10;*p<.05. countries and from rich countries into poor countries.27 the size of the migrant network has a positive and statis- In all four specifications,we find that migrant networks tically significant influence on both portfolio and FDI. have a positive and statistically significant effect on Neither education level nor sector of employment is bilateral investment. significantly different from zero in the rich-rich country Table 6 repeats the subsampling and includes the pairs.These variables are significant and positive-in a measures of educated migrants.These results are sug- pattern consistent with Table 4-for rich-poor coun- gestive of our existing patterns:even controlling for try pairs.On its face,this seems plausible:investment education level and employment in the FIRE sectors, firms likely have multiple offices across a wide array of rich countries(consider the number of cross-border offices of a firm like AIG or JP Morgan).It is also 27 We do not look at investment from poor countries into rich coun- likely that rich countries have more transparent busi- tries for numerous reasons.First,there are relatively few migrants ness practices and government regulations than poorer from rich countries residing in poorer countries.Second,it is likely countries,resulting in a lower need for private informa- that the information asymmetries facing investors looking to invest tion.However,this result also points to the important in rich countries would be small.Finally,and perhaps most impor- tant,both CPIS and OECD data are very limited when looking at role that all migrant groups play in funneling portfo- investment from poor countries into rich countries. lio investment from rich countries into poor countries, 594Familiarity Breeds Investment August 2010 TABLE 4. Educated Migrants Portfolio FDI Log(migrant stock from 0.389∗∗ 0.344∗∗ 0.394∗∗ 0.439∗∗ 0.376∗∗ 0.439∗∗ d in s) (0.0986) (0.0835) (0.0996) (0.130) (0.126) (0.131) % with tertiary education 1.376 1.414 2.087∗∗ 2.096∗∗ (0.901) (0.909) (0.840) (0.846) % in management/FIRE 3.121∗∗ 3.696∗∗ −0.413 0.462 (1.485) (1.496) (1.458) (1.499) Log(product of GDPs) 0.0882 0.0949 0.0875 0.140∗∗ 0.155∗∗ 0.140∗∗ (0.0578) (0.0583) (0.0577) (0.0539) (0.0570) (0.0538) Log(distance) −0.636∗∗ −0.607∗∗ −0.616∗∗ −0.904∗∗ −0.899∗∗ −0.901∗∗ (0.310) (0.309) (0.310) (0.384) (0.393) (0.386) Common border −0.636 −0.675 −0.629 0.140 0.0562 0.141 (0.574) (0.576) (0.574) (0.877) (0.885) (0.876) Official common language 0.0106 0.0847 0.00951 1.545∗∗ 1.621∗∗ 1.544∗∗ (0.397) (0.394) (0.399) (0.686) (0.682) (0.685) Correlation of growth rates −0.156 −0.143 −0.153 0.176 0.182 0.177 (0.273) (0.276) (0.274) (0.373) (0.370) (0.374) Common exchange 1.017∗ 1.046∗ 1.006∗ −0.0837 −0.0210 −0.0855 rate peg (0.602) (0.615) (0.599) (0.847) (0.848) (0.844) Dual taxation treaty 0.754∗∗ 0.748∗∗ 0.750∗∗ 0.895∗ 0.877∗ 0.895∗ (0.354) (0.361) (0.355) (0.467) (0.466) (0.467) Preferential trade 0.000456 0.106 0.0551 0.301 0.263 0.309 agreement (0.619) (0.632) (0.627) (0.648) (0.681) (0.656) Common legal heritage −0.0787 −0.0869 −0.0788 0.643 0.606 0.644 (0.219) (0.225) (0.221) (0.409) (0.413) (0.409) Common dominant religion 0.805∗∗ 0.821∗∗ 0.806∗∗ 0.275 0.295 0.275 (0.306) (0.312) (0.311) (0.448) (0.454) (0.449) Genetic distance 0.000173 0.0000791 0.000137 0.0000412 −0.000134 0.0000360 (0.000593) (0.000548) (0.000583) (0.000702) (0.000736) (0.000711) Log(bilateral telephone 0.160 0.171∗ 0.163 0.220 0.231 0.220 volume) (0.102) (0.0970) (0.101) (0.189) (0.195) (0.189) Constant −63.70 −68.74 −63.33 −116.2∗∗ −127.4∗∗ −116.0∗∗ (51.86) (52.23) (51.76) (47.85) (50.57) (47.76) Observations 1,993 1,993 1,993 1,365 1,365 1,365 Adjusted R2 0.840 0.839 0.840 0.651 0.649 0.651 FDI, foreign direct investment; GDP, gross domestic product. Robust standard errors in parentheses. Dependent variable: log(portfolio investment) in columns 1–3; log(FDI) in columns 4–6. Robust standard errors clustered by both origin and destination country. All models include both origin and destination dummy variables. ∗ p < .10; ∗∗ p < .05. countries and from rich countries into poor countries.27 In all four specifications, we find that migrant networks have a positive and statistically significant effect on bilateral investment. Table 6 repeats the subsampling and includes the measures of educated migrants. These results are sug￾gestive of our existing patterns: even controlling for education level and employment in the FIRE sectors, 27 We do not look at investment from poor countries into rich coun￾tries for numerous reasons. First, there are relatively few migrants from rich countries residing in poorer countries. Second, it is likely that the information asymmetries facing investors looking to invest in rich countries would be small. Finally, and perhaps most impor￾tant, both CPIS and OECD data are very limited when looking at investment from poor countries into rich countries. the size of the migrant network has a positive and statis￾tically significant influence on both portfolio and FDI. Neither education level nor sector of employment is significantly different from zero in the rich–rich country pairs. These variables are significant and positive—in a pattern consistent with Table 4—for rich–poor coun￾try pairs. On its face, this seems plausible: investment firms likely have multiple offices across a wide array of rich countries (consider the number of cross-border offices of a firm like AIG or JP Morgan). It is also likely that rich countries have more transparent busi￾ness practices and government regulations than poorer countries, resulting in a lower need for private informa￾tion. However, this result also points to the important role that all migrant groups play in funneling portfo￾lio investment from rich countries into poor countries, 594