Working Paper: NBER ID: w15765
Authors: Robert Kaestner; Ofer Malamud
Abstract: This paper uses data from the Mexican Family Life Survey (MxFLS) to examine the patterns of selection of male, Mexican migrants to the United States. We confirm previous findings that Mexican migrants are selected from the middle of the education distribution, but show that there is no evidence for selection of migrants on cognitive ability. We demonstrate that migrants are also selected from the middle of the observed skill distribution, as measured by predicted wages. However, controlling for proxies of the costs of migration, we find substantially less evidence of "intermediate selection" on observed skill. We find little evidence for selection on unobserved skill, with or without controls for the costs of migration. Finally, we show directly that the decision to migrate is highly correlated with differential returns to observable skill and the costs of migration. Overall, these findings are consistent with the predictions of the canonical model of migration.
Keywords: Migration; Self-Selection; Mexico; Labor Market
JEL Codes: J24; J61
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Mexican migrants are primarily selected from the middle of the education distribution (I25) | Evidence of intermediate selection (C52) |
| No evidence for selection based on cognitive ability (D91) | Migrant and non-migrant distributions of cognitive test scores are virtually identical (J61) |
| Decision to migrate is strongly correlated with differential returns to observable skill (J61) | Migrants are more likely to move when the return to their observable skills is higher in the U.S. than in Mexico (J61) |
| Controlling for proxies of migration costs diminishes evidence of selection from the middle of the observed skill distribution (J69) | More complete controls could reveal different selection patterns (C99) |