Table 3.
Ols and Within-Household Estimates of the Return to Migration. Households with Two or More Members in the Matched Sample
| Dependent variable = ln(earnings); Coefficient on = 1 if migrant | |||
|---|---|---|---|
| Full sample, 1865 | Rural, 1865 | Urban, 1865 | |
| Panel A. Unweighted | |||
| OLS | 0.545 (0.027) |
0.607 (0.034) |
0.384 (0.044) |
| Within household | 0.511 (0.035) |
0.508 (0.045) |
0.508 (0.057) |
| Chi-squared | 1.49 | 7.47 | 8.31 |
| p-value | 0.2218 | 0.0063 | 0.0039 |
| N | 2,655 | 1,823 | 832 |
| Number of migrant-stayer pairs | 326 | 167 | 159 |
| Panel B. Weighted | |||
| OLS | 0.586 (0.029) |
0.609 (0.033) |
0.443 (0.067) |
| Within household | 0.542 (0.039) |
0.529 (0.042) |
0.561 (0.049) |
| Chi-squared | 2.13 | 4.60 | 5.65 |
| p-value | 0.1441 | 0.0320 | 0.0175 |
| N | 2,241 | 1,666 | 306 |
| Number of migrant-stayer pairs | 269 | 140 | 129 |
Notes: Each cell contains coefficient estimates and standard errors from regressions of ln(earnings) on a dummy variable equal to one for individuals living in the United States in 1900. Regressions also include controls for age and age squared. In each panel, the first row conducts an OLS regression for the restricted sample of households that have at least two matched members in the dataset and the second row adds household fixed effects. Panel B contains results from regressions weighted to reflect the urban status (full sample only), asset holdings, and occupational distribution of fathers in the full population. We conduct chi-squared tests of the null hypothesis that the OLS and within-household coefficients are equal.