Robustness: Addressing Potential Endogeneity of Local Sex Ratios
Dependent Variable | Stock-Market Participation | |||||
---|---|---|---|---|---|---|
(1) | (2) | (3) | (4) | (5) | (6) | |
First son (β1) | −0.228*** (0.063) | −0.221*** (0.072) | −0.251*** (0.064) | −0.251*** (0.093) | −0.314*** (0.089) | −0.170 (0.221) |
Sex ratio (β2) | −0.022 (0.176) | −0.034 (0.168) | −0.032 (0.181) | −0.023 (0.172) | 0.052 (0.175) | −0.068 (0.155) |
First son × sex ratio (β3) | 0.337** (0.160) | 0.344** (0.167) | 0.320* (0.187) | 0.339** (0.170) | 0.273* (0.157) | 0.334* (0.179) |
Average household wealth, k | 0.022 (0.028) | −0.007 (0.031) | ||||
First son × average household wealth | −0.015 (0.048) | −0.060 (0.039) | ||||
Average household income, k | 0.002* (0.001) | 0.003*** (0.001) | ||||
First son × average household income | −0.000 (0.001) | −0.003*** (0.001) | ||||
Gender wage differential, m-f, k | 0.001 (0.005) | −0.015*** (0.005) | ||||
First son × gender wage differential | 0.011** (0.005) | 0.031*** (0.006) | ||||
Proportion with insurance | 0.212 (0.155) | 0.091 (0.187) | ||||
First son × proportion with insurance | 0.067 (0.174) | 0.088 (0.193) | ||||
Average # children | −0.083** (0.035) | −0.039 (0.042) | ||||
First son × average # children | 0.066 (0.043) | 0.035 (0.047) | ||||
Hausman test p-value | 0.780 | 0.993 | 0.653 | 0.877 | 0.568 | 0.671 |
Observations | 4,363 | 4,363 | 4,363 | 4,363 | 4,363 | 4,363 |
R-squared | 0.270 | 0.273 | 0.273 | 0.272 | 0.271 | 0.328 |
Other controls | Yes | Yes | Yes | Yes | Yes | Yes |
First son × other controls | Yes | Yes | Yes | Yes | Yes | Yes |
Prefecture fixed effects | Yes | Yes | Yes | Yes | Yes | Yes |
First son × prefecture fixed effects | Yes | Yes | Yes | Yes | Yes | Yes |
↵Notes: Data on county-level sex ratios are from the 2010 China population census. Data on other variables are from the 2013 CHFS. Results are estimated using Equation 1 based on our sample of CHFS households. The null hypothesis of the Hausman test is that the effect of high sex ratios is equal to that in the baseline—Column 1 of Table 3. Other controls include various parental and household characteristics—both parents’ age, education, hukou, political status, and occupational dummies, plus age of the first child, region of residence, and ethnicity. Interactions of the first-son dummy with these variables, as well as prefecture fixed effects, are also controlled for. Regressions are weighted by CHFS sampling weights. Standard errors clustered at the county level are given in parentheses. *p< 0.1, **p<0.05, ***p< 0.01.