Table 5.
Dynamic models: 1987–1998 sub-sample
| (1) | (2) | |
|---|---|---|
| Lagged Gini | 0.575*** (0.049) | 0.476*** (0.103) |
| Election dummy | −0.015*** (0.005) | −0.021* (0.011) |
| Per-capita net income (1000 Yuan) | 0.001 (0.001) | |
| Per-capita net income squared | −7.35e–06 (8.44e–06) | |
| Log village population | 0.049 (0.070) | |
| Share of emigrant workers (%) | 0.0001 (0.0002) | |
| Unemployment rate (%) | 0.002* (0.001) | |
| CV of household size | 0.026 (0.026) | |
| CV of average schooling years of household adults | −0.150 (0.107) | |
| CV of per-capita household landholding | −0.023* (0.012) | |
| CV of household wage earners | 0.006*** (0.002) | |
| Constant | 0.002*** (0.000) | 0.002* (0.001) |
| Year Dummies | Y | Y |
|
| ||
| m1 | −4.01 | −3.44 |
| m2 | −1.67 | −1.50 |
| Sargan p-value | 1.000 | 1.000 |
Notes: The dependent variable is the Gini coefficient. The data range is from 1987 to 1998. After first differencing and taking lags 442 observations for 47 villages are used in the estimations. All results are estimated using the two-step GMM method proposed by Arellano and Bond (1991), and Bond (2002). Year dummies are included in all models. Both regressions treat elections endogenous. The timing of a province’s adoption of the OLVC and its interactive terms with the number of surnames and the percentage of population of the largest surname in a village are used as instruments for village elections. The linear probability model is assumed for the first-stage estimation. Figures in parentheses are robust standard errors.
indicate the 10%, 5%, and 1% significance level, respectively.