| (1) | (2) | (3) | (4) |
---|
| Full sample | 2007–2009 | 2010–2012 | 2013–2015 |
Panel A: IV-RE regressions |
Treatment var. (Trt) | 0.031** (0.006) | 0.003 (0.020) | 0.008 (0.006) | 0.073** (0.018) |
Other controls? | Yes | Yes | Yes | Yes |
Panel B: GLS-RE regressions |
Treatment var. (Trt) | 0.078** (0.012) | 0.018 (0.013) | 0.030** (0.011) | 0.174** (0.041) |
Other controls? | Yes | Yes | Yes | Yes |
Observations | 17,214 | 2468 | 11,192 | 3554 |
- Notes: Robust standard errors in parenthesis, **p < 0.01, *p < 0.05, + p < 0.1
- The dependent variable is weight-for-age zscore. Estimated treatment effects from regressions using the full sample, 2007–2009, 2010–2012 and 2013–2015 birth cohorts presented. The treatment variable (Trt) is a dummy equal to one if the child had access to the midday meals program. Panel A provides results for the IV-Random Effects model and Panel B provides results for GLS-Random Effects Model. Other controls include child age, age squared, gender, zscore of first weight, pcare, zscore of birthweight, ethnicity, CDC status and birth-year fixed effects either as direct controls or instruments