Table 3.
Total energy | Distribution of total energy coefficient |
sig > 0 | sig < 0 | |||
---|---|---|---|---|---|---|
# of sp. | p25 | p50 | p75 | %-Sig | %-Sig | |
Panel A: Height | ||||||
All IV | 546 | −0.0182 | 0.0099 | 0.0288 | 35 | 0 |
All over-identified IV | 525 | −0.0179 | 0.0107 | 0.0289 | 36 | 0 |
CD > 1 P-val HJ > 5 | 137 | −0.0090 | 0.0034 | 0.0170 | 15 | 0 |
CD > 3 P-val HJ > 5 | 21 | 0.0009 | 0.0231 | 0.0438 | 57 | 0 |
Panel B: Weight | ||||||
All IV | 546 | −0.0061 | 0.0059 | 0.0159 | 36 | 0 |
All over-identified IV | 525 | −0.0036 | 0.0060 | 0.0159 | 38 | 0 |
CD > 1 P-val HJ > 5 | 129 | −0.0024 | 0.0050 | 0.0233 | 32 | 0 |
CD > 3 P-val HJ > 5 | 36 | 0.0142 | 0.0230 | 0.0239 | 83 | 0 |
CD = Robust Kleibergen-Paap F statistic, P-value, J = P-value of Hansen J stat × 100.
1st column: # of specifications that meet criteria; 2nd–4th col: percentile of distribution of estimated coefficients.
5th (6th) column: percent of estimated coefficients that are positive (negative) and significant at 5% significance level.
1st row: all specifications; 2nd row: all over-identified specifications for which # of IVs># of endogenous variables. Other rows include all specifications satisfying the indicated criteria based on the CD and HJ tests.
All specifications include breastfeeding, diarrhea, sex, age, and age squared as covariates and a seasonal dummy for the Philippines, and lagged height and lagged weight, both of which are treated as endogenous.
Height coefficients are divided by 1000 for presentation purposes.