Estimation Results

Table 3 [ PDF 91.9KB | 5 pages ] provides the OLS, 2SLS, and GMM estimates of the determinants of the proportion of children 6-24 years old who are attending school using both male and both female or same sex for the first two births as instruments, respectively. The positive effects of the number of children on the proportion of children 6-24 years old coming out in the OLS regression is suspect, because of the expected endogeneity of the number of children in this equation as per the quantity-quality of children trade-off literature. The data set confirms this endogeneity with F-values for the Wu-Hausman Test and Chi- Square values for Durbin-Wu-Hausman Test, indicating high significance implying a rejection of the null hypothesis that the number of children variable is exogenous in this equation. Thus, more consistent estimates are either the 2SLS or GMM estimates. Given that the presence of heteroscedasticity as indicated by the Pagan-Hall Test, the GMM estimators would give efficient estimates although magnitude wise the estimates are very similar. Given the z values of the estimates, the estimates using same sex as instruments are not as significant as the ones generated from using both male and both female as instruments. Thus, the more reliable estimate of the impact of the number of children on the proportion attending is the GMM estimate of about 15 percentage points average decline per additional child. The GMM estimate, however, also needs to be appreciated in the light of the significant over-identification statistic indicating some correlation between the instrument and the error term. Given the difference in the dependent variable used in this study and the other studies, the results cannot be compared directly.

The other results confirm most of the results from previous studies. The older the parents are, the lower is the proportion of children attending school. The higher the education of parents, the higher is the probability that children attend school. It is note worthy that the impact of mother’s education has about the same impact as father’s education. Other studies have shown that the mother’s education has higher impact on the education of children. Residing in urban areas has no distinct impact on school attendance. The availability of a school, indicated by the proportion of barangays with schools, has a positive impact on school attendance, although this is only true for elementary schools but not for secondary schools. The income variable is insignificant. The regional dummy variables are expected to pick-up whatever area-specific influences on school attendance are not contained in the availability of schools. The national capital region (NCR) is the reference area. The positive (negative) significant value would mean a higher (lower) proportion of children attending in that particular region compared to the NCR, on the average, after controlling for all the other variables.

The first stage results are given in Table 4 [ PDF 91.9KB | 5 pages ]. It shows the significance of the either both male, both female, and same sex as determinants of the number of children. Their usefulness as instruments is further validated by the significance of the partial R-square for the instruments, with F values of 14.8 for the both male and both female and 21.9 for the same sex instrument. It is worth noting that both male and both female have a slightly higher partial R-square of 0.0025 compared with same sex that has a partial Rsquare of 0.0018.

Estimation results of models that include the interaction of the number of children and per capita income quintiles are given in last three columns of Table 3 [ PDF 91.9KB | 5 pages ]. The interaction terms are all significant. The results highlight the regressive impact of the number of children on school attendance. For the poorest quintile, the impact of each additional child is a -18% reduction in the proportion of children 6-24 that are attending school, which is higher that the average impact mentioned earlier. The estimates for the other income quintiles are -11.8% (-17.8+6.0), -12.0% (-17.8+5.8), -12.1%(-17.8+5.7), -12.4% (-17.8+5.4) for the second to the fifth income quintile, respectively.

Finally, estimates for different age groups approximating the different grade levels, namely, elementary (6-12), secondary (13-16) and tertiary (17-24) are also done. The estimates for the 6-12 age groups show that the impact of the number of children is not significant, either on the average or across socioeconomic classes (Table 5 [ PDF 91.9KB | 5 pages ]). For the secondary and tertiary education age groups, however, the number of children has significant negative affects on school attendance. The results for the other variables are similar to the results for the total 6-24 age group so no further explanation will be provided. Again this GMM estimate has to be appreciated given the indication of correlation between the instrument and the error term, as indicated by the significance of the over-identification statistic.

Table 6 [ PDF 91.9KB | 5 pages ], summarizes the impacts and computes these as percentage changes relative to the current recorded proportion of children that are attending school. The table clearly shows the regressiveness of the impact of the number of children on school attendance. It is noteworthy that the regressiveness of the impact rises as one goes up the age groupings corresponding to the different levels of the education ladder. The poorest income quintile always has a higher negative impact compared with the other socioeconomic groups. For instance, for the poorest quintile in the 6-24 age group, each additional child will decrease the proportion of children attending school by -24%, while for the richest quintile this is only -16%. For the tertiary age group, the impact of the poorest quintile is -77%, while for the richest quintile this is only -22%.

Download this Discussion Paper [ PDF 152.9KB| 19 pages ].

Comment(s)

There are [1] comment(s) for this entry. Post a comment.

1. grace quimpan
   (posted 26 August 2005 / 02:10:47 PM)
   
   Yes your right! but how will they will go to school if their parents has no job to afford their school finances. Some parents are working but the salary they get is not enough even for their food.