Page 40 - IJPS-3-2

Desta CG

Table 4. Proportion of variance explained for maternal work by the number of children and other covariates

Full sample Urban sub-sample Rural sub-sample
Model Wald χ 2 2 2 2 2 Wald χ 2 2
2
2
(LR χ for OLS) Prob > χ Wald χ (LR χ for OLS) Prob > χ (LR χ for OLS) Prob > χ
Exogenous 28.32 0.005 23.56 0.018 22.60 0.030
probit (%)
Ivprobit (%) 24.51 0.022 16.08 0.237 14.35 0.130
N 493 248 245
Source: Survey data (2010 and 2013).
Table 5 shows parameter estimates for the ivprobit model (and exogenous probit model). It is worth noting, at this
juncture, that this study does not intend to discuss coefficients from control variables (see Appendix B for coefficients
from control covariates).
The table consists of three panels. Each panel compares results for the rural and urban locations. The first panel shows
results for all households that differ only in their rural-urban location. The second and the third panels show results for
households that differ by the age group of their children, in addition.
For the first panel, first, the ivprobit estimate has all positive coefficients for both the rural and urban locations,
suggesting that an increase in the number of children is associated with an increase in the probability of the mother’s
work participation for the households, despite the lack of statistical significance for the ivprobit coefficients. While the
lack of statistical significance for the ivprobit coefficients as opposed to those using the exogenous model is consistent
also with several other previous research, the lack of difference in coefficient signs by rural-urban location is surprising
because the difference in the employment structure between the rural and the urban economies is expected to respond to
the effect of the number of children differently for the rural and the urban locations. However, this difference becomes
fairly apparent once the lifecycle effect is considered by categorizing households according to age groups of their
children (the last two panels of Table 5).
Table 5. Parameter estimates for maternal work participation by the number of children (with control variables)

Full sample Urban sub-sample Rural sub-sample
Group of households Model
Coef. p > z Coef. p > z Coef. p > z
0.0918 -0.2156 0.1568
Exogenous probit 0.070 0.061 0.004
(0.0321) (0.0452) (0.0425)
All households Ivprobit 0.1671 0.418 0.0304 0.8412 0.113
(0.2031) (0.1549) 0.315 (0.1456)
N 493 248 245
0.1903 0.1843 0.1497
Exogenous probit (0.1102) 0.021 (0.3201) 0.321 (0.1222) 0.107
Households with children 0.3349 -0.0989 0.498 0.7008
of ages < 10 years Ivprobit 0.420 0.092
(0.9742) (1.7025) (0.5079)
N 217 99 118
0.9963 -0.0845 0.2111
Exogenous probit (0.1515) 0.088 (0.0852) 0.541 (0.0981) 0.026
Households with children 0.0711 0.1252 -0.9932
of ages ≥ 10 years Ivprobit 0.476 0.084 0.566
(0.1845) (0.3602) (6.961)
N 276 149 127

Note: Standard errors are reported in parenthesis. Source: Survey data (2010 and 2013).

The second panel of the table shows results for mothers with children of ages less than ten years. The third panel shows
results for those mothers with children of ages ten years or older. For the urban sub-sample, the ivprobit coefficient is
negative for the second panel, suggesting, as expected, that large number of young children decreases the probability of the
mother’s work participation, although it is not statistically significant. In the third panel, the ivprobit coefficient is positive
and statistically significant at p = 0.084, suggesting that for urban mothers with more adult children, the negative effect of
the number of children disappears and contributes positively. For the rural sub-sample, the ivprobit coefficient is positive
and statistically significant for the second panel at p=0.092, suggesting that large number of young children increase the
mother’s probability of work participation. By contrast, for the third panel the ivprobit coefficient is negative, suggesting a
34 International Journal of Population Studies | 2017, Volume 3, Issue 2