explain why parents increase child expenditure with economic development despite decreasing
effectivity of child expenditure on child survival. Thus, the variable h can be thought of human
capital in a broad sense, consisting mainly of health and nutrition when income is low and of
schooling expenditure at high incomes.
Table 1: Independent Variable: Child Survival Rate
Variable |
Const. |
Income |
Latitude |
Health |
(Latitude × Health) |
Coefficient |
0.456 |
0.118 |
0.00206 |
0.00665 |
-0.000312 |
t-value |
13.69 |
13.04 |
3.76 |
2.31 |
-3.51 |
R2 |
0.73 | ||||
Variable |
Const. |
Income |
Latitude |
Health |
(Income × Health) |
Coefficient |
0.252 |
0.183 |
0.00038 |
0.0473 |
-0.0127 |
t-value |
3.77 |
9.90 |
1.66 |
4.18 |
-4.36 |
R2 |
0.74 |
Child survival rate: 1- under five mortality rate, income: GDP per capita in current
international dollars, health: total health expenditure as fraction of GDP, all taken
from World Bank (2004). Latitude: average absolute latitude taken from Masters and
McMillan (2001). Data for 137 countries for the year 2000.
The proposed model therefore suggests that child survival depends positively on income per
capita and latitude (summarizing fundamental factors), positively on the fraction of income
spent on health (the controllable factor), and negatively on an interaction term between health
expenditure and latitude indicating that health expenditure is less effective in favorable geo-
graphic environments. A cross-country regression performed with data from World Bank (2004)
and Masters and McMillan (2001) cannot reject our hypothesis. As shown in Table 1 all coeffi-
cients assume the predicted sign on a significance level of 95 % or higher. The table also shows
an alternative specification testing whether health expenditure is less effective at high income
levels. Again, all coefficient assume the predicted signs although latitude is now significant only
at the 90% level.
Summarizing, child survival is given by
π = π + (1 — π)λh , (1)
where the parameter λ > 0 measures effectivity of child expenditure. Note that the marginal
effect of child expenditure on survival, (1 — π)λ, is large when fundamental survival probability,
π, is low. Fundamental child survival depends on the state of development proxied by income