Hypothesis 4:
The Effect of Education 
The argument has been made that widely distributed education makes for greater equality because it reduces the differential return to human capital and provides access for a larger proportion of the population to the higher incomes accruing to the educated. Adelman and Morris and Ahluwalia found the postulated relationship between education and equality 
The spread of education proves generally to be significantly related to equality in this study, as in other studies, as long as the regression includes only the Kuznets Curve and economic variables. In these regressions both tstatistics and Fstatistics are highly significant whether the dependent variable is the Gini coefficient or the share of the poorest 40 percent. However, once sociopolitical variables are taken into account, education is no longer statistically significant. This is the mirror image of the earlier discussion of the significance of dualism: in Eastern Europe education is widespread; in dualistic society it is concentrated on the foreign elite and its local clients. That is, sociopolitical and educational variables are correlated and when the former are added, the latter loses significance. 
The coefficient and significance of education drop further when regional dummies are introduced. But the reason why the regional variables take away significance may be that the regions differ in educational level. So education declines in statistical, but probably not in real, economic significance, when regions are introduced. 
In the more limited regressions, the coefficients for education are quite high. Comparing countries with 10 percent and 90 percent enrolment ratios — and only a few countries exceed those limits, since secondary school rates are included — the estimated Gini differs by about 0.11 and the share of the poorest 40 percent increases by an estimated 4.9 percent. Education alone explains much of the variance in income distribution in these regressions. But these coefficients probably overstate the effect of education, since they come from regressions with only a limited number of variables. Using the most complete regressions, excluding regional variables, the effect on the Gini is a small .056 and on the share a more impressive 2.3 percent. These results suggest that spread of education benefits particularly the poorest groups in the population. This is quite a reasonable conclusion. For most countries, enrolments in both primary and secondary schools had reached 30 percent during the period covered by this study, and the mean was 62 percent. So a large proportion of the middle income groups had been educated for some years, given the proportion of young people in the labor force. (The education data are for five years earlier than the income distribution data.) A further spread of education therefore means a spread to the poor in most countries and could, therefore, help them disproportionately. However, the maximum increase in enrolment which it is realistic for most countries to achieve is 20—40 percent — the difference between Guatemala and Costa Rica, or Pakistan and Sri Lanka. The feasible improvement in education is likely to increase the income share of even the poorest group by only about one percent. This would represent a 5percent increase in their typical share of income, a modest improvement 
.
.



