Globalization, Redistribution, and the Composition of Public Education Expenditures

indicated by the overidentication tests in table 2, the instruments perform well with respect
to this test in all models. The relevance tests, however, indicate that the set of instruments
may be weak in the model for secondary education expenditures when clustered standard
errors are used.

According to the results in table 2, the estimate for δ is significantly positive in the model
for primary, insignicant in the model for secondary, and signicantly negative in the model
for tertiary education expenditures when the interaction term is not included (the FE 1 mod-
els). These results seems to suggest that globalization is positively associated with primary,
unrelated with secondary, and negatively related with tertiary education expenditures.

The remaining models in table 2, however, reveal that these results are driven by the
eects of globalization in developing countries. The interaction eect is, as predicted by
the REH hypothesis, always positive for primary and always negative for tertiary education
expenditures. In fact, the interaction eect is even significantly dierent from 0 (this, however,
is not necessary to confirm the theory which only predicts that the interaction eects will be
weakly positive or negative, respectively). The signs of the interaction eects thus confirm
Hypothesis REH.

The second hypothesis, AEH, is also confirmed by our results. As the row entitled Glob.
at Dev=1 shows, the marginal effects of globalization in developing countries, δ + 7, is signif-
icantly positive in the model for primary and significantly negative in the model for tertiary
education expenditures. The effect of globalization in industrialized countries, δ (when the
developing country dummy is 0), on the other hand, is always insignificant. In other words,
globalization has apparently led to a relative increase in funding for primary and a reduction
in funding for tertiary education in developing countries, while displaying a “neutral” effect
in the industrialized world.

Note also the the marginal effect of globalization in the model for secondary education
expenditures has always, both for industrialized and developing countries, a value that is
in between the estimates for the marginal effects in the models for primary and tertiary
education expenditures.

With respect to the other control variables, we find that an increase in education expen-
ditures increases spending for all three educational programs since the coecient for this
variable is consistently positive. Apart from total education expenditures, only the popula-
tion and GDP per capita variable are sometimes significant. When countries become more
prosperous, they seem to spend more on primary and secondary education. When countries
become more populous, on the other hand, they seem to spend less on primary education.

at higher levels of education, a higher gross enrollment rate at some lower level is therefore not necessarily
negatively correlated with total education expenditures.

13

<<   12   13   14   15   16   17   18   >>

More intriguing information