in the tertiary education expenditures model. The only exception is the Budget model for
primary education expenditures, where the interaction eect is close to 0 (but still positive).
On the other hand, the Budget model performs well for tertiary education expenditures.
Overall, we consider this as a conrmation of the REH hypothesis.
The AEH is also largely conrmed. The coecients in the row entitled with Glob. at
Dev=1 are always positive for primary education expenditures and always negative for ter-
tiary education expenditures. For tertiary education expenditures, however, the coecient is
insignicant in the Budget model, but still negative. Note also that the overidentication test
is rejected in the Outlier model. This is probably due to the large reduction in the sample
size.
Overall, both sets of robustness checks largely conrm the baseline results and thus rearm
the validity of the REH and AEH hypotheses.
4 Conclusion
In this paper, we have analyzed whether globalization is related to the composition of public
education expenditures. We rst developed a theoretical model which suggested that glob-
alization causes governments in developing countries to adjust public education expenditures
away from higher and towards lower education. For industrialized countries, on the other
hand, the model suggested an ambiguous eect.
Based on these theoretical results, we formulated two empirical hypotheses. The Relative
Education Hypothesis stated that developing countries will increase their spending for primary
education and reduce their spending for tertiary education at least as much as industrialized
countries. The Absolute Education Hypothesis, on the other hand, stated that developing
countries will increase their spending for primary and reduce that for tertiary education with
deepening globalization. In the empirical part of our paper, we found robust evidence, i. e.,
in the baseline models and in sensitivity analyses, for both hypotheses.
Our results thus show that globalization aects the structure of public education expendi-
tures. It apparently shifts the composition of education expenditures towards lower education
in developing countries, which might, in the long run, benet the poor by increasing their
productivity. We therefore conclude that governments in developing countries react to glob-
alization by taking steps that mitigate, to some extent, its potentially harmful distributional
consequences.
In industrialized countries, on the other hand, globalization seems to have been largely
neutral for the composition of public education expenditures. Since the distributional conse-
quences of globalization for low-skilled individuals in industrialized countries are apparently
not alleviated by shifts in the composition of public education expenditures, additional mea-
sures might be necessary (Sinn, 2005).
19