countries face conictive incentives, however, one reasonable hypothesis is that globalization
will be “neutral” for the composition of education expenditures in this group of countries.
3 Empirical results
3.1 Empirical model and data
In this section, we test the implications of our theoretical model by estimating several variants
of the following general empirical model
y = ai + ωt + β x + δ Globalization + 7 Globalization × Dev + e, (10)
where y is a measure for public expenditures for either primary, secondary, or tertiary educa-
tion, ai country xed eects, ωt year xed eects, x a vector of control variables (see below),
and e the error term.
In order to test the implications of our theory, we include a measure for globalization
and its interaction with a dummy variable that is 1 for “developing” countries9 . We are
interested in the estimate for δ and 7, the coefficients attached to the globalization variable
and its interaction with the developing country dummy: δ is the effect of globalization in
industrialized countries, whereas δ + 7 is its eect in developing countries. Note that we do
not have to explicitly include the developing country dummy into this model because of the
country xed effects.
The theoretical model leads two related but distinct empirical predictions. First, the fact
that the effect of globalization on the composition of education expenditures is ambiguous in
industrialized but unambiguous in developing countries implies that with deepening global-
ization, developing countries will increase their spending for primary education and reduce
their spending for tertiary education at least as much than industrialized countries. This
implies that the estimate for 7 should be weakly positive (i. e., 7 ≥ 0) for primary and weakly
negative (i. e., 7 ≤ 0) for tertiary education. We label this prediction the Relative Educa-
tion Hypothesis (REH) because it refers to the relative effect of globalization in developing
countries as compared to industrialized countries.
The theoretical model also leads two a second prediction which we label the Absolute Edu-
cation Hypothesis (AEH). The fact that equation 9 in the theoretical section is unambiguously
negative for developing countries suggests that these countries will increase their spending for
9 Any classication of countries as industrialized or developing is of course arbitrary. We take a conservative
approach and classify the wealthy “Western” countries cum Japan and South Korea as industrialized. This
implies that even OECD countries such as Turkey or Poland are classied as “developing”. Therefore, the
term developing as used in this paper should not be understood as being synonymous with, for example, the
Least Developed Countries (LDC). It should rather be understood as encompassing all countries except the
most wealthy. See table 7 in the appendix for a complete list of how we have classied the countries in our
sample.