2 The model
2.1 Framework
Consider a country that is populated by two groups: individuals who are endowed with a low
level of ability аь (group L), and individuals who are endowed with a high level of ability
ɑ// > аь (group H). These different levels of ability may originate from differences in inborn
intelligence, or in the skills and opportunities that an individual “inherits” by being raised
within a particular social class.
With respect to the masses of the two population groups, we assume that M[L] > M[H].
There are thus more low- than high-ability individuals, which is probably a realistic assump-
tion for most countries. We also normalize M[H] = 1.
Each individual i ∈ I, I = {L, H} produces an output yι by combining her endowed
ability ɑʃ with a government-provided productivity-enhancing public good gɪ. We assume
that the production function of individual i is of the Cobb-Douglas type; it is given by
уг = ɑʃ g}~β, ∀i ∈ I = {L,H} with 0 < β < 1.
In our context, it is appropriate to think of gʃ as different types of education expenditures.
That is, gb could be thought of as spending for lower education. A high value for gb would
then signal the government’s willingness to increase the income of the low-ability individuals
by improving their educational opportunities. Conversely, g∣∣ can be thought of as higher
education expenditures, which increase the productivity of the high-ability individuals, but
are of no direct use to the low-ability individuals.
Even though expenditures for lower education might in reality also directly benet the high-
ability individuals, any such benets are likely to be smaller than those they would obtain
through higher education expenditures. The assumption that the high-ability individuals do
not benet at all from lower education and the low-ability individuals not at all from higher
education is a simple way to capture the fact that different societal groups prefer different
compositions of the public budget. Incorporating cross benets into the model would not
qualitatively affect the argument we want to make, but would add several variables to the
algebra.
It is apparent that industrialized countries have, inter alia, a relative abundance in high-
skilled labor, whereas developing countries are relatively abundant in low-skilled labor. This
implies that in a world of closed economies, wages for low-skilled labor should be relatively
high in industrialized and relatively low in developing countries, while wages for high-skilled
labor should be relatively low in industrialized and relatively high in developing countries.
One important effect of globalization is that rms in industrialized countries gain access
to the low-skilled labor pool in developing countries, for example by relocating their produc-
tion plants. This increases the demand for low-skilled workers in developing countries, while
decreasing it in industrialized countries. Therefore, the returns to low-skilled labor are likely