EDUCATIONAL OUTCOMES IN OECD COUNTRIES
19
the United States is excluded, the coefficient on tertiary education is much small and
insignificant again. The United States is well-known for its extensive tertiary education
system, and Figure 2 already indicated that the United States has the strongest positive
residual in the growth model. While this might be an indication of growth-enhancing
effects of its high-quality higher-education system, the lack of robustness in the sample
without the United States suggests that it might rather be an indication of the high-skilled
immigrant population that it attracts, of a particular set of economic institutions (not
captured by our institutional measures), or of any other idiosyncrasy of the US
economy.15
An additional interesting pattern emerges when employing the measures of different
skill dimensions. When the basic-skill share is used as an alternative skill measure to the
average skills, years of tertiary schooling reach significance (column (5)). When, by
contrast, the top-skill share is used, the coefficient on years of tertiary schooling
becomes smaller and loses any significance (column (6)). This pattern is indicative that
years of tertiary education proxy for the share of students with high-level skills. When
the United States is disregarded, the coefficient on years of tertiary schooling declines to
close to zero in this specification (column (7)).
While the possibility to distinguish the effects of different dimensions of the skill and
schooling distribution again is limited in these small samples, some basic patterns prove
clear. First, the significant effect of cognitive skills is extremely robust to consideration
of any quantitative measure of different levels of school attainment. Second, the finding
of Vandenbussche, Aghion, and Meghir (2006) of a particular effect of tertiary
attainment in rich countries is not robust once the focus is on long-run growth
experiences and educational outcome measures are taken into account. Of course, this
does not mean that learning beyond the secondary level does not matter. Rather, in the
spirit of a lifecycle interpretation where early skills facilitate the development of
subsequent skills (Heckman (2006); Cunha, Heckman, Lochner, and Masterov (2006)), it
means that outcome measures of learning in school are a good predictor for the
accumulation of further skills in life and the capacity to deploy these skills effectively.
5. CALCULATING THE ECONOMIC VALUE OF EDUCATION REFORMS
The results so far indicate that educational outcomes have a strong effect on long-run
economic growth of OECD countries. However, they do not tell us directly how much
improvements in educational outcomes would actually be worth. In particular, the
growth-rate effects do not map linearly into the economic value of any education reform
in a country, not least because different time lags are involved between successful
reform in the education system today and the improvement of skills in the national
workforce.
In this section, we therefore perform simulation analyses that use the estimates from
the previous sections to project what the results mean for the economic impact of
15
See Aghion et al. (2010) for additional analyses of tertiary education in Europe and the United States.