EDUCATIONAL OUTCOMES IN OECD COUNTRIES
13
Conceptually, it is an open question whether any connection between human capital
and institutions stems from human capital causing better institutions or the opposite. But
at least within the group of OECD countries, that part of institutional variation that is not
related to cognitive skills is not related to long-run growth, whereas that part of skill
variation that is not related to institutions remains a strong predictor of long-run growth.
Column (3) adds the measure of openness to the model. Property-rights security and
openness to trade are individually and jointly insignificant in predicting long-run OECD
growth, whereas cognitive skills remain strongly significant. Again, the insignificance
in the institutional measures does not mean that institutions are unimportant for long-run
growth. Rather, they point to the fact that the OECD countries share broadly similar
institutions, so that this kind of institutional variation is unlikely to account for much of
the substantial variation in long-run growth in this rich-country sample. For example,
while there is some variation in the openness measure among OECD countries, it is very
specific and limited. Most OECD countries are coded as open throughout the period of
observation. The exceptions are Mexico, New Zealand, and Turkey that had substantial
periods of being more closed, but the differences between the openness of these and the
remainder of the OECD explain little of OECD growth differences.
This argument is underlined by results of the full-country sample (column (4)). The
institutional measures enter jointly significantly in explaining long-run growth
differences among the 50 countries with available data. However, as the specification
reported in column (5) indicates, there is a significant difference in the institutions-
growth nexus between the OECD countries and the non-OECD countries. The
interaction between an OECD indicator and protection against expropriation brings the
institutional effect close to zero in the OECD sample, and the difference is marginally
significant (at the 13 percent level). The OECD indicator and the interaction jointly
reach statistical significance at conventional levels.12
In sum, property-rights and free-trade institutions help us understand long-run growth
differences between rich and poor countries, but they do not contribute to our
understanding of long-run growth differences within the group of rich countries. By
contrast, the significant effect of cognitive skills on long-run growth in the OECD
sample is robust to the inclusion of the institutional measures.
3.2. Regulation of product and labor markets
While the variation in fundamental property-rights and free-trade institutions may be
limited among OECD countries, a substantial literature has stressed significant
differences in how OECD countries regulate their product and labor markets. For
example, Nicoletti and Scarpetta (2003) show that short-run growth experiences across
OECD countries are related to product market regulations, and Cingano, Leonardi,
Messina, and Pica (2010) find that employment protection legislation is associated with
12 An interaction between openness and the OECD indicator does not reach significance when added to this model.