EDUCATIONAL OUTCOMES IN OECD COUNTRIES
29
the workforce), the difference between the neoclassical and the endogenous growth
model has relatively minor political meaning. In either case, the value of the education
reform is very large, at 288 percent or 211 percent of current OECD GDP. Note that in
the neoclassical projections, the value of this reform as a share of current GDP differs
across countries, because the convergence term means that projected growth rates
depend on the level of GDP reached.
To illustrate the specific dynamics of the neoclassical projections, it is worth
discussing a few details of the trajectories of the projection. Initial growth rates (with or
without reform) in 2010-11 differ across countries, depending on how far away they are
from their steady state. The United States is projected to grow at 1.1 percent initially,
whereas the simple mean of OECD-country growth rates is higher at 1.9 percent due to
many countries’ room for catch-up.22 Due to the convergence process, average OECD-
country growth is down to 1.6 percent in 2090 without the reform, ranging from 1.0-1.8
percent across countries. With reform, the range is 1.2-2.0 percent, with an average of
1.7 percent. By 2104, the average is down to 1.5 percent, and by 2130, all countries’
growth rates have converged to between 1.3 and 1.6 percent without the reform and to
1.4-1.7 percent with reform.
In the neoclassical projections, the difference in average growth rates between the
scenarios with and without reform grows to a maximum of 0.28 percentage points in
about 2060 and then declines back to 0.18 percentage points in 2090 and 0.06 percentage
points in 2150. (Compare this to the endogenous growth model where the long-run
growth rate stays 0.47 percentage points higher starting in 2070.) By 2300, the
difference in growth rates would be below 0.004 percentage points in each country.
Thus, while the difference ultimately converges to zero, the model parameters imply that
this convergence process takes a very long time to take full effect.
In reform Scenario II, where each country improves to the test-score level of Finland,
the present value of the reform amounts to $180 trillion in the neoclassical model,
compared to the previous $275 trillion in the endogenous growth model - or more than
four times (rather than six times) current GDP. Note that this scenario implies that in the
very long run, each country converges to the same steady-state level of per-capita GDP,
as test scores are the only variable influencing the steady-state level in our model.
However, in 2090 the per-capita GDP of the most advanced country would still be 70.6
percent higher than that of the least advanced country. By 2150, this difference would
be down to 19.7 percent, and by 2300 to 1.2 percent.
The present value of reform Scenario III, which brings all students to a minimum level
of 400 PISA points, is $187 trillion in the neoclassical projections, rather than the $226
trillion of the endogenous-growth type projections.23 Again, the difference over our time
22 Based on our model which depicts only effects of test scores and reached levels of income, Luxembourg and Norway - the
two countries with the highest current levels of GDP per capita - are projected to converge to a lower balanced growth path
(both without and with reform) and thus initially have the lowest growth rates. If these countries can keep their current
advantage in per-capita relative to the other OECD countries in the future for reasons outside our model, this would increase
the projected value of the educational reform in these countries.
23 In the underlying growth regression with log per-capita GDP as control variable, the coefficient on the proportion of the
population with scores higher than 400 is 6.478 and the coefficient on log initial income is -2.107 (both significant at the 1
percent level).