income of country i to be higher under free trade than in autarky or, more generally,
the larger the share t of traded goods in the economy. Further, since the ability of
country i to attract innovation in sector j depends on its share in world production
of that sector, which in turn depends on country size, the model suggests that the
impact of θi on productivity should be higher in larger countries. More precisely, as-
suming that a single country is “small” compared to the world economy, but “large”
compared to the subset of countries specialized in the same range of goods, these
implications can be derived formally and summarized as:
where y is real GDP per worker and y is the world average. The Hrst inequality
follows directly form (24) and (25). To derive the second, note that what matters to
attract better technologies is the population-weighted average of the index of IPRs
protection in a given region R of similar countries, θ∕1> = (]At∈∕1>θiLi) ∕∑i∈R Li.
Since the main results of the paper hinge critically on these interactions, testing the
sign of the cross-partial derivatives in (26) provides a way to assess the empirical
plausibility of the model. Predictions on the overall effect of IPRs seem instead
less useful to evaluate the theory. Although the model implies that raising θ should
always have a positive effect on productivity, this result relies heavily on the simpli-
fying assumption that θ does not affect the monopoly distortion in the South.
d Ш > 0
∂θi∂t ^
and
d y > 0
∂θi∂Li
(26)
3 Empirical Analysis
To test the inequalities in (26), measures of labor productivity, IPRs protection,
openness to trade and size have been collected for a panel of countries from 1965
to 1995. Labor productivity is proxied by real GDP per worker (GPDW) from the
Penn World Table 6.0 (PWT6.0). Two important determinants of productivity are
also included in the analysis: the stock of physical capital per worker (KL), again
from PWT6.0, and the fraction of working age population with at least secondary
schooling as a proxy for human capital (HL), from Barro-Lee. As for trade openness,
two different measures are considered: the Sachs and Warner (1995) index, which
is a dummy taking value one if a country is classified as open, and the trade share
22