the largest plants in these industries generally appear to have reached a minimum efficient scale.
Thus, given that large plants account for a disproportionate share of sectoral output, industrial
expansion does not induce large gains in scale efficiency. More precisely, the mean output growth
was more than 50% in the period 1984-1990. But despite this substantial output growth, the
exploitation of scale economies accounts for only the 0.55% rise in average productivity. These
results suggest that the focus of trade models with imperfect competition and representative firms
on the gains from scale economies exploitation may be somewhat misplaced, since these gains
appear modest in magnitude.
b) Output share reallocations.Thiseffect accounts for more than 1% rise in average productivity.
Although this figure is quite modest, it suggests that output share reallocations among firms
with different productivity levels might be empirically more relevant than scale efficiency effects,
and hence that firm heterogeneity can be a key determinant of the efficiency gains from trade
liberalization.
c) Residual effect. Most of the average increase in productivity comes from the catch-all residual
effect, which accounts for 9.6% rise in average productivity. This implies that most of the estimated
overall efficiency gains (11.16%) are indeed left unexplained.
The above results must be interpreted with caution, because of the hurdles involved in the
methodology used by Tybout and Westbrook, which are common, however, to most analyses
of the efficiency gains from trade liberalization. The main methodological problems involved in
estimating the productivity gains from trade reform can be summarized as follows (see Pavcnik,
2002).
1) Identification of the trade effects. In most studies, the identification of trade effects relies on
the comparison of plant productivity before and after a trade policy change. As a consequence,
this approach attributes productivity changes originating from other sources to trade policy. To
see how serious this problem can be, note that most studies use data covering only a short time
period after trade reform, which implies that the estimates of productivity growth can be heavily
15