Agglomeration Premium and Trading Activity of Firms
Gábor Békés*, Péter Harasztosi
Last modified: 2012-08-11
Abstract
While most empirical studies in economic geography document a steady
and positive
correlation between regional density and firm productivity, the impact
is not homogenous across firms. Importantly, the recent international
trade literature showed that trading firms are different in terms of
workforce, size or productivity. We argue that externalities that
determine density premium for firms will be affected by firms'
involvement in trade. Indeed, firms active in international trade may
employ a different bundle of resources and be organized differently so
that they would appreciate inputs and information in a different
fashion and intensity. Using Hungarian manufacturing firm level data
from 1992-2003 at a 150 micro-region level, we show that the
elasticity of agglomeration on productivity is much larger for traders
then for non-traders. As firms' trade participation is endogenous to
firm performance, we offer various treatment methods of this
endogeneity issue. We find that our key result is robust and well
above the gap suggested by simple self-selection models.
and positive
correlation between regional density and firm productivity, the impact
is not homogenous across firms. Importantly, the recent international
trade literature showed that trading firms are different in terms of
workforce, size or productivity. We argue that externalities that
determine density premium for firms will be affected by firms'
involvement in trade. Indeed, firms active in international trade may
employ a different bundle of resources and be organized differently so
that they would appreciate inputs and information in a different
fashion and intensity. Using Hungarian manufacturing firm level data
from 1992-2003 at a 150 micro-region level, we show that the
elasticity of agglomeration on productivity is much larger for traders
then for non-traders. As firms' trade participation is endogenous to
firm performance, we offer various treatment methods of this
endogeneity issue. We find that our key result is robust and well
above the gap suggested by simple self-selection models.