| Abstract: |
This paper is in the spirit of Marshall (1920), who raised the question of how
economic distance affects a firm's productivity, focusing upon the role of
idea sharing in relation to technological knowledge or information between
firms. In order to quantify the degree of knowledge spillover or information
sharing, we take the production function approach. Assuming core-periphery
structure around automobile assemblies surrounded with auto-parts suppliers,
we estimate plant-level production functions of the Japanese auto-parts
suppliers, where productivity function depends upon the degree of information
sharing measured by both geographic plant location and membership of
technological cooperation associations. We take econometric issues of
cross-sectional dependence of productivity and a simultaneity problem between
inputs, applying methods to the standard OLS and GMM estimators. Positive
technological externalities are seen in general and for independent plants,
the fact which is robust to specifications of the production functions.
Agglomeration effects are however rarely observed for relation-specific or
cooperative plants. Some of them cost substantial negative externalities. Once
a simultaneity problem is econometrically considered, instead of increasing
returns, decreasing returns to scale emerge in cases of total materials.
Agglomeration, if any, could be brought about not by increasing returns to
scale, but by productivity spillover among suppliers proximate to automobile
assemblies. |