The heterogeneity of agglomeration effect: Evidence from Chinese cities

Using annual surveys of industrial firms in China from 1998 to 2007, this paper applies the non‐linear least squares (NLS) method based on a grid search to analyse the effect of city size on firm total factor productivity (TFP). The results show that overall, the agglomeration effects in large cities, but not selection effects, significantly promote improvement in firm TFP. The optimal agglomeration scales of different industries differ as follows: those of capital‐ and technology‐intensive industries are larger than those of labour‐intensive industries. The agglomeration effects are also robust to different spatial areas without considering administrative boundaries. An inverted U‐shaped relationship exists between firm size and agglomeration effects, while the relationship between firm age and agglomeration effects is U‐shaped. State‐owned firms experience weaker agglomeration effects than non‐state‐owned firms. Cities higher in the administrative hierarchy and those with special economic zones have stronger agglomeration effects. However, cities higher in the administrative hierarchy and those with a larger economic development zone index can provide more resources to increase the survival rate of low‐productivity firms; thus, selection effects are not significant in Chinese cities.