The two faces of urbanisation and productivity: Enhance or inhibit? New evidence from Chinese firm‐level data

Abstract

This study employs industrial survey data from China's National Bureau of Statistics covering Chinese manufacturing enterprises during the period 1998–2007 to examine the impact of urbanisation on total factor productivity (TFP) in various industries. In recent years, urbanisation development has varied greatly across China. So Chinese cities can be divided into three urbanisation categories based on the proportion of the urban population to the total population: highly urbanised areas (60 per cent and over), moderately urbanised areas (30–60 per cent), and low urbanised areas (0–30 per cent). We estimate industrial TFP levels across these three categories using the Levinsohn–Petrin semi‐parametric estimation method. We also divide regional productivity into a productivity index and an industry composition index. We use aspects of these indexes to analyse the impact of urbanisation on TFP. The results confirm that urbanisation can lead to the gathering of economic activities, which in turn generates a positive impact on TFP by reducing transportation cost, promoting new technology spillovers, and encouraging a higher degree of specialisation. Further, the empirical results indicate that the highest TFP does not always occur in highly urbanised areas—most of the industries with the highest TFP are in moderately urbanised areas. These findings have important policy implications regarding how to improve the TFP of enterprises in order to generate scale effects.

Independent versus Spillover Innovation: A Cross-Country Inspection of Contingent Learning-by-Exporting

“…1 We will then refer to the two types of productivity as OP (Olley-Pakes) and LP (Levinsohn-Petrin) productivity respectively, depending on the estimation method. Both Olley-Pakes and Levinsohn-Petrin methods have long been adopted, even recently, to measure microlevel productivity (Howell, 2016;Kuosmanen and Kuosmanen, 2021;Wang et al, 2022;Wu et al, 2022;Huang and Du, 2023) and confirmed with their validity.…”