The relative contributions of location and preferential policies in China's regional development: being in the right place and having the right incentives

Abstract

A large part of the literature on provincial inequality in China has found it necessary to include regional dummies in the provincial growth regressions. A smaller but vocal part of the literature has emphasised the granting of preferential policies to explain the faster growth of the coastal provinces. We replace the regional dummies with a measure of the ability to participate in international trade (Geography), and a preferential policy index (Policy). We find that geography and policy had about equal influence on coastal growth (3 percentage points each). Geography affected growth with a much longer lag than policy, however. The policy index was highest for the metropolises (Beijing, Shanghai, and Tianjin) and lowest for the central and northwestern provinces. The preferential policies are to a large extent “deregulation policies” that enabled marketization and internationalization of the coastal economies and allowed them to become more like their East Asian neighbours (and competitors). The weak (statistically insignificant) support for conditional convergence is in line with the existence of institutions that retard the income convergence process generated by the movement of labor and capital and by the Stolper–Samuelson mechanism. The household registration system ties the peasants to the land, the monopoly state bank system favors borrowing by state enterprises, and local protectionism reduces inter-provincial trade. Clearly, these institutions need to be deregulated. An effective strategy to develop the western provinces must therefore encompass physical capital formation, human capital formation, and institutional capital formation.

Introduction

Regional disparity has increased quite markedly in the 1990s. Fig. 1 shows the coefficients of variation of per capita provincial incomes constructed from two samples.¹ The first sample consisted of 28 provinces that had complete income data for the 1952–1998 period, and the second sample differs from the first by omitting Beijing, Shanghai, and Tianjin, the three metropolises that have province-level status. The coefficients of variation of GDP per capita (measured in 1995 prices) from these two samples are denoted Cov28 and Cov25, respectively. Cov25 differs from Cov28 by having a smaller dispersion in regional incomes and not showing an upward trend during the 1966–1978 period. These two differences mean that the three metropolises have always been substantially richer than the other provinces, and that the gap between these two groups widened substantially during the period of orthodox socialist economic management. This paper will focus on the common finding in Cov25 and Cov28 that there is a clear upward trend in provincial income inequality from 1992 onward, and that the 1998 level of provincial income disparity is highest since 1952.

This recent rise in regional inequality has elicited significant policy responses from the government. The budgets for infrastructure investments in the poor provinces have increased substantially every year, and a Western Region Development Office has been established under the State Council (the Chinese cabinet) to formulate a comprehensive development strategy and to coordinate its implementation.

The origin and consequences of China's regional disparity growing has been extensively studied in recent years.² One prominent view is that preferential policy treatment of the coastal provinces, especially the establishment of Special Economic Zones (SEZs), was largely to blame for this sustained rise in regional disparity. According to Hu Angang, an advocate for the abolition of SEZs: “If Deng Xiaoping knew the disparities were as big as they are, he would be more militant than I am in trying to eliminate them… In America, the deep differences between the North and the South more than 100 years ago led to the Civil War,” and “We must cease subsidizing rich coastal cities. Preferential treatment should be reserved for the poor.”³ Hu Angang's diagnosis is consistent with the common finding in provincial growth regressions that the coastal variable has a positive coefficient that is statistically significant, e.g., Jian, Sachs, and Warner (1996), Chen and Fleisher (1996), Zhang (2001), and Bao, Chang, Sachs, and Woo (forthcoming).

Our thesis here is that the high coastal growth has been due to more than the preferential policies; it came also from advantageous location that enabled export-oriented industrialization; we propose to quantify the relative contributions of geography and preferential policy.⁴ The other major innovations in this paper are to construct a preferential policy index, explore the links between topographical features and income level, show that the geography and policy variables affect income growth with different time lags, and to quantify the impact of geography and policy variables on provincial growth rates in the 1996–1999 period.