Statistical Analysis of Income Distribution in the Jiangsu Province of China

Abstract

This article analyzes changes in the distribution of household income in rural areas of China between 1980 and 1986. The period is interesting because important organizational changes were introduced in Chinese agriculture in 1978 and extended gradually during the 1980s. The results show that the average household income per head increased rapidly and so did income inequality. Also, the shape of income distribution changed markedly and by 1986 assumed the almost perfect log-normal shape observed in other developing countries. The degree of income inequality increased considerably and by 1986 was similar to that observed in India, Iraq and Kenya. Methods of estimating the Lorenz curve and the associated Gini coefficient, when income distribution is not log-normal, are also considered. This paper presents information about the distribution of household income in rural areas of the Jiangsu Province of China, analyzes the data and compares the results with similar studies using Indian, Iraqi and Kenyan data (Rajaraman, 1975; Kmietowicz, 1984; Kmi- etowicz & Webley, 1975). The Chinese results come from household budget surveys conducted in 1980, 1983 and 1986. The period covered is interesting as it corresponds to the gradual introduction of reforms in farm management by the government in 1978. The analysis of the data shows that incomes increased substantially during the period and so did their inequality. The shape of income distribution also altered significantly. These changes occurred gradually during the period and by 1986 the degree of income inequality and the shape of the distribution were similar to those observed in India, Iraq and Kenya. The income distributions for the three years and results of their basic statistical analysis, providing measures of central tendency, dispersion and skewness, are presented in Section 2. The changes in the degree of income inequality are analyzed in Section 3. Lorenz curves are obtained for the three distributions and the trapezoidal method is used to estimate the Gini coefficients which measure the degree of income inequality. The results for the Jiangsu Province are compared with other estimates for the Province and China as a whole. They are also compared with results for other developing countries. In Section 4, the log-normal distribution, which was found to describe well the distribution of household income in rural areas of India, Iraq and Kenya, was fitted to the three distributions. The results show that the shape of the distribution changed substantially during the period and by 1986 had an almost perfect log-normal form. The degree of income inequality as measured by the Gini coefficient also increased during the period and by 1986 was similar to that found in India, Kenya and Iraq in the 1960s and 1970s. Section 5 deals with the estimation of the Gini coefficient when the underlying distribution of income is not log-normal and the actual distribution is unknown. The procedure involves statistical estimation of parameters of several functions used to describe the Lorenz curve. The functions are so chosen that the Gini coefficient can be deduced from the estimated parameters. The functions employed are those used recently by Rao and Tam (1987). The results are rather inconclusive, showing that accuracy of the estimates depends greatly on the function used. The Kakwani-Podder model gives the best results, but the estimates of the Gini coefficient based on the