Rural household expenditure patterns in China, 1981-1992: An empirical study

This dissertation presents an econometrically-oriented empirical study of the rural household expenditure patterns in China. Using the annual survey data from 1981 to 1992 on per capita consumption expenditure of Chinese rural households, the study examines the empirical behavior of Chinese farmers on four major consumption groups--food, clothing, housing, and miscellaneous--in the framework of Engel curve analysis. For each of four consumption groups, the Engel curve in each individual year during the time period of 1981 to 1992 was estimated, which provided a quantitative description on the observed changes in expenditure patterns from year to year over the past twelve years. The estimates of the elasticities with respect to total expenditure for neither food consumption nor clothing consumption showed an evident tendency to increase or decrease in terms of the entire time period because the shifts of the annual Engel curves for these two groups seemed irregular. By contrast, the estimates of the elasticities with respect to total expenditure for both housing consumption and miscellaneous consumption displayed some roughly identifiable tendencies of change; i.e., the housing estimate seemed to have a tendency to fall and the miscellaneous estimate seemed to have a tendency to rise--especially after 1986. Further investigations demonstrated that, for any consumption group under study, the observed differences in elasticities derived from the different annual data were statistically insignificant. As a result, it was suggested that there was no significant change in consumption expenditure patterns of Chinese farmers from 1981 to 1992. The properly pooled estimations showed that, during the time periods under study, the elasticity with respect to total expenditure at 0.787 (or 0.792) for food, 0.599 for clothing, 1.431 for housing, and 1.411 for miscellaneous. A remarkable feature of this dissertation in terms of methodology is that it represents the first application of the bootstrap methodology for testing the equality of the models\u27 parameters when the F statistic is used. The study presents two methods for creating bootstrap resamples, discusses the advantage and disadvantage each method processed, and shows the empirical results obtained by using each method