Demand Analysis and Tax Reform in Pakistan

Pakistan, like many LDCs, derives most of government revenue from indirect taxation. However, the system of taxes and subsidies has grown up piecemeal over the years. Previous exercises in price reform for Pakistan have been forced to make very restrictive assumptions about consumer preferences, and have typically used demand systems that prejudge what are the desirable directions of price reform. In this paper, the methodology of Deaton (1988, 1991) is extended and applied to the 1984-85 Household Income and Expenditure Survey. A theory of quality variation based on separable preferences is developed, and the implications for welfare and empirical analysis laid out. The prices of oils and fats and of sugar do not vary very much in the survey data, and the symmetry and homogeneity restrictions from the theory play an important part in obtaining sharp estimates of own and cross-price elasticities. The parameter estimates suggest that there are significant cross-price elasticities between the high-calorie foods and the presence of these substitution patterns means that the effects of potential price reforms are quite different from those that would be estimated using the traditional assumptions. Based on demand patterns alone, it would be desirable to raise government revenue by raising the consumer price of rice. However, in Pakistan it is not generally possible to decouple the producer and consumer prices of rice.