Market Shares and Price-Quality Relationships: An Econometric Investigation of the U. S. Automobile Market

Abstract

The objective of this paper is to investigate the role of marketing competition in the NorthAmerican automobile market. This is accomplished by estimating a market-share demand function. The main difficulty with the analysis of demand for automobiles is heterogeneity of model types. Quality differences among different models, both at a point in time and through time cannot be ignored, since price and sales depend on them. One way to remedy this situation is to apply the hedonic method by regressing the price of a model on its characteristics. In this way, the implicit prices for each of the qualitative attributes are derived and then used, free of quality-differences, in the second stage estimation of a market-share demand function. There has been considerable interest in estimating demand functions for automobiles. However, most of the previous research has focused on total automobile demand. In this approach, equations on total sales are estimated without taking into account differences in quality characteristics of different model types. These studies usually relate the depreciation of the service level of the currently held stock of automobiles to the demand for new stock. This leads to a very general stock adjustment [7] or generalized acceleration framework [10]. The best known time-series studies of the aggregate automobile demand using different variants of this approach are those of Chow [2], Suits [15], Wykoff [19], and Westin [18]. The estimates of price elasticities from these studies are about - 1, and income elasticities range from 1.5 to 4. Previous research on automobile demand allowing for qualitative differences in characteristics among different models is much less extensive. The most important research effort in this direction is the paper by Cowling and Cubbin [4]. They explained market-share movements for different car manufacturers by changes in quality adjusted prices using data from England for the period 1957-1968. Their quality adjusted price vector is proxied by the residuals from hedonic price regressions, using the data on manufacturer's models. The idea that the residuals from the regression of prices of differentiated goods on their qualitative attributes are the quality adjusted prices, was suggested by Griliches [8]. Cowling and Cubbin used, as their quality-adjusted price for a specific manufacturing firm, the mean residual computed by taking into account all the models offered by