Technical progress and the evolution of prices of semiconductor manufacturing equipment since the 1970s

This is an empirical study on the relationship between the price and quality of the products of the fast-changing hi-tech industries. The objective of this dissertation is to estimate the hedonic (quality-adjusted) price index for semiconductor manufacturing equipment.;Four contributions are made in this dissertation. First, a database consisting of the prices and characteristics of different models of semiconductor manufacturing equipment is assembled. Second, hedonic price functions and hedonic price indexes are estimated for semiconductor manufacturing equipment. Third, computational methods for the construction of hedonic price indexes are reviewed and some misconceptions about these methods are corrected. Finally, a model for the estimation of the hedonic price function and indexes in a non-perfectly competitive market is developed.;As a part of the efforts of data collection, we studied the semiconductor manufacturing technologies, analyzed the market shares of the different types of equipment, and identified representative equipment and their characteristics.;The functional form for the hedonic price function was chosen from three flexible functional forms on the basis of hypothesis tests. A Chow test was conducted to determine the sample for the regressions. The estimated hedonic price index for semiconductor manufacturing equipment shows an annual rate of increase of 1.0 percent. Pure price increase contributes only 8 percent of the increase in the nominal price of semiconductor manufacturing equipment from 1978 to 1994. Quality improvements, which are the results of technical progress in the semiconductor manufacturing equipment industry, contribute the other 92 percent.;Among the computational methods used for the construction of hedonic price indexes, the dummy variable method has been considered to be simple but inaccurate because it is assumed that the effects of market weights and differences in characteristics cannot be incorporated. We show that properly formulated, the dummy variable method yields the same formula as the imputation method. Under the assumption that the hedonic price function is separable with respect to time, market weights and differences in characteristics have no effect on the hedonic price index and the simple dummy variable method yields the same answer as more sophisticated methods.