Although computers have long been studied in terms of their changing price/performance ratio, the issue of accounting for performance in computer systems has not been adequately addressed. This paper addresses the topic in three ways. First, a survey of IS Managers and business "power-users" of personal computers was conducted to empirically determine the attributes of computer systems that provide value to users; these results guide subsequent choices regarding the operationalisation of user value. Second, an index of system performance was developed from published performance benchmarks and used as a direct measure of performance in the hedonic function. Third, a set of technical proxies was shown to adequately reproduce the performance index derived above, and was used in an alternate specification of the hedonic function. Using data on IBM-PC compatible laptop and desktop systems, price indexes were constructed using both approaches to performance measurement. The results demonstrated that both approaches yielded good explanatory power and nearly identical estimates of the rate of quality adjusted price change in PC systems. Thus, the set of technical proxies could be used to operationalise performance in a larger data set for which direct performance measures are unavailable. For the 1990s, laptop PCs were found to have decreased in quality adjusted price at an average of 39% per year while the corresponding figure for desktop PCs was 35% per year.

Although computers have long been studied in terms of their changing price/performance ratio, the issue of accounting for performance in computer systems has not been adequately addressed. This paper addresses the topic in three ways. First, a survey of IS Managers and business "power-users" of personal computers was conducted to empirically determine the attributes of computer systems that provide value to users; these results guide subsequent choices regarding the operationalisation of user value. Second, an index of system performance was developed from published performance benchmarks and used as a direct measure of performance in the hedonic function. Third, a set of technical proxies was shown to adequately reproduce the performance index derived above, and was used in an alternate specification of the hedonic function. Using data on IBM-PC compatible laptop and desktop systems, price indexes were constructed using both approaches to performance measurement. The results demonstrated that both approaches yielded good explanatory power and nearly identical estimates of the rate of quality adjusted price change in PC systems. Thus, the set of technical proxies could be used to operationalise performance in a larger data set for which direct performance measures are unavailable. For the 1990s, laptop PCs were found to have decreased in quality adjusted price at an average of 39% per year while the corresponding figure for desktop PCs was 35% per year. / Business, Sauder School of / Graduate