An econometric analysis of the desktop computer hardware market

摘要

The purpose of this study was to compare six competing econometric models which depict the relationship between hardware characteristics and machine cost for the desktop computer market. The Box-Cox methodology and multiyear data were used to facilitate this comparison. The analysis validated that the Box-Cox methodology is a viable means for evaluating competing model formulations within the field of information systems research. The results were consistent with past research that suggested a double natural log model formulation for representing the functional relationships among variables when modeling machine cost as a function of hardware attributes. Further, the more complex power transformation model formulations suggested by the Box-Cox methodology did not significantly outperform the more traditional and simpler double natural log model. More specifically, the results indicated that variables related to primary memory and microchip tedinology have the largest impact on machine cost. Additionally, variables related to madiine connectivity, machine expandability, and year of observation were also found to be significant explanatory variables for machine cost.