A SIMULTANEOUS EQUATION MODEL OF COST PERFORMANCE RELATIONSHIPS IN LARGE-SCALE COMPUTING

Cost/performance studies of central processing units have important implications for the organization of computing services. Recent studies have seriously challenged conclusions that economies of scale exist in CPU's and have rejected earlier findings that their average costs decrease with the square root of their power. The authors have argued that previous research is seriously flawed by its use of single equation models, and that performance is not simply an exogenous variable(s). A system of simultaneous equations is employed to measure the causal relationships between hardware characteristics, technology, cost, and their interactions. Parameter estimates suggest that economies of scale exist for large computer systems. The simultaneous equation model is used to define the cost-performance equilibrium and, for various hardware characteristics, a "design path" for optimal configuration.