REGRESSION ANALYSIS VERSUS LINEAR PROGRAMMING IN THE ANALYSIS OF PRICE‐QUALITY RELATIONSHIPS: AN APPLICATION TO THE DETERMINATION OF MARKET SHARES

A linear programming approach to the measurement of quality is presented as an alternative to the traditional hedonic approach. The linear programming approach has 3 distinct advantages. As a boundary method, inefficient observations do not affect the estimation of the boundary, nor do they affect the benchmark by which the relative efficiency of particular observations is ranked. The approach does not impose an arbitrary functional form upon the data, and it is more flexible than the hedonic method in that it can handle multiple outputs and inputs. The 2 techniques are compared in an empirical analysis of disaggregated market share determination in the UK automobile market in 1982. The sample consists of 125 passenger cars, both domestic and foreign. While the 2 techniques provide parallel insights into the role of price and quality in marketing decisions, the linear programming approach leads to much better determined estimates of the quality-adjusted price elasticity and a slightly greater overall explanatory power.