Essays on the use of distance functions in empirical studies : efficiency measurement and beyond

Graduation date: 2005This dissertation explores the intricacies associated with the use of distance\ud functions in empirical studies. It focuses on the measurement of advertising\ud efficiency and investigates the properties of the models that seek to approximate\ud the abatement costs of socially undesirable outputs. The first manuscript is\ud devoted to the development of the algorithm that can be used to measure the\ud efficiency with which firms market their brands in the presence of advertising by\ud rivals. An empirical illustration is carried out using the data from the U.S. brewing\ud industry. The second study analyzes the difficulties associated with the accurate\ud approximation of the abatement costs of socially undesirable outputs. It contrasts\ud the results from a variety of different shadow-pricing models, each of which relies\ud on a different type of distance functions that are used to approximate the polluting\ud technology. The shadow prices of sulfur dioxide are computed using linear\ud programming techniques and the data from the U.S. electric utility industry. The\ud third manuscript shows how a generalized method of moments (GMM) algorithm\ud can be used to estimate the parameters of certain types of distance functions; the\ud empirical illustration is carried out using the data set from the second study.\ud The first manuscript illustrates that advertising spillovers are important in\ud brewing and shows that the estimates of marketing efficiency are inaccurate when\ud spillover effects are present and ignored. The second study shows that the shadow\ud price estimates of socially undesirable outputs are not invariant to the assumptions\ud regarding the parametric form of production technology and can in fact be\ud predetermined by selecting a specific model. Finally, the third study established\ud the legitimacy of the GMM procedure as a choice of an algorithm for the shadow\ud pricing of undesirable outputs