| In recent years, structural analysis consisting of specific demand, cost and strategic hypotheses, has emerged in the mainstream for merger analysis of markets with product differentiation. The research can be broadly categorized into two methodologies. The first major line of research estimates the demand functions and the derived elasticities to predict the post-merger price levels and welfare effects (for examples, Baker and Bresnahan [2], Hausman, Leonard and Zona [32], Shapiro [53], and Nevo [42]). The second methodology proposed by Werden [65] estimates the necessary marginal cost reduction to offset the merger effects. The first theme of our dissertation demonstrates that under the assumption of non-cooperative, oligopolistic price competition, both approaches are consistent based on results showing the extra market power generated by mergers.; The estimation of substitution effects is a necessary and important step in empirical analysis for a differentiated market. Willig [66] first introduced a simple two-good formula to calculate the substitution effect. The second theme of our dissertation explores the estimation and shows that our measure is adequate for many applications and is simple and feasible. Furthermore, we develop the measure for multi-product application.; In recent merger settlements, the Federal Trade Commission (FTC) and Department of Justice (DOJ) have employed divestiture orders. Yet the effectiveness of a divestiture order is hard to evaluate. The third theme of our dissertation formulates a way to evaluate the effectiveness of divestitures from a game-theoretic point of view.; The fourth theme of the dissertation develops a formulation attempting to identify potential collusion facilitated by repeated game. We illustrate that (1) substitution effects among differentiated products influence collusive behaviors between firms and (2) in a diversified market the coexistence of both high substitution ratios (substitution effects) and high price-cost margins over a period of time likely result from collusive interactions rather than one-period Bertrand outcomes. This approach is then applied to the case study of the U.S. long-distance telecommunication market during 1987–1994.; The fifth theme of the dissertation pursues structural analysis in identifying possible collusion for the long-distance market, following an approach introduced by Gasmi, Laffont and Vuong [23]. We propose its various non-collusive and collusive behaviors in detail. All these models have been estimated by the Generalized Method of Moments (GMM). (Abstract shortened by UMI.)... |
