Mobile Edition
Print
Get the Mag
Weekly UpdatesPerloff, Jeffrey M., Larry S. Karp, and Amos Golan, Editors. Estimating Market Power and Strategies. Cambridge, UK: Cambridge University Press, 2007, 352 pp., $88.99.
From my point of view a text on empirical approaches to market power has been sorely needed. Given this view, the effort by Perloff, Karp, and Golan is significant offering. However, the scope of the area raises the typical question faced by most books: Do we write about everything or focus our attention on a few topics. The uneasy compromise in this book is to survey the literature by providing an overview of most of the techniques discussed and more fully developing a handful of approaches. Such an approach may somewhat shift the market for the book (i.e., other experienced researchers will find the text very useful given their ability to fill in the gaps but may limit the classroom use of this book).
Focusing on the book's content: I would divide the book into three major sections. Section 1 reviews the basic and traditional approaches to the estimation of market power models including Structure-Conduct-Performance (SCP), single market formulations of the Lerner formulation, and models of residual demand. The second section focuses on applied game theory and formulations from these models that can be estimated. The third section then focuses on the estimation of strategy models using generalized maximum entropy.
Taking each section in turn, the first section is the most familiar to those of us who have been traditionally interested in the area of imperfect competition. Specifically, the model provides a critique of the standard SCP formulations including discussions of the measurement of return and industrial structure. The discussion of the traditional approaches is followed by a discussion of recent work by Sutton who extends the SCP to examine the effect of promotion and market growth on the sector's performance. The following chapters (chapters three and four) in my opinion are the best empirically developed, probably for two reasons. First, these models are the ones I remember from graduate school and have worked with over time. Second, the models are easily formulated from basic economic theory and follow the typical econometrics we develop in the classroom (if you include generalized instrumental variable and generalized method of moments approaches). The fact that these formulations are reasonably well known allows the authors to demonstrate the potential fragility of market power models using simulation analysis. In addition, the presentation of the differentiated demand model is quite well done, but I have to admit a certain preference for a Rotterdam formulation over the Linear Approximation to the Almost Ideal Demand System.
The second section of the book develops empirical models game theoretic solutions focusing primarily on dynamic games. It is in this section that the magnitude of the task facing the authors becomes apparent. Specifically, they must start with a mini-course or primer in dynamic game theory. Although this primer is well written and very useful (and probably necessary for most readers), it strays somewhat from the empirical focus of the book. Chapters seven and eight then follow the development of the dynamic game in chapters five and six. Chapter seven presents several empirical models for the general dynamic game. It is at this point that a good background in optimal control theory (or at least Bellman formulations) is required. It also highlights the usefulness of generalized method of moments in the estimation of control conditions. Chapter eight then examines empirical models based on the linear-quadratic control formulation. Consistent with the difficulties associated with estimating these models, few if any empirical results are presented for general control formulation whereas chapter eight has several empirical results based on previous research by the authors into coffee and rice markets.
The third section of the book focuses on the estimation of strategy models using entropy approaches. The approach outlined involves segmenting continuous price and investment strategies (typically investment in advertising stock) into a series of discrete points. Each of these discrete points is then a potential strategy in response to the same grid for the other firm. In one example discussed in the book, the pricing strategy for American and United Airlines are estimated. Specifically, probabilities for the pricing response for American Airlines on a given route are estimated based on United Airlines announced prices and other exogenous factors. Given that the response is a probabilistic statement, the response grid can be estimated using a multinomial probability formulation such as a multinomial logit or by entropy. Finally, the authors derive the restrictions on the multinomial structure that are consistent with the Nash-in-price equilibrium and demonstrate how this restriction can be tested.
In general, I found the entire manuscript interesting and potentially useful. Although the first section is probably more fully developed from an empirical perspective, I think that the empirical procedures in sections two and three will probably have the greatest impact on empirical work in the area of market power and strategy. These two sections will "move the goalposts" in our discipline.
Charles B. Moss
University of Florida
FEED