On Merger Simulation With Logit Model

Although we have enacted Anti-Monopoly Law on August 31th 2007. We still do not formulate relevant rules and regulations, made specific provisions and specifications, so that the Law has troubles in application. Therefore, the introduction of Anti-Monopoly Law does not mean that the study and research of anti-trust is drawing to an end, but only just to draw open.Merger regulation is one of important contents of Anti-Monopoly Law. The mainstream analysis of merger is traditional structural analysis which concentrates on market share and market concentration ratio. With the development of unilateral theory and econometric techniques, so as the availability of market data is increasing, there is an obvious trend to use merger simulation method to analyze merger. The studies on merger simulation currently have made fruitful achievements. A wide range of merger simulation models are emerging, including but not limited to linear demand models, log-linear demand models, AIDS and PCAIDS models, simple Logit models, nested Logit models, random-coefficients Logit models, multi-level demand estimation as well as auction models. There are several cases that applied merger simulation to analyze in United States and Europe, such as Interstate Bakeries/Continental Baking Co. case, Kimberly-Clark/Scott Paper case, Staples/Office Depot case, Oracle/People Soft case.The emergency of merger simulation has meaning, it can not only provides a scientific complement for structure method, but also can be used to measure cost efficiency, calculate compensating marginal cost reductions, provide a rough cut for economic analysis of merger investigation, enhance or weaken the conclusions of other analytical methods, analyze the countervailing factors. Market share can not reflect the competition significance in market with product differentiation, so it is difficult to reasonably define the relevant market, merger simulation provides a right choice at this point. The emergence of merger simulation has changed the singleness of merger analysis, increased the diversity.Nevertheless, there are some problems of the employment of merger simulation as an antitrust instrument, the most obvious one is that the results predicted by merger simulation heavily depend on the demand model. Even using the same data, different demand models will obtain distinct results. In order to analysis whether the merger would result in significant impediment to effective competition, we need to identify the right model which is the most appropriate. The right model must not only capture the actual competition and demand, but also be understanding by others. Unfortunately, we can't do it in reality. moreover, there are still some problems, such as data availability, markups that link wholesale and retail prices, results rely on assumptions, neglect of non-quantifiable and long-run competitive effects as well as too expensive.Consequently, although merger simulation has a number of advantages, it also exists many defects, so this paper suggests that we should stick to the following principles in the use of merger simulation:the Daubert principle, the multiplicity principle, the prudence principle and the mutual auxiliary principle. The Daubert principle refers that the simulation must be completed by experts who get good education and training with rich knowledge, skills and experiences, furthermore, the simulation model must be fitting the actual situation of the case and is reasonable in the range of relevant economics. The multiplicity principle refers that we should consider the results of distinct models during simulation to examine whether these results are consistent and determine whether the merger will cause unilateral effects. The prudence principle requires that we should analyze carefully at every moment, not only in data collection and processing, variable selection, model selection and construction, but also in estimate methods, parameters estimation and the whole process of price effect calculation. The auxiliary principle means that merger simulation should used in combination with other methods, as auxiliary tools, mutual verification, and consider the results of different methods. So we could say that the purpose of merger simulation is not for calculating the price effects precisely, but providing a new tool and richening the diversity of merger analysis.To illustrate the above problems and explore the application of merger simulation in the anti-monopoly enforcement, this paper using Chinese air conditioning industry retail monitoring data in March of 2009 to March of 2010 to estimate each brand's demand and cost in the industry based on simple Logit model and nested Logit model, and then simulate the price effects of mergers that mergering in pairs between the five biggest ones, this paper compare the results of simple Logit model, nested Logit model, structure method and PC AIDS.This paper reached the following conclusions:Firstly, the results of merger simulation heavily rely on the demand function that was selected, different models get different even opposite conclusions, though with the same data; Secondly, this paper found that price elasticity play a very important role in merger simulation, so that we should preserve prudence in every links. Although merger simulation could offer a precise price effects, the exact numerical value does not have much significance, and we should take more attention on the direction of results; Thirdly, this paper verify that both the mergered firms and non-merger firms will raise prices, but the price increase of non-merger firms much lower than the mergered ones, and both of them are affected by the market shares before the transaction; Fourthly, merger simulation models seriously affected by its' own drawbacks, for example, the simple Logit model and nested Logit model are affected by the IIA, which affect the final results; Fifth, we could improve the accuracy of merger investigate through considering various results of different methods; Finally, products with lower price or higher market share have lower marginal cost and higher price cost margin.