Research On Estimable Models Of Discrete Static Games

The proposal of game theory promotes the development of microeconomic theory and has important applications in many fields of microeconomic theory.Neither individuals act in a vacuum,each faces many interrelated decisions,most of these decisions involve strategic interactions.The interrelated nature of these decisions suggests modelling them as strategic games.The traditional microeconomic theory can not be applied in the empirical analysis of games because it can not involve strategic interactions between players.Estimable models of games are useful tools in the empirical analysis of games which involve strategic interactions between players.It is an important expansion of the traditional microeconomic theory.Estimable models of games have high theoretical value and application prospect.The purpose of estimating models of games is to estimate players’ payoff function.This paper studies estimating models of discrete static games which concern choices made from a finite set of alternatives,where players’ payoffs from making each choice depend on the decisions of other players.Estimable models of Discrete static games involve information setting that are either complete or incomplete.We study estimating models of discrete static games of complete information and estimating models of discrete static games of incomplete information respectively.First,we study to estimate models of discrete static games of complete information.We construct estimable models of discrete static games of complete information,illustrate the coherency problems raised by the existence of multiple equilibria,introduce two methods to solve the coherency problems and estimate models of discrete static games of complete information.The first method solve the coherency problems by aggregating to a different set of equilibriums which are robust to a unique equilibrium,and the second method solve the coherency problems by placing the sequential moves which guarantee a unique equilibrium.Then,we use Monte Carlo simulations methods to obtain the finite sample performance of two methods.As an application of estimable models of discrete static games of complete information and two methods,we study firms’ entry games in the retail discount industry in America.Next,we turn to estimate models of discrete static games of incomplete information.We construct estimable models of discrete static games of incomplete information,illustrate the fixed point problems,introduce the two-step pseudo maximum likelihood method(2S-PML)to estimate models of discrete static games of incomplete information,propose the nested pseudo likelihood method(NPL)to estimate models of discrete static games of incomplete information,study the asymptotic properties of the NPL estimator,and use Monte Carlo simulations methods to obtain the finite sample performance of the NPL method.The NPL method is a recursive extension of the 2S-PML method,and it performs better than the2S-PML method in finite samples.Finally,we apply estimable models of discrete static games of incomplete information and the NPL method to the retail discount industry in China,study market structure and competitive interactions between firms in the retail discount industry in China.The results show that the market’s population number and economic developmant level has a significant positive effect on firm’s profits,the distance from market’s locations to firm’s headquarters has a significant negative effect on firm’s profits,the competitive interactions between firms has a significant negative effect on firm’s profits,and the competitive interactions between firms has different effects on different firms,the presence of rivals can reduce a firm’s profits,firms prefer to shield their profits by choosing sufficiently spatial different market’s locations.Then,we propose the semi-parametric 2S-PML method to estimate models of discrete static games of incomplete information,study the asymptotic properties of the semi-parametric 2S-PML estimator,and use Monte Carlo simulations methods to obtain the finite sample performance of the semi-parametric 2S-PML method.We also propose the semi-parametric NPL method to estimate models of discrete static games of incomplete information,study the asymptotic properties of the semi-parametric NPL estimator,and use Monte Carlo simulations methods to obtain the finite sample performance of the semi-parametric NPL method.Finally,we construct estimable models of discrete static games of incomplete information,apply the semi-parametric 2S-PML method and the semi-parametric NPL method to the problem of determining the factors that govern the assignment of investment rating by analysts.The results show that analysts’ investment rating is more optimism for the companies which have higher growth,well-established,smaller earnings volatility or having investment banking relationship with analysts’ brokerages.And analysts raise their investment rating proportionally to the investment rating they expect from other analysts.The peer effects are significant in the assignment of investment rating by analysts.