TESTING A CLASS OF SEMI- OR NONPARAMETRIC CONDITIONAL MOMENT RESTRICTION MODELS USING SERIES METHODS
提出一种新检验方法，用于检验参数化涉及未知条件期望函数的条件矩约束模型，适用于不完全信息静态博弈等离散选择模型，并通过沃尔玛和凯马特市场进入案例验证模型有效性。
This paper proposes a new test for a class of conditional moment restrictions (CMRs) whose parameterization involves unknown, unrestricted conditional expectation functions. Motivating examples of such CMRs arise from models of discrete choice under uncertainty including certain static games of incomplete information. The proposed test may be viewed as a semi-/nonparametric extension of the Bierens (1982, Journal of Econometrics 20, 105–134) goodness-of-fit test of a parametric model for the conditional mean. Estimating conditional expectations using series methods and employing a Gaussian multiplier bootstrap to obtain critical values, the test is shown to be asymptotically correctly sized and consistent. Simulation studies indicate good finite-sample properties. In an empirical application, the test is used to study the validity of a game-theoretical model for discount store market entry, treating equilibrium beliefs as nonparametric conditional expectations. The test indicates that Walmart and Kmart entry decisions do not result from a static discrete game of incomplete information with linearly specified profits.