Advances in Economics and Econometrics: Theory and Applications: Bootstrap methods in econometrics: theory and numerical performance

摘要: The bootstrap is a method for estimating the distribution of an estimator or test statistic by resampling one's data. It amounts to treating data as if they were population purpose evaluating interest. Under mild regularity conditions, yields approximation that at least accurate obtained from first-order asymptotic theory. Thus, provides way substitute computation mathematical analysis calculating difficult. maximum score Manski (1975, 1985), developed Ha..rdle et al. (1991) testing positive- definiteness income-effect matrices, and certain functions time- series (Blanchard Quah 1989, Runkle 1987, West 1990) are examples in which difficult bootstrapping has been used alternative.1 In fact, often more finite samples than approximations but does not entail algebraic complexity higher-order expansions. it can provide practical improving upon approximations. First-order theory gives poor distributions statistics with sample sizes available applications. As result, nominal levels tests based on critical values be very different true levels. information matrix White(1982) well-known example large finite- distortions level occur when (Horowitz 1994, Kennan Neumann 1988, Orme 1990, Taylor 1987). Other illustrations given later this chapter. tractable reduce eliminate statistical tests.