On the applications of resampling methods in the economic sciences (Q2743215)

This Ph.D. thesis is concerned with the application of resampling procedures in the economic science and thus located between the fields of statistics and economics. Resampling procedures are statistical techniques which reproduce a given sample given certain rules for the drawings. To be a useful tool it is assumed that the given sample is a sufficiently close approximation of the unknown population. Although the drawn samples result from the given sample, these samples will in general not be identical to the realizations of the random variables of the basic sample. Therefore, the drawn samples are interpreted as random samples and thus allowing to apply statistical procedures based on probability theory. Using this technique allows to apply probabilistic procedures in economic science, a field in which (repeated) experiments are rather seldom.NEWLINENEWLINENEWLINEAfter the introductory chapter, the author surveys different procedures in the context of resampling techniques, distinguishing between the jackknife- and the bootstrap procedures. The bootstrap procedures are divided into parametric and nonparametric approaches. In chapter 3 the author investigates the current state of research in the field of resampling with special emphasis given to the problem of statistical inference. Topics in this chapter are bootstrap significance tests, bootstrap confidence intervals as well as bootstrap regression. Chapters 2 and 3 are concerned with the formal aspects of resampling. The following two chapters ask for possible fields of application in economic science (chapter 3) first, and then discuss some economic examples, demonstrating the usefulness of the resampling procedures. These examples include empirical ML-estimators, sampling distributions of economic indicators, confidence intervals for economic time series as well as confidence intervals for econometric models. Chapter 6 concludes and presents some areas of future research.NEWLINENEWLINENEWLINEGiven that resampling techniques become a more important tool in applied economics the author gives a comprehensive survey of the current state of these approaches and thus opens a door for applied researchers to use these rather new techniques to improve the statistical inference in economic research and therefore to provide the public with more reliable results than have been able in former times.