Semi-and Non-Parametric Estimation of Econometric Duration Models

点击翻译按钮获取中文摘要

Empirical findings from the same data set can change drastically when the underlying statistical assumptions are changed. This is because many econometric procedures can give misleading results when an incorrect model is assumed. One class of models, where it has been demonstrated that the empirical findings can depend on arbitrary statistical assumptions, is models where the variable of interest is the duration of some event such as length of time unemployed. The purpose of this project is to carefully study the structure of commonly used econometric duration-models, in order develop estimation strategies that are robust in the sense that few auxiliary assumptions have to be made for them to have desirable statistical properties. To study the usefulness of the estimation strategies, data will be generated from a variety of models. Then estimators will be constructed and applied to the data to see whether the true underlying model can be determined. This project is mainly theoretical, but it will also have important implications for empirical economic. When a model is estimated, it is often necessary to make arbitrary assumptions in order to make the estimation computationally feasible. It is then important to understand to what extent the results are driven by these assumptions, and whether the results can be expected to be robust to changes in the assumptions. The methods developed in this project will help empirical researchers perform this task for econometric duration-models.