Research On Semiparametric and Nonparametric Estimation of Econometric Models

9307677 Horowitz计量经济学最近的许多研究都关注于发展半参数和非参数推理方法，以减少推理所需的先验和通常不可检验的假设的数量，从而减少由于错误假设而导致的不正确推理的可能性。该项目旨在扩展半参数和非参数方法在几个方面具有重要的应用。这些包括转换模型的半参数和非参数估计，数据中有错误的模型，以及有损耗的面板数据。将开发具有重要经济应用价值的新方法。该项目在以下领域开发了新的方法和经济应用：(1)因变量未知变换的回归模型的估计；(2)数据存在误差的回归模型的非参数估计；(3)磨损对面板数据推断影响的非参数估计；(4)利用基于自举的临界值对广义矩法（GMM）估计的过识别限制进行检验；(5)半参数估计量和非参数估计量的带宽选择。＊＊＊

9307677 Horowitz Much recent research in econometrics has been concerned with developing semi- and nonparametric methods of inference in order to reduce the number of a priori and, often, untestable assumptions needed for inference and, thereby, reduce the potential for incorrect inference due to erroneous assumptions. The project aims at extending semi- and nonparametric methods in several ways that have important in applications. These include the semi- and nonparametric estimation of transformation models, models with errors in the data, and panel data with attrition. New methods will be developed that have important economic applications. The project developed new methods and economic applications in the following areas: (1) Estimation of a regression model with an unknown transformation of the dependent variable; (2) Nonparametric estimation of regression models with errors in the data; (3) Nonparametric estimation of the effects of attrition on inferences derived from panel data; (4) Use of bootstrap-based critical values for the test of overidentifying restrictions on generalized-method-of moment (GMM) estimation; and (5) Bandwith selection for certain semiparametric and nonparmetric estimators. ***