【主讲】Jian Zhang (University of Wisconsin-Madison)
【主题】Nonparametric Identification and Estimation of Transformed Model with Additive Range
【时间】2016年1月11日 (周一) 15：30-17：00
【地点】上海财经大学经济学院楼801室
【语言】英文
【摘要】This paper studies nonparametric identification and estimation of transformed model with additive range. Special cases of such a model include the transformed additively separable model, the single index model, and the transformed partial linear model. First, we identify our model with a strategy allowing discrete independent variables and non-differentiable transformation function which are motivated by a class of economic applications. This is distinct from the related literature which typically imposes continuity on independent variables and differentiability on transformation function to identify the model. Specifically, we identify the model up to location and scale under a key range assumption which requires the range of transformation to be additive and overlapping. When the range of transformation is not overlapping but a union of K disconnected intervals, we show that our model can be identified up to K location-scale normalizations. Second, after imposing the differentiability of transformation function, we propose a nonparametric estimator for the parameter of interest, and establish its asymptotic properties. Monte Carlo experiments show that our estimator performs well in finite samples.
