Average and Quantile Effects in Nonseparable Panel Models
研究了在时间同质性条件下，不可分面板模型中平均效应和分位数效应的识别与估计，给出了非参数和半参数模型下的边界，并提出了适用于面板数据的推断方法。
Nonseparable panel models are important in a variety of economic settings,\nincluding discrete choice. This paper gives identification and estimation\nresults for nonseparable models under time homogeneity conditions that are like\n"time is randomly assigned" or "time is an instrument." Partial identification\nresults for average and quantile effects are given for discrete regressors,\nunder static or dynamic conditions, in fully nonparametric and in\nsemiparametric models, with time effects. It is shown that the usual, linear,\nfixed-effects estimator is not a consistent estimator of the identified average\neffect, and a consistent estimator is given. A simple estimator of identified\nquantile treatment effects is given, providing a solution to the important\nproblem of estimating quantile treatment effects from panel data. Bounds for\noverall effects in static and dynamic models are given. The dynamic bounds\nprovide a partial identification solution to the important problem of\nestimating the effect of state dependence in the presence of unobserved\nheterogeneity. The impact of $T$, the number of time periods, is shown by\nderiving shrinkage rates for the identified set as $T$ grows. We also consider\nsemiparametric, discrete-choice models and find that semiparametric panel\nbounds can be much tighter than nonparametric bounds.\nComputationally-convenient methods for semiparametric models are presented. We\npropose a novel inference method that applies in panel data and other settings\nand show that it produces uniformly valid confidence regions in large samples.\nWe give empirical illustrations.\n