The Local Polynomial Method And Its Application In Semiparametric Conditional Moments

As a nonparametric statistic method ,Local polynomial regression has advantagesover other nonparametric methods.It not only can estimate the regression function'sderivative ,but also can overcome the edge of the kernel estimator,and the local linearestimator has asymptotic minimax e?ciencies among all linear estimators.This paper hasdone mainly two aspects work based on local polynomial method.First,in order to deal with lacking of data in real application,this paper uses theridge regression method to improve the local polynomial method,then uses it to analyzethe economic operation by analyzing the a?ection of investment and consumption toeconomic growth.Second,this paper use local linear regression to estimate conditional moment m(X,α)in semiparametric method,and gain the estimateαn by SMD(sieve minimum distance)method.In some assumptions ,this paper proves the consistency and convergence rate,inaddition ,proves the asymptotic normality of the parametric partθn.We give the asymp-totic covariance for statistic inference ,and prove its consistency.To obtain the e?cientestimator,this paper introduce the three-step procedure in Ai and Chen(2000,2003),andBlundell(2007)'s sieve profile method.