Suntory and Toyota International Centres for Economics and Related Disciplines (STICERD) LSE RSS Contact Us YouTube Twitter


Econometrics Paper
Nonparametric Censored and Truncated Regression
Arthur Lewbel and Oliver Linton
April 2000
Paper No' EM/2000/389:
Full Paper (pdf)

Tags: semiparametric; nonparametric; censored regression; truncated regression; tobit; latent variable

The nonparametric censored regression model, with a fixed, known censoring point (normalized to zero), is y = max[0,m(x) + e], where both the regression function m(x) and the distribution of the error e are unknown. This paper provides estimators of m(x) and its derivatives. The convergence rate is the same as for an uncensored nonparametric regression and its derivatives. We also provide root n estimates of weighted average derivatives of m(x), which equal the coefficients in linear or partly linearr specifications for m(x). An extension permits estimation in the presence of a general form of heteroscedasticity. We also extend the estimator to the nonparametric truncated regression model, in which only uncensored data points are observed. The estimators are based on the relationship ?E(yk\x)/?m(x) = kE[yk-1/(y > 0)x ], which we show holds for positive integers k.