26, 27, 28 ноября 2012 года. Аудитория 615 ИППИ
Valentin PATILEA (CREST-ENSAI, France)
Nonparametric goodness-of-fit checks for conditional moment restrictions
Many statistical models could be defined through conditional moment equations: nonlinear mean regressions, quantile regressions, transformation models, some benchmark econometric models… The parameters of the model are supposed to be identified by the conditional moment equations. A natural question is then whether the model is correctly specified or not, that is do exist parameters satisfying the conditional moment equations?
We review the main approaches introduced in the literature and we focus on kernel smoothing based ones which yield test statistic with standard normal asymptotic critical values. Next, we show how the kernel smoothing based nonparameric testing could be extended to: (a) nonparemetric significance testing; (b) incomplete observations (censored data); and (c) high-dimension data case (functional covariates).
| 24.11.2012 | |