Estimation of the Transformation Function in Fully Nonparametric Transformation Models with Heteroscedasticity
share
Completely nonparametric transformation models with heteroscedastic errors are considered. Despite their flexibility, such models have rarely been used so far, since estimators of the model components have been missing and even identification of such models has not been clear until very recently. The results of Kloodt (2020) are used to construct the first two estimators of the transformation function in these models. While the first estimator converges to the true transformation function at a parametric rate, the second estimator can be obtained by an explicit formula and is less computationally demanding. Finally, a simulation study is followed by some concluding remarks. Assumptions and proofs can be found in the appendix.
READ FULL TEXT