2020 Fiscal Year Annual Research Report
A sample selection model with a monotone selection correction function
| Project/Area Number |
19K13666
|
|---|
| Research Institution | University of Tsukuba |
|---|
Principal Investigator |
YU ZHENGFEI 筑波大学, 人文社会系, 助教 (40774758)
|
|---|
| Project Period (FY) |
2019-04-01 – 2021-03-31
|
| Keywords | Monotonicity / Shape restriction / Tuning-parameter-free / Sample selection model / Ordered response model / Accelerated failure time / Semiparametric |
|---|
| Outline of Annual Research Achievements |
FY2020's research evaluates the practical performance of the proposed sample selection estimation method that uses a monotone control function (MCF). Its theoretical properties have been studied in FY2019. Monte Carlo simulations show that the MCF estimator, which does not require any tuning parameters (such as the bandwidth) yields smaller mean square errors than the conventional kernel estimator in finite samples, even when the latter uses a cross-validation bandwidth. Then the MCF estimator is applied to a US dataset (Merged Outgoing Rotation Groups of Current Population Survey for 2013) to study the wage equation of women. In this example, the MCF estimator produces shorter confidence intervals than the kernel estimators. This year's research also extends the ``estimation with the monotonicity restriction'' to ordered response models.
Overall, the research project incorporates a monotonicity restriction to popular semiparametric models in economic researches such as the sample selection model. The resulting estimator frees the practitioners from choosing tuning parameters. The proposed estimator for coefficient parameters obtains the desirable root-n convergence rate, although the intermediate estimator for the nonparametric part converges at a slower rate. The project provides practitioners with a new easy-to-implement estimation method for several semiparametric models.
|
