2018 Fiscal Year Annual Research Report
A study of estimation and inference of the density discontinuity
| Project/Area Number |
17K13713
|
|---|
| Research Institution | University of Tsukuba |
|---|
Principal Investigator |
YU ZHENGFEI 筑波大学, 人文社会系, 助教 (40774758)
|
|---|
| Project Period (FY) |
2017-04-01 – 2019-03-31
|
| Keywords | Density discontinuity / Manipulation test / Empirical likelihood / Regression discontinuity |
|---|
| Outline of Annual Research Achievements |
FY2017's research provides a mean square error optimal bandwidth selector for the parameter which is a function of densities at both sides of a threshold (for example, the density ratio). It allows bandwidths to differ at both sides. FY2018's research provides an optimal bandwidth selector that minimizes the coverage error for testing a function of densities at both sides of a threshold. It covers the manipulation test for regression discontinuity designs as a special case. The testing approach is based on empirical likelihood whose moment conditions come from a minimum contrast problem. The two year's researches together address the bandwidth selection problem for estimation and inference of density discontinuity, which can be applied to the validity check of regression discontinuity designs and to estimate agent's response to a policy change.
Simulation studies confirm that the the proposed optimal bandwidths yield smaller mean square errors for the estimation problem, better size and power properties for the testing problem than existing methods. Empirical examples illustrate their usefulness in real data sets. The research results have been presented at various conferences. The preliminary result is published in Economic letters. A paper containing the main results is accepted by Journal of Business & Economics Statistics. The computer codes for the proposed testing approach will also be available online.
|
| Remarks |
This paper is accepted by Journal of Business & Economics Statistics.
|
Research Products
(6 results)
