Calibrating the Confidence Intervals in Threshold Regression
Grant Data
Project Title
Calibrating the Confidence Intervals in Threshold Regression
Principal Investigator
Professor Yu, Ping
(Principal Investigator (PI))
Co-Investigator(s)
Professor Phillips Peter
(Co-Investigator)
Duration
30
Start Date
2021-01-01
Completion Date
2024-06-20
Amount
320000
Conference Title
Calibrating the Confidence Intervals in Threshold Regression
Keywords
Confidence Interval, Threshold Regression, Uniform Validity
Discipline
Econometrics and Business Statistical Methods
Panel
Business Studies (B)
HKU Project Code
17501420
Grant Type
General Research Fund (GRF)
Funding Year
2020
Status
Completed
Objectives
1) We will develop a new method of calibration to modify the existing likelihood ratio confidence interval for the threshold point so that the nominal coverage is well matched. 2) Further modifications will be designed to adapt existing Bonferroni-type confidence intervals for the regular regression coefficients in threshold regression and theoretical justification for these modifications will be developed. 3) Simulation studies will be conducted to check the validity and finite sample performance of these new methods. 4) Empirical applications will be performed to assess the impact of these modifications in practical work with real data.