Generalized Least Squares Model Averaging and Confidence Set around Model Averaged Estimate

2015 Fiscal Year Final Research Report

• Project/Area Number: 25780148
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Economic statistics
• Research Institution: Otaru University of Commerce
• Principal Investigator: LIU QINGFENG 小樽商科大学, 商学部, 教授 (60378958)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: Model Averaging / Model Selection / Optimality / Information Criterion / Confidence Set

Outline of Final Research Achievements

In order to reduce the risk of estimation, we proposed several model averaging methods. The asymptotic properties of those methods were investigated and simulation studies were conducted. The simulation results show that our methods work well and performs better than alternative methods in finite samples. Those methods can be widely applied to various fields, including high dimensional data analysis, macroeconomic and financial market forecasting and forecasting for natural science.
We proposed two model-averaging method, the heteroscedasticity-robust method (HRCp) and the generalized least squares model averaging method (GLSMA) for linear regression models with heteroscedastic errors. Moreover, we constructed a confidence set around model averaging estimate, and conducted simulation studies to check its finite sample properties.

Free Research Field

計量経済学