Bayesian quantile regression wiht endogeneity for various type of data

2017 Fiscal Year Final Research Report

• Project/Area Number: 15K17036
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Economic statistics
• Research Institution: Chiba University
• Principal Investigator: Kobayashi Genya 千葉大学, 大学院社会科学研究院, 准教授 (00725103)
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: 分位点回帰 / ベイズ統計学 / 非線形回帰

Outline of Final Research Achievements

In this project, we developed the Bayesian quantile regression models with endogeneous covariates and estimation methods for the proposed models. The proposed model consists of two equation as in the conventional instrumental regression where the second stage regression includes the residuals from the first stage regression in order to remove the bias from the endogeneity. We also considered introducing nonlinear effects of endogenous covariates on the quantiles of the response variable. Although the nonlinear quantile regression is a flexible approach to quantile estimation, the estimates are knwon to be unstable in when the sample size is small or when the quantiles in the tails are estimated. We replace the asymmetric Laplace distribution, whose shape is severely restrictive for a data distribution, with the generalised asymmetric Laplace distribution for improved estimation performance.

Free Research Field

ベイズ統計学