動態追蹤資料分量迴歸

Abstract

本研究計劃預計應用分量迴歸方法於動態追蹤資料具有固定效果的模型, 以下簡稱 DPQR 模型, 此模型可以分析解釋變數對不同分量下的被解釋變數的異質性效果。 由於在 DPQR 模型下, 落後期被解釋變數具有內生性, 本研究計劃將應用配適值方法 (fitted value ap- proach) 來解決內生性偏誤的問題。 因此, 在 DPQR 模型下, 本人將建構一個兩階段的估 計步驟, 其中第一個步驟為先對具內生性的落後期被解釋變數進行估計, 並得出其配適值, 接著第二步驟再利用此配適值來取代落後期被解釋變數, 然後進行追蹤資料分量迴歸之估 計。 此方法由於相當簡單可行, 因此預計將會被廣泛使用。

This project applies a quantile regression method for dynamic panel data model with fixed effects, henceforth the dynamic panel quantile regression (DPQR) model, for revealing heterogeneity effects of regressors on the dependent variable. In the DPQR model, the lagged dependent variable is endogenous due to the existence of fixed effects. This project adjusts for the bias of endogeneity by adopting a “fitted value” approach, and develops a two-stage estimation procedure for the DPQR model with fixed effects. The first step of the two-stage estimation consists of estimation of a fitted value for the lagged endogenous dependent variable. The second step is to replace the endogenous variable in the model by its fitted value, and to run a quantile regression for panel data directly to obtain the estimators. The fitted value approach, by simply running two-stage regressions, is appealing in that it is generally applicable and easy to implement.