CMStatistics 2022: Start Registration
View Submission - CFE
A1655
Title: Bayesian reconciliation of the return predictability Authors:  Borys Koval - Vienna University of Economics and Business (Austria) [presenting]
Leopold Soegner - Institute for Advanced Studies (Austria)
Sylvia Fruehwirth-Schnatter - WU Vienna University of Economics and Business (Austria)
Abstract: A stable vector autoregressive (VAR) system comprising asset returns, the dividend-price ratio, and dividend growth allows one to pin down the question of return predictability to the value of one particular parameter of a restricted VAR model. This restricted VAR model is used, and return predictability is investigated in a Bayesian context. We adapt two new priors, a Jeffrey's prior and a prior based on the reduced-bias estimator, and compare our Bayesian estimation routine to other Bayesian (e.g., uniform and Reference prior) and frequentist approaches proposed in the literature by means of an extensive simulation study. In terms of root mean square error (RMSE), mean absolute error (MAE), and credible interval coverage, the approach proposed in this article leads to superior performance relative to ordinary least squares estimation, a Frequentist reduced-bias approach, and Bayesian estimation using priors proposed in the literature. We apply our methodology to S\&P 500 data and find strong evidence for return predictability after properly accounting for the correlation structure and imposing theory-motivated restrictions on the dividend-price ratio.