CEU eTD Collection (2014); Dyachenko, Artem: Sequential Bayesian Learning in the Present Value Model

CEU Electronic Theses and Dissertations, 2014
Author Dyachenko, Artem
Title Sequential Bayesian Learning in the Present Value Model
Summary In this thesis, I re-estimate the present value model with hidden variables of Binsbergen and Koijen
(2010) using the sequential Monte Carlo algorithm instead of the maximum likelihood procedure. The latent variables are expected returns and expected dividend growth. First, I show that in-sample forecasts are more optimistic than real-time, out-of-sample forecasts. Second, out-of-sample returns are close to being unpredictable, which implies that returns may follow a martingale process.
Supervisor Fulop Andras
Department Economics MA
Full texthttps://www.etd.ceu.edu/2014/dyachenko_artem.pdf

Visit the CEU Library.

© 2007-2021, Central European University