CFE 2015: Start Registration
View Submission - CFE
A0272
Title: Nonlinear dynamic interrelationships between real activity and stock returns Authors:  Henri Nyberg - University of Turku and University of Helsinki (Finland) [presenting]
Markku Lanne - University of Helsinki (Finland)
Abstract: The aim is to explore the differences between a linear Gaussian structural VAR (SVAR) and noncausal vector autoregressive VAR models in capturing a potentially nonlinear real activity-stock return-relationship. Unlike the conventional and commonly used linear SVAR model, the noncausal VAR model is capable of accommodating various nonlinear and non-Gaussian dependencies characteristic of stock returns, and, therefore, it is likely to produce real activity and stock market shocks different from those implied by a SVAR model. In quarterly U.S. data (1953-2012), we find evidence in favor of noncausality suggesting that the stock return is an insufficient proxy for news affecting real activity and stock prices. In addition, especially when conditioning on the state of the business cycle, GDP growth is found more important for the determination of stock returns than in a linear SVAR model.