Estimation of Assets Returns Predictability in Europe and its Implications on Individual Assets Management (FEJON)

FEJON is a study contract with L'Observatoire de l'Epargne Européenne and Eric Jondeau and Michael Rockinger of University of Lausanne.

The objective of this research is to extend the existing literature that deals with forecasting returns' distribution and associated asset allocation, by using European data, and to investigate forecasting ability at various time horizons by using different types of explanatory variables: macroeconomic variables in the long run and a time-series model in the short run. So far, most models focus on a given time horizon, and then use a calibration for one type of data. We wish to extend this by superimposing the information contained in various frequencies. In such a model low frequency data are likely to affect expected returns and high frequency data the distribution of returns (variance, skewness, and kurtosis). In our setting, by using Bayesian Monte-Carlo techniques, it is also possible to perform portfolio optimization over a multi-period time horizon. We also aim at incorporating in our research state-of-the-art technology that will address the issue of parameter uncertainty as well as a better description of investor's preferences that will explicitly take into account possible preferences for portfolios with less extreme variability, yet with a particular upside potential.

The second objective is to evaluate the consequences for asset management in the longer run, if there is parameter uncertainty and investor's preferences that take explicitly into account the possibility of extreme events (crashes). Parameter uncertainty will in general lead to more precautious allocations. Furthermore, the longer time horizon of allocations will lead to hedging demands. A related issue is a specific preference for rightward skewed assets and portfolios. Investors with such preferences may give up some diversification while expecting to catch those assets that will exhibit rare but important windfalls.

The combination of predictability at different time horizons as well as of parameter uncertainty and more general preferences should allow for a better understanding of how to perform portfolio allocation.