Once a year, 5-year forecasts of the second moments (volatilities and correlations) of returns are prepared for different asset classes. At a workshop the results will be presented and explained to the principal.
Modelling the evolution of the second moments in dependency on macroeconomic trends will be instrumental in preparing a long-term forecast of volatilities. The advantage is that cyclical movements proceed relatively slowly and thus lend themselves to forecasting. This approach has found support in the literature. There are, for example, empirical results that the volatility of inflation expectations is linked, among other things, with the amount of inflation. Since returns are meant to include an expectation component regarding future inflation, there is reason to assume that such a volatility of inflation expectations is transferable to a volatility of yields. Since the progress of inflation is relatively slow in comparison to the development of financial market prices this should have a forecasting potential. Also other macroeconomic factors might be useful here. Interim results indicate, for example, that there could be a connection between the volatility of money-market interest rates and the growth rate of US industrial production.
The forecasting will progress in three blocks. In the financial econometrics part because of the large number of asset classes and the partially short time series for returns the dimension of the asset classes will be reduced to the major factors (risk drivers) with the help of a dynamic factor model specially developed for this purpose; the second moment of the factors will be described by means of a multivariate volatility model. Then macroeconomic determinants that have been singled out as important will be forecasted. Finally, with their help, the second moments of the financial-market factors will be calculated or simulated for the forecast period, from which the second moment of asset classes will be calculated back.
RSS