On the Usefulness of Financial Variables Realised Volatility for Recession Forecasting and Business Cycles Turning Points Dating

The main goal of this paper was to check usefulness of introducing measures of the financial markets risk into multivariate forecasting and business cycle dating models to improve their predictive and turning points detection power. Realised volatility was selected as market risk synthetic measure and introduced into two recession dating algorithms: Harding & Pagan (2002) mechanical procedure and Markov Switching Dynamic Factor Model (MS-DFM) with mixed frequencies and missing data handling. In the theoretical part of the article mathematical background of the realised volatility concept and MS-DFM model were presented. It was also described how the output of the MS-DFM model can be used to date turning points. This approach to local maxima detection was compared with Harding and Pagan competitor algorithm. In the practical part of the paper recession detection improvements stemming from introduction of realised volatility measures into MS-DFM model/Harding & Pagan procedure were examined for US and four Western Europe countries (Germany, France, United Kingdom and Italy) in the time span of 20 years between 1990 and 2010