Long term vs. short term comovements in stock markets: the use of Markov-switching multifractal models.

Empirical techniques to assess market comovements are numerous from cointegration to dynamic conditional correlations. This paper uses the fractal properties of asset returns and presents estimations of Markov switching multifractal models [as MSM] to give new insights about short and long run dependencies in stock returns. The main advantage of the model is to allow for the derivation of several indicators of comovements on heterogenous lasting horizons. Empirical applications are performed for four stock indices (CAC DAX FTSE NYSE) at daily frequency between 1996 and 2008.