Coauteurs : N. Boussaha (USTHB, Alger), A. Khalfi (USTHB, Alger)
Résumé : In this paper, we propose a new threshold model to deal with the asymmetry and leverage effect in the volatility of financial time series. Our model generalizes the classical threshold stochastic volatility model. By introducing a new flexible and smooth regime-switching mechanism, we avoid the sudden jump in the log-volatility imposed by the classical model. More precisely, we propose a new threshold model that we call Buffered Threshold Stochastic Volatility (BTSV) model. After the definition of our model, we derive a sufficient condition for the strict stationarity of the model. Then, we develop a Sequential Monte Carlo method to estimate the parameters. Next, we analyse the performance of the proposed method through a simulation study. Finally, we apply our new methodology to model the Honeywell International Inc and Standard and Poor Composite 500 stock market indices. This empirical analysis shows that BTSV model is a competitive alternative for the asymmetry modelling in financial returns analysis.