Arnaud Dufays | Université Catholique de Lille (original) (raw)

Papers by Arnaud Dufays

Research paper thumbnail of A Bayesian method of change-point estimation with recurrent regimes: Application to GARCH models

Journal of Empirical Finance, 2014

We present an estimation and forecasting method, based on a differential evolution MCMC method, f... more We present an estimation and forecasting method, based on a differential evolution MCMC method, for inference in GARCH models subjected to an unknown number of structural breaks at unknown dates. We treat break dates as parameters and determine the number of breaks by computing the marginal likelihoods of competing models. We allow for both recurrent and non-recurrent (change-point) regime specifications. We illustrate the estimation method through simulations and apply it to seven financial time series of daily returns. We find structural breaks in the volatility dynamics of all series and recurrent regimes in nearly all series. Finally, we carry out a forecasting exercise to evaluate the usefulness of structural break models.

Research paper thumbnail of Marginal likelihood for Markov-switching and change-point GARCH models

Journal of Econometrics, 2014

Change-point models are useful for modeling time series subject to structural breaks. For interpr... more Change-point models are useful for modeling time series subject to structural breaks. For interpretation and forecasting, it is essential to estimate correctly the number of change points in this class of models. In Bayesian inference, the number of change points is typically chosen by the marginal likelihood criterion, computed by Chib's method. This method requires one to select a value in the parameter space at which the computation is performed. Bayesian inference for a change-point dynamic regression model and the computation of its marginal likelihood are explained. Motivated by results from three empirical illustrations, a simulation study shows that Chib's method is robust with respect to the choice of the parameter value used in the computations, among posterior mean, mode and quartiles. However, taking into account the precision of the marginal likelihood estimator, the overall recommendation is to use the posterior mode or median. Furthermore, the performance of the Bayesian information criterion, which is based on maximum likelihood estimates, in selecting the correct model is comparable to that of the marginal likelihood.

Research paper thumbnail of Commodities Volatility and the Theory of Storage

CORE Discussion Papers, 2012

One implication of the theory of storage states that commodity price volatility should increase w... more One implication of the theory of storage states that commodity price volatility should increase when inventories are low. We document this volatility feature by estimating asymmetric volatility models for 16 commodity return series, on the period 1994-2011 and show how to account for this feature in Value-at-Risk forecasting. Our contribution is threefold:(i) This study is the first to investigate systematically the volatility implication of the theory of storage for a large panel of commodity types (agriculturals, metals, precious ...

Research paper thumbnail of Estimating and forecasting structural breaks in financial time series

We present an algorithm, based on a differential evolution MCMC method, for Bayesian inference in... more We present an algorithm, based on a differential evolution MCMC method, for Bayesian inference in AR-GARCH models subject to an unknown number of structural breaks at unknown dates. Break dates are directly treated as parameters and the number of breaks is determined by the marginal likelihood criterion. We prove the convergence of the algorithm and we show how to compute marginal likelihoods. We allow for both pure change-point and recurrent regime specifications and we show how to forecast structural breaks. We illustrate the efficiency of the algorithm through simulations and we apply it to eight financial time series of daily returns over the period 1987-2011. We find at least three breaks in all series.

Research paper thumbnail of Specific Markov-switching behaviour for ARMA parameters

We propose an estimation method that circumvents the path dependence problem existing in Change-P... more We propose an estimation method that circumvents the path dependence problem existing in Change-Point (CP) and Markov Switching (MS) ARMA models. Our model embeds a sticky infi nite hidden Markov-switching structure (sticky IHMM), which makes possible a self-determination of the number of regimes as well as of the speci cation : CP or MS. Furthermore, CP and MS frameworks usually assume that all the model parameters vary from one regime to another. We relax this restrictive assumption. As illustrated by simulations on moderate samples (300 observations), the sticky IHMM-ARMA algorithm detects which model parameters change over time. Applications to the U.S. GDP growth and the DJIA realized volatility highlight the relevance of estimating di fferent structural breaks for the mean and variance parameters.

Research paper thumbnail of A Bayesian method of change-point estimation with recurrent regimes: Application to GARCH models

Journal of Empirical Finance, 2014

We present an estimation and forecasting method, based on a differential evolution MCMC method, f... more We present an estimation and forecasting method, based on a differential evolution MCMC method, for inference in GARCH models subjected to an unknown number of structural breaks at unknown dates. We treat break dates as parameters and determine the number of breaks by computing the marginal likelihoods of competing models. We allow for both recurrent and non-recurrent (change-point) regime specifications. We illustrate the estimation method through simulations and apply it to seven financial time series of daily returns. We find structural breaks in the volatility dynamics of all series and recurrent regimes in nearly all series. Finally, we carry out a forecasting exercise to evaluate the usefulness of structural break models.

Research paper thumbnail of Marginal likelihood for Markov-switching and change-point GARCH models

Journal of Econometrics, 2014

Change-point models are useful for modeling time series subject to structural breaks. For interpr... more Change-point models are useful for modeling time series subject to structural breaks. For interpretation and forecasting, it is essential to estimate correctly the number of change points in this class of models. In Bayesian inference, the number of change points is typically chosen by the marginal likelihood criterion, computed by Chib's method. This method requires one to select a value in the parameter space at which the computation is performed. Bayesian inference for a change-point dynamic regression model and the computation of its marginal likelihood are explained. Motivated by results from three empirical illustrations, a simulation study shows that Chib's method is robust with respect to the choice of the parameter value used in the computations, among posterior mean, mode and quartiles. However, taking into account the precision of the marginal likelihood estimator, the overall recommendation is to use the posterior mode or median. Furthermore, the performance of the Bayesian information criterion, which is based on maximum likelihood estimates, in selecting the correct model is comparable to that of the marginal likelihood.

Research paper thumbnail of Commodities Volatility and the Theory of Storage

CORE Discussion Papers, 2012

One implication of the theory of storage states that commodity price volatility should increase w... more One implication of the theory of storage states that commodity price volatility should increase when inventories are low. We document this volatility feature by estimating asymmetric volatility models for 16 commodity return series, on the period 1994-2011 and show how to account for this feature in Value-at-Risk forecasting. Our contribution is threefold:(i) This study is the first to investigate systematically the volatility implication of the theory of storage for a large panel of commodity types (agriculturals, metals, precious ...

Research paper thumbnail of Estimating and forecasting structural breaks in financial time series

We present an algorithm, based on a differential evolution MCMC method, for Bayesian inference in... more We present an algorithm, based on a differential evolution MCMC method, for Bayesian inference in AR-GARCH models subject to an unknown number of structural breaks at unknown dates. Break dates are directly treated as parameters and the number of breaks is determined by the marginal likelihood criterion. We prove the convergence of the algorithm and we show how to compute marginal likelihoods. We allow for both pure change-point and recurrent regime specifications and we show how to forecast structural breaks. We illustrate the efficiency of the algorithm through simulations and we apply it to eight financial time series of daily returns over the period 1987-2011. We find at least three breaks in all series.

Research paper thumbnail of Specific Markov-switching behaviour for ARMA parameters

We propose an estimation method that circumvents the path dependence problem existing in Change-P... more We propose an estimation method that circumvents the path dependence problem existing in Change-Point (CP) and Markov Switching (MS) ARMA models. Our model embeds a sticky infi nite hidden Markov-switching structure (sticky IHMM), which makes possible a self-determination of the number of regimes as well as of the speci cation : CP or MS. Furthermore, CP and MS frameworks usually assume that all the model parameters vary from one regime to another. We relax this restrictive assumption. As illustrated by simulations on moderate samples (300 observations), the sticky IHMM-ARMA algorithm detects which model parameters change over time. Applications to the U.S. GDP growth and the DJIA realized volatility highlight the relevance of estimating di fferent structural breaks for the mean and variance parameters.