Estimating a Markov Switching with Time Varying Transition Probability Model: An Extension to the Filardo-Gordon’s Algorithm (original) (raw)
Related papers
A Markov Switching Regime Model of the Brazilian Business Cycle
SSRN Electronic Journal, 2000
Previous studies have shown that linear models are incapable of capturing business cycle dynamics with accuracy. This has brought interest in non-linear models such as the Markov switching (MS) regime technique, which can distinguish business cycle recession and expansion phases, and is sufficiently flexible to allow different relationships to apply over these phases. This technique can be used to simultaneously estimate the data generating process of real GDP growth and classify each observation into one of two regimes (i.e. lowgrowth and high-growth regimes). In this study, we investigate the dynamics of the Brazilian business cycle using a Markov regime switching vector autoregressive model (MS-VAR). The study was developed using time-variable transition probabilities (TVTP) and for comparison and validation, fixed transition probabilities (FTP) between regimes. In order to capture cyclical fluctuations of the Brazilian GDP, we use the yield spread as a leading indicator. Most of the results obtained with the MS-VAR-TVTP are according to the expected. We show that the model is adequate to predict short-term cyclical fluctuations in the Brazilian economy. We also estimate an MS-VAR model with FTP in order to validate the TVTP model. We confirm the relevance of the yield spread in the estimation of the model parameters and the transition probabilities.
Tunisia by taking the transition probabilities as endogenous in a Markov switching framework. Using Matlab programming of the Gibbs algorithm, Bayesian analysis allowed us to deal with the hidden Markov process with variable transition probabilities. Showing a persistent state, we obtained a positive relationship between previous and current regimes. These are presented as information leading to the variability of probabilities transition cycles. Furthermore, an anticipated increase in France would have a delayed effect on the business cycle in Tunisia because of domestic rigidities and institutional constraints as to depolarization. During recession in France, the persistence of expansion phases compared to recession seems to be verified in the Tunisian context. This type of application is not abundant in the empirical literature in Tunisia. Based on the various robustness tests (Vuong, 1989; Ang and Bekeart, 2002), the supremacy of MS-TVTP models over FTP in the treatment of cyclical fluctuations in Tunisia is shown.
Evaluating Currency Crisis : A Bayesian Markov Switching Approach
2006
In this paper we examine the nature of currency crisis. In line with Jeanne (1997) and Jeanne (2000), we provide an empirical support of the view that both fundamentals and annimal spirits play an important role in the genesis of currency crisis. We do so by employing an out-of-sample forecasting exercise to analyse the Mexican crisis in 1994. We also extend the empirical framework suggested by Jeanne and Masson (2000) to test for the hypothesis that currency crisis was driven by sunspots. To this end we contribute to the existing literature by comparing Markov regime switching model with time-varying transition probabilities with two alternative models. The rst is a Markov regime switching model with constant transition probabilities. The second is a linear benchmark model. Empirical results show that Markov regime switching model with time varying transition probabilities outperfoms both linear and nolinear alternative models but it fails to predict the Mexican currency crisis in...
Regimes Markov models with endogenous transition probabilities: Modeling fluctuations in Tunisia
This paper is an advanced analysis of the cyclical industry in Tunisia by taking the transition probabilities as endogenous in a Markov switching framework. Using Matlab programming of the Gibbs algorithm, Bayesian analysis allowed us to deal with the hidden Markov process with variable transition probabilities. Showing a persistent state, we obtained a positive relationship between previous and current regimes. These are presented as information leading to the variability of probabilities transition cycles. Furthermore, an anticipated increase in France would have a delayed effect on the business cycle in Tunisia because of domestic rigidities and institutional constraints as to depolarization. During recession in France, the persistence of expansion phases compared to recession seems to be verified in the Tunisian context. This type of application is not abundant in the empirical literature in Tunisia. Based on the various robustness tests (Vuong, 1989; Ang and Bekeart, 2002), the supremacy of MSTVTP models over FTP in the treatment of cyclical fluctuations in Tunisia is shown.
Modeling Business Cycle Fluctuations through Markov Switching VAR:An Application to Iran
2016
IN this paper, the Iranian Business Cycle characteristics were investigated via uni-variate and multivariate Markov-switching specifications. By using Hamilton (1989) and Krolzig (1997) (MS-VAR) models, we examined the stochastic properties of the cyclical pattern of the quarterly Iranian real GDP between 1988:Q2 - 2008:Q3. The empirical analysis consists of mainly three parts. First, two kinds of alternative specifications were tried and we were adopted best specification with respect to various diagnostic statistics. Then, selected models were tested against their linear benchmarks. LR test results imply strong evidence in favor of the nonlinear regime switching behavior. Furthermore, the multivariate specification with various macro aggregates and changing variance parameter outperformed the other MS models with reference to one-step ahead forecasting performance. With this specification, we can detect the three recessionary periods experienced by the Iranian economy between 1988...
Business Fluctuations and Cycles, 2007
Duration dependent Markov-switching VAR (DDMS-VAR) models are time series models with data generating process consisting in a mixture of two VAR processes. The switching between the two VAR processes is governed by a two state Markov chain with transition probabilities that depend on how long the chain has been in a state. In the present paper we analyze the second order properties of such models and propose a Markov chain Monte Carlo algorithm to carry out Bayesian inference on the model’s unknowns. Furthermore, a freeware software written by the author for the analysis of time series by means of DDMS-VAR models is illustrated. The methodology and the software are applied to the analysis of the U.S. business cycle.
Exchange Rates and Markov Switching Dynamics
Journal of Business & Economic Statistics, 2005
This article presents a systematic and extensive empirical study on the presence of Markov switching dynamics in three dollar-based exchange rates. A Monte Carlo approach is adopted to circumvent the statistical inference problem inherent to the test of regime-switching behavior. Two data frequencies, two sample periods, and various specifications are considered. Quarterly data yield inconclusive evidence-the test rejects neither random walk nor Markov switching. Monthly data, on the other hand, offer unambiguous evidence of the presence of Markov switching dynamics. The results suggest that data frequency, in addition to sample size, is crucial for determining the number of regimes.
Business cycle analysis with multivariate Markov switching models
2004
The aim of the paper is to describe the cyclical phases of the economy using multivariate Markov switching models. The class of Markov switching models can be extended in two main directions in a multivariate framework. In the first approach, the switching dynamics are introduced by way of one common latent factor . In the second approach, developed by Krolzig (1997), a VAR model with parameters depending on one common Markov chain is considered (MS VAR). We will extend the MS VAR approach allowing for the presence of specific Markov chain in each equation of the VAR (Multiple Markov Switching VAR models, MMS VAR). Dynamic factor models with regime switches, MS VAR and MMS VAR models allow for a multi-country or a multi-sector simultaneous analysis in the search of common phases which are represented by the states of the switching latent factor. Moreover, in the MMS VAR approach we explore the introduction of correlated Markov chains which allow us to evaluate the relationships among phases in different economies or sectors and introduce causality relationships, which allow a more parsimonious representations. We apply the MMS model in order to study the relationship between cyclical phases of the industrial production in the U.S. and Euro zone. Moreover, we construct a MMS model in order to explore the cyclical relationship between the Euro zone industrial production and the industrial component of the European Sentiment Index (ESI).
Bayesian Estimation of Stochastic-Transition Markov-Switching Models for Business Cycle Analysis
2010
We propose a new class of Markov-switching (MS) models for business cycle analysis. As usually done in the literature, we assume that the MS latent factor is driving the dynamics of the business cycle but the transition probabilities can vary randomly over time. Transition probabilities are generated by random processes which may account for the stochastic duration of the regimes and for possible stochastic relations between the MS probabilities and some explanatory variables, such as autoregressive components and exogenous variables. The presence of latent factors and nonlinearities calls for the use of simulation-based inference methods. We propose a full Bayesian inference approach which can be naturally combined with Monte Carlo methods. We discuss the choice of the priors and a Markov-chain Monte Carlo (MCMC) algorithm for estimating the parameters and the latent variables. We provide an application of the model and of the MCMC procedure to data of Euro area. We also carry out a real-time comparison between different models by employing sequential Monte Carlo methods and some concordance statistics, which are widely used in business cycle analysis.