In the present paper we discuss Bayesian estimation of a very general model class where the distribution of the observations is assumed to depend on a latent mixture or switching variable taking values in a discrete state space. This model class covers e.g. finite mixture modelling, Markov switching autoregressive modelling and dynamic linear models with switching. Joint Bayesian estimation of all latent variables, model parameters and parameters determining the probability law of the switching variable is carried out by a new Markov Chain Monte Carlo method called permutation sampling. Estimation of switching and mixture models is known to be faced with identifiability problems as switching and mixture are identifiable only up to permutations of the indices of the states. For a Bayesian analysis the posterior has to be constrained in such a way that identifiablity constraints are fulfilled. The permutation sampler is designed to sample efficiently from the constrained posterior, by first sampling from the unconstrained posterior - which often can be done in a convenient multimove manner - and then by applying a suitable permutation, if the identifiability constraint is violated. We present simple conditions on the prior which ensure that this method is a valid Markov Chain Monte Carlo method (that is invariance, irreducibility and aperiodicity hold). Three case studies are presented, including finite mixture modelling of fetal lamb data, Markov switching Autoregressive modelling of the U.S. quarterly real GDP data, and modelling the U .S./U.K. real exchange rate by a dynamic linear model with Markov switching heteroscedasticity. (author's abstract) / Series: Forschungsberichte / Institut für Statistik

In the present paper we discuss the problem of estimating model likelihoods from the MCMC output for a general mixture and switching model. Estimation is based on the method of bridge sampling (Meng and Wong, 1996), where the MCMC sample is combined with an iid sample from an importance density. The importance density is constructed in an unsupervised manner from the MCMC output using a mixture of complete data posteriors. Whereas the importance sampling estimator as well as the reciprocal importance sampling estimator are sensitive to the tail behaviour of the importance density, we demonstrate that the bridge sampling estimator is far more robust in this concern. Our case studies range from computing marginal likelihoods for a mixture of multivariate normal distributions, testing for the inhomogeneity of a discrete time Poisson process, to testing for the presence of Markov switching and order selection in the MSAR model. (author's abstract) / Series: Report Series SFB "Adaptive Information Systems and Modelling in Economics and Management Science"

This dissertation studies various issues related to regime switching and DSGE models. The methods developed are used to study U.S. business cycles. Chapter one considers and derives the limit distributions of likelihood ratio based tests for Markov regime switching in multiple parameters in the context of a general class of nonlinear models. The analysis simultaneously addresses three difficulties: (1) some nuisance parameters are unidentified under the null hypothesis, (2) the null hypothesis yields a local optimum, and (3) the conditional regime probabilities follow stochastic processes that can only be represented recursively. When applied to US quarterly real GDP growth rates, the tests suggest strong evidence favoring the regime switching specification over a range of sample periods. Chapter two develops a modified likelihood ratio (MLR) test to detect regime switching in state space models. I apply the filtering algorithm introduced in Gordon and Smith (1988) to construct a modified likelihood function under the alternative hypothesis of two regimes and I extend the analysis in Chapter one to establish the asymptotic distribution of the MLR statistic under the null hypothesis of a single regime. I also apply the test to a simple model of the U.S. unemployment rate. This contribution is the first to develop a test based on the likelihood ratio principle to detect regime switching in state space models. The final chapter estimates a search and matching model of the aggregate labor market with sticky price and staggered wage negotiation. It starts with a partial equilibrium search and matching model and expands into a general equilibrium model with sticky price and staggered wage. I study the quantitative implications of the model. The results show that (1) the price stickiness and staggered wage structure are quantitatively important for the search and matching model of the aggregate labor market; (2) relatively high outside option payments to the workers, such as unemployment insurance payments, are needed to match the data; and (3) workers have lower bargaining power relative to firms, which contrasts with the assumption in the literature that workers and firms share equally the surplus generated from their employment relationship.

A relação entre Variáveis Macroeconômicas e o Retorno de Ações é de alta importância para pesquisas econômicas e financeiras, já que, quando descoberto, um mecanismo de conhecer ou prever o impacto dessas variáveis oportuniza uma melhor performance de investidores no mercado acionário. Nesse sentido, nosso trabalho testa nove variáveis macroeconômicas (Preço de Commodities, Taxa de Desemprego, Inflação, Agregados Monetários, Taxas de juros, Relative Money Market Rate (RMM), Produção Industrial, Hiato do Produto (GAP) e Taxa de juros dos EUA) contra o retorno real do Ibovespa, empregando regressões lineares, como tradicional na literatura, e modelos de mudança de regime markoviana (MSM), para avaliar melhor o impacto e poder de previsão do retorno sob uma economia tão perturbada por planos econômicos e crises financeiras. Além disso, realizamos uma rigorosa avaliação do poder preditivo através de testes dentro e fora da amostra, incluindo avaliações dos coeficientes estimados defasados, critérios de Informação de AIC e BIC, Razões de Erro Quadrático Médio e o Erro Absoluto Médio e testes de encompassing de Diebold e Mariano (1995), de Clark e Mccracken (2001) e de Mccracken (2007), combinados aos novos valores assintóticos de Clark e Mccracken (2001,2004). Os resultados indicam que o Ibovespa possui dois regimes, e que a variável Hiato do Produto se destaca por ser a variável mais significativa e de maior poder de previsão, tanto nos modelos lineares como nos nãolineares. Além dessa, a variável RMM, também se mostrou capacitada para prever o retorno quando estimada no MSM, assim como as variáveis inflação e agregados monetários também apresentaram poder preditivo quando acompanhados da variável GAP. Entretanto, Produção industrial e taxa de juros não tiveram qualquer evidência de capacidade preditiva. Por fim, nos horizontes trimestrais e semestrais, os MSM tiveram dificuldade de encontrar os diferentes regimes, e por isso, não conseguiram se mostrar sistematicamente superiores aos modelos lineares. / The relationship between Macroeconomic Variables and stock returns is of high importance for economic and financial research because, when discovered, a mechanism to know or predict the impact of these variables allows a better performance of investors in the stock market In this sense, our research tests nine macroeconomic variables (Commodities Prices, Unemployment Rate, Inflation, Money Stock, Interest Rate, Relative Money Market Rate (RMM), Industrial Production, Output Gap (GAP) and United States Interest Rate) versus the Ibovespa Real Stock Return, with linear models, as in traditional literature, and with Markov Switching Models, to gauge the impact and the predictive power of the assumption of an economy so troubled by economic plans and financial crises. In addition, we conducted a rigorous predictive ability evaluation by testing in-sample and out-of-sample, including a lagged coefficient estimated evaluation, information criteria of Akaike and Schwarz, Mean-square Error, Absolute Mean Error and encompassing tests of Diebold e Mariano (1995), Clark e Mccracken (2001) and Mccracken (2007) combined with the new asymptotic values of Clark e Mccracken (2001,2004). The results indicated that the Ibovespa has two states and the Output Gap variable stands out for being the most significant variable and with the greatest predictive ability for both linear and nonlinear models. Besides, the RMM variable has also shown to be able to predict the stock return when estimated in the MSM. Furthermore, the inflation and money stock variable also presents predict ability when estimated models is addicted with GAP variable. Industrial production and interest rates had no evidence of predictive ability. Finally, in the quarterly and semiannual horizons, the MSM had difficulty in finding the different regimes, and therefore failed to show themselves consistently higher than the linear models.

Submitted by Jordano Vieira Rocha (jordanorocha@hotmail.com) on 2015-04-30T08:48:24Z No. of bitstreams: 1 Dissertação - Jordano Vieira Rocha.pdf: 1057882 bytes, checksum: 1ba84113f5ec0c31d9c99f3bebe4714d (MD5) / Approved for entry into archive by Suzinei Teles Garcia Garcia (suzinei.garcia@fgv.br) on 2015-04-30T13:02:56Z (GMT) No. of bitstreams: 1 Dissertação - Jordano Vieira Rocha.pdf: 1057882 bytes, checksum: 1ba84113f5ec0c31d9c99f3bebe4714d (MD5) / Made available in DSpace on 2015-04-30T17:23:54Z (GMT). No. of bitstreams: 1 Dissertação - Jordano Vieira Rocha.pdf: 1057882 bytes, checksum: 1ba84113f5ec0c31d9c99f3bebe4714d (MD5) Previous issue date: 2015-04-07 / This work assesses the forecasts of three nonlinear methods — Markov Switching Autoregressive Model, Logistic Smooth Transition Autoregressive Model, and Autometrics with Dummy Saturation — for the Brazilian monthly industrial production and tests if they are more accurate than those of naive predictors such as the autoregressive model of order p and the double differencing device. The results show that the step dummy saturation and the logistic smooth transition autoregressive can be superior to the double differencing device, but the linear autoregressive model is more accurate than all the other methods analyzed. / Este trabalho avalia as previsões de três métodos não lineares — Markov Switching Autoregressive Model, Logistic Smooth Transition Autoregressive Model e Autometrics com Dummy Saturation — para a produção industrial mensal brasileira e testa se elas são mais precisas que aquelas de preditores naive, como o modelo autorregressivo de ordem p e o mecanismo de double differencing. Os resultados mostram que a saturação com dummies de degrau e o Logistic Smooth Transition Autoregressive Model podem ser superiores ao mecanismo de double differencing, mas o modelo linear autoregressivo é mais preciso que todos os outros métodos analisados.

Submitted by Wellington Gonçalves Rodrigues (wgrodrig@gmail.com) on 2015-08-25T05:50:39Z No. of bitstreams: 1 Dissertação_Wellington_Rodrigues_Final.pdf: 461618 bytes, checksum: 3cd66ac5a3470a56ed4bc6b0442878ca (MD5) / Approved for entry into archive by Renata de Souza Nascimento (renata.souza@fgv.br) on 2015-08-25T16:26:57Z (GMT) No. of bitstreams: 1 Dissertação_Wellington_Rodrigues_Final.pdf: 461618 bytes, checksum: 3cd66ac5a3470a56ed4bc6b0442878ca (MD5) / Made available in DSpace on 2015-08-25T16:59:05Z (GMT). No. of bitstreams: 1 Dissertação_Wellington_Rodrigues_Final.pdf: 461618 bytes, checksum: 3cd66ac5a3470a56ed4bc6b0442878ca (MD5) Previous issue date: 2015 / The goal of this paper is to identify the occurrence, duration and transition probabilities of different monetary policy regimes in Brazil since the implementation of the inflation-targeting regime in 1999. To estimate the reaction function of the Central Bank of Brazil, a forward looking Taylor Rule for an open economy is adopted and a Markov Regime Switching methodology applied in order to allow for endogenous regime switches in monetary policy. The results indicate the presence of three distinct monetary policy regimes since the implementation of the inflation-targeting regime in Brazil. The first regime occurs during 21% of the studied period and is characterized by a discretionary approach not following the Taylor Rule and a stronger focus in the reaction to the output gap. A balance in the weights given to the output gap and inflation expectations deviation from target and the adherence to the Taylor Rule characterizes the second regime, which is present in 67% of the time. The third regime is characterized not only by its adherence to the Taylor Rule but also by a stronger focus in the reaction to the deviation of inflation expectations from the target, being present in 12% of time. / O presente trabalho busca identificar a ocorrência, duração e probabilidades de transição de diferentes regimes na condução da política monetária no Brasil a partir da implantação do sistema de metas de inflação em 1999. A estimação da função de reação do Banco Central do Brasil é realizada a partir de uma Regra de Taylor forward looking para uma economia aberta, onde utilizamos a metodologia Markov Regime Switching para caracterizar de forma endógena os diferentes regimes de política monetária. Os resultados obtidos indicam a ocorrência de três regimes distintos de política monetária a partir da implantação do sistema de metas de inflação no Brasil. O primeiro regime ocorre durante 21% do período estudado e se caracteriza pela não aderência ao princípio de Taylor e discricionariedade da autoridade monetária, que reage demonstrando maior sensibilidade ao hiato do produto. O segundo regime é o de maior duração, ocorre durante 67% do período estudado, e se caracteriza pela aderência ao princípio de Taylor e equilíbrio nos pesos atribuídos pelo Banco Central tanto ao hiato do produto como ao desvio das expectativas de inflação com relação à meta. Já o terceiro regime ocorre durante 12% do período estudado e se caracteriza não somente pela aderência ao princípio de Taylor, como também por uma maior aversão ao desvio das expectativas de inflação com relação à meta.

Submitted by EDUARDO HINTERHOLZ (eduh17@gmail.com) on 2015-08-26T19:57:33Z No. of bitstreams: 1 DissertaçãoFinal.pdf: 1431279 bytes, checksum: dea2c0cdc148ed945cdfc8b33e86f668 (MD5) / Approved for entry into archive by Suzinei Teles Garcia Garcia (suzinei.garcia@fgv.br) on 2015-08-26T20:02:31Z (GMT) No. of bitstreams: 1 DissertaçãoFinal.pdf: 1431279 bytes, checksum: dea2c0cdc148ed945cdfc8b33e86f668 (MD5) / Made available in DSpace on 2015-08-27T13:12:19Z (GMT). No. of bitstreams: 1 DissertaçãoFinal.pdf: 1431279 bytes, checksum: dea2c0cdc148ed945cdfc8b33e86f668 (MD5) Previous issue date: 2015 / This work proposes a method to examine variations in the cointegration relation between preferred and common stocks in the Brazilian stock market via Markovian regime switches. It aims on contributing for future works in 'pairs trading' and, more specifically, to price discovery, given that, conditional on the state, the system is assumed stationary. This implies there exists a (conditional) moving average representation from which measures of 'information share' (IS) could be extracted. For identification purposes, the Markov error correction model is estimated within a Bayesian MCMC framework. Inference and capability of detecting regime changes are shown using a Montecarlo experiment. I also highlight the necessity of modeling financial effects of high frequency data for reliable inference. / Este trabalho propõe um método para examinar variações na relação cointegração de preços de ações preferenciais e ordinárias da bolsa brasileira através de mudanças de regime no sentido de Markov. Este modelo tem como objetivo contribuir tanto para futuros trabalhos em negociações de pares ('pairs trading') quanto, principalmente, para aplicação em descoberta de preços visto que, condicional nos estados, é pressuposta estacionariedade no sistema. Desta maneira seria possível a extração de medidas de 'parcela de informação' (IS) baseadas na representação de médias móveis de um modelo de correção de erros Markoviano, estimado através de um ferramental bayesiano do tipo MCMC por questões de identificação. A validade do modelo no sentido de capturar as variações de regime é demonstrada através de experimento de Montecarlo, bem como é evidenciada a necessidade da modelar não normalidades na distribuição dos dados de alta frequência visando inferência.

Este trabalho tem objetivo de estudar os ciclos de crédito em cinco países da América Latina – Brasil, Chile, Colômbia, México e Peru - usando modelos com mudança de regime markoviano univariados e multivariados. Alguns dos modelos são capazes de captar períodos de crises bancárias nos países individualmente datados em Laeven e Valencia (2008, 2012) e Reinhart e Rogoff (2008), enquanto o modelo multivariado capta uma dinâmica comum nos países estudados. O ciclo que o modelo multivariado revela está de acordo com conhecidos períodos de expansão e contração da taxa de crescimento do crédito real ao setor privado conhecidos na literatura, em especial o boom da primeira metade da década de 1990 e sua desaceleração subseqüente. / This paper aims to study credit cycles in five Latin American countries in a Markov Switching Approach with univariate and multivariate models. The univariate models, for some countries, identified periods of banking crises dated in Laeven and Valencia (2008; 2012) and Reinhart and Rogoff (2008) while the multivariate model captured a common dynamic in those countries studied. The cycle revealed with this model is in accordance with known periods of expansion and contraction of the growth rate credit in Latin America, in special the early 1990’s boom and it’s subsequent slowdown.

O objetivo desta dissertação é avaliar o comportamento dos principais parâmetros da economia brasileira através da estimação de um modelo DSGE (Dynamic Stochastic General Equilibrium) de economia aberta usando métodos bayesianos e permitindo mudanças de regime markovianas de determinados parâmetros. Utilizando o modelo DSGE desenvolvido por Justiniano e Preston (2010) e o método de solução do modelo Markov Switching DSGE (MS-DSGE) proposto por Farmer et al. (2008), este trabalho encontrou superioridade nos ajustes dos dados dos modelos que incorporaram mudanças markovianas, rejeitando a hipótese de parâmetros constantes em modelos DSGE para a economia brasileira. / The goal of this dissertation is to evaluate the behaviour of the main parameters of the Brazilian economy through the estimation of a DSGE (Dynamic Stochastic General Equilibrium) model of open economy using Bayesian methods and allowing Markov switching of certain parameters. Using the DSGE model developed by Justiniano and Preston (2010) and the method of solution of the Markov Switching DSGE (MS-DSGE) model proposed by Farmer et al. (2008), this work found superiority in the settings of the data of the models that incorporated Markov switching, rejecting the hypothesis of constant parameters in DSGE models for the Brazilian economy.

A relação entre Variáveis Macroeconômicas e o Retorno de Ações é de alta importância para pesquisas econômicas e financeiras, já que, quando descoberto, um mecanismo de conhecer ou prever o impacto dessas variáveis oportuniza uma melhor performance de investidores no mercado acionário. Nesse sentido, nosso trabalho testa nove variáveis macroeconômicas (Preço de Commodities, Taxa de Desemprego, Inflação, Agregados Monetários, Taxas de juros, Relative Money Market Rate (RMM), Produção Industrial, Hiato do Produto (GAP) e Taxa de juros dos EUA) contra o retorno real do Ibovespa, empregando regressões lineares, como tradicional na literatura, e modelos de mudança de regime markoviana (MSM), para avaliar melhor o impacto e poder de previsão do retorno sob uma economia tão perturbada por planos econômicos e crises financeiras. Além disso, realizamos uma rigorosa avaliação do poder preditivo através de testes dentro e fora da amostra, incluindo avaliações dos coeficientes estimados defasados, critérios de Informação de AIC e BIC, Razões de Erro Quadrático Médio e o Erro Absoluto Médio e testes de encompassing de Diebold e Mariano (1995), de Clark e Mccracken (2001) e de Mccracken (2007), combinados aos novos valores assintóticos de Clark e Mccracken (2001,2004). Os resultados indicam que o Ibovespa possui dois regimes, e que a variável Hiato do Produto se destaca por ser a variável mais significativa e de maior poder de previsão, tanto nos modelos lineares como nos nãolineares. Além dessa, a variável RMM, também se mostrou capacitada para prever o retorno quando estimada no MSM, assim como as variáveis inflação e agregados monetários também apresentaram poder preditivo quando acompanhados da variável GAP. Entretanto, Produção industrial e taxa de juros não tiveram qualquer evidência de capacidade preditiva. Por fim, nos horizontes trimestrais e semestrais, os MSM tiveram dificuldade de encontrar os diferentes regimes, e por isso, não conseguiram se mostrar sistematicamente superiores aos modelos lineares. / The relationship between Macroeconomic Variables and stock returns is of high importance for economic and financial research because, when discovered, a mechanism to know or predict the impact of these variables allows a better performance of investors in the stock market In this sense, our research tests nine macroeconomic variables (Commodities Prices, Unemployment Rate, Inflation, Money Stock, Interest Rate, Relative Money Market Rate (RMM), Industrial Production, Output Gap (GAP) and United States Interest Rate) versus the Ibovespa Real Stock Return, with linear models, as in traditional literature, and with Markov Switching Models, to gauge the impact and the predictive power of the assumption of an economy so troubled by economic plans and financial crises. In addition, we conducted a rigorous predictive ability evaluation by testing in-sample and out-of-sample, including a lagged coefficient estimated evaluation, information criteria of Akaike and Schwarz, Mean-square Error, Absolute Mean Error and encompassing tests of Diebold e Mariano (1995), Clark e Mccracken (2001) and Mccracken (2007) combined with the new asymptotic values of Clark e Mccracken (2001,2004). The results indicated that the Ibovespa has two states and the Output Gap variable stands out for being the most significant variable and with the greatest predictive ability for both linear and nonlinear models. Besides, the RMM variable has also shown to be able to predict the stock return when estimated in the MSM. Furthermore, the inflation and money stock variable also presents predict ability when estimated models is addicted with GAP variable. Industrial production and interest rates had no evidence of predictive ability. Finally, in the quarterly and semiannual horizons, the MSM had difficulty in finding the different regimes, and therefore failed to show themselves consistently higher than the linear models.