Esta tese é composta por três ensaios sobre a estimação bayesiana de modelos DSGE com fricções financeiras para o Brasil. O primeiro ensaio tem o objetivo de analisar como a incorporação de intermediários financeiros num modelo DSGE influenciam na análise do ciclo econômico, bem como uma política de crédito pode ser utilizada para mitigar os choques no mercado de crédito sobre a atividade. O governo brasileiro expandiu o crédito na economia através das instituições financeiras públicas tendo como custo o aumento da dívida pública. Para isso, foi estimado um modelo inspirado em Gertler e Karadi (2011) para avaliar o comportamento da economia brasileira sob a influência de uma política de crédito. Política de crédito mostrou-se efetiva para mitigar os efeitos recessivos de uma crise financeira que atinja a cotação dos ativos privados ou o patrimônio das instituições financeiras. Contudo, a política monetária tradicional se mostrou mais eficiente para a estabilização da inflação em momentos de normalidade. O segundo ensaio consiste na estimação de um modelo DSGE-VAR para a economia brasileira. A parte DSGE consiste em uma economia pequena, aberta e com fricções financeiras na linha de Gertler, Gilchrist e Natalucci (2007). A estimação do modelo indicou que flexibilização do espaço paramétrico possibilitado pelo modelo DSGE-VAR proporcionou ganhos em relação ao ajuste aos dados em relação a modelos alternativos. O exercício também obteve indicações de que os choques externos apresentam impactos significativos no patrimônio e no endividamento das firmas domésticas. Esse resultado fortalece a evidência de que um canal importante de transmissão dos movimentos da economia mundial para a o Brasil ocorre através das firmas. Por fim, no terceiro ensaio tem como foco a transmissão dos choques no spread de crédito bancário para as demais variáveis da economia e suas implicações para a condução da política monetária no brasil. Para isso, foi estimado um modelo DSGE com fricções financeiras para a economia brasileira. O modelo é baseado em Cúrdia e Woodford (2010), que propuseram uma extensão do modelo de Woodford (2003) para incorporar a existência de um diferencial entre a taxa de juros disponíveis aos poupadores e tomadores de empréstimos, que pode variar por razões tanto endógenas quanto exógenos. Nessa economia, a política monetária pode responder não somente às variações na taxa de inflação e hiato do produto através de uma regra simples, como também por meio de uma regra ajustada pelo spread de crédito da economia. Os resultados mostram que a inclusão do spread de crédito no modelo Novo Keynesiano não altera significativamente as conclusões dos modelos DSGE em respostas a perturbações exógenas tradicionais, como choques na taxa de juros, na produtividade da economia e no dispêndio público. Porém, nos eventos que ocasionam a deterioração da intermediação financeira, por meio de choques exógenos sobre o spread de crédito, o impacto sobre o ciclo econômico foi significativo e a adoção de uma regra de política monetária ajustada pelo spread pode conseguir estabilizar a economia mais rapidamente do que uma regra tradicional. / The present thesis is a collection of three essays on Bayesian estimation of DSGE models with financial frictions in the Brazilian economy. The first essay intends to investigate how the incorporation of financial intermediaries in a DSGE model influences the analysis of the economic cycle, as well as how the credit policy can be employed to mitigate the effects of shocks in the credit market on the economic activity. The Brazilian government expanded the credit in the economy through public financial institutions, which resulted in an increase of public debt. it estimated a model inspired by Gertler and Karadi (2011) to evaluate the performance of the Brazilian economy under the influence of a credit policy. Credit policy was effective to mitigate the recessionary effects of a financial crisis that affects the valuation of private assets and the net worth of financial institutions. However, the traditional monetary policy was more efficient for the stabilization of inflation in times of normality. The second essay consist of a DSGE-VAR model for the Brazilian economy. The DSGE model was estimated for a small, open economy with financial frictions, in line with Gertler, Gilchrist and Natalucci (2007). The results indicates that the estimation of DSGE-VAR provides an advantage for the data fitting in comparison to alternative models. In addition, the results indicate that external shocks have significant impacts in the equity and debt of domestic firms. This result strengthens (supports) the evidence that an important channel of transmission of the movements of the world economy for the Brazil takes place through productive sector. The third essay analyze the transmission of shocks in the banking credit spread for the other variables of the economy and its implications for the conduct of monetary policy in Brazil. We do so by estimating a DSGE model with financial frictions for the Brazilian economy. The model is based on Cúrdia and Woodford (2010), who proposed an extension of the model Woodford (2003) to incorporate the existence of a differential between the interest rates available to savers and borrowers, which can vary by both endogenous and exogenous reasons. In this model, monetary policy can respond not only to changes in the inflation rate and output gap through a simple rule, but also through a rule set by the credit spread of the economy. The results show that the inclusion of credit spread in the New Keynesian model does not significantly changes the conclusions of DSGE models in traditional responses to exogenous shocks, such as shocks in the interest rate, in the productivity of the economy and in public spending. However, in the events that cause the deterioration of financial intermediation through exogenous shocks on the credit spread, the impact on the business cycle was significant and the adoption of a monetary policy rule set by the spread can achieve a faster stabilization of the economy than a traditional rule.

Esta tese é composta por três ensaios sobre a estimação bayesiana de modelos DSGE com fricções financeiras para o Brasil. O primeiro ensaio tem o objetivo de analisar como a incorporação de intermediários financeiros num modelo DSGE influenciam na análise do ciclo econômico, bem como uma política de crédito pode ser utilizada para mitigar os choques no mercado de crédito sobre a atividade. O governo brasileiro expandiu o crédito na economia através das instituições financeiras públicas tendo como custo o aumento da dívida pública. Para isso, foi estimado um modelo inspirado em Gertler e Karadi (2011) para avaliar o comportamento da economia brasileira sob a influência de uma política de crédito. Política de crédito mostrou-se efetiva para mitigar os efeitos recessivos de uma crise financeira que atinja a cotação dos ativos privados ou o patrimônio das instituições financeiras. Contudo, a política monetária tradicional se mostrou mais eficiente para a estabilização da inflação em momentos de normalidade. O segundo ensaio consiste na estimação de um modelo DSGE-VAR para a economia brasileira. A parte DSGE consiste em uma economia pequena, aberta e com fricções financeiras na linha de Gertler, Gilchrist e Natalucci (2007). A estimação do modelo indicou que flexibilização do espaço paramétrico possibilitado pelo modelo DSGE-VAR proporcionou ganhos em relação ao ajuste aos dados em relação a modelos alternativos. O exercício também obteve indicações de que os choques externos apresentam impactos significativos no patrimônio e no endividamento das firmas domésticas. Esse resultado fortalece a evidência de que um canal importante de transmissão dos movimentos da economia mundial para a o Brasil ocorre através das firmas. Por fim, no terceiro ensaio tem como foco a transmissão dos choques no spread de crédito bancário para as demais variáveis da economia e suas implicações para a condução da política monetária no brasil. Para isso, foi estimado um modelo DSGE com fricções financeiras para a economia brasileira. O modelo é baseado em Cúrdia e Woodford (2010), que propuseram uma extensão do modelo de Woodford (2003) para incorporar a existência de um diferencial entre a taxa de juros disponíveis aos poupadores e tomadores de empréstimos, que pode variar por razões tanto endógenas quanto exógenos. Nessa economia, a política monetária pode responder não somente às variações na taxa de inflação e hiato do produto através de uma regra simples, como também por meio de uma regra ajustada pelo spread de crédito da economia. Os resultados mostram que a inclusão do spread de crédito no modelo Novo Keynesiano não altera significativamente as conclusões dos modelos DSGE em respostas a perturbações exógenas tradicionais, como choques na taxa de juros, na produtividade da economia e no dispêndio público. Porém, nos eventos que ocasionam a deterioração da intermediação financeira, por meio de choques exógenos sobre o spread de crédito, o impacto sobre o ciclo econômico foi significativo e a adoção de uma regra de política monetária ajustada pelo spread pode conseguir estabilizar a economia mais rapidamente do que uma regra tradicional. / The present thesis is a collection of three essays on Bayesian estimation of DSGE models with financial frictions in the Brazilian economy. The first essay intends to investigate how the incorporation of financial intermediaries in a DSGE model influences the analysis of the economic cycle, as well as how the credit policy can be employed to mitigate the effects of shocks in the credit market on the economic activity. The Brazilian government expanded the credit in the economy through public financial institutions, which resulted in an increase of public debt. it estimated a model inspired by Gertler and Karadi (2011) to evaluate the performance of the Brazilian economy under the influence of a credit policy. Credit policy was effective to mitigate the recessionary effects of a financial crisis that affects the valuation of private assets and the net worth of financial institutions. However, the traditional monetary policy was more efficient for the stabilization of inflation in times of normality. The second essay consist of a DSGE-VAR model for the Brazilian economy. The DSGE model was estimated for a small, open economy with financial frictions, in line with Gertler, Gilchrist and Natalucci (2007). The results indicates that the estimation of DSGE-VAR provides an advantage for the data fitting in comparison to alternative models. In addition, the results indicate that external shocks have significant impacts in the equity and debt of domestic firms. This result strengthens (supports) the evidence that an important channel of transmission of the movements of the world economy for the Brazil takes place through productive sector. The third essay analyze the transmission of shocks in the banking credit spread for the other variables of the economy and its implications for the conduct of monetary policy in Brazil. We do so by estimating a DSGE model with financial frictions for the Brazilian economy. The model is based on Cúrdia and Woodford (2010), who proposed an extension of the model Woodford (2003) to incorporate the existence of a differential between the interest rates available to savers and borrowers, which can vary by both endogenous and exogenous reasons. In this model, monetary policy can respond not only to changes in the inflation rate and output gap through a simple rule, but also through a rule set by the credit spread of the economy. The results show that the inclusion of credit spread in the New Keynesian model does not significantly changes the conclusions of DSGE models in traditional responses to exogenous shocks, such as shocks in the interest rate, in the productivity of the economy and in public spending. However, in the events that cause the deterioration of financial intermediation through exogenous shocks on the credit spread, the impact on the business cycle was significant and the adoption of a monetary policy rule set by the spread can achieve a faster stabilization of the economy than a traditional rule.

Submitted by isabela.moljf@hotmail.com (isabela.moljf@hotmail.com) on 2016-08-10T12:49:00Z No. of bitstreams: 1 eldertiagodacostasouza.pdf: 2456774 bytes, checksum: c3b48a393ed4f366bf8bd81d38993b9d (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2016-08-10T13:03:58Z (GMT) No. of bitstreams: 1 eldertiagodacostasouza.pdf: 2456774 bytes, checksum: c3b48a393ed4f366bf8bd81d38993b9d (MD5) / Made available in DSpace on 2016-08-10T13:03:58Z (GMT). No. of bitstreams: 1 eldertiagodacostasouza.pdf: 2456774 bytes, checksum: c3b48a393ed4f366bf8bd81d38993b9d (MD5) Previous issue date: 2016-03-03 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / O principal objetivo desta dissertação é estudar os efeitos da interação entre as políticas fiscal e monetária sobre as variáveis macroeconômicas da economia brasileira. Para tal analisou-se a significância dos diferentes modelos de DSGE, por meio da metodologia DSGE-VAR, que implica na estimação do parâmetro de ajustamento do modelo (lambda), conforme Del Negro e Schorfheide (2004, 2006, 2009). Os resultados mostram que o modelo DSGE Bayesiano, com preços rígidos, setor externo e com a interação entre as políticas, é aquele que tem a melhor aderência aos dados reais. Destarte, foram estimadas as funções impulso-resposta, que corroboram a importância da inter-relação entre as políticas fiscal e monetária. A partir daí, avaliou-se o regime de dominância praticado no Brasil, sob a metodologia proposta por Leeper (1991). A combinação dos valores dos parâmetros resulta na proposição de dominância monetária, no período que compreende o primeiro trimestre de 2002 e o terceiro trimestre de 2015. Ademais, estudou-se o regime de dominância, relacionado ao período anterior e posterior à crise do subprime. Os resultados mostram que o regime de dominância não foi alterado. No entanto, a política fiscal tem se tornado mais ativa.Verificou-se que no período pós crise, tanto os choque de política fiscal quanto os de política monetária foram maiores, ou seja, intensificou-se o uso dos instrumentos fiscais. / The main aim of this dissertation is to study the effects of the interaction between fiscal and monetary policies on the macroeconomic variables of the Brazilian economy. Therefore we analyzed the significance of different DSGE models through the DSGE-VAR method, which involves the estimation of lambda model fitting parameter, as Del Negro and Schorfheide (2004, 2006, 2009). The results show that Bayesian DSGE models with sticky prices, external sector and the interaction between policies, is one that has the best grip to the actual data. Thus, were estimated the impulse response functions, which confirm the importance of the interrelationship between fiscal and monetary policies. From there, we evaluated the dominance regime practiced in Brazil, under the method proposed by Leeper (1991). The parameter values combination results in monetary dominance proposition in the period from the first quarter of 2002 and the third quarter of 2015. Besides, we estudied the dominance regime related to period prior and posterior to the subprime crisis. The results show that the dominance regime has not change. However, fiscal policy has became more active. It was found that in the post crisis period, both the shock fiscal policy as the monetary policy were higher, i.e., intensified the use of fiscal instruments.

In recent years the New Keynesian framework has become widely used to identify the relationship between monetary policy, inflation, the business cycle and welfare. Most commonly in these models inertia in prices are introduced only through the aggregate supply side which generates a short run non-neutrality of money. This thesis begins with an investigation into the impact of sticky prices on the macroeconomic equilibrium through aggregate demand. We show that in models of price stickiness among differentiated goods aggregate consumption deviates from the conventional Euler equation due to relative price distortions. This has some non-negligible implications: there are additional inflation effects, which enter through aggregate demand, that lower the response of the marginal cost and dampen responses of inflation and output; products' price elasticity of demand affects equilibrium output and inflation dynamics independently of supply factors; monetary policy responses are smoother than in the conventional new Keynesian models, particularly the more competitive are the products markets. In chapter 2 we continue with an investigation into the impact that the aforementioned channel has on welfare and monetary policy under various regimes. Specifically, we compare our results with the benchmark New Keynesian model with a cost channel for alternative levels of competition in the goods market. When the central bank is assumed to follow a Taylor rule we find, contrary to the standard New Keynesian literature, that welfare losses ultimately fall as the goods market becomes more competitive. Furthermore, there are additional adverse implications for welfare coming through an exaggerated stabilisation bias associated with discretionary policy in our model version. A move to optimal commitment implies significant additional gains compared to the standard literature by; eliminating this amplified stabilisation bias and; reducing further the fall in output gap and inflation fluctuations at the time of shock. The final part of this thesis develops a Generalised Taylor economy to include a financial market. This finance sector is characterised by savings contracts to households and loan contracts to firms, both of which are differentiated by the duration for which their interest rate remains fixed. Additionally, a time varying external finance premium on loan rates is introduced through an endogenous probability of firm default. Using break-even conditions we show that the fixed markup on loan rates is dependent on, the expected default risk over the lifetime of the contract, and, spillovers from the unexpected losses of current "locked in" financial contracts that must be accounted for in the zero profit condition of the commercial bank. Our results indicate that inertia in loan and savings rates dampens the responses of monetary policy and the business cycle whilst generating a procylical loan rate spread. In contrast, risk of default amplifies the business cycle and delivers a countercyclical loan rate spread. The overall impact of these two channels on the direction and magnitude of loan rate spreads, spillovers to new contracts and the dynamics of the business cycle, are shown depend on the type of shock hitting the economy.

Chapter 1- How well-off or worse-off a country can be by joining a currency union in the presence of structural heterogeneity and idiosyncratic shocks? In light of the proposed creation of a currency union for the Economic Community of the West African States (ECOWAS), we develop a three-region DSGE model to explore the question. We divide the ECOWAS into three regions-Nigeria, the existing WAEMU (West-African Economic Monetary Union), and the rest. Considering two monetary regimes (monetary union and monetary independence), we assess the heterogeneity in the responses to country-specific productivity and terms-of-trade shocks in these two regimes, as well as the costs related to the loss of monetary independence. Our results indicate that shocks hitting a given region generate cross-border spillover effects, whose sign and magnitude depend not only on the nature of the disturbance but also on its origin and on the monetary policy regime considered. Moreover, the propagation of shocks across regions is magnified under the monetary union regime. Shocks hitting Nigeria's economy tend to have a more destabilizing effect on the other regions, especially when they are inside the union. Our results also suggest that the proposed monetary union for the ECOWAS region can potentially lead to welfare improvement for all the members, but the magnitude of the welfare gain is relatively small. Chapter 2- In this chapter, we develop a multi-region New-Keynesian Dynamic Stochastic General Equilibrium (DSGE) of the West-African countries to provide a quantitative analysis of intergovernmental fiscal transfers in the context of the proposed creation of a monetary union. We assess the potential role of fiscal transfers in the stabilization of business cycle fluctuations in the projected monetary union in the presence of idiosyncratic shocks. Starting from a baseline scenario with no fiscal transfers among the regions, we analyze the dynamic and welfare impacts of full and partial fiscal equalization schemes with nominal tax revenue sharing within the union. We consider adverse productivity and term-of-trade shocks. Our simulation results suggest that the transfer mechanism is an efficient stabilizing tool. However, the stabilization property of the fiscal transfer system hinges upon the full or partial nature of the compensation system. Moreover, the ability of the transfer system to absorb the negative effects of idiosyncratic shocks depends not only on the type of shock but also on the size of the region directly affected. Chapter 3- We analyze in this chapter the macroeconomics effects of fiscal policy shocks in the Economic Community of West African States (ECOWAS). To that end, we use a Global Vector Autoregression (GVAR) model, which allows us to assess both the within country and the cross borders spillover effects of the fiscal shocks. For the dynamic analysis, we consider negative country-specific public spending and revenue shocks affecting Nigeria as well as regional public spending and revenue shocks affecting two groups of countries in the area, namely the West African Economic and Monetary Union (WAEMU) and the Rest of ECOWAS (RECOWAS). We provide evidence of considerable cross-country heterogeneity in fiscal spillovers; for instance, spillovers are high for fiscal shocks affecting Nigeria, while the cross-border spillover effects on Nigeria are weak for shocks affecting WAEMU and RECOWAS. Our results also suggest that fiscal policy is very relevant in stimulating real output in each of the ECOWAS countries but limited for the cross-country output stimulation.

In the aftermath of the recent financial crisis, an environment of historically low interest rates and extensive household indebtedness in the OECD countries have triggered a vivid debate on whether central banks should react to house price fluctuations in their pursuit of monetary policy. In Sweden, a period of low policy rates and house price inflation was halted when the central bank increased the interest rates in 2010. This paper studies whether the Riksbank reacted to house prices in 1993-2013. Using Bayesian methods and quarterly data, I estimate a DSGE model with patient and impatient households, where the central bank reacts to house price inflation. The results suggest that the Riksbank did respond to house prices during the sample period. The findings are robust and plausible from an economic point of view.

The primary objective of this is to confront the DSGE model (Ireland, 2011) with data in an attempt to evaluate its empirical adequacy. The perspective used for this evaluation is based on unveiling the statistical model (structural VAR) behind the DSGE model, with a view to test its probabilistic assumptions vis-a-vis the data. It is shown that the implicit statistical model is seriously misspecified and the information from mis-specification (M-S) testing is then used to respecify the original structural VAR in an attempt to achieve statistical adequacy. The latter provides a precondition for the reliability of any inference based on the statistical model. Once the statistical adequacy of the respecified model is secured through thorough M-S testing, inferences like the likelihood-ratio test for the overidentifying restrictions, forecasting, impulse response analysis are applied to the original DSGE model to evaluate its empirical adequacy. At the end, the same inferential procedure is applied to the CAPM model. / Ph. D.

京都大学 / 0048 / 新制・課程博士 / 博士(経済学) / 甲第18036号 / 経博第489号 / 新制||経||268(附属図書館) / 30894 / 京都大学大学院経済学研究科経済学専攻 / (主査)教授 柴田 章久, 教授 中嶋 智之, 准教授 敦賀 貴之 / 学位規則第4条第1項該当 / Doctor of Economics / Kyoto University / DFAM

Tableau d'honneur de la Faculté des études supérieures et postdoctorales, 2006-2007 / Ce mémoire développe un modèle dynamique stochastique d'équilibre général suivant les traces de Ireland (2001), Dib (2003) et Dib, Gammoudi et Moran (2005). Le modèle est estimé à partir de séries chronologiques canadiennes via des techniques économétriques bayésiennes qui consistent en la simulation des paramètres du modèle à partir de leur densité a posteriori à l'aide de l'algorithme Metropolis-Hastings. Un test de stabilité du modèle est ensuite effectué en comparant ses prévisions hors-échantillon lorsqu'il est estimé avec deux échantillons différents. Le premier échantillon de données va de 1981T1 à 2000T4, tandis que le second débute en 1991T1 afin de tenir compte de l'introduction du régime de cible d'inflation par la Banque du Canada. Nous obtenons deux résultat importants. Premièrement, les valeurs estimées des paramètres diffèrent de manière significatives selon que le premier ou le second échantillon est utilisé. Deuxièmement, la capacité de prévision du modèle ne dépend pas de l'échantillon ayant servis à l'estimation. / This thesis develops a dynamic stochastic general equilibrium model (DSGE) in the line of Ireland (2001), Dib (2003), and Dib, Gammoudi, and Moran (2005). The model is estimated with Canadian time series via Bayesian techniques by combining the likelihood of its state-space representation with prior information and simulating parameter values from their posterior density using the Metropolis-Hastings algorithm. A stability test is then performed on the model by comparing its out-of-sample forecasting ability when estimated on two different samples. The first sample runs from 1981Q1 to 2000Q4, whereas the second starts in 1991Q1 to take into account the inflation targeting regime introduced by the Bank of Canada. Our main finding is that although the parameter estimates related to the monetary policy change significantly following the policy shift, the model's forecasting ability remains unaffected.

We apply the data cloning method developed by Lele et al. (2007) to estimate the model of Smets and Wouters (2007). The data cloning algorithm is a numerical method that employs replicas of the original sample to approximate the maximum likelihood estimator as the limit of Bayesian simulation-based estimators. We also analyze the identification properties of the model. We measure the individual identification strength of each parameter by observing the posterior volatility of data cloning estimates, and access the identification problem globally through the maximum eigenvalue of the posterior data cloning covariance matrix. Our results indicate that the model is only poorly identified. The system displays bad global identification properties, and most of its parameters seem locally ill-identified. / Neste trabalho aplicamos o método data cloning de Lele et al. (2007) para estimar o modelo de Smets e Wouters (2007). O algoritmo data cloning é um método numérico que utiliza réplicas da amostra original para aproximar o estimador de máxima verossimilhança como limite de estimadores Bayesianos obtidos por simulação. Nós também analisamos a identificação dos parâmetros do modelo. Medimos a identificação de cada parâmetro individualmente ao observar a volatilidade a posteriori dos estimadores de data cloning. O maior autovalor da matriz de covariância a posteriori proporciona uma medida global de identificação do modelo. Nossos resultados indicam que o modelo de Smets e Wouters (2007) não é bem identificado. O modelo não apresenta boas propriedades globais de identificação, e muitos de seus parâmetros são localmente mal identificados.