We develop a risk factor model based on data from a large number of portfolios spanning multiple asset classes. The risk factors are selected based on economic theory through an analysis of the asset holdings, as well as statistical tests. As many assets have limited historical data available, we implement and analyse the impact of regularisation to handle sparsity. Based on the factor model, two parametric methods for calculating Value-at-Risk (VaR) for a portfolio are developed: one with constant volatility and one with a CCC-GARCH volatility updating scheme. These methods are evaluated through backtesting on daily and weekly returns of a selected set of portfolios whose contents reflect the larger majority well. A historical data approach for calculating VaR serves as a benchmark model. We find that under daily returns, the historical data method outperforms the factor models in terms of VaR violation rates. None yield independent violations however. Under weekly returns, both factor models produce more accurate violation rates than the historical data model, with the CCC-GARCH model also yielding independent VaR violations for almost all portfolios due to its ability to adjust up VaR estimates in periods of increased market volatility. We conclude that if weekly VaR estimates are acceptable, tailored risk factor models provide accurate measures of portfolio risk. / Vi bygger en riskfaktormodell givet en stor mängd portföljer innehållande flera olika typer av tillgångar. Riskfaktorerna väljs ut baserat på ekonomisk teori genom en analys av portföljernas innehåll samt genom statistiska test. Eftersom många tillgångar har en liten mängd historisk data tillgänglig implementerar vi och analyserar effekterna av regularisering i faktorregressionen. Två parametriska metoder för att beräkna Value-at-Risk (VaR) utvecklas baserat på faktormodellen: en med konstant volatilitet och en med volatilitetsuppdatering genom CCC-GARCH. Metoderna utvärderas med bakåttestning på daglig och veckovis avkastning från utvalda portföljer vars innehåll reflekterar den större majoriteten. En historisk data-baserad metod för att beräkna VaR används som referensmodell. Under daglig avkastning överträffar historisk data-modellen faktormodellerna med avseende på frekvensen av VaR-överträdelser. Ingen modell resulterar dock i oberoende överträdelser. Under veckovis avkastning å andra sidan ger båda faktormodellerna mer exakta överträdelsefrekvenser än historisk data-modellen, där faktormodellen med CCC-GARCH också ger oberoende överträdelser för nästan alla portföljer, tack vare modellens förmåga att justera upp VaR-estimaten i perioder av högre volatilitet på marknaden. Sammanfattningsvis ger skräddarsydda riskfaktormodeller goda riskestimat, givet att det är acceptabelt med veckovisa beräkningar av VaR.

This thesis establishes methods to quantify and explain uncertainty through high-order moments in time series data, along with first principal-based improvements on the standard autoencoder and variational autoencoder. While the first-principal improvements on the standard variational autoencoder provide additional means of explainability, we ultimately look to non-variational methods for quantifying uncertainty under the autoencoder framework. We utilize Shannon's differential entropy to accomplish the task of uncertainty quantification in a general nonlinear and non-Gaussian setting. Together with previously established connections between autoencoders and principal component analysis, we motivate the focus on differential entropy as a proper abstraction of principal component analysis to this more general framework, where nonlinear and non-Gaussian characteristics in the data are permitted. Furthermore, we are able to establish explicit connections between high-order moments in the data to those in the latent space, which induce a natural latent space decomposition, and by extension, an explanation of the estimated uncertainty. The proposed methods are intended to be utilized in economic and financial factor models in state space form, building on recent developments in the application of neural networks to factor models with applications to financial and economic time series analysis. Finally, we demonstrate the efficacy of the proposed methods on high frequency hourly foreign exchange rates, macroeconomic signals, and synthetically generated autoregressive data sets. / Master of Science / This thesis establishes methods to quantify and explain uncertainty in time series data, along with improvements on some latent variable neural networks called autoencoders and variational autoencoders. Autoencoders and varitational autoencodes are called latent variable neural networks since they can estimate a representation of the data that has less dimension than the original data. These neural network architectures have a fundamental connection to a classical latent variable method called principal component analysis, which performs a similar task of dimension reduction but under more restrictive assumptions than autoencoders and variational autoencoders. In contrast to principal component analysis, a common ailment of neural networks is the lack of explainability, which accounts for the colloquial term black-box models. While the improvements on the standard autoencoders and variational autoencoders help with the problem of explainability, we ultimately look to alternative probabilistic methods for quantifying uncertainty. To accomplish this task, we focus on Shannon's differential entropy, which is entropy applied to continuous domains such as time series data. Entropy is intricately connected to the notion of uncertainty, since it depends on the amount of randomness in the data. Together with previously established connections between autoencoders and principal component analysis, we motivate the focus on differential entropy as a proper abstraction of principal component analysis to a general framework that does not require the restrictive assumptions of principal component analysis. Furthermore, we are able to establish explicit connections between high-order moments in the data to the estimated latent variables (i.e., the reduced dimension representation of the data). Estimating high-order moments allows for a more accurate estimation of the true distribution of the data. By connecting the estimated high-order moments in the data to the latent variables, we obtain a natural decomposition of the uncertainty surrounding the latent variables, which allows for increased explainability of the proposed autoencoder. The methods introduced in this thesis are intended to be utilized in a class of economic and financial models called factor models, which are frequently used in policy and investment analysis. A factor model is another type of latent variable model, which in addition to estimating a reduced dimension representation of the data, provides a means to forecast future observations. Finally, we demonstrate the efficacy of the proposed methods on high frequency hourly foreign exchange rates, macroeconomic signals, and synthetically generated autoregressive data sets. The results support the superiority of the entropy-based autoencoder to the standard variational autoencoder both in capability and computational expense.

La tesi di dottorato, composta da tre capitoli, si concentra sulla valutazione delle dinamiche di inflazione in Georgia e sulla previsione dei tassi di cambio nominali per i Paesi della European Eastern Partnership attraverso l’utilizzo di moderne tecniche econometriche. Nel primo capitolo, abbiamo svolto un’indagine sui modelli di previsione dei tassi di cambio e dell’inflazione. Questa indagine rivela che i modelli “factor-based and time-varying parameter” generano migliori previsioni rispetto ad altri modelli. Nel secondo capitolo, abbiamo approfondito le dinamiche di inflazione in Georgia utilizzando la New Keynesian Phillips Curve ibrida, inserita all’interno di un quadro di un modello “time-varying parameter (TVP)”. Una stima del modello TVP con volatilità stocastica mostra la persistenza di un’inflazione bassa durante il periodo 1996-2012. Un’analisi più approfondita dal 2003 mostra una volatilità crescente dell’inflazione. Inoltre, le stime del parametro evidenziano che la componente forward-looking del modello è importante a seguito dell’adozione di inflation targeting da parte della NBG a partire dal 2009. Nel terzo capitolo, abbiamo costruito dei modelli fattoriali, “Factor Vector Autoregressive” per prevedere i tassi di cambio nominali per i Paesi dell’European Eastern Partnership. Questi modelli prevedono meglio i tassi di cambio nominali rispetto ad un processo naïve come il random walk. / The Ph.D. thesis consist of three chapters on evaluating inflation dynamics in Georgia and modeling and forecasting nominal exchange rates for the European Eastern Partnership (EaP) countries using modern applied econometric techniques. In the first chapter, we survey of models those produce high predictive powers for forecasting exchange rates and inflation. This survey reveals that the factor-based and time-varying parameter (TVP) models generate superior forecasts relative to all other models. In the second chapter, we study inflation dynamics in Georgia using a hybrid New Keynesian Phillips Curve (NKPC) nested within a time-varying parameter (TVP) framework. Estimation of a TVP model with stochastic volatility shows low inflation persistence over the entire time span (1996-2012), while revealing increasing volatility of inflation shocks since 2003. Moreover, parameter estimates point to the forward-looking component of the model gaining importance following the National Bank of Georgia (NBG) adoption of inflation targeting in 2009. In the third chapter, we construct Factor Vector Autoregressive (FVAR) models to forecast nominal exchange rates for the EaP countries. This study provides better forecasts of nominal exchange rates than those produced by the random walk process.

Diese Dissertation befasst sich mit europäischer und US-amerikanischer Konjunkturgeschichte und Marktintegration im 19. und 20. Jahrhundert. Zur Analyse von konjunkturellen Schwankungen stellt sie der weitverbreiteten Historischen Volkswirtschaftlichen Gesamtrechnung (VGR) die Methode dynamischer Faktoranalyse zur Seite, die dazu beiträgt, die begrenzten historischen Zeitreihen effizient zu nutzen. Die nationale und internationale Entwicklung von Weizenmärkten seit dem Ende der Napoleonischen Kriege wird mit einem multivariaten dynamischen Faktormodell untersucht. Spektralanalyse wird zur Berechnung frequenzspezifischer Kohärenz von historischen Börsenindizes und konkurrierenden Schätzungen des Nationalprodukts in Deutschland zwischen 1850 und 1913 herangezogen. Ein wichtiges Ergebnis ist, dass Finanzdaten die Datierung der Konjunktur im Deutschen Kaiserreich erleichtern, was auch durch die Ergebnisse der Faktoranalyse bestätigt wird. Der verwendete Aktienindex, einzelne reale Konjunkturindikatoren und der dynamische Faktor korrelieren eng miteinder. Die Bildung sektoraler Sub-Indizes zeigt, dass der Übergang von einer landwirtschaftlich zu einer industriell geprägten Volkswirtschaft vermutlich früher geschehen ist als Beschäftigungsanteile aus der Historischen VGR vermuten lassen. Die Untersuchung der U.S.-Konjunktur ergibt die Annahme zeitvariierender Strukturparameter eine Erhöhung der Konjunkturschwankungsbreite nach dem 2. Weltkrieg verglichen mit der Zeit vor dem 1. Weltkrieg. Für die Weizenmarktintegration in Europa zeigt sich, dass die Entwicklung vor der Mitte des 19. Jahrhunderts schneller voran ging als danach, was eine Neuinterpretation der Rolle von Technologien wie dem Metallrumpf und dem Dampfschiff sowie dem Eintritt Amerikas als Weizenproduzenten nahelegt. / This thesis addresses historical business cycles and market integration in Europe and America in the 19th and 20th centuries. For the analysis of historical business cycles, the widely used methodology of historical national accounting is complemented with a dynamic factor model that allows for using scarce historical data efficiently. In order to investigate how national and international markets developed since the early 1800s, a multivariate dynamic factor model is used. Spectral analysis helps in measuring frequency specific correlation between financial indicators and rivaling national income estimates for Germany between 1850 and 1913. One result is that the historical stock market index used helps to discriminate between competing estimates of German national income. A dynamic factor estimated from a broad time series data set confirms this result. Sub-indices for agriculture and industry suggest that the German economy industrialized earlier than evidence from national accounting shows. The finding for the U.S. business cycle is that relaxing the assumption of constant structural parameters yields higher postwar aggregate volatility relative to the period before World War I. Concerning market integration, it is found that European wheat markets integrated faster before mid-19th century than after. Thus, the impact of the metal hull and steam ship as well as the relevance of American wheat for the world wheat market have perhaps been overstated.

Diese Arbeit besteht aus vier Essays, die empirische und methodische Beiträge zur Messung von Konjunkturzyklen und deren Zusammenhänge zu demographischen Variablen liefern. Der erste Essay analysiert unter Zuhilfenahme eines Bayesianischen Dynamischen Faktormodelles die Volatilität des US-amerikanischen Konjunkturzyklus seit 1867. In dem Essay wird gezeigt, dass die Volatilität in der Periode vor dem Ersten Weltkrieg und nachdem Zweiten Weltkrieg niedriger war als in der Zwischenkriegszeit. Eine geringere Volatilität für die Periode nach dem Zweiten Weltkrieg im Vergleich zu der Periode vor dem Ersten Weltkrieg kann nicht bestätigt werden. Der zweite Essay hebt die Bayesianischen Eigenschaften bezüglich dynamischer Faktormodelle hervor. Der Essay zeigt, dass die ganze Analyse hindurch - im Gegensatz zu klassischen Ansätzen - keine Annahmen an die Persistenz der Zeitreihen getroffen werden muss. Des Weiteren wird veranschaulicht, wie im Bayesianischen Rahmen die Anzahl der Faktoren bestimmt werden kann. Der dritte Essay entwickelt einen neuen Ansatz, um altersspezifische Sterblichkeitsraten zu modellieren. Kovariate werden mit einbezogen und ihre Dynamik wird gemeinsam mit der von latenten Variablen, die allen Alterklassen zugrunde liegen, modelliert. Die Resultate bestätigen, dass makroökonomische Variablen Prognosekraft für die Sterblichkeit beinhalten. Im vierten Essay werden makroökonomischen Zeitreihen zusammen mit altersspezifischen Sterblichkeitsraten einer strukturellen Analyse unterzogen. Es wird gezeigt, dass sich die Sterblichkeit von jungen Erwachsenen in Abhängigkeit von Konjunkturzyklen deutlich von den der anderen Alterklassen unterscheidet. Daher sollte in solchen Analysen, um Scheinkorrelation vorzubeugen, zwischen den einzelnen Altersklassen differenziert werden. / The thesis consists of four essays, which make empirical and methodological contributions to the fields of business cycle analysis and demography. The first essay presents insights on U.S. business cycle volatility since 1867 derived from a Bayesian dynamic factor model. The essay finds that volatility increased in the interwar periods, which is reversed after World War II. While evidence can be generated of postwar moderation relative to pre-1914, this evidence is not robust to structural change, implemented by time-varying factor loadings. The second essay scrutinizes Bayesian features in dynamic index models. The essay shows that large-scale datasets can be used in levels throughout the whole analysis, without any pre-assumption on the persistence. Furthermore, the essay shows how to determine the number of factors accurately by computing the Bayes factor. The third essay presents a new way to model age-specific mortality rates. Covariates are incorporated and their dynamics are jointly modeled with the latent variables underlying mortality of all age classes. In contrast to the literature, a similar development of adjacent age groups is assured, allowing for consistent forecasts. The essay demonstrates that time series of covariates contain predictive power for age-specific rates. Furthermore, it is observed that in particular parameter uncertainty is important for long-run forecasts, implicating that ignoring parameter uncertainty might yield misleadingly precise predictions. In the fourth essay the model developed in the third essay is utilized to conduct a structural analysis of macroeconomic fluctuations and age-specific mortality rates. The results reveal that the mortality of young adults, concerning business cycles, noticeably differ from the rest of the population. This implies that differentiating closely between particular age classes, might be important in order to avoid spurious results.

Diese Arbeit besteht aus vier Essays, die empirische und methodische Beiträge zu den Gebieten der Finanzmarktökonomik und der Makroökonomik liefern. Der erste Essay beschäftigt sich mit der Spezifikation der Investoren verfügbaren Informationsmenge in Tests bedingter Kapitalmarktmodelle. Im Speziellen schlägt es die Verwendung dynamischer Faktoren als Instrumente vor. Diese fassen per Konstruktion die Information in einer Vielzahl von Variablen zusammen und stellen daher intuitive Maße für die Investoren zur Verfügung stehenden Informationen dar. Es wird gezeigt, dass so die Schätzfehler bedingter Modelle im Vergleich zu traditionellen, auf einzelnen Indikatoren beruhenden Modellvarianten substantiell verringert werden. Ausgehend von Ergebnissen, dass die Zentralbank zur Festlegung des kurzfristigen Zinssatzes eine große Menge an Informationen berücksichtigt, wird im zweiten Essay im Rahmen eines affinen Zinsstrukturmodells eine ähnliche Idee verwandt. Speziell wird die Dynamik des kurzfristigen Zinses im Rahmen einer Faktor-Vektorautoregression modelliert. Aufbauend auf dieser dynamischen Charakterisierung der Geldpolitik wird dann die Zinsstruktur unter der Annahme fehlender Arbitragemöglichkeiten hergeleitet. Das resultierende Modell liefert bessere Vorhersagen US-amerikanischer Anleihenzinsen als eine Reihe von Vergleichsmodellen. Der dritte Essay analysiert die Vorhersagekraft der Zinsstrukturkomponenten "level", "slope", und "curvature" im Rahmen eines dynamischen Faktormodells für makroökonomische und Zinsdaten. Das Modell wird mit einem Metropolis-within-Gibbs Sampling Verfahren geschätzt, und Überraschungsänderungen der drei Komponenten werden mit Hilfe von Null- und Vorzeichenrestriktionen identifiziert. Die Analyse offenbart, dass der "curvature"-Faktor informativer in Bezug auf die zukünftige Entwicklung der Zinsstruktur und der gesamtwirtschaftlichen Aktivität ist als bislang vermutet. Der vierte Essay legt eine monatliche Chronologie der Konjunkturzyklen im Euro-Raum vor. Zunächst wird mit Hilfe einer verallgemeinerten Interpolationsmethode eine monatliche Zeitreihe des europäischen BIP konstruiert. Anschließend wird auf diese Zeitreihe ein Datierungsverfahren angewandt, das kurze und flache Konjunkturphasen ausschließt. / This thesis consists of four essays of independent interest which make empirical and methodological contributions to the fields of financial economics and macroeconomics. The first essay deals with the proper specification of investors’ information set in tests of conditional asset pricing models. In particular, it advances the use of dynamic factors as conditioning variables. By construction, dynamic factors summarize the information in a large number of variables and are therefore intuitively appealing proxies for the information set available to investors. The essay demonstrates that this approach substantially reduces the pricing errors implied by conditional models with respect to traditional approaches that use individual indicators as instruments. Following previous evidence that the central bank uses a large set of conditioning information when setting short-term interest rates, the second essay employs a similar insight in a model of the term structure of interest rates. Precisely, the dynamics of the short-term interest rate are modelled using a Factor-Augmented Vector-Autoregression. Based on this dynamic characterization of monetary policy, the term structure of interest rates is derived under the assumption of no-arbitrage. The resulting model is shown to provide superior out-of-sample forecasts of US government bond yields with respect to a number of benchmark models. The third essay analyzes the predictive information carried by the yield curve components level, slope, and curvature within a joint dynamic factor model of macroeconomic and interest rate data. The model is estimated using a Metropolis-within-Gibbs sampling approach and unexpected changes of the yield curve components are identified employing a combination of zero and sign restrictions. The analysis reveals that the curvature factor is more informative about the future evolution of the yield curve and of economic activity than has previously been acknowledged. The fourth essay provides a monthly business cycle chronology for the Euro area. A monthly series of Euro area real GDP is constructed using an interpolation routine that nests previously suggested approaches as special cases. Then, a dating routine is applied to the interpolated series which excludes business cycle phases that are short and flat.

La thèse étudie la dynamique et la persistance de l’inflation dans les pays en développement, particulièrement ceux des pays de la Zone UEMOA, en mettant en exergue les spécificités de ces économies. Le premier chapitre, consacré à l’évaluation de la persistance, révèle que le degré de persistance de l'inflation est faible dans ces pays, ce qui constitue un atout pour les autorités monétaires. Dans le chapitre 2, il a été défini un cadre théorique plus approprié à l’analyse de la persistance de l’inflation dans les pays de la sous-région. L’approche a permis de montrer que le degré de persistance de l’inflation dans ces pays ne dépendait pas uniquement des politiques monétaire et de change, mais aussi négativement du poids du secteur vivrier local dans l’économie. Dans le chapitre 3, la thèse analyse les écarts d’inflation dans les pays membres de l’UEMOA, en examinant la β-convergence des différentiels d'inflation. Les estimations révèlent que, d’une part, les écarts d’inflation se sont fortement réduits à l’intérieur de l'Union et que, d’autre part, ils restent fortement persistants avec la zone Euro. Le chapitre 4 est consacré à l’évaluation du rôle des différents facteurs et utilise ensuite une spécification spatiale en panel, pour tester les effets de contagion entre pays. Les estimations indiquent une prédominance des facteurs globaux et des effets de contagion entre pays dont l'ampleur dépend du poids des exportations de chaque pays vers les autres pays de la sous région. / This thesis examines the inflation dynamics and persistence in developing countries, especially in the UEMOA zone, highlighting the specificities of these economies. The first chapter, reveals that the inflation persistence degree, in these countries, is low which represents an asset to the monetary authorities. In Chapter 2, it was defined a more appropriate theoretical framework to analyze the inflation persistence in the countries of the sub-region. The approach allowed to demonstrate that the inflation persistence degree in these countries is not only dependent on monetary and exchange rate policies, but also negatively to the weight of local food sector in the economy. Chapter 3, analyzes the inflation differentials in the UEMOA member countries, by examining the β - convergence of inflation differentials. Estimations show that the inflation differentials are greatly reduced within the Union and they are highly persistent with the Euro zone. Chapter 4, is devoted to assessing the role of various factors and then uses a spatial panel specification to test the spillover effect between countries. Estimations indicate a predominance of global factors and contagion between countries whose magnitude depends on the weight of exports to other countries in the sub-region.

Nous introduisons un ensemble de modèles à facteurs latents dédié à la génomique du paysage et aux tests d'associations écologiques. Cela comprend des méthodes statistiques pour corriger des effets d'autocorrélation spatiale sur les cartes de composantes principales en génétique des populations (spFA), des méthodes pour estimer rapidement et efficacement les coefficients de métissage individuel à partir de matrices de génotypes de grande taille et évaluer le nombre de populations ancestrales (sNMF) et des méthodes pour identifier les polymorphismes génétiques qui montrent de fortes corrélations avec des gradients environnementaux ou avec des variables utilisées comme des indicateurs pour des pressions écologiques (LFMM). Nous avons aussi développé un ensemble de logiciels libres associés à ces méthodes, basés sur des programmes optimisés en C qui peuvent passer à l'échelle avec la dimension de très grand jeu de données, afin d'effectuer des analyses de structures de population et des cribles génomiques pour l'adaptation locale. / We introduce a set of latent factor models dedicated to landscape genomics and ecological association tests. It includes statistical methods for correcting principal component maps for effects of spatial autocorrelation (spFA); methods for estimating ancestry coefficients from large genotypic matrices and evaluating the number of ancestral populations (sNMF); and methods for identifying genetic polymorphisms that exhibit high correlation with some environmental gradient or with the variables used as proxies for ecological pressures (LFMM). We also developed a set of open source softwares associated with the methods, based on optimized C programs that can scale with the dimension of very large data sets, to run analyses of population structure and genome scans for local adaptation.

Submitted by Tomoharu Maeda Junior (tomoharu.maeda@gmail.com) on 2012-09-11T19:06:07Z No. of bitstreams: 1 DissertacaoMPFE-TMJ.pdf: 2327119 bytes, checksum: e86dad879e97ba7ee62edb2eafde4556 (MD5) / Rejected by Suzinei Teles Garcia Garcia (suzinei.garcia@fgv.br), reason: Prezado Tomoharu, Foi alterado o título da dissertação, porém não informado em Ata é necessário seu orientador informar. Título anterior: PREVISÃO DA ESTRUTURA A TERMO DE TAXA DE JUROS DO BRASIL UTILIZANDO MODELO DE FATORES COM CORREÇÃO DE ERROS Att. Suzi 3799-7876 on 2012-09-11T19:48:31Z (GMT) / Submitted by Tomoharu Maeda Junior (tomoharu.maeda@gmail.com) on 2012-09-12T13:14:24Z No. of bitstreams: 1 DissertacaoMPFE-TMJ.pdf: 2327119 bytes, checksum: e86dad879e97ba7ee62edb2eafde4556 (MD5) / Approved for entry into archive by Suzinei Teles Garcia Garcia (suzinei.garcia@fgv.br) on 2012-09-12T13:31:49Z (GMT) No. of bitstreams: 1 DissertacaoMPFE-TMJ.pdf: 2327119 bytes, checksum: e86dad879e97ba7ee62edb2eafde4556 (MD5) / Made available in DSpace on 2012-09-12T13:37:49Z (GMT). No. of bitstreams: 1 DissertacaoMPFE-TMJ.pdf: 2327119 bytes, checksum: e86dad879e97ba7ee62edb2eafde4556 (MD5) Previous issue date: 2012-08-14 / O objetivo do presente trabalho é verificar se o modelo que combina correção de erros e fatores extraídos de grandes conjuntos de dados macroeconômicos produz previsões mais precisas das taxas de juros do Brasil em relação aos modelos VAR, VECM e FAVAR. Para realizar esta análise, foi utilizado o modelo sugerido por Banerjee e Marcellino (2009), o FAVECM, que consiste em agregar o mecanismo de correção de erros ao modelo proposto por Bernanke, Boivin e Eliasz (2005), o FAVAR. A hipótese é que o FAVECM possuiu uma formulação teórica mais geral. Os resultados mostram que para o mercado brasileiro o FAVECM apresentou ganhos significativos de previsão para as taxas mais longas e horizontes de previsão maiores. / The objective of the present work is to examine if the model that combines error correction and factors extracted from large macoeconomic data sets offers a higher forecasting accuracy of the interest rate in Brazil when compared to VAR, VECM and FAVAR. In order to conduct this analysis it was used the econometric methodology introduced by Banerjee and Marcellino (2009), the FAVECM, which allows for the inclusion of error correction terms in the model introduced by Bernanke, Boivin and Eliasz (2005), the FAVAR. The hypothesis is that the FAVECM has several conceptual advantages given it is a nesting (or has a more general) specification. The results show that, for the Brazilian market, the FAVECM presented significant gains in forecasts for longer maturity rates and for longer prevision horizons.

Comovement is ubiquitous in financial markets. The evolution of asset characteristics, such as price, volatility or liquidity, exhibits a high degree of correlation across assets---a phenomenon that in this thesis will generically be denoted with the term comovement. The origins of such comovement are legion. In their investment decisions, economic agents are not only influenced by their idiosyncrasies---a large part of investment motivations are shared over a population. Demographics or the political situation can generate constraints that are similar for a large number of people. A country's geography can greatly influence the sectors in which it is most productive, which implies that many people are sometimes subject to the same risk factors. Moreover, it is well known that mimesis is part of human psychology, and that people mimic their peers even when taking personal decisions. For these reasons, and many more, financial markets have a very systematic character, and studying the nature and intensity of such comovement is important from a risk management point of view. <p>This thesis studies comovement in financial markets under three dimensions. First, I consider comovement in equity liquidity. The liquidity of an asset is the ease with which that asset can be bought or sold. Liquidity can be measured in various ways and the first chapter concludes that market movements of two different liquidity measures have the same origin. Second, I study the impact correlation comovement on the price of stocks. The correlations between stock returns and the market return evolve through time and are correlated themselves. The effect of this correlation comovement on asset prices is however ambiguous and there is not enough evidence to depict a clear image. Finally, I develop a model to investigate contagion dynamics in the secondary market for European sovereign bonds over the past two years. More particularly, I study whether changes in the bond price of one specific country have an impact the next day on the average bond price in Europe. The study concludes of that bonds of France, Ireland, Portugal, Spain and Italy have been most contagious, whereas the much more volatile Greek bonds have had little impact on the other European countries. / Doctorat en Sciences économiques et de gestion / info:eu-repo/semantics/nonPublished