This thesis investigates the use of realized volatility features from high frequency data in com- bination with neural networks to improve forecasts of the yield curve of government bonds. I use high frequency data on futures of four U.S. Treasury securities to estimate the Nelson-Siegel yield curve and realized variance of its parameters over the period of 25 years. The estimated parameters are used in prediction of the level, slope and curvature of the yield curve using an LSTM neural network and compared to the Dynamic Nelson-Siegel model. Results show that the use of realized variance and neural network outperforms autoregressive methods in prediction of the level and curvature in daily and monthly forecasts. The yield curve of government bonds itself has a predictive power on multiple macroeconomic variables, therefore improvements in its forecastability may have broader implications on forecasting the overall state of the economy.

A solução da maioria dos problemas práticos enfrentados por administradores financeiros passa pela identificação prévia do custo de oportunidade para investimentos de diferentes prazos e riscos. Este trabalho busca, no conjunto de seus capítulos, realizar uma avaliação crítica das propriedades da estrutura temporal de taxas de juros em reais e de sua utilização como variável exógena fundamental na análise financeira. Sem a pretensão de esgotar qualquer dos temas abordados, procurou-se estabelecer a curva de juros para investimentos livres de riscos em moeda nacional e, através de um conjunto de testes empíricos e observações informais de séries de dados de mercado, identificar peculiaridades que possam invalidar a implementação no Brasil de modelos desenvolvidos no contexto internacional. Dados os aspectos característicos do mercado doméstico evidenciados nos estudos, foram apresentados modelos explicativos tanto para a formação das taxas prefixadas de período quanto para a determinação da remuneração de operações financeiras indexadas à taxa referencial TR. / The solution to most of the problems facing financial managers requires prior identification of the cost of money for different maturities and risks. This paper aims, in its overall content, to examine the Brazilian currency yield curve’s properties and its supporting role in financial analysis. With no intention of exhausting any of the tackled subjects, the Real risk-free term structure was defined and a set of empirical tests performed to identify, with the support of additional data observation, local market’s peculiarities that might prevent international models from being accurately applied in Brazil. Given the domestic market’s distinguishing features emphasized in the studies, models were proposed to explain how short term interest rates are determined in the marketplace for derivatives, and to allow the pricing of financial instruments indexed to the so called TR benchmark (Taxa Referencial).

Cette thèse présente plusieurs extensions relatives aux modèles affines positifs de taux d'intérêt. Un premier chapitre introduit les concepts reliés aux modélisations employées dans les chapitres suivants. Il détaille la définition de processus dits affines, et la construction de modèles de prix d'actifs obtenus par non-arbitrage. Le chapitre 2 propose une nouvelle méthode d’estimation et de filtrage pour les modèles espace-état linéaire-quadratiques. Le chapitre suivant applique cette méthode d’estimation à la modélisation d’écarts de taux interbancaires de la zone Euro, afin d’en décomposer les fluctuations liées au risque de défaut et de liquidité. Le chapitre 4 développe une nouvelle technique de création de processus affines multivariés à partir leurs contreparties univariées, sans imposer l’indépendance conditionnelle entre leurs composantes. Le dernier chapitre applique cette méthode et dérive un processus affine multivarié dont certaines composantes peuvent rester à zéro pendant des périodes prolongées. Incorporé dans un modèle de taux d’intérêt, ce processus permet de rendre compte efficacement des taux plancher à zéro. / This thesis presents new developments in the literature of non-negative affine interest rate models. The first chapter is devoted to the introduction of the main mathematical tools used in the following chapters. In particular, it presents the so-called affine processes which are extensively employed in no-arbitrage interest rate models. Chapter 2 provides a new filtering and estimation method for linear-quadratic state-space models. This technique is exploited in the 3rd chapter to estimate a positive asset pricing model on the term structure of Euro area interbank spreads. This allows us to decompose the interbank risk into a default risk and a liquidity risk components. Chapter 4 proposes a new recursive method for building general multivariate affine processes from their univariate counterparts. In particular, our method does not impose the conditional independence between the different vector elements. We apply this technique in Chapter 5 to produce multivariate non-negative affine processes where some components can stay at zero for several periods. This process is exploited to build a term structure model consistent with the zero lower bound features.

La crise financière de ces dernières années a relancé le débat sur le caractère dit « sans risque de défaut » des obligations souveraines. Face aux enjeux économiques et financiers, les établissements de crédit et les Institutions Financières ont du revoir les méthodes d’évaluation des obligations. Cette thèse a pour objectif la modélisation et la Gestion des prix obligataires et s’articule autour de quatre points. Dans le premier point, nous avons présenté les approches théoriques portant sur les modèles traditionnels des taux d’intérêt. Dans le second point, nous avons conçu un modèle test nommé M-M en discrétisant les modèles à temps continu du taux d’intérêt court et en recourant aux modèles de la famille GARCH. Ce modèle est construit en incorporant les effets niveau des taux d’intérêt à court terme et GARCH (1,1). Les résultats de l’estimation du modèle M-M suggère la nécessité de tenir compte des deux effets pour la modélisation des rendements des bons du Trésor américain. Le troisième point consiste à extraire les facteurs que l’on peut interpréter comme le niveau, la pente et la courbure. Ces facteurs sont extraits à partir de deux modèles qui sont des extensions dynamiques de la fonctionnelle de Nelson et Siegel. Les courbes des taux utilisés sont celles des Etats-Unis, de la France et de l’Afrique du Sud. La présence de l’Afrique du Sud dans cette étude est due à notre envie de traiter la structure par terme des taux d’intérêt d’un pays africain et aussi son économie émergente. A l’aide des proxies, et d’une ACP sur la courbe des taux de ces trois pays, ces facteurs ont été analysés sur la base de leur qualité d’ajustement. Le dernier point a pour but de traiter les indicateurs macroéconomiques et financiers qui peuvent expliquer les facteurs endogènes extraits. / The financial crisis of recent years has re-opened the debate of the so-called "risk-free" government bonds. Faced with economic and financial issues, credit institutions and financial institutions had reviewed the methods of bonds evaluation. The aim of this thesis is the modeling and management of the bonds prices and is organized on four points. In the first point, we present theoretical approaches on traditional models of interest rates. In the second point, we design a test model named M-M by discretizing the continuous-time models of the short interest rate and using the GARCH family models. This model is constructed by incorporating the level effect of the short term interest rates and GARCH (1,1) effect. The M-M estimation results suggest considering both effects for modeling Treasury bills yields. The third point determines the factors that can be interpreted as the level, slope and curvature, these factors are extracted from two models that are dynamics extensions of the Nelson and Siegel functional. We use Yield Curves of the United States, France and South Africa. The presence of South Africa in this study is due to our desire to treat the term structure of interest rates in an African country which is an emerging economy as well. These factors were analyzed on the basis of their goodness of fit. The last point aims to address macroeconomic and financial indicators that can explain the endogenous factors.

Les marchés futures de matières premières existent depuis des siècles. Néanmoins, depuis le début du 21e siècle, le développement en parallèle de la financiarisation et de marchés futures sur une matière première non-stockable (l’électricité) a bouleversé leur fonctionnement.Les trois essais de cette thèse étudient théoriquement et empiriquement les marchés futures de matières premières dans différentes conditions de fonctionnement.Le premier essai est une étude empirique qui montre l’existence de l’effet Samuelson sur les marchés futures d’électricité. Ce faisant, il montre que le stockage n’est pas une condition nécessaire à l’existence d’un tel effet.Le second essai est un modèle qui montre comment le comportement dynamique des prix d’une matière première stockable sur un marché futures segmenté du reste de l’économie est impacté par ses caractéristiques physiques, et notamment par le coût de stockage.Enfin, le troisième essai est un modèle qui montre que la financiarisation modifie la fonction de partage des risques des marchés futures de matières premières, et ce, quelle que soit la maturité concernée. / Commodity futures markets have a long history. However, since the beginning of the 21st century, both the financialization process and the development of futures markets on a non-storable commodity (the electricity) have shake up their functioning.The three essays of this thesis study theoretically and empirically commodity futures markets in different situations of functioning.The first essay is an empirical study that shows that the Samuelson effect exists on electricity derivative markets. As a consequence, it shows that storage is not a necessary condition for such an effect.The second essay is a model that shows how the dynamic behavior of storable commodity prices on a segmented futures market is affected by its physical characteristics, and more precisely by the cost of storage.Further, the third essay is a model that shows that financialization changes the risk sharing function of commodity futures markets, whatever the concerned maturity.

This study investigates how term structure dynamics of VIX futures can be exploited forabnormal returns. To be able to access volatility as a tradeable asset, the trading strategiesonly trades ETFs which are designed to replicate the movements of VIX futures index. Itis established that such ETFs are unsuitable for buy-and-hold investments because of thenegative roll yield it usually suffers, caused by the slope of the VIX term structure.Consequently, these conditions create opportunities for strategies that use direct andinverse VIX ETFs to be profitable. The study is a quantitative study that uses historicalprice data to back test three different trading strategies. The strategies are tested over theperiod 11-oct-2011 to 31-mar-2020. The authors have deliberately chosen to delimit thestudy by not testing the performance of the ETFs, not statistically test the risk-adjustedreturns and not perform a regression to calculate optimal hedge ratios for the strategies.The results from this study shows that its possible for strategies that exploit the termstructure dynamics of VIX futures to generate abnormal returns.

[pt] Evidências empíricas de estimativas de modelos VAR em forma reduzida mostram que houve uma mudança na maneira que a curva de juros brasileira reage à choques de política monetária. Para melhor entender a razão desta mudança, estimamos um DSGE linearizado, acrescido de uma estrutura à termo para as taxas de juros, sobre dois períodos amostrais para verificar quais parâmetros da economia poderiam causado essa mudança. O método de linearização envolve um termo de ajuste que permite a existência de prêmio à termo e gera um estado estacionário ajustado pela volatilidade. Nós discutimos as evidências empíricas, comparamos o método de solução com outro métodos mais tradicionais e estimamos um modelo com preferências Epstein-Zin usando métodos bayesianos. Nós encontramos que nosso modelo estrutural é capaz de capturar algumas das mudanças de comportamento, que é causada principalmente por um menor coeficiente associado à inflação na regra de juros e por maior persistência dos choques monetários. / [en] Empirical evidence from reduced form VAR estimates shows that there has been a change in the way that the Brazilian yield curve reacts to a monetary policy shock. To better understand the sources of this change we estimated a linearized DSGE model with a term structure of interest rates over two sample periods to see what parameters of the economy might have caused the change. The linearization method is augmented with a risk adjustment term in order to generate a positive term spread and a risk-adjusted steady state. We discuss the empirical evidence, compare the solution methods with other traditional methods and estimate a model with Epstein-Zin preferences using Bayesian methods.We find that our structural model is capable of capturing some of the changes of behavior, and it is caused mainly by a smaller inflation coefficient of the interest rate rule and higher persistence of monetary policy shocks.

During the recent years, numerous so-called Buy Now, Pay Later companies have emerged. A type of financial institution offering short term consumer credit contracts. As these institutions have gained popularity, their undertaken credit risk has increased vastly. Simultaneously, the IFRS 9 regulatory requirements must be complied with. Specifically, the Probability of Default (PD) for the entire lifetime of such a contract must be estimated. The collection of incremental PDs over the entire course of the contract is called the PD term structure. Accurate estimates of the PD term structures are desirable since they aid in steering business decisions based on a given risk appetite, while staying compliant with current regulations. In this thesis, the efficiency of Machine Learning within PD term structure modeling is examined. Two categories of Machine Learning algorithms, in five variations each, are evaluated; (1) Deep Neural Networks; and (2) Gradient Boosted Trees. The Machine Learning models are benchmarked against a traditional Markov Chain model. The performance of the models is measured by a set of calibration and discrimination metrics, evaluated at each time point of the contract as well as aggregated over the entire time horizon. The results show that Machine Learning can be used efficiently within PD term structure modeling. The Deep Neural Networks outperform the Markov Chain model in all performance metrics, whereas the Gradient Boosted Trees are better in all except one metric. For short-term predictions, the Machine Learning models barely outperform the Markov Chain model. For long-term predictions, however, the Machine Learning models are superior. / Flertalet s.k. Köp nu, betala senare-företag har växt fram under de senaste åren. En sorts finansiell institution som erbjuder kortsiktiga konsumentkreditskontrakt. I samband med att dessa företag har blivit alltmer populära, har deras åtagna kreditrisk ökat drastiskt. Samtidigt måste de regulatoriska kraven ställda av IFRS 9 efterlevas. Specifikt måste fallisemangsrisken för hela livslängden av ett sådant kontrakt estimeras. Samlingen av inkrementell fallisemangsrisk under hela kontraktets förlopp kallas fallisemangsriskens terminsstruktur. Precisa estimat av fallisemangsriskens terminsstruktur är önskvärda eftersom de understödjer verksamhetsbeslut baserat på en given riskaptit, samtidigt som de nuvarande regulatoriska kraven efterlevs. I denna uppsats undersöks effektiviteten av Maskininlärning för modellering av fallisemangsriskens terminsstruktur. Två kategorier av Maskinlärningsalgoritmer, i fem variationer vardera, utvärderas; (1) Djupa neuronnät; och (2) Gradient boosted trees. Maskininlärningsmodellerna jämförs mot en traditionell Markovkedjemodell. Modellernas prestanda mäts via en uppsättning kalibrerings- och diskrimineringsmått, utvärderade i varje tidssteg av kontraktet samt aggregerade över hela tidshorisonten. Resultaten visar att Maskininlärning är effektivt för modellering av fallisemangsriskens terminsstruktur. De djupa neuronnäten överträffar Markovkedjemodellen i samtliga prestandamått, medan Gradient boosted trees är bättre i alla utom ett mått. För kortsiktiga prediktioner är Maskininlärningsmodellerna knappt bättre än Markovkedjemodellen. För långsiktiga prediktioner, däremot, är Maskininlärningsmodellerna överlägsna.

Market makers such as large banks are exposed to market risk in fixed income by acting as a counterparty for customers that enter swap contracts. This master thesis addresses the problem of creating a cost-effective hedge for a realistic swap book of a market maker in a multiple yield curve setting. The proposed hedge model is the two-stage stochastic optimisation problem created by Blomvall and Hagenbjörk (2020). Systematic term structure innovations (components) are estimated using six different component models including principal component analysis (PCA), independent component analysis (ICA) and rotations of principal components. The component models are evaluated with a statistical test that uses daily swap rate observations from the European swap market. The statistical test shows that for both FRA and IRS contracts, a rotation of regular principal components is capable of a more accurate description of swap rate innovations than regular PCA. The hedging model is applied to an FRA and an IRS swap book separately, with daily rebalancing, over the period 2013-06-21 to 2021-05-11. The model produces a highly effective hedge for the tested component methods. However, replacing the PCA components with improved components does not improve the hedge. The study is conducted in collaboration with two other master theses, each done at separate banks. This thesis is done in collaboration with Swedbank and the simulated swap book is based on the exposure of a typical swap book at Swedbank, which is why the European swap market is studied.

The initiative by the EU to combat global warming through the introduction of a cap-and-trade system for greenhouse gas emissions in 2005, known as the EU Emissions Trading System (ETS), resulted in the inception of a new financial market. The right to emit one tonne of CO2-equivalents, as well as derivatives on this right, have become commodities, traded both through exchanges and over the counter. A relevant question thus becomes how a market maker trading these derivatives should hedge their exposures. This thesis examines how stochastic optimisation can be used to hedge a portfolio of futures, forwards, and emissions rights in the EU ETS, while taking into account market microstructre effects such as transaction costs. This is done through the implementation of a Stochastic Programming (SP) model that weights portfolio risk against transaction costs,where the entire term structure of futures is Monte Carlo-simulated. The term structure of the futures is analysed by decomposing futures prices into the spot price, the term structure of the risk free interest rate, and the term structure of the convenience yield, as was first done by Working (1948). These are estimated using the non-parametric optimisation framework of Blomvall (2017), where EUROIS contracts and ICE EUA futures are used as benchmark instruments. It was found that the method results in smooth yield curves with small repricing errors, thanks to suitable parameter calibration through 3-fold Cross Validation for both curves. From day-to-day changes in the resulting curves, three systematic risk factors for each curve, that capture more than 98% of the variance during the analysed period from October 2012 to March 2018, were found with PCA. These factors were then fitted to univariate GARCH-models, and normal mixture copulas. This model allows for the hedging problem to be solved with SP. For the out-of-sample period from March 2018 to April 2022, the results show promise, as the portfolios hedged with SP are considerably less volatile than both a statically hedged and an unhedged portfolio. Furthermore, for some values of the parameter weighting risk and costs, these portfolios yield mean variance efficiency.