Cette thèse traite de la modélisation des dérivés de crédit et se compose de deux parties: La première partie concerne le modèle à densité, récemment proposé par El Karoui et al. où on fait l'hypothèse que la loi conditionnelle de temps de défaut sachant la filtration référence est équivalente à sa loi (non-conditionnelle). Sous cette hypothèse, nous donnons des démonstrations différentes (et plus simples) aux résultats déjà existant dans la théorie du grossissement initial et progressif des filtrations. En outre, nous présentons de nouveaux résultats comme par exemple le théorème de représentation prévisible pour la filtration progressivement grossie dans le cas multidimensionnel. Nous proposons ensuite plusieurs méthodes pour construire des modèles à densité, dans les cas unidimensionnel et multidimensionnel. Enfin, nous montrons que le modèle à densité est une approche efficace pour la couverture dynamique de produits dérivés de crédit multi-name. Dans la deuxième partie, afin d'étudier le risque de contrepartie dans un contrat de CDS, nous avons proposé un modèle markovien dans lequel des défauts simultanés sont possibles. Le wrong-way risk est donc représenté par le fait que, à moment de la défaillance de la contrepartie, il y a une probabilité strictement positive pour que l'entité de référence fasse défaut aussi. Nous commençons par considérer une chaîne de Markov à quatre états correspondant à deux noms; Dans ce cas simple, nous obtenons des formules semi-explicites pour la plupart des quantités importantes, comme le prix, la CVA, l’EPE ou les ratios de couverture. Nous généralisons ensuite ce cadre pour tenir compte du risque de spread en introduisant des facteurs stochastiques; nous traitons un modèle copule Markovien avec des intensités stochastiques. Nous abordons également la question de la couverture dynamique du CVA avec un CDS écrit sur la contrepartie. Pour l'implémentation du modèle, nous spécifions les intensités par des processus affines, ce qui compte tenu de la propriété copule dynamique du modèle, rend la calibration de ce modèle efficace. Les résultats numériques sont présentés pour montrer la pertinence du comportement de la CVA dans le modèle avec les faits stylisés du marché. / This thesis deals with credit derivatives modeling and consists of two parts: The first part concerns the density model, recently proposed by El Karoui et al., where the standing assumption is that the conditional law of default time given the reference filtration is equivalent to its (non-conditional) law. Under this assumption, we provide alternative (and simpler) proofs for some existing results in the theory of initial and progressive enlargement of filtrations. Also, we present some new results such as the predictable representation theorem for progressively enlarged filtration in the multidimensional case. We then propose several methods to construct density models, in both one-dimensional and multidimensional cases. Finally, we show that the density model is an efficient approach for dynamic hedging of multi-name credit derivatives. In the second part, a Markov model is constructed for studying the counterparty risk in a CDS contract. The wrong-way risk in this model is accounted for by the possibility of the simultaneous default of the reference name and of the counterparty. We start by considering a Markov chain model of two reference credits, the firm underlying the CDS and the protection seller in the CDS. In this set-up, we have semi-explicit formulae for most quantities of interest with regard to CDS counterparty risk like price, CVA, EPE or hedging strategies. We then generalize this framework to account for the spread risk by introducing stochastic factors, so that, we deal with a Markov copula model with stochastic intensities. We also address the issue of dynamically hedging the CVA with a CDS written on the counterparty. For model implementation, we consider three different affine specification of the intensities, which in view of the dynamic copula property of the model, make calibration very efficient. Numerical results are presented to show the adequacy of the behavior of CVA in the model with stylized features.

The global over-the-counter derivatives market reached a staggering 14.5 trillion US dollars in gross market value at the end of December 2007. Although OTC derivatives are extremely useful and versatile in transferring risks, it appears to be a double-edged sword. For every derivative transaction concluded in the OTC market, there are two parties involved – each of which is exposed to the other defaulting on the agreed terms and conditions of the contract. Counterparty credit risk is defined as the loss that will be incurred in the event that a counterparty fails to honour its financial obligations. This dissertation provides an overview of counterparty credit risk measurement from a theoretical point of view and puts an emphasis on the demonstration of the current solutions used in practice to address this problem. The author applies a bottom up approach to the problem by defining counterparty credit risk exposure on a contract (single-trade) level and expands this definition on a step-by-step basis to incorporate portfolio effects, such as correlation among underlying market variables as well as credit risk mitigation techniques, such as netting and collateral agreements, in measuring counterparty credit risk exposure on a counterparty level. The author also discusses related concepts which impact counterparty credit risk such as wrong-way risk and proposes an enhancement to the framework introduced by Finger (2000) for incorporating wrong-way risk into existing measures of counterparty credit risk exposure. Finger‟s framework is enhanced by the introduction of a structural model approach which can be used in establishing a functional and intuitive relationship between the probability of default of the counterparty and the underlying market variable to the derivative contract under consideration. This approach is also applied to a typical South African situation through the use of Monte Carlo simulation. The topic of counterparty credit risk modelling is a very relevant topic in modern finance, especially since the advent of Basel 2 which this dissertation also touches on in terms of the applications of counterparty credit risk modelling and how this relates to the minimum regulatory capital requirements set by bank regulators. Copyright / Dissertation (MSc)--University of Pretoria, 2009. / Mathematics and Applied Mathematics / unrestricted

This thesis analyzes a continuity between the changes of the exchange rates of the cryptocurrencies and the dynamics of the created content on the Internet by their users. The main goals of this thesis are to analyze unstructured data focusing on the development of the topics for different cryptocurrencies and to interconnect the key topics with the changes of the exchange rates of this cryptocurrencies. In order to successfully obtain goals I describe process and methods of the analysis unstructured data, explain the issues of the cryptocurrencies, identify a source of the downloading data and choose some cryptocurrencies for subsequent analysis. The contribution of this thesis is the identification of the development key topics in the chosen source, finding topics, which are the most under discussion about the chosen five cryptocurrencies. Also this thesis detects the existence of the reaction of the users to the changes of exchange rates of the chosen cryptocurrencies and describes it. Particular contribution of this thesis is a practical demonstration of the application the elected technology for analyzing unstructured data. Also additional contribution is the methods using in the analytical part. Structure of this thesis is divided to the three main parts. The first part describes the analysis of unstructured data theoretically and explain the issues of cryptocurrencies. The second part devotes a draft of the solution for creating an analytical part. The third part maps the most discussion cryptocurrencies in the chosen source, identifies the development of the key topics for five chosen cryptocurrencies Bitcoin, Nxt, Dash, Monero and Counterparty and interconnects the key topics this cryptocurrencies with changes their exchange rates in relation to American dollar.

The first part of the dissertation aims to determine whether and how variances in reinsurance relationships impact insurers' financial performance during the sample period of 2002-2012. Such impact on insurers' financial performance is measured by accounting measurements of ROA and ROE and by the efficiency scores (cost, revenue, and profit) estimated using data envelopment analysis (DEA). This essay analyzes how the usage of captive reinsurance affects life insurers’ firm performance using multivariate regression model. Results show that firm performance is negatively related to captive reinsurance arrangements. The second essay analyzes the value effects of mergers and acquisitions (M&As) in the global insurance industry by conducting an event study of M&A transactions that occurred during the period of 1990-2014, including two M&A waves before the financial crisis and the M&A activities after it. Our results show that (1) M&As are value-enhancing for both acquirers and targets over the whole sample period; (2) for acquirers, within-border transactions are more likely to be value-enhancing, while for targets, both cross-border and within-border transactions are value-enhancing; and (3) for acquirers, the cross-industry M&As are more likely to be value-enhancing, while for targets both cross- and within- border M&As are value-enhancing. / Business Administration/Risk Management and Insurance

The aim of this thesis is to address the implications of Basel III regulation on counterparty credit risk. We analysed the development of OTC market, we addressed systemic risk and the way how central counterparties could mitigate or spread the contagion among banks. We used simulated data to develop a stress test model to find out the impact of counterparty credit risk on banks' capital requirements, in case the interest rate increased extensively. Six pos- sible scenarios of interest rate levels were developed with ascending order of the IR level. From these scenarios we computed the exposure levels and credit valuation adjustment (CVA) as the market value of counterparty credit risk. We came to the following conclusions: (1) Czech banks have enough capital to withstand any interest rate increase in any scenario. (2) Banks with high expo- sure to derivatives like Bank of America, Citibank and JP Morgan would face severe problems if the interest rate increased. (3) There is no direct correlation between credit valuation adjustment and interest rate, the CVA increases faster with the increase of the interest rate.

In the thesis we analyse changes in derivatives valuation after the financial crisis and their impact on behaviour of financial institutions. We focus mainly on the changes related to counterparty credit risk and valuation adjustments. We describe in economical terms the relationship between counterparty credit risk and traditional credit risk, we also introduce management and modelling of this risk. In second part of the study we analyse the regulatory framework, in particular new capital requirement and mandatory central clearing of OTC derivatives. We discuss inconsistencies between regulatory and internal approaches to the counterparty risk measurement and also significant systemic risk connected to central counterparties. Finally we investigate the impact of changes in derivatives valuation on banks in both the EU and the Czech Republic. Specifically we are interested in optimal approach to entering into derivative trade.

After the financial crisis of 2008 regulators found that the derivative market, where financial institutions traded OTC derivatives with each other, played a significantrole in triggering the crisis. This led to the emergence of Counterparty Credit Risk(CCR) which is used to measure the exposure banks have to their counterparties. In simple terms CCR is a mix of Market and Credit risk which defines the risk that your counter party will go into bankruptcy. CCR involves the risk factors used in market risk since all of the derivatives are based on underlying assets such as interest rate and currencies. The thesis will focus on how one can backtest individual risk factors driving the value of OTC derivatives. We will present different Monte Carlo simulation techniques that are being used to simulate and represent all possible future outcomes for the risk factors. In order to better understand the performance of a chosen model and how to adjust the calibration window for the ingoing parameters, two different approaches are presented,Quantitative Backtesting and Statistical Backtesting. As an extension to this, a portfolio of interest rate Swaps are backtested whose value are driven by the evolution of the underlying risk factors. The backtesting ofthe portfolio is done with netting. The time horizon for the backtesting procedureis 2010-2020 giving the user up to 261 independent observations with a forecast length of 14 days. Both of the backtesting methods provide the practitioner with a graphical results guiding the user to choose an appropriate model and calibration method for simulating the risk factors. We found that a combination of the two approaches provides the best result. Hence, no backtesting method is superior the other. Instead they complement each other and should be used simultaneously. Using the two backtesting methods one can find a model that perfectly fit the underlying distribution of risk factors, theoretically. However, one should be careful since there will always be uncertainty about the future and there is no guarantee that tomorrow will follow historical evolution exactly.

Machine learning is a rapidly growing field within artificial intelligence that an increasing number of individuals and corporations are beginning to utilize. In recent times, the financial sector has also started to recognize the potential of these techniques and methods. Nasdaq Clearing is responsible for managing the clearing business for the clearinghouse's members, and the objective of this thesis has been to explore the possibilities of using machine learning to replicate a subpart of the SPAN model's margin call calculations, known as initial margin, in the commodities market. The purpose of replicating SPAN's initial margin calculations is to open up for possibilities to create transparency and understanding in how the input variables affect the output. In the long run, we hope to broaden the insights on how one can use machine learning within the margin call processes. Various machine learning algorithms, primarily focused on regression tasks but also a few classification ones, have been employed to replicate the initial margin size. The primary objective of the methodology was to determine the algorithm that demonstrated the best performance in obtaining values that were as close as possible to the actual initial margin values. The findings revealed that a model composed of a combination of classification and regression, with non-parametric algorithms such as Random Forest and KNN, performed the best in both cases. Our conclusion is that the developed model possesses the ability to effectively compute the size of the initial margin and thus accomplishes its objective. / Maskininlärning är ett snabbt växande område inom artificiell intelligens som allt fler individer och företag börjar använda. Finanssektorn har nu också börjat undersöka hur dessa tekniker och metoder kan skapa värde. Nasdaq Clearing hanterar clearingverksamheten för clearinghusets medlemmar och syftet med denna uppsats har varit att undersöka möjligheterna att använda maskininlärning för att replikera en del av SPAN-modellens beräkningar av marginkravet som kallas Initial Marginal. Syftet med att replikera SPANs initiala marginberäkningar är att öppna upp för möjligheter att skapa transparens och förståelse för hur inputvariablernapåverkar outputen. På sikt hoppas vi kunna bredda insikterna hur maskininlärningslösningar skulle kunna användas inom "Margin Call"- processen. De metoder som användes för att replikera storleken på Initial Margin var olika maskininlärningsalgoritmer, främst fokuserade på regressionsuppgifter men några klassificeringsalgoritmer användes också. Fokus i metoden var att hitta vilken algoritm som presterade bäst, det vill säga den algoritm som predikterade närmst de faktiska värdena för Initial Margin. Resultatet visade sig vara en modell som kombinerade klassificering och regression, där icke-parametriska algoritmer såsom Random Forest och KNN var de som presterade bäst i båda fallen. Vår slutsats är att den utvecklade modellen har en god förmåga att beräkna storleken på Initial Margin och därmed uppfyller den sitt syfte.

Цель исследования – реализация проекта по автоматической проверки контрагентов с использованием современных технологий и методов машинного обучения, выявление проблем конкурентного окружения и разработка способов их решения. Исследование также направлено на определение эффективности такого проекта по автоматизации процесса по сравнению с ручными методами проверки контрагентов и определение его потенциальной экономической выгоды. Объект исследования: процесс автоматизации проверки контрагента на предприятии. Основными результатами работы являются выделенные текущие проблемы в процессе автоматизации проверки контрагента, разработанные пути их решения и собственный проект по автоматизации процесса. Практическая значимость исследования заключается в применении авторских предложений по совершенствованию процесса автоматизации проверки контрагента на предприятия. / The purpose of the study is to implement a project to automatically check counterparties using modern technologies and machine learning methods, identify problems in the competitive environment and develop ways to solve them. The study also aims to determine the effectiveness of such a project to automate the process compared to manual methods of verifying counterparties and determine its potential economic benefits. Object of study: the process of automating the verification of a counterparty at an enterprise. The main results of the work are the identified current problems in the process of automating the verification of a counterparty, developed ways to solve them and our own project to automate the process. The practical significance of the study lies in the application of the author’s proposals for improving the process of automating the verification of counterparties at enterprises.

This dissertation investigates the relationships between reinsurance activities and primary insurers' financial performance in U.S. property-liability insurance market from several perspectives. The first essay investigates the relationship between ceding insurer performance and the affiliation, domicile, and authorization of its counterparties. Specifically, we provide empirical evidence that ceding insurer financial performance is positively related to the use of affiliated reinsurance, the use of foreign reinsurance, and the use of affiliated reinsurance that is ceded to the low-tax, lightly-regulated domiciled counterparties; and negatively related to the use of unauthorized reinsurance. These results are consistent with the cost of information asymmetry theory. The second essay investigates reinsurance counterparty relationships in U.S. property-liability insurance. Firm-specific characteristics determinants of counterparty relationships are examined. We also analyze the relationship between firm performance and reinsurance counterparty relationships. We find that concentration in reinsurance counterparties, especially in unaffiliated counterparties, is adversely related to insurer performance due to higher information asymmetry. On the other hand, relationship with foreign counterparties is positive related to performance, suggesting the foreign reinsurers may have a favorable position in terms of tax treatment, specialized service, among other factors. / Business Administration/Risk Management and Insurance