[pt] Nesta dissertação avalia-se o grau de associação entre pares de excessos de retornos, simultâneos e defasados no tempo, usando-se o conceito de cópulas. Cópulas assimétricas são ajustadas aos pares de distribuições de retornos e coeficientes de dependência de cauda, as medidas de interdependência e contágio baseadas nessas cópulas, são calculados para 10 pares de índices de mercados. Tais coeficientes balizam a escolha do par de ativos com melhor desempenho em períodos de estresse. Se excessos defasados são incluídos, então estes coeficientes também indicam a direção e intensidade de propagação das crises (contágio). Os resultados encontrados na nossa investigação mostram que a técnica utilizada é eficaz na montagem de carteiras em que se pretende aproveitar os ganhos extremos conjuntos dos ativos e, ao mesmo tempo, evitar perdas extremas conjuntas. O uso de retornos defasados, porém, foi um artifício pouco producente, refletindo possivelmente o contágio quase instantâneo entre os mercados financeiros mundiais, nos dias de hoje. / [en] In this dissertation we evaluate the degree of association between pairs of excess of returns, simultaneous and lagged, using the concept of copulas. Asymmetric copulas are fitted to 10 pairs of distributions of returns of world markets índices. From these copulas coefficients of tail dependence are obtained for the right and left tails. Isong those coefficients as measures of cross dependence and contagion between markets one can pick the pair of returns that show the best performance in periods of stress. If lagged excess of returns are included, then these coefficients provide information on the direction and intensity of the contagion spread. Our results have shown that such technique isd efficent in constructing a portfolio in which one wants to take advantage of joint extreme gains of pairs of returns and, simultaneously, avoid losses associated with the occurrence of joint negative extremes. The use of lagged returns in this context hás shown no extra gain, maybe reflecting the fact that, nowadays, the spread of contagion between world financial markets is almost instantaneous.

Urbanisation alters the hydrologic response of a catchment, resulting in increased runoff rates and volumes, and loss of infiltration and base flow. Quantification of uncertainties is important in hydrologic designs of urban infrastructure. Major sources of uncertainty in the Intensity Duration Frequency (IDF) relationships are due to insufficient quantity and quality of data leading to parameter uncertainty and, in the case of projections of future IDF relationships under climate change, uncertainty arising from use of multiple General Circulation Models (GCMs) and scenarios. The work presented in the thesis presents methodologies to quantify the uncertainties arising from parameters of the distribution fitted to data and the multiple GCMs using a Bayesian approach. High uncertainties in GEV parameters and return levels are observed at shorter durations for Bangalore City. Twenty six GCMs from the CMIP5 datasets, along with four RCP scenarios are considered for studying the effects of climate change. It is observed that the uncertainty in short duration rainfall return levels is high when compared to the longer durations. Further it is observed that parameter uncertainty is large compared to the model uncertainty. Disaggregation of precipitation extremes from larger time scales to smaller time scales when the extremes are modeled with the GPD is burdened with difficulties arising from varying thresholds for different durations. In this study, the scale invariance theory is used to develop a disaggregation model for precipitation extremes exceeding specified thresholds. A scaling relationship is developed for a range of thresholds obtained from a set of quantiles of non-zero precipitation of different durations. The disaggregation model is applied to precipitation datasets of Berlin City, Germany and Bangalore City, India. From both the applications, it is observed that the uncertainty in the scaling exponent has a considerable effect on uncertainty in scaled parameters and return levels of shorter durations. A Bayesian hierarchical model is used to obtain spatial distribution of return levels of precipitation extremes in urban areas and quantify the associated uncertainty. Applicability of the methodology is demonstrated with data from 19 telemetric rain gauge stations in Bangalore City, India. For this case study, it is inferred that the elevation and mean monsoon precipitation are the predominant covariates for annual maximum precipitation. For the monsoon maximum precipitation, it is observed that the geographic covariates dominate while for the summer maximum precipitation, elevation and mean summer precipitation are the predominant covariates. In this work, variation in the dependence structure of extreme precipitation within an urban area and its surrounding non-urban areas at various durations is studied. The Berlin City, Germany, with surrounding non-urban area is considered to demonstrate the methodology. For this case study, the hourly precipitation shows independence within the city even at small distances, whereas the daily precipitation shows a high degree of dependence. This dependence structure of the daily precipitation gets masked as more and more surrounding non-urban areas are included in the analysis. The geographical covariates are seen to be predominant within the city and the climatological covariates prevail when non-urban areas are added. These results suggest the importance of quantification of dependence structure of spatial precipitation at the sub-daily timescales, as well as the need to more precisely model spatial extremes within the urban areas. The work presented in this thesis thus contributes to quantification of uncertainty in precipitation extremes through developing methodologies for generating probabilistic future IDF relationships under climate change, spatial mapping of probabilistic return levels and modeling dependence structure of extreme precipitation in urban areas at fine resolutions.

The thesis is focused on extreme value theory and its applications. Initially, extreme value distribution is introduced and its properties are discussed. At this basis are described two models mostly used for an extreme value analysis, i.e. the block maxima model and the Pareto-distribution threshold model. The first one takes advantage in its robustness, however recently the threshold model is mostly preferred. Although the threshold choice strongly affects estimation quality of the model, an optimal threshold selection still belongs to unsolved issues of this approach. Therefore, the thesis is focused on techniques for proper threshold identification, mainly on adaptive methods suitable for the use in practice. For this purpose a simulation study was performed and acquired knowledge was applied for analysis of precipitation records from South-Moravian region. Further on, the thesis also deals with extreme value estimation within a stationary series framework. Usually, an observed time series needs to be separated to obtain approximately independent observations. The use of the advanced theory for stationary series allows to avoid the entire separation procedure. In this context the commonly applied separation techniques turn out to be quite inappropriate in most cases and the estimates based on theory of stationary series are obtained with better precision.

The US sub-prime mortgage crisis greatly affected not only the US economy but also other countries in the world. This thesis employs the extreme value theory and Value at Risk (VaR) analysis to assess the impact of the US sub-prime mortgage crisis on various stock markets of the MSCI indexes, including 10 countries and 7 areas. It is reasonable to guess that VaR value should increase after the crisis. The empirical analyses on these indexes conclude that (1) the American market indexes not only do not agree with the guess after the crisis but four American indexes are identical; (2) not all the Asia market indexes consist with the guess; (3) the European market indexes agree with the guess; (4) MSCI AC PACIFIC, NEW ZEALAND, and AUSTRALIA consist with the guess; (5) the behavior for the positive log returns is different from that for the negative returns in some MSCI indexes. Over speaking, the impacts of US sub-prime mortgage crisis on those countries are not the same.

Les précipitations et les débits des cours d'eau constituent les deux variables hydrométéorologiques les plus importantes pour l'analyse des bassins versants. Ils fournissent des informations fondamentales pour la gestion intégrée des ressources en eau, telles que l’approvisionnement en eau potable, l'hydroélectricité, les prévisions d'inondations ou de sécheresses ou les systèmes d'irrigation.Dans cette thèse de doctorat sont abordés deux problèmes distincts. Le premier prend sa source dans l’étude des débits des cours d’eau. Dans le but de bien caractériser le comportement global d'un bassin versant, de longues séries temporelles de débit couvrant plusieurs dizaines d'années sont nécessaires. Cependant les données manquantes constatées dans les séries représentent une perte d'information et de fiabilité, et peuvent entraîner une interprétation erronée des caractéristiques statistiques des données. La méthode que nous proposons pour aborder le problème de l'imputation des débits se base sur des modèles de régression dynamique (DRM), plus spécifiquement, une régression linéaire multiple couplée à une modélisation des résidus de type ARIMA. Contrairement aux études antérieures portant sur l'inclusion de variables explicatives multiples ou la modélisation des résidus à partir d'une régression linéaire simple, l'utilisation des DRMs permet de prendre en compte les deux aspects. Nous appliquons cette méthode pour reconstruire les données journalières de débit à huit stations situées dans le bassin versant de la Durance (France), sur une période de 107 ans. En appliquant la méthode proposée, nous parvenons à reconstituer les débits sans utiliser d'autres variables explicatives. Nous comparons les résultats de notre modèle avec ceux obtenus à partir d'un modèle complexe basé sur les analogues et la modélisation hydrologique et d'une approche basée sur le plus proche voisin. Dans la majorité des cas, les DRMs montrent une meilleure performance lors de la reconstitution de périodes de données manquantes de tailles différentes, dans certains cas pouvant allant jusqu'à 20 ans.Le deuxième problème que nous considérons dans cette thèse concerne la modélisation statistique des quantités de précipitations. La recherche dans ce domaine est actuellement très active car la distribution des précipitations exhibe une queue supérieure lourde et, au début de cette thèse, il n'existait aucune méthode satisfaisante permettant de modéliser toute la gamme des précipitations. Récemment, une nouvelle classe de distribution paramétrique, appelée distribution généralisée de Pareto étendue (EGPD), a été développée dans ce but. Cette distribution exhibe une meilleure performance, mais elle manque de flexibilité pour modéliser la partie centrale de la distribution. Dans le but d’améliorer la flexibilité, nous développons, deux nouveaux modèles reposant sur des méthodes semiparamétriques.Le premier estimateur développé transforme d'abord les données avec la distribution cumulative EGPD puis estime la densité des données transformées en appliquant un estimateur nonparamétrique par noyau. Nous comparons les résultats de la méthode proposée avec ceux obtenus en appliquant la distribution EGPD paramétrique sur plusieurs simulations, ainsi que sur deux séries de précipitations au sud-est de la France. Les résultats montrent que la méthode proposée se comporte mieux que l'EGPD, l’erreur absolue moyenne intégrée (MIAE) de la densité étant dans tous les cas presque deux fois inférieure.Le deuxième modèle considère une distribution EGPD semiparamétrique basée sur les polynômes de Bernstein. Plus précisément, nous utilisons un mélange creuse de densités béta. De même, nous comparons nos résultats avec ceux obtenus par la distribution EGPD paramétrique sur des jeux de données simulés et réels. Comme précédemment, le MIAE de la densité est considérablement réduit, cet effet étant encore plus évident à mesure que la taille de l'échantillon augmente. / Precipitation and streamflow are the two most important meteorological and hydrological variables when analyzing river watersheds. They provide fundamental insights for water resources management, design, or planning, such as urban water supplies, hydropower, forecast of flood or droughts events, or irrigation systems for agriculture.In this PhD thesis we approach two different problems. The first one originates from the study of observed streamflow data. In order to properly characterize the overall behavior of a watershed, long datasets spanning tens of years are needed. However, the quality of the measurement dataset decreases the further we go back in time, and blocks of data of different lengths are missing from the dataset. These missing intervals represent a loss of information and can cause erroneous summary data interpretation or unreliable scientific analysis.The method that we propose for approaching the problem of streamflow imputation is based on dynamic regression models (DRMs), more specifically, a multiple linear regression with ARIMA residual modeling. Unlike previous studies that address either the inclusion of multiple explanatory variables or the modeling of the residuals from a simple linear regression, the use of DRMs allows to take into account both aspects. We apply this method for reconstructing the data of eight stations situated in the Durance watershed in the south-east of France, each containing daily streamflow measurements over a period of 107 years. By applying the proposed method, we manage to reconstruct the data without making use of additional variables, like other models require. We compare the results of our model with the ones obtained from a complex approach based on analogs coupled to a hydrological model and a nearest-neighbor approach, respectively. In the majority of cases, DRMs show an increased performance when reconstructing missing values blocks of various lengths, in some of the cases ranging up to 20 years.The second problem that we approach in this PhD thesis addresses the statistical modeling of precipitation amounts. The research area regarding this topic is currently very active as the distribution of precipitation is a heavy-tailed one, and at the moment, there is no general method for modeling the entire range of data with high performance. Recently, in order to propose a method that models the full-range precipitation amounts, a new class of distribution called extended generalized Pareto distribution (EGPD) was introduced, specifically with focus on the EGPD models based on parametric families. These models provide an improved performance when compared to previously proposed distributions, however, they lack flexibility in modeling the bulk of the distribution. We want to improve, through, this aspect by proposing in the second part of the thesis, two new models relying on semiparametric methods.The first method that we develop is the transformed kernel estimator based on the EGPD transformation. That is, we propose an estimator obtained by, first, transforming the data with the EGPD cdf, and then, estimating the density of the transformed data by applying a nonparametric kernel density estimator. We compare the results of the proposed method with the ones obtained by applying EGPD on several simulated scenarios, as well as on two precipitation datasets from south-east of France. The results show that the proposed method behaves better than parametric EGPD, the MIAE of the density being in all the cases almost twice as small.A second approach consists of a new model from the general EGPD class, i.e., we consider a semiparametric EGPD based on Bernstein polynomials, more specifically, we use a sparse mixture of beta densities. Once again, we compare our results with the ones obtained by EGPD on both simulated and real datasets. As before, the MIAE of the density is considerably reduced, this effect being even more obvious as the sample size increases.

許多觀察到的時間序列資料，多呈現高峰厚尾（leptokurtic）的現象，本文引用時間序列資料為Paretian分配之假設，估計各個國家股價指數報酬率於不同頻率資料下之最大級數動差，以觀察其厚尾程度。實證結果發現，各個國家指數報酬率於不同頻率資料下之四級以上動差大部分存在，且不隨資料之頻率不同，而有不同的表現。由此可推論，各個國家股價指數報酬率之歷史分配，其離群值之活動並不嚴重。接著，利用樣本分割預測檢定（Sample Split Prediction Test）來檢定所觀察各個國家股價指數報酬率於同一樣本期間內，其左右尾之厚尾程度是否一致，及檢定所觀察各個國家指數報酬率於跨期間左尾或右尾之厚尾程度是否穩定。在同一樣本期間，檢定時間序列之左右尾之厚尾程度是否一致之檢定中，發現各個國家指數報酬率在所觀察樣本期間內，其左右尾之厚尾程度大致相同；而在跨期間之樣本分割預測檢定中，發現各個國家指數報酬率在像是1987年10月美國股市大崩盤、1990年至1991年間之波斯灣戰爭、1997年亞洲金融風暴等事件前後，其左（右）尾之厚尾程度有顯著差異。最後提出Cusum of Squares檢定，係用於檢定一時間序列資料在所觀察之樣本期間內，其非條件變異數是否為一常數。 Cusum of Squares檢定之檢定結果顯示，本文之各個國家指數報酬率在所觀察之樣本期間內，其非條件變異數並非為一常數。進一步觀察各個國家指數報酬率之Cusum of Squares圖，並綜合前述跨期間樣本分割預測檢定之結果，可推論在處理較長樣本期間之時間序列資料可能遇到結構性變動之情況時，跨期間之樣本分割預測檢定及Cusum of Squares檢定可提供結構性變動可能發生之時點。

To estimate the risk of a loss occurring for insurance takers is a difficult task in the insurance industry. It is an even more difficult task to price the risk for reinsurance companies which insures the primary insurers. Insurance that is bought by an insurance company, the cedent, from another insurance company, the reinsurer, is called treaty reinsurance. This type of reinsurance is the main focus in this thesis. A very common risk to insure, is the risk of fire in municipal and commercial properties which is the risk that is priced in this thesis. This thesis evaluates Länsförsäkringar AB's current pricing model which calculates the risk premium for Risk XL contracts. The goal of this thesis is to find areas of improvement for tail risk pricing. The risk premium can be calculated commonly by using one of three different types of pricing models, experience rating, exposure rating and frequency-severity rating. This thesis focuses on frequency-severity pricing, which is a model that assumes independence between the frequency and the severity of losses, and therefore splits the two into separate models. This is a very common model used when pricing Risk XL contracts. The risk premium is calculated with the help of loss data from two insurance companies, from a Norwegian and a Finnish insurance company. The main focus of this thesis is to price the risk with the help of extreme value theory, mainly with the method of moments method to model the frequency of losses, and peaks over threshold model to model the severity of the losses. In order to model the estimated frequency of losses by using the method of moments method, two distributions are compared, the Poisson and the negative binomial distribution. There are different distributions that can be used to model the severity of losses. In order to evaluate which distribution is optimal to use, two different Goodness of Fit tests are applied, the Kolmogorov-Smirnov and the Anderson-Darling test. The Peaks over threshold model is a model that can be used with the Pareto distribution. With the help of the Hill estimator we are able to calculate a threshold $u$, which regulates the tail of the Pareto curve. To estimate the rest of the ingoing parameters in the generalized Pareto distribution, the maximum likelihood and the least squares method are used. Lastly, the bootstrap method is used to estimate the uncertainty in the price which was calculated with the help of the estimated parameters. From this, empirical percentiles are calculated and set as guidelines to where the risk premium should lie between, in order for both the data sets to be considered fairly priced. / Att uppskatta risken för en skada ska inträffa för försäkringstagarna är svår uppgift i försäkringsbranschen. Det är en ännu svårare uppgift är att prissätta risken för återförsäkringsbolag som försäkrar direktförsäkrarna. Den försäkringen som köps av direkförsäkrarna, cedenten, från återförsäkrarna kallas treaty återförsäkring. Denna typ av återförsäkring är den som behandlas i denna avhandlig. En vanlig risk att prisätta är brandrisken för kommunala och industriella byggnader, vilket är risken som prissätts i denna avhandlnig. Denna avhandling utvärderar Länsförsäkringar AB's nuvarande prissättning som beräknar riskpremien för Risk XL kontrakt.Målet med denna avhandling är att hitta förbättringsområden för långsvansad affär. Riskpremien kan beräknas med hjälp av tre vanliga typer av prissättningsmodeller, experience rating, exposure rating och frequency-severity raring. Denna tes fokuserar endast på frequency-severity rating, vilket är en modell som antar att frekevensen av skador och storleken av de är oberoende, de delas därmed upp de i separata modeller. Detta är en väldigt vanlig modell som används vid prissättning av Risk XL kontrakt.Riskpremien beräknas med hjälp av skadedata från två försäkringsbolag, ett norskt och ett finskt försäkringsbolag.Det huvudsakliga fokuset i denna avhandling är att prissätta risken med hjälp av extremevärdesteori, huvudsakligen med hjälp av momentmetoden för att modellera frekvensen av skador och peaks over threshold metoden för att modellera storleken av de skadorna.För att kunna modellera den förväntade frekvensen av skador med hjälp av moment metoden så jämförs två fördelingar, Poissonfördelingen och den negativa binomialfördelningen. Det finns ett antal fördelningar som kan användas för att modellera storleken av skadorna. För att kunna avgöra vilken fördeling som är bäst att använda så har två olika Goodness of Fit test applicerats, Kolmogorov-Smirnov och Anderson-Darling testet.Peaks over threhsold modellen är en modell som kan användas med Paretofördelningen. Med hjälp av Hillestimatorn så beräknas en tröskel $u$ som regulerar paretokurvans uteseende. För att beräkna de resterande parametrarna i den generaliserade Paretofördelningen används maximum likliehood och minsta kvadratmetoden. Slutligen används bootstrap metoden för att skatta osäkerheten i risk premien som satts med hjälp av de skattade parametrarna. Utifrån den metoden så skapas percentiler som blir en riktlinje för vart risk premien bör ligga för de datasetten för att kunna anses vara rättvist prissatt.

This thesis investigates applying the semiparametric method Peaks-Over-Threshold on data generated from a Monte Carlo simulation when estimating the financial risk measures Value-at-Risk and Expected Shortfall. The goal is to achieve a faster convergence than a Monte Carlo simulation when assessing extreme events that symbolise the worst outcomes of a financial portfolio. Achieving a faster convergence will enable a reduction of iterations in the Monte Carlo simulation, thus enabling a more efficient way of estimating risk measures for the portfolio manager.  The financial portfolio consists of US life insurance policies offered on the secondary market, gathered by our partner RessCapital. The method is evaluated on three different portfolios with different defining characteristics.  In Part I an analysis of selecting an optimal threshold is made. The accuracy and precision of Peaks-Over-Threshold is compared to the Monte Carlo simulation with 10,000 iterations, using a simulation of 100,000 iterations as the reference value. Depending on the risk measure and the percentile of interest, different optimal thresholds are selected.  Part II presents the result with the optimal thresholds from Part I. One can conclude that Peaks-Over-Threshold performed significantly better than a Monte Carlo simulation for Value-at-Risk with 10,000 iterations. The results for Expected Shortfall did not achieve a clear improvement in terms of precision, but it did show improvement in terms of accuracy.  Value-at-Risk and Expected Shortfall at the 99.5th percentile achieved a greater error reduction than at the 99th. The result therefore aligned well with theory, as the more "rare" event considered, the better the Peaks-Over-Threshold method performed.  In conclusion, the method of applying Peaks-Over-Threshold can be proven useful when looking to reduce the number of iterations since it do increase the convergence of a Monte Carlo simulation. The result is however dependent on the rarity of the event of interest, and the level of precision/accuracy required. / Det här examensarbetet tillämpar metoden Peaks-Over-Threshold på data genererat från en Monte Carlo simulering för att estimera de finansiella riskmåtten Value-at-Risk och Expected Shortfall. Målet med arbetet är att uppnå en snabbare konvergens jämfört med en Monte Carlo simulering när intresset är s.k. extrema händelser som symboliserar de värsta utfallen för en finansiell portfölj. Uppnås en snabbare konvergens kan antalet iterationer i simuleringen minskas, vilket möjliggör ett mer effektivt sätt att estimera riskmåtten för portföljförvaltaren.  Den finansiella portföljen består av amerikanska livförsäkringskontrakt som har erbjudits på andrahandsmarknaden, insamlat av vår partner RessCapital. Metoden utvärderas på tre olika portföljer med olika karaktär.  I Del I så utförs en analys för att välja en optimal tröskel för Peaks-Over-Threshold. Noggrannheten och precisionen för Peaks-Over-Threshold jämförs med en Monte Carlo simulering med 10,000 iterationer, där en Monte Carlo simulering med 100,000 iterationer används som referensvärde. Beroende på riskmått samt vilken percentil som är av intresse så väljs olika trösklar.  I Del II presenteras resultaten med de "optimalt" valda trösklarna från Del I. Peaks-over-Threshold påvisade signifikant bättre resultat för Value-at-Risk jämfört med Monte Carlo simuleringen med 10,000 iterationer. Resultaten för Expected Shortfall påvisade inte en signifikant förbättring sett till precision, men visade förbättring sett till noggrannhet.  För både Value-at-Risk och Expected Shortfall uppnådde Peaks-Over-Threshold en större felminskning vid 99.5:e percentilen jämfört med den 99:e. Resultaten var därför i linje med de teoretiska förväntningarna då en högre percentil motsvarar ett extremare event.  Sammanfattningsvis så kan metoden Peaks-Over-Threshold vara användbar när det kommer till att minska antalet iterationer i en Monte Carlo simulering då resultatet visade att Peaks-Over-Threshold appliceringen accelererar Monte Carlon simuleringens konvergens. Resultatet är dock starkt beroende av det undersökta eventets sannolikhet, samt precision- och noggrannhetskravet.

The global financial crisis that took off in 2007 gave rise to several adjustments of the risk regulation for banks. An extensive adjustment, that is to be implemented in 2019, is the Fundamental Review of the Trading Book (FRTB). It proposes to use Expected Shortfall (ES) as risk measure instead of the currently used Value at Risk (VaR), as well as applying varying liquidity horizons based on the various risk levels of the assets involved. A major difficulty of implementing the FRTB lies within the backtesting of ES. Righi and Ceretta proposes a robust ES backtest based on Monte Carlo simulation. It is flexible since it does not assume any probability distribution and can be performed without waiting for an entire backtesting period. Implementing some commonly used VaR backtests as well as the ES backtest by Righi and Ceretta, yield a perception of which risk models that are the most accurate from both a VaR and an ES backtesting perspective. It can be concluded that a model that is satisfactory from a VaR backtesting perspective does not necessarily remain so from an ES backtesting perspective and vice versa. Overall, the models that are satisfactory from a VaR backtesting perspective turn out to be probably too conservative from an ES backtesting perspective. Considering the confidence levels proposed by the FRTB, from a VaR backtesting perspective, a risk measure model with a normal copula and a hybrid distribution with the generalized Pareto distribution in the tails and the empirical distribution in the center along with GARCH filtration is the most accurate one, as from an ES backtesting perspective a risk measure model with univariate Student’s t distribution with ⱱ ≈ 7 together with GARCH filtration is the most accurate one for implementation. Thus, when implementing the FRTB, the bank will need to compromise between obtaining a good VaR model, potentially resulting in conservative ES estimates, and obtaining a less satisfactory VaR model, possibly resulting in more accurate ES estimates. The thesis was performed at SAS Institute, an American IT company that develops software for risk management among others. Targeted customers are banks and other financial institutions. Investigating the FRTB acts a potential advantage for the company when approaching customers that are to implement the regulation framework in a near future. / Den globala finanskrisen som inleddes år 2007 ledde till flertalet ändringar vad gäller riskreglering för banker. En omfattande förändring som beräknas implementeras år 2019, utgörs av Fundamental Review of the Trading Book (FRTB). Denna föreslår bland annat användande av Expected Shortfall (ES) som riskmått istället för Value at Risk (VaR) som används idag, liksom tillämpandet av varierande likviditetshorisonter beroende på risknivåerna för tillgångarna i fråga. Den huvudsakliga svårigheten med att implementera FRTB ligger i backtestingen av ES. Righi och Ceretta föreslår ett robust ES backtest som baserar sig på Monte Carlo-simulering. Det är flexibelt i den mening att det inte antar någon specifik sannolikhetsfördelning samt att det går att implementera utan att man behöver vänta en hel backtestingperiod. Vid implementation av olika standardbacktest för VaR, liksom backtestet för ES av Righi och Ceretta, fås en uppfattning av vilka riskmåttsmodeller som ger de mest korrekta resultaten från både ett VaR- och ES-backtestingperspektiv. Sammanfattningsvis kan man konstatera att en modell som är acceptabel från ett VaR-backtestingperspektiv inte nödvändigtvis är det från ett ES-backtestingperspektiv och vice versa. I det hela taget har det visat sig att de modeller som är acceptabla ur ett VaR-backtestingperspektiv troligtvis är för konservativa från ett ESbacktestingperspektiv. Om man betraktar de konfidensnivåer som föreslagits i FRTB, kan man ur ett VaR-backtestingperspektiv konstatera att en riskmåttsmodell med normal-copula och en hybridfördelning med generaliserad Pareto-fördelning i svansarna och empirisk fördelning i centrum tillsammans med GARCH-filtrering är den bäst lämpade, medan det från ett ES-backtestingperspektiv är att föredra en riskmåttsmodell med univariat Student t-fördelning med ⱱ ≈ 7 tillsammans med GARCH-filtrering. Detta innebär att när banker ska implementera FRTB kommer de behöva kompromissa mellan att uppnå en bra VaR-modell som potentiellt resulterar i för konservativa ES-estimat och en modell som är mindre bra ur ett VaRperspektiv men som resulterar i rimligare ES-estimat. Examensarbetet genomfördes vid SAS Institute, ett amerikanskt IT-företag som bland annat utvecklar mjukvara för riskhantering. Tänkbara kunder är banker och andra finansinstitut. Denna studie av FRTB innebär en potentiell fördel för företaget vid kontakt med kunder som planerar implementera regelverket inom en snar framtid. / Riskhantering, finansiella tidsserier, Value at Risk, Expected Shortfall, Monte Carlo-simulering, GARCH-modellering, Copulas, hybrida distributioner, generaliserad Pareto-fördelning, extremvärdesteori, Backtesting, likviditetshorisonter, Basels regelverk

Le fonctionnement d'un système, de façon générale, peut être affecté par un incident imprévu. Lorsque cet incident a de lourdes conséquences tant sur l'intégrité du système que sur la qualité de ses produits, on dit alors qu'il se situe dans le cadre des événements dits extrêmes. Ainsi, de plus en plus les chercheurs portent un intérêt particulier à la modélisation des événements extrêmes pour diverses études telles que la fiabilité des systèmes et la prédiction des différents risques pouvant entraver le bon fonctionnement d'un système en général. C'est dans cette optique que s'inscrit la présente thèse. Nous utilisons la Théorie des Valeurs Extrêmes (TVE) et les statistiques d'ordre extrême comme outil d'aide à la décision dans la modélisation et la gestion des risques dans l'usinage et l'aviation. Plus précisément, nous modélisons la surface de rugosité de pièces usinées et la fiabilité de l'outil de coupe associé par les statistiques d'ordre extrême. Nous avons aussi fait une modélisation à l'aide de l'approche dite du "Peaks-Over Threshold, POT" permettant de faire des prédictions sur les éventuelles victimes dans l'Aviation Générale Américaine (AGA) à la suite d'accidents extrêmes. Par ailleurs, la modélisation des systèmes soumis à des facteurs d'environnement ou covariables passent le plus souvent par les modèles à risque proportionnel basés sur la fonction de risque. Dans les modèles à risque proportionnel, la fonction de risque de base est généralement de type Weibull, qui est une fonction monotone; l'analyse du fonctionnement de certains systèmes comme l'outil de coupe dans l'industrie a montré qu'un système peut avoir un mauvais fonctionnement sur une phase et s'améliorer sur la phase suivante. De ce fait, des modifications ont été apportées à la distribution de Weibull afin d'avoir des fonctions de risque de base non monotones, plus particulièrement les fonctions de risque croissantes puis décroissantes. En dépit de ces modifications, la prise en compte des conditions d'opérations extrêmes et la surestimation des risques s'avèrent problématiques. Nous avons donc, à partir de la loi standard de Gumbel, proposé une fonction de risque de base croissante puis décroissante permettant de prendre en compte les conditions extrêmes d'opérations, puis établi les preuves mathématiques y afférant. En outre, un exemple d'application dans le domaine de l'industrie a été proposé. Cette thèse est divisée en quatre chapitres auxquels s'ajoutent une introduction et une conclusion générales. Dans le premier chapitre, nous rappelons quelques notions de base sur la théorie des valeurs extrêmes. Le deuxième chapitre s'intéresse aux concepts de base de l'analyse de survie, particulièrement à ceux relatifs à l'analyse de fiabilité, en proposant une fonction de risque croissante-décroissante dans le modèle à risques proportionnels. En ce qui concerne le troisième chapitre, il porte sur l'utilisation des statistiques d'ordre extrême dans l'usinage, notamment dans la détection de pièces défectueuses par lots, la fiabilité de l'outil de coupe et la modélisation des meilleures surfaces de rugosité. Le dernier chapitre porte sur la prédiction d'éventuelles victimes dans l'Aviation Générale Américaine à partir des données historiques en utilisant l'approche "Peaks-Over Threshold" / The operation of a system in general may at any time be affected by an unforeseen incident. When this incident has major consequences on the system integrity and the quality of system products, then it is said to be in the context of extreme events. Thus, increasingly researchers have a particular interest in modeling such events with studies on the reliability of systems and the prediction of the different risks that can hinder the proper functioning of a system. This thesis takes place in this very perspective. We use Extreme Value Theory (EVT) and extreme order statistics as a decision support tool in modeling and risk management in industry and aviation. Specifically, we model the surface roughness of machined parts and the reliability of the associated cutting tool with the extreme order statistics. We also did a modeling using the "Peaks-Over Threshold, POT" approach to make predictions about the potential victims in the American General Aviation (AGA) following extreme accidents. In addition, the modeling of systems subjected to environmental factors or covariates is most often carried out by proportional hazard models based on the hazard function. In proportional hazard models, the baseline risk function is typically Weibull distribution, which is a monotonic function. The analysis of the operation of some systems like the cutting tool in the industry has shown that a system can deteriorated on one phase and improving on the next phase. Hence, some modifications have been made in the Weibull distribution in order to have non-monotonic basic risk functions, more specifically, the increasing-decreasing risk function. Despite these changes, taking into account extreme operating conditions and overestimating risks are problematics. We have therefore proposed from Gumbel's standard distribution, an increasingdecreasing risk function to take into account extreme conditions, and established mathematical proofs. Furthermore, an example of the application in the field of industry was proposed. This thesis is organized in four chapters and to this must be added a general introduction and a general conclusion. In the first chapter, we recall some basic notions about the Extreme Values Theory. The second chapter focuses on the basic concepts of survival analysis, particularly those relating to reliability analysis by proposing a function of increasing-decreasing hazard function in the proportional hazard model. Regarding the third chapter, it deals with the use of extreme order statistics in industry, particularly in the detection of defective parts, the reliability of the cutting tool and the modeling of the best roughness surfaces. The last chapter focuses on the prediction of potential victims in AGA from historical data using the Peaks-Over Threshold approach.