Esta dissertação revê as metodologias de teste de estresse em sistemas financeiros e descreve uma análise de cenário e um teste de estresse aplicado ao Brasil. Os cenários macroeconômicos são modelados por vetores auto-regressivos e o teste de estresse por um probit ordenado com efeitos aleatórios. Dados para o Brasil no período de 11/2002 a 11/2007 são usados para estimar os cenários macroeconômicos. A experiência brasileira de 2002 e início de 2003 parecem particularmente interessante para um teste de estresse por incluir uma grande volatilidade de mercado com taxas de inadimplência e perdas bancárias acima da média. A introdução de cenários macroeconômicos no teste de estresse do sistema financeiro brasileiro e o uso de regressões de dados categorizados com dados em painel são a principal contribuição deste trabalho. O modelo pode ser estendido para usar dados para um setor industrial especifico para identificar potenciais riscos de concentração de empréstimos. / This dissertation reviews financial system stress-testing methodologies and describes a scenario analysis and macro stress testing applied to Brazil. The macroeconomic scenarios are modeled by a vector autoregressive and the stress testing by a random effects ordered probit panel. Data for Brazil over the time period from 11/2002 to 11/2007 is used to estimate the macroeconomic scenarios. The Brazilian experience in 2002 and early 2003 appears particularly suited for macro stress-testing as it includes a great market volatility with significantly higher than average default rates and banks\' losses. Introducing macroeconomic scenarios in Brazilian financial system stress-testing and using categorical regression with panel data are the main contributions of the dissertation. The model can be extended to use industrial specific sector data to stress in order to identify potential risks of loans\' concentration.

This thesis deals with questions of financial stability in financial systems, monitoring, analyzing and assessment of financial stability from views of various national and international financial institutions. Enhanced attention is paid to the role of central bank in the area of financial stability. Central bank plays significant role in practice of monetary policy, regulation and supervision, prevention of financial instability and financial crisis solution. In this work is also mentioned Financial stability report issued by Czech National Bank and various reports issued by international institutions (e.g. BIS, ECB, IMF). These international institutions exercise activities to the support of financial stability. Thesis deals closely with instruments of assessment of financial stability. These instruments are indicators of financial stability and stress tests.

La mise en place de Solvabilité II conduit les actuaires à s'interroger sur la bonne adéquation entre modèles et données. Aussi, cette thèse a pour objectif d'étudier plusieurs approches statistiques, souvent méconnues des praticiens, permettant l'utilisation de méthodes multi états pour modéliser et gérer les risques individuels en assurance. Le Chapitre 1 présente le contexte général de cette thèse et permet de faire positionner ses principales contributions. Nous abordons les concepts de base liés à l'utilisation de modèles multi-états en assurance et décrivons les techniques d'inférence classiques adaptées aux données rencontrées, qu'ils soient markoviens ou non-markoviens. Pour finir, nous présentons comment il est possible d'utiliser ces modèles pour la gestion des risques de crédit. Le Chapitre 2 se concentre sur l'utilisation de méthodes d'inférence non-paramétriques pour la construction de lois d'incidence en assurance dépendance. Puisque plusieurs causes d'entrée sont susceptibles d'intervenir et d'intéresser les actuaires, nous nous concentrons sur une méthode utilisée pour l'estimation de modèles multi-états markoviens en temps continu. Nous comparons, dans un second temps, ces estimateurs à ceux utilisés classiquement par les praticiens tires de l'analyse de survie. Cette seconde approche peut comporter des biais non négligeables car ne permettant pas d'appréhender correctement l'interaction possible entre les causes. En particulier, elle comprend une hypothèse d'indépendance ne pouvant être testée dans le cadre de modèles à risques concurrents. Notre approche consiste alors à mesurer l'erreur commise par les praticiens lors de la construction de lois d'incidence. Une application numérique est alors considérée sur la base des données d'un assureur dépendance / With the implementation of the Solvency II framework, actuaries should examine the good adequacy between models and data. This thesis aims to study several statistical approaches, often ignored by practitioners, enabling the use of multi-state methods to model and manage individual risks in insurance. Chapter 1 presents the general context of this thesis and positions its main contributions. The basic tools to use multi-state models in insurance are introduced and classical inference techniques, adapted to insurance data with and without the Markov assumption, are presented. Finally, a development of these models for credit risk is outlined. Chapter 2 focuses on using nonparametric inference methods to build incidence tables for long term care insurance contracts. Since there are several entry-causes in disability states which are useful for actuaries, an inference method for competing risks data, seen as a Markov multi-state model in continuous time, is used. In a second step, I compare these estimators to those conventionally used by practitioners, based on survival analysis methods. This second approach may involve significant bias because the interaction between entry-causes cannot be appropriately captured. In particular, these approaches assume that latent failure times are independent, while this hypothesis cannot be tested for competing risks data. Our approach allows to measure the error done by practitioners when they build incidence tables. Finally, a numerical application is considered on a long term care insurance dataset

Panic Disorder (PD) has been associated with abnormalities in the respiratory system for a long time, and treatment programs aimed at reversing these abnormalities have been developed. Panicogenic effects of biological challenges have been shown to be altered after successful treatment. Furthermore, there is evidence that anxious non-PD patients show similar responses to these challenges and hence may benefit from some kind of breathing training (BT). To test these assumptions, we recruited 45 PD patients, 39 Episodic Anxiety (EA) patients who suffered from subclinical panic attacks, and 20 non-anxious controls (NAC). Patients were randomized to one of two versions of a 4-week therapy with BT, either lower or raise end-tidal pCO2, or a waiting list (WL). Before and after treatment, participants underwent in randomized order a Voluntary Hypoventilation (VHO) test and a Voluntary Hyperventilation (VHT) test in which they were asked to either lower or raise their pCO2 while psychophysiological measures were recorded. Each test consisted of 3 segments: 1 min baseline, 3 min paced breathing, and 8 min recovery. Before treatment, PD and EA patients were more anxious, distressed, tense, and worried than NAC, and felt more dizziness, chest pain, and nausea during the laboratory assessment. However, increases in psychological symptoms or physiological sensations from baseline to the paced breathing segments were not different between groups. The two tests produced similar changes except that anxiety and dizziness increased more during the VHT than VHO. We replicated baseline breathing abnormalities previously reported for PD patients, namely greater respiration rate, tidal volume instability, and number of sighs. However, analyses did not find that patients recovered slower to either challenge. After treatment, both therapies improved on the main outcome measure. Furthermore, BT affected baseline pCO2, resulting in lower levels in the hypocapnic groups and higher levels in the hypercapnic groups without affecting any other measures. We conclude that baseline respiratory abnormalities are specific to PD. However, data suggest that the manipulations might have been too weak to elicit other previously reported group differences. Breathing training was equally effective for the lower and raise BT. Hence, factors unrelated to modifying one’s pCO2 must have accounted for the symptomatic improvement. Breathing training should not be restricted to PD but be applied to all patients suffering from anxiety attacks.

This thesis deals with stress testing of the banking sector. Stress tests are a set of analytical tools used to test the resilience and financial stability of the banking sector. At the beginning of the work financial stability and systemic risk impact not only in the form of sys-temically important financial institutions are discussed. Followed by a chapter on stress tests, which describes historic development of stress testing approaches to testing of individual banking risks and their implementation in the form of stress testing. Stress testing methodo-logy is described primarily from the perspective of the Czech National Bank, the importance of banking regulation and supervision in Basel III is also presented. In the final chapter of the thesis there are commented results of Czech National Bank's stress tests, and EU wide stress tests 2014, launched in cooperation with European Banking Authority, European Central Bank and the European Systemic Risk Board.

The direct effect of spouse presence during graded exercise testing on anxiety and performance has not been previously delineated. Therefore, the purposes of this study were to (a) ascertain if spouse presence during graded exercise testing affects state anxiety or physiological performance variables, and (b) determine differences in psychological status between cardiac patients and healthy adults.

Contexte : Les tests de marche de 6 minutes (TM6) et de stepper de 6 minutes (TS6) permettent d’évaluer et de suivre en routine la tolérance à l’effort des patients porteurs pneumopathies interstitielles idiopathiques fibrosantes (PII-f). En général, les programmes de réhabilitation respiratoire (RR), améliorent la tolérance à l’effort et la performance à ces tests, la qualité de vie et les symptômes chez les patients atteints de PII-f, qu’ils soient réalisés en centre ou au domicile des patients. Peu d’études rapportent un maintien des bénéfices de la RR plusieurs mois après la fin de celle-ci chez les patients PII-f.Objectif : L’objectif général de la thèse était d’évaluer comparativement les adaptations cardioventilatoires au TM6 et TS6 chez les patients PII-f à différentes étapes de leur prise en charge en RR (pré et post-RR). Nous avions également pour objectif majeur d’évaluer la faisabilité et l’efficacité d’un programme de maintenance réalisé dans des structures proposant des activités physiques adaptées (APA) proches du domicile des patients PII-f sur le maintien des bénéfices à six mois post-RR.Matériel et méthodes : L’ensemble des patients atteints de PII-f ont réalisé un programme de RR au domicile de 8 semaines. Durant la période de suivi post-RR, les patients avaient le choix de bénéficier d’un suivi en APA dans une structure extérieure au domicile (groupe APA) ou de continuer la pratique d’une activité physique régulière en autonomie (groupe contrôle). Pré et post-RR, nous avons mesuré la tolérance à l’effort (TM6 et TS6) avec mesure des paramètres cardioventilatoires, les fonctions pulmonaires, la qualité de vie, la dyspnée, l’anxiété/dépression et la motivation des patients. Ces évaluations ont été réalisées aussi après 6 et 9 mois de suivi post-RR.Résultats : Parmi les 21 patients PII-f recrutés, 19 PII-f ont réalisé le programme de RR et sont revenus pour l’évaluation après 6 mois de suivi. Le TM6 et le TS6 induisent des adaptations cardioventilatoires différentes chez les patients atteints de PID, notamment une réponse ventilatoire supérieure au TS6 susceptible de jouer un rôle important sur la moindre désaturation en O2 observée comparé au TM6. A la suite d’un programme de RR, nous avons observé une amélioration des capacités physiques chez les patients PII-f. De manière individuelle, cependant, 58% des patients ne s’amélioraient pas post-RR, et 32% ne continuaient pas de pratiquer régulièrement les exercices physiques recommandés post-RR. La mise en place d’un programme de maintenance dans différentes structures locales proposant des APA proches du domicile chez des patients PII-f est réalisable et permet d’optimiser le maintien des bénéfices post-RR, qu’ils soient répondeurs ou non-répondeurs à la RR.Conclusion : Nos travaux ont souligné l’importance du choix du test d’effort, de la typologie de la PII-f et de la sévérité de l’hypoxémie sur les adaptations cardioventilatoires à l’effort, les adaptations physiologiques et l’évolution de la tolérance à l’effort post-RR. De plus, nous avons constaté que la réalisation de programmes de maintenance en APA, en s’appuyant sur des structures locales est réalisable et favorise un maintien des bénéfices à long terme chez les patients PII-f. Ainsi, l’évaluation des freins à l’activité physique chez un tiers des patients PII-f qui reste non-observant et arrête la pratique d’exercices physiques post-RR, est ainsi d’actualité. De plus, étudier plus précisément les différents profils de patients PII-f (répondeurs/non-répondeurs) pourrait être intéressant afin de comprendre les mécanismes potentiellement responsables de l’absence d’amélioration des capacités physiques post-RR chez les patients non-répondeurs, qu’elle soit liée à la pathologie, la médication ou une éventuelle pathologie musculaire associée. / Context: The 6-minute walk test (6MWT) and 6-minute stepper test (6MST) are routinely performed to measure and monitor exercise tolerance in patients with fibrosis interstitial idiopathic pneumonia (f-IIP). In those patients, exercise capacity, quality of life and symptoms are usually improved after a pulmonary rehabilitation (PR) program which can either be performed in a rehabilitation centre or at home.Aim: First, we aimed, to compare the cardio-pulmonary adaptations during a 6MWT and a 6MST in patients with f-IIP, before and after PR. We aimed to assess patients’ adhesion to physical activity maintenance programs which are offered and located near the f-IIP patient's homes. And their efficiency in maintaining benefits observed post-PR.Methods: All the f-IIP patients included followed a 8-week PR program at home. During the post-PR follow-up period, patients voluntarily chose to perform physical activity in a structure proposed by the investigator (APA group), or by themselves at home (control group). Before and after the PR program, we quantified the patient's exercise capacity (6MWT and 6MST) and measured cardiopulmonary parameters during both tests, pulmonary function at rest, quality of life, dyspnoea at rest and after each exercise, anxiety/depression and patients’ motivation. The same evaluations were repeated at 6 and 9 months post-PR.Results: Among the 21 recruited f-IIP patients, 19 finished the PR program and had an evaluation at 6 months post-PR. The 6MST was characterized by a higher minute ventilation compared with the 6MWT, and this may have contributed to the lower O2 desaturation also observed during the 6MST. Following the PR program, there was a mean improvement of exercise tolerance in f-IIP patients. However, among the patients, 58% showed no improvement in the 6MWT distance post-PR nor in the cardiorespiratory parameters during the test, and 32% did not continue to practise a regular physical activity post-PR. Finally, the individual support from the investigators for the patients to follow a maintenance program in local structures, near the f-IIP patient's home, probably contributed to their voluntary inscription in these programs, which allowed the patients to maintain the post-PR benefits, whether they patients initially responded to the PR program or not.Conclusion: Our works emphasized the importance of selecting an appropriate test, according to the fixed objective; of the f-IIP clinicopathological entities, and of the hypoxemia severity on exercise cardiorespiratory and physiological adaptations and changes of exercise tolerance post-PR. Moreover, we observed that patients from the APA group regularly attended the chosen physical maintenance program in local structures, which contributed to the maintenance the post-PR benefits in f-IIP. The evaluation of physical activity engagement barriers, that lead to the non-participation of a third of f-IIP patients to physical activities post-PR, remains to be conducted. Further studies should also focus on the explanation for the absence of improvement of exercise tolerance and cardiorespiratory parameters in about half of f-IIP patients post-RR, either due to their pathology, medication, or possible muscle disorders.

Submitted by Thiago Corrêa (thiago.strava.correa@gmail.com) on 2017-06-27T00:53:21Z No. of bitstreams: 1 Dissertacao Thiago Strava Correa.pdf: 2528557 bytes, checksum: 43d7258add4d6a8bd0ae8469e0948e7c (MD5) / Approved for entry into archive by Suzinei Teles Garcia Garcia (suzinei.garcia@fgv.br) on 2017-06-27T12:04:57Z (GMT) No. of bitstreams: 1 Dissertacao Thiago Strava Correa.pdf: 2528557 bytes, checksum: 43d7258add4d6a8bd0ae8469e0948e7c (MD5) / Made available in DSpace on 2017-06-27T12:56:20Z (GMT). No. of bitstreams: 1 Dissertacao Thiago Strava Correa.pdf: 2528557 bytes, checksum: 43d7258add4d6a8bd0ae8469e0948e7c (MD5) Previous issue date: 2017-05-29 / Esta dissertação propõe o uso de modelos de análise de estilo para a previsão da distribuição dos retornos de carteiras condicionais a cenários estressados de fatores de risco como uma alternativa aos tradicionais modelos de avaliação total. Dentre os seis modelos de análise de estilo cuja capacidade preditiva é testada, destacam-se os modelos quantílico composto e não-linear, que além de obterem os melhores resultados são ainda pouco explorados pela literatura de gestão de risco. / This dissertation suggests the use of style analysis models for the forecasting of portfolio returns’ distribution conditional to stressed scenarios of risk factors. Among the six style analysis models which had their forecasting capacity tested, the composite quantile and the non-linear quantile models stand out by their quality and lack of documentation in the risk management literature.

Les tests de résistance, qui évaluent la capacité des banques à soutenir la détérioration de la situation économique et financière, sont devenus un outil qui aide les banques centrales à exercer leur nouveau pouvoir de supervision et à promouvoir un système financier stable. En outre, la récente crise financière mondiale a déplacé le centre d'attention de la supervision financière d'une perspective microprudentielle, basée sur la résilience des institutions individuelles, à une perspective plus macroprudentielle, qui englobe la résilience globale du système financier. Par conséquent, les tests de résistance microprudentiels mettent l'accent sur le rôle traditionnel du capital bancaire en tant que coussin de protection contre les pertes, tandis que les tests de résistance macroprudentiels se concentrent sur le système bancaire dans son ensemble. La crise financière mondiale a également souligné le rôle crucial du risque de liquidité dans la détérioration de la stabilité du système financier international. Le premier article passe en revue le premier test de résistance de la BCE, en le comparant à la littérature et aux autres tests de résistance effectués de par le monde, et contextualise ses résultats. Il donne un aperçu des tests de résistance et une liste d'améliorations potentielles. Le second article cherche à construire une méthodologie qui couvre certaines lacunes (effet de spillovers et interactions entre solvabilité et liquidité) identifiée dans le premier chapitre et l'utilise pour évaluer la fragilité actuelle du système bancaire de la zone euro selon différents scénarios. Enfin, le troisième chapitre analyse le lien entre la solvabilité et les coûts de financement. / Stress tests, which evaluate banks' capacity to withstand deteriorating economic and financial condition, have become a tool that helps central banks to fulfil their new supervisory power and promote a stable financial system. Additionally, the global financial crisis shifted the perspective of financial supervision from a microprudential perspective, based on the resilience of individual institutions, to a more macroprudential perspective, which encompasses the whole financial system resilience. Hence, microprudential stress tests emphasize the traditional role of bank capital as a buffer against loss, shielding the deposit insurance agency, while, macroprudential stress tests focus on whether the banking system as a whole has the balance, sheet capacity to support the economy. The global financial crisis also highlighted the crucial role of liquidity risk in undermining the stability of the international financial system. The first paper reviews the first ECB stress test, comparing it to the literature and other stress tests conducted, and contextualize its results. It provides an overview of stress tests as a tool and give a list of potential improvements. The second paper, seeks to construct methodology that covers some shortcomings (spillovers and interaction between solvency and liquidity) identified in the first chapter and uses it to assess the current fragility of the Eurozone banking system to various scenarios. Finally, the third chapter analyses the link between solvency and funding costs as a potential source of second round losses in the stress testing framework.

This master's thesis is focused on the issues of stress testing in the context of financial stability. It consists of two major parts. The first part deals with methodology of general stress tests with special regard to stressed indicators and relevant risks. In the second part, the difficulties and shortcomings of general stress tests are analysed with respect to the dynamic expansion of financial instruments and markets. A special attention is paid to integrate liquidity risk and contagion risk into the current stress testing framework. Furthermore, the alternative instruments for increasing the complexity of banks financial soundness analysis are presented. It can be said that the system of financial stability analysis before the crisis was not sufficient because it did not reflect on all risk exposures. The main contribution of this thesis is the organized presentation of possible solutions which would help to enhance the quality of stress testing outputs and to a certain extent unify these outputs as well.