Natalia Nolde

The global financial crisis of 2007-2009 revealed the importance of systemic risk: the risk that may destabilize the global economy due to financial contagion. Accurate assessment of systemic risk would not only enable regulators to introduce suitable policies to mitigate the risk, but also allow individual institutions to monitor and mitigate their vulnerability. An effective measure of systemic risk should be able to capture the co-movements between a financial system (or market) and individual financial institutions. One popular measure of systemic risk is CoVaR. In this thesis, a methodology is proposed to compute dynamic forecasts of CoVaR semi-parametrically within the classical framework of multivariate extreme value theory (EVT). According to the definition, CoVaR can be viewed as a high quantile of a conditional distribution where the conditioning event corresponds to large losses of an institution. The idea of our methodology is to relate this conditional distribution to the tail dependence function. We develop an EVT-based framework to estimate CoVaR statically by combining parametric modelling of the tail dependence function to address the issue of data sparsity in the joint tail regions and semi-parametric univariate tail estimation techniques. The performance of the methodology is illustrated via simulation studies and real data examples.

View record

Banks that use the advanced measurement approach to model operational risk may struggle to develop an internal process that produces stable regulatory capital over time. Large decreases in regulatory capital are scrutinized by regulators while large increases may force banks to set aside more assets than necessary. A major source of this instability arises from the loss severity selection process, especially when the selected distribution families for severity risk categories change year-to-year. In this report, we examine the process of selecting severity distributions from a candidate distribution list within the guidelines of the advanced measurement approach, propose useful tools to aid in selecting an appropriate severity distribution, and analyze the effect of selection criteria on regulatory capital. The log sinh-arcsinh distribution family is added to a list of common candidate severity distributions used by industry. This 4-parameter family solves issues introduced by the 4-parameter g-and-h distribution without sacrificing flexibility and shows promise in outperforming 2-parameter families, reducing the frequency of severity distribution families changing year-to-year. Distribution parameters are estimated using the maximum likelihood approach from loss data truncated at a known minimum reporting threshold. Our severity distribution selection process combines truncation probability estimates with Akaike Information Criterion (AIC), Bayesian Information Criterion, modified Anderson-Darling, QQ-plots, and predictive measures such as the quantile scoring function and out-of-sample AIC, and we discuss some of the challenges associated with this process. We then simulate operational losses and calculate regulatory capital, comparing the effect on regulatory capital of selecting loss severity distributions using AIC versus quantile score. A combination of these two criteria is recommended when selecting loss severity distributions.

View record

Estimation of multivariate quantile regions with very small probabilities, referred toas risk regions in this report, plays an important part in various applications. Yet,it is a difficult problem since such regions contain hardly any or no data. Existingmethods address the problem only for heavy-tailed distributions or for bivariatedistributions with non-degenerate exponent measure that do not include the casesof asymptotic independence.In this report, we propose a more flexible framework to supplement existingapproaches to risk region estimation by allowing tails of the underlying distributionto be light as well as covering situations of tail independence. In particular,we concentrate on a class of distributions assumed to have a density function withhomothetic level sets. In simulation studies, reasonable performance of our proposedmethod is demonstrated. We also present two real-life applications to furtherillustrate the flexibility and performance of our method.

View record

The Autoregressive Stochastic Volatility (ARSV) model is a discrete-time stochastic volatility model that can model the financial returns time series and volatilities. This model is relevant for risk management. However, existing inference methods have various limitations on model assumptions. In this report we discuss a new inference method that allows flexible model assumption for innovation of the ARSV model. We also present the application of ARSV model to risk management, and compare the ARSV model with another commonly used model for financial time series, namely the GARCH model.

View record

Forecasting of risk measures is an important part of risk management for financial institutions. Value-at-Risk and Expected Shortfall are two commonly used risk measures and accurately predicting these risk measures enables financial institutions to plan adequately for possible losses. Point forecasts from different methods can be compared using consistent scoring functions, provided the underlying functional to be forecasted is elicitable. It has been shown that the choice of a scoring function from the family of consistent scoring functions does not influence the ranking of forecasting methods as long as the underlying model is correctly specified and nested information sets are used. However, in practice, these conditions do not hold, which may lead to discrepancies in the ranking of methods under different scoring functions.We investigate the choice of scoring functions in the face of model misspecification, parameter estimation error and nonnested information sets. We concentrate on the family of homogeneous consistent scoring functions for Value-at-Risk and the pair of Value-at-Risk and Expected Shortfall and identify conditions required for existence of the expectation of these scoring functions. We also assess the finite-sample properties of the Diebold-Mario Test, as well as examine how these scoring functions penalize for over-prediction and under-prediction with the aid of simulation studies.

View record