Mario V. Wüthrich

Copula convergence theorems for tail events

Abstract Tail dependence is studied from a distributional point of view by means of appropriate copulae. We derive similar results to the famous Pickands–Balkema–de Haan Theorem of Extreme Value Theory. Under regularity conditions, it is shown that the Clayton copula plays among the family of archimedean copulae the role of the generalized Pareto distribution. The practical usefulness of the results is illustrated in the analysis of stock market data.

The Structural Modelling of Operational Risk via Bayesian inference: Combining Loss Data with Expert Opinions

To meet the Basel II regulatory requirements for the Advanced Measurement Approaches, the bank’s internal model must include the use of internal data, relevant external data, scenario analysis and factors reflecting the business environment and internal control systems. Quantification of operational risk cannot be based only on historical data but should involve scenario analysis. Historical internal operational risk loss data have limited ability to predict future behaviour moreover, banks do not have enough internal data to estimate low frequency high impact events adequately. Historical external data are difficult to use due to different volumes and other factors. In addition, internal and external data have a survival bias, since typically one does not have data of all collapsed companies. The idea of scenario analysis is to estimate frequency and severity of risk events via expert opinions taking into account bank environment factors with reference to events that have occurred (or may have occurred) in other banks. Scenario analysis is forward looking and can reflect changes in the banking environment. It is important to not only quantify the operational risk capital but also provide incentives to business units to improve their risk management policies, which can be accomplished through scenario analysis. By itself, scenario analysis is very subjective but combined with loss data it is a powerful tool to estimate operational risk losses. Bayesian inference is a statistical technique well suited for combining expert opinions and historical data. In this paper, we present examples of the Bayesian inference methods for operational risk quantification.

Tail Dependence from a Distributional Point of View

The dependence structure in the tails of bivariate random variables is studied by means of appropriate copulae. Weak convergence results show that these copulae are natural dependence structures for joint tail events. The results obtained apply to particular types of copulae such as archimedean copulae and the Gaussian copula. Further, connections to multivariate extreme value theory are investigated and a two-dimensional Pickands–Balkema–de Haan Theorem type is derived. Finally, a counterexample showing that the tail dependence coefficients do not completely determine the dependence structure of bivariate rare events is provided.

The Quantification of Operational Risk Using Internal Data, Relevant External Data and Expert Opinion

To quantify an operational risk capital charge under Basel II, many banks adopt a Loss Distribution Approach. Under this approach, quantification of the frequency and severity distributions of operational risk involves the banks internal data, expert opinions and relevant external data. In this paper we suggest a new approach, based on a Bayesian inference method, that allows for a combination of these three sources of information to estimate the parameters of the risk frequency and severity distributions.

Market-consistent actuarial valuation

Introduction.- Stochastic discounting.- Valuation portfolio in life insurance.- Financial risks.- Valuation portfolio in non-life insurance.- Selected topics.

Model Uncertainty in Claims Reserving within Tweedie's Compound Poisson Models

In this paper we examine the claims reserving problem using Tweedies compound Poisson model. We develop the maximum likelihood and Bayesian Markov chain Monte Carlo simulation approaches to fit the model and then compare the estimated models under different scenarios. The key point we demonstrate relates to the comparison of reserving quantities with and without model uncertainty incorporated into the prediction. We consider both the model selection problem and the model averaging solutions for the predicted reserves. As a part of this process we also consider the sub problem of variable selection to obtain a parsimonious representation of the model being fitted.

Asymptotic Value-at-Risk Estimates for Sums of Dependent Random Variables

We estimate Value-at-Risk for sums of dependent random variables. We model multivariate dependent random variables using archimedean copulas. This structure allows one to calculate the asymptotic behaviour of extremal events. An important application of such results are Value-at-Risk estimates for sums of dependent random variables.

THE MEAN SQUARE ERROR OF PREDICTION IN THE CHAIN LADDER RESERVING METHOD (MACK AND MURPHY REVISITED)

We revisit the famous Mack formula [2], which gives an estimate for the mean square error of prediction MSEP of the chain ladder claims reserving method: We define a time series model for the chain ladder method. In this time series framework we give an approach for the estimation of the conditional MSEP. It turns out that our approach leads to results that differ from the Mack formula. But we also see that our derivation leads to the same formulas for the MSEP estimate as the ones given in Murphy [4]. We discuss the differences and similarities of these derivations.

Credibility for the Chain Ladder Reserving Method

We consider the chain ladder reserving method in a Bayesian set up, which allows for combining the information from a specific claims development triangle with the information from a collective. That is, for instance, to consider simultaneously own company specific data and industry-wide data to estimate the own companys claims reserves. We derive Bayesian estimators and credibility estimators within this Bayesian framework. We show that the credibility estimators are exact Bayesian in the case of the exponential dispersion family with its natural conjugate priors. Finally, we make the link to the classical chain ladder method and we show that using non-informative priors we arrive at the classical chain ladder forecasts. However, the estimates for the mean square error of prediction differ in our Bayesian set up from the ones found in the literature. Hence, the paper also throws a new light upon the estimator of the mean square error of prediction of the classical chain ladder forecasts and suggests a new estimator in the chain ladder method.

Prediction Error of the Multivariate Chain Ladder Reserving Method

Abstract In this paper we consider the claims reserving problem in a multivariate context: that is, we study the multivariate chain-ladder (CL) method for a portfolio of N correlated runoff triangles based on multivariate age-to-age factors. This method allows for a simultaneous study of individual runoff subportfolios and facilitates the derivation of an estimator for the mean square error of prediction (MSEP) for the CL predictor of the ultimate claim of the total portfolio. However, unlike the already existing approaches we replace the univariate CL predictors with multivariate ones. These multivariate CL predictors reflect the correlation structure between the subportfolios and are optimal in terms of a classical optimality criterion, which leads to an improvement of the estimator for the MSEP. Moreover, all formulas are easy to implement on a spreadsheet because they are in matrix notation. We illustrate the results by means of an example.