A key feature of best practice prudential supervision of financial institutions is the use of a risk rating system to formalise the outcome of supervisory reviews and ongoing monitoring processes. The Australian Prudential Regulation Authority (APRA) implemented the Probability and Impact Rating System (PAIRS) in 2002. Given the favourable economic conditions in which PAIRS was developed and has so far operated, any form of validation using backtesting methods is prevented. Consequently, this thesis seeks to develop a framework with which to evaluate and better understand the PAIRS risk rating system for authorised deposit-taking institutions. Specifically, we specify and estimate models in which the risk ratings are related to the statistical data that supervisors have access to when forming their expert judgement assessments of the PAIRS risk components. Whereas prior studies have generally focused on the overall supervisory rating, we model the primary components of the PAIRS rating (inherent risk, management and control risk, and capital support risk) as well as the aggregate risk of failure rating. Using a sample of ratings from 2002 to 2006, we find that the statistical data is able to explain much of the variability in ratings for credit unions and building societies (CUBS) and Australian and foreign subsidiary banks but not foreign bank branches. As expected, the regressions are stronger for inherent risk and capital support risk ratings than management and control risk ratings. However, supervisors?? consideration of adverse qualitative factors adds considerable explanatory power to a model based solely on statistical data, particularly for management and control risk ratings. We also model the determinants of supervisory exceptions and capital adequacy breaches over 1992 to 2006 and find that the risk indicators associated with a higher likelihood of an exception and/or breach are generally consistent with the risk indicators associated with supervisory risk ratings. The outcomes of the thesis have a number of policy implications and practical applications. For example, the estimated models have the potential to be used as a quality and consistency tool to detect rating outliers within PAIRS. We also propose some improvements to APRA??s exception reporting system for CUBS.

A key feature of best practice prudential supervision of financial institutions is the use of a risk rating system to formalise the outcome of supervisory reviews and ongoing monitoring processes. The Australian Prudential Regulation Authority (APRA) implemented the Probability and Impact Rating System (PAIRS) in 2002. Given the favourable economic conditions in which PAIRS was developed and has so far operated, any form of validation using backtesting methods is prevented. Consequently, this thesis seeks to develop a framework with which to evaluate and better understand the PAIRS risk rating system for authorised deposit-taking institutions. Specifically, we specify and estimate models in which the risk ratings are related to the statistical data that supervisors have access to when forming their expert judgement assessments of the PAIRS risk components. Whereas prior studies have generally focused on the overall supervisory rating, we model the primary components of the PAIRS rating (inherent risk, management and control risk, and capital support risk) as well as the aggregate risk of failure rating. Using a sample of ratings from 2002 to 2006, we find that the statistical data is able to explain much of the variability in ratings for credit unions and building societies (CUBS) and Australian and foreign subsidiary banks but not foreign bank branches. As expected, the regressions are stronger for inherent risk and capital support risk ratings than management and control risk ratings. However, supervisors?? consideration of adverse qualitative factors adds considerable explanatory power to a model based solely on statistical data, particularly for management and control risk ratings. We also model the determinants of supervisory exceptions and capital adequacy breaches over 1992 to 2006 and find that the risk indicators associated with a higher likelihood of an exception and/or breach are generally consistent with the risk indicators associated with supervisory risk ratings. The outcomes of the thesis have a number of policy implications and practical applications. For example, the estimated models have the potential to be used as a quality and consistency tool to detect rating outliers within PAIRS. We also propose some improvements to APRA??s exception reporting system for CUBS.

In this paper we study parameter estimation via the Expectation Maximization (EM) algorithm for a continuous-time hidden Markov model with diffusion and point process observation. Inference problems of this type arise for instance in credit risk modelling. A key step in the application of the EM algorithm is the derivation of finite-dimensional filters for the quantities that are needed in the E-Step of the algorithm. In this context we obtain exact, unnormalized and robust filters, and we discuss their numerical implementation. Moreover, we propose several goodness-of-fit tests for hidden Markov models with Gaussian noise and point process observation. We run an extensive simulation study to test speed and accuracy of our methodology. The paper closes with an application to credit risk: we estimate the parameters of a hidden Markov model for credit quality where the observations consist of rating transitions and credit spreads for US corporations.