Abstract
This article examines the epistemology of risk assessment in the context of financial modelling for the purposes of making loan underwriting decisions. A financing request for a company in the paper and pulp industry is considered in some detail. The paper and pulp industry was chosen because (1) it is subject to some specific risks that have been identified and studied by bankers, investors and managers of paper and pulp companies and (2) certain features of the industry enable analysts to quantify the impact of specific risk events of a given dimension on a company's future financial performance. While companies in other industries may be subject to similar risk factors, the impact of risk events may be more difficult to gauge in those industries. The ability of financial analysts to model the impact of a risk event, and hence quantify a credit risk, increases the predictive accuracy of the model. I argue that bankers and regulators should recognise the uncertainty associated with unquantifiable credit risk in financial models, and they should view this uncertainty as a credit risk factor in and of itself. Evaluating the relative degree to which credit risk is quantifiable in financial models is a potentially significant yet largely unrecognised tool for credit risk management. I consider some possible applications of this assessment tool for managing risk within the banking industry.
Acknowledgements
I thank the participants of the April 2011 Epistemology of Modelling and Simulation Conference at the University of Pittsburgh, especially Thomas Breuer (PPE Research Center), Louise Comfort (University of Pittsburgh) and Leonard Smith (London School of Economics), for stimulating comments and discussion. I also thank Thomas Hickey (National Association for Insurance Commissioners) and Mark Wunderlich (Union College) for helpful suggestions. I am especially indebted to Shelley Yu (Société Générale) for many helpful conversations and her unwavering common sense.
Notes
Notes
1. In the technical sense, a risk is a non-zero chance of a negative outcome, but the possible outcomes and their probabilities are known. Uncertainty, by contrast, is a situation in which either the range of possible outcomes or their probabilities – or both – are unknown. Two classic sources for discussion of the distinction are Knight (Citation1921) and Keynes (Citation1937).
2. Uncertainty in our estimates is sometimes called ‘measurable uncertainty’. Our estimates (perhaps under idealisation assumptions) of the likelihood of the occurrence of events and their magnitudes may be only partially determined probabilities, which would typically be expressed by a probability interval or a probability distribution. A discussion of the difference between true and measurable uncertainties can be found in Clements and Hendry (Citation2002, pp. 1–18).
3. The ceteris paribus clause is needed to allow for situations where factors leading to uncertainty in models may be non-comparable.
4. The case is based on an actual credit request, but some details have been changed or omitted to facilitate presentation and to avoid disclosure of any non-public information. The projection models were based on historical information from the company's 10-K and 10-Q statements.
5. An illustration of a 10-factor bank internal rating system can be found in Saunders and Allen (Citation2010, pp. 300–301). As the authors discuss there, banks may employ either a one-dimensional system that assigns an overall rating to a loan or a two-dimensional system that assesses separately the borrower's overall creditworthiness and the loss severity of the individual loan.
6. For a current review of quantitative risk models, see Saunders and Allen (Citation2010).
7. The Kamakura reduced form model (Kamakura Corporation Citation2008) is discussed in some detail in Chapter 5 of Saunders and Allen (Citation2010, pp. 98–116).