Credit risk optimization using factor models

We study portfolio credit risk management using factor models, with a focus on optimal portfolio selection based on the tradeoff of expected return and credit risk. We begin with a discussion of factor models and their known analytic properties, paying particular attention to the asymptotic limit of a large, finely grained portfolio. We recall prior results on the convergence of risk measures in this “large portfolio approximation” which are important for credit risk optimization. We then show how the results on the large portfolio approximation can be used to reduce significantly the computational effort required for credit risk optimization. For example, when determining the fraction of capital to be assigned to particular ratings classes, it is sufficient to solve the optimization problem for the large portfolio approximation, rather than for the actual portfolio. This dramatically reduces the dimensionality of the problem, and the amount of computation required for its solution. Numerical results illustrating the application of this principle are also presented. JEL Classification G11     

Recent developments in consumer credit risk assessment

Consumer credit risk assessment involves the use of risk assessment tools to manage a borrower’s account from the time of pre-screening a potential application through to the management of the account during its life and possible write-off. The riskiness of lending to a credit applicant is usually estimated using a logistic regression model though researchers have considered many other types of classifier and whilst preliminary evidence suggest support vector machines seem to be the most accurate, data quality issues may prevent these laboratory based results from being achieved in practice. The training of a classifier on a sample of accepted applicants rather than on a sample representative of the applicant population seems not to result in bias though it does result in difficulties in setting the cut off. Profit scoring is a promising line of research and the Basel 2 accord has had profound implications for the way in which credit applicants are assessed and bank policies adopted.     

Development and application of consumer credit scoring models using profit-based classification measures

This paper presents a new approach for consumer credit scoring, by tailoring a profit-based classification performance measure to credit risk modeling. This performance measure takes into account the expected profits and losses of credit granting and thereby better aligns the model developers’ objectives with those of the lending company. It is based on the Expected Maximum Profit (EMP) measure and is used to find a trade-off between the expected losses – driven by the exposure of the loan and the loss given default – and the operational income given by the loan. Additionally, one of the major advantages of using the proposed measure is that it permits to calculate the optimal cutoff value, which is necessary for model implementation. To test the proposed approach, we use a dataset of loans granted by a government institution, and benchmarked the accuracy and monetary gain of using EMP, accuracy, and the area under the ROC curve as measures for selecting model parameters, and for determining the respective cutoff values. The results show that our proposed profit-based classification measure outperforms the alternative approaches in terms of both accuracy and monetary value in the test set, and that it facilitates model deployment.     

A semi-analytical method for VaR and credit exposure analysis

In this paper, we discuss new analytical methods for computing Value-at-Risk (VaR) and a credit exposure profile. Using a Monte Carlo simulation approach as a benchmark, we find that the analytical methods are more accurate than RiskMetrics delta VaR, and are more efficient than Monte Carlo, for the case of fixed income securities. However the accuracy of the method deteriorates when applied to a portfolio of barrier options.     

Samsung card lending model

Samsung Card Lending Model (SCLM) analyzes cash flow in individual accounts and measures the level of company-wide risk. Serving as a risk and portfolio management model in the consumer lending business, the main features of SCLM are as follows. Default ratios such as intrinsic balance default probability and annual default ratio are computed using the past, present, and future cash flows of accounts. The provision is shown as the total sum of write-offs. The size of capital required is determined by default probability distribution. The price for new accounts is quoted based on cash flow simulations reflecting future business environments. SCLM has shown good performance in Samsung card consumer lending business since the Korean credit card crisis of 2003.     

Importance sampling for integrated market and credit portfolio models

A sophisticated approach for computing the total economic capital needed for various stochastically dependent risk types is the bottom-up approach. In this approach, usually, market and credit risks of financial instruments are modeled simultaneously. As integrating market risk factors into standard credit portfolio models increases the computational burden of calculating risk measures, it is analyzed to which extent importance sampling techniques previously developed either for pure market portfolio models or for pure credit portfolio models can be successfully applied to integrated market and credit portfolio models. Specific problems which arise in this context are discussed. The effectiveness of these techniques is tested by numerical experiments for linear and non-linear portfolios.     

Time will tell: behavioural scoring and the dynamics of consumer credit assessment

This paper discusses the use of dynamic modelling in consumercredit risk assessment. It surveys the approaches and objectivesof behavioural scoring, customer scoring and profit scoring.It then investigates how Markov chain stochastic processes canbe used to model the dynamics of the delinquency status andbehavioural scores of consumers. It discusses the use of segmentation,mover–stayer models and the use of second- and third-ordermodels to improve the fit of such models. The alternative survivalanalysis proportional hazards approach to estimating when defaultoccurs is considered. Comparisons are made between the wayscredit risk is modelled in consumer lending and corporate lending.     

Optimality of myopic strategies for multi-stock discrete time market with management costs

The paper studies multi-stock discrete time market models with serial correlations and with some management costs. We found a market structure that ensures that the optimal strategy is myopic for the case of either power or log utility function.     

Subagging for credit scoring models

The logistic regression framework has been for long time the most used statistical method when assessing customer credit risk. Recently, a more pragmatic approach has been adopted, where the first issue is credit risk prediction, instead of explanation. In this context, several classification techniques have been shown to perform well on credit scoring, such as support vector machines among others. While the investigation of better classifiers is an important research topic, the specific methodology chosen in real world applications has to deal with the challenges arising from the real world data collected in the industry. Such data are often highly unbalanced, part of the information can be missing and some common hypotheses, such as the i.i.d. one, can be violated. In this paper we present a case study based on a sample of IBM Italian customers, which presents all the challenges mentioned above. The main objective is to build and validate robust models, able to handle missing information, class unbalancedness and non-iid data points. We define a missing data imputation method and propose the use of an ensemble classification technique, subagging, particularly suitable for highly unbalanced data, such as credit scoring data. Both the imputation and subagging steps are embedded in a customized cross-validation loop, which handles dependencies between different credit requests. The methodology has been applied using several classifiers (kernel support vector machines, nearest neighbors, decision trees, Adaboost) and their subagged versions. The use of subagging improves the performance of the base classifier and we will show that subagging decision trees achieve better performance, still keeping the model simple and reasonably interpretable.     

On the construction of non-affine jump-diffusion models

We describe a method for construction of jump analogues of certain one-dimensional diffusion processes satisfying solvable stochastic differential equations. The method is based on the reduction of the original stochastic differential equations to the ones with linear diffusion coefficients, which are reducible to the associated ordinary differential equations, by using the appropriate integrating factor processes. The analogues are constructed by means of adding the jump components linearly into the reduced stochastic differential equations. We illustrate the method by constructing jump analogues of several diffusion processes and expand the notion of market price of risk to the resulting non-affine jump-diffusion models.     

No arbitrage in a simple credit risk model

In a simple credit risk model we find an equivalent condition to the no-simple-arbitrage principle and show that it is insufficient for an equivalent martingale measure to exist.     

Measuring retail company performance using credit scoring techniques

This paper discusses models for evaluating credit risk in relation to the retailing industry. Hunt’s [Hunt, S.D., 2000. A General Theory of Competition. Sage Publications Inc., California] Resource–Advantage Theory of Competition is used as a basis for variable selection, given the theory’s relevancy to retail competition. The study focuses on the US retail market. Four standard credit scoring methodologies: Naïve Bayes, Logistic Regression, Recursive Partitioning and Artificial Neural Network, are compared with Sequential Minimal Optimization (SMO), using a sample of 195 healthy companies and 51 distressed firms over five time periods from 1994 to 2002.     

Tracking models and the optimal regret distribution in asset allocation

We present a tracking model for asset allocation that tracks desired investment goals. The model is shown to be optimal with respect to an investor's ‘regret distribution’, the cumulative distribution of the difference between the revenue under perfect foresight and that possible without foresight. Relationships with Markowitz mean/variance models are also explored.     

Precautionary measures for credit risk management in jump models

Sustaining efficiency and stability by properly controlling the equity to asset ratio is one of the most important and difficult challenges in bank management. Due to unexpected and abrupt decline of asset values, a bank must closely monitor its net worth as well as market conditions, and one of its important concerns is when to raise more capital so as not to violate capital adequacy requirements. In this paper, we model the trade-off between avoiding costs of delay and premature capital raising, and solve the corresponding optimal stopping problem. In order to model defaults in a bank's loan/credit business portfolios, we represent its net worth by Lévy processes, and solve explicitly for the double exponential jump-diffusion process and for a general spectrally negative Lévy process.     

Enhanced indexation based on second-order stochastic dominance

Second order Stochastic Dominance (SSD) has a well recognised importance in portfolio selection, since it provides a natural interpretation of the theory of risk-averse investor behaviour. Recently, SSD-based models of portfolio choice have been proposed; these assume that a reference distribution is available and a portfolio is constructed, whose return distribution dominates the reference distribution with respect to SSD. We present an empirical study which analyses the effectiveness of such strategies in the context of enhanced indexation. Several datasets, drawn from FTSE 100, SP 500 and Nikkei 225 are investigated through portfolio rebalancing and backtesting. Three main conclusions are drawn. First, the portfolios chosen by the SSD based models consistently outperformed the indices and the traditional index trackers. Secondly, the SSD based models do not require imposition of cardinality constraints since naturally a small number of stocks are selected. Thus, they do not present the computational difficulty normally associated with index tracking models. Finally, the SSD based models are robust with respect to small changes in the scenario set and little or no rebalancing is necessary.     

Assessing Markov property in multistate transition models with applications to credit risk modeling

Multistate transition models are increasingly used in credit risk applications as they allow us to quantify the evolution of the process among different states. If the process is Markov, analysis and prediction are substantially simpler, so analysts would like to use these models if they are applicable. In this paper, we develop a procedure for assessing the Markov hypothesis and discuss different ways of implementing the test procedure. One issue when sample size is large is that the statistical test procedures will detect even small deviations from the Markov model when these differences are not of practical interest. To address this problem, we propose an approach to formulate and test the null hypothesis of “weak non‐Markov.” The situation where the transition probabilities are heterogeneous is also examined, and approaches to accommodate this case are indicated. Simulation studies are used extensively to study the properties of the procedures, and two applications are to illustrate the results.     

Handling CVaR objectives and constraints in two-stage stochastic models

Based on the polyhedral representation of Künzi-Bay and Mayer [Künzi-Bay, A., Mayer, J., 2006. Computational aspects of minimizing conditional value-at-risk. Computational Management Science 3, 3–27] , we propose decomposition frameworks for handling CVaR objectives and constraints in two-stage stochastic models.     

Modelling Credit Risk in the Jump Threshold Framework

ABSTRACT

The jump threshold framework for credit risk modelling developed by Garreau and Kercheval enjoys the advantages of both structural- and reduced-form models. In their article, the focus is on multidimensional default dependence, under the assumptions that stock prices follow an exponential Lévy process (i.i.d. log returns) and that interest rates and stock volatility are constant. Explicit formulas for default time distributions and basket credit default swap (CDS) prices are obtained when the default threshold is deterministic, but only in terms of expectations when the default threshold is stochastic. In this article, we restrict attention to the one-dimensional, single-name case in order to obtain explicit closed-form solutions for the default time distribution when the default threshold, interest rate and volatility are all stochastic. When the interest rate and volatility processes are affine diffusions and the stochastic default threshold is properly chosen, we provide explicit formulas for the default time distribution, prices of defaultable bonds and CDS premia. The main idea is to make use of the Duffie–Pan–Singleton method of evaluating expectations of exponential integrals of affine diffusions.     

The credit risk+ model with general sector correlations

We consider an enhancement of the credit risk⁺ model to incorporate correlations between sectors. We model the sector default rates as linear combinations of a common set of independent variables that represent macro-economic variables or risk factors. We also derive the formula for exact VaR contributions at the obligor level.