Default Prediction Models Research Articles

ZETA™ Methodology and Variation in the Systemic Risk of Default: Accounting for the Effects of Type II (False Negative) Errors Variation on Lending

Olgiati Stefano - University of Bergamo, Department of Management, Economics and Quantitative Methods. 
 Danovi Alessandro - University of Bergamo
 The loan manager - dealing with one single borrower at a time and being responsible for that single decision to lend - is exposed to the idiosyncratic risk of default of his customer just like the physician is exposed to the risk of a wrong diagnosis with our strep throat. At the same time – if we do not expect the strep throat diagnostic test kit to change - we would still expect that physician reading that test to become more careful – or update his prior beliefs – about his diagnoses when a flu epidemic is likely to kick in with a certain estimated probability (likelihood). However, this has not been the case with loan management - there is in fact some consensus that before 2007 a reduction in the standards of idiosyncratic risk assessment by lenders - prior to risks pooling - coupled with a worsening of the systemic risk scenario, is partly to blame for the well known 2007-2008 financial crisis, with some of the blame falling also on the incapacity of actuarial mathematical models (test kits) to update worst case scenarios or be calibrated continuously on the basis of variation in the likelihood of default of the underlying risks pool.The authors of this paper argue that, on the other hand, a standard Bayesian transformation of the ZETA bankruptcy prediction methodology introduced by Altman in 1968-1977 allows for a continuous a posterioriupdate of conditional Type I and II errors due to variation in the systemic likelihood of default. The Bayesian transformation can be used both to condition the loan manager’s prior decision (generally based on Basel II-compliant Internal Rating Based system or Credit Agency’s Rating) and to update such decision on the basis of any posterior hypothesis (based on actuarial frequentist assumptions of conditional hazard rates) regarding the creditworthiness and the probability of default of an underlying pool of securities.At the same time – under a Bayesian framework - the ZETA diagnostic test can be conditioned on the new evidence introduced by other tests to increase the total sensitivity of the default prediction models (IRB ratings, TTC ratings, logit, probit, neural) to update the commercial bank’s lending decisions.A ground-state, static meta-analysis of Altman’s et al. ZETA original article (1977) reveals that the odds of the commercial bank detecting a default after the ZETA score has been introduced (post-test) is 13.2 times more effective than the a priori prediction. Under the same assumptions, the odds of the commercial bank detecting a survival after (post-test) the ZETA score has been introduced is 12.2 times more effective than the a priori. Integration of the ZETA model with other default prediction models reaches a credibility interval of CI ≥ 95% when the updated likelihood of default is equal to 60%. As expected, the Efficiency Comparison Test ECZETA=.00243 is invariant under the Bayesian transformation.

A Multi-industry Default Prediction Model using Logistic Regression and Decision Tree

The accurate prediction of corporate bankruptcy for the firms in different industries is of a great concern to investors and creditors, as the reduction of creditors' risk and a considerable amount of saving for an industry economy can be possible. Financial statements vary between industries. Therefore, economic intuition suggests that industry effects should be an important component in bankruptcy prediction. This study attempts to detail the characteristics of each industry using sector indicators. The results show significant relationship between probability of default and sector indicators. The results of this study may improve the default prediction models performance and reduce the costs of risk management.

A Proposed Model for Prediction of Industrial Sickness

The objective of this study is to examine the performance of default prediction model: the Z-score model using discriminant analysis, and to propose a new prediction model on a data set of 30 defaulted and 30 solvent companies. Financial ratios obtained from corporate balance sheets are used as independent variables while solvent/defaulted company (ratings assigned) is the dependent variable. The predictive ability of the proposed Z score model is higher when compared to both the Altman original Z-score model and the Altman model for emerging markets. The research findings establish the superiority of proposed model over default discriminant analysis and demonstrate the significance of accounting ratios in predicting default.

An Empirical Analysis of Default Prediction Models: Evidence from lndian Listed Companies

This paper attempts to evaluate the predictive ability of three default prediction models: the market-based KMV model, the Z-score model using discriminant analysis (DA), and the logit model; and identifies the key default drivers. The research extends prior empirical work by modeling and testing the impact of financial ratios, macro-economic factors, corporate governance and firm-specific variables in predicting default. For the market-based model, the author has extended the works of KMV in developing a suitable algorithm for determining probability of default (PD). While for the KMV model, the continuous observations of PD are used as the dependent variable, for the accounting-based models, ratings assigned are the proxy for default (those rated ’D’ are defaulted and rated ‘AAA’ and ‘A’ are solvent). The research findings largely support the hypothesis that solvency, profitability and liquidity ratios do impact the default risk, but adding other covariates improves the predictive ability of the models. Through this study, the author recommends that accounting –based models and market based models are conceptually different. While market-based models are forward looking and inclusion of market data makes the default risk quantifiable; to make the PD more exhaustive, it is important to factor in the information provided in the financial statements. The conclusions drawn are that the disclosures in financial statements can help predict default risk as financial distress risk is likely to evolve over time and will be reflected in financial statements beyond accounting ratios. Moreover this will also help divulge “creative accounting” practices by corporates.

AN INSURANCE MODEL FOR THE PROTECTION OF CORPORATIONS AGAINST THE BANKRUPTCY OF SUPPLIERS BY USING THE BLACK-SCHOLES-MERTON MODEL

Many banks provide supply-chain finance solutions that might include insurance services that further mitigate trade risk such as the default of suppliers. This study proposes the development of an insurance model that uses the Black-Scholes-Merton Model (BSM) for default prediction and risk pooling management techniques as a way to reduce the risk due to supplier bankruptcy and estimate an insurance premium that banks can use to charge this service to their customers. In order to demonstrate the use of the proposed insurance model, a sample of companies is selected from the New York Stock exchange and data for historical stock prices from the CRSP database (Center for Research in Security Prices) is collected in order to calculate the probability of bankruptcy of a sample of sup- pliers from different industries by using the BSM model. Twelve pools of companies of different sizes are created and a VBA program for Excel is developed in order to calculate probability of bankruptcy tables of companies be- longing to the different pools. A Monte Carlo simulation to simulate the impact on risk and expected losses on the number of insurance policies sold is implemented with the use of simulation software. The results show that the simulation is useful to estimate the number of sold policies required in order to reduce the risk to a minimum level and predict with a high level of certainty the losses due to bankruptcy of suppliers. The expected losses for a risk pool can be used by a financial institution in order to price an insurance contract that hedges a company against the risk of default of suppliers.

Bank Default Prediction: A Comparative Model using Principal Component Analysis

Bank default prediction continues to draw attention given the ongoing effects of the recent financial crisis. Seminal works have found that structural models are better predictors of default. In this paper I argue that accounting models predictive ability have been weakened due to the multicollinearity problem and propose principal component analysis to improve the accounting model. The paper then compares accounting and structural default prediction models using a logit analysis and further evaluates the performance of a combination of accounting and structural default models to predict default. The paper uses panel data on US banks from the Federal Deposit Insurance Corporation database between 1995-2012 and the analysis is developed on 519 defaulted bank years and 5,965 non defaulted bank years. The accounting model is improved and outperforms the structural model; the study also finds that a combination of both models performs better than any one model at predicting default in the US banking system.

Corporate governance characteristics and default prediction modeling for small enterprises. An empirical analysis of Italian firms

The aim of this paper is to analyze how the relationship between corporate governance mechanisms and business failure changes in small enterprises (SEs) compared to larger firms.Logistic regression was applied to a sample of 934 Italian SEs, and a SE default prediction model built based on both financial ratios and corporate governance characteristics. The accuracy rates obtained by this model were then compared to those from a second model, based on the same sample of firms, which used only financial ratios as predictive variables.The findings are the following: i) CEO duality, owner concentration, and a reduced number of outside directors on the board (no more than 50%) are significantly and negatively correlated with small company default and ii) corporate governance variables significantly improve the SE default prediction accuracy rates.

An Alternative to the Standardized Approach for Assessing Credit Risk Under the Basel Accord

The current standardized approach for assessing credit risk under Basel III depends on the ratings assigned by credit rating agencies (CRAs). This approach presents three problems. First, definitions of ratings of CRAs on assessing the likelihood of default and recovery rates are not uniform. Second, because CRAs assign the rating according TTC (through-the-cycle) ratings, the ratings of CRAs less accurate in predicting short term defaults and react slowly to credit events. Third, there are not many companies whose rating are assigned by CRAs in Japan. To improve the standardized approach under Basel III, we propose a new method to evaluate credit risk without CRAs. We analyze how companies' financial and non-financial attributes affect default and construct a default prediction model using the annual reports of companies listed on the TSE (Tokyo Stock Exchange) from fiscal 2003 to fiscal 2009. Furthermore, using this model, we construct new scoring and rating model. The results indicate that our rating anticipates the likelihood of default at least as accurately as the ratings by CRAs.

An Empirical Analysis of Default Risk for Listed Companies in India: A Comparison of Two Prediction Models

The objective of this study is to examine the performance of two default prediction models: the Z-score modelusing discriminant analysis, and the logit model on a dataset of 60 defaulted and 60 solvent companies. Financialratios obtained from corporate balance sheets are used as independent variables while solvent/defaulted company(ratings assigned) is the dependent variable. Furthermore, for logistic regressions, an attempt is made to combinemacro variables and dummy industry variables along with accounting ratios. The predictive ability of theproposed Z score model is higher when compared to both the Altman original Z-score model and the Altmanmodel for emerging markets. The research findings establish the superiority of logit model over discriminantanalysis and demonstrate the significance of accounting ratios in predicting default.

A default prediction model for Italian SMEs: the relevance of the capital structure

Small and medium enterprises (SMEs) play a fundamental role in the economy of many countries, including Italy. The simplicity of the financial structure of Italian SMEs, relying mostly on bank loans, makes them riskier than their larger counterparts and this induces banks to develop default prediction models specifically addressing the management of SMEs’ credit relationship. Building on this framework, this article aims to analyse the determinants of the default probability of a sample of 9208 Italian limited liabilities SMEs in a time frame of 3 years, over the period 2006 to 2010. Specifically, this article adopts a logistic regression model estimated on a database provided by the Centrale Rischi Finanziari (CRIF), an Italian credit rating agency. The results show that, among the observable financial and economic characteristics of the firms, the capital structure (both in terms of internal and external funds and in terms of the source of external financing) and interest expenses are more relevant than economic variables as determinants of SMEs’ default.

Default Prediction for Real Estate Companies with Imbalanced Dataset

When analyzing default predictions in real estate companies, the number of non-defaulted cases always greatly exceeds the defaulted ones, which creates the two- class imbalance problem. This lowers the ability of prediction models to distinguish the default sample. In order to avoid this sample selection bias and to improve the prediction model, this paper applies a minority sample generation approach to create new minority samples. The logistic regression, support vector machine (SVM) classification, and neural network (NN) classification use an imbalanced dataset. They were used as benchmarks with a single prediction model that used a balanced dataset corrected by the minority samples generation approach. Instead of using prediction- oriented tests and the overall accuracy, the true positive rate (TPR), the true negative rate (TNR), G-mean, and F-score are used to measure the performance of default prediction models for imbalanced dataset. In this paper, we describe an empirical experiment that used a sampling of 14 default and 315 non-default listed real estate companies in China and report that most results using single prediction models with a balanced dataset generated better results than an imbalanced dataset. Keywords—Default prediction, Imbalanced dataset, Real estate listed companies, Minority- sample generation approach

MCDM APPROACH TO EVALUATING BANK LOAN DEFAULT MODELS

Banks and financial institutions rely on loan default prediction models in credit risk management. An important yet challenging task in developing and applying default classification models is model evaluation and selection. This study proposes an evaluation approach for bank loan default classification models based on multiple criteria decision making (MCDM) methods. A large real-life Chinese bank loan dataset is used to validate the proposed approach. Specifically, a set of performance metrics is utilized to measure a selection of statistical and machine-learning default models. The technique for order preference by similarity to ideal solution (TOPSIS), a MCDM method, takes the performances of default classification models on multiple performance metrics as inputs to generate a ranking of default risk models. In addition, feature selection and sampling techniques are applied to the data pre-processing step to handle high dimensionality and class unbalancedness of bank loan default data. The results show that K-Nearest Neighbor algorithm has a good potential in bank loan default prediction.

Application of Business Risk Prediction Model: Based on the Logistic Regression Model

Credit risk is one of the three components making up financial risk. Under the New Basel Capital Accord,default risk has been listed as the most important factor for credit risk among all elements that affect risk ofcredit. Banks in China currently leave large quantities of cash idle due to difficulty in loan recovery. This essayfirst analyzes the distributional features of variables’ cross-section data concerning the default rate. Based oncredible data, this research then undertakes the choice of an appropriate default prediction model. The BinaryLogistic Regression Model is adopted here to build the default rate model of business credit risk and to analyzethe risk information generated, in hopes of helping banks find the correct loaning strategies.

The effect of internationalisation on modelling credit risk for SMEs: Evidence from UK market

This study considers domestic and international small and medium-sized enterprises (SMEs) of the United Kingdom separately while modelling their default risk. To establish the empirical validation, separate one-year default prediction models are developed using dynamic logistic regression technique that encapsulates significant financial information over an analysis period of 2000 to 2009. Almost an identical set of explanatory variables affect the default probability of domestic and international SMEs, which contradicts the need for separate default risk models. However, the lower predictive accuracy measures of the model developed for international SMEs motivate us to compare the weights of regression coefficients of the models developed for domestic and international firms. Test results confirm that four out of the nine common predictors display significant statistical differences in their weights. However, these differences do not contribute to the discriminatory performance of the default prediction models, given that we report very little difference in each model's classification performance.

Distance-to-default measures and determinants for systemically important financial institutions

Purpose – This study aims to compare two distance-to-default methods, data-transformed maximum likelihood estimation and “naïve”, that are suitable for financial institutions. The links between these measures and asset size, Tier 1 and Tier 2 capital ratios, non-performing assets and operating efficiency have been examined and an alternative default risk measure has been introduced. Most of the market-based distance-to-default measures are not appropriate for banks due to their unique debt structure. Design/methodology/approach – The author has compared two distance-to-default measures and has identified their accounting determinants using Pearson’s correlation and regressions with clustered standard errors. The sample of the US-based systemically important financial institutions covers the period from 2000 to 2010. Findings – Non-performing assets and operating efficiency are found to be statistically and economically significant determinants of both distance-to-default measures. Tier 1 capital ratio is not a significant indicator of default risk. Practical implications – The results emphasize the importance of using a combination of market-based default risk measures and accounting ratios in default prediction models for the financial institutions. Originality/value – This paper identifies accounting determinants of two distance-to-default measures for large financial institutions, before and during the 2008 financial crisis. It introduces a spread between two measures as an alternative default risk indicator.

The Effect of Internationalization on Modeling Credit Risk for SMEs: Evidence from UK Market

This study considers domestic and international small and medium-sized enterprises (SMEs) of the United Kingdom separately while modelling their default risk. To establish the empirical validation, separate one-year default prediction models are developed using dynamic logistic regression technique that encapsulates significant financial information over an analysis period of 2000 to 2009. Almost an identical set of explanatory variables affect the default probability of domestic and international SMEs, which contradicts the need for separate default risk models. However, the lower predictive accuracy measures of the model developed for international SMEs motivate us to compare the weights of regression coefficients of the models developed for domestic and international firms. Test results confirm that four out of the nine common predictors display significant statistical differences in their weights. However, these differences do not contribute to the discriminatory performance of the default prediction models, given that we report very little difference in each model’s classification performance.

Parametric and non-parametric combination model to enhance overall performance on default prediction

The probability of default (PD) is the key element in the New Basel Capital Accord and the most essential factor to financial institutions’ risk management. To obtain good PD estimation, practitioners and academics have put forward numerous default prediction models. However, how to use multiple models to enhance overall performance on default prediction remains untouched. In this paper, a parametric and non-parametric combination model is proposed. Firstly, binary logistic regression model (BLRM), support vector machine (SVM), and decision tree (DT) are used respectively to establish models with relatively stable and high performance. Secondly, in order to make further improvement to the overall performance, a combination model using the method of multiple discriminant analysis (MDA) is constructed. In this way, the coverage rate of the combination model is greatly improved, and the risk of miscarriage is effectively reduced. Lastly, the results of the combination model are analyzed by using the K-means clustering, and the clustering distribution is consistent with a normal distribution. The results show that the combination model based on parametric and non-parametric can effectively enhance the overall performance on default prediction.

The Next Generation of Default Prediction Models: Incorporating Signal Strength and Dependency

The financial market crisis has exposed a weakness in predicting defaults. A new wave of models must be more powerful and should be more successful in predicting multiple defaults. I argue that such models incorporate signal strength, cross and serial dependencies. The conditional default probability is the most powerful predictor and I demonstrate that heteroscedastic probits with random effects represent a flexible framework to estimate the conditional probability. The lower the variance the higher is the signal strength of the underlying credit score. I derive an approximate likelihood procedure, including asymptotic standard errors, and apply it successfully to a dataset.

Models for predicting default: towards efficient forecasts

Purpose– The purpose of this paper is to assess and compare the forecast ability of existing credit risk models, answering three questions: Can these methods adequately predict default events? Are there dominant methods? Is it safer to rely on a mix of methodologies?Design/methodology/approach– The authors examine four existing models: O-score, Z-score, Campbell, and Merton distance to default model (MDDM). The authors compare their ability to forecast defaults using three techniques: intra-cohort analysis, power curves and discrete hazard rate models.Findings– The authors conclude that better predictions demand a mix of models containing accounting and market information. The authors found evidence of the O-score's outperformance relative to the other models. The MDDM alone in the sample is not a sufficient default predictor. But discrete hazard rate models suggest that combining both should enhance default prediction models.Research limitations/implications– The analysed methods alone cannot adequately predict defaults. The authors found no dominant methods. Instead, it would be advisable to rely on a mix of methodologies, which use complementary information.Practical implications– Better forecasts demand a mix of models containing both accounting and market information.Originality/value– The findings suggest that more precise default prediction models can be built by combining information from different sources in reduced-form models and combining default prediction models that can analyze said information.

Combining accounting data and a structural model for predicting credit ratings: Empirical evidence from European listed firms

Ratings issued by credit rating agencies (CRAs) play an important role in the global financial environment. Among other issues, past studies have explored the potential for predicting these ratings using a variety of explanatory factors and modeling approaches. This paper describes a multi-criteria classification approach that combines accounting data with a structural default prediction model in order to obtain improved predictions and test the incremental information that a structural model provides in this context. Empirical results are presented for a panel data set of European listed firms during the period 2002–2012. The analysis indicates that a distance-to-default measure obtained from a structural model adds significant information compared to popular financial ratios. Nevertheless, its power is considerably weakened when market capitalization is also considered. The robustness of the results is examined over time and under different rating category specifications.