Accuracy Of Default Predictions Research Articles

Tourism SME default: A note on predictors

This note addresses the lack of progress on small and medium enterprises (SMEs) default prediction in the tourism sector. Using the case of Croatian tourism SMEs, this note applies Altman's recently introduced Omega Score model to (i) predict SME default, (ii) explore and contrast important variables for default prediction in tourism vs. other sectors, and (iii) adapt the original Omega Score to tourism sector specifics. Financial ratios have historically been prominent among the predictors in scholarly investigations. However, our findings underscore the significance of supplementing these ratios with other predictors of tourism SME default, specifically with information on (i) creditor payment behaviour and (ii) employee characteristics and relationships with them. Furthermore, our research demonstrates that a refined selection of six variables, accompanied by adjusted weightings, can significantly enhance default prediction accuracy. The results provide valuable lessons for the strategic management of tourism firms.

Inter-industry network and credit risk

As previous literature has documented cross-industry returns and tail risk predictability, especially during a financial crisis, this research investigates the effects of industries' position within an economy, inter-industry connectedness, and industry returns on credit risk using a reduced-form approach. We employ an aggregate measure of tail risk emitted from an industry to capture its outgoing connectedness with other industries, and our empirical results show that the outgoing connectedness of some highly central industries positively impacts all sample firms’ default probabilities, controlling for a variety of firm-specific and macroeconomic variables that are well-known to be related to corporate defaults. In sum, our empirical results support that industry network risk helps explain corporate default and improves default prediction accuracy.

Long-horizon predictions of credit default with inconsistent customers

We developed a decision support framework for default predictions that addresses two common issues: inconsistent customers and predictions of future defaults. We developed a T−m default prediction model using multivariate adaptive regression splines to address the methodological challenges. We confirm that this model outperforms typical approaches in terms of default prediction accuracy. Furthermore, an empirical application of our new framework involving unique data on defaults among Chinese-listed companies yields several substantive insights. Owing to the high interpretability of our predictions, we identify certain industry sectors that should receive high (and low) credit risk assessments. In addition, our research has important implications for the investment decisions of financial institutions and investors and government regulations.

Improving Credit Risk Assessment through Deep Learning-based Consumer Loan Default Prediction Model

This study aims to enhance credit risk identification, improve loan borrower review efficiency, and increase default prediction accuracy rate using data mining and machine learning techniques. The study also employs deep learning to develop a consumer loan default prediction model that minimizes credit risks and ensures consistent development. The researchers collected data from a survey of 1000 participants, stratified into local and foreign banks, and selected the top 11 banks based on turnover and customer volume. To construct the machine learning model, Keras, a neural network library that runs on TensorFlow, was utilized. The model predicts loan applicant default likelihood. The study's practical implications demonstrate a noteworthy success rate of customer default prediction, which can significantly benefit banks. The model was evaluated on a test set of 250 records and achieved a test set accuracy of 95.2%, correctly predicting the default state of 238 out of 250 respondents.

Predicting SMEs’ default risk: Evidence from bank-firm relationship data

This paper uses a probit model on a unique dataset of 13,081 Italian firms and 111 co-operative banks involved in the lending process to provide empirical evidence suggesting that the use and violations of credit lines and long-term loan overruns predict one-year and two-year probability of default (PD). The analysis controls for balance sheet indicators and time varying bank characteristics, captured by bank-time fixed effects. When combined with accounting data, credit-related indicators obtained from private internal banking sources improve small and medium-sized enterprises’ (SMEs) default prediction. The marginal benefit of the bank-firm specific information is also assessed by comparing the default prediction accuracy of a model that incorporates accounting information with that of a full model including private information. In terms of heterogeneity, the association between the balance sheet indicators and data on bank-firm relationships and default probability can vary across sectors and geographies. This highlights the importance for banks of specific analysis to better assess risk at the firm level.

Numerological Heuristics and Credit Risk in Peer-to-Peer Lending

People often use heuristics as mental shortcuts when making financial decisions. Although the literature typically considers heuristics as behavior biases, we explore how different types of heuristics differ from one another. Through peer-to-peer lending data, we observe that borrowers who use limited attention when applying for loans tend to choose round loan amounts, simplifying the decision-making process but compromising accuracy. This round-number heuristic decreases their funding success rate and increases the probability of default. On the other hand, some borrowers select loan amounts in “lucky numbers” that superstitious lenders favor. This lucky-number heuristic caters to the lenders’ preference, thus increasing the borrowers’ funding success rates and reducing the likelihood of default. Our paper demonstrates that borrowers select heuristics based on their motives, leading to varying consequences. We also show that heuristics are not all the same, and people’s choice of heuristics provides insight into their characteristics and can predict decision outcomes. For instance, factoring in heuristic usage information improves default prediction accuracy in our setting. Our findings can be beneficial to practitioners in refining the underwriting and screening of borrowers and loans.

What should lenders be more concerned about? Developing a profit-driven loan default prediction model

Reliable and effective loan default risk prediction can help regulators and lenders effectively identify risky loan applicants and develop proactive and timely response measures to enhance the stability of the financial system. Traditional prediction models concentrate more on improving loan default prediction accuracy, while neglecting to take profit maximization as the goal and evaluation measure of model construction. In this study, a novel profit-driven prediction model is proposed, taking a profit indicator as the optimization objective of the Bayesian optimization to optimize the hyperparameters of the predictor-categorical boosting. The Shapley additive explanations (SHAP) value is then calculated to further interpret the relationship between the input variables and the predicted values. Based on two datasets from Renrendai and Lending Club, the experimental results and statistical test indicate that the proposed model achieves the highest profit-related evaluation metrics values, with the mean average extra profit rate values of 3.0872% and 2.1858% respectively, and the mean Profit values of 5168.8762 and 352.9787 in two datasets respectively. SHAP value further reveals the key factors that will impact predictive output, which provides more valuable information for platforms and lenders for identifying possible defaulters.

EFFECT OF THE COMPANY RELATIONSHIP NETWORK ON DEFAULT PREDICTION: EVIDENCE FROM CHINESE LISTED COMPANIES

The default risk of listed companies not only threatens the interests of enterprises and internal staff but also leads the investors to face significant financial losses. Thus, this study attempts to establish an effective default prediction system for better corporate governance. In present times, it is not uncommon for a senior manager to serve in two or more companies. Our contribution has threefold. First, we construct an indicator system of default prediction for Chinese listed companies by considering the company relationship score. Then, we reversely infer the optimal ratios of the default and nondefault companies’ degrees of influence on their related companies with the maximum area under the curve (AUC). Third, the empirical results show that the default prediction accuracy is improved by using our indicator system that includes the company relationship score.

What Best Predicts Corporate Bank Loan Defaults? An Analysis of Three Different Variable Domains

This paper aims to compare the accuracy of financial ratios, tax arrears and annual report submission delays for the prediction of bank loan defaults. To achieve this, 12 variables from these three domains are used, while the study applies a longitudinal whole-population dataset from an Estonian commercial bank with 12,901 observations of defaulted and non-defaulted firms. The analysis is performed using statistical (logistic regression) and machine learning (neural networks) methods. Out of the three domains used, tax arrears show high prediction capabilities for bank loan defaults, while financial ratios and reporting delays are individually not useful for that purpose. The best default prediction accuracies were 83.5% with tax arrears only and 89.1% with all variables combined. The study contributes to the extant literature by enhancing the bank loan default prediction accuracy with the introduction of novel variables based on tax arrears, and also by indicating the pecking order of satisfying creditors’ claims in the firm failure process.

Understanding Credit Risk for Chinese Companies

In response to the recent elevated corporate credit risk environment in China, we develop a probability of default (PD) measure based on corporate bond defaults using the machine learning technique. We document a large pricing effect of corporate credit risk based on our PD measure in the primary and secondary corporate bond markets especially following the first bond default in 2014. In the cross section of corporate bond returns, there is a positive credit risk premium even after controlling for common risk factors. Finally, stocks of low PD firms outperform those of high PD firms during the COVID-19 pandemic.

Combining multiple probability predictions in the presence of class imbalance to discriminate between potential bad and good borrowers in the peer-to-peer lending market

Credit risk scoring predictions represent an effective guide for lenders to discriminate between potential good (who will repay the loan) and bad (who will default) borrowers in the online social lending market. A common characteristic of such a market is a lower percentage of defaulted borrowers than non-defaulted borrowers; thus, the sample is class imbalanced. Class imbalance may affect the accuracy of default predictions, as classifiers tend to be biased towards the majority class (good borrowers). We analyse the default prediction performance when combining class rebalancing methods with different regression and machine learning techniques. We also propose to combine multiple probability predictions to improve the predictive performance. The analysis is based on a book of loans (with a three-year term) funded in the 2010–2015 period though the online platform of Lending Club. The results show that some measures of predictive accuracy tend to improve when the scoring models are trained using a rebalanced, rather than an imbalanced sample, except when the extreme gradient boosting approach is applied. Finally, we find that combining multiple probability predictions via regularised logistic regression may help to improve the predictive accuracy.

On assessing the relative performance of default predictions

AbstractWe compare the accuracy of default predictions as, for instance, produced by professional rating agencies. We extend previous results on partial orderings to nonidentical sets of obligors and show that the calibration requirement virtually rules out the possibility of some partial orderings and that the partial ordering based on the ROC curve is most easily achieved in practice. As an example, we show for more than 5,000 firms rated by Moody's and S&P that these ratings cannot be ranked according to their grade distributions given default or nondefault, but that Moody's dominate S&P with respect to the ROC criterion and the Gini curve.

Default prediction models: The role of forward-looking measures of returns and volatility

Abstract This paper proposes a variant application of the Merton distance-to-default model by employing implied volatility and implied cost of capital to predict defaults. The proposed model’s results are compared with predictions obtained from three popular models in different setups. We find that our “best” model, which contains forward-looking proxies of returns and volatility outperform other models, carries a default prediction accuracy rate of 89%. Additional analysis using a discrete-time hazard model indicates the pseudo-R2 values from regression models that include the two forward-looking measures are as high as 51%. Overall, our results establish the informational relevance of implied cost of capital and implied volatility in predicting defaults.

Mandatory IFRS Adoption and Changes in Default Prediction Accuracy

The study tests for changes in default prediction accuracy following the country-level switch to International Financial Reporting Standards (IFRS) in 20 European Union (EU) and nonEU jurisdictions. Using a default prediction model that combines both accounting and market inputs, we find that compared to a control group of non-adopters, IFRS adopters do not benefit from the international standards in terms of default prediction accuracy. Further cross-sectional analyses of companies from adopting jurisdictions in the post-adoption period show that voluntary adopters and companies domiciled in jurisdictions with large distances between their local GAAP and IFRS have higher default prediction accuracy. Higher enforcement is associated with higher default prediction accuracy in the EU and with lower default prediction accuracy in non-EU jurisdictions. The results add to the literature on the consequences of IFRS adoption and speak to the performance of default prediction models across jurisdictions, given a change in accounting regulation.

Comparing the accuracy of default predictions in the rating industry for different sets of obligors

We generalize the refinement ordering for well calibrated probability forecasts to the case were the debtors under consideration are not necessarily identical. This ordering is consistent with many well known skill scores used in practice. We also add an illustration using default predictions made by the leading rating agencies Moody’s and S&P.

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.

The Informational Relevance of Forward-Looking Measures of Returns and Volatility in Forecasting Defaults

This paper proposes a variant application of the Merton distance-to-default model by employing implied volatility and implied cost of capital to forecast defaults. The proposed model's results are compared with predictions obtained from the three popular models in different setups. We find that our best model, which contains both forward-looking proxies, outperforms all other models with a default prediction accuracy rate of 89%. Additional analysis using a discrete-time hazard model indicates that the psuedo-R² values from regression models that include the two implied measures are as high as 51%. Overall, our results establish the informational relevance of implied volatility and implied cost of capital in predicting defaults.

SMEs Default Prediction Model: An Analysis of the Existing Literature

The ability to accurately and consistently predict small firms’ default would be beneficial for both lending banks and firms and hence could help to build a more stable and prosperous economy. In this study we use more than 30,000 UK small and medium sized enterprises (SMEs) covering the period from 2000 to 2008 and from 22 manufacturing industries to analyse the existing accounting-based default prediction model for UK SMEs. We tested the prominent SMEs default prediction model introduced by Altman et al. (2010) using logistic regressions to establish their reliability and prediction power for UK sample. The evaluation of the SMEs model has been carried out by comparing the accuracy of the model in predicting default events with Ohlson model which is mainly developed for large firms. Moreover, different sample tests revealed that SMEs’ credit risk behavior is unique and it is crucial to build the default prediction models specifically for these firms to gain the most accurate and reliable outcome. This study also sheds light on effect of economic cycles on SMEs default prediction accuracy. Finally, our results ascertain that controlling for the effects of industry and ownership structure significantly enhances the accuracy of the default prediction model for UK SMEs.

Combining structural models and accounting‐based models for measuring credit risk in real estate companies

PurposeThe purpose of this paper is to, first, analyse to what extent the default probability based on structural models provides additional information and that accounting ratios do not contemplate. Second, to design hybrid models by including the default probability from structural models as explanatory variable, in addition to accounting ratios, in order to evaluate the differences in the accuracy of default predictions using an accounting‐based model and a hybrid model.Design/methodology/approachThe authors calculated the scores from the accounting models annually during the period from 2003 to 2007 and estimated several structural models.FindingsThe results show that the market information obtained from the structural models includes additional information not reflected in the accounting information. Also, it can be concluded that including default probability from structural models as an explanatory variable allows the out‐sample predictive capacity of accounting‐based models to be improved.Practical implicationsThe study highlights the importance of combining a structural model with an accounting model rather than expending energy on determining which of the two provides a greater predictive capacity. In fact, recent literature demonstrates no superiority of one approach over the other because both approaches capture different aspects related to the risk of bankruptcy in companies and they should be combined to improve credit risk management.Originality/valueThis study expands on the existing literature on the probability of business failure in the real estate sector. The authors present a comparative analysis of the accuracy of default predictions using accounting‐based models and hybrid models which will consider the default probability implicit in market information.

On the Determinants of the Implied Default Barrier

We use the maximum likelihood (ML) estimation approach to estimate the default barriers from market values of equities for a sample of 762 public industrial Canadian firms. The ML approach allows estimating simultaneously the asset's instantaneous drift, volatility and the barrier level, when the firm's equity is priced as a Down-and-Out European call (DOC) option. We find that the estimated barrier is positive and significant in our sample. Moreover, we compare the default prediction accuracy of the DOC framework with the KMV-Merton approach. Using probit estimation, we find that the default probability from the two structural models provides similar in-sample fits, but the barrier option framework achieves better out-of-sample forecasts. Regression analysis shows that leverage is not the only determinant of the default barrier. The implied default threshold is also positively related with financing costs, and negatively to liquidity, asset's volatility and firm's size. We also find that liquidation costs, renegotiation frictions and equity holders bargaining power increase the implied default barrier level.