Predicting the outcome following bankruptcy filing: a three-state classification using neural networks

This paper uses artificial neural networks (ANNs), multi-state ordered logit and nonparametric multiple discriminant analysis (NPDA) for predicting the three-state outcome of bankruptcy filing. The study compares the classification accuracy of these procedures. It differs from previous studies on predicting financial distress by focusing on the firm after the filing of bankruptcy using accounting data, market data, and court-related information. Following the filing and through court approval the bankruptcy is resolved as firms are either acquired by other firms, emerging as independent operating entities, or liquidated. Distinguishing this three-state outcome is more complex than discriminating between healthy and financially distressed firms. Models suggested in previous studies for predicting the two-group financial distress perform poorly for our three-state scenario. Therefore, we develop models which focus on characteristics relevant for the bankruptcy resolution. We use a sample of 237 publicly traded firms which have complete data. For the entire sample and estimation samples, ANNs provide significantly better three-state classification than logit and NPDA. However, for some holdout samples the differences in classification accuracies are statistically insignificant. © 1997 John Wiley & Sons, Ltd.     

Empirical comparison of hazard models in predicting SMEs failure

This study aims to shed light on the debate concerning the choice between discrete-time and continuous-time hazard models in making bankruptcy or any binary prediction using interval censored data. Building on the theoretical suggestions from various disciplines, we empirically compare widely used discrete-time hazard models (with logit and clog-log links) and the continuous-time Cox Proportional Hazards (CPH) model in predicting bankruptcy and financial distress of the United States Small and Medium-sized Enterprises (SMEs). Consistent with the theoretical arguments, we report that discrete-time hazard models are superior to the continuous-time CPH model in making binary predictions using interval censored data. Moreover, hazard models developed using a failure definition based jointly on bankruptcy laws and firms’ financial health exhibit superior goodness of fit and classification measures, in comparison to models that employ a failure definition based either on bankruptcy laws or firms’ financial health alone.     

Can Language Predict Bankruptcy? The Explanatory Power of Tone in 10‐K Filings

We examine whether the language used in 10‐K filings reflects a firm's risk of bankruptcy. Our sample contains 424 bankrupt U.S. companies in the period 1994–2015 and we use propensity score matching to find healthy matches. Based on a logit model of failing and vital firms, our findings indicate that firms at risk of bankruptcy use significantly more negative words in their 10‐K filings than comparable vital companies. This relationship holds up until three years prior to the actual bankruptcy filing. With our investigation, we confirm the results from previous accounting and finance research. 10‐K filings contain valuable information beyond the reported financials. Additionally, we show that 10‐Ks filed in the year of a firm's collapse contain an increased number of litigious words relative to healthy businesses. This indicates that the management of failing firms is already dealing with legal issues when reporting financials prior to bankruptcy. Our results suggest that analysts ought to include the presentation of financials in their assessment of bankruptcy risk as it contains explanatory and predictive power beyond the financial ratios.     

BANKRUPTCY DISCRIMINATION WITH REAL VARIABLES

This paper reconsiders the accepted usage of nondeflated financial ratios in statistical models to differentiate between failed and nonfailed firms. Non-deflated ratios are hypothesized to inadequately reflect inter-temporal macroeconomic fluctuations that affect the ability of firm's to survive. Using a sample of 124 oil and gas companies between the period 1982–1988, the going concern assumption is evaluated with statistical logit models using either nondeflated or deflated financial ratios. Deflated company ratios are created by transforming data with price indices or by creating market value ratios. Empirical results suggest that a superior bankruptcy early warning model is developed for the oil and gas industry by creating real financial and reserve ratios and by introducing external factors, such as oil prices, interest rates and accounting method, as independent predictors. Overall classification accuracy is approximately 95 percent.     

Modeling default risk with support vector machines

Predicting default risk is important for firms and banks to operate successfully. There are many reasons to use nonlinear techniques for predicting bankruptcy from financial ratios. Here we propose the so-called Support Vector Machine (SVM) to predict the default risk of German firms. Our analysis is based on the Creditreform database. In all tests performed in this paper the nonlinear model classified by SVM exceeds the benchmark logit model, based on the same predictors, in terms of the performance metric, AR. The empirical evidence is in favor of the SVM for classification, especially in the linear non-separable case. The sensitivity investigation and a corresponding visualization tool reveal that the classifying ability of SVM appears to be superior over a wide range of SVM parameters. In terms of the empirical results obtained by SVM, the eight most important predictors related to bankruptcy for these German firms belong to the ratios of activity, profitability, liquidity, leverage and the percentage of incremental inventories. Some of the financial ratios selected by the SVM model are new because they have a strong nonlinear dependence on the default risk but a weak linear dependence that therefore cannot be captured by the usual linear models such as the DA and logit models.     

Strategic Bankruptcies and Price Reactions to Bankruptcy Filings

This study investigates whether the stock market differentiates between firms that file bankruptcy petitions for strategic reasons and firms that file bankruptcy petitions for financial reasons. We perform both univariate and regression tests on a sample of 245 firms that filed Chapter 11 bankruptcy petitions between 1981 and 1996. After controlling for bankruptcy outcome, probability of bankruptcy, firm financial condition, and firm size, we find that, in the period around bankruptcy filing, firms that file bankruptcy petitions for financial reasons have significantly larger stock price declines than firms that file bankruptcy petitions for strategic reasons.     

Owner liability and financial reporting information as predictors of firm default in bank loans

We examine the effects of owner liability and non-accounting and financial accounting information on the probability of default as defined in Basel II in bank loan contracted by non listed firms. We model default as a function of owner liability and accounting and non-accounting information of non-listed firms, drawing on 43,117 annual accounts of 16,029 firms over a 7-year period. Our estimations based on mixed logistic regressions with random parameters show that the predicted default probability of full-liability firms is 0.72 times that of limited liability firms. The likelihood ratio test for omitted variables confirms the additional predictive ability of liability status over and above other non-accounting and financial accounting information. A Heckman self-selection model does not indicate sampling bias. The particular definition of default used in the study enables the findings to be generalizable across other institutional contexts.     

A hybrid bankruptcy prediction model with dynamic loadings on accounting-ratio-based and market-based information: A binary quantile regression approach

While using the binary quantile regression (BQR) model, we establish a hybrid bankruptcy prediction model with dynamic loadings for both the accounting-ratio-based and market-based information. Using the proposed model, we conduct an empirical study on a dataset comprising of default events during the period from 1996 to 2006. In this study, those firms experienced bankruptcy/liquidation events as defined by the Compustat database are classified as “defaulted” firms, whereas all other firms listed in the Fortune 500 with over a B-rating during the same time period are identified as “survived” firms. The empirical findings of this study are consistent with the following notions. The distance-to-default (DD) variable derived from the market-based model is statistically significant in explaining the observed default events, particularly of those firms with relatively poor credit quality (i.e., high credit risk). Conversely, the z-score obtained with the accounting-ratio-based approach is statistically significant in predicting bankruptcies of firms of relatively good credit quality (i.e., low credit risk). In-sample and out-of-sample bankruptcy prediction tests demonstrated the superior performance of utilizing dynamic loadings rather than constant loadings derived by the conventional logit model.     

Bankruptcy and Institutions: Theory and Empirical Evidence from Korea and the United States

Using a simple z-score bankruptcy model, this article explores the relationship between bankruptcy threshold and institutions. The z-score threshold for bankruptcy is found to be higher in countries with stronger institutions. To test this claim, a cross-section data set of 86 Korean firms and 60 US firms from 1991 to 2001, extracted from a panel data set, is used. The empirical finding that the z-score bankruptcy threshold in the United States (which has better quality of institutions than does Korea) is higher than that in Korea is consistent with the prediction of the model. Additionally, having examined bankruptcy laws of the two countries, it is found that filing a petition for bankruptcy is easier and debtors rights are better protected in the United States than in Korea, which suggests that the bankruptcy laws of Korea and the United States may be partially responsible for the difference in the z-score threshold for bankruptcy.     

Sources of gains to shareholders from bankruptcy resolution

Using a logistic regression model, we identify the characteristics of firms whose shareholders are likely to benefit from bankruptcy resolution. That is, winners (losers) are firms whose shareholders experience positive (negative) excess returns after bankruptcy filing. We find that winners are relatively smaller firms with higher proportions of convertible debt, tend to file for bankruptcy for strategic reasons, have low share-ownership concentration, and suffer comparatively larger pre-filing stock price declines. Among winners, shareholder returns are greater for firms that have higher levels of private debt and research and development (R&D) expenditures, and operate in more concentrated industries. In addition, our analysis indicates that an ex ante trading strategy of purchasing bankrupt stocks with a greater than 50% probability of being a winner on the day after bankruptcy filing and holding the stocks for a year, on an average, can generate average compounded and excess compounded holding-period returns of +71% and +42%, respectively.     

系统性风险与企业财务危机预警——基于前沿机器学习的新视角

本文采用Logit回归模型以及随机森林模型、梯度提升模型等前沿机器学习方法,深入考察系统性风险指标对我国企业财务危机的预测能力。结果表明,系统性风险对中下游企业的财务危机具有显著的预测能力,而基于因子分析构建的系统性风险指标,结合随机森林模型可取得更好的预测效果。本文进一步区分财务危机的不同成因并发现,基于随机森林模型和Logit回归模型的预测框架能够对我国大多数财务危机事件进行有效预警。在此基础上,本文对我国上市企业监管提出相关建议,从而为完善金融风险处置机制提供一定参考。     

Audit committee quality,auditor independence,and internal control weaknesses

In this paper we investigate the relation between audit committee quality, auditor independence, and the disclosure of internal control weaknesses after the enactment of the Sarbanes-Oxley Act. We begin with a sample of firms with internal control weaknesses and, based on industry, size, and performance, match these firms to a sample of control firms without internal control weaknesses. Our conditional logit analyses indicate that a relation exists between audit committee quality, auditor independence, and internal control weaknesses. Firms are more likely to be identified with an internal control weakness, if their audit committees have less financial expertise or, more specifically, have less accounting financial expertise and non-accounting financial expertise. They are also more likely to be identified with an internal control weakness, if their auditors are more independent. In addition, firms with recent auditor changes are more likely to have internal control weaknesses.     

Beyond bankruptcy: Does the US bankruptcy code provide a fresh start to entrepreneurs?

This paper assesses the extent to which the US bankruptcy system is effective in providing small businesses a “fresh start” after a bankruptcy filing. I use data from the 1993, 1998 and 2003 National Survey of Small Business Finances to explore how firms fare after a bankruptcy filing. On the positive side, previously bankrupt firms are not any more burdened than the average small firm by problems relating to profitability, cash flow, health insurance costs, or taxes. Further, the fact that these firms are surviving several years after the filing is itself a testament to the efficient functioning of the US bankruptcy system. It suggests that the bankruptcy system goes a long way toward helping businesses recover after a bankruptcy filing.     

Bankruptcy prediction: Application of the Taylor's expansion in logistic regression

The purpose of the present study is to test whether Taylor's series expansion can be used to solve the problem associated with the functional form of bankruptcy prediction models. To avoid the problems associated with the normality of variables, the logistic model to describe the insolvency risk is applied. Taylor's expansion is then used to approximate the exponent of the logistic function, or the logit. The cash to total assets, cash flow to total assets, and shareholder's equity to total assets ratios operationalize the factors affecting the insolvency risk. The usefulness of Taylor's model in bankruptcy prediction is evaluated applying the logistic regression model to the data from the Compustat database. The classification accuracy in the test data for the first and second years before bankruptcy show that the classification accuracy of a simple financial ratio model can be increased using the second-order and interaction terms of these ratios. However, in the third year, for the test data, Taylor's expansion is not able to increase the classification accuracy when compared with the first-order model.     

Trading activity around chapter 11 filing

This paper uses institutional ownership data and order flow information to document and explain equity trading patterns prior to chapter 11 bankruptcy filing.We provide a model that predicts trading activity prior to filing which results from a difference of opinion amongst different types of investors about whether the firm should be liquidated. We then test trading data to show that trading activity is elevated around chapter 11 filing as the model predicts. We show how institutional holdings change around filing and that chapter 7 firms appear relatively more attractive to institutional investors than emerging firms around filing.     

Predicting Corporate Failure Using a Neural Network Approach

This paper investigates the performance of Artificial Neural Networks for the classification and subsequent prediction of business entities into failed and non-failed classes. Two techniques, back-propagation and Optimal Estimation Theory (OET), are used to train the neural networks to predict bankruptcy filings. The data are drawn from Compustat data tapes representing a cross-section of industries. The results obtained with the neural networks are compared with other well-known bankruptcy prediction techniques such as discriminant analysis, probit and logit, as well as against benchmarks provided by directly applying the bankruptcy prediction models developed by Altman (1968) and Ohlson (1980) to our data set. We control the degree of ‘disproportionate sampling’ by creating ‘training’ and ‘testing’ populations with proportions of bankrupt firms ranging from 1% to 50%. For each population, we apply each technique 50 times to determine stable accuracy rates in terms of Type I, Type II and Total Error. We show that the performance of various classification techniques, in terms of their classification errors, depends on the proportions of bankrupt firms in the training and testing data sets, the variables used in the models, and assumptions about the relative costs of Type I and Type II errors. The neural network solutions do not achieve the ‘magical’ results that literature in this field often promises, although there are notable 'pockets' of superior performance by the neural networks, depending on particular combinations of proportions of bankrupt firms in training and testing data sets and assumptions about the relative costs of Type I and Type II errors. However, since we tested only one architecture for the neural network, it will be necessary to investigate potential improvements in neural network performance through systematic changes in neural network architecture.     

Review of Categorical Models for Classification Issues in Accounting and Finance

Recent studies have extensively used the logit or probit models for classification problems in accounting and finance. More than 289 articles in prestigious journals have used these or similar methods from 1989 through 1996. This paper reviews several categorical techniques and compares the performance of logit or probit with alternative procedures. Intuitive and mathematical explanations of how the models examined differ in terms of underlying assumptions and other attributes are provided. The alternative techniques are applied to two substantive research questions: predicting bankruptcy and auditors' consistency judgements. Four empirical criteria provide some evidence that the exponential generalized beta of the second kind (EGB2), lomit, and burrit (all new to the accounting and finance literature) improve the log-likelihood functions, and the explanatory power, compared with logit and other models. EGB2, lomit and burrit also provide significantly better classifications and predictions than logit and other techniques.     

Firms’ performance under different bankruptcy systems: a Europe–USA empirical analysis

This study empirically analyses the effect that the bankruptcy law has on firms’ performance based on its financial situation. To do this, we considered the different types of efficiency and their influence on firms’ value. The study was carried out for Germany, Spain, the United States, France and the United Kingdom. We applied System‐GMM estimation to dynamic panel data. The main results show that under creditor‐oriented systems, there is a decrease in the value of both financially distressed firms and those filing for bankruptcy.     

基于非线性组合模型的P2P网贷平台危机预警研究

针对P2P网贷平台现金流较大、利润率较低和财务数据获取困难的特点,构建基于平台交易真实数据的危机预警评价指标体系和组合预测模型.将传统的财务评价指标转换成网贷平台交易数据指标,运用邻域粗糙集属性约简的方法对采集的数据指标进行降噪和约减处理,再基于机器学习理论引入神经网络、支持向量机和Logit回归等模型对数据进行训练.通过分组进行单模型和组合模拟预测,提高了新的破产指标下各模型预测的准确率.     

Earnings Management in Chapter 11 Bankruptcy

We study the impact of earnings management prior to bankruptcy filing on the passage of firms through Chapter 11. Using data on public US firms, we construct three measures of earnings management: a real activities manipulation measure (abnormal operating cash flows) and two accounting manipulation measures (discretionary accruals and abnormal working capital accruals). We find that, controlling for the impact of factors known to influence earnings management and firm survival in bankruptcy, earnings management prior to bankruptcy significantly reduces the likelihood of Chapter 11 plan confirmation and emergence from Chapter 11. The results are driven primarily by extreme values of earnings management, characterized by one or two standard deviations above or below the mean. The findings are consistent with creditors reacting positively to unduly conservative earnings reports and negatively to overly optimistic earnings reports. We also find that the presence of a Big 4 auditor is associated with a higher incidence of confirmation and switching to a Big 4 auditor before filing increases the incidence of emergence.