Predicting Corporate Bankruptcy: An Evaluation of Alternative Statistical Frameworks

Corporate bankruptcy prediction has attracted significant research attention from business academics, regulators and financial economists over the past five decades. However, much of this literature has relied on quite simplistic classifiers such as logistic regression and linear discriminant analysis (LDA). Based on a large sample of US corporate bankruptcies, we examine the predictive performance of 16 classifiers, ranging from the most restrictive classifiers (such as logit, probit and linear discriminant analysis) to more advanced techniques such as neural networks, support vector machines (SVMs) and “new age” statistical learning models including generalised boosting, AdaBoost and random forests. Consistent with the findings of Jones et al. ( 2015 ), we show that quite simple classifiers such as logit and LDA perform reasonably well in bankruptcy prediction. However, we recommend the use of “new age” classifiers in corporate bankruptcy modelling because: (1) they predict significantly better than all other classifiers on both the cross‐sectional and longitudinal test samples; (2) the models may have considerable practical appeal because they are relatively easy to estimate and implement (for instance, they require minimal researcher intervention for data preparation, variable selection and model architecture specification); and (3) while the underlying model structures can be very complex, we demonstrate that “new age” classifiers have a reasonably good level of interpretability through such metrics as relative variable importances (RVIs).     

Entropic approximate learning for financial decision-making in the small data regime

Financial decision-making problems based on relatively few observations and several explanatory variables can be problematic for the common machine learning (ML) tools, since they cannot efficiently discriminate the relevant information. To investigate the challenges of this “small data” regime, we employ several state-of-the-art ML methods for predicting whether three selected stocks from the Swiss Market Index will outperform the market, by using, as classification features, a set of commonly used technical indicators. We show that the recently introduced entropic Scalable Probabilistic Approximation (eSPA) algorithm significantly surpasses its competitors in both prediction accuracy and computational cost. We then discuss the interpretability of the employed ML methods and suggest some statistically derived heuristics to select the most appropriate and parsimonious financial decision-making candidate model.     

Features selection,data mining and finacial risk classification: a comparative study

The aim of this paper is to compare several predictive models that combine features selection techniques with data mining classifiers in the context of credit risk assessment in terms of accuracy, sensitivity and specificity statistics. The t‐statistic, Battacharrayia statistic, the area between the receiver operating characteristic, Wilcoxon statistic, relative entropy, and genetic algorithms were used for the features selection task. The selected features are used to train the support vector machine (SVM) classifier, backpropagation neural network, radial basis function neural network, linear discriminant analysis and naive Bayes classifier. Results from three datasets using a 10‐fold cross‐validation technique showed that the SVM provides the best accuracy under all features selections techniques adopted in the study for all three datasets. Therefore, the SVM is an attractive classifier to be used in real applications for bankruptcy prediction in corporate finance and financial risk management in financial institutions. In addition, we found that our best results are superior to earlier studies on the same datasets.     

Machine learning for financial transaction classification across companies using character-level word embeddings of text fields

An important initial step in accounting is mapping financial transfers to the corresponding accounts. We devised machine-learning-based systems that automate this process. They use word embeddings with character-level features to process transaction texts. When considering 473 companies independently, our approach achieved an average top-1 accuracy of 80.50%, outperforming baselines that exclude the transaction texts or rely on a lexical bag-of-words text representation. We extended the approach to generalizes across companies and even across different corporate sectors. After standardization of the account structures and careful feature engineering, a single classifier trained on 44 companies from 28 sectors achieved a test accuracy of more than 80%. When trained on 43 companies and tested on the remaining one, the system achieved an average performance of 64.62%. This rate increased to nearly 70% when considering only the largest sector.     

Universal features of price formation in financial markets: perspectives from deep learning

Using a large-scale Deep Learning approach applied to a high-frequency database containing billions of market quotes and transactions for US equities, we uncover nonparametric evidence for the existence of a universal and stationary relation between order flow history and the direction of price moves. The universal price formation model exhibits a remarkably stable out-of-sample accuracy across a wide range of stocks and time periods. Interestingly, these results also hold for stocks which are not part of the training sample, showing that the relations captured by the model are universal and not asset-specific.

The universal model—trained on data from all stocks—outperforms asset-specific models trained on time series of any given stock. This weighs in favor of pooling together financial data from various stocks, rather than designing asset- or sector-specific models, as is currently commonly done. Standard data normalizations based on volatility, price level or average spread, or partitioning the training data into sectors or categories such as large/small tick stocks, do not improve training results. On the other hand, inclusion of price and order flow history over many past observations improves forecast accuracy, indicating that there is path-dependence in price dynamics.     

Spillovers in risk of financial institutions

ABSTRACT

We analyse the total and directional spillovers across a set of financial institution systemic risk state variables: credit risk, real estate market risk, interest rate risk, interbank liquidity risk and overall market risk. We examine the response of the spillover levels, within the set of systemic risk state variables, to a number of events in the financial markets and to initiatives undertaken by the European Central Bank and the Bank of England. The relationship between the time-varying spillovers and policy-related events is analysed using a multiple structural break estimation procedure and looking at the temporary increases in the spillover indices. Our sample includes five European Union countries: core countries France and Germany, periphery countries Spain and Italy, and a reference country, the UK. We show that national stock markets and real estate markets have a leading role in shock transmission across selected state variables. However, the role of the other variables reverses over the course of the crisis. We document that the total and net spillover indices react strongly to the events relating to financial assistance packages in Europe.     

大数据应用于信用评分模型的实践与启示

传统征信业务必因大数据而发生改变,大数据将为现有征信体系增加海量数据来源并推动普惠金融的发展。但是,由于存在个人隐私权保护、信贷风险控制及管理等限制因素,大数据技术最终如何实现与征信业务的完美结合以及究竟对传统征信业带来何种程度的影响,仍需要时间的检验。     

Market‐based estimates of implicit government guarantees in European financial institutions

I exploit the price differential of credit default swap (CDS) contracts written on debts with different levels of seniority to measure the implicit government guarantees enjoyed by European financial institutions from 2005 to 2013. I determine that the aggregate guarantee increased substantially during the recent financial crises and peaked at an average of 89 bps in 2011. My analysis suggests that the extent of implicit support depends on the type of financial institutions and there exists a eurozone effect. Further investigation of feedback relationship shows that the guarantee implicitly offered by a government positively ‘Granger causes’ the sovereign's default risk.     

Dynamic prediction of hedge fund survival in crisis-prone financial markets

This study focuses on dynamic changes in survival probabilities over the lifetimes of hedge funds. To model such probabilities, a mixed Cox proportional hazards (CPH) model-specifically, a survival/hazard model with time-varying covariates and fixed covariates- is employed. Resulting dynamic survival probabilities show that the mixed CPH model provides significantly higher accuracy in predicting hedge fund failure than other models in the literature, including fixed covariate CPH models and discrete logit models. Our results are useful to investors and regulators of hedge funds in crisis-prone financial markets.     

Teaching auditing using cases in an online learning environment: the role of ePortfolio assessment

While teaching auditing using cases is regarded as an effective approach, spatial separation of students and teachers in online contexts can restrict the application of case teaching. This study examines an undergraduate auditing course implemented to address this challenge by integrating case teaching with ePortfolio assessment. Students’ written ePortfolio submissions and scores were analysed. Results show that despite spatial separation of the online learner from peers and teachers, integrating case teaching with ePortfolio assessment elicits learner behaviour desirable in online auditing courses. This approach enables online learners’ self-directed engagement as compared to instructor-led case teaching in conventional teaching contexts. Based on a new pedagogical approach for teaching auditing trialled in reduced (or absence of) face-to-face interaction, this study informs course design in auditing. It demonstrates that active student engagement, which presupposes an instructor’s role to facilitate student involvement in case discussions, can be implemented in online teaching of auditing.     

The role of attribute selection in Deep ANNs learning framework for high‐frequency financial trading

In financial trading, technical and quantitative analysis tools are used for the development of decision support systems. Although these traditional tools are useful, new techniques in the field of machine learning have been developed for time‐series forecasting. This paper analyses the role of attribute selection on the development of a simple deep‐learning ANN (D‐ANN) multi‐agent framework to accomplish a profitable trading strategy in the course of a series of trading simulations in the foreign exchange market. The paper evaluates the performance of the D‐ANN multi‐agent framework over different time spans of high‐frequency (HF) intraday asset time‐series data and determines how a set of the framework attributes produces effective forecasting for profitable trading. The paper shows the existence of predictable short‐term price trends in the market time series, and an understanding of the probability of price movements may be useful to HF traders. The results of this paper can be used to further develop financial decision‐support systems and autonomous trading strategies for the financial market.     

Explaining the (non-) adoption of advanced data analytics in auditing: A process theory

Audit firms are increasingly engaging with advanced data analytics to improve the efficiency and effectiveness of external audits through the automation of audit work and obtaining a better understanding of the client’s business risk and thus their own audit risk. This paper examines the process by which audit firms adopt advanced data analytics, which has been left unaddressed by previous research. We derive a process theory from expert interviews which describes the activities within the process and the organizational units involved. It further describes how the adoption process is affected by technological, organizational and environmental contextual factors. Our work contributes to the extent body of research on technology adoption in auditing by using a previously unused theoretical perspective, and contextualizing known factors of technology adoption. The findings presented in this paper emphasize the importance of technological capabilities of audit firms for the adoption of advanced data analytics; technological capabilities within audit teams can be leveraged to support both the ideation of possible use cases for advanced data analytics, as well as the diffusion of solutions into practice.     

Archival data of financial analysts' earnings forecasts in the euro zone: Problems with euro conversions

In multi-country studies, researchers frequently extract data in a single currency rather than in native currencies. This approach can be misleading for financial analysts’ forecasts in the euro zone when researchers are using the IBES database. We suspect that forecasts of earnings before the birth of the euro on January 1, 1999 are kept in national currencies, although they are supposed to be displayed in euros, which can severely distort results concerning earnings forecast accuracy. We propose a simple procedure for checking for the existence of this error, as well as a quick solution to overcome it.     

Properties of Accounting and Finance Information and Their Effects on the Performance of Bankers and Models in Predicting Company Failure

When the number of cues provided to a banker for a decision is increased it may (1) increase their information load (number of relevant cues), (2) increase their data load (number of irrelevant cues), and (3) reduce their uncertainty. Models, on the other hand, are not affected by information or data load. The results from this research show that as the number of cues provided to bankers increases, uncertainty reduces, data load increases, but information load is unaffected. The uncertainty reduction increases the decision accuracy of both the bankers and models. Due to the data load experienced by the bankers but not the models, the models have superior performance. The implications for future practice and research are discussed.     

The roles of alternative data and machine learning in fintech lending: Evidence from the LendingClub consumer platform

There have been concerns about the use of alternative data sources by fintech lenders. We compare loans made by LendingClub and similar loans that were originated by banks. The correlations between the rating grades (assigned by LendingClub) and the borrowers’ FICO scores declined from about 80% (for loans originated in 2007) to about 35% for recent vintages (originated in 2014–2015), indicating that nontraditional data (not already accounted for in the FICO scores) have been increasingly used by fintech lenders. The rating grades perform well in predicting loan default. The use of alternative data has allowed some borrowers who would have been classified as subprime by traditional criteria to be slotted into “better” loan grades, allowing them to obtain lower priced credit.