Assessing the financial health of organizations remains a topic of great interest to economists, financial institutions, and invested stakeholders. For more than a century, research into financial distress has focused primarily on traditional applications of statistical analysis; however, modern advances in computational efficiency have created a significant opportunity for more sophisticated approaches. This thesis investigates the application of artificial intelligence on company bankruptcy prediction. The proposed neural network model is evaluated using the Polish Companies Bankruptcy dataset and yields a 5-year prediction accuracy of 96.5% and an AUC (area under receiver operating characteristic curve) measure of 92.4%.

[pt] O objetivo desta pesquisa é propor um modelo estatístico que possa estimar a probabilidade de ocorrência de falências ou concordatas em empresas de Internet. Após as recentes e drásticas perdas de capital em investimentos nas empresas desta nova indústria, instituições financeiras, pessoas físicas e todos os investidores desejam ter o conhecimento da real situação financeira das empresas denominadas pontocom. Esta pesquisa selecionou empresas norte-americanas que pediram falência ou concordata nas Cortes Norte-Americanas de Falências,entre 1999 e 2001, e empresas que não o fizeram, por amostragem de conveniência, que possuem ações listadas em bolsa e operam no e-commerce, isto é que vendem seus produtos ou serviços através da Internet. Utilizou, ainda, as demonstrações financeiras destas empresas para identificar, por intermédio de um teste T de amostras independentes, as variáveis mais significantes na discriminação dos dois grupos de empresas observados na amostra: o de empresas falidas e o de não-falidas. Analisadas as distribuições estatísticas das variáveis,o modelo de regressão logística demonstra ser o mais apropriado à pesquisa, por não possuir a premissa de normalidade multivariada. A conclusão final da pesquisa é a proposição de um modelo estatístico que indica a probabilidade de uma empresa de Internet falir ou não, com índice R2 de Nagelkerke de 0,887, percentual máximo de acerto na classificação de 97,4 por cento e que utiliza ainda não utilizadas em pesquisas anteriores similares. / [en] The objective of this research is to propose a statistical model that could estimate the probability of occurrence of bankruptcy for Internet companies. After the recent and drastic losses of investment capital in companies in this new sector of the economy, financial institutions, individuals and all investors wish to know the real financial position of these companies called dotcom. This research selected American companies that have filed a petition under the United States Bankruptcy Code, between 1999 and 2001, and companies which have not done it, by convenience sampling, that list their shares on stock markets and operate in e-commerce, i.e. companies that sell their products or services through the Internet. The financial statements of these companies were also used to identify,by analyzing a T test of independent samples, the most significant variables for discriminating the two observed groups in the sample: the bankrupt and the nonbankrupt companies. After analyzing the variables statistical distributions, a logistic regression model revealed to be the more appropriate for the research, for not having the multivariate normality assumption. The conclusion of this research proposes a statistical model which indicates the probability of an Internet company becoming bankrupt or not, with a Nagelkerke R Squared of 0,887, and an overall percentage of correct prediction of 97,4 percent. The model uses several variables not previously included in similar previous financial difficulties prediction models.

A causa delle conseguenze che il fenomeno comporta, sia sul piano finanziario sia sul fronte dell'economia reale, l'analisi e la previsione delle insolvenze societarie continua a rappresentare un argomento attuale nell'ambito della ricerca economica. I recenti sforzi condotti dal Comitato di Basilea verso la diffusione di criteri di valutazione del rischio di credito più precisi ed oggettivi, hanno ulteriormente accresciuto l'importanza della materia. L'obiettivo del presente studio è l'analisi del fenomeno del fallimento sul territorio italiano, al fine di valutare quali variabili sono più efficaci nell'individuazione di una situazione di dissesto dell'impresa. Per l'analisi si sono sviluppati dei modelli di previsione delle insolvenze in grado di individuare i segnali early warning di dissesto finanziario. L'analisi econometrica è basata su un campione ampio ed originale di fallimenti rilevati negli anni 2003 e 2004: a tal fine sono stati costituiti dei campioni comparabili di imprese fallite e non fallite ed è stato verificato, mediante l'applicazione di una metodologia logit, il potere previsivo di diversi indici di bilancio e di variabili di tipo non finanziario. I risultati ottenuti sono stati validati su un campione hold-out. L'analisi si evidenzia l'importanza delle caratteristiche del settore di attività nel determinare la forma del processo di fallimento: i modelli sector specific ottengono risultati migliori rispetto ai modelli generali stimati. Inoltre, alcuni fattori comuni ai diversi settori di attività si dimostrano particolarmente efficaci nella previsione dei dissesti aziendali: l'età, il livello di leverage e la composizione del debito d'impresa, così come la sua redditività. / Due to the consequences that the phenomenon entails both on the financial and real sides of the economy, the analysis and prediction of corporate failures continue to be a current topic in economic research. The recent efforts laid by the Basel Committee towards the diffusion of more precise and objective ways of assessing credit risk have further increased the importance of this matter. The purpose of the study is to analyse the bankruptcy phenomenon among Italian firms, in order to assess what firm-specific and industry variables are more important in determining corporate failure events. We develop a bankruptcy prediction model that aims at detecting early signals of financial distress. The econometric analysis is based on a wide and unique sample of recent failure events: comparable sets of bankrupt and non-bankrupt firms are identified and several prior balance-sheet and economic indicators are tested for their power in predicting failure probabilities in a logit modelling framework; model performances are cross-validated on hold-out samples. The analyses provide evidence of the importance of industry membership in determining and shaping corporate failure processes: sector-specific models produce a better assessment of financial distress than general ones. Also, some common factors emerge as important predictors of corporate collapse across different industries: age, gearing and the composition of a firm's debt, as well as its capability of generating profits.

Bakgrund: Många företag går i konkurs varje år vilket är förknippade med kostnader för de enskilda intressenterna och för samhället i stort. För att kunna vidta eventuella åtgärder innan konkursen är ett faktum är det av intresse att veta om någon av de modeller som forskare tagit fram för att förutspå konkurser faktiskt fungerar. Syfte: Syftet med denna undersökning är att ta reda på om det går att applicera någon av ett urval av etablerade konkursmodeller på svenska industri- och tillverkningsföretag. Teori: Studien kommer att testa tre olika forskares modeller och metoder: Altmans, Platts och Platts samt Pompes och Bilderbeeks. Metod: I denna studie kommer enbart en deduktiv forskningsansats att användas och datainsamlingen är kvantitativ då nyckeltal hämtas från de aktuella företagens årsredovisningar. Urvalet baseras på de företag som ansökte om konkurs under år 2011 och de som representerar kontrollgruppen har slumpmässigt valts ut bland de företag inom avgränsningen som inte gått i konkurs det aktuella året. Resultat och slutsats: Altmans och Platts och Platts modeller visar sig inte vara applicerbara på svenska företag. Dock är vissa av Pompes och Bilderbeeks nyckeltal tillämpliga till att använda för konkursprognostisering för svenska företag. / Background: Companies are going bankrupt every year which is associated with costs for individual parties with interests in the company and for society in general. To be able to take any action before bankruptcy is a fact, it is interesting to know if any of the models that scientists developed to predict bankruptcies actually works. Purpose: The purpose of this study is to determine whether it is possible to apply a selection of the established bankruptcy models on Swedish manufacturing companies. Theory: The study will test three different researchers' models and methods: Altman's, Platt's and Platt's, as well as Pompe's and Bilderbeek's. Methodology: In this study, only a deductive research approach will be used and the data collection is quantitative since the ratios are obtained from the relevant companies' financial statements. The selection is based on the companies that filed for bankruptcy in 2011 and the firms which represent the control group were selected at random among the companies within the delimitation that didn't go bankrupt during the current year. Result and conclusion: Altmans and Platts and Platts models turn out not to be applicable on Swedish companies. Some of Pompes and Bilderbeeks ratios are relevant for use in bankruptcy prediction for Swedish companies though.

Prediction of corporate bankruptcy (or distress) is one of the major activities in auditing firms’ risks and uncertainties. The design of reliable models to predict distress is crucial for many decision-making processes. Although a variety of models have been designed to predict distress, the relative performance evaluation of competing prediction models remains an exercise that is unidimensional in nature. To be more specific, although some studies use several performance criteria and their measures to assess the relative performance of distress prediction models, the assessment exercise of competing prediction models is restricted to their ranking by a single measure of a single criterion at a time, which leads to reporting conflicting results. The first essay of this research overcomes this methodological issue by proposing an orientation-free super-efficiency Data Envelopment Analysis (DEA) model as a multi-criteria assessment framework. Furthermore, the study performs an exhaustive comparative analysis of the most popular bankruptcy modelling frameworks for UK data. Also, it addresses two important research questions; namely, do some modelling frameworks perform better than others by design? and to what extent the choice and/or the design of explanatory variables and their nature affect the performance of modelling frameworks? Further, using different static and dynamic statistical frameworks, this chapter proposes new Failure Prediction Models (FPMs). However, within a super-efficiency DEA framework, the reference benchmark changes from one prediction model evaluation to another one, which in some contexts might be viewed as “unfair” benchmarking. The second essay overcomes this issue by proposing a Slacks-Based Measure Context-Dependent DEA (SBM-CDEA) framework to evaluate the competing Distress Prediction Models (DPMs). Moreover, it performs an exhaustive comparative analysis of the most popular corporate distress prediction frameworks under both a single criterion and multiple criteria using data of UK firms listed on London Stock Exchange (LSE). Further, this chapter proposes new DPMs using different static and dynamic statistical frameworks. Another shortcoming of the existing studies on performance evaluation lies in the use of static frameworks to compare the performance of DPMs. The third essay overcomes this methodological issue by suggesting a dynamic multi-criteria performance assessment framework, namely, Malmquist SBM-DEA, which by design, can monitor the performance of competing prediction models over time. Further, this study proposes new static and dynamic distress prediction models. Also, the study addresses several research questions as follows; what is the effect of information on the performance of DPMs? How the out-of-sample performance of dynamic DPMs compares to the out-of-sample performance of static ones? What is the effect of the length of training sample on the performance of static and dynamic models? Which models perform better in forecasting distress during the years with Higher Distress Rate (HDR)? On feature selection, studies have used different types of information including accounting, market, macroeconomic variables and the management efficiency scores as predictors. The recently applied techniques to take into account the management efficiency of firms are two-stage models. The two-stage DPMs incorporate multiple inputs and outputs to estimate the efficiency measure of a corporation relative to the most efficient ones, in the first stage, and use the efficiency score as a predictor in the second stage. The survey of the literature reveals that most of the existing studies failed to have a comprehensive comparison between two-stage DPMs. Moreover, the choice of inputs and outputs for DEA models that estimate the efficiency measures of a company has been restricted to accounting variables and features of the company. The fourth essay adds to the current literature of two-stage DPMs in several respects. First, the study proposes to consider the decomposition of Slack-Based Measure (SBM) of efficiency into Pure Technical Efficiency (PTE), Scale Efficiency (SE), and Mix Efficiency (ME), to analyse how each of these measures individually contributes to developing distress prediction models. Second, in addition to the conventional approach of using accounting variables as inputs and outputs of DEA models to estimate the measure of management efficiency, this study uses market information variables to calculate the measure of the market efficiency of companies. Third, this research provides a comprehensive analysis of two-stage DPMs through applying different DEA models at the first stage – e.g., input-oriented vs. output oriented, radial vs. non-radial, static vs. dynamic, to compute the measures of management efficiency and market efficiency of companies; and also using dynamic and static classifier frameworks at the second stage to design new distress prediction models.

Submitted by Roberto Shanrey Chieh (robertochieh@gmail.com) on 2018-08-24T13:44:12Z No. of bitstreams: 1 Dissertação MPE - Roberto Chieh - v23.pdf: 961097 bytes, checksum: 37ad879de70e42544f67413eccdf0401 (MD5) / Rejected by Joana Martorini (joana.martorini@fgv.br), reason: Prezado Roberto, boa tarde. Por gentileza a justar os seguintes itens: . Retirar o acento da palavra "Getúlio" . Na folha de assinaturas, retirar a frase "Folha de Aprovação" Após a realização dos ajustes, peço que faça uma nova submissão. Att, Joana Martorini on 2018-08-24T17:41:38Z (GMT) / Submitted by Roberto Shanrey Chieh (robertochieh@gmail.com) on 2018-08-24T17:49:02Z No. of bitstreams: 1 Dissertação MPE - Roberto Chieh - v24.pdf: 961253 bytes, checksum: 1b330f244eb5945c4c15d9d2165215e2 (MD5) / Approved for entry into archive by Joana Martorini (joana.martorini@fgv.br) on 2018-08-24T17:57:16Z (GMT) No. of bitstreams: 1 Dissertação MPE - Roberto Chieh - v24.pdf: 961253 bytes, checksum: 1b330f244eb5945c4c15d9d2165215e2 (MD5) / Approved for entry into archive by Isabele Garcia (isabele.garcia@fgv.br) on 2018-08-27T13:48:01Z (GMT) No. of bitstreams: 1 Dissertação MPE - Roberto Chieh - v24.pdf: 961253 bytes, checksum: 1b330f244eb5945c4c15d9d2165215e2 (MD5) / Made available in DSpace on 2018-08-27T13:48:01Z (GMT). No. of bitstreams: 1 Dissertação MPE - Roberto Chieh - v24.pdf: 961253 bytes, checksum: 1b330f244eb5945c4c15d9d2165215e2 (MD5) Previous issue date: 2018-07-31 / O objetivo dessa dissertação é identificar os melhores modelos para prever falência de empresas dos setores de mineração, óleo e gás no período entre 1998 e 2017. Em termos metodológicos, buscou-se estimar um modelo de regressão logística para prever as falências das empresas por meio de indicadores financeiros. Estimam-se modelos com dados anuais e trimestrais utilizando informações dos últimos três, dois e um ano anteriores às falências, contados a partir de um ano antes da formalização da falência. Conclui-se que o melhor modelo é aquele que utiliza as informações mais recentes, do último ano, e com dados trimestrais. As variáveis de patrimônio líquido sobre passivo total e fluxo de caixa de investimentos sobre passivo total se destacaram dentre os demais indicadores, sendo somente a primeira significativa em todos os modelos. O melhor modelo teve 79,1% de acerto geral e 85,5% de acerto para as empresas que faliram. / The objective of this study is to identify the best models for predicting bankruptcy of companies from the mining, oil and gas industries between 1998 and 2017. It was estimated a logistic regression model to predict business failure given their financial indicators. It was estimated models with yearly and quarterly information figures using figures from the last three years, last two years, and also last one year prior to the year just before the formalization of the bankruptcy event. The results show that the best model is the one using the most recent information, from the last one year, and using quarterly available data. The ratios total equity to total liabilities and cash flow from investments to total liabilities are the most important indicators to predict bankruptcy, even though only the first one is significant in all models. The best model correctly predicted 79.1% among all firms and 85.5% of the firms that went bankrupt.

The aim of this study is to evaluate feasibility of using Bayes classifiers for predicting corporate bankruptcies. The results obtain show that Bayes classifiers do reach comparable results to then more commonly used methods such the logistic regression and the decision trees. The comparison has been carried out based on Czech and Polish data sets. The overall accuracy rate of these so called naive Bayes classifiers, using entropic discretization along with the hybrid pre-selection of the explanatory attributes, reaches 77.19 % for the Czech dataset and 79.76 % for the Polish set respectively. The AUC values for these data sets are 0.81 and 0.87. The results obtained for the Polish data set have been compared to the already published articles by Tsai (2009) and Wang et al. (2014) who applied different classification algorithms. The method proposed in my study, when compared to the above earlier works, comes out as quite successful. The thesis also includes comparing various approaches as regards the discretisation of numerical attributes and selecting the relevant explanatory attributes. These are the key issues for increasing performance of the naive Bayes classifiers

Research question: This thesis analyzes the relationship between Altman’s Z’-and Z’’ score model in order to investigate the suitability of the models on the swedish small and medium companies during 2015-2019. Furthermore, analysis of previous research key figures has been examined in more detail to identify which of the individual key figures can be categorized as an early warning signal. Purpose: The purpose with this study is to explore which and when early warning signals can be read in annual reports between inactive and active Swedish tourist companies and to investigate the relationship between the accuracy of bankruptcy prediction models between active and bankrupt companies. Method: The study uses a quantitative method with a deductive approach, z-test and onesided analysis of variance ANOVA to analyze the accuracy of bankruptcy prediction models and identify how the key figures differ between active and inactive companies. Conclusion: The study shows that Altman's Z 'and Z' scores predict bankruptcies better than specify continued operations, are best suited for active companies and have the highest accuracy one year in advance than a longer period. On the other hand, 7 out of 15 key figures examined have identified significant average value differences between active and bankrupt companies, where some had a higher value for bankrupt companies and others had lower ones. / Problemställning: I denna studien har relationssamband mellan Altmans Z’- och Z’’- scoremodell analyserats för att undersöka hur modellerna lämpar sig på den svenska små- och medelstora konkur-respektive aktiva turistföretag mellan 2015-2019. Ytterligare har analys av tidigare forsknings nyckeltal undersökts närmare för att identifiera vilka av de enskilda nyckeltalen kan kategoriseras som en tidig varningssignal. Syfte: Studiens avsikt är att utforska vilka och när tidiga varningssignaler kan utläsas i årsredovisningar mellan inaktiva och aktiva svenska turistföretag samt undersöka relationssambandet gällande konkursprediktionsmodellers träffsäkerhet mellan aktiva- och konkursföretag. Metod: I studien används en kvantitativ metod med en deduktiv ansats, z-test och ensidig variansanalys ANOVA för att analysera konkursprediktionsmodellers träffsäkerhet samt identifiera hur nyckeltalen skiljer sig mellan aktiva och konkursföretag. Slutsats: Studien visar att Altmans Z”och Z’-score förutser konkurser bättre än preciserar fortsatt verksamhet, lämpar sig bäst på aktiva företag samt har den högsta träffsäkerhet ett år i förväg än längre period. Däremot har 7 av 15 undersökta nyckeltal identifierat signifikanta medelvärdesskillnader mellan aktiva och konkursföretag där några hade ett högre värde gällande konkursföretag och andra hade lägre.

This diploma thesis deals with the issue of bankruptcy prediction of small and medium-sized enterprises operating in the manufacturing industry in selected Central European countries. The theoretical part of the thesis defines the concepts related to the prediction of bankruptcy and methods of model creation. The analytical part of the work includes testing the accuracy of selected bankruptcy model by other authors and creating a new bankruptcy model. The accuracy of the newly created model is then compared with the accuracy of selected models by other authors.

This thesis deals with evaluating the financial situation of the company LIPOELASTIC a.s. through deployment of specific tools of financial analysis. The data necessary for financial analysis comes from financial statements of the company. Structure of the thesis is divided onto three parts. Theoretical part contains a set of crucial terms, methods and workflows which are important for designing the individual parts of financial analysis. These methods are practically applied in the practical part. Last part includes the proposals and measures for enhancing financial situation of the company.