Analysis of a Bankruptcy for a Firm - Essay Example

Analysis of a Bankruptcy for a Firm Introduction: Bankruptcy for a firm is its inability to meet its contractual commitments (Damodaran A., 2008). Good corporate finance recommends a certain amount of debt in its capital structure. The two most important factors in a bankruptcy are the extent by which a firm’s debt exceeds its assets, and its inability to pay its debt as it comes due. Even firms with no debt can go bankrupt if they are unable to pay salaries to their employees, or they do not have enough working capital for their operations. Bankruptcy of a company can have far reaching implications for its creditors, suppliers, Government, and all other stakeholders related to it. As a result of this it becomes imperative to predict if any firm has chances of going bankrupt. Accurate forecasting of bankruptcy enables a firm to take corrective actions, and thereby reducing losses, and possibly even prevent bankruptcy. Hence, bankruptcy prediction is a topic of great interest, and attracting a lot of research. Bankruptcy Prediction Models: There are two kinds of bankruptcy prediction models, generally found in literature. The first is the accounting based models, which include logistic regression models. The second category is the market based model, which include Merton Models. Fitzpatrick (1931) used the approach of using ratio analysis to compare bankrupt and successful firms. His univariate model of using 13 ratios to indicate failure was first attempt of such kind to predict industry failures. However, no significant relationship could be established between the model and failure. The work done by Beaver (1966) is considered as the first pioneering work in the field of bankruptcy prediction. He proposed that the firm can be seen as a “reservoir of liquid assets, which is supplied by inflows and drained by outflows. (…) The solvency of the firm can be defined in terms of probability that the reservoir will be exhausted, at which point, the firm will be unable to pay its obligations as they mature”. Beaver used 30 ratios to develop a univariate model. These ratios were applied on 158 companies, half of them as bankrupt and the other half as successful firms. The finest ratios were the “working capital funds flow/ total assets”, and “net income/ total assets”. These ratios correctly identified 90%, and 88% of the cases. This study was followed by Altman’s model (1968, 1983). He applied multiple discriminant analysis to 33 pairs of bankrupt and successful firms. He proposed that bankruptcy could be explained by using a combination of 5 financial ratios. His proposed Z-score is still treated as an indicator of a company’s financial performance. Altman (1993) revised his earlier model, to propose a “four variable Z-score”. These statistical methods however, suffer from flaws like the assumptions of linearity, normality, and independent among input ratios. Ohlson (1980) is considered the first attempt to apply Multiple Logistic Regression (Logit) on the problem of bankruptcy prediction. He used a representative sample of 105 bankrupt and 2058 non-bankrupt firms. He claimed that his study had an important advantage that previous studies did not explicitly considered the timing issue. Recently, Artificial Intelligence approaches have become more popular for Bankruptcy prediction. These techniques include inductive learning, and Neural Networks. The suggestion of using Neural Networks as a complementary tool to classify credit risk was first suggested by Barker (1990), and Marose (1990). Neural Network model have the ability to incorporate a very large number of features in an adaptive nonlinear model. Later, other authors (Wilson and Sharda, 1994; Yang, 1999) found ANNs as a promising technique for bankruptcy prediction. O’Leary (1998) after his study of application of neural networks to corporate failure found that the results were at least as good as those generated by other techniques. For a neural network to give accurate results, it is imperative, that it is provided with good-quality examples with error-free data. Inductive learning is an Artificial learning technique that is based on learning from examples. Here, a system extracts knowledge from samples, and generate a general rule. This technique uses algorithms such as the Classification and Regression Trees (CART) to generate a tree type structure. Inductive learning is free from parametric and structural assumptions that underline statistical methods, and hence it is better than probit or logit analysis (Shaw and Gentry, 1990) Rule induction Algorithm classifies objects into specific groups on the basis of observed characteristics ratios (Meesier and Hansen, 1998). They developed an algorithm for bankruptcy prediction rule which worked for 87.5% of the holdout data set. When compared to a linear discriminant model, a logit model, the ID3 algorithm, and the k-Nearest Neighbor approach, NN models’ performance showed more precise results (Tam and Kaing, 1992). The duo compared 59 failed and 59 non-failed banks data from 1985 to 1987. Fletcher and Goss (1993) compared NN model’s performance to a logit regression model, and found that although NN models had higher prediction rates, the training model required for building NN models was also higher. Support Vector Machines (SVMs) has also been introduced as a technique for bankruptcy prediction. When compared to NN, MDA, and Learning vector quantization, SVM obtained the best results (Fan & Palaniswami, 200). SVM obtained the best results (70.35–70.90% accuracy depending on the number of inputs used), followed by NN (66.11–68.33%), followed by LVQ (62.50–63.33%), and followed by MDA (multivariate discriminant analysis) (59.79–63.68%). Conclusion Bankruptcy prediction of a firm is a very step for its creditors, suppliers, banks, government and other stakeholders. With a downward trend in the economy, and many companies filling for bankruptcy, the research for bankruptcy prediction has seen an increased interest from the academia as well as organizations alike. The research for bankruptcy prediction started with use of univariate models using simple financial ratios. It has slowly evolved over the last 15 years, and many new algorithms, based on Artificial intelligence have been proposed. Some of the recent researches have shown the superiority of methods like NN, SVM, LVQ, and MDA over earlier ones. These methods can be used to predict bankruptcy with 1-2 years, and in some cases eve 5 years. The only limitation these methods have is that they require good accurate data to draw out accurate assumptions. With more advanced intelligent systems developing, it is apparent, that new research will enable us to predict bankruptcy accurately in the times to come. References: Altman, E.I. (1968). Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy. The Journal of Finance, 23(4):589-609. Barker, D. (1990) Analyzing financial health: integrating neural networks and expert systems, PC AI, 4(3), pp. 24–27. Beaver, W.H. (1966). Financial Ratios as Predictors of Failure. Journal of Accounting Research, Vol. 4, Empirical Research in Accounting: Selected Studies 1966 (1966), pp. 71-111. Blackwell Publishing Damodaran, A. (2008). Corporate Finance: Theory and Practice. (Second Edition, pp 542). New Delhi: Wiley India. Fan, A., Palaniswami, M. (2000). A new approach to corporate loan default prediction from financial statements. Proceedings of the computational finance/forecasting financial markets conference, London (CD), UK. Fitzpatrick, P.J. (1931). Symptoms of industrial failures. Catholic University of American Press. Fletcher, D., & Goss, E. (1993). Forecasting with neural networks: an application using bankruptcy data. Information and Management, 24(3), 159–167. Marose, R. A. (1990) A financial network application, AI Expert, May, pp. 50–53. Messier, W., & Hansen, J. (1998). Inducing rules for expert system development: an example using default and bankruptcy data. Management Science, 34(12), 1403–1415. O’Leary, D. (1998) Using neural networks to predict corporate failure, International Journal of Intelligent Systems in Accounting, Finance & Management, (7), pp. 187–197. Ohlson and James, A. (1980). Financial ratios and the probabilistic prediction of bankruptcy. Journal of Accounting Research, 18(1):109-131 Shaw, M., & Gentry, J. (1990). Inductive learning for risk classification. IEEE Expert. 47–53. Stickney, C. (1990). Financial Statement Analysis: A Strategic Perspective. San Diego, CA: Harcourt Brace Jovanovich Publishers. Tam, K., & Kiang, M. (1992). Managerial applications of neural networks: the case of bank failure predictions. Management Science, 38(7), 926–947. Wilson, R. L. and Sharda, R. (1994) Bankruptcy prediction using neural networks, Decision Support Systems, 11, pp. 545–557 Yang, Z. R. (1999) Probabilistic Neural Networks in bankruptcy prediction, Journal of Business Research, 44, pp. 67–75. Read More