Bankruptcy Prediction and Accounting Ratios - Research Paper Example

I am doing FTSE all share travel and leisure industry. Please see attached spreadsheet, I retrieved data from datastream advanced Define researchquestions and objectives based on bankruptcy prediction. 2. Write abstract. 3. Write Introduction. 4. Write Methodology sections. (please refer to references section below) 5. Define Hypotheses and test by using SPSS software to do cross-sectional anlaysis based on spreadsheet attached. (Ohlson model) 6. Conclusion Here is Ohlson model. Example: Y = -1.3 - .4 Y1 + 6.0 Y2 - 1.4 Y3 + .1 Y4 -2.4 Y5 - 1.8 Y6 + .3Y7 -1.7 Y8 - .5Y9 where: Y1 = log(total assets/GNP price-level index); Y2 = total liabilities/total assets; Y3 = working capital/ total assets; Y4 = current liabilities/current assets; Y5 = one if total liabilities exceed total assets, zero otherwise; Y6 = net income/ total assets; Y7 = funds provided by operations/total liabilities; Y8 = one if net income was negative for the last two years, zero otherwise; Y9 = measure of change in net income; Y = overall index. The literature on predictive models can be divided into three main catagories: 1. Accounting ratios only 2. Accounting ratios + market driven factors 3. other models not involving either 1 or 2 above For each catagory, there are several issues to consider: a. which time period/ Is the model sensitive to the time period chosen' b. what are the time interval used for accounting ratios and other factors' c. how is the tested being done' What is the methodology' d. Which countries are being considered' e. what is the predictive rate' 2.1. Methodologies using Accounting ratios 9 Acknowledgement Abstract Introduction In the fields of accounting and finance, researchers have studied bankruptcy since the work of Beaver (1966,1968) and Altman (1968). Previous studies have actively used either accounting ratios or market ratios based measures. Most recently, artificial intelligence models have been devised, leading to the development of so-called neural network techniques whereby computers are used to simulate thought processes and identify behavior patterns. "The implication is that UK analyst referring to one of the well known bankruptcy identification models would in any one year correctly classify the dozen or so listed industrial companies which will fail, but will incorrectly identify about 120 of the remaining 600 as likely to go bankrupt. In fact, analysts who might use the models t help them produce their credit ratings are likely to try them out before relying on them and making them self-fulfulling. It therefore seems unlikely that a misclassification error rate of 1 in 5 for surviving listed companies would be acceptable, even allowing for the substantially greater costs of incorrectly identifying a bankrupt company as sound when compared to those of misclassifying a surviving company as a prima facie failure." Richard Morris This paper examines whether accounting based measures effectively capture publicly available information about a firm's probability of bankruptcy. Section 2 Section 3 describes model and research methodology which includes details about the sample selection procedures, variable estimation and descriptive statistics are reported in section 4. Section 5 present and discuss the results, while Section 6 summarizes and concludes the paper. Also include a list of variables in Appendix A. 1. Literature Review 1.1 Accounting ratios Professor Edward Altman invented a model called Z-Scores by applying multivariate formula to forecast bankruptcy probabilities of the firms over 30 years from 1965-1999. In 2000, he extended his research throughout the year 1999 by improving accuracies of 96% one period prior to bankruptcy to 70% five annual reporting periods prior. Ohlson (1980) also developed a bankruptcy prediction model with logit analysis using a number of bankruptcy firms that were traded on NYSE and AMSE during the 1970s. Begley Joy et al (1997) critised the estimation models of Altman (1968) and Ohlson (1980) were not performed well by using 1980's data. Their empirical research found consistent hypotheses which are: not only debt usage increases Type II (misclassification a non-bankrupt firm as bankrupt) errors but also bankruptcy law changes in 1980's increase Type I (misclassification a bankrupt as non-bankrupt) error when original models are used. Their research concluded that Ohlson (1980)'s bankruptcy prediction model performed stronger than Altman (1968) model. Similarly, Philosophov, Vladimir L, Philosphov Leonid V (1999) argued that Altman (1982)'s linear discriminant analysis is not useful for bankruptcy prediction. They developed probabilistic approach to find optimal debt to equity ratio for capital structure and bankruptcy prediction in the future. They assessed the corporate share value which depends on optimal debt/ equity ratio affecting both returns and probabilities of bankruptcy. They recommend a joint use of theoretic models and empirical data could be appropriate. The comparison between Altman (1968), Ohlson (1980) models and three other Canadian prediction models made by Boritz, J et al (2007) indicate that Altman has the lowest performance than others. Their sample consists of 266 Canadian companies failing for the period 1987-2002 and also contains a matched sample of non failed companies. Gricea John S, Ingramb Robert W. (2001) claimed that Altman's Zscores model is not useful for predicting bankruptcy in recent time and manufacturing firms' predictions are more accurate than non-manufacturing firms. Interestingly, their findings are positive for predicting financial distress other than bankruptcy costs even it is originally intended for predicting bankruptcy. Mossman C E et al (1998) compared four kinds of bankruptcy models of financial statement ratios, cash flow, stock returns; return standard deviations based on 1980-1991 results. They recommend a new challenge towards full use of all available data in a better model for the bankruptcy process. Share prices fluctuations are in an obvious way to look at distressed listed companies claimed by Morris Richard (1998). However, Gilbert R L et al (1990)'s study show that there are overlapping financial characteristics of bankrupt and distressed firms, therefore other factors such as incentives of managers, stockholders, etc should be considered when decision made by distress firm. The estimation model by Korteweg, Arthur G (2007) used a panel dataset of monthly market values of debt and equity over 10 years. He then generalized empirical model for default probability, except leverage, such as Z-scores and credit ratings. The discussion of Haber Jeffry R, College Iona (2005) addresses new models should be developed to predict bankruptcy and apply practically. The evaluation process should be identified how the model classifies the category of go bankrupt soon and will bankrupt. The study of Ward Terry J (2007) suggests that various response variables used for financial distress are different measure and he discouraged a dichotomous bankruptcy measure as the poorest form of financial distress. Recently, empirical study of Hui Huang, Jing-Jing Zhao (2008) found significant relationship between corporate governance aspects and indirect financial distress costs listed companies in China. They suggest distress companies should improve corporate governance to enhance their financial circumstances. 1.2 Accounting ratios and Market-driven factors Black and Scholes (1973) model prove that risk-free interest rate is the correct discount factor and assumptions absence are based upon investor's risk preferences. In 2004, Hillegeist, Stephen A et al agreed the market based measure by Black & Scholes (1973), Merton (1974)'s credit risk assessment models are outperformed than z scores and O scores model. Interestingly, Taffler R, Agarwal V (2008) compares UK based z score Taffler (1984) model with market based models over 17 year period from 1985-2001. They used Hillegeist et al (2004) and Bharath and Shymway (2004) market based models which are based on Black and Scholes (1973) and Merton (1974). They concluded that accounting based approach provides economic benefit over market-based. Shumway T (2001) argued that previous single-period models give inconsistent estimates. Therefore he used combination of traditional accounting ratios along with market-driven variables to produce more accurate forecast than others. The research suggested joint use of theoretical models and empirical research is more appropriate than statistical data. By extending the work of Shumway (2001), Nam C W, et.al (2008) used a discrete-time model to investigate the hazard rates and consider time-varying covariate vectors in macroeconomic environment changes for the listed Korean Stock Exchange (KSE) companies. He Y et al (2005) predicted bankruptcy prediction of small firms by using Ohlson's (1980) and Shumway's (2001) models with the data traded on over the counter (OTC) market in the 1990s. Their results indicate that Shumway (2001) gives impressive prediction accuracy over one year prior to bankruptcy. 1.3 Other models Initially, empirical research with univariate analysis and considered the impact of bankruptcy on stock returns (Beaver, 1966). In order to achieve reliability, he then develops cross-validation test by splitting the sample into two subsamples in 1968. In the year 2006, Cochran, James et al used calendar-time model to investigate bankruptcy of dot com companies and identified key predictors of company failure. Additionally, they revealed two important findings that liquidity is more important than profit potential for one year prior to bankruptcy, but for three year prior, result is reversed. Pindado Julio et al also developed a new model called ex-ante to estimate the probability of financial distress and offered to use in different economic and legal issue. Their model provides accuracy of classification for US, UK and Germany listed companies and finds the probability of financial distress is related with firms' assets, trade-off between generating funds and financial expenses during the financial year. Haber R Jeffry, College Iona (2005) strongly recommended that bankruptcy models should predict more than 94% accuracy and new models have to be developed in the case of classifying companies eligible to file for bankruptcy, but not to. A new model should look at potential bankruptcy companies to file, if they are not, would be counted as errors. The investigation of forecasting bankruptcy hazard rate models for US companies over the period 1962-1999 from Chava S & Jarrow R A (2004) showed that using monthly observation intervals enabled to improve the bankruptcy prediction than yearly. They agreed Shumway's (2001) harzard model as opposed to Altman (1968) and Zmijewski (1984) single-period models. They also found consistent that market efficiency is one of the key determinant for bankruptcy model with respect to publicly available information by demonstrating when market variables are already took account in the bankruptcy model, accounting variables add less predictive power. The evidence of small and medium UK firms provided by Franks and Sussman (2005) concluded that direct costs of bankruptcy for small UK firms are relatively high, followed by the research of 542 small UK firms in bankruptcy and distress. Their paper makes contribution towards a contractualist approach to corporate bankruptcy and provides an operational description in UK bankruptcy process. The comparison analysis of cash flow based models (CFB) with Zeta and Z models made by Aziz Abdul et al (1988) had shown that overall accuracy is equal. But CFB is more likely to predict five year prior to bankruptcy than Z model. Similarly, CFB can give three year prior warnings before the event than Zeta. CFB model based not only on fixed set of variables according to theories but also build up the investigation towards more efficient models. 2. Data and Methodology The approach to examining bankruptcy of travel and leisure sector industry is influenced by prior research favorable for choosing potential predictors. Addressed by the discussion of Haber Jeffry R, College Iona (2005), new models should be developed to predict bankruptcy and apply practically. The evaluation process should be identified how the model classifies the category of go bankrupt soon and will bankrupt. In particular, Altman (1968) , Ohlson (1980) use pure accounting ratios in their models to predict failure. In contrast, Cochran, James et al (2006) used Cox PH model with time varying and cross-sectional data to investigate bankruptcy of dot com companies and identified key predictors which make few assumptions to predict company failure. Nevertheless, Grice J S & Dugan M T (2001) warned the researchers to know limitations of models and use those models carefully because they found out time intervals is important criteria for the accurateness of each model and empirical research should emphasize on financial distress, not just bankruptcy. 2.1. Methodologies using Accounting ratios Altman (2000) applied Multiple Discriminant Analyis (MDA) model which considers credit worthiness, corporate financial distress, investment risk portfolio for managers and investors; others include going-concern situations for internal and external aspect of US firms. In 1980, Logit model developed by Ohlson used a set of independent variables which were financial statement ratios and determined probability of bankruptcy by creating an overall score. Mossman C E et al (1998) indicates that ratio model offers the best single model in the year immediately prior to bankruptcy. Data Methodology The successful bankruptcy prediction models such as Ohlson (1980) and Shumway (2001) has developed to determine the probability of bankruptcy. Ohlson discovered a logit model based on a set of independent variables which were financial statement ratios. Data samples The sample data include 35 firms from UK FTSE All Share index companies. The necessary financial statement ratio samples obtain from Datastream, which is available from Birkbeck college library. Ohlson (1980) developed his model with selective predictors and used logistic analysis with the measurement of nine indicators. He took account into firms' size, leverage, liquidity and performance. Y = -1.3 - .4 Y1 + 6.0 Y2 - 1.4 Y3 + .1 Y4 -2.4 Y5 - 1.8 Y6 + .3Y7 -1.7 Y8 - .5Y9 (2) where: Y1 = log(total assets/GNP price-level index); Y2 = total liabilities/total assets; Y3 = working capital/ total assets; Y4 = current liabilities/current assets; Y5 = one if total liabilities exceed total assets, zero otherwise; Y6 = net income/ total assets; Y7 = funds provided by operations/total liabilities; Y8 = one if net income was negative for the last two years, zero otherwise; Y9 = measure of change in net income; Results and discussions This section reports the findings of test generated by Ohlson model. This section reports the findings of test generated by Ohlson model. Profile Analysis: Variable 2006-2007 2005-2006 2004-2005 British Airways Mean SD MEAN SD Mean SD 2006-2007 2005-2006 SIZE 3.49 0.68 3.49 0.72 3.47 0.74 4.56 4.55 TLTA 0.73 0.26 0.68 0.21 0.69 0.38 0.83 0.76 WCTA -0.05 0.16 -0.03 0.19 -0.06 0.19 0.02 -0.02 CLCA 1.94 2.48 1.72 1.55 1.81 1.03 0.94 1.07 NITA 0.09 0.08 0.07 0.15 0.10 0.26 0.04 0.02 FUTL 0.24 0.24 0.22 0.23 0.23 0.23 0.12 0.12 INTWO 0.12 0.34 0.00 0.00 0.00 0.00 0.00 0.00 OENEG 0.06 0.25 0.06 0.24 0.09 0.29 0.00 0.00 CHIN 0.25 0.44 0.01 0.37 0.06 0.53 0.28 0.32 Test of Ohlson Model on SPSS: The test has been compared between the mean values of all companies and the absolute values of British Airways for 2006-2007 and 2005-2006. The output is shown in the following tables. With a 95% confidence interval, the mean and standard deviation is shown in first table and the lower and upper bounds for the companies is presented in the second table of SPSS test results. Hypotheses H0: Altman's z scores model predicts more accurate percentage in bankruptcy costs than Hybrid Neural Network Model H1: opposite of the above statement As Hsieh, Wen-Kuei, et al (2006) developed 4 models of Hybrid Neural Network in predicting bankruptcy probability which provided high accuracy rate of 86% in model 1 alone. Their experimental result indicated hybrid neural network model performed well using Taiwan bankruptcy data. Test whether their model work for UK companies. Find out the percentage of accuracy for 1 year, 2 years and 3 years respectively prior to bankruptcy. Conclusions References According to existing literature, no single model in the world can differentiate between bankrupt and non- bankrupt firms. Previous studies suggest different uses for Altman E I (2000), Predicting financial distress of companies: Revisiting the Z-Score and Zeta models Avaiable at http://pages.stern.nyu.edu/'ealtman/Zscores.pdf Beaver, W H (1968), Market prices, financial ratios and the prediction of failure, Journal of Accounting Research, 6, 179-192 Begley J, Ming J, Watts Susan G (1997), Bankruptcy classification errors in the 1980s: An empirical analysis of Altman's and Ohlson's Models, Review of accounting studies, vol 1, No 4. Chava S, Jarrow R A (2004), Bankruptcy Prediction with Industry Effects, Review of Finance, v. 8(4), p 537-69' Gilbert, Lisa R, Menon K, Schwartz, Kenneth (1990), Predicting Bankruptcy for firms in financial distress, ournal of Business Finance & Accounting; Vol. 17 Issue 1, p161-171 Grice J S, Dugan M T (2001), The Limitations of Bankruptcy Models: Some Cautions for the researcher, Review of Quantitative Finance and Accounting Researcher, Volume 17(2) pp. 151-166 Gricea John S, Ingramb Robert W. (2001), Tests of the generalizability of Altman's bankruptcy prediction model, Journal of Business Research 54 (2001) 53- 61 Haber Jeffry R, College Iona (2005), Assessing How Bankruptcy Prediction Models Are Evaluated, Journal of Business & Economics Research; January 2005 Volume 3, Number 1 87 A Hsieh, Wen-Kuei., Liu, Shang-Ming., Hsieh, Sung-Yi., (2006), Hybrid Neural Network Bankruptcy prediction: an integration of financial ratios, intellectual capital ratios, MDA and Neural Network Learning, JCIS-2006 Proceedings, Advances in Intelligent Systems Research Hillegeist, Stephen a, Keating, Elizabeth K, Cram P Donald & Lundstedt, Kyle G (2004), Assessing the probability of bankruptcy, Review of accounting studies, Springer Netherlands, 9, 5-34 Hui Huang, Jing-Jing Zhao (2008), Relationship between Corporate Governance and Financial Distress: An Empirical Study of Distressed Companies in China, International Journal of Management Vol. 25 No. 3 Haber R Jeffry, College Iona (2005), Assessing how bankruptcy prediction models are evaluated, Journal of Business & Economics Research, Vol 3, No1 87 Korteweg Arthur G. (2007), The cost of financial distress across industries, Working paper series, Available at http://papers.ssrn.com/sol3/papers.cfm'abstract_id=945425 Morris, Richard, (1998), Bankruptcy prediction models: Just how useful are they'; Credit Management FindArticles.com. Accessed at 29 Nov. 2008. http://findarticles.com/p/articles/mi_qa5308/is_199805/ai_n21421946 Nam C W, Kim T S, , Park N J & Lee H K (2008), Bankruptcy prediction using a discrete-time duration model incorporating temporal and macroeconomic dependencies, Journal of Forecasting, vol 27 iss 6, pg 493-506 Ohlson, J. (1980). Financial ratios and the probabilistic prediction of bankruptcy. Journal of Accounting Research, 19, 109-131 Philosophov, Vladimir L, Philosphov Leonid V (1999), Optimization of corporate capital structure A probabilistic Bayesian approach, International Review of Financial-Analysis; 1999, Vol. 8 Issue 3, p199-214 Pindado Julio, Rodrigues Luis, Torre Chabela de la (2006), International Evidence: Estimating the Probability of Financial Distress, Working paper series Shumway Tyler (2001), Forecasting bankruptcy more accurately: A Simple Hazard model, Journal of Business, January 2001, 101-124. Ward, Terry J (2007), The impact of the response measure used for financial distress on results concerning the predictive usefulness of accounting information, Academy of accounting and financial studies journal, Sep 2007 Available at http://findarticles.com/p/articles/mi_hb6182/is_3_11/ai_n29363363/pg_1'tag=artBody;col1 Cochran James J, Darrat Ali F, Elkhal Khaled (2006), On the bankruptcy of internet companies: an empirical inquiry, Journal of Business Research, Vol 59, Issue 10/11, p 1193-1200 Aziz Abdul, Emanuel David C, Lawson Gerald H (1988), Bankruptcy prediction- an investigation of cash flow based models, Journal of Management Studies, Vol. 25 Issue 5, p419-437 Boritz, J. Efrim; Kennedy, Duane B.; Sun, Jerry Y (2007), Predicting business failure in Canada, Accounting Perspectives, 2007, Vol. 6 Issue 2, p141-165, 25p Agarwal Vineet, Taffler Richard (2008), Comparing the performance of market-based and accounting-based bankruptcy prediction models, Journal of Banking and Finance 32(2008), 1541-1551 He Y, Kamath R, Meirer H H (2005), An empirical evaluation of bankruptcy prediction models for small firms: an over-the-counter market experience, Academy of Accounting and Financial Studies Journal, Jan 2005. Mossman C E, Bell G G, Swartz L M, Turtle H (1998), An empirical comparison of bankruptcy model, Financial Review, 33, 35-54 Read More