| type="text/javascript"> | | | | in combination (Altman, Marco, and Varetto 1994). |
| INTRODUCTION 2 MEASURING FINANCIAL | | | | Altman's Z Score Based on multiple discriminate |
| HEALTH 2 FINANCIAL DISTRESS 2 FACTORS | | | | analysis (MDA), the model predicts a company's |
| AFFECTING FINANCIAL HEALTH 3 Capital | | | | financial health based on a discriminant function of |
| Structure and Capital Adequacy 3 Operating Cash | | | | the form: |
| Flows and Cost Structure 4 Earnings Capacity 4 | | | | Z=0.012X1+0.014X2+0.033X3+0.006X4+0.999X5 |
| Liquidity 4 Asset Conversions — "Growing | | | | Where: X1=working capital/total assets |
| Broke" 5 Asset Utilisation Efficiency/Turnover 5 | | | | X2=retained earnings/total assets X3=earnings |
| Strategic Position 5 PREDICTING FINANCIAL | | | | before interest and taxes/total assets |
| DISTRESS 6 FAILURE PREDICTION MODELS 7 | | | | X4=market value of equity/book value of total |
| Altman's Z Score 8 Logit Analysis: The Model 9 | | | | liabilities X5=sales/total assets The Z-Score model |
| Other Statistical Failure Prediction Models 10 The | | | | (developed in 1968) was based on a sample |
| Gambler's Ruin Models 10 Alternative Models - | | | | composed of 66 manufacturing companies with |
| Artificial Neural Networks 12 CONCLUSION 12 | | | | 33 firms in each of two matched-pair groups. The |
| REFERENCES 13 Introduction A company trying | | | | bankruptcy group consisted of companies that |
| to achieve its business plan faces problems similar | | | | filed a bankruptcy petition under Chapter 11 of |
| to those faced by a driver embarking on a long | | | | the United States bankruptcy act from 1946 |
| trip. The likelihood that car and driver will reach | | | | through 1965. Based on the sample, all firms |
| their destination is dependent on: 1) how much | | | | having a Z-Score greater than 2.99 clearly fell into |
| fuel is in the car's tank upon starting out, 2) the | | | | the non-bankruptcy sector, while those firms |
| car's fuel efficiency, 3) how many service stations | | | | having a Z-Score below 1.81 were bankrupt. |
| will be available to refill the car's fuel tank along | | | | Altman subsequently developed a revised Z-Score |
| the way and 4) whether the car's fuel tank is | | | | model (with revised coefficients and Z-Score |
| large enough to cover unexpected accidents, | | | | cut-offs) which dropped variables X4 and X5 |
| delays, and detours along the way. Similarly, | | | | (above) and replaced them with a new variable |
| whether or not a company survives in a highly | | | | X4 = net worth (book value)/total liabilities. The |
| competitive business environment is dependent | | | | X5 variable was dropped to minimise potential |
| upon: 1) how financially healthy the corporation is | | | | industry effects related to asset turnover. Around |
| at its inception, 2) the company's ability (and | | | | 1977, Altman developed jointly with a private |
| relative flexibility and efficiency) in creating cash | | | | financial firm (ZETA Services, Inc.) a revised |
| from its continuing operations, 3) the company's | | | | seven-variable ZETA model based on a combined |
| access to capital markets, and 4) the company's | | | | sample of 113 manufacturers and retailers. The |
| financial capacity and staying power when faced | | | | ZETA model is allegedly far more accurate in |
| with unplanned cash shortfalls. Measuring Financial | | | | bankruptcy classification in years 2 through 5 with |
| Health There is no single measure of financial | | | | the initial year's accuracy about equal. However, |
| health. Ideally, solvency could be measured along a | | | | the coefficients of the model are not specified |
| continuum in the same way that fuel sufficiency | | | | (without retaining ZETA Services). The ZETA |
| can be measured using a car's petrol gauge. Full | | | | model is based on the following variables: |
| health would equate with having a full tank of fuel. | | | | return on assets stability of |
| Poor health would be equivalent to showing an | | | | earnings debt service |
| empty tank. As healthiness progressively | | | | cumulative profitability liquidity/current |
| decreased, the solvency gauge would register | | | | ratio capitalisation (five year average of |
| movement in the direction of relative insolvency. | | | | total market value) size (total tangible |
| Ultimately, as healthiness continues to decline, the | | | | assets) Logit Analysis: The Model Application of |
| solvency gauge would hopefully flash a warning | | | | the logit model requires four steps. 1. a series of |
| light. Since, in the real world, no single measure of | | | | seven financial ratios are calculated. 2. each ratio is |
| financial health exists, proxies that measure | | | | multiplied by a coefficient unique to that ratio. This |
| various aspects of solvency are often combined | | | | coefficient can be either positive or negative. 3. |
| to estimate a company's healthiness at a point in | | | | the resulting values are summed together (y). 4. |
| time. Financial Distress As a financially healthy | | | | the probability of bankruptcy for a firm is |
| company becomes more and more financially | | | | calculated as the inverse of (1 + ey). Explanatory |
| distressed, it ultimately enters an area of great | | | | variables with a negative coefficient increase the |
| danger. Changes to the company's operations and | | | | probability of bankruptcy because they reduce ey |
| capital structure (ie. restructuring) must be made | | | | toward zero, with the result that the bankruptcy |
| to remain healthy. Apple Computers' attempts in | | | | probability function approaches 1/1, or 100 |
| recent years to restructure its operations to | | | | percent. Likewise, independent variables with a |
| survive in the highly competitive computer | | | | positive coefficient decrease the probability of |
| hardware business is a good example of a | | | | bankruptcy (Stickney 1996). Table 1 shows the |
| company trying to dramatically restructure itself in | | | | financial ratios used in the logit model and their |
| order to maintain solvency. Continued decreases | | | | respective coefficients. TABLE 1 — Financial |
| in financial health ultimately lead to insolvency and | | | | Ratios used in Logit Model FINANCIAL RATIO |
| then potentially, bankruptcy. Available evidence | | | | COEFFICIENT + 0.23883 Average Inventories |
| suggests many companies do not adequately | | | | Sales - 0.108 Average Receivables/Average |
| attempt to resolve their financial health problems | | | | Inventories - 1.583 (Cash + Marketable Securities) |
| until it is too late to avoid bankruptcy. Factors | | | | Total Assets - 10.78 Quick Assets/Current |
| Affecting Financial Health Capital Structure and | | | | Liabilities + 3.074 Income from Continuing |
| Capital Adequacy Companies finance their | | | | Operations/(Total Assets - Current Liabilities) + |
| long-term operations primarily through two | | | | 0.486 Long-Term Debt/(Total Assets - Current |
| sources of capital - debt and equity. One of the | | | | Liabilities) - 4.35 Sales/(Net Working Capital + |
| most important financing decisions a company | | | | Fixed Assets) + 0.11 y = Sum of (Coefficient * |
| makes is the proportion of debt to owner's equity | | | | Ratio) Probability of Bankruptcy = 1/(1 + ey) |
| in the company's capital structure. Summary | | | | Other Statistical Failure Prediction Models Many |
| measures of a company's capital structure include | | | | additional bankruptcy prediction models have been |
| the company's debt to equity ratio (D/E) and | | | | developed since the work of Beaver and Altman. |
| debt to total capital ratio (D/(D+E)). Interest and | | | | Lev (1974), Deakin (1977), Ohlson (1980), Taffler |
| principal payments on debt must be paid from | | | | (1980), Platt & Platt (1990), Gilbert, Menon, and |
| operations before any payments can be | | | | Schwartz (1990), and Koh and Killough (1990) |
| distributed to equity holders (in the form of | | | | amongst others have continued to refine the |
| dividends or share buy-backs). Therefore, the | | | | development of multivariate statistical models. |
| interest and principal, which must be paid on debt, | | | | Almost all of these traditional models have been |
| are considered fixed-costs of operations. From an | | | | either matched-pair multi-discriminate models or |
| operational point-of-view, the extent of the | | | | logit models. A 1997 study by Begley, Ming and |
| burden of these fixed obligations can be | | | | Watts concludes: "Given that Ohlson's original |
| measured relative to the company's continuing | | | | model is frequently used in academic research as |
| ability to pay the fixed obligations. A frequently | | | | an indicator of financial distress, its strong |
| used measure of a company's ability to cover its | | | | performance in this study supports its use as a |
| interest payments is its earnings before interest | | | | preferred model." The Gambler's Ruin Models |
| and taxes and before depreciation and | | | | Wilcox (1971 and 1976), Santomero (1977), Vinso |
| amortisation (EBITDA) to its interest expense. A | | | | (1979) and others have adapted a gambler's ruin |
| company is financially distressed whenever its | | | | approach to bankruptcy prediction. Under this |
| EBITDA is less than its interest expense. | | | | approach, bankruptcy is probable when a |
| Financial leverage involves the | | | | company's net liquidation value (NLV) becomes |
| substitution of fixed-cost debt for owner's equity | | | | negative. Net liquidation value is defined as total |
| in the hope of increasing equity returns. As | | | | asset liquidation value less total liabilities. From one |
| demonstrated by Higgins and others, financial | | | | period to the next, a company's NLV is increased |
| leverage improves financial performance when | | | | by cash inflows and decreased by cash outflows |
| things are going well but worsens financial | | | | during the period. Wilcox combined the cash |
| performance when things are going poorly. | | | | inflows and outflows and defined them as |
| Therefore, increasing the ratio of debt to equity in | | | | adjusted cash flow. All other things being equal, |
| a company's capital structure implicitly makes the | | | | the probability of a company's failure increases, |
| company relatively less solvent (on the downside) | | | | the smaller the company's beginning NLV, the |
| and more financially risky than a company without | | | | smaller the company's adjusted (net) cash flow, |
| debt. Capital adequacy relates to | | | | and the larger the variation of the company's |
| whether a company has enough capital to finance | | | | adjusted cash flow over time. Wilcox uses the |
| its planned future operations. If the company's | | | | gambler's ruin formula (Feller, 1968) to show that |
| capital is inadequate, then it must either be able | | | | a company's risk of failure is dependent on; 1) the |
| to: 1) successfully issue new equity, or 2) arrange | | | | above factors plus, 2) the size of the company's |
| new debt. The amount of debt a company can | | | | adjusted cash flow at risk each period (ie. the size |
| successfully absorb and repay from its continuing | | | | of the company's bet). Using a more robust |
| operations is normally referred to as the | | | | statistical technique, Vinso (1979) extended |
| company's debt capacity. Capital adequacy is | | | | Wilcox's gambler's ruin model to develop a safety |
| normally evaluated by looking at the company's | | | | index. Based on input concerning the variability of |
| operational cash flow projections and its | | | | expected contribution margin amounts, the index |
| projections of capital needs. When companies | | | | can be used to predict the point in time when a |
| undertake major new projects or undergo a | | | | company's ruin is most likely to occur (called first |
| significant financial restructuring they often | | | | passage time). The statistics used in gambler's ruin |
| perform financial feasibility studies to determine | | | | approaches are somewhat formidable (especially |
| whether the company has the financial capacity | | | | to the average reader). However, both Wilcox |
| to undertake the project and whether the | | | | and Vinso richly describe some of the factors |
| company will be able to repay all future debt | | | | which most affect business failure. For example, |
| payments once the project is built. Operating | | | | Wilcox states: "The (cash) inflow rate ... can be |
| Cash Flows and Cost Structure All other factors | | | | increased through higher average return on |
| being equal, companies that can consistently | | | | investment. However, having a major impact here |
| generate positive cash flows from operations will | | | | usually requires long-term changes in strategic |
| remain relatively more solvent than those that | | | | position. This is difficult to control over a short |
| cannot. This requires that operating cash inflows | | | | time period except by divestitures of peripheral |
| (collections or sales) consistently exceed operating | | | | unprofitable businesses...The average outflow rate |
| cash outflows (costs). Companies which | | | | is controlled by managing the average growth |
| experience erratic cash outflows and inflows are | | | | rate of corporate assets. Effective capital |
| relatively more risky because they are less likely, | | | | budgeting ... requires resource allocation |
| in one or more time periods, to be able to cover | | | | emphasising those business units, which have the |
| fixed expenses/outflows. Companies which have | | | | highest future payoff. The size of the bet is the |
| a higher proportion of fixed costs to variable | | | | least understood factor in financial risk. Yet |
| costs are also relatively more risky and relatively | | | | management has substantial control over it. |
| less solvent than companies with a relatively | | | | Variability in liquidity flows governs the size of the |
| lower proportion of fixed costs in their operating | | | | bet. This variability can be managed through |
| cost structure. Earnings Capacity All other things | | | | dividend policy, through limiting earning variability |
| being equal, companies with higher relative | | | | and investment variability, and through controlling |
| earnings and higher relative returns on investment | | | | the co-variation between profits and |
| will remain more solvent than their less fortunate | | | | investments...True earnings smoothing is attained |
| competitors. The most commonly used financial | | | | by control of exposure to volatile industries, |
| measures of earnings capacity are earnings | | | | diversification, and improved strategic position." |
| before interest and taxes (EBIT) and net income. | | | | Vinso supports Wilcox's emphasis on cash flow |
| Liquidity Adequate liquidity is a further necessary | | | | processes and stresses the importance of debt |
| component of solvency. Frequently used liquidity | | | | capacity: "Before deriving a mathematical model |
| measures include: a) working capital (current | | | | for determining the risk of ruin, it is necessary to |
| assets minus current liabilities), b) current ratio | | | | describe the process. First, a firm has some pool |
| (current assets divided by current liabilities), and c) | | | | of resources at time = 0 of some size U0, which |
| quick ratio (cash, marketable securities and | | | | are available to prevent ruin (similar to Wilcox's |
| accounts receivable divided by current liabilities). | | | | beginning NAV). Then, earnings come to the firm |
| To evaluate liquidity, each of the assets and | | | | from revenue(s)...less the costs incurred in |
| liabilities on a company's balance sheet should be | | | | producing the revenues. There are two types of |
| evaluated for liquidity. Current assets are those | | | | costs to be considered: variable, which change |
| which will likely be converted to cash within one | | | | according to the stochastic nature of the revenue |
| year or less. Current liabilities are those which | | | | sources, and fixed costs, which do not vary with |
| must be paid within one year. However, when a | | | | revenue but are a function of the period. So, |
| company becomes financially distressed, even | | | | revenue less variable costs...can be defined as |
| assets which are normally considered current | | | | variable profit (which is available to pay fixed |
| assets (accounts receivable and stock, for | | | | costs). If Ut is less than zero, ruin occurs because |
| example) may become relatively "illiquid". | | | | no funds are available to meet unpaid fixed |
| Long-term assets, in general, are far less liquid | | | | costs...These definitions, however, ignore debt |
| than current assets. Some longer-term assets | | | | capacity, if available, which must be included as |
| may be very "illiquid". Also, as stated above, often | | | | the firm can use this source without being forced |
| a company's long-term liabilities can become | | | | to confront shareholders, creditors or |
| immediately due and payable if the company | | | | bankruptcy,...debt holders or other creditors will |
| violates contractual debt covenants or other | | | | force reorganisation if a firm is unable to meet |
| obligations. Wilcox (1976) argues that net liquidation | | | | contractual obligations because working capital is |
| value provides a solid conceptual basis for | | | | too low and the firm cannot obtain more debt." |
| evaluating a company's liquidity. Net liquidation | | | | Alternative Models - Artificial Neural Networks |
| value is defined as total asset liquidation value less | | | | Since 1990, another promising approach to |
| total liabilities. Wilcox (1976) applies what he calls | | | | bankruptcy prediction, based on the use of neural |
| typical (not definitive) valuation multipliers to | | | | networks, has evolved. Artificial Neural Networks |
| balance sheet assets to arrive at representative | | | | (ANN) are computers constructed to process |
| asset liquidation values: Cash Equivalents | | | | information, in parallel, similar to the human brain. |
| 100% Other Current Assets 70% | | | | ANN's store information in the form of patterns |
| Long Term Assets 50% Wilcox (1976) | | | | and are able to learn from their processing |
| shows that a company becomes bankrupt when | | | | experience. Unlike MDA and logit analyses, ANN's |
| net liquidation value is reduced to zero. Asset | | | | impose less restrictive data requirements (the |
| Conversions — "Growing Broke" Asset and | | | | requirement for linearity, for example) and are |
| liability conversions are continuously ongoing in any | | | | especially useful in recognising and learning |
| dynamic business. Operationally, the company is | | | | complex data relationships. Recent ANN |
| selling its products thereby creating cash inflows. | | | | bankruptcy prediction studies include those of Bell, |
| Alternatively, sales may be made on credit, | | | | et al. (1990), Hansen & Messier (1991), Chung & |
| increasing the company's accounts receivable. | | | | Tam (1992), Liang, et al. (1992), Tam & Kiang |
| Concurrently, inventories are produced and sold | | | | (1992), Salchenberger (1993), Coats & Fant |
| and production and operating expenses are | | | | (1993), Fanning & Cogger (1994), Brockett, et al. |
| incurred to continue operations. If a company's | | | | (1994), Boritz, et al. (1995), and Etheridge & |
| inventories and accounts receivable grow faster | | | | Siriam (1995 and 1997). Research has shown that |
| than the corresponding growth in the company's | | | | ANN's offer a viable alternative to other more |
| sales and accounts payable, liquidity will be | | | | traditional methods of bankruptcy prediction. The |
| negatively affected. Strategic asset conversions | | | | ability of the model to learn allows for the |
| are also ongoing, but with less regularity. Decisions | | | | constant re-calibration and validation of the model, |
| to invest in 'bricks and mortar' and other | | | | which helps increase classification rates. From a |
| long-term investments are made and debt and | | | | theoretical perspective, ANN's are more desirable |
| equity are obtained to supply the capital needed | | | | because they make fewer assumptions about the |
| to pay for them. Slowly but surely, companies | | | | data normality and linear separability. One of the |
| can 'go broke' when assets are converted to less | | | | main disadvantages of ANN's is the inability to |
| liquid forms over a sustained time period. This can | | | | assign intuition the network weights. Another |
| happen when the company's assets grow faster | | | | disadvantage is that the model might simply |
| than the company's sales (often the case for | | | | memorise the data as opposed to forming a |
| many start-up companies). When this happens, | | | | general set of classification rules, which can cause |
| the company becomes more highly leveraged and | | | | estimates on future samples to be less reliable. |
| less solvent. Similarly, a company whose long | | | | Conclusion Future research in bankruptcy |
| term investment decisions do not pay off in | | | | prediction should analyse the economic and |
| terms of planned operating returns (thus | | | | institutional factors that can impact the reasons |
| increasing fixed cost structures and decreasing | | | | for bankruptcy. Jones (1987) indicated that the |
| operating cash flows), will become less solvent. | | | | lack of homogeneity in the motivation for a |
| Asset Utilisation Efficiency/Turnover Those | | | | bankruptcy filing might complicate the modelling |
| companies, which survive, use their human and | | | | effort. Although normally motivated by an effort |
| capital assets relatively efficiently. That is, they | | | | to resolve severe financial problems, a firm may |
| have relatively higher returns on investment (ROI) | | | | file for bankruptcy primarily to void a union |
| and higher returns per employee than less | | | | contract or for other legal reasons (Jones 1987). |
| successful competitors. They achieve relatively | | | | Another area where models can be improved is in |
| higher returns through superior asset | | | | catering for predictor variables other than financial |
| management (capital and human assets) and | | | | ratios may prove beneficial. For example, |
| through superior strategic positioning. In the | | | | measures of management experience, |
| absence of aggressive asset management, | | | | management expertise, or other behavioural |
| companies must usually resort to wholesale asset | | | | aspects that impact the operations of the firm |
| divestitures and/or are forced to restructure to | | | | could be significant in a bankruptcy prediction |
| fund their continuing operations. Strategic Position | | | | model. Additionally, including variables that control |
| Schoffler (Buzzell and Gale, 1987) and others have | | | | for a changing economic environment may |
| documented the high correlation between positive | | | | provide valuable insights for predicting bankruptcy. |
| returns on investment and such factors as: 1) | | | | Bibliography References Altman, Edward I. |
| higher relative market shares, 2) relative product | | | | Corporate Financial Distress. New York, NY: John |
| quality and 3) lower relative capital intensity. | | | | Wiley and Sons, 1983. Altman, Edward I. (1968) |
| Companies that have strong strategic market | | | | Financial Ratios, Discriminate Analysis and the |
| positions are more likely to experience higher | | | | Prediction of Corporate Bankruptcy, The Journal |
| relative returns on investment than their | | | | of Finance. Altman, Edward I. Homepage of |
| competitors. These positive returns, in turn, | | | | Professor Edward I. Altman, New York, NY: Stern |
| increase the solvency of the market leaders. | | | | School of Business. Available at Altman, Edward I, |
| Those competitors that have lower market | | | | Giancarlo Marco, and Franco Varetto (1994) |
| shares or lower product quality are less likely to | | | | Corporate Distress Diagnosis: Comparisons Using |
| achieve industry average returns and are thus | | | | Linear Discriminant Analysis and Neural Networks |
| more likely to become less solvent in the future. | | | | (the Italian Experience), The Journal of Banking |
| Predicting Financial Distress In America, each year | | | | and Finance. Altman, Edward I. and Thomas P. |
| approximately one percent of all firms required to | | | | McGough (1974) Evaluation of a Company as a |
| file with the Securities and Exchange Commission | | | | Going Concern, The Journal of Accountancy. |
| file for bankruptcy. The American Bankruptcy | | | | Beaver, W., 1967, Financial Ratios as Predictors of |
| Institute reports that around 50,000 businesses | | | | Failures, in Empirical Research in Accounting, |
| filed for bankruptcy in 1997. Attempts to develop | | | | Journal of Accounting Research. Begley, J., Ming, J., |
| bankruptcy prediction models began seriously | | | | Watts, S., 1997,Bankruptcy Classification Errors in |
| sometime in the late 1960's and continue through | | | | the 1980s: An Empirical Analysis of Altman's and |
| today. At least three distinct types of models | | | | Ohlson's Models, Review of Accounting Studies. |
| have been used to predict bankruptcy: a) | | | | Bell, T.B., G.S. Ribar and J. Verchio, 1990, Neural |
| statistical models (univariate analysis, multiple | | | | Nets Versus Logistic Regression: A Comparison of |
| discriminate analyses [MDA]), and conditional logit | | | | Each Model's Ability to Predict Commercial Bank |
| regression analyses, b) gambler's | | | | Failures," Boritz, J.E., D.B. Kennedy and A.M. |
| ruin-mathematical/statistical models, and c) artificial | | | | Albuquerque, 1995, Predicting Corporate Failure |
| neural network models. Each of these models is | | | | Using a Neural Network Approach, Intelligent |
| discussed below. Most of the publicly available | | | | Systems in Accounting, Finance and Management. |
| information regarding prediction models is based | | | | Brockett, P.L., W.W. Cooper, L.L. Golden and U. |
| on research published by academics. Commercial | | | | Pitaktong, 1994, A Neural Network Method for |
| banks, public accounting firms and other | | | | Obtaining an Early Warning of Insurer Insolvency. |
| institutional entities (ratings agencies, for example) | | | | The Journal of Risk and Insurance. Buzzell, R.D., |
| appear to be the primary beneficiaries of this | | | | Gale, B.T., 1987, The PIMS Principles Linking |
| research, since they can use the information to | | | | Strategy to Performance, New York: The Free |
| minimise their exposure to potential client failures. | | | | Press. Chung, H.M. and K. Y. Tam, 1992, A |
| While continuing research has been ongoing for | | | | Comparative Analysis of Inductive-Learning |
| almost thirty years, it is interesting to note that | | | | Algorithms, Intelligent Systems in Accounting, |
| no unified well-specified theory of how and why | | | | Finance and Management. Coats, P.K. and L.F. Fant, |
| corporations fail has yet been developed. The | | | | 1993, Recognizing Financial Distress Patterns Using |
| available statistical models derive merely from the | | | | a Neural Network Tool, Financial Management. |
| statistical optimisation of a set of ratios. As stated | | | | Cook, Roy A. and Jeryl L. Nelson. A Conspectus |
| by Wilcox (1973) the lack of conceptual | | | | of Business Failure Forecasting, Available at |
| framework results in the limited amount of | | | | Deakin, E., Business Failure Prediction: An Empirical |
| available data on bankrupt firms being statistically | | | | Analysis,, 1977, in E. Altman and A. Sametz, eds., |
| 'used up' by the search before a useful | | | | Financial Crises: Institutions and Markets in a Fragile |
| generalisation emerges. How useful are these | | | | Environment, New York: Wiley. Etheridge, H.L. and |
| models? Almost universally, the decision criterion | | | | R.S. Sriram, 1995, A Neural Network Approach to |
| used to evaluate the usefulness of the models | | | | Financial Distress Analysis, Advances in Accounting |
| has been how well they classify a company as | | | | Information Systems. Fanning, K. and K.O. Cogger, |
| solvent or non-solvent compared to the | | | | 1994, A Comparative Analysis of Artificial Neural |
| company's actual status known after-the-fact. | | | | Networks Using Financial Distress Prediction, |
| Most of the studies consider a type I error as the | | | | Intelligent Systems in Accounting, Finance and |
| classification of a failed company as healthy, and | | | | Management. Gilbert, L.R., Menon, K., and |
| consider a type II error as the classification of a | | | | Schwartz, K.B., 1990, Predicting Bankruptcy for |
| healthy company as failed. In general, type I | | | | Firms in Financial Distress, Journal of Business |
| errors are considered more costly to most users | | | | Finance and Accounting. Hansen, J.V. and W.F. |
| than type II errors. The usefulness of fail/non-fail | | | | Messier, 1991,Artificial Neural Networks: |
| prediction models is suggested by Ohlson (1980) | | | | Foundations and Application to a Decision Problem, |
| "...real world problems concern themselves with | | | | Expert Systems with Applications. Jones, F. L. |
| choices which have a richer set of possible | | | | 1987. Current Techniques in Bankruptcy Prediction. |
| outcomes. No decision problem I can think of has | | | | Journal of Accounting Literature. Koh, H.C. and |
| a payoff space which is partitioned naturally into | | | | Killough, L.N., 1990, The Use of Multiple Discriminant |
| the binary status bankruptcy versus | | | | Analysis in the Assessment of the Going-concern |
| non-bankruptcy...I have also refrained from making | | | | Status of an Audit Client, Journal of Business |
| inferences regarding the relative usefulness of | | | | Finance and Accounting. Lev, B., 1974, Financial |
| alternative models, ratios and predictive systems... | | | | Statement Analysis, A New Approach. Englewood |
| Most of the analysis should simply be viewed as | | | | Cliffs, N.J.: Prentice-Hall. Liang, T.P., J.S. Chandler, I. |
| descriptive statistics - which may, to some | | | | Ha |
| extent, include estimated prediction error-rates - | | | | This article is free for republishing |
| and no theories of bankruptcy or usefulness of | | | | Source: |
| financial ratios are tested." Subject to the | | | | |
| qualifications expressed above, bankruptcy | | | | |
| prediction models continue to be used to predict | | | | Related Articles |
| failure. Failure Prediction Models The early history | | | | A good dental insurance policy can make your |
| of researchers' attempts to classify and predict | | | | smile brighter |
| business failure (and bankruptcy) is well | | | | Business Insurance |
| documented in Edward Altman's 1983 book, | | | | Consumer credit 101- What You Absolutely Need |
| Corporate Financial Distress. Statistical prediction | | | | to Know |
| models are more generally better known as | | | | Apartment Loan 101 |
| measures of financial distress. Three stages in the | | | | User guide to payday loan terminologies What |
| development of statistical financial distress models | | | | Your Mama Never Told You About Business |
| exist: 1. univariate analysis, 2. multivariate (or | | | | Credit Cards |
| multi-discriminate [MDA]) analysis, and 3. logit | | | | Drive away the car of your dreams |
| analysis. Univariate analysis assumes that a single | | | | Kippers or red herrings? |
| variable can be used for predictive purposes | | | | Ethical finance: who benefits from our spending? |
| (Cook and Nelson 1998). The univariate model as | | | | Save Time With a Bridge Loan |
| proposed by William Beaver achieved a moderate | | | | Related Articles |
| level of predictive accuracy (Sheppard 1994). | | | | Alternative ways to avoid payday loan |
| Univariate analysis identified factors related to | | | | Help to get out of debt |
| financial distress, however, it did not provide a | | | | Do you know what you're worth? |
| measure of the relevant risk (Stickney 1996). In | | | | Forex Trading Best Practices |
| the next stage of financial distress measurement, | | | | Lead Companies, Eight Features To Consider |
| multivariate analysis (also known as multiple | | | | Residual Income - What It Is and Why You Need |
| discriminant analysis or MDA) attempted to | | | | It |
| overcome the potentially conflicting indications that | | | | Life Insurance Settlements-Capitalizing On Your |
| may result from using single variables (Cook and | | | | Policies |
| Nelson 1998). The best-known, and most-widely | | | | What Your Mama Never Told You About Debt |
| used, multiple discriminant analysis method is the | | | | Consolidation Services |
| one proposed by Edward Altman. Altman's | | | | Personal loan 101- What You Absolutely Need to |
| z-score, or zeta model, combined various | | | | Know |
| measures of profitability or risk. The resulting | | | | Residential Income Property Financing: Part 2 of 3 |
| model was one that demonstrated a company's | | | | Government loans for beginners |
| risk of bankruptcy relative to a standard. Altman's | | | | Government grant writing for beginners |
| initial study proved his model to be very accurate; | | | | Life Settlement: Towards A Free Market for Life |
| it correctly predicted bankruptcy in 94% of the | | | | Insurance |
| initial sample (Altman 1968). Despite the positive | | | | The Hitchhiker's Guide To Insanity |
| results of his study, Altman's model had a key | | | | 7 Years At The Keyboard |
| weakness; it assumed variables in the sample | | | | Shorn of the debt responsibility |
| data to be normally distributed. If all variables are | | | | Disability insurance 101- What You Absolutely |
| not normally distributed, the methods employed | | | | Need to Know |
| may result in selection of an inappropriate set of | | | | Are you running the risk of an uninsured business? |
| predictors (Sheppard 1994). Chistine Zavgren | | | | Beware get a "Business insurance" |
| developed a model that corrected for this | | | | "Where Is America Headed To?", New York |
| problem. Her model used logit analysis to predict | | | | Millionaire Wonders
|
| bankruptcy. Due to its use of logit analysis, her | | | | 4 things to watch out for when choosing a |
| model is considered more robust (Lo 1986). | | | | mortgage company |
| Further, logit analysis actually provides a probability | | | | |
| (in terms of a percentage) of bankruptcy. Also, | | | | Ask a Question About this Article |
| the probability calculated might be considered a | | | | >> WHO MADE DIECAST MODEL |
| measure of the effectiveness of management (ie. | | | | TROLLEYBUS 1.50 (approx ... |
| effective management will not lead a company to | | | | >> Is there such a thing as comfortable high |
| the verge of bankruptcy). During the 1980s and | | | | heels? |
| 1990s, the trend has been to use logit analysis in | | | | >> How does forming a limited liability |
| favour of multiple discriminant analysis (Stickney | | | | corporation ... |
| 1996). More recently, logit analysis has been | | | | >> Do you think that doomsday predictions |
| compared to a more advanced analytical tool, | | | | that are about to occur in the year of 2012 are |
| neural networks. Research has found that the | | | | real? |
| approaches perform similarly and should be used | | | | |