FSA 7120 Module 5 Final Project: Measuring and Predicting Financial Stress Example

Early Warning or False Alarm? Measuring Financial Stress With Altman's Z-Score and Ohlson's O-Score

Student Name

Master of Business Administration, Thomas Edison State University

FSA-7120: Financial Statement Analysis

Mentor Name

Month Day, Year

Early Warning or False Alarm? Measuring Financial Stress With Altman's Z-Score and Ohlson's O-Score

Introduction

Companies rarely fail without warning. Before a bankruptcy filing, profits usually turn to losses, working capital thins and borrowing grows relative to assets. The accounting literature has tried for six decades to turn those signals into models that sort companies likely to fail from those likely to survive. This final project reviews that literature, applies its two best-known models to two companies and assesses whether managers can use them as early warning tools, the purpose the course assigns to them. The companies are composites: Harlow Fixtures, the Trenton, New Jersey, lighting manufacturer analyzed throughout the course, and Corbin Signal, a manufacturer of commercial signage with sales of $300 million that has lost money for two years.

The paper proceeds in four steps. It first reviews how the research on distress prediction developed and what each model added. It then computes the Altman Z-score and the Ohlson O-score for both companies from their 2024 statements, showing every input so that a reader can reproduce the results. Third, it compares the results and sets out the limits of both models. Finally, it recommends how the management of each company should use the scores. The data for Harlow come from the statements analyzed in Modules 1 through 4; Corbin's statements were prepared for this project to resemble a manufacturer that has lost money for two years and depends on its lenders' patience.

How Distress Prediction Developed

Beaver (1966) began the modern work by comparing 79 failed firms with matched survivors and testing ratios one at a time. Cash flow relative to total debt was his best single discriminator, with signs visible as much as five years ahead of failure, although predictions grew less reliable the further out he looked. Beaver's method had a weakness: a firm could look strong on one ratio and weak on another, and the analyst had no rule for combining them.

Altman (1968) solved that problem with multiple discriminant analysis, which weights several ratios into a single score. Using 33 bankrupt manufacturers and 33 matched survivors, he selected five ratios covering liquidity, accumulated profitability, operating profitability, market valuation of solvency and asset turnover. The resulting Z-score classified 95 percent of the original sample correctly one year before failure. Its accuracy fell to 72 percent two years before failure, and later tests on new samples found lower accuracy still, a sign that a model fitted to one sample flatters itself.

Ohlson (1980) replaced discriminant analysis with logistic regression, which produces a probability of failure rather than a score and makes fewer assumptions about how the ratios are distributed. His model used nine variables, including size, indebtedness measured as total liabilities to total assets, working capital, liquidity, profitability, funds from operations, a flag for negative equity, a flag for two consecutive years of losses and the change in net income. Using more than 2,000 firms, he showed that size was one of the strongest predictors and that the timing of data mattered: earlier studies had sometimes used statements released after the bankruptcy itself.

Shumway (2001) argued that both earlier approaches were static, since each used one observation per firm and ignored how long a firm had been at risk. His hazard model used every firm-year and added market variables such as past stock returns and the volatility of returns. He found that about half of the accounting ratios used by Altman and Ohlson lost their predictive power once market variables were included. The lesson for this paper is that accounting-based scores are useful but incomplete: the stock market often sees distress before the statements show it.

Later work compared accounting models with models built on market prices. Hillegeist et al. (2004) estimated the probability of bankruptcy from an option pricing model that treats equity as a call option on the firm's assets, using the market value and volatility of equity. Their market-based measure carried more information about subsequent bankruptcies than either the Z-score or the O-score, although the accounting scores still added some information of their own. For an analyst, the practical lesson is that accounting scores and market signals are complements, and a company whose scores look safe while its share price and bond spreads deteriorate deserves attention.

The Altman Z-Score

For publicly traded manufacturers, the original Z-score is Z = 1.2X1 + 1.4X2 + 3.3X3 + 0.6X4 + 1.0X5. X1 divides working capital by total assets, X2 does the same for retained earnings and X3 for operating earnings before interest and taxes; X4 compares the market value of equity with total liabilities, and X5 measures asset turnover as sales over total assets. Altman placed firms scoring above 2.99 in a safe zone and firms below 1.81 in a distress zone, with a gray area between.

Harlow's inputs come from its 2024 statements: working capital of $104 million, retained earnings of $140 million, operating income of $43.2 million and sales of $480 million on total assets of $412 million, with a market value of equity of $396 million against $212 million of liabilities. Corbin's statements show working capital of $12 million, retained earnings of $8 million, operating income of $3 million and sales of $300 million on total assets of $280 million, with equity valued by the market at only $30 million against $236 million of liabilities.

The weights explain the result. The largest weight, 3.3, falls on operating return on assets, and Corbin's near-zero return removes most of the score. The market value term also separates the two firms sharply: investors value Harlow's equity at almost twice its liabilities and Corbin's at about an eighth. Harlow's score of 3.41 places it in the safe zone even after the earnings quality adjustments of Module 2; recasting operating income to $35.2 million lowers X3 to 0.085 and the score to about 3.35, still safe.

Two practical choices affect the result. First, X4 uses the market value of equity rather than book value, so the score responds to investors' views between reporting dates; for Harlow, the market value of $396 million is 22 million shares at the December 2024 price of $18. Second, X3 uses earnings before interest and taxes, which keeps the measure independent of how the company is financed and taxed. Both choices also create weaknesses. A temporary fall in the stock market lowers every firm's score, and operating income inherits the earnings quality concerns raised in Module 2, which is why the recast calculation matters.

The Ohlson O-Score

Ohlson's model adds nine weighted variables to a constant of negative 1.32, with the coefficient of each variable shown in the table below, and the probability of failure equals e to the power O divided by one plus e to the power O. SIZE is the natural log of total assets deflated by a price index with 1968 as the base year. This paper measures total assets in thousands of 1968 dollars, dividing current dollars by 6.39, the approximate rise in the GDP price index since 1968. Because the size term depends on this choice, the level of the probability should be read with care, while the comparison between the two firms, measured the same way, is reliable.

Harlow's probability of 2.5 percent sits below the 3.8 percent cutoff that Ohlson reported as minimizing classification errors in his sample, so the model also classifies it as a survivor. Corbin's probability of 52 percent is driven by its liabilities, which equal 84 percent of assets and carry the largest positive weight, by two straight years of losses and by funds from operations that barely register against what it owes.

Where the Models Agree and Where They Fall Short

Both models place Harlow in the healthy group and Corbin in the distressed group, which is the result an analyst would expect from reading the statements. Their agreement here hides four limits that managers should understand.

First, both were estimated on manufacturers and data from earlier decades. Ratios such as sales to total assets differ by industry, so service and technology firms score differently for reasons unrelated to distress. Second, both rely on reported accounting figures and inherit any earnings management in them. Corbin's managers, facing covenant pressure, have every reason to choose estimates that lift reported income, which would raise its score without improving its position. Third, neither model sees off-balance-sheet commitments, pending litigation or the loss of a key customer, events that often trigger failure. Fourth, the scores are static. Shumway (2001) showed that market variables such as falling stock returns add predictive power, and the market's low valuation of Corbin's equity already signals distress more quickly than its annual statements can.

Using the Scores as Early Warning Tools

For Harlow's management the scores give assurance but should not be ignored. The most useful practice is to compute both scores each quarter, track the trend rather than the level and investigate any decline of more than half a point in Z. The X3 term ties the score directly to operating profitability, so the gross margin risk identified in the Module 3 forecast is also the main threat to the Z-score.

For Corbin's management the warning is urgent. A credible response would act on the variables that drive the scores: sell surplus assets to reduce liabilities, cut working capital tied up in slow inventory, restore positive funds from operations by closing loss-making product lines and negotiate with lenders before a covenant breach removes the choice. Equity analysis, the subject of Module 5, adds a further check. Investors who value Corbin's equity at $30 million are pricing a substantial probability of failure, and management should treat that price as information rather than as an insult.

To make the scores useful rather than ceremonial, each company should build them into its governance. The controller should compute both scores each quarter from the same data used in the quarterly close, report them to the audit committee beside the covenant ratios from Module 4 and explain any change of more than half a point in Z or more than five percentage points in the O-score probability. The board should agree in advance on the actions that a decline will trigger, such as a review of capital spending, a liquidity plan or early talks with lenders, so that the warning leads to action before the options narrow.

Conclusion

Ratio-based prediction has moved from single ratios to weighted scores, probabilities and hazard models that add market data. Applied to two composite manufacturers, Altman's Z-score and Ohlson's O-score agree: Harlow, with a Z of 3.41 and a 2.5 percent probability, is sound, and Corbin, with a Z of 1.27 and a 52 percent probability, is in distress. The models are best used as early warning tools that start an investigation, tracked over time, checked against earnings quality and paired with market signals, rather than as verdicts.

References

Altman, E. I. (1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. The Journal of Finance, 23(4), 589-609. https://doi.org/10.1111/j.1540-6261.1968.tb00843.x

Beaver, W. H. (1966). Financial ratios as predictors of failure. Journal of Accounting Research, 4, 71-111. https://doi.org/10.2307/2490171

Hillegeist, S. A., Keating, E. K., Cram, D. P., & Lundstedt, K. G. (2004). Assessing the probability of bankruptcy. Review of Accounting Studies, 9(1), 5-34. https://doi.org/10.1023/B:RAST.0000013627.90884.b7

Ohlson, J. A. (1980). Financial ratios and the probabilistic prediction of bankruptcy. Journal of Accounting Research, 18(1), 109-131. https://doi.org/10.2307/2490395

Shumway, T. (2001). Forecasting bankruptcy more accurately: A simple hazard model. The Journal of Business, 74(1), 101-124. https://doi.org/10.1086/209665