Strategic Performance Management and Business Valuation · Corporate Failure

Failure Prediction Models and Ratio Analysis for CMA Final

Updated 11 October 2026 · Fact-checked

Failure prediction models use financial ratios or qualitative scores to warn that a company may fail. Beaver tests one ratio at a time. Altman, Springate and Taffler combine several ratios into one score. Argenti's A-score is qualitative. To solve questions, compute the ratios, apply the cut-off, and state a reasoned conclusion.

Understand Other Failure Prediction Models and Ratio Analysis

A company rarely fails overnight. Warning signs show up in its accounts and in how it is managed, often years earlier. Failure prediction models try to turn those signs into a clear signal.

There are two broad types. A univariate model looks at one ratio at a time. William Beaver (1966) compared failed and non-failed firms and found that some ratios separated them well. Cash flow to total debt was the best single predictor. Others he tested were net income to total assets, total debt to total assets, and working capital to total assets. Failed firms showed weaker ratios, and the gap widened as failure came closer.

A multivariate model combines several ratios into one score using weights, typically found by multiple discriminant analysis. Altman's Z-score is the best known. Springate (1978) used four ratios and a cut-off score. Taffler (UK, 1977 onwards) built a model with ratios such as profit before tax to current liabilities, current assets to total liabilities, current liabilities to total assets and a no-credit interval. The result is a Z-type score, where a low or negative score signals risk.

Argenti's A-score is different. It is a qualitative, scored model. It looks at management defects (for example an autocratic chief executive, a weak board, poor finance function), mistakes (such as overtrading, high gearing, a big project that goes wrong) and symptoms (such as creative accounting and falling morale). Defects, mistakes and symptoms are scored, and a high total, above a threshold, signals danger. Argenti described a typical path: defects lead to mistakes, which lead to symptoms, then failure.

Ratios also work as plain warning signals without a formal model. Typical red flags are a falling current ratio, a falling interest cover, rising gearing, slowing inventory and debtor turnover, recurring losses, negative operating cash flow and growing overdue creditors. Use them as a trend over several years, not as a single year reading.

Key rules to remember

Springate model: S = 1.03A + 3.07B + 0.66C + 0.4D; A = working capital ÷ total assets; B = EBIT ÷ total assets; C = profit before tax ÷ current liabilities; D = sales ÷ total assets. A score below 0.862 classifies the firm as likely to fail. Confirm the weights given in the question, and use those.
Beaver's best single ratio: Cash flow to total debt = (net income + depreciation) ÷ total debt; Lower values signal higher failure risk. Beaver tested one ratio at a time.
Altman Z-score (original, listed manufacturers): Z = 1.2X1 + 1.4X2 + 3.3X3 + 0.6X4 + 1.0X5; Zones: above 2.99 safe, 1.81 to 2.99 grey, below 1.81 distress. Use only if the question supplies or asks for it.
Argenti A-score logic: Defects + Mistakes + Symptoms = A-score; Qualitative marks are assigned to each part. A higher total means higher risk. Use the cut-offs given in the question or in your study material.
Common warning ratios: Current ratio = current assets ÷ current liabilities; Interest cover = EBIT ÷ interest; Gearing = debt ÷ (debt + equity); Judge by direction over several years, not one figure.

How to solve Other Failure Prediction Models and Ratio Analysis questions

Use this method for numerical and descriptive questions on failure prediction.

1Identify which model the question asks for: univariate (Beaver), multivariate (Altman, Springate, Taffler) or qualitative (Argenti).
2Write the formula with the weights given in the question. Do not rely on memory if data or weights are supplied.
3Extract the inputs from the balance sheet and profit statement. Compute each ratio separately and show it.
4Multiply by the weights, add up the score and compare it with the cut-off.
5State the classification in one line: likely to fail, grey or safe.
6Support it with ratio trends such as liquidity, gearing and interest cover.
7Add a short caveat on limitations, then give a recommendation, such as restructuring, cost control or closer monitoring.

Quickest way: Ratio table, score, verdict

When to use it: Use when a numerical question gives financial data and asks whether the company is likely to fail.

Draw a small table with each ratio and its weight.
Compute the ratio to three decimals and the weighted value.
Sum the values and compare with the cut-off.
Write the verdict and one reason from the weakest ratio.
If the question is qualitative, list defects, mistakes and symptoms in three bullets and link them as a chain.

Common mistakes in Other Failure Prediction Models and Ratio Analysis

Mixing up univariate and multivariate models.
All of them use ratios, so they look alike.
Fix: Remember: Beaver takes one ratio at a time. Altman, Springate and Taffler combine ratios into a score.
Calling Argenti's A-score a ratio model.
Students assume every failure model is numerical.
Fix: Argenti is qualitative. It scores management defects, mistakes and symptoms.
Using wrong inputs, such as total assets for current liabilities.
Rushing through the ratio definitions.
Fix: Write each ratio's numerator and denominator before calculating.
Treating the score as a certain prediction.
A cut-off makes the result look final.
Fix: Say the company is classified as likely to fail or safe, then note limitations and the need for further analysis.
Listing limitations without specifics.
Students write generic points.
Fix: Name them: models are built on past samples of specific industries and countries, rely on historical accounting data open to window dressing, ignore qualitative factors, and cut-offs may not suit Indian firms.
Judging from a single year.
Only one year of data is given.
Fix: Comment that trends over several years give stronger signals, and say what extra data you would want.

Worked examples

Example 1

A company has total assets ₹100 crore, working capital ₹10 crore, EBIT ₹8 crore, profit before tax ₹5 crore, current liabilities ₹40 crore and sales ₹120 crore. Using the Springate model S = 1.03A + 3.07B + 0.66C + 0.4D with cut-off 0.862, classify the company.

Show the solution

A = 10 ÷ 100 = 0.10, so 1.03 × 0.10 = 0.103.
B = 8 ÷ 100 = 0.08, so 3.07 × 0.08 = 0.2456.
C = 5 ÷ 40 = 0.125, so 0.66 × 0.125 = 0.0825.
D = 120 ÷ 100 = 1.20, so 0.4 × 1.20 = 0.48.
S = 0.103 + 0.2456 + 0.0825 + 0.48 = 0.9111.
Compare: 0.9111 is above 0.862.

Answer: S = 0.911 (rounded), above the cut-off of 0.862, so the company is classified as not likely to fail. The margin is small, so monitor profitability and liquidity closely.

Example 2

Explain how Beaver's univariate approach differs from Argenti's A-score, and state two limitations of using failure prediction models.

Show the solution

Beaver: compares individual ratios of failed and non-failed firms. Each ratio is tested alone, and cash flow to total debt was found the best predictor.
Argenti: scores qualitative factors in three groups, defects, mistakes and symptoms, and a high total warns of failure.
Difference: Beaver is quantitative and single-ratio, based on published accounts. Argenti is qualitative and looks at management and causes.
Limitation 1: models rely on historical accounting data, which may be manipulated and is backward looking.
Limitation 2: weights and cut-offs come from specific samples, so they may not fit other industries, periods or countries, and a single ratio or score ignores qualitative factors.

Answer: Beaver tests one ratio at a time using financial data. Argenti's A-score is a qualitative scoring of management defects, mistakes and symptoms. Both are limited by their sample basis and historical or subjective inputs, so they should support judgement, not replace it.

Exam tips

For MCQs, match the name to the type: Beaver is univariate, Altman, Springate and Taffler are multivariate, Argenti is qualitative.
In numerical questions, use only the weights and cut-offs printed in the question. Show each ratio line by line to earn method marks.
Always end with a verdict and a one-line recommendation, as written answers are marked for application.
Keep a ready list of four limitations to add to any descriptive answer.
Link this topic with the Altman Z-score page, since questions often ask you to compare the models.

Practice questions from Corporate Failure

Other Failure Prediction Models and Ratio Analysis: frequently asked questions

What is Argenti's A-score model?

It is a qualitative model that scores management defects, mistakes and symptoms of failure. A higher total suggests a higher chance of failure. It focuses on causes in management, not only on financial ratios.

What did Beaver's model show?

Beaver's univariate study compared single ratios of failed and non-failed firms. He found that cash flow to total debt was the best single predictor. The difference between the two groups widened as failure got closer.

What are the limitations of the Altman Z-score model?

It uses historical accounting data, which can be manipulated. Its weights come from a sample of US manufacturers in a particular period, so they may not suit other industries or countries. It ignores qualitative factors, and the grey zone gives no clear answer.

Which financial ratios are used to predict corporate failure?

Common ones are working capital, profitability, cash flow and leverage ratios, such as working capital to total assets, EBIT to total assets, cash flow to total debt and interest cover. The exact set depends on the model.