For educational purposes only; not investment advice. Investing may result in loss.
Direct answer
The Altman Z-Score is a financial-distress screening model that combines five accounting and market ratios into one score. The classic public manufacturing-company version is:
Z = 1.2X₁ + 1.4X₂ + 3.3X₃ + 0.6X₄ + 1.0X₅
where X₁ is working capital divided by total assets, X₂ is retained earnings divided by total assets, X₃ is EBIT divided by total assets, X₄ is market value of equity divided by book value of total liabilities, and X₅ is sales divided by total assets.
The score is useful for risk screening, not for certainty. Altman’s 1968 model was estimated with multiple discriminant analysis on 33 bankrupt and 33 non-bankrupt manufacturing firms, matched by industry and asset size within the study’s historical setting. Its in-sample and later test performance does not turn the score into a current bankruptcy probability or a forecast of the date or legal form of a filing. A low score does not prove bankruptcy, and a high score does not eliminate refinancing, liquidity, fraud, industry, accounting, or market risk.
What the five variables measure
X₁ = working capital / total assets captures a balance-sheet liquidity dimension. Working capital is current assets minus current liabilities. A negative value can point to funding pressure, although some business models operate with structurally low or negative working capital; that business-model explanation does not by itself establish safety.
X₂ = retained earnings / total assets captures cumulative profitability after losses and distributions relative to the asset base. It was associated with firm age in the original discussion, but retained earnings is not a direct age measure: dividends, reorganizations, accounting rules, acquisitions, and accumulated deficits also affect it.
X₃ = EBIT / total assets captures operating earning power before interest and taxes relative to assets. Its coefficient was estimated jointly with the other variables in the discriminant model; the numerical weight is not a causal claim and should not be interpreted in isolation.
X₄ = market value of equity / book value of total liabilities captures a market-value cushion relative to recorded obligations. Use the market capitalization for a stated date aligned as closely as practicable with the balance-sheet date, and document the share count and equity classes included. It can move quickly with price and dilution and may embed information already reflected in other inputs.
X₅ = sales / total assets captures asset turnover. It is highly sensitive to industry, revenue presentation, asset intensity, acquisitions, inflation, and accounting policy; a software company, manufacturer, utility, and retailer can have very different normal turnover.
For the classic listed-manufacturer formula, later presentations commonly label scores below 1.81 a distress zone, scores from 1.81 through 2.99 a gray zone, and scores above 2.99 a safer zone. These labels are not calibrated probabilities or guarantees, and boundary conventions can differ. They must not be transplanted to Z’, Z’’, vendor-modified, private-company, or emerging-market variants with different variables, coefficients, constants, or calibration samples.
Worked example
Assume a manufacturing company reports:
- current assets:
$600 million - current liabilities:
$400 million - retained earnings:
$300 million - EBIT:
$120 million - total assets:
$1.0 billion - market value of equity:
$800 million - total liabilities:
$500 million - annual sales:
$1.5 billion
Using one consistent currency and reporting perimeter, the variables are:
X₁ = ($600m - $400m) / $1,000m = 0.20
X₂ = $300m / $1,000m = 0.30
X₃ = $120m / $1,000m = 0.12
X₄ = $800m / $500m = 1.60
X₅ = $1,500m / $1,000m = 1.50
The score is:
Z = 1.2×0.20 + 1.4×0.30 + 3.3×0.12 + 0.6×1.60 + 1.0×1.50
Z = 0.24 + 0.42 + 0.396 + 0.96 + 1.50 = 3.516
Now suppose the stock price falls so the equity market value drops to $300 million and EBIT falls to $40 million. Then X₄ = 0.60 and X₃ = 0.04. Holding other inputs constant:
Z = 0.24 + 0.42 + 0.132 + 0.36 + 1.50 = 2.652
The score worsens because operating profit and the market-value cushion declined. The second calculation holds every other input fixed, so it is a sensitivity illustration rather than a forecast; in reality, cash, liabilities, sales, assets, and retained earnings may also change. A stock-price decline and lower EBIT can reflect some of the same new information, so the model should not be read as five independent causal signals. Debt maturity, cash access, covenants, industry conditions, and management actions still matter.
Limits and practical checks
- Use the correct model version. The classic formula was estimated for listed manufacturers; private, financial, utility, service, non-U.S., and later-period firms require validation or a deliberately selected alternative, not an automatic coefficient or cutoff swap.
- Define every input and reporting perimeter. Reconcile current assets, current liabilities, retained earnings, EBIT, total assets, total liabilities, sales, market capitalization, currency, units, fiscal period, discontinued operations, and restatements before comparing companies or dates.
- Check accounting quality. One-time gains, leases, revenue recognition, inventory write-downs, pensions, goodwill and other impairments, supplier finance, factoring, acquisitions, and off-balance-sheet commitments can affect interpretation or comparability.
- Do not ignore timing. Balance-sheet values are point-in-time, sales and EBIT cover a period, and market capitalization is observed on a market date. Avoid combining stale statements, trailing data, and a current share price without clearly labeling the mismatch.
- Market value is market-sensitive, not mathematically circular. A falling share price directly reduces
X₄; document whether the move reflects firm-specific information, broad market conditions, illiquidity, or share-count changes. - Compare with liquidity evidence: cash, revolver access, debt maturities, covenants, free cash flow, and credit ratings.
- Track a consistently calculated trend, not only one number. A falling series can be informative, but changes in model version, fiscal year, accounting policy, currency, units, or corporate perimeter can create a false trend.
- Combine the score with 10-K risk factors, MD&A, footnotes, and cash-flow analysis.
Common misconceptions
“The Z-Score predicts bankruptcy exactly.” It is a discriminant screening model, not a calibrated probability, legal conclusion, or forecast of an exact filing date.
“Higher is always better across every industry.” Asset turnover, working-capital structure, accounting, and the appropriate model version vary widely. Scores are not automatically comparable across unlike firms.
“A company with a low score is uninvestable.” Low scores flag distress risk. Securities can still have value, but the required return, downside risk, and capital-structure position become central.
“The formula replaces reading filings.” The variables come from filings and market data; they do not replace understanding the business, debt terms, or cash-flow path.
Related topics
Authoritative sources
- Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy - Journal of Finance (2026-08-07)
- How to Read a 10-K/10-Q - Investor.gov (2026-08-07)
- Beginners’ Guide to Financial Statements - SEC (2026-08-07)