In 1968 a young professor at New York University asked himself whether a bankruptcy could be seen coming in the figures. Edward Altman compared 33 bankrupt companies with 33 healthy ones and found five ratios that made the difference. His formula, the Z-score, is still used almost sixty years later by banks, credit insurers and bankruptcy trustees. This article explains how it works, what the three zones mean and where the model gets it wrong.
The best-known case of such a distortion is the management company: little turnover, a lot of cash, and therefore a Z-score that wrongly lands in the distress zone.
What the Z-score measures
The Z-score is not an accounting figure but a statistical model. Altman searched for the combination of ratios that most sharply separated bankrupt companies from healthy ones, two years before the bankruptcy. The result is a weighted sum: each ratio gets a weight that indicates how much it contributes to the prediction.
There are three versions. The original from 1968 was meant for listed manufacturing companies and uses the market value of the shares. Because Belgian SMEs have no share price, Jaarrekening.be works with the 1983 version for unlisted companies, the Z'-score, which uses the book value of equity. A third version from 1995 drops the turnover ratio for service companies and companies in emerging markets.
The formula: five ratios, five weights
The Z-score for unlisted companies looks like this:
| Ratio | Calculation | Weight | What it measures |
|---|---|---|---|
| X1 | Working capital / balance sheet total | 0.717 | Liquidity: is there money left after the short-term debts? |
| X2 | Equity / balance sheet total | 0.847 | Buffer built up from retained profit |
| X3 | Operating profit (EBIT) / balance sheet total | 3.107 | Return on assets, the heaviest weight |
| X4 | Equity / total debts | 0.420 | Solvency: how much can the company lose before creditors are at risk? |
| X5 | Turnover / balance sheet total | 0.998 | How hard the assets work |
Two remarks for Belgian annual accounts. For X2 Altman used retained earnings, an item the Belgian balance sheet does not show separately. Jaarrekening.be therefore uses equity as an approximation. And an abbreviated format does not disclose turnover; the gross margin (code 9900) then serves as the substitute for X5. Which format a company files is explained in micro, abbreviated or full format.
The three zones
| Z-score | Zone | Meaning |
|---|---|---|
| Above 2.9 | Safe zone | The probability of bankruptcy within two years is small |
| 1.23 to 2.9 | Grey zone | Undecided. Look at the trend and the other signals |
| Below 1.23 | Distress zone | Companies in this zone statistically often go bankrupt within two years |
Altman reported that his model correctly identified about nine out of ten bankruptcies one year in advance. That figure comes from American manufacturing companies of fifty years ago; later studies on European SMEs come out lower, but the score remains one of the best simple predictors there is.
Worked example: a Belgian wholesaler
Take a wholesaler in building materials with these figures from the annual accounts (the codes refer to the full format):
| Item | Code | Amount |
|---|---|---|
| Balance sheet total | 20/58 | €2,000,000 |
| Current assets | 29/58 | €1,200,000 |
| Amounts payable within one year | 42/48 | €900,000 |
| Equity | 10/15 | €600,000 |
| Total debts | 17/49 | €1,400,000 |
| Operating profit (EBIT) | 9901 | €120,000 |
| Turnover | 70 | €3,000,000 |
The five ratios and their contribution:
- X1: working capital is 1,200,000 minus 900,000 = 300,000, divided by 2,000,000 = 0.15. Times 0.717 = 0.11
- X2: 600,000 / 2,000,000 = 0.30. Times 0.847 = 0.25
- X3: 120,000 / 2,000,000 = 0.06. Times 3.107 = 0.19
- X4: 600,000 / 1,400,000 = 0.43. Times 0.420 = 0.18
- X5: 3,000,000 / 2,000,000 = 1.50. Times 0.998 = 1.50
Together: Z = 2.23. The wholesaler is in the grey zone. Not alarming, but no reason to grant unlimited credit either. Note how heavily the turnover ratio weighs: without X5 the score would be 0.73. That is at once the model's main weakness.
Where the model goes wrong
The Z-score is designed for companies that have machines, inventories and turnover. In three situations it gives a distorted picture:
- Holdings and management companies. Many financial assets, little or no turnover: X5 is almost zero and the score lands in the distress zone while nothing is wrong. Why so many SMEs have such a structure is explained in what is a holding company.
- Real estate companies. A large balance sheet total full of buildings depresses all ratios. The score is low, but the collateral is real.
- Young growth companies. A scale-up with losses and a lot of cash scores badly on X2 and X3, while a recent capital increase actually shows confidence.
Conversely, a company with a fine Z-score can still collapse because of something that appears in no ratio: the loss of one major customer, a dispute with the tax authorities or a director who leaves. That is why you always combine the score with the five signs of an approaching bankruptcy.
Z-score or Health score?
Jaarrekening.be shows both on the company page, and they measure different things. The Health score weighs solvency, liquidity and profitability over three financial years into a score out of 5: it rewards stability and punishes fluctuations. The Z-score looks at one financial year but also takes turnover and operating profit into account.
The most powerful reading is the combination: a Health score that falls together with a Z-score that slides from the grey zone to the distress zone is a company where you had better limit your outstanding invoices. Two years in a row below 1.23 at a company with real turnover and inventories: that is the moment to deliver only against advance payment.
In short
- The Z-score combines five ratios from the balance sheet and income statement into one predictor of bankruptcy.
- Above 2.9 safe, below 1.23 distress, in between grey.
- Read the trend over three years, not one number.
- For holdings, real estate and services without turnover the score is unreliable.
- Look at it together with the Health score and the qualitative signals before you grant credit.