Few 40-year-olds can boast the ongoing success of the Z-score, a bankruptcy-prediction model introduced by New York University professor Edward Altman in 1968. The model, which assigns weighted values to five financial ratios to assess the likelihood of bankruptcy for manufacturing and industrial firms, has become an industry standard for detecting trouble before it happens.
The model’s five ratios, in order of importance, are EBIT/total assets, retained earnings/total assets, working capital/total assets, net sales/total assets, and market value of equity/total liabilities. Each ratio is multiplied by its assigned weight and the totals are added together for the Z-score, with higher scores indicating better financial health. Two variations of the model cover private companies, for which market value of equity must be estimated.
While banks, credit-rating agencies, and other professors have developed over 100 additional models over the years—in part to cover non-industrial firms in the service and financial sectors—the Z-score continues to be a cornerstone for its sectors, mostly in its original form.
What’s the secret of its success? “The Z-score is useful because it’s easy to do,” says Robert Tormey, a partner in Tatum LLC’s restructuring practice. That’s particularly true for public companies, where the necessary data is easily available. Further, the model “is based on an awful lot of real data, bank loans, which many others aren’t,” he adds.
The Z-score has also stayed relevant because Altman “has remained a thought leader in the space of credit risk and default,” Tormey says, noting, for example, that Altman flagged the looming problems with credit-default swaps early.
In fact, the Z-score may just be hitting its prime in the current credit crunch. Increasingly, auditors, customers, and vendors are all applying the model to their clients and partners to see how risky any given relationship is, says Tormey, adding that smart CFOs will “want to apply the Z-score to their own businesses to see just how overextended they are” before taking on a new client or embarking on an expansion. In fact, Joseph Calandro, a risk specialist who has used the tool in consulting both to firms and investors as well as a part-time finance professor at University of Connecticut, wrote a paper last year looking at how the Z-score could be used as a turnaround tool.
CFO.com recently spoke with Altman, now the Max L. Heine professor at NYU’s Stern School of Business, to find out what he’s done to help the Z-score maintain its youthful allure.
Is the Z-score still relevant?
The model would not be used as much as it is if it was not still fairly representative. I have recalibrated it with up-to-date data, but I haven’t published it because I sell that to hedge funds and other investors. (The old one is in the public domain and doesn’t get a dime—I’m not going to make that mistake again!) Still, it’s not necessarily going to be more accurate when used for any given company. I haven’t been convinced that the new models are any better than the old models.
Broadly speaking, what are some of the major changes you’ve made?
Well, the zones [how a Z-score translates into a prognosis for a company] are out of date; I’m the first to admit that. Now I use bond-rating equivalents of a score, so a Z-score above 5.5 is triple-A, for example, while a Z-score of 1.8, the old cut-off for the distressed zone, is equivalent to a B. You have to go down into a negative score to into the D range. It used to be that 0 to 1 was terrible.
Essentially, [determining the likelihood of bankruptcy is] a three-step approach: you calculate the Z-score, assign a bond-rating equivalent, and then you can assess the probability of default over one to 10 years. You plug the bond-rating equivalent into a matrix of probabilities of default for each category based on 40 years of data. It’s a mortality rate approach, which is economy-neutral. It doesn’t consider whether next year will be a recession or an expansion; it’s more of a smoothing. [Altman maintains a proprietary database of default rates and updates it annually.]
People can read more about it in my books and articles, including Corporate Financial Distress and Bankruptcy (Wiley, 2005) and Managing Credit Risk (Wiley, 2008).
I’ve also done some recalibrations to remove the importance of outliers [in the data set that underpins the model], mostly by doing logarithmic transformations.
Credit-rating agencies have come under a lot of fire recently for being irrelevant. What’s the benefit of this approach over the bond ratings the agencies assign?
The Z-score is unemotional, there are no human adjustments. It’s also a point-in-time measurement, so it doesn’t take into account how a company will perform over the cycle [as ratings agencies do], which makes it more volatile, and more likely to change every quarter as it’s updated for a company’s financial reports.
What do you foresee for 2009?
The number of bankruptcies is going to go up; how much higher it will be depends on how serious the recession is. In terms of high-yield bonds [and not based on the Z-score analysis], I’m expecting a default rate of 6-8 percent in the next 12 months, where the historical average is around 4 percent.