Dr Edward Altman has an international reputation as an expert on corporate bankruptcy, high yield bonds, distressed debt and credit risk analysis. His Z-score model for predicting bond default and recovery rates is widely used throughout the finance industry. Dr Altman has been a consultant to several government agencies, major financial and accounting institutions and industrial companies and has testified before the US Congress and the New York State Senate. In an interview with FinanceAsia, he explains the logic behind the Z-score model and how recent events have made it more popular as a predictive tool.
Briefly explain the idea and logic behind your Z-Score bankruptcy-prediction model.
The Z-score is a fundamental financial indicator model which I started to develop 40-years ago for my doctoral thesis at UCLA. It is a multi-variate model which contains five inputs, each objectively weighted by a statistical programme, that help predict the health of a company. These are: liquidity, solvency, profitability, leverage and activity (see below for formula). It was originally devised for publically-quoted US manufacturing firms, and has proven accurate 80%-90% of the time for predicting corporate bankruptcies at least one year in advance.
So, step-by-step, how is the model constructed?
It goes in three stages. First, I built the model and then back tested it. This showed that a Z-score of less than 1.8 is a reliable predictor of distress in a company. Second, I look at the average Z-score for firms within each bond rating category, from A down to D, in order to get a bond rating equivalent. And then finally, I consult a table - also used by rating agencies - that shows the cumulative default probabilities for each rating category and use that to determine a mortality rate for new bond issues. The model is clearly a leading indicator, for example when it suggested General Motors should be downgraded to sub-investment status two years before the major rating agencies actually did so. Investors should also follow the three-step process.
How widely has it been adopted by credit market specialists?
It is very widely used. It's simple, accurate and free. The basic model is available in software packages and is used by media and financial data provides such as Bloomberg and Standard & Poor's. But, additionally, proprietary models have been created by and for private customers - for instance, banks and hedge funds - containing additional variables. Most big banks have their own versions, sometimes preferring to use local data. In 1995, I built one specifically for analyzing both manufacturing and non-manufacturing companies in emerging markets which has proved to be a successful forecaster in Latin America and elsewhere, as well as private companies and leveraged buy-outs. The one variable it doesn't contain is the "activity" measure, and the remaining variables were recalibrated.
Why didn't Z-scores predict the huge problems in banks and other financial institutions?
Simply because the Z-score model cannot, and nor can variations of it, be used in a predictive way for financial companies. The main problem is that banks are particularly vulnerable to macroeconomic external shocks, including as we've seen, the arbitrariness of governmental support. Another example is liquidity, which although critical for any financial enterprise, is more often than not subject to system-wide shocks rather than any intrinsic problem within an individual bank. On the other hand, we have been able to use the four-variable model with companies such as Ford and General Motors which combine a core manufacturing business with large financial services subsidiaries - as I have shown in recent testimonies to the US Congress.
How is the model linked to the credit default swap (CDS) market? Are they complementary or alternatives?
Whereas the Z-score model updates its five variables every quarter, the CDS market is continuous, and updates instantly to new information. On the other hand, it is more volatile and can easily get distracted by "noise".
Was the model used by market practitioners for creating and investing in CDOs and other credit-linked structured products?
I wish it had been. Unfortunately, bankers just used standard bond ratings without examining the bonds closely enough and relied on simple historic average default and recovery rates. Rating agencies did a better job understanding the risks inherent in corporate CDOs than consumer CLOs, such as subprime mortgages.
Have recent events in the credit markets put the model under particular stress and, if so, is the model still valid?
Because the model predicted so many credit events so far in advance, it is if anything, more popular than ever.
What is your model predicting now? Which industries are especially distressed and are likely to become more so?
We've just witnessed the worst year for distressed investment, with the defaulted bond and loan index falling 45% in 2008. Especially damaged sectors include retail and consumer durables in general which are susceptible to macro-factors, and most of all those that were subject to leveraged buy-outs between 2003 and 2007. More than 50 US LBOs are currently bankrupt. And coincidently - or not - the model forecast a 4.64% default rate for US junk bonds for last year, and the actual rate is looking like 4.6%-4.7%! But it's likely to get worse.
When do you think debt markets will recover? What conditions need to be satisfied before they do?
A multiple regression analysis shows a strong correlation between corporate bond yield spreads and default rates a year later, so the default rate in 2009 could be as high as 11%-to-15%. The best time to buy distressed securities is after default rates have peaked, so the best opportunities probably won't arrive until later this year or even 2010.
How useful are Z-scores for analyzing Asian debt?
I've never believed in the de-coupling arguments. Where the US leads, Asia - and the rest of the world - follows. Of course, there are sometimes data collection and reliability problems for analysts of Asian companies to overcome due to opacity or auditing shortcomings. But, as we saw with Enron, Worldcom and others, those difficulties also arise in the US.
Altman Z-Score Bankruptcy Model
Z = 1.2X1 + 1.4X2 + 3.3X3 + .6X4 + .999X5
X1 (liquidity) = Current Assets - Current Liabilities Total Assets
X2(solvency) = Retained Earnings
Total Assets
X3 (profitability) = Earnings Before Interest and Taxes
Total Assets
X4(leverage) = Market Value of Equity
Total Liabilities
X5 (activity) = Sales (no. of times, e.g., 2.0x)
Total Assets
Professor Altman is the Max L. Heine Professor of Finance at the Stern School of Business, New York University and Director of Research in Credit and Debt Markets at the NYU Salomon Center for the Study of Financial Institutions. Among many awards earned during the last 25 years, Dr. Altman was named Laureate 1984 by the Hautes Etudes Commerciales Foundation in Paris for his accumulated works on corporate distress prediction models and procedures for firm financial rehabilitation and awarded the Graham & Dodd Scroll for 1985 by the Financial Analysts Federation for his work on default rates on high yield corporate debt and was inducted into the Fixed Income Analysts Society Hall of Fame in 2001, made President of the Financial Management Association (2003) and a FMA Fellow in 2004. He has also published or edited two-dozen books and over 130 articles in scholarly finance, accounting and economic journals.
Dr. Altman will be appearing in Hong Kong at FinanceAsia's upcoming "Distressed and Troubled Asset Investing Summit" this 28-29 April. For more info, please visit: www.financeasia.com/distressed
