Commentary: Why Bonds Are No Longer a Safe Harbor

By Heather Timmons

Bonds, especially investment-grade ones, are a refuge from the roiling stock market, right? Not so fast. Already, $20 billion in corporate bonds have defaulted in 2001--half of them rated investment grade not more than 12 months before they tanked. And defaults are going to get worse, say credit agencies and risk assessors.

The news comes at the tail end of a record quarter for the bond market. Corporations issued $150 billion in bonds since the start of 2001, nearly triple the volume of a year ago. Banks have significantly tightened lending standards in the past year, and equity markets have virtually dried up, leaving companies to rely on the bond market for cash.

Corporate bond buyers had better make sure they know what they're getting into, since things have gotten riskier. Just ask American Express Co. (AXP ) On Apr. 2, it said it will miss first-quarter earnings projections, largely because of a $185 million write-down to its junk-bond portfolio.

NO SAFEGUARDS. Holders of better rated bonds may find themselves in the same predicament. KMV LLC, a San Francisco risk assessment specialist, and rating agency Moody's Investors Services say investment-grade bonds are eight times more likely to default now than they were in 1997. Standard & Poor's Corp. (like BusinessWeek, part of The McGraw-Hill Companies) is expecting a rise in overall defaults, too. "Everything suggests that 2001, and perhaps 2002, will be worse than 2000," says S&P credit analyst Leo Brand.

The jump in investment-grade defaults predicted by KMV and Moody's is particularly troubling. Because these bonds aren't expected to go bad, they don't carry the safeguards that junk bonds do. The latter often require borrowers to have more collateral on hand or allow for early repayment if a company's performance slips. So when an investment-grade bond fails or sinks to junk status, investors often take bigger hits.

Rating agencies blame this year's high rate of defaults on so-called fallen angels--quality companies that ran into trouble but were expected to survive. Consider Southern California Edison, a public utility. Based on meetings with management, historical data, and business conditions in the power industry, Moody's, Fitch Inc., and S&P all rated SoCal's debt investment grade as recently as December. In February, SoCal defaulted on $6 billion after falling prey to the same cash crunch that hit other California utilities. "Rating agencies assumed that policymakers would intercede," says Mariarosa Verde, director of loan products at Fitch.

So far, predictive credit models have helped users duck some defaults that analysts at rating agencies missed. KMV, for example, has a quantitative credit model that aims to predict company defaults strictly by the numbers, using data such as stock prices, volatility, liabilities, and assets. In February, 1999, the KMV model showed that floor manufacturer Armstrong Holdings Inc. had slipped below junk status--though it was still rated investment grade by other agencies. In subsequent months, KMV's model showed that Armstrong's credit rating was slipping further. Ratings agencies didn't change their calls until July, 2000. On Dec. 6, a unit of the company filed for bankruptcy. Armstrong's bonds are now rated D, the lowest level.

Rating agencies argue that models like KMV's predict only short-term credit quality, not an investor's likelihood of being repaid in the long term. Nonetheless, Moody's has introduced a similar model, and Fitch is developing its own predictive models.

Bond specialists agree credit quality just isn't as predictable as it used to be. "The pace of the business cycle has made credit evaluation more challenging," says Thomas G. Maheras, a Citigroup (C ) vice-chairman and Salomon Smith Barney's head of global fixed income. If these predictive models are right, the future pace of defaults will provide an even bigger challenge.

Timmons covers banking and finance from New York.

    Before it's here, it's on the Bloomberg Terminal.