How Credit Risk Is Factored in Corporate Bond Spreads

The financial markets have for long grappled with a disconnect between the high credit spreads of corporate bonds and the relatively low returns earned by corporate bond investors in recent decades, despite historically suffering only moderate default losses, on average. A recent Wharton paper cracks this puzzle, paving the way for a clearer perspective on the long-run performance of corporate bonds and the risks for investors who hold them.

The paper, titled “Reconstructing a Century of U.S. Corporate Bonds: Credit Risk in Historical Perspective,” is co-authored by finance professors Nikolai Roussanov of Wharton, Mohammad Ghaderi of the University of Kansas, and Sebastien Plante and Sang Byung Seo of the University of Wisconsin-Madison.

Investors in corporate bonds would expect bond prices and their returns to reflect two types of risks by benchmarking them against similar-tenured Treasury bonds, which are the safest around. One is the risk of default, for which they would expect a credit risk premium. The other is the risk stemming from interest rate fluctuations and captured by the duration of the bonds, which is compensated with a term premium.

A Disconnect in Credit Spreads

But that logic hasn’t held up consistently for investment-grade corporate bonds over different time periods, the paper found. Its analysis of corporate bond pricing data over 128 years — from 1895 to 2022 — showed that the credit risk premiums are insignificant and small for investment-grade bonds in recent history spanning the past 50 years until COVID, but they are significant over longer periods of 100 years or more.

The paper explains the puzzle of the apparent contradiction between the (large) credit spreads and the observed (low) returns on corporate bonds in the recent data (in excess of treasuries).

In theory, credit spread and the credit risk premium should track each other closely for investment-grade bonds, where expected default losses are very small, the paper pointed out. Yet the estimated credit risk premium exceeds the average credit spread in the first half (1947-1985) but falls well below it in the second half (1986-2022).

Over the long term, going back to 1947 (postwar sample) or even further back to 1926, the estimated credit risk premium is substantially larger and highly significant across all rating categories, the paper noted. Credit spreads include both credit risk premia and expected default losses on bonds.

That finding shows a clear and unbroken relationship between credit risk and its compensation, the paper noted. In fact, historical yield spreads between U.S. corporate and government bonds are large relative to bond defaults and losses, and they vary substantially over business cycles, showing a “sizable” premium for bearing credit risk, the paper continued.

The credit spreads in the long sample were too large to be just justified by historical default rates or losses, Roussanov said. He explained that the credit spread gap represents not just the default rates, but also compensation for the risk that investors bear.

“From the standpoint of an investor looking at historical returns and credit risk, corporate bonds are quite an attractive component of a portfolio.”— Nikolai Roussanov

The long-run sample showed that corporate bonds that are more exposed to stock and corporate bond market returns, as well as shocks to industrial production growth and inflation, earn substantially higher expected returns. “Crucially, this variation is driven almost entirely by the credit risk premium,” the paper pointed out.

Solving the Puzzle

The paper narrowed down the source of the puzzle in recent history, or in the last 50 years, to two factors. One is that the last 50 years is “highly unrepresentative” compared to the full historical sample period, which encompasses multiple cycles of widening and tightening of credit spreads. The effect of that in the last 50 years is that the estimated credit risk premium falls far below the average credit spread.

The other explanation is the role of a measurement bias in research literature relating to the callability feature in corporate bonds, which allows issuers to refinance their debt when interest rates decline. When these bonds are incorrectly matched to long-duration Treasuries, the effect is of inflating the estimated term premium, which understates the credit risk premium. This measurement bias is important in the post-1986 sample because of the persistent decline in interest rates in that period.

“The recent behavior of corporate bonds is unusual and a phenomenon that is mostly associated with the secular decline in interest rates since 1982,” Roussanov said. By contrast, bond investors earned significant amounts of credit risk premia in the earlier period, he added. Corporate bonds typically have lower tenures, or durations, compared to Treasuries, and their sensitivity to interest rates increases as their durations get longer, pushing payoffs farther into the future, he explained.

The Future of Bond Investing

“From the standpoint of an investor looking at historical returns and credit risk, corporate bonds are quite an attractive component of a portfolio,” Roussanov said. “Just based on that risk-return trade-off, corporate bonds are more attractive than government bonds. They occupy a potentially more important place in an investor’s portfolio than you would guess just by their fraction of total market capitalization.”

Roussanov noted that corporate bond investing is very different now from what it was 100 years ago or 50 years ago. The market now has bond ETFs (exchange-traded funds), and many investors trade in a basket of bonds at a portfolio level rather than trading in individual bonds. All of that improves liquidity.

“But the fundamental reasons for having a credit risk premium are still there because in a big economic crisis, corporate bonds will suffer defaults, and the risk of defaults will drive prices down and spreads up,” Roussanov said. “It happened during the great financial recession of 2008, and it could happen again. So, this is a risk that investors do bear and want to be compensated for.”

One important takeaway from the study is that credit spreads, on average, predict future bond returns, just as for stocks, where dividend yields and price-earnings ratios predict future returns.

While it is true that high credit spreads in bonds predict higher defaults in the future, in that setting, investors tend to also earn disproportionately higher returns. What actually happens is credit spreads spike more than warranted in response to fears of rising defaults, Roussanov explained. “The subsequent defaults are not large enough to eat away at that extra credit spread that investors earn, and so they earn higher returns.” Establishing this predictive relationship reliably requires data spanning a long period of time, which was not available until now.

The study’s findings could help the corporate bond market achieve better price discovery as well, “because more data is obviously better for training models,” Roussanov said. “We’ve had interest from various quantitative trading firms, and some of them act as market makers for bonds. Having more data will help potentially do that more efficiently.”

The paper empirically disentangles the credit risk premium and the term premium for corporate bonds by overcoming the limitations of long-run bond return data. It resolves that by showing that the credit risk premium is positive and economically significant.