Swedroe: Default Risk Doesn’t Pay

There are many well-documented anomalies in finance. Among them is the surprisingly small return that investors historically have earned for taking credit risk in fixed-income markets—the default premium, as measured by the difference in returns between long-term Treasurys and long-term corporate bonds, has been only about 0.3%—and that stocks with a higher risk of defaulting on debt have produced lower returns.

Chris Godfrey and Chris Brooks—authors of a September 2015 paper, “The Negative Credit Risk Premium Puzzle: A Limits to Arbitrage Story”—contribute to the literature and our understanding of why this anomaly persists.

Reasons Behind An Anomaly

The explanation lies not only in investor behavior—specifically, a preference for lottery-ticketlike positive skewness (the small probability of winning big)—but also in the well-documented limits to arbitrage, which prevent arbitrageurs from correcting mispricings.

Stocks with high credit risk tend to be stocks with the following characteristics: high idiosyncratic volatility, high illiquidity, wide bid/ask spreads and low turnover. These traits hinder the ability of arbitrageurs to correct overpricing because they increase costs and create incremental risks. For example, if a stock is illiquid with low turnover, arbitrage is more difficult, because these properties increase the time it takes for an arbitrageur to enter or exit a position in size.

It’s important to understand that arbitrage is more likely to be constrained on the side of the trade requiring a short sale. This occurs because selling a share short requires that the share be borrowed from a willing counterparty, and the facility for stock lending in sufficient size could be limited. “Lottery ticket” stocks tend to be under-owned by institutional investors, who are the traditional lenders of equities (which allows for shorting).

The limited supply, which drives up the cost of borrowing shares (the short-seller pays a borrowing fee), also limits the ability of arbitrageurs to correct mispricings. Thus, more optimistic investors are able to impact an asset’s price better than those with more pessimistic views. The net effect of the difficulty of shorting relative to a long position is that overpricing is more prevalent. Thus, stocks can remain persistently overpriced, and therefore suffer low returns.

The Disposition Effect

Godfrey and Brooks provide another insight into how overpricing can develop from what is known as the “disposition effect.” They explain: “Disposition investors are those subject to the behavioural bias of the disposition effect …. One of the central predictions of the disposition effect is that such investors will tend to ride losses but sell gains preferentially …. Where disposition investors own a Winner stock, they will tend to increase selling pressure in the stock, as they attempt to lock in a sure gain; conversely, where they own a Loser stock, they will tend to decrease selling pressure in the stock, as they preferentially retain the stock, hoping to ride out their losses. Where limits-to-arbitrage factors frustrate the actions of arbitrageurs in correcting such overpricing, we should expect that disposition investors will cause Loser stocks to become overpriced, and that this overpricing will increase with the proportion the stock’s owned by disposition investors.”

They then suggest that “one reason high credit risk Loser stocks become overpriced in the first place is that high credit risk stocks are owned disproportionately by disposition investors.”

The authors also note that the disposition effect can reduce liquidity in loser stocks and increase trading costs, making it more difficult for arbitrageurs to correct mispricings. In their study, Godfrey and Brooks present evidence showing that high credit risk stocks tend to be disproportionately owned by disposition investors. And disposition investors tend to be individuals, not institutions.

Research has found that institutional investors have a tendency to divest stocks that undergo declines in creditworthiness. Since individual investors are more prone to exhibit the disposition effect, the implication is that high credit risk stocks will be held disproportionately by disposition investors. Importantly, the disposition effect can help explain negative momentum, as disposition investors only slowly react to negative information.

Findings from the study “The Disposition Effect and Underreaction to News” by Andrea Frazzini, published in the August 2006 issue of The Journal of Finance, provide supporting evidence. Frazzini found that “bad (good) news travels slowly among stocks with large unrealized capital losses (gains), generating large subsequent returns for the overhang spread portfolio. Post-event returns of the negative overhang spread are not significantly different from zero. When negative news hits securities trading at large paper losses, it generates a severe post-event drift. Similarly, subsequent returns are large for positive news stocks trading at large gains.” In other words, the disposition effect helps explain momentum in stocks.

Limits To Arbitrage

Clifford Asness, Andrea Frazzini and Lasse Pedersen—authors of a 2012 paper, “Leverage Aversion and Risk Parity”—provide some valuable insights into the limits-of-arbitrage problem.

They write: “Assuming that some market participants are unable or unwilling to use leverage is not unrealistic. Leverage simply presents a risk that investors want to be compensated for bearing. Further, to obtain and manage leverage requires the acquisition of a certain ‘technology.’ Indeed, obtaining leverage requires getting financing, using derivatives, and establishing counterparty relations. Managing leverage requires, among other things, adjusting margin accounts and trading the portfolio dynamically over time, among other things. Our capital markets offer plenty of examples of investors that are not allowed (or choose not) to use leverage to increase their returns. For example, the majority of mutual funds and many pension funds are not allowed to borrow or are limited in the amount of leverage they can take. In addition, mutual fund families typically provide suggested asset allocations for low- to high-risk-tolerant investors. The high-risk recommendations rarely use leverage but, rather, suggest a very high concentration in equities.”

There’s another risk of shorting that should not be overlooked. While losses from being long are limited to 100% of your investment (assuming no leverage), losses from shorting are unlimited.

The Evidence

Godfrey and Brooks used the stock universe of all U.K.-domiciled stocks, active and dead, whose primary listing is or was on the London Stock Exchange Main Market or the Alternative Investment Market from June 30, 1987 through April 30, 2012. The authors chose the U.K. market because the U.K. has bankruptcy laws more favorable to creditors than those of the United States. Thus, U.K. equity holders typically expect to make a negligible recovery in bankruptcy resolution processes.

The pricing model they used included market beta, size, value and credit risk (using the well-known z-score, which measures the likelihood of default). They also examined the impact of illiquidity, turnover, momentum and bid/offer spreads. Following is a summary of their findings:

• High credit risk stocks are held disproportionately by disposition investors.
• High credit risk stocks have significantly higher levels of idiosyncratic risk than low credit risk stocks.
• Loser stocks have significantly higher idiosyncratic risk than winner stocks.
• The negative credit spread anomaly turns out to be a story about the apparently unexplained underperformance of high credit risk, extreme loser stocks.
• High credit risk stocks have higher levels of illiquidity than low credit risk stocks, and loser stocks are significantly more illiquid than winner stocks.
• High credit risk stocks have significantly lower turnover than low credit risk stocks.
• High credit risk stocks have significantly wider spreads than low credit risk stocks and loser stocks have significantly wider spreads than winner stocks.
• The measure of illiquidity is significantly negatively priced, with less illiquid stocks earning higher ex-post returns than more illiquid stocks. That’s in line with what would be expected if it were acting as a limit to arbitrage, preventing overvalued, illiquid stocks from being shorted down to fair value. Similarly, the bid/ask spread has a significantly negative price. Stocks with wider spreads have lower returns than stocks with narrower spreads. And, in line with the limits-to-arbitrage story, high-turnover stocks earn higher returns than low-turnover stocks.
• The book-to-market premium is most significant where limits-to-arbitrage factors are more severe.
• The negative credit spread is only significant where arbitrage is most constrained. In other words, the negative credit spread is only significant for the decile of widest credit risk, lowest turnover, highest illiquidity and smallest size.

Another interesting finding of the study was that high credit risk stocks are not smaller than low credit stocks, showing that firms do not have high credit risk simply because they are small.

Summary

The negative cross-sectional pricing of credit risk in equities has been a persistent anomaly within the traditional asset pricing literature. The field of behavioral finance provides us with a helpful explanation for the anomaly—the disposition effect. And the evidence shows the risk and real-world costs of implementing arbitrage strategies that would prevent mispricings from occurring serve to create hurdles that allow for mispricings to persist.

Godfrey and Brooks summarize their findings this way: “Idiosyncratic volatility, illiquidity, turnover and average spread behave as a limits-to-arbitrage explanation would predict.” They concluded that their research provided “a parsimonious explanation of the outstanding anomaly of the negative pricing of credit risk among equities.”

Their conclusion follows closely the conclusions from the aforementioned paper, “Leverage Aversion and Risk Parity.” It’s not only about leverage aversion, but also about the factors that make leverage harder and/or more expensive to implement.

For investors, the implications are striking. If you want or need higher expected returns from your portfolio, you shouldn’t go seeking those higher expected returns from the default premium. Instead, consider looking for ways to increase your exposure to the factors that show persistent and pervasive premiums.

This means the premium should be robust (they hold for various definitions) and have a logical explanation for why it should be expected to persist. The factors that meet these criteria are limited and include: beta, size, value, momentum, profitability/quality and term.

Larry Swedroe is the director of research for The BAM Alliance, a community of more than 140 independent registered investment advisors throughout the country.