This post is available as a PDF here.
Before we dive into this week’s commentary, we want to extend a very heartfelt thank you to everyone who nominated us for ETF.com’s 2016 ETF Strategist of the Year Award.
The award ceremony was held on Thursday night and we were fortunate enough not to leave empty handed!
We’re incredibly honored and humbled and would like to thank ETF.com, our partners, our advisor community, and all the readers of our research commentary for this distinction.
Now without further ado, back to the regularly scheduled programming!
- A fundamental tenet of investing is that excess risk requires excess reward.
- Over the last 15 years, investment grade corporate bonds have delivered little-to-no excess returns in comparison to safer U.S. Treasuries.
- The lack of out-performance has caused some to ask whether taking credit risk is “worth it.”
- We argue that in collecting risk premia, we act as insurers. If we do not adequately predict future risks, the premium we charge will be too little.
- The 2000s were an unprecedented period of risk and the fact that the ex-post realized credit premium was zero may simply be an indicator that the market underestimated risk, not that there is no risk premium to be gained in the future.
A fundamental tenet of finance is that greater risk requires the expectation of greater reward. Otherwise, why bother bearing excess risk at all?
In equities, this idea is rolled up into the concept of risk premia: factors that identify unique sources of risk that investors are compensated for bearing. The obvious example is the equity risk premium itself: the excess return offered to investors for investing in equities instead of safer assets like U.S. Treasuries. The value and size factors are also often considered to be risk premia. Value compensates investors for holding more distressed companies, while size compensates investors for holding less liquid companies.
In many ways, risk premia can be thought of as the premium an investor collects for offering an insurance service.
For example, in the case of value stocks, investors are looking to reduce their exposure to what are perceived to be distressed companies. To bear this extra risk, a value investor demands an excess reward.
If investors were risk neutral, then the premium demanded by a value investor would exactly compensate them for the expected loss they would incur from holding more distressed stocks that are more likely to go bankrupt.
Investors are not risk neutral, however: they are risk averse. A value investor will not just want to earn a premium that is expected to offset expected losses and, ultimately, leave them in the same place they were before. Rather, they will want to earn a premium that creates excess return relative to the broad market.
At least, that’s the narrative as to why value investors have historically earned an excess return premium to broad equity markets. Evident in hindsight and over the (very) long run, the exact expected premium is difficult-to-impossible to calculate in real time. Few are willing to look at the market today and say, “the insurance premium being offered for offloading your distressed companies is X%.”
Bonds, however, are different. With bonds, we can calculate the exact premium being offered for taking more risk by looking at the spreads between riskier and less risky issues.
For example, we can compare BAA coupon yields versus Treasury rates. The difference is the current premium demanded by investors for holding investment grade bonds. How this spread varies will be largely based on the risk appetite of investors as well as their forward expectations of default and recovery rates.
Source: Federal Reserve Bank of St. Louis. Calculations by Newfound Research.
Over the last 30 years, the average spread has been 235 basis points (“bps”). In the post-recovery period of the Great Recession (12/31/2009 and after), that spread has climbed to 278bps. This climb may be due to an increased perception of risk or an increased aversion to it among investors.
One might think, then, that an investor should have earned about a 2.35% premium investing in investment grade bonds, minus the cost of any defaults that occurred.
Plotting the iShares iBoxx $ Investment Grade Corporate Bond ETF (LQD) versus the iShares 7-10 Year U.S. Treasury ETF (“IEF”), however, gives a very different result.
Source: Yahoo! Finance. Calculations by Newfound Research.
While LQD technically beat IEF over the period, the average monthly excess return between the two is not significant from a statistical perspective. In other words, yes LQD beat IEF, but it can be explained entirely by short-term randomness and not long-term trend.
It is worth stopping and noting that this analysis is not perfectly apples-to-apples for a number of reasons.
- LQD holds a large number of A-rated bonds, which will have a smaller spread to Treasuries than the BAA-rated bonds.
- There are duration and convexity differences in the two products that will cause interest rate changes to potentially be a larger driver of return differences.
- Index turnover (due to defaults, new issuance, reinvestment, prepayment, et cetera) differences can make the path that interest rates and credit spreads take over time a large influencer on returns.
Nevertheless, such performance in recent years has caused some to ask whether stretching for the credit premium is worth it. Not only has it failed to compensate investors for the added risk, but it is also highly correlated with equity risk. We can see, for example, that in 2008 while IEF had a “pop” due to flight-to-safety behavior among investors, LQD sold off. Holding investment grade bonds within a portfolio, then, potentially sacrifices valuable diversification benefits from less-risky U.S. Treasuries.
If we take a step back, however, and consider our earlier definition of risk premium – the compensation we demand for bearing extra risk – we need to consider the role of risk in the equation. In acting as insurer, it is our responsibility to estimate the probability and magnitude of future losses and charge an adequate premium to cover those losses. If, for some reason, we under-estimate, then the premium we charge will be too little.
The 2000s were marked by significantly higher-than-average default rates among investment grade bonds. While this caused a rise in credit spreads, the “insurance premium” for those bonds that were defaulting had already been set in prior years.
Thought of another way, the fact that investment grade corporate bonds kept up with U.S. Treasuries during a period of time that saw the worst economic recession since the Great Depression may actually be an indicator that the credit premium is alive and well. Investors are notoriously poor predictors, so the fact that the premium charged was large enough to offset the unprecedented realized risk during this period could be viewed as impressive.
While spreads-minus-defaults is a commonly cited way of measuring the credit premium, it may ultimately be a flawed methodology as the excess credit returns investors realize do not correlate perfectly. However, a February 2015 paper titled The Credit Risk Premium, Asvanunt and Richardson (both from AQR) finds that after adjusting for term risk, the ex-post credit premium is positive and statistically significant.
Concluding Food for Thought
This analysis raises an interesting question: does the absence of ex-post excess returns necessarily imply the absence of a risk premium?
On the one hand, we can argue that the proof is in the eating of the pudding. Without past evidence of a realized excess return, it is hard to argue that it will exist in the future.
On the other hand, risk premium rely specifically on the market’s ability to estimate and price risk. If the market underestimates future risk for some reason, the premium earned will be consumed by losses.
The difference lies, we would argue, in whether the underestimation is serial or idiosyncratic. The former case implies that there is no structural compensation: the market pays you exactly enough to offset the increased risk you bear. The latter case simply implies a one-off, unexpected event happened and the market mispriced it.
The credit premium, for now, seems to fall in the latter case. The last 15 years may have been a rough patch.
“That’s some rough patch,” you may be saying. But consider that the value factor and size factor for equities have had similar rough patches, with drawdowns lasting 15.6 and 26.6 years respectively. Even the equity risk premium can struggle to materialize at times; remember that the 2000s were largely a lost decade for U.S. equities.
In fact, we’ve argued that these rough patches are necessary for the sustainability of the factor premia: if collecting the premia was seen as easy, more investors would seek to do it and the size of the premia would be driven down. Precisely because the premia is not guaranteed is why, over the long run, it can be a significant contributor to returns.