Saturday 2 March 2013

Guest post - Long-Term Corporate Credit Returns

"The mind ought sometimes to be diverted that it may return to better thinking." -  Phaedrus

Please find below a great guest post from our good friend Paul Buigues, Head of Research at Rcube Global Macro Research. In this post Paul goes through the differences between equity and credit, as well as looking at the potential outcomes for long-term corporate credit returns:

Equity is an infinite claim on the free cash flows generated by a company. Due to the inherent uncertainty of future cash flows, relying on a pure valuation framework to predict equity returns is often a disappointing experience, for both specific companies and equity as an asset class.

Corporate debt, on the other hand, is generally a finite claim on a predetermined stream of cash flows (coupons + principal repayment). Broadly speaking, the range of potential outcomes for debt is binary: 
- Either the company is still alive at the debt’s maturity, in which case the investor is fully repaid the principal on top of having received coupons (or CDS premia).
- Or the company defaults or restructures at (or before) maturity, in which case the investor recovers a certain amount after the default/restructuring (e.g. around 40% on average for senior corporate bonds).

From this point of view, corporate credit is a much simpler bet than equity, especially for investors that hold credit instruments until maturity. We will call these investors (usually large institutions) static investors. They still constitute a large percentage of corporate credit investors, despite a general trend towards mark-to-market accounting. 
Calculating the P&L of a static investor is very straightforward. For example, if an investor sells protection on a credit index containing 100 issuers, with a one year maturity, he will: 
- Receive the CDS premium (S), and   
- Pay the difference between par and the recovery rate (1-R) for each default (i.e. 100 times P defaults, P being the default probability). 

When we equate both legs, we obtain the famous formula that links credit spreads, default probabilities and recovery rates: 

S = P (1-R)
S = credit spread, P = annualized default probability, R = recovery rate

Although this formula is only valid from a static investor’s point of view, thanks to the “law of one price” it can also be used to value credit as an asset class. This formula’s main interest is in the fact that, despite the numerous and complex issues surrounding corporate credit (subordination levels, bond callability, legal definition of credit events, etc.), it enables the easy translation of spreads into implied (or breakeven) default probabilities, and therefore makes spreads interpretable from a macroeconomic standpoint.

However, when we refer to credit as an asset class, we generally consider a rolling strategy that consists of using the proceeds of coupon and debt repayments to acquire newly issued credit instruments (this is how credit indices are built and most corporate credit funds are managed). 

These investors - which we will call rolling investors - generally have a mark-to-market viewpoint, and are therefore also sensitive to changes in credit spreads (i.e. changes in expectations of future defaults and recovery rates as well as changes in investors’ risk aversion). 

In the long term however - especially for static investors - only default rates matter, assuming stable recovery rates . Therefore, we will first look at the long-term history of default rates, focusing on US high-yield issuers, since they have the deepest history and are the most central to our discussion: 
Nota bene: Credit specialists might object that there is a negative correlation between default rates and recovery rates (see for example Altman & All, The Link between Default and Recovery Rates - September 2003). However, we lack proper long-term history on recovery rates, and will therefore assume stable recovery rates overall.

We notice that periods of high defaults tend to cluster together (in a very similar way to volatility). These spikes in default rates correspond to well-known periods of recession:

- The Great Depression during the 1930s
- The relatively mild recession of 1969-1970
- The early 1990s recession
- The tech crash of the early 2000s (before which the telecommunication-media-telecom sector represented up to 40 per cent of high yield indices) 
- The “Great Recession” between 2007 and 2009

However, it is worth noting that not all recessions have coincided with spikes in default rates. For example, between WWII and 1968 (during the ‘Trente glorieuses’), default rates remained very low, but there were five recessions according to the NBER count.

That being said, since credit investors generally own bonds or CDSs with a maturity of 5-10 years, we should look at cumulative prospective default rates of 5-10 years rather than 12-month trailing default rates.

Understandably, variations in annualized default rates are much milder when we look at longer horizons, topping at 9.4% for 5 year periods, and 6.1% for 10 year periods. Assuming a recovery rate of 40%, this translates into maximum annual losses of around 5.6% (= 9.4% * 0.6) when we consider 5 year periods, and 3.7% for 10 year periods.

Schematically speaking, as we’ve seen, credit investors capture the spread and endure subsequent default losses. If credit markets were highly efficient, credit spreads would therefore be priced in order to compensate for future default losses (plus a relatively stable risk premium).  

However, as we can see on the following chart, the relationship between spreads and realized forward 5 year losses is rather loose: 

If we analyse the data shown above, we can make at least three observations:

1) Since 1920, average high-yield spreads (5.02%) have been much higher than average annualized default losses (1.73%), implying an average risk premium of around 3.30% per year. Capturing this risk premium was actually the sales pitch used by Michael Miken in the early 1980s to promote high yield bonds. However, as we will see, this risk premium has receded since the mid-80s. 

2) The range of default losses is much more limited than the range of credit spreads (which went up to 20% in 1932 and 2008). Credit spreads clearly overshoot when times are bad.  

3) Finally, we can observe that there is no visible link between spreads and 5 yr forward default losses, especially when we consider the period since 1970. This implies that, despite their alleged high degree of sophistication, credit investors have a very weak predictive power on future default rates. This is confirmed by the following regression analysis:

Current spreads have virtually no correlation with actual future default losses. They are therefore driven by something else (risk aversion, greed/fear cycle). Corporate credit investors actually seem to care a lot about one thing: current (i.e. trailing 12-month) default rates.

We interpret this as evidence that credit investors are collectively subject to an extrapolation bias.
When default rates are high, credit investors behave as if default rates were going to stay high for the next 5-10 years. They liquidate their portfolios in panic (or because they are forced to do so). This snowball effect leads to spread levels that have no economic rationale. At the height of the latest credit crunch, corporate high-yield spreads were pricing a 33% annualized implied default probability over the following five years (20% / ( 1 – 0.40 )), which is around four times the maximum five year annualized default probability during the Great Depression!

Inversely, when default rates are low, credit investors believe that stability is the norm, and start piling up on leverage, inventing new instruments to do so (CLOs, CDOs, CPDOs etc.). This recklessness leads to malinvestment, and sows the seeds of the next credit crisis. 

Spreads moves between June 2007 and October 2008 (from 250bp to 2000bp in just 16 months) were a great illustration of this manic-depressive behaviour (which can also be related to Minsky’s model of the credit cycle).

Since spreads are weakly related to future default rates, we can hypothesize that a contrarian approach should work well for corporate credit instruments, at least from a long-term perspective.    

To check this assumption, we compare the spread of the BAML US High Yield Master II Index (H0A0) with the forward excess returns of the same index against 10 yr Treasuries. 

Nota bene: Using rolled CDS indexes (CDX, ITRAXX) would be an alternative data set. However, these indexes lack historical depth as they are only 10 years old).

Since 1986, this risk premium has averaged 1.55% per year, which is around half of the 3.30% risk premium observed since 1920 (as always, risk premiums have the nasty habit of diminishing once they are discovered and exploited).

Upon visual inspection, we can observe that there’s a rather strong link between current spreads and future HY excess returns vs. Treasuries:

This is confirmed by the following regression analysis (R2 = 0.47) 

This means that, even for a rolling investor (whose returns are also driven by mark-to-market spread moves), initial spreads explain nearly half of 5yr forward returns


Benjamin Graham’s famous allegory of a “Mr. Market” who alternates between periods of depression and euphoria applies especially well to corporate credit investors. In addition to having a bipolar disorder, corporate credit investors are afflicted by a severe case of myopia, as they focus on current default rates, rather than trying to estimate realistic future default rates. 

As a consequence, spreads themselves are a very good indicator of long-term forward returns, for both static and rolling investors. The last three decades gave investors three extraordinary opportunities to expose themselves to credit risk, as credit spreads reached levels that made no economic sense (we believe that average high-yield spreads above 1000bp are irrational, unless we’re on the verge of a global catastrophe that would wipe out the entire economy).

As of end of February 2013, the situation is obviously much less clear-cut. 

US high yield spreads (CDX HY) are currently around 435, close to European Crossover spreads (455). 

From our point of view, these spread levels are still attractive from a static valuation point of view, since they correspond to an annualized implied default probability of around 7% over the next 5 years.

These default rates have only occurred twice during the last 100 years:

From a valuation standpoint, we believe that there’s still at least a 50bp spread tightening potential for both US and EUR High Yield. 

When it comes to high yield bonds however (which are still the main instrument through which investors gain corporate credit exposure), many commentators lament on the low yields that are currently being offered (less than 6%). 
Indeed, anyone who has worked in corporate credit is understandably worried when he sees a single-B issuer like Tenet Healthcare being able to issue bonds with a coupon of 4.5%.

However, as in many debates, we believe this is just a question of vocabulary.
For us, credit is spread, not yield. A high-yield bond is a bet on the issuer’s creditworthiness combined with a bet on risk-free interest rates. Consequently, people who rely mostly on the low yield argument to justify their bearishness on high yield bonds should concentrate their hostility against Treasuries.

Since most institutional investors cannot avoid an exposure to bonds, we believe it still makes sense for them to capture spreads by going long on high yield bonds instead of Treasuries.

For absolute return investors, it is very easy to hedge the interest rate component of the bond (like we did last week), or to opt for pure credit instrument.  

To conclude, since we are tactical investors, we do not place too much weight on valuation considerations.  Unless we’re very far away from fair value, we see the valuation factor as a gentle breeze that can go either for or against us. Therefore, we would not hesitate to go against the breeze should our tactical indicators (especially lending standards) indicate we should do so. 

"An investment in knowledge pays the best interest." - Benjamin Franklin

Stay tuned!

No comments:

Post a Comment

View My Stats