The credit risk premium: does it pay off?

Transcription

1 By: Van Luu, Ph.D., Senior Research Analyst, EMEA Lloyd Raynor, Senior Consultant, Consulting & Advisory Services, EMEA JANUARY 2011 The credit risk premium: does it pay off? In this paper we revisit the case for having a strategic allocation to credit. We conclude that the case remains valid but that investors should be mindful of the potentially long holding period required. We also conclude that the credit risk premium is impacted by perceived economic conditions, but that further indicators should be used to have confidence in an active timing strategy. Even before the financial crisis of 2007 to 2009, we were aware of the credit tilt of many active fixed income managers. In addition, pure global credit portfolios have become increasingly popular among our clients given the increased emphasis on using government bonds for liability hedging as opposed to return seeking purposes. Both of these facts only highlight the importance of understanding whether the credit risk premium, the premium of credit over government bonds, has actually paid off over the long term. The violent swings in credit excess returns experienced in the last three years have justifiably resulted in this subject receiving yet more attention. This paper provides a concrete answer to the question of whether a strategic allocation to credit has paid off for investors. We also address the question of timing exposure to the credit markets, and whether perceived economic conditions may be of relevance for an active timing strategy. We conclude that the credit risk premium is impacted by perceived economic conditions, but that further indicators should be used to have confidence in an active timing strategy. THE ARGUMENT FOR CREDIT Conceptually, a strategic allocation to credit risk, or more narrowly to corporate bonds, should be considered if the spread, on average, at least compensates for the default risk over government bonds. However, the literature on corporate bond pricing (for example Longstaff et al, 2005) has pointed to the existence of an additional compensation for holding riskier debt from liquidity, spread volatility and other risk premiums. If investors time horizons are sufficiently long, they can reap those premiums either by being persistently overweight the credit asset classes in their fixed income portfolios or via strategic allocations to credit. Russell Investments // The credit risk premium: does it pay off?

2 SCOPE OF THIS PAPER In this paper, we focus on whether a strategic allocation to corporate bonds passively yields excess returns over government bonds that are due to risk premiums embedded in the instruments in the long term. This question has been addressed in a past Russell research note by Tillberg (2005), using US BarCap index data from 1988 to This paper makes two contributions over and above updating previous studies of US credit returns. First, we proxy the excess returns to investment grade credit going back to 1926 by using the Moody s time series of Baarated corporate bond yields and defaults (see Appendix for details). This longer history enables us to have greater confidence in our estimates of the long-term risk premium of corporate bonds over Treasuries. Second, we examine the behaviour of credit excess returns through many economic cycles, using the OECD Composite Leading Index to divide economic activity into four re-occurring phases. This allows us to speculate on the best and worst periods to have a strategic allocation to credit, taking into account the difficulty of identifying the prevailing economic environment and transaction costs associated with altering credit exposures. THE CREDIT RISK PREMIUM Short-term track record The excess return of credit relative to government bonds needs to be positive over the long run to make a strategic allocation to the asset class a rational strategy. The first estimate of long-term credit excess returns we examine is the monthly time series of BarCap US Investment Grade Corporates starting in 1988 and finishing at the end of 2010, shown in Figure 1. Based on this relatively short sample the arithmetic mean annualised excess return of Investment Grade (IG) corporates in that period is 0.32% pa and the geometric mean is 0.24% pa. We proxy the excess returns to investment grade credit going back to this longer history enables us to have greater confidence in our estimates of the long-term risk premium of corporate bonds over Treasuries. Figure 1: US Corporate Investment Grade Excess return 8 6 Geometric Mean Annual Excess Return: 24 bps Arithmetic Mean Annual Excess Return 32 bps U.S. Corporate I t t G d Source: Barclays Capital For the equivalent BarCap High Yield excess return series, the arithmetic and geometric means are higher at 2.07% pa and 1.57% pa respectively. Given the more limited outperformance of IG corporates in particular, the short history of corporate bond excess returns based on BarCap may not be considered to lend strong support to a strategic allocation to credit. Russell Investments // The credit risk premium: does it pay off? / p 2

3 Figure 2: US Corporate High Yield Excess return 15 Geometric Mean Annual Excess Return: 157 bps Arithmetic Mean Annual Excess Return 207 bps U.S. Corporate High Yi ld Source: Barclays Capital Long-term track record Although less precise in its measurement, our proxy of excess returns before 1988 is desirable for extending the history significantly and helping us to gain greater confidence in the existence of a credit risk premium. As it turns out, the evidence is more strongly in favour of positive excess returns based on the longer sample. The arithmetic average of annual excess returns between 1926 and 2010, which are shown in Figure 3, is 1.42% pa, while the geometric average is 1.21% pa. The standard deviation of annual excess returns is substantial at over 650 basis points, resulting in an information ratio of around 0.2 and underlining the high risk associated with a strategic allocation to credit. Figure 3: Long excess return history of Baa Corporates p Geometric Mean Annual Excess Return: 121 bps Arithmetic Mean Annual Excess Return 142 bps The short history of corporate bond excess returns based on BarCap may not be considered to lend strong support to a strategic allocation to credit, but the evidence is more strongly in favour of positive excess returns based on the longer sample Source: Russell Investments Calculations, Barclays Capital Pay-offs are also heavily dependent on the sample period. The 1950s, 1960s, 1980s and 1990s were good decades for credit returns, while the 2000s were absolutely disastrous. Occasional and large negative returns from a permanent allocation raise the question of pay-off periods, i.e. the holding period historically required to ensure that excess returns are in positive territory. Russell Investments // The credit risk premium: does it pay off? / p 3

4 In Figure 4, we have plotted the trailing 10- and 20-year annualised (geometric mean) returns to illustrate this point. Before the most recent global crisis, 20-year excess returns never dipped below zero while 10-year returns were only slightly negative on two occasions in the 1970s. However, the financial market dislocation changed that dramatically. Both 10- and 20-year annualised returns declined well below zero, reaching -3% and -1.5% respectively at their lowest point. Even a 20-year holding period has historically been insufficient to guarantee positive excess returns for the unluckiest cohort of credit investors. Avoiding the worst periods for credit would thus certainly be worthwhile although it is far from certain whether one could identify those periods in real-time and implement a timely reduction in the strategic allocation to credit. In the next section, we turn to the issue of timing the allocation to credit based on economic conditions. Figure 4: Payoff periods of strategic allocations to Baa rated credit 5% 4% 3% 2% 1% 0% -1% -2% -3% Annualised 10Y excess Returns Annualised 20Y excess Returns -4% Even a 20-year holding period has historically been insufficient to guarantee positive excess returns for the unluckiest cohort of credit investors. Source: Russell Investments Calculations, Barclays Capital TIMING THE CREDIT RISK PREMIUM The style cycle Can we do better than have a constant strategic allocation to credit? In theory we could if we succeeded in forecasting the periods that deliver the best and worst excess returns and positioned the portfolio accordingly. While there are many ways to forecast credit excess returns, we use the concept of the style cycle which we have previously applied to equity factor returns (Luu and Sondhi, 2010). The basic idea is simple and is based on taking a leading indicator of economic activity, in our case the OECD Composite Leading Index for the United States, and dividing it up into four cycle phases (or regimes) according to the following rule: Expansion : CLI >100 and CLI rising vs. last month Slowdown : CLI>100 and CLI falling vs. last month Recession : CLI<100 and CLI falling vs. last month Recovery : CLI<100 and CLI rising vs. last month Russell Investments // The credit risk premium: does it pay off? / p 4

5 The OECD CLI has the nice property of being relatively smooth and phases tend to last for around 6 to 7 months on average, i.e. once a change from one phase to another is recorded one can be reasonably confident that the new regime is going to be in place for around half a year although there is certainly some variation around that mean duration, as shown in Table 1. Table 1: Duration of various cycle phases, Jan May 2010 (mths) Duration (since 1960) Expansion Slowdown Recession Recovery Median Mean Standard Deviation Max Min Source: Russell Investments, OECD Is there return diversification across the four cycles? For the cycle regime indicator to be useful in timing the allocation to credit, the mean excess returns to credit need to be sufficiently different across the four phases of the cycle. In what follows we study the excess returns of Baa-rated corporate bonds, constructed exactly as described earlier, but at a monthly frequency starting in 1961, matching the data available for the OECD CLI. The box plot in Figure 5 gives a first visual impression of the distribution of credit excess returns by cycle phase. Figure 5 1 Baa excess returns by economic cycle regime Expansion Slowdown Recovery Recession Source: Russell Investments, OECD The first observation to make is, perhaps unsurprisingly, that the variation of excess returns within a cycle regime is very high compared to the magnitude of mean excess return for each regime. This is important as it impacts on the statistical significance of the mean returns in each cycle regime. In short, this means that the averages for the Expansion and Recovery phases are significantly (in a statistical sense) positive while the means for Slowdown and Recession are indistinguishable The OECD Composite Leading Indicator has the nice property of being relatively smooth and phases tend to last for around 6 to 7 months on average. 1 The orange line inside the box represents the median, the blue dot the mean, and the box itself the first and third quartiles (middle 50 percent of the data). The difference between these quartiles is called the interquartile range (IQR), and the vertical lines outside the IQR represent a further 1.5x the IQR away from the median in each direction. Outliers are represented by circles and asterisks. The grey shaded region around the median displays approximate confidence intervals for the median (under certain restrictive statistical assumptions). The bounds of the shaded area are defined by the median +/- 1.57*IQR/n, where n is the number of observations. Shading is useful in comparing differences in medians; if the notches of two boxes do not overlap, then the medians are, roughly, significantly different at a 95% confidence level. Russell Investments // The credit risk premium: does it pay off? / p 5

6 from zero. Having a strategic allocation to credit during Expansions and Recoveries is therefore likely (over the very long term) to have outperformed compared to the same portfolio stance during Slowdowns and Recessions. Importantly, excess returns during Recession periods are also by far the most volatile, i.e. the range of possible outcomes is greater than for the other regimes (even when discarding the near -12% underperformance in one particular month). The obvious conclusion from these observations is to avoid having a strategic allocation to credit in Recessions in particular, since it has not had a positive pay-off in the long-term while being extremely volatile. Credit exposures during Slowdowns should probably be avoided as well, although the dispersion of excess returns is not nearly as great as during Recessions. In theory, if it were possible to forecast cycle regimes with any accuracy, strategic allocations to credit should be concentrated in Expansions and Recoveries, with Expansion benefitting from a lower volatility or excess returns and Recoveries from slightly higher mean excess returns. A visual inspection of the excess return indices for each cycle regime in Figure 6 reveals that, although tiny on a monthly basis, the differences in mean excess returns can be economically meaningful over very long-time horizons. Over a 50- year period, the difference between credit excess returns in the best-performing regime, Recovery, and the worst-performing, Recession, is around 75 percentage points. Figure 6 also shows that even in Expansions and Recoveries, there are quite a few episodes when the cumulative performance has been negative, despite the favourable pay-off in the long run. If it were possible to forecast cycle regimes with any accuracy, strategic allocations to credit should be concentrated in Expansions and Recoveries. Figure 6: Baa excess return indices by regime Recession Slowdown Recovery Expansion Source: Russell Investments, Moody s, Barclays Capital, OECD At the margins, knowledge of the prevailing economic regime and of the average historical behaviour of credit excess returns in those regimes may inform an investor s actions. In practice, to have stronger confidence in an active timing strategy, further indicators should be taken into consideration, which have been missing from the simple framework described here. Spreads and other measures of value have a role in complementing the cycle-based indicators and, last but not least, qualitative assessments of prospective credit excess returns from conversations with external managers and other market intelligence should be factored in as well. Russell Investments // The credit risk premium: does it pay off? / p 6

7 CONCLUSIONS In this paper, we have revisited the case for having a strategic allocation to credit based on the notion of a credit risk premium. The evidence from a shorter sample of Barclays Capital excess return indices arguably does not lend strong support to it, but a longer-term series of approximated excess returns of Baa-rated corporate bonds dating back to 1926 provides greater confidence in the existence of a credit risk premium. However, investors who implement a strategic allocation to credit should be aware of the potential for substantial tracking error of the credit risk premium (relative to government bonds) and the long pay-off periods required. When the most recent financial crisis was taken into account in our study, even a 20-year holding period did not ensure that investors reaped the benefit from a strategic allocation to credit although such an allocation will pay off in most 20-year periods. Credit has of course rebounded significantly following the financial crisis, with the Barclays Global Aggregate Credit Index 2 returning a cumulative 29.7% (GBP hedged) from 31 October 2008 to the end of Indeed, we urged investors to maintain their allocations to credit in our thought piece, Give credit where credit s due, of January We also argue that a relatively simple economic regime indicator based on the OECD CLI contains useful information about prospective credit excess returns and could in theory be used for deciding when to allocate to credit. However, we would strongly recommend that other indicators are used as well to have confidence in an active timing strategy. While mean excess returns in two of the four regimes, namely Expansions and Recoveries, are significantly positive and significantly different from those for the other two (Slowdowns and Recessions), the variability around those means is very large. At the margin, awareness of the prevailing economic regime may also aid in controlling risk, for example by tempering the credit weights during the Recession phase of the cycle. RELATED READING Brakebill, Keith (2010); US Fixed Income Beliefs: Challenged and Still Strong, Russell Investments IM&R Research Note Luu, Van; Kelly-Scholte, Sorca; Raynor, Lloyd (2009): Give credit where credit s due! Understanding current opportunities within credit. Russell Research Longstaff, Francis.; Mithal, Sanjay; Neis, Eric (2005): Corporate Yield Spreads: Default Risk or Liquidity? New Evidence from the Credit Default Swap Market, The Journal of Finance, Vol. 60, No. 5 (Oct., 2005), pp Tillberg, Brad (2005): The Role of Credit Tilt Strategies in US Corporate Bond Portfolios, Russell Investments IM&R Research Note 2 The Barclays Global Aggregate Credit Index only contains investment grade issues. Russell Investments // The credit risk premium: does it pay off? / p 7

8 APPENDIX - DATA The most precise measures of credit excess returns currently available are supplied by the fixed income index providers such as Barclays Capital or Citigroup. However, the available history for these excess return series versus government bonds is fairly short, extending back to 1988 for the BarCap investment grade indices for example. In order to study the long-term risk premium of an asset class, it is desirable to have a much longer history than 22 years. For the purpose of extending the history, we proxy excess returns of investment grade corporate bonds by using information from Moody s on corporate bond yields, defaults and credit losses, as well as data on 10- year Treasury bonds. This yields a series of annual and monthly excess returns dating back to We then splice the proxy series with the BarCap series for US Baa-rated corporates at the inception date of the BarCap series (June 1988). In approximating excess returns before June 1988, we calculate the total return of a hypothetical Baa-rated corporate bond with a 10-year maturity and a yield equal to the Moody s yield series. We subtract from that return the credit losses of Baa-rated bonds, which are also available from Moody s from 1982 onwards. Before 1982, we approximate credit losses by linear regression with the default rate as the explanatory variable. Finally, the total return of 10-year US Treasuries is deducted from this loss-adjusted total return of Baa-corporates to arrive at our excess return series. We compute an annual series starting in 1926 and a monthly series starting in 1961, the latter of which is used in conjunction with monthly economic data to investigate the timing of the credit bet. In addition to the constructed Baa-rated series, we also look at BarCap-provided excess return indices for investment grade and high yield corporate bonds, which both start in Our indicator of the economic cycle is the monthly OECD Composite Leading Index (CLI) for the United States, which starts in 1961 and is subject to a 2-month publication lag, i.e. the March value of the CLI will only become available in May. In back-testing our timing rules, we explicitly account for the publication lag. For more information: Call Russell Investments on +44 (0) visit Disclosures This material is not intended for distribution to retail clients. This material does not constitute an offer or invitation to anyone in any jurisdiction to invest in any Russell product or use any Russell services where such offer or invitation is not lawful, or in which the person making such offer or invitation is not qualified to do so, nor has it been prepared in connection with any such offer or invitation. Unless otherwise specified, Russell Investments is the source of all data. All information contained in this material is current at the time of issue and, to the best of our knowledge, accurate. Any opinion expressed is that of Russell Investments, is not a statement of fact, is subject to change and, unless it relates to a specified investment, does not constitute the regulated activity of advising on investments for the purposes of the Financial Services and Markets Act The value of investments and the income from them can fall as well as rise and is not guaranteed. You may not get back the amount originally invested. Any forecast, projection or target is indicative only and not guaranteed in any way. Any past performance figures are not necessarily a guide to future performance. Any reference to returns linked to currencies may increase or decrease as a result of currency fluctuations. Any references to tax treatments depend on the circumstances of the individual client and may be subject to change in the future. Issued by Russell Investments Limited. Company No Registered in England and Wales with registered office at: Rex House, 10 Regent Street, London SW1Y 4PE. Telephone Authorised and regulated by the Financial Services Authority, 25 The North Colonnade, Canary Wharf, London E14 5HS. Russell Investments // The credit risk premium: does it pay off? / p 8