Corporate bond balancing act
Corporate bonds are the asset class of the moment. They are being touted as a panacea that will address many of the problems associated with pension fund investment. We look at the risks associated with different bond investment strategies and how that impacts on the solvency risks of a defined benefit (DB) pension fund.
There are probably three main catalysts for the heightened interest in corporate bonds.
The behaviour of equity markets over recent years has come as a reminder of the risks associated. Some commentators may perceive other assets such as corporate bonds or property as less risky.
The yield pick-up offered by corporate bonds over government bonds in a low interest rate environment can appear particularly attractive.
Chart 1 shows how US corporate bond yields have fared against US risk-free bond yields historically.
The choice of a credit-risky (AA) discount rate by the accountants for the purpose of calculating balance sheet liabilities has led some commentators to postulate that a credit-risky portfolio may actually be less risky than a government bond portfolio when we are considering an asset/liability example.
Of course, along with the apparent attractions of corporate bonds comes some fairly significant risks. These risks can be managed, but not removed, through careful portfolio construction. Further, when we are considering an asset/liability dynamic, such as that in a DB pension fund, we need to take into account not only the uncertainties of fund returns but also the way in which assets and liabilities relate. Let us first consider the management of fund return risk and then asset/liability risk.
Bond fund returns will vary over time under the influence of three main risk ‘drivers’:
1. Credit quality. Many of the assets held will have been issued by companies. The price and yield of corporate bonds is linked to the credit quality of the issuer. Generally, the lower the credit quality of the issuing company the higher the bond yield. Bond issuers can (and do) default. In this situation the bond investor may lose all future promised coupon and redemption payments or some part may be recovered. Whilst outright default is clearly a risk, for many bond investments the most common credit impact is when companies have their credit rating upgraded or downgraded. A rating change will change a bond’s price and this will impact on the price volatility of a corporate bond.
Bond fund managers are able to manage this risk by holding a diversified portfolio. Fund returns will depend on the level of concentration of assets. Consider two funds. One holds a single credit-risky bond, the other 100 credit-risky bonds. If we assume that credit events (default or credit rating change) are rare, then the first fund will only be affected by credit events very occasionally. However, when an upgrade or downgrade does take place, the effect on the fund will be large. By contrast, the second fund will frequently be affected by credit changes. There are, after all, lots of different issuers in the portfolio. However, the effect on the aggregate fund associated with each credit event will be less, because each position is smaller.
2. Interest rate sensitivity. Fixed income prices move in line with general changes in interest rates. The sensitivity of an individual bond will depend on the pattern of cash flows promised by the bond issuer to the bond investor. In general, bonds with long-dated cash flows will be more sensitive to interest rate movements than bonds with short-dated cash flows.
3. Credit spread risk. Corporate bond yields are higher than yields on similar default-free bonds. This difference – the bond’s spread – is determined by investors’ views of the risks associated with holding corporate bonds and the risk premium they demand for bearing those risks. The spread varies over time. In periods of (surprising) economic slowdown the risk of bonds defaulting or being downgraded generally increases for all bonds, resulting in a fall in the value of credit-risky bonds and a consequent increase in the spread. In periods of strong economic growth the opposite effect can generally be seen. This risk, as with interest rate sensitivity cannot be diversified away.
So, from the above, we can see that individual corporate bond funds could be very different depending on how much credit and interest rate risk is carried. Some are high yielding and very credit risky. Others are invested in long-dated assets – making them prone to changes in interest rates, whilst a few may adopt very high levels of concentration.
We can illustrate the scale and relative risks of different strategies by using a model. We could consider different credit quality portfolios, the risks created, and the impact of different diversification or rebalancing strategies.
Chart 2 looks at the spread of annualised returns we can expect, over a 10-year period, for different initial bond portfolios and rules. (We use the Barrie & Hibbert Stochastic Bond and Pension Fund Models, UK 2002 calibration. We show the 1st to 99th percentile simulations.) We have used bonds of a similar maturity and coupon and considered starting portfolios of differing credit quality. The rules are:
o Rule 1: Only hold two bonds (equally weighted) and don’t rebalance if they are downgraded or default.
o Rule 2: Hold 12 bonds (equally weighted) and don’t rebalance if they are downgraded or default.
o Rule 3: Hold two bonds (equally weighted) and rebalance to an A bond if any bonds are downgraded below this rating.
Interpreting this analysis we could derive that, on average and after allowing for downgrades and defaults, returns improve as we move down the credit curve. These rewards come with higher risks of course. A two-stock single A portfolio has a 1% chance of returning less than 0.7% pa compared to 3.5% pa for a government bond portfolio. On the upside there is a 1% chance that this corporate portfolio returns 7.0% pa and the government portfolio 6.3% pa.
More interesting is the effect that the rules on diversification and rebalancing have on the spread of returns. The impact of increasing holdings (Rule 2) and having a minimum credit exposure (Rule 3) appear fairly similar. Extending this, the implication is that for managers with different skills there are a number of effective structural solutions that could be overlaid to create similar risk profiles. So the stock-picker with a concentrated portfolio and a simple rebalancing rule can be compared to the diversified manager.
But what do these different choices mean when we are considering an asset/liability problem where the valuation of the liabilities themselves may be linked to prevailing financial market conditions? Let us use FRS 17 as an example. Under this accounting valuation for DB pension funds, liabilities are discounted at AA swap rates. So any emerging surpluses will depend on swap rates and fund values. They will also be sensitive to cashflow mismatches. This happens in pension funds because:
q We cannot truly predict liability cashflows. Changes in mortality experience for groups or individuals, salary inflation and pensioner choices are examples of factors that lead to uncertainties in predicting outgoings.
q Scarcity of matching assets. Even if we could exactly predict liability outgoings, creating matched assets would be difficult.
Using the same investment choices as earlier we have modelled FRS 17 surplus in 10 years’ time for a hypothetical pension fund where contributions are 25% of salary and the current FRS 17 liability is about £60m (e88m). Currently there is a small FRS 17 surplus. Chart 3 shows our results.
There are a few surprising and some counterintuitive results to consider. Again, there is an improvement, on average, as we move down the credit curve. Single A portfolios are, on average, about £2m more solvent than a government-backed portfolio. More broadly, the spread of outcomes is a lot less accentuated than was evidenced in chart 2. Specifically, however, AAA (Rule 2) and even A (Rule 2) look less risky than a government bond investment. How could this be?
To answer this we need to go back to one of the main drivers of fund returns in bond portfolios that was mentioned earlier – credit spread risk. Now think about the liability valuation. It is dependent on AA discount rates. If AA spread rates rise (and government bond yields remain the same), then liabilities fall – as does the value of corporate bonds and vice versa. Changes in AA credit spreads are normally a reflection of more general shifts in corporate yields. So, there will be correlated changes across the credit spectrum.
We can surmise that, when considering the emergence of FRS 17 surplus, corporate bond investments, of a certain nature, can have a dampening effect on volatility that will not be experienced through government bond investment. Of course, both government and corporate bond investments will expose the fund to interest rate risks associated with cashflow mismatch and corporate bonds will also be exposed to default and downgrade risks.
In the end, the choice of an appropriate corporate bond portfolio is a balancing act. The appropriate exposure to credit spread movements needs to be countered by an appropriate level of diversification or credit rebasing. However, in a world where liabilities and pricing are calculated on a credit-risky basis (eg FRS 17 and annuity portfolios), it may be important that the assets too are sensitive to similar dynamics.
Andrew Barrie is an executive director of consultants Barrie & Hibbert in Edinburgh