[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries
.

Pensions and fixed-income investment

W hether bonds should make up a larger
proportion of pension scheme assets
has been a topical question ever since the Boots pension scheme moved £1.3bn of equities into bonds. In addition, the arrival of FRS17 has led to a growing awareness that pensions are like bonds in nature, and so the best match for them is, unsurprisingly, a portfolio of bonds or bond-like investments.
When looking at bond asset classes (government bonds, investment-grade corporate bonds, and high-yield corporate bonds) and other risky asset classes (I use equities as a proxy), it is helpful to consider the risk and return characteristics in the absence of liabilities. In terms of risk-adjusted returns, the asset classes are not greatly differentiated. However, the correlations between them are more interesting. Despite the fact that the three bond asset classes are all comparable in terms of risk and return (they all exhibit much lower risk and return than do equities), government bonds and investment-grade corporate debt are highly correlated with each other, whereas high-yield corporate debt is not highly correlated with any other asset class. This makes it a good diversifying asset in low-risk portfolios.

A match for the liabilities?
When liabilities are brought into the analysis, the minimum risk portfolio is even more dominated by government bonds (provided the liabilities are valued using government bonds), although the portfolio can vary by funding level and the method by which solvency is measured (cash surplus or funding level, for example).
It is possible to use historical returns from equities and bonds to construct efficient frontiers. Unsurprisingly, government bonds dominate the lower-risk portfolios, equities the higher-risk. However, a more interesting fact is that, if the time period under investigation is varied, the composition of the efficient frontier between these two extremes changes drastically: although the broad split between equities and bonds (corporate and government) remains reasonably stable, the divisions between the bond asset classes vary hugely. Perhaps the conclusion is that asset-liability modelling should be used only for the big decisions (ie the matching/non-matching split), while other decisions should be based on more pragmatic approaches.
The usefulness of the various bond asset classes can also be assessed in terms of the income they produce. Here, it can be seen that despite the volatility in capital values, the income produced by a portfolio of high-yield corporate debt is actually quite stable; indeed, it is comparable to the income from investment-grade corporate debt or government bonds. The government bonds appear to be particularly bad value when judged under this criterion.

Mismatching risks still remain
It is worth noting that investing in bonds without taking the details of the liabilities into account can leave substantial mismatch risks. The risks of currency and nature (fixed or index-linked) are broadly appreciated. Few UK pension schemes would believe themselves to be properly matched if they were invested in unhedged non-sterling bonds. The split between fixed-income and index-linked bonds is a fundamental decision that is normally well considered.
However, duration is not often given the attention it deserves, with the recommendation frequently being that the investment should be in a pooled fund benchmarked against a long-bond index. There will be a residual risk if the duration of the liabilities exceeds the duration of the bonds underlying the index chosen; given that the duration of, for example, the over-15-year gilt index is only around 13 years, this may well be the case. If it is also borne in mind that active bond managers rarely take duration bets of more than about a year, it should be clear there can be significant risks for immature pension schemes which get the duration of their bond portfolios wrong.
There are several alternatives to investing in these index-based pooled funds. The first is to hold a portfolio of bonds of the appropriate duration. However, there are only a few very long bonds in issue. A solution for large schemes is to control duration through the use of interest rate swaps, as some are beginning to do. However, this route is not open to smaller schemes. For them, having a range of duration-specific funds available would help, so that pension schemes could mix and match as appropriate. The funds themselves could use interest rate swaps as required.
Whether or not interest rate swaps become more widely used, the other factors outlined above mean all types of bonds will continue to attract interest from pension schemes for the foreseeable future.

03_11_03.pdf