Pensions and Bonds: The End of the Affair?

Inigo Fraser Jenkins:
Hello, I’m Inigo Fraser Jenkins.

David Hutchins:
And I’m David Hutchins.

Inigo Fraser Jenkins:
In this video, we make the case that the role of bonds and pension allocations is set to change. In part, this is because the proportion of pension assets run on a defined contribution basis will rise relative to defined benefit. But this is also due to larger macro forces, such as greater longevity and our expectation of the unwinding special conditions since the early 80s of high growth and quiescent inflation. The result is that the ability to offer, and any expectation of receiving, a guaranteed income is going away. Our conclusion is that pension systems are set to make a significant shift out of bonds. This is not to say there is no role for bonds, but a different one and in smaller size. Defined benefit plans are largely heading down the path of buyouts; the main action in changing demands of pension asset allocation is in defined contribution funds.
We have outlined in other research why we think that equilibrium inflation will remain at a higher level. This is a consequence of large forces, such as deglobalization, high levels of public debt, the cost of the energy transition and demographics. This leads to the prospect of lower returns and higher inflation. There is a stark conclusion for someone earning a median salary and paying 8% of that each year into a traditional target-date structure. If that person planned to retire at 65, they would likely face a hardship outcome. In the light of needing a given level of real returns, a position in nominal government bonds is not necessarily low risk.
In addition, we make the case that bonds may be less able to play a diversifying role. The negative stock-bond correlation in recent decades is relatively unusual in a longer-term context, and if there are extra sources of structural inflation that are not linked to growth, this could change the relationship between equities and bonds.

David Hutchins:
One of our key findings is that defined contribution plans will not fill the gap left by defined benefit plans and wealth managers in demand for long-dated bonds. This is despite the fact that it can be shown that longevity risk is an insurance problem and not a wealth-management problem for all but the wealthiest of retirees. But whilst insurance products like annuities demand long-dated bonds, they cannot be efficiently provided to retirees en masse without introducing other significant risks of contractual misalignment, misselling, inflation and investment inefficiency.
Whereas accumulation-defined contribution plans have been able to use default strategies to overcome these risks, the same approach applied to annuities actually amplifies them. Annuities are individual investment products whose return is based on the individual retiree’s characteristics, such as their lifestyle, dependence, wealth, health, occupation and required income shape.
Default annuities, where no individual engagement or underwriting is embedded in the process, can be shown to create socially unfair wealth transfer from the poorest 80% of society to the wealthiest. Even if we assume this risk can be mitigated, our research shows that annuities should only be purchased by retirees after age 75. In fact, this was a conclusion we first published over 15 years ago, back in 2009.
Whilst higher yields and bonds may have reduced its optimal age marginally, the combination of [the] high economic cost of buying bonds too early, the need for inflation protection and retiree needs for flexibility still [makes] that original conclusion sound. If this advice to annuitize at a later age was taken on board by all [defined contribution] DC  retirees, we estimate the demand for bonds, as well as their duration, would fall by over 30%. We also believed that annuities and, hence, long-dated bonds that embed an inflation uplift only in periods of high inflation, above a high watermark of say, 2.5%, would better match the shape of retiree needs. But these just simply don’t exist today. There are alternative types of insurance, such as guaranteed lifetime withdrawal benefit insurance, which our last paper on retirement income showed can efficiently overcome many of the objections I’ve raised, but they, too, work by looking to avoid long-dated bonds.
And finally, the whole case for so-called collective DC plans, as proposed in the UK today, is predicated on holding risky assets, not bonds, for longer than either [defined benefit] DB  or even well-managed DC can. So our conclusion is [that] DC demand for long-dated bonds is likely to be significantly lower than DB plans has been in recent times at a potentially high cost to governments in terms of their cost and stability of borrowing.

Inigo Fraser Jenkins:
What does all this mean for asset allocation? The changed macro circumstances and the need to protect purchasing power far into the future imply that DC pensions have to increase their allocation to real assets. We see equities as part of the list of real assets, even if our expected real return on equities is lower than the historic average.
In the note, we also discussed the range of other real assets that may be appropriate. In addition, it seems likely that retiring later is an inevitable conclusion in many cases. Not only does this mean there is a longer time to contribute, but it also means there is a longer time available to stay invested in risk assets when the scale of the investment pot is larger, as opposed to de-risking just when the pot size starts to reach a more significant level. We show in the note that staying invested in risk assets longer than was traditionally the case can materially increase the likely level of real returns available and is more attractive than taking leverage.
Overall, the options for retirement systems in general are raise retirement age, increase contributions, shift the burden onto future generations (and that is morally questionable) or let pensions take more risk. Only a government can adjudicate on these issues. Moreover, even in jurisdictions where pensions are funded, governments are ultimately on the hook if there’s a shortfall. That is why, ultimately, the design and possible risk levels in DC plans is a political question. In this light, and given the options available, letting pensions take more risk might be relatively attractive for regulators. However, the bond allocation of pension systems not only has implications for strategic asset allocation, it also raises questions about the financing of governments with high debt levels and high levels of issuance. Who will buy these bonds if pension systems require less? Thank you for your time.