Pensions and Bonds: The End of the Affair?

What You Need to Know

The role of bonds in pension systems is set to change. The large default allocation to bonds by pension funds has been a function of the large proportion of defined benefit (DB) funds and, until recently, quiescent inflation. We do not think either of these forces still holds. The combination of increased longevity, higher equilibrium inflation and lower growth rates implies that the strategic asset allocation of pension systems is likely to change.

To be clear, there is still a role for bonds, but via active fixed income or as part of longevity insurance, which is very different from large passive holdings of long-duration government bonds. We explore the options that pension systems have to adapt their strategic asset allocations in the face of a new investment regime.

Pension systems have built up huge allocations of government bonds. This has in part reflected the structure of pension systems (a historically large proportion of DB funds) and what has until recently been quiescent inflation. We argue that both of these forces have changed, that the share of pension assets in defined contribution (DC) plans will continue rising, and that the equilibrium level of inflation seems set to be elevated. The implication: pension systems in aggregate are or should be poised to make a significant allocation shift out of bonds and into other assets over the next decade.

There’s a stark conclusion in applying our forecasts for asset-class real returns to someone earning a median salary and paying 8% of it each year into a simple target-date structure that de-risked in the mid portion of their career before retiring at 65. That person, if they are early in their career, would face a “hardship outcome” below the minimum level deemed necessary for retirement. We have taken the title of this note from Greene. It seems appropriate to reflect the view expressed in his novel that a sense of unhappiness is much easier to convey than one of happiness. The intent of the structuring of retirement systems should, we argue, be minimizing the risk of unhappiness for many at the prospect of their life in retirement.

Putting aside for the moment the minutiae of optimal asset allocation, our overarching point is that the ability to offer (and indeed any expectation of receiving) guaranteed incomes is going away. This is, inter alia, the consequence of 100 years of improved life expectancy, as well as of the unwinding of a special set of macro conditions in the second half of the 20th century (mainly quiescent inflation and strong real growth). In addition, the fall in birthrates to below the replacement rate both lowers expected economic growth rates and makes it infeasible to attempt to transfer the cost of retirement to future generations (which would be morally questionable, anyway). The consequence of this shift is that nominal liability managers are in terminal decline. A combination of increased longevity, higher inflation and lower growth implies that a change in asset allocation is needed, including the option of buying longevity insurance.

Lower returns on equities, positive correlations with bonds, higher inflation and greater longevity force DC plans to make uncomfortable compromises. The options are: later retirement, higher contributions, lower retirement income or higher investment risk. There is another potential path for the system overall, albeit not for individual funds: to dump the risk onto later generations. Countries that don’t even attempt to fund retirement, such as Italy, do this already. However, this approach raises profound questions of intergenerational fairness and is doubly hard given shrinking working-age populations and that younger cohorts are less well off than older cohorts were at the same age. We will touch on these issues in the follow-on note on the politics of retirement, given the greater propensity of older, retired people to vote.

We want to be clear at the outset: we are not suggesting that pension funds should not hold any bonds; it’s more that their role in pension allocation is changing. There is a role for bonds within longevity insurance/pooling, but as we will show, this may become a limited role that requires a new type of bond. There is also an important role for liquidity (although overall liquidity needs are small). Strategically, earning a default “premium” is a potentially attractive element within a broad array of asset and factor risk premiums, and our view that investors must increase allocations to alpha in a low-return world leaves ample space for active fixed income. But in all of this, the overall pension exposure to passive longer-duration government bonds is going to be significantly attenuated.

If pension funds are set to have fewer government bonds, what asset classes are set to benefit? We think the main shift will be an increase in the strategic asset allocation (SAA) toward real assets (we include equities as a real asset). Private assets overall will likely see an increased allocation, too (in part reflecting both investor needs and the change in the source of marginal capital raising in the economy). There will likely also be increased allocations to strategies that seek to address longevity risk.

Many aspects of this issue have more to do with politics than with finance. Large pools of pension assets that seek to invest in real assets will be very tempting targets for politicians trying to influence or direct this capital to a long list of pet projects. These might include infrastructure, national champions, and environmental, social and environmental (ESG) goals. At the same time, any reduction in pension fund holdings of government bonds raises questions about both the absolute cost of borrowing and bond volatility, as the demand for shorter-term bonds leads to more frequent refinancing. The question is already asked: Who on earth is expected to buy the large amounts of debt set to be issued in the coming years?

There is another political aspect—where the risk sits for funding retirement. There has been a mass transfer of this risk onto individuals in recent decades. Arguably, this was acceptable when it was easy for individuals to buy simple and cheap products (passive 60:40 funds, say) that strongly outperformed inflation with a low level of risk. We argue that is no longer the case, which raises questions of whether the “social compact” implicit in the transfer of retirement risk onto individuals is still intact. As we discuss in this article, the choices for pension systems in aggregate are: force people to retire later, lower real payouts in retirement, dump the problem onto future generations (which would only work if mass migration were allowed to offset declining populations in developed economies) or allow the pension system to take more risk. Put in these stark terms, the latter is far easier politically than the other options. Expect treasury departments to lean on pension regulators accordingly.