Replies to Barr, van Ewijk, Heath, Karnein and Schokkaert

1. Risk Pooling

My commentators raise several challenges to my conception of a collective pension scheme as a multigenerational corporate body that efficiently pool risks across generations. ¹ I begin with Joseph Heath’s most fundamental challenge: his claim that ‘the intergenerational cooperative system does not actually pool risks (which leads me to register a mild complaint about the title of Otsuka’s book, for suggesting otherwise)’. Rather, this ‘retirement system’ generates returns on savings during people’s working years – i.e. the portion of their earnings they contribute to a pension fund – through the productive investment of those savings, the compounding returns of which they enjoy during their retirement years.

Heath is right to draw attention to the central role played by the productive investment of pensions contributions. I wish I had registered its importance in Otsuka (2023). Moreover, I am in agreement with him that the compounding of returns over time on the productive investment of pensions contributions is other than intergenerational risk pooling. ² Nevertheless, I would still maintain that collective pensions pool risks across generations. As I indicate in the last paragraph of Section 1 of ‘Risk pooling, reciprocity, and voluntary association’ (this volume), the creation of a corporate body that will endure indefinitely across multiple generations is what renders it safe to remain heavily invested in productive assets – both publicly traded equities (stocks and shares) and illiquid private equity (such as the infrastructure into which large pension schemes invest). Pension schemes are thereby able to fully reap the risk premium that equities enjoy over bonds, which is the higher expected return necessary to induce people to invest in the former rather than the latter, to compensate for its greater risk, to which they are averse. ³ Even if, as Heath maintains, the smoothing of variable returns on equities is unlikely to be over more than a generation, one still needs to create a multigenerational corporate body in order to make possible the persistent heavy investment in equities over which the returns are smoothed.

Casper van Ewijk and Nicholas Barr both discuss the recent experience of the Netherlands, which might be thought to cast doubt on the prospects of pooling risks across generations in practice rather than theory. However, many of the problems to which they point can be traced to Dutch adherence to valuations tethered to government bond yields, along lines of the ‘financial economics approach’ which I critique in Otsuka (2023: Ch. 2). On the contrasting ‘actuarial approach’ which I endorse, pensions liabilities are valued as the long-term expected return on a portfolio weighted heavily towards equities. ⁴ The funding crises into which collective pensions were plunged in the Netherlands and elsewhere – such as the UK, which I discuss in Otsuka (2023: Ch. 2) and Barr also discusses in his comments – were the result of a lengthy cratering of government bond yields between the global financial crisis of 2008–09 and the surge of inflation in 2022. By contrast, falls in the value of equities during that period were comparatively short-lived and swiftly recovered. A valuation based on returns on equities, with responsibility for making good any underfunding spread across a number of years, would have necessitated far less drastic rises in contributions or cuts in benefits. ⁵

I would, however, acknowledge the need, on such an equities-based approach to funding and valuation, to make pension benefits conditional on returns on equities being sufficient. Such conditionalization avoids what Barr describes as a flawed corner solution of a hard defined benefit pension promise in which all financial risk is borne by the sponsoring employer, and none by the members who receive a pension. With this better solution, both the inflation-indexed increases in pensions in payment and the revaluation of accrual during active members’ working years would be conditional on returns on investments in the pension fund being good enough over the long run. ⁶ Such conditional indexation and revaluation would make it possible to reap the premium of a greater level of investment in equities rather than inflation-index-linked bonds and other fixed income assets, since the pensions liability would no longer be so closely linked to a hardwired level of indexation and revaluation. Equities would provide a ‘match’ to the pensions liability by fitting the liability to the assets rather than the assets to the liability.

The approach to risk pooling across generations that I endorse requires a steady stream of incoming contributions from new members to the collective scheme who continue to replace retired members in roughly the same numbers. I agree with van Ewijk that one will need something along the lines of a centralized society-wide pension scheme, rather than a decentralized plurality of smaller schemes, to realize this. This needn’t be pay as you go (PAYG). It could instead be funded. In Otsuka (2023: Ch. 2), I maintain that the UK’s large, multi-employer Universities Superannuation Scheme (USS) offers a model for such a funded arrangement. The expansive reach and ‘last man standing’ mutuality of USS provides a prototype for the delivery of pensions across society more generally. Here the pooling of risks across geographic space would facilitate its pooling across succeeding generations as well.

Erik Schokkaert draws attention to a challenge which would remain in replenishing a collective scheme with newcomers even when its membership encompasses an entire society. He points to the difficulties posed by demographic shifts within many societies in the dependency ratio of pensioners to working-age individuals, arising from a low replacement rate of old by young. While there are no easy, lasting or fully adequate solutions to this problem, I think it worth mentioning that funded pensions now provide more promise than PAYG in managing such demographic changes. This is because they are able to address a decline in the size of the domestic workforce by the ‘importing’ of ‘labour’, not ‘directly, through immigration’, but rather ‘indirectly by exporting capital to countries with a young labour force’ (Barr 2020: 184), where this involves investment in shares of companies in such countries. Given the growth of the human population on a global level – which is forecast to rise from its current level of 8.2 billion to a peak of 10.3 billion in the mid-2080s before gradually declining to 10.2 billion by the end of the century ⁷ – countries with low replacement rates will be able, for several more decades, to tap into funding for the pensions of their elderly through the investment of assets in emerging markets with growing populations of workers. ⁸

2. Reciprocity

I maintain in Otsuka (2023: Ch. 4) that, so long as the state provides all workers with a pension that meets their basic needs for income in retirement, there is a sound reciprocity-based case for the mutually beneficial risk pooling of collective occupational pensions above that floor even if they mirror unchosen inequalities in lifetime earnings. I grant Schokkaert’s point that such an arrangement would fall short of the full realization of the luck egalitarian element of justice. But I would maintain that when everyone has enough so that nobody is in need, the demands of equality should not override unequal improvements for all above this floor. Rather, such an arrangement properly trades off two distinct and conflicting elements of justice: one of which is luck egalitarian and the other grounded in the mutual advantage of reciprocity, where each party voluntarily brings his pension contributions to the collective, risk pools these resources with the resources of others, and then gets back in proportion to what he puts in.

I argue that such risk pooling is a form of Rawlsian reciprocity in the sense of mutual advantage from a benchmark of equality. I maintain that, even if this benchmark does not involve an equal distribution of wealth and income, it can encompass agreements where all are treated as equals insofar as nobody exploits asymmetries in bargaining power arising from their unequal shares of resources. I thereby invoke what I take to be a recognizably and distinctively Rawlsian concept of reciprocity as fair terms of social cooperation for mutual advantage among parties who regard one another as equals. Anja Karnein’s insightful discussion reveals that there are nevertheless fundamental respects in which my interpretation and application of this concept departs from Rawls’s. I apply this concept to choices of individuals to join cooperative arrangements against an alternative of going it alone in society, whereas Rawls applies it to an assessment of different all-encompassing basic structures of society, where these are not contrasted with an alternative (e.g. a state of nature) involving non-cooperation. Moreover, on my understanding of mutual advantage as involving rational self-interest, I depart from a Rawlsian understanding which appeals to a notion of reasonable agreement involving the fair adjudication of the competing claims of different individuals. I accept Karnein’s judgement that I should therefore disclaim extensive fidelity to Rawls. Nevertheless, I would maintain that, however Rawlsian it may or may not be, my interpretation of reciprocity offers a clear, straightforward and intuitive concept which provides what I take to be the most compelling and uncontroversial case for collective pensions: the comparative wastefulness and inefficiency, to the detriment of each, of the alternative in which individuals go it alone. ⁹

3. Voluntary Association

I accept Schokkaert’s challenge to my claim that the known longevity risk of different individuals is roughly equal from the perspective of young adults at the beginnings of their working lives. Rather, those from less privileged socioeconomic backgrounds will be known to have a lower life expectancy. Moreover, this knowledge would provide them with reason to opt out of a collective occupational pension which provides an annual retirement income for life, in favour of the alternative of investment in an individual retirement savings account that provides them with a lump sum to draw down and enjoy during their anticipated shorter period of retirement.

In response to this problem, we can take on board Schokkaert’s suggestion that a package that combines a state pension that meets everyone’s basic needs with a collectively funded occupational pension that provides further annual retirement income can be justified within my ex ante rational insurance approach. Schokkaert notes that even those from more privileged backgrounds who have higher expected earning power would need to guard against the risk of losing their ability to work on account of illness or disability to which we’re all susceptible. A basic state pension would ensure that they, as well as everyone else, would have enough to survive in later life even if their earnings turn out to be insufficient to fund this via an occupational pension. Therefore, even if they expect that they will enjoy higher earnings out of which they would end up overpaying contributions into their basic state pension, it would remain rational for them to opt into a package that includes such a safety net. Those from less privileged backgrounds with known lower earning potential would have even stronger reason to opt into an arrangement that includes the safety of a basic state pension. The fact that this pension would provide a guarantee against their heightened risk of poverty during their retirement years would provide a significant counterweight to the aforementioned reason to opt out of this arrangement that is provided by the fact that they can expect fewer years in retirement to enjoy annual pension income.

I also accept Schokkaert’s point that it would not be in the ex ante self-interest of the children of the very most wealthy to opt into a collective pension scheme. They could simply fall back on the high wealth of their parents in response to any contingency. This segment of society is, however, sufficiently small that it would not be necessary to compel their participation in such a scheme in order to facilitate risk pooling among the large remainder of society. Moreover, tax relief that selectively applies to collective pensions that generate annual income in retirement, but not to the accumulation of a bequeathable or otherwise transferable lump sum from an individual retirement savings account, would both mitigate such accumulation and transfer of wealth and help to ensure the rationality of voluntary participation in such collective pensions. Such selective tax relief combined with default enrolment would ensure sufficient voluntary participation in such collective pensions by means of the carrot of financial incentives, while avoiding the coercive stick of penalties for non-enrolment. Collective pensions can therefore be sustained by the voluntary participation of individuals in the mutual association of an enduring corporate body that pools risk across both space and time on fair terms of social cooperation.