Not Just for Breakfast Any More

j. mark iwry, is a nonresident senior fellow at the Brookings Institution and a visiting scholar at the Wharton School. william gale is a senior fellow and wclaire haldeman is a senior research assistant at the Brookings Institution. david john is a senior policy advisor at the AARP Public Policy Institute and a nonresident senior fellow at the Brookings Institution. Their research was supported by the philanthropy, Arnold Ventures.

Illustrations by ben kirchner

Published April 26, 2021


It does sound tasty — perhaps like a croissant married to a millefeuille. Alas, the tontine is not a pastry but an investment vehicle, a reinvention of an almost forgotten type of financial contract dating back to the 17th century. And though they aren’t edible, tontines could help solve a major problem in retirement planning.

The Origin Story

Back up for a moment. Households face two related financial challenges as they prepare for retirement: saving enough while they are working, and spending their savings wisely after they’ve retired. Policymakers have paid a lot of attention to the first issue, less to the second.

Drawing down assets in retirement presents a fundamental dilemma. Consume too much too fast and you might end up living in your daughter’s spare bedroom. Consume too little too slowly and you may end up leaving a windfall to your unworthy nephew.

This timing problem is only getting worse. Not only have life spans increased, stretching retirement years, but ever more retirees depend on defined contribution (DC) retirement savings plans like 401(k)s and IRAs that accumulate cash nest eggs, rather than defined benefit (DB) pensions that pay out as steady income for life — that is, as an annuity.

You generally do have the option of taking at least part of that DC nest egg at retirement and buying your own annuity from an insurance company. While this can often be a very good choice, it might be — or might seem — less than optimal. Commercial annuities can be expensive: interest rates are low, annuities are subject to insurance regulations and reserve requirements that raise costs, and sellers’ commissions can be high and may contain hidden fees. Other financial products, such as managed payout funds, can facilitate the planned decumulation of wealth with higher average returns and lower expenses, but don’t guarantee the specific amount of each monthly payment and, in any case, are not available in most retirement plans.

Prudent householders who want to make sure they won’t outlive their savings might, however, have another option. We explore the potential for a different financial product — a tontine — that could help retirees cope with the financial uncertainty linked to unknowable life spans. The “tontine principle” — that survivors benefit financially from the death of their fellow participants in an investment pool — can evoke strong reactions. It has inspired murder plots in novels, movies and even a Simpsons episode. But stripped to basics, there’s nothing sinister about it.

Over the centuries, tontines have been marketed to finance everything from colonial- era infrastructure projects to European wars. And yes, tontines once even financed Americans’ retirements. They were quite popular in the United States in the late 1800s and early 1900s, but were regulated into oblivion in response to corrupt insurance company management.

Now they are creeping back. A number of countries have created pension plans that incorporate tontine principles. The attraction, explained more fully below, is clear: tontines can provide higher returns than annuities. To be sure, they don’t come with the guarantee of a fixed monthly payment offered by plainvanilla life annuities. But they could fill a gap for those who would like higher returns than commercial annuities offer while still enjoying considerable protection against outliving their assets.

Tontines were popular in the United States in the late 1800s and early 1900s but were regulated into oblivion in response to corrupt insurance company management.
How Tontines Work

The basic idea is that a group of people pool assets, and when members of the group die, their shares of the pool are divided among the living. For example, suppose 10 people irrevocably invest $100,000 each to create a $1 million investment pool. The pool buys, say, a bond that yields 4 percent interest and pays the entire $40,000 in interest as an annual dividend. While all members are alive, each receives $4,000 per year ($40,000 divided by 10) as a “base return.”

When a member dies, the departed’s annual base return is divided among the survivors. Thus, after the first death, each surviving member receives an additional $444 ($4,000 divided in ninths) per year as a “mortality credit” in addition to their $4,000 base return. The total annual interest of $40,000 from the bond stays the same, of course. But as there are fewer and fewer survivors, the annual payout to each survivor grows. When the group reaches an end point specified in the original pool contract, such as when only three of them are still alive, the $1 million in principal is divided equally among them, and the arrangement terminates.

Everyone in the pool gets income for life (except the last three survivors, who receive a very big lump sum) and a predictable minimum return while living. But there’s a catch. Theory and evidence suggest that retirees prefer income that is constant or somewhat declining as they age. And although each participant in a tontine is ensured income for life, it would be unsuitable as a primary source of retirement income for those who would rather spend more of their assets in the early years of retirement and less later on. Remember, in a tontine, payments to survivors rise rather than fall.

Wheels Within Wheels

To address this preference, the retirement experts Moshe Milevsky and Thomas Salisbury of York University in Toronto propose a hybrid tontine in which the annual return to survivors stays constant over time, even as the group shrinks. Their approach is simple and elegant: Invest in a portfolio that generates a base investment return that declines with projected survivorship over time, so that total return for survivors (base return plus mortality credits) stays roughly constant over time.

Two other retirement experts, Jonathan Forman of the University of Oklahoma and Michael Sabin (an unaffiliated researcher), propose a differently structured means to the same level-payout end. Their level-payout tontine offsets the increasing mortality credits by having the investment pool annually repay members a declining portion of their initial contributions.

Either of these approaches better matches most retirees’ preferences by providing more income than the simple tontine initially, and then maintaining the same high total payout as the number of survivors shrinks, rather than escalating over time.

In practice, the pace at which the members of a relatively small group actually die will deviate from statistical expectations. So, if the declining base return schedule is set in advance based on expected mortality, the total periodic return to survivors will vary from year to year. But if the pool is sufficiently large (say, 1,000 members), the expected variation in payouts should be small because the number of surviving members at any given time is quite predictable — except, of course, in the sort of pandemic we now face.

Moreover, one could take this a step further. Adjusting the rate of decline in the base return annually, based on the actual (as opposed to predicted) mortality experience of the pool members could almost entirely eliminate variability in the total return.

In the examples above, each tontine member is assumed to have identical life expectancy (or “mortality risk”). In those circumstances, division of payments among survivors would be actuarially fair if they were divided equally for each dollar invested. But if the members have differing life expectancies, this approach favors those likely to live longer and discourages the participation of those with short life expectancies.

A tontine covering people with different life expectancies could still be designed to avoid favoring some participants over others, though it would get a bit more complicated. Richard Fullmer, founder of Nuovo Longevita Research, a pension and retirement consultancy, shows that, to achieve an actuarially fair tontine, for each participant alive at the start of each period, the participant’s expected gains (expected mortality credits times probability of surviving) should equal the participant’s expected losses (the account balance times the probability of dying). [Trust us on this one: we did the math and it works.] By appropriately allocating mortality credits, the pool can permit people with differing life expectancies and differing initial balances to participate on terms that are fair to everyone.

Yes, But...

In theory, then, retirement tontines can work. But how about in practice? While modern tontines would be substantially different from the products that fell out of favor over a century ago, their legal status remains unclear. Except for the original New York state law’s prohibition of tontines, which actually prohibited only tontine-style investments that pay out less frequently than annually, and except for a couple of states that explicitly bar tontines, both state and federal law are generally silent on tontines, so we assume that current law generally would not flatly prohibit all tontinelike arrangements (that is, investment pooling and allocation of mortality credits).

Pensions with tontine features are offered in several high-income economies and are authorized generally in the European Union. But the U.S. remains mostly on the sidelines.

Indeed, there are precedents. For example, the law has long allowed “participating life annuities,” a cousin to the tontine that pairs a basic income guarantee with a variable surplus distribution dictated by the actual mortality of the pool of investors. Similarly, some life annuities periodically adjust payouts based on the mortality experience of the pool of annuitants, so the participants — not the insurance company — bear the risk if people in the pool live longer than expected.

Pensions with tontine features are offered in several high-income economies and are authorized generally in the European Union. But the United States remains mostly on the sidelines. To date, there is little evidence that the U.S. insurance and pension planning industries are interested in tontines.

Certainly, misperceptions about the nature of tontines should not be a deal breaker.

As is now plain to readers who have gotten this far, the survival-contingent payout that underpins the tontine implies that survivors benefit from the death of others. But while the notion of profiting in this way makes some people squeamish, it should not be a concern. Tontine participants have no incentive to wish harm to co-participants when the group participating is large or, in any case, when members’ identities are kept confidential from other members. And films and novels aside, there have been no recorded episodes of actual attempts by members of a tontine to eliminate fellow members.

Ironically, commercial annuities — and even the Social Security program — regularly pool mortality credits without controversy. In those cases, mortality credits are less visible because they go through the government or the insurer, which also takes a portion of the credit as increased margin. By contrast, mortality credits in tontines directly and transparently benefit the survivors rather than being obscured by the involvement of a financial intermediary that may keep a share of the gains

Nor should complexity be a stumbling block. Tontine-style pooling can be straightforward, transparent and fair. Of course, in practice, there would be complications from accurately estimating mortality risk, adjusting the base return and investing in equities to increase expected returns. But the real question is how tontines rank against other retirement income vehicles. Tontines would be more transparent than many current annuity products, which contain bells and whistles and use brand-specific terminology that make it difficult to compare products and prices.

Tontines could also provide the spousal protection that other retirement vehicles offer. “Tax-qualified” retirement plans are required to protect the interests of spouses. Definedbenefit pension plans must pay death benefits to surviving spouses unless they formally waive that right. In 401(k)s and most other defined-contribution plans, a deceased participant’s spouse inherits the participant’s full account balance unless the spouse earlier agrees in writing to the designation of a different beneficiary.

Tontines outside tax-qualified plans could provide similar protections. Indeed, a tontine might also be arranged to provide that base payments and mortality credits continue being paid to the surviving spouse until that person’s death, at which point the deceased couple’s interest in the pool would be reallocated to surviving participants.

One potential stumbling block applies to tontines, but also to other retirement products. An equitable allocation of mortality credits would ensure that tontines do not discriminate against people with shorter life expectancy or in favor of those with longer life expectancy. Life expectancy varies with age, gender, race, ethnicity, income level, marital status and other factors. And, as discussed above, if a tontine is designed to take all these factors into account, it can generate equitable allocations of mortality credits.

Racial disparities in life expectancy present a particular problem, as well as for any other retirement income product that pays benefits until a person dies (including Social Security, commercial annuities and definedbenefit pensions).

Existing law may, however, forbid some adjustments of this type even if the law is intended to protect traditionally disadvantaged groups, like African-Americans, who tend to have shorter life expectancies. For example, commercial annuities and other insurance products are generally allowed to engage in price discrimination because of life expectancy differences based on age and gender, but not race. In contrast, employer-sponsored pensions are not allowed to discriminate in benefits or contributions based on gender or race and can discriminate only because of life expectancy differences based on age in certain specific ways.

For tontines that are sponsored by an entity other than an insurance company or an employer-based retirement plan, it is unclear how current law would apply to all the possible adjustments for mortality risk. Racial disparities in life expectancy present a particular problem here, as well as for any other retirement income product that pays benefits until a person dies (including Social Security, commercial annuities and defined-benefit pensions). We highlight this issue because it would have to be addressed in implementing tontines.

The End (or at Least Planning For It)

Everything old is new again, as the adage goes. Tontine-inspired arrangements to finance retirement are receiving more attention around the world because they are both efficient and transparent; mortality credits accrue to pool members directly and traceably. While commercial annuities can guarantee a specified income for life, tontine pooling offers more expected income for the premium dollar plus meaningful — although less — protection against outliving savings.

In addition to lower cost on account of transparency, tontine-type arrangements require less expensive regulation, avoiding the financial drag affecting commercial annuities: the costs of an entrenched distribution network and reserves needed to cover systematic longevity risk while maintaining their annuity payment guarantees. For many people, commercial annuities’ guaranteed monthly payment amounts and complete protection against outliving one’s assets may not be worth the extra cost involved, especially as historically low interest rates have depressed the monthly returns available on commercial annuities.

Public policy is only beginning to grapple with the challenges of effectively managing retirement security in a world of 401(k)s, IRAs and lump-sum payouts rather than defined- benefit plans. Whether tontines can penetrate the retirement finance market in the teeth of legal uncertainties and other loose ends remains to be seen. But given the limited diffusion of simple fixed-income life annuities and of managed payout funds, tontineinspired structures could offer an attractive alternative that should have a place among the options to finance retirement income.

main topic: Pensions