Other views

Using tontines to solve public pension underfunding

Tontines are investment vehicles that could replace large pension plans, public or private, or those of smaller employers, and could be used to provide retirement income for participants.

These “tontine pensions” would have several major advantages over most of today's pension programs, especially with the challenge of maintaining full funding.

In a simple tontine, a group of investors pool their money to buy a portfolio of investments, and, as investors die, their shares are forfeited, with the entire amount going to the last surviving investor. On the television show “M*A*S*H” for example, Col. Sherman T. Potter, as the last survivor of his World War I unit, got to open the bottle of French cognac he and his buddies had bought. Over the years, this last-survivor-takes-all approach has made for some great fiction where nefarious characters try to kill off the rest of the investors and “inherit” the fund.

Of course, tontines can be designed to avoid such mischief. For example, a tontine could be designed to make periodic distributions to surviving investors and to solicit new investors to replace those that have died. Structured in this way, a tontine could operate into perpetuity; and it could be used to create a fully funded tontine pension.

In a tontine that makes periodic distributions, forfeitures might be divided equally among the survivors. For example, four people might each contribute $1,000 to a tontine, and when one dies, each survivor would get $333.33. Unfortunately, this simple approach can be unfair — for example, because it favors younger investors who are likely to live longer and so receive more distributions.

Instead, distributions should be made in unequal portions, carefully chosen to provide fair bets for all investors. For example, imagine that our four $1,000 investors are age 65, 70, 75 and 80. According to the Social Security Administration's 2009 unisex life expectancy table, our 65-year-old has an 18.88-year life expectancy and a 1.3% probability of dying before reaching age 66; our 70-year-old has a 15.22-year life expectancy and a 2% death probability; our 75-year-old has an 11.89-year life expectancy and a 3.2% death probability; and our 80-year-old participant has an 8.95-year life expectancy and a 5.2% death probability.

With some elementary mathematics, we can use these death probabilities to determine a fair way to divide any dying member's $1,000 investment among the survivors. For example, if the 80-year-old participant is the first to die, the 65-year-old should get $187.64, the 70-year-old should get $300.51, and the 75-year-old should get $511.85. The younger investors should get less as they are likely to live longer and so receive more distributions.

Tontines can also be designed to take into account differing levels of contributions and to properly account for investment earnings. Certainly, those who survive the longest would get better than average returns, while those who die young might not even recover their initial investments. On average, however, each investor could expect to recover their initial contribution and any returns on that investment. In short, this tontine provides a fair bet.

Two more enhancements are needed to create a tontine pension plan. First, in order to provide regular rather than sporadic distributions, we will need more investors. For example, with 5,000 investors, survivors could pretty much count on getting some money every month. Also, rather than making distributions whenever a member dies, we should just make monthly distributions. If a member is alive at the end of the month, he or she will get a share of the forfeitures from those who died that month. If a participant dies during the month, they will get nothing, as the balance in their account was distributed to the remaining survivors.

Second, in order to provide level distributions throughout retirement, we need to cancel out the age-related back-loading inherent in a fair tontine. For example, that is what occurred when the 80-year-old in our four-investor example died: The 75-year-old survivor got $511.85, but the 65-year-old survivor got just $187.64.

Our approach is to use a “self-payback” mechanism that increases the distributions made to younger survivors and decreases the distributions to older survivors. Basically, if a retiree is alive at the end of the month, they get a distribution equal to the balance in his or her tontine account divided by the appropriate monthly annuity factor. For example, if an employee retired at age 65 with $100,000 in a tontine pension, we estimate they would get level distributions of around $850 a month for life ($10,000 a year).

Alternatively, a tontine pension could be designed to make inflation-adjusted distributions to retirees. Assuming a 3% inflation rate, our hypothetical 65-year-old retiree would get around $660 in the month she retired, but eventually her inflation-adjusted distributions would exceed the $850-a-month distributions of the level tontine pension.

Now, assume a large employer decides to contribute 10% of salary to a tontine pension plan for all of its employees. Consider an employee who started working for that employer at age 35 and retired at age 65. We estimate that he would earn a level tontine pension that would initially replace around 55% of final earnings or, alternatively, an inflation-adjusted tontine pension that would replace around 43% of final earnings.

As in a defined contribution plan, participants could be allowed to direct the investment of their individual accounts, but we believe the sponsor of a tontine pension plan would do a better job. Unlike a defined contribution plan, however, there would be no option for lump-sum payouts: Benefits would be paid out only as monthly distributions that would closely resemble the monthly payments from an actuarially fair variable annuity. As with both defined contribution plans and variable annuities, monthly tontine pension distributions would fluctuate with the value of the underlying investments, although that variability could be reduced by designing the plan to smooth distributions over, say two or three years, and, in any event, individuals can always self-budget to smooth their consumption in the face of variable income.

We acknowledge plan sponsors might face some challenges in explaining how tontine pensions actually work. From a participant's standpoint, however, a tontine pension would look like a variable annuity, and a “smoothed” tontine pension would look like a traditional defined benefit pension or fixed annuity.

All in all, tontine pension plans would have three major advantages over today's pension programs. First, unlike traditional pension plans — which are frequently underfunded, tontine pension plans would always be fully funded, just as defined contribution plans are funded, through regular contributions equal to, say, 10% of salary. This fully funded feature should make tontine pension programs particularly attractive to public and private employers with underfunded pension funds. While replacing a traditional pension plan with a tontine pension plan would do nothing to reduce the plan's unfunded liability, the sponsor would never again have to worry about underfunding attributable to future benefit accruals.

Second, unlike traditional pensions — where the plan sponsor must bear all the investment and actuarial risks, with a tontine pension the plan sponsor would bear neither of those risks, as both are implicitly passed on to the participants.

Finally, tontine pension plans would be inexpensive to operate. The underlying investments could be managed by a cadre of active and passive investment managers, just as they are in a defined benefit or defined contribution plan. But, unlike a DB plan, no actuary would be needed, as monthly distributions would automatically approximate an actuarially fair variable annuity. Also, unlike a commercial annuity, no money would need to be set aside for insurance company reserves and profits. In fact, we believe a discount stock broker could manage a tontine pension plan for as little as 0.3% of assets under management (e.g., 0.1% for a passive stock index fund and another 0.2% for the tontine pension management and record-keeping functions); and that means tontine pension plans would be able to provide significantly higher lifetime income benefits to retirees than current pension plans or annuities.

Jonathan B. Forman is the Alfred P. Murrah professor of law at the University of Oklahoma, Norman, and Michael J. Sabin is an independent consultant in Sunnyvale, Calif. They are the authors of “Tontine Pensions: A Solution to the State and Local Pension Underfunding Crisis,” forthcoming in the University of Pennsylvania Law Review.

This article originally appeared in the June 9, 2014 print issue as, "Using tontines to solve public pension underfunding".