In 1693, as related by economist Moshe Milevsky in his book King William’s Tontine, King William of England (actually a Dutchman and formerly William of Orange) sold to investors an interesting financial product. Each buyer was a member of a pool of buyers that immediately received a “dividend” on their investment of 10% per year. After a specified number of years (seven, to be exact) the “dividend” yield was reduced to 7%. So, for example, supposing that each of 1000 members of a pool invested 100 crowns, the total payments to the pool members would be 10000 crowns (1000*100*10%) for first few years and then 7000 crowns after the reduction in the dividend. As an individual investor, your payments would continue as long you remained alive, and would stop at your demise. This makes the investment analogous to a life annuity. But there is a kicker. As each member of the pool passed away, the payments to the remaining members went up. This is because the King committed to paying out the same total sum each year until everyone in the pool had expired. Thus, the last remaining individual would have a final year payment of 7,000 crowns, massively higher than the initial annual payment of 100 crowns. Actually, the plan called for total pool payments to decline twice; first at year seven when the dividend was reduced from 10% to 7%, and second once the number of remaining individuals in the pool dropped to 7. So, the largest individual payments actually made were 1,000 (7,000/7) crowns.
This product is called a “tontine” after the Italian inventor Lorenzo de Tonti, who died in 1684. It has a long and rich history. In 1785, Treasury Secretary Alexander Hamilton proposed to President Thomas Jefferson that a tontine structure be used to pay down the enormous Revolutionary War debt. Although President Jefferson didn’t jump on the idea, maybe it is time for a reappraisal. Federal government sponsored tontines could solve two very significant financial problems today – providing financial security to retirees and paying off today’s giant federal debt. Here’s how that might work. Instead of issuing Treasury bonds, suppose the government issues Treasury Tontines. Suppose a pool of 1000 recent retirees invests $1 million each in the Treasury Tontine. Instead of receiving 2% interest as they would on Treasury Bonds, the annual income could be a lot greater. The Treasury needs only make an estimate of the age of the last of the 1000 people to pass away. This number is not hard to estimate, and will be somewhere in the range 105-115. Suppose the retirees are 65 and the estimated age of the last to die is 115. This means the Treasury can schedule pool payments for a fifty year period. The wherewithal for the payments is the $1 billion proceeds plus normal interest that the Treasury would otherwise be paying. Also, it is reasonable to allow total pool payments to decline at a modest rate, lower than estimated pool mortality rates (this is done so the last few members of the pool don’t receive giant payments of millions of dollars per year). The key idea would be that as members of the pool died, the payments to each of the survivors would increase. There would be a guaranteed initial annual payment, probably around 5% of the investment, but whether that payment rose or fell over time would depend on the actual mortality experience of the pool.
The potential benefits of the scheme are considerable. First, the investors will receive a life annuity. Granted, the amount is not fixed and could rise or fall. This is because the longevity risk is shared by members of the pool. This is the second benefit. Since there is no need for an outside guarantor, the chance of plan failure is remote. There are no counter-party problems. Finally, after the final person in the pool has expired, the government obligation has expired as well.
For a variety of reasons, tontines developed a bad reputation. One problem, although apparently never documented, is that members of a pool have a financial incentive to see the demise of other members of the pool. Economists have looked closely at the death and survival rates of people who were part of a pool and see no evidence of foul play.
A variety of a tontine was a very popular savings product sold by insurers in the US in the mid-1800s. The essence of the plan was that each member of a pool would make an annual payment into a fund, the accumulated funds would be invested by in the insurance company, and after twenty years the funds would be distributed to the members of the pool who were both still alive and who had continued making payments for all twenty years. If a pool member failed to make an annual payment, the member’s policy was said to have “lapsed” and his stake in the fund eliminated. Thus, there were multiple sources of investment “return” – the actual return on funds invested plus a larger share in the fund as others either died or let their policy lapse. The picture of savers losing everything because they were unable to continue making payments each year, or simply forgot to make one payment, sparked enormous negative publicity and resulted in legislation that effectively killed the product.
Author and professor Milevsky is one of several academics that is promoting the idea of “tontine thinking” or risk sharing as an ideal model for retirement planning. By letting actual payouts be adjusted for faster or slower mortality rates, longevity risk can be effectively managed. According to Milevsky, the best example of this today is the teachers’ retirement plan TIAA-CREF. Although cleverly avoiding the use of the work “tontine”, TIAA-CREF annuities capture the risk sharing idea. If the mortality rate in a pool is greater than predicted, annuity payments will rise, and conversely if the actual mortality rate is less than predicted. Since the sponsoring entity is not guaranteeing a specific annuity, the entity does not have to spend economic resources hedging against longevity risk, or maintaining high levels of capital, or charging enormous fees that massive hedging or capital levels would require. Because of this, Milevsky estimates that professor retirees (like himself one day) receive annuity payments roughly 20% greater than they would be if offered in a traditional life annuity form. He believes that most retirees will accept a modest amount of uncertainty regarding the exact payment, if on average it is 20% greater and the underlying economic structure is more solid.
Obviously, in view of the TIAA-CREF example, it is not required that sponsors of tontines be public entities. Aside from potential legal obstacles, there is nothing to prevent private issuers from sponsoring tontine pools. In fact, maybe this is to be preferred. The beauty of the tontine is to replace guarantees by risk sharing, but political pressures tend to push government programs into the guarantee business.