Learning from ‘Tontines’ to Build Modern Retirement Income Schemes

The Brookings Institution hosted a thought leadership conference dedicated to the topic of retirement income, featuring such prominent speakers as Professor Richard Thaler, of the University of Chicago Booth School of Business, and Phyllis Borzi, former Assistant Secretary for Employee Benefits Security at the U.S. Department of Labor.

The day kicked off with a panel of experts who explored the history of retirement income, highlighting the fact that the challenges facing retiring workers today are by no means new or novel. And, as the panel pointed out, neither are the types of solutions being debated by academics and policymakers.

David John, a nonresident senior Brookings fellow and senior strategic policy adviser for the AARP Public Policy Institute, said the retirement income planning challenge has literally for centuries been the most complex financial decision an individual faces in a capitalistic society.

“The simple fact is that this topic is difficult,” John said. “In 2019, the data shows more than seven in 10 Americans don’t know how to implement a retirement income plan—and that’s just the proportion who will admit it. When you test their skill sets, most of those who say they can make a plan aren’t going to always make optimal decisions. So this begs the question, can we automate the same way we did with accumulation? That’s something a lot of people are asking themselves right now.”

John observed that this is not a U.S.-only issue. Every country with a developed retirement savings system is focusing on this problem of converting savings into income.

“The Australians have the same worry about running out of money despite the fact that they are required to save for retirement,” John said. “As a result, they are requiring all plans in their national system to offer income options by 2024. In the United Kingdom, the House of Commons has recommended a similar solution. In Canada, we have seven major pension stakeholders calling for longevity risk pooling at a massive scale. In New Zealand, they have a triennial review of the KiwiSaver system, and retirement income solutions are on their agenda for the next meeting.”

According to John, there is growing global interest in retirement income solutions built around managed payout funds.

“Shell Oil has this kind of a mechanism for their employees in the Netherlands,” he observed. “It is basically an active investment fund with a high proportion of equities, but it also has a significant amount of countercyclical hedging investments to limit great losses during market downturns. Retail funds in the U.S. offer some of these features already.”

John recommended U.S. policymakers consider promoting a three-pronged approach to retirement income. Workers could enter a managed payout fund starting at age 55 or 58. This would form the basis of their nondiscretionary retirement spending, and supplemental investment accounts could be used by those with sufficient means and with an interest in taking more investment risk. Finally, individuals would purchase a longevity annuity kicking in later in life. This could be structured as a qualified longevity annuity contract, or an individual could choose to delay annuitization, perhaps purchasing the annuity at age 75 with payments set to start at age 85.

Another speaker on the panel, Moshe Milevsky, professor at the Schulich School of Business, York University, zoomed into the interesting topic of “tontines,” which are a type of historical annuity structure that was first put into well-documented practice as far back as the 1600s. Commonly, tontines were used by governments to fund wars or other foreign exploits, especially in France and the United Kingdom.

“In these income schemes, the individual would give, say, 100 pounds to the government, and in return he would essentially get installment payments for life, with interest,” Milevsky explained. “The approach resembled an indexed annuity, fascinatingly. So, there has been a really deep history of all this and that shows how complex the problem is. Centuries ago, people had already developed very sophisticated systems of budgeting mortality and longevity risk against annuity payments, and systems to divide up shares of any profits or interests.”

One unique feature of early tontines was that they were often structured as closed “syndicates,” meaning that once a tontine money pool started paying out income streams, the size of the income stream going to the individuals grew each time one member of the syndicate died. By the same token, payments stopped when the last syndicate member died. The practical effect of this was that the income streams from tontines tended to start out modest and then grow to be quite large for the select few people who survived longest among the syndicate membership.

“The system basically had a sum of income that would stay the same each year, but it was being paid out to a shrinking group of people over time,” Milevsky said. “The winner at the very end got a huge income. At the end the principal is gone, importantly. It’s basically amortized.”

This system would not be practical in the modern context, Milevsky said, because of the back-loaded nature of the payouts for any given individual in the tontine. However, modern “tontine theorists” have developed models that would address these issues and deliver smoother income streams for members. 

“We can contrast this approach against a standard lifetime income annuity. As more people pass away over time, the sum of the total payments for a group lifetime annuity actually goes down, but the payments are stable for individuals,” Milevsky said. “On the tontine side, it’s basically the opposite. I also like to point out that, in the 18th century, Adam Smith was writing in favor of tontines. Alexander Hamilton was another big fan of these.”

On Milevsky’s analysis, “tontine thinking is more important than actually using them.” What he means is that tontines very clearly demonstrate that “mortality credits are to be thought of as an asset class.”

“When we are pooling people’s longevity risk, mortality credits are an alpha. Not everyone likes to think about it this way, but it’s true,” he said. “What this means in practice is that, if you’re willing to put your money in the insurance pool and accept the risk that you’ll die early, you will get a much greater return should you live to your hoped-for point of longevity. The lesson is that we need to stop mixing up annuities as a topic with mortality credits as a topic. I’ve noticed that explaining the tontine first makes annuities shine in a new life. Annuities are tamer; the income stream is stable and you don’t have to worry about people knocking each other off to increase their own payments.”

The final panelist on this topic was Michael Davis, head of defined contribution plan specialists for T. Rowe Price and a former deputy to Phyllis Borzi during her time at the head of the Employee Benefits Security Administration.

“Annuities can play a meaningful role for many people, and it’s critical to explain there is a variety of vehicles that can be used,” Davis said. “Given the documented heterogeneity in retirement spending patterns, Americans should have the right to choose how they structure and spend their retirement income.”

Offering a history lesson of his own, Davis pointed out that, until just the last five or 10 years, retirement plan sponsors were not very interested in keeping participant assets in their plans post-retirement. But this has evolved significantly and sponsors and participants are highly interested in in-plan income solutions.

“We already see a clear trend that more and more dollars are being left in the plan,” Davis observed. “I think part of that is the conversation about fiduciary protections, and the interest among plan sponsors to achieve and maintain scale for positive pricing benefits. Given all this, the behavior of retirees is becoming an important point of focus for providers like us and for our clients. Some have argued that, because of Social Security, U.S. retirees are over-annuitized. We find this is true for only the very lowest earners. For the great many working Americans that is simply not true. A single solve for these different needs is not the right way to go, I should add.”

Davis expects the retirement system will rapidly evolve to give greater access to systematic withdrawal programs, bond ladders, tontines, deferred and immediate annuities, and managed payout funds.

“Our firm is focused a lot on managed payout funds and we are going to launch a retirement 2020 income fund this year,” Davis noted. “This will be a target-date product but it’s not a qualified default product. Participants will have to select into this fund within five years of retirement.”