Moshe Milevsky is an Associate Professor in Finance at the Schulich School of Business at York University, and he is one of the world’s leading authorities on retirement income. Professor Milevsky recently published and presented an academic paper on tontines and he is in the process of writing a book on the same topic.
AD: Can you provide some historical context on regulatory issues and why tontines disappeared from the U.S. financial services landscape in the early 20th century? Were the problems related to plan participants, plan sponsors, or both?
Moshe Milevsky: The main problem in the late 1800s and early 1900s was insurance companies who sold "tontine insurance" and mismanaged the funds. The investment fees were too high, the company managers paid themselves too much, the company's bribed regulators and in the end the consumer was left with meager returns. Sound familiar?
AD: How do the past regulatory issues impact the present, and are perceived regulatory obstacles actually still relevant? One can understand how moral hazard might come into play at the participant level, but at the sponsor level how is it really that different from regulating a bank, insurance company or any other custodian of funds?
Moshe Milevsky: Tontine insurance is illegal in most (if not all) U.S. states, so the regulatory obstacle is still on the books. Either the law has to change, or innovators will have to convince regulators that 21st century tontines are quite different from 19th century tontine insurance -- despite the shared name.
AD: Why would state departments of insurance regulate tontines if they are not technically insurance products?
Moshe Milevsky: It is debatable whether 21st century tontines would be considered insurance or not. Yes, there is pooling of risk and the payout is linked to lifespan, so I can see the argument for it being regulated as insurance.I presume insurance companies will want it in their court. On the other hand, there is no need to maintain reserves or capital, so an economist might argue that it is not insurance, per se. I suspect this will be a battle between Vanguard (not insurance) and Metlife (yes insurance) and might reach the supreme court, if this idea catches on.
AD: Are there examples of contemporary tontines that are thriving? If so, where do they exist and how have these plans avoided the pitfalls that led to extinction in the U.S. market?
Moshe Milevsky: One example is the Israeli pension system. They adjust annuity payouts based on actual mortality experience, in relatively transparent way. They call it participating annuity, but I prefer tontine-like. Note that TIAA-CREF also uses such a system. A number of countries in Scandinavia have adopted similar approaches. There is some work in Australia to reintroduce along the same lines.
AD: Are there contemporary examples of tontines that are structured for social welfare optimization as discussed in your paper?
Moshe Milevsky: See above answer.
AD: With an income annuity, longevity risk is removed for the annuity owner but retained by the insurance company. With a tontine, longevity risk is eliminated for both the plan sponsor and participant. Why the difference and how does this work?
Moshe Milevsky: Basically, the pool (i.e. plan sponsor) uses the LAW of LARGE NUMBERS to diversify idiosyncratic longevity risk and the aggregative (systematic) longevity risk is absorbed by the pool itself.
AD: You discuss the inherent cost advantage of a tontine relative to income annuities and the loading thresholds at which risk averse participants would prefer tontines. Can you summarize what this cost advantage looks like, and what is the typical insurance loading on an income annuity?
Moshe Milevsky: The typical loading is 3% to 15% depending on which mortality table you use to determine the mark-up. My understanding is that new capital requirements in Europe (Solvency II) will increase this number by another 10%, which -- to me -- makes tontines more appealing.
AD: Are there any specific considerations or requirements regarding the nature of the “pool” of tontine participants (e.g. age diversification, geographic diversification, etc)?
Moshe Milevsky: The pool must be relatively homogenous in age, gender and health. Otherwise it gets quite tricky how to adjust payouts based on unique mortality rates. Lets keep it simple. My guiding principle is that a tontine annuity should not require an actuary to manage!
AD: What is the minimum number of participants for a viable tontine?
Moshe Milevsky: I would say that at 50 participants you can run a variable tontine and at 500 you have basically eliminated the idiosyncratic component.
AD: What is the ideal legal structure or vehicle for a tontine? Is there, for example, a legal plan and plan document? Are there special purpose vehicles that are similar to what is used for asset securitizations or is it simply a custodian that’s involved?
Moshe Milevsky: Custodian with a simple rule: Split the annual income amongst the survivors. This was done 300 years ago without much fuss.
AD: Are there any similarities between tontines, pass-through mutual funds and the notion of Limited Purpose Banks as advocated by Laurence Kotlikoff?
Moshe Milevsky: Yes. I consider Limited Purpose Banking to be the banking equivalent of the tontine. The insurance company takes no risk and acts at intermediary. And, like Larry's Limited Purpose Banks, I suspect (sadly) that they both stand the same chance of resurrection!
AD: How are tontine assets managed? Is it simply a matter of asset-liability matching or attempting to maintain real purchasing power through TIPS or some other inflation hedge? Would there be any equity exposure?
Moshe Milevsky: The funds would be managed like any other index fund. You (as the annuitant) would pick the asset mix you want, and agree to a tontine overlay.
AD: You discuss subjective mortality in your paper. Doesn’t subjective mortality already come into play in the existing annuity market with the existence of longevity annuities? In other words, can’t individuals who believe themselves to be healthier than the general population choose back-loaded annuity payouts in the form of longevity insurance?
Moshe Milevsky: Yes. The concept of subjective vs. object mortality is well established in the annuity market. It drives pricing, as it would for tontines.
AD: Thanks very much for your time Moshe.