Key regulatory features of a tontine

This article describes our views on a regulatory framework for a tontine. Much of this article was provided in our June 2017 submission to Treasury's Retirement Income Policy Division with respect to the Discussion Paper ‘Development of the framework for Comprehensive Income Products for Retirement’.

We consider the concept of pooling longevity risk to be very important in meeting longevity needs of retirees. Our analysis shows that payments from a tontine based longevity pool (otherwise known as a ‘GSA’) could become a key source of income for those people who live to at least 90 years of age. It can provide a modest source of income for people who live into their 80s.

We see little advantage, and potential significant disadvantages, to a person entering into longevity risk pooling prior to age 70. Although it seems simpler, it may not be appropriate for a person to take out a longevity product on retirement when they are still aged in their 60s. If a healthy person retires at 65, their mortality rate may be less than 1%. If so, they are risking a lot of capital for a small additional income at that age.

Our analysis suggests it is very difficult for a person to accumulate sufficient wealth to be self-sufficient in the situation when they retire around ages 60-63 and then live to more than 87 years of age. For example, if a person were to live to 100 years of age after retiring at age 62, the ratio of retirement to working years would increase from 25:38 (71%) to 38:38 (100%). Without government welfare support this would be debilitating to most people and a large strain on family & friends. Any government welfare support would require taxing the income of (younger) others which would only result in lower accumulated wealth in retirement for those people. As such, we need a commonality based solution to this problem, and we believe it can be eloquently provided by the ‘tontine’.

We consider the tontine to be insurance. A proper definition of insurance is a form of cover that provides indemnity for a mutualisable risk event over time and space. A tontine fits this definition of insurance. The tontine is essentially a reverse life insurance arrangement within a mutualised pool. If a person dies, their contributed capital is distributed amongst the other living people in the pool as a ‘survival credit’. The allocation of this ‘survival credit’ requires specialised understanding of equity between members of the pool, similar to the allocation of bonuses for with-profit life insurance policies.

The calculation of the survival credits paid to each living person in a period is based on that person’s contributed capital and their deemed mortality rate for that period. Those with more capital and higher deemed mortality rates receive a greater proportion of the ‘survival credit’ distribution. The determination of deemed mortality rates for each member requires specialised understanding of mortality rates, similar to those used to determine term life insurance premium rates.

The deemed mortality rate (qi) is based on a best estimate of the mortality rate for an individual using set criteria (i.e. annuitant selection, age, sex, select period), and is only used to equitably distribute the ‘survival credits’. This approach for the allocation of ‘survival credits’ allows the tontine to be ‘open-ended’. This way the tontine does not need to experience a decline in the number of participants over time.

The payment of ‘survival credits’ reflects the actual mortality rate of people in the pool, allowing the concept to be a mutualisable risk event. If the pool is of sufficient size to remove much of the idiosyncratic death risk of the cohort, the tontine participant, if still alive, should receive distributions from the tontine at a rate similar to that of their deemed mortality rate (qi) divided by their probability of surviving the period (1-qi).

It should be noted that the tontine only provides insurance coverage for the idiosyncratic risk of living, and it does not provide insurance coverage for the systemic risk of improving / deteriorating mortality of the cohort within the pool. Under our definition of insurance, the systemic risk is not an insurable event, as it is not a mutualisable risk. Insurers providing this type of cover are prone to failure. For this reason we do not support retirement products for people with life annuities and/or deferred life annuities provided by a sole insurer. The concentration risk to the failure of a single insurer for potentially a significant weighting of a member’s retirement savings is not appropriate.

The payment profile of a life annuity is typically described in either nominal or real terms. To provide sufficient income for longevity risk (essentially for those years over 85 years of age), an individual would typically be required to invest a significant proportion of their wealth into a life annuity. An investment in a deferred life annuity would require a much lower proportion of their wealth for a person at retirement and therefore reduces the concentration risk to the failure of an insurer. A participant in the tontine would not need to put all of their wealth into the tontine. The proportion of wealth put into the tontine would depend on a participant’s personal circumstances, including their level of wealth and current health status. The proportion of wealth put into a tontine would be significantly less than a life annuity and somewhat less than a typical deferred life annuity.

It should be acknowledged that life annuities and deferred life annuities are not fully scalable due to the systemic longevity risk provided by those products. There is a capacity limit to the longevity liabilities that life insurance and life reinsurance companies will be willing to price and to take on. As the tontine is a mutualisable risk product, it has no such capacity limits.

It is not possible to offer a ‘withdrawal benefit’ under a true tontine as the contributed capital cannot be withdrawn if a person suspects they might die in the near future. That would be unfair on the other participants in the pool who are putting capital at risk. Products are likely to be designed to hide this fact, either by adopting hybrid structures or using misleading terminology. Product disclosure should make it clear that the product is a communal pooling arrangement where on death a person’s remaining investments are surrendered to the pool.

It should be made clear to members that a longevity product provides benefits on survival, but has costs (i.e. forfeiture of capital) on death. It should not be promoted as a ‘better’ product, with higher income than an account based pension. The uplift to payments from a longevity product should be recognised in disclosure as compensation for putting capital at risk on death. It would be inappropriate to compare the income levels of longevity products to account based pensions without acknowledging the capital loss on death for the longevity products.

Product disclosure communication should be clear and up-front. For the longevity component it should be made clear that the member will forfeit capital. It should not be masked in the form of a withdrawal penalty.

The longevity product itself cannot provide flexibility. If people were able to withdrawal monies when they became ill, then that would be to the detriment of others in the pool.

Portability is not possible for a tontine as it could be unfair on other members of the pool. It may encourage activity where another provider provides soft or hard encouragement for less healthy lives to switch into their pool, to make it more attractive for its members. It would be better to encourage the formation of a market with specialists who can appropriately offer a member monies for their policy based on an individual health assessment undertaken by the specialist.

If a person survives the period within the tontine, they would expect a ‘survival credit’ equal to the capital put at risk multiplied by qi/(1-qi), plus their share of the investment return on the assets held by the tontine, less any expenses and charges. For equity reasons, survival credits should be allocated to surviving members at least yearly and should be allocated on an equitable basis requiring an actuarial certificate. Details of how ‘survival credits’ are determined must be disclosed so that members can make an informed decision if they should join the pool.

Where a person’s deemed qi is less than 1.0% (roughly age 70 for healthy persons), we consider the expected ‘survival credit’ to be insufficient to justify participation in a tontine. Prior to expenses a survival credit of under 1.01% (i.e. 1.0%/99.0%), compares poorly to ‘downside’ of losing the full amount allocated to the tontine. We are concerned some product providers may encourage members to join tontines at too early an age.

As a tontine is essentially a reverse life insurance arrangement, other than for tax planning reasons, there seems to be very little justification for someone participating in a tontine when concurrently holding life insurance cover. This is a mis-selling risk that should be monitored by product providers and regulators. Those people with life insurance cover should be discouraged from taking a longevity product unless they have sought appropriate financial advice.

Typically, only a healthy life would join a tontine. This is due to the factors used to determine a person’s deemed mortality rate used in the actuarial allocation of the ‘survival credits’. If only basic factors such as age, sex and select period are used, it makes economic sense for participants to only join if they consider themselves to have close to or better than average life expectancy than the rest of the pool. That said, it may be possible to broaden the factors used to determine the deemed mortality rate or alternatively to offer a ‘bad lives’ tontine which would be restricted to those with poor records of health, or work records in certain industries.

The reason why the tontine can provide an elegant solution to those who live to an old age is that expected mortality rates typically rise sharply from ages 85-90 onwards and therefore provide a good reverse life insurance ‘premium’. Up until age 70-75, expected mortality rates tend to be quite low, so the ‘premium’ received from the reverse life insurance is too modest for most to place any capital at risk.

Using a 3% real return on investments and 75bp p.a. expenses, with annuitant factors applied to Australian life tables (ABS 2010-12), the tontine allows for higher annual payments in the early stages of retirement, with much greater self-sufficiency from 85 onwards when the regular account based pension starts to deplete. If a person was to live into their 90s, the timeframe for participating in a tontine is around 20 to 30 years. As mortality rates rise with age, if you they were to live to 100, around 2/3rds of the payment from the tontine component (i.e. survival credits) would be received from age 90 onwards.

Given the high mis-selling risk of longevity based products it is important that members have chosen a product in an informed manner. For longevity products this may only be discovered on the death of a member, particularly in the case where the member has a living spouse or dependents. As such, in the documentation it must be clear that capital is lost on death for the longevity product. ‘Clever’ products that disguise this fact should be discouraged. Longevity based products may not be suitable for distribution via some direct channels.

Keeping fees and charges low (including those embedded in longevity products) is very important. Caps on fees may be appropriate. It should be noted that asset fees are likely to be incurred by longevity based product providers, similar to the level charged for account based pensions. Administration of the longevity based product will occur additional fees and charges. This may include unit pricing, custodian, actuarial, regulatory etc.

The capital put at risk in the tontine must be invested in suitable wealth that will still be able to provide income over this period. As the tontine is likely to be a major source of a person’s income during the period of 90-100 years of age, the quality of the assets backing the tontine must be high.

The difficulty of finding high quality assets to back a tontine should not be underestimated. Fiat based assets, such as nominal government bonds, carry a significant risk that they won’t provide the purchasing power required to the retiree over the 20-30 years of the tontine.

Our view is that the highest quality of wealth is the gold coin, closely followed real (gold) bills of exchange. Given the duration of the product and the requirement for high quality wealth, there is an argument for an asset allocation of a tontine to gold coin and real (gold) bills of exchange. Depending on the marginal investment return from ‘gold mortgages’ and ‘gold equities’, there is also a case for also including those assets in the tontine. Sadly, at the moment, there is no available market for real (gold) bills of exchange or for gold mortgages. There are, however, some high quality exchange traded companies that could be classified as ‘gold equities’. These should not be confused with gold mining companies. As such, if offered today, a tontine could be backed by gold coin and ‘gold equities’. The asset mix of gold coin and ‘gold equities’ could be managed in a form of arbitrage by the tontine provider.

When considering retirement products trustees should be particularly aware of counterparty risks and fraud. Default risk of an annuity provider should be considered and quantified with respect to the implications for a member’s retirement income. Fraud risk will be heightened with the lack of liquidity of longevity based products. Trustees must have confidence that high quality assets back the products.

Third party certification of the assets should be required to be provided to members on at least an annual basis. It may be appropriate for assets held outside of a life insurance environment to be held in the name of a custodian, or by a trusted third party.

The level of longevity protection could be capped to limit any risk of bias in the pool. Based on basic retirement needs, a maximum amount of around 800 1/4oz gold coins (currently circa A$400,000) per member seems sufficient exposure to a longevity product.

The main lesson from defined benefits schemes is that product guarantees are difficult and expensive to provide. Life annuity and deferred life annuity providers are likely to face similar difficulties. There is likely to be a cap on the level of systemic longevity risk that can be provided by insurers and reinsurers. Similarly, if payments are guaranteed on a nominal or inflation adjusted basis, the costs of providing assets of sufficient quality is likely to result in a poor product offer. Annuity providers may be encouraged to mis-match assets and liabilities in order to offer more attractive products. This increases the risk of default for an insurer. This is a specific concern if a member has a significant exposure to that longevity product provider.

Given the lack of portability for longevity based products, the structure for holding those assets needs special consideration. In Australia, the assets of life annuities and deferred life annuities are held in statutory funds, with the funds and entities regulated by APRA. Consideration should be given to requiring separate statutory funds for life annuities to reduce the risk of depletion of funds from losses in other classes of business.

The assets of tontine based products need to carry similar protections to statutory funds. It may be appropriate that they are regulated by APRA. As individual members can not withdraw funds there is a heightened risk of a Ponzi-style scheme developing with longevity products.

It will be important to monitor any fees and charges made to the assets of longevity based products. It would be preferable if tontines were structured as mutuals, with members controlling the management company.

If significant fees and charges are allowed to be taken from longevity based products then ‘income efficiency’ will be significantly diminished. This would be particularly the case for members who are under the age of 70-75, where expenses may dominate any survival benefit.

Members must be able to determine, through a vote, which entity has the management responsibilities for a tontine. The management company for a tontine should not be allowed to claim penalty for termination at contract end by members, and management contracts should be limited to three years. For example, a vote of 75% of members should be sufficient to change management companies of a tontine.

Entry into the longevity protection should not be the default option as this would be unfair on those people with lesser health. For unhealthy lives a longevity product may not be appropriate. This does not just include those individuals with ‘a terminal medical condition’. It essentially includes all those who are likely to have a higher mortality rate than used by the pool’s actuaries to determine their longevity benefits. Depending on the fund and the pool this may be a significant proportion of members. This is why product with a longevity component  must be ‘opt-in’. It is not possible for the Trustees to make an assessment if a member is less healthy than what the deemed mortality rate for determining longevity benefits will be for that individual.

As longevity protection products are dependent on member mortality rates, and mortality rates are typically under 1% prior to age 70, there is not a high priority for members to be engaged with longevity products until closer to that age. Indeed there are mis-selling risk if members are encouraged to take out longevity product protection at too early an age.

The payoff profile of default life insurance is quite different to default longevity products. With default life insurance, a member pays a small amount for a large payout on death. With default longevity products, a members receives a small amount to put a large amount of money at risk. Some product providers will try to mask this reality using features such as return of (nominal) capital on death before 75, but underlying such features is essentially a reversal of the product features the member has purchased (while still paying the product fees!).

Individuals need to determine if a longevity product is right for them. This likely requires both medical and financial advice.