Catherine Donnelly discusses the new drawndown products that increase returns through longevity credits.
Elderly people outliving their pension savings is a genuine risk in a drawdown world. As the workplace pensions system gets under way in the UK, but with default contribution rates too low to provide a reasonable chance of a comfortable retirement, this problem will increase.
New products that pool longevity risk can reduce the chance of a poor retirement income. Pooling enables pensioners to generate returns that are in addition to their investment returns. While the risk of outliving one's retirement savings is not eliminated, it is mitigated. This is a massive benefit in a drawdown environment.
The reason for the reduction in longevity risk is simple: the members of the pool become the beneficiaries of each other. When one of the pool members dies, the value of their allocated assets is shared out among the pool's participants. The value received by each member is called a longevity credit.
A longevity credit is paid only when someone dies. In this way, pensioners gain from pooling if they remain alive. They do not lose any of their money, except upon their own death.
As a simple example, the pooling can operate in a unit-linked fund environment. Pensioners can choose units as they would normally, but with the chosen units earmarked to be part of the pool. Upon the death of a member, each pool participant receives a proportion of the value of the deceased member's units. This could be paid out as a cash sum or re-invested in additional units, according to the stated preferences of each participant.
In the past, the popular choice for guaranteeing a lifetime income was the life annuity contract. The insurance company bears all the risk, the cost of which will be passed onto the customer. Is it necessary for all these risks to be eliminated for the customer, though? If customers benefit from the pooling of lives while maintaining investment control of their assets, they can reduce the risk of outliving their assets without having to pay for costly investment and longevity guarantees.
Consider a 67-year-old man who has the choice of investing £100,000 in either a drawdown product or a pooled product. He wishes to have a an income of £5,000 per annum, withdrawn at the start of each year. For simplicity, assume there are no investment returns. Suppose that the longevity credit received is equal to its expected value and is added to his wealth at the start of each year.
Figure 1 shows the development of his fund value, according to how much of his asset value is assigned to the pool. Pooling 50% of his fund value at all times, he never runs out of money while alive. Pooling 35% of his total fund value, he exhausts his fund at 101 - still a valuable result. In the drawdown product, he has no money left at 87, having had 20 years' payment of £5,000 per annum.
Figure 1. Fund value at each age for a pensioner who withdrawers £5,000 per annum and has no investment returns, according to what percentage of their assets is allocated to the pool.
The example above is, of course, idealised. In practice, longevity credits will be volatile in size and frequency. As investment returns are also volatile, the calculation of the overall risk of a pensioner's financial position needs to take account of the volatility of investment returns and longevity credits together.
The volatility of the longevity credit depends on the composition of the pool. Numerical results in Donnelly, Guillén and Nielsen (2014) show that pools can operate with only a few hundred members, assuming moderate heterogeneity in the age-wealth membership profile.
What about the pensioner's investment choice? Is it restricted, and do they have to invest the same way as everyone else in the pool? Not at all. One could choose to invest entirely in equities, another in cash. With explicit mortality pooling methods, the decision to pool assets with others is separate from the decision of how to invest those assets.
The separation of the return on assets into two parts - investment return and longevity credit - allows further innovation. For example, an insurer could offer to remove the downside risk that the longevity credit is less than expected, for a fee. This is explored in Donnelly and Young (2017).
While the benefits of mortality pooling are clear, what about the practical implementation?
To ensure fairness among members, a reasonable estimate of everyone's mortality risk is needed. The estimate of mortality risk can be updated as often as desired, to reflect current expectations of mortality.
However, adverse selection must also be considered. A person who is in excellent health may receive a larger longevity credit than they should, on an actuarially fair basis. The opposite goes for someone who is poorly.
One solution is to have a waiting period before people can join the pool - such as 10 years. This takes the viewpoint of solidarity: at the outset, few will know if they will be sick or healthy in 10 years' time. If the pensioner dies before the period has elapsed, all their asset value is paid to their estate, rather than to pool participants. After the waiting period has elapsed, rather than trying to track everyone's health status, average mortality risk estimates could be used to calculate longevity credits.
A product that pools longevity risk, Mercer LifetimePlus, has already been launched in Australia. While the pooling mechanism is not revealed in the product literature, its existence demonstrates that there is a commercial appetite for such products.
In an IFoA autumn webinar, Catherine Donnelly will outline methods of explicitly pooling mortality risk. The study of new mortality pooling products is part of the IFoA's Actuarial Research Centre (ARC), within the funded project ‘Minimising longevity and investment risk while optimising future pension plans’. It is led by Catherine Donnelly of Heriot-Watt University and Jens Perch Nielsen of Cass Business School, City University London.
Keeping it in proportion
Several mechanisms for sharing out longevity credits are known. The general approach is that each member of the pool receives a credit that is proportional to their own risk of dying and their own asset value (where only the assets allocated to the pool are counted).
An attractive feature for any mechanism is actuarial fairness: members should not expect to lose or gain from their participation in the pool. This means that the expected value of the longevity credits received by a member while they are alive should equal the expected value of their own asset value conditional on their death. The advantage to members is in the timing of when they gain and when they lose. They 'gain' while they remain alive, and only 'lose' upon their death.
The mechanism proposed by Donnelly, Guillén and Nielsen (2014) is actuarially fair and works for any group of participants, no matter what their individual asset values or mortality risk. This and other mechanisms, and how they can be integrated into products for retirement, are studied within the IFoA funded project 'Minimising longevity and investment risk while optimising future pension plans' as part of the Actuarial Research Centre.
Catherine Donnelly is an associate professor in the Department of Actuarial Mathematics and Statistics at Heriot-Watt University, Edinburgh.