Helping clients understand annuity value

What’s theoretically rational for an investor may not be what most people actually do. This could be because of behavioural factors, or there could be deeper, and quite rational factors.

So it is with lifetime annuities. Academic research suggests they’re superior to other products in providing a lifelong income, and the earlier they are bought, the better it is for clients.

But, the behaviour of consumers, at least American consumers, runs contrary to simple logic. Instead of buying annuities when they are young and policies are cheap, they take the plunge when they are, on average, 62- years-old. Further, most buy term-certain annuities, rather than lifetime ones. With a term-certain annuity, payouts can be made for 10, or 20 years, and if the annuitant dies before the end of the term there could be a lump-sum payment to a beneficiary.

By contrast, with a lifetime annuity, payouts only cease at death.

An analysis of this annuity puzzle was one of the presentations this week at the annual conference of the Individual Finance and Insurance Decisions Centre, which focused on retirement income analytics, including securitization models for annuities and sustainable withdrawal rates from a retirement-income portfolio.

But a hard look at the annuity puzzle determines the 62-year-olds may be rational, says Eytan Sheshinski, a professor at both Hebrew University and Princeton University. These late-entry consumers are likely to get 95 or 96 cents on the dollar. By contrast, in view of the mortality tables applied to the general population, the average consumer will only get 70 cents on the dollar.

This illustrates the problem of adverse selection. When an insurance company sets up an annuity pool, it knows nothing about an individual’s characteristics. As those who expect to be long-lived become annuitants, they are less profitable — hence riskier — for the insurance company. And, when annuities are bought later in life, prices will reflect this adverse selection.

Of course, the inverse of adverse selection is self-selection: individuals chose the policies from which they can most benefit. Buying annuities young mitigates that problem — with lower chances of early death, annuities are cheap. But other concerns come into play: Many clients are hesitant to lock up their money since they might need it for health care or a stay in a nursing home.

To solve the problem, Sheshinski suggests refundable annuities. Young people could buy something along the lines of a ladder of annuities, each with a preset refund price. Should they need money later, they could cash out the annuities, starting with the highest valued ones.

A second option, for those concerned about future health issues, would be to bundle an annuity with life insurance. The life insurance would pay out more for current needs, while the annuity benefit would be reduced. Sheshinski calls this a “life care annuity,” and it makes sense from an insurer’s perspective because out-of-pocket health expenses probably correlate negatively with longevity — although they could be a function of income, Sheshinski notes. It’s also possible to use these structures to provide annuities that leave bequests.

Pricing Risk

Still, in offering annuities, insurers have to hedge their risks, not least to provide a more liquid market such as Sheshinski suggests.

Modelling annuity risk provides an opportunity to expand risk-transfer markets, suggests Michael Sherris, a professor at the University of New South Wales. While demographic mortality models are widely available, the key is transforming them into financial models so that risk can be priced.

There are models for financial risk pricing used in the reinsurance world and through modest issues of insurance-linked securitizations. Transferring the longevity risk inherent in annuities could be achieved on contemporary securitization methods, for example collateralized debt obligations.

But CDOs are out of favour because the collateral dried up during the financial crisis. With a longevity bond, the collateral would be government bonds, not the assets that underlie mortgages, auto loans and credit card debt. And, unlike the banks which sponsored CDOs with dodgy collateral, “you want it managed by someone who is reliable,” Sherris adds.

Still, the securitization would proceed the same way as a CDO, with senior, mezzanine and junior or equity tranches. There would be progressively higher interest coupons to reflect higher chances of default. In this instance, default occurs when the insurer’s actual payouts exceed the expected payouts as provided by a model; thanks to longer-lived annuitants.

Sherris’s efforts on expected payouts involve mortality data on Australians aged 60 to 89 from 1971 to 2004. He finds the volatility of mortality increases substantially after age 85, but is relatively stable from age 60 to 80. That variability is less than the equity market volatility of 20% to 30%, or interest rate variability of 15% to 20%. Nevertheless, at 5% to 10%, it’s a significant factor in whether insurers price longevity risk adequately or not.

Managing Longevity

But, whether or not insurers can develop a broader market for transferring risk, individuals still have to manage their longevity risk. And that discussion turns on sustainable withdrawal rates from a retirement income portfolio, to obviate the risk of ruin. It also encompasses the minimum individuals need to save before retirement to provide an adequate income.

Nabil Tahani and Chris Robinson, professors at York University’s School of Administrative Studies, have modelled sustainable withdrawal rates.

To map out sustainable retirement spending, financial planners need to know the date of death, rate of return and rate of consumption. These can be assumed by methods such as using mortality tables for projected death, taking historical rates of return and putting a cap on consumption (traditionally, 70% of pre-retirement income). But these assumptions don’t capture the randomness of death, returns and consumption.

The withdrawal scenario assumptions in traditional models typically apply stochastic methods to simulate the random walk of mortality and rates of return, but not to consumption. Tahani and

Robinson say that consumption could be random too, albeit in some instances dependent on rates of return. Thus, a particularly good year on the markets might lead to higher spending, and a bad year to cutting back. Then again, consumption in retirement may also be on the decline, compared to consumption before retirement.

That analysis yields three scenarios. Someone with a constant spending pattern — what they call the “socialite” — could count on a 3% withdrawal rate. Someone who might face increasing expenditures in retirement, thanks to ill health, would be constrained to a 2% rate. They call this the “uninsured.” Finally, there’s the “gardener,” who expects declining expenditure over all, compared to his or her work life. That person could withdraw up to 6%.

In a similar exercise, Tahani and Robinson address the relationship between initial wealth and future savings and its bearing on how much equity exposure consumers should have. Take a 40-year-old couple with $100,000 saved. Their goal at 65 is $500,000 to supplement their pensions. They plan to save $7,000 a year. What proportion should be in equities?

Assuming the risk-free rate delivered by T-bills, they can’t reach their goal with no exposure to equities. They would need a balanced portfolio, with up to 90% exposure to equities. And , even then, the probability of a shortfall is 32%. In itself, Tahani and Robinson argue, this calculation is a valuable exercise for communicating risk to clients.

But there is an optimal solution. Increasing annual savings from, say, $7,000 a year to $9,500 raises the probability of reaching the goal to 85%, while reducing equity exposure to 11%.

That seems to be Tahani and Robinson’s benchmark: clients should have an 80% chance at reaching their goals. Their approximation is not perfect — “but it does a great job” in cutting to the chase.