Annuities are insurance products that provide a payment stream for as long as the insured
individual is alive. According to a new study by Professor Eytan Sheshinski, published in
the January 2007 Economic Journal, competitive insurance markets tend to lead to
differential pricing of annuities according to the ‘risk class’ or longevity prospects of the
buyer. These prices entail lower consumption and later retirement of longer-lived individuals
relative to shorter-lived individuals.
Much work has been undertaken by economists to understand and rationalise the ‘annuity
puzzle’: the fact that the markets for private voluntarily purchased annuities are extremely
thin. In the United State, for example, only 3% of retirees hold annuities issued by
insurance firms (according to Brown et al’s 2001 report, The Role of Annuity Markets in
Financing Retirement).
One common explanation is that social security mandatory contributions and indexed
benefits ‘crowd-out’ private annuity markets. Since the social security system provides a
‘replacement rate’ (the ratio between pension benefits to income during the working phase
of life) of between 28% in the United States and 40% in the UK (for the average
participant), this cannot, in itself, be an adequate explanation.
A 2004 survey of attitudes to annuitisation among midlife individuals conducted in the UK
by Gardner and Wadsworth – Who Would Buy an Annuity? An Empirical Investigation –
found that over half of the respondents would never want to annuitise. By far, the dominant
reason for choosing not to annuitise was a preference for flexibility. Uncertainty about
health and income were considered most important in the need for such flexibility.
Indeed, it is difficult to predict individual survival probabilities and other risks early in life.
Consequently, risk-averse individuals have an interest in insurance against such risks,
which involve transfers across ‘states of nature’. In private markets for annuities, firms
attempt to classify purchasers according to their 'risk class' or longevity prospects. In
particular, they sell annuities to those in poor health at lower prices for a given pension
level.
The UK has a thriving and growing market in ‘impaired life’ annuities. These offer additional
income in return for evidence supporting substandard lifespan (it is estimated that in excess
of 10% of all pension annuities in the UK are written on this basis). The Brown et al report
finds evidence that those who purchase annuities closely before retirement (in the United
States, the average age of annuity holders is 62) have significantly higher than average life
expectancy. This 'adverse selection' leads to high annuity prices that reflect the low
mortality rates of annuitants.
Differences in mortality rates across different risk classes provide strong inducement for
competitive annuity suppliers to price annuities separately for each risk class. The result is
often a ‘separating equilibrium’ in which prices support a profit break-even condition for
each risk class.
Professor Sheshinski’s study demonstrates that under certain conditions such a competitive
equilibrium is inferior to an optimum social security system with a mandated retirement age
independent of risk class, which provides lifetime benefits at a uniform price.
This stark conclusion has to be qualified when considering cases where the ‘first best’
optimum involves different retirement ages and consumption levels in different states of
nature (due, say, to morbidity effects). In these cases, a social security system that does
not provide adequate flexibility in choosing the retirement age and the level of contributions
may be inferior to a competitive annuity market.
The major results of this study are as follows:
1. When utility of consumption and labour disutility are independent of survival
probabilities, the first best allocation entails uniform consumption, transfers
from shorter- to longer-lived individuals and an optimum retirement age
independent of risk class realisation.
2. A social security system that provides uniform benefits and a mandated
retirement age is potentially first best.
3. Competitive insurance markets tend to yield a separating equilibrium with riskclass
pricing of annuities. These prices entail lower consumption and later
retirement of longer-lived relative to shorter-lived individuals. In these
circumstances, optimum social security dominates the competitive equilibrium.
In more general cases, it may be impossible to rank (for all preferences and
any survival function) the competitive equilibrium and optimum social security.
ENDS
Notes for editors: ‘Optimum and Risk-class Pricing of Annuities’ by Eytan Sheshinski is
published in the January 2007 issue of the Economic Journal.
Eytan Sheshinski is at the Hebrew University of Jerusalem.
For further information: contact Eytan Sheshinski on +972-2-5883144 (email:
mseytan@mscc.huji.ac.il); or Romesh Vaitilingam on 07768-661095 (email:
romesh@compuserve.com).