«Abstract: The UK requires individuals with individual accounts to annuitize before the age of 75. Using a time series of annuity prices and ...»
Table 3 provides a summary of market rates for level annuities on April 30, 1999. The table records the annual payment offered against a purchase price of £10,000 by age and gender. We report information separately on open-market options and compulsory purchase annuities, and consider the market average (median), the fifth best rate in the market, and the top rate. Payments are made monthly in advance and there is no guarantee period.
From the top panel we see that a 65-year old man purchasing a typical £10,000 level annuity in the market for open market options could expect to In addition to these factors, the precise rate depends on payment frequency (monthly, semi-annually, or annually), whether payment is in advance or in arrears, whether there is a guarantee period, and, in the case of joint life annuities, the size of survivor benefits. We concentrate on policies where payment is monthly in advance, with no guarantee and no reduction in survivor benefits.
The exact number of providers in each market varies by type of annuity.
receive an annual payment of £851 for life. A woman of the same age could expect to receive £749 for life, or about 12 percent less, on account of higher longevity. A 65-year old couple (where the man is 65 and the woman 60) would receive £623.9 a year for life, or about 27 percent less than a single male. Not surprisingly, annuity payments generally increase with age, but the increase is not as large as may be expected. Thus a 70-year old man would receive an annual payout of £1013 a year, almost 20 percent more than a man five years his junior. For women, age also raises the annual payment, but by less than for men. A 70-year old woman could expect to receive around 14 percent more than a woman 5 years her junior. For joint and survivor annuities, the increase is even lower. A 70-year old couple (where the man is 70 and the woman 65) would receive about 10 percent more than a couple 5 years younger.
Table 3 gives an indication of the degree of dispersion at the upper end of the market for open market options. In general, there is a small difference (1-2 percent) between the average rate and the fifth best rate. Differences between the average and the top rate are greater, of the order of 6-8 percent.
This suggests that the top half of the market tends to be fairly clustered, and the benefits of shopping around for the best rate in this segment of the market do not vary significantly by age and sex.
Table 3 also suggests that annuity rates do not vary significantly between open market options and compulsory purchase annuities. Indeed the rates in the two markets are fairly close, and the pattern of dispersion is also broadly similar. The best rates in the compulsory purchase market, however, appear higher than in the market for open market options.
Mortality: We use two types of information on survival probabilities in the value for money calculations. The first relates to the expected mortality experience of the general population, and is based on tables prepared by the Government Actuary's Department (GAD).38 The second relates to the expected mortality experience of individual annuitants and is derived from the latest report of the Continuous Mortality Investigation Bureau.39 Both sets of tables are cohort tables as opposed to base (or period) tables and are discussed in further detail below.
The life tables available from GAD incorporate the latest expectations on mortality improvements among the general population.40 The starting point is a base table incorporating the mortality experience of the general population during the period of the latest Census in 1991. Mortality improvements are grafted on to this table, derived both from improvements in mortality in the recent past and a reasonable guess as to likely improvements in the future.41 We use tables for the UK population and cohorts attaining the ages of 65, 70, and 75, in 1998-99. The figures provided in Table 2 represent the expected mortality of individuals aged 65 in 1998-99.
For annuitants, we use the tables derived from CMI Reports No. 16 and
17. CMIR No. 16 provides base tables derived from the experience of annuitants during the period 1991-94 (base year=1992), while CMIR No. 17 includes projection factors to allow for improvements in mortality with the passage of time. The improvement factors are based on trends since 1975, subject to certain consistency requirements.42 We use the base tables and projection factors to estimate the expected mortality of cohorts attaining the GAD provides actuarial advice to government departments both in the UK and overseas on matters such as social insurance, state pensions and insurance company supervision.
The CMI Bureau carries out research into the mortality and morbidity experience of individuals covered by long term risk contracts issued by life offices in the UK and the Republic of Ireland. Since 1973, its findings have appeared in regular CMI Reports.
We are grateful to Steve Smallwood of the Government Actuary's Department for providing these data.
Broadly speaking, the mortality improvements approach a rate of 0.5 percent per annum over a 40year horizon. While these are the current predictions, they may be revised upwards or downwards in the light of new information on the pace of mortality improvements.
Mortality at each age is assumed to decrease exponentially over time to a limiting value, with the speed of convergence depending on age. At t=∞, the reduction in mortality varies between 45-70 percent. See Part 6, CMIR No. 17, available from http://www.actuaries.org.uk.
ages of 65, 70 and 75 in 1998-99. The figures for individuals attaining 65 in 1998-99 are reported in Table 2.
The CMI Reports separate the mortality experience of two types of annuitants: ‘pensioners’ and ‘annuitants’.43 Previously, pensioners included all individuals with pensions and annuities policies with life assurance offices in the UK. However, CMIR No. 16 and 17 separate those with Section 226 retirement annuities (‘annuitants’) from other pensioners. According to these reports, the mortality experience of the two groups is quite different, with annuitants displaying lighter mortality than pensioners. However, the reasons for these differences are not entirely clear. For the purposes of this paper, we use the pensioners' tables for both markets.44 Interest rates: We use two sets of interest rates in our analysis. The first is the term structure of yields on UK government bonds on April 30, 1999.
These were obtained from the Bank of England and are based on their estimate of zero coupon nominal and real yield curves. The data are for maturities between 2 and 25 years. For earlier and later years, we assume that the yields are similar to immediately adjacent periods. So the 1-year yield is assumed to be the same the 2-year yield, and yields beyond 25 years the same as the 25-year yield.
In addition to this riskless term structure, we also consider the term structure of yields on corporate bonds on the same date.46 The data provide information on the average difference in yields (swap spreads) between corporate bonds and government bonds for maturities between 2 and 10 years. We add this difference to the term structure of zero coupon nominal The relevant mortality tables are referred to by their acronyms: for pensioners, PML/PMA (lives or amounts); for annuitants, RMV/RMD/RMC (vested, deferred, combined).
In general, using the annuitants tables for valuing the open market options makes OMOs appear uniformly better in terms of value for money, with MWRs typically greater than 1.00.
For discounting nominal cash flows we use the term structure of nominal yields. For RPI-indexed annuities, we treat the payout as fixed in real terms and discount using real interest rates.
We are grateful to Vasant Naik of Lehman Brothers for making these data available to us.
yields to obtain the term structure of yields on corporate bonds. We assume that the yield differential for maturities less than 2 or greater than 10 years is the same as in adjacent years.
Money's Worth Ratios Tables 4 through 7 present our main findings on the value for money of annuities in the UK. Computations use the average (median) payout rates on April 30, 1999 to estimate money’s worth ratios (MWRs) by sex and age. We consider both single and joint-life annuities, and consider the market for open market options and compulsory purchase annuities separately. In each table we use at least two sets of cohort mortality tables, one for the general population and one for the population of annuitants. For the annuitant population, we generally use the ‘pensioners’ PML92 table with mortality improvements unless stated otherwise.
Table 4 examines money’s worth ratios for level annuities using two sets of interest rates, the term structure implicit in the zero coupon government bond yield curve, and the related term structure of corporate bond yields. In Table 5, we examine whether the value for money varies by type of annuity, focusing on level, escalating, and indexed annuities. In Table 6 we consider whether annuity costs vary by policy size, comparing the value for money of a £10,000 policy with a £100,000 policy. In both Tables 5 and 6 we use the risk-free term structure. The implications of using a more risky term structure are easy to infer on the strength of Table 4. Finally, in Table 7, we examine changes in the value for money over time.
Before discussing the tables in greater detail, it is worth summarising our main findings.
• First, the average retiree facing the mortality prospects of the general public, would perceive a significant financial cost in purchasing an annuity from a typical life assurance company. This cost varies from 10-12 pence in the £, depending on age and sex. (In other words, money’s worth ratios lie in the range of 0.88-0.90.) Much of this cost is accounted for by selection effects, that is, the higher than average longevity of the annuitant population. Our estimates suggest that around 7 pence of the ‘cost’ is due to effect of selection. (Money’s worth ratios using annuitant life tables are in the range of 0.95-0.97.) By implication, the remaining 3-5 pence is due to remaining factors, including the administrative charges of life assurance companies. These findings are also consistent with those of Finkelstein and Poterba (1999). It is not clear to what extent the selection effects are due to the socio-economic characteristics of the annuitant population rather than individual behaviour.
• Second, using higher discount rates to account for the possible risks associated with future income streams from annuities would give us higher cost ratios (lower money’s worth ratios.)
• Third, the costs of annuitization vary significantly by product type, particularly between nominal and real annuities. The average retiree can pay an additional 8-10 pence in the £ for purchasing inflation protection in addition to longevity insurance. (Money’s worth ratios are typically 0.08lower for RPI-indexed annuities than for level annuities.) It is difficult to relate these additional costs to the additional risks borne by insurers, since insurers may cover their inflation risk by investing in inflationindexed government bonds. The higher costs may have to do with selection effects. As the real payout does not fall over time with RPIindexed annuities, they would tend to attract those with higher expectations of longevity. This is not something we can investigate fully as we do not have annuitant tables by product. However, we observe that the additional cost of inflation protection is just as high for the average annuitant as for the average member of the population.
• Fourth, annuitization costs fall with policy size. The difference in cost between a £10,000 policy and a £100,000 policy is about 2-3 pence in the £.
Assuming that larger policies attract annuitants with higher longevity, these findings suggest that the saving in administrative and other costs probably overwhelm selection effects in larger policies.
• Fifth, we do not observe any trend in value for money of annuities, at least in the last five years.
• Sixth, there do not appear to be any significant differences in costs in the market for open market options compared to compulsory purchase annuities.
We now turn to a more detailed discussion of our findings. Table 4 shows the value for money for level annuities for men and women aged 65, 70 and 75. In general, the money’s worth ratios are all less than 1.00 implying that an average retiree is likely to face a cost in converting an accumulated pension fund into an annuity. We begin by considering the value for money when zero coupon yields are used to discount future payments. For a 65-year old man facing the mortality prospects of the general public, the MWR is
0.897. For a 70-year old man in similar circumstances, the MWR is lower, at
0.875. Among men, the value for money generally declines with age. The same pattern is upheld among women (although to a lesser degree than men), and among couples. Controlling for age, we also observe that the MWRs are lower for men than for women, and are the greatest for couples.47 Throughout our analysis we evaluate MWRs among couples assuming their individual mortality probabilities are independent of each other. However, evidence suggests that individual mortality probabilities among couples displays positive dependence. This would alter our calculations slightly.
Since annuitants live longer than the general public, using the annuitant tables should give us uniformly higher MWRs. This is borne out in Table 4. For example, a 65-year old man would assess the value per premium £ at 89.7 pence if he perceives himself to be facing general mortality risks, and at 96.6 pence if facing the mortality risk of the annuitant population.
Likewise, a 65-year old women would perceive the value per premium £ at 97 pence as opposed to 94 pence if her risk of dying corresponded to the annuitant public. The observation on declining MWRs with age is upheld when annuitant tables are used. However, controlling for age, it is no longer the case that value for money is higher for women than for men. In fact, the reverse is found to be true.