Managing Risks in Retirement
In the long run, the equity premium means that investments in equities produce lower pension costs. The price is significant short-term volatility: falls of 30 per cent and more within a year have been commonly experienced.
Once they have retired, it is likely to be difficult for fund members to re- enter the employment market to make up for investment losses. Members therefore need protection against such significant falls in income after retire- ment. This protection should probably begin before the retirement date, as plans for retirement presumably take some time to implement. Superannuation funds therefore need investment instruments that offer protection against investment fluctuations over this relatively long, but finite, period.
The strategy employed by most DC funds and their members is to use a combination of investments in order to give an optimum mix of security, infla- tion protection and participation in the equity premium. To the extent that the equity premium arises from the correlation of share prices with the universe of consumption, it may be possible to develop other assets that share this risk without the short-term volatility. One possibility would be to develop invest- ments that provide a cash flow that could match pensions, and was linked to a wage index. This is the approach taken by the new Swedish notional DC national scheme. Asher (1994) suggests a method for utilizing housing finance.
In the absence of these instruments, the next section describes how volatile investment returns can be smoothed to produce a more acceptable income flow, and provides a family of algorithms to do this fairly and efficiently.
member’s final salary. Given that final salary can be distorted by recent promotions, or a reduction in working hours related to ill health or impending retirement, it is not clear, however, that this is an appropriate benchmark.
Delaying Action
One approach to smoothing investment returns is to delay the decision to change crediting rates until the end of some time period. This is not as alarm- ing as it first appears. Unit trusts will normally delay transacting until the end of a day. Many unit-linked DC funds will allocate units at the end of a month, while non-linked funds will apply the same crediting rate for a year at a time.
In almost every case, however, they will reserve the right to calculate unit prices, or change the crediting rate, at some intermediate time if market condi- tions change and some participants may otherwise be disadvantaged. This could happen, for instance, if market values dropped dramatically. In this case, failing to recalculate unit prices will mean that departing members are given more than a fair share of the assets – to the detriment of remaining members.
The inequity may be aggravated, and the solvency of the fund perhaps threat- ened, if some members are able to elect to leave in order to take advantage of the artificially high unit price. This risk of some members making elections that harm others is called anti-selection in the actuarial literature. Delaying action should, however, be seen more as an administrative convenience than a method to address market fluctuations.
Smoothed Bonuses
Many DB and DC funds historically operated through policies with life insur- ance companies. The companies declared smooth bonuses on these policies to distribute investment and other profits. DC funds often declare their own smoothed bonus or crediting rates.
One actuarial method of determining the bonus is to begin with a calcula- tion of a smoothed value of assets. Head et al. (2000) provide a description of their development and justification in the context of DB funds. One common approach is for the value of shares to be determined by discounting future divi- dends at the actuarial valuation rate. An alternative is to assume that share prices will (immediately) return to a level that reflects some long-term aver- age dividend yield. Other approaches to smoothing involve some averaging of the market value of assets.
To the extent that these actuarial values for the assets differ from their market values, they are, arguably, unrealistic. Actuaries who use the method respond that their smooth actuarial value represents a more realistic estimate of the long-term value of assets than the market price. Market prices may well
Smoothing investment returns 149
be overly influenced by the needs or views of those who happen to be trans- acting at that time.
Even if this were true, three major problems arise. The first is the possibil- ity of anti-selection. Incoming policyholders or members benefit if the differ- ence between market and actuarial value is positive; maturing policyholders and exiting members gain when it is negative. They thus have the incentive to make decisions that disadvantage other members.
The second is that when market values are below actuarial values, the fund is exposed to the risk that a combination of further falls in market values and an outflow of funds may lead to an unacceptable drop in the benefits available to the remaining members, or even to insolvency if the fund is entirely exhausted. This problem may be further aggravated by anti-selection. Fund trustees and company managers are therefore understandably reluctant to allow a shortfall of assets to become significant. This reduces their ability to cushion falls in benefit payments – which is the original reason for smoothing.
The third problem is more subtle. As a result of the reluctance to pay bene- fits in excess of the market value of the underlying assets, most smoothing involves limiting the increase in the crediting rate when market values rise rapidly. This creates a tendency for smoothing reserves to build up with no obvious methodology of release. In the absence of protection for policyhold- ers or members, the resulting surplus is open to expropriation by stronger and opportunistic parties. Asher (1991) discusses one such incident, where an
‘orphan estate’, built up from the contributions of previous generations of poli- cyholders, was expropriated by a new shareholder.
Clay et al. (2001) discuss UK with-profit policies and make various recom- mendations as to disclosure and the protection of vulnerable parties. They believe, however, that smoothing necessarily requires discretion on the part of the fund’s governing body, in order to ensure equity and solvency. The discus- sion of their paper reflected widespread doubt as to whether the three prob- lems discussed above can be resolved without using an objective algorithm for smoothing.
Alternative Algorithms
Thomson (1997) suggests an algorithm that optimizes measures of smoothness and solvency in order to produce both an investment policy and a crediting rate for a DC fund. The results appear to avoid negative bonuses (which are not easily understood by members), but produce more volatile bonus rates than would be common practice.
Khorasanee and Ng (2000) propose an arrangement where the contribu- tions made by, or on behalf of, active members are explicitly reallocated to retiring members in order to allow for smoother returns. As pointed out in the
discussion by Sze (2001), this approach does not appear to be fair, particularly if investment returns are poor and the fund is declining.
Blake et al. (2003) investigate the benefits of a smoothing algorithm that truncates higher and lower returns so that bonuses lie within a limited range.
Blake et al. do not, however, provide a justification for the method. All three methods are vulnerable to anti-selection.