Actuarial Thought on Pension Fund Investment Strategy (1957-1985)

As noted above, the exceptional UK inflation experience that accompanied the First World War and its aftermath provided a salutary lesson to pension actuaries on the vulnerability of pension fund solvency to inflation shocks. But this period of extraordinary inflation proved short-lived: between the years of 1921 and 1941, the UK CPI actually fell. However, the post-Second World War period saw the return of high inflation, and, perhaps equally importantly, high levels of inflation volatility and uncertainty. By the mid-1950s, there was a growing sense that the prevailing political and economic environment made this inflationary experience the new normal rather than merely a transient post-war adjustment. The political focus on full employment and the increasing strength of trade unions appeared to make the prospect of a permanent inflationary environment quite plausible.

Pension actuaries of the post-Second World War era recognised the inflation risk exposure that arose when investing in long-term nominal fixed interest assets to back salary-linked liabilities. At this time, there was no index-linked gilt market. The only available ‘real’ asset classes were equities and property. Actuarial thought in pensions started to embrace the use of equities as a form of inflation hedge in the mid-1950s. As we have seen in earlier sections, this was also a period of increased equity investment for life offices, and the ‘cult of the equity’ was more generally in full flight. Pension funds were not immune to this trend—between 1945 and 1954, UK pension schemes’ average asset allocations to equities increased from 10 % to around 30 %.^[1] The inflation hedging argument may have driven this, or may have simply provided an intellectual rationale for it.

An actuarial paper by McKelvey, ‘Pension Fund Finance’, published in 1957, was amongst the first to argue for substantial equity investment by pension schemes on the basis of equities’ inflation hedging ability.^[2] The quality of inflation hedging that equities could provide was accepted as a matter of some uncertainty by McKelvey. He constrained his advocacy of equity investment to open rather than closed pension funds, and emphasised that it was the inflation-protection of long-term dividend growth (rather than equity market values) that he believed was important for such funds. He posited that the ‘propaganda value’ of balance sheet asset valuations of open funds was of no great importance. In essence, McKelvey was arguing that dividends provided a form of real cashflow match for salary-linked benefits. Volatility in real equity price levels could be ignored as an inconsequential market foible for an open pension fund that would never need to sell them. He emphasised that, in real terms, long-term nominal fixed income was the risky asset class for backing salary-linked pension schemes:

The question now is not, as it used to be, dare we put more than 10% in equities? It is, dare we leave more than 50% in fixed interest investments?^[3]

J.G. Day published a paper in the Journal of the Institute of Actuaries in 1959 that provided further support for substantial equity investment by final salary pension funds.^[4] Day emphasised how final salary pension liabilities had different characteristics to life office liabilities—specifically being real rather than nominal, and having longer duration. He argued that these characteristics made pension funds more suited to equity investment than life offices and noted the high levels of equity allocation that life offices had made in recent years. He recognised that equity investment would expose pension funds to more market value volatility, but, like McKelvey, he argued that this was not fundamentally important. He was unequivocal in his advocacy of equities as ‘the basic investment’ of final salary pension funds:

A pension fund by the nature of its liabilities is so vulnerable to both inflation and any movements towards higher wages and salaries and a higher standard of living that, if one takes the view for the normal pension fund, it is income that matters and provided that income is secure then fluctuations in the value of the portfolio do not matter, then in the author’s opinion one should come down on the side of equities (or property) as the basic investment for a pension fund.^[5]

The classic papers on pension fund valuation of the early 1960s that were discussed above also contained further advocacy of equity investment for final salary pension schemes, primarily on the basis of inflation hedging. For example, in the discussion of Heywood and Lander, Hemsted commented:

Rising standards derive from increased productivity which in turn is based on increasing capital backing per employee. Much of this increased capital arises from retained earnings by industrial companies and these retentions in a period of expansion produce increasing equity asset values per share. There is therefore reason to think that ordinary shares have a built-in correction not only for falling money values [price inflation] but also for increasing standards of living [real salary growth].^[6]

Day and McKelvey’s 1963 paper was unsurprisingly supportive of pension fund equity investment given the positions they had taken in their individual papers of the late 1950s. They wrote:

It has been widely and authoritatively suggested that, to meet the dangers of inflation, pension funds should invest in equities. It has been argued that with inflation a business’s real assets increase in money terms ... so that, although inflation may initially cause difficulties and distortions, an equity investment will eventually have increased in earning power in rough proportion to the effect of inflation (or very often rather more, as prior charges lose in value).^[7]

All of the above development in pension fund thinking had taken place under the premise that the raison d’etre of the pension fund was to generate asset income that, together with a stable pattern of contributions from the employer and possibly also the members, would reliably fund liability outgo as it fell due. (There was also an implicit assumption that the uncertainty in long-term real dividends was much less than could be inferred from shortterm equity market price volatility.) The idea of the pension fund as a vehicle whose primary purpose was to ensure the security of members’ accrued benefits was recognised by some actuaries, but sat largely in the background of the development of actuarial thinking on pension funds before the 1970s. However, this perspective would occasionally be considered and could lead to substantially different conclusions. Plymen noted in the discussion of Day’s 1959 paper:

With any but the strongest employers the contingency existed in a slump that the employer might be bankrupt or there might perhaps be severe staff redundancy. Under those conditions, the employer would be unable to subsidize a pension fund prejudiced by depreciation of its ordinary shares. He imagined that that would be regarded as by far the greater threat, so that fixed interest securities would tend to be preferred. He maintained that the strength and security of the employer should be the major factor in determining the equity percentage.^[8]

Plymen was arguing that the nature of the asset risk that mattered to pension funds was complex: asset price falls did not matter so much if the sponsor was healthy; what really mattered was how the pension fund assets behaved in the conditions that were associated with sponsor bankruptcy—that’s when the assets were really needed. Such thinking would play an increasingly influential role in actuarial thinking on pension funds, their valuation and their asset strategy over the following decades. With the benefit of hindsight, it is perhaps striking that it was never more than a footnote in the thinking of the 1950s and 1960s, especially when sophisticated thinking by leading actuaries such as Plymen was in circulation.

Equity allocations continued to grow in the UK defined benefit pension fund sector, both as a result of the strong growth of pension funds’ equity holdings over the 1960s, and with the investment of new contributions. By 1975, two thirds of UK final salary pension funds were invested in equities and property.^[9] The extraordinary volatility of UK equities in 1973-1975, however, provided actuaries with a reminder that equity returns could be driven by factors other than inflation. In his faculty presidential address of 1977,^[10] R.E. Macdonald noted:

The market collapse of 1974 finally provided convincing evidence that any connection between the values of shares and the rate of inflation was mainly fortuitous.^[11]

The 1973-1975 period certainly opened eyes to the degree of volatility that was possible in real equity price levels (despite the exceptionally high inflation rate, UK equities fell by over 70 % in 1973-1974; they then more than doubled in 1975). But McKelveys and Day’s papers had been careful not to suggest that real equity returns were not volatile: rather, they argued it did not matter if they were. Their argument was that real dividend growth could be expected to be stable, and that was what mattered for pension funds (open ones, at least). But the experience of the early 1970s also proved that this faith in the stability of the real value of dividends was misplaced: UK dividend pay-outs fell by around 45 % in real terms between 1970 and 1974 and did not fully recover their real 1970 value until the late 1980s.^[12]

So the UK experience of the early 1970s showed that the much-vaunted inflation-protection of equity dividend income was a myth. Nonetheless, actuarial wisdom remained unperturbed. In his Journal paper^[13] of 1977 on pension fund investment, Holbrook was sanguine on this episode of equity market volatility, noting:

The catastrophic fall in UK equity markets in 1973 and 1974 reflected a serious crisis of confidence, which has not been wholly restored by the subsequent rise. With the benefit of hindsight it may now be said that the main importance, for a long-term investor, of the exceptionally low market level in 1974 was the opportunity it presented to buy shares cheaply.^[14]

Thus, equities and property became the dominant asset class choices of defined benefit pension funds. This was doubtless in part influenced by the wider ‘cult of the equity’ that developed from the second quarter of the twentieth century onwards, and, from a specific actuarial perspective, because they were the only viable real asset class choice of the era with which to back undoubtedly real liabilities. The question of how much protection from short-term inflation shocks or long-term inflationary eras can be provided by equities (relative to bonds) is a question that has been much contested by actuaries and economists for decades. Whilst it is intuitive that the real characteristics of an equity claim implies some form of inflation protection, the question remains of how material a driver of returns (and dividends) inflation is relative to the many other factors that drive the behaviour of equities (over the short and long term).

A quantitative framework for analysing liability-orientated investment strategy was proposed by A.J. Wise in his 1985 paper ‘The Matching of Assets to Liabilities’.^[15] Wise defined a matching portfolio in a very particular way, which encompassed a much wider potential set of portfolios than the ‘absolute matching’ of cashflows that was first defined by Haynes and Kirton back in 1952: Wise’s matching portfolio was defined as the asset portfolio that minimised the volatility of the ‘ultimate surplus’ of the book of business, i.e. the residual (positive or negative) asset value after the final liability cashflow had been paid. Importantly, he added the stipulation that the matching portfolio could not be re-balanced over the projection horizon.

Wise’s matching portfolio could differ from a cashflow matching portfolio for a number of reasons: the universe of investable assets may not be granular enough to exactly cashflow match (i.e. there may not be a bond with a maturity date exactly coinciding with the liability cashflow); the liability cashflow could be exposed to factors such as salary inflation that may not be directly obtained in available assets; the liability cashflows may have non-linear features that are not present in available assets (and that could not be produced by a dynamic rebalancing strategy for the asset as such strategies had been outlawed by Wise’s matching portfolio definition).

Wise’s stipulation that the matching portfolio must be a static one came more than a decade after Black-Scholes-Merton had shown how fundamental the idea of dynamic rebalancing was to liability matching (which in Black- Scholes-Merton was just a means to arbitrage-free pricing by replication). The static assumption also ran counter to the actuarial idea of immunisation introduced by Redington in 1952 which focused on what might be called instantaneous matching and recognised the need to rebalance through time to maintain the match. But Wise believed that there was something inherently imprudent about including dynamic rebalancing assumptions in a risk modelling exercise. In introducing his paper at the Staple Inn discussion, he considered practices in the life field where modelling was comparatively more developed than in pensions, and asked:

When assessing the solvency of a life office is it safe to assume that the office will always be able to take advantage of investment conditions in the future so as to keep itself in an immunised position? Would it not be safer to identify a minimum risk matching portfolio for solvency purposes on the assumption of no future investment activity except as required by new money?^[16]

I.C. Lumsden, a senior life actuary, responded:

It does seem to me unrealistic to consider matching portfolios only on the assumption of passive investment policies, and a pity in particular to ignore the one dynamic strategy — immunisation — which has gained our acceptance in the past.^[17]

Wise went on to propose that the market value of the matching portfolio as defined above for pensions liabilities could be considered as a form of liability valuation. On the surface this seemed quite natural, and not dissimilar to the economic idea of arbitrage-free pricing by replication. But Wise’s definition of a matching portfolio was profoundly different from an economist’s notion of a replicating portfolio. Wise’s matching portfolio could leave the matcher significantly exposed to market risks—so whilst the portfolio might generate an expected surplus of zero, it may still have a significant economic cost. The identification and cost of Wise’s matching portfolio could also be a function of subjective assumptions about long-term asset risk premia—again a feature that ran contrary to the properties of market-based valuation techniques.

It could be said that much of the confusion that has historically arisen between actuaries and financial economists has been due to a lack of clarity between the ideas of expected funding costs (based on expected long-term asset returns) and replication costs (based on the current market prices of replicating assets). Wise’s matching portfolio definition and its use in liability valuation could only add to this confusion. Wise compared pension fund liability valuations produced by his matching portfolio concept with the more straightforward market-consistent approach of discounting liability cashflows with the observed market yield curves. He expressed the traditional actuarial discomfort with such direct use of market prices:

This [market-consistent value] result might be considered realistic on the grounds that current redemption yields on secure fixed interest stocks are the best available guide to future economic conditions. Whether this is so is a somewhat philosophical point; it is certainly not generally accepted.^[18]

For pension actuaries, uncoupling expected funding costs and economic values appeared interminably challenging. But the distinction would only become more relevant.

• [1] McKelvey (1957), p. 136.
• [2] McKelvey (1957).
• [3] McKelvey (1957), p. 121.
• [4] Day (1959).
• [5] Day (1959), p. 130.
• [6] Hemsted, in Discussion, Heywood and Lander (1961), p. 365.
• [7] Day and McKelvey (1963), p. 107.
• [8] Plymen, in Discussion, Day (1959), p. 152.
• [9] Holbrook (1977), p. 58.
• [10] Macdonald (1977).
• [11] Macdonald (1977), p. 9.
• [12] Dimson, Marsh and Staunton (2002), p. 152.
• [13] Holbrook (1977).
• [14] Holbrook (1977), p. 17.
• [15] Wise (1985).
• [16] Wise (1985), p. 65.
• [17] Lumsden, in Discussion, Wise (1985), p. 71.
• [18] Wise (1985), p. 55.