In the Guest Viewpoint column of IPE March 2012, Kees Cools and Anton van Nunen claimed that the current calculation used in assessing the health of a pension scheme is incorrect. They also claimed it had forced pension schemes to sell ‘cheap’ equities in favour of ‘expensive’ sovereign bonds, and that this selling has depressed prices of equities.
As for calculating the health of a pension fund, their arguments ignore conventional academic theory developed by Nobel laureates and distinguished academics such as Robert Merton, Myron Scholes, Fischer Black, John Hull, Stephen Ross and Richard Roll, regarding future financial claims and the reality of a $1.2 quadrillion derivatives market, based on those fundamentals. So much for the claim by Cools and van Nunen that ‘economic theory’ is very clear that future liabilities are to be discounted by the expected investment return.
We will be arguing two simple principles: that risk-free pricing is academically and practically the prudent way of assessing the economic value of a pension scheme; and that evaluating the regulatory solvency of a pension scheme with a risk-free benchmark is the most prudent and therefore correct approach for all stakeholders and the supervisor DNB.
Cools and van Nunen seem to confuse two different processes - the strategy setting and the monitoring. In the strategy-setting process it is justified to take views on long-term return-and-risk characteristics. Given the risk level it can afford, the pension fund can construct an asset mix and assume a long-term expected return higher than what one would expect from more or less risk-free investments like AAA sovereign bonds. Pension funds use ALM and strategic asset allocation studies to determine appropriate risk levels for their investments and to design an asset mix accordingly.
Once the strategy is implemented, the risk the pension fund runs is that the realised investment returns will be less than expected and not be enough to cover the growth in the value of the liabilities. So it is important to monitor the development of the actual value of the investments in relation to the actual value of the liabilities. Where taking views and assuming positive long-term returns was justified, even necessary, in the strategy-setting process, it is totally inappropriate to use expectations in monitoring, which is a measurement process.
Consider, for example, a football team that, after conceding a goal, states: ‘Actually we are 3-1 ahead as we expect to score three goals this game’. If you incorporate positive expectations and assumptions in your measurement (valuation) you are fooling yourself dangerously. Feel free to decide on how to act on a measurement, but don’t tamper with the measurement itself.
Put in a pensions context, consider two hypothetical pension funds, A and B, with liabilities of just one cash flow of 100 in 30 years’ time, and assets consisting of one risk-free zero coupon bond that pays exactly 100 in 30 years’ time. The coverage ratio is 100% and stays 100% as long as liabilities and investments don’t change: the investments and liabilities match exactly.
Cools and van Nunen will agree on the 100% coverage ratio: the value of the liabilities discounted by the expected return of the zero coupon bond equals per definition the value of the bond. The disagreement between us starts as fund A decides to sell the bond and buy equities from the proceeds. Fund B doesn’t change anything. We would argue that the coverage ratio is still 100% for both funds; nothing changed in the value of the investments or the liabilities.
According to Cools and van Nunen, however, the coverage ratio of fund A would increase substantially and instantaneously. Assuming a 30-year risk-free interest rate of 3% and a long-term expected equity return of 4%, the coverage ratio would jump to 134%. However, reality is never as you expected it to be. Monitoring the coverage ratio and taking action in time depending on its development is an essential part of managing a pension fund, precisely because a pension fund should be run on a long-term basis. A small difference between realised and expected return will have substantial consequences over the long term. Cools and van Nunen rhetorically state that €1 invested in equities 1802 would be worth almost €1m now, assuming an annual return of 6.8%. If the annual return turns out to be just half-a-percent less, 6.3%, the same €1 invested in 1802 would be worth less than €375,000 now.
Ask a member of a pension scheme about their risk tolerance and the answer will be ‘none’. This is simply because the upside of any risk is not shared with the members, but with the sponsoring entity. The maximum payout a pension scheme member can expect is an income based on last or average earned salary. Most members of pension schemes, with the exception of schemes of very recent history in The Netherlands, will expect with certainty for such a payment to be honoured. Calculating the coverage ratio, using a risk-free interest rate, disambiguates the certainty implied by the member’s expectations.
In finance, such surety is always valued in an arbitrage-free environment; options, futures, and, under market consistent embedded value, guarantees in insurance savings contracts, are all valued against risk-free benchmarks. Whether the future deliverable is IBM stock, crude oil or, indeed, a pension, doesn’t make a difference. Economic theory and practice are very clear.
Risk-free pricing is academically and practically the prudent way of valuing the health of a pension scheme, and evaluating the regulatory solvency of a pension scheme with a risk-free benchmark is the most prudent and therefore correct approach for all stakeholders and the supervisor DNB, whose objective it is that members’ future benefits are preserved as much as possible.
Where a scheme’s coverage ratio falls below a trigger, the DNB takes a closer look and considers whether to recommend possible actions for the fund, such as additional sponsor contributions, de-risking the scheme, or reducing member benefits. The DNB does not force a scheme to take a course of action (although the glassworkers’ scheme, compelled to sell some gold holdings, is a famous exception). As Cools and van Nunen suggest, the DNB can encourage schemes, explicitly or implicitly, to reduce financial risk where their coverage ratio is near or below the DNB’s trigger level. However, they overstate both the pressure that schemes come under and the effect of de-risking on asset prices.
Cools and van Nunen’s argument that Dutch pension schemes de-risking creates a vicious circle by impacting asset prices so as to further lower scheme coverage ratios, necessitating more de-risking, is hard to believe. Most schemes have diversified portfolios across different geographies and asset classes. These schemes making marginal changes to their portfolios over a period of time will have little price impact on global asset markets with plenty of nimble and valuation-driven investors. Cools and van Nunen also make the simplification that the only way a scheme can de-risk is to move from equities to bonds. For example, they could use derivatives to reduce their duration mismatch (and therefore the coverage ratio required by the DNB) without having to sell equities; or use corporate bonds to reduce the duration mismatch while maintaining a significant risk premium exposure.
Economic theory, practice and common sense all require the future claim of a pensioner to be valued against the risk-free rate. What is killing the pension system is not the thermometer used to assess the financial health of a pension scheme, but the continued inability and unwillingness to act.