Under (classical) ALM, the industry has so far given little attention to the fact that the downside risk (in the funding ratio) is far more expensive than the upside most of the time. It mainly looks at expected outcomes (like average contribution rate) and probabilities (like shortfall probability), but does not incorporate risk-averseness as discounted in financial markets. However, since Black and Scholes (1973), we all know that (at-the-money) put options are more expensive than call options. Fair-value ALM uses risk-averseness as discounted in financial markets to value cash flows of pension funds. This will lead to better ALM decisions: improved choices for contribution, indexation and investment policy and more sustainable pension deals.
The following example might be helpful in illustrating this insight. A pension fund has €130 assets, €100 liabilities (and thus a funding ratio of 130%) and an equity/bond split of 50/50, as well as full indexation and a fixed contribution rate of 17.5% of pay.
With classical ALM methodology, we can calculate that after 15 years there is a probability of slightly more than 10% that the fund will be underfunded1 (and thus almost 90% of being more than fully funded).
Fair-value ALM2 shows us that protecting the downside in the year 2020 would cost €8, while the substantial upside above funding ratios of 100% would only be worth €35. The reason is that in fair-value ALM negative future scenarios are attributed far more weight than positive scenarios. Risk-neutral valuation will correct for the (equity) risk premium in discounting future cash flows: high expected returns of €100 invested in equities will have the same present value as the low expected returns from €100 invested in bonds, when corrected for the higher risk of equities.
In economic bad times, hedging the downside in financial markets is expensive. In other words, a (hypothetical) bank or reinsurance company taking over this risk from the pension fund would have to provide a put option on equities underperforming the (indexed) liabilities.
The fund could decide that the shortfall probability is too high for its taste. Let us assume that the fund increases contributions to 20% when the funding ratio is below 130% and will cut it to 15% when the funding ratio is higher. The shortfall probability would indeed drop by a couple of percentage points and at the same time, the average contribution rate would decline by 0.2%. In the classical ALM world these are all positive signs. Fair-value ALM though, shows us that the present value of contributions will increase by e2; so in this example, a lower average contribution rate actually results in a higher present value (of e117)!
Increasing contributions in bad economic times will have larger negative consequences for the fund’s members than similar increases in good times. This is because it will have a far larger financial impact on employers and employees than rising contribution rates in economic good times. In bad economic times, stakeholders most likely will also be hit by other financial problems, such as lower profits, falling equity prices or lower wage growth. A rising contribution rate would therefore be a larger disappointment in bad times.
The advantage of fair-value ALM is that it incorporates people’s risk-averseness. It shows that employers and employees have a significantly lower appetite for paying higher contributions in economically poor times, even when the average contribution rate over a long period under such a policy may be lower. One can draw similar conclusions for conditional indexation (not discussed in this article). A bonus of fair-value ALM is that it allows calculating the impact of policy changes on various stakeholders (also not discussed), that is, it answers the question who will gain and who will lose from increasing contributions in bad times and to what extent.
1 In the graph, we included percentiles for 1, 5, 95 and 99% as well as the median.
2 Fair-value ALM uses risk-neutral valuation to calculate the present value of all the cash flows of the fund. See for example ‘Techniques for market-consistent valuation of contingent claims’ by Hibbert et all. in ‘Fair value & pension fund management’ in Kortleve et all. (2006)
Niels Kortleve is manager, actuarial projects and special accounts at PGGM
Assume you have a 50% probability to become a millionaire and 50% probability to become unemployed. An insurance agent is trying to sell you two products. The first product promises €1,000 if you lose your job, the second product will pay €1,000 if you become a millionaire. For which product would you pay the most?
Classical ALM’s answer is: both products have the same price, because both products have the same probability and the same pay-out of €500. But the fair-value ALM approaches states otherwise: a payment when you are unemployed is far more attractive than a payment when you are millionaire. So most people would be willing to pay substantially more for the first product.
The same methodology can be applied to the cash flows of pension funds. Hedging underfunding will be expensive since underfunding will occur in economically bad and expensive times – you could say when many people are unemployed. Overfunding on the other hand, will most likely happen in good times, when many people will be wealthy already and will therefore have a relatively low present value.