Adina Grigoriu asks, is pro-cyclical risk management necessarily a cost - or can it be an unexploited source of performance?
EIOPA’s call for advice on the review of the IORP directive has brought forward many questions. One that both pension funds and their advisers raise is: “What is the cost of risk management?” But isn’t there an even a better question to ask: “What is the cost of not managing risks?”
For many investors, the answer to this question should trigger risk management measures immediately, independently from legislation. Aren’t we all, in one way or another, paying for those that took risks bigger than they could afford? The management of assets with little or no link to liabilities continues to have dramatic repercussions that can go as far as the lack of pensions for people who have worked their entire life.
First, let us acknowledge that there is a difference between risk measurement and risk management. The second only makes sense in the light of the investor’s objectives and constraints.
Consider the example of a long-term investor whose only risk constraint is never to lose more than 10% from his portfolio’s highest level. His manager will likely offer him three investment alternatives (in a long-only world) - strategic asset allocation (SAA), tactical asset allocation (TAA) or dynamic risk management (DRM).
SAA has, for a long time, been the most widespread option in the industry. In this approach, the manager will try to find, in the light of past market experience, the optimal allocation between different asset classes and regularly rebalance. TAA is often used in addition to SAA to move the asset allocation away from the SAA benchmark, based on return expectations on the underlying asset classes.
DRM has its roots in portfolio insurance techniques. It introduces an often forgotten concept - the investor’s margin for error. The principle is straightforward: the asset allocation is not based on the past behaviour of asset classes or on forecasting, but is defined and adjusted dynamically, based on the investor’s margin for error. When the margin for error is substantial, the investor can take more risk then when it is small. For instance, the margin for error of a pension fund whose liabilities are bigger than its sponsor’s market capitalisation is very small. Even small movements in the funding level below a certain level could lead to the liquidation of the parent company.
Regulation seems to be encouraging this third approach. What does that mean for the long-term investor?
Let us return to our investor that has a maximum drawdown limit of 10% and only invests in UK stocks and UK gilts, with a limit of 60% to equities. We compare the three approaches over the period between the end of December 1998 and the end of August 2011. The portfolios are rebalanced on a monthly basis.
Approach 1, SAA. We study, over the period, the evolution in terms of assets and in terms of on-going drawdown of all portfolios with a fixed allocation to equities, ranging from 0-60%. We find that in terms of returns, the asset level reached by those portfolios is almost identical. However, in order to comply with a 10% maximum drawdown, a SAA portfolio should not have held more than 30% in equities during this time. This prompts two observations:
• This is ex-post risk management and if the future does not look like the past, the risk limit might not be respected in the future
• The percentage allocated to equities is relatively low, given the long time investment horizon of the investor and the 60% limit allowed.
Approach 2, TAA. We also study a TAA strategy where the fund manager invests to the maximum limit in equities (here 60%) when the manager thinks that they will outperform bonds in the following month, and 100% into bonds in the contrary. We simulate the decisions taken by 1,000 fund managers, all having the same success rate, and measure their drawdowns and performances.
For what is considered in the market to be a good success rate, 7/12, the performances of these asset managers vary greatly with the moment when they are right or wrong. Some of them barely protect capital during the considered period, others multiply it four times.
From a risk prospective, the occurrence of drawdowns greater than the investor’s limit is substantial. Even when we repeat the study with manager success rates up to 11/12, some still breach the 10% maximum drawdown limit. Alone, the tactical approach does not seem well-suited to the presence of risk constraints.
Approach 3, DRM. Finally, we look at a DRM approach where the allocation to the risky asset depends on the drawdown budget available at all times and a multiplier determined in a stochastic environment. Hence, with a 10% drawdown limit and a multiplier of six, the allocation to equities is 60%. If the current ongoing drawdown is, say, 3% (meaning 10% - 3% = 7% left), the allocation to equities is no longer 60% but 7% multiplied by six, or 42%.
Figure 1 shows DRM strategy evolution compared with that of the underlying assets, the protection floor and the allocation to equities. This helps visualise the fact that the allocation to equities only depends on the distance between the DRM strategy and the protection floor, which accounts for the investor’s risk constraint.
The comparative results are as follows: over the period considered, DRM yields 6.4% compared with 5% for SAA. The over-performance of DRM over SAA with the same risk limits persists in a stochastic environment, when comparing many possible outcomes for the evolution of the assets. As for TAA, a 7/12 success ratio comes with a 27% drawdown and makes the strategy ineligible in the presence of risk constraints.
This example shows that it is indeed possible to manage risk constraints ex-ante while delivering better long-term performance. The second point usually comes as a surprise. An investor might assume that the pro-cyclical investment decisions that result from DRM - or regulation for that matter - would negatively affect performance. After all, it means selling low and buying high. Actually, because markets exhibit momentum, this type of approach performs better, as it allows for a dissymmetric management of returns: as negative returns are cut out, the whole distribution shifts to the right (figure 2).
The management of risk budgets can take more complex forms and include all kind of constraints: minimum funding ratio; capital requirement optimisation; capital preservation at a given horizon, and so on. It can also be used as a very effective tool by underfunded pension schemes to get back to a fully-funded situation while minimising contributions.
When we compare the contributions needed to get from an underfunded to a fully-funded situation using the most widespread techniques, dynamic asset and liability management - the extension of DRM to an asset and liability management world - requires 30-50% less in contributions, depending on the scenarios.
For example, in a stochastic environment, if the contributions required amount to 100 in 50% of the scenarios with an SAA approach, they amount to 105 with traditional LDI, to 95 with a de-risking (or ‘flight path’) approach - but to just 65 with dynamic asset and liability management. Of course, those results can vary depending on the particular situation of each pension fund. Trustees should ask for comparative studies of the impact of those strategies on their fund.
No matter the investor’s constraints, risk management represents an unexploited performance vector more than a cost. Quite surprisingly, adding an extra risk constraint often leads to a performance improvement.
Adina Grigoriu is CEO of Active Asset Allocation International Consulting