After years of near-neglect, asset allocation is now centre stage. The pensions world has (re-) discovered its importance for investment performance, liability matching and risk management.
How quickly things can change. Only a few years back the view was widespread that there was not much value added to be gained from asset allocation. Instead, in the long bull market of the 1990s, pension boards and consultants increasingly focused on areas such as stock selection, core-satellite, active versus passive, fund manager selection and so on.

Why the comeback?
Why such a comeback of asset allocation? It is not difficult to find the main reasons in real world ‘shocks’:
q Market movements: The trigger was the combination of falling stock markets and falling interest rates in the first years of this millennium. For many pension plans, this meant heavy falls in assets at a time when liabilities gained momentum;
q Funding issues: Funding levels often fell to critical levels, leading to higher contribution demands on sponsors, interventions by the regulators and nervousness of plan members;
q Risk awareness: It was suddenly realised that big risks were not managed adequately, while too much time was spent on comparatively marginal investment issues.
As a result, stakeholders of many pension plans not only felt uneasy about the asset split of their pension scheme but also questioned the way asset allocation decisions were taken.
So, asset allocation has moved to the top of the agenda. However, it is less clear what exactly to do with it. The debate is on. Part of the heated discussion is still about dogmatic questions. What is the real god: bonds or stocks (or even the newcomer hedge funds)? Beatifically, some of the debate has moved on to more pragmatic questions: How to do it, who should do it, and what exactly to invest in?
It is worth discussing first some of the key points from the dogmatic debate pro and contra the two main asset classes, equities and bonds.
The main argument for stocks is the long-term outperformance versus bonds and cash. Looking to the future, a substantial projected ‘equity risk premium’(often 3% or higher) does the trick. For actuaries and consultants, it pushes quantitative asset-liability-models (ALM) to produce high equity weightings. For sponsors, it reduces the projected cost of pensions, which in turn often leads to more generous benefit promises for members.
This view worked quite well for a long time. To a large extent, pension funds with high equity weightings financed themselves. In the 1990s, many Continental European plans also raised their equity weightings. More recently, however, this view has come under criticism for a variety of reasons:
q The equity dogma has not turned out to be bear-market proof.
q The long-term liability matching qualities of equities are being questioned.
q How much of an inflation protection can stock markets provide? What about deflation protection?
q There are conceptual and practical problems with the established ALM tools used to model expected returns and risks, for example, the occurrence of ‘too many’ extreme events. Some trustees have found themselves well below the statistical ‘worst case scenario’.
As a result, several experts now believe that equities are a bad (or no) fit for pension obligations and therefore too risky to have in any substantial proportion.

Matching bonds
Now on to the case for bonds. The relevant risk of a pension fund is the mismatch of assets and liabilities. The new dogma favours bonds as the natural investment for pension funds. The main argument is that ‘pensions are bond-like’, as the cash flows of pensions and bonds are very similar.
It is interesting to note that the concept of liability-matching is not new in financial theory or in practice (for example, by insurance companies). Also, in several countries pension funds have always been seen as ‘insurance-like’ arrangements and therefore more conservatively invested, even during the bull market.
This view is appealing in particularly to those concerned about the security of pensions. However, there are issues which are heavily discussed, especially the following:
q There are conceptual reservations about bonds being a perfect match: liabilities are a ‘moving target’ as biometric factors, benefits and regulation change all the time;
q Financing pensions is likely to become more expensive in the longer term with a bonds-dominated asset allocation;
q Are there enough long-dated and inflation-linked bonds on the markets to satisfy the increasing demand from pension funds?
q Investment banks offer swaps and new derivative instruments for cash flow matching, but at what price and risk?
q The paradigm shift away from equities comes too late. And anyway, what’s wrong with higher beta returns from stocks if you can afford the risk?
In a nutshell, the move back towards bond has so far been slower in practice than in people’s talk. Having gone through the last wave of dogmatic debate between the two schools, a new sense of pragmatism seems to take hold among the practitioners, more concerned with finding the best specific solution for their pension fund.

Framework for allocation
Sensible pension plans take asset allocation from first principles and start with discussions such as: What is the asset side supposed to achieve? What is the tolerance for risk? Is there an understanding between the pension plan, sponsor and members about the key objectives, including risk and return targets?
General debates about asset allocation often neglect important real world constraints although they are often the most determining factor. The most obvious examples are quantitative restrictions on asset class weightings imposed by law.
Another crucial influence stems from solvency regulation. For example, a requirement to be fully funded is a major limitation to asset allocation policy which asks for applications of option theory and the use of derivatives in order to guarantee daily solvency. On the other hand, a buffer fund that has no withdrawals for many years may be in the position to develop a nearly unconstrained long-term investment policy.
Furthermore, the debate has recently focused a lot on the impact of changing accounting regulation for sponsors. Clearly, the move towards mark-to-market rules puts the sponsors’ (balance sheet and contribution) risk to the forefront (which is different!).
This in turns raises the whole question of the relationship with the sponsor: How strong is it financially and how strong is the commitment to the pension plan? Who decides about contributions? What is the balance of power in case of takeovers or bankruptcy? This is just a small sample of questions that need to be clarified before making any asset allocation decision.

Who does it?
Moving on to the strategic decision-making, the recent difficult period has also led to discussions about the appropriate governance of investment matters.
Who is making strategic asset allocation decisions? In a DB plan, what is the main pension board willing, and able, to delegate? In DC, how is asset allocation effectively determined? What are the rules of play for life-style and default funds?
Clearly, different countries have developed very different solutions. At one extreme, the main board gets involved in short term asset allocation bets. At the other extreme, strategic decisions are effectively undertaken by external advisers. Time for a re-think about who should do what!

What approach to take?
Only when the top level questions - objectives, risk, constraints and governance - are clarified, it is possible to develop a sensible asset allocation policy.
A decision needs to be made about which approach is being taken as a starting point: ALM based on the classic risk/return approach or (insurance-type) liability-based investing (LBI)? The answer will often naturally emerge from discussion of the earlier top-level points.
On LBI, for example, the debate has only started on the best way to approach it. Is exact cash flow matching possible and reasonable? What mix of ‘immunisation’ strategies is best? Should fund managers be given liability driven benchmarks, and with it a great deal of the strategic responsibility for managing the funding ratio?
Other people advocate some sort of absolute return benchmarks, like a cash-plus rate, or even inflation-indexed benchmarks. On what basis make such decisions? Should mandates then be much longer than in the past? What would be an appropriate fee structure?
Whichever approach to strategic investment, active managers a trying to bring themselves back into the loop through the notion of ‘alpha investing’. In which area to seek best for additional returns? How much risk can be incurred?
We are currently experiencing a broadening of the investment universe available to pension funds. And funds make increasing use of it:
q More international investing, including emerging markets;
q Sub-classes of traditional assets, for example, small cap stocks, style investing, high yield, and many more;
q Old asset classes, newly discovered, such as property, or newly financially packaged, such as commodities;
q Private equity in various forms;
q The long and very diverse list of hedge funds strategies;
q Overlay strategies, as in currency or tactical asset allocation;
q Other opportunities, such as art funds or infrastructure assets.
New asset classes and sub-classes are being proposed by the day and this, unfortunately, already has the smell of the next investment over-hype. Should you hold them all? Unlikely – but how to choose?
How much of each? In a computer simulation, it is easy to claim some risk/return benefits from adding any additional asset class to the pension fund portfolio. It is more difficult to say how much to hold of each, even in theory.
In practice, some very important constraints come into play. One of them is the additional time and resources required for researching, selecting, managing and controlling a changing fund set-up. How many pension funds undertake such a cost-benefit analysis of adding asset classes?
To make it worthwhile, people often recommend a minimum of 5% when investing in new asset classes. But how many such slices of the cake can a pension plan digest? Also, by adding new asset classes one by one, the whole risk-return characteristics are changed every time. New clusters of risks are being produced which are probably not well understood.

How to change over time?
There is an even more difficult issue that relates to the time dimension of asset allocation:
q As markets move, the asset allocation drifts away from the neutral position. Is there a rebalancing policy in place, and what is it? What happens in case of significant market movements?
q Even strategic asset allocation (SAA) needs to reviewed from time to time. People now find that a three-year, even a one-year adjustment period is too inflexible. However, how to make sure that decision-makers are not carried away by short-term market fashions or panic?
q In light of these problems, there is a call for a more ‘dynamic asset allocation strategy’(DAA). But how is it implemented and how often is it itself reviewed?
There are very different views about the merits of tactical asset allocation. There are important questions to be answered. For example: Are you sure TAA is not already undertaken by any of your manager (often not explicit!)? How does TAA interact with the strategic changes in SAA or DAA? Any timing of a strategic change has a tactical dimension: Who is accountable for this? How to hang all these things together in your risk control exercise?
These are challenging times in pensions investment. The new focus on closer liability matching, while still seeking to squeeze the best returns from financial markets at times of low nominal growth, opens a whole new range of
questions on the optimal asset allocation policy.
At least, the current difficulties seem to have spurred a better dialogue: Investment banks and managers are learning fast about the requirements of pension funds, while trustees are improving their understanding of modern investment instruments.
Georg Inderst is an independent consultant based in London