In his novel The Alchemist, Paulo Coelho tells the story of a shepherd who leaves home in search of a treasure. Along the way he meets a beautiful gypsy woman, a man who calls himself a king and an alchemist. No one knows what or where the treasure is but they all point him in the direction of his quest; ultimately, the treasure is found right where he started. This could easily be the tale of the average pension fund manager who in their search for more alpha may have missed the treasure within.
Since investors realised that the returns from classic asset classes and static strategic asset allocations (SAA) would not deliver the returns required to either maintain or improve the funded status of pensions, there has been a flurry of activity. Every journal is rife with discussions about separating alpha from beta or hedge funds, and even fund of funds. The marketing hype has overtaken a clear evaluation of what these strategies may offer, and product positioning has allowed vendors to charge enormous fees.
Unfortunately, all of this has happened as economic growth has slowed and hence the budgets for pension fund managers has been cut dramatically, even while many of these funds have paid external managers fees that are outrageous relative to the contribution to total fund value. As a result trustees and pension fund managers should be attentive to all the decisions in their portfolio structure, as the potential to generate meaningful excess returns is easily achievable and money spent on internal operations may be a lot more rewarding than hiring the next best alpha provider.
Figure 1 highlights a typical pension fund investment decision process (IDP), based on the Dutch pension industry. It has been acknowledged that the SAA contributes 90% of the variability of returns, but too many funds go from an asset-liability study straight to focusing on which manager to hire, even though manager selection is the bottom most investment decision in this tree!
There are a number of additional decisions in structuring the fund (asset allocation, sector/regional allocation or even style) that impact value at the total fund level every period. Most funds have ignored the value they can add from making these decisions in an informed manner on an ongoing basis and hence may spend too much money on managers whose impact on total fund returns may be marginal. Moreover, in making these structuring decisions effectively, funds can compound alphas.
There are three basic models through which a pension fund may be run:
q Head in the sand model (low cost/low risk – benchmark returns). Index all assets at the bottom most tier of the tree and undertake naïve rebalancing. This model is going to lead to benchmark returns, low tracking error and potentially require additional contributions;
q Grass is greener on the other side model (low internal/high external cost – moderate return potential). Funds seem to be drifting to this model by keeping staff lean and hiring consultants, hedge funds, fund-of-funds or active managers. Returns will potentially be higher (along with manager fees), but this is predicated only on the quality of the manager selection effort. Staff will need to not only make many good hiring decisions, but also good manager allocation decisions. Implicitly, trustees seem to be willing to pay external staff more than internal;
q The alchemist model (moderate internal cost/low external cost – highest return potential). Good staff are hired, empowered and compensated because the trustees realise that good decisions by staff can add more value than good external managers and at much lower cost. Moreover, with good process and controls, as decisions roll up the decision tree, alpha should compound. Many Dutch and Swedish funds have implemented GTAA programmes internally. We are proposing to go a step further and use the inputs from the GTAA program to structure the entire fund appropriately in every period.
We assumed that the fund had made good manager decisions and that the managers contributed 42 bps to total fund returns and that benchmark choice (benchmark risk) added another nine bps. We then suggested to the client that there were three ways in which the fund could be run:
q Let the portfolio drift. Hire external managers and let the allocations to various styles and asset classes be the result of how the portfolio drifts based on market movements of various asset classes;
q Naïve quarterly rebalancing. Funds can follow some form of rebalancing, based either on regular timing or on some ranges around the SAA. These alternatives are largely substitutes trying to track the SAA performance with minimal risk and fit in with the grass is greener on the other side philosophy. Pension fund staff are convinced or told that they cannot time markets, but must realise that they are implicitly taking an active view on markets when such choices are made;
q Informed decision-making. Staff develop and test views on which style, region and asset class is likely to do best in any period, and implement these views to make sure that the portfolio is structured appropriately for the next period. Here, a pension fund is being operated like a sophisticated asset manager/hedge fund, and risk management is ensured through good decisions being made at every level of the fund. The ideas to manage the decisions in the fund were in the public domain and they are uncorrelated with the SAA. However, every decision in the IDP is made on an informed basis.
Table 1 illustrates the contribution of performance under the three approaches. The drift approach erodes all the manager alpha, the quarterly rebalancing approach preserves the same, while the informed decision approach adds meaningful returns to the bottom line – greater than the manager alpha for considerably lower cost 2.
Once clients understand that decisions in a portfolio cannot be viewed in isolation (eg, asset class style), but rather how they impact decisions above and below them, they can actually see alphas compound. Table 2 shows the expected alphas from testing rules in isolation (ie, the GTAA type analysis shown as pure strategy excess) and what happens when these rules are embedded in portfolio trees (contribution to portfolio). By overweighting good decisions lower in the portfolio and underweighting poor ones below, alphas can compound up an IDP as shown in the third column.
More interestingly, these allocation decision alphas are achieved more cost effectively than the external manager alphas. To prove this, funds need only calculate the aggregate fees paid to external managers as a ratio of the alpha generated by these managers. The same ratio for allocation decisions will show greater alpha for less cost. That may explain the ability of external managers to have high overheads.
If funds are not going to do the job of managing the entire portfolio (with all the decisions and manager choices) in an informed and intelligent manner and expending resources to make decisions effectively, the costs incurred by the plan sponsor may be substantial. The alchemist model focuses attention back on the decisions that can have the greatest impact on fund performance, allow for alpha compounding and represent the cheapest source of alpha, the above advantages notwithstanding.
Table 1: Results are extracted from an analysis of these funds using the AlphaEngine system (www.mcubeit.com). Results are for an illustrative 10-year period between July 1995 and June 2004. They do not include transactions costs and the usual caveats to simulated performance apply. US LC/SC stands for the decision between large and smallcap US equities.
2The Norwegian Petroleum Fund is on record stating that even though external managers handle less than 20% of the assets, they account for over 50% of the cost.
Arun Muralidhar is chairman at Mcube Investment Technologies LLC