Tactical asset allocation (TAA) is the talk of the town again. Once known as market timing (before that term became associated with more unsavoury practices), the basic tenet is to buy the market when it’s low, and sell high. No one doubts that this would add significant value if done right. The big question is, can anyone do it right consistently?
In many ways, TAA is strikingly similar to another well-known acronym: SETI the Search for Extra-Terrestrial Intelligence. Even the acronym can be twisted to suit current purposes: Search for Exploitable TAA Inspiration.
Fortunately, the rest of the parallels are less contrived. Both activities involve hordes of PhDs doing extensive quantitative analysis. Both are driven by a combination of fear, greed and curiosity. Both involve billions of dollars. The most important similarity is more subtle, however. Experience tells us that in neither case does one start from neutral: you are inherently either a believer or a cynic. And where you start materially influences what you find.
Enter Agents Mulder and Scully from the X-Files. Mulder was the believer, convinced that the truth is out there. In the context of TAA, if you are a believer, your basic tenets are most likely that markets do not move together, and that efficiency within a market does not necessarily imply efficiency between markets. You are also likely to believe that economic and financial forces drive prices, which can be forecast to some extent. This leads you to believe that it is likely that value can be added by TAA.
Agent Scully was the cynic, whose approach was always to examine the data, and see if the facts could only be caused by extra terrestrials. In the case of TAA, the researcher must prove that any successful TAA is not random noise. Because if you can’t prove the existence of TAA skill, does it exist? This data-driven, ‘scientific’ approach is favoured by most professional academics and some, but not all, asset consultants.
Both perspectives can be valid if pursued with integrity. Each also has its drawbacks. The believer must be careful not to misinterpret every apparent success as skill; luck is a possibility that must be seriously considered. Neither can the cynic be allowed to wait until the accumulation of evidence is so great that it ceases to be controversy at all. By that time, investment markets will have competed away all the opportunity for profit and the phenomenon will have passed into history books.
The key question boils down to whether investment managers can predict future market movements. There is no absolutely rigorous way to test this: there are as many time horizons, asset class definitions and investment approaches as there are investment managers. However, IPE has been conducting a survey, the Investment Managers’
Expectations Indicator, since February 1997. This compiles managers’ expectations for market movements over the following six to 12 months.
Managers are asked whether they foresee a rise, fall or stabilisation in a range of equity, fixed income and currency markets. The data comprises 76 overlapping time periods. Only managers who responded to over 75% of total questions asked were included in the analysis that follows. Manager forecasts were compared to the following
six-month index returns in an attempt to identify forecasting skill. Russell, in association with IPE, has reviewed the track records of individual participants, and the group as a whole.
Managers do not appear to display significant skill at predicting market movements. The average across all managers, time periods and markets is 37%. This is not significantly different from an equally weighted random outcome.
Looking more closely at the numbers on a market by market basis does yield some interesting information: Table 1 shows the average accuracy of managers by market. The variation in successful prediction is considerable across markets, and the rate of success is not always lowest where you would expect it.
One expectation that is met is that success rates in fixed income are higher than in the equity.
A likely explanation is that movements of fixed income markets are more easily forecast, because they are more directly affected by interest rates and inflation. However, the disparity within the currency markets is quite striking: the success rate of the $/£ rate is nearly double that of the $/e, which at 27% is quite near the lowest level of accuracy across all the markets. If we examine the markets during different time periods, further interesting elements appear: Table 2 shows the annual average accuracy of managers by market. One of the most striking things is the wide range of accuracy. The highest level of accuracy is 59%
and the lowest 13%.
The only markets with accuracy levels above 50% are $/£ and sterling fixed income. There are none in the equity markets.
The success rates on a manager-by-manager basis (names withheld to protect the innocent) show a similarly discouraging result. Only three managers achieved an average success rate across markets above 50%, which is not statistically differentiable from a random result at a 5% confidence level. For purely voyeuristic purposes, we will just note that the highest level of success by a manager in any one market is 92% and the lowest is 0.
Why are the numbers like this? We believe that there are four main reasons that inhibit manager’s ability to employ successful TAA strategies:
q Complexity of the decision: The information portfolio managers use to make market-timing decisions comes from diverse sources, and requires understanding of economic and tax information from across countries and security types. Standard practice for relating this information varies widely, yet analysts must turn it all into comparable data. Cultural and political events may also influence market movements, yet they are very difficult to quantify, let alone compare.
q Inability to predict the future: The fact that past or current market conditions are fertile ground for market timing does not help us much for the future. This is because the factors that can affect future market movements are infinite. Consider the impact of the unpredictable SARS epidemic and the series of accounting scandals. Such events make the best analysis valueless to an investment manager. Much smaller, apparently less significant, events also have the potential to cancel out months of excellent analysis.
These first two may not appear so serious from first glance. The complexity of a TAA decision may not be much greater than one taken by a global equity manager. Furthermore, any active management is a form of predicting the future – although on a stock rather than an asset class level. The problem is the impact of an incorrect bet across an entire portfolio.
q Increased risk by ‘betting the farm’ on one decision: Russell advises its clients to use many small decisions as the main source of excess returns. This approach is not unique; see for instance the “pop-finance” Fundamental Law of Active Management promulgated by Grinold and Kahn. However, the nature of TAA means that investment managers will make relatively few key decisions every year. This leads to yet another complication – understanding how good a manager is becomes much more difficult. The timeframe used must be much longer than the normal three to five years. Even the new strategy, GTAA, where managers take bets across a wider range of markets and currencies, suffers from this limitation, albeit to a lesser extent.
q Problems with the data: There are a series of important caveats to add to the analysis above. First, the period over which the accuracy was measured (seven years) is short. While it looks like markets may have gone through a full cycle, it has certainly not been a normal market cycle. Also, the sample of managers was small (58), and restricted to European-based managers canvassed by IPE.
While clearly a distinguished list, it is not all-encompassing. The list also excluded managers who answered fewer than 75% of all questions. Furthermore, the initial survey gave no clear indication on how to quantify what markets going up, down or staying flat meant, so we have had to make some un-testable assumptions. Although varying the definition of what was considered to be a stable market did not materially affect the outcome of the research, it may have affected manager’s responses.
Next, the questions were asked in absolute terms, rather than in relation to each other. In a TAA world, managers would need to know what markets would do in relation to others, rather than in isolation.

Consequences for pension funds
The main consequence of TAA decisions is clearly the impact of bad performance on the pension plan funding. Consultants and trustees painstakingly establish pension plans’ allocations using finely-tuned models over months at a time. Market timing leads pension schemes to move away from these strategically set allocations. While some might consider it a worthwhile gamble, so many pension funds are currently under-funded that trustees may be wiser to avoid any situation that could compound funding problems.
Despite results from this simple analysis, some managers may have legitimate TAA skill. If you only examine the data, you are likely to conclude that the empirical evidence does not sustain an argument for TAA.

Skill present?
However, the fact that a large group does not, on the whole, show unambiguous evidence of skill does not mean it is not present. In fact, there are glimpses of success in the currency markets. However, success may have been swamped by bad luck in the period under review. Or it may manifest itself sporadically. You cannot be 100% sure. Sometimes, Mulder is right and you just have to go with your hunch.
The big problem, though, is the discontinuity between what TAA as described here is trying to achieve, and what investors should use it for. TAA should not be a source of return that can override all other sources of return. Rather, if it is to be adopted at all, it should be seen as additional and complementary to a well diversified, carefully allocated portfolio.
Alexandra de Zwart is marketing associate and Scott Donald is director, marketing and product development at the Russell Investment Group in London