Tactical asset allocation as a specialist mandate
Most practitioners know it and it was even proven scientifically by Brinson, Hood and Beebower with their 1986 Financial Analysts Journal-paper ‘Determinants of Portfolio Performance’, namely, asset allocation is responsible for the bulk of an investor’s performance or portfolio risk profile.
However, it is strange to see that most institutional investors are rather reluctant in giving away specialist tactical asset allocation (TAA) mandates. Most of them have an asset-liability management (ALM) consultant optimise the long-term asset mix and have specialist managers run mandates on the asset level. Active shorter-term ‘playing’ with asset weights around the longer-term ALM ‘ideal portfolio’ weights is either not done at all, or it is done on an ad-hoc basis. What are the reasons for this paradox?
In this article we will address some of the issues involved and also present the results of a quantitative study we did. Our conclusion is that the lack of structural attention for TAA issues is not warranted both from a return and risk point of view.
It is sometimes stated that active TAA-play is too risky. Brinson and others argue that at least 50% of portfolio performance (and probably even much more) can be explained by active TAA ‘bets’. Comparing the asset- and country-choice decisions of a TAA-mandate with that of an active equity mandate we see that the investable universe consists of a much smaller number of securities. This means it is more likely that specific unexpected mistakes, ‘negative outliers’, in any individual month will not be compensated by above-average results elsewhere – ‘positive outliers’. This is especially so when active bets were large. It is true that active TAA bets are more risky when looking to individual monthly observations. However, there are two counter-arguments against refraining from active TAA because of risk:
o First of all, because the impact of a TAA bet is relatively large it is not necessary to take huge bets. Smaller deviations from longer-term ideal weights or benchmarks will suffice. The correlation coefficients between various asset classes – or countries within an asset class – are relatively smaller than the inter-stock correlation in an equity mandate.
o Secondly, an active TAA strategy is something that one should install for at least a sufficiently long period before judging it or even cancelling it. ‘Time diversification’ is relatively more important when having less opportunities to diversify between various asset-class bets in individual months.
We sometimes hear people stating that active TAA is too costly. They normally refer to the cost of moving the overall portfolio and not necessarily narrowly defined transaction costs related to trading or the asset management fees. It is argued that it is impossible to move large chunks of underlying portfolio value back and forth between cash, bonds, equities, real estate or any other asset category. With the current sizes of major institutional investors the buying and selling might be too tedious and costly.
Different managers are running individual portfolio components, so this would involve ordering one manager to sell part of his assets and move the resulting cash to another manager. This can also be quite costly because we are talking about ‘big trades’ that could have substantial market impact. Research by Markowitz & Van Dijk (see panel below) shows that this might partly hold. However, as stated before, we are talking about relatively small deviations between long-term ideal asset category weights and the short-term optimal TAA weights;
TAA bets are not ‘short-term’ and ‘technical-analysis like’ as are speculative positions in individual stocks. The chance of a quick reversal that might disturb the regular asset management process is relatively small.
We do acknowledge that things might be tedious and complicated for institutional investors that have a large component of their money managed externally by specialist managers at the asset category level. This does not mean that it is impossible to run this type of strategy with internally managed money. And even when talking about external mandates it is still a cost-benefit thing. We are not sure that all parties that state ‘TAA is too costly’ have carried out such a cost-benefit analysis.
TAA operations can be run very cost-efficiently when using futures overlays. The cost of futures trading is relatively low and it helps avoid the hassle of individual managers moving assets back and forth to each other. Besides that, the management fees charged by TAA managers are relatively low when comparing them with a) active equity managers; and/or b) the relative rewards of a ‘well-managed’ TAA mandate both from a return and risk point of view.
It is sometimes stated that markets are so efficient that active TAA decisions are not warranted. But it is strange that a lot of people who say this actually run active mandates on the individual asset category level. If markets are efficient on the aggregate level, aren’t they on the individual security level? Of course, it is not easy to beat the market, but good managers with a rigorous framework for data collection and interpretation should be able to extract information that is not available to others at the same time. No doubt they will make mistakes too, but as shown by UCLA’s Professor Mordechai Kurz and others, ‘efficiency’ should be interpreted as ‘every investor getting the return-risk trade-off that matches his or her level of information-availability’. This means that markets as a whole could be efficient for the ‘average market player’ with some investors beating the markets and others under performing. It also means that it is not true that the group of outperformers is completely ‘random’ and that no one can outperform structurally. Above-average information/skill will result in an above-average return/risk trade-off. A metaphor from the chess world will help explain this. The average player in a tournament will always have a score of 50%. However, when world champion Kasparov participates it is highly unlikely that his above-average skill and information-processing qualities will not give him an above 50% score. Though he might under perform as a result of other factors in an individual tournament, his average will still be above 50% as long as his information processing framework is ‘above average’. Analysing the results of active TAA-managers in practice and studying academic efforts to create TAA-frameworks we see that ‘good’ TAA systems:
o Provide a structured approach in which return and risk are taken into account simultaneously;
o Decisions are always taken in relation to expected return/risk combinations for various asset classes; the portfolio status quo; portfolio restrictions and the long-term ALM or strategic benchmark.
When comparing these results to activities of institutional investors in practice it is striking that a lot of them seem to work with too low an equity allocation. Recently we see a lot of signals that institutional investors are adjusting themselves. This is mainly related to demographic factors (aging population), but we do feel that a better understanding of the (long-term) difference between equity and fixed-income returns, even after correcting for risk plays a substantial role too.
Another remarkable thing is that those active institutional investors that do implement TAA strategies in many cases can be considered to be ‘too active’. Contrary to the results of our research they are trading quite heavily (futures or underlying asset) irrespective of the predicted state. We find that the bulk of trading activity should be limited to those situations in which extreme ‘states’ are predicted, a top-quintile or a bottom-quintile return-estimate for equities. We feel that this is to a large extent behavioural. Investors are often over confident and fail to acknowledge that their predictions are not good enough to really believe that ‘a slightly above-average state’ could not be just an average state. Only the extreme predictions have sufficient explanatory power to warrant substantial shifts in the portfolio composition or the futures overlay. It is also remarkable that the essential link between strategic allocation (ALM) and TAA is often not visible at all in TAA mandates. In many instances it is limited to a month-by-month or quarter-by-quarter comparison with and deviation from some kind of combined world index (equities and stocks). No doubt the weights in that combined index are often based on strategic equilibrium weights for both the bond and equity component of an underlying portfolio, but what is lacking is an interplay between TAA and ALM. The longer-term consists of a series of shorter-terms: information from TAA analysis can be of substantial importance for the longer-term allocation. Many institutional investors know what we mean when thinking about the decline of the Nikkei during the beginning of the nineties. It took a long time before the continuous TAA signals that something was structurally wrong were incorporated in longer-term analyses. A lot of institutional investors correctly felt that there might be something structural going on here but did not have the correct framework for the integration of ALM and TAA.
We feel that this is going to change. In a way one could say that until now the relative negligence in terms of structured active TAA policies was nothing more than the majority of institutional investors stating that ‘passive’ is best when it comes to TAA. Just like with equities or bonds, both passive and active management do have their attraction. When not having explanatory power one should be passive, but when having above-average information availability one should act. Our analysis shows that the general optimum kind incorporates both.
The next five years will probably see substantial growth in terms of interest in TAA and assets under management for managers with a proven TAA system might grow. However, this will only happen if TAA managers are able to interact with institutional investors in such a way that links can be created with strategic policy. It is also essential that the TAA policy will consist of relatively small bets. One of the problems might be that a good TAA manager that stays relatively passive in ‘about average’ states will actually be blamed for doing so by the sponsor and therefore be induced to trading. That is obviously not in the interest of both parties. There is a lot at stake here and the only solution can be relatively I) low base fees; II) a growing interest in performance-related fees; and III) tight monitoring from a return-risk point of view. We feel confident that an important, often forgotten field of investments is about to be rediscovered.
Harry M Markowitz and Erik L van Dijk presented their findings at a Palladyne seminar on ALM and TAA, Amsterdam, November . A handout of their presentation is available from Erik L van Dijk, CEO Palladyne Asset Management (firstname.lastname@example.org)