Rationale & Best Practice:
A guide to Selecting Quantitative Global Macro and GTAA Strategies
 
  
 

Many pension funds, endowment funds, and individual investors are concerned that equities—typically their largest asset allocation— will provide lower average returns over the next decade. In this environment, many investors have questioned the wisdom of thinking about asset allocation solely in strategic terms, and have shown renewed interest in strategies that constantly profit from evolving economic conditions and the temporary mispricings that result among individual geographies and asset classes.

qCIO's Global Macro strategy provides such a solution when implemented as an overlay program (GTAA). Our Global Tactical Asset Allocation (GTAA) solution helps our clients make their strategic asset allocation (SAA) responsive to the changing macroeconomic conditions without altering their long-term allocations. In that configuration, our dynamic strategy actively tilts our client's SAA to systematically exploit inefficiencies or temporary imbalances in equilibrium values among different asset or sub-asset classes while continuously complying with our client's portfolio constraints. Our clients profit from the opportunistic adjustments in qCIO's investment views in response to the changing patterns of risk and reward in the global financial markets through our advanced quantitative analysis of global pricing trends, business cycles, volatility levels and multiple other macro-economic signals.

Further, we elaborate on the principles upon which qCIO's quantitative Global Macro strategy are based, and provide a review of the best practices for developing or selecting a quantitative Global Macro or GTAA strategy. Specifically we review the components of a robust model, relevant qualitative and quantitative evaluation metrics, and the tools and processes needed to make optimal decisions regarding quantitative Global Macro and GTAA strategies.


I/ SAA, GTAA, and traditional active strategies side by side

    The case for SAA
SAA, also known as policy asset allocation, is the establishment of a long-term target allocation in major asset classes such as stocks, bonds, and cash based on portfolio objective, risk tolerance, and time horizon. Over time, SAA is the most important determinant of the total return of a broadly diversified portfolio with limited market timing. Studies support empirically the dominance of SAA in determining total return and return variability.

    The case for GTAA
TAA attempts to add value to SAA by over-weighting those asset classes or sub-asset classes that are expected to outperform on a relative basis and under-weighting those expected to underperform. In a TAA model, financial and economic variables (“signals”) are used to predict performance and assign relative short-term asset-class weightings. While a traditional TAA model generally consists of domestic stocks, bonds, and Treasury bills, GTAA models usually i
nclude assets such as currencies, commodities, and other alternative investments on a global basis. Some GTAA models can decompose to the sub-asset-class level to include growth and value stocks or corporate bonds and Treasury securities. While GTAA is an active strategy based primarily on systematically timing the market, it can be carried out through security selection or indexed investments.


Economically, GTAA is based on the assumption that relative returns among asset classes will diverge temporarily from equilibrium levels, allowing the opportunity for excess returns from systematic (generally contrarian) strategies. 

A well-
designed GTAA strategy will recognize an imbalance and suggest an underweight to an overpriced equity market, for example, resulting in value added for a portfolio. GTAA strategies are different fromsecurity-selection strategies in terms of both risks and benefits. It is helpful to understand these differences before implementing a strategy and choosing a manager. 

— Sources of returnThe source of return for a GTAA strategy can be illustrated through comparison with other strategies. GTAA strategies attempt to add value by timing systematic (or market) risk factors and over-weighting those asset classes that are expected to outperform. A passive indexed strategy also derives return from systematic risk factors. However, with a passive strategy, investors are simply compensated for assuming the market risk resulting from the variation in factors such as interest rates, term-structure shifts, specific business and industry shocks, and unexpected inflation shocks. In other words, passive returns result from beta, and GTAA returns attempt to produce alpha through a bet on systematic risk. Like GTAA, security-selection strategies attempt to produce alpha, but alpha is a bet on idiosyncratic or firm-specific risk, as opposed to systematic risk.

    Origins of TAA 
When compared with security-selection strategies, the usual method of timing systematic risk entails different risks and potential benefits. Understanding these differences is critical to the successful implementation of a GTAA strategy.

Historically, the biggest difference has been the the
available opportunity set, a limitation that has been partially alleviated in recent years. Tactical Asset Allocation (TAA) first emerged in the 1970s as an attempt to turn academic advances in quantitative finance into practice. In its earliest incarnation TAA focused exclusively on the US market allowing just three pair-wise comparisons (stocks-bonds, stocks-cash, and bonds-cash) in which to source potential mispricings. But as investor interest in global diversification increased, TAA likewise went abroad. Because pricing inefficiencies generally occur more frequently across geographies, the expansion into a number of non-US markets opened up a new breadth of opportunity for GTAA managers to potentially exploit. With access to the stock, bond and currency markets of more countries or regions a GTAA manager could effectively expand the opportunity set from three markets to approximately 35. Additionally GTAA managers assess each market simultaneously and incorporate two levels of investment decision: relative valuations across asset classes and across countries. For instance, an investment decision on the US equity market would question the attractiveness of US equity relative to US bonds, as well as relative to all other non-US equity markets in the global opportunity set. By matching up similar pair-wise comparisons, both within (intra-country) and across (inter-country) markets, 238 discrete views become available, each with a unique array of correlations and risk/reward scenarios. While the number of opportunities is limited by the number of countries and asset classes that managers can over- or underweight, GTAA managers have access to the most diverse and uncorrelated opportunity set. Unlike correlations among individual securities, correlations among asset classes are low, by definition. Low correlation facilitates independent bets and reduces the chance that two bets in one strategy will effectively cancel each other out.

Another difference is that single signal predictability was generally lower with TAA strategies.  GTAA managers have directly addressed this limitation by analyzing return differentials across asset classes and countries (as opposed to outright performance) and by utilizing advanced quantitative methodologies incorporating global pricing trends, business cycles, volatility levels and other macro-economic signals to allow for significant signal to noise ratios. 

While the combination of these two factors —lower predictability and a limited number of possible bets— used to make GTAA challenging, advances in methodologies and the unique nature of the opportunity set GTAA managers have access to more than outweighs those past limitations. GTAA strategies offer two additional benefits relative to security-selection strategies. Global Macro and GTAA strategies are implemented with low transaction costs and in highly liquid instruments. With other active strategies, trading individual securities typically involves significant transaction costs and the potential for side pockets resulting from investments in highly illiquid assets; while tactical shifts in major asset classes can be carried out with liquid futures contracts on an index, resulting in very low transaction costs and the highest possible liquidity. Note that tactical sub-asset allocation or global strategies may not all share this benefit. Liquid futures contracts on an index may simply not be widely available for all styles, segments, or markets.


II/ Evaluating a GTAA strategy

Although the successful implementation of a GTAA strategy is often portrayed as simple, it is actually difficult. As mentioned earlier, predictability is low and opportunity sets are limited. As a result, investors must be careful in selecting managers or strategies. They should understand a strategy’s information signals; for example, how a manager determines over- and under-weightings and what makes the strategy durable. It is important to use appropriate qualitative and quantitative performance measurement criteria and to identify strategies with relatively low costs. Some best practices for selecting or developing a TAA strategy follow. 

    Understanding how the forecasts are created. 
Success with a TAA strategy is largely dependent on constructing a good model. The first step in developing an overall TAA model is to forecast excess returns by constructing models that attempt to predict asset-class returns using a set of explanatory variables or signals. Running tests over a sample period will help to reveal the strength of the signals and the overall explanatory power of a model. Models may have varying predictive strengths during different periods. As a result, multiple predictive models are typically required to consistently add value. Models should also be dynamic; that is, they should change with structural changes or other factors that permanently affect signal strength. The text box below summarizes the commonly used TAA signals, their rationale, and the time periods over which they are expected to add value. A good forecasting model must include multiple economically meaningful signals and have a verifiable research process that follows a reasonable method to identify meaningful signals:

1/ Economically meaningful signals. Economically meaningful signals are those with rational, intuitive explanations for their expected predictive power. For example, the term spread, as an indicator of the business cycle,is intuitive and rational. Typically, the yield curve is positively sloped, meaning that long-term interest rates are higher than short-term ones. According to the term structure of interest rates, a positively sloped yield curve is compensation for the higher risk of locking in longer-term bonds and the uncertainty of inflation (and therefore the direction of interest rates) in the future. Investor expectations about the future economic environment affect the amount of this risk compensation and, therefore, the shape of the yield curve. Since market risk premiums and firms’ cash flows are linked with the future economic environment and the business cycle, a TAA strategy may benefit from systematically following a term-spread indicator.

Understanding the commonly used TAA signals
Investors should understand the rationale behind commonly used TAA signals and the time periods over which such signals are expected to add value so they can judge the skill of a manager and be comfortable with the types of bets made in portfolios.

— The “Fed model” signals:
This involves a model that compares earnings yields (the inverse of the price/earnings [P/E] ratio) to nominal bond yields to determine the relative attractiveness of equities over bonds. The underlying notion is that stocks and bonds compete for the same dollars and the higher returning asset class should be over-weighted in the portfolio. Clifford Asness (2003) detailed several drawbacks to this approach.
— Business-cycle/macroeconomic signals:
These signals attempt to find value added by timing the business-cycle-related variation in market risk premiums and firms’ earnings. Widely used signals include the term spread (the yield differential between long- and short-term bonds), the credit spread (the yield differential between high- and low-credit-rated corporate bonds), “unexpected inflation,” and industrial production. Business-cycle variables are typically implemented over intermediate time horizons.
— Fundamental-valuation signals:
One approach involves using fundamental firm-valuation metrics, such as dividend yield, book/market ratio, and P/E ratio, to determine relative valuation. Another approach is to use top-down or bottom-up cash-flow valuation methods (Damodaran, 2002). An example of the latter would be to use the dividend discount model to reverse engineer the required rate of return from market prices and projected dividend growth rates. Fundamental-valuation signals are typically implemented over intermediate time horizons.
— Momentum signals:
These signals attempt to add value by following the short-term momentum in markets. Typical momentum signals include technical indicators, earnings growth, and changes in trading volumes. Momentum signals can be at odds with fundamental or business-cycle signals at times, such as during the technology stock bubble of the late 1990s. However, when momentum signals are appropriately combined with fundamental or business-cycle signals, they can produce complementary strategies.
— Sentiment signals:
These signals attempt to add value through a contrarian strategy that looks for extreme levels of sentiment, such as consumer confidence and margin borrowing, to identify deviations from equilibrium returns. Sentiment signals tend to be implemented over intermediate time horizons.

2/ Absence of data mining. In addition, the manager should be able to confirm that he or she did not “mine the data,” the process of rerunning the model with modified signals until a desired result is reached and presenting only those results for the sample period. Data mining, while producing impressive results, runs the risk that the model will not work in real time. Out-of-sample tests of the strategy, such as in other time periods or countries, can help to confirm that the strategy’s success is not simply the result of fitting the model to explain one historical period.

3/ Rational decision process. Finally, like the signals themselves, the decision process for determining which signals are included and how they are combined should be economically reasonable. Consider a forecast that relies on a high R-squared to determine whether a signal is predictive and should be included. Despite a high R-squared, an economic rationale may suggest that the signal should not be used. A simple example of this is the “Super Bowl effect,” which predicts that the value of the Standard & Poor’s 500 Index will increase if an old National Football League team wins the Super Bowl or decline if an old American Football League team wins (example from Clifford Asness [1996]). The explanatory power of which team wins can be quite strong, but of course, there is no plausible rationale for a predictive relationship between football games and equity returns.

A good model will have a method for ensuring that the selection and testing process is economically reasonable. An example of this is the Bayesian method. Bayesian probability calculations involve assumptions based on economic theory or intuition, which are combined with what is learned from the data (Campbell and Thompson, 2005). For the model, the approach might be to assume that the prospective equity risk premium is positively related to dividend yield, for example. This would mean constraining the forecast model to disregard any results where the relationship between the dividend yield and the equity risk premium was negative. This approach provides a qualitative, rational overlay to the statistical-measurement process. It allows the model to disregard those results that strongly conflict with economic intuition.

    Understanding assets' weightings and determining and how risks are controlled.
Another factor to evaluate is a model’s asset weighting optimization process. First, over- or under-weightings should be proportional to the combined strength of the model’s information signals. If a strategy recommends a large over-weighting in a particular asset class, the manager should be able to demonstrate the model’s strong predictive power in that regard. Investors should avoid a manager who makes big bets based on weak signals. Second, over- or under-weightings should be made with the appropriate constraints on deviation. Constraints can be based on ad hoc decision rules. For example, a rule could state that if the projected equity risk premium is greater than 10%, then a manager should overweight equities by 15%. Or if optimization is used to determine asset-class weightings, constraints can be implemented with models that “tame” the optimizer. Traditional mean-variance approaches are highly sensitive to expected returns and often result in very large weightings in particular asset classes. The Black-Litterman Model (Black and Litterman, 1992) is an example of a model that corrects for these extreme results. This model starts with equilibrium expected returns and then moves away from them based on the volatility and correlations of each asset class and the degree of confidence in each forecasting model. The results of the model tend to be less extreme than traditional mean-variance approaches.

    The durability of value added.
Evaluate performance to determine what makes a strategy durable and how the back-tested performance compares with the real-time record. The following checks on a model help to determine this:

— Review the manager’s historical asset allocation relative to the benchmark. This process is critical to determine whether the model’s excess returns are simply a result of bias toward the historically higher-returning asset class, such as a strategic over-weighting in equities, or of predictive power (Lee, 1998). The manager’s historical allocations should not be largely different from the benchmark’s allocations for long periods. TAA involves short-term over- or under-weightings to capture price discrepancies. For instance, a longer-term over-weighting in equities would likely produce an excess return relative to the benchmark, but the source of the return would simply be the equity risk premium.

— Review significant events. It is important to review a model’s test results during periods marked by significant events—those that cause large divergences among asset-class returns— such as the U.S. stock market crashes in October of 1987 and 2008 or the August 1998 Russian debt-default crisis. If stock and bond returns are very close, the relative benefit of an over-weighting in either asset class is small. If their returns diverge, there are greater opportunities for bets to pay off and, of course, greater opportunities for loss. Historically very few events have caused returns to diverge significantly. However, these events highlighted the success or failure of a strategy. When the volatility of the equity risk premium is high, there are more opportunities to add (or subtract) value, which can result in higher (or lower) excess returns (Arnott and Miller, 1997; Lee, 1998).

— Examine a model’s results in different periods. Reviewing a model’s results in many different periods can help reveal whether the model is likely to be enduring. A manager’s back testing may show excess returns, but these may be the result of data mining, that is, running tests again and again until excess returns emerge. Out-of-sample results, those not presented by the manager, highlight any signal degradation. Signals may weaken over time. Also, it is not uncommon for a signal to produce significant excess returns in one period but not another. It is important to understand why the value added from the strategy was not consistently repeatable. The changes in the macroeconomic environment, weakening of the signals, or just bad luck are possible reasons that should be explored. A robust strategy should produce significant excess returns whether it is applied to in-sample or out-of-sample periods. A final metric to consider for TAA strategies over different periods is risk-adjusted return. TAA strategies can provide excess return at the expense of a disproportionate increase in volatility.

— Examine a model’s results over the intermediate term. Finally, it is important to examine a model’s results over the appropriate time period. Short-term predictability requires very informative signals. There is simply too much noise around short-term returns to find a pattern to determine which asset class will outperform. TAA strategies cannot be expected to work day-to-day or month-to-month. Over the intermediate term (perhaps three-year periods), it is more likely that some pattern or cycle related to economic variables, momentum, or change in investor sentiment can be gleaned and exploited.

    The appropriate quantitative and qualitative performance-measurement criteria.
As with any strategy, when evaluating a TAA strategy, it is useful to combine qualitative and quantitative performance-measurement criteria. In terms of quantitative measurement, the two metrics that deserve the most attention are the historical information ratio and the t-statistic associated with the historical average return. A historical information ratio calculated over five or more years can provide an indication of the risk/return trade-off that the strategy has offered and may continue to offer if the historical relationships hold. The t-statistic measures the consistency of the average historical excess return. A t-statistic greater than 2 indicates that the historical excess return has been and probably will continue to be very consistent—most likely because of the skill of the manager and the durability of the strategy— if the historical relationships hold.

In terms of qualitative criteria, we believe investors should evaluate any manager in terms of four principles: people, philosophy, process, and performance. Questions to ask include:
— How have all the manager’s strategies or funds performed, not just the ones being touted?
— Who is performing the work? How long has the team been in place?
— What is the manager’s approach to investments? Is it easily understood?
— How long has the strategy operated? Does it have at least a three-year track record?

Quantitative criteria help distinguish skill from luck in the historical performance of a TAA manager, while the qualitative criteria outlined above help improve the likelihood that historical relationships will continue to hold in the future. For example, the information ratio for all the manager’s strategies would provide more insight about manager skill than that for the one or two strategies being marketed. Likewise, information about the makeup and tenure of the team producing a particularly high information ratio would be important to determine whether the ratio is likely to be replicable.

    Assessing costs.
As with any investment, cost affects returns and should be evaluated. The higher the implementation and investment management costs, the higher the threshold for the success of a TAA strategy. Costs are impacted by portfolio turnover, which can be high for TAA strategies. Turnover results in different periods are shown for the business cycle/ macro strategy discussed earlier. Turnover, like signal strength, varies over time, and is not necessarily correlated with excess return.

The salient question is whether the excess returns cover the costs of the higher turnover. Break-even cost is the estimated trading cost that will eliminate excess return and therefore also generate loss against the benchmark.

TAA strategies are typically implemented with liquid futures contracts. This can be a benefit relative to security-selection strategies because transaction costs for futures trading in liquid markets are typically lower than those for trading individual securities. However, futures trading for TAA strategies requires expertise to be cost-efficient, particularly when a global TAA is implemented. For example, to maintain the benchmark position, hedging ratios must be accurately calculated. For equities, this is the process of measuring the sensitivity of changes in the futures index to changes in the benchmark. For fixed income securities, the process is more complex. In addition, global TAA and tactical subasset allocation strategies may require trading in illiquid futures markets, resulting in higher costs.
SelectionFile type iconFile nameDescriptionSizeRevisionTimeUser