Summary

  • The power of strategic asset allocation is best harnessed when future asset class behavior is relatively certain and diversification opportunities abound.
  • Uncertainty around rising rates and the current monetary policy environment may call both of these criteria into question.
  • Holding all else equal, tactical asset allocation (“TAA”) is most likely to add value in environments where diversification opportunities are scarce.
  • We believe that best practices for TAA include: (1) Using TAA as a complement to core strategic allocations, (2) incorporating TAA in such a way that overall strategic allocations are not skewed, and (3) using multiple TAA signals / managers.

As of Friday, we are more likely than not to see a rate hike by the end of the year.  The chart below plots the probability of the Fed Funds target being set to certain levels at upcoming FOMC meetings. The market-implied chance of seeing at least one rate hike before year end is 74%.  These probabilities are based on CME Group 30-Day Fed Funds futures prices and can be studied in more detail here.

pic1

Source: CME

The 10-year Treasury rate closed Friday at 1.85%, 48bps above the all-time low of 1.37% from July.

Future interest rate movements continue to be the subject of endless debate.  For our tactical strategies that are driven by momentum, rising rates are not necessarily a concern in isolation.  Evidence suggests that the performance of trend-following strategies has been robust across interest rate regimes.

However, we believe that in all the interest rate talk, an equally important topic - how current monetary policy and interest rate dynamics will affect future stock/bond correlations – has been neglected. While expected returns can play an important part in portfolio construction, correlations are essential for determining diversification effects, and assuming the past will be the same as the future can be dangerous.

Over the last two decades, we have gotten used to negative stock/bond correlations.  Negative stock/bond correlations, however, are far from a sure thing.  In the next graph, we plot rolling 1-year correlations between S&P 500 returns and changes in the 10-year Treasury yield.  We see that correlations have varied significantly over time.

pic2

Data Source: Federal Reserve Bank of St. Louis.  Calculations by Newfound Research.  Past performance does not guarantee future results. 

IMPORTANT NOTE: Positive numbers imply a positive relationship between stock returns and interest rate changes.  Positive correlation between stock returns and interest rate changes is equivalent to negative correlation between stocks returns and bond returns.

The future correlation between core stocks and bonds has tremendous implications for both strategic and tactical asset allocation.

Impact of Correlations on Strategic Asset Allocation

From a strategic perspective, correlation changes can significantly alter the composition of the Sharpe-optimal portfolio.  Using J.P. Morgan’s capital market assumptions, we computed the mix of U.S. Equities and 10-Year U.S. Treasuries that maximizes the portfolio’s expected Sharpe Ratio for three different stock/bond correlations: -0.5, 0.0, and +0.5.  At lower correlation levels, 10-Year Treasuries receive significant allocations as a result of their diversifying potential.  When correlations rise, the diversification benefit does not outweigh the cost of allocating capital from a higher expected return asset class (stocks) to a lower expected return asset class (bonds).

pic3

Data Source: J.P.Morgan.  Calculations by Newfound Research.  Allocations illustrative and are not to be considered investment advice.

Impact of Correlations on Tactical Asset Allocation

The impact of correlations on tactical asset allocation is a little more nuanced.  To explore the impact, let’s use a simple example.  Assume there are two securities.  Both securities have expected excess returns of 5% and volatilities of 10%.

The Sharpe-optimal portfolio is a 50/50 blend regardless of the correlation between the two securities (assuming the correlation is not exactly equal to 1).  While correlation does not affect the portfolio composition, it does impact the volatility and the Sharpe Ratio of the portfolio.  As correlation decreases, more diversification is available.  As a result, the portfolio is able to achieve the same return with less risk, increasing risk-adjusted return (e.g. Sharpe ratio).

pic4

Calculations by Newfound Research.  This is a hypothetical example that does not reflect any actual investments or asset classes.  Credit to AQR for inspiring our use of this framework. See here.

Now let’s assume that we want to implement a fully flexible tactical asset allocation overlay that can adjust the weight in each security from 0% to 100% (the sum of the weights in each security must still sum to one).  For now, we will assume that the tactical asset allocation makes completely random decisions (“Random TAA”). 

Completely random decisions will be right half of the time and wrong half of the time.  As a result, Random TAA will have the following impact vs. the strategic 50/50:

  • Expected Return: Unchanged.
  • Volatility: Higher. In our simple example, the 50/50 portfolio is both the Sharpe-maximizing and volatility-minimizing portfolio.  The Random TAA will on average move the portfolio away from 50/50 and so will increase volatility.
  • Sharpe Ratio: Lower. Unchanged expected return plus higher volatility equals lower risk-adjusted returns.

The size of the Random TAA’s cost (lower Sharpe ratio) will be a function of the correlation between the two securities.

When correlations are low, the diversification benefit foregone by moving away from the 50/50 portfolio is high.  At a correlation of -0.5, Random TAA reduces the Sharpe ratio by 29% (from 1.00 to 0.71).

When correlations are high, there is little diversification to be achieved and so the Random TAA approach is not much worse than the strategic 50/50.  For example, Random TAA only reduces the Sharpe Ratio by 5% (from 0.58 to 0.55) when the correlation between the two securities is +0.5.

pic5

Calculations by Newfound Research.  This is a hypothetical example that does not reflect any actual investments or asset classes.  

To this point, we have assumed that the TAA makes random decisions.  While this is unrealistic, it is still a helpful construct.  Why?  Because we can use it to compute how skillful a TAA provider must be in order to breakeven.  We define breakeven as the information ratio that the TAA manager must achieve so that the portfolio (strategic 50/50 plus TAA overlay) achieves the same Sharpe ratio as the strategic 50/50.

The breakeven information ratio will depend on correlation:

  • Higher correlations --> Less diversification foregone by TAA --> Lower hurdle for TAA to add value
  • Lower correlations --> More diversification foregone by TAA --> Higher hurdle for TAA to add value

pic6

Calculations by Newfound Research.  This is a hypothetical example that does not reflect any actual investments or asset classes.  

Unfortunately, information ratios are not very intuitive.  Below we transform the breakeven information ratios to breakeven hit rates.  The breakeven hit rate defines the percentage of trades that must be right (i.e. make money) for the TAA to breakeven vs. the strategic portfolio.  As a reference point, the hit rate for each of our 5% return / 10% volatility asset classes is 69%.

pic7

Calculations by Newfound Research.  This is a hypothetical example that does not reflect any actual investments or asset classes.  

We think this simple model underscores two points that we constantly stress in regards to tactical asset allocation:

  1. While we believe tactical allocation has a place in most portfolios, it is not a substitute for a well-diversified strategic core.
  1. The relative attractiveness of tactical and strategic asset allocation will depend on the diversification opportunities available. When diversification opportunities abound, strategic asset allocation is hard to beat.  When diversification opportunities disappear, tactical asset allocation can shine.  [Note: We think this point is very relevant to the ongoing discussion on timing of equity factors.  Unlike many long only asset classes, equity factors have a more consistent history of maintaining diversification to each other.  More diversification means that the hurdle to overcome for factor timing to beat a diversified factor portfolio is much higher].

If you believe that the current interest rate environment will lead to increased asset correlations, then it may be time to reconsider how you are employing tactical asset allocation within your portfolio.

The model we’ve outlined here can also be used to provide intuition around other best practices for using TAA.

How you use TAA can matter just as much as picking the right manager or approach.

Let’s say we have our optimal 50/50 portfolio and we want to incorporate at 25% TAA sleeve.  We need to decide where we want to fit this 25%.  We could take it from asset A, asset B, or a combination of the two.

The correct usage is to fit in the TAA such that we stay as close as possible to our 50/50.  Why?  Because the farther we move away from the 50/50, the more diversification we give up, raising the hurdle for the TAA to add value.  In this example, we’d allocate 37.5% to asset A, 25% to TAA, and 37.5% to asset B.  The TAA serves as a pivot around the optimal 50/50 strategic portfolio.

If instead we use the TAA in place of only one asset class (i.e. 25% asset A / 25% asset TAA / 50% asset B), the breakeven information ratio doubles[1].

Real world takeaway: If you have a TAA strategy that allocates between stocks and bonds, it should in many cases be used a pivot between your core strategic stock and bond holdings. This assumes that the initial strategic portfolio is optimal.  Another sound usage for TAA is as a tool to move a sub-optimal portfolio closer to being optimal.  Consider a moderate client that hates owning bonds due to fears of rising rates.  The client is allocated 75/25 between stocks and bonds even though the optimal portfolio for the client’s risk profile is 50/50.  For this client, allocating 25% away from equities and towards a TAA strategy can make sense because it will on average tilt the portfolio further towards the optimal 50/50 portfolio.  In this case, bringing the original portfolio closer to the optimal allocation outweighs the increased information ratio hurdle.

Use multiple TAA managers with differentiated investment processes.

Diversifying across TAA managers with complementary investment processes (i.e. one manager that uses momentum, one that uses valuation, etc.), accomplishes two things:

  1. Lowers the likelihood of larger tactical bets, which are precisely the allocations that sacrifice the most diversification.
  1. Lowers the hit rate that any individual manager must achieve for TAA to add value at the portfolio level.

pic8

Calculations by Newfound Research.  This is a hypothetical example that does not reflect any actual investments or asset classes.  

When reading the above graph, it’s important to note that while hit rates may be more intuitive, information ratios are the better tool for evaluating how much TAA skill is needed to breakeven in each scenario.  When four independent TAA signals/managers are used, the breakeven information ratio declines from 0.19 to 0.10, implying that each individual signal or manager only needs about half of the skill.

Addressing real world complexities

As with any simplified example, we can’t hope to perfectly represent the real world.  In particular, we think there are three topics worth considering that are not addressed by the simple model:

  1. The model assumes that asset returns are normally distributed with constant means, volatilities, and correlations. In reality, asset classes can exhibit much more complex behavior.  It also assumes that investors seek to maximize return per unit of volatility.  In reality, investors may care about other criteria as well (i.e. limiting drawdowns, tracking certain benchmarks).  In these cases, TAA - especially those that focus on downside risk management – may be particularly valuable.
  2. The model assumes that the optimal strategic asset allocation is known. It reality, we can never be sure that our strategic allocation is optimal for the future. It is entirely possible that an investor’s initial strategic allocation is actually suboptimal and that TAA moves the portfolio to a more diversified state.  In this case, TAA may improve risk-adjusted returns even if its information ratio is zero or negative.
  3. The model does not address the possibility that the performance of a given TAA signal or process may be related to the correlation between the asset classes it tilts between. If the TAA does better when correlations are higher, then TAA becomes even more (less) attractive when correlations are higher (lower).  If the TAA does better when correlations are low, then it’s possible that low correlation environments may be more attractive for TAA even though breakeven hurdles are higher.

Conclusion

The ability of TAA to add-value to a portfolio will depend on the correlations between asset classes in its universe.  When asset class correlations are low, the cost of tactically moving away from the optimal weights is high in the sense that valuable diversification is foregone.  This increases the hurdle that TAA must clear to positively contribute to risk-adjusted returns.  When asset class correlations are high, this opportunity cost is significantly lower and TAA may be able to add significantly more value.

These conclusions rely on the assumption that the strategic portfolio is optimal.  In the real world, the strategic portfolio may be sub-optimal and asset class behavior and client expectations may be significantly more complex.  In these cases, conclusions may change.

We believe that TAA should be used as a complement to core strategic asset allocations and that users should combine signals and/or managers with unique approaches.

Client Talking Points

  • The power of strategic asset allocation is best harnessed when future asset class behavior is relatively certain and diversification opportunities abound.
  • Uncertainty around rising rates and the current monetary policy environment may call both of these criteria into question.
  • Tactical asset allocation, including strategies driven by trend-following, can be a strong complement to core strategic allocations.
  • All else being equal, the potential benefit of tactical asset allocation will be greatest when diversification opportunities are scarce (e.g. 2008).

 

[1] We assume correlation of 0.0.