Tactical Asset Allocation, Part I

by: Geoff Considine September 29, 2008 | about stocks: DJP / EEM / EFA / IDU / IGE / IVV / TIP    

Geoff Considine

About this author:
Become a Contributor Submit an Article
  • Font Size:
  • Print

The term “market timing” means different things to different people.  Several basic truths are unassailable.  First, most investors hurt their performance by chasing hot asset classes.  Second, there is evidence for momentum effects in asset class performance—which suggests that it might be possible to time the market successfully.  Third, the price you pay for something matters.  In the institutional world, market timing is typically referred to by a more respectable name: Tactical Asset Allocation [TAA].

TAA means that a portfolio manager considers the current valuation of an asset when considering its potential impact on the total portfolio.  In other words, I am not talking about momentum investing or any sort of high-frequency buying and selling.  TAA is a form of market timing, but implies something considerably more nuanced than most investors’ understanding of that term. 

When the real estate market was at its peak, it certainly made sense for an investor to consider that real estate was at an all time high as he/she considered adjusting an asset allocation plan.  To ignore the recent performance is foolhardy.  Even Nobel Laureate Bill Sharpe, a long-time proponent of efficient market theory, says that the current price of assets is a crucial consideration:

One of my favorite diatribes is, "How can you possibly do your asset allocation without looking at the market values of the asset classes out there today?" That's probably the most valuable information you can get as to the future prospects of those asset classes, and what astounds me is the fact that many people—many of them very sophisticated—do asset allocation without even checking to see what the relative values of the outstanding shares in, say, European securities, U.S. securities, emerging markets, etc., are. (Bill Sharpe, interview at IndexUniverse.com, 2008)

Dr. Sharpe’s “sophisticated people” who ignore current price levels are the proponents of   a pure “policy portfolio.”  Portfolio management using a policy portfolio means that you determine your targeted percentages and rebalance back to these percentages periodically—regardless of market behavior in the interim.

In a practical sense, TAA means that even if you have a long-term plan to hold 10% of your portfolio in an asset class, you may decide that it makes sense to hold less of that asset class if it has undergone a prolonged period of out-performance (and vice versa).

Strategic Asset Allocation and Tactical Asset Allocation

There is a long-standing debate among experts as to the relative role of Strategic Asset Allocation [SAA] and Tactical Asset Allocation [TAA].  SAA is the process of developing a portfolio based on long-term expected returns and correlations between

assets.  Altering a portfolio’s basic asset allocation by looking at relative valuations of assets is commonly referred to as Tactical Asset Allocation [TAA].  To quote Rob Arnott, TAA is “a focus on measured departures from the policy mix that are deemed likely to garner rewards.”   

For the interested reader, I recommend a useful analysis of TAA published by Vanguard’s institutional arm which explains TAA as follows:

Tactical asset allocation [TAA] is a dynamic strategy that actively adjusts a portfolio’s strategic asset allocation [SAA] based on short-term market forecasts. Its objective is to systematically exploit inefficiencies or temporary imbalances in equilibrium values among different asset or subasset classes. Over time, strategic long-term target allocations are the most important determinant of total return for a broadly diversified portfolio. TAA can add value at the margin…

The entire document from Vanguard is well worth reading in providing a balanced summary of TAA.

Rob Arnott and Peter Bernstein (among many others) have concluded that TAA plays an important role in portfolio management.  On the other side of the divide, we have John Bogle.  Mr. Bogle posits that the key steps in portfolio management are the following:

  1. Maintain an appropriate policy portfolio
  2. Diversify to the maximum possible extent
  3. Hold investment costs to the bare-bones minimum.
  4. Be realistic with your return expectations
  5. Hold on to your hats!

This five step process is a nice summary of the pure SAA model of portfolio management.  It is certainly true—as Mr. Bogle often asserts—that this form of low-cost low-turnover investing has historically provided better returns than most portfolio managers will achieve. 

My view of TAA is consistent with the quotes from Rob Arnott and from Vanguard.  You start with a target asset allocation (the SAA part of the process) that maximizes diversification benefits in the portfolio.  As I have discussed previously, there is considerable evidence that a truly well-diversified portfolio can add 2% or so in annual return relative to a generic stock / bond index mix.  TAA comes in when you look at the relative valuation and performance of the various asset classes you want in your portfolio.  Emerging markets have a place in a portfolio, but there have been times in recent years when you could only buy into these markets at very high prices—and the high price to be paid made emerging markets less attractive. 

I am a believer in Reversion To the Mean [RTM].  No asset class can out-perform forever.  Quantext Portfolio Planner [QPP] incorporates RTM as a key construct.  QPP projects long-term average returns for asset classes.  My approach to TAA is to look at the returns of an asset over recent years and to compare those returns to the expected long-term returns calculated by Quantext Portfolio Planner.  If the asset class has delivered returns below the projected returns, there is solid potential for a period of above-average returns---and the reverse is also true.

An Example

Rather than keeping this theoretical, I want to use a concrete example from an article that I published back at the start of November of 2007.  In this analysis, I selected a range of sector and broad index ETFs, calculated their trailing three-year returns, and used QPP to project long-term expected returns for each of these.  I proposed that a useful indicator of the timeliness of purchasing one of these sectors was to look at the three-year future returns that would be required to bring the combined trailing three year and future three year average return into line with the QPP-projected average return.  Asset classes that had out-performed their expected returns would be likely to generate below-average returns in the future, and vice versa.  This indicator is a bet on Reversion To the Mean [RTM] for each asset class. 

Back in late October of 2007, as I was writing this article, Quantext Portfolio Planner suggested that most asset classes that I analyzed were over-valued.  Over-valued asset classes may still be quite attractive in terms of projected return—even if they are likely to deliver less than in recent years.  When I sorted the list of ETFs by the three-year future returns that would, when averaged with the trailing three years, lead to an average over both periods equal to the QPP-generated long-term expected return, I got the following results—see original article:

 

There is no reason to believe that asset classes will average to their long-term expected returns over a six year period, but this calculation can serve as an estimate of the projected returns—combining the risk-return balance of an asset class with its relative over- or under-valuation. 

When we calculate the total returns of each of these ETFs (and one ETN: DJP) from November 1, 2007 through September 23, 2007 (when I am writing this), we obtain the following results:

 

These results are sorted as in the original article.  The ETFs in the top third (6 out of 19) in terms of projected returns to bring the returns back into line with the QPP-expected return have averaged -12% since November 1, 2007.  The ETFs in the bottom third have average -21% over the same period.  The median returns for each group are generally consistent.  The spread of 9%-10% in performance between the top third and the bottom third shows the impact of tactical asset allocation via Reversion To the Mean [RTM].

There were three asset classes / ETFs for which QPP projected that the next three years would need to have negative returns to bring the trailing returns into balance with long-term expected returns generated by QPP: IXP (global telecom), EFA (developed international), and EEM (emerging markets).  These three have all delivered very negative returns this year.  An approach to TAA that simply under-weighted these asset classes would have reaped benefits. 

Of the asset classes / ETFs that were projected to have the highest returns, the worst performer has been IIH (internet)—with a return of -30%.  That said, QPP estimated an annualized standard deviation in return (going forward) of -52% using data through October 2007.  The internet has always been a volatile sector and IIH, in particular, is concentrated in very risky stocks.  In other words, IIH was a very high risk proposition and thus any sensible asset allocation would have limited exposure to IIH to a tiny portion of the portfolio.  EEM, by comparison, has a projected standard deviation of 25% at that time.  EFA and IXP had projected standard deviations of 13% and 15%, respectively.  Note: QPP users can easily verify these results, running QPP with default settings for the three years through October, 2007. 

Now, nobody is going to be thrilled with a return of -10% (the median for the third of ETFs which TAA indicated were most under-valued) --- except when you consider that this is in the context of a drop in the S&P500 of -20%.  Let me be clear, however, that I am not endorsing a strategy of simply buying or selling individual asset classes based on an indicator.  The value proposition of TAA is in combination with Strategic Asset Allocation (SAA).  Simply looking at the spread between projected ‘fair’ returns and trailing returns does not tell you anything about risk and return of a total portfolio—this is where SAA is the key.  To the extent that an investor tilted his or her asset allocation plan based on these projections, returns would tend to be higher.

Combining Strategic and Tactical Asset allocation

To combine SAA and TAA, we would be looking to build a portfolio with following properties:

  1. Highly diversified
  2. Targeted risk level
  3. Low expenses
  4. Projected return greater than trailing return

The TAA component comes in with number (4).  This approach is a far cry from a high turnover ‘timing’ model and is broadly consistent with both the standards of practice for TAA and with a long view based on a consistent balance between risk and return over long periods (i.e. SAA).  If you can buy the assets you want in your strategic asset allocation at a discount, this will boost your long-term performance.  One of my favorite case studies of a portfolio that meets all of these criteria was based on Berkshire Hathaway’s twenty top equity holdings.  This study, published almost exactly two years ago, found that Berkshire’s holdings were both well-diversified and were projected to deliver higher returns in the future than in recent years: they captured both SAA and TAA.  The two years since have borne out the results of this study—and the details of these results will be discussed in detail in a follow-up article. 

 

Related Articles

This article has 14 comments:

  •  
    Sep 29 05:37 PM
    this is TAA in motion:

    SMN UltraShort Basic Materials ProShares22.92%
    EEV UltraShort MSCI Emerging Markets ProShares20.96%
    QID UltraShort QQQ ProShares10.34%
    SIJ UltraShort Industrials ProShares9.86%
    DGP PowerShares DB Gold Double Long ETN9.73%
    SDS UltraShort S&P500 ProShares9.64%
    IEF iShares Lehman 7-10 Year Treasury Bond Fund9.47%
    TWM UltraShort Russell2000 ProShares7.08%
    Reply | Link to Comment
  •  
    Sep 30 01:50 PM
    To phdinsuntanning:

    If you can time the direction of the market well enough to consistently be short when the market is going down (and vice versa), you don't need to worry about things like SAA, TAA, or risk management--you will make such large returns that you will own a large investment bank in no time at all. My writing is for those of us who lack your powers.
    Reply | Link to Comment
  •  
    Sep 30 02:05 PM
    I find your article interesting and I’d like to see how your approach worked over substantially more time periods. However, I think mean reversion has some statistical issues. Example: you flip a coin that comes up heads 10 times in a row, what do you bet the next flip will be? Trend followers would bet heads ... mean reversion tells us to bet tails. What is the right answer? The right answer is ... it doesn't matter. No one knows coin flipping is random; the next flip has exactly the same odds as any other flip 50/50 heads/tails. The stock market is random and on average over the long term neither a simple trend following or a simple mean reversion analysis will give you the edge. The market is more complicated than that. Mr. Bogel's approach seems to be the wiser approach here. He is basically recommending Modern Portfolio Theory (MPT).

    However, I do believe that the way advisors generally apply MPT today could be improved with the result leading to improved risk adjusted returns. First lower costs (per Mr. Bogel) could certainly be achieved with index ETFs. Second, standard deviation should be replaced as a measure of risk, using some measure of downside risk only. Standard deviation considers upside and downside volatility as equally bad. Yet, no investor worries about unusual upside returns. Third, large Growth, large Value, Mid-cap, and small-cap growth/value asset classes have correlations converging on one. This type of diversification doesn’t reduce risk especially in down markets. Investors and their advisors need to focus on a new group of asset classes with truly low correlation.

    These are principals being brought to the forefront of academic theory by Behavioral Finance and Post-Modern Portfolio Theory (Post-MPT). For more information on investments using Post-MPT principals visit isectors.com. There you will find examples of managed investment allocation models with a history of outperforming the market using these simple approaches to apply MPT more effectively. In addition, if you’re looking for more on the “science” (financial theory) behind the Post-MPT investment allocation models …read “Practical Applications of Post-Modern Portfolio Theory” also found on the iSectors® website.
    Reply | Link to Comment
  •  
    Sep 30 02:49 PM
    perhaps a simpler approach is the Point and Figure methodology which is probably the oldest in investment methodology.
    It helps you determine three things;
    1. what is the risk level of the total market.
    2. what sectors are performing best relative to all others
    3. What is the risk level of individual stocks within each of the best sectors.

    The foundation of the method is relative strength. I find it very helpful, once learned. just a suggestion. Many books/sites availble to teach it.
    Reply | Link to Comment
  •  
    Oct 01 09:38 AM
    yes Geoff, maybe is some fruit that I got in the jungle, by the way I like your academic approach, just take care of black swans. Jokes apart do you know any SAA or TAA model for preference shares? (stocks which provides a specific dividend that is paid before any dividends are paid to common stock holders, and which takes precedence over common stock in the event of a liquidation). Some good oportunities are around in financials...
    Reply | Link to Comment
  •  
    Oct 01 02:07 PM
    Geoff, since Vernl brought it up, why does QPP use standard deviation to measure/analyze risk as opposed to downside deviation? It seems to me that an investment that has a high standard deviation but low downside deviation would be quite different than an investment with identical standard deviation but higher downside deviation. If you are using normal distribution of returns, as I believe QPP is, then standard deviation and normal distribution may not paint the most accurate picture. Just curious.
    Reply | Link to Comment
  •  
    Oct 01 02:59 PM
    Mynion:

    Okay--here's the argument: Volatility (up or down) represents the markets uncertainty as to how to value the future earnings stream of a company--this uncertainty is a measure of risk. An investment with very high skewness (asymmetry between upside and downside) would have important implications but realized volatility is a measure uncertainty and the magnitude of changes in opinion in the overall market. Frankly, I have nothing against modeling skewness but every increase in statistical complexity brings its own challenges. Estimating skewness is harder than estimating variance because skew is a cubed statistic...a few data points can easily sway the stats. I do lean towards the simplest possible models--you are correct--largely because additional parameters lead to their own issues.

    Geoff
    Reply | Link to Comment
  •  
    Oct 01 03:34 PM
    Thanks Geoff.

    Would you say, then, that reducing volatility as measured by standard deviation, by default reduces risk as measured by downside deviation?

    If so, constructing a portfolio that maximizes return for risk, as measured by standard deviation, would still be inherently efficient.

    In regards to TAA, which the article is about, have you read "A Quantitative Approach to Tactical Asset Allocation" by Mebane T. Faber (it is available on cambriainvestments.com I believe)? Just wondered what you thought of the research.

    Thanks Again! Keep up the great work!
    Reply | Link to Comment
  •  
    Oct 02 12:53 PM
    Hi Goeff,

    Another interesting and informative piece.

    Regarding the internet etf, IHH, I notice that there was an odd downward move in the price of the index on one day in May. There was barely a corresponding volume blip or even the typical explosion in volatility one typically sees after a movement of this size. Of course, arbs sometimes come in to exploit the premiums or discounts on these etfs, but then you see the volume spikes and volatility. There are 14 stocks in the index and I looked at the price action of the top 10 of them on that day. There was no corresponding price move in any of them. Perhaps the etf was either restructured, reorganized or split.

    In any case, there are odd changes in securities prices because of cap distributions and reorganization, spin outs and what not. I’m wondering how qpp deals with these odd events to ensure robust and accurate data in the program?

    Thank you,

    JMorace
    Reply | Link to Comment
  •  
    Oct 02 12:53 PM
    That's IIH......
    Reply | Link to Comment
  •  
    Oct 04 02:11 PM
    Geoff: i hope you don't get tired of my posts, but i have to tell you, once again, great article!!!!!!!
    Reply | Link to Comment
  •  
    Oct 06 12:54 PM
    I'm new to this forum, so hopefully this is an ideal place to start with a comment.

    If one were able to buy MVV or any of the 2X long ETFs on margin they'd in effect be 4X long the index. Correct? So, in the event the investor might time that purchase such that the index subsequently rallies +10% , the investor might be up about +40%. Of course, this assumes the risk that the index might continue falling, and they are on the hook X4 to the ongoing losses that index might continue sustaining. This is not for the risk averse investor, however it seems like it might be a tempting investment tactic for those who are able to handle the risk.

    After seeing QLD fall from $83.55 on 8/15/08 (at its 200 DMA resistance) to the $40 area now, the probability of a +10% rally in the coming weeks seems rather good. Anyone?

    I originally attempted to post this comment in response to the 7/11/06 article by Greg Newton seekingalpha.com/artic...
    which mentioned the then soon to be introduced 2X ETFs linked to the major indices

    2x Long the NASDAQ-100 (QLD)
    2x Long the S&P 500 (SSO)
    2x Long the S&P MidCap (MVV)
    2x Long the Dow Jones Industrial Average (DDM)

    2x Short the NASDAQ-100 (QID)
    2x Short the S&P 500 (SDS)
    2x Short the S&P MidCap (MZZ)
    2x Short the Dow Jones Industrial Average (DXD)
    Reply | Link to Comment
  •  
    Oct 07 12:01 PM
    Vernl, downside risk only measures the standard deviation below the mean. In the real world, a loss is a peak-to-trough drawdown, so standard deviation is inherently superior as it is the probability weighted maximum point above the mean to the minimum point below the mean. If the maximum historical drawdown is what one is worried about, then a simple heuristic ( 1 / maximum drawdown in % ) can be used to allocate.
    Reply | Link to Comment
  •  
    Oct 12 03:43 PM
    Geoff,

    If you believe in reversion to the mean, it would seem that IIH is a great buy right now, wouldn't it? Also, have you ever considered taking a short position on ETFs that QPP predicts will have a significant negative return?

    Thanks,
    Ben
    Reply | Link to Comment
More Comments on DJP »
Top Rated Comment Streams:

Numbers are net rating-

See all Top 100 »
More by Geoff Considine

Investment Resources
Money Management
Research
Seeking Alpha provides these ads as a service to its leading contributors
Articles on related themes
Asset Allocation with ETFs
Portfolio & Risk Management
Closed-End Funds
Commodity ETFs