Open Journal Systems

(1)

ORIGINAL ARTICLE

TAA Properly Defined

Robert A. Brown

Chief Investment Officer and Senior Executive, Integrated Financial Partners, Inc.300 Fifth Avenue, 3rd Floor, Waltham, United States

Abstract: Tactical Asset Allocation (TAA) has generally been misspecified, oversold, and subsequently underdelivered. Nevertheless, TAA offers a series of highly attractive investment attributes when adviser/client expectations are proper-ly set and the strategy is appropriateproper-ly positioned as a portion of a comprehensive investor solution. This article’s objec-tive is three-fold. First, to identify the attracobjec-tive investment attributes of TAA relaobjec-tive to passive buy & hold. Second, to quantify or parameterize these relative advantages so that users can better assess the relevance of TAA for their own specific needs. Third, this article’s last objective is to describe the give-ups or tradeoffs associated with TAA, so that it can be properly understood, communicated, and therefore applied to the correct portion of an investor’s aggregate port-folio.

Keywords: Tactical Asset Allocation; policy implications; TAA tradeoffs

1. Introduction

Tactical Asset Allocation (TAA) has generally been misspecified, oversold, and subsequently underdelivered. Consequently, TAA has earned a poor reputation with many advisers and their clients. This is unfortunate since TAA offers a series of highly attractive investment attributes when adviser/client expectations are properly set and the strate-gy is appropriately positioned as a portion of a comprehensive investor solution. This is not to suggest that TAA is a superior approach to other investment solutions, only that it has clear relative advantages with respect to certain invest-ment attributes but at the cost of certain relative disadvantages with respect to other attributes. In other words, there are tradeoffs.

Despite recent disenchantment with TAA, it remains a popular strategy both within the advisory and institutional channels, amassing between $100 and $200 billion of AUM. The purpose of this article is NOT to provide evidence that TAA does or does not add value.Instead, it starts with the observation that TAA’s use within the wealth management industry (both institutional and retail) is both large and rapidly growing. This article’s objective is to more adequately define the attributes, both pro and con associated with TAA. Specifically, our objective is three-fold. First, to identify the attractive investment attributes of TAA relative to passive buy & hold. Second, to quantify or parameterize these relative advantages so that users can better assess the relevance of TAA for their own specific needs. Third, to describe the give-ups or tradeoffs associated with TAA, so that it can be properly understood, communicated, and therefore ap-plied to the correct portion of an investor’s aggregate portfolio. Given the large asset pools currently managed using various TAA processes, these objectives are neither academic nor overly artificial.

In brief, it will be shown that the benefit provided by TAA is in the realm of risk mitigation. That it reduces maxi-mum drawdown, monthly volatility, and probability of loss. But these benefits come with tradeoffs relative to passive, Copyright © 2019 Robert A. Brown

doi: 10.18686/fm.v4i1.1097

This is an open-access article distributed under the terms of the Creative Commons Attribution Unported License

(http://creativecommons.org/licenses/by-nc/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

(2)

fixed-weight buy & hold. First, tracking error increases. Second, the frequency of monthly underperformance becomes commonplace. Third, the longevity of sequential monthly underperformance becomes pertinent. A failure to adequately understand both the significant benefits and tradeoffs associated with TAA will only lead to misrepresentation, incorrect application, and inevitable disappointment. None of which is necessary.

2. Evaluative framework

TAA takes many forms. But most rely on two simple observations. First, that markets trend. In other words, once a market starts going up, it has a tendency to keep going up. Similarly, when a market starts going down, it has a ten-dency to keep going down. Second, that bear market declines tend to last for a long period of time. This trending be-havior opens up the possibility for TAA to enhance portfolio bebe-havior but without having to rely on forecasts of the fu-ture. Instead, it relies on markets continuing to trend.

This article explores the simplest of TAA trending models, one which relies on a market’s current level relative to its short-term moving average. To examine TAA’s robustness, we examine numerous different asset categories, combi-nations of asset categories, and different time periods. To identify and quantify TAA’s benefits and costs, we compare it to relevant passive, fixed-weight buy & hold performance benchmarks. Through this comparison, we provide evidence that TAA’s benefits span all asset categories, all geographies, and all time periods. Because we analyze 97 years of data, our evaluation spans several quite different inflationary (and deflationary) environments. Since investors care about ―net‖ returns (what they take home after the effects of inflation), our analysis is conducted exclusively in real space (after inflation has been subtracted out).

This article adopts the position that the benefits of TAA are founded on the observation that markets trend. How-ever, a robust examination of this position requires that we also examine how TAA behaves if markets do not trend. With this objective in mind, later in the article we examine the behavior of TAA in trendless markets, those defined by return distributions that are independent and identically distributed (iid). The results of this ―counter-analysis‖ are star-tling and serve to further reinforce the veracity of this article's findings that TAA is both robust and attractive but relies critically on markets’ inherent trending attributes.

Finally, we end with a review of TAA’s fundamental tradeoffs: those attributes that must be sacrificed or given up in order to earn the benefits of tail-risk mitigation and reduced monthly volatility. This article provides evidence that the downsides include low benchmark correlation, frequency of underperformance, and longevity of underperformance.

The policy conclusions drawn from this article’s findings are several. First, TAA may be an attractive solution for those who prefer tail risk mitigation. Second, the risk management properties of TAA, may make it the preferred strate-gy for those with longer investment time horizons, such as a dozen years or more. Third, advisers who need a solution for all seasons should stay strictly away from all things TAA.

3. Analysis

The analysis presented herein, is based exclusively on live monthly total return data. All data was provided by Global Financial Data (GFD). The period of evaluation encompassed the last 97 years, spanning 1/31/1920 through 1/31/2017. The objective was to examine the relative behavior of TAA across different asset categories and different timeframes. For this reason, we examined 31 different asset categories:

- U.S. stocks - 5 definitions (4 indices and 1 equal-weighted composite of the 4) - Non-U.S. stocks - 7 definitions (6 indices and 1 equal-weighted composite of the 6)

- U.S. bonds - 6 definitions (3 treasuries, 2 investment grade corporates, and 1 equal-weighted composite of the 5) - Non-U.S. bonds - 6 definitions (5 indices and 1 equal-weighted composite of the 5)

- Commodities - 1 definition

- Combinations - 6 definitions (1 global stock, 1 global bond, and 4 balanced mixes)

By examining the behavior of TAA across stocks, bonds, and commodities and pursuing this analysis on a domes-tic, international, global, and balanced basis, we are able to demonstrate the robustness of TAA across a broad range of asset categories. Moreover, in order to show TAA’s attractive attributes across differing time periods, we broke our

(3)

analysis into two time windows, i.e., before 3/31/1969 and after 3/31/1969.

Because investors are concerned with what their money can actually purchase, all of the analysis presented herein was conducted using inflation-adjusted real returns. This was particularly important because the inflationary and defla-tionary environments have varied considerably over the last 97 years. The inflation adjustment was made using the Consumer Price Index, All-Urban Consumers, Not-Seasonally Adjusted.

The TAA model examined here is the simplest possible. In a real world situation, more sophisticated and nuanced models would be constructed. We adopted the simple modeling approach in order to maintain transparency and help support the robustness of the examination. For each of the 31 different asset categories, the TAA model adopted was a simple binary ―in‖ or ―out‖ decision process. If the inflation-adjusted total return growth of a dollar level of the asset category was sufficiently above its short-term moving average, then the model was 100% ―in‖ the applicable asset class. Otherwise, it reverted 100% to 90-Day U.S. Treasury Bills.The number of months over which the moving averages were calculated was based on what worked effectively for each asset category. The range of time windows examined ranged from as short as eight months to as long as eighteen months.

To evaluate TAA’s performance attributes, we compared each model to a customized plausible mance benchmark. For this benchmark, we used a passive, fixed-weight buy & hold mixture of the asset category in question and the 90-Day U.S. Treasury Bill. The weighting between these two assets was customized so that the return on the TAA model (over the entire time period, 1/31/1920 through 1/31/2017) was exactly the same as for the pas-sive, buy & hold benchmark. By constructing the benchmark in this fashion, we can immediately identify whether the TAA model contributed any risk management benefits relative to passive buy & hold (or not). Through this compari-son, we isolate risk and tracking attributes of TAA versus buy & hold.

4. Findings- A Representative asset category

A representative asset category to begin with is provided by the S&P 500 Utilities Index. Exhibit 1 provides the summary statistics comparing the TAA model for utilities versus its passive, fixed-weight performance benchmark. Over the last 97 years, the TAA utilities model delivered a real return of 6.77%. A passive, buy & hold mixture of cash and utilities (using monthly rebalancing) with 108% allocated to utilities and -8% allocated to cash, generated the same 6.77% real return. Thus, we use this levered mixture (of cash and utilities) as the comparative benchmark. The TAA model allocates 100% to utilities if the current level of the index is more than 4.45% above its 17-month moving aver-age, otherwise it reverts to T-Bills.

Exhibit 1 shows the risk characteristics for these two portfolios (TAA vs. benchmark) and segments these statistics into two halves, before 3/31/1969 and after 3/31/1969.For each period, we show the annualized standard deviation, maximum drawdown, and the likelihood (probability) that over a randomly selected 12-month time window, the portfo-lio’s return would be less than the identified percentage return. As supported by this data, the TAA model delivered re-markably large reductions in each of these risk metrics during both the first and second timeperiods.

The remainder of Exhibit 1 focuses on tracking error. The objective here is to identify the tradeoffs associated with using a TAA strategy - for nothing in life is free, including risk-mitigation. This section of the table begins with correla-tion. Note that the correlation between the TAA model and its benchmark comes in at a remarkably low 0.63. In practi-cal terms, the investor in such a strategy will experience a very different ride from that provided by passive buy & hold, and must be prepared in advance for this alternate journey.Similarly the tracking error between the TAA model and its benchmark is unusually high with a 15.8% annualized standard deviation. This level of tracking error is the norm for TAA strategies. Once again, users must be prepared for this alternate journey and not be discomforted by the lack of similarity with buy & hold index comparisons.

(4)

The bottom of Exhibit 1 focuses on the frequency and longevity with which the TAA model underperforms its pas-sive benchmark. In the case of utilities, the longest number of consecutive months that the TAA model underperformed the benchmark was twelve (over the last 97 years). Moreover, the frequency with which the utilities TAA model under-performed relative to its benchmark, for different length windows (ranging in length between one and fifteen months), was between 58% and 72% of the time. Users of TAA models must appreciate that the benefits of TAA are only realized over investment time horizons long enough to include a market downturn. It is during this downturn that the TAA solu-tion more than makes up for its frequent underperformance during ―up markets.‖

Exhibit 1’s conclusion is that a simplistic TAA model for utilities results in a significant reduction in maximum drawdown, month-to-month volatility, and likelihood of loss. This risk mitigation benefit comes at the cost of having to adopt a long enough investment time horizon (one that includes a down-market) for the benefits of risk mitigation to overcome the performance drag during up-markets.

Findings-Across asset classes and geographies

Let’s next progress to international stocks by examining a representative example. Exhibit 2 provides the results for a non-U.S. stock composite consisting of equal weights Australia, Finland, Sweden, Germany, England, and France. These specific countries were chosen because quality monthly total return data exists back to 1/31/1920. Here as before, the benchmark was chosen with weights between the asset category (non-U.S. stock composite) and 90-Day U.S. Treasury Bills so as to generate the exact same 97-year return as with the TAA model. In this case, a 7.39% per annum real return was generated.The TAA model allocates 100% to non-U.S. stocks if the current level of the index is more

TAA Fixed-Weight, Buy&Hold (108% S&P 500 Utilities Total

Return Index 55 and -8% 90-Day T-Bills)

Return (in %) 6.77 6.77

Standard Deviation (in %) 13.6 23.6

Maximum Drawdown (in %) -47.3 -83.6

-17% 4 15

-15% 4 17

-13% 5 18

-11% 7 20

-17% 0 11

-15% 0 11

-13% 0 12

-11% 0 14

Correlation (TAA to Buy&Hold) 0.63 na

Tracking Error (in %) 15.8 na

Longest number of consecutive months when TAA underperformed Buy&Hold, ever experienced 12 na

1-month 58 na

5-months 69 na

6-months 71 na

7-months 72 na

8-months 72 na

9-months 74 na

10-months 73 na

11-months 74 na

12-months 74 na

13-months 73 na

14-months 72 na

15-months 72 na

Exhibit 1

Percentage of all time windows, X-months long, during which TAA underperformed Buy&Hold

S&P 500 Utilities

Return

Risk

During the first half of the period (before 3/31/1969)

Percentage of all possible 12-month windows when return was less than:

During the second half of the period (after 3/31/1969)

(5)

than 3.045% above its 11-month moving average, otherwise it reverts to T-Bills.

As before, the reduction in risk, whether characterized by maximum drawdown, month-to-month volatility, or probability of loss, was remarkably reduced, both during the first half and during the second half. Similarly, as before, these benefits come at the cost of poor tracking and a certain longevity and frequency of underperformance for time periods that fail to include a market downturn.

Moving on to bonds and focusing on the domestic market, we examine 30-Year U.S. Treasury Bonds in Exhibit 3. With this asset category, the requisite performance benchmark required a 130.4% allocation to 30-Year Treasuries and a -30.4% weighting to 90-Day T-Bills. Both the TAA model and its benchmark earned a 3.43% annualized real return. The TAA model allocates 100% to 30-Year Treasuries if the current level of the index is more than 0.54% above its 9-month moving average, otherwise it reverts to T-Bills.

TAA Fixed-Weight, Buy&Hold (109.1% Non-US Stocks and -9.1%

90-Day T-Bills)

Return (in %) 7.39 7.39

-17% 4 9

-15% 5 10

-13% 6 11

-11% 6 14

-17% 1 9

-15% 1 12

-13% 2 14

-11% 3 17

1-month 58 na

5-months 68 na

6-months 69 na

7-months 69 na

8-months 69 na

9-months 69 na

10-months 68 na

11-months 67 na

12-months 67 na

13-months 66 na

14-months 66 na

15-months 64 na

Exhibit 2

Non U.S. Stocks Composite

Return

Risk

(6)

Here as before, the TAA model delivered significant risk mitigation across all three measures and within both time periods. However, the maximum drawdown during the period prior to 3/31/1969 was reduced to only -43.6% from the passive benchmark’s -58.0%. This observation provides a telling comment on TAA models over the long span of time. Although they have successfully reduced both maximum drawdown and monthly volatility, they have in no way elimi-nated drawdown. Such an objective (drawdown elimination) is far too much to expect, and TAA shows little to no propensity to contribute to such a lofty goal.

Moving on to international bonds, we examine a composite benchmark consisting of equal weights Sweden, France, Australia, Italy, and Canada. In each case, we selected 10-year government bond indices to represent these countries. Counties were selected because quality return data was available back to 1920. Exhibit 4 provides the summary statistics.The TAA model allocates 100% to non-U.S. government bonds if the current level of the index is more than 3.56% above its 10-month moving average, otherwise it reverts to T-Bills.

TAA Fixed-Weight, Buy&Hold (130.4% USA 30-year Government

Bond Return Index and -30.4% 90-Day T-Bills)

Return (in %) 3.43 3.43

-17% 0 1

-15% 1 3

-13% 2 6

-11% 2 9

-17% 0 9

-15% 1 12

-13% 1 15

-11% 2 18

1-month 55 na

5-months 62 na

6-months 61 na

7-months 63 na

8-months 63 na

9-months 62 na

10-months 62 na

11-months 61 na

12-months 60 na

13-months 60 na

14-months 59 na

15-months 58 na

Exhibit 3

U.S. 30-Year Treasury Bond

Return

Risk

(7)

Now for the first time, the performance benchmark required a positive weighting for 90-Day T-Bills. In this case, 94.5% was allocated to non-U.S. government bonds and 5.5% to U.S. T-Bills. Across the 31 asset categories examined herein, 68% required a positive allocation to T-Bills for the TAA model’s benchmark, with a median cash weighting of 7.40%. In other words, TAA models should properly be viewed as ―balanced accounts‖ that generally deliver a long-run return lower than that of the asset category within which they invest. Here as before, the risk mitigation prop-erties are pleasing. Note how the maximum drawdowns during the first and second periods decrease from -71.1% down to -36.5% and -74.7% down to -31.0%, respectively.

We complete our examination of traditional asset categories by examining the S&P GSCI Commodities Index in Exhibit 5. This asset category presented a special case that is worth noting. The commodity TAA model delivered an annual real return of 4.10% over the last 97 years. However, no passive buy & hold mixture of commodities and 90-Day Treasuries exists that would have generated an identical return. Neither the returns on commodities nor cash are high enough. Further levering cash (back into commodities) does not work due to the detrimental effects of commodity vola-tility.

Across the 31 asset categories examined herein, six provided circumstances where no fixed-weight combination of T-Bills and the applicable asset category could deliver the same real return as that provided by the applicable TAA model. As mentioned above for commodities, the return was too low and the volatility too high for solution. The other five cases were the stocks of Finland, Germany, and France and the 10-year government bonds for France and Italy. In each of these five cases, the detrimental effects of WWII on these countries’ stock and bond markets, drove returns so

TAA Fixed-Weight, Buy&Hold (94.5% Non-US Bonds and 5.5%

90-Day T-Bills)

Return (in %) 2.64 2.64

-17% 0 5

-15% 1 5

-13% 1 8

-11% 2 10

-17% 0 2

-15% 0 3

-13% 0 4

-11% 1 7

1-month 51 na

5-months 51 na

6-months 53 na

7-months 53 na

8-months 54 na

9-months 55 na

10-months 55 na

11-months 56 na

12-months 57 na

13-months 57 na

14-months 58 na

15-months 58 na

Exhibit 4

Non U.S. Bonds Composite

Return

Risk

(8)

low that, once again, no passive fixed-weight combination of the applicable asset category and T-Bills could match the applicable TAA model’s return.

As with the prior examinations of domestic and international stocks and bonds, commodities also benefit from a significant reduction in risk via a TAA approach. This benefit arose in both the earlier and the later time periods, with maximum drawdowns shrinking from -43.3% down to -25.8% and from -69.3% down to -34.0%, respectively.

These attractive features of TAA, hold across a wide cross section of asset types and geographies. Exhibit 6 pro-vides summary statistics across all 31 asset categories and the median and mean results are shown at the bottom of the table.

TAA Fixed-Weight, Buy&Hold (70% S&P GSCI Total Return Index

and 30% 90-Day T-Bills)

Return (in %) 4.10 2.11

-17% 0 3

-15% 0 6

-13% 2 7

-11% 3 9

-17% 3 9

-15% 4 11

-13% 6 12

-11% 8 15

1-month 46 na

5-months 46 na

6-months 46 na

7-months 46 na

8-months 47 na

9-months 46 na

10-months 47 na

11-months 49 na

12-months 48 na

13-months 49 na

14-months 48 na

15-months 49 na

Exhibit 5

S&P GSCI Commodities

Return

Risk

(9)

Asset categories are ordered within the table so that those with the greatest risk mitigation resulting from TAA ap-pear at the top and those with the least apap-pear at the bottom. Note that the top eight categories are all stocks, both do-mestic and international. Similarly, the four at the bottom of the table are all dodo-mestic and international bonds.

The median results are worth noting. The typical TAA model relies on an eleven-month moving average and its associated passive buy & hold performance benchmark allocates 7.4% to T-Bills. The absolute reduction in the likeli-hood of losing more than -11% (in any 12-month window) was 6.5% prior to 3/31/1969 and 7.4% after 3/31/1969. Sim-ilarly, the absolute reduction in the probability of losing more than -13% was 5.5% prior to 3/31/1969 and 6.3% after 3/31/1969. And as constructed, this is without any reduction in return over the aggregate time period.

If markets don’t trend

This article contends that the causality driving TAA’s significant reduction in maximum drawdown and month-to-month volatility is the trending behavior of stocks, bonds, commodities, and currencies, both domestic and international. To examine this contention, we examined seven asset categories, but only after first removing their trend-ing characteristics.

To construct the seven de-trended asset categories, we determined the geometric mean real return and the standard deviation for a series over the entire 97 years. Next, we accessed an independent and identically distributed (iid) distri-bution (a trendless distridistri-bution) and adjusted its mean and standard deviation so that it generated the exact same return and volatility as the original (trending series).

Exhibit 7provides the results for the Synthetic iid S&P 500 Utilities asset category. The results are interesting. The application of a TAA strategy to this trendless time series meaningfully decreased return and increased risk (both max-imum drawdown and volatility).

Asset class Number of months trending calculated over Allocation to 90-day U.S. T-bills (in %) Return improvement (in bps) Absolute reduction in standard deviation during

1st half (in %)

Absolute reduction in standard deviation during

2nd half (in %)

Absolute reduction in frequency of 12-month windows losing more than -11% during 1st half (in %)

Absolute reduction in frequency of 12-month windows losing more than -11% during 2nd half (in %)

U.S. Stocks Composite 11 -5.76 0 9.5 6.2 10.0 11.6 11.4 11.8

S&P 500 Utilities 17 -8.04 0 10.0 6.6 13.0 13.2 12.9 12.3

Dow Jones Industrials Average 12 7.40 0 6.1 4.6 10.7 11.1 9.7 9.7

S&P 500 Transportation 10 8.34 0 10.5 6.1 12.3 7.7 10.6 7.2

Australia ASX Accumulation-All Ordinaries 8 0.46 0 5.0 8.1 10.4 8.4 10.6 10.5 OMX Helsinki All-Share Gross (Finland) 14 20.00 396 3.7 3.3 7.6 16.7 8.3 15.5 OMX Stockholm Benchmark Capped Gross (Sweden) 10 2.79 0 6.0 7.7 10.9 14.9 9.3 14.2

Germany CDAX 10 35.50 966 65.6 0.1 12.3 8.8 10.2 8.6

Sweden Government Bond 8 -66.11 0 7.8 11.4 4.8 23.0 4.8 20.9

Non-U.S. Stocks Composite 11 -9.09 0 18.3 8.1 7.4 14.1 5.5 12.5

France 10-year Government Bond 10 25.00 325 6.6 0.0 17.8 6.9 15.8 5.3

UK FTSE All-Share 10 16.54 0 3.2 5.3 6.5 7.4 6.0 7.6

U.S. 30-year Government Bond 9 -30.38 0 3.0 6.7 6.9 15.6 4.0 14.6

S&P 500 10 18.82 0 5.4 3.5 7.2 9.7 6.3 9.1

France CAC All-Tradable 10 32.00 116 4.9 1.1 4.0 7.9 5.5 9.5

Australia 10-year Government Bond 10 3.39 0 4.9 3.6 4.4 9.1 3.5 8.4 Italy 10-year Government Bond 13 37.00 266 5.1 0.0 9.3 4.7 9.0 2.6 Non-U.S. Government Bonds Composite 10 5.50 0 3.1 3.2 8.6 6.7 6.3 4.0

S&P GSCI Commodities 11 30.00 199 0.7 0.0 5.8 7.4 5.6 6.3

Global Bond Composite 17 -29.92 0 2.2 3.5 4.8 6.7 4.6 5.6

U.S. 60/40 Stk/Bnd Composite 8 11.13 0 4.0 2.4 5.3 5.3 4.9 4.2

Global 60/40 Stk/Bnd Composite 12 7.80 0 3.8 2.2 4.2 4.9 3.0 3.9

U.S. 60/35/5 Stk/Bnd/Commodities Composite 12 17.03 0 3.2 2.3 6.7 4.9 5.6 3.2

Dow Jones Corporate Bond 18 -11.56 0 1.7 2.4 1.8 5.3 0.9 4.4

Global Stocks Composite 10 26.08 0 4.3 1.8 4.4 5.4 3.3 4.4

U.S. Bonds Composite 14 -4.42 0 0.5 1.8 0.7 4.2 0.0 2.8

Global 60/35/5 Stk/Bnd/Commodities Composite 12 13.82 0 3.3 1.8 3.9 3.7 2.6 3.7

U.S. 10-year Government Bond 16 3.52 0 0.7 1.5 0.4 3.2 0.0 1.8

U.S. 5-year Government Note 18 -9.51 0 0.5 2.1 1.1 1.8 0.2 0.9

U.S. Total Return AAA Corporate Bond Index 18 -10.15 0 0.4 2.0 0.0 4.7 -0.2 3.0 Canada 10-year Government Bond 10 16.24 0 0.3 1.1 0.9 3.0 -0.2 1.6

Median 11 7.40 0 4.0 2.4 6.5 7.4 5.5 6.3

Mean 12 4.95 73 6.6 3.6 6.6 8.3 5.8 7.4

(10)

As a consequence, the performance benchmark was forced to adopt a high cash allocation, i.e., 46.4% Synthetic iid S&P 500 Utilities and 53.6% T-Bills. The TAA model’s month-to-month volatility (as measured by standard deviation) increased from 8.7% to 12.4% and 8.8% to 13.8%, during the two periods, respectively. Maximum drawdown was sim-ilarly worsened, increasing from -38.0% to -55.6% and from -27.5% to -57.3%, respectively.These results strongly support the contention that a market’s trending characteristics are what establishes the potential for value-added TAA.

By examining additional asset categories, using this same approach, first de-trending and then applying a TAA framework, we can provide additional evidence of this contention. Exhibit 8 provides the results for the de-trended 30-Year U.S. Treasury Bond.

TAA Fixed-Weight, Buy&Hold (46.4% synthetic iid utilities and

53.6% 90-Day T-Bills)

Return (in %) 4.08 4.08

-17% 3 1

-15% 6 2

-13% 8 2

-11% 11 4

-17% 1 0

-15% 2 1

-13% 3 1

-11% 8 3

1-month 49 na

5-months 52 na

6-months 53 na

7-months 54 na

8-months 54 na

9-months 55 na

10-months 54 na

11-months 54 na

12-months 55 na

13-months 55 na

14-months 56 na

15-months 56 na

Exhibit 7

Synthetic iid S&P 500 Utilities

Return

Risk

(11)

With this iid distribution (trendless) return stream, TAA severely damaged returns. And to such an extent, that the fixed-weight passive benchmark was 44.4% Synthetic iid U.S. 30-Year Treasury Bonds and 55.6% T-Bills. The deterio-ration to the maximum drawdown was particularly stark, having fallen from -24.7% to -49.7% in the first period and from -17.6% to -32.7% in the second period. Moreover, monthly volatility almost doubled, climbing from 4.7% to 7.4% in the first, and from 4.6% to 7.8% in the second.

We conclude our analysis of trendless asset categories with a third, i.e., global commodities. As shown in Exhibit 9, TAA modeling so severely damaged returns, that the required passive performance benchmark required a 96.8% alloca-tion to T-Bills. Moreover, all of the risk metrics were severely impacted, i.e., maximum drawdown, monthly volatility, and probability of loss.

TAA Fixed-Weight, Buy&Hold (44.4% synthetic iid 30 year

treasury and 55.6% 90-Day T-Bills)

Return (in %) 1.83 1.83

-17% 0 0

-15% 1 0

-13% 2 0

-11% 4 0

-17% 0 0

-15% 0 0

-13% 0 0

-11% 1 0

1-month 49 na

5-months 50 na

6-months 50 na

7-months 51 na

8-months 50 na

9-months 51 na

10-months 51 na

11-months 50 na

12-months 52 na

13-months 51 na

14-months 52 na

15-months 51 na

Exhibit 8

Synthetic iid U.S. 30-Year Treasury Bonds

Return

Risk

(12)

We summarize our analysis of the behavior of TAA strategies when applied to trendless markets (iid return distri-butions) in Exhibit 10. Here we report the summary statistics for seven distinct asset categories, each constructed as a synthetic iid return stream. This analysis spans stocks, bonds, and commodities, both domestic and international. Of special note are the median synthetic results. For the median case, the moving average used ten months, the appropriate passive performance benchmark allocated 55.64% to 90-Day T-Bills, standard deviations increased by 3.7% and 3.9% in the first and second periods, respectively, and the probability of loss greater than -11% (over a random 12-month window) increased by 5.3% and 4.2%, respectively.

But perhaps more telling, all seven de-trended asset categories realized severe degradation by the application of a TAA strategy. This is seen most clearly through the median allocation to T-Bills of 56%. We conclude that TAA’s potent risk mitigation properties are the direct result of the trending attributes across virtually all market segments and

geog-TAA Fixed-Weight, Buy&Hold (3.2% synthetic commodities and

96.8% 90-Day T-Bills)

Return (in %) 0.90 0.90

-17% 2 0

-15% 4 0

-13% 9 1

-11% 13 2

-17% 1 0

-15% 2 0

-13% 2 0

-11% 7 0

1-month 51 na

5-months 51 na

6-months 52 na

7-months 52 na

8-months 52 na

9-months 53 na

10-months 53 na

11-months 54 na

12-months 54 na

13-months 55 na

14-months 55 na

15-months 55 na

Exhibit 9

Synthetic iid GSCI Commodities

Return

Risk

Tracking Asset class Number of months trending calculated over Allocation to 90-day U.S. T-bills (in %) Return improvement (in bps) Absolute reduction in standard deviation during

1st half (in %)

Absolute reduction in standard deviation during

2nd half (in %)

Absolute reduction in frequency of 12-month windows losing more than -13% during 2nd half (in %) Synthetic 30-Year U.S. Treasury 9 55.64 0 -2.7 -3.2 -3.9 -1.1 -2.1 -0.2 Synthetic 10-Year Australian Government Bond 10 50.25 0 -2.6 -2.0 -1.2 -0.2 -0.7 0.4 Synthetic Dow Jones U.S. Investment Grade Corporate Bonds 18 35.30 0 -1.2 -1.5 -0.2 0.0 0.0 0.0 Synthetic S&P 500 Utilities 17 53.57 0 -3.7 -5.0 -7.2 -4.7 -5.8 -1.6

Synthetic S&P 500 10 56.26 0 -4.5 -3.9 -1.9 -12.5 -0.7 -6.5

Synthetic U.K. Stocks 10 72.72 0 -7.4 -7.3 -10.9 -3.7 -8.3 -1.9

Synthetic GSCI Commodities 11 96.83 0 -8.5 -9.5 -11.4 -7.0 -7.2 -2.3

Median 10 55.64 0 -3.7 -3.9 -3.9 -3.7 -2.1 -1.6

Mean 12 60.08 0 -4.4 -4.6 -5.3 -4.2 -3.5 -1.7

(13)

raphies. TAA Tradeoffs

As shown previously, TAA powerfully mitigates both tail risk and month-to-month volatility. These two benefits span markets, geographies, and time periods. Nevertheless, this benefit comes at a cost. This cost has three key attrib-utes: tracking, frequency, and longevity. We touch on each in turn.

Tracking-Whether focusing on correlation or the standard deviation of tracking error, TAA solutions deviate widely from passive fixed-weight benchmarks. This observation is well demonstrated in Exhibits 1-5 where the correlations ranged from 0.78 to 0.47 and where the standard deviations of the tracking errors ranged from 21.2% to 6.8%. These numbers demonstrate the extent to which TAA solutions deliver return patterns that have little to do with their underly-ing exposures.

Frequency-The frequency, month-to-month, with which TAA strategies underperform their passive, fixed-weight brethren can be difficult to swallow for the unprepared. Once again, Exhibits 1-5 demonstrate this best. For one-month periods, the frequency of underperformance across these five asset categories ranged from a low of 48% to a high of 58%. In a similar fashion, for 15-month long time windows, the frequency of underperformance ranged between 49% and 72%.

Longevity-Finally, the longevity with which TAA solutions can consecutively underperform passive, fixed-weight buy & hold, one month after another, is well demonstrated by Exhibits 1-5. The maximum number of months when this happened (over the last 97 years) ranged between ten and twenty months.

Policy implications

Policy implications are relatively clear cut. Each is the direct result of the behavioral attributes of TAA strategies. There are five policy implications.

First-TAA is not a return enhancement strategy. This is best demonstrated by examining the mean and median cash levels within the comparative passive performance benchmarks appearing at the bottom of Exhibit 6. In a clear majority of cases, TAA reduced the level of return below that of the underlying asset category over the last 97 years. Unfortu-nately, this is one of the most common sources of miscommunication and misselling within the advisory community. Setting client expectations that TAA is expected to enhance returns instead of diminish returns, serves only to misguide, and in the fullness of time generates disenchantment.

Second-The contribution or benefit of TAA is risk mitigation. This benefit has three forms: reduction in maxi-mum drawdown, reduction in monthly volatility, and reduction in the probability or likelihood of loss. As before, this is once again an example of how TAA is often missold. At times, it is suggested that TAA has the ability to eliminate or largely eliminate drawdown. Such a result is neither practical nor possible. Moreover, although all three forms of risk mitigation are tremendously valuable, many clients seldom come to appreciate such risk mitigation until they experi-ence a bear market decline. Exhibits 11 and 12 provide the summarized results for the proportionate reductions in maximum drawdowns and monthly volatility. This pair of exhibits portrays the median results for the five major asset categories and for balanced and/or global solutions. For each of these distinct asset categories and for both time peri-ods, TAA modeling delivered significant risk reduction along two different risk metrics.

(14)

Third-TAA delivers low correlation and high tracking error. This is once again a fundamental source of miscom-munication between advisers and clients. Often they inadequately communicate the extent to which TAA will march to a different drummer, deviating significantly from any fixed-weight passive index. Failure to adequately explain this issue in advance, and only compare it to an appropriate benchmark, is an endless source of trouble and the inevitable

0.0 7.1 14.3 21.4 28.6 35.7 42.9 50.0

6.0 11.4 16.9 22.3 27.7 33.1 38.6 44.0

Proportionate reduction in annualized standard deviation (in %)

P

ro

po

rt

io

na

te

reduc

tio

n

in

ma

xi

m

um

drawdo

wn

(in

%)

U.S. stocks

Non-U.S. stocks Non-U.S. bonds Commodities

U.S. bonds

Balanced

First Half - Before 3/31/1969

Exhibit 11

40.0 43.1 46.3 49.4 52.6 55.7 58.9 62.0

0.0 4.9 9.7 14.6 19.4 24.3 29.1 34.0

Proportionate reduction in annualized standard deviation (in %)

P

ro

po

rt

io

na

te

reduc

tio

n

in

ma

xi

m

um

drawdo

wn

(in

%)

U.S. stocks Non-U.S. stocks Non-U.S. bonds

Commodities

U.S. bonds

Balanced

Second Half - After 3/31/1969

Exhibit 12

(15)

dissatisfaction of clients using TAA. The solution is twofold. First, compare against the appropriate benchmark, one comprised not of fixed-weight passive indices, but instead of a peer universe of TAA managers. Second, continually evaluate whether the TAA strategy in question is delivering on its objective, i.e., reduction in maximum drawdown, monthly volatility, and probability of loss over 12-month windows.

Fourth-TAA frequently underperforms passive buy & hold performance benchmarks for periods that fail to include a market downturn. This frequency problem is well reported in Exhibits 1-5. Once again, this is a fourth source of mis-communications between advisers and their clients and the failure to set appropriate expectations. This relative under-performance does not detract from TAA’s inherent attraction. Instead, it identifies oneof TAA’s inherent attributes, i.e., that TAA requires the inclusion of a market downturn for it to prove out its worth. The solution, as before, is to utilize correct benchmarking so as not to mislead the client.

Fifth-TAA will at times underperform passive fixed-weight benchmarks one month after the next in a consistent and consecutive fashion. This is the longevity problem. The parameterization of this challenge is again well identified in Exhibits 1-5. This attribute does not undermine the usefulness and attractions of TAA, instead it simply identifies one of the tradeoffs associated with the strategy. For the adviser and client, it requires correct performance benchmarking so that erroneous conclusions are avoided. Specifically, TAA strategies should be compared against peer universes of other TAA managers.

5. Conclusions

TAA is all about trend following. TAA does not enhance returns, instead it lowers returns. But when compared to appropriate balanced performance benchmarks (mixtures of the underlying asset category and cash equivalents), TAA adds surprisingly large risk-mitigation benefits. These risk mitigation benefits are of three types. First, a reduction in maximum drawdown. Second, a reduction in monthly volatility. Third, a reduction in the probability of significant loss during a given year.

These risk reduction benefits span stocks, bonds, and commodities, both domestic and international. They also span both pure exposures and blended mixes. Of equal importance, these risk mitigation benefits also span different time periods, both recent and past. These conclusions are further supported by examining trendless markets (true ran-dom walks). For such iid markets, the benefits of TAA disappear.

But as in life, no benefit comes without its associated tradeoffs. TAA is no different. These tradeoffs take three forms, i.e., tracking, frequency, and longevity. Users of TAA must compare their realized performance against appropri-ate performance benchmarks to keep from misunderstanding or being misled by these three ―costs.‖ This requires that TAA users compare their performance against a peer universe of other TAA managers. It also requires that they make determine if they have received the benefits that TAA is designed to deliver, i.e., reductions in maximum drawdown, monthly volatility, and likelihood of loss.

Finally, given the fundamental benefits of TAA and its associated tradeoffs, users of TAA should adopt the appro-priate investment time horizon. Such horizons are not measured in a year or two. One highly attractive approach for the use and application of TAA is the so-called time-segmentation or bucket-approach to investing. Under such approaches, TAA should be used exclusively within the 10-to-15 year and the 15-to-20 year buckets. By doing so, the investor adopts the required time horizon for TAA to show its worth, i.e., mitigate risks during a market downturn. Horizons of between zero and nine years are inappropriate for TAA.

References

1. Amenc, N., F. Goltz, and A. Grigoriu. ―Risk Control Through Dynamic Core-Satellite Portfolios of ETFs: Applica-tions to Absolute Return Funds and Tactical Asset Allocation.‖ The Journal of Alternative Investments, (Fall 2010). 2. Arnott, R. and T. Miller. ―Surprise! TAA Can Work in Quiet Markets.‖ Journal of Investing, Vol. 6, No. 3 (1997),

pp. 33-45.

3. Asness, C. ―Global Tactical Asset Allocation.‖ Goldman Sachs Pension and Endowment Forum, (September 1996). 4. Blitz, D. and P. Van Vliet. ―Global Tactical Cross-Asset Allocation: Applying Value and Momentum Across Asset

Classes.‖ The Journal of Portfolio Management, (Fall 2008).

(16)

6. de Freitas, E., C. Barker, and E. Hehn, ed., ―ETFs—Tactical Asset Allocation Tools.‖ In Exchange Traded Funds: Structure, Regulation and Application of a New Fund Class. Berlin: Springer, 2006.

7. de Silva, H. ―Modern Tactical Asset Allocation.‖ CFA Institute Conference Proceedings Quarterly, (December 2006), pp. 1-10.

8. Faff, R., D. Gallagher, and E. Wu. ―Tactical Asset Allocation: Australian Evidence.‖ Sage Journals, (December 1, 2005).

9. Grossman, S., and Z. Zhou. ―Optimal Investment Strategies for Controlling Drawdowns.‖ Mathematical Finance, Vol. 3, No. 3 (1993), pp. 241-276.

10. Jensen, G., R. Mercer, and J. Johnson. ―Tactical Asset Allocation and Commodity Futures.‖ Journal of Portfolio Management, Vol. 25, No. 4 (2002), pp. 100–111.

11. Lee, W. ―Risk and Return Characteristics of TAA Under Imperfect Information.‖ Journal of Portfolio Management, Vol. 25, No. 1 (1998), pp. 61-70.

12. Theory and Methodology of Tactical Asset Allocation, New York, N.Y.: John Wiley and Sons, 2000.

13. Lewis, N., J. Okunev, and D. White. ―Using a Value at Risk Approach to Enhance Tactical Asset Allocation.‖ Journal of Investing, (Winter 2007).

14. Lundin, M. ―On the information ratio of tactical asset allocation.‖ Journal of Asset Management, Vol. 4, No. 5 (February 2004).

15. MacBeth, J. ―Tactical Asset Allocation: Pros and Cons.‖ Financial Analysts Journal, Vol. 49, No. 6 (1993), pp. 30-43.

16. Mattu, R., V. Naik, M. Devarajan, and M. Sharif. ―Tactical Asset Allocation With Macro Views.‖ PIMCO Quanti-tative Research, (October 2012), pp. 1-11.

17. Philips, T., G. Rogers, and R. Capaldi. ―Tactical Asset Allocation: 1997-1994.‖ Journal of Portfolio Management, Vol. 23, No. 1 (1996), pp. 57-64.

18. Sapp, T., and A. Tiwari. ―Does Stock Return Momentum Explain the Smart Money Effect?‖ Journal of Finance, Vol. 7, No. 6 (2004), pp. 2605-2622.

19. Solow, K., M. Kitces, and S. Locatelli. ―Improving Risk-Adjusted Returns Using Market-Valuation-Based Tactical Asset Allocation Strategies.‖ FPA Financial Planning Association, (2012).

20. Tokat, Y. ―A Primer on Tactical Asset Allocation Strategy Evaluation.‖ Vanguard Investment Counseling & Re-search, (2006), pp. 1-12.

21. Weigel, E. ―The Performance of Tactical Asset Allocation.‖ Financial Analysts Journal, Vol. 47, No. 5 (1991), pp. 63-70.