Another look at Tactical Asset Allocation
Warning: Modest geekery in this post - but some very valuable nuggets of portfolio management wisdom too. Don't read if you're already feeling a bit drowsy. This is a long one. I've written a bit about tactical asset allocation (TAA for short) in past posts and thought it would be a nice time to revisit it since I've spent a lot of time this past year with creating smart ways to use it in getting better returns with lower risk as an investor.
TAA is basically a way to actively manage an asset allocation in effort to increase returns, reduce risk, or both. The antonym (opposite) of TAA is static asset allocation. Static asset allocation is the most commonly used way to manage a portfolio whereas an investor puts a specific amount of their portfolio into various asset classes and rebalances to that static allocation periodically (monthly, quarterly, or annually for example).
A moderate/balanced investor might use a static asset allocation like this:
25% Government bonds 20% Corporate bonds 15% Large company US stocks 10% Mid-sized company US stocks 10% Small-sized company US stocks 5% Developed country international stocks 5% Emerging country international stocks 5% Real estate 5% Cash or equivalents
Every rebalancing period the investor (or their advisor/money manager) would rebalance back to these percentages if one area outperformed/underperformed the others causing the portfolio to become out of balance.
The positive to the approach is it's simple, requires little to no work, can be low cost if done correctly, and since it trades infrequently it usually won't produce much in the way of short term capital gains (and thus the taxes that can go along with them).
The downside is if the markets perform poorly for an extended period of time a static portfolio is likely to produce losses, sometimes significant. In a year like 2008, for example, this type of mix would have produced a 20-30% decline in portfolio value and thus required a 30-40% increase just to get back to break even. It also forces investors to endure having money in a losing asset class (like real estate from 2007 to 2009) regardless of how obvious it seems one should not own it.
So back to TAA...
TAA seeks to strike a balance between being simple and efficient; but also manage the downside in bad markets and accentuate returns in up markets. If done correctly it can work great, but does have it's drawbacks to be sure.
Now is when the geekery really begins.
When I set out to test the merits of an investing strategy I think "backtesting" is a great way to see how the strategy would have performed in various market condition. Backtesting is when we apply a set of mechanical rules to investments to see what the strategy would have chosen, how much it would have chosen, and when it would have sold. Backtesting is not the same as live trading and does have limitations - namely whether or not an investor would have been able to actually stick to the strategy as it is built.
In backtesting TAA I think it's important to define our goals. In this study my goals were pretty simple:
Can we effectively reduce the risk of investing by moving out of asset classes showing undesirable characteristics?
Can we effectively increase returns by over-weighting the asset classes showing the most desirable characteristics?
Can we accomplish items (1) and (2) without excessive trading or costs?
Can we build a mechanical model that allows us to accomplish items (1), (2), and (3) to remove emotion from the decision making process?
I'll show the full study in a moment - but first want to say that I believe all 4 items above can be accomplished with TAA and the merits for use in portfolios are overwhelmingly strong.
In order to keep TAA simple and low cost I've determined it is best done with only a few asset classes. So whereas I used 9 asset classes in the example for a static allocation I think TAA can be done with just 5. To make this happen we just need to break down investments into their most simple form. In other words, instead of having 5 stock asset classes, why not just use 1?
My proxy for stocks was the Vanguard Total Stock Market Index Fund. I choose it because it automatically looks at all the equity asset classes I'd broke down in the static allocation example (large, small, US and abroad) but does so with just one very low cost index fund. I did the same for bonds by using the Vanguard Total Bond Market Index Fund for bonds, and the Vanguard REIT Index Fund for real estate. Cash is just cash, so that stayed the same.
The proxy I choose for the 5th asset class though was a bit different. I choose to use the State Street Gold ETF (exchange traded fund) for the 5th asset class as it offers a completely uncorrelated asset class to stocks, bonds, and real estate. And truthfully, I think most investors would agree that having some exposure to gold at times can prove to be a very useful asset allocation component.
To be able to prove that TAA can significantly outperform static allocation I also created a comparable benchmark that looked at gold so as not to create an apples to oranges comparison. I also wanted to see how a balanced investor might use TAA versus static asset allocation, so my exact proxy is as follows:
50% Bonds 16.5% Stocks 16.5% Real estate 16.5% Gold
I also created a proxy for this example that resembles more closely how many investors might invest and used the following ultra simple breakdown:
50% Bond 50% Stocks
If you're wondering why so simple...well, because it is. What we're trying to do with this study is look at the merits of one method of investing (buy and hold with occasional rebalancing) versus TAA.
For the TAA strategy I wanted to create a fair test, so it works in this very basic manner:
50% of the portfolio must always be in either bonds or cash 50% of the portfolio can "tactically" shift to varying degrees of stocks, bonds, real estate, gold, or cash
To be sure I'm totally clear on this - the TAA model can never get any more aggressive than the static allocation benchmark - but can move entirely to the cash when it feels market investments (or bonds for that matter) are too risky for investing. It can also use just stocks and not real estate, gold, or bonds for the other half; or any other mix of one or all 4 asset classes.
The rules for TAA in this test are very, very simple.
Check to see if the asset class' 50 day price average is greater than its 200 day price average (the Golden Cross)
If it passes item (1) add the asset classes 6 month return and 12 month return to measure momentum
If the sum of the 6 month return and 12 month return are greater than 0% it is included in the allocation
If nothing passes items (1), (2), and (3) the portfolio moves to cash
Repeat the process just once per month to update the TAA allocation for the next month (so never more than 12 allocation changes per year)
For clarity, here's how that might work.
Let's say all asset classes pass items (1), (2), and (3); and the returns are all the exact same. If that were the case the TAA model would have the exact same weightings as the static asset allocation model. But if stocks, for example, had a 6 month return + 12 month return that was twice as high as the other asset classes the model would suggest 50% in bonds, 25% in stocks, and the remaining 25% split equally between real estate, gold, and additional bonds.
If I'm losing you here - don't worry. If you really want to know all the metrics of the test I'll provide an excel spreadsheet with the nitty gritty details. The main thing to take away from this post are the results of the study and how it can help your portfolio.
The reason for keeping using this simple method is to make sure we are over-weighting the asset classes that pass a risk check, but also have the most positive momentum. In general, once an asset class establishes itself as a leader for 6 to 12 months it will remain a leader for multiple years. Having a simple way to gauge this strength (just adding the 6 and 12 month returns) accomplishes this quite beautifully.
To keep this post reasonably short (may have already failed in that department, sorry) I'll show the results of the stud for the past 6 years. I actually went back much, much farther - but most of us only really remember the most recent market conditions.
Without further ado - here's how our TAA strategy stacks up against our static benchmark visually:
The blue line is the hypothetical growth of TAA from 12/2005 to 11/2011, the red line is static asset allocation (as I described was built for proxy #1), and the green line is the 50% stocks/50% bonds (proxy #2).
To be sure I was being extra fair I also assumed the TAA portfolio had a whopping 2.5% annual fee per deducted from the return on a pro-rated monthly basis.
As you can see, TAA outperformed quite nicely.
For the number lovers out there here's how this all breaks down:
[table id=5 /]
I should also note that the volatility as measured by standard deviation was about 30% lower for TAA versus both benchmarks - which means it was able to increase returns and do so while simultaneously lowering risk. Not too bad.
Another very nice benefit of TAA is that it really rarely goes to cash. Instead it just tends to ebb and flow between being a balanced portfolio (stocks, bonds, real estate, and gold) to a mostly bond portfolio. The reason this is beneficial is that cash is paying virtually nothing in interest at the moment. While that may change down the road, it is nice to be able to earn a 6.73% yield while sitting out of the stock market - which is what the Vanguard Total Bond Market Index Fund is yielding right now.
I'm sharing all of this for 2 reasons:
I'm going to make TAA a monthly feature on my blog - similarly to my RPA (recession probability analytics). This means each month I'll update the returns of TAA versus the two benchmarks shared in this post as well as exactly what the holdings are each month.
In doing this research over the past year I've developed an even better version than what I'm publishing for free here on the blog for the use of my firms clients. So if you're already a client expect to see a healthy dose of TAA in your portfolios soon. The rules used for our managed accounts are more refined, tested more thoroughly, and deliver a bit more consistency than the blogs free version I'm sharing for research purposes too.
Finally, for those wondering what TAA would have suggested these past couple of turbulent months:
August - 53% bonds, 9% stocks, 13% real estate, 25% gold September - 54% bonds, 5% stocks, 7% real estate, 34% gold October - 62% bonds, 0% stocks, 0% real estate, 38% gold November - 60% bonds, 0% stocks, 0% real estate, 40% gold (this is what the model would be holding as of 11/1/2011 and for the balance of this month)
For a historical perspective TAA would not always be nearly as heavy in bonds and gold. That's just where the best risk adjusted returns are these days and the model is designed to see that and take advantage of it. In 2006 real estate would have been the largest growth asset and gold would have been quite small.
During the past few months though with its most recent allocations TAA is up 1.32% while the stock market is down nearly 5%. I'd say for now a balanced, yet tactical approach to asset allocation is doing exactly what it is supposed to do.
Expect more on this topic next month, and as always - if you know someone who might benefit from this read as well as keeping up with the strategy as a new monthly feature of my blog feel free to pass this along.