April 29, 2013
Asset Allocation & Rebalancing Review | 29 April 2013
Divergence. That's a good description for the major asset classes so far this year. The spread between the best performers and the laggards is relatively wide on a year-to-date basis through April 26, offering a mix of opportunity and risk. Par for the course, perhaps, but the divergence is quite striking nonetheless.
Consider how the numbers stack up on an equal-weighted basis. Imagine that on December 31, 2012 we bought everything, weighted the portfolio equally, and let Mr. Market run. The resulting asset allocation as of last Friday:
US stocks and REITs are leading the horse race for the moment, closely followed by foreign equities in developed markets. Meanwhile, commodities, emerging-market stocks, and government bonds in developed markets are the weakest members of the major asset classes on a year-to-date basis so far in 2013, based on our list of proxy ETFs.
For another view, here's how the year-to-date total returns compare, sorted by performance in descending order:
Graphing the relative returns for the year so far offers a more striking comparison, as the chart below shows. In this case, each of the proxy ETFs listed above is indexed to an initial value of 100 at last year's close and left to wander at Mr. Market's discretion.
The wide array of results may appear dramatic, but it shouldn't be too surprising. Diversifying across asset classes, after all, is a strategy that seeks to exploit the expected set of varied return and volatility correlations. By that standard, the year so far offers a rich opportunity set. The relatively wide spectrum of results at the moment implies that rebalancing's abilities for juicing performance, reducing risk, or both are intact.
As usual, however, there is doubt about whether the past will repeat and so the divergence, for all its implied opportunity, isn't easily accepted at face value. Maybe it's different this time. The impressive historical results that accompany a simple rebalancing strategy for the major asset classes always looks suspect in real time. As a pair of analysts (John Kiskiras and Andrea Nardon) remind in a recent study--"Portfolio Rebalancing: A Stable Source of Alpha?"--the "essential condition" for success with rebalancing is mean reversion. Robust volatility and correlation help, of course, and sometimes by a lot, but these are mostly "amplifiers" for mean reversion.
It's never really clear if mean reversion will prevail for any two asset class pair, which is one of the reasons for holding a broad set of assets. In a world of limited ex ante insight, diversification is an obvious risk-management tool. But in an asset allocation context, rebalancing is no less critical. Asset allocation and rebalancing together are a powerful combination, much more so than if one is used in isolation with the other. There's always doubt about how, or even if, history will repeat, however. That's a valid concern, and it's probably one reason why investors generally have such a hard time exploiting this strategic pair in something close to optimal conditions. But that's also why the rebalancing bonus has proven to be so impressive and persistent on an historical basis: the crowd tends to leave quite a lot of the associated opportunity on the table. Why? Let's just say that it's hard to be a contrarian.
That brings us back to the current divergence in asset class performance so far this year, which again raises the perennial question: Is it different this time? No one really knows, of course. That's why there's a risk premium associated with asset allocation and rebalancing in the first place. Just as Rick mistakenly went to Casablance for the waters, if you thought rebalancing and asset allocation provide a free lunch you too have been misinformed. Ah, but you'll always have Paris.
Posted by jp at April 29, 2013 4:50 AM
Great question. The short answer is that an asset class is a group of assets that share common characteristics. Yes, the definitions are inevitably warm and fuzzy, but the broad categories are beyond debate: stocks, bonds, real estate (REITs) and commodities. Each of these can and often should be broken down into more granular sets, in part because narrower definitions are valuable for rebalancing purposes.
You are right, of course, that the set of definitions one chooses will influence the asset allocation and the resulting risk and return profile. This is especially true for an equal weighted portfolio.
On that note, I'm trying to be agnostic by tracking all the broad asset classes that are available for purchase via ETFs. Beyond that basic approach there's a fair amount of subjectivity. Breaking up, say, equities into more pieces raises the overall weight in an equal weighted portfolio, for instance. But since equal weighting is effectively a model-free strategy, it's not clear what would constitute an objective set of asset class definitions. Should stocks be listed as one input or 3, or even 10? And how should we weight them? The same question applies to all the asset classes.
One solution is to use a model, such as the CAPM, to choose weights. But that comes with baggage too.
In any case, the first order of business is monitoring a broad set of all the investable asset classes. Things get tricky real quick after that.
Posted by: Anonymous at May 1, 2013 2:16 PM
Thanks for the great work on the blog. How do you define an asset class? Also, when equal weighting, don't you indirectly introduce a bond bias since there are 8 ETFs out of 14? Just trying to understand your methodology.
Posted by: NCAdv at April 30, 2013 10:09 AM
I guess you could say that we're always rebalancing into different asset classes, albeit in varying degrees, depending on the time frame. To the extent you have confidence in distinguishing why they're different, and what that implies for forecasting risk and return, you adjust your asset allocation and rebalancing plans accordingly. What's old is new, but no less challenging.
Posted by: JP at April 30, 2013 6:55 AM
Isn't the US Bond Market Asset Class (i.e. BND) becoming skewed by the Fed purchases, so that it no longer represents the "Bond Market." Wouldn't the bond market index be composed differently if the Government had been buying all the components of BND, rather than just US Treasuries it now buys. We would seem to be rebalancing into a different asset class - a US treasury class rather than a bond market class.
Posted by: W J McGurn at April 29, 2013 11:32 AM