Rebalancing is a investing staple of the financial planning world. The execution of a rebalancing strategy helps to ensure that the client's asset allocation does not drift too far out of whack, as without such a process a portfolio holding multiple investments with different returns would eventually lead to a portfolio that increasingly favors the highest return investments due to compounding. Yet in practice, most financial planners often discuss rebalancing not only as a risk-reduction strategy (by ensuring that higher-return higher-volatility assets do not drift to excessive allocations), but also as a return-enhancing strategy. However, in reality, there is nothing inherent about rebalancing that would be anticipated to generate higher returns... unless you get the market timing right.
The inspiration for today's blog post comes from a comment made in response to a prior blog post discussing the market timing aspects of rebalancing strategies. In the commented response, a critic suggested that rebalancing should be viewed as the very opposite of market timing, because it is an a priori strategic planning process, rather than something that is responding to market movements as they occur.
Portfolio Rebalancing - Risk Management Or Return Enhancement?
In response, first of all, I will agree that there is an aspect of rebalancing that I would "purely" call risk management; it's about reallocating the portfolio back to the target allocation, essentially combating the portfolio allocation drift that would otherwise occur over time when you have multiple investments with different returns. However, as my examples from my prior rebalancing blog post show, there is nothing favorable to returns about doing that. In fact, rebalancing adversely affects returns, to the extent that the investor continuously sells the higher return asset to keep it from creeping to an ever-higher allocation.
Yet many, many advisors talk about rebalancing not just as a technique to manage risk and prevent portfolio drift, but as a return enhancer. As my examples illustrate again, though, the return-enhancing aspect of rebalancing is not natural to the process, per se; it only occurs if you get the timing favorably. So at the least, if we're going to insist on viewing rebalancing as purely a risk management function to prevent portfolio drift, that's fine, but it should never be referred to as a "return enhancing" strategy in that context. It would actually be a return-reducing, risk-management technique; not that that's bad, if your goal is to focus on the risk management aspects, but let's at least call a spade a spade.
So where did this idea come from that rebalancing is return-enhancing? As I've shown, the only return-enhancing aspect of the exercise is if you get the timing right on the rebalancing, relative to when assets are "zigging and zagging". As the comment to my prior blog post noted, common techniques of rebalancing include those that rebalance at some fixed time interval (e.g,. once a year), or some percentage deviation from target (e.g., when stocks shift 20% above their intended allocation)... yet how does one choose what time interval, or what percentage deviation? The answer that I see in the research? We try to choose a time interval or percentage deviation that is most likely to give us favorable market timing in the transaction. The percentage deviation research is focused on determining how likely assets are to deviate before mean-reverting, so that you can time the rebalancing trade at the point of maximum deviation. The time interval rebalancing research is generally focused on trying to determine the typical time span before momentum trends reverse, or the overall periodicity of market cycles, again so that our time interval lines up in a manner that is likely to produce favorable timing results.
Improving Returns With Rebalancing Is Sensitive To Market Timing?
So simply put, yes you can apply the rebalancing process on a purely mechanistic basis, but I have yet to see someone make the decision about what time interval or percentage-deviation targets to use for that mechanistic process without implicitly or explicitly doing so because they're trying to get the market timing right. And either way, anyone who has an expectation that rebalancing will produce higher returns, not lower returns, must implicitly be assuming they will get the timing right more often than they get it wrong. If you merely assumed that it was random - that you were just as likely to get example 2 (rebalancing leading to lower returns) as you were example 3 (rebalancing leading to higher returns) in my prior post - then you should assume that rebalancing, on average, will generate as many good trades as bad, no excess return due to timing, and overall result in a net reduction in return due to transaction costs.
Yet again, in practice, I don't hear anyone talk about rebalancing as a return-reducing risk management technique. We talk about it as return-enhancing. But the reality is that it only enhances returns when you get the timing right, and all the research that I've seen into fixed-time-interval and percentage-deviation rebalancing is essentially trying to explicitly or implicitly mine the historical data to figure out what time interval or deviation will produce the most favorable transaction timing.
Granted, I don't think that's necessarily a bad thing. There's some good research out there to show, over lots of different investment environments, what percentage-deviations or time intervals tend to result in transactions that yield favorably timed results. But once again, let's call a spade a spade; if you expect enhanced returns from rebalancing - and if you actually get those enhanced returns - it would appear to be a result of systematically getting the market timing right more often than you got it wrong. In other words, you picked a rebalancing process that gave you good market timing results.
So what do you think? Given that even a purely mechanistic rebalancing process requires some decision, up front, about what the timing will be - and in practice, that timing can increase or decrease returns - is there still an aspect of market timing in rebalancing? And if you don't expect any favorable market timing in your rebalancing transactions, does that mean you set an expectation for your clients that on average, rebalancing will control risk but at a cost of generating no increased returns while also decreasing returns due to transaction costs? Is that a fair characterization of rebalancing?
Ron Lieber says
Return reducing as compared to what, though? Or maybe the better question is “From what?” The alternative to this is individual investors chasing “market beating” returns and individual stocks (or managers) or their advisers trying to time the market (or chase non-existent alpha) rather than dutifully rebalancing (on whatever schedule). So yes, you’ll pay a few shekels to rebalance (or none with ETFs on certain platforms). But I’d still argue that the alternative isn’t so hot either. Am I missing something here?
Michael Kitces says
In the simplest case, it would simply be return-reducing relative to not-rebalancing (where allocations also drift), and would be expected to generate lower returns than the index-benchmark equivalents by the transaction costs necessary to rebalance back to those index benchmarks.
Mind you, I don’t necessarily think that this structure is bad, by any means. For many investors, it’s far better than how they will execute some of the alternative options.
My point here is simply to break apart this widely held viewpoint that rebalancing enhances returns. There’s no particular reason to expect that it would… except to the extent that you’re actually “good” at market timing, or conversely are “good” at picking a rebalancing methodology that systematically creates those “good market timing” situations.
There’s nothing wrong with just rebalancing, without any return-enhancing goal or expectation, simply because it’s an effective way to manage risk. But if that’s what we’re doing, let’s at least call a spade a spade, and stop talking about rebalancing as a strategy that ALSO enhances returns… because it doesn’t, unless you get the timing right.
And if we think an investor can pick a rebalancing process that will, systematically, get the timing right and enhance returns… maybe we need to revisit our beliefs about how efficient markets are and whether alpha can, in fact, be created in a systematic manner?
Rob Bennett says
There are several layers of confusion evidencing themselves here, Michael. I don’t mean on your part in particular. I see you as trying to sort through the confusion with the questions you have raised. The confusion is in the conventional understanding of how stock investing works. I believe that we need to look skeptically at some long-accepted fundamentals to make true sense of all this.
The suggestion in your blog entry is that rebalancing reduces risk but also often reduces return. This is consistent with the conventional thought that it is only by taking on more risk that we assure ourselves of higher returns (the idea here is that a higher return is _compensation_ for taking on more risk.
What if we are wrong about this fundamental point? This is what I believe to be the case. I believe that the general understanding of how risk and return relate to each other is today poor.
The goal should be to get the allocation percentage right. If you get it right, you increase your overall return while also diminishing your overall risk. This is not an either/or. You don’t have to take on added risk to get more return and you don’t have to give up return to lower your risk.
An investor who should be at 60 percent stocks when a particular valuation level applies is going to increase his return and diminish his risk by moving from either 70 percent stocks or 50 percent stocks to 60 percent stocks. The thing you are trying to protect against when you limit your risk is the possibility of losing money. Lowering your risk means limiting your losses, which is of course just another way of saying increasing your return. The investor who was at 50 percent stocks when he should have been at 60 percent stocks was in a sense increasing his risk as well as diminishing his return by going with the wrong allocation. By earning lower returns than someone in his circumstances should be earning he causes himself to fall behind in his quest to accumulate the assets needed for retirement in a reasonable amount of time and thereby imposes pressures on himself at later times to take on more risk than he can bear to make up for the shortfall.
Rebalancing is a good thing when it moves you in the direction of your proper stock allocation. The problem with the way in which rebalancing is generally practiced is that there is no way to know whether it is moving you in the right direction or not. Most people rebalance to return to a stock allocation that they intend to follow at all valuation levels. But if the risk and return associated with stocks changes with each change in valuations (I believe it does), there is no single stock allocation that is right at all times. So the rebalancing exercise may be moving you in the right direction or in the wrong direction, depending on the valuation level that happens to apply.
For example, say that an investor began investing at some time between 1996 and 2008, when valuations were sky high. He went with a 70 percent stock allocation, which is far too high for those sorts of valuation levels. Prices fall by 20 percent. That would cause his stock allocation to fall, which is a good thing. But if the investor rebalances to get back to 70 percent stocks, he will be moving his allocation back to a dangerous level by doing so (it won’t be as dangerous as it was prior to the 20 percent price drop, but given how high prices were at that time, they would still be on the high side after a 20 percent price drop).
On the other hand, if prices went up 20 percent, rebalancing would mean selling stocks, which would be a step in the right direction (not a big enough step, to be sure, but a positive step all the same).
Rebalancing is not inherently good or bad. It is being at the right stock allocation that is good and being at the wrong stock allocation that is bad. Whether rebalancing helps you get closer to the right stock allocation depends on whether you took valuations (which affect both risk and return and both in the same direction) into consideration when setting your return.
“I have yet to see someone make the decision about what time interval or percentage-deviation targets to use for that mechanistic process without implicitly or explicitly doing so because they’re trying to get the market timing right.”
Then you haven’t ever visited “Bogleheads” or ever read anything written by John Bogle.
(A condition which, I find literally beyond credulity, Michael.)
But, let’s take the other side of the coin: You overtly claim that most rebalance strategies overtly include a market cycle timing component to electing the interval.
Please provide references for those, would you please? I assume these will also be representative of a larger number, representing the majority of “rebalancers.”
Sorry, Michael, but I do not believe you can provide such evidence for your bold and direct assertions.
Michael Kitces says
You can take a look at “Opportunistic Rebalancing: A New Paradigm For Wealth Managers” from the Journal of Financial Planning as a starting point on this topic. Many planners view this as THE leading research paper in the financial planning world about how to execute rebalancing properly. The author of the paper also created “iRebal”, the leading rebalancing software in the industry, which in turn spawned several competitors in the rebalancing software space, all oriented around how to most effective execute rebalancing transactions.
And to say the least, the focus of the paper is entirely in line with the focus of the discussion I’ve presented here, starting with the very title of the paper – “OPPORTUNISTIC Rebalancing” – with a primary purpose of trying to determine how much EXTRA RETURN can be generated by an ‘optimal’ rebalancing strategy.
You can see the paper at http://www.tdainstitutional.com/pdf/Opportunistic_Rebalancing_JFP2007_Daryanani.pdf
Michael Kitces says
As a follow-up, just to give a somewhat arbitrary but very “classic” example of our industry literature discussing rebalancing, see:
The atricle cites the common, standard reasons I heard repeated throughout the industry (that you seem incredulous about, given your comments here), including “portfolio rebalancing is a safety mechanism that protects your portfolio from becoming dominated by over-valued assets” (i.e., it’s a valuation-traded tool!?) and “The second reason why you would want to use portfolio rebalancing is that it’s an automated mechanism for buying (relatively) low and selling (relatively) high.”
The point is not to dissect this article in particular, but simply to point out that this is a VERY standard perspective on rebalancing in the industry.
Yes, I realize that Bogle has some different views on this. But bear in mind that my primary audience is not individual investor “Bogleheads” – it is financial planners, who read the kinds of articles I am citing here and who, by and large, have a deep-seated (and I believe, incorrect) belief that one of the primary benefits of effective rebalancing is to enhance returns.
Let’s see what the “greatest” investors do to maximize return while minimizing risk (of loss, not volatility). Investing “champs” over long periods of time tend to be value investors. How do they succeed? Essentially by rebalancing – by selling all or a portion of securities which have risen to or above some measure of full value. Maybe most financial planning practitioners can’t be great stock pickers, but they can pay great attention to value at purchase. By consistently buyinhg cheaply, they quite naturally hold on until they either know they’ve made a selection or timing error or that their securities are headinhg toward or above fair value, at which point they’ll sell and repeat the investing process. I thinkRob had it right when he suggested rebalancing on valuation rather than by time is a portfolio return optimizer. As a fiduciary of occasionally nervous clients, it is job 1 to provide excellent risk adjusted returns. Doing so entails buying securities at “safe” (you can’t escape judgement calls) levels and selling at “nosebleed” levels or at breakdown. The point? If we limit risk first and look for gain second, we have a better chance to achieve excellent long term results. So..do we “rebalance” to reduce risk or to maximize returns? Like love and marriage/coupledom, they are intertwined. This said, how many value investors lost their jobs during the heydays of the 90’s? We may talk rationally here inh this forum, in this time, but there are times when being rational will cost you your job.
“Doing so entails buying securities at “safe” (you can’t escape judgement calls)…”
Ah, but there is the rub – I disagree that you can’t escape making topical judgment calls under the pressure of some latest supposed information!
I am dead set on personally trying to avoid those judgment calls altogether, to the degree possible! And buying the whole market, holding it for Warren Buffet’s favorite interval (i.e. ‘forever’), rebalancing occasionally back to the predetermined target allocation, dollar cost averaging as I add new money (I can almost here your professional CFP snorts over my amateurish views, from here!), and generally AVOIDING flavor-of-the-month approaches…
THIS is pretty darn close to my own ideal, with a view towards capturing most of the market return, and NOT towards some sort of optimizing strategy to get over on everyone else.
I don’t expect or ask that others follow that approach. Do whatever you like!
I only ask that others BE AS HONEST with what they are doing or proposing as I am. That’s not too much to ask, is it?
Michael Kitces says
Indeed, this is my point. I find the standard expression of rebalancing – that it is NOT market timing, yet is expected to maintain or increase returns – to be logically inconsistent. Either it does NOT systematically execute timing in a favorable manner, which would REDUCE returns on average by transaction costs (while maintaining a target allocation for risk management purposes), or it DOES execute with favorable timing and thereby allows rebalancing to generate higher returns.
But it can’t actually be both, despite the widely held views to the contrary. Challenging that conventional wisdom is the point of these blog posts.
You seem to have construed what I have written as being some kind of universal endorsement that all people should TRY to be excellent rebalancing market timers. I have made no such assertion, at all. The point here, again, is simply to say “If one claims that rebalancing will generate higher returns, then that person must also be claiming that he/she can execute an effective market timing strategy.” The two are inseparable, even though most planners unintentionally claim otherwise.
You are a clever debater, but you seem to attempt to draw me as having taken the position that I am somehow anti-rebalancing, when I am certainly not.
I merely asked you to provide citations for your assertion that trying to use market TIMING signals to establish the interval is predominant, instead of, say ‘annually’ or ‘quarterly’, but instead you give me a citation and include a definition which frankly undoes your own assertions:
“The article cites the common, standard reasons I heard repeated throughout the industry (that you seem incredulous about, given your comments here), including “portfolio rebalancing is a safety mechanism that protects your portfolio from becoming dominated by over-valued assets” (i.e., it’s a valuation-traded tool!?) and “The second reason why you would want to use portfolio rebalancing is that it’s an automated mechanism for buying (relatively) low and selling (relatively) high.”
The word “Automated” is key there, Michael. “Automated” means based on a pre-established criteria, that progamatically instructs the holder (or his agent) to make the necessary trades to establish stasis. NOT MARKET TIMING.
And I guess saying it is ‘valuation based’ is supposed to be a barb as well? I certainly believe in valuations. I just don’t know how to make money trading on someone else’s impression of what a ‘fair’ valuation is at a particular tactical moment in time, which is the very raison d’etre for a mechanistic, pre-established re-balancing plan, IMHO.
I readily admit I am not a planner or finance professional, just a layman. But you don’t need to be a weatherman to see which way the wind blows, so throwing out a journal article or two won’t satisfy the request that you establish what you claimed: That most rebalancers are doing so on an interval derived from timing cyclical market behavior, UNLESS those articles explicitly make that point themselves. I will take the time to research the articles to see if they indeed do that. I doubt that will be the case, since empirical evidence is that far and away, the most common rebalance interval for those who chose one, is ‘annually’ just because it is an easy, naturally occurring anniversary.
Annual re-balancing to a preordained asset allocation is hardly market timing, Michael. It is, in fact, the antithesis of it, and requires no high-priced or clever finance professional to shave money out of your portfolio in order to get it done. Perhaps that is the *real* issue you are fighting?
Michael Kitces says
I’m sorry that apparently I continue to fail to make my point clear.
My point is simply this:
If you assume that rebalancing is done on a purely random, automated, utterly-not-based-on-timing process, then you are asserting that you expect your rebalancing on average to produce ZERO additional return, while paying transaction costs, which means in the end your expected return from rebalancing is LOWER than the expected return by not rebalancing by at least the average amount of transaction costs. This cost is the tradeoff that you pay for maintaining your equity exposure at the targeted levels, as a form of risk management.
I am not saying that approach is good. I am not saying that approach is bad. What I am saying is that I have never met a financial planner who ever, in any way, expressed the idea that rebalancing was, on average, an REDUCTION in expected returns. It is always expressed in our industry literature as a net POSITIVE. Yet in reality, the ONLY way it can be a positive is to execute it with effective timing.
And in practice, what I see from financial planners across the country is that they WANT to generate higher returns by getting the timing better. That is why there is so much anxiety about the optimal rebalancing time interval, or the optimal rebalancing deviation threshold. Almost by definition, if you care at all about what time interval or deviation threshold is, you must be trying to get better timing on your execution to generate higher returns.
I realize this is not how YOU manage your money, or what YOU are looking for in a money manager. But the reality is that these are the dominating views in the industry, and in point of fact my goal here is to challenge those views and make it clearer to planners that you can’t have it both ways: either you believe rebalancing reduces returns in exchange for ‘risk management’ (keeping allocations on target), OR you believe rebalancing increases returns because you can determine a timing interval or deviation threshold that will systematically provide better-than-random effective market timing for the transactions.
But despite the fact that it is a widely held belief, we shouldn’t state that rebalancing is market-timing-neutral AND that it increases returns; that is logically inconsistent, which is the point of these blog posts.
The very title: “Opportunistic Rebalancing: A New Paradigm For Wealth Managers” kinda told me:
1) We are talking about something new, not the current status quo,
2) This will be someone who backtested some data, applied some algorithm and is going to propose it can beat mechanistic quarterly or annual re-balancing.
I was neither disappointed, nor surprised:
1) Jan 2008, a PhD CFP™, working for TD Ameritrade, tells us to throw out quarterly or annual mechanistic re-balancing becasue he has a better mousetrap. Even though a 2005 study shows that “91 percent of re-balancers use a structured method to rebalance, 9% due it ad hoc BASED ON MARKET CONDITIONS. Over 85% due it quarterly, semiannually or annually” (added emphasis mine)
From the paper: ‘The problem with quarterly or annual rebalancing is that the date are arbitrary, therefore we cannot possibly expect to catch the juiciest buy-low/sell-high opportunities.”
Michael, I consider my point made, via use of your own referenced paper. When I hear language like that from someone eyeballing MY money, I don’t walk away, I RUN; just like I do when they start talking about optimizing their own fees, etc.
I am not looking for someone to play timer trying to catch the ultimate ‘juicy’ trading opportunity with some algorithm that might be trading “daily, weekly, or biweekly” (from the paper) to try to capture market highs and lows. I believe that like most early retirees, I am looking for prudent capital appreciation and preservation, with a reasonable expectation of risk control.
“Juicy” doesn’t fit.
Kitces: “Either it does NOT systematically execute timing in a favorable manner, which would REDUCE returns on average by transaction costs (while maintaining a target allocation for risk management purposes), or it DOES execute with favorable timing and thereby allows rebalancing to generate higher returns.”
A classic false dichotomy, and I am forced to reiterate that such verbal gamesmanship does not suit either you individually or the discussion at large.
Going about it as I outline and as most actually practice: chose a method for re-balancing, bands or interval, apply it, and go about your business (like watching this Superbowl!!!!), will indeed almost certainly return a bit of a tailwind, as has been often noted, if not exhaustively documented already.
One need not TIME the market to profit from the random (or other) volatility, as you rake off your profits from the [temporarily] outperforming class, and set everything back to where you intend it to be, for yet another largely UNPREDICTABLE excursion… with only the end game in relative assurance, but the ‘winners’ for any one individual interval remaining unknown and unknowable…
While your goal is obtained — keeping an allocation proportioned for the risk you are prepared to take.
No timing required, nor desired.
Michael Kitces says
“…will indeed almost certainly return a bit of a tailwind, as has been often noted, if not exhaustively documented already…”
This is my POINT! Why do we often note a bit of a tailwind? BECAUSE WE HAVE DATA-MINED A TIMING INTERVAL WHICH HAPPENS TO LINE UP WITH HIGHER RETURNS.
There is no theoretical basis for why rebalancing sales should increase returns any more often than they decrease returns. If it’s merely random and unknowable, half the time you should conduct a sale that happens to produce higher returns (you bought low and sold high), and half the time you should rebalance while the market just continues higher (you end out selling low before the market rises further).
If you are assuming that rebalancing trades will systematically cause buys that produce higher returns more often than buys that reduce return (as I showed in the original examples, both are equally possible), then you must imply that there is some systematically exploitable timing process occurring in the markets, which you are taking advantage of by choosing the “right” timing interval and not the “wrong” one.
So again, to state it simply: if you asserts a belief that volatility is random (as you do here), why would there be any basis to assume that you will systematically have more favorable rebalancing trades than unfavorable ones? If the process is random, then on average the favorable sales should match the unfavorable ones, and your AVERAGE result from rebalancing should be market returns less transaction costs.
That’s not to say rebalancing is bad. It’s simply to make the point that you can’t claim volatility is completely random and inexploitable, and then also claim that rebalancing will somehow magically result in more favorable trades than unfavorable ones. There is no theoretical basis to claim that both would work simultaneously. In fact, the entire essence of assuming that volatility is random is to assume that you CANNOT systematically get favorable trades more often than unfavorable ones. In which case, rebalancing does NOT provide a tailwind; only FAVORABLY TIMED rebalancing does.
Michael, you wrote:
“There is no theoretical basis for why rebalancing sales should increase returns any more often than they decrease returns. If it’s merely random and unknowable, half the time you should conduct a sale that happens to produce higher returns (you bought low and sold high), and half the time you should rebalance while the market just continues higher (you end out selling low before the market rises further).”
I gather that rebalancing trades that reduce returns are losing return to momentum, while those that increase returns are riding on mean reversion. If that’s the case, then it seems to me that your argument hinges on momentum and mean reversion effects being of equal magnitude. Is that really the case? Although, assuming volatility is random, rebalancing should produce positive (mean reversion driven) returns and negative (momentum driven) returns with equal frequency, can we say that the momentum effects will be as strong as the mean reversion effects? Do you know if there is any data on this?
Kitces: “In fact, the entire essence of assuming that volatility is random is to assume that you CANNOT systematically get favorable trades more often than unfavorable ones. In which case, rebalancing does NOT provide a tailwind; only FAVORABLY TIMED rebalancing does. ”
“the intrinsic rebalancing potential of any asset pair is the difference between its mean variance and covariance.
The beneficial effect of poorly correlated assets on portfolio risk, undoubtedly appreciated since antiquity, was mathematically formalized only forty years ago by Markowitz.
The expected return of a rebalanced portfolio is not accurately represented by simple arithmetic weighting of individual asset returns. This is particularly true of assets which have high standard deviations and are poorly correlated. ”
Michael Kitces says
Thanks for sharing the Bernstein paper. Looks interesting. Will read through tomorrow if I can.
Yes do read it, but I’ll tell you in advance that many have criticized Bernstein’s approach since he put it out, so do look for those remarks as well for a counter-point.
I want to tell you that I appreciate the individual responses you’ve provided to my probably overheated criticisms, and to apologize for seeming to assign ulterior motives to this effort by you. I know you are a person of high integrity, but I guess I’ve seen just one too many timing schemes get unveiled as the next-best thing, so when I saw timing applied to re-balancing (which is fine – play your own cards your own way!) but along with it, the apparent position that anyone who re-balances and doesn’t lose money while doing it must be a de facto market timer… well, I don’t buy it. I may be the last one doing such an unfashionable thing, but I personally am putting my money where my mouth is and merely trying to capture as much of the market return as is possible through a diversified and risk-aware passive approach. I think I’ll continue to do so.