As the end of the year approaches, many people start to think about whether to rebalance their investment portfolios or not. But most people don’t consider exactly why they rebalance. It turns out that rebalancing is not the slam dunk decision that the finance industry often claims. I just added Article 8.7 on Rebalancing to the Investing Over Time series. The article explores the good, bad, and ugly of rebalancing, and this post covers the central concepts.
A chief attribute of investing mindfully is to buy a relatively simple set of diversified stock funds (and for the older investor some cash or possibly bonds) and hold them until you absolutely need to spend the money. But like most simple-sounding plans, successful implementation requires following a few important details. One of the most important of these details is rebalancing.
What is rebalancing?
The question of rebalancing arises because the asset allocation in any diversified portfolio will drift over time. The classic example is a portfolio with a combination of stocks and bonds. Because stocks typically have higher returns over time, as your portfolio grows, you will end up with a greater proportion of stocks to bonds. Here’s an example graph from Michael Kitces using average annual returns that shows a 50/50 stock/bond portfolio would be 80 percent stocks after 30 years.
Standard advice is that you should periodically reset (rebalance) your portfolio to its original allocations to correct for this drift over time.
Be wary of standard advice
Asset allocation is about selecting proportions of assets that achieve a suitable trade-off of returns and risk of permanent loss. The typical investing cook book says that if you don’t rebalance, then the trade-off of returns and risk you devised will evolve over time into some completely different trade-off. That is, you’re not actually doing what you originally intended. But mindful investing can examine both the original intent and the long-term consequences of any selected approach. We can’t simply assume that the evolution of a portfolio over time is undesirable. Perhaps portfolio evolution tends to boost returns or reduce risks or both.
Another reason to be suspicious of standard rebalancing advice is that it’s often coming from the finance industry. Rebalancing necessarily means more transactions than a pure buy-and-hold approach. And more transactions result in more fees going to the finance industry. Many studies have found a direct negative correlation between investing costs and performance. And small increases in annual costs can have a huge erosive effect on portfolio performance over time.
Zen of rebalancing
Some researchers have been questioning the wisdom of rebalancing recently. Andrew Miller points out that, almost by definition, rebalancing is a form of short-term market timing. (Almost all short-term market timing is doomed to failure for both mathematical and behavioral reasons.) For example, looking at the graph above, if you decide to rebalance away from stocks and back toward bonds after a period of 10 years, you’re making an implicit market-timing decision to favor bonds over stocks in the next period (however long that may be). But you have no guarantee that this renewed balance will improve your risk/return profile in the next period, because markets are unpredictable. Similarly, avoiding rebalancing can be viewed as just the opposite market timing decision, where you implicitly assume your current balance will continue to perform well in the next period.
Market timing is doomed to failure, and yet doing nothing and doing something (rebalancing) are both a kind of market timing. This sounds like a riddle with no answer. Buddhism, as the origin of mindfulness, is famous for examining life’s paradoxes for meaning and truth. This graphic shows some examples (which I paraphrase for fear of misquoting someone).
*I’m taking liberties here, because Yin and Yang isn’t a Buddhist creation. It originates from ancient China.
And you don’t have to resort to spiritualism to find potential truths that are paradoxical. Science offers plenty of paradoxical truths that are experimentally verified:
- A photon is both a particle and a wave, and it is neither.
- The particle and the observer are separate. But the mere observation of a particle changes its position and state.
- The “solid” chair you’re sitting on is mostly nothing, because all atoms contain mostly empty space.
You may disagree with some of these observations, and some of the apparent contradictions may feel more like a matter of semantics. For example, some scientists would say photons are simply quanta, which erases the apparent contradiction between particle and wave properties. That’s fine. Whether you agree with these observations or not, the practitioner of mindfulness is wonderfully prepared to explore potentially paradoxical truths. So, perhaps it’s possible that both rebalancing and not rebalancing are “correct”. Or perhaps, it depends on the circumstances, and sometimes rebalancing can be “good” and sometimes it can be “bad”.
We could simulate many different circumstances and try to derive conditions that favor rebalancing, but fortunately, others have done most of the work for us. First, existing research shows that rebalancing stock/bond portfolios usually has a different outcome than rebalancing diversified stock portfolios. Michael Kitces presents these dueling return graphs for a stock/bond portfolio (top) versus a stock portfolio (bottom) containing a combination of large and small cap stocks.
These are 30-year rolling returns. For example, if you invested your portfolio starting in 1926, you can look at 1926 on the horizontal axis and see the annualized return produced by that investment from 1926 through 1956. (That’s why the graph ends at 1984, which includes the next 30 years through 2014.) The gray shaded zones are times that rebalancing (the blue line) provided better returns than not rebalancing (the “buy-and-hold” orange line).
You’ll note that sometimes rebalancing improves returns and sometimes it doesn’t. This is our first clue that the case for rebalancing might be circumstantial. If nothing else, the merits of rebalancing depend on when you happen to invest. However, these graphs focus solely on return performance and ignore the risk side of the investing equation.
You can run these kinds of simulations for many different periods and combinations of assets to examine both the returns performance and risks, or combine them into “risk-adjusted” returns. William Bernstein ran simulations like this back in the 1990s and Vanguard conducted a useful study in 2010. This table summarizes the findings. (I borrowed this concept from Kitces, but I’m using a different format.)
|Portfolio Type||Impact on Returns||Impact on Risk|
|Assets with Similar Returns and Volatility (e.g., diversified stocks)||Increased Returns||Little Change|
|Assets with Different Returns and Volatility (e.g., stock/bond portfolio)||Decreased Returns||Decreased Risk|
For a portfolio of diversified stocks, rebalancing usually results in increased returns with little increase in risk. Bernstein termed these additional stock returns the “rebalancing bonus”. For portfolios of stocks/bonds, rebalancing usually decreases returns and risks. In this case, there is no “rebalancing bonus” for returns, but it still results in improved “risk-adjusted” results. Bernstein’s work also clarifies that the rebalancing bonus for stock portfolios becomes bigger as the volatility of the stock components increases and their correlation decreases. This is our second indication that the case for rebalancing is circumstantial; the usefulness of rebalancing depends on what’s in your portfolio.
How significant is the rebalancing bonus?
Rebalancing performs better with certain types of portfolios and at certain times. When it does perform well, how big is the rebalancing bonus? You’ll note in the two Kitces graphs above that rebalancing appears to provide slim improvements in returns (and only at select times).
However, many people focus on the improvements in “risk-adjusted” returns provided by rebalancing. I’ve not used the term “risk-adjusted” much in Mindfully Investing articles, because a mindful approach isn’t heavily influenced by typical measures of “risk”, which are all about routine volatility. The real risk that we should avoid is a permanent loss when it comes time to spend invested money. And long-term investors are mostly vulnerable to volatility in a relatively narrow 5 to 10-year window around the time when spending starts (e.g., the start of retirement). However, the mindful approach cannot completely ignore volatility as a measure, because large decreases in volatility (and associated large increases in risk-adjusted returns) would decrease the chances of a permanent losses in the vulnerable period.
Let’s examine whether two frequently cited studies found large improvements in risk-adjusted returns due to rebalancing. One is the 2010 Vanguard study I already mentioned, and the other is a 2017 study by Baker et al. There are multiple methods to calculate risk-adjusted returns. I’m going to use a common-sense measure called return-risk ratio that simply divides the average annualized return by the annual volatility (both in percent). This can be viewed as the percent return achieved for each percent in volatility experienced. Let’s say your portfolio gets a 10% annual return with 20% volatility (standard deviation). That means you get half a percent return for every percent of “risk” experienced. The higher the result, the better the risk-adjusted return. Here’s a table compiling results from the Vanguard 2010 and Baker et al. 2017 studies.
|Study and Method||Annualized Return (%)||Annualized Volatility (%)||Risk-Adjusted Return|
|Baker et al. 2017|
|Volatility Based Range||5.15||10.84||0.48|
|Target Dated Fund||4.88||9.68||0.50|
The Vanguard study uses a portfolio of 60% U.S. stocks and 40% U.S. bonds. The Baker et al. study uses a portfolio of 30% large cap U.S. stocks, 10% small cap U.S. stocks, 15% international stocks, 30% U.S. bonds, 10% commodities, and 5% cash. There are some funky sounding names for the rebalancing methods here. But you don’t have to understand each method, because our main purpose is to see if any rebalancing approach provides substantial risk-adjusted return improvements.
In the Vanguard study, the best improvement in risk-adjusted terms is just 0.1% of a return for each percent in volatility as compared to our “no rebalancing” benchmark (in red). In the Baker et al. study, the best risk-adjusted improvement is a mere 0.04%. Further, the meager Vanguard study improvement includes an actual decrease in annualized return as compared to no rebalancing (see the first column). And for the Baker et al. study, we receive an increased return for only the last two rebalancing approaches noted with quotes around the names. (These last two strategies are theoretical and complicated approaches based on momentum trends that involve a lot of market-timing. And as Andrew Miller points out, it’s highly questionable that either approach is implementable, particularly by an individual investor.) And as shown in the second column, the reductions in routine volatility over a no rebalancing approach are small in both studies: in the 2% or less range.
Ben Carlson says that if you want to win any argument about markets, just use a different historical period for the analysis. We’ve already seen that rebalancing didn’t work well in certain periods. Could it be that these studies’ poor results were caused by the historical periods used? Dan Sotirof asked exactly this question and calculated the rebalancing bonus (increased or decreased returns) for a 60/40 U.S. stock/bond portfolio (and other stock/bond portfolios) over multiple historical periods as shown in this graph.
For long-term investing (20 or more years), Sotirof found more periods where rebalancing harmed returns. The longer the duration, the more likely that this “rebalancing penalty” occurred. His results were similar for other stock/bond portfolios. Because investing mindfully means investing for the long-term, these longer duration results are the important ones.
Rebalancing stock portfolios
None of the above studies involved stock-only portfolios. The greatest proportion of stocks we’ve seen in any analysis so far is just 60%. Bernstein found that rebalancing works best for more volatile stock-only portfolios, especially where the various stock holdings turn out to be relatively uncorrelated. The mindful investing plan heavily favors stocks over bonds and cash. So, it seems like we should examine rebalancing specifically for stock-only portfolios.
Wesley Gray studied stock-only portfolios to see how much of a rebalancing bonus was created by using stock components with lower correlation and higher volatility. Here’s the key graph from the study.
The correlations of the stock components in these portfolios range from 60% to 89%. Because most stock components tend to be correlated or even highly correlated, the lowest value here of 60% is significant for its relative lack of correlation with other stocks, even though 60% might not sound particularly uncorrelated. The volatility (standard deviation) of these stock components range from 20% to 36%, which is relatively volatile even by stock standards. In general, as you move right to left in the portfolios shown in the graph, the correlation decreases and the volatility increases. As expected, all the stock-only portfolios have a rebalancing bonus (increased returns), and the portfolios with the greatest volatility and lowest correlation get the biggest bonus.
Getting back to our key question, how big is the rebalancing bonus for stock-only portfolios? Well, the low-end of 0.26% annualized return increase is kind of disappointing. No doubt this small additional return will compound and spur a considerably higher portfolio value over many years. But stop and compare 0.26% to other compounding factors that commonly erode portfolio returns such as inflation (usually at least 1 or 2%) or costs of actively managed funds (at least in the 1% range), or taxes (around 1% median for active funds). This lowest rebalancing bonus won’t be enough to offset any one of these costs.
But what about the two portfolios on the right part of the graph? The increased returns range from 0.66% to over 1.24%. Isn’t that significant? I would say, yes. But you must examine the portfolios involved. The study picked out only the most volatile 50 to 100 stocks from broader stock indices to obtain components with nose-bleed volatility (26 to 36% annual standard deviation). This is in the same range as some of the most volatile stock indices available, such as emerging markets and world-wide small caps. While mindful investing places less emphasis on routine volatility, the wild rides represented by these concentrated stock portfolios can’t be entirely ignored. And we still can’t predict the future. As we saw in the very first graphs in this article, there’s still a very real chance that the rebalancing bonus may not materialize in your investing life, even using these highly volatile stock-only portfolios.
Rebalancing appears to work well in limited cases. Transaction costs impose an additional limitation on rebalancing. Rebalancing usually means selling winners (assets that grew substantially) and buying losers (assets that grew slowly or shrunk). Each sell and buy event comes with fees and other, sometimes hidden transaction costs. And capital gains tax may be due after selling winners in taxable accounts.
Vanguard examined these costs and noted that they’re difficult to explicitly quantify. That’s because the costs are often included “indirectly in advisory fees or reflected as trading restrictions”, not to mention other smoke and mirrors employed by the finance industry. For example, in some 401Ks, rebalancing is allowed for no “additional” cost, but it’s hard to tell whether the total 401K fees would be less if rebalancing were not available. Even if you choose the simplest case of a taxable account that charges a flat fee for each trade, the percentage cost for rebalancing a $100,000 portfolio will be much higher than the percentage cost for rebalancing a $1,000,000 portfolio. And when it comes to tax costs, you get totally different answers depending whether you are using tax-advantaged or taxable accounts.
I’d hoped to conduct a cost-benefit analysis directly comparing rebalancing costs to any rebalancing bonus for different scenarios. However, it appears highly unlikely that such an analysis would be generally relevant.
Before we can say much more about rebalancing costs, we need to recognize that there are a plethora of possible rebalancing methods including:
- Rebalancing at set time intervals
- Rebalancing based on thresholds of percent change in portfolio allocations
- 0% threshold
- 1% threshold
- Combinations of time and threshold rebalancing
- Many others – see examples in the Baker et. 2017 study
As a proxy for costs, Vanguard examined how the frequency of transactions associated with various time and threshold methods contributed to “the number of rebalancing events” and the “annual turnover” (how much of the portfolio was sold and bought in any given year). Unsurprisingly, they found that as the time intervals and thresholds decrease, the larger the turnover and number of events, as shown in this table from the study.
Vanguard also looked at another often-recommended method, which they termed “income” rebalancing. This is where you use only stock dividends, bond interest payments, and any other account interest when rebalancing the portfolio. Because it’s often (although not always) “re-invested” anyway, there is no added cost to investing income back into loser assets. I would add that the same is true of any regular savings you are adding to your account (as part of regular 401K contributions, for example). These regular contributions would also typically be invested in something anyway and can be invested in loser assets with no added cost. The merits and reasons for re-investing stock dividends and bond interest payments is discussed more in Article 8.8 (coming soon). With income rebalancing you may not have sufficient funds to completely rebalance your portfolio at any given time. The idea is to get as close to your target balances as possible using the income and new contributions that are available.
If you choose to rebalance, these cost observations lead to some rules-of-thumb for minimizing rebalancing costs, which can be separated into the young and “old” investor types discussed in Articles 3 and 8. The point of any rebalancing is to manage your total investing return/risk profile, even if those investments occur across several accounts such as a 401K, a HSA, and a taxable account. Accordingly, these rules take advantage of the characteristics of each type of account.
Cost minimization for the young investor – For young investors we have the following mindful rules:
- If your investing within a 401K or similar plan, in most cases, you’re already paying the plan’s costs regardless of rebalancing decisions. Other than abiding by plan restrictions on frequent trading, you can rebalance as much as you want at no added cost. Some 401Ks also offer rebalancing as an automatic feature.
- If you must sell winners for some reason, then do so in tax-advantaged accounts to avoid taxable events.
- For both independent tax-advantaged and taxable accounts, you should avoid the order fees associated with selling winners, and instead, only augment losers with income and new contributions. If you can’t reach your target balance with the available funds, get as close as possible. In the case of taxable accounts, this rule also avoids potential tax events associated with selling winners.
Cost minimization for the old investor – In general, the young investor rules also work for older investors, where applicable. But there are some additional nuances:
- If you use investment income for living expenses, then you obviously can’t rebalance using those same funds. Also, most retired folks aren’t making new contributions to accounts. Therefore, you should be more reticent to rebalance, given the selling and buying will create two cost events and might also create a tax event if done in a taxable account.
- If you need to sell investments for living expenses, take them from winners (preferably in a Roth IRA to avoid capital gains) and avoid any flip-side buying cost.
- After age 70.5, you must take required minimum distributions (RMDs) from tax-deferred retirement accounts. Take RMDs from winners.
- If you plan to make charitable contributions (which can also have tax advantages) or gifts to family, take those from winners as well.
The existence of many potential rebalancing methods raises the question of which one might be best in terms of returns and risk as well as costs. For returns/risks, the tables above show that all these method variations usually result in pretty small differences in returns, volatility, and risk-adjusted returns. So, if you find yourself in the limited situations where rebalancing might make sense, you probably shouldn’t expect one method or another to provide outstanding return or risk improvements.
However, coming back to costs, Kitces compared various methods and recommends a threshold-only approach. This is because a time-based approach can cause you to transact relatively small amounts, which generally means higher and/or more frequent cost events. He defines a preferred threshold method called “tolerance bands” and has some suggestions about how to define them. However, he never directly recommends specific bands for general application.
Similarly, Vanguard vaguely suggests that either time or threshold-based methods can be used, and the key is to reduce costs by not having the time-period or threshold too small. They conclude that annual to semi-annual rebalancing or using about a 5% threshold might be best. However, even after close examination of their Table 7 above (and all the other tables in their study), I can’t discern the rationale for rebalancing semi-annually or at a 5% threshold. For example, these methods result in an annual turnover of around 2%, which is comparable to some of the smaller time and threshold levels examined in their study. In contrast, annual or 10% threshold balancing results in turnover closer to 1.5%, with very similar return and volatility results as shown in Table 7.
Given it appears impossible to come up with a consistently applicable cost-benefit comparison, these researchers understandably balk at giving specific methodology conclusions. Vanguard goes so far as to say:
- “Just as there is no universally optimal asset allocation, there is no universally optimal rebalancing
Thus, the mindful conclusion regarding rebalancing methods is necessarily simplistic. Namely, if you’re going to rebalance, then use the rules listed above and relatively infrequent rebalancing events to minimize transaction costs as much as possible. This will maximize any rebalance bonus you may (or may not) eventually realize.
We’ve discovered that:
- Standard investing advice assumes that rebalancing is productive.
- However, rebalancing versus pure buy-and-hold are mirror image market timing decisions. One is not inherently better than the other.
- It’s mindful to consider that rebalancing may be both “good” and “bad”, or effective only under specific circumstances.
- The effects of rebalancing vary over different periods and with different portfolios; rebalancing effectiveness is circumstantial.
- There is often no “bonus” (increased returns) for rebalancing portfolios of stocks and bonds, particularly when investing over many years. And risk-adjusted improvements are also usually small.
- There is a rebalancing bonus for stock-only portfolios, but it’s most prominent for portfolios of relatively uncorrelated stock components that are concentrated in highly volatile areas. And the bonus may disappear in any given period as correlations change over time.
- It appears impossible to create a generally applicable cost-benefit analysis for various rebalancing methods and scenarios.
- No one type of rebalancing method appears superior from a return/risk perspective.
- Any rebalancing method being considered should be closely evaluated to minimize transaction costs specific to that situation and investor type.
- Use of income and ongoing contributions for rebalancing is a key cost minimization technique.
- Minimizing the number of transaction and taxable events will help maximize the rebalancing bonus you may (or may not) eventually realize.
Mindful contemplation has revealed the answer to the rebalancing paradox. While buy-and-hold and rebalancing are both technically a form of market timing, rebalancing often adds additional costs, only yields tangible benefits for select types of portfolios, and may fail to produce the sought-after bonus in any given investing period. Rebalancing sounds like a paradox, but in reality, the main issue is that the benefits are confined to a narrow set of circumstances.
Rebalancing is probably reasonable in two potentially overlapping circumstances. First, if you decide for other reasons to invest in a stock-only portfolio composed of multiple highly volatile stock components, then rebalancing that portfolio is probably worthwhile and may provide some additional return.* Even in this case, minimizing the number of rebalancing events may be key to obtaining the elusive rebalancing bonus. Because the cost side is murky, we can’t determine an exact best rebalancing method for highly volatile stock-only portfolios. But my admittedly qualitative judgement is to rebalance no more often than annually and only when allocations have changed by a 10% or greater threshold.
Second, if you have a mostly stock portfolio that is almost entirely within a 401K (or similar) account that charges no additional costs for periodic rebalancing, there is probably little harm in rebalancing. This may result in additional return depending on the volatility and ongoing correlation changes in your various stock components. Because there are no added costs in this scenario, the rebalancing method is not particularly important. Pick something that’s allowed by your plan (like annual rebalancing) or simply opt for the automatic rebalancing feature, if available.
For stock/bond and similar combination portfolios, the mindful conclusion is that rebalancing is unwarranted. In these cases, rebalancing provides slightly better risk-adjusted returns, but mostly by decreasing volatility incrementally. This small volatility decrease probably won’t help you sleep much better at night, and the rebalancing will simultaneously slow your portfolio’s growth. Further, the added costs of rebalancing may undermine even these limited benefits.
*I’ll note that this is a defensibly mindful portfolio, although it’s a rather extreme investing approach. I concluded in Articles 7.1 and 7.3 that a stock-heavy portfolio, which is moderately diversified across a range of common stock types, is all that’s needed for most investors. The degree to which you push your stock balance into highly volatile components is mainly a question of whether you are mindful enough to withstand the periodic pain of dramatic portfolio draw downs and can avoid panic selling that ruins the whole plan.