Hosting an industry conference? Ask us about including it in this ticker?
What do you think of our site upgrade?

Why Asset Allocation Doesn’t Matter In The Long Run

June 24
00:03 2015

(The following is the fourth of a special five part series meant to be shared by professionals and non-professionals alike. This particular series covers
only one of the 7 Deadly Sins Every ERISA Fiduciary Must Avoid.)

Asset allocation cannot produce consistent results when applied to short-term time periods, whether the duration is a single year or an entire decade. The real world evidence of this was revealed in our previous installment. But this 1430512_98601880_forty_yard_line_stock_xchng_royalty_free_300conclusion should not surprise anyone. For years, industry pundits have questioned the use of “tactical” – or short-term – asset allocation. Speaking with Morningstar’s Christine Benz in 2010, Jason Zweig, a personal finance columnist for The Wall Street Journal, said, “There are tactical asset allocation mutual funds that have been around for a long, long time. There are also many more that no longer exist because they were closed down, because the people with multimillion dollar budgets and supercomputers, and the world’s best investing software failed at it. So I don’t think that the average investor is likely to succeed at it, and I’m really not persuaded that most professionals will either.”1

Why Long-Term Asset Allocation Might be a Better Alternative
One of the corollaries of asset allocation is the need to rebalance. This is theoretically most effective when practiced over extremely long time periods. The focus on the long-term is critical because, as 2002 and 2008/2009 attest, there always remains the possibility of a short-term “anomaly.” Mark Lund, author of The Effective Investor and located in Draper, Utah, says, “The secret to making asset allocation work is having structured funds in the portfolio mix, rebalancing, and staying with it for the long run. One must know that there will be years when the market goes down but discipline is what wins the game.”

Establishing a fixed asset allocation with regular rebalancing is said to offer the advantage of systematizing the “buy low/sell high” philosophy that investment gurus have long touted as the secret to attaining a winning investment record. Robert R. Johnson, President and CEO, The American College of Financial Services located in Bryn Mawr, Pennsylvania, says, “Asset allocation is not only a valid concept, it is an essential part of a client’s investment plan. It is the rough guide and as part of an investment policy statement, the contract between an advisor and client. Target asset allocations help the client and advisor navigate rough waters. If, for instance, a client has a target stock/bond mix of 60/40, when equity markets have outperformed bond markets, periodic rebalancing will ensure that an investor’s asset allocation doesn’t vary too much from the target. It provides a discipline of buying the asset class that has underperformed and selling the asset class that has outperformed. Straying too far from the target asset allocation can subject the client to unintended risks.”

Again, as with the short-term hypothesis in Part III of this series, it’s easy to conduct an experiment that tests a similar long-term hypothesis.

The Long-Term Test
Like the short-term test, we used the Ibbotson data for annual stock and bond returns from 1926 – 2013. Unlike the short-term test, we’ve kept the mix limited to stocks and bonds. Also unlike the short-term test, we won’t guise this test in a story loosely based on characters from a classic Hollywood movie. This time we’ll go straight to the data.

First, we had to decide upon an appropriate long-term time period. Here, for reasons made obvious in Part V of this series, we picked a 40-year time frame. Next, we determined which stock/bond allocations to use. We went with 10% increments from 100% stocks to 100% bonds, figuring this was the only sure way to include the favorite balances of 70/30, 60/40, 50/50, and 40/60. There were a total of 49 40-year periods. The hypothesis here states there exists some allocation – not either 100% stocks or 100% bonds – that will provide the optimal portfolio mix. Here’s how they charted:

40-Year Period Annual Returns
for Different Stock/Bond Asset Allocations

What’s interesting with this graph is that it shows the best allocation is no allocation – just put everything in stocks. Across the board, the 100% stock allocation has the highest high annual return, the highest median annual return, and the highest low return. This suggests, for 40-year time periods – not an unusual time duration for someone saving for retirement – the best returns comes from not using asset allocation.

And before you can say “risk-adjusted returns,” a statistical analysis shows the 100% stock portfolio possesses one of the lowest standard deviations. Only 90/10 (0.90%) and 80/20 (0.88%) have a lower standard deviation than 100% stocks (0.99%). All other asset allocations have standard deviations in excess of 1.00%, with the figuring increasing progressively until you have 100% bonds, which has a standard deviation of 2.34%.

We can conclude, with no rebalancing, a portfolio consisting of 100%  stocks both performs better and is nearly as, if not more, reliable than all other stock/bond asset allocations.

But this leaves us with a question – What about rebalancing? Does rebalancing improve performance? That’s another hypothesis we can easily test.

The Rebalancing Test
In this test we pick one of the more popular asset allocations – the 60% stock/40% bond mix. Again using the annual data provided by Ibbotson from 1926-2013, this time we rebalanced annually. We used a 5% variance to trigger a rebalancing. In real life, to avoid the “blinker” problem (i.e., excessive trading caused by too small a rebalancing trigger), professionals will often use a larger variance (like 5%) before rebalancing. In our case, if at the end of the year stocks exceeded 65% of the portfolio, we sold down to 60% and put the balance in bonds. Likewise, if at the end of the year bonds exceeded 45% of the portfolio, we sold down to 40% and put the balance in stocks.

Remember, the concept of rebalancing is to “sell high” and “buy low.” The hypothesis would therefore state that, by rebalancing, the buy low-sell high asset class trading would yield a higher return. We ran the numbers with the 60/40 split to test this hypothesis. Here are the results:

40-Year Period Annual Returns for 60% Stock/40% Bond
Asset Allocation Mix Comparing With and Without Rebalancing

Again, the results of this test will disappoint the asset allocation believer. It turns out an investor who does not rebalance will receive a higher return in all areas compared to an investor who rebalances. Needless to say, if rebalancing a stock/bond asset allocation mix doesn’t beat the results of the static case, then it certainly won’t beat the 100% stock mix. Again, in this variation of the long-term test, asset allocation fails.

Why does rebalancing fail? It appears, because stocks routinely (and to a significantly higher degree) perform better than bonds. That means annual rebalancing has you selling stocks in more years than you’re selling bonds.

Analyzing the results
More than a half century ago, a group of soon-to-be Nobel Prize winners made a guess. Built on the Capital Asset Pricing Model, The Efficient Frontier, and the primacy of rational markets, it became known as Modern Portfolio Theory. The consequence was asset allocation. We computed the consequences of that guess. We just now compared those consequences to experiment. The consequences disagreed with experiment. Therefore, it – asset allocation – is wrong. It’s that simple. It doesn’t make a difference how beautiful the guess was. It doesn’t make a difference how smart the proponents of asset allocation are, that its creators won a Nobel Prize, or that a lot of really famous – and even successful people – will go to their grave believing in asset allocation. It disagrees with experiment. It’s wrong.

If not short-term, if not long-term, does asset allocation offer any value? Intuitively, it seems as though we’re missing something. The practice of “asset allocation” existed long before Modern Portfolio Theory, portfolio optimization, and high-end computing power. There has to be a reason why it’s been used for so long. It had to have added some value to investors.

Maybe it was the way it was used. Maybe it would be beneficial to review how old-time portfolio managers determined whether to invest in stocks or bonds. Maybe, just maybe, knowing this might be the best way professional (human) advisers will be able to survive the coming Robo-Advisor apocalypse.

We’ll cover this in our next and final installment.

Part I: 7 Deadly Sins Every ERISA Fiduciary Must Avoid: The 5th Deadly Sin – Asset Allocation
Part II: How’d an Innocent Fiduciary Like You End Up Asset Allocating?
Part III: Asset Allocation’s Greatest Failure: Short-Term Investing
Part IV: Why Asset Allocation Doesn’t Matter In The Long Run
Part V: The Hows, Whys, and Right and Wrong Way to Use Asset Allocation

Are you interested in discovering more about issues confronting 401k fiduciaries? If you buy Mr. Carosa’s book 401(k) Fiduciary Solutions, you’ll have at your fingertips a valuable reference covering the wide spectrum of How-To’s (including information on the new wave of plan designs) every 401k plan sponsor and service provider wants and needs to know. Alternatively, would you like to help plan participants create better savings strategies? You can buy Mr. Carosa’s latest book Hey! What’s My Number? How to Improve the Odds You Will Retire in Comfort right now at your favorite on-line or neighborhood book store.

Mr. Carosa is available for keynote speaking engagements, especially in venues located in the Northeast, MidAtantic and Midwestern regions of the United States and in the Toronto region of Canada.


1Benz, Christine, “Zweig: Tactical Asset Allocation Has a Lot to Prove,” Morningstar, September 29, 2010

About Author

Christopher Carosa, CTFA

Christopher Carosa, CTFA


  1. Dick Purcell
    Dick Purcell June 26, 15:28

    Chris, Monte Carlo shows that with return-rate variations of the sizes of those portfolios, to assess the distribution of the final results we need several thousand 40-year samples. But from your span of years, the number of non-duplicative samples you can produce is only 2.

    You claim to have 49 samples, which is way too few. Worse, your 49 samples are almost wholly duplicative — EG, sample 2 is 39/40ths the same as sample 1.

    Worse, your 49 duplicative samples grossly overweight your middle years of bond-adverse high inflation. EG, only 1 of your 49 samples includes 1926, but most of your samples include the high-inflation bond-adverse ’70s.

    For your sweeping conclusions, that evidence is WAY too weak.

    Dick Purcell

  2. Christopher Carosa, CTFA
    Christopher Carosa, CTFA Author June 26, 16:00

    Dick, thanks for the comment. Unfortunately, it appears the data disagrees with you. Take a look at what Roger Ibbotson says in the previous installment “the standard error goes down with time, not the number of observations.” Also, using rolling 40-year spans represents all relevant age groups in that time-frame, so as to account for any misleading “snapshot-in-time” errors. Finally, since this represents the complete set of historical data, the sample size is not relevant. As far as the “middle years of bond-adverse inflation,” while bonds did underperform in the late 1970s, the were up double digits in 10 of the next 20 years for the longest sustained bull market bonds ever had (this doesn’t include the run of double-digit years from 2007 to 2001). Meanwhile, stocks were down 9 out of 13 pre-WWII Depression years, not to mention the “Lost Decade” of the early 2000s when stocks had no growth. (BTW, that’s the precise reason why rolling time periods are used – they tend to mitigate issues like this that arise from the “accident of the calendar.”)

    In the end, the numbers don’t lie. The fact remains, and really cannot be disputed, during these real actual years, when investing over a 40 year span, the best asset allocation is 100% stocks. The only way to conclude differently is to not use, or worse, misuse, real data.

  3. Dick Purcell
    Dick Purcell June 26, 17:22

    Chris, I don’t mean to say your observations have no value. On the positive side, they reflect stocks-asset-class reversion toward the mean, which standard Monte does not reflect — and which was strong in the USA 20th century.

    But Monte’s demonstration of need for way more samples, when there are such large variations, IS valid. Heck, that need was revealed two centuries ago by Jacob Bernoulli’s law of large numbers.

    I see your data as so woefully insufficient it’s like being permitted to deal only one hand of cards, and getting 3 aces, and basing conclusions on that. There too, you could say “The data don’t lie” — but you wouldn’t have enough data for roaring conclusions.

    It’s better to recognize the woeful insufficiency of the data for your analysis-and-conclusions, ending up with far more uncertainty instead of your conclusions.

    The kind of analysis I’d rely on is many thousands of Monte Carlo samples, with various assumptions built in as to how stocks reversion might work in the future.

    Dick Purcell

  4. John Olsen
    John Olsen June 27, 17:22

    Using 49 periods that overlap hugely and generalizing from that statistically tiny sample demonstrates nothing that can be relied upon.

  5. Dick Purcell
    Dick Purcell June 29, 00:47

    Chris, you’ve done some especially good stuff recently — for example your recent survey showing “best interest” to be so lacking in specificity it permits almost anything.

    So I don’t like to attack. But this column is just foolishness.

    1. The right measure is what people invest FOR, net $ purchasing power for years of future needs and goals — not average return rate for the individual year.

    2. A realistic plan has the general pattern of (a) put in $X each work year, or (b) take out $Y each retired year, or (a) and then (b) — not just invest an initial $X for 40 years.

    3. When (1) and (2) are considered, sequence of returns becomes a major consideration. Your unrealistic example and wrong result measure fail to reflect that consideration.

    4. Did I mention absurdly inadequate sample size, cited previously? You claim 49 but those are mostly duplicates of each other as well a absurdly too few. Several thousand are needed. Of separate non-duplicate samples your span of history can provide only 2.

    5. Please don’t rely on Ibbotson saying standard error goes down with time. The uncertainty that counts, $ for the goals at the end, goes UP with time. That name can boast 1990s leadership in spreading ALL of the principal ways that MPT has been misapplied ever since, right up to fi360 today: (a) Using the label “expected return” for a return that is not expected; (b) Diverting the focus from the investor’s best interest, $ for his $ goals, to his short-term fears and % return rate for the individual year; (c) Presenting the dishonest appearance that any fool gamble WITHIN an asset class is faithful investment in the whole diversified asset class.

    Chris, bury this one. “Asset allocation is dead” is dead. It’s asset allocation’s Ibbotson-to-fi360 MIS-application that should be put to death. With your willingness to challenge conventional thinking, you can be a leader in that.

    Dick Purcell

  6. Chuck Miller
    Chuck Miller June 30, 10:36


    Rebalancing a risk reduction tactic. You lock in some gains that might escape if the market tanks. It’s not a return maximizing tactic. A return maximizing strategy would be 100% equity as recent studies indicate.

    In a personal retirement plan, rebalancing readjusts your risk and return to levels that are more sustainable over the long term. You never know when a Greece will happen.

    It’s a marathon you must win.

  7. Christopher Carosa, CTFA
    Christopher Carosa, CTFA Author July 02, 15:05

    Dick, as always, thanks for the comments. This would make an excellent topic under the practice category for – the private peer-to-peer community will be launching offically on July 15th. Be sure to join when you get the invitation email. Also, I know you made this comment prior to the publication of the fifth and final installment of this series. You might want to read that one.

Only registered users can comment. Login is sponsored by…

Order Your 401k Fiduciary Solutions book today!

Vote in our Poll


The materials at this web site are maintained for the sole purpose of providing general information about fiduciary law, tax accounting and investments and do not under any circumstances constitute legal, accounting or investment advice. You should not act or refrain from acting based on these materials without first obtaining the advice of an appropriate professional. Please carefully read the terms and conditions for using this site. This website contains links to third-party websites. We are not responsible for, and make no representations or endorsements with respect to, third-party websites, or with respect to any information, products or services that may be provided by or through such websites.