ASSET ALLOCATION DETERMINES 94% OF YOUR INVESTMENT RETURNS ... FALSE
The claim that asset allocation determines 94% of returns continues to be widely used by industry in selling both (surprisingly) active and passive products. It is a sales pitch. The advice industry needs a 'service' to sell. The asset allocation recommendation
- is a cheap product to offer. It is computer generated from simple data from a standard questionnaire.
- cannot be 'proved' right or wrong in hind-sight.
- fits into the parameters of good advertisements - a simple and easy to understand product.
- does not involve any other professional (like a stockbroker) who might take away your custom.
- purposefully creates the misunderstanding that your returns will increase after proper asset allocation.
So of course the industry thinks it is really important! It's really important to their business model.
The asset allocation decision does not determine returns. Your expected returns are determined by what assets you own, their risk and your ability to stay-the-course when their prices drop. Different asset classes have different risk - e.g. equity is more volatile than debt. Generally, the higher the risk of the asset class, the higher return that is expected. You increase your expected returns by increasing the portion of more risky assets you own. The asset allocation decision should position you along the risk-return continuum. It does not increase your risk-adjusted return.
The benefit of asset allocation: The asset allocation decision process should attempt
to measure your personal tolerance for risk (falling prices). Theoretically, to maximize returns you hold 100%
high-risk/high-return equities. This does not work in practice because
once the threshold of your tolerance for risk is passed, and you are
tested by market losses, your decisions will become counter productive.
You will not realize the long-term returns because in the short-term
you commit the sin of selling at market lows.
In real life your optimal
asset allocation puts you just inside the far edge of your tolerance for volatility. The process always reduces the theoretically possible returns to allow for your own risk tolerance. It optimizes the returns that you personally can
You need not allocate between asset classes to adjust your risk. Within each asset class there are securities with widely differing risks. Within common stocks there are steady blue chips offering decent returns and volatile small caps promising huge returns. Same for bonds. Government bonds are very safe but low yielding, while high yielding corporate debt can be very risky.
Problem 1: How accurately do the advisors measure
your personal risk tolerance? They have a computer program and a series
of factual questions, but your answers give no insight into your risk
tolerance. None of the questions ask you to assess your possible
reactions to loss, or evaluate your mental toughness.
The computer program spits out an allocation between equities, debt,
real estate and commodities. But there is no way to back test how
optimal that allocation really was. There is no way to test "how you
would have reacted" given a different allocation - "what your returns
would have been" if the allocation had been different.
In April 2011, many years after this article was originally written, the Journal of Financial Planning finally (a LITTLE late) admitted that "The overwhelming majority of people ... actually have no clue what risk tolerance is until it whacks them in the face. It really can’t be shaped other than through repeated experience." "Risk tolerance questionnaires. They’re silly."
No one but you yourself can find out how much risk you can stomach. You can only find out by experience. Since falling markets only happen every few years, you must move slowly into risky assets, allowing time to show how you react in falling markets. Sitting down with an adviser owning only GICs, and walking out holding 50% common stocks because you were given that asset allocation is nuts.
Problem 2: The concept of asset allocation is
grounded in presumptions of long-run returns and risks for different
asset classes. In the long-run stocks outperform bonds, value stocks
outperform growth stocks, small caps outperform large caps. But in the
long-run we are all dead.
In the medium term: E.g. It took 14 years before the market got back
to its starting point after the 1929 crash. Fourteen years of 0%
returns for equities. E.g. From 1966 to 1982 (16 years) the Dow Jones
returned a total 1.2% loss. In the mean time inflation increased 7%.
History is full of these exceptions to the rules. No one knows what
future markets will be like. Portfolio allocation between asset classes
is not going to help predict it. Nor will it protect you from the
Problem 3: If you accept that the point of asset
allocation is to manage your risk tolerance, then rebalancing is
necessary. The risky assets, with a higher theoretical return, will
grow faster than the others. Their weighting in the portfolio will grow
until your threshold for risk is exceeded.
Academic studies show that more frequent rebalancing decreases
returns. Higher returns come from less frequent rebalancing, even to
four year intervals. The conclusion generally accepted is that
rebalancing should be done only when the risk profile gets seriously
out of whack, because you pay for peace of mind with lower returns.
Similar to the misrepresentation that asset allocation increases returns, it is also said that rebalancing increases returns. You can see the actual results from Canadian data in the graphs on Sheets 14 and 15 of this spreadsheet. There is no income gain. Rather the costs involved reduce returns.
Problem 4: It is now common to hear people say "Oh,
my returns were much lower than the market index this year ... but
that's OK because my financial plan shows I don't NEED to earn great
returns in order to meet my objectives". Wrong, wrong, wrong. Equity
markets are unpredictable. You need to make hay while the sun shines because it will rain tomorrow. Accepting
sub-standard returns today will not earn you higher returns next year. You
must make use of the better years to offset the lesser years. Don't be misled by averages.
Problem 5: The advisors' emphasis on asset
allocation detracts from the more important issue of removing emotions
from the management of equity portfolios. Your emotional reaction to falling values is hugely influenced by self-knowledge and self-control -
by creating an investing 'process' that acknowledges your emotional
vulnerability and controls it. See the discussion at Emotions Destroy Returns
Problem 6: The academic research on the best asset allocation for sustainable retirement funding has only been modeled for 30 year timeframes. In reality many people retiring at 55 will live to 100. That is a 45 year timeframe. For more on the retirement funding see Retirement - Got Enough?.
Problem 7: Asset allocation advice always comes in the form of "percent in asset class 1, percent in asset class 2, etc. But the first allocation everyone should make is measured in dollars, not percentages - the specific dollars you will need on a specific date. E.g. if your savings are earmarked for a child's university tuition, a recommendation to hold 10% bonds, without specifying the maturity date or dollar value, is useless. The price volatility of an asset exposes you to the risk that its value is depressed at exactly the time you need to cash out. When you need surety, you must hold the type of asset that can guarantee that cash.
Problem 8: Nearly none of the asset allocation models incorporate the mortgage on your personal residence. Yet it makes no sense to purchase 4% Government Bonds while owing 5% on a mortgage. Both are risk-free. There would be an immediate gain from selling the bonds to pay down the mortgage. Recommendations on asset allocation almost always tell you to buy debt.
Problem 9: Nearly none of the asset allocation models incorporate your ownership of a home, or annuities, or a company pension, or government benefits like Canadian CPP. These should offset allocations to debt because they are so risk-free and certain.
The "94% of Investment Returns" assertion comes from Brinson, Hood, and Beebower (1986, 1991) that stated, “…investment policy [static asset allocation] dominates investment strategy (market timing and security selection), explaining on average 93.6 percent of the variation in total plan return.” This original statement has been distorted two ways. First, Brinson was measuring variation in returns (volatility), not returns. Second, his word 'explains' [from regression analysis of correlation] has been changed into 'determines' [implying causation]. Levi Folk's article tries to explain what (little) was really meant by the paper. Roger Ibbotson's short piece tries to explain what was actually stated and proved, and more importantly what was NOT.
It is generally accepted that the paper's choice of metrics means it can conclude only that the average pension in the sample adhered very closely to its asset allocation policy and used broad diversification within asset classes. In other words they were closet indexers. The closer your portfolio mimics the index being benchmarked, the closer the volatility of your returns will be to that benchmark. When you own the benchmark ETF for each asset class then 100% will be determined by your asset allocation.
What is the impact of asset allocation on returns? That is the question most people want answered. Ibbotson concludes that for passive buy-and-hold indexing type investors, returns will be 100% determined by the returns of the asset classes they own - as you would expect. You buy the index - you get the index returns. "The idea that asset allocation policy in aggregate explains 100 per cent of the “level” of return before cost is not a profound idea." (Bill Jahnke). If you include enough portfolios within your sample, your sample BECOMES the market index benchmark, so of course there is 100% explanatory power.
For short-term traders in individual stocks, asset allocation is less important - as you would expect. Daily volatility - no matter what stock or sector or asset class - becomes most important. When the market as a whole has a good day, most stocks rise together, but selective stocks may rise more, or less, or even fall in price.
Surz concludes essentially the same. Asset allocation "explains approximately 100% of investment returns. If a manager succeeds in adding value, this can decrease to as low ... as 75% on a risk-adjusted basis. On the other hand, if the manager fails to add value, policy can explain as much as ... 165%." In other words the returns of an actively managed portfolio will be rooted in the index's returns with an overlay of
- 33% upside if successfully beats the index [(1-75%)/75%] , or
- 39% downside if underperforms the index [(1-165%)/165%].
A group of academics have approached the issue by trying to measure the relative importance of the decision (policy asset allocation) versus decisions on market timing (tactical asset allocation) and security selection. They measure the dispersal of outcomes resulting from different decisions. The most important decision is the one where the spread between the best and worst percentile of excess returns is largest - where there is more opportunity for improving returns by making a good decision, and risk of destroying returns by making a bad decision. These authors come to different conclusions - especially when comparing actual decision outcomes vs. theoretically possible decisions. References - Tokat, Kritzman, Assoe, Assoe
Another side of this issue falls within the active vs. passive portfolio management debate - which outperforms? It should be noted that twenty years after the original Brinson publication, the co-author Hood revisits their original paper. He makes the point "Nothing in the original paper suggests that active asset management is not an important activity." This comment is important because the paper has became part of the pro-passive camp's arsenal.
None of all that academic work has any bearing at all on your own decision on what asset classes to own or in what proportion.