Is it possible for a portfolio to be too diversified? There are certainly a lot of articles making that claim. Investopedia even has one called The Dangers of Over-Diversifying Your Portfolio that concludes like this: “Diversification is like ice cream: it’s good, but only in reasonable quantities.”
Is the only free lunch in investing really just a fattening dessert? That depends on what type of investor you are.
Only active investors can overdiversify
The concept of overdiversification is only meaningful to investors bent on beating the market. If you’re a fund manager who is trying to outperform an index, it’s true that holding 100 stocks will make your job even harder than it already is. The more stocks you hold, the smaller the impact of your decisions, both good and bad. You might make a brilliant call on a company that doubles in value, but if that company makes up just 1% or 2% of your fund, the effect may be trivial.
An extremely overdiversified active fund manager is called a closet indexer: he or she holds a portfolio that closely resembles the benchmark, while charging fees that can be 20 times higher than an index fund. The chance of a closest indexer beating the benchmark over any meaningful period is close to zero.
A manager with a more concentrated portfolio—perhaps just 15 or 20 companies—at least has a fighting chance. When each stock makes up 5% to 8% of the portfolio, the big winners will have a meaningful impact. Indeed, there is some evidence that the less a portfolio resembles the index, the more likely it is to outperform. A well-known Canadian proponent of concentrated portfolios is Steadyhand: their Small-Cap Equity Fund, for example, includes just 17 stocks.
Index funds don’t dilute quality
So if you’re an active manager whose job depends on outperforming the market, then it clearly makes sense to be concerned about overdiversification. But if you’re a regular Joe or Jane who isn’t concerned about pleasing an investment committee, can you be too diversified? The answer is no.
It’s possible for a Couch Potato portfolio to be inefficient: for example, if you own two or more index funds that track similar benchmarks, your portfolio will be unnecessarily complicated and you’ll incur extra costs to rebalance. If an index fund incurred excessive trading costs trying to replicate a huge and unwieldy benchmark, that would also be a problem. But that’s not what most people mean when they talk about portfolios being “overdiversified.” A more typical argument goes like this:
“If you own 1,000 stocks you will have eliminated specific or unsystematic risk but your portfolio will not own the highest quality or best stocks… You might be better off owning 25 quality stocks among a variety of industries than own [sic] 50 quality stocks and 950 mediocre stocks that pull your portfolio performance down.”
This is nothing more than the tired argument that index investing is a poor strategy because “you get the bad companies along with the good ones.” While that sounds sensible, it assumes that investors can reliably identify the 25 “best stocks” from a universe of 1,000 or more, and that any added returns from these stocks will be sufficient to cover the cost of the research, transactions and taxes over the long term. This is possible, of course, but decades of data make it clear that it’s extremely unlikely, and that the overwhelming majority of investors fail in the attempt.
The equity ETFs in the Complete Couch Potato give investors exposure to about 10,000 stocks in more than 40 countries. Does this include “mediocre stocks that pull your portfolio performance down”? I suppose it does. But it also includes all of the companies that will ultimately deliver the highest returns. Until someone figures out a way to identify these two groups in advance, you give yourself the best odds by owning the whole market.
If diversification is like ice cream, then it’s fat-free and sugar-free with chunks of broccoli. The more you eat, the better off you’ll be.
{ 21 comments… read them below or add one }
As long as we’re limiting ourselves to investing in stocks, where there is little reason to expect one stock to outperform another, you’re right that there is no such thing as over-diversification. However, some investors buy real estate, commodities, or even collectibles with the justification that they are diversifying, but they ignore the fact that different asset classes have different return expectations. The benefits of diversifying can make it a good idea to buy an asset class with slightly lower return expectations than stocks, but if the return expectation gap is too large, the diversification benefit isn’t enough to close the gap.
@Michael: Great point. I like to ask the question, “Why am I adding this asset class to my portfolio?” One reason is to lower volatility because of low correlation with other asset classes: this might lower expected returns (bonds or cash), though not necessarily (foreign stocks). The other reason is that it should raise my expected returns, even if that means an increase in volatility (emerging markets). If you’re adding an asset classes that does neither, then that’s bad diversification.
Commodities and collectibles could fit into that category. Another example might be global bonds without currency hedging: these may increase volatility with no corresponding rise in expected return.
Congratulations. You’ve managed to turn my one free lunch into the most vile sounding dessert I could imagine. Chunks of broccoli? Ugh. I almost always enjoy your writing Dan, but this time I threw up a little in my mouth.
@Raman: On second thought, maybe that wasn’t the best metaphor. Spinach, maybe?
On of the ironies of the Investopedia article you linked to, CCP, is its failure to admit that if one were as smart as Warren Buffett, one might have a sufficiently informed opinion about many more stocks. But since most people don’t have the time to be this well informed, Buffett also supplies us with the insight that the average investor should be in a broadly diversified index fund. I guess Buffett, like Marx and the Bible, can be selectively understood too.
Regardless, I thought it was more important to diversify across types of investment (fixed income, domestic and foreign equity) than to worry about what was contained in each of those groupings.
@JAH: I’ve always found it interesting that stock pickers try to emulate Warren Buffett. For at least 15 years Buffett himself has been encouraging individual investors not to try what he does and to use index funds. It’s not a matter of people not having the time to be well informed: it’s that we don’t have the skill or the resources that he has at his disposal. Your chances of being the next Warren Buffett are about the same as being the next Wayne Gretzky.
A passive investor should invest across asset classes and within asset classes. The difference is that you only need perhaps three to six asset classes in a portfolio, but you may need hundreds or thousands of securities within each of those asset classes to get proper diversification.
Too not overdiversify, I’ve been using iShares sector ETFs to beat the Index.
Index:
XIC -3.84% YTD -TSX
I use T/A to choose which sectors to be in or out.
Sectors:
XRE +5.34% YTD – Reits
XHC +4.61% YTD – Health Care
XST +4.53% YTD – Consumer Staples
Returns were as of June 15, 2012
iShares has a print add out now about diversification touting REITs XRE, commodities CBR, and global infrastructure CIF. These are the kinds of things that pension funds hold along with a healthy amount of real return bonds (and some have timberland). Also pension funds have been increasing exposure to private equity concerns and hedge funds because of low yields.
Question: does a really diversified larger portfolio require more asset classes?
I am thinking beyond a range of duration bonds both government and corporate, and the major world equity indexes:
preferreds, high yield bonds, emerging market bonds, domestic REITs, Global REITs, infrastructure, MLPs or their equivalent (like high yield energy – I have seen some pension funds run high yield MLP type portfolios in energy), munis (for US investors) ,real return bonds (TIPs), small caps and value tilt indexes and perhaps the use of momentum strategies.
I would imagine it would only make sense to have such broad diversification with a larger portfolio. Is it even necessary in such a case? Just wondering what the family office or the Oxford Club type big guns do with their portfolios and if it has any relevance to CP strategy if balances get larger some day.
@Andrew: First let me make the point (which I know you’re aware of) that my post was about individual stocks within an asset class, not about multiple asset classes within the overall portfolio. Clearly there is a big difference. You can’t have too many stocks, but can you have too many asset classes? Absolutely.
Remember that institutional investors are operating in a different universe. Rick Ferri and Larry Swedroe have written about how asset classes must not only have positive expected returns and low correlation with others in the portfolio, but must also be practical. Private equity, timberland, MLPs, infrastructure etc. may all be legitimately good investments for endowments and pension funds with billions to manage, but you and I cannot get access to them in a cost-effective way. (Buying a timber ETF does not give you the exposure that Harvard and Yale get by owning enormous tracts of land.) So we should just ignore them.
Let’s remember that there is a danger of making portfolios more complicated than they need to be. Once you have, say, 10 asset classes in a portfolio, it is hard to imagine that you could add anything else that would make a significant difference in terms of lowering volatility or raising expected returns. One has to ask why you would make things more complicated for differences that will, in a best case scenario, amount to a few basis points over the long term, and are just as likely to impair your performance.
Looove the last sentence…awesome!
Dan I was mentioning the other asset classes because they are supposed to increase risk adjusted return by lowering correlations but some of them are more or even highly correlated to the equity market (and bond markets) in certain financial conditions.
I get the idea of practicality but was wondering why the big guns would include such types of investments. I also get the idea of excess complication for marginal gain (or being worse off because of management issues).
I bucket money in 4 classes; as cash and cash like instruments, non and low equity correlated, equity correlated and absolute return. I would include indexed pensions in the absolute return including CPP/OAS. What I find however is that a lot falls under the equity correlated and sometimes certain classes, like preferreds, act like equities and sometimes they act like bonds. Some equities, like low beta common, also behave differently than broad indexes. And we didn’t talk about other equity correlated such as rate resets, floaters, debentures, warrants, sector rotation strategies and various mechanical options strategies like managed futures (I think theres a Canadian ETF for that even!). So many equity correlated investment classes if you look on iShares site.
DanB – I’m late to reading your post and to this discussion but you are on the money again with this piece. I would like to address a couple of the comments above.
On the issue of how big the return gap can be before diversification loses its benefit, I think it’s hard to generalize. One way to look at this, however, is simply to require diversification to improve risk-adjusted returns. I’m going from memory here but ten years ago I took a close look at some of the hedge fund index data and I did a little test in diversification. The “short bias” hedge funds at the time boasted something like a 2% per year total return (over ten or so years). This was a far cry from the standard S&P 500 which boasted returns of 10%-12% annually.
Despite the huge return difference, adding as much as 30% short bias to a long-only index portfolio would have done wonders to risk adjusted returns. This was because of the “short” hedge funds’ negative correlation (and beta) to the index but a positive return. And while this was all based on history but it demonstrates that it’s the combination of returns and correlations that will determine the diversifying power.
On the notion of using institutional ideas like real estate, infrastructure and the like for diversification, I just wrote a blog post about this. And you are right that it just hasn’t worked.
http://thewealthsteward.com/2012/06/desperately-seeking-income/
@CCP:
This question is only indirectly related to “overdiversification” — suppose that in your attempt to spread the asset classes around, you inadvertently achieve some overlap? For instance in a hypothetical TFSA where the proceeds are projected to be not required for more than a 20 year horizon, so that the investor can be relatively fearless and aggressive in “Park and Keep”, with yearly $5000 contributing and rebalancing.
Suppose we aimed for 30% DEX Corporate Barbell (KXF), and tried to split the remaining 70% equally into Canadian Small Cap Equities (e.g. iShares XCS), Unhedged US Small/Mid Cap (e.g. Vanguard VXF) and Unhedged All-world ex-US Small/Mid Cap (e.g. VSS), which also captures emerging markets.
The above allocation seems to be complicated by the presence of Canadian Equities in the last mentioned ETF, which was intended to capture the world minus USA, but the Canadian investor seems to be so inconsequential as to not warrant consideration when designing these ETF’s. So VSS includes 16.7% of its content listed as “North America”, which by default must mean Canada. In fact checking the top ten listings, the largest 8 are Pembina Pipeline Corp, SXC Health Solutions and Franco Nevada, Viterra, Metro inc, Baytex, Progress Energy Resources and Aberdeen Asset Management, (respectively Canadian, Canadian startup, but headquartered in Illinois, Canadian, multinational but largely Canadian holdings, Canadian, Canadian, Canadian, Canadian, Canadian).
Does this create an overweighting in Canadian content? Or, rephrasing my question, is this overlap enough to worry about? If so, can it be dealt with by just eye-balling the overlap and adjusting the ratios from the original 70/3% each to, say, 20%, 25%, 25% (assuming that about 1/6 of the last asset is actually Canadian content, and 1/6 x 25% is about 4% and change)? Looking back at my figures, I now realize the difference is really small change; but as an exercise, especially in preparation for a future situation where the overlap might actually be significant, is my solution sound?
@Oldie: I would say that the overlap caused by the Canadian content in VSS (or VXUS, for that matter) is likely to be trivial. However, one solution might just be to forget about XCS and use only VSS for global small-cap coverage.
@CCP:
Yeah, I thought so, from the viewpoint of getting the right amount of Canadian Equity. However, given that the other 2 Equity blocks are in non-hedged US dollars, does 20% Canadian Equity EFT priced in Canadian Loonie added to the 30% Bond Loonie-priced ETF make for more sensible overall currency balance?
@Oldie: Not sure if you are aware of this, but holding Canadian stocks in a US-denominated ETF does not give you exposure to the US dollar. See the last section of this post:
http://canadiancouchpotato.com/2011/04/04/currency-hedging-in-international-funds/
This confuses everybody.
@CCP:
Yes, thanks for reminding me what you had covered in an earlier seminar! From what I remembered, the easiest way to understand the principle was to consider only our currency at this end i.e. the Loonie, and then the initiating currency at the other end, ignoring the intermediary currency/currencies. So the All-world Ex-US EFT’s exposes us Canadians to whatever mix of currency the underlying stocks of the “foreign” index are sold in, including our own Loonie (which makes those stocks a wash as far as currency goes). For exactness, though, would the VXF portion of the portfolio count towards US$ exposure?
@Oldie: Yes, VXF is 100% US dollar exposure.
@CCP:
To return to your original suggestional tweak to my hypothetical long horizon TFSA portfolio:
“forget about XCS and use only VSS for global small-cap coverage”
I see that it has the merit of dispensing with one separate asset class (Canadian Equities) to manage and pay for in trades and periodic rebalancing, while still capturing Canadian Equities as 16% of the Global Except US Small Cap Asset Class VSS. This hearkens to your earlier “Does Home Bias Ever Make Sense” (May 22, 2012) post; the refined portfolio even without a specific Canadian Equity ETF has more than enough globally proportioned Canadian weighting to be rationally effective regarding diversification over the long haul. This makes really good sense.
Regarding diversification, I know that you have previously discouraged holding unhedged US Bond ETF’s in a portfolio on the basis that despite being Bond based, it increased volatility due to the superimposed factor of currency fluctuation which overwhelmed the dampening effect of being a bond fund.
What if the investor was prepared to accept the increased volatility of the overall portfolio for the time being. Looking at the unhedged US Bond ETF and unhedged US Equity ETF (say in a 40:60 ratio to each other) as an integral subunit, say, a mini self-contained US Couch Potato module buried within your envelope CP portfolio. Your yearly or 6 monthly Bond-Equity rebalancing would be done internally, between these two US funds, without the need for intermediary currency transactions (Unless, I suppose, the total combined US$ amounts varied enough from your original prescription to warrant an injection of Canadian Loonie converted in, or Extraction of US$ converted out to balance) . Would the accumulated nickel and dime savings over 20 years more than compensate for the increased short-term currency volatility caused by the addition of the US Bond ETF?
I ask because my TSFA does not allow holdings of US dollars as cash in the account. I am aware of your previously described strategy of doing all the currency exchange and US ETF purchasing external to the TFSA and then swapping in kind into the fund, but that seems like a lot of work if it can be avoided conveniently, especially on repeated occasions in the next 20 years.
XIC is mainly Financial, Energy and Materials (74%) Heath is only 2.2%
XSP has 13% Health. Do you think would it be worth adding extra Health care ETF (XHC) to the portfolio, since aging and rapid, new medical discovery.
Add 5% XHC?
Peter
@Peter: There’s no question that XIC is poorly diversified across sectors. But I feel the best way to improve the situation is to keep only about a third of your equity portfolio in Canada and have significant holdings in US and international equities.
{ 4 trackbacks }