How much diversification is enough?
Updated: September 1, 2015
Mom told us to dress warmly, eat well, wear clean underwear and always hold a diversified investment portfolio. Well, the first three were certainly right up there on most mothers’ lists of instructions – maybe only financial planners’ mothers mentioned the last one.
You can think of diversification (buying several stocks) as putting your eggs in several different baskets. But, how do you know that you’ve picked enough baskets?
It gets worse. Some stocks are so different from each other that we classify them into distinct asset classes. Examples are US Large Capitalization (e.g., the S&P 500), International, and Commercial Real Estate. You can think of stocks as little baskets, fitting into the bigger baskets that are asset classes. Do you need even more stocks to cover the various asset classes?
Intuitively, you might think that buying 10 technology company (tech) stocks doesn’t offer as much diversification as buying 10 stocks from different industries, say 1 tech stock, plus the stocks of one automobile manufacturer, a bank, an airline, an oil company, a shoe marketer, an entertainment company, etc.
But, how do you know 10 is enough? And, how do you know the 10 stocks are “different” enough?
And what about mutual funds? Don’t they provide built-in diversification? Can’t just a few mutual funds be enough?
By now, you’ve read enough questions to deserve some answers. After our answers, there’s more discussion to look at if and when you have more time.
- How many asset classes? Sensible Financial™ uses seven equity (stock) asset classes: US Large Cap Growth and US Large Cap Value, US Mid Cap, US Small Cap, US Commercial Real Estate, International Equity, and Emerging Markets Equity1.
- How many stocks per asset class?
- You need 10 to 20 randomly selected stocks in each equity asset class to get most of the diversification benefit available from that class.
- You can obtain the full amount of diversification available from any asset class by buying an index mutual fund that holds all of the stocks in that asset class.2
- How many stocks in all? You would need to hold 70 to 140 stocks to come close to taking advantage of the diversification opportunities available in equities.3
- It’s a lot of work to hold 70 to 140 stocks. You have to decide
- which stocks to buy;
- how much of each stock to buy; and
- how to rebalance the portfolio – how much to change your holdings of each stock to balance risk and return as the prices of the stocks change.
That’s a good deal of work just for asset classes. For individual stocks, it’s close to a full-time job. It’s a lot of work to hold 70 to 140 stocks. You have to decide
- It’s expensive to hold 70 to 140 stocks. Start with commissions 10 to 20 times the commissions required to buy one mutual fund for each asset class. Commissions for rebalancing are higher, too.
- It’s a lot of work to hold 70 to 140 stocks. You have to decide
Those are the answers. And now, if you want more detail, we’ll move on to the discussion we promised.
Three factors enter into your asset allocation decision. Readers of these newsletters are already familiar with two of them, return and risk. The third factor is correlation, the tendency of returns to move together (think “co-relation”). Diversification means allocating your resources into assets with uncorrelated returns (returns that move independently of each other), so that when one asset goes down, the other may go up, or when one goes up only a little, the other may go up a lot.
Diversification changes the relationship between return and risk in your portfolio. You know that to target more return, you have to accept more risk. By diversifying, however, you can target more return with the same risk, or target the same return with less risk. In a diversified portfolio, the uncorrelated risks offset each other, leaving you with only the asset class or undiversifiable risk, the risk you can’t avoid. By diversifying within an asset class, you get a return close to the asset class return, at a risk close to the asset class risk. An individual asset in that class has the same expected return, but higher risk – the asset class risk plus its own.
Here’s a simple example. Suppose that you can invest in two stocks, each in the same asset class. Each stock has the same expected return (10%), and the same risk (also 10%). Suppose further that you invest half of your portfolio in each stock. If the two stocks are perfectly correlated (correlation = 1), the portfolio risk is also 10% – there is no diversification benefit, because there is no diversification. The two stocks are effectively the same. On the other hand, if the correlation is zero, the portfolio risk is only about 7% – the portfolio risk is reduced by almost a third!4 The expected return is still 10%, though. The same expected return, and less risk – that’s why diversification is so valuable!
Why does it work that way? Think about two uncorrelated stocks, each with the same expected return and risk. If you put half of your portfolio in each, your expected return will not change, (the return of one has no relationship with the return of the other). Your risk will be different, however. Risk is the spread of returns – one way to think of risk is the chance of extremely positive or negative returns. When one of the stocks has an extremely large or small return, the other, being uncorrelated with it, is likely to have a more normal return. Therefore, total portfolio return will be closer to the expected return of each stock more often and the portfolio will be less risky.
Asset classes are groups of securities that have high correlation with securities in their own group, and lower correlations with securities in other groups. You can think about the return on a particular security as having two pieces: 1) the return on its asset class (making the security a member of its asset class) and 2) its own particular special return (making the security different from all the others in its class). The graph illustrates for fictional Stocks A and B.
Each asset class has an expected return, an expected risk, and expected correlations with all of the other asset classes.5 Each security has a return (the asset class return plus its own special return), risk (the asset class risk plus its own), and correlations with all of the other securities in its class. These correlations tend to be relatively large, because every security in the class shares with the others in the class a part of its return – the asset class return.
Two stocks in different asset classes have relatively low correlations with each other (their asset class returns are different), and relatively high correlations with stocks in their own asset classes (their asset class returns are the same). For example, Microsoft and Ford (two US Large Cap Growth stocks) are likely to be more correlated with each other than with, say, British Petroleum (an International stock). And, some closely related stocks within asset classes (say BellSouth and Verizon) have even higher correlations. That’s why you should pick stocks randomly to obtain diversification – you want unrelated stocks, and random selection is more likely to provide them than selecting, say, stocks you might have heard a lot about recently.
Finally, what about mutual funds? Unfortunately, portfolios of mutual funds may be little more diversified than portfolios of individual stocks. Clients sometimes bring us portfolios containing as many as 10 or 20 mutual funds that turn out to cover only one or two asset classes. Once we look inside each mutual fund, and list the stocks that each one holds, we find a lot of overlap. More overlap means less diversification – different mutual funds in the same asset class are the same big basket for diversification purposes (perhaps decorated with different ribbons).
In summary, if you want a diversified portfolio of stocks, you need
- Asset classes different enough from each other (large cap domestic equity and domestic commercial real estate are more different from each other than domestic automakers and domestic utilities);
- Stocks in enough different asset classes (leaving out asset classes misses diversification opportunities – leaves risk higher than it needs to be);
- Stocks different enough from each other inside each asset class (a technology stock and a retail stock, say, rather than two technology stocks);
- Enough different stocks in each asset class (for example, the chart illustrates that ten stocks have much smaller risk than one, but not much more risk than fifty).
There are ongoing discussions about what constitutes an asset class. That’s another way of saying that there is a lack of professional agreement about how many asset classes there are. Sensible Financial’s reading is that the following categories are pretty well established:
- Large and small cap (big and small companies) 
- Growth and value (high priced stocks and bargain stocks) 
- Commercial Real Estate 
- Each of these asset classes could be found in different countries (US, International, and Emerging Markets) 
- Commodities  (a global asset class)
That could lead you to as many as 16 different equity asset classes – (2×2+1)x3 + 1. For example, you’d have Emerging Markets Small Growth, US Large Value, International Commercial Real Estate, etc. In practice, in asset allocation we make decisions based upon the availability of data that allows us to assess the inter-correlations of all of the different asset classes, and a lot of the data is either unavailable, or hard to get. Furthermore, the asset class has to be investable – you have to be able to buy it (I’m not aware of an Emerging Markets Commercial Real Estate Fund, for example). Our assessment of inter-correlation and investability has led us to the conclusions we outlined at the beginning of this discussion – seven or eight equity asset classes.
If lack of correlation is so important and valuable, why not look for other uncorrelated asset classes, beyond stocks? In fact, this is one of the great quests in investment life, like the search for the Holy Grail or the Fountain of Youth. Investment advisors are always seeking assets with uncorrelated returns, but unfortunately they are hard to find.6
In addition, uncorrelated returns aren’t enough. Think about stocks and rocks. The return on rocks is probably pretty uncorrelated with the return on stocks.7 When stocks go down, rocks probably don’t go down very much. Unfortunately, they don’t go up much, either.8 And it would be pretty expensive to hold enough of them to balance out the rest of a good-sized portfolio (what would $100,000 worth of rocks look like, and where would you put them?). So, rocks aren’t likely to be one of the components of your diversified portfolio.
By the way, returns are the major problem for bonds as diversifying assets. Bond returns are relatively uncorrelated with stock returns, which is good. Bond returns are also relatively low, which is not so good.
The bottom line? Diversification is a GOOD THING. It allows you to reduce risk, by mixing the risks of one stock or asset class with the risks of other, relatively unrelated ones. And, it’s easier than you might think. Using index funds as the key components of your diversification strategy:
- Ensures accurate diversification (limits/eliminates overlap among portfolios in the same asset class, and assures you that you are getting the asset classes you want).
- Keeps your investment costs (commissions, research, monitoring, etc.) lower as you hold one index fund per asset class rather than 10 or more stocks.
The downside? When people are bragging at cocktail parties about the killing they made in Integrated SuperSoft and Nuclear BioPlastics, you won’t have much to say unless you are willing to make it up. Index funds are boring. On the other hand, you won’t have to regret the pasting you took in InterPet and WorldWideWebbing.com, either.