Higher Stock Market Concentration Does Not Mean Higher Risk

Diversification is the only free lunch in finance. Does that mean that the increased concentration of the U.S. stock market is taking away our free lunch? No. Higher concentration in the stock market does not obviously imply higher risk.

Here’s the concern, as posed in a recent installment of the excellent series of University of Chicago surveys of eminent economists:¹

With some measures of concentration by market capitalization within broad US stock market indices at an all-time high, investors seeking a well-diversified passive equity portfolio should consider alternatives to market-cap-weighted indices.

Now, as I’ve argued previously,² this is actually a trick question; my answer would have been that the premise is false because most concentration measures are nowhere near an all-time high. I regret to report that none of the respondents answered this way.³ But let’s consider the logic of the question. Is high concentration obviously a bad thing from the perspective of shareholders?

No. Concentration by itself is a poor measure of risk, and you should not be making investment decisions based only on concentration.

I’m open to the idea that the market is quite risky today. After all, it appears we are about to undergo a major economic transformation driven by AI, one that will greatly impact existing publicly traded firms, with both huge upside and huge downside possibilities for stock market investors. But this risk has little to do with stock market concentration.

I’m also open to the idea that the US stock market will experience something similar to the 1999 tech-stock mania, and perhaps we’re already in the initial stages of a speculative bubble. There are many symptoms of bubbles, but stock market concentration is not one of them. Right now, the main bubble currently occurring in markets is a bubble in irrationally exuberant commentary about the Magnificent Seven; in the past 12 months, the FT has mentioned this term more than 211 times.⁴

Maybe we are entering a new era of AI-driven markets requiring bold new thinking. But when it comes to stock market concentration, we are certainly not in a new era, and what we really need is bold old thinking from the 1950s: portfolio theory.

Let’s think it through. Suppose you have 100 assets and you construct a portfolio from these assets. Are more concentrated portfolios always riskier? No, obviously not. What if the 100 assets consist of T-bills plus 99 stocks selected by a committee of crypto bros, TikTok finfluencers, and convicted felons?⁵  Wouldn’t it be less risky to hold 99% of your wealth in T-bills, despite the highly concentrated nature of the resulting portfolio?

You can’t just look at portfolio weights and immediately intuit the portfolio risk without knowing something about the nature of the assets. What’s important is not the weights, but the weights in combination with the variances of the returns and the covariances of the returns with each other. That’s portfolio theory 101.

For those of you who know matrix math, portfolio risk is w’Ωw (for those of you who don’t know matrix math, just fake it). Notice how you cannot just look at w (the weights) and make any statements about risk, because there is this other thing in there that must be important because it is a Greek symbol.

OK, forget matrix math, let’s consider another example. Suppose I have two portfolios. The first portfolio consists of all the individual stocks in the S&P 500, held using market cap weights. The second portfolio consists of a single security, thus this portfolio is 100% concentrated. Is the second portfolio riskier than the first portfolio? Not if the second portfolio consists of SPY, the ETF for the S&P 500. The two portfolios have identical risk.

Consider AT&T.* After many years of legal battles, the U.S. Department of Justice broke up the monopoly and forced AT&T to split into seven independent firms. On December 31, 1983, AT&T was the second largest stock in the U.S. stock market. On January 1, 1984, “Ma Bell” ceased to exist and was replaced by seven “Baby Bells:” Bell Atlantic, Bellsouth, etc. As a result, measured concentration in the US stock market went down overnight. But did this action, however good for economic efficiency and consumer welfare, really make the stock market safer for investors holding the market portfolio? Did diversification really go up? I don’t think so. Indeed, you could argue that a monopolistic phone company is less risky for its owners than seven competing Baby Bells.

So, why doesn’t risk change much in 1984? Because the risk of the stock market reflects two things. First, the risk of the underlying fundamentals. Second, the risk induced by possible mispricing, that is, departures of prices from fundamentals. If you are holding the value weighted stock market, those are the risks you are exposed to. The risks don’t necessarily go up or down just by bundling or unbundling securities into different combinations.

Speaking as a former University of Chicago professor, I will put on my efficient markets hat. The price of AT&T reflects the fundamental risk of the telephone business. Breaking it up into pieces only changes the stock price risk if it somehow changes the nature of the operating cash flows. That’s the logic of the Modigliani-Miller framework: you can’t make risk disappear just by splitting it up into different components; the risk has to go somewhere.

Now let me take off my efficient markets hat, since I am a former University of Chicago professor. Could the breakup of AT&T cause greater or lesser mispricing? Maybe the increase in the number of securities somehow changed the nature of the market? Maybe it will unleash a frenzy of speculation? Sure, I guess, in which case decreased concentration would cause total risk to go up, not down. Even in an inefficient market, concentration does not measure risk.

Let’s do a thought experiment. Suppose a newly invigorated Department of Justice starts an anti-trust crusade, and forces each of the Magnificent Seven to break up into seven independent firms. Amazon is broken into seven Baby Bezos, while Microsoft is broken into seven micro-Microsofts. So now instead of the Magnificent Seven we have the Magnificent 49. Hocus pocus, abracadabra, presto chango, the stock market is now less concentrated as measured by the weight of the seven largest components. Is the stock market now obviously less risky and more diversified than before? I don’t think so.

Amazon is already an extremely diversified firm, diversified across geography, business lines, and technology. The seven Baby Bezos would include a digital commerce firm, a cloud computing firm, a movie studio, and so on. So if the stock market is currently assigning higher weight to economically diverse firms like Amazon, it could well be that today’s higher stock market concentration reflects lower economic risk and lower total market risk.

One argument against the Magnificent Seven has to do with risk associated with specific individuals. Here’s the FT’s Unhedged podcast:⁶

If you’re buying the S&P 500 you are like, almost by definition taking a huge mega super long position on like the fate of Microsoft and Google and Meta and so forth… your retirement fund like depends on how Satya Nadella runs Microsoft, right… Is market exposure with highly concentrated markets really market exposure properly conceived of?

I agree that there is risk in having one individual in charge of a substantial chunk of market cap. But let me make two comments. First, whatever your concerns about the CEO of Microsoft, he controls far less of the U.S. economy and far less of the U.S. stock market than the CEO of GM did in 1953. He probably has less political power than the CEO of GM, who in 1953 became Eisenhower’s Secretary of Defense and was widely misquoted as saying “What's good for General Motors is good for the country."

Second, if you are concerned about the risk of specific individuals, market concentration is not the best way to measure that risk. Consider Carlos Ghosn, at one time the head of three different auto companies (Renault, Nissan, and Mitsubishi) and currently an international fugitive. Having separate, publicly traded companies is no guarantee that your firm is not exposed to Ghosn risk. Similarly at one point, Jack Dorsey was the CEO of two publicly traded companies, Twitter and Square.

Elon Musk currently controls Tesla, The Boring Company, SpaceX, Neuralink, The Microblogging Site Formerly Known as Twitter, and I don’t know how many other companies. Let’s take two scenarios. First, suppose they were all publicly traded as separate stocks. Second, suppose they all merged into one giant conglomerate. Do you think a stock market containing separate Elon-controlled stocks really has a lot less Elon risk than a stock market with a single large Elon-controlled conglomerate? Elon’s gonna Elon, in either scenario.

What do we learn from this Elon example? One key to understanding portfolio risk is covariance. Reducing concentration is not helpful unless it reduces weighted covariance. Risk doesn’t come from portfolio weights, it comes from portfolio weights interacting with returns.

Let me try one last comparison. Suppose you have invested all your wealth in a house but you are worried that it will burn down. So, you split the house into 3 condos, all of which you own. Does that reduce risk? No way. If there is a fire, the flames don’t care that you’ve arbitrarily relabeled parts of the house to reduce your portfolio concentration. The returns from the condos are all correlated with each other, and fire risk is not reduced.

I hope I have convinced you that concentration does not measure risk. The right way to measure risk involves return volatility, return covariances, and firm-level characteristics that predict future volatility (such as price relative to fundamentals).

I assure you that I’m not an efficient market zealot who believes that market cap weights are always best. There are many smart ways to deviate from market-cap weights, such as overweighting safe, cheap stocks in a risk-aware manner. But concentration is not the key defect of market-cap weights.

So let me go back to the survey question. Another way to answer would be: Yes, investors should “consider alternatives to market-cap-weighted indices.” But that statement is true not only when the market is concentrated, it is also true when markets are not concentrated. That statement is true every day of the week.

* Any mention of a company in this writeup is for illustrative purposes only and is not a recommendation to buy or sell a specific security.