What exactly is your problem with stock index concentration?

Ever since artificial intelligence was invented eight months ago, people have been writing about the rising concentration of stock indices. The usual way to start is with an observation that the trillion-dollar IT club makes up approximately a quarter of the S&P 500 index by weight — because of AI hype, or possibly, the result of a rotation by investors after the same stocks sold off a year earlier.

Everyone says this much concentration is bad, because obviously. But there’s not much agreement as to the specific reasons for that badness. Here’s one suggestion, from Liberum strategists Joachim Klement and Susana Cruz:

In a more concentrated index, stock pickers have a harder time outperforming the index because outperformance becomes more and more a question of timing the performance of the largest stocks rather than selecting the best performing stocks.

And if you’re a stock picker, fair enough. Fund managers tend to be underweight the heavyweights and overweight other stuff. When funding positions, there’s a necessary trade-off between benchmark hugging and alpha generation. So, for example, an active manager with convictions might this year have been buying Coinbase with cash they freed up by selling Nvidia.

Equity-market wealth creation comes from a tiny proportion of winners, as a general rule. And in a time like this, when passive and lazy money is flowing into value-weighted indices that are already top heavy, any fund manager with convictions is pretty much guaranteed to underperform. Liberum does some numbers:

Nvidia contributed 4.9% of the 18.3% total return of the S&P 500 in the 12 months to July. Add the next four contributors (Apple, Microsoft, Meta, and Broadcom) and investors end up with two-thirds of the return of the entire index simply because the largest of the large stocks in the S&P 500 had a good year. Meanwhile, the S&P 495 (ex the 5 largest contributors) generated a return contribution of 5.2% in the 12 months to July or roughly the same as Nvidia alone.

The situation is similarly striking for the FTSE 100 where HSBC accounts for 1.5% of the 9.2% total return of the last 12 months. Add the two oil majors, Flutter Entertainment and 3i and one again has explained more than half the total return of the index.

Other arguments are available for why stock-market concentration might be bad. Here are a few:

Maybe an index lacking diversification is more volatile, or at least overly exposed to single themes? It can be argued that the size of Apple, LVMH and AstraZeneca relative to their host markets mean US, European and UK equity trackers are disproportionately reliant on sales of iPhones, handbags and treatments for non-small cell lung cancer, respectively. Overall, though, there’s not much evidence that volatility rises or falls in tandem with concentration: papers from 2008 on the FTSE 100 and a 2022 on the US+Brics can find no link.

Maybe an index of superstars and also-rans is a symptom of weak competition? As policymakers and academics tackle the broader topic of corporate concentration and antitrust enforcement, is it useful to revive the once-popular conceit that the stock market is an economic proxy?

For example, here’s a 2021 paper that finds: “stock markets dominated by a small number of very successful firms are associated with less efficient capital allocation, sluggish initial public offering and innovation activity, and slower economic growth.” And here’s another that finds no evidence to support this idea, saying the premise “incorrectly defines markets and competition, excludes most competitors, ignores important market dynamics and leads to erroneous results.” So, mixed findings.

Maybe the clustering effect drives the biggest stocks to overvalued levels? This is often claimed but is tricky to test. Hindsight makes it easy to argue that General Electric, for example, probably shouldn’t have held on so long to the title of S&P’s 500’s biggest stock. But markets are best considered efficient until proved otherwise and many thousands of broker notes are available that will give reasons why Apple, LVMH and AstraZeneca remain screaming bargains. If anyone has a failsafe way to identify which stocks are too cheap or expensive, please get in touch.

Maybe a narrow rally is inherently more fragile than a broad one? This sounds reasonable. The durability of herd behaviour rarely gets a positive press irrespective of its breadth, and relevant research we’ve found focuses mainly on the short-term outlook for returns. But OK, fine.

In that case, how should we be defining narrow? As Liberum data shows (and as SocGen showed earlier in the week), the UK and European markets are fairly concentrated relative to history, but not quite at the US’s “unprecedented” levels. And the levels of concentration that came before the last three market crashes seem unremarkable:

Maybe increased stock-market concentration simply doesn’t mean that much? Is it possible that becoming top-heavy just sort of happens as an index grows old? That’s the premise of a recent paper from Lisa Goldberg, of BlackRock’s Aperio research group.

Goldberg’s hypothesis is that big stocks keep getting bigger because they’re big: the same power-law distribution effects that apply to things like wealth inequality and urban population growth are also relevant for stock index constituents.

Using a Zipf distribution — the fat-head-thin-tail flavour of power law that killed a million dotcom start-ups — imbalances in weighting really aren’t that extreme, she argues. Or rather, they weren’t 18 months ago:

Maybe, instead of getting bogged down in definitions of what’s normal, it’s perhaps useful to remember that value is one of many ways to see the world. It just happens to be the one the financial industry prefers.

Value-weighted indices are great for passive investments because rebalancings are few, so dealing costs are low. In contrast, it makes more sense for active managers to use an index that weights each member equally, because this more closely resembles what stockpickers are meant to be doing. A 2014 article from S&P Global makes the case that “since equal-weight benchmarks reveal what can be achieved by a random, meaningless selection and weighting process, any genuinely alpha generating process should beat such a benchmark.”

But equal-weight benchmarks are rarely used because they’re too efficient. Active managers, given the choice of failing to hit a challenging target or an impossible one, choose to play on easy mode. The charts below are from S&P’s Happy 20th Birthday letter to the Equal-Weight Index.

An equal-weight index represents its average stock. A cap-weighed index represents a dollar invested in the market, which is usually the less challenging measure because the average stock usually beats the index. That doesn’t mean most stocks beat the index (they don’t) or that most active fund managers beat the index (they don’t). It just means that most years, random luck would be working in their favour.

This held true for a long time. It was once possible, as a famous study found, to pick a portfolio at random from S&P 500 members and outperform the index with a 99.9 per cent probability.

But momentum has mattered more than value since early in the pandemic, which interrupted the usual helpful interplay between small and large caps, and meant equal-weight indices stopped outperforming.

Concentration is the result. It doesn’t really matter much if it’s because of IT hype or the death throes of a liquidity-fuelled rally. What matters is that a thousand monkeys with dartboards have been finding it increasingly difficult to chance upon assets that might beat their preferred benchmark (before fees, friction, etc) and, as Liberum correctly notes, they don’t like that one bit.