Concentration over Representation

Standard Statistics Co & Poor's Publishing Co merger back in 1941 had a lot to do with the American Railroad Journal managed by Varnum Poor. The merger increased the number of the list of stocks to 490. This was a significant jump and leapfrogged the merged entity into the top spot among its competitors. But the Indexing company's leadership had little to do with its underlying methodology of building a benchmark. The 1871 Laspeyresian bias and calculation convenience still echo in the market capitalization method. The bias of overweighting the winner (with every price change) leads to concentration in our Indexes, references, and benchmarks, a real challenge for modern finance, and one which is intrinsically linked with the Google Paradox.

In 2014 Google decided to issue a new class of shares. The indexing company decided to include both GOOG and GOOGL in the Index, choosing to increase the number of components in the S&P 500 to 501. Google’s A, B, and C class of shares is about Google founders wanting to keep voting control over the company. Every founder worries about the control of their startup and more so when it becomes the world’s most important company. My thoughts are not about governance, activism, or reigning big tech but how it affects our Indexes, the ones we believe are sacred and unassailable. S&P had some difficulty deciding as it had to think through the integrity of the indexing process. 

“Per the S&P U.S. Indices Methodology, companies that have more than one class of common stock outstanding are represented only once in an index. The stock price is based on one class, usually the most liquid class, and the share count is based on the total shares outstanding. An investable weight factor is applied to ensure that only the publicly available share float is included in the index…

…S&P Dow Jones Indices' US Index Committee reviewed Google's stock split with major index fund managers, index traders and other market participants. Based on this review, the US Index Committee chose to temporarily include both share classes in the S&P 100 and S&P 500 to reduce the risk of possible market volatility related to distribution of the Class C shares and the resulting index adjustment.“

S&P DJ Indices Press Release - Feb 3, 2014

An Asset Owner and Investor would like to have a say in how the company is run, be it ESG issues, or other issues that accompany the role of large-cap companies in societal impact and governance. For S&P 500, the paradox was whether it should have chosen concentration above representation. If it takes both classes of shares, it increases the concentration of Google in the S&P 500, and if it takes one of the classes, it favors representation over overweighting a component.

If the Indexing company would have chosen representation, it would have sent a message to its leading holdings that if they followed the Google way with more classes of shares, they will be underweighted accordingly. But then winners are contagious, everyone wants them, as they create a snowball effect, which makes the prices go up. Why would an indexing method that epitomizes winner amplification want to choose representation over concentration? Not so surprisingly, the S&P Indexing team chose to override its methodology and go for the concentration (overweight) alternative.

Consequently, this created a bigger predicament for the Indexing company. The top 10% of the S&P 500 companies. i.e. 50 stocks own 51% of the S&P 500 value. If this set of companies assume the Google precedent and start issuing new classes of shares (some of them have done so) to get voting control or some other corporate objective, they know the next steps. S&P 500 will choose concentration above representation, which means there is a probability that accommodating these new share classes would mean overweighting what’s already overweight at the expense of the other not-so-important and underweighted 450 companies, reaching to the super concentrated stage, where 10% of the components own 80% of the S&P 500 value. If this would happen, S&P 500 would become one of the most concentrated passive portfolios in the world. Passive nomenclature would seem misfit in such a case.

One would wish that the risk of the concentration problem was limited to our Indexes. Unfortunately, Index funds are designed to mirror the Indexes and hence follow the overweighted winners. Eventually what starts as a small flap from a butterfly, becomes impossible to control risk, which travels from the method to the index, to the ETF, into our pensions and our economic life, making us wonder, how did our ETFs become so risky? How did we not see this all coming? How did we miss all this bias, risk, and erroneous mathematics of overweighting the bias as it becomes bigger?

The S&P 500 market capitalization weighting method is not designed for fair representation, it’s a way for everyone to ride the emotional bus called mass hysteria aka running faster and faster after the winner, creating irrational exuberance when sentiment leaves value behind and starts imagining its journey to the moon. Concentration is like sugar, intoxicating in the short term and wealth-destroying in the longer term.

"Concentration is like sugar, intoxicating in the short term and wealth-destroying in the longer term"

The AI machines that are expected to take over stock markets and make global asset owners and investors rich beyond their dreams are destined to fail if they don’t grasp the harmful effects of selection and concentration. Before we think of intelligence and conscious machines, we have to confront the banality of our economic assumptions. Challenging an assumption is where common sense begins. Amplifying the winner effect is not mathematics, it’s a herding mechanism, which can never be an optimal way to represent a market. Fiddling with rules is a naive way to fix a legacy problem.