3. Passive flows: what they do to the price, and not

Level 11: The structure of the index

This is the module where the most nonsense gets talked in both directions. Part of the industry maintains that indexing has broken price formation and is inflating a mechanical bubble. The other side replies that index funds barely trade and that the price is still set by active investors. Both positions handle correct data and reach opposite conclusions, almost always because they confuse two different quantities.

Ownership and volume are not the same thing

This distinction is worth half a module:

• Ownership is a balance: what fraction of the shares in existence is in the hands of index vehicles. It is a balance sheet figure.
• Volume is a flow: what fraction of a day's trades originates from an index vehicle. It is an activity figure.

The magnitudes are very different. Index ownership of the US stock market has grown spectacularly and today is counted in a high double-digit percentage of the market, with estimates varying a great deal depending on whether you count only index funds or also institutional portfolios managed against an index. By contrast, those vehicles' participation in daily traded volume is estimated at a small fraction, on the order of low single digits, because an index fund only trades when money comes in or goes out, when the composition changes or when there are dividends.

Someone wanting to prove that indexing dominates the market cites ownership. Someone wanting to prove it is irrelevant cites volume. Both are right about their figure and neither has answered the question, because the price is set by whoever trades, but the willingness to trade depends on who owns.

The argument that it does distort

Jeffrey Wurgler has spent years documenting the economic consequences of indexing. The line of the argument is this:

• When a security enters a widely tracked index, it starts moving more to the rhythm of the index and less to the rhythm of its own news. Comovement appears that did not come from the business.
• If a growing share of capital buys without looking at the price, demand stops being price-sensitive in that segment, and anyone wanting to sell has to move the price more to find a buyer.
• Xavier Gabaix and Ralph Koijen formalized this idea with the inelastic market hypothesis: their estimates suggest that a dollar invested permanently in the equity market raises aggregate value by a multiple of several dollars, on the order of five. If that is approximately correct, flows matter far more than the classical view assumed.

The argument that it does not

The counter is just as solid and cannot be waved away:

• An index fund is a price taker, not a price setter. It buys what it is told to buy, when it is told, and usually does so at the close. It expresses no opinion.
• The price is set at the margin: the last trade between someone willing to sell and someone willing to buy. And the vast majority of volume still comes from participants who do look at the price.
• If indexing created systematic, detectable distortions, correcting them would be an extraordinarily profitable business, and there is a great deal of capital devoted to looking for exactly that. The more obvious the supposed distortion, the faster it gets eaten.
• The Gabaix and Koijen multiplier, if it is correct, applies to any flow, not only the indexed kind. It is not a property of indexing: it is a property of a market with inelastic demand.

What can be stated with caution

Three defensible conclusions, far more modest than the headlines:

• Flows move prices more than classical theory said, and that holds for all flows.
• Indexing changes who is left on the other side and when they trade: it concentrates activity on particular dates and moments, above all in the closing auction.
• The argument about whether 'there is too much passive money' has no known numerical answer, and be suspicious of anyone who gives you an exact figure for the point beyond which the market stops working. What is true is that, if the search for correct prices is done by fewer and fewer people, the reward for doing it well would tend to rise, and that rise would attract capital back. It is an equilibrium with feedback, not an unchecked spiral.

And there is a fourth, more uncomfortable one: the effect need not be uniform. In the index's largest holdings, with billions traded every day, the indexed presence is a drop in an ocean of activity. In the index's smaller constituents, where indexed ownership can represent a very high fraction of the free float and daily volume is modest, the same flow weighs a great deal more. Talking about 'the effect of indexing' in the singular hides the fact that it is probably negligible at one end of the index and perceptible at the other.

None of this licenses saying the index is inflated or trading accordingly. It is analysis, not a recommendation.

Real case: the morning of August 24, 2015

That Monday, the US market opened with a violent fall after the yuan devaluation. In the first few minutes, the S&P 500 fell on the order of 5% and trading halts multiplied: more than a thousand halts in securities and exchange-traded funds were counted in a single morning.

The interesting part is what happened to many exchange-traded funds. Several of them, including some tracking broad and highly liquid indices, traded at the open far below the value of the basket of shares they contained, with deviations that in some cases exceeded 20%. It was not that the companies were suddenly worth 20% less: it was that many of their shares were not open for trading and the market makers, unable to value the basket, widened their spreads or withdrew.

Within an hour, most of those deviations had closed. The conclusion is not that index vehicles break the market, nor that they are harmless. It is more precise and more useful: the wrapper and the contents can separate when liquidity disappears, and that separation resolves itself when liquidity returns. Anyone confusing that mechanical episode with the market's judgment about the value of five hundred companies will be watching the wrong film.