What moves an index

Two quantities, moving independently 

An equity index level can be decomposed, at any moment, into the aggregate earnings its constituents are expected to produce and the multiple of those earnings the market is currently paying. The decomposition is an identity rather than a theory: the level is the product of the two, so if the level is known and one factor is known, the other follows. Its usefulness is that the two factors respond to entirely different news, and separating them explains most of what otherwise looks like an index moving for no reason.

The earnings channel 

The first factor moves on evidence about company profits, and the evidence arrives in two forms. Reported results are the direct form: during a reporting season, constituents publish what they actually earned and, more importantly for the index, what they expect to earn next. Aggregate activity data is the indirect form. Output, orders, employment and consumption releases change the earnings the market expects before any company has confirmed anything, which is why an index responds to a data release that mentions no company at all.

Two properties of this channel are worth holding. It is slow relative to the second factor: expectations for aggregate earnings move over weeks and quarters, not in seconds. And it is not evenly distributed across constituents, so a report from a company carrying a large index weight reaches the level far more than a report from a small one, and a reporting season is therefore a sequence of unevenly sized events rather than a uniform process.

The discount rate channel 

The multiple is, in large part, an expression of the rate at which future earnings are discounted back to the present. A higher risk free rate makes a given stream of future profits worth less today and simultaneously makes the alternative of holding government debt more attractive, and both effects push the multiple down. A lower rate does the reverse. This is the channel through which a central bank decision, an inflation release or a movement in government bond yields reaches an equity index without touching a single company's business.

The channel is fast, and it is unequal across constituents in a way that is entirely predictable from their profiles. A company whose value rests on profits expected many years out is more sensitive to the discount rate than one earning steadily today, because more of its value sits further away and is discounted harder. This is the mechanism behind the observation that growth heavy indices respond to rate news more than broad or value heavy ones, and it is a property of arithmetic rather than of sentiment.

It also explains why the same economic release can reach an index in two opposing directions at once. A strong activity report raises expected earnings and, at the same time, raises the expected path of policy rates. The first factor rises and the second factor falls, and which of the two dominates on a given day is not determined by the release itself. Attributing an index movement to a release without separating the two channels is the most common analytical error in this class.

The risk premium 

The multiple contains more than the discount rate. It also contains the compensation investors require for holding equities rather than government debt, and that requirement is not constant. It widens when uncertainty about the distribution of outcomes rises, which is why an index can fall while both expected earnings and policy rates are unchanged. Geopolitical events, banking stress, elections and any development that widens the range of plausible futures act primarily through this term.

This is also the channel in which indices across regions move together. Earnings are local and the discount rate is national, but the appetite for risk is global, so during a broad repricing of uncertainty the correlation between otherwise unrelated national indices rises sharply. Correlations measured in calm conditions consistently understate correlations during stress, which is a well documented property of the class rather than an anomaly.

Composition, and the currency underneath 

Two further drivers are properties of the specific index rather than of markets in general, and they explain most of the divergence between indices on the same day.

• Sector composition. An index inherits whatever mix of industries its constituent list happens to contain, and a shock to one industry reaches each index in proportion to that industry's weight. An oil price movement reaches an energy heavy national index directly and a technology heavy one barely at all, from the same news at the same moment.
• Concentration. Where a weighting rule allows a few constituents to dominate, the index increasingly reports those companies, and a single earnings report can move the whole level. The published level does not disclose this, and the constituent weights have to be consulted to know it.
• Revenue geography. A national index whose largest members earn most of their revenue abroad responds to the exchange rate through translation, which is the mechanism behind the long observed tendency of some national indices to move opposite to their own currencies.
• The currency of denomination. An index is quoted in the currency of its market, so a result must be converted into the account currency. In markets with volatile currencies this conversion can account for more of the outcome than the index movement itself.

Flows that are index events, not market events 

A final group of movements originates inside the index machinery rather than in the market it measures, and they are distinguishable because they are scheduled and published in advance.

1. Reviews and reconstitutions. When a provider adds or removes constituents, or revises a free float factor, every fund tracking the index must trade the difference on the effective date. The divisor is adjusted so the index level itself is continuous, so the flow lands on the constituent share prices rather than on the index arithmetic.
2. Derivative expiries. On the dates when listed index options and futures settle, the volume transacted in the constituents rises sharply and settlement prices are determined by defined auction procedures. The elevated activity is a calendar feature.
3. Ex dividend clustering. A price return index declines mechanically as constituents distribute cash, and where a dividend calendar clusters, the cumulative effect over a season is visible on the chart without anything having been lost.
4. Session boundaries. Between one cash close and the next cash open no constituent price is being set, so a full period of accumulated news is expressed in a single step at the reopen rather than continuously.