A consensus estimate is a single figure that summarizes the forecasts of a group of professional forecasters for a number that has yet to be reported. In equity markets it usually refers to the average of analysts' predictions for a company's earnings per share, revenue or other financial measures for a coming quarter or year. In economics it refers to the average forecast for a data release such as employment, inflation or growth.
The consensus matters because markets price in what is expected. When the actual figure is published, the reaction depends less on whether the number is good in absolute terms than on how it compares with the consensus. The term has no connection with consensus in the blockchain sense, which concerns agreement among computers on a network.
How it is compiled
Analysts employed by brokers and independent research firms build financial models of the companies they cover and publish forecasts, often alongside an analyst rating. Data providers collect these forecasts and calculate summary statistics, most commonly the mean and sometimes the median, together with the highest and lowest estimates, the number of contributors and the standard deviation. A separate summary of recommendations, the analyst consensus rating, is produced in the same way.
Providers differ in which analysts they include, how quickly they drop stale forecasts and how they treat unusual items, so the consensus for the same company can vary slightly from one source to another. Most try to put all estimates on the same basis as the adjusted figure the company itself emphasizes, which often differs from the audited accounting result. For economic indicators, news agencies survey bank and research economists in the days before each release and publish the median.
Earnings surprises
The difference between a reported figure and the consensus is called a surprise and is usually stated as a percentage:
- is the actual reported figure.
- is the consensus estimate, taken as an absolute value in the denominator.
If analysts on average expected earnings of 2.00 dollars per share and the company reports 2.10, the surprise is percent, described as a beat. A result of 1.90 would be a miss of 5 percent. Share prices often move sharply on such differences, although management's comments about future periods can outweigh the reported number.
Research going back to the 1960s has found that share prices tend to keep drifting in the direction of an earnings surprise for weeks afterwards. This pattern, known as post-earnings-announcement drift, is one of the longest-studied challenges to the efficient market hypothesis.
Uses
Investors use consensus figures in several ways. Valuation measures such as the forward P/E ratio divide the share price by consensus earnings for the next year. Changes in the consensus, called estimate revisions, are tracked as a signal in their own right, since upward revisions have historically been associated with better share performance. Aggregated across an index, consensus forecasts give a picture of expected corporate earnings for the market as a whole.
The spread of individual estimates is also informative. A narrow range suggests that analysts agree and that the business is predictable, while a wide range signals uncertainty. For economic data, traders position ahead of releases such as the consumer price index according to the consensus, and the size of the market move afterwards is roughly related to the size of the miss.
Company guidance and the expectations game
Companies influence the consensus through guidance, the forecasts and ranges that management provides. In the United States, Regulation FD, adopted in 2000, bars companies from giving material information to selected analysts privately, so guidance is delivered publicly. Analysts tend to cluster around it.
A well-documented pattern is that estimates start high a year or more ahead, are reduced as the reporting date nears, and end at a level the company can slightly exceed. A majority of large US companies report earnings above the consensus in a typical quarter, which reduces the information in a simple beat. Traders sometimes refer to an unofficial whisper number, the result the market is thought to expect in reality, which may sit above the published figure.
Limitations
- Herding. Analysts are reluctant to stand far from their peers, since being wrong alone is costly to a career. Estimates therefore bunch together and can all miss in the same direction, a tendency related to anchoring bias.
- Optimism and conflicts. Long-range forecasts have tended to be too high. Analysts' employers may have investment banking relationships with the companies covered, and access to management can depend on a favourable stance.
- Thin coverage. For small companies the consensus may rest on two or three estimates, and a single change moves it substantially.
- Definitions. Comparisons break down when the company reports on a different basis from the one analysts forecast, for example on core earnings that exclude certain costs.
- Turning points. Consensus forecasts are slow to recognize recessions and recoveries, and errors are largest when conditions change abruptly.
For these reasons the consensus is best understood as a record of what the market expects, not as an independent prediction of what will happen.