To Be, or Not to Be…Surprised: Part Two

In a previous post, we discussed how the interactions between security analysts and the companies they follow often lead to a biased measure of consensus expectations. Here, we focus on what the term consensus implies, what it does not explain, and how it relates to earnings surprises.

As noted in Part 1, there is an implicit expectation of small positive earnings surprises. This manifests in the fact that small earnings surprises are more prevalent than would be expected compared to the assumption that if expectations were truly unbiased, surprises (positive and negative) would follow close to a normal distribution. Our long-term data indicates that companies in our large capitalization universe have reported positive surprises of at least one standard deviation above consensus at a higher rate (44%) than would be expected from a normal distribution. Interestingly, this bias toward positive earnings surprises is asymmetrical. The percentage of companies announcing negative surprises equal to or greater than one standard deviation is fairly close to the percentage expected in a normal distribution.

An important question then is: what is meant by “consensus” expectations? The term sounds like it encapsulates the popular opinion of the investment community, but, it is just a simple average of the forecasts of the analysts that publish estimates on the company. Consensus is just one point along a spectrum of expectations. It does not tell us anything about the dispersion of estimates. As estimate dispersion widens, the average becomes less indicative of analysts’ expectations. This is important when evaluating actual results once they are reported. An actual result can miss the average point of the estimates, yet still fall within a statistically expected range.

Standardized measures of an earnings surprise are more useful to accurately assess how big or small the difference really was. A large earnings miss, relative to consensus, may not pressure a stock if there were a high degree of uncertainty beforehand. Likewise, an earnings surprise of just a few pennies can thrust a stock price higher if there was a narrow range of expectations. Remember, there is an inherent expectation that actual results will beat expectations, so normalizing filters out “noisy” surprises that arise from biased estimates and earnings management.

It is important to view earnings announcements through an appropriate lens. When digesting earnings results, remember the consensus is just an average of a broad array of possibilities. Standardized results provide a useful measure to evaluate earnings results relative to each of those expectations.

Read more in To Be, or Not to Be…Surprised >