Viewpoints
How Useful Are Consensus Estimates in Small Cap?

Timely insights from portfolio managers and industry experts on key financial, economic and political issues.

The views expressed in these posts are those of the authors and are current only through the date stated. These views are subject to change at any time based upon market or other conditions, and Eaton Vance disclaims any responsibility to update such views. These views may not be relied upon as investment advice and, because investment decisions for Eaton Vance are based on many factors, may not be relied upon as an indication of trading intent on behalf of any Eaton Vance strategy. The discussion herein is general in nature and is provided for informational purposes only. There is no guarantee as to its accuracy or completeness.

Topic Category
Content Type
Brand
The article below is presented as a single post. Click here to view all posts.

By J. Griffith Noble, CFACo-Director of U.S. Small/Mid-Cap Equity, Eaton Vance Equity and Michael D. McLean, CFA Co-Director of U.S. Small/Mid-Cap Equity, Eaton Vance Equity

Boston - In September 2018, we shared what we believed to be interesting research around the magnitude of annual forecasting error by Wall Street analysts. We found that these consensus earnings estimates for constituents of the Russell 2000 small-cap equity index were off by 47% on average over the eight-year study from 2010 to 2017. After four years and a big economic downturn, we revisit the analysis here.

In the original blog, we used the study as evidence of inefficiency in the small-cap market, the pitfalls of focusing on short-term results and "anchoring" to a wildly inaccurate consensus. In this post, we provide an update to the initial study with a refined methodology and a significantly extended time period.

Our original post also hoped to dispel the myth of quantity equals quality — the industries with the highest number of analysts, Energy and Tech, had the least accurate estimates, which still holds. With this update, we want to share our application of the results and our belief that focusing on quality businesses with a long-term perspective and an appropriate valuation framework is a better way to potentially generate outperformance and exploit inefficiencies in the market.

Perhaps not surprisingly, the inaccuracy we witnessed over the original eight-year study was similarly large over a 22-year period. The year-by-year errors in consensus estimates are displayed below, where once again we see the median error was sizeable at 40%. The same sector pattern was also present in the updated study, with Utilities only off by 14% on average versus Energy at 94%.

Small cap 072222

Source: Eaton Vance Research. Data from 1999 to 2020 as of 7/26/22. Provided for illustrative purposes only.

As a reminder, if company A was expected to generate $1.00 of EPS entering 2020 but eventually reported $1.40 in a positive scenario, or $0.60 in a negative scenario, we measure that as a 40% estimate error (see methodology below).

With the market's fixation on short-term earnings beats and misses relative to Wall Street estimates, one might look at the results of this study and conclude that exploiting this inefficiency simply requires more accurate forecasting than consensus. To such a simple solution we say... good luck!

More importantly, we believe this "solution" misses the point. Overconfidence can often lead to short-hand valuation approaches that produce overly precise point-in-time price targets based on the P/E ratio multiplied by earnings per share.

In reality, if we knew what earnings would be with 100% foresight, that still wouldn't tell us what we should be willing to pay for a business. The appropriate valuation methodology — never mind what multiple we should place on earnings — depends on the fundamental quality characteristics of that business: durability/variability of the earnings, growth potential, return on invested capital, the cost of capital and the conversion of those earnings into cash flow over the long term.

Rather than pursuing false precision, investors may be able to tilt the odds in their favor by embracing uncertainty and asking what would be a realistic range of outcomes for a business in an uncertain world. What if everything goes right for the company? Where could its cash flow and valuation go? What if the investment thesis is flawed? What is the potential downside scenario? Understanding a business and its drivers can position investors for a more complete view of a stock's potential than trying to pinpoint earnings.

It's very difficult to accurately forecast earnings, and Wall Street estimates certainly provide a large data set to prove that they widely miss the mark year in and year out. Rather than simply trying to be more accurate, we believe the way to exploit inefficiencies in the small-cap market is by focusing on businesses rather than earnings. Embrace uncertainty with a thoughtful framework that accepts unknowns, while taking a much more comprehensive approach to determining what a business is worth in the long term.

Bottom line: Earnings are just one piece of the puzzle, and we believe investors should beware of overemphasizing short-term earnings expectations at the expense of gaining a full understanding of a company's long-term economics.

IMPORTANT INFORMATION

Russell 2000® Index is an unmanaged index of U.S. small cap stocks.

It is not possible to invest directly in an index. Past performance is no guarantee of future results.

METHODOLOGY

Eaton Vance Small/Mid-Cap Equity Team performed analysis to compare beginning of calendar year forward adjusted earnings per share consensus analyst estimates for the constituents of the Russell 2000 Index against the actual adjusted earnings per share results reported for each index constituent over a 22-year time frame from 1999 to 2020. We calculated the percent difference (actual earnings divided by earnings estimates, minus one) between the estimates and actuals to demonstrate the extent to which consensus analyst estimates diverged from actual earnings. We took the absolute value of each percentage differential and excluded any estimates that were off by more than 300% to eliminate outliers. Taking the median of the absolute values across this data distribution yielded our final result.