Representativeness in Growth Forecasting

Representativeness can heavily influence investor decisions and analyst forecasts. This cognitive bias leads us to assume that past performance will persist, making it a common source of error in growth rate estimation.
Consider this example, shown in the chart below:

1️⃣ Initial Growth Expectations:
Analysts build portfolios based on long-term earnings growth forecasts. The first two bar for each portfolio reflects the annual growth rate from the previous 5 years and forward looking 5 years estimations (correlated).

2️⃣ Projected vs. Actual Growth:
The subsequent three bars show the actual growth per annum over 1, 3, and 5 years following these forecasts.

Analysts often assume that companies with strong historical growth will continue on the same trajectory, and conversely, that poor performers will remain so.

Key Observations:

Overestimating Growth Persistence:
Analysts tend to project high growth for previously high-growth companies, effectively thinking, “this company has been great, so it will continue to be great” or “this underperformer will always underperform.” This is a clear example of representativeness bias, akin to the well-known Linda problem in psychology.

Ignoring Mean Reversion:
Analysts overlook that earnings growth typically reverts to the mean over a 5-year horizon. The base rate of mean reversion is high, meaning that low-growth companies tend to catch up, while high-growth companies often slow down. Despite initial appearances, low-growth portfolios generate nearly as much long-term growth as their high-growth counterparts.

This analysis highlights a crucial lesson for investors: judging a company by its past growth can be misleading. The tendency to ignore mean reversion and place undue weight on historical performance is a mistake many analysts make, but one that can be avoided with a more evidence-based approach.

Sourse: James Montier, Behavioural Investing.