Testing for Underestimation Bias in Consensus Revenue Forecasts

Summary

The document describes a question about whether analysts’ consensus revenue forecasts tend to fall below companies’ eventual quarterly results. Forecasts are refreshed monthly during the quarter, so the data could be examined by comparing each forecast with the realized revenue for its period. The proposed explanation is that companies benefit in share price when they beat expectations, potentially giving analysts or managers an incentive to set cautious estimates.

No statistical test, sample findings, or evidence of systematic bias is provided; the document asks how such a bias might be demonstrated. A useful analysis would need to define the forecast horizon and error measure, then test forecast errors across enough periods and companies. It would also need to account for forecast revisions, company or sector differences, and the possibility that revenue surprises reflect new information rather than deliberate underestimation. The incentive described is a hypothesis, not proof of biased forecasts.

Key ideas

• The document asks whether consensus revenue estimates systematically fall below realized quarterly revenue.
• Forecasts are updated monthly before each quarter ends, creating observations at different lead times.
• The suggested incentive is that a company may benefit when actual results exceed expectations.
• The document presents no test or evidence, so underestimation remains a hypothesis to investigate.

Tags

# Presence of underestimation bias in consensus earnings predictions


# Presence of underestimation bias in consensus earnings predictions












I am working on a financial data that entails forecasted revenue a company generates over a fiscal quarter and the actual revenue for that quarter.

```
import pandas as pd

df = pd.DataFrame({'Period': ["Q3'16", "Q1'17", "Q2'17","Q3'17"],'Predicted M1':[1000,1026,1023,1024],'Actual Earnings':[1010,1030,1020,1026]})
```

Every month throughout the quarter I generate a new forecast for what revenue will be at the end of the quarter.

A little background on the data. These are consensus earnings predictions/estimate for a quarter. If a company beats these predictions they enjoy a positive share price change. This gives an incentive for analyst/manager to under predict those earnings. Hence majority of predictions are underestimated.

My question is this: How can I prove/show presence of such biases in the forecast.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.