ACCT3013 Lecture Notes - Lecture 13: Keynesian Beauty Contest, Utility, Observational Error
![](https://new-preview-html.oneclass.com/L7W93pro8dXxmolo4B48j1VKzkbBJv2e/bg1.png)
NATALIE WU
1
ACCT3013 FINANCIAL STATEMENT ANALYSIS WEEK 13
Behavioural Aspects
Forecasting Biases
• Forecasting for FSA requires synthesising all newly acquired information with existing
knowledge, public opinion and personal experience / understanding.
• Forecasting depends on not only your ability to forecast, but also on your access to better
quality information, both in terms of accuracy and timeliness.
• Forecasting is mostly driven by data, and involves the evaluation of content, volume and quality
of information about business fundamentals.
• However, forecasting for FSA is inherently subjective: opinions are shaped according to what
one believes and how one perceives the question at hand, and thus it is underlined by cognitive
biases.
Ideal Forecasting Environment
From a market perspective, we do not care whether you or someone else gets it right. What is
important is that:
• We have many forecasters working on the same problem at the same time.
• All forecasters try their best to get fair valuations without any underlying systematic bias, e.g.
from any vested interest.
• All forecasters work independently from each other without cross-references or time-
dependencies.
If the above conditions are met, then the Law of Large Numbers and Central Limit Theorem are met
and the average valuation forecast (the consensus) is an unbiased estimate of the truth.
Actual Information Processing
1. Form initial estimate using public information: read annual reports, quarterly reports and
analyses by financial journalists.
2. Consider other available information: observe other forecasts from other analysts (industry,
government, academia) – can be public or private information.
3. Revise your estimate by benchmarking on other analysts: benchmark expectations on the
basis of all other information provided up to that point.
4. Issue forecast and recommendation: aim to get close to the consensus in order to minimise
penalty, maintain good reputation and attain consistent rewards over time.
Steps 2-4: iterative process.
Keynesian Beauty Contest
• Rewards are given to those who get close to the consensus and not to the true answer.
• It is important to estimate what everyone else would forecast.
• It implies that even if you have high quality private information, then you have no incentives to
issue forecasts on the basis of this information because you will then deviate from the
consensus.
• This also means that you give more weight to other forecasts that are closer to each other (with
less variance) and less weight to forecasts that are far away from each other, hence larger bias.
Document Summary
From a market perspective, we do not care whether you or someone else gets it right. If the above conditions are met, then the law of large numbers and central limit theorem are met and the average valuation forecast (the consensus) is an unbiased estimate of the truth. Issue forecast and recommendation: aim to get close to the consensus in order to minimise penalty, maintain good reputation and attain consistent rewards over time. Keynesian beauty contest: rewards are given to those who get close to the consensus and not to the true answer. It is important to estimate what everyone else would forecast. Behavioural economics and behavioural finance explain that financial analysts, investors, managers, regulators and everyone else involved in capital markets are economically irrational and their decision-(cid:373)aki(cid:374)g pro(cid:272)ess is largely u(cid:374)derpi(cid:374)(cid:374)ed (cid:271)y (cid:858)(cid:271)ehavioural (cid:271)iases(cid:859) i(cid:374) (cid:272)o(cid:373)pariso(cid:374) to the (cid:272)lassi(cid:272)al utility theory.