‘How to Eliminate Bias’ by Bijan Tabatabai, Finance and Management, December 2009

The existence of bias in forecasts is a subject very close to our hearts and one that we regularly cover in our course sessions on risk analysis and financial planning. The evidence is very strong that many forecasts in most companies are biased, sometimes one way, sometimes the other. Pessimistic bias - making sales forecasts deliberately low or cost budgets deliberately high - is usually found in the annual plan so that the forecaster can later receive praise for over achievement; optimistic bias - high sales and low costs - is likely to be found in the context of investment appraisals when a project team wants to get their proposal through the system.


The article starts with a helpful definition of an unbiased forecast - one where there is a 50/50 chance of achieving the actual result, where it is just as likely that the actual will be over or under the forecast. There is then the obvious but important observation that bias is caused by human behaviour; maybe a manager who likes to receive praise for over delivery, or a manager who needs to get a project accepted.

The author then produces a list of types of bias which is rather too long and involves categories that overlap too much. The most significant category in our experience is ‘motivational bias’- the impact of incentives in the organisation to either over or under achieve. Also interesting is the category of ‘anchoring bias’ which occurs when a team only listens to selective data that reinforces their plans. What is not mentioned in the article is the important point that these types of bias are often driven by the corporate culture or a particular management style which it is possible - though usually difficult - to change.

There is then coverage of what companies can do to eradicate bias from their forecasts and two approaches are highlighted; either you can adjust manager’s estimates where bias is suspected, or you can change the process to eliminate the behaviour that causes it in the first place. However, having posed these two alternatives, the author does not make this important choice; instead he goes on to talk about the tools he has used within Unilever to identify bias. The quote from Unilever confirms this - ‘the toolkit provides simple but statistically sound rules to spot bias’.

Though I am sure that the toolkit works well, I would have liked to see mention of the important issue that we cover on our courses for Unilever and others - the role of the financial person as business partner. The good financial manager will be aware of the corporate pressures, will know the personalities of his or her management colleagues, will take into account past performance compared to plan and will try to steer each manager towards unbiased forecasting.

The author’s technical toolkit will help but in the end this is a behavioural problem that requires a behavioural solution and the financial person has a big part to play in steering forecasts towards a consistently objective and unbiased level.

To read this article go to:

http://www.icaew.com/index.cfm/route/169551/icaew_ga/en/Faculties/Finance_and_Management/Publications/Finance_and_management/How_to_eliminate_bias_in_forecasts