Home Blogs Asian Chemical Connections The flawed “science” of forecasting

The flawed “science” of forecasting

Business, China, Economics, Markets, Polyolefins
By John Richardson on 20-Nov-2007

Maybe I’ve been to too many conferences this year, and indeed over the last decade, and have seen too many forecasts go wrong.

Call me cynical, or plain wrong, but………..……the key to success in a chemical company planning department or in a consulting firm might be to produce numbers with decimal points in order to give the impression that you have been precise with your calculations.

Round numbers lack the feeling of credibility because the world is seen, quite rightly, as a complex place. The temptation could therefore be to add a 0.1 or a 0.3 if your calculations have actually generated, say, an estimate of exactly 10% demand growth for polyethylene in China in 2007.

It’s not just a case of the suspiciously nearly always un-round numbers that give cause for concern.

In the ten years I’ve been covering this industry I haven’t once heard somebody ask for details of the assumptions behind the numbers presented by consultants and chemical companies during conferences and seminars. This is extraordinary when you consider I have seen almost as many PowerPoint slides as there are demand and price forecasts that have been proved wrong.

But maybe all this takes place behind closed doors during question and answer sessions or in rigorous cross-checking of every internally and externally produced report that lands on the desks of project planners.

I doubt this very much, firstly because of the volume of numbers that need to be fed into any model for predicting the profitability of a new project. Any one individual might have a view on polycarbonate demand growth in China and India, but Vietnam?

Leaving Vietnam up to the experts as it’s only a relatively small country and won’t make that much difference to the overall global supply and demand balances might be the temptation.

But if enough Vietnams are added together, you might end up with an error significantly bigger than a few decimal points added to round numbers for the sake of authenticity.

Secondly, company executives quite often want their hunches confirmed and therefore look for supporting rather contradictory evidence. In other words, they will happily swallow any number (preferably not a whole number) which helps them get a proposal past their board of directors.

As I said, this might all be far too cynical, grossly unfair and ill-informed. But can you honestly say you have challenged every single number that’s come across your desk?

And why, if the “science” of forecasting is indeed a science, do company research departments and consultants keep getting it wrong time and time again?