Sunday, December 31, 2006
The forecasting league table
Posted by David Smith at 10:59 AM
Category: David Smith's other articles
SCORING SYSTEM: GDP growth: 3 points for 2.5%-2.9%; 2 points for 2.2%-3.2%; 1 point for 1.9%-3.5%. Inflation: 3 points for 2.4%-2.8%; 2 points for 2.1%-3.1%; 1 point for 1.8%-3.4%. Current account: 1 point for within £5 billion of outturn. Unemployment: 1 point for 0.9m-1m. Base rate: 1 point for direction (up). Bonus: 1 point for growth exceeding inflation. Blank = No Forecast for this variable * percentage of GDP ** Unemployment rate (Labour Force Survey) *** Unemployment level (LFS)
An interesting collation of macro forecasts, but does it really tell us anything of value? Afterall, no group or individual consistently outperforms the consensus...or if they do, they surely shouldn't be publishing their predictions for all to see !
Fair point, though over the years I've been doing this there are some forecasters who consistently do better - or worse - than the consensus. It's a bit of end-of-year fun, and those who do badly can always look forward to the inevitable data revisions that will prove they were right all along. There's also a spooky tendency for people to produce their best forecasts just before leaving the organisation for which they did it.
Do you have similar tables for the last few years, to identify whether there is a genuine trend for some forecasters to be more consistently right than others over time?
Hi David, It is indeed just a bit of year end fun, and your additional observations are interesting and quite funny. I suppose you could look at the Bloomberg records of top economic forecasters and track their performance against the consensus, as some people do with equity forecasters. If any of this public information is found to have an advantage, you could trade this 'edge' outright via the relatively new economic derivatives that allow you to trade directly the outcome of economic data.
I have noted the publication of many forecast tables to be self-serving to the financial industry in that they allow the top performers to self-publicise and tout their performance to the buy-side, while those who perform badly just keep quiet about it. It's all part of the illusory forecasting merry-go-round.
Yes, in response to that, and to answer Nick's point, I did do an exercise two or three years ago assessing the long-term records of the forecasters. I can't lay my hands on it so far but when I do I'll post it here.