This is an update to my series of posts (2009Q1, 2009Q2, 2009Q3) that uses Statistics Canada's estimates for monthly GDP estimates (available for the first two months of the previous quarter) and the LFS data for the last month of the quarter to provide an estimate for GDP growth in the previous quarter. Statistics Canada will be releasing their numbers on March 1.
The mean of this distribution is 4.0%, its standard deviation is 0.5%, and the interquartile range is [3.7 , 4.4].
This is lower than the preliminary US estimate, and since this model doesn't provide a breakdown of aggregate demand, I don't know to what extent this projected increase may be driven by the replenishment of inventories.
Looks like others agree with you (maybe they borrowed your model? ;-)):
"But Friday, economists, who are notoriously stubborn about changing their predictions, found themselves boosting their outlooks for Canada. Among them were analysts at CIBC World Markets and BMO Nesbitt Burns, both of whom predicted the economy would grow at a 4-per-cent annual rate in the fourth quarter."
http://www.theglobeandmail.com/report-on-business/economy/recovery-points-to-summer-rate-hike/article1450086/
Posted by: Nick Rowe | January 29, 2010 at 09:17 PM
Huh. Maybe they'll develop the habit of providing error bands, too.
Posted by: Stephen Gordon | January 29, 2010 at 09:36 PM
I've got a Q4 forecast of 3.9%. (2 SE confidence bounds = LB: 3.3; UB: 4.5%)
Posted by: brendon | January 29, 2010 at 10:50 PM
Hmm. It may not be a good sign that so many people have almost exactly the same numbers...
Posted by: Stephen Gordon | January 30, 2010 at 08:59 AM
or we're all getting really good at forecasting.....
Posted by: brendon | January 30, 2010 at 06:38 PM
I'm not sure we'll ever see SE bands in the Bay Street forecasts. I have heard that most of the banks use Global Insight spreadsheets (which is why they their forecast output has the same format, save CIBC and Scotia), which may not provide standard errors.
Posted by: brendon | January 30, 2010 at 06:43 PM
That sounds like what I've heard. The banks don't really do forecasts, because that would involve hiring - and paying for - a team of econometricians. Instead, they subscribe to forecasting services that generate projections based on the subjective inputs that clients provide (US economy, oil prices, etc). Does that correspond to your understanding?
Posted by: Stephen Gordon | January 30, 2010 at 07:06 PM
Banks get their forecasts from subscriptions services? Data based based on subjective inputs from *clients*?
Oh God. I need a drink.
Posted by: Patrick | January 30, 2010 at 09:33 PM
I think your impression of the forecasting process is largely correct. They use the same model, but with different assumptions. They may employ some short-run indicator models as well.
I have limited experience with Bay Street economists, but in the one conversation I've had with a Bay Street chief economist he mentioned that they have available a small version of Global Insight's Canadian Macro model. I also didn't get the impression that he knew much about forecasting (I'll let you guess which one it was).
Posted by: brendon | January 31, 2010 at 01:17 AM