I'll begin with some general remarks and then perhaps pass the floor to my colleague Tiff. He gave an excellent speech on this topic just a couple of weeks ago at the Economic Club in Toronto.
When you use your model to construct these forecasts, of course you have all the data in place and a structure around them. As we went through time, we were noticing that exports were not recovering as rapidly as our model had predicted in line with the actual evolution of foreign economies. At the time we believed that this was a temporary thing and we still believe fundamentally that it is temporary.
When you do your forecasts, you assume that over the next year or two the error term that you're generating will actually go back to normal. That's how a forecaster would do this.
If the error persists long enough, you begin to look for deeper reasons and then assess whether they are temporary or permanent. The sorts of reasons that we put our finger on basically look at the mix of growth in the U.S. in particular, which is not classic, and not every sector has contributed to growth yet. That's something we can look forward to.
The second thing is that, this being a non-typical cycle, the export sector lost a lot of companies, some 20% of exporting companies, and a lot of other companies downsized. The conditions that will bring back the export path are more demanding than would normally be the case because of the length of this cycle. So it is taking longer than we saw in the past.
Mr. Macklem may have a follow-up.