Tyler Cowen thinks about the industry for forecasts. He makes me wonder how much private work he's done, as I think he seems to really miss the point of private economic modeling.
I must say--remember that a firm will often purchase a number of difference forecasts, which will disagree, as well as produce their own forecasts. Does one expect a firm to "believe" all of these forecasts? I don't think so. Would a firm cherry-pick one forecast that they choose to believe? Then, why purchase the others?
No, I think forecasts are more used to consider a range of possibilities. None is particularly more believable necessarily, but having a variety of forecasts gives us a starting point in talking about what may happen in the future and assumptions we may consider using.
Particularly in the light of the recent recession, there is a lot of distrust in models anyways, even if they're considered necessary.
In that vein, Free Exchange tells me about a new NBER paper actually taking a stab at the financial aspect of macroeconomic modeling.