groundhog day," reflecting some comments and email. Here is a pretty up to date graph of real GDP, the CBO's current assessment of "potential" and the previous trendline, which is a pretty good approximation to what the CBO thought potential was just before the crisis. The surprising thing about this recession is how sadly-diminishing expectations of "potential" are behind the closing of the gap, rather than GDP rising to meet potential.
The forecast of a swift recovery really had nothing to do with DSGE vs. VAR modeling, new Keyensian vs. old Keynesian models or anything else deeply technical. Any model that embodies something like the CBO's previous assessment of potential will say that pretty soon the economy recovers to potential.
The one exception would be the permanent income forecasting model, which ignores the CBO's "potential" altogether and says future GDP converges to whatever nondurable plus services consumption is now. The latter is a random walk and consumers reveal their forecasts of future GDP by what they're eating now. Such consumption fell sharply with the financial crisis and has followed a similar stair step. True blue stochastic growth models might behave similarly, as they have no "potential" concept. But I don't think any policy-oriented forecasters are using either approach at the moment.
A good project would be to assemble groundhog day plots for the CBO's GDP forecasts and their potential GDP forecasts.