The Fair Model and the Crisis

I've been trying to figure out a way to argue the case for more short-term stimulus in a way that might appeal to those worried about deficits over the longer term. It is likely that a stimulus that reduces the duration and intensity of the recession will result in a lower national debt over the medium-term horizon (say, 5-10 years). This is because the longer and deeper the recession is, the more government revenue will suffer.

I thought to run Ray Fair's macro model to quantify what government debt might look several years in the future, with and without a stimulus package now. Empirical forecasting models like this one suffer from the Lucas critique, but they're the best we have.

My intents were stymied, however, by the fact that the most baseline run of the model doesn't forecast a severe recession. Here's what the model's forecast memo of October 30 says:

Real GDP Growth and the Unemployment Rate: The model is predicting no growth in the next two quarters, 1.7 percent in 2009:2, and 2.3 percent in 2009:3. The unemployment rate rises to 6.9 percent in 2009:3.
....
The negative wealth effect from the fall in stock prices ... is one of the reasons the model is predicting no growth for the next two quarters. There is, however, nothing in the initial conditions and likely paths of the exogenous variables that suggests there is going to be a deep and prolonged recession. It may be, of course, because of the current financial crisis that consumers and investors are concerned enough or restricted enough in their ability to borrow that they decrease their spending by huge amounts. These potential "animal spirits" and "credit crunch" effects are not the type that the model can capture. If, for example, people have been spooked by the dire warnings of policy makers and significantly cut back their spending, the economy could be much worse in the next few quarters than the model is predicting.

Real housing prices are exogenous in the model, and the baseline scenario has prices continuing to drop slowly and stabilizing at about 5% below current values. I tried running the model with a bigger decline in housing prices (which creates a larger wealth effect in the model), but it didn't make much difference for the trajectory of the economy beyond a couple quarters down the line.

The model's forecast does make me wonder just for a moment if the dangers of a deep recession have been overstated. But I am fairly confident that the main reason the model doesn't forecast a more severe recession is that, reflecting the weaknesses of macro modeling, the model doesn't have a way of incorporating the "credit crunch" effect.

(Note that I'm not actually a macroeconomist, and this was first stab at tinkering with the Fair model, so take all my conclusions with a grain of salt.)