About Those GDP Numbers

See Munger and McMegan.

Look, we've spent something like 4-5% of GDP in stimulus, from the American Recovery and Reinvestment Act (a.k.a ARRA a.k.a. stimulus bill) to the bailouts. (You can track ARRA spending here and bailout spending here). We've spent about $120bn in ARRA and $575bn in other bailouts. That represents ~ 5% of GDP spent in deficit by the government: $700bn/$14.25tn ~ 5%. Those expenditures affect GDP evaluations, since GDP = Consumption + Investment + Government + Net Imports. If GDP's "natural" state rate now if government spending had stayed constant is negative 1-2%, then that translates into roughly 4% growth: -1 + 5 = 4. We actually got 3.5%, which is right in the back-of-the-envelope range.

But it's unclear if that means anything good. If the fiscal multiplier is greater than one, as CEA head Christina Romer believes, then that spending should have led to higher reported GDP than we actually saw (something like 5-6%, using Romer's estimate of the multiplier as close to 1.5)*. If the multiplier is less than or equal to one, as these folks argue, then at best this quarter's GDP report is a statistical mirage based on an accounting identity rather than actual economic improvement. If the multiplier is close to zero or even negative, as Robert Barro thinks, then economy is in great shape indeed.

I tend to take the middle view: the multiplier is probably close to 1 most of the time (although context is important), and I think the present circumstance backs that view up. In the 2nd quarter, GDP fell by about 1%. The government spends about 5% of GDP in deficit, and the next quarter GDP grows by about 3.5% despite falling employment. Coincidence?

This is one reason why talk of a "jobless recovery" is at least partially missing the boat: there hasn't actually been a recovery yet. We've just pushed some numbers from one side of an accounting ledger to other by taking on more debt and called it progress. It's not.

*Yes, I know that the lags matter too. But these, too, are unclear and you could just apply the same argument to next quarter or several.