Monday, January 9, 2012

On the ineffectiveness of a fiscal stimulus

When is a fiscal stimulus going to work? If you listen to economists these days, it would be difficult to know. One reason is that protagonists quickly become sidetracked by the associated political discourse. The other is that one would first need to define the fiscal stimulus. Depending on whether it is provided by tax rebates, public expenses or tax delays makes a big difference. Also, who is targeted matters a lot. Thus one needs to get the specifics of the fiscal stimulus to give a proper answer. The general rule, though, is that a fiscal stimulus works best if the credit or liquidity constraint households obtain additional cash. They are going to spend it right away. The other households would reduce consumption only slightly in anticipation of future increased taxes due to a wealth effect. In the aggregate, consumption would then go up, but not a lot. But there can be other circumstances where it would work better.

Thorsten Drautzburg and Harald Uhlig study a situation similar to the one now in the Unites States: nominal interest rates getting very close to zero. Then, they claim the fiscal multipliers are still rather small, but may be positive despite distortionary taxation. Now contrast this with the results of Lawrence Christiano, Martin Eichebaum and Sergio Rebelo, who argue that the fiscal multipliers become very large (and positive) when nominal interest rates are close to zero. Who is to believe? Both use a rather elaborate DSGE model, in both cases estimated. Both use the Calvo fairy. Or a Taylor rule. The first has financial frictions, while the second has investment adjustment costs, which should come to the same. If anybody can figure out why the results are so different, I would appreciate it.

It particular it would be interesting to understand why now would now be the wrong time to institute an austerity regime. To me, it would even make perfect business sense: why not build infrastructure when the cost in terms of financing and labor are low? Keep in mind that every unemployed worker that is hired does not need unemployment benefits...

5 comments:

Vilfredo said...

I find it intriguing that the authors of the first paper are at the University of Chicago, while all three authors of the other are at cross-town rival Northwestern University.

Anonymous said...

I haven't looked into it deeply at all, but my gut instinct is that the the financial and investment frictions employed by the two sets of authors do not come out the same, in the sense that they do not map to the same wedges in business cycle accounting terminology. Financial frictions often manifest as labor wedges, whereas the investment adjustment costs should show up as investment wedges.

Kansan said...

I am rather puzzled that Harald Uhlig does not mention why he gets a different result. The other paper is only mentioned in passing. He should be doing a better job selling his paper when a close competitor is getting very different results and was published in the JPE.

Anonymous said...

Uhlig should respond here.

daniels said...

This is explained rather well in Michael Woodford's paper on the "simple analytics of the gvt expenditure multiplier". The main difference between big multipliers and small multipliers papers at the ZLB is between the assumed duration of the fiscal stimulus plans. If the rise in government spending is concentrated during the ZLB regime, we get big multipliers. If government spending increases are hard to reverse, then future increases in taxes (even more if they are distortionary like in uhlig et al) act like a negative news shock depressing private spending today, which gets amplified at the ZLB. Fernandez Villaverde's and some co-authors' paper arguing that (credible?) promises of future structural reforms can be highly stimulative at the ZLB is also related.
DS.