Or, are there nonlinearities in the real (macro) world?
Figure from IMF Fiscal Monitor, April 2012.
The authors describe the results underpinning this graph thus:
The model finds significant evidence that the impact of fiscal policy on economic activity varies with the business cycle and that the effect of fiscal policy on output is nonlinear. Average fiscal multipliers in G-7 countries are significantly larger in times of negative output gaps than when the output gap is positive (Figure A1.2). Results from a simple linear model are very much in line with averages identified in the previous literature, as shown in Table A1.1. Assuming, in line with recent fiscal adjustment packages in advanced economies, that two-thirds of the adjustment comes from spending measures, a weighted average of spending and revenue multipliers in downturns yields an overall fiscal multiplier of about 1.0.
In line with the bulk of the previous literature (including the survey by Spilimbergo, Symansky, and Schindler, 2009), short-term spending multipliers are found to be significantly higher than revenue multipliers. This can be explained with basic Keynesian theory, which argues that tax cuts are less potent than spending increases in stimulating the economy, since households may save a significant portion of the additional after-tax income. (p.35)
The idea of nonlinear responses makes sense to me. In the context of the standard aggregate demand/aggregate supply framework, the distinction can be illustrated as follows:
Figure 1: Linear (in logs) aggregate demand and linear aggregate supply vs. nonlinear/kinked aggregate supply curve.
Note that variability in aggregate demand (AD) around potential GDP, yFE, (denoted by the gray arrow) induces smaller increases in output when above full employment than below. (This by the way reminds me of a 1995 paper by my coauthor Guy Meredith, with Doug Laxton, documenting asymmetry in inflation adjustment to slack. In other words, this idea has been around for a while).
What I find of interest are the US results.
Excerpt from Figure A1.3 from IMF Fiscal Monitor, April 2012. Left side bars are for positive spending shocks at 4 and 8 quarters; right side bars are for negative revenue shocks at 4 ad 8 quarters.
In other words, government spending has large output effects, particularly in the presence of a negative output gap, once nonlinearities are allowed. Tax revenue changes have much smaller impacts (which are nonsignificant statistically in the linear specification).
Our findings suggest that all of the extensions we developed in this paper—controlling for expectations, allowing responses to vary in recession and expansion, and allowing for different multipliers for different components of government purchases— all have important effects on the resulting estimates. In particular, policies that increase government purchases have a much larger impact in recession than is implied by the standard linear model, even more so when one controls for expectations, which is clearly called for given the extent to which independent forecasts help predict VAR policy “shocks.” Given the historical experience of the US economy, our preferred estimates of the government spending multiplier are between 0 and 0.5 in expansions, and between 1 and 1.5 in recessions.
One interesting implication of these results, beyond the fact that further spending cuts now would be disastrous, is that the hundreds of billions of stimulus during the 2004-08 period –- during which the output gap was slightly positive — could have been better spent now. (After all, publicly held Federal debt increased by $3.4 trillion going from 2001Q1 to 2009Q1.)
Figure 2: Log GDP (blue) and log potential GDP (gray). NBER recession dates shaded gray. Source: BEA, 2012Q1 advance, and CBO, Budget and Economic Outlook, Jan. 2012, and NBER.
Or, as I wrote nearly six years ago:
What we can be certain of is that the choices made by policy makers in the past five years have circumscribed our ability to manage a downturn.
This post originally appeared at Econbrowser and is posted with permission.