Thursday, December 22, 2011

The Shrinkage of Expansionary Austerity

-- Posted by Neil H. Buchanan

After months of arguing against austerity measures in the U.S. and Europe -- both of which are in the midst of extended slumps that threaten to get worse, with interest rates near zero -- I have understandably received some feedback from readers who have asked for my take on the empirical evidence that supposedly supports the idea of "expansionary austerity." This is the claim that a government can cut government spending (and possibly increase taxes as well), in an effort to cut deficits, yet still see the economy expand, because businesses will be so excited about the government's reduced footprint on the economy that they will expand their investment spending and hire more workers. The evidence to support this idea (not really a theory, but just an empirical assertion about the response of business decision makers to particular stimuli) was and is unconvincing.

There have been some attempts to mount a serious academic defense of expansionary austerity. A Working Paper from 1999 that was written by Alesina and Ardagna (along with two co-authors), and another Working paper from 2009, again written by Alesina and Ardagna, are often held up as empirical proof that austerity is expansionary. Alesina was certainly eager to present the evidence in that light, writing a widely-discussed op-ed in the Wall Street Journal last year.

As one might imagine, those results have been widely contested. I wrote a Dorf on Law post last September, discussing the latest Alesina work. (I am not saying that the readers who contacted me should have known about that post. We welcome new readers all the time, making it helpful to revisit issues.) Today, I extended those earlier thoughts in a column on Verdict in which I directly addressed the weak theoretical and empirical underpinnings behind expansionary austerity.

In my new column, I first described why the Alesina results are simply irrelevant to our current situation. In the first Working Paper, for example, Alesina et al. write: "[M]any fiscal contractions have been associated with higher growth, even in the very short run. Similarly, economic activity slowed during several episodes of rapid fiscal expansions." Notice that these authors are simply describing episodes where expansion followed spending cuts or tax increases (and vice versa), without identifying whether the countries in question were already experiencing slumps when the austerity measures were adopted. For example, they write: "[I]ncreases in public spending increase labor costs and reduce profits." This would be true for most economies that are in relative health, but not at all for countries in slumps. Additional spending by the government today would neither increase labor costs, nor reduce profits, because of the slack in the economy.

Similarly, this approach fails to control for situations in which something else (other than a burst of business confidence-led spending) took up the slack left by the government. Basically, this empirical work -- even if we take it seriously on its own terms, which might not be wise -- does not show that any country has successfully moved out of a deep slump by cutting government spending, unless some deus ex machina like an export surge saves the day (as happened in Ireland in 1987).

I further explained in today's column that any theory that could back up expansionary austerity is extremely tenuous, because it requires businesses to respond not just in the right direction, but to a sufficient degree, to offset the contraction. In that way, it is very similar to the claims that tax cuts pay for themselves, which require not just that economic activity expand in response to tax rate cuts, but that the expansion be sufficient to outweigh the lost revenue from the rate cut. One could imagine a world in which people's responses are qualitatively and quantitatively sufficient to support such counter-intuitive economic claims, but the world in which we live never delivers what is needed to save those claims.

Two additional thoughts occurred to me as I wrote this week's Verdict column. First, we are having a surprisingly animated debate about whether there are any -- any -- examples to support the idea of expansionary austerity. As I argue, the empirical support just keeps shrinking, as we see Europe and the UK suffer the effects of broad austerity programs. The best that the other side of the debate can offer is present-day Ireland and Latvia, neither of which stands up to even a moment's scrutiny as a model for U.S. policy. And honestly, what would the other side say if Barack Obama started talking about how his economic policies must be followed because evidence from Ireland and Latvia proves that he is right? At best, therefore, we are having a debate about whether there are any exceptions to the Keynesian prediction that contractionary policy contracts the economy.

The second thought that occurred to me was that it does not really matter if the work by Alesina and his co-authors was ever intended to be a serious defense of expansionary austerity -- where "serious" would mean that it could stand up at least to the fundamental objections that have been leveled against it. (Economists disagree all the time, of course, but this particular run of claims contains the types of errors that normally do not survive the professional review process.) What matters it that there is now a series of papers with a Harvard economist's name on them, which anti-government believers can point to as proof that their preferred point of view is backed up by "evidence."

This is very similar to the debate over the empirics of the death penalty. For years, serious empirical work failed to find that the death penalty deters murders. The results were so uniform that one working paper that I read a few years ago (which I cannot find on-line) reasonably concluded that the only possible avenue of further empirical inquiry for defenders of the deterrence hypothesis would be to try to find sub-categories of crimes that might be subject to deterrence. In other words, it might be possible that the death penalty deters contract killings but not murders of passion, and more fine-grained empirical analysis might yet detect such an effect.

Notwithstanding that state of affairs, a group of economists published within the last decade a paper claiming to have found not just a deterrent effect in the data, but a huge deterrent effect (along the lines of 13-16 murders prevented by every execution, if I recall correctly). This claim was NOT based on more fine-grained empirical analysis, but rather on the usual sort of empirical game-playing that makes people distrust statisticians. (Indeed, a famous 1983 article, "Let's Take the 'Con' Out of Econometrics," used the death penalty as the prime example of how one can manipulate statistical methods to reach a predetermined conclusion.)

Does it matter that the new paper showing the big deterrent effect is unserious? Yes and no. It matters if one really wants to understand how the death penalty might affect potential murderers' behavior. If all one wants, however, is to be able to say that "there are studies out there that prove my point," then the new paper is a godsend. Sure enough, the new deterrence claims were picked up enthusiastically, not just by politicians, but by some legal scholars who were all too happy to promote the supportive result.

I understand that we should always be open to new and surprising empirical findings. That should not, however, blind us to the shortcomings of the analyses that purport to prove what has never before been proven. In any event, when it comes to the claims that austerity programs are the key to a return to prosperity, the evidence is simply not there.