My previous post on multipliers (http://oxdown.shadowproof.wpengine.com/diary/3484) looked at the evidence, and gave references, and asserted that there is no evidence that tax cuts were a surer or quicker stimulus than spending. We are not sure which has the “bigger” effect. Taken over their life whole course, all we know is that they are about the same magnitude, on average, over the last 30 to 50 years. But the pattern did seem to be clear that spending cuts provided a bigger initial impact, and their peak effect was bigger. The US had a very recent experiment with a tax-based stimulus that indicated it would be weaker that statistical estimates indicate because people are very eager to reduce their debt burden and most of the first round of the stimulus is saved, rather than spent.

Therefore I thought the weight of the evidence says that spending is the most reliable and appropriate stimulus.

Menzie Chinn of the Econbrowser blog kindly provides a discussion of the Great Multiplier Debate(http://www.econbrowser.com/archives/2009/02/why_canat_we_al.html) that I recommend for the confused. I don’t think he gets to the core of the disagreements between macroeconomists (IMHO, the short answer is “No, macroeconomists cannot all get along because of deep differences in assumptions about how the economy works and methods" ”). I am working on how to explain that in a simple way.

Chinn presents his favored model (from the IMF, so it has all sorts of bells and whistle built in for the conservative economists out there) shows that spending will provide a bigger, quicker and more reliable stimulus. These model based estimates can go into more detail than purely statistical estimates. It is Chapter 2 from the IMF World Economic Outlook (http://imf.org/external/pubs/ft/weo/2008/01/pdf/c2.pdf)

The model’s estimates show that public infrastructure investments are expected to have a bigger and quicker impact than either spending in the form of transfers, or labor tax cuts, assuming either that the Fed raises or does not raise interest rates (though the later is by far the likelier scenario).

One reason that some economists have expressed doubt that a stimulus might work as expected is that statistical estimates show a much weaker effect after 1980 than before. No one really knows why. The IMF report quotes results from more work by Roberto Perotti to demonstrate this reduction in effectiveness. If this stimulus works as statistical estimates after 1980 indicate, spending may have very small effects but tax cuts will not. The bottom line, again, I think a spending stimulus is by far the safer course because it still provides some small stimulus while tax cuts may reduce GDP, which would be a very dangerous effect right now.

Perotti did look into possible reasons for the decline in the effectiveness of fiscal policy, and concluded that it was most likely due to a change in the response of interest rates to fiscal policy after 1980 (I think that is mainly due to changes in monetary policy). Assuming interest rates don’t respond in this crisis as they have over 1980 to 2000 (a good assumption for short run), then the response, assuming results from the sample after 1980, is about half that of before 1980. So, lower than expected but still better than tax cuts.

The estimates of the fashionable statistical methods used by Perotti and others are also sensitive to the policy environment in ways that cannot be modeled easily. A big thing that happened after 1980 that falls into that category: the increasing use of automatic monetary policy rules by the Fed (the Taylor Rule, for example). That probably makes the Perotti estimates themselves too low, and Perotti may not be able to fix things by simply saying “OK, I can fix things up by running simulations so that more spending does not increase the interest rate” The results may still be too low. So there is another reason to suspect that recent estimates of the effectiveness of fiscal policy is too low.

BTW, the IMF model Chinn discusses was partly based on Perotti’s and similar estimates.

So, the weight of the historical evidence still says, IMHO:
Public spending on infrastructure > transfer payments > tax cuts
(I would extend infrastructure to spending on human capital, such as health and education).

This conclusion points towards more aid to state and local governments, many of whom are putting maintenance and new investment projects on hold due to lack of funds.

I will repeat my disclaimer that I am an economic statistician, not a macroeconomist. References and links not found in the text are at the bottom.

PS: I am donning my most blandly objective possible self in my posts on the economic debate. As a normal person and citizen I am shocked by some of the stuff that is going on. Other than to say Newsweek prints horrible economic stories and columns so misleading and simple minded as to be harmful to the public discussion, I will stifle myself.

The Perotti estimates mentioned above are from
Estimating the effects of fiscal policy in OECD countries, Roberto Perotti
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=637189
(a free version, almost identical of final one)

Note: edited for my ever present typos.

wesgpc

wesgpc

5 Comments