More Evidence on Why the Stimulus Didn't Work

Polls show that most Americans don't believe that the stimulus package worked, but debate continues among economists. The most debated issue is the size of government purchases multiplier. Suppose that the government purchases multiplier is 1.5. Then Economics 1 students learn that the change in GDP due to an increase in government purchases is found by multiplying the change in government purchases by 1.5. That is:

change in GDP =1.5 times change in government purchases

Government purchases include spending on items such as infrastructure, law enforcement, and education, but do not include interest and transfer payments. (A derivation is in Ch. 23 appendix of the Taylor-Weerapana principles text.)

The example of 1.5 is at the upper range of estimates, and was used in a paper by Christina Romer and Jared Bernstein to estimate the impact of the American Recovery and Reinvestment Act of 2009 (ARRA). However, John Cogan, Volker Wieland, Tobias Cwik, and I found that the multiplier in the case of ARRA was much smaller, around .7. Robert Barro argues that it is zero. So there is debate.

But few have focused on the second term in the above multiplier formula: the change in government purchases due to ARRA. John Cogan and I have been tracking data on the changes in government purchases since ARRA was passed, using a new data series provided by the Commerce Department. We just finished a working paper reporting the details of our findings, which provide additional evidence that the stimulus has not worked and, just as important, on why it has not worked.

Despite the gigantic $862 billion stimulus package, the change in government purchases due to ARRA has been immaterial to the economic recovery: government purchases increased by only 2 percent of the $862 billion package ($18 billion). Infrastructure was even less at $2.4 billion. There has been almost no change in government purchases for the multiplier to multiply. It’s no wonder people don’t think the stimulus worked. And the size of the multiplier is largely irrelevant!

Our research looks at both federal and state and local purchases. Federal purchases due to ARRA reported by the Commerce Department are very small. We also find that large ARRA grants to the states did not increase state and local government purchases at all. To check our results we traced where the grant money went (it went mainly to reduce state borrowing) and we considered counterfactuals (in the absence of ARRA, government purchases would likely have been higher). The chart below summarizes the findings with all the government purchases at the federal level. The implication of this research is not that the stimulus program was too small, but rather that such countercyclical programs are inherently limited by feasibility constraints of the federal system. John Cogan and I first reported these results in a preliminary way over a year ago in a September 16, 2009 a Wall Street Journal article with Volker Wieland, entitled “The Stimulus Didn’t Work”. Though ARRA data were then only available through the second quarter of 2009, it was clear to us that government purchases were not contributing to the recovery, and we reported that “there is no plausible role for the fiscal stimulus here.” Many dismissed our conclusion, saying it was too soon to judge. Another year of data has confirmed our results as we explain in our new working paper.

A New Normal for Monetary Policy?

A year and a half ago when the Fed’s extraordinary quantitative easing (QE) was shifting from emergency liquidity programs to large scale asset purchases, we convened a conference at Stanford’s Hoover Institution to discuss the shift. Jim Hamilton, of UC San Diego, in his talk Concerns about the Fed's New Balance Sheet and Peter Fisher of Blackrock in his talk The Market View expressed serious concerns about the extraordinary policies and the use of the Fed’s balance sheet to finance them. Don Kohn, then Fed Vice-Chair, attended and defended the Fed’s position

One concern expressed at the time (March 2009) was that such extraordinary measures would become a "new normal" for monetary policy, in which the Fed would not restrict its massive doses of QE to times of panics and other emergencies. Such a new normal would likely breed uncertainty and reduce the Fed’s independence, eventually leading to economic instability and inflation. I put it this way in my paper in the book, Road Ahead for the Fed, which came out of the conference:

“The danger I see is that as the recovery begins, or after we are a couple of years into it, people may feel that it’s not fast enough, or there is an unpleasant pause. Either could generate heavy pressure on the Fed to intervene…. Why would such interventions only take place in times of crisis? Why wouldn’t future Fed officials use them to try to make economic expansions stronger or to assist certain sectors and industries for other reasons?”

Many Fed officials dismissed the concerns about such a scenario, saying that the crisis was unique. Yet this is exactly the scenario that is now playing out. Sure enough, the recovery paused, and lo and behold, there is a QE2 in the works.

Today’s Roubini Global Economics newsletter is ominous. It predicts that after QE2 the Fed will “announce QE3 (and eventually even QE4).” After Road Ahead for the Fed we published another book Ending Government Bailouts as We Know Them. Perhaps the title of the first book should have been The End of Monetary Policy as We Know It

Cash for Clunkers in a Macro Context

More empirical studies are demonstrating that temporary fiscal stimulus actions are a poor way to get the economy moving again on a sustainable basis. More permanent and predictable policies are much better. I demonstrated this problem in the case of the stimulus packages of 2008 and 2009, but perhaps the clearest case is the “cash for clunkers” program of July and August 2009. To illustrate the problem quantitatively, it is useful to put some new micro-empirical results of Atif Mian and Amir Sufi into a macroeconomic context, as I do with the following charts.

The first chart shows disposable personal income along with personal consumption expenditures in the United States. In previous work I focused on how the two big bulges in disposable income due to the stimulus programs failed to jump start consumption. But here I focus on the cash for clunker program, which clearly did change consumption.
Using the Mian-Sufi results, which are based on a comparison of different regions of the United States, I estimated the amount by which total personal consumption expenditures first increased as people were encouraged to trade in their clunker and purchase new cars, and then declined because many of the trade-ins were simply brought forward. To make this increase and subsequent decrease easier to see, the second chart focuses on personal consumption expenditure during the period of the program. You can see that consumption rises above what it would have been without the program and then actually falls below what it would have been. Some argue that bringing forward purchases like this is exactly what such programs are supposed to do, but the graph makes it very clear that the offsetting secondary effects occur so quickly that the net result is an insignificant blip in the recovery. The impact is not sustainable.
An important result of Mian and Sufi is that the positive effects are completely offset in a few months, as you can see in the picture. But even if they were not offset, the graph raises serious doubts about how such a program could sustain a recovery. Suppose that the red line never dipped below the blue line. We would still see growth simply picking up for a month and then slowing down again. That is not sustainability.

Beware of Announcement Effects When Assessing Policy Interventions

The effects of exchange market interventions are frequently estimated by looking at what happened on the day of the announcement of the intervention or of the intervention itself. But my observation—based in part on experience running the international division of the U.S. Treasury and engaging in financial diplomacy with Japan and other countries—is that announcement effects can be very misleading as an estimate of the overall effect of interventions because the impacts can wear off with no announcement or reverse interventions. The recent experience with the September 14 announcement of an exchange market intervention by the Bank of Japan is an important case in point as clearly indicated by this chart of the yen-dollar exchange rate. The yen did noticeably depreciate against the dollar on the day that the intervention was announced and took place, but that has already been reversed.

This is one of the reasons why I think it is unwise to rely on announcement effects to assess the impact of central bank asset purchase programs as in Gagnon et al. Better to look over longer periods of time where you can control for other factors as in this paper with Johannes Stroebel.

Meltzer’s History Lesson

I recently had the pleasure of reading, and then writing a review of, Allan Meltzer’s monumental A History of the Federal Reserve, Volume 2 which is published in Book 1 and Book 2. The lesson from this thorough 2,112-page history (volumes 1 and 2 together) deserves careful consideration by policymakers today.

It is a history of policy successes and policy failures. The failures are the Great Depression of the 1930s, the Great Inflation of the 1970s, and the Great Recession of recent years. The successes are the Great Disinflation of the early 1980s and the Great Moderation which succeeded it. What caused these successes and failures? Meltzer focuses on two types of policy errors: (1) succumbing to “political interferences or pressure” and (2) basing policy on “mistaken beliefs.” Failure comes from making one or both of these errors; success comes from avoiding them.

He argues that the Great Depression was mainly the second source of error: mistaken beliefs about the real bills doctrine. The Great Inflation was a combination of both types of errors, but failure to resist political pressure dominated because when beliefs changed in the 1970s, policies did not. The Great Disinflation was marked by an absence of both types of errors as Paul Volcker regained independence and restored basic monetary fundamentals about the impact of changes in the money supply and interest rates. The Great Moderation was a period where independence was solidified and rules-based policy, grounded in fundamentals, was followed. The Great Recession was a return to a combination of both kinds of errors, a departure from rules-based policies that worked in the Great Moderation and a loss of independence as the Fed engaged in fiscal and credit allocation policy.

Meltzer’s historical research thus leads him to conclude from the past that “Discretionary policy failed in 1929-33, in 1965-80, and now,” and to recommend for the future that “The lesson should be less discretion and more rule-like behavior.” While I registered some disagreements with parts of Meltzer’s history in my review article, I think his overall conclusion and recommendation are largely correct.

Trading Places: HIPCs and HIICs

I thought of the movie Trading Places when I saw the term HIIC in the headline of today's Wall Street Journal article by Kelly Evans. The new term refers to the "Heavily Indebted Industrialized Countries" and of course to the exploding debt of these countries--including the United States. It was not so long ago that the main concern in the international community was the debt of the "Heavily Indebted Poor Countries," or the HIPCs; these low income countries were the focus of the debt relief, or the "drop the debt," movement.

Remarkably the debt of the advanced countries is now higher and growing more rapidly than the debt of the lower income countries, as I show in this chart based on data from the IMF's Fiscal Monitor of last May. The switch seemed to take less time than it took to change the P to an I. It's good news for the lower income counries, but not such good news for the industrialised countries which obviously have to get back on track.