Category Archives: Stimulus

Why the Federal Government Shouldn’t Use the States to Implement Fiscal Stimulus

[C]hannelling the stimulus package through state governments exposed it to agency costs, free-riding problem, and political expediency. As a result, the stimulus has failed to meet its objectives at the state level. The lesson is that fiscal stimulus should be conducted centrally.

That’s Robert Inman, writing over at Vox. He summarizes forthcoming research with Philadelphia Fed economist Gerald Carlino (I don’t believe the draft is on line yet). They find:

[A]n income multiplier for federal transfers to states of only 40 cents for each dollar of federal aid even after 20 quarters.

He also sums up his solo paper on states in fiscal distress:

ARRA’s assistance was largely distributed as a per capita transfer. Projected fiscal deficits did lead to more assistance, but ARRA covered at most $0.25 of each dollar of projected state budgetary shortfalls. The other important determinant of ARRA funding was whether the state’s Senators had membership on an important congressional committee making fiscal policy. Controlling for state population, deficits, and committee membership, the state’s rate of unemployment at the time of passage had no statistically significant impact on the level of assistance.

These results largely corroborate Veronique deRugy’s work.

The “Things Would Have Been Worse” Excuse

The University of Chicago’s Casey Mulligan has a great post and an interesting chart over at the NYT’s Economix blog:

Here is the thrust of his argument:

  • In promoting the Stimulus, the Administration had assumed an employment multiplier such that for every 10 people hired under the Stimulus, another 6 jobs would be created. Thus, they said that if the law passed, the unemployment rate would be down to 7 percent by now.
  • The fact that the unemployment rate is still around 10 percent means one of two things: a) the stimulus didn’t work, or b) the stimulus did work and the recession was simply much worse than the Obama Administration thought.
  • This spring, the Census hired a bunch of new workers, providing a fresh opportunity to test the fiscal stimulus hypothesis. At the same time that the Census went on its hiring binge, non-Census worker employment seems not to have budged much at all.
  • Professor Mulligan then does something very clever: He takes the Administration at its word. He assumes—as they do—that for every 10 government employees hired, another 6 jobs were created. In order to reconcile this assumption with the fact that overall employment didn’t increase much at all, he concludes that the economy must have been hit (once again!) with some extraordinary countervailing contraction right at the very time that this government stimulus was taking place. What terrible luck! 

How Long is the Long-Run?

The arguments for fiscal stimulus are essentially short-run. Most Keynesian economists concede that fiscal stimulus is not free and that we pay for it in the long run (here is one estimate of the tradeoff between short-run gains and long-run costs; here is another). Because fiscal stimulus relies on a short-run/long-run tradeoff, many in Washington are trying very hard right now to extol the virtues of short-sightedness. We should worry about boosting the economy today, their argument goes, and worry about the economic cost of that boost at some later, unspecified, date (presumably when the laws of public choice are repealed and politicians have all become far-sighted, benevolent omniscients).

A new study, however, suggests that the long-run may not be all that long in coming:

We examine the ability of the government to increase consumption by evaluating the impact of the 2009 “Cash for Clunkers” program on short and medium run auto purchases.  Our empirical strategy exploits variation across U.S. cities in ex-ante exposure to the program as measured by the number of “clunkers” in the city as of the summer of 2008. We find that the program induced the purchase of an additional 360,000 cars in July and August of 2009.  However, almost all of the additional purchases under the program were pulled forward from the very near future; the effect of the program on auto purchases is almost completely reversed by as early as March 2010 – only seven months after the program ended.  The effect of the program on auto purchases was significantly more short-lived than previously suggested.  We also find no evidence of an effect on employment, house prices, or household default rates in cities with higher exposure to the program.


Stimulating $19,000,000,000 in New State Taxes

Among the proposals the president is considering in his latest round of stimulus is an additional $50 billion in transportation spending. It is not clear, but near as I can tell from reading the reports, the money would be channeled through the states. Now, one might think that federal lump-sum grants to state governments would offset the states’ own expenditures, allowing them to cut their own taxes. In fact, according to new research, a $50 billion grant to state governments is likely to cause them to raise their own taxes by about $19 billion.

A well-documented phenomenon known as the “flypaper effect” has found that when the federal government makes grants to state governments, the latter do not decrease their own expenditures by anywhere close to the level of the grant. The idea is that the money sticks where it lands; hence the funny-sounding name. Now, new research by the University of West Virginia University’s Russell Sobel and George Crowley shows that it is even worse than previously thought. 

Prior studies had focused on state spending reactions in the year in which grants were made. But Sobel and Crowley’s research (based on data from 50 states over a 13 year period) finds that even after the federal money goes away, state and local governments tend to permanently increase their own spending and taxation in order to keep the newly-funded programs running. (If this argument sounds familiar, it is because Governor Sanford actually made it back when the last stimulus was passed).

What is the magnitude? For every $1.00 in federal grants that a state receives today, the state can be expected to increase its own future taxes by somewhere between $0.33 and $0.42 in the future. Let’s assume it is the average of this range and that future state taxes will rise by $0.38 for every $1.00 in federal aid. If this is the case, then an additional $50 billion in “aid” to the states will cause states to raise their own future taxes by about $19 billion.

There is Nothing So Permanent as Temporary Stimulus

It is interesting to note that not even the most-ardent Keynesians are willing to claim that permanent fiscal stimulus makes any sense whatsoever. That is, the stimulative effects of debt-financed spending—whatever they may be—are only fleeting. Hence, the perennial Keynesian calls for ”temporary, targeted and timely” stimulus spending. But is that the way it happens? 

Last summer, the Congressional Budget Office issued a report showing federal spending as a share of GDP skyrocketing well above its 40-year average. At the time, the CBO projection seemed to indicate that the spending would, indeed, be temporary. By 2013, the report claimed, spending as a share of GDP would be within a few percentage points of the 40-year average (though still above it).

One year later, CBO has issued another report. This one shows spending as a share of GDP remaining will above its 40-year average for the foreseeable future. Note, however, that they still show spending as a share of GDP declining somewhat in the coming years. Below, I show both projections on the same graph (click on the graph to make it larger). I plan to update this graph next summer. 


Build America Bonds: A Transfer from the Taxpayer to the Non-Taxpayer

The [Build America Bonds] program can be interpreted as a wealth transfer from the natural holders of municipal bonds, who are individual U.S. taxpayers, to corporations, pension funds, and foreign investors not subject to individual U.S. income taxes.

That is Andrew Ang, Vineer Bhansali, and Yuhang Xing in a new NBER working paper.

The BABs were a part of Stimulus II. They are a new way for municipalities to finance capital projects. Like the traditional method, BABs effectively subsidize state and local borrowing, but the mechanics are slightly different. Traditionally, the interest on muni bonds is not subject to federal taxation. BABs are taxed, but the federal government subsidizes 35 percent of the interest payment. 

(Note that under both cases, local governments receive a special privelege that private borrowers do not.) Ang, Bhansali and Xing trace out some of the less-obvious consequences of shifting the method of finance.

Can Government Put Idle Resources to Good Use?

“Big chunk of economic stimulus yet to be spent by state, local governments”

This is according to the Washington Post. They write:

The $862 billion package was divided roughly in thirds among tax cuts, aid to states and the unemployed, and investments in infrastructure, health care and other areas. The first two have delivered most of their boost, but much of the investment spending is moving far more slowly. At the end of July, nearly 18 months after the stimulus passed, more than half of the $275 billion in investments had yet to be spent.

Meanwhile, from up in Alaska, I read that that state is slated to receive a piece of the new state aid passed last week, despite the fact that it doesn’t seem to need it:

Alaska is eligible to receive $23.5 million dollars of the $10 billion total, according to the U.S. Department of Education. Alaska’s allocation comes despite a healthy revenue surplus this year that has allowed the state to actually increase school funding.

So money is not being spent that should be spent; meanwhile, other money is being spent that shouldn’t be. Stimulus defenders will say that any time a big organization spends a lot of money, mistakes are bound to be made. And they are right. But I would argue that these sorts of mistakes are far more likely when the organization in question is the government.

Remember the Keynesian story: remove idle resources from the private sector, put them to work, and the economy will grow (provided government borrowing doesn’t crowd-out valuable private investments and provided consumers don’t hold back in anticipation of future tax increases). But the whole Keynesian idea rests on the notion that governments can effectively put resources to higher-valued use than the private sector. This requires government to know what is and is not valuable; to know when and where to spend the money; and to know that – had it been borrowed by someone else in the private sector – the money wouldn’t have been put to better use. Moreover, government has to do all of this without the benefit of the signals that help the private sector allocate resources: no profits, no losses, (almost) no price mechanism.

Unemployment Insurance, Take II

In response to my post earlier this week about unemployment insurance being stimulative, Harry Moroz over at Huffington Post, makes a good point

I had cited evidence showing that—contrary to conventional Keynesian expectations—those with lower net wealth and those with lower incomes actually have lower marginal propensities to consume compared with high-wealth, high-income people.

According to Moroz, I “wholly conflated ‘the poor’ with ‘the unemployed.’ ” Fair enough. Unemployment and low income are not the same. But research by the Center for Labor Market Studies at Northeastern University shows that the correlation is extremely strong (in fact, as Veronique de Rugy has pointed out, the blogosphere has lit up lately with posts about the high unemployment rates among low-income people):


So: low income people are more likely to be unemployed; and according to the Sahm, Shapiro, and Slemrod study, low-income workers seem not to have high marginal propensity to consume. Putting these two facts together, I would be surprised if unemployment insurance were particularly stimulative.   

In my mind, the central argument comes down to three points:

  1. Keynesians will argue that transfer payments to the unemployed will—through the magic of the multiplier—lead to a boost in aggregate demand. I tend not to put a ton of stock in this because many estimates of the multiplier are relatively low and the latest estimates of the multiplier are even lower. Also, as I argued in my last post, I don’t see a lot of evidence to indicate that the unemployed or the poor have really high marginal propensities to consume (and probably no higher than those from whom the revenue is obtained through taxation or borrowing). The bottom line: We can quibble about which estimate is right but it seems that many proponents of stimulus are over-confident in their assessment that fiscal stimulus works. Given the ambiguities in both the theoretical and empirical research, I’d say a little humility is in order.    
  2. Even if we take the Keynesian multiplier arguments at face value, we must acknowledge that there are other forces at work. In the most basic economic model, if you tax work and subsidize non-work, then on the margin you should expect less work. And, indeed, numerous studies have found that increasing the length of potential unemployment benefit duration increases the average length of the unemployment spell. We may not like this result, but as Alan Blinder notes, we have to acknowledge what this is what the research shows. This must be weighed against the Keynesian result in #1 above. 
  3. The final point is a long-term one. Compared with other countries, the U.S. has significantly lower long-term unemployment rates. Moreover, the unemployed in the U.S. tend to remain so for shorter periods than in other countries. At the same time, U.S. unemployment insurance replaces a much smaller fraction of income and does not last as long (see charts below). Numerous studies have found this is no coincidence: the difference in European and U.S. unemployment experiences seems to be due to the relative dynamism of the U.S. labor market. Compared with Europe, we have relatively low taxes on labor, limited regulation of employment, and limited duration of unemployment benefits. I believe that if we really want to decrease the likelihood of unemployment and the length of the average person’s unemployment spell, then the best thing we can do is ensure that ours continues to be a dynamic labor market.  Ironically, extending unemployment benefits may very well make that more difficult. 


Is Unemployment Insurance Stimulative?

Alan Blinder has an interesting article in today’s Wall Street Journal. 

In it, he says that the Obama Administration is on the right policy track in its attempt to extend unemployment benefits, create more fiscal stimulus, and permit the Bush tax cuts to expire for people earning more than $250,000. 

He makes a claim that has become increasingly popular: policies that tax the rich and redistribute to the poor are not only compassionate, they are stimulative. There was a time when those on the left talked about a tradeoff between redistribution and growth. But Blinder and others now argue that redistribution is, on net, stimulative; that it is possible to have one’s cake and eat it too. 

Blinder begins by conceding a point to the opponents of more generous unemployment insurance. He writes:  

[L]onger-lasting benefits dull the incentive to seek work, which in turn drives up unemployment. Economic research suggests they are right.

But, he says, “one shouldn’t exaggerate the magnitudes.” Furthermore, he sees reason to believe that unemployment benefits can be stimulative. The key to this reasoning is his assertion that the poor are more likely to spend a marginal dollar than the wealthy. That’s why we can tax the wealthy, redistribute to the poor, and see a net gain.

He writes:

[C]onsider three different ways to add a dollar to the budget deficit: increase unemployment benefits by $1, give a $1 tax cut to someone earning $50,000 a year, or give a $1 tax cut to someone earning $5 million a year.

While the immediate impacts on the budget are identical, the near-term spending impacts are not. The unemployed worker struggling to make ends meet will likely spend the entire dollar right away. The $50,000 earner probably will spend the lion’s share of it, saving just a bit—that’s what most Americans do. But the $5,000,000 earner probably will save most of the new-found dollar.

Blinder is referring to the “marginal propensity to consume.” Keynesians have long-argued that the poor have higher marginal propensities to consume than the wealthy. That is, Keynesians believe that if you tax a wealthy guy and redistribute the revenue to a poor guy, the economy will actually grow in the short run. Why? The wealthy guy wasn’t going to spend that money (or at least not much of it) anyway. He was just going to let it sit in his bank account (never mind that savings makes its way into aggregate demand as investment—buy Keynesians have other stories for why that doesn’t work). The poor person, however, is different. He will go out and spend that dollar right away, leading to a multiplier in terms of growth.

In my mind, this makes theoretical sense. The problem is: it doesn’t seem to be true. And President Bush’s Stimulus I provides the evidence. Economists Claudia Sahm, Matthew Shapiro and Joel Slemrod studied the way people spent the stimulus checks that were sent out in the first half 2008. Using data from the Reuters/University of Michigan Survey of Consumers, they found that spending patterns were “strongly at odds with the conventional wisdom.” It turns out that the poor were actually less likely to spend their 2008 stimulus checks than the wealthy. What’s more, analysis of the 2001 stimulus found much the same thing.

Now there may very well be humanitarian reasons for unemployment insurance (I’ll leave it to others to debate those). But is seems to me that the data are making it increasingly more difficult to argue that redistribution through unemployment benefits is both humanitarian and stimulative.

The Government (Un)employment Effect

A few hours ago, the Obama Administration released a new report estimating that Stimulus II saved or created about 3 million jobs. Shortly thereafter, my colleague, Veronique de Rugy, testified before Congress on the impact of the stimulus. She argued, among other things, that more realistic estimates show that fiscal stimulus tends to do more harm than good.

All of this talk about jobs reminds me of one of Veronique’s recent posts:

Since the beginning of the recession (roughly January 2008), some 7.9 million jobs were lost in the private sector while 590,000 jobs were gained in the public one. And since the passage of the stimulus bill (February 2009), over 2.6 million private jobs were lost, but the government workforce grew by 400,000.

I will leave it up to you to draw conclusions.

In the spirit of drawing conclusions: one body of research suggests that the conclusions are not happy ones. A number of studies have examined the relationship between government employment and private employment, concluding that the former crowds out the latter. Using data from 19 countries over 17 years, Horst Feldmann (2006), for example, examined the relationship between a large government sector and unemployment. He found:

[A] large government sector is likely to increase unemployment. It appears to have a particularly detrimental effect on women and the low skilled and to substantially increase long-term unemployment.

What is more, Feldmann is not the only one to come to this conclusion. As he reports in his literature review:

Several empirical studies suggest that an increase in government expenditure impairs labor market performance. For example, Karras (1993) observed negative employment effects of government spending in eight countries in his sample of 18 countries. Yuan and Li (2000) came up with the same result for the US. In a cross-country study of 15 major industrial countries, Abrams (1999) found that the government expenditure ratio was positively related to the unemployment rate. Christopoulos and Tsionas (2002) examined the relationship between the government expenditure ratio and the unemployment rate for 10 European countries over the period 1961 to 1999 and found that there was unidirectional causality from government size to unemployment rate.

The magnitude of the government (un)employment effect is not trivial. Looking at a sample of OECD countries for 40 years, Algan, Cahuc, and Zylberberg (2002) found that the “creation of 100 public jobs may have eliminated about 150 private sector jobs”