I began this blog at the end of March after the stimulus bill had become law. I had been struck by how much the stimulus debate had focused on whether the bill was efficient. (It clearly was not.) There was much less discussion of whether the stimulus would be effective, and of the timing of the macroeconomic boost.

Everyone wants to know when the U.S. economy will start growing. I will focus on a related question: when will the stimulus law begin to have a significant positive effect on U.S. economic growth? And could it have come sooner if the Administration had done something different?

I believe the Administration made an enormous mistake in its legislative implementation of the stimulus. As a result, the boost to GDP will come six to nine months later than it needed to (maybe more). Given the President’s desire to do a large fiscal stimulus, and given his policy preferences, he could have had a different bill that would have been producing significant GDP growth beginning now, rather than in the middle of next year. That’s a huge mistake with real consequences for the U.S. and global economies.

To illustrate this point, let me classify four types of fiscal stimulus:

  1. a permanent tax cut;
  2. a temporary tax cut;
  3. one-time checks to people independent of their tax liabilities; and
  4. increased government spending through federal and state bureaucracies: infrastructure, energy spending, etc.

There is of course a fifth option: no fiscal stimulus law.

If you’re going to do a fiscal stimulus (big if), the best kind is a permanent tax cut. It is effective, efficient, and fast:

  • effective – People spend a large proportion of a permanent tax cut. This is derived from Milton Friedman’s “permanent income hypothesis.”
  • efficient – People spend their own money on themselves, so they waste very little of it, and they spend it on things that matter to them. Again, see Milton Friedman.
  • fast – Checks are delivered quickly, and people spend most of their own money soon after they get the check.

This was part of the short-term logic behind the 2003 tax cut, which we designed to foster both short-term and long-term economic growth. I also have a strong general policy preference for lower taxes rather than more government spending, but that’s a separable question from how it works as short-term stimulus.

In 2008 we knew we could not get a Democratic Congress to enact a permanent tax cut. Q: Do you then go for a temporary tax cut, or do nothing? The President thought the risks of an economic slowdown in 2008 were significant enough that it made sense to pursue a (second best) temporary tax cut with the Congress.

Like the 2003 law, the 2008 law got the bulk of its short-term GDP boost by advancing tax refunds from the year to come, and delivering them as checks from the IRS to taxpayers. As in 2003, the checks were delivered to taxpayers in the summer (mid-June to early-August), and consumers immediately started spending a portion of their rebates.

Because the 2008 law was a temporary tax cut, taxpayers spent a smaller proportion of it than anyone would have liked. While designing the law, we assumed about 1/3 would be spent, and much of that fairly quickly. The rest would be saved, which is also good but doesn’t help short-term GDP growth. Economists agree that GDP in Q3 and Q4 of 2008 was higher than it otherwise would have been because of the 2008 stimulus law. It was efficient, fast, yet only partially effective, with a smaller GDP boost than we would have liked:

  • efficient – People were again spending their own money on themselves. You get very little waste, and people know what they want and need.
  • fast – Checks were delivered quickly, and much of the spending that did occur happened in Q3, with some in Q4, and with very little left by Q1 of 2009.
  • only partially effective – Because it was a temporary tax cut, people saved a lot of their checks, as we expected. Still we got a GDP bump in Q3 and Q4, and in retrospect we certainly needed it.

The 2008 law was mostly (2) from my list above – a temporary tax cut. Some of the money went to (3), checks to people who didn’t pay income taxes. This was necessary to reach a compromise with a Democratic Congressional leadership that placed a high priority on the distributional effects of the law. Speaker Pelosi insisted that poor people who owed no income taxes still get “rebate” checks, and that high-income taxpayers get nothing. So the 2008 stimulus law was mostly (2) with a little bit of (3).

Now fast forward to January of 2009, when President Obama proposed an enormous fiscal stimulus. The President’s mistake was in largely deferring to Congress on the composition of the stimulus bill. Rather than allowing Congress to pump hundreds of billions of dollars through slow-spending and inefficient bureaucracies, the President should have insisted that Congress instead send all the funds directly to the American people and let them spend it quickly and efficiently. Given his policy preferences, he could have directed a large share of those funds to poor people who don’t pay income taxes. He could have again mislabeled these payments as “tax cuts,” or just correctly labeled them as one-time entitlement payments. I would not have liked that policy, but it would have generated a faster macroeconomic boost than what he allowed Congress to do instead.

Let’s compare the two scenarios. The enacted 2009 stimulus is:

  • effective (eventually) – Most of the spending through government bureaucracies will (eventually) increase GDP. Some of the funds transferred to State governments will be used to offset State spending or tax cuts that otherwise would have occurred, so there’s a loss. But clearly the proportion of the $787 B that will eventually increase GDP will be high, and much higher than if all the funds were given to individuals and families.
  • inefficient – It will be inefficient in two senses. The spending represents the policy preferences of legislators (and all their ugly legislative deals and compromises), rather than the choices of hundreds of millions of Americans who presumably know better how they would like money spent on them. The spending will also be wasteful, and we are starting to see signs of this in the press.
  • s-l-o-o-o-w – CBO says that $25 B of spending had gone into the economy by May 22nd. That’s less than 4% of the total budgetary impact of that bill. Other news reports suggest that about $40 B is in the economy if you include the revenue side. Remember that almost all of the 2008 stimulus was in private hands by August 1. We will get very little GDP boost from fiscal stimulus in Q3 of 2009, and not much in Q4 either. The stimulus will begin to ramp up in Q1 of next year, and be in full swing by Q2 and Q3 of 2010.

Had the President instead insisted that a $787 B stimulus go directly into people’s hands, where “people” includes those who pay income taxes and those who don’t, we would now be seeing a stimulus that would be:

  • partially effective but still quite large – Because it would be a temporary change in people’s incomes, only a fraction of the $787 B would be spent. But even 1/4 or 1/3 of $787 B is still a lot of money to dump out the door. The relative ineffectiveness of a temporary income change would be offset by the enormous amount of cash flowing.
  • efficient – People would be spending money on themselves. Some of them would be spending other people’s money on themselves, but at least they would be spending on their own needs, rather than on multi-year water projects in the districts of powerful Members of Congress. You would have much less waste.
  • fast – The GDP boost would be concentrated in Q3 and Q4 of 2009, tapering off heavily in Q1 of 2010.

Why did the President not do this? Discussions with the Congress began in January before he took office, and he faced a strong Speaker who took control and gave a huge chuck of funding to House Appropriations Chairman Obey (D-WI). I can think of three plausible explanations:

  1. The President and his team did not realize the analytical point that infrastructure spending has too slow of a GDP effect.
  2. They were disorganized.
  3. They did not want a confrontation with their new Congressional allies in their first few days.

I think the Administration now recognizes this problem. Last month when they released a CEA paper “Estimates of Job Creation from the American Recovery and Reinvestment Act of 2009,” the paper danced around the timing of job growth and government outlays in 2009 and 2010. Tips for reporters: (1) ask the Administration to give you OMB estimates of quarterly cash flows for the stimulus law, and (2) ask them to give you the quarterly GDP and job growth estimates behind this CEA paper. I know the first one exists, and I’d bet heavily the second does as well.

Fortunately, CBO Director Doug Elmendorf just gave a presentation titled “Implementation Lags of Fiscal Policy” to the IMF’s conference on fiscal policy. All of the following data are from his presentation.

The final 2009 stimulus law broke down like this:

10-yr total

% of total

Discretionary spending (highways, mass transit, energy efficiency, broadband, education, state aid)

$308 B


Entitlements (food stamps, unemployment, Medicaid, refundable tax credits)

$267 B


Tax cuts

$212 B



$787 B


The problem is that only 11% of the first line (discretionary spending) will be spent by October 1 of this year. In contrast, 31-32% of the entitlement and tax cuts lines will be out the door by that time. (I have questions about the speed of the entitlement part. The bulk of that is Medicaid spending, and it’s not clear to me that a Federal payment to a State means the cash is immediately flowing into the private economy.)

If we extend our window to October 1, 2010, then less than half the discretionary spending will be out the door, while almost 3/4 of the entitlement spending and all of the tax cuts will be out the door and affecting the economy. The largest part of the stimulus law is therefore also the slowest spending part. This is fine if you’re trying to increase GDP growth over the next 2-4 years. If you’re going for short-term GDP growth, it makes no sense.

Director Elmendorf drills down further into discretionary spending and shows that defense spending happens quickly, highways and water extremely slowly:

  • If you allocate $1 to defense spending, 65 cents has been spent within one year.
  • If you allocate $1 to highway spending, 27 cents has been spent within one year.
  • If you allocate $1 to water projects, only 4 cents has been spent within one year.

In fact, the infrastructure spending in the stimulus law will peak in fiscal year 2011, which goes from October 1, 2010 to September 30, 2011. That’s too late from a macro perspective.

The Director further points out that the 2009 stimulus law created many new programs. This slows spend-out, as it takes time to create and ramp up the new programs.

The Administration has made much of working with federal and state bureaucracies to find “shovel-ready” projects to accelerate infrastructure spending. All of my conversations with budget analysts suggest this claim is tremendously overblown, and Director Elmendorf asks, “Is this practical on a large scale?”

The 2009 stimulus law will increase U.S. economic growth. But the actuals are matching the budget analysts’ projections for the speed at which that effect will occur.

I would not have liked a stimulus law that would have given cash to people who didn’t pay income taxes. But from a macroeconomic perspective, we need the faster economic growth now. Had the President and his team insisted on giving money to people (taxpayers or not) rather than to bureaucracies, we would be seeing a huge growth spurt in Q3 and Q4 of this year.

It is sad that instead we have to wait until the middle of next year because the White House deferred to Congressional desires to spend on infrastructure. This strategic mistake was avoidable, and the recovery will be delayed because of it.