Monday, August 23, 2010

It's all in the assumptions

Lawrence Lindsey explains why the study that the press and Obama administration have been relying on to show that the Democrats' stimulus plan saved jobs.
A recent paper by Alan Blinder and Mark Zandi claims that if not for the response of the federal government, the unemployment rate would be 15.7 percent, far higher than the current 9.5 percent. The press quickly reported that this vindicated the Obama stimulus plan. But the fact is that most of the positive effects cited in their paper came not from the stimulus but from stabilizing actions of the Federal Reserve, the FDIC, and TARP.

The paper argued that fiscal stimulus enacted under both Presidents Bush and Obama lowered the unemployment rate by 1.5 percentage points. But it did not measure either the number of people who found work or the effectiveness with which the Obama stimulus created jobs. Instead, it assumed through the use of economic modeling that the recently enacted stimulus was roughly as effective, dollar for dollar, as similar provisions in the past. It then multiplied the past measures of job creating effectiveness by the number of dollars in the current plan and added the result to the current unemployment rate.

This is the economic equivalent of assuming there are 1,000 angels on the head of a pin, observing that we have 10 pins, and therefore calculating that we must have 10,000 angels. The math is fine.
Pretty clever not to measure what actually happened just what was theorized to perhaps happen.

So we're back to the original argument against the way the package was structured in the first place. It was designed to be ineffective.
There are ample reasons for this lack of success. National Economic Council chairman Larry Summers argued that stimulus should be “timely, targeted, and temporary.” But the package that passed was neither timely nor targeted and today Congress is faced with making many of the stimulus programs permanent because unemployment remains stubbornly high.

The bill was not timely because the bulk of the funds were disbursed through the cumbersome government contracting process—and often made doubly complicated because the funds were then channeled through state and local governments. Weekly data collected in 2009 (since discontinued) showed a very consistent $7 billion of stimulus disbursed every week starting in the second quarter of 2009. If you are one of the 6.8 million persons unemployed for six months or more, this slow pace of disbursement is anything but timely.

Nor was the bill targeted, at least in any economically sensible way. It was written not by Larry Summers or Christina Romer, but by Democratic members of congressional appropriations committees, based on the normal political logrolling and reward process. This is the group that notoriously brought us “bridges to nowhere” in the past. The bill was, moreover, rushed through without much review or oversight. One may remember that the stimulus bill was the one that authorized the payment of bonuses to AIG executives, a fact not discovered until well after the bill was signed. This was then followed by days of publicly discussed “mystery” about how such a provision was included. Well-designed targeting would not have included such provisions.
As Lindsey notes, the actual unemployment growth is quite close to the original Christina Romer plot of what would happen, especially after he adjusted for the actual unemployment point when the stimulus was enacted. What Lindsey and others recommended 18 months ago of reducing payroll taxes for both the employer and employee now seems like a much better approach than what the Democrats actually did.
Many of us who supported the administration’s call for a stimulus in early 2009 recommended the reduction of the payroll tax for both employers and employees, something with the same net revenue effect as what was passed. Such a payroll tax cut would have provided an incentive at the margin for continued work and employment for more than 90 percent of the labor force. The tax provision in the actual stimulus that passed did so for less than 15 percent of the labor force, and the spending provisions impacted only 2 percent of the labor force even under the administration’s assumptions. That is bad targeting.
File that under "If only..."