Where's the Stimulus Hangover?

Copyright, The New York Times Company

Last week’s final report on gross domestic product for 2010 provides a fresh opportunity to evaluate the stimulus law passed two years ago. The data and economic reasoning suggest that the effect of government spending on G.D.P. was minimal at best.

As planned, almost all of the tax cuts and public spending increases from the American Recovery and Reinvestment Act of 2009 are finished. The Obama administration and its supporters promised that the fiscal stimulus law would create or save more than three million jobs by now. Their stated intention was to provide government spending while the economy was weak, then end the extra spending as the economy recovered.

But instead of adding jobs, employment is now about two million below what it was when the law was passed in February 2009.

Some of us think that the fiscal stimulus made a bad situation worse, and that employment would have grown, or fallen less, if the stimulus law had not been passed. The Obama administration contends that, apart from the stimulus law, the economy was in worse shape than anyone expected, and that the law kept the employment drop to two million, rather than a potential drop of more than five million.

While the increase in the stimulus by design coincided with economic weakness, the stimulus decline did not coincide with economic strength. Unemployment rates remained high, and employment, home prices and the Federal funds rate remained low as stimulus spending was winding down (as this profile of stimulus spending shows; note that we are now in the middle of fiscal year 2011).

If Keynesian stimulus advocates are correct, economic growth should have been sharply reduced when stimulus spending slowed.

I use real G.D.P. results from the Bureau of Economic Analysis to measure actual economic growth through the end of 2010. In order to compare the results with the Keynesian theory, I assume a government spending multiplier of 1.5, as the Obama administration did when it projected the impact of the law.

Such a multiplier means that each additional dollar in government spending adds $1.50 to G.D.P., and each dollar subtracted from government spending subtracts $1.50 from G.D.P.

Because we know that the economy would have been weak in the first few quarters of the stimulus regardless of the law, I do not begin the measurement until the fourth quarter of 2009, when the president’s Council of Economic Advisers declared that the stimulus law had successfully started a slow recovery.

If the advisers were right, economic growth should have increased further when government spending grew still faster in the next couple of quarters, and then economic growth should have been less as government spending grew more slowly later in 2010.

I have illustrated the Keynesian-multiplier-1.5 as a red line in the chart below, and the actual results as blue squares. The blue square at the end of the red line is the data for the fourth quarter of 2009. If the multiplier of 1.5 held up, all of the data for the subsequent quarters should have appeared on the red line. (The quarters represented by the squares are not in chronological order.)

Instead, actual G.D.P. growth ended up below the red line and, more important, the quarters with more government spending growth tend to be those with less G.D.P. growth.

Stimulus advocates lament that the stimulus law was too small and was significantly offset by shrinking state and local government spending. But my chart measures total government spending at all levels.

We can see from the chart that real government spending did in fact grow rapidly at times and grow slowly at other times: the actual growth rates range from less than 1 percent per year to more than 7 percent per year.

The red line shows that the range was wide enough to, according to the 1.5 multiplier, make G.D.P. growth rates of almost 9 percent per year (technical note: as drawn, the red line does not have a slope of 1.5 because, in terms of growth rates, the slope is the product of 1.5 and the ratio of government spending to G.D.P.).

A number of Keynesians outside the Obama administration would distinguish government spending on “transfers to individuals” from government spending on goods and services (among other things, government spending on goods and services is automatically counted in G.D.P.; transfers are not).

My chart’s green line shows an alternate Keynesian hypothetical based on a multiplier of 0.75, which might represent a smaller multiplier for transfers.

(Because the Obama administration’s original projection made no distinction between purchases and transfers to individuals – even though it knew that much federal spending would be the latter and some federal grants to state and local governments would allow those governments to make transfers – the 1.5 hypothetical is the appropriate one for evaluating their promises of stimulus results, even it is not appropriate for evaluating other Keynesian theories).

The blue squares showing actual results for our economy do not fit anywhere in the cone formed by the two Keynesian hypotheticals, suggesting that, contrary to the Keynesian promises, the stimulus law did not noticeably increase G.D.P. and might even decrease it.

After all, the stimulus spending penalized success, since its benefits — for example, extending unemployment insurance — were aimed at people and businesses with low incomes, and not at those who were working and/or achieving a certain income level. So it would be no surprise if the result was to keep incomes below what they would have been — as in other cases, a counterproductive result of a well-intentioned program.

Perhaps you think that government spending does its stimulation with a lag, but the Keynesian theories do not fit the lagged data any better. The chart below is the same as the one above except that government spending growth is measured in the quarter prior to the G.D.P. growth.

Again, the data fail to fall in the cone predicted by the 0.75 to 1.5 range of Keynesian multipliers. (Further variations on these charts provide no better results).

Recent G.D.P. growth results are just one way to attempt to measure the amount of stimulus the stimulus act provided. But the longer we go without any vivid empirical demonstration of the stimulus law’s potency, the more we are driven either to reject Keynesian theory or to accept it solely as a matter of faith.