Thursday’s WSJ opinion article “Stimulus Spending Doesn’t Work” should be titled “In the long run, stimulus spending doesn’t work because we are all dead.” The analysis of stimulus over averaged periods by Robert Barro and Charles Redlick seems to ignore Keynes famous guidance on economic analysis: “The long run is a misleading guide to current affairs. In the long run we are all dead.”
The article looks at three types of stimulus: Defense Spending, Non-Defense Spending, and Tax Cuts. There seem to be errors in each analysis but I don’t have access to the actual work so I could have the details wrong. Apologies in advance and an invitation to the authors for corrections.
First of all, Defense Spending is not intended for stimulus, so including it in an analysis of stimulus spending is wrong unless your intent is to prove the obvious. They probably proved that you should not increase defense spending if your intent is stimulus, but they averaged in times that just don’t apply to today. WWII should not have been included in this analysis as it was a nondiscretionary event involving total takeover of the economy by the government and conscription of every able bodied man 18 to 26 into the military or military activities. Modern defense spending is discretionary, at least at the margin where stimulus spending would take place so Vietnam, Cold War and Iraq spending are most representative of the effects of increased defense spending. These should be analyzed as discrete events to determine if starting a war gives more stimulus that just building non-productive weapons.
They stated that non-defense spending was more important than defense spending as stimulus and then dismissed analysis of it as “problematic” because government spending goes up with GDP increases. This is a surprising problem for Dr. Barro to have because he is a trained physicist as well as an economist and must be familiar with the concept of different physical states. It would be just as problematic to try to judge the properties of water without measuring the temperature that determines if it is ice, liquid or steam.
Once you recognize that the economy has different states, it is a pretty easy issue to get around and handled everyday in economic research. The Romers did a nice job in their analysis of the effect of taxes (see link below) by looking at the legislative reason for the tax change. One even easier method is comparing non-defense deficit spending as a percent of GDP to percent GDP growth. I’ve previously done a chart (below) of the total budget deficit (including defense) vs. % GDP growth and you can see a strong mirror image pattern that indicates strong negative correlation. This means that as total deficit stimulus spending becomes more negative, GDP grows faster. Removing defense spending should make this analysis indicative of the stimulative effect of just non-defense spending.
(Note: I originally did this analysis to point out that excessive deficit spending during a time of full employment from 2003-2007 seems to have resulted in the current recession. Too much of a good thing is bad.)
The tax analysis contains the biggest misuse of averages. The tax analysis looks at only the period from 1950 to today during which top tax rates were being cut from 84% to 35% on income and 25% to 15% on capital gains. FICA did increase from 2.2% to 10.8% but it covers only part of earned income so it is offset by income tax cuts. During this period, the marginal tax rate generally moved down and GDP generally grew so this is going to generate a correlation. This correlation is meaningless if the Great Depression and WWII is left out as this is when taxes were raised to high levels. It sounds like they did look at of discrete periods in this analysis but found reasons to discard some periods. The Romers in The Macroeconomic Effects of Tax Changes: Estimates Based on a New Measure of Fiscal Shocks did a better job of looking at the effect of tax changes during the post 1950 period though they did not directly address the issue of the general decline in tax rates.
The real bottom line is that because averages were used and data was excluded, the analysis does not indicate anything. The data they have may hold the secret to stimulus spending but they have not mined it yet.
The day this article came out, the market fell 2.5% from the sudden realization that things were not as good as thought. The claim in the WSJ that stimulus does not work eliminated the only tool left to moderate the decline and probably made the market decline worse.
Disclosure: None for this article