by Mike Kimel
Corporate Tax Rates and Unemployment
Cross posted at the Presimetrics blog.
Update: Reader JzB has noted that a large drop in the top marginal rate that appears in graph two may not have occurred. At this point I think I made a transcription error in pulling the data. I will check the numbers when I get a chance and redo the post. Apologies.
I’ve been kind of swamped, low on sleep, and doing a few book related things in my few waking hours that don’t work or parenting (buy a copy of Presimetrics!!!!), so posting has been light. But I thought I’d do a quick and dirty post today about a hot topic – the effect that taxes on businesses have on unemployment. The usual argument is that the lower the taxes on businesses, the more money they keep and pump back into, well, doing business, and thus, the more people they end up hiring. But is it true?
Now, since the talk right now is about cutting payroll taxes in particular, ideally I’d use that data. However, in a quick perusal at the IRS’ site, all I found was the corporate marginal tax rate. However, the folks who suggest tax cuts as a way to boost hiring aren’t particular – most of them feel any tax cuts will lead to more hiring. So let’s check that, at least, shall we?
Figure 1 below shows the data used in this post; the top corporate marginal tax rate (obtained from the IRS) is on one axis and the unemployment rate for individuals sixteen years and over (from the Bureau of Labor Statistics) is on the other axis. The latter series begins in 1947, but I decided to start with 1948 just to be far enough off from WW2 to avoid that effect as much as possible.
Now, consider the correlation between the top corporate marginal tax rate and the unemployment rate. If it is true that lower taxes = lower unemployment, the correlation between the two series should be positive. A positive correlation means the series should move more or less in the same direction; as tax rates rise, unemployment rises, and as tax rates fall, unemployment falls.
If the correlation is, in fact, negative, that means that lower unemployment tends to happen when tax rates are higher. Correlation may not be causation, but it would be very hard to argue that cutting taxes on corporations leads to lower unemployment if we do not see a positive correlation between the two series.
Now, obviously, it may take time for tax rates to do whatever magic they might have. So Figure 2 looks at the correlation between the top corporate marginal tax rate and the unemployment rate in the same year, the unemployment rate the next year, etc., all the way through ten years out. Its really hard to see how the effect of tax rates should last beyond a couple of years, but I figured I’d be thorough and put up the figures. I’ll take a pass at interpreting them, but feel free to reach your own conclusions.
Additionally, because whatever effect tax rates might have on unemployment might change over time, each correlation is computed several times: once for the entire 1948 – 2009 sample, a second time for 1960 – 2009, a third time for the period since 1970, a fourth for the period from 1980 and a fifth time for the 1990 – 2009 period. (I didn’t look at just post-2000 because the top corporate rate has been frozen during that period.)
So what does this say? Here’s my interpretation. If you look at the entire data set, the correlation between tax rates and unemployment appears to negative. That is to say, lower unemployment rates tend to be associated with higher corporate tax rates, not lower ones. Fast forward to the 1970s, and you see some of the oft-stated effect; higher taxes do seem to be associated with lower unemployment, and that effect becomes even stronger when you focus on the 1980s and beyond. However, the whole thing falls apart when you move into the 1990s and beyond, when the correlation is clearly negative, at least for the early years. The effect in out years is positive, but once again, its hard to see how the effect would lag that long. If I had to guess, the 1970s and 1980s were an aberration but I don’t have time to develop that thought right now.
My second interpretation… if I wasn’t some random guy who enjoys looking at data in an attempt to understand the world, but rather someone determined to reach a certain pre-determined conclusion, I’d pick my time sample very, very carefully.
How do you see it?