I’ve had a number of posts recently looking at the effect of reductions in the tax burden during Presidential administrations or during (or just following) recessions and how those affected subsequent growth. In each case, cutting the tax burden did not lead to faster growth. In fact, the data shows that, contrary to theory, real economic growth tends to be slower following cuts in the tax burden than following hikes in the tax burden.
As noted in my last post , one common criticism, that there aren’t enough observations, is mooted by the fact that state and local level data seems to show the same thing. But it occurs to me that I haven’t done this sort of an analysis at the state and local data in a while, so that’s what I’m going to do in this post.
The data I’m going to use in this post comes from the Tax Foundation. Here’s how the Tax Foundation describes itself:
The mission of the Tax Foundation is to educate taxpayers about sound tax policy and the size of the tax burden borne by Americans at all levels of government. From its founding in 1937, the Tax Foundation has been grounded in the belief that the dissemination of basic information about government finance is the foundation of sound policy in a free society.
I would add to that – from what I can tell, their primary goal seems to be to convince people that lower taxes produce faster economic growth, which is something that contradicts all the data I’ve looked at so far. Thus, though I don’t normally use data from advocacy groups, I figure it adds an interesting dimension to the analysis in this case.
Anyway, the data I will use comes from this file , and it shows the state & local tax burdens and the per capita income for every state plus the District of Columbia for every year from 1977 to 2008. Despite my trepidation, I’m going to trust the data.
Now, showing results that involve three decades of data for all 50 states and the D of C (henceforth “the states”) graphically – which is what Michael Kanell and I tried to do in Presimetrics – isn’t easy. Here is what I did in this case. For every year for every state, I computed the change in the state & local burden for the two subsequent years, and the growth rate in per capita income for the six years after that. (Per capita income is not adjusted for inflation, but for the purposes I’m using it, that doesn’t matter.) Thus, in 1977, I computed the change in the state & tax burden for Alabama between 1977 and 1979, and the annualized change in per capita income from 1979 to 1985. The same thing was done for every state, and for each state, it was performed for every year for which data was available. Regular readers will recognize that I’ve used the 2 years of tax changes followed by 6 years of growth in the past when looking at data at the Presidential level.
Now, in some years, tax cutting was the norm. In other years, most states hiked taxes. So the trick is to compare relative performances. If the tax cutting proponents are right, in general, the states that reduced tax burdens the most (or increased them by the least) in any given two year period should have the fastest growth in per capita income in the next six year period. To check that, in each year, I divided states into three buckets with 17 states each. The 17 biggest tax cutters over the next two years are placed in one bucket, the next 17 appear in the next bucket, and the last 17 are placed in the next bucket. Obviously, a state won’t be in the same bucket every year. The median growth rate in per capita income in the subsequent six years is computed for each bucket, and buckets are ranked according to whether they produced the fastest, second fastest, or third fastest growth rates in per capita income.
For example… from 1977 to 1979, most states were in tax cutting mode. The 17 states that cut taxes the most are placed in the “biggest tax cutters” group, the next 17 states are placed in the “medium tax cutters” group, and the 17 remaining states (including the D of C) are placed in the “smallest tax cutting” group. Now, the biggest tax cutters produced a median annualized growth rate in per capita income from 1979 to 1985 of 7.8%. By contrast, the median tax cutters from 1977 to 1979 enjoyed a median annualized growth of per capita income from 1979 to 1985 of 7.9%. On the other hand, the smallest tax cutters produced a median growth in per capita income of 8.3% from 1979 to 1985. Thus, for the 1977 year, we rank the smallest tax cutters as first, the median tax cutters as second, and the biggest tax cutters as third.
The following graph summarizes our results for all the years available:
The table is interpreted as follows: the smallest tax cutters (i.e., those who cut taxes the least or raised them the most in any given year) produced the fastest economic growth about 42% of the time, second about 29% of the time, and third 29% of the time. By contrast, the middle group came in first in one third of all occasions. The ”tax cuttingest” group came in first place a mere quarter of the time; in half of all years, it came in last place.
Many conclusions are reasonable from this. However, concluding that states that cut taxes have produced the fastest growth rate in per capita income is not one of them. After all, in general, the states that cut tax burdens in a two year period have, as often as not, turned in the worst performance when it comes to growth in per capita income in the following six years.
Tying everything back to the work I’ve been showing in recent months (and which supports our findings in Presimetrics) – looking at data over the length of Presidential administrations since 1929 (when national accounts data became available), those Presidents that cut tax burdens in the early years tended to have worse real economic performance in later years than those that raised tax burdens. Similarly cutting tax burdens during or just following a recession produces slower, shorter recoveries. Now we find that cutting tax burdens isn’t the prescription at the state level either.
I’ll repeat something I’ve noted before – any economic theory for which just about every observation is a special case is wrong.