Last Friday, I wrote here on the "uselessness of economic development." Using The New York Times’s new database on state and local economic development incentives, I found no association between these incentives and key measures of economic performance and found virtually no association for the fifty states.
One thing is clear about this topic: it gets people talking. The Times’s series itself sparked an onslaught of debate, and my analysis here on Cities prompted a lively debate in the comments as well as some illuminating email exchanges from various experts in the field.
Kenneth Thomas, a leading expert on economic development incentives at the University of Missouri-St. Louis who has written two books on the subject, took issue with some of the items the Times included in its $80 billion estimate for incentives. The Times database, Thomas wrote to me in an email, "which is the basis for [its] $80 billion a year estimate (my most recent is $70 billion), is badly flawed because 5/8 of that total comes from sales tax breaks, which for the most part prevent tax cascading rather than representing a subsidy. I personally think some sales tax exemptions do count as subsidies because they are specific (to an industry, for example), but some economists think the right answer is that no sales tax breaks are subsidies." He also wrote a post on this, and directed me to another, by economic development expert Timothy Bartik of the W.E. Upjohn Institute for Employment Research, questioning similar aspects of the Times estimate.
Fortunately, the Times's database is sortable by incentive type. So, my Martin Prosperity Institute team recalculated the incentives, this time subtracting the category for "sales tax refund, exemptions or other sales tax discounts." This is the largest single category of incentives, as Thomas notes, mounting to $51.4 billion dollars. Taking it out leaves an estimated $29 billion in total incentives. The MPI's Zara Matheson mapped these revised figures for total incentives and incentives per capita.
The first map charts the total incentives across the U.S. Texas again leads with $4.2 billion in incentives. California is second with $3.8 billion, followed by New York with $3.2 billion, Michigan $1.8 billion, Louisiana $1.7 billion, Pennsylvania $1.4 billion, and Kentucky $1.3 billion.
The second map charts the revised figures for incentives per capita. Louisiana leads at $386.03, and Kentucky is second ($299.18). Michigan is third ($183.92), followed by West Virginia ($182.99), Texas ($166.90), New York ($164.95), D.C. ($150.08), Massachusetts ($148.08), Washington ($132.17), Connecticut ($131.15), Pennsylvania ($111.78), California ($101.63), and Wisconsin ($100.37).
The MPI’s Charlotta Mellander ran a new correlation analysis for these revised figures on incentives. The analysis remains preliminary. There are missing values for some states, and as usual, I reiterate that correlation does not mean causation. Two things stand out.
First off, the states that spend more on incentives spend more on all types of them. Our revised estimates are highly correlated with the original Times figures (with a correlation of .82). The correlation for the two types of incentives on a per capita basis is also statistically significant, though somewhat lower (.56).
And, second, incentives still do not have any meaningful relationship to the economic performance of states. The key findings of our analysis remain the same as before. Even when we take out sales tax and related tax refunds, we find no relationship between incentives and any meaningful measure of economic performance. As before, there is no statistically significant correlation to economic output per capita, none with wages, none with income, and none with educational attainment, measured as college grads as a share of adults.
The correlation we previously found between incentives and the poverty rate now disappears when we used revised figures for incentives without tax refunds. In reviewing these revised findings, Thomas wrote in an email to me that while the Times database still has some limits, they probably would not affect this revised analysis.
The only caveat here is that the database has very uneven coverage of local incentives, so you are measuring state only for some states vs. state/local for others. This probably won't change things too much, as there are only two states I know of for sure where local incentives are higher, Missouri and California.
This provides additional evidence of the inefficacy — what I dubbed the "uselessness" of — state and local economic development incentives.