A new report by Tim Bartik and George Erickcek at the Upjohn Institute provides a novel approach to estimating the effects of the Michigan Economic Growth Authority's tax incentives, and perhaps other fiscal incentive programs. Unfortunately, their findings fail to provide an accurate assessment of the program's efficacy.
Their study was performed largely due to their dissatisfaction with earlier studies by the Mackinac Center and the Anderson Economic Group. I am a co-author of two of these studies, and have reviewed the new Upjohn Institute study.
The Center has published two empirical analyses of the MEGA program, one in 2005 and one in 2009. The first study used a "fixed effects" model — commonly used in economic literature — to measure MEGA's impact. The second analysis used what's known as a "shift-share" examination that included some subsequent statistical regression work. Both concluded that MEGA is ineffective.
In their new study, Bartik and Erickcek dismissed our lengthy and detailed analysis as "flawed" with a short commentary that counties in Michigan that had "low MEGA usage" within their borders made a poor "control group."
In other words, all of Michigan suffered economically, and the counties in which the largest levels of MEGA grants were provided were big losers of jobs. Of course, Bartik and Erickcek are right to be concerned about this potential problem, which economists call "endogeneity," but not in the case of these studies.
First, we specifically corrected for endogeneity in our control group through our use of the instrumental variable model approach. This technique has been frequently employed by scholars published in other peer-reviewed journals. It is used to mitigate the fact that we don't have the luxury of double blind experiments. Economists have used this technique for nearly 30 years.
Second, we used a counterfactual approach, which means we tested what happened to construction expenditure in the counties receiving MEGA dollars. If our study suffered this endogeneity problem, we might find that the MEGA dollars did nothing to construction spending, especially during Michigan's decline. However, MEGA dollars did indeed spawn new construction jobs, at the cost of about $123,000 in tax credits offered per temporary job. We made this very clear in our report. Unfortunately, we found no impact (positive or negative) of MEGA on warehousing or manufacturing jobs.
We are fairly confident of our results, but there's no need to trust our findings. The 2005 study was peer reviewed by economic development economist Peter Fisher, a scholar who Bartik and Erickcek source approvingly in their own paper. Moreover, the findings of our 2005 study were repackaged and submitted to the highly respected academic journal "Economic Development Quarterly."
The journal staff has put the paper through a completely new peer review by scholars unknown to me and my Center colleagues, and they readily accepted it for publication early next year. Ironically, Bartik and Erickcek comprise two of the three primary editors of this journal. It is odd that these two authors might be prepared to wholly dismiss our research with a few paragraphs on the one hand, but their journal accepted its findings and agreed to publish them on the other.
As for the Mackinac Center's 2009 study, which comes to the same basic conclusion as our 2005 study, we used a time-tested technique that is a de facto fix to the endogeneity problem. We used a shift-share approach which isolates national and industry specific trends from those changes that are happening inside a specific county. For example, if a Michigan county had a high proportion of jobs in auto manufacturing, we wouldn't confuse job losses that affected that industry with something else (say onerous taxes) that specifically led to job losses in that county. We find it a bit puzzling to be criticized for the careful use of a time-tested approach to measure MEGA's impact on Michigan's counties.
Bartik and Erickcek have tried something fairly novel in the historical treatment of tax incentives and should be applauded for the innovation. They employed a widely used regional economic model produced by REMI Inc. to estimate what would have been the total effect in Michigan if all the MEGA credits earned were truly responsible for all the jobs created by businesses in Michigan that received grants. They then extrapolated the effect based upon a recent study by Don Bruce, Bill Fox and Mark Tuttle at the University of Tennessee.
This approach is interesting, and has a great deal of promise. There are some troubling matters, however, related to the study. First, Bruce, Fox and Tuttle's 2006 study estimated tax elasticities for states, including Michigan. These are aggregate elasticities, not firm specific responses to tax abatements. What they estimate is the response to an entire economy of a tax cut or increase, not how much new economic activity will result from an individual firm applying for and receiving a credit.
That's significant, since the firms receiving the credit are already doing business in Michigan or considering doing so and thus might expand or relocate here anyway. Looking past this difficulty, Bartik and Erickcek use their estimates and assign a "batting average" to Michigan of roughly 1/12, which if mixing metaphors is allowed, gives the program a record that is painfully similar to that of the Detroit Lions.
There is another problem — but this one is for MEGA administrators — and that is the fact that Bartik and Erickcek estimate that as few as 8.2 percent of the MEGA winners were actually induced to expand or relocate in Michigan as a result of the MEGA incentive. In other words, 91.8 percent of MEGA deals were not the deciding factor in a business executive's decision to expand in or move a facility to Michigan, as required by MEGA law. In fact, the law reads, "except for a qualified high-technology business, the expansion, retention, or location of the eligible business will not occur in this state without the tax credits offered under this act." So, are the vast majority of MEGA deals even in compliance with the law?
This new paper is a novel approach to examining incentive programs, but it has its own shortcomings. Worse, the authors in their criticism overlook the fact that the Center's 2005 paper directly corrected for their concerns and its 2009 paper came to the same basic conclusion: MEGA is an ineffective program.
In the end, policymakers and Michigan taxpayers might be expected to be puzzled by a half dozen technical studies which seemingly come to different answers about the impact of MEGA. The results range from findings that MEGA created one temporary job for every $123,000 in tax credits offered, to its most positive finding that only one in 12 jobs claimed by the program wouldn't have happened anyway. In reality, from the standpoint of assessing the efficacy of a program, these studies all say about the same thing -MEGA hasn't been smart public policy for Michigan. Further, if the performance of this program is a success, we all might be interested in asking: Just what does failure look like?
Michael J. Hicks, Ph.D., is associate professor of economics and director of the Center for Business and Economic Research at Ball State University and an adjunct scholar with the Mackinac Center for Public Policy, a research and educational institute headquartered in Midland, Mich. Permission to reprint in whole or in part is hereby granted, provided that the author and the Center are properly cited.