GOP prof uses sketchy study to claim M&A credit works

You may have seen this report by UW Professor Noah Williams, which has been tweeted out and referred to by WisGOPs in the last couple of days. The paper deals with the state's Manufacturing and Agriculture Tax Credit, and the media dutifully reported its findings, in no small part because Williams' findings went against pretty much any economic report we’ve seen over the last 4 years, which has Wisconsin consistently lagging its neighbors and having stagnant manufacturing growth.

So I dug into Williams’ paper, and started by looking at how he defines Wisconsin’s “border counties”, where much of the growth has taken place.

A map of the counties of Wisconsin and its neighboring states is shown in Figure 2.1. I now focus on differences across these borders. My sample includes 21 border counties in Wisconsin (Buffalo, Burnett, Crawford, Douglas, Florence, Forest, Grant Green, Iron,Kenosha, La Crosse, Lafayette, Marinette, Pepin, Pierce, Polk, Rock, St. Croix, Vernon, Vilas, and Walworth), each of which are matched with their paired county (or counties if the borders of multiple overlap) in Minnesota (Carlton, Chisago, Goodhue, Houston, Pine, St. Louis, Wabasha, Washington, and Winona), Iowa (Allamakee, Clayton, and Dubuque), Illinois (Boone, Jo Daviess, Lake, McHenry, Stephenson, and Winnebago), and Michigan (Dickinson, Gogebic, Iron, and Menominee). Thus the sample includes a total of 43 counties in 21 groups.

Right away, this should set off your radar as a problematic sample, and not only because Williams didn't include Trempealeau County as a "border county" (which might come as news to the Trempealeu Hotel on the Mississippi). The M&A credit is for the entire state, not just border counties, so wouldn't it be worthwhile to compare it state-to-state?

Williams' sample also includes the Wisconsin counties of Kenosha and Rock, which suffered the largest devastation in the state during the Bush Recessions of the 2000s. Rock County unemployment was over 13% in early 2010 and was consistently around 2% higher than 2002’s level in 2012, while Kenosha County unemployment peaked at 12.1% in 2010, and was also 2% higher in 2012 than 10 years prior.

Because Wisconsin border counties lost more jobs than other places before 2012, it made it easier for it to “snap back” and show improvement in the 4 years after that. And Williams uses that fact to “prove” Wisconsin grew faster, and advance his case. This is from Williams’ own paper.

Annual growth rate for manufacturing jobs in border counties
2002-2012
Wisconsin -2.39%
MN-IA-IL-MI -2.01%

2013-2016
Wisconsin 1.78%
MN-IA-IL-MI 0.90%

And then Williams makes up a “difference” stat that tries to make it look like the M&A credit is the reason for this stronger manufacturing recovery in Wisconsin, when it may be nothing more than snapback (notably, we do not see a comparison of the total job change between 2002-2016).

But in isolation, the job growth in those border counties has been impressive since 2012, and Williams says the bump in manufacturing lifts the numbers for the rest of the state, and assumes that the rest of the state also added manufacturing jobs due to the M&A credit.

The estimated 6.6% increase in manufacturing employment from the border counties implied a 4.3% increase in the interior, and thus a 4.6% increase statewide. This suggests that manufacturing employment would have fallen in the absence of the MAC, and I estimate that by September 2016 the MAC accounted for a cumulative increase of 20,819 manufacturing jobs statewide.

Basically, Williams assumes that instead of gaining 13,000+ manufacturing jobs over the last 4 years, we would have lost 7,000+ manufacturing jobs in that same time period. So apparently the manufacturing sector is booming in the state- which might come as news to all those Wisconsin blue-collars that voted for Trump to “turn the economy around.”

As for spillover effects, Williams says the manufacturing gains raised overall employment in the state as well.

Figure 4.3 plots the estimated cumulative statewide increase in manufacturing, non-manufacturing, and total employment due to the MAC. There we see that the employment gains in the manufacturing and non-manufacturing sectors were roughly equal. The estimated percentage gains in the non-manufacturing sector were substantially smaller, but non-manufacturing employment is roughly 4 times as large as manufacturing employment, resulting in a commensurate increase in total jobs. In total, I estimate that by September 2016 the MAC accounted for a gain of 42,161 jobs statewide, a 1.8% increase in private employment.

If that’s true, then Wisconsin’s underlying economy is even shittier than I thought it was. I say this because Wisconsin didn’t come close to outperforming its neighbors in job growth between 2012 and 2016, and was well behind the US rate of growth as the Obama Recovery hit full steam. In fact, the Walker jobs gap tripled in the first 4 years that the M&A credit started and expanded, from less than 38,500 at the end of 2012 to more than 115,500 by the end of 2016.



And of course, while Williams may say that jobs were created through this giveaway, he doesn’t mention the huge price tag that went with it. As the Wisconsin Budget Project noted earlier this year, the M&A tax handout is estimated to cost the state a total of $768 million through the end of this fiscal year, and will cost another $654 million in the next budget.



Even if you buy Williams’ argument that this credit has added 42,000 jobs in the last 4 years, is that worth over 3/4 of a billion dollars? I would think that investing that money in fixing roads (for example) and funding the schools would have gone much further, and reduced the future costs that are now resulting from that inaction. And it would have made Wisconsin a much more attractive place for talent to locate, instead of relying on this strategy of one-time giveaways to donors corporations that have failed to make our manufacturing wages competitive with Wisconsin’s neighboring states (note that Williams does not mention wages in his analysis).

Lastly, while this is a study from a UW professor, and the numbers seem to have some legitimacy at first glance, it is not UW-sanctioned research. If you look closely at Williams’ press release, it says the following.

The Center for Research on the Wisconsin Economy (CROWE) was created to support and disseminate economic policy research, with a particular focus on the Wisconsin economy and state-­‐level economic policy issues. (Final campus approval is pending.)

CROWE’s goal is to better understand the economic outcomes and the impacts of policies at the state level, and make economic research accessible to policymakers, businesses, and community groups in Wisconsin and around the nation.

If you try to look up CROWE in the UW directory, it’s not even mentioned as a future initiative. I find that intriguing. Also intriguing is that Williams’ CV lists numerous pro-GOP, supply-side articles such as “Marco Rubio’s Tax Plan Provides Relief for all Families”, and a 2015 article he wrote for Forbes magazine called “Under Scott Walker, Wisconsin Has Prospered – Keep That in Mind for 2016.”

So that whole meme that WisGOPs will try to use in the coming weeks and months to claim “See, the research shows M&A credit added 42,000 jobs! It's working!” Be skeptical. Be VERY skeptical.