In addition to the estimates based on all counties in the continental U.S., we also estimate our model for the manufacturing, construction and transportation and warehousing industries with smaller samples. These include Michigan and its border states, and separately, Indiana and its border states. We now turn to the discussion of those results.
This sample includes counties from Michigan, Indiana, Ohio and Wisconsin. Of these states, only Ohio did not have right-to-work legislation passed by 2018, the year from which our employment data is collected. Manufacturing accounts for roughly 23% of total private employment in this sample. The direct effect of right-to-work laws on a typical right-to-work county in this sample is to increase manufacturing employment as a percentage of total private employment by roughly 26%, although this finding is not statistically significant.
A brief digression on statistical and economic significance is in order. Some researchers choose to rely only on findings that meet the conventional threshold of statistical significance. In doing so, they discount all results that are not statistically different from zero. Others, including, for example, economist Deirdre McCloskey, argue that economic significance of point estimates matter, too. These should be considered, especially if the effect is large, as is the case with Michigan’s results.
We include and emphasize the magnitude of right-to-work’s impact on manufacturing in Michigan border counties here. The lack of statistical significance in our state-specific example may simply be a function of the limitations in our dataset due to the relatively recent adoption of right-to-work in Great Lake States. The observations in our dataset have only had about six years to be influenced by the adoption of right to work. Some scholars argue it could take 10 years of data to properly measure the full impact of right-to-work laws on an economy, so inconclusive results may be reasonably expected.[40]
One result that was statistically significant and large was the spillover impact of a right-to-work law on counties in non-right-to-work states, in this sample, Ohio. Our model shows a 30% decline in manufacturing employment share in Ohio counties. While right-to-work laws were not associated with an overall increase in manufacturing employment share in this region of study, these results suggest there was a realignment of manufacturing employment opportunities toward right-to-work counties in the region.
Similar results were observed in both the construction and the transportation and warehousing industries. That is, the direct effect of right-to-work is to increase employment share in right-to-work counties and by a large amount: 30.0% in construction and 42.5% in transportation and warehousing. The indirect effect was also large and predicted: Employment share in Ohio counties was 32.8% lower in construction and 68.6% lower in transportation and warehousing. In the construction industry, the effects across the region cancel each other out such that there is only a small net effect. In transportation and warehousing, the net effect is relatively large in magnitude and negative, suggesting that total employment share in the four-state region is lower by potentially 26%.