Much of the public discourse and media analysis of the surprise outcome of the 2016 US presidential election has emphasized the role of manufacturing workers. This paper examines the importance of manufacturing jobs and job loss as determinants of voting patterns using county-level voting data from recent presidential elections. The share of employment in the manufacturing sector and long-run manufacturing job loss at the county level are not statistically significant in explaining the change in Republican vote shares from 2012 to 2016, when controlling for standard voting determinants. However, the change in the Republican vote share is positively correlated with manufacturing in predominantly white counties and negatively correlated with manufacturing in ethnically diverse counties, with these effects roughly offsetting each other. The paper further shows that this polarization between white and nonwhite manufacturing counties is more closely associated with polarizing candidates than a polarized electorate.
The data underlying this analysis are available here.