In the wake of last week’s election, we wondered how much impact jobs had on the election. Using BLS data for the states (current as of September), we ran correlations between state labor market data and state voting data. It’s worth noting that votes are still (seriously, still) being counted in California, so these aren’t necessarily official vote totals, but they’re currently the best we’ve got so far, courtesy of Cook Political Report’s tracker which you can find here. For labor market data, we used the unemployment rate, the change in the unemployment from the year prior, and the change in employment versus the year prior. To measure voting, we used the Democrats’ margin in each state in 2016, the change in that margin versus 2012, and the change in total votes cast (for all candidates) versus the year before.
As shown, the best predictor of any change in our three vote outcomes was growth in employment. While none of the 9 correlations we ran are very strong, by far the strongest predictor of how states voted was growth of employment; higher employment growth favored Dems, lower employment growth favored the GOP. Again, this wasn’t an extremely strong predictor, only accounting for about 12% of the total variation in voting margins. But it was the most important one of the variables we tried. Below we show a scatter plot of the two variables.
A weaker, but still present, correlation was between the unemployment rate and turnout as measured by the change in votes cast per state versus 2012. Higher unemployment rates were associated with higher voter turnout, and vice-versa. Below we chart a scatter plot of the two variables.