Googling around it looks to me that this paper by Ed Glaeser and Bryce Ward, “Myths and Realities of American Political Geography,” got only cursory attention on the blogs, which is really too bad, because it's just terrifically illuminating. If you've been following this type of thing, you know interesting tidbits like: church attendance predicts voting Republican better than income these days; rich people in rich states tend to vote Democratic and rich people in poor states tend to vote Republican. But I did not know this:
Industrialization 85 years ago is an astonishingly good predictor of social and cultural attitudes today across states and a good predictor of support for the Democratic Party at both the state and county levels. As the share of the workforce in 1920 in manufacturing increases by one percentage point, the share of respondents today believing that AIDS is punishment declines by .28 percentage points, the share believing that military strength is the best way to peace declines by .16 percentage points, and the share supporting John Kerry at the state level increased by .42 percentage points.
Religious and political attitudes are better predicted by industrialization and immigration 100 years ago, than by the history of slavery and religion.
Glaeser and Ward hypothesize that the ethnic and religious heterogeneity of a largely immigrant workforce packed into densely-populated urban manufacturing centers created a strong incentive for the emergence of ideologies that minimized conflict by creating a climate of tolerance.
[I]f different religious or ethnic groups are prevented from using the power of the state to disenfranchise, enslave or kill each other, and if there exists a powerful group that benefits from eliminating conflict [i.e, the capitalists employing immigrant labor], then diversity can eventually lead to a watering down of core religious tenets or ethnic animosities.
That's right. Industrial capitalism civilized religion, moderated tribalism, and created a moral and political culture in which Democratic politics thrives. Places far from manufacturing centers, or places that industrialized later, are more likely to now be home to more socially conservative evangelical Christians and Republicans. Glaeser and Ward argue that the middle of the 20th Century, in which economic issues took precedence over social issues in determining party affiliation, was an anomaly and we have lately returned to form.
Glaeser, together with Jesse Shapiro and Giacomo Ponzetto, has come up with a theory for predicting when we'll see median-voter defying political extremism. Basically, if you can signal to a large constituency relatively radical intentions you can excite them and get an advantage in turnout. But you need to be sure that you're inspiring your base more than arousing the opposition, so your communication needs a way to be narrowcast to the target group. If such a group is large enough (but less than half the population), its views can largely determine the main divide in electoral politics. Evangelical churches seem to serve this function for Republicans. And Glaeser and Ward suggest labor unions may have once done the same for Democrats, which may have pushed issues of economic distribution to the forefront of electoral politics. But as union membership has receded, so too have the electoral rewards of narrowcasting relatively extreme economically left-wing views to them.
This theory then provides us with two hypotheses for the changing importance of economic and social issues in American politics and for the realignments throughout the 20th century. One candidate is the rise and fall of unionization in America. At the beginning of the century, unions were a small part of the population. Only in small areas of the population did they provide an opportunity for targeting a significant fraction of the population. In mid-century, they rose to over 30 percent of all workers and today they are back down to 12 percent (Troy 1965, http://www.laborresearch.org).
The rise and fall of unionization corresponds reasonably with the connection between income and Republicanism shown in Figure 10. The middle decades of the 20th century were the high point of unionism and they were also the high point of the correlation between income and Republicanism. During this time period, the Democratic Party had access to the labor unions and this created an incentive for Democrats to move to the left on economic issues to get support in this important base. The rise and decline of unions provides at least one possible reason why economic issues rose and then fell in importance.
So de-unionization may play a role in explaining why rich people, especially those living in the tolerant milieu of the historical seats of heavy industry, are increasingly Democrat-leaning these days.
This is truly fascinating. All I can say about it is that Paul Krugman says something a lot different, so this must be wrong.