by Scott Winship
Apologies for the lack of posts, but I’m realizing that when I have a roundtable discussion to coordinate, posting has to take a backseat. I’ll do better though. Please. Don’t go.
What’s the matter with Kansas? Possibly nothing, according to a National Bureau of Economic Research working paper by Duke economist Jacob Vigdor [pdf] (go Blue Devils!). This conclusion isn’t that novel, but rather than pointing to cultural issues as the reason that working class voters vote Republican, Vigdor argues that voters’ well-being depends not only on their standard of living but on their living standards relative to others in their reference groups.
I’ll be honest with you – you don’t want to read this paper unless you love the Greek alphabet. It has a deceptively catchy title – “Fifty Million Voters Can’t Be Wrong” – but a whole lotta math. But here’s the gist. Research indicates that individual evaluations of well-being depend on how others are doing. Vigdor considers two types of “others” you and I might compare ourselves to – people in a similar economic situation as us, and people who are geographically near us. He proposes a mathematical model relating political preferences to income and income relative to a reference group. Theory predicts a number of ways that relative income could affect political preferences, and real-world trends and patterns can be cited that are consistent with these predictions. If the predictions are borne out and accurately reflect reality, then the finding that relative income affects political preferences can explain why the working and middle classes tend to vote Republican and why they have not become more Democratic than they have as inequality has increased in recent decades.
Whew! Got that? OK, let me break it down more slowly.
Vigdor’s theory of relative income predicts that the more income inequality there is in a geographic area, the more support there will be for redistribution among the poor and among the rich who are altruistic. The idea is that when a poor person looks around and sees rich people, she is more inclined to support redistribution than if she looks around and sees only poor people. When an altruistic rich person encounters lots of poor people, she will be more likely to support redistribution than if she only comes across other rich people. Among the working and middle classes, support depends on how many rich people there are, how many poor people, and how altruistic voters are. With more rich people, for instance, the working and middle classes will support redistribution because they will benefit. When Vigdor estimates the key “parameters” in his mathematical model using statistics, he finds that the estimates are consistent with these predictions.
An implication of Vigdor’s findings is that one reason support for the Democratic Party among the working and middle classes failed to increase more as inequality grew is that segregation between the rich and everyone else has been on the rise for several decades. That means that today, the poor and the working and middle classes are less likely to see rich people when they look around than they were in, say, the 1960s.
Additional support for Vigdor’s theory comes in his finding that the relationship between income inequality and support for Al Gore in 2000 is stronger in urban counties than in rural ones. That is, in cities, people can look around and see whether there are many or few rich people, while in rural areas with low population density, it’s more difficult to do so.
Vigdor’s theory also predicts, and his data supports the predicton, that poor, working-class, and middle-class voters should have been less likely to vote for Gore the more poor people there were in their county. Rich individuals should have been – and were – less likely to vote for Gore the more high income people there were.
Finally, and perhaps most importantly for the Kansas question, Vigdor’s theory predicts that if voters compare themselves to people who are in a similar economic situation, then working- and middle-class voters should be less likely than poor or rich voters to support the Democrats. That’s because of the particular way that income is distributed in the U.S.
As income increases from $0 to a working-class income, the number of people at each income level gets larger and larger. That means that more often than not, when people in this income range (“the poor”) compare themselves to other poor people, they will find that there are more poor people doing better than them than worse. They will thus tend to support redistribution.
On the other hand, as income increases from a working-class income to an upper-middle-class income, the number of people at each income level gets smaller and smaller. When people in this income range (“the working and middle classes”) compare themselves to other similarly-situated people, they will find (more often than not) that there are more working- and middle-class people doing worse than them than better. Consequently, they will tend to oppose redistribution.
Finally, as income increases from an upper-middle-class income to an upper-class income, the number of people at each income level continues to get smaller and smaller, but the decline is not very steep. When the rich compare themselves to their peers, they will tend to find that there are nearly as many people doing worse than them as there are doing better. The rich will tend to be indifferent toward redistribution.
These predictions about support for redistribution are also supported by Vigdor’s data. Vigdor notes that since the New Deal, the income distribution in the U.S. has changed so that more people fall in the “working and middle classes” range where comparing oneself to one’s peers will produce opposition to redistribution. He also speculates that if, in the post-Civil-Rights-era South, poor whites began to compare themselves not to other poor whites but to even poorer blacks (who would not have been considered a proper reference group during Jim Crow), then southern whites would have become less redistributionist and would have moved into the Republican column, which is of course what happened.
It’s important to note that – as with all models – the estimates produced are accurate only to the extent that the model accurately depicts reality. The point of Vigdor’s analysis is not that his estimates are the final word or that relative income is the be-all, end-all, but that under fairly basic assumptions about how different factors affect political preferences, a relatively simple model applied to real-world data confirms the predictions made before he began playing around with the data. The evidence implies that Democratic underperformance among working- and middle-class voters is due in part to the tendency of people to compare themselves to others and to a number of social patterns that made this tendency prevent people from becoming more Democratic.
Incidentally, I remain a Blue Devils basketball fan even though a) their engineering school denied my undergraduate admissions application and b) Coach K is allegedly a big conservative. Don’t forget to check out the roundtable discussion on redistricting, electoral competition, and targeting of districts.
by Scott Winship