Rising inequality is associated with a shift toward conservatism for both the wealthy and the poor:
Does Inequality Make People More Conservative?, Monkey Cage: Yes, according to some new research (pdf) from Nathan Kelly and Peter Enns. They rely on a a yearly measure of "policy mood" from 1952-2006. This is an omnibus summary of the public's ideological leaning, liberal to conservative. (See the graph and corresponding Excel file at Jim Stimson's homepage.) They also draw on a specific measure of the public's support for welfare. The question is whether and how both measures respond to inequality.
Their first main finding: increases in inequality are associated with a conservative shift in mood and increasing opposition to welfare. (For more on why this would be true, see this paper (pdf) by Roland Benabou.)
Their second main finding: increases in inequality are associated with a conservative shift among both the wealthy and the poor.
One natural objection: perhaps some citizens, and especially poorer citizens, just do not realize that inequality has increased. But the third main finding contradicts this: over time, the poor are actually more likely to perceive increased inequality than do the wealthy.
Kelly and Enns offer some further speculation on why, in particular, the rich and poor respond in parallel to rising inequality:
Despite the fact that parallelism is not driven by lack of information about income inequality, we think it is possible that the way information about distributional outcomes is framed is important. This idea is rooted in Gilens's …argument is that during good economic times news stories focus on individualism (enhancing opposition to welfare) and during bad economic times stories emphasize people being down on their luck (enhancing support for welfare).
Given that rising inequality since the 1970s has been driven in large part by gains at the top of the income distribution, media frames over this period may have increasingly emphasized stories of individualism, thus generating a negative link between rising inequality and public opinion liberalism. The decline in inequality prior to the 1970s, by contrast, was driven primarily by increasing incomes at the bottom of the income distribution and may have generated stories emphasizing government's role in education and job creation. This could explain why declining inequality up to the 1970s pushed public opinion in a liberal direction.
See the paper for some further discussion and appropriate caveats.
That explanation doesn't ring very true to me, but I don't have anything better to offer. Any ideas?
(When you are puzzled by a regression result, the tendency is to question the statistical technique or the quality of the data. So, in that tradition, the autocorrelated error structure for the regressions in Table 2 where the results are disaggregated by income could be signaling a misspecified model. For example, citing the paper they cite to justify the correction for autocorrelation, "analysts should view autocorrelation as a potential sign of improper theoretical specifcation rather than just a narrow violation of a technical assumption." If the model is, in fact, misspecified then the results cannot be trusted. I'd also prefer to see the autocorrelation corrected by adding more lags to the error correction model rather than using the Prais-Winsten estimator as in the paper -- or at least try adding more lags as an alternative -- and see if it makes a difference for the results. That's the more usual correction to the autocorrelation problem in the applications of the error-correction models I'm familiar with.)
