Saturday, April 25, 2015

Red State Blue State

A comment that ran away with me. Ed Kilgore has a good post which you should read. He notes that blue states have better outcomes than red states and begins to discuss how hard it is to derive useful conclusions from this pattern. Read his post. I think this excellent post is also important. The Rosenberg article and Wise post are also important. Your criticisms of Wise are valid (I'm sure Wise would agree). I'd add another -- which came first the goose or the golden egg ? I think it very likely that economic under development and social dysfunction cause conservatism. In particular, I think that dysfunction causes fear and disapproval of the poor which causes hatred of welfare. I also think that social dysfunction causes people to think they (and others) should get right with God and traditional morality.

I think conservative policy causes poor economic performance and social dysfunction. But I don't think the effect can be measured by looking at correlation, because I think there is a vicious cycle of bad outcomes causing conservatism which causes bad outcomes.

The possibly unsolveable problem is that the two valid criticisms imply opposite strategies. If one is worried about reverse causation, one must control for the state of the state in the past -- that is look only at short run changes. If one is convinced that short run changes have a lot to do with, say, the price of petroleum, one must look at long term averages.

I think the solution is to look at long run effects of policy shifts especially if it is hard to see how local conditions caused the shift. One example is different effects of a Federal shift based on the interaction of a new Federal law with old state choices (the ARRA (Obama stimulus) medicaid provisions sent different amounts of money to different states because of their existing programs -- this makes it possible to show that the ARRA stimulated the economy ( pdf and cites in it) .

It is now known that food stamps to parents of baby girls (and female fetuses) cause reduced obesity when the girls grow up

Here the point is that valid statistical analysis points us just where we want to go -- it is possible to evaluate the effects of policy (including long term effects). It is not possible to evaluate the effects of a pro-moonlight and magnolias ideology in general. Evidence about specific policies is more useful, because maybe we can reach agreement about specific policies while we will never agree about exactly how one gets right with God.