Max Black is Back

So the law & econ workshop paper last week was "Adding Substance to Section 5 of the Voting Rights Act," by David Epstein. When a state redistricts, it has to get clearance from the courts. In the past, the courts have forced states to create districts with high percentages of minority voters, so that the minorities have at least some representation in the legislature (federal or state). These days many people feel that this is counterproductive, since concentrating like-minded voters in one district is a way to diminish their actual political clout.

Epstein argues that there are 2 kinds of representation: descriptive and substantive. Descriptive representation happens when minorities elect their preferred candidate. This means that their candidate wins both in the primary and the general election. If minorities lose in the primary but then vote for the winner in the general election, that's not descriptive representation. Substantive representation just means winning actual votes within the legislature. There is some tension between these two kinds of representation, since minorities might get more descriptive representation if they're concentrated in a few districts, but more substantive representation if they're spread out.

My main objection to the paper is that it measures substantive representation by tracking the votes of the representatives of heavily minority districts. So for instance, imagine that a district is 65% black and the representative won both the primary and the general election with black support. That legislator is taken to represent black interests, and substantive representation is measured by how many times the legislature votes along with that person. Similarly, to tell whether another legislator is "black-friendly," says Epstein, just tracks how often that person votes with the black caucus.

Aside from the fact that a vote count is a crude measure (because some votes matter more than others), this strikes me as a poor proxy for substantive representation. It so happens that black-supported legislators are usually Democrats. The results then show that white Democrats are fairly black-friendly, while Republicans (white or black) are not. The problem is that this measure over-states the friendliness of Democratic legislators. Put simply, we don't know whether the white Democrats are voting with the black interests or whether the black-supported representatives are voting with the Democrats. If black-supported legislators were somehow forced to vote Democratic in all votes (say, with financial or institutional pressure), the Democrats wouls appear to be very black-friendly without in fact changing their votes at all. In reality, of course, it's not so extreme. Black-supported legislators probably work with white Democrats so that each group scratches the other's back. The point is that we don't know the extent of this compromise; the numbers may reflect 100% Democratic support for black positions while the reality is that black-supported legislators often compromise and vote for white Democratic interests.

I asked Epstein about this at the talk, and his response was essentially that compromise is part of the political process. No doubt this is true, and no doubt it helps blacks to be in a position to take advantage of the give and take of politics. Still, Epstein's measure of substantive representation is likely to be overstated. He thinks courts can allow states to trade away descriptive representation in return for increased substantive representation, but his measure of substantive representation is flawed. This calls the whole tradeoff into question; courts should be aware that if they use Epstein's measure, they could trade away descriptive representation and get less than they expect in return.