Bailout Vote Analysis: Theory

by David Epstein and Sharyn O'Halloran

OK, now we've seen that the defection from the wings in the bailout vote was true in fact. But is it true in theory?

Let's start by defining the issue space. We'd argue that there are two dimensions: the amount of money made available for the bailout, and the amount of help to homeowners. (Good thing there are two dimensions, since you need more than one to generate strange voting patterns like we saw on Monday.)

Now let's assign preferences to three groups of House members: Conservative Republicans (CR's), Moderates (M's) and Liberal Democrats (LS's).

In this instance, the CR's and LD's were both against the idea of a bailout, so they agreed on dimension 1. The Democrats didn't like it because they thought it was socialism for the rich and capitalism for the poor; the Republicans didn't like it because they just thought it was socialism.

Moderates from both parties supported the bailout as necessary to save the financial system. And Democrats were more favorably disposed towards helping homeowners than were Republicans.

Overall, then, we posit that preferences over the two dimensional policy space looked like this:

 

Next, we place the status quo and alternative (bill) in the space. The status quo had no bailout and no homeowner help, while the bill had (lots of) money for the bailout and some degree of homeowner protection. So now the picture looks like this:

Finally, we see who votes for the bill and who votes for the status quo. Assuming that members vote for the alternative closest to their ideal point, we draw a line joining the status quo and the bill, then add its perpendicular bisector, or cutting line (L), like this:

And voila! We have a defection from the wings: CR's and LD's voting against the bill, and moderates voting for it, with the mass in the wings outweighing the mass in the middle so the bill goes down.

One lesson we can learn just from looking at this diagram is that the CR's are going to be hard to get on board, because their ideal point is at (or near) the status quo. Only a very watered down bailout package with little help for homeowners could get their votes.

What else can we do with this picture? Well, we can try to predict how different alternative bills would fare. If the administration wants to keep the size of the bailout fixed, it can only move the proposal along the vertical dimension: either more or less assistance to homeowners.

The next diagram shows the result of changing the proposed bill to one with more help for homeowners (B') or one with less help (B''):

The bill B' might indeed increase overall support (as indicated by the new cutting line L'), but at a cost: the new supporting coalition is more clearly partisan than before, as some LD's are added and even more CR's pull away.

On the other hand, moving to B'' (and the corresponding L'') keeps the vote bipartisan, but with a larger coalition rejecting the bailout altogether. Not a big help.

The one extra possibility is to stay with the current bill but move everyone's preferences to the right, as shown here:

This looks great: bipartisan support and the bill passes. But how to move preferences?

Well, two forces already at work may do the trick. First, public pressure for the government to do something will raise legislators' tolerance for the bailout, especially if the damage done to credit markets becomes apparent to the average voter.

Second, an injection of "face-saving measures" might insulate legislators from complaints that they are not protecting their constituents, allowing them to take a more pro-bailout position than they have to date.

(These are tricky, though, since they require conservative members telling their constituents that the revised bill is more conservative than its predecessor, and liberal representatives telling their constituents the opposite. We have to hope that voters in the two groups never actually talk to each other.)

In fact, it seems that this latter strategy is the one the Senate is pursuing, voting today on a bill that includes include tax breaks for businesses and alternative energy and higher government insurance for bank deposits. We await the outcome with more than theoretical interest.

Comments

[Paul Lehto]

and, I heard on CNN, funding for mental illness the same as physical illness in some fashion was added to the bill. Late night comics should have a field day with that one. But it may be tough to get Americans to laugh too much.

This is an interesting analysis, but "Defining" the issue space with a mere "argument", as you probably know, is building the house on stilts. But, I think it still yields some interesting stuff.

I think for folks on the wings, it wasn't just the distance from the ideal, but also the existence of poison pills. one of these is Sec. 119, which only allows constitutional claims as lawsuits, but then even if you win on a constitutional claim, it's immediately subject to "automatic stay" by will of Congress, if the Secretary appeals and asks for a continuance of the stay. Given the pre-existing preference for the status quo in stay analysis and litigation generally, plus the force of congressional intent, the risk's very high that even proven constitutional violations would wait a long time to be resolved, since nothing in the bill expedites appeals (it does expedite trial court matters).

Therefore, somebody really seriously believing in the Constitutional rights of all, and these exist on both the Left and Right, could find this a violation of their oath to uphold the Constitution and vote against for that poison pill reason alone. This would be difficult to explain to the public, but may be a major troubling factor in the conscience of a legislator.

Question: All models simplify, but just like elections are concatenations of millions of forces with disparate reasons, so are congressional votes with an engaged public. Can you say for sure that your datamaps are meaningful, even though we know for sure that they oversimplify by defining the issue space thus?