I started reading An Economic Theory of Democracy with high expectations, but I quickly became frustrated. The book has some good points but there were numerous parts that disappointed me.
I liked the main point of the book, which is that in two party democracy parties converge on the views of the median voter (wiki entry). The broad point that uncertainty can make parties diverge was also a well made, although I thought that the specific mechanisms the book used to make the point were not very good. I suspect that one of the major contribution was viewing political parties as working to reach political office for the private rewards of political service, prestige, pay and seedier rewards, but I tend to discount this because I am so familiar with this argument.
I was disappointed by the lack of mathematical models used in the book, by the presence of omissions and by the use of numerous arguments which I thought were poor. I was probably wrong to expect formal models to be used in the book because those are generally reserved for academic papers, but I want to give examples of what I perceived as omissions and faulty reasoning.
First, although the low individual returns to voting are repeatedly discussed, the idea that intelligent voting is a public good was completely omitted. There are two distinct ways to think about this. If assumes that voters vote in their material self-interest, then voting intelligently for those material self-interests is a public good to those who share those material interests. If one assumes that voters vote altruistically, then intelligent voting is a public good to everyone. Either way, intelligent voting is very likely to be under-provided. The book continually skirts this issue; it mentions the fact that any given vote has a very small probability of affecting the outcome and therefore has an extremely low individual return, and it mentions that government is largely in the business of providing collective goods, but it does not connect these ideas.
Second, I was disappointed that Downs’ work failed to apply the logic he had developed earlier about party convergence to multiparty systems. Downs argues that multiparty systems lead to less centrist government than two party systems, but uses faulty reasoning to reach this conclusion; his logic works like this:
Consider 5 parties on a one dimensional spacial political spectrum (see figure 1), and assume that each party receives an even proportion of the vote.
Figure 1. The distribution of political parties with equal voting power.
No one party or even two parties can rule by themselves; three parties must form a coalition in order to form a government. Downs assumes that party C can come together with the two parties on either extreme (A/B and D/E) to form a government which is centered on either middle extreme party (B and D), and therefore produces more extreme political results than two party democracy.
However, this logic ignores individual party incentives. Assuming that parties A/B and D/E automatically form coalitions (for simplicity), according to Downs’ own logic, the coalitions at both extremes have incentives to offer party C the chance to be in coalition with a policy point which is closer to C’s ideal point than the one center point in the offer given by the coalition on the other side, because the parties want to be in office. These incentives ensure that the equilibrium policy point converges on the median party’s (party C in this case) ideal point.
I think Downs’ error here lies in that he thinks of people voting to “select a government” and not to influence which government is selected. He does not seem to consider how parties in a multiparty system can act as good agents for voter’s political desires.
My overall opinion of An Economic Theory of Democracy is much like that of The Calculus of Consent. Both books include numerous dubious arguments, but both manage to make some very important arguments. I like both books in retrospect, even if I didn’t appreciate them very much when I was reading them.