G. Bingham Powell

G.B. Powell's book "Elections as Instruments of Democracy" researches precisely what its title promises: elections as ways for citizens to have their preferences represented in a parliament. Because the procedure through which these preferences at election time get translated into representation is the voting system, Powell analyzes the two main voting 'visions' and their respective performance in actually doing what they promise, namely to represent people's wishes. This means also that he only goes into the way in which voting systems represent people in given countries empirically - he does not go into the "fairness" of certain aspects of the voting systems themselves, such as FPTP's tendency to actually not count around 70% of the vote in common elections.

These two viewpoints on voting systems are the majoritarian one, usually implemented as some form of first-past-the-post voting (or with runoff, like in France), and the proportional one, implemented as proportional representation or a (regionally) mixed system, like in Germany or Italy. Using extensive data from over 150 elections in more than 20 countries, Powell first reviews both the majoritarian and the proportional 'vision' according to their own standards. For the majoritarians, this is that the voters must clearly be able to identify which government they're going to get by voting, and that the will of the majority must be represented over that of the minority. For the proportionalists, this is that the voters must all be represented equitably in accordance with their popular support.

Using a system of (somewhat arbitrary) weighing of various criteria related to each vision's objectives, Powell shows that each is relatively good at doing what it wants to do. Still, the majoritarians come off more poorly than the proportionalists already, since in practice a given party rarely actually achieves a majority of all votes cast, and the distortions created by first-past-the-post voting actually enables the second-most popular party overall to gain majority representation, as happened in New Zealand in 1993: the National Party got 35% of the popular vote and an absolute majority in parliament.

But then Powell has to do the hardest task, and that is to meaningfully compare the voting systems in accordance with a common standard. He does this elegantly by measuring several criteria that are supposedly shared widely by supporters of both visions: effective representation and closeness of government to the median voter's preferences. The former is measured by looking at how the voters' preferences are actually weighed in the government policies, not by going into each policy everywhere individually, but by ranking the government parties or coalitions on a left-right scale. When weighed against various aspects of political rules that allow non-government parties a certain say as well (shared committees, veto powers in Senate, etc.), one can get a sort of 'weighted average' of the country's effective policy stance at a given point, and measure this against the self-identification of the voters.
The latter in turn is measured by looking at the median voters' preferences and then weighing this against the median legislator within the government (coalition).

Now some of the weights may seem somewhat arbitrary, but Powell's enormous data quantity and his neutral stance towards the actual content of policies (he avoids all pitfalls of having to measure the "leftistness" or "rightistness" of individual policies), as well as the way in which his data matches with a lot of prior political science work by Lijphart, Strom and others, lend his conclusions significant weight. In the end, Powell demonstrates that the proportional systems score systematically vastly better on scales of effective representation, closeness to median voter, and even considering that some of the common ways of measuring are themselves already put in majoritarian terms. One can have issues maybe with the left-right dimension's usefulness (Powell discusses this but claims the literature shows it has good predictive power), as well as the odd assumption he seems to make that people supporting a proportional vision tend to be more opposed to direct democracy and to be more "elitist", but the conclusions are clear as can be. The proportional voting system is the better one.… (més)