Peter Kellner has set out the case for the Alternative Vote (AV) as the preferred objective of electoral reform. I don't find his case at all convincing. Here's why.
Peter is of course absolutely right to point out that there is no perfect electoral system. At least, there isn't if you have more than one criterion for what makes a good electoral system. As soon as you admit more than one criterion, there is every chance the two could conflict and so you have to face a trade-off in which one electoral system gives you more of one thing you want and another gives you more of whatever else it is you want.
So at the outset, I want to eschew any argument for PR that treats proportionality as the sole criterion of importance and then tells us that (obviously) PR meets this criterion better than AV or what we have now.
Having firmly eschewed the 'PR absolutist' position, let's set out the non-absolutist case for PR as an alternative to AV.
(1) Giving some (non-trivial) weight to proportionality as an expression of fairness.
First, I am struck that in listing his criteria for a good electoral system, Peter seems to give no weight at all to proportionality as such. His fourth criterion - ensuring some representation for minorities - implies some limits on how disproportionality can express itself, but does not really amount to an acceptance of the desirability of proportionality as such. Now, while there can indeed be reasonable disagreement about how to weigh various criteria, Peter's apparent view that proportionality as such should have no weight strikes me as unreasonable.
Proportionality has a clear claim to being an important, weighty criterion because it is linked to the idea of individuals' votes counting equally in determining the outcome of an election. Take the infamous example of the 1983 general election in which Labour got roughly 10 times the number of seats in the House of Commons compared to the Liberal/SDP Alliance even though they had almost equal shares of the popular vote. Any Alliance voter had a quite reasonable grievance in this situation - their vote had so much less influence on the final composition of the Commons than that of a Labour voter.
Proportionality matters because it embodies the demand for a certain kind of fairness in the way votes get translated into representation in the legislature.
Once we allow proportionality to be a weighty consideration, however, the case for AV is hugely weakened. In some recent general elections (1997, 2005), AV would have delivered even more disproportional results than the current system. Even if we accept all of Peter's other criteria for judging electoral systems, if we include proportionality as a desirable feature too, with some weight, it is hard to see how one can justify AV over, say, 'AV plus', which has a proportional element.
(2) PR has better consequences than majoritarian electoral systems.
If the fairness consideration which animates PR proponents carries little weight for Peter, what does matter in his view are the consequences that an electoral system can be expected to have. His implicit strategy is to point to various bad consequences that PR allegedly has compared to AV.
This is fair enough. On my 'non-absolutist' position, I accept that the specific fairness which comes with proportionality can in principle be outweighed if it results in sufficiently bad consequences (unstable government, disproportionate influence to small parties, etc.)
Supporters of PR can and do respond to this line of argument in a number of ways. On the one hand, one can argue that to some extent worse consequences are worth it for the sake of greater fairness - though this won';t get any traction with Peter as he doesn't seem to want to give proportionality as such any weight. Or one can argue that the bad consequences are less likely than is being alleged. Or one might argue that they can be substantially avoided if the principle of strict proportionality is qualified in some way(s), e.g., by setting thresholds of the popular vote which parties have to get over to get any representation.
A fourth line of response, however, is to argue that, relative to majoritarian electoral systems like our present system and AV, PR has good consequences.
As someone on the left, one consequence I care a lot about is social justice, as expressed, for example, in the distribution of income. Back in 2006, the political scientists Torben Iversen and David Soskice published a paper in the American Political Science Review (arguably the leading political science academic journal in the world) which examined, theoretically and statistically, how different types of electoral system impact on redistribution and income inequality. David Soskice also wrote an article exploring the implications of the paper for Prospect.
The main result of Iversen and Soskice's analysis is striking. They find that, controlling for other relevant factors, PR electoral systems generate more redistribution and inequality reduction than majoritarian systems. (Australia, with its AV system, is included in the analysis as a majoritarian system.) The effect is substantial and statistically significant. (The chance that the statistical association they find is pure fluke is less than 5%.)
Why is this? Well, a large part of the answer is that PR systems simply generate more years of centre-left government on average than majoritarian systems. To be sure, left-wing parties rarely govern alone in PR systems; they hold office in coalitions with other parties. But periods of government by coalitions with gravity to the centre-left are far more common in PR systems than are periods of single party rule by the left in majoritarian systems.
Iversen and Soskice explain this finding in terms of a theoretical model which looks at how different electoral systems affect the possibilities of cross-class/cross-party governing coalitions. The basic intuition is that under PR systems, middle-class centre parties are drawn to coalitions with left working-class parties, making for centre-left political domination. By contrast, under majoritarian systems, the same middle class group tends to prefer straight right-wing parties to straight left-wing parties - this tending to be the restricted choice on offer. This produces a politics in which the centre-right is dominant.
So, let's take stock.
There is an important consideration of fairness which ought to be given some weight in evaluating electoral systems. I would say, going a little further than I did above, that this consideration establishes a reasonable presumption in favour of PR. This presumption can be defeated if proportionality generates bad consequences (and many PR supporters accept this when they argue for qualifications to strict proportionality to reduce the risk of certain bad consequences). However, we have good evidence to show that along one dimension of crucial concern to the left - economic inequality - PR systems tend to produce better outcomes than majoritarian systems (of which AV is an example). So rather than defeating the proportionality presumption, a consequentialist analysis seems to reinforce it.
On balance, therefore, I am not convinced by Peter's argument that we should endorse AV. On the contrary, we should endorse PR.