the proportional representation fallacy
a lot of people in the electoral reform community intuitively believe that p.r. (proportional representation) is desirable in and of itself. but this is fallacious. any electoral system is only “good” insofar as it optimizes for human welfare, i.e. “utility efficiency”.
warren smith, a princeton math phd who is widely regarded as one of the top experts on electoral systems, has written a synopsis of the empirical research on p.r., as well as its political viability in the united states. one of his key points, which is largely applicable to other countries like the u.k. and canada is that you probably won’t be able to achieve proportional representation without first escaping two dominant parties. and that requires a better single-winner voting method, like approval voting or even score voting.
but beyond political viability, even if we focus purely on whether p.r. is even desirable, we can only compare p.r. against the typical choose-one plurality voting, aka c.o.p.v. or “first past the post” voting method. we cannot compare p.r. to advanced single-winner non-proportional voting methods like approval voting, because we don’t have the data. (we’ll soon have some evidence, since st louis adopted approval voting and portland, or recently adopted proportional representation, but even that will be only city-level and hard to extend to international country comparisons.) so here, i want to extend smith’s primarily empirical analysis with a more theoretical analysis via a thought experiment.
imagine the simplest conceivable proportional representation scenario. the electorate is divided into 3 equally-sized ideologically homogeneous groups we’ll call “left”, “center”, and “right”, and they’re electing 3 representatives to their city council. (you could multiply this number arbitrarily, e.g. 99 representatives with 33 members in each faction, but for the sake of simplicity we’ll just use 3 representatives.)
proportional representation would imply that we elect a member of the left, center, and right.
but imagine instead we use a non-proportional method like score voting, which elects all 3 representatives from essentially the exact center of the electorate. you can see this depicted in fancy graphical computer simulations here.
now, which of these two legislatures is better, the proportional one or the centrist one? both have an approximately identical ideological average, and thus would fall in roughly identical positions on any particular issue. one might expect the proportional body to consider more diverse perspectives and thus land on more informed and thus optimal policy. on the other hand, one might expect the centrist body to be more cooperative and efficient on account of their ideological similarity. and of course they already have an incentive to consider the needs and perspectives of their diverse constituencies, given the use of a voting method which elects candidates who best approximate the overall consensus views.
the bottom line is, this is an empirical question. we could only know for sure which kind of electoral system would produce better results through years of large scale political use and empirical analysis of human welfare metrics from life expectancy to reported happiness metrics. if there’s anything we can logically prove here, it’s that there’s no logical proof of which is better. you just have to get data. and the most exhaustive analysis of the data shows no conclusive evidence that proportional representation outperforms the best single winner methods.
the “proportionalist” fallacy is to simply assert that “without proportional representation, the centrists will get 100% of the seats despite earning only 33% of the vote”. but i’ve just shown why that’s mathematically meaningless. at the end of the day, the legislative outcomes come down to the overall ideological average of the entire legislature. it’s not about largely arbitrary binary abstractions like “blue party” and “red party”. it’s about policy outcomes and how they affect human welfare, empirically.
my personal suspicion is that the centrist consensus-favoring methods will perform much better. this is because i believe the human-welfare-maximizing policy proscriptions largely fall in the political center. for instance, we know that the most utilitarian economic policy redistributes as much wealth as possible while also minimizing deadweight loss. this means we want pigovian taxes, land value taxes, and universal basic income. but the political right is generally against wealth redistribution, and the political left is generally in favor of redistributive mechanisms which create deadweight loss, such as minimum wage, rent control, income and wealth taxes, and all manner of obviously economically irrational policies. the objectively optimal policy regime looks roughly center-left “neoliberal”. and even if having people from the left and right might “average out” on where to set certain “dials”, it may fundamentally fail to land on optimal policy in a categorical way. i.e. we don’t debate on how big the ubi should be, but instead we just don’t have a ubi at all. or carbon taxes, or land value taxes. in other words, centrism isn’t just better because people get along better and more efficiently craft policy with less identity politics and fox news nonsense. it’s intrinsically better because it’s where most of the best policy comes from anyway.
