What’s so great about Proportional Representation, anyway?
Loosely speaking, proportional representation (PR) means having the preferences of the electorate be proportionately reflected in an elected body. These preferences can be about partisanship, ideology, geography, race, religion, or anything else a sizable chunk of the electorate deems important. So if the electorate cares primarily about partisan affiliation and 44% of voters prefer the Pink party, 37% prefer the Orange party, and 19% prefer the Yellow party, a proportional outcome for a 5-winning election would be to elect two Pink candidates, two Orange candidates, and one Yellow candidate.
(More rigorous definitions of proportional representation are also possible. For a fully partisan system, deviations from perfect proportionality can be measured using the Gallagher index. When voters aren’t perfect partisans and you’re evaluating voting methods that are candidate-centered (e.g. Single Transferable Vote, a.k.a. Proportional Ranked Choice Voting) instead of a Party List variant, defining proportionality is more complicated and controversial but still feasible. However, the loose informal definition will suffice for our purposes here.)
Intuitively, PR seems extremely fair. It makes sense that a party that wins 35% of the vote should win 35% of seats. But this tends to be a superficial argument since it is not immediately clear why this fairness is a good thing. So let’s take a closer look at the consequences of proportional representation. Why is it good? Can it have negative repercussions? And is PR really necessary, or can something else do just as good of a job?
The obvious point is worth emphasizing: PR means that minorities get representation. Smaller political parties win seats under PR when they wouldn’t under single-member districts (Australia offers a particularly clear example of this), and racial/ethnic minorities (along with women) have an easier time winning seats.
There are a couple of caveats. First, minorities may fail to get represented if minority voters don’t vote for the candidates in their minority; PR means that voter preferences get reflected proportionately, not characteristics such as race, religion, or even partisan affiliation. Second, a minority must be sufficiently large. As I wrote previously,
Whether a minority is large enough is determined primarily by district magnitude, which is the number of representatives elected from each district. We can talk about district magnitude without actually having districts — we just treat the entire jurisdiction (be it a city or an entire country) as a single district. A district magnitude of 1 means having single-winner elections — and a voting bloc needs to comprise over 50% of voters to ensure representation. With a district magnitude of 2, a voting bloc typically only needs the support of over 1/3 of voters to get represented, and with a district magnitude of 5 (as would be common if the Fair Representation Act is passed) a mere 1/6 of voters are needed. With a Party List voting method, it is natural to elect all the representatives in a single election without districts, so the district magnitude can be well over 100 and even small minorities can ensure that they are represented.
(Another limitation: Many countries that use Party List methods have a threshold of the popular vote share that a party must receive (say, 3.25%) in order to win any seats.)
Why might we want more minority representation? Some arguments:
- Some policies may have very different effects on different groups. If a group is unrepresented, an elected body may fail to properly consider the harms or benefits of a policy on a group that is disproportionately affected by it.
- Even if an elected body properly considers how a policy impacts a minority, that doesn’t mean an elected official actually cares about those effects. Minority representation entails the representation of values as well as the representation of beliefs.
- People like feeling that they’re represented, and they like having an elected representative who shares their views to whom they can voice their concerns. (I find this to be a relatively weak argument since it doesn't appear to involve policy outcomes or anything like that, but maybe representation reduces societal tensions or something.)
The downside of more minority representation is that minorities you find abominable — far right, far left, or other fringe ideologies—are also more likely to win seats.
Ending the Two-Party System
Proportional representation means that more parties win seats, so you don’t have a two-party system.
A two-party system makes politics zero-sum; whatever damage you cause to the opposing party is a benefit to yours. Scorched-earth smear campaigns are savvy politics. If you’re in office and have a way to hurt the country such that your party will take 45% of the blame and the other party will take 55% of the blame, hurting the country like this will boost your electoral prospects.
But with more parties, dragging an opposing party into the mud and causing them to lose a seat doesn’t mean that you’ll gain it. Having two parties at one another’s throats opens the door for both of them to lose ground to a third party, so scorched-earth politics make a lot less sense strategically. To be clear, having more viable parties doesn’t come close to eliminating the incentive for negative campaigning or any other kind of dirty politics — but it does at least weaken these incentives.
Another reason why having more parties reduces polarization is that the more parties there are, the less likely a single party is to win an outright majority. Without a majority, parties need to negotiate and form coalitions in order to govern. Alienating other parties to the point that they’re unwilling to form coalitions with you isn’t good politics; merely the possibility of forming a coalition with another party reduces the incentive to attack them.
But it’s also possible to have too many parties. Politics is compromise. With two parties (and no separation of powers) all the compromising happens at the ballot box: voters select the one of the two parties that looks better than the other, even if they both seem terrible. Then the party with the majority does whatever it likes; no need to include the other party in anything (unless there’s a supermajority requirement). But with a lot of parties, voters don’t need to compromise very much — if they have over a dozen options, chances are there will be a party that’s a pretty good match for their values. But the parties themselves will have to compromise a lot. All the dealmaking and jockeying for advantage means that it can take months to form a government, and this dealmaking can seem more important than the actual election. The difficulties of maintaining stable coalitions that include over a half dozen parties are a major reason why most countries with a Party List system limit the number of parties in parliament with thresholds that parties need to clear in order to win any seats.
Ultimately, I think both the American two-party system and the Dutch many-party system are problematic, so an intermediate option is best: At least three parties, but no more than about six.
(What about a two-party system with either a supermajority requirement or a lot of checks and balances that require both parties to work together in order to govern effectively? Here again you run into the problem of zero-sum politics — if being part of a functional government makes the other party look good and causes them to win more seats, that means that this functional government is making your party win fewer seats, so you’re better off with there being a non-functional government. If exactly one party is dragging everything down and refusing to compromise, voters do have some power to punish this intransigence at the ballot box — but determining which party is being unreasonable is quite difficult when at least one of the parties is constantly trying to muddy the waters, so using supermajority requirements to sabotage good governance is a pretty reasonable move. Especially if the other party holds the White House and voters are biased towards giving the President’s party most of the blame/credit for everything the government does.)
Gerrymandering and Geography
One problem with single-winner districts is the prospect of gerrymandering, i.e. redrawing district boundaries for political advantage. Gerrymandering involves two complementary tactics: packing and cracking. You want to pack opposing voters into a few districts that they win by overwhelming margins so that as many of their votes as possible get wasted. And you want to crack the opposition in the remaining districts by giving them as large of a minority of voters as possible in each of those districts while ensuring that your own party still has the most supporters there. In a perfect world, you’d win most districts 51%-49% while losing a handful of districts 0%-100%. (In reality, of course, you’ll be unable to pack the opposition with 100% efficiency, and you’ll need to have some room for error in case your opponents become more popular. More realistically, you’ll try to win cracked districts by something like 15 percentage points and you’ll lose packed districts by maybe 50 percentage points.)
Proportional representation — whether achieved through an at-large election or multi-winner districts — obviates the problem of gerrymandering. With at-large elections, the mechanism is obvious: without districts, there’s nothing to gerrymander. But with multi-winner districts, it’s a bit subtler. Imagine there are fifteen representatives being elected and you’re an omniscient gerrymanderer who can arrange voters however you like, and your party has about 34% support and the opposing party has 66% support.
- With 15 single-winner districts, you can fill 5 districts entirely with opposing voters and then win the remaining 10 by a razor-thin margin, thereby winning 5 more seats than the opposition.
- With three 5-winner districts, this is still the best you can do: Fill one district entirely with opposing voters, and have a narrow majority in the other two districts. You win 6 seats (three from each district where you have a majority), and the other party wins 9 seats (five from the packed district and two from each of the other districts). This is a slight deviation from perfect proportionality (which would have you only winning 5 seats), but it’s nothing next to what can happen with single-winner districts. (You could get slightly more disproportionality in a scenario with three 5-winner districts, you winning two districts with districts with 50.01% of the vote (for three seats each) and getting 30.01% in the district you lose (for two seats), winning a majority of seats despite having a bit under 44% of the vote. But this is the most extreme scenario possible, doesn’t allow any room for error, and is still nowhere near as bad as what you get with single-winner districts.)
To be fair, proportional representation isn’t the only way of dealing with gerrymandering; independent redistricting committees can also be effective. But independent redistricting committees can’t help against disproportionate outcomes that naturally result from geography.
First, there’s the rural/urban divide. Suppose a state’s major cities are 80% Democratic and the rest of the state is 60% Republican. The most natural way to draw a map is to give these cities their own districts and have other districts that cover the more rural part of the state. But this functions exactly like a Republican gerrymander! By packing urban Democrats and cracking rural Democrats, any map that isn’t designed around boosting the Democrats will yield a disproportionate outcome that favors the Republicans.
Alternatively, a minority party can be spread out evenly throughout a state. The abstract of a study on disproportionate outcomes in Massachusetts:
Republican candidates often receive between 30 and 40 percent of the two-way vote share in statewide elections in Massachusetts. For the last three Census cycles, MA has held 9–10 seats in the House of Representatives, which means that a district can be won with as little as 6 percent of the statewide vote. Putting these two facts together, one may be surprised to learn that a Massachusetts Republican has not won a seat in the U.S. House of Representatives since 1994. We argue that the underperformance of Republicans in Massachusetts is not attributable to gerrymandering, nor to the failure of Republicans to field House candidates, but is a structural mathematical feature of the distribution of votes. For several of the elections studied here, there are more ways of building a valid districting plan than there are particles in the galaxy, and every one of them will produce a 9–0 Democratic delegation.
It is impossible to address this disproportionality with single-winner districts.
The problems stemming from gerrymandering are obvious since it is basically a method of rigging elections; it undermines democratic legitimacy and can serve as a useful tool for a constitutional coup. More generally, having geography determine election outcomes is at odds with majority rule, i.e. the notion that a party or coalition with the support of a majority of voters should be able to govern.
Even if you happen to get proportional results with single-winner districts, not all of those districts will be equally competitive. There will be swing districts and safe districts; political parties will be incentivized to cater to the needs of voters in swing districts, but not those in safe districts. Voters in safe districts wield no real influence since election outcomes in these districts are, by definition, a foregone conclusion.
With the extremely high district magnitude that comes with at-large Party List elections (say, over 100), there is always uncertainty over the exact number of seats that will be won by each party, and this uncertainty entails competition and influence. With multi-winner districts that have a much lower district magnitude, there is still much more competition than you get with single-winner districts, but you can also have something akin to safe districts.
Suppose, for simplicity, that there is a two-party system and a district magnitude of five. As in the case of single-winner districts, a district is competitive if it’s split roughly 50–50 between the two parties; here, “competitive” means that either party could win three seats, even though both parties are near-certain to win at least two. But unlike the case of single-winner districts, a five-winner district is also competitive if it’s split roughly 70–30 or 90–10, where the competition is over whether the majority party wins a fourth seat (for the 70–30 split) or all the seats (for the 90–10 split). A district can still be non-competitive (in terms of partisan control) if it’s split 60–40 or 80–20, however. The more seats that are up for election, the greater the number of competitive splits, until they blur together and all elections become competitive.
But that’s with a two-party system with no significant differences between candidates within a party. With more parties or more kinds of candidates, every district can be competitive, even with a district magnitude of five.
Suppose a district is 60% Democratic and 40% Republican. With five winners, this is as non-competitive as a district can get; it’s expected to elect three Democrats and two Republicans. But what if the Republicans run a pro-choice moderate (like Susan Collins) as one of their candidates? Such a candidate may attract a great deal of support from the most conservative third of usually-Democratic voters. Or Democrats could run a moderate who is more akin to Joe Manchin to try to win a fourth seat. Or a far-left Green candidate could wrest a seat from the Democrats, a Libertarian running on a platform of deregulation and market-based climate action could make a splash, etc. With more than two ideologies available, there are many possibilities for competition.
Another factor: with single-winner districts, a popular incumbent may be so heavily favored for reelection that no one else even bothers to run. This can be easily observed in looking at city council elections where some council members are elected from wards and others are elected at-large (most often using Block Plurality). You’ll probably find a lot of uncontested single-winner races, but for the multi-winner at-large races there tend to be at least twice as many candidates as there are seats to fill, and I’ve never seen a truly uncontested multi-winner contest. (At least, this is what I found from looking at dozens of municipal elections in Colorado.)
Why is more competition good? With fewer non-competitive districts, there are fewer people that political parties aren’t incentivized to care about. This is a strong argument, but it’s the only argument I’m confident in. There are other arguments that could be valid, but for which I have substantial skepticism:
- More competition means higher voter turnout. (I’m confident that this is true, but I’m not convinced that having higher turnout really matters.)
- With more competition, it sounds like it should be easier to oust corrupt incumbents. (The flip side is that, with proportional representation, corrupt incumbents need a smaller share of the vote to stay in power. This argument may be completely wrong.)
- More competition means that voters have more options. (I don’t consider this intrinsically important, and it sounds more like something that makes voting feel nice than like something that yields better policy outcomes or reduces polarization.)
The downside of having more competition is that needing to reach out to more voters means needing more money; there are significant cost savings in being able to identify a bloc of voters as unimportant and forgoing any outreach to them. More competition makes politics more expensive, so having proportional representation may increase the role of money in politics. (Also, candidates going door-to-door or showing up at local events is less viable with multi-winner districts since multi-winner districts tend to be bigger.)
Is proportional representation necessary?
Those are the benefits of proportional representation: better minority representation, an end to the two-party system, no more elections getting decided by gerrymandering or geography, and elections being more competitive. But is PR actually needed to realize these benefits, or can they be achieved just as effectively through other means?
- In general, good minority representation and eliminating the importance of geography can’t be achieved without something that is at least extremely similar to PR. Single-winner districts can sometimes be drawn with the goal of favoring minorities, but this is impossible with evenly distributed groups such as Republicans in Massachusetts.
- Ending the two-party system might be achievable through a major cultural shift or by switching to a better single-winner voting method, but this seems unlikely. Australia has a two-party system only in the chamber that is elected from single-winner districts, despite using Instant Runoff Voting instead of Plurality (and the shortcoming of Instant Runoff Voting compared to other single-winner voting methods aren’t the problem either).
- Getting more competition and eliminating non-competitive “safe” districts probably requires something like PR too. Having more candidates can also be achieved with (multi-winner but non-proportional) Block Plurality, but such a district can still be a collection of safe seats in a partisan sense; a five-winner district that is 70% Democratic will always elect five Democrats under Block Plurality, even if there is competition over which Democrats get elected.
So we need something that’s at least very similar to PR to realize these benefits. But what about using a proportional voting method that incorporates a desirable bias of some sort?
A pro-minority bias — one that gives smaller parties or minority factions more seats than their share of their vote — seems tempting. But it’s exploitable, as larger parties can subdivide in order to receive the benefits given to small parties, and it weakens majority rule by giving coalitions that include a lot of small parties an unfair advantage over coalitions composed of just a few larger parties.
But a pro-centrist bias seems workable. Suppose a five-winner district is 40% Left, 20% Center, and 40% Right. The proportional outcome is to elect two Left candidates, two Right candidates, and one Center candidate. But what if you elected three Center candidates, one Left candidate, and one Right candidate instead? This isn’t all that proportional, but it’s still brings all the benefits of proportional representation. All political factions get representation, there isn’t a two-party system, there’s no reason for things to be less competitive, and it needn’t reintroduce a dependence on geography when there are many such districts.
But a pro-centrist bias may weaken majority rule. If 51% of voters are Democrats and a voting method has a pro-centrist bias, this bias may cause the median representative to be a centrist independent instead of a Democrat. But this should be a mild violation of majority rule since it would elevate centrists instead of the opposing party. And whether a given candidate or party is centrist is determined in part by Democratic voters; if all of them think that the Middle of the Road Party is no better than the Republicans, this party wouldn’t be considered centrist for the purpose of a voting method’s pro-centrist bias.
Of course, a pro-centrist bias can only go so far; it wouldn’t work to elect only centrist candidates since that would entail major ideologies going unrepresented. (A pro-extremist bias wouldn’t prevent the benefits of PR from being realized either, but it doesn’t seem desirable.)
Ultimately, I think it’s important to have a voting method that fits the common-sense idea of proportional representation. But this is a large and fuzzy target; a pro-centrist bias is compatible with it, as is (perhaps to a lesser extent) denying representation to especially small factions, as can be done via thresholds under Party List or by having a relatively low district magnitude. How close a proportional voting method comes to the center of this target has little bearing on the extent to which it yields the benefits of proportional representation, so, when it comes to comparing proportional voting methods, the question of how proportional they are is relatively unimportant.