Proportional Representation

(Skip to summary)   (Separate "executive summary")   (Return to main page)


The election of a representative from a single-member district falls to the candidate receiving the highest vote. There is no representation for groups supporting other candidates, except insofar as the person elected [chooses to] give voice to the members of such groups. Hence, direct representation is not afforded minority groups. To give them such representation, various European countries, with Norway taking the lead in its Constitution of 1814, have provided for systems of proportional representation, which necessarily operate in plural-member districts. If, for example, seven persons are to be elected to a... legislative body, proportional representation... may be accomplished by having three members from one party with 42% of the total vote, two from another party with 30%, and one each from two other parties sharing equally the remaining number of votes.

The previous paragraph is copied from a 1964 Encyclopedia Americana article by Dr. Spencer D. Albright, University of Richmond, VA.

Range voting is not a PR system

Range voting, which we only recommend for single-winner elections, thus is not a form of proportional representation according to Albright's definition above. If there are 1000 Range Voting elections between two parties A and B with 66% and 34% of the voters, then Range Voting will elect 1000 party-A winners and zero party-B winners (assuming the voters hew strongly enough to their party lines) – which is very disproportional.

However, there is an alternate form of Range Voting which is designed for multiwinner elections and which is designed to exhibit proportionality. It is called Reweighted Range Voting (RRV) and is described in paper #78 here and more simply on this web page. There is also an unconventional PR voting system somewhat related to range voting, called "asset voting," discussed in paper #77 there.

Is PR constitutional or obtainable in the USA?

An earlier version of this page argued that PR's constitutionality was unclear in view of a long sequence of random-seeming Supreme Court decisions. However in Sept. 2007 we changed our stance to say that PR is constitutional.

Proportional representation would be essentially unobtainable in US Federal and state governments if it required a US Constitutional (or State Constitutional) amendment, since that would take 2/3-supermajority votes, which either of the two major parties would be willing and able to block forever. (And it definitely is forbidden for US congress elections by current Federal laws.)

Historical note on US laws concerning districting vis-a-vis PR: In 1996, congresswoman Cynthia McKinney (who later ran as the Green Party candidate for US President) wrote, but failed to pass, bill HR 2545, which would have overridden previous US law that had outlawed multi-member districts, i.e. had made PR illegal. She re-introduced a similar bill, HR 1189, in 2001. It failed again. Then she tried again with HR 2690 in 2005. It failed yet again. According to McKinney's bills, it is the 1967 law discussed below which mandates single-winner districts. Chronological list of laws:
1842 Apportionment act:
Representatives "should be elected by districts composed of contiguous territory equal in number to the number of Representatives to which each said state shall be entitled, no one district electing more than one Representative." (5 Stat. 491.)
1872 Apportionment act:
Districts should contain "as nearly as practicable an equal number of inhabitants." (17 Stat. 492.)
1901 Apportionment act:
Districts must be made up of "contiguous and compact territory and containing as nearly as practicable an equal number of inhabitants." (26 Stat. 736.) [None of the preceding standards were ever enforced. This one was largely and often dramatically ignored.]
1929 Permanent Apportionment act:
No districting standards. (46 Stat. 21.)
1967 "An Act for the relief of Doctor Ricardo Vallejo Samala and to provide for congressional redistricting," approved 14 December 1967 as Public Law 90-196:
Congress reimposed the requirement that Representatives must run from single-member districts, and forbade "at large" (i.e. statewide) representatives. (81 Stat. 581.) Various states at various times in US history had elected congressmen at large (e.g. CA in 1861) or via multimember districts (examples), which indicates they, while illegal, are constitutional. This 1967 law remains in effect as of 2009. (A quote from it.) It was codified as Title 2, U.S.C.§2c. It arose in a peculiar way as a Senate amendment to a House-passed private immigration act – H.R. 2275, 90th Congress, "an act for the relief of Dr. Ricardo Vallejo Samala, and provide for congressional redistricting." No hearings were held or reports issued.

Indeed, PR was tried in several US city governments (most notably New York City 1936-1947 and Cincinnati 1924-1957) and the two major parties eventually both fought against it and played an essential role in its abandonment in almost all the US cities that tried it. (As of 2000, only a single US city – Cambridge MA – still uses PR.)

However since it is constitutional, it could be enacted by mere act of congress, requiring a mere 51% vote plus presidential approval. This is not "essentially unobtainable" but merely "extremely difficult." I'd still say it won't happen so long as there is 2-party domination because the two major parties will both block it. But it at least becomes possible it might happen if for some unlikely reason one party wanted it.

The whole single-winner district-representative concept nowadays seems built in to the whole structure of the US federal and most state governments, and indeed there are laws (e.g. voting rights act) about the allowed ways to draw district boundaries which take it as a given that there will be a district from which comes a representative, and where it is generally regarded as unconstitutional to disobey that law (an opinion that has been stated by high courts). Nevertheless, it is not explicitly stated in the (Federal) constitution that representatives must come from single-member districts, and multimember districts have been used in the past.

Range voting, however, is attainable without constitutional amendments. Note to PR advocates in the USA: if you want PR, we very much doubt you are going to get it without first getting Range Voting as a preparatory step (which in turn will gradually diminish 2-party domination). If you think IRV is a good stepping stone to PR, then you are wrong because IRV leads to 2-party domination and the two parties can and will always block PR. Range voting offers a far better chance of breaking 2-party domination and is a stepping stone to Asset Voting and Reweighted Range Voting, two apparently-superior forms of PR.

Another point: to those who think PR will break "2-party domination" – although you are on the right track, note that PR cannot ever be used for senators, the president, or state governors, nor for congressmen from 1-rep states like Alaska and Wyoming (and PR also would have little effect on 2-party domination in 2-rep states and indeed might be deleterious to democracy if adopted in such a state).

In principle, you could overturn the whole US setup which specifies, e.g, two senators per state elected in staggered terms – thus allowing PR to extend its reach into the senate. However, in practice, that will never happen because the senators will block that change forever. Hence PR can, at best, cover the congressmen from large states, but not the small-state ones, and not any senators, governors, or presidents. This would be a limited kind of PR and plausibly would exacerbate current power imbalances between large and small states.

Therefore, PR might, even if adopted wholesale thoughout the USA, still lead to pretty heavy 2-party domination (e.g. Senate 100%) unless the whole government structure is completely redone, which seems unlikely to happen.

PR systems

The most common PR proposals are

  1. Multiwinner voting systems such as Hare/Droop STV are used to choose the MPs from multimember districts (or other voting systems such as my own asset voting and reweighted range voting might be better, papers #77 and 78 here). Hare/Droop STV is advocated by the British Electoral Reform Society.
  2. Party-list systems: you don't vote for people, you vote for political parties. Each party wins an amount of seats proportional to its vote count. Each party then awards them to its candidates selected in descending order from a prepublished list.
  3. "Mixed" systems, e.g. elect 75% of parliament with some single winner system, but then the votes for each party (nationwide popular totals) are used to "top up" the remaining 25% of the parliament seats from party lists in such a way as to achieve PR. This is advocated by the Hansard society and both the Germans and New Zealanders (2002) do something like that. (Update: A 2010 poll suggests the New Zealanders will repeal their MMP system in an upcoming referendum. Later update: they did not repeal it.) Mixed systems can try to get the best of both worlds (single-winner quality and PR representativeness for minorities) to at least some degree, but at the cost of forcing political parties down our throats and rendering "independent" (non-party affiliated) politicians a thing of the past. Think that doesn't matter, we'll just go with parties? Then you may not have thought through the implications. New Zealand enacted a "party hopping" law saying that anybody leaving a party has to be tossed out of parliament (and arguably this requirement is forced by the logic of the MMP system). That causes, essentially, massive party loyalty to be built into and enforced by the law itself – disagreeable!
  4. In some sense the ultimate PR system is simply not to have any votes and voters at all; the legislature is simply selected randomly! For some reason, although this is commonly done (modulo "challenges") for juries, I am unaware of anyplace today that does it for legislatures. This practice was introduced in Ancient Athens where a machine was invented for generating the randomness.

PR pluses and minuses

Ignoring the fact that PR is unobtainable in the USA – is it, abstractly, a good idea? Quite probably. It has been criticized, though – single-winner systems also have some advantages. For example, the German "Weimar Republic" had a party-list PR system. It suffered massive unemployment and hyperinflation and then it gave Hitler and his Nazi party the largest share of the seats in the Reichstag (although still not a majority). Hitler then was able to use that power base plus some extra-legal maneuvers to take over completely. So PR did not work out well in that particular case.

Here is a list of commonly-claimed pluses and minuses for PR. The minuses mainly are taken from Hain's and Katz's anti-PR books; the pluses mainly from Amy's pro-PR book or his PR library web site.

PR minuses PR pluses

In Hain's view the most important thing is politicians have to be held accountable by voters and a lot of PR systems (both party list and STV) sever or weaken the "vital links between local communities and their elected MPs." He cites a magazine in the 1980s that published a month before the election (based on opinion polls) the names of 448 candidates who were "certain to be elected," with only 129 out of 577 seats in doubt. This proved extremely accurate. The point is, if somebody high in a party list is a scoundrel, there is no way for voters to make him pay in a party list system. Hain in particular regards Israel (where the entire country is just one "district") as a textbook example of total dysfunctionality of party-list PR.

Even in systems (e.g. Holland) where you can vote for parties and for individuals (thus allowing somebody far down a party's list still to be elected) there is still a huge benefit to being high on the party list – this guarantees election rather than it merely being possible. That causes everybody to vote in partyline bloc fashion – the coalition versus the others – since if they vote against the deals made by their party leaders they fear demotion on the list. When that happens, the whole ideal PR strives for, is thrown in the garbage!

Hain attacks STV as actually not generating very good proportionality, e.g. cites Ireland 1984 where Labour got 8% of votes but zero seats thanks to cutoffs and the small cardinality of each district. (Party list and combination systems both would have given Labour some seats.) STV would have given Labour a seat in a 13-member district, but elections with 13 winners and maybe 50 candidates, seem impossible for voters to keep track of. On the other hand it has been argued that STV by zeroing out only sufficiently small parties, can produce beneficial "centrism" and "majority consensus" and favor parties that are more "unified" (although some would deride that as favoring "robotic" parties; and any bias against small parties has the bad consequence that it prevens new parties from being "born" and "surviving infancy" thus effectively shutting down new ideas).

Many PR systems are complicated and cannot run on simple voting machines, whereas many single-winner systems are simple.

Hain says PR is more likely to lead to "coalition" governments. Then "centre parties benefit" and tend to hold sway, while sometimes small fringe groups get tremendous power to make or break coalitions. And the parties can become polarized with the most-polar party members getting promoted high on their party's lists. Katz criticizes PR as consequently leading to "gridlock" governments dominated by "interparty bickering." Professor Harrie de Swart told me he thinks that in many-party coalition governments like Holland 2008 another problem is lack of responsibility: who is responsible for doing anything? They can't be identified and they can't be penalized. Both Katz and Hain worry that (as a further consequence of all these factors) party-list PR systems can lead to instability of the government. And indeed, party-list PR led to takeovers by both Hitler in Germany and the Fascists in Italy, and something somewhat similar may be happening in 2000 Israel.

PR leads to extremist parties getting seats whereas with single-winner they'd not be able to get seats; this clogs up parliament with wackos who can never cooperate. (E.g. Israel has never had a majority party. France abandoned PR after a brief experiment of using it to do their 1986 elections.)

In contrast, single-winner voting systems can lead to stable 2-party domination. The 2-party stranglehold part may not be so good, but at least it is highly stable. (Incidentally, it is a myth that single-winner systems always lead to 2-party domination. In fact, the plurality plus top-2-runoff (2nd round) single-winner system is well known – based on great experience in many countries – usually to lead to more than two powerful parties.)

With PR maybe minorities can "get some representation" but so what? They are still in the minority and still will lose exactly the same battles they would have lost anyhow with 0% representation (assuming there is bloc voting, which often happens under PR since there is a premium on party loyalty in party-list systems) so in practice it doesn't make much difference. Further, as we saw, even with party-list PR, women were only getting 7-26% representation, so proportional aint necessarily so proportional! So it's a con game, don't fall for it.

Proportionality, and only proportionality, truly represents the electorate and what it wants. With single-winner, minority views and groups can be "zeroed out" of power; that can be even worse with gerrymandering. Sometimes even majority groups can get zapped: e.g. under single-winner systems in 1983 women were 52% of voters but 3.5% of British MPs (and during 1945-1990 it ranged from 3 to 5% in both Britain, Canada, USA, and New Zealand). That is highly disproportional! (Similarly, in 1983 there were zero black MPs although 4.5% of the British population is black, and in 2001 there were zero black US Senators although about 12% of the US population is black.) Hain on page 70 gives a table showing party-list PR indeed causes more women to be elected (7-26%). Also Amy observes that "In the 1994 German election, the percentage of women elected in the single-member districts was 13% – about the same as in the USA – while the number elected from the party list PR contests was 39%." But Hain's table also indicated PR-STV produces only a "marginal" improvement for women (4-9% for Malta & Ireland).

So with true PR, the legislature will vote for what society wants, thus benefiting society.

And we'd get some Green, Libertarian, and Conservative congresspeople, and actual changes. In Germany's national elections of 1994, the Green Party emerged as the third largest, increasing their legislative seats by six times, from 8 to 49. The pacifist Party for Democratic Socialism won 30 seats. Women increased their legislative seat count to 176, which is 26%. That is real change, and it's representation the USA just doesn't have.

With PR we often get a lot more than two political parties.

Single-winner systems often lead to permanent machine domination by one or two parties that can last for 50 years and totally shut other groups out of power, e.g. Democrat "Jim Crow" control of the US South 1900-1960 with blacks totally shut out of power and oppressed. In the 1996 elections for the U.S. House of Representatives, the Democrats won 66% of Massachusetts votes, but received 100% of the state's ten seats while the Republicans got 33% of the votes but were zeroed out, receiving no representation. (That same year in Oklahoma, Republicans won 61% of the vote and won all six seats. In the British elections of 1951 and February 1974, parties won more seats with fewer popular votes than its opponents.) Instant Runoff Voting does not fix this problem, e.g. Hain claims in 1948 in Alberta Canada an IRV election gave all the seats to one party despite only 58% support for it, and in 1967 in Victoria Australia the Liberals won 3 times as many seats as Labour despite fewer top-rank votes. [In British Columbia Canada in 2001, the votes were 57.6% Liberal, 21.6% NDP, and 12.4% Green, but the NDP won only 2 seats and the Greens zero.]

Turnout would increase: Voters in PR countries typically turnout at rates of 70-80%, compared to 50% or less in the USA. (But turnout probably would also increase if the USA were to use range voting, which is not PR, but rather a superior single-winner system.) But Radek Sikorski in his article On the brink of chaos [National Review 43,22 (Dec 1991) 24-25] pointed out that in Poland under PR, turnout was 40% (versus previously without PR 60%). He opened his article with "If ever the evil of PR needed to be demonstrated, then the Polish elections have proved the case once and for all." There were over 100 parties participating in these elections, resulting in 18 seats for the "Friends of Beer" party, etc.

But many single-winner voting systems tend to yield 2-party domination, which in the USA combined with gerrymandering nd senate "filibuster" rules to produce massively a gridlocked and useless congress around 2010-2014 which broke records for non-productivity.

Stephen Hill says PR would also serve as a campaign finance reform: Since it requires a lower percentage of votes for a party or independent candidate to win a seat, it also lowers the amount of money those candidates need to spend to win a seat. (But I'm unsure that made sense.) And it would make the media cover alternative-party views for a change.

Single-winner systems often lead to dishonest "strategic voting," which causes crazy and undesirable effects; whereas the proportionality guarantee in PR systems perhaps causes voters to be more honest. But: There are some common forms of strategic voting in STV systems called "free riding" where you intentionally do not vote for popular candidates in order to maximize your vote weight for the candidates that are less certain.

PR systems are largely or entirely immune to gerrymandering.

Many books have been written on the subject with no clear conclusion being reached. I have an idea that my asset voting proposal (#77 here) may be able to yield a better kind of PR subject to fewer criticisms that it leads to battling extremist factions and/or gridlock. However, at present that is highly speculative.

My personal view is I am unsure that PR is an improvement although I'm slightly biased toward "yes," but I know that range voting is an improved single-winner system because we can measure it with Bayesian Regret and the results are very convincing. (It is a lot easier to understand and measure the relative quality of single-winner voting systems since we can use Bayesian Regret, but no analogous quantity is known for multiwinner systems.)

What does economics say?

Margit Tavits found a clear statistical relationship: as democracies become "more PR," their governments become larger (as measured by their expenditures as a fraction of GDP). This finding was highly statistically significant – 99.9% confidence (see her plot on page 39). So if you are one of those who says "big government is bad," stay away from PR!

Lijphart found the more-PR democracies had more economic equality, i.e. less wealth disparity. So if you are one of those who says "economic inequality is bad," support PR.

What if we try big statistical analyses to see which leads to a better economic situation – "consensus" (i.e. PR) or "majoritarian" democracy?

The history of attempts to do that has been messy.

Arend Lijphart devoted a substantial fraction of his life to just this, writing several books on the subject. In 1994 Lijphart [Democracies – forms, performance, and constitutional engineering, Europ. J. Political Research 25,1 (Jan 1994) 1-17; Electoral Systems and Party Systems: A Study of Twenty-Seven Democracies, 1945-1990, Oxford University Press 1994] found no significant differences between PR and non-PR democracies in either peace, inflation, or economic growth, but found that PR democracies had higher employment – which sounded good until you noted from page 10 of his book the employment "advantage" vanished under a different analytic method.

However, this changed. M.L.Crepaz in 1996 [Consensus vs Majoritarian Democracy, Comparative Political Studies 29 (1996) 4-26] and Lijphart again but now with new, bigger and better, statistical studies [Patterns of Democracy: Government Forms and Performance in Thirty-Six Countries New Haven CT: Yale University Press, 1999; Reflections – dimensions of democracy, Europ. J. Political Research 31 (1997) 195-204 ] claimed that PR produced both higher employment and lower inflation (supposedly slight, but clear, advantages for PR). Consequently, Lijphart proclaimed: "The results of the statistical analysis were clear and almost completely unequivocal... overall consensus (i.e. PR) democracy is superior to majoritarian democracy... I have no second thoughts about these conclusions."

Sounds great – at last a clear conclusion that PR is better. Right? Wrong!

Liam Anderson [The implications of institutional design for macroeconomic performance – Reassessing the claims of consensus democracy, Comparative Political Studies 34,4 (May 2001) 429-452 ] hit back by pointing out that the economic improvements that Lijphart attributed to PR, could instead be explained by (1) greater central bank independence and (2) greater corporatism. "Corporatism" refers to the idea that corporations and labor unions (with greater or lesser government supervisory involvement) enter a negotiated tradeoff assuring labor quiescence and wage-demand moderation in exchange for employment guarantees and increased influence for labor in setting economic and social policy. Notion discussed in D.R. Cameron: Social democracy, corporatism, Labor quiescence, and the representation of economic interests in advanced Capitalist society, pp. 143-178 in JM Goldthorpe (ed.) Order and conflict in contemporary capitalism, Oxford Clarendon Press 1984. Indeed, Anderson's statistical study showed that

So this is precisely the opposite of Lijphart's conclusion: Anderson finds that making a democracy more-PR while holding central bank independence and corporatism fixed instead worsens both inflation and unemployment!

OK, great – we now have a fairly clear conclusion that PR is worse. Right? Again, no.

See, the government, by making laws and creating policies, has a high degree of influence on how independent the central bank is, and how much corporatism there is. So you could argue that the PR governments caused those things, in which case Anderson misled us – really making the government PR has a good economic effect. But since this influence is only partial, perhaps the bad effects outweigh the good... or maybe not...

So in summary, it is all as clear as mud! After all the sound and fury from Lijphart, Crepaz, and Anderson, my net conclusion is nobody won and I see no clear evidence that the PR/not-PR choice has any effect on inflation, unemployment, or economic growth.

The very fact that neither PR nor non-PR (mainly plurality-voting-based) democracies exhibit obvious superiority over the other suggests to me that a democracy based on range voting could be significantly superior to either (since there is good reason to believe RV will be significantly superior to plurality-based majoritarian democracy).

PR versus single-winner – the fundamental conflict

Single-winner Proportional representation

Better quality of the winner and more accountability and scrutiny by the voters.

Better (in the sense that is more proportional) representation of the electorate.

This illustrates the desire-conflict in a nutshell. The ultimate in "perfect" representation is to use random selection, but then you get perfectly average people running the government. The Hare/Droop-STV-PR scheme sounds like it is designed to counter that criticisms by taking account to at least some degree of individual candidate quality – but it can actually refuse to elect landslide beats-all winners because it prefers instead to elect 5 approximately equally-bad "representatives." In contrast, with a good 5-way single-winner election method, you get somebody above 80th percentile in perceived ability (among candidates), running the country (in fact on average 83rd percentile) – better. You want to get both best possible quality & accountability and best possible representation, but no known voting system delivers both at the same time, so it's "pick your poison."

Return to main page