Undercover Economist

Why democratic elections are always flawed

I sometimes wonder if we expect more than we should from democracy

Most Britons are unhappy with the result of the UK general election. That is the logical conclusion, given that 63 per cent of voters cast their vote for someone other than David Cameron’s Conservatives. Nevertheless, the Conservatives surprised even themselves by winning more seats than every other party put together.

From the point of view of the losers, this state of affairs seems outrageous. It is sometimes said that splits on the left of British politics have prevented what should have been a solid leftwing majority and allowed the rightwing views of a minority to prevail. Yet the right is also split: the Conservatives and the UK Independence Party attracted more than 50 per cent of the vote between them. Leftwingers frustrated by the election result should blame the voters before they blame the voting system.

Critics of the British voting system do have a point. An analysis by Jack Blumenau and Simon Hix of the London School of Economics suggests that the disparity between votes cast and seats won has been widening for many decades, the consequence of the large number of votes now cast for smaller parties.

Since each seat is decided separately, votes cast for losing candidates simply do not count. It is possible to stack up a hefty pile of such votes while winning only a single seat — just ask Ukip’s Nigel Farage, who failed to be elected despite leading a party that attracted 3.9 million votes. The Conservatives earned about 34,000 votes per seat won, and Labour about 40,000 votes. The Scottish Nationalists needed just 26,000 votes per seat. Nick Clegg’s derided Liberal Democrats required more than 300,000 votes for each of their eight seats. Supporters of both Ukip and the Liberal Democrats might well feel disenfranchised, as might the Greens and the many Scots who voted for parties other than the SNP. Still, rules are rules and everyone knew the rules before they started to play the game.

Yet rules can be changed. And perhaps they should be. But to what? Clever schemes abound: the D’Hondt method offers something close to proportional representation while maintaining a link to local constituencies; the Borda count attempts to measure the strength of preferences; the alternative vote, AV, is designed to allow people to cast a conditional vote for whichever of several parties might find itself with a chance of winning.

AV was decisively rejected by the British in a referendum in 2011. But perhaps referendums themselves need looking at. Consider a referendum on an issue such as gay marriage. A small number of people — gay people who might wish to get married — have a tremendous interest in liberalisation. But no matter how strongly they feel, they get just a single vote each and so they have had to wait while the weakly held views of the majority slowly move in a tolerant direction.

Glen Weyl, an economist, argues that in such cases we might want to hold a referendum that allows people to express their strongly held beliefs by buying multiple votes at increasing cost: one vote costs $1; two votes cost $4; 1,000 votes cost $1m. Weyl calls this idea “quadratic voting”. It has some appealing theoretical properties but to the layperson it looks alarming. Expect to see it used in TV talent shows.

I am all in favour of improving institutions when we can but I sometimes wonder if we expect more than we should from democracy. There are two deep reasons why democratic elections are always flawed.

The first is that voters are, quite rationally, rather ignorant about politics. Sensible people vote to express themselves or out of a sense of duty, not because they harbour the illusion that it might be their vote that swings the entire election. Quite sensibly, then, people who devote hours to researching a new phone will not waste time researching which party to support.

The second reason is Nobel laureate Ken Arrow’s “impossibility theorem”, one of the most celebrated and misunderstood results in economics. Arrow’s theorem is often described as showing that there is no voting system that will reflect what society truly prefers. Arrow actually showed something more profound: that it makes little sense to speak of what “society truly prefers”. That very idea is incoherent. And those who expect that a democratic election will ever give society what it “truly prefers” will have to get used to disappointment.

Written for and first published at ft.com.