Disproportionality part II: France

(As explained yesterday in part 1, this week I’m examining the numbers from the British and French elections; next week we’ll get to the politics of it all.)

It’s one of the sad curiosities of the Australian political scene that if you asked a bunch of journalists which major democracy has an electoral system most like Australia’s, probably no more than one in ten would give you the right answer. It is of course France. It elects its lower house of parliament entirely from single-member districts that are a similar size to ours (averaging about 75,000 voters vs 114,000 here), and using a two-round system that approximates the effect of our preferential voting.

The main difference is that Australian voters have to exercise their choices all at once instead of spread across consecutive weekends. (That’s why the Americans call our system “instant runoff voting”.) That means less opportunity for reflection, but it also means that we can in effect run multiple rounds instead of just two. Losing candidates are eliminated one by one instead of all together, so a third (or, conceivably, fourth) placed finisher can occasionally leapfrog the top two to win a seat – something that isn’t possible in a classic two-round system.

But France has a refinement for that: provided that they meet a threshold of 12.5% of enrolment, other candidates can compete in the runoff if they choose. This year, an exceptionally large number met that test: 316 of them (311 thirds and five fourths), of whom 91 stayed in for the second round, creating 89 three-cornered contests and one four-cornered.*

And sure enough, one of the third-placegetters managed to pull off a win: in the third district of Ardèche, independent centre-right candidate Fabrice Brun trailed in the first round behind National Rally and the New Popular Front. But in the runoff, benefiting no doubt from the centrist candidate who had been eliminated, he just sneaked ahead, beating the far right by 35 votes – the narrowest margin in the country.

This highlights the downside of the French system: if more than two candidates stay in, the winner in the second round might not have a majority of the vote. Brun won with just 34.6%, the lowest of any, but of the 90 three- or four-cornered contests, only 11 were won with an actual majority. Twenty-four of the winners had less than 40%. (Detail of all the results is available at Le Monde; only minimal French required.)

Because of the success of the other parties in building a broad republican alliance, a large number of candidates were elected from second place – far more than you would ever see in an Australian election. There were 172 of them: 98 from the centre, 41 from the left, 32 from the centre-right and one regionalist; none at all from the far right.

Yesterday I produced a table showing the lack of proportionality in the British election; here’s the corresponding one for France (I’m using the interior ministry’s totals but merging a few lines for clarity):

Party	Votes	%	Actual	St-Lag.	d’Hondt
National Rally	9,379,092	29.3%	125	168	170
New Popular Front	9,042,485	28.2%	180	162	164
Ensemble	6,820,446	21.3%	159	123	124
Republicans (anti-RN)	2,106,166	6.6%	39	38	38
Republicans (pro-RN)	1,268,822	4.0%	17	23	23
Miscellaneous centre-right	1,154,785	3.6%	27	21	21
Miscellaneous left	490,898	1.5%	12	9	8
Miscellaneous centre	391,423	1.2%	6	7	7
Extreme left	366,594	1.1%	0	7	6
Regionalists	310,727	1.0%	9	6	5
Reconquest	238,934	0.7%	0	4	4
Independent Greens	182,478	0.6%	1	3	3
Independents	142,871	0.4%	1	3	2
Miscellaneous far right	149,789	0.5%	1	3	2
Radicals	12,434	0.0%	0	0	0

You can see immediately that the distortion, while still there, is much less serious. That’s partly because this was a close election – single-member districts have their worst effect in landslides – but it’s also because of the preferences. With left and centre in effect exchanging preferences, they both did better than either would have on its own, and the far right, getting no preferences to speak of, did correspondingly worse.

Further down there is some inequity, with both extreme left (basically Trotskyists) and Reconquest (the right-wing equivalent) missing out, while the regionalists, whose vote by definition is very geographically concentrated, did much better. But it’s very mild compared to the British case.

Why did the pundits get the result so wrong? The commentary we get here mostly comes from British and American sources; since they have no experience of preferential voting, it’s no surprise that they fail to appreciate its impact. Even in France, though, the projections made after the first round were hopelessly inaccurate. For five different pollsters, the mid-range of their estimate of the far right’s strength ranged from 224 to 275; even the bottom of the lowest range was 204.

They improved a bit during last week, as it became clear how much republican solidarity was being shown by the other parties, but up until polling day every one of them was still tipping a far right plurality, typically at around 200 seats. The final figure of 143 came as a shock to all of them.

But it shouldn’t have. This is not a case where voters changed their minds between the two rounds; they stuck with their original party, and where it had withdrawn or been eliminated they mostly followed its recommendation. As I summed up on Monday morning:

In effect, this election resolved itself into a two-party contest, the far right versus everyone else. “Everyone else” won a big majority of the vote … so not surprisingly, given a system of single-member districts, it won an even bigger majority of seats.

.

* Two of the withdrawals, including one of the fourth-place finishers, happened after the deadline, so some sources list Maine-et-Loire 5th and Rhône 8th as three- and four-cornered respectively. But because candidates provide their own ballot papers, even late withdrawals in France are extremely effective, so I have included both of those.