On Sunday 11th June, the first round of the French legislative election was held. On Sunday 18th the second round took place. Given the results of the previous week, Sunday’s election provided few surprises. There were some notable individual results: Marine Le Pen, the leader of the Front national (FN), was elected, even if her party did badly overall; Jean-Luc Mélenchon, the leader of the left party, La France Insoumise (LFI), was also returned and his party won enough seats to constitute a group in parliament, giving him speaking time; the former Socialist (PS) prime minister, Manuel Valls, was also returned, though only by a whisker and as a non-aligned candidate, indeed the Socialists had actively campaigned against him; Nathalie Kosciusko-Morizet, who is a high-profile figure from the right-wing Les Républicains (LR) and who had been the victim of an attack in the street while campaigning during the week, an attack that left her unconscious for a while, was defeated. However, the main event was the huge majority won by President Macron’s La République en Marche (LREM) party. Winning just 28.2% of the votes cast in the first round seven days before, the party ended up with about 300 seats in the 577-seat legislature after the second round. With the support of its MoDem ally, which won about 4.2% of the vote at the first round, President Macron now has the support of over 350 deputies in the legislature. This nice figure from Laurent de Boissieu’s blog neatly captures the many different components of the new French Assembly, but also indicates the huge majority for LREM and MoDem.
How did this happen? After all, before the first round of the presidential campaign, between the two rounds, and immediately after Macron’s victory, there were fears or claims that his party would not win a legislative majority and that he would not be able to govern, dragging France back to the bad old days of the Fourth Republic. Worse still, there were those who thought that he would face a period of cohabitation.
This was not the worry of a few isolated individuals. After the first round of the legislative election, L’Express put up a nice montage of politicians who argued that cohabitation was inevitable. But it wasn’t just politicians. At a certain point, Twitter got in a total fuss about the likelihood of cohabitation, though that’s what Twitter does.
But not everyone was so worried. Matthew Shugart said that the idea there would be a period of cohabitation was “nonsense“. And modesty almost, but not quite, forbids me from noting that we adopted a similar argument here.
What we have witnessed is instructive from a political science point of view. There is a well developed literature on how the legislative party system is shaped by direct presidential elections. (Anyone wanting a copy of the article with the literature review should just e-mail me). To simplify only a little, this work shows that when legislative elections follow shortly after the direct election of a powerful president, they typically return a presidential majority. This is exactly what we saw in France in 2017. For sure, the general argument is probabilistic, not deterministic. But the association is strong. The probability is high. So, the academic work hasn’t just generated something amounting to a reasonable guess that a certain outcome would occur. It suggested that there was a very good chance that Macron would get at least a working majority. In the end, he won the support of a huge majority, bigger than most academics had expected. The literature, though, was basically right. Why?
Well, the academics who have investigated this topic have made their argument on the basis of a statistical relationship, but they have also identified certain causal mechanisms to explain why we should expect honeymoon legislative elections to return a presidential majority. These mechanisms are all very general. They don’t always easily apply to specific countries. That’s all we can expect in large-n studies. However, and at the risk of committing an egregious ecological fallacy, the France 2017 case illustrates how these causal mechanisms can play out under local-level conditions.
We know that presidential elections are often the catalyst for party system realignments. This has been true in France before, but the evidence that this was going to be a realigning election was present even before the presidential election had finished. The election was catastrophic for the PS. It was hopelessly split and faced a strong challenge to its left. Going into a honeymoon legislative election in such a weakened state did not bode well for the PS. The presidential election also generated splits within LR. There were those, like the former prime minister, Alain Juppé, who were willing to work with LREM in a future Assembly, whereas there were others who were not. Going into a honeymoon legislative election in such a divided state did not bode well for LR. The FN was also in trouble. Le Pen did well to get through to the second ballot of the presidential election, but she did not perform as well as expected. The party’s support had been slipping even prior to her disastrous presidential debate with Macron. In the end, she was decisively beaten at the second round. After the election, there were reports that Le Pen was exhausted; the party was demoralised; there were also splits within the FN, even though it had done historically well. So, going into a honeymoon legislative election in such a state did not bode well for the FN either. In other words, presidential elections upend party systems. We saw how this general idea played out specifically in France in 2017.
A similar point applies to abstention rates. We know that abstention rates are higher in honeymoon legislative elections relative to the presidential election. We also know that it is typically the voters of the parties that lost the presidential election who stay at home. So, even when the presidential election does not generate a party system realignment of the sort that we saw in France in 2017, we should still expect the new president’s party to be the biggest beneficiary of the higher abstention rate at the legislative election. Again, this is exactly what happened in France. But it’s what we would expect to have happened.
There was a further element too. Macron’s victory at the presidential victory was bigger than expected. Thus, he had momentum. Once in office, he also had some excellent photo opportunities, meeting European and world leaders, even upstaging Donald Trump in the handshake stakes. There were one or two relatively minor concerns with his government, but by and large he kept his presidential promises in terms of government formation. In other words, presidential elections give the victor the potential to act, well, presidentially. This presidential lustre can rub off on to the president’s party at the legislative election. This is exactly what happened. In other words, like other presidents in a similar context, Macron benefited at the legislative elections from being the newly-elected president.
Of course, there are always local, idiosyncratic conditions. The electoral system clearly exaggerated the gains for LREM. But LREM was particularly well placed to benefit from the system. As a centrist party, it could win the support of right-wing voters who wanted to keep out left-wing candidates in LREM/left second-round duels; it could win the support of left-wing voters who wanted to keep out LR candidates in LREM/LR duels; it could also win the support of pretty much everyone in LREM/FN duels. So, strategically, it was better placed than some parties in equivalent situations. This particularity helped to inflate its majority. Also, Macron was not a long-time incumbent who had just been re-elected. He was a new figure and for some he did generate an enthusiasm for a new form of politics. In France 2017, all these local conditions worked in favour of his party at the legislative election. In other cases, they might not be present, helping to ensure that the relationship between presidential elections and legislative elections is not deterministic.
We are encouraged to talk confidently about our work (that’s Twitter again!), even when we do not always have grounds to be as confident as all that. More than that, we only have to look at opinion polling to see that even in an area where there has been a huge amount of research, where the sample is very large, and where there is competition in the academic market, we can still get things wrong. So, we should be modest about what we claim and certainly what we predict. However, we were on strong grounds to claim that cohabitation was very unlikely in France in 2017. We have an idea about the general processes. The local conditions were ripe. In short, politicians and Twitter didn’t need to get in such a fuss.