When the biggest party doesn’t form the government

Charlotte Minvielle, a former candidate for the New Popular Front (NFP), has an article in the Guardian in which she criticises Emmanuel Macron’s reluctance to consider a government led by the NFP, and calls for a Sixth Republic. (Hat-tip to Declan Gaffney for drawing my attention to it).

Although I think the NFP should have the first chance to form a government, and although I believe that parliamentarism is superior to both presidentialism and semipresidentialism, I disagree with Minvielle when she writes

“In any parliamentary democracy, the political force with the most elected representatives is allowed to form, and lead, a government. This is not the case in France, an anomaly in a region predominantly characterised by parliamentary democracies such as the UK, Spain and Germany”

This is not true, even in the countries Minvielle mentions. Betwen 1972 and 1983 the CDU/CSU was the largest political force in the Bundestag, but between 1974 and 1982 the Social Democratic Party governed with the FDP.

Coalitions excluding the largest party are moderately common in Scandinavia, where Minvielle has campaigned. For example: the Social Democrats won the most seats in the 2022 Riksdag election, but the present Swedish government is a right-wing coalition led by the Moderates.

We can go beyond citing specific counterexamples and show the rate at which the largest party fails to enter government. I’ll do this using data from ParlGov, which collects data on election results and government participation in a large set of (mostly parliamentary, mostly European) democracies.

suppressPackageStartupMessages({library(tidyverse)})
pg_url <- "https://parlgov.org/data/parlgov-development_csv-utf-8/view_cabinet.csv"
if (!file.exists("view_cabinet.csv")) {
    download.file(pg_url,
                  destfile = "view_cabinet.csv")
}
dat <- read.csv("view_cabinet.csv")

I’m going to exclude caretaker governments and pre-1945 governments.

dat <- dat |>
    filter(caretaker == 0) |>
    mutate(election_date = as.Date(election_date),
           start_date = as.Date(start_date)) |>
    filter(election_date > as.Date("1945-01-01")) |>
    dplyr::select(country_name_short,
                  cabinet_id, cabinet_name, start_date,
                  election_id, election_date, 
                  party_name_short, cabinet_party, prime_minister, seats)

Within each cabinet, I’m going to keep only the largest party. I’ll then ask whether that largest party is in the government.

largest_party <- dat |>
    group_by(cabinet_id, election_id) |>
### Start largest first
    arrange(desc(seats)) |>
    ### take the top row within each cabinet
    slice(1) |>
    ### remove that grouping
    ungroup() |>
    ### just focus on a single variable rather than the whole dataframe
    pull(cabinet_party) |>
    ### get the mean of this variable
    mean()

round(largest_party * 100)

[1] 83

Thus, for governments that ParlGov knows about between 1945 and the end of 2023, around 83% featured the largest party – which of course means that 17% didn’t.

This figure includes governments that formed immediately after an election and governments that formed after a post-election cabinet collapsed. Let’s see how the figure changes when we focus on post-election cabinets only.

largest_party <- dat |>
    group_by(election_id) |>
    ### pick rows from the earliest starting government
    filter(start_date == min(start_date)) |>
    group_by(election_id, cabinet_id) |>
### Start largest first
    arrange(desc(seats)) |>
    ### take the top row within each cabinet
    slice(1) |>
    ### remove that grouping
    ungroup() |>
    ### just focus on a single variable rather than the whole dataframe
    pull(cabinet_party) |>
    ### get the mean of this variable
    mean()

round(largest_party * 100)

[1] 85

When we look only at governments that formed after an election, 85% of governments included the largest party. This means that even in this most favourable case, around one in every seven governments doesn’t include the largest party.

When something happens in one in seven of the cases you’re analysing, it can’t be dismissed as an exception to a general rule. Although there are good reasons why the largest party has a claim to enter into government, that claim isn’t conclusive. We elect legislatures to decide by majority, and if you cannot find a legislative majority to support you that is a problem for your potential government, whether or not your government includes the largest party.