Scotland and Wales-some simulations

For those that didn't notice, elections to Scotland, Northern Ireland and Wales's devolved Parliaments, as well as to various local authorities in England and the mayoralty and assembly of Greater London, took place on May 5.

In 1997, referendums took place in Scotland and Wales on whether country-wide Parliaments should be introduced to govern the two regions. In both cases, the proposal was successful. Further referendums were held in Northern Ireland (where the assembly was part of a comprehensive peace deal) and London in 1998, where devolved assemblies were introduced.

In order to gain cross-party support for these assemblies, the Blair Labour government decided against using a single-member plurality electoral system to elect members of the Scottish and Welsh assemblies. Instead, it was decided to use a mixed-member proportional representation system.

In London, there are fourteen members elected in single-member districts, and eleven members elected from closed party lists. Voters have two votes; one for their single-member district candidate, and one for a party list. In the districts, the candidate with the most votes is elected. Following this, all the list votes are tallied up, and the eleven party list seats are allocated using the D'Hondt method of party list proportional representation.

However, it is modified somewhat. The number that the party's vote is divided by is increased by the number of single-member districts the party has won. So, if a party wins five single-member districts, its vote will be divided by six at the beginning of the D'Hondt count. A party needs to win 5% of the vote to win any party-list seats.

A very similar electoral system is used for Scotland and Wales. Scotland is divided into eight districts. In each of these, eight to ten members are elected in single-member districts, and seven members are elected using party-lists. An identical system to the one used in London is used to allocate the party-list seats: however, there is no threshold. This is made up for by the substantially reduced number of seats to be elected in these districts. If we add all the seats in a district together, we get an effective threshold of about 4.4%; however, this is probably lower in practice, and may depend on the results of the single-member districts.

Wales is identical, except the districts are somewhat smaller. Each district elects between seven to nine members in single-member districts, and four members in the party-list system. This makes it very much less proportional

Scottish results

A couple of months ago, I wrote this blog post on decoy lists in Scotland. As it turns out, Nicola Sturgeon does not read my blog, and as a result no party in Scotland tried the decoy list strategy. While the Greens only ran district candidates in a small number of districts, this was more so that they could focus their efforts on winning list votes.

However, what if the parties had used decoy lists? The below chart shows what would have happened if only the SNP had used decoy lists; that is, their list candidates and district candidates were not affiliated to each other. 

Compared to the actual results, the SNP would gain thirty seats. The Conservatives would lose thirteen seats, Labour eleven, the Greens five, and the Liberal Democrats one. Rather than their current position of governing in minority, with support from one of the opposition parties needed to back legislation, the SNP would have a comfortable majority. 

But what if other parties had followed the SNP's strategy? This would make Scotland effectively mixed-member majoritarian, an electoral system where the aim is to add some proportionality onto a majoritarian system. The goal of mixed-member proportional representation is to make the entire result proportional.

The SNP would only gain twenty-four seats this time, with their list gains balanced out by the ability of the Conservative and Labour parties to make the same sort of gains. However, Labour would lose ten seats and the Conservatives would win nine; Labour's substantial weakness in the constituencies would be exaggerated. The Liberal Democrats would win the same number of seats, as most of their seats were won in constituencies. The Greens would be, proportionally, the biggest losers; they would lose all but one of their seats.

Following the election, and the SNP's failure to win a majority, Alex Salmond stated on his weekly radio show that the electoral system used in Scotland was responsible for the result. In an interview that appears fairly incoherent, Salmond claimed that the 'variations' that led to the SNP not winning a majority were the fault of the regional list system.

But is this the case? Well, we can test this. The below charts show the results of the election using a national list system; to be precise, such a system would allocate all the list seats in one district, covering the entirety of Scotland. As such a system would be very proportional, there are two simulations; one with a 5% threshold, and one without.

It's quite clear that Salmond is wrong, at least in that sense. The national list system would not have given the SNP a majority; in fact, in both simulations they would be in an even weaker position. They would have won only the 59 seats they won in the constituencies. As you can see, there are few other differences; the Liberal Democrats and Greens would both gain a couple of seats, as they were disadvantaged by having their vote spread out in the districts, while UKIP, which won only 2% of the nationwide constituency vote, would win two seats without a threshold, and none with one.

In the same article, Salmond also supports the introduction of a single-vote MMP system. Under such a system, voters would have one vote, for a constituency candidate. That vote, however, would count as a party vote for the party that constituency candidate was affiliated to.

Were such a system implemented in Scotland, it would have a number of effects. For a start, it would be bad news for the Green Party. The Greens only ran a few constituency candidates, and focused their efforts on the list. In most constituencies, the Greens have little chance of election, and Green voters would be pushed towards tactically voting for a more competitive contender. This would mean the Greens would win fewer party votes, and thus substantially fewer seats.

At the same time, tactical voting in the constituencies would go down somewhat. Despite the above, some people would still vote Green in the constituencies, but they would obviously vote for someone else if there was no Green candidate, which is the case in most places. Abolishing the list ballot would mean the Greens would be forced to run a candidate in every seat. The same would go for weak Liberal Democrats, Conservative and Labour candidates; partisans for these parties would be less likely to switch to more viable constituency candidates closer to their party affiliation if that meant that they could not vote in any way for the party of their first choice.

Wales

The results of the Welsh elections were, on the face of it, rather unusual. Labour's vote in the constituencies dropped by 7.6% and their regional vote dropped by about 5%, but they only lost one seat. So, how did this happen?

Well, there are a number of reasons for this. First of all, the big gainer in these Welsh elections was UKIP. This Eurosceptic party has gained substantial support in recent years, and while it failed to make much of a mark in Scotland, at the 2015 general election the party polled 13.6% in Wales. 

Rather than focusing their attention on the party-list seats, like the Scottish Greens, UKIP ran constituency candidates in all but two constituencies. This meant that the opposition vote became more split. Labour also had an efficient vote spread; only 16% of the vote for Labour was cast in the constituencies that they did not win (which constituted 26% of the electorate).

This on its own does not cause disproportionality. For example, in Scotland, the Scottish National Party received 80% of the district seats, off only 46.5% of the vote. Wales had a similar imbalance, with 67% of the seats for Labour for 35% of the vote. However, Wales was twice as disproportional as Scotland, using the Gallagher Index.

The key issue causing disproportionality in Wales is the very low number of party-list seats. In Wales, the percentage of list seats is very low for an MMP jurisdiction.

What if Wales had a similar amount of proportional seats to Scotland? I have simulated this by adding three seats to each region, thus making the percentage of list seats 46.67% (compared to 45.74% for Scotland).

Labour would be the main loser, falling from one seat short of a majority to just 40% of the Assembly. Plaid Cymru, the Conservatives and UKIP would be slightly better off. The Liberal Democrats would be the big gainer, winning four seats, up from just one. 

Overall, the Gallagher Index of disproportionality for Wales would decrease from 13.03 to 7.25, closer to Scotland's 5.6. In an area like Wales, where one party is dominant and the opposition is heavily split, increasing the number of party-list compensatory seats would be a sensible step to ensure an accountable regional government.