Tuesday, April 27, 2010

STV: A Study

As a Conservative supporter, I am deeply worried about the consequences of a coalition government and the inevitable ensuing clamour for a change in the electoral system.

Make no mistake: First Past the Post (FPtP from now on) is by no means perfect, but it allows the electorate to send a strong message to the government but simultaneously does not offer fringe parties support, which, in my view, they simply do not warrant. In Britain, it has produced an exceptionally stable conveyor belt of governments since 1945, all of which, with the exception of the February 1974 government (and eventually the October ‘74 and the 1992 governments), had overall majorities. One can argue the merits of these governments’ policies at length, but one surely cannot dispute that each of these governments was the preferred choice of the majority of those who cast their vote; and, in all probability, the choice of the majority of those who chose not to.

With the Lib Dems flying (dangerously) high in the polls, hovering around 30%, all bets are off as far as projections are concerned. Using a Uniform National Swing (UNS) to calculate changes in seats is a complete waste of time. It is exceptionally difficult to know if the Lib Dems’ support is evenly spread: if their new support originated primarily from young voters, there is a good chance that much of it is concentrated in university seats, for example.

If, on the other hand, their vote is increasing across the country, they stand a greater chance of doing more damage to the two main parties’ prospects. Labour will fear that the Lib Dems will make progress in their heartlands – the most recent PoliticsHome/YouGov regional poll indicates that the Lib Dems are polling 35% in the North East of England, up 12 points on the 2005 election. The Tories will fear that the Lib Dems will poll well in suburban and small town seats in the North and East Midlands, which is where the election will be won and lost.

It is simply not good enough for the Tories to think that if there is a Lab->Lib Dem swing, they will ‘slide through the middle’. The Lib Dems may be in third place in many seats which the Tories need to win, but their vote is nowhere near as derisory as it was in the past. A Labour->Lib Dem swing starves the Conservatives of the extra votes they need and may allow the Lib Dems to win swathes of seats, some of them from third place.

So, in the event of a hung parliament, which now looks extremely likely, barring exceptional events, the Lib Dems’ demand for a more proportional voting system may be irresistible. It is my task to explore the outcome of an election conducted under the Single Transferable Vote system: the least popular candidates’ votes are transferred, round by round, to the candidates placed higher. The candidates who claims the most accumulated votes wins.

My forecasting software, UK Elect 6.4, will only allow me to forecast an election using a ‘Two Round’ system. This means that parties polling less than the party placed 3rd will have their vote transferred without any detail. Whilst imperfect, this system does allow us to gain a rough idea of what might happen under STV.

Let’s take the 2005 election first:

Original result (old boundaries):

Lab  35.0%, 355 seats, 55.0% of total seats

Con 32.4%, 198 seats, 30.7% of total seats

LD 22.0%, 62 seats , 9.6% of total seats

Result: Lab Maj 66

Result using STV (old boundaries):

Lab 35.0%, 356 seats, 55.1% of total seats

Con 32.4%, 171 seats, 26.5% of total seats

LD 22.0%, 86 seats, 13.3% of total seats

Result: Lab Maj 68

From these figures we can see that the Lib Dems might have won 26 more seats under STV than FPtP in a 2005 election scenario. But what about the latest voting intention?

Possible 2010 result based on PoliticsHome/YouGov regional data (data entered on a regional basis, not UNS)

Con 33%, 276 seats, 42.5% of total seats

LD 30%, 106 seats, 16.3% of total seats

Lab 28%, 235 seats, 36.2% of total seats

Result: Con Short by 48

Same data using STV (final round):

Con 33%, 196 seats, 30.2% of total seats

LD 30%, 161 seats, 24.8% of total seats

Lab 28%, 260 seats, 40.2% of total seats

Result: Lab Short by 64

As we can see from this STV projection, Labour gain a very large number of seats because it is assumed that most Lib Dem voters would have Labour as a second preference. Whether or this is the case or not in this election remains to be seen – Labour look set to have their number of seats reduced dramatically, perhaps to under 200 under FPtP.

As a Conservative, I am horrified that Labour might continue to govern with the support of the Lib Dems. If Lib Dem voters are voting on an anti-politics basis, I doubt very much that they would choose Labour to be their second choice in an STV scenario. In a different climate, I have no doubt that such a system would make it extremely difficult for the Tories to win outright.

It’s interesting to explore how far ahead the Tories would need to be under an STV system to win outright. Let’s take the figures from the 1992 general election: Con 41.9%, Lab 34.4%, LD 17.8%.


Con 275 (42.3% of total seats)

Lab 282 (43.4% of total seats)

LD 60 (9.2% of total seats)

Result: Lab Short by 42 (1992 FPtP Result: Con Maj 21)

Once again, the Tories are punished in seats where they do not outpoll the sum of Labour and Lib Dem votes.

What about the 1983 general election figures? Con 42.4%, Lab 27.6%, LD 25.4%.


Con 309 (47.5% of total seats)

Lab 212 (32.6% of total seats)

LD 95 (14.6% of total seats)

Result: Con Short by 15 (1983 FPtP Result: Con Maj 144)

As we can see, even with 14.8 point lead, the Tories still cannot win outright under STV, starting from their current base. With figures of Con 44%, Lab 27%, LD 25%, they would scrape home with a majority of 8 with a 17-point lead. As I have shown, Labour can be only 3 points ahead and win an outright majority with STV because of assumed Lib Dem second choice votes.

And what of the Lib Dem position? Well, they certainly do an awful lot better under STV, because they take a lot of Conservative second choice votes in places where Labour can’t win, but this is by no means a perfectly proportional system, and it still under-represents their vote. It does, however, retain the constituency MP, which, if removed under pure PR, would surely not be a popular choice in the current climate.


Anonymous said...

"One can argue the merits of these governments’ policies at length, but one surely cannot dispute that each of these governments was the preferred choice of the majority of those who cast their vote"

Actually, no party has been elected with a majority of the popular vote since 1935, when the Conservates under Stanley Baldwin achieved 55%. Since then, winning parties have tended to achieve results in the low 40%. On this factual basis, I think that one can dispute the statement 'that each of these governments was the preferred choice of the majority of those who cast their vote'.

Perhaps you are conflating winning of the majority of seats in the House of Commons (which is the usual case) with a majority of votes cast?

Andrew James said...

Anon. yes, I see your point. Post-war governments have indeed been elected by fewer than 50% of the popular vote - no government has been elected by an absolute majority in terms of the popular vote since Baldwin. Perhaps I could amend my quote by replacing 'majority' with 'plurality'.

My underlying point remains: postwar FPtP election results have always reflected the electorate's overall wish for continuity or change - continuity in 1959, 1966, 2001 and change in 1979 and 1997. Labour's derisory 35% in 2005 probably comes close to testing this principle.