Time to enact the boundary changes for a more level playing field at next General Election

An outcome of the expenses scandal in 2009 both the Conservatives and Liberal Democrats had pledged in their 2010 manifestos to reduce the number of constituencies. The outcome was Schedule 2 of the Parliamentary Voting System and Constituencies Act 2011, with the reduction from 650 to 600, along with a review to achieve more equal constituency sizes. The Boundary Comission published its initial recommendations in 2016, then, following extensive consultations, published revised recommendations in 2018. Given the historically low esteem with which the House of Commons is currently held, and the deadlock on the Brexit issue, perhaps there should be a coming together in the House of Commons to enact these changes to show that they can work together to produce a more level playing field in general elections. The main block on such changes is that greater fairness will lead to shifts in the makeup of Parliament.

2018 Review Changes by Region

The Boundary Commissions 2018 recommendations, published on 05/09/18 are spread across sites for Wales, Scotland, Northern Ireland and England. Using the results of the 2017 General Election I constructed a table to show the proposed constituency changes by region.

Figure 1 : Proposed Constituency Changes by Region. Additionally is an apportionment of the seats won by Conservative and Labour at GE 2017 on a national and regional basis. As the decrease smallest decreases in seat numbers tend to be in the regions where the Conservatives are stronger, GE 2017 fought on the new boundaries may have given the Conservatives an majority.

All regions will lose seats in the proposed boundary changes, but the proportionate changes vary considerably. Wales will lose over a quarter of its seats, going from 40 to 29 seats. At the other extreme, the South East, which is already the biggest region, will lose just one seat. The result of implementing the changes would make the average number per constituency across regions far more equal than currently. It should be noted that both Scotland and Wales will still have broadly the same representation per capita in Westminster as England, despite having devolved parliaments.

More detailed analysis from Electoral Calculus

The above gives a general high level impression. The Electoral Calculus website provides a far more detailed analysis through its “Make your Prediction” tool. I first plugged in the results of the 2017 election for the major parties based on the 2017 boundaries at regional level.

Figure 2: Electoral Calculus prediction for 2017 General Election based on 2017 boundaries. The Speaker is included in the Conservative Party Numbers

This fairly accurately produces the 2017 result at the top level with a hung parliament. The predictor slightly overstates the Labour seats, fails to predict any seats for Plaid Cymru and understates the Lib-Dem seats for the same reason. That is the support for the parties is more concentrated than the program allows. Plugging in the national results produces the prediction of a small Conservative majority, much for the same reason. The program does not properly allow for concentrated regional or local support.

More interesting is the prediction based on the 2018 boundaries with 600 seats.

Fig 3 : Electoral Calculus’s estimate of the 2017 General Election result against an estimate based on regional party vote share.

The detailed Electoral Calculus analysis predicts the Conservatives would have been just one short of a majority if the 2017 General Election has been fought on the 2018 600 seat boundaries, or two short if the Speaker is excluded. The national predictor shows a majority of 12. These figures are very similar to my rough estimates above, with a slight Conservative bias in top level forecasts. Given that both the main parties have seen losses of MPs, on these figures neither would see a large net loss of sitting MPs, provided that both Labour and Conservatives were in a similar position in the polls to 2017. However, this is not the case. Electoral Calculus, based on opinion polls from 03 Sep 2019 to 27 Sep 2019, predict a Conservative majority of 12. Plugging in the national vote shares into the predictor, I get a Conservative majority of 26. Using the 2018 Boundaries, with 600 seats the majority increases to 54. Conservative seat share rises from 52.5% to 54.0%, whilst Labours seat share falls from 32.6% to 30.8%. A more level playing field works in the Conservatives favour. However, the bias in the predictor means that the difference is likely smaller.

Fig 4 : Based on opinion polls from 03 Sep 2019 to 27 Sep 2019, Electoral Calculus national GE predictor based on current 650 seat parliameny and 2018 constituency boundaries

However, voting for fairer boundaries should not be based on immediate polling. One would hope that HM’s Official Opposition would have ambitions of winning and election in the future. What would be the impact of swapping the poll positions of Conservatives and Labour? Figure 5 does just that.

Fig 5 : Same assumptions as fig 4, with the exception that polling positions of Conservatives and Labour are reversed.

If the polling positions of Conservatives and Labour were reversed then the Labour Party would be in a similar position to under the existing system and under the 2018 constituency boundaries with 600 seats. This is however a likely scenario under the current circumstances. Allowing for a Conservative bias in the Electoral Calculus estimates, Labour would likely obtain an overall majority. If Conservative votes fell away, they would mostly go to the Brexit Party, whilst gains for Labour would come from the Liberal Democrats. In the final figure I also assume that the Brexit Party are up 5% and the Lib Dems down 5%.

Fig 6 : Same assumptions as fig 5, with the exception that Brexit Party are up 5% and Lib Dems down 5%.

This marginally improves Labour’s position at the expense of the Lib Dems, sufficient to gain a slim overall majority. Again, allowing for the Conservation bias in the figures, the majority would be less marginal.

The Labour party bias under the current boundaries

Although the Electoral Calculus figures have a slight Conservative bias, the Boundary Comission changes will favour the Conservatives over Labour. Figure 1 indicates part of the issue. In Wales, the North West and the North East are regions have both larger than average falls in the number of seats and where Labour have a clear majority of the seats. Conversely in the Eastern, East Midlands, South West and South East Regions have both much smaller than average falls in seat numbers and where the Conservatives have a clear majority of the seats. The current constituency boundaries have a Labour Party bias that will be rectified. I have created a couple of charts from the General Election 2017 results, one of which amplifies the current Labour bias in the voting figures.

Fig 7 : From the 2017 General Election results, bands of the percentage of valid votes gained by each party. The upper chart is the count of seats in each band. The lower chart shows the average constituency size.

In 2017, the Conservatives achieved at least 50% of the vote in 243 seats. For Labour it was 222 seats. The Conservatives gained 40-50% of the vote in 135 seats, winning 71. Labour gained 40-50% of the vote in 89 seats, winning 35. The problem for Labour is that their vote is more highly concentrated than the Conservatives. Thus they need a higher share of the national poll, on a uniform swing, to gain a parlimentary majority than the Conservatives under current boindaries. Put another way, on average the Conservatives gained 55.46% of the valid vote in the constituencies they won, whereas Labour gained 59.33%. Of course the greater spread of votes across constituencies works in the Conservatives favour in winning elections, but against them relatively if party support drops below 25%. Conversely the current boundaries work in Labour’s favour in the event of a poll collapse, but it is mostly due to their concentrated support.

The lower chart illustrates the problem for Labour from the boundary changes. Whereas the average constituency size where the Conservatives recieved > 50% of the vote in 2017 is 75,400, for Labour it is 70,500. Under the 600 seat Parliament, whereas the Conservatives would retain 243 such seats, Labour would expect to only recieve 208 seats, a loss of 14. For the seats gained with less than 50% of the vote the Conservatives would expect to go from 74 to 71, whilst Labour from 40 to 38. Thus in a 600 seat parliament the Conservatives would expect to have 314 seats, as against 317 seats in a 650 seat Parliament. Labour would go from 40.3% of the seats to 41.0% on the same rough calculation.

Finally, there is another paradox. Although, on average, Labour constituencies have a smaller electorate than Conservative ones, of the 31 seats with an electorate over 85,000, 14 are Labour. There are three Labour seats with over 90,000 voters – Bristol West, West Ham and Manchester Central. It is not an even picture across the country.

Concluding Remarks

It is now eight years since legislation was passed to reduce the number of constituencies from 650 to 600, along with making them more equal in size. Had the 2017 election been faught on this more level playing field it is most likely that the Conservatives would have been returned with a small majority rather than losing that majority. The impact of enacting the changes would be to counter the relative discrimination that much of the South and East of England has in general elections due to having larger average constituencies. Most of all it would be of benefit to those in the 31 constituencies with over 85,000 voters, listed below.

Fig 8 : The gemeral election 2017 results for the 31 constituencies with electorate of over 85,000.

Kevin Marshall