Eliminating cross-county constituencies
Mar 19, 2017 16:43:12 GMT
YL, Pete Whitehead, and 2 more like this
Post by East Anglian Lefty on Mar 19, 2017 16:43:12 GMT
Earlier this week, I suggested that if you eliminated cross-county constituencies, most of the least popular constituencies offered by the Commission would rapidly fall by the wayside. This thread is an attempt to either demonstrate that, or to prove I had no idea what I was talking about.
At present the 5% rule and the practice of assigning seats to European regions necessitates the crossing of county boundaries. But that's a new system (and one that will rapidly make less and less sense, as we'll be leaving Europe and what little regional planning we do do these days doesn't generally use those administrative geographies.) Prior to the 6th Review, seats were assigned to preserved counties (broadly speaking, former metropolitan counties on the one hand, and shire counties and unitaries created from them on the other.) So what if we went back to that system, but with a little more concern for equalising electorates?
My methodology is as follows:
1. I assign an integer number of seats to each preserved county by simple rounding (you could use the harmonic mean, or St-Lague, or D'Hondt, but the former creates additional seats and the latter two don't change the outcome much but take more effort to work out.)
2. I work out the average size of a constituency in the preserved county, and make sure all my proposed seats are within 5% of that.
3. I also make sure they're within 10% of the national average.
4. If this is impossible (it might be with some small counties, though I'm going to follow the practice of earlier reviews in treating Rutland as Leicestershire and doing Herefordshire and Worcestershire together), then and only then can you combine two neighbouring counties.
I'm taking the North-East as my test case, because it's a small region I've frequently worked with and because the geographical constraints minimise the temptation to gerrymander.
Northumberland
Northumberland has an electorate of 232,448, which entitles it to 3.11 constituencies. That's within the 5% rule and it is possible to assign three seats wholly in the county under the present restrictions, but I figured I might as well take advantage of a little extra flexibility to draw neater seats:
Berwick & Ashington (75,406)
Hexham & Morpeth (76,906)
South East Northumberland (80,136)
The allowable seat range under my criteria would be 73,609 to 81,357, so there's no great difficulty involved.
Naturally the first seat won't be to everybody's taste and I can readily see why (though I'm increasingly happy with it, because regardless of the social differences it's much easier to get from Berwick to Ashington staying in the seat than it is to get from Cramlington to Hexham.)
But if you don't like it, you can always shift round Cramlington, Ashington and Morpeth for essentially the same result.
Tyne & Wear
Tyne & Wear has an electorate of 792,738, good for 10.60 constituencies. Again, you could do this under the current rules (though it'd look horrible), but if you use my criteria then the acceptable range is 68,464 to 75,670.
Within that you can just about assign two seats to North Tyneside, though you need to move three wards in each direction to get the two existing seats to work and look vaguely reasonable.
Sunderland could mathematically have three whole seats, but it's tight enough that I suspect it's not possible in practice and if it is it'll look abominable.
Gateshead is exactly the right size for two seats, but Newcastle is too small for three and unless somebody starts building a new bridge quickly you can't combine it with South Tyneside.
Taking all that into account, here's what I came up with:
North Tyneside (75,507) - actually the successor to Tynemouth, but this is the least terrible name I could think of
Tynemouth & Wallsend (75,538)
Newcastle Upon Tyne Central (68,944)
Newcastle Upon Tyne North (74,496)
Newcastle Upon Tyne West & Blaydon (70,829)
Gateshead (68,913)
Jarrow & Felling (69,909)
South Shields (70,109)
Washington & Sunderland West (71,848)
Sunderland Central (71,232) - unchanged
Houghton & Sunderland South (75,413)
Despite the name (it doesn't feel appropriate to call a seat containing Elswick Newcastle East), Newcastle Central is the abolished seat. In reality, Chi Onwurah would be eligible for North and Catherine McKinnell would be eligible for North and West. If you prefer, you could do a Tyne Bridge seat, in which case Gateshead would probably be the seat to nominally disappear.
County Durham
County Durham including Darlington has an electorate of 452,644, which entitles it to 6.05 seats. One of those can cover Darlington and the other five the current County Durham UA. The range here is from 79,213 to 71,669, so fairly close to the nationwide range.
Darlington more or less draws itself, but then you have to decide which Durham seat to eliminate. The natural candidate is Sedgefield, as it's already small before you consider the 10,000 electors it loses to Darlington. Another 12,000 electors then get Easington up to size. The problem is dividing up the rest without splitting Spennymoor three ways or tearing City of Durham away from its hinterland. This is the best I could manage without ending up with Barnard Castle in the same seat as Consett. I'm perfectly willing to believe there's a better solution:
Darlington (74,929) - co-extensive with the UA.
Bishop Auckland & Aycliffe (78,467)
North West Durham (73,830)
North Durham (77,303)
City of Durham (74,529) - I'd feel a lot happier about this if I'd managed to get Deerness ward in as well.
Easington (73,586)
Sedgefield would be the eliminated seat, and Phil Wilson wouldn't have an automatic claim to any other seat.
Cleveland
Yes, I know Cleveland isn't a county any more. But it was still used as a preserved county by the 5th Review, so I'm going to keep using it.
The area has an electorate of 396566, which entitles it to 5.30 constituencies. Round that down to five, and you've got a range from 82,446 (because no seat can be more than 110% of the nationwide quota) to 75,348.
There are 165,794 electors north of the Tees, which is too much for two constituencies, so the river has to be crossed. None of the authorities can stand alone.
Hartlepool stays whole, but it can't get up to the local quota without venturing into Billingham (or Stockton, I guess). However, you can then fit the rest of Stockton and Billingham into a single seat. Eaglescliffe goes with Thornaby and you then have to decide where in Middlesborough you want to find 30,000 electors. 20,000 electors from Redcar & Cleveland then end up in a Middlesborough-based seat.
This would be my best attempt at least-change, though in practice it'd look a lot neater if you split the Greater Eston area:
Hartlepool (78,310)
Stockton-on-Tees (79,572)
Thornaby & Middlesborough West (81,281)
Middlesborough East & Guisborough (78,977)
Redcar (78,426)
There are, I will confess, a fair amount of unlovely seats there - Teesside is, if anything, even worse this way. But there are, I'd argue, fewer utterly daft constituencies and this approach does make it easier to consider minimum change criteria, which reduces the likelihood of monstrosities such as you get every time the BCE decides to put Barnard Castle somewhere daft. However, I suspect the benefits are more obvious when we're dealing solely with shire counties, so that's where I'll turn to in my next few attempts.
At present the 5% rule and the practice of assigning seats to European regions necessitates the crossing of county boundaries. But that's a new system (and one that will rapidly make less and less sense, as we'll be leaving Europe and what little regional planning we do do these days doesn't generally use those administrative geographies.) Prior to the 6th Review, seats were assigned to preserved counties (broadly speaking, former metropolitan counties on the one hand, and shire counties and unitaries created from them on the other.) So what if we went back to that system, but with a little more concern for equalising electorates?
My methodology is as follows:
1. I assign an integer number of seats to each preserved county by simple rounding (you could use the harmonic mean, or St-Lague, or D'Hondt, but the former creates additional seats and the latter two don't change the outcome much but take more effort to work out.)
2. I work out the average size of a constituency in the preserved county, and make sure all my proposed seats are within 5% of that.
3. I also make sure they're within 10% of the national average.
4. If this is impossible (it might be with some small counties, though I'm going to follow the practice of earlier reviews in treating Rutland as Leicestershire and doing Herefordshire and Worcestershire together), then and only then can you combine two neighbouring counties.
I'm taking the North-East as my test case, because it's a small region I've frequently worked with and because the geographical constraints minimise the temptation to gerrymander.
Northumberland
Northumberland has an electorate of 232,448, which entitles it to 3.11 constituencies. That's within the 5% rule and it is possible to assign three seats wholly in the county under the present restrictions, but I figured I might as well take advantage of a little extra flexibility to draw neater seats:
Berwick & Ashington (75,406)
Hexham & Morpeth (76,906)
South East Northumberland (80,136)
The allowable seat range under my criteria would be 73,609 to 81,357, so there's no great difficulty involved.
Naturally the first seat won't be to everybody's taste and I can readily see why (though I'm increasingly happy with it, because regardless of the social differences it's much easier to get from Berwick to Ashington staying in the seat than it is to get from Cramlington to Hexham.)
But if you don't like it, you can always shift round Cramlington, Ashington and Morpeth for essentially the same result.
Tyne & Wear
Tyne & Wear has an electorate of 792,738, good for 10.60 constituencies. Again, you could do this under the current rules (though it'd look horrible), but if you use my criteria then the acceptable range is 68,464 to 75,670.
Within that you can just about assign two seats to North Tyneside, though you need to move three wards in each direction to get the two existing seats to work and look vaguely reasonable.
Sunderland could mathematically have three whole seats, but it's tight enough that I suspect it's not possible in practice and if it is it'll look abominable.
Gateshead is exactly the right size for two seats, but Newcastle is too small for three and unless somebody starts building a new bridge quickly you can't combine it with South Tyneside.
Taking all that into account, here's what I came up with:
North Tyneside (75,507) - actually the successor to Tynemouth, but this is the least terrible name I could think of
Tynemouth & Wallsend (75,538)
Newcastle Upon Tyne Central (68,944)
Newcastle Upon Tyne North (74,496)
Newcastle Upon Tyne West & Blaydon (70,829)
Gateshead (68,913)
Jarrow & Felling (69,909)
South Shields (70,109)
Washington & Sunderland West (71,848)
Sunderland Central (71,232) - unchanged
Houghton & Sunderland South (75,413)
Despite the name (it doesn't feel appropriate to call a seat containing Elswick Newcastle East), Newcastle Central is the abolished seat. In reality, Chi Onwurah would be eligible for North and Catherine McKinnell would be eligible for North and West. If you prefer, you could do a Tyne Bridge seat, in which case Gateshead would probably be the seat to nominally disappear.
County Durham
County Durham including Darlington has an electorate of 452,644, which entitles it to 6.05 seats. One of those can cover Darlington and the other five the current County Durham UA. The range here is from 79,213 to 71,669, so fairly close to the nationwide range.
Darlington more or less draws itself, but then you have to decide which Durham seat to eliminate. The natural candidate is Sedgefield, as it's already small before you consider the 10,000 electors it loses to Darlington. Another 12,000 electors then get Easington up to size. The problem is dividing up the rest without splitting Spennymoor three ways or tearing City of Durham away from its hinterland. This is the best I could manage without ending up with Barnard Castle in the same seat as Consett. I'm perfectly willing to believe there's a better solution:
Darlington (74,929) - co-extensive with the UA.
Bishop Auckland & Aycliffe (78,467)
North West Durham (73,830)
North Durham (77,303)
City of Durham (74,529) - I'd feel a lot happier about this if I'd managed to get Deerness ward in as well.
Easington (73,586)
Sedgefield would be the eliminated seat, and Phil Wilson wouldn't have an automatic claim to any other seat.
Cleveland
Yes, I know Cleveland isn't a county any more. But it was still used as a preserved county by the 5th Review, so I'm going to keep using it.
The area has an electorate of 396566, which entitles it to 5.30 constituencies. Round that down to five, and you've got a range from 82,446 (because no seat can be more than 110% of the nationwide quota) to 75,348.
There are 165,794 electors north of the Tees, which is too much for two constituencies, so the river has to be crossed. None of the authorities can stand alone.
Hartlepool stays whole, but it can't get up to the local quota without venturing into Billingham (or Stockton, I guess). However, you can then fit the rest of Stockton and Billingham into a single seat. Eaglescliffe goes with Thornaby and you then have to decide where in Middlesborough you want to find 30,000 electors. 20,000 electors from Redcar & Cleveland then end up in a Middlesborough-based seat.
This would be my best attempt at least-change, though in practice it'd look a lot neater if you split the Greater Eston area:
Hartlepool (78,310)
Stockton-on-Tees (79,572)
Thornaby & Middlesborough West (81,281)
Middlesborough East & Guisborough (78,977)
Redcar (78,426)
There are, I will confess, a fair amount of unlovely seats there - Teesside is, if anything, even worse this way. But there are, I'd argue, fewer utterly daft constituencies and this approach does make it easier to consider minimum change criteria, which reduces the likelihood of monstrosities such as you get every time the BCE decides to put Barnard Castle somewhere daft. However, I suspect the benefits are more obvious when we're dealing solely with shire counties, so that's where I'll turn to in my next few attempts.