Picking the Players: our next government

Who will form our next government? © Daily Telegraph

My posts over the two months have focussed on trying to understand likely seat shares for each party following the General Election. A week on Friday, we will know the answer. However, the process of forming a government could take much longer, and we could be living with the ramifications of that for a full five years.

The key to this is the parliamentary maths: it is all important. There is one simple convention: the monarch must appoint as Prime Minister whoever is most likely to be able to command the confidence of a majority MPs in the House of Commons.

That does not necessarily mean whoever is the leader of the party that won the most votes or seats.

There are 650 seats in the House of Commons. For a Prime Minister to command an overall majority -able to outvote everyone else, if united in opposition – the target is 326. 326 seats gives a majority of two, because that is the margin by which it outvotes the 324 in opposition to it. For every additional MP a government party gains over the 325 “tie” line, it decreases their opposition by one, and increases their governing share by one, and therefore increases their majority by two. It is different when a party is short of a majority. When commentators and journalists talk about Party ‘X’ being ‘Y’ seats “short” of a majority they are talking about how many extra MPs would be needed for them to reach the magic 326 number.

In practice, a majority of two is generally considered to be insufficient. That’s because, over the course of a 5 year parliament, some MPs may resign their seats, defect to another party, suffer a party or parliamentary suspension for poor conduct, or even die, resulting in by-elections where the seat is lost to the governing party. In addition, the governing party is likely to have a number of what they would consider more “awkward” MPs. These MPs are more likely to rebel, or abstain, from supporting the government on matters they feel strongly about, making it uncertain if they can pass key measures in their programme.

As an example: poor old John Major won the 1992 General Election with a majority of 21 (336 seats to 315 seats) By the end of the parliament, just prior to the 1997 General Election, he’d lost so many MPs to resignations or by-elections he was without any majority at all.

2010 General Election – Recap

Before considering the 2015 General Election result, it’s worth casting our minds back to the 2010GE result:

UK GENERAL ELECTION 2010	SEATS	CHANGE
Conservative (includes Speaker)	307	+97
Labour	258	-91
Liberal Democrat	57	-5
Democratic Unionist Party	8	-1
Scottish National Party	6	0
Others	14	0

Whisper it, but Labour actually did relatively well. I (along with the betting markets, and virtually everybody else) expected Labour to be down to well under 250 seats. In fact, I expected them to be in the 230s. Conversely, I expected the Liberal Democrats to be on at least 75 seats, and the Conservatives to be in the 290s.

The Conservatives were lucky that with a highly effective targeting strategy, superior resources, good groundwork and, in several places, a convenient LD-Labour vote split that they were able to walk away with 307 seats. Their parliamentary seat shares outperformed the “swing” that the national vote shares (Con – 36.1%, Lab – 29% and Liberal Democrat – 23% and Others 11.9%) implied by about 10-15 seats.

This bonus was absolutely crucial for the coalition negotiations that followed. The Labour and Liberal Democrats could only muster 315 MPs between them. This was greater than the Conservatives on 306 MPs, excluding the Speaker, but was not quite enough to provide a realistic alternative stable government. However, it wasn’t impossible, and that prospect alone was enough to frighten the Conservatives into giving a concession on an AV referendum.

Had the seat shares been just slightly different – for instance, in line with the BBC swingometer (giving 291 Con, 266 Lab and 64 LD) – then a Labour-LD coalition, with an overall majority of 8, would have been a real option. The Conservatives could easily have been shut out of power, or forced into such serious concessions on voting reform by the Liberal Democrats that David Cameron could not get them past his party. He might have been left with no option but to resign.

Looking the other way, had the Conservatives gained just a further extra 9 seats from Labour and the Liberal Democrats, to take them to 315 seats themselves, then a Labour-Liberal Democrat coalition would never have been a go’er. They’d always have been outvoted by the Conservatives, and the Conservatives could either have done a deal with the DUP to run a minority government or give far less generous terms for a Liberal Democrat coalition, which would not have included a referendum on voting reform.

This really does show just how much difference a handful of seats can make to the government we get, and the policies it pursues. And that’s critical to assessing what may happen in GE2015.

In the event, the Conservatives ended up slap bang in the middle. They went into coalition with the Liberal Democrats commanding a total of 363 seats between them, and a government majority of 76 (363-325 * 2 = 76). Upon dissolution of this current parliament, on 30th March 2015, the coalition had  358 MPs, and still commanded a majority of 66.

2015 General Election – Party positions

Almost everyone (correctly) expects a hung parliament this year. However, unlike last time – when there were only three real players, and the only question was how heavily Labour would be defeated – there are now multiple players and no-one knows what the result will be. Minor parties have been busy outlining their post-election negotiation positions. A summary (of my understanding of the current positions, they seem to be tweaked regularly) is below.

Notes:

• Coalition is a formal agreement where the party agrees to officially become part of the government, and take ministerial positions.
• Confidence and supply can be a formal arrangement, but only that the party will support the government in votes of ‘confidence’ and to ‘supply’ the votes needed to pass its finance bills. In effect, it will vote to keep the governing party in office, and to pass its budgets. On everything else, it is free to make up its own mind. It does not take ministerial positions.

	Conservatives		Labour
Option:	Confidence and Supply	Coalition	Confidence and Supply	Coalition
UKIP	Yes	Yes	Yes+	No
DUP	Yes	No	Yes	No
Green	No	No	Yes	Yes**
SDLP	No	No	Yes	No
PC	No	No	Yes	Yes
SNP	No	No	Yes	No
Liberal Democrats*	Yes++	Yes	Yes++	Yes

*Assumes Nick Clegg remains Liberal Democrat leader. He has ruled out coalition deals with SNP or UKIP

**Green coalition with Labour yet to be ruled out, but very unlikely

+UKIP stated they can’t see circumstances in which they would prop up Labour but are yet to formally rule out any arrangement whatsoever.

++Liberal Democrat strong preference for coalition over confidence & supply

Two things leap out from this table. First, the Greens/SNP/Plaid Cymru may well act in concert, all favouring propping up Labour and consulting with one another. And, second, the Conservatives only really have the DUP, the Liberal Democrats, and UKIP to call upon as allies. The last would be considered a highly unreliable ally by the Conservatives, and possibly a divisive one.

I have not considered Respect (George Galloway) or Lady Sylvia Hermon, but both would probably incline towards Labour.

Given the projected numbers within the House of Commons, that gives Labour an advantage.

Finally, it should also be noted that Sinn Fein have consistently refused to take their seats in the House of Commons, on principle, and are highly likely to do so again this time. For all intents and purposes this removes their 5 seats from the House of Commons parliamentary equation. This lowers the target bar from 326 seats to 323 seats ( (650-5)/2 = 322.5, rounded to 323 seats) to give an effective overall majority of one.

2015 General Election – Possible seat ranges

I have already given a prediction for the election in my last post.

However, this already looks out-of-date, particularly given the ever growing strength of the SNP in Scotland. But we are blessed, or cursed depending on your point of view, with a surfeit of data in this election. That allows us to really drill down and understand how things might play out in each constituency.

Using all the data, I am as confident as I possibly can be (90%+) that seat movements will be within the following boundaries:

Seat changes	Min	Max
Conservatives losses to Labour	14	50
Conservative gains from Labour	0	4
Conservative losses to LDs	0	1
Conservative gains from LDs	6	20
Labour gains from LDs	7	12
SNP gains from LDs	7	9
SNP gains from Labour	21	41
SNP gains from Conservative	0	1
UKIP gains from Conservative	1	9
UKIP gains from Labour	0	6
UKIP gains from Liberal Democrats	0	1

Adding in the other minor parties, who I haven’t bothered to list separately, this gives the following “spread” for the bounds of seat ranges for all the parties:

All parties – forecast seat spreads:	Min	Max
Conservative	252	316
Labour	233	301
Liberal Democrats	16	38
UKIP	1	16
SNP	34	58
DUP	8	9
Green	0	1
Plaid Cymru	2	5
Respect	0	1
UUP	0	1
SF	4	5
Alliance	0	1

2015 General Election – Government thresholds 

The government will be either Conservative or Labour led, and so will the PM. My assessment of the formation thresholds for each are below:

Conservative-led governments:

• Stable Conservative minority: 320 seats or more. At this level the Conservatives would be only 3 seats short of an effective overall majority. In practice, this would probably suffice for a 5-year parliament. Almost every single MP in the House of Commons would need to unite to defeat it. All it would need to do on anything remotely that contentious would be to secure the abstention of the DUP. It would be able to implement almost all of its manifesto unhindered. However, I assess this as beyond the reach of the Conservatives, unless Labour have a truly awful night and make virtually no advance against them at all.
• Conservative minority with DUP support: 315 seats or more. At this level they would have to cut a deal (formal or otherwise) with the DUP to vote for some measures. That shouldn’t be a problem. There is considerable overlap between the parties on defence, Europe, immigration, the economy and the Union. Such an arrangement would give the Conservatives, in practice, a waver thin effective overall majority at 323 or 324 seats.  An EU referendum on Conservative terms would easily pass the House of Commons, with no concessions needed to the Liberal Democrats. However, this government would be vulnerable to defections, rebellions and by-election defeats. If Cameron chose this then, as the parliament progressed, it might struggle to get its programme through without additional deals with other parties. Unless UKIP tallied 5 seats or more, I don’t think they’d be a part of any arrangement. And, if they’d done that well, it’s unlikely the Conservatives would be in such a good position in the first place.
• Unstable Conservative minority with ad-hoc support: 310 seats or more. This is the lowest point at which the Conservatives still have, in my opinion, an option of going it alone. A deal would need to be struck with the DUP, and possibly the UKIP MPs too, to get it over the 320 vote line. It would be unstable, likely to become more so as the parliament progressed, and highly vulnerable to ambush. However, it might survive for a year or two purely because other parties (such as the Liberal Democrats) could keep it in office simply by abstaining on the Queen’s Speech and votes of confidence. Key votes would need to be negotiated on a case-by-case basis.
• Stable Second Conservative-Liberal Democrat coalition/supply and confidence: 300-315 seats. If the Conservatives clock between these number of seats, I expect Cameron to try and go for a second Con-LD coalition. The numbers work on a wide range of LD seat outcomes, from 20 (just do’able) to 30+ LD MPs, giving an effective overall majority of 14; however, it only works if Nick Clegg remains as Liberal Democrat leader. Furthermore, the challenge here is making such a coalition practicable. The LDs would be so heavily depleted in numbers, there simply wouldn’t be that many of them left to take ministerial positions and departments. In practice, I think a way would be found, but I think a coalition would really only have legs if the LDs had 25+ MPs in the parliament. Otherwise, it would be Confidence and Supply. Cameron would have to give real concessions to get the EU referendum bill through, the extension to right-to-buy, immigration reforms and an inheritance tax cut. The Lib Dems would want the STV voting system introduced for local elections, votes for 16 year olds, House of Lords reform rekindled, and an amelioration of welfare spending reductions as outlined in their manifesto. But on most other areas, an agreement shouldn’t be too difficult.
• Unstable Conservative-Liberal Democrat coalition/supply and confidence: 294-300 seats. This really is the lowest level at which the Conservatives can be sure of retaining some role in government. At 294 seats, the Liberal Democrats and Conservatives combined should be able to muster to within the 315-325 seat band. They probably need a combined a total of 320 seats to make that work effectively, there will be rebellions and splits, but it’s do’able if the LDs retain around 25 seats. All the caveats of the stable deal above apply. A soft deal with the DUP to passively support, or abstain, would probably be needed too. I suspect some increased funding to Northern Ireland would do the trick.

Labour-led governments: Labour have more options, simply because there are more minor parties that would not vote against them and risk putting the Conservatives in power. There is history here. So, the SNP, Plaid Cymru, the Greens and the SDLP will all prop up Labour. “Independents”, such as George Galloway and Lady Sylvia Hermon, would be unlikely to bring it down either. The DUP will also cut a deal if their numbers are needed, and they see advantage in it. However, all those moving parts also make Labour led governments more multi-faceted, and harder to predict, so I am less certain about composition:

• Labour minority government: 310 seats plus. Labour could go it alone on these figures, with no deals with the SNP or the Liberal Democrats, if they chose to do so. Ad-hoc support from Plaid Cymru, the Greens and SDLP and the DUP would be sufficient to govern.  Labour would be able to implement almost all of its manifesto, with concessions here and there. Once again, I assess this as beyond their reach unless Labour hold up well in Scotland and the Conservatives have a truly awful night.
• Stable Labour-Liberal Democrat coalition/supply and confidence: 285-310 seats. My assessment is that Labour will do everything possible to avoid dealing with the SNP. The two parties detest one another, and Labour know that the SNP will be unreliable and unpredictable partners. If the numbers stack-up so they don’t have to rely on the SNP, they will swallow their pride and turn to the Liberal Democrats. This will be easier if Nick Clegg is no longer leader -and he may step down anyway to facilitate such a deal, if he’s not defeated – but it probably won’t be the deal breaker. Even if a Labour-Liberal Democrat pact is short of an overall majority, it’d probably work on any combined numbers above 310 seats, possibly even slightly less, for the reasons outlined above. Minor parties would be unlikely to overthrow it. Instead, they’d prefer to passively support, or abstain, and extract concessions where they can. But its programme for government would be complex, and possibly unpredictable, as the Liberal Democrats sinewed Labour on fiscal discipline, whilst the minor parties pitched for (probably) more regional financial goodies.
• Unstable Labour minority government with tacit SNP support: 260-285 seats. If Labour drop out in this range, which is actually perfectly possible, then a minority government with tacit SNP support is most likely. Ed Miliband has ruled out a coalition, as has Nicola Sturgeon, and whilst the SNP have stated they’d be happy to strike some sort of formal ‘confidence and supply’ deal, Labour have recently been at pains to emphasise that they wouldn’t touch it. In reality, I suspect if the hard arithmetic of parliamentary reality bites, then they will. I have ruled out the Liberal Democrats of propping up such a deal, on the basis of Nick Clegg’s comments, but (once again) if he is not leader, it’s not impossible. Then again, if the SNP do get 50 seats or more, this arrangement would stand by itself and the Liberal Democrats wouldn’t be necessary.

No-man’s land:

No man’s land arises when no party has any clear mandate to govern and/or has no obvious path to form a stable government. Unfortunately, this is quite likely.

Typical no man’s land seat ranges include:

• Conservatives on between 280 and 290 seats
• Labour on between 250 and 260 seats

One way of thinking about this is to consider what the Conservative/Labour seat tallies would have been in this election if the SNP surge had not occurred.

The gap between the Conservatives and Labour was only 49 seats at GE2010. So, without the SNP surge, and assuming gains from the Liberal Democrats giving neither main party any net advantage, a simple switch of 25 seats from Labour to Conservative would have put Labour just ahead. The SNP is likely to cost Labour (at the very least) 21 seats in Scotland, and that’s if the SNP have a very bad night. It’s actually more likely the SNP cost them the full 41. You’ll see that, if you add those to the Labour seat ranges above, that’d have put Labour and the Conservatives at close to seat parity. Which explains why this is no-man’s land: Scottish ex-Labour voters voting SNP this time are doing so because they expect their SNP MPs to be better at holding a Labour government to account, in Scotland’s interest, than any Scottish Labour MP.

I simply don’t know what the government outcome would be in no man’s land. If you want to investigate yourself, you can play with this excellent FT coalition calculator here. Any Labour/SNP arrangement would be extremely weak, and vulnerable to a Conservative ambush whenever the SNP abstained. Even with all of the SNP votes behind Labour, such a government would be well short of a majority.

Conversely, a Conservative-led government could be brought down at almost any time. Even with Liberal Democrat support, it wouldn’t command much more north than 300 MPs. One possible test is whether Conservative/Lib Dem MPs combined could outvote Labour/SNP combined. But I don’t expect that to wash either, because Plaid Cymru, Greens and the SDLP will work to block the Conservatives out of power. Perhaps a ultra-weak Conservative PM would be left in office – by Labour abstaining on the Queen’s Speech – until Labour elected a new leader. And then brought down.

I’d expect a second election within 12-18 months. Another possibility is a grand coalition, between Labour and the Conservatives, but I don’t expect that to be tried until a second general election has also failed to deliver a decisive outcome.

Betting

For what it’s worth, my personal views on the results are as follows:

• Labour will do very well in metropolitan areas, such as London and Manchester
• The Conservatives will do well in the south-west, against the Liberal Democrats
• The Conservatives will hold up ok in the Midlands battlegrounds
• Labour will pick off several key seats from the Conservatives in the North-West
• The SNP will clean-up Scotland

So..

• The Conservatives won’t drop below 270 seats
• Labour will not clock higher than 290 seats
• The Liberal Democrats will be below 30 seats (but could have a disaster and drop below 20 seats)
• UKIP will be below 6 seats
• SNP will be above 45 seats

That makes my betting choices as follows:

• Hung Parliament – the price for a hung parliament is still 1.12 on Betfair exchange. This is ridiculous, and basically buying free money, with a 10%+ tax-free return on your money inside 10 days. I’ve fully played out this market, but this is still exceptionally good value. It should be 1.03-1.05, now, so, if you have deep pockets, you might want to consider.
• Next government – back Labour Minority at 2.8, or Conservative Minority at 6.6 or Con-LD coalition at 4.9 on Betfair exchange. Only the Liberal Democrats will be forming a formal coalition, and I don’t think Labour will do well enough to make a LD-Labour coalition a runner. Only the Conservatives and Labour can form minority governments. And I don’t think there’s much evidence to suggest anything else will be a winner. Labour minority is the most likely, but please do your own research. I am on all of those options.

Conclusions

I will be watching the exit poll, due to be released at 1opm next Thursday night, very carefully indeed to see where it puts both main parties. It will be our first real clue as to the result and, I suspect, within +/-10 seats for both the main parties nationwide.

I’ll be looking for the Conservatives above 290 seats, or Labour above 285 seats. If we don’t get either of those, we will have something messy:

• Labour must take at least 30 seats from the Conservatives just to be sure of standing still on their 2010GE result of 258 seats
• Labour will take between 15 and 40 seats from the Conservatives. If it I had to call it now, I’d say around 30 seats.
• But the Conservatives cannot afford to drop more than 20 seats to Labour to be sure of staying in government – so David Cameron has a fight on his hands to stay as PM
• Conservatives and Liberal Democrats combined will need to sum to 320 seats for a stable government
• If Labour and the SNP seat totals (combined) are more than 5 seats ahead of the Conservatives just by themselves (say Con at 290 seats stays in government with Lab at 255 + SNP at 55 seats) there will be all sorts of arguments about legitimacy from Scotland. The opposite also applies, vice-versa, if Labour and the SNP form a government, with the Conservatives having a clear majority in England.
• There could be several resignations within 24 hours of the result – Ed Miliband will go quickly if he does worse than his GE2010 seat total and can’t become PM
• If David Cameron doesn’t remain as PM, he will also go quickly
• If we get an unstable government, expect new leaders to be elected soon, and parties put on an election footing by Autumn

I fear the great British public are deeply disillusioned (cynical?) this time, and want everyone to lose, and no-one to win. By a strange process of osmosis, they have a habit of normally getting what they want, so I expect a result similar to the one below:

FINAL Casino Royale FORECAST:UK GENERAL ELECTION 2015	Seats	Change
Conservative	289	-18
Labour	259	+1
Liberal Democrat	24	-33
Democratic Unionist Party	9	+1
Scottish National Party	51	+45
UKIP	3	+3
Others	15	+1

And I very much doubt David Cameron would remain as Prime Minister on those numbers. But we’ll see what happens next week..

That’s it folks, see you on the other side.

Good luck.

The Primary Battleground: Labour v. Conservative

Photo: AP / Eddie Mulholland

There is now less than one month to go until the General Election. However, despite this being the most multifaceted election in living memory, it is worth remembering that the Conservatives and Labour currently hold over 85% of all the seats in the House of Commons. Under almost every conceivable scenario, this will remain the case after 8th May.

Constitutional convention demands that the Prime Minister is appointed by the monarch on the basis of whoever is most likely to be able to command a majority in the House of Commons. Since both the Conservatives and Labour will possess (by far) the greatest numbers of seats in parliament in comparison to all the other parties, one of these two men will be Prime Minister after 8th May. The question is which.

The answer to that question will be determined through the dozens of Labour-Conservative duels that are currently taking place up and down the country. In previous blogposts I have examined the prospects of the Liberal Democrats in England and Wales, and Scotland as a distinct nation, separately.

Today, it is the turn of those key Labour-Conservative marginals in England and Wales: what I have deemed the ‘primary’ battleground. 

The first step in doing the analysis to define the seats that make up the battleground. Using the excellent UK polling report Conservative defence list, which ranks seats by percentage majority, I have produced my own list of 75 key Conservative-Labour marginals.

I have produced this by stripping out Scotland, and any Conservative seat facing a nominal Liberal Democrat challenge. Although the Liberal Democrats do have a shot at Watford, and possibly an outside chance at Oxford West and Abingdon and Montgomeryshire, I do not expect them to pose much of a threat to sitting Conservative MPs. You may disagree, but due to their collapse in the national polls and their need to focus almost all their resources on defending existing seats, I can’t see them building on their 2010 vote here. I have also ignored any Labour targets above their 75th best Conservative prospect. This is on the basis that Labour are simply not doing well enough in the national polls to justify analysis beyond that point, which would represent over a 6% swing from Conservative to Labour.

Therefore, the most marginal seat on my list (no. 1) is North Warwickshire, with a majority of just 54 votes, and the least (no. 75) is Battersea, with a majority of 5,977.

It is worth recalling the conclusions from my two previous posts at this point. Labour started with a ‘Scottish handicap’ on my earliest forecast, which increased the nominal Labour-Conservative seat gap in England and Wales from 48 seats to 75 seats. I then went on to forecast that the Liberal Democrats would surrender 13 seats to the Conservatives, and 10 to Labour, in England and Wales. Therefore, the gap for Labour to close is 78 seats – if they can strip the Conservatives of 39 seats, or more, Labour will be the largest party.

Analysis

I decided to approach this in two ways: (1) through a “bottom-up” forecast (using data from Ashcroft constituency polls, and any local information on the prospects of the candidates; to try and ‘predict’ the result in each individual seat), and, (2) by a “top down” forecast (using predicted shares of national votes on the day, with suitable adjustments for incumbency and regional factors).

For both approaches, I also added a couple of extra data points: whether the seat was a Tory 40:40 target, or not, and whether the candidate was an incumbent MP, or not.

Bottom-up

I was led by the Ashcroft individual constituency polls for the bottom-up analysis. These have been taken in most of these key marginal seats over the last year. They are extremely useful in assessing the underlying level of support for each of the main parties. However, although the methodology is sound, they need to be taken with a pinch of salt. They are only snapshots at a point in time, as Ashcroft himself concedes, and the true picture may since have varied wildly with the national and local campaign efforts. Further, it is much harder to poll a seat of 55,000-70,000 electors accurately with a sample size of 1,000 than it is a national electorate in the tens of millions. The margin of error is much higher, and the true picture at that point in time could have been anywhere within +/- 5% of the numbers reported.

Nevertheless, it’s the best data we have to go on. What I’ve done in terms of analysis is very crude. I’ve annotated, next to each Ashcroft poll, the month in which it was taken. Using the UK polling report ‘poll of polls’ data I’ve then averaged the Labour lead for that month next to it. The national polls have moved over the last 9 months, they are now broadly level rather than reflecting a 3-4% Labour lead, so doing this allows me to adjust for the predicted national effect on 7th May. I’ve then reviewed each seat in turn with an ‘instinctive’ pass. Generally, I’ve discounted any seat with an Ashcroft poll Labour lead of >10% as a lost cause and given a Tory hold where the lead is <5%.

Model 1 – Casino Royale adjusted Ashcroft constituency polls (England and Wales): Conservatives lose 33 seats, Labour pick up 32 seats and UKIP 1 seat.

I think these figures look plausible. I have 11 extra Conservative “holds”, and correspondingly fewer Labour gains, over what Ashcroft found. However, two of these seats he did find to be statistical ties and the other nine were Labour gains by less than 5%.  The seats are: Stockton South, Morecambe and Lonsdale, Brighton Kemptown, Hove, Halesowen & Rowley Regis, Pudsey, Nuneaton, South Swindon, Croydon Central, Norwich North and High Peak.

In general, I’ve found them to be Tory holds due to opinion poll movements, and strong local candidates and campaigns. However, even on these figures, the Conservatives would be down to 285 seats after the election, and almost certainly out of power.

There are a possible further five Conservative holds using this method: Amber Valley, Lincoln, Northampton North, Keighley and Wirral West. However, I did not count these due to strong Labour challenges in these seats and fear of skewing the results.

Top-down

I started this by forecasting forward my predicted GB vote shares on polling day. This is based on a 4% Conservative lead GB-wide. Although the average of current polls show a statistical tie, there have been a few telephone polls in recent weeks showing the Conservatives on 36%, and one online YouGov poll showing them on 37%. In recent electoral history, the poll showing the Conservatives on the best vote share has tended to be the most accurate. In addition, there has been a corresponding pattern of Conservative governments doing better than Labour oppositions in the final weeks before election day. Peter Kellner at Yougov is in a similar place with these assumptions.

So I expect further relative movement to the Conservatives by 7th May, and a slight differential turnout in their favour, sufficient to deliver the following vote result on the day:

GB ONLY	2015 %	2010 change +/-%
Conservative	36.27%	-0.6%
Labour	32.28%	2.6%
Liberal Democrat	8.59%	-15.0%
Scottish National Party	3.78%	2.1%
Plaid Cymru	0.56%	0.0%
Green	3.99%	3.0%
UK Independence Party	13.31%	10.2%
Others	1.22%	-1.2%

By my reckoning, this would give numbers for England and Wales as follows:

ENGLAND ONLY	2015%	2010 change +/-%
Conservative	39.00%	-0.6%
Labour	32.85%	4.8%
Liberal Democrat	9.10%	-15.1%
Green	4.00%	3.0%
UK Independence Party	13.90%	10.4%
Others	1.15%	-2.5%

WALES ONLY	2015 %	2010 change +/-%
Labour	36.50%	0.3%
Conservative	24.50%	-1.6%
Liberal Democrat	6.00%	-14.1%
Plaid Cymru	11.00%	-0.3%
UK Independence Party	18.00%	15.6%
Green	2.20%	1.8%
Others	1.8%	-1.6%

This then needs to be applied to seats. What I’ve done is to apply some standard splits to the 2010 general election voteshares for each party in each seat. I’ve then “tested” the results by backfitting them to see if they deliver, in aggregate, a similar national voteshare to those voteshares predicted above (which they broadly do, but UKIP needs a boost from new voters to make up the correct national total)

The standard vote splits (applied to the 2010 constituency totals) are below:

LD Standard Split
LD	34.00%
Tory	16.00%
Labour	31.00%
UKIP	8.00%
Green	9.00%
Others	2.00%

Labour Standard Split
LD	1.00%
Tory	2.00%
Labour	87.00%
UKIP	5.00%
Green	3.00%
Others	2.00%
	100.00%

Con Standard Split
LD	2.00%
Tory	86.00%
Labour	1.00%
UKIP	9.00%
Green	1.00%
Others	1.00%
	100.00%

BNP 2010 Standard Split
LD	1.00%
Tory	5.00%
Labour	7.00%
UKIP	70.00%
Green	1.00%
Others	1.00%
BNP	5.00%
NO VOTE	10.00%
	100.00%

UKIP 2010 Standard Split
LD	1.00%
Tory	9.00%
Labour	3.00%
UKIP	85.00%
Green	1.00%
Others	1.00%
	100.00%

When I do this, and apply it to the 75 key marginals, I get the following:

Model 2a – Casino Royale predicted voteshares (England and Wales): Conservatives lose 25 seats; Labour gain 25 seats.

Clearly, this method is a bit simplistic. I am basically applying a form of uniform national swing (UNS). This isn’t likely to be very accurate – as it takes no account of regional or local factors, or any incumbency bonus – but it does give a broad indication of how things might stack up.

Let’s refine it a bit. For my second attempt on the top-down, I have allocated a 1% first-time incumbency bonus to any Conservative MP re-standing for election in 2015 who was first elected in 2010. I haven’t done anything fancy here, I’ve simply taken the total turnout for the seat and given 1% of that (extra) to the Conservative candidate. It might be a bit less, it might be a bit more, but broadly it gifts them between an extra 400-600 votes, which seems about right for a new MP who’s been working the seat hard for 5 years, helping people out along the way. I now get the following:

Model 2b – Casino Royale predicted voteshares with incumbency (England and Wales): Conservatives lose 14 seats, and Labour pick up 14 seats.

However, that’s still not quite right. We know UKIP have regions of concentrated strength (in the east of England, and around Kent, Essex and near some northern English cities in particular) and Labour in London, where there appears to be a much greater Conservative -Labour swing, and where 45% of all Labour party members are based. We also know that the Lib Dems are likely to suffer a lower vote share where they are not in contention this time, as they get squeezed by both main parties.

So, I’ve done a final model adjusting for all these factors. My figures are below:

Model 2c – Casino Royale predicted voteshares with incumbency, LD adjustment, UKIP regional adjustment and London adjustment (England and Wales): Conservatives lose 23 seats, Labour pick up 23 seats, and UKIP pick-up 1 seat.

You can see a detailed breakdown here.

I’m still not fully happy with this final model. Some results just don’t look right, such as Cannock Chase staying Conservative, Carlisle looking very close, and Stockton South going Labour. But I’m not going to manipulate the model to suit my intuition. Even if I spent a disproportionate amount of time on still yet further analysis, I don’t see how I could make a forecast (any forecast) that’s more accurate. This election is so close, and the dynamics so complex, it might simply come down to turnout on the day.

However, 23 seats is more-or-less in between my 14 seat and 32 seat book-ends, so I’m going with it for the national seat forecast. May God have mercy on my soul.

Adding this to my overall forecasts so far (with some adjustments for Plaid and UKIP, which I haven’t gone into on this thread) gives this grand UK total forecast for 7th May 2015:

UK WIDE 2015
Party	2015 Seats	Gains	Losses	2010 net change +/-
Conservative	294	15	28	-13
Labour	261	34	31	3
Liberal Democrat	25	0	32	-32
Democratic Unionist Party	9	1	0	1
Scottish National Party	40	34	0	34
Sinn Fein	5	0	0	0
Plaid Cymru	4	1	0	1
Social Democratic & Labour Party	3	0	0	0
Green	1	0	0	0
Alliance Party	0	0	1	-1
UK Independence Party	6	6	0	6
Respect-Unity Coalition	1	1	0	1
Others	1	1	1	0

Notwithstanding the slightly tentative forecast, we can make some broad conclusions:

• Whichever way I cut it, Labour look on course to pick up between 15 and 35 seats from the Tories
• Incumbency will be pretty crucial for determining if sitting Tory MPs hold on, or not
• The collapse of the 2010 GE Liberal Democrat vote will break in Labour’s favour, and cost the Conservatives seats, but not as many as one might think
• UKIP look like they’ll be squeezed in these main battles, except in areas where they are naturally strong
• These forecast seat gains are short of what’s needed for Labour to become the largest party
• But David Cameron will really struggle to remain Prime Minister on these numbers anyway – personally, I think he has to keep his direct Labour losses to 20 or less

In terms of betting opportunities, there’s fairly slim pickings here in terms of value. Given the unpredictable nature of precisely which seats may fall in the top 30 key marginals, I’m not betting on too much. Once again, it comes down to your confidence as a punter. However, my own personal picks (all on a successful Conservative defence) are below:

• Stockton South – 11/10 on the Conservatives with Corals. James Wharton is well regarded in his constituency and has been working hard over the past five years. I expect him to do better with his incumbency bonus than almost any other Conservative MP in an ultra-marginal seat. Survation, conducting a poll for Unite, even had him ahead in November. I think it’ll be close, but he’ll cling on.
• Blackpool North and Cleveleys – 5/6 on the Conservatives with William Hill. All of my models have the Tories holding this seat. This looks this value to me.
• Morecambe and Lonsdale – 11/8 on the Conservatives with Ladbrokes. I understand the local Conservative MP is well-known and fairly well respected. Almost all of my models show a hold.
• Hove – 2/1 on the Conservatives with Betfair Sportsbook. This seems a bit long. You are ‘gambling’ if you back this, but I expect the result to be very close and for the Conservatives to hold. Just.
• Halesowen & Rowley Regis – 6/5 on the Conservatives with Corals. This looks like a hold to me. I’m on.
• Bristol North West – 8/15 on the Conservatives with Corals. As far as I know, there has been no constituency poll here. The Liberal Democrats are in 2nd place, with a Labour a strong 3rd. But it should be a relatively easy Conservative hold.
• Cleethorpes – 4/6 on the Conservatives with Corals. A long way down the target list, but this should be a comfortable Conservative hold. I’m on.

Health warning: please do not rely on my model for your own betting, and please do your own research.

Thanks for reading.

In my next post, I’ll examine the possible compositions of the next government, and the various thresholds each party will need to achieve in order to form it.