According to a number of commentators, the result of the 2010 general election heralded a new era in UK politics (Curtice Reference Curtice2010). For three decades nearly every election produced not only a clear winner – albeit based on only a minority of all of the votes cast – but also a single-party government with a workable Parliamentary majority (1992 became an exception because, although Major had an initial majority of 21, splits within his party made governing increasingly difficult). The 2010 result, which led to the formation of the country’s first coalition government since 1945 involving the Conservative and Liberal Democrat parties, was interpreted as the first of a probable new sequence which would either produce no majority party in the House of Commons or one party would emerge with a small majority that could not be sustained over a full five-year Parliament. Coalition government could become the norm, it was argued (see, for example, Huhne Reference Huhne2014); McLean (Reference McLean2012: 20), for example, concluded that whether or not Britons love coalitions – still a subject of debate – nevertheless in the foreseeable future (he was writing in 2011) ‘the UK is likely to have to make the choice between coalition government and minority government quite often, irrespective of its electoral system’. Furthermore, there could be much uncertainty as governments fell and operation of the Fixed Term Parliaments Act, 2011, made it unclear whether new coalitions would be formed mid-cycle, or minority administrations would be created, or snap general elections would be needed. In an era of economic volatility, such uncertainty would be unwelcome.
This is not the first time such a shift has been proclaimed – although in this case it is based on the ‘facts’ of a general election result rather than the ‘maybes’ of opinion poll data in the months preceding a contest. There were arguments in 1981–2, for example, that the next election (to be held by May 1984) could result in a hung parliament (see Butler Reference Butler1983): the outcome was a Conservative landslide! The expectation was not fulfilled then because although the Alliance got a substantial share of the votes cast – almost as many as Labour – most of them were wasted because of an inefficient geography to their distribution; Labour won nine times as many seats – and formed a substantial opposition – because of deep strengths in many of its heartlands. (Those strengths were undermined in subsequent years in many areas by the deindustrialisation of much of northern Britain under Thatcherism, the decimation of the coalfields, the neutering of the trades unions, and the selling-off of much of the council house stock – delivering another Conservative landslide four years later and stimulating changes initiated by Kinnock, carried forward by Smith and brought to their conclusion by the emergence of New Labour under Blair.)
A year before the scheduled 2015 general election there were many uncertainties – some pointed to by the opinion polls – which made the outcome then difficult to predict. Would the Conservative share of the vote at least equal that of 2010, even increase if economic recovery continued apace, helping it to maintain its hold on the 40 marginal seats it feared losing and perhaps enable it to win a few of its targets, and be able to form a government on its own – albeit with only a small majority? Would the Liberal Democrat vote collapse, perhaps enabling the Conservatives to win some of that party’s vulnerable seats and/or helping Labour advance elsewhere? What share of the vote would the UK Independence Party (UKIP) get, compared to its performance at the 2014 European Parliament elections – and if it was at all substantial, which other parties would be hurt most? (Kellner Reference Kellner2014). And finally, whatever the outcome of the 2014 independence referendum, how would the Scots vote six months later? Even if they voted for independence, because the scheduled date for secession was not until March 2016 the Scots electorate would have a vote at the 2015 UK general election. Because Scotland has returned a clear majority of Labour MPs at recent UK general elections, the outcome of the 2015 contest could depend very substantially on how many are elected then (the alternative would presumably be more Scottish Nationalist Party (SNP) MPs – and would they enter a coalition arrangement with any of the three ‘parties of the union’? – because enhanced Conservative and Liberal Democrat support seemed very unlikely); and if Scotland became independent a year later the UK government could then fall and, given the requirements of the Fixed Term Parliaments Act, 2011, a premature general election might be necessary.Footnote 1
All these uncertainties have their own geographies – where voters switch support, from and to whom, is crucial to the 2015 election outcome, and creating such geographies is central to the parties’ campaigning strategies. Rather than concentrating on those unknown geographies, however, this article focuses on two other aspects of the UK’s (in effect, Great Britain’s) electoral geography – ‘malapportionment and gerrymandering, UK style’ – and party political attempts to use them to influence the outcome of the 2015 contest.
Malapportionment and gerrymandering are normally associated with US electoral politics; they are cartographic abuses whereby one party – in charge of the redistricting (or redistribution) procedure in a State – seeks to advance its electoral cause at its opponents’ expense. Much research has demonstrated the extent to which these strategies have been deployed there and their substantial electoral impact. They are not believed to be part of the British electoral scene, however, because since 1944 redistricting has been undertaken by neutral Boundary Commissions operating a set of enacted Rules for Redistribution in an entirely non-partisan way (for a full discussion of the Commissions, see Rossiter et al. Reference Rossiter, Johnston and Pattie1999). In such a situation, any partisan benefit that accrues from the introduction of a new set of constituencies should be entirely serendipitous, but research reported over three decades ago – some of the mathematically most sophisticated ever applied to this area of work in the UK (Gudgin and Taylor Reference Gudgin and Taylor1979) – showed that partisan outcomes from such non-partisan procedures could be the norm. The parties know that and, to a greater or lesser extent, have sought either to operate within the rules or to modify them to gain electoral advantage. Illustrating that activity, with particular reference to the Parliamentary Voting System and Constituencies Act, 2011, and its implementation, is the focus of this article.
MALAPPORTIONMENT – GB STYLE
Malapportionment, in the British case, refers to differences between constituencies in their number of electors. It generates two types of concern. The first is where substantial differences in constituency size mean not only considerable variations in MPs’ workloads – particularly on issues raised with them by individual residents and interest groups – but also differences in representation ratios (residents of a constituency with 50,000 electors, for example, could claim they are significantly under-represented compared to those in another with only 25,000). The second type of concern arises where that inequality favours one party rather than others in the translation of votes into seats. A party whose support is concentrated in relatively small constituencies will get more seats, relative to its overall share of the votes cast, than another party which gets the same share of the votes but its support is concentrated in the larger constituencies (Johnston et al. Reference Johnston, Pattie, Dorling and Rossiter2001).
The latter concern is the one addressed by party strategists seeking to maximise their returns from any given share of the votes – although they may deploy arguments based on the first concern when arguing for rule changes. And those bases for such concern have existed in the UK ever since Parliaments were convened: the Rules for Redistribution first formalised in 1944 did not remove them.
The UK has had malapportionment over the period 1944–2011 for three reasons (Johnston et al. Reference Johnston, Pattie, Dorling and Rossiter2001). First, from the outset the allocation of seats to the country’s four constituent parts – England, Northern Ireland, Scotland and Wales – has not been commensurate with their electorates, although there is no evidence this was done for partisan reasons (the draft rules were drawn up by an all-party Speaker’s Conference, to reflect the status quo ante).Footnote 2 In effect, England was under-represented. That inequality increased as England’s population grew more rapidly than that of the other parts of the UK, a situation partly exacerbated by the intricacies of the rules which almost obliged the Boundary Commissions for Scotland and Wales to increase their number of MPs at most redistributions. It was somewhat stemmed in the case of Scotland when the 1998 devolution legislation required the Scottish Commission to use the same average electorate (quota) as the English when undertaking its next redistribution, resulting in the reduction of Scottish MPs from 72 to 59 in 2004. There was no commensurate change in Wales, however, and at the 2010 general election the average Welsh constituency had 55,767 electors, compared to 70,203 for England. (Scotland’s average was 65,287, lower than England’s because a rule allowing Commissions to take account of special geographical considerations permitted them to create much smaller constituencies than the norm for the Highlands, the Western Isles and for Orkney and Shetland.)
The second reason is that the Commissions have recommended constituencies of unequal size within each country, in part because the rules require them to take ‘special geographical considerations’ into account. They have interpreted this as recommending smaller constituencies (by number of electors, but larger by area) in areas of low population density. Over time, the number of constituencies created using this rule declined and at each subsequent redistribution the Commissions successfully reduced the variation in electorates while respecting the other rules.
Thirdly, the geography of population change over the period between redistributions was far from even. In general, inner city constituencies experienced substantial (relative if not absolute) decline whereas the outer suburbs and towns beyond suburbia grew relatively rapidly so over time the variation in constituency electorates increased. This tendency could have been reduced in its impact if redistributions were conducted frequently, which was the original intention: the House of Commons (Redistribution of Seats) Act, 1944, specified redistributions every seven years. The first was reported in 1947 and – after some amendments in 1949 – the constituencies that emerged were used at the 1950 and 1951 general elections. A further redistribution then produced new constituencies for the 1955 election, to which many MPs reacted negatively. They didn’t want their constituencies altered that frequently and the Conservative government amended the legislation in 1958 so that redistributions occurred only every 10–15 years. It also downgraded the importance of electoral equality as a criterion on which the redistributions within each country were to be based: continuity of representation of defined communities (usually identified as local authorities) was to dominate, and constituency electorates only had to be as ‘equal as is practicable’ within the required implementation of those other criteria.
All three of these causes of malapportionment, especially the first and third, have had substantial impacts on general election results since 1955, to Labour’s benefit (Johnston et al. Reference Johnston, Pattie, Dorling and Rossiter2001). It has become by far the dominant party in both Scotland and Wales and benefited accordingly in the translation of votes into seats. Further, because its main areas of electoral strength are in urban Britain it has benefited between each pair of redistributions from the many declining electorates there. Indeed, the (fairly accurate) ‘conventional wisdom’ until the 1990s, shared by the parties and commentators, was that Labour suffered a net loss of around 20 seats after each redistribution, as constituencies were removed from the declining urban cores and re-allocated to areas of Conservative strength in the expanding outer suburbs and beyond.
These pro-Labour advantages in the translation of votes into seats because of malapportionment are one element of a wider pattern of bias in recent British election results. The largest party at each election post-1945 in terms of votes has been allocated a larger percentage share of the seats than of the votes – a disproportionality considered normal in single-member plurality election systems. That disproportionality has not always treated each of the two main parties equally, however: there was bias favouring one over the other. This has been demonstrated using a statistic devised to answer the question: if each party had the same share of the votes cast, would each also get the same share of the seats? (The procedure is fully elaborated in Johnston et al. Reference Johnston, Pattie, Dorling and Rossiter2001; see also Thrasher Reference Thrasher2012.) In the 1950s and 1960s the answer was no, and the beneficiary was the Conservative party, by as many as 50 seats (i.e. if the Conservatives and Labour had equal shares of the votes cast, the Conservatives would have won more seats). In the 1970s and 1980s there was very little bias, but in 1992 there was a major change: thereafter, not only did the volume of bias increase – to a maximum of 142 seats – but it now favoured Labour. Part of that pro-Labour bias – identifiable because the measure could be decomposed into various contributing factors – was due to malapportionment.
Awareness – in a general if not a mathematical sense – of the impact of malapportionment stimulated attempts to achieve partisan gain by manipulating the procedure in some way. In 1969, for example, Labour – then in government – calculated it could lose around 20 seats if the Boundary Commissions’ recommended new constituencies were used at the next election. It used an unfinished local government reorganisation to argue that the new constituencies should not be introduced. This was contested by the Conservatives, so the Home Secretary, James Callaghan, introduced the orders creating them but his party whips instructed their MPs to vote them down: they did, and the 1970 election was held in the old seats – producing the largest malapportionment bias component ever for the entire period 1950–2010. Labour again attempted to prevent the Commissions’ recommended new constituencies being implemented in 1982. It was not in government, so channelled its opposition through the courts – arguing that the Commissions had failed to apply the rules fully with regard to equal electorates. It lost; the 1983 election was fought in the new constituencies, and the pro-Labour bias due to malapportionment fell relative to 1979.
Labour was not alone in seeking to manipulate the malapportionment bias. Having secured a narrow victory in 1992 the Conservatives wanted to ensure that their prospects next time were as rosy as possible. Their Boundary Commissions Act, 1992, required the Commissions to speed up their ongoing redistribution so that new constituencies were in place in time. They did, and the 1997 contest was fought in them but – as for Labour in 1970 and 1983 – it didn’t ensure victory. (For full details of all three attempts to promote malapportionment, see Rossiter et al. Reference Rossiter, Johnston and Pattie1999.)
The 2011 Legislation
The 1997, 2001 and 2005 general elections saw by far the greatest pro-Labour bias: at the first of those contests with equal vote shares Labour would have won 82 more seats than the Conservatives; at the second, the difference would have been 142 seats; and in 2005 it would have been 112. The Conservatives slowly realised the disadvantage they were suffering under: after 2001 some – notably Andrew Tyrie (Reference Tyrie2004) – were arguing that the rules should be changed. After 2005 it became party policy to do so, and draft legislation was prepared, having been prefigured twice; a draft Bill debated in the Lords but not in the Commons proposed a maximum 5 per cent variation for all constituencies around a single UK-wide quota; and an amendment to a 2010 Labour Constitutional Reform Bill proposed a 3.5 per cent maximum variation (Johnston and Pattie Reference Johnston and Pattie2012; Johnston et al. Reference Johnston, McLean, Pattie and Rossiter2009).
The Conservatives’ strategists realised that variations in constituency electorates were neither the sole nor the major cause of the pro-Labour bias at those three elections, but were convinced that it could be substantially reduced. The Bill introduced in July 2010 proposed three major changes to the Rules for Redistribution that had applied over the previous 50 years:
-
∙ There was to be a single UK electoral quota, thereby removing England’s under-representation in the House of Commons;
-
∙ All constituencies, with four special exceptions, were to have electorates within +/−5 per cent of that quota, thereby ensuring electoral equality across the entire UK; and
-
∙ There was to be a redistribution every five years, with the new constituencies approved by Parliament 18 months before each general election (whose dates were to be determined by the Fixed Term Parliaments Act, 2011), thereby removing Labour’s advantage from population changes between redistributions.
The other criteria which were dominant under the previous rules – fitting into the geography of local authority boundaries; reflecting local community ties; maintaining continuity wherever possible with the previous constituencies; and special geographical considerations – could only be taken into account so long as the arithmetic criterion was met. This meant that the new constituency map to be in place for the 2015 election would differ very considerably from that used in 2010.
The Conservatives exacerbated that potential difference by a further clause in the Bill: the number of constituencies, currently 650, was to be fixed at 600 (a 7.7 per cent reduction). Thus the number of Welsh MPs would be reduced by 25 per cent, from 40 to 30, for example, reflecting the Principality’s over-representation under the previous regime. The reductions were smaller elsewhere but nevertheless, as was widely recognised, this additional change meant that in many parts of the UK the pattern of representation would be even more different in 2015 than it had been over previous decades because malapportionment was to be removed (Rossiter et al. Reference Rossiter, Johnston and Pattie2013).
GERRYMANDERING – UK STYLE
Malapportionment was not the sole cause of Labour’s significant advantages in the operation of the electoral system from 1992 on: it benefited more from both the geography of abstentions (much higher in its heartland seats than in those traditionally won by the Conservatives) and the efficiency of the geographical distribution of its votes. That concept of efficiency can be explained by categorising votes into three types: wasted, surplus, and effective. Wasted votes are those a party gets in constituencies where it loses, contributing nothing to its tally of seats; surplus votes are gained in constituencies that it wins, but are additional to those needed for victory – i.e. one more than the second-placed party’s total; and effective votes are those that win the seats. Thus in a seat contested by two parties where A gets 25,000 votes and B 15,000, all of B’s votes are wasted; 15,001 of A’s are effective and its remaining 9,999 are surplus.
A party’s electoral strategy should aim to maximise its effective votes and minimise the other two: its goal should be to ‘win small but lose big’, though not playing it too cleverly, misjudging any local situations and losing some that it should win (a small majority can readily become a small loss). However, the parties do not draw the constituency boundaries, so such strategies are only open to them insofar as they decide not to campaign too hard in either seats they know they are going to lose or those where victory is certain (tactics that all employ: Johnston and Pattie Reference Johnston and Pattie2014).
Why then was there such a substantial efficiency component to the pro-Labour bias at the 1997–2005 elections? In part, this was a natural reflection of the geographies of party support, especially those for Labour and the Conservatives. There are some places where each is so strong that it is virtually certain to win all of the seats there. In Coventry, for example, Labour won 50.1 per cent of votes cast in 2005, giving it an easy victory in each of the city’s three seats: five years later its vote share slipped to 44.5 per cent, but it still had a comfortable victory in each of the three (new) seats. The same was true in many other areas: Labour won a larger share of the seats (even if it did not win them all) than of the votes: indeed it would be very difficult to devise a set of three constituencies for Coventry that would not all be won by Labour providing it won more than around 40 per cent of the votes. Elsewhere, the same held true for the Conservatives although – particularly after the rise of the Liberal Democrats in many parts of England – rarely across an entire county or borough. (Johnston et al. Reference Johnston, Borisyuk, Thrasher and Rallings2012 show that at recent elections, both the Conservative and Labour parties – but not the Liberal Democrats – can expect to win a seat where they get more than 40 per cent of the votes cast.)
Elsewhere, there is no single party hegemony and relatively small changes in vote shares can lead to larger shifts in the allocation of seats. In Swindon, for example, Labour won 42.0 per cent of the votes to the Conservatives’ 37.6 per cent in 2005 and held both of the town’s seats. Four years later, their respective vote shares were 32.3 and 43.2 per cent, with the Conservatives victorious in both constituencies. A relatively small change in vote share was exaggerated in the allocation of seats (another feature generally associated with single-member plurality electoral systems: Rae 1971) with substantial shifts in the share of each party’s votes that were effective, wasted and surplus.
If each of the two parties has parts of the country where it dominates the voting pattern and so is almost certain – save a major shift in vote support – to win the vast majority of the seats there, then those two sets of bias should balance each other out. Where one wins out over the other comprehensively in areas like Swindon, however, then the bias should be strongly in its favour. That happened to Labour’s advantage in 1997, 2001 and, to a considerable extent, 2005. But the bias then was much more in Labour’s favour than it was for the Conservatives in their 1983 and 1987 landslides. One reason for that was tactical anti-Conservative voting at Labour’s landslide victories under Blair. In many seats Labour wasted fewer votes (i.e. ‘lost bigger’) than it otherwise might have done because of its supporters switching tactically to the Liberal Democrats; elsewhere it won others by small margins (i.e. with few surplus votes – ‘winning small’) that it might otherwise have lost with many wasted votes if it hadn’t attracted tactical votes from Liberal Democrat supporters.
There was a final reason. Gerrymandering the boundaries isn’t part of the redistribution procedure but the parties are deeply involved in it, and direct much effort at enhancing their electoral prospects through the constituency-building process. The House of Commons (Redistribution of Seats) Act, 1958, formalised the public consultation procedure. After publication of a Commission’s provisional recommendations for an area (usually a county or borough) any interested persons or bodies could submit written representations and if sufficient people expressed negative concerns a Local Inquiry was convened at which not only the pros and cons of the recommendations but also those of alternative configurations could be debated. The Labour party soon realised this was an opportunity to promote its electoral prospects and in preparation for the Fourth Periodic Review, which recommended the constituencies used first at the 1997 election, it prepared schemes for every part of the country and developed supporting arguments to be presented to the Assistant Commissioners (using arguments within the rules and without ever indicating the possible partisan implications of its desired changes). It completely outflanked the Conservatives – who complacently assumed that they would again benefit by some 20 seats from the redistribution – and in some places significantly changed the electoral complexion. In effect, the party used a strategy akin to that of the gerrymanderer, except that they didn’t control the final outcome and could only seek to influence the Commission-cartographers’ decisions – ‘gerrymandering by consultation’.
The Conservatives determined not to be beaten again, and their 2011 Bill proposed changes to the public consultation procedure accordingly. They initially proposed abolition of Local Inquiries; there were to be written representations only. Labour strenuously opposed this in the Lords and the government eventually yielded, introducing a system of non-confrontational Public Hearings to be held during the period for written representations (Johnston et al. Reference Johnston, Pattie and Rossiter2013). The Conservatives had already committed considerable resources to developing schemes for each part of the country so that by the time the Commissions published their provisional recommendations they were prepared – and, as it turned out, much better prepared than Labour – to press their partisan claims during the consultations, as Labour had been 20 years previously.
THE (UNFINISHED) SIXTH PERIODIC REVIEWS OF UK PARLIAMENTARY CONSTITUENCIES, 2011–13
The Parliamentary Voting System and Constituencies Act, 2011, became law in February 2011 and the Commissions, having already done much preliminary work, formally started their first reviews under the new Rules for Redistribution in March: final sets of recommended constituencies had to be delivered to Parliament via the relevant Secretaries of State by October 2013. (Previous Reviews had taken much longer, and the government feared that if it had retained the previous Local Inquiry procedure the exercise would not have been completed in time for the new constituencies to be in place for the 2015 general election.) Full sets of provisional recommended constituencies were published for England, Northern Ireland and Scotland in late 2011 and for Wales in early 2012. The 12-week periods for written recommendations started on publication date, and all of the scheduled Public Hearings were held during that period. The Commissions spent spring and summer of 2012 considering the arguments and evidence presented, and in the autumn published their revised recommendations, on which there was a short period for written representations. These were received and considered, and by early 2013 much had been done to finalise the recommendations.
They were never published, however, because Parliament brought the review to a premature end. Angered by David Cameron’s campaigning against the alternative vote referendum and failure to deliver his party’s support for House of Lords reform, the Liberal Democrat leader, Nick Clegg, announced in July 2012 that he would instruct his MPs and peers to vote against implementation of the Commissions’ recommendations when they came to Parliament in October 2013. This meant that they were very unlikely to be approved, as Labour had already indicated its intention to do likewise. The reviews continued, however, until in January 2013 an amendment to the Electoral Registration and Administration Bill was successfully introduced by four peers that would delay implementation of the new Rules for Redistribution until 2016. The Conservatives failed to overturn it in the Commons, where Labour and Liberal Democrat MPs combined to retain it in the Act. The Reviews were immediately halted, which meant that the 2015 election would be fought in the same 650 constituencies as the 2010 contest.
Although the Reviews were not finished, they came close enough to their conclusion for an evaluation of the impact of both the Conservatives’ removal of the malapportionment elements to the UK’s electoral system and its preparations for gerrymandering the redistribution, UK-style, by consultation. Using a methodology developed by Colin Rallings and Michael Thrasher, and implemented by Anthony Wells, it is possible to rerun the 2010 general election as if it had been held in any alternative set of constituencies. This has been done not only for the Commissions’ provisional and revised sets of recommendations but also for the alternatives proposed during the initial consultation period by the Conservative, Labour and Liberal Democrat parties.Footnote 3
As anticipated, though more so than many MPs had expected, the Commissions’ recommendations involved much more substantial changes to the map of constituencies than had been the case at previous redistributions, because of the combination of the single UK quota, the +/−5 per cent limit to variations around that, and the reduction in the number of seats. The disruption had three main features:
-
∙ The much greater fragmentation of many of the existing constituencies than at previous reviews;
-
∙ The much larger number of constituencies than at previous reviews that crossed major local government (i.e. county and borough) boundaries; and
-
∙ The splitting of many identified communities between two or more constituencies and the inclusion in many constituencies of separate communities between which there were few links.
The extent of the change was greatest in urban England, where the relatively large wards used as the building blocks for the constituencies presented a major problem that the Boundary Commission was not prepared to address by splitting wards (unlike its Scottish and Welsh counterparts, which did so without stimulating much dissent).Footnote 4 Leeds provides an excellent example. At the 2010 election it had seven MPs: six represented constituencies comprising Leeds wards only, and the other represented three Leeds wards combined with two from Wakefield. Under the new rules it was entitled to 7.1 MPs, which could have been rounded down but it was impossible to create seven constituencies using the city’s 33 wards as building blocks. Instead the Commission recommended eight constituencies, just three of which comprised five Leeds wards only. Two others crossed the Leeds–Bradford boundary, one the Leeds–Kirklees, another the Leeds–Wakefield; the last one combined four Leeds wards with five much smaller ones in the Harrogate district of North Yorkshire. This fragmentation of the city not only caused much concern within its boundaries but also outwith them. North Yorkshire had eight constituencies, all of them with electorates within the new allowed size variation, but the decision to link some Harrogate and Leeds wards, plus similar West–North Yorkshire boundary crossings further south, meant that all but one of North Yorkshire’s eight seats were scheduled for changes – much to the consternation of their MPs who thought they would be unaffected.
The changes in Leeds and the knock-on effects not only in North Yorkshire but elsewhere in West Yorkshire (notably the towns of Batley, Dewsbury and Wakefield which were to lose coherent representation) stimulated many representations and suggested alternative configurations – with two from the Conservatives.Footnote 5 The Assistant Commissioners responded by agreeing to recommend that the eight North Yorkshire constituencies be unchanged from their existing configuration, which meant that they came up with an entirely new set of recommendations for Leeds. Instead of the eight provisionally recommended it now proposed nine. Three of them comprised Leeds wards only (and none of them the same as any of the first set); additionally it proposed three that all crossed the Leeds–Bradford boundary, two that both combined wards from Leeds and Wakefield, and one linking two Kirklees wards with three in south-west Leeds.
Such disruption to the existing constituencies was unwelcome to many MPs and their local parties, but did it achieve the national party’s strategicians’ aims? Table 1 shows the results of bias analyses for the actual results of the 2010 election in Great Britain’s 632 constituencies then (Northern Ireland’s 18 are omitted) plus those for the two sets of Boundary Commission recommendations – if the 2010 election had been held in the proposed 584, plus those in the schemes proposed by each of the three main parties in their representations responding to the provisional recommendations.Footnote 6 In that table a positive number indicates bias towards Labour and a negative shows pro-Conservative bias. It shows that the malapportionment effect was entirely removed by insisting that all but four of the constituencies had electorates between 72,810 and 80,473. The first number at the top of the malapportionment column indicates that, in the constituencies that were actually used for the 2010 general election, that bias component would have been worth 18 seats to Labour if it and the Conservatives had obtained equal vote shares at that election (and the number and geography of votes for the Liberal Democrats and other parties, along with the number of abstentions, had remained unchanged); in any of the five alternative schemes shown in the other columns that advantage had totally gone.
The Bias Components in the Result of the 2010 General Election in Great Britain, and in Various Alternative Configurations of 584 Constituencies
Note: A positive bias score favours Labour, a negative score favours the Conservatives.
Key to bias components: M = malapportionment; A = abstentions; G = geography; TP = third parties; Σ = total. The sum of the four elements does not always equal the total bias shown in column Σ because of interaction effects.
Turning to the geography component, at the actual 2010 election neither party was advantaged – suggesting that, as in the 1970s and 1980s, each had approximately the same share of effective votes. In the Commissions’ provisional recommendations, however, there was a pro-Conservative bias of 13 seats: their recommended redistribution serendipitously favoured the Conservatives. The figures in the next three columns indicate the extent to which the parties’ attempts to ‘gerrymander through consultation’ could have changed that; the clear conclusion is that, in terms of maximising the outcome, the Conservatives’ scheme would have been the most successful. It would have more than tripled that bias component, to 41 seats, and resulted in an overall small pro-Conservative bias (as against 27 for Labour in the Commissions’ proposals). Labour’s scheme, on the other hand, would have halved the pro-Conservative efficiency bias from the Commissions’ proposals and slightly increased its overall advantage.Footnote 7
It is very unlikely, of course, that the Commissions – or in England’s case the nine teams of Assistant Commissioners, whose revised recommendations were accepted in their entirety by the Commission itself – would be totally swayed by one party’s alternatives and entirely reject the others’ (or those suggested in non-party representations).Footnote 8 But the data in the final row of Table 1 indicate that the Conservatives were most successful in influencing the Commissions’ revised recommendations. The pro-Conservative efficiency bias component increased from 13 to 18 between the Commissions’ provisional and revised recommendations and the overall pro-Labour bias reduced by more than one-third.
Recent developments in the measurement of bias have extended the approach adopted here to the study of three-party systems, which the British situation approximated in 2010 (Borisyuk et al. 2010; Thrasher et al. 2011). This gives a measure of each bias component separately for each of the three parties – with a positive value indicating that the party concerned is advantaged by that component and a negative value that it is disadvantaged. Table 2 gives the values for the geography component only – the focus of this analysis of ‘gerrymandering by consultation’. The first row shows that at the actual 2010 general election both the Conservatives and Labour were equally advantaged by that component and the Liberal Democrats substantially disadvantaged. The Commissions’ provisional recommendations did not alter that situation significantly – given the smaller number of seats; the Conservatives and Labour both benefited substantially from having their votes distributed more efficiently than the Liberal Democrats, a reflection of the situation in 2010 whereby the country (basically England) did not have one three-party system but rather three two-party systems (Johnston and Pattie Reference Johnston and Pattie2011). Each of the three parties’ alternative schemes would have substantially altered the direction of the bias if implemented, however: if the Conservatives’ scheme had been adopted in its entirety, for example, it would have given that party a 25-seat advantage over Labour; Labour’s scheme would have given it an 11-seat advantage over its main opponent (in both cases with the Liberal Democrats’ disadvantage being exacerbated); and if the latter’s scheme had been adopted, that disadvantage would have been reduced to less than in the Commissions’ provisional proposals, with the Conservatives having a 10-seat advantage over Labour. The Commissions’ revised recommendations favoured the Conservatives, with a 14-seat advantage over Labour, and the Liberal Democrats disadvantage was considerably reduced relative to the provisional recommendations. The Liberal Democrats and Conservatives, in that order, were the clear beneficiaries from the ‘gerrymandering by consultation’
The Geography Bias Component in Three-party Bias Analyses of the Result of the 2010 General Election in Great Britain, and in Various Alternative Configurations of 584 Constituencies
Note: A positive bias score favours the named party.
The bias figures are, in one sense, statistical science fictions – a good measure of the bias in the system but not a straightforward indicator of the ‘real’ impact of the changes: who would win and by how much? Table 3 shows the number of seats that each party won in 2010 (when Great Britain had 632 MPs) and its estimated number in each of the various schemes. The first row shows the Conservatives’ 48-seat lead over Labour in the House of Commons elected in 2010, but lack of a majority over all parties (for which it would have needed 326).Footnote 9 The Commissions’ provisional recommendations substantially changed the former situation, giving the Conservatives a 68-seat lead over Labour in a smaller House and making it only two seats short of an overall majority.
The Result of the 2010 General Election in Great Britain, by Number of Seats Won, and in Various Alternative Configurations of 584 Constituencies
If the Conservatives’ alternative scheme had been implemented in full by the Commissions their lead over Labour would have been almost twice that of the actual 2010 result – 93 seats; and they would have had a small majority over all parties. Labour’s scheme would have reduced the Conservatives’ lead to only 47 seats, and denied them a majority; the Liberal Democrats’ scheme – somewhat surprisingly since this was the first redistribution at which they had played a major role in the public consultation process – would also have been quite successful from their point of view, increasing their number of seats from the 46 proposed by the Commission to 56 as well as (almost certainly serendipitously) somewhat reducing the Conservative–Labour gap and denying the former party a majority. The Conservatives clearly succeeded over Labour – as the bias calculations suggested. If the Commissions’ revised recommendations had been implemented – and it is very likely that the final round of public consultation would have resulted in only a few changes from the published revised recommendations – then the Conservatives would have won with a small majority in 2010 and a lead over Labour of 70 seats.
The Conservatives clearly outflanked Labour in this procedure. But in many ways the biggest winners were the Liberal Democrats, who previously had only been half-hearted in their approach to redistributions (Rossiter et al. Reference Rossiter, Johnston and Pattie1999). For this review, a national official was given the task of coordinating the preparation of alternative schemes and supporting evidence by regional office teams, all of which invested substantial effort into the task; among the main parties, they were the only one to propose ward-splitting in England in order to protect some of their currently-held seats, such as Portsmouth South.Footnote 10 And they were successful: they increased their projected seat tally by five between the Commissions’ provisional and revised recommendations, compared to the Conservatives’ increase of just three (which was much smaller in relative terms: Table 3). To a considerable extent, the Conservatives’ gain over Labour through the ‘gerrymandering by consultation’ procedure came about because of the Liberal Democrats’ success in promoting changes to the Boundary Commissions, almost entirely in England.
These are, of course, only estimates – with unknown errors – but the overall picture they paint strongly sustains the Conservative strategy that the disruption to the constituency map created by their decisions to reduce the number of MPs and change the Rules for Redistribution to remove malapportionment, plus their careful preparation of evidence for the Commissions, was justified. If the Commissions’ final recommendations had been delivered to and accepted by Parliament in October 2013 the Conservatives would have gone into the 2015 election in a stronger position than if it was to be fought in the same constituencies in which they failed to win outright in 2010. They could have been defending a hypothetical majority in half of Great Britain’s seats. Furthermore, the disruption to the map would have probably negatively affected Labour (and even more so the Liberal Democrats) more than the Conservatives. The Conservatives have much stronger – and wealthier – local parties than either of the other two (Johnston and Pattie Reference Johnston and Pattie2014), and the requirement to restructure their local organisations could well have put greater stress on Labour and Liberal Democrat ability to mount intensive local campaigns than on the Conservatives. For them, therefore, the aborted redistribution was an opportunity clearly missed – because of internal coalition politics: removal of malapportionment plus successful ‘gerrymandering through consultation’ would have significantly altered the dynamics of the 2015 general election campaign.
But what about 2020? As the legislation currently stands, a new redistribution using the rules specified in the 2011 Act will begin in 2016 – again with a reduction of MPs from 650 to 600; it should recommend new constituencies in October 2018, to be used at the general election scheduled for May 2020. That would almost certainly be as disruptive of the current map as the aborted redistribution discussed here – if not more so: the allocation of seats across the four countries and England’s regions could well be changed;Footnote 11 the ward structure in many parts of England at least will be different from that in 2011, thanks to the work of the Local Government Boundary Commission for England, and there are uncertainties as to how accurate and complete the electoral roll will be after the introduction of Individual Electoral Registration. If the Conservatives form a majority government in 2015, they may well decide that the disruption is worth it to strengthen their position for 2020, and convince their MPs and regional and party organisations accordingly. If they form another coalition, they will have to decide what policy concessions to make to their partners to ensure support for the exercise. (Those concessions might include some minor changes to the rules and/or retaining the number of MPs at 650.) If Labour is in power after 2015 (alone or in coalition) it might decide either to repeal the 2011 legislation and return to the previous Rules for Redistribution – which, because of the length of time their implementation took at recent reviews, could mean the same constituencies being used in 2020 as in 2010 – or to amend the 2011 rules to its advantage (and perhaps also repeal the Fixed Term Parliaments Act).
CONCLUSIONS
The importance of geography to the UK’s electoral system has become increasingly apparent in recent decades. The shift from a two- to a three-party system in the post-1980 decades in effect meant that England was divided into three different parts, each with a different two-party system and a very-much minority third party. Within that new electoral map, the parties increasingly targeted their intensive campaigning – much of it centrally-directed – on a relatively small number of marginal seats. And they learned that how that map is redrawn can be crucial to their electoral prospects.
This article has illustrated the last of those lessons with reference to biases in election results introduced by malapportionment and gerrymandering – UK style. Analysis of the prematurely ended redistribution under a new set of rules crafted to promote the Conservatives’ interests has shown how removal of malapportionment and well-crafted ‘gerrymandering by consultation’ could have placed them in a much stronger position to fight the 2015 general election than their failure to get that redistribution completed means. If, as many believe, the result of that election will be at best a small minority government for either Labour or the Conservatives (with the possibility of that being hard to sustain, as the post-1976 evidence for Labour and post-1992 for the Conservatives suggests), or another coalition, or a minority (Conservative or Labour) government, the importance of the biases inherent to the current system will be re-emphasised. In any knife-edge election results, the (direct and indirect) influence of geography could be profound.
ACKNOWLEDGEMENT
This article, which is a revised version of the Government and Opposition/Leonard Schapiro Memorial Lecture, given on 14 April 2014 at the annual conference of the Political Studies Association, draws almost entirely on much-enjoyed collaborative research conducted over some 35 years with David Rossiter and Charles Pattie; I am grateful to them for their comments on a draft of the manuscript. I am also grateful to Galina Borisyuk, Colin Rallings and Michael Thrasher for collaboration on the measurement of bias in three-party systems, and for their algorithm estimating general election results in alternative constituency configurations: all of the estimates presented here using that algorithm were produced for us by Anthony Wells.
