The Politico's Guide to Electoral Reform in Britain

Patrick Dunleavy, Helen Margetts and Stuart Weir The Politico's guide to electoral reform in Britain

Book section

Original citation: Originally published in Dunleavy, Patrick, Margetts, Helen and Weir, Stuart (1998) The Politico's guide to electoral reform in Britain. Politico's Publishing, London, UK. ISBN 190230120X

This version available at: http://eprints.lse.ac.uk/62253/

Available in LSE Research Online: June 2015

LSE has developed LSE Research Online so that users may access research output of the School. Copyright © and Moral Rights for the papers on this site are retained by the individual authors and/or other copyright owners. Users may download and/or print one copy of any article(s) in LSE Research Online to facilitate their private study or for non-commercial research. You may not engage in further distribution of the material or use it for any profit-making activities or any commercial gain. You may freely distribute the URL (http://eprints.lse.ac.uk) of the LSE Research Online website.

the Guide to ELECTORAL REFORM in Britain

Patrick Dunleavy, Helen Margetts and Stuart Weir First published in Great Britain 1998 by Politico’s Publishing 8 Artillery Row London SW1P 1RZ England Telephone 0171 931 0090 Email [email protected] Website http://www.politicos.co.uk Copyright Patrick Dunleavy, Helen Margetts and Stuart Weir 1998 The right of Patrick Dunleavy, Helen Margetts and Stuart Weir to be identified as authors of this work has been asserted by them in accordance with the Copyright, Design and Patents Act 1988 A catalogue record for this book is available from the British library ISBN 190230120X Printed and bound in Great Britain by Colourworks Typesetting and cover design by Tony Garrett All rights reserved. No part of this publication may be reproduced or transmittted in any form or by any means, electronic or mechanical including photocopying, recording or any information storage or retrieval system, without prior permission from the publishers. This book is sold subject to the condition that it shall not by way of trade or otherwise be lent, resold, hired out, or otherwise circulated without the publisher’s pior consent in writing in any form of binding or cover other than that in which it is published and without a similar condition including this condition being imposed on the subsequent purchaser.

2 + The Politico’s Guide to Electoral Reform in Britain Contents

About this report 4

Acknowledgements 5

INTRODUCTION 6 Jenkins in a nutshell 6

FIRST PAST THE POST 9 Deviation from proportionality in 1997 and 1992 9 Electoral deserts 13 The virtues and vices of FPTP 14 The Jenkins arguments in perspective 15

COALITION GOVERNMENT 20 The idea of the party mandate 20 How effective are coalition governments? 23

AV-PLUS – THE BASICS OF THE JENKINS SCHEME 25 Constituency and top-up seats under the Jenkins scheme 25 Calculating the effects of AV-Plus 28 The supplementary vote 32

HOW AV-PLUS WOULD HAVE WORKED IN 1997 AND 1992 33

VOTES AND SEATS UNDER AMS IN 1997 AND 1992 43

THE ‘DANGER’ OF SPLIT-TICKET TACTICAL VOTING 48

THE REPRESENTATION OF SMALL OR ‘EXTREME’ PARTIES 53

THE TREATMENT OF NORTHERN IRELAND 55

CONCLUSIONS – A BOLD AND INGENIOUS SOLUTION 59

References 62

About the Democratic Audit 63

Inside the back cover there is a map of the Jenkins proposals, showing top-up areas, seats won, etc. See full explanation on the back page opposite.

The Politico’s Guide to Electoral Reform in Britain + 3 About this report About the authors This guide is primarily an expert Patrick Dunleavy is Professor of Gov- analysis of the electoral and political ernment at the London School of effects of the scheme proposed by the Economics and Dr Helen Margetts is Independent Commission on the Vot- Lecturer in Politics at Birkbeck Col- ing System (the Jenkins Commission) lege, London.. They acted as for elections to the House of Com- consultants to the Independent Com- mons. It compares the Jenkins mission on the Voting System and scheme, known as AV-Plus, with the previously also to the Government current ‘first past the post’ electoral Office for London on preparing the system, because the Commission’s electoral systems for the election of home-grown scheme, known as AV- the new Mayor of London and the Plus, will be put to the public as an London Assembly. alternative to first past the post elec- Stuart Weir is Senior Research Fel- tions in a referendum. low at the University of Essex and The new scheme could not be in- Director of the Democratic Audit. His troduced until after the next election. latest book, Political Power and Demo- The government has not yet an- cratic Control in Britain (with David nounced the date of the referendum, Beetham) on the power of government but Lord Jenkins has indicated that in the UK, was published by he would regard it as a ‘betrayal’ if it Routledge on 19 November 1998. were delayed until after the next gen- Dunleavy, Margetts and Weir have eral election. collaborated on modelling election re- This guide is based on research sults in the United Kingdom since undertaken by Patrick Dunleavy and 1991. In 1992, they published the first Helen Margetts for the Jenkins Com- study, Replaying the 1992 General mission; previous research on the Election: how Britain would have voted 1992 and 1997 general elections; and under alternative electoral systems a special study of mixed voting sys- (LSE Public Policy Group). In 1997, tems. These previous studies were the Democratic Audit published their variously commissioned and funded follow-up study, Making Votes Count, by the Joseph Rowntree Reform Trust, which analysed re-runs of the 1992 the Economic and Social Research and 1997 elections under alternative Council, and the Joseph Rowntree electoral systems. When it became Charitable Trust. clear that the Independent Commis- sion was considering a ‘mixed’ electoral system, the Audit published a further study of the results of such a system as Making Votes Count 2. Full details of these and other election studies will be found in the references on page 62. 4 + The Politico’s Guide to Electoral Reform in Britain Acknowledgements Commission – for Scotland, Professor This report would not have been pos- David Denver, University of Lancas- sible without the help and assistance ter; for Wales, Dr Barry Jones, of a wide range of people since 1991. University of Wales, Cardiff; for First, we are very grateful to the Eco- Northern Ireland, Dr Sidney Elliot, nomic and Social Research Council Queen’s, Belfast, and Professor and the Joseph Rowntree Charitable Brendan O’Leary, LSE. Trust for funding both the ICM pub- We are grateful to Paul Laughlin, lic opinion surveys on which the 1997 of RTE Dublin, for permission to use studies are based and the extensive data collected by Ulster Marketing computer analysis involved. (The Surveys Ltd on voters’ second and ESRC Award Number was subsequent preferences in the 1998 N000222253.) The 1992 research was Northern Ireland Assembly election, funded by the Joseph Rowntree Re- and to Richard Moore of UMS for his form Trust. help in supplying the data. We owe special thanks to David Jane Pugh, of the LSE Geography Shutt, chairperson of the Charitable Drawing Office, designs and produces Trust’s Democratic Panel, panel mem- the excellent maps which lend focus bers and Steve Burkeman, Trust and colour to our reports. Tony Garrett secretary; to Professor Lord Smith of designed and produced this report Clifton, chairperson of the Joseph with his usual skill. Rowntree Reform Trust, and Trust Patrick Dunleavy, members; to Nick Sparrow, of ICM, Helen Margetts, for his ideas and enthusiasm for our Stuart Weir idea in 1992 and 1997. Those polling November 1998 companies that tendered for the polling contract in 1997 provided a wealth of helpful advice and suggestions, especially Nick Moon from NOP, Simon Orton of BMRB and Brian Gosschalk of MORI. Our simulations relied on the STV Election Computer Program, designed by David Hill, of the Electoral Reform Society. In 1997, Dr Pippa Norris, from the Kennedy School of Government at Harvard University, kindly provided us with basic election data. Various colleagues in political sci- ence gave good advice on top-up areas and constituency pairings during the consultancy work for the The Politico’s Guide to Electoral Reform in Britain + 5 INTRODUCTION

n 29 October 1998, the In which is a significant element in Brit- dependent Commission on ain’s political culture. Indeed, the Othe Voting System (the Commission specifically says that it Jenkins Commission) proposed radi- does not wish to impose ‘a coalition cal change in the way Britain elects habit’ on the country. the House of Commons. The Commis- This report analyses the major fea- sion, chaired by Lord Jenkins of tures of the AV-Plus system which the Hillhead, recommended a two-vote Commission has recommended mixed electoral system, AV-Plus, as should be put to the British public in the best alternative for Britain to the a referendum as an alternative to first current ‘first past the post’ (FPTP), or past the post voting. We compare AV- plurality-rule, system for general elec- Plus with first past the post, but not tions. This ‘home-made’ system, with standard alternative voting sys- AV-Plus, was designed to meet the tems (for which see our previous four criteria laid down for the Com- writings); show how AV-Plus would mission by the government – that the have worked in 1992 and 1997 in alternative system they recommended some detail, including the degree of should offer greater voter choice, de- distortion in its results; and provide a liver stable government, maintain the broad assessment of its likely electoral link between MPs and local constitu- effects. encies, and produce broadly proportional results. Jenkins in a nutshell The Commission’s AV-Plus scheme Under the Commission’s scheme, AV- belongs to the broad family of mixed Plus, most MPs – 80 to 85 per cent – electoral systems, generally known as would continue to be elected in local additional member systems (AMS), constituencies, which would be rather but has been designed to build on fea- larger than now. Electors would cast tures of the existing FPTP system and their first vote for a constituency can- British political culture. In particular, didate under the alternative vote (AV) to keep strong local links and make it system, not FPTP as in the AMS possible for the larger political parties schemes for Scotland, Wales and the still to win a working majority of seats London Assembly (as well as AMS in the House of Commons on a mi- systems abroad). Under AV, voters nority of the popular vote. This number the candidates in order of objective is dictated in part by the preference on their ballot papers. If a aversion to coalition government candidate gets a majority of first-pref- 6 + The Politico’s Guide to Electoral Reform in Britain erence votes, he or she is elected. Oth- the remaining 36 – in London, cen- erwise, the bottom-placed candidate tral Scotland and large metropolitan is eliminated and his or her second- areas, as well as the two Northern Ire- preference votes are distributed land areas – would each elect two among the other candidates, and this top-up MPs. process continues from the bottom up The choice of locally identifiable until one candidate secures a simple top-up areas is a significant innova- majority of votes. tion. It is designed both to reduce In addition, electors would cast a central party control of the choice of second ‘party’ vote for some 98 to 132 candidates and their place on the top-up MPs (15 to 20 per cent of the party lists, and to provide both local total in the House of Commons). The accountability and a broad local link Commission leaves open exactly what for top-up MPs. the final proportion of top-up MPs to Local AV elections, plus the correc- local members should be. The pur- tive top-up mechanism, will, the pose of the additional layer of MPs is Commission argues, substantially in- to reduce the inevitable dis- crease voter choice. Voting under AV proportionality of the local election for local candidates frees voters from results. The more to-up MPs there are, having to face the ‘tactical vote’ choice the more accurately the composition between their first-preference candi- of the House of Commons would re- date or party and the most acceptable flect voters’ wishes; the ‘classic’ AMS of the candidates likely to win the seat. scheme has equal numbers of local They can vote in order of preference, and top-up MPs and achieves close knowing that their second and third to pure proportionality. At the same preference votes may still count if time, there would be fewer local MPs their first-preference candidate is serving larger constituencies if the knocked out. The second party vote size of the Commons is held constant also helps voters to avoid the same sort (as Jenkins intends it should be). of dilemma. The Commission also in- Under Jenkins, the top-up MPs sists that party lists in the second would not be elected nationally or ballot should be ‘open’, not ‘closed’, regionally, as under most proportional thus giving voters the choice of either representation (PR) systems, but from a straight party vote or a vote for a counties and equivalent-sized metro- specific individual candidate on the politan districts in England, Scottish list. and Welsh Euro-constituency areas, The Commission’s report states and two top-up areas in Northern Ire- that elections by the alternative vote land. There would be 80 top-up areas will ensure that all constituency mem- in total – 65 in England, eight in Scot- bers have majority support in their land, five in Wales and two in constituencies – which is not now the Northern Ireland. Of these areas, 44 case with over 40 per cent of existing would have a single top-up MP and MPs. However, there are major objec- The Politico’s Guide to Electoral Reform in Britain + 7 tions to their choice. First, AV can pro- establishment of a new independent duce considerably more dispro- Electoral Commission, to oversee portional results than FPTP, as indeed electoral administration and referen- it would have done had it been in use dums. It accepts that AV-Plus could alone in 1997. Secondly, critics – who not be introduced until after the next include Lord Alexander, a Commis- election and recommends that if the sion member – object that it gives too scheme is put in place, it should be much weight to ‘lower grade’ second, reviewed only after two elections have third and perhaps further choices. taken place and that any fundamen- The Commission also makes an tal change, such as a change in the important recommendation on one ratio of top-up MPs to local members cause of bias in current electoral ar- or a return to FPTP, should not be in- rangements. Strictly speaking, it troduced without a further would be possible to achieve some referendum. sort of parity between voters in differ- Overall, the Commission has given ent constituencies if they were all of a priority to two main element of the broadly equal size. But the UK has existing system – locally-based MPs four Boundary Commissions, one for and single-party government – over each of the ‘home countries’ and they the criterion of ‘broad proportionality’ apply quite different ‘electoral quotas’ while extending voter choice. This to determine the size of constituen- priority is evident in the Commission’s cies; and Scotland, Wales and own summary: Northern Ireland are all entitled to a ‘Our recommendation would have minimum number of constituencies – produced single party majority Gov- which for instance means that Scot- ernment in three out of the last four land’s quota is 20 per cent lower than elections, with the only exception that for English constituencies. These [1992] being a parliament which, even inequalities make for greater dispro- under the old system, exhibited many portion in election results. of the features of uncertain command. The government has already de- It is therefore difficult to argue that cided that the Scottish Commission whet we propose is a recipe either for will be able to ignore the quota for a predominance of coalitions or for Scotland. To reduce bias, the Jenkins producing a weakness of government Commission recommends that there authority . . . .’ should be a single electoral quota for the UK as a whole. The Commission also proposes that, as far as possible, the ratio of constituency to top-up MPs should be equal in the four nations of the United Kingdom. The Commission makes several other major proposals, including the 8 + The Politico’s Guide to Electoral Reform in Britain FIRST PAST THE POST

s the public will be invited to tionality’? Political scientists have choose between AV-Plus and developed many different possible Athe existing system in the indicators of electoral system perform- proposed referendum, it is important ance, but serious comparative work on to discuss and analyse the strengths electoral systems has tended to focus and weaknesses of first past the post on the concept of ‘deviation from pro- (FPTP) elections, or the plurality-rule portionality’. Table 1 shows how we system (to use the correct name). calculated deviation from proportion- A primary duty of any electoral sys- ality (DV) in the 1997 general election tem is that it should represent the in Great Britain. (We exclude North- votes – or wishes – of the electorate as ern Ireland here because it has a quite effectively as possible. The key criti- separate party system.) cism of plurality-rule in Britain is that Table 1: Deviation from proportionality in the shares of seats in the House of the 1997 election Commons which the political parties Party % votes % seats deviations receive are quite different from their (1) (2) (1)-(2) shares of the popular vote in general Conservatives 31.4 25.7 – 5.7 elections. Further, the relationship Labour 44.4 65.4 + 21.0 between seat shares and vote shares changes markedly from one election Liberal Democrat 17.2 7.2 – 10.0 to the next. It is often said that the Scottish National Party 2.0 0.9 – 1.1 results are unfair between parties – Plaid Cymru 0.5 0.6 + 0.1 some parties get large returns in seats Referendum Party 2.7 0 – 2.7 for relatively few votes, others may Others 1.7 0.2 – 1.5 collect a significant overall vote, but Total 100 100 receive very few seats at all. But as Total deviations (ignoring + or – signs) 42.1 the Jenkins report states, ‘the major “unfairness” count against FPTP is DV score = total deviations/2 21% that it distorts the desires of voters’ Largest deviation (for Labour) 21% (para 6). We simply subtract the percentage Deviation from proportionality in of seats a party gained in the Com- 1992 and 1997 mons from its percentage vote share In institutional terms, the British sys- to give a deviation for each party. tem is not ‘proportional’. But how Then we add up the deviations for all exactly should we measure ‘propor- parties (discarding their plus or mi- The Politico’s Guide to Electoral Reform in Britain + 9 nus signs, which would otherwise when, as the Liberal-Social Democrat mean that they cancel each other out) Alliance, it gained 26 per cent sup- and divide by 2. This gives a devia- port in 1983. tion from proportionality (DV) score The British deviation from propor- of 21 per cent for the 1997 general tionality score has been among the election. largest recorded amongst liberal de- This figure can be simply under- mocracies for the last 25 years. In stood as the fraction of MPs who are western Europe, proportional repre- not entitled to their seats in the legis- sentation systems commonly achieve lature in terms of their party’s national scores of 4-8 per cent – a level only share of the vote. Under a pure pro- briefly recorded in Britain during the portional representation system, then, two-party era of the 1950s. Similarly over one fifth of seats would switch to in the USA, where there is a perfect a different party – a score which has two-party system in Congressional been much the same in most elections elections, the deviation from propor- since the mid-1970s, when substan- tionality is very stable at around 7 per tial Liberal and other third party cent. So the British system is broadly voting became an established feature three times worse at translating votes of British politics. The DV score for into seats accurately than the main the 1997 Labour landslide is slightly countries against which we tend to below the 1983 deviation from propor- measure our democracy. The major tionality score of 23 per cent, when countries which still achieve high Margaret Thatcher won a large ma- deviation scores like Britain’s are jority over Michael Foot’s divided former imperial territories which re- Labour party. tain plurality-rule elections, especially The almost unique feature in Ta- Canada, Malaysia and India. ble 1, however, is that Labour’s gains However, even the deviation from (its huge ‘winner’s bias’) did not come proportionality score for Britain as a solely from third parties, but in large whole does not tell the whole story. part also from the under-representa- This figure is almost always mislead- tion of the Conservatives – only the ingly low if compared with other third time since 1918 that they have countries, because areas of pro-Con- obtained fewer seats than their share servative deviation in the south-east of the votes, the other occasions be- are partly offset by areas of pro-La- ing in 1945 and (marginally) in 1966. bour deviation in Scotland and the The Liberal Democrats were the most north. In 1992, the national DV score under-represented party in 1997, but was just 17.4 per cent, but far higher in fact they did relatively well by com- scores than this were common in most parison with the past. In most regions. Across south-east England elections since 1970, the party has the Conservatives won 97 per cent of achieved only 3 per cent of seats, seats in 1992 on the basis of 55 per whatever its share of the vote – even cent of the votes (leaving all other 10 + The Politico’s Guide to Electoral Reform in Britain The Politico’s Guide to Electoral Reform in Britain + 11 12 + The Politico’s Guide to Electoral Reform in Britain parties virtually unrepresented). So slumped from 43 per cent in 1992 to the deviation from proportionality in just over 31 per cent – and the plural- the south-east was 43 per cent – just ity-rule system tends to discriminate about as high as it is possible to get heavily against parties whose support inside a liberal democracy (see 1992 falls below about a third. In 11 out of map). In 1997, the DV score was the 18 regions we used in our 1997 higher than the national level of 21 election analysis, the Conservatives per cent in 12 of 18 regions (see 1997 fell badly below the 33 per cent mark, map). In central Scotland, another and they were ahead of Labour in only very high DV score (42 per cent) re- three regions (south-east and south- flected the Labour party’s unfair west England and East Anglia). political domination (although with A further sign of their crisis was the fewer seats at stake than in south-east growth of regions where the Con- England). servatives gained no MPs at all (as in Thus British voters experience an Scotland and Wales) or hardly any electoral system which is far more MPs (as in all the great urban areas unfair than the national figure would of England). For the Jenkins report, suggest – on average, the votes of these ‘electoral deserts’ represent a more than one in four voters (28 per major failing of first past the post elec- cent) did not count in 1992 when it tions. The report points out that came to the allocation of seats in the Labour experienced a similar fate in House of Commons. In 1997, the ero- the 1980s, being excluded from the sion in Conservative voting and the more rapidly growing and prosperous Liberal Democrat breakthrough in southern half of the UK. South of a south-west England reduced high DV line from the Wash to the Severn es- scores in critical southern areas where tuary, there were only three Labour there is a large number of seats, but seats outside London in both 1983 and still nearly one in four voters (23 per 1987; and no predominantly rural cent) found that the electoral system constituencies in England elected a ignored their votes in allocating seats. Labour MP. The report is highly criti- Only in south-west England did first cal of the ‘geographically divisive’ past the post deliver reasonably pro- effects of FPTP, commenting that ‘such portional results – a surprising apartheid in electoral outcome is a outcome given the fairly even three- heavy count against the system which way split of votes in that region in produces it. It is a new form of 1997. Disraeli’s two nations’ (para 31). It is the same properties of FPTP Electoral deserts which tend to make it hard to allow Scrutiny of regional voting patterns third party support to express itself. in 1997 reveals one major element in As we have shown, the rise in elec- the poor electoral performance of the toral support for the Liberals and their Conservatives. Overall, their vote successor parties which has de-stabi- The Politico’s Guide to Electoral Reform in Britain + 13 lised plurality-rule elections since the wanted government. 1970s has not been rewarded with an ● It offers unorthodox MPs a degree equal rise in their representation in of independence from excessive party Parliament. Plurality-rule elections control, provided (as many do) that work in two-party political systems, they can retain the support of their as in the USA, but Britain has now local party. ceased to be a two-party system. By 1974, the low Liberal shares of the Its deficiencies (or vices) derive, the report states, from its natural ten- vote common in the 1950s heyday of two-party politics had grown to nearly dency to disunite rather than to unite 20 per cent of the popular vote, but the country. The report lists the following vices: the Liberals still won only 2.2 per cent of the seats in the House of Commons. ● FPTP exaggerates movements of In 1983, the Alliance got only 3.5 per opinion and, when they are strong, cent of the seats after winning 26 per produces mammoth majorities in the cent of the vote. Even in 1997, with House of Commons (for Labour in all the benefit of tactical voting, it still 1945, 1966 and 1997; for the Tories, got only 7 per cent of the seats for in 1959, 1983 and 1987). Landslide nearly 17 per cent of the vote (see Ta- majorities do not, in general, conduce ble 1). to the effective working of the House of Commons. The virtues and vices of FPTP ● Recent large majorities have been The Jenkins report seeks to summa- secured with smaller percentages of rise the virtues and vices of the the popular vote in 1987 and 1997 plurality-rule system. The report lists than in the 1940s and 1950s, largely its virtues as follows: because third parties have taken ● It is said to be familiar and simple larger shares of the vote. to use. ● Third parties are however grossly ● It gives each MP a direct relation- under-represented in Parliament un- ship with a particular geographical less they have a relatively narrow area and encourages them to try to focus, like Plaid Cymru and (less serve all their constituents well, how- markedly) the SNP. Thus perversely, ever partisan they may be at third parties with a broader appeal – Westminster. a ‘favourable factor from the point of ● It usually (though not invariably) view of national cohesion’ – are heav- leads to single-party government and ily discouraged. this outcome may be seen as assist- ● FPTP creates ‘electoral deserts’ (see ing quick decisions and sustained above). policy lines. ● FPTP narrows the terrain over ● It enables the electorate sharply which the political battle is fought, by and cleanly to rid itself of an un- creating an essential election contest 14 + The Politico’s Guide to Electoral Reform in Britain in about 100-150 marginal in normal substantial list of deficiencies. This circumstances. Many voters in ‘safe’ approach underpins its essential case seats may thus pass their entire adult for AV-Plus – which is presented as a lives without ever voting for a winning compromise between FPTP and a candidate or even influencing a result. more pluralist and proportional alter- This has a knock-on effect on turnout native. The compromise approach is at elections. built into the very criteria which ● At local level MPs are increasingly framed the Commission’s inquiries: returned to Westminster on a minor- the criteria of ‘stable government’ and ity vote. In the 1950s, some 14 per cent the ‘constituency link’ serve as a po- of MPs won their seats on less than litical code for the current system; and 50 per cent of the local vote. In the ‘voter choice’ and ‘broadly propor- two 1990s elections, the figure has tional’ for more pluralist systems. risen to 44 per cent – nearly half of all But the Commission’s attempt to MPs. provide a balance of argument for and against plurality-rule means that the ● There is ‘some, but not overwhelm- report fails to subject the virtues which ingly strong evidence’ that FPTP is it lists to thorough analysis – though less good at producing parliamentary they are noted elsewhere in passing representation for women and for eth- – and curiously understates the ma- nic minorities than are most jor structural fault of FPTP under proportional systems. contemporary British conditions. This The perverse effects of FPTP are fault renders it unfit for further serv- proliferating fast (see Political Power ice. As we shall argue, like the royal and Democratic Control in Britain, yacht Britannia, our electoral system Stuart Weir and David Beetham, is obsolete. Routledge 1998, pp. 54-5). The First, the MP’s constituency link. Jenkins report cites two – in 1951, There is no doubt about the huge in- Labour lost the election even though crease in correspondence between they out-polled the Conservatives and MPs and their constituents, both over actually won a majority of the popu- political issues nationally and locally, lar vote; in February 1974, the and constituents’ own problems. But Conservatives won most votes but the closeness of the MP-constituency Labour took power with more seats. link is usually exaggerated and the Jenkins report tends to take it at face The Jenkins arguments in value. Opinion polls suggest, for ex- perspective ample, that only about half the The Jenkins report is clearly anxious population can name their MP and a to appease pro-FPTP sentiment in its major study, published in 1992, found summary of the system’s virtues and that only about one in ten people had vices, while concluding that the case contacted their MP in the previous for it has to be tested against a very five years. Considerably more people The Politico’s Guide to Electoral Reform in Britain + 15 take their problems to local council- England means that the local and wel- lors. Further, the idea that MPs are fare roles of MPs is set to diminish. dependent on their constituents rather None of this is meant to deny the than on their parties is a political value of a local link for MPs, but sim- myth. Party loyalty by MPs to their ply to warn against giving it undue party, in government or opposition, is prominence and to signal the need to the keystone of their political role and ensure that MPs perform their wider determines all or most of their con- duties more effectively. Here Jenkins duct. does usefully suggest that top-up MPs The role of MPs as ‘problem-solv- might take a greater interest in the ers’ has not been closely studied. It is scrutiny of legislation, which is noto- not known whether they provide an riously badly performed by existing effective service to the minority of constituency MPs. people who approach them. But it is The Jenkins report does question almost certainly patchy, and probably the idea that single-party government less effective than that which ex- is the traditional norm in British poli- panded and more accessible tics and considers evidence from a few Ombudsman services, as well as a chosen countries, notably Ireland and well-resourced network of Citizen’s Germany, which can be shown to have Advice Bureaus, could provide. achieved stable government and no- There are other considerations too table economic growth in the postwar which arise from confusions about the period. So far as tradition is con- indeterminate role of MPs. The evi- cerned, the report points out that for dence is that backbench MPs are 43 of the past 150 years, Britain has grossly over-stretched and many turn been governed by overt coalitions; for to ‘constituency work’ as a tangible another 34 years the government of satisfaction in a badly-defined and the day depended on the votes of one often frustrating career. Former La- or more other parties; and for another bour MP John Garrett suggests that nine a government technically hold- whips, ministers and civil servants ing an overall majority actually had encourage MPs to act as local advo- an uncertain command over the cates because they know that House, the most recent example be- constituency overload can drive out ing John Major’s 1992-97 persistent scrutiny; and critics of the government. Thus, for more than half Commons, such as Kate Jenkins, are of this period, Britain has not been scornful anyway about the ability of ruled by the traditionally strong sin- MPs properly to perform their duty to gle-party government which FPTP make the executive accountable. seeks to produce; and the Commis- Finally, the advent of devolved gov- sion argues that British history shows ernment in London, Scotland, Wales that single-party government is not a and Northern Ireland, and at least the necessary prerequisite for effective prospect of regional government in action. 16 + The Politico’s Guide to Electoral Reform in Britain However, the report fails to chal- electorate that actually makes – and lenge the strongly-held idea that unmakes – governments in the UK. coalition government is necessarily Mrs Thatcher’s governments in the weaker and less desirable than single- 1980s, all-powerful politically, were all party government, to subject it to returned on minority votes of some- rigorous analysis, and to review the thing over 40 per cent in turn, and experience of coalition government in ‘unwanted’ by nearly 60 per cent of more than a few western democracies. the electorate. The fact is that that We briefly do so in the next section of the desires of the electorate are dis- this report. torted in the first instance by the There are two further weaknesses unrepresentative nature of the results in the Jenkins approach. The first is of the electoral system itself, and then that, by default, it does not fully ex- further distorted by the supreme amine the claim that is often made power which the political system con- that the British electorate has direct fers upon the leading party (or not) control over the process of forming once it has won its majority. governments in the United Kingdom, Finally, it ought to be acknowl- and endorses the view that the elec- edged, as US politics constantly toral system ‘enables the electorate reminds us, that first-past- the-post sharply and cleanly to rid itself of an elections can deliver very reasonably unwanted government’. The weak- proportional results time after time. In ness of both claims is that the the United States, only 7 per cent of ‘electorate’ evidently does no such members of the House of Representa- thing. Certainly voters do have a di- tives are elected for parties who are rect say in the formation and dismissal not entitled to their seats in terms of of governments, but it is a minority of their share of the national vote (com- voters which performs the first func- parable to the best European PR tion, and a government can continue systems, which hit between 4-8% on in office against the wishes of the the same ‘deviation from proportion- majority at election after election. The ality’ score). And virtually all US ‘electorate’ is a political phantom. On Representatives enjoy clear majority the only occasion on which a major- support in their districts. But the USA ity of the electorate voted in postwar is the only large country where first- Britain for a particular party to hold past-the-post elections work in this power, in 1951, that party lost the elec- way, because it is the only perfect two- tion and the party with fewer votes party system in the world. Everywhere was returned to power. Could it really else in liberal democracies party sup- be said with any degree of truth or port is fragmenting over time, and logic that the ‘electorate’ had spoken? first-past-the-post elections cannot Thereafter every government has cope. been elected on a minority of the In Canada the system is now dan- popular vote: so it is a minority of the gerously erratic, projecting the The Politico’s Guide to Electoral Reform in Britain + 17 Conservatives at the last election from parties obtained significant vote having a Parliamentary majority to shares. holding just three seats. In India first- Most media commentators and past-the-post means that party seat many academics are blind to the new shares also yo-yo dramatically with structure of British politics. In their small shifts in votes, and enormous minds politics is still a two-party af- local and regional corruption has fair. Like first-past-the-post itself, they been stimulated by the strong elec- treat all the voters for third parties toral insulation of MPs. In Malaysia with contempt. Liberal Democrat ar- the system supports a regime where guments for fairness for their voters executive power has become unsta- and party are dismissed as special ble and civil rights are in jeopardy. pleading. But they will have to open And these are now the only substan- their eyes and minds sooner rather tial countries that still use the British than later. For the former Roy system. Jenkins’s ‘mould’ really is broken. The In Britain, as we have seen, the structural changes which prevent the disproportionality score is commonly system from more accurately reflect- three times higher than in the USA; ing the choices that people actually and only just over half of all MPs en- make are not going to go away. The joy majority support from their consistent voting patterns and trends constituents. These are poor levels of of the last 28 years will not suddenly electoral legitimacy. But there is a go into reverse. We live in a post-mod- strong structural reason for these ma- ern age, and the former two-party jor deficiencies. Since 1972, politics of Britain will go on being frac- thousands of opinion polls, 26 years tured – indeed, the fracturing will take of municipal elections, four Euro- worse forms if voting reform is de- elections, and seven successive layed. If the electoral system stays general elections have shown one fifth unchanged, it can only be a matter of or more of the vote going to third and a few years after 2001 before the pro- fourth parties – to the Liberal Demo- portionally elected Scottish crats, to the SNP in Scotland and Parliament will so dominate electoral Plaid Cymru in Wales, or to the legitimacy in Scotland that the Com- Greens (notably in the 1989 Euro elec- mons is reduced to a farce in Scottish tions). In 1997, a record-breaking 4.4 eyes, and the ratchet for Scottish in- per cent of the vote even went to fifth dependence will take a further and sixth parties (such as the Refer- powerful turn. endum Party and the UK In the last 15 years Britain has be- Independence Party). And in North- come a more middle class society; ern Ireland, the previous umbilical years in school have lengthened; and connection to the mainland party sys- the numbers of graduates have soared tem has been completely severed: in – all factors that used to predict the PR 1998 Assembly elections 12 greater election turnout. But overall 18 + The Politico’s Guide to Electoral Reform in Britain turnout has stubbornly not increased – instead it plunged in the 1997 general election by six percentage points to a postwar low. Voting in the 1998 municipal elections fell back by a staggering 10 to 15 percentage points right across the country, and dramatically in Labour heartlands. The underlying propensity to vote in Brit- ain is in decline, with serious effects in inner-city areas. Turnout level will not easily be rebuilt, and could all too easily spiral further downwards. Jenkins identifies the elements of the structural causes of the higher levels of distortion in Britain’s elections to Parliament, but fails explicitly to nail his significant proposals for reform to a structural shift in British politics which will not fade away.

The Politico’s Guide to Electoral Reform in Britain + 19 COALITION GOVERNMENT

ere we briefly review the on electoral and political effects which general state of knowledge are relevant to the Jenkins choice, AV- Habout single party and coa- Plus, which is for example deliberately lition government, focusing on the designed to minimise central party arguments that coalition governments control over the choice of candidates. deprive electors of effective choice of and control over government and are The idea of the party mandate less effective than single-party gov- A major justification for FPTP in Brit- ernments. The trouble is that debate ain is that single-party governments, in Britain assumes that there is a sharp even though elected on a minority distinction between single party and vote, can deliver on the programmes coalition governments; and that pro- that they put to the electorate because portional representation and coalition their artificial majority in the House government are a uniform political gives them the power to deliver on phenomenon. Thus, opponents of PR their mandate. The party mandate can argue both that coalition govern- thus offers a mechanism for linking ments are inherently unstable (citing electoral preferences to government Italy) and too stable (citing Germany). action through the central party role The fact is that there are many differ- in both. We should note the positive ent kinds of coalition in western side of mandate arrangements, even Europe alone. Some are more effec- when they do rest on a plurality rather tive than others. Some emerge clearly than a majority of the vote. They have from electoral verdicts, others do not. two interrelated strengths: There are general truths about ● electors know what they are voting most proportional systems. By their for and can cast their vote so as to very nature, they generally reflect the enhance the chance of their preferred wishes of voters far more accurately party forming the government and than the British system does; and carrying through its programme those systems which employ party ● lists do generally give central party the election result does secure the organisations a greater degree of con- electoral choice of at least the largest trol over the selection of candidates minority among the electorate. than is considered appropriate in the However, the mandate idea under United Kingdom. But different sys- FPTP involves often unexamined de- tems produce different results and in ficiencies. The full operation of the this brief review we will concentrate mandate really demands that Parlia- 20 + The Politico’s Guide to Electoral Reform in Britain ment be totally subordinate to the gov- arrangements, though less so in frag- ernment formed by the majority party. mented systems with larger numbers All opposition parties can or should of parties. The most obvious case is do is try to rally popular support for where parties form explicit electoral alternative programmes in light of the alliances before the election. As the next election. Attempts to give Parlia- Social Democrats and Greens have ment real powers of investigation or just done in Germany, the allies pro- control subvert the idea of the man- claim their intention of serving date. But should Parliament not together in government if they win the consist of more than a venue for the election. Sometimes they even issue debate between government and op- a common ‘Programme of Govern- position, designed to influence the ment’ so electors know what next election? This idea is all the more government policies they are voting questionable because the government for. Thus, election alliances can sub- only has a plurality of votes – nowa- stitute quite effectively for single days as little as 42 or 43 per cent – parties, particularly if one alliance of which only the operation of FPTP parties is lined up against another, so transforms into a legislative majority. electors are able to choose between Thus, it can well be argued that a two clear-cut opposing alternatives. popular majority has voted against the Such alliances have been common in party which forms the government – Germany, where the Free Democrats and the mandate which it is empow- have generally formed an alliance ered to push through Parliament. with one or other of the two larger By definition, where parties can parties, the Christian or Socialist rarely hope to form a single-party gov- Democrats. They also occur in Ire- ernment, as in most countries with PR land, though not so frequently, with elections, the parties with different Fine Gael and Labour allied against programmes coming together after an Fianna Faìl. In Sweden and Norway election to form a coalition do not the ‘bourgeois’ parties usually state seem to have a mandate in the same their intention of forming a govern- sense, even if they can agree on a ment together if they get a majority. common programme (we shall see In other countries, all the parties they often can). However, the major- of the left join forces against an alli- ity often seems ‘manufactured’ in the ance of all the right-wing parties, with sense that one or even all of the part- the intention of producing either a left ners may have lost votes in the or a right government depending on election and still form a government; which tendance gets a popular major- and also in the sense that it may be ity. In this case, electors have a choice the product of unseen political deal- between left and right priorities, ings. though it is not always clear exactly Yet most countries using PR sys- which parties will be in and out of tems do have something like mandate government. In Norway and Sweden, The Politico’s Guide to Electoral Reform in Britain + 21 the contest is essentially between the choice, than the classic mandate bourgeois, centre-right alliance, and theory – particularly as the coalition the left, represented by Social Demo- will come closer to representing the crats or Labour, the large party, and choices of voters from parties ex- small left-socialist or Communist par- cluded from government than ties. If the left gets a majority a single-party plurality rule. Further, single-party Social Democratic or La- formal electoral alliances of two or bour government will form, supported more parties do offer electors a clear by the smaller left party. The outcome choice between two different pro- of the vote is blurred a little by the grammes and usually produce a question of which bourgeois govern- government with a genuine majority. ment will form in the case of a This is secured, however, by party centre-right majority – which not may strategies and behaviour, not by for- not always include all the allied par- mal constitutional arrangements ties. Further research by political scien- An academic study of the influence tists like Professor Ian Budge, of the of electors over the making of govern- University of Essex, has shown that ments between 1950-90 in 16 nations coalitions do not make it harder for found that electors determined the parties to keep their commitments, formation of nearly every government owing to agreements among the part- in countries using plurality-rule elec- ners which let each pursue their own tions – the UK, USA, Canada, differing priorities. A comparative Australia and New Zealand (to 1994). study in ten liberal democracies of the But they also determined the making relationship between election priori- of most governments in Ireland, Ger- ties and government spending found many, Sweden, Norway and (just) that coalitions in Germany, Sweden Denmark and exercised partial con- and Austria were more likely to stick trol over most others. closer to their manifesto priorities On the positive side, coalition gov- than single-party governments in Brit- ernments are likely to include the ain, Australia and the USA, while also ‘middle party’ which represents the satisfying a broader section of society ‘average’ elector, and to avoid putting than UK governments are able to do government in the hands of a fairly (see further Stability and Choice: a extreme plurality, which the majority review of single party and coalition might even be said to oppose. The government, by Ian Budge, Demo- problem for FPTP in Britain, based on cratic Audit Paper No. 15, 1998). a plurality rather than a real majority AV-Plus is designed to avoid some of the popular vote, is that it cannot of the disadvantages of coalition poli- guard against this happening (as in tics, as seen from a British perspective, the 1980s). Thus coalitions are gen- though as we have seen, these can be erally more likely to satisfy a wider greatly exaggerated. The fact that it concept of representation, or voter is partially rather than fully propor- 22 + The Politico’s Guide to Electoral Reform in Britain tional means that coalition govern- growth figures for the United King- ments will be rarer in the UK than in dom and the USA and broadly western Europe. But this brief review comparable countries in Europe suggests that British voters have little (France, Germany, Italy) which have to fear from coalition governments coalition governments, for the period and something to gain. The partners 1960-90. The growth figures are in coalitions here are likely to seek to shown before and after the oil shocks honour individual party programmes, of the mid-1970s. What they reveal is or pre-election agreements; any coa- common knowledge. Britain’s growth litions are likely to reflect a majority rate lagged behind the coalition coun- of voters; and the prospect of parties tries in the earlier period, as did that with extreme views gaining power of the USA. After the oil shocks the will diminish. figures became more equal, but there is no sharp divide between countries How effective are coalition with different forms of government. governments? Britain and France parallel each other National well-being is hard to meas- as do the USA and Italy. German ure directly. It is almost impossible to growth temporarily slowed down but trace out exactly what is due to gov- jumped again at the end of the dec- ernment action and what is due to ade (4 per cent in 1989). Growth in other, often structural and institutional GDP hardly decides the matter of factors. So in talking about the gen- course. Various ‘quality of life’ meas- eral effectiveness of coalitions we can ures, recording the adverse affects of only offer relevant but not conclusive growth, show Britain performing bet- evidence. Where this evidence is most ter than Japan but hardly better than convincing, however, is in destroying Germany and France. any idea that there is a prima facie The general point is, however, that case against the effectiveness of coalitions in general. Indeed, if there is a Table 2: Rates of Growth of GDP in prima facie case to be made, it is Comparable Countries with and without against the general effectiveness and Coalition Government, 1962-88* efficiency of single-party govern- Average GDP rate of increase for: ments. Erring on the side of caution, 1962-1972 1977-1988 however, it is probably best to say that France 4.7 1.6 there are only limited grounds for Germany 3.6 1.3 claiming greater effectiveness of one Italy 3.9 2.2 side over the other. UK 2.2 1.8 The most commonly used measure USA 3.0 2.3 of national well-being is economic growth, as measured by the annual *Annual Percentage Rates of Increase in Gross Domestic Product) increase in Gross Domestic Product. Source: Ian Budge, ‘Relative Decline as a Political Issue’, Table 2 presents average annual Contemporary Record, vol 7, No. 1, (Summer 1993), p. 5. The Politico’s Guide to Electoral Reform in Britain + 23 even under a Thatcherite regime which concentrated the powers of single-party government to an unusual degree and focused on economic growth, Britain hardly stands out as exceptional. Even at best, it lags behind the country, Italy, whose coalitions could most justly be criticised for weakness and indecisiveness. On the general historical record, too, postwar British governments do not appear as more generally success- ful than German – or indeed Italian – governments which in two decades brought the country out of occupation and defeat to prosperity and military security. Broadening the comparison, one could not say that Scandinavia, often under minority governments or coalitions, offers less quality of life than Britain; or the Low Countries either. Such general comparisons are about as far as we can take the matter and must be severely qualified. They certainly do not indicate, however, that coalition governments are less effective than single-party governments.

24 + The Politico’s Guide to Electoral Reform in Britain AV-PLUS – THE BASICS OF THE JENKINS SCHEME

e go on in the next section, last two elections, and compare the ‘How AV-Plus would have results of the Jenkins Commission’s Wworked in 1997 and 1992’ choice, using the alternative vote (AV) to test the Commission’s scheme un- for elections of constituency MPs, der the quite different political with the results of using plurality- conditions of 1997 and 1992. This sec- rule, or first past the post, elections to tion of the guide explains the basics choose them. of the scheme and how the local and The three top-up area schemes ex- top-up seats are distributed in Great amined are: Britain, and goes on to describe the ● Scheme A, with 112 top-up seats in methodology which has been em- Great Britain (17.5 per cent of the to- ployed to test the scheme, and a tal number of MPs); variant using FPTP for the local elec- ● Scheme B, with 128 top-up seats (20 tions. Much of the detailed per cent of the total) explanation of the methodology is fairly technical in nature. We set it out ● Scheme C, with 96 top-up seats (15 here because we believe in transpar- per cent of the total). ency, but it may be passed over by The three schemes use the same set readers who are not expert in such of 78 top-up areas in Great Britain to matters. The section as a whole is group together Westminster constitu- based on the evaluation of the Com- encies for the purposes of allocating mission’s final AV-Plus scheme one or top-up MPs. (We discuss the presented to the Commission by special problems raised by Northern Patrick Dunleavy and Helen Ireland below; see pp. 55-8.) As the Margetts, its academic consultants. final scheme was defined very late on in the Commission’s processes, we Constituency and top-up seats present data for the FPTP elections in under the Jenkins scheme slightly different and older versions of The Commission decided finally to schemes A, B and C. The differences present the government with three in terms of seats for the different par- alternatives of the same basic scheme, ties are very small. But readers should varying only in the ratios of constitu- note that the allocation of top-up seats ency to top-up MPs. The three options between England, Scotland and use the same basic structure of areas. Wales varies slightly more between We test them under the quite differ- the final and earlier versions of the ent conditions of 1992 and 1997, the three schemes. The Politico’s Guide to Electoral Reform in Britain + 25 The top-up areas, as defined by the Table 3: The distribution of top-up areas Commission, are: by total number of MPs in each area, Great Britain and Northern Ireland 1. Counties in England outside met- Total MPs in No. of top- % of top- ropolitan areas. Four very large top-up area: up areas up areas counties – Kent, Essex, Lancashire Four 1 1.3 and Hampshire – are split in half. This Five 6 7.5 step simplifies the size range of top- up areas, and very importantly avoids Six 10 12.5 creating too large a variation between Seven 12 15.0 top-up areas in the threshold levels Eight 15 18.8 at which parties would win seats, a Nine 13 16.3 feature otherwise likely to have a se- Ten 10 12.5 verely distorting effect on party Eleven 12 15.0 behaviour) Twelve 1 1.3 2. Sub-divisions of London and the All areas 80 100.2 % metropolitan counties, which fit with district or borough boundaries, so far only four MPs, including one top-up as possible. MP; and another county (Stafford- 3. The existing top-up areas to be used shire) with 12 MPs in total, including in the Scottish Parliament and Welsh two top-up MPs. The Commission Assembly elections in 1999, which could have removed these discrepan- correspond to the Euro-constituency cies, by merging Northumberland boundaries recommended by the re- with a neighbouring county (such as spective Boundary Commissions for Durham, creating another 11-seater) the two countries in their thorough and splitting Staffordshire up into two 1996 reports. top-up areas (with six seats in each). But it finally decided to tolerate such The resulting areas (with two ex- minor anomalies in order to maintain ceptions) all contain at least five historical areas. current parliamentary constituencies, The basis for allocating top-up and no more than 11 constituencies. seats in all three schemes is that ar- Table 3 shows the number of top-up eas with fewer Westminster areas in the Commission’s schema by constituencies have only a single top- the number of existing MPs. The most up MP, while larger areas are common total of MPs each areas allocated two top-up MPs. The differ- would have is eight (which is also the ences between the A, B and C median size). schemes arise simply from the thresh- The distribution around the me- old at which a top-up area qualifies dian is fairly symmetric, except for a for two top-up MPs. The allocation small bulge of 12 areas with 11 MPs rule is: each. There is a single historic county in the scheme (Northumberland) with (i) Calculate the number of top-up 26 + The Politico’s Guide to Electoral Reform in Britain Table 4: Top-up MPs and areas with two top-up seats across Great Britain

Scheme A: 112 seats

All seats Top-up Top-up Two-seat Top-up seats areas seats top-up areas as % of all GB 641 78 112 34 17.5

England 529 65 92 27 17.4 Scotland 72 8 13 5 18.1 Wales 40 5 7 2 17.5

Scheme B: 128 seats

All seats TU areas TU seats Two-seat TU seats as TU areas % of all GB 641 78 128 50 20.0

England 529 65 106 41 20.0 Scotland 72 8 14 6 19.4 Wales 40 5 8 3 20.0

Scheme C: 96 seats

All seats TU areas TU seats Two-seat TU seats as TU areas % of all GB 641 78 96 18 15.0

England 529 65 79 14 14.9 Scotland 72 8 11 3 15.3 Wales 40 5 6 1 15.0

MPs needed in each component coun- ured by number of electors), until all try of the UK to maintain a constant available top-up seats in that country ratio of top-up to total MPs, (20, 17.5 have been allocated. or 15 per cent according to which Thus, under scheme A the 27 top- scheme is adopted); up areas in England with most electors are assigned two top-up seats (ii) Assign one top-up seat to each each, plus the five most populous ar- top-up area in each country; eas in Scotland and the two most (iii) Assign a second top-up seat to Welsh areas. Table 4 shows how the the most populous top-up areas in number of top-up seats would vary each country, ranked in descending across the three countries in Great order of their population size (meas- Britain and across the different The Politico’s Guide to Electoral Reform in Britain + 27 Table 5: The distribution of top-up areas representative because of the small by size of electorate number of cases.) Electors per top-up area: No. of top-up areas Calculating the effects of AV-Plus Below 300,000 people 1 The calculations conducted here use 300000 to 399,000 people 11 data from the 1997 and 1992 election 400000 to 499,000 people 19 to project forward results for the Com- 500000 to 599,000 people 21 mission’s chosen voting scheme. 600000 to 699,000 people 11 However, readers need to be aware of 700000 to 799,000 people 13 the simulation methods used and Over 800000 people 2 some of their key limitations. We wanted to compare the effects of the schemes. The size of the Commis- Commission’s AV-Plus scheme with sion’s top-up areas in terms of those of an equivalent AMS scheme population size is shown in Table 5. – that is, the Jenkins scheme using The distribution of top-up areas by the FPTP in local elections rather than AV. size of the electorate has a peak of 21 (The Jenkins scheme belongs broadly areas between 500,000 and 600,000 within the AMS family of electoral electors, tapering off fairly symmetri- systems.) This modified version of cally to either side. Only one top-up AMS would be the scheme which the area (Northumberland) has fewer dissenting Commission member, Lord than 300,000 electors, while two (Staf- Alexander of Weedon QC, was in ef- fordshire and Devon) have rather fect recommending in his ‘note of more than 800,000 electors. reservation’. The average (mean) number of However, our data from 1992 and electors per top-up area is 546,000 1997 are only completely accurate for across Britain (for further information, a ‘classic’ AMS scheme (with half the see the LSE/Birkbeck report, The Per- seats local and half top-up MPs) formance of the Commission’s which pairs existing Westminster con- Schemes for a New Electoral System, stituencies. Since the Commission’s by Patrick Dunleavy and Helen scheme involves retaining local seats Margetts, LSE Public Policy Group, for five out of six seats (or four out of 1998, p. 10). However, the median fig- five) – an utterly different proportion ure (at 525,000) is probably a more – our approach has been as follows: reliable guide to average sizes over- 1. We defined a set of local seats for all, since it is not distorted by higher ‘classic’ AMS within the Commis- numbers in a few areas. The mean fig- sion’s 78 top-up areas, and projected ure in England is slightly larger at an outcome in terms of both local and 560,000, while the means for Scotland top-up seats for each such area. Note and Wales are considerably lower at that the AMS pairings used here are 493,000 and 440,000 respectively. (In distinctive because the constituencies these two countries the median is less have had to be paired so that all pair- 28 + The Politico’s Guide to Electoral Reform in Britain ings fall within the 78 top-areas. The the number of MPs per area and the AMS pairings here are quite differ- point where two top-up seats limit is ent from those used in Making Votes placed. Table 6 (over) shows that both Count 2: Mixed Electoral Systems in top-up areas with seven or fewer (Democratic Audit Paper No. 14, MPs, and in areas with ten or more 1998). The pairs of seats there only MPs, the top-up MPs’ share of all MPs had to fit within 11 or 18 broad regions is constant across the three schemes. used to allocate top-up MPs. But in top-up areas with eight or nine 2. We established the difference be- MPs this share fluctuates sharply tween each party’s local seats in the across the schemes. In Scheme A, the 100 per cent local scheme (that is, the share of top-up MPs varies from 12.5 general election constituencies) and to 22 per cent; in Scheme B, from 14- in the 50:50 scheme. From there we 25 per cent; and in Scheme C. from computed a marginal increment (or 11-20 per cent. decrease) in local seats which each 4. For the actual proportion of local party in each top-up area would re- seats in each top-up area under the ceive, as the proportion of local seats Commission’s scheme we then inter- in that area grows from 50 to 100 per polated a local seats projection for that cent in single percentage steps. This scheme. An essential assumption here marginal increment stage is key be- is that there is a linear relationship cause in some top-up areas with odd between changes in the proportion of numbers of total seats, our ‘classic’ local seats used under various AMS AMS solution will in fact involve schemes and in the seats won by par- fewer than half of seats being local ties. If this relationship is not in fact seats. For instance, in a seven-mem- completely linear, some distortions ber area, we had to form two AMS will arise in our estimates. pairs of seats, plus an AMS triple, giv- 5. In a number of cases the model ing only three local seats.* makes conflicting arithmetical predic- 3. A small additional complication is tions of the number of seats that would that the percentage of all seats in each be won by two different parties in a top-up area varies considerably in the single top-up area – for instance, re- Commission’s schema, depending on quiring that two parties win the same

* For technical readers, a detailed listing is available from LSE which provides a details of all the current Westminster parliamentary constituencies in each of the top-up areas. This listing also gives a column for an ‘AMS pairing’, indicating which constituencies have been paired together to compute the ‘classic’ AMS local seats data. There are also a fair number of cases where three constituencies have had to be joined here to fit within the top-up area boundaries. We have tried to pair socially similar constituencies and to preserve so far as possible a diversity of party representation in the paired local seats. But it is important to bear in mind that these are analytic pairings for research purposes only. The actual local seat boundaries which would be set under the Commission’s scheme in action will not resemble these pairings, but be much closer to the 100 per cent local scheme seats in use at present. The Politico’s Guide to Electoral Reform in Britain + 29 local seat. These cases have to be re- go to very conspicuously under-rep- solved judgementally, making resented parties. Note that the impact reference to the geographic pattern of the top-up seat allocations still and extent of party support in that top- tends to correct for any substantial up area in the base election year, and biases in estimating local seats distri- the fit between these variables and the butions – because in the end AMS likely AMS constituencies under the seats must get as close to matching Commission’s scheme in that top-up vote shares as the number of available area. top-up MPs allows. 6. Having estimated local seat out- Our procedures for simulating AV- comes under the Commission’s Plus outcomes are essentially the scheme, we then allocated the one or same as for a Jenkins version of AMS, two top-up seats in each area using with one significant difference. We the d’Hondt allocation rule (already used data from our 1992 and 1997 legislated for use in the Scottish Par- surveys on how people filled in AV liament, Welsh Assembly and ballot papers across the regions of European Parliament elections). Great Britain (13 in 1992 and 18 in Again this method is slightly differ- 1997) to run AV contests in the exist- ent from our previous work (which ing Westminster constituencies, and allocated top-up seats using a generic in our AMS-paired constituencies. ‘minimising DV’ allocation rule). The These 100 per cent and 50 per cent distinctive impact of the d’Hondt rule local seats outcomes under AV were is relatively restricted, however, be- then substituted in the simulation, cause there are so few top-up seats in instead of the plurality-rule results each area, and these seats normally used with AMS.

Table 6: Top-up MPs as a percentage of all MPs in a top-up area across different sizes of top-up area

Total number of MPs Number of top- Top-up MPs as a % Comment in a top-up area up seats of all MPs in the area Five One 20% All schemes Six One 17% All schemes Seven One 14% All schemes Seven Two 28% A few areas, varies by scheme Eight One 12.5% Some areas, varies by scheme Eight Two 25% Some areas, varies by scheme Nine One 22% A few areas in Scheme C Nine Two 22% Most areas, all schemes Ten Two 20% All schemes Eleven Two 18% All schemes

30 + The Politico’s Guide to Electoral Reform in Britain In estimating all these outcomes the 1997 constituencies, instead of we have worked directly from the con- going back to reconstruct the whole stituency votes in the two general constituency map as we did in Mak- elections. We have therefore arrived ing Votes Count and Making Votes at a whole top-up area vote under both Count 2: Mixed Electoral Systems. The AV-Plus and an AMS version of reason for this change is fundamen- Jenkins by aggregating up votes from tally logistical: Dunleavy and the component constituencies, since Margetts could not calculate the re- this data is much more fine-grain than sults for 1992 within the any we could generate from our sur- Commission’s 78 areas in any other vey responses. Nonetheless this is an way within the deadlines they were important limitation. In some of our working to. We have reservations work, notably a publication Devolu- about the accuracy of the re-calcu- tion Votes (LSE Public Policy Group, lated BBC/ITN data, but since the 1997) dealing with the likely results Commission’s schemes use 78 top-up under AMS of elections to the Scot- areas instead of 18 regions, there tish Parliament and Welsh Assembly, would also have been large and una- we compared simulation outcomes voidable inaccuracies in remodelling using actual votes only with outcomes the 1992 constituency pattern to fit obtained using more voting patterns these schemes (which of course are projected from our survey data. Some drawn up for the 1997 pattern of con- significant differences arise because stituencies). On balance, we do not of vote-splitting effects (See the sec- believe that using the BBC/ITN data tion, ‘Split-ticket tactical voting’ produces less accurate figures. We below). Under the Commission’s sys- have carefully cross-checked the re- tem, voters will have two separate sults here and in our published work ballot papers, and they may choose to and found no substantial divergences. vote differently at the constituency But readers should note that the ta- level and for top-up MPs. If highly bles of 1992 results add up to 641 organised or differential ‘tactical vot- constituencies, and not the 634 con- ing’ should emerge on a significant stituencies actually existing at that scale, there will be some variations time. from the results reported in the main Finally, it is important to stress that tables here The potential impact is the simulations in this guide provide greater with AV-Plus than with an only a snapshot picture of how the AMS variant. Commission’s schemes would have We have also used a simpler way operated under 1997 and 1992 condi- of modelling the results for the 1992 tions. They do not (and cannot) take general election than we did in pre- into account the dynamic effects of the viously published work. We have used introduction of a new system. How- the BBC/ITN data showing the 1992 ever, on a number of issues below, we results re-calculated on the basis of do cite evidence relevant to thinking The Politico’s Guide to Electoral Reform in Britain + 31 through such dynamic effects where produced identical seats outcomes to it is possible to measure their poten- AV, both within the current 641 West- tial in quantitative terms. But it is minster constituencies, and in our important to bear in mind is that dy- paired constituencies. In 1992, there namic effects will certainly introduce was a single constituency (Inverness, divergences from the estimates pub- Nairn and Lochaber) where the Sup- lished here. These estimates are plementary Vote would have stymied presented in precise quantitative a third-placed candidate who would terms, but we do not mean to imply have won under AV. There were two that an electoral system can be fine- other cases in 1992 where a similar tuned by design decisions. Actual effect came close to happening but in results under a new electoral system the end just did not, and one similar will reflect the infinite capacity of both ‘near miss’ in 1997. But the only clear voters and political parties to re-as- difference between SV and AV out- sess and change their behaviour in comes was the single seat change in response to new conditions. 1992. Essentially therefore, SV-Plus would perform just as AV-Plus would. The supplementary vote Since SV has however different quali- There is another electoral system, the ties from AV, and is regarded by some supplementary vote (SV), which is a experts as being the superior system, close cousin of the Commission’s we are not eliminating it from this choice for local elections, the alterna- survey of the Jenkins recommenda- tive vote. Like AV, SV is used for tions. In all subsequent tables in the elections in single-member constitu- guide, results quoted for AV-Plus will encies, but instead of numbering all apply also for SV-Plus. or some of the candidates in order of preference, voters simply make a first- preference and second-preference choice. Again, a candidate who gains more than half the first-preference votes cast is elected. But if no-one passes this threshold, then only the two top candidates remain in the contest and the second preferences of the eliminated candidates are distributed between the top pair. The surviving candidate with most votes is then elected, and of course second preference votes for eliminated candidates do not count. Our research has demonstrated that in the 1997 general election, SV 32 + The Politico’s Guide to Electoral Reform in Britain HOW AV-PLUS WOULD HAVE WORKED IN 1997 AND 1992

e now examine how the In 1997, Liberal Democrats gave the three versions of Jenkin’s largest share of their second prefer- WAV-Plus – schemes A, B ences to Labour, so that in and C – would have performed under constituencies where the Liberal 1997 conditions. This general election Democrats ran third, Labour was ad- is a difficult one for any electoral sys- vantaged and the Tories lost out – an tem to cope with for two reasons. First, effect apparent both in actual West- 4.4 per cent of the electorate voted for minster constituency contests re-run small parties with diminutive vote under SV or AV and in our paired shares, which are unlikely to secure AMS constituencies. In areas where representation under any electoral the Liberal Democrats ran second, system. This figures is twice as high they could generally rely on attract- as the British norm of around 2 per ing the bulk of third-placed cent for smaller parties. Second, the candidates’ second preferences, leading party (Labour) was 14 per- whether Conservatives or Labour. centage points ahead of the second Table 7 shows the overall outcomes party (the Conservatives), gaining in of the Commission’s ‘middle scheme’ effect 1.5 times their principal rival’s – scheme A with 112 top-up MPs (or vote and so greatly strengthening the 17.5 per cent of all MPs). The table possibility of a ‘leader’s bias’ effect. shows both the local and top-up seats Under AV-Plus, local members are the parties would have won under AV- elected under the alternative vote to Plus in 1997, and the parties’ total ensure that all constituency MPs win number of seats, for England and majority support in their locality. The English regions, Scotland and Wales. simpler version of AV, the supplemen- The table also shows the deviation tary vote (SV; see p. 32), achieves the from proportionality on a regional and same aim. In British politics, the ex- national basis. Note that the results tent to which either of these are grouped by countries and English majoritarian systems make a differ- regions, not by top-up areas (for ence compared with plurality-rule is which, see Table 9). Thus, in 1997, basically governed by the distribution Labour would have gained virtually of the second preferences of Liberal all their seats under AV-Plus at the Democrat voters (and to a lesser de- local level, the 357 local seats in Great gree, of voters for minor parties too). Britain being enough in fact to secure The Politico’s Guide to Electoral Reform in Britain + 33 18.8 18.3 11.7 11.3 19.5 16.9 12.9 15.4 19.7 15.3 04000300 154770170 184880074 9000067800 1214250051 90100 563922001177.9 21100 163940059 41300 81130022 71200 142640044 60000 114050056 72810036 5005050100 547911072 52843040 Local seatsCon Lab LD Nat Other Con seats Top-up Lab LD Nat Other Con Lab LD All seats Nat Other Total DV 277 223 24 5 0 39 17 50 6 0 316 240 74 11 0 641 8.7 Seats won under AV-Plus with 112 (17.5%) top-up in re-run 1997 election (Great Britain) Seats won under AV-Plus Table 7: Table South WestSouth East MidlandsWest East Anglia 3East Midlands 50 14 7 & Humb’sideYorks 32 25 39North 6 10 14 4 0North West 3 10 25 0 39London 0 0 2 0 1 3 0 ENGLAND 0 5 0 1 0SCOTLAND 47 0 0 28WALES 0 7 3 1 93 48 BRITAINGREAT 0 282 0 0 5Compare: 57 1 47 0 1992 GREATBRITAIN 0 93 0 0 9 357 0 28 10 1 69 6 3 9 11 0 3 64 1 10 0 22 74 0 10 0 23 5 157 0 292 79 167 0 367 1 92 529 14 1 11.6 641 12.9 34 + The Politico’s Guide to Electoral Reform in Britain an overall majority in the House of top-up seats (15 per cent). Scheme B Commons (where the winning post is is significantly more proportional, 330 seats). The Conservatives would with deviation from proportionality at still be under-represented, gaining just over 11 per cent – almost halving almost the same total of seats (167) as the DV score in the actual general they did in the actual general elec- election. However, the scheme would tion (165), while the Liberal still be well above the fully-propor- Democrats would double their seat tional 4 to 8 per cent range for DV count from 46 actual seats in 1997 to scores. The main change would be to 92 MPs under AV-Plus. reduce the number of Labour local The deviation (DV) scores in Table seats, with the extra top-up MPs go- 7 measure how many MPs would hold ing primarily to the Conservatives seats not justified in terms of their who would claim ten more MPs over- party’s share of the vote in their re- all than in the actual general election. gion or the country as a whole. A Labour’s total would fall a bit to 359 fully-proportional system in 1997 seats for Great Britain – still enough would have given a result within the for an overall victory – and the Lib- 4 to 8 per cent range, and probably eral Democrat total would scarcely towards the top end of that range. In change. With just 96 top-up MPs (15 the actual 1997 election under FPTP, per cent), disproportionality would the national DV score was much rise appreciably to nearly 15 per cent, higher, at 21 per cent. As usual, the offering the smallest reduction on the DV levels are higher at a regional actual general election DV score. La- level than across Britain as a whole, bour would be the clear beneficiary but they reduce to a lower national here, winning a major increase in lo- figure principally because a bias in cal seats (370 in all) at the expense of the Conservatives’ favour in southern all the other parties, though the Lib- England partly offsets a bias towards eral Democrats would be only Labour in its heartlands. Under the marginally affected. Commission’s scheme, the overall But what would happen at local national DV score would have been level? Table 10 (pp 38-40) shows the 12.9 per cent – a score which is three- results for scheme A (the 17.5 per cent fifths less disproportional than the top-up) at the level of the Commis- actual FPTP result in 1997, but still sion’s 78 top-up areas in mainland above the 4-8 per cent levels most Britain. For each area, we show the other European countries achieve. vote shares of the major parties, their Tables 8 and 9 (pp. 36-37) show projected holdings of local and top- what would have happened in 1997 up seats, and their total seat shares. conditions under the Commission’s We use the d’Hondt counting method two other schemes – scheme B, with to determine the allocation of top-up one in five Commons seats going to seats. (Interested readers can work out top-up MPs and scheme C with 96 the number of top-up seats a party has The Politico’s Guide to Electoral Reform in Britain + 35 16.8 18.3 11.3 14.1 14.1 11.5 15.4 18.3 12.8 0000 1314240051 0100 163940059 1200 143750056 04000300 164670170 184880074 88900 563922001177.9 3120041300 910300229.4 142640044 60100 72720036 6006060100 646812072 62743040 Local seatsCon Lab LD Nat Other Con seats Top-up Lab LD Nat Other Con Lab LD All seats Nat Other Total DV Table 8: Seats won under AV-Plus with 128 (20%) top-up in re-run 1997 election (Great Britain) 8: Seats won under AV-Plus Table South WestSouth East MidlandsWest East Anglia 3East Midlands 48 14 6 & HumYorks 31 24 39North 6 10 13 0North West 3 25 0 9 4London 0 0 0 1 36 1 0ENGLAND 5 0 10 1 0 3SCOTLAND 46 0 27 0 0 WALES 10 7 3 1 0 90 48 BRITAINGREAT 0 275 0 0 5Compare: 54 10 1 46 0 1992 GB TOTAL 0 90 0 0 8 348 0 27 268 11 1 65 6 216 3 9 24 11 0 3 73 5 1 11 0 0 25 85 0 11 41 0 26 24 6 57 163 0 6 286 79 0 175 0 359 309 1 91 240 529 15 81 11 1 10.4 641 0 641 11.6 7.6 36 + The Politico’s Guide to Electoral Reform in Britain 18.8 18.3 11.7 11.3 21.3 16.9 18.6 19.4 21.1 17.8 9000056600 1214250051 90100 554022001177.9 11100 163940059 41300 81130022 70100 142640044 60000 114140056 70300 72810036 80300 125160170 155180074 5004050000 548910072 52933040 Local seatsCon Lab LD Nat Other Con seats Top-up Lab LD Nat Other Con Lab LD All seats Nat Other Total DV Table 9: Seats won under AV-Plus with 96 (15%) top-up in re-run 1997 election (Great Britain) 9: Seats won under AV-Plus Table South WestSouth East MidlandsWest East Anglia 3East Midlands 50 14 7 & HumYorks 34 25 39North 7 10 16 0North West 3 10 25 0 4London 0 0 2 0 1 41 0 ENGLAND 0 5 0 1 3SCOTLAND 51 0 0 28 0WALES 7 3 1 0 94 51 BRITAINGREAT 293 0 0 0 5Compare: 59 1 48 0 1992 GB TOTAL 0 94 0 0 9 370 0 29 284 1 71 6 230 3 9 26 0 3 56 5 1 0 8 0 18 66 0 8 31 0 18 14 45 4 150 6 0 301 0 77 160 0 378 315 1 244 89 529 71 13 11 1 13.3 641 0 641 14.6 9.2 The Politico’s Guide to Electoral Reform in Britain + 37 Table 10: Seats won under AV-Plus in re-run 1997 election with 17.5% top-up, by top-up

Vote shares Local seats Top-up areas Con Lab LD SNP PC Other Con Lab LD Nat Other Dorset 41.8 18.8 34.1 0.0 0.0 5.3 01600 Wiltshire 40.2 28.0 26.2 0.0 0.0 5.6 02300 Gloucestershire 39.4 33.9 22.5 0.0 0.0 4.2 13100 Bristol and Bath 32.7 36.5 26.3 0.0 0.0 4.5 14300 Somerset 36.5 17.4 40.6 0.0 0.0 5.4 00400 Devon 36.8 25.9 31.3 0.0 0.0 6.1 13500 Cornwall 30.4 17.1 44.0 0.0 0.0 8.6 01300 Essex South West 39.8 40.1 15.3 0.0 0.0 4.8 44000 Essex North East 40.9 31.6 22.1 0.0 0.0 5.4 31200 Oxfordshire 38.0 31.7 24.7 0.0 0.0 5.6 31100 Berkshire 42.2 28.5 24.6 0.0 0.0 4.7 32100 Buckinghamshire 43.7 30.6 21.2 0.0 0.0 4.5 42000 Hertfordshire 40.6 39.7 16.0 0.0 0.0 3.7 45000 Bedfordshire 38.6 44.0 12.8 0.0 0.0 4.6 23000 Surrey 46.2 22.3 24.5 0.0 0.0 7.0 70200 Kent East 39.2 37.7 17.1 0.0 0.0 6.1 22100 Kent West 41.2 36.8 17.0 0.0 0.0 5.0 45000 East Sussex 39.4 29.2 24.0 0.0 0.0 7.4 23200 West Sussex 44.7 24.3 25.6 0.0 0.0 5.5 51100 Hampshire Solent 35.7 33.5 24.9 0.0 0.0 5.9 22300 Hampshire North 46.3 21.2 28.8 0.0 0.0 5.2 51100 Warwickshire 38.7 43.8 13.9 0.0 0.0 3.6 13000 Hereford & Worcs 41.0 32.6 21.9 0.0 0.0 4.5 23200 Shropshire 37.2 39.7 20.5 0.0 0.0 2.7 03100 Staffordshire 33.7 51.3 10.7 0.0 0.0 4.2 19000 Dudley & Sandwell 26.5 55.9 9.8 0.0 0.0 7.9 06000 Birmingham 28.4 54.6 12.8 0.0 0.0 4.3 18000 Coventry & Solihull 32.9 47.0 13.6 0.0 0.0 6.6 13000 Wolverhampton/ Walsall 33.2 54.4 8.3 0.0 0.0 4.2 14000 Suffolk 37.6 40.2 17.6 0.0 0.0 4.6 24000 Cambridgeshire 42.0 34.5 17.9 0.0 0.0 5.6 32100 Norfolk 36.7 39.9 18.2 0.0 0.0 5.1 14100 Lincolnshire 42.4 36.9 17.5 0.0 0.0 3.1 42000 Northants 40.4 45.0 11.1 0.0 0.0 3.4 14000 Leicestershire 36.8 43.8 15.1 0.0 0.0 4.3 34100 Notts 30.5 54.3 10.9 0.0 0.0 4.3 18000 Derbyshire 29.5 53.6 13.8 0.0 0.0 3.1 17000 North Yorkshire 40.0 32.8 23.0 0.0 0.0 4.2 32100 Humberside 30.4 50.4 15.8 0.0 0.0 3.3 16100 W Yorkshire: Leeds 28.0 55.3 12.9 0.0 0.0 3.8 07000 W Yorkshire: Bradford 33.0 49.6 13.2 0.0 0.0 4.3 06000 Continued overleaf

38 + The Politico’s Guide to Electoral Reform in Britain areas (GB) won in any area by simply subtract- All seats after top-ups ing the local seats from the total seats Con Lab LD Nat Other DVV figures). Notice that again there are 1160 0 40.9 some high and very high DV scores 1230 0 29.1 in many top-up areas. Even though 2310 0 16.1 these distortions are often 3430 0 7.2countervailing and even out at na- 1040 0 39.4 tional level, they are real enough at 3350 0 15.6 the level at which voters actually feel 1130 0 18.9 the experience of voting. Despite the 4510 0 10.1 effect of AV, which institutionalises 3220 0 8.4 ‘tactical voting’ of the kind which oc- 3210 0 13.6 4220 0 8.2curred in 1997, with Labour and 4210 0 13.5 Liberal Democrats voters making the 5510 0 10.6 other party their second preference, 3300 0 17.5 and thus damaging the Conservatives 7220 0 17.4 in many seats, a notable feature of the 3210 0 10.8 results is that both the Conservatives 4520 0 9.8and Labour have an MP in all the top- 3320 0 9.3up areas, except for four. The 5120 0 17.8 Conservatives are utterly excluded 3330 0 8.4 from three top-up areas in central 5220 0 11.0 2300 0 17.5 Scotland and Labour in the Somerset 3320 0 8.0top-up area. The Liberal Democrats 1310 0 20.3 have no MP in 25 top-up areas, mostly 3900 0 23.7 in Labour heartlands territory. They, 1600 0 29.8 too, gain no seats in the three central 2810 0 18.1 Scotland top-up areas from which the 2300 0 20.2 Tories are excluded; and indeed eight 2400 0 12.5 of Glasgow’s ten seats go to Labour 2410 0 16.9 (the SNP claims the other two). The 3310 0 9.2 SNP would gain an MP in four out of 3410 0 10.8 eight top-up areas in Scotland, while 4210 0 14.7 2400 0 21.6 Plaid Cymru would win seats in two 4510 0 9.4out of five top-up areas in Wales. Thus, 2810 0 18.4 the undesirable phenomenon of ‘elec- 2710 0 16.4 toral deserts’, noted by the 3320 0 6.7Commission (see p. 13), is greatly di- 3610 0 9.6minished. 1700 0 32.2 For the sake of comparison, Tables 1600 0 36.1 7, 8 and 9 all include a row below the 1997 results showing how the parties The Politico’s Guide to Electoral Reform in Britain + 39 Table 10 continued Vote shares Local seats Top-up areas Con Lab LD SNP PC Other Con Lab LD Nat Other W Yorkshire: South 25.9 56.9 12.7 0.0 0.0 4.5 07000 South Yorks (Sheffield & Rotherham) 16.5 59.7 20.3 0.0 0.0 3.5 06100 South Yorks (Barnsley & Doncaster) 16.9 66.3 11.0 0.0 0.0 5.9 05000 Cleveland 25.2 62.4 9.8 0.0 0.0 2.6 05000 Tyne & Wear South 14.6 69.4 11.8 0.0 0.0 4.3 07000 Tyne & Wear North (inc Newcastle) 21.2 63.8 11.9 0.0 0.0 3.2 04000 Durham 17.6 68.5 9.7 0.0 0.0 4.2 06000 Cumbria 33.5 45.8 16.5 0.0 0.0 4.1 14000 Northumberland 22.7 48.7 25.0 0.0 0.0 3.7 02100 Liverpool & Wirral 19.1 63.5 12.8 0.0 0.0 4.7 07000 Merseyside North 20.6 60.0 16.5 0.0 0.0 3.0 05100 Manchester North 25.8 54.2 16.7 0.0 0.0 3.3 06100 Manchester West 23.6 62.3 10.9 0.0 0.0 3.3 07000 Manchester East 23.0 53.1 20.2 0.0 0.0 3.7 16100 Cheshire 33.4 46.5 12.3 0.0 0.0 7.8 26001 Lancashire South 29.6 55.2 10.8 0.0 0.0 4.3 06000 Lancashire North 39.3 42.5 14.8 0.0 0.0 3.4 24000 South East London 36.6 41.3 16.7 0.0 0.0 5.4 36000 South West London 34.6 33.7 28.2 0.0 0.0 3.5 03400 South Central London 23.6 57.9 14.3 0.0 0.0 4.2 08100 North West London 32.8 53.1 10.1 0.0 0.0 3.6 18000 North London 32.9 52.3 11.4 0.0 0.0 3.4 18000 North East London 28.2 56.9 8.3 0.0 0.0 6.6 08000 North Central London 27.4 53.5 12.7 0.0 0.0 6.5 27000 Scotland: South 22.6 43.4 13.4 19.1 0.0 1.6 05210 Scot Highlands 16.2 27.0 27.7 26.7 0.0 2.4 02310 Scot N E 22.4 30.9 18.9 26.1 0.0 1.7 04220 Scot Mid & Fife 21.1 40.0 12.6 25.3 0.0 1.0 05120 Scot Central 10.4 59.4 5.2 23.4 0.0 1.6 08000 Scot West 18.4 52.0 9.4 20.2 0.0 1.3 08000 Lothians 19.2 45.9 14.9 18.4 0.0 1.5 06100 Glasgow 8.5 60.2 7.3 19.4 0.0 4.6 09000 Wales North 24.3 46.7 11.8 0.0 14.2 3.1 06110 Wales Mid 20.7 37.8 18.4 0.0 20.1 3.0 03220 Wales South West 15.0 65.6 10.7 0.0 5.7 3.0 06000 South Wales Central 20.3 58.1 11.8 0.0 5.6 4.2 06000 South Wales East 16.7 66.1 9.4 0.0 4.2 3.6 07000 Great Britain 31.5 44.4 17.2 2.0 0.5 4.4 93 357 69 9 1

40 + The Politico’s Guide to Electoral Reform in Britain would have performed under the All seats after top-ups same version of AV-Plus in the 1992 Con Lab LD Nat Other DV general election. The 1992 election 1700 0 30.6 was a more balanced one, with the Conservatives on nearly 43 per cent 1620 0 8.9support, Labour on 35 and the Liberal Democrats just over 18 per cent. The 1500 0 17.1 1500 0 20.9 distribution of Liberal Democrats’ sec- 1700 0 18.1 ond preferences was more even, with most support going to the Conserva- 1400 0 16.2 tives nationally and Labour gaining 1600 0 17.2 more Lib-Dem second preferences 2400 0 20.8 than the Tories in only a few regions. 1210 0 3.7 Both Conservative and Labour voters 1710 0 14.3 gave the Liberal Democrats majority 1510 0 11.5 2610 0 12.5 backing at the second preference 2700 0 15.5 stage. In none of the three AV-Plus 2620 0 6.9scenarios would the Conservatives 3610 1 9.3have won the overall majority which 2600 0 19.8 FPTP gave them in the actual elec- 2410 0 14.6 tion in 1992. With a 17.5 per cent 4610 0 13.2 top-up in 1992, both the Conservatives 2340 0 16.3 and Labour would have won most of 2810 0 14.8 their seats locally, but would both have 3800 0 19.4 scored well at the top-up stage as well. 2810 0 20.5 2800 0 23.1 The Liberal Democrats would have 3710 0 10.2 won two thirds of their 74 seats at the 1521 0 21.0 top-up stage, but their overall total in 1231 0 16.7 seats would have been lower under 1422 0 16.9 AV-Plus in 1992 than in 1997, even 1512 0 15.6 though their vote was slightly higher 0802 0 20.6 in 1992. The DV score for Scheme A 0801 0 37.6 in 1992 – 8.7 per cent – would be only 1611 0 20.7 slightly outside the full proportional- 0901 0 29.8 ity range (see Table 7). 1611 0 24.0 1322 0 21.6 A 20 per cent top-up scheme in 1600 0 17.2 1992 would have fallen within the 1610 0 14.8 fully-proportional range with a DV 1700 0 19.3 score of 7.6 per cent (see Table 8). The 167 367 92 14 1 12.9 Liberal Democrats would have increased their share of seats to 81, 57 of them won at the top-up stage, and The Politico’s Guide to Electoral Reform in Britain + 41 their gains would have been largely their appeal to voters. Protected in at the expense of the Conservatives. their safe seat areas and able to se- Reducing the proportion of top-up cure majorities on low shares of the seats to 15 per cent would primarily vote, parties under plurality rule had have benefited Labour in 1992 (see few incentives not to try and bounce Table 9). Labour’s seat total would ‘extreme’ or unpopular policies into actually rise by four compared with effect, relying on their core support- the other two schemes, with most of ers alone. Party behaviour can be their gains coming from the Liberal expected to change under AV-Plus, Democrats. The DV score would rise and hence the higher distortions that above 9 per cent, a result that is out- arise under AV-Plus rather than an side the full proportionality range of AMS equivalent using FPTP for the 4 to 8 per cent even in a relatively trac- local elections (see next section be- table election. low), would in practice be lower than Although the AV-Plus system has in our re-modelled 1997 results. advantages in terms of expanding voter choice and in giving MPs the legitimacy of majority support in their constituencies, it can under some circumstances carry a heavy penalty in terms of greater disproportionality. In elections when one of the major parties is particularly disliked, AV elections facilitate joint action against it by voters supporting all the other parties. In 1997 the Conservatives would have lost even more heavily under AV than FPTP because around 61 per cent of voters wanted them out of power, whereas in 1992 no similar conditions applied. Thus AV-Plus (or SV-Plus) automates tactical voting in the local constituency contests. Con- sidering previous elections, we would expect that in 1983, for instance, a very similar effect would have severely penalised Labour under AV-Plus. Defenders of AV-Plus might well argue, however, that these strong effects arose under plurality rule because there were not such strong incentives for parties to maximise 42 + The Politico’s Guide to Electoral Reform in Britain VOTES AND SEATS UNDER AMS IN 1997 AND 1992

n AMS version of the Jenkins elected with less than majority sup- schema could have two port – as 47 per cent of MPs were in Aadvantages over AV-Plus. It the 1997 general election. would be more proportional (as we In 1997, an AMS version of the show below) than AV-Plus in 1997 Commission’s scheme A (with 113 conditions and similar elections. And top-up seats, because it is a slightly it would allow UK elections to be earlier variant of the scheme that is standardised more on a ‘British AMS’ being tested) would have given the model, since other new electoral sys- Conservatives almost 30 more seats tems on the mainland (for Scotland’s than the similar AV-Plus scheme – 194 Parliament, the Welsh National As- as against 167 (see Tables 11 and 7). sembly and the Greater London Two thirds of these seats would have Assembly) share key features. All of come from the Liberal Democrats and them use a double-vote AMS ballot, the other third from Labour. At 10.8 with FPTP elections for local mem- per cent, the DV score would have bers; all have a majority of members been two percentage points lower elected in local constituencies and than AV-Plus, although still outside fairly small top-up areas; all use the the fully proportional range. As for d’Hondt allocation rule to distribute 1992, the share of seats among the seats between parties; and all in ef- major parties differs only slightly be- fect have relatively high thresholds for tween the AMS version and its third parties seeking to win seats. An AV-Plus equivalent (Tables 11 and 7), AMS version of the Commission and the DV score improves only by scheme, as advocated by Lord Alex- less than half of one percentage point. ander, the dissenting member, would At top-up level under the AMS ver- arguably fit quite closely within this sion of scheme A (though with a wider model, whereas an AV-Plus slightly different distribution of seats scheme introduces a discordant ele- between top-up areas), ment. disproportionality ranges very widely Against these points, of course, the from an almost perfectly proportional ability of voters to indicate preferences result in Shropshire (3 per cent) to a at the local constituency stage is re- still highly disproportional outcome in stricted to one vote (creating tactical Dorset (45 per cent). While these bi- voting incentives to misrepresent their ases partly offset each other, they may preferences for some voters), and MPs still be alienating at the point of vot- in local constituencies could still be ing and receiving the local result. The Politico’s Guide to Electoral Reform in Britain + 43 15.9 14.1 17.0 20.3 7300 22 07 22036 1007 27200 0807 40 0006 45812072 30130 Local seatsCon Lab LD Nat Other Con seats Top-up Lab LD Nat Other Con Lab LD All seats Nat Other Total DV The share of seats under AMS with 113 (17.6%) top-up in the re-run 1997 election, Great Britain Table 11: Table South WestSouth East MidlandsWest East AngliaEast Midlands 19 & HumYorks 61 12 11North 29 37North West 12 12 12London 7 0 0 6 24 7ENGLAND 0 0 0 40 0 0SCOTLAND 6 2WALES 0 0 4 0 1 0 48 8 27 2 6 BRITAINGREAT 138 0 0 0 2Compare: 47 1 1 270 0 0 3 0 1992GB TOTAL 0 6 0 138 4 28 5 0 45 0 6 12 345 287 1 4 0 0 0 0 2 0 8 30 221 37 0 7 0 0 1 0 5 0 3 4 3 16 1 7 0 23 43 10 0 3 0 0 0 5 3 63 1 0 18 0 0 0 0 7 0 56 0 9 0 12 6 15 12 0 35 40 16 6 37 4 0 12 30 0 19 9 0 0 4 0 0 15 26 40 0 0 0 0 13 59 0 0 40 3 8 181 0 18 51 8 0 4 0 0 117 48 0 194 279 59 0 0 47 8 8.5 68 12.5 0 44 354 317 9 14.9 0 0 236 56 77 0 1 1 75 11.3 15 13 0 529 70 19.5 1 0 74 641 641 14.4 9.6 14.0 10.8 8.3 44 + The Politico’s Guide to Electoral Reform in Britain 13.2 15.9 11.3 14.1 11.3 12.9 14.0 17.0 15.3 0400 184790074 42400 22131600516.5 6050001300 183650059 32300 12730022 80400 152630044 50200 143750056 80600 72630036 144780170 7008071100 745812072 72823040 Local seatsCon Lab LD Nat Other Con seats Top-up Lab LD Nat Other Con Lab LD All seats Nat Other DV The share of seats under AMS with 127 (19.8%) top-up in the re-run 1997 election, Great Britain Table 12: Table South WestSouth East MidlandsWest 18East AngliaEast Midlands 60 11 12 & HumYorks 12 27 36North 12 12 0 7 0North West 6 24 0 6London 0 0 0 0 37 0 0 ENGLAND 0 6 0 2 1SCOTLAND 47 0 0 26 4 0WALES 8 136 2 6 1 0 47 BRITAIN 261GREAT 13 0 0 0 5Compare: 28 0 1 45 0 1361992 GB TOTAL 0 0 0 0 333 8 0 27 276 1 37 4 218 1 7 64 16 10 0 48 3 33 1 5 11 20 0 44 0 0 62 0 12 0 34 0 117 45 21 8 63 13.4 184 0 272 8 72 0 198 0 345 1 310 82 529 239 15 79 1 13 8.9 641 0 641 9.4 7.6 The Politico’s Guide to Electoral Reform in Britain + 45 14.9 15.9 11.3 21.3 14.1 18.6 15.4 22.5 20.3 41400 23121600518.5 6040000300 183740059 32300 12730022 60200 152630044 50100 124130056 70300 72720036 90400 135150170 174890074 5005050000 549711072 53014040 Local seatsCon Lab LD Nat Other Con seats Top-up Lab LD Nat Other Con Lab LD All seats Nat Other Total DV 291 227 17 6 0 25 15 54 6 0 316 242 71 12 0 641 9.0 The share of seats under AMS with 1OO (15.6%) top-up in the re-run 1997 election, Great Britain Table 13: Table Seats outcomes in the 1997 election re-run under AMS, with 100 (15.6%) top-up seats (Scheme C) South WestSouth East MidlandsWest 19East AngliaEast Midlands 62 11 12 & HumYorks 12 30 37North 12 12 0 7North West 0 7 24 0 6London 0 0 0 0 41 0 0 ENGLAND 0 6 0 2 1SCOTLAND 51 0 0 27 2 0WALES 8 139 2 5 1 0 48 BRITAIN 276GREAT 11 0 0 0Compare: 5 28 0 1 491992 GB TOTAL 0 139 0 0 0 0 355 7 0 30 1 36 6 10 1 64 0 42 4 1 35 8 18 0 35 52 0 0 8 0 117 35 0 5 13.4 181 0 284 63 191 0 363 71 1 529 15 1 10.6 641 12.2 46 + The Politico’s Guide to Electoral Reform in Britain However, the electoral desert effect cure seats in 28, or 36 per cent, of the would again be greatly diminished, mainland top-up areas (36 per cent). with Conservative and Labour MPs in In 1992 conditions, there are too few virtually all top-up areas (though the top-up seats under scheme C to push Liberal Democrats would not win a the national result into the fully pro- seat in 25 out of 78 areas). (See the portional range. LSE/Birkbeck report on the Commis- sion’s schemes, Dunleavy and Margetts 1998, pp. 34-7 for the full results.) Moving on to scheme B for 127 top- up MPs (19.8 per cent of the total), the slightly increased number of top- up MPs would reduce the DV score under 1997 conditions to just 9.4 per cent, or 2.2 percentage points fewer than for the AV-Plus version of scheme B (see Tables 12 and 8), and appreciably closer to full proportionality. In 1997, the Liberal Democrats and the Conservatives would have gained equally from the additional top-up MPs, lowering the size of Labour’s overall majority from 47 to 29 seats. By contrast, in 1992, the DV score for the 20 per cent AMS scheme would be the same as that for AV-Plus, lying just inside the fully proportional range. Table 13 shows the results for AMS using an earlier version of scheme C with 100 top-up seats. In 1997 conditions, it again performs more fairly than the equivalent AV- Plus scheme, but its DV score of over 12 per cent makes it nearly 3 percentage points worse than the AMS scheme B and 1.4 points worse than scheme A. Most of the top-up seats lost under scheme C would be in urban areas so that Labour’s representation is particularly boosted. The Liberal Democrats would not se- The Politico’s Guide to Electoral Reform in Britain + 47 THE ‘DANGER’ OF SPLIT- TICKET TACTICAL VOTING

he double-ballot is an impor local seats and far fewer top-up seats tant feature of the AV-Plus is especially vulnerable to large-scale Tscheme, allowing people to tactical voting if people choose to cast vote one way at local level and another their votes for top-up MPs differently at the top-up stage. However, it is ar- from their constituency votes. gued that giving people two votes Kellner’s basic argument was that offers scope for ‘split-ticket’ tactical Labour voters in top-up areas where voting which would damage both the their party has won a surfeit of local greater proportionality of the new seats and cannot hope to secure top- scheme and its ability to avoid ‘elec- up seats could shift their top-up votes toral deserts’. This section gives the to the Liberal Democrats in suffi- results of a sensitivity analysis con- ciently heavy numbers to squeeze out ducted for the Commission by Patrick one or more possible Tory top-up MPs Dunleavy and Helen Margetts on the in these areas. He claimed that about basis of the AV-Plus scheme as it ex- a third of Labour voters behaving in isted in mid-September 1998, with this way could have relegated the 113 top-up seats in Great Britain, al- Conservatives in 1997 to third place located in 78 top-up areas. (This in the Commons, even though they scheme is essentially the same as the were clearly second in the popular Commission’s scheme A, but the seat vote. figures in Table 14 below differ by five There are strong reasons for doubt- seats from those in the up-to-date Ta- ing whether such a scenario, or its ble 7 above.) Conservative equivalent, is likely to Any electoral system can create occur in practice. First, Kellner as- incentives for tactical voting, even if sumed that the top-up areas would be it is fully proportional. But some sys- regional, whereas the far smaller top- tems can be particularly sensitive to up areas would be less susceptible to the effects of the strategic behaviour large-scale manipulation and the ef- of voters. FPTP is certainly one of fects of any organised tactical strike these, as Michael Portillo and other would be more contained. will Fur- Tory MPs In ‘safe’ seats know to their ther, with only one or two top-up MPs cost. The columnist Peter Kellner ar- per area, a second-placed party un- gued (in the Evening Standard of 7 represented in local seats may well September 1998 and on BBC2’s have piled up a significant lead in Newsnight the same day) that an AV- votes over the third-placed party – re- Plus system with a large majority of quiring a lot of tactical voting for this 48 + The Politico’s Guide to Electoral Reform in Britain lead to be over-turned under the Much tactical voting misfires because d’Hondt counting system (which the voters simply get it wrong, or be- tends to favour large parties). cause of countervailing tactical The culture of Britain’s political decisions by different groups of vot- parties also runs strongly against any ers. such organised strategy, even though We have assessed how vulnerable local parties did in 1997 covertly en- to ‘split-ticket’tactical voting AV-Plus courage tactical voting. It would also would have been in the very different be an unusually confident party which conditions of 1997 and 1992. We ex- decided that it would cream off all the amined all 78 top-up areas in Great local seats – and then determined to Britain in both election years to iden- ask its voters to vote for another party. tify those where There are clear risks in such a strat- ● the leading party wins all or the egy for Labour (say). Could party bulk of local seats, so that voters could organisers be sure that the only effect foresee that this party would not be of encouraging its voters to switch to able to win a top-up seat; and the Liberal Democrats at the top-up stage would be to squeeze out the To- ● the leading party has a convincing ries? For over quite a short period, the lead in terms of vote shares over the perceived viability of the Liberal second-placed party within that top- Democrats may improve, as indeed it up area. This criterion ensures that the has already done under FPTP in seats political dominance of the leading like Cheltenham, as a result of con- party is visible to voters, since it is not scious and unconscious tactical voting just an artefact of hard-to-anticipate by Labour supporters. They could seat allocations. then become serious rivals to Labour For each such seat we calculated for local seats, especially if local Tory the minimum amount of tactical vot- voters also began to see the Liberal ing by the leading party’s supporters Democrats as a more electable ‘oppo- which would be necessary to raise the sition’ party. This kind of process has third-placed party above the second already occurred at local government and thus to win a top-up seat if abso- level in Labour strongholds such as lutely everyone who could vote Liverpool and Sheffield. Finally, the tactically did so. In 1997, there were ability of political parties to organise 35 areas where Labour vote-switching such complex manoeuvres must be in could in such extreme circumstances doubt. cost the Tories a seat, and two cases As for the voters themselves, not where the reverse effect could apply. many more than one in ten attempt to In 1992, there were 21 cases where vote tactically, and fewer still achieve Labour voters, and 12 where Con- their objective. First, they must vote servative voters, could bring about in the right direction in the right con- seat losses for their opponents. stituencies to achieve their goal. What is important, however, is the The Politico’s Guide to Electoral Reform in Britain + 49 likely level of vote-switching required bour supporters would actually im- to trigger these losses. In 1997 for in- prove the DV score, by raising the stance if a more realistic 10 per cent representation of the Liberal Demo- of voters for Labour and the Conserva- crats to more proportional levels and tives in vulnerable top-up areas ‘split’ reducing in a more even-handed way their votes, the Conservatives would the over-representation of the other have lost nine seats and Labour one, two main parties. The DV score falls giving the Liberal Democrats ten more rapidly at all levels of tactical voting seats. In 1992, the Conservatives and, at Kellner’s 30 per cent point, the could have lost five seats and Labour score would actually be halved. So three, adding eight seats to the Lib- AV-Plus system would then be oper- eral Democrats’ total. It is only at an ating within the fully-proportional extreme level of tactical voting, such range! Thus split-ticket tactical vot- as the 30 per cent assumed by Peter ing is not at all the great danger for Kellner, that there is a significant ef- democracy which the ingenious fect. At that level, 23 Conservative and Kellner supposed (see Dunleavy and two Labour seats could have been lost Margetts, LSE-Birkbeck 1998, for in 1997, and 17 Tory and eight Labour more detailed figures). in 1992. The Conservatives in fact There would however be one ad- would have been badly hit in 1997 if verse consequence. Split-ticket voting more than 15 per cent of Labour vot- of this kind would somewhat offset the ers split their votes tactically. effect which an AV-Plus electoral sys- But what overall damage would tem would otherwise have in concerted split voting inflict on the improving the representation of other proportionality of elections under AV- parties in areas of one-party domi- Plus? As we have explained above, it nance (Jenkins’s ‘electoral deserts’). is possible to measure deviation from Substantial split-ticket tactical voting proportionality (see p. 9) and give it a would erode AV-Plus’s ability to score (the ‘DV score’) for the sake of broaden the regional base of the Con- comparison. In 1997, we calculate that servative and Labour parties (though, the DV score under AV-Plus would of course, it would not increase the have remained basically unchanged leading party’s predominance). right up to levels of split-voting just But it is worth noting as well that below Kellner’s 30 per cent. Only at split-ticket voting would primarily re- or above that level – with 30 per cent flect the expansion of voter choice of Labour and Tory supporters voting which the scheme allows. The Con- tactically for top-up MPs and doing servatives lost out badly in 1997 so with remarkable accuracy, all vot- because a large body of voters for ing the right way in the right other parties were also very hostile to constituencies – would a tiny adverse them. Had the policies and perform- effect reveal itself. In 1992, split-ticket ance of the Conservatives in tactical voting by both Tory and La- government reflected more sensitively 50 + The Politico’s Guide to Electoral Reform in Britain what the public wanted, this marked ing effects will take time to emerge, hostility would not have existed. It as a result of ‘social learning’. Over might be argued that the adoption of the same period, however, political an electoral system which penalises parties will also have time to modify parties which get so out of touch with their campaigning strategies to guard people’s wishes would expand politi- against or to exploit tactical voting cal accountability to the people, and potential. is to be preferred over a system, such Overall, we judge that between 5 as FPTP, which encourages parties to and 15 per cent of voters as a whole try merely to ‘bounce’ the people’s are likely to choose in divergent ways wishes by gaining a Commons major- between the constituency and the top- ity by winning over a 42-44 per cent up area votes under the Commission’s minority of the electorate. AV-Plus scheme. The level of ‘net’, or But how much split-ticket tactical effective, tactical voting is likely to be voting is really likely? Our figures are below 10 per cent. There is undoubt- based on the assumption that the tac- edly some potential for the Liberal tical voting would be undertaken by Democrats to advance at the top-up perfectly informed voters who could stage, but there could also be accurately foresee the results in their countervailing effects reflecting that top-up area. They would need to know party’s weaker hold over its voters and far more than they require now vot- stronger local than national identity. ing in a single constituency under The net effects on the parties’ overall FPTP. Split-ticket tactical voters would national seat total are likely to be need to know the political conditions small. in their local seat and the likely out- There is a simple remedy to wide- comes in another four to ten seats in spread split-ticket tactical voting – a their top-up area. But many voters shift to a one-vote mixed system. But won’t have this level of information this would directly restrict an impor- in many areas. It may be fairly clear tant element of voter choice in the to Labour supporters in Glasgow for Commission’s schema simply to avoid instance, but could the same be said a theoretical problem which is any- for the party’s voters in north east way unlikely to distort election results London? and may even improve their propor- A change in the way we vote will tionality. clearly set in train dynamic effects, but It is worth recording the evidence as we have stressed in all our work showing that voters appreciate and on electoral reform, these effects can use the greater degree of choice made not be scientifically estimated. It is possible by split-ticket voting. More very likely that voters in the first elec- than one in three voters in New Zea- tion under a new system will be very land’s first election in 1996 under a unclear how it is going to work. It is ‘classic’ Additional Member System also probable that any net tactical vot- (with a 50:50 division of local and top- The Politico’s Guide to Electoral Reform in Britain + 51 up seats) made different choices at the two stages of the AMS ballot, even though most parties stood to gain at least some top-up seats across the country. One in eight voters in Ger- many also exercise their right to split their vote. In our 1997 work on modelling the future of the Scottish Parliament and Welsh Assembly (Devolution Votes, Democratic Audit, 1997), we found that 15 per cent of voters were prepared to split their vote in Scotland. Subsequent opinion polls in Scotland have begun to show similar levels of split-ticket voting intentions.

52 + The Politico’s Guide to Electoral Reform in Britain THE REPRESENTATION OF SMALL OR ‘EXTREME’ PARTIES

ne of the most common criti formal threshold, of the kind which cisms of proportional elec Germany has, to protect the body poli- Otoral schemes is that they tic from very small parties gaining an fragment the political spectrum by undue influence in the House of Com- increasing the ability of very small mons on a fraction of the popular vote. parties to secure a seat in the legisla- Germany, in fact, applies the quite ture. There is a particular fear that a high threshold of 5 per cent of the to- party representing an extreme popu- tal vote, victory in a local seat, or list, ethnic identification or racist before it can qualify for a top-up seat. viewpoint might gain a parliamentary However, Germany and other nations bridgehead and perhaps exert an in- which set a threshold for entry to their fluence out of all proportion to its parliaments operate fully proportional following simply by virtue of it hold- schemes, and the essence of AV-Plus ing a tactical, or even ‘hinge’, position is that it deliberately is not fully pro- in Parliament. The quite atypical ex- portional, precisely to support the ample of Israel is often held up as a larger parties and to make single- warning against any kind of propor- party government the norm. tional system. Dunleavy and Margetts undertook We do not intend to enter into the extensive analysis of the effective merits of such fears, or the underly- thresholds created under AV-Plus on ing political viewpoint to which they behalf of the Commission. They con- give expression, except to point out centrated on the largest top-up areas that the advent of new parties, as with – with 11 MPs – because the larger the Greens in Germany, or the ad- constituencies are, the more likely are vance of a small party such as the small parties to win seats. Here we Scottish Nationalists in Scotland, can present a brief summary of their reinvigorate the political process and analysis (for further detail, see their could improve parliamentary repre- report to the Commission, LSE/ sentation of our increasingly pluralist Birkbeck 1998). Their figures are society. Here we concern ourselves theoretical and assume a proportional with the question of whether the Com- allocation of local seats within the top- mission’s AV-Plus scheme requires a up areas. Very broadly, within the

The Politico’s Guide to Electoral Reform in Britain + 53 larger top-up areas where their best chances of gaining seats are to be found, small parties would have to secure more than 8 per cent of the vote in the area, or some 28,000 votes, to be sure of gaining a seat. The lowest percentage of the total vote which would have secured a top-up seat in the simulated election results prepared for the Commission was the 10.9 per cent scored by the Liberal Democrats in Nottinghamshire. These are in effect very high thresholds and the Commission has rightly concluded that there is no need to set a formal figure. There is no danger of an extremist party scaling the effective thresholds built into its scheme. We have examined the voting histories of the National Front and British National Party in national and local elections and they have never polled the levels of voters which would be required across a whole top- up area to gain a seat in the House of Commons under AV-Plus.

54 + The Politico’s Guide to Electoral Reform in Britain THE TREATMENT OF NORTHERN IRELAND

he Commission wanted to rec calculate results for a 50:50 ‘classic’ ommend a voting system AMS or AV-Plus system. Twhich could be applied with We examined four possible ar- equal validity throughout the United rangements of the 18 seats in Kingdom, but Northern Ireland al- Northern Ireland, which would allow ready had a distinctive voting system an AMS or AV-Plus scheme to be im- of its own, separate from that for the plemented. They are: UK as a whole. The single transfer- ● a province-wide scheme with 14 able vote (STV), which is the system local seats, plus four top-up seats (22 used in Ireland, is also used for elect- per cent of all MPs); ing the Northern Ireland Assembly, ● local authorities and MEPs in the a scheme with two top-up areas, province. However, the Commission one a ten-seater, the other with eight found that representative organisa- seats (including two top-up seats tions in Northern Ireland also wanted each); to have the same system as applied ● the same scheme of a ten-seater and throughout the UK (even though an eight-seater top-up area, but this some wanted that system to be STV). time giving the eight-seater a single As is well known, Northern Ireland MP (thus providing only three top-up has a completely separate and distinc- MPs in the province), comprising 16.7 tive party system in which the main per cent of all MPs; line of cleavage has long been a sec- ● a scheme with three six-seat top-up tarian one. Assessing alternative areas, each including one top-up MP electoral outcomes here is compli- (again 16.7 per cent of all MPs). cated by the larger number of observable parties, with at least 12 We had no reliable data on the sec- principal parties that might win seats ond or subsequent preferences of under a reasonably proportional elec- voters at the 1997 general election. toral system. The constituencies Hence we can only present AMS pro- nearly all have idiosyncratic local fea- jections for this election. We obtained tures which set them apart from their information on second and subse- neighbours, not least sharply varying quent preferences for the Northern balances of electors on either side of Ireland Assembly election, held un- the sectarian line. This feature of lo- der STV in 1998, from an exit poll of cal politics had clear implications for 2,193 respondents (with a weighted pairing of constituencies necessary to value of 1,637 people). The poll was The Politico’s Guide to Electoral Reform in Britain + 55 carried out as people left the Assem- of deviation is the over-representation bly election polling stations by Ulster of the Ulster Unionists in local seats Marketing Services for RTE, the Irish (though, with AV, these results reflect state broadcasting corporation. This the transfer of support from other un- election marked a big change in the ionist parties). This result differs little politics of the province; the Unionist from that of a province-wide AV-Plus bloc split in two with ‘YES Unionists’ election. The DV score is the same, backing the peace agreement, and but the SDLP would gain one seat ‘NO Unionists’ opposing it. The vote fewer and the Ulster Unionists one for the Ulster Unionist Party fell to a more in a province-wide contest. In record low of 21 per cent, and the an election using three six-seater con- Democratic Unionist Party achieved stituencies, with one top-up seat each, 18 per cent. It is too early to say if this the DV score rises by ten percentage effect is a one-off, or a harbinger of a points, chiefly because the Alliance longer-term change. We are able to Party does not win a seat and Sinn provide an AV-Plus projection for Fein get only two. The DUP and Ul- Northern Ireland, based on the 1998 ster Unionists gain one seat each from election, though people were voting these changes. for a devolved assembly and not for An alternative way of assessing the Westminster, and the different mix of results is to look at the outcomes by issues and responsibilities across the blocs rather than by parties (Table 15). two bodies may have occasioned dif- This enables us to compute a ‘bloc’ ferent expressions of preferences. DV score which compensates for the Further, voting behaviour under STV presence of smaller parties, and to may be systematically different from suggest how legitimate the election that under AV, and we have by defini- results are likely to seem. The Com- tion excluded from our analysis mission’s two top-up area schema cuts ‘loyalist’ voters who chose only to vote the projected DV score by more than for their first preference party. half to 4.6 per cent, by comparison Table 14 shows the result of the with a notional FPTP outcome (10.2 Assembly election replayed under the per cent), and a province-wide AV-Plus model that the Commission scheme to 5.4 per cent. By contrast, recommends for use in Northern Ire- the three six-seater scheme yields a land – two top-up areas with two DV score even higher than under top-up MPs each. The deviation from FPTP. proportionality (DV) score is very high It is possible to re-run the 1997 at 17.9 per cent, reflecting in part the elections in Northern Ireland under fact that almost 15 per cent of the ini- an AMS equivalent of the Jenkins AV- tial vote went to small parties which Plus scheme (i.e., using FPTP for the did not win a seat, and would be de- local seat elections instead of AV). nied a seat under almost any electoral (This cannot be done for AV-Plus it- system. Otherwise, the main source self because we have no useful data 56 + The Politico’s Guide to Electoral Reform in Britain Table 14: The 1998 Assembly election re-run under AV-Plus with a ten-seater and an eight seater top-up areas each inc two top-up MPs

SDLP SF DUP UU APNI UKU PROG UDP CON Other Others NLP Total DV union score Local seats 4 3 2 6 0 0 0 0 0 0 0 0 14 Top-up seats 1 0 2 0 1 0 0 0 0 0 0 0 4 Seats under AV-Plus 5 3 3 6 1 0 0 0 0 0 0 0 18

% Votes shares 22 17.6 17.8 21.3 6.5 4.5 2.5 1.5 0.2 2.8 3.2 0.1 100 % Seat shares under AV-Plus 27.8 16.7 16.7 33.3 5.6 0.0 0.0 0.0 0.0 0.0 0.0 0.0 100 17.9

Notes: SDLP=Social Democratic and Labour Party; SF Sinn Fein; DUP Democratic Unionist Party; UU Ulster Unionist Party; APNI Alliance Party of Northern Ireland; UKU UK Unionist Party; PROG Progressive Unionist Party; UDP Ulster Democratic Party; CON Conservative; Others includes the Workers Party, Women’s Coalition, independents; Other Union – includes all other unionist parties; NLP Natural Law Party.

Table 15: Results of an AV Plus re-run of the 1998 Assembly Election by bloc

The Commission’s scheme (2 top-up areas with 2 top-up MPs each) Nationalist bloc Non-aligned Unionist bloc Bloc DV & other parties score Seats under plurality rule 8 0 10 Seats under AV Plus 8 1 9 % vote share 39.6 9.6 49.6 % seats under plurality rule 44.4 0 55.6 10.2 % seats under AV-Plus 44.4 5.6 50 4.6

A province-wide scheme with four top-up seats

Seats under AV Plus 7 1 10 % seats under AV Plus 38.9 5.6 55.6 5.4

Three six seat top-up areas, each with one top-up MP

Seats under AV Plus 7 0 11 % seats under AV Plus 38.9 0 61.1 10.8 on voters’ second and subsequent more evenly distributed and the de- preferences.) With two top-up areas viation from proportionality would be of ten and eight seats (including two far lower than in the actual election – top-up seats each), the exact equiva- at 14.9 per cent as against 26.8 per lent of the Commission’s AV-Plus cent under FPTP. A province-wide model, the outcomes for the national- AMS scheme would give the nation- ist parties and unionists would be alist parties three more seats than The Politico’s Guide to Electoral Reform in Britain + 57 FPTP, take three seats from the Ulster the votes for small parties, unlikely to Unionists, and allocate one seat to the secure representation under STV, we Alliance. Accordingly, it would be would expect its performance to be even more proportional, at 10.5 per worse than either AV-Plus schemes or cent. (Bear in mind in that 4 per cent their AMS equivalents. of the vote went to small parties which may well not win seats under any con- ceivable electoral system.) Assessed on a bloc basis, both AMS schemes would be far lower than the actual bloc DV under FPTP in 1997 (21.5 per cent), but the AMS equivalent to the Commission’s two top-up area scheme is, at 10.5 per cent, twice as high as under the province-wide version (4.2 per cent). The province-wide scheme again represents the two main blocs accurately and awards a seat to the Alliance. The Alliance keeps its seat under the two top-up area scheme, but the balance between the two main blocs tips back by two seats towards the unionist side. These results strongly suggest that having four top-up seats makes an important contribution to producing more proportional outcomes. The province-wide scheme, using either FPTP or AV for the local seats, is clearly more proportional than other models, including the Commission’s own, and always gives the Alliance a seat, thereby broadening party representation. However, province-wide top-up MPs in Northern Ireland would be more remote representatives than in the rest of the country, and given this consideration, the Commis- sion scheme has enough to commend it too. It ought not to be assumed, finally, that STV would produce more proportional results. Given the size of 58 + The Politico’s Guide to Electoral Reform in Britain CONCLUSIONS

he Jenkins Commission has lition governments may result. In produced a bold and in- years when the electorate has reached Tgenious solution to its difficult a decisive view about the need for a brief . It has retained the much-prized government change, AV-Plus will cut constituency links, but has nonethe- back the unwon landslide parliamen- less devised a way of securing some tary majorities which voters dislike parliamentary representation for the and which make a mockery of the idea vast majority of voters, wherever they of opposition in the Commons. But it live in the country. The incentives for will still allow for single-party govern- voting will be strengthened, and vot- ments whenever a party gains around ers will be more able to exercise 44-45 per cent of the vote with a strong effective choices, both locally and at lead over the second-placed party. the top-up area level. AV-Plus is well adapted to the con- The AV-Plus scheme is not fully tours of British party politics since the proportional and to the extent that it early 1970s when the current pattern is not, elections in Britain will con- of alignments was first defined. In the tinue to allocate seats to parties who two 1974 elections, perhaps in 1979 have not won them on their share of and certainly in 1992, AV-Plus elec- the vote. But it will be significantly tions would have brought forth either more proportional, and more reliably coalition governments, or minority proportional over different kinds of Conservative or Labour governments. elections, than plurality rule. It will This may seem a radical change, but spread representation. In 1997, it under FPTP the two 1974 elections would have given the Conservatives produced first a minority Labour gov- eight seats in Scotland, six in Wales, ernment, and then a marginal Labour and 28 in the metropolitan areas of government which soon lost its small England outside London – all zones overall majority. And though in 1992 where they won no or almost no rep- John Major won a working majority resentation under plurality rule. which lasted for five years, his own Similarly, in 1992, Labour would have intra-party coalition fragmented over gained 25 seats across south-east Europe, and the Conservatives only England and East Anglia (instead of enacted the Maastricht Treaty after six) and the Liberal Democrats 14 receiving crucial Parliamentary sup- seats (instead of none). port from the Liberal Democrats. In In close elections, the new system 1983, 1987 and 1997, the Jenkins will be nearly proportional, and coa- scheme would certainly have pro- The Politico’s Guide to Electoral Reform in Britain + 59 duced the same single-party govern- prevented the danger of ‘PR sclero- ments, although with smaller overall sis’ in British government, and majorities and much stronger incen- retained incentives for bold and tives for the governments to pursue broad-based popular leadership in the policies which commanded a broad Conservative and Labour parties to level of public agreement. seek overall command. British voters Thus it is only 1979 which might will still have the ability to ‘throw the have been completely different under rascals out’, but they will also escape AV-Plus, and we do not have suffi- the danger of large artificial majori- ciently good data on people’s second ties creating ‘electoral dictatorship’ preferences to be able to say whether and narrowly-based ‘strong’ govern- or not a Tory majority government ment being foisted on a public most would have resulted. of whom have not voted for it. In many ways, then, the Jenkins The key to realising this balancing scheme can claim to represent the trick was the decision to go for local ‘best of both worlds’. It creates the top-up areas with a limited quota of stronger incentives for people to use top-up MPs. This bold move cuts their votes which have meant that PR through the common aversion to elec- elections in Europe generally have toral systems which create the ‘flocks’ much stronger levels of turnout than of list MPs, beholden only to party in Britain. It prevents political parties managers and wheeling well above from developing strategies to appeal local accountability. Instead, every only to a large minority of the elector- top-up MP will be selected for a local ate, and it vitiates for ever the attempt area, often with an established histori- by people of many different ideologi- cal identity, good public visibility, and cal persuasions to capture one of the generally existing party organisations great British parties by organisational capable of organising a democratic means and then use its historical selection process for these candidates. weight to try and ‘bounce’ the elec- In just over half the top-up areas, there torate into accepting unpopular will be a single top-up MP, who will policies. always be the representative of the At the same time, the Commission main opposition party in areas where was alert to the danger that, in British one party is dominant. Where the conditions, a fully- proportional sys- Conservatives and Labour more or tem could vest too much political less share local seats evenly, the Lib- power in the hands of a large centrist eral Democrats, SNP or Plaid Cymru, party, the Liberal Democrats – mak- will win the top-up seat. In the 36 ar- ing them a ‘hinge’ party that would eas with two top-up seats, these third be required permanently for coalition or fourth parties will usually take one governments to be formed. By allow- of these extra places if they have not ing for single-party governments to already won a local seat. alternate with coalitions, Jenkins has None of these advantages means 60 + The Politico’s Guide to Electoral Reform in Britain that the Jenkins proposals are ‘per- people voting for more marginal can- fect’. No electoral scheme ever is. The didates, and the weight generally use of AV at local level is not good given to such preferences, would not news for the Conservatives. In 1997, be able to confuse the debate. Further, the Conservatives would still have SV is to be used for the election of been badly under-represented, why is London mayors and any other may- why Lord Alexander’s minority report ors thereafter. argues the case for a FPTP local base For smaller parties, like the Greens instead, and argues it well. An AMS or an anti-Europe party, it would be equivalent of AV-Plus would be more hard to win seats under the Jenkins proportional than AV-Plus itself. In formula, since the effective threshold 1997, it would have given the Tories would be about 8 per cent of the top- 25 more seats, as it would specifically up vote in the largest areas where work more proportionally in the hos- their chances of winning would be tile conditions of 1997 (though not greatest. Where local seats are not in necessarily so, say, when the Tories practice proportionately allocated, or are in the ascendant, as in 1983). It top-up areas are smaller, the real would also be more consistent with threshold may be as high as 11-12 per the AMS schemes intended for elec- cent. Such thresholds will be hard to tions to the Scottish Parliament, and climb, but at least smaller parties now Welsh and London assemblies, and have far more of an incentive to try perhaps in local government elections and build up support, and they will as well. Lastly, the FPTP element is be assisted by the use of AV in con- familiar and simple to use. stituency contests where people can But an AMS scheme would also split their votes and give, say, a Green have its drawbacks. It would not ex- candidate the nod with a first prefer- tend voters’ choices in constituency ence vote. elections, as AV (and the supplemen- Another virtue of the realism inher- tary vote) would; and so it would ent in the Jenkins package is the recreate tactical voting problems for space it leaves for recommendations many voters. It would also mean that to evolve as the public debate on elec- many local MPs would continue to toral reform unfolds. This is take their seats on a minority of the particularly true of the Commission’s local vote. decision not to choose between the Another alternative might be to three varying ratios of top-up to con- substitute the supplementary vote for stituency MPs, but to leave the final AV. SV-Plus would produce exactly choice to be debated further. Our the same results in terms of seats and analysis shows that the 80:20 mix government as AV-Plus, but the bal- would significantly improve the over- lot papers and counting would be all proportionality of the scheme simpler. Arguments against ‘wacky’ without substantially reducing the second and subsequent preferences of strength of local representation or the The Politico’s Guide to Electoral Reform in Britain + 61 chances of single-party governments References coming into power. The consequences for any one party would be negligi- Budge, I, Stability and Choice: a review of single-party and coalition government, ble. But adding a mere 15 extra top-up Democratic Audit Paper No. 15, 1998 MPs would reduce the number of top- Dunleavy, P, Margetts, H, and Weir, S, up areas with only one top-up MPs to Replaying the 1992 General Election: How a minority (29 out of 80) and reduce Britain would have voted under Alterna- deviations from proportionality appre- tive Electoral Systems,, LSE Public Policy ciably in all kinds of electoral Group and Joseph Rowntree Reform conditions. It would systematically Trust, July 1992. strengthen the representation of un- Dunleavy, P, Margetts, H, and Weir, S, derdogs and reduce rather more the Devolution Votes, Democratic Audit Pa- over-representation of locally-domi- per No.12, 1997 nant parties. Dunleavy, P, Margetts, H, and Weir, S, A government commitment to an Making Votes Count 2: Special Report on Mixed Voting Systems 80: 20 mix in the referendum on the , Democratic Audit Paper No.14, 1997. Jenkins proposals would be a sign of a party in tune with the more diverse Dunleavy, P, Margetts, H, and Weir, S, Open or Closed List Voting for the Euro- politics of contemporary Britain. But pean Parliament Elections: the ‘State of the before that we require a commitment Nation’ report, Joseph Rowntree Reform to a referendum before the next elec- Trust (also available from the LSE PPG), tion. February 1998. Dunleavy, P, Margetts, H, O’Duffy, B, and Weir, S, Making Votes Count: replaying the 1990s General Elections under Alter- native Electoral Systems, Democratic Audit Paper No. 11, Dunleavy, P, and Margetts, H, Report to the Government Office for London: Elect- ing the London Mayor and the London Assembly, LSE Public Policy Group and Birkbeck Public Policy Centre, 1998 Dunleavy, P, and Margetts, H, Report to the Independent Commission on the Vot- ing System: The Performance of the Commission’s Schemes for a New Electoral System, LSE PPG and Birkbeck Public Policy Centre, 1998 Dunleavy, P, Hix, S, and Margetts, H, Counting on Europe: PR and the 1999 Euro-Elections, Adamson Associates, Brussels (also available from LSE PPG), January 1998

62 + The Politico’s Guide to Electoral Reform in Britain About the Democratic Audit

he Democratic Audit of the ternational courses and seminars on United Kingdom inquires into democracy and freedom. Tthe quality of democracy and The Audit is based at the Human political freedom in the UK. The Rights Centre, University of Essex, Democratic Audit has published two Colchester, Essex CO4 3SQ and is major studies. Political Power and sponsored by the Joseph Rowntree Democratic Control in Britain, by Charitable Trust. You can find out Stuart Weir and David Beetham more about the Audit by writing to: (Routledge), analyses the power of The Democratic Audit, government in this country, its open- PO Box 18000, ness and accountability. Three Pillars London N1 7WW. of Liberty, by Francesca Klug, Keir Starmer and Stuart Weir, published in September 1996 (Routledge), audits the protection of civil and political rights in the United Kingdom. The Democratic Audit is continu- ing its inquiries into democracy and political freedom throughout the period of this government. The intention is to follow up these two “benchmark” studies at regular intervals, so that the quality of democracy and political freedoms in the UK can be measured over time. The current plan is to pub- lish the first follow-up report at the end of this Labour government’s term in office. The Democratic Audit has published ground-breaking reports on quangos in the UK and cooperated with Channel 4 Despatches on a docu- mentary on advisory quangos, Behind Closed Doors. The Audit undertakes consultancy and educational work in the UK and abroad, and organises in- The Politico’s Guide to Electoral Reform in Britain + 63 The map

The pocket in the back cover opposite contains a map, showing the 80 top-up areas which would be established in the UK under the Electoral Commission’s AV-Plus scheme. Top- up areas may have one or two top-up MPs. The map also shows how many seats the parties would have won in these areas in 1997 under the Com- mission’s ‘middle’ scheme, split 82.5:17.5 between constituency and top-up MPs. Two other schemes, with a mix of 80: 20 and 85: 15, are also put forward for debate. The index shows the number of seats which each party would hold in each area. The coloured areas indicate that a particular party would hold more than half the seats in the area (i.e., local and top-up seats). In 1997, Labour would have won over half the seats in 47 areas (coloured red); the Conservatives would have a similar majority in five areas (blue) and the Liberal Demo- crats in three (yellow). In the rest of the country (grey), no one party would be dominant.

64 + The Politico’s Guide to Electoral Reform in Britain