AN ANALYSIS OF PROPOSALS FOR CONSTITUTIONAL CHANGE

IN NEW ZEALAND

Prepared by

Penelope Brook Cowen, CS First Boston New Zealand Limited,

Tyler Cowen, George Mason University, and

Alexander Tabarrok, George Mason University

for the New Zealand Business Roundtable

September 1992

CONTENTS

Foreword The Perils of Setting a Constitutional Order

by Richard A. Epstein

Chapter 1 Approaching Constitutional Reform

1.1 Introduction

1.2 Why Democracy?

1.3 Analytical Framework

1.4 Normative and Positive Analysis

1.5 Overview

Chapter 2 Constraints on Policy-Makers in the New Zealand System

2.1 Introduction

2.2 Elections in Westminster Systems

2.3 The Role of Caucus

2.4 International Constraints

2.5 Ideology and Personal Policy Preferences in Parliament

and the Bureaucracy

2.6 Concluding Remarks: When Does a Westminster System

Work Well?

Chapter 3 Proportional Representation

3.1 Introduction

3.2 Preferential Voting

3.3 The Single Transferable Vote

3.4 List and Partial List Systems

3.5 Mixed-Member Proportional and Supplementary Systems

3.6 Proportional Representation and Westminster Systems

3.7 Concluding Remarks

Annex The Possibility of Perverse Results Under the Single

Transferable Vote

Chapter 4 Unicameralism and Bicameralism

4.1 Introduction

4.2 Determinants of the Strength of a Second Chamber

4.3 Bicameralism and the Flavour of Democracy

4.4 Second Chambers in Majoritarian Parliamentary Systems

4.5 Conclusions

Chapter 5 An Analysis of Referenda

5.1 Introduction and Definition of Terms

5.2 Analysis

5.3 Procedural Options

5.4 The New Zealand Indicative Referenda

5.5 Recommendations

Chapter 6 Summary and Concluding Remarks

6.1 Introduction

6.2 Summary of Findings

6.3 Further Options for Constitutional Reform

6.4 The Imperfection of Politics: Alternatives to Purely

Constitutional Reform

6.5 Concluding Comments

References

The authors wish to thank Susan Begg, Hon David Caygill, Richard Epstein, Tom Gilligan, Gary Hawke, Sir Kenneth Keith, Stephen Jennings, Roger Kerr, Chris Meads, Sir Geoffrey Palmer, Mark Toma and Bryce Wilkinson for their comments on earlier drafts of this study. Tyler Cowen would also like to acknowledge the financial assistance of the International Institute at George Mason University.

The views expressed in this study are not necessarily those of CS First Boston New Zealand Limited.

1.0 APPROACHING CONSTITUTIONAL REFORM

1.1 Introduction

The New Zealand system of government is currently under serious re-evaluation. Comprehensive constitutional reform in the near future is possible and perhaps even likely. Over the next 12 or so months, New Zealanders will be asked to consider three kinds of constitutional reform.

In September 1992, there will be a referendum on the replacement of the present first-past-the-post electoral system with some form of proportional representation. In addition to indicating their preferences between these two general modes for choosing MPs, voters will be asked to indicate their favoured form of proportional representation, choosing between preferential voting, the single transferable vote, a supplementary system, and a mixed-member proportional system. If the 1992 referendum produces a majority in favour of a switch to proportional representation, a further referendum will be held at the time of the 1993 general elections. In this referendum, voters will be asked to choose between the top-rated form of proportional representation and first-past-the-post.

Secondly, a 1993 referendum is also proposed on the reintroduction of a second chamber of parliament or "senate". There has as yet been no formal indication of how the members of such a second chamber might be selected, or what powers they might have over legislation.

Thirdly, a Citizens Initiated Referenda Bill is likely to be passed into law by the end of 1992. The Bill provides for non-binding referenda to be held where an officially sanctioned petition succeeds in attracting the signatures of 10 percent or more of eligible voters.

Constitutional reforms have been the subject of much recent discussion in New Zealand. Proponents of reform point to concerns about broken promises and weak accountability of political decision-makers to the electorate. More generally, they cite opinion poll evidence of a wide distrust of politicians and political institutions. What has been missing from much (though not all) of the debate has been a careful analysis of how successfully the proposed reforms are likely to be in addressing these sorts of concerns. Reform alternatives have received little systematic examination since the Royal Commission report of 1986.

In this study, we attempt to analyse the likely effects of the three proposed reforms. We take as our starting point an examination of the current Westminster parliamentary system. We focus on the incentives that operate under this system, on the premise that policy outcomes are shaped by the way in which constitutional rules mould the exercise of politicians' interests. We then consider how the present incentives would be changed by proposed reforms.

1.2 Why Democracy?

Our treatment of potential electoral and constitutional reforms considers democratic forms of government only. This focus represents our conviction that democracy is the only feasible system compatible with long-run freedom and prosperity. We see democracy as offering the following benefits:

• politicians have incentives to implement policies favoured by a large number of individuals;

• the citizenry can vote out politicians who act contrary to the national interest;

• democracy provides for an orderly succession of power;

• non-democratic forms of government require costly suppression, given that many individuals prefer democracy;

• no single individual can enrich himself or herself greatly at the expense of the rest of the nation; and

• democracy allows for free and open discussion of public policies.

These potential benefits of democracy are among the standards we will use to evaluate constitutional alternatives for New Zealand. While democratic forms of government are highly imperfect, we agree with Winston Churchill that democracy may be the worst form of government, but all other forms are worse yet.

Autocracy, the primary alternative to democracy, does not provide satisfactory incentives and liberties. Autocratic governments do not generally have the incentive to promote either freedom or prosperity. The autocrat knows that economic growth and a wealthy citizenry ultimately spell the end of autocracy. For this reason, autocrats tend to favour economies based upon special concessions and privileges, rather than free competition. The energies of an autocrat are usually devoted to suppression of alternative sources of power (both economic and political), rather than the creation of wealth.

We do find several examples of enlightened autocrats who have governed with a reasonable degree of benevolence and led their countries through periods of rapid economic growth (for example, Singapore's Lee). Nonetheless, benevolent autocrats are the exception; the incentives of autocracy favour control and attract control-oriented individuals. Furthermore, the autocratic system of government cannot ensure that the successors of a benevolent autocrat will be benevolent as well.

1.3 Analytical Framework

We analyse electoral reform by focusing on the operation of incentives. Incentives and constraints are two sides of the coin which shapes government policy; each incentive can be seen as offering a corresponding constraint. Incentives originate from voters, special interest groups, politicians, the bureaucracy, international forces and the like, and influence the policies that governments implement.

More specifically, we consider six incentives influencing government behaviour and consider how differing electoral systems affect the scope and influence of these incentives. These incentives overlap to varying degrees, but in our judgment represent the most important forces affecting government behaviour.

The six incentives we consider are:

(i) The incentive to mirror the preferences of the median voter

Electoral competition encourages politicians to propose and implement policies that appeal to voters. Under winner-takes-all or first-past-the-post electoral systems, politicians tend to converge to a centrist ideology and pitch their campaigns to the median voter. The median voter is the voter who stands in the middle of the political spectrum.

The incentive to appeal to the median voter can be illustrated by considering a one-dimensional political spectrum with different ideological positions aligned from left to right. Voters and candidates take positions along points of the spectrum. We assume, for the moment, that voters are distributed evenly across all political ideologies.

Left ---------------Median--------------- Right

With this distribution of preferences, politicians maximise votes and the chance of election by taking positions in the middle of the political spectrum.

Incentives for the election of politicians close to the views of the median voter are not absolute. Politicians may be elected on one set of promises and implement another. Furthermore, governments based upon regional or district elections will represent some weighted average of the medians in each district, rather than the median of the nation as a whole.

Other, more technical issues may prevent the median voter incentives from holding. If voters are clustered at the extreme left and right points of the spectrum, for instance, politicians have little incentive to gravitate towards the middle. Similarly, median voter incentives break down when political issues cannot be represented on a uni-dimensional ideological spectrum. In other cases, the electoral mechanism deliberately weakens the influence of the median voter by according representation to minority parties (i.e. proportional representation).

Despite these caveats, we still see median voter incentives as an operative force in most democracies. Changes in the strength of this incentive are a primary difference between Westminster systems and proportional representation.

(ii) The incentive to serve special interest groups and lobbies

Governments have incentives to appeal not only to voters but also to organised political interests and lobbies. These interests and lobbies include trade unions, business groups, farming associations, ethnic minorities, and so forth. Special interests may offer support, funding, favourable publicity and perks to politicians. All of these factors give politicians an incentive to weight electoral constraints less heavily and consider the views of organised minorities.

The term "special interests" may have either favourable or pejorative connotations. One person's "special interest" is another person's "endangered minority". We do not mean to prejudge when the influences of special interests are desirable. By special interests we simply mean minority groups who express their intensity of preference through organised political action.

Mancur Olson (1971, 1982) argues that special interest privileges are an important cause of economic stagnation. For Olson, the problem is that smaller, more concentrated lobbies often have more influence than more numerous but diffuse lobbies. When the benefits of a policy are concentrated across a relatively small number of individuals and the costs are diffused across a larger number of individuals, interest group support for the policy will tend to exceed the opposition to the policy. One hundred individuals who benefit greatly from a policy have an incentive to organise an interest group or lobby, whereas one million individuals who lose by a small amount do not.

Government does have incentives to favour the preferences of small, well-organised minorities and lobbies and the policies demanded by these minorities are not necessarily the policies that are good for the nation as a whole. Nonetheless, minority lobbying also may serve as a useful check upon the desires of a majority in a democracy. We accept Olson's point without concluding that all special interest activity is harmful.

(iii) The incentive to maximise revenue and redistribute resources from productive citizens to the government

Governments wish to maximise their tax revenue intake, all other things being equal. Higher revenue implies more money to spend on favoured projects, more political and bureaucratic perks, and greater influence and prestige for politicians. To varying degrees, government agents have an incentive to expropriate resources from citizens to serve their own interests, rather than the interests of the citizenry in general.

The citizenry, of course, does not necessarily prefer that the government maximise tax revenue. Nonetheless, when electoral constraints and checks and balances (both explicit and implicit) are weak, governments have latitude to act in this manner. As we argue below, different electoral and constitutional regimes affect the strength and nature of this tendency.

(iv) The incentive to favour particular regions and districts at the expense of other regions and districts

Political systems sometimes favour the redistribution of resources from one region to another. If representatives from one region are especially powerful, for instance, they can use their power to extract resources from the country at large. Examples of such behaviour are pork barrel policies, or subsidies or tariffs that favour particular regions. In other cases, regions may profit at the expense of other regions through the operation of special interest groups apart from the electoral process. Regionalism may also take the form of representatives focusing upon constituency service, rather than proposing policies to address national issues.

The extent to which electoral and special interest incentives operate at a national or regional level is largely a function of the electoral system. As we will see below, some regimes, such as preferential voting, encourage the predominance of local and regional interests, and other regimes, such as the mixed-member proportional system, focus electoral and special interest incentives upon the national level. Similarly, the regional focus is stronger in federal than in unitary systems.

(v) The incentive for politicians to indulge their own policy preferences or ideology

To the extent that other constraints and incentives are not binding, politicians will implement their own "ideology" or vision of what is good for the country.

Different electoral systems will give politicians varying degrees of freedom to indulge their policy preferences. The influence of political ideology may be either positive or negative. We do not attempt to resolve the age-old debate of whether politicians should serve as voters' agents (doing what the voters believe to be best) or as the voters' trustees (doing what the politicians believe to be best for the voters), but we do stress the importance of this distinction for evaluating constitutional reform.

(vi) The incentive to respond to international constraints

The need to compete in world markets limits the policy freedom of a country. Poor economic policies will damage international competitiveness and, in the limit, encourage the flight of resources to other political jurisdictions. International constraints operate through political desires to be re-elected, please special interest groups, and maximise government revenue.

International constraints play an especially powerful role in small, open economies such as New Zealand. To the extent that resource flight is easy, or that international trade comprises a large percentage of gross national product, international constraints channel the direction of political incentives. In the chapters that follow, we examine how different electoral systems affect the ability of politicians to respond to these international constraints, and, to some degree, shape the role that these constraints play.

1.4 Normative and Positive Analysis

Our analysis is primarily positive in nature and focuses upon how different electoral and constitutional regimes affect the relative strength of political incentives. We do not offer a running commentary on whether these changes in political incentives would be desirable or undesirable. When judging the desirability of different electoral systems, we do not use any single overriding normative standard or benchmark. Our primary purpose is to present an analytical framework for considering the relevant issues, rather than to persuade the reader to accept any particular set of normative recommendations. We do, however, offer our own recommendations and normative evaluations in the concluding remarks section of each chapter.

Throughout our normative remarks, we treat electoral and constitutional regimes as the relevant choice variables. Particular policies are not a choice variable. In other words, we do not consider options that first institute so-called "desirable" policies and then lock these policies into place with constitutional reforms. This procedure begs the question, given the constitution-level choices New Zealand is facing. The question is not whether we should institute policy X and lock this policy into place, but which constitutional regime will provide the best incentives to implement policy X in the first place. Furthermore, even if we could lock the "best policies" into place today we cannot do so for the future. The policies that will be adopted in the future will depend not only upon the policies adopted today, but also upon the constitutional regime chosen.

We also do not wish to burden the electoral system with enforcing or encouraging a very specific set of policy views. Specific policy preferences should be debated through democratic forums, rather than built into the electoral system. In our view, a desirable electoral system should allow the government to respond effectively to external pressures, both from the citizenry and from international competition. A good pragmatic outcome is a government which presides over a high rate of economic growth and respect for civil liberties, while providing policies that are satisfactory to a wide range of the citizenry. While such outcomes are not always possible, we can be biased against electoral systems that make such outcomes less likely.

1.5 Overview

Each of the chapters of this study considers the operation of a particular constitutional system or component of a constitutional system. Chapter 2 analyses the incentives that operate under the current Westminster parliamentary system in New Zealand. Westminster systems allow for rapid change but these changes are also subject to constraint. In a Westminster system, electoral constraints, intra-party constraints, and international constraints are the primary determinants of policy. We consider also the extent to which politicians can foist their favoured policies upon the electorate in a Westminster system. Chapter 2, which is analytical in its thrust, is essential for understanding the later comparisons with alternative electoral mechanisms.

Chapters 3 through 5 examine electoral and constitutional alternatives to the current New Zealand system, focusing upon the possibilities currently being debated. Chapter 3 considers the four forms of proportional representation on the September referendum: preferential voting, the single transferable vote and the mixed member proportional and supplementary systems. We address how each reform would affect the incentives of individuals in government and the nature of government policy.

The four forms of proportional representation under consideration would have significantly different effects upon New Zealand government. The mixed member proportional system, for instance, would herald a move towards the coalition governments found on the European continent. The effects of the supplementary system would be less radical, and would depend upon the number of seats allocated to proportional representation. The single transferable vote, used for the Irish legislature, would lead to weak party structures and incentives for local constituency service. Preferential voting would leave the basic structure of the Westminster system intact but would tally second-place preferences and allow these preferences to count when no party wins more than fifty percent of the vote. We compare these different reforms with each other and with the Westminster system. We find both pros and cons when comparing the Westminster system with these electoral alternatives, but favour the mixed member proportional system over the other forms of proportional representation and the Westminster system over the mixed member proportional system.

Chapter 4 focuses upon whether New Zealand should reinstitute a second legislative chamber. We describe how a second chamber must be constituted differently from, and share comparable powers with, the first chamber if it is to act routinely to check legislative outcomes. We then consider the implications for representation and accountability of instituting this kind of strong bicameralism. In New Zealand, the introduction of a strong second chamber would fundamentally alter the nature of accountability in government, and undermine the effective operation of the Westminster system. A weak second chamber would, by contrast, be compatible with the maintenance of a Westminster system. However, the capacity of a weak second chamber to enhance the operation of the present system, at reasonable cost, would be strictly limited.

Chapter 5 deals with citizens' referenda and considers alternative forms such referenda might take. We examine the incentives created by different referenda, and how these incentives would affect government policy. The form of referendum proposed in the New Zealand Citizens' Initiated Referenda Bill gives the citizenry only weak powers and would have little effect upon New Zealand government. Other, more binding, forms of referenda would, in general, make government more representative of majority opinion, place additional constraints upon politicians, weaken the protection of minority rights, and weaken the role of political parties. We recommend that the present proposals for citizens' initiatives be made more accessible, but remain non-binding. We also recommend the introduction of binding protest referenda, which would enable legislation to be struck down within a limited time after its passage, and the use of binding, government-controlled referenda on constitutional issues.

Chapter 6 summarises our findings and canvasses possibilities for further research. We briefly consider alternative constitutional reforms which might enhance the accountability of politicians and the quality of policy without altering the fundamental features of the New Zealand Westminster system. Without denying that improvements are possible, we note that the ability of any programme of constitutional reform to improve upon the workings of democracy is limited. Choice between systems of government is a choice between necessarily imperfect mechanisms for identifying, aggregating and satisfying the interests of diverse voters. In this context, we note that at least some of the present concerns about New Zealand's system of government may be better addressed by redefining the scope of government, rather than altering the rules by which government is conducted. However, we argue that the proper scope of government should and indeed must, in an existing democracy, ultimately be subject to a democratic test.

2.0 CONSTRAINTS ON POLICY-MAKERS IN THE NEW ZEALAND SYSTEM

2.1 Introduction

Government in New Zealand is based on a Westminster model inherited from the UK. The key characteristics of the particular version of the Westminster model that has developed in New Zealand are as follows:

• There is no entrenched, written constitution. The constitution consists, instead, of a combination of statutes, procedural rules and conventions.

• Parliament is the source of sovereignty. The party forming the majority in parliament constructs a cabinet from its elected members. Cabinet members are collectively and individually responsible to parliament. The executive and legislative divisions of government are linked through the cabinet.

• Party discipline is central to the conduct of parliament. Conscience votes are used rarely, and party constitutions and whipping are used to facilitate party unity in voting on legislation.

• Parliament is unicameral, and policy-making powers are largely concentrated at the national level.

• There is plurality rule. Members of parliament are elected from single-member districts, under a first-past-the-post voting system.

• The bureaucracy is largely anonymous, career-oriented and non-partisan.

• The judiciary is independent, and has no direct powers of legislative review.

The popular perception of the Westminster system is that it gives governments - or a few key members of the executive - largely unlimited power to define policy. Policy changes can be radical, numerous and rapid. The relative ease with which policy may be changed in a Westminster system should not, however, be confused with an absence of constraints on policy decisions. The relative paucity of formal constitutional checks and balances does not imply that no checks and balances exist. Policy-makers cannot choose freely from a completely open set of policy options.

In this chapter, we show how policy options are constrained in the New Zealand system of government. We concentrate on four, interrelated, determinants of policy outcomes.

First, we consider the role of elections in shaping the policies that governments pursue. In a unicameral parliamentary system where members are elected on a first-past-the-post basis, governing parties generally have a clean majority, conferring a considerable degree of discretion over policy. However, voters can also, routinely, overturn governments in elections. Relatively small shifts in voter support can make the difference between winning and losing an election. Competition for electoral favour results in a continuing constraint on the behaviour of government and opposition alike. The nature of this constraint is discussed in Section 2.2.

Secondly, governments are constrained by the need to maintain the unity of their own party in the legislature. On a day-to-day basis, the ability of a government to secure the necessary parliamentary support for its policies depends not on convincing the opposition, but on the support of its own caucus. By convention, caucus presents a united front in parliamentary debate and voting. Caucus support must ultimately, however, be earned rather than assumed. Backbenchers concerned with their prospects of re-election will generally be unwilling to support policies they expect to be unpopular. Unless a government is willing to make concessions in its policy stance to reflect the concerns of its backbenchers, its ability to deliver a unified vote in parliament will be weakened. The daily constraints imposed by caucus further support the tendencies of elections to produce policies favoured by the median voter. The way in which caucus members continually enforce the electoral constraint is discussed in Section 2.3.

Thirdly, governments face a range of international constraints, both political and economic, which limit their policy options. Especially in a small, open economy such as New Zealand, policy must take account of the need to compete for resources in international markets. Policies that result in the loss of capital or labour resources, weakening economic performance, will tend to cost electoral support and endanger political perks. The impact of the flow of resources between countries on policy options is discussed in Section 2.4.

Fourthly, the policies that governments implement are affected by ideas and ideologies. In a Westminster system like New Zealand's, ideological preferences can shape policy in two ways. First, governments with particular policy preferences can influence electoral support for those policies by explaining their rationale and timing their implementation carefully. Favoured policies will be implemented to ensure, as far as possible, that the electorate sees the benefits in time to reward the government at the next election, and the policy itself shifts resource flows to increase support for the policy, once the election comes. Secondly, the preferences and ideologies of the bureaucracy, which has the benefit of permanence, can to a certain extent shape or moderate policy. These two conduits of ideology into policy are discussed in Section 2.5.

In a Westminster system, the power of citizens is concentrated in their ability to overturn and replace governments at election time. The degree of discretion that a government can exercise over policy is determined by the way in which pressures outside of elections combine to sharpen the electoral constraint, and also by its ability to shape the electoral constraint.

The pressures described in this chapter will vary in their impact between countries and over time. The ease with which policy can be changed is likely to be more or less constant, so the degree of policy discretion will vary. In Section 2.6, we discuss how the different forces described in Sections 2.2 through 2.5 interact. We then consider the conditions under which the Westminster system is most likely to produce effective and responsive government.

2.2 Elections in Westminster Systems

Politics in democratic states are characterised by relatively peaceful, institutionalised competition between political parties. Citizens, voting in parliamentary elections, are the arbiters of this competition. The exercise of a vote is the primary formal means by which the success of one or other party is tagged to citizens' preferences and perceptions. The prospect of having to compete in general elections brings other, less formal and continual, kinds of competition in its wake.

The significance of elections as a constraint on governments and a determinant of policy varies among constitutional systems, but is most potent in majoritarian and unicameral systems such as New Zealand's. In this section, we discuss the way in which electoral competition affects policy.

In a unicameral parliamentary system where members are elected by a first-past-the-post system, voters have an unmatched ability to fire and replace governments. Governing parties routinely have a clean majority of seats. And governments are also, quite routinely, overturned in elections. In a first-past-the-post system, relatively small shifts in voter support can make the difference between winning and losing an election. Governments are for this reason particularly susceptible to swings in public opinion. The result, in Popper's words, is "a continual process of self-criticism" by the party in power and the party that could, at the next election, assume that power.

Parties compete both for direct influence over policy, and for electoral favour. In a Westminster system, the ability of an opposition to influence legislation directly is limited. This is particularly so when majorities are solid, conscience votes are rare, and whipping is strictly enforced. However, it is possible for an opposition, as well as a government, to use the formal and informal institutions of parliament to shape and inform public opinion in ways that affect their electoral prospects. These institutions include parliamentary debates, question time, private members' bills and the select committee system. In addition, politicians can use a variety of extra-parliamentary means to shape the electoral contest, ranging from media coverage and debate to constituency work. The nature of the electoral constraint is also shaped by the role of political parties in selecting and supporting candidates and debating broad policy positions.

2.2.1 Constraints of Parliamentary Convention

Parliamentary convention constrains policy through several different media, including debate, select committees and the requirements of parliamentary procedure. Each of these conventions operates outside of a written constitution, but nonetheless constrains the choices of politicians.

Parliamentary debate shapes and legitimises policy, and maintains political accountability to electors. In particular, parliamentary debate provides a regular forum through which both the government and the opposition can advertise their position on policy issues to voters. Through the press, the reporting of parliamentary debates transmits information about policy to the public. Debates also allow routine scrutiny of the government by the press and the electorate. On the whole, governments prefer to avoid battles in the debating chamber, and moderate legislation accordingly. Concessions to opposition arguments may not be apparent in the course of a debate, but may well be implicit in a government's subsequent actions (Robinson, 1978b).

Question time and the opportunity to introduce private members' bills provide opportunities for criticism and for the extension of parliamentary debate into areas outside the government agenda. Neither is likely to have a significant, immediate impact on the path of policy. Question time provides for increased public scrutiny of the conduct of ministers and their departments, but easily becomes formalised. Private members' bills are rarely successful in the sense of passing into legislation. However, together with the media coverage they generate, they do inform the electorate.

The select committee process is another convention through which backbench politicians from both the government and the opposition can influence legislation.

The select committee process has been greatly strengthened over the past 30 years (beginning with the establishment of the Public Expenditure Committee in 1962). Traditionally, select committees heard petitions and examined proposed legislation. It is only since 1979 that the majority of bills have been referred to select committees as a matter of course.

Select committees play roles in constraining and shaping policy that extend beyond those implied by their formal powers. At their best, they provide for informed discussion of policy issues in a manner not possible (or at least unlikely) in formal parliamentary debates. These committees help to transmit information from the general public to the government, and allow citizens to exert an influence on policy in a more specific and detailed way than is possible through the ballot box. They also provide a forum for a particular kind of competition between government and opposition MPs. Select committee hearings give the participating MPs an opportunity to develop their image (establishing their concern and expertise) before interest groups. Further, although select committees are numerically dominated by government members, and it is relatively rare for an opposition member to be given the chairmanship of a select committee, they provide a venue in which compromise is possible (if not necessary). Finally, cabinet ministers proposing bills, and wishing to be supported through the select committee process, are likely to be influenced by the composition of the relevant committee, both opposition and government.

At a more routine level, the ability of governments to pass legislation is constrained by the requirements of parliamentary procedure. Some matters may be dealt with by means of regulations, and some may be dealt with under parliamentary urgency. But the great majority of policy must be implemented through the normal legislative process, set out in the standing orders. For complex legislation, this process can be protracted, with extensive public consultation and/or expert review. Parliament may not be in session when a government decides a new piece of legislation is needed, or that an existing piece of legislation should be amended. When parliament is in session, there may be no room on the legislative programme. If time can be found, the prospect of scrutiny that attends parliamentary debates and select committee proceedings may be a disincentive to pursuing a change whose political acceptability is in doubt (Mulgan, 1989).

2.2.2 Political Parties and the Electoral Process

So far, we have concentrated on how the operation of parliament reinforces the electoral constraint between elections. In this section, we turn to the role of political parties in shaping the electoral process, and the policy options available to politicians.

Parties simplify the democratic process for electors. The existence of parties provides electors with an ideological screen with which to sort and identify candidates. In a first-past-the-post parliamentary system, the existence of parties also gives voters a means of signalling where they want policy-making power to lie. Selection of a local MP therefore reflects not only an assessment of his or her personal merits, but also more general preferences about which party should be in government. Consistency between what parties appear to offer at elections, and what their parliamentary wings do when they achieve power, is important to the stability of the system and the credibility of candidates. Party ideology, party manifestos, and the bonds formed between party members provide a framework within which to operate and a basis for mutual commitment and cooperation for MPs. The importance of party affiliation to electoral success (re-election of party dissenters who subsequently stand as independents is rare) provides an incentive for tolerance and flexibility.

Broad political parties are most influential in the activities surrounding general elections. Some of this influence is direct. For example, both major parties in New Zealand have a direct role in candidate selection. In both parties, too, non-parliamentary members are involved in the formulation of election policy. Parties also exert an indirect influence on the policy platforms adopted by their parliamentary wings. Parties provide much of the on-the-ground support for electoral candidates, raising funds, organising meetings, checking the electoral roll, and canvassing and courting voters. Their enthusiasm for these tasks will be influenced by the extent to which they expect successful candidates, whether in government or in opposition, to pursue the policies endorsed by the wider party. As elections approach, parliamentary parties, and the government in particular, will wish to avoid alienating those who have signed up to support them.

The role and influence of parties on the electoral process may change over time, as the conventions of political campaigning change. For example, election campaigns in New Zealand in recent years have become increasingly presidential in nature, with leadership receiving a proportionately higher emphasis vis-a-vis policy than formerly. (The depiction of elections as contests between leaders is, indeed, consistent with the reality of the executive role in the policy-making process.)

Recent elections have also seen greater diffidence about the use of firm and credibly binding manifestos. Whereas prime ministers like Kirk and Muldoon had used manifestos as a constraint on policy and a tool for disciplining wayward members of caucus, Labour in the 1980s placed relatively little emphasis on specific policy promises. The economic policy on which Labour campaigned in 1984 was vague - a state of affairs that can only partially be explained by the fact that it was contesting an early election. In 1987, Lange campaigned explicitly on the basis that the government's record, not a manifesto, should be assessed by voters. This did not appear to cost Labour electoral support; indeed, the increased majority won by Labour in the 1987 election could be regarded as an endorsement of the approach (Mulgan, 1990). In the 1990 election, there was a return to more formal manifestos. Manifestos remain, however, schedules of promises that it may be prudent to keep, but which can be broken.

However important party unity may be for practical reasons, governments are ultimately accountable not to their parties, but to the electorate. The party check on policy is ultimately subordinate to the check imposed by elections. In the normal run of things the positions adopted and promoted by parties act as a constraining influence on their parliamentary wings. In times of crisis, however, when radical policy changes are required, parliamentary parties may depart from the policies represented by their party base. Their political survival will then depend on the extent to which they can convince voters that a departure from the party's traditional policy platform is broadly beneficial.

2.3 The Role of Caucus

In Section 2.2, we discussed a number of ways in which competition between political parties shapes and constrains the policies that governments can develop in Westminster systems. But however potent the threat of electoral loss may be, the fact remains that competition between parties or factions in the period between elections is weaker than in congressional and other mixed systems. A government may be embarrassed in parliamentary debate, may make strategic compromises in select committees, and may be sensitive to media coverage. Ultimately, however, a party with a parliamentary majority need not heed the wishes of the opposition.

In a system where the winner, at least in principle, takes all, the stakes shift from inter-party competition to intra-party competition. The prime minister and perhaps only a handful of key cabinet ministers appear to have near boundless power. Becoming part of the executive, or at least influencing it, becomes crucially important. As the discretionary power available to the executive increases, so do the forces influencing and bidding for this power. For this reason, constraints are not removed but shifted to another form - intra-party and caucus constraints.

In this section, we discuss the ways in which competition within a governing party can operate to shape the policies endorsed by the executive. We focus on competition for the leadership and for cabinet positions, on the role of caucus, and on the role of such institutions as collective responsibility in the cabinet, caucus secrecy and whipping. In the process, we show how caucus can filter electoral pressures through to the executive on an ongoing basis.

2.3.1 Competition for Executive Positions

The power to initiate policy, and to ensure that policy passes, lies primarily with the prime minister and the cabinet. The greater the power of this group, the more intense will be competition for membership. Competition within caucus for party leadership and for positions in the cabinet will, in turn, shape the policy positions adopted both by the successful and by aspirants.

2.3.1.1 Party Leaders

The leadership of a party's parliamentary wing is determined by the vote of caucus. Caucuses can and do force or encourage the removal of leaders they think are under-performing. The small size of New Zealand caucuses, and the fact that it is caucus, rather than the wider party, that elects the parliamentary leader, make party leaders in New Zealand more vulnerable than their colleagues in, say, Britain or Canada.

The last two decades in New Zealand afford ample evidence of the vulnerability of party leaders. John Marshall resigned as leader of the National Party in 1974, lacking the ability to gain the support of the majority of his caucus. Robert Muldoon, after surviving one challenge to his prime ministership in 1980, was ousted from his leadership of the (by then) opposition in 1984. His successor, Jim McLay, was replaced by Jim Bolger in 1986. The experience of the Labour party has been equally turbulent. In 1983, Bill Rowling declined to stand for re-election to the party leadership when a loss to David Lange appeared almost certain. In 1989, David Lange resigned from the prime ministership in the face of caucus criticism over his handling of economic policy issues and, in particular, the removal of Richard Prebble and Roger Douglas from the cabinet. His successor, Geoffrey Palmer, resigned prior to the 1990 election, in the face of increasing dissent in caucus and the prospect of electoral defeat.

In selecting a prime minister (or potential prime minister), caucus members are influenced by a range of criteria. They are generally interested in the candidates' leadership qualities and personal priorities, as well as their reputations for standing by promises - in particular, promises made within the caucus. They are interested in the way in which candidates are likely to allocate cabinet posts. And they are interested in whether candidates are likely to take account of the interests of non-cabinet members of caucus. Candidates perceived as unwilling to listen, unable to compromise, overly prone to favouritism, or simply maverick, are unlikely to succeed. Successful candidates who turn out to be these things are, in the medium to longer term, unlikely to survive.

Competition for leadership increases the accountability of leaders. The reputation of incumbent leaders is constantly assessed, often by comparison with rivals. If a leader's performance wavers, rumour, plotting and threats of replacement flourish. The positive result of these threats is that leaders are compelled to listen both to backbenchers and to influential colleagues (Weller, 1985). Unlike the US congressional system, for example, the performance of leaders is assessed by party members on an ongoing basis. The monitors of executive power hold an ongoing threat of dismissal in their hands. In addition, a party with a reputation for ignoring backbenchers will weaken its electoral prospects, because voters prefer candidates who will get a fair hearing in caucus.

2.3.1.2 Cabinet Ministers

Competition for cabinet positions can also reinforce the accountability of the executive to backbench interests.

The methods by which cabinets are selected differ between the two main parties in New Zealand. In Labour governments, caucus elects members to a defined number of cabinet positions. Cabinet rankings, and specific portfolios, are then decided by the prime minister. In National governments, the prime minister decides the membership of the cabinet, as well as cabinet rankings and the allocation of portfolios. In both cases, however, the prime minister is in practice constrained by the preferences of caucus. A National prime minister is unlikely to have much greater freedom than his or her Labour counterpart in deciding who will be in the cabinet. And both will be influenced by the preferences and concerns of caucus in deciding how portfolios are allocated. (In addition, both are likely to face the practical constraint of finding a sufficient number of suitable candidates in a small caucus.)

To ensure the passage of favoured policy programmes, prime ministers need to maintain caucus unity. The creation and reinforcement of unity will be a critical factor in deciding cabinet seniority and allocating portfolios. Cabinets are likely to include not only loyal supporters of the prime minister, but also potential or declared challengers. Potential dissenters, once included in the cabinet, will be bound by the collective responsibility of cabinet. (Backbenchers, by contrast, are free, within bounds, to criticise government policy.) At the same time, the inclusion of dissenters in the cabinet, and the assurance that this offers their caucus supporters of some influence over cabinet policy, can assist in the promotion of caucus unity. The cost of this assurance is an increase in trade-offs within the cabinet, implying a possible moderation of policy. The prime minister's ability to reshuffle his or her cabinet, including the power to dismiss recalcitrant ministers, provides some assurance that the power of dissidents in the cabinet will not be exploited - in particular, by contraventions of the conventions of collective responsibility and cabinet secrecy.

As recent experience in New Zealand has shown, the unity of the cabinet and of caucus can break down, and do so publicly. However, the rarity of such events, given the tremendous scope for disagreement among members even of the same party, attests to the strength of competitive and institutional forces for unity. Unity does not imply that the cabinet and caucus rubber-stamp all policies desired by the executive. Rather, the executive proposes policies that will succeed in achieving a relative degree of unity.

2.3.2 The Day-to-Day Powers of Caucus

After a cabinet has been selected, the primary means by which backbenchers can influence the direction of policy is by attempting to influence ministers. Their main bargaining counter is cabinet reliance on caucus support for the passage of legislation. Cabinet must make at least some concessions to backbench influence if it is to retain the degree of caucus support and unity on which the real power of the cabinet depends. The principal forum for this formal trade in support and influence is the weekly caucus meeting, but in practice the barter is continual.

Caucus and caucus committee meetings bring together the cabinet and backbenchers to debate policy issues under a rule of secrecy. The secrecy of caucus deliberations, like the secrecy of cabinet deliberations, enables at once frankness in the discussion of differences, and the presentation of a unified face to the outside world. These conventions are reinforced by the whipping system, which delivers a unified vote in the House. In the case of the Labour party, the party constitution requires unified voting.

Even if the cabinet has the numbers to impose its will on caucus unilaterally, a policy of systematically ignoring caucus would be unsustainable, given the threat to party unity. Secrecy is a convention subject to change. If backbenchers are satisfied with their ability to influence the cabinet through caucus, they are likely to abide by this convention. The threat of breaking secrecy confers on them a degree of power to achieve such influence. Cabinet intransigence towards wider caucus opinion would encourage backbenchers to impose pressure on the cabinet directly, for example through leaks to the media, or through pressure groups or the parliamentary opposition (Breton, 1991).

Caucus members also retain the option of crossing the floor to vote with the opposition on policies that they oppose. In practice the use of this option is rare. It is, of course, most effective where the government's majority is small. For example, the 1981-1984 National government, which had a one-vote majority, was particularly susceptible to backbench dissent. Its perceived inability to govern, following a series of floor-crossings by both government and opposition members, was used as a justification for an early election in 1984,.

The size of the cabinet vis-a-vis caucus is also important. The smaller is the cabinet relative to caucus, the greater is the likely influence of backbenchers as a group, but the smaller is the cost of individual backbenchers crossing the floor. Where caucuses themselves are small, sheer camaraderie may be a uniting factor. Where caucuses are habitually large relative to cabinets, as is the case in large parliaments, the prospects of attaining cabinet rank will be small, and the energies of backbenchers may be devoted more to select committee activities than to attempts at influencing policy through caucus.

Unanimity in party voting in the House, and silence on matters of disagreement, necessitate flexibility in the policy line taken by the prime minister and his or her key ministers in the face of opposition from their parliamentary colleagues. The apparently fragile conventions of secrecy and party discipline can compel greater accountability in policy-making than is at first apparent.

2.4 International Constraints

So far, we have concentrated on domestic constraints on policy makers. Governments' policy options are also shaped by international constraints. In this section, we concentrate on the role played by economic constraints, and in particular on the pressures created by the potential for resource flows between countries.

Through the present century, the emphasis in election campaigns has increasingly been on economic performance. Governments seek support and are judged on the capacity of their policies to generate economic growth and prosperity. The performance of governments in this sense depends on their ability to attract and retain resources in increasingly competitive international markets.

Historically, governments have taken two broad kinds of approach to this problem. The first has been to attempt to seal the borders, whether through the brutally autarchic approach of a Burma or the strictly enforced restrictions on exit of the former communist countries, or through systems of exchange controls and trade barriers. Autarchy of varying force was the prevailing approach in both socialist and so-called market economies through the middle decades of this century. Autarchic policies held strongly in New Zealand, for example, through from the late 1930s to the early 1980s.

The second approach is to enable resources to move in and out of a country relatively easily, but to design policy so that they will tend to come and stay. (Freedom of exit is a crucial component, based on the recognition that the ability to attract resources, and the cost at which they can be attracted, will reflect the ease and costs of exit.) This is the approach that has been adopted in New Zealand since the mid-1980s. It is also being attempted, to a lesser or greater extent, in many other countries, from Latin America to Central Europe.

The nature and strength of resource flow constraints on governments will differ according to the approach adopted. Strongly protectionist governments may, at least in the medium term, succeed in sealing themselves off from international pressures on their policy mix. Rigid exchange controls and a fixed exchange rate, for example, may inhibit the flow of capital out of the country in the event of an adverse policy development. Generous production subsidies may retain investment that would otherwise flow elsewhere. In the longer run, however, resources will flow into and out of even highly protected economies in response to internal policy developments. Evidence on this point is provided not only by the economies of central and eastern Europe, but also by New Zealand's own experience in the late 1970s and early 1980s. Heavily protectionist policies, reinforced in the wake of Britain's entry to the EC and the oil crises of the 1970s, imposed increasing fiscal and efficiency costs, reflected in persistent fiscal deficits, capital outflows, and heavy dependence on the government.

In addition, governments are sensitive to international ratings of their performance. In particular, country ratings applied by agencies such as Moody's and Standard and Poor's have a direct effect on the cost of funds to governments. Strenuous efforts may be taken to maintain a high rating, particularly in the areas of fiscal and monetary policy. These factors will tend to reinforce natural pressures arising from resource flows.

International constraints should not be thought of as a force contrary to the operation of democracy, or as a force that requires politicians to violate the wishes of voters. With the exception of explicit written treaties, international constraints do not have independent force apart from how they influence the economy and voters' pocketbooks. International constraints do, however, induce wealth-maximising outcomes in areas where voters are likely to be ignorant of economic policy issues. For instance, voters may not be informed about the withholding tax on foreign investment, but international constraints will encourage politicians to adopt policies that will increase foreign investment and domestic real wages.

2.5 Ideology and Personal Policy Preferences in Parliament and the Bureaucracy

Although we have focused so far upon the constraints operating in a Westminster system, the Westminster system does also provide considerable scope for politicians to implement policies of their own choosing. Political agents may indulge their individual policy preferences either by influencing the electoral mechanism or by operating through parts of government that are not subject to election (for example, the bureaucracy). Starting with elections and then moving on to the bureaucracy, we examine how politicians may enjoy independent latitude and then consider how this independence affects the content and nature of policy.

The ultimate binding constraint in parliamentary systems, the electorate, has the power to recall the government only once every three years in New Zealand. Between elections, governments may attempt to change the preferred policies of the electorate. While the electoral constraint is ultimately binding once in operation, the government can direct and influence preferred policies by changing constraints. By influencing electoral preferences, politicians have the scope to implement their own views of what is good for the country. When speaking of government influence over electoral preferences, we mean the ability of the government to shape the economic environment.

Governments may influence electoral preferences in several ways. First, governments may influence electoral preferences by inducing resource flows that are hard to reverse. A typical example is the liberalisation of capital markets in New Zealand in the mid-1980s. The electorate may or may have not favoured capital market liberalisation prior to reform, but the reforms themselves created a constituency for maintaining reforms, once in place. Firms developed greater international linkages, jobs were created to manage capital flows, the floating exchange rate increased the strength of some sectors, and so on. The losers from the new policies were still in opposition, but a new class of well-identified beneficiaries had been created.

Governments can also implement their preferred policies by attempting to constrain their successors. Successive governments can be (partially) constrained by the judicious timing and use of policies. A current government, for instance, may try to constrain its successors by exposing them to international pressures that limit the room for action. The institution of a market-determined flexible exchange rate in 1985, for instance, performed this function. The costs of fixing the exchange rate subsequently increased, as the reimposition of capital controls would have imposed considerable practical problems and reduced confidence in New Zealand financial instruments and securities.

A government can also attempt to enshrine the institutions by which a preferred policy is implemented. For example, the Reserve Bank Act of 1989 attempts to create a durable and transparent mechanism for monetary stability. It gives the Governor of the Reserve Bank the responsibility for meeting pre-specified inflation targets. While the act is vulnerable to amendment or repeal, the legislative process required to change the Act is more cumbersome and overt than that required to modify an unlegislated policy. Perhaps most importantly, once the Act is in place, the political costs of resuming an inflationary monetary policy increase greatly. In order to resume inflation, the New Zealand government would have to acknowledge specifically that inflation was now its goal, and would be spending the considerable reputational capital it has built up through the Act.

The second method of government influence over electoral preferences involves the timing of policy so that policy benefits are apparent before elections (and costs appear after elections). By using the lag between implementation and the next election to educate voters about the benefits of change, governments have scope to shift voter sentiments and preferences to accord with their own policy agenda.

For example, by implementing the Employment Contracts Act in the first year of its term, the present National government increased the probability that beneficial effects in terms of productivity improvements and employment would be felt before the 1993 election. In the process, it increased the probability that a constituency opposed to centralised wage bargaining would be well-established before the election.

An example of the relative vulnerability of "late" policy initiatives is afforded by the Employment Equity Act 1990, which established an Employment Equity Commission with responsibilities for equal employment opportunities and comparable worth. This legislation came into force on the eve of the 1990 election, which saw the replacement of a Labour government with a National government. National had given indications some time before the election that it would repeal the legislation, and duly did so in December 1990. The new Employment Equity Commissioner had only a very limited period in which to establish an institutional domain and constituency. Comparable worth claimants had little time to lodge a large body of cases with the Commission. Further, there had not been time to see any of these cases through to their conclusions; the Commission did not have a chance to build up a list of satisfied customers. The costs of change were therefore relatively low.

Only some kinds of policy change are directly susceptible to this kind of manipulation. For example, governments will be wary of pursuing policies whose costs become evident early, but whose benefits are likely to accrue beyond the date of the next election. A number of policies that involve major restructuring of the economy may fall into this category. Adverse employment effects are felt early, obviously and painfully, while the wider benefits of increased productivity and standards of living may take some years to be recognised. The shorter the period between elections, the more binding this constraint on policy is likely to be. To the extent that radical policy programmes pay off more slowly than relatively incremental changes, policy will be more conservative the shorter is the parliamentary term.

Governments will be more willing to implement policies whose benefits are felt early, but whose costs are disparate and fall late, even if they expect their successors to reverse these policies. For example, a government may selectively distribute import quotas or production subsidies to favoured constituents, to the longer-term detriment of the wider community, even if it recognises that a successor may redistribute (or eliminate) this patronage. This propensity will, however, be checked if constituents have a preference for relatively small, stable benefits over large, possibly short-lived, ones.

Westminster systems, with their strong reliance upon electoral constraints, and weak reliance upon explicit division of powers, provide considerable scope for politicians to pursue their own policy preferences. Herein lies the source of many complaints about the Westminster system. The implicit New Zealand constitution, while reflecting preferences tightly through the electoral mechanism, does not prevent politicians from influencing electoral preferences through manipulating the economic environment.

Versions of this complaint are heard from both sides of the political spectrum. On one side, those who favour a relatively unregulated market economy argue that latitude for political ideology allowed various New Zealand governments to pursue welfare and protectionist policies that aggravated New Zealand's economic slide by reducing the flexibility of the economy. Once New Zealand governments had instituted protectionism, for instance, pressures were created that made this protection difficult to remove.

On the other side of the political spectrum, those who favour a more interventionist role for the state blame the Westminster system for the free market reforms of the Roger Douglas period. Many of these reforms, once instituted, have proved relatively difficult to reverse. Even if the electorate did not approve of these reforms initially, the electorate may, however, come to recognise their longer-term benefits, or at least be unwilling to bear the costs required to reverse the reforms.

2.5.2 The Bureaucracy

The potential for elected politicians to pursue their personal policy preferences produces pressures for other independent sources of power in government. While the New Zealand government does not possess a judiciary with the formal power to overturn legislation, the bureaucracy is a power centre independent of the elected government to a certain degree. As such, the bureaucracy may play two roles: constraining the executive and implementing its own policy preferences.

The bureaucracy receives its power from providing politicians with policy advice and implementing (or ensuring the implementation of) policies that governments adopt. Both of these powers can affect the final outcomes experienced by citizens. Policy advice is important because ministers rely heavily upon their departmental advisors for information and insights. Maintaining a good relationship with these advisors is important for eliciting information and developing effective policies. Policy implementation is important because rules and regulations must be drawn up in detail and enforced if they are to prove effective.

The bureaucracy has both advantages and disadvantages when engaged in a conflict with ministers over a policy programme. By virtue of its permanence, the bureaucracy has a potentially strong advantage over politicians in developing and defending a particular ideological stance. As the ambit of government policy has grown, the potential for the bureaucracy to influence policy has also grown.

Nonetheless, the bureaucracy is also heavily dependent upon the government. The bureaucracy relies upon ministers to ensure that its favoured policies receive a fair hearing in the cabinet and in the House. Ministers also control the flow of public funds to departmental activities and defend (or criticise) the actions of the bureaucracy to the Minister of Finance at budget time.

Recent reforms have increased the autonomy but also the accountability of the bureaucracy. By replacing convention with contract as a means of handling relationships with departmental heads, the reforms have in principle given ministers a greater degree of control over policy outcomes. The contractual approach gives departments greater freedom of choice about how to achieve the outputs required by ministers, but also makes them more clearly accountable for the delivery of these outputs. The reforms should be seen in the light of considerable growth in the public sector that had, over a number of decades, increased the chance for bureaucrats to redirect, rather than simply stabilise, the policy-making process. In this sense, the reforms can be seen as shifting the balance away from departmental discretion over policy outcomes, in favour of ministerial discretion, while at the same time providing greater transparency and susceptibility to public scrutiny.

2.6 Concluding Remarks: When Does a Westminster System Work Well?

Since all known sets of constitutional rules are demonstrably imperfect, we must ask comparative questions: Is the New Zealand system, as it can be expected to operate, as good as or better than other feasible constitutional arrangements? And would we answer differently if New Zealand were larger (or smaller), or more (or less) economically developed, or suffering a greater (or lesser) problem with status quo policies?

The absence in New Zealand of an entrenched, written constitution, and of any formal separation of powers, does not imply an absence of checks and balances on executive power. Rather, policy-makers in New Zealand are constrained by a range of mostly informal checks and balances.

Governments are compelled to contest hair-trigger elections. To win, they must try to select policies whose political benefits are timely and exceed their costs. They must perform credibly and reasonably consistently in the rough and tumble of parliamentary and media debate. They must retain the loyalty of party faithful, and/or win new and committed constituencies.

To pass legislation, prime ministers and cabinets must consistently secure the support of their caucus. Support and the appearance of party unity come at a cost in terms of influence - influence over the actual allocation of positions in the executive, and influence over the policies pursued by the executive.

These constraints indicate several notable advantages of a Westminster system. Policy-makers can respond to events in a flexible and timely fashion. Furthermore, the medium of intra-party constraints involves a continual funnelling of electoral pressures upon the executive. The prime minister and cabinet are given the ability to implement favourable policies, and they are held accountable for their actions.

The potential for problems in a Westminster system arises when a misguided executive can implement preferred policies that are undesirable for the nation as a whole. As we have seen above, the preferences of the electorate can be manipulated even when electoral constraints remain operative. Depending upon the preferences of the leaders, policy can be manipulated in either a favourable or unfavourable direction. In evaluating a Westminster system, the critical issues are therefore how different constraints interact, and the initial situation of the country in question.

In any system, day-to-day checks and balances arise from the need for promoters of a policy to convince others of its worth. The costs of convincing others determine the magnitude of the checks and balances imposed by any particular system. These costs differ between systems.

In a Westminster system, the key group that the executive must convince, at least on a day-to-day basis, is the caucus. Especially in a small country such as New Zealand, caucus itself it likely to be small, and is bound by common party affiliation. It is also conveniently gathered in the weekly caucus meetings. While the costs of securing caucus support may vary between policies, on average they will be low relative to other systems. Legislation can be passed relatively cheaply and rapidly. Legislative packages tend also to be consistent; in particular, expenditure decisions are more likely to be made in close connection with revenue decisions than in, say, a presidential-congressional system.

Although the bulk of our comparative analysis will come in later chapters, it is worth noting here that alternative constitutional arrangements involve higher costs of passing and agreeing upon legislation than does a Westminster system. In the presidential-congressional system of the US, for example, the president must win the support of both houses of congress and avoid a judicial veto.

Under many forms of proportional representation, the passage of legislation requires approval of a coalition government. Although this requirement is similar to the requirement of caucus unity in a Westminster system, it will generally be more costly to achieve. Coalitions in multi-party systems are generally manufactured after, not before, elections, and lack the adhesive provided by common party membership. While coalitions of parties may survive a number of electoral cycles, there is less assurance that they will do so than there is that party members in a Westminster system will remain with the party. Nor is membership of a majority party such an important prerequisite for electoral success as in a Westminster system. In addition, coalitions in multi-party systems are not bound by the same conventions of secrecy or whipping found in Westminster systems. As a result, favoured legislation is likely to be bought at a price of larger, and more overt, trade-offs in terms of both the detail of the legislation and other legislative packages.

2.6.1 The Interaction of Constraints

The greater ease of passing legislation in a Westminster system implies that successful Westminster systems are those accompanied by implicit checks and balances that operate in the proper direction. Furthermore, Westminster systems are most effective when radical policy changes may be required. We isolate the following factors as favourable to the operation of Westminster systems:

• a well-informed electorate;

• a small, open economy willing to subject itself to international competition; and

• the necessity for policy change in the future.

We have noted that the electoral check is particularly strong in a first-past-the-post parliamentary system, because it enables the ready, even routine, removal and replacement of governments. Elections of this kind will work to promote good government if they compel political parties to compete on the basis of the quality of their policies. A prerequisite is an electorate that can discriminate between policy programmes - that can make reasonably sensible assessments about the quality of performance and promises. Voters do not need to understand the detail and merits of, or have a view on, every policy implemented or proposed by a past or prospective government. (This is precisely the kind of information that representative government economises on.) But they do need to know, in general terms, what the government and opposition stand for, and how they are likely to perform in practice.

International factors may be more or less binding in their ability to constrain legislation. For example, as illustrated by the policies of the late 1970s to early 1980s, it is feasible to attempt to shut out international pressures by pursuing a form of economic autarchy. In this instance Westminster systems tend to compound bad legislation, at least in the short run.

The experience of New Zealand through the post-war period suggests that policies leading to a gradual erosion of economic performance can be sustained, and reinforced, over many years. The policies implemented under the 1975-1984 Muldoon administration in the wake of Britain's entry to the EC and the oil crises of the 1970s were ultimately an extension of a general policy approach that had been in place since the 1930s. The assurance that international constraints will, in the long run, compel policy changes, may provide scant comfort.

In the longer term, international constraints did prove binding and the Westminster system responded with deregulatory policies from the mid-1980s. Given that the New Zealand economy has now been opened to the world, we expect the Westminster system in the future to provide superior performance than it did through say, the 1960s and 1970s. A general effect of the economic policies implemented since the mid-1980s has been to reduce barriers to resource flows in and out of New Zealand. The result is that governments are more constrained than previously to compete with other jurisdictions. If their policies are sound, resource flows into the country are likely to reinforce their positive domestic effects. Poor policies bring a risk of resource outflows - both the withdrawal of foreign resources, and the departure of domestic resources. In this sense, the policy changes of the 1980s brought a fundamental (beneficial) change in the nature and strength of constraints on policy-makers.

In today's economic climate, we see international constraints as playing a binding role in the short run. In the 1980s New Zealand made a fundamental policy decision to compete in the world economy rather than pursue autarchy. The disciplines for good decision-making are now powerful. In addition to explicit policy decisions, general improvements in information and communication technologies have improved resource mobility and tightened international constraints. As long as we expect these disciplines to hold, we can be relatively sanguine about the results that will be produced by a Westminster system.

The stabilising function that we have attributed to international resource flows is likely to operate with differing force among countries of different sizes - even if their basic policy programmes are similar. In a country with large, diverse, internal markets, for example, international constraints are less likely to be binding than in small countries heavily dependent on imports and exports. In small, open economies, too, there is likely to be more need to respond rapidly and flexibly to external shocks. (For example, Britain's entry to the EC had a considerably greater impact on New Zealand than on larger commonwealth countries trading with Britain.) On average, Westminster systems might be expected to perform better in small, open countries than in large ones.

Finally, Westminster systems also work best in situations where substantial policy changes are required. The Westminster system permits rapid responses to major policy problems or exogenous shocks. Although Westminster systems do not guarantee that the legislation passed will be in the proper direction, they do at least minimise the chances of legislative deadlock. Alternative systems, such as that in the US, have found legislative deadlock a serious problem when trying to address important issues such as the budget deficit or education policy.

To pursue this contrast further, the US constitution can be seen as designed to preserve and build upon favourable initial conditions. Its inherent conservatism may be desirable where the existing body of law (for example, an inherited system of common law) is sound, and where destabilising exogenous shocks are expected to be rare or absent. But such conservatism will be less appropriate from other starting points. For example, the introduction of a constitution and separation of powers on the American model may unduly restrict the implementation of a deregulatory programme in a new democracy in Eastern Europe, or an underdeveloped country with a history of interventionist colonial law.

If the suitability of a Westminster system is in part a matter of where a country starts from, it follows that the appropriateness of this system for a particular country could change over time. The likely need for change will depend in part on the sensitivity of the Westminster model's performance to such factors as size and openness. It will also depend on the capacity for evolutionary changes within Westminster systems. In the absence of entrenched, written constitutions, Westminster systems have proved capable of sustaining significant evolutionary changes: the extension of suffrage, the growth of the party system, the emergence of caucus as a forum for policy debates, and the extension of the select committee system, for example. The scope for ongoing evolutionary change should therefore be taken into account in deciding the case for revolutionary constitutional reforms. It is possible that in many instances constitutional problems can be addressed more effectively by evolutionary change than by direct reform.

3.0 PROPORTIONAL REPRESENTATION

3.1 Introduction

The forthcoming referendum offers four electoral systems as alternatives to the current Westminster system: preferential voting, the single transferable vote, mixed-member proportional systems, and supplementary systems. Each system allocates legislative seats according to a numerical formula. Except for preferential voting, each of these formulae offers representation to candidates and parties who do not receive a majority or plurality of the vote in any single district. The Westminster system, in contrast, awards district seats on a winner-take-all plurality basis.

To summarise each system very briefly, the mixed member system allocates seats in the legislature in proportion to party votes. The supplementary member system maintains the current first-past-the-post seats but adds some seats allocated in proportion to party vote. The single transferable vote allows voters to rank all candidates in their district and uses a complicated numerical formula to determine the winner. Preferential voting maintains the basics of the Westminster system intact, but allows second-place votes to count towards determining the winner if no party wins more than fifty percent of the vote.

Mixed-member proportional and supplementary systems are similar in their effects, as we shall see below; we refer to these as partial list systems. We discuss this point in more detail below. Partial list systems are distinct from the single transferable vote, which is based upon quite different principles and creates systematically different effects. Under partial list systems party structures are strengthened and government by coalition replaces the Westminster "winner-take-all" tendency. Under the single transferable vote party structures are weakened and representatives emphasise local benefits and constituency service. Preferential voting would not alter the current Westminster system substantially.

Most Western democracies use some form of proportional representation, with the exceptions of the Anglo-American countries (to varying degrees). The exact form of proportional representation varies with each country. The mixed-member proportional system is found in Germany, the single transferable vote is used in Ireland, preferential voting is used for the Australian lower house, and the supplementary system is now used in the fledgling democracies of South Korea and Hungary. We also discuss list systems in general, which are used in many countries in either pure or modified form.

Many countries which currently have proportional representation once had other forms of democracy but switched electoral systems, most commonly in the early years of this century. In Table 3.1 we summarise the history of adoption of proportional representation - where the system was adopted, when, and for what reason.

Moves to proportional representation tend to be irreversible. Once proportional representation is adopted both small parties and the leaders of large parties (who control party lists) have a strong stake in the system. With the exception of France, which has changed constitutions several times, no countries have voluntarily abandoned proportional representation for alternative forms of democracy.

3.1.1 Overview and Summary of Results

Each of the four suggested electoral reforms would change the scope and direction of government activity. To present an overview of these changes, we summarise how each reform would change the relative strength of the factors influencing government behaviour. Again, we do not attempt in the body of the text to judge whether these results are desirable; we save such judgments for the concluding remarks. The focus of our attention is changes in the balance of power. The following outline is a summary and overview of the analysis to follow.

Preferential voting would not change the status quo dramatically. For the other reforms, the factors influencing government behaviour would change in the following manner:

1. Government mirrors the preferences of the median voter.

All forms of proportional representation would weaken the influence of the median voter. In the multi-party system produced by partial list systems, no single party attempts to stake out the middle ground. The single transferable vote also weakens the influence of the median voter but less so than partial list systems.

2. Government is influenced by special interest groups and lobbying.

Both partial list systems and the single transferable vote have ambiguous effects here. In Westminster systems interest group pressure is channelled through the majority party. Partial list systems channel interest group pressures through multi-party coalitions instead. The single transferable vote would place these pressures upon individual legislators and away from party organisations.

3. Government maximises revenue and redistributes resources from productive citizens to itself.

Partial list systems weaken the ability of government to redistribute wealth from one group of citizens to another or to the government. The use of multi-party coalitions implies that policies must be supported with a greater degree of consensus than under the Westminster system. Under the single transferable vote the ability of the legislative to check the executive is weakened and less consensus is needed to pass policies.

4. Government provides benefits to particular districts and regions, at the expense of other regions.

The single transferable vote markedly increases the incentives for political representatives to seek wealth redistribution at the local and regional levels. Partial list systems, in contrast, favour policies with a national rather than regional focus.

5. Government indulges the policy preferences ("ideology") of politicians in office.

Partial list systems have ambiguous effects here. Under partial list systems political parties differentiate themselves by ideology to a greater degree but, once in office, face greater explicit checks and balances. Westminster systems, in contrast, give greater influence to the ideologies of voters rather than politicians, but also place weaker explicit checks upon a ruling government. The single transferable vote weakens the influence of politicians' ideology by redirecting political effect towards constituency service and regional policies.

6. Government finds its hand forced by international constraints.

Both partial list systems and the single transferable vote decrease the ability of a government to respond rapidly and flexibly to changing international constraints. Through its clear demarcation of government power and responsibility, the Westminster system has the greatest scope in this regard. International constraints play a smaller role under other electoral alternatives.

The remainder of this chapter is organised as follows. Section 3.2 considers preferential voting, a system that would not alter the status quo drastically. Section 3.3 examines the single transferable vote, the "outlier" form of proportional representation. We examine the incentives created under the single transferable vote and focus upon the properties of the single transferable vote which differ from other forms of proportional representation.

Sections 3.4 through 3.6 provide the bulk of the comparative analysis. Section 3.4 considers the operation of pure and partial list systems. Although the pure list system is not on the forthcoming referendum, an understanding of the pure list system is essential to grasp the derivative mixed-member proportional and supplementary options. Section 3.5 compares directly the mixed-member proportional and supplementary member alternatives in cases where the number of supplementary seats is significant. Section 3.6 provides a general comparison of proportional representation and the Westminster system of government.

3.2 Preferential Voting

Preferential voting, sometimes called alternative voting, is based upon single member constituencies and a winner-take-all electoral formula. The current Westminster system would remain intact with a single exception. Voters would choose not only a first choice, but also subsequent choices in order of preference (second, third, etc). When no candidate gains more than fifty percent of the vote, these additional rankings are used to determine the winner.

The most prominent use of preferential voting is found in Australia, where the system has been used to elect the Federal House of Representatives since 1918. Preferential voting is used also in the lower houses of many Australian states and for the Irish presidency.

The specific formula behind preferential voting is the following. If no candidate wins more than fifty percent of the first preference votes, the additional voter rankings are used to determine a winner. (In contrast, the status quo based upon plurality simply awards the seat to the candidate with the largest number of votes). The candidate with the fewest first place votes is eliminated. The second place choices of the voters who preferred the eliminated candidate are now elevated to first place votes. If a candidate now has more than fifty percent of the vote, that candidate is elected. If not, we perform the same step, eliminating the candidate with the next smallest number of first place votes. Lower choices of these voters are elevated once again, until one candidate has won fifty percent or more. If subsequent eliminations do not result in one candidate having fifty percent or more, the seat is awarded to the candidate with the largest number of votes.

Under preferential voting, whether voters must rank all candidates can be mandatory or optional (both have been tried in Australia). Whether universal ranking is mandatory or voluntary does affect the operation of the mechanism. When universal ranking is optional, many individuals will choose only one, or perhaps two candidates. The rationale for preferential voting is partially vitiated, and a candidate can be elected without having fifty percent of the vote. When universal ranking is mandatory, voters, with their lower choices, may be helping to elect candidates whom they do not favour. Moreover, some percentage of voters will refuse to rank all candidates and thus hand in ballots that do not count.

Preferential voting may affect actual outcomes because resort to second and third place preferences is likely to occur frequently in New Zealand. Not since 1951 has a government received a majority of the votes cast by electors. Although preferential voting will often give the same winner as plurality voting, it need not always do so. Most likely, had New Zealand used preferential voting the results of some elections would have been reversed.

Preferential voting tends to favour parties that are the second choices of those inclined to vote for a third party. A voter can choose a third party without "wasting" his or her vote if the third party has no chance of winning. If no candidate wins more than fifty percent of the votes and the third party is eliminated, this voter's second choice will now count towards determining the final outcome.

For this reason, preferential voting tends to favour third (or additional) parties. Because voters can choose third parties without "wasting" their vote, third parties are likely to have more success in marketing their candidates to voters. Preferential voting thus encourages a greater diversity of opinion in electoral contests and increased party competition before an election.

The greater diversity of opinion, however, does not necessarily imply an electoral contest that better reflects the preferences of the voters. The number of parties will still be small and these parties will not reflect all shades of voter opinion. In fact, the major parties may be less inclined to stake out controversial or innovative positions. The use of a ranking scheme favours candidates that generate the least opposition, rather than those candidates that command the widest support. Being chosen second rather than first by a voter involves no disadvantage if the party chosen first is dropped. Parties that are chosen last, however, will not obtain additional votes in this fashion.

Unlike most other forms of proportional representation, preferential voting does not significantly increase the likelihood that third parties will participate in governing coalitions. Once all the votes are tallied, seats are still handed out according to a winner-take-all formula. The only difference with the status quo is that the winner is computed on a different basis. Third and smaller parties have an easier time winning votes, but unless these parties attract widespread support, these additional votes need not translate into additional seats.

Furthermore, preferential voting also weakens third parties in one important respect. Larger parties need no longer consider third parties an electoral threat. If a larger party knows that a third party cannot split its vote by dividing the electorate (second place votes will still count), the larger party will be less inclined to make policy concessions to neutralise the influence of the smaller party. In this regard smaller parties have less influence, and the incentive to start smaller parties decreases. Alternatively, the incentive to repair a split in the party is reduced.

For these reasons, the encouragement given to small parties under preferential voting is ambiguous. As we see in Australia, third and smaller parties are present but have not attained the influence of third parties on the European continent, where traditional forms of proportional representation are used. In Australia, preferential voting can be said to have encouraged the Liberal-Country (now National) party alliance (and thus a governing role for the National party), but large parties are not necessarily dependent upon small ones, as in, for instance, Germany. We are thus likely to favour alternative voting if we wish to encourage small parties slightly, but not too much.

Preferential voting does not necessarily remedy the kind of inequities associated with a first-past-the-post electoral system. While the status quo allows candidates to be elected with less than fifty percent of first place votes, preferential voting also creates possible inequities. These inequities arise because preferential voting potentially treats second, third, and lower place votes as equal to first place votes.

In the 1966 election in Australia, for instance, the Liberal-Country party won a large parliamentary majority with a tiny majority of the popular vote. In the Canadian province of Alberta in 1948, preferential voting allowed one party to win all the seats in the legislature with only 58 percent of the first place votes. Similarly, in the Australian state of Victoria in 1967, the Liberals won three times as many seats as Labour despite having fewer first-place votes.

For the reasons discussed above, we do not see preferential voting as a significant alternative to the status quo. While preferential voting is not dominated by the status quo, neither does preferential voting offer clear-cut advantages. Throughout the remainder of this chapter, we will focus our attention upon those electoral alternatives that would systematically modify the current Westminster system.

3.3 The Single Transferable Vote

The single transferable vote was a popular reform proposal in the late nineteenth and early twentieth centuries, especially in Anglo-American countries. John Stuart Mill and Thomas Hare were especially prominent as founders of the single transferable vote movement.

The single transferable vote allocates seats in the legislature according to voter ranking of available candidates. Voters are given a list of all eligible candidates and list these candidates from most favoured to least favoured. The information contained in these rankings is then used to determine electoral results (the exact formula is explained below).

In addition to its national use in Ireland (since 1920, and formerly in Northern Ireland), the single transferable vote has also been used in Malta, for local elections in Tasmania (since 1907), and to elect the upper house in Australia (since 1949). In New Zealand the single transferable vote formerly was used for some local elections (e.g. in Christchurch city). We focus upon the effects of the single transferable vote when adopted fully at the national level.

Electoral systems based upon the single transferable vote tend to produce the following effects:

• voters can express preferences for more than simply their favourite candidate or party;

• representatives are focused towards constituency service and district policies, rather than national policies;

• political parties are weak, non-ideological, and subject to frequent infighting;

• the ability of the legislature to check the executive is weak;

• most voters are confused by the mechanics of the single transferable vote; and

• sophisticated voters have an incentive to manipulate the system and vote an order which is not their true preference.

We now consider the operation of the single transferable vote in more detail.

3.3.1 The Single Transferable Vote Formula: A Technical Digression

The election of representatives under the single transferable vote is based upon the concept of a quota, sometimes called the Droop quota. Representatives who satisfy the Droop quota (explained further below), based upon information from voter rankings, achieve election. Each voter is first given a ballot and then asked to rank as many candidates as he or she wishes (first choice, second choice, third choice, etc.). The single transferable vote system attempts to ensure that as many voter preferences as possible are counted, even when a voter's first choice is not elected.

The Droop quota hurdle is the smallest number of first-place votes sufficient to ensure election. If we take the case of a nine-member district with one hundred voters, the Droop quota is eleven; it would be impossible for more than nine candidates to each receive eleven first-place votes .

Once all the ballots are tallied, the single transferable vote elects those politicians who have eleven or more first-place votes; i.e. those who satisfy the Droop quota. Assuming that not all nine seats have been filled, further computation is required to fill the additional seats.

We now take the voters whose first choice votes did not matter and count their second-place vote as a first-place vote. Specifically, we examine the second-choice votes of two groups of voters: those who do not see their first choice elected and those whose votes are "extra" or "surplus" votes for candidates who did win. In the latter case if a candidate receives seventeen first place votes but requires only eleven, six of these supporters will now have their second vote transferred and counted elsewhere.

After transferring unused votes we add these "new" first-place votes and see which candidates meet the Droop quota from this pool. Those who do are elected.

If not all the seats have been filled, another step in the procedure is used. We then drop out the votes of that candidate with the smallest number of first-place votes. If a voter's first choice has been dropped, that vote is then allocated to the next candidate on the voter's list. We compute the Droop quota once more and choose those representatives who meet the quota. We then drop the candidate with the second smallest number of first-place votes and transfer those votes, and so on, computing the Droop quota each step of the way. The process continues until we have elected the chosen number of representatives.

The single transferable vote can be applied to any number of legislative seats, even as small as one (at which point the system approaches preferential voting). The Royal Commission Report suggests five seats per district and Ireland uses between three and five seats per district. Any number of candidates can run for the chosen number of seats.

In Ireland, the results of the single transferable vote tend to be roughly proportional. That is, if a party receives twenty percent of the first-place votes cast, that party tends to receive close to twenty percent of the seats in the legislature. The difference between first-place votes cast and seats received rarely exceeds five percentage points (e.g. twenty percent of first-place votes and fifteen percent of the seats; see Penniman, 1978, p. 26). The percentage of voters who see their first choice elected is usually about seventy percent (Penniman, 1978, p. 30).

Once representatives have been elected, a ruling government is formed either through a coalition or through a majority party. In section three of this chapter we consider the operation of coalition governments in more detail. Here, we focus upon those features of the electoral mechanism which distinguish the single transferable vote from other forms of proportional representation.

3.3.2 Complexity of the Single Transferable Vote

The complexity of the single transferable vote allows voters to express an additional range of preference. Voter expression of preference is not limited to the selection of a first choice but encompasses the entire range of candidates. There is also less likelihood that a vote will be "wasted" on losing candidates. Even if the first choice of a voter is not elected, the preferences of that voter still have input towards determining the final outcome.

The greater expressiveness of the single transferable vote, however, comes at the price of a complex electoral mechanism. The single transferable vote is subject to two criticisms resulting from its complexity. First, under the single transferable vote a majority of voters are unaware of how their electoral system works. In this regard the single transferable vote does not use a very representative process, even if the final allocation is roughly proportional to the percentage of first-place votes cast.

The complexity of the single transferable vote also may affect outcomes if voters cannot mark their ballots accurately. Data on voter confusion are difficult to find because the ballots used in modern elections are typically secret. Nonetheless, available evidence suggests that voter confusion is likely to affect ballot markings adversely under the single transferable vote.

A study of the Irish election of 1973 implies that many voters simply ranked candidates in alphabetical order past a certain point on their ballots. A random sample of the Irish population showed that surnames with the first letter A, B, or C occurred 20.3 percent in the general population yet 33.6 percent in the legislature. Similarly, surnames with the letters P-Z occurred 19.4 percent in the general population but only 11.9 percent in the legislature (Chubb, 1982, p. 161). Voters appear to have been overwhelmed by the complexity of the choices offered, although Ireland had been using the single transferable vote for many years.

Another study, this one of the single transferable vote in New York, produced similar conclusions. The city of New York experimented with the single transferable vote for its city council from 1937 to 1947, when the system was abandoned as unsatisfactory. A study by the New York State Constitutional Conventional Committee concluded that voter confusion on ballots was rampant. In the borough of Brooklyn, for instance, thirty-one percent of the ballots were either invalid or unusable. On many other ballots, confused voters had voted alphabetically or selected a favourite candidate and then selected all of the names directly below that candidate on the ballot.

The second criticism points out the possibility of manipulative or "strategic" voting under the single transferable vote. Those voters who do understand the workings of the single transferable vote do not always have an incentive to rank the candidates according to their true preferences. Irish parties are aware of the potential for vote manipulation and sometimes instruct and persuade their supporters to take advantage of the system in this regard (Gallagher, 1986, pp 255, 273).

In some situations, voters have an incentive to rank their favourite candidate last rather than first. In other cases, a candidate will fail to achieve election, even if he or she would defeat all other candidates in a one-on-one run-off election.

Examples and a proof of these propositions can be found in Annex A. Nonetheless, the intuitive reasoning behind these results is evident. For instance, the method of eliminating candidates with the smallest number of first-place votes gives voters an incentive to falsify their candidate rankings. Consider the case of a voter who prefers Smith to Jones but realises that Smith is a heavy favourite and that Jones is in danger of early elimination. That voter may shift his or her first-place vote to Jones even though Smith is preferred. In contrast, first-past-the-post systems, as long as they are restricted to two major parties, do not provide this incentive, albeit by limiting voter choices. The single transferable vote provides no guarantee that votes reflect true preferences.

3.3.3 The Single Transferable Vote and Political Incentives

The effect on political incentives is perhaps the greatest difference between the single transferable vote and list and partial list systems. Whereas list systems focus political incentives upon national problems and strengthen political parties (see below), the single transferable vote focuses representative attention on local issues and weakens the role of political parties.

The direct appeal to voters under the single transferable vote changes the incentives of legislators. Legislators tend to focus upon constituency service and district-specific policies rather than national politics. Voting is on the district level and campaigns must focus upon satisfying district needs. Politics becomes less a matter of ideology and more a matter of "delivering the goods" to constituents.

The emphasis upon constituency service is illustrated by the Irish experience with the single transferable vote. In Ireland politicians complain frequently that they are required to cater continually to constituency whims and preferences. Politicians have little time to address national problems or think strategically about future policy in broad terms. In the Irish Dail, a deputy frequently is called a "constituency messenger" (Calder, 1988, p. 65). Irish representatives spend considerable time helping their constituents deal with government bureaucracies (Chubb, 1982). Similar tendencies operate in the single transferable vote in Tasmania and Malta (Gallagher, 1986, pp. 266-7).

The tendency of the single transferable vote to encourage constituency service (as opposed to national policy) is even stronger than under current electoral institutions. Under a Westminster system, party constraints and party unity play some role in focusing the attention of representatives upon national issues.

The single transferable vote weakens political parties by requiring members of the same party to compete against each other in the same election. Rather than using candidate selection processes and political parties to determine which party members have first crack at elected office, the single transferable vote allows voters to decide. In effect, the single transferable vote combines candidate selection and run-off into one comprehensive election. We should expect no more party unity under the single transferable vote than we find in candidate selection processes; candidates in the same party are more likely to feel rivalry than unity.

In Ireland we even find examples where sitting representatives prefer to fill empty seats with members of the opposing party; filling the seat with a member of one's own party would increase the degree of competition faced at the next election. When two candidates from the same party are running in the same district, they tend to divide the district into personal bailiwicks to minimise the chances of a direct clash (Katz, 1984, pp 143-4).

Parties find it difficult to develop coherent ideological platforms that all or most members adhere to. Since representatives focus upon providing private benefits for their constituents, they are less inclined to adhere to a party line or consistent ideological position. Furthermore, candidates will differentiate themselves from competing members of the same party and not just from the opposition party. Re-election requires competition against (part of) a legislator's party, rather than unity with a legislator's party. For these reasons, in Ireland, political parties exist primarily in the form of local branches which become active shortly before elections.

The weakness of parties and the emphasis upon constituency service decrease the ability of the legislature to check the power of the executive branch of government. Members of the Irish parliament rarely challenge the executive, not because they are loyal to party ideology, but rather because they are preoccupied with constituency service. The government dominates parliament considerably more than in other countries (Gallagher, 1980, pp 501-2). Intra-party constraints are weak under the single transferable vote.

Because the single transferable vote de-emphasises ideology and party structure, the single transferable vote does not necessarily give rise to a multi-party system and coalition government. The lessening of the ideological content of politics makes it more difficult (and less rewarding) to start a third party. Upstart politicians who wish to run for election are less likely to find themselves ideologically excluded from the major parties. These politicians will find it simpler to attach their name to a prevailing party rather than start their own organisation. Furthermore, the small number of seats in a district (from three to five in Ireland; the Royal Commission recommends five for New Zealand) limits the room for party proliferation.

The compatibility of the single transferable vote with a two-party system is illustrated by the historical evidence. Tasmania, for instance, has had more than eighty years of the single transferable vote (since 1907) but is still essentially a two-party system (Wright, 1984, p. 133). The Country party, which traditionally represented farming interests, found little support in Tasmanian local politics. Farming interests instead choose to support farmers who run on one of the two major tickets; the weakened party structure under the single transferable vote makes this possible (Lakeman, 1984, p. 46). Similarly, attempts to form a strong National party in Tasmania have not succeeded.

Maltese politics also has been limited often to two dominant parties (Katz, 1984, p. 140). In Ireland, the fate of third parties has been mixed. The Irish system evolved from a multi-party system to two parties (many commentators have interpreted Fine Gael and Labour as a single party; see Katz (1984), for instance). Since the 1980s there has been a blossoming of third parties but most of these parties are a reflection of the polarisation of the Irish conflict, rather than an intrinsic result of the single transferable vote.

3.4 List and Partial List Systems

We consider now the use of party lists to allocate seats (the pure list system) and systems which combine use of lists with first-past-the-post selection criteria (partial list systems). We compare and contrast the properties of various list systems to set the stage for our later comparison of partial list systems and the Westminster system.

The list system is the purest form of proportional representation. In its simplest form, the list system allocates seats in parliament according to the percentage of the vote each party receives. If three parties receive forty, forty, and twenty percent of the vote total, for instance, these parties will receive forty, forty, and twenty percent of the parliamentary seats, respectively. Voters choose a party rather than an individual representative and the party organisation determines which individuals will fill the allotted seats. Henceforth arises the name "list system"; parties rank potential representatives in a list from which elected members are drawn.

The list system may be used at either the national or district level. When used at the national level, the list system puts all votes into the same pool. The matching of votes to parliamentary seats is performed across this national pool. National versions of the list system are found in the Netherlands and Israel, the two countries with the "purest" list systems.

When used at the district or regional level, list systems pool votes within each district. Districts are awarded multiple seats and the distribution of seats within each district is determined by the percentages obtained in each district pool.

New Zealand, as a country with a small electorate, is a more likely candidate for use of national lists rather than regional lists. Most countries with proportional representation, however, use some version of a regional list system. Although the Royal Commission report focuses upon national list systems, the scope of the list system to be considered remains an open choice.

3.4.1 Differences Between National and District List Systems

The choice between national and district list systems influences the incentives that operate under proportional representation. National list systems favour the following features:

• strong party organisation at the national level;

• incentives to propose national policies rather than regional policies; and

• increased representation for small parties.

Regional list systems, in contrast, favour the following features:

• slightly weaker incentives for national policies;

• weaker national party organisation and stronger regional party organisation; and

• weaker representation for small parties.

Both district and national list systems link a politician's chances of election to his or her standing on the party list. Politicians in the upper reaches of party lists are virtually assured of election, whereas the probability of election decreases as one moves d