Citizens, Elections, and Policy Making: Understanding the Role of Elections in Democracy, Apuntes de Ciencia de la administración

¡Descarga Citizens, Elections, and Policy Making: Understanding the Role of Elections in Democracy y más Apuntes en PDF de Ciencia de la administración solo en Docsity!

G.

BINGHAM

POWELL,

JR.

of Elections as Instruments

(^) Democracy

MAJORITARIAN

AND

PROPORTIONAL

VISIONS

Yale

(^) University

(^) Press

LondonNewHaven&

„

Copyright©

(^2000)

(^) by Yale University. All rights reserved. This book may

(^) not

(^) be

reproduced, in whole

(^) or (^) in (^) part,

(^) including illustrations, in any form (beyond

(^) that

copying permitted by Sections

(^107)

(^) and

(^108)

(^) of (^) the (^) U.S.

(^) Copyright Law

(^) and

(^) except

by reviewers for the public press),

(^) without

(^) written permission from the

Printedpublishers.

(^) in the

(^) United

(^) Stares

(^) of (^) America.

Library

(^) of (^) Congress Cataloging-in-Publication

(^) Data

Powell,

(^) G. (^) Bingham.

Elections as instruments

(^) of (^) democracy:

(^) majoritarian

(^) and

(^) proportional

visions/ G. Bingham

(^) Powell,

(^) Jr.

p.

cm.

Includes bibliographical references

(^) and

(^) index.

ISBN

(^) 0-300-08015-

(^) (cloth: alk. paper) -

ISBN

(^) 0-300-08016-

(^) (paper:

(^) alk. paper)

(^) Elections.

2. Democracy.

representation.

4. Majorities.

JF

.P

2000

321.8-dc

3. Representative government

(^) and

(^) Proportional representation.

(^) Tide.

99-

A catalogue record for this

(^) book

(^) is (^) available from the British Library.

The

(^) paper

(^) in this

(^) book

(^) meets the guidelines for permanence and durability

(^) of (^) the

Committee

(^) on (^) Production

(^) Guidelines for Book Longevity

(^) of (^) the Council

(^) on

10 Library Resources.

(^9) 8 7

(^6) (^5) (^4) (^3) 2

1

5 4 3 2 I

List of Tables Contents

vu

List of Figures

1x

Acknowledgments

x

Part

(^) I Citizens, Elections, and Policy Making

Elections as Instruments of Democracy

Constitutional Designs as Visions

(^) of (^) Majoritarian

(^) or

Proportional Democracy

20

Part

(^) II Responsiveness: Connecting

(^) Votes,

(^) Governments,

and

(^) Policymakers

Accountability: Conditions for Citizen Contra!

Conditions for Mandates: Identifiability and Majority

A Vision of Dispersed Political

(^) Power:

(^) Authorized Representation

in (^) Policy

(^) Making

V

4

Citizens, Elections, and Policy Making

such citizen influence

(^) is (^) a (^) good

(^) thing,

(^) that

(^) elections should

(^) not

(^) only provide

symbolic reassurance,

(^) but

(^) also genuinely serve as instruments

(^) of (^) democracy.

Elections are

(^) not

(^) the only instruments

(^) of (^) democracy. They

(^) must

(^) be helped

by (^) other

(^) organizations

(^) and

(^) by rules

(^) that

(^) encourage communication

(^) and

(^) coop-

eration.

(^4) But elections seem

(^) to (^) be the critical democratic instruments. They

claim

(^) to (^) establish connections

(^) that

(^) compel

(^) or (^) greatly encourage the policy-

makers

(^) to (^) pay

(^) attention

(^) to (^) citizens. There

(^) is (^) a widespread consensus

(^) that

(^) the

presence

(^) of (^) competitive elections, more

than

(^) any

(^) other

feature, identifies a

contemporary

(^) nation-state as a democratic political system.

(^5)

The

apparent

(^) consensus

(^) that

(^) elections are significant conceals deep dis-

agreements

(^) about

(^) whether

(^) and

(^) how

(^) they serve

(^) to (^) link citizens

(^) to (^) policymak-

ers. These disagreements are partially normative; they reflect different ideals

(^) of

the relationship between citizens

(^) and

(^) policymakers. They are partially concep-

tual, reflecting different understandings

(^) of (^) how

(^) the preferences

(^) of (^) citizens can

be aggregated. They are partially empirical,

(^) grounded

(^) in alternative theories

about

(^) what

(^) kinds

(^) of (^) institutional arrangements will best serve

(^) to

link the

people

(^) and

(^) their representatives.

Following a familiar

tradition

in comparative analysis, I

group

the ap-

proaches

(^) to (^) elections

(^) and

(^) democracy

(^) into

(^) two

(^) great camps: majoritarian

(^) and

proportional.

(^6) These are the

(^) contemporary

(^) expressions

(^) of (^) Tocqueville's

(^) "two

opinions which are as

(^) old

(^) as the

(^) world,"

(^) quoted

(^) above in the epigraph. Major-

itarianism tries

to use elections

(^) to

bring the

(^) power

of (^) the people directly

to (^) bear

(^) on (^) policymakers. Proportionalism establishes

(^) an (^) alternative, positive

democratic ideal,

(^) rather

(^) than

(^) just

(^) "limiting"

(^) majorities, a goal which can be

(and has been) espoused by those

(^) who

(^) are opposed

(^) to (^) democracy as

(^) weil

(^) as by

those sympathetic

(^) to (^) its fundamental aims.

This comfortable language oversimplifies

many

(^) complex arguments

(^) and

distinctions, including the roles

of (^) electoral laws

and

policy-making con-

straints. But it captures some very fundamental assumptions

(^) and

(^) their im-

plications. Moreover, as we shall see in the

(^) next

(^) chapter, the

(^) two

(^) primary

(^) types

of (^) constitutional designs in

(^) contemporary

(^) democracies can be understood as

having election rules

(^) and

policy-making rules

(^) that

(^) reflect either the major-

itarian

(^) or (^) proportional

(^) vision. In the analysis

(^) to (^) come I shall try

(^) to (^) identify the

contrasting elements in the visions

(^) and

(^) to (^) test their empirical expectations

against performance in real elections.

Democratic Visions

of

Citizens

and

Policymakers:

Concentrated or Dispersed

Power?

In writing this

(^) book

(^) I was (naively) surprised

(^) to (^) discover

(^) how

(^) hard

(^) it (^) is

to (^) use consistent language

(^) about

(^) the empirical claim

(^) of (^) competitive elections

Elections

(^) as (^) Instruments

(^) of (^) Democracy

5

to (^) be instruments

(^) of (^) democracy. I

(^) w~s_tor!

(^) bet~ee_n

(^) the

(^) formulatiqn

(^) that

(^) elei;:-

tions enabled citizens

(^) to (^) control

(^) policymakers

(^) and

(^) the formulation

(^) that

(^) elec-

tions enabled citizens

(^) to (^) influence

(^) policymakers. Initially, I

(^) thought

(^) that

(^) the

difference was one

(^) of (^) magnitude -

that

(^) control

(^) was

(^) a claim

(^) of (^) greater effect,

while influence implied a lesser effect. After

(^) much

(^) wrestling with the whole

array

(^) of (^) associated concepts, I have come

(^) to (^) think

(^) that

(^) the language points

(^) to (^) a

more fundamental divergence in conception

(^) of (^) the processes linking citizens

and

(^) policymakers.

I shall refer

(^) to (^) this divergence in conception as involving

(^) two

(^) visions

(^) of

elections as instruments

(^) of (^) democracy. Each unites a distinctive image

(^) of (^) the

electorate, a closely associated normative concept

(^) of (^) appropriate

(^) citizen influ-

ence,

(^) and

(^) an (^) empirical model

(^) of (^) the working

(^) of (^) electoral

(^) and

(^) legislative in-

stitutions. These constitute ways

(^) of (^) looking

(^) at (^) election processes as

(^) weil

(^) as

theories

(^) about

(^) how

(^) those processes

(^) work

(^) in practice. In a

(^) shorthand

(^) that

(^) is

generally consistent with recent usage in the empirical literature, I refer

(^) to

these as majoritarian

(^) and

(^) proportional

(^) visions

(^) of (^) elections

(^) and

(^) democracy.

The

(^) language

(^) of (^) elections as instruments

(^) of (^) control

(^) seems

(^) to (^) be associated

with a vision

of (^) concentrated

(^) policy-makirig

(^) power

(^) whose

(^) exercise can be

made the target

(^) of (^) citizen action.

(^) The

(^) elected officeholders are able

(^) to (^) make

and

(^) implement policies. Responsibility for policy

(^) is (^) obvious

(^) to (^) everyone.

(^) FLQ!ll

one perspective the citizens use elections

(^) to (^) choose b.etween prospective teams

of policymakers.

(^) Froill'

(^) another

(^) p'erspective

(^) the citizens

(^) ~~e

(^) eiections

(^) t~ (^) .rewar<l

or (^) punish the incumbents, While there are some very

(^) important

(^) disparities

between the

(^) forward~

(^) and

(^) backward-looking views, they

(^) both

(^) presume con-

centrated policy-making

(^) power

(^) exercised by officials

(^) who

(^) are the objects

(^) of

citizen electoral behavior. In the use

(^) of (^) elections

(^) to (^) control these powerful

policymakers, it

(^) is (^) the citizen majority

(^) that

(^) should, normatively, prevail over a

minority

(^) who

(^) supports the opposition.

(^) lt (^) is (^) the

(^) domination

(^) of (^) the

(^) majority

that

(^) gives such a vision its status as a democracy.Jn Tocqueville's words,

(^) "The

very essence

(^) of (^) democratic government consists in the absolute sovereignty

(^) of

the (^) majority"

([

5]

(^) 264). Hence, the

(^) term

(^) majoritarian

(^) to

~efer

(^) to (^) this

vision.

7

In the majoritarian vision

(^) of (^) citizen control, concentrated policy-making

power

(^) is (^) not

(^) undesirable. In fact, concentrated

(^) power

(^) is (^) necessary,

(^) although

not

(^) sufficient, for citizen

(^) cont~ol.

(^) If (^) power

(^) is (^) dispersed

(^) among

(^) officials, offices,

and

(^) issues, then policy making

(^) must

(^) be the outcome

(^) of (^) complex bargaining

between winners

and

(^) losers, ins

and

outs. As a result

(^) of (^) such dispersion,

retrospective responsibility can be difficult

(^) to (^) pinpoint,

(^) and

(^) elections

(^) may

(^) bear

only a tenuous relation

(^) to (^) the formation

of (^) winning policy coalitions.

(^) The

directness

(^) and

(^) clarity

(^) of (^) the connections

(^) that

(^) make

(^) this vision attractive de-

pend

(^) on

(^) concentrated political

(^) power

(^) that

(^) citizens can control.

6

Citizens, Elections, and Policy Making

On (^) the other band, the language

(^) of (^) elections

(^) as (^) instruments

(^) of (^) citizen influ-

ence

(^) is (^) more often associated with a vision

(^) of (^) dispersed policy-making power.

of the visionIn this vision elections play a more indirect role in policy making. The essence

(^) is (^) that

(^) the election brings representative agents

(^) of (^) all the factions

outcomes determining directly the makeupitarian approach views elections as mechanisms for tight control, with electioneach other in a flexible and accommodative fashion. The concentrated, major-in the society into the policy-making arena. These agents then bargain with

(^) of (^) the policymakers

(^) who

(^) will make

the representation of all points of view brought into an arenaall policies between elections. The dispersed influence counterpart emphasizes

(^) of (^) shifting policy

coalitions.

(^8)

Two important arguments underlie the claim

(^) of (^) the proportional influence

vision.

(^9) They offer related but conceptually distinct reasons for dispersing

power among representatives

(^) of (^) all groups:

I. izen choices, and election rulesElections are clumsy instruments. The intersection of party offerings, cit-

(^) is (^) complex. Many kinds

(^) of (^) distortions may

litical powerposition. Thus, using elections as a one-stage device for concentrating po-oric of election campaigns may make it difficult to locate the true majorityintervene between citizen preference and electoral victory. The heated rhet-

(^) is (^) hazardous (especially given the rules used in practice in

most majoritarian systems).

(^) lt (^) is (^) safer

(^) to (^) elect a legislature of representa-

This argumenttives and !et these representatives bargain to find the most preferred policy.

(^) is (^) essentially an empirical challenge to the working

(^) of (^) ma-

joritarian institutions in practice. In chapters

(^6) (^) and

(^8) (^) I shall test its validity.

In a democracy the preferences of all citizens,

(^) not

(^) just an electoral majority,

should be taken into account in the making of

(^) policies.

(^10) Even if they^

represent the citizen majority position on all the issues, a majority

(^) of (^) repre-

sentatives should

(^) not

(^) ride roughshod over the preferences (especially if

intense)

(^) of (^) the minority. The best guarantee that the majority will take ac-

count of minority preferences

(^) is (^) to give the minority some valuable policy-

checks and balances across institutions, and so on).making power (by consensual policy-making mies, regional governments,

(^) As

(^) John

(^) Stuart Mill

argued,

"Human

beings are only secure from evil

(^) at (^) the-Eänds

(^) of (^) others

in proportion as they have the power of being, and are,

(^) self-profecfing"

([:i86r]

(^11) If^ (^) one accepts its assumptions, this argument requires

a different democratic connection, a different standard

(^) of (^) democratic per-

formance, from

(^) that

(^) of the majoritarian

(^) or (^) concentrated power vision.

(^12)

I shall explore various implications

(^) of (^) these visions later in the book. At this

point I

(^) want

(^) only to emphasize the presence

(^) of

(^) these quite disparate general

Elections

(^) as (^) Instruments

(^) of Democracy

7

visions of the kind of connection

(^) that

(^) elections can and should establish be-

tween citizens and officials. In

(^) part

(^) these are two empirical

(^) th~ories

(^) about the

kinds

(^) of (^) arrangements

(^) that

(^) will usually result in poÜcymakers

(^) being

(^) ~o~;;ected

to (^) the voters. The proponents of concentrated power are

(^) more

suspicious

more desirousof the autonomy of elected representatives, less concerned about minorities,

(^) of (^) seeing that elites are clearly accountable to voters. The pro-

long as citizens havecreated by elections), less worried about the autonomy of policymakers asponents of dispersed power are more suspicious of majorities (especially those

had

(^) a role in selecting them, and less worried about

negotiated inaction. To (^) some extent one can test predictions about the relation between concen-

trated or dispersed power arrangements and electoral connections.

(^) One

(^) can

see (^) the extent

(^) to (^) which -

and

(^) why-each

(^) vision succeeds

(^) or (^) fails. But as two

(^) vi-

concentrated power vision focuses on majorities, assumingpolicymaker relationship, they do not always imply the same hypotheses. Thesions having something a little different in mind in conceptualizing the citizen-

(^) that

(^) one can rea-

sonably identify what citizens

(^) want-

(^) or (^) at (^) least

(^) who

(^) emerges from the election

with the citizens'

(^) support-and

(^) the problem of elections

(^) is (^) to (^) make policy-

makers follow that citizen directive. The dispersed power vision tends

(^) to (^) as-

sume

(^) that

(^) citizens are not a homogeneous bunch, and the main problem of

elections

(^) is (^) to (^) see that everybody and everybody's views get taken into account

in policy making.

(^) To (^) some degree, then, empirical predictions about the nature

ables and not really be alternative theoriesof the citizen-policymaker relationship will focus on dissimilar dependent vari-

(^) about

(^) achieving the same goal.

Elections and Policymakers: A

"Voter's-Eye"

View

Conceptions of how elections permit citizens to exert influence are sur-

prisingly varied. Different images of citizen control direct

(^) our

(^) attention to

different aspects of the connection

(^) that

(^) elections may create between the vot-

ers and the elected.

(^) Some

(^) of these connections may be complementary and

litical conditions.mutually supportive. Others require radically diverse, even contradictory, po- A (^) "voter's-eye"

(^) view of elections as instruments of democracy affords a

build from empirical studiesfresh outlook on issues debated in the theoretical literature and will also help

(^) of (^) voters. Assume for the moment

(^) that

(^) citizens

are fully participating in elections and are making the best choices they can.

(^13)

In (^) what

(^) ways can these choices influence the basic directing decisions

(^) of (^) gov-

ernment? Two dimensions of choice seem suggestive

(^) (fig.

(^) LI).

The vertical dimension

(^) of (^) citizen choice involves the target of voting. I am

rn

Citizens, Elections, and

(^) Policy

(^) Making

policy making

(^) that

(^) will take place after the election. The voter needs

(^) to (^) choose

an agent who he

(^) or (^) she believes will act

(^) as (^) the informed voter would have

acted if bargaining for him-

(^) or (^) herself. Ideally, it should be an agent whose

preferences mirror the voter's own across many issues, including issues

(^) that

have

(^) not

(^) yet arisen. Presumably, it should

(^) be (^) an agent with good bargaining

skills. In the lower-left-hand cell, the voter

(^) is (^) evaluating the past behavior

(^) of (^) her

or (^) his incumbent representative(s) in influencing policy as the voter would

his constituents in 1774, in which he arguedhave wished. The retrospective focus recalls Edmund Burke's famous letter to

(^) that

(^) they should trust him to

act in their best interests, subject only

to their retrospective oversight

17

-^

thus my characterization

(^) of (^) this combination

(^) as (^) representative trustee. This

role

(^) is (^) shown in parentheses because it

(^) is (^) given little attention in subsequent

chapters.

(^18)

The various visions

(^) of (^) the salient features

(^) of (^) the election will emphasize

different kinds

of (^) information for the voter and different kinds of policy-

making conditions for them to

(^) work

(^) weil.

(^) One

(^) can develop these further

(^) by

explicitly associating them with models

(^) of (^) citizen control that appear in theo-

retical and in ordinary language discussions

(^) of (^) the roles of elections in democ-

racy. These models draw attention to intermediate conditions

(^) that

(^) seem to

(^) be

necessary if elections are to serve as instruments

(^) of (^) democracy.

Later (especially in chapters 3, 4, Accountability, Mandates, Authorized Representation

(^) and

(^) I will describe the presence

(^) or

absence

(^) of (^) conditions that several models imply are important for elections. I

will sketch them here to prepare for

(^) that

(^) analysis. This sketch identifies more

of the roleexplicitly the relation between the two great normative and conceptual visions

(^) of (^) elections in empowering citizens and the alternative conceptions

of The simplest and perhaps most fundamental role^ electoral processes from the voter's-eye viewpoint.

(^) of (^) elections

(^) is (^) the evalua-

tion

(^) of (^) the incumbent government. Citizens consider the performance of an

incumbent party of policymakers and decide

(^) to (^) keep them in office

(^) or (^) throw

them out. Walter Lippmann wrote more than half a century ago,

(^) "To

(^) support

the Ins when things are going weil; to support the

(^) Outs

(^) when they seem to be

going badly, this

(^) is (^) the essence of popular

(^) government."

(^19) Elevated into a^

full-fledged model, Lippmann's argument corresponds

(^) to (^) the ideal

(^) of (^) elections

as enforcing

(^) accountability

(^) on the

(^) part

(^) of (^) governments.

(^) As

(^) a model the ac-

countability idea has been attractive

(^) at (^) two levels. At the more basic level it has

seemed a kind

(^) of

(^) minimal criteria

(^) of

(^) what

(^) elections must do: they must offer

Elections

(^) as (^) Instruments

(^) of (^) Democracy

II

citizens a periodic opportunity to change the policymakers.

(^) lt (^) is (^) a model

(^) that

appeals to those who desire clear citizen control yet are skeptical

(^) about

(^) the

capability of the citizens to form opinions

(^) on (^) complex policy issues.

The accountability model

(^) is (^) also appealing

(^) to (^) theorists who find it difficult

to (^) accept the collective electorate as able

(^) to (^) choose positive policy directions

across multiple issues. William Riker (1982a) puts this very

(^) plainly:"The

(^) es-

sence

(^) of (^) the liberal interpretation

(^) of (^) voting

(^) is (^) the notion that voting permits

the rejection

(^) of (^) candidates

(^) or (^) officials who have offended so many voters

(^) that

they cannot win an

(^) election."

20 Riker argues^

(^) that

(^) we can expect no more

(^) of

elections than this ability to permit the rejection of incumbents.

21 The com-^

plexities of public opinion make it impossible

to (^) interpret future-oriented

bentsvoter choices as endorsing governinent policies. The minimal veto on incum-

(^) is (^) the essence

(^) of (^) elections as instruments of control. Citizens will have an

influence because they will be able,

(^) at (^) least occasionally,

(^) to (^) reject policymakers

who are doing the wrong thing. The role

(^) of (^) competitive elections as enabling

unhappy citizens

(^) to (^) throw the rascals

(^) out

(^) has great appeal. Although many

theorists would like elections

(^) to (^) do even more, few would

(^) want

(^) to (^) abandon

this role entirely. To (^) other theorists the simple accountability model can be even more power-

ful

because the anticipation of possible rejection shapes the policies of the

incumbents.

22 In this view the power^

(^) of (^) the voters to

(^) throw

(^) the rascals

(^) out

(^) will

view of the uninterpretabilitywith voter review in mind. Despite the difficulties that social choice theorists'pressure on all incumbents to worry about the next elections and make policynot only keep obvious rascals from remaining in office, but also create a

(^) of (^) elections creates for general applicability

(^) of

such a model, there are certainly examples in both theory and practice

(^) that

suggest the power that policymakers' anticipation

(^) of (^) citizen desires has.

Part

(^) of (^) the appeal

(^) of (^) the accountability models

(^) is (^) their simplicity and the

limited knowledge

(^) that

(^) is (^) required of the electorate. Citizens do not have

(^) to

worry about the credibility of proposals and promises. But there

(^) is (^) one thing

that the citizens do have

(^) to (^) know: who was responsible for policy making.

They cannot make retrospective judgments

(^) about

(^) the incumbents unless it

(^) is

clear which incumbents made the policies. Closely linked

(^) to (^) this idea

(^) is (^) the

notion that the citizens need

(^) to (^) have a clear opportunity

(^) to (^) vote against those

incumbents. The critical condition for this model, then,

(^) is (^) clarity

(^) of (^) respon-

sibility,

a clarity

(^) that

(^) is (^) relevant in electoral terms.

(^) If (^) a single, unified party

had control

(^) of (^) all the policy-making resources and the citizens can vote for

or (^) against

(^) that

(^) party in the election, then clarity

(^) of (^) responsibility would be

high. But if policy-making resources were divided among numerous parties

(^) or

groups

(^) or

(^) if the policy-making coalitions changed from issue

(^) to

(^) issue, then it

I (^2)

Citizens, Elections,

(^) and

(^) Policy

(^) Making

would

(^) be (^) difficult for citizens

(^) to (^) use elections as instruments

(^) of (^) accountability.

They would

(^) not

(^) know

(^) whom

(^) to hold responsible. Moreover, accountability

requires that voter rejection

(^) of (^) the incumbents

(^) be (^) followed

(^) by

(^) their actually

losing policy-making power. In chapter 3

(^) will explore the degree

(^) to (^) which

sponsibilityelectoral conditions in contemporary democracies in fact offer clarity of re-

(^) and

(^) decisive incumbent replacement.

A second major role of elections focuses

(^) on (^) citizens looking ahead

(^) to (^) choose

future governments (see

(^) fig.

r. (^) r). This

(^) is (^) the top right cell in the figure: pro-

spective choice

(^) and

(^) national government

(^) as (^) a target. There

(^) is (^) a special body of

theoretical literature associated with this idea too.

(^) lt (^) is (^) often called the idea of

electoral mandates.

13 The focus^

is not (^) only on the incumbents, but on the

opposition party

(^) and

(^) the policy alternatives presented

(^) by (^) both. For example,

in an influential statement written nearly fifty years ago, a committee

(^) of (^) Amer-

political parties, including an effective opposition partyican political scientists discussed an ideal party system with strong, cohesive

(^) that

(^) "acts

(^) as a critic of

which are necessary for a true choice in reaching public decisions.the party in power, developing, defining and presenting the policy alternatives

24

The origins of this idea are discussed more fully in chapter

(^) 4, (^) but for the

moment it

(^) is (^) enough

(^) to (^) note some of the essentials of the model. First, it must

be possible for the voters

(^) to (^) be (^) able

(^) to (^) identify alternative future governments

at (^) the time

(^) of (^) the election. Although many writers in the tradition

(^) of (^) mandates

and

(^) responsibility have assumed only two parties, this need not be so.

(^) One

(^) can

imagine competing teams of party coalitions. But there should

(^) be (^) good reason

for the voters

(^) to (^) anticipate a tight connection between a vote for a party or

team of parties and the governments subsequently formed.

(^) If (^) voters think that

the formation

(^) of (^) policymaker coalitions

(^) is (^) disconnected from shifts in their

support-

(^) either because it

(^) is (^) predetermined

(^) or (^) because it

(^) is (^) incomprehensibly

complex-they

(^) have little reason to use their votes strategically to shape

(^) fu-

ture policy making. Second, this model requires

(^) that

(^) the winning electoral party (or coalition) in

fact dominate policy making after the election.

(^) If (^) the policy-making rules in

election, the impact of the voters' choicethe society force the electoral winners to negotiate with the losers after the

(^) is (^) weakened

(^) or (^) lost.

(^) discuss these

issues more fully in chapter

(^4) (^) and then describe the degree to which mandate

conditions are offered in elections in contemporary democracies. Both accountability

(^) and

(^) mandates fall within the vision of concentrated

power. They are two faces

(^) of (^) majoritarian control. The vision of dispersed po-

voters. The retrospective-prospective distinctionlitical power and citizen influence suggests rather different conditions for the

(^) is (^) not so powerfully associ-

ated with models

(^) of (^) representation and influence, although it appears in a lim-

ited way in the argument as

(^) to

(^) whether representatives should serve as trustees

Elections

(^) as (^) Instruments

(^) of (^) Democracy

13

or delegates of the citizens they represent.

(^25) More essential conditions, in my^

view,

(^) focus on each

(^) of (^) the two stages

(^) involved-

(^) the election and the post-

election bargaining. These two stages are discussed more fully in chapter

At the election level, as can be inferred from the lower-right-hand box in

figure

(^) r.1,

(^) voters will be choosing representatives

(^) to (^) bargain for them in policy

making. In this model it

(^) is (^) critical

(^) that

(^) each group of voters be able to find a

candidate

(^) or (^) party who will have their confidence and understand their views.

Equally significant, the voters' choices must result in the election

(^) of (^) represen-

tatives in proportion

(^) to (^) the size

(^) of (^) the group

(^) of (^) voters.

(^) As (^) the model assumes

collective government carriesthat policy-making influence begins as the representatives bargain (not as a

(^) out

(^) its promises

(^) or (^) anticipates future sanctions),

citizen influence will depend directly

(^) on (^) all the groups

(^) of (^) voters getting propor-

tional representation for all are so essentialtional representation. These linked concepts of desirable choices and propor-

(^) to (^) this vision of democracy that

they are often valued

(^) as (^) if they were democracy in their own right. Mill sum-

marizes this in the title

to (^) chapter

7 (^) of (^) Considerations on Representative

Government:

(^) "OfTrue

(^) and

(^) False Democracy; Representation of All, and Rep-

resentation of the Majority

(^) Only"

(^) ([1861]

(^) 1958,

(^) 102).

(^) He has no

(^) doubt

(^) that

representation

(^) "of

(^) the majority

(^) only"

(^) is (^) "false,"

(^) not

(^) "true,"

(^) democracy. Dis-

proportionate representation

(^) is

presumed

(^) to

imply a failure in democracy

itself. In chapter

(^) discuss the extent to which proportional representation of

votes

(^) is (^) achieved in elections in contemporary democracies.

At the postelection stage, the vision of dispersed power and citizen influence

tends

(^) to (^) assume at least flexibility in forming coalitions among all representa-

tives. There

(^) is (^) some tension here,

(^) as (^) observed earlier

(^) (see

(^) note 9 above), be-

tween theorists who

(^) see

(^) this flexibility

(^) as (^) a means for allowing different major-

ities to form on different issues and those who hope

(^) that

(^) it will allow minorities

to have some influence on all issues. In my examination in chapter

fall short

of identifying conditions

(^) that

(^) discriminate between these two views, preferring

election. This conditionof voters are provided with proportionate policy-making influence after theto look merely at the degree to which authorized representatives of each group

(^) of (^) authorized representation in policy making

(^) is (^) the

most removed from the voters of any

(^) that

examine. This distance reflects the

two-stage nature of the dispersed power and citizen influence vision.

(^) lt (^) reflects

the autonomy

(^) that

(^) this vision necessarily confers on authorized representa-

tives, who must serve as their voters' agents as new issues are debated.

Responsiveness

and

Representation

Even aside from the differences between control and influence, questions

about elections and the role

(^) of (^) citizens are hard

(^) to (^) answer, in

(^) part

(^) because they

r

Citizens, Elections, and Policy Making

a way that

(^) is (^) directly responsive to citizens' votes. I shall refer to this connec-

tion as responsiveness in the emergence

(^) of (^) policymakers.

Focusing on responsiveness defined in this way has both strengths and

weaknesses. The great strength

(^) is (^) that

(^) it builds on firm, largely indisputable

evidence about the choices made by citizens.

(^) If (^) we find that those electoral

we are certainly justified in being skepticalauthorizations seem to have little relationship to the people making policies,

(^) about

(^) the claims of elections

(^) to

provide citizens with even indirect participation in policy making. The problem with relying on votes

(^) to (^) measure citizens' preferences, how-

ever,

(^) is (^) that

(^) citizens can vote only for the candidates and parties offered to

offices and agreements among partiesthem. Usually they can choose only one party, although sometimes multiple

(^) about

(^) policy-making participation can

expand their options. But

(^) not

(^) knowing how satisfactory were the options

provided

(^) by (^) the party system, we do

(^) not

(^) know

(^) how weil the party they chose

can represent their preferences.

(^) We

(^) know

(^) only that relative to the alternatives

andin

(^) the given circumstances it seemed the most preferable.

28 We^ (^) know

(^) even

less about the parties

(^) that

(^) citizens did

(^) not

(^) select. Were some of these almost

(^) as

satisfactory

(^) as (^) their given choice, so

(^) that

(^) these parties would be acceptable

substitutes

(^) to (^) represent their position in policy making? Were others totally

distasteful? To get

(^) at (^) these questions we have

(^) to (^) go beyond the simple assump-

tions

(^) that

(^) voting choices are all we can

(^) know

(^) about

(^) citizens' preferences.

(^) We

must measure those preferences directly.

(We

must incorporate the A to B

linkage in figure

(^) I.2,

(^) rather than beginning with stage

B.)

In the second approach taken below, in

(^) part

(^) 3, I shall use the device of the

left-right scale

(^) to (^) attempt

(^) to (^) measure citizens' preferences and parties' posi-

tions. The left-right scale has huge advantages over simple vote choices.

(^) We

can

(^) see

(^) whether the chosen party

(^) is (^) really very close to the voters' preference.

We (^) can also see if the parties

(^) that

(^) make policy are very close

(^) to (^) the voters,

regardless

of (^) the specific voting authorization connection.

If (^) this measure,

which I shall refer to

(^) as (^) representational congruence

(^) between voters and pol-

icymakers,

(^) is (^) very strong, then elections seem to be performing weil

(^) as (^) instru-

ments

(^) of (^) democracy.

(^) On

(^) the other hand, if congruence

(^) is (^) poor, if the policy-

makers are quite far from the citizens, then elections are performing less weil. Although representational congruence seems plausible

(^) as (^) evidence

(^) that

(^) elec-

tions are performing weil as instruments

(^) of (^) democracy, it contains hazardous

assumptions. The questions

(^) or (^) techniques used to assess citizen preferences

must yield an accurate measure of those preferences.

(^) If (^) the left-right scale used

here

(^) is (^) not meaningful to the citizens,

(^) or (^) if citizens' preferences across many

issues are

(^) not

(^) being adequately aggregated in that scale,

(^) or (^) if the citizens are

not

(^) responding candidly,

(^) or

(^) if the survey sample

(^) is (^) not

(^) a good one,

(^) or

(^) if the

Elections

(^) as (^) Instruments

(^) of (^) Democracy

r

may appear to be close to what citizenscitizens use to place themselves, then one can commit grievous errors. A party placement of the parties on the scale does not reflect the same issue positions as

want

(^) when in reality some

of (^) its

policies are abhorrent to them, and vice versa. Of (^) course, if both responsiveness

(^) to (^) elections in choosing policymakers and

citizens' preferences leadthe representational congruence between the policymakers' positions and the

to similar assessments

about

(^) the performance

(^) of

elections

(^) as (^) instruments

(^) of (^) democracy, then we would be more confident in

our assessments.

(^) As (^) their likely sources

(^) of (^) error vary somewhat, finding com-

mon results would be reassuring.

(^) As

(^) we shall see, there are some grounds for

optimism

(^) about

(^) our parallel findings, but also some serious problems in com-

paring them.

In chapter The Subsequent Chapters

(^2) I continue my introduction

(^) to (^) the problem

(^) of (^) elections as

instruments of democracy

(^) by (^) examining the constitutional arrangements un-

constitutional arrangements to the alternative visions -der which elections and policymaker recruitment proceed. I link the basic

majoritarian control

and proportional influence -

of the role of elections in democracy.

Part

2,^ (^) containing chapters

3-6,

(^) examines the electoral experience in de-

important conditions for the two visionsmocracies over the past twenty-five years to see how frequently the most

(^) of (^) the controlling roles

(^) of (^) elections

were satisfied. lt builds on the assumption

(^) that

(^) citizen voting choices are the

critical starting point for evaluating democratic performance.

(^) Under

(^) the con-

centrated power vision, the majoritarian view, I shall look in chapters 3 and

(^4)

at accountability, the eviction

(^) of (^) unpopular incumbents, and election man-

dates.

(^) Under

(^) the dispersed power vision, I shall look in chapter

(^) at (^) election

sider the degreechoices, proportional outcomes, and authorized representation. I also con-

(^) to (^) which realization

(^) of (^) these conditions

(^) is (^) associated with the

constitutional designs discussed in chapter

(^) 2.

Scholarly literature

and

(^) political debate frequently refer to the value

(^) to

empirical investigation will be to measure the presencetoral mandates, authorized representation for all citizens. The first step in mydemocracy of such conditions as clarity of responsibility, interpretable elec-

(^) or (^) absence of such

conditions. The results are interesting in their

(^) own

right, even before one

attempts

to assess their relation

to (^) the connections

of (^) responsiveness and

right, in chaptercongruence. While these conditions are often valued as good in their own

(^6) (^1) (^) evaluate more directly the performance

(^) of (^) constitutional

designs in shaping responsiveness in choosing policymakers, a criterion of

I 8

Citizens, Elections, and Policy Making

democratic performance valued in each of the great visions.

(^) As (^) we shall see,

responsiveness, but only in theirthe empirical manifestations of each vision are relatively successful in creating

(^) own

(^) terms, a result that

(^) is (^) encouraging for

democracy but does little

(^) to (^) resolve competing claims between the ideals.

Part

(^) 3 assumes

(^) that

(^) in analyzing elections as instruments

(^) of (^) democracy we

preferences. In this section I seek the sources of the creationcan move behind citizens' voting choices and build directly on their political

(^) or (^) failure of

representational congruence. Chapter

(^7) (^) discusses the left-right scale as

(^) ade-

vice for measuring citizens' preferences, so

(^) that

(^) they can

(^) be (^) compared to the

positions

(^) of (^) parties in governments and influential policy making.

(^) lt (^) also spec-

each vision. I shall argue that proximity to the position of the median citizenifies the normative expectations and empirical hypotheses associated with

(^) is

a criterion valued in each vision, especially

(^) by (^) majoritarians. Chapter

(^8) (^) exam-

on the left-right scale. Chapterlinking legislatures and governments to the position of the median citizenines the successes and failures of the majoritarian designs and processes in

(^) 9 (^) revisits the proportional influence processes

ence, taking accountand shows their strengths and limits in representing the median citizen prefer-

(^) of (^) the influential policymakers as weil

(^) as (^) the govern-

tational congruence between citizens and policymakers.ments. Each vision expects its ideal electoral process to create good represen-

(^) As

(^) we shall see, they

are (^) not

(^) equally successful. The proportional influence designs enjoy a surpris-

ing advantage. Chapter

(^10)

(^) summarizes the findings of the book from a slightly different

governments andmance as measured by vote and preference connections between citizens andpoint of view. Shifting from elections to country averages, it shows the perfor-

by^ (^) majoritarian and proportional criteria.

(^) lt (^) also considers

relations addresses an important puzzleexplicitly the relations between desirable outcomes. Consideration of these

(^) about

(^) the writings on elections.

(^) On

one level, students

(^) of (^) democracy assume

(^) that

(^) competitive elections make a

difference. They are loathe

(^) to (^) accept a claim

(^) of (^) democracy

(^) by (^) rulers, however

weil intentioned, who do not permit electoral competition.

(^) On

(^) another level,

the same students

(^) of (^) democracy are perpetually complaining

(^) about

(^) the failure

of democratic elections

(^) to (^) provide citizens with adequate opportunities

(^) to

iansconfusing paradoxes of decentralized parties and divided government. Ital-exert the right kind of control in their society. Americans complain about the

(^) and

(^) Hollanders complain

(^) about

(^) government coalitions whose formation

representationseems divorced from election outcomes. The English complain about under-

(^) of (^) some voters

(^) and

(^) government dictatorship between elections

by (^) the representatives

(^) of (^) others. Some

(^) of (^) these complaints are indirect forms

(^) of

pleading for special policies. Others, however, genuinely point

(^) to

(^) respects in

Elections

(^) as (^) Instruments

(^) of (^) Democracy

I (^9)

which elections seem

(^) to (^) fail in offering conditions necessary for them

(^) to (^) serve

as A conclusion^ instruments of citizen control.

(^) that

(^) emerges most forcefully from the approach taken in this

book

(^) is (^) that even success in using elections as instruments of citizen control

political systems that have perfected most fully one(or influence) usually involves a trade-off between desirable processes. Those

(^) of (^) the major processes

rhrough which citizens seem to use elections for influence

(^) do (^) so (^) at (^) the expense

of one of the other processes. This trade-off

(^) is (^) not only a matter

(^) of (^) flaws in

constitutional design or human imagination (although we shall

(^) see

(^) some

(^) of

rhese).

(^) lt (^) is (^) built into the tension between concentrated and dispersed power

for policy making

(^) and

(^) the desirable consequences

(^) of (^) each. Students

(^) of (^) elec-

tions and democracy will always have something

(^) to (^) complain

(^) about

(^) because

electoral roles.no set of election arrangements can satisfy conditions for all the desirable

(^) We

(^) also see, however, that there

(^) is (^) much evidence

(^) of (^) the relative

success of elections performing as instruments

(^) of (^) democracy.

22

Citizens, Elections, and Policy Making

the combinations of election rules

(^) and

(^) decision rules in the twenty democ-

racies. The combinations enable

(^) us (^) to classify most of these constitutions

(^) as

primarily majoritarian

or

proportional, although other combinations are

quite possible.

The

Sample

of

Democracies

The sample of democracies in this study

is not

a random sample of

possible democracies

(^) or (^) even a random sample of democracies in the world

obtain relevant datatoday. Rather, it comprises all the working democracies for which I could

about

(^) citizen preferences and party alternatives for a

substantial set

(^) of (^) elections in the past twenty-five years

(^) or (^) so. Inevitably, this

thehas meant a heavy concentration of democracies from Western Europe, plus (^) United

(^) States, Canada, Japan, Australia, and New Zealand.

These countries share a number of features, including a high level of eco-

mocracynomic development and (except for Japan) a Western cultural heritage. De-

(^) is (^) now generally accepted as the appropriate form of political system.

have been around for most of the relevantThey have substantial democratic experience; most of their political parties

(^) Jives

(^) of most voters. The party

develop meaningful national organizations.leaders have had time to learn the intricacies of the constitutional rules and to

(^) All

of them except the

United

tialparty choices offered in electoral competition can be summarized to a substan- States operate under some variation of a parliamentary constitution. The

extent

(^) on a left-right

(^) or (^) interventionist-conservative dimension under-

stood

(^) by (^) voters and politicians alike.

There are substantial advantages in these commonalities. They control for

considerations such as a threat of military intervention

(^) or (^) an ongoing civil war

that

(^) can vitiate the role

(^) of (^) elections

(^) as (^) instruments of democracy. They ease the

other settings. Thisgeneralizing from the democratic experiences analyzed here to democracies inthat connect elections and public policies. Nonetheless, one must be careful inproblems of finding equivalent measures for some of the interesting variables

(^) is (^) obvious in the case of some of the !arge disparities in eco-

nomic and political conditions.

(^) lt (^) also applies to new electorates, new party

systems, and variations on the constitutional rules. In the concluding chapter

shall try

(^) to (^) be more specific about the limitations,

(^) as (^) weil

(^) as (^) the advantages.

The Election Ru/es

and

the Visions

of

Democracy

Elections may be shaped

(^) by

(^) many kinds

(^) of (^) rules. The election rules or

election laws refer here

(^) to (^) the rules

(^) of (^) representation

(^) by (^) which the votes of the

citizens are aggregated

(^) to (^) determine the winning candidate(s).

(^) As

(^) almost all

Constitutional Designs

23

critical authority to authorize public policies.representatives to the popular house of the national assembly, which has the these democracies are parliamentary systems, I shall concentrate on electing

EXPECTATIONS

(^) ABOUT

(^) EtECTION

(^) RULES

(^) AND

(^) POLITICAL

(^) PARTIES

A !arge political science literature

(^) is (^) devoted to analysis

(^) of (^) the conse-

quences of the election laws. The landmark in this literature

(^) is (^) Maurice Du-

verger's assertion in

(^) r (^) 9 5 4 (^) that

(^) single-member district plurality (first past the

post) rules tend

(^) to

produce two-party systems, an observation known

fa-

mously

(^) as (^) Duverger's Law.

(^3) Under

(^) single-member district plurality rules the

nation

(^) is (^) divided into geographic constituencies; in each constituency candi-

the representative.more votes than any other candidate, a plurality, wins the district and becomesdates compete to win a single representative position. The candidate winning

(^) As

Riker has shown, effects favoring two-party

(^) or (^) two-

1982b).British debates about franchise extensions in the nineteenth century (Rikercandidate competition under these rules were widely suspected during the

(^) That

(^) single-member districts could unfairly disadvantage some candi-

dates

(^) or (^) parties was also known sufficiently weil to encourage the various

nineteenth-century inventions

(^) of (^) alternatives

(^) -districts

(^) with multiple repre-

verger hypothesizedsentatives chosen according to some rule of proportional representation. Du-

(^) that

(^) proportional representation election rules would en-

cal evidence for them.ing these ideas and suspicions into a scientific theory and offering some empiri-courage multiparty systems. Duverger, properly, gets the credit for generaliz-

(^4)

The political science election law literature since Duverger contains both

seldom found together.theoretical and empirical analyses, although until fairly recently these were

(^5) There are three elements in the theoretical analysis:

citizensthe decisions of parties as they offer candidates in the election; the decisions of

(^) as (^) they vote; the rules that aggregate the citizen partisan choices

(^) to

fixed (although they may,determine the winning representatives. In a given election the election rules are

(^) of (^) course, be modified over time for partisan advan-

elementtage or other reasons); the citizen and party decisions interact. The essential

(^) is (^) that

(^) the election rules limit the number of representatives from a

district, so only slightly more than

(^) that

(^) number

(^) of (^) candidates have a realistic

chance of winning. In a single-member district, only the two leading candidates have much

chance

(^) of (^) winning, unless the difference in support between the second

(^) and

third

(^) is (^) very slight.

(^) If (^) a district has three or four

(^) or (^) five

(^) representatives, this

same logic applies to the fifth, sixth, and seventh candidates.

(^) If (^) the citizens

know this and know the rough prospects

(^) of (^) the candidates, they may vote

"strategically,"

(^) that

(^) is, vote only for candidates who have a reasonable chance

24

Citizens, Elections, and Policy Making

of (^) winning rather than wasting their votes on candidates who, however ap-

pealing their ideas, have no chance.

(^) If (^) parties know roughly the prospects of

they willthe candidates, and especially if they expect some citizens to vote strategically,

(^) not

(^) waste effort running candidates in situations with

(^) poor

(^) pros-

pects. Rather,

(^) to (^) maximize their overall chances, they will

(^) work

(^) out

(^) agree-

ments with similarly oriented parties

(^) to (^) divide the districts in which

(^) to (^) offer

candidates

(^) or (^) join to form !arger parties.

In practice there are many possibilities for making Strategie mistakes, de-

pending on the information available and the numbers

(^) of (^) potential competi-

tors. In early democratic elections we expect all kinds of mistakes

(^) by

(^) candi-

dates, parties,

(^) and

(^) citizens, mistakes

(^) that

(^) can lead to very strange representa-

tion. (Striking examples appeared in early elections in

(^) Poland

(^) and

(^) Russia, in

which a third

(^) or (^) more of the citizens voted for parties that failed to gain

ber of candidates competing in a district willfairly low threshold.) Eventually, however, we expect in theory that the num-legislative representation, even with proportional representation rules and a

(^) be (^) reduced toward

(^) no (^) more than

the district magnitude (its number

(^) of (^) representatives) plus one (Cox

(^) 1997).

Thus, in single-member district elections, we expect theoretically

(^) that

(^) with

candidates. Moreover, the greater the magnitudecandidates will run in each district. The greater the magnitude, the morethe eventual development of an equilibrium of expectations only two serious

of (^) a district, the less the

consequences for representation

(^) of (^) a strategic mistake concerning a single

pothesisparty. These expectations underpin Duverger's Law and its counterpart hy-

(^) about

(^) !arge district magnitudes and multiple parties.

6

Although the theoretical logic underpinning Duverger's Law assumes that

rules still produce only one winner per district regardlessvoter support to consolidate around only two candidates in a district, thesesingle-member district election rules will encourage party competition and

(^) of (^) the number com-

district's outcome, even if they constitutepeting. Thus, the voters for the losing party are always unrepresented in the

(^49) (^) percent

(^) of (^) its district's electorate.

If (^) the same party finds itself in this situation across most of the districts, the

ture. (The famous American inventioncollective outcome will find its voters badly underrepresented in the legisla-

(^) of (^) gerrymandering the district bound-

advantageary lines for the advantage of one party and disadvantage of another takes

of (^) this possibility.) For single-member district election rules to

districts (see the discussion intional outcomes in each district must compensate for each other across all theavoid serious disproportionality of seats in the legislature the many dispropor-

(^) Powell

(^) and

(^) Vanberg forthcoming). There

(^) is (^) no

guarantee

(^) that

(^) this will happen even if strategic anticipation reduces the num-

bers

(^) of

(^) candidates

(^) to

(^) two in each district.

Constitutional Designs

(^2) 5

If

(^) party and citizen strategies do not reduce the number of competing parties

to (^) two, the effect of the single-member district winner-take-all rules may still

tions in vote-seat relations. The former possibility maytive majorities for one of them. They may also create some spectacular distor-greatly reduce the number of parties in the legislature and even create legisla-

(^) be (^) helpful for realizing

the majoritarian vision; the latter possibility

(^) is (^) disconcerting to the propor-

tional vision.

Such

majorities and such distortions are also possible with

particular distributions of party support across constituencies, itmultimember-district proportional representation, but seem less likely. With

(^) is (^) also possi-

ble (^) that the party winning the most votes in the election will not win the most

both visions.seats in the legislature. This outcome would be normatively disconcerting to There are, of course, many variations on the simple dichotomy of single-

date attains some basic threshold (usuallyresentation. Single-member districts may use multiround runoffs if no candi-member district plurality rules and multiple-member district proportional rep-

(^) 50 (^) percent) in the first round, as in

France and in some American primary elections.

(^) Or

(^) they may allow voters to

losers (and even winners in some systems)indicate a !arger preference ordering among candidates, with wasted votes of

to be transferred to candidates

offer varying district magnitudes (fromgetting more support, as in Australia. Proportional representation systems

(^) in Ireland to a single, national

district of

(^) l 50 (^) representatives in the Netherlands), multiple tiers of candidates

chosen in different ways, alternative formulae for calculating the

"unused"

remainders

(^) of (^) votes, ordinal preference possibilities, and various thresholds of

representatives assignedrelationship between the number of voters in a district and the number ofover, whatever the system, proportional outcomes will depend on an equitableminimal amounts of voter support necessary to achieve representation. More-

(^) to (^) it. (^7)

Empirically, we can simplify the many complexities

(^) of (^) election rules by

drawing on the analysis of election laws presented

(^) by (^) Lijphart

(^) (1994)

(^) in his

study of election laws in twenty-seven democracies from

(^1945)

(^) to (^) 1990.

(^) Lijp-

hart analyzes each of the systems

(^) of (^) election laws used in these elections and

combines various features into his concept

(^) of (^) the

"effective

threshold"

of

representation. The effective threshold attempts to capture the difficulty

(^) that

(^) a

conceptsmall party faces in achieving legislative representation through elections. The

(^) is (^) most easily understood when applied to a proportional representa-

intion system that literally has a minimum threshold for national representation: (^) Sweden

(^) a party must win

(^4) (^) percent of the national vote

(^) to (^) gain any represen-

tation in the legislature; in Germany a party must win

(^) percent

(^) of (^) the national

vote (or three individual district seats) to gain any

(^) of

(^) the !ist seats in the

Table

(^) 2.r.

Election

(^) Rufesand

(^) Consequences in Twenty Democracies:

Fewer

(^) Parties

(^) and

(^) Greater Disproportionality

Number Effective

Number

Legislative

of (^) Parties

(^2)

Vote/Seat

of

Elections Rules

(^1)

Votes

Seats

Disproportionality

(^3) Elections

Multimember districts -

pure proportional representation

Austria

( 2.6)

(

Belgium

( 4.8)

( 9)

Denmark

( 1.6)

(11)

Finland

( 5.4)

(

Germany

( 5.0)

(

*Greece

( 3.3)

(

ltaly

( 2.0)

(

Netherlands

( .7)

(

*Norway

( 4.0)

(

Sweden

( 4.0)

(10)

Switzerland

( 8.5)

(

Multimember districts -

increased potential distortion

*France

(12.0)

(

*Greece

(16.1)

( 4)

Ireland

(17.2)

(

Japan

(16.4)

(

*Norway

( 8.9)

( 5)

Spain

(10.2)

( 5)

Single-member districts Australia

(35.0)

(11)

Canada

(35.0)

(

*France

(35.0)

(

NewZealand

(35.0)

(

United Kingdom

(35.0)

(

United States

(35.0)

(

*Major

(^) changes: Greece in

(^) 1989-

(^) only;

(^) Norway

(^1989)

(^) ff, (^) France

(^1986)

(^) (PR).

(^) Austria

(^) and

Sweden used higher

(^) PR (^) thresholds only for the very first election covered. ltaly 1994 excluded.

"Effective^1 N=155.

(^) Threshold"

(^) (Lijphart 1994,

(^) 25-30)

(^) in parentheses.

"Effective^2

(^) Number

(^) of (^) Parties"

(^) from Laasko

(^) and

(^) Taagepera

(^) 1979·

Gallagher disproportionality measure (Lijphart 1994,^3

(^61) (^) ).

28

Constitutional Designs

29

1 . Number

(^) of (^) Parties.

(^) As (^) we read the down three groups of election rules in

ipatingthe table, we expect to find fewer parties competing in the elections (antic-

(^) that

(^) high thresholds will make it

(^) hard

(^) for small parties

(^) to (^) win) and,

The election averages show us how weil this expectationian vision of democratic elections usually assumes two-party competition.especially, fewer parties winning legislative representation. The majoritar-

(^) is (^) met. The num-

winning votes and seats, respectively. For both votesbers in the first two data columns show the effective number of parties

(^) and

(^) seats this number

is (^) the equivalent of the simple number

(^) of (^) parties if their sizes were similar.

(^) lt

is (^) a rough guide to the number

(^) of (^) competing parties that takes account of

parties' relative size

(^) and

(^) asymmetry.

(^11)

The averages in the table

(^) fit (^) our

(^) expectations fairly weil. The (effective)

number

(^) of (^) parties winning votes generally declines

(^) as (^) we read down the

categories in the table. However, we

see (^) that

only the

(^) United

(^) States

(^) is

the electoral competitiontruly a two-party system in this sense. Even in the bottom-most category,

(^) is (^) usually between the equivalent

(^) of (^) three parties.

In chapter 8 we shall see that this seems

(^) to (^) create problems in achiev-

ing some

(^) of (^) the putative benefits

(^) of (^) two-party electoral competition. Ex-

election,In all countries fewer parties win seats in the legislature than run in thepectations about parties in the legislature are fulfilled more satisfactorily.

(^) but

(^) the reduction (comparing the first and second columns un-

der parties)

(^) is (^) slight in the low-threshold PR systems, quite substantial in

some

(^) of (^) the countries in the bottom categories. Because of the dual effect

of generally fewer parties running

(^) and

elimination of

(^) some

party rep-

resentation, the

(^) number

of parties in the legislature declines sharply

(^) as

we descend the categories in the table. There are effectively

(^) about

/2^1

parties in the legislatures of the pure PR systems, 3

(^) to

4 (^) parties in the

intermediate threshold cases, and only

(^2) /2^1 (^) parties in the single-member

the (effective) number of parties winning votesdistrict systems. The correlation between Lijphart's effective threshold and

(^) is

-.43,

(^) while the corre-

lation with number

(^) of (^) parties in the legislature

(^) is

-. (^) 5

(^) To

(^) this extent,

majoritarian processes as we move down the table. In chapters 3 andwe are justified in considering the election rules more likely to encourage

(^4) (^) I

directly.ber of competing parties to consider accountability and mandates moreshall explore this relation in much more detail, moving beyond the num- We (^) can also

(^) see

(^) in table

(^) 2.1,

(^) however,

(^) that

(^) various special circumstances

of history and the nature of the society create interesting exceptions

(^) to

these strong general patterns. Austria has only an average of

(^) 2. (^7) (^) parties

competing in the elections in this period, despite PR with a low threshold;

30

Citizens, Elections, and Policy Making

France has effectively

(^) or more parties, even using single-member districts.

Constitutional design

(^) is (^) not the only factor shaping the party system.

Vote-Seat Disproportionality.

(^) The proportional vision presumes multiparty

competition, so that a variety of points of view can

(^) be (^) represented, and the

fair representation of all parties in the legislature.

(^) We

(^) have already seen that

ofthe third data column, the table shows a measure of the disproportionalitythe number of competing parties does decline as we move down the table. In (^) the relation between votes and seats,

(^) that

(^) is, the degree to which the

distribution

(^) of (^) party seats in the legislature fails

(^) to (^) correspond to the votes

won

(^) by (^) the parties in the elections. While there are many measures of dis-

11

proportionality, the number here

(^) is (^) based on the ratios of vote percentage to

11

considered the seat percentage for the parties, in a particular formula (Gallagher's) usually

(^) bestand

(^) most widely used in vote-seat proportionality anal-

ysis.

(^) Various empirical analyses (for example, Rae

(^) 1967,

(^) Lijphart

(^) 1994)

have shown the strong relation between effective threshold

(^) and

(^) vote-seat

tionality as we read down the groupsdisproportionality. Therefore, we expect to find more vote-seat dispropor-

(^) of (^) the table. The theoretical relation

is (^) somewhat more complex than sometimes assumed, as I shall discuss in

chapter

(^) 9 (^) (also

(^) see

(^) Powell

(^) and Vanberg forthcoming,) but the expectations

here are consistent with the Standard findings. Indeed, the outcomes are strikingly consistent with these expectations

and with Lijphart's

(1994)

(^) results on effective threshold and dispropor-

2.3,tionality. The average disproportionality numbers for the three groups are

(^) 6.o, and

(^) l r.8.

(^) The correlation between effective threshold

(^) and

(^) dispro-

portionality

(^) is (^).

(^) (N

(^) = (^) l (^) 5 5 ). (^) lt (^) is (^) notable that the single-member district

system with the least misrepresentation in vote-seat terms

is the (^) United

Stares, whose disproportionality

(^) is (^) still higher than

(^) that

(^) in any

PR

(^) system

with under a

(^) ro (^) percent threshold. Low threshold

PR

(^) systems are without

todoubt creating much fairer party representation in the legislature and are, (^) that

(^) extent, encouraging

(^) of (^) the proportional influence vision.

I need

(^) not

(^) add to the large literature on consequences of election laws

(^) at (^) this

point. Table

(^) 2.

(^) (and the literature discussed above

(^) andin

(^) the notes) should

(^) be

sufficient to justify the classification

(^) of (^) electoral systems by the election mies.

visions. However, we need alsocharacteristics of the election more precisely linked to the various democraticable, offers a useful starting point as we begin in the next chapter to examineThis classification, as well as Lijphart's continuous effective threshold vari-

(^) to (^) consider features

(^) of (^) the decision rules for

the policy-making process

(^) that

(^) follows the elections.

Constitutional Designs

(^3) r

The Policy-Making

Rufesand

the Visions

of

Democracy

The elections allow citizens to choose the potential policymakers. Bur

general set of mles. I call these mles about howafter the elections those representatives act to make policies following another

(^) to (^) make policies the decision

rules.

(^) As

(^) in the case of the dection mles, important features of the decision

or even in less formal norms of behavior. Forrules may appear in a single constitutional document, in ordinary legislation,

(^) our

(^) purposes, the critical point

(^) is

encourage policy-making processes closer to the proportional vision.rhat some decision mies encourage majoritarian policy making, while others In (^) many ways the twenty countries in this book share quite similar decision

assembly chambersrelatively centralized parliamentary systems with single or similarly electedof the final power to authorize public policies. Moreover, most of them arerules. They are all democracies in which an elected national assembly has most

(^) and

(^) limited

(^) or (^) relatively inactive judicial review systems. I

shall refer to these

(^) as (^) pure parliamentary systems. The similarity

(^) of (^) the deci-

sion mies

(^) is (^) helpful in giving

(^) us (^) substantial confidence in the comparability of

the degree of sharing or concentrating power in the assembly. Then, Iing a critical dimension of the decision mle common to all these democracies,to democracies with different kinds of decision mies. I shall begin by consider-our measures and analyses. However, it limits the applicability of the analysis

(^) turn

(^) to

decision mle features relevant to only a few of these countries.

CONCENTRATION

(^) OR

(^) DISPERSION

(^) OF

(^) POWER

(^) IN (^) THE ASSEMBLY

Such countries as Britain, Sweden, and the Netherlands are pure parlia-

mentary systems. Authority to make public policies

is (^) concentrated in the

rulesnational assembly. Nonetheless, these countries vary strikingly in the decision

(^) that

(^) shape the policy making of the assemblies. In the usual language

(^) of

and the opposition parties within the assembly. Ithe assembly. They also regulate the relation between the government partiesment, whose cabinet controls the bureaucratic agencies of the executive, andparliamentary politics, these mles regulate the relation between the govern-

(^) want

(^) to distinguish between

rules that encourage a single parliamentary majority

(^) to (^) control policy making

and rules

(^) that

(^) encourage all parties to have influence on policy making and

implementation proportional to their size. I take for granted here

(^) that

(^) political parties in mature parliamentary systems

are generally very cohesive in their voting behavior.

Obviously, when legisla-^

tors

(^) in (^) the governing parties frequently break ranks, this Jack of party disci-

pline will create opportunities for opposition parties. This feature

(^) of

(^) legislative

Table

(^) 2.2.

Legislative Decision Ru/es: Concentrating Governmental

(^) Power

Country

Formal Committee Structure

(^1)

Over

(^) Ten

Standing

Committees

Corres

(^) ponding

to Government Departments

OppositionChairs toCommittee

Rules facilitate opposition influence

(^) in (^) legislature

Austria

Yes

Shared

Belgium

Yes

Shared

Denmark

Yes

Shared

Germany

Yes

Shared

Netherlands

Yes

Shared

Norway

Yes

Shared

Spain

Yes

Shared

Sweden

Yes

Shared

Switzerland

Yes

Shared

Rules encourage some dispersal of influence

(^) in (^) legislature

Canada

Yes

No

ltaly

Yes

No

Japan

Yes

No

United

(^) States

Yes

No

Finland

No

Shared

Rules support domination of legislature

(^) by (^) government

Australia

No

No

France

No

No

Greece

No

No

Ireland

No

No

NewZealand

No

No

United

(^) Kingdom

No

No

Associated Features of Assembly Rules

2

Government:

MediumMedium AgendaAssembly Controls Low

Medium Low

MediumMedium HighLow

-^ Low

High - HighHighHigh- Low

Limits

AmendmentCommittee

NoNoNoNoYesNoYesNoNo

(Some) YesYesYes(Yes) No(No)(Yes)No Yes(Yes)

From^1

(^) IPU

(^) 1986, except for Greece, which limited

(^) number

(^) of (^) committees when democracy

was reinstituted in late

(^) l97os;

(^) see (^) Mattson

(^) and

(^) Strom 1995,

(^261)

(^) ff. (^) and

(^) Damgaard

(^199)

(^) 5, (^) 31 (^) r.

From Doering^2

(^199)

(^) 5, (^) 225

(^) (l-III=High;

(^) IV (^) =Med;

(^) V-VI=Low),

(^) based

(^) on (^) expert assessments;

Doering

(^199)

(^) 5, (^) 236; parentheses are my interpretation, based

(^) on (^) Doering's original source

(IPU

(^) 1986, table 33.4). For

(^) Canada

(^) also see the special issues

(^) of (^) Parliamentary

(^) Government

43, (^) 44 (June

(^) and

(^) August 1993).

34

Constitutional Designs

3 5

Belgium, Denmark, Norway,

(^) and

(^) Sweden

(^) and

(^) the

(^) popular

(^) houses in Germany

and the Netherlands. Variations on this arrangement give more committees

(^) to

the government,

(^) but

(^) still a substantial number

(^) to (^) the opposition in Austria,

divide committee chairs only between the

(^) two

(^) major parties in Spain,

(^) and

(^) give

each substantial party exactly

(^) two

(^) chairs in Switzerland. Thus, in the

top

group the combination

(^) of (^) many committees whose specialization corresponds

to (^) government departments and sharing chairs with the opposition suggests

substantial opportunities for the opposition

(^) to (^) influence legislation.

In (^) the other countries, the opportunities seem

(^) tobe

(^) fewer. The middle group

may have stronger committees,

(^) but

(^) the government dominates the committee

chairs.

(^) Only

(^) intraparty

(^) or (^) intracoalition divisions (admittedly fairly

(^) common

in (^) Italy,

(^) Japan,

(^) and the

(^) United

(^) States, as we shall see later) can spread oppor-

tunities for the opposition members. Finland shares the committee chairs,

(^) but

the committees themselves are weak. In the countries

(^) at

the

bottom

(^) of (^) the table, we find the combination

of

committee systems

(^) that

(^) are greatly disadvantaged relative

(^) to (^) the government

cedures, may be givenspecial committees, such as committees overseeing particular government pro-and committee chairs largely in government hands. (In a few countries some

(^) to (^) the opposition.) We

(^) would

(^) expect

(^) that

(^) in these coun-

tries the opposition sho,1ld have little opportunity

(^) to (^) influence legislation un-

less

(^) it can persuade the government

(^) of (^) the rightness

(^) or (^) electoral appeal

(^) of (^) its

arguments. The

(^) two

(^) columns

(^) on (^) the right

(^) of (^) the table present additional information

about the assembly mies

(^) that

(^) offer some validation

(^) of (^) our

(^) expectations and

the classification.

(^) We

(^) do

(^) not

(^) use them for the classification itself as most are

available only for the European countries included in the Doering (r 99

(^) 5)

volume. Legislative mies

(^) that

(^) directly bear

(^) on (^) the power

(^) of (^) the governm_ent

versus the committees in shaping legislation can

(^) show

(^) the strength

(^) or (^) weak-

ness

(^) of (^) the classification based simply

(^) on (^) number

(^) and

(^) jurisdictional specializa-

tion. The analysis

(^) of (^) Mattson

(^) and

(^) Strom (

r 99

(^) 5) (^) suggests

(^) two

(^) dimensions

(^) of

committee influence, one linked

(^) to (^) control

(^) of (^) the assembly's agenda

(^) and

(^) the

other linked

(^) to (^) the ability

(^) of (^) the committees

(^) to (^) amend legislation submitted by

the government. The

(^) two

features shown explicitly here lie

(^) at (^) the heart

of

those

(^) two

(^) dimensions

(^) and

(^) are fairly self-explanatory.

The analysis by Doering (

(^) 5, (^) 225)

(^) classifies the government's control over

the assembly agenda

(^) on

(^) a seven-step ordinal ranking from most government

control

(I

(^) = (^) government alone determines the plenary agenda)

(^) to (^) least govern-

ment control (VII

(^) = (^) chamber

(^) itself controls the agenda). The table collapses

those into a Low-Medium-High classification. There

(^) is (^) a fairly

(^) good

(^) fit (^) with

the general committee stmcture. The government dominates the legislative

(^) 6

Citizens, Elections, and Policy Making

only in Switzerland agenda in all the countries in the bottom group in the table, while elsewhere

(^) is (^) there more than moderate government control over the

agenda. (In Switzerland, over

(^) percent

(^) of (^) the legislators are members

(^) of

parties directly represented in the government.) The column on limitations on the committee amendment powers

(^) is (^) also

drawn from Doering's analysis,

(^) but

(^) here he interprets the

IPU

(^) reports

rather than experts in his own study.

(^1) (^) have

(^) put

(^) in parentheses my

(^) interpreta-

tion

(^) of (^) the same data, after reviewing Doering's classification. The main

(^) dis-

tinction in the rules

is (^) whether the committee can completely replace the

government's initial wording with a new draft text

(^) or (^) whether it

(^) is (^) restricted

(^) to

mostamendments of the government's original text within general constraints. In

of (^) the countries in the top group

and

(^) even in the middle group, the

committees are free

(^) to (^) substitute their own drafts for the government's

(^) origi-

nal language. In Austria a substitute text may be offered, but it must

(^) be (^) consid-

ered against the original, as in the

U.S.

(^) House. The exceptions are committees

legislature shown in the previous column), Japan, in Denmark, the Netherlands (despite the maximum agenda freedom of the

(^) and

(^) to some extent Canada.

constrainedIn contrast, committees in all the countries in the bottom group are severely

(^) to (^) work

(^) with the government's original text. The purpose

(^) of (^) com-

mittees in most

(^) of (^) these countries

(^) is

not

primarily shaping of policies but

oversight

(^) of (^) the government bureaucracy. Thus, the government's capacity to

control the assembly agenda and

(^) to (^) limit committee amendments generally

used to classify the committee systems.coincides with the organization and chair distribution arrangements that were

(^16)

INSTITUTIONAL

(^) FEATURES

(^) THAT CAN

(^) DISPERSE

(^) POWER

OUTSIDE

(^) THE

(^) NATIONAL ASSEMBLY

Most

(^) of (^) the democracies studied here are parliamentary systems

(^) featur-

ing relatively centralized political authority. In such democracies the main

(^) fea-

tures

(^) that

(^) organize the concentration or dispersion of power are the election

rules and the rules for making policy inside the legislature, which

(^1) have been

discussing. The dependence of the executive on the legislature in

(^) parliamen-

tary systems fuses control of both and contributes to high levels

(^) of (^) legislative

the legislative opposition will have influence on government-originatedthe formation of governments; the committee rules help determine whethervoting cohesion in governing legislative parties. The election rules help shape

(^) legis-

lation.

(^1) (^) have devoted most attention

(^) to (^) these two types

(^) of (^) rules because they

dominate

(^) our

(^) sample of democratic constitutions. In a few

(^) of (^) these

(^) democ-

racies, however, the decision rules offer points

(^) of (^) institutional influence

(^) out-

side the assembly. When one party

(^) or (^) coalition of parties controls the legisla-

Constitutional Designs

7

rure and another controls some outside veto point, then shared influence -

if

only in the form of

(^) deadlock-

(^) is (^) inevitable. Several of these possible veto

institutions are

(^) worth

(^) brief comment:

I. Independent Executives.

(^) The most famous constitutional distinction in

(^) de-

mocracies is,

(^) of (^) course,

(^) that

(^) between parliamentary and presidential

(^) sys-

arrangements in many of the new democracies tems, a contrast complicated by the explosion of mixed legislative-executive

(^) of (^) the

(^) l99os.

(^) A separately

over legislation, authorityelected executive with independent policy-making powers (such as a veto

(^) to (^) issue decrees,

(^) or (^) ability to dismiss the

(^) legisla-

ture) can give rise to divided government situations with dispersed

(^) oppor-

tunities for influence.

(See

(^) Shugart and Carey [1992, 155] for a !ist

(^) of

presidential powers in different systems.) Among

(^) our

(^) twenty systems,

(^) how-

ever, only the

(^) United

(^) States

(^) clearly has a strong presidential constitution.

The French president has

(^) had

(^) a sufficiently independent role to

(^) be (^) consid-

ered a point for opposition influence under the

(^) "cohabitation"

(^) situations of

the late

l98os

and mid-199os.

(^17) In both cases we must take account of^

divided government possibilities in examining the role of elections as

(^) instru-

ments

(^) of (^) democracy in policy making.

Separate Legislative Chamber

(^) with

(^) Independent Selection Base

(^) and

(^) Veto

Powers.

(^) Many

(^) democracies have more than one legislative chamber. For

our purposes, these are interesting

(^) as (^) a strong basis for opposition influence

and if its basis of electiononly if the second house can effectively veto legislation passed by the first

(^) is (^) likely

(^) to (^) create different majorities (see Lijphart

1984, chapter 6). The British House

(^) of (^) Lords and the Canadian

(^) Senate

(^) no

were elected bylonger have serious veto influence. The Belgian and Italian upper houses

PR

(^) with rules very similar

(^) to (^) those

(^) of (^) the lower houses

of our democraciesand simultaneous elections, thus resulting in similar party balances. In four

(^) -Australia,

Germany, Switzerland, and the

(^) United

States

the upper house has sufficient power to influence policy

(^) signifi-

cantly if controlled

(^) by (^) the opposition. Australia and Germany offer a

(^) num-

ber of instances of such opposition control

(^) that

(^) we shall have to consider in

analyzing the postelection situations.

Federal Systems.

Regional governments with independent bases of

(^) elec-

tion and policy making can

(^) be (^) sources of influence for opposition

(^) politi-

hause,^ cal parties. Where such governments are represented directly in the upper

(^) as (^) in Germany, we shall already have captured some of the effect.

But regional governments also have some direct influence over policy

(^) mak-

ing and its implementation in Germany as weil

(^) as (^) in Australia, Austria,

Canada, Switzerland, and the

(^) United

(^) States. Regional governments have