1_8 Carolien Van Ham Why Do Elections Fail _2

8/13/2019 1_8 Carolien Van Ham Why Do Elections Fail _2

http://slidepdf.com/reader/full/18-carolien-van-ham-why-do-elections-fail-2 1/32

www.electoralintegrityproject.com

Working Papers Series

Carolien van Ham

(University of Twente)

Wh d El i F il?

This paper develops an explanatory framework to understand (1)

whether elections fail due to lacking capacity or due to intentional manipulation, (2) whether elections fail due to long ‐term structural factors or rather short ‐term contingent factors.

SYNOPSIS



Why do Elections Fail?

Explaining election integrity in third and fourth wave regimes

Carolien van Ham ab

a Centre for the Study of Democracy, University of Twente, Enschede, The Netherlandsb Electoral Integrity Project, University of Sydney, Sydney, Australia

[email protected]

AbstractIn the wake of the third and fourth wave of democratization, the number of countries holding regular elections

for executive and legislative offices sharply increased: over 85% of the world’s states now select the ir national

leaders through elections. However, while global norms for elections increasingly converged, global practice

shows a widely varying “menu of manipulation”. Thus, while most states hold formally democratic elections and

commit to international s tandards for elections, the de facto quality of elections ranges from “free and fair”

elections with genuine contestation between parties/candidates to “façade” elections that are marred by

i l ti d f d I light f th i i l d l t h l t l i t g it i i i gl



1. Introduction

Almost a year ago, on May 23 2012, Egyptians held their first presidential elections

with multiple candidates in over 30 years. While it is still very unclear where the region is

heading, after the post- 1974 “third wave” and the post -1989 “fourth wave” of

democratization, hopes are that the Arab spring might come to constitute a new ‘wave’ of

transitions away from authoritarianism (Huntington 1991, Doorenspleet 2000). At the same

time however, shortly before the landmark Egyptian elections, fourth wave president Joseph

Kabila managed to stay in power in Congo after an election that was so deeply flawed and so

obviously rigged that it could hardly be called an exercise in democracy. Nor is Kabila alone

in knowing his way around the toolbox of electoral fraud as the recent elections in Venezuela,

Kenya and Russia have shown.

Clearly, holding elections has become common practice in the vast majority of

countries around the world. i However, while global norms for elections increasingly

converged, global practice shows a widely varying “menu of manipulation” (Norris 2009,

Hyde 2011, Global Commission on Elections, Democracy and Security 2012, Schedler 2002,



relatively late to pick up the topic of electoral integrity. Several scholars did question the

standards used by election observers to evaluate elections (Geisler 1993, Anglin 1998,

Carothers 1997), and others initiated attempts to better define the notion of “free and fair”

elections (Elklit and Svensson 1997, Pastor 1998), however the real upsurge in scholarly

attention seems to have taken place only in recent years (cf. Elklit 1999, Mozaffar and

Schedler 2002, Schedler 2002a, Lehoucq 2003, Elklit and Reynolds 2005, Simpser 2005,

Boda 2005, Katz 2005, Calingaert 2006, Hartlyn and McCoy 2006, Hartlyn et al. 2008, Birch

2007, 2011, Alvarez, Hall and Hyde 2008, Collier 2009, Lopez-Pintor 2010, Darnolf 2011). ii

However, now the field is developing rapidly, with important advancements made in

terms of conceptualizing electoral integrity, mostly based either on democratic theory or

international legal norms for elections (Elklit and Svensson 1997, Elklit 1999, Elklit and

Reynolds 2005, Schedler 2002a, Mozaffar and Schedler 2002, Lindberg 2006, Munck 2006,

2009, and Boda 2005, Katz 2005, Goodwin-Gill 2006, Davis-Roberts and Carroll 2010, EC-

NEEDS 2007, Global Commission on Elections, Democracy and Security 2012, Norris 2013).

Also, several cross-national databases have been developed that measure electoral integrity in

third wave regimes (and sometimes also first and second wave regimes) (Elklit and Reynolds



collected data on electoral integrity for over 800 elections from 1974 to 2009 in 97 third and

fourth wave regimes in Central and Eastern Europe, Former Soviet Republics, Sub ‐Saharan

Africa, South America and Central America (Van Ham 2012).

The paper is set up as follows: in section 2 I shortly discuss the concept of election

integrity and the way it is conceptualized in this paper. In section 3 I outline the elements of

the theoretical framework, discussing the administrative versus political dimension of election

integrity, and identifying the distant, intermediate and proximate causal factors that may

explain variation in election integrity across countries and over time. Section 4 subsequently

discusses the data and methods used, and section 5 presents and evaluates the results of the

empirical analyses. Section 6 concludes with a reflection on the results and suggestions for

further research.

2. Conceptualizing election integrity

What does it mean to “get elections right” (Schedler 2002)? Existing definitions of the

concept range from posit ive definitions such as “free and fair” elections, “clean” elections,



opposition intimidation, etc. During election-day, aspects such as access to polling stations,

vote secrecy and access of election monitors are important. Examples of irregularities are lack

of polling stations in rural areas, vote-buying and intimidation of voters, ballot-box stuffing,

etc. Finally, after election-day important aspects are counting and tabulating the vote,

resolving election related complaints, and publishing the election results. Here, problems that

can occur are: biased counting of votes, not publishing disaggregate results, etc. (Elklit and

Reynolds 2005, Calingaert 2006, Birch 2011). Clearly, elections are complex logistical

operations, and can fail at any step in the process (Norris, forthcoming). Hence ordering the

electoral process by the sequential steps taken before, during and after election day helps to

ensure that all relevant aspects are taken into account, mapping the full “menu of

manipulation” (Schedler 2002, Mozaffar and Schedler 2002).

Secondly, election integrity can be undermined both by intentional manipulation as

well as by administrative incapacity. The key difference between positive and negative

definitions (as listed above) is that positive definitions tend to use a broader conceptualization

of election integrity including both administrative irregularities and intentional manipulation,

while negative conceptualizations focus on a more narrow set of irregularities, emphasizing



What do we know and not know (yet) about electoral integrity? In the emerging

literature on the topic, a number of socio-economic, institutional and cultural explanatory

factors have been identified. Regarding socio-economic factors, lower levels of economic

development and economic inequality appear to increase the frequency of vote-buying,

election violence, and electoral malpractice in general (Von Borzyskowski 2011, Brusco,

Nazareno and Stokes 2004, Bratton 2008). Vote-buying specifically seems to be targeted at

poor and low educated voters (Bratton 2008, Birch 2011). In addition, some authors have

found electoral fraud to be more common in rural areas than in urban areas (Bratton 2008,

Hicken 2007). vi Moreover, the so- called “resource curse” is significantly related to electoral

malpractice as well (Birch 2011). vii Turning to institutional factors, electoral systems and

electoral management bodies have been found to affect election integrity. Birch (2007)

suggests that majoritarian electoral systems are associated with higher levels of electoral

malpractice, while other authors have suggested that candidate-centered electoral systems and

electoral systems with small districts might be associated with higher levels of electoral fraud

(Hicken 2007, Stokes 2007). Hartlyn et al. (2008) find that non-partisan independent Electoral

Management Bodies significantly improve the quality of elections. Apart from socio-

economic and institutional aspects, electoral competition might augment the likelihood of



they fit together in a comprehensive explanatory framework. In the next paragraphs I present

some lines of thought on how to develop such an explanatory framework, exploring (1)

whether elections fail due to lacking capacity or due to intentional manipulation; and (2)

whether elections fail due to long-term structural factors or rather short-term contingent

factors.

Why do elections fail ? The admini str ative and pol it ical dimension of election integr ity

While it is common to think about election integrity in terms of the more blatant forms

of manipulation such as stuffed ballot boxes and manipulated vote counts, elections can and

do fail for reasons of administrative incapacity as well. For example, in a large country with

poor infrastructure, the distribution of ballot boxes and other election material can pose an

enormous challenge (in the 2011 elections in DR Congo, election material was transported by

plane to regional hubs, and then often further transported by bike and foot to the more remote

rural areas). Likewise, in a post-conflict country with a poor civil registry and a large number

of displaced citizens, generating an accurate voter register often involves as much, or more,

logistics than organizing the elections itself (as the census and voter registration exercise in



Building on insights from social-psychology and game theory, I propose that the

perceived need to cheat depends crucially on the stakes of the electoral race. The stakes of the

race are determined by, firstly, what is won through elections, secondly, the degree to which

elections are a zero-sum game, and thirdly, the closeness of the race. Regarding what is won

through elections, not all elections are the same, and the stakes of the electoral race are likely

to be higher if (a) elections give access to concentrated power (i.e. presidential elections in

political systems with highly concentrated executive power), (b) elections give access to

extraordinary rule-making power (i.e. elections after a regime transition, elections for

constituent assembly), and (c) elections give access to state resources (i.e. elections in

countries with small private sectors or abundant natural resources). Hence, the stakes of the

electoral race are higher if what is won through elections is more than ‘just’ public office. For

example, in the early years after a regime transition, winning office provides more extensive

access to rule-making as in times of normal government, as the constitution and other

legislation are still being drafted. This allowed many early fourth wave presidents in for

example Sub-Saharan Africa to expand their presidential powers after the first and second

multi-party elections (Van der Walle 2002), and allowed parties that won the early elections

to have decisive influence on new constitutions, as is illustrated by many cases in Sub-



opposition. Likewise, winner-take-all electoral systems are also likely to increase the zero-

sum nature of the race, and even more so in single-member districts than in multi-member

districts, as the winner(s) of the elections gains access to legislative office to represent the

district constituency as a whole, while the loser(s) remain unrepresented until at least the next

election. Indeed, Birch (2007) has shown that electoral malpractice is higher in countries with

majoritarian electoral systems. Apart from institutional characteristics, the structure of society

increases the stakes of the race as well, if societal divisions are reflected in electoral

competition. For example, in a society that is ethnically or religiously heterogeneous, and in

which such group characteristics are a basis for electoral competition, the trade-off between

winners and losers may be sharper than in a more homogenous society. As Wrong (2009)

describes in the case of Kenya, if winning elections means privileged access to state resources

for one’s own ethnic group at the expense o f other ethnic groups, and losing means being shut

out of such economic opportunities until the next elections, winning the elections becomes of

vital importance. Similar dynamics might occur in the aftermath of a civil war, where

electoral competition often reflects the previously warring factions, and winning or losing the

elections takes on a wholly different meaning, as Almeida (2010) describes in the case of

Angola.



detection are higher, political actors may be less likely to cheat (Birch 2011, Lopez-Pintor

2010). However this may vary in different regime types, depending on the domestic and

international legitimacy costs of cheating. For example, if electoral fraud is obvious, the new

incumbent lacks domestic legitimacy which might lead to civilian protests and even armed

rebellions or attempted coup d’etats. Obvious fraud might also affect international legitimacy,

and possibly lead to political pressure and/or economic sanctions by international

organizations and states. However, the degree to which legitimacy is of any concern to

regimes differs markedly. For example, domestic legitimacy costs are lower in electoral

authoritarian regimes where incumbens have alternative sources of power (i.e. supportive

selectorate, strong internal security apparatus), compensating for lacking electoral legitimacy

towards citizens. x Likewise, international legitimacy costs are likely to be lower for countries

that are economically powerful, dispose of natural resources, or trade predominantly with

non-democratic countries (Levitsky and Way 2010, Kelley 2010, Donno 2010). xi

In addition to the likelihood of detection, and whether or not detected cheating has

legitimacy costs, possibilities to cheat also depend on whether cheating is actively

counteracted and punished. This implies for example an independent electoral management



Building on the explanatory factors identified in the previous paragraph, we can

construct an explanatory model of election integrity that takes the shape of a ‘funnel of

causality’, starti ng with the more distant historical, socio-economic and cultural factors

influencing election integrity, and then shifting to intermediate and proximate causes. Doing

so allows us to differentiate structural causes of election integrity, i.e. distant, non-changing

factors, such as previous experiences with democracy or colonial legacy, from more

immediate causes of election integrity that may change from one election to the next, such as

the presence of election observers. xii Figure 2 below gives an overview of the explanatory

variables.

Figure 2 about here

Considering the distant causes, these are mainly historical and structural socio-

economic factors. The first four variables relate to administrative capacity. If countries have

held democratic elections in their (recent) past, as was the case in Greece and many of the

Latin American transitions, there may be a degree of institutional memory and knowledge



factors refer rather to the political dimension of election integrity. Above we argued that the

stakes of the race are higher in heterogenous societies, especially if societal divisions form a

basis for electoral competition. Ethnic or religious fractionalization, as well as a history of

civil war, might generate such societal divisions, and potentially undermine election integrity.

Another distant causal factor increasing the stakes of the race is the presence of natural

resources, and ownership by the government of these resources, also referred to as the

‘resource curse’.

Turning to intermediate causes, these refer mainly to institutions. Effectiveness of the

state bureaucracy is an indicator of administrative capacity. As noted above, the independence

of the judiciary and electoral management body may be effective both in preventing

intentional manipulation of elections as well as administrative irregularities. Finally,

presidentialism and majoritarian electoral systems are hypothesized to increase the stakes of

the electoral race, and thereby undermine election integrity.

As regards proximate causes of election integrity, the only factors relating to

administrative capacity included here are whether there was a peacekeeping mission present



existing datasets strongly disagreed were coded by the author). This resulted in election

integrity data for over 800 elections from 1974 to 2009 in 97 countries in Central and Eastern

Europe, the Former Soviet Union, Sub ‐Saharan Africa, South America and Central

America. xiv Apart from the data on electoral integrity, data for which no precise indicators

were available in the QoG dataset, such as the structure of the party system, opposition and

incumbent vote and seat shares, legislative electoral systems, and experiences with civil war

were gathered by the author on the basis of information from the Electoral Handbooks on

Africa, South and Central America, Central and East Asia and Europe edited by Dieter

Nohlen (Nohlen 1999, 2001, 2005, 2010), as well as the data handbook by Rose and Munro

(2003) on Central and Eastern Europe, and the Parline Database of the Inter-Parliamentary

Union (IPU).

The dependent variable election integrity is measured as a continuous variable ranging

from 0 indicating low election integrity to 1 indicating high election integrity. xv As for the

independent variables, previous democratic experience with elections is measured as a

dummy variable indicating whether or not countries had held elections that met the minimal

international standards before their third wave transition. Colonial legacy is a dummy variable



separating governmental and mixed from independent EMBs. The independence of the

judiciary was measured using data from the CIRI dataset (and to limit endogeneity with

election integrity this variable was measured 2 years before the elections (Cingranelli and

Richards 2010). xvi Presidentialism was measured by a dummy variable indicating whether the

political system is presidential or parliamentary. The electoral system is a dummy

distinguishing majoritarian from proportional systems in the case of legislative elections and

absolute majority from plurality systems in the case of presidential elections.

Finally, regarding the proximate causes, the presence of peacekeeping operations in

the country at the time of the elections was coded as a dummy based on UN data, and another

dummy indicated whether the country received election assistance for these particular

electons (based on Bjornlund 2004). As regards the type of elections, presidential elections

and constituent assembly elections are both coded as dummies. Closeness was measured by

the difference between the incumbent candidate/party vote/seat share and the largest

opposition candidate/party vote/seat share in the elections preceding the actual elections. xvii

The reason for this is that opposition and incumbent strength are to a certain extent

endogenous to election integrity: in clean elections, opposition parties are more likely to gain



statistically significant (negative) association between for example economic development

and electoral integrity might indicate that economic development is driving electoral integrity,

or the other way around. Over-time analyses within the same country allow for a more precise

assessment of causality: does economic development in the year(s) before an election lead to

higher electoral integrity (correcting for other time-changing independent variables)? The

power of such analyses is that it holds non-changing variables constant (by virtue of the

within- country analyses, ‘fixing’ the country -level variation), allowing for a ‘ceteris paribus’

test of the effect of variables that do change over time. However, the bottleneck of these

analyses is that inferences are based on the number of observations within each country, and

hence the number of time-points (T) needs to be large enough to allow for sufficient degrees

of freedom to draw statistical inferences. Moreover, by holding non-changing variables

constant, the effects of such variables are corrected for, not tested.

Ideally, one would have data on many countries and on many elections within each

country, in order to be able to both model the average level of election integrity for each

country as well as the over-time development in election integrity within countries. Time-

series cross- sectional analyses (TSCS) typically allow for such a combination of ‘between -



countries of the so- called ‘fourth’ wave of democratization 1. An alternative way of modeling

data with this structure as proposed by Beck (2008) is to think of the data as having a

hierarchical structure, with elections (level 1) being nested within countries (level 2). Time-

changing variables, i.e. variables that take on different values with each election, can then be

considered as level 1 variables, and non-changing variables, i.e. variables that are the same for

each election within a country, as level 2 variables. Of course, also with this modeling

approach one needs to test for serial autocorrelation and heteroskedastic errors, but the

approach seems to fit the particularities of the data better.

5. Results

Table 2 shows the results of the empirical analyses. Model 1 shows the results for

distant factors. Clearly, previous experience with organizing elections strengthens election

integrity, and election integrity also appears to gradually improve over time as more

successive elections are being held. Countries with higher levels of economic development

also tend to have higher integrity elections, and colonial legacy appears to have only a weak

and insignificant effect. As regards structural political factors, we hypothesized that the



majoritarian electoral systems for legislative elections reach statistical significance, however

in the more fully specified models 4 and 5, presidentialism also results significant.

Table 2 about here

Model 3 presents the results for proximate factors. Here, factors strengthening

administrative capacity appear more weakly related to election integrity. Election assistance is

in the expected direction but does not reach statistical significance in this reduced model, in

more fully specified models it does turn out significant. Peacekeeping missions do not

significantly affect election integrity and moreover the coefficient is opposite to what was

expected, however this may be due to a selection effect (peacekeeping missions being sent to

countries with weak administrative capacity in the first place). Turning to political factors, the

results seem to confirm the importance of the likelihood of detection in reducing electoral

manipulation. Election integrity is significantly higher if there is independent media and if

there is a larger number of opposition parties and / or candidates participating in the elections.

The type of elections seem to matter less, as neither presidential elections nor constituent

assembly elections are significantly associated with lower election integrity (the effect for

constituent assembly election is however close to one-tailed significance). Finally, model 4



Model 5 demonstrates that these results remain robust even if we include further

controls for election integrity in the previous election and regional dummies. Model 5 also

demonstrates the effect of closeness in the previous elections, measured as the vote/seat

differential of the incumbent and main opposition party or candidate in the previous elections

(the coefficient is close to one-tailed significance). I expected that closeness would increase

the stakes of the electoral race and hence undermine election integrity. However, the negative

coefficient appears to indicate that the closer the race in the previous election was, the higher

election integrity is in the current elections. Hence closeness seems to constrain electoral

manipulation, rather than fuelling it. This finding will require further analysis however.

Another interesting finding in model 5 is the effect of election integrity of the previous

elections on the current elections. Here, contrary to the “self -reinforcing” mechanism of

increasing election integrity as countries hold successive elections, for which we also find

evidence, this result seems to point at the existence of a “lock -in mechanism” of election

integrity, i.e. high election integrity reproduces high election integrity in the next elections

and vice versa. Comparing the coefficients for election experience with previous election

integrity seems to indicate that the effect of previous election integrity is about twice as strong



relative importance of long-term structural versus short term contingent causes of election

integrity. If election integrity is to a large extent driven by past experiences with democracy,

and structural social and economic conditions, there is not much that can be done to

strengthen election integrity. If on the other hand, election integrity is driven by factors that

are more amenable to change, such as the type and functioning of institutions, or the presence

of independent media and observers, this leaves more agency for citizens and organizations

seeking to improve the integrity of their elections.

A preliminary empirical test of these ideas appears to indicate the importance of both

structural and more contingent explanations of election integrity. Structural factors

undermining election integrity include historical factors (history of civil war, lack of previous

democratic experience, previously fraudulent elections), as well as socio-economic factors

(social fractionalization, low economic development, resource curse), and institutional factors

(presidentialism, majoritarian electoral systems, non-independent judiciaries). These factors

shape both the administrative capacity to organize elections as well as the incentives for

political actors to engage in intentional manipulation. Once these structural factors have been

accounted for, more contingent explanations seem to point at the importance of the presence



can different actors be expected to engage in intentional manipulation, and what types and

combinations of irregularities are more likely under what conditions. Such disaggregate and

conditional analyses would not only advance academic understanding of this topic, but would

also generate more useful advice for practitioners, both in terms of mapping the ‘menu of

manipulation’ as well as gathering more concrete insights in how to strengthen election

integrity.



Figures and tables

Figure 1. Election integrity: differences in kind and degree



Table 2. Explaining election integrity in third wave regimes 1974-2009Model 1 Model 2 Model 3 Model

4 Model 5Distant Intermediate Proximate Full model Full w controls

Distant factors Administrative dimension

Prev. democratic experience (0-1) 0.154** 0.096*0.060+

(0.050) (0.039) (0.036)Colonial legacy – British (0-1) 0.030 0.045 0.013

(0.071) (0.060) (0.060)Election sequence (1-20) 0.007* 0.007*

0.006+(0.003) (0.003) (0.003)

Economic development 1.36e -5** 8.29e -6+ 3.83e -6**GDP/cap t-2yrs (118-26193) (5.06e -6) (4.68e -6) (4.53e -6)

Political dimension

Resource curse t-2yrs (0 – 99.70) -0.001** -0.001** -0.001**

(0.001) (0.000) (0.000)

Ethnic fractionalization (0 – 0.93) -0.155 -0.037 0.003(0.116) (0.095) (0.089)

Religion fractionalization (0.01-0.86) -0.085 -0.198* -0.167+

(0.125) (0.100) (0.087)



Political dimension

Presidential elections (0-1) 0.018 -0.002 -0.008(0.017) (0.019) (0.018)

Constituent assembly elections (0-1) -0.021 0.023 -0.094(0.054) (0.053) (0.065)

Independence media t-2yrs (0-1) 0.222*** 0.185*** 0.115***

(0.030) (0.030) (0.031)Number of opposition candidates 0.008*** 0.007*** 0.003+(0-28) (0.002) (0.002) (0.002)

Exports to high income economies 0.002*** 0.001**0.001

t-2yrs (0 - 99.75) (0.000) (0.000)(0.001) Closeness in previous election (0-100)-0.001

(0.000)Lagged election integrity (0-1) 0.246***

(0.035)Region Former Soviet Republics (0-1) -

0.174**(0.065)

Constant 0.604** 0.736*** 0.345***0.516*** 0.519***

(0.081) (0.057) (0.040) (0.093)(0.110)

N level 1 (elections) 837 837 837 837 667N level 2 (countries) 90 90 90 90 89Variance explained level 1 2% 2% 7% 9% 9%



25

Figure 2. Explaining election integrity: distant, intermediate, and proximate causal factors



26

Appendix – Conceptualizing election quality a Authors I. Background concepts II. Systematized concept III. Indicators

Conceptualizations of democracy Attributes Components of attributesSchumpeter (1975) - Political leaders are elected by citizens

- Competition between political leaders for citizens’ votes 1. Elected officials

2. Periodicelections

3. Participation

4. Competition

1.1. Object of choice: mostpowerful collective decision-makers selected throughelections.1.2. Consequences ofchoice:election results areirreversible & elected officialshave control over policy-making.

2.1. Elections held at regularintervals.

3.1. De jure right to vote3.2. De facto conditions for the

use of the right to vote

Continued on next page.

- Elected legislature- Elected executive (directly or indirectly)

- Winners of elections assume office- Elected officials have discretion over

policy- making, influence of “reservedpower domains” is limited.

- Elections held periodically, according toconstitution/electoral law.

- De jure universal & equal suffrage- De facto possibility to participate:

- Possibility to form preferences:- Freedom of expression- Alternative sources of information- Voter education

- Possibility to cast a vote:- Freedom of movement- Freedom from

intimidation/coercion- Voter registration (register,

registration requirements)- Polling station access- Secret balloting

- Votes cast are counted:- Accurate vote counting

Dahl(1989)

- Elected officials- Free and fair elections- Inclusive suffrage- Right to run for office- Freedom of expression,- Alternative information- Associational autonomy

Huntington (1991) - Elections for most powerful decision-makers- Fair and honest elections- Periodic elections- Universal suffrage- Free competition

Przeworski et al.(1996)

- Ex ante uncertainty- Ex post irreversibility- Repeated elections

Schmitter and Karl(1991)

Dahl’s 7 criteria plus 2: - Absence of reserved power domains- State sovereigntyConceptualizations of election quality (concept-based)

Hermet et al.(1978)

- Freedom voters- Competition candidates- Effects elections on policies

Elklit &Svensson(1997)

- Free elections (before, during & after polling day)- Fair elections (before, during & after polling day)

O’Donnell (2001) - Competitive, Free, Egalitarian, Decisive, InclusiveLindberg(2006a)

- Participation- Competition- Legitimacy

Munck(2009)

- Inclusive elections- Clean elections

- Competitive elections- Elective public offices

Lopez-Pintor(2010)

- Political framework: periodic elections, free expression of the will of thepeople- Free and fair elections: right of the people to vote, guarantees for secretballot

and genuine elections



27

- Right to franchise: right of universal & equal suffrage- Right to participate: right to stand for election & assume office

Authors I. Background concepts II. Systematized concepts III. IndicatorsConceptualizations of election quality (process-based) Attributes Components of attributes

Schedler (2002a) - Object of choice

- Range of choice- Formation of preferences- Agents of choice- Expression of preferences- Aggregation of preferences- Consequences of choice

4. Competition

5. Freeelections

6. Fair elections

4.1. De jure right to run for office4.2. De facto conditions for the useof

the right to run for office

5.1. Freedom to form preferencesand

vote free of coercion (forvoters)5.2. Freedom to run in electionsand

compete for popular supportfree

of coercion (for parties/candidates)

6.1. “Unbiased application of rules”of

electoral competition &electoral

governance before, duringand

after elections.

- De jure multi-party/candidate elections- De facto possibility to compete:- Registration requirements

- Ballots accurate- Freedom from

intimidation/coercion- Freedom of movement,

expression,assembly, association

- Alternative sources ofinformation /

media coverage- Use of state resources

- Campaign finance regulation

- Campaign activities- Access to dispute adjudication

- Freedom of movement, expression,information.

- Freedom of movement, expression,information, assembly, and association.

- Impartial Electoral Management Body- Implementation according to electorallaw:

- Before elections:- Constituency & polling district

demarcation- Ballot paper design- Registration of voters and

parties- Campaign regulation

- During elections:- Polling

Mozaffar &Schedler (2002)

- Rule-making (choosing rules of electoral game)- Rule application (organizing electoral game)- Rule adjudication (certifying results & solving disputes)

Elklit & Reynolds(2005a)

- Legal framework- Electoral management- Constituency& polling district demarcation- Voter education- Voter registration

- Ballot paper design & party/candidate registration- Campaign regulation- Polling- Counting and tabulating the vote- Resolving election related complaints, v erication & certication- Post-election procedures

Birch(2008, 2011)

- Manipulation election rules- Manipulation voting process / election administration- Manipulation of vote choiceInternational legal conventions b

UniversalDeclaration ofHuman Rights(UDHR 1948, art.21, 1 & 3)

- Everyone has the right to take part in the government of his country, directlyor

through freely chosen representatives;- The will of the people shall be the basis of the authority of government, thiswill

shall be expressed in: periodic and genuine elections, which shall be byuniversal

and equal suffrage, and shall be held by secret vote or by equivalent freevoting

proceduresInternational - Every citizen shall have the right and the o pportunity, […] without



28

Covenant on Civiland Political Rights(ICCPR 1966,art.25a&b)

unreasonablerestrictions: to take part in the conduct of public affairs, directly or through

freelychosen representatives; to vote and to be elected at genuine periodic

elections, whichshall be by universal and equal suffrage and shall be held by secret ballot,guaranteeing the free expression of the will of the electors.

- After elections:- Counting and tabulating the

vote- Resolving election complaints- Post-election procedures

a. The studies by Anglin (1998), Pastor (1999), Van de Walle (2003), Hartlyn et al (2008), Donno (2010), Kelley and Kiril (2010) and Hyde and Marinov (2010) are not included herebecause they do not provide an elaborate conceptualization of election quality, but rather give an operational definition based on how wide-spread, intentional and consequentialirregularities are, or, in the case of Pastor, whether all major parties accept the results. b. For reasons of space, I only report the conditions mentioned for elections in the UDHR andICCPR. Other instruments are described in the text.Notes

i Over 85% of the world’s states now select their national leaders through elec tions (Cross-National Time-Series Data Archive).ii This is interesting since, as Lehoucq describes, electoral fraud was by no means uncommon in the early years of first wave democracies. In his excellent review, Lehoucqmentions historical accounts of electoral fraud in the US, Latin America and Europe at the end of the 19th and early 20th century. Accounts range from voter intimidation,vote buying and stuffing the voter registry with ineligible voters, to pressure from landlords, employers and the church, in countries ranging from the UK and Ireland, toGermany and Spain (Lehoucq 2003). The fact that these countries by now have had such long experiences of democratic rule, in which electoral processes have come to beseen as straightforward administrative procedures, probably has contributed to the “illusion of transparency” and the concomittant neglect of the topic of electoral governancein research on the third wave regimes (Guennifey 1993, Mozaffar and Schedler 2002, Lehoucq 2003, Pastor 1999).iii Most inferential work has been done in the form of case studies, with Lehoucq and Molina’s (2002) study of Costa Rica, Eisenstadt’s (2004) and Magaloni’s (2006) workon Mexico, and Nyblade and Reed Chuo’s (2008) research on Japan as excellent examples (cf. Leh oucq 2003 and Birch 2011 for a more extensive overview of case studieson electoral fraud). Also, more extensive research has been done on specific types of electoral fraud, most notably on vote-buying and to a lesser extent on election violence.Vote buying has been analyzed in a mix of country-studies, qualitative analyses and surveys, as well as experiments (cf. Schaffer 2007; Brusco, Nazareno and Stokes 2004;Calvo and Murillo 2007; Vicente 2010; Bratton 2008; and a forthcoming volume by Stokes et al.). As regards election violence, the literature is expanding rapidly, howevermost research seems to have concentrated on Sub-Saharan Africa, using both survey and experimental methods to gauge the extent and causes of election violence (Bratton2008, Collier and Vicente 2011, Von Borzyskowski 2011).iv As Mair points out: “However understandable it is to avoid the apodictic free and fair, it is even more problematic to specul ate on voter intentions and election results. Howelse can the voter’s intention be registered than by an election, and how can this take place correctly other than in a free and fair election?” (as quoted in Bjornlund). Mair,

International Election Observation ; Bjornlund, Beyond Free and Fair , 119.v Remarkably however, as the Appendix shows, despite the disagreement about election integrity at the conceptual level, when it comes down to measurement scholarsidentify very similar indicators. These indicators mostly match the stages of the electoral process described above, i.e. identifying irregularities that can occur before, duringand after election day,vi However, Lehoucq (2007) proposes that different types of electoral fraud might be common in rural versus urban areas, as voter intimidation might be relatively easier inrural areas were traditional patron-client relationships are more common, while in urban areas politicians might need to resort to other methods of manipulation such as

buying votes; and Molina and Lehoucq (1999) also show in the case of Costa Rica that such regional patterns may shift over time as the consequence of changes in the natureof political competition.vii Of course, the resource curse not only provides incumbents with funds to finance electoral manipulation, it also gives incumbents more leeway to disregard international

pressure for holding high integrity elections.



29

viii As Bjornlund (2004) describes, the degree of involvement of the international community in elections varies strongly, ranging from providing limited technical training andadvice, to large-scale assistance in organizing the elections. Examples of the latter are the first elections in Bosnia-Herzegovina after the Dayton Accords, the elections inAfghanistan and the first elections in DR Congo. As far as I am aware, thusfar little comparative research exists evaluating the effectiveness of international assistance toelections for strengthening election integrity, a topic that will be taken up in the third part of the Electoral Integrity Project.ix However, increased competition may also limit possibilities to engage in fraud as parties monitor each other. Moreover, the effect of closeness of elections on electionintegrity may only hold in electoral democracies, as in electoral authoritarian regimes successive fraudulent elections have generated such a non-level playing field that close

elections are unlikely to occur, and yet election integrity continues to be undermined (presumably to maintain the status quo ).x As Simpser (2005) has suggested, in these regimes overt fraud may in fact be a use ful ‘signalling’ tool for incumbents to demonstrate their strength to voters and opposition,deterring opposition participation – and votes- in future elections.xi For example, for many countries in the Commonwealth of Independent States, that have important trade and aid relationships predominantly with Russia, holding flawedelections might not be that costly, as clean elections are not a condition for these relationships to thrive. The increasing economic influence of China in Sub-Saharan Africamight have a similar effect on third wave regimes there (Levitsky and Way 2010). On the other hand, for countries in Central and Eastern Europe, holding clean elections wasone of the many conditions that the European Union required to be met for accession, presu mably increasing legitimacy costs of fraud.xii This also allows for the evaluation of the relative importance of these different factors in explaining election integrity, a s well as the use of different theoretical ‘lenses’ tolook at election integrity. For example, distant causal factors allow us to evaluate more historical and structuralist explanations, intermediate causal factors relate toinstitutional explanations, and proximate causal factors allow for rational choice explanations.xiii The debate is still ongoing whether these differences are due to natural conditions in former colonies that either stimulated or discouraged settlement or rather due to thespecific characteristics of the colonizing powers. As such, it has often been argued that former British colonies tend to have achieved higher levels of democratic qualitycompared to former French, Spanish and Portuguese colonies, due to the development of democratic political institutions, the traditions of British common law, and thedevelopment of a ‘democratic culture’ under British rule.xiv Since election integrity measures de facto variation in election integrity for elections that meet, at a minimum, de jure international norms for elections, I first selected allelections that met the minimal international criteria held in third and fourth wave regimes between 1974 and 2009. Conditions for de jure democratic elections are met ifelections are held for both legislative and executive (directly or indirectly), if elections are irreversible, and if elections allow for de jure multi-party or multi-candidatecompetition and de jure universal suffrage. The first criterion implies that both the executive and the legislative should be selected through elections, either directly orindirectly. This means that elections held under military governments holding only parliamentary elections, such as Brazil between 1964 and 1982, or elections held undermonarchy governments holding only parliamentary elections, such as Swaziland, are excluded. The second criterion means that elections immediately followed by a coupd’etat and/or civil war that prevent the elected government from taking power are excluded. Elections held under conditions o f civil war where nevertheless the electedgovernment was able to assume po wer are included, as are elections held just after a coup d’etat. The third criterion implies that elections should be held u nder de jure multi-

party conditions, i.e. it should be legally possible for multiple parties or candidates to contest elections. This means that elections held under a constitution forbidding theformation of political parties, such as Sudan between 1989 and 1998, or allowing for only one state party, like Malawi between 1966 and 1993, are excluded. Finally, thefourth criterion implies that elections should allow de jure universal suffrage, i.e. it should be legally possible for all adult citizens to participate in elections. This means thatelections where suffrage was restricted on the basis of gender, race, income or literacy were excluded. The absence of a de jure secret ballot is also considered as a suffragerestriction, as public voting limits the possibilities for opposition supporters to cast a free vote. Elections where secret ballots were not legally required were hence alsoexcluded. Elections with suffrage restrictions for non-adults, detainees, and specific occupations (such as police or army officials) were included. Note that these criteria aremeant to capture the full range of variation in election integrity, thus mere ly specifying the minimal conditions that need to be met for elections to be “de jure democratic”.This means that elections held under conditions of de facto limitations on party competition and voter participation are included, as long as the de jure conditions specifiedabove hold. (cf. Van Ham 2012 for a full description of this dataset).



30

xv The existing data were merged in a single dataset and combined after extensive validity tests. I subsequently coded missing elections and recoded elections on which therewas strong disagreement. Elections were coded based on election observation reports, academic articles, human rights reports, and news media reports. For these scores I useda hierarchy of sources, with the academic and election observation reports with well-established methodologies and professional observer teams given priority over reports innews media. I coded elections along a 3-point ordinal scale, based exclusively on the frequency and spread of irregularities, rather than intentionality or perceivedconsequences for the outcome. Elections were coded using the following categories: 0 = widespread irregularities, whether intentional or not, that seriously affected theconduct of the elections in a large part of the country; 0.5 = considerably frequent irregularities, but not so widespread that the election is outright rejected by observers; 1 =

no irregularities or some isolated, non-frequent irregularities. For more information on the merging procedure, validity tests and the coding of missing and disagreed electionssee Van Ham (2012). For the purpose of the analyses presented here, I use a continuous variable of election integrity that takes the average of the election integrity scores forelections on which there was low disagreement, and my own scores for elections that had missing data and high disagreement. I reran the analyses using a dichotomousdependent variable that recodes elections with widespread and considerably frequent irregularities in one category of “flawed” elections and compared to “clean” elections(i.e. elections with no or isolated irregularities). The results are essentially similar to the ones presented here.xvi Note that for a number of variables, the values tested are indicated as lagged values (t-2), indicating that these variables will be measured 2 years before the elections. Thisis to prevent endogeneity or reversed causality. For example, an increase in international aid in the year of the election might be the result of a clean election, rather than acause of it. Taking the level of international aid some time before the elections as an indicator helps to diminish such reverse causality, even if not completely solving it. Thetime lag of 2 years is chosen to also reduce the likelihood of potential anticipation effects. For example, if elections will take place in the following year, and are expected to

be fraudulent or problematic, foreign investors might already retreat in anticipation and hence international trade might drop before elections, as a consequence of theanticipated low integrity of the electoral process, rather than causing such lower integrity (and in fact I expect trade to have a positive effect). Finally, a time lag of 2 yearsalso reduces the likelihood of temporal overlap between the measurement of independent variables and the electoral process, which, counting the campaign period prior toelections as well as adjudication of complaints and the final publication of results, can take up to anywhere between 3 to 6 months, and sometimes even longer (Lindberg2006).xvii Note that closeness is measured as the vote differential for presidential elections, but due to a large proportion of missing data for vote shares in legislative elections, as theseat differential for legislative elections.xviii Of course, many things change between elections, among which the popularity of candidates and parties, so this is a very imperfect measure of closeness. However,lacking comparative poll data that would allow us to assess parties’ and candidates’ popularity during t he election campaign, this seems the best alternative.xix Data on EMB independence was collected by IDEA in 2006, when many countries that had had problems with election integrity had adopted formally independent EMBsin reaction to those problems, blurring the effects of this variable.xx Note that bureaucratic effectiveness, presidential electoral system type and the presence of international observers had to be dropped from the full model due to multi-collinearity. Bureaucratic effectiveness is highly collinear with economic development and bureaucratic effectiveness were highly collinear, indicating that economicdevelopment is indeed a good proxy for administrative capacity. Economic development was left in the full model as it is more precisely measured. Presidential electoralsystem type was collinear with presidential elections, since the formed was already insignificant in a the reduced model 2 I left it out. Finally, the presence of internationalobservers is collinear with election assistance, however, while election assistance remains significant in the full model, the effect of international observers is insignificant,and hence the latter was dropped.



31

1_8 Carolien Van Ham Why Do Elections Fail _2

Documents

Transcript of 1_8 Carolien Van Ham Why Do Elections Fail _2