The Conditional Effect of Electoral Systems and Intraparty Candidate Selection Processes on Parties' Behavior

• Author: Yael Shomer
• DOI: http://doi.org/10.1111/lsq.12141
• Date: 01 February 2017
• Published date: 01 February 2017

YAEL SHOMER

Tel-Aviv University

The Conditional Effect of Electoral

Systems and Intraparty Candidate

Selection Processes on

Parties’ Behavior

How do electoral systems and intraparty candidate selection procedures affect

party unity? In this article, I distinguish theoretically and empirically between electoral

systems and candidate selections and argue that the influence of selection processes on

parties’ behavior is conditional on electoral system (and vice versa). Measuring parties’

unity using Rice and weighted Rice scores, and applying hierarchical models to a new

data set of 249 parties in 24 countries, I find support for the claim that the influence of

selection processes on behavior is greater under electoral systems that encourage perso-

nal vote-seeking incentives than under electoral systems that encourage party

centeredness.

Maintaining a unif‌ied party voting record is an important objective

for both party leaders and rank-and-f‌ile members. Unif‌ied parties aid pol-

icy implementation and improve stability in both parliamentary and

presidential systems. Without party unity, in presidential systems, the

president’s base of legislative support becomes less stable and predict-

able, whereas in parliamentary systems government survival might be

threatened. Additionally, a unif‌ied party record helps enhance the party’s

electoral performance, as it creates a clear brand name with which voters

identify. Lastly, within the legislative arena, a unif‌ied party record

reduces transaction costs in log rolling, increases policy-making capabil-

ities, decreases legislative unpredictability, and amplif‌ies legislative

eff‌iciency. Indeed, disciplined-unif‌ied parties are regarded as one of the

pillars of the responsible party government doctrine (Ranney 1954).

What explains the degree to which parties behave as a unif‌ied

actor? Scholars have long argued that electoral systems and intraparty

candidate selection processes (i.e., the manner by which a party decides

who can represent it on its ballot) shape the balance legislators strike

between emphasizing their unique personal reputation and the collective

LEGISLATIVE STUDIES QUARTERLY, 42, 1, February 2017 63

DOI: 10.1111/lsq.12141

V

C2016 Washington University in St. Louis

reputation of their parties. As legislators seek to maximize their probabil-

ity of getting (re)-elected, they adhere to the incentives and constraints

electoral systems and candidate selection procedures produce and

choose to either vote with their copartisans and sustain a unif‌ied party

voting record or vote against them weakening their party’s unity (Carey

and Shugart 1995; Hazan 2000).

This article seeks to examine how extralegislative institutional

arrangements—i.e., electoral systems and intraparty candidate selection

processes—affect parties’ behavior and legislators’ tendency to maintain

a unif‌ied party record. To date, most scholars consider the effect of elec-

toral systems and selection processes separately. It has been argued that

the more permissive the electoral system is, the less unif‌ied parties will

be (see Andr

e, Depauw, and Shugart 2014). Similarly, inclusive and

decentralized candidate selection processes have been hypothesized to

induce a dis-unif‌ied party voting record (Bowler, Farrell, and Katz 1999;

Hazan 2000; Hazan and Rahat 2006). However, numerous examples

might cast doubt on this simplistic depiction.

The Brazilian left-wing parties’ (e.g., the Worker’s Party) high lev-

els of unity present a puzzle. Brazil’s open-list PR system should have,

according to the electoral systems’ separate-effect hypothesis, induced

undisciplined and dis-unif‌ied parties. However, whereas some scholars

(e.g., Ames 1995) characterize Brazilian parties as undisciplined, others

(e.g., Figueiredo and Limongi 2000) f‌ind highly unif‌ied parties, espe-

cially within the left wing (Mainwaring and Lin

an 1997) (see Amorim

Neto [2002] for discussion about the debate). Why would parties under

the Brazilian permissive electoral system enjoy high levels of unity?

Mainwaring and Lin

an argue that restrictive selection procedures might

solve this puzzle and claim that even under open-list PR systems

“[i]f central party leaders control the selection process, individual

legislators ...are more likely to toe the party line” (1997, 476).

Similarly, despite variation in selection procedures used in

Argentina (De-Luca, Jones, and Tula 2002) and despite expectations for

provincial effects on parties’ behavior, Jones and Hwang f‌ind that “no

provincial effects are detected” (2005, 273). They explain the seemingly

unif‌ied parties’ and legislators’ behavior using the negative agenda-

control thesis presented by the American cartel theory (Cox and McCub-

bins 2005). However, I suggest a different explanation—a conditional

theory: given the closed-list PR electoral system Argentina uses, we

would not expect local primaries and provincial control to affect legisla-

tors’ behavior and parties’ unity levels dramatically.

As a last motivating example I will present Israel. Israel’s parties

used over the years numerous forms of candidate selection procedures

64 Yael Shomer

from the most exclusive and centralized manner, in which one leader (or

rabbi) selects the list, to an inclusive selection via party primaries. If the

separate effect of selection processes on parties’ unity levels is correct,

one should have observed that those parties that select via primaries

exhibited lower levels of unity compared to parties selecting in a

centralized restrictive way. However, this conclusion is not supported by

data (Rahat 2007).

1

To solve these puzzles and explain why other empirical tests of the

separate effect of elections and selection on parties’ behavior yield

inconclusive support, I suggest a theory about the combined conditional

effect of elections and selections. Both electoral systems and candidate

selection processes have the theoretical separate ability to induce a uni-

f‌ied collective voting record. Thus, if the electoral system incentivizes

legislators to toe their party’s line, the effect of the candidate selection

procedure, and especially democratized processes, will be muted. This is

because the electoral system will ensure a unif‌ied party record regardless

of the type of selection processes a party uses. Under these conditions,

therefore, party leaders may afford democratizing candidate selection

processes, knowing their negative effect on party unity will be minimal.

By the same token, if the manner by which legislators are selected by

their parties guarantees they maintain a unif‌ied voting record, then the

effect of electoral systems on behavior would be minimal. Party leaders,

however, may use restrictive selection processes to attenuate the effect

of electoral systems that encourage legislators to break the party line,

and ensure—even under permissive electoral systems—a unif‌ied voting

record. When selection processes are centralized and restrictive, party

leaders may either apriorichoose a party slate that holds similar ideo-

logical preferences, or they may threaten to punish reluctant legislators

by not renominating them in the next election cycle. Hence, restrictive

candidate selection processes may overcome the personal vote-seeking

incentives produced by an electoral system. Therefore, I claim that the

inf‌luence of selection processes on parties’ unity levels is conditional on

the type of electoral system a country has and similarly, that the effect of

electoral systems on a party’s voting record is conditioned by the manner

by which the party selects its candidates.

After presenting the current state of the literature, which motivates

the necessity for the elaborated conditional hypothesis, I test the condi-

tional effect of electoral systems and selection processes on party unity

using a varying-intercept varying-slope hierarchical model, where par-

ties are nested within country-sessions. This multilevel structure allows

me to account for the fact that selection procedures and electoral systems

are distinct institutions that operate usually at distinct levels: Candidate

65Electoral Systems and Selection Processes