Presidentialization of political leadership, Spanish Style: its
effects on the reselection of losing top candidates in the
Spanish Comunidades Autónomas
Javier Astudillo
[email protected]
ECPR General Conference
Glasgow
September 2014
Draft version
Abstract
In parliamentary democracies political parties constitute one of the most important
resources that political leaders have to implement their policy preferences. Yet, parties
have also played a vital role in their prior selection as “top candidates”, and in their
replacement if they lost their elections. Controlling party machines seemed to be a key
asset for starting political leaders to have a second chance after failing in their first
contest. A recent proposal suggests, however, that the role of parties in general and their
machines in particular has changed considerably in “old parliamentary democracies”,
while the extent of changes remains unclear among “new” ones. What this means for
the fate of losing candidates remains a fairly understudied topic. This study is a first
step in its analysis. Based on a dataset built by the author, we have analyzed the
candidates of the main national parties for the premiership of the 17 Spanish
Comunidades Autónomas that lost their first regional contest. Our statistical analysis
shows that, after controlling for other factors, being party chair of the regional branch of
one of the national parties significantly increases chances of “reselection”. In Spain, our
example of a “new” democracy, party machines still play a fundamental role for starting
an enduring political leadership.
1
Introduction1
Under what conditions do politicians repeat as “top candidates” in parliamentary
democracies after losing their first election? What assets do losing candidates need to
control if they want to repeat the experience of throwing their hat into the electoral ring?
In an alleged time of party decline, does controlling their machines still protect nascent
political leaders in the face of electoral setbacks?
Political leaders need to attain the highest executive offices to lead. In most democracies
this means that parties must have previously selected them as their candidates for those
offices. Furthermore, if they lose the elections, and they wish to run again, they must be
reselected as candidates. Party intervention in their reselection or replacement is thus a
vital part of the game.
In western parliamentary politics, this remains an understudied topic. So far, studies
about reselection have focused on incumbents (Wayne and Baker 1986; Fisher III and
Herrick 2002). A few scholars have analyzed losers’ fate (Taylor and Boatright 2005;
Carsey and Berry 2014), but they have exclusively focused on US legislative
candidates, where party machines have tended to play a lesser role in candidate
recruitment. With respect to parliamentary democracies, we only have the recent
proposition that the new criteria used by parties for the recruitment of political leaders
can make them less “durable” in the face of electoral defeats (Webb, Poguntke and
Kolodny 2011). It is not clear, however, if in some parties, especially those in “new
democracies”, control of the party machine can still protect political leaders from a bad
start in the electoral arena.
This paper attempts to assess the current salience of chairing the so-called “party central
office” - the extra-parliamentary organization which is the governing body of the party
machine (Katz and Mair 1993) on first-time losers’ chances of renomination as top
candidates at the regional level in Spain. By doing so, it seeks to evaluate to what extent
we can generalize the alleged transformation of the role of parties in top candidate
1
This study has been possible thanks to a research grant of the Instituto Carlos III-Juan March de
Ciencias Sociales.
2
recruitment and replacement processes in “old parliamentary democracies” to “new”
ones, and at the same time to contribute to the literature about the power sources of an
enduring political leadership.
In the next section we discuss the arguments about the loss of salience of the party
organization on politicians’ selection and replacement as top candidates, and why we
focus on the Spanish Comunidades Autónomas. In section three we explore the meaning
of “losing” in parliamentary democracies. In section four we present the basic
descriptive information about the top regional candidates in Spain. In section five we
present our set of hypotheses to explain losers’ fate. Following that, we offer the results
of our empirical analysis. The final section concludes.
The alleged declining role of political parties in the recruitment and replacement
process of political leaders
The literature about “political leadership” tends to focus on the heads of the executives,
presidents or prime ministers. These studies have shown that political parties are a
fundamental factor in political leaders’ capacity to carry out their policy preferences
(Elgie 1995). Those heads of governments who do not count on the collaboration of
their legislatures cannot implement their policies, and in parliamentary systems their
own survival is also jeopardized. Parties have traditionally been one of the most
important instruments for obtaining this collaboration (Cheibub 2009).
In a democracy these offices are directly or indirectly “elected”. This means that
prospective political leaders must not only win their elections, but in most countries
they must have previously been selected as candidates by the political parties. While
many aspirants are never selected, some of those who are selected are only given one
chance, yet others are nominated again after failing their first contest. The role that
political parties play in executive candidate selection and replacement therefore seems
to be a fundamental element for explaining the beginning and duration of a top
leadership career, just as important as subsequent party support in implementing
leaders’ preferred policies or their capacity to obtain citizens’ approval. However, its
3
study needs further research, especially in the case of parliamentary democracies
(Helms 2005).
The multiple roles that parties play in the recruitment of legislative candidates in
parliamentary countries are well established (Gallagher Marsh 1988; Norris1997; Hazan
and Rahat 2010). We know that in these countries political parties have played a vital
role not only because their machines controlled until recently the selection of
candidates, but because having a long record of voluntary work for the party
organization, or “party service”, was a requisite for recruitment (McAllister 1997).
Through this process, elected officials were socialized into the party’s culture and
practices, reducing “the need to utilize the available disciplinary measures (…) in order
to keep its elected representatives in line” (Hazan and Rahat 2010:148). In addition, the
loyalty of the incumbent legislators to the organization was a key element to be
reselected as candidates (Wessels 1997).
Less attention has been paid, however, to the recruitment of candidates for the head of
the executive in parliamentary systems: the prime ministers 2. It is true that there is a
developing literature about the democratization of the process of selecting political
leaders (Scarrow, Webb and Farrell 2002). Nevertheless we know much less about the
criteria parties use to recruit them, or about their role in the reselection or replacement
of those experiencing electoral defeats.
The few studies approaching this topic suggest that recently, at least in the old
parliamentary democracies, this role has experienced profound transformation as a
consequence of the “presidentialization” of parliamentary elections as well as the
increasing salience of the so-called “party in public office” -meaning both the
representative of the party in parliament and government (Katz and Mair 1993). If, as
we will see, the first development points to a loss in the salience of having a strong
“party service” for the recruitment of political leaders, with consequences for their
2
The literature about “party leadership” summarily tells us that political parties tend to select their “party
leaders” as their “top candidates”, - also known as “prime ministerial contenders/candidates” or
“electoral leaders”- so they become, if elected, prime ministers. The problem is that, either this is a
tautology (all “top candidates” are regarded as “party leaders”), or this is not always the case.
4
durability, the second potentially has even deeper implications. It would involve a loss
of relevance that controlling the party machine has for political leaders experiencing
electoral setbacks. However, it is far from clear to what extent we can generalize the
transformation of parties’ role for losing leaders to all kind of political parties and,
above all, to all parliamentary democracies.
But before entering into the argument, we must first explain who we are referring to by
“executive candidates” in this kind of democracy, because in their parliamentary
elections, all “candidates” are de jure “legislative” candidates. In fact, in parliamentary
democracies, by definition, there are no proper “executive candidates” who run in any
popular election. Prime ministers are selected by legislatures (Lijphart 2012).
Nevertheless most parties identify now their “top candidate” (“spitzenkandidat/in” in
German) in the electoral campaigns. They are usually their informal candidates for
being prime minister or chancellor3.
Reaching the point of having “prime ministerial candidates” in parliamentary elections
has been a long process. Firstly, as Cox’s study of the nineteenth-century British
parliament shows (Cox 1988), the legislative elections were transformed, in practice,
from a selection of individual legislators, on a rather non-party basis, to the selection of
a collective body, the government, on increasingly party lines. Later, and especially in
those countries where prime ministers were mainly recruited among the members of
parliament, political parties started to explicitly identify in their electoral campaigns to
whom, among all their legislative candidates, they wanted to assign the position of
prime minister should they win the elections. As a result, these “top candidates” became
their de facto “prime ministerial candidates”. This process was reinforced because the
mass media increasingly focused on these “executive” candidates (Farrell and Webb
2002). Prime-ministers who had been presented by their parties as “top candidates” in
the parliamentary elections started to talk about having a “popular mandate”, just as
presidents in presidential democracies have. At the same time those prime-ministers
duly elected by their parliaments, but who had not been previously presented as parties’
3
See, for example, the German news about the main parties’ candidates for chancellor for the 2013
Bundestag elections in http://www.bundestagswahl-bw.de/spitzenkandidaten_2013.html (retrieved on
June 3th, 2014).
5
top candidates, started to be regarded by political pundits and journalists as
“unelected”4. Parliamentary elections had finally become “presidentialized” (Poguntke
and Webb 2005).
The interesting thing is that according to Webb, Poguntke and Kolodny (2011) this
process of “presidentialization” of elections in parliamentary countries ought to not only
have created this informal figure of being a “prime ministerial candidate”, but at the
same time it ought to have modified the criteria that political parties use to recruit
political leaders, with the consequence, however, of weakening the resilience of those
leaders whose parties obtained unsatisfactory electoral results under their leadership.
In the classic mass party of parliamentary democracies before the “presidentialization of
politics” the last step of a politician’s successful internal party career was obtaining the
highest office of the extra-parliamentary organization, the party chair, thanks to a
meticulous process of building a power base within the organization (Ware 1996). They
were also recruited as the party’s aspirant for the prime minister office, although they
had not been necessarily presented as such in the parliamentary elections5. Political
leaders were just as much party creatures as any ordinary backbencher.
It is true that under the old recruitment criteria of political leaders parties risked
supporting dull aspirants for the prime minister office, expert only in winning
negotiations in smoke-filled rooms. At a time when people voted for ideological or
party attachment reasons, or parties were willing to sacrifice votes for ideological
purity, this was not such a problem. However, “mainstream” parties could no longer
afford such recruitment criteria when winning elections became their paramount goal
and the personal qualities of their now “top candidates” started to be highly valued by
4
See for example BBC news “Italians keen to gamble on unelected Matteo Renzi”
http://www.bbc.com/news/world-europe-26218177 (retrieved on May 30th, 2014).
5
In the Westminster parliamentary countries, the “party chair” usually is not the “leader of the party”.
This role is the hands of the leader of the parliamentary party, and they are the ones who become prime
ministers. The arguments about the decline role of the extra-parliamentary organization do not apply in
these cases.
6
citizens. According to Webb et al now parties seek to select top candidates that they
believe will give them the highest chances of winning the elections (Webb, Poguntke
and Kolodny 2011). This also means for them that a long internal party career is no
longer a requisite to be selected6.
Yet these changes of political leaders’ recruitment criteria, reflected in their declining
partisan features, will already have effects on political leaders’ durability in the face of
electoral defeats. These scholars suggest that these leaders are “less likely to survive
electoral defeats than their precursors” (Webb, Poguntke and Kolodny 2011:9). They
offer a double reason for this loss of resilience. It is not only that if they lose, they have
belied their parties’ expectations on them. It is also the fact that, whereas aspirants for
the prime minister office in the past “were safely entrenched in their parties” (Webb,
Poguntke and Kolodny 2011:9), they usually headed the extra-parliamentary
organization, now the top candidates, because of their limited party experience, no
longer control the party arenas where dissatisfaction with the electoral results is
mounted. They are the perfect scapegoat for the party’s defeat (Andrews and Jackman
2008). In brief, aspirants for the political leadership now no longer have to spend their
energies in burdensome party work and climbing through the internal party offices, but
at the cost of increasing their vulnerability in the face of electoral defeats.
We must note, however, that according to these scholars, a long internal party career is
not necessary to be selected as “top candidate”. But this does not mean that having that
party background is incompatible with being selected. Following their explanation, we
would expect that those current top candidates who by whatever reasons are still
“entrenched” in their parties, and have become party chair before getting their
nomination, could control the post-election challenges to their leadership by using their
internal power resources. As such, they would be as resilient as their precursors. If this
is so, we should expect that the fate of present top candidates in parliamentary
democracies is contingent upon their electoral fortune only when they are not
“entrenched” within their parties’ machines. Is this so?
6
This collateral consequence of candidates having limited party experience has also been supported by
other scholars such as Russell J. Dalton, David M. Farrell, and Ian McAllister, Political Parties and
Democratic Linkage (Oxford: Oxford University Press, 2011).
7
Things are not so simple, however, because in current parliamentary democracies the
power that controlling the party machine provides also seems to be quite uncertain from
a theoretical point of view. On one hand, as part of their argument about the current
“cartelization” of political parties, Katz and Mair have suggested that the increasing
reliance of political parties on the state resources has involved a shift in parties’ internal
distribution of power. The “party in public office” has managed to subordinate the
“party’s central office” because most of the states subsidies goes to the former. If this is
so, the change of the role and salience of party machines is even deeper. It is not only
that a long party service is no longer necessary for being selected as top candidate, it is
that controlling the extra-parliamentary party organization is not as useful as in the past.
Therefore losing top candidates that chair the party’s central office may be just as
fragile as those who do not if, for example, dissatisfaction with them appears among the
parliamentary party.
On the other hand, Katz and Mair themselves do not expect the relegation of the party’s
central office to be universal. They expect this relegation to be more pronounced in the
“mainstream parties” than in the so-called “fringe parties” because of their higher
accessibility to public subsidy and their office-seeking orientation. Similarly, Van
Biezen has suggested that in the so-called “new democracies” of South and Eastern
Europe the party’s central office emerges “as the most predominant” reflected in the
fact that the largest part of the state subventions is allocated to the extra-parliamentary
organization (Van Biezen 2000). This is so, she explains, because, given their weakly
developed party loyalties and lack of party institutionalization, political leaders in new
parliamentary democracies use the extra-parliamentary organization to maintain unity in
the party in public office. If true, then it is not clear that controlling party machines has
lost its salience in all parliamentary countries. Consequently, in the “new democracies”
we should expect that, ceteris paribus, current losing top candidates who are still the
party chairs are more likely to repeat as candidates than those who do not hold that
extra-parliamentary office. Is this theoretical expectation correct?
In order to test the last expectation, we focus on the candidates for the premiership of
the 17 Spanish regional governments who were “defeated”, referred to here as “losers”.
In addition, inspired by the only two empirical studies about losers’ fate (Taylor and
8
Boatright 2005; Carsey and Berry 2014), we will study only “first-time losers” (that is
candidates who ran for an elective office for their first time, and lost) who were in
opposition7. We thus focus on a critical moment of prospective political leaders: their
first elections as “top candidate”. Yet our goal is not to fully explain losers’ repetition in
itself, for as Taylor and Boatright have underlined (Taylor and Boatright 2005), often
intangible and whimsical factors affect this matter.
Spain, one of the countries analyzed by Van Biezen, is a good case to study the current
role of the party machine on losers’ survival in “new democracies”. First of all, its
current parliamentary monarchy was established in its constitution of 1978, and the
presidentialization of elections is said to have been present since the reestablishment of
democracy (Gunther, Montero, and Botella 2004). Therefore we would expect Spanish
parties to give high salience to whom they select as “top candidate”. Secondly, Spain is
also interesting because its party system has been mainly composed in the last decades
of two “mainstream” national parties, the social democrat PSOE and the conservative
PP, who have alternated power since 1982, and a smaller further-left coalition,
Izquierda Unida (IU)8. Therefore, we can additionally check if there are differences
between “mainstream” and “fringe parties” on the role of extra-parliamentary party
organizations on losers’ durability9.
However, studying the national politics of just one country has an insurmountable
limitation: we have more variables than cases. Fortunately, one of the recommended
ways to increase the number of observations relevant to our theory is to study
subnational units (King, Keohane, Verba, 1994). By studying the regional level we can
increase the number of observations while holding factors that vary across nations.
Obviously, studying regional politics has its own implications that we will consider
7
Cheibub and Przeworski (1999) show that running for the first time being already incumbent prime
minister affects negatively their chances of reelection by the people. This can also influence their chances
of reselection by their parties.
8
The rest has been almost exclusively “regional” parties.
9
For a definition of “fringe party” see Kai Arzheimer, “Fringe Parties” in J. Alt, G. Kurian, J. Alt, S.
Chambers, G. Garrett, M. Levi, and P. McClain, The Encyclopedia of Political Science (CQ Press, 2011).
9
when accounting for losers’ chances of renomination: the role of regional elections as of
a second-order type, and parties’ internal distribution of power.
Before presenting the descriptive data about the top regional candidates in Spain, we
must address a previous question: what does “losing an election” mean in a
parliamentary system?
Which “prime ministerial candidates” are regarded as “losers”?
In presidential systems after the polls are closed and the votes are counted, it is (almost)
certain who will become the next president and, therefore, which candidate has won the
elections. In parliamentary systems, if when the polls are closed and votes are counted,
no party, or pre-electoral coalition of parties, has obtained an absolute majority of
legislators, it is just the beginning of the game. A “top candidate” can belong to the
most voted party, with a plurality of MPs, and still be left in opposition.
So a first possible definition of a “winner” (and therefore a “loser”) is a top candidate
who becomes prime minister in the first government formed after the elections,
regardless of whether or not s/he is the candidate of the party who obtained the highest
number of legislators. The rest are “losers”.
Another feature of parliamentary systems, their collective or collegial executives, makes
things a bit more complex. Let us think about a real example taken from the Spanish
regional elections. For the 2003 elections in the Comunidad Autónoma of Cantabria in
Northern Spain, the regional branch of the Spanish socialists selected a new candidate,
Lola Gorostiaga, for the regional premiership. The Socialist Party, except for a brief
period at the beginning of the 90s, had never been in power in that region. After the
elections, the PSOE finished, as usual, as the second most voted party. However, the
most voted party − the Spanish Conservatives (PP) – again failed to obtain an absolute
majority. In the previous government, they had held the presidency in a coalition with
the Cantabrian regionalists (PRC), the third largest party in the region in both votes and
seats. This time Gorostiaga offered the regionalists the opportunity to preside over a
coalition government with the Socialists. The PRC accepted her offer, and she became
10
vice-premier of the new regional government. Can she be regarded as a “loser”? We
doubt it. Yet, according to our first definition she was.
In this example the PSOE successfully managed to enter the regional government.
Should we then consider all candidates whose parties are not able to enter the
government as “losers”? Let us think, for a moment, about candidates without real
expectations of becoming the next regional premier, and whose parties had no previous
legislators in the regional assembly, but who manage to obtain a seat for themselves,
and perhaps even for other colleagues. Are they going to be regarded as “losers” by
their own party? The appraisal they usually receive makes us doubt it. So from this
perspective we could now consider all top candidates whose party enters the first
government formed after the election as “winners”, or, if the party had never held
legislative seats, those who managed to obtain their seat. The “losers” will be the rest.
Which definition of “loser” is better? The first definition is perhaps too broad because,
owing to the strict interpretation of who a winner is (just those who become premiers),
we risk considering as “losers” candidates who were actually regarded as winners by
their own parties, while our second definition of loser is perhaps too narrow, because
with our lenient definition of who is a winner we risk excluding candidates who were in
fact regarded as losers by their parties. As a result, we must decide whether we prefer to
accept the error of possibly including some winners within our losers’ pool (type I), or
the potential error of excluding some of the losers from our pool (type II). We choose to
take a cautious stance here and avoid the type-I error. The second definition gives us the
highest probability of studying only losers, at the risk of excluding some of them 10.
The selection of “regional premiership candidates” in Spain
Spain is composed of 17 Comunidades Autónomas (CAs). The constitution of 1978
guarantees their exclusive competences in certain domains and concurrent competences
with the national government in others. Their political structure is formally
parliamentary, although their “presidentialized practice” has also been underlined (Aja,
10
In the multivariate analysis we will check if results vary according to the definition used.
11
1999). This means that each region has its own assembly, directly elected by the people,
which in turn selects the regional government. For 13 CAs, regional elections take place
at fixed intervals of 4 years.
The PSOE and the PP have dominated the regional party systems in most CAs.
Approximately 77 per cent of all regional governments have been headed by a premier
belonging to one of these two parties. The PP and PSOE have amassed around 65-75
per cent of the vote at the regional elections, whereas the average vote of IU at the
regional elections has been around 6 per cent11.
Regional institutions have also stimulated both the creation of regional branches of the
national parties and the emergence of parties exclusively located in some regions. The
regional branches of the national parties have their own organs of self-rule. The chairs
of the extra-parliamentary organization are formally elected by their respective regional
congresses, but in the mainstream parties, the PP and the PSOE, their national organs
closely monitor this process. In addition, the “candidates” for the premiership of
regional governments, are selected in a different process and normally by the regional
extra-parliamentary organs, although the ones belonging to the PP and PSOE must
receive the formal approval of the national party organs. Finally, since the late 90s, both
the PSOE and IU provide in their statutes for party primaries to select the top candidate.
However, their use in practical terms has been very limited (Astudillo 2014).
We have studied all the regional elections and the top candidates of these three national
parties for each of the 17 CAs from the early 1980s up to 201212. That means that we
have 427 candidates, but only 243 different individuals because many of them run more
than once (on average a candidate has run 1.8 times). Our dataset was built using
information from a variety of sources: party and public websites, a review of the
Spanish press, parties’ internal documents, and scholarly books.
11
Nevertheless, this party does nominate candidates for the premiership of the regional governments. See
for example the internal Statutes of IU in Aragon.
12
We have excluded the “regional” parties because of their low number of cases and the complexities
they introduce.
12
In order to have a better idea of who were these top candidates, we present their main
characteristics at the time of their first contest.
[Table 1 here]
In table 1 we can see that they are not atypical. Like regional legislative candidates
(Coller 2008), they are middle-aged men, have been affiliated to their parties for around
15 years and had public office experience of around 9 years. The majority of them were
party chairmen, but in every party a substantial amount were not. Most of them were not
in power, and were not selected in party primaries. We can also see that there are no
significant differences between candidates who were party chair and those who were
not, except in their party membership13.
The first column of Table 2 shows the proportion of oppositional first time losers each
party has had. As expected, this percentage is lower for the two mainstream parties,
while around 45 per cent of losers belong to our “fringe” party, IU.
[Table 2 here]
As we drop the losers of the most recent elections, since we do not know yet if they will
repeat or not, we retain 158 cases. Table 3 shows that over 40 per cent of them had at
least one second chance. We are not studying a rare event.
[Table 3 here]
The descriptive data also show that the two mainstream parties seem to present slightly
different features in comparison with the “fringe” party. As expected, the former tend to
present a higher percentage of candidates who were not party chair. On the other hand,
and contrary to our expectations, the mainstream parties do not get rid of losers in a
higher proportion than the “fringe” party. Notwithstanding this, why do some of them
repeat at least once and others do not?
13
We have rerun the analysis excluding those candidates who were not party members because this
feature could affect non-chair losers” chances of reselection. Results did not change.
13
Variables and hypothesis
Dependent variable: losers running again or not
Our dependent variable is whether an opposition candidate for the premiership of the
Spanish CAs who “lost” his or her first contest repeats in the next regional elections as
his or her party’s candidate (coded 1), or not (coded 0).
Main independent variable: being “party chairman”
Earlier we argued that in the case of a “new democracy”, such as Spain, we should find
that those candidates who control the party’s central office do have a higher probability
of surviving an electoral defeat and repeat as a candidate. As a proxy for the control of
this extra-parliamentary party organ we study if the candidate is the “party chair” or not
at the moment of their first contest14.
H1: losers are more likely to repeat if they are party chairmen at the time of their first
contest.
In our previous discussion, we also saw that controlling the party’s central office may
be less important in the case of the mainstream parties than in “fringe” parties. We have
thus built a dummy to distinguish between the PP and the PSOE (coded 0), and the IU
(coded 1). We have also studied an interaction effect between being party chair and the
class of party.
14
It is true that this proxy can be problematic. In some parties people who have just arrived in the party
can be first selected as party chair and immediately thereafter selected as prime ministerial candidate.
This is not our case, however. On average, those party chairmen that were selected as candidates had been
affiliated to their parties for around 12 years before they climbed at the top of the organization.
14
H2: the effect of being party chairman is stronger in “fringe” parties than in
“mainstream” parties.
Control variables
In our explanation of why electoral defeats may affect the chances of candidates being
renominated, we have so far held two assumptions that are debatable: (1) that losers
themselves will wish to run again, and that (2) they will have to confront other aspirants
to the next nomination. These factors may not only influence losers’ chances of
repetition, but they may also be associated with our main independent variable. We
therefore have to control for their impact. For example, those candidates who are not
party chair may be “amateur politicians”, easily disappointed by their first setback 15.
Candidates who are party chair may have a stronger probability of repeating, not
because they are party chair, but because they have a stronger character.
We know that the first assumption – losers will want to try again – is debatable even for
winners (Wolak 2007). Consequently, inspired by a common practice in the study of
incumbents’ retirement, we have controlled, first, for personal characteristics such as
“political ambition” (Fowler 1996; Herrnson 1997)16.
Political ambition and office holding experience
Running an electoral campaign is not an easy job. It requires a high level of
commitment, and candidates must have a strong desire for the potential benefits they
obtain through winning to compensate for these costs (Fowler 1996). As a result, those
who possess a high degree of ambition will be more willing to repeat their contestation
experience. In addition, the literature suggests that political ambition and office holding
experience are linked (Taylor and Boatright 2005). Consequently, we use as a proxy for
political ambition the number of years between candidates holding their first elective
15
We have already seen, however, that in Spain there are no significant differences in public office
experience between candidates who chair their extra-parliamentary organization and those who do not.
16
Following common practice we have also controlled for losers” age.
15
office (at any level) and their first contest for executive public office. Our concrete
expectation is:
H3: losers are more likely to repeat the higher the number of years they have held an
elective office (at any level) at the time of their first contest for executive office.
Secondly, in politicians’ personal cost-benefit analysis on running again, they can also
gauge their chances of winning by judging how the political and economic context, at
the time of the elections, affects those chances (Wolak 2007; Taylor and Boatright
2005)17. The assumption here is that a context favorable for incumbents discourages
losers from rerunning.
Public perception of the national government
However, as we are studying regional, second order elections, voters may use these
elections to have a mid-term say on the performance of the national government (Reif
and Schmitt 1980). Therefore our regional losers’ chances of winning the next elections
will be affected by citizens’ positive (negative) perception of the national incumbents
(Carsey and Wright 1998). Our concrete hypothesis is:
H4: the higher the national prime ministerial approval at the time of the next elections,
the higher the probability of a loser belonging to a party in national government
running again, and the lower the probability of a loser belonging to a party that is not
in national government doing so18.
17
It is true that losers” evaluation of trying a second time can be influenced by other options they have in
politics and the private sector, and we may know what losers did after not repeating. But we would not
know if obtaining another political office was the cause (losers preferred those offices to rerunning) or the
consequence of quitting (they got them as a “consolation prize” from their parties).
18
In Catalonia and the Basque Country the approval of the regional incumbent can also affect losers’
cost-benefit analysis. We have rerun the analyses without these two CAs to see if their exclusion affects
the test of this hypothesis. They do not.
16
The performance of the regional economy
The economic voting model suggests that governments are punished (or rewarded) for
bad (or good) economic conditions at election time. However, the fact that we are
dealing with a multilevel context may significantly alter how the economic voting
model holds. So we should specify the level of “economic performance” that citizens
will look at (regional or national levels), and which incumbents are held accountable
(the national or the regional ones). The study by Anderson shows that the public will
take into account the economic results at the regional level, but they will hold regional
politicians that belong to the party of the national incumbents accountable for those
results as a way of showing (dis)approval “of the performance of the national
government in a subnational election” (Anderson 2006). Here we use as a proxy, the
evolution of regional unemployment rate, given by the Instituto Nacional de
Estadística, between losers’ first elections and the following ones.
H5: the higher the increase of the regional unemployment level, the lower the
probability of a loser belonging to a party in the national government running again,
and the higher the probability of a loser belonging to a party that is not in national
government doing so.
Opportunity windows: the type of regional cabinet
Another contextual factor that may also affect losers’ desire to run again is a specific
peculiarity of parliamentary systems: prime ministers and governments can change
without a vote being cast. This means that, depending on the type of government
formed after an election, losers can become winners before new elections are called. If a
defeat is not necessarily seen as definitive, this can in turn influence their willingness to
continue in politics and, ultimately, to run again. To make it simple we distinguish here
between single-party majority cabinets and other situations:
H6: losers are more likely to repeat when no single party has an absolute majority than
when it does.
Party’s electoral evolution under the loser
17
So far we have also been assuming that, irrespective of a loser’s desire to repeat or not,
obtaining the following nomination will be a highly contested issue. That is why being
party chairman puts a candidate in a better position to defeat other aspirants to the
party’s nomination. However, parties may be so satisfied with the electoral results
obtained under their “losers” that nobody dares to challenge their renomination if they
decide to run again.
The issue here is how parties gauge the impact of their candidates in their results.
Following the study by Andrews and Jackman (2008), we suggest that they look at their
“electoral evolution”. That is, we compare a party’s results with its losers, in time “t”,
with previous results, in “t-1”, under a different candidate (keep in mind that we are
studying only first-time losers). In addition, we do not look at absolute but relative
increases. A five-point percentage increase involves a quite different relative growth
when the previous vote share was five per cent than when it was 40.
H7: losers are more likely to repeat the more positive, in relative terms, the electoral
evolution of the party is under their candidacy.
Other control variables
We have already stated that we are studying regional politics. This level of government
can have additional peculiarities that we should take into account. For example, some of
our parties have a multi-level structure with different degrees of “self-rule” for their
regional branches. In some parties, the regional organization can be highly autonomous
in selecting its candidates for regional premier. In others they may have to accept the
“suggestions” received from their national colleagues.
This variable is important to include in our model because it could explain the
association between being party chair and loser repetition. When regional party
branches’ self-rule is low, regional losers’ fate is decided by their national leaders, and
perhaps they select their favorite “henchmen” in the regions, first as chairmen of the
party branch, and later as “candidates” for the regional premiership. As a result, if
18
national leaders want them to repeat, in spite of having lost, they will do so, not because
they hold that party office, but because of their connection with the national party elite.
However, it is not clear to us if, on average, national leaders will want their regional
losers to repeat or not. That is why we do not have a predetermined average expectation.
In addition, given that there is no equivalent to a “Regional Authority Index” (Hooghe,
Marks, and Schakel 2010) for political parties, we have used proxies. First, the regional
institutional score for the “self-rule” dimension at the time of a candidate’s first
election19 and second, we have included in the analysis whether the party of the regional
candidate was in office at the national level. Being in the national government is usually
presented as a factor that weakens regional branches of national parties and strengthens
national leaders (Fabre 2008).
Finally, we have also introduced a two-period dummy variable to control for the
consolidation of both the new regional institutions created after the return of democracy
in Spain, as well as the party branches. The first period is from the first regional
elections, held from the beginning of the 1980s, to the end of 1993. In 1993, regional
governments were at least 10 years old and the last “Autonomic Pacts” to speed up the
decentralization process, signed by the PSOE and the PP, had started to be
implemented. The second period is from the beginning of 1994 until the latest regional
elections, in 201220.
Empirical Findings
To test the hypothesized relations, given that our dependent variable is binary (a loser
repeats or does not), we conducted a logit model. As the residuals for all losers within a
19
We have also used as proxy the electoral weight of “regional” parties instead of using the RAI. Results
did not change.
20
We did not include a variable referring to how candidates are formally selected because there is almost
no variation across time, regions and parties. In the same vein, given that for 13 out of 17 CAs there is a
fixed time-span between elections, we did not include this variable either.
19
Comunidad Autónoma may be correlated with each other, and the variance of the
residuals is likely not to be constant across CAs, we have used robust estimates of the
standard errors, clustering by CA21.
In addition, we have built two models. In the first one we wanted to see the effect of
being party chair while holding constant the class of party. In the second model we
include the interaction between being party chair and the class of party to assess if the
effect of being party chair is stronger, weaker, or does not matter in the case of fringe
parties compared to mainstream parties.
[Table 4 here]
The first model shows that, as suggested, heading the party organization has a clear
positive and significant effect on the likelihood of repeating in the following election. In
order to have a more substantive idea of the results, we have calculated the predicted
probabilities of being party chairman and not being so, while holding the rest of the
variables at their means. In the first case the probability is 53 per cent, in the second
case 23 per cent. Being party chair increases losers’ probability of repeating by 29
percentage points.
The rest of variables are not significant except for the relative evolution of the party
under the loser’s first contest. As expected, the greater the electoral evolution (in
relative terms), the higher the losers’ chances of repeating. Under certain circumstances
losers may not have to fight to get their renomination. If they lose, but the party is
satisfied with them, they repeat at least a second time.
Graph 1 gives an idea of how important controlling the party machine is on the
likelihood of losers running again, taking into account the electoral evolution. We
compare the effect of improving in relative terms the party results obtained by the losers
when they are party chair with the same effect when they do not hold that office. In both
21
The other approach for dealing with clustering, multilevel modeling, cannot be used given the low
number of cases within clusters as well as regions (see also Primo, Jacobsmeier, and Milyo, 2007).
20
situations, the likelihood increases. But the chances of repeating among those
candidates who are not party chair and improve electoral results by 25 per cent are more
than ten points lower than the chances of losers who are party chair, but reduce party
results by 25 per cent. Holding their party top office clearly offers candidates some
protection from bad electoral performance.
[Graph 1]
However, the second model shows a result that goes against our initial expectation. The
effect of being party chair decreases when the party is “fringe”, not “mainstream”. After
calculating the predicted probabilities, being party chair when the party is mainstream
increases the probability of repeating by 39 percentage points, but for our “fringe” party
it is just a 7-point increase. In the conclusion we suggest why this may be the case22.
We still must admit that it is very unlikely that parties select their party chairmen as
candidates randomly23. This unexplained factor could also be linked to the losers’
chances of repeating. For example, it is possible that the higher the number of years a
party is out of power, the higher the impatience of a party for returning to it, and the
more intransigent it will be in the face of a new defeat. Party chairmen could foresee
this and avoid running as candidates in these situations, not only because of the risk of
losing, but because their defeat could damage their continuation as party chair (Andrews
22
We have also checked if these results varied according to the definition used. In the first model they do
not, but in the second model the interaction effect between type of party and being party chair is not
significant. However, according to the first definition all IU top candidates are “losers” whereas probably
some of them were regarded as “winners”.
23
According to the “presidentialization” argument the selection of party chairs as top candidates ought to
depend on their electoral appeal. Unfortunately there are no public opinion polls about this matter. We
know, however, that parties do have their own private surveys about the popularity of their potential
candidates.
21
and Jackman 2008). This is not the case. In fact, the higher the number of previous
defeats, the higher the probability that the party chair will run as top candidate24.
Conclusions
In this paper we have analyzed the understudied reselection and replacement process of
candidates for prime ministerial office in parliamentary democracies. We already know
that political parties are a vital factor in explaining political leaders’ capacity to carry
out their policy preferences. However, we know much less about their role in the
incipient moments of their leadership. We have thus tried to assess if, at least in the case
of a “new democracy”, controlling the party machine still plays a role in the “durability”
of nascent political leaders in the face of electoral setbacks. The presidentialization of
elections and, above all, the alleged new salience of the “party in public office” would
suggest that the role of party machines has declined. It is not only a question of the
expansion of party primaries to select top candidates. It is that a long record of internal
party service may not be needed to be selected as top candidate, and above all, that
controlling the party machine now offers less protection faced with a bad electoral
performance. However, there is reason to believe that this is not the case in the “new
democracies”. Furthermore, the paper has also addressed two understudied issues: the
meaning of being a “prime ministerial candidate” in parliamentary democracies, and the
meaning of these candidates “losing” their elections.
Using the top candidates for regional executive office in Spain as a case study, we tested
empirically this expectation about “new democracies”. We admit that we may have
been very inclusive regarding which candidates are considered as “winners”. However,
this procedure gives us more guarantees that in our pool of “losers” no candidate
regarded as a “winner” by his or her party is included. The data clearly confirm the
salience of controlling the party machine in this democracy. First-time losers who are
still party chairmen have a higher probability of running again, around 29 percentage
points more than their fellow losers who are not. It is true that “improving” electoral
24
In addition, the number of previous defeats, controlling for party office holding and electoral evolution,
does not predict if a candidate will repeat or not. Data is not shown.
22
results also helps losers’ chances of rerunning, but this does not diminish the salience of
being party chair. Losers who manage to increase their party share by 25 per cent but
who do not hold that office still have less chance of repeating than losers who do, and
who reduce their parties’ share by the same amount.
We have also found that being party chair matters less in the case of our “fringe” party,
IU. We suggest that this may be the case because in the new democracies the leaders of
the extra-parliamentary organs of mainstream parties have a greater access to state
resources than their counterparts in fringe parties. This is not the situation in old
democracies because these resources go to the parliamentary party. However, explicit
comparisons of the effect of being party chair across types of parties and democracies
are needed to settle this issue.
Given the novelty of the topic, we do not claim to have settled the issue of under what
conditions nascent political leaders repeat after losing their first elections in
parliamentary democracies. Cross-national studies are needed to confirm some of our
findings, as well as to assess the influence of other factors that do not exist in Spain or
do not vary much. Nevertheless this study is a step in the analysis of the role of parties
in the (re)selection of “prime ministerial candidates”, and so far we conclude that in this
country, as an example of a “new democracy”, we do not see a decline in salience of the
extra-parliamentary organization on the durability of an incipient top political career in
the face of electoral setbacks, as can be suggested is the case for older democracies.
Controlling this party organ is still an asset. In Spain political parties are as important to
explain the beginning of political leaders’ career at the top, as they are to explain their
capacity to carry out their policy preferences later on.
23
Tables and graphs.
Table 1: Main features of regional top candidates who run for their first time
All
Party chair
Not party
chair
Men (%)
89.7
91.2
87.4
Age (mean of years)
44.4
43.3
46.3
6.2
-
15.8
Years affiliated to their party (mean of years)
15.0
15.6
13.6
Years holding a public office (mean of years)
9.0
9.0
8.3
First-timers already in government (%)
7.4
6.8
8.4
Candidate selected by primaries (%)
4.9
4.7
5.3
60.9
-
-
-
-
58.7
-
-
54.5
-
-
Not affiliated to the party (%)
Party chairmen (%)
Party chairmen by parties
PP
PSOE
IU
68.1
N
243
148
95
Source: dataset created by the author
Table 2: Characteristics of first time, oppositional losers (second definition only)
Percentage of losers by parties
Party composition of losers (%)
PP
69.0
28.0
PSOE
73.0
26.3
IU
87.9
45.7
Total
77.8
100
24
Table 3: Percentage of first time, oppositional losers that repeat by party
Not repeat
PP
PSOE
IU
Total
Repeat
Total
58.3%
41.7%
100%
(28)
(20)
(48)
55.0%
45.0%
100%
(22)
(18)
(40)
61.4%
38.6%
100%
(43)
(27)
(70)
58.9%
41.1%
100%
(93)
(65)
(158)
(N)
Source: dataset created by the author
Table 4: Factors explaining first time, oppositional loser’s running again
Model 1
Model 2
Repeat vs not repeat
Repeat vs not repeat
Constant
6.047 (3.914)
6.952 (4.000)
Party-chairman (dummy)
1.302 (.395)***
1.824 (.423)***
Age
-.041 (.031)
-.049 (.032)
Office holding experience
.009 (.026)
.013 (.024)
Party at national government (dummy)
.242 (1.671)
.046 (1.933)
Unemployment increase
-.022 (.045)
-.024 (.044)
-.073 (.080)
-.066 (.089)
-.224 (.459)
-.200 (.464)
-.045 (.323)
-.032 (.366)
Type of government (dummy)
.120 (.492)
.120 (.514)
Relative electoral evolution
.012 (.006)**
.013 (.006)**
Degree of “self-rule” of region
-.469 (.336)
-.576 (.337)
Class of party (dummy)
-.424 (.583)
.647 (.921)
Party-chairman*kind of party
-
-1.529 (.765)**
Time period
1.048 (1.060)
1.293 (1.050)
Pseudo R
.18
.19
N
141
141
Unemployment*Party at national government
Evaluation national Prime Minister
Evaluation national PM*Party at national government
2
Robust standard errors in parenthesis, ***p < .01, **p < .05 Party chairman: chairman (1), not (0); Party
at national government: at national government (1), not (0); Type of government: no party obtained an
absolute majority (1), yes (0); Class of party: fringe party (1), mainstream parties (0).
25
Losers' predicted probability of repeating
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
Graph 1. Predicted probabilities of renomination for losing candidates.
-25
-20
-15
-10
-5
0
5
10
Relative electoral evolution
Not party chair
CI 95%
15
20
25
Party chair
CI 95%
26
Bibliography
Astudillo, Javier (2014) ‘La Selección de los Candidatos a las Presidencias de los
Gobiernos Autonómicos’ in Francesc Pallarés (ed.), Elecciones Autonómicas, 20092012. Madrid: CIS (forthcoming).
Aja, Eliseo (1999) El Estado Autonómico. Madrid: Alianza Editorial.
Anderson, Cameron (2006) ‘Economic Voting and Multilevel Governance: a
comparative individual-level analysis’, American Journal of Political Science, 50: 2,
449--463.
Andrews, Josephine, and Robert Jackman (2008) ‘If Winning isn’t Everything, Why Do
They Keep Score? Consequences of Electoral Performance for Party Leaders’, British
Journal of Political Science, 38, 657--675.
Carsey, Thomas, and William Berry (2013) ‘What’s a losing party to do? The calculus
of contesting state legislative elections’, Public Choice, Published online: 09 April
2013.
Carsey, Thomas, and Gerald Wright (1998) ‘State and National Factors in Gubernatorial
and Senatorial Elections’, American Journal of Political Science, 42:3, 994—1002.
Coller, Xavier (2008) ‘Political Elites in Federalized Countries: the case of Spain (19802005), ICPS, Working Paper, núm. 268.
Cox, Gary (1988) The Efficient Secret: The cabinet and the development of political
parties in Victorian England. Cambridge: Cambridge University Press.
Cross, William, and André Blais (2012) Politics at the Centre: the Selection and
Removal of Party Leaders in the Anglo Parliamentary Democracies. Oxford: Oxford
University Press.
27
Detterbeck, Klaus (2013) ‘The Rare Event of Choice: Party primaries in German Land
Parties’, German Politics, 22:3, 270—287.
Elgie, Robert (1995) Political Leadership in Liberal Democracies. New York: St.
Martin Press.
Farrell, David, and Paul Webb (2002) ‘Political Parties as Campaign organizations’, in
Russell Dalton and Martin Wattenberg (eds.), Parties Without Partisans. Oxford:
Oxford University Press.
Fowler, Linda (1996) ‘Who Runs for Congress?’, Political Science and Politics, 29:3,
430--434.
Gallagher, Michael, and Michael Marsh, eds. (1988) Candidate Selection in
Comparative Perspective. London: Sage
Gunther, Richard, Montero, José Ramon, and Joan Botella (2004) Democracy in
Modern Spain. New Haven: Yale University Press.
Hazan, Reuven, and Gideon Rahat (2010) Democracy within parties: candidate
selection methods and their political consequences. Oxford: Oxford University Press.
Helms, Ludger (2005) Presidents, Prime Ministers and Chancellors: Executive
Leadership in Western Democracies. New York: Palgrave MacMillan.
Hooghe, Liesbet, Marks, Gary, and Arjan Schakel (2010) The Rise of Regional
Authority: a comparative study of 42 democracies. London: Routledge.
Herrnson, Paul (1997) ‘United States’, in Pippa Norris (ed.), Passages to power.
Cambridge: Cambridge University Press.
Katz, Richard, and Peter Mair (1993) ‘The Evolution of Party Organizations in Europe:
the three faces of Party Organization’, American Review of Politics, 14, 593--617.
28
Katz, Richard, and Peter Mair (2002) ‘The Ascendancy of Party in Public Office: Party
Organizational Change in Twentieth-Century Democracies’, in Richard Gunther,
Montero, José Ramón, and Juan Linz (eds.), Political Parties: Old concepts and new
challenges. Oxford: Oxford University Press.
Lawless, Jennifer (2012) Becoming a Candidate: political ambition and the decision to
run for office. Cambridge: Cambridge University Press.
McAllister, Ian (1997) ‘Australia’, in Pippa Norris (eds), Passages to power.
Cambridge: Cambridge University Press.
Norris, Pippa, ed. (1997) Passages to power: legislative recruitment in advanced
democracies. Cambridge: Cambridge University Press.
Poguntke, Thomas, and Paul Webb, eds. (2005) The Presidentialization of Politics: A
comparative study of modern democracies. Oxford: Oxford University Press.
Primo, David, Jacobsmeier, Mathew, and Jeffrey Milyo (2007) ‘Estimating the Impact
of State Policies and Institutions with Mixed-Level Data’, State Politics and Policy
Quarterly, 7:4, 446—459.
Reif, Karl, and Hermann Schmitt (1980) ‘Nine second-order national elections: a
conceptual framework for the analysis of European elections results’, European Journal
of Political Research, 8:1, 3--44.
Taylor, Andrew, and Robert Boatright (2005) ‘The Personal and the Political in Repeat
Congressional Candidacies’, Political Research Quarterly, 58:4, 599--607.
Van Biezen, Ingrid (2000) ‘On the Internal Balance of Party Power: Party organizations
in New Democracies’, Party Politics, 6:4, 395--417.
Webb, Paul, Poguntke, Thomas, and Robin Kolodny (2011) ‘The Presidentialization of
Party Leadership? Evaluating Party Leadership and Party Government in the
Democratic World’, Paper presented to the APSA conference, Seattle, September 2011.
29
Wessels, Bernhard (1997) ‘Germany’, in Pippa Norris (eds.), Passages to power.
Cambridge: Cambridge University Press.
Wolak, Jennifer (2007) ‘Strategic Retirements: the influence of public preferences on
voluntary departures from congress’, Legislative Studies Quarterly, 32:2, 285--308.
30