Why Not Parties? Electoral Markets, Party Substitutes, and Stalled Democratization in Russia Author(s): Henry E. Hale Source: Comparative Politics, Vol. 37, No. 2 (Jan., 2005), pp. 147-166 Published by: Ph.D. Program in Political Science of the City University of New York Stable URL: http://www.jstor.org/stable/20072880 . Accessed: 10/04/2014 12:18 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . Ph.D. Program in Political Science of the City University of New York is collaborating with JSTOR to digitize, preserve and extend access to Comparative Politics. http://www.jstor.org This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PM All use subject to JSTOR Terms and Conditions
21
Embed
Why Not Parties? Electoral Markets, Party …fdjpkc.fudan.edu.cn › _upload › article › files › 22 › 41 › bd6cf9e...Why Not Parties? Electoral Markets, Party Substitutes,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Why Not Parties? Electoral Markets, Party Substitutes, and Stalled Democratization in RussiaAuthor(s): Henry E. HaleSource: Comparative Politics, Vol. 37, No. 2 (Jan., 2005), pp. 147-166Published by: Ph.D. Program in Political Science of the City University of New YorkStable URL: http://www.jstor.org/stable/20072880 .
Accessed: 10/04/2014 12:18
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp
.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].
.
Ph.D. Program in Political Science of the City University of New York is collaborating with JSTOR to digitize,preserve and extend access to Comparative Politics.
http://www.jstor.org
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
Comparative studies have stressed two fundamental factors that drive the process of
political party development: prominent social cleavages and state (especially elec
toral) institutions. While some have divided theorists into "cleavage" and "institu
tional" camps, most leading scholars have in fact held that both combine to shape the evolution of parties despite variations in emphasis. Their and other classic works
also share an assumption that politicians have incentive to organize collectively in
parties in order to achieve political ends and that these activities are conditioned,
channeled, and ultimately rendered successful or unsuccessful by social cleavages and institutions. Leading scholars differ, however, on whether strong parties are an
equilibrium result in societies first making the transition to democracy. While
Duverger and many after him have stressed the power of incentives from cleavages and institutions over time to impel politicians to form dynamic and programmatic
parties, others like Kitschelt place more emphasis on transitional factors and
sociopolitical institutional legacies that can inhibit the development of strong, pro
grammatic parties.1
While both of these approaches to transitional party development clearly capture
important parts of the story, it is argued here that both can be usefully supplemented in the many countries of the world where law permits candidates to run as indepen dents (that is, without party affiliation). In these systems, parties will fully penetrate electoral politics only when virtually all serious candidates for office find it in their
interest to run as party nominees rather than as independents. In such a situation, candidates are effectively consumers of goods and services that will help them con
nect with voters in order to be elected. Parties can be seen as suppliers who emerge to meet this candidate demand. Most theories typically stop here, looking only at
parties as potential suppliers. Basic economic theory, however, says that production decisions for any good (an outcome of supply and demand) are not made in a per
fectly isolated market involving only that particular good, but in the context of
broader markets in which demand for this good can be dampened by competition from "substitute" goods that can meet some or most of the same needs, if imperfect
ly.2 Thus, in the realm of transitional polities, when parties are seemingly stunted in
their growth, the reason may lie in the existence of "party substitutes" that are able
147
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
to out-compete parties in the provision of electoral goods and services under certain
market conditions. The implication is that explicit theoretical attention must be paid not only to the incentives that institutions, social cleavages, and sociopolitical lega cies provide to political parties and those who would lead or join them, but also to
the way these same structures interact with variation in local patterns of meso-level resource concentration that can provide the basis for party substitutes. Paying explic it theoretical attention to such factors promises to improve the accuracy of predic tions (and explanations) of party development in many transitional societies.
The potential of market logic can be illustrated by applying it to an important
empirical puzzle that has left area specialists and comparative theorists divided: the case of political party development in Russia. Some researchers insist that Russian
parties are weak and stagnating, pointing to their lack of penetration of state organs and apparently weak regional organizations. Others argue that Russia's parties are
growing and are already quite strong for a country at its stage of democratization,
citing survey evidence that voters increasingly base their electoral choices on party
signals and that "transitional" party identification is steadily emerging. The market
approach makes it possible to see how, in an important sense, both camps are right.
Party loyalties and organization are in fact emerging, but parties have failed to close
out the electoral market and penetrate state organs less because they face the wrong institutional or societal incentives than because they face stiff competition from
other forms of political organization (party substitutes). The power of this model is
demonstrated by examining the impact of two important forms of party substitute,
provincial political machines and politicized financial-industrial groups, on party
development in the single member district (or simply district) elections to the
Russian parliament (Duma) in 1999. Quantitative analysis of an original dataset on
candidates and voting in these elections lends confirmation to this answer to the puz zle.
Conceptualizing the Electoral Market
In countries with electoral systems that allow candidates to run for office as inde
pendents, including the United States and India, party penetration of the political
system requires the vast majority of promising candidates to choose to run for office
as party nominees. In such systems, it is reasonable to make several assumptions in
developing theory. Voters vote for candidates who they think will represent their
interests (broadly conceived). Candidates, in turn, want to be elected and thus com
pete to show voters that they can best represent voter interests.3 Also, a relatively
plentiful supply of people will decide to run for office for any number of reasons, as
Cox notes is often the case in new democracies.4 Candidates' desires to gain votes
lead them to want political goods and services that help them get elected.
148
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
Political parties are a particularly important supplier of these goods and services
demanded by candidates. A party is defined here, after work by Sartori and
Schattschneider, as an enduring association of people who identify themselves by a
public label and who are joined together under this label for the primary purpose of
winning control of the government by means of presenting their own candidates in
elections for public office on the basis of a common platform.5 Comparative theoret
ical research has focused on three major electoral benefits that parties are said to
provide to candidates: organizational support, reputation (a "brand name"), and
material resources.6
These dynamics, taken together, constitute a kind of market. The "demand side"
of the production equation in the market for electoral goods and services includes
those factors affecting the amount of electoral goods and services that candidates
(consumers) are willing to buy from parties at a given price. The "supply side" refers
to factors that affect the amount of electoral goods and services that producers (in this case, parties) are willing to supply to candidates at a given price.7 Parties operat
ing in this market face some important constraints. For one thing, the number of
nominations they can meaningfully supply is limited by the number of districts and
mandates at stake in the election. Parties' provision of electoral goods and services is
also constrained by costs associated with building organization, eliciting collective
action, and developing reputation, none of which can be taken for granted.8 Even once parties are created, they always face real limits on the amount of resources they can mobilize to respond to competition in any one campaign. Competition is also
imperfect, in part because the goods proffered by political parties are not exactly the same. Some specialize in "ideational" goods (reputation) in which brand name dis
tinctions are critical, whereas others concentrate more on "administrative" products such as campaign services and financing.9
It would be a mistake to try to predict the size of the "party industry" (that is, the
degree to which parties are created and become important suppliers of electoral
goods and services) by focusing exclusively on parties as the producers of electoral
goods and services demanded by candidates. The demand side of a market equation reflects not simply the desire for a good potentially on the market, but also the avail
ability and relative price of substitutes for these goods.10 As the price for the good in
question rises, more people will turn to available substitutes that can satisfy the same
basic demand, albeit imperfectly. If the price of a cappuccino goes up, price-sensitive consumers may decide to settle for a regular coffee rather than absorb the new addi
tional cost of the fancier beverage. In the context of political party development, the
implication is that one must look not only at factors affecting the supply of and
demand for party-provided goods and services when estimating levels of party
development, but also at the potential availability of substitutes for these electoral
goods and services. To be sure, there is no reason to believe that parties are the only institutional forms capable of providing organizational support, reputation, and
149
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
material resources. It is therefore necessary to be aware of the possibility of party substitutes. A critical implication is that parties will close out the electoral market,
coming to dominate the democratic system, only when they establish themselves as
the sole credible suppliers of electoral goods and services. Likewise, parties can fail
to close out the electoral market even when they are in fact successful and powerful
providers of electoral goods and services; if party substitutes prove to be equally
mighty, candidates may continue to choose them even over strong political parties. Sources of cross-national variation in the strength of party substitutes vis-?-vis
parties are beyond the scope of the present analysis. Nevertheless, it can be briefly noted here that the degree to which party substitutes successfully compete against
political parties will depend on the nature of party substitutes available in a given
society, which in turn hinges on the particular pattern of concentration of those polit ical and economic resources that can potentially be mobilized for electoral purposes. In particular, the transition from what Kitschelt and his coauthors have called "patri
monial communist" regimes has tended to produce certain clientelistic regional social institutions and resource clusters that have proven to facilitate the rise of party substitutes in these more than in other postcommunist countries.11 It is crucial to
note, however, that the availability of party substitutes will also be directly affected
by the nature of political, including electoral, institutions. Indeed, proportional repre sentation systems typically rule out independent candidacies, effectively granting
parties a monopoly on ballot access, whereas most single member district systems do not.
Party Development in Russia
Russia is interesting as a case study of the market model not only because it is an
emerging democracy with single member districts, but also because it has generated a great deal of disagreement over how to interpret developments in its party sys tem.12 If the market model can help reconcile these positions, it will have demon
strated impressive explanatory power and will provide prima facie grounds for test
ing the model in other contexts.
On the one hand, broadly consistent with comparative scholars like Kitschelt who
argue that postcommunist political and social legacies often militate against strong
programmatic parties, a vast body of research argues (or even takes for granted) that
Russian parties are weak.13 Governors ran for reelection as party nominees only 3
percent of the time during the period 1995-2000; the mean share of a region's legis lators elected through party nomination was just 12 percent in the period 1995-2000; and less than 50 percent of the deputies chosen in the 1999 district
Duma elections were major party nominees.14 Some cite the frequent appearance
150
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
and disappearance of parties from election to election and signs that parties lack
regional structures or programmatic platforms.15
Parties are said to be underdeveloped for the following reasons. Voters are suspi cious of the very idea of "party" after their experience with Communist Party rule.16
The Soviet regime destroyed the social cleavages and related social infrastructure
that are said, following Lipset and Rokkan, to give birth to parties, and the transition
has failed to create stable new ones.17 Russian political institutions do not provide the proper incentives for party formation.18 Few organizational resources are avail
able for party leaders to use to build structure.19 A rise in postmaterialist values, the
atomizing effects of television, and other factors related to "modern European soci
ety" have weakened parties in Russia, just as they have driven a decline of parties in
European politics.20 Russia's governors have intentionally kept them weak, so as, for
example, to protect personal interests in "partial reform."21 Potential activists have
too little economic opportunity available to them and thus tend not to take the politi cal risks involved in party building.22 Russian President Boris Yeltsin's personal decision to neglect or subvert party formation nipped Russian parties in the bud.23
And Russia's particular form of postcommunist transition has produced an "oli
garchic capitalism" in which those with resources do not need parties and those
without them are too fragmented and disoriented to organize parties effectively.24 On the other hand, a different group of scholars, extending the perspective of
those comparative theorists who stress the tendency of structural incentives to pro duce strong parties, have provided systematic (mostly survey-based) evidence for the
following arguments.25 Russia's parties distinguish themselves clearly in the minds
of voters on the basis of issues.26 Voters are in fact choosing parties based on these
issue positions.27 Partisan voting patterns are grounded in postcommunist social
cleavages, as the classic Lipset-Rokkan framework would predict.28 Party loyalties are forming in the Russian population.29 This "transitional partisanship" is one of the
most important determinants of voting in Russia.30 Party activists show signs of
being ideological "true believers" rather than mere political opportunists.31 Parties
act in a distinctly coherent manner in the Duma, influencing the activities of legisla tors in important ways.32
There are several reasons why Russian parties might show consistent signs of
weakness at the same time many find evidence that they are taking root in mass
political culture. For one thing, some of the "pessimistic" studies were based on
older evidence from Russia's first multiparty parliamentary elections (1993). Second, most of the optimistic analyses are based on mass surveys of potential Russian voters, whereas most of the pessimists focus on macropolitical outcomes
(such as low party penetration of provincial organs). Third, perhaps as a result, these
researchers may be implying different definitions of party "strength" and "weak
ness." Survey research can tap into mass attitudes about parties but can not directly
151
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
assess the relative capacity of party organizations themselves. Studies of macro out
comes without micro-level data usually can not rule out the possibility that low party
penetration of state organs and high party volatility might result from factors other
than thin party organization, undeveloped reputation, or even a lack of popular sup
port. The pattern of evidence cited by the two camps, in fact, suggests that both
interpretations might be largely right. Parties may enjoy significant resonance in the
population and be coherent organizationally at the same time that forces other than
intrinsic weakness hinder them from actually dominating elections and the political
system more generally.
To test the interpretation that party substitutes can help account for the many
apparent signs of Russian party weakness, this article focuses on two particular kinds of party substitute: governors' political machines and politicized financial
industrial groups. They are appropriate as case studies because they are sufficiently
powerful and interested in electoral outcomes that they should behave as party sub
stitutes if the market model is to be helpful at all. While it may not be surprising for
Russia-watchers to learn that governors or "oligarchic" groups matter, it would be
striking to show that Russian party weakness is in part a result of these party substi
tutes' own political strength and not, as many Russia specialists have believed, a con
sequence of parties' own failures to organize or connect with voters due to postcom munist Russian political culture or perverse electoral institutional incentives.
This article seeks primarily to establish the usefulness and promise of the theoret
ical approach, leaving more comprehensive, cross-national testing to other work. It is
appropriate, then, to focus here on the case of the 1999 district Duma elections.33
Russia employs a mixed parliamentary election system, with 225 seats distributed
among party lists according to proportional representation and another 225 seats
allocated in district elections. While most studies of the relationship between parti
sanship and voting behavior in Duma elections have focused on the proportional rep resentation half, the only candidates voters can select there are partisan.34 Voters in
the district elections, in contrast, have the option of casting ballots for independent instead of party candidates. Partisan votes in district races thus reflect a choice to
"go partisan," not a lack of an affirmative alternative as is the case in the party list
contest.35 The district Duma elections, therefore, are well suited to testing a theory of party development that treats party votes and party affiliation as something to be
explained, not taken for granted. While one could also examine other single member district elections in Russia,
Duma races have certain advantages as a case study. Unlike regional executive and
legislative elections, they take place across the whole of Russia on the very same
day, helping to control for timing-related factors. Moreover, Duma contests feature
far greater numbers of districts and candidates than elections for any other Russian
organ taken individually. They thus foster greater confidence that observed patterns
152
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
are not merely random. The 1999 Duma election cycle, Russia's third, is appropriate to study because parties had had some time to develop reputations and organization,
making it possible to test whether their failure to dominate the electoral market has
more to do with their own intrinsic or strategic failings or, as suggested here, compe tition from party substitutes. Finally, the 1999 district Duma races are an appropriate case study because the necessary data are available to examine patterns of party behavior and party substitutes.
Suppliers on Russia's Electoral Market
There is good reason to believe that governors' political machines and politicized financial-industrial groups do in fact compete with parties to provide at least some
of the three main electoral goods and services that theorists usually assume parties
provide (organization, reputation, and material resources).
Governors' Political Machines The collapse of the Soviet regime and Russia's
transition from Communist rule left a great deal of power in the hands of the leaders
(to whom I refer broadly as "governors") of Russia's eighty-nine federal regions.36
By 1999 virtually all governors were chosen in direct popular elections and had built
up some very powerful political machines capable of providing locally many of the same political goods and services that parties are expected to provide, often much
more effectively than parties themselves.37 Strong governors had the most powerful local organizations in Russia, consisting of strict hierarchies of executive officials, all beholden to the governor for their jobs and kept in line by locally controlled secu
rity structures. Governors could also provide material resources on a large scale
since they typically strictly regulated major financial flows in their regions through state ownership of local enterprises, fiscal levers, the coercive use of license-issuing
authority, inspections, and other powers. While governors did not usually develop brand names and symbols that could help more directly associate a candidate with a
governor's reputation, they could appear frequently with their candidates on local
media or (illegally) send their representatives to polling places to influence voters.
These regional political machines could not be called parties themselves because
they did not express their primary aims in terms of an explicit label encompassing their candidates and did not develop public and coherent platforms on which their
candidates ran. These functions are some of the very ones that lead scholars to con
sider parties vital for democratic development and state stability. Explicit labels and
public platforms, for example, are critical in efficiently organizing choice for voters
and facilitating accountability.38 For the purpose of examining broad patterns, when terms such as "gubernatorial
support" are used here, they refer to any form of action on the part of the governor
153
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
designed to promote the victory of one particular candidate in any given election
contest. While future studies may gain comprehensive access to data on the different means that different governors use to back their favorites, this analysis is forced to
limit itself to looking only at patterns of who is supported regardless of how the sup
port is manifested. The core data on gubernatorial backing come from the East-West
Institute, which asked its network of Russian regional observers to identify the can
didate (if any) whom the governor supported in each Duma district in 1999.39 These
data were supplemented by identifying explicit statements that a given governor sup
ported a given candidate in two additional sources: reports from Radio Svoboda's
extensive (if not comprehensive) reporting on the Duma district races and summary information on these contests made available to the author but prepared for internal
use by a major Russian political party.40 Where at least one of these sources clearly identified a candidate as being supported by a governor and no other source contra
dicted this assessment, the candidate was coded as being governor-backed. In those
thirteen districts where at least one source contradicted at least one other source,
judgment calls were made based on the content of their reports. When there was
insufficient information in these reports to make a judgment, preference was given to the East-West Institute assessments, since its evaluative process was the most sys tematic.41
According to this data collection exercise, a governor backed a candidate in 181
of the 224 districts in the 1999 Duma elections, and most of this support was not
partisan in nature.42 Only nineteen of the eighty-eight governors were found in this
exercise to have supported a single party's nominees in more than half of their Duma
districts. The remaining governors either backed nominees from multiple (sometimes
opposing) parties in different districts, were not observed backing any candidate in a
large share of their districts, or, most frequently, supported candidates who were not
major party nominees. Of course, a candidate's ideal strategy would be to obtain
support from both a major party and a governor, something that the data indicate one
candidate successfully did in 39 percent of the districts.
Politicized Financial-Industrial Groups Starting with the USSR's legacy of large state industrial conglomerates, Russia's privatization process left intact huge monop olies controlling immense financial flows. These concentrations of resources formed
the basis for a series of financial-industrial groups that by the late 1990s had the
potential and the will to have a major impact on Russian politics. While some
observers have focused on financial-industrial groups' indirect role as sponsors of
political parties and individual candidates, perhaps their most interesting role is as
party substitutes in electoral politics.43 For the first time with the 1999 parliamentary election cycle, several major financial-industrial groups made conscious decisions
not to work primarily through alliances with a political party or to seek political
154
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
influence exclusively through the lobbying of legislators that had already been elect
ed. Instead, they compiled secret slates of their "own" district candidates in the 1999
Duma elections to elect maximally loyal and sizable representations in the new par liament.
Politicized financial-industrial groups have tended to rival political parties in pro
viding candidates with organization and especially material resources, although they have usually been (intentionally) weaker than parties in lending their candidates a
brand-name reputation since they want to avoid a public image of corruption. Politicized financial-industrial groups themselves are not parties because their pri
mary organizational purpose is economic; they do not seek to govern or to win out
right control of the polity, because their candidates typically do not publicly run on
the basis of the company label and because they usually do not make known the
"platform" that is the company's condition for lending them support. National politicized financial-industrial groups reputed to have their own district
candidate lists in the 1999 Duma race included Transneft, LUKoil, Russian
Aluminum, and the Interros Group, many of which were associated with some of
Russia's most prominent business "oligarchs." While in many cases these companies backed loyal local notables, top leaders of major financial-industrial groups at times even ran for office themselves. Some of the most impressive candidate resumes were
advanced by SibNeft' coowner Roman Abramovich, Logovaz Group owner Boris
Berezovsky, Slavneft' head Vasily Duma, and ONAKO chief Rem Khramov, all of
whom ran as independents. Of equal or often greater importance in many regions in
1999 were regional financial-industrial groups, sometimes linked to national ones
but often in competition with them in particular districts.
While evidence is less forthcoming than for governors, the internal party report and to a lesser extent Radio Svoboda and written accounts of the campaign process
produced by the East-West Institute's observers frequently indicated politicized financial-industrial group involvement in the district Duma contests.44 A candidate
was coded as being a politicized financial-industrial group candidate if reports from at least one source met two criteria: there was evidence not merely of a link between a candidate and a major firm but of the active support of the leadership of the firm
for the candidate's run for office (counting a run by the firm's CEO or owner as
"active support"); the firm in question was either national in scope or a major part of
a region's economy (such as the automobile giant KamAZ in Tatarstan), not an ordi
nary factory.45 Due to the often secretive nature of politicized financial-industrial
group activity, it is not possible to be certain that the dataset has captured all recipi ents of politicized financial-industrial group support, so these results must be treated
as only a preliminary cut.46
Politicized financial-industrial groups were found to have backed a total of seven
ty-seven candidates in 1999. While this figure is a only a small share of the 2,226 candidates on the ballot in this election, they competed in sixty-one separate dis
155
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
tricts, thereby being in a position to affect a quarter of the contests. When politicized financial-industrial groups are combined with governors (one of whom covers each
district), party substitutes as a whole were in a position to influence party prospects and associated election outcomes significantly. Indeed, like governors, politicized financial-industrial groups tended to operate independently of the major political
parties. Of the seventy-seven politicized financial-industrial group candidates, only
twenty-eight also happened to be nominated by a political party.
A Statistical Test
If the market model accurately captures a real dynamic, empirical support should be
found for the following three hypotheses in data on the 1999 Duma elections. (1) Candidates nominated by a major party should have tended to get larger percentages of the vote than those without such a nomination, ceteris paribus. (2) Candidates
with the active backing of major party substitutes (governor's political machines and
politicized financial-industrial groups) should have tended to get larger percentages of the vote than those without this backing, ceteris paribus. (3) There should be a
negative correlation between the vote totals of gubernatorial political machine candi
dates and the vote totals of party nominees across those districts where governors do
not back party candidates.
Hypotheses 1 and 2: Method To test the first two hypotheses, two regressions were run using the dataset described above, which includes information on all candi
dates competing in the 1999 district elections to the Duma.47 For both regressions the percentage of the district vote that each district candidate won in the election is
the quantity to be explained. To test the first two hypotheses directly, dummy vari
ables were included indicating nomination by Russia's seven major parties as of
1999 (Communist Party, Fatherland-All Russia, LDPR, Our Home is Russia, Union
of Right Forces, Unity, Yabloko) as well as active support by the two party substi
tutes in focus here (governor, politicized financial-industrial group). See the
Appendix for a more technical explication of the employed method.
There are at least two reasons why hypotheses 1 and 2 might be true, however.
For one thing, parties, governors' political machines, and politicized financial-indus
trial groups are expected to improve the vote-getting capacity of their candidates.
The market model also anticipates that parties and party substitutes will compete for
candidate recruitment, seeking to attract the candidates that are a priori most likely to get the most votes. Strong parties in developed democracies are widely held to do
both. To the extent that a party or party substitute is strong, therefore, it should be
expected to display a mix of "making vote-winners" and "attracting vote-winners."
While the specific ratio of "making" to "attracting" is not of particular importance
156
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
to the present argument, it should be reasonably established that at least some signif icant "making" is going on. It is also important, however, to rule out a third possibil
ity not anticipated by the market theory, that strong candidates are coming forward on their own and that parties or party substitutes are then simply "rubber-stamping" them, lending them no significant electoral support.
In two ways, this statistical analysis isolates the degree to which parties and gov ernors are "making" rather than "attracting" or "rubber-stamping" vote-winners.
First, it includes a series of control variables that indicate whether a given candidate
possesses certain traits or personal resources that might be expected to lead to elec
toral success regardless of that candidate's affiliations or backers.48 These controls
capture whether a candidate was an incumbent Duma deputy (elected either in the
district or in the proportional representation half of the voting); a state official (at the
federal, regional, or local level); a regional or local legislator; or primarily employed in the military, the intelligentsia, agriculture, politics (as with full-time party lead
ers), blue-collar (proletarian) work, sports (sometimes linked to mafia groups), the
railroad industry (associated with a shadowy Kremlin network), or private business
(as distinct from politicized financial-industrial groups in the way described above
and in note 45). A variable coding gender (female) is also incorporated into the
analysis, as are variables indicating nomination by the three minor parties that round
out Russia's top ten vote-getters in the 1999 proportional representation competition: Communists for the USSR, Women of Russia, and the Pensioners Party.
The second way in which this analysis controls for the personal strength of candi
dates so as to isolate the effects of parties and party substitutes is to include a variable
that indicates whether regional experts considered particular candidates to be "key contenders" prior to the election. The East-West Institute, in a predictive exercise, asked its network of Russian regional observers to identify candidates who were key contenders in nearly every district prior to the election. These predictions were then
augmented with similar assessments conducted in a large number of districts by Radio Svoboda's regional correspondents and in almost all districts by a political
party whose internal data were made available to the author. Candidates were thus
coded as contenders if any one of these three sources named the candidate as a realis
tically possible winner or a primary challenger. These expert assessments should be
expected to capture more intangible but very important candidate assets like charis
ma, popularity, and name recognition, as well as tangible assets like personal wealth
that might be known locally but may not be evident from a candidate's official biogra
phy. To the extent that candidates backed by parties or party substitutes perform better
than candidates with similar biographies above and beyond local expert expectations, it can be said with a great deal of confidence that these forms of political organization contribute directly to their candidates' success and are not succeeding only in candi
date recruitment (or in identifying good candidates to rubber-stamp). It is critical to note, however, that the text of the reports filed by many of these
157
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
correspondents indicated that they judged candidate prospects not solely by their per sonal assets, but also by the backing of strong parties or party substitutes.
Regressions including the contender control variable, therefore, are likely to under
state the real level of party and party substitute vote-winning capacity in the district
elections. These regressions, then, are best interpreted as establishing a lower bound
to the real average degree to which parties and party substitutes were helping candi
dates win votes, with an upper bound provided by those equations that do not include
this control variable. For this reason, both sets of results are reported in Table 1 :
columns 1(a) and 1(b) leave out the contender control, and columns 2(a) and 2(b) include it.49
Table 1 Correlates of Candidate Performance in the 1999 Duma District Elections
Party Substitutes Governor PFIG
Major Parties Communist Party Yabloko Union of Right Forces Fatherland-All Russia Unity LDPR Our Home is Russia
Controls Communists for USSR Women of Russia Pensioners Party Incumbent (SMD) Incumbent (PR)
Military Intelligentsia
Agriculture Politics Proletarian Federal Official Regional Official
Regional Legislator Mayor Local Official Local Legislator Railroad
Sports Business Female Number
Experts See as Contender F? N *
p<J0 ** p <.05 ***
p<.01
1(a) 1(b) Impact of factor on % district vote (95% confidence Coefficient interval}_ (standard error)
Hypotheses 1 and 2: Results Attention is first drawn to the bolded columns of
Table 1, which report the estimated magnitude of the impact of the different parties and party substitutes (and control factors) on the vote totals likely to be won by oth
erwise average candidates in their districts. Most strikingly, both parties and party substitutes are found to have played large roles in Russia's 1999 district elections,
confirming hypotheses 1 and 2. An average Duma candidate who gained the backing of an average governor in 1999 could accordingly expect to win, on average, an
additional 7.1 to 11.7 percent more of the district's vote, with the difference between
the figures in 1(a) and 2(a) depending on whether experts' assessments of candi
dates' prospects were controlled. Similarly, politicized financial-industrial groups were associated with a mean electoral bump of between 5.3 and 8.3 percent.
Table 1 also reports strong evidence that parties, too, provided significant elec
toral support for their candidates. That Communist Party nomination was correlated
with an electoral boost of nearly ten percentage points is perhaps the least surprising
finding given the undisputed strength of the party's organization and brand name.
More remarkably, nominees of the strongest liberal-market parties, Yabloko and the
Union of Right Forces, tended to garner 3.7-5.4 and 5.6-6.9 percent more of the
vote, respectively, than did otherwise similar candidates. Given that over two-fifths
of the 1999 district Duma races were decided by 7 percent of the vote or less and
that roughly half of these races involved winning margins of 3 percent or less, the
electoral advantage associated here with major party nomination appears quite sig nificant in real world terms. Also surprising is the clear finding that the nominees of
parties often presented as purely dependent on friendly governors and lacking in
independent organization and reputation, Fatherland-All Russia and Unity, systemat
ically outperformed other candidates with similar biographies by average margins of
2.7-6.9 and 4.6-10.7 percent, respectively, even controlling for gubernatorial sup
port. Two major parties had no significant effect on their nominees' chances:
Vladimir Zhirinovsky's LDPR (which initially went out of its way to include reputed criminals on its party list in what appears to have been a fundraising tactic) and Our
Home Is Russia (the party of former prime minister Viktor Chernomyrdin, widely believed responsible for the Russian financial collapse of 1998).
Table 1 also indicates that it can be concluded with a high degree of confidence
that parties and party substitutes both mattered. The parenthesized figures in
columns 1(a) and 2(a) report 95 percent confidence intervals. Thus, given this infor
mation, it can be said with 95 percent confidence that an average governor's backing
brings an average district candidate somewhere between 9.2 and 14.2 percent more
of the district's vote. In no instance does "zero impact" fall into the range of possi bilities in which there is 95 percent confidence for the five most important parties discussed in the previous paragraph or for the two forms of party substitute consid
ered here. While the figures in the columns labeled (b) need not be of concern here
(see the Appendix for interpretation), it is important to note the patterns of asterisks:
159
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
the presence of three asterisks by the figures for both kinds of party substitutes and
the five key parties indicates that, if the real effects of these forms of political orga nization were in fact zero, the likelihood that patterns would have taken the form that
they actually did would be less than 1 percent. The control variables also identify some interesting patterns. Perhaps most impor
tant, the expert predictors were generally accurate; they successfully identified can
didates that received, on average, 7 percent more of the vote than would have been
predicted by the "objective" factors included directly in the equation. This accuracy lends credence to the argument that the expert assessment variable captures candi
dates' "personal votes" reasonably effectively and thus that the remaining correlation
found between vote shares and party nomination or party substitute backing reflects
their "making," rather than just "attracting" or "rubber-stamping," of vote-winners.
Adding to the conclusion that parties matter, some of Russia's minor parties were
even correlated with slightly better than expected showings for their district candi
dates. Russia also appears to feature a significant incumbent advantage, especially for district-elected incumbents who earned a mean 3.7 to 10.1 percent more of the
vote than did otherwise similar candidates. Candidates employed in agriculture, in
the railroad industry, or as regional officials, regional legislators, mayors, or local
legislators also performed better than did other candidates who differed only in lack
ing these jobs. Finally, there is some slight evidence that voters reacted negatively to
simple business representatives (those not from politicized financial-industrial
groups) seeking to win office.
Hypothesis 3: Method To confirm the third hypothesis, there must be evidence that
parties are failing to close out the electoral market at least partly because of the com
petition they face from party substitutes rather than because of their own structurally induced incapacity to make inroads into an institutionally open political space. These
different interpretations can be operationalized in a testable manner. If the "intrinsic
weakness" theory is right, no correlation would be found between the overall perfor mance of party nominees and that of gubernatorial political machine and politicized financial-industrial group candidates. To say that the major parties were failing purely due to such qualities as poor organizational capacity, a lack of resources, the absence
of meaningful reputation, or institutional incentives is to say that the share of the vote
that major parties "earned" collectively for their candidates through nomination in an
average district was not constrained in any way by that part of the vote earned collec
tively by the party substitutes studied here. Indeed, all of the variables noted above
(including parties and party substitutes, not to mention the controls) explained only 51 percent of the variation in district candidate vote totals, as indicated by the estimat
ed value of R2, at least suggesting that parties had room to expand independently of
party substitutes but failed to do so. If, in contrast, the supply and demand theory is
correct, a negative correlation would be found between the performance of important
160
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
party substitutes (as reflected in the share of ballots won by the candidates they
backed) and the vote totals of major party nominees. According to this interpretation,
many factors always go into voting decisions, including such intangibles as personali
ty and voter physical preferences; the 49 percent of the variation found to be
"unclaimed" by parties, governors, or politicized financial-industrial groups likely reflects such intangibles. Just because parties and party substitutes combined do not
explain 100 percent of the variation in voting, therefore, does not mean that they are
not in competition for that share of the vote that various political service providers
actually can hope to capture for their candidates.
To test these rival interpretations, the tobit statistical technique is employed. Tobit
makes use of the information that the cumulative vote percentage won by major par ties in each district (the dependent variable) is effectively "censored" at the value of
zero?that is, total party performance could not register as less than zero no matter
how bad the situation got for parties.50 The fact that in fourteen of the races consid
ered here major parties nominated no candidates and hence received zero votes
means that more accurate results are likely to be obtained by treating the data as cen
sored.51 Leaving other sources of cross-district variation in party performance for
future research, the testing here is limited to the narrow third hypothesis advanced in
this article. Is there a negative correlation between the performance of major party nominees and that of party substitute candidates? While the result will not be fully conclusive since district-level data are generally not available to detect all possible influences on voting for party candidates, it will be highly suggestive and an impor tant step beyond prior knowledge. The test was conducted in those 120 of the 224
districts where no major party nominee happened to be backed by a governor or a
politicized financial-industrial group. This test is the most appropriate one since the
argument is explicitly interested in the many districts where governors and party nominees were clearly in competition.52
Hypothesis 3: Results Table 2 reports the findings. As predicted by the supply and
demand theory, there is a strongly negative correlation between the vote totals of
candidates backed by governors and politicized financial-industrial groups and those
of the seven major parties' nominees in those 120 districts where governors did not
back major party nominees. The coefficients allow some idea of just how powerfully
governors and politicized financial-industrial groups can dampen party performance. When a governor's candidate won an extra one percent of a district's vote, the share
of this vote garnered by Russia's seven biggest parties is found to have dropped by about one-third of one percent. Politicized financial-industrial groups appear to
"crowd out" parties even more powerfully. For every one percent of the vote won by a politicized financial-industrial group candidate, Russia's major political parties are
found to lose over half of a percentage point of the district's ballots.
161
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
Table 2 Correlates of District Totals of the Vote for Major Party Nominees
Governors, political machines: - 0.33***
PHGs: - 0.57***
Pseudo-R2 ?.U3
Number of observations 120 ***
P < -01
Conclusion
Parties may fail to penetrate a transitional polity not only when they are intrinsically weak, but also when they are strong but happen to face similarly strong competitors
(party substitutes) in the market for goods and services that help candidates get elected. This argument reconciles what otherwise appear to be strikingly contradicto
ry findings in the Russian case. Researchers who find that parties play a large role in
structuring voter choice and do so on the basis of ideology and social cleavages are
right. But those who report extremely thin party penetration of most state institutions
are also right. Russia's parties have failed to close out the electoral market not so
much because they are weak in and of themselves, but because they face heated
competition from party substitutes like governors' political machines and politicized financial-industrial groups that are in fact quite strong.
An important implication for comparative scholarship is that theories of institu
tions and broad social cleavages need to be supplemented with theories explaining variation in patterns of party substitutes. This variation will certainly depend on
electoral institutions, although it will also lead to the study of meso-level concentra
tions of power and resources (legacies of the dictatorial past and the transition itself) that are not broad enough to constitute social cleavages but that entrepreneurs might use to build party substitutes. Research into the formation of strong party systems in
new democracies should, therefore, more directly address how parties come to close
out the electoral market. This query is relevant not only to Russia, but to all coun
tries whose electoral systems permit independents to run. Indeed, independent candi
dates were a major part of political life in states as diverse as the U.S. of the 1820s
and India just after independence. The reasons why their position changed raise
some of the most important questions in explaining democratic development.53
Appendix: Statistical Method
The dependent variable for hypotheses 1 and 2 is a percentage. It is thus bounded
between zero and one hundred and can be interpreted as a composition, a particular
162
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
allocation of a district's total vote between one candidate and the others. Following
techniques developed and tested for analyzing compositional data, the dependent variable is transformed into a log ratio, the natural logarithm of a candidate's vote
total divided by the share of the vote not received by the candidate. Since log ratios are unbounded, they facilitate the method of ordinary least squares (OLS), which assumes no bounds.54 Since observations are not expected to be completely indepen dent within election districts but are expected to be independent across districts, the
Huber-White estimator of variance is used to obtain accurate estimates of statistical
confidence in the results and a control variable (Number) is also included indicating how many candidates were contesting a given district.55
While the actual coefficients (along with standard errors and p-values) are report ed in columns (b) of Table 1, it is helpful to use the software Clarify to obtain esti
mates of the average gain in the percentage of a given district's vote that an average candidate can expect to receive when his or her score on the listed factor changes from its minimum to its maximum value as well as the associated 95 percent confi
dence interval.56 Clarify works by using stochastic simulation to generate (in this
case) 1,000 estimates of the parameters, setting the independent variables at the
appropriate values, and then calculating the desired quantities from the 1,000 result
ing simulations. These results are reported in columns (a) of Table 1.
163
NOTES
The author would like to thank Naomi Wachs for research assistance, Robert Orttung for some important
data, Indiana University's Russian and East European Institute for funding through the Andrew M. Mellon
Foundation Endowment, and the many people who read and provided helpful feedback on earlier drafts in
various fora, including a workshop organized by the Program on New Approaches to Russian Security. 1. John H. Aldrich, Why Parties? (Chicago: University of Chicago Press, 1995); Gary Cox, Making
Votes Count (New York: Cambridge University Press, 1997); Maurice Duverger, Political Parties (New York: Wiley, 1954); Herbert Kitschelt, "The Formation of Party Systems in East Central Europe," Politics
and Society, 20 (March 1992), 7-50; Herbert Kitschelt et al., Post-Communist Party Systems (New York:
Cambridge University Press, 1999); Seymour Martin Lipset and Stein Rokkan, eds., Party Systems and
Voter Alignments (New York: Free Press, 1967); Rein Taagepera and Matthew Soberg Shugart, Seats and
Votes (New Haven: Yale University Press, 1989). 2. Israel M. Kirzner, Competition and Entrepreneurs hip (Chicago: University of Chicago Press,
1973); Paul A. Samuelson, Economics, 8th ed. (New York: McGraw-Hill, 1970). 3. Anthony Downs, An Economic Theory of Democracy (New York: Harper & Row, 1957). 4. See Regina Smyth, "The Political Implications of Mixed Electoral Systems" (draft, 1998); Cox,
pp. 151-52.
5. Giovanni Sartori, Parties and Party Systems, vol. 1 (New York: Cambridge University Press,
1976), pp. 58-64; E. E. Schattschneider, Party Government (Westport: Greenwood Press, 1970), pp. 35-37.
6. Aldrich; Cox; Kitschelt et al.; and Smyth. 7. Samuelson, p. 59.
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
8. Aldrich; Cox; David M. Kreps, "Corporate Culture and Economic Theory," in James Alt and
Kenneth Shepsle, eds., Perspectives in Political Economy (New York: Cambridge University Press, 1990),
pp. 90-143; Mancur Olson, The Logic of Collective Action (Cambridge, Mass.: Harvard University Press,
1965); and Daniel S. Treisman, "Dollars and Democratization," Comparative Politics, 31 (October 1998), 1-22.
9. Smyth; Stephen E. Hanson, "Ideology, Uncertainty, and the Rise of Anti-System Parties in Post
Communist Russia," Journal of Communist Studies and Transition Politics, 14 (March-June 1998), 98-127.
10. Kirzner, pp. 104-7; Samuelson, pp. 58, 409, 415-16.
11. Henry E. Hale, "Explaining Machine Politics in Russia's Regions," Post-Soviet Affairs, 19 (July
September 2003), 228-63; Kitschelt et al.; Herbert Kitschelt and Regina Smyth, "Programmatic Party Cohesion in Emerging Postcommunist Democracies," Comparative Political Studies, 35 (December
2002), 1228-56.
12. Operationally, what this study defines above as a "party" is formalized by the legal recognition of
the ministry of justice of the right of the organization to nominate candidates officially in Duma elections.
13. Michael McFaul, "The Perils of a Protracted Transition," Journal of Democracy, 10 (April 1999),
4-18; Peter Reddaway, "Instability and Fragmentation," Journal of Democracy, 5 (April 1994), 13-19;
Richard Rose, "Mobilizing Demobilized Voters in Post-Communist Society," Party Politics, 1 (January
1995), 549-63; Richard Rose, "A Supply Side View of Russia's Elections," East European Constitutional
Vlasti Sub"ektov Rossiiskoi Federatsii 1997-2000, vols. 1, 2 (Moscow: Ves' Mir, 2001). 15. Kitschelt et al.; Kitschelt and Smyth; Rose, "A Supply Side View"; Richard Rose and Neil Munro,
Elections without Order (New York: Cambridge University Press, 2002). 16. Jerry F. Hough, "The Failure of Party Formation and the Future of Russian Democracy," in
Timothy J. Colton and Hough, eds., Growing Pains (Washington, D.C.: The Brookings Institution, 1998),
pp. 685-88; Michael McFaul, Russia's Unfinished Revolution (Ithaca: Cornell University Press, 2001), p.
315; Richard Sakwa, "The Development of the Russian Party System," in Peter Lentini, ed., Elections and
Political Order in Russia (New York: Central European University Press, 1995), p. 184.
17. Hough, "The Failure of Party Formation and the Future of Russian Democracy," pp. 688, 696;
McFaul, Russia's Unfinished Revolution, p. 316; Sakwa, pp. 191-92.
18. Hough, "The Failure of Party Formation and the Future of Russian Democracy," p. 691; John T
Ishiyama, "Political Integration and Political Parties in Post-Soviet Russian Politics," Demokratizatsiya, 1
(Spring 1999), 200; Sakwa, p. 169; Smyth. 19. Ishiyama, p. 200.
20. Sakwa, p. 190.
21. Slider; Stoner-Weiss, "The Limited Reach of Russia's Party System." 22. Kelly M. McMann, "The Personal Risks of Party Development," in Joel C. Moses, ed., Dilemmas
of Transition in Post-Soviet Countries (Chicago: Burnham Publishers, 2002). 23. Hough, "The Failure of Party Formation and the Future of Russian Democracy," pp. 699-700;
McFaul, Russia's Unfinished Revolution, pp. 315-17.
24. McFaul, Russia s Unfinished Revolution, pp. 317-19.
25. Aldrich; Cox; Duverger; Lipset and Rokkan; Taagepera and Shugart.
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
26. Ted Brader and Joshua A. Tucker, "The Emergence of Mass Partisanship in Russia, 1993-1996,"
American Journal of Political Science, 45 (January 2001); Thomas F Klobucar and Arthur H. Miller,
"Party Activists and Party Development in Post Soviet Democracies," paper presented at the annual meet
ing of the American Political Science Association, Boston, August 28-September 1, 2002; Stephen White,
Richard Rose, and Ian McAllister, How Russia Votes (Chatham: Chatham House, 1997). 27. Miller et al.; White, Rose, and McAllister.
28. Stephen Whitefield and Geoffrey Evans, "Class, Markets and Partisanship in Post-Soviet Russia:
1993-96," Electoral Studies, 18 (1999), 155-78; Whitefield, "Partisan and Party Divisions in Post
Communist Russia," in Brown, ed., pp. 235-43.
29. Brader and Tucker; Timothy J. Colton, Transitional Citizens (Cambridge, Mass.: Harvard
University Press, 2000), pp. 112-15; Arthur H. Miller et al., "Emerging Party Systems in Post-Soviet
Societies," Journal of Politics, 62 (May 2000), 455-90; Arthur H. Miller and Thomas F Klobucar, "The
Development of Party Identification in Post-Soviet Societies," American Journal of Political Science, 44
(October 2000). 30. Colton.
31. Klobucar and Miller.
32. Thomas F. Remington, "Political Conflict and Institutional Design," Journal of Communist Studies
and Transition Politics, 14 (March-June 1998), 201-23; Steven S. Smith and Thomas Remington, The
Politics of Institutional Choice (Princeton: Princeton University Press, 2001). 33. On these elections, see Yitzhak Brudny, "Continuity or Change in Russian Electoral Patterns?," in
Brown, ed., pp. 154-78.
34. Colton; Miller et al; Miller and Klobucar; Brader and Tucker.
35. Robert Moser, "Independents and Party Formation," Comparative Politics, 31 (1999), 148.
36. Not all regional leaders have this formal title, but it is used here for convenience. Hale,
"Explaining Machine Politics in Russia's Regions"; Steven L. Solnick, "Russia's 'Transition,'" Social
Research, 66 (Fall 1999), 789-824; Stoner-Weiss, "Central Weakness and Provincial Autonomy"; Daniel
Treisman, After the Deluge (Ann Arbor: University of Michigan Press, 1999). 37. Henry E. Hale, "Machine Politics and Institutionalized Electorates," Journal of Communist Studies
and Transition Politics, 15 (December 1999), 70-110; Henry E. Hale, "Bashkortostan," in Colton and
Hough, eds.; and Stoner-Weiss, "The Limited Reach of Russia's Party System." 38. Samuel P. Huntington, Political Order in Changing Societies (New Haven: Yale University Press,
1968); Seymour Martin Lipset, "The Indispensability of Political Parties," Journal of Democracy, 11
(January 2000), 48-55; Mair.
39. Observers were given only the general instruction, not specific criteria by which to make their
assessments.
40. Radio Svoboda transcripts can be obtained at http://www.svoboda.org/archive/elections99. 41. Cases for which all three sources gave no indication were coded as "governor supports no one."
Cases where only one source covered the district and where that source gave only an indication (not a
definitive statement) of a gubernatorial endorsement were coded as gubernatorial support. A regression
applying stricter coding criteria (dropping the kinds of cases mentioned in this note) obtained essentially the same findings as those reported here.
42. The eighty-ninth region, Chechnya, and its district (the 225th) are excluded.
43. Colton and Hough; Juliet Johnson, A Fistful of Rubles (Ithaca: Cornell University Press, 2000);
Treisman, "Dollars and Democratization."
44. These sources were supplemented by reports in The Moscow Times, Dec. 21, 1999, ISI Emerging
Markets; Nezavisimaia Gazeta, Oct. 19, 1999, ISI Emerging Markets; and anonymous regional analysts in
personal communication with the author.
45. In the author's judgment. A separate category for business representatives was included as a control.
46. If there is any bias resulting from this coding process, it is likely to be in the direction of underesti
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions
mating the impact of politicized financial-industrial groups. There are three likely reasons for overlooking a politicized financial-industrial group candidate. The politicized financial-industrial group was very effective in keeping its secret; support was not large; and the district is remote. Candidates of the first
type can be expected to be more effective than average since the reason for secrecy is that public suspi cion of corruption is seen to be a negative. Instances of the second type do not affect the argument since it
concentrates theoretically on cases of major support. As for the third, anecdotal evidence suggests that
politicized financial-industrial groups operating in remote areas are more effective than elsewhere
because there is less local money to put up resistance and because, if they do identify themselves, they can credibly promise to increase the region's wealth.
47. Nineteen of the 2,226 candidates are dropped in this study because some values are missing. 48. On the importance of personal candidate resources, see Smyth and Moser.
49. PFIG is the abbreviation for politicized financial-industrial group. 50. On tobit and why the coefficients it estimates can, with this particular kind of data, be read the
same way as OLS coefficients, see Lee Sigelman and Langche Zeng, Political Analysis, 8 (Spring 2000), 167-82.
51. These zero values also rule out the use of log ratios as done above.
52. Naturally, since governors' candidates are party nominees in the other 104 districts, the shares of
the vote for "governors' candidates" and for "party nominees" will vary together there. Only nineteen
governors consistently backed the same party in such instances, though. When joint candidate totals are
not counted as evidence of party strength but only of governors' strength, the correlation of course
remains strongly negative. The very important question why governors support party candidates in some
regions but not others is beyond the scope of this article but is treated by the author elsewhere. This article
is interested in what happens in the majority of cases, where governors do not back party candidates.
53. Aldrich; Richard P. McCormick, The Second American Party System (Chapel Hill: University of
North Carolina Press, 1966); Myron Weiner, Party-Building in a New Nation (Chicago: University of
Chicago Press, 1967). 54. Jonathan N. Katz and Gary King, "A Statistical Model for Multiparty Electoral Data," American
Political Science Review, 93 (March 1999), 15-32; Michael Tomz, Joshua Tucker, and Jason Wittenberg, "An Easy and Accurate Regression Model for Multiparty Electoral Data," Political Analysis, 10 (Winter
2002), 66-83.
55. See the StataJ software manual, [U] 23.11, "obtaining robust variance estimates." The command
used is cluster (). The cost of this method is a small loss of efficiency, expected to be negligible in this
case given the large number of observations examined.
56. Gary King, Michael Tomz, and Jason Wittenberg, "Making the Most of Statistical Analyses," American Journal of Political Science, 44 (April 2000), 347-61; Michael Tomz, Jason Wittenberg, and
Gary King, CLARIFY: Software for Interpreting and Presenting Statistical Results, Version 2.1, Stanford
University, University of Wisconsin, and Harvard University, 2001, available at http://gking.harvard.edu.
This content downloaded from 202.120.82.40 on Thu, 10 Apr 2014 12:18:14 PMAll use subject to JSTOR Terms and Conditions