Insider-Outsider Politics in Industrialized Democracies The Challenge To Social Democratic Parties

Insider-Outsider Politics in Industrialized Democracies: The Challenge to Social Democratic
Parties
Author(s): David Rueda
Source: The American Political Science Review, Vol. 99, No. 1 (Feb., 2005), pp. 61-74
Published by: American Political Science Association
Stable URL: http://www.jstor.org/stable/30038919
Accessed: 22-11-2015 04:58 UTC
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].
American Political Science Association and Cambridge University Press are collaborating with JSTOR to digitize, preserve
and extend access to The American Political Science Review.
http://www.jstor.org
This content downloaded from 202.94.83.110 on Sun, 22 Nov 2015 04:58:50 UTC
All use subject to JSTOR Terms and Conditions
American Political Science Review Vol. 99, No. ~ February2005
Insider-OutsiderPoliticsin IndustrializedDemocracies:
TheChallengeto SocialDemocraticParties
DAVID RUEDA University of Oxford
In much of the political economy literature,social democratic governments are assumed to defend
the interests of labor. The main thrust of this article is that labor is divided into those with secure
employment (insiders) and those without (outsiders). I argue that the goals of social democratic
parties are often best served by pursuing policies that benefit insiders while ignoring the interests of
outsiders. I analyze Eurobarometer data and annual macrodata from 16 OECD countries from 1973 to
1995. I explore the question of whether strategies prevalent in the golden age of social democracy have
been neglected and Left parties have abandoned the goal of providing equality and security to the most
vulnerable sectors of the labor market. By combining research on political economy, institutions, and
political behavior, my analysis demonstrates that insider-outsider politics are fundamental to a fuller
explanation of government partisanship, policy-making, and social democracy since the 1970s.
Comparative political economists generally agree that the objectivesof social democraticgovernments
that social democratic parties are the defend- are best served by pursuingpolicies that ignore the
ers of labor. The persistence of widespread un- interestsof outsiders.
employment witnessed under social democratic Disaggregatinglaborinto insidersandoutsiderspro-
governments since the early 1970s, however, power- motes the explorationof three topics of importanceto
fully conflicts with this assumption. Moreover, the pop- the comparativepoliticaleconomyliterature.The first
ular press has reported with increasing frequency that one has to do with the transformationin partystrate-
the distinctiveness of some of the economic policies gies resultingfromnew voterdemandsin industrialized
once championed by social democratic and conser- democracies.Whilethe relevanceof otherfactors(like
vative parties has been lost. How can these seeming lower economic growth, demographicor production
anomalies be explained? The answer, I argue, lies in changes, the emergence of post-Fordism,increasing
challenging the notion that social democratic govern- internationalization,and competitionfrom industrial-
ments represent the interests of labor. izing countries) has been recognizedfor some time,
The traditional conception of social democratic my analysis makes clear the significanceof insider-
policy-making rests on the assumption that labor is outsiderpreferencesas a determinantof government
affected disproportionately by unemployment. But in policy.The secondtopic is relatedto the verynatureof
the following pages I argue that labor is divided into socialdemocracy.Ourassumptionsaboutthe strategies
two segments: those with secure employment (insiders) of leftist parties have not changedsubstantiallysince
and those without (outsiders). Since the early 1970s, in- the golden age of social democracy(whenequality,so-
siders have become insulated from unemployment. Not cial protection,and economic growthwere perceived
only do they enjoy high levels of protection, they also as compatible).My analysisquestions these assump-
benefit from the fact that outsiders act as a buffer bear- tions and providesa fuller understandingof the limi-
ing the brunt of fluctuations in the business cycle. In tations and opportunitiesfaced by social democrats
response to the increasingly significant differences be- in the post-oil crisesera. The finaltopic concernswhat
tween insiders and outsiders, social democratic govern- the goals of social democracyshould be. This arti-
ments have transformed their policy goals. The anoma- cle demonstratesthat, in the presence of conflictbe-
lies mentioned above can be explained by considering tween differentgroupswithinlabor,social democratic
governmentsoften do not promote the interests of
the weakest membersof society.The insider-outsider
David Rueda is University Lecturer in the Department of Pol- modelopens the doorto a debateaboutthe desirability
itics and International Relations, University of Oxford and Fel- of this outcome.
low at Merton College. Merton College, Oxford OX1 4JD, UK
([email protected]).
Earlier versions of this article were presented at the 2000 Inter- THE INSIDER-OUTSIDER PARTISANSHIP
national Conference of Europeanists, at the 2002 Annual Meeting MODEL
of the MPSA, and at workshops in Yale University and Cornell
University. I would like to thank Nancy Bermeo, Joseph Foudy, Like much of the literature that explores the rela-
Geoffrey Garrett, Burcu Gezgor, Torben Iversen, Peter
Katzenstein, Lane Kenworthy, Anirudh Krishna, Peter Lange, tionshipbetween partisangovernmentand economic
Michael McDonald, Herman Schwartz, Sidney Tarrow, Christopher policy,I understandpoliticalpartiesto have electoral
Way,MichaelWallerstein,BruceWestern,andHeidi Young,as well objectivesas well as commitmentsto ideology and to
as threeanonymousreviewers,andLee Sigelmanfortheircomments historicallymeaningful groups of voters. As Powell
and suggestions. I am especially grateful to Christopher Anderson,
WalterMebane,andJonasPontussonfor theirconstructivecriticism (1982) has pointed out, the existenceof a relationship
and encouragement.For their financialsupportof this project,I between "strong,continuingexpectationsabout par-
thank the Social Science ResearchCouncil,the FondazioneLuigi ties and the interestsof social groupsnot only creates
Einaudi, Cornell University, and SUNY Binghamton. easily identifiablechoices for citizens,it also makes it
61
Insider-OutsiderPolitics in IndustrializedDemocracies February2005
easier for parties to seek out their probablesupport- modelthereforeis thatsocialdemocraticgovernmentis

ers and mobilizethem at election time"(116). History associatedwithhigherlevels of employmentprotection
and ideology,however, are not enough. Elections in- legislationbut not of ALMPsor PLMPs.
evitablyrevolve aroundissues (like employmentpro- Dividing labor into insidersand outsidersalso has
tection or labor market policies) because issues give implicationsfor the strategies of conservative gov-
political meaning to partisanattachmentsand social ernments.Like many other authors,I consider con-
divisions(Dalton 2002, 195). servative parties to depend on a core constituency
Unlike most of the comparativepolitical economy that consistsof upscale groups (employers,the upper
literature,I do not conceptualizelabor as a homoge- middle-class,and the business and financialcommu-
neous politicalactor.I sharean interestin disaggregat- nity).Paradoxically, the insider-outsidermodelimplies
ing laborwith some recentworkson the determinants that, in some cases,conservativegovernmentsmay be
of party strategies and individualpolicy preferences able to pursue labor market policies that are more
(e.g., Iversen and Soskice 2001 and Kitschelt 1994, attractiveto outsidersthan those promotedby social
1999). My analysisis based on two propositions:that democrats.Having the upscale groups as their core
laboris dividedinto insidersandoutsidersandthatthe constituencymakesit difficultfor conservativeparties
interests of insiders and outsidersare fundamentally to promote the interests of insiders.But ignoringin-
different.1 siders allowsconservativesto engage in some policies
I define insiders as those workerswith highly pro- unavailableto social democrats.As mentionedabove,
tected jobs. They are sufficientlyprotectednot to feel outsidersfavor lower levels of insiderjob protection
greatly threatened by high levels of unemployment. legislation.By reducinginsiderjob protection,conser-
Outsiders,on the otherhand,are eitherunemployedor vative parties may attractsome outsiderswhile rein-
holdjobs characterizedby low salariesandlow levels of forcingthe supportof theircore constituency(upscale
protection,employmentrights,benefits,and social se- groupswhowantflexiblehiringandfiring).Whilelower
curityprivileges.The interestsof these two groupsare employmentprotection is favored by both outsiders
fundamentallydifferent because insiders care about andmembersof the upscalegroups,this is not the case
their own job securitymuch more than about the un- regardinglabor market policies. Higher levels of ac-
employmentof outsidersand outsiderscare aboutun- tive andpassivelabormarketpoliciesrepresenthigher
employmentand job precariousnessmuch more than taxes and a more intrusiverole for governmentin the
about the employmentprotectionof insiders. economy.Because of this, upscale groups (and there-
While dividinglabor into insidersand outsidershas fore conservativegovernments)are not interestedin
some precedentsin both the economics and the po- the promotionof labormarketpolicies.
liticalscience literature,integratingthis divisioninto a My model predicts,then, the followingpartisandif-
coherentconceptionof partisanshipandpolicy-making ferencesregardingeconomicpolicies:(1) partisanship
representsa completely new endeavor.It is my con- will significantlyaffect pro-insiderpolicies-I expect
tention that social democraticpartieshave strong in- social democraticgovernmentsto be associatedwith
centives to consider insiders their core constituency. higherlevels (andconservativeones with lowerlevels)
There are historical and ideological reasons for this of employmentprotection;and (2) partisanshipwill
but there is also the fact that the other group within not significantlyaffectpro-outsiderlabormarketpoli-
labor, outsiders,tends to be less politicallyactive and cies (neither social democraticnor conservativegov-
electorally relevant (as well as less economicallyin- ernmentswill promoteALMPsand PLMPs).
dependent) than insiders.I further argue that social It is importantto emphasizethat the existence of
democraticgovernmentswill side with their core con- two distinctgroupswithinlaboronly affectsthe strate-
stituencywhen faced with the choice between insiders gies of partisan governmentswhen there is a con-
and outsiders.Insidersare benefited by higher levels flict between insidersand outsiders.The coincidence
of employmentprotectionlegislationwhile, inasmuch of insider and outsidergoals is possible in some pol-
as lower protection facilitates hiring, outsiders are icy areas. For instance, some parts of the welfare
not. Consequently,the main policy objectiveof social state (like health care or education) may benefit in-
democraticpartieswill be the continuationor increase siders and outsidersequally.Insofar as both insiders
of insiderjob security.Higher levels of labor market and outsiders need these welfare services, they will
policy, on the other hand, benefit outsiders,but not join in support of governmentsthat promote them.
insiders.Both active labor market policies (ALMPs) In this respect, the insider-outsidermodel does not
and passive labor market policies (PLMPs)have the modify the conventionalassessmentof social demo-
potential to benefit insiders,but more directly,they cratic governments as prolabor (insiders plus out-
mean higher taxes and low-wage competition (more siders). The argumentspresentedin this paper, how-
on this below). The implicationof my insider-outsider ever, integrate this conventional assessment into a
more general and more accurateexplanationof the
politicalfactorsinfluencingpartisangovernment.Put
1 There are two frameworks that inspire the model that I propose: moreboldly,althoughboth frameworkspredictsimilar
work on dual labor markets (such as Berger and Piore 1980 and outcomes for some social policies, only the insider-
Doeringer and Piore 1971) and the economic insider-outsider ap-
proach emphasizing the differences between the employed and
outsider model provides an explanation of the rea-
the unemployed(see, for example,Blanchardand Summers1986, sons why partisanshipmattersto some policiesbut not
LindbeckandSnower1988,and Saint-Paul1996). others.
62
American Political Science Review Vol. 99, No. 1
INSIDERS, OUTSIDERS, AND THE INDIVIDUAL PREFERENCES IN THE

TRADITIONAL VIEW OF PARTISANSHIP INSIDER-OUTSIDER MODEL
The main approach to the relationship between polit- My argument is fundamentally concerned with the rela-
ical parties and policy in comparative political econ- tionship between government partisanship and policy,
omy can be categorized as the traditional partisan- and consequently, this is the focus of the quantitative
ship school. Its authors-Alt (1985) and Hibbs (1977) exploration I develop in the next section. First, how-
being the most cited examples-believe that social ever, I provide some survey data that preliminarily sup-
democratic governments will promote the interests of port my expectations about the preferences of insiders,
labor, while conservative ones will satisfy the demands outsiders, and upscale groups. I should emphasize the
of upscale groups. Labor is assumed to be dispropor- illustrative nature of this section's analysis. I am not
tionately affected by unemployment, and as a conse- presenting a systematic test but rather some initial ev-
quence, social democratic governments are expected to idence to demonstrate the plausibility of my model's
design economic policies that promote employment. assumptions about individual preferences.3
Inflation is assumed to disproportionately influence The importance of a permanent job in my definition
upscale groups so conservative governments are ex- of insiders (as opposed to fixed-term or temporary con-
pected to promote policies that reduce price increases. tracts) and the need for questions related both to policy
The model proposed in this article should be con- preferences and to labor market status limit the data
sidered a transformation of the traditional partisan I use to one survey: Eurobarometer 44.3 (February-
approach. I agree with the traditional partisanship au- April 1996).4 Although this Eurobarometer survey
thors in considering parties to have economic goals provides only a snapshot of individual preferences, it
fundamentally related to those of their core constituen- does allow me to develop insider-outsider codings that
cies. My insider-outsider analysis departs from their closely address my claims.
framework in its identification of the electorates that I define insiders as employed full-time with a perma-
parties are interested in attracting. The difference re- nent job or as those with part-time or fixed-term jobs
sults from a disagreement about how unemployment who do not want a full-time or permanent job. This
affects labor. group includes individuals with permanent contracts
It is highly misleading to categorize all labor as vul- (defined as not having a time limit). Outsiders are then
nerable to unemployment. Since the early 1970s, two defined as those who are unemployed, employed full-
factors have substantially decreased unemployment's time in fixed-term and temporary jobs (unless they do
threat to labor. First, because of the growth and sta- not want a permanent job), employed part time (un-
bility of the late 1960s as well as the social unrest less they do not want a full-time job), and studying.
and union activism that characterized the early 1970s, Students are included in the outsider category both
firms accepted highly restrictive tenure and severance because they have no certainties about their future
pay arrangements (see Bentolila and Bertola 1990 and employment (even those who hope to become insid-
Blanchard et al. 1986). As a consequence, a consid- ers or upscale managers can end up unemployed) and
erable proportion of labor became significantly insu- because in some cases they may have extended their
lated from unemployment. Second, as "insiderness" education because of difficulties entering the labor
emerged, so did "outsiderness." The unemployment market.5 The upscale group category, finally, contains
growth experienced by many OECD countries dur- those individuals who are self-employed (profession-
ing the post-Oil Shock crises contributed to the in- als, owners of shops, business owners, and managers)
crease in outsiders. But labor supply shocks caused by as well as employed managers.
larger numbers of women entering the labor force and The insider-outsider partisanship approach rests on
a general intensification of international competition some assumptions about the preferences of insiders,
and working time flexibility2 are also part of the story. outsiders, and upscale groups. These preferences are
One of the consequences of these developments is a summarized in Figure 1.
dramatic increase in part-time work and temporary There are two dimensions represented in Figure 1:
contracts. The great majority of these jobs, however, labor market policy and employment protection. I have
pay poorly, are concentrated in low-skilled activities, placed the three groups on these dimensions accord-
and possess minimal rights and benefits. More impor- ing to the preferences specified in the model. On the
tantly, the precariously employed and the unemployed first dimension, insiders (who enjoy a high degree of
are the main group to suffer the consequences of eco- job protection) are considered to be less affected by
nomic fluctuations (being hired in good times and laid
off in downturns).
To assume that unemployment disproportionately 3 A more systematic analysis can be found in Rueda 2001 and Rueda,
harms labor as a whole and that social democratic forthcoming and is the focus of ongoing research.
governments therefore need to focus on the problem 4 The sample includes the following countries: Austria, Belgium,
Denmark, Finland, France, West and East Germany, Great Britain,
of unemployment is clearly inaccurate when analyzing Greece, Ireland, Italy, Luxembourg, the Netherlands, Northern
the post-1973 period. Ireland, Norway, Portugal, Spain, and Sweden.
5 In Rueda 2001, I analyze whether the preferences of groups within
the three categories (insiders, outsiders, and upscale) are homo-
geneous. My results show that students share the preferences of
2 See, for example, Dore 1994. outsiders.
63
themselves closer to the other side of the spectrum.

FIGURE 1. Theorized Preferences It is clear that lowering employment protection leg-
LaborMarketPolicy Preferences islation directly attacks the interests of insiders. The
preferences of outsiders are justified by their belief that
Upscale Insiders Outsiders
lower employment protection will facilitate their exit
from unemployment and precarious employment. Ac-
cording to many, employment protection "is, in effect,
a tax on work-force adjustments" (OECD 1999, 68),
More
Resources and as such it may inhibit firms from shedding labor
intoPolicy in economic downturns but also from hiring in peri-
ods characterized by good performance. The upscale
groups (especially employers and managers) clearly
EmploymentProtectionPreferences benefit from the flexibility of lower levels of employ-
ment protection.
Upscale Outsiders Insiders To test whether the preferences presented in
Figure 1 are accurate I turn to the Eurobarometer sur-
More
vey. Figures 2a and 2b reflect the responses of insiders,
Resources outsiders, and upscale groups to two questions that
intoPolicy address active labor market concerns. In the first ques-
tion, respondents were asked whether they would tend
to agree or disagree with the following statement: "The
unemployment and less interested in dedicating more government should offer a guarantee of training, or a
resources to labor market policies. Outsiders are most job, to all young people leaving school." Responses that
vulnerable to unemployment and therefore more con- agreed were given a 10 and those that disagreed were
cerned about active and passive labor market policies. given a 0. The x-axis in Figure 2a measures the mean re-
As for the upscale groups' preferences their position is sponse of the three groups. As theorized, outsiders are
justified by their desire to reduce the taxes that pay for strong advocates of guaranteed training and jobs for
these policies and a general inclination to limit the role young people, while both insiders and upscale groups
of government in the economy. believe this is a much lower priority.The expected pref-
There are reasons for insiders to favor higher levels erences are also confirmed when the question moves
of labor market policy. Insiders face some probability away from the relatively abstract issue of job and train-
of losing their jobs (when companies become econom- ing guarantees to a more concrete issue. Respondents
ically unviable, for example) and labor market policies were asked whether they would tend to agree or dis-
can reduce the intensity of job searches by outsiders agree with the following statement: "I would be ready
and therefore reduce competition for wages. But the to pay more tax if I were sure that it would be devoted to
reasons for insiders to oppose labor market policies creating new jobs." Again, agreements were given a 10
are more powerful. An increase in the levels of active and disagreements a 0. The x-axis in Figure 2b measures
or passive labor market policies, after all, represents a the mean response of the three groups. Although this
higher tax burden for insiders. Additionally, some of time the scores are lower, the relative positions of the
these policies may, if successful, promote the entry into groups are almost identical. As expected, outsiders are
employment of individuals who can underbid insiders' clearly in favor of employment promotion measures
wage demands.6 As Saint-Paul (1998) has argued, when even if they imply increasing taxes, while insiders and
insiders feel protected enough not to significantly fear upscale groups display much lower levels of support.
unemployment, lack of support for ALMPs may result The numbers in Figures 2a and 2b are suggestive, but
from the insiders' interest in being sheltered from low an initial assessment of their statistical significance can
wage competition (162).7 help confirm their meaningfulness. I estimate Pearson
On the second dimension, insiders are expected to chi-square statistics to test whether a significant rela-
be strongly in favor of employment protection, while tionship exists between being an insider, outsider, or
upscale groups and outsiders are expected to place upscale individual and holding the specific opinions
about employment promotion contained in the figure.
The null hypothesis in this test is that there is no asso-
6 See Calmfors 1994 and Saint Paul 1998. For the relationship be- ciation. The results suggest (at better than a 99% level
tween the effects of active and passive labor market policies, see of confidence) that a relationship does exist between
Calmfors 1993. For an analysis showing that the effects of ALMPs on insider-outsider-upscale status and employment pro-
labor market competition may be dependent on whether they target
particular individuals, see Calmfors and Lang 1995. Regarding the
motion preferences. This is the case for the numbers in
latter point, the implication for my analysis is related to one of the both figures so the significance of the association does
starting assumptions: that ALMPs target outsiders. not depend on the question.8
7 It is clear that the emergence of employment protection described
in the previous section amplifies insider-outsider differences. The
higher the level of protection, the more important the insider-
outsider conflict, as one group (insiders) increasingly pays the costs of 8 For a more systematic test of the implications of the insider-
ALMPsandPLMPswhilethe other(outsiders)increasinglybenefits outsider model for employment promotion preferences, see Rueda,
fromthem. forthcoming.
64
FIGURE 2a. Employment Preferences
Upscale
Outsiders
Insiders
7.2 7.4 7.6 7.8 8.0 8.2 8.4 8.6

Guaranteed Job and Training Mean (10 = Agreement)
FIGURE 2b. Employment Preferences
Upscale
Outsiders
Insiders
4 4.2 4.4 4.6 4.8 5 5.25 5.4

taxes for Employment Mean (10=Agreement)
FIGURE 2c. Employment Protection Preferences
Upscale
Outsiders
Insiders
5 5.2 5.4 5.6 5.8 6 6.2 6.4 6.6

Mean of "Job Security is Very Important"Question (10 = Agreement)
Note: Data for Figures 2a, 2b, and 2c from Eurobarometer 44.3 (February-April 1996). Weights provided by the Eurobarometer were
used so that the samples were representative of the countries from which they were drawn. Data for Norway are missing.
The numbers in Figures 2a and 2b are a good repre- Turning now to the third dimension in my analy-
sentation of ALMP preferences but they do not reflect sis, Figure 2c depicts the job security preferences of
concerns about the levels of passive labor market poli- insiders, outsiders, and upscale groups. Respondents
cies. This is because of the absence of any question in were asked the following question: "Foryou personally,
the survey addressing PLMPs. The questions asked are how important do you think each of the following is in
either too general (Is social welfare a necessity?) or choosing a job?" Respondents were then given several
related to areas excluded from PLMPs (Should health characteristics that they could rate from very important
care or education be guaranteed?). We can turn, how- to not important at all. Responses that considered a se-
ever, to the analysis of PLMP preferences in Boeri, cure job very important were given a 10 and those that
B6rsch-Supan, and Tabellini 2001, which shows that did not were given a 0. Again, the mean preferences of
the preferences in my model are in fact reasonable. the three groups confirm the hypothesis in Figure 1. As
They find that the individuals I define as outsiders expected, insiders are most concerned about job secu-
would be ready to accept higher costs (i.e., taxes) rity, while outsiders and upscale groups are much less
in return for more unemployment insurance. Insiders likely to consider job security very important. Whether
and upscale groups do not seem to share these pre- an individual is an insider, an outsider, or a member
ferences. of the upscale groups does prove to be a statistically
65
significant determinant of his/her attitude toward job fair dismissal, lay-offs for economic reasons, severance
protection. A Pearson chi-square test shows that the payments, minimum notice periods, administrative au-
association between these two variables is significant thorization for dismissals and prior discussion with la-
at better than the 99% level of confidence. bor representatives" (OECD 1994, 69). The previous
section made clear the reasons behind the preferences
of insiders, outsiders, and upscale groups. Because in-
THE EFFECTS OF PARTISANSHIP
ON POLICY siders and upscale groups have opposing interests and
they are the core constituencies of social democratic
The data presented in the previous section illustrate and conservative parties, the insider-outsider model
that my partisanship model's expectations about indi- implies the existence of marked partisan difference re-
vidual preferences are reasonable. In the next pages I garding job protection.
explore whether parties do in fact develop policies in I use two different measures of employment protec-
line with these preferences. I focus on two measures: tion legislation. The first one is the mean for the 1980s
labor market policies and employment protection leg- and the 1990s of the OECD's overall protection against
islation. dismissals index. The index is constructed by averaging
the scores obtained by each country in three categories:
"procedural inconveniences which the employer faces
The Dependent Variables when trying to dismiss employees; notice and sever-
Labor Market Policies. PLMPs provide unemploy- ance pay provisions; and prevailing standards of and
ment compensation, whereas active ones are aimed penalties for unfair dismissal" (OECD 1999, 54). Con-
at reducing unemployment by shaping the supply, de- ceptually, this index is ideal for testing my hypotheses.
mand, and mobility of labor. The OECD data used in It suffers, however, from the important practical limita-
my statistical analysis include unemployment benefits tion of being available only as a summary value for the
as the main component of PLMPs. The ALMP measure 1980s and the 1990s. I want to use yearly data that allow
encompasses the following five areas: (1) public em- a significant increase in the number of observations
ployment services and administration, (2) labor market and in the complexity of the estimated models. For this
training, (3) youth measures, (4) subsidized employ- reason, I also use a measure of the number of months of
ment, and (5) measures for the disabled. severance pay a blue-collar worker with 10 years of ser-
While PLMPs (as an important element of the wel- vice receives upon termination without cause.9 Cause
fare state) have received quite a lot of attention in is illustratively explained by Lazear (1990) as generally
the comparative political economy literature, the re- meaning "for reasons having to do with the worker's
lationship between ALMPs and government partisan- own shortcomings, and it must be extreme. A reading
ship has been underexamined. Economists have been of the rules suggests that in most countries, dismissal
concerned mostly about their effects on employment with cause requires the kind of evidence necessary to
and have generally ignored the role of partisanship withdraw an American academic's tenure" (708).
in promoting different levels of ALMP. In political
science, some authors have explored the relationship The Explanatory Variable:Government
between partisanship and ALMPs but some important Partisanship
questions have been left unanswered. The effects of
divisions within labor, in particular, have not been ana- The government partisanship measures used in my
lyzed in detail since the starting point for most political analysis attempt to capture the ideological position of
scientists is to consider ALMPs one more measure that governments in relation to a left-right continuum. Two
social democratic parties will employ to benefit labor variables are needed for the construction of these mea-
(see, e.g., Boix 1998 and Janoski 1994, 1990). sures: one that reflects the presence of parties in gov-
As explained in more detail in the previous section, ernment and another that measures their ideological
outsiders are considered the main beneficiaries of labor characteristics. The operationalization of the first vari-
market policies in the model proposed in this article. able has been relatively straightforward in the compar-
Before the widespread adoption of employment pro- ative politics literature. But important questions sur-
tection in the early 1970s, the interests of insiders and round the measurement of party ideological positions.
outsiders regarding labor market policy were closely Assessments of left-right party positions are based
aligned. Insider vulnerability to unemployment was on two sources: expert opinions and party manifestos.
higher and social democratic governments could pro- These two measures imply a different set of complica-
mote labor market policies that favored outsiders. The tions. Expert opinions are produced from surveys that
emergence of employment protection causes the inter- are administered rarely and that may be interpreted
ests of insiders and outsiders to diverge. As ALMPs and differently in different national contexts (Gabel and
PLMPs increasingly become policies that insiders pay Huber 2000). Data extracted from party manifestos, on
the costs of while outsiders receive the benefits from, the other hand, can be criticized for being a reflection
social democratic governments become less likely to of what parties say to win elections, and not necessarily
promote them and partisanship becomes insignificant. of what they will do once they have won them.
Employment Protection Legislation. Employment 9 For an analysis of the close relationship between severance pay and
protection legislation affects "the rules governing un- overall employment protection, see OECD 1994.
66
In this paper, I try to avoid some of these compli- right or between the Christian democrats and the con-
cations by using two measures of government parti- servatives. Considering the two policies emphasized in
sanship. The main results reported below use parti- the previous pages, this seems uncontroversial. The
san cabinet composition as measured by Tom Cusack free-market philosophy of liberals places them close
(1997).10The reason for this choice is mostly practical. to conservatives regarding employment policy and job
Many analyses dealing with the effects of government protection.13
partisanship on policy use this measure and I want
my results to be easily compared to those obtained
by other authors. After presenting results using cabi-
Other Variables
net partisanship, I verify the conclusions of my initial Labor Market Institutions. Both because of their di-
analysis with a measure of government partisanship rect involvement in industrial relations (negotiations
that uses party manifestos to assess a party's left-right covering work regulations and wages being the most
position.11For the construction of this government par- clear examples) and because of their capacity to in-
tisanship measure, a party's average left-right position fluence political parties, the behavior of unions is a
is multiplied by its cabinet weight. In this case, the relevant factor in a government's decisions over policy.
cabinet weight is the proportion of parliamentary seats I emphasize two labor-related factors: the centraliza-
that parties in governments posses. This is similar to tion/coordination of wage bargaining and union den-
Cusack's measure because governments tend to "ap- sity.14
portion their cabinet portfolios to parties in simple
proportion to the relative percentage of seats held by International and Financial Openness.i There are
each in the lower house of the legislature" (Powell 2000, two contradictory accounts of the effects of interna-
173). tionalization on partisan politics. First, there is a large
I would like to address a final point with refer- literature suggesting that growing levels of interna-
ence to the government partisanship measures used tional openness and interdependence result in a blur-
in the analysis. As mentioned above, the variables ring of partisan differences caused by the inability of
calculate the ideological position of governments in social democratic parties to produce policies that do
relation to the partisan composition of cabinets. This not conform to market forces (see, e.g., Iversen 1996
means that parties other than the social democratic and and Scharpf 1991). Then there are some authors who ar-
conservative ones influence the weighted partisanship gue either that international forces do not affect some
measure.12This, however, does not affect the insider- partisan differences (like Boix 1998 and Garrett and
outsider partisanship model or the findings described Lange 1991) or that they actually have strengthened
below. Regarding partisan options to the left of the the influence of partisanship on policies and economic
social democrats, communist parties are strongly pro- outcomes (Garrett 1998). The results presented in the
insider. Most communist parties in Europe rely even following pages do not address whether international
more on the support of insiders (through both their dependence limits the autonomy of governments. In-
votes and their participation in unions) than social stead I look at the great variance of economic policy
democratic ones and they have been just as reluctant within the sample and try to assess the factors that
to integrate the interests of outsiders. The government are responsible for it. My hypothesis is that, once the
participation of communist parties in the sample that I influence of internationalization is controlled for, par-
analyze is limited enough, in any case, not to affect the tisanship will account for this variance only in the case
conclusions made about the influence of partisanship of employment protection policies.
over policy. Government Debt.16 I introduce government debt
When considering Christian democratic parties (usu- into the analysis as a measure of the availability of
ally placed in the moderate right), the implication of
my analysis is that these parties promote policies that
can be placed in between social democracy and conser- 13 The differences between liberals and conservatives (in individ-
vatism for the dimensions I focus on. This seems real- ual freedom issues such as abortion and divorce) pertain to policy
istic. Liberal parties are placed either in the moderate dimensions not discussed in this paper.
14 See Iversen 1999 for a complete specification of the centraliza-
tion/coordination variable. To capture the inertia associated with
institutional change, I use a moving average of the yearly values
10 Higher figures signify more conservative government. Cusack (present and previous four years). It should also be noted that values
groups parties into five families, multiplies each family's share of for the last two years in the time series were extrapolated. The union
cabinet portfolios by its weight, and sums the products. See Cusack density measure used in the regressions represents employed union
1997 for further details. members as a percentage of the employed labor force. The pre-1990
11 This variable relies on party programs for the codification of pol- figures were taken from Visser 1996; post-1990 figures were provided
icy emphases and it was produced by the Comparative Manifestos by Bernhard Ebbinghaus (Max-Planck Institute).
Project. Source for all countries but Japan: McDonald and Mendes 15 International openness is measured as imports plus exports as
2001. Data for Japan were created by the author. Sources: Com- a percentage of the GDP Source: OECD electronic database and
parative Manifestos Project left-right party index and Woldendorp, OECD Historical Statistics 1960-95. Financial openness is measured
Keman, and Budge 2000. Given the variance of the ideology measure, as the sum of several indexes for financial restrictions. For details,
the 1973-95 average is used in the analysis. see Armingeon, Beyeler, and Menegale 2002.
12 Although I refer to social democratic and conservative parties in 16 Source: Franzese 1998. Given the possibility of endogeneity
the Results, it would be more accurate to refer to the partisan options (higher levels of policy causing higher debt), I use a one-year lag
as left and right. for this variable.
67
resources affecting a government's choices. One widely the recent comparative political economy literature, I
accepted interpretation of the policy changes of the estimate panel-corrected standard errors.
early 1980s, for example, is that many governments I therefore regress the dependent variables on a
had reached unsustainable levels of public debt (see one-year lag of the dependent variable, independent
Schwartz 1994). I use the level of consolidated central variables, and period dummies. That is, I estimate the
government debt as a percentage of GDP as the indi- following equation:
cator and include it only in the regressions with ALMP
and PLMP as the dependent variables (since any re- Yit = XWi,t-1 + i fkXkit + T + Uit,
lationship between resource limitations and legislation k
affecting employment protection seems unlikely). where i refers to the cross-sectional units, t to the time
Unemployment.17 Some authors have argued that units, k to the number of independent variables, r to
policies simply result from increasing needs-whether the time period intercepts, f to the slopes of the ex-
demographic, economic, or other (see, e.g., Wilensky planatory variables, yi,t-1 to the lagged dependent vari-
1975). I engage these arguments by controlling for the able, and the x's are the independent variables.
effects of unemployment. It is important that my results The only regression not run according to these spec-
are not affected by needs. I want to be able to conclude, ifications is the one that uses two time averages of the
for example, that social democratic governments do protection index per country instead of yearly data.
not promote high levels of ALMPs regardless of the Given the small n, in this case I did not estimate panel-
size of unemployment. Unemployment also acts as a corrected standard errors and simply ran a bivariate
proxy for the number of outsiders in an economy. It OLS regression on the variable of interest.
is essential for this article's conclusions that the re-
sults control for the size of the outsider group, since RESULTS
the insider-outsider partisanship model maintains that
social democratic governments will appeal to insiders Table 1 provides the estimates for the determinants
even when outsiders are numerous. of labor market policies. Here, it is most important to
point out that, as hypothesized, cabinet partisanship is
GDP Growth.1s Most analyses of economic policy not significant as an influence on the levels of ALMP
include a measure of economic growth. This is particu- or PLMP. The table clearly shows that whether a gov-
larly important here because of the need to control for ernment is social democratic or conservative makes no
the effects of growth on the behavior of governments. difference to the levels of ALMPs promoted. It is also
important to emphasize that these results contradict
METHODOLOGY the conventional wisdom, and much of the existing
literature, regarding the influence of partisanship on
I use annual data from 16 countries from 1973 to 1995 active policies (see, e.g., Boix 1998, Janoski 1990, 1994,
and present ordinary least squares (OLS) results.19The and Swank and Martin 2001). This article represents
pooled data significantly increase the number of obser- a considerable improvement on most of those done in
vations and therefore allow me to test more complex the past because of the number of cases and factors
causal models. I also include a lag of the dependent included in the analysis.21
variable among the regressors. Since the chosen de- Table 1 also reveals that government partisanship is
pendent variables exhibit noticeable time stability, the an insignificant determinant of passive labor market
introduction of a lagged dependent variable provides policies. As hypothesized, social democratic govern-
a better dynamic model in which the influence of the ment is not associated with greater levels of PLMP.
previous year's values is explicitly assessed. To further These results reinforce the conclusions of the regres-
capture cyclical factors I include time period dummies sion analyzing ALMPs: when insiders do not share the
in the regressions.20 goals of outsiders, social democratic governments do
Beck and Katz (1995,1996) have proposed a method not promote pro-outsider policies. Other authors have
that produces consistent standard errors estimates observed results similar to those presented in Table 1.
in the presence of panel heteroscedastic errors. Since In a very direct fashion, they are confirmed by the anal-
their recommendations have been widely followed in ysis in Moene and Wallerstein (2003).22 They analyze,
among other things, the effects of Right government
17 The measure used is the standardized unemployment rate for all
countriesbut Austria,Denmark,and Switzerland.For these three 21Most existing analyses rely on a very limited number of
countries,I use regularunemploymentrates.Sources:OECD His- observations-Boix's (1998) regressions,for example,range from
toricalStatistics1960-95and1960-97.Giventhe possibilityof endo- 18 to 21 observations(75-79), and Janoski's(1994) from 36 to 38
geneity,I use a one-yearlag. (70-78). Thisseverelylimitsthe possibilityof systematicallytesting
18 GDP growth is measuredas year-to-yearpercentagechanges. alternativehypotheseswhile simultaneouslycontrollingfor other
Source:OECD electronicdatabaseand OECD HistoricalStatistics relevantfactors(theseregressionstypicallyhaveno morethanthree
1960-97. variables), which in turn introduces great caveats into the significance
19 The countries are Australia, Austria, Belgium, Canada, Denmark, of the results.Othersdo not systematicallyassessthe validityof their
Finland, France, Germany, Italy, Japan, the Netherlands, Norway, claimsacrosscountriesand throughtime, whichlimitsthe general-
Sweden,Switzerland,the United Kingdom,andthe United States. izabilityof theirconclusions(see, for example,the country-specific
20The periods are 1973-79, 1980-84, 1985-89, and 1991-95. The analysis in Janoski 1990 and 1994).
excludedreferenceyearis 1990. 22 And also by the analysis of the United Kingdom in King 1995.
68
TABLE 1. The Determinants of Labor Market influencing PLMPs.24 However, as hypothesized, the
sign of the coefficient indicates that social democratic
Policy, 1980-95
government is not associated with greater levels of
ALMPs PLMPs PLMP The sign seems to suggest that Right govern-
Constant .037 .446
ment (rather than Left government) is in fact associ-
(.119) (.186)
.378 .009 ated with more generous PLMPs.
Laggeddependentvariable .921 .939 A number of authors have argued that governments
(.048) (.025) dominated by Christian democratic parties promote
<.001 <.001 generous welfare state policies (see, e.g., Hicks and
CabinetPartisanship .015 .013 Swank 1992 and Huber and Stephens 2001). Since this
(.024) (.022) could be the explanation for the results using cabi-
.268 .275 net partisanship, I run a regression following the same
UnionDensity .001 .001 specifications but with Christian democratic share of
(.001) (.001) cabinet seats instead of government partisanship.25
.044 .288 Christian democracy turns out to be insignificant as a
BargainingCentralization .132 .150 determinant of PLMPs. This means that the insignif-
(.131) (.145) icance of partisanship effects observed in Tables 1
.157 .152 and 2 is not caused by ignoring Christian democratic
International
Openness .001 .001
effects.
(.000) (.001)
.083 .167 The second column in Table 2 substitutes the ex-
FinancialOpenness -.005 -.018 pert opinion-based measure of partisanship for one
(.008) (.013) that uses party manifesto data. While the results in
.262 .083 Table 1 are confirmed regarding ALMPs, the use of this
Lag of GovernmentDebt -.099 -.145 alternative measure results in a significant partisan-
(.046) (.070) ship effect on PLMPs. Although the sign of the coef-
.016 .019 ficient indicates that more social democratic govern-
Lag of Standardized .002 .005 ments promote higher levels of passive labor market
UnemploymentRate (.005) (.008) policies, the size of the coefficient (-.002) suggests
.375 .281
GDP Growth -.019 -.099 that the substantial effect of this variable is extremely
(.006) (.009) small.
.001 <.001 Some authors have argued that strong labor move-
N 171 209 ments allow leftist governments to promote low un-
R2 .93 employment policies (see Alvarez, Garrett, and Lange
.97
Source: OECD Social ExpendituresDatabase 2000, except 1991 and Garrett 1998). The lack of partisanship effects
Switzerlandand Austria(onlyfor PLMP),OECD Employment in Table 2 could result from misspecifying the rela-
Outlook. tionship between labor market institutions and gov-
Note:ALMPand PLMPspending as % of GDP.Highervalues ernment partisanship. I therefore test the Left labor
of the CabinetPartisanshipvariablemean more conservative hypothesis in the third column in Table 2. I substitute
governments.Numbersinboldare estimatedcoefficients;num-
bers in parentheses are theirpanel-correctedstandarderrors; cabinet partisanship for Garrett's (1998) variable cap-
numbersin italicsare p-values fromone-sided t-tests. Period turing the interaction between social democratic gov-
dummyestimates are not reported(availableupon request). ernment and the power of labor. Since Garrett's Left
labor power index is not available after 1990 (or for
Switzerland and Australia), I recreate the index by
on the major categories of welfare state spending and adding together social democratic government, the
do not find a partisan effect on an aggregate measure level of centralization/coordination of wage bargain-
of ALMPs and unemployment benefits. ing, and union density. This is a good proxy for Garrett's
In Table 2, I estimate some alternative models to index since it is highly correlated to it (the correlation
confirm the results presented in Table 1 (only the vari- coefficient is .91). Following Garrett, interactions with
able of interest is reported). The first column repre- international and financial openness are introduced,
sents the same regressions used for the main results while union density and bargaining centralization are
with the inclusion of country fixed effects. Following
Hsiao (1986), I include country dummies to control
for those influences that are country specific and that 24 Nickell (1981) demonstrates that, with short panel data, OLS esti-
could affect the accurate estimation of the variables of mation of models with lagged dependent variables and fixed effects
interest (country-specific omitted variables).23Cabinet produces biased coefficients. I therefore check these fixed-effects
partisanship is an insignificant determinant of ALMPs results by using a two-stage instrumental variable procedure similar
to the one explained in Rueda and Pontusson 2000. The findings
(as was the case with Table 1) but a significant variable regarding ALMPs are confirmed but government partisanship loses
significance as a determinant of PLMPs in the instrumental variable
model (confirming the main results in Table 1).
25 Results not reported but available from the author. Source for
23 Dummies for all countries are present in the regressions with fixed Christian democratic share of cabinet seats: Duane Swank, Political
effects. I ran these regressions without a constant. Science Department, Marquette University.
69
TABLE 2. The Determinants of Labor Market Policies, Alternative Models

CabinetPartisanship, GovernmentPartisanship,
ModelwithCountry Measuredwith LeftLaborPower
FixedEffects PartyManifestoData Test
ALMPs PLMPs ALMPs PLMPs ALMPs PLMPs

.035 .047 -.001 -.002 .000 -.002
(.029) (.030) (.001) (.001) (.003) (.003)
.116 .061 .212 .047 .452 .221
Note: See Table 1 for details about the regressions. Highervalues of the Cabinet Partisanshipvariablemean
more conservativegovernments.Numbersin bold are estimatedcoefficients;numbersin parentheses are their
panel-correctedstandarderrors;numbersin italicsare p-valuesfromone-sided t-tests.Allotherestimates are not
reported(availableupon request).
eliminated from the model. When analyzing both

ALMPs and PLMPs, the Left labor power variable is TABLE 3. The Determinants of Employment
insignificant (as expected). The openness interactions Protection, 1973-95
are also insignificant. It seems, therefore, that even Severance OverallEmployment
when the influence of social democratic governments, Pay Protection
wage bargaining, and union strength is put together, Constant .314 3.957
the main observations made about Table 1 stand. (.317) (.841)
The analysis presented in Tables 1 and 2 follows .161 <.001
the standard procedures of recent comparative politi- Lagged dependent .964
cal economy work relying on pooled time series data. variable (.054)
The models, however, ignore the question of whether <.001
the impact of social democratic government is subject CabinetPartisanship -.074 -.661
(.041) (.271)
to a lag. The results presented in the previous pages .037 .008
are therefore vulnerable to the criticism that they do UnionDensity .002
not show a significant government partisanship effect (.002)
simply because they do not capture social democratic .171
influences only discernible after some time. To address Bargaining -.378
this issue, I run the regressions in Table 1, substituting Centralization (.378)
yearly cabinet partisanship for a 2-, 5-, and a 10-year .159
average. The main results hold and government parti- International -.002
sanship was found to be an insignificant determinant Openness (.001)
of ALMPs and PLMPs no matter the lag. .036
FinancialOpenness -.004
If the first two tables make clear the lack of a pro- (.016)
outsider orientation in governments of the Left when .398
examining the determinants of employment protection Lag of Standardized .017
legislation, their pro-insider behavior seems equally Unemployment (.009)
obvious from the analysis in Table 3. Leaving aside Rate .036
the lagged dependent variable, only two factors are GDP Growth .021
significant at better than the 95% level of significance: (.020)
partisanship, international openness, and the previous .145
year's unemployment rate. As expected, social demo- N 310 32
cratic governments are significantly associated with R2 .92 .14
higher severance pay. In other words, social democracy Note: Higher values of the Cabinet Partisanship variable mean
has promoted higher employment protection legisla- more conservative governments. Severance Pay Notes: The
data are the correction of Lazear's figures by Addison, Grosso,
tion (or protected insiders) during the period of time and Teixeira, updated for the 1992-95 period using OECD
under analysis. This is all the more meaningful when (1999). See Lazear 1990 and Addison and Grosso 1997. Num-
we remember that in many OECD countries unem- bers in bold are estimated coefficients; numbers in parenthe-
ployment increased dramatically during this period. ses are their panel-corrected standard errors; numbers in italics
The substantive significance of these findings is per- are p-values from one-sided t-tests. Because of missing data,
Canada is not included in these regressions. Estimates for pe-
haps best illustrated by simulating the effects of a riod dummies are not reported (available upon request). Overall
change in government partisanship while keeping the Employment Protection Notes: Source: OECD (1999), Table 2.2,
other variables constant. A change in the partisanship Panel B. All entries are bivariate OLS estimates. Numbers in
measure from a rightist government to a leftist one bold are estimated coefficients; numbers in parentheses are
their standard errors; numbers in italics are p-values from one-
(from a 4 to a 2 in Cusack's index) would be associated sided t-tests.
with an immediate increase in employment protection
70
those in the 1980s. To explore this issue, I test whether

TABLE 4. The Determinants of Employment the results hold in a cross-sectional design (with only
Protection, Alternative Models 16 observations). I regress employment protection in
CabinetPartisanship, GovernmentPartisanship, the 1990s on each country's cumulative measure of
ModelwithCountry Measuredwith
FixedEffects PartyManifestoData government partisanship from 1964 to 1995. Although
- .076 .000 the significance level of government partisanship di-
(.050) (.002) minishes, the results in Table 2 are confirmed in this
.066 .464 analysis. Social democratic government is still found
Note:See Table3 fordetailsaboutthe regressions.Higherval- to be associated with more insider protection (in spite
ues of the CabinetPartisanshipvariablemean moreconserva- of the small number of observations, the variable is
tive governments.Numbersin bold are estimatedcoefficients; significant at better than the 90% level).28
numbersin parenthesesare theirpanel-correctedstandarder- As was the case with the labor market policy models,
rors;numbersin italicsare p-values fromone-sided t-tests. All the analyses presented in Tables 3 and 4 do not address
otherestimates are not reported(availableupon request).
whether the influence of government partisanship is
subject to a lag. To explore this issue, I follow the
strategy described before. In this case, however, the ob-
equal to .148 of a month of severance pay. The long-
jective is different. In the labor policy analysis, the ex-
term effect of this change in government partisanship
ploration of different lags was meant to check whether
would equal 4.111 months of severance pay.26
the statistical insignificance found was the result of not
I have complemented this analysis with a bivariate
giving government partisanship enough time to affect
regression that takes the OECD's overall employment policy. In the employment protection case we know
protection index as its dependent variable. Since only
that government partisanship has a significant imme-
figures corresponding to the "late 1980s" and "late diate effect but we want to see whether there is also a
1990s" exist, I regress overall employment protection
long-term one. I run the regressions in Table 3 substi-
on cabinet partisanship averages of the 1973-87 and
tuting yearly cabinet partisanship for 2-, 5-, and 10-year
1988-95 periods. The goal, given the limitations in
averages. Government partisanship was found to be a
terms of both the number of observations and the na-
significant determinant of employment protection with
ture of the regressor, is simply to confirm the previous
the 2-, and 5-year lags, but an insignificant one when
severance pay findings. And the results do just that.
the lag was longer. These results suggest that the effects
As in the first severance pay regression, social demo- of government partisanship on employment protection
cratic governments are strongly associated (surpassing
are strongest in the short term.
the 99% significance level) with more protection for
insiders.
Table 4 tests the robustness of the employment pro- CONCLUSION
tection conclusions by exploring some alternatives. The
first column indicates that a model with country fixed Going back to the initial questions that motivated
effects produces similar results to those in Table 3. this article's analysis, the previous pages have demon-
Government partisanship is a significant influence over strated that insider-outsider politics are fundamental
employment protection legislation when we control for to a fuller explanation of government partisanship,
country-specific omitted variables.27This is, however, policy-making, and social democracy since the 1970s. It
not the case in the second column, when the measure of is also clear that recent social democratic governments
partisanship based on expert opinions is substituted for have not promoted some of the policies we would ex-
one that uses party manifesto data. It seems that when pect. The strategies prevalent in the golden age of social
we look into what parties say (and not what they do), democracy have been abandoned and the provision of
there is not a connection between social democratic equality and security to the most vulnerable sectors of
government and protection policy. The fact that social the labor market has been sacrificed to satisfy other
democratic government would decide not to empha- objectives. My evidence suggests that, in the presence
size their pro-insider strategies regarding employment of insider-outsider conflict, there is a strong tempta-
protection (especially in times of high unemployment) tion for social democratic governments to implement
is not necessarily surprising. inegalitarian policies.
Tests for robustness in the regression using the Herbert Kitschelt (1999) has accurately pointed out
OECD's overall employment protection index are that the study of party strategies requires a "bridge
complicated, due to the small number of observations across the familiar divide between students of com-
in the sample. In Table 3, it is assumed that the observa- parative political economy and parties and elections"
tions for each country in the 1990s are independent of (318). The model I have presented attempts to do this
by combining research on political economy, institu-
tions, and political behavior. Political economists often
26 On the assumption that the effects of a one-unit change in a par- assume that the preferences of parties are exogenous
ticular variable persist, the long-term effects of such a change can
be computed by dividing the value of the coefficient for the variable
and stable and that they are mediated by institutional
of interestby one minus the coefficientfor the lagged dependent configurations that produce differentiated political and
variable.
27 Government partisanship, however, becomes insignificant in the
instrumental variable model. 28 Results available from the author.
71
economic outcomes. The analysis in this article, how- democratic parties. Perhaps more importantly, there
ever, focuses on the transformation in party strategies is also the question of what outsiders can do when
that results from new voter demands. I show that eco- ignored by social democratic governments. Starting in
nomic factors affect the preferences of particular elec- the 1970s, most Western democracies have experienced
torates and, as a consequence, the strategies of partisan the emergence of powerful antisystem parties (espe-
governments. cially extreme right ones). This article's arguments sug-
My analysis also provides a useful framework with gest that the losers in the labor market arena (outsiders
which to analyze more recent partisanship debates. who are not the focus of traditional left or right par-
There is an ongoing discussion, in both academic and ties) may have reasons to turn away from mainstream
more general circles, about whether a "third way" has options.
emerged as a comprehensive philosophy uniting so- Before closing, I would like to briefly explore three
cial democratic parties around a coherent set of policy topics that may not have received enough attention in
alternatives. According to some analysts, a third way my analysis: the degree of security enjoyed by insiders,
does indeed exist and it is distinguished by an emphasis the perceived effectiveness of labor market policies,
on employment promotion as a goal and by the use of and the existence of other policy options. The impor-
active labor market policies (see, e.g., Giddens 1998). tance of the first issue cannot be understated. This
The arguments presented in this article emphasize the article's partisanship model, after all, is based on a
importance of taking into consideration insider- fundamental difference in the way insiders and out-
outsider distinctions to understand these policy devel- siders are vulnerable to unemployment. To the extent
opments. Taking this article's framework as our guide, that insiders are protected from unemployment, their
we would speculate that a number of factors could interests will be significantly different from those of
make pro-outsider policies more attractive to social outsiders. Factors that increase insiders' vulnerability
democratic governments. Among them an increase in to unemployment, however, will align their interests
the number of outsiders past a possible critical point, with those of outsiders. In Rueda (forthcoming) I ex-
a weakening of unions, or even a decrease in the plore this question in some detail. Using a case study
level of "insiderness" come to mind. In fact an argu- and an analysis of surveys and macrodata, I show that
ment could be made that in the United Kingdom, it the unemployment vulnerability of insiders is indeed
was the influence of these last two factors that facil- a significant factor affecting the likelihood that social
itated the emergence of Blair's third way. The mod- democratic governments will promote labor market
eration of unions in the late 1990s and the influence policies. A hierarchical analysis of individual prefer-
of Thatcherism are surely factors influencing the poli- ences shows that insiders are significantly more likely
cies of New Labour. Thatcher, Giddens (1998) argues, than outsiders or members of the upscale groups to
"attacked established institutions and elites.... The have high preferences for employment protection (as
Labour Party and its intellectual sympathizers first of shown here in a less systematic manner). Insiders, how-
all responded largely by reaffirming old left views. The ever, experience a drastic increase in their labor mar-
electoral setbacks the party suffered by so doing, how- ket policy preferences as soon as they feel vulnerable
ever, necessarily stimulated a new orientation" (ix). to unemployment. As implied by the insider-outsider
This article's arguments shed some light on two addi- model, the analysis of aggregate data also shows that
tional partisan issues. First, a number of scholars have social democratic governments are more likely to pro-
explored the reasons behind the decline in partisan- mote higher levels of labor market policies only when
ship observed at the individual level in recent years insiders become more vulnerable to unemployment.
(Dalton 2002). My analysis suggests that insider- As for the second topic, the economics literature
outsider differences help explain why some people provides us with some evidence that higher levels
do not feel represented by mainstream political op- of ALMP promote lower levels of unemployment
tions. It is reasonable to assume that outsiders would (OECD 1994; Jackman, Pissarides, and Savouri 1990).
be increasingly less likely to identify with parties that The existence of scholarly analyses demonstrating the
do not defend their interests. At the aggregate level, beneficial effects of ALMPs, however, is not as rel-
analysts have also perceived a progressive decline in evant as the general acceptance by policy-makers of
the political prominence of social democracy (see, e.g., ALMP as a tool against unemployment. This accep-
Pontusson 1995). My results suggest that this view may tance has been widespread in recent years. Transfer-
be inaccurate and that partisanship is still a powerful ring public resources into active labor market policies
determinant of policy. We cannot observe these effects, has been an objective repeatedly endorsed by third
however, if we do not look at the right policies. I argue way social democrats and OECD labor ministers. As
that only by understanding insider-outsider differences Martin (1998) points out, it has also become part of
will we perceive the true influence of partisanship on the EU's official strategy to decrease unemployment
policy-making. since the Essen Summit in December 1994 (12). The
The second partisan issue relates to the costs of social degree to which the perceived effectiveness of ALMPs
democratic strategies. The model presented in this ar- has influenced social democratic strategies should be
ticle opens the door to a debate about the implications the focus of further research. It is in any case clear
of emphasizing the interests of insiders. Given some that social democratic governments do promote labor
recent electoral setbacks in industrialized democracies, market policies when insiders are threatened by unem-
it is uncertain insider strategies are beneficial to social ployment (see Rueda, forthcoming).
72
Regarding the existence of other policy options, a Boix, Caries. 1998. Political Parties, Growth and Equality. New York:
commitment to full employment (a traditional Key- Cambridge University Press.
Calmfors, Lars. 1994. "Active Labour Market Policy and Unemploy-
nesian macroeconomic strategy for social democrats) ment: A Framework for the Analysis of Crucial Design Features."
surely poses less of a conflict between insider and out- OECD Working Papers. Paris: OECD.
sider preferences. There is, however, an important liter- Calmfors, Lars. 1993. "Lessons from the Macroeconomic Experience
ature showing the difficulties social democratic govern- of Sweden," European Journal of Political Economy, 9 (March):
ments face when trying to develop Keynesian policies 25-72.
Calmfors, Lars and Harald Lang. 1995. "Macroeconomic Effects
after the early 1970s. Up to that point, social demo- of Active Labour Market Programmes in a Union Wage-Setting
cratic efforts to reduce the inequality and insecurity Model," The Economic Journal 105 (May): 601-18.
of the most vulnerable sectors of the labor market Cusack, Thomas. 1997. "Partisan Politics and Public Finance." Public
while more generally promoting growth and employ- Choice 91 (March): 375-395.
ment had been very successful. The challenges posed Dalton, Russell. 2002. Citizen Politics. New York: Chatham House.
Doeringer, Peter B., and Michael Piore. 1971. Internal Labor Markets
by rational expectations (Alesina 1989) and increasing and Manpower Analysis. Lexington, MA: Heath.
levels of internationalization (Alt 1985) are often iden- Dore, Ronald. 1994. "Incomes Policy: Why Now?" In The Return to
tified as the reasons for the end of the golden age of so- Incomes Policy, ed. Ronald Dore, Robert Boyer, and Zoe Mars.
cial democracy. Even accepting the relevance of these London: Pinter.
Europa Yearbook. Various years. London: Europa.
challenges, some options are still open to social demo- Franzese, Robert. 1998. "The Political Economy of Public Debt."
cratic governments. Labor market policies (especially Paper presented at the Wallis Conference on Political Economy.
active ones) can be used by partisan governments to Northwestern University, November 1998.
promote employment, growth, and equality in an envi- Gabel, Matthew, and John Huber. 2000. "Putting Parties in Their
ronment that impedes demand management. I want to Place." American Journal of Political Science 44 (January):94-103.
Garrett, Geoffrey. 1998. Partisan Politics in the Global Economy.
conclude this article by pointing out that it is precisely New York: Cambridge University Press.
because of the importance of these policies that an Garrett, Geoffrey, and Peter Lange. 1991. "Political Responses to
accurate understanding of insider-outsider differences Interdependence: What's 'Left' for the Left?" International Orga-
becomes crucial. My analysis emphasizes some of the nization 45 (Autumn): 539-64.
considerable difficulties confronting social democratic Giddens, Anthony. 1998. The Third Way. London: Polity Press.
Hibbs, Douglas. 1977. "Political Parties and Macroeconomic
policy-makers who are interested in equality. The ac- Theory." American Political Science Review 71 (December):
knowledgment of these difficulties may be the first step 1467-87.
in finding truly solidaristic solutions. Hicks, Alexander, and Duane Swank. 1992. "Politics, Institutions
and Welfare Spending in Industrialized Democracies, 1960-82."
American Political Science Review 86 (September): 658-74.
Hsiao, Cheng. 1986. Analysis of Panel Data. New York: Cambridge
REFERENCES University Press.
Huber, Evelyne, and John Stephens. 2001. Development and Crisis
Addison, John T., and Jean-Luc Grosso. 1996. "Job Security Provi- of the Welfare State. Chicago: University of Chicago Press.
sions and Employment: Revised Estimates." Industrial Relations Iversen, Torben. 1999. Contested Economic Institutions. New York:
35 (October): 585-603. Cambridge University Press.
Alesina, Alberto. 1989. "Politics and Business Cycles in Industrial- Iversen, Torben. 1996. "Power, Flexibility and the Breakdown of
ized Democracies." Economic Policy 8 (April): 55-98. Centralized Wage Bargaining." Comparative Politics 28 (July):
Alt, James. 1985. "Political Parties, World Demand, and Unemploy- 399-436.
ment." American Political Science Review 79 (December): 1016- Iversen, Torben, and David Soskice. 2001. "An Asset Theory of Social
40. Preferences." American Political Science Review 95 (December):
Alvarez, Michael, Geoffrey Garrett, and Peter Lange. 1991. "Gov- 875-93.
ernment Partisanship, Labor Organization and Macro-Economic Jackman, Richard, Christopher Pissarides, and Savvas Savouri.
Performance." American Political Science Review 85 (June): 539- 1990. "Unemployment Policies." Economic Policy 5 (October):
56. 449-90.
Armingeon, Klaus, Michelle Beyeler, and Sarah Menegale. 2002. Janoski, Thomas. 1990. The Political Economy of Unemployment.
Comparative Political Data Set 1960-2001. Institute of Political Berkeley: University of California Press.
Science, University of Berne. Janoski, Thomas. 1994. "Direct State Intervention in the Labor Mar-
Beck, Nathaniel, and Jonathan Katz. 1995. "What to Do (and Not ket." In The Comparative Political Economy of the Welfare State,
to Do) with Time-Series Cross-Section Data." American Political ed. Thomas Janoski and Alexander Hicks. New York: Cambridge
Science Review 89 (September): 634-47. University Press.
Beck, Nathaniel, and Jonathan Katz. 1996. "Nuisance vs. Substance." King, Desmond. 1995. Actively Seeking Work? Chicago: The Univer-
Political Analysis 6: 1-36. sity of Chicago Press.
Bentolila, Samuel, and Giuseppe Bertola. 1990. "Firing Costs and Kitschelt, Herbert. 1994. The Transformation of European Social
Labour Demand: How Bad Is Eurosclerosis?" Review of Eco- Democracy. New York: Cambridge University Press.
nomic Studies 57 (July): 381-402. Kitschelt, Herbert. 1999. "European Social Democracy Between Po-
Berger, Suzanne and Michael Piore. 1980. Dualism and Discon- litical Economy and Electoral Competition." In Continuity and
tinuity in Industrial Societies. New York: Cambridge University Change in Contemporary Capitalism, ed. Herbert Kitschelt, Peter
Press. Lange, Gary Marks, and John Stephens. New York. Cambridge
Blanchard, O. J., et al. 1986. "Employment and Growth in Europe." University Press.
In Restoring Europe's Prosperity, ed. O. J. Blanchard. Cambridge, Lazear, Edward. 1990. "Job Security Provisions and Employment."
MA: MIT Press. QuarterlyJournal of Economics 105 (August): 699-726.
Blanchard, 0. and L. Summers. 1986. "Hysteresis and the European Lindbeck, Assar, and Dennis Snower. 1988. The Insider-Outsider
Unemployment Problem." In S. Fisher (ed.), NBER Macroeco- Theory of Employment and Unemployment. Cambridge, Mass.:
nomic Annual. Cambridge: MIT Press. MIT Press.
Boeri, Tito, Axel Borsch-Supan, and Guido Tabellini. 2001. "Would Martin, John. 1998. "What Works among Active Labour Market
You Like to Shrink the Welfare State? A Survey of European Policies." OECD Labour Market and Social Policy Occasional
Citizens." Economic Policy 32 (April): 7-50. Papers, No. 35. Paris: OECD.
73
McDonald, Michael and Silvia Mendes. 2001. Parties in Parlia- Rueda, David, and Jonas Pontusson. 2000. "Wage Inequality and
ments and Governments,1950-1995. Political Science Department, Varieties of Capitalism." World Politics 52 (April): 350-83.
Binghamton University-SUNY. Saint-Paul, Gilles. 1998. "A Framework for Analysing the Political
Moene, Karl Ove, and Michael Wallerstein. 2003. "Earnings In- Support for Active Labor Market Policy."Journal of Public Eco-
equality and Welfare Spending." World Politics 55 (July): 485- nomics 67 (February): 151-65.
516. Saint-Paul, Gilles. 1996. "Exploring the Political Economy of Labour
Nickell, Stephen. 1981. "Biases in Dynamic Models with Fixed Ef- Market Institutions." Economic Policy, 23 (October): 264-315.
fects," Econometrica 49 (November): 1417-26. Scharpf, Fritz. 1991. Crisis and Choice in European Social Democ-
OECD. 1994. The OECD Jobs Study. Paris: OECD. racy. Ithaca, NY: Cornell University Press.
OECD. 1999. Employment Outlook. Paris: OECD. Schwartz, Herman. 1994. "Small States in Big Trouble." World Poli-
Pontusson, Jonas. 1995. "Explaining the Decline of European Social tics 46 (July): 527-55.
Democracy." World Politics 47 (July): 495-533. Swank, Duane, and Cathie Jo Martin. 2001. "Employers and the
Powell, Bingham. 1982. Contemporary Democracies. Cambridge, Welfare State." Comparative Political Studies 34 (October): 889-
MA: Harvard University Press. 923.
Powell, Bingham. 2000. Elections as Instruments of Democracy. New Visser, Jelle. 1996. "Unionization Trends Revisited." Centre for
Haven, CT: Yale University Press. Research of European Societies and Industrial Relations,
Rueda, David. 2001. "Government Partisanship and Economic Amsterdam.
Policy: Insider-Outsider Politics in Industrialized Democracies." Wilensky, Harold. 1975. The Welfare State and Equality. Berkeley:
Ph.D. Diss. Cornell University. University of California Press.
Rueda, David. N.d. "Social Democracy and Active Labor Market Woldendorp, Jaap, Hans Keman, and In Budge. 2000. Party Govern-
Policies: Insiders, Outsiders, and the Politics of Employment Pro- ment in 48 Democracies (1945-1998). Boston: Kluwer Academic
motion." British Journal of Political Science. Forthcoming. Publishers.
74

Insider-Outsider Politics in Industrialized Democracies The Challenge To Social Democratic Parties

Uploaded by

Copyright:

Available Formats

Insider-Outsider Politics in Industrialized Democracies The Challenge To Social Democratic Parties

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Insider-Outsider Politics in Industrialized Democracies The Challenge To Social Democratic Parties

Uploaded by

Copyright:

Available Formats

Insider-Outsider Politics in Industrialized Democracies: The Challenge to Social Democratic

easier for parties to seek out their probablesupport- modelthereforeis thatsocialdemocraticgovernmentis

INSIDERS, OUTSIDERS, AND THE INDIVIDUAL PREFERENCES IN THE

themselves closer to the other side of the spectrum.

FIGURE 2a. Employment Preferences

7.2 7.4 7.6 7.8 8.0 8.2 8.4 8.6

FIGURE 2b. Employment Preferences

4 4.2 4.4 4.6 4.8 5 5.25 5.4

FIGURE 2c. Employment Protection Preferences

5 5.2 5.4 5.6 5.8 6 6.2 6.4 6.6

TABLE 2. The Determinants of Labor Market Policies, Alternative Models

ALMPs PLMPs ALMPs PLMPs ALMPs PLMPs

eliminated from the model. When analyzing both

those in the 1980s. To explore this issue, I test whether

You might also like