By Paul Pennings
To be cited as:
Pennings, P. (2010) Exploring Variations in the Political Discourse on Public
Sector Reforms, 1981-2005, Public Management Review 12 (2): 173-
190.
Abstract
This paper explores the patterned variations in the references in election manifestos of political parties in OECD countries to market oriented reforms of the public sector, irrespective whether these references are in favour of these reforms or not. It is expected that these variations are structured by institutional features which are related to national, partisan and sectoral differences.
The empirical analysis shows that the national differences between parties are influenced by their membership of ‘families of nations’ since the adoption of market principles is expected to be ideologically more acceptable in, for example, the Anglo-
Saxon world than in Scandinavia. The recent differences between the main party groups are modest, which means that these reforms have become equally ’important’ for the established party families that dominate the governments in the selected
OECD countries. The differences between policy sectors are partly due to their relation with the welfare state. Most references to reforms are made in the policy sector infrastructure which reflects the numerous attempts to liberalise and privatise this sector. The increase of references to reforms in some sectors that are related to the welfare state (e.g. social affairs and health care) does not coincide with less public expenditures due to the path dependency of spending in these sectors. The diffusion of public sector reforms does not lead to convergence between parties in the sense that national, partisan and sectoral differences become smaller over time.
0

Public sector reform seeks to maximize efficiency and effectiveness in the public sector. This article focuses on one of the several waves of public sector reform, namely the introduction of market elements in the public sector. It has been introduced in an increasing number of policy sectors in which (semi-) public goods and services are produced. This paper departs from the overarching research question: do all political parties refer to public sector reforms in similar or in differing ways over time due to patterned variations in groups of parties and families of nations?
This question is related to the ongoing debate on ‘Does Politics Matter’ for public policy-making (Hood 1995; Keman 2002: 166-212). Most researchers conclude that party policy positions do have a conditional effect on public policy-making. In case of public sector reforms such a consensus does not exist because it is not strongly connected to the left-right distinction that structures most of the party competition in the post-war era (Pierson 2001: 417). Does this mean that party differences are unrelated to public sector reforms so that there is an overall consensus among parties and do parties differ on this?
The research is guided by a number of hypotheses that will be empirically tested.
These hypotheses are in line with the existing literature (Schmidt 2000; Huber and
Stephens 2001; Swank 2002) but they have never been combined into a longitudinal and comparative analysis of party manifestos based on the saliency theory of party competition (Budge 2001: 78-85). It is hypothesized that public sector reforms first emerged in the Anglo-Saxon world and that its diffusion in public policy-making depends on the type of welfare state, the party family, and the type of policy sector.
First, we will discuss in more detail the hypotheses and research design. The subsequent sections provide an empirical analysis of references to public sector reform per family of nations, party family, time period and policy sector. In addition, we examine the consequences of public sector reform: does it lead to lower public expenditures and to policy convergence?
The main goals of public sector reform are to reduce costs, increase effectiveness and efficiency in order to make the public sector more businesslike, ’work better and cost
1

less’ and more client-oriented (Esping-Andersen 1996; Nascholt 1996; Clayton and
Pontussen 1998; Hood 2000; Lane 2000). Although there are different routes in public sector reform, the focus is on the most dominant one which favours the diffusion of market values in the public sector so that the operation of public programs in general and in public enterprise in particular is steered by the market mechanism (Lane 1993: 148; Hodge, 2006). Often this type of public sector reform is related to the discourse on welfare state adjustment (Esping-Andersen 1996; Huber and Stephens 2001; Pierson 2001; Schmidt 2002; Swank 2002). Public sector reform is part of a broader change from the dominant interventionist and Keynesian policies of the first part of the 20th century (’the golden age’) to more market friendly and less state controlled public policy-making since the 1980s (’after the golden age’) (Scharpf and Schmidt 2000: 24-85).
The hypotheses depart from the central assumption that the number of references that parties make to public sector reform is characterised by patterned variations due to national, sectoral and partisan differences. This expectation is subdivided into five hypotheses:
Hypothesis 1: Public sector reform will spread first in the Anglo-Saxon world
(1970s/1980s), then influence European parties (1980s/1990s) and only after that enter the Scandinavian manifestos (1990s and onwards) (Huber and Stephens 2001:
113-201; Schmidt 2000: 235-301).
Hypothesis 2: Liberal and conservative parties do refer earlier and more often to public sector reform than social democratic and socialist parties because advocates of public sector reform propagate a retrenchment of the welfare state.
Hypothesis 3: There is an interaction between the family of nations and the party family influences. This means that social democratic parties in, for example, Spain, the Netherlands and Sweden will refer to public sector reform in different manners and at different time points because they are rooted in different welfare states
(Esping-Andersen 1996: 1-27; Huber and Stephens, 2001: 113-201). However, the social democratic party group as a whole is expected to differ from the liberal group that stressed to need for public sector reforms earlier.
Hypothesis 4: The closer a policy sector is related to the welfare state, the less its
2

discourse will be penetrated with references to market depending also on the family of nations and the accompanying type of welfare state. Infrastructure, for example, is expected to get more co-mentions with market principles than social welfare. But in the Anglo-Saxon world references to market principles in the domain of social welfare are expected to be more frequent than in Scandinavia where a more generous welfare state is prevalent.
Hypothesis 5: If a sector is at the core of the welfare state it is expected that public sector reform will be modest or absent and public expenditures do increase to some extent. If it is not strongly related to the welfare state it is expected that public sector reform will be more prevalent in manifestos and that public expenditures may decrease.
The distinction between policy areas follows the main divisions which are made among ministries (Woldendorp, Keman and Budge 2000: 21-22). Not included are the policy areas Defence, International Affairs, Industry/trade and Interior because they are either mainly internationally oriented, or less relevant for public sector reforms because they are not directly dealing with matters that concern the public sector and public services. Included are also agriculture and infrastructure because the former is or was heavily subsidised (just like culture) and the latter has been engaged into some far reaching privatisation projects (post, railways, electricity etc.). All other included sectors are predominantly oriented towards to production of public goods: safety, health, housing, environment, education, welfare, etc.
Although party manifestos are hardly read by voters, they do reflect the priorities of parties that mediate between electors’ preferences and public outcomes. Party responsiveness is at the heart of democratic decision-making as it concerns the ways in which political parties translate problems (i.e. situations which cannot be solved by means of societal self-regulation or are considered as a public assignment) and issues into programs and, when in government, translate the programs into public or collective decision-making (McDonald, Budge and Pennings 2004: 845-846).
It should be stressed that the priorities which are expressed in manifestos are not always translated into actual policy-making by the parties that enter government. The political-institutional context impacts both on the content of manifestos and the actual policy-making by party governments (Klingemann, Hofferbert and Budge 1994: 20-
35). Although the call for public sector reform is expected to be higher in Anglo-
3

Saxon world than in the Scandinavian world, this does not mean that the level of actual welfare state adjustment towards less state and more market involvement follows the same pattern. Existing research has shown that, for example, in Sweden the degree of public sector reform in the 1990s is relatively high, whereas the party pledges of the social democrats do not emphasise the need for such reforms (Clayton and Pontusson 1998: 90-97). This discrepancy can partly be explained by the saliency theory of party competition (Budge and Farlie 1983; Budge 2001; Petrovic, Benoit and Hansen 2003) which assumes that parties try to render selective emphases by devoting most attention to the types of issues that favour themselves and give correspondingly less attention to issues which favour their opponents. This may evoke the paradoxal situation that social democratic parties in highly developed welfare states may be less inclined to stress public sector reform in manifestos in order to maintain their credibility, but at the same time do initiate and cooperate on those reforms when in government in order to make the welfare state less vulnerable to external threats (Green-Pedersen 2001).
In addition, within families of nations there are also significant differences due to political institutional differences, such as the formal and informal rules of decisionmaking, which might facilitate or block partisan influence. Rational parties are limited in pursuing their interests because they must act within the rules of the political game to achieve their individual utility which yields optimal (not maximum) results (Keman
1999: 251). For example, if a right government comes into office in a period in which there are both strong unions and a corporatist intermediation structure, then it is much harder to impose public sector reforms that favour the market. Hence, the feasibility of political choice depends on the institutional room to manoeuvre that may constrain actors in maximising their own needs.
Finally, the type and degree of public sector reform also depends on the difficulties that parties have to overcome in order to reach consensus on this often publicly contested topic. Green-Pederson (2001) shows in a comparative study of retrenchment in Denmark and the Netherlands that a larger extent of retrenchment has been implemented in the Netherlands because of the dominant role of a pivotal centre party, the Christian Democratic Appeal, in its system of party competition which produced party consensus about welfare-state retrenchment. The Dutch labour party had to accept welfare reforms to regain government power and Dutch governments were able to use this consensus to frame the reform agenda as an economic necessity.
In Denmark, the strong opposition from the social democrats made public sector
4

reform a contested issue during the right-wing governments from 1982 to 1993.
Following the social democrats' return to power in 1993, a party consensus emerged, leading to actual reforms. Cox (2001) has argued that such reforms did not happen in
Germany because the debate remained polarized, with many important political actors rejecting the suggestion that reform was either necessary or desirable.
These examples suggest that there is no one-to-one correspondence between the discourse on public sector reform and the degree of actual reform. Nonetheless, the discourse itself is an important and relevant angle to study public sector reform because it reveals the ideological party positions on the topic which is one of the main bases of party behaviour, although the actual behaviour is shaped by the institutional environment. An important argument in favour of using manifestos as a data source is that they enable longitudinal and cross-sectional comparisons. The available handcoded party policy positions of the Manifesto Research Group (MRG) (as reported in
Klingemann et al. 2006) include a number of categories that are vaguely related to public sector reform (such as efficiency), but never in direct relation to the policy sectors that are listed in Table 1. This is important as every co-occurrence between market principles and a public sector domain is counted as reference to public sector reform. Such inferences cannot be made from the MRG data set.
As it is not feasible to recode all manifestos manually, the only way to extract this information is by means of automated content analysis. This has been made possible by the digitisation of the manifestos which are collected by the Manifesto Research
Group (Budge et al. 2001; Pennings and Keman 2002; Klingemann et al. 2006).1
INSERT TABLE 1 ABOUT HERE
All documents have been translated into English so that they can be coded with one single categorisation dictionary that is used to count the references to market principles in all main public policy areas. These references do not include explicit mentions of the phrase ‘public sector reform(s)’ (since there are two of them), but references to words that are typical for the discourse on public sector reform, such as efficiency, effectiveness, deregulation, privatisation, marketisation, liberalisation, competition etc. The number of references to market principles are counted per policy sector and compiled in a database that enables a cross-national longitudinal analysis.
The categorisation dictionary allows one to change specific words, word patterns, or expressions, to the ten categories listed in Table 1. The number of key words and
5

phrases varies per policy sector because some descriptions of sectors need to be more elaborate than others due to the inclusiveness of these sectors. Social welfare or environment, for example, are very broad and incorporate more aspects than housing
(which is more focussed). The dictionary is validated by means of the keyword-incontext (KWIC) procedure which highlights the keywords in Table 1 within the context in which they are used. No word or word string was allocated to more than one coding category.
The dictionary is used to extract so-called co-occurrence matrices with the software package WordStat by cross-tabulating the ten categories.2 The unit of analysis is the frequency of co-mentions - of policy areas and references to market principles - per party per election year per policy sector. This format allows us to aggregate (in all the tables that follow) across parties, time periods and countries in order to reveal spatial and temporal variations. The co-occurrences are analysed by counting the number of references made to public sector reforms per policy area. The calculations are based on this total number as a percentage of all references in manifestos in order to take the size of the manifesto into account. If the number of ‘hits’ on the main key itself is only five or lower, it is set to zero because small numbers may occasionally lead to a very high percentages (even 100%) if the manifesto is only a few pages long.
The co-occurrence matrices are stacked into a pooled data file with a total of 4712 rows covering 81 parties in the period 1981-2005 in 19 OECD countries. Excluded are parties that started after the middle of the 1980s because these parties do not exist long enough to have experienced the rise of public sector reform. A shortcoming of the coding technique is that it is unable to detect implicit references to terms related to market principles within paragraphs or sections since it will only count explicit references within sentences. For this reason, the reported numbers do underestimate the real amount of references to market principles and must be seen as proxies for the degree to which parties do co-mention policy domains and market values. The coding technique causes systematic errors: reproducible inaccuracies that are consistently in the same direction and cannot be analyzed statistically. Since we are not mainly interested in calculating the exact number of references made to market principles in the public sector, but more specifically in the spatial and temporal variations in these references, these systematic errors do not hamper our research goal.
Not only the coding technique, but also the parties may be a cause of ’error’. Since party manifestos are strategic documents, parties may have incentives to minimise the number of references to market values in the public sector, especially when the degree
6

of public support for this is low (Kitschelt 2001; Vis 2008; Vis and van Kersbergen
2007). This is not surprising since there are several reasons why parties may deemphasise the introduction of market principles in the public sector and hide their
‘real’ position on this issue. First, the discourse on public sector reform tends to blur the left-right distinction which puts parties at the risk of becoming indistinguishable from each other (Hood 1995; Lane 2000; Pierson 2001). Second, given its controversy in relation to distributive politics and social policy, public sector reforms may threaten party cohesiveness by fuelling debates between left and right wings within parties. In order to avoid this, parties may deliberately avoid the topic, as is also the case with other controversial themes, such as European integration (Marks and Steenbergen,
2004: 170-171). Third, in some countries public sector reforms are implemented at local level (Germany and Sweden, for instance, have a ’local welfare state’) and consequently debated at the sub national level of government (Montin 2000;
Wollmann 2000; Mattei, 2007).
Families of nations are groups of countries with shared national attributes, like historical background, language, culture and religion (Castles 1993; Roller 2005).
Examples are English-speaking, German-speaking and Scandinavian countries. This common background may facilitate or hamper the diffusion of certain policy reforms and ideas. The concept of ‘families of nations’ provides an angle to perceive countries as interdependent actors which is important to understand variations in diffusion mechanisms (Braun and Gilardi 2006).
In case of public sector reform, not only language matters but also the different types of welfare capitalism (Esping-Anderson 1990: 9-34; Schmidt 2000: 234-235). Esping-
Anderson has divided Western countries into three types, depending on the degree of de-commodification and principles of stratification, namely Liberal, Conservative and
Socialist types. These types are strongly tied to families of nations, since the liberal type is predominantly present in the Anglo-Saxon world, the conservative or corporatist type in continental Western Europe and the socialist type in Scandinavia.
This discussion on varieties of welfare capitalism may help to explain why parties within some clusters relate differently to public sector reforms than parties in other clusters.
7

However, it is not possible to equate families of nations with types of welfare states.
The division into types of welfare regimes has been questioned because it results from some crude measurement criteria which are arbitrary, biased (expenditure considerations), incomplete and sometimes inconsistent (Scruggs and Allan 2006).
Castles (1993: 93-126), for example, has argued that Australia is an example of liberal de-commodification and socialist stratification. He also proposed a fourth world of radical welfare capitalism, consisting of Australia, New Zealand and the UK, with a high trade union density, but a low non-right incumbency.
We opt for a broad definition of family of nations as clusters of countries on the basis of their historical background, language, culture and religion. Four families are distinguished: Scandinavia, Western Europe, Latin Europe, Anglo-Saxon countries.
This broad categorisation suits our purpose to distinguish patterned variations in the diffusion of public sector reform in established OECD countries. In addition to these four families, some authors add a fifth one, namely Switzerland and Japan (Castles
1993; Roller 2005). Since Japan is not part of the data collection, this fifth category is not feasible (and it is not very instructive to have a category with solely deviant cases).
The Latin group needs some more explanation. These four countries are thought to have a common linguistic background and Roman Catholicism as the prevalent religion. Within this group, France and Italy are of course different from Spain and
Portugal which are more typical for the ancient culture of the Mediterranean and a delayed economic, social, and political modernization. France is exceptional in several respects, with its combination of statism, republicanism, presidentialism and majoritarian electoral system. But in broad political, social and cultural terms (i.e. the degree of political instability, party systems with both strong conservative and socialist parties, a non-Anglo-Saxon orientation in terms of language, clientelist social relations, Latin welfare states), France and especially Italy do seem to be more part of a Mediterranean than a Western Continental group of nations (Ferrera 1996; Pridham and Lewis 1996; Colomer 2008: 94-134).
INSERT TABLE 2 ABOUT HERE
In all the tables that follow, the election years are grouped into five year periods. It is expected that these periods present different stages in the rise of public sector reform: a preliminary stage, a stage of early diffusion, a stage of widespread diffusion and a stage of stabilisation or slight decrease.
8

Table 2 shows the references to public sector reforms in manifestos per period and per
‘family of nations’. These references are highest in Latin Europe, followed by
Western Europe and then the Anglo-Saxon group. Its spread in the Scandinavian countries is relatively modest and takes place in a later stadium (apart from Norway, mostly after 1995) (Schmidt 2000: 256-266).
There are significant within-group differences which indicate a number of interesting exceptions within those families of nations. Within the Scandinavian group the
Norwegian parties show relatively high scores that are comparable with the means of
Western European parties. This is partly explained by the Norwegian social democracy that adopted many of the neo-liberal ideas. Especially the Labour government in 2000-2001 carried through extensive market reforms, such as the privatisation of the state telecom (Telenor) as well as the state oil company (Statoil).
In addition, the entire hospital sector was restructured into a new market oriented model and public services at municipal level were put under competitive tendering.
(Lyngstad 2008: 70-73). This is also reflected in the manifestos. The Norwegian social democrats are quite sympathetic to introducing more efficiency in the public sector, given statements such as: ‘The Labor party favours a well run and efficient public sector, and is impatient to develop the welfare society accordingly’ (2005). In
Finland, Sweden and Denmark public sector reform mainly emerges after 1995 in the manifestos and the number of references are relatively low. Whereas in Norway public sector reform is mostly characterized by centralising tendencies, in Sweden the dominant policy implementation strategy is decentralisation, although this may change per policy sector and time period (Montin 2000). This would explain the difference between these countries.
Within Western Europe the Swiss parties are consistently below the mean.
Nonetheless, most Swiss parties seem to be in favour of public sector reform, given some of the statements made in their manifestos. Only the Social democrats have some reservations: ‘With an efficient national insurance system, the inefficient bureaucracy and the expensive pseudo competition of the today's health insurance companies must be eliminated’ (2003).
Because at least two of the four Latin countries started the process of welfare state building decades later than the other OECD countries, their development of welfare state programs, in many instances, is still incomplete. The fast growth of their welfare states more or less coincides with the diffusion of market principles in social and public policy-making. In addition, their expansion of the welfare state, contrary to the
9

situation in other European States which profited from the continuous economic development of the 1960’s, occurred in a period of general economic recession with stressed the need of efficiency. In Italy there is an additional factor, namely the sceptical attitude of most parties towards the efficiency of the Italian public administration. Within Latin Europe, the French parties do show relatively few references to public sector reform, which is in line with the French state-led production regime of social and economic welfare (Schmidt 1996).
Within the Anglo Saxon world the American and New Zealand parties started to make references to public sector reform, quickly followed by the UK and Ireland. In
Australia and Canada these references are below the group mean until 1995. Whereas in New Zealand the monetarist and Hayekian doctrines shaped policies from 1984 onwards, the deregulatory thrust was weaker in Australia, partly due to the influence of a working relationship between unions and the Labour movement (the Australian
Accord) (Castles 1996: 105). Canada provides a counterpoint to the American model since it has avoided the American disease of rising poverty and inequality largely via social transfers. Only by the 1990s a large public debt has forced Canada to cut social expenditures (Myles 1996: 118).
The table demonstrates that the references to public sector reform in manifestos do not show a linear trend. It is difficult to account for all fluctuations between decennia since many factors may have played a role in this, such as party strategic considerations, the importance of particular issues in election campaigns, the internal consent on public sector reform and the size of manifestos. But the general trend goes up and occasionally goes down again after 1995. It seems that the references to market principles in the public sector mainly go down in 1996-2005 in those cases where we find the highest public sector reform scores in 1986-1995. Similarly, it goes up in 1996-2005 in those cases with relatively lower scores in 1986-1995 (i.e. When the level of public sector reform in 1986-1995 is dichotomised into lower and higher than 1.5% and the growth in 1996-2005 is also dichotomised into growth/no growth, the correlation is –0.44, p=0.037). It is too early to conclude from this that public sector reform has been a temporary hype that will fade away soon.
Political parties are grouped into party families depending on their ideological
10

affiliation. We adopt the division of parties into party families that is used in the MRG project: communism, social democracy, liberalism, Christian democracy, conservatism and agrarian parties. The nationalist, ethnic and regional parties are excluded due to a low number of cases (lower than 5). As Table 3 shows, the degree to which party families refer to public sector reforms is quite similar. However, since this table gives no information about the direction of the references (i.e. whether they are for or against) it is not directly clear how these references should be interpreted.
INSERT TABLE 3 ABOUT HERE
A ‘Keyword in Context’ analysis can clarify whether party groups are in favour of or against public sector reform. It turns out that mainly the communist references to public sector reform should be interpreted as predominantly negative ones. Note that only eight communist parties are included with a long tradition in a country, which is mainly the case in Scandinavia and Latin Europe. From the ‘Keyword in Context’ analysis it becomes clear that these eight communist parties are opposed to public sector reform as far as it implies privatisation and marketisation of the public sector.
It is remarkable that social democratic parties do not refer less to public sector reform than liberal parties, which goes contrary to our hypothesis. Public sector reform has been equally ‘important’ through all stages of diffusion for the four most established party families that dominate the governments in the selected OECD countries. An explorative ‘Keyword in Context’ analysis learns that some social democratic parties are critical or ambivalent about the results of public sector reform in some policy areas, but overall their position towards public sector reform is quite positive. This is not surprising since public sector reform has some resemblance with ‘third way politics’ which has been adopted by many social democratic parties in the 1990s.
Examples of shared principles between public sector reform and the ‘third way’ are the reconstruction of social welfare systems to give people incentives to work and partnerships with the non-government sector (Giddens 1998; Bonoli and Powell
2004).
After 1996 there is hardly any variation in the degree to which parties from different party families refer to public sector reform in their manifestos, which is a clear sign of diffusion, since parties recognize the importance and relevance of public sector reform, irrespective whether they are against, in favour or take a nuanced position.
The percentage of references to public sector reform has become two to six times higher in the period after 1995 compared to the period before 1976. The strongest
11

increase was between 1980-1975 and 1976-1985.
More relevant than party family differences per se are the differences between party families per family of nations because these patterns show us whether the number of references to public sector reform is influenced by the national context. It is hypothesised that there is an interaction between the family of nations and the party family influences. Table 4 shows that this is the case. The Scandinavian party groups refer less to public sector reform than their counter parts in other families of nations, in particular the Anglo-Saxon and the Latin group.
INSERT TABLE 4 ABOUT HERE
This is a sign of diffusion, since the references to public sector reform started in the
Anglo-Saxon world and consequently gradually spread over the parties in the other groups of nations. After 1975 references to public sector reform reach even a higher level in some policy areas in non-Anglo-Saxon families of nations. References to public sector reform are only more frequent in agriculture and (until 1996) in infrastructure in the Anglo-Saxon countries. In some other areas, like social affairs, public order, education and culture, the parties in Western and Latin Europe make often even more references to public sector reform.
As specified in Table 1, all policy sectors are included which produce public goods or private goods that are heavily subsidized with public money and/or predominantly publicly regulated. Table 5 shows that most references are made to reform in the domain infrastructure. This reflects the numerous attempts of liberalisation and privatisation in this sector. In agriculture the number of references to reform is relatively low and rather stable. In the welfare related sectors there is often a significant rise in the references to public sector reform. Education, for example rises from 0.6 to 1.3 and health care from 0.3 to 0.9. The differences among these ‘welfare sectors’ are higher than expected. In health care and housing references to reform are
12

quite modest compared to social affairs and public order that show quite stable scores over the years.
INSERT TABLE 5 ABOUT HERE
The table also reports the change in public expenditures between the periods 1976-
1985 and 1986-1995. There is no significant statistical relationship between the references to reform in manifestos in the periods 1976-1985 and 1986-1995 on the one hand and public expenditures on the other hand. More references to reform do not coincide with lower expenditures. However, there are patterned variations per policy sector. In health care and social affairs the expenditures do increase. In agriculture, education and infrastructure the expenditures decrease. This pattern suggests that reforms in policy sectors that are more market oriented ‘by nature’ coincide with decreasing public expenditures. Reforms in sectors that are per definition or traditionally part of the welfare state does not so easily coincide with less public expenditures, due to path dependency, although there are exceptions (Kittel and
Obinger 2003).
Policy diffusion is often associated with convergence. In the case of public sector reform this would mean that the differences between countries, party families, parties, policy sectors etcetera become smaller over time because it emerges eventually in all manifestos in a similar manner. In order determine to what extent these groups become more similar, the Eta statistic is used. Eta equals the square root of the sum of squares for an interval variable y between classes (categories) divided by the total sum of squares. The numerator and denominator in this formula have meanings as in
ANOVA, and to the extent that x and y are linearly or nonlinearly related, the numerator will be as large as the denominator and Eta will approach 1.0. Eta2 is the percent of variance in the dependent variable (in our case: references to public sector reforms) explained linearly or nonlinearly by the independent variable.
13

INSERT TABLE 6 ABOUT HERE
Table 6 shows that diffusion does not imply convergence since the between-groups differences become either consistently larger over time or they remain small or nearly absent. The table demonstrates that the differences between parties, nations and sectors are important to understand the variations in the references to public sector reform. The between-groups differences are small before the 1980s because references to reform are hardly present in election manifestos: 87 per cent of all entries are zero. In 1996-2005 only 49 per cent is zero which indicates that far more parties are referring to public sector reform. The table also shows that the betweengroup differences are small: the explained variance is never higher than 25 per cent.
This means that there are significant within-group differences so that references to reform by parties within countries, party families and policy sectors do vary significantly. As a consequence, although there is diffusion, public sector reform does not penetrate the political discourse of parties in all countries to the same degree and in the same manner (Schmidt 2002).
Do parties differ on how frequently they refer to public sector reform? The hypotheses that deal with this question depart from the central assumption that references to reforms in the election manifestos in OECD countries is characterised by patterned variations due to national, sectoral and partisan differences. The empirical analysis shows that the references to reforms in established OECD countries is indeed characterised by patterned variations along these lines. Public sector reform did spread first in the Anglo-Saxon world (1970s/1980s), then influence European parties
(1980s/1990s) and only after that enter the Scandinavian manifestos (1990s and onwards) (hypothesis 1). Liberal and conservative parties refer earlier and more often to public sector reform than left parties (hypothesis 2). However, we should be aware that only explicit references to market principles at the sentence level are counted. As a consequence, the percentage of references to reforms is underestimated. If we leave this aside, the increase of references towards public sector reforms is quite significant.
14

In some decades, the number of references is several times higher than in the preceding decade(s).
The national differences among parties are partly influenced by their membership of ‘family of nations’ (hypothesis 3) since public sector reform is more prevalent, for example, in the Anglo-Saxon world than in Scandinavia (with Norway as the exception). The diffusion process is characterised by patterned variations that are related to the type of welfare state and its phase of development. Since the rise of the welfare state in Latin Europe coincides with the rise of public sector reform, both seem to be intrinsically related which leads to relatively many references to public sector reform in the party manifestos in Latin Europe (with France as an exception).
Within families of nations differences between members are often related to differing types of welfare states.
The differences between policy sectors are partly due to their relation with the welfare state (hypothesis 4). In several welfare related sectors there is often a significant rise in the references to public sector reform. Education, for example, rises from 0.6 to 1.3 and health care from 0.3 to 0.9. The differences among the welfare sectors are higher than expected. In case of health care and housing, the number of references is quite modest compared to social affairs and public order. Most references are made in the policy sector infrastructure. This reflects the numerous attempts of liberalisation and privatisation in this sector. In agriculture the number of references is lower than expected and hardly varies over time.
The emphasis on market principles in some sectors that are strongly related to the welfare state does not coincide with less public expenditures due to the path dependency of spending in these sectors (hypothesis 5). The references to market principles in the public sector also do not lead to convergence between parties per country, family of nations or party family because diffusion means that parties more and more start mentioning market principles and therefore become more different from each other compared to the situation that hardly any party refers to public sector reforms.
15

References
Bonoli, G. and Powell, M. eds (2004) Social Democratic Party Policies in
Contemporary Europe, London: Routledge.
Braun, D. and Gilardi, F. (2006) Taking 'Galton's Problem' Seriously. Towards a
Theory of Policy Diffusion, Journal of Theoretical Politics , 18:3 pp298-322.
Budge, I. (2001) Theory and Measurement of Party Policy Positions in I. Budge, H.-
D. Klingemann, A. Volkens, J. Bara, E. Tanenbaum (Eds) Mapping Policy
Preferences. Estimates for Parties, Electors, and Governments 1945-1998 .
Oxford: Oxford University Press.
Castles, F.G. ed. (1993) Families of Nations: Patterns of Public Policy in Western
Democracies , Aldershot: Dartmouth.
Castles, F.G. (1996) Needs-Based Strategies of Social Protection in Australia and
New Zealand, in G. Esping-Andersen, (Ed.) Welfare States in Transition.
National adaptations in global economies, . London: Sage.
Clayton, R. and Pontusson, J. (1998) Welfare-State Retrenchment Revisited:
Entitlement Cuts, Public Sector Restructuring, and Inegalitarian Trends in
Advanced Capitalist Societies, World Politics, 51:1 pp67-98.
Colomer, J. ed. (2008) Political institutions in Europe, London and New York:
Routledge (3 rd
edn).
Cox, R. H. (2001) The Social Construction of an Imperative: Why Welfare Reform
Happened in Denmark and The Netherlands but not in Germany. World
Politics, 53: 3 pp463–98.
Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism , Cambridge:
Polity Press.
Esping-Andersen, G. ed. (1996) Welfare States in Transition. National adaptations in global economies , London: Sage.
Ferrera, M (1996) The ‘Southern Model’ of Welfare in Social Europe,
Journal of
European Social Policy.
6:1 pp17-37.
Giddens, A. (1998) The Third Way. The Renewal of Social Democracy, Cambridge:
Polity Press.
Green-Pedersen, C. (2001) Welfare-state Retrenchment in Denmark and the
Netherlands, 1982-1998. The Role of Party Competition and Party Consensus,
Comparative Political Studies.
34:9 pp963-85.
16

Hodge, G. ed (2006) Privatisation and Market Development: Global Movements in
Public Policy Ideas , Cheltenham: Edward Elgar.
Hood, C. (1995) The ‘New Public management’ in the 1980s: Variations on a theme,
Accounting, Organizations and Society.
20: 2/3 pp93-109.
Hood, C. (2000) Paradoxes of Public Sector Public Management, Old Public
Management and Public Service Bargains, International Public Management
Journal.
3:1 pp1-22.
Huber, E., and Stephens, J.D. (2001) Development and Crisis of the Welfare State.
Parties and Policies in Global Markets, Chicago and London: The University of Chicago Press.
Keman, H. (1999) Political Stability in Divided Societies: A Rational-Institutional
Explanation, Australian Journal of Political Science.
34:2 pp249-68.
Kitschelt, H. (2001) Partisan Competition and Welfare State Retrenchment: When do
Politicians Choose Unpopular Policies? in P. Pierson ed. The New Politics of the Welfare State , Oxford: Oxford University Press.
Kittel, B. en Obinger, H. (2003) Political Parties, Institutions, and the Dynamics of
Social Expenditures in Times of Austerity. Journal of European Public Policy
10, pp20–45
Klingemann, H. D., R. I. Hofferbert and Budge, I. (1994) Parties, Policies and
Democracy, Boulder, CO: Westview Press.
Klingemann, H.-D., Volkens, A., Bara, J., Budge, I. and McDonald, M. (2006)
Mapping Policy Preferences II: Comparing 24 OECD 24 CEE countries, 1990-
200 3, Oxford: Oxford U.P.
Lane, J.-E. (1993) The Public Sector: Concepts, Models and Approaches, London:
Sage.
Lane, J.-E. (2000) Public Sector Reform. Rationale, Trends and Problems , London:
Sage.
Lyngstad, R. (2008) The Welfare State in the Wake of Globalization .
The case of
Norway, International Social Work.
51:1, pp69-81.
Marks, G. and Steenbergen, M.R. Eds. (2004) European Integration and
Political Conflict , Cambridge: Cambridge University Press.
Mattei, P. (2007) Transforming the Local Welfare State in Italy, West European
Politics.
30:3 pp595-620.
17

McDonald, M. D, Budge, I. and Pennings, P. (2004) Choice versus Sensitivity: Party
Reactions to Public Concerns, European Journal of Political Research.
43:6 pp
845-68.
Montin, S. (2000) Between Fragmentation and Co-ordination. The Changing Role of
Local Government in Sweden, Public Management.
2:1 pp1-23.
Myles, J. (1996) When Markets Fail: Social Welfare in Canada and the United States in G. Esping-Andersen (ed.) Welfare States in Transition. National Adaptations in Global Economies, London: Sage.
Naschold, F. (1996) New Frontiers in Public Sector Management , Berlin-New York, de Gruyter.
Pennings, P. and Keman, H. (2002) Towards a New Methodology of Estimating Party
Policy Positions, Quality & Quantity.
36:1) pp55-79.
Petrovic, J.T., Benoit, W.L. and Hansen, G.J. (2003) Issue Ownership and Presidential
Campaigning, 1952-2000. Political Science Quarterly, 118 pp599-626.
Pierson, P. (2004) Politics in Time: History, Institutions, and Social Analysis ,
Princeton University Press.
Pridham, G. and Lewis, P.G. eds. (1996) Stabilising Fragile Democracies: New Party
Systems in Southern and Eastern Europe, London: Routledge.
Roller, E. (2005) The Performance of Democracies. Political Institutions and Public
Policy , Oxford: Oxford University Press.
Scharpf, F.W. and Schmidt, V.A. Eds (2000) Welfare and Work in the Open
Economy. From Vulnerability to Competitiveness . Oxford: Oxford University
Press.
Schmidt, V. A. (1996) From State to Market? The Transformation of French
Business and Government, Cambridge: Cambridge University Press.
Schmidt, V. A. (2000) Values and Discourse in the Politics of Adjustment, in: F.W.
Scharpf and V.A. Schmidt (Eds) Welfare and Work in the Open Economy.
From Vulnerability to Competitiveness , Oxford: Oxford University Press.
Schmidt, V. A. (2002) Does Discourse Matter in the Politics of Welfare State
Adjustment?, Comparative Political Studies. 35:2 pp168-93.
Scruggs, L. and Allan, J. (2006) Welfare-state decommodification in 18 OECD
Countries: A Replication and Revision, Journal of European Social Policy.
16:1 pp55-72.
Smith, A. (1776/1981) The Wealth of Nations , Harmondsworth: Penguin Books.
18

.
Swank, D. (2002) Global Capital, Political Institutions, and Policy Change in
Developed Welfare States , Cambridge: Cambridge University Press.
Vis, B. (2008) Biting the Bullet of Steering Clear? Politics of (Not-) Unpopular
Welfare State Reform in Advanced Capitalist Democracies, Amsterdam:
Dissertation series VU University Amsterdam.
Vis, B. and van Kersbergen. K. (2007) Why and how do Political Actors Pursue
Risky Reforms?, Journal of Theoretical Politics.
19:2 pp153-172.
Woldendorp, J, Keman, H. and Budge, I. (2000) Party Government in 48
Democracies (1945-1998). Composition, Duration, Personnel , Dordrecht:
Kluwer.
Wollmann, H. (2000) The Development and Present State of Local Government in
England and Germany, a Comparison, in H. Wollmann, E. Schröter (eds.)
Comparing Public Sector Reform in Britain and Germany , Aldershot: Ashgate.
19

Tables
Table 1. Overview of the categorisation dictionary which is used to identify the references to public sector reforms in nine policy areas
Category
KEY
Description
References to terms and phrases that are part of the discourse on public sector reform, in particular market principles. All cooccurrences between this ‘key’ and the following policy areas are interpreted as references to public sector reforms and are counted at the sentence level.
Agriculture
Culture
Fisheries; food (safety); farmers
Multi-culturalism; leisure; music; sports; press; film
Education Academia; schools; training; students
Environment Animals; emission; water; flora; fauna; forest, climate
Health
Housing
Hospitals; doctors; health care; mental health
Tenants; home buyers
Infrastructure Airways; energy; ICT; minerals; rail roads; cars; Telecom
Public order Judicial branch; crime; law enforcement; police; prisons
Social welfare Pensioners; social security; disabled; families; income protection
Note: See for the full list of words: http://home.fsw.vu.nl/pjm.pennings/wordstat_PMR.txt
20

Table 2. References to public sector reforms per decade and ‘family of nations’,
1981-2005 (%)
Country 81-85 86-90 91-95 96-00 01-05 Family of nations
Scandinavia Denmark
Finland
0.0
0.0
0.2
0.0
0.0
0.0
0.2
0.2
0.3
0.0
Norway
Sweden
0.9
0.0
1.3
0.0
1.4
0.0
1.4
0.1
1.4
0.4
0.30
1.4
0.53
1.4
0.67
1.8 Western
Europe
Mean
Austria
0.23
0.5
0.50
0.6
Belgium
Germany
Switzerland
Mean
Latin Europe France
1.0
0.7
Netherlands 1.4
0.6
0.88
0.0
1.0
1.0
1.8
1.5
1.23
0.6
1.3
1.6
1.8
0.9
1.42
0.5
1.4
1.9
1.8
0.8
1.34
0.1
1.1
2.1
0.9
0.7
1.25
0.6
Italy
Portugal
Spain
Mean
Anglo-Saxon Australia
Canada
Ireland
New
Zealand total mean total n
UK
US
Mean
2.0
0.7
2.3 2.1
1.29 1.09
0.4
0.9
1.3
1.0
1.3
1.8
1.03 0.96
0.85 0.91
866
1.7
0.2
0.6
0.6
1.0
1.1
1.4
1.2
1.5
2.4
2.1
1.61
0.8
0.6
0.5
1.6
1.7
1.8
1.07
1.0
1095 895
1.3
2.2
2.0
1.60
1.1
1.3
1.4
0.5
1.9
1.2
1.21
1.16
903
1.2
1.3
1.03
1.08
952
0.5
1.4
1.7
0.4
1.0
2.3
2.1
1.72
21

An univariate ANOVA of public sector reform by Family of Nations shows that
F=99.5. Sign.=.000. Eta=0.20.
Reading example: The mean number of references to market principles in the public sector in Danish manifestos in the period 2001-2005 is 0.3% of all references made in these manifestos.
22

Table 3. References to public sector reforms per decade and party family (%)
81-85 86-90 91-95 96-00 01-05 mean n (nr. of
Communist parties
Social democratic parties 0.9
Liberal parties
Christian democratic parties
0.8
1.0
0.7
0.5
1.0
0.9
0.9
0.9
1.2
0.8
1.2
1.4
1.1
1.1
1.2
1.3
1.2
1.1
1.0 parties)
0.92 484 (8)
1.10
0.98 799 (15)
1.01
1458 (24)
668 (11)
Conservative parties
Agrarian parties
F sign.
1.1
0.1
2.29
0.03
1.2
0.6
2.80
0.01
0.9
0.6
3.70
0.00
1.2
0.9
1.96
0.06
0.9
0.7
2.8
0.01
1.07
0.59
4.7
0.00
761 (13)
328 (6)
Reading example: The mean number of references to public sector reforms in Communist manifestos in the period 2001-2005 is 1.3% of all references made in these manifestos.
23

Table 4. References to public sector reform per decade, party family and family of nations (%)
Communist parties Scandinavia
Social democratic parties
Latin Europe
Scandinavia
81-85
0.2
1.1
0.1
86-90
0.3
0.6
0.6
91-95
0.2
1.5
0.2
96-00
0.8
1.7
0.3
01-05
0.9
1.6
0.5
1.5 1.4 1.2 1.1 Western
Europe
0.9
Latin Europe 1.4
Liberal parties
Anglo-Saxon 1.1
Scandinavia
Western
Europe
0.2
1.0
1.3
0.3
0.3
1.4
1.9
1.5
0.2
1.4
1.8
0.9
0.5
1.3
1.9
1.1
0.5
1.4
Christian democratic parties
Conservative parties
Latin Europe
Anglo-Saxon
Scandinavia
Western
Europe
Latin Europe 1.9
Anglo-Saxon 0.6
Scandinavia
2.2
1.5
0.1
0.9
0.6
1.3
0.7
0.2
0.9
3.0
1.8
0.7
1.2
0.4
0.2
1.6
2.3
0.0
0.4
.
1.2
0.3
1.6
1.3
0.7
.
1.6
0.5
1.3
1.5
.9
Agrarian parties
Latin Europe 0.8
Anglo-Saxon 1.3
Scandinavia 0.0
Western
Europe
0.6
1.3
1.7
0.4
2.0
1.4
1.2
0.3
1.7
1.2
1.4
0.4
1.1
Anglo-Saxon 0.0 0.7 0.7 1.5 0.5
Reading example: The mean number of references to References to public sector reform in Communist manifestos in Scandinavia in the period 2001-2005 is 0.9% of all references made in these manifestos.
1.4
0.7
0.5
1.8
24

Table 5. References to public sector reform per policy sector in relation to the change in public expenditures between the periods 1976-1985 and 1986-1995 (%)
76-85
Agriculture 0.6
Culture
Education
0.3
0.6
86-95
0.6
0.4
1.0
96-05
0.6
0.6
1.3 change (1) change (2)
-0.2
.
-0.2
-0.8
.
-.07
Environment 0.5
Health Care 0.3
Housing 0.1
Infrastructure 1.9
Public order 1.4
0.9
0.5
0.1
2.0
1.4
1.2
0.9
0.1
2.2
1.4
.
+0.2
.
-0.5
-0.3
.
+0.2
.
-1.5
-0.3
Social affairs 1.3 1.6 1.8 +.08 +1.1
An univariate ANOVA of public sector reform by Policy Sectors shows that F=97,
Sign.=.000, Eta=0.32.
Change (1)= change in expenditures as % GDP; change (2)=change in expenditures as % of total government expenditures.
Latin Europe only includes France and Italy.
Reading example: The mean percentage of references to Public sector reform in
Agriculture in the period 1996-2005 is 0.6%.
Source on expenditures: IMF, Government Finance Statistics Yearbook , various years.
25

Table 6. The development of between-group differences over time (Eta)
Category 81-85 86-90 91-95 96-00 01-05
Countries
Family of nations
0.36
0.21
0.37
0.18
0.44
0.32
0.41
0.24
0.49
0.26
Party families
Parties
0.14
0.45
0.13
0.50
0.14
0.51
Policy sectors 0.15 0.34 0.36
* These scores are statistically not significant.
0.09*
0.48
0.44
0.15
0.56
0.37
26

Notes
1
The digitised party manifestos (from 1960 onwards) are made available upon request at http://manifestoproject.wzb.eu under certain conditions regarding their usage. The data are part of the Comparative Electronic Manifestos Project
(http://home.fsw.vu.nl/pjm.pennings/) directed by Paul Pennings and Hans Keman,
VU University Amsterdam
, The Netherlands in cooperation with Zentralarchiv für
Empirische Sozialforschung (ZA) at the University of Cologne. This project has been financed by the Netherlands Organisation for Scientific Research (project # 480-42-
005) and partly by ZA.
2
See for a description of this software: www.simstat.com.
27