Why and How Do Politicians Do Risky Reforms? On the Basic Cause and Blame Avoidance Mechanisms of Welfare State Reform Paper Prepared for the Workshop Blame Avoidance and Blame Management: Institutional and Policy Implications of the ECPR Joint Sessions, Granada, Spain, April 14 19, 2005 Barbara Vis and Kees van Kersbergen Department of Political Science Vrije Universiteit Amsterdam De Boelelaan 1081 NL-1081HV Amsterdam The Netherlands Tel + 31 20 598 6821 Email: B.Vis@fsw.vu.nl Abstract The dynamics of welfare state reform are theoretically not well understood. In this paper we ask why and how politicians do risky reforms? Drawing on prospect theory we identify the basic cause of reform: policy-makers only take risks as the outcomes of the status quo are no longer perceived as gains but as losses. Prospect theory furthermore predicts that the condition under which welfare state reform takes place, i.e. when the status quo equals a loss, is identical across countries. To find the underlying causal mechanisms of reform, we turn to the theory of blame avoidance. We assess empirically the theoretical consequences of both theories and conclude, first, that politicians do not do risky reforms unless they consider that upholding the status quo is no longer a political (electoral) option and, second, that of the many strategies available, some are especially likely to be chosen by politicians.
1 Introduction One of the major puzzles of contemporary welfare state research is why and how drastic or radical reforms are taking place in spite of the institutional mechanisms and political resistance that work against change. Empirically, the field is very rich with detailed single case studies, comparative projects and large-n studies that have produced a host of hypotheses on the why s and how s of welfare state reform. On a theoretical level, however, there seems to be a problem with understanding the dynamics of reform, particularly with identifying the basic cause and the necessary causal (micro-) mechanisms that explain the link between cause and effect (reform). The central question of this paper is: why and how do politicians do risky reforms? Mainstream welfare state research informs us that it is plausible to assume that welfare state reform is always politically risky, so that there must be a very strong or convincing reason why politicians will embark upon risky reforms in the first place. We propose that prospect theory, which specifically deals with choice under uncertain and risky conditions, can give us a better grounding of the basic cause of reform. For a solution to the problem of causal mechanisms, we look for inspiration in the literature on blame-management and blame avoidance that particularly focuses on how political actors try to avoid the negative repercussions of their reform measures. We try to assess the empirical relevance of our theoretical findings against comparative data on welfare states over time. The paper is structured as follows. In section 2 we shortly discuss major studies of contemporary welfare state research and identify the basic theoretical problems that contribute to the puzzle of reform. In section 3 we turn to prospect theory to find the necessary condition under which risky welfare state reforms occur. Section 4 is devoted to the theory of blame avoidance and gives us the strategies of reform. Section 5 studies the empirical bearing of the theoretical findings. Section 6 is the conclusion. 2 Puzzles in welfare state development Theories from various intellectual backgrounds in the late 1980s and (early) 1990s forcefully argued and predicted the definitive crisis and final breakdown of the welfare state. They pointed to formidable challenges such as ageing populations, sluggish economic growth, long-term unemployment, changing family structures, the transformation of life cycle patterns, the postindustrialization of labour markets, the erosion of 1
systems of interest intermediation and collective bargaining, the rise of new risks and needs, and international pressures (globalisation and European integration). Such challenges and pressures necessarily would bring about major structural revisions as they threatened the welfare state s viability and subsistence. However, empirical reality contradicted these theories since the welfare state continued to exist, albeit perhaps functioning at a lower level of social and economic performance. As a result, the major explanatory puzzle in the mid-1990s became the theoretically unexpected but empirically observed persistence of the major institutions of the welfare state. Paul Pierson, one of the field s leading researchers, aptly summarized the problem: Underlining the severe pressures confronting mature welfare states does not, however, imply that the expected result is a collapse or radical retrenchment of national welfare states. Major policy reform is a political process, dependent on the mobilization of political resources sufficient to overcome organized opponents and other barriers to change (Pierson 2001: 411, our emphasis). Theoretically, and in accordance with mainstream political science, institutionalist approaches provided part of the solution to this puzzle by identifying some of the crucial mechanisms that explained welfare state persistence. Another part of the solution was found in the politics of blame avoidance (Weaver 1986). Taken together, a theoretically plausible answer was formulated to the puzzle. Major policy reform was not only very difficult because of the heavy weight of policy legacies, but also politically highly risky as it adversely affected (parts of) the electorate and vested interests. Major welfare state reform efforts time and again failed because of two major pro-status quo forces: first, the broad (electoral) support for core social programmes and accordingly the political unpopularity of cutbacks; and second, the rigidity of welfare state institutions that results from path dependent development and veto points that have the capacity to obstruct reform. Paul Pierson (1996: 178) argued that frontal assaults on the welfare state carry tremendous electoral risks and that retrenchment as a political issue should not be misunderstood as the mirror image of the growth of the welfare state. Welfare expansion usually generated a popular politics of credit claiming for extending social rights and raising benefits to an increasing number of citizens (i.e. the old politics), while austerity policies tended to affront voters and networks of organized interests and therefore triggered a politics of blame avoidance (i.e. the new politics of the welfare state). In other words, the post-1945 welfare state had produced an entirely novel institutional and political context. Once welfare programmes were solidly established, they 2
created their own programme specific constituencies of clients and professional interests. As a consequence, the emergence of powerful groups surrounding social programs may make the welfare state less dependent on the political parties, social movements, and labor organizations that expanded social programs in the first place (Pierson 1996: 147). Specialized social programmes in the policy areas of social housing, health care, education, public assistance, social security, and labour market management developed into institutionally separated and functionally differentiated policy domains. Therefore, a general weakening of social democratic and Christian democratic parties and the trade union movement the main historical supporters of welfare state expansion was unlikely to translate into a corresponding weakening of social policy. Supported by strong popular attachments to specific policies, professional policy networks were able to muster substantial veto powers against reform efforts. Moreover, given the political salience and popularity of social policy, it was not easy to turn a political preference of dismantling the welfare state into an electorally attractive proposition. Shifting the goals from expansion to retrenchment imposed tangible losses on concentrated groups of voters in return for diffuse and uncertain gains (Pierson 1996: 145). On average, retrenchment advocates thus confront a clash between their policy preferences and their electoral ambitions (idem: 146). Other researchers (e.g. Huber and Stephens 2001) empirically corroborated the Pierson-thesis and concluded that for austerity and retrenchment (public employment in Scandinavia is the exception) class and politics matter less and less, because an institutional rather than a political logic governed the adaptation of welfare states. However, by the late 1990s, a new puzzle was emerging: if institutional sclerosis and political deadlock were indeed so powerful in precluding change, why were there nonetheless empirical instances of (sometimes substantial or even radical) welfare state restructuring? What explained the reforms that did occur in spite of the theoretically plausible predictions of log jam? Most new analyses start from the institutionalist argument, and then try to add specific mechanisms of conditions under which substantial reform is possible. So, researchers stress the role of coalition politics (Bonoli 2000), the potential and risks of concentrated political power (e.g. in France, Great Britain and Switzerland; Bonoli 2001). They look at the particular institutional configurations that determine retrenchment (Swank 2001) or the domestic institutional refraction of international pressures (Kitschelt et al. 1999).They highlight the effect of the party system, party competition 3
and party strategy (Levy 1999; Ross 2000; Green-Pedersen 2001; Kitschelt 2001) and point to the role of framing, discourse, policy learning and ideational leadership in overcoming electoral and institutional resistance against major policy reform (Cox 2001; Schmidt 2000; see Green-Pedersen and Haverland 2002; Van Kersbergen 2002). The current state of affairs is such that many researchers are struggling to understand the dynamics of institutional resilience and reform. Indeed, many theoretical propositions and empirical hypotheses are proposed (and even tested), but all seem to face serious problems in elaborating the necessary causal (micro-) foundation of their argument. Let us give two examples. Scharpf (2000: 116ff) explains that the large-scale comparative project that he directed with Schmidt did not succeed in finding plausible political and institutional explanations of national adjustment strategies. He points to the unsatisfactory explanatory power of existing theories and explain their own failure to provide a better alternative by arguing that the irreducible contingency of the problem-institutions-actorspolitics-policy nexus (Scharpf 2000: 117) of national reform experiences is too complex to allow for causal inferences. But part of the complexity and contingency that the Scharpf Schmidt project suffered from was theoretically self-imposed, namely by conceptualizing in the best institutionalist tradition the causal relationship between institutions (or institutional conditions) and policy and policy outcomes as inherently of an contingent character. It seems to exclude the possibility of formulating a general theory from which specific empirical hypotheses can be derived. As Scharpf (2000: 22) himself put it: even though institutional differences may provide very powerful explanations for the success or failure of national policy responses, these explanations can be formulated only with regard to specific types of challenges. Scharpf s theoretical exercise is therefore phrased in the typical wording of challenges, vulnerabilities and institutional capabilities. It gives us the general picture of what kind of economic and social challenges national welfare states are facing, how vulnerable they are to the whims of the world market and how well adjusted the national institutional arrangements are. But it gives us very little to hold on to when we want to explain, for instance, why countries similar in institutional layout and facing comparable challenges, nevertheless perform radically different in reforming their welfare arrangements (for an interesting attempt see Kuipers 2004). A capability is a disposition, but for a causal argument one needs information on the strategy with which political actors attempt to realise the potential and avoid the likely backlash of policies that aim to reform the status quo. 4
A second example is taken from a co-authored work of one of us. Hemerijck and Van Kersbergen (1999) attempted to identify some of the mechanisms of reform: next to patching up institutions and transposing their use, they focused on policy learning. They argued that ideas and institutions are intimately related: institutions give us ideas; ideas are embodied in institutions. Policy learning is therefore institutionally nested. The institutional context of the welfare state is complex and consists of various policy domains, namely macroeconomic policy, industrial relations, social security and labour market policy, they may be loosely or tightly coupled. These domains vary in policy content and are governed according to different institutional logics. The relative ease and success of policy learning is contingent on the structure and character of the linkages between the different policy domains. This means that in some institutional contexts paradigmatic changes in ideas and policies happen slower and are less visible than in other contexts. They then proposed to view policy learning as informed by 1) policy failures, 2) anomalies and 3) the extent to which new information and insights into the relationships of cause and effect in various policy domains generate political windows of opportunity for reform. But although they were aware that policy making is a matter of both puzzling and powering, they treated reform to a large extent as an almost exclusively policy rather than also political issue and accordingly had very little to tell about the conditions under which those responsible and accountable for reform could be successful both in a policy sense and a political sense. In other words, theories of policy learning failed to problematise the political feedback mechanisms that either preclude learning to occur in the first place or inhibit action on the basis of new failure induced insights. So, on the one hand we have a body of literature that explains the difficulty, if not impossibility, of welfare state reform by pointing to institutional resistance and electoral risks, and on the other hand we have various fragmented hypotheses from policymaking, organizational and political-economy theories that try to identify the mechanisms, conditions and strategies of welfare state reform. It seems to us that we need stronger theory that can specify much better how under conditions of a viscous institutional context and a limited room to manoeuvre political actors succeed to overcome the resistance against change. And for this we turn to other sources of inspiration, particularly prospect theory and the literature that deals with blame-management and blame avoidance, in order to understand better the mechanisms that affect the political management of hazardous welfare state reforms. 5
3 Specifying the conditions of welfare state reform I: prospect theory Expected utility theory as commonly used in economics and in schools that employ economic theories such as rational choice institutionalism make predictions that do not adequately describe how people actually make choices under conditions of risk and uncertainty. More than 25 years ago, Kahneman and Tversky (1979) already offered an alternative, prospect theory, which was meant to provide a better account of choice under risk. As Levy (2003: 215) aptly summarized: Prospect theory deviates from expected-utility theory by positing that the way people frame a problem around a reference point has a critical influence on their choices, and that people tend to overweight losses with respect to comparable gains, engage in risk-averse behavior with respect to gains and risk-acceptant behavior with respect to losses, and respond to probabilities in a non-linear manner. In Political Science, prospect theory has been mainly applied in the field of International Relations (and occasionally in Comparative Politics) (see Quattrone and Tversky 2000; special issue Political Psychology 2004, issues 2 and 3), but precisely because of its stress on choice under conditions of risk and uncertainty we expect prospect theory to be particularly relevant for improving our theoretical understanding of the choices that politicians make when trying to reform welfare state programmes. There are a number of reasons why in the face of risk decision-making does not involve a rational deliberation over so-called end-states (as in expected utility theory), but rather a choice between gambles (i.e. prospects). First, there is the certainty effect, which means that people overweight outcomes that are considered certain, relative to outcomes which are merely probable (Kahneman and Tversky 1979: 265). If there is a chance of 50 per cent that one can win 300 and certainty that one wins 75, expected utility theory predicts that people choose the first option (expected utility of option 1 = 0.5 * 300 + 0.5 * 0 = 150, which is higher than the expected utility of option 2 = 1.0 * 75 = 75). However, experiments show that a theoretically unexpected but statistically significant high number of people choose the second option. Second, there is the reflection effect, which suggests that the preference order in the negative domain is the exact mirror (hence the term reflection ) of the preference order in the positive domain. This implies that how choices are framed matters for how preferences are ordered. Framing is ( ) the process by which a communication source constructs and defines a social or political issue for its audience (Nelson et al. 1997: 221). For example, when people are faced with the fact that there is 10 per 6
cent unemployment, they probably are ready to accept tough policy measures to fight unemployment. However, when this fact is reformulated in positive terms, namely that employment is at 90 per cent, they probably are less inclined to accept harsh measures to increase further the level of employment. In other words, framing the same fact differently reverses preferences. Finally, there is the isolation effect, which tells us that ( ) to simplify a choice between alternatives, people often disregard components that the alternatives share, and focus on the components that distinguish them (Kahneman and Tversky 1979: 271). Hence, voters disregard those elements in the manifestos of political parties that they have in common (e.g. retrenchment of health care) and focus on the distinguishing aspects of the programme. This implies that voters do not consider political alternatives as a whole (i.e. a party electoral manifesto), but focus on a comparison of political differences. A vote for a party manifesto, therefore, cannot be taken as a contribution to the mandate of that party. Given these psychological processes, prospect theory states that there are two phases in the choice process: an editing phase and an evaluation phase. During the editing phase, the decision is simplified by means of all kind of editing operations like coding in gains and losses, established against some neutral reference point. In the evaluation phase, a choice is made between the prospects. These are just the basics of prospect theory: the theory is more complex (complexity that we do not need for our purposes) and surely more complex than expected utility theory. The added value is, of course, that prospect theory better reflects how choices are made. Prospect theory namely argues that policy-makers avoid risks as long as the outcomes of the status quo are perceived as gains, i.e. as still acceptable or tolerable. Paraphrasing Berejekian (1997: 793), the theory yields two predictions. First, governments will opt for the certainty of the status quo, when they view the status quo as a gain (their position of power) and are confronted with a choice between a) the status quo (no reform) and b) some gamble (reform) with both a positive expected value (e.g. electoral gain) and some smaller risk of loss (electoral punishment smaller than the expected gain). Second, governments will opt for the gamble, when they view the status quo as a loss and are confronted with a choice between a) the status quo (no reform) and b) some gamble (reform) with both an expected value of further loss (further electoral loss) and some smaller prospect for improvement (an electoral reward smaller than the expected loss). 7
Hence, a corollary of this reasoning is that governments will only embark upon drastic welfare state reforms with risky electoral repercussions ( gamble ) if the status quo is considered a loss. Prospect theory therefore gives us the necessary condition under which welfare state reform can occur, namely that governments view the status quo as a loss. In contrast to mainstream welfare state theory that holds that welfare state reform varies according to the different institutional logics of welfare state regimes (liberal, social democratic, conservative), prospect theory predicts that reform takes place under similar conditions (status quo equals a loss) and therefore applies to all welfare states irrespective of their regime-type. Paraphrasing Berejekian (1997: 789) governments in a gains frame pursue absolute gains and are unwilling to engage in risky reform efforts, while governments in a losses frame pursue relative gains and are more willing to accept the risks of reform. This theory, then, gives us an explanation for the fact that governments embark upon risky reforms at all, that is for the empirical fact for which institutional approaches and especially rational choice institutionalism that employs expected utility theory have no account. Stated differently, prospect theory gives us the basic cause of hazardous reforms. However, the theory tells us little about a) the causal mechanisms (except inductivist psychological findings of laboratory experiments) that connect cause (the losses frame) and effect (reform) and b) other possible conditions that enable or restrain action (e.g. institutions), c) problems of choice at the aggregate rather then the individual level. We hope to find clues for solving these problems in what is rapidly becoming a theory of blame avoidance. 4 Specifying the conditions of welfare state reform II: the theory of blame avoidance Irrespective of whether or not policymakers find themselves in a losses frame, for their decisions they can be influenced by different and mutually exclusive kinds of motivations: 1) credit claiming (assuming vote seeking), 2) good policy (assuming policy seeking) and 3) blame avoidance (assuming vote seeking) (Weaver 1986: 372). Which motivation or strategy is chosen, is affected by the structure of party competition. Weaver (1986: 378) argues that the number of parties or candidates competing for votes ( ) may influence whether a party stresses credit-claiming or blame avoiding in its electoral appeals. In a two-party system ( ), the best strategy is probably to take ambiguous stands and duck divisive issues (i.e., to minimize blame) to avoid offending 8
marginal voters. In a multi-party system, on the other hand, some parties may be better off by taking pointed, controversial positions (credit-claiming) in order to build a distinctive political base and avoid becoming lost in a crowded field. However, we think that in certain contexts, especially those in which established social rights are object of reform, the motivation of blame avoidance seems most relevant to focus on. Policymakers are here most likely to search for blame avoidance strategies because it promises to attain the goal that they have been forced to set (reform), whilst avoiding the well-known risk of blame that accompanies such hazardous reform. As Weaver (1986: 372) argued: ( ) when push comes to shove, most officeholders seek above all not to maximize the credit they receive but to minimize blame. But can blame avoidance strategies in this sense be politically (rather than policy) effective? There are arguments that support a positive answer. The idea of negativity effects in political behaviour is important here. A negativity effect refers to the greater weight given to negative information relative to equally extreme and equally likely positive information ( ) (Lau 1985: 119). There appears to be a rising negativity bias both among voters and politicians (Hood 2002: 20). In addition, Pierson (2001), among others, has correctly argued that the benefits of reforms are widely dispersed electorally, while the losses tend to be concentrated among voter groups. We take this to imply that voters who are negatively affected by welfare state reforms remember such reforms much better and longer than reforms that affect them positively. Or, as Weaver (1988: 21) states, voters ( ) are more sensitive to what has been done to them than what has been done for them (italics in original). So, how to do things to voters and get away with it? Many empirical studies in one way or another refer to strategies of blame avoidance as a solution to policymaking under uncertain and risky conditions. Hood (2002), focussing on the link between the risk game and the blame game, discusses three strategies of blame management. First, there are presentational strategies or impression management, that is selecting arguments to minimize or avoid blame, for example in choosing between excuses intended to mitigate blame and justifications designed to turn blame into credit (Hood 2002: 16). Second, there are policy or substantive strategies, i.e. the selection of policy positions to minimize or avoid blame, for example choosing between policies that support risk creators as against those that support risk victims (idem). Third, through agency strategies, that is the selection of institutional arrangements to minimize or avoid blame, for example in choosing between direct control and delegation (idem: 17). Hood s point is that ( ) agency strate- 9
gies ( ) if successful ( ) eliminate the need for presentational or policy bias to achieve blame avoidance ( ) (2002: 17). In our interpretation, however, these strategies are not so much mutually exclusive, but rather hierarchically nested. That is to say, policymakers are most likely to try, first, to choose an institutional arrangement that minimises or avoids blame, second, if this fails, to alter policy positions, and finally, to present and phrase policies in a manner which minimises or avoids blame. This means that we do need a more detailed understanding of blame management or blame avoidance strategies, because empirically they may very well overlap. The theory of blame avoidance, which can be traced back to Downs (1957) theory of electoral competition, has had a clear focus on explaining why governments have a strong incentive to refrain from pursuing policies that are unpopular among voters. It was, in fact, one of the theoretical underpinnings of the welfare state resilience thesis. Interest group theory at the same time had been informing us that in addition to resisting to vetoing reforms, strong organized interests also reinforce the blame avoidance incentives of governments because trade unions and elderly lobbies are able to mobilize large segments of voters in elections (Hering 2003: 4, referring to Weaver 1986: 394 5). Indeed and interestingly, the resistance-against-change theorem was definitively one of the specific consequences of the politics of blame avoidance that Weaver (1986: 394) pointed to: Blame avoidance ( ) helps to explain why policies are so difficult to change, even if they fail. If policymakers and their constituents perceived costs and benefits symmetrically, they would be willing to change policies quite freely, at least as long as the new policies promised at least as high a surplus of concentrated benefits over costs as the status quo. But substantial vested interests often develop around programmes. Because costs and benefits are perceived asymmetrically, policymakers fear that new policies will not win them as much support as dismantling the old ones will lose. However, we think that of the strategies that Weaver identified (see table 1) to explain why policy change is so difficult, most of them may very well be strategies that can be employed to simplify or facilitate reform. 10
Table 1 Strategy Blame avoiding strategies Approach to avoiding blame (1) Agenda limitation (2) Redefine the issue (3) Throw good money after Bad (4) Pass the buck (5) Find a scapegoat (6) Jump on the bandwagon Prevent blame generating by keeping potentially costly choices from being considered Prevent blame generating by developing new policy options which diffuse or obfuscate losses Prevent or delay blame generating by providing resources to prevent constituencies from suffering losses Deflect blame by forcing others to make politically costly choices Deflect blame by blaming others Deflect blame by supporting politically popular alternative (7) Diffuse blame by spreading it among as many policymakers as possible Circle the wagons (8) Prevent blame generation by keeping credit claiming opportunities that conflict with policy preferences from being considered Stop me before I kill again Source: Weaver (1986: 385), table 3. Strategy 1, limit the agenda, basically tries to avoid that the very issue of a hazardous reform is politicized and therefore is not an option if reform is considered necessary. Strategy 2, redefine the issue, is primarily an attempt to spread the costs of reform and can just as likely be used as a reform strategy. Strategy 3, throw good money after bad, is basically a strategy that tries to soften the pain of reforms by offering a small compensation and can also be a strategy to help reform. Strategy 4, pass the buck, is delegation of the blame and includes also automatic government (see Weaver 1988). This, again, is a strategy for implementing reform that may help to minimise negative consequences. Strategy 5, find a scapegoat, is shifting the responsibility for reform measures. As indicated in the context of prospect theory, the European Union and especially the Economic and Monetary Union (EMU) are probably the best examples of scapegoats in welfare state reforms. Strategy 6, jump on the bandwagon, is swinging opinion towards the direction of the winning group. In the context of welfare state reform, this seems a strategy that either is not likely to occur or reinforces the difficulty of reform. Strategy 7, circle the wagons, is used when blame is inevitable and aims to share as much blame as possible, for instance by including the opposition in a reform package that is highly unpopular. Finally, Weaver s strategy 8, stop me before I kill again, is restricting the discretion over the choice. The EU s stability and growth pact 11
and the introduction of an independent European Central Bank are examples of such a strategy. This strategy is also a reform reinforcer. In sum, most blame avoidance strategies that were originally thought primarily to work against reform, can, in fact, be employed by politicians to ease the reform efforts that they undertake. 5 Assessing the empirical relevance of the theoretical findings We have now established that prospect theory gives us the reason why hazardous reforms take place and that the theory of blame avoidance provides us with a list of strategies with which politicians can deflect the negative feedback of such reforms. Next, we search for ways to find out whether these theoretical results have empirical plausibility. We start with prospect theory. On the basis of the three psychological biases ( certainty effect, isolation effect, and reflection effect ), we deduced that politicians introduce drastic reforms only when the status quo is considered a loss. Is it possible to assess the empirical relevance of this supposition? Obviously, this would involve a careful conceptual discussion of what exactly is a drastic reform, a painstaking operationalization of the concept and devising a clever way of measuring it. This is still a highly controversial issue among welfare state researchers and is also known as the dependent variable problem (for an overview see e.g. Green-Pedersen 2004). In addition, it would include a comparable exercise for the main independent variable, namely risky electoral repercussions in order to be able to establish if the status quo can be considered a loss. Budget cuts that reverse the trend towards increasing public debt and budgetary deficits can be considered drastic reforms because of the expansionary political and institutional logic normally present in public finances. Moreover, such cuts can also be considered as very risky, because of the high visibility and salience of fiscal policy, not only during phases of government formation, but also during the yearly political presentation of the government s plans. Another argument is that the criteria for entry into the EMU have forced the governments of EU member states (and other states whose currency is strongly linked to the Euro) to follow more radical adjustment policies than those that seemed electorally safe from the politician s short term perspective. In the context of national social policy changes, the EU already functioned as a whipping-boy for the politics of blame 12
avoidance in some countries (Anderson 1995), but increasingly became a major point of reference for the Thatcherite TINA (there-is-no-alternative) arguments that underpinned proposals of more radical shifts in national policy. National governments that tried to meet the harsh criteria of Maastricht, in our view, automatically ended up in a losses status quo. Therefore, budgetary cuts that reverse the trend towards increasing public debt and budget deficits indicate the losses context. Table 2a Budget deficits 1985 2000, selected countries 1985 1990 1995 2000 1985 90 1990 95 1995 00 Austria - 2.6-2.4-5.3-1.7 0.2-2.9 3.6 Belgium - 10.2-6.8-4.3 0.1 3.4 2.5 4.4 Denmark 0-1.0-2.3 2.5-1.0-1.3 4.8 Finland 3.3 5.3-3.7 7.0 2.0-9.0 10.7 France - 3.0-2.1-5.5-1.3 0.9-3.4 4.2 Germany - 1.1-2.0-3.3 1.1-0.9-1.3 4.4 Greece - 11.6-15.9-10.2-1.8-4.3 5.7 8.4 Ireland - 10.3-2.8-2.2 4.5 7.5 0.6 6.7 Italy - 12.7-11.8-7.6-0.6 0.9 4.2 8.2 Netherlands - 4.1-5.7-4.2 2.2-1.6 1.5 6.4 Portugal - 7.2-4.9-4.5-3.0 2.3 0.4 1.5 Spain 0 0-6.6-0.6 0-6.6 7.2 Sweden - 3.9 3.8-7.7 3.7 7.7-11.5 11.4 Great Britain - 2.9-1.6-5.8 3.9 1.3-4.2 9.7 Notes: A negative entry means that there is a deficit; a positive entry means that there is a surplus; changes are absolute changes. Source: To be added. If we look at Table 2a, we observe that of the 14 countries for which we had data, 9 were still increasing their budget deficits between 1985 and 1995. These 9 countries, however, clearly reversed this trend as a result of which every single country was reducing its budget deficit measured over the period 1995 2000. This policy reversal can be taken to indicate that there is indeed a losses frame and that there is therefore some empirical confirmation for the basic cause of risky welfare state reform. 13
Table 2b Public debt 1985 2002, selected countries 1985 1990 1995 2002 1985 90 1990 95 1995 02 Australia.. 22.6 42.2.... 19.6.. Austria 49.8 57.9 69.4 60.0 8.1 11.5-9.4 Belgium 118.8 125.2 129.8 106.8 6.4 4.6-23 Canada 66.3 73.5 99.2.. 7.2 25.7.. Denmark 74.9 65.8 73.9 49.0-9.1 8.1-24.9 Finland.. 14.5 58.1 43.2.. 43.6-14.9 France 37.9 39.5 59.4 57.1 1.6 19.9-2.3 Germany 41.6 42.0 59.1 60.5 0.4 17.1 1.4 Greece 47.2 89.0 108.7 99.8 a 41.8 19.7-8.9 Ireland 99.7 92.6 80.8 43.0 a - 7.1-11.8-37.8 Italy 82.1 103.7 123.1 107.0 a 21.6 19.4-16.1 Netherlands 68.7 75.6 75.5 64.5 6.9-0.1-11 Norway 34.6 32.4 41.1.. - 2.2 8.7.. Portugal 57.0 65.3 65.9 53.2 8.3 0.6-12.7 Spain 48.6 48.5 68.4 59.3-0.1 19.9-9.1 Sweden 64.6 42.9 77.2 58.0 a - 21.7 34.3-19.2 Great Britain 59.4 39.1 58.9 42.0-20.3 19.8-16.9 United States 53.5 60.9 68.3.. 7.4 7.4.. a 2001. Notes: Debt is general government gross public debt as a percentage of nominal gross domestic product; changes are absolute changes. Sources: Data for 1985 1995: Comparative Political Data Set 1960-2002 [OECD Economic Outlook 75, 2004]; data for 2002: Schuknecht (2002). Table 2b summarizes the development of the public debt and basically we find the same pattern as in table 2a. Of the 14 countries for which we had data for all the years, 1o were still increasing their debt between 1985 and 1995. All countries, except Germany, however, clearly reversed this trend as a result of which 9 countries were reducing their public debt measured over the period 1995 2002. Also this policy reversal can be taken to indicate that there is indeed a losses frame and that there is therefore some empirical confirmation for the basic cause of risky welfare state reform. As with respect to prospect theory, we need to ask if it is at all possible to assess the empirical relevance of the mechanisms of blame avoidance that may help to explain how the need to reform (in the context of the losses frame) is successfully translated. Let us, first, briefly discuss the empirical applicability of the reform concept. Pierson (2001) criticized mainstream welfare state theory for not capturing fully the multi-dimensionality of reform. He therefore reconceptualized reform as having three dimensions. First, cost containment is the attempt to keep balanced budgets through 14
austerity policies, including deficit reduction and tax moderation. Cost containment is a feature of all welfare states. Second, recalibration consists of reforms which seek to make contemporary welfare states more consistent with contemporary goals and demands for social provision (Pierson 2001: 425). This is an unlikely route for blame avoidance, because it belongs more to the expansion phase of the welfare state and is therefore more suitable for credit claiming. Levy (1999) has argued that some policies can turn vices into virtues. However, this is an option that can only occur if there are very (regime) specific vices that already lead to serious blaming, such as the disability scheme in the Netherlands (see Kuipers 2004). Finally, recommodification is the attempt to restrict the alternatives to participation in the labour market, either by tightening eligibility or cutting benefits (Pierson 2001: 422). Pierson (2001) argued that each regime (social democratic, liberal or conservative) is characterized by its own specific new politics of welfare state reform. So, in the liberal regime voters are less likely to be attached to the welfare state than in the conservative or social democratic models. Since the level of decommodification is fairly low already here, further restricting the alternatives to participation in the labour market (i.e. recommodification) is the politically most feasible feature of welfare state reform. In the social democratic welfare regime, voters are highly attached to, and dependent on, the welfare state. Recommodification is not so much on the political agenda of reform, but if only because of the sheer size of the public sector cost containment is. The conservative regime is probably the most ill-adapted model of the three worlds of welfare capitalism, as a result of which recalibration and cost containment are the two dimensions of reform that dominate. Here the issues are how to stimulate job growth in the underdeveloped service sector and how to contain the exploding costs of pensions, disability, and health. In contrast to Pierson, however, we think that like cost containment recommodification is a feature of all welfare reform efforts, irrespective of the regime, especially given the basic cause of reform (the losses frame) that we identified above. These are at the same time by far the most risky strategies, as a result of which the politics of blame avoidance should be most prominent here. In our empirical analysis, we therefore concentrate on recommodification and cost containment. Can we convert the expectations of blame theory to what we expect on the basis of mainstream welfare state theory? The strategy of redefining the issue, i.e. spending the costs of reform, can be recognized in recommodification policies that are presented as cost containment. Cost containment, while the demand for social expendi- 15
tures remains the same or increases, implies recommodification. We may assume that no government will ever defend reform policies of popular social programmes as explicitly aiming at reducing the level of protection from the market, that is as explicitly recommodifying. Instead, harsh retrenchment policies will always be portrayed as necessary efficiency measures (for impopular social programmes the opposite would apply). Although this is not uncontroversial, we conceptualise recommodification as the opposite of decommodification (cf. Bonoli 1998) and define the latter as ( ) a citizen s relative independence from pure market forces (Esping-Andersen 2000: 353). To measure the extent of decommodification, Esping-Andersen s (1990) index is widely used. This index incorporates 1) benefit replacement rates, 2) the number of qualifying weeks; 3) the numbers of waiting days; 4) the duration of the benefit; and 5) the coverage of the benefit for unemployment insurance, sick pay and pensions. The higher the score on this index, the higher the extent of decommodification. We therefore take that a lowering of this score through time indicates recommodification. Table 3 summarizes the development of the decommodification index between 1980 and 2002. On average, the changes in the three welfare regimes are rather small. 1 In the conservative regime, the extent of decommodification increased somewhat between 1980 and 1985 and then stabilised. In the social democratic regime, we see an increase between 1980 and 1985 as well, but here there is some recommodification between 1985 and 1990 and stabilisation from 1990 onwards. In the liberal regime, the extent of decommodification is stable until 1990, then converts into recommodification between 1990 and 1995 and decommodification between 1995 and 2002. 1 To make the averages as adequate as possible, we have not included Italy and Switzerland in any of the regimes but present their data separately. Whilst Italy is often placed in the conservative welfare regime, it has been convincingly argued that the Southern European welfare states have their own logic and should therefore be seen as a separate welfare regime (Ferrera et al. 2000). Unfortunately, Scruggs (2004) does not provide data for the other Southern European countries (Spain, Portugal, Greece) so we cannot include this regime. Switzerland is often placed in the conservative regime and, equally often, in the liberal regime. The former is probably correct for the period 1980 early 1990s; the latter probably for the early 1990s onwards. 16
Table 3 Decommodification index 1980 2002, selected countries 1980 1985 1990 1995 2002 Liberal Welfare Regime Australia 20.1 19.2 18.3 18.5 17.9 Canada 25.8 25.2 25.2 25.1 25.8 Ireland 21.8 25.8 22.2 23.1 28.9 New Zealand 23.8 24.5 24.4 22.0 22.9 Gr. Britain 22.9 22.4 23.8 24.2 24.7 US 18.6 18.6 19.0 18.1 18.1 Average 22.2 22.6 22.2 18.5 23.1 Conservative Welfare Regime Austria 27.8 28.4 27.3 28.1 28.8 Belgium 30.5 29.5 30.2 30.4 30.1 France 27.7 30.0 30.3 29.6 27.0 Germany 30.7 30.8 30.5 30.9 30.2 Netherlands 31.8 34.7 33.7 33.5 34.6 Average 29.7 30.7 30.4 30.4 30.1 Social Democratic Welfare Regime Denmark 33.0 34.2 33.3 35.0 34.9 Finland 27.9 31.4 30.4 30.3 30.1 Norway 33.5 36.7 34.8 35.0 37.3 Sweden 36.4 41.0 36.9 34.5 32.5 Average 32.7 35.8 33.9 33.7 33.7 Italy 20.6 24.0 24.2 24.3 26.7 Switzerland 32.2 32.3 30.2 28.4 21.9 Notes: See for the construction of the decommodification index Esping-Andersen (1990) and Scruggs and Allan (2004). Source: Scruggs (2004); averages own calculations. This picture of rather unspectacular changes transforms in a mixed picture if we focus on the whole time series 1980 2002 instead of on snapshots through time, and on individual countries instead of on welfare regimes averages. 2 What we can identify on the basis of these time series is, first, a group of countries (Austria, Belgium, and Denmark), where the increase of decommodification levels off over time, but where recommodification is not a feature of reform. Second, there is another group (Australia, New Zealand, Sweden, Switzerland) that is clearly characterised by recommodification. Third, the data show that there is a group of countries where either nothing changes or 17
the level of decommodification keeps increasing (Canada, Finland, Germany, Ireland, Netherlands, United Kingdom). The final group, then, consists of the countries that have very peculiar structural breaks in their development (France, Italy, Norway, United States). As stated, the decommodification index has different components so that a variety of reform measures can result in a change in the index. Most of the components changes are fairly transparent (replacement rates, waiting days, benefit duration, and benefit coverage) and are thus not likely candidates for blame avoidance. One, however, is much more non-transparent (qualifying weeks) and is thus a likely candidate for blame avoidance. The recently publicised Welfare State Entitlement Data Set (Scruggs 2004) presents data on the different components of the decommodification index, allowing us to see what caused the trend towards recommodification and the structural breaks identified above. Lets us begin with the countries displaying recommodification. In Australia, the only component that declines substantially is pension coverage, i.e. the portion of those above official retirement age who receive a public pension (Scruggs 2004). In New Zealand, the replacement rates for sick pay and unemployment insurance (UI) were increased between 1981 and 1986 (potentially indicating decommodification) and were then lowered (potentially indicating recommodification); the waiting days for UI were doubled in 1982 and subsequently remained the same; the waiting days for sick pay, finally, were doubled in 1991. In Sweden, a lot of changes took place but the steadily lowering of the replacement rates of UI, sick pay and pensions were probably critical. Also in Switzerland, there is a large number of changes of which the reduced sick pay coverage, measured as the percentage of the labour force with sick pay insurance (Scruggs 2004), is the most substantial. But also the increase in UI waiting days and the increased number of qualifying weeks for pensions are important here. In sum, only in Switzerland, some of the change in the decommodification index is caused by the invisible measure, qualifying weeks. Is this different in the countries that display peculiar structural breaks? The structural break in France s decommodification occurs in 1982, when the level suddenly increases substantially and remains stable until around 1996. Then another break occurs and decommodification decreases to a lower level than in the pre 1982 period. In 1982 it is the qualifying weeks for UI 2 Figures on which the statements that follow are based are available upon request from the authors. 18
that increase, but this recommodifying trend is offset by increased UI duration. The break in 1996 is caused by increasing UI waiting days and a lower sick pay duration. The Italian decommodification level is very low until 1983. After the break, there is a sudden increase and then stabilization between 1983 and 2000, when a further increase commences. Higher UI replacement rates cause the index to increase and offset these benefits extra waiting days. It is the withdrawal of these waiting days that caused the index s additional increase in 2000. The Norwegian decommodification index displays an increasing trend until the sudden decrease in 1990 that continues to the again sudden increase in 1998. It is difficult to precisely identify the causes of these changes because the different components show little alterations. However, the higher score in 1998 can be attributed to an extension of the UI duration. The American decommodification pattern shows a structural break in 1990, namely a decrease in the index. Like the Norwegian index, also here it is difficult to distinguish what instigated the structural rupture. Probably the lowered UI replacement rates and the lowered coverage of this benefit are the primary suspects. All in all, again, the only non-transparent measure, the qualifying weeks, are not important in the reforms. The strategy of throwing good money after bad, i.e. softening the pain of reforms, is difficult to translate into the terms of mainstream welfare state theory and also hard to operationalise and measure for quantitative comparison. However, we know that in many countries various cost containment measures were introduced varying from tightening the eligibility criteria and lowering the benefit replacement rates to the abolishment of whole social programmes that were subsequently followed by special measures to compensate partly the disadvantageous impact on income equality and poverty. However, cross-national studies of aggregate data and of survey data on income (the Luxembourg Income Study) indicate that generally speaking both incomes inequality and poverty have increased (see e.g. Huber and Stephens 2001: chapters 6 and 7). One may infer from this that most probably throwing good money after bad does not fully compensate retrenchment, but is used as a political strategy for its high symbolic positive potential. Pass the buck, i.e. delegation of the blame and/or automatic government, is another important strategy in welfare state reform. Retrenchment policies very often take the form of national framework laws that specify general objectives for lower level government to attain. To redefine competencies and delegate them to lower levels of gov- 19