Introduction
Over the past decades, much of the institutional literature on policy changes has emphasized that policy is characterized by long periods of stability periodically interrupted by exogenous shocks, which allow for more or less radical change. Institutions are seen as static and constraining. For instance, for welfare institutions, the ‘new politics’ literature has emphasized that welfare programmes always have their own constituencies, which makes it difficult to dismantle the welfare state. By doing so, institutional frameworks for policy analysis have been better at explaining continuity than change.
However, a growing number of scholars (Campbell, Reference Campbell2002; Pfau-Effinger, Reference Pfau-Effinger2005; Schmidt, Reference Schmidt2008, Reference Schmidt2010; Béland, Reference Béland2009; Béland and Cox, Reference Béland and Cox2011) have argued that the role of policy ideas and discourse should be taken seriously in explaining policy changes. This ‘discursive’ or ‘ideational’ institutionalism takes ‘account of the substantive content of ideas and the interactive processes by which ideas are conveyed and exchanged through discourse…including frames, narratives, myths, collective memories, stories, scripts, and more’ (Schmidt, Reference Schmidt2010: 3). Consequently, it defines institutions as cognitive constructs and structures. Unlike other forms of institutionalism more interested in the impact of external shocks, discursive institutionalism emphasizes endogenous ideational processes as the explanation for change. It represents a dynamic approach to policy change, in which change (or continuity) is possible through ideas and discursive interaction in politics.
Previous literature on discursive institutionalism is mainly based on theoretical arguments, but empirical evidence is so far rather scarce. In addition, prior empirical evidence has concentrated on economic policies (e.g. Hall, Reference Hall1993; Blyth, Reference Blyth2002), and those who have examined other policy areas (e.g. Béland and Hacker, Reference Béland and Hacker2004; Hiilamo and Kangas, Reference Hiilamo and Kangas2009; Kuivalainen and Niemelä, Reference Kuivalainen and Niemelä2010; Niemelä and Saarinen, Reference Niemelä and Saarinen2012) have focused mainly on policy documents in order to explore ideational shifts and the agenda-setting of different political actors. Even though these studies have enriched our understanding considerably by pointing out that a systematic analysis of ideational processes is necessary in order to get a full picture of institutional change, they have not thoroughly explained the mechanisms of how ideas or discourses matter. As Schmidt (Reference Schmidt2010: 21) argues, ‘[t]he research agenda for DI [discursive institutionalism], therefore, should not just be to seek to convince political scientists theoretically that ideas and discourse matter…but to show empirically how, when, where, and why ideas and discourse matter for institutional change, and when they do not’.
This case study provides an empirical contribution to the research agenda described above. Our aim is to analyse (1) how different actors use ideas to frame a policy alternative and hence, make their own alternative more appealing to the public, and (2) how effective different frames are, that is, to what extent competing frames are able to shape public opinion on the implementation of a specific policy change. Thus, on the basis of Campbell's (Reference Campbell2002) typology of ideas, this article explores ideas as frames, that is, normative or cognitive beliefs that are located in the foreground of policy debates. The question is how actors try to make their ideas ‘true’ (Somers and Block, Reference Somers and Block2005).
Discourse about the administration of social assistance in Finland provides a useful example of a specific policy issue that has brought about several competitive frames. Finnish last-resort social assistance is a municipal function. In the field of public policy, Finland is one of the most decentralized social policy regimes in the European Union: local authorities have far-reaching powers as well as significant budgetary independence. National legislation and directives from the Ministry of Social Affairs and Health provide a general framework within which each municipality can act according to its own decisions. As a consequence, there is a great deal of diversity in the local application of national legislation on social assistance, with practices varying from liberal to more stringent ones. Where the practices are more stringent, there is a substantial number of clients that do not receive the benefits to which they are formally entitled. Due to this failure, the proposition that the administration of social assistance should be centralized, that is, transferred from the 336 municipalities to one nation-state-level organization (the Social Insurance Institution of Finland, Kela), has repeatedly appeared on the political agenda. The political debate concerning centralization has continued with varying intensity for over 20 years. Our aim is to examine the effects of the various frames, ideational devices, which political actors have used in this debate in order to garner public support for their policy initiatives.
Our theoretical starting points will be discussed in more detail in the next section. Thereafter follows a historical survey that describes how arguments for and against the state model have been framed in political struggles. Historical scrutiny results in four frames, which refer to different ideational aspects of social assistance. The first, social work frame, pertains to the connection between social work and giving money and advice to the needy: the frame emphasizes that in the state model, the connection between monetary transfers and social work would disappear, leaving the recipients on their own and without the helpful advice given by social workers. The second frame, the income transfer machine, emphasizes that the state model, relying only on nation-wide legislation, would automatically deliver money to people without properly screening whether or not they really are in need, whereas the municipal model is more stringent. Thus, the formal practices of the state model would increase costs. Both of these first two frames are used by the opponents of the state model.
However, proponents of the state model argue that due to harsh needs testing based on local traditions and conditioned by municipal finances, needy people do not always get benefits. The centralized state model would provide people with better access to benefits to which they are entitled. We label this frame as a rightfulness frame. In fact, the second and third frames refer to the same phenomenon, but appeal to different ideational underpinnings (increasing costs vs. rightful treatment). The last frame, equality, emphasizes that the state model would apply the very same standards everywhere in the country and guarantee equal treatment regardless of one's place of residence, whereas in the current model the variety of municipal practices unavoidably leads to unequal treatment of clients.
The subsequent examination proceeds at two levels. First, we look back into history to see what the ideational frames used in policy debates are. We show that on two occasions there was an ideational shift in favour of a policy change. Second, after having constructed the historical frames we evaluate their effectiveness in an opinion survey. We use an experimental survey design with an opinion survey in which questions were formulated according to their historical frames in order to see how effective they are in present-day popular opinion on the issue. The different frames provided the respondents either with additional information on the subject, pertaining to increasing costs and problems with social work, or with references to more abstract ideas of equal treatment and rightfulness. Our general finding is that framing matters, which is in line with findings from earlier studies (e.g. Baumgartner and Jones, Reference Baumgartner and Jones1993), but we also show that some frames are more effective than others. The abstract frames appealing to moral sentiments are more effective than frames relying on factual arguments. The moral frames have wider valence (cf. Stokes, Reference Stokes1992), that is, there is a much broader agreement on moral values than on more precise factual statements. Whereas abstract moral ideas had the strongest impacts on opinions and increased the certainty of the responses, references to escalating costs and problems with social work often led to ambiguity and confusion, making it more difficult for respondents to formulate their opinion. These empirical results of the survey are presented in the penultimate section.
Furthermore, we argue that successful framing alone is not enough to bring about a policy change. Two conditions will be discussed: (1) timing and (2) political power. First, in the literature on institutional change punctuated equilibrium is a central concern. For long periods institutions are stable, but this equilibrium is occasionally punctuated by factors such as wars or sudden economic crises. As a consequence an institutional change occurs. In the Finnish case, the deep economic crises of the 1990s and 2008 formed a critical period that might have punctuated the equilibrium. But as we shall show, the popular idea of shifting from municipal to state-based social assistance was threatened by the gloominess of the crises. The opponents of the state model successfully used their ‘income transfer machine’ frame, and in both cases the reform proposals faded out. Second, ideas as such are not enough. As argued by Max Weber (Reference Weber1989 [1904–1905]: 90), those ideas that can be effective in history must be supported by predominant social groups. If the frames are not supported by powerful policy actors, the ideas will be inefficient. This notion brings power of ‘old politics’ back into policy-making (Korpi, Reference Korpi1978; Pierson, Reference Pierson1994; Korpi and Palme, Reference Korpi and Palme2003). The concluding section discusses our findings on the role of ideas, crises, and politics.
The role of frames in the policy process and public opinion
In ‘old politics’, when explaining differences in welfare state institutions, emphasis was laid on political power, whereas the proponents of ‘new politics’ pinpointed the role of institutions and path dependency created by previous social policy solutions. Institutions were seen as sticky and hard to change. However, what we often see in real world is institutional change that sometimes can take rapid and abrupt forms (see e.g. Kangas et al., Reference Kangas, Lundberg and Ploug2010). To explain changes, institutionalists introduced the concept of punctuated equilibrium, that is, in some moments the institutional stickiness is melted as a result of external shocks or changes in ideas or the interaction of these two factors. In this study, we regard the relationship between ideas, external shocks, and ‘old politics’ as dialectic. External shocks – like economic crises – may fortify or nullify some ideas and ideas must be adhered to by strong groups to have political effects (Geertz, Reference Geertz1972: 314).
In political debate, politicians constantly frame their proposals by wrapping them in a proper ideational disguise. The premise of framing theory is that a policy issue can be viewed and interpreted from a variety of perspectives, and framing ‘refers to the process by which people develop a particular conceptualization of an issue or reorient their thinking about an issue’ (Chong and Druckman, Reference Chong and Druckman2007: 104). Previous research on framing and its effects can be roughly divided into two branches: studies on discursive institutionalism and studies on opinion formation. These two branches overlap but have different focuses and areas of interest. Our study is in the intersection of the two research orientations.
The first branch, discursive institutionalism, analyses policy debates in order to examine the formation and impacts of framing. In these studies the focus is on different attributes of the issue, on arguments in favour of policy prescriptions, or on actors involved in policy debates (Surel, Reference Surel2000; Druckman, Reference Druckman2001; Schmidt, Reference Schmidt2002; Baumgartner and Mahoney, Reference Baumgartner and Mahoney2008; Hiilamo and Kangas, Reference Hiilamo and Kangas2009).
From an ideational perspective, frames can be understood as normative concepts that elites use to legitimize policy choices to the public (Campbell, Reference Campbell2002). Politics is often competitive, fought between ideological factions and interests, and the issues that are debated are framed in opposing terms. Framing is an essential part of the political battle – or of the ‘struggle of interpretation’ as Pfau-Effinger (Reference Pfau-Effinger2005) labels it – in which linguistic concepts and symbols give conceptual meaning to an issue and steer political debate. Ideational framing is a strategic process that aims to create the basis for political decisions and to help political actors legitimize decisions to their constituencies. Further, with the help of framing, political actors may create, that is, socially construct, the need to reform. Their strategic frames often draw on society's cultural repertoire in order to sell policy ideas or alternatives to the public and to key interest groups (Cox, Reference Cox2001; Béland, Reference Béland2009). This is especially important in the stages of issue/problem definition and agenda-setting, that is, when determining which questions are qualified as important enough to be placed on the political agenda.
Hence, policy actors who are able to establish the frame to be used exercise decisive power on the motivational basis of policy debates (Fairclough, Reference Fairclough1989; Bourdieu, Reference Bourdieu1991). These ideas constitute paradigms that form the basis for cognitive and ontological assumptions. This means that policy ideas can also be regarded as path dependent as, for example, Keynesianism and monetarism were in economic thinking.
We can also speak of hegemonic discourses: certain frames or ways of thinking can become ‘naturalised’ and achieve the status of common sense, and in consequence become invisible even to the actors that launched them (Gramsci, Reference Gramsci1971; Bourdieu, Reference Bourdieu1991; Baumgartner and Jones, Reference Baumgartner and Jones1993; Steensland, Reference Steensland2006). Good examples of this phenomenon are the abovementioned economic meta-paradigms that have epistemic privilege as Somers and Block (Reference Somers and Block2005) express it. Due to the hegemony of certain ideas, in order to change something in society, the structures and content of thinking must also be changed. However, with regard to mechanisms that explain how and when ideas or discourse matter, ideational path dependency means that policy proposals must be framed in normatively acceptable ways. Ideas must fit into the general ideological climate. This is where the ability to refer to and embellish central values and symbols embedded in a society's cultural repertoire is essential (Cox, Reference Cox2004; Béland, Reference Béland2009). Instead of referring to a specific group interest, political actors must refer to more widely accepted cultural values such as freedom, social justice, and equality, and wrap their messages in these frames.
In the second branch of research on framing, focusing on public opinion, the main purpose is to examine how opinions are formulated and how competitive frames impact public opinion (Smith, Reference Smith1987; Rasinski, Reference Rasinski1989; Kangas, Reference Kangas1997). From the perspective of policy change, it is crucial to examine the impact of frames on public opinion because, as argued above, framing strategies are used by policy actors to sell their proposals to the public. Further, public opinion research has consistently shown that opinions change easily depending on how questions are phrased. This concerns the context more than the content of the information (van Gorp, Reference van Gorp2007). By presenting information in a new context, people may realize connections they had not been aware of before. This may lead the respondents to reconsider and possibly revise their opinions (Tourangeau and Rasinski, Reference Tourangeau and Rasinski1988).
Contextualization and additional information are particularly important when respondents are confronted with questions they have not thought of, or if they are asked to express their opinion on issues about which they have no specific knowledge or information. It is here that political framing's use of phrases, metaphors, and symbols become crucial. As individual voters are unable to take into account all possible aspects and collect all possible information on different policy options – or if they simply do not know enough about the issue – they react to these metaphors and use them as shortcuts for their political choices. The nature of the frame is therefore crucial.
In many cases, opinion surveys present respondents with questions on issues about which they have very little knowledge. One way to circumvent this problem is to include some clarifying information in the survey. However, in some cases the information given may result in ambivalence and hesitation, which is mirrored in the increasing number of ‘I don't know’ responses. The additional information may complicate the picture too much and hence increase uncertainty. We will provide empirical examples of this.
Much of previous empirical research on the impact of framing on opinion formation on welfare attitudes has concentrated on attitudes towards redistribution and claimants of welfare benefits, as well as on the social legitimacy of different types of social benefits and government spending. However, there is a lack of studies that focus on frames and their effects on specific policy proposals. Yet studies of this kind are important in order to get a fuller understanding of the impacts of framing. Thus, we need studies on the mechanisms that explain how ideas or discourses matter for institutional change. Discourse on the implementation of social assistance in Finland offers a good case for the experimental study of debate on a specific policy issue in which several competitive frames have been used to garner support both for and against the policy change. In the next section, we move on to give a short historical summary of how centralizing the administration of social assistance has been debated in Finnish politics.
Social assistance on the political agenda
Most of the basic income transfer schemes in Finland are organized at the state level, that is, they are handled by the Social Insurance Institution (Kela). However, in keeping with early poverty relief tradition, last-resort social assistance is organized by municipalities and handled by municipal social workers. National legislation provides a frame of reference within which the municipalities can have their own practices of paying out social assistance, depending on their local traditions and financial possibilities. Due to local procedures and to the complicated application process, the non-take-up rate of social assistance is rather high: estimations vary between 40% and 50% (Bargain et al., Reference Bargain, Immervol and Viitamäki2007). In other words, 40–50% of those who, in principle, have the right to social assistance do not receive the benefit. There are also concerns that social assistance applicants are treated unequally, because the application process and the severity of means-testing vary depending on the municipality of residence.
Concerns related to non-take-up (rightfulness) and equal treatment (equality) were brought up in several expert committees and working groups in the late 1980s. In 1989, the Finnish Ministry of Finance assigned a working group to investigate measures to improve the accessibility of social services. As a part of this project, the working group organized a temporary experiment in one medium-sized town, in which both primary social benefits and last-resort social assistance could be applied for in the local Kela office (Ministry of Finance, 1991: 27–28). This experiment was soon followed by two broader experiments: the first took place in 1993 in 12 municipalities, and the second from 1995 to 1997 in 34 municipalities. In both experiments, the goal was to simplify the application process by allowing claimants to submit their social assistance applications directly to Kela offices instead of municipal social workers.
The results from the experiments were generally considered to be positive. The application processes became more flexible and processing times shorter, while claimants found making their submissions at Kela offices to be less stigmatizing than doing so at municipal social welfare offices. (Committee Report, 1993: Appendix 2; Ministry of Social Affairs and Health, 1996). In 1997, the management group responsible for the latter experiment recommended that the administration of the fixed part of social assistance should be transferred from municipalities to the state level. This way, applying for social assistance would be made easier and citizens would receive equal treatment. As a consequence, the non-take-up rate would decrease, as the benefit would be guaranteed homogenously as a social right by a state-level agency and not on the basis of varying municipal practices. (Ministry of Social Affairs and Health, 1997).
The experiments coincided with a historically severe economic depression, which led to a rapid increase in social assistance applications: between 1990 and 1996, the number of persons receiving social assistance nearly doubled, rising from 314,000 to 610,000, which was almost 12% of the population (Virtanen and Kiuru, Reference Virtanen and Kiuruc2010). Taking into account the economic situation, the representatives of the Ministry of Finance rejected the state model, claiming that the facilitation of the application process would inevitably lead to a further increase in social assistance expenses (Ministry of Social Affairs and Health, 1997: 50–53). Easy access to the benefit would turn the social assistance scheme into an income transfer machine that would more or less automatically distribute money to all applicants. There was also a fear that in the state model the connection between financial aid and social work would be lost. Thus, the frames relating to increased expenses and the connection to social work were pitted against the frames emphasizing equal treatment and rightfulness. As Finland was only starting its recovery from a deep depression, the fear of increasing costs became a strong argument in the political discourse, and there was no political will to go through with the extensive reform. In the end, only the Centre Party was in favour of the state model. The experiments were cancelled by the end of 1997, and the Act on Social Assistance was instead reformed to include stronger work incentives.
The question of the state model resurfaced on the political agenda in 2007, when the government appointed the Committee for reforming social protection. Its mission was to propose substantial reforms to basic income security and taxation, with the goal of improving work incentives, reducing poverty, and providing a sufficient level of basic security in all phases of life. The state model was eagerly discussed in the committee's reports released in 2009. The framing of the topic was almost identical to that seen 15 years earlier: the committee brought up the ideas of equal treatment (equality), problems with the low take-up rate of social assistance (rightfulness), facilitation and speeding up of application processes, and improving the resources of social work. The majority of the committee, including representatives of the Ministry of Social Affairs and Health, the Social Insurance Institution, the Centre Party and the Green League were all in favour of the state model. However, representatives of the Ministry of Finance, the conservative National Coalition Party, and all labour market organizations (employers’ as well as wage earners’ organizations) were against the state model, pointing to increasing costs (the income transfer machine) and the loss of connection to social work. Due to internal disagreements, the committee was unable to make any proposals on the issue (Ministry of Social Affairs and Health, 2009). Opposition parties were not represented in the committee. Of these excluded parties, the Social Democrats were against the state model, whereas the Left Alliance was more split on the issue, with some of its members strongly in favour and some strongly against it. The True Finns did not voice strong opinions on the model (for a description of the Finnish political parties, see footnote 2).
When the centralization of social assistance made its way to the political agenda in the 1990s and again in the late 2000s, the arguments supporting the state model were similarly compelling. Although centralization was supported by experts as well as the public, attempts to reform the system proved futile. There are a number of explanations. Both times that reform seemed possible, Finland experienced a severe economic recession that lent more credibility to arguments related to increasing costs than to arguments related to expanding social rights. In these two cases, the external shocks contributed to institutional continuity instead of causing a policy change. The shocks ‘punctuated’ the innovative ideas. Moreover, labour market organizations, which are very influential in Finnish policy-making, have opposed the state model. Unlike politicians, the social partners do not need to compete for voters – instead, they can use their institutional power to veto even popular policy proposals.
Historically, the political discussions on the state model have utilized four main frames. On the one hand, there is a discursive continuum that builds on equality (all clients should be treated equally) and specificity (everyone has their specific needs that must be taken into account individually). On the other hand, there is a continuum between social rights (everyone should get what the legislation promises) and costs of social assistance (from a financial perspective, the high non-take-up rate is a good thing). Thus, legalism based on social rights competes with cost-awareness. The fourfold table in Figure 1 summarizes the discursive frames used in political debates on the issue.
Figure 1 Frames of the implementation of social assistance.
The proponents of the municipal model have argued that Kela's impartial, mechanical decisions are too sloppy and do not take means-testing seriously enough. Therefore, many of those not eligible now would qualify for benefits under the state model, which would increase costs considerably. This idea of state model as an income transfer machine is effectively used as a framing device against the centralized model. Another frame supporting the municipal model builds on the connection between monetary payments and social work. In the municipal model, all those who need social assistance must face a social worker who gives them advice and helps them back towards economic independence. The consequent argument in this social work frame is that the state model would destroy the connection between social work and the monetary benefit, and hence the advice/control of the social worker would disappear and claimants would be left on their own.
As the historical overview indicated, supporters of the state model have utilized two different but interlinked frames: rightfulness and equality. In the rightfulness frame, the municipal system is blamed for having too much variation due to differences in local practices: access to help does not depend on the law, but rather, on local circumstances. Legal regulation is therefore often nullified by local decisions and many of those who by law should be entitled to benefits do not receive them, resulting in a high rate of non-take-up. The equality frame, in turn, emphasizes that due to variation in municipal practices, claimants living in different municipalities are not treated equally. Because Kela, as a national actor, would have the very same homogenous procedures throughout the country, the state model would guarantee the equal treatment of people in need regardless of their municipality of residence.
The debate on which agency is the proper one for delivering social assistance benefits is a good example of political framing. In practical policy-making, each actor tries to frame the issue in a way that promotes their own political goals (positive framing) and undermines their opponents’ arguments (negative framing) (see Chong and Druckman, Reference Chong and Druckman2007: 120–121). As described above, while proponents of the municipal model have presented the state model in an income transfer machine or social work frame, supporters of the state model have referred to rightfulness and equality frames.
Research design: questions and data
The aim of the study is to experimentally examine what kind of an impact the different frames used in the policy debates have on respondents’ attitudes to social assistance administration in Finland. What are the most effective discursive devices to manipulate public opinion? Is it better to appeal to facts or to moral sentiments? What will happen if we increase the amount of information contained in the questions?
The data are based on telephone surveys conducted in late 2008. The data are representative of the Finnish population between 15 and 79 years of age. To measure the impact of the frames, the sample was stratified into five different groups. The unframed questionFootnote 1 was posed to 1500 respondents who could answer on a 5-degree scale (1, absolutely; 2, maybe; 3, maybe not; 4, absolutely not; 5, I do not know). The margin of error in this part was ±0.9–3.2 percentage points.
In addition, this same question was framed in four different ways using the arguments that were found most common in the political discourse. These frames were then presented to four separate sub-samples, each of them consisting of 500 respondents, so that each sub-sample answered a different question. The income transfer machine and social work frames were meant to support the municipal model, while the rightfulness (non-take-up) and equality frames were hypothetically in favour of the state model. The answer options were the same as in the unframed question and the margin of error in these sub-samples was ±1.3−4.5 percentage points. The literal translation of the questions was as follows:
1. Income transfer machine: ‘Today municipalities are responsible for paying out social assistance. It has been discussed that the administration of social assistance should be transferred to Kela. However, this might entail easier access to benefits, which might consequently increase economic costs. What is your opinion: Should the duty of paying out social assistance be transferred to Kela, although the costs might increase?’
2. Social work: ‘Today municipalities are responsible for paying out social assistance. The paying out of social assistance is also linked to social work. It has been discussed that the administration of social assistance should be transferred to Kela. This might result in the loss of connection between money transfers and social work. What is your opinion: Should the duty of paying out social assistance be transferred to Kela, although the connection between social work and the monetary benefit would disappear?’
3. Equality: ‘Today municipalities are responsible for paying out social assistance. There are differences between municipalities in the practices of paying out benefits. It has been discussed that the administration of social assistance should be transferred to Kela. What is your opinion: Should the duty of paying out social assistance be transferred to Kela, in order to bring about equality amongst all citizens?’
4. Rightfulness (non-take-up): ‘Today municipalities are responsible for paying out social assistance. There are various local practices and many do not receive the benefits to which they are entitled. It has been discussed that the administration of social assistance should be transferred to Kela. What is your opinion: Should the duty of paying out social assistance be transferred to Kela, in order to avoid non-take-up?’
Results
According to the basic, unframed question, a slight majority of respondents are in favour of the state model: 57% of the answers were in the affirmative (‘absolutely’ or ‘maybe’). Only 25% were against the model, whereas as many as 18% could not formulate an opinion (see Table 1). Women were slightly more in favour of the model than men. When it comes to the respondents’ political affiliations,Footnote 2 supporters of the Left Alliance took the most positive view of the model, followed by voters of the conservatives and the Social Democratic Party (SDP). The lowest support was found among the Greens and the True Finns. The voters of the Left Alliance were more polarized in their opinions than other respondents: almost 30% of them were against the state model. The proportion of the ‘I don't know’ answers was the lowest among the voters of the Left Alliance (<4%) and the highest among the True Finns (28%). All in all, the proportion of those who did not take a stance on the issue was rather high: more than a fifth of the respondents found the question too difficult to answer.
Table 1 Support (%) for the state model in different frames
Yes = in favour of the state model; No = against the state model; DNK = do not know; SDP = Social Democratic Party.
Viewed against a historical background, the results are interesting. Even though the Centre Party was the most eager proponent of the state model, its voters were no more in favour of it than the supporters of other parties. Interestingly enough, grassroots conservatives were the second firmest supporters of the state model, despite the fact that the conservative National Coalition Party itself has held a more negative opinion towards the centralized model, as has the employers’ central organization, the Confederation of Finnish Industries, which is politically close to the party. The same goes for the Social Democrats, who often have shared the negative attitude of the Central Organisation of Finnish Trade Unions.
As can be seen in the Table, framing matters a lot. The reference to increased economic costs (income transfer machine) decreased the state model's support to 42%, while the share of negative opinions increased it to 35%. The voters of the Centre Party and the conservatives were the most worried about the costs, and among them municipal practice was more popular. The social work frame also eroded the popularity of the state model: the answers were rather evenly distributed (37% for and 35% against). This frame had an especially strong appeal for women who, unlike men, expressed more support for the municipal system than the centralized one.
Reference to equal treatment (the equality frame) – which the centralized model would include – increased support for the model: as many as 70% were now ready to transfer the administration of social assistance to Kela, and again women were more affected by the frame than men. The most positive opinions in this frame were seen among the conservatives. The impact of the rightfulness (non-take-up) frame was also significant – now 71% of the respondents gave their consent to the state model, the left-wingers, the Social Democrats and the True Finns being the most eager and the conservatives being the most sceptical. In fact, among the conservatives, this frame produced the second lowest support rate for the model, second only to the income transfer machine frame.
The results presented in Table 1 are based on simple bivariate cross-tabulations. In order to take into account the impacts of various confounding variables, the respondents’ socio-economic status, educational attainment, income, gender, and age were controlled for in order to see the ‘pure’ impact of the frames and of political affiliation. We applied the general linear modelling (GLM) approach.Footnote 3 For space considerations we only show results on interactions between party affiliation and the frame. The GLM analyses showed that the frame was the most important factor (sig. = 0.000), while gender was not significant (sig. = 0.840). High-income earners were more reluctant to support the state model than those in lower income brackets, but the association was not significant (sig. = 0.520). Neither education (sig. = 0.120) nor socio-economic status (sig. = 0.520) was significant in relation to opinion, although those who were outside the labour market (pensioners, students, and home-makers) showed more support for the state model than the other groups.
Party affiliation as such was not significant (sig. = 0.678), but inspection of the interaction (sig. = 0.107) between party and frame gave some interesting results. These observations are visualized in Figure 2, which displays the interaction between opinion, party, and frame after controlling for age, gender, educational attainment, socio-economic position, and income. The figure can be interpreted as follows: the higher the value of the marginal mean, the more positive the attitude towards the state model.
Figure 2 Support for the centralized state model (estimated marginal means). SDP = Social Democratic Party.
The strong impact of framing can easily be seen in the figure. As a rule, the impact of the frames tended to be similar among all respondents. Some interesting deviations can be seen, however. Those who vote for the conservatives reacted differently to some of the frames. Although the equality frame seems to have been an effective device to increase support for the state model among the conservatives, they reacted very negatively to the rightfulness (non-take-up) frame, very differently to the other respondents. It is possible that the conservatives see the non-take-up issue in light of increasing costs – cost containment comes first – while the others interpret the situation from a more moral (rightfulness) perspective: claimants should get what they are entitled to, regardless of the costs.
From a policy-making point of view, the inspection of ‘I don't know’ answers is of utmost interest. Table 1 shows that nearly 18% of the respondents had no opinion on the unframed question. Hinting at the extra costs that the state model would possibly cause increased the uncertainty to 24% and referring to social work had an even stronger impact, increasing the share to 28%. Interestingly, if we frame the question with references to moral aspects (rightfulness and equality), the share of uncertain respondents drastically decreases to <13%.
Hence, the interim conclusion could be that the frames pertaining to the increase in costs and to the mismatch between social work and paying out benefits created confusion among the respondents. The cost aspect had an especially strong effect on the voters of the Left Alliance, who in the initial unframed question were the strongest supporters of the state model. Among them, the income transfer machine frame increased the share of ‘I don't know’ answers to one-third, which is a significant increase from 4% yielded by the unframed question. Also the social work discourse created some ambiguity among the left-wingers. The same goes for the True Finns.
For all parties, the moral framing led to the lowest proportion of uncertain responses. Supporters of the Left Alliance were absolutely certain in their opinions when it came to the equality and rightfulness frames: the share of ‘I don't know’ answers was now zero! Again, a similar trend was visible among the True Finns: about 30% of them were uncertain when faced with the unframed question or with the social work frame, but the moral framing brought the share down to 6%. The pattern was the same for other parties as well, however within them it was less pronounced.
These results reflect the fact that for the respondents, it is easier to construct an opinion on the basis of moral sentiments (something is right or wrong) than factual statements (do we need social work in the context of paying out money, or do we need to care about costs when decreasing the non-take-up rate of social assistance). The extra information included in the factual questions seems to have led to increasing uncertainty instead of clarifying the picture (cf. Festinger, Reference Festinger1957). This is true in particular for the voters of the True Finns and the Left Alliance – while the latter group was the most eager proponent of the state model and had the smallest share of uncertain answers in the unframed question, discourses linked to increasing costs and to social work put left-wingers in a difficult position, increasing uncertainty.
On the level of political rhetoric and agenda-setting, our results indicate that to be successful, a politician should avoid presenting too many facts or approaching the issue from too many different and difficult viewpoints. Instead, one has to simplify the issue and effectively refer to moral sentiments that appeal to people's views on what is right and what is wrong.
Discussion
The starting point of the article was in the growing field of discursive institutionalism, which emphasizes the importance of ideas in policy implementation. This specific case study, based on experimental survey design, focused on the centralization of the administration of social assistance in Finland. The aim of the study was to empirically examine how ideas and competing discursive frames are used in policy-making and which kind of framed ideas have the strongest effect on people's opinions on the subject. The results show that framing matters. In principle, a majority of the population is in favour of the state model, but references to increasing costs (due to the decrease in non-take-up) and to the absence of social work weaken support for the policy change.
Even though the majority of the population is in favour of the centralized model, the way in which the question is presented has a huge impact on the strength of the support and on the respondents’ rates of uncertainty. When the political discourse was based on increasing costs or social work frames, support for the state model decreased. It is interesting in this context that almost one-fourth of the respondents expressed no opinion on the issue: it is evident that the additional information provided by the frames increased uncertainty among the respondents. If, instead of ‘factual’ statements, moral sentiments (equality and rightfulness) were referred to, the state model received an overwhelming majority and, more importantly, the share of hesitant respondents significantly decreased. Thus, referring to costs or to social work created a state of ambiguity, as respondents who were in favour of the model were confronted with information that was against their original stance. The references to social justice had the opposite effect. Thus, some ideas that frame the political discourse are more effective than others. At the increasing level of abstraction towards moral sentiments, policy ideas become more effective. First, they diminish uncertainty; on the basis of their moral judgement, people can take a strong stand on an issue on which they would, under other circumstances (e.g., when confronted with a number of factual statements), be more ambivalent. Morality surpasses ignorance. Second, moral frames have a wider valence (cf. Stokes, Reference Stokes1992), that is, there is a much broader agreement on moral values than on more precise factual statements.
Our starting point was Schmidt's (2010: 21) idea to ‘show empirically how, when, where, and why ideas and discourse matter for institutional change, and when they do not’. The historical part of our analysis displayed that there are indeed some persistent frames that have crystallized or institutionalized ideas of equality and rightfulness, as well as cost containment and the right to receive help from a social worker. Thus, the study is a success story in its illustration of the impact of political ideas on people's attitudes: by applying different ideational underpinnings, elites try to and can manipulate public opinion and garner support for their policies. However, it is also a story of failure to successfully use this efficient discursive device to implement a policy change. Why was this the case?
Nearly all expert committees as well as practical experiments spoke in favour of the state model. The motivation for this proposal was linked to the rightfulness and equality frames, and the ideational turn was taking place. However, this was not enough. Social policy institutions are characterized by path dependency and inertia, and therefore they are rigid and resistant to change. This inertia is occasionally punctuated by external shocks, which facilitate policy changes. In the case of Finnish social assistance, the external shocks worked in favour of institutional continuity. As stated above, the discursive pendulum was swinging in favour of rightfulness and equality on two occasions: in the late 1990s when the results of the major experiments became available, and in the mid-2000s when a major government committee was appointed to reform Finnish social policy and to simplify the processes related to basic social benefits. However, external shocks thwarted any possibilities for reform. In the early 1990s, the country experienced the deepest economic recession in its history, and again in the late 2000s, when the reform committee delivered its recommendations, the global economic crisis effectively disarmed the discourse used in support of the state model. In contrast, the idea of cost containment became generally accepted political wisdom. Thus, the economic crises lent more credibility to the frame that emphasized decreasing costs, which better fit in the general ideological climate of the two periods of macro-economic problems (cf. Somers and Block, Reference Somers and Block2005).
However, it is not entirely clear that the implementation of the new model would have taken place even in the absence of these external shocks, since some powerful actors have always opposed the state model. While the shift in discourse in favour of the model probably softened the attitudes among the reluctant political parties, the conservatives in particular, the strongest opposition was found outside the political sphere: the social partners were against the state model. They are not as attuned as politicians to discourses and policy ideas. Although politicians depend on voters, the social partners have their own institutional power that can easily override the will of politicians. Votes are tallied, but power resources rule.
In sum, the policy discourse failed to bring about policy change. First, there were external shocks, which could have opened a window of opportunity for change. In this particular case, however, faced with circumstances of high uncertainty, Finnish policy-makers did not have the political courage to implement a new system and instead reverted to the old model. The external shocks, which, in the institutionalist discussion, have been argued to facilitate changes may, under certain circumstances, inhibit them, as the case study herein illustrates.
Second, the political discourse failed because certain important actors were against the reform. Due to the deterioration of the economy, they were able to successfully utilize the frames regarding increasing costs and social work. Discursive frames, particularly those pertaining to austerity, acquired a hegemonic position as the ‘politics of responsibility’, as contrasted with the ‘politics of distribution’.
This notion of power brings the debate back to ‘old politics’ vs. ‘new politics’, the use of power and power resources in policy-making (Korpi, Reference Korpi1978; Pierson, Reference Pierson1994; Korpi and Palme, Reference Korpi and Palme2003). When emphasizing the role of ideas, the ideational approach tends to explain policy changes primarily in terms of changes in discourses, often neglecting the old issue of political power. Our approach combines the ‘old’ power resource approach with the ‘new’ ideational view. When arguing, as we have done, that discourses give meaning and interpretation to social phenomena, there is no intention to refute the importance of the role of political and other societal actors; the approach herein is dialectical (or interactionist). In other words, the ideas that can ‘become effective forces in history’ (Weber, Reference Weber1989 [1904–1905]: 90) must be implemented by predominant bearers, or as Geertz (Reference Geertz1972: 314) succinctly writes: Ideas must ‘be carried by powerful social groups to have powerful social effects; someone must revere them, celebrate them, defend them, impose them. They have to be institutionalised in order to find not just an intellectual existence in society, but, so to speak, material as well’. In the case of Finnish social assistance, the ideas failed to materialize because there were no predominant or sufficiently powerful elites to revere, defend, and impose them.
Acknowledgement
We want to thank Jenni Blomgren, Heikki Hiilamo, Claus Wendt and three anonymous reviewers for their valuable comments and Leena Rautjärvi for language checking.
