AUSTERITY POLICIES AND POLITICAL INVOLVEMENT. EVIDENCE FROM A PANEL SURVEY IN SPAIN ( ) Jordi Muñoz

Transcription

1 AUSTERITY POLICIES AND POLITICAL INVOLVEMENT. EVIDENCE FROM A PANEL SURVEY IN SPAIN ( ) Jordi Muñoz jordi.munoz@uab.cat Guillem Rico guillem.rico@uab.cat Eva Anduiza eva.anduiza@uab.cat Universitat Autonoma de Barcelona Paper prepared to be presented at the IPSA conference, Madrid 8-12 July 2012 Work in progress - First draft Abstract The purpose of this paper is to see whether austerity measures taken by the Spanish government since 2010 (particularly pension and social benefits cuts, salary reductions for public sector employees), have changed individuals political involvement. We expect that people that have been particularly affected by these measures (those perceiving some kind of public subsidy, those employed in the public sector) will be more likely to increase their levels of political involvement than citizens that have not suffered these measures as directly. It is to be expected that, in the wave of political protests and mobilization produced by the economic crisis, those affected by the fiscal cuts will increase their levels of interest in politics, will increase the frequency of expression of their political dissatisfaction, and will be more likely to participate in politics. We test these expectations using data from an ongoing online panel survey that includes about 1,800 Spanish individuals between 16 and 45 years of age carried out between 2010 and

2 1. Introduction In May 2010 the Spanish government, led by socialist Jose Luis Rodriguez Zapatero, made what The Economist called a spectacular U-turn in its economic policies. While during the first months after the 2008 crisis, public expenditure was seen as the proper recipe to face recession and several stimulus plans were put in place by the government, in 2010 things looked very different. A budget deficit over 10% of GDP, markets souring on the public debt, and EU and Obama s pressure finally changed Zapatero s mind. On the 12th of May 2010, he announced the implementation of a harsh austerity package that was expected to save 15bn euros and cut deficit to 6% in Public employees salaries, public investment plans, pensions and social benefits were drastically reduced. These changes had immediate and important consequences on citizens life, and the kind of consequences that are expected to influence their political attitudes. Protest in the streets spread the following months, with a sharp increase in the number of demonstrations and strikes. According to official data, the Spanish Ministry of Interior (2011) received information about 21,941 demonstrations in 2010, which represents an 18% increase with respect to The wave of protests seems to have gained momentum with the indignados (15M) wave of protests that took place in spring and Summer 2011, as well as with other demonstrations called by trade unions against austerity policies, including a general strike in September 2010 and more recently another one in March Is there a causal connection between the turn towards austerity and the aggregate increase in political participation? One could concieve several research designs aimed at addressing this question. In this paper we focus on the micro foundations of such relationship, and try to assess whether these austerity measures taken by the Spanish government in 2010 affected citizens levels of subjective political involvement. In order to identify the policy feedback effect, we analyze whether those groups of citizens that are especially affected by the austerity policies show a distinct evolution in their levels of political involvement vis à vis the rest of the population. We expect that people that have been particularly affected by these measures (those working for the public sector or perceiving some kind of public subsidy) will be more likely to increase their levels of political involvement than citizens that have not suffered these measures directly. It is to be expected that, in the wave of political protests and mobilization produced by the economic crisis, those affected by the public spending cuts will increase their levels of interest in politics, the frequency of expression of their political dissatisfaction, will 2

3 be more likely to participate in politics and, eventually, will also increase their levels of internal political efficacy. We test these expectations using data from an ongoing online panel survey that includes about 1,800 Spanish individuals between 16 and 45 years of age. The panel so far includes 3 waves, with fieldwork carried out in November 2010, May 2011, and November Our results indicate that public workers and subsidy recipients increased their levels of political involvement vis à vis the rest of the population during the period covered by the panel survey, for most of the indicators used, at least in the short term. 2. Theory While political involvement is usually considered to be affected by contextual variables, the literature on how economic policies affect it is relatively scarce. This is probably a sign of how un-connected the literatures on mass political behavior and public policy have been so far. As a consequence, it is hard to find established arguments on what the consequences of austerity policies may be for political engagement. In an attempt to bridge political behavior and public policy analysis, Mettler and Soss (2004) identify several reasons why policies are expected to affect citizens political engagement, which could be applied to the analysis of the consequences of austerity policies. First, policies provide resources for mobilization and participation. Second, policies provide instances of political learning that shape political beliefs. Third individuals affected by a program may become active on related topics to protect their benefits, as they find material incentives for doing this. Depending on the argument used, the expectations as how austerity should affect engagement are different. First, extending the resource model of political participation (Verba et al 1995), austerity policies basically imply a reduction in the resources available to face the costs of political participation. The consequences of austerity for the individuals affected by the spending cuts would operate in the same direction as economic adversity (versus economic growth): that is, a reduction in political involvement due to a reduction in the resources available. Plenty of evidence confirms the relationship between economic development and political participation, following this resource-based argument (Barnes and Kaase 1979, Ronsenstone 1982, Pacek 1994, Norris 2002, Dalton and van Sickle 2005). Accordingly, we would thus expect austerity 3

4 measures such as salary reductions, pension or other social benefits cuts to reduce political involvement. According to this theoretical perspective, when facing the consequences of austerity policies in their own pockets, citizens will try to focus in solving their own economic problems rather than increase their political involvement and collective action. Some classic works on political involvement have analyzed the policy feedback effects of social programs on citizens and their consequences for public participation in public life following this perspective. Wolfinger and Rosenstone (1980) argue that the resources provided by agricultural programs explain the outstanding high turnout of farmers in the US. Verba, Schlozman and Brady (1995) show that those receiving means-tested-benefits (Medicaid, housing subsidies, food stamps) participate more than those receiving non-mean-tested benefits (Medicare, student loans, veterans benefits). Rosenstone and Hansen (1993) also consider relevant the effect of social security as a resource. Also following resource theory, Mettler (2002) analyzes the effects of a very generous and inclusive program for WWII veterans. She distinguishes, following Pierson (1993), a resource-based effect (recipients get more resources such as education, increasing their capacity to participate) and an interpretative effect (recipients receive information and meanings that make them more predisposed to participate), both favoring increasing civic engagement. This interpretive effect introduces a second reason why austerity should affect political involvement. Some welfare programs can be considered sites for adult political learning as through them citizens get in touch with government institutions, for good and for bad. Soss (1999) finds that, depending on the characteristics of the program, these interpretive effects may not always work in the same direction. In his analysis, being a participant in a welfare program may have positive or negative consequences for participation and external efficacy, but tend to increase internal efficacy anyway. The interaction may provide at the same time the skills to deal with government authorities and a sense that government institutions are (or are not) responsive. The same argument could be extended to public employees (see Parry et al 1992), as working for the public sector implies a closer contact with government and thus may increase internal efficacy, but also affect the perceptions on how (un)responsive the political system is. The consequences to be expected from austerity according to this argument are not entirely clear, as they would be contingent on the specific type of policy and on whether interaction with government is affected by the cuts or these only represent less 4

5 money. But in general we can assume that the message conveyed by a reduction in welfare programs and salaries does not favor political engagement. However, a third argument -on the opposite direction- can be put forward. As Campbell et al. already acknowledged, economic interest has long been seen a primary motive impelling political action (Campbell et al. 1960, 381). Economic grievances, dissatisfaction and deprivation are expected to stimulate people to protest (Gurr 1968, 1970). Austerity policies that imply welfare reductions and public employees salary cuts put the burden of the costs of the economic crisis on specific sectors of the population. Citizens may then act to try to change this course of action by increasing their levels of political engagement. We could expect that austerity policies would have an effect on the incumbent vote (but see Alesina et al 2012), although there is evidence as well linking fiscal adjustments to protest participation at the aggregate level. Ponticelli and Voth (2011), using a long dataset covering Europe between 1919 and 2009, show how budget cuts increase frequency of protest events (demonstrations, strikes, riots and political assassinations), though their paper does not develop the theoretical arguments behind this relationship. The micro foundations of their findings should be visible in the case we analyze here. Austerity policies are particularly expected to produce mobilization by grievances, following a logic that is different than the introduction of a welfare benefit. While the introduction of a benefit may increase resources and feelings of internal efficacy, the reduction or withdrawal of these benefits may not have the symmetric opposite effect. The acquired learning is not erased with the withdrawal, predispositions towards participation may have changed and remain at least for a while. Though losers of social benefit programs or those affected by a salary reductions loose resources, in the short term other factors may offset the effect of this loss of resources over political participation (Flavin, Griffin 2009). Cuts are expected to produce strong emotional reactions in terms of increased anxiety and outrage, and these motivational impulses (Lodge, Taber, and Weber 2006) have been found to affect participation and engagement (Kenney 2010, Tam Cho, Gimpel, and Wu 2006, Valentino et al. 2011, Brader 2005, 2006). To sum up, austerity policies may operate in different ways regarding political involvement. They could affect the level of resources available and hence reduce levels of political involvement. They can also affect political beliefs, self-perceptions on the competence to 5

6 interact with government, and expectations about political system responsiveness, with different and sometimes contradictory consequences for political involvement. Finally they can produce grievances and emotional reactions that increase levels of political involvement and mobilization, especially when they are directed to specific social groups. 3. Empirical strategy Identifying the policy effects on political attitudes and behavior is not an easy task. Multiple confounding factors might lead to biased estimations. In order to try to overcome this difficulty, we use a difference-in-differences type of design, in which we measure the outcome at different points in time for treated and untreated individuals. In this case the treatment condition the policy- is not easy to specify, since austerity packages have deep and wide consequences on most social and economic sectors. However, these effects are not homogeneous, and some groups in the population are especially affected by the spending cuts. We use this heterogeneity in order to proxy treatment by using indicators of being especially affected or threatened by the spending cuts: public sector workers (whose salary was reduced at least by 5%) and recipients of a public subsidy (that were frozen, threatened and, in some cases, access conditions tightened). In the case of austerity policies that occur in the midst of a deep economic crisis, it is especially challenging to separate the effect of worsening economic conditions from the pure policy effect, since both austerity and recession not only co-occur but are causally related presumably in a reciprocal pattern of relationship. This problem is not easy to overcome. However, in order to address it, we replicated our analyses with two independent variables that identify those especially affected by the economic crisis: respondents with low income (under 900 a month) and those unemployed. If our argument is correct, and the different paths followed by subsidy recipients and public employees are related to the introduction of austerity policies and not just to the economic turmoil, low-income and unemployed respondents should not display a distinct pattern of involvement. On the contrary, were our results driven by a crisis-effect instead of a policy-feedback effect linked to austerity, we should observe the same, or even starker pattern of increasing involvement among these two groups. Our empirical analysis, however, suffers from two serious limitations: on one hand, the indicator that identifies those receiving a public subsidy does not allow for a disaggregation of 6

7 different groups of recipients (students, unemployed, disabled or ill people, widows, etc.). This poses a problem since not all programs were equally affected by the spending cuts and therefore we might be loosing precision. In the case of public sector workers the variable is a better proxy, since the link between working for the public sector and having a salary reduction is clear. Second, and more importantly, our panel started a few months after the first budget cuts were announced. This might seriously affect our ability of identifying the policy effects since at our starting point they might already have taken place. However, as austerity has been, since 2010, an ongoing policy in Spain that is still unfolding, we expect its effects to be visible even beyond the first shock in May In any case this would make us underestimate the real effect of a policy change. 4. Data and variables We use an ongoing online panel survey that interviews Spanish citizens, starting in November Respondents were selected from an on-line survey pool set up through active recruitment of potential subjects in the main commercial on-line services and websites in Spain. 1 Since Internet use is sharply decreasing with age, the survey was directed only to respondents between 16 and 45 years of age. Quotas were applied to the sampling process in order to reduce biases related to the non-probability nature of our sample. However, there are still important deviations from a representative sample (see appendix). The dependent variable is political involvement. In order to provide a complete picture, we use four distinct indicators, two behavioral and two attitudinal: political interest, internal efficacy, political participation and expression of political ideas online. Political interest is measured using the common 4-point scale, ranging from not interested at all to very interested. To measure internal efficacy, we use an indicator of agreement with the statement It is difficult to understand what happens in politics. For the sake of interpretability, we reverse the scale so higher values correspond to higher levels of efficacy. Political participation is a cumulative index of political activities that includes signing petitions, contacting politicians, buying or boycotting products for political, ethical or environmental reasons, donating money to political causes, taking part in demonstrations and strikes. The on-line political activity index is composed by the following items: signing online petitions, making online donations, 1 Further details can be found at 7

8 expressing opinions in social networks, blogs and websites, and sending s with political content. The main independent variables are receiving a public subsidy (pension, grant, housing subsidies, etc.) and working for the public sector. Due to the age exclusion in our data, retired persons are excluded from our analysis, and therefore the scope of our conclusions will be limited: we are not referring to the public employees/subsidy recipients in general, but just to those below 45 years of age. In order to separate the policy effect from the effect of the economic crisis, we introduce a dummy variable that identifies unemployed respondents, as well as a variable that measures their income. As additional controls, we include ideology (on a 0-10 left-right scale), education and age in our models. Tables 1 and 2 report the mean values of our main variables across the first three waves of the panel (the fieldwork of the fourth wave will take place in May 2012). About 60% of those perceiving a subsidy in our sample are women, while non-perceivers are more balanced in terms of gender. Those who get a subsidy are slightly older than those who do not. They also hold lower levels of education. With regard to left-right ideological self-placement, they became more conservative after wave 1, so that in waves 2 and 3 they stand significantly more to the right than the group of non-perceivers. Differences between recipients and the remainder are not consistent across all four indicators of political involvement. People that perceive a subsidy hold a slight advantage in internal political efficacy but display similar levels of interest in politics. Higher levels of participation are visible in the first two waves but vanish in wave 3, as the general election of November 2011 gets closer. Recipients also lead in online political activity, but only in wave 2 the difference achieves statistical significance. Table 1: Variable means by subsidy reception Gender (0 male -1 female) Subsidy 0.61*** 0.58*** 0.61*** No subsidy Age (16-45 years) Subsidy 31.27** No subsidy Education (1 lowest-11 highest) Subsidy 7.53*** 7.66** 7.54*** No subsidy Left-right (0 left-10 right) 8

9 Subsidy * 4.57* No subsidy Internal efficacy (0 low-10 high) Subsidy * No subsidy Political interest (1 none-4 much) Subsidy No subsidy Participation (0 low-6 high) Subsidy 1.99** 2.11*** 1.73 No subsidy Online political activism (0 low-6 high) Subsidy *** 1.54 No subsidy * p<0.1, ** p<0.05, *** p<0.01 (between-group differences) As shown in Table 2, the proportion of women is lower among respondents employed in the public sector than is for the rest of the sample. Public workers also are slightly older and more educated. With regard to our dependent variables, public workers show higher levels of interest in politics and both offline and online political participation, but they fall behind in terms of internal efficacy. Table 2: Variable means by public sector employment Gender (0 male -1 female) Public workers 0.43*** 0.44** 0.42*** Other Age (16-45 years) Public workers 32.84*** 33.49*** 33.71*** Other Education (1 lowest-11 highest) Public workers 8.39*** 8.41*** 8.52*** Other Left-right (0 left-10 right) Public workers Other Internal efficacy (0 low-10 high) Public workers 4.04*** 3.78*** 3.94 Other Political interest (1 none-4 much) Public workers 2.59*** 2.61*** 2.57*** Other Participation (0 low-6 high) Public workers 2.06*** 2.07*** 2.01*** Other Online political activism (0 low-6 high) Public workers 1.66*** 1.70*** 1.56 Other * p<0.1, ** p<0.05, *** p<0.01 (between-group differences) 9

10 The above comparisons reveal significant differences across the groups defined by our treatment measures. Yet the social composition of subsidy recipients and public sector workers differs markedly, as does their pattern of political involvement, which suggests that these positions have disparate influences on the various dimensions of political engagement. One thing they definitely have in common is the fact that both groups have been affected, albeit to varying degrees, by the austerity policies undertaken since Thus, our design poses the withdrawal hypothesis to a stringent test. We expect for voters directly affected by these measures to respond in similar directions despite their distinct predispositions and underlying attitudes and levels of political participation. To the extent that austerity cuts display similar effects on both groups, we can be confident that it was indeed the cuts that made a difference, and not factors related to the different composition of these groups. 5. Analysis and discussion In order to estimate whether the levels of political engagement of public employees and subsidy recipients followed a distinct pattern than the rest of the respondents during the 12 months of our panel, we estimate a set of fixed-effects panel models (one for each of the four dependent variables discussed above) that include interaction terms between the dummy variable that identifies these groups and the 2 nd and 3 rd waves of the panel (wave 1 is the reference category). The models incorporate the controls we already discussed: education, age, income and unemployment. Table 3: Subsidy recipients, fixed effects models (1) (2) (3) (4) Internal efficacy Participation Political interest Online political activism b/se b/se b/se b/se Wave *** -0.09** -0.04** -0.10** (0.07) (0.04) (0.02) (0.04) Wave *** -0.12*** -0.03* -0.08* (0.07) (0.04) (0.02) (0.04) Subsidy -0.31* * 0.01 (0.16) (0.10) (0.04) (0.09) Wave 2*Subsidy ** *** (0.18) (0.11) (0.05) (0.10) Wave 3*Subsidy 0.52*** (0.20) (0.12) (0.05) (0.12) Education (0.05) (0.03) (0.01) (0.03) Ideology (LR) -0.11*** ** (0.03) (0.02) (0.01) (0.02) Wage * (0.03) (0.02) (0.01) (0.01) Unemployed (0.12) (0.08) (0.03) (0.07) Constant 5.05*** 1.93*** 2.44*** 1.73*** (0.44) (0.27) (0.12) (0.25) 10

11 N N_g r2_w r2_b rho * p<0.1, ** p<0.05, *** p<0.01 Table 3 displays results for the models regarding public subsidy recipients. The first thing to note is that during the period covered by our panel there has been a general trend towards a decrease in political engagement in every single indicator used here. This might be an effect of the worsening economic conditions or other contextual circumstances, but is especially remarkable taking into account that the third wave of the panel took place a few weeks before the general election, a context that a priori might be expected increase the levels of political engagement. This is however consistent with the saturation effect that has been found in some analyses of the consequences of electoral campaigns in Spain (Anduiza 2005). Beyond this general time-trend, we observe no main effect for recipients of public subsidies in time 1: they were not any different, in terms of engagement, than the rest of the population in November 2010, except for political efficacy: they were less efficacious at the beginning of our panel. However, the interaction terms show how their evolution was not parallel to the rest of the population, at least for some indicators. Political participation and online political activism increased in wave 2 for subsidy recipients, while they decreased for the rest of the population. Political efficacy, on the other hand, increased for recipients only in the third wave. This different sequence might be suggesting that participation had a lagged impact on efficacy, but checking this hypothesis would require further analysis beyond the scope of this paper. Finally, political interest does not show a distinct pattern of evolution for recipients: the slight decrease is common to both subsidy recipients and the rest of our sample. Figure 1 graphically represents the interactions shown in table 3. It shows how the evolution in political participation and on-line activism is different in wave 2 for subsidy recipients (red line), while it tends to converge with the rest of the population in wave 3 (blue line). Regarding efficacy, it shows how in wave 3 it increased slightly for recipients while continued its descending path for non-recipients. 11

12 Figure 1: Subsidy recipients and political involvement. Linear predictions and 95% CI: Subsidy recipients and political involvement. Linear predictions and 95% CI Internal efficacy Participation Political interest On-line activism Table 4 presents the results for the public sector employees models. The specification is the same as above. We expect, as well, that they increase their levels of engagement more (or decrease less) than the rest of the population. The story expressed by the models is roughly similar to the case of subsidy recipients, albeit with some differences. The overall demobilizing path shows up in this models as well, while the effect of working for the public sector at time 1 is not statistically significant except for internal efficacy: public workers had higher levels of internal political efficacy. In this case, the interaction term of public workers and wave 2 has a statistically significant but negative coefficient, indicating that the levels of internal efficacy dropped more sharply for public sector employees between wave 1 and 2 than for the rest of the sample. This runs counter our expectations, and might be related to the fact that in wave 1, five months after the salary cuts were announced, public employees had a higher level of efficacy, so the policy effect would be already present in our first measurement. The fact that we do not have a true pre-treatment measure limits our ability to draw firm conclusions. 12

13 Table 4: Fixed effects models. Public sector workers (1) (2) (3) (4) Internal efficacy Participation Political interest Online political activism b/se b/se b/se b/se Wave *** -0.09** -0.06*** -0.09** (0.07) (0.04) (0.02) (0.04) Wave *** -0.13*** ** (0.08) (0.05) (0.02) (0.04) Public 0.32** (0.14) (0.08) (0.04) (0.08) Wave 2*Public -0.25* ** 0.15* (0.14) (0.09) (0.04) (0.08) Wave 3* Public (0.15) (0.09) (0.04) (0.09) Education (0.05) (0.03) (0.01) (0.03) Ideology (LR) -0.10*** ** (0.03) (0.02) (0.01) (0.02) Wage * (0.03) (0.02) (0.01) (0.01) Unemployed (0.12) (0.07) (0.03) (0.07) Constant 5.02*** 1.89*** 2.43*** 1.70*** (0.44) (0.27) (0.12) (0.25) N N_g r2_w r2_b rho * p<0.1, ** p<0.05, *** p<0.01 Participation does not show a significantly different pattern of evolution for public employees, while political interest shows a slight increase for public workers in Wave 2 while it declines for the rest of the sample. On-line activism, as in the case of subsidy recipients, increased in wave 2 and declined less in wave 3 for public workers than the rest, showing also a distinctive pattern. Figure 2 represents, again, the different evolutions for both groups. 13

14 Figure 2: Public employees and political involvement. Linear predictions and 95% CI Public employees and political involvement. Linear predictions and 95% CI Internal efficacy Participation Political interest On-line activism In order to consider the alternative explanation that relates our results to the crisis itself and not to austerity policies, we replicated our analyses for two groups especially harmed by the economic turmoil. Following the same logic we have discussed in the previous analyses, if we were capturing just a crisis effect and not a specifically policy-related effect on involvement, we should observe the same, or even more stark patterns of relative increase of political involvement among those especially affected by the recession. In this case, we proxy overexposure to the crisis through two dummy variables, indicators of vulnerability: low income (under 900 a month) and unemployment. Tables 5 and 6 display the same models presented above for subsidy recipients and public workers with these two new variables instead of those that intended to capture the policy effects. Therefore, again, we fitted fixed effects models with an interaction between each wave and our independent variable of interest. 14

15 Table 5: Fixed effects models. Low income (1) (2) (3) (4) Internal efficacy Participation Political interest Online political activism b/se b/se b/se b/se Wave *** *** (0.08) (0.05) (0.02) (0.04) Wave *** -0.13*** -0.05** (0.08) (0.05) (0.02) (0.05) Low income ** (0.12) (0.07) (0.03) (0.07) Wave 2*Low income -0.28** (0.13) (0.08) (0.03) (0.07) Wave 3*Low income -0.22* (0.13) (0.08) (0.04) (0.08) Education (0.05) (0.03) (0.01) (0.03) Ideology (LR) -0.11*** ** (0.03) (0.02) (0.01) (0.02) Constant 4.89*** 2.07*** 2.43*** 1.59*** (0.43) (0.26) (0.11) (0.25) N N_g r2_w r2_b r2_o rho * p<0.1, ** p<0.05, *** p<0.01 Table 6: Fixed effects models. Unemployment (1) (2) (3) (4) Internal efficacy Participation Political interest Online political activism b/se b/se b/se b/se Wave *** * (0.07) (0.04) (0.02) (0.04) Wave *** -0.10** (0.07) (0.04) (0.02) (0.04) Unemployed (0.16) (0.09) (0.04) (0.09) Wave 2* Unemployed (0.17) (0.10) (0.04) (0.10) Wave 3* Unemployed (0.18) (0.11) (0.05) (0.10) Education (0.05) (0.03) (0.01) (0.03) Ideology (LR) -0.11*** ** (0.03) (0.02) (0.01) (0.02) Constant 4.94*** 2.05*** 2.40*** 1.66*** (0.42) (0.26) (0.11) (0.24) N N_g r2_w r2_b r2_o rho * p<0.1, ** p<0.05, *** p<

16 The models show that the distinctive increases in levels of political involvement we observed for subsidy recipients and public workers are not present in the case of the groups of the more vulnerable groups. Low-income and unemployed respondents followed mostly the same evolution in terms of political involvement as the rest of the population, as expressed by the non-statistically significant coefficients for the interaction terms. There is just one exception to this rule: low-income respondents became significantly less efficacious during the year covered by our panel survey. This shows how, in this case, worsening economic conditions had an alienating effect on the most vulnerable sector of the population, which is consistent with previous research, as discusses above. 6. Conclusions Our data indicate a general trend towards a declining level of political engagement in both the attitudinal and behavioral indicators during the 12 months covered by our panel survey (November 2010-November 2011), that were months of worsening economic conditions and the introduction of increasingly harsh austerity policies. However, these trends are not homogeneous, and specific population groups show distinct patterns of evolution. Two groups that were disproportionally affected by the austerity packages introduced by the Spanish government in May 2010 followed a distinct pattern of evolution than the rest of the population: public sector employees (that lost at least 5% of their salary) and subsidy recipients. Their levels of political engagement increased, or declined less than those of the rest of the population during the analyzed period. This might be indicating that, even if the general context, deeply conditioned by the economic turmoil, is leading to growing political apathy, those groups with specific grievances or that might have felt specially threatened by the government s political decisions, reacted to the introduction of the austerity packages and eventually experienced an increase in their levels of political participation, efficacy or interest. Certainly, our results are not as clear-cut and consistent as we expected: the patterns are not homogeneous across all the indicators of political involvement, and the magnitude of the effects is very small, although this is indeed coherent with the well-known stability of attitudes such as political interest (Prior 2009). We also see that some of the diverging patterns do not appear to consolidate over time and in some cases the groups converged with the rest of the population after two waves. 16

17 How should we read this weakness of our results? Our understanding is that they are related to problems with the data rather than with our theoretical expectations. The fact that the panel started five months after the first cuts were announced, as well as the limitations of the sample in terms of its representativeness of the population might be severely limiting our ability to identify at the individual level the increased political mobilization of specific groups that can be observed at the aggregate level. But, nonetheless, we have identified a mobilizing effect of spending cuts on the groups that were specifically targeted by austerity, an effect that counteracted an overall demobilizing effect related most probably to a deepening crisis. While entitlement of social rights and benefits has been shown to empower citizens through an increase in their resources, the withdrawal of rights and benefits does not seem to have, at least in the short term, the opposite effect. On the contrary: fiscal adjustment and spending cuts appear to have increased (in relative terms) the political efficacy and involvement of those that were more directly targeted by the cuts. Despite the empirical difficulties, related to the nature of the data used and the problems in identifying the groups more affected by the austerity policies, in this paper we have made an effort to advance towards a more clear specification and estimation of the micro foundations of the aggregate relationship between fiscal adjustments and political participation. In order to estimate policy effects on political attitudes and behavior, a true pre-post design with a clear identification of the groups affected by the policy is desirable. Our analysis provides insights on the micro-foundations of the macro findings that relate fiscal adjustments to an increase in protest among the population. We have seen how, in a context of worsening economic conditions that tends to alienate citizens from politics, those groups that were specificall targeted by the spending cuts showed distinct patterns in terms of engagement. This has implications on the expected consequences of the ongoing adjustment programs, that might lead to further episodes of protest, at least in the short term: the lack of stability of the effects after twelve months is also a relevant finding. Longer panels and better data might provide insights on whether the empowering effect of spending cuts is juts temporary and then an alienating effect dominates or, on the contrary, grievances and emotional reactions lead to a sustained increase in political involvement of the targeted groups. 17

18 References Alesina, Alberto; Dorian Carloni and Giampaolo Lecce (2012) The Electoral Consequences of Large Fiscal Adjustments in: Alberto Alesina and Francesco Giavazzi, editors Fiscal Policy after the Financial Crisis, University of Chicago Press Anduiza Perea, E. (2005). Campaign Effects in the Spanish Election of Journal of Elections, Public Opinion & Parties, 15(2), doi: / Barnes, S. H., & Kaase, M. (1979). Political action: mass participation in five Western democracies. Sage Publications. Campbell, A., & Center, U. of M. S. R. (1980). The American voter. University of Chicago Press. Dalton, R. J., & van Sickle, A. (2005). The Resource, Structural, and Cultural Bases of Protest. CSD Working Papers. Recuperado a partir de Flavin, P., & Griffin, J. D. (2009). Policy, Preferences, and Participation: Government s Impact on Democratic Citizenship. The Journal of Politics, 71(02), Kinder, D. R., & Kiewiet, D. R. (1979). Economic Discontent and Political Behavior: The Role of Personal Grievances and Collective Economic Judgments in Congressional Voting. American Journal of Political Science, 23(3), doi: / Mettler, S., & Soss, J. (2004). The Consequences of Public Policy for Democratic Citizenship: Bridging Policy Studies and Mass Politics. Perspectives on Politics, 2(01), Ministerio del Interior, 2011, Anuario, Madrid: Ministerio del Interior Norris, P. (2002). Democratic Phoenix: Reinventing Political Activism. Cambridge: Cambridge University Press. Pacek, A. C. (1994). Macroeconomic Conditions and Electoral Politics in East Central Europe. American Journal of Political Science, 38(3), doi: / Parry, G., Moyser, G., & Day, N. (1992). Political Participation and Democracy in Britain. Cambridge University Press. Ponticelli, J. & Voth, H.J. (2011) Austerity and Anarchy: Budget Cuts and Social Unrest in Europe, Discussion Paper No Rosenstone, S. J. (1982). Economic Adversity and Voter Turnout. American Journal of Political Science, 26(1), Soss J Lessons of Welfare: Policy Design, Political Learning, and Political Action, The American Political Science Review, Vol. 93, No. 2 (Jun., 1999), pp Verba, S., Schlozman, K. L., & Brady, H. E. (1995). Voice and equality. Harvard: Harvard University Press. Wolfinger, R. E., & Rosenstone, S. J. (1980). Who votes? Yale University Press. 18

19 Appendix. Characteristics of our panel Percentages Differences 1st wave of panel CIS INE Panel-INE cis-ine panel-cis Men 50,5 51,3 51,3-0,7 0,0-0,8 Women 49,5 48,7 48,7 0,7 0,0 0, years of age 26,9 18,6 19,3 7,6 *** -0,7 8,3 *** ,1 18,6 17,8-0,7 0,8-1, ,0 21,4 21,0-0,1 0,4-0, ,6 22,4 21,1 1,5 1,3 0, ,4 18,9 20,7-8,3 *** -1,8-6,5 *** Primary education or less 0,2 6,2 9,4-9,1 *** -3,1 *** -6,0 *** Lower secondary 8,6 29,6 25,1-16,5 *** 4,5 *** -21,0 *** Upper secondary 26,3 24,2 26,0 0,3-1,8 2,1 Vocational 17,6 15,7 14,6 3,1 *** 1,2 1,9 University 47,3 24,3 25,1 22,3 *** -0,8 23,1 *** Up to inhabitats 49,0 46,6 49,6-0,6-3,0 2, ,8 36,9 35,2-1,4 1,7-3,1 More than inhabitants 17,2 16,5 15,2 2,0 * 1,3 0,7 Proportion of people reciving a public subsidee * p < 0,05; ** p < 0,01; *** p < 0,001 19