Poor versus rich countries: a gap in public attitudes towards fiscal solidarity in the EU

Abstract Existing research has primarily focused on the role of utility and identity in shaping individuals’ European Union (EU) preferences. This article argues that macroeconomic context is a crucial predictor of attitudes towards transnational financial assistance, which has been omitted from previous analyses. Using data from the 2014 European Election Studies (EES) Voter Study for 28 EU member states, this article demonstrates that citizens living in poorer EU countries are less willing to support fiscal solidarity than their counterparts in more affluent countries. Country affluence serves as a heuristic, moderating the relationship between individual-level utility and identity considerations and willingness to show solidarity to member states with economic difficulties. When a country does not fare well economically, citizens’ views on providing help to others remain negative, irrespective of individual-level utilitarian and identity considerations. Our findings have implications for understanding the decision-making calculus underlying preference formation.

prepared to engage in a collective response to EU member states' economic problems. Existing literature has addressed this question primarily through a focus on individuals' social dispositions, including identity and cosmopolitanism. The key finding is that culturally open individuals are more likely to also be supportive of international bailouts (Bechtel et al. 2014;Kuhn et al. 2017;Verhaegen 2018). Other perspectives relate to the effect of party cues, elite dissent, left-right economic ideology (Kleider and Stoeckel 2019;Stoeckel and Kuhn 2018) and the role of regional factors, such as corruption and quality of governance (Bauhr and Charron 2018). None of the above contributions focus on the role of context in shaping support for transnational financial assistance. This is surprising given that the costs and benefits of within-EU financial assistance differ from one member state to another. We contend that a country's macroeconomic situation serves as a heuristic shaping the extent to which citizens are willing to share resources. We advance the literature byfor the first timeexamining the mechanisms underlying the direct relationship between macroeconomic context and preferences for offering financial assistance to EU member states in need. We further theorise the moderating effect of context on the relationship between individual-level utility and identity considerations on the one hand, and support for financial assistance on the other.
We posit that citizen preferences for transnational financial assistance relate to an evaluation of how their state may be affected by offering such help. Citizens living in poorer member states do not view within-EU solidary measures as a collective responsibility, but rather as primarily the obligation of richer member states. They are less likely to support offering transnational financial assistance to member states in need, although their countries may potentially become recipients of such funds. If they contribute towards assisting another EU member state, they will have to share scarce domestic financial resources. If, on the other hand, they receive external financial assistance, they will have to abide by EU conditionality, which is associated with austerity and could potentially place them at a competitive disadvantage. Crucially, the domestic context moderates the impact of key individual-level explanations, including utility and identity, on citizen support for transnational financial assistance. In poorer EU member states, citizens oppose financial assistance irrespective of utility and identity factors. Put differently, the relationship between individual-level utility and identity considerations, on the one hand, and preferences for transnational financial assistance, on the other, is only a rich-country phenomenon.
We test our hypotheses using data from the 2014 European Election Studies (EES) Voter Study for 28 EU member states. Incorporating context yields substantial theoretical and empirical insights. First, by showing that public opinion reflects country-level divides, often irrespective of individual-level utility and identity considerations, our work casts serious doubt on classical rational and cultural theories of attitude formation and political choice. We show that economic prosperity matters directly in explaining support for offering financial assistance to EU countries in need. This runs counter to literature that finds no significant effects of macroeconomic contextual variables on citizen attitudes towards such assistance (e.g. Bauhr and Charron 2018;Daniele and Geys 2015). The article demonstrates the limitations of identity and utilitarian theories: in poorer EU member states support for cross-border transfer of resources is evenly low irrespective of personal economic and affective considerations. This finding calls for a significant revision of the two core approaches of preference formation. Second, we empirically show that the structure of citizens' attitudes towards offering financial assistance to EU member states in economic difficulties does not fully reflect that of general preferences for European integration. The same individuals in different countries might support European integration but may not be prepared to support potentially costly policies. We thus establish the importance of examining Euroscepticism through its different dimensions, and the existence of conflicting policy preferences among European citizens. Our findings have important implications for policy debates, pointing to the difficulty of addressing Euroscepticism with a one-size-fits-all strategy.
Modelling support for fiscal solidarity: the role of domestic economic prosperity Whereas bailouts related to the euro crisis might have been prominent in media and political discussions, the EU has provided financial assistance to both Eurozone and non-Eurozone member states. Within the context of the European Monetary Union (EMU), member states have established intergovernmental stabilisation mechanisms, such as the European Financial Stabilisation Mechanism (EFSM) and later the European Stability Mechanism (ESM) to preserve financial stability across the euro area. Unlike contributions to the EU budget, this financial assistance does not relate to net fiscal transfers. It takes the form of a loan linked to macroeconomic conditionality 'to ensure that Member States receiving such assistance implement the necessary fiscal, economic, structural and supervisory reforms'. 1 In the context of the ESM, these loans constitute an implicit form of risk sharing, as the ESM has an AAA þ rating. Countries can borrow at lower interest rates, butto the extent that loans are repaiddebtor countries can also make profits (De Grauwe and Ji 2013). Although member states contribute to the EU's programmes, acting as guarantors, the ESM raises the funds that it lends through the financial markets, which gives the EU the role of a 'borrower'. The EU has also set up programmes of assistance to non-EMU member states, offering Balance of Payments assistance in the form of loans. To date, Cyprus, Greece, Ireland, Portugal, Spain, Romania, Hungary and Latvia have participated in various EU-managed financial programmes, often in collaboration with the International Monetary Fund (IMF). These programmes and bailout packages should not be confused with permanent EU transfers and redistributive polices (see e.g. Mattila 2006).
Our point of departure is that country context is key to understanding preferences for transnational financial assistance. Donors and beneficiaries are first and foremost governments and countries, rather than individuals or groups of individuals. Countries that receive external financial assistance do not allocate that funding to the benefit of a specific social class. Cross-border financial assistance in the EU aims to address imbalances at the country level, rather than income inequality at the individual or group levels. During the crisis, the key question related to whether those EU member states with a current account surplus would bail out those facing severe economic and financial difficulties. Similarly, in the context of foreign aid, typically prosperous countries tend to give higher amounts of financial assistance measured as a percentage of their GDP compared to struggling economies (Chong and Gradstein 2008).
This suggests that an individual's probability to oppose or support fiscal aid requires taking into account geographic variation. We posit that citizens employ the national macroeconomic context, and specifically wealth, in order to assess the costs and benefits of such financial assistance. Our approach speaks directly to research that focuses on the domestic foundations of international politics and shows that the national context has a 'cueing' or 'benchmarking' effect on how Europeans evaluate the EU, its institutions and policies (De Vries 2017;Rohrschneider 2002;Rohrschneider and Loveless 2010;S anchez-Cuenca 2000). Specifically, this literature suggests that domestic political institutions and performance are important antecedents of EU preferences. The EU tends to be too complicated and too distant for the majority of citizens, who form their preferences using information shortcuts provided by their domestic context (Anderson 1998). For example, S anchez-Cuenca (2000 finds that the EU project is more attractive to individuals who live in countries with higher levels of corruption. European integration becomes a solution to domestic problems that cannot be resolved within the nation state. In addition, citizens in countries with high-quality domestic institutions tend to be more Eurosceptic and more dissatisfied with EU democracy (Hobolt 2012;Rohrschneider 2002). National conditions matter because they serve as benchmarks against which the EU is judged (De Vries 2017;Hobolt 2012). Examining specifically the effect of national economic conditions on EU support, the literature has pointed towards two directions. On the one hand, assuming that European integration is associated with prosperity, as a country's standard of living increases, its citizens are expected to become more pro-EU (Eichenberg and Dalton 1993;see Gabel and Whitten 1997 for mixed results). This relationship was observed during the pre-Maastricht period (Eichenberg and Dalton 2007). On the other hand, given that poorer member states are the main beneficiaries of EU funds, their citizens tend to be more pro-EU compared to those living in net contributor countries (Anderson and Reichert 1995;Hooghe and Marks 2006;although Hobolt 2012 finds no effect of growth and net EU transfers on satisfaction with EU democracy). In sum, the intuition is that when national political and economic institutions work poorly, people tend to be more supportive of the EU, and this relationship is especially strong in the post-Maastricht period.
We also posit that individuals employ cues from their domestic economic context in order to evaluate whether they support their country's contribution to EU-wide fiscal solidarity. Citizens are sensitive to their collective economic circumstances. However, our intuition regarding the effect on preferences for within-EU aid is the reverse. We argue that citizens living in less affluent EU member states are likely to oppose transnational financial assistance initiatives for two reasons. First, poorer countries tend to have more severe budgetary constraints. As such, their citizens may perceive national resources as scarce and prefer to spend them domestically (Lengfeld et al. 2015: 9-10). This is in line with work that expects that, in Germany, support for international bailouts 'should be lower in net receiving regions, since residents might worry that international financial transfers would come at the expense of transfers to their own region' (Bechtel et al. 2014: 838; see also Gomez 2015;Kleider and Stoeckel 2019;Kuhn et al. 2017). Note that Slovakia, an EU peripheral and relatively unprosperous country, opposed expanding the European Financial Stability Facility (EFSF) mechanism under the justification that it was an expensive measure. 2 Second, we know that domestic conditions affect the salience and politicisation of issues in a given country (Rohrschneider and Loveless 2010). EU programmes of financial assistance are often tied to tough conditions and external intervention in a member state's domestic finances. Citizens in poorer countrieswhich are potential recipients of fiscal aidare likely to more prominently consider the austerity-related effects of such assistance. Austerity is often linked to slower economic growth, downward pressures on domestic welfare spending, unemployment, and a general deterioration of citizens' wellbeing. This is in line with Nicoli (2019) who finds that citizens living in countries in financial distress are more likely to oppose measures towards European economic governance, as these entail further austerity measures and fiscal supervision. We also know that the EU's legitimacy has been undermined in crisis-ridden countries where the decline of trust in the EU has been particularly severe (Armingeon and Ceka 2014;Foster and Frieden 2017).
If a country's inability to absorb the shock of an economic crisis is considered to be partly related to structural imbalances inherent in the EU system, then the citizens of poorer countries might see that it is the obligation of wealthier member states to address such imbalances by bearing the costs of bailing out countries in severe economic difficulties. Therefore, citizens in poorer member states view EU fiscal aid with scepticism. If they pay towards assisting another country, they will have to share precious and finite domestic financial resources; if they end up receiving assistance, they will have to abide by EU conditionality, which could potentially place them at a substantial competitive disadvantage.
H1: Citizens in poorer EU member states are less likely to support offering financial help to EU member states in severe economic and financial difficulties.

The conditioning effect of a member state's economic prosperity
The direct effect of a member state's economic prosperity is only part of the story. Citizens may experience the same economic conditions but may perceive them differently depending on their individual dispositions. We argue that country affluence may also moderate the relationship between individual-level considerations and willingness to show solidarity to member states with economic difficulties. The literature puts forward at least two key theories at the individual level that may explain such preferences, i.e. utility and identity (Bechtel et al. 2014;Hobolt and De Vries 2016). We advance the literature by showing that the relationship of two key models of preference formation to attitudes towards transnational financial assistance is nuanced.
Subjective economic evaluations are likely to influence an individual's attitude towards the EU (e.g. Anderson and Reichert 1995;Braun and Tausendpfund 2014;Gabel and Palmer 1995;Gabel and Whitten 1997;Hooghe and Marks 2006;Serrichio et al. 2013). Here we may distinguish between sociotropic and egocentric perspectives. Citizens who do not feel confident about their country's economy are less likely to support integration (e.g. Anderson 1998;Gabel and Whitten 1997;Hooghe and Marks 2006). In addition, the human capital hypothesis suggests that citizens in different socioeconomic situations may experience integration differently, with those with low levels of human capital (low-skilled; low-educated) less likely to support the EU (e.g. Gabel 1998a; Gabel and Palmer 1995;Hooghe and Marks 2006;Serrichio et al. 2013). Applying these perspectives to the context of attitudes towards transnational financial assistance, we similarly expect that citizens who negatively evaluate their national economy and those with low levels of human capital will oppose the policy. This is simply because such individuals are likely to perceive a much more intense resource conflict between domestic and international redistribution. 3 These mechanisms, however, are expected to work differently across member states based on country affluence. The material costs to the national economy might weaken the impact of individual-level considerations. In poorer member states, individuals who negatively evaluate their country's economy and those with lower levels of human capital are not the only ones unlikely to support offering financial help to EU member states in severe economic and financial difficulties. Given that resources are particularly scarce in poorer EU member states, we suggest that people may oppose offering financial assistance to another country irrespective of their evaluations of the economy and their socioeconomic placement. In such countries, the majority of citizens may be concerned about the economic costs of bailouts on the country as a wholeeven those with a positive outlook of the national economy. In addition, austerity and cuts do not only negatively affect individuals with low levels of human capital. Downward pressures on the economy can also affect those educated individuals in managerial positions who may see a relative decline in their incomes and purchasing power.
H2: Individual-level utility effects on support for fiscal solidarity are weaker in poor EU member states.
Going beyond the utilitarian argument, research has shown that affective factors have a strong effect in understanding people's preferences (see Bechtel et al. 2014;Kuhn and Stoeckel 2014;Lengfeld et al. 2015). Factors related to cultural openness, such as cosmopolitanism (Bechtel et al. 2014), identity (Kuhn and Stoeckel 2014) and moral reasoning (Lengfeld et al. 2015) may affect willingness to support transnational assistance. General feelings of cosmopolitanism are dispositions that may positively affect attitudes towards economic openness in terms of both trade and immigration (e.g. see Hainmueller and Hiscox 2007). In addition, feelings of European identity, or inclusive forms of nationalism more generally, positively relate to support for the EU and European economic governance (e.g. see Carey 2002;Hooghe and Marks 2009;Kuhn and Stoeckel 2014;McLaren 2002). Affective attitudes toward European integration are closely related to citizens' willingness to make sacrifices to help other member states (Gabel 1998b). We also anticipate weak feelings of European identity to be negatively correlated with support for offering financial help to EU member states that need it.
We expect, however, that domestic macroeconomic context dampens the effect of identity. We posit that the identity dimension of attitudes towards fiscal solidarity will have smaller effects in poor countries. This is because the perceived costs of transnational financial assistance will outweigh the impact of European attachment. In poorer member states, domestic economic conditions may act as a 'buffer' diluting the effect of identity on support for providing financial assistance to the EU's struggling economies (Garry and Tilley 2009;Kuhn and Stoeckel 2014).
H3: Individual-level identity effects on support for fiscal solidarity are weaker in poor EU member states.

Data and methods
In order to test the theoretical expectations, we use individual-level data from the EES Voter Study 2014 (Schmitt et al. 2015) for 28 EU member states combined with country-level macroeconomic data. The EES Voter Study is an EU-wide survey carried out using a nationally representative sample of approximately 1000 interviews in each member state. The total sample size in the analysis is 30,064 respondents. The interviews were conducted in May and June 2014. Although the study has been carried out every five years since 1979, this is the first time that the respondents were explicitly asked to indicate their attitudes towards fiscal solidarity. This entails that the analysis lacks a time dimension by design. While driven by data availability, we acknowledge that relying on a single cross-section remains a major limitation of this study. However, we think that the timing of the survey allows us to make valid inferences on preferences for cross-national fiscal solidarity because it coincided with the discussion on whether Greece should receive additional financial support from international institutions, which was a highly politicised question. While we are not able to directly test the extent to which citizen opinions may change over time, previous research indicates that the salience of monetary integration, amplified by the euro crisis, is positively correlated with stable informed opinions (Hobolt and Wratil 2015). The topic continues to be relevant against the ongoing debate over establishing a euro area fiscal union. The survey question was formulated as follows: To what extent do you agree or disagree with the following statement: In times of crisis, it is desirable for the [country] to give financial help to another European Union Member State facing severe economic and financial difficulties.
The answers were given on a four-point scale (totally agree, tend to agree, tend to disagree, totally disagree). Due to our substantive interest in opposition to fiscal solidarity, we recoded the variable as binary, where 0 ¼ agree and 1 ¼ disagree. 'Don't knows' are excluded (for descriptive statistics see Online Appendix A1). 4 To test the robustness of our results, we replicate the analyses using an alternative dependent variable, measuring general opposition to the EU. The reason for this is to test the possibility that public preferences for the general direction of European integration differ from those for specific EU-related issues (Vasilopoulou and Talving 2019). To measure EU support, we utilise a survey item, which asks respondents whether they consider their country's membership of the EU a good thing, a bad thing, or neither. The variable was recoded in a similar manner to our original dependent variable, with 1 referring to negative attitudes and 0 to all other categories. 'Don't knows' are excluded. The correlation between the two dependent variables is weak (r ¼ 0.20), 5 suggesting that attitudes towards fiscal solidarity differ from general EU opinion.
We test the aggregate-level economic explanation (H1) by relying on a country's macroeconomic situation, measured as GDP per capita in the year prior to survey fieldwork. We utilise Eurostat statistics on GDP per capita in Purchasing Power Standards (PPS), which allows for meaningful volume comparisons between countries. Before modelling, log-transformation was used for the variable to improve the normality of the distribution. This also allows us to consider the possibility that the relationship between GDP levels and the outcome variable is non-linear. As a robustness check, we run the models using payments from the EU budget as a proxy for the domestic economic situation. We expect that citizens in countries that have received more financial benefits from the EU are more likely to consider the austerity-related effects of such assistance and are therefore more opposed to it. To test this, we use a measure of operating budgetary balance (OBB) (percentage of Gross National Income [GNI], 2013), which reflects the difference between what a country receives from and pays into the EU budget. 6 A negative budgetary balance indicates that a country contributes more to the EU budget than it receives from it, and a positive balance that it receives more than it contributes, i.e. the direction of this variable is opposite to the GDP variable. Log-transformation is also used for this variable. The results of the models using OBB are presented in Online Appendix A2. We finally replicate the findings using two more robustness checks, where country macroeconomic context is operationalised using Eurostat data for the unemployment rate (for 2013) as well as GDP per capita change (from the crisis year of 2007). The results appear in Online Appendix A3.
The utilitarian dimension of attitudes towards transnational financial assistance (H2) is operationalised at the sociotropic level using subjective perceptions of national economic performance. Respondents were asked to assess whethercompared to 12 months agothe economic situation in their country has become a lot better, a little better, stayed the same, become a little worse, or a lot worse. This is a standard measure of economic positions in economic voting studies (Lewis-Beck 1988). Although a bulk of evidence confirms that perceptions of the national economy influence political attitudes more than personal material wellbeing (e.g. Kuhn and Stoeckel 2014 on support for European economic integration; Mansfield and Mutz 2009 on attitudes towards free trade; Stegmaier and Lewis-Beck 2013 on political support), we also test the utilitarian explanation at the egocentric level. Unfortunately, survey data do not include personal economic assessments; therefore, we operationalise the egocentric dimension via educational attainment and occupational skills. We assume that higher education and better occupational skills lead to higher levels of human capital. Level of education is measured with number of years spent in full-time education. The answers were recoded into categories 'no full-time education', 'up to 15 years', '16 to 19 years' and '20 years or older'. Respondents still studying were assigned to these categories according to their actual age. Occupational status is measured using the pre-coded categories included in the EES Voter Study, including selfemployed, managers, other white-collar workers, manual workers, house persons, unemployed, retired, and students. These categories are added as dummy variables.
The identity dimension (H3) is operationalised via attachment to Europe. Respondents in the survey were asked to indicate whether they feel attached to Europe, using a scale where 1 ¼ yes, definitely, 2 ¼ yes, to some extent, 3 ¼ not really, and 4 ¼ not at all. We reversed the scale, with higher values indicating a stronger feeling of European identity.
Drawing from the international redistribution and Euroscepticism literatures, the models include standard sociodemographic controls, e.g. age (in full years), gender (0 ¼ male, 1 ¼ female), and social class (1 ¼ the lowest and 10 ¼ the highest level in society). We control for ideology through respondents' self-placement on the left-right scale (0 ¼ left, 10 ¼ right), as preferences for within-EU fiscal aid might be absorbed by the traditional ideological conflict. We additionally employ a squared term of left-right ideology as we expect the effect to be curvilinear, with support for fiscal solidarity being lower among citizens placed at the fringes of the political spectrum. Similar patterns have been observed in studies on international redistribution (e.g. Bechtel et al. 2014) and EU attitudes (e.g. Steenbergen et al. 2007). We further include a control variable that measures preference for domestic redistribution of wealth to account for the possibility that support for redistribution between countries reflects support for redistribution between individuals (e.g. Kleider and Stoeckel 2019). We also control for domestic proxies to account for the possibility that fiscal solidarity reflects approval of the national government (Armingeon and Ceka 2014). At the aggregate level, we include a dummy for countries that received financial assistance from the EU to adjust for possible confounding effects on support for fiscal solidarity. 7 The dichotomous nature of our main and alternative dependent variables requires the use of logistic regression modelling. To account for the clustered structure of data where respondents are nested within countries, we employ multilevel modelling techniques.

Results
Descriptive data show that on average 45% of respondents disagreed that in times of crisis it is desirable for their country to give financial help to EU member states in economic and financial difficulties; 50.8% agreed with the statement and 4.2% did not take a position. These attitudes vary greatly across countries. In Slovakia, for instance, 67.1% of respondents indicated their opposition to fiscal solidarity, whereas in Sweden only 16.1% shared similar views (Figure 1). We may also observe that compared to Western Europe, opposition to providing financial aid is stronger among the Southern and Eastern member states which would potentially benefit more from solidarity measures. Attitudes towards solidarity measures clearly differ from EU opinions more generally. At 13.8%, opposition to the EU is much lower compared to nearly half of Europeans disapproving of fiscal aid. Patterns of country variation diverge between the two dependent variables as well, with levels of overall Euroscepticism being highest in Britain and lowest in Estonia. These tendencies indicate that the two items do not capture different facets of the same reality, but that they are indeed separate phenomena.
Moving on to the multilevel logistic regression analyses, we examine how competing sets of explanations influence attitudes towards providing help to countries in difficulties. Results of the additive model are presented in Model 1 in Table 1. At the contextual level, we firstly test the impact of the national macroeconomic situation on the likelihood of supporting fiscal solidarity (H1). The results suggest that support levels are indeed partly determined by country affluence. Citizens living in wealthy countries are less opposed to fiscal assistance than their counterparts in poorer member states where domestic resources are limited (see also Kuhn et al. 2017;Lengfeld et al. 2015). Substantive interpretation of the results is eased by looking at average marginal effects, which at -0.12 suggest that the likelihood of opposing fiscal solidarity decreases by 0.12 percentage points with every one-unit increase in the logged GDP per capita variable. However, since the model is non-linear and thus the marginal effect may not be constant across all levels of country affluence, we focus our interpretation on predicted probabilities for specific values of the wealth variable. For instance in Bulgaria, which is the poorest country in the sample, the mean predicted probability of opposing fiscal solidarity is 53%, but the likelihood remains much lower at 33% in the wealthiest country, Luxembourg. Note that the mean predicted probability of rejecting transnational financial assistance is rather high in countries where the wealth levels remain below average even if they have received financial assistance, e.g. Latvia (49%) and Greece (48%).
At the individual level, we are interested in utilitarian and identity concerns. In line with previous findings (e.g. Kuhn and Stoeckel 2014), the utilitarian dimension demonstrates significant effects mostly at the sociotropic level: citizens who positively evaluate the economy are more supportive of fiscal aid, whereas negative assessments of the economy lead to scepticism towards helping other countries. One of the two egocentric measures, i.e. education, yields significant effects: compared to less-educated  individuals, citizens with medium and high levels of education are much less likely to reject transnational financial assistance. Occupational skills, however, play a smaller role in attitude formation. The results also indicate that identity helps explain variation in opposition to fiscal assistance, with stronger attachment to Europe yielding lower levels of resistance. On average, then, preferences are determined by economic and cultural attitudes at the individual level, as well as levels of prosperity at the national level. In addition, the outcome is associated with some of the control variables, such as social class, ideological dispositions and government approval. 8 Considering the large contextual variation in attitudes towards transnational financial assistance, we next examine the moderating effects of country affluence on individual-level considerations. We expect to find weaker socioeconomic (H2) and identity (H3) effects in poor member states. To test this, we include interaction terms between GDP per capita and micro-level measures of the socioeconomic and identity dimensions. The results for our key economic variable, i.e. national economic perceptions, are shown in Model 2 in Table 1 and on the upper panel of Figure  2. Economic issues influence views on fiscal solidarity differently depending on the affluence of a country that a person lives in. In countries with low levels of wealth, attitudes are not defined by utilitarian concerns. Average marginal effects largely fall on the zero line, indicating that the effects are not statistically significant. In richer countries, in turn, opinions are conditional on individual-level economic factors: opposition to fiscal solidarity varies across population strata, being highest among people with negative economic assessments (Figure 2). The substantive importance of the results is emphasised by predicted probabilities. In Luxembourg, the wealthiest country in the sample, the mean predicted probability of rejecting fiscal solidarity ranges from 20% to 54% depending on respondents' economic evaluations. In Bulgaria, however, the country with lowest GDP levels, the chances of opposing fiscal aid reach 48-55%, irrespective of personal economic considerations. Put differently, citizens who consider the economy a lot better than a year ago, i.e. the group that is typically expected to support European integration, have a 20% likelihood of opposing fiscal solidarity in Luxembourg, but as high as 48% in Bulgaria. Similarly, optimistic economic perceptions are associated with a much higher probability of rejecting fiscal aid in Greece (39%) and Latvia (42%) than they are in wealthy countries, for example Germany or Sweden (both 30%). In short, in less affluent member states opposition is strong across all economic appraisal groups. Similar patterns appear when we look at egocentric measures of the utilitarian dimension. In wealthy countries, people in certain occupational categories that are associated with higher levels of human capital, such as students, are significantly less likely to express pessimistic views on fiscal solidarity compared to their counterparts in other occupational groups. In poorer member states, however, occupational status fails to explain variation in the dependent variable (Model 1 in Online Appendix A5). The results are even clearer when we operationalise the human capital dimension using educational attainment. Belonging to the highest educational group significantly decreases the likelihood of opposing fiscal solidarity in rich countries but does not distinguish between support levels in member states with scarce financial resources (Model 2 in Online Appendix A5).
Next, we explore contextual variation in identity effects. Model 3 in Table 1 and the lower panel of Figure 2 demonstrate that, similar to economic issues, affective attitudes only explain variation in disapproval of fiscal assistance in countries with high GDP. In Luxembourg, which is the wealthiest country in the sample, predicted probabilities range from 17% for people with strong European identity to 67% for those who do not identify with Europe at all. In Bulgaria, on the other hand, which is the poorest country, the probability of opposing transnational financial assistance remains high at 49-62%, with no substantial differences between groups. Citizens are very likely to reject fiscal solidarity also in poorer debtor countries such as Greece and Latvia even if they strongly identify with Europe (39% and 42%, respectively). The same cannot be said about wealthy countries such as Germany (28%) or Sweden (27%), where a strong European identity clearly lowers opposition. 9 These results are in line with Verhaegen (2018), who demonstrates that respondents with strong European identity who live in crisis-ridden countries do not support offering financial help to countries in need.
Our argument on contextual variation is centred on the notion that citizens in less prosperous countries are likely to reject fiscal solidarity because they are reluctant to share domestic resources that are already scarce. The second part of our argument posits that views on transnational financial assistance are related to poor countries potentially being on the receiving end of that assistance. EU conditionality is associated with harsh and widely unpopular austerity measures. Our analysis thus far focuses on a country's socioeconomic standing as a proxy for domestic context. To directly address the second part of the argument, we test the robustness of the results by utilising a measure of payments from the EU budget. We expect to find thatsimilar to countries with low levels of wealthresidents in member states that have received more financial benefits from the EU express scepticism towards fiscal solidarity, irrespective of their personal economic and affective positions. The results in Models 1 and 2 in Online Appendix A2 suggest that this is the case. With significant interaction effects in both models, the results show that individual-level economic and identity effects are dampened when a country's operating balance budget is high, i.e. it is a net receiver of EU budgetary payments. Under such circumstances, poor economic perceptions and weak attachment to Europe do not define opposition to fiscal aid as clearly as they do in member states that are net contributors to the EU budget. Finally, we repeat the analyses using economic growth from the crisis year as well as unemployment rate as two alternative indicators of the domestic context. The results by and large confirm our key findings: individual considerations differentiate between attitudes towards fiscal aid in those European countries that have positive economic growth or low unemployment, but are not sufficient to explain variation in others, i.e. those member states with large negative economic growth or high unemployment (Online Appendix A3).
Our findings suggest that views on fiscal solidarity differ from those on EU integration more generally. We have shown that in poor member states, resistance to offering financial assistance is homogenously high, even among population segments that are typically considered optimistic towards EU integration, such as those with better socioeconomic status and strong attachment to Europe. We carried out an additional test to account for the possibility that attitudes towards financial assistance and EU support are two separate phenomena. We repeated the analyses using opposition to a country's membership of the EU as the dependent variable. Comparing the models for both dependent variables, we would expect to find different patterns of support and opposition. Models 1 and 2 in Online Appendix A6 mostly support this expectation. Unlike in the fiscal solidarity models, there are no contextual differences in the extent to which economic opinions determine opposition to the EU. Quite the contrary, economic effects perform in a similar manner in poor and in wealthy countries, with negative economic assessments leading to higher levels of Euroscepticism. The identity effects somewhat resemble those found in fiscal solidarity models in that strong attachment to Europe does distinguish between EU attitudes in countries with higher GDP levels, but the impact remains smaller compared to that in our original models. Altogether, our analysis indicates that in contrast to general support for EU integration, which can be mostly explained at the individual level, attitudes towards fiscal aid are embedded in between-country differences, thus creating different support patterns across member states. This suggests that it is important to examine public opinion on specific EU-related issues separately from general EU support because attitudes are structured differently across these dimensions.

Discussion
This article has theorised and empirically substantiated the effect of context on public attitudes towards offering financial help to EU member states in need. First, we find polarisation of attitudes between richer and poorer member states. Citizens in countries that are economically worse off show significantly lower levels of support for providing financial assistance to the EU's struggling economies, even though these are the countries that would presumably benefit more from it. This is because, given resource scarcity, individuals in poorer countries prioritise the domestic compared to the transnational level. In such countries, in addition, programmes of external financial assistance are more likely to be associated with austerity and downward pressures on social spending. Second, individuals with similar predispositions but living in different countries take widely varying views on fiscal solidarity. Micro-level utility and identity considerations do not impact upon preferences for transnational financial assistance in poorer countries. This relationship is a rich country phenomenon. In wealthy countries the individual-level socioeconomic and cultural divide is sharper, whereas in poorer member states providing financial aid to others is perceivedby all social groupsas an unnecessary burden to an economy that is already under strain.
We have proposed that attitudes towards EU fiscal solidarity should be better understood through the prism of contextual conditionality. This helps explain some of the puzzles of political behaviour, i.e. why some individuals oppose policies that both benefit them individually and align with their cultural predispositions. A focus on context adds explanatory power to these individual views. Our findings have severe implications for both research and policy making. First, our approach provides leverage to the long-standing debate about the relative explanatory power of rational choice and cultural approaches. Context influences the ways in which a decision is framed, which has a direct effect on attitude formation and decision making. Individuals' cost-benefit analyses of European integration and its various policies are not consistent across different member states. Equally, the concept of collective identity and its effect on EU policy approval also varies depending on country. In poorer EU member states, context serves as a heuristic, which is strong enough to dampen the effects of utility and identity. Our model calls for a significant re-evaluation of two key theories of political behaviour and EU preference formation. This is very important, as current literature has prioritised the identity/cosmopolitanism perspective in explaining such preferences (e.g. Bechtel et al. 2014;Kuhn et al. 2017;Verhaegen 2018), which plays no significant role in poorer EU member states. Although citizens in countries with low wealth levels tend to be recipients of EU financial assistance, they are less willing to support such initiativesa relationship that is also evident among countries that have benefited from EU aid. Our work endeavours to initiate a new research agenda that examines preferences for financial assistance not only in richer but also in poorer EU member states. Thus far, case study research on preferences for transnational financial assistance focuses on Germany (Bechtel et al. 2014(Bechtel et al. , 2017Stoeckel and Kuhn 2018; see also Kuhn et al. 2017 on Germany and the UK), arguing that 'the bailouts have faced significant popular backlash in donor countries as voters denounced the transfer of billions of taxpayer funds to prop up other countries' economies' (Bechtel et al. 2017: 864). We have shown that opposition may also be found in poorer member states, which are potential or actual recipients of these funds. Our results demonstrate that the dynamics of support for such initiatives might be different in these member states and that findings related to Germany might not be directly applicable in such contexts. Future research should therefore also investigate preferences for transnational financial assistance in other EU member states.
Second, our results place an important caveat on existing research. On the one hand, citizens living in core EU member states tend to prefer institutional integration over Eurobonds, as a means of limiting their fiscal exposure to weaker countries (Nicoli 2019), i.e. a form of 'conditional solidarity'. On the other hand, citizens living in crisis-ridden countries view further EU economic governance integrationi.e. EU supervision of national budgets, policy coordination and the creation of a finance ministeras a way out of the crisis. They tend to prefer Eurobonds instead because the latter decrease their countries' borrowing risks and costs (Kuhn and Stoeckel 2014;Nicoli 2019). However, we show that citizens in poorer member states might also associate economic governance with austerity and fiscal supervision and thus oppose aspects of it. They tend to view the question of financial assistance as a potential obligation of wealthy nations to address EU-level structural imbalances by bearing the costs of bailing out countries in crisis. In poorer countries with budgetary constraints, the additional risk of austerity serves to further undermine solidarity. Taken together, these findings illustrate conflicting attitudes towards the various dimensions of EU fiscal cooperation and suggest that EU political initiatives on such issues may suffer a two-fold challenge. First, in richer member states, support for economic governance and fiscal solidarity might weaken given the strong presence of far-right Eurosceptic parties that politicise issues of deservingness across borders. Second, in poor countries that are potential recipients of EU funds, burden-sharing may not be considered as their primary responsibility, which may ultimately further solidify the idea that questions of economic malaise should be addressed at the domestic level. For example, the poor-rich distinction allows us to contextualise the result of the 2015 Greek referendum on whether the country would accept the proposals of the EU/IMF. Despite the dual potential risk of default and Grexit, citizens rejected the lenders' reform plans. The main reason behind the result related to opposition to externally imposed austerity, and a general feeling that a 'No' vote would ultimately lead to renewed negotiations (Tsatsanis and Teperoglou 2016).
European solidarity, of course, is a multidimensional phenomenon that differs from national solidarity (Baute et al. 2018;Nicoli 2015). With specific reference to within-EU financial assistance, it should be noted that systemic resilience is contingent upon common solutions, cross-national unity, and the understanding of mutual responsibility and interdependence (see also Ferrera and Burelli 2019). This in part depends on the extent to which rich countries' solidarity is perceived as 'conditional', i.e. tied to conditionality and austerity, and promoting political inequality through a distinction between the 'rulers' and the 'ruled'. It equally depends on the political willingness of poorer member states to position themselves at the heart of European integration and also engage in some form of risk sharing. Our findings indicate that to ensure popular acceptance of new cross-European initiatives, therefore, political elites may need to frame the need for financial assistance differently, depending on country wealth. Such policies are more likely to succeed if initiated, spearheaded and promoted by political actors across the union, rather than solely from richer 'motor' EU member states.
Finally, we show that the same individuals in different countries might support European integration but may not be prepared to support costly policies. In doing so, we highlight the importance of examining Euroscepticism going beyond questions of general 'diffuse' support for the system. Investigating attitudes towards EU issues as objects of 'specific' support reveals that there are different constellations of preferences on different policy domains. This suggests that the 'winner-loser' structure is uniform neither across EU member states nor across policy areas (Vasilopoulou and Talving 2019), and points to the challenge of addressing Euroscepticism with a one-size-fits-all strategy.

Notes Funding
This article draws upon research funded by an ESRC Future Leaders grant, ES/ N001826/1.