ABSTRACT
The utilitarian approach to pro-EU attitudes – noting that citizens establish their preferences based on their perceived self-interest – remains the dominant one in the social science literature on these attitudes. Yet previous work following this approach has overlooked the role of subjective financial insecurity. Based on prospect theory and marginal utility theory, we argue that individuals who feel financially insecure determine their preference for further European unification in terms of the gains and losses for themselves and that, since they are disproportionately sensitive to economic losses, they display more risk aversion and oppose further macro-political changes in the form of further European unification. Using hybrid models and 15 waves of a representative panel survey conducted in the Netherlands and covering 2008–2023, the evidence strongly supports our expectation. Controlling for individual education, occupational status, individual income, gender and age, subjective financial insecurity is cross-sectionally and longitudinally related to support for European integration. People generally feeling financially insecure (those who over time increased their feeling of financial insecurity) display significantly less support for further European unification than people who generally feel financially secure (people who over time did not feel more financially insecure).
While seminal theoretical works on European integration were published shortly after the creation of the European Coal and Steel Community (e.g. Deutsch et al.1957; Haas 1958), social scientists started to gain interest for the topic of citizens’ EU support in the mid-1990s (Eichenberg and Dalton 1993; Anderson and Reichert 1995; Gabel and Palmer 1995; Anderson and Kaltenthaler 1996; Gabel and Whitten 1997; Gabel 1998). Since then, a plethora of empirical studies have advanced our understanding of citizens’ attitudes towards the European Union and its integration project.
As the European integration project was originally restricted to the creation of an economic community, social scientific work on citizens’ EU support has paid special attention to the relevance of individuals’ direct material interests as determinant of EU support. The first empirical studies – including the ones of Gabel, Anderson and colleagues (Eichenberg and Dalton 1993; Anderson and Reichert 1995; Gabel and Palmer 1995; Anderson and Kaltenthaler 1996; Gabel and Whitten 1997; Gabel 1998) – investigated the power of various material factors (such as personal socio-economic position or national economic factors like GDP) in explaining citizens’ EU support. This utilitarian approach assumes that citizens who obtain or perceive some benefit from EU integration will be more likely to support it, while citizens who experience the costs of EU integration are more likely to oppose it. Many empirical studies confirm the persisting relevance of this utilitarian approach for understanding attitudes toward the EU (Foster and Frieden 2021).
The utilitarian perspective has faced major challenges since the 2000s by, first, a cultural approach and, later on, an elite cueing approach. Since the beginning of the 2000s scholars consider the role of values and (national) identity in shaping EU support (e.g. McLaren 2002; Hooghe and Marks 2004; McLaren 2004). Next to this cultural turn, political scientists also investigate the role of national public debates and elites in cueing citizens on their attitudes toward the EU (e.g. Hooghe and Marks 2005).
While acknowledging the contributions of the cultural and cueing approaches, this manuscript revisits principles of the utilitarian approach. The utilitarian, value and collective identity perspectives constitute indeed complementary analytical tools to understand citizens’ attitudes toward the EU. We specifically explore a single research question: What is the role of subjective financial insecurity in explaining citizens’ support for further EU unification? We define subjective financial insecurity as the negative feeling derived from the perception of not having enough financial resources to meet one's needs.
We contribute to the utilitarian approach by focusing on the role of perceptions of one's financial situation. Extensive literature on social theory and social psychology has shown that perceptions of social reality – what Bourdieu et al. (1991) called ‘spontaneous sociology’ – influence individual attitudes and behaviours (Thomas 1938; Eveland et al.2008). We draw on prospect theory (Kahneman and Tversky 1979) and diminishing marginal utility theory (Gossen 1983 [1854]) to argue that citizens who feel financially insecure tend to be more risk-averse and prefer the status quo over changes. According to Kahneman and Tversky (1979), people perceive the outcome of a decision in terms of the gains and losses such an outcome would mean for their current personal situation. Furthermore, diminishing marginal utility theory (Gossen 1983 [1854]) predicts that losses will be relatively more difficult to bear for people who are economically deprived. Consequently, they tend to be more risk-averse and less likely to favour any change than well-off people. We argue that this mechanism also applies to socio-political attitudes. We therefore expect subjective financial insecurity to be negatively associated with support for further EU unification.
The focus on the subjective financial situation overcomes the restrictive consideration of objective conditions that has characterised most work on the utilitarian approach (e.g. Anderson and Reichert 1995; Gabel and Palmer 1995). It also overcomes the limitations of work on prospective and retrospective considerations (e.g. Hooghe and Marks 2004) by addressing perceptions of personal circumstances that are relatively stable and non-conjunctural. Continuous advancements in statistical techniques and the availability of more sophisticated survey data – such as panel survey data – enable us to refine the main findings of pioneering empirical studies on citizens’ EU support. For this purpose, we have analysed the Dutch survey panel database Longitudinal Internet Studies for the Social Sciences (LISS) and the 2008–2023 waves.
The Netherlands is one of the six founding EU member states and until recently a ‘Europhile nation’, given its relatively strong popular EU support (van den Hoogen et al.2022a). Dutch public opinion on the EU has become more divided and more negative since 2010 as a result of the Eurozone crisis (de Vreese et al.2017). Like many EU member states, the Netherlands has a prosperous economy with a mature and generous welfare state. This means that Dutch residents feeling financially insecure are far from being in a situation where they do not have anything to lose.
The use of the Dutch survey panel database LISS enables us to decompose the effects of subjective economic insecurity on citizens’ support of EU unification into within- and between-individual effects (Allison 2009; Schunck 2013; Wooldridge 2015). Estimating within-individual effects allows us to assess the net effect of subjective financial insecurity on support for EU unification controlling for time-constant characteristics (e.g. gender, cohort of birth or stable personality traits such as tendency of risk avoidance or loss aversion) and theoretically-relevant time-varying factors (e.g. occupation, employment status, net income). This, in turn, provides us with much more precise estimates of the effect of subjective financial insecurity than previous studies based on cross-sectional survey data. Between-individual effects, in contrast, can better capture long-term effects. In a nutshell, our results show that both between- and within-individual measures of subjective financial insecurity are significantly and negatively associated with support for further European unification1 even when controlling for the between- and within-individual effects of respondents’ objective socio-economic position.
In what follows, we first describe the stand of the empirical literature on the effect of subjective financial insecurity on support for further European unification. We then present a theoretical mechanism that could explain why subjective financial insecurity is negatively associated with support for further unification. Lastly, we describe the data and methodological approach before interpreting our results and concluding.
Previous studies on the role of subjective financial insecurity for EU support
Gabel and Whitten (1997) were among the first to highlight the role of subjective evaluation of respondents’ economic situation for understanding EU support.2 They argue that citizens’ perception of the economy might be inconsistent with the objective economic reality (Gabel and Whitten 1997). Their analysis based on Eurobarometer data shows that (independently from objective measures of the socio-economic position) citizens who consider their household's financial situation to have improved over the last 12 months are significantly more likely to support EU integration (Tucker et al.2002; Christin 2005). Other studies also based on Eurobarometer data assessed the effect of individuals’ future expectations regarding a deterioration of their household's financial situation on EU support and come to similar conclusion (Díez-Lanchas et al.2021): both retrospective and prospective perceptions of the household's financial situation have proven significantly associated with EU support independently from objective measures of the socio-economic position.
In their landmark work on losers and winners of globalisation, Kriesi et al. (2008) also highlighted the complementarity of objective and subjective measures of socio-economic position in explaining support of opening up of national borders for political integration: those whose life chances were traditionally protected by the national boundaries perceive the weakening of national border as a threat to their social status and their social security (Kriesi et al.2008). Accordingly, feeling of economic insecurity is likely to be a key determinant of opposition to the EU, even when considering individuals’ objective socio-economic position. We define subjective financial insecurity as the negative feeling derived from the perception of not having enough financial resources to meet one's needs. It is often characterised by the belief that one may not be able to pay bills or cover unexpected costs (Bossert and D’Ambrosio 2013; Hacker et al.2013). Please note that according to our working definition, subjective financial insecurity is not merely a neutral assessment of one's household's poor financial situation, but it also involves the emotions attached to it. Subjective financial insecurity usually involves negative feelings such as anxiety or dissatisfaction.
Previous studies on EU support assessing the role of subjective economic evaluation use two main types of measurement. The first is the one used in the aforementioned study of Gabel and Whitten (1997) and asks respondents to evaluate retrospective or prospective short-term changes in their household's financial situation.3 These items are provided by the Eurobarometer and can be answered with a three- to five-point Likert scale. Cross-sectional studies using either the retrospective version (Gabel and Whitten 1997), the prospective version (Hooghe and Marks 2004; Díaz-Lanchas et al.2021) or an index combining both (Tucker et al.2002; Christin 2005) consistently highlight a significant effect of each of these measurements on EU support even when controlling for the objective socio-economic position.
These items, however, constitute a poor measurement of subjective financial insecurity for two main reasons. First, a retrospective or prospective evaluation of the household's financial situation assesses change without considering respondents’ current level of economic deprivation. Experiencing an improvement of the financial situation over the last 12 months does not necessarily imply a comfortable financial situation at the time of the survey. Vice versa, facing a deterioration of the financial situation in the last 12 months does not necessarily imply a state of economic insecurity at the time of the survey. Items on change of the financial situation over time without considering the absolute financial situation constitutes thus a poor measurement of economic insecurity.
Second, and more importantly to our purpose, these items on retrospective and prospective evaluation of the household's financial situation are very close to the objective measurements of socio-economic position, as they ask respondents to give a neutral assessment of their financial situation. They thus fail to encompass how respondents feel about their financial situation – an emotional evaluation component. Subjective feeling about one's own economic situation does not however necessarily depend on the objective level of income or wealth: respondents belonging to the high-income segment of the population might feel economically dissatisfied (Citrin et al.1975). Citizens perceive and judge the same economic situation differently, which highlights the importance of considering citizens’ subjective judgement of their economic situation in addition to their objective socio-economic position when assessing the overall role of economic situation on political attitudes.
The second main type of item used in studies on EU support provides a more accurate measure of subjective economic insecurity as it incorporates a subjective judgement of the household's financial situation (‘how would you judge the current financial situation of your household?’ with a four-point answer scale: very bad, rather bad, rather good, very good). Cross-sectional studies using this item from Eurobarometers consistently show that the subjective judgement of the household's financial situation is significantly associated with EU support, even when controlling for respondents’ objective socio-economic position (Levy and Phan 2014; Teney et al.2014). Despite the measurement improvement provided by this second type of item over the prospective and retrospective financial evaluation, it can still be considered as a rather poor measurement of subjective economic insecurity for two main reasons. First, the explanatory power of an item tapping at subjective economic dissatisfaction can be increased by including a larger number of answer categories (Rosenstone et al.1986; Dominitz and Manski 1996). Thus, the quality of a measurement of subjective financial insecurity would be improved by the inclusion of a larger answer scale. Second, combining items about subjective insecurity regarding economic ‘flows’ -such as income- as well as economic ‘stock’ – such as wealth – provides a more comprehensive measurement of the overall subjective economic dissatisfaction (Hacker et al.2013). The use of a better measurement of subjective economic dissatisfaction can indeed largely increase its power in explaining political attitudes (Wroe 2016).
Following this reasoning, we instead use an index composed of two eleven-point-Likert scale items measuring (i) how satisfied is the respondent with her/his financial situation and (ii) if it is hard or easy to live off the household income. Our measure of subjective financial insecurity constitutes therefore an improvement over the traditional measures used in previous studies on EU support.
Theoretical mechanism behind subjective financial dissatisfaction and support for EU unification
Whereas prior work on the link between subjective financial insecurity and support for EU unification draws on inadequate indicators of subjective financial insecurity, it also generally fails to discuss the theoretical mechanism of this link. These studies usually highlight the importance of considering respondents’ subjective perception or evaluation of their economic situation, as ‘any utilitarian calculations made by an individual are necessarily conducted through his perceptions of the surrounding economic world’ (Levy and Phan 2014: 569). Yet previous work does not clarify exactly how subjective financial dissatisfaction affects attitudes toward the European Union independently from the effect of the objective socio-economic positions.
In this section, we combine principles of prospect theory (Kahneman and Tversky 1979; Kahneman 2011) and diminishing marginal utility theory (Gossen 1983 [1854]) to present a theoretical mechanism for the link between subjective financial insecurity and attitudes towards EU integration. We argue that a feeling of financial insecurity leads to risk aversion, which, in turn, implies opposition to changes, including those resulting from further EU unification.
According to Kahneman and Tversky (1979), people perceive the outcome of a decision in terms of gains and losses defined with a reference point (usually the current state), rather than in terms of final stage. This tendency helps us understand how respondents answer to our attitudinal item on a further EU unification. In answering this item, respondents do not imagine the final political state such an EU unification process would lead to. Rather, they tend to evaluate the kind of changes a further European unification would imply in terms of losses or gains to their current situation.
Based on this principle of prospect theory, we expect respondents to answer the question on support for further EU integration in terms of two poles: preference for (1) changes (further unification) or for the (2) status quo (unification has gone too far). Favouring the status quo implies choosing the alternative that provides certainty, as any deviation from the status quo implies uncertainty regarding the consequences of non-status quo options (Eidelman and Crandall 2012). This means that favouring status quo means being risk averse (Anderson 2003: 143; see also Kahneman et al.1991). Supporting further European unification means accepting the risks arising from the uncertainty of departing from the status quo. More specifically, respondents selecting this option do so under the expectation that the upside risks trump the downside risks. The corollary of this reasoning is that respondents who consider that EU unification has gone too far are more risk-averse than respondents who favour further EU unification.
Preferences for changes or for the status quo regarding EU integration depend, in turn, among others, on one's feeling of financial insecurity. According to the diminishing marginal utility theory (Gossen 1983 [1854]), people who feel financial insecurity tend to be more sensitive to potential losses that might result from change than people who feel economically well-off. Indeed, any kind of financial loss will be more difficult to bear for people who feel financially insecure than for those who are economically satisfied. Losing a part of your income when you perceive that you cannot currently cope with your financial situation has more dramatical implications for your economic well-being than losing a part of your income when you feel economically well-off. This, in turn, implies that people who feel economically insecure ought to be more risk-averse than people who feel economically well-off, since any financial loss will weigh more dramatically on their economic well-being.
Although we are not aware of empirical studies on the link between subjective financial insecurity and risk aversion, being loss averse has been shown to be positively and significantly correlated with preference for status quo policies (Moshinsky and Bar-Hillel 2010). Moreover, several works assess the role of objective economic insecurity. Dohmen et al. (2011) show, for instance, that low income respondents from the German Socio-Economic Panel are significantly more risk-averse than high income respondents. Moreover, Guiso and Paiella (2008) highlight with Italian survey data that respondents who are more likely to face income uncertainty or liquidity constraints tend to be significantly more risk-averse. Lastly, Díez Medrano and Braun (2012) point to the significant negative association between income and support for economic protectionism.
If we apply the mechanism described by diminishing marginal utility theory to our item on EU unification, we can expect respondents who feel financially insecure to oppose further EU unification to a larger extent than respondents who feel financially well-off. Indeed, respondents who feel financially insecure will tend to avoid any risks resulting from change (and favour instead the maintenance of the status quo). Being risk-averse applies to anyone who feels financially insecure, regardless their objective socio-economic position. Hence, the theoretical mechanism we described applies because of respondents’ subjective feeling regarding their own financial situation, not because of their objective position in the socio-economic space.
Theoretical mechanism between subjective financial insecurity and opposition to further EU unification.
Theoretical mechanism between subjective financial insecurity and opposition to further EU unification.
Data and methods
This manuscript leverages the LISS Panel, which is a probabilistic yearly panel of individuals living in 5000 households of the Netherlands.4 LISS has been active since 2007–2008 and up to date includes 15 waves. We utilise waves 1–15 in our analysis (covering 2008–2023). LISS has several favourable characteristics for analyses of pro-European attitudes. All 15 considered LISS waves include a questionnaire item on support for further European unification. They also include indicators of dimensions commonly stressed by the literature on European attitudes (i.e. age, educational level, employment status, gender, income and occupational group), as well as an index of two questions on subjective economic deprivation.
The outcome of the following analysis involves support for the perception that European unification has gone too far versus the perception that it should go a step further – an item drawn from European Social Survey and subject to extensive research. This LISS item has already been successfully utilised in recent work on the role of different education fields (Kunst 2020), generational differences (Ringlerova 2019) and the consistency of between- and within-person attitudes (Brandt 2021). The European unification is available for 80,034 individual-wave observations corresponding to 15,839 different respondents.
Our single prediction is that attitudes towards European unification are linked to subjective financial insecurity. We measure the latter through a latent factor constructed with two LISS items concerning (i) how satisfied is the respondent with her/his financial situation and (ii) if it is hard or easy to live off the household income. Both items are strongly correlated (r = .735, p < .05). To facilitate the interpretation, we reverse coded both variables. Using principal components factor analysis (PCA), we afterwards construct a single-item latent variable subjective financial insecurity based on the predicted values of the PCA (eigenvalue = 1.735, explained proportion = 86.73%).
The models control for six dimensions noted as significant predictors of pro-European attitudes in extant research. Women and older age groups have been shown to be moderately less enthusiastic about European unification (McLaren 2002; Clements 2009). All models thus control for female and four age groups (16–29, 30–44, 45–59, 60–74 and 75+). The individual level of education has proven the most reliable predictor of pro-EU attitudes (Reference removed for review; Foster and Frieden 2021). We control for this dimension by distinguishing respondents with secondary or lower, intermediate vocational, higher vocational and university education. According to the utilitarian approach, employment status, occupational class and income also shape these orientations. Employed respondents, those who are not unemployed, who hold upper-class occupations and have higher income have more to gain from European unification, which under rational action principles ought to affect their preferences (Gabel 1998; Gabel and Palmer 1995; Gabel and Whitten 1997). We therefore control for status of employed, unemployed, five occupational classes (professional with academic training, professional supervisor, intermediate professional, other service professional and manual worker) and income.
Individual purchasing power is captured through CPI-adjusted, net individual income logged in the LISS Background Module, which captures the net individual income and imputes missing values. We adjust the values to 2015 real income to control for inflation changes over the observed period and – due to the strong right-hand skew – use the logged transformation. The Supplementary materials file includes definitions and operationalizations of all variables. Table A1 in the Supplementary materials file provides descriptive statistics.
For the main following analyses, we estimate linear hybrid models. Hybrid models are increasingly common in research on socio-political attitudes and have been utilised by previous work on attitudes towards European unification (Fernández and Eigmüller 2018; Kuhn et al.2021). They leverage the fact that panel surveys capture both variation over time in individual characteristics and variation between respondent characteristics to decompose within and between effects in a single set of estimates. This decomposition is achieved through the estimation of separate coefficients for two dimensions: one reflecting strict within-individual changes and another one reflecting between-individual differences. The former is captured in the fixed effects (FE) component of the model and the latter is captured in the between effects (BE) component (Allison 2009; Schunck 2013; Wooldridge 2015).
The FE component includes variables that reflect deviations from the person-specific variable mean (i.e. demeaned values) to assess the association between within-respondent changes in an independent variable and the outcome. In other words, it includes variables whose values are positive (negative) if they are higher (lower) than the individual's mean over all waves she/he participates in. Through this process, hybrid models provide estimates equivalent to standard FE models, which have the important advantage of preventing biases in parameter estimates caused by time-constant unobserved heterogeneity (Halaby 2004; Wooldridge 2015). The BE component, instead, includes variables that are time-constant (e.g. female) or person-specific means (e.g. mean individual value of subjective financial insecurity) to assess the covariation between constant individual characteristics. The BE component is, thus, equivalent to cross-sectional analyses. Estimates in the BE component have the limitation of being potentially biased due to time-constant unobserved heterogeneity. Yet they have the important property of reflecting between-respondent variation, which is commonly a large source of variation in panel data. They are also better equipped at capturing long-term effects of explanatory factors.
Descriptive results
We begin the presentation of the results with a descriptive analysis. If Dutch citizens do not display substantial between-respondents and within-respondents variation in their support for European unification and their subjective financial insecurity, an examination of their relationship would not be warranted. If, moreover, both variables (subjective financial insecurity and support for European unification) do not have a bivariate relationship, the estimation and interpretation of hybrid models would not be warranted.
In line with previous work (Lubbers and Japers 2011; de Vreese et al.2017), LISS data indicates that the Dutch public opinion displays noticeable division over the advantages of deepening European unification. In all waves the modal value in attitudes towards European unification is the intermediate category (3), but on average as many as 23.87.1% and 5.69% of all respondents take the extreme positions of considering that European unification has ‘gone too far’ or that ‘it should go further’, respectively (Figure A1). Also, importantly, the evidence indicates non-negligible within-case variation in support for European unification (Figure A2). Most position-takings remained stable across consecutive waves (54.97%), but as many as one in every five position-takings decreased (22.77%) and increased (22.26%) their support over time, respectively.
Similar patterns can be documented on the level of subjective financial insecurity. Evaluations of personal financial insecurity differ substantially between respondents across all waves (Figure A3). A FE model with wave as the only prediction indicates that on average, respondents have over time become significantly less satisfied with their financial situation (β = −.025, p < .005). Support for European unification has decreased significantly over time (β = −.004, p < .005), although the effect magnitude is small. The evidence thus indicates substantial between-respondent and within-respondent variation in both key dimensions: subjective financial insecurity and support for further European unification.6
Bivariate relationship between European unification and subjective financial insecurity.
Bivariate relationship between European unification and subjective financial insecurity.
Multivariate results
To assess the relationship between subjective financial insecurity and the outcome, we estimate seven models (Table 1). Due to the possibility that subjective financial insecurity could mediate the association of objective socio-economic factors (e.g. individual income or individual education) and the outcome, models 1 and 2 only include objective socio-economic factors to identify whether they are related to the outcome. Models 3–7 include subjective financial insecurity and different groups of objective conditions (education, employment status, class status and income) with model 7 as the final full model. We begin the analysis with the between-effects component, which is equivalent to standard cross-sectional OLS and provides a baseline estimate of the association of subjective financial insecurity with the outcome.
. | Model 1 . | Model 2 . | Model 3 . | Model 4 . | Model 5 . | Model 6 . | Model 7 . |
---|---|---|---|---|---|---|---|
Fixed effects component | |||||||
Intermediate vocational education (ref. secondary educ. or less) | −0.020 | −0.055 | −0.034 | −0.037 | −0.047 | −0.031 | −0.045 |
(0.036) | (0.042) | (0.047) | (0.047) | (0.050) | (0.048) | (0.052) | |
Higher vocational education | 0.044 | 0.022 | 0.069 | 0.057 | 0.055 | 0.057 | 0.038 |
(0.037) | (0.042) | (0.047) | (0.047) | (0.051) | (0.049) | (0.053) | |
University education | 0.089* | 0.062 | 0.068 | 0.048 | 0.054 | 0.044 | 0.012 |
(0.048) | (0.058) | (0.063) | (0.063) | (0.074) | (0.064) | (0.075) | |
Employed | 0.014 | 0.022 | 0.009 | ||||
(0.019) | (0.019) | (0.021) | |||||
Unemployed | −0.013 | 0.011 | 0.003 | ||||
(0.028) | (0.031) | (0.032) | |||||
Professional with academic training (ref. manual worker) | 0.066 | 0.057 | 0.058 | ||||
(0.047) | (0.052) | (0.053) | |||||
Professional supervisor | −0.076* | −0.100** | −0.099* | ||||
(0.045) | (0.050) | (0.051) | |||||
Intermediate professional | 0.009 | 0.006 | −0.000 | ||||
(0.030) | (0.035) | (0.036) | |||||
Other service sector | 0.001 | −0.023 | −0.018 | ||||
(0.031) | (0.036) | (0.037) | |||||
Individual income | 0.004 | 0.005 | 0.006 | 0.005 | |||
(0.003) | (0.004) | (0.004) | (0.005) | ||||
Subjective financial insecurity | −0.030*** | −0.029*** | −0.032*** | −0.029*** | −0.031*** | ||
(0.007) | (0.007) | (0.008) | (0.007) | (0.008) | |||
Between effects component | |||||||
Intermediate vocational education (ref. secondary educ. or less) | −0.012 | −0.010 | 0.022 | 0.045* | 0.008 | 0.024 | 0.022 |
(0.021) | (0.023) | (0.023) | (0.023) | (0.024) | (0.024) | (0.025) | |
Higher vocational education | 0.442*** | 0.381*** | 0.406*** | 0.429*** | 0.342*** | 0.416*** | 0.358*** |
(0.021) | (0.026) | (0.023) | (0.024) | (0.027) | (0.024) | (0.028) | |
University education | 0.824*** | 0.712*** | 0.771*** | 0.794*** | 0.665*** | 0.792*** | 0.690*** |
(0.029) | (0.037) | (0.032) | (0.033) | (0.040) | (0.033) | (0.041) | |
Employed | −0.112*** | −0.116*** | −0.124*** | ||||
(0.025) | (0.023) | (0.027) | |||||
Unemployed | −0.223** | −0.061 | −0.031 | ||||
(0.089) | (0.090) | (0.097) | |||||
Professional with academic training (ref. manual worker) | 0.334*** | 0.273*** | 0.282*** | ||||
(0.045) | (0.049) | (0.050) | |||||
Professional supervisor | 0.240*** | 0.190*** | 0.202*** | ||||
(0.041) | (0.044) | (0.046) | |||||
Intermediate professional | 0.190*** | 0.170*** | 0.182*** | ||||
(0.026) | (0.028) | (0.029) | |||||
Other service sector | 0.115*** | 0.112*** | 0.117*** | ||||
(0.026) | (0.028) | (0.029) | |||||
Equivalized income | −0.016*** | −0.008 | −0.010** | −0.001 | |||
(0.004) | (0.005) | (0.005) | (0.006) | ||||
Subjective financial insecurity | −0.149*** | −0.153*** | −0.145*** | −0.146*** | −0.144*** | ||
(0.010) | (0.010) | (0.011) | (0.011) | (0.011) | |||
Female | −0.018 | −0.030** | −0.037** | −0.023 | −0.032* | −0.034* | −0.036** |
(0.017) | (0.015) | (0.018) | (0.017) | (0.017) | (0.019) | (0.018) | |
Age 30–44 (ref. age 18–29) | −0.174*** | −0.163*** | −0.165*** | −0.182*** | −0.175*** | −0.182*** | −0.177*** |
(0.019) | (0.018) | (0.020) | (0.021) | (0.021) | (0.023) | (0.022) | |
Age 45–59 | −0.247*** | −0.234*** | −0.249*** | −0.255*** | −0.263*** | −0.255*** | −0.250*** |
(0.020) | (0.019) | (0.021) | (0.023) | (0.023) | (0.025) | (0.024) | |
Age 60–74 | −0.247*** | −0.237*** | −0.271*** | −0.256*** | −0.286*** | −0.252*** | −0.251*** |
(0.021) | (0.020) | (0.024) | (0.024) | (0.025) | (0.026) | (0.025) | |
Age 75+ | −0.262*** | −0.243*** | −0.295*** | −0.266*** | −0.315*** | −0.263*** | −0.254*** |
(0.028) | (0.027) | (0.032) | (0.031) | (0.032) | (0.033) | (0.031) | |
Wave FE | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
Constant | 2.592*** | 2.794*** | 2.698*** | 2.710*** | 2.771*** | 2.610*** | 2.778*** |
(0.026) | (0.028) | (0.034) | (0.028) | (0.030) | (0.032) | (0.041) | |
Observations | 66,932 | 75,870 | 63,748 | 59,429 | 59,427 | 54,302 | 56,909 |
Number of individuals | 13,235 | 15,177 | 12,733 | 11,949 | 11,949 | 10,969 | 11,520 |
. | Model 1 . | Model 2 . | Model 3 . | Model 4 . | Model 5 . | Model 6 . | Model 7 . |
---|---|---|---|---|---|---|---|
Fixed effects component | |||||||
Intermediate vocational education (ref. secondary educ. or less) | −0.020 | −0.055 | −0.034 | −0.037 | −0.047 | −0.031 | −0.045 |
(0.036) | (0.042) | (0.047) | (0.047) | (0.050) | (0.048) | (0.052) | |
Higher vocational education | 0.044 | 0.022 | 0.069 | 0.057 | 0.055 | 0.057 | 0.038 |
(0.037) | (0.042) | (0.047) | (0.047) | (0.051) | (0.049) | (0.053) | |
University education | 0.089* | 0.062 | 0.068 | 0.048 | 0.054 | 0.044 | 0.012 |
(0.048) | (0.058) | (0.063) | (0.063) | (0.074) | (0.064) | (0.075) | |
Employed | 0.014 | 0.022 | 0.009 | ||||
(0.019) | (0.019) | (0.021) | |||||
Unemployed | −0.013 | 0.011 | 0.003 | ||||
(0.028) | (0.031) | (0.032) | |||||
Professional with academic training (ref. manual worker) | 0.066 | 0.057 | 0.058 | ||||
(0.047) | (0.052) | (0.053) | |||||
Professional supervisor | −0.076* | −0.100** | −0.099* | ||||
(0.045) | (0.050) | (0.051) | |||||
Intermediate professional | 0.009 | 0.006 | −0.000 | ||||
(0.030) | (0.035) | (0.036) | |||||
Other service sector | 0.001 | −0.023 | −0.018 | ||||
(0.031) | (0.036) | (0.037) | |||||
Individual income | 0.004 | 0.005 | 0.006 | 0.005 | |||
(0.003) | (0.004) | (0.004) | (0.005) | ||||
Subjective financial insecurity | −0.030*** | −0.029*** | −0.032*** | −0.029*** | −0.031*** | ||
(0.007) | (0.007) | (0.008) | (0.007) | (0.008) | |||
Between effects component | |||||||
Intermediate vocational education (ref. secondary educ. or less) | −0.012 | −0.010 | 0.022 | 0.045* | 0.008 | 0.024 | 0.022 |
(0.021) | (0.023) | (0.023) | (0.023) | (0.024) | (0.024) | (0.025) | |
Higher vocational education | 0.442*** | 0.381*** | 0.406*** | 0.429*** | 0.342*** | 0.416*** | 0.358*** |
(0.021) | (0.026) | (0.023) | (0.024) | (0.027) | (0.024) | (0.028) | |
University education | 0.824*** | 0.712*** | 0.771*** | 0.794*** | 0.665*** | 0.792*** | 0.690*** |
(0.029) | (0.037) | (0.032) | (0.033) | (0.040) | (0.033) | (0.041) | |
Employed | −0.112*** | −0.116*** | −0.124*** | ||||
(0.025) | (0.023) | (0.027) | |||||
Unemployed | −0.223** | −0.061 | −0.031 | ||||
(0.089) | (0.090) | (0.097) | |||||
Professional with academic training (ref. manual worker) | 0.334*** | 0.273*** | 0.282*** | ||||
(0.045) | (0.049) | (0.050) | |||||
Professional supervisor | 0.240*** | 0.190*** | 0.202*** | ||||
(0.041) | (0.044) | (0.046) | |||||
Intermediate professional | 0.190*** | 0.170*** | 0.182*** | ||||
(0.026) | (0.028) | (0.029) | |||||
Other service sector | 0.115*** | 0.112*** | 0.117*** | ||||
(0.026) | (0.028) | (0.029) | |||||
Equivalized income | −0.016*** | −0.008 | −0.010** | −0.001 | |||
(0.004) | (0.005) | (0.005) | (0.006) | ||||
Subjective financial insecurity | −0.149*** | −0.153*** | −0.145*** | −0.146*** | −0.144*** | ||
(0.010) | (0.010) | (0.011) | (0.011) | (0.011) | |||
Female | −0.018 | −0.030** | −0.037** | −0.023 | −0.032* | −0.034* | −0.036** |
(0.017) | (0.015) | (0.018) | (0.017) | (0.017) | (0.019) | (0.018) | |
Age 30–44 (ref. age 18–29) | −0.174*** | −0.163*** | −0.165*** | −0.182*** | −0.175*** | −0.182*** | −0.177*** |
(0.019) | (0.018) | (0.020) | (0.021) | (0.021) | (0.023) | (0.022) | |
Age 45–59 | −0.247*** | −0.234*** | −0.249*** | −0.255*** | −0.263*** | −0.255*** | −0.250*** |
(0.020) | (0.019) | (0.021) | (0.023) | (0.023) | (0.025) | (0.024) | |
Age 60–74 | −0.247*** | −0.237*** | −0.271*** | −0.256*** | −0.286*** | −0.252*** | −0.251*** |
(0.021) | (0.020) | (0.024) | (0.024) | (0.025) | (0.026) | (0.025) | |
Age 75+ | −0.262*** | −0.243*** | −0.295*** | −0.266*** | −0.315*** | −0.263*** | −0.254*** |
(0.028) | (0.027) | (0.032) | (0.031) | (0.032) | (0.033) | (0.031) | |
Wave FE | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
Constant | 2.592*** | 2.794*** | 2.698*** | 2.710*** | 2.771*** | 2.610*** | 2.778*** |
(0.026) | (0.028) | (0.034) | (0.028) | (0.030) | (0.032) | (0.041) | |
Observations | 66,932 | 75,870 | 63,748 | 59,429 | 59,427 | 54,302 | 56,909 |
Number of individuals | 13,235 | 15,177 | 12,733 | 11,949 | 11,949 | 10,969 | 11,520 |
Note: Robust standard errors in parentheses; ***p < 0.01, **p < 0.05, *p < 0.1.
The BE component of models 1 and 2 indicates that individuals with higher education, upper-class occupations, and not employed have, on average, higher levels of support for European unification than otherwise. Yet mean individual income is not consistently positively related to the outcome in models 1–2. These results are in line with previous work based on cross-sectional data, which documented a strong positive association between objective socio-economic status and pro-European positions (Foster and Frieden 2021).7 Also, interestingly, gender does not shape support for European unification, while age proves especially influential as younger individuals are significantly more supportive than other age groups. Once we include the variable subjective financial insecurity, the between-individual associations of education and occupational status remain strongly significant.
More importantly, in models 3–7 the between-respondent relationship between subjective financial insecurity and support for further European unification is strongly negative and significant. Support for further European unification is lower for respondents who feel financially insecure than for those who do not. The coefficients of occupational status decline moderately after including subjective financial insecurity in the models, suggesting that the latter do partially mediate the influence of the former. However, this does not apply to the income and education variables. The coefficients of education (models 6–7) and income (models 6–7) do not decline substantially after including the variable subjective financial insecurity, suggesting that financial insecurity does not mediate the influence of income. In sum, regarding the BE component, average subjective financial insecurity is robustly negatively associated with the outcome and this relationship does not clearly mediate the association of income, although it partially mediates the association of occupational status with the outcome.
Predicted Value in European integration at Different Levels in Subjective financial insecurity mean and Subjective financial insecurity demean.
Predicted Value in European integration at Different Levels in Subjective financial insecurity mean and Subjective financial insecurity demean.
Results regarding FE components display fewer significant associations. Gaining access to university education, becoming employed, changing one's occupational class or increasing one's income are not consistently and significantly related to changes in European unification. This is very relevant evidence because it indicates that the robust cross-sectional associations between, on the one hand, education and occupational class and, on the other hand, pro-EU attitudes documented by prior work does not hold when we assess the relationship in longitudinal terms. Either (a) differences between respondents in education and occupational status are related to pro-EU attitudes due to the association with a third set of time-invariant factors (e.g. personality traits or early socialisation) or (b) the significant association of average objective socio-economic conditions and the outcome occurs with a substantial temporal lag. We return to this issue in the Discussion.
More importantly, in the FE component subjective financial insecurity is negative and strongly significant in all models 3–7. Even after controlling for 10 other time-changing (demeaned) variables, subjective financial insecurity retains a significant association with the outcome (model 7). Respondents who over time observed larger increases in their subjective financial insecurity display significantly larger over time decreases in their support for European unification. A comparison of the between- and within-individual effects (model 7) of subjective financial insecurity indicates that large difference in the coefficients. As depicted in the right-hand subplot of Figure 3, a standard deviation increase in subjective financial insecurity demeaned is associated with a .015 point decrease in European unification.
Also, interestingly, the FE coefficient of individual income does not decline after controlling for subjective financial insecurity indicating that the latter does not mediate the role of the former.8 Using hybrid models, Table 1, therefore, indicates robust between- and within-individual associations of perceived financial insecurity with support for further European unification.
We conduct three sensitivity analyses. First, since 28.89% of all individual-wave cases are excluded in model 7 due to missing values in independent variables, the results may be sensitive to the large number of missing values. We therefore replicate the models in Table 1 imputing missing values in all independent variables.9 These additional results are available on Table A2. In these models with imputed values, the FE and BE coefficients of subjective financial insecurity remain having robust between- and within-individual negative associations with the outcome. The evidence is, thus, strongly consistent with the core expectation of this study. Second, using a one-wave lag in all independent variables, the FE and BE coefficients of subjective financial insecurity decline moderately although they are still significant at the standard 5% level (Table A3). Third, the subjective financial insecurity variable could be capturing the influence of respondents’ prospective economic situation (Hooghe and Marks 2004). We thus replicate model 7 in Table 1, including an indicator of prospective personal economic situation. This additional evidence (Table A4) indicates that the BE and FE associations of subjective financial insecurity with the outcome occur independently from prospective personal perceptions.
Discussion
This manuscript contributes to the literature in EU attitudes by highlighting a factor still unexplored by the utilitarian approach: individual current subjective financial situation. Using hybrid models and panel data for the Netherlands from the LISS panel, it explores whether – independently from objective socio-economic situations – feeling financial insecure is cross-sectionally and longitudinally related to support for European unification. The analysis yields three main findings.
First, when considering differences across respondents in their stable characteristics, objective socio-economic position proves consistently related to EU attitudes. Respondents who have at least some university education or generally work as higher-grade professionals display significantly larger support for European unification than respondents with secondary education or less and manual workers, respectively. These results are in line with extensive cross-sectional research (Gabel 1998; Gabel and Palmer 1995; Gabel and Whitten 1997; Foster and Frieden 2021).
Second, and more interestingly, respondents’ occupational status and attaining university prove lacking a longitudinal association with the outcome. Put differently, respondents who over time became higher-grade professionals or underwent university education do not display significantly larger increases in support for European unification. This constitutes an important finding, because almost all previous work on EU attitudes relies on cross-sectional data which generates results potentially biased due to unobserved heterogeneity. The combined findings of across-respondents significant and within-respondent non-significant associations between, on the one hand, European unification and, on the other hand, education and certain occupations, indeed suggest that these across-respondent associations may capture the influence of one or several unobserved time-constant factors correlated both with the outcome and these objective socio-economic conditions. Following the lead of Kuhn et al. (2021), further research should explore what background variables (e.g. parental education, early childhood experiences, personality traits or even physiological conditions) may constitute those unobserved major factors.
Third, and most importantly, both in across-case and within-case terms, feeling financially insecure has negative and significant associations with support for further European unification. People generally feeling financially insecure display significantly less support for further European unification than people who generally feel financially secure. People who over time increased their feeling of financial insecurity, moreover, display significantly less support for further European unification than people who over time did not feel more financially insecure. The relevance of subjective financial situation becomes especially notable when we consider that it is the only considered factor with robust across-case and within-case associations with support for European unification.
The robust link between subjective financial insecurity and support for European integration is consistent with our theoretical model regarding the conditions that facilitate support for policy and political change. Indeed, prospect theory (Kahneman and Tversky 1979) and diminishing marginal utility theory (Gossen 1983 [1854]) are useful theories for understanding decision making not only regarding personal economic-related situations but also socio-political matters. Following the prospect theory, we argue that attitudinal questions on socio-political changes are interpreted by respondents in terms of gains and losses implied by such a socio-political change. In addition, based on diminishing marginal utility function theory, we argue that subjective financial insecurity implies a higher tendency of risk aversion, which translates into preferences for the status quo and opposition to (socio-political) changes.
Based on this theoretical model, future research could explore how generalised is the financial insecurity-EU attitudes link and whether financial insecurity also undermines support for other policy changes like the signing of new international treaties, reforms in certain social policy programmes or regulations on immigration. More generally, the results of this study suggest the potential of cointegrating principles of the utilitarian approach to pro-European attitudes with social-psychological models of personal perceptions to advance our understanding of the forces shaping preferences towards European integration.
Furthermore, our theoretical mechanism is likely to be restricted to EU members with a relatively flourishing economy, as in the case of the Netherlands. This is because in contexts of harsh economic downturns where citizens perceive that their economic situation cannot further worsen, they are likely to exhibit risk-seeking behaviour. In such situations, taking risks only implies upside risks for citizens who are totally deprived and have nothing left to lose. Moreover, the principle of diminishing sensitivity applies to the evaluation of wealth change (Kahneman 2011: 285): a unitary loss of wealth is less painful when one has lost a large amount of wealth than when one has not lost much of it. Future comparative research would be necessary to test the extent to which our theoretical mechanism based on risk aversion also applies to citizens from countries facing a harsh economic downturn or whether citizens from such countries instead favour risk seeking in their attitudes toward European unification.
These results notwithstanding, this study has limitations pertaining to construct validity and internal validity. While our dependent variable measures attitudes toward social change correctly (with one item pole specifically referring to further European unification), it provides a less precise measurement of preferences for the status quo. We interpret the other item pole, ‘European unification has gone too far’, as being in favour of the status quo. Indeed, the vast majority of Dutch citizens are in favour of their country's EU membership. It is therefore unlikely that respondents answering ‘European unification has gone too far’ wish for European disintegration, as only 12% of Dutch respondents from the 2018 ESS data answered that they would vote for their country to leave the EU in the event of a referendum. However, we cannot exclude that this statement could also be understood as support for European disintegration. Future research should assess the application of our theoretical mechanism to attitudinal items that explicitly measure both preferences for socio-political change and preferences for the status quo. Moreover, due to data limitations in the LISS, the models were not able to include control variables on certain social values (e.g. universalism) that may be related to the outcome and the potentially subjective financial situation. Finally, our analysis did not test the theorised mechanism that financial insecurity fosters general risk and change aversion among the considered population. Despite these shortcomings, we show that the application of an interdisciplinary perspective on classical research questions in political sociology – such as the association between subjective financial insecurity and EU attitudes – is a fruitful research avenue to advance academic debate both empirically and theoretically.
Acknowledgements
We are grateful to Stephan Dochow-Sondershaus, Carl-Friedrich Elmer, Antonio Jaime-Castillo, Giuseppe Pietrantuono and participants in the 2023 Annual Conference of the World Association for Public Opinion Research (WAPOR) for their feedback regarding previous versions of this study. We are also grateful to the two anonymous referees for their insightful comments, which have helped to improve the manuscript.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Footnotes
We hereafter use ‘European unification’ and ‘European integration’ as synonyms.
Although support for further European integration and EU support are not synonymous and sometimes even have different determinants (e.g. van den Hoogen et al.2022), in this section we focus on the latter as it has been the main theme of the debate between utilitarian and non-utilitarian approaches to EU-related attitudes.
‘How does the financial situation of your household now compare with what it was 12 months ago?’ for the retrospective version and ‘What are your expectations for the next 12 months: will the next 12 months be better, worse or the same, when it comes to the financial situation of your household?’ for the prospective version.
In this paper we make use of data of the LISS (Longitudinal Internet studies for the Social Sciences) panel administered by Centerdata (Tilburg University, The Netherlands).
Replication files for all models are available at https://figshare.com/s/96b93cb0687759e58acb.
The average trends in these two dimensions may differ from the patterns found in the multivariate analysis as the latter control for compositional characteristics of the population.
Van den Hoogen et al. (2022) conducted a cross-sectional study of the configuration of pro-European attitudes in the Netherlands and reported non-significant associations between income and the two considered outcomes: being an instrumental pragmatist (supportive of EU membership for strict instrumental reasons) or a non-federalist (supportive of EU membership but not supportive of EU enlargement).
Table 1 shows that the professional supervisor variable has a negative and significant effect on the FE component and a positive and significant effect on the BE component. Few people change their social class during their professional lives. Hence one possible explanation for these contradictory findings is that this FE estimate may be influenced by measurement error and the limited statistical power of the test.
Imputed values were estimated with the program Blimp (Enders 2022; Keller and Enders 2022). Blimp uses a Bayesian fully conditional specification and latent response variables to treat categorical variables. We estimate 5 imputed datasets (with 20,000 burn-in iterations and 5 final iterations in 5 chains) and the values were clustered by the respondent's ID to reflect the two-levels structure of the data. After the imputation, the models were estimated with Stata.
References
Céline Teney is Full Professor of Macrosociology at the Institute of Sociology, Freie Universität Berlin. Her research interests encompass political sociology, European integration, and migration and integration.
Juan J. Fernández is Associate Professor of Sociology at the Department of Social Sciences, University Carlos III of Madrid. His research focuses on European integration, gender & politics and social policy.
Author notes
Supplemental data for this article can be accessed online at https://doi.org/10.1080/14616696.2023.2271982
Edited by Yannis Tsirbas
Replication files are available at https://figshare.com/s/96b93cb0687759e58acb.