The International Inequalities Institute (III) was launched in 2015 and will provide co-ordination and strategic leadership on inequality research across LSE.

Dec 10, 2015

Image Credit: https://pixabay.com/p-825602 CC0 Public Domain

European Identity and Redistributive Preferences

by Joan Costa-Font & Frank Cowell

LSE International Inequalities Institute Working Paper 3

Introduction

We still know relatively little about what shapes preferences for redistribution. The standard political economy theory of redistribution (Meltzer and Richard 1981, Romer 1975), linking the expansion of gross income inequality to increased demands for redistribution, has not been consistently validated in practice (Georgiadis and Manning 2012, Gouveia and Masia 1998). Other explanations focus on the presence of biased perceptions of redistribution,1 the expectations of offspring social mobility (Banabou and Ok 2001), the influence of ethnic fractionalisation (Alesina et al. 2001) or the role of genetics.

Here we focus on the influence of social identity on preferences for redistribution, that is, the presence of common reference points (prescriptions) acting as social norms (Akerlof and Kranton 2000) which influence behaviour within the social group; the identity mechanism then confers some sense of social solidarity on the members of the group. If identity plays this role and solidarity is determined within the context of the nation, a move from that setting to another will affect people’s preferences for redistribution. However, it may be that the development of a European identity affects the extent of solidarity and individual experiences within a wider community. The move from the default of national solidarity expression may be pro-redistributive.

Identity might contribute to the development of cognitive biases insofar as a person’s reference group is not the whole population but that of his group, or his country. People in relatively rich countries may perceive themselves as being poorer than they really are, not so much because of an information bias, but because their reference point is based on the social group they identify with, and not necessarily the whole population. So an important question for empirical purposes is that of identifying whether an exogenous change in reference point, such as the relevant spatial dimension of identity (Europe v national), exerts an influence on distributional preferences. Ignoring identity and relying on an individualistic model of self-interested demand for redistribution will underestimate the benefits of redistribution itself.3 Processes of regional integration offer a unique natural experiment to examine such a question in the field.

Social identities shape individuals’ preferences by defining a “sense of belonging” to a club good that appears in people’s utility functions (Akerlof 1997). Accordingly, an individual suffers disutility from deviating from his or her category norms, which induces behavior and influences preferences has wide- ranging implications for welfare economics that conforms to those norms (Akerlof and Kranton 2000). The extent to which identity is feeling part of a group triggers more positive social evaluation towards the group (Cremer and Vugt 1999, Gaertner et al. 1989). The substitution of a national currency by a common currency (the euro) may have triggered some salience to the European project resulting in a greater weight of the European component of people’s identity; at the same time identity may remain highly valued as a position good, especially for European countries that did not enjoy the club status with their own national attachments. Identification with a polity largely depends on the status of the groups compared to the alternative possible status (Roccas 2003).

Our focus in this paper is on individuals’ redistributive preferences, and we claim that the development of a European identity resulting from institutional reforms such as the introduction of a common currency provides quasi- experimental evidence to examine it. Europe is the ideal setting to study changes in identity, given that the progressive integration process exerts effects on welfare-state institutions, which in turn can influence the existing welfare institutions by affecting people’s redistributive preferences. The unique experience of the setting up of a single currency exerted a non-neutral effect on European attachment as measured by confidence in the EU, and reduced national pride. This result is consistent with other findings that indicate that European identity explains satisfaction with democracy (Hobolt 2012). Similarly, Risse (2010) finds that people who identify themselves as European are more likely to identify with the values of tolerance and democracy.

The intuition behind the paper is that, when identity is defined by a “broader other,” people are more likely to express a preference for true redistribution (redistribution in small communities might be partially explained by exchange motivations instead). Becoming part of the Eurozone club in a setting where redistribution is primarily undertaken by national welfare states should not change the individual’s expectation of benefiting from redistribution, and should primarily affect the rescaling of people’s spatial identity. However, there is a potential reverse causality that should be taken into account: a revival of anti- European nationalism (which we proxy here by national pride) is underpinned by anti-immigration attitudes; also there might be a problem with omitted- variable bias if some confounding variables correlated with identity. In this paper we propose an instrumental-variable strategy that takes advantage of the adoption of a common currency (a largely exogenous decision to individuals themselves). We focus on countries that adopted the common currency only after its inception so that we can observe a period before and after being part of the common currency. We use other instruments to measure the robustness of our results. Finally, we use different subsamples to ascertain whether the results hold beyond the specific country sample.

Our paper brings together different strands of the literature. We incorporate some findings from the European politics literature suggesting that some aspects of national identity are substituted for, with the expansion of European identity. This not only changes people’s affiliations but also preferences towards equality. In a more competitive setting, such as that of a wider European Union, wide inequalities are likely to emerge and so the role of redistributive mechanisms becomes more important. We contribute to the literature on preference for redistribution and the limitations of the Meltzer and Richard approach. Third the paper contributes to the role of identity in influencing economic behaviour (Akerlof and Kranton 2000, 2005). If changes in institutions affect people’s preferences by changing their identification and collective affiliations then policy needs to be focused more strongly on such indirect effects. Finally, this paper extends the findings of Luttmer (2001), suggesting that preferences for redistribution change with the share of the poor in a region, as Eurozone enlargement to central and Eastern Europe might have exerted an impact on preferences for redistribution. However, we argue that the mechanism for such an effect is channeled through identity.

Section 2 provides the background to the analysis of this paper. Section 3 describes the data and methods, section 4 presents results and the paper concludes with section 5.

Background

There are two important branches of the economics literature that connect to the approach that we use in this paper: the literature on redistributive preferences and the literature on the economics of identity.

2.1 Preferences for redistribution

Economic approaches to redistribution such as Meltzer and Richard (1981) typically assume that people’s position in society determines their preferences and often disregard how people’s social groupings influence preferences. But groupings are important: for example, ethnically diverse societies exhibit less class conflict or, if they do, it is more rare as ethnicity or identity add additional dimension to the political spectrum away from purely economic or redistributive questions (Lee and Roemer 2006).

In the last twenty years we have seen an increasing interest in examining how multiculturalism and diversity influence preferences for redistribution. Alesina et al. (2001) find that ethnic fractionalisation exerts an influence on redistributive preferences in the context of the United States where the default is not a consolidated welfare state as in Europe. Luttmer (2001) finds a negative relationship between diversity and preferences for redistribution: people’s preferences for redistribution are interdependent in the sense that preference is influenced by the characteristics of other individuals around them. People appear to be more likely to redistribute to the groups they identify with, be that identification based on ethnicity, religious group, social class, region or something else.

Preferences for redistribution have been found to be related to voting behaviour and political ideology, to people’s own self-interest, to their evaluation of inequalities, and to their perceptions of the “leaky bucket”, the efficiency of the transfer mechanism (Krawczyk 2010).

Furthermore, redistributive preferences may reflect cultural differences (Luttmer and Singhal 2011) and political differences. But these differences are not exogenous or immutable and may be associated with the phenomenon of identity.

2.2 Identity

“Identity” refers to mechanisms through which individuals become attached to each other by creating a sense of belonging (Tajfel 1978). Akerlof and Kranton (2000) consider identity as an externality on people’s actions triggered by the presence of common social norms: these are common reference points that can shift over time. Collective identities are the expressions of different cultures which can be an important source of preference endogeneity (Bowles 1998) and a recent survey suggests that they can explain individuals’ solidarity attitudes (Costa-Font and Cowell 2015). People may alter their behaviour to conform to other people’s expectations and social norms (Asch 1951) beyond their narrow personal self-interest.10

Social identity can have inward effects on the person, and an outward effect on the group (Mayer and Palmowski 2004). Clearly a person may be associated with multiple groups and, as a result, reveal multiple identities — for example regional and European identities. Some identities attributed to a person may conflict with each other and even become “oppositional” (Battu and Zenou 2010), but others may not. Identities have been seen as a multidimensional social categorization that can be primed by certain circumstances or events. Easton (1975) distinguishes instrumental and affective support for political institutions. If an institution is perceived as being instrumentally beneficial, the attachment to that institution would be expected to increase. Inglehart and Raabier (1978) have put forward the theory of cognitive mobilization whereby education exerts an effect on individuals’ cosmopolitan identity.

Consider the connection with redistributive preferences discussed in section 2.1. National or social identity can act as a “social tie,” which in turn operates in enhancing support for the welfare state (Costa-Font and Cowell 2015).11 Redistribution is one of the central features of welfare states: maintenance of redistributive institutions largely depends on individual support for taxing higher incomes more heavily and targeting expenditures to social need. Since such activities are typically associated with nations, the question arises whether support for redistributive institutions and programs varies with the rescaling of individuals’ identities to both supranational and subnational bodies.

Clearly this is of particular interest with reference to a supranational organisation such as the European Union.

2.3 European identity

In principle European identity could play a role similar to that of American identity, uniting people by transcending borders, and especially racial divisions (Transue 2007). A superordinate identity eliminates the effects of parochialism, country nationalism and group identity. The “European project” certainly raises interesting questions in connection with the mechanisms of redistribution and perceptions of identity. With European integration, the efficient level of redistribution scales up to the European rather than the country level (Cassela and Fray 1992) and is likely to change the strength of people’s attachments to state sovereignty as the institutions in member countries become locked into this emerging structure (Eichengreen 2008).

The rise of a European common identity acts on people’s attitudes as a pro- redistributive force that confronts the existence of own-nationality bias (Lowes et al. 2015). This is, perhaps, to be expected as spatial identities are potentially rescaled from solely national to the supranational, European, level. However, within this structure there is a variety of identities — national, regional, European — and we know little about the relations between these identities, whether they are complementary, substitute or independent. So it is not clear a priori whether the priming of an identity (as mentioned in section 2.2) would exert an external effect on others.

However, among the variety of priming events that might be considered relevant, one of the most important would be the setting up of a common currency. The introduction of a single currency encompasses the reduction of one of the most important old symbols of national identity; so one should expect it to exert an influence on people’s identification: the euro exemplifies the strategy of burning one’s boats. For many countries joining the euro club has meant a way to improving their status worldwide, and hence it implied positive social externality. Support for the euro has remained stable, even through the recent crisis; but whether such (largely exogenous) externality leads to stronger preference for redistribution is an empirical question.

2.4 Our approach

In this paper we take advantage of an institutional reform, the adoption of the single currency, which we argue has had a symbolic effect on priming European identity. The introduction of the euro and its effects when the national currencies were effectively replaced would be expected to have had an effect on attitudes and preferences. We can test whether that effect was stronger for countries that joined the euro initially than for the rest.

But capturing identity empirically is not a simple task. Most studies rely on survey questions which identify some component of a “latent European identity.” Some evidence finds that the stronger is the feeling of national identity, the weaker is support for the European Union (Carey 2002). So here we use both national pride and confidence in the European Union to proxy the underlying European identity. Our identification strategy hinges on taking advantage of cross-country variation in preferences and collective identification (social identity) over time. A key challenge is to control for potential omitted variables. Indeed, cohort effects are important because individuals of the same cohort share similar experiences and observable similar constraints.

Given that our results are affected by a number of potential individual characteristics, we look at the presence of heterogeneous effects and subsample analysis and robustness checks such as examining the role of additional instruments (such as years of citizenship education to instrument European Union confidence and medals in the Olympic games to instrument national pride, as well as peer effects).

Data and Methods

Our primary dataset on preference for redistribution and identity is the European sample of the World Values Survey, also known as European Values Survey for the period 1981–2014. The dataset provides with a series of repeated cross sections observations on the ideas, beliefs, preferences, attitudes, values and opinions of citizens all over Europe. We have employed records of individual specific redistributive preferences, alongside rich measures of spatial or geographic identity that are recorded in the European Values Survey. Specifically, we use self-reported measures of individual’s preference for redistribution, which have been validated in previous studies.

We focus on a set of countries that joined the European Union after 2004. Not all countries are covered in each survey wave, but the years range from 1981 to 2014 (for details see the summary statistics in the Appendix). Overall we are left with a fairly large sample of 27,376 respondents. There are several advantages of using such a sample. First, it allows one to identify the effect of joining the common currency (adopting the euro), in contrast to using the total sample of European Union countries; the founding countries of the euro substituted the currency almost at the same time and hence there is not enough variability to exploit. The second advantage of using a sample of those that joined the euro after 2004 is that there is likely to be an attraction for joining the euro club (“institutional arbitrage” in the spirit of Guiso et al. 2014) which plays out in terms of boosting European identity and hence “widening the spatial identity beyond the national reach.” Finally the introduction of the common currency was an unexpected effect within the time frame of the survey questions (4–6 years), and so it is unlikely that anticipation effects (on the final success of an economy in joining a common currency) could threaten the identification of the effect on spatial identity.

As noted, the redistributive preference question is extensively used in the literature, and refers to a general question about redistribution without specifying the level of authority responsible to make incomes more equal. This way, it can be argued to be institutionally neutral. By contrast, national pride relates to restricted loyalties to national groups which depend on the perceptions of status of national communities. Hence, in this paper we hypothesize that becoming fully part of a larger community (for example by joining the euro area) would be expected to weaken the effect of national pride. Finally, we use confidence in the European Union. This is a different question from trust in the working of European institutions: it captures in a multi-question format the perceptions of individuals in post-2004 European countries of their degree of attachment to the European Union. This question has been found to correlate well with other attitudes towards Europe in the sample.

Tables A.1 and A.2 in the Appendix show the samples and countries included in the dataset alongside the main sample characteristics such as the average age, the percentage of women, education attainment, family characteristics and size and political affiliation. The sample size of each country is about 1000 respondents. Table A.3 displays the sample size of the survey waves which is larger for 1991- 98 than the rest. Tables A.4 to A.6 show the distribution of the main study variables and Table A.7 the proportion of countries that have adopted the euro in the total sample (13%).

Our identification strategy relies on selecting a sample of countries that have progressively joined the European Union for which we can identify a period before and a period after they adopt the euro as a currency. In doing so, it is important to understand how different this subsample of countries is from the rest of the European Union member states. Figures A.1 to A.6 (in Section A.3 of the Appendix) provide plots of our variable of interest (preferences for redistribution) and identity variables for the subgroup of countries that joined the EU after 2004 and those that joined before. In each case there appears to be little difference between the values in the two subsamples when plotted across interview years, but there appears to be evidence of convergence when we plot over sample waves. We may conclude that examining the subsample of countries is likely to allow us to identify the effect of an identity change in redistributive preferences.

Perhaps the most obvious problem in examining the effect of identity on redistributive preferences is the endogeneity of identity measures, and specifically the possibility of reverse causality whereby identity could be viewed as the effect of the existence of redistributive institutions. Furthermore, there might be unobservable variables that intermediate the association between identity and redistributive preferences. In order to account for the non-random changes in identity, we use an instrumental variable (IV) strategy that exploits the exogenous variation of a key institutional change, namely the introduction of the euro, which does not directly affect redistributive preferences unless it is by changing people’s identification with Europe (the excludability condition). We also test for the so-called monotonicity condition to test whether the introduction of the euro did indeed affect identity in the expected sign and that the effect is strong (relevance condition) which is generally observed by examining the joint significance of first-stage estimates in a 2SLS (Staiger’s condition).

Our IV strategy identifies the local average effects of the impact of identity changes resulting from the introduction of a common currency. In addition, we employ a battery of other instruments to examine whether the sign comparisons and results are equally robust. Finally, we undertake some placebo tests to make sure our results are not spurious.

We have estimated reduced forms of the effect of identity on redistributive preferences. Our identification rests on a combination of cross-sectional, time and cohort variation. In some specifications we run cohort-specific regressions to examine the potential cohort-specific effects. Country and time-specific trends are controlled for, as they could be driving the results. The regression strategy includes a quadratic trend to control for all those macroeconomic factors that are varying and exhibit a trend in time. Other country-specific time factors are expected to be captured by country fixed effects.

The total number of observations is 27,376. Our main dependent variable refers to redistributive preferences measured as before. Our treatment variable of interest refers to the two variables capturing the effect of identity, namely national pride and confidence in the EU. We include a long list of controls including demographics, income and socioeconomic, household size and employment status and we identify whether individual are immigrants to the country. The omitted categories in the regressions are male, elementary or lower education, all other marital statuses, no children, all other employment statuses and no immigrant status.

Results

4.1 Preliminary Evidence

Figures 1 and 2 provide data on the cohort and time trends on preference for redistribution in the sample of countries examined in this study. The cohort trend indicates that those individuals over 55 are more likely to support redistribution. Importantly, redistributive preferences have progressively become more salient in people’s attitudes in recent survey waves. This effect is not just an artefact of the most recent wave.

Figures 3 and 4 examine similar trends in national pride indicating an age component to it, which is in line with the hypothesis of Europeanisation as reducing national pride: Europeanisation may have reduced national attachment. Some research finds time trend identification with Europe in EU countries (Fligstein et al. 2012), but identification appears to be largely dependent on the economic performance of Europe, particularly unemployment (Guiso et al. 2014). Some recent evidence finds that Eastern European countries exhibited a comparable or even higher identification with Europe which is in large part explained by the large minority groups in many of those European member states.

Figures 5 and 6 examine cohort and time trends on EU confidence. Measuring the importance of confidence in the EU is important, because being European can mean different things across countries whilst confidence with the EU is a commonly accepted construct. Interestingly cohort trends show that younger cohorts are more likely to identify with Europe, exactly the opposite trend to that of national pride. In contrast, we find that time trends suggest a slight reduction in EU confidence. This result is consistent with the idea advanced by Fligstein et al. (2012) that European identity is a class-based phenomenon directly linked to the transnational mobility benefits of the common EU market. However, the rise of European identity might be the effect of educational attainment and increasing cultural interconnection. To disentangle such effect we need additional regression analysis.

Figure 6: Confidence in the EU — time trend

Our baseline specification is the following:

𝑅𝑖 = 𝛾0 + 𝛾1𝐼𝑖𝑡𝑟 + 𝛾2𝑋𝑖𝑡𝑟 + 𝛾3𝑦𝑖𝑡𝑟 + 𝛿𝑟 + 𝜃𝑡 + 𝜀𝑖𝑡𝑟

where 𝑅𝑖𝑡𝑟 refers to the preference-for-redistribution response by an individual 𝑖, interviewed at time 𝑡 and in country 𝑟. The variable 𝑡𝑟 refers to a variable indicating individuals’ European identity, measured as the individual identification with their country (national pride) or Europe (confidence in the European Union). All specifications include a vector of individual characteristic 𝑋𝑖𝑡𝑟 which includes age, gender, schooling, civil status, size of the area of residence;15 𝑦𝑖𝑡𝑟 is included to control for changes in income a la Meltzer and Richard as well as unemployment as potentially driving the results; 𝛿𝑟 refers to country fixed effects to control for common background of individuals residing in each country, 𝜃𝑡 refers to a wave- (time-) specific effect to control for age- specific trends in redistributive preferences and 𝜀𝑖𝑡𝑟 is a random term. where euro refers to a dummy variable taking the value 1 if the country 𝑟 has adopted the common currency at time 𝑡 and a vector of individual characteristics (𝑍𝑖 ). As a rule-of-thumb the F-test of such a first regression should exceed the value 10 for the instrument to be strong enough to meet the relevance condition.

In addition, the excludability condition refers to the absence of a correlation between the error term and the instrument. This condition cannot be tested empirically, but we do address some issues concerning this assumption by testing the effects of alternative instruments that follow a similar rationale, and examining different of suggestive evidence on its plausibility.

All regressions have been estimated using OLS to ease the interpretation of coefficients, and robustness checks include specifications using alternative techniques dealing with the categorical nature of the data, including a binarised identity and redistributive preference variable so as to interpret the dependent variable as a probability. Robust standard errors are clustered at the country level and descriptive statistics are provided in Tables A.1 and A.2 of the Appendix

In order to estimate the 2SLS equation we employ a first-stage equation capturing the impact of the proposed instrument on the identity questions:

𝐼𝑖𝑡𝑟 = 𝛼0 + 𝛼1euro𝑡𝑟 + 𝛼2𝑋𝑖 + 𝛼3𝑍𝑖 + 𝜇𝑖𝑡𝑟

Baseline regressions

Table 1 reports the regression results to explain redistributive preferences by change in national pride and EU confidence. These are OLS results that do not take into account all the potential problems of reverse causality and omitted- variable bias. We provide different specifications with different controls and the beta coefficients to interpret the results as the effects of a one-standard- deviation change. The effects of income are as expected, indicated by a negative and significant coefficient.

A one-standard-deviation increase in national pride is found to reduce redistributive preferences by the same magnitude (6%) as a one-standard- deviation increase in income. As expected, younger individuals are more likely to support redistribution. A one-standard-deviation increase in the population in tertiary education reduces preferences for redistribution by 13%. So the effect of education appears to be twice the size of the effect of income. This is an important result, given the focus in the literature on the Meltzer and Richard type of approach. Indeed, this coefficient is important as it can explain why the income ranking of the median voter would not exert the predicted influence.

Initially, Columns (1) and (6) report only the coefficient for national pride and confidence in the EU respectively. The regressions contained in columns (2) and (7) report the effect after the introduction of a quadratic time trend to account for potential underlying trends that could be driving the coefficients. Columns (3) and (8) contain the effect of adding additional controls for income and employment, and finally Columns (4) and (9) contain the effect resulting from the additional control for town size. All regressions contain country fixed effects. Overall, the coefficients for national pride exhibit little variation in its size. Importantly, unemployment which is a variable that is found to explain European identification and trust in European institutions (Guiso et al. 2014) does not exert an influence on preferences for redistribution. From all the covariates reported here the most important determinant of redistributive preferences appears to be education.

Table 2 reports a 2SLS instrumental variable (IV) analysis that controls for reverse causality and unobservables. Now confidence in the EU switches its coefficient to being positive and significant and exhibits the same coefficient size but with the opposite sign to that of national pride (both are statistically significant). Again, the effect’s size indicates that one standard deviation of national pride reduces preferences for redistribution by an amount similar in size (but opposite in sign) to that of an increase in confidence in the EU. Tests all reject the hypothesis of exogeneity, and the F-test of the first stage regression both exceed 10 suggesting that instruments are not weak. Furthermore, the instrument (join the euro) exhibits the expected sign. As expected, looking at the first stage regression we find that women, older individuals, married and people without tertiary education, unemployed and from smaller towns are more likely to exhibit national pride.

As expected the coefficient for tertiary education remains strongly significant and negatively associated with redistributive preferences. Interestingly, income and tertiary education have an opposite effect on national pride. Again all regressions control for country fixed effects. Some important differences across specifications when national pride and confidence in the EU are estimated refer to the effect of age, which only the effect of age squared turns out to be significant when explaining national pride. In contrast, age exhibits a reverse nonlinear effect in explaining confidence in the EU and redistributive preferences. The pattern of coefficients in Table 0 remains in Table 1 when redistributive preferences are binarised. The interpretation is that the probability of supporting redistribution declines by 2.5% if individuals exhibit national pride, and increases by the same magnitude if they have confidence in the EU. Tables 6.4 and 11 in the Appendix show that this effect is primarily driven by older-age and low- income individuals. On this basis we can conclude that the the instrumental strategy we employ appears to provide consistent and robust results, given that alternative instruments provide similar results.

Robustness checks

Tables 4 and 5 report the regression estimates using different instruments. In Table 4 we use alternative instruments of EU confidence such as a dummy indicating whether the country has joined the EU, average confidence in the EU (of other countries in the sample) to predict confidence and average pride (of other countries in the sample) to predict pride. The exogeneity and F tests coefficients all suggest they are not weak instruments and the coefficients are all exhibiting the same sign although vary in terms of the impact. An expansion of EU confidence exerts a larger positive effect now than the negative effects of national pride. Then in Table 3 we examine other instruments such as the duration of citizenship education, which turns out to be a weak instrument, and the number of medals in the Olympics games which was a strong instrument for national pride and results in the IV analysis showing a significant and comparable coefficient as in other regressions estimates.

Table 6 shows that the estimates predicting redistributive preferences are robust to the inclusion of political preferences. As before, the inclusion of different instruments suggests the same reverse-sign effect in the coefficient for EU confidence which now turns positive when including the political control in the estimates with the alternative instruments used in Table 3. As expected, the more to the right an individual positions himself the less likely he is to support redistribution. All estimates exhibit an important nonlinear trend which captures among other effects that of time-varying macroeconomic and contextual effects.

Conclusions

The rescaling of spatial identity in the context of European integration processes is potentially an important effect underpinning changes in redistributive preferences. This paper has provided evidence to support that claim.

However, unlike previous research that mainly stresses the importance of group identity, we have argued that the scale of geographical identification matters. Specifically, the development of a European identity appears to weaken national pride. But how important is this new collective identity in shifting preferences for redistribution? The answer to this question could help to explain the limited evidence of median-voter explanations for redistributive preferences. Indeed, in addition to the information problems people face in identifying their position in the income distribution, when making redistributive judgments, they appear to react to change in the spatial scale of reference. So the expansion of the European integration process together with the introduction of a common currency exerts a non-neutral influence in the context that influences redistributive preferences.

This paper has specifically examined the effects of spatial identity in those countries that joined the EU after 2004 where we can identify the introduction of reforms expanding European integration and the effect of joining the single currency. Unlike the case of the founding countries of the Eurozone we can identify the effect of joining a European club more precisely using the recent- joiners sample. We find a positive (negative) impact of European identity (national identity) on preferences for redistribution. The effect of identity is comparable in size to the effect of income and is only exceeded by the effect of tertiary education.

These results indicate that institutional changes involving symbolic features that define one’s identity, in this case the currency, can exert an impact on people’s attachments, and more specifically can underpin the formation of a person’s identity. We find robust evidence suggesting that the introduction of the euro as a common currency in countries that joined the EU after 2004 increased people’s confidence in the European Union, and reduced the importance of national pride. Similarly, this evidence is replicated when other potential identity instruments are examined. Using an instrumental variable strategy, we find that the exogenous change in European identity resulting from a common currency increases people’s preference for redistribution. These results are consistent with the previous identity literature.

Among the policy implications of these findings it appears that there are important positive knock-on effects on redistributive preferences associated with furthering European integration. The lessening of national pride is more likely to lead to changes in individuals’ reference points which influence the way they form preferences for redistribution.

The International Inequalities Institute (III) based at the London School of Economics and Political Science (LSE) aims to be the world’s leading centre for interdisciplinary research on inequalities and create real impact through policy solutions that tackle the issue. The Institute provides a genuinely interdisciplinary forum unlike any other, bringing together expertise from across the School and drawing on the thinking of experts from every continent across the globe to produce high quality research and innovation in the field of inequalities. We are home to a Masters programme (MSc in Inequalities and Social Science) and Doctoral programme funded by the Leverhulme Trust.

Website: www.lse.ac.uk/III
Twitter: @LSEInequalities

For Appendixes, see pdf version at www.lse.ac.uk/IIIpapers

References

Akerlof G. A. 1997. Social distance and social decisions. Econometrica 65,1005–1027.
Akerlof G. A. and Kranton R. E. 2000. Economics and identity. The Quarterly Journal of Economics 115, 715–753.
Akerlof G. A. and Kranton R. E. 2005. Identity and the economics of organizations. Journal of Economic Perspectives 19, 9–32.
Alesina A., Glaeser E. and Sacerdote B. 2001. Why doesn’t the United States have a European-style welfare state? Brookings Papers on Economic Activity (2), 1–70.
Asch S. E. 1951. Groups, Leadership and Men, Chapter Effects of group pressure upon modification and distortion of judgement. Carnegie Press.
Battu H. and Zenou Y. 2010. Oppositional identities and employment for ethnic minorities: Evidence from England. The Economic Journal 120, 52_71.
Benabou R. and Ok E. 2001. Social mobility and the demand for redistribution: The POUM hypothesis. Quarterly Journal of Economics 116, 447_487.
Bowles S. 1998. Endogenous preferences: The cultural consequences of markets and other economic institutions. Journal of Economic Literature 36, 75_111.
Carey S. 2002. Undivided loyalties: Is national identity an obstacle to European integration? European Union Politics 3, 387_413.
Carlsson F., Johansson-Stenman O. and Khanh Nam P. 2014. Social preferences are stable over long periods of time. Journal of Public Economics 117, 104–114.
Casella A. and Frey B. 1992. Federalism and clubs. towards an economic theory of overlapping political jurisdictions. European Economic Review 36, 635–646.
Charness G., Rigotti L., and Rustichini A. 2007. Individual behavior and group membership. American Economic Review 97, 1340–1352.
Chen Y. and Li S. X. 2009. Group identity and social preferences. The American Economic Review 99, 431–457.
Corneo G. and Grüner H.P. 2002. Individual preferences for political redistribution. Journal of Public Economics 83, 83–107.
Costa-Font J. and Cowell F. A. (2015). Social identity and redistributive preferences. Journal of Economic Surveys 29, 357–374.
Cremer D. D. and Vugt M. V. 1999. Social identification effects in social dilemmas: A transformation of motives. European Journal of Social Psychology 29, 871–893.
Cruces G., Perez-Trugliad R. and Tetaz M. 2013. Biased perceptions of income distribution and preferences for redistribution. Journal of Public Economics 98, 100- 112.
Durante R., Putterman L. and van der Weele J. 2014. Preferences for redistribution and perception of fairness: An experimental study. Journal of the European Economic Association 12, 1059–1086.
Lindqvist, E. and Östling R. 2013. Identity and redistribution. Public Choice 155, 469–491.
Lowes, S., Nunn N., Robinson J., and Weigel J. 2015. Understanding ethnic identity in Africa: Evidence form the implicit association test (iat). Working Paper 20885, NBER.
Luttmer E., Singhal F. and M. 2011. Culture, context and taste for redistribution. American Economic Journal: Economic Policy 3, 157–179.
Luttmer, E. F. P. 2001. Group loyalty and the taste for redistribution. Journal of Political Economy 199, 500–528.
Mayer, F. and Palmowski J. 2004. European identities and the EU — the ties that bind peoples of Europe. Journal of Common Market Studies 42, 573–598.
Meltzer, A. H. and Richard S. F. 1981. A rational theory of the size of government. Journal of Political Economy 89, 914–927.
Norton, M. and Ariely D. 2011. Building better America — one wealth quintile at a time. Perspectives on Psychological Science 6, 9–12.
Risse, T. 2010. A Community of Europeans? Transnational Identities and Public Spheres. Ithaca, NY: Cornell University Press.
Roccas, S. 2003. The effects of status on identification with multiple groups. European Journal of Social Psychology 33, 351–366.
Romer, T. 1975. Individual welfare, majority voting, and the properties of a linear income tax. Journal of Public Economics 4, 163–185.
Shayo, M. 2009. A model of social identity with an application to political economy: Nation, class, and redistribution. American Political Science Review 103, 147–174.
Tajfel, H. 1978. Differentiation between social groups: Studies in the psychology of intergroup relations. Academic Press.
Transue, J. E. 2007. Identity salience, identity acceptance and racial political attitude: American national identity as uniting force. American Journal of Political Science 51, 78–91.
Zakharov, A. and E. Ponarin 2013. Genetic factors and preferences for redistribution. Technical report, Mimeo