Did Decentralisation Affect Citizens’ Perception of the European Union? The Impact during the Height of Decentralisation in Europe

: The aim of this paper is to assess the extent to which different levels of decentralisation across regions of the European Union (EU) affected citizens’ perceptions about European integration over the period 1973–2002. The paper uses Eurobarometer Surveys to explore by means of multinomial logistic regressions whether decentralisation was an important factor behind the varying perceptions about Europe. Two dimensions of decentralization—political and fiscal—are considered in the analysis, alongside several compositional and contextual effects. The results of the analysis show that fiscal decentralisation was fundamental for citizens’ support for European integration, while there is limited evidence that political decentralisation played a similar role. Hence, while fiscal decentralisation may have given prominence to the economic benefits of European integration, political decentralisation was more associated with its economic costs. Taking into account that history matters, this paper raises potentially interesting insights for the design of policies aimed at promoting social cohesion.

European economy, in particular. This will allow us to assess whether there was a potential trade-off between 'localism' and 'Europeanism'. The key hypothesis driving this paper is that, taking into account that the balance between the costs and the benefits of fiscal and political decentralisation differs among individuals, decentralisation affected and determined people's socio-political perceptions about Europe during the period of analysis.
This paper focuses on the period 1973-2002 for the following reasons. First, the European integration process, which took off with the creation of the European Coal and Steel Community in 1952, is characterized by many historical events. During this long period, Europe has faced many internal and external social, economic and political challenges, e.g., from the construction of the Berlin Wall in the summer of 1961 and the financial problems in the European Economic Community in 1965 to, more recently, the shocks related to the 2004 enlargement, the post-2008 global economic and financial crisis, and the 2015 refugee crisis. It is known that during a recession/crisis periods, citizens have been more sceptical about the benefits of the European integration; while during an economic boom, citizens are more likely to support European integration (Jones et al. 2016;Kuhn and Stoeckel 2014;Daniele and Geys 2015;Toshkov and Kortenska 2015). Hence, the historical events, shocks and challenges of Europe have created a high degree of heterogeneity regarding the perception of citizens about the benefits of the European integration and the EU Single Market. This varies not only across European countries, but also over time. Regarding the historical events, the period 1973-2002 could be characterized by a relatively higher degree of homogeneity than the period 2003-2020. During the period 1973-2002, citizens generally had a positive view about the EU. Prior to the two big EU enlargements of 2004 and 2007, which changed the landscape of Europe, and the challenges of the economic, financial and refugee crises, the EU-despite some ups and downs-was widely regarded as a beacon of democracy, prosperity, and political stability.
Second, the period between the early 1970s and the early 2000s was characterised by a rapid drive towards decentralisation across many countries of Europe. This was the period when, on top of the already decentralised Austria and West Germany, Belgium and Spain experienced deep transformations in their territorial organisation. Devolution of powers and resources was also important in Italy, while France and the UK had more cautious and, in the case of the latter, asymmetric processes of decentralisation. It was also a period in which countries in central and Eastern Europe, such as Poland, started to toy with granting more authority and resources to regions (Rodríguez-Pose and Gill 2003;Rodríguez-Pose and Bwire 2004).
Third, this study focuses on a long historical period (30 years) for Europe which has been characterized by the deepening of integration, the enlargement to the South, the Single European Act, the Single Market, the reform of the Common Agricultural Policy, the Treaty on European Union, The Treaty of Amsterdam, and the Treaty of Nice. The period 2003-2020 includes key major events which have drastically shaped European citizens' perception about the EU. According to Tselios and Tomaney (2018), some parts of Eastern Europe that joined the EU in 2004 and in 2007 are striving to become fully European, and considerable gaps remain between 'Western' and 'Eastern' institutions. Recent shocks related to the post-2008 financial crisis as well as to the 2015 refugee crisis and Brexit have raised concerns about future of European integration. Moreover, recent studies show that differences in public support for European integration have increased after enlargement and the economic crisis. Kuhn and Stoeckel (2014) and Daniele and Geys (2015) underline the role of the economic crisis in increasing opposition to European integration. Likewise, Toshkov and Kortenska (2015) have argued that increases in immigration have undermined public support for integration in the EU, while de Vries (2013) uncovered the existence of considerable differences in public support for integration between East and West. Wyplosz (2015) showed that the crisis exposed deep governance issues and construction flaws in the Eurozone, further denting the support for European integration. All these factors, along with the most recent 2020 national lockdown, have radically transformed the nature and expectations of the EU (Tselios and Tomaney 2018).
This paper makes several contributions to the field. First, from a theoretical and empirical perspective, this is, to the best of our knowledge, the first study which attempts to assess whether Second, this study examines the impact of decentralisation on research outcomes from a microeconomic perspective, because the dependent (response) variable refers to the characteristics of the European citizens/respondents, i.e., to their perceptions. This approach differs from the existing empirical studies of the impact of decentralisation, as these studies have tended to focus on research from the macroeconomic perspective, because the dependent (outcome) variable refers to the characteristics of the European countries, such as economic growth and economic disparities. Moreover, these macroeconomic studies often reaching contradictory results (e.g., Canaleta et al. 2004;Davoodi and Zou 1998;Fisman and Gatti 2002;Iimi 2005;Kyriacou and Roca-Sagalés 2011;Rodríguez-Pose and Ezcurra 2010;Rodríguez-Pose and Gill 2004;Tselios et al. 2012). This study, resorting to a microeconomic perspective, covers a large sample-over 1 million individuals and over a long period of 30 years-and can shed some light on the contradictory conclusions stemming from macroeconomic analysis, as the interest on the effects of decentralisation on people's attitudes and behaviour has been negligible.
Third, this study will demonstrate the extent to which changes in fiscal and political decentralisation across European countries have changed European citizens' perception towards the EU. We assume that the more the European citizens feel European, the more likely they are likely to support greater European wellbeing and cohesion. Hence, this paper complements the existing studies which discuss the association between decentralisation and cohesion. Despite some exceptions (e.g., Bjørnskov et al. 2008;Díaz-Serrano and Rodríguez-Pose 2012;Rodríguez-Pose and Maslauskaite 2012), it has been often assumed that decentralization to lower government tiers can be detrimental for overall economic cohesion and beneficial for local economic cohesion. However, our knowledge about the effects of political and fiscal devolution and regional autonomy on social cohesion and people's perceptions about Europe remains limited and patchy.
Fourth, from a policy point of view, understanding whether and to what extent fiscal and political decentralisation affected citizens' perception of the EU over 1973-2002 and taking into account that history matters, the findings of the paper could be used by decisionmakers to improve the design of European policies, as the quest for higher fiscal and political decentralisation, and possibly the quest for regional autonomy, could be seen as an alternative for social cohesion policies.
The paper proceeds as follows. Because of the complexity of the issue and the relative absence of a coherent theoretical background on which to base the empirical investigation, the next section discusses the pros and the cons of fiscal and political decentralisation which shape people's attitudes towards the EU. Section 3 presents the data, including the descriptive statistics. Section 4 is devoted to the model and method and to the discussion of the econometric results. The final section synthesises the empirical results, draws some implication for policy and discusses directions for future research.

The Benefits and Costs of Decentralisation under the Umbrella of the European Integration
The EU has for a long time recognised the developmental role of local authorities and generally supported decentralisation as an instrument to promote the European ideal and achieve better economic outcomes. In this respect, the European Commission (2013) stresses the importance of local authorities in partner countries for enhanced governance and more effective human and economic development outcomes.
Fiscal and political decentralisation have therefore often been regarded as a key determinant for the advancement of equity and efficiency at the local, regional, national and European level. 1 Yet, 1 Fiscal decentralisation refers to the degree to which central governments transfer fiscal autonomy to subnational government entities, while political decentralisation refers to the power of subnational governments to undertake the political functions of governance . the levels of decentralisation across countries and regions of Europe remain extremely uneven. While some countries have taken limited steps towards decentralisation, others have transferred a substantial amount of fiscal and political power, and their regions and cities enjoy a considerable degree of regional autonomy. Over the last four decades, the trend towards decentralisation has been widespread across Europe (Rodríguez-Pose and Gill 2003), yet the process has remained widely uneven (Hooghe et al. 2016). For instance, the European countries with the highest degree of decentralisation (i.e., Austria, Belgium, Germany, Italy, and Spain) had or have introduced institutional, political, law-making and constitutional reforms that make their regions important actors in many areas of public policy. This high degree of decentralisation in an economy is generally justified by the principle of efficient allocation of resources and the equal distribution of wealth (Musgrave 1959). Other European countries have remained, in contrast, highly centralised. Regions in the Baltic states, Greece, Ireland or Portugal-with the exceptions of the Azores and Madeirahave very little autonomy and are often nothing more than an administrative division.
According to the literature in economics and political science, fiscal and political decentralisation can have both a positive and a negative impact on the economic performance of a country and on the wellbeing of its citizens.
On the one hand, the transfer of authority and resources to subnational tiers of government can lead to improvements in the efficiency of an economy (Tiebout 1956;Oates 1972). Local governments are often considered to be more capable of matching in a more adequate way public spending (i.e., public goods and services) to the heterogeneous preferences of the citizens living in different regions than central governments (Rodríguez-Pose and Ezcurra 2010). Decentralisation can also mobilise underused resources creating competition among subnational governments, thus delivering better policies (Rodríguez-Pose and Ezcurra 2010). Local governments are deemed to have information advantages over central governments regarding the needs and wants of locals (Ezcurra and Pascual 2008;Tselios et al. 2012; Tselios and Tompkins 2019) and local officials considered better able to design and implement policies at the local level than officials in remote central governments (de Mello 2011;Lessmann 2009;Tselios et al. 2012). Moreover, smaller constituencies are regarded as being more homogeneous and more easily controllable than larger ones, where voter power is diluted (Wyplosz 2015;Olson 1971). Thus, decentralisation can bring government closer to the people, thus promoting greater voice, accountability and participation ). Participation at a local level can result in the development of self-identity and local empowerment (Robinson 1988). Greater decentralisation can also foster greater transparency and highlight differences in the provision of public goods and services across territories (Rodríguez-Pose and Ezcurra 2010). In addition, greater transparency limits the opportunities for corruption, reducing the risk of an elite capture of the returns of public policies (Weingast 2009). Local governments have also political incentives to deliver growth and to achieve higher levels of economic development to lure votes for their parties and remain in power (Weingast 2009;Ezcurra and Pascual 2008). Generally, the presence of externalities and of returns to scale, information asymmetries and heterogeneous preferences favour decentralisation (Wyplosz 2015).
On the other hand, decentralisation and regional autonomy may undermine the capacity of the central state to exercise an equalizing role (Prud'homme 1995). Local governments may face skill shortages, due to problems in attracting suitably qualified officials and decision makers (Prud'homme 1995). If, in an absence of strong local accountability, local vested interests are powerful, decentralisation could facilitate elite capture of the returns of public policies and, consequently, increase social fragmentation and decrease the wellbeing of citizens (Bardhan and Mookherjee 2006;Neyapti 2006). In addition, in poor regions with low economies of scale, fiscal and political decentralisation may not result in a better matching of public goods and services to the needs of the population (Prud'homme 1995; Rodríguez-Pose and Ezcurra 2011). Decentralisation is associated with additional bureaucratic costs derived from the proliferation of layers of government. Poor countries can ill-afford these additional costs linked to institutional redundancy (Escaleras and Register 2012;Tselios and Tompkins 2017). Finally, local traps may be a consequence of a narrow focus on local perceptions and knowledge and become an obstacle to finding solutions from other locations (Brown and Purcell 2005).
Overall, there is no an ex-ante evaluation of the balance between the costs and benefits of fiscal and political decentralisation. The impact of these costs and benefits on the welfare of European citizens and their attitudes and beliefs about Europe differs not only between citizens but also between countries and over time.

Perceptions and Public Attitudes toward European Integration
What determines the public support for, and opposition to, European integration? Do the costs and benefits of decentralisation, which differ among individuals, explain variation in perception of the EU? The theory of socio-political perception can help us understand how people feel about Europe. It is widely known that perception is a synthetic and complex process involving many different psychological factors. Singer (1952) and Lakey et al. (1996) divide the determinants of perception into two categories: the personal characteristics (compositional effects), such as sex, age, education of people, and the environmental characteristics (contextual effects), which refer to the broader socioeconomic and political environment where the person lives, such as the macroeconomic condition of the country. In this paper, we examine whether fiscal and political decentralisation, which is an environmental influential factor, shapes people's perceptions and attitudes about European integration, after controlling for personal characteristics.
Following World War II, European Economic integration was driven by two forces: a) the expected collective benefits of rebuilding West European economies and b) the need to deliver peace and enhance international security (Gabel and Whitten 1997). The benefits from European integration are multidimensional as there are many political (e.g., European policy and security), economic (e.g., European economic integration, freedom of movement of goods, capital services and workers, international competition), and cultural (e.g., European identity) reasons. However, European integration is also bound to bring about a number of costs, both for individuals and firms-e.g., challenges to domestic producers-as well as for nation-states-e.g., loss of powers in areas like competition law and monetary policy. European integration can also have consequences for the spatial distribution of wealth and prosperity. Most of these benefits and costs can be measured with objective indicators, but citizens use other subjective and multidimensional criteria to support or oppose European integration. Taking into account both the 'objective' economy as well the 'subjective' economy, a key question is whether we need further European integration. The EU not only invests in deeper European integration, through new approaches to European integration (Schout and Jordan 2007), but, through specific policies, such as the European Cohesion policy, it also aims to reinforce solidarity among Europeans and to strengthen the sense of belonging to the EU (Tsaliki 2007).
According to Gabel (1998), five theories provide a theoretical basis for explaining the differences in public attitudes towards European integration: (a) the theory of cognitive mobilization (Inglehart 1970a), where high cognitive mobilization-that is, high political awareness and well-developed skills in political communication-enables the identification of individuals with the EU; (b) the theory of political values (Inglehart 1970b), where the political attitudes of people are shaped by the conditions they endured during their formative years; (c) the theory of utilitarian appraisals of integrative policy (Gabel and Palmer 1995), where citizens at different levels of income are confronted with different costs and benefits from European integration, and consequently, their support for integration is related to their expected potential welfare gains; (d) the theory of class partisanship (Inglehart et al. 1987), where citizens adopt attitudes towards integration following political party lines; and (e) the theory of government support (Franklin et al. 1994), where support for integration and for national government go hand in hand.
The potential influence of fiscal and political decentralisation on citizen's perception of the EU is mainly related with the theory of utilitarian appraisals of integrative policy, because EU citizens experience different costs and benefits from fiscal and political decentralisation and, hence, these differences shape their perceptions and attitudes toward European integration, after controlling for Economies 2020, 8, 38 7 of 26 several individual and national characteristics. On the one hand, decentralisation through 'localism' may reinforce the opposition to Europe, and on the other hand, decentralisation through 'Europeanism' may reinforce the support for Europe.
Overall, individual perceptions and attitudes are determined by a thorough analysis of the distributional outcomes of economic integration (Chalmers and Dellmuth 2015). Decentralisation provides differential welfare benefits and costs for EU citizens which reflect differences in variation in support for, or opposition to, Europe. The transfer or resources and powers to subnational tiers of government affects, by way of being filtered through the individual's predispositions, individual support for European integration (Chalmers and Dellmuth 2015).

Data and Variables
To study the influence of fiscal and political decentralisation on perceptions and attitudes towards the EU, data both on individuals and the places where they live is required. The individual data used in the analysis stems from The Mannheim Eurobarometer Trend File 1970-2002(Schmitt and Scholz 2005, which accumulates data from the European Commission's Eurobarometer Surveys. The surveys monitor the public opinion in the EU countries at times through approximately 1000 face-to-face interviews per country. The Mannheim Eurobarometer Trend File data set (edition v2.0.1), used for this study, includes data from the European Communities Studies 1970 up to Eurobarometer 57.2 (2002). That is a total of 86 waves and over 1,000,000 individual entries.
The Mannheim Eurobarometer Trend File 1970-2002 provides data for people's perceptions and attitudes towards the Common Market. The Common Market refers to the stage in the European integration process leading to the Single Market, where the basis for the removal of all the barriers to intra-Community trade were being set. 2 This implies the free movement of people, goods, capital and services across borders. In this survey, people were asked to indicate their positive, negative, or neutral view about the European Community. Two questions included in the surveys are crucial for this study: (a) 'Generally speaking, do you think that (your country's) membership of the European Community is 'a good thing', 'neither good nor bad thing' or 'a bad thing'?' (hereafter, EU membership variable); and (b) 'If you were to be told tomorrow that the European Community have been scrapped, would you be 'very sorry ' about it, 'indifferent' or 'very relieved'?' (hereafter, Regret variable). These questions refer to the two sides of the same coin. The first question has to do with the bright side of European integration, while the second question has to do with the consequences of a potential demise of the EU. Both the EU membership variable and the Regret variable are categorical.
The Mannheim Eurobarometer Trend File 1970-2002 also provides a Common Market index variable which combines the EU membership variable with the Regret variable. This index measures the respondents' perception and attitude towards economic European integration. Individuals with a positive attitude to membership of the EU or who would regret a potential implosion of the EU, are classified as 'strong supporters of economic integration'. Respondents with positive opinions of the European Community on only one variable are 'moderate supporters of economic integration'. Those with undecided or indifferent to integration, are included in a 'neutral' category. Respondents displaying a negative view of the EU in either the EU membership variable or the Regret variable are put in a 'moderate opponents of economic integration' category. Finally, respondents with negative views on both variables are 'strong opponents of economic integration'. The Common Market index variable is categorical. Figure 1 displays (a) the distribution of the EU membership variable by country, (b) the distribution of the Regret variable by country, and (c) the distribution of the Common Market index variable by country. 3 The citizens of the six original members of the current EU-Belgium, France, Germany, Italy, Luxembourg and the Netherlands-traditionally held a more positive view than negative view about the EU. Other countries with historically high positive views of the EU were Ireland, Portugal, Spain and, until the outbreak of the crisis, Greece. In contrast, the Austrians, Danes, Finns, Swedes, Brits and, outside the EU, Norwegians were generally more sceptical about the benefits of EU integration. It could be argued that citizens of countries with a longer membership of the EU were more likely to have a positive view about European integration than the dwellers of countries that became members later, with the exception of the United Kingdom. Therefore, the year of membership is likely to matter for people's perceptions.  In the analysis, country-level fiscal and political decentralisation are connected to the dataset of the views and characteristics of individuals across Europe. Two indicators of fiscal decentralisation are used: (a) the subnational share in total government expenditure (hereafter, FiscDec (expenditures) variable) and (b) the subnational share in total government revenue (hereafter, FiscDec (revenues) variable). Both indicators have been calculated using time-series collected in the IMF Government and Finance Statistics database. Although these indicators can be considered to reflect the dimensions and complexity of the process of fiscal decentralisation, they have been criticised for not properly identifying the 'true' level of fiscal autonomy awarded to subnational governments, for not differentiating, among other things, between tax and non-tax revenue, and for failing to determine the share of transfers are conditional or unconditional (Rodríguez-Pose and Ezcurra 2011). The level of political decentralisation is proxied using Hooghe et al.'s (2016) regional authority index (known as RAI index, hereafter, PolDec (RAI index) variable). The RAI index is the sum of the scores of two domains: self-rule, which refers to the authority exercised by a regional government over those who live in the region (score: 0-18), 4 and shared-rule, which refers to the authority exercised by a regional government or its representatives in the country as a whole (score: 0-12). 5 The RAI index measures political decentralisation along several dimensions and allows for variation over time (score: 0-30) . Figure 2 displays the degree of fiscal and political decentralisation by country between 1973 and 2002. It shows that Denmark, Germany and Sweden had the highest levels of fiscal decentralisation and Portugal, Luxemburg and France had the lowest levels. Moreover, there were no great differences between the FiscDec (expenditures) variable and the FiscDec (revenues) variable. As for 4 Reflecting (i) the degree of the autonomous power rather than deconcentrated power of regional governments (score: 0-3); (ii) the competence areas that are the responsibility of regional governments (score: 0-4); (iii) the independent capacity to tax of regional governments (score: 0-4); and (iv) the borrowing capacity of regional governments (score: 0-3); and (v) the presence of independent legislature and executive at regional levels (score: 0-4). 5 Calculated as the sum of the extent to which (i) regions influence the national legislative process (score: 0-2); (ii) regions influence national policy in intergovernmental meetings (score: 0-2); (iii) regions influence the geographical allocation of national tax revenues (score: 0-2); (iv) regions are capable of influencing subnational and national borrowing constraints (score: 0-2); and (v) regions affect constitutional change (score: 0-4).   Table 1 displays the mean of fiscal and political decentralisation for the EU membership, Regret, and Common Market index variables. The majority of citizens living in countries with high fiscal decentralisation considered EU membership to be less valuable and, as a consequence, were less likely to support European integration. This panorama changed when political, instead of fiscal, decentralisation was considered. Citizens of countries with high political decentralisation felt either neutral or positive about the EU. As for the Common Market index, the majority of citizens in high fiscal decentralisation countries were strong opponents of further economic integration, while those from countries with high political decentralisation were more prone to support further economic integration. Obviously, this   To examine whether fiscal and political decentralisation influenced people's perceptions towards the EU, we control for both compositional and contextual effects. To rule out compositional effects, the gender, age (in quadratic form) and education (measured by responses to the question: 'How old were you when you finished your full-time education?') of every individual is controlled for. The data stem from The Mannheim Eurobarometer Trend File 1970-2002 To rule out contextual effects, GDP per capita (at 2005 constant prices (in millions of 2005 US$)) and total factor productivity (tfp) (at 2005 constant prices) are included in the analysis. The origin of the data are the Penn World Tables (PWT) (Feenstra et al. 2015). Table 2 displays the descriptive analysis of the control variables. 48.26 percent of respondents are men, the mean age of those included in the sample is 43.37, and 25.81 percent of respondents had only basic levels of education. The mean GDP per capita of the countries included in the analysis is 29,285.69 and the mean tfp, 0.94. We do not control for other key individual control variables such as occupation, family composition and happiness due to the low number of observations.

Multinomial Logistic Regressions
The final database includes data for 1,116,576 individuals (i), living in 16 European countries (s) over 30 years  (t). The dependent variable is a categorical (polytomous) variable. In these variables the categories, C, take the values of 0, 1, …, C − 1 (i.e., C = 3 for the EU membership variable and the Regret variable, and C = 5 for the Common Market index variable). The two main explanatory variables are FiscDec (the proxy for fiscal decentralisation) and PolDec (the proxy for political decentralisation). In addition, a vector of control variables at individual level (ContInd) and a vector of country-level control variables (ContCount) are included in the analysis.
In the multinomial logistic regression model we assume that the observations of the dependent variable are independent from each other and represent a truly random sample of the population. The dependent variable has a multinomial distribution with the probability parameters A maximum likelihood (ML) estimation is used in the logistic regression. We resort to iteration in order to adjust the values of the estimated parameters until obtaining the ML value for the estimated parameters. The slope coefficients depict, by holding the model constant, the effect of a unit of change in an explanatory variable on the predicted logits with the other variables. The slope denotes, when holding the other variables constant, how a change in one unit in an explanatory variable affects the log of the odds. To explain the concept of 'log of odds', we need first to move from probability to odds, and to log of odds. The probability of a category j for a variable which ranges from 0 to 1. The odds of a category j for a variable ( ( ) ( ) ) is defined as the ratio of the probability of category j over the probability of category 0. The transformation from probability to odds is a monotonic transformation, this implies that the odds rise whenever the probability of category j increases and vice versa. The odds range from 0 to infinity. The log of odds ranges from negative infinity to positive infinity.
The multinomial logistic regression model is weighted by national population size. Weighting is normally used to increase the representativeness of the statistics and serves to highlight the extent to which each observation counts in the analysis. The parameters of interest become accurate population estimates, because the sample is a simple random sample of the total population. Since the weighted correlation coefficient between fiscal decentralisation and political decentralisation is high (i.e., above 0.5), the possible separate effects of fiscal and political decentralisation on the views of the individuals consulted about their country's membership of the EU is calculated. Moreover, when examining the effect of the FiscDec (revenues) variable, the economic development of the country cannot be controlled for due to the high weighted correlation between the FiscDec (revenues) variable and the GDP per capita variable. In all models, which are presented in Tables 3-5, the likelihood ratio chi-square with the p-values tell that the model has a significantly more accurate fit than an empty model (i.e., a model with no predictors included). 7 Finally, since weights increase the standard errors of estimates and introduce instability into the data, the analysis has also been 7 We have also calculated the relative risk ratio (rrr). This is the the ratio of the probability of selecting one potential outcome category over the odds of choosing the baseline category. The rrr can be provided upon authors' request. conducted using unweighted regressions-included in Appendix A 8 -in order to compare the weighted and unweighted results. No significant difference in the results is found between both approaches.
(a) The influence of decentralisation on perceptions of the EU Table 3 presents the multinomial logistic regression results for EU membership as the dependent variable. This table displays the effects of fiscal and political decentralisation on the views of individuals about their country's membership of the EU, after controlling for sex, age, education as well as national GDP per capita and tfp. Country-specific and time-invariant effects, such as size, borders, position, physical endowments and year of membership of the EU are controlled through the use of dummies. Year dummies are also included in the analysis to control for time-period specific effects.   Regression 1 shows that a one-unit increase in the FiscDec (expenditures) variable was associated with a 0.3619 decrease in the relative log odds of viewing the EU as 'a bad thing' vs. 'neither good nor bad' and a 0.2641 decrease in the relative log odds of viewing the EU as 'a good thing' vs. 'neither good nor bad'. In other words, Regression 1 shows that the higher the subnational share of total government expenditure, the lower both the negative (i.e., 'a bad thing') and the positive ('a good thing') view of individual citizens about their country's membership of the EU. Therefore, any rise in the share of subnational expenditure in total government expenditure seems to negatively have influenced the two extremes of the spectrum. In any case, the negative view outweighs the positive one (0.3619 > 0.2641), meaning that, if anything, fiscal decentralisation was more associated with a decrease in the level of attachment to the EU. Regression 2 indicates that a rise in the FiscDec (revenues) variable was associated with an increase in the odds of the EU being 'a good thing' vs 'neither good nor bad' (0.4022), and with a decline in the log odds of it being 'a bad thing' vs 'neither good nor bad' (0.2975). Therefore, increases in the share of subnational government revenue relative to total government revenue weakened the negative position and strengthened the positive position of European citizens towards the EU. Overall, the influence of fiscal decentralisation on popular perceptions of EU membership is sensitive to the proxy for fiscal decentralisation used in the analysis, with the results sensitive to whether fiscal expenditure or revenues are considered.
With regard to political decentralisation, proxied by the regional authority index (RAI index), the estimation in Regression 3 reveals that a one-unit increase in political decentralisation was associated with a 0.0025 increase in the relative log odds of viewing Common Market as 'a good thing' vs. 'neither good nor bad', with a p-value < 0.1. However, there was a much higher increase (0.0571) in the relative log odds of viewing the EU as 'a bad thing' vs. 'neither good nor bad', with a p-value < 0.01. Hence, the negative view outweighed by far the positive one. Therefore, greater political decentralisation was connected to a worsening of people's views about the EU.
(b) The influence of decentralisation on regret Table 4 presents the results of the multinomial logistic regression, using Regret as the dependent variable. The coefficients the pitch the potential effects of fiscal and political decentralisation on the views of individuals concerning a potential demise of the EU.   Regression 1 signals that a one-unit rise in the subnational share in total government expenditure was connected with a 0.5030 decline in the relative log odds of feeling 'very relieved' vs. 'indifferent' in the event of a disintegration of the EU. There was an even higher decrease (1.1637) in the relative log odds of feeling 'very sorry' vs. 'indifferent'. Thus, a rise in the share of total government expenditure at a subnational level negatively influenced both ends of the spectrum, as was the case in Table 3, but, in contrast to the previous results, the positive view about the EU ('very sorry') clearly outweighed the negative one ('very relieved') (1.1637 > 0.5030). Regression 2 indicates that a one-unit increase in the subnational share of total government revenue was associated with a 0.5642 decrease in the relative log odds of feeling 'very relieved' vs. 'indifferent'. This points to the fact that an increase in fiscal decentralisation-proxied by the FiscDec (revenue) variable-was associated with a decrease in the negative view about the EU.
The findings for political decentralisation in Tables 3 and 4 are cut from the same cloth. A oneunit increase in political decentralisation was associated with a 0.0305 increase in the relative log odds of feeling 'very sorry' vs. 'indifferent' in the case of a demise of the EU, but with a much higher increase (0.0900) in the relative log odds of feeling 'very relieved' vs. 'indifferent'. Hence, the general trend seems to be that political decentralisation dented citizens' support for Common Market.
(c) The influence of decentralisation on the Common Market index Table 5 presents the multinomial logistic regression results, using Common Market index as the dependent variable.   Regression 1 exposes that an increase in the subnational share of total government expenditure had a negative effect on both ends of the attitudes towards the EU spectrum (both the strong supporters and the strong opponents of economic integration). Therefore, an increase in fiscal decentralisation, proxied by the FiscDec (expenditure) variable, pushed European citizens towards a more neutral view of the economic benefits and costs of the European integration. Regression 2 shows than an increase in the subnational share of total government revenue had a negative influence on the strong opponents of economic integration. Rises in fiscal decentralisation, proxied by the FiscDec (revenue) variable, improved the perceptions of citizens towards European integration. Finally, Regression 3 presents the influence of political decentralisation on perceptions of citizens towards economic aspects of European integration. This regression shows that a one-unit increase in political decentralisation was connected with a 0.0920 rise in the log odds of 'strong opposition' vs. 'neutral position', a 0.0338 increase in the relative log odds of 'moderate opposition' vs. 'neutral position', a 0.0155 decline in the relative log odds of 'moderate support' vs. 'neutral position', and a 0.0193 rise in the relative log odds of 'strong support' vs. 'neutral position'. Taking the magnitude of the coefficients into account, it can be said that political decentralisation as a whole-as was the case in Tables 3 and 4-dented citizens' support for European integration. (

d) The influence of controls on European integrations
The results of the controls (Tables 3-5) are discussed only briefly. All control variables shaped perceptions of the EU by those individuals considered in the different surveys. First, women tended to be more extreme in their positive and negative views of the EU (i.e., good or bad thing) than men. Second, the identification, both positive and negative, of the young and the elderly with the EU tended to be more extreme than that of adults. Third, education was a key factor behind EU support: the higher the level of education of a respondent, the higher his/her support for the EU. Fourth, individuals living in richer countries were generally more favourable towards the EU, relative to poorer countries. In addition, fifth, the influence of tfp was sensitive to the proxy for decentralisation.

Discussion and Conclusions
This paper has provided a first glimpse at the connection between fiscal and political decentralisation and the attachment of European citizens to the EU over 1973-2002. The empirical analysis shows that decentralisation affected individuals identify with the EU. The model controlled for age, sex, GDP per capita, education, tfp, and country-and time-specific effects. Fiscal decentralisation is proxied by the share of total government revenue spent by subnational governments and was a factor in reinforcing citizens' support for European integration. If instead of focusing on revenue, we consider expenditure, fiscal decentralisation had a more neutral role on citizen's identification with the EU. Hence, from citizens' point of view, an increase in the share of total government revenue spent by subnational governments gave prominence to the economic benefits of European integration, while an increase in political decentralisation to the economic costs. Political decentralisation, in contrast and regardless of the variable considered, operated in only one direction: greater political decentralisation seems to have undermined the attachment of European citizens to the EU. Therefore, the objective of promoting 'Europeanism' may have converged with the objective of 'localism' through revenue autonomy, but may have diverged with the objective of 'localism' through the capacity of subnational tiers of government to implement the independent policies and governance.
The analysis in this paper is not without limitations, which mainly have to do with the choice of the period of analysis. When compatible data are made available, the analysis could be extended to include the period 2003-2020. This analysis of the period 2003-2020 should explore the impacts of the key historical events towards the European integration, such as the 2004 and 2007 enlargement, which was at the root of a weakening of cohesion in Europe; the 2008 financial and economic crisis increasing opposition to European integration; the 2015 refugee crisis; Brexit; and, more recently, the unknown spatial, social, economic and political impacts of COVID-19. Although this paper covers the period 1973-2002, it can provide an empirical framework for additional research. It would be valuable not only to refine our estimates by considering data spanning longer periods (comparing the two periods), but also to experiment with alternative proxies for decentralisation.
Despite the limitations of this study, the empirical evidence presented in this paper raises potentially interesting insights for the design of European, national and regional policies and for social cohesion in the EU. Policy-makers could use the findings of the period 1973-2002 for the design of European policies, as history always matters. Hence, this study is also relevant within the context of today's strained EU environment. If the EU is still today to promote the well-being of its citizens and countries through greater European integration ('Europeanism'), on the one hand, while promoting the importance of local authorities in member countries ('localism'), on the other, ignoring the implications of fostering decentralisation for citizens' attachment to the European project may leave decision-makers saddled with impossibly ambitious and impractical agendas. The results of the analysis show that the two objectives ('Europeanism' and 'localism') were related for the period 1973-2002 but did not necessarily go in the same direction. Revenue autonomy stimulated support for the EU, but this is not necessarily the case of political autonomy. Although this paper considers an historical period for the EU, policies always have a role to play when there is a trade-off between different objectives. There is, therefore, a need to look for optimal policies to achieve both objectives. The analysis presented in this paper, despite its limitations, raised several issues and can constitute the foundation for further research into the issues addressed and specifically how decentralisation affects people's perceptions towards Europe.
Author Contributions: Conceptualization, investigation, data collection, methodology, validation, writingoriginal draft preparation, V.T.; supervision, writing-review and editing, A.R.-P. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.