Crisis and Public Support for the Euro, 1990–2014

Abstract

This contribution analyses the evolution of public support for the single European currency, the euro, from 1990 to 2014 for a 12-country sample of the euro area (EA-12), focusing on the most recent period of the financial and sovereign debt crisis, starting in 2008. We find that citizens’ support for the euro on average was marginally reduced during the first six years of the crisis, and that support has remained at high levels. While the pronounced increase in unemployment in the EA-12 throughout the crisis has led to a marked decline in trust in the European Central Bank (ECB), it is only weakly related to support for the euro.

Originally published in: Felix Roth, Lars Jonung, and Felicitas Nowak-Lehmann D. Crisis and Public Support for the Euro, 1990–2014. Journal of Common Market Studies, Vol. 54, No. 4, 2016, pp. 944–960.

We have benefited from comments by participants at the 14th Göttinger Workshop ʻInternationale Wirtschaftsbeziehungenʼ in 2012, the fourth International IFABS Conference in 2012, the annual conference of Swedish economists in 2012, the 2013 EPCS Meeting, the 14th annual SNEE conference in 2013, the XVI Applied Economics Meeting in 2013, the EcoMod conference in 2013, the 21st International Conference for Europeanists in 2014 and meetings of DG ECFIN’s research fellowship initiative in 2014/2015. We thank three anonymous reviewers as well as Richard Baldwin, Michael Bordo, Eugénia da Conceição-Heldt, Barry Eichengreen, Michael Ehrmann, Javier Estrada, Stephan Klasen, Eric Ruscher, and Charles Wyplosz for valuable comments. Felix Roth is grateful to Raf van Gestel for excellent research assistance and to the Stiftung Mercator for funding the research project entitled ʻHas the crisis in the Eurozone undermined citizens’ support for European Monetary Union and the euro?ʼ. Earlier versions of this paper were published as CEPS Working Document No. 358 in 2011 and Working Paper 2012:20 of the Department of Economics at Lund University in 2012.

1 Introduction

Ever since the plans for a European Monetary Union and a single European currency were announced, social scientists have explored the determinants of public attitudes towards the new currency (see e.g., Banducci et al., 2003, 2009; Brettschneider et al., 2003; Deroose et al., 2007; Gärtner, 1997; Guiso et al., 2014; Hobolt & Leblond, 2009, 2014; Hobolt & Wratil, 2015; Kaltenthaler & Anderson, 2001). This study falls into this area of research by analysing the longest time series collected to date for public support for the single currency, covering the period 1990–2014 for a 12-country sample of the euro area (Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal and Spain – the EA-12).

We analyse the period 1990–1998 solely on a descriptive basis, before focusing on the period since the establishment of the euro (1999–2014) in the econometric analysis, making a distinction between the pre-crisis years 1999–2008 and the crisis years 2008–2014.¹ Our study takes its inspiration from the observation that citizens’ trust in the European Central Bank (ECB) fell significantly during the financial and sovereign debt crisis that started in 2008 (see e.g., Ehrmann et al., 2013; Roth et al., 2014; Wälti, 2012). This raises the question: has the euro, the currency supplied by the ECB, also suffered a loss in public support due to the crisis, similar to the fall in trust in the ECB?^2,3

In line with the literature (Guiso et al., 2014, p. 1, and Hobolt & Leblond, 2014, p. 132, and Hobolt & Wratil, 2015, p. 238), our analysis reveals that on average there is no empirical evidence of a significant erosion of citizens’ support in times of crisis. It remains largely unchanged. However, in contrast to the above-mentioned literature, we detect distinct differences in the time series of public support within the individual EA-12 countries in times of crisis. Estimating our panel time series data with the help of a fixed-effects dynamic feasible generalized least squares (FE-DFGLS) approach, we detect that these differences do not seem to be affected by inflation or growth of GDP per capita, but they are – by and large – negatively affected by unemployment.

For the quarter century covered by our analysis, the euro has always on average been supported by a majority of the citizens in the euro area and, since its introduction in 1999 – aside from short periods in Finland and Greece – in each individual member state of the EA-12. The suggestion that ‘the global economic crisis has sapped support for the euro’ (Jones, 2009, p. 1085) finds little empirical support – at least within the first six years of the crisis that we examine.

The remainder of this contribution is structured as follows: Section 2 discusses the role of public support for the euro. Section 3 considers the measurement of public support for the euro and describes the aggregated and individual country patterns. Section 4 specifies the econometric model, the research design and the data utilized. Section 5 presents the econometric results and Section 6 discusses the empirical findings. The last section concludes.

2 The Role of Public Support for the Euro

Why study public support for European monetary unification and the single currency, the euro? Our reply is straightforward: public support plays a crucial role in determining the sustainability of the euro. The glue that holds a monetary union together is the political will to maintain a single currency. The costs and benefits of the euro as perceived by the public are reflected in their support for the currency. By analysing public support, we are able to understand the factors that impact the sustainability of the euro. We highlight three strands of the argument below to support this view.⁴

First, according to Banducci et al. (2003, p. 686), citizens’ support for European monetary unification and the euro is critical to evaluate the future of European integration and the potential to move towards supranational governance. Similarly, Kaltenthaler and Anderson (2001, p. 141) argue that citizens’ support for the euro functions as ‘a crucial test case for whether and why European citizens may be willing to transfer power from the nation state to European institutions’. Following the sovereign debt crisis, De Grauwe (2010, 2014) argues that only deeper European political integration would guarantee the long-term success of EMU (Economic and Monetary Union).

Second, Jonung (2002, pp. 413–421) and Bordo and Jonung (2003) stress that citizens’ support is crucial for the political legitimacy of the euro. Weak political legitimacy will undermine the political unity behind it, thus eroding the glue that holds the monetary union together.

Third, high levels of citizens’ support can be interpreted as a shared sense of a ‘commonality of destiny’, which Baldwin and Wyplosz (2009, pp. 327–329) argue is a prerequisite for the smooth functioning of a currency union. The absence of such sentiment will likely lead to the dissolution of a currency union in the medium to long run. In this context, De Grauwe (2014, p. 133) argues that an important prerequisite for the proper functioning of a currency union might be a socially determined variable, such as solidarity, in contrast to the standard economic criteria found in the theory of optimum currency areas.

In sum, public support for the common currency is an important determinant of the sustainability of a monetary union. Traditionally, this aspect is neglected in assessments of the monetary policy of a nation state, as the existence of the national currency is taken as a self-evident fact. We find no studies, for example, of the popularity of the dollar or the pound. Such popularity data are only available – as far as we know – for the euro. Thus, our data are unique in the context of international comparisons.

3 Public Support for the Euro

3.1 Measuring Public Support for the Single Currency

We construct our measure of public support for the euro from data on responses to Eurobarometer (EB) surveys⁵ carried out bi-annually between 10–11/1990⁶ and 6/2014 (EB34–EB81). Here, the survey respondents were asked their opinion on several statements: ʻPlease tell me for each statement, whether you are for it or against it’. One statement was: ‘A European Monetary Union with one single currency, the euro’. The respondent could choose from the following answers: ‘For’, ‘Against’ or ‘Don’t Know’. The exact wording of the survey question was adjusted over time in response to the development of the monetary union (see Appendix 2).

The use of this survey question underlies the literature on public attitudes towards the single currency (see e.g., Banducci et al., 2003, 2009; Kaelberer, 2007; Kaltenthaler & Anderson, 2001). Following Gärtner (1997, pp. 488–489), we focus on the average percentage of net support measured as the number of ‘For’ responses minus ‘Against’ responses to the above question on the country level.⁷

3.2 Support in the EA-12

Figure 4.1 shows average net support for the single currency in the EA-12 country sample from 10–11/1990 to 6/2014.^8,9

Fig. 4.1

Average Net Support for the Single Currency in the EA-12 Countries, 1990–2014

Notes: Fig. 4.1 is based on 546 individual observations at the country level. Since the figure depicts net support, all values above 0 indicate that a majority of the respondents support the single currency. Data for EB45 are not available.

Data sources: Aggregated data from 1990 to 2014 include observations from EB34–EB81. The aggregate from 1990 to 1994 is only based on EA-10 countries, that is, EA-12 excluding Austria and Finland.

In Fig. 4.1, we identify four distinct phases in the history of the euro during the period 10–11/1990 to 6/2014. The first one covers the 1990s up to the actual establishment of the euro area in January 1999, when irrevocably pegged exchange rates were introduced among the euro area members. This period is characterized by a steady decline of net support from 47% in 3/1991 to 17% in 2–6/1997, with a rapid increase in net support to 51% until 10–11/1998. Whereas the average net support remained positive, net support was indeed negative (≥ −40) in Austria and Finland (1995–1997) and in Germany (1992–1997).

The second period starts with the introduction of the euro as a bookkeeping entity in January 1999 and ends with the launch of the euro as a fully-fledged currency in January 2002. Initially, net support deteriorated by 20 percentage points to 31% until 11–12/2000, increasing again to 55% in 3–5/2002. Net support was negative (≥ −4) only in Finland (1999–2000).

Our third period starts when the euro entered into actual circulation in January 2002. Whereas net support declines to 38% in 10–11/2003, from this time onwards until 3–5/2008, net support remains stable at an average mean level of 41% and a standard deviation of 3.5%. Net support was negative (≥−7) only in Greece (2005–2007).

Our fourth period begins with the financial and sovereign debt crisis in September 2008. An average mean level of 37% paired with a standard deviation of 4% from 10–11/2008 until 6/2014 suggests that there is no evident break between the pre-crisis and crisis periods. Net support has been positive in each individual EA-12 country. In 6/2014, in the sixth year of the crisis, net support is positioned at 41% (in total values, 67% support the euro vs. 26% who are against it). The summary statistics for the four phases of the population-weighted aggregated times series and the individual observations at the country level are shown in rows 3–10 in Table 4.A1, Appendix 3.

Figure 4.1 and its underlying single-country patterns in Fig. 4.A1 clarify five facts. First, on average, there always existed a majority of EA-12 citizens who supported the euro over the 25-year period. Second, since the establishment of the euro area in 1999, aside from short periods in Finland and Greece in pre-crisis times, a majority of citizens in each member state of the EA-12 supported the euro, even during the crisis. Third, in 6/2014, in the sixth year of the financial crisis and the fourth year of the sovereign debt crisis, on average, there is actually a slight increase (1 percentage point) in popular support for the euro compared to the pre-crisis period in 3–5/2008. Fourth, in 6/2014 a large majority of EA-12 citizens still supported the euro as a whole (net support >40), and the same was the case in each individual member state of the EA-12 (net support ≥25). Fifth, in comparison to a significant decline in net trust in the ECB since September 2008, net support for the euro remained almost stable in the EA-12 on average; it declined only slightly in most individual countries and even increased in some countries (Figs. 4.A4 and 4.A5, Appendix 3).

Fig. 4.A1

Net support for the single currency, EA-12 countries, 1990–2014

Data sources: EB34–EB81.

Fig. 4.A2

Net support for the single currency, non-EA countries, 1990–2014

Data sources: EB34–EB81.

Fig. 4.A3

Net support for the euro, six EA countries that joined EMU after 2001, 2004–2014

Data sources: EB62–EB81.

Fig. 4.A4

Average net support for the euro and net trust in the ECB, EA-12 countries, 1999–2014

Notes: Net support/trust levels for/in the euro/ECB above 0 indicate that a majority of citizens supports/trusts the euro/ECB. Net trust levels in the ECB below 0 indicate that a majority of citizens mistrust the ECB. The dashed lines distinguish the second, third and fourth periods in the history of the euro. Values on the x-axis depict the month(s) of fieldwork in the respective year.

Data sources: EB51–EB81.

Fig. 4.A5

Net support for the euro and net trust in the ECB, EA-12 countries, 1999–2014

Notes: Net support/trust levels for/in the euro/ECB above 0 indicate that a majority of citizens supports/trusts the euro/ECB. Net trust levels in the ECB below 0 indicate that a majority of citizens mistrust the ECB. The dashed lines distinguish the second, third and fourth periods in the history of the euro. For Greece, the time series is displayed from 2001 onwards.

Data sources: EB51–EB81.

3.3 Support across Member States

To analyse the impact of the crisis on net support for the euro across member states, Table 4.1 compares net support before the crisis (3–5/2008) with the level recorded in the sixth year of the crisis (6/2014) (and the respective change between these two periods). Following Roth et al. (2014, p. 308), values are displayed for EA-12, EA-4, EA-8, and each individual EA-12 country, as well as for a non-EA and an EA-6 country sample.

Table 4.1 Net support and changes in net support in the EA-12, EA-4, EA-8 and individual EA-12 countries, 2008–2014

Fig. 4.A6

Unemployment and net support for the euro, EA-12 countries, 1999–2014

Notes: The left-hand y-axis depicts the scale for unemployment ranging from 0 to 30. The right-hand y-axis depicts the scale for net support for the euro ranging from 0 to 80. The dashed line distinguishes the pre-crisis situation from the crisis times. The 0-line is adjusted for unemployment. For Greece, the time series is displayed from 2001 onwards.

Data sources: EB51–EB81 and Eurostat.

Fig. 4.A7

Inflation and net support for the euro in the individual EA-12 countries, 1999–2014

Notes: The left-hand y-axis depicts the scale for inflation ranging from −2 to 4. The right-hand y-axis depicts the scale for net support for the euro ranging from 0 to 80. The dashed line distinguishes the pre-crisis situation from the crisis times. The 0-line is adjusted for inflation. For Greece, the time series is displayed from 2001 onwards.

Data sources: EB51–EB81 and Eurostat.

Fig. 4.A8

Unemployment and net trust in the ECB, EA-12 countries, 1999–2014

Notes: The left-hand y-axis depicts the scale for unemployment ranging from 0 to 30. The right-hand y-axis depicts the scale for net trust in the ECB ranging from −100 to 100. The dashed line distinguishes the pre-crisis situation from the crisis times. The 0-line is adjusted for unemployment. For Greece, the time series is displayed from 2001 onwards.

Data sources: EB51–EB81 and Eurostat.

Fig. 4.A9

Inflation and net trust in the ECB in the individual EA-12 countries, 1999–2014

Notes: The left-hand y-axis depicts the scale for inflation ranging from −2 to 4. The right-hand y-axis depicts the scale for net trust in the ECB ranging from −100 to 100. The dashed line distinguishes the pre-crisis situation from the crisis times. The 0-line is adjusted for inflation. For Greece, the time series is displayed from 2001 onwards.

Data sources: EB51–EB81 and Eurostat.

Fig. 4.A10

Inflation and inflation perceptions, EA-12 countries, 2003–2011

Notes: The left-hand y-axis depicts the scale for inflation ranging from −2 to 4. The right-hand y-axis depicts the scale for inflation perception ranging from 0 to 0.6. The dashed line distinguishes the pre-crisis situation from the crisis times. The proxy for individual perceptions about inflation in the context of the national economic situation (PNE) has been available within the standard EBs from 3–4/2003 (EB59) onwards.

Data sources: EB59–EB75.

According to Table 4.1, similar to the findings in Fig. 4.1, average net support in the EA-12 actually increased by 1% point, from 40% to 41% from 3–5/2008 to 6/2014. Only a small difference can be observed between an EA-4 and an EA-8 country sample, with a slight drop in the EA-4 by 3% points and a small increase in the EA-8 by 1% point. The marginal drop in the EA-4 is driven by a decrease in Spain (−13), with a strong increase of 38% points in Greece leveling out the more pronounced decline of 22% points in Ireland. The marginal drop in the EA-8 is driven by Germany, France, and Italy, which either follow their pre-crisis paths with only slight declines of 3% and 2% points (in France and Italy) or even manage to augment support (in Germany), with an increase of 12% points. One EA-8 country, Belgium, registered a noteworthy decline of net support, with a 10%-point decrease.

Thus, although most EA-12 countries follow a stable time pattern, we detect distinct differences within the time series between the EA-4 and EA-8 countries. By focusing almost exclusively on country aggregates, these individual country differences have not been discussed in the existing literature (Guiso et al., 2014, p. 21, Hobolt & Leblond, 2014, p. 132, Hobolt & Wratil, 2015, p. 244–245). The results of Table 4.1 contrast sharply with comparable data showing net trust in the ECB, which declined by 52% points in the EA-12 and by 89% points in the EA-4, with Spain, Ireland, and Greece losing 103%, 71%, and 58% points in net trust, respectively (Table 4.A2, Appendix 3).

In contrast to the stable support in most EA-12 countries, support outside the euro area declined sharply by 27% points, from 1% in 3–5/2008 to −26%.¹⁰ The most pronounced declines occurred in the Czech Republic and Sweden, with respective values of −60% and −59%. In the UK, whereas the decline was still moderate, with a value of −17%, the level of −66 in 11/2012 is the lowest within the EU-27 over 1990–2014. Within the member countries that joined the euro after 2001, support increased by 13% points.¹¹ We focus on the original EA-12 countries while controlling for the robustness of the econometric results with an EA-18 country sample.¹²

4 Empirical Approach

4.1 Model Specification

So far, we have described public support for the euro. Next, we analyse channels potentially influencing support for the euro. We estimate net support for the euro as a function of inflation, growth in real GDP per capita and unemployment.¹³ This approach is solidly embedded within three strands of research: 1) the literature on popularity functions (Bellucci & Lewis-Beck, 2011, pp. 192–194, Nannestad & Paldam, 1994, pp. 215–216); 2) a recent study on trust in the ECB (Roth et al., 2014, pp. 306–307); and 3) a study that suggests further work on the macro-economic impact on the popularity of the euro (Banducci et al., 2009, p. 564). Thus, in our baseline model (4.1), aggregated net support for the euro is estimated as a function of inflation, growth of GDP per capita, unemployment, and macro-economic control variables deemed to be important:

$$ \mathrm{Support}\_{\mathrm{euro}}_{it}={\alpha}_i+{\beta}_1{\mathrm{Inflation}}_{it}+{\chi}_1{\mathrm{Growth}}_{it}+{\delta}_1{\mathrm{Unemployment}}_{it}+{\phi}_1{Z}_{it}+{w}_{it}, $$

(4.1)

Support_euro_it is the aggregated net support for the euro in country i during period t; Inflation_it, Growth_it, Unemployment_it and Z_it are, respectively, inflation, growth of GDP per capita, unemployment and macro-economic control variables, such as the change in the euro/US dollar exchange rate for country i during period t. α_i depicts a country-specific constant term, and w_it is the error term. Since we utilize an FGLS (Feasible Generalized Least Square) estimation approach, time dummies are not included within our baseline estimation.

4.2 Research Design

We proceed in two steps. First, support for the euro is studied from a macro perspective with a focus on feedback effects between support for the euro and the overall economic situation. Eq. (4.1) is estimated with an EA-12 country sample for 1999–2014, with a total number of 331 observations.¹⁴ Due to a lack of monthly or quarterly data on inflation, GDP and unemployment, it is not possible to cover the period prior to the establishment of the euro (1990–1998). Focusing on the period from 1999 onwards allows us also to compare our econometric results with those from other studies.

With t = 31 and n = 12 and thus with a ratio of t/n = 2.58, Eq. (4.1) is estimated via panel time series estimation. The analysis differentiates between a pre-crisis (1999–2008) and a crisis period (2008–2014). The matching between the macroeconomic variables and the Eurobarometer data follows a procedure proposed by Wälti (2012, p. 597).

Second, to corroborate the findings between inflation and support for the euro from the macro-econometric analysis in regressions 1–3 in Table 4.2, support for the euro is examined from a micro point of view using 136,587 individual observations. In this step, emphasis is put on inflation perceptions, controlling for the personal characteristics of the interviewee (age, gender, education, employment and legal status, and political attitudes) as well as perceptions concerning the employment and economic situations.

Table 4.2 Inflation, unemployment, GDP per capita growth, and net support for the euro: FE-DFGLS estimations (aggregated level), 1999–2014

4.3 Data Used

Data on support for EMU and the euro and trust in the ECB are taken from the biannual Eurobarometer survey. For the descriptive analysis, aggregated data on support for EMU and the euro from 1990 to 2014 include observations from EB34 (10–11/1990) to EB81 (6/2014).¹⁵ For the econometric analysis at the aggregated level, data on support for EMU and the euro and trust in the ECB from 1999 to 2014 include observations from EB51 (3–4/1999) to EB81 (6/2014). Monthly data on inflation (the change in the harmonized index of consumer prices) and unemployment rates are from Eurostat. Unemployment data were seasonally adjusted. Monthly data on GDP¹⁶ and population¹⁷ are taken from Eurostat’s quarterly database. To gain monthly observations, data on GDP and population were interpolated.¹⁸ Monthly data on the exchange rate of the euro vis-à-vis the US dollar are based on Eurostat data. A summary of the data utilized for the descriptive analysis (1990–2014) is given in Table 4.A1, Appendix 3; and data for the econometric analysis at the macro level from 1999 to 2014 are given in Table 4.A3, Appendix 3.

Data for the econometric analysis of individual observations are obtained from the ZACAT service from GESIS–Leibniz Institute for the Social Sciences and have been merged for the period 1999–2011; they include observations from EB51 (3–4/1999) to EB75 (5/2011). The merged variables utilized include support for the euro, inflation perceptions, and socio-economic background variables, including age, gender, education, legal and employment status, political attitudes and perceptions concerning the employment and economic situations. A summary of the descriptive statistics of all variables is given in Table 4.A4, Appendix 3.

5 Econometric Results

5.1 Macro Analysis

We estimate Eq. (4.1) by means of DOLS (dynamic ordinary least squares),¹⁹ a method that permits full control for endogeneity of the regressors (Stock & Watson, 1993; Wooldridge, 2009).²⁰ To correct for autocorrelation,²¹ we apply an FGLS procedure.²² Both applications lead to the following Eq. (4.2), representing our FE-DFGLS approach (the detailed steps leading from Eqs. (4.1)–(4.2) are explained in Appendix 4).

$$ {\displaystyle \begin{array}{l}\mathrm{Support}\_{\mathrm{euro}}_{it}^{\ast}={\alpha}_i+{\beta}_1{\mathrm{Inflation}}_{it}^{\ast}+{\chi}_1{\mathrm{Growth}}_{it}^{\ast}+{\delta}_1{\mathrm{Unemployment}}_{it}^{\ast}+{\phi}_1{Z}_{it}^{\ast}+\\ {}\sum \limits_{p=-1}^{p=+1}{\beta}_{2p}\Delta {\mathrm{Inflation}}_{it-p}^{\ast}+\sum \limits_{p=-1}^{p=+1}{\chi}_{2p}\Delta {\mathrm{Growth}}_{it-p}^{\ast}+\sum \limits_{p=-1}^{p=+1}{\delta}_{2p}\Delta {\mathrm{Unemployment}}_{it-p}^{\ast}+\\ {}\sum \limits_{p=-1}^{p=+1}{\phi}_{2p}\Delta {Z}_{it-p}^{\ast}+{u}_{it}\end{array}} $$

(4.2)

with α_i being the country fixed effect and Δ indicating that the variables are in first differences. Inflation, growth, and unemployment turn exogenous and the coefficients β₁, χ₁, δ₁and ϕ₁ follow a t-distribution. This property permits us to derive statistical inferences on the impact of inflation, growth and unemployment.²³ The asterisk (*) indicates that the variables have been transformed (purged from autoregressive processes) and that the error term u_it fulfils the requirements of the classical linear regression model (i.e., it is free from autocorrelation).

Table 4.2 shows the econometric results for Eq. (4.2) within our EA-12 country sample. When analysing the full sample (3–4/1999 to 6/2014) with 331 observations, in regression (1) unemployment is significantly (95% confidence level) and negatively (−1.4) related to support for the euro.²⁴ In contrast to unemployment, inflation and growth of GDP per capita are insignificantly related to support for the euro when estimating our full sample.²⁵

We have argued that the pre-crisis period (3–4/1999 to 3–5/2008) should be kept distinct from the crisis period (10–11/2008 to 6/2014); accordingly, regressions 2 and 3 split the full sample into a pre-crisis period and a crisis period to explore the impact of the crisis on popular support for the euro. Splitting the full sample reveals that the significant effect of unemployment (−1.4) on net support for the euro is driven by the crisis period, in which unemployment is negatively (−1.6) and significantly (95% confidence level) related to net support for the euro.²⁶ In contrast, whereas inflation is insignificantly related to support for the euro in times of crisis, it is strongly negatively (−12.7) and highly significantly (99% confidence level) related to net support for the euro in the pre-crisis period.^27,28,29

The relatively weak coefficient of −1.6 between unemployment and net support for the euro in times of crisis is in clear contrast to a much larger coefficient of −6.6 between unemployment and net trust in the ECB in times of crisis (see Table 4.A9, Appendix 3). Thus, in times of crisis, an increase in unemployment exerts an effect on net trust in the ECB that is more than four time stronger compared to the effect on net support for the euro. Whereas the pronounced increase in unemployment rates in the EA-12 throughout the crisis – with the exception of Germany, but in particular in the EA-4 – has led to a significant decline in net trust in the ECB, it has only led to a slight decline in net support for the euro in the EA-12, in particular in Spain and Ireland.³⁰ It even followed opposite trends in Greece at the beginning of the crisis. Interestingly, whereas the reduction of unemployment rates in Germany was positively associated with a significant decline in net trust in the ECB, it is negatively associated with an increase in net support for the euro, and thus contributes to the weak negative evidence between unemployment and net support for the euro.³¹

We are confident that our econometric analysis has not omitted any important variables, having found that our time series are cointegrated.³² However, to address concerns over missing variables, we include the change in the euro/US dollar exchange rate, as Banducci et al. (2003, p. 694; 2009, p. 571), Brettschneider et al. (2003, p. 50) and Hobolt and Leblond (2009, 2014, p. 137) stress its importance for support for the euro. The inclusion of the change in the euro/US dollar exchange rate in Table 4.A10, Appendix 3 does not alter our results in any substantial manner – although the growth of GDP per capita renders significant (95% confidence level) in times of crisis. This observation confirms previous empirical results, which find a positive and significant relationship between the change in the euro/US dollar exchange rate and net support for the euro in pre-crisis times. In times of crisis, however, it is insignificantly related to net support for the euro.

To corroborate our results for the complete euro area, we include an EA-6 country sample. Estimating an EA-18 country sample, as shown in Table 4.A11, Appendix 3, does not change the key econometric results in any substantial manner, although inflation is rendered significant (95% confidence level) in times of crisis.

5.2 Micro Analysis

To extend our study of the relationship between the official inflation rate and net support for the euro from regressions 1–3 in Table 4.2, and as Banducci et al. (2009) suggest that the actual economic situation – as summarized in official economic statistics – does not necessarily accord with the perceived economic situation, in Eq. (4.3) we examine the support for the euro based on a probit model and individual data, to account for citizens’ perceptions towards inflation.³³ The data set at hand does not allow us to track individuals over time. The equation for the probit model is expressed as follows:

$$ P\left(\mathrm{Support}\_{\mathrm{euro}}_{jit}=1\right)={\alpha}_i+\beta \mathrm{Inflation}\kern0.5em {\mathrm{PC}}_{jit}+\chi \mathrm{Economic}\kern0.5em {\mathrm{PC}}_{jit}+\delta \mathrm{Unemployment}\kern0.5em {\mathrm{PC}}_{jit}+\phi\;{Z}_{jit}+{\gamma}_t+{\varepsilon}_{jit}, $$

(4.3)

where P represents the probability with which the euro is supported. The dependent variable (Support_euro_jit) represents the support for the euro for individual j in country i at time t and takes on 1 if the individual supports the euro and 0 if the individual does not support the euro. Inflation,³⁴ Economic and Unemployment PC_jit represent the inflation, economic and unemployment perceptions for the national economic situation or personal economic situation for individual j in country i at time t. Z_jit represents micro controls including age, gender, education, employment and legal status and political orientation for individual j in country i at time t; α_i represents the country fixed effects; γ_t represents the time-fixed effects; and ε_jit represents the error term.

To corroborate the findings between inflation and net support for the euro, Table 4.3 only displays and analyses the value for the β-coefficient in Eq. (4.3) (in other words, the impact of the perception of inflation on support for the euro). Controlling for the above-mentioned specification with a maximum number of 136,587 individual observations, inflation perceptions, in contrast to the official inflation rate, have the expected negative effect in all three samples (Full Sample, Before Crisis and Crisis) for the national economic situation (regressions 1, 2 and 3) as well as the personal economic situation in times of crisis (regression 6). As the values depict the marginal effects, the interpretation of the coefficient in times of crisis is as follows: an individual who identifies inflation to be an important issue either for the national economy or for his/her personal economy in times of crisis is around 2.9 or, respectively, 4.4% less likely to support the euro than an individual who has not identified inflation to be an important issue.

Table 4.3 Inflation perceptions and support for the euro – probit analysis (individual level), 2003–2011

5.3 Previous Findings

How do our econometric results for the crisis period 2008–2014 square with previous findings? First, in contrast to Hobolt and Leblond (2014, p. 141), we find a significant and negative relationship between unemployment and net support for the euro in times of crisis.^35,36 Second, similar to previous empirical findings (Roth et al., 2014, p. 310), this negative relationship is four times smaller than the one between unemployment and net trust in the ECB. Third, in contrast to Banducci et al. (2009, p. 571)³⁷ and Hobolt and Leblond (2014, p. 141),³⁸ we find a significant and negative relationship between inflation and net support for the euro in pre-crisis times, in line with underlying theoretical literature (Kaelberer, 2007, p. 626). The negative relationship, however, is insignificant in times of crisis. Fourth, at the micro level, we are able to confirm the negative relationship between inflation perception and support for the euro in pre-crisis times, as found by Banducci et al. (2009, p. 576), and we also find a similar negative relationship in times of crisis. These differences in the empirical results suggest the need for further research on the determinants of public support for the euro.

6 Discussion

Our econometric results invite a number of comments concerning the future of the euro. First, how should we interpret the support for the euro in light of the theoretical literature and the first six years of the crisis? Following the arguments by Banducci et al. (2003, p. 686) and Kaltenthaler and Anderson (2001, p. 141), support for the euro within the EA-12 during the crisis period 2008–2014 suggests that there may be scope for further political integration to strengthen the sustainability of the single currency, as argued by De Grauwe (2010, 2014). Following Jonung (2002) and Bordo and Jonung (2003), support for the euro during the crisis indicates that the political glue necessary for the euro is present within the EA-12. In a similar vein, according to the arguments of Baldwin and Wyplosz (2009, pp. 327–329) and De Grauwe (2014, p. 133), the key prerequisite for the smooth functioning of a currency union – the sense of a ‘commonality of destiny’ or solidarity – is still present within the EA-12.

Second, how should we interpret the fact that the ECB bears the brunt of the blame for the unemployment crisis in the EA-12, as opposed to the actual euro? One could argue that euro-area citizens simply continue to want the euro as their currency and do not hold the euro per se responsible for the unemployment crisis. Instead, they blame policymakers and their institutions. Consequently, the decline in trust in the ECB is part of a larger decline in systemic trust due to the crisis, including institutions of democratic governance at the European and national level (see e.g., Ehrmann et al., 2013; Roth et al., 2013). It may be the case that citizens support the euro because the euro is a binary regime from which exit would have worse consequences than staying in (Guiso et al., 2014, p. 32, Hobolt & Leblond, 2014, p. 142). In contrast, the ECB is a policy-making institution that is held accountable by citizens for the crisis.

Finally, the fact that the euro – a centerpiece of European integration – still finds support during the crisis should be viewed as a necessary condition for its survival. The future will show if this support is sufficient to guarantee its existence.

7 Conclusions

Five findings emerge from the analysis presented in this contribution.

First, the analysis covering the 25-year period from 1990 to 2014 for the EA-12 country sample shows that, on average, a majority of citizens has supported the single European currency.

Second, since the establishment of the euro in January 1999, aside from short time periods in Finland and Greece in the pre-crisis period, a majority of citizens in each individual member state of the EA-12 supported the euro, even during the crisis period 2008–2014.

Third, the crisis only slightly dented support for the euro in most EA-12 countries and even increased it in some. This finding contrasts with the evolution of net trust in the ECB, which declined in a pronounced way due to the crisis.

Fourth, the difference between net support for the euro and net trust in the ECB during the crisis can largely be explained by changes in unemployment rates. Whereas the pronounced increase in unemployment rates in the EA-12 during the crisis – with the exception of Germany, but in particular in the EA-4 – has led to a significant decline in net trust in the ECB, it only led to a slight decline in net support for the euro in the EA-12.

Fifth, whereas an insignificant relationship was detected between inflation and net support for the euro during the crisis at the macro level, a negative link was found between citizens’ perceptions towards inflation and support for the euro.

Finally, the support shown for the euro both before as well as after the crisis suggests that one of the most important prerequisites for a sustainable monetary union is present within the EA-12. The future will show how well European policymakers manage to sustain this support for the single currency.

Appendices

Appendix 1: Individual Country Time Series for the EU-27, 1990–2014

Appendix 2: A Detailed Breakdown of the Questionnaire over the 25-year Time Period

Over the course of the 25-year period examined in this study, the question of whether a common, single currency is supported has been slightly modified. The wording of the proposals in EB34 to EB37 reads: ‘Within this European Economic and Monetary Union, a single common currency replacing the different currencies of the Member States in 5 or 6 years’ time.’ The wording of the question from EB38 to EB40 reads: ‘There should be a European Monetary Union with one single currency replacing by 1999 the (national currency) and all other national currencies of the Member States of the European Community.’ After ratification of the Maastricht Treaty, the wording in EB41 was changed to: ‘(…) Member States of the European Union and European Community.’ From EB42 onwards, the ‘European Community’ was dropped. From EB44 onwards, the ‘by 1999′ was dropped. From EB46 onwards, the ‘euro’ is introduced and the wording ‘European Monetary Union’ is taken out. From EB48 onwards, the word ‘should’ is replaced by ‘has’. From EB54 onwards, the wording ‘replacing the (national currency) and all other national currencies’ is dropped and the wording ‘European Monetary Union’ is reintroduced. In addition, from EB54 onwards the word ‘statement’ replaced the word ‘proposal’. From EB56 to EB72 and onwards, ‘There has to be’ is dropped. The question in EB56 to EB72 represents the wording as highlighted within our main text. From EB73 onwards, the ‘European Monetary Union’ was replaced by the ‘Economic and Monetary Union’.

As we are of the opinion that these changes in the framing of the question over time are related to context and time and are not responsible for any significant changes in the responses, we ignore these slight modifications to the survey questions. A similar argument is made by Banducci et al. (2003, p. 690) for the time-period 1990–2000. However, in contrast to Banducci et al. (2009, p. 570), who argue that an alteration in the wording of the question prevented them from comparing the 1990–2000 period with the 2000–07 period, we believe that the introduction of the new wording in EB 54 in 11–12/2000 (as described above) should not prevent an empirical analysis from making a comparison. The change in values from 4–5/2000 to 11–12/2000, with a decrease of 7% points from 38% to 31% is well situated within the pattern of the 25-year time series (see here also the summary statistics in row 1 in Table 4.A1, Appendix 3).

Table 4.A1 Summary statistics for the descriptive analysis

Table 4.A2 Net trust and changes in net trust in the ECB in EA-12, EA-4, EA-8 and individual EA-12 countries, 2008–14

Table 4.A3 Summary statistics for the macro analysis, 1999–2014

Table 4.A4 Summary statistics for the micro analysis, 2003–11

Table 4.A5 Pesaran’s CADF panel unit root tests, EA-12 countries

Table 4.A6 Kao’s residual cointegration test, EA-12 countries

Table 4.A7 Sensitivity analysis between unemployment, inflation and net support for the euro: FE-DFGLS estimations (aggregated level), 1999–2014

Table 4.A8 Correlation coefficients between unemployment, inflation and net support for the euro, net trust in the ECB in the EA-12 countries, 1999–2014

Table 4.A9 Inflation, unemployment, GDP per capita growth and net trust in the ECB: FE-DFGLS estimations (aggregated level), 1999–2014

Table 4.A10 Inflation, unemployment, GDP per capita growth, change in euro/US dollar exchange rate and net support for the euro: FE-DFGLS estimations (aggregated level), 1999–2014

Table 4.A11 Inflation, unemployment, GDP per capita growth and net support for the euro: FE-DFGLS estimations (aggregated level), EA-18, 1999–2014

Appendix 3: Descriptive Statistics and Test Results

Appendix 4: Detailed Steps Leading from Eqs. (4.1)–(4.2)

In the baseline model (4.1), aggregated net support for the euro is estimated as a function of inflation, growth of GDP per capita, unemployment and macro-economic control variables deemed of importance:

$$ \mathrm{Support}\_{\mathrm{euro}}_{it}={\alpha}_i+{\beta}_1{\mathrm{Inflation}}_{it}+{\chi}_1{\mathrm{Growth}}_{it}+{\delta}_1{\mathrm{Unemployment}}_{it}+{\phi}_1{Z}_{it}+{w}_{it}, $$

(4.1)

where i represents each country and t represents each time period; Support_euro_it is the net support for the euro in country i during period t; Inflation_it, Growth_it, Unemployment_it and Z_it are respectively inflation, growth of GDP per capita, unemployment and macro-economic control variables, such as the change in the euro/US dollar exchange rate for country i during period t. α_i depicts a country-specific constant term and w_it is the error term. As we utilize a Feasible Generalized Least Square (FGLS) estimation approach, time dummies are not included within our baseline estimation, as they are mutually exclusive with FGLS using EViews (version 7.2).

1.1 The Issue of Endogeneity

When running regressions such as in Eq. (4.1), one must be aware of the possibility that the right-hand side variables (inflation, growth and unemployment) might be endogenous (affected by a common event) or stand in a bi-directional relationship with support for the euro (a low level of support might lead to a self-fulfilling prophecy, speeding up and worsening an existing downturn). Therefore, we estimate the model by means of dynamic ordinary least squares (DOLS), a method that controls for the endogeneity of the regressors (Stock & Watson, 1993; Wooldridge, 2009).

It can be shown that by decomposing the error term and inserting the leads and lags of the right-hand side variables in first differences, the explanatory variables become (super-) exogenous and the regression results thus become unbiased. The baseline regression, which does not control for endogeneity and reflects a situation whereby all adjustments have come to an end, has already been depicted in Eq. (4.1) above. Within Eq. (4.1) w_it is the iid-N error term with the properties of the classical linear regression model. Controlling for endogeneity requires the decomposition of the error term w_it into the endogenous changes of the right-hand side variables, which are correlated with w_it(the changes in the variables) and the exogenous part of the error term υ_it; with

$$ {\displaystyle \begin{array}{l}{\mathrm{w}}_{it}=\sum \limits_{p=-1}^{p=+1}{\beta}_{2p}\Delta {\mathrm{Inflation}}_{it-p}+\sum \limits_{p=-1}^{p=+1}{\chi}_{2p}\Delta {\mathrm{Growth}}_{it-p}\\ {}+\sum \limits_{p=-1}^{p=+1}{\delta}_{2p}\Delta {\mathrm{Unemployment}}_{it-p}+\sum \limits_{p=-1}^{p=+1}{\phi}_{2p}\Delta {Z}_{it-p}+{\upsilon}_{it}\end{array}} $$

(4.1a)

Inserting Eq. (4.1a) into Eq. (4.1) leads to the following Eq. (4.1b) in which all explanatory variables from the baseline model can be considered exogenous:

$$ {\displaystyle \begin{array}{l}\mathrm{Support}\_{\mathrm{euro}}_{it}={\alpha}_i+{\beta}_1{\mathrm{Inflation}}_{it}+{\chi}_1{\mathrm{Growth}}_{it}+{\delta}_1{\mathrm{Unemployment}}_{it}+{\phi}_1{Z}_{it}+\\ {}\sum \limits_{p=-1}^{p=+1}{\beta}_{2p}\Delta {\mathrm{Inflation}}_{it-p}+\sum \limits_{p=-1}^{p=+1}{\chi}_{2p}\Delta {\mathrm{Growth}}_{it-p}+\sum \limits_{p=-1}^{p=+1}{\delta}_{2p}\Delta {\mathrm{Unemployment}}_{it-p}+\\ {}\sum \limits_{p=-1}^{p=+1}{\phi}_{2p}\Delta {Z}_{it-p}+{\upsilon}_{it}\end{array}} $$

(4.1b)

with α_i representing country fixed effects and Δ indicating that the variables are in first differences; the error term υ_itshould fulfil the requirements of the classical linear regression model. Inflation, growth and unemployment become exogenous and the coefficients β₁,χ₁, δ₁ and ϕ₁ follow a t-distribution. This property allows us to draw statistical inferences on the impact of inflation, growth and unemployment on support for the euro.

1.2 Omitted Variables and Autocorrelation

Having found that net support for the euro and the economic variables (inflation, growth and unemployment) are non-stationary and cointegrated, we can be confident that omitted variables (which are lumped together in the error term) do not systematically influence our long-run relationship between support for the euro and macroeconomic variables. Omitted variables could include: socio-political factors such as positive attitude towards EU membership (Banducci et al., 2009; Hobolt & Leblond, 2014), mass media attention (Brettschneider et al., 2003), consumer confidence (Hobolt & Leblond, 2014), or macro-economic variables of importance, such as the change in the euro/US dollar exchange rate and the interest rate (Banducci et al., 2003, 2009; Hobolt & Leblond, 2014), as well as social indicators,³⁹ such as measures of income inequality and poverty rates, which have most likely deteriorated within the periphery countries of the EA-12.

Even though the error term is stationary [I(0)], a characteristic of cointegration, autocorrelation of the error terms might still be a problem that must be fixed. We do so by applying the two-step FGLS procedure. In a first step, we collect the ${\hat{\upsilon}}_{it}$s from Eq. (4.1b), which has been estimated by means of DOLS. Thereafter, we estimate ρ₁ the first-order autocorrelation⁴⁰ coefficient, via OLS based on Eq. (4.1c).

$$ {\hat{\upsilon}}_{it}={\rho}_1{\hat{\upsilon}}_{it-1}+{u}_{it} $$

(4.1c)

Since the coefficient ρ₁ is usually unknown (as in our case), it has been estimated (giving us ${\hat{\rho}}_1$) by means of the Cochrane–Orcutt method (Pindyck & Rubinfeld, 1991), which is an FGLS procedure. In a second step we transform all variables of Eq. (4.1b), which can be described by the following formulas (4.1d):

$$ {\displaystyle \begin{array}{l}\mathrm{Support}\_{\mathrm{euro}}_{it}^{\ast}=\mathrm{Support}\_{\mathrm{euro}}_{it}-{\hat{\rho}}_1\mathrm{Support}\_{\mathrm{euro}}_{it-1},\\ {}{\mathrm{Growth}}_{it}^{\ast}={\mathrm{Growth}}_{it}-{\hat{\rho}}_1{\mathrm{Growth}}_{it-1},\\ {}{\mathrm{Inflation}}_{it}^{\ast}={\mathrm{Inflation}}_{it}-{\hat{\rho}}_1{\mathrm{Inflation}}_{it-1},\\ {}{\mathrm{Unemployment}}_{it}^{\ast}={\mathrm{Unemployment}}_{it}-{\hat{\rho}}_1{\mathrm{Unemployment}}_{it-1},\\ {}{Z}_{it}^{\ast}={Z}_{it}-{\hat{\rho}}_1{Z}_{it-1}\end{array}} $$

(4.1d)

where the differences of the explanatory variables are transformed in exactly the same way as the variables in levels. Correcting for autocorrelation in the error term via FGLS leads to Eq. (4.2):

$$ {\displaystyle \begin{array}{l}\mathrm{Support}\_{\mathrm{euro}}_{it}^{\ast}={\alpha}_i+{\beta}_1{\mathrm{Inflation}}_{it}^{\ast}+{\chi}_1{\mathrm{Growth}}_{it}^{\ast}+{\delta}_1{\mathrm{Unemployment}}_{it}^{\ast}+{\phi}_1{Z}_{it}^{\ast}+\\ {}\sum \limits_{p=-1}^{p=+1}{\beta}_{2p}\Delta {\mathrm{Inflation}}_{it-p}^{\ast}+\sum \limits_{p=-1}^{p=+1}{\chi}_{2p}\Delta {\mathrm{Growth}}_{it-p}^{\ast}+\sum \limits_{p=-1}^{p=+1}{\delta}_{2p}\Delta {\mathrm{Unemployment}}_{it-p}^{\ast}+\\ {}\sum \limits_{p=-1}^{p=+1}{\phi}_{2p}\Delta {Z}_{it-p}^{\ast}+{u}_{it}\end{array}} $$

(4.2)

with α_i being the country fixed effect and Δ indicating that the variables are in first differences; * indicating that the variables have been transformed (purged from autoregressive processes) and that new error term u_it(${u}_{it}={\upsilon}_{it}-{\hat{\rho}}_1{\upsilon}_{it-1}$) fulfils the requirements of the classical linear regression model (it is free from autocorrelation). Equation (4.2), which is an improved version of Eq. (4.1b), represents the fixed effects dynamic feasible generalized least squares (FE-DFGLS) approach.