Introduction

Crises are times for political leadership. Critical historical junctures in which a society's core values are threatened and its political and economic order is upset call for and favour the emergence of strong political agents who can take matters in their hands, reduce uncertainty and guide the community out of the mire (Ansell et al. Reference Ansell, Boi, ‘t Hart, Rhodes and ‘t Hart2014). In particular, situations of crisis call for charismatic leaders: extraordinary political actors able to assuage people's stress and anxiety by presenting inspiring visions and solutions for the future, so breaking the collective impasse and creating the conditions for new social orders to materialise (Tucker Reference Tucker1968; Pillai Reference Pillai1996).

The European Union (EU), and most notably the Eurozone, has been in a state of economic, political and institutional crisis for almost a decade now and yet leadership – let alone charismatic leadership – is still in scarce supply among its key players. Due to their limited resources and clout, common political institutions (in the first place, the European Commission, the EU's executive body) struggle to fulfil this role beyond wide‐ranging statements whose effectiveness remains questionable (e.g., the ‘Five Presidents’ Report’; Juncker Reference Juncker2015) – or narrower and more concrete yet negligible actions (e.g., the recent ‘Juncker investment plan’). Neither is there much effective leadership at the level of member states. The guidance provided by Germany – the EU's ‘natural hegemon’ – has been intermittent at best, and in any case largely unimaginative and centred on the preservation (and strengthening) of the Eurozone's existing legal and economic paradigm (Bulmer Reference Bulmer2014; Schoeller Reference Schoeller2016). At the same time, alternative loci of (potential) leadership, such as Italy or France, are either too weak, both structurally and politically, or have yet to prove to be credible candidates to lead Europe out of its predicament (Longo & Murray Reference Longo and Murray2015).

Amid this vacuum of political leadership, it is a formally technocratic actor that has emerged to provide some clear and effective responses to Europe's crisis – namely the European Central Bank (ECB). ‘The dramatic days of the euro crisis are, thank God – or thank Mario Draghi, now behind us,’ former European Parliament president Martin Schulz commented in early 2015 (Schulz Reference Schulz2015). While Schulz's take on the end of the crisis might have been overly optimistic, it is certainly fair to say that the ECB's actions in defence of the euro – culminating with President Draghi's oft‐cited ‘whatever it takes’ pledge – are the closest thing to effective supranational crisis leadership the Eurozone and the EU as a whole have seen in recent years (Verdun Reference Verdun2017). Needless to say, in the eyes of many this has come at the cost of an increased ‘politicisation’ of the ECB and its policies.

This article aims to investigate whether and to what extent the leadership of the ECB, and more precisely its presidency, can be characterised as charismatic by looking at a particular aspect of it – namely language. The question of ECB charisma is important from both analytical and normative standpoints: for one thing, it bears on the transformation of EU governance and political equilibria resulting from the euro crisis. For another, it relates to the issue of the democratic legitimacy of the Bank in the context of its increasingly political role.

The article proceeds as follows. It begins by providing an overview of the ECB's actions to tackle the euro crisis, and highlights how these can be seen as configuring the Bank as a ‘euro leader’. Building on this premise, the article proceeds to examine the applicability of the concept of ‘charismatic leadership’ to the ECB case, and introduces the topic of language as a signal of charisma. Thereafter the research design is presented, along with the results of a software‐assisted content analysis of 560 speeches of the ECB president, showing that the Bank's presidency has indeed emerged as a rhetorically charismatic actor as a result of the euro crisis. The final section of the article discusses the implications of these results and concludes.

The ECB's role in tackling the euro crisis

The ECB's responses to the euro crisis have been analysed in detail elsewhere (e.g., Beukers Reference Beukers2013; Micossi Reference Micossi2015), so only a summary will be presented here. In the early phases of the international financial crisis, following the September 2008 collapse of Lehman Brothers, the ECB acted through a number of instruments with the main objective of tackling liquidity problems within the Eurozone. In the first place, between October 2008 and May 2009 it lowered interest rates from 4.25 to 1 per cent. In addition, the Bank extended the maturity of its refinancing operations and relaxed collateral requirements for loans to the financial sector. In the spring of 2009 the ECB also launched the first instalment of its new Covered Bond Purchase Programme (CBPP1), planned to continue for one year and for a total amount of €60 billion.

The improvements in the Eurozone's financial situation resulting from these measures proved short‐lived: in early 2010, revelations about Greece's previous budget data falsifications marked the beginning of the country's public debt crisis. The ECB's response to the Greek crisis was mainly twofold. First, in May 2010, the Bank introduced a Securities Market Programme (SMP) to carry out sterilised purchases of Greek bonds in secondary markets. The SMP represented the ECB's ‘first foray into distressed sovereign bonds markets’ (Micossi Reference Micossi2015: 13) and was later renewed several times to also assist Ireland, Portugal, Italy and Spain. Furthermore, in the same period the ECB also started having a more direct role in dealing with the Greek crisis by joining the ‘troika’ of creditors – together with the European Commission and the International Monetary Fund (IMF) – that would negotiate and monitor the implementation of Greece's structural reforms in the context of its international bail‐out packages. Later on, this ECB function would be generalised and embedded legally with the establishment of the European Financial Stability Facility and the European Stability Mechanism.

The year 2010 also marked the beginning of the ECB's practice of sending letters to governments whose economies were receiving assistance to request certain measures and reforms in exchange. In October and November, for instance, two such letters were sent to Ireland asking the country to apply for a bail‐out and threatening to cut the Emergency Liquidity Assistance (ELA) that was propping up the country's financial institutions. Similar letters were sent to the Italian and Spanish governments the following summer asking for a series of structural reforms (e.g., in the area of pensions and labour regulations) as a condition for buying the countries’ bonds under the SMP. Spain and Italy were also the main beneficiaries of two rounds of Very Long Term Refinancing Operation (VLTROs) implemented in December 2011 and February 2012, respectively. Finally, in October 2011, the ECB launched a second one‐year CBPP for a total of €40 billion.

In November 2011 Mario Draghi took over the ECB presidency from Jean‐Claude Trichet. One of Draghi's first moves was to lower the Bank's refinancing rate (previously raised by Trichet) from 1.5 to 1.25 per cent and then to 1 per cent – initiating a path of steady rate reduction that has continued to this day. Draghi's moves did not prevent yield spikes in Spanish and Italian government bonds, which were caused by the markets questioning the solvency of the two countries. The risk of default in Spain and Italy was key to Draghi declaring, on 26 July 2012, that the ECB was ready to ‘do whatever it takes to preserve the euro’. The speech preluded the launch, in September, of the Outright Monetary Transaction (OMT) programme for the unlimited purchase of government bonds, conditional on reform programmes on the part of the assisted countries. An augmented version of the SMP (which it replaced), the OMT proved to be so effective as to reassure financial markets without ever being actually employed.

In March 2013 yet another banking crisis erupted in Cyprus, to which the ECB responded most notably by pushing the country to agree on an EU‐IMF package under the threat to cut its ELA funds. The Bank would use this leverage instrument again two years later with Greece, following the decision by the new government led by Alexis Tsipras to hold a national referendum on the terms of a third bail‐out agreement. In the meantime, in June 2014, the ECB announced new measures to tackle the Eurozone's liquidity problems, primarily the introduction of the Targeted LTROs (four‐year loans open to the non‐financial sector, renewed for a second round in March 2016) and a third round of CBPP. In the same month the ECB also applied, for the first time, negative interest rates.

On 22 January 2015 in a further move to counter the Eurozone's low inflation, the ECB announced its first programme of quantitative easing (QE), involving the purchase of €60 billion worth of financial assets per month between March 2015 and September 2016, for a projected total of €1.1 trillion. In December 2015 the QE was extended to last until March 2017, and expanded by an additional €360 billion, raising the monthly purchases to €80 billion. A second extension came in December 2016, with the doubling of the QE's original amount to €2.2 trillion, to be achieved by the end of 2017 (with the monthly purchases brought back to €60 billion).

Charisma and the ECB

The concept of ‘charisma’ was introduced in the study of politics and society by Max Weber, who borrowed it from the realm of religion. For Weber (Reference Weber, Roth and Wittich1978: 241), charisma is

Weber juxtaposes charismatic leadership and authority to two other forms of rule: traditional and rational‐legal. While the former is based on history and tradition, and the latter on laws, charismatic authority hinges on the ‘attitude of awe’ that the leader inspires among followers due to his (perceived) exceptional abilities (Spencer Reference Spencer1973). Implied in this definition is also the dependence of charismatic authority on performance, whereby obedience to the leader is withdrawn if the latter is seen to fail in his functions, or in any case he loses his aura of invincibility in the followers’ eyes (Weber Reference Weber, Roth and Wittich1978; also Tucker Reference Tucker1968).

In its ideal conception, charismatic authority is both an individual characteristic and a transitional state – often between traditional and modern legal systems – which should therefore disappear with the modernisation and rationalisation of society: ‘in its pure form charismatic authority may be said to exist only in statu nascendi. It cannot remain stable, but becomes either traditionalized or rationalized, or a combination of both’ (Weber Reference Weber, Roth and Wittich1978: 246). At the same time, Weber theorised the possibility of a ‘routinisation’ of charisma through, among other mechanisms, what he called ‘charisma of office’ – namely the transformation of charisma from a purely individual quality to an inherent attribute of a certain office (e.g., the papacy or a throne), so that the person holding that office acquires some dose of charisma just by virtue of his position (in addition, of course, to any charisma deriving from his personal characteristics).

While some later scholars have criticised what appears as a contradiction in Weber's theory (e.g., Tucker Reference Tucker1968), others have picked up and elaborated on these ideas. Shils (Reference Shils1965), for instance, has taken issue with historicist readings of charisma, and posited that the latter can and does exist in dispersed form in any rational‐legal social system. It is especially those institutions that are high up in the system that carry charisma because of their proximity to the ultimate values and principles ordering society and its members, and their consequent ability to defend and preserve such values, but also change them. In linking the metaphysical and the earthly order, Shils continues, these offices perform a function analogous to the connection between God and people that the charismatic leader performs in the original, religious meaning of the term.

Whether a product of individual traits or the consequence of an office, charismatic leadership is particularly likely to appear and thrive in situations of social crisis. Junctures in which common values are at stake, institutions are under pressure and unable to offer clear answers to people's needs, are fertile ground for the emergence of leaders able to respond to collective and individual anxieties by ‘mak[ing] sense of an incomprehensible reality’ and providing ‘guides for action and a promise for the future’ (Spencer Reference Spencer1973: 345). The charismatic leader, Tucker (Reference Tucker1968: 742) notes, ‘is one in whom … the promise or hope of salvation – deliverance from distress – appears to be embodied. He is a leader who convincingly offers himself to a group of people in distress as one peculiarly qualified to lead them out of their predicament.’

At the same time, situations of crisis make it easier for charismatic leadership to deliver on its promise by loosening societal and structural constraints, so expanding the space for leaders to break old organisational and institutional frames, and promote and eventually establish new ones (Pillai Reference Pillai1996). In this sense, charismatic leadership often overlaps with (although it remains distinct from) the idea of ‘transformational leadership’ (Bass Reference Bass1985; Yukl Reference Yukl1999).

More recently, organisational scholars have further explored the dynamics and mechanisms of charisma, particularly during crises, working from two different but complementary angles. Drawing mainly on management studies and social psychology, and on a large body of (often questionnaire‐based) empirical research, a number of scholars have spelled out the ways in which charisma plays out and affects social action in various settings. By presenting resonant ideas, articulating a vision and future objectives, and in some cases providing role models, charismatic leaders are able to respond to and stimulate certain follower needs in times of uncertainty, for example, by improving followers’ self‐esteem and self‐perception, increasing their sense of belonging and loyalty to a group, and providing templates for desirable behaviour. This in turn increases followers’ commitment to common goals and motivation for action and change (House Reference House, Hunt and Larson1977; Shamir et al. Reference Shamir, House and Arthur1993; Choi Reference Choi2006).

Another strand of scholarship has elaborated on the coordinative aspects of charismatic leadership, building on insights from economics, political communication and game theory. Social crises are here depicted as situations akin to Rousseauian stag hunt games, in which the collapse of old frames of reference and behavioural guides leaves the members of a group uncertain about each other's preferences and hence unable to coordinate for collective action. Charismatic leadership is able to break this strategic uncertainty by means of ‘values‐based, symbolic, and emotion‐laden’ signaling (Antonakis et al. Reference Antonakis, Bastardoz, Jacquart and Shamir2016: 304). Signals have a number of functions: they indicate that the (potential) leader is willing to act as the first mover to break the social impasse, they convey certain skills (so making leadership credible) and they indicate the direction of social action by transmitting a vision and commitment to a course of action. Needless to say, viewing charisma as signaling places great emphasis on the importance of communication – both verbal and non‐verbal – in shaping and strengthening leadership (Frese et al. Reference Frese, Beimel and Schoenborn2003; Antonakis et al. Reference Antonakis, Fenley and Liechti2011; Reference Antonakis, Bastardoz, Jacquart and Shamir2016; Reh et al. Reference Reh, Quaquebeke and Giessner2017).

Research design and empirical analysis

As stated at the outset of the article, the objective of our research is to establish if the crisis leadership of the ECB, and particularly its presidency, can be described as rhetorically charismatic. This broad research question in turn implies two causal sub‐questions: the first is whether the euro crisis has led the ECB presidency to use a more charismatic language; and the second is whether such a shift – should it be detected – has been mediated by individual factors. As mentioned in the foregoing, it is particularly under Mario Draghi's presidency that the ECB has taken the lead in crisis management and has been perceived, and often criticised, as politicised. Therefore, should an increase in the ECB's charismatic rhetoric in the crisis years be observed, we must ask whether this is the effect of the crisis itself on the holder of the ECB presidency office (no matter who this person is) or, on the contrary, just the result of Draghi's own personal characteristics.

To answer these questions, we have conducted a content analysis of the entire corpus of ECB presidential speeches from the inception of the Bank in 1998 to 31 December 2016.Footnote ² The corpus consists of 560 speeches, of which 129 are by the Bank's first president Wim Duisenberg, 312 by Jean‐Claude Trichet and 119 by Mario Draghi. While speeches are not the only way in which the ECB communicates, we have both theoretical and methodological reasons to focus on this particular corpus. In the first place, speeches are the ECB's main form of public communication, hence the place where the charismatic leadership of the Bank's presidency, if any, is most likely to be matched by verbal markers and signals. Second, and from a more methodological standpoint, speeches present the important advantages of accessibility, plenitude and relative standardization – all of which facilitate and strengthen empirical analysis and comparison.Footnote ³

To analyse the texts we have used the DICTION 7 software (Hart Reference Hart and West2001). Created specifically to analyse the tone of political discourse in speeches and other types of texts, DICTION codes text on 31 pre‐defined variables using in‐built dictionaries and assigns scores reflecting the presence and weight of dictionary terms. Generally speaking, dictionary‐based software analysis presents methodological pros and cons. Among the latter are difficulties in detecting nuances in language and differentiating between contexts. These shortcomings, however, are compensated by the software's ability to process large text corpora (as needed in our case) and its high levels of coding reliability.

Relying on a well‐established method (Bligh et al. Reference Bligh, Kohles and Meindl2004; Bligh & Robinson Reference Bligh and Robinson2010; Davis & Gardner Reference Davis and Gardner2012; Bastardoz et al. Reference Bastardoz, Jacquart and Antonakis2015; Olsson & Hammargård Reference Olsson and Hammargård2016), we have analysed the corpus of ECB speeches by combining DICTION variables into seven composite constructs, building on the linguistic characteristics of charisma listed in the previous section. Table 1 summarises the seven constructs and the corresponding formulas of DICTION variables, along with sample words for each variable and text examples extracted from various ECB presidential speeches.Footnote ⁴

Table 1.

The seven linguistic constructs of charisma

The constructs can be kept separate for a fine‐grained analysis of charisma's components or aggregated into a single index of charismatic rhetoric by subtracting the value of tangibility from the sum of the six remaining constructs.Footnote ⁵ Figure 1 presents a first overview of the charismatic content of the corpus under examination by plotting the year‐by‐year mean value of the speeches’ aggregate indicator of charisma. A first visual inspection reveals a general increase in the charismatic content of ECB speeches over time, and particularly in the past decade or so, in correspondence with the financial crisis years. Interestingly, there also seems to be a considerable drop in charismatic rhetoric in the past couple of years, which might be accompanying Europe's recent return to more stable and favourable economic conditions.

Figure 1.

The ECB presidency's charismatic rhetoric over time, 1998–2016.

The next step in answering our research questions is to assess the statistical significance of the observed increase in charismatic rhetoric from before to during the crisis period, so as to establish whether this variation is likely to reflect a true difference in the underlying populations. In the absence of statistical significance, the null hypothesis that the underlying populations are the same, and that the observed increase is due to sheer chance, cannot be safely discarded.

Broadly speaking, there are two ways in which such a test can be run. The first, following much of the quantitative literature on charismatic rhetoric (e.g., Bligh et al. Reference Bligh, Kohles and Meindl2004, Reference Bligh, Merolla, Schroedel and Gonzalez2010; Davis & Gardner Reference Davis and Gardner2012; Schroedel et al. Reference Schroedel, Bligh, Merolla and Gonzalez2013; Olsson & Hammargård Reference Olsson and Hammargård2016), relies on univariate and multivariate analyses of co‐variance (ANCOVA and MANCOVA), two techniques for testing the significance of mean differences in a dependent variable of interest (in our case, charismatic language) across samples, while also controlling for a number of factors (covariates) that might be related to that variable. In recent years a part of the charisma scholarship has moved away from the analysis of covariance to propose the regression discontinuity design (RDD) as an alternative method (e.g., Antonakis et al. Reference Antonakis, Bendahan, Jacquart and Lalive2010; Bastardoz et al. Reference Bastardoz, Jacquart and Antonakis2015; Steffens et al. Reference Steffens, Peters, Haslam and van Dick2017). Compared to ANCOVA/MANCOVA, the RDD allows the modeling of the assignment variable (in our case, time) in the regression model – thus tackling endogeneity problems resulting from the non‐random nature of the sample distribution – and concentrates on the significance of differences in the dependent variable around the cut‐off point between groups, rather than throughout samples (Lee & Lemieux Reference Lee and Lemieux2010).

While recognising the merits of the RDD, we believe that the analysis of covariance remains, on balance, a better method for our case, for two reasons. First, although our group selection method is clearly not random, we have no reason to expect time to have, in and of itself, an independent effect on the charismatic language of the ECB presidency. As such, the endogeneity that the RDD is designed to solve should not be too problematic an issue in our case. The second and more important reason has to do with the nature of the global and European financial crisis, which escapes the identification of a single, unequivocal starting point, but has instead unfolded more gradually over time. Accordingly, a method that treats the division between periods as more conventional than absolute and looks at the samples in their entirety is more appropriate than one placing most of the weight on the cut‐off point based on the expectation of a sudden change in the dependent variable (charismatic language). In the remainder of this section we therefore present analyses of covariance as our primary identification strategy and use the RDD as an auxiliary technique, aimed to check the robustness of our results.

For our tests we divide the period under examination into two sub‐periods, ‘pre‐crisis’ and ‘crisis’, placing the cut‐off point conventionally at 15 September 2008, the day in which the investment bank Lehman Brothers filed for bankruptcy, starting the chain of events making up the global financial crisis. As discussed above, this was also the time during which the ECB began to deploy the first ordinary and extraordinary instruments to tackle the effects of the crisis in Europe. Our division places 312 speeches in the pre‐crisis period and 248 in the crisis one.

As a first test, we performed a univariate analysis of covariance (ANCOVA) comparing the aggregate measure of the ECB's charismatic rhetoric (the dependent variable) across the two periods (the independent variable), including the total number of words in each speech as a covariate, to control for speech length.Footnote ⁶ As Table 2 shows, the test determines that the difference in the mean values of the aggregate index of charisma across periods is statistically significant (F(1, 557) = 20.221, p < 0.001). This confirms our general hypothesis that in the euro crisis years there has been an increase in the charismatic rhetoric of the ECB presidency.

Table 2.

The euro crisis and the ECB presidency's charismatic rhetoric: Aggregate results

Notes: Pre‐crisis period N = 312; crisis period N = 248. ^***p < 0.001.

To add a more fine‐grained examination of these results, we have then looked into the seven linguistic constructs taken individually by means of a one‐way multivariate analysis of covariance (MANCOVA) on the adjusted mean value of each construct in the two periods. As in the previous test, the total number of words was included as a covariate. The multivariate test confirms the results we obtained for the aggregate charisma index, attesting significant differences between the two time‐periods (Wilks's Lambda = 0.840, F (7, 551) = 14.994, p < 0.001). As Table 3 shows, the test determined that four out of the seven constructs – Collective focus, Temporal orientation, Adversity and Tangibility – changed in the expected direction as a result of the crisis (i.e., the first four increased, while Tangibility decreased). Conversely, results for Action, Followers’ worth and Similarity to followers show no statistical significance, which means we cannot exclude that the observed variation is due to chance. At least as far as the last two constructs are concerned, a preliminary explanation – which, however, would require further testing – might be that because the ECB is not an elected body, the rhetorical aspects that relate most closely to its relationship with the European citizenry (the ‘followers’) are less important in its charismatic discourse than it might be the case for a traditional political actor or institution.Footnote ⁷

Table 3.

The euro crisis and the ECB presidency's charismatic rhetoric: Results by single construct

Notes: Pre‐crisis period N = 312; crisis period N = 248. ^**p < 0.01; ^***p < 0.001; ns = non‐significant value.

As already noted, the above tests by themselves do not distinguish between the direct effect of the euro crisis on the charismatic rhetoric of the ECB presidency and the effect that might have been produced by individual factors, regardless of the crisis. More precisely, the analysis does not tell us to what extent the observed increase in charismatic rhetoric might be an independent effect of the personal characteristics of Mario Draghi, who has held the ECB presidency for over five of the roughly eight years that we have classified as the crisis period. The lag between the beginning of the crisis and the Draghi presidency, however, allows us to separate individual dynamics from the effects of the crisis itself, in order to answer this question. We have done so in two ways. First, we have isolated the presidency of Jean‐Claude Trichet (1 November 2003 to 31 October 2011) and run a one‐way ANCOVA with the same characteristics as above but limited to the corpus of his speeches, dividing them between those held in the pre‐crisis period (183) and those held in the crisis years (129). The test results, summarised in Table 4, show a statistically significant increase in the charismatic rhetoric of Trichet's speeches in the crisis period (F(1, 309) = 9.875, p < 0.01). In other words, the effect of the crisis on the ECB president's charismatic language remains even once we control for personal factors by holding them constant.

Table 4.

The euro crisis and the ECB presidency's charismatic rhetoric: Trichet's speeches

Notes: Trichet pre‐crisis period N = 183; Trichet crisis period N = 129. **p < 0.01.

We, then, have run a further mirroring test by isolating all presidential speeches held during the crisis period, and checking for the significance of the difference between those by Draghi (119) and those by Trichet (129), so as to hold the crisis factor constant (thus controlling for it) and focus on the effect, if any, of individual characteristics. As Table 5 shows, there is no significant difference between the charismatic content of the two presidents’ corpora of crisis speeches. This rather surprising result shows that, at least linguistically speaking, Mario Draghi cannot be said to be a more charismatic president than Jean‐Claude Trichet. Moreover, it reinforces our finding of a relationship between the euro crisis and the charismatic rhetoric of the ECB presidency.

Table 5.

Individual characteristics and the ECB presidency's charismatic rhetoric: Trichet and Draghi

Notes: Trichet crisis period N = 129; Draghi crisis period N = 119. ns = non‐significant value.

Like before, for this test we have looked more closely at the seven individual constructs so as to establish whether the aggregate picture shown above hides some interesting fine‐grained rhetorical differences between Draghi and Trichet – put differently, whether the two presidents have been linguistically charismatic in different ways as a result of the crisis. Table 6 shows the results of a MANCOVA run on the mean values of the seven individual constructs from the crisis speeches of Draghi and Trichet. Interestingly, the analysis confirms the overall similarity between the rhetoric of the two presidents: while some values are higher for Draghi and others for Trichet, none of these differences is statistically significant, with the sole exception of Temporal orientation, which is higher and statistically significant for Draghi.

Table 6.

Individual characteristics and the ECB presidency's charismatic rhetoric: Trichet and Draghi – results by single construct

Notes: Trichet crisis period N = 129; Draghi crisis period N = 119. ***p < 0.001; ns = non‐significant value.

Robustness check: Regression discontinuity design

As discussed above, by placing the emphasis on the cut‐off point between the pre‐crisis and the crisis period, the regression discontinuity design is better suited to analysing the effects of sudden and unequivocal crisis events (such as a natural disaster or a terrorist attack). Given the different configuration of the global and European financial crises, we do not expect our results to be perfectly replicable by means of a RDD. However, if our results are robust enough, they should hold, at least in part, with an alternative model specification.

Following the same setup as above, we have first tested the effect of the crisis on the aggregate value of the ECB's charismatic language at the cut‐off point of 15 September 2008, modelling our regression to account for the interaction between time and the time‐centred grouping factor (‘pre‐crisis’ and ‘crisis’). Accordingly, we included three independent variables: period, time and the interaction term between period and time (see also Bastardoz et al. Reference Bastardoz, Jacquart and Antonakis2015), where period is a binary variable with a value of 0 and 1, and time is the number of days separating the date of each speech from the cut‐off date. The total numbers of words of the speeches was included as a covariate to control for speech length. Finally, we added a categorical variable, president, to control for any effect of the individual characteristics of the three ECB presidents. The regression is presented visually in Figure 2.

Figure 2.

The euro crisis and the ECB presidency's charismatic rhetoric: Regression discontinuity plot.

On the whole, the RDD results are consistent with those of the analyses of covariance presented above: as Table 7 shows, the level of charismatic rhetoric of the ECB presidency displays a statistically significant increase at the crisis cut‐off point (β = 14.240, p < 0.05), while the identity of the president does not have any statistically significant effect on charismatic language. Furthermore, the RDD shows that time, whether by itself or interacted with period, is not correlated with the dependent variable, confirming our initial expectations.Footnote ⁸

Table 7.

RDD estimates for charisma and individual constructs (SUR)

Notes: N = 560. Standard errors in parentheses. *p < 0.05; **p < 0.01; ***p < 0.001.

Next, we performed an analogous test on the seven charismatic constructs taken individually by means of a seemingly unrelated regression (SUR) to keep track of the disturbances between the different dependent variables (Bastardoz et al. Reference Bastardoz, Jacquart and Antonakis2015). For this operation we relied on the systemfit package for R (Henningsen & Hamann Reference Henningsen and Hamann2007). As shown in Table 7, when disaggregated, our results are not perfectly aligned with the previous ones: of the four constructs on which the crisis had a significant effects in the MANCOVA, only two – Collective focus and Adversity – remain significant in the RDD.

Read together with the outcome of the aggregate regression, these results indicate that, to the extent that we are looking for an immediate effect of the global financial crisis (represented here by the collapse of Lehman Brothers) on the ECB presidency's charismatic language, this effect is still to be found, but is carried mostly by the change in two of the seven components of charisma. The remaining two constructs, on the other hand, have moved more gradually over time, so escaping significance when tested through the RDD.

All things being considered, the fact that most of our ANCOVA/MANCOVA analysis (and in particular the aggregate‐level one) holds when replicated with the different specification of the RDD is, we believe, a good indicator of the robustness of our results.

Discussion and conclusion

To recap, our content analysis of the ECB presidential speeches corpus has found a solid correlation between the euro crisis and the charismatic rhetoric of the Bank's presidency, as measured through the aggregation of DICTION's seven linguistic constructs. The correlation holds well when the analysis is conducted on the individual constructs, four of which show statistically significant variation from the pre‐crisis to the crisis period. Finally, by means of two additional analyses controlling, respectively, for individual factors and period, we have been able to rule out any significant mediation of the effect of the euro crisis on the ECB presidency's charismatic rhetoric on the part of President Mario Draghi's personal characteristics. The overall consistency between our ANCOVA/MANCOVA analysis and the outcome of the alternative RDD specification lends additional support to the robustness of our results.

Once again, we should be clear about the limitations of this study, which has focused on a particular aspect of charismatic leadership – namely language – and has examined the ECB presidency's charismatic rhetoric via the analysis of a specific type of communication embodied by presidential speeches. That said, given the key role of public communication in the dynamics of charismatic leadership, the reliability of language as an indicator of charisma, and finally the robustness of our empirical results, we feel confident in concluding that our study supports the hypothesis that the euro crisis has led to the emergence of the ECB presidency as a charismatic euro leader.

These results in turn suggest that the EU has conformed to the expectations of charismatic leadership theory, but has done so in an unusual way, for the charismatic response triggered by the crisis appears to have taken place in what is at least formally a technocratic body rather than among political actors – in the first place, supranational ones. In this respect our study must be read together with Olsson and Hammargård's (Reference Olsson and Hammargård2016) analysis of the public rhetoric of the European Commission between 2006 and 2011, which has found a negative correlation between the severity of the financial crisis and the deployment of charismatic language on the part of the Commission president, reflecting, more generally, the latter's fast retreat from a leadership role in tackling the euro crisis (for a more qualitative take, see also Pansardi & Battegazzorre Reference Pansardi and Battegazzorre2016).

Our findings have a number of analytical and normative implications for the governance of the EU. Analytically, this study confirms the trend of ‘politicisation’ of the ECB that has been observed time and again in recent scholarly and journalistic accounts of the Bank's anti‐crisis measures (e.g., Scicluna Reference Scicluna2014; Fleming Reference Fleming2015; Wyplosz Reference Wyplosz2015; Högenauer & Howarth Reference Högenauer and Howarth2016). Unlike most of these accounts, however, we do not regard politicisation necessarily as a negative phenomenon. Instead, we see it more neutrally as a process whereby the ECB has stepped up its own role in the context of, and as a response to, a paucity of crisis leadership among other players. This process in turn can be interpreted negatively as the ECB's opportunistic overstepping of its limits and mandate, or positively as an act of responsibility and guardianship vis‐à‐vis the Eurozone (and the EU as a whole) as the latter's future seemed increasingly at risk (on this, see, e.g., Torres Reference Torres2013). At a minimum, this alternative interpretation of politicisation as charismatic leadership should be added to the existing ones to further enrich the political and academic debate on this topic – not least because it lends itself to being tracked empirically through language, as we have done here, so mitigating the ‘essentially contested nature’ of politicisation as a concept (Gallie Reference Gallie1956).

Our analysis of the ECB as a rhetorically charismatic actor is also important normatively. The foremost problem linked to the political role of the ECB is its lack of democratic legitimacy. The Bank's highly independent setup, the argument goes, has a counterweight in the strict mandate – to ensure price stability – given to it by the Treaties. As soon as the ECB starts exceeding the limits of its mandate (or at any rate is perceived to do so) to influence the Eurozone's economic policy beyond its monetary policy mission, the question is immediately raised of ‘who elected the ECB’ to take such initiatives. In Weberian terms, this can be formulated as a problem of rational‐legal authority: when acting outside of its statutory powers, the Bank loses, by definition, the authority conferred to it by the Treaties without being able to make up for it with the sort of authority deriving from an electoral mandate (or, more realistically, from acting on the instructions of another body endowed with such a mandate). However, to the extent that charisma is considered an alternative basis of legitimacy, along the lines suggested by Weber, the problem posed by the ECB's lack of rational‐legal authority is mitigated, at least in part. Put differently, as long as the ECB presents itself, acts and is perceived as the saviour of the Eurozone and the EU more generally, its deficit of democratic accountability might be regarded as less of a problem. In this article we have explored empirically the specific communicative side of charisma as a source of authority. Future analytical and normative work on the ECB should expand on these ideas by looking at different aspects of charisma and, more generally, examining the question of alternative bases of political legitimacy in times of crisis.

Yet charismatic authority is very dependent on the performance of the charismatic leader. In this sense, the ECB's legitimation through charisma is intertwined with the widespread notion, illustrated in the introduction to this article, that the Bank so far has been the main bulwark against a breakup of the euro. Should this cease to be the case – either for exogenous reasons, such as a new deterioration of the European economy, or following changes in the ECB's own conduct and policies, such as the future phasing out of the quantitative easing programme – the democratic legitimacy problem will come back in full force and call for a more stable solution in the form of a retreat of the Bank to a purely technocratic role or, alternatively, a radical restructuring of its governance architecture to connect it more closely to the Eurozone's political authorities, ultimately increasing the ECB's democratic accountability.

Acknowledgements

This article was written as part of the project ‘REScEU – Reconciling Economic and Social Europe’, funded by the European Research Council (advanced grant no. 340534) and led by Maurizio Ferrera, whom we wish thank for his support and feedback. We are also grateful to Rulof Burger and Alessandro Pellegata for precious suggestions on the statistical analysis, and Nicola Borri, Anna Viterbo, Cornel Ban, Josef Hien, Furio Stamati and three anonymous reviewers for helpful comments.