Tracing leadership: the ECB’s ‘whatever it takes’ and Germany in the Ukraine crisis

Abstract

Political scientists face problems when assessing a leader’s impact: how can we know that a policy outcome or institutional change is caused by leadership? This article argues that in addition to relying on comparisons and counterfactuals, we need to trace the causal mechanisms by which leadership affects outcomes. Therefore, the article proposes a way to trace leadership and applies it to two cases of EU crisis management: the European Central Bank’s role in announcing Outright Monetary Transactions in the eurozone crisis, and Germany’s role in shaping the EU’s response to the Ukraine crisis. Systematic process-tracing shows that both actors provided leadership ‘by default’. However, while the ECB had to combine the provision of knowledge with unilateral action in order to overcome the eurozone crisis, Germany could use manifold bargaining-based strategies and thus became the EU’s de facto agenda setter and main representative in managing the Ukraine crisis.

Keywords:

• Leadership
• process-tracing
• causal mechanism
• European Union
• eurozone crisis
• Ukraine crisis

Leadership studies are an important subdiscipline of political science. Yet when it comes to determining a leader’s impact, political scientists still face methodological difficulties: how can we know that a policy outcome or institutional change is caused by leadership? So far many political scientists have assessed a leader’s impact by making recourse to comparisons or counterfactuals (e.g. Beach and Mazzucelli 2007; Elgie 1995; Helms 2012a). Both methods yield plausible arguments as to why a certain outcome depends on the respective actor in charge. However, they cannot reveal the supposed causal mechanism linking the leader’s actions to the outcome. For this we need process-tracing.

This article suggests a way to trace the process resulting in a leader’s impact. Furthermore, it illustrates the added value of this approach by applying it to two prominent instances of EU crisis management: the European Central Bank’s (ECB) announcement of Outright Monetary Transactions (‘whatever it takes’) in eurozone crisis management, and the role of Germany in shaping a common EU response to the Ukraine crisis and mediating the Russia‒Ukraine conflict. Hence, the article’s contribution is twofold. On a theoretical level, it puts forward a conjecture and a causal mechanism on how leaders influence policy or institutional change. On an empirical level, the article draws on 26 semi-structured elite interviews to offer insights into the motivations and strategies of powerful actors in EU crisis management.

The article first outlines the comparative approach to assessing a leader’s impact and identifies its drawbacks. In a second step, it proposes systematic process-tracing as a complement to comparisons and counterfactuals by developing a theoretical conjecture and a causal mechanism on how leaders influence outcomes. The third step consists in the empirical application.

How to assess a leader’s impact?

In 1978, James MacGregor Burns claimed that ‘leadership is one of the most observed and least understood phenomena on earth’ (Burns 1978: 2). Forty years later, research on political leadership is still under-theorised (Hartley and Benington 2011: 211). This concerns also the question of whether observable outcomes are really caused by a leader, or whether they would also have come about in the absence of any leadership (e.g. Jervis 2013). With regard to EU negotiations, for example, Moravcsik (1998, 1999a, 1999b) considers the provision of leadership redundant due to almost symmetrical access to information and ‘low’ bargaining costs relative to the gains of integration. Beach (2005) and Tallberg (2006), however, come to quite different conclusions on the same question.

Burns defined leadership as a power-based process targeted at the achievement of ‘reciprocal betterment or … real change’ (Burns 1978: 434). Also other seminal definitions of leadership, such as those of Kindleberger (1981: 243), Young (1991: 285), Underdal (1994: 178) or Nye (2010: 306) agree that leaders help a group overcome collective problems. The article builds on these definitions by understanding leadership as a process in which an actor influences a group’s activities in such a way as to achieve an institutional or policy change to the benefit of the entire group (see Avery 2004: 22; Schoeller 2019: Ch. 2).

In order to find out whether a given policy or institutional change has really been brought about by a leader, many leadership studies make implicit use of process-tracing (for an overview, see Schoeller 2019: Ch. 2). However, when it comes to the explicit employment of methods, existing research has mainly relied on comparisons and counterfactuals (Jervis 2013: 158; see e.g. Beach and Mazzucelli 2007; Tömmel 2019). In this article, I therefore argue that the explicit application of ‘theory-testing process-tracing’ (Beach and Pedersen 2013: 56–60) constitutes a valuable innovation for the study of leadership. Before I elaborate on the advantages of process-tracing, I briefly present some drawbacks inherent to comparisons and counterfactuals.

If researchers seek to assess a leader’s impact, they may examine whether two situations bring about different outcomes if the only major variable that changes is the leader in charge (e.g. Helms 2012b: 6–14). Such comparisons can deliver compelling evidence for a leader’s impact if most other situational circumstances are controlled for. However, situations in which only the leader changes, while all other circumstances remain the same, are fairly rare. One may even argue that it is impossible to find different leaders in identical situations because actors are a constituent part of their environment: whenever the leader changes, the situation also changes. If two situations are not identical, however, a comparison cannot provide certainty that a given outcome is caused by the leader. Next to synchronic comparisons, one can examine different leaders over time (Helms 2012b: 13; Jervis 2013: 158–60). Still, besides potential problems of endogeneity, these diachronic comparisons rely on counterfactual reasoning as the researcher needs to speculate what would have happened if there had been no change of leader.

Due to a lack of cases that allow for controlled comparisons, leadership scholars are often forced to rely on thought experiments (Helms 2012b: 8). Counterfactual reasoning can yield plausible arguments as to why an outcome would be different (or not), had there been no actor providing leadership. However, as we can never know what would have happened if there had been (no) leadership in a given case, counterfactuals are not sufficient for assessing a leader’s impact empirically.

Most importantly, both comparisons and counterfactuals are based on logic of congruence, as the researcher correlates a given outcome (policy or institutional change) with the presence of a potential leader. This implies that we see only the alleged cause and the outcome, but we do not have any evidence pointing to a causal relationship between the two. A noted congruence could therefore be spurious. Thus, in order to find out whether a given outcome is really the result of leadership, we need to study causal mechanisms (e.g. George and Bennett 2005: 201).

Tracing a leader’s impact

When tracing a leader’s impact, we know ex ante about both institutional or policy change (outcome) and a potential leader in charge (presumed cause). What we need to uncover is the causal mechanism linking the alleged leader to the outcome. These are typical criteria for ‘theory-testing process-tracing’ (Beach and Pedersen 2013: 11–22): In order to elaborate and test a causal mechanism, we first need a theoretical conjecture on how an outcome may be related to a leader. Second, we need to explicate single steps of the expected causal mechanism as well as their observable manifestations. Third, we should determine what kind of inference we can draw from the observables (Van Evera 1997: 30–4). In the following, I will go through these three steps to model a causal mechanism based on a rationalist power-based approach to political leadership.

Theoretical conjecture

From a rationalist perspective, leadership is a response to collective action problems¹ that are not regulated by institutions (Beach 2005: Ch. 2; Schoeller 2019: Ch. 3; Tallberg 2006: Ch. 2). These situations cause a demand for leadership to improve the efficiency of collective action or overcome the problem (= common goal). If leaders translate their power resources into strategies, they can help a group deal with its collective action problems (Nye 2010: 306; Underdal 1994: 178f.; Young 1991). By using privileged information and/or institutional prerogatives (resources) to set the agenda (strategy), for example, a leader may bring about agreements that otherwise are prevented by imperfect communication, high transaction costs, or free-rider dilemmas. By investing material resources into package deals, by contrast, a leader may compensate potential losers of a deal (strategy) and thus overcome distributive impediments. The online appendix provides a detailed description of leadership strategies, required resources and observable manifestations.

Hence, power resources are a necessary condition for exercising leadership (Ahlquist and Levi 2011: 5; Burns 1978: 434; Endo 1999: 16–26). What connects them to an outcome that is collectively superior to the default outcome resulting from the respective constellation of preferences and fall-back options (see Moravcsik 1998) is the use of leadership strategies. While I have to refer the reader elsewhere for a more detailed elaboration of this argument (Schoeller 2019: Ch. 3), we can state here the following conjecture: if there are actors in a position of power, who translate their resources into strategies, they may bring about policy or institutional change to the benefit of the entire group (common goal).

This process requires additional steps, though. First, the actors need to have an interest in deploying their resources (Frohlich et al. 1971: 7). Second, in most cases the would-be leaders need to signal their leadership offer or claim by explicating where they want to bring the others, and why this is a common goal. At this point, leaders normally face opposition: if all actors had not only a common goal, but even the same preferences on how to reach it, they could simply move on without anyone taking the lead (see Kindleberger 1981: 243). Thus, in order to overcome opposition, leaders need to change the strategic preferences (positions) of followers by using strategies. In the following sub-section, I will integrate these single steps into a causal mechanism and explicate their observable manifestations and strength of evidence.

Causal mechanism and observable manifestations

In line with the theoretical argument formulated above, the presumed cause for a policy or institutional change is an actor in a position of power. Power resources can be differentiated into material, institutional and non-material resources (see Schild 2019). Material resources comprise military and economic capabilities. Institutional resources are rights of agenda setting, decision making, implementation and evaluation. Non-material resources include privileged information, expertise, reputation and credibility (e.g. Beach 2005: 27–9).

The causal mechanism leading from an actor in a position of power to institutional or policy change starts with the would-be leaders’ interest in deploying their resources. This may be expressed in the form of strategy papers or public speeches, but also ex post through confidential face-to-face interviews (for the particular observables, see Table 1 and online appendix). However, as any actor can express such an interest without actually taking the lead, such statements provide a low level of ‘uniqueness’: finding this evidence does not necessarily mean that our conjecture is right. Moreover, actors that take the lead may conceal their real motivation. Therefore, such evidence also offers only a low level of ‘certitude’: even if we do not find it, our conjecture may be right. Hence, the first step of our causal mechanism qualifies only for a so-called ‘straw in the wind’ test: relevant statements affirm the plausibility of our conjecture, but neither confirm nor disconfirm it (Beach and Pedersen 2013: 102; Van Evera 1997: 32; see online appendix for the different test types).

Table 1. Causal mechanism.

	X	Causal mechanism (leadership)				Y
Theoretical conjecture	Actor in a position of power	Interest to deploy resources	Offer or claim for leadership	Use of strategies	Change of reluctant actors’ positions	Policy or institutional change
Empirical manifestations	Resources relative to other actors: (1) material:  (a) economic  (b) military capacities (2) institutional: procedural rights (3) non-material:  (a) information  (b) reputation	Written or oral expressions of interest	Communications indicating the intention to invest resources into reaching a common goal	Agenda-management Provision of focal point Arena-shifting Arena-linking Coalition-building Pre-negotiations Unilateral action ‘Leading by example’ Representation Provision of common knowledge	Ex post approval of leader’s decision or change of respective statements	Difference between ex ante and ex post situation
Observables (see online appendix for details)	(1a) GDP, current account balance, public debt and deficit; financial means (1b) Troop strength, nuclear power (2) legal texts, written rules of procedure (3a) issue briefs, position papers, number of specialised staff (3b) subjective statements (e.g. interviews, press, declarations)	Strategy papers, face-to-face interviews, speeches	Speeches, press releases, (government) declarations, open letters	Depending on strategy, see online appendix	Public statements, strategy papers, common conclusions, interview statements	Legislative texts, legal acts, rules of procedure, treaties, international or inter-institutional agreements, practices
Test strength		Straw-in-the-wind test	Straw-in-the-wind test	Hoop test	Straw-in-the-wind test

Source: own illustration (inspired by Beach and Pedersen 2013).

The second step of the causal mechanism consists in an offer or claim for leadership. This can be observed through open declarations indicating the leader’s intention to reach a common goal, such as speeches, press releases or open letters. Taken by themselves, such statements do not necessarily mean that the outcome was caused by that actor (‘low uniqueness’). Moreover, an actor may become a leader even without having claimed that role explicitly (‘low certitude’). Hence, evidence for a leadership offer or claim qualifies only as a ‘straw in the wind’ test.

In order to convince followers to adopt a proposed institutional or policy change, a leader needs to employ strategies. They include agenda management, arena shifting and linking, coalition building, pre-negotiations, unilateral action, ‘leading by example’ and the provision of common knowledge (Table 1; see online appendix and Table A1 for a detailed explication and observable manifestations). While it is true that other actors may also use these strategies (‘low uniqueness’), no actor can provide leadership without making use of some kind of strategy (‘high certainty’). Hence, this part of the causal mechanism qualifies for a ‘hoop test’: if we do not find evidence for the use of strategies, we cannot assume that the outcome was caused by a leader (Beach and Pedersen 2013: 102f.; Van Evera 1997: 31).

If leaders employ their strategies successfully, we should observe that at least some reluctant actors change their positions. This may become visible through an ex post approval of the leader’s decision or a change in the respective statements (e.g. interviews, common conclusions, strategy papers). As actors can also withhold their opinion, or – vice versa – change their positions independently from a leader’s action, such evidence provides neither a high level of ‘uniqueness’ nor ‘certitude’, but it increases the plausibility that leadership has taken place (‘straw in the wind’).

Table 1 illustrates the entire causal mechanism. None of the observable manifestations alone is sufficient for claiming that an outcome was caused by a leader (‘smoking gun’). Yet evidence for multiple straw-in-the-wind tests also strongly corroborates a conjecture (Collier 2011: 826). Thus, if we find evidence in favour of all single pieces of the causal mechanism, we have a quite unique signature indicating the provision of leadership. By contrast, if we do not find evidence indicating at least the use of strategies, it is implausible that a policy or institutional change was caused through leadership.

Empirical application

This section applies the proposed mechanism to two cases of EU crisis management. The first covers the ECB’s announcement of Outright Monetary Transactions (OMT), which is regarded as the decisive move that saved the Economic and Monetary Union (EMU) from collapse (De Grauwe and Ji 2015). The second case focuses on Germany’s role in the Ukraine crisis, since Germany was seen as the leading actor in shaping and representing the EU’s common position in the conflict (Fix 2016; Krotz and Maher 2016; Speck 2016).

While in both cases it has been argued that the respective actors provided leadership (e.g. Verdun 2017; Wright 2018), there has been no systematic analysis of the causal steps transforming their power and actions into outcomes. Given the plausibility of a causal link between X (actor in a position of power) and Y (policy or institutional change), the two cases constitute ‘typical cases’, which are best suited to either probe a causal mechanism or show ‘that there are no plausible causal mechanisms connecting this independent variable with this particular outcome’ (Seawright and Gerring 2008: 299; see Beach and Pedersen 2013: 146f.). In other words, ‘typical cases’ are particularly useful when congruence between an alleged cause and an outcome is established in the literature or ‘empirically known ex ante, but the causal mechanism is unknown and needs to be probed’ (Meissner and McKenzie 2018: 7). Substantively, both cases are highly relevant as they represent decisive instances of crisis management in the respective policy areas. The fact that the two cases differ in so many other aspects contributes to illustrating the general applicability of the proposed approach.

In addition to speeches and press releases, I draw on 26 semi-structured interviews with high-ranking officials from the European Commission, the Council of the European Union, two Permanent Representations in Brussels, the ECB, the German Ministries of Finance, Foreign Affairs, Economic Affairs and Energy and the Federal Chancellery. Elite interviews ‘potentially offer a more direct measure of a causal mechanism’ (Beach and Pedersen 2013: 134) if the interviewees actually participated in the case under investigation. At the same time, there is a risk that the respondents do not tell the ‘truth’. On the one hand, I addressed this problem when formulating the questionnaire: for example, I avoided questions implying social desirability or suggesting causal relationships. On the other hand, I triangulated the data with other sources: wherever possible, I checked for each step of the causal mechanism whether the interview data was congruent with the information obtained from public speeches or secondary literature. In order to further increase internal validity, I guaranteed the respondents strict confidentiality and selected interviewees who were actively involved in the relevant cases.

The ECB’s ‘whatever it takes’

When in July 2012 government bond spreads reached an unprecedented height and fuelled speculations about a collapse of the eurozone, ECB President Draghi declared that ‘the ECB is ready to do whatever it takes to preserve the euro’ (President of the ECB 2012). Following this statement, the ECB announced on 2 August that it ‘may undertake outright open market operations’ (ECB 2012). As a consequence, government bond spreads narrowed, which provided immediate stability to the eurozone (Altavilla et al. 2014).

The announcement of OMT represents a policy and an institutional change. First, OMT widened the ECB’s spectrum of policy instruments, as a programme providing for the unlimited purchase of government bonds was unprecedented (Salines et al. 2012: 668). Second, the announcement constitutes an informal institutional change as the ECB thereby became the eurozone’s de facto lender of last resort (De Grauwe and Ji 2015: 752).

Yet is the OMT announcement really an instance of leadership (see Verdun 2017)? If, for instance, the ECB acted without the intention to save the single currency, was forced by other actors, or announced OMT without convincing member states to ‘follow’, the suspension of the crisis would be an unintended side effect rather than leadership. Process-tracing can show whether the OMT announcement was actually an instance of leadership.

According to the proposed leadership conjecture, the cause for policy or institutional change is an actor in a position of power. As regards material resources, the ECB is extremely powerful as it enjoys ‘unlimited firepower’: being the only actor that can create money in the eurozone, the ECB’s material capacities exceed those of member states by far (Interview 14). If it comes to institutional resources, the ECB is the only authorised decision maker in EU monetary policy. In economic policy, by contrast, the ECB’s institutional resources do not include any formal rights of decision making, but it enjoys participation, consultation and monitoring powers (Interviews 21, 24; Beukers 2013: 1583–90; Art. 127.4 TFEU). With regard to non-material resources, the number of administrative staff working in the area of EMU can be used as a proxy for expertise. According to written answers from summer 2015, the ECB at the time employed 523 staff members, followed by the Commission (391) and Germany (117). This assessment is corroborated by statements of ECB officials, who attest that the ECB’s expertise allows it to influence member states’ policy decisions and even concrete legislation (Interviews 14, 21, 22, 24). In sum, at the time of the OMT announcement, ‘the ECB was the only actor within EMU that had the capacity to respond quickly and effectively in order to prevent a collapse of the Eurozone’ (Krampf 2016: 467).

This allows us to trace whether the ECB provided leadership when announcing OMT. The first part of the causal mechanism regards the actor’s interest. Most interviewees stated that the ECB’s interest in announcing OMT consisted primarily in restoring its own effectiveness and preserving the single currency (Interviews 2, 5, 11, 13–15, 18–24, 27). While the ECB’s institutional survival, which is bound to the survival of the euro, was thus the main benefit of announcing OMT, the ECB was also interested in increasing control over member states’ budgetary policies by making the use of OMT subject to a consolidation programme with the European Stability Mechanism (ESM) (Interviews 3, 11, 13, 14, 19). Hence, the ECB had a strong interest in taking the lead.

The second step of the causal mechanism is an offer or claim for leadership. Indeed, ECB President Draghi explained that it ‘is now time … for the ECB to … assume greater responsibility for the functioning, or even the survival, of the system’ (Euractiv 2012). But already the statement ‘to do whatever it takes to preserve the euro’ (President of the ECB 2012) defines any ensuing action as a contribution to the eurozone’s common goal. Although these statements qualify only as ‘straws in the wind’, they increase the plausibility that the ECB has purposefully taken the lead when announcing OMT.

As a third step, we expect the ECB to employ strategies convincing reluctant ‘followers’. While the ECB is formally independent, the announcement would only be credible with the support of member states: since the ECB can conduct OMT only in countries that agreed on an ESM programme, member states can prevent OMT by not signing such a programme. Moreover, a lack of member state support would damage the ECB’s credibility in the financial markets and thus its effectiveness (Interviews 8, 14). Therefore, the ECB prepared the OMT announcement through exchanges of views with member states and EU institutions. More precisely, ECB officials explained monetary policy and macroeconomics both publicly and in all relevant EU fora, thereby defining possible solutions and advocating their preferred options (Beukers 2013: 1581). In particular ECB President Draghi was ‘pro-active’ (Interview 2) in communicating with eurozone governments and the Presidents of the European Council, Eurogroup and Commission (Interviews 1, 2, 10, 22, 25–7). This was regarding especially the German government: ‘Mr Draghi worked informally with Ms Merkel, carefully testing what might be acceptable’ (Spiegel 2014; also Blackstone and Walker 2012). In doing so, the ECB provided ‘full information’ (Interview 24) on possible actions and outcomes, framed the problem, and defined possible solutions (Interviews 1, 2, 21, 24). Declaring to governments that in order to save the single currency, either the governments themselves or the ECB would have to act as a lender of last resort, made OMT acceptable (see Bastasin 2015: 397). However, the ECB did not disclose its concrete plans or when to launch them.

The provision of common knowledge was a precondition for obtaining the reluctant member states’ approval. Only after the ECB had informed member state governments about the options for action and their consequences did it move on to unilateral action. As a top official in the German Finance Ministry made it clear, by announcing OMT the ECB took a unilateral decision which actually should have been taken in the ECONFIN Council (Interview 27). An ECB official explained this approach: ‘There were only two alternatives: Either the ECB appearing as lender of last resort to governments or Germany moving on towards Eurobonds or Redemption Fund or something to stop the crisis’ (Interview 22). After the ECB had unilaterally decided to appear as lender of last resort by announcing it would ‘do whatever it takes’, reluctant member states had no choice but to back the ECB’s decision (Blackstone and Walker 2012). If they had protested against the ECB’s announcement, they would have re-fuelled speculations against the euro, increasing the risk that the ECB had to act on its promise and buy government bonds. This would have been the worst outcome possible from the perspective of the reluctant creditor states. They understood this because the ECB had explained it beforehand (see above; Interview 2). Hence, the ECB first provided common knowledge and then resorted to unilateral action to make the OMT announcement credible and effective. Finding this piece of evidence was necessary for claiming that the ECB provided leadership in announcing OMT (‘hoop test’).

If it is true that the ECB provided leadership, we should find evidence that reluctant states eventually changed their positions. Some governments like Germany, Austria and the Netherlands were initially against the ECB’s announcement (Interviews 1, 2, 13, 26). However, ‘those who were sceptical decided not to be vocal’ (Interview 2). This can be seen most clearly in the case of Germany (Interviews 10, 11, 13, 26). While it welcomed the signal effect of the announcement, the German government fought against the actual activation of OMT behind closed doors (Interviews 25–27). After the Governing Council’s decision of 2 August, German officials clearly conveyed in all fora that an activation of OMT ‘would not be welcome and that the ECB would be well advised not to go ahead’ (Interview 10). In public, however, Germany changed its sceptical attitude. Both Chancellor Merkel and Finance Minister Schäuble welcomed the ECB’s decision even against the resistance of the powerful Bundesbank. In the words of Bastasin: ‘Berlin had broken with the Bundesbank. Draghi had the cover he wanted’ (Bastasin 2015: 399; see also Blackstone and Walker 2012: 4f.).

Taking the single pieces of evidence together, we can plausibly affirm that the OMT announcement was an instance of leadership. The ECB’s position of power, its interest to deploy its resources, the explicit claim for leadership, the use of strategies, and finally the reluctant actors’ change of preferences corroborate this claim.

Germany and the Ukraine crisis

When in 2014 the Ukraine crisis escalated to violence, Russia annexed Crimea and a war broke out in Donbass, observers noted that ‘Germany decided to play a leadership role’ (Speck 2016: 4; see Fix 2016; Krotz and Maher 2016: 1058f.; Matthijs 2016: 146–8; Wright 2018). Considering the EU’s common goal in this conflict, namely to maintain stability at its borders, this was a challenge: the EU needed to contain Russia’s hegemonic aspirations in Eastern Europe and defend the rule-based European peace order while preventing the conflict from escalating into an open war.

Accordingly, the observable outcomes are twofold. First, the EU came up with a homogeneous response towards Russia, which consisted in early and consistent sanctions.² This was a unanimous policy change in the relations towards Russia (Interviews U2, U3, U5; Speck 2016: 1). Second, the Minsk agreement of 12 February 2015 (‘Minsk II’) represented an institutional change in the conflict between Russia and Ukraine. Although it was at best a ‘brittle ceasefire’ (Matthijs 2016: 148), it helped prevent the conflict from escalating into open war (Interview U6; Speck 2016: 3). Moreover, only through Minsk II did Russia accept the role of a ‘conflict party’, whereas before it insisted on being a ‘mediator’ (Interview U6). However, is the homogenous EU position and Minsk II really a result of German leadership?

The conjectured cause for a leader-initiated outcome is an actor’s power position. To start with Germany’s material resources, the country is no nuclear power and lacks a strong army, but it is the EU’s largest economy, export champion and main creditor (Matthijs 2016: 141). In terms of institutional resources, Germany does not stand out from other member states. In the EU’s common foreign and security policy (CFSP), decisions are made unanimously and agenda management takes place in EU bodies such as the European External Action Service, the Political and Security Committee and COREPER. Moreover, the High Representative (HR) is responsible for representing and coordinating the EU’s external action. As regards Germany’s non-material resources, the country’s reputation as a ‘civilian power’ provides it with elevated credibility (Matthijs 2016: 142f.).

Although Germany’s economic and non-material resources may partly compensate for its military weakness, it is not an ideal leadership candidate in CFSP. Compared to other potential candidates, however, Germany may become a ‘leader by default’ (Helwig 2016: 27). In the specific case, the UK was unavailable because of its retrenchment in EU affairs and difficult relationship to Russia (Interview U1), France did not claim leadership (Interview U4) and the EU institutions (Commission and HR) ‘failed’ to provide leadership (Interview U5; Krotz and Maher 2016: 1058–62). Hence, Germany was best placed to take the lead in the Ukraine crisis.

The first step of the causal mechanism consists in a leader’s interest to deploy resources. Being an economic power with only small military capabilities, Germany’s prosperity depends eminently on a liberal and rule-based order. As for instance the German Minister of Defence explained: ‘As a major economy and a country of significant size we have a strong interest in international peace and stability’ (Von der Leyen 2014). Hence, Germany had a particular interest in defending the European peace order against Russia’s hegemonic aspirations (Interview U2). At the same time, it was essential to avoid an open war in the EU’s neighbourhood, which could imply spill-over effects to Germany (Interviews U3, U5, U6). As the German chancellor made it clear a few days before she negotiated ‘Minsk II’ together with her French, Russian and Ukrainian counterparts: ‘None of us is interested in … a confrontation with the risk of uncontrollable escalation’. She linked Germany’s ‘long-term goal of a common economic area from Vladivostok via Lisbon to Vancouver’ to ‘overcoming the Ukraine crisis on the basis of international law’³ (Merkel 2015). In particular, when the US Congress issued a resolution to deliver arms to Ukraine, Germany’s fear of an uncontrollable escalation increased. To prevent the USA from putting these plans into practice, the German government conceived that a renewed ceasefire together with a concrete implementation plan (later ‘Minsk II’) was in its immediate interest (Interview U6). In sum, Germany was not only interested in deterring Russia through sanctions, but also had a particular interest in peaceful conflict management. This increases the plausibility that Germany provided leadership in the Ukraine crisis (‘straw in the wind’).

With regard to the second step of the proposed mechanism – a claim for leadership – German Foreign Minister Steinmeier declared that Germany ‘wants to be and will be a driving force for a common European foreign, security and defence policy’.⁴ Referring to concrete cases, he also mentioned the Ukraine conflict, for which ‘there must not be a violent solution’⁵ (Steinmeier 2014). More concretely, in a speech entitled ‘Leadership from the Centre’, Minister of Defence Von der Leyen not only expressed an unambiguous claim for leadership, but also related it explicitly to Germany’s role in Ukraine crisis management:

‘Is Germany ready to lead?’ My answer is: Yes, we are. … One example is the Ukraine crisis: It commits NATO and the EU and the OSCE. … In each of these three organisations Germany has demonstrated appropriate commitment at an early stage. … the contribution of the Bundeswehr to implementing the decisions taken at the NATO summit in Wales is just as indispensable as the untiring commitment of the Federal Government to … ensure that the EU adopts a common position with regard to Russia. Germany is thus at the centre of transatlantic and European policy. (Von der Leyen 2015: 2, 4f.)

Although these statements in themselves cannot ‘prove’ German leadership, they increase the plausibility that Germany took the lead in Ukraine crisis management (‘straw in the wind’).

The third step concerns the use of leadership strategies. One such strategy is agenda setting. Germany has acted as an informal agenda setter in the mediation process. Not only has it been ‘leading’ in providing papers that present options and propose solutions (Interview U5). The Chancellery also provided the roadmap for the Minsk process (Interviews U4, U5). It has thereby become a usual procedure that Germany drafts the agenda, coordinates it with France, and then searches for agreement with Russia before application (Interview U4). As regards EU sanctions, Germany provided a focal point for agreement. It assumed a ‘balancing’⁶ (Interview U5) and ‘intermediating’⁷ (Interview U2) role between ‘hawks’ and ‘doves’ and between more and less affected member states (Interview U1; Seibel 2015: 67). Thus, Germany proposed sanctions that were asymmetric regarding their negative repercussion within the EU, but were therefore politically viable (Interview U4).

In so doing, Germany showed credible commitment by assuming a major share of the negative economic consequences against the resistance of its own business lobbies (Interviews U2, U5, U6; Fix 2016: 120f.). The fact that Germany was willing to forgo economic gains for the sake of sanctioning Russia demonstrated to other member states that the preservation of a liberal peace order was paramount for Germany (Interview U2). Convincing others to join a common project by signalling credible commitment through the assumptions of costs is one way of leading by example (Hermalin 1998).

Another strategy is pre-negotiation in smaller subgroups. While formats varied (Interviews U3, U4), the Franco-German tandem emerged as the pre-eminent constellation. There has been daily coordination with France via e-mails and phone calls at different levels (Interview U3). The bilateral co-ordination helped find a compromise on sanctions because France often did not share preferences with Germany. While French interests were representative for Southern member states, Germany was closer to Central European and Baltic countries (Interview U2). Thus, the eventual compromise considered the interests of all member states. The Franco-German cooperation resulted in the so-called ‘Normandy format’ (Ukraine, Russia, Germany, France). Although Germany and France did not have a formal mandate, they informally represented the EU through this format (Interviews U2, U3). Merkel used it in February 2015 to negotiate the Minsk agreement. These negotiations were preceded by two meetings in Kiev and Moscow, and one meeting in Berlin at the level of state secretaries (Interviews U4, U6).

If agreement cannot be reached in one negotiation arena, a leader can switch arenas (arena-shifting). In July 2014, Merkel decided to deal with the Ukraine crisis in the Chancellery and thereby shifted the issue from Foreign Ministers to the Heads of State and Government (Interviews U2, U4, U5, U6). Accordingly, Minsk II was negotiated at the level of leaders, too.

Based on their power resources, leaders may modify the followers’ course of action by changing their own behaviour (unilateral action, Underdal 1994: 183). Although Germany had never been a driving force behind economic sanctions (Interview U5), it changed its position after the downing of passenger flight MH17 in July 2014. Germany’s changeover caused a ‘pull effect’ and made more reluctant states like France and Italy change their position as well (Bunde 2016: 182f.). Unilateral action as a leadership strategy is related to ‘moving first’: by taking the initiative, an actor reduces the possible moves of others and may even establish path dependency. In summer 2014, Germany moved first and thus became the ‘driving force’⁸ (Interview U4) in the negotiations leading to Minsk II. Specifically, Germany did not settle for the status quo but repeatedly initiated summits (Interviews U4, U5). Thus, Germany became the EU’s de facto representative in mediating the Russia‒Ukraine conflict and reported to other member states (Interview U4).

Hence, there is ample evidence that Germany made use of leadership strategies. Without such evidence it would not be plausible to claim that Germany provided leadership in shaping sanctions and mediating the Russia‒Ukraine conflict (‘hoop test’).

Finally, if Germany provided successful leadership in shaping sanctions, we should find evidence that some member states changed their positions so that agreement became possible. As regards non-economic sanctions in spring 2014, there was ‘great solidarity’ (Interview U6) among member states (Interviews U1, U2, U6). Yet when in July 2014 the German government started to support economic sanctions, it faced more resistance (Fix 2016: 117; Speck 2016: 8f.). While Poland, the Baltic states, Sweden, Denmark and partly the UK were in favour of further sanctions, a counter-coalition including Italy, France, Austria, Czech Republic, Slovakia, Hungary, Bulgaria, Greece and Cyprus initially opposed them (Interviews U1, U2, U4). However, three days after a video conference, which along with the US, Germany and the UK included reluctant Italy and France (Speck 2016: 9), all member states agreed on economic sanctions, so that the Council could adopt them on 31 July 2014.

In summary, we find observable manifestations for all steps of the causal mechanism. The case study reveals strong evidence in particular for the mechanism’s crucial piece, the use of leadership strategies. Systematic process-tracing thus confirms that Germany provided leadership in shaping and representing the EU’s response to the Ukraine crisis.

Conclusions

If we want to be sure that a policy or institutional change is the result of leadership, we need to trace the underlying causal mechanism: while comparisons or counterfactuals can point to congruence between a potential leader and a given outcome, process-tracing can reveal the causal mechanism in between. Therefore, this article proposes a theoretical conjecture and a related causal mechanism to trace a leader’s impact. Moreover, two case studies show how this approach can fruitfully be applied to the study of EU politics.

To be sure, the proposed causal mechanism is far from perfect. In particular, its single steps could be further specified in future research. For example, one might explicate which leadership strategy is appropriate given a particular collective action problem and available power resources. Furthermore, one might ‘unpack’ the single strategies by splitting them into subordinate parts of the causal mechanism. Both refinements would also help improve the mechanism’s test strength: while the combined evidence of multiple ‘straw-in-the-wind’ tests still provides a quite unique signature for a causal mechanism, the question of how to obtain ‘smoking gun tests’ remains crucial (and unanswered). However, by specifying a causal mechanism (increase of internal validity), the researcher risks losing generalisability (external validity). Since this article aims to illustrate a general way forward on tracing a leader’s impact, it prioritises external over internal validity. Future research, which focuses on a more specific theoretical proposition or empirical scope, should do the opposite.

Finally, the main purpose of the procedure proposed by this article is to check whether it is plausible that a given outcome results from leadership. Therefore, the two case studies serve primarily to illustrate the added value of applying rigorous process‐tracing to leadership research. In doing so, they necessarily disregard parts of real-world complexity. A complete explanation of outcomes, instead, would require the consideration of further factors. In order to fully understand the announcement of OMT, for instance, one would need to take into account the ECB’s negative experience with its prior programme of limited capacity (‘Securities Markets Programme’) and the infamous letters sent to Italy and Ireland, revealing a lack of control over the risk of its bond purchases (Schoeller 2019: 168f). Another important aspect would be the role of individual leadership: Draghi’s leadership style or Merkel’s personal relationship to Putin may indeed be crucial explanatory factors for the respective outcomes. While the proposed theory-testing procedure focuses on the leadership of collective actors only, it remains open for future extensions such as the adaptation to the study of individual leadership or ‘explaining-outcome process-tracing’ (Beach and Pedersen 2013: 63–7).

Acknowledgements

I would like to thank Femke van Esch and Henriette Müller for their valuable feedback on earlier versions of the article. Moreover, I am indebted to Tobias Bunde and Wilfried Jilge for very informative background talks on the Ukraine crisis, and I am grateful for constructive comments provided by the participants of the ECPR Joint Sessions workshop on ‘Leadership in EU Politics and Policy-Making’ (Nicosia 2018) as well as three anonymous reviewers.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Notes on contributors

Magnus G. Schoeller

Magnus G. Schoeller is a postdoctoral researcher at the Centre for European Integration Research (EIF), Department of Political Science, University of Vienna. He is the author of Leadership in the Eurozone and a co-author of European Parliament Ascendant (both Palgrave Macmillan 2019). His articles have appeared in journals such as the Journal of European Public Policy or the Journal of Common Market Studies. [[email protected]]

Notes

1 ‘Collective action problem’ refers to a situation in which everyone would be better off cooperating but diverging interests make cooperation fail.

2 For an overview, see https://europa.eu/newsroom/highlights/special-coverage/eu-sanctions-against-russia-over-ukraine-crisis_en (accessed 27 August 2018).

3 Translated by the author.

4 Translated by the author.

5 Translated by the author.

6 Original: ‘ausgleichend’.

7 Original: ‘vermittelnd’.

8 Original: ‘treibende Kraft’.