![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
ABSTRACT
The recent resurgence of inflation in Europe has led the ECB to increase interest rates and phase out asset purchase programmes designed to address the effects of the Great Financial Crisis. This article investigates how the ECB has adjusted its logic of responsibility throughout this series of crises. Using a topic model and in-depth analysis of speeches, we examine the ECB’s strategic framing of linkages related to inflation during three historical periods: the Central Bank Independence (CBI) era (1998–2011), the secular stagnation era (2011–2021), and the new inflation era (2021-). Our findings indicate that modifications made to the CBI’s linkages during the secular stagnation era shaped the ECB’s framing of the new inflation era in a novel way. However, despite acknowledging difficult policy tradeoffs, which they used to downplay in the past, ECB policymakers have continued to reframe its initial imperative of responsibility in the hope of avoiding policy discussions on regime change.
Acknowledgement
We would like to thank the editors of the special issue ‘Beyond Responsibility vs. Responsiveness,’ Tiago Ramalho Moreira, Amandine Crespy and Vivien Schmidt, as well as the participants in the different workshops organised to prepare this special issue. We would also like to thank the participants of ISPOLE’s 2023 research day for their useful feedback, as well as the three anonymous referees.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
2 Treaty on the Function of the EU (TFEU), Article 127.
5 Careful about their regulatory reputation, independent institutions favour discourse and policy stability and refrain from abrupt changes (Carpenter, Citation2010).
7 In our study, words and expressions mean unigrams (like ‘price’), bigrams (‘price stability’) and trigrams (‘maintain price stability’).
8 For example, the words with the highest values associated with topic 14 are ‘price stability’, ‘stability’ ‘objective’ and ‘maintain price stability’. This indicates that topic 14 is about the ECB’s objective of price stability.
9 For instance, a paragraph may be strongly associated with both a topic on price stability and one on fiscal policy.
10 Ferrara et al. (Citation2022) do not look at ECB’s public speeches but at European Parliament hearings of ECB presidents.
11 Unsupervised method has been already used on different objects to identify frames: Klüver and Mahoney (Citation2015) use K-means clustering method to attribute to different European interest groups’ documents a cluster (consider as a ‘frame’); Fligstein et al. (Citation2017) use topic modeling to uncover various frames used by the Federal Open Market Committee to talk about monetary policy.
12 We select ECB’s speeches that display mentions of ‘inflation’ a number of times at least twice the number of pages of the speech. We include in this calculation all the terms that are composed of the chain of characters ‘inflation’, i.e. ‘disinflation’, ‘inflationary’, etc. See the online appendix (‘2. Step 1: Creating the corpus’) for further details on how we have tested different thresholds.
13 The emergence of new topics in speeches do not necessarily indicate new framing strategies about inflation. For example, central bankers could mention fiscal issues in paragraphs that are unrelated to inflation. Similarity measures help us to monitor how the evolution of peripheral topics is linked to the evolution of framing strategies related to inflation. Supplementary Table 2 lists the relevant topics with the highest correlation for each of our three periods. See online appendix (‘4.2 Calculating similarities’) for details on similarity measures.
14 The choice of pre-processing steps may be crucial for the results produced by a topic model (Denny & Spirling, Citation2018). We discuss our choices at length in the Appendix (section ‘Step 2.b: Evaluating different models’).
15 See details about quantitative and qualitative validation in the online appendix (‘Step 2.b’).
16 Online appendix (‘5 Information on all the topics’) lists all the topics, their top words, as well as the most representative speeches and paragraphs for each period.
17 Two other prevalent topics refer to technical specifications related to the CBI model: inflation targeting (topic 102) and the definition of medium term inflation (topic 49). See online appendix for details.
18 For each date of publication of a speech, we average the strength of association between the paragraphs published at this date and the topics we are interested in. We thus obtain the prevalence of these topics over time. The figures with topics are built by estimating the relationship between dates and prevalence, using a local polynomial regression with a ‘span’ value of 0.2. The result is a smoothed representation of the data allowing us to observe the overall trend of various topics. See the online appendix (‘Step 3: Analysing the results’) for further explanations.
19 See also the prevalence of Topic 7 on wages during this period.
20 ‘Stepping up structural reforms in these areas is indispensable for improving the euro area’s unsatisfactory growth potential and its ability to create employment’ (Issing, Citation2004, p. 2).
21 ‘This recalibration of our asset purchases … helps to maintain the necessary degree of accommodation and thereby to accompany the economic recovery in an appropriate way’ (Draghi, Citation2017c, p. 4).
22 Topic 47 refers to ‘underlie; headline; headline inflation; core; measure’. For example, ‘the strengthening of economic activity has yet to find correspondence in inflation developments: underlying inflation and domestic price pressures remain subdued’ (Praet, Citation2017, p. 5).
23 Constancio explained that ‘the negative demand shock could be the delayed result of the fiscal policy tightening since 2010, or possibly the consequence of weakened economic sentiment after two recessions’ (Constâncio, Citation2018, p. 5).
24 Topics 30, 58, 67. Topics 35, 47 and 60 have a high prevalence during the new inflation era but they were also prevalent during the CBI or the secular stagnation eras.
25 The evolution of topic 60 about ‘anchoring inflation expectations’ (), which is prevalent during both the CBI and the new inflation era but not during the secular stagnation era, corroborates this finding.
26 These citations come from the most representative speeches in topic 103 for the last period. This topic picked up first in late 2019 and during the Covid period: prevalent speeches in this period defended the need for active and expansive fiscal policies to sustain economic growth and help EU countries to recover.
Additional information
Notes on contributors
Aurélien Goutsmedt
Aurélien Goutsmedt is a F.R.S-FNRS post-doctoral researcher at UcLouvain.
Clément Fontan
Clément Fontan is Professor in European Economic Policies at UcLouvain.