COVID-19 and party competition over the EU: Italy in Early Pandemic Times

ABSTRACT

Despite receiving much attention in the literature, existing analyses on the impact of Covid-19 on European societies and politics do not investigate the consequences for party competition over the European Union (EU) dimension. To this end, this article asks whether the pandemic affected the salience of the EU issue and the related party positions in Italy during the ‘first wave’ of the crisis. The analysis relies on an original CrowdTangle dataset comprising around 24000 posts from parties’ official Facebook pages, which are thematically coded and subsequently employed in time-series cross-section regression models. The findings show that the pandemic caused a significant increase in salience and polarisation of the EU issue in the Italian party system.

KEYWORDS:

• COVID-19
• Italy
• European integration
• party system
• party competition

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Data availability statement: The article is based on an original dataset, made up of original Facebook data obtained by employing the CrowdTangle platform, Manifesto Project (MARPOR) data, Chapel Hill Expert Survey (CHES) data, and independently collected data. The data and software syntax for the analyses will be provided once the article will be accepted for publication. Acknowledgements: This article is the outcome of an equal collaboration between the authors. Previous versions of the article have been presented at the ‘Emotions and political attitudes in the context of COVID-19’ workshop during the European Consortium for Political Research (ECPR) 2021 Joint Sessions of Workshops hosted by the Charles University of Prague; at a 2021 workshop entitled ‘The causes and consequences of EU (dis)integration’ hosted by the Institute for European Studies of the Vesalius College and Utrecht University; and at the 10th Conference of the ECPR Standing Group on the European Union hosted by LUISS University. We want to thank all the participants and convenors of the three events, in particular Ekaterina Lytkina, Margaret Samahita, Nathalie Hofstetter, Enzo Loner, Olesya Tkacheva, Erik Niemand, Hans Vollard, Martin Huysmans and Sven van Kerckhoven. We are particularly indebted to Björn Bremer, Giorgio Malet, Vincenzo Emanuele, Maurits Meijers, Luca Carrieri, and Lorenzo De Sio for their helpful comments and suggestions on earlier drafts of this article.

2 The ‘Letter of the nine’ is available at https://www.governo.it/sites/new.governo.it/files/letter_michel_20200325_eng.pdf. The reported passages are from p. 3 of the document.

3 Although both neo-functionalism and post-functionalism are mainly concerned with the drivers and scope of European integration, their theoretical claims offer relevant insights into the structuring of party competition at the national level.

4 For instance, some relevant far-right parties, such as Forza Nuova, could not be included due to their absence from social media during the analysed time frame.

5 The parties are: Articolo Uno, Azione, Cambiamo con Toti, CasaPound Italia, Fiamma Tricolore, Forza Italia, Fratelli d’Italia, Italia Viva, Lega - Salvini Premier, Movimento 5 Stelle, Partito Comunista, Partito Democratico, Più Europa, Potere al Popolo, Sinistra Italiana.

6 Whilst social media are becoming key for parties’ political communication strategies, more traditional instruments (such as electoral manifestos) are still relevant. This is recognised by the press release assumption, which does not imply that parties will neglect traditional tools when it comes to delivering political messages to citizens, but rather that social media will function as a complementary instrument that boosts parties’ presence in the media (see, e.g., De Sio et al., 2017).

7 The intercoder reliability tests were carried out on a random sample, constituting 10% of the overall dataset as per the indications of the methodological literature (e.g., Lombard et al., 2010). Other coefficients, including Krippendorf’s Alpha, confirm the robustness of the reported statistical measure. All details on these tests are reported in Table A2 in the Appendix.

8 Specifically, an LR test of panel-heteroskedasticity, a Wooldridge test of serial correlation and a Hausman test between fixed and random effects were performed for both Model 1 and Model 2. In both cases, the first two diagnostic tests were significant at p<.001, confirming that the panels are heteroskedastic and the observations serially correlated. However, the Hausman test was never significant, hence allowing the use of random effects.

9 Our results are confirmed by all the robustness checks we have performed. More specifically, these are panel-data fractional regressions performed through both a population-averaged model by employing general estimating equations and a multilevel generalised linear model for Model 1; a panel-data tobit regression and the same specification as in the main model by taking the log transformation of the dependent variable for Model 2. These different robustness checks account for the differences in the dependent variables, i.e. a proportion in Model 1 and a continuous variable in absolute values in Model 2. The robustness checks for Model 1 and Model 2 are reported, respectively, in Table A3 and Table A4 in the Appendix.

10 The MARPOR dataset version employed is 2020b.

11 Analogous tests illustrate that no other regression model in this article is affected by issues of multicollinearity.

12 This categorical variable takes on the following values: (0) between the beginning of the analysed time frame and the first key event (9 February-9 March); (1) following the start of nationwide lockdown on 9 March and until the second key event (10 March-9 April); (2) following the Eurogroup agreement on establishing the Recovery Fund on 9 April and until the third key event (10 April-18 May); (3) following the announcement of the Franco-German initiative for the Recovery Fund on 18 May and until the fourth key event (19 May-28 May); and (4) following the EU Commission proposal on the Recovery Fund on 28 May and until the end of the analysed time frame (29 May-3 June).