These days, hardly a discussion about markets goes by without somebody bringing up political risks. Italy's budget and Brexit are perennial topics, of course. But at least in our experience, concerns about elections to the European Parliament (EP) are also on the rise. We have only just started our own analysis, using similar approaches to those we developed to analyze and attempt to predict national election outcomes in recent years. Our initial results suggest several reasons to feel reassured.
Below, we describe:
How it works: Is there such a thing as a "European election?"
What can we (or anybody else) already say about the likely outcome? And how we intend to cover the rest of what promises an unusually exciting election campaign.
On May 23-26, "Europe" will elect the 705 members to the European Parliament (MEPs). And if you stumbled over the use of Europe in the previous sentence, good for you. We will return to the question of whether there is such a thing as a "European election" in Section 2. The sentence mentioning "705" MEPs also assumes Brexit takes place and that the UK's departure will be sufficiently certain, come May 23, for the country to leave to avoid having elections that day. We will touch on those topics in Section 3.
In the meantime, it is worth highlighting that since its creation, the EP has continuously gained new powers. The election outcome will help determine the composition of the European Commission and influence other important upcoming nominations, notably to the European Central Bank (ECB). In most policy areas, the EP's influence is now on an equal footing with the European Council (consisting of each member state's head of government). To adopt or amend European Union (EU) legislation, the final wording generally has to be approved by both the EP and the Council. (Legislation can only be proposed by the European Commission.)
In addition, results of the European elections can play an outsized role in shaping the political discourse within member states. A good example can be found in the 2014 results. In the UK and France, it marked a break-through moment for Nigel Farage's UK Independence Party (UKIP) and Marine le Pen’s Front National (FN). Both UKIP and FN topped their respective national polls. Despite some advance warning in the polling, observers were stunned by what looked like "political earthquakes". Later results indeed confirmed some shifts in the tectonic plates of national politics, and not just in France and the UK. In France, the 2014 EP results showed how vulnerable the two traditional mainstream parties were, though the eventual beneficiary in the 2017 presidential election turned out to be Emmanuel Macron, not Ms. Le Pen. In Spain, the left-wing populist Podemos party won 8% in the 2014 election to the EP, foreshadowing its strong showing in two subsequent national general elections. In Italy, the equally populist but ideologically more eclectic Five Star Movement came in second with 21%, before emerging as the clear winner in the Italy's March 2018 parliamentary elections. Over the past year, support for both Podemos and Five Star has eroded. In both instances, this partly appears to reflect the rise of right-wing populist forces, Vox in the case of Spain and Lega, Five Star's far-right coalition partner, in the case of Italy.
From the experiences since 2014, it is understandably tempting to extrapolate as to what might happen in May. In coming weeks, you will probably encounter plenty of confident predictions of a populist wave at the elections to the European Parliament. Based on the experience of Trump and Brexit, English-speaking commentators will be looking for early signs of another shift to the far-right.
The European reality is likely to be far messier. Statistically, historically and even in terms of the election rules, there is no such a thing as a European election. Elections take place via national lists (plans for pan-European lists have been shelved). All countries are using (various versions of) proportional representation, but the details vary quite a bit between member states. Seats to the EP are allocated under a (fairly loose) system of degressive proportionality. For example, in small countries such as Malta and Luxembourg, one MEP typically represents 100,000 inhabitants or even less. This compares to more than 800,000 inhabitants per MEP for the largest member states, Germany and France. Polling quality also varies widely across (and within) member states. (For example, polling accuracy tends to be quite dismal in Italy, but quite good in the Netherlands.]) Moreover, most pollsters only ask about voting intentions for the EP elections in the last few weeks ahead of the vote. In the meantime, polling aggregators such as https://pollofpolls.eu/ and https://europeelects.eu/ep2019/ tend to rely heavily on hypothetical voting intentions for national parliaments.
To be sure, some data is generally better than no data! But in the context of the EP elections, using hypothetical voting intentions for national parliaments greatly increases the scope for electoral surprises. Voters have historically tended to treat EP elections very differently from how they treat national elections. Voter turnout was generally lower than in elections to national parliaments. Elections to the EP have also tended to be shaped by national issues; voters frequently treated them largely as an opportunity to punish national governments for reasons typically unrelated to European integration. In the jargon of political scientists, EP elections are best described as second-order national elections, i.e. of lesser importance than elections to national parliaments. From the perspective of voters, parties and the media there appears to be less at stake.
All this tends to make forecasting for any given member state very uncertain. To illustrate just how uncertain, consider the 2009 European election in Estonia. And before complaining that such a small country could not possibly matter, keep in mind that it is precisely the smaller countries that tend to be "overrepresented" in the EP. By 2009, Estonia had just started to feel the effects of the global economic crisis and there was wide-spread dissatisfaction with the political status quo. Well before the election date in May 2009, it was already clear who the big loser was likely to be: Estonia's left-of -center Social Democratic Party (SDE). SDE had come on top at the EP elections in 2004, with 37%. By Estonian standards, that had been a huge upset. In national votes, SDE has never come anywhere near its 37% share in the 2004 EP elections before or since, instead seeing its share fluctuate between about 7% and 17%. Following a brief and unhappy stint of SDE in government from 2007 onwards, the bigger question was which of the other parties might benefit at the EP in 2009. The answer, as it turned out, was none of the established parties. The big winner was an independent candidate, Indrek Tarand. With 25.8% of the vote, he was just a whisker behind the Centre Party, at the time the main opposition party. Mr. Tarand had first made his mark as a student dissident in the 1980s. An outspoken former civil servant, he ran a strong anti-establishment campaign on a budget of less than 2000 euros.
If history is any guide, it is likely that the 2019 elections to the EP may see some similar upsets. Within at least a few member states, anti-establishment candidates or parties are likely to do well, despite not showing up much in the polls as yet. This is partly because of the structural features we already mentioned. In second-order elections, smaller protest groupings historically have tended to do comparatively better than larger groups. Also, even insurgents on a tight budget can do quite well by gaining momentum in the last few weeks before a vote takes place (regional elections in federal countries such as Germany are another good example). That, in turn, can generate plenty of free media publicity. Such uncertainty at the level of individual country results does not necessarily mean plenty of surprises on the European level. This is because statistically speaking, elections to the EP are likely to once again be 27 largely uncorrelated national elections in the various member states over a long weekend from May 23-26, 2019. In fact, Europe wide forecasts are likely to be quite resilient against even fairly large polling misses in individual countries. To see why, return to Mr. Tarand's success in 2009. With just above 102.000 votes, he lagged behind the Centre Party by 1046 votes. Centre secured 2 MEP. But, given the vagaries of Estonia's seat allocation that year, Mr. Tarand and three smaller parties each secured 1 MEP – in the case of the SDE with just 8.7% or 35.000 votes.
This is an extreme example of a broader forecasting problem. Depending on precisely how each country allocates its seats, even very large polling errors sometimes have surprisingly little impact, especially in smaller member states with several medium-sized parties or lists. Conversely, small polling misses can sometimes make a big difference. That is especially so for larger member states, which set hurdle rates any party needs to clear to secure any MEPs. For example, Italy has a 4% minimum threshold for EP elections smaller parties have to clear. And, according to the polling average of https://pollofpolls.eu/IT/61/ep2019-in-italy both the pro-European, Alliance of Liberals and Democrats for Europe (ALDE) affiliated list Europe+ and the far-right Brothers of Italy are at about 4% each. That translates into about 8 seats in total, or about 10% of all Italian seats being up for grabs, meaning that even without any significant polling error, Italian seat forecasts might be widely off the mark.
Sources: Eurostat, European Parliament, DWS Investment GmbH as of March 2019
In a recent report, the European Council on Foreign Relations (ECFR), a think tank, made the case for how a big populist wave might threaten European policy-making. Based on a recent in-depth survey of the European Union's 27 remaining member states (after the UK leaves), the ECFR describes: "How anti-Europeans plan to wreck Europe." We will probably return to the report a few more times in coming months. For now, suffice it to say we rather doubt its basic premise. Given all the above, we are more than a little skeptical about claims that right-wing populists look set to make significant gains at the 2019 EP elections in terms of the aggregate seat share.
Assuming Brexit takes place, the most obvious effect will likely be a significant weakening of two of the three rightwing Eurosceptic parliamentary groups, due to the loss of the Tories and UKIP. As for the third, current Eurosceptic grouping, the Europe of Nations and Freedom (ENF), seat gain projections are largely dependent on the governing Lega doing extremely well in Italy. To sustain a narrative of a broader populist surge, you also need to assume that Lega can do so without hurting its coalition partner Five Star too much. That might happen, of course. For larger parties, Italy has frequently seen polling errors in the double digits. Such potential for error, however, exists in both directions. It is equally possible that the combined share of Lega and Five Star is currently overstated in Italian polls. In Italy and elsewhere, populist (erstwhile-protest) parties are now part of the government in several member states. In such situations, these parties have tended to underperform polling averages in the past due to voter dissatisfaction with the new status quo and difficulties in turning out their core voters. Tantalizingly, there are also some early signs in German polling that the far-right Alternative for Germany (AfD) might be struggling to mobilize for the EP. This might reflect a certain amount of fatigue among German voters sympathetic to the AfD's cause. After all, there have been and will continue to be plenty of other opportunities for German voters to punish the governing coalition in regional elections. So why "waste" too much attention, time and organizational resources on the EP elections?
All this is based on the current polling data and assumes the UK will not take part in the EP elections. The data also suggests heavy losses of seats for the centre-left Progressive Alliance of Socialists and Democrats (S&D). This reflects the weakness of the Social Democratic Party of Germany (SPD) and several of its sister parties in many of the larger (remaining) member states. The centre-right European People's Party (EPP) is also likely to suffer some, but more limited losses. This certainly suggested a more fragmented EP. Based on current polling, the biggest gains might well accrue to the centrist socially liberal, broadly pro-European and pro-market ALDE. In various forecasting models, its expected seat gains are fairly broad based, thanks to Emmanuel Macron's La Republique en Marche collaborating with this group, gains by Spain's Ciudadanos and the rebound of Germany's Free Democratic Party (FDP) since 2014, as well as strong showings of liberal parties in a number of the smaller and medium-sized member states. In terms of EU policy-making, much will remain uncertain, even once the election results are known. This is because among the anti-establishment parties in particular, it remains unclear who will be willing and able to work with whom.
We will, no doubt, have more to say on their prospects, as the election gets nearer. In the meantime, we would add our usual disclaimer: a lot can happen in the next two-and-half months until election day. To this, we would add an additional caveat for the 2019 EP elections. This concerns Brexit. With the Brexit date looming on March 29, there is still no clarity as to whether, when and how the UK will leave. Even a decision that results in the UK taking part in the EP elections can hardly be ruled out, though the EU will probably try to avoid this in case of an extension to Article 50. And in terms of the likely direction of European politics, the impact of Brexit is hard to overstate. For one thing, a recent analysis of VoteWatch Europe suggests it will probably increase the regulatory burden of businesses in the remaining EU members, given on how British MEPs and governments of all stripes have voted in the past.
We will return to these issues as and when Brexit visibility – hopefully – improves in coming weeks. Already, Brexit appears to have boosted the popularity of the EU in the remaining member states – and sharply diminished the attractions of exiting in other, erstwhile Eurosceptic countries such Denmark. This creates dilemmas for formerly Eurosceptic parties around the continent, who have lately softened their tone when it comes to membership referenda in their own countries. Such softer tones might make it harder to mobilize for the EP elections. Conversely, a hard, chaotic Brexit or even the mere prospect of a swing to the far right causing economic harm could help mobilize broadly EU friendly forces. Such tactics have worked well in the past, most recently for Reform, the ALDE affiliated winner in the March 3 parliamentary election in Estonia. Which brings us back to Indrek Tarand, the surprise populist runner-up in Estonia's 2009 EP elections. Now about to finish his second EP term, Mr. Tarand may run again, this time probably for the Social Democratic SDE. In the last two EP terms, he caucused with the Greens–European Free Alliance. His story should, perhaps, serve as a reminder that yes, there might well be some populist surprises once again, come May 2019. It is far less clear that such surprises will necessary strengthen the Eurosceptic camp, however broadly defined.
2. For both the ECB and the European Commission, however, note that by far the more important determinant is the changing composition of national EU governments, rather than that of the European parliament. Such analyses suggest a much more diverse Commission, in terms of party affiliation. In particular, Europe's liberals look set to make sizeable gains, mostly at the expense of the main centre-right and centre-left parties, reflecting the changing composition of national governments. See: https://www.foederalist.eu/2019/02/bunt-wie-nie-parteipolitische-zusammensetzung-eu-kommission-2019.html
4. The UK Independence Party is a Eurosceptic, right-wing political party in the United Kingdom.
5. For the full 2014 results in all EU member states, see http://www.europarl.europa.eu/elections2014-results/en/country-introduction-2014.html
6. See, e.g. https://www.independent.co.uk/news/world/europe/european-elections-2014-marine-le-pen-s-national-front-victory-in-france-is-based-on-anguish-rage-9436394.html; https://www.reuters.com/article/us-eu-election-france/far-right-national-front-stuns-french-elite-with-eu-earthquake-idUSBREA4O0CP20140525
7. On possible reasons, for Podemos, see https://www.theguardian.com/commentisfree/2019/feb/19/podemos-spanish-politics; for Italy, see: https://www.bloomberg.com/opinion/articles/2019-02-26/italy-s-populist-insurgents-five-star-are-collapsing
8. Vox is a right-wing populist party in Spain.
9. To get an up-to-date overview of national polls in all 28 EU member states (plus Switzerland and Norway), an excellent and extremely useful source is https://pollofpolls.eu/ It also provides a useful daily seat track in terms of the implied composition of the European Parliaments. For reasons spelt out in the text, such seat projections should at most serve as a starting point, more than two months ahead of the elections.
10. For a basic overview on various national rules, see: http://www.europarl.europa.eu/news/en/press-room/elections-press-kit/1/facts-and-figures-on-electoral-process
11. Under the Lisbon Treaty, each country is allocated at least 6 MEPs, but no more than 96 MEPs. In between, "middle-sized" member states are allocated a proportionally larger number of seats than larger member states.
12. For these reasons alone, forecasting the EP is a bit like forecasting elections to the U.S. Senate, only trickier in that apart from general opinion surveys such as Eurobarometer run by the EU itself, there are very few measure of pan-European public opinion.
13. For our previous takes on polling it Italy, see [Insert link to An Italian Muddle 2/27/2018, https://dws.com/insights/cio-view/emea-en/an-italian-muddle/]. For the Netherlands, see [Insert Link to "When (not) to trust the opinion polls", 3/13/2017, https://dws.com/insights/cio-view/emea-en/european-elections-2017/
14. For early evidence of this tendency, see Reif, K. and Schmitt, H. (1980) "Nine Second-Order National Elections: A Conceptual Framework for the Analysis of European Election Results". European Journal of Political Research, Vol. 8, No. 1, pp. 3–45.
15. Useful more recent studies of EP elections are best described as Second-Order National elections include: Freire, A. and Santana-Pereira, J. (2015) "More Second-Order than Ever? The 2014 European Election in Portugal", South European Society and Politics, Vol. 20, No. 3, pp. 381-401; Schmitt, H. (2005) The European Parliament Elections of June 2004: Still Second-order?, West European Politics, Vol. 28, No. 3, pp. 650-679, Available online at in draft form at: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.522.3159&rep=rep1&type=pdf; Schmitt, H., Sanz, A. and Braun, D. (2008) “The Micro-Foundations of SecondOrder Election Theory: Theoretical Reconstructions and Empirical Tests Based Upon the European Election Study 2004,” Paper presented at the 2008 CSES Workshop, University of Manchester, December 9-11. Gagatek, W. (2010) "The 2009 Elections to the European Parliament – Country Reports", Firenze, European University Institute; in particular chapter 1 by Trechsel, A., " How much ‘second-order’ were the European Parliament elections 2009?", available as an ebook at: http://cadmus.eui.eu/bitstream/handle/1814/13757/EUDO_2009-EP-Elections_CountryReports.pdf;jsessionid=AF4E897149363BB8B61C9F78F821FAF7?sequence=1 Nielsen J.H., Franklin M.N. (2017) "The 2014 European Parliament Elections: Still Second Order?". In: Hassing Nielsen J., Franklin M. (eds) The Eurosceptic 2014 European Parliament Elections. Palgrave Macmillan
17. It owed much to low turnout of just 27% (see http://vvk.ee/varasemad/ep04/tulemus/osavott_eng.html for details), as well as the popularity of SDE's 2004 lead candidate, former foreign minister Toomas Hendrik Ilves, who in 2006 went on to become Estonia's 4th president.
18. Both Estonians biggest parties, Eesti Keskerakond (Centre Party) and Eesti Reformierakond (Reform Party) the centrist Alliance of Liberals and Democrats for Europe (ALDE) in the EP – the main fissure between the two is that Centre is more accommodating to Estonia's Russian minority. For more details on governance in Estonia, see: https://freedomhouse.org/report/nations-transit/2018/estonia
19. For further analysis, see: Ehin, P. and Solvak, M. (2010) "Party voters gone astray: explaining independent candidate success in the 2009 European elections in Estonia", Paper prepared for the PIREDEU Final User Community Conference, Brussels, 18-19 November 2010
20. For results, see: http://www.vvk.ee/ep09/index.php?id=11208; the somewhat strange distribution of seats among the smaller parties appear to be due to the fact that Mr. Tarand technically ran as a one-person closed-list (see Ehih and Solvak (2010), cited above). So, more than half the votes he won were effectively wasted, in terms of parliamentary representation, allowing the SDE to hold on to one of its MEPs with just under 35.000 votes.
21. It is a transnational alliance between two European political parties, the Alliance of Liberals and Democrats for Europe Party and the European Democratic Party.
22. However, such national polling misses with regards to smaller parties often are likely to tend to cancel each other out for Europe-wide seat forecasts. More than half of the remaining member states effectively have thresholds and there is no reason to think polling errors might to be correlated across various countries.
23. The survey, by Pawel Zerka & Susi Dennison, was published on February, 2019. It is accessible here: https://www.ecfr.eu/specials/scorecard/the_2019_European_election. A brief outline of the ECFR's main arguments can be found here: https://www.ecfr.eu/article/commentary_how_europes_populists_can_win_by_losing
24. There is a tendency of pollsters to overcorrect past (real or imagined) polling misses.
25. Depending on how one classifies and counts various parties and governing arrangements, about 1/3 of EU member states are already governed, to a greater or lesser extent, by right-wing populists.
26. In particular, AfD tends to do worse in polls for the EP elections than when the question is hypothetically asked for national parliamentary elections. On the https://pollofpolls.eu/DE/98/ep2019-in-germany polling average, it is at about 10% at the time of writing for the EP election, compared to 13% in national elections (Für Übersetzung: Sonntagsfrage).
27. Another useful source for analysis and to catch-up on the state of the race is https://www.votewatch.eu/; for their two most recent updates, see: https://www.votewatch.eu/blog/ep-2019-increasingly-fragmented-parliament-fresh-projections/ and: https://www.votewatch.eu/blog/eu-elections-countdown-state-of-candidates-manifestos-projections/
28. See, for example, https://pollofpolls.eu/EU
29. For more details, see: https://europeelects.eu/2019/01/30/four-scenarios-for-the-european-election/ as well as: https://www.votewatch.eu/blog/ep2019-group-discipline-under-pressure-post-elections/
33. See, for example, http://cphpost.dk/news/danish-desire-to-leave-eu-nosedives-after-brexit.html; http://cphpost.dk/news/danes-largely-positive-about-eu-new-poll-shows.html; https://www.independent.co.uk/news/uk/politics/brexit-denmark-poll-eu-referendum-consequences-a8286296.html; https://www.politico.eu/article/how-the-afd-could-save-the-eu/
35. For an in depth recent interview, see: https://news.err.ee/912054/interview-indrek-tarand-back-from-europe-and-running-for-home-elections