Expressive voting with booing and cheering: Evidence from Britain

Highlights

• Expressive behaviour may involve ‘booing’ disliked parties as much as ‘cheering’ preferred party.

• Analyse British Election Survey to understand turnout decision.

• “Booing” opponents increasingly more important than “cheering” preferred party.

Abstract

Previous work on Expressive Voting has focused on the desire of voters to express what they are for and thus who they are. But, often also as important, is the desire of voters to express what they are against, and who they are not. In this paper we extend the standard formulation of Expressive Voting to account for this possibility. Using data for the UK we find empirical evidence that the desire to boo has been increasingly important at recent elections. The implications of this for recent trends in political polarization are discussed.

Introduction

The theory of Expressive Voting (Brennan and Lomasky, 1993; Hamlin and Jennings, 2011) suggests that voters care not only about the outcomes of elections but also about expressing their opinions, beliefs, or conscience. As discussed by Hamlin and Jennings (2019), the domain of expressive motivations is ‘large and nuanced’ in a similar manner to conventional instrumental concerns. Thus, voters may want to express what they are, and they may want to express what they are not. They may be motivated to vote by a desire to cheer a political party whose platform reflects their views and identity, and by a desire to boo parties whose platform does not.¹ This paper argues empirically that voters' desire to express what they are not has been as, if not more, important than the desire to express what they are at recent British elections, and increasingly so.

We suggest this finding may offer an additional perspective on the well-documented recent increases in political polarization and populism in many western democracies. Across countries and political systems, the last decade has witnessed the emergence of electorally-successful populist and anti-establishment politicians. While, the specifics naturally vary, many countries have experienced the following:

• Increased political polarization

• Populism and anti-establishmentism

• Involvement by movements of both the political right and left.

There are many competing explanations for each of these trends individually and for them as a set. Some appeal to pocketbook voting explanations and the effects of austerity including Fetzer, 2019 for the UK, Autor et al. (2016) for the US, and Dal Bó et al. (2018) for Sweden. Others emphasise the limited explanatory power of the internet and social networks (Allcott and Gentzkow, 2017; Boxell et al., 2017). Similarly, in the UK at least, there was a limited role of migration (Becker et al., 2017) although xenophobia and anti-Muslim sentiment predict support for the far-right EDL.

This paper posits that the changing importance of booing and cheering is one explanation, perhaps reflecting increasing antipathy towards politics and politicians. This explanation requires neither changes in voter preferences nor pocketbook concerns, but instead changes over time whether voters are more likely to vote to express their preference or to express what they are not. Put another way, differences in whether voters’ political behaviour is driven by cheering what they are for or booing what they are against. We provide empirical evidence that booing is an important motivation, and at recent elections more important than cheering.

Brennan and Lomasky (1993), Schuessler (2000), Brennan (2008), Hamlin and Jennings (2011) argue for an expressive theory of political behaviour. In this view voters decide whether and how to vote depending both on the outcome that will occur if their vote were decisive and also based on their return from expressing their opinion, conscience, or beliefs. This departure from the standard view in rational choice where voters have preferences only over outcomes is reminiscent of the large literature on the, by now, well documented preferences for fairness identified by Fehr and Schmidt (1999), Dawes et al. (2007), and Tricomi et al. (2010). In the same way that that literature argues individuals have preferences over economic behaviour beyond their economic self-interest, the theory of expressive political behaviour suggests they have preferences over political behaviour beyond their narrow self-interest. Notably, while in a market setting there are reasons to believe that these other preferences may be normally a comparatively small aspect of behaviour as the opportunity cost may be expected to be high. In the case of elections where voting is private and unlikely to affect outcomes, opportunity costs will be smaller.²

The analysis of Brennan and Lomasky (1993) distinguishes between cheering the party or group you support and booing the one you do not. But, focuses on the case of two parties and thus this distinction has limited empirical content. Brennan (2008) argues that strategic voting may be understood in expressive terms, and there the distinction between cheering and booing has more bite.³ Importantly, outside of the US where three or more parties contesting elections is the norm, the comparative desire of voters to cheer versus boo may imply different outcomes.⁴

This paper analyses British data for general elections over the period 1922–2017 to establish some stylised facts about the relative importance of cheering and booing. We find that:

• booing is comparatively important

• that the relative importance of cheering and booing have changed over time

• the importance of booing seems to be higher after the 2008 Financial Crisis.

Note, viewed from the perspective of expressive behaviour it is the decision to vote that is key and one feature of the data is that turnout varies considerably over time. Individuals are not strategic in this view, and so the outcome of an election hinges on which groups can be motivated to vote.⁵ Fig. 1 reports aggregate UK turnout rates for the last 95 years and we note firstly that turnout has varied considerably over time from a high of nearly 84% in 1950 to historic lows of just under 60% in 2001. Secondly, we note that turnout is relatively volatile with differences in turnout rates between elections of around 10% in several cases. Long-run changes in turnout rates will represent a combination of different non-consequentialist (Shayo and Harel, 2012) factors encompassing warm glow and civic duty explanations as well as changes in information acquisition (Feddersen, 2004, Feddersen and Sandroni, 2006) and social preferences and information (Myatt, 2015) or strategic protest voting (Myatt, 2017). To the extent that these explanations rest on fundamentals of voters preferences it is harder for them to explain short-run changes in turnout rates as normally we regard such fundamentals as, at most, slowly changing if not fixed. In general, there is every reason to believe that voters' behaviour will reflect both expressive and instrumental motivations. But, in this paper we will assume for clarity, and given that the elections we study are large, that there is little instrumental benefit from voting and thus that voters’ behaviour reflects expressive motivations.

At this point it useful to briefly rehearse the key features of UK political history over the last century. Fig. 2, Fig. 3 plot the vote shares and number of seats in the UK Parliament over the same 95 year period as Fig. 1. The blue and red lines depict the two largest parties, the Conservative and Labour parties. Our interest is in the fluctuating importance of the gold line which plots the varying success of what was initially the Liberal Party and is now the Liberal Democrat party, and the purple line which plots the fortunes of ‘other’ parties, most importantly United Kingdom Independence Party (UKIP) (in terms of votes but not seats) in recent years, as well as the green line plotting the importance of the Scottish and Welsh nationalist parties. The detail has been extensively documented and analysed by historians and political scientists, for example Sked and Cook (1979), but our interest is in the macroscopic story that following the ‘Strange Death of Liberal England’ Dangerfield, ([1935] 2017) prior to 1918, the period 1922–1945 saw the Labour party usurp the Liberals as the main opposition. The Cold War period then saw a stable, Duvergian, two party duopoly with almost all seats until the beginning of the 1980s when key members of the Labour party split to form a separate, more centrist party, and in time joined with the remainders of the Liberal party to form the Liberal Democrats. Also relevant is the growth in the Nationalist parties, and crucially from the late 1990s onwards, the populist anti-EU UKIP. Thus, the UK has become, contra Duverger, a multiparty democracy with coalition government in 2010 and minority government in 2017 (see Fig. 3).

Coincident with these changes in the party landscape has been a precipitous decline in the perception of politicians and political parties. The British Social Attitudes Survey Lee and Young (2013) documents that

Back in 1986, only 38 per cent said they trusted governments “to place the needs of the nation above the interests of their own political party”. By 2000, this had more than halved to just 16 per cent. After rising somewhat, it returned to a similar low in the immediate wake of the MPs' expenses scandal of 2009 and, at 18 per cent, the latest figure is only a little better.

This, and declines in trust in, and respect for, politicians may explain why voters are increasingly less motivated by expressing who they like, and more by expressing who they like least. This is consistent with our finding that booing has become more important in recent years, and this change may explain in part the proliferation of parties – booing one's political opponents reduces the viability of large centrist ‘big tent’ parties.

This paper is organised as follows. The next section introduces the British Election Survey data we use, how we measure expressive voting and our empirical strategy. Section 3 presents the results of this analysis. We discuss possible interpretations of our research in section 4. Finally, section 5 briefly concludes.