Increasing attention is being paid to the distinction between ‘instrumental’ and ‘expressive’ choice in the political literature, and specifically in that part of the literature devoted to voting behaviour. The idea of ‘expressive voting’ captures the idea that voting may be motivated by concerns other than a concern for the eventual outcome of the election – concerns that are more directly and immediately linked to the act of voting, or of voting for a particular candidate or option, itself. A now-standard line of argument in support of the idea of expressive voting in large-scale elections begins with the observation that for an ordinary member of a large electorate, their individual vote is extremely unlikely to determine the electoral outcomes. Any ‘instrumental’ calculus that focuses on the expected benefits associated with the outcome of the election, and admits that voting is at least somewhat costly, is therefore likely to show that voting is irrational. By emphasizing aspects of the act of voting, or of voting for a particular candidate or option, that do not depend on the outcome of the election, voting may be portrayed as individually rational; and such aspects have been labelled ‘expressive’.Footnote 1
The basic idea here seems clear enough: voting, or voting in a particular way, may ‘express’ some aspect of the voter's beliefs, values, ideology, identity or personality regardless of any impact that the vote has on the outcome of the election, and such ‘expression’ may be valuable to the individual in its own right and so provide sufficient motivation to vote. This basic idea needs considerable further development if it is to offer more than a general critique of the standard instrumental model. Expressive theory must move beyond the mere logic of the idea of expressive voting and the contrast with instrumental voting, in order to face a range of issues concerned with behaviour other than voting, the content of the relevant expression and the behavioural and normative implications of expressive behaviour.
Since much of the literature to date has concentrated on the basic contrast between expressive and instrumental voting,Footnote 2 it has exhibited many different approaches to the content of expressive behaviour. Responding to this variety, a common criticism of the expressive idea is that its myriad possibilities make it rather ad hoc and lacking in specific predictive and normative bite. We agree that no single, clear definition of the content of expressive behaviour has achieved general acceptance; instead, there are several competing accounts. Recently, for example, Hillman has offered a definition of expressive utility as that which derives from confirming identity.Footnote 3 While his approach has much in common with the discussion to follow, and we certainly agree that identity is an important element in understanding expressive behaviour, we disagree with the claim that confirmation of identity exhausts the potential sources of expressive utility; rather, we see identity as just one possible source of expressive benefit. The ideas of duty, morality, deception and self-deception (as well as other ideas) all feature alongside issues associated with identity. For reasons to be discussed in the next section, we do not pursue the specification of a formal utility function; rather, we seek to interrogate the nature of expressive behaviour in more general terms. Partly because of the variety of approaches adopted, there is also no consensus on the implications, either positive or normative, that can be drawn from the analysis of expressive behaviour. Indeed, some discussions of expressive voting focus only on the decision to vote, and the implications for electoral turnout, rather than the decision of how to vote and the implications for electoral outcomes.
In response, this article provides a more detailed definitional account of expressive behaviour and, with definitions in place, a discussion of the foundational content of expressive choice distinguishing between a number of cases and relating these cases to the specific theories of expressive choice in the literature. We also discuss the normative implications of the various theories.
This article is intended, in part, to survey the literature on expressive choice. However, it is distinct from a number of recently published articles which set out to survey the literature on voting turnout/participation and which refer to the expressive idea as one of several approaches to this topic.Footnote 4 We focus on expressive motivations and behaviour across the range of political behaviour, but even when limited to the area of voting, our perspective differs from that in the turnout/participation literature. The primary concern in that literature is explaining why individuals vote at all rather than how they vote. By concentrating on the content of expressive choice, rather than the logic of expressive choice, we will focus on how individuals vote and, more generally, how they behave.
While this article has a survey aspect, it is intended to be much more than that. Our first major aim is to expand the understanding of expressive behaviour beyond the specifics of voting in large-scale elections and beyond the sharp contrast with instrumental behaviour. We will argue throughout that expressive ideas are potentially relevant across all institutional arrangements and in a wide variety of contexts, and that it is the interaction between expressive and instrumental consideration, rather than the contrast between them, that is important.
Our second major aim is to use the variety of substantive accounts of expressive behaviour not as a critique of the operationalization of the expressive idea, but rather as an indicator of the richness of the expressive domain. In the instrumental domain, the idea that individual interests are complex and context dependent is well known, and while it may be appropriate in some models to compress interests into one or two variables (such as income or wealth maximization), it is understood that this stands in place of a more detailed and nuanced account of interests that becomes relevant if we wish to explore behaviour in greater detail. Similarly, in the expressive domain, we argue that the range of expressive concerns is wide and complex, and that there may be tensions between different expressive concerns. It is precisely in the recognition of this large and nuanced domain of expressive concerns that the chief benefit of incorporating the expressive idea into the discussion of rational accounts of politics may be seen. Expressive motivation is not a simple idea deployed to escape from the paradox of voting and resolve some seeming difficulties in the theory of voting; rather expressive motivations open up a new area of study which allow rational choice techniques to be employed in ways that more accurately reflect the meaning and symbolic significance of much political behaviour.
The next section is devoted to providing a more precise and useful definition of expressive choice, one that is capable of application in a variety of settings. In debating a variety of approaches to explaining voting in rational terms, Dowding concludes that the main reason why non-instrumental explanations for voting find little favour with some political scientists and political economists is not so much that they find it tautological or lacking in predictive power, but that the critics have a ‘desire for deeper reasons’.Footnote 5 An important aspect of the following assessment of the various theories of expressive action is whether they do more than simply state the possibility of an expressive motivation. That is, can they provide the required ‘deeper reasons’ that would underpin any particular expressive motivation? And, indeed, would such ‘deeper reasons’ satisfy a reasonable definition of what is required for a choice to be expressive?
In the section after the one that follows, we will survey the various theories of expressive choice that have emerged and the related empirical work. We identify three broad categories of expressive theories: relating in turn to expressing identity, expressing moral views, and expressing social pressures, ‘rational irrationality’ and self-delusion. With these three broad accounts in place, we will then turn to questions of the efficiency or inefficiency of expressive behaviour and its institutional implications. The final section offers concluding comments.
Expressive behaviour: towards a definition
A major reason why the idea of expressive behaviour has received so much attention in the analysis of voting is that its underlying logic seems both clear and attractive.Footnote 6 While the specific content of expressive choice is contentious, the basic idea of what it means for a choice to be expressive appears relatively uncontroversial. However, we suggest that this view is a little too optimistic, and that definitional aspects of expressive behaviour need rather more careful consideration. We also suggest that limiting attention to the voting context can be unhelpful in identifying a more general definition of expressive behaviour.
There is a basic difficulty in providing a simple statement of the distinction between instrumental and expressive accounts of behaviour within a rational framework. If an act is rational, it is explicable in terms of the achievement of some purpose, and such a purpose can be associated with (net) benefit or ‘utility’. In this very general sense, all rational action is ‘instrumental’: action is a means towards the achievement of specified purposes. However, the distinction between instrumental and expressive accounts of behaviour that concerns us here operates at a slightly finer granularity. First, distinguish between direct and indirect accounts of choice/action, where a direct account focuses attention on some property of the choice/act itself as the source of motivation, while the indirect account focuses on some more remote outcome that follows (logically, causally or probabilistically) from the choice/act. Secondly, within the class of direct accounts, distinguish between two types of benefit.Footnote 7 First, what we will refer to as consumption benefits. These are the kind of benefits that are familiar in any act of final consumption; when I eat a favourite fruit, there may be indirect motivations and explanations in terms of, say, the health benefits of eating fruit, but there may also be direct benefits in terms of the sheer pleasure in the act of eating the fruit. In contrast to these direct consumption benefits, consider the sub-class of direct benefits that derive not from the consumption aspect of the act/decision, but from its symbolic or representational aspect: not from the act, but from its meaning. It is this sub-class of direct benefits that are engaged in expressive accounts of behaviour.
We emphasize that not all direct or intrinsic benefits are ‘expressive’ in nature. Consider the link between the basic idea of expressive voting and Olson's classic Logic of Collective Action. In Olson's discussion, a fundamental contrast is between those benefits associated with the group that are dependent on collective action and subject to possible free-riding, and those benefits of group membership that are directly accessible to individuals and that can therefore act as selective incentives for individuals to join the group. These selective incentives play a very similar role in Olson's theory to the role played by expressive benefits in expressive voting theory – in both cases, they focus attention on the direct benefits that are individually accessible – but there is no sense in which Olson's selective incentives must be expressive in nature. Indeed, the standard examples are simple consumption benefits.Footnote 8 This is not to say that expressive ideas play no part; it may be that one selective incentive that relates to joining particular groups is the desire to identify with that group. The point here is simply that the link between the idea of a direct benefit that flows from the performance of an act and an expressive benefit is not automatic. The expressive idea identifies a subset of all possible direct, intrinsic or performative benefits.
With these ideas in place, we might attempt to construct comparative statements of the instrumental and expressive cases in what might be termed their ‘pure’ forms. In the instrumental case, acts/choices have no symbolism or meaning in themselves, so that individuals act/choose in ways that respond only to indirect benefits and direct consumption benefits. They act/choose in such a way that the acts/choices maximally serve their interests (whether narrowly or broadly defined). By contrast, in the purely expressive case, the individual responds only to the meaning of the act/choice, so as to act/choose in a way that maximally expresses the individual. Indirect and consumption benefits are irrelevant in such a purely expressive account; individuals undertake action Z in order simply to expresss some relevant meaning bound up in Z (and, perhaps, being seen to Z).Footnote 9
But, of course, ‘pure’ cases are rare. Most cases involve both instrumental and expressive considerations. In the ‘all-things-considered’ choice, individuals respond to all types of benefit, giving each the appropriate weight. The various considerations may reinforce each other or pull in different directions, so that there may be trade-offs between expressive and instrumental considerations, just as there may be trade-offs between different instrumental considerations, or between different expressive considerations. It should be clear that, in our view, expressive concerns are best conceptualized as a proper subset of the concerns that will be considered in a fully rational analysis of the all-things-considered evaluation and choice; and exactly the same may be said for instrumental concerns. Each is a part of the whole.Footnote 10
Just as the idea of expressive benefits should be seen as narrower than the idea of direct benefits, so the simple idea of ‘expression’ may be in another way too broad. Consider my behaviour when I accidentally hit my thumb with a hammer. I may cry out in pain, and that cry may naturally be termed an expression of my pain. But is this the sort of expression that we are concerned with in developing a theory of expressive behaviour? Our approach is to place the theory of expressive behaviour firmly within the rational choice approach, so that the types of expression that concern us are those that relate to motivating rational action. We might term this subset of expressions the set of motivating expressions. Now, in the case of my hammering, the possibility of a painful blow certainly provides me with reason to be careful, but the idea of the expression of pain (as distinct from the pain itself) plays no obvious motivational role.Footnote 11
So far, then, we have done no more than mark out the territory that we believe the theory of rationally expressive behaviour seeks to occupy. It aims to focus attention on the potential motivating effects of certain forms of expression that attach directly to actions or choices. The theory may be seen both as capable of offering distinctive understandings of particular situations (such as voting) where instrumental rational choice theory fails to convince; and of contributing to the more general understanding of rationality in a wider range of situations. As we have stressed, the all-things-considered rational choice should be seen as including all direct and indirect benefits (and costs). Expressive considerations may not be relevant in all choice situations, or may be of vanishingly small importance in some situations, but the general idea that expressive ideas may be relevant alongside more instrumental considerations is important; not least since it points to the idea that expressive and instrumental motivations are best seen as joint inputs into an overall analysis of behaviour, rather than alternative models.
In any given situation where both expressive and instrumental considerations are relevant, the action that would be chosen on expressive grounds may differ from the action that would be chosen on either instrumental grounds or as the result of the all-things-considered choice, but this is not necessarily so. To the extent that expressive and instrumental arguments pull in different directions, it is common to locate the trade-off between them as part of standard economic price theory.Footnote 12 We should expect to see essentially expressive choice triggered more readily in those situations where the ‘price’ of such behaviour is low; where ‘price’ is construed in terms of instrumental considerations forgone. In cases where the decision maker faces all-things-considered choice, expressive considerations will be relevant but the price that has to be paid to choose expressively in terms of the more instrumental benefits forgone may be higher or lower depending on the circumstances. In cases such as voting, where action can be argued to be divorced from further outcomes, the price of expressive behaviour is low, and we should expect it to dominate.Footnote 13
However, the voting situation presents a number of features that frame the definition of expressive behaviour, and we should be careful in teasing them apart. Here, we consider three further issues relevant to defining expressive behaviour: the first concerns the structure of collective action problems and the link to the general idea of inconsequential behaviour; the second concerns the potential audience for expressive behaviour; and the third concerns the relationship between expressed views and true views.
As noted, the standard argument for expressive voting starts from the presumption that voting in large-scale elections poses a collective action problem that renders individual votes essentially irrelevant to the determination of the outcome. It might then be supposed that expressive behaviour is limited to situations that may be characterized as collective action problems. But that would be a mistake.
For example, consider a situation in which an individual decides to write to a local newspaper to complain about some feature of local life and perhaps to suggest a remedy. How might this behaviour be explained? The standard instrumental line would have to be that the individual sees this as a means of generating a desirable outcome: perhaps the implementation of the suggested remedy. But the expressive line would suggest that the behaviour is best understood simply in terms of venting dissatisfaction, or identifying with the critical position, and that the observed behaviour might be expressively rational, even if the individual knew in advance that writing to a newspaper would have absolutely no impact on the situation about which they had complained. Here, several aspects of the standard voting story are absent: there is no background belief that decision making relative to the particular aspect of local life is made by reference to the aggregate level of letter writing and there is no necessary reference to interactions amongst large numbers of individuals, so this is not a collective action problem. In this case, the act of letter writing just is inconsequential (ex hypothesi), and it is this fact that both invites and suggests an expressive account.
This example suggests that the case of expressive voting may not always be a good guide to the more general class of expressive behaviour, in so far as the case of voting focuses our attention on issues such as collective action and the large number case which are not defining aspects of expressive choice.Footnote 14 And this point is at least reinforced by considering other analogies and examples that are commonly used within the expressive voting literature. Brennan and Buchanan as well as Brennan and Lomasky compare expressive voting to cheering at a sports match, and to dinner-party conversation.Footnote 15 Both examples are taken to describe situations in which the action to be explained is both ‘inconsequential’ and literally ‘expressive’ in terms of the use of voice. While we agree that these are examples of expressive behaviour, we are not convinced of the analogy with voting. In the dinner-party case, the general idea is simply that I might express support for ideas, positions or policies that I might not support instrumentally. But here again, a number of key features of the voting case are missing. There is no valid assumption that the aggregate of dinner-party conversations is causally effective in bringing about any outcome: there is no necessary reference to the size of the dinner party (or the number of dinner parties) and so on. All there is, and all that there needs to be, is a plausible claim that dinner-party statements are (largely) inconsequential with respect to the apparent content of those statements. If I say ‘I support X’, it has no significant effect on whether X comes about. And the whole point of the dinner-party analogy, presumably, is that this inconsequential setting may produce statements that are at variance with the actions that the same individual would undertake under instrumental choice. In this way, the dinner-party story speaks directly to the ‘How to vote?’ question, rather than to the ‘Why vote?’ question, in that it addresses the question of what we say at dinner parties, rather than why we speak at all.
In the case of cheering at sports events, even if we accept that cheering may be instrumentally effective in increasing the probability that your team wins to at least some extent, and that the large numbers involved at a sports venue generates a collective action, free-rider problem in regard to adding your voice to the cheering,Footnote 16 it seems that the most that can be offered here is an account that addresses the ‘Why cheer?’ question rather than the ‘How to cheer?’ question. After all, it is hard to see a Manchester United fan cheering against his team and explaining his behaviour on the expressive grounds that it did not make an instrumental difference!
The point here, then, is that the construction of a case which approximates ‘pure’ expressive choice requires the suppression of all (or almost all) instrumental considerations. Only when action is inconsequential, in this sense, will action be purely expressive. Collective action problems may be one way of generating examples, but they are not the only way. Furthermore, this does not imply that expressive considerations are only present or relevant in such ‘pure’ settings. Imagine a member of a committee facing a decision on some issue, and start from the premise that if one focused only on instrumental considerations this person would favour action X, but that if one focused only on expressive considerations she would favour action Y. Now, one question is, what action would this person favour all-things-considered? That is, what would she choose if she were fully decisive? Note that the answer here is not necessarily X. Expressive benefits remain benefits even if the individual faces consequential choice and it may be that the expressive costs associated with X are just too high. If the issue is expressively salient, it might be that she would choose Y or it might be that once all things have been considered and given their due weight, she would choose some third option Z that offers a better balance of instrumental and expressive benefits than either ‘pure’ choice. Even in purely private choice (that is, in a committee of one), an individual may value an expressive concern more highly than instrumental concerns and so take decisions that are best understood in expressive terms even though the individual was fully decisive.
But a second question is how the committee member's action might vary in different institutional settings, particularly as the size of the committee varies. Here, we have a clearer answer. If we start with a committee of one member, we must conclude that she would adopt her all-things-considered choice, whatever that is. Increasing the committee size will emphasize the expressive at the expense of the instrumental, so that beyond some limit she will support Y.
These two questions, and their answers, point to two different aspects of expressive choice. First, including issues of expressive concern into an analysis of rational political choice may yield different outcomes from those that would be revealed by a more rigidly instrumental approach, even if the institutional circumstances were not such as to approximate to the case of ‘pure’ expressive choice. If expressions matter, they can affect behaviour; and building this feature into our definitions and analysis may be important in many cases. Secondly, institutional circumstances will be important in influencing the balance between instrumental and expressive issues as they appear to actors, and so directly influence behaviour under those institutional circumstances. In this way, the articulation of the expressive/instrumental distinction helps us to approach the idea of the endogeneity of political behaviour with respect to political institutions.
In cases such as dinner-party conversation, it may appear that the mere fact that the act under consideration is a speech act is sufficient to engage with the expressive idea, but this is misleading. Consider the case of an election speech made by a professional politician. Here, it seems clear that the motivation for the speech, and for the content of the speech, is instrumental and it is the instrumental nature of the speech that raises the potential for the content of the speech to be deceptive, since the politician will face an incentive to say whatever will increase the probability of his election, and to say different things to different audiences. Nevertheless, there may be a connection between the (instrumental) use of political rhetoric by politicians and expressive ideas. If the politician recognizes the potentially expressive nature of voting decisions by the electorate, he will use his speech to signal in relation to those expressive concerns. The use of emotive language, appeals to patriotism or morality, as well as tactics such as the vilification of opponents may all provide examples. In such cases, political rhetoric, which may or may not be deceptive, may be viewed as the vehicle for expressive concerns.Footnote 17 The distinction between an election speech and dinner party conversation illustrates the difference between recognizing an expressive aspect of an instrumental action, and offering an essentially expressive explanation. The case of the election speech also illustrates the fact that it is often the interaction between expressive and instrumental considerations that is important.
Our second concern relates to the visibility of the behaviour under consideration. While the act of voting is at least somewhat public, the secrecy of the ballot typically assures us that the content of our vote is private. By contrast, many of the examples of expressive behaviour already given (writing to newspapers, cheering and dinner party conversation) are all essentially public, so that we can immediately see the possibility of them being expressive in nature. This raises the question of whether an act must be at least somewhat public in order to qualify as an expressive act; in short, does an expressive act require an audience?
One possibility is that the actor may form her own audience; i.e. that expression can, at least sometimes, be self-directed. This is sometimes linked with the argument concerning expression as a form of identification, where identifying with some position or cause combines elements of self-identification and identifying oneself to others. In the context of voting, the lack of an obvious audience in a secret ballot might be thought to undermine the incentive to vote expressively, but here we might point to a slightly more complicated, expressive argument. Suppose that I have an expressive desire to support a particular political position, or particular political cause. Even if we admit that merely voting for that position or cause cannot in itself count as ‘expressing’ myself because of a lack of relevant audience, I can surely hold that I may wish to express my political views in all sorts of public arenas, and the only way in which I can make these expressions while maintaining a degree of internal consistency and integrity is to vote expressively. Here, then, the vote is not itself expressive, but it is a precondition for expression.Footnote 18
Another version of the idea that you may be your own audience in matters of expression arises in the discussion of cases such as philanthropy. For example, Andreoni argues that individuals derive private benefits from a ‘warm glow’ associated with giving, and it is for this reason that philanthropy occurs on a much greater scale than standard economic theory would lead us to expect. To the extent that expressive voting is analogous to the ‘warm glow’ in charitable giving, it is clear that the relevant audience is oneself.
While it seems reasonable to allow the possibility of being your own audience, is it plausible to allow the possibility of there being no audience at all? Here, we think that the answer is no, although we accept that this is largely a matter of stipulation. It is difficult to see how the idea of a motivating expression can generate the required motivational force if it is has no possible audience. Note that we are here adding the rider of a ‘possible’ audience; it may well be the case that a motivating expression operates on the basis of an intended (or perhaps even a hoped for) audience that never in fact materializes, so that it is the intended possible audience that matters in building an explanation of the underlying behaviour, and the fact that there was no actual audience may be neither here nor there.
Our third concern relates to the issue of whether expressed views hold any particular relationship to truly held views. There is much debate to be had in this area, but we offer some simple initial thoughts. First, it seems inappropriate to begin from a position that identifies either purely instrumental or purely expressive preferences/views/opinions with true preferences/views/opinions as a matter of definition. The more reasonable and less restrictive starting point seems to be one that recognizes that any individual at any time is likely to hold a range of preferences, views and opinions – both instrumental and expressive – where there is no necessary requirement of absolute coherence. From this starting point, it might seem that a context that brings both instrumental and expressive considerations to bear on decision making, each with its appropriate weight, is one which would allow the individual to reach an all-things-considered decision that might be as close as we are likely to get to reflecting some idea of ‘true’ or ‘fully considered’ preferences/views/opinions.Footnote 19 And in circumstances that privilege either instrumental or expressive concerns at the expense of the other, we are likely to reveal only a limited subset of the full range of preferences/views/opinions. In this sense, neither instrumental nor expressive preferences may be taken as ‘true’, while each reflects an element of some underlying truth. Similarly, both instrumental and expressive actions may be deceptive in the sense of being intended to deceive. We will return to discuss issues of deception and self-deception in a later section.
So, at this stage, we offer the following three defining aspects of expressive behaviour and the distinction between expressive and instrumental behaviour.
(1) Behaviour is expressive to the extent that it reflects, wholly or partly, underlying concerns that derive directly from the meaning or symbolic significance of actions or choices themselves, rather than their indirect consequences or consumption benefits. Expressive concerns sit alongside instrumental concerns within a structure of overall rational choice.Footnote 20 Institutional contexts will influence the balance between instrumental and expressive considerations in particular cases. Pure cases arise when one class of consideration is entirely suppressed. More generally, there may be a trade-off between instrumental and expressive considerations.
(2) Expressive behaviour is to be understood relative to an audience, either directly or indirectly, intended or anticipated, and allowing for the fact that an individual may, in at least some circumstances, be their own audience. This is to indicate that the specification of the intended audience, as well as the specification of the actor, may be required to understand expressive behaviour fully. I may have good reasons to express myself very differently to different audiences, even though my underlying concerns (both instrumental and expressive) are constant.
(3) Both expressive and instrumental preferences and beliefs may be ‘true’ in the sense that they are held on the basis of full information and full consideration, even when they are in conflict. Neither instrumental nor expressive concerns should be viewed as definitionally more foundational than the other. The normative status of expressive (or indeed, instrumental) behaviour is a matter for further analysis, and will need to account for the possibility of distorted or manipulated behaviour.
It should be clear that these three statements do not serve to characterize expressive motivations or expressive behaviour fully – they serve only to provide a structure within which such expressive behaviour can be understood and analysed. In order to complete the definitional exercise, the fundamental questions of the content of expressive motivations must be confronted.
Theories of expressive choice
We now present a survey of the various substantive theories of expressive choice that have emerged, and the empirical work associated with them. As already noted, most of the work on expressive choice has been developed in the context of the discussion of voting, but in what follows reference will be made to theories and applications that bear no direct relationship to mass elections.
We begin with a version of the expressive account which provides a reason for voting, but carries no implications for how to vote. This is the idea of ‘expressive choice as doing one's duty’ developed by Riker and Ordeshook.Footnote 21 This approach proposes that voters express their respect for duty through voting. However, since there is clearly no duty to vote for any particular candidate or option, duty itself can have no impact on how to vote. On this account, the expressive value of ‘doing one's duty’ gets around the paradox of voting without challenging the results of the standard instrumental model in relation to electoral outcomes. Of course, when applied outside the field of voting, the idea of duty as an expressive motivator of actions may carry a wide range of different implications depending on what is considered to fall within the remit of ‘duty’. We will return to the link between expressive choice and moral choice below.
The version of expressive voting theory based on civic duty seems to draw some support from the evidence for strategic voting that is normally interpreted in instrumental terms.Footnote 22 This tells us that where a voter ranks candidates A, B, C in order of all-things-considered desirability, but where A is known to have no chance of winning the election, a strategic voter will vote for B rather than A in an attempt to prevent C from winning. This certainly seems to be an instrumental explanation for strategic voting, since the voter would seem to be committed to a logic which relies on the possibility that their vote is decisive so that failure to vote for B might allow their least preferred candidate C to win. Theories of expressive choice that also address the how (and not just why) question in relation to voting might therefore appear to have some difficulty explaining strategic voting. However, Brennan rebuts the claim that evidence of strategic voting provides evidence against expressive voting and advances two arguments that might reconcile strategic voting and expressive voting.Footnote 23 First, that voting is a serious undertaking and that the voter might consider it frivolous and irresponsible to vote for A, if A is widely thought to have no chance of winning. Secondly, expressive choice can be about booing as well as cheering. There may be greater expressive value in booing for C (by voting for B) than in cheering for A.
Brennan's defence of an expressive understanding of strategic voting seems to accept the idea that the expressive ranking of A, B and C is the same as the all-things-considered ranking. But clearly this need not be the case. An alternative explanation of apparently strategic voting might rely on differences in these rankings. Therefore, a particular pattern of voting might appear to be ‘strategic’ when considered in terms of the all-things-considered ranking, but it is revealed to be straightforwardly rational when considered in terms of the expressive ranking alone (which would still allow for Brennan's possibility of booing rather than cheering). This raises the questions of what factors are likely to determine an expressive ranking in the first instance, and under what circumstances instrumental and expressive rankings are likely to diverge. In order to categorize expressive accounts in terms of the broad nature of their claims regarding the content of expression, we begin by considering variations on the theme of expressing identity, before considering the possibility that morality may provide the relevant content for expressive behaviour. Finally, we consider the idea that the content of expression may derive from social pressures, ignorance or illusion. With these three broad accounts in place, we will then turn to questions of the efficiency or inefficiency of expressive behaviour and its institutional implications.
Expressive Choice as Identity Based
The general idea of considerations of identity playing an important role in the economic analysis of decision making has developed in recent years.Footnote 24 The link with expressive behaviour is that some actions directly express the actor's identity (or the identity they wish to project) and this provides a route to explaining those actions. As already noted, Hillman defines all expressive behaviour as being associated with acts motivated to confirm identity and builds a model that explores the trade-off between expressive and instrumental motivations in a variety of settings.Footnote 25 Here, we take the rather different route of considering several different aspects of what it may mean to express identity.
(1) Social identification
Schuessler proposed the idea that what motivates voters is how many others vote for a particular option or candidate, and who the other voters are.Footnote 26 In this way, identity is confirmed through association with specific groups of others. In simple terms, voting for X identifies you with the set of people who vote for X. Numbers may matter because you would not wish to identify with too small a group, but also because too large a group may dilute the value of the expressive attachment. But the particular identity of the voters that you seek to identify with, rather than just their number, may be more relevant. If all voters are seeking purely to associate themselves with groups of other voters, using the candidates or electoral options only as points around which to congregate as a mutually identifying club, there may be many possible equilibria, with the prospect of instability, tipping points and bandwagon effects. However, if there are at least some voters who might be considered as partisans, in the sense that their motivation for voting points to voting for a specific option or candidate, these partisans may provide focal points around which others cluster, thus reducing the tendency to instability.
Candidates and political parties, faced with such voters, have a clear incentive to appeal to groups that would also provide them with a winning level of support. Generally, parties will want to present themselves as a club with an attractive combination of membership type and membership numbers. A prediction deriving from Schuessler's approach is that expressive behaviour can help to explain the impact of negative campaigning and the polarization of voters. Negative campaigns focus on identifying the character of other parties, and attacking that character so as to make it seem an unattractive club to join. Seen from the perspective of any single party, negative campaigning will be a useful weapon in reducing the attractiveness of rival clubs. However, negative campaigning by all parties will make all of the available ‘clubs’ less attractive, implying lower turnout overall. And, of course, it will be the least committed members of the electorate who will be dissuaded from voting by negative campaigning and therefore the remaining voters are more likely to be partisans, so that parties are more likely to be polarized.
Rotemberg provides an alternative account in which voters identify with individuals whom they agree with.Footnote 27 This model builds on two psychological tendencies. First, people tend to be altruistic towards individuals who agree with them; and, secondly, individuals gain in self-esteem from discovering agreement. One challenge to the idea that voting is expressive is the correlation between voter turnout and the closeness of the election. This correlation might suggest that voters are behaving instrumentally and the higher turnout reflects an increase in the (admittedly small) probability of being decisive. Interestingly, Rotemberg argues that this correlation can be explained by reference to the psychological tendencies that motivate his model, as voters receive greater psychological benefits in close rather than one-sided elections. In this way, Rotemberg argues, the correlation between the closeness of the election and turnout can be provided with a basis in the logic of expressive voting.
Hamlin and Jennings extend the idea of the motivating force of social identification from the domain of voting to the more general setting of the formation of political groups.Footnote 28 Here, the target is not just the emergence of groups, but the explanation of conflict between such groups, and the emergence of group leaders who may mediate that conflict. Brennan and Pettit provide a further variant on the theme by exploring the idea of esteem (and self-esteem) as a motivator of behaviour.Footnote 29 On their account, it is the quest for the good opinion of others (or oneself) that provides a basic driving force behind behaviour that might be considered expressive.
(2) Identification with parties or candidates
Brennan and Hamlin put forward the idea that expressive choice may be related more to identifying directly with parties, candidates or political positions rather than with groups of other voters.Footnote 30 They suggest that elections may be dominated by issues that capture expressive interest, and that these issues may exist in a domain that differs from that which accounts for instrumental concerns. But even if the domain of expressive concerns is similar to the domain of instrumental concerns, the distribution of expressive preferences may differ markedly from the distribution of instrumental preferences.
In the Brennan and Hamlin model, voters vote for positions/candidates that are sufficiently close to their own ideal point in the expressive domain. If there are no candidates within a certain distance of their ideal point, they do not vote. As in the standard spatial model, candidates can adjust their platforms strategically to attract voters. This leads to a result that is roughly equivalent to the standard median voter theorem in the setting of instrumental voting, although it emphasizes the significance of the modal voter rather than the median voter.
However, the Brennan and Hamlin argument leads to an empirical prediction that distinguishes their expressive model from the parallel instrumental model. Although parties/candidates face similar incentives in the two models, and will tend to converge on ‘moderate’ positions, the decision of whether to vote is very different in the two models. Brennan and Hamlin predict that, in the expressive case, moderates will vote with extremists abstaining, while the standard instrumental model suggests the opposite. In an instrumental model, moderates are non-voters because they are broadly indifferent across the alternative candidates, while in an expressive model extremists are non-voters because they are alienated.Footnote 31
Greene and Nelson set out to test this prediction and find that extremists are as likely to vote as moderates and thus argue that Brennan and Hamlin's prediction does not hold.Footnote 32 But Greene and Nelson effectively rule out instrumental voting by assumption, so that their model is inconsistent with Brennan and Hamlin's prediction at the level of design. A more appropriate test of the Brennan and Hamlin prediction would be to check the nature of motivation for extremists and moderates. If the former are instrumentally motivated and the latter expressively motivated, then the Brennan and Hamlin prediction is supported. Drinkwater and Jennings conduct such an analysis and find support for the Brennan and Hamlin prediction.Footnote 33 Calcagno and Westley also find evidence in support of Brennan and Hamlin's thesis by considering the effect of primaries on turnout in general elections.Footnote 34 Closed primaries lead to greater divergence between general election candidates than open primaries. An instrumental account of voting predicts that turnout should be higher in states with closed primaries, while the expressive account of voting would predict that the greater identity with the convergent candidates should lead to higher turnout in states with open primaries. Calcagno and Westley find that turnout in a general election is greater the more open the primaries.Footnote 35
The normative implications of the identity-based models surveyed so far are unclear. They predict limited convergence in an, as yet, unspecified expressive dimension. One strong implication from the Brennan and Hamlin model, however, is that global instability (due to cycling) is implausible. Quite simply, political positions that are too far from voter's expressive preferences will never be selected.
Expressive Choice as Moral Choice
(1) Voting for moral or desirable characteristics
In a further paper, Brennan and Hamlin reconsider the argument for representative democracy over direct democracy in expressive terms.Footnote 36 In instrumental theory, the argument for representative democracy is essentially of a second-best nature: direct democracy would be superior to representative democracy but for the costs involved and other issues of feasibility. The expressive reconsideration is two-pronged. First, direct democracy may be rendered problematic by inefficient expressive choice on policy issues (discussed below). But secondly, and more positively, in the case of representative democracy, voters may pick out moral or other desirable qualities in their representatives that particularly suit them for political office, so that representatives are on average more moral or more talented or more competent than would be the case if they were drawn from the population in some way that was statistically representative. The institutional implications are striking.Footnote 37 In contrast to the heavy emphasis on imposing constraints on government that normally characterizes the public choice literature,Footnote 38 there is less need to be concerned about the principal/agent problem between politicians and the citizenry. The focus of attention shifts, at least to some extent, from the imposition of constitutional constraints on politicians assumed to be self-interested, towards the design of institutions that select politicians with appropriate characteristics.
The idea of the expressive selection of moral, competent or otherwise desirable politicians or policies also links with the idea of understanding individual political motivations more in terms of dispositions and commitments rather than pure preferences.Footnote 39 The combination of politicians who can credibly commit to particular dispositions, and voters who select politicians at least partly on the basis of their disposition, reinforces the idea that constitutional arrangements that empower politicians rather than constrain them may be warranted.
But the link from this aspect of the expressive literature to the wider debate on constitutional design has another element. To the extent that constitutions themselves are approved by popular voting, we might expect the expressive argument to apply to the choice of constitutional provisions themselves, as well as the operation of everyday politics within the chosen constitutional structures.Footnote 40 This link reminds us that constitutions are themselves the outcomes of political processes and must be seen as endogenous.
(2) Voting for merit goods
Brennan and Lomasky argue that a distinction can be made between ‘expressive choice’ (political choice), ‘instrumental choice’ (market choice) and fully reflective or all-things-considered choice.Footnote 41 The normative appeal of the idea of individual autonomy (and consumer sovereignty) rests on the basis of fully reflective choice, but choices reflect institutional settings, so that the institutional framing of decisions is significant in explaining the decision. In choosing between institutional settings, politics versus the market, for example, we should be aware that both politics and the market are ‘frames’ and that, in at least some cases, expressive choice in a political frame may yield a closer approximation to fully reflective choice than instrumental choice in a market frame.
How then do we identify such cases? One example relates to merit goods.Footnote 42 Brennan and Lomasky argue that the more expressive political environment may be more appropriate than the market in such cases, since the market can be expected to consistently under-supply merit goods. Furthermore, to the extent that a political mechanism is used, the act of contemplation prior to a vote may bring more reflective preferences to the fore, such that they may have a subsequent effect on how individuals behave in market choices. This implies that reflection contributes to an expressive choice, which may in turn influence instrumental choice, pointing to a possible route by which the interaction between expressive and instrumental choices may be mutually informative.
(3) Voting for redistribution: generosity and altruism
Redistributive taxation may result from voting even where each individual is narrowly self-interested and purely instrumental.Footnote 43 But there is a major strand of the literature on redistribution that starts from the presumption of some degree of altruism or generosity understood as a concern for the welfare, or income level, of others.Footnote 44 Once a degree of altruism is in place, the expressive possibility is clear, and the expressive aspect of political choice may be important. The basic story, then, is that if we compare the situation in which redistribution is a matter of private philanthropy with the otherwise similar situation in which a public redistributive scheme may be enacted via a popular vote, we would expect significantly more redistribution in the latter case. And this may happen for three reasons: first and most obviously, some voters will expect to benefit directly from redistribution and will have an instrumental reason to vote for it (although to the extent that the election is large, this may not be sufficient to cause them to vote). Secondly, some of those who expect to contribute to the redistributive scheme but who are altruistic to a sufficient extent will have reason to vote to achieve their all-things-considered desired outcome (although, again, this may not be sufficient to cause them to vote). But thirdly, all of those who are altruistic to at least some degree will recognize the opportunity to express their altruism, and here the fact that the election is large will ensure that there is little cost to such expression and so encourage them to vote.
The argument could be taken further, to the point where no actual altruistic motivation is required for individuals to vote for redistribution since such a vote offers uncharitable individuals the (cheap) opportunity to appear charitable.Footnote 45 This further step of the argument takes us back to the idea of voting to confirm identity, and the possibility of projecting an identity that is flattering but deceptive. Hillman extends this argument further, beyond the idea of the uncharitable voting for redistribution to the idea of private charitable giving itself, particularly in the context of international aid.Footnote 46 He argues that private philanthropy (rather than voting for public philanthropy) may reflect an expressive attempt to self-identify as a generous person, which does not require genuine generosity or altruism as a motivation.
There have been numerous empirical studies of the expressive case for voting for redistribution. Carter and Guerete find only weak evidence, but Fischer builds on that study to find considerably stronger evidence in support of the expressive account. Later papers confirm this support.Footnote 47 Interestingly, Tyran does not find direct support for the expressive view of redistribution, but does find support for the view that voters tend to approve proposals if they expect others to support them.Footnote 48
Social Pressure, Information and Illusion
We have argued that the idea of true preferences need play no role in the definition of expressive behaviour. Consider the issue of redistribution, where we might observe an individual choosing ‘selfishly’ in their private or market-based activities, but choosing more ‘benevolently’ in the political arena. It would be difficult, perhaps impossible, to identify whether the ‘true’ preferences of the individual are ‘selfish’ or ‘benevolent’ (or some mix of the two), since we argue that each institutional ‘frame’ elicits a particular behaviour from some underlying set of preferences (both instrumental and expressive), rather than directly ‘revealing’ true preferences. Given that market choice is decisive and responsible, it might be argued that it reflects an all-things-considered perspective and thus does reveal ‘true’ preferences. Indeed, this would seem to be the position often adopted in the literature. However, if every institutional setting is a potentially distorting frame, there may be no neutral setting which allows ‘true’ underlying preferences to be directly revealed. This point is highlighted further if we reconsider the discussion regarding merit goods where we argued that the political arena may be the more likely arena to elicit ‘true’ preferences in at least some cases.
A background assumption here is that behaviour is equally informed, free and autonomous whether it is in a market setting or in a political setting, and it is partly because of this symmetry assumption that we conclude that neither setting is superior in providing a ‘fully revealing’ context for choice. If it could be argued that one setting was systematically inferior to the other in terms of information, or autonomy, or in some other relevant way, this would certainly be relevant to the overall consideration of the relationship between preferences revealed under any particular institutional framework and ‘true’ preferences. We will now consider examples where constraints are imposed upon political action such that political behaviour which may seem to be expressive may also be interpreted as artificially or exogenously constrained. In these examples, the statement that expressive behaviour may not reflect ‘true’ expressive preferences is explicitly built into the discussion.Footnote 49
Consider again the distinction between examples like cheering at a sports match or participating in a dinner party conversation, on the one hand, and voting on the other. As already noted, one key difference lies in the identification of an audience. Where a specific audience is in place, it may be, as Kuran argues, that individuals are pressurized to conform to that audience.Footnote 50 On the one hand, the anonymity of voting may reduce its expressive significance by limiting the direct audience; on the other hand, anonymity seems to protect the voter from social pressure, thus increasing the likelihood that the expressive content of a vote is in some relevant sense ‘authentic’.
Where expressions are public, there may be hidden costs in the form of social pressures that distort the expression made. And while the secret ballot may provide some insurance against such pressures, voting is not the only politically relevant form of expressive behaviour. Many actions such as attending demonstrations or political meetings, engaging in political debate, indeed almost all aspects of ‘political participation’ are likely to engage expressive behaviour.Footnote 51 Since these forms of behaviour are often essentially public, the question arises as to whether there is a way of separating authentic from socially constrained expressions.
In the examples explored by Kuran (such as the support of repressive regimes in the pre-1989 East European countries), political equilibrium is highly unstable as the views expressed are not truly held and we should expect bandwagon effects (as happened in the post-1989 East European countries). In contrast, one might expect views that are truly held to be more stable. So, the stability of political equilibria, where equilibrium clearly features mass support and thus expressive behaviour, might offer a potential test of how authentic the underlying political expression is.
If we observe individuals engaged in collective action where we are confident that the views expressed are not the result of distorting social pressure, can we be sure that the opinions displayed are expressive? There are at least two further challenges. These are the roles played by information and illusion.
In addition to the paradox of voting, Downs famously drew attention to the idea of rational ignorance.Footnote 52 Given the low probability of being decisive, there is a low incentive to become informed about issues. Caplan extends this idea by developing the idea of ‘rational irrationality’ to suggest that while voters may happily incur the low costs of voting, they may also express ill-informed and biased opinions which, when aggregated, can lead to inefficient policies.Footnote 53 Of course, it is the claim of bias, rather than the simple claim of relatively ill-informed opinion that is important to this argument. A crucial finding in Caplan's supporting empirical work is that citizens untrained in economics have systematically biased beliefs relative to those trained in economics.Footnote 54 He argues that the untrained hold these biases because people desire certain beliefs and will hold these beliefs even when they run contrary to evidence or expert opinion.
Beliefs are viewed as normal goods: when their price is low, demand for them will be high. The price will be low in situations where the belief has no direct day-to-day implications for the individual concerned. There will be many areas in which individuals face low-cost decisions between alternative beliefs and may ‘choose’ their beliefs to fit with their other preferences and prejudices. But if individuals vote by expressing such beliefs, they may have important social and political implications in the aggregate.
Rational irrationality differs from the simple case of rational ignorance because of the role of the idea of choosing beliefs and the possibility that this leads to bias rather than just ignorance. Rational irrationality is clearly a member of the same family of ideas as expressiveness, as the argument is driven by the underlying idea of the inconsequential nature of certain choices, but Caplan is careful to distinguish the two: ‘In expressive voting theory, voters know that feel-good policies are ineffective. Expressive voters do not embrace dubious or absurd beliefs about the world … In contrast, rationally irrational voters believe that feel-good policies work.’Footnote 55
Therefore, a further condition would need to be fulfilled in order to judge a vote to be expressive of true preferences rather than rationally irrational ones; we would need to check how well informed the voter is. One suspects that this issue may be similar to social pressure. If voting is both expressive and ‘rationally irrational’, making information available might be expected to result in a rapid and significant shift in the political equilibrium. If, by contrast, voting is an expression of truly-held expressive preferences, the political equilibrium will be much more stable.Footnote 56
A further challenge to a conclusion that voting is expressive of truly held preferences stems from Akerlof's analysis of illusion.Footnote 57 Our everyday interactions may influence the preferences that we bring to politics. For example, where a market failure exists due to free-riding, those who engage in free-riding may justify it to themselves through the process of cognitive dissonance, so that when an attempt is made to correct the market failure through the political process, voters may vote to maintain the inefficiency since they have already justified their actions to themselves. This is a particularly thorny problem. Voters would appear to be expressing their true beliefs, but these beliefs are the result of a prior act of self-deception. Once again, we would simply note that any attempt to locate ‘true’ preferences is likely to run into a wide range of difficulties of this type, involving the endogeneity of preferences through both voluntary and involuntary processes. While we certainly accept the difficulty in saying anything very clear about true preferences, we do not think that this difficulty arises, or is seriously exacerbated, as a result of distinguishing between expressive and instrumental behaviour.
Inefficient Expressive Choice?
If some arguments paint a picture of expressive behaviour that seems normatively benign and even desirable, there is also an alternative picture available that portrays expressive choice in terms of prejudice, fear and intemperate reaction. Two ideas emerge from this more negative conception of expressiveness.
A dark side of expressiveness seen as the confirmation of identity relates to the fact that identity is often confirmed by reference to a rival or ‘other’ group, perhaps in forms that result in inter-group conflict, so that the expressive nature of political choice may help to explain conflict which might be avoided by instrumental behaviour.Footnote 58 Kaempfer and Lowenburg explore the role of nationalist attachment in cases of international sanctions.Footnote 59 The traditional argument for economic sanctions has been an instrumental one: sanctions may bring favourable policy change in the target country by imposing, or threatening, economic harm. But the instrumental case for economic sanctions often appears weak. Kaempfer and Lowenburg argue that while some pressure groups in the sanctioning country gain instrumentally from sanctions, most citizens gain purely expressive benefits by taking a stand against the targeted country even though the collective stand may lead to high costs for these citizens. The imposition of sanctions may paradoxically strengthen the objectionable government in the targeted country as the citizens there expressively ‘rally around the flag’.Footnote 60 Glazer models the role of anger in party political competition and uses this to explain possible divergence in the position of the competing parties.Footnote 61 Tyran and Engelmann, in an experimental study, investigate consumer boycotts and find that consumers often approve of boycotts for essentially expressive reasons in situations where there are no effective instrumental reasons.Footnote 62
Expressive choice may also help to explain inefficient redistribution without recourse to complex explanations based on information asymmetry or transfers to maintain numbers within an interest group. Consider Acemoglu and Robinson's discussion of inefficient redistribution.Footnote 63 They focus on the inability to form binding commitments as the trigger which leads interest groups to seek redistribution inefficiently through specific policy interventions rather than efficiently through cash transfers. While this theory may explain a significant fraction of the support for inefficient transfers (namely the support by members of the recipient groups), it is also the case that such policies are often supported by wider groups of voters who are not direct beneficiaries of the policy, and who may actually be disadvantaged by such policies. Where an inefficient redistribution policy receives a high level of support, it would seem likely that a large proportion of those supporters are not material beneficiaries of the policy.Footnote 64 An expressive account offers an explanation for such support and also suggests why such voters may respond particularly positively to inefficient redistribution policies rather than cash transfers, since the relative transparency of cash transfers will make the cost of the policy more salient. A key point here is that political entrepreneurs, who make instrumental choices (because they are in decisive or directly consequential positions), may manipulate collectives to support positions that actually leave them worse off, but benefit the interest group that the political entrepreneur represents.Footnote 65
Clearly expressive voting and political behaviour can produce inefficient and even disastrous outcomes. There can be no sustainable claim that expressive behaviour is always a force for good in the world.Footnote 66 Equally, there can be no sustainable claim that expressive behaviour is always a force for bad in the world. This is an appropriate point to underline the idea that political outcomes are always the result of the interaction between expressive and instrumental behaviour. Even if many ordinary citizens can be expected to act expressively in many political situations, there will typically be some individuals who will face strong incentives to act instrumentally: not least professional politicians. To the extent that professional politicians are in the business of being elected, they will face incentives to appeal to voters in any way that is available to them, and this will include presenting themselves and their policy positions in expressively salient ways, and engaging in expressive rhetoric. This will involve a whole series of trade-offs. Some voters may be expressively drawn to candidates who exhibit certain characteristics that they wish to identify with, even where these characteristics may not relate directly to political positions or policies. Others may be expressively drawn to support moral positions. Still others may be drawn to express anger or display hostility to external groups – and so on. We might expect politicians of various types to emerge to reflect this range. The recognition of the relevance of expressive motivations does not make the analysis of politics simpler; rather, it shifts the debate away from a focus on the domain of interests (in all their diversity) and towards a focus on the domain of expressed opinions (in all their diversity).
Institutional Implications
Expressive choice provides a rather different perspective on the issue of institutional design. Once we recognize that at least some political behaviour may be expressive in nature, and that expressive concerns are of relevance, we must reconsider the more standard arguments for the design of political institutions.
The nature of this reconsideration is to institutionalize the more ‘positive’ aspects of expressive behaviour, while neutralizing the more ‘negative’ aspects. But this can only be translated into real institutional design/reform to the extent that we can identify structures and institutions within a society that are more likely to prompt some particular kinds of expressions as opposed to others. The interesting question then is the balance between welfare-increasing and welfare-decreasing aspects of expressive choice and whether there is anything that can be done by way of institutional design to select for the former and against the latter.
The emphasis on institutional design has formed the cornerstone of the normative approach taken by public choice and constitutional political economy. Since The Calculus of Consent was published, the argument has been made forcefully that political outcomes are best seen as functions of the particular political institutions and rules-of-the-game in place, and that the construction of an artificial social welfare function will not resolve disputes.Footnote 67 On this view, hope lies in finding more basic agreement on institutions and the rules of the game, as citizens might choose them behind a ‘veil of uncertainty’. This insight is not fundamentally altered by an awareness that much political behaviour is likely to be expressive in nature. All that would seem to be required is that the role of expressive choice is given full consideration when institutions are designed.Footnote 68
While recognizing the importance of expressive concerns in the normative analysis of institutional design, Brennan and Hamlin highlight the problem that follows from acknowledging the presence of expressive choice in the positive analysis of the process of institutional design.Footnote 69 If institutional or constitutional design is itself subject to a process that encourages expressive behaviour in circumstances where we have reason to believe that expressive views may depart from all-things-considered views, we have little reason to have confidence that appropriate institutions will emerge. Large numbers of citizens being asked to support a constitutional proposal in a referendum may reject it expressively even though they may have accepted it instrumentally, or on the basis of an all-things-considered evaluation. Brennan and Hamlin argue that subjecting constitutional proposals to a popular vote may undermine well-designed rules. Perhaps these proposals should be decided by small (but representative) groups, which might be more likely to take an all-things-considered view. Crampton and Farrant make explicit the potential problem that such a small group might design institutions that enrich themselves if they are not fully representative in a relevant sense.Footnote 70 Therefore, a trade-off may exist between the problem of expressiveness, on the one hand, and allowing too much room for the narrow self-interest of unrepresentative groups, on the other. More recently, Brennan and Hamlin point to the significance of written versus unwritten constitutions, in that written constitutions may provide more clearly specified rules, but are more likely to be infected with rhetorical appeal and heavy symbolism that may limit the operational efficacy of the constitution.Footnote 71 Written constitutions are to be seen as ‘expressive documents’ that are used to express identity or ideology as well as to specify the rules of the political game, and in this sense there may be some support for an emergent or unwritten constitution.
More optimistically, Jennings argues that if a constitution must be passed by referendum, then additional institutional apparatus may be required within that constitution that may not have been required from a purely instrumental perspective.Footnote 72 The argument is illustrated by reference to the 1998 Belfast Agreement, where it could be argued that aspects of that agreement were included primarily to stave off expressive rejection at the stage of the popular referendum.
We have focused on the idea that expressive concerns can be relevant at the constitutional level of institutional design as well as at the political level of policy choice within given institutional arrangements. But we began this section with the contrast between the principal–agent conception of the design of political institutions that is recommended by the standard instrumental account of political behaviour, and the broader conception that accompanies the expressive perspective. This should not be taken to indicate that the principal–agent idea becomes irrelevant in a more expressive world. Rather, it is no longer the only game in town.Footnote 73 At least two further ideas become relevant. The first is the general idea of selection and the role of political leaders.Footnote 74 Politicians, and leaders in all arenas, may be selected for particular characteristics, rather than simply as the embodiment of a package of policy measures. And to the extent that the characteristics selected for have normative dimensions (not least in terms of the motivations of candidates, or characteristics such as honesty), we might expect the selection mechanism to carry direct normative implications. On this reading, politics clearly runs the risk of institutionalizing an adverse selection problem, but also has the potential to institutionalize a more positive selection game.
The second general idea rendered salient by the discussion of expressive behaviour might be referred to as feedback or reinforcement effects. Once we recognize that variations in the institutional environment can be important in influencing both how we express our political preferences and what political preferences we express, it is a short step to building this idea into our thinking on institutional design, so that we may favour those institutional structures that elicit the most ‘positive’ or relevant aspects of our motivation. In some cases, this may imply institutions that avoid individually inconsequential behaviour and encourage what might be loosely referred to as ‘responsible’ choice. Such institutions may carry the flavour of the market. However, in other cases, it will be necessary and appropriate to design our institutions in such a way as to elicit political expressions, and here it may be important to structure our political institutions so as to avoid some aspects of expressive behaviour while encouraging others.
Concluding comments
In exploring the idea of expressive behaviour, we have both attempted to understand the essential structure of the expressive argument and to illustrate the great variety of substantive ideas that can be articulated within the expressive framework. We will not attempt to summarize or review the wide range of points made, but will restrict ourselves to just two final thoughts that reflect the two major aims identified in the introduction.
In working towards a definition of expressive behaviour, we have stressed the idea that expressive and instrumental aspects of motivation sit alongside each other as parts of all-things-considered motivation. Expressive preferences, like instrumental preferences, reflect a valid part of our motivations, but only a part. Different institutional settings may elicit responses that reflect different parts of our motivational structure, and we should recognize this fact both when analysing and evaluating the outcomes achieved under different institutional settings and when designing and reforming the institutions that frame our behaviour. Of course, some specific expressive concerns may be manipulated, or the result of deception (including self-deception) or social pressures; but the same is true of some instrumental concerns. There is nothing in the mere fact that a concern is expressive (or instrumental) that either grants that concern special normative salience, or implies lesser behavioural significance. Expressive motivations apply everywhere, not just in voting or in collective action problems where instrumental issues are muted by the inconsequential nature of individual behaviour. Of course, the relative weight given to expressive considerations will vary systematically with the institutional settings which frame behaviour, but some expressive concerns may be strong enough to determine behaviour in the most consequential settings.
The broad range of substantive ideas that might be relevant within the category of the expressive may, at first sight, seem to restrict the value of the expressive insight since there can be no easy argument that expressive behaviour is always of a particular type, or always carries a particular normative implication. However, closer consideration recognizes that the variety of ideas within the expressive domain is no more problematic than the variety of preferences in the instrumental domain. What is important is that the heterogeneity of expressive consideration, as well as the heterogeneity of more instrumental interests, is reflected in our political and institutional analysis at an appropriate level of granularity. Of course, this implies that politics is complex, and that the appropriate design of political institutions is subject to a large variety of considerations and trade-offs, but it also allows us to recognize the value of a range of institutional and political mechanisms that might appear rather mysterious under a purely instrumental understanding of politics. In this way, the expressive literature both adds to the diversity of political problems recognized within this branch of political economy, but also adds to the diversity of potential solutions to those problems.