The Problem of Narrow Identities

On the Possibility and Necessity of a Liberal Identity Politics

Very short summary: This essay argues that identity politics isn’t the problem from a liberal perspective; the trouble starts when identities narrow and turn exclusive. Drawing on social identity theory and a few economic models, I show how in‑group bias can appear without material conflict, how identification can override self‑interest, and why exclusivity can seem rational yet be normatively suspect. The claim is that liberal democracies need citizens to cultivate overlapping identities and settle conflicts through public reason, not allegiances that reward exclusivity over pluralism.

---

Talks of “identity politics” are nowadays common. Political practices, discourses, and ideas are increasingly tied to collective or social identities such as race, ethnicity, religion, or gender. Political claims are made based on people’s membership in groups and categories. Reasons justifying the acceptance or rejection of laws are grounded in the history of groups and norms attached to them. Identity politics is predominantly geared toward demands of recognition, often (though not exclusively) by groups whose members (and their ancestors) consider having been treated unfairly in the past and not granted the rights that should be theirs.

If the phenomenon is not new, the concept has taken a new meaning over the last couple of decades as it has been increasingly associated with criticisms of liberal democracy and the promotion of illiberal forms of politics. Even relatively balanced and sophisticated treatments such as Francis Fukuyama’s argue that identity politics challenges cosmopolitan liberal values that emphasize universal human attributes beyond divisions of class, gender, ethnicity, and so forth.[1] The emerging picture is that, for liberal intellectuals and political actors at least, identity politics is perceived as a threat to liberal society and a constitutive feature of the various forms of populism that seek to destroy it. On the other hand, as with the label “populist,” few of those who endorse a conception of politics that liberals would call “identity politics” embrace the label themselves.

I want to argue here that the debate around identity politics has taken a caricatured shape that doesn’t serve the liberal side. Liberals are making the same mistake they made in the 19th century and again today with national identity. The contrast between cosmopolitan and identity politics doesn’t properly reflect what is at stake in the contestation of the liberal nature of democracy nor what is constitutive of the liberal conception of politics. That people’s political claims, and a society’s political morality, depend on a broad range of social identities is not inherently problematic from a liberal perspective. Quite the contrary: (self-)categorization and a sense of belonging in groups and communities are required to avoid the complete disengagement from public life that, as Alexis de Tocqueville presciently noted, contains the germ of democratic despotism.[2] Rather, the danger lies in the increasingly narrow self-categorization that underlies individuals’ social identities. In other words, identity politics becomes a genuine problem when it is coupled with narrow identities.

There is a large body of social-psychology literature that shows how intergroup behavior can be determined by identity-based affiliations, even when those affiliations are essentially arbitrary. So-called “minimal group paradigm” studies pioneered by social psychologist Henry Tajfel have experimentally established minimal conditions under which intergroup bias is likely to appear.[3] The original experiment famously pitted two groups of schoolboys against each other by randomly telling each of them that they were either “overestimators” or “underestimators” of the number of dots in a display (in another experiment, they were randomly told that they were part either of the “Klee” team or “Kandinsky” team). While such arbitrary assignment is, at first sight, unlikely to create any genuine sense of belonging, the schoolboys subsequently displayed in-group favoritism in the distribution of “rewards for participation.” Remarkably, this happened even though there were no indirect interactions between the teenagers and self-interest considerations were not at play.

These findings have been replicated dozens of times under different conditions. Minimal group paradigm experiments establish the robustness of in-group favoritism under many forms of largely arbitrary categorizations for multiple outcome measures (e.g., allocation tasks, evaluation ratings, measures of attitude and identification). They serve as the backbone for the development of “Social Identity Theory,” a research program in social psychology that aims to uncover the mechanisms relating group identification and group behavior. The research program has established that in-group favoritism and group biases can emerge even in the absence of conflicting material interests as a result of cognitive processes of categorization and social comparison.[4] Economists have built on these theoretical and experimental results to study further the relationship between social identity and various economic behaviors and outcomes in what is nowadays known as “identity economics.”[5]

“Castle and Sun,” Paul Klee (1928)

Now, this doesn’t lead to any particular normative conclusion. Insofar as in-group favoritism is likely to be related to values such as solidarity or reciprocity, it is a vector of social cooperation and a remedy to collective action problems. Of course, in-group biases also increase the salience and intensity of intergroup conflicts. If the stakes of the conflict are high, we may imagine that group identification may trigger intergroup violence. It could be argued, however, that this risk should not be exaggerated. First, in-group favoritism weakens, and sometimes almost completely disappears in experiments when considerations of material self-interest are introduced.[6] That suggests that pure (self-)categorization alone is not a strong driver of identity-based group behavior. Second, in light of the fact that categorization and group identification seem to be relatively easy to induce through artificial means, we could expect that each individual has multiple and simultaneous group identifications that, in the aggregate, largely “neutralize” each other.

Regarding the first point, evidence is complex. Theoretical models and statistical studies indicate, for instance, that in real-life settings, identity considerations can prevail over material self-interest. A notable contribution in this spirit is by political economist Moses Shayo, who shows that preferences for redistribution depend on whether individuals predominantly identify with their social class or their nation.[7] In Shayo’s model, poor households who identify with their social class unsurprisingly favor relatively high levels of taxation and redistribution. But those for whom national identification is predominant prefer lower levels, as they give more weight to the status of their reference group (the nation). The model helps account for significant empirical patterns in democracies, such as national identification is more likely among poorer than richer populations and for a strong negative statistical relationship between the level of redistribution and the prevalence of national identification. The point is that self-interest, narrowly understood, doesn’t necessarily trump identity considerations, far from the contrary.

As for the second remark, this depends on the mechanisms that determine social identification. While in experiments, social identities can relatively easily be “primed” in individuals, outside the lab which identities are effective depends on a complex set of factors. While in principle we all have multiple identities (I’m French, European, middle-aged white man, husband, university professor, member of different overlapping academic communities, Substack writer, cat owner, and so on), that doesn’t entail that they all take the same importance for us or that their importance doesn’t change through time. Schematically, we may imagine two general classes of social identity equilibria. In the first class of equilibria, all individuals maintain multiple social identities rather than belonging to a single group. In these “multiple identities equilibria,” identity considerations are statistically likely to be balanced and, even if in-group biases exist, they are mostly neutralized at the aggregate level. However, there is a second class of equilibria where each individual adopts a single social identity and takes the other ones as irrelevant. In some of these “narrow identities” equilibria, everyone identifies with the same group, while in others, the population is divided between different groups with which different individuals strongly identify.

The economists Partha Dasgupta and Sanjeev Goyal offer an interesting theoretical model that explores the conditions that make the emergence of this last type of narrow identities equilibria more likely.[8] In their two-group model, payoffs from group memberships are an increasing function of one’s absolute and relative group size. In situations where one’s payoffs decrease as one’s relative group size decreases (e.g., in conflicts over land), individuals who opt for multiple identities (i.e., they identify with groups 1 and 2) generate negative externalities for the other members of their own group. If membership costs are small, a socially inefficient multiple identities equilibrium can prevail; if membership costs are high, we obtain a socially efficient narrow identities equilibrium.

A distinctive feature of Dasgupta and Goyal’s model is therefore that identifying exclusively is, in some situations, not only rational but also collectively optimal. Now, this result entirely depends on the existence of external costs related to groups’ relative size. There clearly are situations where multiple group memberships are unimaginable because of those negative externalities, e.g., in the case of a military conflict between two countries. In a scenario like this, we easily understand why people may be asked to choose a side by other group members. It’s either/or. But, if we put aside war-like situations, the collective rationale for narrow identities is less obvious. Of course, at the personal level, each of us is likely to experience one day or another “conflicting identities” that demand we make difficult choices between incompatible prescriptions and allegiances. In such situations, a rational move may indeed be to make a strong and irreversible (behavioral, if not cognitive) commitment to some of them. However, the negative externalities we are talking about are mostly non-material. They are not about the fact that one group will be military overpowered by the other. They result from preferences (values) and normative beliefs that have no intrinsic justification.

One may not like that someone from their religious group marries someone else from a different religious group. One may sincerely think that this is “wrong” according to their religious principles. One may think, and eventually be right about this, that tolerating interreligious marriages will weaken one’s cult and, over the long run, contribute to its disappearance. Still, does it make it “socially optimal” that individuals are barred from marrying someone from a different religious group, or more generally from changing religious affiliations? The problem lies in the fact that you cannot do a normative analysis without committing with respect to the kind of considerations that should count. In (normative) economics, social optimality is traditionally evaluated against individuals’ preferences. In the same way, what is called “externality” is always identified and measured in terms of people’s preferences. Narrow identities are socially optimal when people’s preferences are such that they would prefer that their social group be bigger than other competing social groups, or more generally, when they would prefer that others do not belong to several groups. The question is: should we grant any normative significance to this kind of preferences?

In any case, this makes it clear why even if social identities are not a problem for liberal politics, narrow identities potentially are. “Being a member of a democratic society” is itself a form of social identification. In some cases, this social identity can conflict with prescriptions arising from other group memberships. As a French citizen, you may understand and eventually accept that you’re not supposed to ostensibly display religious symbols in some public settings and cannot legitimately demand the public celebration of your religious beliefs. As the member of a religious group, you may eventually regret this and even militate to change the law. That’s probably okay as long as you still abide by it. Conflicts between multiple identities must be resolved and, from a liberal perspective, the default assumption is that they must be resolved in accordance with the general rules of liberal politics and the idea of public reason. Still, as I said above, we don’t want people to narrowly identify with their identity as a citizen of a liberal state. Other social identities are not only legitimate from a liberal perspective, they are also valuable and a relevant source of normative demands. Without them, there is no public life. At best, it leads to democratic despotism. At worst, the most cohesive and politically aggressive groups will sooner or later take over political power.

The bottom line is that, from a liberal perspective, there is no acceptable narrow identities equilibrium, of any kind. Narrow identities are incompatible with the principles and institutions of a diverse, pluralistic, and open society. In such a society, individuals should learn to cultivate multiple identities, navigate the conflicts between them, and acknowledge that, in the last resort, the umpire must be the liberal state and the underlying processes of public reasoning.

---

[1] Francis Fukuyama, Identity: The Demand for Dignity and the Politics of Resentment (Farrar, Straus and Giroux, 2018). Francis Fukuyama, Liberalism and Its Discontents, International Edition (Profile Books, 2022).

[2] Alexis de Tocqueville, Democracy in America and Two Essays on America, ed. Isaac Kramnick, trans. Gerald Bevan (Penguin Classics, 2003).

[3] Henri Tajfel et al., “Social Categorization and Intergroup Behaviour,” European Journal of Social Psychology 1, no. 2 (1971): 149–78.

[4] Though I should note that Social Identity Theory is not without critics and that experiments relaxing some “minimal group” criteria led to less clear conclusions. In particular, when considerations of material self-interest are introduced, in-group favoritism is almost completely eliminated. See Gary Charness and Yan Chen, “Social Identity, Group Behavior, and Teams,” Annual Review of Economics, no. 12 (2020): 691–713.

[5] The foundational article is George A. Akerlof and Rachel E. Kranton, “Economics and Identity,” The Quarterly Journal of Economics 115, no. 3 (2000): 715–53. For an excellent and recent survey of this literature, see Moses Shayo, “Social Identity and Economic Policy,” Annual Review of Economics, no. 12 (2020): 355–89.

[6] See what I say in footnote 4 above.

[7] Moses Shayo, “A Model of Social Identity with an Application to Political Economy: Nation, Class, and Redistribution,” American Political Science Review 103, no. 2 (2009): 147–74.

[8] Partha Dasgupta and Sanjeev Goyal, “Narrow Identities,” Journal of Institutional and Theoretical Economics (JITE) / Zeitschrift Für Die Gesamte Staatswissenschaft 175, no. 3 (2019): 395–419.