NOTE-1311: Economics and Identity

Contribution

It brings identity into economic analysis as an argument of the utility function. Identity is tied to social categories, each with prescriptions for behaviour and an ideal of physical and other attributes, and with higher or lower status. The paper builds a small game around it and applies it to three settings: sex segregation in jobs, poverty and social exclusion, and the division of housework. In each, it says, identity “substantively changes conclusions of previous economic analysis” (p. 715).

Key insight

Behaviour that looks self-destructive from outside can be utility maximizing once self-image is counted. People “behave in ways that would be considered maladaptive or even self-destructive by those with other identities”, to “bolster a sense of self or to salve a diminished self-image” (p. 717). And because one’s identity depends on what others in one’s category do, there is a new kind of externality: a man in a dress threatens other men’s identity, and they may respond (p. 717).

Assumptions

• The utility function (p. 719). U_j = U_j(a_j, a_−j, I_j), with I_j = I_j(a_j, a_−j; c_j, ε_j, P). c_j is j’s assignment of everyone, herself included, to categories; ε_j her given characteristics; P the prescriptions. “Identity” names both the self-image and the assigned categories (p. 718).
• Prescriptions, not norms (n. 2, p. 716). Norms in economics are obeyed for fear of punishment. Prescriptions are followed “for the most part, to maintain their self-concepts”.
• As-if maximization (p. 719 and n. 6). People need not be aware of their motives; motives may be unconscious. Friedman’s methodology is invoked.
• Psychodynamic psychology (p. 728). Breaking internalized rules causes anxiety (the superego); identity must be “defended against anxiety”. Identification is internalizing values “such that her actions should conform with the behavior of some people and contrast with that of others”. Cognitive dissonance is offered as an alternative basis (n. 28). Self-categorization theory is not used; Tajfel and Turner (1979) are cited only for a review.
• Identity is partly chosen (pp. 719–720, 725–726). Categories are more or less ascriptive; choice “may be proscribed for others”.

Key results

• The prototype game (pp. 727–730, Figure I). All see themselves as Green; Greens should do Activity One. Each person has a taste for one activity, worth V. Person Two, with a taste for Two, loses Is by doing it. Person One then loses Io and may respond, at cost c, inflicting L. Payoffs (One, Two): Activity One (V, 0); Two, no response (V − Io, V − Is); Two, response (V − c, V − Is − L).
  • Four subgame-perfect outcomes (p. 730): deterrence when c < Io and Is < V < Is + L; response without deterrence when c < Io and Is + L < V; no response when c > Io and Is < V; no Activity Two at all when Is > V.
  • Comparative statics (pp. 730–731). A tax T > Io − c on responding ends deterrence; total utility moves from V to 2V − Is − Io. A rhetorical campaign can raise Is and Io. The conflict is Sen’s Paretian liberal: one cannot protect both the deviant and those disturbed by the deviance.
• Evidence that identity enters payoffs (pp. 720–727).
  • Minimal-group and Robbers Cave findings; the authors grant these “do not show the existence of a ‘self’ or this identification per se” (p. 720).
  • A telephone vignette survey, 70 households from the Fremont, CA phone book (Table I, p. 724). A carpenter teases the first female carpenter. 84% rated “he feels less masculine” at least somewhat likely; 84% “he and his friends will not be able to joke around”; 36% fear of a lower wage; 13% unfairness of her lower wage.
  • Southern men insulted in Nisbett and Cohen’s experiments feared the experimenter thought less of their masculinity, though they would never meet again (p. 725).
  • Alumni giving to one’s own college, against giving where the marginal return is highest (p. 722).
• Gender and work (pp. 732–737). A woman in a “man’s job” loses Is; male coworkers lose Io and may act against her. Firms may create a parallel “woman’s job”; competitive firms underinvest in new job categories when associations are economy-wide.
  • Two-thirds of the labour force would have had to switch jobs for parity in 1970, unchanged since 1900; 53% by 1990 (p. 732).
  • 1970 shares female: secretaries 97.8%, nurses 97.3%, elementary teachers 83.9% against secondary 49.6% (p. 734).
  • After the women’s movement: the median job-tenure gap fell from 3.3 years (1968) to 0.4 (1998); of 45 occupations 0.0% female in 1970, one was under 1% female in 1990; women rose from 24.6% to 52.7% of accountants and 4.5% to 24.5% of lawyers (p. 735). With no change in technology or endowments, “tastes” are “the leading suspect” (p. 736).
• Exclusion and poverty (pp. 737–745, Figure II). Members of an excluded community choose Green (dominant culture, work) or Red (oppositional). Greens lose r, the degree of exclusion.
  • An All-Green equilibrium exists if and only if r < I_oR (p. 741).
  • Exclusion alone, r > 0 with no peer externalities, makes some choose Red (p. 741). Middle-class out-migration or lost jobs raise the Red share (pp. 742–743).
  • Red “self-destructive” behaviour “is not the result of individual ‘irrationality’” (p. 742).
  • Readings of residential Job Corps against the non-residential JOBSTART, of selective schools, and of affirmative action’s rhetoric as raising or lowering r (pp. 743–744).
• The household (pp. 745–748, Figure III). Panel Study of Income Dynamics, 1983–1992, slightly over 29,000 couple-years (n. 63). A Tobit fit of husbands’ share of housework on their share of outside hours: about 10% when they do all the outside work, rising to no more than 37% as their share falls (p. 746). Equal-bargaining comparative advantage cannot fit the left side; adding the prescriptions that men not do “women’s work” and earn more than their wives can.

Claims

id	claim	strength	support
C1	Identity, a self-image tied to social categories and their prescriptions, can be represented as an argument of utility, affected by one’s own and others’ actions	stipulated: a modelling proposal, motivated by examples	§II
C2	Others’ violations of one’s category’s prescriptions are an externality that can provoke costly responses against the violator	weak to moderate: a game, a 70-household vignette survey and ethnographies	§§II–III, Table I
C3	Gender-typed identity explains occupational segregation and its decline after the women’s movement better than human-capital or labour-attachment models	weak: consistent aggregate trends, reached by eliminating technology and endowments; no test against rivals	§IV
C4	Social exclusion alone can lead some in an excluded community to choose an oppositional identity and forgo work	a theoretical result; the derivation is not in the paper (n. 57)	§V, Figure II
C5	Gender prescriptions explain why wives who earn more still do more housework, which comparative advantage cannot	moderate as a stylized fact (29,000 couple-years); the explanation is fitted, not tested	§VI, Figure III

Concepts

• identity: “a person’s sense of self” (p. 715); in the model, both a self-image and one’s assigned categories (p. 718).
• prescriptions (P): the behaviour and ideal attributes appropriate to each category in each situation; kept to maintain a self-concept, not for fear of sanction.
• Is and Io: the loss of identity from one’s own, and from another’s, violation of prescriptions.
• identity externality: a change in one person’s identity utility caused by another’s action.
• situation: who is matched with whom, and in what context; it may fix which categories are salient (n. 30).
• Red / Green: an oppositional identity that rejects the dominant culture, and one that adopts it at the cost of never being fully accepted (r).

Connections

• Sugden’s objection to the device (NOTE-1288). This paper is the standard example of what Sugden rejects: keeping individual maximization and folding the social motive into the payoffs. Sugden’s point that a preference can always be posited to fit a choice applies here; the authors grant that utility theory is agnostic about motives (p. 719).
• Self-verification (NOTE-1256). Swann et al. explain willingness to die for the group by the motive to keep one’s self-views true. This paper’s identity utility is the same kind of motive: following prescriptions to “maintain their self-concepts” (p. 716).
• Self-categorization theory (THEORY-1114). Little overlap. Its categories and prescriptions resemble social identities and their norms, and its “situation” gestures at salience (n. 30). But the mechanism is internalized rules and anxiety, not depersonalization, and the self-image is the individual’s.
• Categories, not group entities (TERM-096). Every identity here is membership in a category with a shared prescription. No group is modelled as a whole with a good of its own.
• Sen (1985), “Goals, Commitment, and Identity”, not held. The authors note that Sen discusses identity but does not put it into a utility function (n. 1). Sen’s point that commitment drives a wedge between choice and welfare is the alternative they do not take up.

Bearing on the record

• The owner’s thesis (CLAIM-491). The claim says this paper makes the thesis true of an economist’s agent by construction. That is half right. These are inferences; the paper says nothing about selflessness.
  • What it makes true by construction is that identity-driven action maximizes the agent’s utility. That is “every motive is my motive”, the trivial reading CLAIM-492 separates from the thesis.
  • The self in the utility is a self-image: how far one’s own traits and acts, and others’ acts, fit the prescriptions for one’s categories. It is not a self that includes the group’s good. No group’s welfare enters anyone’s utility. Io is a loss when a fellow member deviates, and the remedy is to harm that member.
  • So the paper’s “larger self” yields conformity and policing, not benevolence toward the group. Its one example of giving to a group, alumni giving (p. 722), is explained as loyalty and identification, not modelled.
  • It agrees with Swann et al. on what the served self is: a self-concept to be maintained. If the thesis recruits this paper, “self-interest” means keeping a self-image, not furthering a welfare.
  • On whether identity utility is welfare, the paper is two-minded. It treats utility as an as-if representation of choice (n. 6), yet sums identity gains and losses in “total utility” (p. 730), with the usual caveats (n. 29). Only on the second reading is identity-driven sacrifice self-interest in a substantive sense.
• The triviality objection (CLAIM-492). The paper offers a criterion of the self’s extent that is not caring, but no evidence that it works apart from caring. Inferences again.
  • The criterion is category assignment with shared prescriptions (c_j, P). It is fixed partly by others: categories are “more or less ascriptive”, the excluded “could never be a ‘true’ Green” (p. 748), and a person’s identity can depend on how others categorize her (n. 5). That can come apart from what she cares about.
  • But the force of identity is in Is and Io, free parameters. They are never measured; they are read off the behaviour they explain. The paper’s content comes from structure: payoffs attach to categories, are shared by their members and move together when prescriptions change. The housework asymmetry is the one place where that structure predicts what a taste for housework would have to assume.
  • The authors grant the gap: the experiments they rely on “do not show the existence of a ‘self’ or this identification per se” (p. 720).
• Batson’s objection (CLAIM-494). Not addressed. Nothing here concerns empathy or helping individuals.
• A dated note is proposed for CLAIM-491 (section 3 below), to correct its line about this paper. No THEORY filed. No ML instruction.

Limitations

• Illustrative evidence. Examples, ethnographies and one vignette survey of 70 households. No test sets the identity model against a rival on data it was not built to fit.
• Derivations withheld. The workplace model’s full specification (n. 34), the poverty model’s analysis (n. 57) and the household regression’s equations (n. 63) are “available from the authors upon request”.
• Parameters unobserved. Is, Io, r and the prescriptions are inferred from the behaviour they explain.
• Elimination as evidence. The women’s-movement argument rests on ruling out technology and endowments, not on measuring changed prescriptions.
• Typos.
  • “from from” (n. 4, p. 719).
  • “how others interpret i’s actions” for j’s (n. 5, p. 719).
  • “Nisbett and Cohn” (p. 723); the reference list has Cohen.
  • “Levine [1997]” (n. 13, p. 723); the reference list has 1995.
  • “Pierson 1986” (p. 734); the reference list has 1989.
  • Notes 48 and 49 (p. 736) give the same citation, 442 F.2d 385 (5th Cir.), cert. denied 404 U.S. 950 (1971), for Phillips v. Martin-Marietta and for Diaz v. Pan American. It is Diaz’s.
  • “Wetherell [1998]” (n. 70, p. 749); only Wetherell 1996 is listed.
  • Kandori’s Review of Economic Studies volume is printed “LXIX”; for 1992 it should be LIX (p. 751).
  • “This respect is no less diminished by” (n. 11, p. 721) means “no less” or “not diminished”, not both.
  • “Peter Howett” in the acknowledgements is probably Peter Howitt.

Open questions

• Can Is and Io be measured apart from the choices they explain, for instance by asking about anxiety at deviating? Only then does category membership become a criterion of the self that can be tested against caring.
• Does identity utility ever take in the group’s welfare, rather than members’ conformity? The later identity-economics literature (Akerlof and Kranton’s 2010 book, work on identity in organizations) may say; the record holds none of it.