“Identity Economics” by George Akerlof and Rachel Kranton: A Rambling Review


I am neither a sociologist nor an economist , so instead of qualifying statements throughout  – how boring would that be? – I am making one big boring disclaimer here: claims made here about the faults of various social sciences should be read with my amateur status in mind, and I’d appreciate comments from any passing professionals if they see big mistakes in what I say.

This is a long essay (4,000 words) and even those few who do want to read it won’t find it easy in blog form, so here is a PDF: Download Identity Economics.

Two Solitudes

For the fifty years since Gary Becker first applied economic techniques to social issues such as dysfunctional families and crime, the social sciences have been inhabited by two solitudes, seemingly incapable of communication. Sociologists and cultural theorists talk of ideology, identity, hegemony and discourse; economists deal in rational choice, individual tastes, incentives and the mathematics of game theory. Sociologists suggest that society shapes the individual; economists that individual traits shape society. Many economists come from a right-wing and market-friendly outlook; mainstream sociology has a more left-wing perspective.

Many people have talked about this gulf; for example Jon Elster has argued that sociologists’ methods of explanation (functionalism in particular) are invalid, that methodological individualism can be a viable way of thinking even for Marxists, and that it should not carry with it the taint of right-wing political individualism that it often does. But the exciting thing about work of Kranton and Akerlof is that they don’t just talk about the gulf, they do something about it, while respecting both traditions of thought. This is surely an important achievement.

Identity Economics

Identity Economics explains many persistent social patterns like gender-based segregation in the workforce or the high rates of dropout from school among African Americans. It does so using the tools of rational choice – game theory, based on utility maximizing individuals and their interactions — but it looks to the social phenomenon of identity as an important source of individual utility. Society is shaped by individuals who are shaped in turn by society. In a series of three long papers Kranton and Akerlof described an economics of identity and applied it to some important problems; they have now pulled together material from these papers and present the core ideas in book form.

Kranton and Akerlof investigate two environments where identity plays an important role: work and school. They also investigate two of the most obvious and important sources of identity in modern North American or European life: gender and race. In each of these cases, they find that adding identity to standard economic models provides convincing explanations for many different patterns of behavior and diagnose some serious social problems.

Let’s take an example: why do “many African Americans seem to choose courses of action that middle-class white and black Americans consider disastrous” such as adopting a life of crime even though “crime does not pay” economically, and or dropping out of school even though “the rate of return to skill acquisition for blacks is, if anything, higher than for whites” [p99]? Traditional economic theories, they argue, cannot help us understand these patterns.

A&K start with a “boilerplate economic model”: there are black and white workers and individual workers must decide whether or not to work at a given wage. But in addition to the economic outcomes, this choice carries with it a cost or benefit based on identity. They identify a set of pre-existing social categories that individuals may choose to identify with, and each social category carries with it a set of prescriptions, which may be behavioral norms or categorical ideals.

In this case the categories are insider and outsider.

White workers are all, by definition, insiders. Black workers can choose whether to integrate and join the dominant majority as insiders. Alternatively, they can be outsiders, who remain apart and adopt an identity in opposition to the insiders. [p. 103]

The norm for an insider is to work for firms in this economy. Outsiders feel they should not be so submissive.

An individual African-American in this model faces a choice of social category (“I’ll always be an outsider…”) and a choice whether to conform to or to deviate from the prescriptions for that category (“…but I’ll do what it takes to get a job”). Costs and benefits accompany these choices because people internalize norms and ideals: “people follow norms much of the time because they want to do so. This conception does not correspond to economists’ usual view. To date, economists have mostly seen norms as sustained by external forces: people follow norms because if they do not, they could be punished in some way.” [p. 33] Violating the internalized norms of one’s category produces anxiety and stress.

A&K quote journalist Jill Nelson after a job interview at the Washington Post to illustrate how identity manifests itself:

I’ve also been doing the standard Negro balancing act when it comes to dealing with white folks, which involves sufficiently blurring the edges of my being so that they don’t feel intimidated, while simultaneously holding on to my integrity. There is a thin line between Uncle-Tomming and Mau-Mauing. To fall off that line can mean disaster: On one side lies employment and self-hatred; on the other, the equally dubious  honor of unemployment with integrity.

As they walk through their approach, identity reveals itself as a convincing and self-consistent approach to complex problems. It explains why the best choice available to some African Americans is to adopt an oppositional identity and to opt out of the mainly white job market — even though they may be economically better off to adopt an insider identity and pursue a job. Adopting the identity of an insider route may gain them more money, but they will always fail to conform to what an “ideal insider” should be, and this failure will exert a cost (“blurring the edges of my being”). Adopting the identity of an outsider but pursuing a career in the white world will also gain them money, but they will suffer anxiety and as a result of “Uncle-Tomming” – what Glen Loury calls a “dissonance between my self-concept and the socially imputed definition of who I am supposed to be… balancing my desire not to disappoint the expectations of others with a conviction that one must strive to live authentically.” [link]. Adopting an outsider identity and staying outside the working world will lose money, but will allow them to sustain a sense of living authentically. In this way, identity economics suggests that self-destructive behavior in marginalized communities is not the result of individual irrationality, or even of peer pressure to conform with existing destructive behavior, but can be seen as the best individual response to a world of low economic endowments and, among other things, to a high degree of social exclusion by the “insider” group. The refusal of whites to permit black workers
to become true insiders is a key factor in the utilitarian calculus.

This is quite a different explanation for high incarceration rates, high out-of-wedlock birth rates, low employment rates, and a divergence in outcomes for African Americans [p. 99] to Becker’s model or economics models based on statistical discrimination or on peer-driven destructive behaviour.

The argument here has been descriptive, but in the original papers A&K set up a more formal structure and create a game for which the oppositional identity, with its associated prescriptions, can be an equilibrium. In the same way, gender prescriptions in the workplace, oppositional identities in the classroom, and employee decisions whether or not to identify with their employer all lead to consequences that, while familiar, have been difficult to explain from an economic point of view. Kranton and Akerlof are very respectful about the arguments made by Becker in his very different analyses of discrimination, but come to conclusions that are very different: while Becker has long argued that competitive markets will erode discrimination, A&K say that it is no surprise it took social movements to make real changes to gender or race-based discrimination.

The response to these papers has been, from what I can tell, lukewarm. Akerlof’s standing has given the arguments a profile that they otherwise would not have achieved, and there has been follow-up work by others, but ten years after the first paper was published, identity is a long way from being a part of mainstream economic thought and game theory is a long way from the mainstream of sociology. Why has the response been so muted, and what does each discipline have to gain from this venture?

What economics has to gain from identity

In the original papers A&K pursue a full, albeit qualitative, game theoretical treatment and show that for some magnitudes of costs there are mixed solutions: in particular, if the cost of “working in the white world but not ‘making it’” is large then it is rational for some black people to choose the insider identity, and it is rational for some to adopt an oppositional outsider identity, and to drop out of school at an early age.

At a lecture at the LSE in 2007 [TranscriptSlides], Akerlof said “I know that all of the economists in the audience have three thoughts. The first is that all of this gives up the parsimony of standard economics. The second question is whether identity is, or is not, observable. The third question concerns the endogeneity of identity.”

I think there are two separate concerns behind the “parsimony” complaint. One is that some economists, having trained themselves to see the world through the lens of earlier models, have a blind spot for other factors and therefore think they are unimportant. There is a story of a group of impatient economists who offered $5 to the person ahead of them in the check-out line if they could switch places, and were surprised that the person was offended. These people have trained themselves to live in a world different from the one the rest of us live in. Steven Levitt of Freakonomics fame fell into this category, but recognized the role of identity when he was caught in a traffic jam:  

I had read their [A&K’s] papers, but in general have such a weak sense of identity that I never really understood what they were talking about. The first time I really got what they meant was when I realized that a key part of my identity was that I was not the kind of person who would cut in line to shorten my commute, even though it would be easy to do so, and seemed crazy to wait for 15 minutes in this long line. But, if I were to cut in line, I would have to fundamentally rethink the kind of person I was. [Link]

The phrase “I am not the kind of person who…” is helpful when thinking about where identity intrudes into your own decision making. If you have sat in a car at a red light at an obviously-empty intersection, waiting for it to change rather than go through against the light, well that’s an identity-based decision, following the norms of the community you live in. I could easily take paper and pens from work, but I am not the kind of person who does that.

Beyond such decisions, identity affects us whether we like it or not. We have internalized what it means to be the gender we are, what it means to be the race we are, so deeply that we usually do not notice it.

Parsimony is a technical issue too. One of the disappointments of game theory in the social sciences is that researchers found that anything beyond the most simple game delivered too many answers: an answer for every taste perhaps. The subtleties and grey areas around issues of identity seem like natural candidates for generating such multiple equilibria. Nevertheless, and I can’t convince you of this here, it seems to me that by staying with a qualitative approach, by phrasing their problems with care, and by introducing just enough identity into their models, Kranton and Akerlof have managed to find a way through the maze, to construct consistent, logical
explanations for social patterns of real importance. This is a delicate and impressive balancing act; it remains to be seen if their success can be reproduced in other problems, or if the approach becomes limited by society’s complexity.

The second question is to ask where it comes from, this “identity”? Is it measurable? Here again, the problem is as much about the asker as the answer. Too many economists are hardcore reductionists who would rather look inside the brain for sources of utility (see the sillinesses of neuro-economics) than open their eyes to the world around them, and who see the work of other social sciences as fluffy, ill-defined, and inconsequential. But any evolutionary system resists purely reductive explanations. To explain dam-building behaviour in beavers we need to look not just at the genes they inherited but at the landscape they live in: the large and the small are tightly coupled and neither is “more fundamental” than the other. And if it is true that identity is difficult to quantify, and that the internal costs of breaking the norms of our identities are difficult to place on the same scale as dollars and cents, well that’s why all those other social scientists spend entire careers on the problems. The question reveals a deeply ingrained lack of respect for non-economic social sciences. Too many physics-envying economists seem to want to solve massively complex problems from first principles, but this agenda (doomed to fail) forces them to ignore massive amounts of useful observation. The semi-empirical approach of much theory of chemistry, in which the role of theory is to relate seemingly disparate observations and to see patterns where we did not see them before, is a better model for the social sciences.

The third question is to ask why identity must be exogenous (imported into the model from outside as a pre-defined force that constrain individual actions), rather than endogenous (emerging from the choices of the individuals). The identities and prescriptions associated with gender and race have developed over centuries and over continents. They did not evolve within the short timescale of a particular scenario under the microscope (employment choices of women in the first decade of the 21st century North America). And the prescriptions associated with white/black or male/female are not, of course, symmetrical: these are not arbitrary labels, but are the product of long and specific histories, which lie outside the scope of the situation being analyzed.

Most of Kranton and Akerlof’s efforts in promoting Identity Economics have been aimed at fellow economists, trying to convince them of the validity of the approach. But there is another group who could benefit from their work, and that’s sociologists.

What sociology has to gain from identity economics

The response from some members of the other social sciences has been the opposite to that of skeptical economists. Instead of dismissing identity as fluffy and unobservable, they have decried the approach as obvious. Oh look, economists have finally noticed that identity matters! And they think it’s a big deal! Well lah-di-dah, tell us something we don’t know. [See comments at the Chronicle of Higher Education review for examples]

Such a response misses the point. First, Akerlof and Kranton make no claims to have discovered identity: they explicitly and repeatedly credit other social scientists with identifying it, exploring it, and making it a rich concept that they can use. The accusations of appropriation are empty. Second, they are doing something with identity that other social scientists have not done, and to ignore this is cutting off their nose to spite their face.

So what do sociologists, who recognize and understand identity, have to gain from identity economics? One answer is a single word: equilibrium, and my main disappointment with A&K’s book is that in simplifying their research they have avoided, even once, mentioning this crucial concept.

Back in 1965 Mancur Olson’s The Logic of Collective Action showed that there are good reasons why “rational, self-interested individuals will not act to achieve their common or group interests” [p2]. If the benefits of collective action require all members of the group to share the load, then there is a commons problem where the best choice for individuals may be to free ride. It can make absolute sense for people to act as if they believe on thing, while believing another. It may, to take an example at random, make sense for individuals to shop at Wal-Mart
even while they wish the store had never been built. Why? Because one decision (to allow Wal-Mart to be built or not) is a collective one, and the other decision (to shop at Wal-Mart or not) is an individual one.

Sometimes it has seemed to me as if all the avant-garde contortions of post-Marxist, post-modern political thought could have been avoided if sociologists had learned a little game theory. Instead, believing that it is in the interests of subjugated classes to throw off their shackles and yet seeing no shackles being thrown, they have resorted to more and more baroque explanations for this seeming contradiction. Terry Eagleton’s Ideology: An Introduction describes
several of these explanations for why “people invest in their own unhappiness”.[xiii] The only common thread is that the populace fails to see the state of the world as it really is (otherwise they would act to change it): false consciousness is to blame; Habermas claims that consumerism bypasses meaning to engage the passive subject subliminally [37]; Althusser says that thought is conducted within the terms of an unconscious ‘problematic’ which silently underpins it and which ‘constitutes the limits of what we are able to utter and conceive’ [137]; Lacan argues that we live in a state of self-deception in which we have suppressed our true desires [53]; hegemonic discourses make subversive thoughts impossible to express, and so on. Yet simple game theory would have cut this Gordian knot. Eagleton writes that “If it is rational to settle for an ambiguous mixture of misery and marginal pleasure when the political alternatives appear perilous and obscure, it is equally rational to rebel when the miseries clearly outweigh the gratifications.”[xiv] But this ignores the collective action problem: just because it would be good for a group to rebel does not make it sensible for an individual to rebel. The contradiction is not a contradiction at all: was all this baroque theorizing really necessary?

That is overstating the case, of course. Olson may have explained why we can believe one thing and act in a seemingly contradictory way, but not all social patterns are reducible to tragedies of the commons. Identity-driven patterns are some of those that could not be explained by collective action logic alone.

I think I can see why A&K chose to avoid spelling out the game theoretic portions of their investigations. To do so would have made a much longer and more difficult book, and so would have limited the audience. Yet as it stands the book presents what is really a theoretically sound, logically consistent methodology as if it were an identification of important factors in a situation coupled with rough and ready inferences. It underplays the work they have done. That work is still available in the papers, of course, but if the book is the way the ideas are transmitted, readers are missing an important piece of the puzzle.

What game theory brings is an explanation of why certain patterns persist, and the introduction of identity extends these explanations to many important patterns. It is all very well to say that there are norms associated with the roles of women and men in the workforce, but such a statement is descriptive only. It does not explain how such norms sustain themselves, or why norms sometimes collapse (gays cannot marry, animals have rights). Identity economics spells out, qualitatively at least, how norms can be sustained in some cases while not in others. If the equilibrium of a model is one in which nobody chooses to be an outsider, the social category of “outsider” will vanish: left-handed people will cease to be sinister, for example.

Sociologists could benefit from Identity Economics because it clarifies the question of how and why identities and their prescriptions are sustained. Without the notion of equilibrium sociological explanations of how systems of oppression sustain themselves without explicit violence are vulnerable to the skeptical “why do people not simply choose otherwise?” that has sent them to psycho-analysis, structuralism and other sources of unconvincing obfuscation. But this question is not as damaging as it seems, once we realize that “because it is in their self-interest not to” is a perfectly reasonable answer.

A second reason for sociologists to pay attention to Identity Economics is that game theory forces the researcher to be explicit and precise about what they mean by certain terms. So identity is a set of social categories, which carry prescriptions, which have costs and benefits. This is a simpler description of identity than I have seen anywhere else, and Kranton and Akerlof achieve this clarity because they have to make the concept fit into a game theory framework. Perhaps this misses some of the richness of sociological theory, but what it misses in richness it gains in clarity. And some of the “richness”, I suspect, is really just fuzziness anyway.

It is natural to ask, if identity can be drawn into an economic framework, whether other sociological concepts may be too. Despite its clarity, the pictures that Kranton and Akerlof draw should be seen as a beginning, not a closed and polished object.

Identity is a rich concept, and there are several associated concepts that K&A have not explicitly drawn on but which clearly impinge on their theory. Here are a few random thoughts to finish off. Ideologies are, Eagleton says, action-oriented sets of beliefs. They are beliefs that must be “capable of furnishing their adherents with goals, motivations, prescriptions, imperatives, and so on.“ [47] That is, ideology can be seen as the concept that connects social categories to their prescriptions. By studying identity, Kranton and Akerlof are studying ideology. A sentence from Eagleton reads simultaneously as if it could belong in an K&A paper, and also as if it comes from another planet: “A mode of domination is generally legitimated when those subjected to it come to judge their own behaviour by the criteria of their rulers. Someone with a Liverpool accent who believes he speaks incorrectly has legitimated an established cultural power.” [55] Alternatively, we can say they have adopted an insider identity, but fail to meet the ideal set of prescriptions associated with that identity. Kranton and Akerlof do not investigate class-based identity, but it exists in many cultures and has its own consequences, as anyone who was told not to “act above their station” or that some goals were “not for the likes of you” will recognize.

While ideological beliefs may be the glue that connects social categories to prescriptions, Eagleton asks doubtfully if it makes sense that people would hold beliefs that they know to be untrue? This misrepresents the problem of knowledge in a social context. I may not hold beliefs that I know to be untrue, but I and many others do publicly subscribe to beliefs about which I have real doubts. Why would I do such a thing? Because I’m ignorant, and so I adopt the belief of people who have similar outlooks to myself: people who help to define the prescriptions for people in the social category to which I have elected to belong. I read and follow people whose arguments echo my own prejudices, in the absence of conclusive evidence I will continue to conform to common knowledge. To articulate beliefs carries with it a cost.

The distinction between identities and the actions associated with them is not always clear. Judith Butler’s argument that gender is a “performance” suggests that even the most seemingly hardwired identities are ones that we can choose. From the tiny bit I know, Butler goes all psycho-analytical as to why people would adopt various identities, but I wonder if a little game theory might help to clarify why, in some cultures, gender is strictly defined and binary while in others it is more fluid and multifaceted.

So you can see that I found much to admire here. I do wish the authors had brought more of the game theory background into the book; the notion of equilibrium would have made some passages more convincing to those who have not read the papers, but to this outsider it seems that both economists and sociologists have a lot to gain from studying this work: let’s just hope it doesn’t go too much against the prescriptions associated with their social categories as economists or sociologists to do so.

Bookmark the permalink.


  1. Great review, Tom. Thanks: I am going to get the papers. As far as your disclaimer, l think you forgot more economics than most of my co-economists ever knew, as the phrase goes. Can we look forward to a 2nd edition of NMSW with a new chapter: “Rachel and George visit Whimsley?

  2. Rather to my surprise, I find myself more sympathetic to the “question [of] whether identity is, or is not, observable” than you seem to be. I take it to be about whether or not there are ways of determining what the available identities are, and who holds which, that are logically independent of the behaviors that they are supposed to explain. If not, it would seem to be only too easy to make identity (or norms or culture…) into a sort of dormitive property. One might well feel that people who make this critique yet believe in explaining all behavior by maximizing “utility” are not so much living in a glass house and throwing stones as breaking one of their own walls so as to throw the shards at their neighbors, but still it’s not without force. This doesn’t really apply to the examples you cite from Akerlof and Kranton, of course. (I’ve read neither the book nor the papers, just lived through many iterations of this argument when it’s been about norms and culture.)

  3. Thanks Kevin. The final chapter of NMSW, the 1st and only edition, does address the Identity Economics papers. But “Rachel and George visit Whimsley” would have been a better title.

  4. I looked up “dormitive property” and found it linked to arguments like “people tip because of the utility they derive from tipping”. So yes, I can see that identity could be used to justify arguments like “people tip at restaurants because they have adopted the identity of a class of person for whom tipping is the right thing to do.” It’s certainly not without its dangers.
    They key, then, is whether identities and their prescriptions can be observed outside the context of the game. I think this is one reason why making it an exogenous property in the models is a good move.
    My brother pointed out that I under-reviewed the book, and he’s right: I didn’t actually go through a lot of what the book says. But a big part of the work that they did was to scour sociological, literary, psychology and other sources to define the categories they identified and the prescriptions associated with each (not so difficult in the case of gender and race; more difficult in the cases of workplace and schools). It’s that secondary research that made the papers and, to a lesser extent, the book impressive to me. Kranton and Akerlof accepted that, properly done, identity is not a concept to be used as seems convenient, but one to which many sociologists have contributed by endless observing and interviewing. I would agree that there is an obligation to read that observational material – to establish the fact of identities and their prescriptions independently of the economic argument – before invoking identity in an economics kind of way.

  5. “If you have sat in a car at a red light at an obviously-empty intersection, waiting for it to change rather than go through against the light, well that’s an identity-based decision, following the norms of the community you live in.”
    I think norms = identity is an overstatement. Imagine that a norm is so ubiquitous and entrenched that it doesn’t even occur to anyone that there might be another “kind of person”. Do we still claim decisions according to that norm are based on “identity”?

  6. Not entirely sure what you mean. An example might help. Norms, in this view, are the prescriptions associated with an identity. One could feel constrained to follow norms even if one does not have ideas about alternative identities.

  7. This was extremely good. I don’t know if I can wait for the paperback.

  8. Very nice, thoughtful review. Thanks.

  9. just out of idle interest, do they mention Roland Fryer whose written lots on “acting white”? if not, and you don’t already know him, you might be interested in his stuff.

  10. I should also add – great review, much appreciated. Like Cosma, I think concerns about observability are more about wondering how the theory can be tested, and less about disrespecting soft science.

  11. Being an economist who studies identity and who read a lot of sociology in grad school, I am very interested in Akerlof and Kranton’s work and your response.
    However, there is a flaw in both your response to economists and sociologists which highlights why many economists have found their work to be very interesting but somewhat unsatisfactory, more an excellent first step rather than a complete theory.
    (Though I must admit that while I have read most of their academic articles on the subject, I have not read this book yet.)
    As AK acknowledge, economsts are skeptical about the exogeneity of identity. While their mathematical models tend to assume it is exogenous for tractability sake we know that not only are identities changeable (a rural/black/southern/woman is sometimes black, sometimes a woman, sometimes rural, sometimes southern, sometimes a combination of those 4). Thus their analytical model does not give us any guidance to perhaps the most important feature of identity (though I do acknowledge they do talk about these issues more informally).
    Similarly, when you say the main contribution for sociologists is to aprpeciate equilibrium and how identities and norms form. We agree. (My co-authors and I have for years been working for years on such a theory.) Akerlof and Kranton’s highlight the importance of these questions but have not pushed as far analytically on the important questions of identity and norm formation.

  12. Time will tell about the observability concern, I guess.
    They mention Fryer briefly (in a footnote to Chapter 7), and he references them in An Empirical Analysis of ‘Acting White’ (PDF). I didn’t of him, so thanks.

  13. If you’ve read the articles I don’t think there is a lot to be gained from the book. By marking identity as exogenous A&K definitely leave a lot of questions open but this seems like a good thing to me. The big identities (race, gender, class) are formed over centuries and over continents, and to ask that they be explained in a theory of current society sounds very ambitious, to say the least. And it’s not clear to me that economics is the tool to use to explain it. Every field of study has to take some results from other fields for granted, and maybe this is something economists just have to take from the other social sciences.
    I have more sympathy when it comes to the smaller identities – their studies of workplace and school, for example, where identities are formed and transmute quickly and on a local scale. Like a lot of big words “identity” covers many distinct phenomena – maybe too many.
    Thanks for the comment, and I’ll definitely look with interest at your papers.

Comments are closed