Social Distance

Social Distance and Social Decisions
Author(s): George A. Akerlof

Source: Econometrica, Vol. 65, No. 5 (Sep., 1997), pp. 1005-1027
Published by: The Econometric Society
Stable URL: https://www.jstor.org/stable/2171877
Accessed: 18-04-2020 06:59 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
The Econometric Society is collaborating with JSTOR to digitize, preserve and extend access
to Econometrica
This content downloaded from 62.201.255.151 on Sat, 18 Apr 2020 06:59:54 UTC
All use subject to https://about.jstor.org/terms
Econometrica, Vol. 65, No. 5 (September, 1997), 1005-1027
SOCIAL DISTANCE AND SOCIAL DECISIONS
BY GEORGE A. AKERLOF1
A model of social distance is presented that is useful for understanding social

decisions. Status and conformity in previous models are discussed, and then a generaliza-
tion is described. In this generalization agents have inherited positions in social space and
an expected value of trade between two individuals as a function of the difference in their
initial positions. An example of this system is constructed in which there is class stability.
Agents who are initially close interact strongly while those who are socially distant have
little interaction. In this example inherited social position, which may be interpreted as
social class, plays a dominant role. The relevance of this model to social decisions such as
the choice of educational attainment and childbearing is discussed in the context of
specific ethnographic examples. Class position may play a dominant role in these deci-
sions.
KEYWORDS: Social interaction, education, fertility, urban economics, social class.
1. INTRODUCTION
TRADITIONAL ECONOMICS HAS BEEN BASED on methodological individualism.

Until quite recently, with some rare exceptions, it has not been appreciated that
this method can be, or perhaps I should say, should be, extended in describing
social decisions to include dependence of individuals' utility on the utility or the
actions of others. Except under rare circumstances, such interactions produce
externalities. These externalities typically slow down movements toward socially
beneficial equilibria but in the most extreme cases they will create long-run
low-level equilibrium traps that are far from socially optimal. Much of the
existing literature is analytically complicated and the models appear quite
special. It may therefore be of some use to strip the logic to the bone. I will do
so in the form of three rudimentary models that will demonstrate the externali-
ties involved in social interaction. One model presented is analogous to the
Newtonian theory of gravity; the special case of this model that illustrates the
low level trap is analogous to the three-body problem. A potentially productive
outcome of this discussion is a characterization of groups that are of sufficient
size to encompass the externalities so that coordination within those groups can
attain the social optimum.
The emerging theory of social'interaction, including some of the recent work
of Becker and coauthors, gives an alternative to the early work of Becker, in
l This paper was prepared for the Fisher-Schultz Lecture of the 1995 Econometric Society
meetings in Tokyo. The author would like to thank William Dickens, Rachel Kranton, Paul
Krugman, Paul Romer, and Janet Yellen and three anonymous referees for invaluable help. He
would also like to thank the Canadian Institute for Advanced Research, the Brookings Institution,
and the National Science Foundation, under Research Grant Number SBR 94-09426, for financial
support.
1005
1006 GEORGE A. AKERLOF
which social decisions were based on individual values.2 Social interaction

theory explains why social decisions-such as the demand for education, the
practice of discrimination, the decision to marry, divorce, and bear children, and
the decision whether or not to commit crimes-are not simple choices based
primarily on individual considerations. There is a significant difference between
these social decisions and the conventional economic decision-making epito-
mized in intermediate microeconomic theory as choices among alternative fruits
available at the supermarket.
This lecture will explain why rational choice analysis of social decisions must
take into account the externalities involved in social decision-making. In princi-
ple, the analysis is an extension of Becker's earlier work; but in important
special cases the incorporation of these social factors into rational choice
analysis results in behavior that more closely corresponds to the intuition of
sociologists than of economists. For example, the abstract models presented
here will explain the existence of social class and linguistic dialect.
The key difference between social decisions and conventional economic
decisions (e.g., the choice of fruits) is that the social decisions have social
consequences whereas economic decisions do not. While my network of friends
and relatives are not affected in the least by my choice between apples and
oranges, they will be affected by my educational aspirations, my attitudes and
practices toward racial discrimination, my childbearing activities, my marriage or
divorce, and my involvement in drugs. All of these activities will affect who I am
in an important way, and thus how I associate with my friends and relatives, as
well as who those friends may be. As a consequence, the impact of my choices
on my interactions with other members of my social network may be the primary
determinant of my decision, with the ordinary determinants of choice (the direct
additions and subtractions from utility due to the choice) of only secondary
2Previous highlights in linking social interactions with economic theory include Schelling (1971),
Loury (1977), Jones (1984), Frank (1985), Axelrod (1986), Kolm (1994), Benabou (1993, 1996), Crane
(1991), Durlauf (1993), Bernheim (1994), Greif (1994), Brock and Durlauf (1995), Ellison and
Fudenberg (1995), Glaeser, Sacerdote, and Scheinkman (1996). Epstein and Axtell (1996) have
developed a computer simulation program that is capable of generating models with social interac-
tion.
Becker's early work (for example, Becker (1971, 1964, 1968, 1973, 1974) Becker, Landes, and
Michael (1977), and Becker and Murphy (1988)) did not take into account social interactions; in the
intermediate stage of his career he showed special circumstances in which social interactions, if
present, were not important since the externalities could be captured within the group. More recent
work by Becker, has moved into the area where social interactions play a key role. For example,
Becker (1991) has explained the popularity for crowded restaurants through social interaction and
Becker and Murphy (1993) have examined the implications of advertising.
Of course sociology is all about the importance of social interactions and the whole extensive
literature on social networks concerns the implications of social interactions. The classic theoretical
perspective on this is given by Merton (1968). Any review of the literature should start there. Of
course the dominant theme of social psychology is also social interaction.
Probably the model closest to the work here is the tradition model in Jones (1984). I have also
written four previous papers on this topic (Akerlof (1976, 1980, 1985), Akerlof and Yellen (1994)).
SOCIAL DISTANCE 1007
importance. A proper theory of social decisions then must first spell out their
consequences for social exchange.
In social decisions externalities abound. Externalities are important either
when people try to distance themselves, in social space, from their friends and
relatives (status seeking), or alternatively, when they try to move themselves
closer (conformist behavior). I will first construct a pair of general models that
demonstrate these externalities. I will then describe a specific model that
abstractly roots people in social space, and explains the stability of class
structure. The predictions of this model accord with the ethnographic and
biographical sketches of life in the United States inner city that I will review
It turns out that the examination of ethnographic and biographical sketches is
necessary in order to discern the presence of social interaction. Empirical
estimates tend to demonstrate that neighborhood effects are statistically signifi-
cant and important3 and there is no disagreement at all about the importance of
family variables on behavior, but there is a fundamental problem of identifica-
tion in interpreting both the neighborhood and, similarly, the family background
effects, as evidence of social interaction. Borjas (1995) has found that the slow
rate of convergence for different ethnic groups can be explained mainly by
neighborhood fixed effects; Crane (1991) has found that approaching the bottom
of neighborhood quality there is a jump in the incidence of social problems for
individuals with fixed socioeconomic characteristics; and Case and Katz (1991)
have found the behavior of individuals in poor areas of Boston is correlated with
the behavior of others in the same and adjacent one or two block neighbor-
hoods. In each of these cases the evidence is consistent with social interaction,
but could also have a variety of other explanations (see Manski (1993)). The
neighborhood characteristics may be predictive of unobserved individual charac-
teristics that affect behavior but were omitted from the prediction equations;
Evans, Oates, and Schwab (1992) show that endogeneity of neighborhood choice
could account for the observed neighborhood effects without any effects from
social interaction. A further problem of identification arises if neighborhood
characteristics are indicative of exogenously determined neighborhood charac-
teristics that affect children's performance-such as expenditures on schooling
-even though the effect on performance does not operate through social
interaction. The analogous problems of identification occur in the interpretation
of coefficients of family characteristics in regressions of performance. This
identification problem can only be resolved at a sufficient level of detail that it is
possible to impute individuals' motives. For this reason we shall turn to ethno-
graphies and biographies, which entail a level of thick description at which it
may be possible to discern unambiguously the presence of social interaction.
3 These results conform to the one-sentence recapitulation of previous work by Evans, Oates, an
Schwab (1992) that "many studies find that these peer group effects exist and are quite important."
For an alternative view of the importance of neighborhood effects in empirical studies, see the
survey by Jencks and Mayer (1990).
2. SIMPLE MODELS OF STATUS AND CONFORMISM
Status seeking and conformism can be illustrated by a pair of simple reduc-

tionist models. In each model the agent chooses a variable x to maximize an
indirect utility function. As in Robert Frank's Choosing the Right Pond (1985), in
the case of status behavior utility depends positively on the difference between
the individual's own status and the status of others. In contrast, in the case of
conformist behavior, utility declines as distance between the individual's behav-
ior and that of everyone else increases. Both of these twin models have
representative agents, so that, in equilibrium everyone behaves in exactly the
same way and "the behavior of everyone else" is well defined. These models
show, as is already well known, that if people are either conformist or status-
seeking, their behavior generates important externalities. Although the equilib-
rium outcomes are the same for all individuals in these models, they represent
status-seeking or conformism because ex ante people would prefer to be
respectively better than others in the status-seeking model, or more like others
in the conformist model.
The Status Model
In the status model the individual chooses the status-producing variable x to

maximize the indirect utility function
(1) U=-d(x-x)-ax2+bx+c.
The person loses utility in amount d(x -x) insofar as she falls behind everyone
else in her choice of x, where x is the choice of everyone else. In addition, x
has an intrinsic value to her of -ax2 + bx + c.
Faced with this decision problem, each individual, in equilibrium, chooses
(2) x=(b+d)/2a.
Because of the competitive race for status the value of x exceeds its optimum by
the amount d/2a. The externality is similar to what occurs in the overfishing of
a lake or of an ocean: Status seeking people fail to take full account of the
consequences of their own social positioning on the welfare of their friends and
relatives, just as fishermen fail to internalize the effect of their behavior on the
availability of fish for others.
The Conformist Model
We shall now examine the alternative case-of conformity-in which the

individual wants to minimize the social distance between herself and others.4 In
this case she does not seek to be better than other people, but instead wants to
4The classic work on the economics of conformism is Jones (1984).
be as much like them as possible. I will later explore some of the reasons why
individuals want to conform.
In the twin model on conformity, the utility function is
(3) U=-d1x-xI-ax2+bx+c.
The agent loses utility dlx - ii from failing to conform to others. As before, x
has an additional intrinsic utility of - ax2 + bx + c.
And, in equilibrium, since everyone is alike,
(4) x =x.
In this model, there are multiple equilibria as long as d is greater than zero.
In fact the range of values of x between (b - d)/2a and (b + d)/2a constitutes
a zone of equilibria for given d. As d, the parameter describing the taste for
conformity, increases, the equilibrium zone will expand. The reason for multipl
equilibria is quite simple. For the individual there is a discontinuity in the
marginal utility of x. For x less than x, an additional unit of x increases the
conformist component to utility by d. However, beyond x an additional unit of x
reduces the conformist component of utility by d. Because of this discontinuity
the choice of x at xi is no longer a marginal decision; over a range of parameter
values, the individual would choose x at x. A mathematical consequence of this
accumulation of choices of x at x because of the discontinuity of marginal
utility is a range of possible equilibria-for both x and x.
Over the range (b - d)/2a <x < (b + d)/2a a marginal change in one of the
parameters that affect utility-a, b, c, or d-will have literally no effect
whatsoever on the equilibrium value of x. Thus what seems like a truism-that
people pursue education, or discrimination, or, as the parodies of Becker's work
have alleged, brushing one's teeth (Blinder (1974)), up to the point where the
marginal cost is equal to the marginal benefit-is not a truism. In this model,
over some range of parameter values, this standard result does not even apply.
The welfare differences between status seeking and conformity are clear in this
example. In the seeking of status there is a clear tendency to overindulge in the
status-producing activity; it is overproduced by d/2a. With conformity, the
tendency to mimic the status quo can result in either underproduction or
overproduction of x, in amount ranging from -d/2a to d/2a.
Quadratic Utility
It is important to recognize that the existence of multiple equilibria in the

conformist model is sensitive to the choice of utility function. As the distance
between the representative individual and others goes to zero, the marginal
utility of moving closer, in the utility function given by (3), does not fall to
If, in contrast, the utility function were quadratic, of the form
(5) U=-ax2+bx+c-d(x T)2,
there would be only one equilibrium value of x-precisely at the optimum value
of x = b/2a. Although there is conformist behavior with this quadratic utility
function, the multiple equilibria disappear because the marginal utility of x is
nowhere discontinuous. I have considered utility functions of this sort-with the
marginal utility of decreased distance vanishing at zero-to be a special case. As
discussed below, in an analogue to a gravity model the marginal utility of moving
closer at zero distance would be at the opposite extreme-not zero, but infinity.
3. A MODEL OF SOCIAL DISTANCE
The "representative agent" models presented so far ignore individual differ-

ences and as a result make it difficult-if not impossible-to illustrate the
existence of subgroups that behave differently from the majority of the popula-
tion. The potential existence of such groups-subgroups in the population with
their own norms and values-is one of the most important consequences of
social interaction theory. We shall now introduce sufficient heterogeneity into
the previous models to show how social interaction can produce such groups,
which can be interpreted as stable subcultures or social classes.
In broad terms, the new heterogeneous model can be described as follows. As
before, each individual must choose a decision variable x and there is an
inherent benefit associated with each potential choice of x. To introduce
heterogeneity in social interactions, I shall let individuals occupy different
locations in social space. Social interaction, which is represented as mutually
beneficial trade between individuals, will increase with proximity in this space.
Current social location is acquired and dependent on the decision x, but past
social location for each individual is inherited. Both the acquired and the
inherited social position affect social interactions-reflecting the social aspects
of the adage on my friend's living room pillow: "It is better to be nouveau than
not riche at all." Finally, to complete the model, individuals are given static
expectations about the positions to be occupied by others in social space.
Such a model can portray stable groups in low level equilibrium traps because
individuals' incentives to choose x to conform with those whose inherited social
locations are close may overwhelm their incentives to choose x for intrinsic
reasons. This is a model of conformist behavior because those who are closer in
social space are more likely to interact. Therefore there is an incentive to
conform, with a further bias toward conforming to those who are initially close
by in inherited distance; just as firms, geographically, have an incentive to
remain close to their current customers, individuals too have good reason not to
abandon their relatives and current friends, who, by analogy, are their current
customers for social exchange.
Because one dimension of social location is geographic, the concept of social
proximity yields a generalization of a geographic model.5 Sociologists have a
5The concept of social geography and its implications are inspired by the work of Krugman (1990
1991a, 1991b, 1992) on economic geography.
whole lexicon of their own for what might be considered "social geography."
Important concepts in sociology are, for example, social networks and reference
groups, usually comprised of individuals who are socially close.
Although the trades influenced by position in social space may be economic,
we conceive of them primarily as social. We shall give a positive value in the
model to the benefits from social interaction, as would occur from love and
friendship. However, with no change in the behavior of the model, with the
inclusion of a negative constant term the social interactions could yield negative
benefits. Such negative benefits from social interaction may also reflect reality
since not all social exchange contributes positively to utility and fear of negative
sanctions, due, for example, to jealousy and envy, are potentially as important a
motive for conformity as the desire for the positive benefits of love and
friendship. The jealousy and envy of friends, relatives, and neighbors (see Mui
(1995)) result in the same incentives to keep close to one's origins as the positive
benefits portrayed in the model, and, irrespective of whether the total returns
from social exchange are positive or negative, the model gives the same negative
marginal incentives of social distance and similar analyses of behavior.
Benefits from Trade and Social Location
To generate examples of the behavior I have described it will be necessary to

give functional form to the benefits of social interaction between individuals at
different locations in social space. I shall follow the common practice of basing
trade, and, correspondingly, also the benefits of trade, on a "gravity" model. In a
pure gravity model the trade between two countries is proportional to the GNP's
of the respective countries (analogous to their respective mass) and inversely
proportional to the square of the distance between them. In practice, to
estimate such models (for example, Frankel and Wei (1993)) some modifications
are usually made to this pure formulation. For our purposes we shall make two
modifications. In the gravitational model in physics (Feynman (1963, p. 7-1))
with the force of attraction inversely proportional to the square of the distance,
attraction increases without bound as the distance between the two masses
approaches zero. Adoption of this exact formula to represent social exchange
would place excessive benefits on proximity in social exchange. To dampen the
effects of proximity I will therefore modify the formula so that as distance
approaches zero the benefits from trade will rise but not without limit. I also
want the concept of social distance to be a bit richer than would be captured in
a pure gravity model since I want social exchange to depend jointly on the
differences between peoples' current positions and also their inherited positions.
A formulation that incorporates both of these desirable modifications to the
pure gravity model assumes that trade depends on the inverse of the product of
a constant plus the inherited social distance and a constant plus the acquired
social distance. This product substitutes for the square of the distance in the
canonical gravitational model of trade. In the pure case in which the constants
are both zero and there is no distinction between the acquired and the inherited
social position, this will be an exact gravity model of the benefits from trade.
These modifications of the gravity model to the volume and benefits to trade
yield the expected value of the benefits of trade between i and j as: e/(f +
do ij)(g + de ij), where do ij is the initial social distance between i and j and d
is the expected final social distance between i and j. This functional form has
several beneficial features. First, as in the earlier conformist model, in the
neighborhood of d = 0, there is still some marginal benefit from closer relations,
but not infinite, as in the unmodified gravity model. Second, as either the initial
or the acquired social distance between a pair of individuals increases, the value
of social exchange between them declines asymptotically toward zero.
Intrinsic Returns to Choice of x
The variable x plays two roles in this model. First, it determines acquired
social position; but the variable x also has an intrinsic economic value. For
example, education is an important determinant of earnings. As in the twin
models of status and conformity, we posit that the intrinsic value of x is
-ax2 + bx + c. By focusing on the intrinsic returns to social choice, previou
models of social decisions-such as the choice of education, childbearing, and
discrimination-have ignored the important social externalities that are embod-
ied in the returns from social exchange.
Expectations and Choice of x. The problem confronting each individual i is to
choose x1i contingent on her initial social position, xoi. In order to make this
decision the individual must form expectations about the position of her
potential trading partners in social exchange. Many outcomes are possible
depending upon how these expectations are formed. The simplest assumption is
static expectations that the acquired social position of all the other individuals
will coincide with their initial position. With such static expectations about
social position, d, ', i's expected acquired distance between herself and j wil
lxii -x0jl. In Jones' (1984) model of tradition, a similar assumption-that
the population (of workers) is new in each generation-plays a similar role, as
each new generation finds itself conforming to the traditions of the older,
inflexible half of the population.
In sum, each respective agent i chooses the respective value of x1i to
maximize
(6) i= Ee/[(f+IXoi-xojl)(g+ ix -xIojl)] + [-axi+bx1+ c].

joi
A solution to this maximization determines x1i for all individua

the value of the xoi's.
The key question to ask is whether or not the equilibrium of such a model is
welfare maximizing. Long-run welfare is maximized if each x is equal to b/2a.
Are there situations where the vector of xi's does not converge to a vector
whose elements are all uniformly b/2a? The answer is clearly yes: depending on
initial endowments many possible equilibria are obtainable.
We shall first give a simple example where convergence to a social optimum
does not occur. We shall then describe ethnographic evidence that shows that
the clustering illustrated by our example accords with experience, leading to
nonoptimal choices in education, child-bearing, and racial discrimination.
4. A SIMPLE EXAMPLE OF CLASS STABILITY
Figure 1 depicts a simple three-person example. This figure shows the inher-
ited social positions of three persons: 1, 2, and 3. In this figure the inherited
social distance between 1 and 2 is small, but the inherited distance between 1
and 3 and also between 2 and 3 is large. In addition, 3's initial position is close
to the social optimum b/2a.
If 1 and 2 are initially fairly close to each other and 3 is fairly distant-as
pictured-and if the value of social exchange is sufficiently high relative to the
intrinsic value of x, there is one stable solution in which 1 and 2 will exchange
each other's positions while 3 will choose a point that is close to the economic
optimum, only slightly influenced by the possibilities of trade with 1 and 2,
because they are socially distant. The proof of this proposition is given in the
Appendix. This outcome occurs if xo1 and x02 are sufficiently close to each
other, if x03 is sufficiently distant, and if the value of social exchange relative to
the marginal intrinsic value of x is sufficiently high. And, since person 3 does
not much value trade with persons 1 or 2 since she is initially so socially distant
from both of them, she chooses a value of x that is close to the economic
optimum value of b/2a.
Commentary
The manner in which expectations are formed plays an important role in

determining the nature of the equilibrium of the model. With rational expecta-
tions, either in the three-person model or in its generalization described above,
the social optimum with x equal to b/2a for all individuals is a possible
equilibrium. Indeed, when expectations can be altered so that everyone expects
the social optimum to be achieved, in fact it will be. Later we shall describe two
"miracles" of social intervention in which expectations about the proper mode
of behavior were at least arguably changed for everyone in the relevant social
I I ... I
xOI x02 x03 b/2a
FIGURE 1.-The three-person model.
network, resulting in equilibria close to the economic optimum. But the social
optimum need not be the only rational expectations equilibrium; for example,
the solution I have described to the three-person model with static expectations
will also be an equilibrium with rational expectations if persons 1 and 2 are close
enough.
For simplicity we have represented only two time periods in the model: the
initial time period, denoted 0, and the time period in which the choice is made,
denoted 1. An acquired value of each xi will be chosen conditional on the
vector of initial endowments xi. If the values of xi that are acquired in a gi
period become the initial values that are used in next period's choices, this
model will describe the full dynamic path of the xi vector.
The conformist social distance model is based on an analogy with the
Newtonian model of gravity. Because of the mutual advantages of possible social
exchanges, people are motivated to move toward others who are close to them
in social space. In the model this is seen in the behavior of 1 and 2, who are
close to each other. Like twin stars, 1 is attracted to 2, while 2, symmetrically, is
attracted to 1. Because of the analogy with the physical model of gravity we may
also expect some of the other astronomical consequences of the Law of Gravity
to be mirrored in social space. Perhaps the most obvious consequence of gravity
is the existence of concentrations of mass such as stars and planets. The
analogue in a social distance conformist model is the existence of subcultures.
While it is useful to point out the similarities between the social distance
model and the Newtonian model of gravity, the analogy is not complete and the
differences are also of some interest. We have already discussed the modifica-
tions made from the analogue to a pure gravity model in our formula for the
gains from social interaction. We have made a further modification to the
Newtonian system by deriving the law of motion of the system from the equation
for "force" in quite a different fashion. The gravitational law of motion in
physics is derived from the formulae for force by the assumption that "an object
responds to a force by accelerating in the direction of the force by an amount
that is inversely proportional to the mass of the object" (Feynman (1963, p. 7-1)).
Instead of deriving such a differential equation for acceleration, the law of
motion in our system comes from a difference equation, in which each person
chooses his/her best social position on the assumption that the position of
others will remain unchanged. This difference in construction of the law of
motion may qualitatively change the nature of long-run equilibria. If we had
assumed that the system was initially at rest and derived the law of motion in
the social location model analogously to the physical gravitation model, the long
run would always collapse to the single point with all individuals choosing x at
the social optimum, equal to b/2a. However, with the assumed choice-theoretic
law of motion of our social system, it is quite possible not to get such a collapse.
Each person chooses where he/she wants to be next period; they choose their
point given where everyone else was in the previous period; there are solutions,
as just pictured, in which there is no convergence to the social optimum. In the
previous example if persons 1 and 2 were located at exactly the same point in
social space, there would be a stable long-run equilibrium with persons 1 and 2
at that point and person 3 close to the social optimum, b/2a.
It is also useful to note that the preservation of a stable system does not
necessarily depend on the property that the marginal utility of getting closer
falls to zero as the distance approaches zero. We could easily imagine a situation
in which individuals have no mass, but there are groups massed at given points
in social space. Each individual finds that his/her optimum choice of x is
exactly the value of his/her inherited social location. It is a nonmarginal
decision to part from his/her subgroup because they have a measurable mass
even though the marginal value of departing from a single individual just a little
bit happens to be zero.
The astronomical analogy of concentrations of mass in stars and planets offers
insight into important features of the social and economic landscape, particu-
larly the division of society into different social classes with distinctive manners
and customs. As discussed in the introduction, the existence of social clustering
due to social interaction is hard to establish from econometric evidence. But the
existence of stable dialects for subgroups of the population can only be inter-
preted as due to the clustering of social interactions such as modeled in this
paper. (See also Durlauf (1993) and Brock and Durlauf (1995).) Thus dialects act
as a diagnostic for social interaction. In discussing the model, we emphasized
that the variable x could be interpreted as the choice of education, but
alternatively, x could represent pronunciation or language, with the model
showing theoretically why there might be a correlation between a choice
variable, such as dialect, and subcultural membership. Massey and Denton
(1993) relying on the intuitive notion that differences in dialect reflect social
distances argue that the increasing disparity between White and Black English
in the United States signals growing effective racial segregation. William Labov,
a modern-day version of Shaw's Professor Higgins, has painstakingly analyzed
the differences between White and Black speech.6 The studies of Labov and his
colleagues reveal that Black English of different metropolitan areas has con-
verged, while it has been simultaneously diverging from Standard American
English. Massey and Denton have interpreted these findings as a symptom of
the slow pace of integration in the United States.
In the model we pictured both members of the conforming group, 1 and 2, as
choosing too little of the variable x, in contrast to 3, who chose close to the
optimum quantity. Viewing x as education, the model predicts that education
may be underpurchased (relative to the welfare optimum) by the conforming
group, which we shall later identify with inhabitants of the urban ghetto.
Similarly, the variable x could represent drug consumption, with the model
symmetrically predicting overconsumption in the ghetto relative to the welfare
optimum. This could occur if 1 and 2 conform to each other's overconsumption
6See Labov (1972, 1975) and also Labov and Harris (1986). Also see Baugh (1983).
of the variable x, drugs, while the distant 3 chooses her value of x close to the
economically maximizing level.
5. CLASS STABILITY AND THE ECONOMICS OF THE UNDERCLASS
As we have emphasized, a natural interpretation of x is educational attain-

ment. The motivation for this interpretation comes from ethnographic observa-
tions. To this day the classic ethnography of working class youth is William
Whyte's (1955) Street Corner Society based on observations in the Italian North
End of Boston in the late 1930's. Whyte observed two groups of boys, dubbed
for contrast "the corner boys" and "the college boys." In contrast to the
college-bound group Whyte's corner boys mainly spent their time hanging
around-in part, of course, because this was the Great Depression and jobs
were not available. Follow-up studies showed that they later drifted into factory
jobs. In the course of his description Whyte mused on the theoretical question:
what was the difference between the upwardly mobile college boys and the
stay-at-home corner boys? In particular, Whyte wondered why Doc, the leader
of the streetcorner gang, was not upwardly mobile. Because of his intelligence,
and natural curiosity, which Whyte considered exceptional, school would have
been easy for him. Whyte concluded that Doc did not seek extra education out
of loyalty to his group, whom he would be abandoning were he to advance
beyond them educationally. Doc would also consider himself as deserting the
neighborhood. In contrast Whyte viewed the "college boys," not as exemplars of
the American dream of getting ahead, but rather as coldly disloyal to their
neighborhood, and also potentially to each other. This is Doc's own view about
his lack of social mobility:
"I suppose my boys have kept me from getting ahead ... But if I were to start over
again-if God said to me, 'Look here, Doc, you're going to start over again, and you can
pick your friends in advance,' still I would make sure my boys were among them-even if
I could pick Rockefeller and Carnegie ... Many times people in the settlement ... have
said to me, 'Why do you hang around those fellows?' I would tell them, 'Why not? They're
my friends."' (Whyte (1955, p. 108)).
The behavior of Doc and his corner boys corresponds, but not exactly, to the
characterization of persons 1 and 2 in the three-person model. We might
imagine that Doc is person 1, who in the absence of the other corner boys
(collectively characterized as person 2) would choose a point close to the social
optimum b/2a. Instead, because he wants to maintain his social contacts with
the corner boys, he chooses a point close to them. There is a difference between
the model and reality: in the model in the absence of being held back by their
peers, individuals would all choose a point close to b/2a. In reality in Whyte's
ethnography only Doc is restrained; the other corner boys would choose a point
much closer to their origins. To make the model conform to reality requires a
minor modification; we might imagine that in each succeeding generation one of
persons 1 or 2 would of his own volition choose a point close to b/2a, but not
the other. This loyalty to the social network in each generation trumps the
evolution toward the social optimum that would otherwise occur.
Whyte's interpretation of Doc's motivation translates into a theory of the
demand for education that differs substantially from the standard model (due to
Becker). While it may be true that Doc, like everyone else, pursues his
education up to the point where the marginal benefit of extra education is equal
to the marginal cost, the relevant cost includes not just lost wages, tuition, and
the disutility of schoolwork-the ordinary interpretation of those costs. In
addition the cost of the additional education includes the lost contacts with
others. In the standard model with lost wages, tuition, and the disutility of the
additional effort as the major costs of education, educational choices involve few
externalities. In contrast, if the cost of education includes disutility from
deviation from others in one's social network, the potential for large externali-
ties is apparent. These externalities explain why even with generous scholarship
aid available, students from low-education backgrounds commonly drop out too
early while those from high-education backgrounds may stay in school too long.
The model, as epitomized in Doc's response to Whyte's question, explains
how social structure-in this case the social structure of the white working class
-tends to reproduce itself. It explains why there is less mixing in American
society than might be expected if purely individual incentives combined with
purely individual values.
Similar conclusions arise from Lee Rainwater's classic study (1970) of the
Pruitt-Igoe housing projects in St. Louis in the late 1960's. This study presaged
the subsequent developments in the Black inner city over the next 25 years.
Rainwater's research questions the "old" sociological paradigm of discrimina-
tion in which "white cupidity [created] black suffering (1970, p. 3)." Such "old"
sociological reasoning is embodied in the economic models of Becker where
white discrimination coefficients cause decreased demand for Black labor, and
possibly also for the products of Black-owned capital-so that Black incomes
are significantly reduced. According to the "new" sociology (Rainwater (1970, p.
4)), which Rainwater's ethnography supports, "white cupidity creates structural
conditions highly inimical to basic social adaptation to which Negroes adapt by
social and personal responses which serve to sustain the individual in his
punishing world but also to generate aggressiveness toward the self and others
which results in suffering directly inflicted by Negroes on themselves and on
others." In these personal and social responses individuals conform to the norms
of the ghetto society, and the cycle of poverty is passed from one generation to
the next. The meaning of this theoretical perspective, and its correspondence
with the special three-person model above, can be illustrated by the biographies
of two of Rainwater's subjects.
The first of many portraits of residents of the Pruitt-Igoe housing project in
Rainwater's Behind Ghetto Walls is of Thomas Coolidge, a 21-year old Afro-
American laborer. We first meet Coolidge as a young married father employed
at a drive-in diner, living in the projects, but with his life basically in order, as
indicated by the tidiness of his apartment. Over the next year Coolidge's life
disintegrates as he becomes childishly dependent on his parents after losing his

job and his wife, who has pursued numerous affairs with other men. This is how
Coolidge theorizes about ghetto life (p. 19):
"This place is like a jungle but you can't get away from it so you submit yourself and
submitting yourself is the worst thing that you can do. Submitting is like letting yourself do
what other people want you to do. It's like tilting the scale, if you have the bad on one side
and the good on the other and the bad outweighs the good then the bad comes out best in
the end. Now if you are one of the good and you're constantly coming in contact with one
of the bad and there are a whole lot of bad and very few good, then you don't have any
other choice but to go along with the bad whether you want to or not" (italics added).
Why do people, in Coolidge's words, "do what other people want them to
do?" In the face of poverty and joblessness the other world is too hard to obtain,
and the social pressure to conform is simply too great. This interpretation of
Coolidge's predicament, consistent with Coolidge's own view, conforms to the
three-person model in which individual 1 and individual 2 respectively conform
to one another.
Coolidge's story typifies how men were captured in the orbit of ghetto life.
The comparable dynamics for women is epitomized by the life-story of Alice
Walker, who becomes a single-parent teenage mother, despite early ambitions
otherwise. We first meet Alice Walker at the age of 15 when she is aspiring to
become a secretary and move out of the ghetto. In fact Alice becomes pregnant
at an early age. To have persisted in her prior ambitions would have kept her
from participating in her friends' web of activities centering around boyfriends
and babies. Between lonely ambition and poverty among friends, Alice, under-
standably, like so many others, chose friends and companionship. This is exactly
what the three-person model, in this case referring to the choice to have
children at a young age, is meant to illustrate.
Ethnographies, like Rainwater's and Whyte's, tend to sample ordinary people.
In Behind Ghetto Walls especially we meet young people, typically hopeful for
the future, and then observe how the forces for conformity with the life of the
ghetto turn these hopes sour, as these young people follow everyone else around
them. In contrast, biographies of those who have escaped the ghetto or working
class offer a different perspective. If our theory of social conformity is correct,
upward class mobility, especially out of the ghetto, must be fraught with
difficulties that are only surmounted by the exceptional.
Probably the most revealing autobiography in this respect is Manchild in the
Promised Land, the autobiography of Claude Brown (1965), who grew up in
Harlem in the 1940's and 50's. By the age of eight he had already been shot
while attempting a robbery, had set his own house afire, had been hit by both a
car and a bus, and been thrown into the Harlem River. By luck he was sent to
the Wiltwyck School for Boys, a special reform school where under the influence
of the counselors, he decided to reform. He worked days, attending remedial
high school at night; but in order to make it he found it necessary to remove
himself physically-to Greenwich Village-away from his family and friends.
A biography of an Afro-American youth from Harlem and an autobiographi-

cal essay by a Mexican-American from Sacramento both provide additional
insight into the difficulties experienced by those achieving upward mobility
through education. The biography (Anson (1985)) was motivated by newspaper
reports describing the shooting of an African-American youth in the Morning-
side Heights neighborhood of New York in an attempted robbery of a plain
clothes policeman. The case drew national attention because the young man had
just graduated with honors from Phillips Exeter Academy, perhaps the nation's
most elite preparatory school, and had just won a full four-year fellowship to
Stanford. Robert Anson decided to investigate fully the circumstances leading to
this tragedy. The boy, Eddie Perry, had taken part in the so-called Special
Program of his local New York public school. Over the course of 21 years 230
talented students from this program had been sent to 72 different prep schools
(Anson (1985, p. 47)). The program has achieved great success, as judged by a
comparison of the careers of its graduates to those of their peers left in the
inner city. However, when Anson investigated the fate of Eddie Perry he
discovered that the success of the program had been achieved at the cost of
considerable psychological pain, because the students did not fit naturally into
either their old world of the inner city or their new world of the prep school. In
the lingo of the model their social distance at prep school was too great for
psychological comfort; but then when they returned home to the inner city they
were, similarly, ill at ease. A close mentor of Eddie's, who had also gone to prep
school from the inner city, later graduating from Yale, explained the psychologi-
cal tension:
"So that leaves people on the street. And how in the hell is he supposed to talk to them?
... this kid couldn't even play basketball. They ridiculed him for that, they ridiculed him
for going away to school, they ridiculed him for turning white. I know because he told me
they did." (Anson (1985, p. 205)).
This mentor viewed Eddie's death as a suicide induced from the stress of living
in two separate and dissonant cultural worlds.
Richard Rodriguez' autobiographical essay (1982) depicts the clash in cultures
between the customs and language of his Mexican-American home and the
Gringo culture that he learned in school and at college. Rodriguez recounts with
feeling that his family (especially his extended family) considered him increas-
ingly alien, just as Eddie Perry's peers in Harlem no longer considered him
Black, because of what he had learned at prep school. Rodriguez recounts that
as English became his dominant language:
"Pocho then they called me. Sometimes, playfully, teasingly, using the tender diminutive
-mi pochito. Sometimes not so playfully, mockingly, Pocho. (A Spanish dictionary defines
that word as an adjective meaning 'colorless' or 'bland.' But I heard it as a noun, naming
the Mexican-American who, in becoming an American, forgets his native society.)"
(Rodriguez (1982, p. 29)).
Although they lived at opposite ends of the continent and came from different
ethnic backgrounds, and although one life ended in tragedy while the other
continues as a remarkable success, Eddie Perry and Richard Rodriguez bore

psychological costs of educational attainment that were strikingly similar: in the
one case there was the cost of being labeled a White man in a Black neighbor-
hood, in the other case there was the cost of being labeled a Gringo by his
extended family. For both there was also the difficulty of living in the culturally
alien white world. These were perhaps the costs that Doc, a generation earlier,
had decided not to bear when he had made the opposite choice-to be a
streetcorner boy, and not a college boy.
Stack's All Our Kin (1974) documents the intensity of social exchange for poor
Black families and shows how these social exchanges are disrupted by upward
social mobility. Stack details how these families swap and share a wide variety of
things-furniture, cars, meals, childcare services, places to live, and, in times of
distress, each other's children. In 1970, at the time of Stack's field work in an
industrial town close to Chicago, almost 10 percent of Black children were living
with neither of their biological parents. She reports survey results that 20
percent of Black children in the poor area of this town were living with relatives
other than their mothers.
Stack gives an example of how going middle class results in the disruption of
these exchange relations, especially with respect to the mutual sharing and
discipline of children (Stack (1974, pp. 76-79)). She describes the relations
between two sisters, Ethel and Wilma, and among their children and grandchil-
dren. A third sister, Ann, and her brood make an appearance. Ann has become
middle class and the distance between her and her sisters is especially clear in
the mutual care of their respective children. For example, in an emergency,
Ann's daughter Vilda let Georgia (Wilma's daughter) take care of her child
Betty in return for pay (ten dollars per day.) But instead of allowing her to use
her best judgment as childcare person and aunt, Vilda gave instructions to
Georgia not to yell at Betty or to spank her. Georgia responded to this lack of
confidence in her parenting abilities by preventing Ann from disciplining her
own daughter. When Ann tried to prevent one of Georgia's daughters from
chasing one of her own grandchildren with a red hot poker at a family barbecue,
Georgia responded: "You won't let me touch your granddaughter, so don't you
tell my child what to do." Stack interprets this event as more than the clash
between two different personalities, but as symptomatic of the clash between
two different life styles.
These snippets of ethnography, autobiography, and biography, of course, do
not statistically prove that people try to conform to the social norms of their
friends and relatives. Rather, such vignettes provide token reminders of what
most of us already appreciate instinctively, which is the reality underlying the
social distance/social exchange model of demand. Relative to education, we
know that the education sought by most people will be the education that meets
the approval of friends and relatives. Like children on the merry-go-round who
look up to see if anyone is watching,7 youth who are attaining an education look
7The image here of course comes from Erving Goffman.
around to see if their work is being appreciated by the adult and teenage worlds
around them. The absence of a favorable response takes away the fun.
Economists have modeled the demand for children as if they were consumer
durables. But children are a special kind of consumer durable whose enjoyment
is enhanced by hobby clubs of other mothers and adoring relatives who share
the pleasures of the new models. And, if everyone else is a member of a baby
club, it is lonely to stay out. Thus, for example, Alice Walker chose to be a
welfare mother rather than a secretary because the road to becoming a secre-
tary was lonely, whereas being on the dole made her one of the gang.
6. FURTHER OBSERVATIONS WITH POLICY IMPLICATIONS
In the social distance model described above each person chooses her
respective position in social space under the static expectation that the social
positions of others would remain fixed. In the equilibrium this assumption
turned out to be roughly true, since 1 and 2 were close and they merely
interchanged positions. And 3 did not change her position much either. How-
ever, in special situations people may expect others to move in tandem rather
than to remain in place, generating a motive for social movement that is
normally absent. Interventions in a closed environment that attempt to change
the social position of an entire social network simultaneously and, as a conse-
quence, to alter the expectations of the individuals in the network about their
neighbors' behavior can be quite successful in generating major social change.
In terms of the model if everyone is expected to change their social position to
the economic optimum, then that optimum will be the new equilibrium.
In practice the best known intervention where such a change in expectations
most likely occurred is Eugene Lang's famous offer to give a college scholarship
to every student of a sixth grade class in Harlem. Of the 51 students who
remained in the New York area, 40 were considered likely to go to college six
years later. Even more remarkable, all of these students could easily have
obtained either scholarships or loans in the absence of Lang's program. (See
Ellwood (1988, pp. 125-126).)
How then should we account for this success? One possibility is that Lang was
successful because he put a lot of resources into the program and actively
engaged a dedicated social worker who organized group activities, enlisted the
support of parents, and also intervened at students' times of crisis. Lang did not
just offer the money and then walk away from the children to await later claims
for scholarship money, if any.
An alternative explanation is that the experiment was successful because the
students formed a cohesive group in which each member received reinforcement
from others who, like themselves, were on the academic track toward graduation
from high school. In terms of the model each individual student would expect to
be more isolated from her peers if she dropped out than if she kept up with her
school work. Additionally, these students were perceived by the community as
the recipients of a rare bit of good luck. By pursuing this unusual opportunity,
the community would not identify members of the class as departing from
behavioral norms.8 Both of these explanations -in terms of group dynamics and
in terms of signaling-conform to our model of social distance. The students
took advantage of Lang's rash promise because they could do so without
creating great social distance between themselves and their social network.
Another social intervention commonly cited for remarkable results occurred
in two New Haven schools in the 1970's. In this case too the intervention
reversed the normal incentives governing behavior. When James Comer and his
team (1980) first came to the New Haven schools they found a great deal of
anger among teachers, students, administrators, and parents-all of whom were
frustrated because they felt that they were being unfairly treated. Comer and his
assistants used human relations tools to teach responses that were more con-
structive than anger. To give one concrete example, Comer taught methods for
dealing with misbehavior resulting from children's frustration. He relates the
story of an eight-year-old boy who was sent North by his family from North
Carolina and was dropped off at school by an aunt on her way to work. When
the boy was left in the classroom by the school principal without any introduc-
tion, the teacher showed some annoyance at the additional burden. In anger, the
eight-year-old kicked the teacher and ran out of the classroom. By creating a
school environment in which the teacher was trained to find out the reason for
the child's unhappiness and misbehavior, the teacher and the student were able
to reach a mutual understanding: the teacher would respect the boy's feelings
and the boy would behave. Once the lesson had been learned that understand-
ing, not anger, was the way to deal with difficulties, academic success followed.
Comer tells a vignette that shows that this message had been appreciated by
the students (as well as the faculty and parents). When a student new to the
Comer school had his foot stepped on, he raised his fists, ready to fight. Told by
another youngster "Hey man, we don't do that here," the transfer student read
the expressions on the faces of the others around him and dropped his fists
(1988, p. 219). By creating a school spirit in which everyone reacted to anger
with the attitude "we don't do that here," each member of the school commu-
nity feared, like the transfer student, at least some degree of ostracism if he did
not conform.
In terms of the model, in the old equilibrium anger was acceptable and often
contributed to an individual's acceptance in her respective social group. In this
equilibrium anger is uninhibited because it contributes positively to social
exchange. In contrast, in the new situation, after Comer's intervention, each
person understands that anger will have the opposite effect: a show of temper
would reduce acceptance by one's peers and even by children on the playground.
By reversing the social code for acceptance or rejection, a new socially optimal
equilibrium was achieved in which learning could be the focus of attention.
8 This observation corresponds to Bernheim's interpretation of conformist behavior. See Ber

heim (1994).
7. CONCLUSIONS
Although the models presented here are only extensions of the early Becker
framework for social decisions, it is my hope that the incorporation of these
social externalities provides a broader and more accurate framework for the
rational analysis of social choice. In contrast to standard economic models the
social distance approach provides insight into sociological phenomena including
class structure and patterns of behavior such as dialect.
The pervasive externalities that influence decisions in the social distance
model have implications for economic policy. Because group interactions are an
important influence on individual decisions, the analysis of social programs must
include an evaluation of an intervention's impact on group interactions and not
just the direct effects of the program. For example, in education the returns to
programs such as art and athletics cannot be measured simply by their direct
effects on grade point averages and added earnings of the participants. Similarly,
the pros and cons of tracking in the schools depends upon the group interac-
tions that are thereby engendered. Comer's school intervention suggests that
with harmonious social relations, academic achievement may be easy to attain,
even in schools in the most disadvantaged neighborhoods. Conceivably, there
could be high payoffs to drama programs that enable students to learn an alien
culture and language while curiously pretending that they are not.
Several recent books and reports (see, for example, Eisenhower Foundation
(1993) and Schorr (1988)) have endorsed a community-wide, multiple-solution
approach to the problems of the inner city. The theory of social interactions
might be interpreted as providing a rationale for such an approach. But the
analysis here suggests that the community is endogenously defined in terms of
peoples' sense of social location. What may appear as a community to an outside
reformer (a city neighborhood, for example) may be too large a unit in which to
encompass the social interactions involved in social exchange. As Comer has
shown, these externalities may be possible to capture in small, near total
institutions such as schools.
Dept. of Economics, University of California at Berkeley, Berkeley, CA 94720,

U.S.A.
and
The Brookings Institution, 1775 Massachusetts Ave., N. W, Washington, D.C.
20036, U.S.A.
Manuscnipt received August, 1995; final revision received November, 1996.
APPENDIX
This appendix will show that in the three-person example person 1 chooses x at the initial
position of person 2 and, similarly, person 2 will choose x at the initial position of perso
and x02 are sufficiently close to each other, if x03 is sufficiently distant, and if the value of social
exchange relative to the marginal intrinsic value of x is sufficiently high. And, if person 3, who is
socially distant from persons 1 and 2, does not much value trade with persons 1 or 2 she will choose
a value of x that is close to the economic optimum value of b/2a.
The mathematics is surprisingly simple: Consider person l's choice of x1l. We shall show that
under the appropriate conditions it will be chosen at x02. This variable will be chosen at the point
where the derivative of U1 turns from positive to negative. The derivative is well-defined at all but
the two points, x1l =x02 and x,1 -x03, where instead there are left-hand and right-hand derivatives,
but of different magnitudes. According to (6) in the interval x1l <x02,
(7) U1 = [e/(f+ (XO2 -x01))][1/(g - (xll -X02))]
+ [e/(f + (x03 -x01))][1/(g - (x1l -x03))] - ax1 + bx11 +
Note that the quantities x1l -x02 and x1l -x03 are both negative in this range, since
also x1l <x03.
Differentiating (7) we find that in the range x1l <x02,
(8) 8U1/8xll = [e/(f+ (XO2 -x01))][1/(g - (Xll -X02)) ]
+[e/(f+ (X03 -x01))][1/(g- (Xll -X03))2] + [-2axll +b].

In this range each of the square-bracketed terms is positive. Remember that - 2ax1l + b
since in our description of xo1 and x02, 1 and 2 were both underinvesting in x. In consequ
Xll ?X02.
For the range x02 <x11 <x03, the value of U1 is
(9) U1 = [e/(f+ (XO2 -x01))][1/(g + (xll -X02))]
+[e/(f+ (x03 -x01))][/(g9- (x1l -x03))] - axh2 + bx1l + c.

And in this range,
(10) 8U1/dxll =-[e/(f+ (X02 -Xoi))][1/(g + (Xll -X02))2]

+[e/(f+ (X03 -x01))][1/(g- (Xll -X03))2] + [-2axll +b].
Note that the sign of the first square-bracketed term changes at x1l =x02 from positive at xO-2 to
negative at x0+. If the distance between x03 and xo1, is sufficiently large, and if the intrinsic value of
x is sufficiently small relative to the value of the exchange, then the first term dominates the sign of
(10) and 8U1/8xll < 0 at x1l =x2. Since 8U1/8xll > 0 at x1l =x-2, this is a point where the sign
of the derivative changes-discontinuously from positive to negative. Such a change of sign does not
automatically guarantee that x02 is the optimal value of x1l for person 1. 8U1/dxll may turn
positive as x1l gets closer to x03: the first term in square brackets becomes smaller in absolute value
as x moves away from x02; and the second term in square brackets becomes larger as x moves
closer to x03. However, if the initial distance between x02 and xo1 is sufficiently small, and,
important, if the initial distance between xo1 and x03 is sufficiently large, 8U1/8xll will not cha
sign from minus to plus as x1l approaches x03, guaranteeing that x02 is preferred to any point in
range, x02 <x11 <x03.
Finally, in the range x1l > x03, the value of U1 is
(11) U1 = [e/(f+ (X02 -Xo1))][(1/(g + (xll -X02))]
+ [e/(f + (x03 -x01))][1/(g + (x1l -x03))] -a41 +bx1 +?c.
And in this range we find
(12) dU1/8xll =-[e/(f+ (X02 -XO1))][(1/(g+ (Xll -X02)) ]
-[e/(f+ (x03 -x01))][1/(g+ (xll -x03)) ] + [-2axll +b].
There is no guarantee that 8U1/8xl

for x02 <x11 <x03; nevertheless, if the
small, 8U1/8xll will be negative thro
value of x is sufficiently small relative to the value of social exchange, and if 1 and 2 are sufficiently
distant from 3 and also from b/2a, the optimal value of x1l will be x02.
A similar proof will show that under these same conditions the optimal value of x12 will be xo1,
and x13 will be chosen close to b/2a.
REFERENCES
AKERLOF, G. A. (1976): "The Economics of Caste and of the Rat Race and Other Woeful Tales,"
Quarterly Journal of Economics, 90, 599-617.
(1980): "A Theory of Social Custom, of Which Unemployment May Be One Consequence,"
(1985): Discriminatory, Status-Based Wages among Tradition-oriented Stochastically Trading
Coconut Producers," Journal of Political Economy, 93, 265-276.
AKERLOF, G. A. AND J. L. YELLEN (1994): "Gang Behavior, Law Enforcement and Community
Values," in Values and Public Policy, ed. by H. J. Aaron, T. E. Mann, and T. Taylor. Washington,
D.C.: Brookings Institution Press, 173-209.
ANSON, R. S. (1985): Best Intentions: The Education and Killing of Edmund Perry. New York: Random
House.
AXELROD, R. (1986): "An Evolutionary Approach to Norms," American Political Science Review, 80,
1095-1111.
BAUGH, J. (1983): Black Street Speech: Its History, Structure and Survival. Austin: University of Texas
Press.
BECKER, G. S. (1964): Human Capital: A Theoretical and Empirical Analysis with Special Reference to
Education. New York: Columbia University Press.
(1968): "Crime and Punishment: An Economic Approach," Journal of Political Economy, 76,
169-217.
(1971): The Economics of Discrimination, Second Edition. Chicago: University of Chicago
Press.
(1973): "A Theory of Marriage: Part I," Journal of Political Economy, 81, 813-846.
(1974): "A Theory of Marriage: Part II," Journal of Political Economy, 82, S11-S26.
(1991): "A Note on Restaurant Pricing and Other Examples of Social Influences on Price,"
Journal of Political Economy, 99, 1109-1116.
BECKER, G. S., E. M. LANDES, AND R. T. MICHAEL (1977): "An Economic Analysis of Marital
Instability," Journal of Political Economy, 85, 1141-1187.
BECKER, G. S., AND K. M. MURPHY (1988): "A Theory of Rational Addiction," Journal of Political
Economy, 96, 675-700.
(1993): "A Simple Theory of Advertising as a Good or Bad," Quarterly Journal of Economics,
108, 941-964.
BENABOU, R. (1993): "Workings of a City: Location, Education, and Production," Quarterly Journal
of Economics, 108, 619-652.
(1996): "Heterogeneity, Stratification, and Growth: Macroeconomic Implications of Commu-
nity Structure and School Finance," American Economic Review, 86, 584-609.
BERNHEIM, B. D. (1994): "A Theory of Conformity," Journal of Political Economy, 102, 841-877.
BLINDER, A. S. (1974): "The Economics of Brushing Teeth," Journal of Political Economy, 82,
887-891.
BORJAS, G. J. (1995): "Ethnicity, Neighborhoods, and Human-Capital Externalities," American
Economic Review, 85, 365-390.
BROCK, W. A., AND S. N. DURLAUF (1995): "Discrete Choice with Social Interactions I: Theory,"
Working Paper No. 5291, National Bureau of Economic Research, Cambridge, MA.
BROWN, C. (1965): Manchild in the Promised Land. New York: Macmillan.
CASE, A. C., AND L. F. KATZ (1991): "The Company You Keep: The Effects of Family and
Neighborhood on Disadvantaged Youths," NBER Working Paper No. 3705, National Bureau of
Economic Research, Cambridge, MA.
COMER, J. P. (1980): School Power: Implications of an Intervention Project. New York: The Free Press.
(1988): Maggie's American Dream: The Life and Times of a Black Family. New York: New
American Library.
CRANE, J. (1991): "The Epidemic Theory of Ghettos and Neighborhood Effects on Dropping Out
and Teenage Childbearing," American Joumal of Sociology, 96, 1226-1259.
DURLAUF, S. N. (1993): "Nonergodic Economic Growth," Review of Economic Studies, 60, 349-366.
EISENHOWER FOUNDATION, MILTON S. (1993): Investing in Children and Youth, Reconstructing Our
Cities: Doing What Works to Reverse the Betrayal of American Democracy. Washington: The Milton
S. Eisenhower Foundation.
ELLISON, G., AND D. FUDENBERG (1995): "Word of Mouth Communication and Social Learning,"
Quarterly Joumal of Economics, 110, 93-125.
ELLWOOD, D. T. (1988): Poor Support. Poverty in the American Family. New York: Basic Books.
EPSTEIN, J. M., AND R. AXTELL (1996): GrowingArtificial Societies: Social Science from the Bottom Up.
Washington: Brookings Institution Press.
EVANS, W. H., W. E. OATES, AND R. M. SCHWAB (1992): "Measuring Peer Group Effects: A Study of
Teenage Behavior," Joumal of Political Economy, 100, 966-991.
FEYNMAN, R. P. (1963): The Feynman Lectures on Physics. Reading, MA: Addison Wesley.
FRANK, R. H. (1985): Choosing the Right Pond: Human Behavior and the Quest for Status. Oxford:
Oxford University Press.
FRANKEL, J. A., AND S.-J. WEI (1993): "Trade Blocs and Currency Blocs," Working Paper No. 4335,
National Bureau of Economic Research, Cambridge, MA.
GLAESER, E. L., B. SACERDOTE, AND J. A. SCHEINKMAN (1996): "Crime and Social Interactions,"
GREIF, A. (1994): "Cultural Beliefs and the Organization of Society: A Historical and Theoretical
Reflection of Collectivist and Individualist Societies," Joumal of Political Economy, 102, 912-950.
JENCKS, C., AND S. E. MAYER (1990): "The Social Consequences of Growing Up in a Poor
Neighborhood: A Review," in Inner City Poverty in the United States, ed. by L. E. Lynn and
M. G. H. McGeary. Washington, DC: National Academy Press, 111-186.
JONES, S. R. G. (1984): The Economics of Conformism. Oxford: Basil Blackwell.
KOLM, S. C. (1994): "The Theory of Reciprocity and of the Choice of Economic Systems: An
Introduction," Investigaciones Economicas, 18, 67-95.
KRUGMAN, P. R. (1990): "Increasing Returns and Economic Geography," NBER Working Paper No.
3275, National Bureau of Economic Research, Cambridge, MA.
(1991a): "Cities in Space: Three Simple Models," NBER Paper No. 3607, National Bureau of
(1991b): "First Nature, Second Nature, and Metropolitan Location," NBER Research
Working Paper No. 3740, National Bureau of Economic Research, Cambridge, MA.
(1992): "A Dynamic Spatial Model," NBER Working Paper No. 4219, National Bureau of
LABOV, W. (1972): Language in the Inner City: Studies in the Black English Vernacular. Philadelphia:
University of Pennsylvania Press.
(1975): "The Logic of Nonstandard English," in Black American English: Its Background and
Usage in the Schools and in Literature, ed. by P. Stoller. New York: Dell Publishing.
LABOV, W., AND W. A. HARRIS (1986): "De Facto Segregation of Black and White Vernaculars," in
Current Issues in Linguistic Theory 53: Diversity and Diachrony, ed. by D. Sankoff. Amsterdam:
John Benjamins Publishing, 1-24.
LOURY, G. C. (1977): "A Dynamic Theory of Racial Income Differences," in Women, Minorities, and
Employment Discrimination, ed. by P. Wallace and A. LaMond. Lexington, MA: Lexington Books.
MANSKI, C. F. (1993): "Identification of Endogenous Social Effects: The Reflection Problem,"
Review of Economic Studies, 60, 531-542.
MASSEY, D. S., AND N. A. DENTON (1993): American Apartheid: Segregation and the Making of the
Underclass. Cambridge, MA: Harvard University Press.
MERTON, R. K. (1968): Social Theory and Social Structure (enlarged edition). New York: Free Press.
MONTGOMERY, J. D. (1994): "Mainstream Norms, Cognitive Dissonance, and Underclass Behavior:
Revisiting Tally's Corner," Rationality and Society, 6, 462-488.
Mui, V. L. (1995): "The Economics of Envy," Journal of Economic Behavior and Organization, 26,
311-336.
RAINWATER, L. (1970): Behind Ghetto Walls: Black Families in a Federal Slum. Chicago: Aldine.
RODRIGUEZ, R. (1982): Hunger of Memory: The Education of Richard Rodriguez. New York: Bantam.
1982.
SCHELLING, T. C. (1971): "Dynamic Models of Segregation," Joumal of Mathematical Sociology, 1,
143-186.
SCHORR, L., WITH D. SCHORR (1988): Within Our Reach: Breaking the Cycle of Disadvantage. New
York: Doubleday.
STACK, C. B. (1974): All Our Kin: Strategies for Survival in a Black Community. New York: Harper &
Row.
WHYTE, W. F. (1955): Street Comer Society, Second Edition. Chicago: The University of Chicago
Press.

Social Distance

Uploaded by

Copyright:

Available Formats

Social Distance

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Social Distance

Uploaded by

Copyright:

Available Formats

Social Distance and Social Decisions

Author(s): George A. Akerlof

SOCIAL DISTANCE AND SOCIAL DECISIONS

A model of social distance is presented that is useful for understanding social

KEYWORDS: Social interaction, education, fertility, urban economics, social class.

TRADITIONAL ECONOMICS HAS BEEN BASED on methodological individualism.

which social decisions were based on individual values.2 Social interaction

2. SIMPLE MODELS OF STATUS AND CONFORMISM

Status seeking and conformism can be illustrated by a pair of simple reduc-

The Status Model

In the status model the individual chooses the status-producing variable x to

The Conformist Model

We shall now examine the alternative case-of conformity-in which the

4The classic work on the economics of conformism is Jones (1984).

It is important to recognize that the existence of multiple equilibria in the

(5) U=-ax2+bx+c-d(x T)2,

3. A MODEL OF SOCIAL DISTANCE

The "representative agent" models presented so far ignore individual differ-

Benefits from Trade and Social Location

To generate examples of the behavior I have described it will be necessary to

Intrinsic Returns to Choice of x

(6) i= Ee/[(f+IXoi-xojl)(g+ ix -xIojl)] + [-axi+bx1+ c].

A solution to this maximization determines x1i for all individua

4. A SIMPLE EXAMPLE OF CLASS STABILITY

The manner in which expectations are formed plays an important role in

5. CLASS STABILITY AND THE ECONOMICS OF THE UNDERCLASS

As we have emphasized, a natural interpretation of x is educational attain-

disintegrates as he becomes childishly dependent on his parents after losing his

A biography of an Afro-American youth from Harlem and an autobiographi-

continues as a remarkable success, Eddie Perry and Richard Rodriguez bore

7The image here of course comes from Erving Goffman.

6. FURTHER OBSERVATIONS WITH POLICY IMPLICATIONS

8 This observation corresponds to Bernheim's interpretation of conformist behavior. See Ber

Dept. of Economics, University of California at Berkeley, Berkeley, CA 94720,

Manuscnipt received August, 1995; final revision received November, 1996.

(7) U1 = [e/(f+ (XO2 -x01))][1/(g - (xll -X02))]

+ [e/(f + (x03 -x01))][1/(g - (x1l -x03))] - ax1 + bx11 +

(8) 8U1/8xll = [e/(f+ (XO2 -x01))][1/(g - (Xll -X02)) ]

+[e/(f+ (X03 -x01))][1/(g- (Xll -X03))2] + [-2axll +b].

(9) U1 = [e/(f+ (XO2 -x01))][1/(g + (xll -X02))]

+[e/(f+ (x03 -x01))][/(g9- (x1l -x03))] - axh2 + bx1l + c.

(10) 8U1/dxll =-[e/(f+ (X02 -Xoi))][1/(g + (Xll -X02))2]

(11) U1 = [e/(f+ (X02 -Xo1))][(1/(g + (xll -X02))]

+ [e/(f + (x03 -x01))][1/(g + (x1l -x03))] -a41 +bx1 +?c.

And in this range we find

(12) dU1/8xll =-[e/(f+ (X02 -XO1))][(1/(g+ (Xll -X02)) ]

-[e/(f+ (x03 -x01))][1/(g+ (xll -x03)) ] + [-2axll +b].

There is no guarantee that 8U1/8xl

You might also like