European Sociological Review VOLUME 20 NUMBER 1 FEBRUARY 2004 1–11 1
European Sociological Review vol. 20 no. 1 © Oxford University Press 2004; all rights reserved
Would Perfect Mobility
be Perfect?
Adam Swift
This paper explores the key normative issues raised by empirical research into social mobility
and meritocracy. Typically, sociologists working in this area are motivated by a concern
with matters of social justice and equality of opportunity, but that concern tends to be
rather vague and diffuse, which makes it difficult to assess the normative relevance
of their findings. Surveying, in an accessible manner, five issues familiar to political philoso-
phers that clarify the significance of sociologists’ results, this paper explains why a regime of
‘perfect mobility’ is not an appropriate benchmark for evaluating the extent to which a
society offers its members social justice or equality of opportunity. Some of the mecha-
nisms that produce an association between the social position of parents and children are
unobjectionable and would exist even in an altogether just society. Sociologists do not endorse
perfect mobility. But neither are they clear about the variety, and normative significance,
of the various mechanisms that tend to generate inequalities in mobility chances.
This paper explores some of the normative issues raised by
empirical research into social mobility and meritocracy.
Sociologists working in this area regard what they do as
relevant to matters of social justice in general, and of equality
of opportunity in particular. They often acknowledge that
their research is motivated, at least in part, by a normative
interest in such matters. But that interest tends to be rather
vague and diffuse. Masters of precision when it comes to
measuring and analysing the empirical phenomena they
study, they are, typically, less sure in their analytical control of
concepts such as ‘equality of opportunity’ or ‘life-chance’.
Moreover, they are sometimes suspicious of attempts by
others to treat normative issues with similar seriousness,
holding – less or more consciously – that such issues are not
amenable to intellectually respectable investigation. Rejecting
that view, this paper continues my attempt to build bridges
between the technically complex, sophisticated, and rigorous
work done by social scientists on the one hand, and the
technically complex, sophisticated, and rigorous work done
by political philosophers on the other (Marshall et al., 1997;
Swift and Marshall, 1997; Swift, 2000).
Of course, social justice and equality of opportunity are
not the only things that mobility researchers are interested
in. One can quite coherently study the determinants of
individuals’ well-being, or unpack the mechanisms that
generate an association between the levels of advantage
enjoyed (or suffered) by parents and children, without
regarding one’s work as having any normative significance.
Conceptualizations of stratification may be informed as
much by an explanatory as a normative interest. These
two goals can come apart, and it could very well be the
case that the intellectual tools best suited to understand-
ing class formation and action, or to elucidating the
processes generating patterns of social mobility conceived
as movement up or down a continuous hierarchy, are not
also those most appropriate for investigation of the extent
to which societies offer their members social justice.
Nonetheless, it remains the case that those working on
social mobility are typically motivated by some kind of
interest in the value of equality of opportunity, formulating
their positions in terms that have moral resonance and
taking their findings to be relevant to the assessment of
2 SWIFT
public policy. So it matters that we have a clear under-
standing of what their research does and does not show.
The conception of a society in which people’s destina-
tions are quite independent of their origins often acts as
an implicit benchmark in such research. Whether the
association between origin and destination is measured by
conditional probabilities or odds ratios between social
categories, or by correlations between positions forming a
continuum in stratification space, it is widely held that
low associations indicate greater ‘openness’ or ‘fairness’,
more ‘equality of opportunity’ or ‘social justice’. This way
of thinking can seem to suggest that, in an altogether just
society, origins and destinations would be entirely unre-
lated. The normative ideal is a society in which where a
person ends up is random with respect to where she starts
out. On this view, ‘perfect mobility’ would be perfect.
Of course, no sociologist has been foolish enough to
endorse such an extreme view. It is, typically, the extent
of the statistical association between origins and destina-
tions, not the mere fact that there is one, which is taken
as evidence of injustice or of inequality of opportunity
(e.g. Goldthorpe et al., 1980: 252). The point of this paper
is not to construct and knock down a straw man. Nor is
it to deny that mobility research has given us strong
prima facie evidence of real injustice in the distribution
of opportunities as between those with different social
origins (or at least as between some of them). Its aim,
rather, is to identify and elucidate, with the aid of tech-
niques and arguments familiar to political philosophers,
some of the reasons why complete statistical indepen-
dence is an implausible ideal – reasons which sociologists
recognize, but recognize only in a somewhat hazy way. It
begins by running through five key issues that mobility
researchers might want to keep clearly in mind when
thinking about the normative implications of their research.
It ends by drawing some of these together in order to
explain precisely why perfect mobility would not be perfect.
Issue 1: Why Care About
Unequal Chances of Mobility
Between Positions Rather
Than the Extent to Which
Those Positions Are
Unequal?
I have discussed elsewhere (Swift, 2000) some of the limit-
ations, from a normative perspective, of the influential
‘class analysis’ approach to social mobility developed by
Erikson and Goldthorpe (1992). Some of the classes in
their schema are said to be hierarchically ordered rela-
tive to one another, but others are not. (For discussion
of the significance of non-vertical mobility for social jus-
tice, see Marshall et al., 1997: 219–221.) And even where
they are so ordered, it remains the case that the class cat-
egories are only indirect and ordinal measures of that
which, because it is distributed unequally between them,
allows us to say that occupants of those class positions are
unequal with respect to anything, rather than just different.
This means that we need to be very careful when taking
class mobility rates as evidence relevant to claims about
equality of opportunity. Mobility rates between classes – as
measured by odds ratios – tell us solely and specifically
about the distribution of chances, as between those of
different class origins, of achieving and avoiding particular
class destinations. They tell us nothing about the distribu-
tion of opportunities in any more general sense.
To see why, it may help to invoke a pedantic formula
that can help to clarify our thinking about the distribu-
tion of opportunities:
x is unequal to y with respect to the opportunity to get
(or to become) z
The purpose of the fomula is to focus attention on the
specificity of the claim in question. (See Roemer, 2000
for an alternative way of urging the importance of careful
specification.) Two societies – or the same society at
different times – can manifest identical patterns of mobility
between class positions, yet distribute other kinds of
opportunity in very different ways.
Consider the wholly hypothetical case in Figure 1. For
all we know, the odds ratios measuring class mobility
could be exactly the same in 2000 as in 1900. But it would
be wrong to infer that there had been no change in
inequality of opportunity. Comparing those from different
class origins, all of the following statements are true:
1. Everybody is better off.
2. There are more chances for upward mobility.
3. The gap between the positions that those of different
origins tend to end up in is smaller.
4. The distribution of opportunities to achieve an
absolute level of goods (e.g. level 4) is more equal.
5. The distribution of opportunities to achieve a level
2/3 that of the best-off class is more equal.
6. The distribution of opportunities to buy goods is
more equal.
This last point depends on observing that the rungs of
the ladder, movement between which counts as mobility,
WOULD PERFECT MOBILITY BE PERFECT? 3
are themselves constituted by bundles of opportunities
(such as opportunities to buy goods). Even if there were
no mobility at all between the two classes, still it would
be true that the opportunity to buy goods was more
evenly distributed, as between those born into different
class origins, in 2000 than in 1900. The class structure
may be no more ‘open’ than it was. ‘Social fluidity’ may
be constant. But the opportunity to do some of the
things that matter to people – the resources available to
them – are more equally distributed as between children
of different class origins.
Though this presents the critique as specifically directed
against a ‘class analysis’ approach to mobility, it should
be clear that the issue raised applies more widely. However
we conceive the positions between which individuals
move, those positions are going to have associated with
them, perhaps be constituted by, bundles of opportunities.
Some of those will be mobility opportunities – opportu-
nities for movement to other positions – but many of
them will not. If mobility theorists are interested in the
distribution of opportunities as between those born into
different social origins, they need to recognise that posi-
tions are characterized by differing amounts and kinds
of opportunity. Some of those opportunities are not
opportunities to move between positions but opportu-
nities to do other things. Any overall index of the extent
to which a society offers its members equality of oppor-
tunity should take all kinds of opportunity into account.
Opportunities for movement between positions could
be getting more equal while the opportunities constitut-
ing those positions are getting less so (or the converse).
(For a similar distinction developed by economists see
van de Gaer et al., 2001. For an innovative attempt to
operationalise the kind of index suggested here, using a
novel ‘human capital’ approach to the measurement of
social position, see Gershuny, 2002a, 2002b.)
Social justice is not simply about securing equal
chances of access to unequally rewarded positions. Fair
access is an important part of that story, but it is not all
of it. It matters also that the inequalities in rewards
attached to those positions are themselves justified
(Marshall et al., 1997: 15–17). Once we notice that the
unequal rewards that characterize different positions
can themselves be conceived as opportunities (such as
opportunities to buy goods), it no longer makes sense
to regard conventional mobility research as telling us
all we need to know about the distribution of opportu-
nities as between those born into different social origins.
It may tell us about the distribution of opportunities
for movement between positions, but it is silent on the
distribution of all the other kinds of opportunity that
matter.
0
1
2
3
4
5
6
7
1880 1900 1920 1940 1960 1980 2000 2020
Date
St
an
da
rd
o
f
L
iv
in
g
Working class
Middle class
Key
Figure 1 Class sizes and standards of living: hypothetical changes
4 SWIFT
Issue 2: Statistical Chances
Versus Chances as
Opportunities
But does conventional mobility research tell us even
about the distribution of opportunities for movement
between positions? To make claims about the distribution
of those opportunities involves a speculative inference.
The data used tell us not about the distribution of oppor-
tunities as between those of different origins, but about
the distribution of outcomes. It is true that one cannot
achieve an outcome without having had the opportunity
to achieve that outcome. But the converse does not hold.
One can perfectly well have the opportunity to achieve an
outcome that one does not in fact achieve. Indeed, one can
have the opportunity to ‘achieve’ an outcome that one
makes no effort to attain and the attainment of which one
would not even regard as an ‘achievement’. This means
that, on their own, measures of association between where
people start out and where they end up cannot even tell us
about the distribution of opportunities for mobility.
There is a crucial ambiguity in the term ‘chances’.
Sociologists claim that their research tells us about the
distribution of mobility chances. This is correct if we
understand ‘chances’ as ‘statistical probabilities’. For
example, it is true that, statistically speaking, the chance
of a working-class child ending up in a service-class job
is less than that of a service-class child. But that fact, on
its own, tells us nothing about whether two such children
had the same chance of ending up in such a job, about
whether they both had the opportunity to do so. Here,
of course, a ‘chance’ is something that a person might or
might not take, or perhaps be able to take – because they
don’t want it or, perhaps, because they don’t have the
ability to take it.
From a normative perspective, what matters are chances
as opportunities, not chances as statistical probabilities.
What we care about is not whether people from different
origins have the same statistical chance of ending up in
particular destinations, but whether they have the same
opportunity to do so. It is the opportunity set that counts.
But mobility research tells us nothing about that. So
sociologists should be more cautious than they typically
are when they claim that their research has uncovered
inequalities of opportunity. They may have uncovered
inequalities in the statistical chances of those with different
origins coming to occupy different destinations. But
they trade on implicit assumptions when they infer, and
sometimes imply, that those inequalities are problematic
from a normative perspective.
Sociologists have typically taken their findings as
evidence of a variety of obstacles to class mobility – from
overt discrimination to inequality in access to education
and other resources. But conventional mobility data are
equally consistent with the hypothesis that those of
different origins differ solely in their preferences –
whether for destinations, or the education needed to
achieve them, or both. Unless we assume that all members
of society are equally motivated to seek upward mobility –
with what counts as ‘up’ being given by the sociologist – and
that they are equally motivated to seek the means necessary
for upward mobility, we cannot tell whether there is any
normatively troubling inequality of opportunity, as
opposed to an unequal distribution of (i.e. simply different)
preferences. (On education specifically, see Murphy, 1990.)
Issue 3: Explaining
Preferences
Suppose we did the kind of research needed to discover
whether inequalities in statistical chances of mobility
were indeed due either to differences in preferences for
different destinations, or to differences in preferences for
the kinds of education needed to achieve them. Suppose
it turned out that they were. Society is full of children
who want the same kind of job as their parents or who,
though wanting what they regard as a ‘better’ job, prefer
not to do the schoolwork that would help them get it.
Having read the previous section, we might infer from
this that there was no inequality of opportunity – merely
different people making different choices from the same
opportunity set. We can, after all, explain the differential
destinations ‘achieved’ by those from different origins
without invoking any obstacles to mobility.
But this would be too quick. Most obviously, the mere
fact that people chose their destinations, and got the
destinations they chose, would not be enough to show
that they had the same opportunity set. It matters, for that,
whether they could have achieved the same outcomes if
they had wanted to. Consider a son of working-class
parents who never aspires to anything other than a
working-class job. In fact, he would be happy to get such
a job because that way he would avoid downward mobility
into the underclass (Goldthorpe, 2000: 161–181) and,
because of economic growth, could reasonably expect to
have a better life than his parents had. This explains why
he ends up in a working-class job. But it does nothing to
show that he had any chance of ending up anywhere
else. The same goes for a daughter who does want to
climb the social ladder but does not want to work to pass
WOULD PERFECT MOBILITY BE PERFECT? 5
exams at school. Perhaps she would not have passed
those exams, or at least not achieved the grades requisite
for social ascent, even if she had tried. In that case she
did not really have a chance to climb the ladder, even
though she never came up against the obstacle that
would have prevented her.
Suppose we find that the distribution of outcomes
corresponds exactly to the distribution of choices for
those outcomes. Children of parents in positions that
sociologists characterize as relatively advantaged
(e.g. ‘service class’) tend to choose to pursue such positions
themselves. Children of parents in positions that socio-
logists characterize as less advantaged (e.g. ‘working
class’) tend to choose those. It matters normatively, and
for a proper understanding of the causal mechanisms in
operation, where those choices come from. To see why,
contrast the following two cases:
In case A, the reason why the two kinds of children
make different choices is because they both want to follow
in their parents’ footsteps. As it happens, the playing-
field is level, so that they both face the same opportunity
sets, and all know this to be the case. Both sets of chil-
dren are fully informed about the rewards attaching to a
wide range of jobs, about their probabilities of achieving
them, about the risks they run in pursuing any particular
strategy. It just happens that their desire to do the kind
of job their parents do is their dominant goal. Under one
description, they share the same preferences – to do what
their parents do. Under another, their preferences are very
different – one set wants to be salaried managers or pro-
fessionals, the other to do manual work for a weekly wage.
In case B, the reasons why the two sets of children
make different choices are quite different from those in
case A. They have different resources at their disposal, and
they are competing on a playing-field tilted in favour of
some and against others, so the chances of their achieving
different jobs – and the costs of trying but failing to do
so – are unequal. Moreover, because of biases and
distortions in the way information is available to the two
groups, less advantaged children underestimate their
chances of climbing the ladder, perceiving the playing-
field to be even less level than it actually is. On top of
this, let us suppose, there is also some adaptive preference
formation. Their preferences for different kinds of jobs
adapt to the perceived probability of their getting them
and the costs and benefits judged likely to result from
strategies devoted to their pursuit.
In case A, it is hard to see that there need be anything
normatively problematic about the inequality in chances
of ending up in different kinds of jobs. The pattern of
outcomes does not manifest or result from any injustice
or unfairness. In case B, however, the mere fact that the
children have chosen to pursue strategies that lead to
their pursuing jobs similar to those of their parents hardly
means that their society is above normative criticism. To
the extent that we understand the choices people make as
endogenous to distributions that are themselves unjust
(or perhaps simply unequal), the fact that people are get-
ting what they choose cannot be the end of the story.
How we decide when choices, and especially prefer-
ences, are adaptive is, of course, a difficult question and
one central to the normative evaluation of many kinds
of empirical inequality. Progress in answering it would
be of interest to those troubled by gender inequality no
less than to class analysts (Nussbaum, 2001). I do not
have any answers, but a few distinctions may help to
clarify what is at stake.
First, it matters whether the empirical factors and
probabilities to which choices may adapt are accurate or
not. A society in which less advantaged children accu-
rately perceive that social ascent would be an uphill
struggle is surely worse than one in which they mistakenly
believe it to be so – though of course it matters why they
have the mistaken perception, whether anybody is
responsible for that fact, and so on. And the policy
implications are different. To the extent that misperception
is the problem, the solution is better information. If they
really are struggling against the odds, other measures are
needed. (For a discussion which excellently brings out
the complexity of educational choices, see Gambetta,
1987: 71ff and conclusion.)
Second, it matters whether the causal conditions to
which choices adapt are themselves unjust or merely
unequal. Choices that are normatively troubling, because
inequality-producing, can be endogenous to inequalities
in the distribution of resources that are not themselves
unjust. Suppose it is fair when, facing the same opportunity
set, some parents choose to work hard and earn money
while others choose not to. The resulting inequality
between parents may well be just. But the mere fact of
that inequality – however just in itself – could make
different strategies rational for their children. This is
simply a special case of the familiar problem faced by
those who would seek to insulate equality of opportunity
from justified inequalities of outcome.
Third, in so far as different educational or occupa-
tional choices result from different preferences, it matters
why children differ in their preferences. We should be
clear that, in case B, it is primarily choices that are adaptive,
not preferences. As that case was described, there was
also an element of adaptive preference formation or sour
grapes. Less advantaged children, who would have
6 SWIFT
wanted to pursue more ambitious careers but decide not to
try, come to believe that they do not want those careers
anyway. But it is differences in perceived probabilities, and
assessments of costs and benefits, not differences in prefer-
ences, that do most to explain the two groups’ different
choices. In case A, by contrast, it is precisely preferences
that differ. Indeed, the fact that the case is merely one of
‘different preferences’ is what grounds the suggestion that
the resulting outcomes are normatively unproblematic.
The thought is that all children face the same opportunity
set and simply choose different options within it.
But of course it is relevant why children’s preferences
tend to reflect their parents’ occupation. The case stipulates
that they have good information about a wide range of
occupations, so we cannot accuse their parents of dis-
torting the formation of their preferences by withholding
information. Presumably the tendency results from the
differential effects of intra-familial interaction, differential
exposure to particular kinds of stimuli such as friends of
the family, and so on. But, in that case, it may be thought
unfair that some children get to meet and be inspired by
high-achieving professionals while others, though having
accurate information, never benefit from that kind of
immediate, ambition-enhancing and preference-affecting
interaction. Parents may be doing nothing wrong when
they or their friends act as role models for their children.
Nonetheless, some may find morally problematic the
distributive outcomes that such processes of preference-
formation tend to produce.
Issue 4: Equality of
Opportunity and Luck
Egalitarianism
A fourth issue concerns the way in which the ideal of
equality of opportunity is understood, and has implications
for the reasons we might have to value meritocracy. Both
concepts – ‘equality of opportunity’ and ‘meritocracy’ –
are widely invoked by sociologists. Their core thought is
that a society in which people’s opportunities for social
ascent or descent are equal (i.e. based on merit alone) is
better than one in which irrelevant characteristics – such
as their race, gender or class of origin – influence their
chances. This perspective is, however, called into question
by some political philosophers, who have developed a
more radical conception of equality of opportunity.
Engagement with that alternative conception may make
a difference to the way in which sociologists think about
the value of meritocracy.
This more radical approach – luck egalitarianism –
holds that inequalities are unjust to the extent that they
result from factors beyond people’s control, factors for
which individuals cannot be held responsible. From this
perspective, it is indeed unfair if the social position of
her parents makes a difference to someone’s chances in
life. But it is no less unfair if that difference is made by her
natural ability. Both are ‘morally arbitrary’ external infl-
uences that cannot justify inequalities either in outcomes
or in the opportunities to achieve unequal outcomes.
Luck egalitarians would seek to level the playing-field not
only between rich and poor children but also between
those lucky and unlucky enough to possess, or not to pos-
sess, productive capacities (Cohen, 1989; Dworkin, 2000).
Suppose we define as ‘meritocratic’ an allocation mech-
anism whereby people get better- or worse-rewarded jobs
on the basis of their true ability, undistorted by ascribed
characteristics such as race, gender or class. It is, for luck
egalitarians, unclear why this should be regarded as an
improvement, morally speaking, on an allocation mech-
anism in which people’s social origins influence their
rewards. Sociologists who care about social mobility tend
to value equality of opportunity between children of the
same level of ability born to parents occupying different
positions in the stratification system. Poor children
blessed with natural ability should have the same chances
in life as rich children with the same good fortune. Rich
children who lack ability should suffer downward
mobility – and it is a problem that their parents are well-
placed to protect them from sliding down the ladder.
But the luck egalitarian wants to know why a clever poor
child should end up better placed than a stupid rich one.
Notice that we can formulate this objection in terms
of the pedantic formula proposed above. For example, it
is conventional to care about equality of opportunity as
between x and y, with x = ‘children from working-class
origins’ and y = ‘children from service-class origins’.
Why not care about equality of opportunity between
x = ‘children lacking in natural ability’ and y = ‘children
with lots of natural ability’? Notice also that it matters
how we specify z – what it is that people should have
equal opportunity to achieve. A luck egalitarian does not
have to think that all people – adroit as well as clumsy –
should have equal opportunity to become brain surgeons.
She need only think that they should have equal oppor-
tunity to earn the same income (Roemer, 2000: 29–31).
For luck egalitarians – as for followers of Hayek – the
idea that people deserve to be rewarded in accordance
with their productivity cannot withstand scrutiny (Swift,
2001: 39–48). But there is a second reason why we might
value greater meritocracy: it is more efficient when people
WOULD PERFECT MOBILITY BE PERFECT? 7
get jobs on merit than when allocation mechanisms are
distorted by class factors. Meritocracy now is a means to
the end of economic productivity (Sen, 2000). But luck
egalitarians will be very interested in the distribution of
that product. For example, they may value meritocratic
equality of opportunity only in so far as it tends to
further, perhaps maximally to further, the well-being of
those who, through no fault of their own, are the least
advantaged members of society. At this point, of course,
incentive considerations enter the story. For luck egali-
tarians, the ideal would be for people to do the jobs they
are best able to do without anybody receiving a return to
luck. We would separate the allocative and the distributive
functions of the market. The reason usually given for
why we cannot do this is that the more productive are
not willing to use their abilities optimally without
receiving greater-than-average rewards (Cohen, 1995).
It is conventional to argue that greater equality of con-
dition is necessary if societies are to move towards more
equal mobility chances as between those born into
different classes of origin. This is because of the difficulty
in insulating children’s equality of opportunity from
parent’s inequalities in outcome. One reason offered for
why Sweden does exhibit greater social fluidity than the
other societies with which it has been compared is
precisely that Sweden has taken greater steps to reduce the
gap between the various starting positions (Goldthorpe,
2000: 255–256). In so far as this is conceived as an argu-
ment for reducing the gap – in so far as the thought is
that we should reduce inequality of condition because
doing so promotes equality of opportunity – luck egali-
tarians will think that it gets things the wrong way
round. Greater fluidity between class positions is valuable
only instrumentally, because it is an efficient means of
creating economic product and thereby, if we can get
our distributive mechanisms right, of helping the
unluckiest members of society. It may be that we should
indeed reduce the gap in order to promote efficient
mobility, but, perhaps paradoxically, the reason for
doing so is, ultimately, that this is itself a means to
improve the absolute position of those who are least
advantaged because most unlucky.
Issue 5: The Family and
Unequalizing Mechanisms
Of course, most people reject luck egalitarianism. Many
think that it is unfair if people’s prospects depend on
their class background but that it is not unfair if those
prospects depend – in part – on their natural ability,
even where it is accepted that both are a matter of luck
(for relevant empirical evidence see Marshall et al.,
1999). There are of course interesting further issues here.
How should we deal with effort? Isn’t it the case that a
person’s productive capacities depend both on what
they are born with and how they choose to develop what
they are born with? It is hard to tell exactly what can be
concluded from empirical research into popular beliefs
about justice. Even the most carefully worded questions
on questionnaires leave crucial indeterminacies (Swift,
1999). Still, it seems likely that most people believe, or
think they believe, that some differential reward is due to
the exercise of attributes that people possess or fail to
possess as a matter of brute luck (Miller, 1999: 61–93).
But, quite apart from this fundamental divide over
what philosophers call constitutive luck – kinds of luck
that constitute people as the individuals they are (Hurley,
2002) – it is natural to make further normative discrimi-
nations between the mechanisms that tend to produce
unequal outcomes. Even those who disapprove of private
education, or economic bequest and inheritance, do not
typically also disapprove of bedtime stories or other
kinds of familial interaction that tend to give some lucky
children advantages over less fortunate others. Hard-liners
may seek to neutralize or compensate the unequalizing
effects of these various mechanisms by which relatively
advantaged parents tend to have relatively advantaged
children. It may be that, while it would be wrong to prevent
parents from reading bedtime stories to their children,
or engaging them in intellectually stimulating conversa-
tion at meal times, or introducing them to their high-
achieving friends, we would be justified in doing what
we can to prevent the children who receive such treatment
from deriving unfair advantage over others who do not.
One way to do this would be to organize things so that
everybody got the same level of advantage, irrespective
of how they had been treated by their parents. Another
would be to accept unequally advantaged positions, but
allocate people to them randomly. Of course, in anything
but an egalitarian utopia (or dystopia), we must expect
non-random inequalities of outcome. And in that case it
matters hugely that some mechanisms by which parents
convey advantage to children warrant greater respect,
morally speaking, than others.
Clearly the family is of crucial importance. Political
philosophers and social scientists have long known that
the family is an obstacle to equality of opportunity
(e.g. Rawls, 1971: 301ff; Fishkin, 1983). Sociologists have
now started to unpack the mechanisms by which the
family exerts its unequalizing effects – though they have
not yet approached (and may never be able to approach)
8 SWIFT
the level of detail relevant to normative discussion.
Could we ever, for example, separate out the effect of
informal intra-familial interaction (like bedtime stories or
mealtime conversations) and similar processes likely to be
regarded as permissible by any political philosopher
from other, more controversial mechanisms (such as
informed school choice in a context where choice of
school matters to children’s chances of success)?
But it is not as if political philosophers know what to
say about the family either (but see Valentine and Lipson,
1989, and Munoz-Darde, 1999, for some relevant argu-
ments). There is, as yet, little serious work attempting to
identify which transmission processes are justified,
which objectionable, and which morally required. What
we need, on the normative side, is a theory of the family
that tells us what parents are and are not justified in
doing for their children, and what kinds of intergenera-
tional transmission within the family we, as citizens,
would be justified in permitting and preventing. We
need a theory that tells us why we would do well to abolish
private schools but do badly to ban bedtime stories, and
why this would remain true even if the difference
between getting and not getting bedtime stories was
more influential for a child’s prospects than the difference
between going to a private or a state school. There are
big and difficult issues here, concerning the scope of
legitimate – and perhaps required – parental partiality
and how it meshes with any concern for equality of
opportunity (Swift, forthcoming).
Would Perfect Mobility
be Perfect?
It can be tempting to think that social justice would
require that children from all social origins would have
equal chances of achieving and avoiding all destina-
tions. This would be what sociologists call ‘perfect
mobility’ – a mobility regime in which where you start
off has no influence on where you end up. Though none
endorses such a regime, they do tend unproblematically
to assume that low associations between the positions
occupied by parents and children are morally better
than high ones. Let me begin to conclude by drawing
together some of the strands of argument sketched out
above, each of which, in a different way, explains why
some inequality in statistical mobility chances should be
expected even in the ideal society and clarifies the nor-
mative significance of the various mechanisms that
might generate such inequality.
A. Asymmetry Between Downward
and Upward Mobility
Influential rational choice models explaining differential
educational outcomes between those of different origins
posit that people are more concerned to avoid downward
mobility than they are to achieve upward mobility
(Goldthorpe, 2000: 161–205). Suppose that people do
indeed regard downward movement not merely as bad
but as worse than upward movement would be good.
Every time a service-class child skidded and her place in
the top flight was taken by a working-class child, there
would be net loss of welfare. In that case, and other
things equal, the way to maximize social welfare would
be to have no mobility. Everybody should stay on the
same rung of the ladder as that occupied by their parents
(Harding et al., forthcoming: 23–25).
The point discussed as Issue 1 strengthens this obser-
vation. Suppose that people’s reference groups are such
that what they care about is being better off than their
parents. Though noticing that others are even better off
than they are, they are not too bothered about that fact.
What matters – to them – is that they are going to be
better off as adults than their parents, better off than
they were when they themselves were children. Upward
mobility is not needed for them to achieve that goal.
Economic growth will do it for them. Immobility in class
terms, or in terms of percentile in the distribution, or in
terms of rung on the ladder, however that ladder is
conceived, may not be immobility in terms of standard
of living or quality of life. One can have more or better
opportunities than one’s parents without having any
opportunity to move up in terms of social stratification.
The ladder is really an escalator.
Normatively speaking, this consideration is far from
compelling. Even if it were true that people’s welfare
functions worked this way, aggregate social welfare is not
all that matters. Justice matters more. So if giving a work-
ing-class child a fair chance of reaching the top means
worsening the service-class child’s chance of staying there,
we should not worry about that – even if the latter is more
distressed than the former is pleased. It is of course con-
sistent with this normative position that such welfare
functions are crucial for understanding the choices people
make, and the mobility patterns such choices generate.
B. Non-suspect Immobility-producing
Preferences
Issue 3 concerned the possibility that people’s choices,
and perhaps also their preferences, may be endogenous,
WOULD PERFECT MOBILITY BE PERFECT? 9
causally explained by the inequality they tend to reproduce.
However seriously we take that possibility, some tendency
towards immobility would surely result from those
about which there were no grounds for suspicion. Even if
we assume that sociologists correctly rank occupations –
or groups of occupations – in terms of the objective level
of well-being typically enjoyed by those who occupy
them, we still have to allow that some people may, non-
adaptively, prefer ‘worse’ positions to ‘better’ ones. It is
hard to believe that all the sons of farmers who go into
farming do so only because they do not rate their
chances of doing anything else. It is not necessarily irra-
tional to prefer manual work, and the workplace culture
that goes with it, to a job pushing bits of paper round an
office. Some people may prefer to stay in the region
where they grew up, close to family and friends, forsaking
the higher ‘objective’ rewards that would accrue to them
if they were willing to move. Familiar processes of accul-
turation and socialization will surely lead to some
immobility at the level of occupations or types of occu-
pations. Some children want to be like their parents and
some parents want their children to be like them. It
would be at least controversial to regard the contribution
of such mechanisms to inequalities in mobility rates as
indicating any failure of social justice. (One area where
this has policy implications concerns the mobility of
members of ethnic groups. If cultural or religious differ-
ences tend to produce inequalities in statistical mobility
chances – because of different preferences for different
kinds of work – it is clearly problematic to infer that
there is anything morally objectionable going on.)
C. Legitimate Parental Partiality
As discussed as Issue 5, normative reflection on the family
suggests that any society that permits unequal outcomes
should also permit parents to do some things for or with
their children likely to influence their children’s outcome
position in the distribution of advantage. Suppose a society
blocked all transmission mechanisms that blatantly offend
against egalitarian principles and do not fall within the
scope of legitimate parental partiality. Suppose this meant
that there was no private education, no economic bequest
or inheritance. Suppose the society tried hard to compen-
sate unlucky children for their parents’ comparative
inability to provide valuable investment. So there was free
nursery education for all, considerable resources devoted to
helping parents learn how to help their kids, and the like.
Still, familial interactions of the kind that even egalitarians
want to permit as legitimate would tend to produce
unequal outcomes. And it is relatively advantaged parents
who will tend to transmit, through those interactions,
attributes likely to make their children relatively advan-
taged – whether those attributes be intellectual curiosity,
nice social skills, a sense of discipline, or the right stuff genet-
ically speaking. (On the variety of mechanisms explaining
the intergenerational persistence of inequality, and an
attempt quantitatively to decompose that persistence into
various components, see Bowles and Gintis, 2002.)
This is not a conceptual claim. We could, just about,
imagine a world in which it was only badly-off parents
who had valuable characteristics to impart. (For example,
in terms of motivation, in the world as it is, some children
of disadvantaged parents are particularly driven to
succeed.) But, overall, the world we live in is not like that,
and for obvious reasons. It is because they possess certain
characteristics, and others do not, that some parents are
better off than others. Those who possess such charac-
teristics are particularly well placed to convey them –
whether intentionally or unintentionally – to their children.
This kind of legitimate parental partiality would better
be regarded as part of social justice than as a constraint
on its realization. A society that denied its members the
freedom to stimulate their children through engaging
them in conversation at the table would be unjust, even
if it had greater equality of opportunity as between those
born to parents with different levels of cultural capital.
D. Incentive Considerations
Any non-utopian theory will have to take incentive effects
into account. Suppose that nothing deeply moral to do
with individual freedom or human rights or legitimate
parental partiality should prevent us from banning the
bequest and inheritance of property. There is no justice
consideration that requires us to permit that kind of
transmission. It by no means follows that we should
institute a 100% inheritance tax. If parents were prevented
from bequeathing resources to their children, they would
lose a powerful incentive to produce. If we care about
productivity – even if only because we want to use the
product to maximize the advantage of the worst off –
then we have reason to permit some intergenerational
transmission, despite its unequalizing impact and even if
we regard that transmission as fundamentally unjust.
This is still a normative consideration. We have moral
reason to permit it – that reason is the moral urgency of
helping the worst off. Not permitting it will worsen their
position. But this is the kind of moral reason we act on
when we decide to pay the kidnapper’s ransom. It may
be the morally right thing to do but it is not morally
justified all the way down (Cohen, 1995; Sen, 2000).
10 SWIFT
Where exactly to draw the line between C-type and
D-type reasons to permit intergenerational transmission
is a difficult issue. One person’s legitimate partiality will
be another’s selfish demanding of an incentive. That
does not mean, as social scientists tend to assume, that
the answer is purely subjective, with no right answer in
principle being available. It means just what it says:
people disagree. (See Swift, 1999, for more discussion of
social scientists’ misplaced mistrust of political philoso-
phy.) But the conceptual distinction should be clear
enough. C says that parents are acting quite properly – and
we would be acting wrongly if we prevented them – when
they engage in certain kind of advantage-transmitting
activity. D says that, even when they are not acting
properly, and we would be justified in preventing them
if we could do so without inducing negative conse-
quences, those consequences give us moral reason to let
them go ahead. Society would be better, morally speaking,
if people did not demand incentive payments. But it
would be worse, again morally speaking, if we did not
pay them the incentives they demand.
Conclusion
No sociologist thinks that social justice requires perfect
mobility. All are aware that such a view would be crude
and simplistic. On the one hand, equality of opportunity
does not seem to demand the complete absence of any
association between origins and destinations. It requires
that only on the implausible assumption that ability and
motivation of the kind that many think properly deter-
mine people’s destinations are randomly distributed as
between those born into different origins. On the other
hand, equality of opportunity is not the only component
of social justice. There is a tension between equality of
opportunity on the one hand, and respect for the family
on the other, which few would resolve simply by abolish-
ing the latter. While there are families, and those families
are differently located in the distribution of advantage,
children born to different parents are indeed likely to
enjoy or suffer unequal prospects. So equality of oppor-
tunity does not require perfect mobility, and social justice
does not require complete equality of opportunity.
Few sociologists, however, have moved beyond these
rather general observations to explore in greater detail
the normative significance of the various mechanisms
that do, in fact, combine to produce the mobility
regimes their empirical research describes. In a sense, of
course, that is not their task, and this paper is not primarily
intended as a critique of existing sociological practice.
Sometimes, to be sure, sociologists report their findings
in ways likely to mislead – for example, when they infer
conclusions about ‘inequality of opportunity’ per se, when
their evidence supports only claims about the distribution
of statistical chances to achieve and avoid destinations
conceived in particular, and sometimes rather limited,
ways. So it is true that one aim of this paper is to encour-
age an equivalent precision in the formulation of claims
about normatively loaded ideas as is currently achieved
in the specification of their statistical models. Its main
purpose, however, is more constructive – to clarify and
make available to sociologists, in accessible form, some
of the ideas that often, albeit implicitly, inform their
thinking about the significance of their research.
Acknowledgements
I am grateful to the British Academy for the Research
Readership during which this paper was written, and to
Nuffield College for the ideal environment in which to
write it. Thanks are due also, for improving suggestions,
to Harry Brighouse, Diego Gambetta, Jay Gershuny,
audiences at Nuffield College and RC 28, and an anony-
mous referee.
References
Bowles, S. and Gintis, H. (2002). The Inheritance of
Inequality, Journal of Economic Perspectives, 16, 3–30.
Cohen, G. A. (1989). On the Currency of Egalitarian
Justice, Ethics, 99, 906–944.
Cohen, G. A. (1995). Incentives, Inequality and Com-
munity. In Darwall, S. (Ed.), Equal Freedom, Ann
Arbor, MI: University of Michigan Press, pp. 331–397.
Dworkin, R. (2000). Sovereign Virtue. Cambridge, MA:
Harvard University Press.
Erikson, R. and Goldthorpe, J. H. (1992). The Constant
Flux. Oxford: Oxford University Press.
Fishkin, J. (1983). Justice, Equality of Opportunity and
the Family. New Haven, CT: Yale University Press.
Gambetta, D. (1987). Were They Pushed or Did They
Jump?: Individual Decision Mechanisms in Education.
Cambridge: Cambridge University Press.
Gershuny, J. (2002a). Beating the Odds (1): Intergenera-
tional Social Mobility from a Human Capital Perspec-
tive. University of Essex: Institute for Social and
Economic Research.
Gershuny, J. (2002b). Beating the Odds (2): A New Index
of Intergenerational Mobility. University of Essex:
Institute for Social and Economic Research.
WOULD PERFECT MOBILITY BE PERFECT? 11
Goldthorpe, J. H., Llewellyn, C. and Payne, C. (1980).
Social Mobility and Class Structure in Modern Britain.
Oxford: Clarendon Press.
Goldthorpe, J. H. (2000). On Sociology. Oxford: Oxford
University Press.
Harding, D., Jencks, C., Lopoo, L. and Mayer, S. (2004).
Trends in Intergenerational Economic Mobility:
Theories and Estimates for the US since 1960. In
Bowles, S., Gintis, H. and Osborne, M. (Eds),
Unequal Chances: Family Background and Economic
Success. Princeton, NJ: Princeton University Press.
In press.
Hurley, S. (2002). Luck, Responsibility and the “Natural
Lottery”, Journal of Political Philosophy, 10, 79–94.
Marshall, G., Swift, A. and Roberts, S. (1997). Against the
Odds? Social Class and Social Justice in Industrial
Societies. Oxford: Oxford University Press.
Marshall, G., Swift, A., Routh, D. and Burgoyne, C.
(1999). What Is and What Ought to Be: Popular
Beliefs about Distributive Justice in Thirteen Coun-
tries, European Sociological Review, 15, 349–367.
Miller, D. (1999). Principles of Social Justice. Cambridge,
MA: Harvard University Press.
Munoz-Darde, V. (1999). Is the Family to be Abolished
Then? Proceedings of the Aristotelian Society, 99,
37–56.
Nussbaum, M. (2001). Adaptive Preferences, Women’s
Options, Economics and Philosophy, 17, 67–88.
Murphy, J. (1990). A Most Respectable Prejudice:
Inequality in Educational Research and Policy, British
Journal of Sociology, 41, 29–54.
Rawls, J. (1971). A Theory of Justice. Cambridge, MA:
Harvard University Press.
Roemer, J. (2000). Equality of Opportunity. In Arrow, K.,
Bowles, S. and Durlauf, S. (Eds), Meritocracy and
Economic Inequality. Princeton, NJ: Princeton
University Press.
Sen, A. (2000). Merit and Justice. In Arrow, K., Bowles, S.
and Durlauf, S. (Eds), Meritocracy and Economic
Inequality. Princeton, NJ: Princeton University Press.
Swift, A. (1999). Popular Opinion and Political Philosophy:
The Relation between Social-Scientific and Philo-
sophical Analyses of Distributive Justice, Ethical
Theory and Moral Practice, 2, 337–363.
Swift, A. (2000). Class Analysis from a Normative
Perspective, British Journal of Sociology, 51, 663–679.
Swift, A. (2001). Political Philosophy: A Beginners’ Guide
for Students and Politicians. Oxford: Polity Press.
Swift, A. and Marshall, G. (1997). Meritocratic Equality
of Opportunity: Economic Efficiency, Social Justice,
or Both? Policy Studies, 18, 35–48.
Swift, A. (2004). Justice, Luck and the Family: Norma-
tive Aspects of the Intergenerational Transmission
of Economic Status. In Bowles, S., Gintis, H. and
Osborne, M. (Eds), Unequal Chances: Family Back-
ground and Economic Success. Princeton, NJ: Princeton
University Press. In press.
Valentine, P. and Lipson, M. (1989). Equal Opportunity
and the Family, Public Affairs Quarterly, 3, 29–47.
Van de Gaer, D., Schokkaert, E. and Martinez, M.
(2001). Three Meanings of Intergenerational Mobility,
Economica, 68, 519–537
Author’s Address
Adam Swift, Balliol College, Oxford OX1 3BJ, UK. Email:
adam.swift@balliol.ox.ac.uk
Manuscript received: September 2002.
