Inequality, Economic Growth and Social Mobility

Author(s): Richard Breen

Reviewed work(s):
Source: The British Journal of Sociology, Vol. 48, No. 3 (Sep., 1997), pp. 429-449
Published by: Wiley-Blackwell on behalf of The London School of Economics and Political Science
Stable URL: http://www.jstor.org/stable/591139 .
Accessed: 31/10/2012 04:59

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
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 support@jstor.org.

Wiley-Blackwell and The London School of Economics and Political Science are collaborating with JSTOR to
digitize, preserve and extend access to The British Journal of Sociology.

http://www.jstor.org
Richard Breen*

Inequality,economic growthand social mobilityl

ABSTRACT

This paperdevelopsa model of intergenerationalmobilityand intragenerational

inequalitythat allowsus to explore the relationshipbetween economic growth
and socialmobility.The model is used to analysethe neo-liberaltheoryof stratifi-
cation and to assessthe consequencesof some of the criticismsthat have been
made of it. In particular,the intergenerationaltransmissionof wealthand privi-
lege, and the existence of ethnic, gender and other formsof ascriptivedisadvan-
tage,reduceeconomicefficiency,althoughtheydo not alwaysdiminishthe extent
of social mobility. Furthermore,excessive intragenerationalinequality may
inhibit,ratherthanencourage,economic growth.We showthatthere is no neces-
sarylink betweenratesof socialmobilityand levelsof economic growth.This,we
suggest, providesan explanationof why rates of social mobilityshow very little
cross-nationalvariationand display no very evident trend over time towards
greatersocietalopenness.

KEYWORDS:Stratificationtheories;intergenerationalsocialmobility;
intragenerationalinequality;economic growth

INTRODUCTION

This paper addressesthe question of the relationshipbetween inequality,

economic growth and social mobility. This question is of interest, we
suggest,not leastbecause,whileit is widelyassumedthathigh levelsof social
mobilityare necessaryto secure economic growth ('makingthe best use of
the nation'sresources'),it is also assumedthat high levelsof inequalitywill
tend to restrict rates of social mobility. Yet inequality in occupational
rewardsis thought to provide a necessaryincentive structurewhich pro-
motes growth. Thus there is a paradox:both inequalityand mobilityare
good for growth,yet one militates against the other. In our analyseswe
investigatethese assumptionsand conclude that they are likelyto be incor-
rect. In particular,our resultsmaybe added to the small,but growing,body
of literaturewhich suggests that a high level of inequality,rather than
improvingeconomic performance,mayhave negativeeffects (Perssonand
Tabellini 1994). We also find that the relationship between economic
growthand the amount of social mobilityin a societyis largelyindetermi-
nate. We suggest that this may account for why, both cross-nationallyand
through time, changes in patternsof social fluiditydo not displayanyvery

Bril. llo.48
volllme
iolon
Jnl.ol.4;o I.sslle 11o.
3 September 1997 ISSN ()()()7-131.5 O Lotldotl School of Ec()llomics 1997
430 RichardBreen

evidenttrend towardsgreateropennessbut, rather,are betterdescribedas,

in the wordsof Eriksonand Goldthorpe (1992), a 'constantflux'.
In sociology the relationshipbetween inequalityand economic growth
has perhapsbeen most extensivelyaddressedin the neo-liberaltradition.
Although approaches in this tradition have been widely criticized they
neverthelesscontinue to inform both debate (academicand popular) and
policy:accordinglywe take this model as our point of departure.We begin
with a mathematicalformalizationof the neo-liberalaccount of the wayin
which certain elements of economy and society interrelate.Our purpose
in doing this is to investigatethe corollariesof the theory and of elabora-
tions that seek to take into consideration some critiques of its validity.
When we make changes to the model so as to incorporatethese criticisms
we find that manyof the resultsthat flow from the originalmodel are very
much altered. In particularwe show that the outcomes are often con-
siderably less benign and unproblematic than proponents of the neo-
liberalapproachallow.

INEQUALITIES
ASINCENTIVES

The idea thatinequalitiesof rewardsprovidea necessarystructureof incen-

tives is both long standing and enjoys a great deal of contemporarycur-
rency.Thiswasnowheremore evident than in Britainduringthe Thatcher
years,when, for example, reductionsin ratesof personalincome tax, par-
ticularlyfor high earners,werejustified on the basisof the need to provide
incentivesto retain and attractthe most talented personnel.2The idea is
also pervasivein the social sciences, especially,of course, in economics,
where, in the relevantcontexts, individualagents are assumed to seek to
maximizetheir utility,which, in practice,is often a function of monetary
rewards.In sociology the idea is prevalent in neo-liberal treatmentsof
stratification.In some cases this is explicit, as in Davisand Moore (1945)
and other functionalistaccounts (Davis 1949; Parsons 1957), while else-
where it provides the implicit underpinning for macro-sociological
theories. A good example of the latter is the so-called 'liberal theory of
industrialism'(Kerr et al. 1960; Parsons 1964). This posits that, among
other things, economic developmentleads countriesto become ever more
meritocraticin the sense that positionsin societywill increasinglycome to
be filled on the basis of achievement rather than ascription.This, it is
argued,is a functionalnecessityof industrialsociety:the acquisitionbyindi-
vidualsof positionson the basisof anythingother than merit or abilitywill
be sul}optimal both for individual firms competing among themselves
within a countryand for that country'scompetitiveposition vis-a-vis other
nations.Furthermore,as the productionof goods and servicescomes to rely
increasinglyon the applicationof scientific and technical knowledge,so
greaterimportancewillbe attachedto educationalqualifications.Thusedu-
cational and other credentials will play an ever larger part in securing
Inequality,economicgrowthand social mobility 431

positions while ascriptivecharacteristics- such as race, gender and class

background- will decline in importance.
The liberaltheoryof industrialismis couched at the macrolevel and con-
cerns how economic and social structures evolve during development.
Competitionboth within and between nations leads to a greateremphasis
on ensuring that the most 'able' people come to occupy the 'correct'pos-
itions in society. But, to have any claim to plausibility,the theory must be
underpinnedby some account of how this arisesas the resultof 'the action
and interactionof individuals'(Elster1989: 13). The nebliberal theory of
stratificationprovidesone such account,accordingto whichpeople are per-
suaded or encouraged to behave in the appropriatewayby incentives.
The liberal theory of industrialismhas received very little empirical
support.3Researchin socialmobilityhas shownthat,farfrom ratesof social
fluidityincreasingover time (as one might expect given greateremphasis
on achievementat the expense of ascription),no such trend is discernible
(Erikson and Goldthorpe 1992).4 And similarly,while one might have
anticipatedsubstantialcross-nationalvariationin mobilityrates (and par-
ticularlyin ratesof social fluidity)accordingto the degree to which differ-
ent societies had evolved along the path prescribedby the liberal theory,
this is equally hard to discern (Lipset and Bendix 1959; Erikson and
Goldthorpe 1985; 1992). Rather, the sbealled FJH thesis (Featherman,
Jones and Hauser 1975) has largelybeen confirmed.
A basicsimilarityin patternsof social fluidity. . . will be found acrossall
nationswithmarketeconomies and nuclearfamilysystemswhereno sus-
tained attemptshave been made to use the power of the modern state
apparatusin order to modifythe processesor the outcomes of the prb
cesses through which class inequalities are intergenerationallyreprb
duced. (Eriksonand Goldthorpe1987:162)
But although cross-nationaland temporalinvariancein social fluidityhas
been well documented, it has proved difficultto explain, except in rather
general terms.Here recourseis made to the abilityof those in positionsof
power to maintain their privilegeseven in the face of economic changes
and policyinnovations(such as free education). In this paper,however,we
argue that this invariancein mobilityrates can be explained quite simply.
We suggestthat,contraryto earlierbelief, and even within the termsof the
liberaltheory itself, economic growthand the level of social mobilitybear
no necessaryrelationshipto each other. But to explain this resultwe need
to examine more closelythe nebliberal micrblevel theory of stratification
which,we havesuggested,underpinsthe liberaltheoryof industrialism.To
this end we return to the seminalworkof Davisand Moore (1945).

DAVISAND MOORE AND THEIR CRITICS

Davisand Moore (1945) present a functionalisttheory to explain 'the uni-

versalpresence of stratification'(op. cit.: 242).
432 RichardBreen

Social inequality is thus an unconsciously evolved device by which

societies insure that the most importantpositions are conscientiously
filled by the most qualifiedpersons. (op. cit.: 243)
The patternof rewardinequalitiesis in turn governedby twofactors:these
are the relativefunctionalimportanceof positionsand the level of training
or talent required.Thus, a position requiringrelativelylow levels of train-
ing or talentwill be less well rewardedthan one of equalfunctionalimport-
ance that demandshigh levels of one or both of these. This is because the
supplyof applicantsfor positionswill be inverselyrelated to the position's
requirementsin termsof skillsand abilities.Thus the stratificationsystem
evolves in response to the functional importanceof positions and to the
supplyof people qualifiedto fill them.
This theory has been the subject of several important critiques. For
example, Davisand Moore themselves(op. cit.: 244) noted the difficultyof
establishingthe functionalimportanceof positionsin anygivensociety.But
the most telling criticismshave focused on three issues. First,it is argued
that their treatmentof the distributionof rewardsin society ignores the
'powerelement' in stratification.For example
The unequal rewardsin wealth and prestige 'attachedto' the positions
also give theirincumbentsgreateropportunitiesto influence the general
distributionof rewardsin society and to protect or augment their own
privileges.(Wrong1958:774)
Second, given that unequal rewardsare necessary,the theoryis silent as to
the magnitude of such rewards.Nothing rules out the possibilitythat
certain positions are over-rewardedrelative to what would be needed in
order to ensure that a sufficient number of suitablyskilled individuals
would be willing to enter them (Tumin 1953). And third, critics have
pointed to the role of ascriptionand the intergenerationaltransmissionof
rewards,as factorsthatwillinterferewiththe efficientallocationof talented
persons to appropriatepositions (see also Davis1949).
Notwithstanding these and other criticisms, the neo-liberal model
remainsinfluential.In this paper we subjectit to a mathematicalanalysis,
the aim of which is to bring to light the consequences that derive from it
and to show how these change once the model is extended to incorporate
the kinds of criticismsalreadyoutlined. In order to do this the model is
dynamic:it investigatesnot simplythe shortrun or cross-sectionaloutcomes
in terms of economic growthand social mobilitybut also the longer term
(acrossgenerations)resultsassociatedwithparticularaccountsof a society's
stratificationsystem.5Accordingly,we begin with a model of the neo-liberal
theory. Having investigatedits consequences we then extend it. Firstwe
include the possibleeffects of the intergenerationaltransmissionof advan-
tage:inheritancefor short. Second we look at whathappenswhen we allow
for the existence of an incentivestructurethat leads individualsto seek to
occupy positions in a manner that does not maximize what we term
Inequality,economicgrowthand social mobility 433

'economic efficiency'.This typicallyariseswhen groups acquire sufficient

power to shape the allocation of rewardsto their own and their family's
advantage.Lastly,we examine the resultsthat arisewhen ascriptionassoci-
ated with ethnic group and gender are taken into account. In the paper's
concluding sections we draw out some consequences of our results for
policy and for our understandingof the link between social mobilityand
economic development.

FORMALIZINGTHE NEO-LIBERALTHEORY

We begin by assumingthat individualsare endowed with a qualitycalled

'ability',whichis randomlyand normally(mean zero and unit variance)dis-
tributed over the members of each generation. We further assume that
there existsa distributionof positionsand that the aggregateoutput of the
economy depends upon, interalia, the match between the abilityrequire-
ments of these positionsand the abilityof their incumbents.The closer the
match the greaterthe output;hence, output is higher the strongerthe cor-
relation between the distributionsof abilityand position.6In the spirit of
the neo-liberal treatment of stratification,differential rewardsplay the
centralrole of persuadingthose with the highest level of abilityto take up
those positions in which it is most required.We refer to the correlation
between abilityand position as 'economic efficiency'.7
Note that this formulationof the rank ordering of positions in society
overcomes certain common criticismsof Davisand Moore - namely that
functionalimportancecannot be assessedon the basisof empiricalobser-
vationsconcerning, say, the prestige attached to occupations,and that in
any society there will be conflicts as to the importanceto be attached to
variousoccupations.In our formulationall that is required is that there
should in theorybe some wayof orderingpositionsthatwill make optimum
use of the abilitydistributionfor the particulargoal of maximizingecon-
omic growth. The assumptionof one dimension of ability is innocuous
here, since to allow for severaldimensions would complicate the mathe-
maticsbut not alter the conclusions.
Given the existence of this ordering of positions, by what mechanisms
might individualscome to occupy specific positions?Here there are two
issues. First,accordingto neo-liberaltheories, individualsmust be offered
incentives.Hence each position must have a set of rewards(economic and
non-economic) attached to it and the ordering of rewardsshould match
the ordering of positions.Frequently,of course, this will not be so. As the
criticsof Davisand Moore pointed out, the rewarddistributionin society
cannot be assumed so unproblematic.Even leaving aside such compli-
cationsfor the moment, an importantissueis the magnitudeof these incen-
tivesor the degree of rewardinequalitythatis needed to persuadethe most
able to fill the highest rankingpositions.This is the issue we will focus on
initially.
434 RichardBreen

The second issue relates to the way in which, in the labour market,
positions are allocated. Since abilityis unobservablesome proxy measure
must be used. Most commonlysome forms of educationalor other certifi-
cationplaythis role (at leastin the initialallocationof young people to pos-
itions). Ideally these resources (as we shall call them) for acquiring a
position should directlyreflect ability,but in practicethis is unlikelyto be
the case. While abilityinfluences the resourcesthat individualsbring with
them into the labourmarket,class,gender, ethnic group membershipand
other ascribeddifferencescontinue to be important.

THE SIMPLECASE

We begin by writinga simple model in which the resourcesfor acquiringa

position depend only on abilityand the rewarddistributionmatches the
rank ordering of positions as defined earlier. The model has three equa-
tions

Sit = Ait (1)

Cs 13 (2)

Rit = Pit

Since we will later want to look at the relationshipbetween resources,S,

position, P and rewards,R, across successivegenerationswe use the suS
scriptt to index generations.For simplicitywe assumethat the numbersin
each generationare constantand thusa 'family'or 'dynasty'can be thought
of as a sequence of individualslinked acrossgenerations.We use i to indi-
cate families.Thus in this firstversionof the model, resources,Sit,are given
by ability,At, which is assumed to be randomlydistributed.The position,
P, occupied by the ithfamilyin the tthgenerationdepends upon resources
(standardizedto have mean zero and unit variance)and a random com-
ponent, c, which is a drawfrom a standardnormaldistribution.Finally,in
equation (3), each position is linked to a particularlevel of rewardsvia the
parameter ,B (>O) which determines the degree of reward inequality
between positions.8,Bis thus a measure of rewardinequality.Note that it
enters into equation (2) where the larger is ,Bthe smalleris the random
element in allocatingindividualsto positions. This reflects the impact of
incentives:the stronger the incentives to occupy a position (that is, the
greater the inequalityin rewardsassociatedwith different positions) the
closer will be the match between resourcesand positions.9
We can use this model to calculatetwo statistics.The firstof these is the
Inequalitw,economicgrowthand social mobilitw 435

correlationbetween abilityand position, p(A,P), which we label <1). This is

a measureof economic efficiencyin the sense in whichwe earlierused the
term. Economic efficiencyis by definition maximizedif this correlationis
perfect. The second statisticwe label H,defined as one minus the corre-
lation between positions in successivegenerations (that is 1 - P(Pt,Ptl)).
This capturesinter-generationalmobilitysince it measures the extent to
which the position occupied by a familyin one generation is related to its
position in the next. 10
In the simple form of the model given above there is no relationship
between successivegenerationsbecausewe assumethat abilityis randomly
distributedwithineach generation.This absence of heritabilitymeans that
there is complete social mobilitywith H= 1. The size of the correlation
between abilityand position, which defines economic efficiency,depends
on the magnitudeof the randomcomponentin equation (2), namelyc/,B.ll
This becomes smaller, and the correlation becomes larger, as ,B itself
increasesin size. So, simplymaximizingrewardinequalitywill cause econ-
omic efficiencyto increase.
This is a model of societythat might be said to encapsulatea liberal,free-
marketideology. Positionsare allocated on the basis of abilityalone and
while rewarddifferencescan be verygreat indeed they can be justified on
the grounds that they serve to increase economic efficiency and do not
obstructsocialmobility.l2To the extent thatsocialmobilityis less than com-
plete this must be due to the inheritanceof ability.One means of allowing
for this in our model is to replace the assumptionthat abilityis randomly
distributedwith

Ait=zAit-l+(it (

which expresses ability in one generation as a function of that in the

previousgeneration plus a random term. In this case the degree of social
mobilityis inverselyrelated to the strengthoffy, the parametermeasuring
the extent of inheritance of ability.Again, the argument that deviations
from perfect mobilityare due, to some importantdegree at least, to ability
differencesbetween familieslocated in differentpartsof the occupational
distributionor the class structureis often made - Saunders (1994) and
Herrnstein and Murray(1994) being two recent examples of this long
standingtradition.However,when we incorporateinto the model some of
the criticismsof the neo-liberaltheories to whichwe earlierreferred,these
resultsno longer hold.

INTERGENERATIONALTRANSMISSIONOF RESOURCES

As a firststep we incorporatethe intergenerationaltransmissionof rewards

as a factorin shapingthe level of resourcesindividualsbring with them to
the labour market.Once again revertingto our assumptionthat abilityis
436 RichardBreen

randomlydistributed(ratherthan inherited) we can replace equation (1)

with

Sit- (xAjt+ (1 - (x) Rit 1 (la)

This expressesthe resourcespossessedby each individualas a function of

endowments comprisingparental rewardsRjtl and their own ability,Ajt.
The relativeimportanceof each depends on the value of the parameter(x
which can varybetween zero and one.
Making this modest change has some far-reachingconsequences (the
following results are proved in the appendix). First,the extent of social
mobilityis now given by

0 = 1 - (1 -(X) fi (5a)

Whereasin the simplermodel (under the assumptionof non-heritabilityof

ability)mobilitywasalwaysequal to one, now it depends on both (xand ,B.
The greateris the contributionof inherited rewardsto resources(in equa-
tion (la)) relativeto the contributionof ability,the less mobilitywill there
be. This is intuitivelysensible: the greater the contribution of the non-
inherited factor (ability)in determiningresourcesand thus position, the
weakerthe inter-generationalrelationshipbetween positions.But perhaps
more strikingis that, whereasin the simpler model, mobilityand inequal-
ity were unrelated, now the greater is inequalitythe less mobilitywill be
observed.l3 For any given balance of ability and inherited rewardsin
shaping resources,the greater the rewardinequalitythe more immobile
will a societybecome, simplybecausethe variationin inheritedrewardswill
be large relativeto the variationin ability.
Second, economic efficiencywill now be given by
+= (X (5b)
(Js(Jp

Here the impactof increasing(xon efficiencyis the same as on mobility-

it increasesit. Given that efficiencyis defined as the correlationbetween
abilityand position and (xis the parameterthat links abilityand resources
this result is hardlysurprising.However,what is perhapssurprisingis the
effect on efficiencyof increasinginequality.Whereasin the simplermodel
more inequality meant more efficiency, this is now only partiallytrue.
Althoughefficiencyincreasesas inequalitygrowsfrom a relativelylow level,
continued increasesin inequalitywill cause efficiencyto decline.The point
at which this occursis
l/4

*= 1 +(1_a)2 (6)
Inequality,economicgrowthand social mobility 437

,B*is the value of the inequality parameter that maximizes economic

efficiency.Any value of ,Bgreater or less than this representsa level of
inequalitythat is sub-optimalfor economic efficiency.
This result arises because, although inequality contributes to greater
efficiency by providing incentives, it also militates against efficiency by
increasingthe degree of inequalityin inherited rewards,so that the indi-
vidual'sresources,S, come to depend more heavilyon inheritedrewardsat
the expense of ability.As inequalityexceeds ,B*this second effect comes to
dominate the first:inherited rewardscome to exercise an overwhelming
influence. Put in simple terms, even if we accept the neo-liberalnecessity
of incentives to encourage economic efficiency, too great a degree of
inequalitycan have negativeeffectson such efficiency.On the other hand,
if we were to plot the level of economic efficiencyagainst,Bwe would see
that in the area around ,B*the efficiencycurve is relativelyflat due to the
small power term in equation (6). This suggeststhat the loss in economic
efficiencyarisingfrom a ,Bthat is only slightlytoo largeor slightlytoo small
will be relativelyminor. Nevertheless,increasesin inequalityabove ,B*are
unwarrantedin the sense that they will reduce mobility for no gain in
efficiency.l4
FiguresI and II illustratethese points.In FigureI valuesof efficiencyand
mobilityare plotted against,Bfor a valueof (xfixed at 0.5. Thisshowssteadily
decreasing mobilityand a level of efficiency that first increasesand then
declines as ,Bexceeds the criticalvalue of 1.189.At this point- the point at
which economic efficiency is maximized (the correlationbetween ability
and positionis here 0.41)- the level of mobility,givenby 0, is 0.36. In Figure
II efficiencyand mobilityare plotted against(xwith ,Bchosen so as to maxi-
mize efficiencygiven the particularvalue of (x.So the points plotted here
are the maximum values of efficiency and the correspondingvalues of
mobility.Here a steadyincrease in both efficiencyand mobilityis evident
as the contribution of ability to resources increases at the expense of
inherited endowments.

IMPERFECTRELATIONSHIP BETWEENREWARDSAND POSITIONS

If we furtherassumethat the rewarddistributiondoes not correspondper-

fectly to the optimum distributionof positions, then economic efficiency
declines further.Individualschoose to occupy particularpositions, within
the limits set by the resources they possess, in response to incentives. If
these incentivesare such as to rewarda lower ranked position more than
a higher ranked position then we might expect the correlation between
resourcesand position to weaken.We can model such deviationsin at least
two ways.l5First,we assume that these deviationsin rewardsare random
withineach generation.l6In such a situationeconomic efficiencydeclines,
as expected, but mobilityincreases. The reason for this perhaps initially
surprising result is that the pattern of rewards linked to positions is
438 RichardBreen

FIGURE I: Efficiencyand mobilityplottedagainst inequality

constantly changing in successive generations and thus the correlation

between the positions occupied by a familyis weakened,so causing mobil-
ity to increase.
Second, we might instead assume that the deviations in rewardsare
maintainedfrom one generation to the next. So this might correspond,
for example, to a situation in which certain positions were now sys-
tematicallyover-rewarded,relative to their ranking, while others were
under-rewarded,and the families who occupied such positions in one
generation were able to restrictentry such that the same families tended
to occupy these positionsover generations.This would cause social mobil-
ity between positionsto decline, for the obviousreason that we have intro-
duced a further factor linking the positions occupied by the membersof
a given familyover generations.It would also reduce economic efficiency
to the same degree and for the same reasons as in the foregoing case -
namely that the distribution of rewards now no longer matches the
optimum rankingof positions.

ETHNICITYAND GENDER

Thus far we have assumedthat non-randomvariationbetween individuals

in the resourcestheypossessarisesfrom the inter-generationaltransmission
Inequality,economicgrowthand social mobility 439

1.2 - ---- -- -- ---------- ------------------ ------

-Efficiency + Mobility

0.8

0.6

0.4

0.2

o
.05 .10 .15 .20 .25 .30 .35 .40 .45 .50 .55 .60 .65 .70 .75 .80 .85 .90 .95 1.0
a

FIGURE II: Maximumvaluesofefficiency

and corresponding
valuesof mobility
plottedagainsta

of rewardswithina family.However,there are at least two forms of stratifi-

cation that do not arise in this way.For example, ethnic inequalitiescer-
tainlydifferentiatebetween families,but the source of this differentiation
is not the rewardsaccumulatedby the previousgeneration:rather,the dis-
advantagesthat flow from the membershipof a minorityethnic group are
ascriptivenot simplyfrom the point of viewof the individualin the tthgener-
ation (as is the case with inherited resource endowments) but are ascriw
tive from the point of view of the entire dynasty.Thus in modelling this
form of stratification,we must treat it not as a characteristicof individuals
but, rather, as a characteristicof families. Assuming,for simplicity,that
there are two ethnic groups in society,one of which is disadvantagedrela-
tive to the other, we can rewriteequation (la) as follows
Sit= aLAit
+ (1 - (X)Rit-l+ si (lb)
Here vris a parameterwith mean zero, that takesa common (and constant
acrossgenerations)positivevaluefor all membersof the advantagedethnic
group and a common negativevalue for all membersof the disadvantaged
group. Disadvantage,in this specification,is modelled as reducingthe indi-
vidual'sresourcesfor obtaininga position.
The effectsare reasonablystraightforward.l7 The varianceof resourcesis
increased,so reducing the amount of economic efficiencyin the society.
Intuitivelythis is because the relativeinfluence of abilityon resourcesis
440 RichardBreen

weakened by the addition of the ethnic group disadvantage parameter:

members of the disadvantaged group find themselves in positions lower
than those occupied by members of the advantaged group when both have
the same level of ability. Mobility is also reduced because another factor has
been introduced that links the positions of a group of families through
generations. Recall that mobility is maximized, for any level of ,B,the larger
is the influence of ability relative to inherited endowments. The addition
of tr is an extra inherited 'endowment' (a negative endowment from the
point of view of members of the disadvantaged group) that, accordingly,
reduces the overall amount of social mobility.
In contrast to ethnic group inequalities, gender inequality operates
within rather than between families: it is an ascriptive characteristic that
applies to the individual and not to the family or dynasty. But thus far our
model has not distinguished, within a given generation, between indi-
viduals and families because of our assumption that a family in a generation
comprises a single individual. We now relax this assumption. Let us suppose
that each family has two children, a girl and a boy, and let the subscript j
identify the two sexes (j = 0 for men, j = 1 for women). Then we might
replace equations ( 1a) and ( 1b) with
Sijt = cxA,jt + (1 - CX) Rijt-l + bj ( 1c )

Here the resources available to women are less than those available to
men by a quantity, 2b, (80 > 0, 81 = 40) that is meant to capture women's
disadvantage in the competition for positions. The impact of this on econ-
omic efficiency is the same as ethnic disadvantage: arbitrary distinctions in
the distribution of resources serve to diminish economic efficiency. It is also
worth noting that the existence of gender and ethnic disadvantages also
serve to reduce the value of ,B*,making it more likely that a given level of
reward inequality will itself act to reduce economic efficiency.
Once we turn to the question of mobility, however, matters become more
complicated. Some readers will have already noted that, by adopting our
earlier assumption that the individual and the family were identical, we
avoided difficulties that arise in deciding whether the 'unit of analysis' in
measuring mobility should be persons or families. This, of course, has been
a much debated issue in the context of class analysis.l8 Once we introduce
gender into our model several possibilities become open to us: do we define
mobility destination according to the individual's own position or in terms
of the family's position; and do we define mobility origins in terms of the
position of one or other parent or as some weighted average of the two?l9
And if we use individual position to identify both origins and destinations
we must decide whether we want to measure intra-gender mobility (mother
to daughter and father to son) or inter-gender mobility. If we concentrate
on intra-gender mobility we will find identical levels of mobility for both
sexes, for both of which the model given by equations (la), (2) and (3)
applies.20 But such a model is not really capturing the mobility process,
since, for women in particular, their mother's resources are unlikely to be
Inequality,economicgrowthand social mobility 441

a good indicator of the resources available to a woman when she was

growingup. And, of course the same might be said in respect of men and
their fathers, though here the neglect of the mother's resourceswould,
empirically,have been less problematic,at least up until veryrecent times.
Nevertheless,such considerationswouldseem to arguefor takingthe latter
position: that is, using some weighted averageof both parents' positions
from which to measuremobility.
If we do this then we are faced with the necessityof incorporatinginto
our model an account of how individualscome together to form families:
what, in earlier times, might have been termed 'the marriagemarket'.2l
The operation of this marketwill be a significantfactor influencing social
mobility,and to deal with this fullywould requirethat the model be greatly
extended.22However,we can gain some insightsinto the effects of gender
distinctionsusinga verysimpleexampleof a marriagemarket.Supposethat
there is no mobility through marriage:in other words, individualsform
familieswith other individualshavingexactlythe same inherited resources
and suppose further that origin and destinationposition are measuredas
the averageof the positionsoccupied by individualsand their partner.We
then find, comparingthe position of familiesover generations,that mobil-
ity is marginallyless than it was in the simpler model. This is because the
variancesof the resourceand position distributions,being the sums of the
variancesof randomvariables,are now somewhatsmallerthan previously,
so decreasing mobility.If we allow for more marriagemobility through
decreasingthe correlationbetween fathers' and mothers' positions, then
mobility increases because the variance of inherited rewards becomes
larger.
But if, instead, we compare the individual'sposition with that of his or
her familyof origin we will find more mobility,for both women and men,
the greater is the degree of female disadvantage.This is because the cor-
relationbetweenan individual'sposition and the averageposition of his or
her parentswill alwaysbe smaller the larger is the degree of female dis-
advantage(the largeris 8 in equation (lc)). It is paradoxicaland undesir-
able that a measureof mobilityshould increasewhen gender disadvantage
increases,suggestingthat this is not a useful approachto measuringmobil-
ity.
Thus,althoughintroducinggender distinctionsinto our model does not
change the wayin which we measureeconomic efficiency,it leads to some
complicationsin measuringmobility.But regardlessof how we choose to
proceed we find that female disadvantagewill either have no influence on
the amount of mobilityin society or, depending on how we define our
mobilityoriginsand destinations,willserve to increaseit. This followsfrom
the wayin which gender disadvantagehas been included in the model -
namelyas a factor that influences all membersof one sex or the other in
the same way.If we assumedan interactionbetween gender disadvantage
and, say, inherited resources,then this result might not follow. However,
studiesof the mobilityof men and women suggestthatgender disadvantage
442 RichardBreen

does indeed operate much as we have specified, in so far as social fluidity

is very similar for men and women although absolute mobility patterns are
quite different (see, for example, Erikson and Goldthorpe 1992, chapter
7). None of this should be allowed to obscure the central result of this part
of our analysis - which is that gender disadvantage will result in women
having a higher probability of being found lower in the position distri-
bution than men of the same ability. One consequence of this will be to
reduce economic efficiency. At the same time, gender disadvantage will not
act to diminish mobility since, although it shifts downward the distribution
of women over positions, it leaves the covariances between positions in suc-
cessive generations undisturbed.

SOCIAL MOBILITYAND ECONOMIC EFFICIENCY

It is a simple matter to calculate the level of mobility compatible with

maximum economic efficiency in a given society. In the case of the model
given by equations (la), (2) and (3) it is the level of mobility that arises
when we insert ,B*in equation (5a). But the level of mobility is in no way
determined by the level of economic efficiency in a society. A given level of
economic efficiency will arise as a consequence of the values of the two par-
ameters of our model, (x and ,B,23and the same level of efficiency can be
arrived at by different combinations of these two. It follows, therefore, that
the level of mobility in a society will depend not simply on the degree of
economic efficiency but also on the way in which (in terms of the values of
(x and 13)this is arrived at. Thus a given level of economic efficiency can co-
exist with either high or low (in relative terms) levels of social mobility.
Furthermore, if we allow for some inheritance of ability (as given by equa-
tion (4)) then the degree of mobility will be correspondingly less for any
given level of efficiency.
As an example, consider a society in which the correlation between
ability and position (i.e. the value of +, or economic efficiency) is 0.33.
This could be generated by a high level of inequality combined with a mod-
erate level of inheritance of rewards (,B= 2 and 1 - cx= 0.53). This would
yield a level of mobility (0) of .17. Alternatively the same level of economic
efficiency could arise from a low level of inequality and a somewhat larger
contribution of inherited rewards (,B = 1 and 1 - (x = 0.57). In this case
mobility would be 0.37. For intermediate values of the two parameters (,B
= 1.5 and 1 - cx= 0.56) that also generate an efficiency score of 0.33 the
mobility value is .24.
How do these findings square with what is perhaps the strongest result
from empirical research into patterns of social fluidity, namely that, in many
countries, this pattern has changed little and that the degree of variation
between countries is similarly limited - results which, of course, cast con-
siderable doubt on the accuracy of the liberal theory of industrialism? This
theory suggests that competition will lead to a decline in the significance
Inequality,economicgrowthand social mobility 443

of ascription relative to achievement and an increase in social fluidity.

Howeverour resultsshow that this is not a necessaryrelationship.Nations
may well seek to increase economic growth as a result of competitive
pressuresbut this need not lead to anychange in fluiditypatternsnor need
they seek to achieve this throughweakeningthe role of ascription.
The limitedvariationthatwe see in socialfluidity,particularlyamong the
industrializednationsin the CASMINdata (Eriksonand Goldthorpe1992)
would seem in large part due to the fact that these data all come from the
so-called Sgoldenage of capitalism'(Maddison1982) when welfare state
Keynesianismwas institutionalized,albeit to somewhatvaryingdegrees, in
all these countries.24But such variationin socialfluidityas has been shown
to exist can be understood in terms of the parametersof our model. For
example, the Swedishcase is one in which there does appearto have been
some diminutionin the role playedby inheritedendowments(and thus, in
our terms,an increasein the value of cx)and in which income inequalities
were restrained. This contrasts with countries such as Britain and the
Republicof Ireland,where there is little or no evidence of a decline in the
role of inherited endowments and rather more income inequality.The
result is somewhatgreater mobilityin Sweden than in either of the latter
twocountries.Whetherthisled to greatereconomic growthin Swedenthan
in Ireland or Britain would depend on the exact configuration of the
variousparametersin these countries.However,it is certainlythe case that
economic growthhas been greaterin Swedenduringthis century.Overthe
period 191S1987 Britain and Ireland had, respectively,the lowest and
second lowestratesof growthin real productper capitain Europe.By con-
trast,Swedenhad the thirdhighestrateof growthafterFinlandand Norway
(Kennedy1992:7).
A more formal test of the link between equalityand economic perform-
ance is providedby Persson and Tabellini (1994). They report a negative
relationshipbetween inequalityand economic growthin empiricalanaly-
ses based on twosamplesof countries,the firstof which comprisesthe USA
and eight Europeancountriesand spansthe period 1830 to 1985,while the
second includes 56 countriesand coversthe post-SecondWorldWaryears.
In so far as economic growthdepends upon making the optimum use of
the abilitiesof a nation's people, the present paper providessome clues as
to whyhigh levels of inequalitymaybe injuriousto such growth.25

CONCLUSlONS

This paperhas presenteda model of the relationshipbetweensocialmobil-

ity and economic efficiency,where the latterhas been used as a shorthand
to refer to the correlation between ability and the ranking of positions
accordingto their abilityrequirements.We began by constructinga formal
mathematicalmodel that sought to capturethe role playedby incentivesin
neo-liberalaccounts of stratificationin sociology.This led to a scenario in
444 RichardBreen

whichincreasinginequalitystimulateseconomic growthand anydeviations

from perfect mobilityare attributableto the inheritanceof ability.We then
introduced changes to our model that sought to incorporatesome of the
major criticismsmade of such neo-liberalaccounts. On the one hand we
explored the consequencesof the intergenerationaltransmissionof advan-
tage and, on the other, the resultsof admittingthat the patternof reward
incentives may not correspond to that which would optimize the match
between abilitiesand positions.These changes led to some quite different
results.Mostnoticeably,the relationshipbetweeninequalityand economic
efficiencyis no longer linear:it is now possible to have too much inequal-
ity from the point of view of maximizingeconomic growth.The relation-
ship between economic efficiencyand social mobilityis also now shown to
be indeterminate.By this we mean thatverydiSerent ratesof social mobil-
ity can be compatible with a given level of economic efficiency. This is
because, in our relativelysimple model, both are functions of two par-
ametersthat capturethe degree of inequalityin rewardsand the extent to
which inherited endowments determine an individual'sresources. This
providesa relativelysimple explanationof whywe fail to observeany clear
relationshipbetween the economic growth of nations and their rates of
social mobilityeither through time or cross-sectionally.
Our resultssuggest that a policy much favouredby governmentsduring
the 1980sand 1990s- thatof increasinginequalityso as to providea sharper
structureof incentives- will not necessarilycause economic efficiencyto
improve.If the level of inequalityin the societyis alreadyat or around the
valuewe havecalled 13*then anyfurtherincreasesin inequalitywill,at best,
be of no benefitand maybe counter-productive. Thus 13*setsan upperlimit
on the level of Suseful'inequalityand the size of thislimitwill,itself,depend
upon other factors.So, in a societyin whichacis small,the scope for increas-
ing economic growth by increasing inequality will be correspondingly
limited. Similarly,the positive impact of inequalitywill also be restricted
accordingto the extent to whichthe incentivestructuresystematically over-
rewardssome positionsat the expense of others.This typicallyoccurswhen
interest groups are able to shape the reward distributionin their own
favour.On the other hand, an increasein the valueof cxwillalwaysincrease
economic efficiency (as FigureII shows), all other things held constant.26
It is also the case that were gender and ethnic group inequalities to be
diminishedthis would lead to an increasein economic efficiency.Thus an
increasein inequalityis not the only policythatmightbe used to tryto stim-
ulate economic growthand in some circumstancesit may not be the best
wayof seeking to accomplishthis.

(Date accepted:October 1996) RichardBreen

Centre
for SocialResearch
Queen'sUniversity
ofBelfast
Inequality,economicgrowthand social mobility 445

APPEND1X

We begin by consideringthe three equation model

+ ( 1 - a) Rit-l
Sit= aLAit ( l a)

Pit= Sit- St + eit (2)

Rit= Pit

The numberingof the equations is as in the text. Note that if we replace

(la) with (1) this merelyleads to a simpler,special case. Bearingin mind
that both e and A are random and N(O,(r)we can calculate the constant
variancesof the randomvariablesS, P and R. Assuming(for concreteness)
that both e and A have unit variancewe find that
r{=ac2+(1-a)2erfi,

= 1 + 2

(ri = 2 a; = 1 + 2

If we now substituteequation (3) into (la) and the result into (2), and
bearingin mind that S has a zero mean, we get
p _ ttAit + ( 1 - t[) Pit-l + eit
it- (rS s

from which it follows (becauseA and P are independent) that the corre-
lation between abilityand position is given by

4) (rS(rp (5b)

and between Pitand Pit_1 by

(1 - at)l3 (5a)
(rs

and so mobilityis given by

(1 - a)
s

Note thatif we do not assumethat the variancesof A and e are equal to one
the variancesof S and P will change accordinglyand (5b) will be rewritten
as
+ = tCA
SP
446 RichardBreen

The partialderivatives

3 aaoand a

are all linear and reasonablystraightforwardto compute and so are not

reported here.
a(r (a/Sp)
a= a
a((rsap)
{x a
- 1/2
srSsrp
= (ot2= (1 - ot)2) (1 + 2) (1 +

a(asaP)- 1 [_20l2-3 - 2(1 _ a)2 -3 + 2)(1 _ a>2

Therefore - -
a<, , {X2-3_ -3 (1 _a)2 + (1 _a)2
a -ffg(rp sP
And so

a >Oif
a
ot2+ (1 - ol)2> (1 _ an2

ot2+(1-ot)2,,4
(1 - ot)2

from which it followsthat the condition holds if and only if

2 li
4 < 1 + (1 _ a)2

NOTES
Centre for Social Research, Queen's Uni- 1. An earlier version of this paper
versity, Belfast, Northern Ireland, BT7 was delivered as my inaugural lecture
1NN. Phone: ++-44-1232-242427;fax: ++- at Queen's University, Belfast. in June
44-1232-239842;e-mail: R.BREEN@QUB. 1992.
AC.UK 2. Analysesshow that income inequality
Inequality,economicgrowthand social mobility
increased greatly in the UK during the
1980s. For example
Considering the fact that income
distributions often tend to change
slowly, the increase (in inequality)
particularlyin the United Kingdom
must be considered substantial
(Fritzell1993:59, parentheses added)
3. The liberal theory has been subject
to criticism on several grounds, not least
the fact that ascriptionhas shown little sign
of diminishing in importance in industrial
societies:neverthelessit continues to flour-
ish. In particular, one corollary of the
thesis is a convergence between societal
types, and this has received considerable
stimulusfrom the collapse of the commun-
ist regimes (see, for example, Fukuyama
1989).
4. Though a dissenting view can be
found in Ganzeboom, Luijkxand Treiman
(1989).
5. Economists have produced several
models that seek to account for intragen-
erational inequality and intergenerational
mobility. Among the most notable are
Conlisk (1974) and Becker and Tomes
(1979, 1986). Goldberger (1989) provides
a trenchant critique of the latter two
papers.
6. We refer to 'positions' rather than
jobs, since all positions in society, and not
simply those that we call jobs, can, in
theory, be ranked in this way.
7. Technicallywe are assuming a single
good economy and that individuals are
identical with respect to their preferences.
Note that the use of the term 'efficiency'
here needs to be distinguished from its
conventional uses in economics - as in, for
example, the concept of the 'Pareto
efficiency' of a particulardistribution (see
Kreps 1990: 154 for a definition).
8. Just as we assumed only one dimen-
sion of ability so we assume only one
dimension of rewards.As critics of Davis
and Moore (notably Tumin (1953), but
also Davis (1953)) pointed out, rewards
can come in a varietyof forms, and people
might seek to perform particularroles for
a varietyof reasons.The assumptionof one
dimension of rewardsin our model is, once
again, made for the sake of simplicity.
None of the results changes if we assume
several dimensions (e.g economic and
447
non-economic) of rewards, though the
mathematicsbecome more awkward.
9. Provided,of course, that rewardsand
position are perfectly correlated, as they
are in this simple model.
10. Note that we do not need to attend
to the distinction between absolute mobil-
ity and relativemobility (or social fluidity)
which is often drawn (see, for example,
Breen 1985 or Goldthorpe et al. 1980).
This is because in our model the structure
of positions does not change over gener-
ations. Absolute and relative mobility are
therefore identical.
11. The correlation is equal to
-1
1+d
which clearlygoes to 1 as 3 o oo.
12. Though such a perfectlyopen society
may be undesirable on other grounds
(Young 1958).
13. The partial derivatives of o with
respect to (xand ,3 are everywherepositive
and negative, respectively.
14. Galor and Zeira (1993) present a
model in which the distribution of in-
herited wealth negativelyinfluences econ-
omic performance not only in the short
run but also has longer term negativeecon-
omic effects associated with social immo-
bility.
15. Proofsof the resultspresented in this
and the next section of the paper are avail-
able on request from the author.
16. This is very plausible. We are assum-
ing that random factors preclude an exact
match between the economic efficiency
ranking of a position and the rewards it
receives.
17. This case is identical to that dealt
with at the end of the foregoing section as
we can see if we write the equation for P in
full. In the case where we have a disturb-
ance factor fixed over generations we have
p (xAs,+ (1 - (x) Rit-l + uj + 13
while in the case of the two ethnic groups
we have
aA, + ( 1 - cx) Rj,_ 1+ STj+ f jt
as 13
448
Defining sT/v = u the two equations are
identical.
18. For recent summaries of the debate
see, for example, Breen and Whelan
(1994); Breen and Rottman ( 1995, Ch. 7);
Crompton ( 1993, Ch. 4); Dex, ( 1990);
Goldthorpe ( 1990); McCrae ( 1990);
Payne and Abbot ( 1990) and Roberts
(1993) .
19. Assuming that all individuals have
two parents.
20. The reason for this is that, once we
focus on only one gender, the parameter8,
now being a constant, playsno part in com-
puting the correlation between positions
. .
n successlve
.
generatlons.
21. In fact, any measure of mobility
except intra-gender mobility will require
that we attend to the workingof a marriage
market.This is because any such measures
will entail some degree of cross-gender
mobility- for example, we might measure
the mobility of individuals using their
father's (mother's) position as the origin.
This will mean that there is cross-gender
mobility between women (men) and their
father (mother). Similarly, using a
weighted combination of positions (both
parents or persons and their spouse or
partner) will entail the same.
22. Not least would be extending the
model to allow for differential chances of
having a partner and differential fertility.
23. And also of 8 and sTif we include
gender and ethnic group distinctions. In
most of what follows we use the simpler
model for simplicityof exposition.
24. The exceptions are Hungary and
Poland which were state socialist societies.
25. Persson and Tabellini provide a
somewhat different explanation of their
result to the one that follows from our
argument. They suggest that since growth
is a function of individualinvestment,high
levels of inequality are bad for growth
because they lead to political support for
precisely those redistributivepolicies that
blunt incentives to invest. However, their
empirical results do not provide a basis on
which to prefer any particularaccount of
whyinequalityis negativelyrelated to econ-
omic growth. As they themselves note,
their attempts to test their specific theory
are tentative (1994: 617-8). Similar argu-
ments about the political consequences of
RichardBreen
inequality can be found in Perotti (1993:
see also Piketty (1995).
27. Though it might be argued that the
opportunity to pass on to one's children
the returnsthat one has accumulateditself
functions as an important incentive which
might be weakened by any reduction in the
impact of inherited endowments.
BIBLIOGRAPHY
Becker,G. S. and Tomes, N.1979 'An Equi-
librium Theory of the Distribution of
Income and Intergenerational Mobility',
JournalofPoliticalEconomy 87(6): 115>89.
1986 'Human Capital and the Rise
and Fall of Families',JournalofLabourEcon-
omics4(3): S1-S47.
Breen, R. 1985 'A Frameworkfor Com-
parativeAnalyses of Social Mobility', Soci-
ology19(1): 9>107.
Breen, R. and Rottman, D. B. 1995 Class
Stratification:A ComparativePerspective,
London: Harvester-Wheatsheaf.
Breen, R. and Whelan, C. T. 1994 'Gender
and Class Mobility: Evidence from the
Republic of Ireland', Sociology29(1): 1-22.
Conlisk, J. 1974 'Can Equality of Oppor-
tunity Reduce Social Mobility?',American
EconomicReview6( 1): 8s90.
Crompton, R. 1993 Classand Stratification:
An Introductionto CurrentDebates,Cam-
bridge: PolityPress
Davis, K 1949 Human Society,New York:
Macmillan.
1953 'Replyto Tumin', AmericanSocio-
logicalReview18(3): 394-7.
Davis, K and Moore, W. E. 1945 'Some
Principlesof Stratification',AmericanSocio-
logicalReview10(2): 242-9.
Dex, S. 1990, 'Goldthorpe on Class and
Gender', in J. Clarke, C. Modgill and S.
Modgill (eds) JohnH. Goldthorpe: Consensus
and Controversy,London: FalmerPress.
Elster,J.1989 NutsandBoltsfor theSocialSci-
ences,Cambridge:CUP.
Erikson, R. and Goldthorpe, J. H. 1985
'Are American Rates of Social Mobility
Exceptionally High? New Evidence on an
Old Issue',EuropeanSociological Review1( 1):
1-22.
1987 'Commonalityand Variationin
Social Fluidity in Industrial Nations, Part
Inequality,economicgrowthand social mobility
II: The Model of Core Social Fluidity
Applied', EuropeanSociological Review,3 (2):
145-66.
1992 TheConstantElux: A Studyof Class
Mobility in Industrial Societies, Oxford:
Clarendon Press.
Featherman,D. L., Jones, F.L.and Hauser,
R.M. 1975 'Assumptions of Mobility
Researchin the United States:The Case of
Occupational Status',SocialScienceResearch
4: 329-60.
Fritzell,J. 1993 'Income InequalityTrends
in the 1980s:A Five-CountryComparison',
ActaSociologica 36(1): 47-62.
Fukuyama,F. 1989 'The End of History?',
TheNationalInterest16: S18.
Galor,O. and Zeira,J.1993 'Income Distri-
bution and Macroeconomics', Reviewof
EconomicStudies60: 35-52.
Ganzeboom, H., Lliijkx, R. and Treiman,
D. 1989 'Intergenerational Class Mobility
in Comparative Perspective', Researchin
SocialStratifcationand Mobility8: S55.
Goldberger, A. 1989 'Economic and
Mechanical Models of Intergenerational
Transmission', AmericanEconomicReview
79(3): 504-13.
Goldthorpe, J. H. 1990 'A Response' in J.
Clarke, C. Modgill and S. Modgill (eds)
John H. Goldthorpe:Consensusand Con-
troversy,London: FalmerPress.
Goldthorpe, J. H. with LleweLlyn,C. and
Payne, C. 1980 Social Mobilityand Class
Structurein ModernBritain,Oxford: Claren-
don Press.
Herrnstein, RJ. and Murray,C. 1994 The
Bell Curve:Intelligence and ClassStructurein
Amerzcan Life,NewYork:Free Press.
Kennedy, K A. 1992, 'The Context of
Economic Development' in J.H. Gold-
thorpe and C.T. Whelan (eds) TheDeveke
ment of Industrial Society in Ireland,
Proceedings of theBritishAcademy79,Oxford:
Oxford UniversityPress.
Kerr, C., Dunlop, J. T., Harbison, F. and
Myers C. As 1960 Industrialismand Indus-
trial Man: The Problemsof Labourand the
Management of Economic Growth, Cam-
bridge, Mass.:HarvardUniversityPress.
449
Kreps,D. M. 1990 A Coursein Microeconomic
Theory, Harvester Wheatsheaf: Hemel
Hempstead.
Isipset, S. M. and Bendis, R (eds) 1959
SocialMobilityin IndustrzalSociety,London:
Heinemann.
Maddison, A. 1982 Phases of Capitalist
Development,Oxford: Oxford University
Press.
McCrae,S. 1990 'Women and ClassAnaly-
sis', inJ. Clarke,C. Modgill and S. Modgill
(eds) JohnH. Goldthorpe: Consensusand Con-
troversy, London: Falmer Press.
Parsons, T. 1957 'A Revised Analytical
Approach to the Theory of Social Stratifi-
cation' in R. Bendix and S.M. Lipset (eds)
Class,Status and Power,Glencoe Ill.: Free
Press.
* 1964 'Evolutionary Universals in
Society', Amerzcan SociologicalReview29(2):
339-57.
Payne, G. and Abbott, P. 1990 The Social
Mobilityof Women,London: The Falmer
Press.
Perotti, R. 1993 'Political Equilibrium,
Income Distribution and Growth', Revzew
ofEconomicStudies60(4):755-76.
Persson, T. and Tabellini, G. 1994 'Is
InequalityHarmfulfor Growth?',Amerzcan
EconomicRevzew84(3): 60s21.
Piketty,T. 1995 'Social Mobilityand Redis-
tributivePolitics', QuarterlyJournalof Econ-
omics110(3): 551-87.
Roberts, H. 1993 'The Women and Class
Debate' in D. Morganand L. Stanley (eds)
Debatesin Sociologzy,Manchester:Manches-
ter UniversityPress.
Saunders, P. 1994 'Might Britain be a
29 ( 1): 23-41 .
Meritocracy?', Sociologzy
Tumin, M. M. 1953 'Some Principles of
Stratification:A CriticalAnalysis',Amerzcan
Sociologzcal Review18(3): 387-93.
Wrong, D. H. 1959, 'The Functional
Theory of Stratification:Some Neglected
Considerations', American Sociological
Review24(5):772-82.
Young, M. 1958 TheRise of theMerztocracy,
1870-2033, London: Thames and Hudson.