World Bank Reprint Series: Number Eighty-five
A. Berry and R. H. Sabot
Labor Market Performance
in Developing Countries:
A Survey
Reprinted from  World Developmiient, vol. 6, no. 11/12 (1978), pp. 1199-1249



World Develovpmeit Vol. 6, pp. 1199 -1242                      o)3(5 750X/78/l201 1199 S02.00/0
0 Pergamron Press Ltd. 1978. Printed in Great Britain.
Labour Market Performance in
Developing Countries: A Survey
A. BERRY
(U,11 c itr oif' To or to
and
R. H1. SABOT*
Tlhe I'od- d BanIk
Summary. - Nlalfuncrioning of labour markets is often given as a principal e.xplanation of the
widespread povertvin developing countries. Open urban unemployment anid disguised unemploy-
ment in agriculture are generally considered symptoms of the poor pertormance of' the de-
centralized system of allocation of labour time and skills in these countries.
This survey leads us to a much less pessimistic view ot' labour market performance pcri sc,
though obviously imperfections do exist. On the o(ne hand, the shifts of the labour lorce in
response to shifts in demand have been noteworthy and suggeNt, at an acgr"ale level. rather
impressive performance. On the other, a closer look at openLu nl.Vo111)1 nioenl, disguised unemploy-
ment, and other possible types of labour market malfLtnctions ;ugAgsts that titey may be less
serious iiiisallocatiuns than they appear, atzd that only in part can suc:h misallocations be attri-
buted to poor labour market functioning.
Though the laboLur market is the irunicdia ti locuis ol' thie problem of low and stagnant incomes
of workers at the bottom of the distribution, the evidence suggests no causality in this associa-
tion. There is no reason to presume that poverty is a manifestation of labour market tailure.
1. INTRODL(TION                     ventionis to illntinulcce labour market outcomes
at ormini jo urn social cost. We documiient those
Our principle aim in this essay is to assess  acc.nrplishtnents of labour markets which are
the social efficiency   of labour markets in     readily  apparent   when    viewed   from   an
developing countries in allocating the supplv of  aggregate and dynamic perspective, as a hack-
labour time and skills among alternative activi-  ground to our assessment of their performance
ties, on  the basis of a survey of relevant       reco0ding to micro-economic criteria suSggested
tlheoretical and applied scholarly research. We  by the funduamentally static coiieept of econo-
use the term  llabour market' to refer to the    Iic efficiency.
com)plex of interactions between sellers and
emnployers of labour services.t The neoclassical
presumption is that these interactions sirnul-      (a) (;iourIh, strrcturgl clhantge artd taahotoi
taneously  determine the aggregate level and                  ?narkel perIl,rnwrr ce
structure of employnment (and, to the extent
that the market is -onst r:rined or suffers fronm   T Ie  spIr iIl. occLItr lI It.rl,  and  sectr)r:rl
inmperfectiois, of imnemiployrnmen   and   job   redisNtriho tionsl of labour whiichi accomplpany
vacancies) anid the aggregate level ind strtcttrre
of :saes or more generally the t-rms of work.
The nature of labour market outcomes. wvhile      i lrrs article is oine of a series of sur%ey articles
important per se, also yields important clues as  1nc by the le I of I inirad I)crlopnrr'nrd Tihsned
$ la~Iv tile editori or Wforldl Dvlopml>ssrenlt. Trlle viewvs
to the nature of the processes by whlich these   presented are those of the authors and not those of
otiteomes are deternmined. This process must be  tl, in,qitutions with whichi they are associated. Tire
specified in order to achieve both our goal and  authors are gratelul to Shlirin \Velji for assistance in
the goals of governjnents seeking policy inter-  crrnipiline the bibliography.
1 l9t)



120 '                0                   RoIt 1) LVI'I1 PM II NI1
growth and structural ehange are observable        higlher income may be associated witit either
both in i1idividual countries over time and in     increa.sinig or decic:ising femiiale parlicipatiol.
different countries at different levels of' deve-  reflecting the differenit cuiltures of 6iilterenr
lopinent:2                                         eGuntries. lhe dlirection of this trend Venierally
1. The   decreasing share of total outpuit     dlcteriiiiiies the trenid of the :iggregate rate ("I
from  agriculture andc the increasing share from   participat ion.8
mi nLitfact oring and services as nIta Iionail iinctllme  In the CoUIrse Of de%Celopitictit the net flit 0o
rises. is associated with a decrease in the slares  workers, miost of whoimi are following training
of the lallou r force in agrictil tm  anid rtural  and   work   pzathis ditfferenlt  f'rom  o reced ing
areas, implying a net urban migration.3 Simul-      celrLtaltioroS. tend(ls to be towardi more' dynamic
taneously the prop()rtioni of non-leisure time     sectors, geographic areas and enterprisc-. and
devoted   to non-agricultfuril activities in tural  towardl higher sl:illekl 0 t,-,1. iionlls. Most of tllis
areas rises.4 At hiigler levels of nationial incomie,  slhift  is  presu uiiwbly  from i-c Iivt ic  wvith
and especially wheni the nurithiei ot workers in   relat ivcly IoW  prIodUCtivity to activities Witlh
9
agriulgltuLrc is declining absolutely, an increasing  relatively higlh productiVit) .  While this does
share  of rural family    inicomiie comries tromn  not mican that the labour miiarket is efl'iciiei. as
non-agricultural activities.                       efficienicy  is assessed by conipa11Mrinig marginal
2. As per capita income rises, the dieclinie    rather thntl a' erage product ivit ies in losing and
in the share of output from agriculture, where     receiving activities, it does give a more favour-
small   produictive  units  still  predomiinate,   able pictuire of labour imarket perlorma1,n.ce than
tocether with increases in the size ot inaniuac-   iudging the nmarket in poor cointrl`s by the
turing and service eLtuabli-.hmcnlts. produces an  level of workers' incomlies.
increase  in  the  number of employees        as      \Mltulnction ing of the labour market miay
opposed to independLenii workers, in pad f.mily    mintilihiwe to thie scarcity of ci)Imiplenl eot arv
helpers, or employers.5                            resources, low mL1ualiiV technology, and une(lual
3. The increase in the tcchniological comii-    distribtutioni (f reoitres ehicl   in large part,
plexity of non1-dgriclli itrul product ion and in  explaini thle pinIMel 1lcnon of poverty, hut there
the size of establishmenits brings gi cater division  is a priori no basis for a'sLu m1ing that it is a
of labour and hence increases in skill differen-   major factor. Ihe market's efticieney in imield-
tiation and in levels of training. Greater availa-  ing  the  cLhanging  needs of the ccooniy's
bility of educational opportunities, and corres-   productive apparatuis v% itli the chlnging ablities
ponding   shifts in the supply   of persons of     aid pi-referenc. es of labour force  participants
dilterilng educational attainiiemts. are a general  cannot be judged     by  the share of people
feature of development.6                           co nsmniing less tlhani sone arblitrary in ininuinii.
4. The   increase  in the share of mutpu.it     More aplpropi iite criteria  whiether a realloca-
from  services, together with the adopti(on of     tion of labour wouIld pcellllt an increase in
labour-saving technologies in manufacturing as    oUtp)Ut (or an increase in a mlore complicated
income rises, contribUtes to a relative increase   social Wellare functioxl)  are more ilitlictilt to
in the demand for non-production workers. Ihe      apply. But the reporonsiveness of labour alloca-
response, faCilitated by the rise in edLuclatiuiial  tion to changing demand suggesls to Its that the
levels and aspirations, is an increase in the      burden of proof shouldl not be with those who
supply of whlie-collar, professional, and tech-    claimii labour markets are ctficient, but with
nical s%orker., and henee an increase in the       those WIho   1Cege on the basis of ob,servitions of
propuortimio of the labouir force in these cate-   p)ove tv alonc thiat these la:boo r nmarkets are
uieles.                                            serij(msly incfricient. TlmeretoIcl  we adoptl   what
S. The process of sLruLtutral Lhmange is not    is best tcromedi a (hicago Sclhool stLinlC. O(ur
accompanied by systerimtic sllilts in thie rate of  null hypothesis is that, given    the existing
participation in the ni.mle labour force in the    techlnology, sItrucfet e of jlCef'erLnLes, aid stock
central adult ages: tHIis rctimaiwms at a high ievel  of' plysic:l c pital anid land, nio appreci:i b11
and above) as national income inreases     inerement in aggregate economic weltt;re or its
rhere is a tendenIL, however, for the rates to     rate of' growth. is to be hiad1 by such reallocation
ieClin1e in you nger age grouip, as prolmOnigel    of tle labour force as a more perfect lalbour
edlucation delays entry into the labour force,     niairklet coulld lurinlg aout.
and in older age groLu.p,I because men retire
earlier from  wage than non-wage employrnent.
Among femai;les, the rate of particip:itioni in the    ( h) Ialour i market pcYtir1aMocc?LT critcria
youingest and    oldest g-roi pis also  tends to
decrease, though these trends are less consistent     In a,svessing the degree of inefficiLicy of the
than amnonig mnales. In the cenitral age groups.   labour miarket and its imuplicati ns for %%elt'are,



I .AlND)LR 1\01KN 1                                 OU1 KI tR ;\l  I 1,  <II  PI 1 ,( N I Rl S  1201
wCe cmrir!re the actual witli the optimlal alloca-  because labour mlarket researc:h ha, focused
tion of laboour. TIhis is coniventionally defineid as  prcdorrmirrad tlY  on  thiose  inelficiciicres  with
that in which the marginal social productivity     readily apparent sy rrrptorrrl, urban unernploy-
of a g:vcn   type of labour is eqlual in each      miient anld slhor't hours ot work in rural areas
possible use, and the relative marginal produc-   lbeing the nmost notat:.'. TIhlus in Section 3 we
tivity of any two different types cequlals their   cani only partially 'test' oo. null hybpot hcs. as
iarl-irial rates of substitution in each prodlue-  our j'ivestir,a(ion of costs is conifinied to the
tion process. Marginial social product is diefined,  Imiost cotimimionly  investigated  labour irimket'
in turn, as the ineremiienit contributed to the    iIIffiCiCic'Ilk'.
value of the output otf goods aInd services of the    Several other caveats regairding our assess-
society by an ad(lition;al worker plus the net     ment   of   labour   market   performance    are
diiect utility (or minius thie net disuitilitx) of tlle  suggested by the conceptual (questionis heggcd
work itself to the individual. UI'IderlVilln our   by the definition of efficiency. WNe note these
initial presumtiiption of labour market etficiency  lhere  to  avoid  tedious   qualification  sub-
is thie assumpijition that individuals are respmonive  seqUlen i, and to emiphasize thiat (luite (letailed
to changinlg labour iii.muket conlditionls, arld thlat  investigation may he necessary to reach firm
workers tilvidie their timne between leisutre and  coonclusions abllmt the causes, cost. and appro-
s%tork. and choose among alternative occupa-       priate remedial policies ot an apparent labour
tions, industries. anid geographic areas as thotugih  miarket problem. [Ihe firs. sta,ge of any test for
they have a fine regard for marginal costs and     nicfficicnicy in the labour ii.li;kel proceeds in
rettirns. We assume further that the behaviour     one of two ways. lither we seek evidlence of
of 0iiipI -Crs rard rig     f '  ii.r g decisions cani  phenomlena such as open unemployment or
also lie interpreted as an atlerlpt to inai\niiize  wage gaps oetween workers ot cormpalrable skill
net returns and, furtleirniore, that costs and     that thieory tells us are symptois ot sNl teiiiintkic
returns to individlual decir, i0-ri.rkAer- accuir-Aclyv  malfuiiictions,  or,  alterniatively.  we  seek
reflect costs and returns to society.              evideence of' plhenomiena such als illmiiilnu   NNage
UWe do not review    sJihl,irl\ work oin thie  leishIliOll, labour uiinioII or1 imiinpOnll\ powNer,
adequacy ot the profit maximii/a(tion aissumip-    th at thieory tells us siotiuldi catuse the ahbour
tion for business firmis in I D)( as uillost ot whiat  ruarket to mialfutiiction.
little thlere is on this subJeCt is conlce.nemd with  Ihle conivenitional criterion  ot e(quality ol
issues  otiler than  dlccision-mi-inglg and  be-   marginal social prodIuct imiplies an exclusive
haviour in the labour market. We focus .ur        identification of optimal labour alhlcat oii withi
attention in Section 2A on the responsiVenless     maxiniumi   achiievable  giopegat e output.  Illie
of individual producers arid sellers o1 labouir    allocation that is optimal in this sense may be
services to  difterences in prices and n(i   es    associated withi ani 'inequitable' ;dh trihLIlb L)ro  ot
I[libtoghi  behaviour  consistent withi rationail  Income. WVIr I we recognlize thie tegitlilrirev of
decision-making on the piart Of inrdividlu,ls is a  addi.ng the 'best' achievable d11t rubILI1o01 o  f
necessary   I riditi lr. it is byI no mieanis sutfficlelnt  inicomie as another dimension o)f our assessmlient
t(o assume an optimnal allocation of labiouir. It is  of perli'mriiarce, we liave not done so hiere
also niecessary foT all workers to be Price-takers  because it would increase the scope and coml-
anlLd to  be  free  to) enter any   industrv  or   plexity of the survey, and because we dioubt
occupation tor whlichl they are qrdihitekd, and to  that e\isttrig infGrmation would advance Us
compete in it on1 equal terms wNith otlhers  Ilie  very far towardL tihe desired miiappinig of the
rio trLierlrs restrictionis on freedlom of entry and  d ist rIlhLiroIM (of income tiroii variouis .11OhL. Ic r',
iiiequiities in thie tdrrLhiritu it of tippiiitWlies, in  anid labour niarket stirmctiures Ne'ertlhelcss, we
the labour imarkets Of dle etlipilg cotintries arise  recognize thalt oiur neglect of thrs dimiiension otl
tromii a variety of sources. sutcl as nionopsoni0r  labour imartket prrimmrice will redluce coii'i-
poNwer,   di,crii ilrill:t io ii  goverinient  inter-  dLence in our ability to tise the theory of (lie
ventionis,  ard   workers'   associationis.  I lie  opllim nal allocatioil ot resources as the basis for
;rilmci pie aini of SeCtioll 2 is to . (lilt thiese  eonmlfirifirig  lalhorr rmarket  inefficiency b) y
restrictimis andl inequities anrd  the specrt e    lciaireri ti ng observable causes. I'or ex.r iriple.
fornis of labour misallocatioris to which they     tlie costs, in terinis ot aggregate output fore-
give rise. It is not possible to  et lciiil inc for eachi  gone, that a le,il.r lted nimiiiium iwage imnposes
of these misallocations, liowever, whether the     on  the e0cu1io11riy  rniy  be e'xcee&dil by the
social costs are iiiiriiriald or sufficieri Ii higlh to  bierie'it's it yields, ill ternis of a more equitable
refute ouI null hypotlhesis of a high degree of    dStLrirlruiirn of income.
labrur rnarket etficiemiy. In part this is because    Similarly, consideration ot micro-dl    .iiii y  ir,
of, I 111.1lolris otf spicwe. hut imore fundaamentally  aspects ot the process of labour Al(), .1lln



I202                                WORLD I1)1 \'I I 0111 NI
reduces   thle  relibi litI  of  the  Symptom   change. Correspondingly, one ;ssesses sucl
approach   to .dentitfving  inadequate  perfor-  performaince with s,ome trepidation, since there
mance. Phenomena which would be sinmptonia-     is always the risk that what appears inefficient
tic of an inefticient labour market in the static  would in fact be judged efficient (or vice versa)
sense. such as unemployment or wage differen-    were better datar available or otlher relevanit
tials for a given skPW. rna1  not be so when     ConsidLerIations  taken  into  account.   Our
dynamic factors are t ;   into account. Given a  rationale for persevering is that policy decisions
positive relbaionship  hetween lengthl of job    and dlecisions oni the allocation of r[e'LMrL'C for
search  and  level of wages associated    witlh  research are generally based on a mllch more
accepted job offers, unemployment may be the     liimiitedl body of' evidence than that wlich we
mIIost productive use of an individual's time, not  have been able to collect, and that therefore a
only when alternative activities are assessed in  comiprelhenNive  review, tentative  thom.ulgh  its
terms of private costs and benefits but also,     eCucluLsion,s are, may im prove the quiality of
given the prevailing lags in information flows   discu.sion  preceding such  decisions. 13 lotre
and wage adjustin1eits. when they are assrsed    Considering each of the cited sn iptomis ot
with regard to social costs and benefits.       rlos'ille labour mnarket inefficiency and attemp-
Likewise, labour allocationi may     be  both   ting to gauge the imiportanlce ot the various types
privately and socially optimal despite a wage   of loss, we review some of the evidence on the
differential if the low wage earner is in a job  behaviour of the actors in labour markets and
offering greater opportunities for learning by   the nature of the wage dgtermination process.
doing and hence a fa .t -r rate of increase of
earnings.
The degree of complexity     of the task a         2. RA I lONAL INI)lVID)UALS AND
labour market performs should also :ntlllen;C              StIB-0)11 IMAL SYSTIN %IS
our assessment of its efficiency. W'e notedl
above the positive corn elatiorn  between the
level of national inconme and the extent of        (a) 7The rCij III.Si IC%s ot.l.lio it)I-oIa1ers
struc[Luralt change and ditferenti;ition on both   tC di/t.t0Crencs in econtiotic, opportunities
the supply and demand sides of the labour
market. The increasing inportdneT of' matching      I vidence that individual (lecisions on htmLni
skills and job requirements relative to the      capital imi'.estniernts, labour force partiCipation
spatial and sectoral allocative functions of the  and hours of work are based on comlllllex
labour market pres,irrably increases the poten-  cost-bcnefit calcilatWons is not necessary to
tial for inefficiency. Fven more inolnitinl,    suplport thie presumiiption of an eftikient labour
given the pervasiveness of interactionis amongl0 i market. Whetlier decisions are ulided by con-
the produet and factor m-narkets in decentralized  ventional rUleS of tlhumb or even motivated by
economies, is the effect of the structure and    apparently non-ecoi-onoic factors is irrelevant as
efficiency of other markekt on the allocative   long as inldividuials behave   as if they are
tasks of the labour market. To conclude that     attempting to max.mize expected returns.1 I
one labour market is more       efticient than  The long-lield belief that the indigenou.s popula-
another without taking into account the larger  tions of developing countries are not 'economnic
economnic environment is like coinp:pring the    men', in the sense that they appear to pay no
rates of acceleration of autitC mobiles w-ithotut  heed to niirCin.al econioin it incentiveN, can be
saying that the time trial for some of them n   dismiised as a nmanifestation of western etlhno-
were held on a 30" incline. Ihe inalfo ictioninig  cenitrisCni. Nevertlieless, evidtence of internal
of Capital, land or product markets may result  inconsistency  in the   beliaviotir of workers
directly in the mnisallocation of labour. For   would mnake our task of asses.inlg Idbo LI market
exa mple. insufficicnt labour in an indtustry may  performaniie' unmanageable. In thiis reLgard, the
be the conseLlienlce ot its monopoly power. WNe  resl Ik  ot' recent nmicCic-ecunCOInIc eiimpuilical
attempt to account for such interactions and    sti udies are hceairteniing, as they- s1ugg`est thlat
()ther comphjilicaiti hg factors,. but our treaitmnent  workers in ID('s behave in a quite conventional
of them,. reflecting their neglecL in the works of  maimer and are hig,hfly responsive to economic
the researchers on whom  we have relied, is not  stinmli.
nearly as thlorouighi as they deserve.             Most evidence on short- and long-run rela-
Evidently no such simple criterioni as the   tionishlips between thie price of a crop and its
rate of uinemprlo-nment can be used to judge    sales on tlhe mnark"-, for example, stipports the
laboLur market performance. e,.pecially in deve-  conclLusion that farmers responrd to changes in
loping cotuntries uiii(lergoing rapid strulctLral  prices. Indeed, some evidence  from   cross-



! .   I'R \I.\  I T 1 '1 RI ()1R\I.\'( I IN DVII C 1()IING; ('t7N IRItIIX  I 'R203
country comparisons of the price elasticity of    hypothesis that laboutr supply functions are
supply of given crops suiggests that there is no  backward-slopi4g. The phenomnenon was alleged
appreciable  difference  in  dajJustmlellt lags   mainly in those African and Asian countries
between farmers in developing couintries and      where the wage systvimn   of  liploynlieni was
those in hiigh-ilonice countries.12 Recently the  established on a large scale by foreign eniter-
interpretation of resistance to innovation as     prises which, at least initially, experienced
non-economic belhaviour has been challenged       chlronic shortfages of unskilled labour. In Africa.
by evidence that traditional methods are often    sLarLity  of wage labouir persisted   into the
the most appropriate for local factor endow-      1950S.1 5 s I'his past scarcity, however, is not
ments and that many of the techniques pushed      sLufficient to substantiate the claimi. Nor does
upon unwilling smallholders by colonial agricul-  evidence  on  the  reduction  of the scarcity
tural experts were uneconomnic. For example,      problem  support the poplar 11yipothesis tlhat
the land-intensive techniique of sllifting cultiva-  over time, as the CaIshI ecoloin0ly spi-ead. there
tion which is still )revalent in tropical Africa is  was an increasing preference for goods anitd
now seen as economically efficient in contexts    services purchased with caslh, ai'd this increased
in which land is effectively free) l3            the 'nornality' of responses by workers to
The price elasticity of supply of a given crop  econrmic incentives. While economiie miowthl
reflects, among other things. the elasticity of   may be associate(d with clhaniges in preferences.
labour supply, a matter of particular interest    some recent hhtorical research supports the
because of our focus on the labour market.        alternative hypothesis tlhlt scarcity of labour
There are as yet no precise studies of the        was due prinmrily to wx aLge 1 below the equili-
relevant contributions to the price elasticity of  brium  supply price of labour while the recenit
supply of trade-offs between leisure andl work,   eilimination of this problem  was duie to wage
z-goods  arid agricultural  production. wage      iicicieases. Wages were niot hid, iup) by enmllloy !s
labour and agricultural prod(uetion. one crop     dLuring  the  coloniial period. because vari'Ous
and anothler, f.ersonal consumption and market    coercive measures were wvide.ly uised to cOilt-
surplus.  Ihere   is,  ho% c' .e.,  eo isilder.mble  plemiient economiic incentives in generating aii
evidence on the shape of the :iggregat e labour  adequate suipply of labour, and these redutced
supply schedule of rural dwellers.                the costs of a shortftall of workers Nohiinlecring
Until  quite   recently,  the  conventional    for cilll)iOV1ent1 . lt' l he evidence also  Lgg1ests
wisdom, palrticlal1rly in Atrica and Asia, pre-  that trade and innovation in consumiiptioni were
sumiied that schedule to be bjckw %ard-bending in  characteristics of economic behaviour before
the  relevant  ranges  as  a  consequence   of    tlle inculrsion of direct foreign inmesmnente
preference  structures which   place a greater   and that inadequtiate recognition ot the .mnount
value on leisure than on wage goods. comp.red1    of eniill(l ymime1 t in the production ot i-goodsi 8
to industrial societies. This view is not neces-  ledt to  overe-;tii11.mt is of leisLure time.  Ihe
sarily an indictment of the economic radiionahltx  iinprobabililty of a back Jar-slopingp .igregale .lte
of the xorkers:. nor does the imiplicationi oif  labour suppfly curve is underscored bN. the fact,
incomle effects domuinating substitution effects  noted by Berg, that wlhat ilmiglht be true of
in regard to the leisure work trade-ol0 flsggest  individual supply functions was unlikely to be
aniv static inefficienicy  in the allocation of   true of the aggregate functi in, since tlie partici-
labour as conventionllal     defined. However,    patiion rate in wage niiiphsm)vient as well as
such a sehedlule may imply that sectors with the  hours worked WOUlIL be expetedl toi respond to
pot emit al for the imiost rapid growth of output,  imicoiesese in wages."' In sunm, the apparent
prodiuctivity, and savings will be starved for    lresiAleCLTe of idlletiess in the rural sectors ot
labour as attemnpts to attract workers by raising  SOmle diL'VelOping couintries is Im(ore likely to
\agcs have perverse results. (iiven the prevailing  reflect the lOW  iiiimginl.l uLciii 1s to) Iddil ioi.mI
low level of inicorlme, pgovernmint ilmay place a  work  thatn   ay    inort h ()(It)o   pi terenice
lower weight on    leisure tlhan do indlividual   fu ncttionis 201
workert, and t ieret1ore be j1iii i icdl in conclud-  Our comparisons ol Lill'It CICI Cut ntries anld1
ing  that  thle dLeccn11t.1rie1d  labour  market  11ifrerni periods suggest that mniuiatimi, like
allocates labourinelticien ly. Speci.ti ally . if the  inter-sectoral mobility, is a  uitu.allv uni'ersall
sa'ings rate is believed to be suob-optim.il. or it  feature of ccinOIllic growth. Ihe high propor-
it is telt that indiViduill prelerercecs will change  tion ot inigraii H in the labouir supplv to the
after introduction  to a (lifferent  economic    non-agrictultural sector In tlvetClopinu couintlies
environnient,14  such  an argument could be       is the  basts of the rcceiit rapid gri; t1I of'
enteritjinced.                                    micro-economic   research  of the    miiigration
In fa. t. there is little evide[lne to support the  process  TIhe results are an inip,lamir new



1204                                 MR1\ t)l D)FIVI I 0P111 NI
source of evi(ldence on the responsiveness of       Large-scale sample survzys made possible by
workers to economniC stimnuli.                   recent de.velop)mentso in the tvchnol )gy of data
Brigg has reviewed the findings of 27 field    obocessing, suchl as thni-,c recently adm;inistered
surveys of urban and rural communities21     in  In dndonesia, Iran, Kenya, Sierra Leone and
Africa, Asia and    Latin  America, in which       1,zania.2 7 and  deesigned  withl the specific
migrants were asked directly why they moved.     purpose (,t generating a Tnicio-data base appro-
Most of these were administered prior to tlle    priate for the analysis of   iigratnt bellaviouir,
recent surge of interest in migration and were   hiave at least parti.illy resolved the piroblemis of
designed  by   non-economists.   l)espite theit  miissing Ndriables alld excessive lugreg:flion.2
pervasive methodlological weaknesses.-2 hoiw-     Relaxing the constraints iniposted biy the need
ever, the virtual unanimity of their findings on  to use ieaCdilV available data shifts the tocus
nmigrant decision-making is notable. Fconomnic   back to sonme still serious conceptual problems
motivations predomlinLte, and such factors as    regarding variable mneasuremnenit For example.
saccompanying   family  members' or 'seeking     the difference between source and      eceiving
education opportunities' are primary reasons     areas in the incomles of employed workers is a
for imioving in only a small proportioni of cases.  i1n nponen t of all econoric expla nat ions of
karly  behavioural studies,. whiclh estatblislhed  miiigrant behaviour. Most emnpirical analyses
simple negative correlatioins between regionail  hlave relied  on extremely crude proxies for
levels of income and regional rates ot entigra-  incomiie: goods produced for horne consumption
tion, also lend some weiglht to an economic      and the value ot services or amenities subsidized
interpretation of imigration.23                  by the government are frequentlyn e oitted from
If miiigration is viewed as an investmlenit in  consideration, even though as a proportioni ot
human capital2,4 whiclh requires, resources and  total incomiie they are likely to vary signifi-
time to realize a dlilterenti set of emyploylmenit  cantly amwong geographic areas.
and earnings oppori u nitic. then   imInltiviriate  TFlis latter aspect of the imi precision  ot'
analysis will be the lbest '\;a to study it. Ihis  measures of incomiie :ni\ noit be as serious as it
mlore advanced type of behavioural anahllsis of  first appears, at least fromn tIle perOfe.li' t
migration is still in its infancy and there are as  using the re.sUlts of research tor policy planning.
yet significant unresolved data, ine:isuremiernt,  A  priori there is a strong pre- inllpi inn that
and specification problems. Ihe deinographic.    ni igra tion flows, and labour siupply in general,
mobility,  employment     and    incomie  data   are more elastic witlh respect to 1hanlges in
gathered by comn entionnl instrumenlts such as   monetary   than sjecific nomi-ionet ary incen-
censuses, anniual surveys of emEiployment and    tives. Given diverse lpret'eiecLe-. poods provided
eirning-, and   household  buedget surveys, is   by the governmiienit are uinlildy to have the
generally  inadequate  for such   an  exercise.  attraction ot miotiney, the imiost etticienlt instru-
Information  on   key  variables is eithier not  tluelit of exchminue. Rurai resi,denits who lhave
collected  or is spread among a number of        stronig  preferences  0o3r pdrticulal.r aIImlellities
sources which c:i nniot be d'etail. In multi-   provitded  by   the   govcrnimerit  over  Other
variate analyses based on existing dita, there-  couso niption itermis imiay be indifferent to the
fore, the high level of *iggregwtmon may obscure  'currency' in which their increase in incoilie is
a number of significant difterences in patterns  received. if they receivedi higher imoney pay-
ot response   among   migrant !iriss  andl stub-  inents they wouldl in any case exchange it for
groups of the popillat lill  In particular, such  the ameinities.2 9 Others whlo prefter the ni menti-
studties foctis on the relationsliip bet%keeii inter-  ties less will findl it ditIIHtit,LI  if no0t i imIploible ',
re,ional l Hows anid average e;irimipiis clifferen  to ex chlingc their increase in access to the
tials, even thiough the >trollilre of' differentials  amenities for goods they value more. Tlmis, in
b1t s%ecri sotircL and rece[ivimli areas nimi\ differ  the imggregIte thei given increase in rural in-
dramatically as between thle rural rural and     vommies is likelx to be lperceived as,, less wlhen
rural urban flowNs, .md nilt togi intrai-regionil  received in kind ttian in -i h, anid suitl Li 1its
lows miay be mi a ntit atmly as i iii polmta ut  f1'  enmistrining effect on in igi at n0ii is likily to be
rreaiter concern, these studies. rarely analyse  less.
tlo(ws between regionls that are roughlily homlno   Spatial dif'ferences in prices and the resIliultg
geneous in elucattionn, sex arid igC, thirebl    index    niiinib er  problem    pose   another
biasing the assessment of responsiveness to the  problemtl.30  A  elated difficulty follo%%s fromi
e\tent that app)arent spatial differences in     spatial differences in thie proportion of inter-
e-inrinuic oppiortu nities only reflect composi-  mediate goods andl services, whlich facilitate
tional di fle're nces in the labour force and do  other cmisiminptinmi but do not confer utiililtv
not have a bearinog on incentives for uiigra-    directly, in total olnu mpim)lion  't1icievney o;f
lion 2fi                                          eor 1insuitip1oii iin ay also be affecte(id by spatial



I AiB)1 '    1R MAR11 1 '1 I' 1 P]R I  *XN( 1 IN Dl VI I CPI( ( DUN I R11 iS  1 205
dlittcicinces in cliimate or othier aspects of the  elasticities  fromi  recent   st udi e  in  Kenya
pli.sicail environmlent, or in poplulation density,  (rural urban), 1aim,ani.   (rural urban), Vene-
wlhich tmay necessitate inore cxpendituL1C on an    zuela    (inter state) and    India (inter skite)
input suc:h as sewage in one place than another     wvhich lhave corrected for many of the grosser
to achieve the same final obiective, Individuals    iniethodulogical problems that charil leri/Cd
who have achieved the same ob jeLtiVe shoul11 be    earlier multivariate analys'es. In eachl case the
said to have idlentical incomies regardLless of     sign on thle de.stinaition Nvage dusti city is posi-
differences in their levels of expenditure oni      tive and on the origin wage cla,t icily, niegative.
in eriiiedliate'  goods.3'1    Conversely, . the    findllrigs wlichl support the presrsi In tlion that
dlifterence in expenditure levels between rural     decision-miiakers respond   to economilic oppoi -
and utiban workers with the same human a:pital      tun1ities.
endownients mighlit do no iniore thani collr   :-     [IIiis responsiveness does not imply that thie
sate sutclh differentials, rather thalnl bieing a net  illndiVIdIlsIj are incomiie maximizers. hlie cotnven-
addition to welfare.                                tional justiicLItiionl tor ashsuwinlug that entei prlises
Ihese are some of t(le prolIleIHs ot econo-      are profit nixi\inli-ers is not that bLuinessiuieCi
metric analysis of the resplonse of in(dividluals to  ulw,iys miiake econoimiically rational decisioiS;
spatial ditferentials in econolilic opporituniieti  rather it is that, no nmatter what the nmotivation
From   even this brief discusSion it should be      of the decision-miaker, the operaition of the
clear that estimates of the elasticity o1' the rate  market will ensure that only the iniore efficient
of iiiigration with respect to spatial differences  producers, the pi olit maximlizers, survive. With
in income are likelly to be sulh'iect to consider-  regardt to investmiients in hUmlan cai        the
able error. Nce'ithcless as a meanls of testing     mnarket cannot performii this function. An indlivi-
specitic hypotheses on migrant lbelhavi(our and     dlual Who dLecidles not to invest in education or
of assessing economic responsiveness in general     inigration, despiite positive ecu nimmiii  returns.
terms, these studies lprovide a inore rigoloil      dtoes not place his survival ais a participant in
lbase than earlier onies in NOicli in igr, 01it were  (lle labour miarket at risk, lie Niuniph accepts tlie
tqluest ined directly, or those whllicl relied on   li  Iiehhoo  f o f  earning  a  'Xi0 tel intome3le
bivariate behavioural analys!s. I able I presents   l I ipirical siI upport  o t  these  isIerIt Ions  IS
I'able I . Partial incimtO)?  clasticitws rvoln iiiigrctioni JU1tictiofl3 t0,) Ilt'11
I'lya               fan7ania               e'eneIuell             India
IlIIItItnt.Ii)   (lBarnituji anid Sabot)**  (I evv mnd Wads cki)   IC reenwuod,
)pelpedent variable          if                    Uf,,
I)Detination ware I I'j)    +6.7(J'               -1 .'26                +11.94*               + t i
(4.6 1)                                       (2.59)                ('.l2)
Oricill vv aizc (it")        1.5                   1.5(6                   ()1.85                1.24*
'2 (o)/                                       (2.32)                14.48)
.Sntmxr  t . Yiap (1975,.
Notes: 'ul'idicamlt at the five percent level.
I )DtI inuit 11:  A1N  : eira tioll from pilace i to 1
Pj,JY  POPUlation in plIace i, j, respectiwvel.
**II. lIarnumii an(d R. Sabot estimiated a linear lunction, using lifetime dirilimest. iiindlsctuiiieI, I lie elasticities are
calcLilatedt at tihe mean oftlihe variables, using t(ie income ' t1l lfi1 ie lb, 0(1. 24 (destination il, mim. i andl -I1IIi7I)
(origin incomee).
1 M. , reiwooil  s (lcdplendil x riable is 1i,,, r.itIler than .1 1,Pi, lowever, the incomile ocifliccicnis \Vould not
chanae it lhis model vvere re-estimated, usinpY tle rate u1,J,,',, for i'j, included in riglht-liand side of' the equation,
has a LIIL-ltl ilnt of approximately one,
In f"ther wvords, the income coefficient ', is the same tilr
X '  ) X'~
anld 1!,,    '
f'i



1 206                                 WVO RI I) II \' I . OtNlI N I'
provided by the nmlgration ftunction e-tinlated    flow  of labour market informa;ltion, such as
tor Tanzania. Barnum and Sabot denlonstrated       em1lpoy1mlenti  exchanges. trade journals, and
that the positive relationship betweeni rates of   local newspapers, of the poor cotuntries do not
migration and years of sclhooling is only parti-   imiply a .signi'ic antly less efficient comtnunicai-
ally  explained    by   movement     along   the   tions  system. Indeed, on     the basis of the
function; significant shifts of the function also  indirect evidlence of the ri-sponse ot farmers to
contribute.33                                      price  changes   and   of  workers   to  spatial
Thle role of non-economiic faclois in labou-    diflert.nces  in  eirniings  it is reasinable to
allocatioh raises yet another seriouis practical   presiume that the informiial c.-hainels are not
problem    for  thie  assessment  of  allocative  mIuChI less t.ict'iLlenl.
efficiencY. In order to interpret the persistence     TIhus our brief review of the evidence on the
of income differentials among workers with         causes of labour market inefficiency tocuwes on
0iin ilar embodied human capital as signfilying    non-mnarket   or  ill-titUiOli nal  distortliOll . of
the  presence of allceative inefficiency, it is    wages. We cniphasi/e at the outiet that, given
necessary to a.ssuiiie that workers are income     the cuirienit state of knowledge of the ftiunctioIn-
maximizers. If. h(,wever, workers' preferences     ing of labour markets in L DCs, it is not pcsssible
are mnfluenced   by the status of the employ-      to arrive at estimates of the social costs of
ment, or its cleanliness, or the extent to which   labour 1lisallOLti(M.N based on observationi of
it allows them to remain their owIn bosses, then  l1yp)otheCtical cLiJUes. On the other hand, as we
persistent wage dlilferentiadls may only be com-   illustrate in our discussion of open uneriploy-
nensatine   for tdillerenlcs  in  the  non-wage    ment, to deterluirle whether a given pheno-
qjualities of different jobs.                      menon    is synipltoniatic  ot a serious or a
relatively  insignitica ii  problem  freCLqiently
requires specification of its cause.
(11) lW'age distortions and lab(ouir         tUnionis and governinent % ae lpolicies are
nairket in"'ftivitici-               often  identif'ietd as piincipll causes of wage
rigidity, lienel  ot persistent wage distortions
On the evidenlce lust considered. we can        and segmiented labour markets. Biy seginienta-
eliminate  perverse or internally   inconsistent   tion we miiean that some workers are paid more
lbehaviour of workers as a basis for predicting    than  others with    the  same human     cplital
the  persistence of dis4equilibria aririn" from    endowments simply by xituie of their 'sector'
changes on the supply or demand side of the        of einphylomielnt TIhlis situation sihould be dis-
labour miarket. Other potential auses of iaiiiire  tinguished  frolll dLicriiinatiion, whlere some
to eu(llalile the imarginal social produci1:t! ol  -orkers are pai(d mnore than others with the
hoinogenOeLIS labour miay be grouped int( three    samBe  liuim.ai c:ipital endowmiient by virtue of
categories: monopoly power or other inipert cc-    somie non-economiic characteristic sucl as their
tions in   product or other factor markets:        race, relidmio  or sex. In practice the two may be
inadeuate oCMMunnicdtion between emnplhyers,       extremlely difficult to sepiaraie. as segmlecntation
workers and potenltial participanits about es,en-  implies the rationing . htigl1i-s age employm ient
tial labour market informationi; inmi-mnark-et    o(l-'iom tniiics and the criteria alpplietd may be
ioluen;,   on mobility and the wage dleterinina-   discriminiatory.
tion  process. The first of these should irit          hlie '.iniilamity in the rate of increase of
intluence  our ju dgement ot' l.ibour niarket      imlepe  lent   natius w  and  coiuntries  %\ lere
peCNtolMimCnc directly. WN ithi regard to the seconkl  w(rkers are itepiesemitid by unions and, pro-
there  is no   consensus on   what constitutes     teoted bh  riiniimunimii \saLle laws in the post-war
esseoti.al informat ioirl and very little research on  peritdtl is nmore thian coincidenie Hlimnination ol
its rate of dispersion or its accuracy. AHtL-11E it  ilu       kr(1o1llusive practices whlich dip)e1sed
by govermini en I of LDCs direc tl! tt control      ages, anld of the aipalllot t.i.itL su ppoll tIy
inobility, as thie Sotutlh Africans do() "itll their  goicnmlient, I for sutclh practi,N'S was p)erceived
pass  %sItein  or as is (lotie in ilie cenlti.ly   b y tle ness political elites as onle ot' the imimst
plannedl economiiies. arc still rare and uemelm  %all, v  :il able tfruits of political aiitoilnoinyI lilde-
unsuccessful.                                        finIl'llh . I h' ugh erroneously, political rat her
The absence of rigorous researclh on intol      thian economnic torces were believCLl to have
n.it ion flows do .s not seriously jeop:rdiic our  been re'   nm msbihle tor the low waiges of u iskilled
tttenipt to assess labour market performance.     labour. a view   whicehi may well explain the
Research on thlis area in inmd striali7ed ;ocieties  rea1d inc'- of new goverrinments to ad1opt wages
is nuit mnu.hi inore advainced. Fu ithierimi tO the  policies Vith pitll 1 t iallN si iomus econoinic eon-
less hiighl y LI (Ieiped fom iii al Channels for the  seqi icn. for tlle cttk ILenii alloca lion no' l.i1h MIi



LfA`XUR MARKT 1P1 RI ORl\ANCI' IN D1) LI (IING; ( M)N I RIIS          1207
and  for output and     employment growth.     by selecting the most proficienit from the pool
Raising wages above the level determined by    of available workers. Thus part of the wage
the market reduces output and employment by    differential between the protectedi and unpro-
inducing firms to select more capital-intensive  tccted sectors may be matclhed by a skill
technologies than they otherwise would; also,  differential extremiely  difficult to imeasure.
by segmenting the labour imiarket, it can lead  Ilighi wages and good living condiLions may also
simultaneously to scarcity of labour in some   contribute  to  higher productivity  through
unprotected sectors and excess supplies and    better health; high relative wages may pay off
consequent underutilization of labour in pro-  in lower turnover and more positive worker
tected sectors, pheniomiienia we examine in    attitudes. Indeed witlh these latter factors in
greater detail in Section c on open unemploy-  mind, it has been hypotliesized that elploy1ers
ment. Furthermore, it can induce socially un-  themselves may be primnarily responsible for
prodtuctive investment in education and mis-   increases of wages in excess of the level
matches between worker skills and job require-  necessary to attract a sufficient number of
ments. a phenomenon we consider in Section     workers.34  WN'lhen  employers expect wage-
c(iv).                                         related gains from increased productivity and
Though    unions  and   government   wage    decreased  non-wage costs ot labour to be
policies have  undoubtedly   raised  the real  greater than the costs of increased wages. the
incomes of protected groups, there is little   wage per man which minimizes total costs per
empirical evidence as to the further crucial   efficiency unit of labour is greater than tha'
questions of the magnitude of the gap between   which clears the labour :narket. There is some
the protected sector wage and the equilibrium  evidence from Africa, where short-termn eircuilar
supply price of labour, and whether these gains  nmigration between rural and urban areas was
are mainly at the expense of capital or of     characteristic, to support the hypothesis that
unprotected workers. In principle, minimumli   eiployers in the    nianufa]Cturing sector in-
wage legislation differs from union power in   creased wages as a way of stabilizing their
being applicable to a country's wlhole wage    labour force alld justilfying iniestr menit in the
labour force (at least us usually conceived). but  training of the indLustrially disciplined semi-
in fact both are relevant mainly to larger firms  skilled workers necessary for factory employ-
and the government. How much they push up      ment.3 5 A positive relationship between wages
the incomes of the protected group       will  and productivity increases the difficulty not
generally depend- on:                          only of allocating responsibility for observed
(a) The strength and aggressiveness of the  wage differences but also of determlining the
union and, in the case of miniuniiin wages, the  cost to the economiiy of sucli differences. Thle
commitment of the government and its        adverse conseluLences for allocative efficiency
ability to enforce legislation;             noted above may be offset or more than offset
bb) The elasticity of demand for labour by  by imuprovenients in X-efficiency. 6
the relevant employers iif uite inelastic a    Even if we could resolve the problems of
big push for wage increases will te nattralc  measuring wage differentials net of differenices
an inash fodemagnd is a       natural result of  in non-wage qualities of jobs, and assure that
an inelastic  ind is and           result  f  the labouir whose wages are comlpared is hiomo-
price inelasticity of demand for the final   gno,adtedfeecstepouto
routand a low labour .share              geneouls and the dliffierences the product of
product                                      non-imiarket forces rather than  induLe.ld by
(c) Thle size oi' the protected wage sector  enploy-ers, measures of the differenice in wages
relative to other sectors. it it is quite smlall.  between  protected  and unllprotected sectors
wage increases will not induce inflation; if it  would still be a poor gui(le to the social costs of
is large, they will, so that real wage increases  a wage di.stortion. It is now widely acCep)tedl
will be less than money wage increases,      among economists speciahli'ing in the econonmicse
A  uiniple compalrison of wages in prote.- "l  of labour in 'ndlUstrLralietd societies that the
and  unprotected   sectors is inistirfficit to  Lltelrelce  in  wages. for example between
estimaLte the impact of non-niarket tforees, not  uipionized and nni-unionizedl estahlishments, is
only because this procedure fails to take into  a p6oor p,roy for the measure of the gap
account differences amlonig .bs in net psychic  between the c:quilmbrimui market wage and the
benefits or, if the sectors are geograplicully  union wage. The supply price of workers in the
separated, costs of moving. The issue is rather  non-unionized sector may not be the same in
more coinplicLted. For one thing, employers in  the presence of a unionized sector as in its
the protec ted sector, faced with payment of   absence, due to the 'emnuilationl effect', wlich
high wage-s, are in a position to 'skiim the cream'  may  stimulate  wage  demands frol non-



I 2(S                                 \ OR I 1) )I VII [ 1 o\II NT
uinioriized  emnployees. At tlle samiie time the    planation of dispersion. lhe following explana-
prospect of uinionization     may   increase  the    tory variables are norniaPly regarded as denloting
receptivity of enphloyers. On the other hand,        'lhumian  capital', tthat is, the acquiiisition  of
the negative effects of hig'i wvages on labour      pliodlLucti%e k<now0ledge: education, years in cur-
demand in unionized establihmirents and their       renit job., years of past employmient experience,
positike  ettects  oin the aggregatc suLpply   of   andt fornial training. The increasing frequency
workers in the ocLuptlion      in (111estion mlay   with which wage strul.tures in L.DCs are subtjec-
lower the    supply   price  of lOf   1lrll to  non-  ted to mutltivariate analysis largely reflects the
unionized establishments. 7                         spread amlonig thiese countries of techniques tor
hlie samne qualiticar ions apply to an asse,%s-  the  critiCiiln and   foorimlalil.tioll of  duILIcAtion
iuent of the dlirect impact ot minimum     w-ages.  policy b.ised on estimiiates of rates of retLllll  to
But it is somiietimes asserted that the indirect    :nvestnment in edlucation.39 The narrow     tocus
effects of legislatedl wages tor unskilled wor-     of these studies has so far limllited the potential
kers are quantitatisO    the mllost distortionair  .  of this type of analysis to yield imnportaint clues
Whether thlis is so deperinds on the degree of      to other aspects of the inter,cltions between
rigidit v of the wage sirucltuoe, hence the C\Ielll  sellers and einlof'ei s of labour services that
to whi,c1i increases in w,iges at the bottoml Of    detc   ilmne s ages.
the occupational ladder influence wages higiher        TIhe coelticient oin the indl.epltlellt variable
up. Our understanding of intra-firm   taLge struc-  repiesenting years ot schooling. in a IfLnction
ture is at presenit inl.llVeuate to judtge the degree  where the dependent varalble! is thei log of a
and nature of such rigidities,. though it wvould   nmeasure of wdges, can be interpreted as the
seem  impiob.ihle that firmiis would not mlodify    average percentage increas,e in earniings associa
wage structures it the inpmiotio n ot a miinmumi    ted with I or more years (t schtolihng. or as tlhe
wvould otherwise greatly increase tlheir %tage bill.  average gro(ss private rate of return to additionl11,
Whether ejnplkloer, can e\pect to pass higlher      edlclctioln. As in the ilnduslrialiiedl couintrieN,
wages on in the form   ot higfher prices is tllus a  studies tn d0elInJ) 1g ColintliiCs innvarialyl yield
consideration.   In  liighly  protected  modern     a pos1,itiVCe coeftiliCin  on tlle educatllin varn-
sectors, nionopoly and ofit±opol\ power are the     able.4 IThe conventional view of this positis e
noirml. so  such  an  ex\pe,tit imi is plLiu-sible.  association between eduicLionlal attainmnent :an3d
Clearl\ economnic strueture andL ewpecttion'        svages is that schooling is a production process
witlh respect to zencral price inflition are also   for human capital, thouglh perhaps only olne of
rele'.ant here.                                     many.    Without    an   independent    econoiiiki
In a statistical analsis. Jackson               imieasure of schooling ouitput, lhowever, esntina-
inferred  twoi chlaracteristic processes of itnteu-  tion of ai) education l)llkdv iOl<   finction is
relationship between legal imiinimiiumiii wages antl  iinimissihcl. For the samiie reason, we are forced
actual wage levels in poor cotuntries In onie. sa  to rely on a inca  re of inpu ts. generalll time in
in a 'key' sector advance rapidly. the legal      chooul. when we atteiiupt to asseIs the contri-
tminimum is period icki 1 JJ isted to )ollow.V th1LhI  hoution of the cd neat ioinal production i process to
not necessarily in proportiown: -roups between the  worker productivity. Since we cannlot dircleCl
mninimumi and thle wage leaders also strungle to  estimate  in  economic terms how      muchi the
restore their rel.tite waue position. but wvith   education   process   in, reases  an  individu:il's
%ar ilnl  success. In the alternatihe process the  productive capacilty. aniy evidence of a relation-
tiiinimiumii wvage is puislied up by Ilotliii,.il pressure,
mi   e is  i                    . '   hip bet ween years of schooling and p)rlduc-
and worikers pa,: above the minimum  then strise  t               o      t
.avain. with varying success  to i maintain their     y must be
previotis mnargin over that base. 1ither ot' thiese  tiOls. Thle 'screening' hvpothesis. for instance.
mnechanisms tould produce both a conipa.ni:tivel>  stUggests that thie higher p)rodtictivity of wvorkers
wide 'lispersal' of wvte movemenits and consider-  with miore education reflctcs t lie snUCess of tile
able mnsiatiiliti in thie wage hmiera'lr\L. I lev also  school proninotion  systenm in  electing those
ilmiply. of' t .orNs. lilm. il distortions o1f the relative  wvith iniore prod uctive capacity,42 rather t ha n
wage structure are miucih mnore Ilk-el to) occur in  the  direct protivity  enhancing eftctl     of
the underdeveloped ecoinoiies than in the ads an-    dir        o    t   y     hniO
ced ones.3                                       schooling.
Ihle imost comnmorm way of assigning weights
While researclh on the level and structure of    to these alternatve hypotheses on the role of
%%agese lias.not yieltded precise mneasure.s ol the  the education system is to zinclude a measure of
cost of distortions, it has nevertheless provided   'ability', suich  as IQ  in the wkage ftinction.
the basis for a few tentative generalizlatio.ns on  Fconornetric studies of this type have i-ot
the  roles of huaian    capital, laboour inaiket    resolved the issue, in part because of vatriability
segnient:tlion. and   dIiscri nfinamt ion in the ex-  of results. More ftmndanmen tally this approaclh is



I AIBOt'R MARIKFT Pl:RIORMANCI: IN 1)1 '1 ()IPING ('COUN1 RII S        1209
open to the criticismii that it is wrong to assume  tional attainment as a hiring criterion and
an identity between productive capacity and     workers to allow themselves to be graded by
intellectual ability as mneasured by psychologi-  the education system. Where education does
cal tests. Rather the former 'is an unobserved  not enhance productive capacity, that part of
latent variable that both drives people to get  the cost of education which would not be
relatively more school and earn more incoome,   justified on the basis of direct consumption
given sL0ullng. and perhaps also enables and    benefits is a measure of the cost of this labour
inotivates people to score better on various    market inmperfection, but there is no reason for
tests. Basically it is a hypothesis about the   assuining  that investment in    education  is
cause of and a re-interpretation of the correla-  socially unproductive,
tion among the residuals from   individual in-        Education still has social value as a
come, schooling, test scores, and other equa-   signaling device which helps to match abilities
tions. As such, it is only loosely related to   and job requirements, assuming of course that
"ahility" as it is commonly understood by       different qualities of labour are not perfect
psychologists. It could just as well be "energy"  substitutes in production.46 There is no reason
or "motivation".'4 3                            why output gains from improvements in alloca-
There is, however, some indirect evidence    tive efficiency  in which category would also
and a powerful a priori argument in support of  fall the social returns to migration  should be
the hypothesis that at least some part of the   valued less than those from increases in produc-
wage dlit'ferences between groups with different  tive capacity, Presumably, the negative con-
levels of education reflects the role of the    notations of screening for social returns to
education system  as a screening mechanism.     education  are the result of a hypothetical
Presumably graduation from a course should be   comparison between the costs of a conventional
an indicator of productive capacity, one com-   edLucation system  and the drdniatically lower
ponent of which is likely to be 'staying power',  costs of a screening system based, for example.
more useful to an emplover than mere atten-     on a battery of aptitude tests. Such a comipari-
dance for a number of years, Thleref'ore,       son may indicate that a shift in screening
whether the rate of return is higher to those   techniques would increase returns, but it does
coimipleting the final years of a course is a test  not imlply a reduction in current estimates of
of the screeni'ig hypotlhesis. No significant   social returns to education. Indeed even the
difference between 'dlron-outs' and 'conmpleters'  former is in doubt. DLiscussion of superior
in returns to education is found in studies of  techniques begs the question why in LDC
school-leavers in  the  United   States.44  In  economies, in nmany of which there are high
Malav'ia  l however, the earnings t'unction esti-  rates of inntovation and the government does
nmatedi by Mazumdar reveals significantly higher  not lhave a monopoly in the provision of
returns to those completing each of the three   edueational services, market forces have not led
post-primary levels,4 5                         to their adoption. The obvious answers are
It is difficult to conceive of a production  either that the productivity-enlhancing role of
function  for human caplital that does not      education is important at all levels or that
include as an explanatory variable the different  superior screening techniques do not currently
levels of efficierncy witlh which time spent in  exist, because of the differences noted above
school is used, Even if the criteria for promo-  between intellectual ability  and pre-existing
tion on the job    are based entirely on the    productive capacity. It is possible- that both
iccuullMlation of' productive skills rather than  answer'. are correct.
on indicators of pre-existinig productive capa-    The relative importance of the screening and
city, We Would still expect to find a positive  prodlctti%ity enhancing aspects of the educa-
relationship between the latter and edtucationral  tional system  determiiine the cost of a labour
level unless tlhere is zero or negative correlation  market imperfection, not tlle social value of
between education and 'efficiency' or 'ability'  eduL-cation. Cconfirmationi of thle inplort-.nice of
as defined in the prodtuction f'unction. This   this imoperfection does not imply a labour
seems highly unlikely.                          market inefficiency, however, according to our
The screening hypothesis has generated con-   criteria of eLlualization of social marginal pro-
troversy  because its confirmation seenins to   ductivitv. Nevertheless, differences between
imply gross social inefficiency in the market for  countries in the coUtN of correcting an imper-
education. In our view, this interpretation is  fection common to all migllt bear on the issue
incorrect. Imperfect knowledge of pfe-existing  of differences in labour market perfornmance by
differences between   workers in productive     suggesting differences in the magnitude of the
capicit  Tmotivate  emnployers to use educa-    imperfection.



1210                               WORLD DL' VIF'.hLPll-Nr
The length of experience of workers in       enterprises, also for reasons of ability to pay.50
emnploymnent (or its proxy, age) is often found  To  briefly  summarize the conclusions of
to explain nearly as much of the variance in   attempts to test these hypotheses: differences
earnings  as  educational attainment.    This  inr mean earnings are consistent with these
suggests  that  perhaps   a  disproportionate  hypotheeses, but using inultivariate analysis to
amount of attention has been devoted to        control for differences in the compositioni of
formal training as a means of accumulating     the labour force between various categories of
human capital.47 Experience within the firm    employers greatly reduces the independent
generally has more effect on earnings than     impact of firm clharacteristics on wages. Knight
previous experience; this may reflect the extent  and  Sabot found that the contribtution of
to which skills acquired on the job are firm   differences in firm   characteristics to  wage
specific. One mneans of assessing the influence  variance was less than 10%'s of the contribution
of labour market segnmentation on wage disper-  of differences of the personal characteristics of
sion is to include in a wage function a dummy  employees.5I
variable indicating whether the worker is a
casual or a regular employee. The presumption
is that regular employees are protected by       3, SOME SYMPTOMS OF INEF-FICIENCY
governnment legislation governing wages or by   AND ITS QUANTITATIVE SIGNIFICANCE
other institutional interventions, while the
wages of casual employees are determined          Persistent disequilibrium  in the form of a
solely by market forces. Consistent with seg-  large number of workers without enmployment
mentation  theory, Knight and    Sabot have    but willing to work at the going wage, un-
shown that in Tanzania, casuals are paid less  employment in the Pigovian sense, is the most
than regulars solely on account of being casual,  frequently  noted  manifestation  of labour
that they possess less education and that what  market failure in LD)Cs. It is by definition a
education they possess is of less value to theml.  wage sector phenomenion, hence the potential
Analyses  of   discrinmination  in  the labour  for open unemployment increases as the share
markets of developing countries and of its     of the labour force enmployed by others rises, a
contribution to the explanation of variance in  corollary of structural change. Indeed the evi-
earnings are still extrenmely scarce. Wage func-  dence suggests that open unemploymient is
tions estimated in Malaysia and Tqanzania with  relatively unimportant in the poorest societies,
ethnicity and sex as independent variables have  where self-employment predominates, while in
revealed significant differences in wages, parti-  the developing countries its rate increases with
cularly among racial sub-groups, after standard-  the national income. On the basis of this
izing for measured levels of embodied human    relationsliip, of evidence of declining aggregate
capital.48 Economic discriinination might seem  labour force participation, and of disappointing
a natural corollary of the racial divisions in the  rates of growth  of industrial employmtnt
politics of developing countries. At the same   despite high rates of industrial investment and
time the particularly large cultural differences  output growth,52 there have been nurnerous
between ethnic groups suggest the need for      declarations  to  the  effect that the  mal-
caution  in  interp)reting  the  evidence. The  functioning of the labour market has generated
ethnicity dummy nmay simply be a proxy for     a problemn of grave proportions. To quote the
some unmeasured chlaracteristic of the high    Pearson Report, 'The failure to create nleaning-
wage group that increases their productivity.  ful emiiploymnent is the most tragic failure of
There are a variety of hypotheses regarding  developmnenit. All indications are that un-
the relationship of a firml's character to its wage  employment and underutilization of hluman
level that have recently received rather pronmi-  resources have increased in the l1)60s and that
nent attention. It is conventional -wisdomn that  the problenm  will grow even more serious'.53
in developing couLntries wages are higher if the  This sentinment has been eclhoed. both in general
emnployer is a imultinational enterprise botti  terms54 and in specifically African,55 Asian56
becaLIse, being a nmonopoly or collusivc oligo-  and Latin Anmerican57 contexts.
poly and for other reasons, they have the         The presumnption   is that unemployment
ability to pay higher wvages and because, being  poses a double threat in the form of significant
in a politically exposed position, they are     resource costs (the outpuit foregone relative to
willing to do so.49 Similarly large firms are  a situation in which all workers are employed)
thought to pay higher wages than small firms    and welfare costs (the dcnmoralization and
and firms with capital-intensive techlnology are  physical dieprivation of unemployed, workers,
thoought to pay   more than labour-intensive   taken as an indication that a significant propor-



IlABOUR MARKET PF'RIFORMAN(IT. IN DlLVl.OPING COLIN I'RIFS         1211
tion of the population is not sharing in the  slope of the labour supply curve in the long
fruits of economic growth). In additioni, un-  run, that is, the problem of labour absorption
employment is sometimes viewed as a cause of  over several decades or, from a different per-
political instability.58 It is not surprising that  spective, to the magnaitude of labour reserves;
the unemployed should exercise considerable  while the last is more akin to measures of
political influence, nor is this inconsistent with  poverty. The phenorr.'na measured by these
the fact that those at the bottom of the income  various approaches generally differ in causes,
distribution in developing societies are among  social costs, and policies appropriate for reme-
the least likely to engage in illegitimate, or  dial action; by lumping them together we lose
indeed any political action.59 As we shall see,  the ability to make important analytic dis-
the  openly unemployed are niot primarily    tinctions. For this reason we have separated our
drawn from  these low-income groups; on the  discussion of open unemployment from that of
contrary  they  are disproportiona [ely from  other forms of underutilization.
relatively high income families, and are well   We do this despite our recognition of the
educated.60                                  possibility of functional linkages between open
Assertions of the seriousness of the 'unem-  unemployment and other categories of labour
ployment problem', and the conclusions that  surplus. If a rising share of the labour force
invariably follow  on the need for vigorous  engaged in wage employment increases the
government intervention, are generally based on  potential for open unemployment, conversely it
assessinients  of the  magnitude  of under-  should  decrease the potential for disguised
employment or disguised unemployment, as     unemployment. The combination     of rapid
well as open unemployment. Open unemploy-    urban  growth with the increase in urban
ment, it is said, is only the tip of the iceberg. In  unemployment, partiCularly in African and
the ILO's reports on the emnployment problem  Latin American Countries, together with a
in Colombia, Kenya and Sri Lanka, 35-- 40' of  scarcity of labour for eertain jobs in rural areas,
the labour forces are reported to be uneder-  has led some observers to hypothesize that
utilized, though open unenmploynent is only  surplus labour is being tranlsferred from rural to
about 10%',(. The various definitions of under-  urban areas.61 Thus Ramos has argued, pri-
employment are more ambiguous and hence      mnarily on the basis of steady increases in wages
more controversial than those for open un-   in a number of Latin American countries which
employment. Also such assessments are rarely  cannot be explained by government policies or
based on anything but quite aggregate data,  the pressure of organized labour, that there has
despite the near certainty of significant rela-  been no real increase in the excess supply of
tionships between the demlographlic and tem-  labour in Latin Amnerica; rather, the growth of
poral composition of surplus labour and its  open Ullnelluloynient has been associated with a
resource  and  (particularly)  welfare  costs.  decline in disguised unemployment.62
Finally, and perhaps still more significant for
assessing the costs of a given quantum of
surplus labour, there is a tendelncy to abstract       (a) Open uniemploylmenzt
from the precise nature of the inadequacies of
the labour market that give rise to the problem.  Open unemployment is predominantly an
The aggregationi of open unemployment and  urban phenomenon; in the cities of developing
various other forms of underutilization of   countries estimated rates frequently exceed
labour in the    discussion  of enmployment  levels that would trigger nmajor remedial govern-
problems has caused considerable confusion.  ment actions in indiustrialized econiomiiies. (See
While some measures of underutilization rest on  T'able 2.) Protblems of definitioni and measure-
the same analytic foundations as unemploy-   ment do not reduce the gloom    this picture
ment, in that both attempt to quantify the   projects, as there is no evidence that the
excess of working hours supplied relative to  resulting biases are consistently in an upward
demand at the given wage, others are based on  direction. On the contrary, the use of the
comparisons between incomes and the miarginal  conventional criterion of active job search for
product of labour in particular sectors, between  dtJ erriining the number of workers witlhout
hours worked and some mneasure of total time  employmnent who are labour force participants
available, or between actual earnings and some  introduces a consistent downward bias, as it
arbitrary level of income or productivity. None  excludes from the unemllployed the 'discouraged
of these others yields a measure of labour   workers' who would accept a iob at the going
market imbalance, that is the current labour  wage but have given up the searclh beauLIse of
surplus. The first two comiiparisons relate to the  low probabilitv of emiployment.f63 In a stuidy



Table 2. Recent measuremnents of openi unemployment rates, various countries
Asia                                               Africa                                       Latin America
Open unemploy-                                   Open unemploy-                                   Open unemploy-
ment rate                                        ment rate                                        ment rate
Country               Year        Urban      Total        Country       Year      Urban      Total        Country         Year       Urban    Total
India                1971         3.0         3.9     Ghana              1970                6.0       Bolivia            1974                 9.7
Indonesii            1971         4.8        2.2      Tanzania          1971      10.0                 Colombia           1974      10.0
NMalavsia            1967/68      9.9        6.8      Egypt              1971                1.5       Panama             1973                 6.5
Pakistan             1972                     2.0                                                      Trinidad-Tobago    1973                14.0
Sri Lanka            1969/70      16.9       13.2                                                      Uruguay            1973       8.9
Thailand             1969          1.3        0.2                                                      Venezuela           1971      6.0
Turkey               1969         4.9                                                                  Peru                1974      6.5
Korea                1974                    5.4                                                       Brazil             1970                2.0-2.4
Philippines          1971         11.0        5.3                                                      El Salvador         1975     4.9-8.6    5.2
Syria                1973                    4.5                                                       Honduras           1972                 8.0
Taiwan               1972                     1.5                                                      Mexico              1970                3.7
Average Asia*        1975         6.9         3.9     Average Africa                                   Average Latin
(ILO estimate)    1975         6.9         3.9        (ILO estimate)  1975     10.8       7.1          America
(ILO estimate)   1975      6.5       5.1
Source: ILO Yearbook of Labour Statistics (various years); country census and labour force survey statistics; IBRD country economic reports.
*Excluding China and uther centrall) planned Asian economies.



LABOLUR MARKET PERF:ORN      (TNCI IN DEVI:[ OP'ING (COUNTRII S      1 213
of urban labour markets in Tanzania, based on    criteria  for  open  unemployment:    the
a sample survey designed so that the degree of   division of all the individual's time between
sensitivity of the aggregate rate of unemploy-   leisure (non-economic activities) and job-
ment to alternative definitions could be deter-  search; no earned income; and no con-
mined, alternative labour force participation    tribution to output.
criterid  and  criteria  of employnment were    'The first stylized fact suggests that models
applied in order to adjust for 'hidden unem-   treating uniemployment as a generalized pheno-
ployment' and 'hidden employment'. Although    menon for the economy are not useful in a
the variation in the aggregate rate was consider-  disu.ssion of urban unemployment, since they
able, the  lower boundary    was 8': unemn-    cannot explain excess supply in one sector but
ployed.                                        not in another. The second raises the question
Interniittent collection of information on  of whichi of the various models considered are
the  employment    status  of  labour force    consistent, or could be reformulated so as to be
memnbers and a tendency to redefine categories  consistent, with open and as well as disgutised
between labour force surveys or censuses has   unemployment or underemployment. It is diffi-
resulted in time series evidence that is sketchy,  cult to make sense of the notion of long-run
at best. The apparent rise in the rates of urban  explicit uneinployiment of speCifiL members of
unemployment may be a statistical illusion and  the labour force without taking into considera-
for most countries rough constancy over the    tion some mechanism    by which the unem-
last decade a closer approximation to reality.  ployed subsist. Since the unenmployed in LDCs
Nevertheless, increases in the size of the labour  tend to be young, we cannot look to savings as
force in less developed countries imply a rise in  the subsistence meclhanism. Given the absence
the absolute number of unemlployed workers,    of government transfer schemes, we face a
and increasing urbanization implies a rise in the  cloice between direct transfer payments or, in
rate of unemployment when the nationial rather  the case of disguiised unemployment, work-
than just the   urban, labour force is the    sharing or employment creation. Thus, to ex-
denominator, It is not, however, the rate of   plain open unemployment we miust consitler
growth of unemployment that distinguishes the  the circumnstances under which the employed
time trend of unemnployment rates from those   support those without jobs by direct transfers.
prevailing in industrialized countries. Rather it
is that high levels of urban unemployment are a  Deficiencyi of aggregate dematid economny-
chronic, not a cyclical phenomenon. It is this  wide models. In contrast to classical tlheory, in
distinction that leads to grgve interpretations of  which the labour market is characterized by
the magnitude of the problem.                  flexible WsIgCS, popularized veisions of the
Keynesian  framework assume that the real
(i) The inmplicatiotns of alrernatire          wage, the sole determinant of labour supply, is
mnodels of unelnplovmnenlt for social costs   rigid. Thus there is, in effect, a fixed quantum
The social costs of unemployment are said   of labour available. The employment function
to be so high in many countries that planners  translates any decline in aggregate demand, and
should  give high priority to a cure. It is    hence output, into unemployment. This un-
uinfortuinate that such statements are frequently  emiiploymlent is involuntary in that workers are
made on t-e basis of an erroneous diagnosis or  willing to accept, but cannot find, jobs at
without reference to the causes of the problem,  prevailing or lower wages. Keynes himself did
since costs and aplpropriate policy responses  not in fact posit ligid wages, but simply
vary with causes. In this section we classify  dropped the assumnption of perfect information.
unemployment, models in development theory     He assumed that, in diseqlilibrium, qluantities
and assess their implications for social costs and  adxjust faster than pirices. rhus any deviation of
their relevance to the anaivsis of urban un-   aggregate demaLnLId from  the full emnplo)yment
employnment.65 I'he c,riterion for relevance is  level will be amlnlifiedl mathler than Counteracted,
their consistency with two stylized facts estal-  becatuse  the  initial unernploy ment relduces
lished  by  recent  research  in  deeveloping  purchasing power and creates expectations of
countries:                                     further contractions. Aside from  the strong
positive effect of the negative relationship
(a) the coincidence of unemployment in the  between prices and wealth on demand for
urban  sector with 'full employment' or     output, which Keynes recognized as logically
labour scarcity elsewhere in the economy;   possible but socially unjust and inefficienit, a
(b) the fulfilment by a large number of     decline in wages only exacerbates the situation,
urban residlents of the three conventional  as its first consequence is to reduce aggregate



1214                               WORLD DEVElOPM )PI NTr
demand, reinforcing   expectations of defla-    cyclical pattern of pruodtuctioni and employment
tion.6 6                                        in the industrialized countries, they are not
Where unemployment is caused by         dle-  appropriate  to   the   chronic   urban   ull-
ficiency of aggregate demand, it is generally   employment of LDCs.
associated with high levels of national output     Nor can these models explain the specificity
foregone. In crude terms, the cost of unemploy-  of location and the form of unemployment in
ment in industltrial economies, or more generally  LDCs. The coexistence in LI)Cs of urban
of cyclical dowlnturns in economiiic activity, is  surp,lus and  rural scarcity  contradicts the
estimated as the difference between the gross   presumnption  of Keynesian models that the
national product (GNP) in the period and what   direction, if nlot the rate, of chanige in aggregate
it would have been if GNP had advanced          demand is the same in all sectors. The LDCs
smnoothly from  the pre-recession to the post-  may not be prone to deficienicies in agnregate
recession  level, measured  at the point at     demand. Even if they did experienice such
which GNP returns to the trend line of econo-   delficienicies, the difterence between indiustria-
mic growth projected from    the pre-recession  lized and less developed countries in the extent
period. Exercises such as this conducted in the  of wage employment suggests that the resulting
United States and Britain indicate that in most  unemploymnent would be disguised. WVlhere the
downturns output could be increased by con-     wage system   is pervasive, workers who lose
siderably more than the proportion 'full em-    their jobs in an economic downturn have no
ployniient' bears to actual employinent during  alternative imeans of earning a living. Where
any year of the recession,'6 despite the likeli-  there is free entry to the agricultural sector
hood that eniployers lay off their least pro-   (either because of an abundance of land or the
ductive workers first. There are always large   strength of traditional obligatiUns to family
numbers of workers who, tIloughi not clas.sified  nionmbers), or to the urban own-account sector,
as unemployed, would contribLute     more to    nearly all workers without a wage job can find
output in a tighter labour mnarket than they do  other employment. T)eniand defiliency alone
during a slump: those who have withdrawn        cannot account for significant a    persiibLent
from  the labour force because of poor pros-    open unemploymnent in such a situation. Ihere
pects of employment, those who are working      must be an addlitiolzil mechanism, on whiclh
less hours than tHey  desire at the pievailing  more below, to explain why workers do not
wage, and those who teLpora rily accept jobs at  accept the alternative, albelL meagre, earninig
an occupational level lowser than that for which  opportunities.
they are qualified. 'lo these direct losses in     Other economy-wide models can be classi-
production is sometimes added the effect of the  fied according to whether uneniployllment is
dILownturn on the rate of economic growth.      dedluced fromi teclhnological lack of substitu-
As explanation for urban unernployment in    tion, given the factor endowments, or from the
LDC's Keynesian models suffer from    four in-  fixity/stickiness of real wages. We review these
adequacies. They presume that demand de-        categories of models in turn but, because they
ficiencv results not only in the undtteruitilization  share several important features. assess their
of labour but of complementary resources as     relevance and inmplications for the social costs
well. Yet the numerous instances in which       of unemployment as a group.
deficit finlldcing, high rates of inflation and    Tecnlnical lack ot substitution. Two basic
urban uincniplolv uent have co-existed suggest  miiodlels can be distinguislhed here: the static
that this is not true in developing countries.'6  Walrasiani model asSLoCia1ted witlh the work of
Reddaway lhas eniplinsi,ed '. . .that althIougLh  Fckaus  on  factor   proportions, an(d the
there is abundant labour, at least of unskilled  dynamic Hlarrodl l)oinar nmodel.70 Considering
types, a general increase in deanand will not   first the foriner, witlh two factors (capital and
lead to a general increase in output, because   labour), two commiiioodities and production func-
other ooperating tfactors are neededl to work  tions clharacterized  by  leountief-style  fixed
with labour. The traditional one to take is     coefficients, the productioi poOsibility curve is
Capital   i.e. real capital e(lquipnilent- notlhing  characterized by 'flats' along wlhich labour is
much can be done with bare hiand(Is alone' "    unemployed and there is no market imieclianismii
Second, they ahstract from such fundamnental    to ensuie that demnand will take production to
inflUences on output as population growth,      the production point, if it exists, at which both
technological hane and capital accumulation.    factors are fully eniployed. In the dynamic
The level of effective dLeml-and alonie can deter-  mnodel, particularly that developed by Domar,
mine the volumne of eumplovinent only in the    the  average savings ratio and the mlarginial
sho,rt run. While such nmodel.s mnav fit the     apital outpuit ratio are fixed.7 Ilencee the



LABOUR MARKET PERFORMANCE IN 1)1 \I. 1.O'I;NG COUNTRII:S             1215
economic system is geared to a steady rate of  level of minimum wage, but also with both the
growth and since, again, there is no market   product- price ratio and the composition of
mechanism to equilibrate demand and supply    demand for output.
for labour, the rate of growth of production    In these models unemployment, traceable
may well be exceeded by the exogenously       either to technical lack of substitution or to a
determined rate of growth of the (working)    rigid wage in excess of the level required for full
population. The result is an exponential rate of  employment, is sector non-specific. Thus, like
growth in labour unemployment. It was this    dlemand-deficiency models they do not appear
'naive' growth model that Solow rescued from  to be useful for explaining why in many LDCs
the inevitability of unemployment by arguing  uneiiiploy ment is exclusive to urban areas. A
that the choice of technique, the capital-    more fundamental issue is their ability to
output ratio, could shift in response to a    explain open unemployment. The question is
growing availability of labour, as could the  not the pervasiveness of wage employment
savings ratio.72                              fixed coefficients or a minimum rate of re-
Fixitvystickiness ov real wvages. The simplest  muneration maintained by custom could con-
model in this category is the one-sector model,  ceivably characterize non-wage sectors  but
such as that set forth by Meade to analyse the  whether the models can explain why employed
dilemma faced by the monocrop economy of      workers offer, and the workers in excess supply
Mauritius:73 simultaneous dramatic increases in  prefer, direct transfers to work-sharing arrange-
the growth rates of both labour force and     ments as a means of support. Certainly work-
wages. By 1960, past demographic changes had  sharing and  employment creation are not
ensured a 50D' increase in the working popula-  incompatible with an excess labour supply.
tion over the subsequent 1 5 years.           Both sets of models imply that there is a given
A   classical competitive economy which    stock of man-hours available for employment,
maintained full employment of labour should   and demand sufficient to absorb only part of
respond to this relative abundance of labour  that stock; both are silent as to the number of
with rises in the rent of land and the rates of  workers mneeting the  deemand. The excess
profit and interest, and a fall in the real wage  supply of labour could just as easily be mani-
rate. But given a highly skewed distribution of  fested in underemployment in sectors where
land ownership and capital such trends would  coefficients are not rigid or in dlisguised un-
further skew the distribution of income. Al-  employment, workers employed at a wage in
thiough from the perspective of resource alloca-  excess of their marginal product, as in open
tion the wage rate ought to be very low, the  unemployment.
'political awakening  of the   underdog  in     Albstracting from the utility implications of
Mauritius has not unnaturally been associated  eleencGsynary behaviour, from the perspect ixe of
with aggressive trade union action, which has  the .emiployed the distinction between sharing
pushed up the wage rate in the sugar industry as  work and transferring income is that in the
a method of redlistributiing part of the wealth of  latter case they get nothing in return for the
the  island'. Meade predicted   that, unless  income foregone. In the former case, there is a
remedial policies were adopted, unemployment  gain of leisure. Where employment creation is
would result. With a given supply of labour and  feasible there is a gain in output as well as
the marginal product of labour less than the  leisure. This gives rise to the presumption that,
fixed real wage it is clear that capitalist   ceteris paribus, the employed with an obliga-
emiiployers will hire only a proportion of the  tion to support workers in excess supply will
labour force.                                 prefer work-sharing or employment creation to
The deduction of unemployment is much      transfers. There is no incelntive for direct
the same in the two-sector models developed by  transfers such as a commitment by the un-
international trade theorists.74 Brecher added a  employed to make repayment  possibly witlh
(real) wage floor covering the entire labour  interest to oUtweigh the output and leisure
market   to  the  standard, two-commodity     benefits.
(capital-initen-sive, labour-intensive) two-factor  None of the models we have considered
model and derived the feasible (constrained)  provides a basis for such an incentive, because
production possibility curve. Capital is always  in none of them  does openi unemployment
fully utilized, but for a given product-price  offer the prospect of an economic return
ratio there is a locus of profit-maximizing   directly to the job seeker. They are all demand
output equilibria (a Rybczynski line) along   deficiency models in the limited sense that in
which labour is unemployed. This model shows  the economy as a whole, the number of hours
that unemploynment will vary not only with the  offered for work by the labour force, no matter



1 216                             WxORI 1) 11 ' I ()IN1M I N I
how deployed, exceeds the number demanded      labour will reflect the agricultural wage.
by enmplovers. Where there is an aggregate        Unemployment is sector-specific in the
scarcity  of employment opportunities, the     Lewis   model, thougli the    wage   floor is
unemployed worker is best viewed as a vietim   economny-wide, because it is in agriculture that
of circLumstances who would prefer even part-  excess workers cani obtain the mlieans to subsist,
time work at the prevailing wage rate. Thus he  Lewis' view, similar to our own, andl apparently
would have no incentive to offer employed      based on the same reasoning,. is that workers
workers 'willing to share their work' a propor-  will be in disguiised. rather thani open tinem-
tion of future earnings in exclhange for direct  ployinent. When laced witlh the obligation of
transfers.                                     supporting workers, it is to the advantage of
It would be wrong to conclude from this     peasant farmers to cinplop them at the institu-
discussion, however, that deficiencies of aggre-  tional wage rather than simply provide them
gate demand for output, lack of suLstitlltion  with direct transFers. In this way the larmiier
between factors, or wage rigiditie:. do not cause  reduces the cost of support by the value of hlis
labour allocation andi utilization and incomne  Ladditional leisure, and, given a neoclassical
distribution problems in developing countries,  production  function, by  the value  of the
Our point is simply that problems of this nature  worker's output. Even it' the marginal produc-
are unlikely to be manifested in open unemn-   tivity of labour is zero, the farmer should prefer
ployment. When open unemployment is the        to subsidize the worker in eniployinent rather
symptom ibeing diagnosed other miiodels are    than in unemploymnent. Only if the inarginal
iimore likely to provide the appropriate anulytic  praduct of labour is negative or increased
framework. Hlowever, in the event that the     leisure has negative utility would the excess
stringent cunditions for technological lack of  supply  of labour be   nianilesledl in  open
substitution or rigid wages to lead to open     uuc.ili   eut As Sen emlphasized, 'there is no
unemployment were fulfilled, the output fore-  conitradiction het wcen diguise(d unemploynIent
gone per man-year of u nenmplo% mleint would be  and  rationial  behaviour. In a li. n ily-hased
less than in a situatizn of Keynesian urnellliu p, -  peasant economyIiy, uii iEplo% Imlent will natui rall
ment. Indeed, if the technical and wage rigidi-  put on this disguise. A piece of land that can be
ties are accepted as given, the opportunitv cost  cultivated fully by two, iniy actudlly be looked
of labour and hence agglregate resource costs are  after by four, if a fanmily of four working men
zero.                                          ha., '.g  no  other employnment opportunity
ehappens to own it'." 6
mTioug, -wage employees may be constimaimied
Divergietce  between  wvage  andl mnarginal  from  work-slharing. presumably  because of
prodtict. In his seminal papiper, Arthur Lewis was  fixedt hLours of work, Lewis dloes cite monopo-
concerned nmore with the consequenieN ot a      listic comrpetition aunlomig the urban self-em-
supply of labour, perfectly elastic at a conhtant  ployed  and social cmnisciousness aniong em-
real \vagQe, to the modern capitalist sector than  ployers, who hire other'.ise destitute workers
with the cause ot the iinemploy nent in the    whose contribution to output does not cover
agricultural sector which it suggested.7 5 Never-  their wage, as means by which some of the excess
theless. a model is implicit in his discuss,ion. It  supply may be absorbed in the capitalist sector.
is virtually the same as the two-sector model in  If these phenomena assumed significanit propor-
whichii nemiploynient. is the   result of an   tions it would, of course, qualify our cm.ssimtica-
eVinnon1y-%%ide rigid wage in excess of the     tion of divergence betkeen wage and illrginal
niam ket Jleadrig level. Ihe wage in the agricul-  product imiodels as sector-specific. Bult the
tural sector is exogenous to supp,rly and demanaid  p'loints to einphuasie are that, given the assto mop-
whether it equals the average product of the  tIions regarding 1e.s,i,t behaviour in this type
peasant farmiier, as Lew.'is  gge Is. a         otbsis-  of model, uLm)plus laIbour will  crsi be ct-onf'itned
tence' level of conso UptiOl or a cuistomnary   to the urban sct lor. Ior is it likely to be
level of co nsuL mp) tinl is nOtt Of torlida,cnit:il  manifested in open onloy men1
importance   to  the  analysis. In the  'over-    Segmcileltio,n of th1e latlbottr mnarket. There is
populated' economies Lewis had in iiind, these  also a rigid wage in the well-known Harris
interactions would lprodLce a /eio or near zero  olodaro model, but unlilike that of the Lewis
return to labour becLause of the abundMIL1ance: of  model it is not economy wide. Rather, the
labour relative to capital and natural resources.  labour imarkel is segnmentid in the sense noted
Since the modern capitalist sector must draw its  above, that some workers receive higher real
%%orkers from the agricultuiral sector it too is  wages, net of psychic and direct costs of
subier t to the sam111e floor. The supply price of  mlobility, than others with the same level of



LABOUR MARKET PERI ()10X1NCi IN DLVI I OPING ('OUNTRIES              1 217
human capital, simply by virtue of their sector  limaitedI to  a  specific  sector. Yet because
of employment,                                  employment is available elsewhere, he must
Segmentation, however, is no more a suffi-   have voluntarily limited himself.77 With regard
cient condition for open unemployment than is   to subsistence, the worker believes that the
an economy-wide rigid wage. Inideed, it is not  present value of his future income stream is
even a sufficient condition for the derivation of  greater if he doesn't work than if he took a
an excess supply of labour either in the sense of  low-paying job. Because of the prospect of high
an aggregate imibalaiice in the supply of and   wage jobs, the worker can offer those whio
demand    for iman-hours  (as distinct from     support him a return in direct transfers. They
workers) or in the sense of an intra-sectoral gap  expect greater benefits from this arrangement
between the wage and the marginal product of    than they   would get through work-sharing
labour. As long as there is a sector in wlhich  (which would allow    themn more leisure) or
entiy is free, the conlse(qUence of raising the  emlploymiient creation (wwhich would increase
fixed wage to a level above the laissez-faire   output),
equilibrium is simiiply a decline in the flexible  The model is also consistent witlh our other
wage (mpl) to a level at which all the workers  stylized fact of the coexistence in different
who could not find jobs in tlle high wage sector  sectors of labour scarcity and labour surplus. lt
can be absorbed in employment. There is excess  could be mo(lified to incoiporate rural surplus
supply to the high wage sector in the limited   labour in the Lewis sense, by positing a gap
sense that workers employed in the flexible     between a rural wage reflecting incomne-sharinig
wage sector prefer a job in the rigid wage      rather than profit-maximizing beehaviour and
sector, thus the supply curive to the latter is  tlle urban wvage, But clearly the model is, in its
infinitely elastic,                             originall fornm specified so that there is full
The essential contribution o1 the Harris     eimploymcint o0 labour scarcity (job vacancies at
Todaro model to the explanationi of 0memplloy-  the imiarket wage) in the rural sector and
ment is a labour allocation meclhanismi undJer  wuienuploynent in the urbani sector.
which actual wages are not equalized, but the     The model was formulated in hast Africa; it
actual rural wage is e(quated with the expected  is consistent with otlher features of tlle evolu-
urban wage, the latter defined as the (rigid)  tion of the labour inarket there. Urban wages
miinimum wage weighted by the rate of employ-   have not declined in response to the pressure of
ment (i.e. the share of persons in this labour  the excess supply of labour that emerged in the
market who are employed). The flarris -Todaro   1960s and led to rising unemployment rates,
model thus explicitly resolves the inconsistency  Rather, goernmnent interventions and apparentl-
between the predictions of the micro-economi,  ly. in some indListlies, strong positive relation-
theory of the labour market as to the form in  ships between   wage and productivity levels
whiclh an excess supply of labour will manifest  contributed  to their continued rise, which
itself,  and   conventiorfal  macro-econiomic   resulted in 1970 in a rural- urlban inicone
explanations for open unemployment. This        differential for unskilled labour estinmated at
inconsisteucy is resolved through provision of  between 1:4 and 1:8.78 The contributioni of
an incentive to workers to remain in open       migration to aggregate urban labour supply -
unemploymnent rather than accept a job readily  iimigrants comiiprise the majority of the urban
available at a lower wage. The incentive derives  labour force and of its annual increment  also
from the implicit assumptions that employment   suggests a link between migration and urban
and rural residence on the one hand and seeking  labour talauice. Recently the theory has been
an  urban job on the other are mutually         put to more severe tests. Carefully designed
exclusive activities. That is to say, workers   econometric estimlates of nmigration functions
employed in the rural sector have no chance of  have verified its central hypothesis: that the
obtaining a high wage job     or, e.onversely,  probability  of urban en ploymlerit, inrdelpemi-
only those excess work-ers in open unemploy-    dently of the difl'erences in actual rural and
ment have a chance of obtainling a job in the   urban wages, contributes significantly to the
rigid wage sector,                              explanation of variance amuong time periods and
The unemployed worker is not a victirn of    siub-grouips of the rural population in rates of
circumstances, as he is in mondels cliaracterized  urban mnigration.79
by  an aggregate imbalance between labour         One a:pparent weakness of the model is its
supply  and  demand. Rather, this type of       tendency in some couniitries to predict rates of
urernployrnent is a hybrid of 'voluntary' and   unemnployment    significantly  higher  than
'inivoluntary' elements. It would be involuntary  measured rates, but one can increase its realism
if it were taken for grantedL that the worker is  by tinkering with the non-price iumecharmismii by



1 218                              WORLD DVlFl0P1MENT
which  high  wage jobs are rationed among      decline in total employment during the 1950s,
competing job-seekers.80 With a labour supply  and attributes these plhieromiiena in turn tW
function based on expected income, there will  risirng. wages anid  contseuenttt latabour-saving
be open unemploymenit in a segmennlted labour  innovations.83  Ihis explanation, akin to thie
market in equilibrium as long as, ceteris pani-  economy-wide  rigid  wage models, lhowever,
bus, a worker without a job has a higher       does not p,rovide answers to the tjuestions of
probability of obtaining a high wage job than  wly, if inadeqtaute labour deimind is the prob-
does a worker with a lowv wage job. Harris and  lem, both Puerto Rico and Jamaica experiellced
Todaro's assunmption that the probability of an  labour shortages in somiie sectors and why in
employed worker obtaining a high wage job is   Jaiiiaica, where employment rose at a taster
zero is both unrealistic and unnecessary. Simi-  rate, unenmployinent remiiinled higher than in
larly the absolute constraint that rural residence  Puerto Rico.
places on urban job search in the Harris-         In the inter-sectoral misallocation nmodel, by
Todaro model is not e>ential. As lonig as the  conitrast, inemploynent is a functioni of an
probability of obtaining a high wage job is    eaecssive movemnent of labotur from low to high
higher for urban than rural residents, in equili-  wage sectors and not to an aggregate imbalance
brium there will be urban open unemploymenit.  in supply and denmand. so it is consistent with
The implications of relaxing the constraint of  the coexistence of pockets of super-Alundanlce
employment on job search are particularly      and scarcity.84 One way to assess the ielevanceof
interesting. Mazumdar and Sabot have shown     the model is to determine whether unemployed
that if workers in an urban flexible wage sector  workers would accept a jobl at the wage level
have a better chance of obtaining a high wage  prevailing in the low wage sector wvhere employ-
job than rural workers, thouglI not as good as  ment is uvail.ible. In Jlamaica, surveys n1dicateed
the unem]ployeT, then sonie of the workers     that in 1)55 over tour-fifths and in 1966  67
comprising the excess supply to the hiigh wage  over two-thir(ds of job) seekers would not azccept
sector, who o therwise vould select Lineiploy-  work at the wage rate pre%ailing in agricul-
ment, enter the urban free entry sector and    ture.8 s As expected. unenliplo% irent and thie
depress its wage to an equilibrium level llelow  uncertuini prospect of a hiigh wage job was
the rural wage.81 In this case the excess supply  preferred to the :et ta in iprospeL t of a low wage
of labour to the high wage sector assumiies three  job. The model also can explaini why uinenmploy-
formus: hidden rural unemployment, in the      mlent in Jamaica remained higher than in Puerto
sense of workers employed in the rural sector   Rico, despite a more rapid increase in employ-
willing and qualified to work at the rigid wage,  ment opportuniLies. For a giver initer-sectoral
but discouraged fromn seeking such work b' the  income diflferential. the hiigher the rate at which
(:'sts of doing so, urban undercinplox\ neit in  workers are hiired in the hiiglh wage sector the
the limited sense that the transfer of workers  greiter the unemployment rate must be to
fro m the urban flexible wage to the rural sector  equalize expected wages in the two sectors.
would increase national output; and open urban  Finally, the model can explain tle' relativel)
unenmployment.                                 low rate of uinemplluy iient in lIaiti, since for a
This imo0del Of 111nelplolyub nwn111 is not con-  given wage gap and jobt opening rate the larger
fined to intcraction-l between rutral and urban  the high wage sector as a p.r(olportion of total
segmnents of the labour market. It has been    eniploymlenit, at least up to a point, the greater
ap plied in .amfnica . where the wUage gap of   thle level (If uneniplo' mlL-Vtl as a  rprt'0 ionll of
smgnrli lmnce (for a comiiparable level of skill) is  tlhe national laboiir torce.
between  the spatially sep:ira ted ininiing and   Silnce .ori ph  n m   odit imons wiltiri the
s`Lgar'CnL- industries whlich are boith rural. It  urban sector are theinselveo diierse. migration is
resolves the pueile of why the unemoploy neiet  an inessential feature of this g,emle of sementa-
rate is higher tIiere andL in Puerto Rico mind  tion miiodels. In somie countries mliigrat ion is
Trinlidadt  also aelativcly dyniamiic ec oiiies  comist raineil  by dlirect Lo tinos,. in Asia and
than in   la,iti wlich is relalively poor and   Latin Anmericani contexts niigr.lm ts ficluently
economically  stagnant.82   In  Jamaica  and    uinminrism a relatively siim:ll proportion of the
Puerto Rico uneimployineint l iained at hiigh   annuial in:remnient to the urban labour f'orce,
levels llllogllolt Lte 1 SOs and 1 ')(0t),, despite  cithlr be- muse of a relatively high degree of
rapid output growth and, as a consequence of    111lhalniiat i-un 11nId or a narro%  Lilfterential in
emii,r:ition. virtually  stable labour  forces.  incomes between rural anld mil.in areas. Never-
Reynolds exlplains this phen.-1om0enon in Puerto  LtIeless>, given rigidity ol sonie urban wages, as
Rico by ieterence to the low output elasticity  long as the prob:ibilmty at a giveni skill level of
of labour deimind in indiustry that resulted in a  obfiinimvin  a high wage job is greater for those in



LABOUR IARKET 1[ RI ()IRIAN('T IN DIFVI. I OPING COTUNT RII S       1219
unemployment than employment, open unem-     functions in various occupations and a speci-
ployment is the natural result of individuals'  fication of the precise benchmark situation
allocating themselves on the basis of expected  against which the actual allocation of labour is
private economic returns. For the sector in the  measured. If the existing wage rigidities are
aggregate there may be no imbalance between  taken as part of that specification, then it is
the supply and demand for labour, but workers  logical to assume that the unemployed would
will refuse the offer of a low wage job if the  have to be absorbed in low wage activities;
expected gain, in ternms of the present dis-  uniless labour's marginal prodluct were zero in
counted value of expected income from waiting  those activities, an increase in output would
as an unemployed worker in the queue for high  result. There could be a further increase in
wage jobs exceeds the costs - the income     output if the specification involved the highly
foregone  from  a low   wage job      while  trained currently unemployed replacing less
waiting.                                     qualified people, and the latter in turn replacing
One model that abstracts from rural urban  people less qualified than they and so on, But
migration focuses on those changes in the    since in the segmentation model the presullp-
linkage between education and occupational   tion is that complementary factors are fully
levels that are a consequence of rapid expansion  employed, the cost of a given amount of
of the educational system. It provides an    unemployment will be less than in a situation
economic explanation for the unemployment    where Keynesian deficiency of demand prevails.
of educated workers that does not rely on lags  Thus, unless the increase in employment were
in information, rigid expectations or preference  to occur in activities with high marginal pro-
structures biased  against certain types of  duct of labour, which by assunmption it would
work.86 In most LDCs the supply of educa-    not, the increase in national output resultinlg
tional opportunities at any given level is not  from a shift to full employment would be less
determined  by  demand in the occupations    than the ratio of full employment to the
entered by previous school-leavers of that level.  prevailin, level of cnmp)loynient. IndleeL. there
When, as has often occurred in the past two  would be ircLumsta11ceC in which the realloca-
decades, the school system trains more workers  tion of labour among various segments of the
than these occupations can absorb and if wages  labour market would have little if any impact
are rigid, school-leavers are faced with the  on aggregate output. Where unemploymnent is
choice of 'queuing' for a job in the preferred  the result of a distortion in the rural urban
occupation  or of accepting a less-preferred  income differential, for example, the social
(lower wage) job. For some workers expected  opportunity cost of leaving an urban labour
income will be higher in unemployment than in  force participant idle is given by the marginal
relatively low  wage emllploymnelt. Under the  product of labour in the rural areas; if it is
extreme assumption that employment and job   negligible, the resource costs of urban un-
search are mutually exclusive activities, the  employment are insignificant. Even in East
unemlploymnent  of  educated  workers  will  Africa where rural marginal product is clearly
increase and the probability of findirng a pre-  positive, the degree of urbanization is so low
ferred job will decline until the expected wage  that a 10'.; urban unemployment rate translates
in those jobs is reduced to the wage level in the  into a national unemiploymenit rate of under
less-preferred jobs. As in the sinmple segmenta-  2%,, which implies an even lower proportionate
tion model relaxing the assumption regarding  loss of national OlutpuIt. These back of the
the relationship between enmlployment and the  envelope assessments of the magnitude of the
efficiency of job search reduces the etiuilibriumin  resource costs of imneniployment are consistent
rate of unemployment. The filtering down of  with numerous studies, in botlh industrialized
workers with relatively high levels of eduication.  and  developing econonmies, of the cost of
in effect displacing workers with less education  diistortions in botlh factor and product markets.
from  lower level occupations, sets off an   Leibenstein notes 7 studLlies where the cot of'
action reactioni sequence that can gencrate  misallocatioins is less than lt  o  national
unen-iploymient in each occupational segienit of  income.87
the labour market where wages are sorrewhat     The segnmentation models to which we have
rigid downwards.                             referred are search models only in the limiited
For that open   unemployment in LDC's      sense that difierent.als in the probability of
which is the consequence of a rational response  obtaining a high wage job associated with
by workers to labour market segnmentation, the  geographic lot.ation and employment that give
measurement of resource costs involves both a  rise to unemployment in situations of labour
knowledge of the marginal pr-oduict of labour  market segnmentation  are attributed to con-



1220                                VOR I D DVV I. LOPNIMNT
straints on the efficiency of job search. But   use to which it is assumed such people would
search is no more than waiting in a job queue   be put. If the wage rigidities which appear to be
and the constraints appear to be simply on the  a major causal factor are taken as giveni, the
time available for this activity. There is an   otutput that would result from  their employ-
implicit assumption that workers know with      ment   - in  sectors where work is readily
certainty not only the structure of wages and   available -- is unlikely to add more than 1 2''
the job opening rate, but the number of         to national income. This does not imply that
competing job seekers as well. In search models  open unemnploymei,nt is not a social problemii
developed for labour markets in indLlstrialized  worthy of governmelnL remeldial policy. The
countries unemployment is not the result of     poorer the country the less it can aflord to
segmentation, identified with wage rigidities.  ignore an  opportunity to increase national
Rather it results because workers are searching  output without increasing the stock of produc-
for rather than in possession of certain know-  tive resources. Unfortunately, however, it is
ledge regarding the labour market. The search   questionable whether there are policies at the
or learning process is modelled as a process of  disposal of planners that can substanitially
gathering samples of job offers, not in terms of  reduce  unemployment    without   imposing
queuing. Since all wages are generally assumned  countervailing  political or economic costs.
to be flexible, in the absence of uncertainties a  Eliminating the wage distortions that attract
Walrasian full employment equilibrium would     workers into some sectors in excess of employ-
prevail.8 8                                     ment opportunities is the atiost obvious can-
The modificition of these models to apply    didate.91 An additional gain in efficiency to
in developing couiltries holds some promise for  that associated with the reallocation of labour
enriching our understanding of labour market    among sector-s would result from     investors
dynamics and unemploymecnt. particUlarly in     selecting technology  on the basis of more
relation to structural change, but the process  appropriate factor prices. And tlhe elimination
has just begun.89 It is their implications for the  of wage (distortions would contribute to the
resource costs of unemployment wvith which we   achievenment of a distributional goal widely
are concerned. If all uncnpioymcrnt were of the  accepted even in states witlhot any pretension's
search rather than the queuing variety, then the  to socialism, namely that workers with the
unemployed would in fact be self-employed in   'same' level of lhuman capital doing the 'same'
information   collection, and  under certain    work should receive the 'same' pay. While the
conditions such labour would be optim-ally      political cost of reducing the relative economic
allocated,90 Seeking information is a necessary  advantage  of groups with   disproportionate
activity if labour is to be transferred from less  influence is likely to be the operative factor,
to  more   productive  uses, More generally,    tlhere are several economic reasons why altera-
however, there is no presumptioni that the      tion of the wage structure may prove unaccept-
optimal share of a labour force will be engaged  able.
in search, and that of this share an optimnal     Wage cliffs may     be the   r esult. not of
proportion will be searching full time. i.e. in the  arhitrary governmnernt policies or of the pressure
unemployment pool. The efficiency of the        of unions, but of enmployers in some sectors
process depends, among other tlhings, on the    finding that they can reduce labour costs per
accuracy of individuals' guesses as to how much  unit of output by !aising xvwages. The positive
time they should spend searching in order to   relationshlip between wages and productivity
maximize their own expected income stream      that gives rise to this apparently paradoxical
and  on   the way   different people's search   situation may have its root-. in mnotivational
procesies alfect each other. Unitil the nature of  faictors or in a negative relationship in some
the search process is better understood, it will  sectors bet Wee1n wages and la;biour IIII nover and
not be   possible to iuige whether scarclh      hence costs of hiring and training workers.92
unemployment has a fairly high cost, no cost,   Thus elimiinatinig wge d(iSl ion110s IIILN' increase
or whetlher it should optimially be even hiigher  allocative efficiency but as we noted above,
than it is. The fact that some LDC uinemploy-   may also reduce X.efficiency, Fuirthermlore the
ment clearly has a search aspect to it does,    eliminltion of dlistortion in the wage structure
certainly, create some presunmption that its cost  mayl have adverse as well as beneficial dis-
is less than might otherwise appear.            tributional consequences. If, as is likely in the
The general conclusion of this discussion is  urban  sector, the elasticity of substitution
that the resource costs of nmaintaining a pool of  between capilal and labour is less than unity,
openly urneniployed workers are not likely to  the share of wages in total incomlie will be
be very high, althoULghl this does depen(d on the  redutced and the urban wage bill may also be



LABOUR NIARKET PlRI 01RN1A\NCL IN D)LA'1 1'ING; (COLN rRIlIS           1221
reduced if increases in output occasioned by   segmented labour market, since employment
the decline in urban wages do not offset the   opportunities are rationed by means other than
reduction of the wage share.9 3 9 4            the adiustment of wages, those who succeed in
The   discussion  of  distribution  suggests  finding a job overfuilfil their expectations while
another perspective  on   the seriousness of   the unsuCCessfUl underfulfil theirs.
unemployment as a social problem. The loss of     Nevertlheless, the limited evidence available
output associated with unemployment has an     on the characteristics of the openly unem-
obverse side in loss of income; the existence of  ployedi suggests that welfare costs are lower
unemployment    also  leads to   uncertainty,  than they would be in industrialized countries
uneven income streams, and the psychic costs   in the absence of a formal social security
of losing the direct utility derived from work -  system. This is not because the unemployed in
what Sen    calls the recognition  aspect of   developing countries have a greater store of
employnment.9 5 The loss of satisfaction through  assets on which they can'draw. Rather depriva-
income loss and the other costs of unemploy-   tion is limited because, on the one hand, the
ment is by no means evenly distributed. Even in  subjective costs are distributed beyond the
the industrialized countries, where the resource  persons without jobs to those with jobs by
costs of unemployment are demonstrably high,   means of intra-family transfers and, on the
the remedir.l policies, particularly those which  other, because sectors in which incomes are
focus directly on the workers without jobs, are  flexible are available as employers of last resort.
frequently justified by reference to the distribu-  In two contexts that in many other respects are
tion of welfare costs. These costs are generally  dramatically different, detailed studies of the
concentrated among oroups for whom the loss    unemployed support these assertions, In botlh
of income will have s.gnificantly adverse con-  Colombia97   and  Tanzania98   the pool of
sequences  on   consumption, namely     those  unemployed    comprises predominantly    non-
without much accumulated wealth, but with      heads of houselhold, the younig and married
family xesponsibilities. The relevant questions  women who do not bear economic responsi-
in  developing coLuntries are: who are the     bilit) for others and who have a high degree of
unemployed, and to what extent are they able   access to transfer payments from   parents or
to live off their own savings or assistance from  husbands. They can and frequently do renmain
other persons?   As Pigou   enmphasized, the   unemployed for long pericods without suffering
welfare costs are much higlher if it means that  extreme deprivation. Most of the unemnployed
those without jobs 'are reduced to the verge of  are eduicated, which places their parents in the
starvation than if it means merely that the    top half or third of the incorme distributiOnl.
superfluities of some rich men are cut a little to  Thlouglh the evidence is sketclhy it does appear
finance the unemployed. 9 6                    that the unemiployed who are heads of house-
In situations where open unemlployment is    hold and thus likely to have more limited access
best explained  by  segminentation (i.e. wage  to intra-family or other transfer payments, only
rigidity) or search models, it is 'voluntary' in  remain unemployed  for short periods. In a
the sense that there are enmployment oppor-    situation of labour market segmiientation, when
tunities available which the worker declines to  unemployment threatens with deprivatioll an
accept. This undoubtedly has an influence on   individual who has exihaLlstedi his savings and the
the  welfare costs of open unemployment.       generosity of his relatives, he can usually find
However, the extreme argumnlent that voluntari-  employment in the sectors where wages are
ness should be eqluiated with zero welfare costs  IIexible. The comiposition of unenmploymnent
for the society, implying that the composition  observed in ('olombia and Tanzania appears
of the unemployed is irrelevant to the assess-  characteristic of other developing countries,99
ment of welfare costs, is fallacious because it  suiggesting that beiniig With0oUt work is a luxu1ry
confuses ex ante and ex post perspectives of job  only a small proportion of the laboUr rorce can
search. Presumably for all job seekers ex ante,  afford for longer than several niontlis at a timiie.
the expected incomle associated witlh a period    Our under-standing of the causes of opeen
of open unemployinent is greater than expected  unemployment remains rudimentary and the
income in the low     inicomce sectors where   data to assess the number and character of
employrment is available. Individual decisions  unenlploved workers are unavailable in many
are based on this view of the labour market.   developinig counitries. Nevertheless, the weight
However, economic welfare is assessed     by   of existing evidence fails to support the conten-
observing actual income or consuimption, not   tion that increasing Unemnploynicrit in m1.ny
by consideration of alternatives that become   couintries over the last two decades demon-
irrelevant once a decision is reached. In a    strates that the   development pnrocess   has



1222                                WORLD DI.VE LOPNI I Nr
strained  the  allocative  mechanisnms of the       tural labour and the condition of ceteris parihtus,
labour market beyond their ability to cope. But     which hias been adopted by I cibenstein, Viner,
the case that labour market failure has con-        Rosenstein-Rodan, and many odhers.
strained the rate of economic growth and is to      W. A. Lewis' seininal article on (levelopmiient
blame for the persistence of large pockets of    under   labour   surplus,103  and   'ulhseqluent
extreme poverty does not rest only on the        elaborations especially by Fei and Ranis' 04
prevalence and social cost of open unemploy-     heightened   the interest in this issue1l0  by
ment but also on the extent and costs of         providing  a theoretic;l   fraineworkl for the
various forms of disguised unemployment and      phenomenon, and contributed to a wave of
underemploynment. t-o which we now turn.         research   directed   toward   more   adequate
empirical measurement. The essential theoreti-
cal proposition was that, because of income
(b) Disguised unIenplo,vinent           sharing arrangements in family enterprises, the
marginal product of labour in the traditinnal
In sectors where family production units      sector could fall below the going wage in the
predominate, unemploymnent is more likely to     'mnodern' sector, which was characterized by
be disguised than open. Population census data   profit maximizing   prodluction  units. In one
indicate that the share of the labour force      variant of the model, it was assumed that as
classed as paid workers, which ranges from       long as individuals were free to choose whether
around   10%  up to   60--70; or more even       to take a modern sector job, labour would
among LDCs, is usually lower in agriculture      move from the traditional to the modern sector
than in non-agriculture, being virtually negli-  only if its miarginal product in the latter (i.e.
gible in the agricultural sector of Tanzania, the  the wage) was at least as great as its average
Ivory Coast, and Thailand and over half in a     product in the former (average product would
few  countries (e.g. Algeria. Chiile, Argentina);  eqlual incoLm0e if thiere were equal sharing of
for those developing countries on which the      income among all family wtorkers in the tradi-
ILO provides data, the median ratio is abl)ut    tional sector production unllits). The difflerential
.        Since virtually all the early (quantita-  in the iiarginal product of labour between the
tive discussions of labour surplus in    LDCs    two sectors implies that labour is not optimally
related to the agricultural sector, they focuissed  allocated; the opportunity cost of labour to the
on the phenomenon of disguised unemiiploy-       modern sector is low or even zero, so that total
ment, the    first estihriates  of  which  were  income    (including   non-monetary    benefits
strikingly high.'° 1 To quote Kao, Anschel and   related to preferences for living, on the farm) in
Eicher:                                          rural areas would    rise if workers could be
Studies by Buck, Warriner, and Rosentein-Rodan  ';l'ttetl to the modern sector. To6 few workers
in the 1930s and 1940s in less developed countries  nmove because doing so involves losing a share of
presented statistical data for China and Soutlh-  the surpluŁs  - rent or profits  generated by
eastern 1 urope to supgest that a large percenitaze of  family enterprises, This model assumes workers
agricultural labour was idle for substantial periods  are  responsive  to  income  differentials; the
of the year. . . . D)oreen N%arriner...in 1939... in a  misallocation  of  labour  results  from  the
widely quoted study revealed that before World  diflferenit possibilities of incomiie sharing accord-
War It in 'ILastern lurope as a wlhole, one-quarter  ing to whether the individual remains on the
to one-i bird of' tile tarml p1odiuLu ion is surpluts.  fainmily farm or moves to the city.
Ne\t, in 1943, Rosenstein-Rodan wrote that 20 to  To  demonstrate the presence of surplus
25 million of the 100 to 110 million people in  a urucoiivdwemstlow                  htte
l astern and Souiheastern Europe were eithier  labour son   ondceived  we must show  that thle
-whllt) 5  or partially  unemployed. In  1945,  marginal product of lahou r ( NI PL) is low
\tandelbanin estitniated that l'rumn 20 to 27t' of tlle  relative to that found elsewhere in the system
active rural workers in (ireece, Y*ugoslavia, Poland,  or, in cases where few htours are worked, that
Hungary, Rumania, and B3ulgaria were redun-   MPL would be low       withi a smiiall additional
dant. . the widely quoted 1951 U nitved Nations  inl)Lit of hours worked. We muist also shIow that
report by a .:rtoiup ot' c\perts fil,ludint .W. Arthur  income sharing arrangements or other harriers
l.ewiss. T. W* Schuti.. and 1),R. K  ,laldil cited1 liest  to niohility not involving real costs are the
studies and added that it seems 'safe to assume f  reas r
tfor many regions of India and Pakistan, and f Ior  r.a tn[I for persons not working elsewlhere. If
,ertain parts ot' the Philippines and Indonesia, the  the  conicern is sinply' to  demonstrate low
* urplus (tural population ) cannot be less than the  utilization of lab out in some dhsolute sense (ie.
pre-war average for tile last Luropean Region'.  not against the yardstick of miore productive
The experts advanced this definition of disguised  opportunitiesi) then  only low  MI'L (vi.si-t'iis
unemployment: zero marginal produict of agricul-  average incoime, for exaimplc) or low  hours



1. ABOL R IARKET P1 RI ORMA\N(CI IN DIVI I (0I NG(; CO()N IRtIS         1 223
worked are relevant. It is very difficult to find a  (i) Measuremen t of labour surplus ill tle
measure reasonably indicative of how    much    tn-aditionial sector
additional productive work would be forth-          Following the initial attemipts to test for
coming under various conditioni-s (e.g. wage     surplus labour, a number of the original pro-
rates) or of how    much welfare gain wouki ldonents of the concept either clhanged their
result.                                         miinds regardinig its validity, scaled down their
Empirical work has generallyt shown that the  estimates  of   disguised  unemploymnent    or
estimates of 20 40%', excess labour in the       m1oditfied their definitions of the concept. I'he
agricultural sector were too high by most        review by Kao, Aniscliel and I ichler concludLed
interesting definitions: the agricultural labour  that using the origiinal definiLion  zero MPL
force in some countries works more than was      and ceteris parihits (i.e. no increase in capital or
initially  thought, including  non-agricultural  other reorganiizationi in traditional agr-icuLlt urel
activities which sometimes take up a fair share    there was no serious evidenlce of more tllan
of total time available. Currently the issue     about 5%   disguised unemployment anywhlere.
might more aptly be phrased in terms of          Since that time, additional andt more detailed
whether the relevant surpluis (i.e. the share of  studies attempting to measure labour sUrplus
the agricultural labour force whose produictivity  liave  been  undertak-en, and  there is more
would be markedly higher in another sector) is  information on labour utilization in general,
closer to zero or to say 10-15%(. Such 'single   especially for the non-agricultl'l.: sector. Since
figure' measures are, however, misleading. An    the proposition has incasing\ been put for-
adequate specitication of labour supply dis-     ward that much disguisCd unemnploymlent is
tinguishes the different amounts of labour       now  found in urban areas, especial)l in the
offered for different possible tises, and explains  service sector (e.g. retail commerce) it is im-
the reasons for those differncles. Soine of the  plortzant to take a new look at the evidence. WLe
factors determining labour supply to various     turn first to evidence on how    muucl people
possible uses are: preference for working in,    work in the traditional sector, then to evidlence
own enterprlise, other job preferences, costs of  on MPL (mnainly in aigicultiure). 3here are no
transport and lodging elsewhere, costs of chang-  accepted practical criteria for a separation of
ing place of residence. Surplus labour could     traditional and mlo(derni activities. thloughl it is
exist, but be usable only on the family's farm;  uIsually  presumed that econlomie uinits whiclh
thus if l)reference for work there is sitrong.  hire workers are modern in the l.ewis sense, as
more labour might be fortlhconming if more       are independent workers or small groupls of
complementary factors were available to those    professionals and otlher persons working in
farms or if land were redistributed away from    'moderni-type' activities. LUnits fitting neitlher of
large 'modern' farmers. This same labour might   these designations are coMsidered traditional
not be transferable to urban settings by normal  and, m11;asured by inputs of capital and other
market mechanisms.                               factors, are usually smaller.
Further complicating   the labour surplus        The   available information  suggests high
issue is the fact that maximizing labour use may  labour inputs by   persons attached   to the
not maximize output; if the nmarginal product    traditional sectors. Open unemnploylment rates
of capital is higher on large farms, using the   tend to be lower in agriculture, trade and
labour surplus by transferring capital to sinaller  services than  in  other sectors, presumably
farms could lower output, the possibility of an  because a smaller part of the labour force is
emnploymient output    contlict  has  received   usually enmployed for wages in these sectors.
increasing discussion during the past decade.    I)ata on lhours wvorked are seld10omi available for
The central thread   of the argumllents of    agric:ultture;  for  non-agricuiltulre  the  data
Nurkse, Lewis and Ranisr -Fei was that surplus   available from the ILO shows a det'inite decline
labour existed which could be transferrekd and   from low to Middle inColme counltrics. 10 6 More
whose use elsewhere would raise total output.    ietailed  information  on  hours   vorked  by
Potential output gains in the traditional sector  ze,:Wr, available for a smailler *et of comuntries,
and the related question of optimal distribution  indiCCtes that people tend to work longer lhours
of other factors between the modern and          in developing countries than in developed ones.
traditional sectors received little attention in    No general tendency for persons in agrictil-
this early literature, After reviewing iecent    ture to work longer or slhorter hours than in
evidence bearing on    the  extent of labour     non-agricL1lture applears foi the set of countries
surplus in the traditional Lewis sense, we turn  whose    detailed   informiationi  we    have
to this more general question of optilal factor  studied.1 0 7 In Colonihia ( 1970), persons in thie
allocation and its imiplications for productive  rural areas worked longer hours on average than
labour use.                                      urban dwellers:' 08 at the same time, a higher



1224                              WORLD DlsV I OPIMN I NT
share of the urban labour force worked ex-     Auiguist 1957 survey, the mean and median
tremely long hours (70 hr per week and up)     hoLirs worked by employed farm    household
Both these differences appear to be due to     members' 12 were 42.9 and 46.4 respectively
differences in occupation: the two groups witlh  and the percent lby which person hours would
many members working very long hours, sales    have risen had persons available for more work
persons and service workers, are found mainly  actually put in 40hr was 4.24%',. In South
in  urban  areas. Agricultuiral workers have   Korea, August 1964, the mean and median
medium   long  hours, while non-agricultural   hours worked    were 41.5 andL 44.8 respec-
workers along with professionals, teelnicians  tively, the  calculated  percelnt increase  in
and others foundL mainly in urban areas, work  person lhours was 3.63 and 5.96 respectively
shorter hours.                                 By   comnparison. in  mo)re developed  Japan
Short hours of work, like non-participation  (1956) these latter increuises were 0.97 and
in the labour force, may be due to leisure     2.07.' 13
preference  or the time required   for non-      Indication of a desire for more work is not a
economic responsibilities rather than to labour  yes--no issue, of course, and further wvork is
market imbalance. Just as open unemployment    needed  to judge   what respondents mean,
does not include all personi. who are without  especially wlhen they answer in the affirmative.
employment, so visible underemployment does    Somne would take only work in the immediate
not include all persons working short hours.   area; others would be willing to inigrate but
Whether individuals would prefer additional    might require a suibstantial wage premium to do
hours of work at their regular wage is the     so. In India,
relevant consideration, In LDCs where the
question has been asked, not too many have       it hlas ften bieen npoted that even the rural poor
said they would, and the percentage of total     rLportin) llderelnloYnlent andlor availahilit!' for
poteill   ,                                extra lil mllent hiave still not availed themiselves
potential labour hours their un1dereulloyment    fully or the einprlo io et supplied through public
represents appears mondest. As of an A\upgst     constrULtion  works near their villaeos. Some
1973 survey, a total of 14.2'; of emilployed     studies have used qtuestionnaires to probe into the
persons in the Philippines wanted additional     reasons belhind this aplpareiit inconsistency. The
work.'09 As expected, a higher share of people   25th round of NSS indicated that a very higih
working less than 40 hr wanted more work than    proportion of persons from small cultiv-ror and
of those working 40 or more hr: in August        labour houwoCliulds willing to leave village for work
1973  the two shares were 27.8 and 10.4%         where quite keen about employment in public
y  T e f e  vworks provided they were at suitable locations.
respectively. These  figures varied  markedly    The qualifying clause turns out to be rather
between  agriculture (22.1  ana 10.8';) and      important.' I4
non-agriculture (40.0 andL 8.8F)1 10 Ior all
categories, a smaller share of women working a  Other res-honTs sometimes mentioned are the
given number of hours wanted additionial work  short duration of such projects and low wages
than of mnen working the same nunber of           Most of the evidence cited above comes
hours. Of all people wanting additional work,  either from  countries with multiple cropping
70'J were already working 30 hr or more.       patterns or from surveys taken in the summer.
The extent of lost labour inputs rellected in  The sUbstantial seasonal element of agriculture
peoples desire to work more hlours cannot be  in the colder climates and other one-crop
calculated  without estimates of how   much    situations, raises the question of how fully the
more they want to work. Still, the evidence on  labour force keeps employed in the off-season,
the share of employed persons reporting a      Most studies in si. h couiintries have locused
desire for additional work does not suggest that  nmore on available tiite than on desire to work,
total hours would rise dramiatic;lly if their  On the basis of a set of assuniptions about a
preferences were satisf'ied. In the l'llilppines, if  nornmal wvorking year anrd needled labour inlputs
everyone workingi under 40 (48) hir and imdi-  in the house, C'ho esti oaied that abolUt 327' of
cating a desire to work imore had \\orked 40   the po[tntially av:ailahle time of family 'taron
(48) hr, the increase in lper.con hours would  work ers  went unused    in  Korea. I    For
have been about 4.09V, (7.48';) in May 1961    attached wage workers the share was 14.24';. A
and 2.30'' (3.977',) in November 1973.111      nunmber of studies reporting labour inputs to
Colombian figures for 1970, though not qluiite  gricultuti in Africa have shown surprisingly
compdiable. imply   a lower level than the     low levels of man hoots worked per year.' 1 6
Pllilippines in 1973.                          But Byerlee and   Licher note that there is
Siinilar informnationi exists for farm house-  evidence  from  several studies that seasonal
holds in heveral LDCs. In India's August 1956  labour bott.enecks limit future expansion of



I AB0 1t R \IARKI r P1 -RI )0R.NIANC  IN )l\'V I.VLOPIN( (uN'lRIIS      1 225
agricultural prodiuLtionl un1der existing tech-  on smlall farmiis witlh going wage rates as labour
nologies. They also observe that a considerable  surplus theory would sluggest, Buck showed
amount of 'leisure' time is actually spent on    that it was substantially lower on small than on
non-farimi economiiic activities such as crafts and  large farmls. Average incomrle per mlan equivalent
trading. as muclh as 50'e of -'.orking time imiav  was about half as high on snmall farms as on
be spent in these acti%itics.  In a sturvey of 3  large ones ( dependilng on just how the two
villages in Northern Nigeria, Normani f'ountd an  grotups are delined ). Thleir inferior resource
inverse relationship between farmii labour inputs  position led tfamilies on snmall farms to engage in
and off-farm labour ini puts, 18 suggestoing that  miiore off-farm  work and more non-agricul t1oral
off-farm  work is a means of salvaging labour    activities on the farmi; in fact they had less idle
time with a low opportuniity cost. Even thioughll  mionths per able-bodied   man than did large
seasonal labour peaks were a bottleneck to       farms 1 1.6  1 .8)122
agricultural expansion. farmers still spent 31";    In india, S. Mehra1 23 attempted to measure
of their time in the peak monith in off-farmii lhow   imuchl labour could     be   removed   in
employment. Norman specuIates that this might    1956  57 without lowering agricultural output,
correctl% retlect the opportunit'. cost of off-  beginninig with the UiSSU option that the largest
farm labour relative to farmi labour. 1parltiUlarl)  farmis. wvlere the number of days of farm work
since some activities such as tradiing are main-  peer person so emiployed per year was higher had
tained by farmers all year round.1 1 9           no labour surplus: she arrived at nliniillLn and
'I'o summarize, most of the evidence from    rmaximunm estimates of 6.4 and 29l; of the
direct surveys of hours worked indicates high    existing agricultural labour force, where the
labour utilization by farm families. At the samne  inimhumn estimate assumed that if any of a
time, some of the more carefiul studies using    marginal person's time was needed on the farmi
other approaches have concluded that as nmuch    he wotuld not be counted as surplus. and the
as 20 -.30'; of available labour goes tunused.   mlaximiiuml estimate related the total surplus of
Since  a number of these     refer to  African   person years, including fractions, to the labour
countries, for which very little survey data is  force, Melbra was trying to     edasUre labour
availablc, it is not possible to pin down the    which was 'surplus to agriculture',1 24 and since
sources of their apparent differerles. Certainly  persons on small farms do other work, the
there is no evidence that 20 40'; of the labour  corresponding figures for truly surplus labour
force is redundant all the year round. On the    would be lower.' 25 1 26
other hand, we cannot conclude at this point       An approach similar to Mehra's was used in a
that there are no countries where perhaps 10'.'  sample of 1,144 Taiwan farm families (1962),
of available hours are lost ('available' in the  but allowance   was irmade for housekeeping
sense of willingness to work at wages not far    reqluiremenits  and   off'-farm  labour   was
below going rates).                              recorded.1 2 7 The percentage of the potential
labour force (ages 1 2 60) emlployed was an
(ii) .M1at-i'mal produlcihit m in the traditionial  increasing function of farm size; compared to
sector                                           the maximnum utilization level (on farms of'4  5
Even where labour is fairly fully occupied,   chias), that for the smallest farms was only
MPL   may be near zero, especially in small      55%t. With this measurement teclnique, Wu and
productive  units  with  few   complementary     Lee reached an overall figure of I1;) rural
resources. One   metlhod  of qimantilying the    uneinploym nent. While the measuring rod is
amount of labour with very low        marvinal   arbitrary, the dilflerence in elmlployment across
product has been to commmmpare 'needled' with   farmi size, wlhich clearly reflects a differencee in
available labour, But this nmethodology     is   opportunhit for useful activity, can only mean
generally sUsp)eCt, sinCe thIe outside observer is  that the MPL for niany famiilies is below that
unlikelv to be able to delfine the niultiiLILie of  for families on large farmis. This propo,rsition is
relevant micro production functiows in terms of  further supported by the evidlence from Taiwan
the  heterogenoUS  lalhomi . lanid, and capital  and elsewhere of much higler lanid productivity
inputs.' 1 2  A second approach involves use of  on lhe smallest farms, suggesting that MPL is
comlparative data on labour inputs by farm siAe.  pushed down iouch further on these farms.
Buclk's study of Chinese agriculture. 1 21 whiclh   Thle studies jlust cited  were designed to
provided much of the miiaterial for early dis-  locate unnecessary labour by comparing units
cussions of the labour surplus issue, falls in this  assumed not to have a surplus with others
category. He    emphLiasized  the  much  lower   which appear to. An alternative way to get at
average crop output per worker on smiiall farin-, the marginal product of labour is a iiore direct
than on large ones. While not comparing NI l'L   prodcLtion1 fumnctiono approach. Many    sucl



1226                               WORLD DEVI1.1.0PI11 N l
analyses have been carried out. Specification   still slope up  because the income sharing
pitfalls are too numerous to consider in detail.  pattern varies across t'amilies, or because other
Some studies specify a production function      factors such as transportation costs vary. Thus,
under which MPL cannot be zero.some present     while the existence of labour surplus would
the MPL estimiiatedl at point of means128 as    increase the likelihood that real wages would
relevant to the labour surplus issue (when      not rise over time, such a result is clearly
presmnably it is the MI'L of farmis with a-     neither a necessary nor a sufficient condition to
typically high labour,:land ratios whicl is rele-  presume that labour surplus exists. Labour
vant); most fail to distinguish labour and land  surplus mighlt also be expected to decrease the
of diff..rent  qualities; most exclude some     variability of wages over tiimie, perhaps even
factors of production fronm the analysis because  over the seasons. But there seems to have been
of cotmiplementarity   with  labour (thereby    some confusion between the proposition that
probably generating an overestimate of labour's  institutional (non-economic) factors affect the
own marginal productivity); few   attempt to    supply function   of workers to the modern
discuss  the  complex  seasonal activity  and   sector and the stonger proposition that the
variety  of work in the farm    economy. To     wage level is a non-market or institutionally
summarize   what has been     leamed  by the    determined one. The wage is still determined by
production  function  approach it would be      demand and supply; supply as well as demand
necessary  to survey such studies in detail,    can shift under labour surplus conditions. So it
discardinig most and piecing together the results  seems unlikely that many solid conclusions
of the best executed ones. Short of that,       abouit labour surplus will be reached using wage
however, we can still conclude that such studies  data alone.' 3 1
provide some general evidence     against the      While the higher labour/land ratio and the
proposition that large arnounts of labour in    higher land productivity on small farms could
LDCs, even those     with unfavourable land/    be due to a variety of factors such as higher
labour ratios, have zero or insignificanit mar-  land  qualitie., closeness to  market, higher
ginal  productivity:, 29  at  the  same  time   capital/land ratios, and while it is true that the
they  provide   evidence  of differeltials  in  lower share of available labour hours used by
MPL across farms an(d of low MPI. for sizable   families on small farms is partially offset by
blocks of farms.                                greater oft-ta rni emiployement, available evi-
Since the logical pre.tuniption is that such  dence  strongly  suggests that the  marginal
surplus labour as exists is to be found on small  productivity of small farm labour is below that
family farms, production function estimates     of large farm labour. 32 A parallel pattern is
based on data from     snmall and large farms   suiggested by the imiore limited evidence avail-
together are   clearly  hazardous: they  miay   able t'or noni-agricultural activities.' 33  The
provide good overall fits without reflectinig the  evidence seemis to lead to the conclusion that at
characteristics of small farms. Separate analysis  least some of the dif'erence in labour/land or
of small farms or farms with high labour lanid  labour/capital ratios between small ,and large
ratios is a better way to get at the issue. Most  productive units is due to different supply
such studies have indeed found MPL to be        prices, and leads to systematically and sub-
lower on siall farmns. 130                      stantially dif'ferent MPLs. Sen's original con-
A final approach to the detection of labour  Jecture  that land   piroductivity  differentials
sur-plus miay be noted in passing  that which   owed their existence to a dual labour mnarket
attenmpts to relate wage trends to the presence  seem)s,134  in  other words, to  be broadly
or absence of surplus labour. In the labour     supported by the enmerging evidence, A com-
surplus 11o0del, earnings of somle workers in the  pairable plhenomiienoni is foulndl outside agricul-
traditional sector are affected by 'instUtional'  ture as well.
tfactors, that is, they are not (letcr imiimedl ex-  T'he infornu:;ition oni labour use and labour
elusively by the marginal product of labour.    productivity by size of establishnent or farm
One simple variant of the miiodel has taken this  does not demionstrate thiat labour could be
wage as a constant, andl concluded that the     removed from any Unlits without a decrease in
modern   sector faces a lorizonital sNUpply of' (ot1pult   the extremle definition of surplus
labour. In fact, a variable set by non-ecnlommlic  labour. Nor does it bear directly on whether
factors need not remain constant. Furtlher, even  workers with low productiVity would be willing
if the institUtionally determined supply price of  to take jobs elsewhere. Finally it does not, per
each potential worker in the modern sector      se, provide any hints as to whether the labour
were lixe.l, the aggregate supply curve of such  mi.arket imlperfections leatling to diffTerent MPL
workers to the modern sector would probably     across prodUCtion units are due to informiiationi



IBL M).\T.\RI: I11 RI    ANCI IN I)LNV'1LOl'IN(; ('OUIN URIlS        1227
problems, costs   of   mobility, monopsony!    factor which would lhave to be shifted to equate
monopoly, or to the income sharing behaviour   its MPL across productive units without any
on which the main labour surplus hypotheses    other shifting of other factors. f or agriculture,
are based. Since it is obvious that all of these  evidence on the variance of labour/land ratios
phenomena do exist at some level in most       across farms gives clues. It appears that perhaps
countries, they must explain part of the dif-  30 35(' of the Colombian agricultural labour
ferentials, perhaps even the larger part of them.  force would have to be shifted,1 36 while a
It should   be apparent that the labour     much smaller share of India's would.
markets and labour allocationi processes in the   Suchl illiustrative gue.,ses. even if reasonably
agricultural sectors of many LDCs are quite    accurate, leave open the (luestion of what can
complicated, as are those of other traditional or  be done in practice to improve the labour and
'informal' sectors, While simple models like the  other factor markets and how much of the
original labour surplus ones have given us     optimal reallocation  could be aclhieved by
powerful insights, their simplicity when taken  various policy tools. This depends in tum  on
too literally has led to invalid or mlisleading  why labour is so often unevenly distributed
conclusions.  The  intricate  interactions  of  over the comnplemiientary resources. The litera-
demand and supply and the detailed phasing of  ture on the differing economic characteristics
complementary activities over time or among    of farms by size includes extensive discussion of
household members does not imply that the      this issue, imperfections in each of the land,
market is perfect, so one cannot go to the     labour, and capital markets appear to play
extreme of presuming it works 'well', merely   significant roles.137 On the labour side, it
because changes in demand and supply are       seems well established that muclh family labour
reflected fairly quickly in prices and qjuaritities.  has lower opportLunity cost on the family
Labour surplus theory    and attempts to    farm  than elsewhere, because of prefcrences.
quantify the disguisedly unenmployed surplus   transportation costs and income sharing,. l'e
assume inadequate complementary resources      capital market is highlly imperfect because of
for the labour available from farm families. The  high risks, imperfect informiiation, monopoly,
amount of labour which would have to be        etc. The land market is imlperfect hecause of
reallocated to bring MPL to equality between   such factors as costs of vigilance (by owner to
the modern and traditional sectors and the     make sure sharecroppers pay the rent), costs of
corresponding income increase is a measure of  subdivision (if it is possible to rent plots of a
the extent of labour surplus. But factor mis-  large farm), the political danger of having
allocation'1 3  loss may alternatively be reduced  renters on land which they may then claim, and
by improvemient in other factor markets or     so on. Many of these imnperfCctinns arc the stuff
changes in the distribution of other factors; in  of undcrL'CvClo0jmu1enCt; they diminish gradually
recent years the possible gains from channelling  with development but no magic wand can erase
new  inputs to traditional agriculture have    them at a given point of time. More relevant
received increasing attention. The implications  here, there is no presumption that improvement
of moving more complementary resources to a    of labour allocation is more likely to result
traditional sector  require  careful analysis,  from improvements in the labour market than
especially of the incomre sharing phenomenon.  froam improvements in other factor markets.
Conisider for example, a situation where M1'K is  (Capital and land market imnperfections often
particularly highi in this sector and capital is  work in the same direction as labour mnarket
then made available to it to the point xvhere  imperfections to create wide divcrgenL:es across
MPK is equated across sectors. Thle iiiarginal  farms or firms in the ratio of labour to other
product of labour would rise if the traditional  resources. 'I'here is some, though as yet less
sector labour force did not cliange and hours  Nystemmmatic evidence that MPK (arid ndmrginal
worked inight increase, But the income sharing  product of land in the case of agriculture) tends
arrangemients which were the originial source of  to vary in the opposite direction from  MPL
the high labour/otlher resources ratio might now  across produicing units, i.e. that where MPL, is
encouirage people to return to this sector or not  low, MPK  tends to be high and vice versa,
to leave thus reviving the original ditt'erentiall of  \\ lmer-e this is the case, the existing inefficiency
MPL between sectors. Well specified assump-    muist retlect some forumi of immobility on the
tions about labour mobility are clearly needed  part of both labour and capital (or land) and
before very interesting conclusions can be     could be reduced by moming land and capital to
lrawn,                                         the labour, so to speak, as well as by moving
Somne feel for the    iuiagnituide   may  be  the labour to the land and capital. 38
obtained fronm estiniates of the amount of one    What are spileitic possible uses of low pro-



1 228                              WOR   DI.F) IIVLOM'1 I ,T
ductivity labour in agriculture? Activity on the  location gives any new firni an advantage with
farm clearly avoids many of the problems. such  the consumers nearest to it and thus creates a
as transpdrt cost and reluctance to leave the   downxard sloping demeiand curve for it,140
family which lead to labour market itnperfec-   after a certain point the social product of an
tions. Local employment, especially in the      additional firm imiay become quite small, but
off-season, would be next easiest. Perroanenl   tile private returns are still adlequate to attract
cmlli_raliOTn is the Imlost extreme soiution.   new entrants. To judge whethler losses due to
this market structure are large onie may com-
pare the size of labour force enigaged in these
(C) .11l1mupov,listic covnpetition     sectors witlh that of other countries, tryinig to
allow for diffcrerencs in level of development,
The open and disguised unetrnployicnm (or    distributioni of populationi, etc. Income trends
underemploymiient) phenonmena discussed so far  over time in the sectors in question may be
involve low or zero iiiarginial private produc-  useful, since if the labour force so engaged was
tivity of labour. Other forms of iisallocation  growing faster than the dlemand for the services,
may be related, not to low private productivity  average income would be expectedt to fall.
Of that factor but to low 90'i.l productivity.    Little research has thus far been addressed to
one context in which social product is less than  these questions; what has been done  mostly
private is monopolistic competition.           in Latin America   tends to suggest that labour
The combination of fast or very fast urban  wastage of this type has probably not been
growth, the explosion of low income 'slums,    increasing rapidly nor does it constitute a very
the rapid growth of tertiary rather than secon-  signiificanit share of the labour force (5' might
dary industry in urban areas ancd other factors  be anl upper limit), Our earlier discussion of
noted  above   have  led  many  observers to   open miemiployiineMt brought out the greater
hypothiesize that rural and agricultural labour  wage distortion" in some African countries
surplus is increasingly  being tranisferred to  (Fast Africa beinig the most studied) compared
urban areas. P'art of the attraction of urban  to most of Latini America. Since the continents
areas is the availability of somiie hiigh paying  could (difter substantially in the present context
jobs. We considered above the theories whiclh  as well, it is possible that the low    social
explain  urban  unemployment in terms of       produictivity labour force in monopolistically
expectatiJans  of gettillg such jobs. Anotlher  competitive activities is a significant share of'
substanitial chunk of labour .disorption  miiay  the urbani force in some African countries. Still
occtr  in  the  easy  entry  minopolisticully  it wouLld tenid not to loom    large in their
competitive  sectors; this  market structure   nationalmlabotlrtclof.    t
permits imore and more people to get a share of  to    amnlos studty of Latin America led him
the total incomiie generated in a sector, even  to conclude that disguised urban unemploy-
though thiey may not raise total output, or may  mnent was decrerasig over the period of the
raise it by an amount which is less than the    1 950sand 1 960s to wlich his data referred.1 4
income they earn.                              His criterion was the share of salaried employ-
Specifically, tlen, a h% pothesis of interest is  ment, which rose by a median 4.1'. over the 13
that  nuch   labour  surplus  (labour  whose  )countries for which he had data (for varying
marginal  social productivity  is  below  its  Periods within the post-World War I1 period).
inc omic ) is harboured in the growing urban   lie notedl the c.on1ti.mdietion between his con-
'ILSioi   a d   tat   f'   oiie   erlir   o serer142
teri amy sector. In suppor't ot this hlypotlhesis,  elusion and that of Some earlier observers
be-sides the obviously rapid growth of the lov  wlio ten(ded to take the increasing shlare of the
income tertiary sector labour force, is the fact  tertiary sector as a sign of increasing dlisguise(d
that todays Ll)'s have a higher share (if the  ucmloy'nent      lhough the share of salaried
oiun -prim.lr) labour f)rce in tertiary ac&v ties  nworkers is an inadequaLe critericimi in itself, its
than dlid the now developed ' oumit ic at com-  trend  iS clearly  part of the circo mosta ntim
parable stages of developmenit.' 3 9           evidenice which can be adduced on the issue.
Tlhe imajo isstue, as with Lewis typie dis-    In ('olombia, where as late as 1970 less than
agliseL! immemiiploy iment, is not whethier there is  40'  olf the labour force in commerce were
some labour with low margnial social produc-   wage workers and where the commerce share of
tivity in stuch activities - there obviously is in  the total labour force rose from 5t; in the late
any country -- but whether (a) it is quantita-  1910s to about 9'; in 1970,'43 Berry's study
tively signifficant, and (b) it could be put to  revealed increases in the average wage in both
better use. With a mnoploplitlically cmnpenitive  wholesale  and  retail comml1erce dutring  the
market StruCtUre, as in retail omiimcrce where  perio(d of 1954 67 (not a fast grov.tlb period



LABOUR MARKET PERFORMANCE IN DEVELOPING COUNTRIES                   1229
for ~.e economy as a whole). Although evi-
dence was spotty, there seemed no reason to
believe that proprietor earnings had not risen
along with small establishment wages. Even if a
third of the people engaged in commerce,
restaurants and some other services were 'sur-                                   /
plus', it would be hard to build the figure much
above 5%', of the labour force (8-9%k of the    a
urban labour force). Such   a figure would
exceed  the  differential surplus  found  in
Colombia vis-a-vis a typical developed country, -
since some low social productivity labour is         s
always present in the monopolistically competi-      c        Lo        L2       l
tive sectors in a free market system.
Pending detailed defence of the pro.position               Quontity of tabour
that the monopolistically competitive sectors                    Figure 1.
are growing at rates exceeding demand for their
output, or that earnings levels are performing  their being hired would be minimized (at
badly relative to  those  elsewhere in the    SAIL0O) if the low supply price workers were
economy, a best guess is that their relatively  employed. But with the high wage being paid
fast growth is related primarily to demand    OL, workers will compete for the jobs, and
patterns. It has been argued that a major cause  they will be rationed, In the worst possible
of the high share of the tertiary sector in   outcome, only high supply price workers will
today's LDCs by historical standards is the   be hired, and the opportunity cost could be
relatively slower technological change in that  L2BCI1, where L,2L, = OLO. It is clearly
sector  vis-i-vis  agriculture  and  manufac-  possible for this miiismatching loss to exceed the
turing.' 1  In any case, there is no persuasive  loss froml the hiring of too few persons in the
evidence that the phenomenon is supply rather  protected sector (in this case area EIA ).
than demand determined,                          The mismatch that has received the most
attention to date is that related to the use of
educated persons (presumably having high
(d) Education- occupationz and other     opportunity cost) in positions which could be
mismatches                    filled at lower opportunity cost by others. The
problem involves the relationship between the
An excess supply of labour in one part of an  education and occupation levels of labour force
economy, as manifested by open unemploy-      members and, more fundamentally, the degree
inent or by workers in flexible wage jobs     to which the educational system is providing
earning less than they could in some other    workers with the appropriate skills, given the
activity, is one form  of inefficiency in the  composition of the demand for labour.
allocation of labour that can result from the    Ordering the occupational hierarchy on the
segmentation of labour markets. Other ineffi-  basis of the associated educational attainment,
ciencies associated with segmentationi. more  we would expect in virtually all contexts to
difficult to documinent but possibly imposing  find the wage rate positively related to occupa-
greater resource costs on the economies of    tion, since higher wages are required to coIll-
developing coLun1tries, involve the failure of the  pensate workers for the direct costs and income
market to move the 'right' resources into high  foregone during training. This presumes that as
wage sectors, a failure commonly described by  one climbs the occupational scale more training
the termn 'nmismiiatch'. The essence of the prob-  is required for minimum standards of perfor-
lem  can be seen in the diagram, where DD'    mance, Otherwise employers shotuld prefer to
and SS' represelt, respectively, the demand and  hire workers with less training and a lower
supply curves for labour in the protected or  supply price. If the labour market is operating
high wage sector, and where IVm is the wage,  efficiently, workers enter those occupations
set for examnple by legislation. Assuming all  where their skills will have the greatest impact
aspirants have the same productiviLty in the high  on productivity.
wage sector, while each worker's supply price is  Segmiientation can result in a mismatch be-
equal to his social opportunity cost, then if  tween wvorke!rs' skills and actual requirements.
OL0 workcrs are to be hired (as will occur with  Consider the consequences of a minimum wage
the wage at W,,,) the social cost associated with  from  this perspective. The legislation only



1230                              WORLD DEVELOPMENT
applies to the wage sector of the economy and  supply of such opportunities is responsive to
frequently can only be enforced in large firms;  demand and is not determined solely on the
it may lead to a significant difference between  basis of estimates of the future skill require-
protected and unprotected sectors in pay for  ments of employers, let alone estimates of the
work requiring the same level of skills. This  social returns to  various educational pro-
creates an inducement for workers with formal  grammes.'48 Thus labour market segmentation
education to enter the protected sector even if  may lead to a misallocation of resources in the
their training does not contribute to increases  educational system, one manifestation of which
in productivity as much as it might elsewhere.  is the bias toward academic training, despite the
In the extreme a worker with considerable     skew  in the   distribution  of school-leavers
formal education may enter a job for which no  toward nmanual occuLpations.
training is required, because that job is pro-
tected by the minimum wage, while the occupa-
tion for which the worker is best suited pays             4. CONCLUSIONS
less than the minimum in the unprotectedi
sector. 14 4  Inefficiency in the allocation of  If research on labour markets in developing
labour among occupations need not be con-     countries were provided with clear and accurate
fined to the lower rungs of the occupational  data and appropriate methods for answering
ladder. Wage fixing and wage rigidity due to  precisely formulated questions, this survey
government policy in middle and high-level    might be considered too long. Since there are in
manpower posts may result in similar misalloca-  fact few accurate data and considerable dis-
tions of educated labour,                     agreement as to which questions are important
One resource cost of this type of misalloca-  and how best to answer them, it may even be
tion is the output directly foregone because  too short to do justice to the complexity and
educated workers are engaged in activities in  variety of the issues involved. Furthermore,
which their marginal social product is less than  because  the  zfficirilcy  of  unconstrained
in the activitic s in which they would be engaged  markets is an article of faith for some econo-
in the absence of segIllentation. 146 The secon-  mlists, while the pervasiveness of imperfections
dary effects of segmientationi may impose even  and their dire consequences are premises to
more significant costs. The misinatch between  which others give uncritical allegiance, it is
necessary skills and employers' requirements  necessary to exercise great caution in inter-
implies a gap between the private and social  preting the available economic evidence. Never-
returns to investment in various levels of    theless, certain key features of our assessment
education. The gap would persist even if      of the performance of labour markets in deve-
employers  selected  randomly   among   the  loping countries, which we now recapitulate
competing applicants for high wage jobs, as   briefly, appear to be firmly established.
some   workers with 'excess' education   are     Our null hypolltesis, that labour markets
bound to be hired, their skills would have been  function at a comparatively  high level of
better used elsewhere. The gap is greater, the  efficiency, is based on weli-documented aspects
more employers use education as a criterion for  of their macro dynamics. As development
rationing jobs in which formal qualifications  proceeds, time series and cross-country compar-
beyond a certain level are not associated with  isons reveal changes in the occupational, indus-
highler levels of labour productivity.'47 These  trial and spatial distribution of workers which
gaps are almost certain to distort the signals on  are consistent with clhanges in the structure of
the  basis of whiclh educational planning     production and economic growth. The success
decisions are reaclhed. Certainly to the extent  of labour markets in niuhilizing workers for
that the private demand for educational oppor-  new growth-generating projects, and in otlher
tunities is a function of expected economic   ways altering the dlistribtilion of labour services,
returns to invest inents of time and money, it  does not, however, exclude the possibility that
will be inflaLed in the aggregate. If, when   at any given time labour may be nlisallocated
r3tioning employmlent opportunities. empfloyers  and a significant amount of productive poten-
exaggerate the advantages for productivity of  tial be w-asted. Furthermore, though micro-
certain types of educational credenitials, for  economic stuCies of migration and peasant
examiple academic, nis-a-vis others, e.g. voca-  agricuLlture have provided conclusive evidence
tional, then the structure of private demand  that workers respond to economic incentives in
will also be distorted. While in virtually all  allocating their time, this is not in itself
LDC's the government controls most of the     sufficient to confirm  a high level of labour
edLucaLional systen, there is evidence that the  market efficiency.



I 1ABOUR MARKET1 FRI'ORMANCE IN D)FA'V   I OPtNG'( c'tNrI ITI.S          1231
Our conclusion, that in miost countries our   uLncniiployinient and dkguised unemployment
null hypothesis is accepted, rests primarily on  have too often been exaggerated. Exhortations
the micro-economiiic  evidence  of observable    to give the resolution of these problems a
causes  of   misallocation  and   particularly,  priority equal to the achievemilent of a high rate
because of methodological problems in moving     of economic growtlh or, more optimistically, to
directly from identification of causes to assess-  orient strategic planning to take advlantage of
ment of costs, on the evidence relating to       the low wages and high savings a pool of surplus
consequences. This is not to suggest that in     labour offers, seldom  appear justified by the
developing countries labour is optimiially allo-  facts. This is partly explained by the failure to
cated. There is little doubt that income sharing  identify open unemploynment as an allocation
practices sever the relationship between the     or pricing problem  and the assumption tllat,
marginal productivity  of workers and their      because the opportunity cost of leaving a
earnings in family enterprises: that union pres-  worker unemployed is high in industrialized
sure, government legislation and other factors   countries (where coomplementary factors are
segment the labour market; and that in hiring    also likely to be underutilized), it must be high
and wage determination employers discriminate   in developing countries, Underestimation of
for or against certain groups of workers on the  work timie, particularly off the farmn, and an
basis of non-economic characteristics. These     uncritical acceptance of slhort hours as a sign of
and other factors give rise to myriad dliffere nces  surplus labour, explain in part the initial over-
in nmarginal productivity among workers of a     estimates of rural disguised unemployment, As
given type, implying tllat a reallocation of     in the industrialized countries, the resource
labour could increase aLggregate output. Indeed,  costs of any particular form of misallocation,
we have emphasized that, thouigh the explana-   induced by segimienitationi or some othler eause
tion is often sought elsewherv, the irmmediate   of inadeqjuate  perl'ormiiance  of' the  labour
cause of open uinemnpluyinent in developing      market, are not likely to exceed 2"i of GNP. By
countries is frelLuently a malfunctioning of the  contrast, where  unemployment is     due   to
labour market. Generally, open unemoploy mnent   Jeficicncy of aggregate demand, output lore-
is not a symptomii of an aggregate shortfall of  gone per idle labour lforce participant is hiiglher
enploynient opportunities relative to labour     and aggregate costs have been imeasured to be
supply, as conventional demand deficiency.       10  15"; of GNP andc     more. f inally, it is
technical lack of substitUtion, or econoniv-wvide  important to recognize that optimal use of
rigid wage models suggest. Except where it is a  labour may require significant andl often diffi-
genuline  search  phenomenon    (a result   of   cult improvements in other factor mlarkets. If
inadequate information), it is in most cases a   labour-intensive  technologies  were   chosen
symptomi of labour misallocation caused by the   wheiever socially eflficient, the productivity of
decisions of workers, miore and less educated    now surlulIs labour mliglht rise Nubstantially in
alike, to torego available low income emplov-    some countries; but the obstacles to choice and
ment opportUniities and queue for the few high   effective utilization of stuchi teclhnologics grow
income positions available in    a segmnented    increasingly evident witlh the passage of time.
labour inarket.                                     T'his is not to suggest that labour market
The explanation    of labour misallocation    problems are of so little sionificance as to
sh; Id not be sought only in imperfections of    juNtify ignoring tlhemn; only that e\aggerating
the labour imiark-et, however. In many situa-    their IllugnitLude is counterproductive, as it may
Lioln, at least two markets Imiust have imuiperfec-  lead to the adoption of such tlhird-best solu-
tions before resoturce misallocation occurs, and  tions as make-work piogruannmes whose cost
it %k ould be arbitmamy to designate either type of  exceeds their IenCrlits. Of course, some couii-
imperfection as the cause of miiisallocationl. For  tries may suffer everal serious labour imarket
example, misalloeation of labour between small   deficiencies Himnult mimeoulyl, total costs being
anti large luarmmE cani occur only because both  qtuite high; but it is doub't'ul whether, in this
the labour and the landl mlarkets are niperteet.  case, any s ingle remiedial measure would prove
Nor is sirnply douitimenting the presncie ot  sut ticient. Government intervention  may be
nmisallocations adequate to establish that labour  ius,ntled on grounds other than potential gains
(or othieri mlLark-ets in developing countrics are  in oilput ': unenmployment is considered a social
seriouYly inefficient. An  assessmeniit of the   probleml as much becauise of the loss of income
nmagnitude of costs, in terms of output fore-    suffered by indlividuals as because of the loss of
oonc, liszhibutionnal inequitics, or other types of  outplut in the aggregPte. IlOWever, these 9uhjeC-
loss is necessary. Our ikidgeimment is that the  tive costs also appear lower in dteveloping
costs of such    plC1e1n111CnI  as open  urban   countries than they woiuld be in indmistrallizeL



1 23 2                                WORLD 1)l \'I 1.01'1 I NT
countries without a fornmal social security sys-     of labour market performance, we do not view
tem. They are alleviated by intra-family trans-      what has come to be called 'the employment
fers to the unemployed from those with jobs          problem' as unimportant. Rather, the implica-
and by the availability of flexible wage sectors     tion of our study is that the term is a misnomer
as employers of last resort. Workers' views that     because it conjures up an image of labour
segmented   labour markets are mere lotteries        market failure. The slow     rate of growth of
while those where discrimination is practised       workers' income at the bottomn of the distribu-
are unjust, may of course be considered socially    tion, despite    (luite higlh aggregate rates of
dysfunctional by -policy-makers and provide a       gro;wth of output, which often appears to be
justification  for   egovrnment     intervention,    the source   of concern, is better termed       a
Finally, though    we have not attempted to         poverty or distribution problem. When it results
specify   the  links  between    labour   mnarket    froml  the low   rate of increase of demand
pathologies andl the distribution of income, it is   relative to supply of 'unskilled' labour, it is
obvious that elimination of excessive income        apparent that the labour market is the prob-
differences between sectors, geographic areas,      lem's immediate locus: but the most important
and   types of labour     (the  cause   of many      factors inlluLencing the dynamics of supply and
misallocations),  will yield   the   benefits of     demand    for  labour    -  population   growth,
improved    distribution  of income as well as       educational opportunities, the levels of invest-
-greater output. The former may provide suffi-       ment,   technology        are  themselves    only
cient juLstifica:tio; for government intervention    moderately influenced by aberrations in the
where the latter alone does not.                    interaction   of  suppliers and    employers of
There remains one important point to clarify,    labour services.
Though we dismiss exaggerated claims of critics
NOTEIS
1. In developing countries there are notable excep-  force in poor countries, where thie proportion of wage
tions to the con"L'enti,,nal assumptions that suppliers  earners (1LW/L) is low; the two are roughly eqiual when
and employers of labour services are sellers and buyers  LW/L reaches 0.45, according to J. Lacaillon and D.
a-nd can be identified with distinct sub-groups of the  (;ermidis (1975); and the average wage falls below
population. Where agricultural and related activities  average income in developed countries, by 2-5; or
are characterized by r-nrkI4 ceasonal fluctuations in  imore. Neerthleless in economies where LW/L is quite
the inarginal product of labour some workers may be  low, 10'; or so, the wage share is also low, tlhoughl
empnloyers at one time of the year and wage labourers  lligher thani LNV/L, 2()', or so. The share of total
at another. lndeed in several key respects own-account  labour incomeje, wages plus im)puted income of own-
workers, who generally comprise significant propor-  account workers, also rises from roughlylx 40--50',, in
tions of both rural and urban labour forces, fulfil both  poor countries to  70f. or so in industrialized
roles simnultaneously tlhough of coUrse witlhout produc-  countries. This increase, however, is not due to the
ing market transactions as evidence.                redistribution of labour but to the decline of the ratio
rate of retorn to capital/wage rate.
2, D)ata supporting these penerali/ations is in A.
Berry and R. Sabot (1976).                             6. F. Machlup (1975), T. NV. Sclhultz (1961).
3. 11. ('henery and NI. Syr(luin (1975); S. Kiuznets  7. S. Kuznets (1966).
1966).
8. .1. Durand (1975).
4. D. Antderson (1977). [or farm families this is
not nress:irily true durinig early stages of develop-  9. This presumption is supported by general evi-
ment. As suchi laiilies becomc more integratdLI into  dence of higlher average labour productivity in sectors
the market econmIvn thiey may secialize incremsintly  with rapid output growth. See S. Kuznets (1966).
in agricultural output for sale, while purcltsing items
zhey previously provided. S. llymer and S, Resnick    10. Even equality at the margin of the social costs
(1969);S. Resnick (1970).                           and btcncfits of unemployment imniplies an efficient
labour market on'y in a relative sense. Compared to a
5. A corollary of the increase with development in  'first best' ,allcaution ol labour unemployment, though
the proportion of the labour force receivin, remunera-  an optimnal response in a 'second best' world, remains
tion is a rise in wvages as a proportion of national  an indicator of labouir market inefficiency.
income despite the decline of average wages relative to
the average inconmes of non-wage earners. The average  11. See R. Sabot (1978).
wage is well above the average income of the labour



L . B1301'O NI ARKErT l'L RIt ORNMAN(T IN 1)l \'V'l OPING COUNiTR I1.S             1233
12. i-or e\ample, R. Krishlna (1963); J. R. 3ehIr,iani  (1970); and  M. Bienefeld   and   R. Sabot (1971)
(1968); I. R. Dean (1965); V. Dubey (1963); W.          respectively.
lalcon (1964); W. 1). Ilopper (1965); C. C. Malone
(1965); 1). Welsch (1965).                               28. When determining samnple size, underestimation
of the degree of siratificacion required at the analysis
13, (;. Iielleiner (1975).                             stage can result, as it did in the case of the Kenya
sulrvey, in limlits on disaggregation simlilar to those
14. And that higher vvelfare wvill be achieved after  imposed by conventional sources of data.
the change.
29. S. Rottcnberg (1956).
15. R. Sabot (1978).
30. J. Knight (1972) and P. Collier and R. Sabot
16. R. Sabot (1978). Note thiat various steps such as  (1976).
the alienation of 'native' lands were taken vith a view
to lowering the supply price of labour by diminishing    31. P. ('ollier and R. Sabot (1976).
the alternative (opportlunities of poienui.il workers.
While not coercion in the sense that workers were        32. Recall that in the classical labour markets of
forced to do work against their wtill, these steps had  Adam Smith, workers made choices among alternative
the saume ultimiiate e'ffect ol' making labour cheaper  employment opportunities on the basis of total net
than it would otherwvise have been.                     advantage, not in terms of' comparative wages. (S.
Rottenberg (1956)). Since the maximization of in-
17, M. Miracle and    B. l'etter (1970); R. Sabot      comiie is not synonymous witli the nia\nimjiation of
1978).                                                  welfare, there will always be many individuals in
employment wvho have balanced a lower level of
18. S. Ilymer and S. Resnick (1969).                   expected  income   against a higher level of non-
econiomic bene'its in their hluman capital investment
19. 1. Berg (1961).                                    decisions.
20. See G. Becker 1965) and S. Linder (1970) for        33. 11. Barnum and R. Sabot (1976).
detailed theoretical discussions of how, given a fixed
total supply and increasing demand as a consequence      34. See A. Bottomiiley (1969); C'. Bliss and N. Stern
of rising incomes, the a1lh,ition    of time anmong     (1977); J. Hlarris (1971); It. Leibenstein (1966): R.
production and consumiption :mcti%ities and idleniess is  Sabot (1978); and J. Stiglitz (1974).
likely to change.
35. R. Sabot (1978).
21. 1'. Britg (1973), see also 1). B\ erle (1974).
36. D)efilnd as the actual output ol' the resources
22. Only rarely are cotntrol groups, comprised of      or ili/ed by a producer rTl.uie to potential output, l'or
non-migrants withl chIaracierisiics similar to migrants,  a given allocation. See II. Leibenstein (1966).
asked why    they  did not move; the citegorieN of
answers are generally predetermined, and the relia-      37. 11. Johnson and P. Mtieskowski (1970); H. G;.
bility ot' the answers depends on the degree of clarity  Lewis (1963).
of the respondent's conmllrehenision ot' his reasons for
migrating and his ability and willingness to coin-       38. Study   noted in It. Turner and     D. Jackson
municate hliese to enumerators. L ven the best designed  (1970).
of these m'igration stuedies do not yield mleasures of the
elasticity  of migration  with respect to changes in     39. See G. Psachlaropoilos (1973) l'or hibliouriphical
policy or other variables.                              detdik ot nunieroLus of tllese studies.
23. See. for e\ample, W. J. 13arber (1960').            40. The work of M. llaiug (1973) on Tlhailand is an
e\LCPiH0ii in that while measurilg the rate of return to
24. 1, Siistaid (1962).                               education was thle principa.l aimi of' his eslimalion of a
wage   t'unction, he  did  use thic miiicro data base
25. ()t the 14 economletric studies '  ilern:l migra-  generated  for that lpmirp'|'s, to ek\plore. it' only
tion  in Atrica, Asia ani I attin  Armerica  receml1%   eursorily, (actors other than edIi aill ion1ClueneCi,n thle
reviewed by Yap (1975), 10 analysed ini(er-repininl     level otfwge .
tlows from which tneiiher rural rural nor rural urban
flows could   be separated. and only one mial%d sed      41. (. Ns.icharopoulws (1973).
variance in mniaration rates within mduc itional ero lps.
42. K. Arrow ( 1 973).
26. T. P. Schultz (1976).
43. Z. (;rilhLhes (1977).
27. See 1. flarris (1976); J. Scully and NI. Toosie
(1976); f1. Rinpel (1971); 1). lIyerlee and J. Tommy     44. R. Layard and (G. Psacharopoulos (1974).



1 234                                  WORLD DEVEAL011NI'Nr
45, 1). Mazuindar (1977).                          inoperative  or  productive  of grave   social con-
sequences'.
46. If they are pertfect substitutes, education would
be devoid of social value and the correlation between  58. Uor example, J. 1P. Grant (1971): 'It is probably
Iability' and  educati)n level would   nut justify   no accident thiat many of the most severe of these
employers' preferences for relatively inore educated  (political) uphleavals in recent lhistory have occurred in
workers in higher level OucCUpatiu0s.                countries with the higlhest level of wwnmnplw. i:enO.
47. G. I'Sdlmclropoulos (1973); M. laiaug (1973); 1I.  59. J. Nelson (1970).
Thias and M. (arnoy (1972) J. 13. KIniglht and R. 11.
Sabot (1977).                                         60. Though unemployment rates are generally much
hiigher among the educated and relatively highl income
48. 1). %M'azumdar (1977); J. 13. Kniglht and R. 1I.  groups, their small size mieans that the uneduicated and
Sabot (1977).                                        poor may still comprise a significant proportinn of the
pool of unemployed workers. A. lBerry t1975a); R.
49. 1). Liim (1977).                                Sabot (1977b).
50. G. Fields and N. Marulanda (1976); J. Knight     61. L. (;. Reynolds (1969); NI D. Whitaker (1970).
and R. Sabot U1977).
62. J. Ramos (1974).
51. J. Knight and R. Sabot (1977).
63. The discouraged worker lyp)oUthesis of a negative
52. 1). Morawetz (1974);    ' W. Baer and M. Hlerve  relationslhip between the participation and unemploy-
(1966).                                              ment rates fecuses exclusively on thle substitlutionl
cff'cct of standard demand theory, An increase in
53. L. Pearson (1969).                              unmployn'nt nl may draw Udditioin:l wvorkers into the
labour niarket fronIt f'amlilies with mIemibers whio are
54. See for example 1). Mlorse (1970); ' c .ountries  ou(t of work in an attemipt to conmpensate for the
that are undergoing rapid economic growth are still  decline in faimily income. This incomle effect is the
faced with inercimsinvm unemployment. . B, and it is  basis for a competing   i)ypotlhesis. that has been
'virtuially certain that the scale of the problem wOill  sustained empirically by somiie studies in indulstrialized
increase dramatically in the years ahead'; J. P. Grant  countries; 13owen and Finnegan (1969). however,
(1971): '. . . Thcre is a serious and growing unemploy-  conclude that it is swamped by the discouraged
ment problem   in countries fromn one end of the     wvorker effect and raise metlhodological doubts about
developing world to tlle otlher and it is likely to  the studies which fotund it to be important. For
dominate international development in the 1970s as   Bogota, Berry (1977) found that, as in indUstriali7ed
the food issue did in the 1960s'; Vincent Barnett:   ;outntrici, the substitution effect dominates the in-
'TUnemnplom ment (however measure(L) is at too higlh a  come effect for males; for females no significant
lesel in most of the countries and is increasing. A  relationi was found. One e.planaiion for Colombia
number of countries Wvith otherwise    satisfictory  that would s%ugest that this finding applies to other
growth rates continue to suffer increased unemploy-  developing countries concerns the household composi-
ment'; 1). Turnham (1971).                           tion of the uinemiployed, a topic we discuss below. In
industrialized countries the additional workers forth-
55. Nl.  Todaro   (1971): '..A     clironic  urban  coming tend to be secondary workers, women and
uneniployment and mindereumployment problem    has   chiildren, who enter the labour force wlhen primary
emmrc d in tropical Africa. A11t1OUgh there are few  workers, generally heads of houselholds, lose their jobs.
hard  data  on  the  magnitude   of African  urban  W\here household heads comnprise a relatively small
m1netnIpIoN mllelnt, owilln both to conceptutal difficultIies  proportion of the unemployed, as appears to be the
in dulfining unemuipoi) iment. and, more importantly, to  case in the LDCs for wvhich information is available,
the fact that very few studies have been directed to  the rationale for the income effect is weakened.
the problem. the limited evidence available provides
aimple empirical ctnfirutatitn of whiat any informed  64. R. 11. Sabot (1977a).
observer already knows    namtmly that urban unem-
ployment is an extremeel serious probleni'.           65. The taxonomy presented lhere is taketi fromT a
paper by J. laTris and R. Sabot (1976) and muiichl of
56, R. Krishlna (1973). Llnemlploynient in India 'is a  the discumssiotn draws freely from that paper as well.
grave national problem ... which hias defied solution
in spite of two decades of planned deivlopteni'.      66. A. I eijonliurvutd (1968).
57. R. Miller (1971). The imaginitude of uiiemploy-  67. T. Nitirgai ( 1 952) -and W. S. Wuytinsky (1953).
mnent indicates 'that thie labour markets of Latin
America have reached a state of nearly total disintegra-  68. This does not imply that in LOCs capital
tion -- a situation in wlhichl the allocative and pricing  equipment is fully utilized, only that for the most part
mcclh;misnts of urban and rural markets are eitlher  underutilization is due to suchl factors as supply



LABOUR MARKET 'LRI:OR.IANCL IN DEVI LOPING (OUNTR101 S                         1235
boutleicelks, maintenance  problems, etc. See   t.    85. G. Tidrick (1975).
Winston (1971).
86. The evidence suggests that highler uinemployment
69. W. Reddaway      (1963)  also  quoted  in  M.   rates among educated than uneducated workers are
Gersovitz (1974). For similar sceptical views of the  found  IroluulloLut the developing world. R. Sabot
applicability of aggregate demand policies to employ-  (1977b); A. 'Berry  (1975a); ILO    (1972); ILO
ment problems in LDCs see A. Peacock and F. Shaw     (1971); D. Keesing (1975); 1). Mazumdar (1977).
(1972), and V. K. Rao (1970).                        Further dlisa.gregatioun reveals that tlle relationshlip
between education level and rate of imretiiiploymncn is
70. R. Eckaus (1955).                               that of an inverted U   thie une(duicated and workers
with post-secotida ry educationl experience the lowest
71. F. Domar (1957).                                rates. See M. Blaug (1973).
72. R. Solow (1956).                                 87. H. Leibenstein (1957). See also C. D)ougherty
and M. Selowsky (1973). Ilarberger's study (1959) of
73. J. Meade (1961).                                Chile is an exception. Hle concludes that the realloca-
tion of existing resources could raise national outiput
74. While G. Haberler (1950) was the first to assess  by as much as 15':. flowever, he assumes large
the implications of the relationship between minimum  discontinuities between those industries where inputs
wages and employment for the theory of optimal       are located and those industries to whiclh the resources
trade policies, one of which is the weakening of the  would be iiioved in the absencc of distortions and does
proposition that autarky is inferior to free trade, and  not consider the costs of social overlhead capital or
later H. Johnson (1965) and J. Bhagwati (1966) dealt  direct capital necessary to make the mioves involved.
with the same issue, Brecher's (1974) analysis is the
most notable.                                         88. 1K. Phelps (1970); R. Nelson and S. Winter
(1975). In these models if the structure of wages
75. 'Suggested' rather than 'implied' is the appro-  remiained stable, unemnploynient wouild be a transitory
priate term  because wvhat A. Lewis (1954) calls     phenomenon since witlh conuiinued searchl iml'orinmtion
'unlimited supplies of labour' could be the con-     would imlprove and thle siUhjcclive distributioni on the
sequence of emlployers in the capitalist sector, for one  basis of wlhich  individuals make their decisions
reason or another, paying more than the supply price  whether or not to accept a job olfer and the objec Live
of labour.                                           distribtutiin wouild  converge, so  tlhe  returns  to
additional search would decline to zero. However in a
76. A. K. Sen (1966).                               dynamic economy a flexible wage structure is unlike)'
to remain stable.
77. It is not, however, voluntary in the sense
associated with rentiers whose reserve price e.\ceeds  89. J. Hlarris and R. Sabot (1976);and T. Modieliani
the prevailing wage; it may persist even where the   and 1.. Iarantelli (1973).
reserve price of the unemployed is less than the wage
in the low wage sector.                               90. A. Alchian (1970).
78 R. Sabot (1978).                                  91. See J, Hlarris and MN. Todaro (1970).
79. See l{. Barnum and R. Sabot (1977);see also M.   92. With reference to the conte\t of iradirion:tl
Todaro (1976).                                       agiculture II. Leibenstein (1957) suggested a relation-
ship between wages and nutrition as the basis for a
80. (;. Iields (1975) cites evidence (rom a number  poSttive link betweeni wages and prodLicri'.k%, an
of countries whiere the estimated ratio of rural to  hlypthesis that has been     widlely  discussed. See
urh.an incomes is between 1:2 and 1:8 and thus       Bottomley (1969), C. 13liss and N. Stern U1977', J.
predicted  rates  of   urban   uneniploymnemip  are  Hlarris (1971). Ihle rclationsl6ip aiulm' wages, turnover
12.5 50%) and yet tlle lhighest measured rate of     and non-waLe costs of labour lhas been discussed by J.
unemployment is 20(1%.                               1. Stiglitz (1974) with particular reference to l ast
Africa, by R. Sabot (1978).
81. See 1). Mazumdar (1976): R. Sabot ( 1977aL); P.
Collier (1975).                                       93, See for example, M. S. Alhltuwalia (1974). U'nder-
lying the negative interpretation  of' this ehlinve.
82. G. Tidrick (1975).                              however, is tlhe presumption that the increase in the
non-wage slhare of total income all accrues as incomle
83. L. G. Reynolds (1965); L. G. Reynolds and P.    to capitalists. In tact, the increase in profits may
Gregory (1965).                                      induce incrca'ed investmenit, eitlher by capitalists or by
the go.ernmnent whliclh owns the enterprises or has
84, The conmbination of urban uinenmployminct and   taxed away thle increment.
rural labour sc:rcity has also been noted in Sri Lanka
by l . Thorbccke (1973).                              94. If' that el:sticiu% were unity, the moodern sector



1 236                                  WORLD 1)1EvU I   ,0LMENT
wag,e bill would remain unchlanged, so only under the  109. In August 1972 and Noveniber 1973 the figyures
extreme assumption that additional workers woull     were 12.6 andl 12.4 respectively.
receive no wages elsewhere would the total wage bill
fall,                                                110. In other words, a highl share of the low lhouir
workers in   non-agriculture  wanted  more   work,
95, A. K. Sen (1975).                               wlhereas a much smaller share of those in amriculture
did. Thlis is mainly due to the prevalence in ayxiculture
96, From A. 'igou (1933).                           ot' unpaid fanily workers. Of wage earners, a hiighler
share of tlinwe working less than 40 lhr wanted work,
97. A. Berry (1975a).                               over 45'; in aptricultture and arotund 55'; in non-
apriultlinre: of selt-eniplonyed workers, lower but still
98. R. Sabot (1977b).                               substantial shares wanted more work.
99, See ). I'urnlham (1971); MN. B3laug (1973); 1'. J.  ll . Since people were not asked to state how nany
Richards (1971); D. Nla.tumdar (1977).               mlore hours they would likc to work, these guesses
cotuld be eitlher upward or downward biased. It seems
100. Among developed      unuritics the range is also  unlikely that a higlher figure could be much, if at all,
very wide and the median might be about this level.  downward biased.
101. Some coimmentators hiave concluded that rural   112. Wlich incluided 4.77 who did not work during
underemployment was a serious problem in Western     the reference week.
I.urope pTrior to the industrial revolution. C. Clark and
M-. Ilaswell (1970) note that 'tlhis state of affairs, so  113. Hliures for India, Korea, and Japan have been
res tably commnon in the present-day world, was also  calculated  from  tables presented  in  S. lshikawa
this is not generally recognized  the lot of rural  (1967).
Iongland  for a long period, probably for several
centuries, in the past. In 1688 Gregory king estimnaled  114. K. 13ardhlan (1976).
that, out of a total population of 5.5 million, 1.3
million represented "cottagers and paupers" and their  115. Y. S. C(l') (1963).
faiimilies, a seriously under-occutpied rural populdation.
receiving an averaye family income less than half that  116. Byerlce and 1 icher 11972) report: 'Most studies
of a regularly employed labourer. This persistent rural  in rural areas of Africa have found comparatively low
unemployment in Frngland did not disappear until the  labour use in apricultural production. Cleave (1970) in
middle of the nineteenth century, and in Ireland is  a survey of 1.5 micro-level studies of agricultural
present to this day. It required a rapid growth of   production in areas of both high and low manland
urban employment relati.e to population    -more     ratios, found an annual average of little over 1000 hr
rapid  than  that of which    the seventeenlh and    per male adult used in alricultural production. At first
eigh teendtIm centuries were capable  to solve this  sighlt these figures suLggest a  b%ihstantial pool of surplus
proiblem, even in a country so favourably sit.UldL   labour in rural areas whichi can be drawn into
as Englangzd'.                                        T0LI Lietion, . .'.
102. C. Kao, K. Anschel and C. I ielier (1964).     117. As reported by W. Jones (n.d.).
103. A. Lewis (1954).                                1l8. J. Cleave (1970).
104. J. l ei and G. Ranis (1964).                    119. J. Cleave (1970).
105. Note that the concept dated from mulch earlier; - 120. Rosenstein-Rodan, in his study  ot' Southern
the term  appears to have been proposed by Mrs        Italy, made a seriouis attemipt to get this sort of'
Rohinsomn in reference to certain types of urban     intorimuation by questionnaires. D)etining the removable
vworkers in a developed economy; J. Robinson (1936).  surpluis in lpersons rcq,uired for 50 days or less per
year, he estimnated it at 10  1 ," of trhe agricultural
106. A. Berr% anid R. Sabot (1976).                  labour force. Persons not nieeded at all constituted
about 5C`. See Kao, Aknsclhel and Ficlher (1964).
107, LJnfuorLumatel , only the Philippines had scveral
surveys available, corresponding to different seasons  121. J. Buck(1930).
and therefore allowineg somne confidence that figures
used aie representative of year round input levels. 'The  122 J. Buck (1930).
Philippine data, available on a (luiarterly basis, do not
suggest niuch seasonalit) in average labour inputs in  123. S. Melhra (1966). The earlier study of Mujumdar
agriculture or elsewhere.                            (1961) had different and probably greater methodo-
logical difficulties. (See the discussion in Kao,
108. Whereas nearly   5S'.: of the formrier reported  Anschel and 1icher (1964)).
working more than 48 hr, only 45': of the latter did.
See 1,)NI: (1971).                                  124. S. %Tfira (19661.



L ABIOUR MARK IT PFRIORMAN('I IN DIX] I OPING CO(           N I RI I S            1237
125. The calculations involved otlher problems as well.  132. Across countries, one would anticipate that the
lor a discussion see A. K. Sen (1975).                 greater the availability of ol1'-a rni work relative to the
labour force on small farmiis, the lower would be the
126. More recent studies in India include those of K.  labour/;and and output/land differentials across farm
Ahuja (1973) and A. Rudha (1973). The former uses      sizes. This proposition does seem to be borne out.
standardized coeflicient of labour requirement per
acre for each major crop based on farm management      133. For m1nu111faCturing, relevant studies include J.
data related to a number of states, and a labour       Todd (1972). 1For comnmerce, thiey include A. Blhalla
coefficient for maintenance of animals. On this basis  (1970); A. Berry (1972).
labour needed is estimnated; it is compared Nwith
available labour on a 2400 hr per year norm. Rudha's   134. A. K. Sen (1962).
approach involves comparison of labour actually used
(assumed equal to reqtuired) with available person     135. Technically, it is not possible to distiniuiSh IOSS
hours. Nteither takes account of non-auriculltural activi-  due to misallocation  Of labour and that due to
ties or gets at questions of leisure preferences and   misallocation  of other factors; the factors are mis-
other possible determinants of work done.              allocated relative to each other. One can, however,
estimate the output gains from equating one or the
127. C. K. \ku and T. K. Lee 11963), as reported in A.  others' miiarginal productCisitY  aeross all producing
Koo (1969).                                             units.
128. Of the factor inputs.                             136. Based on data presented in A. Berry and M.
Ulrrutia (1976). Their estinmates for the Colombian
129. It would be implausible to expect them to rule    aeri:culttoral sector, as of the early 1960s, suggested
out the possibility of a moderate amount of suclh      that thie rouglhly one-quarter of the labour t'orce on
labour, e.g. 5  10-   of the labour force. Suclh coin-  farms of (1 3 hectares produced only a little over 10%
straining assumliptions as the samie productioln function  of thle outplut.
on aUl farms, honmogeneous factors etc. introduce too
much imprecision.                                       137. For a recenlt review see A. Berry and W. ('line
(1976).
130. G. R. Saini's '1969) recent studI for the Punliab
and U'ttar Pradesli states of India (for years 1955 56  138. Whiere MPl. is low in a given sector and MPK is
and 1956. 57) found an MN:IPL averaging about 20':     equal to or below    its level elsewhlere. the case is
higher on farms hirinig 51)  or more of their total    JifltLren :i; \tilaimatiion of output calls for the demise
labour than on farms hiring less than 25' M. Desai and  of some production units in that sector and the shift
D. Mazumdar (1970), on the other hand, separate pro-   of all factors to other sectors. The implicit as%uniptioll
duction functions for farms not using any hired labour  of most labouir surplus theory   that modern tech-
and for those using it. T he estimated marginal product,  ne   domtninates  led many analysts to expect this
for a West Bengal district in the mid-1950s. was posi-  case to be the typical one. IEmpirical work has
tive for the farms using hired labour but not signifi-  eradually eroded this position, and it is now recog-
cantly dillerent from zero forthoseusingpurely family  nized that traditional sector technology is frequently
labour. I or Colombia, Tllirsk eslinated MPL to be     efficient  under existing   factor proportinns. See
lower on small farmls tllan large ones, but still positiie.  Section 2A.
WV. Thirsk 11972).
139. See I). Turnham (1971).
131. lc ppm's aericultural sector hias been the context
for one discussioni of the implic.inions of wage move-  140. The special -elatiomislmip built uip between buyer
ments. Ihe l-' pidan case provides an int lerling test  and seller cointribuites to tlhis, wlhere the buyer mrust
ol labour surplus; it is a predominantly agricultural  every so often resort to credit, his est.miiflhed record
and nmall farm  econlomyv. As Hlansen has noted, 'the  wvith a seller fheips him with that setler but not much
first 1  ptx 'ian 5-year plan worked VOL11 tlhie j510 1mitinm0  with other ones.
that almost 25 '  of the rural labour ftorce couldi be
perminanenlth  rernoxed  trout aariculture'. A  debate  141. J. Rtamos ( 1974).
between F. B. Hansen (I969) and J. Hanson t197 1
focused on tile ICVel of' Uineinplm ii ei l. whethier the  142. P. 1 laLser 11961).
amount fot ot-farm uork donie by members ot tarmii
tlaniiie, constituited lubst.111til evidence that on-fairmii  143. A. Berry 11972).
MPL was about equal to the wage rate, and whether
seasonal and otlher v.iri.atiomns in the wvage rate dentoti  144. See 1'or example. A. Berry (1978).
strated that it was not non-market determined. In fact,
as noted above. the wage rate can be market delter-    145. See S. I'6iier, and M. Selm,%slkv (19761 for a
mined and variable over the seasons x'itlhut dislrom-  l'ornal e'.pol1itinn of tlhi. argument.
ing the existence of labour surplus in a sector wvhere
less than 40); of the agricultural labour force is      146. Trtie even if the Wrcunine model applie"s.
easssilied aN paid workers.



123 8                                 WORLD DEVELOPIMENT
147. Econometric estimiiates of the social returns to  social returns is more difficult to estimate but in
investment in educationi are not detlated to take   highly segmented labour markets may be quantita-
account of thle mismatch that results from segnienta-  tively  of greater significance. S. Pi'nera and M.
tion. Generally, only the gap between social and    Selowsky (1976).
private costs that arises from government subsidies of
the educational system  is taken into account. G.   148. See E. Edwards and M. Todaro (1974); R. Sabot
Psacharopoulos (1973). The gap between private and  (1977c).
R[*1 l:RENC lS
Ahluwalia, M. S., 'Thie scope for policy intervention',  Berry, A., 'Special problems of policy-making in a
in H. Chenery et al. (eds.), Redistribution wvith   technologically   heterogeneous     agriculture:
Growth (London: Oxford liniversity Press, 1974).    Colombia', in L. Reynolds (ed.), .griculture in
Ahuja, K., 'Agricultural underemployment in Rajas-    Dei'elopmrent TheorY' (New liaven: Yale University
than', ECconomic anid Political hi'ec/dy (September  Press, 1975c).
1973).                                           Berry, A., 'DisgLuised unemployment and the parti-
Alchian, A., 'Information costs, pricing and resource  DiPation rate in urban Colombia', Journal of
unemployment', in E. Phelps, et al., Micr-o E'co01o-  Econiomic Studies, Vol. 4, No. 2 (November 1977).
1iC Ic'oundations o.f Em'nplo yminent anud hnhu1alioln  Berry, A., 'A positive interpretation of the expansion
Thleovr' (New York: W. W. Norton & Company,         of urban services in Latin America with some
1970).                                              Colombian evidence', Journal of Development
Anderson, D., 'Development issues in rural non-farm   Stuidies, Vol. 14, No. 2 (January 1978).
employmilent', Developmient EcoIIoIIics Depart-  Berry, A. and W. Cline, 'Fartm size, factor productivity
ment, IBRI) (\V3shiingtoun D.C.: 25 April 1977).   and technical change in developing countries',
ArroNs, K., 'Highler education as a filter', JouriInal of  minico, IBRI) (January 1976).
PublicEc&onomics, (July 1973).                   B3erry, A. and R. Sabot, 'Labour miarket performance
Baer, W. and M. lIerve, 'Imployment and indus-        in dc-eloping countries', mimeo, IBIRL) (1976).
trialization in developing economnies', Quarteily  13erry, A. and M. Urrutia, Income Distibhution in
Journal of Elconoicsi, Vol. 80, No. 1 (february     Coloimibia (New  Ilaven: Yale University Press,
1966).                                             1976).
Barber, W. J., 'Economic rationality and behaviour  Bhagwati, J., 'The tlheory and practice of commercial
patterns of undeideveloped areas: a case study of  policy', Frank Gralham Memorial Lecture, 1967, in
African economic behaviour in the Rhodesias',      Special Papers int Iiieriiatiornal Econiomizics, No. 8
E'onolzic Developmnent antd Culitural Chlanzge, Vol.  (Princeton: 1966).
8, No. 3 (April 1960).                          13l1hl.i, A., 'The role of services in employment
Bardhan, K., 'A survey of' research on rural employ-  expaansion', Iii w'r,wiminal Labour Rev'iew, Vol. 101,
ment, wages and labour markets in India', mimeo,   No. 5 (NMay 1970).
IBRI) (ivlay 1976).                              Beinefeld, M. and R. Sabot, 'The National Urban
Barnum, H. and R. Sabot, 1figration, E'ducation and   Mobility  Enmployment and Income Survey of
Urban Suirplus Labouir, Employment Series No. 13   Tanzania (NUNFIST)', 3 volumes (Dar es Salaam:
(Paris: Development Centre, OECI), 1976).           Ministry of Economic Affairs and Developmnent
Barnum, H. and R. Sabot. 'I ducatiun, employment      Planning and Economic Research Bureau, 1971).
probabilitics and rural- urban inigration in Tan-  Blaug, M., ducation!i0 anid the Employment Problemi. if
zania', 0.Qjbrd Bulletini of Economics ancd Statistics,  Developing Counltries (Geneva; ILO, 1973).
Vol. 39, No. 2 (May 1977).                       Bliss, C. and N. Stern, 'I conunmic aspects of the
Becker, (., 'A theory of the allocation of time',     connectioln between productivity and consump-
1:Leonomic Jouiirnial, Vol. 75 (September 1965).   tion', miiirieo, I11RI) (1977).
Belhrinian, J. R., Supply Response in Underdeveloped  Bottotnley, A., 'Wage rate determination in under-
,.fgrir'ulffre (Amsterdam: North-Ilolland, 1968).   populated LinlderdCeVelCopedL rural areas', Ecotic*nic
Berg, FK., 'Iackw:crd-oluping labour supply functions in  hitcrnatihncal, Vol. 22, No. 1 (1969), pp. 1 14.
dual economy    -  the African case', Quartel.I  lBowcn, W. and T. I'iiineapai, Thle  Economnicy of
Journzal of kLcononics, Vol. 75 (August 1961).      Lahbouir Force Parricipaioi'c (Priiicetow: Princeton
Berry, A., 'Ulrban labour surplus and the commerce    llniversity Press, 1969).
sector', Yale Ecotnomic Growth Center Dliscussion  Breciier, R., 'Mininmum wage rates and the pure theory
Paper (1972).                                       of intem alion al trade', QuarierlY Journal of
lierry, A., 'Open unumployment as a social problem in  Economzics (February 1974).
urban Colombia: cicyrih and reality', Economic   Brigg, P., 'Some economic interpretations of case
Development andcl Culturlal (Change, Vol. 23, No. 2  studies of urban iiigrationi in developing countries',
(January 1975a).                                   \WVoking Paper No. 217, IBRD (March 1973).
Berry, A., 'Wage employment, dualism  and labour   Buck, J., Chlinlese Farm Economyv (ChicaVo: University
utilization in ('olunbia: trends over time', Journal  of Chlicago Press, 1930).
oJf Devloping Areas, Vol. 19, No. 4 (July 1975b).  Byerlee, D., 'Rural urban migrationi in Africa: tlieory,



LABOUR MARKET PERFORMANCE IN DEVELOPING COUNTRIES                        1239
policy and research implications', International  Surplus Economy  (Homewood, Illinois: Irwin,
Migration Reviev, Vol. 8 (Winter 1974).           1964).
Byerlee, D. and C. Eicher, Rural Employvmnent, AMigra-  Fields, G., 'The private demand for education in
tion  and Economic Development: Theoretical      relation to labour market conditions in less deve-
Issues and Em72pirical Evidence from Africa, African  loped countries', Economic Journal, Vol. 84, No.
Rural Employment Paper No. 1 (East Lansing:      336 (December 1974).
Department of Agricultural Economics, Michigan  Fields, G., 'Rural-urban migration, urban unemploy-
State University, 1972).                         ment and underemployment and job search activity
Byerlee, D. and J. Tommy, 'An tntegrated metho-     in LDCs, Jounial oj Developmnent Fconomnics (June
dology for migration research: the Sierra Leone   1975).
rural-urban migration survey', paper presented at  F[ields, G. and N. Marulanda, 'Intersectoral wage
Research  Workshop  on   Rural-Urban  Labour     structure in Colombia', Economic Growth Center,
Market  Interactions,  Development Economics     Yale University, Center Discussion Paper No. 251
Dept., IBRD (February 1976).                     (August 1976).
Chenery, H. and M. Syrquin, with the assistance of  Gersovitz, M., 'Aggregate demand, the wage gap, and
Hazel  Elkington,  Patterns  of Development,     unemployment in LDCs', Jourinal of Development
1950-70   (London: Oxford    University  Press,  Economnics, Vol. 1, No. 3 (December 1974).
1975).                                         Grant, J. P., 'Mlargina] man: the global unemployment
Cho, Y. S., Disguised Unemployment in Under-        crisis', Foreign Affairs, Vol. 50, No. 1 (October
developed Areas (Berkeley: University of California  1971).
Press, 1963).                                  Gregory, P., 'Wage structures in Latin America',
Clark, C. and M. Haswell, The Econonmics of Sub-    Journal of DevelopingAreas, Vol. 8 (July 1974).
sistence Agricuzltzure (London: NIacmillan, 4th  Grilliches, Z., 'Estimating the returns to schooling:
Edition, 1970).                                  some econometric problems', Eco,zo'netrica, Vol.
Cleave, J., 'Labour in the. development of African  45, No. 1 (January 1977).
agriculture: the evidence frorn farm  surveys',  Haberler, Gv., 'Some problemns in the pure theory of
unpublished Ph.D. dissertation, Stanford University  international trade', EconoImlic Joutrnal (June
(1970),                                          1950).
Collier, P., 'Labour mobility and labour utilization in  Hansen, B,, 'Employment and wages in rural Egypt',
developing  countries',  Oxford  Bulletin  of    Amiierican Econiomic Review, Vol. 59, No. 3 (June
Economics and Statistics Vol. 37, No. 3 (August   1969).
1975).                                         Hansen, B., 'Employmiient and rural wvages in Egypt -
Collier, P. and R. Sabot, 'On the measurement of    reply', Amtlerican Eco0nomic Review, Vol. 61, No. 3
rural-urban income differentials', mimeo, IBRD    (June 1971).
(1976).                                        Hanson, J., 'Employment and rural wages in Egypt: a
DANE:    Departamento   Administrativa  Nacional    reinterpretation', American Economnic Reviewt, Vol.
Estadistico, En7cuiesta dc Hogares, 1970 (Bogota,  61, No. 3 (June 1971).
Colombia: 1971).                               Harberger, A., 'Tsing the resources at hand more
Dean, E. R., 'Economic analysis and African responses  effectively', American Econ7omic Review, Vol. 49,
to price', Journial of Farm Economics (May 196S),  Supplemnent (May 1959).
pp. 402-409.                                  Harris, J., 'Wage rate determination with limited
Desai, M. and D. Mazumdar, 'A test of the disguised  supplies of labour in developing countries', Joturnal
unemployment hypothesis', Economica (February    of Development Studies (January 1971).
1970).                                         Harris, J., 'Migration task force -- draft proposal',
Domar, E., Essays in the Theory of Economic Growtlh  (Indonesia) Paper presented at Research Workshop
(New York: Oxford University Press, 1957).       on   Rural-Urban  Labour Market Interactions,
Dougherty, C. and M. Selowsky, 'Measuring the effects  Development  Economics  Department.  IBRD
of the misaliocation of labour', Review of Econo-  (February 1976).
mzics and Sratistics, Vol. 55 (August 1973),   Harris, J. and R. Sabot, 'Urban unemployment in
Dubey, V., 'The marketed agricultural surplus and   LDCs: towards a more general search model', paper
economic growth in underdeveloped countries',     presented as Research Workshop on Rural-Urban
Econiomic Journlal (December 1963).               Labour Market Interactions, Development Econo-
Durand, J., The Labour Force in Economic Develop-   mics Department, IBRD (IFebruary 1976).
Inent (Princeton: Princeton University Press, 1975).  flarris, J. and M. Todaro, 'Migration, unemiployment
Eckaus, R., 'The factor proportions problem  in     and development: a two-sector analysis', American
underdeveloped  areas',  American  Economic      Econzomzic Reviev, Vol. 60, No. 1 (March 1970).
Review, Vol. 45, No. 4 (September 1955).       Hauser, P., V"rhani:a.io'n in Latinh Am77erica (New York:
Edwards, E. and M. Todaro, 'Education and employ-   Interiiatiounal Documents Service, 1961).
ment in developing countries', in E. Edwards (ed.),  fielleiner, G., 'Smallllolder decision-making: tropical
EmploYment in Developing Nations (New York:       African evidence', in L. Reynolds (ed.),Agriculture
Columbia University Press, 1974).                il Development TheorY (New Haven: Yale Univer-
Falcon, W., 'Farmer response to price in a subsistence  sity Press, 1975).
economy: the case of West Pakistan', 1mericanl  Hopper, W. D., ':.llocation efficiency in a traditional
Economic Reviev, Proceedings (May 1964).          Indian agriculture', Journial of Farm  otnomics
Fei, J. and G. Ranis, Developmenst of the Labour    (August 1965).



1240                                 WORLD DEVlI:.O[)MENT
Hlymer, S. and S. Resnick, 'A model of an agrarian  Lewis, A., 'Economic development with unlimited
economy with non-agricultural activities', Americani  supplies of labour', The Manchester School (22
Economic Review, Vol. 59 (September 1969).         May 1954).
ILO, Towards F'ull Employment. A Programmiine Jier  Lewis, A., The Th1eory, of Econonic Growth (London:
Colombia (Geneva: ILO, 1970).                      George Allen and Unwin, 1955).
ILO, Matching Emnploy'ment Opportunities adl EEx,pec-  Lewis, H. G., Unzionismii and Relative WVages in the
tations. A  Progranmme of Alti     km  ('evion1    United States, 1963   (Chicago: University  of
(Geneva: ILO, 1971).                               Chicago Press, 1963).
ILO, Employ1minent, Incomes andCI EqualitY: A StrateiQy  Lim, D., 'I)o foreign companies pay higher wages than
/br Increasing Productive f&mplo.Vnient in Kenya   their local counterparts in Malaysian manufactur-
(Geneva: ILO, 1972).                              ing?' Journial of Development Lccononics, Vol. 4,
Ishikawa, S., Econzomiic Development in Asiain Perspec-  No. 1 (March 1977).
tive (Tokyo: Kiinokuniya l3ookstore Company,    Linder, S., The Harried Leisutre Class (New York:
1967).                                             Columbia University Press, 1970).
Johnsoni, I-I, 'Optimal policy intervention in the  Machllup, F., Education and Economic (;rowIth (New
presence of domestic distortions', in R. Baldwin ct  York: New York University Press, 1975).
al., Trade, Growth  aid Balantce of Paymevnts    Malone, C. C., 'Some responses of rice farmers to the
(Chicago: Ranid McNally, 1965).                    package programme in Tanjore District, India',
Johnson, H. and P. Mieskowski, 'The effects of       Journal of Farmt Econtomaics (M1ay 1965).
unionization on the size distribution of income: a  Mangahas, M., E. Recto and V. Ruttan, 'Price and
general equilibrium approachl', Quarterlyj Joutrnal of  market relationslhips for rice and corn in the
Econiotmzics, Vol. 84, No.4 (Noveniber 1970).      Philippines', Journal of Farm  Ecotnoics (August
Jones, W., 'Labour and leisure in traditional African  1966).
societies', Social Science Researclh C'otlcil Itemns,  Mazumdar, D., 'The urban informal sector', Woild
Vol. 22, No. 1. pp. 1 6.                           Development Vol. 4, No. 8 (August 1976).
Kao, C., K. Anschel and C. Eicher, 'Disguised unem-  MazumdaT, D., 'The urban labour market and income
ployment in agriculture: a survey', in C. Eicher and  distribtition in peninlsular Malaysia', mimeo, IBRD
L. Witt (eds.), Agriculture in Economic Develop-   (Noveniber 1977).
ment (New York: McGraw Hill, 1964).             Meade, J., 'Mauritius: a case study in Malthusian
Keesing, D., 'Employment and lack of emllploynment in  economics', Economnic Jourinal (September, 1961).
Mvlexico, 1900 1970', Researchl Memorandumn No.  Melbra, S., 'Surplus labour in Indian agriculture',
64, Center for Development Economics, Williams     Inidiani Et ononie Review, Vol. 1 (1966).
College (March 1975).                            Miller, R., 'The relevance of surplus labour theory to
Knight, J., 'Rural-urban migration and income com-    the urban labour markets of Latin America',
parison in Ghana', Oxjbrd Bulletin of Economics    Bulletin Itnternationial Institutte for JLabour Stuidies,
and Statistics (May 1972).                         No. 8 (1971).
Knight, J. and R. Sabot, 'Wages in the manufacturing  Miracle, M. and B. Fetter, 'Backward-sloping labour
sector of Tanzania', mimeo, IBRD (1977).           supply functions and African economic behaviour',
Koo, A., The Role of Laud Reform in Economic          Economic l:)Developmenit antd Cultiural Chanzge, Vol.
Development: A Case Studi- of Taiwan (New York:    18 (January 1970).
Praeger, 1969).                                  Modigliani, F., and E. Tarantelli, 'Generalization of
Krishna, R., 'Farm supply response in India -Pakistan:  the Philips curve for a developing country', Review
a case study of the Punjab region', Econzomic      of Econiomiic Studies, Vol. 40 (April 1973).
Journ-al (September 1963).                       Morawetz, D., 'Employment implications, of indus-
Krishna, R., 'Unemployment in India', Economic and    trialization: a survey', Economic Journal, Vol. 84,
Political Weekl1', Vol. 8 (3 March 1973).          No. 335 (September 1974).
Kuznets, S., Mocdeni Economnic    Growth: Rate,    Morgan, T., Income anid Einployrmvnt. Second Edition
Structure and Spread (New H1aven: Yale University  (New York: Prentice-Hall, 1952).
Press, 1966).                                   Morse, D., 'Unemployment in developing countries',
Layard, R. and G. Psacharopoulos, 'The screening     Political Secinec' Quarterly, Vol. 85, No. 1 (March
hypothesis and the social returns to education',   1970).
Journal of Political Econzomy (1974).           Mujumdar, N., Some Problems of Underemployment
Lecaillon, J. and D. Germidis, 'Economic development  (Bombay: Popular Book Depot, 1961).
and the wage share in national income', Inter.  Nelson, J., 'Peasants in city: miigrationl, urban poverty
national Labouvr Review, Vol. 3, No. 5 1Maya 1975).  and politics in new nations', ht'orlml Politics (April
Leibenstein, H., 'The theory of underemployment in    1970).
backward   economies',  Journlal  of' l'olirical  Nelson, R. and S. Winter, 'Factor price changes and
Economy, Vol. 65 (April 1957).                     factor substitticioni in an evolutionary model'; The
L.eibenstein, H., 'Allocative efficiency  versus X-  Bell Journal of Economics and Mafoanaoenmeni
efficiency', American  Ecoinomic Review  (June     Scicemcc, Vol. 6, No. 2 (Autumn 1975).
1966).                                          Norman, D. W., 'Labour inputs of farmers: a case
Leijonhlmf%ud, A., Otn Keynesian Economics and the    study of the Zaria Province of the Nortlh Central
Economics of Keynes. A     Stidm  itn .1lonu'arv   State of Nigeria', Nigeriant Journtal of Economnic and
Theory  (New  York: Oxford University PIess,       Social Situdies XI, Vol. 1 (1969).
1968).



L.Aliol SMARKET PERI 0RM.%I \NCF IN DEVI (WING COUNTRIES                    1241
Peacock, A. and F. Shaw, Piscal Policy and tlne     Sabot, R. H., The Social Cost of Urban Surpluis
Ellnpldn1;ncvii Problem  in D)eveloping Countries,  Labour (Paris: OECD Development Centre, 1977b).
EIimployment Series No. 5. (Paris: Development   Sabot, R. H., 'Education and thie labour market in
Centre, OFCD, l ebruary 1972).                      Thailand', mimeo, Development Econdmics I)epart-
Pearson, L. B., I)artners in Development, Report of the  ment, IBRD (September 1977c).
C'ommission on International D)evelopment (New   Sabot, R. H., Economic Development atnd Urban
York: Praeger, 1969).                               Migration: A Study of Tan2zania (Oxford: Claren-
Phelps, F, et al, Micro-Econonoic koundiations ot      doni Press, 1978).
EmploYment adlll 1nl7tii)n TheorY (New York: W.  Saini, G. R., 'Resource use efficiency in agriculture',
W. Norton & Company. 1970).                        Indian Journial of lgriciltnural Econlotmics, Vol. 24,
Pigou, A., Thle Tlteory oJ Unenplo iment (London:     No. 2 (April-June 1969).
Macmillan, 1933).                                Schultz, T. P., 'Rural-urban migratioii in Colombia',
Piniera, S. and M. Selowsky, 'Unempiloy iient, labour  Review of Econoinics and Statistics, Vol. 53, No. 2
market segmentation, thie opportunity  cost of     (May 1971).
labour and the social returns to education', World  Sclhultz, T. P., 'Notes on the estimation of migration
Bank Staff Working IPaper No. 233, IBRI) (.lune    decision functions', Paper read at Research Work-
1976).                                             shop on Rural---Urban Labour Market Interactions,
l'saelharopoulo', G., assisted by K. I i nchlille, Retutrnis  Developmenit  Economics  Department,  IBRD
to  Education: .Auz  Inzternlationial Comnpaison   (February 1976).
(London, New    York and Amsterdam: E.1sevier    Schultz, T. W., 'Education and economic growth', in
Scientific Publishing Company, 1973).              Nelson B. Henry (ed.), Social Forces Inlue-iung
Psa.chjrop,1losh, G., 'Schooling.  experience  and   .Amnerican  Ecdtucation  (Chicago: University  of
earnings: the case of an Ll)C'. Journial oflDevelop-  Chicago Press, 1961).
n7enlt k'cotonomies, Vol. 4, No. 1 (March 1977).  ScuLlly, J. and M. Toosie, 'lnterim report on sample
Ramos, J., 'A hleterodoxical interprertion of the     hlouselhold survey  on economic conditions in
employment problem   in Latin America'. (V'orld     Telhran and migration to Tehran', Paper read at
Detrloptcint, Vol. 2, No, 7 (July 1974).            Researclh  Workshop  on  Rural- Urban  Labour
Rao, V. K., 'Investment income and the multiplier in  market   initcraclions,  D)evelopment  Economnics
an umderdevcloped econonmy', in A. N. Agarwala      Department, IBRD (February 1976).
and S. P, Singh (eds.), Thle Economnics of Urban  Sen, A. K., 'An aspect of Indian agriculture', Econo-
Developmzenlt (London: Oxford U'niversity Press,   inkc  kl'c(kli, Vol. 14 (1962).
1970).                                           Sen, A. K., 'Peasants and dualism with or without
Reddaway, WV. B., 'Tlhe economics of underdeveloped   surplus labour', Journal of Political Econom1o'
countries', Thle LEcononzic Journral, Vol. 73, No.  (October 1966).
289 (March 1963).                                Sen, A. K., Enplovnlient. Teclhnolokv anzd Develop-
Rempel. H.. 'Labour migration into urban centres and )7en'1t (London: Oxford University Press, 1975).
urban unem1ployment in Kienya', Ph.D. dissertation,  Si.is,cmid, L., 'The costs and returns to human imigra-
lUniversity ot Wisconsin, Madifon (1971).          tioil', Journal ojfPolitical Ecoinoviv, Vol. 70, No. 5,
Resnick, S.. 'Thle decline of rural industry under    Part 2 (October 1962).
export expansion: a cornparison among Burmiia,   Solowv, R., 'A contribution to the theory of economic
Philippines, Thailand', Joturnial (of E.eonomic Hfis-  growth', QuarterlY Journal ofLEcon omnics (February
tor)r, Vol. 30, No. 1 (March 1970).                 1956).
Reynolds. L. G., 'Wa'ges and employnment in a labour  Stiglhtz. J. F., 'Alternative theories of wage determina-
surplus economy'. American Economnic Review,       tion and unemiployment in LDCs: I - the labour
Vol. 55, No. 1 (March 1965).                       turnover model; II - the efficiency wage model',
Reyniolds, L. G., 'Economic dcvelopmnnlt withl surplus  Quarterly Journ.al o,f Economics, Vol. 88, No. 2
labour: some comlplicatiolls' OVmI'nd Economic     (May 1974),
Papers, Vol. 21, No. 1 (March 1969).             Thias, H, and M. Carnoy, Cost-Benefit Analsis in
Reynolds, L. G. and P'. Gregory, Wages, Pr,,dtolu uiri, ., Education: A Case Study of Ken.ya, World Bank
and Iltdtstrializationi in Puerto Rico (l1mniewoud,  SLatl  Occasional Papers No. 14 (Baltimore and
Illinoki: Richard I). Irwin, IInc., 1965).          London: TeIC Johns ilopkins Press, 1972).
Richards, P. J., L'inplo, mlmnt alnd Unemplot)yinent in  I'hirsk, w., 'The economics of Colomrbian farm mechla-
Cejylon, Employment Series No, 3 (Paris: D)evelop.  nization', Ph.D. dissertation, Y'ale  University
ment Centre. ()I CDl) Novemiiber 19711.             (1972).
Robinson, J., 'l)is:ui'ied utFenmlmloy nmein', Econwnie  Elmorbeeke, I.. 'The em ploymen)t problem: a critical
Journal (June 1936).                                e.alumatinn of four ILO  comprehensive country
Rottenberg, S., 'On clhoice in Xihbour markets', Indus-  reports'. International Laboutr Review, Vol. 107,
trial andtl Laboutr 1% In lation. Review, Vol. 9 I 95(6).  No. 5 (Nim 1973).
RAudh.j, A.. '\arrinaflis, expl\mui1ions for more inten-    G drick  G., '\\aJge spillover and unemployment in a
sive labour input in ssnaller farms: empirical veri-  wage-gap economy: the Jamaican case', EcoIotmic
fication', Economnic anid Political Weeklly (June  Deverlopmlenit anid Cultutral Clhange, Vol. 23, No. 2
1973).                                             (January 1975).
Sabot, R. II., 'The meaning and meastiremtient of urban  Todaro, M., 'Income expectations, rural- urban migra-
stirplus  l.abour',  ()Iflm'I  Economic  Papers     tion and emnploymnen! in Africa', Inriernaliolal
(Nov'eniber 1977a).                                l.aboiur Rvricw, Vol. 104, No. 5 (No-vember 1971).



1242                                WORLD DEVELOPMENT
Todaro,  M., Internal Migration  in  Developing     analysis of the industrial sector', Ph.D. dissertation,
Countries (Geneva: ILO, 1976).                    Purdue University, Lafayette (1970).
Todd, J., 'Efficiency and plant size in Colombian  Winston, G., 'Capacity utilization in economic deve-
manufacturing', Ph.D. dissertation, Yale University  lopment', Economic Journal, Vol. 81, No. 321
(1972).                                          (MaTch 1971).
Turner, H. and D. Jackson, 'On the determination of  Woytinsky, W. S. and Associates, Employmenit atnd
the central wage level - a world analysis', Econo-  Wages in the United States (New York: Twentieth
inic Journal, Vol. 80, No. 320 (December 1970).  Century Fund, 1953).
Turnham, D., withl the assistance of 1. Jaeger, The  Wu, C. K. and T. K. Lee, Thte Demnalnd and Supply of
Emjiplo-yment Problem in Less DevelopedCotntries:  Farmn Labour of the Families of the Vocational
A Review of Evidence, Employment Series No. 1     Students (in Chinese) (Taichung: Chung Hsing
(Paris: Developmtent Centre, OECD, June 1971).   University, 1963).
Welsch, D. E., 'Response to economic incentives by  Yap, L., 'Internal migration  in less developed
Abakalike rice farmers in Eastern Nigeria', Joumal  countries, a survey of the literature', Bank Staff
of Farmn Econonmics (November 1965),             Working Paper No. 215, IBRD (September 1975).
WVhitaker, M. D., 'Labour absorption in Brazil: an



World Development Vol. 6, pp. 1243-1246                  0305-750X/78/1201-1243 S02.00/Q
© Pergamon Press Ltd. 1978. Printed in Great Britain.
Comment
HENRY J. BRUTON
Williams College
Messrs. Berry and Sabot have provided us  their conclusion that the labour markets do
with a comprehensive survey of labour markets  work satisfactorily.
in less developed countries. Their general con-  Given this way of asking the question, it is
clusion is that labour markets in such countries  not surprising that they find little evidence to
seem to function quite well indeed. The authors  support the view that simply reallocating labour
mean by 'functioning well' that labour responds  would result in a significant increase in output
to incentives in a predictable fashion, and that  in many developing countries. There are how-
the operation of labour markets does not seem  ever other questions that may be asked in
to have impeded the changing of economic     appraising the labour market. In particular in a
structure and the increasing of output that is  development context. one can ask liow  the
part of the development process. At the same  functioning of the labour market affects those
time they find many reasons why some mis-    aspects of the economy that Berry/Sabot
allocation of labour does exist, but that the  assume giveni.
costs of this misallocation have been exag-     Perhaps the most important question in this
gerated. They conclude that the frequently   regard is this: Does the labour market affect in
made exhortations to aim policy directly at the  any way the development and appliuation of
Iemployment problem' and away from     the   new technology in the economy? Of course we
achievement of a high rate of economic growth  know very little about this question, but there
rests on a misunderstanding of the factual   are some ideas around that merit attention.
situation. This is an important conclusion and  Two of these may be mentioned.
the authors have made their case carefully. In  Technological change imposes demands on
this note, I discuss three issues that are espe-  the labour force, but in turn the labour market
cially relevant to the Berry/Sabot review and to  can affect the nature of the technological
the conclusion that they reach. My comments  change that does in fact take place. The authors
are more in the nature of elaborations and   note that some firms (especially large ones) in
questions, rather than outright disagreements as  LDCs pay higher wages than the market would
I find their approach and general arguments to  seem  to require. This is done because it
be quite congenial.                          apparently  attracts a more reliable, loyal,
1. The  particular hypothesis that   the  productive worker (see below). The practice
authors investigate is a relatively narrow one:  also produces signals to the firm's management
namely 'given the existing technology, structure  and technicians that labour is expensive. Not
of preferences and stock of physical capital and  only then do firms tend to choose the more
land, no appreciable increment in aggregate  capital-intensive of existing techniques. but
economic welfare or its rate of growth is to be  more importantly they (may) tend to direct the
had by such reallocation of the labour force as  search for new techniques toward increasing
a more perfect labour market could bring     capital-intensity of production and away from
about'. Putting the issue in this way implies  new  labour using technologies. This is an
that the rest of the economy evolves indepen-  example of the general question, does the
dently of the labour market, and then members  operation of the labour market encourage the
of the labour force simply react to the pattern  search for new technology to proceed in the
of demand for labour created. Members of the  'right' direction. While the functioning of the
labour force thus seem to respond rather well labour market is certainly not the sole factor
to the various opportunities for employment  affecting technological search, it is an impor-
created. Where they do not - where there are  tant component. Especially foreign owned or
appearances of labour market failures - the  controlled firms seem  reponsive to labour
authors generally conclude that there exist  market considerations, and it is these firms
sufficiently acceptable explanations to justify  which are most likely to have technicians and
1243



1244                             WORLD DEVELOPMENT
managers who can search effectively for new  of the labour market. In economies where
technologies.                                unemployment   and   underemployment   are
A second general point to be mentioned has  widespread, labour aliocation decisions are
to do with labour's direct role in technological  often -- especially in the public sector -- based
development. To what extent does the func-   on political and equity considerations. Such
tioning of the labour market contribute to the  considerations are very much in order. On the
creation of a labour input that is an effective  other haild, given that employment is offered
carrier of n\ew knowledge. This notion is partly  for these reasons, the effective utilization of
a matter of training, and BerryiSabot refer to,  that labour becomes more relevant. In tlLese
but do not really analyse, the problems of   circumstances, there are of course many issues.
defining and implementing a wage policy in a  but the traditions, attitudes and practices of the
situation where labour productivity is rising due  labour force are surely important.
to on the job training. The wage problem       To summarize this point: It is important to
becomes even more complex when the training  ask the question that Bcrry, Sabot ask about
imposes a cost on the enterprise. Evidently. a  allocation of labour in a situation of given
policy that resolves current allocation problems  technology, structure of preferences. and stock
and that provides inducement to enterprises to  of physical capital. It is however equlally im-
offer and workers to seek the 'right amount' of  portant to ask whether or not the labour
training is desirable. It is not clear that labour  market fuinctions in a way that these -iveas are
markets in LDCs contribute as much to this  themselves affected. Of greatest importance in
issue as they might.                         this connection are the possibilities of the role
Of equal importance is a role for labour in  that the labour market can play in the develop-
capital formation in agriculture. Some attention  ment of new  technology and in the capital
hias recently been given to direct allocation of  formation process. In the medium  term, an
aaricultuLral labour to capital formatlon on the  effective performance bv the labour market in
farms. In some a-ricultuiral areas ot the world  these respects is probably more inportant than
such activity can be extremely productive.   allocation in the short-run conitext.
Given the seasonality of agricultural activity in  2. The second general question that may be
most   countries,  opportunities  for  yield-  raised by the BerryiSabot survey is concerned
increasing investments of this sort would appear  withl the way time is used by economic agents.
significant, and, for the most part, unexploited.  Economists are now  noting with increasing
Where such opportunities do exist, the way the  frequency that the usual threefold classification
labour market (and of course other things,   of the population into employed, unemployed,
particularly  price  policies  for  igriculltural  outside the labour force, is often not very
output) functions is of relevance. If hired  revealing in the LDC. Adding the category,
labour is used, the employer may need help in  underemployed, helps some but not a great
financing wage payments or the labour market  deal. Part of the problem is the large number of
might be made to work in a way such that wage  own account workers, who in any society fit
payments could be postponed until harvest. In  into the classification less well than do wage
this event the hired worker lends the farmer his  earners. Some of these own account people
services. Even for the own account worker    have productive and remunerative jobs, but
incentives may work in a way that such a     others are simply trying to eke out an existence
person does not spend time in these kinds of  and very much available for alternative employ-
capital accumulating activities. The possible use  ment. Also many people appear not to be in the
of labour in rural areas in these kinds of   labour market, but in fact are pursuing other
activities may be more productive and mote   activities because of the absence of employ-
easily implemented than are many rural indus-  ment opportunities. University students are an
trialization efforts.                        obvious example of this category. In some
The point is of considerable generality.  countries (e.g. Sudan) women are generally
Labour practices often result in idleness on the  classified as outside of the labour force, but do
part of workers wvhen there are numerous and  a great deal of the farm work while men move
obvious things that could be done to make    around the country in search of jobs. Also, as
living and working conditions a bit more     Berry, Sabot and others note, the very poor
pleasant as well as more productive. Govern-  cannot afford to be unemployed.
ment employment in particular often fails to   These examples suggest the generalization
utilize labour in as effective a way as is possible.  that  tinemplovment.  underemployment,
Why this is the case is due to many factors, but  outside the labour force categories do not
an important factor is often due to some aspect  measure that which is of greatest interest to the



COMIM IE N T                                 1 245
policy-maker, i.e. the availability of labour for  spend their time and their sources of survival.
economic activities and the welfare implications  3. Wage policy has become a major issue
of an existing allocation of the labour force.  both for the analyst and for the policy-maker.
Instead then of relying on this conventionial  BerryiSabot treat the subject in a quite com-
three-way classification, it may be of greater  plete way, and I wish to add only a complica-
use to focus on questions relating to how the  tion or two that I tlhink deserve more attention.
adult members of the society speiid their time.  rhe aLuthors attaclI importance to trade union
In addition, data on sources of income - as a  activity in pulilling wage rates up in certain
reward for work performed, from a claim based  industries. The exact role of the union however
on recognizable personal relationships, from a  is indirect. Strikes are qtlite infreluelnt in less
claim based on social relationships etc. -- will  developed countries, since unions rarely have
facilitate increased understanding of the weltfare  funds with which to conduct a stri;e. Union
implications of a given use of labour time.   pressure is largely political pressure, and must
Data on how and why the adult members of   be exercised through government policy or
a society spend their time the way they do is  threat of government action. Companies. es-
especially helpful. Such data contribLute to oUr  pecially foreign. Xre of course interested in
understanding  of the extent of 'voluntary'   keeping the government pleased, but to do this
unemployment as opposed to 'involuntary', the  may not always mean paying wages higher than
kind of incentives to which people respond, the  the market dictates.
k-ind of constraints that limit their activity, the  Why then do real wages in the modern sect.ir
way their actual activities contribute to the way  seem to rise in many countries in the face o.
the system functions, and how that contribu-  apparently ample supplies of labour? Berry:
tion may be enhanced within prevailing condi-  Sabot give considerable attention to the argu-
tions.                                        ment that higher wages result in increased
With these more complete data certain      labour productivity, a more reliable, loyal
policy issues become clearer. The mobility of  laboLur force. They also note that xllen xages
labour, the skills and experience of the labour  are pushed uLp for this reason, it is difficult to
force, the 'non-economic' demands on the adult  assess the cost to the economy of any mis-
population, and a variety of other matters are  allocations produced by such increases. This
illuminated. Such illumination then helps us to  argument is convincing in explaining why wage
understand more clearly than we now do the    rates may well be higher in certain key moctern
conditions under which rural industry can     sector activities, especially those dominated by
survive, what kind of transfer payments are   a few large firms. It is considerably less con-
effective, whether or not a given technology is  vincing in explaining why real wages tend to
known or is effective, obstacles to adaptation  rise over time in these sectors. Tlle argument
of new seeds that would raise the marginal   that wages rise in urban centres in response to
product of labour in agriculture. and on and on.  the rising prices of wage goods could of course
In the few instances where a more comprehen-  account for increasing nominal wage rates, but
sive survey on the use of time and the sources of  not real wages.
income has been taken, the results generally     The argument of Turner and Jackson, re-
show a different picture from that which one  ferred to by Berry;Sabot, is helpful, but not
sees  from  looking  only  at employmenti     completely satisfactory. The existence of 'higll'
unemployment/outside the labour force series.  wages in some modern sector activities results
In particular one begins to gain some more   in an increase in the economy-wide minimum
detailed insight into why the labour market   wage to maintain some sort of equity. Then the
works the way it does. It may also be noted   modern sector wages are pushed up to re-
that such detailed surveys seem to show more  establish the 'necessary' differentiaLl. This argu-
employment (and more output) and income       ment implies that the higher productivity and
than do the more conventional employment      the greater loyalty and reliability are due to the
surveys.                                      wage differential, not to the level of real wages.
The point of these observations is not so  Where the increased productivity is attributed
much to criticize the Berry/Sabot reliance on  to improved health and well-being, the differen-
the usual threefold approach, but rather to  tial argument is not evidently applicable. If
emphasize that a number of the analytical and,  emphasis is placed on the more subjective
especially, policy questions rest heavily on the  characteristics of loyalty and reliability, the
ambiguity of the notion of employment in an   explanation in terms of a differential rather
LDC. The elimination of these arabiguities    than level cannot be ruled out a priori, but it is
requires a greater understandiing of how people  difficult to believe that such would be the case



1246                             WORLD DEVELOPMENT
over extended periods of time. Also, of course,  There are other aspects of the Berry/Sabot
minimum wage laws apply in fact to a relatively  treatment of the wage issue that one could
small part of the labour force.              discuss, but the general point is this: The
The role of education is equally trouble-  satisfactory functioning of the laboutr market is
some. The emphasis the authors give to educa-  crucially dependent on the extent to which
tion as a screening device is probably warran-  wage rates reflect the true scarcity of the many
ted. On the other hand, there is evidence to  kinds of labour. The evidence and irguments
suggest that jobs often call for certain educa-  offered by Berry/Sabot plus the points in the
tional requirements, and then pay wages suffi-  preceding paragraphs suggest that wages in most
cient to yield a return on the cost of that  LDCs do not send out very accurate signals,
education. The education does not add to the  Attention to the source of these anomalies and
productivity of the worker, but it is allowed to  to the kind of policies that would correct them
justify the higher wage. One is paid higher  is perhaps the most important aspect of labour
wages because one has graduated from a univer-  market study at the present time.
sity, not because the university attendance
made one more productive. As education
becomes more common, these higher wages        Summnarv. The BerryiSabot survey is a
become more common. Governments are most     useful review of recent discussions of the labour
frequently guilty of this sort of thing. As  market in LDCs. In this comment I have tried
BerryjSabot note, it is difficult to estimate the  to widen their terms of reference a bit to say
cost of using educational institutions as a  something about the relationship between the
screening device to determine who gets these  labour market and technological -developmenit
positions. Casual observations. however, suggest  and the diffusion of the new techniology. Also I
that it is exceedingly high, as the cost includes  have elaborated on the wage issue in a couple of
maintaining the educational institutions as well  ways that seem to be particularly relevant in
as the cost of a distorted wage structure.   many developing countries.



World Development Vol, 6, pp. 1247-1249                    0305-750X/78/1201-1247 S02.00/0
Q Pergamon Press Ltd. 1978. Printed in Great Britain
Comment
MNICHAEL P. TODARO
Centre Jfr Policy StuLdies,
Tlhe Poptldatio,z Co unzcil
Professor Berry and Dr. Sabot are to be    populations will have for future problems of
commended for providing practitioners, re-   urban surplus labour; and, third, the general
searchers and students of development econo-  playing down of the seriousness of the relation-
mics with both a comprehensive and perceptive  ship between the burgeoning private demand
review of the burgeoning literature on labour  for formal educatioln, job rationing mechanisms
market performance in developing countries,  in the 'modern' urban economy and sector-
Their general 'Chicago School' thesis that urban  specific public investment allocations.
unemployment and underemployment is not as
critical a problem as some make it out to be
and that labour markets funlction at a com-     1. WONMEN iN THE LABOUR NIARKET
paratively high level of efficiency is an impor-
tant, if not totally   convincing, counter-     Except for a brief comment on female
argument to those who claim that problems of  labour participation rates, the authors have
urban surplus labour are serious manifestations  neglected in their survey to deal with an issue
of underdevelopment arising out of chronic   of growing importance in the field of Third
structural problems exacerbated by pervasive  World labour market studies - namely how
market imperfections, distortions and malfunc-  these markets funiction for (or against) women
tions. Considering  the importance of the    as opposed to men.' Although Berry and Sabot
structuralist, 'dependence' school of thought, it  deal extensively with the important issue of
is regrettable to note that in such an otherwise  segmfiented labour markets and employment
comprehensive survey the authors do not deal  stratification, they fail to mention or analyse
with recent Marxist critiques of the role and  sex-status as a criterion of stratification in both
functioning of labour markets in developing  job search and job allocation. And yet, the
countries.                                   interplay between sex as a variable of stratifica-
In spite of its attractiveness to economists,  tion and other criteria of group identity. race
Bermr and Sabot's argument is weakened to a  and education is critical to a more complete
certain extent by their tendency to disregard  understanding of human resource allocation in
those real world institutional, cultural and  developing nations.
political factors that induce a less than socially  For example, it has often been noted that
efficient allocation of human resources -    the structure of female opportunities and the
however well markets may function - and by   wage structure of the occupations in which
their general neglect of key   demographic   they tend to settle are determined directly by
considerations. I will restrict my comments to  the structure of male employment oppor-
three aspects of labour market problems in   tunities.2 Urban job opportunities available to
developing nations which the authors have    women, irrespective of their education and
either neglected to review in their surxey or  training, in situations of high and rising male
where they have only    provided a passi'tlg-  surplus labour are typically  restricted  to
comment or two. These issues are: first, the  unrirodvctive service activities in the 'informal'
authors' general neglect of the process whereby  sector. W-oren are in effect expelled from
qualified women are systematically discrimi-  access to modern sector jobs by high male
nated against in labour market allocations and  unemployment and underemployment. Their
the corresponding social costs of such dis-  widely observed lower levels of labour force
crimination; second, the cursory attention given  participation resulting from this job exclusion
by the authors to the important question of  are likely to provide a significanit downward
what effect rapid   population  growth  and  bias to official estimates of urban unemploy-
current age distributions of developing country  ment. Their activities tend to be concentrated
1247



1 248                            WORLD DEVELOPMENT
in  domestic services, vending, retail sales,  tions about rural labour absorption    and
prostitution or low-status clerical positions in  decelerated internal rural-urban migration, the
the public and private sector. Pervasive dis-  problem of efficient labour allocation in cities
crimination blocks female entry into a wide   and towns will surely remain as one of the
range of factory jobs (outside of textiles and  critical dilenmmas faced by Third World planners
electronics in southeast Asia) and, particularly  and policy-makers alike.
from those administrative, managerial, and re-
lated skilled occupations for which their train-
ing and4or natural abilities would qualify them.  S. EDUCATION. JOB RATIONING AND
Flrequently, men with lesser training and lower   PUBLIC RESOUIRCE ALLO('A[IOsN
productivity are selected for these activities,
therebv lowering potential output (and employ-   Among the most controversial and politi-
ment) beyond what non-discriminatory market   cally sensitive issues in the development econo-
forces would dictate.                         mics literature is the one that relates to the
Although the above problem is by no means  -growing recognition that public investment in
unique to developing nations. it is in general  formal education, particularly at the higher
more orornounced. MNoreover, the social costs of  levels, may represent a considerable mis-
this 'iniefficiencv' are likely to rise with time as  allocation of scarce financial resources.4 Many
more and rnore women gain access to educa-    autlhors, including Dr. Sabot, have described
tion. hut are denied access to those jobs for  and  docuimented  the widespread  'filtering
which their training qualifics them, Problems of  down' and 'credentialization' phenomenion that
urban surplus labouir are likely to be exacer-  clharacterizes labour markets in de'.eloping
hated in the coming    decade as increasing   countries.' In situations of urban labour sur-
nunmbers of educated, unattached women in     plus, both public and private, employers tend
Africa and Asia begin to dominate the rural-  to ration jobs on the basis of educatiolnal
urban migration streams. Wlhat is needed, there-  certification or more simply, years of com-
fore, is a new look at urban labour markets and  pleted  schooling  --  often  irrespective  of
rural--urban migration. one in which factors  whether or not such schoolinig is necessary for
that  y,termnicallv  and  exclusively  affect  dLS1tisfLtoLy job performance. Schooling thuts
women are taken into account.3                becomes a necessary criterion for .uCCeCSSfll
modern sector job search and given well-known
urban wage differentials, both by sector and
2. DENIOGRAPHIC DYNAMICS AND URBAN            occupation, the private, expected benefits of
LABOUR MARKETS                    more years of completed schooling greatly
exceed private costs. On the other hand, in the
While Berry and Sabot do on occasion       context of high and growing urban surplus
mention the problem of population growth for  labour, the social benefit/cost ratio of devoting
labour supply, they   fail to give sufficient  additional resources to  formal educational
attention  to this question when discuissinig  expansion as opposed to altcrnative uses such as
whether or not the problem of urban surplus   job creation is less clearly apparent. Hoowever,
labour is as serious an issue as many have    the political pressures arising out of intense
claimed. Given the acknowledged downward      private demand for more publicly financed
bias of urban unemplovmnent statistics, the   school places tend to dominate economic con-
continLeCd  influx  of rural migrants (with   siderations. with the net result that distorted
-rowing numbers of women irr the migration    labouir markets and arbitrary job 'rationing
stream), and the very high    proportion of   mechanisms may impose severe social costs not
current LDC populations under 15 years of age  adequatelv  described  in  the  Berry-Sabot
(averaging from 42 to 47"' in the developing  review.
world), it is clear that however well ITbour     Despite the above comments, this reader
markets may have ftunctioned in the past (and  found  the Berry Sabot paper provocative,
this, in spite of Berry and Sabot, is still a moot  extremelv informative, wvell-organized and, on
point), they will be increasingly strained to  the whole, generally persuasive in many of its
fuinction as well in the future. Since the size of  arguments, It should serve as a valuable intel-
the labour force for the next decade is already  lectual resource in the often complex and
determined by previous birth rates and cohort  confusing field of labour market studies in
sizes, even under the most optimistic assump-  develo'ping nations.



COMMENT                                           1249
NOTES
1. For an enlightening analysis of women and labour  Thadani and Michael P. Todaro, 'Towards a tlieory of
markets in developing nations, see Nadia H. Youssef,  female migration  in developing countries', paper
Women and Work in Developing Societies (Berkeley:   presented  at PopLulation  Association ot' America
Institute of international Studies, Population Mono-  Meetings, Atlanta, Georgia, April 1978.
graph Series, No. 15, 1974). See also Ester Boserup.
'Emplovment of wvomen in developing countries',     4. For an extensive analysis of investment in educa-
Proceedings of the htiternationial Population  Con-  tioin. see John Simmons (ed.). Investment in Educa-
ference, Vol. 1 (Liege: IL'SSP. 1973); and Guy      tion;. .atiional Strateqi' Options for Dei'elopin7g
Standing, Labour Force Participation and Develop-   Countries. (Washington, D. C.: IBRD. June 1976). See
,nent (Geneva: International Labour Office, 1978).  also E. Edwards and M. P. Todaro, 'Educational
demand and supply in the contexit ot growing tin-
2. An examination of this thesis can be found in Paula  employment in less developed countries, WVorld Deve-
E. Hollerbach, 'NMaternal employment and fertility: a  lopmnent, Vol. 1, Nos, 3 & 4 (1973).
new theoretical perspective', Center for Policy Studies.
The Population Council, February 1978 (mimeo), pp.  5. One of the best and most comprehensive analyses
18-23.                                              can be found in Ronald Dore, T)e Diplol1la Disease
(Berkeley: University ot California Press. 197h).
3. For a first attempt to analyse this issue, see Veena