Jorgenson (1967) - Surplus Agricultural Labour and The Development of A Dual Economy

Surplus Agricultural Labour and the Development of a Dual Economy
Author(s): Dale W. Jorgenson

Source: Oxford Economic Papers, New Series, Vol. 19, No. 3 (Nov., 1967), pp. 288-312
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/2662328 .
Accessed: 17/06/2014 12:42
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access to Oxford
Economic Papers.
http://www.jstor.org
This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

All use subject to JSTOR Terms and Conditions
SURPLUS AGRICULTURAL LABOUR AND THE
DEVELOPMENT OF A DUAL ECONOMY
ByDALE W. JORGENSON
I. Introduction
THE pointof departureforthis paper is the well-established empirical
associationbetweenthe degreeofindustrialization and thelevelofecono-
micdevelopment.Thisassociationcharacterizes bothtimeseriesdata for
individualcountriesand international cross-sectionsat a givenpointof
time.Highincomeperheadis associatedwitha relatively largeproportion
of the total populationengagedin industry.Low incomeper head is
associatedwitha predominance ofemployment sector.
in the agricultural
The processof economicdevelopment may be studiedas an increasein
incomeperhead or as an increasein theroleofindustrialactivityrelative
to thatin agriculture.Quantitatively speaking,grossdomesticproduct
percapitaforthe worldexcludingcentrallyplannedeconomiesincreased
from$334to $534in U.S. dollarsof 1958between1938and 1961. In the
sameperiodvalueaddedinindustrial activityincreasedfrom$117to $223.
Similarly,in 1961 grossdomesticproductper capita forindustrialized
countrieswas $1,486U.S. dollarsof 1958;thecorresponding figureforless
industrializedcountrieswas $132percapita. Value added in industry per
capitawas $666forindustrialized countriesand $36 forlessindustrialized
countries.'For bothtimeseriesand cross-sections highlevelsof output
are associatedwitha highdegreeofindustrialization.
The greatdisparityin degreeofindustrialization in developedand less
developedcountriesis mirrored by a bifurcationin theoriesof economic
growth.In modelsof a developedeconomythe analysisis concentrated
on the allocationof nationalproductbetweenconsumption and invest-
ment.Technology maybe characterized by fixedfactorproportions,as in
Harrod'smodelofeconomicgrowth,or by variablefactorproportions, as
I Statistical Officeof the United Nations, Department of Economic and Social Affairs,
The Growthof WorldIndustry, 1938-1961, InternationalAnalyses and Tables, New York,
United Nations, 1965, pp. 194-5. For more detailed support of the association between
degreeof industrializationand the level of economic developmentfortime series,the follow-
ing may be consulted: F. Dovring, 'The share of agriculture in a growing population',
MonthlyBulletin of AgriculturalEconomics and Statistics,8 (1959), pp. 1-11, reprintedin
Carl K. Eicher and Lawrence W. Witt (eds.), Agriculturein Economic Development.New
York: McGraw-Hill, 1964, pp. 78-98; S. Kuznets, 'Quantitative aspects of the economic
growthofnations, II. Industrial distributionof national product and labor force',Economic
Developmentand Cultural Change,5, Supplement (1957). Similar support forthe association
on internationalcross-sectionsmay be obtained from: C. Clark, The ConditionsofEconomic
Progress,3rd ed. London: Macmillan, 1957; H. B. Chenery,'Patterns of industrialgrowth',
American Economic Review, 50 (1960), pp. 624-54; Centre for Industrial Development,
Department of Social and Economic Affairs,A Study of Industrial Growth,New York,
United Nations, 1965.

D. W. JORGENSON 289
in Tinbergen's model.' Technological change may be embodied or dis-
embodied,neutralor biased. Similarly,the model may be closed by assum-
ing that investment is a constant fraction of national output or that
investmentdepends on the distributionof income as in Kaldor's theory
of economic growth.2For presentpurposes the similaritiesof theoriesof
growth for an advanced economic system are more importantthan the
differences.The industrialcompositionofoutput,that is, the proportionof
output generatedby industry,by agriculture,and by the remainingsectors
of the economy,is entirelyignored. The central featureof the process of
increasingincome per capita, namely, an increase in the role of industry
relative to that of agriculture,is left entirelyout of account.
By contrast, in models of a less developed economy the analysis is
concentratedon the relationshipbetween the growthof income and the
growthof population. One branch of the modernliteratureon theoriesof
development originated with the paper 'Economic Development with
Unlimited Supplies of Labour', by W. A. Lewis.3 Lewis postulates that
the fundamentalcharacteristicof certain less developed economies is the
existence of disguised unemployment. Lewis's analysis of the role of the
unemployedin the determinationofwdvages duringeconomicdevelopmentis
strictlyanalogous to that of Marx. Wages are tied to a subsistence level
so that agriculturaloutputper capita remainsconstantso long as disguised
unemploymentpersists. A second branch of the modern theoryof deve-
lopmnent originateswith Harvey Leibenstein's book of 1954.4 The central
resultofLeibenstein'stheoryis theexistenceofa low-levelequilibriumtrap,
a kind of Malthusian equilibrium of population and sustenance. The
Malthusian equilibrium level of income is stable for small changes in
income; to achieve sustained economic growthsomethinglike a massive
infusion of capital is required. As in theories of economic growth,the
industrialcompositionof output is entirelyignoredin theoriesof economic
development.
1 R. F. Harrod, 'An essay in dynamic theory',Economic Journal, 49 (1939), pp. 14-33;
J. Tinbergen, 'Zur Theorie der langfristigenWirtschaftsentwicklung',Weltwirtschaftliches
Archiv, 55 Band (1942), pp. 511-49, translated and reprintedas 'On the theory of trend
movements', in L. H. Klaasen, L. M. Koyck, and H. J. Witteveen (eds.), Janl Tinbergen
SelectedPapers, Amsterdam,North Holland, 1959, pp. 182-221.
2 N. Kaldor, 'A model of economic growth', Economic Journal, 67 (1957), pp. 591-624,
reprintedin Essays in Economic Stabilityand Growth.Glencoe: Free Press, 1960, pp. 259-
300; N. Kaldor, 'Capital accumulation and economic growth',in F. A. Lutz and D. C. Hague
(eds.), The Theoryof Capital. London: Macmillan, 1961, pp. 177-222; N. Kaldor and J. Mirr-
lees, 'A new model of economic growth',ReviewofEconomicStudies, 29 (1962), pp. 172-92.
3 W. A. Lewis, 'Economic developmentwithunlimitedsupplies oflabour', The MAanche8ter
School, 22 (1954), pp. 1039-91;W. A. Lewis, 'Unlimitedlabour: furthernotes', The ]Ianche8ter
School, 26 (1958), pp. 1-32.
4 H. Leibenstein, A Theoryof Economic-Demographic Development.Princeton: Princeton
University Press, 1954; see also: H. Leibenstein, Economic Backwardness and Economic
Growth.New York: Wiley, 1957.

290 SURPLUS AGRICULTURAL LABOUR AND A DUAL ECONOMY
It is widely recognizedthat under contemporaryconditionsmany less
developed countries have importantrelations with developed countries
eitherthroughinternationaltradeor throughtheestablishmentof a modern
'enclave' in an otherwisepurely traditional social and economic setting.'
Either relationshipgives rise to economic and social 'dualism' in which
a given economic or social system consists of two componentparts-an
advanced or modernsector and a less advanced or traditional sector. To
capturethe essence of dualisticdevelopmentit is necessaryto focus on the
association between the degree of industrializationand the level of eco-
nomicdevelopment. A theoryof developmentof a dual economy requires
a theoryof the industrialcompositionof output and its relationshipto the
level of economic development. The process of economic development
must be studied as an increase in the role of industrial activity relative
to that in agriculture. In the development of a dual economy these two
developmentsare intimatelyrelated.
The purpose of this paper is to present a theory of development of a
dual economy,focusingon the relationshipbetween the degree of indus-
trialization and the level of economic development. The theory of de-
velopmentof a dual economy has been approached withinboth classical
and neo-classicalframeworks.The chiefdifference betweenthese two ap-
proachesto the developmentof a dual economyis in conditionsgoverning
the supply of labour to the industrialsector. In the classical approach to
the theory,the real-wage rate is assumed to be fixed in terms of agri-
cultural goods; fromthe point of view of industrylabour is available in
unlimited amounts at a fixed real wage. In the neoclassical approach
labour is never available to the industrial sector without sacrificing
agriculturaloutput. From the point of view of the industrialsector the
real-wagerate rises steadily over time, depending on the rates of techno-
logical progress in both sectors and the rate of capital accumulation.
Disguised unemploymentis assumed to be non-existent. As Lewis points
out, a phase of development characterized by disguised unemployment
may be followedby a phase without unemployment: 'When the labour
surplus disappears our model of the closed economy no longer holds.
Wages are no longertied to a subsistencelevel.'2 Within both frameworks
it is possibleto examine aspects ofindustrializationofstrategicimportance
fordeveloping countriessuch as changes in the structureof output and
1 This point of view is elaborated in my paper, 'The development of a dual economy',

EconomicJournal,71 (1961), especially pp. 309-11. The same point of view is expressed by
Luigi Spaventa, 'Dualism in economic growth', Banca Nazionale Del Lavoro, Quarterly
Review,51 (1959), especially pp. 386-90. An excellent review of the literatureon economic
dualism through 1960 is given by Howard S. Ellis, 'Las economias duales y el progreso',
Revista de Economia Latinoamericana (1961), pp. 3-17.
2
Lewis, 1954, p. 176.

D. W. JORGENSON 291
employmentand changes in the rate of investment,capital intensity,and
factorsubstitution.
II. Development of a dual economy: a classical approach

In presentingthe classical approach to the theory of development of
a dual economy the essential assumption proposed by Lewis, unlimited
supplies of labour at a fixedreal-wagerate, will be retained. In the theory
of developmentof a dual economy,the economic systemmay be divided
into two sectors-the advanced or modern sector, which we will call,
somewhat inaccurately,the industrial sector or manufacturing,and the
backward or traditional sector, which may be suggestively denoted
agriculture. This terminologyhas been used by Lewis and by Fei and
Ranis as well as by the presentauthor.' It is clear that industryincludes
a good many traditional activities and that these activities have many
of the characteristicsof the backward sector; similarly,the agricultural
sector may include a relatively advanced sub-sector. Examples of the
formerwould include small-scaleindustryin Japan; examples ofthe latter
would include plantation agriculturein Asia and agriculturein areas of
European settlementin parts of Africa. Nevertheless,it is usefulto regard
the backward sector as mainly agricultural and the advanced sector as
primarilyindustrial.
Productive activity in each sector may be characterizedby a function
relatingoutput to each of the factors of production-land, labour, and
capital. The special character of the theory of development of a dual
economyis an asymmetryin the productiverelations. The output of the
traditionalsectoris a functionofland and labour alone; thereis no accumu-
lation of capital except in the formof land reclamation. This assumptionis
made by the present author and also by Lewis and by Fei and Ranis.2
Of course, other assumptions are possible. Even in relatively primitive
societies, there are importantuses of capital in agriculturalproduction.3
Capital is accumulated in the formof land reclamationand in the formof
equipment foragriculture,fishing,and hunting. In the study of primitive
societies, saving and investment,ownershipof property,and even credit
cannot be ignored. For present purposes, the assumption of no capital
in agricultureis useful. The essential distinctionis between agriculture
which uses capital produced in the advanced or modern sector and agri-
culture which uses only traditional formsof capital. We will referto an
1 Lewis, 1954, pp. 146-8; Jorgenson,1961, p. 311; J. C. H. Fei and G. Ranis, 'A theoryof
economic developmient', American Economic Review, 51 (1961), pp. 533-4.
2 Lewis, 1954, p. 146; Jorgenson,1961, p. 311; J. C. H. Fei and G. Ranis, Developmentof
theLabor Surplus Economy.Homewood: Irwin, 1964, p. 16.

3 See, for example, the essays in R. Firth and B. S. Yames (eds.), Capital, Saving and
Creditin Peasant Societies. Chicago: Aldine, 1964.

agriculturalsector utilizingmodern forms of capital as commercialized
agriculture.For presentpurposesthe special role of commercializedagri-
culture will be ignored. The resultingtheory of development of a dual
economyis of special relevance to the less developed countries.
It will be assumed that land is fixedin supply. Further,it is assumed
that agriculturalactivityis characterizedby constantreturnsto scale with
all factorsvariable. These assumptionsare made by the presentauthorand
by Fei and Ranis.1 Althoughthereare many ways to account fordiminish-
ing returns,e.g. decliningquality of land as more and more is put under
cultivationas in Ricardo's extensive margin-the initial assumptionthat
land is fixedin supply implies that the diminishingreturnsarise at the
intensivemargin of the Ricardian scheme. In the neo-classical theoryof
development of a dual economy it is assumed that the marginal pro-
ductivityoflabour in agricultureis always positive. In the classical theory
it is assumed that thereis some point at which the marginalproductivity
of labour becomes zero. If population exceeds the quantity at which the
marginalproductivityof labour becomes zero, labour is available to the
manufacturing sectorwithoutloss of agriculturaloutput. This assumption,
made by Lewis and by Fei and Ranis,2 will be retained in the present
versionof the classical theoryof developmentof a dual economy.
Land does not appear as a factor of production in the manufacturing
sector; the level of manufacturingoutput is a function of capital and
labour alone. In manufacturing,expansion ofproductiveactivityproceeds
withconstantreturnsto scale. This appears to be a reasonable assumption,
at least on the basis of evidence from the manufacturingindustries of
advanced economies.3 A second feature of the production functionsfor
agricultureand manufacturingis that each functionwill shiftover time so
that a given bundle of factorswill generatea higherlevel of output at one
date than at an earlierdate. In short,technologicalchange will be assumed
to take place in the mannerindicated by Tinbergenand othercontributors
to the neo-classical theoryof economic growth. A special problem arises
in applyingthis assumptionto the classical theoryofdevelopmentof a dual
economy. For simplicity,it will be assumed that the size of labour force
forwhich the marginal productivityof labour becomes zero remains the
same for all technological changes. Of course, the output of the agri-
cultural sector at this point increases over time as the agriculturalpro-
duction functionshiftsupward.
In the classical approach to the theoryofdevelopmentofa dual economy
populationgrowthis ignoredor shunted aside as a qualificationto the main
I Jorgenson,1961, p. 311; Fei and Ranis, 1964, pp. 15-16.
2 Lewis, 1954, p. 141; J. C. H. Fei and G. Ranis, 'Unlimited supply of labour and the
concept of balanced growth',Pakistan DevelopmentReview, 1 (1961), p. 30.
See Jorgenson,1961, p. 311, and the referencesgiven there.

D. W. JORGENSON 293
argument. Lewis discusses a demographictheoryquite similar to that of
Leibenstein,as outlined above. However, this demographictheoryis not
integratedinto the theoryof economic developmentin a satisfactoryway.
For Lewis's main line of argument it sufficesto assume that unlimited
quantities of labour are available to the industrial sector at a fixedreal
wage; an unlimitedsupply of labour may have its origin in population
growth,but population growthis not affectedby activity in either the
agriculturalorindustrialsectorsuntilthe phase ofdisguisedunemployment
is completed. A similarassumptionis made by Fei and Ranis: 'Population
growthwill be treated as a knownphenomenonexogenous to our model."
This assumptionmust be qualifiedin that so long as the real wage remains
fixed, the consumption of workers consists entirelyof products of the
agriculturalsector. In the wordsof Fei and Ranis: '. . . as a consequence of
the natural austerityconditionarisingfromthe same unlimitedsupply of
labor situation,muchindustrialoutput musttake the formofcapital goods
due to the absence of a domestic market forconsumer goods.'2 For sim-
plicity,it willbe assumed that so long as thereis disguisedunemployment,
population expands at the same rate as the growthof agriculturaloutput.
This is the only assumptionwhichis consistentwiththe view of Lewis and
of Fei and Ranis that the real-wage rate remains fixedand equal to the
initiallevel ofreal incomein the agriculturalsector. At this level ofincome
all of the income of workersin either sector is used for consumptionof
agriculturalproducts.
The chiefdifference between the classical approach to the development
of a dual economy and the neo-classical approach is in the conditions
governingthe supply of labour. In the classical theory,labour is available
to the industrialsector in unlimited quantities at a fixedreal-wage rate,
measured in agriculturalgoods. Lewis suggeststhat it is immaterialto his
argument whetherthe marginal productivityof labour in agricultureis
zero or simply less than the real-wage rate.3 Fei and Ranis distinguish
between phases of development in which the marginal productivity of
labour is zero and in whichthe marginalproductivityof labour is positive
but less than the real wage.4 In the firstof these phases labour may be
supplied to the industrialsector at no loss in agriculturaloutput; in the
second ofthesephases, labour may be supplied to the industrialsectoronly
at some sacrificein agriculturaloutput. In both phases labour is available
to the industrialsector at a fixedreal-wagerate only if the termsof trade
between agricultureand industryremainfixedand ifpopulation growthis
preciselyequal to the growthof agriculturaloutput. If the termsof trade
1 Fei and Ranis, 'A theory of economic development', 1961, p. 550.
2 Fei and Ranis, Developmentof a Labor Surplus Economy, 1964, p. 118.
3 Lewis, 1954, p. 142.

should turn against industry,a constant real wage (measured in agri-
culturalgoods) will imply a risingprice of labour relative to the price of
industrialgoods.
Finally, in the presentversion of the classical approach to the develop-
ment of a dual economy,it will be assumed that saving is equal to total
profitsin the industrialsector. This assumptionis consistentwith Lewis's
observationthat: 'We have seen that if unlimitedlabour is available at a
constantreal wage,the capitalistsurpluswillrise continuously,and annual
investmentwill be a risingproportionof the national income.'" As Lewis
emphasizes: 'Practicallyall saving is done by people who receiveprofitsor
rents. Workers'savings are very small.'2 The presentassumptionimplies
that agriculturalrents, in so far as they exist at all, are exchanged for
goods produced by the industrialsector. The agriculturalproductsrepre-
sented by these rents are then provided to the industrial workers. The
institutionalmechanismby which this transactiontakes place may vary
fromone economy to another. For example, agriculturalrents may be
taxed away and the proceeds spent on governmentallyfinancedinvest-
ment; alternatively,landlords may themselves invest in the industrial
sector, becoming industrial capitalists; finally,landlords may consume
goods produced by the industrialsector so that all investmentis done by
the ownersofindustrialcapital. For presentpurposesit sufficesto assume
that saving is equal to total profits in the industrial sector without
specifyingwhetherthe resultingaccumulation of capital is owned by the
government,the landlords,or the industrial capitalists.
We are now in a positionto lay out a more concreteversionofthe classi-
cal approach to the developmentof a dual economy. To begin the analysis
we consideran economicsystemin whichno developmentofmanufacturing
activity has taken place; all productive activity is concentratedin the
traditionalor backward sector. We will assume that there is some maxi-
mum quantityof labour which may be employedin the agriculturalsector
with positive marginalproductivity; the agriculturallabour force,say A,
is always less than this maximum quantity of labour. If we let Y be the
level of agriculturaloutput and L the fixed quantity of land available to
the economy,then a simple version of the production functionfor agri-
culture,characterizedby constantreturnsto scale withall factorsvariable,
is given by the Cobb-Douglas function:
Y eatLfA1-3,
where eat representsthe shiftfactor correspondingto technologicalpro-
gress. Changes in techniques are assumed to take place at a constant
percentagerate, oa. The constantfrepresentsthe elasticityof output with
I Lewis, 1954, p. 171. 2 Lewis, 1954, p. 157.

D. W. JORGENSON 295
respect to an increase in the supply of land; if the supply of land is fixed
it is possible to choose the originformeasuringthe passage of time so that
the productionfunctioncan be rewrittenin the simplerform:
Y -etA1-P.
For a total population in excess of the maximum quantity which may
be employed at positive marginal productivity, we may distinguish
between the labour force employed at positive marginal productivity,
say A, the agriculturallabour force, and the labour force which is re-
dundant, say R. Then total population is the sum of the agricultural
labour forceand redundantlabour:
P zA+R.
If we representthe maximum labour force which may be employed at
positive marginalproductivityby A+, then the agriculturallabour force
is the minimumof total population and this maximum labour force:
A=~ lnfp
A min A+'
Of course, if the agriculturallabour force is equal to total population,

disguised unemploymentis zero: if the agriculturallabour forceis equal
to the maximum level, A+, redundant labour is equal to the difference
between total population and this maximum level.
Under the assumptionsthat the rate of technologicalprogressin agri-
cultureis positive and that the maximum quantity of labour which may
be employed with positive marginal productivityis fixed over time, the
development of an economy in which all productive activity is con-
centratedin the traditionalor backward sector is simple to describe. At
a constant real-wage rate, measured in agricultural goods, population
increases at the same rate as agricultural output. In the presence of
redundant labour, the rate of growth of agricultural output and popu-
lation is constant and equal to the rate of technological progress in
agriculture,a. In the absence of redundant labour, population growth
can exceed the rate of technologicalprogresssince the rate of growthof
output is equal to the rate of technologicalprogress,a, plus the elasticity
of outputwithrespectto labour, 1 -, multipliedby the rate of growthof
population. With a constantreal-wage rate the rate of growthof popula-
tion is simply odd, a positive quantity. Hence, in an economy in which
there is no redundant labour initially,population will grow at a positive
rate untilthe maximum quantity of labour which can be employed with
positive marginal productivityis reached. After this point the rate of
population growthwill slow to the rate of technologicalprogress,a, and all
incrementsin population will become part of the redundant labour force.

We next consider an economic systemin which developmentof manu-
facturingactivityhas taken place. Conditionsof productionin the manu-
facturingsectormust be described. We have assumed that the production
functionin manufacturingexhibits constant returnsto scale. We have
also assumed that the output of manufacturedgoods fora given bundle of
capital and labour increasesover time. If we denote the quantityofmanu-
facturingoutput by X, the manufacturinglabour force by H, and the
quantityof capital by K, then a simpleversionof the productionfunction
formanufacturingis given by the Cobb-Douglas function:
X = eAtK0M11,
where eAtrepresentstechnological change, as before,and the constant a

representsthe elasticity of manufacturingoutput with respect to an in-
crease in the supply of land.
With respect to the supply of labour to the manufacturingsector, we
have assumed that redundant labour is available to the industrialsector
at a fixedreal wage, measured in agriculturalgoods. We may also assume
that the termsof trade between agricultureand manufacturingare fixed.
If we assume furtherthat competitiveconditionsprevailin manufacturing,
the marginalproduct of labour is equal to the fixedreal wage, measured in
eitheragriculturalor manufacturinggoods. This assumption is made by
Lewis and by Fei and Ranis.1 If we denote the fixedreal wage measured
in manufacturedgoods by w, the marginalproduct of labour in the manu-
facturingsector is then:
A3X X
AM(1 a)M W.
If thereis no redundantlabour, the marginalproductivityoflabour in the
agricultural sector may still be below the real-wage rate, measured in
agricultural goods. However, labour may be transferredfromthe agri-
cultural sector to the industrial sector only by sacrificingagricultural
output. Under these conditionsit may still be assumed that the termsof
trade between agricultureand manufacturingare fixed. This assumption
is made by Lewis.2 Alternatively,it may be assumed that the terms of
trade turn against manufacturing,so that the wage rate measured in
manufacturedgoods increases. This assumptionis made by Fei and Ranis.3
In the presentversion of the classical approach to the theoryof develop-
mentof a dual economy,the termsoftrade betweenagricultureand manu-
facturingcannot be determined endogenously. For simplicity,we will
beginwith Lewis's assumptionthat the termsoftrade are fixed; underthis
assumption the marginal product of labour in manufacturingis fixed.
1 Lewis, 1954, pp. 146-9; Fei and Ranis, 1964, pp. 16-19.
2 Lewis, 1954, p. 142. 3 Fei and Ranis, 1964, p. 209.

D. W. JORGENSON 297
Using the marginalproductivityrelationshipgiven above to eliminatethe
manufacturinglabour force from the production function for manu-
facturing,we may write:
X ()
\W/
e(A/U)tK.
If we assume that saving is equal to the share ofprofitsin the industrial

sector, ignoring depreciation we may set the rate of change of capital
equal to the share of profitsin manufacturingoutput:
K = aX,
so that the rate of growthof capital may be written:
K (1Wa)(lU)/U (A/U)t
Using the productionfunctionand the fact that the output per man re-
mains constant, the rate of growth of manufacturingoutput may be
written: _ -+_A )-{I )(l- A
X ar W
The rate of growthof manufacturingemploymentis, of course, equal to

the rate of growthof manufacturingoutput.
For an economy with total population in excess of the maximum
quantity which may be employed at a positive marginal productivityin
agricultureplus the manufacturinglabour force,thereis redundantlabour.
Total populationis the sum of the agriculturallabour force,the industrial
labour force,and redundantlabour:
P _ A+M+R.
The agriculturallabour forceis the minimumof total population less the
manufacturinglabour forceand the maximum labour forcewhich may be
employed at positive marginalproductivity:
A ~min(PM1
A-mA+
So long as thereis redundant labour in the agriculturalsector, manu-
facturingoutput and manufacturingemploymentgrow at a rate which is
positiveand increasing.Capital in manufacturingalso growsat a rate which
is positiveand increasing,but always less than the rate ofgrowthof output.
This implies that the capital-output ratio is always falling; a similar
resultis obtained by Fei and Ranis.1 Since agriculturaloutputis increasing
at a constantrate, equal to the rate oftechnologicalprogressin agriculture,
population is increasingat this same rate. Whatever the initial value of
1 J. C. H. Fei and G. Ranis, 'Capital accumulation and economic development', American
Economic Review,53 (1963), p. 288.

the rate of growthof manufacturingoutput,this rate of growtheventually
exceeds any fixedrate ofgrowth. The sum ofredundantlabour and manu-
facturingemploymentgrowsat a rate which exceeds the rate of growthof
population; but this rate must fall to the rate of growth of population.
Hence, the rate of growthof manufacturingemploymenteventually be-
comes so large as to forcethe rate ofgrowthofredundantlabour to become
negative and decreasing. Under these conditions redundant labour
eventually disappears altogether. This concludes the descriptionof the
firstphase of developmentwith unlimitedsupplies of labour. The point
at which redundant labour disappears is called the 'Lewis turning-point'
by Fei and Ranis.1
Afterthe Lewis turning-pointis reached the marginal productivityof
labour in the agriculturalsector is positive but less than the real-wage
rate, measuredin agriculturalgoods. Under the assumptionthat the real-
wage rate remainsfixedwhen measured in agriculturalgoods, the rate of
growthof population is equal to the rate of technological change in the
agriculturalsectorless the elasticityof agriculturaloutput withrespectto
labour multipliedby the rate of decline of the agriculturallabour force.
Where wAis the proportionof the agriculturallabour forcein total popu-
lation and w31ithe proportion of the manufacturinglabour force, this
condition on the rate of population growthimplies:
A A
=A wx
AA Wm
+W
or,simply: A _ W31IV
o+(W31r_9)
For this conditionto be satisfied,the manufacturingproportion,WM, must

be such that the rate of growthof the agriculturallabour forceis negative
at the Lewis turning-point.Furthermore,the rate of growthof the agri-
cultural labour force must remain negative until the labour force itself
reaches the level at which the marginal product of labour in agriculture,
measured in agricultural goods, is equal to the real wage. For this it
sufficesto assume that the share of manufacturingin the total labour force
exceeds the elasticityof agriculturaloutput with respect to land. Under
this condition the agriculturallabour forcedeclines at an increasingrate
untilthe marginalproductoflabour is equal to the real wage. At thispoint
a third phase of the development of a dual economy is reached. In this
phase the wage rate of labour is the same in agricultureand in manu-
facturing.
The third phase of developmentof a dual economy under the classical
approach is described by Lewis as follows: 'When capital catches up with

D. W. JORGENSON 299
labour supply, an economyentersupon the [third]phase of development.
Classical economics ceases to apply; we are in the world of neo-classical
economics,whereall the factorsof productionare scarce, in the sense that
their supply is inelastic. Wages are no longer constant as accumulation
proceeds; the benefitsof improvedtechnologydo not all accrue to profits;
and the profitmargindoes not necessarilyincrease all the time.... 1 Fei
and Ranis describethe thirdphase as follows: 'The transitioninto phase
[three] constitutesa major landmark in the developmentalprocess. With
the completionof the transferof the disguisedlyunemployed,there will
occur a switch, forced by circumstancein employer behavior, i.e., the
advent of a fullycommercializedagriculturalsector. This landmark may
be definedas the end of the take-offprocess. We know of no otherway to
establish a nonarbitrarycriterionfor an economy reachingthe threshold
of so-called self-sustaininggrowth.'2 The basic point made by Lewis and
by Fei and Ranis is that a neo-classical theoryof growthforan advanced
economy applies afterthe third phase of developmenthas been reached.
Hence, furtherdiscussion of this phase will be postponed until the neo-
classical theoryof developmentfora dual economyhas been discussed.
Parenthetically,it should be remarkedthat Fei and Ranis attempt to
combine Lewis's notion of disguised unemployment with the critical
minimumefforthypothesis of Leibenstein. Their criterionfor a critical
minimumeffortis that the rate of growthof population must be less than
the rate of growthof the industriallabour force. In the presence of dis-
guised unemployment,this condition is always satisfied,provided only
that the rate of technologicalchange in the industrial sector is positive.
With a positive rate of technological change the rate of growth of the
industriallabour forceeventually exceeds any fixedrate of growth; with
a fixedreal wage, measuredin agriculturalgoods, the growthofpopulation
is limitedby the rate of technologicalchange in the agriculturalsector. In
the absence of disguised unemployment,the critical minimum effort
criterionis satisfiedonlyunder a somewhatdifferent set of conditions. We
will returnto the discussion of this problem afterour review of the neo-
classical theory.
III. Development of a dual economy: a neo-classical approach3

The distinguishingcharacteristicsof the neo-classical theory of the
development of a dual economy are the technology of the agricultural
sector and the conditions governingthe supply of labour. First, in the
neo-classical approach it is assumed that the productivityof labour in
'
Lewis, 1954, pp. 26-27.
2 Fei and Ranis, 'A theoryof economic development', 1961, p. 537.
3 This section is based on my paper 'The development of a dual economy'.

agricultureis always positive so that labour is never redundant. Secondly
it is assumed thatthe real-wagerateis variable ratherthan fixed; wage rates
in the backward sector are assumed to be proportionalto those in the
advanced sector. The interpretationof this relationshipwill be discussed
below. Except for the possibility that labour may be redundant, the
descriptionof technologyforthe agriculturalsector is the same for both
classical and neo-classical theories. In the neo-classical approach there is
no level ofthe agriculturallabour forceat whichthe marginalproductivity
of labour is zero. It is assumed that the agriculturalproductionfunction
forany level of the agriculturallabour forcemay be characterizedby the
Cobb-Douglas productionfunction:
Y eatLfAl-P,
where variables and parameters have the same interpretationas in the

classical approach. Assuming that the supply of land is fixed,this pro-
duction functionmay be rewrittenin the form:
y - eatA-3,
wherey = Y/A is agriculturaloutput per head.
Conditionsof productionin the manufacturingsector are the same as
those of the classical theory. We have assumed that the manufacturing
productionfunctionmay be characterizedby constantreturnsto scale and
that the output of manufacturedgoods fora given quantity of capital and
labour increases over time; a simple version of the manufacturingpro-
duction functionis the Cobb-Douglas function:
X =e tKUM1-or
where variables and parameters have the same interpretationas in the

classical theory. Second, we assume that saving is equal to the share of
profitsin the manufacturingsector; as before,we ignore depreciationso
that the rate of change of capital may be set equal to the share of profits
in manufacturingoutput: K aX
This assumptionis identical to that made in the classical approach.

To close the model forthe neoclassical theoryof the developmentof a
dual economyit is necessaryto describe the allocation of labour between
the backward and advanced sectors of the economy. To simplifythe dis-
cussion we will assume that as agriculturaloutput per head increases, all
output is consumed up to a level of agriculturaloutput per head equal to
the critical value, y+. We assume that once the critical value is attained
all furtherincreasesin consumptionperhead take theformofmanufactured
goods. Under these assumptions agricultural output per head in excess
of the critical value, y+, constitutes a surplus; we may define the

D. W. JORGENSON 301
betweenagricultural
agriculturalsurplus per head, say s, as the difference
output per head and the criticalvalue, y+:
s = y -y.
If agriculturaloutput per head exceeds the criticallevel, part ofthe labour

forcemay be released fromthe land to produce manufacturedgoods with
no reductionin the rate of growthof total population.'
As before,we denote agriculturalpopulation by A and manufacturing
population by M; total population, say P, is the sum of these two com-
ponents: P = A+M.
The demographic theory for the development of a dual economy is as
follows: The net rate of reproductionis the minimumof the rate corre-
sponding to the minimumforce of mortalityE and a rate which corre-
sponds to output offoodper head; the basic demographicrelationshipmay
be written: A
=min('YP
where yA/P is output of food per head for the whole economy and 8 is
the minimumnet reproductionrate equal to the maximum possible force
of mortality(mass starvation) and E is a fixedbirth-ratethat depends on
medical technique and social institutions. For an economy with an agri-
culturalsurplus,total food consumptionis the criticallevel, y+, multiplied
by total population; the proportionof the total labour forceemployedin
agricultureis the ratio of this critical level of agriculturalproductionper
head to the actual level of output per head:
A
y+y~_ A
Y P
Of course,this relationshipholds only when an agriculturalsurplus exists,
that is, if y > y+. Under these assumptions,the relationshipgoverning
the distributionof labour between the backward sector and the advanced
sector may be representedby:
A .{
A-= min{'/.
To study the developmentof a dual economyforthe case in which the

1 The relationshipbetween the existence of an agriculturalsurplus and development of
the advanced sectorhas been discussed by Lewis and by Fei and Ranis. This relationshipis
also discussed by N. Kaldor in his paper, 'Characteristics of economic development',
in Essays on Economic Stabilityand Growth.Glencoe: Free Press, 1960, pp. 233-42. The
necessityofan agriculturalsurplushas been emphasized by William H. Nicholls in his papers,
'The place of agriculturein economic development',in Eicher and Witt (eds.), Agriculturein
Economic Development,pp. 11-44, and 'An agricultural surplus as a factor in economic
development', Journal of Political Economy, 71 (1963), pp. 1-29.

advanced sectoris economicallyviable, we must assume at the outset that
an agricultural surplus eventually emerges, that is, that oX -e> 0,
whichis both necessaryand sufficient forthe emergenceof an agricultural
surplus. The case in which the advanced sectoris not economicallyviable
will be treated subsequently. We assume firstthat the initial level of
agriculturaloutput per head is below the critical level, y+. An industrial
labour force comes into being when agriculturaloutput per head attains
the critical value, y+, that is, when agricultural output attains the
minimum level necessary for population to grow at its maximum rate.
From this point forward,population grows at the maximum rate of net
reproduction,E.
From the fact that population is growingat a constant rate and that
consumptionof food per head is stationary,we obtain the followingex-
pressionforthe growthof the agriculturallabour force:
A = P(0)e[-!10-9ht = A(0)e[E6!1-P't.
Agricultural population may grow, decline, or remain constant, de-
pending on the magnitude of the parametersE, the rate of growthof total
population, and a, the rate of technologicalprogressin agriculture.
The manufacturingpopulation is equal to total population less agri-
cultural population; hence the growthof the manufacturinglabour force
is governedby the followingexpression:
M P(O)[eEt e[1E-1-Plt],
whichis zero at time t 0 and growsat a rate whichis always morerapid
than the rate of growthof total population. To show this we begin with
the assumption that an agriculturalsurplus eventually emerges,namely:
aY-E > 0,
which implies: E-C- < E(I-A),
so that: e 1-a.
The rate of growthof population is greaterthan that of the agricultural

population alone; hence the manufacturinglabour forceis growingat a
rate which exceeds that of total population. The rate of growth of the
manufacturinglabour forceis always decliningand approaches, as a limit,
the rate of growthof population, E.
To study the growth of manufacturing output, it is necessary to
characterizethe process of capital accumulation in the advanced sector of
the economy. The fundamentalrelationshipsinclude the expressiongiven
above for the growthof the manufacturinglabour force,the production
functionforthe manufacturingsector,and the savings function. Combin-
ing these relationshipswe may eliminatethe output of the manufacturing

D. W. JORGENSON 303
sectorand the manufacturinglabour forceto obtain a differential equation
in capital alone:
It = KaP(O)1-aeAt[eet-e[16-0/1it]1-a,
whichis the fundamentaldifferential equation forthe neoclassical theory
of developmentofa dual economy. From thisfundamentalequation it may
be deduced immediately that there is no stationary situation for any
economy in which the advanced sector is economically viable; that is,
provided that there is a positive and growingagriculturalsurplus, the
advanced sectormust continueto grow. The patternof growthof the ad-
vanced sector is determinedby two initial conditions,the size of total
population at the time that the growthof the advanced sectorbegins and
the size of the initial capital stock. Only the initial size of the population
has any effecton the long-runpattern of growth of the economy; the
influenceof the initial size of capital stock eventuallydies out.' Secondly,
it may be shown that there is no critical minimum level of the initial
capital stock requiredforsustained economic growth. Given any positive
initial capital stock, no matterhow small, the existence of a positive and
growingagriculturalsurplus generatessustained economic growth.
For the neo-classicaltheoryofthedevelopmentof a dual economycapital
and output grow at the same rate in the long run, namely,A/(1-oj)+-E,
whereAis the rate of technologicalprogressin industry,1- is the share
oflabour in manufacturingoutput,and Eis therate ofgrowthofpopulation.
Population growsat the rate c; since the share of labour in manufacturing
output is constant, the wage rate of the manufacturinglabour force
eventuallygrowsat the rate A/-1 . In the shortrun the beginningof the
growth of the advanced sector is always characterized by a 'big push',
that is, an extraordinarilyhigh rate of growthof manufacturingoutput.
From the viewpoint of the neoclassical theory of the development of a
dual economy,such a high initial rate of growthmay be interpretedas a
statisticalartifact. Using the productionfunctionforthe advanced sector,
we may derive the relation:
IX A+r N+(I-o)
so that the rate of growthof manufacturingoutput is equal to the rate of

technologicalprogressplus a weighted average of the rates of growthof
capital stock and of the manufacturinglabour force. But the initial rate
of growth of the manufacturinglabour force is essentially unbounded;
this rate of growthdeclinesgradually,approaching a long-runequilibrium
value equal to the rate of growthof total population. The existence of a
statistically observable 'big push' is no evidence for the necessity of a
1 A proofof this propositionis given in 'The development of a dual economy', pp. 330-3.
4520.3 Y

massive infusionof capital fromoutside the system for a 'take-off'into
sustained growth; sustained growth depends on the economic viability
of the advanced sector and not on the initial level of capital stock. The
advanced sectoris economicallyviable if and only ifthereis a positive and
growingagriculturalsurplus.
We have assumed that wage rates in the backward sectorof a dual eco-
nomy are proportionalto those in the advanced sector. Using thisrelation-
ship and the saving functionit is possible to determinethe termsof trade
between agricultureand industry. The balance of trade between agri-
culture and industry requires that the value of labour income in both
sectorsis equal to the value of manufacturingoutput not used for addi-
tions to capital togetherwith the value of total agriculturaloutput. This
balance relation may be written:
wM+HpwA = (1-a)X4+qY,
whereq is the termsof trade betweenagricultureand industryand P is the
constant of proportionalitybetween wage rates in the agriculturalsector
and wage rates in the industrialsector.
The constant of proportionalitymay be interpretedin a number of
differentways. First, in a 'strict' neoclassical theory wage rates in the
two sectors must be equal. In this case the constant of proportionality,
[k,is unity. Alternatively,ifthe process of developmentof a dual economy
is characterizedby a steady flowof labour fromagricultureto industry,
a differentialbetweenagriculturaland industrialwages may be requiredto
sustain this flow.' As a third alternative,if land is owned by the culti-
vators but the full value of the land cannot be realized by outrightsale,
the industrialwage rate must be sufficiently highto cover both labour and
propertyincomefora memberofthe agriculturallabour force.2If nothing
can be realized by the sale of land, the industrialwage rate would have to
be equal to unity divided by the share of labour in total agriculturalout-
put. Otherinterpretationsof the constantof proportionalitycould doubt-
less be given. Provided that p.is a fixedconstantthe balance relationmay
be rewrittenin the form: juwA q Y,
so that: w+A -+
and: = [1z: ] +w
In the long run the rate of growthof the wage rate in manufacturingis
equal to A/1-or, so that the rate of growthof the termsof trade is the sum
I See, forexample, 'The development of a dual economy', pp. 322-3.
2 Lewis, 1954, pp. 148-9.

D. W. JORGENSON 305
of a negative and a positive quantity; hence the termsof trade may turn
in favourof agricultureor industry,dependingon the relative magnitude
of the two quantities.
IV. Beyond disguised unemployment

Where the advanced sector is already in existence, wage rates in the
advanced and backward sectorsmay be taken to be equal, as in the 'strict'
neo-classical approach. Then the neo-classical theoryof the development
of a dual economy may be reinterpretedas a theory of the neo-classical
phase of Lewis's theory of economic development. The growth of the
manufacturinglabour force and manufacturingoutput and the accumu-
lation of capital are described by the relations given above for the neo-
classical theory. However, the initial phases of the development of the
advanced sector are not the same as in the neo-classical theory. In the
classical theorythe phase of redundant labour initiates the development
of manufacturing.This sector develops furtherin the phase of disguised
unemployment,where there is no redundant labour but the marginal
product of labour in the agriculturalsector is below the real wage rate,
measured in agriculturalgoods. Finally, the marginalproducts of labour
in both sectorsare broughtinto equality withthe fixedreal-wagerate. By
this time a certain amount of capital has been accumulated in the manu-
facturingsector. Given the manufacturinglabour force,the second initial
condition for the fundamental relations of the neo-classical theory of
developmentof a dual economy,namely,the size of total population when
agriculturaloutput per head reaches its criticalvalue, y+, can be computed
by insertingthe manufacturinglabour forceinto the equation:
31(t)-P(O)[eft-eIE6Ilf- t],
by insertingtotal population into the equation:

P(t) P(O)eEt,
and by computing P(O) and the origin for the measurement of time.
These constants may then be used to determinethe course of economic
growthin the neo-classical phase of the classical theory of the develop-
ment of a dual economy. Of course,the fundamentalrelationsof the neo-
classical theoryare valid forthe classical theoryonly after the beginning
of the neo-classicalphase.
Up to this point we have consideredonlythe case in whichthe advanced
sectoris economicallyviable. A necessary and sufficient conditionforthe
economic viability of the advanced sector is the eventual emergenceof a
positive and growingagriculturalsurplus. Provided that an agricultural
surplus eventually emerges,the development of a dual economy may be
characterizedin two ways. If there is disguised unemploymentas in the

classical approach, the manufacturingsector develops in three separate
phases. First,manufacturingoutput and employmentgrowat a rate which
is positive and increasing. Capital in manufacturingalso grows at a rate
which is positive and increasing,but always less than the rate of growth
of manufacturing output. Redundant labour eventually disappears.
Secondly, provided that the share of manufacturingin the total labour
forceexceeds the elasticityof agriculturaloutput withrespectto land, the
agriculturallabour forcedeclines at an increasingrate until the marginal
product oflabour is equal to the real wage in both sectors. The realization
of this conditionmarks the end of disguised unemployment. Finally, the
manufacturingsector enters on to the neoclassical phase. This phase is
the same as the phase of 'dualistic' developmentin the neo-classicaltheory,
provided that the initial conditions of the fundamentalrelations of the
neoclassical theory are properlyreinterpreted. If there is no disguised
unemployment as in the neo-classical approach, the backward sector
develops accordingto the fundamentalrelations describingan increasein
agriculturaloutput per head untilthe criticallevel, y+, is reached. At this
level the forceof mortalityreaches its minimumand the net reproduction
rate fortotal population reaches its maximum. From this point forward
the development of the manufacturingsector is described by the funda-
mental relations for capital accumulation and for the growth of manu-
facturingoutput and employment. These relations are the same as those
describingthe neo-classicalphase of developmentof a dual economyin the
classical approach.
We may now consider the case in which the advanced sector is not
economically viable. First, we will describe the neo-classical theory of
developmentforthis case. If capital forthe advanced sectoris already in
existence,the conditionforeconomicviability of this sector,oC- 3E> 0, is
not satisfied. There are two possibilities. First,suppose that a- /E; then
the manufacturinglabour forceis equal to zero and thereis no manufactur-
ing production. Secondly,suppose that a < PE and the initial value of the
manufacturinglabour forceis positive. Then this labour forcedeclines to
zero after which there is no furthermanufacturingproduction. Total
population becomes entirelyconcentratedin the agriculturalsector and
agriculturaloutput per head eventuallydeclinesto that associated withthe
low-level equilibriumtrap.
The classical theoryof development where the advanced sector is not
economically viable at the maximum net reproductionrate is somewhat
more complex. We consider development only in the third or neo-classi-
cal phase. In this phase the developmentof a dual economy is character-
ized by the same fundamentalrelations as in the neo-classical approach.
If the advanced sector is not economically viable, two possibilitiesexist.

D. W. JORGENSON 307
First, if a /E, the existence of a positive manufacturinglabour force
contradicts the fundamental differentialequation for the neoclassical
theory of development of a dual economy. Hence, for the classical ap-
proach this conditionmustbe ruledout by assumption. Secondly,if ace,<
the manufacturinglabour forcebegins to decline as soon as disguised un-
employmentis eliminated. This decline continues until the agricultural
labour forcereaches its maximum level, so that furtherincreases in the
agriculturallabour force are redundant. Throughout the decline of the
manufacturinglabour forcethe real wage in both sectorsremainsconstant
with no disguised unemployment.
With a fixed real wage, measured in agricultural goods, the rate of
population growthmust decline to the rate of technological progressin
agriculturewhenthe agriculturallabour forcereaches its maximum,so that
E = fromthis point forward. At this lowerrate ofpopulation growththe
advanced sectoris always economicallyviable. The labour forcein manu-
facturingbeginsto growat a rate exceedingthat ofpopulation growth,but
eventually declines to this rate of growth. The renewed growth of the
manufacturinglabour force is characterized by a 'big push', that is, an
extraordinarilyhighrate of growthof the manufacturinglabour force. As
in the neo-classical theory of the development of a dual economy, this
high initial rate of growth may be interpretedas a statistical artifact.
The existence of such a statisticallyobservable 'big push' is no evidence
forthe necessityof a massive infusionof capital fromoutside the system
fora 'take-off'.Sustained growthdepends on the economicviabilityofthe
advanced sector at the new rate of population growth and not on the
initial level of capital stock.
In the second phase of growthin the manufacturinglabour force the
labour forcein agricultureremains constant at its maximum level, while
agriculturaloutput grows at the same rate as population. Manufacturing
output and capital stockeventuallyincreaseat the rate A/(1-U)+ca and the
real wage, measured in manufacturinggoods, grows at the rate A/1i-r.
Throughoutthe second phase ofgrowthin the manufacturinglabour force,
the real wage, measured in agriculturalgoods, is increasingat the rate of
technologicalprogressin agriculture,a. The termsof trade between agri-
culture and industryeventually grows at the rate A/(1-a) . This rate -
may be positive or negative, depending on the relative rates of techno-

logical progressin the two sectors.
In the classical theoryof the developmentof a dual economythe phase
of development beginningwith no manufacturingproduction but with
redundant agriculturallabour or disguised unemploymentis character-
ized by a rate of growthof the manufacturinglabour forcethat exceeds
the rate of growth of population. This characterizationis a necessary

consequence of the classical theorywhetheror not the advanced sector is
economically viable at the maximum rate of net reproduction. If the
advanced sector is not economically viable at this rate of population
growth,the initialphase of disguisedunemploymentis followedby a phase
of absolute decline in the manufacturinglabour forcethat terminateswith
the agriculturallabour forceat its maximumlevel and with a reduced rate
ofpopulationgrowth.This phase is followedby a second phase ofgrowthin
the manufacturinglabour force. Again, the rate of growthof the manu-
facturinglabour forceexceeds the rate of growthof population.
We concludethat the criterionfora criticalminimumeffortproposed by
Fei and Ranis, that the rate of growthof population must be less than the
rate of growthof the industriallabour force,provides no indicationwhat-
ever concerning the economic viability of the advanced sector. The
advanced sectoris economicallyviable if and onlyifthereis a positive and
growingagriculturalsurplus,that is o > fE. During the phase of disguised
unemployment,the critical minimum effortcriterionof Fei and Ranis
is satisfiedwhether or not the advanced sector is economically viable.
Where their criterionis satisfied,the eliminationof disguised unemploy-
ment may be followed by sustained economic growthor by a period of
absolute decline in the manufacturinglabour force. Only the existenceof
a positive and growingagricultural surplus assures that growthwill be
sustained.
V. Summary and conclusion

In the precedingsections we have describedtwo alternativeapproaches
to the theory of development of a dual economy. In order to facilitate
comparison of the two approaches, we have attempted to develop both
withinthe same framework.Within this frameworkthe basic differences
between the two approaches are in assumptions made about the tech-
nologyofthe agriculturalsectorand about conditionsgoverningthe supply
of labour. In the classical approach it is assumed that thereis some level of
theagriculturallabour forcebeyondwhichfurtherincrementsin thislabour
forceare redundant. In the neo-classical approach the marginalproducti-
vityoflabour in agricultureis assumed to be always positive so that labour
is neverredundant. In the classical approach the real-wagerate,measured
in agriculturalgoods,is assumed to be fixed'institutionally'so long as there
is disguisedunemploymentin the agriculturalsector. In the neo-classical
approach the real-wage rate is assumed to be variable ratherthan fixed;
it is furtherassumed that at very low levels of income the rate of growth
of population depends on the level of income. These are the basic differ-
ences between the neo-classical and classical approaches to the theoryof
developmentof a dual economy.

D. W. JORGENSON 309
The neo-classical and classical theoriesdifferin the characterizationof
the backward or traditionalsectorofthe economy. These differences have
implications for the behaviour of the backward sector. Among the im-
plications we may note that according to the classical approach, the
agricultural labour force must decline absolutely before the end of the
phase of disguised unemployment;in the neo-classical approach the agri-
cultural labour forcemay rise, fall, or remain constant. The differences
between the two approaches also have implicationsfor the behaviour of
the advanced sector; unfortunately,these implications depend on the
actual behaviour of the terms of trade between the backward and ad-
vanced sectors. In the neo-classical approach the termsof trade may rise
or fall. In the classical approach the termsof trade cannot be determined
endogenously. Alternativeassumptions about the course of the terms of
trade may be made. Correspondingto each assumptionabout the termsof
trade, there is an alternative theory for the behaviour of the advanced
sector. Since any assumption about the course of the terms of trade is
consistentwith the classical approach, the behaviour of the termsof trade
cannot provide a test of this approach. The classical approach may be
tested only by derivingthe implicationsof this approach forthe advanced
sector,given the observed behaviour of the termsof trade, and confront-
ing these implicationswith empiricalevidence.
We have developed the classical theoryin detail only on the assumption
that the terms of trade between the backward and advanced sectors
remain constant. Proceeding on this assumption, we have derived the
followingimplications of the classical approach: (1) output and employ-
ment in the advanced sector grow at the same rate so long as there is
disguised unemploymentin the backward sector; that is, labour pro-
ductivityin the advanced sector remains constant; (2) capital grows at a
slowerrate than output and labour so that capital-outputratio falls; this
result correspondsto that of Fei and Ranis ;1 (3) the rates of growthof
manufacturingoutput,employment,and capital increase duringthe phase
of disguised unemployment. For the neo-classical approach, the corre-
sponding results are: (1) output and capital in the advanced sector grow
at the same rate, asymptotically,so that the capital-outputratio remains
constant; (2) manufacturingemploymentgrows more slowly than either
output or capital so that labour productivityin the advanced sectorrises;
(3) the rates of growthof manufacturingoutput and employmentdecrease
throughoutthe developmentprocess. Since the classical approach reduces
to the neo-classicalapproach afterthe phase of disguisedunemploymentis
completed,the two approaches have different implicationsonly for situa-
tions whereit is alleged that disguised unemploymentexists.
1 Fei and Ranis, 1963, p. 288.

In view ofthe similaritiesbetweenclassical and neo-classicalapproaches
to the development of a dual economy it is not surprisingthat many
implicationsof one model are also implicationsof the other. For example,
both models implythat ifthe proportionof manufacturingoutput to agri-
culturaloutput increases,the share of saving in total incomealso increases.
Thus, eithermodel sufficesto explain an increase in the fractionof income
saved in the course of economic development. The fact that the impli-
cations of the two approaches for the share of saving are identical is of
considerablesignificance.Accordingto Lewis: 'The centralproblemin the
theoryof economicdevelopmentis to understandthe process by which a
communitywhich was previouslysaving and investing [four or five per
cent.] of its national income or less, convertsitselfinto an economywhere
voluntarysaving is runningat about [twelveto fifteen per cent.]of national
income or more. This is the central problem because the central fact of
economicdevelopmentis rapid capital accumulation (includingknowledge
and skills with capital).'1 Both classical and neo-classical theoriesof the
developmentof a dual economy provide an explanation of an increase in
the share of saving. In each case the explanation is based on the relation-
ship between saving and industrial profits. Disguised unemploymentis
neithernecessarynor sufficient to generatea sustained rise in the share of
saving. Ultimately,a sustained increase in the saving share depends on a
positive and growing agricultural surplus and not on the presence or
absence of disguised unemployment.
The role of the industrialsectorin economic developmentis critical for
the eliminationof disguisedunemployment. In the absence of industrial-
ization an economywith redundantlabour is characterizedby population
growthat a rate equal to the rate of growthof agriculturaloutput. Agri-
cultural output grows at the rate of technologicalprogressin agriculture.
All incrementsin population become part of the redundant labour force.
In the presence of industrializationthe rate of growthof manufacturing
employmenteventuallybecomes so large as to forcethe rate of growthof
redundant labour to become negative; redundant labour eventually dis-
appears altogether. The disappearance of disguised unemploymentis,
however, no indication that the industrial sector is economically viable.
If the conditionforpersistenceof an agriculturalsurplus is not satisfied,
the initial phase of disguised unemploymentis followed by a phase of
absolute decline in the manufacturinglabour forcethat terminateswith
the agriculturallabour forceat its maximumlevel and with a reduced rate
of population growth. The conditionthat population grows more slowly
than the manufacturinglabour forceprovides no indication of the econo-
mic viability of the advanced sector at a given rate of population growth.
I Lewis, 1954, p. 155.

D. W. JORGENSON 311
In the theory of development of a dual economy, there is no critical
minimum level of initial capital stock required for sustained economic
growth. Given any positive initial capital stock,no matterhow small, the
existence of a positive and growingagricultural surplus generates sus-
tained growthof the industrialsector. In the longrunthe developmentof
a dual economyis characterizedby a growthin industrialoutput at a rate
equal to the sum of the rate of growthof population and the ratio of the
rate of technologicalprogressin industryto the share of labour in that
sector. Agriculturaloutput grows at a rate equal to the rate of growthof
population; hence the ratio of industrialoutput to agriculturaloutput is
always increasing. Similarly,the rate of growthof the industriallabour
forceis equal to the rate of growthof population in the long run. The rate
of growth of the agriculturallabour force is equal to the differencebe-
tween the rate of growthof total population and the rate of technological
progressin agriculturedivided by the share of labour in agriculture. The
existence of a positive and growingagriculturalsurplus assures that this
rate of growthis less than that of population; accordingly,the ratio of
industrial labour force to agriculturallabour force is always increasing.
Finally, capital and output in the industrialsectoreventuallygrowat the
same rate so that capital per man increases at a rate equal to the rate of
growth of output per man, namely, the rate of technologicalprogressin
industrydivided by labour's share in industry. The share of capital for-
mation in national product is always increasing,ultimatelyapproaching
the share of propertyin the product of the industrialsector.
We conclude that the industrial sector plays a strategic role in the
developmentof a dual economywith or withoutdisguisedunemployment.
Industrial output and industriallabour forceultimatelycome to dominate
a developed economyas a consequence of the shiftin a consumerdemand
fromagriculturalto industrialproducts and as a result of the risingpro-
portion of investment demand in total output as income per capita
increases. However, supply conditionsforthe agriculturalsectormust not
be neglected in any analysis of prospects for industrialization. Unless
technologicalprogress in agricultureis sufficiently rapid to outpace the
growthof population and the forceof diminishingreturns,the industrial
sector may not be economicallyviable.
In the absence of a growingagriculturalsurplus,forcedindustrialization
at fixedreal wages mayresultin a phase of growthin therelativeimportance
ofindustrywithno improvementin levels ofliving,followedby an absolute
reductionin the size ofthe industrialsectoras population growthis forced
down to the Malthusian level consistentwith the increase of sustenance.
Since the criterionforthe persistenceof an agriculturalsurplus depends
on the rate of growthof population, the rate of advance of agricultural

technologyrequired forimprovementin levels of livingis largerthe larger
the rate of population growth. The recent increase of rates of population
growthin low-incomecountries has increased the thresholdfor rates of
improvementin agricultural technology required to sustain industrial
development. Where the conditionforviability of the industrialsector is
not met, any policy forindustrializationmust be accompanied by policies
forpopulation controland forthe introductionof non-traditionalfactors
into the agriculturalsector.'
Berkeley
I T. W. Schultz, TransformingTraditional Agriculture.New Haven: Yale University

Press, 1964. The importanceof policies of this type for low-incomecountrieshas been em-
phasized by Schultz in this book.


Jorgenson (1967) - Surplus Agricultural Labour and The Development of A Dual Economy

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Jorgenson (1967) - Surplus Agricultural Labour and The Development of A Dual Economy

Uploaded by

Copyright:

Available Formats

Surplus Agricultural Labour and the Development of a Dual Economy

Author(s): Dale W. Jorgenson

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

1 This point of view is elaborated in my paper, 'The development of a dual economy',

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

II. Development of a dual economy: a classical approach

theLabor Surplus Economy.Homewood: Irwin, 1964, p. 16.

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

Of course, if the agriculturallabour force is equal to total population,

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

where eAtrepresentstechnological change, as before,and the constant a

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

If we assume that saving is equal to the share ofprofitsin the industrial

The rate of growthof manufacturingemploymentis, of course, equal to

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

For this conditionto be satisfied,the manufacturingproportion,WM, must

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

III. Development of a dual economy: a neo-classical approach3

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

where variables and parameters have the same interpretationas in the

where variables and parameters have the same interpretationas in the

This assumptionis identical to that made in the classical approach.

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

If agriculturaloutput per head exceeds the criticallevel, part ofthe labour

To study the developmentof a dual economyforthe case in which the

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

which implies: E-C- < E(I-A),

The rate of growthof population is greaterthan that of the agricultural

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

so that the rate of growthof manufacturingoutput is equal to the rate of

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

IV. Beyond disguised unemployment

by insertingtotal population into the equation:

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

may be positive or negative, depending on the relative rates of techno-

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

V. Summary and conclusion

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

I T. W. Schultz, TransformingTraditional Agriculture.New Haven: Yale University

This content downloaded from 195.78.108.81 on Tue, 17 Jun 2014 12:42:33 PM

You might also like