Disequilibrium and increasing returns in Adam

Smith's analysis of growth and accumulation

Gaviri C. Reid

I . Introductiori
There exists an enormous secondary literature on all aspects of Adam
Smith's work.' arid in contemplating adding to i t , one must have consid-
erable confidence that there remains something worth saying. We wish to
argue in this article that with respect to Smith's analysis of growth and
accumulation. there are two good reasons for adding to this literature.
Firstly, there is a tendency for writers on Adam Smith's analysis of
growth to emphasise either its 'cheerful' (growing) or 'dull' (declining or
stationary) aspects. By contrast, Smith himself actually analyses-though
perhaps not all in one place-what one would today call a complete
'growth trajectory,' in the course of which growth. decline. and ultimate
stationarity all occur. We will argue that it is both unnecessarily analyti-
cally restrictive cirid textually inaccurate to emphasise any one aspect of
this growth process to the neglect of others.
Secondly, an essential aspect of growth as analysed by Smith. is that it
progresses at a rate which is neither uniform over time nor uniform across
sectors of the economy. Smith's is a disequilibrium lorm of growth. driven
by the increasing returns which are consequential on the division of labour.
with advance in one sector being a prerequisite to advance in another.
A certain amount of formal notation may help make the approach
adopted become clearer to the reader. Consider an ti-sector economy at
time period with sectoral growth rates g , ( t ) ,g 2 ( f ) .. . . g , ( f ) . In our view.
an appropriate representation of Smithian long-run equilibrium is the sta-
tionary state in which there is a zero growth rate in each sector. Formally,
stationarity is attained in time period 7 if g , ( r ) = g 2 ( t ) = . . . g,(t)
= 0 for all r 2 T. However, before this stationary state is reached we have,
typically, g, f g, # . . . # g, implying unequal growth rates across sectors.
Finally, growth is not merely unbalanced across sectors, but progresses at
a variable rate over time for each sector. this being represented by the
functional dependence of each growth rate on time ,qi = g,(r) ( i = I .

Correspondence may be addressed to the author. Dept. of Economics. University of Edin-

burgh. William Robertson Building. George Square, Edinburgh EH8 9JY SCOTLAND.
I . For an up-lo-dm: survey ot' a vast amount of this secondary literature. see Gee 1085.
xx
2, . . . 1 1 ) . This notation has been made precise. even pedantic. in order
to sharply ditferentiate the view taken here troin that ;idopted in a nuinlJer
of' distinguished analyses of Sniithian growth. In Eltis 1984, the sectoral
aspect 01' Smithian growth is suppressed. a single aggregate production
function being used. and for given parameters the growth rate is invariant
over time. 'Thus ;i single constant ,q is implied. In Samuelson 1977 a tiiore
complex picture is presented of ail ri-sector economy. However. growth
rates are the same across dl sectors: ,g, = ,q for all i . Furthermore this
growth rate is constant over time until land scarcity plummets the economv
into the stationary state with all sectoral growth rates falling to zero:
gi = 0 for all i. Oiily in Samuelson 1Y78 is a more satisfactory picture
presented in which the economy follows a trajectory of variable growth.
both over time aiid between sectors, in its approach to the stationary state.
In the sort ot' disequilibrium situations that are being contemplated in
this article, markets may not be cleared. actual and desired levels of in-
ventories may not coincide, and short-period expectations may not be ful-
lilled. Models of this sort have been considered in Jorgensvn 1961, and
riiost notably in Hahn 1963. They have in common, however. the retention
ot the assumption of constant returns to scale. By contrast. we shall argue
that increasing returns arid disequilibrium are characteristic o f Sinithian
growth. Recently. Blitch I983 has drawn attention to Kaldor's recognition
of the extraordinary insights of Young ( 1928) in his extensions ot' Adam
Smith. Kaldor ( 1972) himself made a further contribution to our under-
standing of the Stnithian theory of econornic progress. and in Kaldor I98S
we now have an elegant and cxtended restatement ot the earlier argunient.
l'liere is no doubt that in his enthusiasm to ;rrnplify the merits of Sniith
and Young. Kaldor overstated his own case and made some ill-judged
attacks on certain aspects of neoclassical general equilibrium theory. Hahn
(1974) in his inaugural lecture at the University of Canibridge was to
rnerci lessly snipe at Kaldor tor his errors. Certainly there 1t'er-eerrors on
Kaldor's part. of which the most signilicant was his suggestion that in-
creasing returns are ritwssitril~destructive ot' an Arrow-Debreu type of'
equilibrium. In fact. certain limited occurrences of increasing returns are
permissible: which is perhaps to say 110 more than that i t the extent o f
increasing returns is limited. then the consequences .are not substantial.
However, it is clear that what Kaldor had in mind-and this is very much
in keeping with Smith-was a situation in which increasing returns were
pervasive and o f considerable consequence. In Section I11 below a device
will be introduced that enables one to represent the consequences o f the
division of labour by an appropriately specified 'iiiiplicit' production tunc-
tion which displays increasing returns to scale. I t increasing returns ;ire
strong, the conventional stable equilibriuni ot' general equilibrium analysis
can be transformed into an unstable equilibrium.
 Rvid - Disequilibrium in Adam Smith 89

The nature ot’ Smith’s growth analysis as we would present it is now

evident: its core is a sectoral model. with many. i t not all. sectors subject
to increasing returns, and growth taking place at a rate which is variable
through time and across sectors, in response to sectoral disequilibria. The
final outcome of all this is stationarity.
These two sentences alone define an analytical framework of immense
complexity. Smith, of course. expounded it in purely literary terms, free
even of numerical examples. It is apparent that Smith did consciously think
in terms of systems, these being “inventions of the imagination” ( E P S .
105),* and that he applied his theory of scientific systems to the construc-
tion of an economic ~ y s t e m He. ~ drew an analogy between systems and
machines, saying “A system is an imaginary machine invented to connect
together in the t‘ancy those different movements and effects which are al-
ready in reality performed” (EPS, 66). Though he was not averse to the
mathematical analysis of systems, he recognised the limitations of this
method: “It rarely happens, that nature can be mathematically exact with
regard to the figure of the objects she produces. upon account of the infinite
combinations o f impulses, which must conspire to the production of each
of her effects” ( E P S . 95). Having observed how the astronomers had la-
boured in vain to discover constancy and regularity in the motion of the
heavenly bodies. he is compelled to remark that they were perhaps ex-
pecting too much, for it is “to be found in no other parts of nature” ( E P S ,
96). I t is appropriate, therefore, to look at Smith’s analysis of growth in
terms of an imaginary machine, or as we would say today, an economic
model. Such a model might be mathematical. though in this form it is
unlikely to capture many aspects of reality..
Nevertheless, theorists undaunted as always by the need for heroic
abstractions. have attempted to model Adam Smith’s system, some exam-
ples of these etrorts being Thweatt 1957, Higgins 1959, Adelman 1961,
Hicks 1965. Barkai 1969. Eagly 1974, O’Brien 1975, Eltis 1975, 1984.
Samuelson 1977. 1978, Gee 198 I , 1984, Rosier 1984, and Reid 1985.
Undoubtedly the most rigorous and elegant of these models has been that
of Samuelson 1977. However, as we shall see below, even Samuel-
son’s work might be regarded as a deficient treatment of Smith’s growth
analysis.
The article develops as follows. First, the full growth trajectory is con-
sidered, both in terms of Smith’s original exposition and in terms of at-
tempts by later writers, notably Samuelson, to model this trajectory. It is
concluded that most modelling efforts have neglected important. indeed
2. Throughout this article, where works of Adam Smith are citcd. reference is made to
the Glasgow Edition. using the following system of abbreviations: W N . The icxwlrh na-
tions: EPS. Essays on philosophicd subjects; W , Lecturesoit jiuispriidettce.
3 . For ;1 rcccrit
1:’.fc*!nentof this view, see Raphael 1985.
crucial. aspects of. Smith's analysis. Second. a method of' developing
Siiiitli's growth arialysis is exariiinecl. I t is t'aithful to the original arid
irivolvcs sonte signiiicmt extensions. Its origins lie in Mmng I928 and
Kaldor 1 Y72, arid the riiethod is essentially that of'sectoral clisequilibriiim
growth. Such ;I treatment is not intensely matheniatical. although i t is
certainly analytically rigorous. but it ott'ers. we suggest, the most illunii-
iiating iiieaiis ot' expounding and extending Sniith's growth andysis.
II. 7'1w G r o ~ * tliujcctoty
h
Before turning to contemporary accounts ot' rldani Smith's growth pro-
cess, i t is helpful to take the original development in the Wmlrii c!f'ricrtioris
as the point ot' departure. In his exposition. Smith started from the prin-
ciple of' the division of' labour and moved on to an explanation of' capital
accumulation. fdowever. in the words o f Viner 1928, 116. Sniith's model
is "a coordinated and mutually dependent system of cause and effect rela-
tionships." This being so. we can enter the model at any point without
doing violence to the logic. I t is in fact highly instructive to take the role
of parsimony in permitting accumulation 3s our starting point, moving on
to the result that the division of labour is limited by the extent of the
market. Growth takes place i n an unplanned way and depends on the desire
of individuals for betterment:
An augmentation of fortune is the means by which the greater part of'
men propose and wish to better their cwndition. I t is the means most
vulgar and riiost obvious; atid the most likely w ; i v ot au_enienting their
t'ortune is to save and ;iccuiiiiil;ite sonie part of what tliey q u i r e
(CVN, 341-421.
'I'his willingness to save leads directly to an augriientation of capital. In
Smith's view there was an identity between saving and investment at both
the individual and aggregate levels:
Whatever a person saved from his revenue he adds to his capital. and
either employs it himself in maintaining an additional number of pro-
ductive hands. or enables some other person to do so. by lending it
to him for an interest. that is for a share of the profits. As the capital
of an individual can be increased only by what he saves from his
annual revenue or his annual gains. so the capital of a society. which
is the same with that of all the individuals who compose i t . can be
increased only in the same manner IW N . 337).
'I'he increasc ot. capital arising t'roni parsimony
naturally tends to increase . . . the real quantity ot' industry. the iiuin-
her of productive hands. and consequently the exchangeable value ot'
 Reid - Disequilibriuni in Adam Smith YI

the annual produce of the land and labour of the country. the real
wealth and revenue of' its inhabitants [ W N , 337).
Finally, this increase of annual revenue enlarges markets and provides the
basis for the division of labour and hence an increase in productivity. When
markets are small, there is no inducement to specialization, but as they
enlarge "the power of exchanging . . . gives occasion to the division ot'
labour" and "the extent of this division must always be limited by the
extent of that power, or, in other words. by the extent of the market" ( W N .
31). Thus "the division ot' labour . . . so far as it can be introduced. oc-
casions, in every art. a proportionable increase of the productive powers
of labour" (I!". 15). This process of the progressive expansion of markets.
with increasing scope for the division of labour. is not impeded by mone-
tary restriction, for in Smith's view. the money supply adjusts passively to
the value of the annual produce:
The quantity of money . . . must in every country naturally increase
as the annual produce increases. The value of the consumable goods
annually circulated within the society being greater, will require
greater quantity of money to circulate them [ W N ,340).
This increase of produce will not go on forever, and ultimately the econ-
omy approaches a stationary state which is "consistent with the nature of
its laws and institutions" (WN,1 I 1 ) . The characteristics of the stationary
state were described by Smith as follows:
In a country which had acquired that full complement of riches which
the nature of its soil and climate, its situation with respect to other
countries, allowed it to acquire: which could therefore, advance no
further. and which was not going backwards, both the wages of la-
bour and the profits of stock would probably be very low. In a country
fully peopled in proportion to what either its territory could maintain
or its stock employ, the competition for employment would necessar-
ily be so great as to reduce the wages of labour to what was barely
suthcient to keep up the number of labourers. and. the country being
already fully peopled, that number could never be augmented. In a
country fully stocked in proportion to all the business it had to trans-
act, as great a quantity of stock would be employed in every particu-
lar branch as the nature and extent of trade would admit. The
competition, therefore, would everywhere be as great, and conse-
quently the ordinary profit as low as possible IW N ,1 I I ] .
It is exactly conditions of this sort that are described by Samuelson ( 1978)
in his characterisation of the limiting, long-run equilibrium position in a
classical economy. Unfortunately, as in so much contemporary analysis,
crnphasis is on this stationary state. l'or tnodern economists ;ire most at-
tached to the properties ol'equilibrium positions. 111 Smith this equilibriurii
is discussed, as in the above passage, but only in the context ot' a growth
tra-jectory: the path to equilibriurii is ;IS important as the attainment o f
cquilibrium, Saniuelson's system can be represented by these relation-
s ti i ps .
( K ( f+ I ) - = k(r(r1 - ,-':I
K(f)l/K(f)
I/v(f+I ) - = r r ( w ( t ) -- w"I
N(f)J/N(f)
with
LimlN(i), K ( r ) , r - ( r ) , w ( i ) . .
1 . 0 )

= IN", K*, f k ,t,:k . . .I

k ( O ) = ri(0) = 0 and k ' . ri' > 0. I t is completed by relationships deter-

mining the rate of prolit ( r )on capita (Mand the wage rate ( w) for labour
(N). ;Iproduction function and an adc ng-up relationship. Starred variables
are long-run equilibrium values. and the system can be shown to be glo-
bally stable. In the absence of any form of induced technical progress,
uset'ul illustrations of possible dynamics of this system can be developed-
an issue which Samuelson rather neglected in his enthusiasm for providing
a rigorous analysis of stability. Consider Figure I . in which w* is the
subsistence wage rate and r* is the minimal profit rate. Point A denotes
some initial level of the wage and profit rate ( w ' . r ' ) and point B the
stationary state (,r**. f k ) with \ti' > \\*.': ; r i d r' >: P. Paths trh and ch illus-
trate two different possible routes to the stationary state. Along path t r h .
the wage rate and the rate of profit fall steadily towards stationary state
values, with population adjustment being assumed to take place relatively
slowly. Thus the rate of profit reaches its tiiininial level before the wage
rate has been depressed to a subsistence level. Along path ~ 6 both . the
wage rate and the rate of profit initially rise. with r again adjusting more
rapidly than w. starting to tall earlier, and attaining its stationary state value
sooner. The rationale of path lib is obvious. but that of path c6 invites
further discussion. Now i n Smith technical change is endogenous. it being
caused by the division of labour. By contrast, in Samuelson 1977, 1978.
the process of technical change is exogenous io the system. In typical o.v
cnrhedra style, Samuelson ( 1977. 43) writes:
Now let there be an invention. . . . [ i t ] must transiently raise one. or
more probably both. of the profit rate and the real wage rate . . . but
ultimately . . . the law of' diminishing returns on lixed land operates.
The system relapses into Smith's 'dill1 state.' . . . If inventions keep
recurring, the system goes through a Brownian motion in which profit
rates and real wages average out above their subsistence levels. . . .
 Reid - Disequilibrium in Adam Smith 93

0 r*
Figure I

The rnodel captures the general behaviour ot' economic history these
last two centuries.
In a similar vein Sarnuelson (1978, 1428) talks ot' the way in which "in-
novation plucks the DD' string outwards" (i.e.. the marginal product
curve) 3s a consequence of which there is "a Brownian dance or Schum-
peterian tluctuation of real wages and profits at average levels above the
minima." The insertion of the word Schumpeterian here is telling. and
makes the application of Samuelson's analysis to Adam Smith very
strained. Whilst Samuelson talks elsewhere of inventions being both spon-
taneous and induced, it is clear that the latter category is ignored in a
formal sense. In his world, inventions are conceived of as impincging on
the system from the outside, having been generated by some stochastic
process. A modified version of his production function could be written:
Y = F7K. IV. E )

where E is a random variable. Thus marginal products are themselves I'unc-

tions of this random variable. which implies. assuming perfect competi-
tion, that the w:il wage rate and the rate of' protit art- random variables:
 \I. = ,I:,(E iincl t- = ,/*.(F.

I t is lor this reasori that S m u e l s o i i talks of the economic systeni going

tlirougti t3rowniari motion. By contrast. in Smith the eniptiasis is almost
critirely reversed. with induced innovation being dominant. i t being
achieved by the division o f labour. To Smith. the principal determinant ot'
itinovation was undoubtedly the xcuniulation ot' capital. for "labour cat]
be more and more subdivided in proportion only as stock is iiiore x c u -
inulated" ( W N . 277). This is ignored in a timnal sense by Sariiuelson. and
indeed the specilication of' the production t'unction above is. as eveii he
seenis prepared to hint. iiiore Schumpeterian than Sinithian. The two ap-
proaches can be contrasted by reference to Figure 2. For the purposes of'
geometrical illustration, inputs are regarded ;is being combined in fixed
proportions. s o that it is therefore sensible to talk of a 'dose' of' I *'I bour-
cum-capital, and its corresponding rnarginal product. At time period t the
insrginal physical product curve is MPP,. and for the input level o f o n we
can examine the hrcakdown into wages. rent. and protit. The subsistence
wage ratc is d , the wage rate at time period is \t*, and theretore total
wage payments are w,cciO. Profits are given by area \tj,c.de. and rent hy the
area contained between ed and the MPP, curve. Assuming that the MPP,
curve has not changed for ;I number of previous periods. the path of the
point that determines the split between wages and protits is the negatively
sloped arrowcd line going through (.. Irivention of the sort Sarnuelson

Marginal
I' r od uc I

wt ' 1

wt
W'

Dose of
labour -cum-
0 capital
 Reid . Disequilibriurii in Adam Smith 95

envisaged shifts the rnarginal physical product curve to the right. Such a
new curve is labelled as MPP,+, in Figure 2. This can cause an increase
in the wage rate along the path clz to a level w , + ~whereupon if the
MPP, curve remains stable for a number of periods-or expressed more.
~

formally. no new E term appears in the production function-the wage rate

will fall along the path h k . eventually reaching the subsistence level \ t i * at
input level Om if no new discoveries occur. An alternative representation
of this trajectory is given by the path cb in Figure I . By contrast. in the
Smithian approach, this random perturbation of the wage rate and rate of'
profit would not occur. The marginal product curve would shift contirrir-
ously to the right as the economy expands, and scope for the division of
labour increases, though growth would not necessarily proceed at a uni-
form rate. Furthermore. this growth would not be periodic or intermittent.
If the growth rate were sufliciently rapid, and if the productivity gains from
the division of labour were suthciently great, then the real wage rate might
move along an ascending trajectory for a while, as in Figure 2.
Having considered a variety of paths to the stationary state. let us now
examine the rate at which it occurs, and how adjustment takes place. Sam-
uelson (1978, 1420) in analysing what he calls "the canonical classical
model of political economy" refers to "long-run equilibrium and . . . tran-
sient mowrnenr toward equilibrium" (this writer's emphases). However.
what might sensibly be regarded as transient in a physical system (and of
course the theory of differential equations applied to such systems has been
carried over to economics) is not so sensibly referred to as such for a social
system. which might spend several centuries in this passage to long-ru.n
equilibrium. From the examples which Smith used to illustrate the final
phase of the growth process, in which the stationary state is approached.
it is clear that he envisaged the period of adjustment to this state. in the
absence of institutional evolution, to be very lengthy. As a result of Atkin-
son 1969 we know that the time scale on which growth models converge
can be very long, for reasonable choices of parameters. Economic theorists
are therefore faced with a considerable difficulty. Much of theory is con-
cerned with equilibrium analysis. This is neither mere expediency nor
blinkered thinking. It arises almost of necessity, for the concept of equilib-
rium is more primitive, or essential, than the concept of adjustment to
equilibrium. As Hahn4 has put it:
there is no "primitive" account of the lag structure of all the error-
correcting mechanisms we postulate; that structure is completely ad
hoc. But the behaviour of the system is very sensitive to that struc-
ture. . . . Equilibrium theory, even in its sequential and expectational
form is as robust as its sparse axioms; process theory seems totally

4 . Quoted b\. .hild (from a personal communicntionj Kirzner 1982.

 uiirohust and therefore strictly coritingciit. That is. while tijr equilib-
riuiri analysis t'unctioiial form is aliiiost iriesseiit ial. i t is almost every-
t t i i ng t'( )r process ;I ii;i I y i s .

l7,sseiitial to the viewpoint adopted i i i this article is the notion that Sniith's
;inalysis of growth is a pr*oc.rs.rtheory. Furtherniore. whilst this process
might in the long r u n culniinate i n a stationary state. the system must ti)r
;I long tirne be exhibiting the characteristics o f the process. The error-
c.orrection mechrrnisrns postulated by Saniuelson are 1'amiIiar from the l i t -
crature on passage to equilibrium in a Mlrasian system. However. there
i z II( ) dou h t i n g t lie i r arbi t rar i liess . t: u r t ti e r 111c) re . these at1.i us t iiie11t niec h ;I -
iiisliis do not begin to capture the sort u t process of cuniulative growth
wliicti is rcgarded as central to Sriiithian ;I' .!!,sis in this article.
So far. attentiori has been coiiceiitrated on aggregate niodels. which
indeed is the tendency in much 01' the literature. where contributions in-
clude the work ot' Adelinan 1961. I-liggins 1959. Barkai 1969. and Eltis
1975. 1084. However. there is no doubt that Srnith himself thought in
sectoral terms. A new revivalist school. the Modern Classical. of which
Walsh & Gram 1980 provides the most comprehensive and accessible ex-
position. has once again eniphasised the sectoral approach of the classical
writers. this time bringing to bear on traditional areas of inquiry the full
arsenal ot' contemporary analytical niethods. Adopting a notation close to
that o f Wdsh & Grani 1980. a two-sector yersion of this sectoral inoclel
with unbalanced growth ~iiaybe writtcri

where ( I , , (the typical "tcchnical coetlicicrit") denotes the arncwnt o f the

i.th commodity required to produce one unit of' the j - t h commodity. Other
symbols are ?', the output ot' the i-th sector: 8, the rate o f growth in the i-th
sector; A, the consuniption proportion in the i-th sector. In defining con-
sumption proportions. we follow the tisual practice of not treating mini-
mum subsistence as consumption. but ;IS ;I cost of production. The A ,
adhere to what Dorlinan et al. 1958. 325. calls "extraneous linal con-
sumption" and Ckilsh & Gram IY80. 72. calls "luxury consumption." In
the Von Neuiiiann inodel as usually developed by rnodern economic theo-
rists." the Ai are set to zero. arid growth rates are equal across sectors.
g1 = ,q = ,o. giving
 Reid - Disequilibrium in Adam Smith 97

!'I + (I,? )*J ( I + g) = !'I

( ( I 2 , )., + ( 8)
(1:: -I._.) I + = J'2

which is the case of maximum balanced growth. If positive extraneous

final consumption is permitted, a lower rate of balanced growth will be
attained. The above model is relevant to our discussion because i t is an
analytical framework of this sort that Samuelson 1977 adopted in its re-
markable "vindication" of Adam Smith. In this analysis the economy
grows at a balanced exponential rate, until it runs into the constraint ot'
land. and then it relapses into Smith's 'dull' state with a minimal real wage
rate and protit rate, and maximum rent. One of the difficulties with this
approach. to which Walsh & Gram 1980, 72, draws attention, is that the
assumption that there is long-run equilibrium with unbalanced growth (i.e.
that the g, of the set of quantity relations above are distinct) is inconsistent
with the assumption that the technology exhibits homogeneity of degree
one, this being a characteristic of the Leontief-Sraffa technology used
above. Either unbalanced growth has to be abandoned or else one must
give up constant returns. Both Samuelson 1977 and Walsh & Gram choose
to abandon unbalanced growth. As we shall now argue, that is to remove
an essential component driving Smith's machine of growth.
I I I. The Young-Kaldor Extensions of Smith
In September 1928 Allyn Young read his presidential address to Section
F of the British Association in Glasgow. His topic. 'increasing returns and
economic progress' was fundamental and very deep: perhaps too deep to
be appreciated in his day. Subsequently published in the Economic Jour-
nal, the paper has at its core an analytical construction which even his
leading disciple. Nicholas Kaldor (1972. 1243) finds "at times . . . ob-
scure." The basis of increasing returns in Young's analysis is the division
of labour.
However, Young endowed the concept of the division of labour with a
broader meaning than one finds in Smith. Specifically, Young ( 1928. 537-
39) extended it to industrial differentiation, and to the realising of more
roundabout niethods of production. This line of analysis has been elabo-
rated and amplified in Kaldor 1970, 1972. 1985, a variety of publications
over the last fifteen years. Most notable of these was the Scottish Eco-
nomic Society Lecture delivered at the University of Aberdeen, where
Kaldor ( 1970. 340) emphasised the significance for growth of
the existence of increasing returns-using that term in its broadest
sense-in the processing of activities. These are not just the econo-
mies of large scale production, commonly considered. but the cu-
mulative advantage accruing from the growth of industry itself-the
 development ot' skill and kiiow-tiow; the opportunities for easv coin-
rnunicatioii o t ideas arid experiences: ttie opportunity tor ever-
increasing dittereiitiatiori ot' processes and of' specialization in tiurnan
;ict i v i t ies .
In tiis earliest paper K a I h r W;IS particulx-I!* coiicerned with the way iii
which increasing returns iiiight n o t lead to :ill regions benelitting t'ruiii
inter-regional trade. Subsequently. Kiildor 1972. I 985. have einptiasised
the conditions necessary tor expansionm-y inter-sectoral. or inter-industry
trade. and this theme has been taken up in Reid 1975. ch. 7; 1979.
In Smith. the analysis is incomplete. but certainly the extensions pro-
vided by Young and Kaldor are in harniony with Smith's views, and con-
stitute a logical development of them. Young's analysis runs in sectorat
terms. with the interplay of reciprocal supply and demand leading. he
cIairIis, to growth of a cumulative sort. Kaldor I Y72 tias demonstrated that
Young's analysis too is incomplete, lacking a theory of effective demand.
Such a theory is supplied by Kaldor. q a i i i working within ;I Sniithiaii
I'raniework.
Exploring further these extension< of' Smith. we will return to the e x -
tended arid precise analytical Note to Young's article. Rather than rejecting
i t 3s "obscure." we will regard it as being ot' central iiiiportrliice. A cartl'itl
study of this Note reveals that the whole analytical construction o f ttie
article proper is woven around ;Iset ot' devices due to Pareto lc)27.7 Froin
our standpoint. the Note is full ot' promise. t'or i t starts with ;Iconstruction
that depends on a sectoral view ot the econotiiy. namely. the productioti
possibility frontier.
Suppose th;it goods X and Y are produced t'roni the same set of t x t o r
inputs capital (K)and labour ( L ) according to distinct production t'unc-
tions. Under the assumption of :in increLising opportunity cost ot producing
X in t e r m ot Y as relatively more ,Y is produced. the production possibility
lrontier will have the lorn1 TT' in Figure 3.
In Pareto 3 line such ;IS IT'is always reprded as ;I possible path tliroiigti
the commodity space. Whether a path is. or is not. possible depends. in
tiis terminology, on "obstacles" to transforming one commodity into an-
other. Those "obstacles" are detined by the form of the production func-
tions. Superimposed on the same diagram is a set of collective indifference
curves. U , , U,, . . .
Of course, the indifference curve construction is not found in Smith.
and Young, in employing i t , is clearly engaging in rather a radical exten-
sion of Smith. However, as Hollander f 1973. ch. 4 ) has argued in perwi-
sive detail Smith certainly extensively eriiployed a theory of choice. and
 Rcid - Disequilibrium in Adam Smith 99

0 T'
Figitre 3

his lack of a narrowly defined concept of marginal utility should not be

construed ;IS his rejecting the notion (so central to much of economic
analysis) that utility is a necessary condition tor exchange value. More
recently, Gee & Jarvis 1984, in its formal specification of a Smithian
model. has represented the preference structures of labourers and landlords
by explicit utility functions. There does, therefore. seem to be some merit
in Young's suggested extension. In terms of Pareto's terminology, maxi-
mum ophelirnie is attained at the point a in Figure 3. This is a familiar
way of representing general equilibrium in a two-good, two-factor econ-
omy and is a standard device in trade theory, welfare economics, and so
on. With Young and Kaldor, both of whom are concerned with logical
extensions of the Smithian scheme, this device is used as a point of depar-
ture, for their concern is with disequilibrium.
Youn 1928 directs attention not at the stable equilibrium point a in
4
Figure . but at ways in which the production possibility frontier can be
modified in order to move away from equilibrium. Young talks (p. 533) of
"the counterforces which are continually defeating the forces which make
I00 H i s t o n o/' P o l i t i d Ec*orioniv 19: 1 ( 1987)

lor economic equilibrium" and eiiiphasises ( p . 528) that in the growth

process which he is analysirig "movements away from equilibrium. depar-
tures from previous trends. are characteristic of it." This movement away
from equilibrium is made possible by the division of labour, and this mod-
ifies production functions so that strongly increasing returns become man-
ifest. In Figure 3 this is represented by a modification of the production
possibility frontier TT' to the form TT"which exhibits at least decreasing
opportunity cost in the neighbourhood of' the initial equilibrium point (1.
The circumstances under which this ett'ect can arise are worthy of further
consideration. Certainly they are not treated at all by Kaldor. and perhaps
w e n in Young they are imperfectly understood.
Let the production tunctions for each sector be written
Y = F ( K , L; M )
x = G ( K , L ; Iv)
where M and N are parameters representing the consequences of the di-
vision of labour, Y and X are commodity outputs, and K and L are cap-
ital and labour inputs. Marginal products of capital and labour are as-
sumed positive and diminishing:8 F,, F,, G,, G,- > 0 and F K K FLl,,
, GKK,
G , < 0. Furthermore, the marginal product schedules are assumed to shift
positively with positive variations in the division of labour parameters.
Thus we have F,,, F,.,w, C, G,, > 0. This corresponds to the situation
graphed in Figure 2. where inventions shift the MPP curve positively.
flere. however, this process is continuous rather than intermittent. The
simplest way of expressing this is to make the division of labour terms
themselves dependent on the level of output (i.e.. on the extent of the
market). Thus we can rewrite the above ;is
Y = F ( K , L ; M ( Y ) ) with M ' > 0
X = G ( K , L ; N ( X ) ) with N ' > 0

Now, assuming invertibility of the above two functions we get:

Y = f ( KL)
x = g(K, L)
In the recent international trade literature on increasing returns these have
been described as intplicit production functions. In the presence of the
8 . The specification of such marginal productivity relationships has o f late been sur-
rounded by controversy. Thc device is used in Samuelson 1978 and its lcpitimacy was
examined in great dctail in Hollander 1980. The discussion by Smith in his first hook ot'
the Wecdfhrfnarions is taken t o be one "into which the implications of diminishing returns
are fitted as a special casc" by Hollandcr 1980. S60. Hollandcr thcn proceeds to give four
cxtensivc passages in favour of Smith's recognition 01' diniitiisliing returns. one of which is
the familiar colony quotation rclcvarit t o thc specific case of diminishing agricultural rc-
turns.
 Reid Disequilibrium in Adam Smith 101

division-of-labour etfects indicated above. questions of the existence, or

otherwise. ot' increasing returns using the standard homogeneity technique
should be addressed to these functionsfand 8 , rather than to the functions
F and G .
To illustrate the method adopted, consider the production function for Y
given by F. and write F in the separable form
Y M(U,@ ( K . L ) with M' > 0
=

Now if @ is linearly homogeneous in K and L, and M(Y) has an elasticity

of less than unity, then the corresponding implicit production function, say
Y = cb ( K . L )
exhibits increasing returns to scale, a result given by Helpman & Krugman
1985. As an example, consider the production function with parameters
A,a:

where M(u) = YO s. Thus the relevant elasticity is (dM/dY')/(Y/M)= 0.5,

which, as required. is less than unity. Note that were M a fixed constant.
the above production function would appear to exhibit constant returns to
scale, as the sum of the exponents on the capital and labour variables is
unity. However, inverting to get the implicit production function gives

Now the sum of the exponents on K and L is 2. implying increasing returns

for the implicit production function, even though the original parameter-
ized production function apparently exhibited constant returns before the
division-of-labour effect was investigated.
Now it is known that increasing returns are a necessary but not a suffi-
cient condition for convexity in the production possibility curve, a matter
which has been given detailed treatment by Bator 1957 inter a h . What is
being assumed in Figure 3 is that the increasing returns, due to implicit
production functions at the sectoral level of the above form, are suficiently
strong to induce the decreasing opportunity cost segment on TT" between
Q and 6. There clearly now is a potential advantage to be sought if increas-
ing returns of this nature can be obtained, for the path from u to b is what
Young 1928, 541, calls a "preferred route." using a metaphor much em-
ployed by Pareto, What was a stable equilibrium at u has been transformed
into an unstable equilibrium. However, the argument is not quite as simple
as this. Young argues that the achievement of this extent of increasing
returns is not costless, and that the costs of moving along the "preferred
route" a6 depend not only on the distance moved in commodity space, but
also on the rate at which this movement takes place. Clearly there is no
I02

siiiiple way ot' accounting for these costs ot' movernent a\Vily from equilib-
rium. I t is suggested bv Youiig tliiit tlicv can be iiccoiiiiiiodatcd by suppos-
ing that the iiidift'ercncc inap t w r w i c s "cviitracted" by costs. Tlius. given
m y point in the coiiirnodity spacc. ;I lower level of utility is attained ;it
t h poiiit c.ificJr-a inotlification ot'the prodtiction possibility t'rotitier because
ot. the cost incurred in achieving i t . Oiic is. tlieretore. riiakirig the whole
prefcrcnce field pcith t l e p c t i d c ~ r i r .;I clillicult concept. ivhicli perhaps ex-
plains why Kaldor regarded i t ;IS obsciirc. I-lowever. to sav this is not to
say ttiiit thc notion is fllultv. but ratlicr that it is n o t readily translated i i i t o
sirnple geoinetry. One ci\n capture the tlavour u t the argurnent by consid-
ering the stable equilibriuin point b in Figure 3 at which ;I utility level of'
Us is attained. Now this utility level is relevant to a preference field defined
1.01-the point h. Retore the route f'rom ci to 11 was undertaken. the utility
level corresponding to Us was attained o n the indift'erence curve U,. j i - o n i
the perspe(*riivc $ t h irtitiol e c p i l i h r i r t n i p o i r i t a. Likewise. the utility level
associated with h having been achieved. and the cost of moving from 11 to
1) having been incurrcd, the utility level initially attained o n inditference
curve U , will now be attained on a contracted version otthis inditference
curve to the south west of U , . Otcourse it is not necessary to this argument
that the new stable equilibriuni point h should ever be achieved: indeed
this is to weaken Young's conclusions. tor i t would signal ;IH end to eco-
nomic progrcss. What he suggests is that strong increasing returns ot' tlie
above nature provide a continuous incentive tor rnoving along a preterred
route through the coinmodity space. and away t'rorn any potential static
stable equilibrium. In the longer term. this tendency would be reinforced
by the discovery ot new resources or the growth ot' population which
would tend to move the production possibility froiitier t'urther out into the
commodity space.
There remains a significant difliculty with this analysis. which Kalilor
I972 first attempted to solve. Young argued that provided the increase in
supply of a commodity led to an increase in deinand for a l l other coin-
rnodities. growth would be cumulative. Kaldor pointed out that tor this to
occur, rota1 income must be rising and not merely expenditure on a partic-
ular commodity. In order to explore the conditions under which this would
occur, Kaldor puts emphasis on a figure who had already been significant
in Smith's analysis. the merchant. Smith realised that the merchant w;is
crucial to the development of the cominercizll stage of societal develop-
ment, and that "the slow progress of opulence" had been partly caused by
tlie general contempt towards merchants: "trade of a merchant . . . w a s
depreciated in the beginnings of' societv. . . . This mean and despicable
idea which they had ot' nierchants greatly obstructed the progress ot' coiii-
nierce. The merchant is. as it were. the mean between the nianukicturer
and the consunier" (W. 527). In niure rnodern tcrminolo_ry. Kaldor 1972.
 Reid - Disequilibrium in Adam Smith 103

1247. locates the source of “inside demand” in the activities of the mer-
chant: it is he who reconciles the frequent discrepancy between flow supply
and flow demand by varying his level of stocks. The emergence of the
merchant (as. even. the emergence of the philosopher) is in itself a con-
sequence of the increasing division of labour as society evolves: “A mer-
chant in Glasgow or Aberdeen who deals in linen will have in his
warehouse Irish. Scots and Hamburg linen. but at London there are sepa-
rate dealers in each of these. The greatness of the market enables one to
lay out his whole stock not only on one commodity but on one species of
a commodity and one assortment of it” (W, 355-56). It is false of Kaldor
to assume that Smith neglected the function of the merchant, but certainly
true that. lacking a theory of effective demand, he was not able to perceive
the role of merchanting activities in sustaining growth. And yet. all the
elements are there in Smith’s writings. He gives very full consideration to
the activities of merchants in markets for which prices are prone to fluc-
tuations: “The operations of the speculative merchant are principally em-
ployed about such commodities. He attempts to buy them up when he
foresees that their price is likely to rise, and to sell them when it is likely
to fall” ( W N ,133). Elsewhere, he talks of “traders and artificers” for whom
“it was the manifest interest of every particular class of them, to prevent
the market from being over-stocked, as they commonly express it. with
their own particular species of industry” (WN.141). In the same way as it
was proper for Young to regard his paper on increasing returns and eco-
nomic progress as “variations on a theme from Adam Smith,” so also it is
appropriate for us to view Kaldor’s emphasis on the merchant as a further
variation. Kaldor argues that provided the expansion of supply consequent
on the division of labour leads to an increase not only in the volume. but
also in the value of stocks held by merchants. this increase in induced
investment leads to an income flow which generates an increase in the
effective demand for other goods. In turn, this rise in the effective demand
for other goods creates in those markets further opportunities for the di-
vision of labour, and thus leads to further increasing returns.
The above argument requires a final refinement. What has been said so
far about “inside demand” emphasised competitive markets, which we
shall characterize as agricultural and subject to constant returns. Imper-
fectly competitive markets we shall characterize as being concerned with
manufacturing activities which are subject to increasing returns. In such
markets, it is the firms that generate “inside demand” by varying their own
levels of stock holding. If production varies in response to sales, then an
increase in demand will lead to an increase in both fixed and circulating
capital in the face of a run down of stocks, provided producers expect
markets to grow. It is this increase of capitals that is the induced investment
which enlarges etfective demand. This distinct ion bet ween manufactures
I04

and agriculture. with the t'orrner oti'cririg the morc plentiful opportunitics
lor the division 01' lahour. is entirely iri keeping with Sniith. Thc identili-
cation of dittercnt t'orms of induced irivestriieiit. the one taking place tor
comrnotlities in rcsponse to excess supply. and the other taking place tijr
manufacturers. in the face o f excess dcmand. is the distinct contribution
01' Kaldor. but one which logically t'ollows on t'roin ;I Sniithian view ot'
cconomic progress. with its emphasis on sectoral disequilibria. 'This exten-
sion ot' Smith brings with it a cautionarv note, iind in a more riioderri
context suggests that the 'cheerful' view cannot be supported in an u n -
qualified way. There are two untested assumptions in Kaldor's extension
o f Smith.' First, can it be assumed that merchants will increase not just
the volume. but also the value of stocks in the face of excess supply'!
Second, will tnanufhcturers necessarily be sutiiciently confident to respond
to excess demand by capital formatiori? Both assumptions depend on a
view which sees capitalist development as being at least orderly if not
equilibrating, without prices for cotnmodi ties ever collapsing. and with
cxpansionary long-run prospects tor manut'acturers.
I v. Corlc-llrsiorl
There is much that can be drawn trom the writings of as great a genius
as Adam Smith. Further. there is much that one might wish to draw from
him. so great is his authority. However. this brings with it the danger of
misinterpretation. which might do ;is much damage to great economists of
today as to Smith himself.
The first part of this article concentrated on interpretations of Smith
which seemed to miss something quite essential. The argument advanced
was that any modern restatement of Smith m w t consider a full growth
trajectory, embracing both the 'cheerful' and 'dull' phases of economic
tlevelopnient. Furthermore, i t should be expressed in sectoral terms.
though riot as in the modern classical approach. with constant returns to
scale technologies and uniform growth rates ;icross sectors. Finally. it
should see growth in disequilibrium terms. as, in Smith. i t is clearly ad-
vance in one sector that creates the possibility o! advance in another.
Having dismissed many types of modelling of Smithian growth as in-
appropriate, we suggested an approach which both attempts to be analyt-
ically tight and avoids sidestepping the heart of the matter: disequilibria
between sectors. The Young-Kaldor extensions of Smith are faithful to the
original, constitute a genuine analytical advance, and offer scope for fur-

9 . Kaldor expressed them as requirements on expectations: a ccrtain inelasticity ol' c x -

pccrations regarding prices: and a ccrtain clastic.ity ol' cxpcctations rcgarcling salcs. Tlicsc
are really long-tcrm expectations. and for thc growth process t o continue they iiiust hc
fulfilled. Clearly the fact that stocks are in thc short term being held at unanticipated lcvcls
iiieans that short-krm expcctations arc not ncc.cuarily hciiiy rcaliwcl.
 Reid - Disequilibriuni in Adam Sniith 105

ther development. Modelling sectoral disequilibria is fearfully diHicult.

and may require recoursc to simulation methods i t ;1 model of any size is
constructed. Gee & Jarvis 1984 has already made a preliminary excursion
into this difficult territory. The route will no doubt be hard and long, but
the direction now at least is clear, and is open for dl the stout-hearted to
follow.
T h e origins o t this article l i e in research undcrt;iken b y thc author during visiting professor-
ships at Queen's University. Ontario. and thc Univcrsity of Denver. Colorado. T h c author
wishcs t o cxprcss his thanks tor the congcnial environmcnts w h i c h made this w o r k possiblc.
T h e prcscnt vcrsion ol'thc paper has benetited from seminar prcscntiitions at the Univcrsitics
of' Edinburgh. St. Andrcws. Dundee. Denver. Uppsala. and the London School of Econulii-
ics. I n addition. thc comments of the t'ollowirlg academics h;ive been appreciated: T. 1.1.
M c C u r d y (Queen's University). N. Salvatore (University of Catania). C . Bcnetti ( U n i v c r -
sity of Paris X) and V. C.Walsh (University of Denver). T h e comments of a rcfercc havc
significantly intlucnccd thc linal form of this article. T h e usual disclainiers apply.

References
Abraham-Frois. G.. and E. Berrebi 1979. Theon of \wlite. prices and uccumie
lation. Carnbridgc.
Adelman. I . I96 I . Theories cg' economic growth nml efevelopmenr. Stanford.
Calif.
Atkinson. A . B. 1969. 'The timescale of economic modcls: how long is the long
run'?' Review oj' Economic Studies 35: 137-52.
Barkai. H. 1969. 'A formal outline of a Smithian growth model.' Qiiurter1.v Joitr-
no1 of Ecortomics 831396-414.
Bator. F. M. 1957. 'The simple analytics of wclfarc rnaximization.' Atneric*titi
Economic Review 47:22-59.
Blitch. C. P. 1983. 'Allyn Young on increasing returns.' Journcil of Post Keyws-
ian Economics 51359-72.
Dorfman. R. . P. Samuclson. and R . Solow 1958. Lirtetrr proRrcimmittaqcrrid ecw-
nomic tinci1,vsis. New York.
Eagly. R . V. 1974. The structure oj'economic theon,. Ncw York.
Eltis. W. A. 1975. 'Adam Smith's theory of economic growth.' In Skinner &
Wilson 1975, 426-54.
1 984. The classical theon of economic Rrowth. London.
Gee, J . M. A . 1981. 'The origins of rent in Adam Smith's Wealth ofNations: ail
anti-neoclassical view.' Hisrory of Political Economy 13: 1-1 8.
1985. 'Adam Smith: critical assessments.' Scottish Joirrnal of Politiccrl
Ecorromy 32209-1 8.
and R . J. Jarvis 1984. 'Gauss-Newtonised Smith: a formal model ot'
Smithian economic statics. growth and development.' Paper delivered at thc
1984 History of Economic Thought Conference, Guildford. U.K.
Hahn. F. 1963. 'On the disequilibrium behaviour o f a multi-sectoral growth
modcl.' Ecoiiomic Joicrtial 73:442-57. Reprinted in Hahn 1985. 301-2 I .
1974. 'On thc nature of equilibrium in economics.' Iiiaugural Lccturc.
Cambridge University. Cambridse. Reprinted in Hahn 1984. 43-7 1 .
I 984. Equilibriirni arid macroeconomics. Ox ford.
1985. A.loricy, groic*rhtilid stcihiliry. Oxford.
 Hclptnan. E.. and P. K . Krugman 1085. Altirket stritctttre t i n d fi)rei,qit trtide.
Canibridgc. Mass.
Hicks. J . R . 1965. C'upitczl a d Rrowtlt. Oxford.
I liggins. B . 1959. Econontic dri*elopntent. London.
Ilollandcr. S. 1973. 7 h c econontics t ! / . A h t n S t t i i t l i . London.
1980. 'On Prol'cssor Samuclson's canonical classical niodcl ol' political
cconoiny.' Joitrntil (4'E'c*onomicLiternturc I 8:559-74.
Joncs. K. W.. and P. El. Kcncn. cds. 1084. Iftiidbook t i ' inrcrilcitiotwl c ~ * o i t o n ~ -
ics. vol. 1 . Anistcrdani.
Jorgenson. D. W. 1061 . 'Stability o f :I tlyn:iiiuc input-output systcin.' Rerim*of'
E:cottotnic Stitdius 76: 105- I 6.
Kaldor, N. 1070. 'The case tor regional policy.' Scwtti.rh J ~ i t r oj' ~ ~Po/itic.iil
l
Ec.onorn!~17:337-48.
1972. "Thc irrclevancc ot cquilibriuiii cconomics.' Economic. Joitriicil
82: 1237-55.
1985. Ec*onomic*switltoui equilibrirtrn. Carditi.
Kirzner. I . M.. cd. 1982. Method. proc.'o.v.r( i t i d Aiistrkitt ec*onot?tics.Lexington.
Mass.
O'Brien. D. P. 1975. The clussical economists. Oxford.
Iketo. V. 1927. Manual ofpolirical ec*owtnj. Translation ( 197 I ) of the Frcnch
edition by A n n S. Schwier. London.
Raphael, D. D. 1985. Adam Smith.Oxford.
Reid, G. C. 1975. An anul?tic*ulstud? of price letidership. Ph.D. thcsis. Univcr-
sity of Edinburgh.
1979. 'An analysis of the firm, market structure. and technical progress.'
Scofrish Journal of Political Econoinv 26: 15-32.
1985. 'Adam Smith on accumulation: a formal analysis.' Univcrsity of
Edinburgh Dcpartmcnt ol' Economics Discussion Paper ( 1985: V l ) .
Rosicr. M. 1984. 'On Adam Smith's modcl of' rcproduction and accumulation.'
Discussion Papcr. Univcrsity of Paris. VII.
Saniuclson. P. A . 1947. Iiiitndutions of cwnotnic. tinu1,vsis. Cambridge. Mass.
1977. ' A modem theorist's vindication of Adam Smith.' Antericciti Eco-
nomic Review 67:42-49.
1978. 'The canonical classical modcl ol' political economy.' Joitrnnl of'
Econotnic Literature 16:I4 15-34.
Skinner, A . S.. and T. Wilqon. eds. 1975. Essa.vs on Adam Smith. Oxford.
'I'hweatt. W. 0 . 1957. ' A diagrammatic prescntation of Adam Smith's growth
model .' Sociul Research 24:227-30.
Viner, J. 1928. 'Adam Smith and laisscz-t'airc.' In J. M. Clark et al.. Adatn Sntith
1776-1 926 ( 1928).
Walsh. V., and H . Gram 1980. Classicti1 c i d neoclnssical theories of getiernl
eqitilihriirrn. Oxford.
Young, A . 1928. 'Incrcasing rcturns and economic progress.' E'cmornic Jottrntil
3 8 527- 42.
All refercnces to the work of Adani Smith are to thc Works and correspondence
ofAdam Srnirh undertaken by the University of Glasgow. Works cited in this
;irticlc arc: An enquiry irtto the nature t i r i d causes of the \r*ealthoj nations. edited
by R . H.Campbell and A. S . Skinner ( 1976): Lectures on jiirisprttdence. cditcd
by R . L. Meek, D. D. Raphael. and P. G. Stein (1978); Essa,vs of1 plrilosophiccil
subjects. edited by W. P. D. Wightman ( 1980). All were published by the Oxtord
University Press, Oxford.