Leijonhufvud, “Capitalism and the Factory System,”
in R. N. Langlois, ed., Economic as
a Process: Essays in the New Institutional Economics.
New York: Cambridge University Press, 1986, pp. 203-223.
[Page numbers in brackets.]
utilizes, on the one hand, mathematical techniques and, on the other, thought
experiments, parables, or stories. Progress may stagnate for various reasons.
Sometimes we are held back for lack of the technique needed to turn our stories
into the raw material for effective scientific work. At other times, we are
short of good stories to inject meaning into (and perhaps even to draw a moral
from) our models. One can strive for intellectual coherence in economics either
by attempting to fit all aspects of the subject into one overarching
mathematical structure or by trying to weave its best stories into one grand
This paper attempts to revive an old
parable, Adam Smith’s theory of manufacturing production, which has been
shunted aside and neglected because it has not fitted into the formal structure
of either neoclassical or neo-Ricardian theory. The paper attempts to persuade
not by formal demonstrations (at this stage) but by suggesting that the parable
can illuminate many and diverse problems and thus become the red thread in a
theoretical tapestry of almost epic proportions.
The subject may be approached from either
a theoretical or a historical angle. Regarding the theoretical starting-point,
it is possible to be brief since the familiar litany of complaints about the
neoclassical constant-returns production function hardly bears repeating. The
one point about it that is germane here is that it does not describe production
as a process, i.e., as an ordered sequence of operations. It is more like a
recipe for bouillabaisse where all the ingredients are dumped in a pot, (K, L),
heated up, f(·), and the output, X, is ready. This abstraction
from the sequencing of tasks, it will be suggested, is largely responsible for
the well-known fact that neoclassical production theory gives us no  clue
to how production is actually organized. Specifically, it does not help us
explain (1) why, since the industrial revolution, manufacturing is normally
conducted in factories with a sizeable workforce concentrated to one workplace,
or (2) why factories relatively seldom house more than one firm, or (3) why
manufacturing firms are “capitalistic” in the sense that capital
hires labor rather than vice versa.
Revolutions: agricultural and
The story of the
industrial revolution has often been told around the theme of technical
invention and innovation in spinning and weaving, in steel-making and power
generation, in freight transportation, and so on. Similarly, the agricultural
revolution that preceded it sometimes seems just a long catalogue of new crops,
new rotations, new ways to drain or fertilize land, new techniques of selective
If one looks at the two revolutions
instead from the standpoint, not of technological history, but of a “New
Institutional” history, the agricultural revolution becomes primarily the
story of enclosures and the industrial one the story of the coming of the
factory system and, eventually, of the joint-stock corporation.
It is customary in standard treatments of
l8th century English economic history to hail both these organizational
developments as obvious examples of “progress.” Carl Dahlman (1980,
pp. 209-210) has pointed out that the juxtaposition of the two poses something
of a paradox, for one process seems to be almost the reverse of the other. The
reorganization of agriculture, known as the “enclosure movement,” was
a move away from the collective “team” working of village land. Each
family ended up working their own farm. Correspondingly, it required the
“unscrambling” of joint ownership rights in land held in common (and
of obligations owed to the collective). In the somewhat later reorganization of
manufacturing we have the reverse. The coming of the factory was a move towards
collectively organized modes of production. It replaced the “family
firm” craftshop and the putting-out system. The craftshop run by a master
craftsman with a couple of journeymen and apprentices and with family helpers
had been the dominant type of manufacturing business since the early Middle
Ages. Under the putting-out system, an entrepreneur “put out”
materials for processing at piece rates by workers who usually worked at home.
The factory pulled the workforce in under one roof. Later on, the limited liability
manufacturing corporation arose to pool individual titles to physical capital
in the joint stock arrangement. 
Thus the Dahlman paradox: What is
progress in manufacturing is backwardness in agriculture and vice versa!
The open field system and enclosures are admirably analyzed in Dahlman’s book.
The present inquiry concerns the Factory System.
The Factory System.
tended, of course, to marvel at the new inventions and to be deeply impressed
by the (very visible) role of fixed capital in the new factories. The most
prominent features of the factories were (a) the size of the workforce in one
and the same workplace, and (b) the new machinery. The impulse has been to
explain (a) by (b), that is, to take for granted that the novel spinning-frames,
weaving looms, steam engines, etc., made the explanation for factory
organization of the work almost too obvious to require explicit comment.
Some histories of the industrial
revolution have taken the line that the new machinery explains the factories. The
point has been made, for example, that the early steam-engines, with their low
thermal efficiency, were very large, stationary ones; consequently, if one
wanted to utilize steam-power, one had to pull a sizeable labor force in under
one roof and run the various machines of the factory by belt-transmission from
a single source. The answer suggested in this sort of illustration is that the
new technologies introduced obvious economies of scale (e.g., in power
generation) that led quite naturally to large-scale factory production.
Economies of scale were obviously one
aspect of the story. But they do not make the whole story. Some 150 years
later, small-scale electrical motors removed the basis for the particular type
of scale-economy just adduced — but did not, of course, thereby undermine the
factory system. (At the same time, the economies of scale in generating
electricity were even more formidable than they had been in steam power.) We
might also check some centuries earlier. The fourteenth century Arsenal of
Venice was one of the wonders of the world for the size of the labor-force
concentrated in it. Yet, the organization of shipbuilding in the Arsenal was
not that of a single firm; instead, numerous craftsmen, owning their own tools,
each with a few journeymen and apprentices, operated within the Arsenal and
cooperated via exchange transactions in the building and outfitting of ships.
In short, the famous Arsenal was not a factory and not a firm. 
There are other examples of large
workforces in one location before the industrial revolution. Large woolen
manufacturing workshops existed in England since at least the beginning of the
16th century. Their size would not have been dictated by machine technology.
While medieval mining was in general organized as independent partnerships of
miners, by the sixteenth century, deeper mineshafts with dangerous ventilation
and drainage problems raised the capital requirements in mining beyond the
means of artisan miners. The mines became capitalist firms. Alum, bricks,
brass, and glass, were 17th century examples of technology dictating production
in sizeable establishments.
In these instances, the workplaces were factories and were firms.
The putting out system was also replaced
by the factory system. It exemplified capitalist control of production often
without capitalist ownership of the means of production.
The organization could be large but the workplaces were, of course, small.
It is not all that obvious, therefore,
what role should be assigned to indivisible machinery in explaining the
emergence of the factory as the dominant form of manufacturing enterprise. Some
questions remain. Why, for example, did not the steam engine simply lead many
independent masters to locate in the same workplace (and, perhaps, pay rent for
the right to attach their new-fangled machines to the overhead, steam-powered
The Classical theory of the division of
There is one
contemporary observer whom economists might be particularly inclined to pay
attention to, namely, Adam Smith. The Wealth of  Nations is,
of course, a bit early (1776) for the mechanized, steam powered, relatively
fixed-capital-intensive Factory System to have become established as the wave
of the future. Even so, it is worth
remembering that Smith did not dwell much on machinery as one of the
“Causes of Wealth.” Instead, of course, he made the “division
of Labour” his grand theme. In fact, he treats the role of machinery
as important but as secondary and subsidiary to “increasing division of
labor” in his account of economic progress:
E]very body must be sensible how labour is
facilitated and abridged by the application of proper machinery. It is
unnecessary to give any example. I shall only observe , therefore, that the
invention of all those machines by which labour is so much facilitated and
abridged, seems to have been originally owing to the division of labour….
(Smith, 1776 [1937, p. 9].)
The Classical theory of the division of
labor was greatly advanced by Karl Marx in Das Kapital. In his day, of course, the factory system
was the wave of the present. Marx made the use of machinery the criterion of
“modern industry,” which he associated with factories. At the same
time, however, he emphatically agreed with Smith that mechanization followed
from the division of labor. In Marx’s historical schema, capitalism was
subdivided into a manufacturing period (“from the middle of the 16th to
the last third of the 18th century” (Marx, 1906, p. 369)) and the
subsequent modern industrial epoch. Manufacturing, in Marxist terminology,
resulted from applying the principles of the division of labor to as yet
In Smith’s famous pin-making illustration
of the benefits of the division of labor, two modes of organizing production
were contrasted.  Prejudging matters a little, let us call them
“crafts production” and “factory production” respectively.
In Crafts production, each
craftsman sequentially performs all the operations necessary to make a pin. In Factory
production, each worker specializes in one of these operations so that
“the important business of making a pin is, in this manner, divided into
about eighteen distinct operations which, in some manufactories, are all
performed by distinct hands ….”
(Smith, 1937, p. 4-5.)
Suppose, for illustration, that we have
five craftsmen producing a product that requires five successive operations.
These must be undertaken in temporal sequence running from left to right in
Here each artisan is working at his own
pace and the individuals differ in (absolute and comparative) skill across the
Suppose, next, that we simply rearrange
the work in some given workshop as indicated in Figure 9.2:
People who previously worked in parallel
now work in series. Worker b now performs only operation 2 but does so on all
units of output produced by the team. Each individual now has to work at the
pace of the team. This, obviously, makes supervision of work effort easier.
Note, however, that we do not change the engineering descriptions of the
operations performed, we do not change the tools used, and we do not change the
people involved. We  might expect output to be unchanged as well,
therefore, Yet both Smith and Marx would tell us to expect a large increase in
productivity from this reorganization of the work.
The sequencing of operations is not
captured by the usual production-function representation of productive
activities; nor is the degree to which individual agents specialize. A
production function simply relates a vector of inputs to one or more outputs
without specifying the method by which the tasks involved are coordinated. Thus
Smith’s division of labor — the core of his theory of production — slips
through modern production theory as a ghostly technological change coefficient
or as an equally ill- understood economies-of-scale property of the function.
The economies achieved by switching from
crafts to factory production arise from increased division of labor. In the
above example, labor was entirely “undivided” to begin with, so that
the conversion takes us from individual production to team production.
There are three aspects to this that deserve comment. First, the specialization
of labor in team production will require standardization of product.
Under crafts production, in contrast, the skills and care of individual
artisans will be reflected in non-standard output. Second, serial production
requires coordination of activities in the sense of the time-phasing of
the inputs of individual workers. Third, the labor of individual workers become
complementary inputs. If one work-station on an assembly line is
unmanned, total product goes to zero.
So far we have supposed that the number
of workers and the tools are unchanged and that the only change arises from
their improved coordination. But it is obvious that the conversion from
crafts to factory production will present opportunities to economize on inputs.
The switch is capital-saving. This
is an aspect easily missed. The reorganization of production undertaken to
increase the division of labor will very often also create opportunities for
mechanizing some stage of the process. Hence what we tend to observe is that an
increase in fixed capital takes place at the same time. The impression we are
left with is that productivity increases are normally due to more capital
intensive technology being adopted.
But the pin-making illustration is a counter-example.
In crafts-production, each artisan would
be equipped with a full complement of pin-making tools. Suppose, for
simplicity, that there is a different tool for each of the five stages in the
series. Then, four out of five tools are always idle when artisans work
in parallel under crafts-production. In factory production, only one complement
of tools is needed, not five.
It is possible that the more decisive
capital-saving incentive may be the opportunity to economize on
goods-in-process inventories. Suppose that, under crafts production,
considerable time (and concentration) is lost in switching from one task to the
next. A master craftsman with a thick enough market to allow him to produce in
batches would then perform operation 1 x times, before moving on to operation
2, etc. If his “dexterity” (as the Classical writers used to say) at
each task were equal to that of the specialized factory worker, the factory’s
competitive edge would lie mainly in its lower working-capital requirements.
Economizing on goods-in-process is likely to have been particularly important
in the evolutionary struggle between the factory and the putting-out system.
The switch to factory production will
also save on human capital. No worker need possess all the skills
required to make a pin from beginning to end. Under crafts production, each
individual has to spend years of apprenticeship before becoming a “master
pinmaker.” In factory production, the skills needed to perform one of the
operations can be quickly picked up. The increased productivity resulting from
specialization on simple, narrowly defined tasks is the advantage arising from
increased division of labor most emphasized by the Classical economists.
Correspondingly, the decreased investment in human capital is the disadvantage
that most concerned them.
Horizontal and vertical division of
There are two
dimensions along which the division of labor may be varied. Adam Smith drew
examples from both (without, however, making the distinction clear). The
manufacture of pins illustrates what we will call vertical division of
labor. Recall his observation that “in so desert a country as the
Highlands of Scotland, every farmer must be butcher, baker and brewer for his
own family.” When the growth of the market turns slaughtering, baking, and
brewing into specialized occupations, we have examples of horizontal division
The distinction is seldom drawn in the
literature. This may be in part because those authors, who see the advantages
of division of labor as deriving primarily from the concentration of time,
experience, and ingenuity on part of individuals on a narrower range of tasks,
are looking simply for all the differentiations of functions that the
expansion of markets will allow. Charles Babbage improved on Smith’s statement
of the division of labor by making clear how functional differentiation brings
comparative advantage into play also inside the individual firm.
the master manufacturer, by dividing the work to be executed into different
processes, each requiring different degrees of skill or force, can purchase
exactly that precise quantity of both which is necessary for each process;
whereas, if the whole work were executed by one workman, that person must
possess sufficient skill to perform the most difficult, and sufficient strength
to execute the most laborious, of the operations into which the art is divided.
(Babbage, 1833, pp. 175-76.)
But there are reasons for making the
proposed distinction. An increase in the vertical division of labor requires less
skilled labor at the various stages of the manufacturing process. Increased
horizontal division of  labor does not in general carry this implication
and is perhaps more likely to mean an increase in human capital per worker.
Furthermore, increased horizontal division is a question simply of minimum
economical scale, whereas vertical division of labor results from an
increasing returns to scale technology.
This implication of pin-making technology
may be another reason why the distinction is most often fudged, particularly in
the neoclassical literature. Stigler, in his famous article on the subject,
notes the dilemma bequeathed by classical to neoclassical theory: “Either the division of labor is limited by
the extent of the market, and, characteristically, industries are monopolized;
or industries are characteristically competitive, and the theorem is false or
of little significance. Neither alternative is inviting” Stigler (1951 [1968,
p. 129]). Marx saw the significance of
the distinction very clearly. The consequences of expansion of the market for a
branch of manufacturing, he pointed out, would depend upon the technology. He
distinguished two “fundamental forms,” namely, “heterogenous
manufacture” and “serial manufacture.” The latter, of course,
was exemplified by Smithian pin-making and offered opportunities for vertical
division of labor. As an example of the latter, Marx used watch manufacturing.
All the parts of a watch could be separately manufactured for final assembly.
This “makes it … a matter of chance whether the detail labourers are
brought together in one workshop or not.” Heterogenous manufacture might
be carried out under the putting-out system, therefore. (Marx, 1906, pp.
impetus to exploit the economies afforded by “vertical” division of
labor would seem to explain, therefore, many of the social consequences of the
19th century factory system that have been the object of so much adverse
1. When labor is subdivided vertically, less skill is
required, less versatility as producer is acquired by the individual worker.
 The use of child labor at some workstations often becomes feasible.
2. No normal prospect of promotion or improvement in
social status is to be expected; the unskilled workman does not become a master
of his guild by sticking to his job for many years.
3. More discipline is required and of a sort that
most people will find irksome and that most rural emigrants would have to be
taught; you cannot “work at your own pace,” you have to be on time;
random absenteeism must be subject to relatively severe sanctions.
4. “Alienation from the product”: no worker
can take personal pride in the output or its quality.
this sort do not give one grounds for blundering into the much controverted
subject of the development of standards of living during the Industrial
Revolution in Britain. The point to be made is simply that the competitive
pursuit of the productivity gains afforded by the “vertical” division
of labor will explain many of those conditions in industry that were criticized
by contemporary observers.
The Extent of the Market.
In our simple
five-man example, a doubling of output under crafts production will require a
doubling of all inputs. Under factory production, some economies of scale will normally
be present. In factory production, “the division of labor depends on
the extent of the market” — and so, therefore, do the scale economies
that can be realized. These will be of two kinds.
scale economies. Suppose, in the example, that one of the
workers (worker d at work-station 4, let’s say) is idle half the
time after conversion to factory production. Then double the output can be had
with nine workers, and the flow of work would be organized as in Figure 9.3.
This is the source of increasing returns
emphasized by Georgescu-Roegen as almost universally present in manufacturing
— but not, as all the Classical economists agreed, in agriculture. Even on the
sophisticated assembly lines of a large-scale factory there is almost bound to
be some “factor” (“fund”  in Georgescu-Roegen’s
terminology) whose input-stream is not perfectly continuous. Babbage’s
“master manufacturer” cannot always divide the work so as to
“purchase exactly that precise quantity” of the services of the
factor that is technically required to produce his output. A machine that is
idle half the time cannot be replaced by half a machine employed all of the
time. But it may be possible to double its utilization rate if, say, the
machine can be shared between two parallel assembly lines and the firm
can sell twice the output.
These “parallel scale
economies” are probably never totally exhausted. In our five-stage
example, it might be found, for instance, that worker b is busy only 80 per
cent of the time, in which case a quintupling of output can be had with only a
quadrupling of stage 2 workers. And so on. But it is clear that, if we keep the
number of serial stages constant, these economies of parallel replication
become less and less significant as output is increased. It can in fact be
shown that this is a case of assymptotically constant returns (although with a
non-monotonic approach to the assymptote).
Longer series scale economies. Smith,
Marx, and Mill, however, were thinking more of another source of economies of
scale, namely, increased vertical division of labor. As the extent of the
market grew, opportunities would arise, they thought, for further efficient
subdivision of the production process into a greater number of serial tasks.
This vertical differentiation would not only be efficient in itself but, as it
proceeded, it would open up new possibilities for exploiting scale economies of
the Georgescu- Roegen kind. 
Mechanization and division of labor as a
As one proceeds
with the analysis of this Classical Division of Labor theory, it increasingly
escapes the analytical categories of static neoclassical production theory. The
Classical theory becomes a theory of an evolutionary process, rather
than a theory of the rational choice between known alternatives.
Recall that Smith and Marx both insisted
that the New Division of Labor preceded the mechanization of industry.
They also thought that one led to the other, and they thought it rather
obvious what the causal link was: as one subdivides the process of production,
vertically, into a greater and greater number of simpler and simpler tasks,
some of these tasks become so simple that a machine could do them. The
mental task of analyzing the production process so as to carry through the
division of labor leads to the discovery of these opportunities for
mechanization. Once the principles of the division of labor are mastered, the
discovery of how industry can be mechanized follows.
Mechanization, in turn, will renew the
sources of economies of scale. Suppose each stage of what was previously a five
stage process is subdivided into two. Suppose further that it is then
discovered that stage 4b can be mechanized. But at the old scale of the
enterprise, the 4b machine may be idle 90 percent of the time. In that case,
the most economical scale of production has multiplied tenfold
Differentiation of function: Capital and
The process leads to increasing
functional differentiation of both capital equipment and labor. But in one
respect the consequences are quite different — and it turns out to be a
socially important respect.
Although the tasks that become mechanized
tend to be quite simple, completely standardized tasks, the machines very often
will be extremely specialized to doing just this one task (or series of tasks)
in the production of just one product. This means that they may have no
alternative employment. This differentiation of equipment can be observed also
in simple handtools: ” so soon as the different operations of a labour-process
are disconnected the one from the other, and each fractional operation acquires
in the hands of the detail labourer a suitable and peculiar form, alterations
become necessary in the implements that previously served more than one
purpose…. In Birmingham alone 500 varieties of hammers are produced, and not
only is each adapted to one particular process, but several varieties often
serve exclusively for the different operations in one and the same process”
(Marx, 1906, pp. 374-5.) 
In the course of this vertical
subdivision of the production process, labor gets increasingly unskilled.
The sociocultural consequences are disturbing. Adam Smith gradually became so
convinced that the division of labor tended to produce an unskilled,
illiterate, brutalized proletariat that in the end his Wealth of Nations
contained “Two Views” of the division of labor. (West, 1964;
Rosenberg, 1965). In the early chapters, it was The Source of the Wealth of
Nations. Towards the end of the book, it became the ruination of the laboring
classes. This outlook Marx took over.
From the more narrowly economic
standpoint, the vertical subdivision of production makes the machines
functionally more specialized or “dedicated.” A particular machine,
as a consequence, may have few alternative uses but is also not easy to
replace. With labor, the result is rather different. The individual worker
becomes a “detail laborer,” i.e., specialized in the sense that, when
at work, he performs only one task. But the task is an unskilled one. The
worker, consequently, can be easily replaced and can also easily qualify for
alternative tasks. Thus, increasing specialization has quite different
implications for the competitive position of capital and of labor,
respectively. We will return to this point shortly. .section ‘American and
Japanese traditions in production management’ The American tradition in
production management has made the most of the “static” advantages of
the division of labor: minimal human capital requirements, maximum
“dexterity” in performance of individual tasks, and minimal time lost
in switching between tasks — these are the principles stressed on Henry Ford’s
assembly lines and in Taylorite time-and-motion studies.
Apparently, Japanese production
management violates all of these principles. Each member of a production team
is supposed to learn every work-station on the assembly line. Human capital
input is maximized rather than minimized. But the “dynamics” of the
Smithian evolutionary process are improved. The Japanese teams are better at
“discovering” potential improvements in both products and methods.
The Capitalist Firm
Consider next an idyllic
thought-experiment of so-called team-production:
a number of individuals come together for the purpose of producing a
particular commodity. In the “Original State,” we suppose, there are
no marked distinctions of wealth, power or status among these people. Some of
them will contribute their skills and labor, others will commit themselves to
bring machines to the joint enterprise.
We may assume that they will decide to
take advantage of the Smithian economies of vertical division of labor and so
set up production in the form of a single, long assembly line. For simplicity,
let there be n stages of production, n machines, and n operatives
— one per machine. The product could also be produced by n individual
artisans using a set of simple hand- tools or in k shorter assembly
lines of n/k workers using less-specialized machines. But we presume,
with Smith and Marx, that by setting up on one long assembly line, the
collective effort will produce a larger output with the same resources.
The questions are: How many firms will
there be? Will the typical firm be a “capitalist” one? If so, why?
One can imagine the possibility of n successive
firms, each one buying the output of the stage preceding and selling to the
stage succeeding. In half of these firms (one might also imagine) the owner of
the machine hires the operative and pays him wages, while in the other half the
worker rents the machine he is working with. But these imaginings, of course,
fit singularly ill with the ways in which we find modern manufacturing to be
Since the team utilizes the economies of
scale due to the division of labor, the enterprise earns a joint rent (or a
“surplus,” if you will). Total sales-proceeds exceed the sum of the
earnings that the inputs would find in alternative employments. The joint rent
is a Ssssnake in this paradise. For how is it to be divided? In our
illustration, all the inputs are assumed to be strictly complementary. If one
machine is withdrawn from the assembly line, total output falls to zero. If one
worker is missing, the consequence is the same. Marginal productivities will
not supply the criteria for the distribution of product.
The division of the joint rent becomes a
bargaining problem. Let the members of the collective form coalitions amongst
themselves and bargain against the rest. How well might the various coalitions
do? How stable would we expect them to be?
Consider first how the bargain might go
between the machine owner (“capitalist”) and the operative
(“labor”) at one of the work-stations on the presupposition that the
total sum going to this work-station has somehow been arrived at. Each can
threaten the other to withhold his input so that their joint income will go to
zero. But the bargaining situation is not symmetrical. There are plenty of
unskilled laborers in the market, but few if any substitutes for the
specialized machine. This might make us  suspect a tendency for the
capitalist to walk away with the joint rent, leaving the laborer with a wage
equal to his alternative earnings. But there is also another asymmetry: the
unskilled laborer has many, the specialized machine few, alternative employment
opportunities. If, therefore, the laborer could threaten to “fire”
the machine, his bargaining position would be very strong indeed. The question
becomes who can fire and replace whom? Or: Who owns the work-station, the
machine owner or the operative?
To get a clue to this question, consider
the bargaining situation among the “capitalists.” Each machine owner
can threaten to reduce output and, therefore, everyone else’s earnings to zero
— until a replacement for his machine can be found. But, again, the market for
very specialized machines will be thin, so replacements — and alternative
employments — for them are hard to find. Any agreement about the division of
earnings amongst the machine owners would be extremely unstable. So unstable, in fact, that some organization
of production that avoids the complementarities between the highly specialized
inputs of cooperating owners might be preferred — even at the cost of
foregoing the advantages of the division of labor. To sink one’s capital into
these dedicated machines will not appear to be an attractive investment —
unless some stable organizational form can be found.
The solution, of course, is to prevent
individual capitalists from owning and controlling specific machines. Instead,
a “firm” is formed and any capitalist who joins has to give up
ownership of his machines and accept “shares” in the firm. Thus the
assembly-line is vertically integrated into one firm. We might find a “market-gap”
between firms along the production chain at some stage where the market in the
intermediate product issuing from the stage is thick enough so that firms on
both sides of the gap are safe from “hold-ups.”
The formation of a firm as a solution to
the machine-owners’ bargaining problem has one additional advantage (for them):
it creates a cartel  of capitalists that bargains as one unit against
workers. This cartel will “own” the work-stations. It can fire and
replace workers; the workers cannot threaten to fire and replace the dedicated
machines. The non-unionized worker is not going to come out of that contest
with any part of the joint rent (unless, of course, he has some firm-specific
capital). As long, at least, as unions can be kept illegal, the factory owners
will continue to appropriate all the rent.
Unionization will look like labor’s best
bet in this situation. Workers cannot pool their labor power, as the
capitalists pool their physical capital, in order to hire the machines at a
rental that would leave the joint rent going to the workers of the
labor-managed firm. Labor will not be owned and specialized machinery is not
for hire. The producer cooperative is a possible compromise form but, on the
whole, successful enterprises started as worker partnerships are going to end
up owning capital and hiring labor — which is to say, end up as capitalist
firms. Unions that do succeed in capturing part of the joint rent, on the other
hand, might thereby discourage capital accumulation and the further productive
subdivision of labor and hence weaken the competitive position of the
enterprise over the longer run.
The labor union is a subject on which
economics has a less than secure grasp. In neoclassical economic theory, unions
are just another pernicious form of monopoly. The alternative “labor
relations” tradition tends to reject economic theory and to draw lessons
more friendly to unions from labor history. Perhaps the view of the
manufacturing firm presented here might provide ground on which theoretical and
historical analysis could finally meet?
Fluctuations and growth.
of the pin-making technology is so simple as to be little more than a metaphor.
It is obviously capable of considerable formal elaboration. But at this point the question is whether
there are good reasons to prefer it to that other simplistic metaphor, the
neoclassical production function. The Smithian production function may well
have advantages in areas other than the ones discussed in this essay. It may be
worthwhile, in conclusion, to indicate some of these potential applications.
One of the mainstay stylized facts of
applied macroeconomics is that employment in manufacturing fluctuates less than
proportionally to output over the business cycle. Most macroeconomic models
assume a  neoclassical constant returns to scale technology and most
macroeconomists explain the “Okun’s Law” phenomenon as reflecting the
“hoarding” of labor, in particular of workers with firm-specific
skills, during recessions. According to this hoarding hypothesis, firms keep
workers on during recessions, although they are not needed in production, in
order to make sure their skills are available when business picks up again.
The Smithian increasing-returns
technology suggests a competing hypothesis. Firms that utilize the scale
economies of parallel series (Figure 9.3) will reduce output by shutting down,
say, one assembly line of two. But the workstation that the two lines have in
common cannot be left unmanned. Thus, half the workforce cannot be laid off
when output is cut in half. By the same token, the laid-off worker
cannot by cutting his own wage get the line started up again. Individuals are
not able by marginal wage-cutting to expand the number of production jobs being
offered at the factory in recession.
When the Extent of the Market determines
the Division of Labor, economic growth will bring productivity gains. The
growing economy will show increasing division of labor not only within firms
but among firms. The economy becomes more complex as it expands. When, in our
simple illustration, the work of the five artisans was reorganized into a
five-man factory, the production process became more complex in the
straightforward sense that the number of people cooperating in making any given
unit of output increased. It is this increasingly complex coordination (when it
can be maintained!) of larger and larger numbers of specialists that shows up
as increasing productivity. It is  perhaps over- optimistic to hope
that explicit modelling of division-of-labor production would give us an
econometric handle on the Solow-Denison growth residuals. But it could give us
a better qualitative understanding of how economic development differs from
mere economic growth, which would be worth having. An economist used to
thinking of production in terms of the Smithian division of labor model is
likely to be more impressed with the dangers of protectionism, for instance,
than colleagues whose thinking run in neoclassical or neo-Ricardian channels.
To the welfare losses arising from impediments to trade in constant (or
diminishing) returns models, the Smithian economist
would add not only the static loss of scale economies foregone but also the
dynamic losses of innovative discoveries foregone when the Smithian
evolutionary process is stemmed. While the loss of competitive improvements
never made may be unquantifiable, comparisons between open and closed economies
suggest that they are nonetheless the most significant category of welfare
losses due to protectionism.
The theory of the capitalist factory
outlined here shares elements with other explanations that have been proposed.
It is not to be expected, however, that the proponents of these other theories
will be entirely happy with it. The present theory stresses the complementarity
of inputs as a central problem as do Alchian and Demsetz (1972), but it does
not at all accept their insistence that the bargain between capital and labor
is essentially symmetrical. My story has a great many points in common with
Williamson’s “Organization of Work” (1980) but differs from his in
seeing technological rather than transaction-cost considerations as central.
Finally, like Marglin (1974), I recognize an element of power in the
capital-labor bargain as essential. Marglin would insist, however, that the
capitalists’ control of production has no technological or efficiency
rationale, while I see the capitalists’ power as rooted in the efficient,
Alchian, Armen and Harold Demsetz. 1972.
“Production, Information Costs, and Economic Organization,” American
Economic Review, 62(5), December.
Arrow, Kenneth. 1979. “The Division
of Labor in the Economy, the Polity, and Society,” in Gerald P.
O’Driscoll, Jr., ed., Adam Smith and Modern Political Economy. Ames,
Iowa: Iowa State University Press. 
Babbage, Charles. 1833. On the Economy
of Machinery and Manufactures, London: Charles Knight, third edition.
Buttrick, John. 1952. “The Inside
Contracting System,” Journal of Economic History, 12.
Dahlman, Carl J. 1980. The Open Field
System and Beyond, New York: Cambridge University Press.
Dahmen, Erik. 1971. Entrepreneurial
Activity and the Development of Swedish Industry, 1919-1939, Homewood,
Ill.: Richard D. Irwin.
Georgescu-Roegen, Nicholas. 1972.
“Process Analysis and the Neoclassical Theory of Production,” American
Journal of Agricultural Economics, 54(2): 279-294.
Georgescu-Roegen, Nicholas. 1976. Energy
and Economic Myths. New York: Pergamon Press.
Gioja, Melchiorre. 1815. Nuovo
Prospetto delle Scienze Economiche. Milan: Presso G. Pirotta..
Golden, Claudia and Kenneth Sokoloff.
1984. “Women, Children and Industrialization in the Early Republic:
Evidence from the Manufacturing Censuses,” Journal of Economic History,
Ippolito, Richard S. 1977. “The
Division of Labor in the Firm,” Economic Inquiry, 15(4): 469-492.
Klein, Benjamin, Robert G. Crawford, and
Armen Alchian. 1978. “Vertical Integration, Appropriable Rents, and the
Competitive Contracting Process,” Journal of Law and Economics,
Lane, Frederic C. 1973. Venice, A
Maritime Republic. Baltimore: Johns Hopkins University Press.
Lundberg, Erik. 1959. “The
Profitability of Investment,” Economic Journal 69: 653-677.
Mantoux, Paul. 1962. The Industrial
Revolution in the Eighteenth Century, New York: Harper and Row, revised
Marglin, Stephen A. 1974. “What Do
Bosses Do? The Origin and Functions of Hierarchy in Capitalist
Production,” Review of Radical Political Economics, 6(2): 60-112.
Marx, Karl. 1906. Capital, New
York: Modern Library. Mill, John Stuart. 1964. Principles of Political
Economy. New York: Ashley edition.
Nef, J. U. 1934. “The Progress of
Technology and Growth of Large-scale Industry in Great Britain,
1540-1640,” Economic History Review, 5(1): 3-24.
Pratten, Clifford F. 1980. “The
Manufacture of Pins,” Journal of Economic Literature, 18(1): 93-96.
Rae, John. 1964. Statement of Some New
Principles on the Subject of Political Economy. New York: Augustus M.
Kelley. [First published in 1834.]
Rosenberg, Nathan. 1965. “Adam Smith
on the Division of Labour: Two Views or One?” Economica, 32:
Smith, Adam. 1937. An Inquiry into the
Nature and Causes of the Wealth of Nations. New York: Modern Library
Sokoloff, Kenneth L. 1983.
“Investment in Fixed and Working Capital During Early Industrialization:
Evidence from U.S. Manufacturing Firms,” UCLA Working Paper No. 311,
Sokoloff, Kenneth L. 1984. “Was the
Transition from the Artisanal Shop to the Non-Mechanized Factory Associated
with Gains in Efficiency?: Evidence from the U.S. Manufacturing Censuses of
1820 and 1850,” UCLA (mimeo).
Stigler, George. 1951. “The Division
of Labor is Limited by the Extent of the Market,” Journal of Political
Economy, 59(3), reprinted in Stigler, The Organization of Industry,
Homewood, Ill.: Richard D. Irwin, 1968.
Thompson, E. P. 1967. “Time, Work-Discipline,
and Industrial Revolution,” Past & Present, December. 
Weitzman, Martin L. 1982.
“Increasing Returns and the Foundation of Unemployment Theory,”
Economic Journal, 92: 787-804.
West, E. G. 1964. “Adam Smith’s Two
Views on the Division of Labour,” Economica, 31: 32-32.
Williamson, Oliver E. 1975. Markets and
Hierarchies. New York: The Free Press. Williamson, Oliver E. 1980. “The
Organization of Work: A Comparative Institutional Assessment,” Journal
of Economic Behavior and Organization, 1(1): 5-38.
Lane (1973, esp. pp. 162-65). Production by small “firms” inside a
larger facility remained an important organizational form in manufacturing into
this century. A famous example is the Winchester Repeating Arms Company which
operated in this manner until the outbreak of World War I. Compare Buttrick
Mantoux (1962, pp. 33-36). Mantoux was not willing to count the royal
manufactories sponsored by Colbert in France as forerunners of the industrial
factory system, mainly because they required royal subsidies or patronage for
their continued existence.
(1934). Nef also discusses large plants, such as cannon foundries, in various
is, the individual weaver might own his own loom, for instance. The jobber
would own the working capital (the materials).
 It was
tried. “In the Coventry silk weaving industry the experiment of ‘cottage
factories’ was tried. In the centre of a square surrounded by rows of cottages,
an engine-house was built and the engine connected by shafts with the looms in
the cottages. In all cases the power was hired at so much per loom. The rent
was payable weekly, whether the looms worked or not. Each cottage held from 2
to 6 looms; some belonged to the weaver, some were bought on credit, some were
hired. The struggle between these cottage factories and the factory proper
lasted over 12 years. It ended with the complete ruin of the 300 cottage-
factories.” Marx (1906, p. 503). Marx mentions other examples “in
some of the Birmingham trades.”
also p.86: “The greater their number, the more they naturally divide
themselves into different classes and subdivisions of employment. More heads
are occupied in inventing the most proper machinery for executing the work of
each, and it is, therefore, more likely to be invented.” And, of course,
the opening paragraph itself: “The greatest improvement in the productive
powers of labour, and the greater part of the skill, dexterity, and judgment
with which it is any where directed, or applied, seem to have been the effects
of the division of labour.” (Smith, 1776 [1937, pp. 86 and 1].)
1906, Part IV, Chs. XIV and XV, pp. 368-556. This is, of course, a far more
extensive treatment than we find in Smith. It is far superior to that of J.S.
Mill, who had little of any interest to add to Smith. Cf. Principles,
Book I, Chs. VIII and IX:1 (Mill, 1964, pp. 116-136.) It is worth noting,
however, that Mill too shared the opinion of Smith and Marx that the advantages
of division of labor had precedence over “the introduction of processes
requiring expensive machinery” among the “causes of large
manufactories.” (pp. 132, 135.)
 Kenneth Sokoloff’s study of a large 1832 sample of
manufacturing firms in the U.S. northeast finds that “the evidence serves
to undercut the notion that the early period of industrialization was based on
a proliferation of new, machinery- intensive technologies.” (Sokoloff,
distinction between “manufacturing” and “factory
production” is a perfectly good and useful one. It is omitted here so as
not to burden the paper with too much terminological baggage.
recalls his calculation: “Those ten persons, therefore, could make among
them upwards of forty-eight thousand pins in a day.” Etc. Marx checked on
pin- making in his own day: “… a single needle- machine makes 145,000 in
a working day of 11 hours. One woman or one girl superintends four such
machines and so produces near upon 600,000 needles in a day ….” (Marx,
1906, p. 502.) The most recent report is Pratten (1980): Today, one operative
supervising 24 machines, each of which turns out 500 pins per minute, will make
about 6 million pins in a day.
Georgescu-Roegen especially stresses the failure of neoclassical production
theory to illuminate the fundamental difference between manufacturing processes
and agricultural production processes where nature dictates the time-phasing of
operations. See Georgescu-Roegen (1972), which is reprinted (with several other
essays germane to our subject) in Georgescu-Roegen (1976).
has mustered impressive evidence on the efficiency advantages of small,
non-mechanized factories over craftshops in the early industrialization of the
American northeast. His estimates of total factor productivity show
“factories” with more than five employees to be over 20 percent more
productive than artisanal shops. Cf. Sokoloff (1984, sections III and IV).
will sometimes challenge that impression. Swedish economists will recall the
“Horndal effect” (so named by Erik Lundberg). Horndal was a steelmill
considered outdated by its controlling corporation, which intended to
concentrate production in its more modern plants. Investment in Horndal was
therefore stopped altogether. The expectation, of course, was that in a couple
of years the mill would not cover variable costs. To the consternation of
observers, however, the rate of productivity growth in Horndal kept pace with
that of the rest of the industry for many years. (Lundberg, 1959, pp. 663-64.)
 It was
in fact normal for each craftsman (guild member) to own the tools he was using.
John Rae (as quoted by Mill, 1964, p. 129): “If any man had all the tools
which many different occupations require, at least three-fourths of them would
constantly be idle and useless…” etc.
found that priority for this statement of the advantages of division of labor
belonged to Gioja (1815).
Arrow (1979, p. 156): “This dilemma has been thoroughly discussed: it has
not been thoroughly resolved.” But, surely, there is no genuine dilemma —
just our obstinate collective refusal to draw the obvious conclusion and allow
the empirical reality of increasing returns to displace the convenient
construct of “perfect competition.”
 Cf. esp.,
and Sokoloff (1984) find that, in the first half of the 19th century, even
quite small factories (with five or more employees) were giving a greater share
of jobs to women and children than did artisanal shops.
data suggest that, for non-mechanized factories deriving their competitive
advantage solely from the division of labor, economies of scale would tend to
be very nearly exhausted already in the size-range of 6-15 employees and
totally exhausted at 20. (For the already mechanized textile industries the
scale economies were much stronger and remained significant up to a far larger
scale). (Sokoloff, 1984, section III.)
by Alchian and Demsetz (1972).
 In the
literature on vertical integration this is familiar as the post-contractual
“opportunistic behavior” of Williamson (1975) or the
“hold-up” problem of Klein, Crawford, and Alchian (1978).
speaking, the core is empty since every distribution can be blocked. (It does
not seem helpful to insist that the empty core is a “transaction
cost” problem.) I am especially grateful to Dan Friedman for clarifying
the structure of the bargaining situation for me.
the integration should be vertical does not seem to be necessary in general. In
Dahlman’s theory of the open field system, avoidance of the “hold-up”
problem is the explanation for why the scattering of strips was maintained over
the centuries. With his arable strips scattered, the individual farmer could
not, in some dispute over communal production or distribution issues, threaten
to withdraw and thereby to reduce the benefits of scale economies to the
village as a whole). (Dahlman, 1980, pp. 120-30 and 135-38.)
attempt in this direction is made in Ippolito (1977).
cost curves, which have managed to survive (at least in undergraduate teaching)
in uneasy co-existence with neo- Walrasian theory, have a rather natural fit to
the Smithian technology. In neoclassical production theory, we cannot be sure
that there are any “firms” to talk about. With the Smithian theory,
we at least have no doubts about their existence. Marshall tended to presume
long-run decreasing cost for his firms; this property follows directly from the
increasing returns of Smithian technology. Marshall’s short-run U-shaped
average cost schedule gets its downward-sloping segment by the same argument as
used above in connection with Okun’s Law and its upward-sloping segment, quite
conventionally, from the diminishing marginal product of variable factors when
fixed factors are kept fixed. Pricing in the markets supplied by these firms,
however, should be analyzed in Hicksian, rather than Marshallian terms. We
should expect them to be “fix-price” rather than
 I very
much agree, therefore, with Martin Weitzman that the prevalence of these
increasing returns technologies must be taken into account if one is to
understand the situation of manufacturing workers in a recession. Unemployment
theory, Weitzman argues, must as a first logical requirement explain why
unemployed factor units do not set up in production on their own. In the
Smithian division of labor case the answer is straightforward: the
manufacturing worker simply does not have the skills and knowledge required to
make the product as an artisan. (Cf. Weitzman, 1982.)
example of an important idea which has not found a home in neoclassical theory
but would fit into a Smithian production theory is Erik Dahmen’s
“development block.” In a growing economy, all the component sectors
of a Dahmen block have to be completed before any one of them becomes
economically viable. (Cf. Dahmen, 1971.)
worthy, of course, expects countries with similar factor endowments to export
similar products to each other and would not be surprised is trade of this
description would reach large volume.