Further Thoughts on the Use of Machines

1. the cutthroat knife, a fuel's paradox and fantastical credit

Marx, that wily trickster, shanghaied Bill Sikes, the unkempt, scowling, growling,

murderous, dog-abusing Dickens villain from Oliver Twist, and contrived for him the

following mock-ingenuous plea to an imaginary jury:

…no doubt the throat of this traveling salesman has been cut. But that is not my

fault; it is the fault of the knife! Must we, for such a temporary inconvenience,

abolish the use of the knife? Only consider! Where would agriculture and trade be

without the knife? Is it not as beneficial in surgery as it is in anatomy? And in

addition a willing help at the festive table? If you abolish the knife—you hurl us

back into the depths of barbarism.

The occasion for the ruffian's cameo appearance was a section in Das Kapital dealing with

what Marx labeled "the theory of compensation as regards the workpeople displaced by

machinery" – a topic that had been dear to the hearts of political economists since at least

the anti-machinery riots of 1779 and which remains an article of faith among contemporary

economists in spite of Herr Marx's sarcasm. As University of California economics

professor, Carl Walsh explained, "there is little debate among economists about the long-

run effect of productivity on employment. [...] In the long run, faster productivity growth

should translate into an increase in the overall demand for labor in the economy."

That "there is little debate" may itself be debatable. But what debate there is has a peculiar

configuration. An odd twist was added by Stanley Jevons in 1865 when he requisitioned

the said theory of compensation to answer a question about the supply and demand for

coal. In The Coal Question, Jevons exclaimed, "It is wholly a confusion of ideas to suppose

that the economical use of fuel is equivalent to a diminished consumption. The very

contrary is the truth [emphasis in original]." He went on to explain:

As a rule, new modes of economy will lead to an increase of consumption

according to a principle recognised in many parallel instances. The economy of

labour effected by the introduction of new machinery throws labourers out of

employment for the moment. But such is the increased demand for the cheapened

products, that eventually the sphere of employment is greatly widened. Often the

very labourers whose labour is saved find their more efficient labour more

demanded than before.

The difference between the well-trodden compensation theory and the paradox outlined by

Jevons is that the former is seen as a blessing by its proponents, while the latter is

supposed to be a curse. As is their wont, optimistic economists are inclined to seek out

evidence supporting the first proposition or discounting the second, while pessimists do the

opposite. Confirmation bias ensures that the search for such evidence often succeeds, thus

controversy persists. But is the post hoc increased demand necessarily propter hoc the

cheapened price? It is if we assume the expansion of trade beyond some current,

artificially-fixed barriers, as did Dorning Rasbotham in his 1780 response to the 1779

machinery riots in Lancashire:

There is not a precise limited quantity of labour, beyond which there is no demand.

Trade is not hemmed in by great walls, beyond which it cannot go. By bringing our

goods cheaper and better to market, we open new markets, we get new customers,

we encrease the quantity of labour necessary to supply these, and thus we are

encouraged to push on, in hope of still new advantages. A cheap market will always

be full of customers.

Rasbotham envisioned opening new markets and getting new customers while Jevons did

not specify where the increase in consumption and increased demand would come from.

This latter raises the possibility of an increased demand coming from existing customers in

existing markets. Actually, the two situations are not all that different. After all, what is

"the expansion of trade," other than the expansion of the means for trading? The key to

increased demand, either from new markets or from the increased consumption of existing

customers is purchasing power. 2

Cheap prices may indeed increase the demand for a product but, as Charles D'Avenant

pointed out in 1699, increased demand doesn't always translate into larger total revenue

from sale of that product. A bumper crop in wheat, for example, may lead to a fall in

revenue for farmers and a poor harvest may result in an increase. Furthermore, increased

demand for commodities is not the same as increased demand for labor. So cheapness of

the products is not a sufficient cause for a compensation effect or "rebound," especially

when it is assumed that "the introduction of new machinery throws labourers out of

employment for the moment." Even if their distress is only temporary, for the moment

those unemployed workers have diminished means for purchasing a portion of the

expanded quantity of consumer goods that machinery makes possible.

In a closed system, the loss of income by the displaced workers would lead to a decrease in

demand, not an increase, regardless of the cheapness of products. This decrease of income

– and consequently of demand – would not be made up by the incomes of newly employed

workers constructing the machines. If it did, then the end products would not be cheaper

and we would be back in the same predicament of insufficient means for purchasing the

expanded production. The increased demand thus can only come from outside the closed

circuit of employment and income. But from where?

The answer to that riddle is yet another riddle: credit. D'Avenant described credit as

"fantastical":

Of all Beings that have Existence only in the Minds of Men, nothing is more

fantastical and nice than Credit; 'tis never to be forc'd; it hangs upon Opinion; it

depends upon our Passion of Hope and Fear; it comes many times unsought for,

and often goes away without Reason; and when once lost, is hardly to be quite

recover'd.

But, take heart! This fantastical being resolves the other two paradoxes – of labor displaced

by machinery and the economy of fuel, respectively – and sets the stage for a grand

resolution of the triple paradox, all tied up in a neat bundle!

3

In a 1934 BBC radio address, John Maynard Keynes outlined his objections to the notion,

prevailing among "almost the whole body of organized economic thinking and doctrine of

the last hundred years" that "the existing economic system is in the long run self-adjusting

[…]" Keynes must not have read Marx's section on the theory of compensation because he

attributes to Marxism, the fervent acceptance of the "essential elements" of the orthodox

faith in self-adjustment, with the proviso that Marxists make the "highly plausible

inference" from the doctrine that "capitalistic individualism cannot possibly work in

practice."

What Marx wrote about the theory of compensation – his label for what Keynes called

self-adjustment – however, was this:

The labourers that are thrown out of work in any branch of industry, can no doubt

seek for employment in some other branch. If they find it, and thus renew the

bond between them and the means of subsistence, this takes place only by the

intermediary of a new and additional capital that is seeking investment; not at all

by the intermediary of the capital that formerly employed them and was afterward

converted into machinery [emphasis added].

For his part, Keynes attributed the "fatal flaw" in the orthodox self-adjustment doctrine to

its failure to "develop a satisfactory and realistic theory of the rate of interest":

Now the school that believes in self-adjustment is, in fact, assuming that the rate

of interest adjusts itself more or less automatically, so as to encourage just the

right amount of production of capital goods to keep our incomes at the maximum

level that our energies and our organization and our knowledge of how to produce

efficiently are capable of providing. This is, however, pure assumption. There is

no theoretical reason for believing it to be true [emphasis added].

The difference between Marx's argument and Keynes's is one of perspective. Where Marx

focused on the quantity of "new and additional capital that is seeking investment," Keynes

considered the rate of interest required to encourage the "right amount of production of

4

capital goods." Marx thus presented the case from the perspective of the creditors while

Keynes assumed the point of view of the entrepreneur taking on debt. In both cases, they

targeted the easing of credit, not the cheapening of prices, in order to stimulate new

demand.

In his radio address, Keynes went further, stipulating that when interest rates had been low

enough for a sufficiently long time so as to indicate "there is no further capital construction

worth doing even at that low rate" then "drastic social changes" would become necessary

to increase consumption. "The full employment policy by means of investment," he wrote

to T.S. Eliot, a decade later, "is only one particular application of an intellectual theorem.

You can produce the result just as well by consuming more or working less. Personally I

regard the investment policy as first aid. […] Less work is the ultimate solution."

It is with regard to this projected "ultimate solution" that Keynes's comment about

Marxism inferring "capitalist individualism cannot possibly work in practice" may have

some validity. In the section of the Grundrisse that has come to be known as the "fragment

on machines," Marx highlighted the tendency of capitalism "to reduce labour time to a

minimum, while it posits labour time, on the other side, as sole measure and source of

wealth." This tension he called the "moving contradiction" of capital, a developmental

process that appears to capital as "mere means" of expanded accumulation but in fact

contains "the material conditions to blow this foundation sky-high" because if capital

becomes too successful in its drive to create disposable time "then it suffers from surplus

production, and then necessary labour is interrupted, because no surplus labour can be

realized." So instead of working less, as Keynes envisioned the ultimate solution to

unemployment, it becomes increasingly necessary, under capitalism, for people to work

superfluously.

To recap and sum up the argument thus far: first, it is not cheap prices but easy credit that

stimulates new demand, whether for labor or for the consumption of fuel and other natural

resources. Second, it is not unemployment per se that is the scourge of labor but

specifically unemployment in the face of an increasingly superfluous expenditure of

working time. That's the bad news. The good news is that if it becomes possible to 5

conceive of a great deal of current work as superfluous, then the energy and other natural

resource consumption directly associated with that labor could also be reduced

substantially without a diminution in standards of living.

If that good news story seems a bit of a stretch, look at it this way: formerly we were

confronted with two paradoxes, based on the same principle, whose consequences were

diametrically opposed in terms of desirability. The compensatory employment of labor

displaced by machines was seen as a blessing while the rebound effect from the economy

of fuel was viewed as a curse. From that perspective, solving the problem of

unemployment could only make the problems of resource consumption and environmental

impact worse and vice versa.

Reframing unemployment as primarily a problem of superfluous employment has the

advantage of aligning the environmental and employment dilemmas so that the solution to

one is also the solution to the other. This is not to say that the two problems have somehow

magically been solved or even that the solution is easy – only that a solution has now

become conceivable. It is the knife-wielder, not the knife, which is on trial!

Actual solutions, as opposed to conceptual ones, require a firmer grounding in realism. By

realism, I don't mean resignation or compromise with an unreceptive status quo but

reference to historical, empirical evidence regarding concrete policies and policy

frameworks. There are currently two dominant, competing policy frameworks for

addressing social costs: a market-based approach, exemplified in carbon emissions trading

schemes, and Pigouvian taxation and subsidies, which seek to correct pricing by charging

for the social costs of negative externalities or compensating for the social benefits of

positive ones. The analysis in the next section (forthcoming in Marshall Studies Bulletin)

demonstrates that besides working less being "merely" the ultimate solution to the problem

of unemployment, the analysis of working time provides a fundamental key to

understanding the underlying dynamics of social costs and environmental externalities in

general.

6

2. The hours of labour and the problem of social cost

In "The Problem of Social Cost," Ronald Coase (1960) examined one variety of presumed

market failures – outcomes that Cecil Pigou (1952) had described as “incidental uncharged

disservices” (or uncompensated services) but are now commonly referred to as

"externalities." The incidental quality of these effects makes them a social cost. The

economic analysis Coase challenged and the standard examples he re-examined were taken

from Pigou's discussion in part II of The Economics of Welfare. Coase argued that the

suggested courses of action in the Pigovian tradition – liability, taxation or regulation –

were inappropriate and often undesirable.

Coase claimed that the traditional approach to the problem of social cost "tended to

obscure the nature of the choice that has to be made" (1960, 2). He characterized the

question posed by the approach as "one in which A inflicts harm on B and what has to be

decided is: how should we restrain A?" He objected that the problem was really a

reciprocal one and the real question should be "should A be allowed to harm B or should B

be allowed to harm A? The problem is to avoid the more serious harm."

However, Coase didn't consider the full range of Pigou's examples and analysis. While

Coase’s restatement of the problem may have been appropriate to the specific externality

problems discussed by Pigou in part II, it entirely overlooked the radically different labour-

market problem encountered in part III, in which competitive pressure compels an

employing firm to inflict harm on both itself and its employees and thus regulatory

restraint of the firm (and competing employers) may benefit both.

Along with the majority of the preceding Pigovian tradition, Coase evaded the thorny

questions of working conditions and unemployment. Whatever gains in tractability may be

accomplished by such a maneuver are more than offset by a forfeit of realism and of

insight into the complex interdependency of economic factors in the long period. The

determination of the hours of work provides a particularly compelling example of a

circumstance in which mutual benefit could result from an imposed non-market restraint.

7

In part III of Economics of Welfare, Pigou argued that "after a point, an addition to the

hours of labour normally worked in any industry would, by wearing out the work people,

ultimately lessen rather than increase the national dividend" (1952, 462). That observation

derives from the theoretical exposition performed by another of Alfred Marshall's star

pupils, Sydney J. Chapman. Chapman's theory of the hours of labour (1909) and his

historical study of the Lancashire cotton industry (1904) that foreshadowed it offer a

suggestive counter-example of the largely unrealized potential of Marshall's industrial

economics. This paper argues that Chapman's analysis provides greater insight into the

problem of social cost than does either Coase's or Pigou's.

As Chapman demonstrated, under competitive conditions, employers would tend to prefer

hours of work that exceed the length that would be optimal for output. If an individual

employer and workers were able to negotiate more optimal hours of work, it would involve

a present investment by the employer in the workers future productivity. Well-defined

property rights to that future capacity could not be transferred to the employer and thus the

arrangement could be upset by a future offer of higher wages from a competing employer.

If we assume an optimal length of working day of eight hours for a given technology,

during which an average worker could produce nine units of output but a longer actual

working day of ten hours, during which the same worker produces only eight units, then

Table 1, below, illustrates in simplified fashion the dilemma confronting the progressive

employer seeking to reduce the hours to the optimal level.

Table 1

Month 0 1 2 3 4 5… 12

Units of output 6.4 7.05 7.7 8.35 9 9

Cost per unit 15.63 14.18 12.99 11.98 11.11 12.29

Value of daily output 80.00 88.13 96.25 104.38 112.50 112.50

Daily pay 100.00 100.00 100.00 100.00 100.00 110.62

Difference (per day) -20.00 -11.88 -3.75 4.38 12.50 1.88

8

At an hourly wage of $10, the worker would receive a daily income of $100, which would

translate to a labor cost of $12.50 for each unit of output. Now, assuming that the output on

the first day of the change would drop in proportion to the reduction in hours, that it would

take four months after a reduction in hours for the productivity gains to be fully realized,

and that the subsequent recovery and improvement in productivity would follow a straight-

line path, an employer who continued to pay the old daily rate (to retain the workers)

would incur a higher labor cost per unit of output during approximately the first two and

half months of operation at the new schedule.

During that period, the employer would have invested a total of $557 per worker, an

average of around $10 a day, in anticipation of future productivity gains. Due to the higher

total output, the employer could eventually grant a wage increase but would need to retain

a portion of the revenue from that increased output to recoup expenses from those first two

and a half months. However, a competing employer, who had not invested the initial $557,

could hire away the now well-rested workers with a nearly 2 per cent larger wage

premium. In the real world, where employers don't have perfect knowledge, there would be

even greater uncertainty regarding the amount of the potential productivity gain and the

time and expense it would take to achieve it. On the positive side, the workers might not be

so eager to change jobs just for a wage premium.

Following Chapman, Pigou viewed market failure with respect to the hours of work as

commonplace, observing that, "the evidence is fairly conclusive that hours of labour in

excess of what the best interests of the national dividend require have often in fact been

worked" (Pigou 1952, 465). The chapter on the hours of labour is one of two places in The

Economics of Welfare where Pigou specifically called attention to the divergence between

private net dividend and "the best interest of the national dividend" and consequently

where there is "a prima facie case for public intervention" (p. 331).

Pigou's "Hours of Labour" and Chapman's

Pigou's analysis of the hours of labour in The Economics of Welfare closely followed five

main points of the theory Chapman had presented in 1909 in his presidential address to the

9

Economics and Statistics section of the British Association for the Advancement of

Science, subsequently published in the Economic Journal as "Hours of Labour." In his

analysis, Chapman had referred to a mass of evidence from the 19th century indicating that

reductions in the hours of work had not led to proportionate declines in output and, instead,

had often led to increases. The reduction of hours allowed better rested workers to produce

as much or more in shorter hours. Pigou inferred from the same evidence "that hours of

labour in excess of what the best interests of the national dividend require have often in

fact been worked” (465) In part, this is because, "after a point, an addition to the hours of

labour normally worked in any industry would, by wearing out the workpeople, ultimately

lessen, rather than increase, the national dividend" (462), but also because competition

(along with their own “short-sightedness”) would tend to compel employers to exceed that

point at which additional work diminishes output over the long run.

Both economists referred to the several complicating factors but arrived at the same

conclusion regarding a hypothetical optimal length of working day. For Pigou, the

"essential point" was that "in each several industry, for each class of workers there is some

length of working day the overstepping of which will be disadvantageous to the national

dividend" (464). Similarly, Chapman had concluded that beyond a certain point, each

additional hour of work would contribute to the current day's total output but at the

expense of the following (and subsequent) day's capacity for effort. The intensity of the

work involved, along with the average characteristics of the individual workers, would

dictate the point at which cumulative output would begin to decline and thus the length of

the optimal working day.

The historical evidence also contradicts a standard assumption that self-interest will lead

employers and employees to pursue an optimal working day, from each of their

perspectives and to negotiate a compromise. Chapman's analysis explained why

competition would tend to produce excessively long days. Workers would choose a day

longer than was prudent for their welfare because the prospect of unemployment would

cause them to give higher consideration to immediate earnings than to their long-term

earning capacity. Similarly, because well-rested workers could be lured away by an offer

10

of higher wages from another firm, an employer could never be certain of benefiting from

the short-term restraint that maintaining an optimal workweek would require.

Pigou explained the presumed market failure as follows: "workpeople, in considering for

what hours per day they will consent to work, often fail to take account of the damage

unduly long hours may do to their efficiency" (466). In the case of employers, they "also

often fail to realise that shorter hours would promote efficiency among their workpeople,

and so would redound to their own interest." Furthermore, "except in firms which possess a

practical monopoly in some department of industry, and so expect to retain the same hands

permanently, the lack of durable connection between individual employers and their

workpeople makes it to the employers' interest to work longer hours than are in the long

run to the interest of production as a whole."

The Marshallian Realism of Chapman's The Lancashire Cotton Industry

Chapman's analysis of the hours of work was acknowledged as both novel and canonical

(Hicks 1932, Marshall 1920, Robbins 1929). Pigou's presentation can thus best be regarded

as an accurate paraphrase of that theory that was lax in fully crediting its source. That

characterization is more than a pedantic quibble over originality and citation because of

crucial differences in methodology between Chapman and Pigou. Chapman's theory

evolved out of what Marshall called his "realistic-impressionist" scholarship on the

Lancashire cotton industry – a method of inquiry that Marshall upheld as more suitable to

the subject matter than the abstract, "statical" method employed by Pigou.

The Lancashire Cotton Industry (Chapman 1904) was a self-conscious application of

Marshall’s theory of industrial localization. According to Raffaelli (2004), it was one of

two "very promising steps towards the establishment of a Marshallian school of industrial

economics […]" (211). Chapman's later theory of the hours of labour, published in 1909,

retains distinctive marks of influence from his "realistic-impressionist" study of the

Lancashire cotton industry. Recapping his hours of labour analysis, Chapman (1911)

specified that the ultimate efficiency gains from a shorter working day were "long-period

11

results, which may not be fully realized, and […] are apt to be overlooked by everybody"

(342).

In the preface to The Lancashire Cotton Industry, Chapman explained that the "different

guiding notions" of employers and employed were a "striking feature in the history of the

Cotton industry […]" (p. ii). Some of the most compelling of those notions had to do with

the hours of work and the agitation for shorter hours. Instead of relying on an abstract

analysis of the economics, Chapman scrupulously investigated the workers' and employers'

own view of the question. Chapman had been exposed to the pamphlet literature of the

early English socialist writers, many of whom were involved in labour struggles in

Lancashire, through his studies at Cambridge with Herbert Foxwell. An extensive select

bibliography in the Lancashire study is reminiscent of Foxwell's (1899) bibliography for

The Right to the Whole Produce of Labour, including the apologetic tone of the

introductory remarks. Chapman's astute observations on the views of the advocates and

opponents of the Ten Hours Bill foreshadowed his later theoretical treatment of the hours

of labour:

Sound as were the fundamental ideas for the realization of which the Society for

National Regeneration had been instituted, its propaganda were frequently vitiated

by appeals drawn from the doctrine of the labour fund, as the "lump of labour"

fallacy might be called […]. We must notice, however, that those who advocated

shorter hours, both in this period and later, found also many sound reasons for

their action in the expected effect on the health and comfort of the operatives.

They perceived that high wages were of little value to those who had little time to

spend them. Moreover, the mistakes made by the operatives lay not so much in

their fundamental opinions as in some of the reasons given by them for holding

these opinions (98).

Looking back at that assessment from the perspective of his 1911 recap of his 1909 theory

of the hours of labour, one might conclude that what Chapman perceived as the

"fundamental ideas" of the advocates of shorter hours hearkened to the long-period results

of the measures rather than the propagandistic appeals to immediate effects. In "Hours of 12

Labour," Chapman presented a similar contrast between the guiding "ideals of life" and

possibly fallacious views about "the mechanics of distribution":

It would seem from the records of labour movements as if the operative's fear –

based as much on ignorance as on distrust – lest the longer day should mean no

more pay, though the weekly product would be greater, has protected him against

the injurious consequences of short-sightedness; but I am inclined to think that the

dominant force in these labour movements has consisted in ideals of life, formed

half instinctively, which are unconnected with views, fallacious or otherwise,

concerning the mechanics of distribution. Bad arguments have been used to

justify good ends (365).

Marshall wrote to Chapman, praising his book as "the best monograph of the kind that has

ever been published. It is both a realistic-impressionist study of human life and an

economic treatise" (Whitaker 1996, 93). By contrast, the comments Marshall left in the

margins of his personal copy of Pigou's Wealth and Welfare (1912) express unease with

Pigou's overestimation of "the possibilities of the statical method" (Bharadwaj 1972, 33).

In his Industry and Trade, Marshall (1919) again displayed diffidence about the broader

applicability of the mathematical analysis inherent in the "brilliant work of Edgeworth and

Pigou […]" (605). Marshall cited Chapman's Lancashire study with approval three times in

Industry and Trade, along with two other works by Chapman while only mentioning Pigou

twice, both times in connection with Edgeworth and without citing any specific texts.

For Marshall, realism was not simply a matter of relaxing the constraints of simplifying

assumptions that had been imposed on an abstract hypothesis. According to Chapman,

Marshall viewed theory and realism as "two lines of investigation" that converged. The

evolution of actual economic practices was not something that could be deduced from

abstract principles. It needed to be documented through historical investigation and the

collection of facts. Chapman's study of the Lancashire cotton industry strove for just such a

convergence of realism and hypothesis.

13

In "Distribution and Exchange," Marshall (1898) outlined what he saw as the limits to

mathematical, abstract analysis. In that article he explained that although the mechanical

analogy with its ceteris paribus assumptions may indeed be suitable for the short period, it

is entirely unsuitable for the long-period analysis in which the cumulative effects of

external economies predominate: "If we include in our account nearly all the conditions of

real life, the problem is too heavy to be handled; if we select a few, then long-drawn-out

and subtle reasonings with regard to them become scientific toys rather than engines for

practical work" (52). He argued that it is even more important to know when to quit an

analogy than when to introduce one because they can become an obstacle to judgment in

the long period. In place of mechanical analogies, Marshall prescribed "biological"

analogies for the investigation of long-period phenomena.1

Marshall's notion of external economies – whose cumulative change in the long period

ruled out the broader applicability of mechanical analogies – evolved into incidental

uncompensated disservices or uncharged services and eventually became abbreviated as

negative and positive externalities.2 Those external economies, though, could hardly be

bundled into packages of well-defined property rights and traded on the futures market like

any other commodity. Instead, even though they may be negligible for the purposes of

short period analysis, they are integral to the evolution of industrial organization.

In 1932 J. R. Hicks posited a condition for sidestepping, in analytical technique if not in

basic theory, the type of market failure indicated in Chapman's theory and reiterated in part

III of Pigou's Economics of Welfare. Hicks conjectured that a "very modest degree of

rationality on the part of employers will thus lead them to reduce hours to the output

optimum as soon as Trade Unionism has to be reckoned with at all seriously […]" (217-

18). Hicks thus introduced an analytical simplification that was actually more of a

1 Exactly what Marshall meant by biological analogies, other than a more inductive approach, is beyond the

scope of this paper.

2 "An external economy or diseconomy of production arises when the production of a commodity gives rise

to incidental uncharged services or uncompensated disservices to a third party who is neither producer nor

consumer" (Knox 1960).14

complication – and an institutionally contingent one at that. Moreover, he did it in the

name of "think[ing] back our arguments into a more cumbrous but more realistic form

[…]" (93). Hicks's notion of realism was strikingly at odds with Chapman's or Marshall's.

Restoring the Elided Long-Period View

A narrative ellipsis haunts discussion of the problem of social cost and contemporary

labour economics in a way that few acknowledged sources could hope to. What I mean by

"ellipsis" is not simply an absence of influence but an odd sort of semi-presence that leaves

out precisely the most salient details. Pigou left out explicit credit to Chapman for the

theory of the hours of labour. The Pigouvian tradition, including Coase's critique of that

tradition, disregarded that key part of Pigou's welfare economics that relied on Chapman's

theory. The elided influence of Chapman extends also to Hicks's procedure for evading the

serious complication introduced by Chapman's theory. Contemporary economic analysis

proceeds as if the given hours of work are optimal for output, an assumption that can be

traced directly to Hicks (Nyland 1989, Walker 2007b).

A third instance of ellipsis occurs precisely at the intersection of social cost and labour

economics, in what John Maurice Clark termed the "social overhead cost of labour."

Stabile (1995, 1996) has highlighted the affinities between Studies in the Economics of

Overhead Costs (Clark 1923) and The Economics of Welfare, paying particular attention to

Clark's analysis of the shifting of the fixed costs of labour, which explicitly incorporates

Pigou's observations (in part derived from Chapman) of the effects on future efficiency of

unemployment and poor working conditions. Nearly 40 years later, Walter Oi (1962) took

up the theme of "the treatment of labor as a quasi-fixed factor," a concept Oi attributed to

"J. M. Clark, who dealt primarily with the social cost of unemployment" (554). Following

Oi, the notion of fixed costs underwent a remarkable inversion. Instead of referring to the

cost of sustaining each worker, regardless of whether employed or not, it has become an

employment cost (e.g., "fringe benefits") that doesn't vary with hours worked. The aspects

of cost shifting and of social cost have been omitted. Instead, the existence of these fixed,

per employee costs has become a stock rationale for why reducing the hours of work, by

15

"increasing the overtime premium does not appear to be an effective method of decreasing

unemployment" (Ehrenberg 1971, 206).

In each case – social cost, optimal hours of work and fixed costs – the contemporary

version has left out the core substance of Chapman's analysis. It is precisely the long-

period, cumulative results of external economies that are ignored. These are not theoretical

advances but detours around theory, bolstered by vague impressions that predictions based

on the truncated models have been empirically verified. But these "empirical proofs" may

consist of nothing more substantive than speculative assertions supplementing empirical

analysis rather than the results of the analysis itself. In the case of Ehrenberg conclusion

cited above, for example, his sentence began with the disclaimer, "Our own personal view

[…]"!

"In economics," wrote Paul Samuelson (1951, cited in Boyer and Smith 2001, 207) "it

takes a theory to kill a theory; facts can only dent the theorist's hide." Milton Friedman

(1953, cited in Boyer and Smith 2001, 207) further argued that it wasn't the realism of a

theory's assumptions that mattered but the quality of its predictions. But in the absence of

faithful attention to the history of economic thought, who is to say what the theory actually

said, what in fact it predicted and whether the empirical analysis confirmed the prediction?

3. Unpacking the decoupling tautology: a statistical science

fiction

When University of Arizona Sociology Professor Lane Kenworthy asked, "Is Decoupling

Real?", he was referring to the gap between median family income and per capita GDP in

the U.S. Along much the same lines is the chart below produced by the Economic Policy

Institute's State of Working America, which compares median family income and

productivity growth.

16

The Sandwichman is concerned with another kind of decoupling -- the much touted

decoupling of energy consumption from GDP growth that technological optimists like

Amory Lovins promote as the solution to environmental impact and resource exhaustion

problems. I've got another question: "is decoupling a tautology?" Relative decoupling of

energy consumption per dollar of GDP is a well established fact. What is in dispute is

whether that can be translated into absolute decoupling through imminent technological

breakthroughs.

The short answer is: it can't. The slightly longer answer is it can't because even the relative

decoupling that has occurred over the last 39 years is questionable. Oh, there's no doubt

that energy consumption per dollar of GDP has fallen; what's questionable is the

composition and distribution of that growing GDP.

Even at the aggregate level, there is the question of the increasing proportion of economic

activity that needs to be devoted to repairing the damage done by previous economic

activity -- cleaning up toxic spills, recovering from extreme weather events, etc. This is

what Stefano Bartolini referred to as negative externalities growth and Roefie Hueting

17

called asymmetric entering. This is a kind of decoupling of GDP increase from any

meaningful notion of expanded or improved utility. It is just running faster on a treadmill.

But that's not all. There is also the question of distribution and even the possibility that the

growing gap between GDP growth and median incomes is a structural imperative. Now, by

"structural imperative" I don't mean something that can't be changed -- only something that

isn't going to be budged by moralistic pronouncements about fairness. To show what I

mean by structural imperative, I would like to share a composite chart that integrates the

data from the EPI productivity and median income chart above and data comparing the

energy intensity of GDP to the energy intensity per hour of work. I have added an inverted

productivity series (black dotted line) for reasons that will become clear as the narrative

unfolds.

The first thing that becomes clear from this chart is that productivity, median income and

the energy intensity per hour worked tracked each other closely from 1949 to 1973. The

latter year was chosen as the index year because energy intensity per hour of work peaked

in that year. After 1973, energy intensity per hour of work declined for about ten years and

then remained virtually flat up to the present. Productivity and Median Income diverge

after 1973.18

The inverted productivity series now presents a clue as to what exactly is being

"decoupled" in all this decoupling. With productivity defined as GDP per hour worked,

inverted productivity is hours worked per unit of GDP. Before 1973, inverted productivity

appears as simply the mirror image of productivity, median income and energy intensity

per hour worked. After 1973, though, it tracks energy intensity of GDP, while the stable

energy intensity of hours worked can be understood as the axis around which productivity

and energy intensity of GDP rotate and reflect one another. For all intents and purposes,

then, one could say that "energy intensity of GDP" is a statistical tautology, which itself,

prior to 1973, performed as the axis around which productivity growth translated into

steady gains in median income.

Correlation does not imply causation. Sometimes, however, it reveals a hidden tautology.

Hours, energy consumption, income and GDP are inputs and outputs of an economic

system that transforms some of those into some of these. The inputs don't cause the outputs

any more that cattle "cause" beef, they're just different names for the same thing in a

different state.

The indexes I have plotted in the graph are ratios between inputs and outputs (productivity

and energy intensity of GDP) or inputs and other inputs (hours of work and energy

consumption). Median family income can be thought of as a circular ratio of inputs and

outputs in which both income and the family can be viewed alternatively as either outputs

or inputs -- income provides sustenance for the family; the family supplies labor to

industry in return for income and so on.

Ratios between inputs do not wholly determine ratios between outputs or between inputs

and outputs. Policies do that. But the quantities of outputs are constrained by the quantities

of inputs. It is thus necessary when examining the energy intensity of GDP, for example, to

ask what is happening with the other inputs and the other outputs.

Next, I would like to present a third chart that compares the growth of population, labor

force, employment and hours worked in the U.S. from 1950 to 2009.

19

When I initially chose employment as the numerator of an alternative index, I was aware

that there was a rough coincidence with total population and thus would be similar to a per

capita energy consumption index but would smooth out some of the cyclical variations.

Aggregate hours of work presumably performs this smoothing function even more

precisely. For the last fifty years, though, the growth in employment, labor force and

aggregate hours has been steeper than population growth, with hours reaching a peak in

2000 of nearly 30% more per capita than in 1961. Employment as a percentage of total

population peaked in 2008, even though labor force participation peaked in 2000 because

the later ratio considers only the population 16 years and older.

Something is not adding up. More people are in the labor force as a percentage of

population. They are considerably more productive per hour and they are each working

about as many hours as they did 30 years ago. Yet family incomes are stagnant.

Time for a little statistical science fiction. What might have happened if Americans had

taken a larger fraction of the productivity gains over the last sixty years, say forty percent,

as reductions in working time? The number is not pulled from a hat. It is based on the ratio

between the increase in productivity over the last sixty years and a continuation of the

20

trend in average annual hours of work that prevailed from 1909 to 1961, with the latter

year being the year in which hours worked per capita reached their lowest point.

This what-if scenario makes a number of distinctions and assumptions. One of the key

assumptions, following from Chapman's theory, is that the average annual hours of work

will rarely fall below the number that would be optimal for productivity except in

circumstances of severe economic depression or under strict government regulation. The

assumption that the optimum number of hours will decline progressively as technology

improves is explicit in Chapman's theory. We also introduce a distinction between

productivity and productiveness, where the latter term excludes from its imagined

calculation all "superfluous" products. By superfluous, we have in mind defensive

expenditures, such as military spending and the cost of mitigating environmental damage

done by industrial activity or compensating for congestion. A great deal of litigation,

medical intervention, advertising and political campaign spending could be considered

superfluous. These expenditures do not add to the standard of living or quality of life.

Chart 4 compares the actual average annual hours worked per capita in the U. S. from 1909

to 2009 with a projection of the optimal hours per capita for productivity, based on the

above assumptions.

21

U. S. Annual Hours Worked Per Capita, 1909 - 2009

0

200

400

600

800

1000

1200

Year

Ho

urs

Actual

Optimal

After 1961, not only did average annual hours per capita not decline as productivity

increased, they also increased, from 697 in 1961 to a peak of 899 in 2000. This reflects the

increased participation of women in the labor force without a compensating reduction in

hours per worker. Recall, however, that the green line represents a working time that is

assumed to be optimal for output. That means that the excess hours actually worked

subtract from total output rather than adding to it. According to this projection, if we had

worked an average of 1100 hours a year in 2009 instead of the actual reported 1742, we

would have had more of the fruits of our industry at our disposal.

How much more? Here is where our estimates may begin to seem "fantastical." Let's say,

for the sake of argument, two and a half times as much? Alternatively, taking into account

"productiveness" (goods minus bads), we could work an average of 456 hours a year rather

than 1100 to achieve the same material prosperity that we currently eke out of 1742 hours.

If that sound preposterous, consider that measured productivity – that is, the inflation-

adjusted GDP divided by the total number of hours worked – increased 266% (about two

and a half times) between 1961 and 2009 even as hours worked per capita increased by

nearly 14 percent – and at their peak in 2000, by nearly 29 percent. Where did it all go?

22

With regard to energy consumption, I'm going to be a bit naughty here. Let's say that

instead of first declining and then leveling off after 1973, the energy intensity per hour

worked continued to increase at roughly the same rate as it had from 1949 to 1973 – or for

that matter from 1949 to 1961 – a little more two percent a year. Combining that increased

energy intensity with the radically-diminished average hours of work, 456 a year, would

yield a net savings in total annual energy consumption for 2009 of 30 percent!

Of course it is idealistic to assume that social costs or negative externalities can somehow

be reduced to zero, which is an implicit assumption of the above productiveness

calculation. The purpose of this statistical science fiction, though, was not to specify a final

destination but only to point out the direction in which we should be heading and to

estimate the magnitude of how far off track we are from where we might be if we weren't

traveling in exactly the wrong direction.

23

References

Bator, F. M. 1958. "The Anatomy of Market Failure." Quarterly Journal of Economics,

Vol. 72, No. 3, 351-379.

Bharadwaj, K. 1972. "Marshall on Pigou's Wealth and Welfare." Economica, New Series,

Vol. 39, No. 153, 32-46.

Boyer, G. R. and R. S. Smith. 2001. "The Development of the Neoclassical Tradition in

Labor Economics." Industrial and Labor Relations Review, Vol. 54, No. 2, 199-223

Chapman, S. J. nd. Unpublished Autobiography of Sir Sydney John Chapman. Coll Misc.

664, LSE Archives.

Chapman, S. J. 1909. "Hours of Labour." Economic Journal. Vol. 19 No. 75, 353-373.

Chapman, S. J. 1904. The Lancashire Cotton Industry: A Study in Economic. Development

Manchester: Manchester University Press, 1904.

Chapman, S. J. 1911. Outlines of Political Economy. London: Longmans, Green and Co.

Clark, J. M. 1923a. Studies in the Economics of Overhead Costs. Chicago: University of

Chicago Press.

Clark, J. M. 1923b. "Some Social Aspects: An Application of Overhead Cost to Social

Accounting, with Special Reference to the Business Cycle." American Economic

Review, Vol. 13, No. 1, Supplement, Papers and Proceedings of the Thirty-fifth

Annual Meeting of the American Economic Association, 50-59

Coase, R. H. 1960. "The Problem of Social Cost." Journal of Law and Economics. 3, 1-44.

24

Coase, R. H. 1972. "The Appointment of Pigou as Marshall's Successor." Journal of Law

and Economics. Vol. 15, No. 2.

Coase, R. H. 1975. "Marshall on Method." Journal of Law and Economics, Vol. 18, No. 1,

25-31.

Coats, A. W. "The Appointment of Pigou as Marshall's Successor: Comment" Journal of

Law and Economics 15 1972 487-95.

Daniels, G. W. and J. Jewkes. 1928. "The Post-War Depression in the Lancashire Cotton

Industry." Journal of the Royal Statistical Society, Vol. 91, No. 2, 153-206.

Ehrenberg, R. 1971. "The Impact of the Overtime Premium on Employment and Hours in

U.S. Industry." Western Economic Journal, 199- 207.

Foxwell, H. S. 1899 "Introduction" and "Bibliography" in Menger, A. 1899. The Right to

the Whole Produce of Labour. London: MacMillan.

Friedman, M. 1953. Essays in Positive Economics. Chicago: University of Chicago Press.

Goldberg, V. P. 1976. "Commons, Clark, and the Emerging Post-Coasian Law and

Economics." Journal of Economic Issues, Vol. 10, No. 4, 877-893.

Hicks, J. R. 1932. The Theory of Wages. London: MacMillan.

Kapp, K. W. 1950. The Social Costs of Private Enterprise. New York: Schocken Books

Keynes, J.M. 1924. "Alfred Marshall, 1842-1924." Economic Journal, Vol. 34, No. 135,

311-372.

Keynes, J.M. 1981. Collected Writings Vol. 19 Activities 1922 -1929: The Return to gold

and industrial policy, Cambridge: Macmillan & Cambridge University Press.

25

Knox, F. 1960. "The doctrine of consumers' sovereignty" Review of Social Economy Vol.

18, No. 2, 138-149.

Marchionatti, R. 1995. "Keynes and the Collapse of the British Cotton Industry in the

1920s: A Microeconomic Case against Laissez-Faire." Journal of Post Keynesian

Economics, Vol. 17, No. 3, 427-445.

Marchionatti, R. 2004. "What Don't Economists Know Now That Marshall Knew a

Century Ago?" Journal of Post-Keynesian Economics. Vol. 26, No. 3, 441-460

Marshall, A. 1898. "Distribution and Exchange." Economic Journal, Vol. 8, No. 29, 37-59

Marshall, A. 1919. Industry and Trade. London: MacMillan..

Marshall, A. 1920. Principles of Economics. (8th edition) London: MacMillan.

Medema, S. 2009. The Hesitant Hand: Taming Self-Interest in the History of Economic

Ideas. Princeton University Press, Princeton.

Menger, A. 1899. The Right to the Whole Produce of Labour. London: MacMillan.

Nyland, C. 1989. Reduced Worktime and the Management of Production. Cambridge:

Cambridge University Press.

Oi, W. 1962. "Labor as a Quasi-Fixed Factor." Journal of Political Economy. Vol. 70, No.

6, 538-555.

Pigou, A.C. 1952. Economics of Welfare. (reprint of 1928 third edition) London:

MacMillan and Co.

Raffaelli, T. 2004. "Whatever happened to Marshall’s industrial economics?" European

Journal of History of Economic Thought Vol. 11, No. 2, 209–229.

26

Robbins, L. 1929. The Economic Effects of Variations in the Hours of Labour." Economic

Journal. Vol. 39, No. 153, 25-40.

Samuelson, P. A. 1951a. "Economic Theory and Wages." In D. M. Wright, ed., The Impact

of the Union. Freeport, N.Y.: Books for Liberty Press, 312-42.

Stabile, D. R. 1993. "Accountants and the Price System: The Problem of Social Costs."

Journal of Economic Issues. Vol. 27, No. 1, 171-188.

Stabile, D. R. 1995. "Pigou's Influence on Clark: Work and Welfare." Journal of Economic

Issues, Vol. 29, No. 4, 1133-1145.

Stabile, D. R. 1996. "Pigou, Clark and modern economics: the quality of the workforce."

Cambridge Journal of Economics. Vol. 20, 277-288.

Walker, T. 2000. "The 'Lump-of-Labor' Case against Work-sharing." In L. Golden and D.

Figart, eds, Working Time: International trends, theory and policy perspectives.

London and New York: Routledge, 196-211.

Walker, T. 2007a. "Why Economists Dislike a Lump of Labor." Review of Social

Economy. Vol. 65, No. 3, 279-291.

Walker, T. 2007b. "Missing: The Strange Disappearance of S. J. Chapman's Theory of the

Hours of Labour." Conference paper. Available at SSRN:

http://ssrn.com/abstract=1125543.

Whitaker, J. K. 1996. The Correspondence of Alfred Marshall, Economist: Towards the

Close, 1903-1924 Cambridge: Cambridge University Press.

27