Friday, August 21, 2009

(Can't Stop) Endlessly Spouting Chapman II

by the Sandwichman

In 2001, the Government of Queensland further summarized Sandwichman's summary of Chapman's theory in its submission to the Australian Industrial Relations Commission's Reasonable Hours Test Case. At 400 words, it's the shortest comprehensive summary of Chapman's theory I know of.
5.2.1 Theoretical review

The study of the relationship between work intensity and fatigue owes much to S.J Chapman's theory of the hours of labour, where in 1909 Chapman demonstrated market failure in the determination of working time. This argument initially involves the establishment of a concept of 'optimal hours'. The main points of this argument can be summarised as follows:
  • a mass of evidence indicating that reductions in hours of work had not led to proportionate declines in output;

  • in modern industry fatigue was increasingly less physical in nature and more a combination of psychological and physiological as a result of specialization and increased need for mental concentration;

  • the reduction of hours allowed better-rested workers to produce as much or more in the shorter hours;

  • the total value of the output would initially rise as the working day increased but eventually the total output as well as the output per hour would decline as the working day became so long that it prevented adequate recovery from fatigue for workers;

  • this is the case because, beyond a certain point, each additional hour of work would be contributing to the output of the current day's total output but at the expense of the following day's output capacity; and

  • the intensity of the work involved would dictate the point at which total output begins to fall and thus the length of the 'optimal' working day.
The second half of this argument explores whether the free market can arrive at the 'optimal' length of day, and can be summarised as follows:
  • the maintenance of a long-term optimum by employers would require short-term restraint;

  • each individual employer could never be certain of reaping the benefit of their restraint as another firm could potentially entice the employer's well-rested workers away with a wage premium;

  • therefore the optimal output work time is a form of investment without equity;

  • simultaneously, Chapman assumed that workers would choose a longer working day than was prudent (although not as long as the working day preferred by employers), primarily because of a general short-sightedness that would mean workers would consider their immediate earning capacity more than their long term earning capacity; and

  • the outcome in a free market situation would therefore be one where employers and employees acting in self-interest would each tend to select a working day that was longer than the 'optimal' hours.

11 comments: said...


It is true that there is a negative correlation between real per capita income and working hours, which one would expect with rising productivity and value of leisure time as a result of market forces (not a result that we need to go all the way back to Chapman to understand), although laws limiting hours to 40 hour weeks in most jobs for most high income countries have probably played some role, these laws having come in some time ago in most of these countries.

There are two questions in my mind at this point. The first is, if it is the role of government in modern high income economies to enforce shorter working hours than we have now, how short should they be and how should this be determined? Is this really an enormous crisis or problem?

The second involves arguments that were brought up earlier, claims that presumably forcibly shortening working hours would do such things as increase income equality and reduce business cycles and financial instability. There remains no data in these additional recent posts being presented to support any of those as was requested.

Note, I think one might be able to raise longer term employment rates with such enforced shortenings of working hours, although the evidence on that is not exactly overwhelming, as the example of France in recent years shows, whose move to an enforced 35 hour work week did not noticeably budge the unemployment rate to the best of my knowledge.

BTW, I cannot resist noting that "Chapman" was already one of the top four items in this blog's "Label Cloud" even prior to this recent round of 14 posts on his work. But, of course, his work is clearly more important than anything else to post on...

Sandwichman said...


In answer to your first question, my personal view is that the best way to achieve shorter hours would be through 1. general social approbation for the progressive reduction of hours and 2. collective bargaining, with a keen sense on the part of union negotiators of the importance of continually pressing the issue. As for a government policy, currently there are many policy extant that discourage work time reduction and few that encourage it. A couple of examples of the latter policies would be the right to refuse overtime, the right to petition for adjustment of hours and short-time compensation as a key element in unemployment insurance. Some of the obstacles to voluntary work time reduction include the custom of employer paid health insurance (with no public insurance and tax credits for employer contributions), the income ceilings on payroll tax contributions to social security, the overtime PREMIUM provisions of the FLSA (effectively creating an incentive to workers for overtime rather than a disincentive to employers).

Looking back historically, I'm quite convinced that the current situation is indeed a crisis. In the past, there has often been a willingness to look at WTR as part of a policy response to unemployment. But in the period since the second world war what was formerly the prevailing view among economic reactionaries about the regulation has somehow become the mainstream consensus. Labor unions, which used to be strong advocates have dropped the issue like a hot potato.

I think there is some validity to post-Marxist arguments, such as those of Paolo Virno, that the nature of work has changed so much that old distinctions between work and leisure and work and unemployment no longer carry the same freight they once did. But I don't think such conceptual fuzziness precludes practical measures. I work 24 hours a week at a wage-paying job. I spend roughly an equivalent amount of time on unpaid research and advocacy work. I think a workweek of somewhere in the range of 24 to 32 hours, with annual vacations of five to seven weeks would be a huge improvement over most people's present arrangements.

Second question to follow.

Sandwichman said...

Your second question is about data. There is no silver bullet single data point. For one thing, it is difficult to attribute results to any given factor, especially over the medium to longer run. Bluestone and Rose discussed "The Enigma of Working Time Trends" and I would point to some of their findings as confirming my hypothesis, at least with regard to the negative case (that is longer working times increasing inequality). For example, they compare the percentage growth in annual hours worked versus percentage change in real annual earnings for prime age couples from 1973 to 1988. For high school dropouts, hours increased by 11.6% and real earnings decreased by 8.2% For high school grads, hours increased by 16.1% and earnings increased by 3.7%. For college graduates, hours increased by 16.6 hours and real earnings increased by 32.5%.

What I find persuasive about the above number is the entire context of Bluestone and Rose's discussion in which they detail how they have decomposed the data and talk about why the data source they use is more reliable than some other source.

Some highlights from their conclusions:

Average annual hours (in the US) decreased from 1967 to 1982, then rebounded to return to the 1967 level by 1995.

"Overwork" is most clearly evident in families, with a 600 hour increase in the joint working time of couples.

The family average wage decreased for all families except those headed by someone with at least a four-year college degree.

The long-term trend of translating economic growth at least partially into shorter hours has been reversed.

But I do have to add that data confirming the income equality effects of shorter working time are a bit hard to come by when there are few countries pursuing persistent policies of work-time reduction. Even the famous (and much reviled) French 35-hour work week is rather equivocal in that regard.

Sandwichman said...

And, of course, it's possible to explain longer hours as the result rather than the cause of income inequality as Bell and Freeman do in a 2000 paper. The corollary being that greater income equality will result in shorter hours. Frankly, I don't care which comes first but I would tend to expect a reciprocal relationship, with either virtuous or vicious cycles.

This paper seeks to explain the greater hours worked by Americans compared to Germans in terms of forward-looking labor supply responses to differences in earnings inequality between the countries. We argue that workers choose current hours of work to gain promotions and advance in the distribution of earnings. Since US earnings are more unequally distributed than German earnings, the same extra work pays off more in the US, generating more hours worked. Supporting this inequality–hours hypothesis, we show that in both countries hours worked is positively related to earnings inequality in cross-section occupational contrasts and that hours worked raises future wages and promotion prospects in longitudinal data.

Sandwichman said...

Whoops, I left out the title of the Bell & Freeman article I abstracted from above.

"The incentive for working hard: explaining hours worked differences in the US and Germany"

Barkley Rosser said...


On the Bell and Freeman paper I would tend to agree with the view that the causation is more likely to work the other way around: inequality breeds longer hours. Regarding Bluestone and Rose, we are back to why there has been an increase in inequality, again with me doubtful it is caused by any lengthening of work hours, or significantly or substantially so. The bottom line here is that this period saw rising inequality of earnings around most of the world. Much broader trends at work here, I fear.

Sandwichman said...


As I said in the other conversation, the stock response to empirical evidence would be that "correlation doesn't imply causation..."

First, I would like to point out that in this matter causation doesn't have to go in either one direction or the other. It can go in both directions depending on individual responses to social opportunities. In my view, it's more likely to go in both directions, rather than only in one. I would argue (for historical reasons) that the more important factor is the long hours.

I have a couple of problems with the Bell and Freeman explanation. First, in considering the range of possible explanations, they didn't consider the possibility of longer hours leading to higher inequality other than as a result of increased opportunity for employer observation.

Their approach is consistent with the "standard labor supply model" and thus it is no surprise that you or other economists would "tend to agree with the view..." But the standard labor supply model has been severely questioned both empirically (Pencavel 1980, Altman 2001, Farzin and Akao 2006) and in terms of its theoretical 'pedigree' and/or epistemological coherence (Derobert 2001, Spencer 2003 & 2004, Philp, Slater and Harvie 2005).

Meanwhile, the alternative explanation, which would be consistent with the Chapman theory, has never been directly challenged, just ignored or "set aside for simplicity's sake". I go through this in detail in "Missing, the strange disappearance of S.J. Chapman's theory of the hours of labour."

But leaving aside the Chapman alternative for a moment, consider what Keynes says about what he calls the "classical theory of employment" in his General Theory. There are two postulates: 1. Wages equal to the marginal product and 2. Utility of the wage equal to marginal disutility of quantity of employment. But those postulates only obtain assuming there is no such thing as "involuntary unemployment."

"But to suppose that this is invariably the case would be absurd. For more labour than is at present employed is usually available at the existing money-wage, even though the price of wage-goods is rising and, consequently, the real wage falling. If this is true, the wage-goods equivalent of the existing money-wage is not an accurate indication of the marginal disutility of labour, and the second postulate does not hold good."

Now the standard labor supply model clearly is an implementation of that "second postulate". So unless one is assuming a condition of full employment, there needs to be a much more substantial justification for going with the standard model than "it's the standard model" or "I tend to agree with the view." In other words there needs to be an explicit reason for rejecting the more rigorous theoretical formulation in favor of an ad hoc and widely discredited "orthodox" one.

The really ironic part of preferring the now orthodox model to the Chapman model is that Hicks explicitly justified setting aside Chapman "for the moment" on the grounds that unions were sufficiently strong to enforce any reasonable hours demands they might make. Talk about implicit assumptions!

Barkley Rosser said...


This will be my last comment on this for some time, as, sorry, this issue is simply one that I am not all that worked up about one way or the other. However, a curious aspect of this is implied in your last remark here and is also in some of your earlier posts on Chapman. His argument is actually neoclassical, as Marshall and Hicks liked it, even if he came to be "set aside" whether fairly or unfairly. You criticize lots of economists for having "standard" labor supply models, when in fact your hero is an early promulgator of absolute orthodoxy in labor economics, even if he was an advocate of shorter working hours.

So, which is it, S-man? Are we supposed to be impressed by Chapman's neoclassical orthodoxy or ignore it while we buy into his policy prescriptions?

Sandwichman said...

You're quite right that Chapman's neoclassical credentials are impeccable. What I'm trying to say (and what Chris Nyland said twenty years ago) is that contemporary "neoclassical" economics ignores IT"S OWN THEORY on the hours of labor, which, it turns out, confirms rather than rejects the much maligned position of Marx and of the eight-hour theorists with regard to wages and employment.

Myself, having been unemployed and endured the fathomless indifference and arrogance of many economists toward the issue of unemployment, I do get worked up about the refusal of economists to confront the issue with intellectual integrity (or even much curiosity). Unemployment today is an 'iatrogenic', economist-induced malady.

Anonymous said...

Meanwhile, as unemployment rises, and hours of work fall, income inequality falls:

Anonymous said...

Oh yeah: and productivity is rising as well. Which seems to result not from better rested workers, but the decline in superfluous hours of work. Might be a nobel in this for someone. Just sayin'...