Friday, 13 August 2010

Lump of labour fallacy fallacies.

The lump of labour fallacy is often cited in response to the idea that shortening the working week alleviates unemployment. The latter idea, of course, being that if less work is done by existing employees, that will leave work that can be done by some of the unemployed.

A fair amount of nonsense has been published on this whole subject. I’ll set out the flaw in the shorter working week idea and then deal with various bits of nonsense that surround the subject. Plus I’ll argue that the phrase “lump of labour” is a poor description of the flaw in shortening the working week.

Shortening the working week is actually just ONE of numerous methods of REDUCING LABOUR SUPPLY. That is, a similar “employment creation” argument can be erected in relation to ANY method of reducing labour supply. For example it is sometimes argued that early retirement or delayed entry into the labour market for youths will leave unfilled vacancies which can allegedly be filled by the unemployed. And a nice example of a labour supply reduction scheme from history was the idea advocated by King James I of England (1566 –1625), namely that unemployment could be reduced by shipping the unemployed off to Newfoundland and Virginia. He thought that would reduce the number of unemployed in England.

In view of the variety of labour supply reduction schemes, I’ll refer henceforth to “labour supply reduction” (LSR) rather than to “shortening the working week”, for the most part.

Clearly a significant proportion of economists cannot see the flaw in LSR because the French actually implemented a shortened working week a few years ago because they thought this would reduce unemployment.

The flaw in the argument is actually closely related to the fundamental reason why inflation arises near full employment, and for the following reasons.

Given significantly higher unemployment than normal, it is relatively easy for employers to find the skills they want: the more unemployed there are, the better the chance of finding a specific skill. However, given rising demand and falling unemployment, it becomes progressively more difficult for employers to find skills.

Now when demand for a skill exceeds supply, the wage for that skill tends to rise. And that does not matter too much so long as only a few skills are involved. But given falling unemployment, skill shortages get progressively worse until the point comes where the wage for so many skills or professions is being forced upwards that general inflation ensues. Or the labour market exacerbates whatever inflation already exists.

Now suppose employment in an economy is at a level such that a further increase in demand and employment will cause excess inflation. That level is sometimes called NAIRU (Non Acceleration Inflation Rate of Unemployment) and sometimes called the “Natural Rate” of unemployment. I’ll use the acronym NAIRU for want of any better (rather than because I agree with every detail of the NAIRU theory on the strictly correct definition of the phrase).

Suppose also that employees are forced to work fewer hours per week. That will result, as the advocates of the shorter working week claim, in work being left undone. Employers will then attempt to locate the skills needed to get that work done.

But wait a minute – remember we have assumed that the economy is at NAIRU: the employment level at which locating skills is so difficult that inflation will ensue if employers try to attract those skills by bumping up the pay for said skills!


Of course, the assumption that an economy is at NAIRU is an artificial assumption. But relaxing this assumption does not get the LSR argument anywhere. That is, if unemployment is well above NAIRU, then LSR schemes WILL WORK. Reason is that the above skill shortage point does not apply. But in this situation, employment can perfectly well be raised by increasing demand! That is, there is NO NEED for LSR schemes!

In short there is NO LOGICAL NICHE for LSR schemes.

Labour supply reduction fallacies.

Now for some of the nonsense that surrounds this whole LSR subject.

1. The Wiki article on lump of labour claims that the reason employers may not hire extra workers on the implementation of an LSR scheme is “administrative cost to hiring more workers”. Well, that is a bit vague: there are “administrative costs” involved in doing almost anything.

The Wiki article lacks an explanation as to WHY administrative costs are prohibitively high in some situations (i.e. at NAIRU). The reason, as explained above, is that at low unemployment levels it is difficult to locate skills.

2. Bizarrely, Keynes put in a good word for shorter hours. He said “It becomes necessary to encourage wise consumption and discourage saving, and to absorb some part of the unwanted surplus by increased leisure, more holidays (which are a wonderfully good way of getting rid of money) and shorter hours’’ (p. 323). (“The Long-Term Problem of Full Employment’’, 1943, p. 323). That sentence is complete nonsense.

Keynes is clearly dealing with the paradox of thrift here, that is the fact that if the private sector saves money excessively (rather than spend it), unemployment ensues. Of course, FORCING people to go on holiday and spend money is a solution of sorts. But it’s a daft solution. The private sector only net saves when it thinks it has an inadequate stock of savings (a statement which is so obvious that it shouldn’t need making).

Far better than forcing people to go on holiday is to provide them with the level of savings they want. People can then make up their own minds as to how to split their time between work and leisure.

3. Advocates of shorter working hours often point to various beneficial effects: increased productivity, less stress, the environmental benefits of human beings consuming less. These arguments may well be strong enough to warrant shorter hours. But they have nothing to do with be basic theoretical flaw in LSR set out above.

In other words, there are several possible justifications for shorter hours, but reduced unemployment is not one of them.

4. It is often claimed by opponents of LSR that LSR advocates claim there is a fixed amount of work to be done, hence, for example fewer hours for one lot of people means others can work extra hours or find work. (e.g. see Wiki, point No 1.)

That “fixed amount of work” point is a dangerous point to make because it invites the obvious retort, namely that the total amount of work is NOT fixed in that economies EXPAND most of the time (and occasionally contract).

Samuelson actually attacks the above “fixed amount of work” idea here (p.1) and for precisely the above reason. Samuelson did not seem to realise that he is not attacking a fundamental weakness in LSR, but rather a poor description of the weakness.

In short, the phrase “lump of labour” is not a good description of the basic flaw in LSR.

The real flaw in LSR does not have much to do with “lumps” of anything or “fixed amounts of work”. The real weakness is the assumption that the “skills to vacancy matching process” by some unexplained magic, becomes more efficient just because labour supply is artificially constrained. That assumption is of course nonsense. Perhaps the flaw in LSR should be called the “labour market efficiency flaw”.


  1. Ralph,

    Congratulations! You are half way there in recognizing that the lump of labor fallacy is a poor way of criticizing shorter working time. Now let's see if you can make the next step.

    Shorter working time is NOT fundamentally labor supply reduction (LSR) but labor QUALITY enhancement (LQE). Of course this only applies within certain parameters, which in turn depend on the current state of technology. In the late 19th century and early 20th, it was often found that a reduction in the hours of work led to an actual increase in total output per worker without any other change in technology (See John Rae, Sydney J. Chapman, Josephine Goldmark and Philip Sargent Florence). Even if total output per worker declines somewhat as the result of shorter working time, as long as output per hour increases there is still an enhancement of labor quality (See Edward Denison on this point).

    So, what is the theoretically likely outcome of labor quality enhancement? It is generally agreed among economists that an increase in productivity leads in the long run to expansion of employment. I won't pass judgment on whether this is always necessarily true. But it IS the consensus of economists.

    But wait, there's one more effect to consider. Ceteris paribus, an increases in labor quality reduces labor costs relative to the cost of potential substitutes. And therein is the undeniable contribution of work time reduction to job creation. Again, this effect only applies within a given range of hours where the LQE exceeds the LSR. At an extreme there will be a reduction in both labor supply AND labor quality.

    Where would that golden range fall? We don't have the data but we can do trend projections. Assuming that the trend in hours reduction in the first half of the twentieth century was roughly in line with the optimal hours for output, projection of that trend over the subsequent 50 years would suggest average annual hours in the U.S. of approximately 14% less than the current average. The job creation potential continues, however, to the point of maximum hourly productivity, which I would estimate at around 75% of the optimal hours for total output.

    The reasons for assuming the earlier trend to be optimal but the last fifty years to be a "deviation from trend" is that up until mid-century shorter working time was a priority for labor unions and the hourly wage was the preponderant form of employee compensation. In the second half of the 20th century, unions abandoned the shorter working time strategy and "quasi-fixed" per employee benefits rose sharply as a component of total compensation.

    Hypothetically, then, given current technology, there could be an hours reduction of nearly 36% before the LSR effect would overtake the LQE effect. Such a reduction would presumably be far in excess both of what would be culturally acceptable to most Americans and of the potential unemployed and inactive labor supply that could be attracted into the labor force by the expanded labor demand. Therefore the "problem" of LSR doesn't really appear until well beyond any reasonable proposal for work time reduction.

    See my related PowerPoint presentation that reviews the theoretical arguments, historical trends and projections of those trends, "Climate change, growth and the productivity/employment dividend from shorter hours":

  2. Sandwichman: I had a look at your Power Point presentation on working hours (and other publications). Working hours is obviously your speciality, so I’ll be careful what I say!

    I agree that shorter working hours can lead to improved output per hour. I actually made that point above (point No 3 near the end).

    But as you rightly point out, the exact effect depends on the “state of technology”. It doubtless depends on numerous other things: the type or work, plus the effect will vary from individual to individual.

    Thus I’m in favour of each employer (and employees) working out for themselves what is best in each case, rather than across the board measures. If someone wants to work 60 hours a week, let them. Ditto for someone wanting to work 20 hours.

    I don’t agree with your claim that “an increase in productivity leads in the long run to expansion of employment.” We have had astronomic improvements in productivity over the last century, yet employment levels are much the same now as a century ago.

    Also I don’t agree with your claim that “an increase in labor quality reduces labor costs relative to the cost of potential substitutes.” Reason is that the cost of the substitutes (capital equipment and materials) is itself very largely and ultimately made up of the cost of labour. E.g. if everyone’s wage was doubled in dollar terms tomorrow, the price of everything else from cars to timber would also quickly double as well.

    I’ve seen evidence that when people raise hours worked from about 40 hours a week to 60, most people initially produce more, but after a few weeks they get knackered, and output per week declines to the original 40 hour level. However I’m doubtful about reducing hours to 20 (except for those who WANT to work part time). This idea gains support from your “US weekly hours 1850-2010” chart, which shows hours declining from 65 in the 1800s to about 40 in about 1940, and then staying at that level.

    My guess is that that is explained by the fact that in the 1800s people had to work about 60 hours to get the basic essentials. Nowadays, there is not that pressure, so people work the hours they want, more or less. And most choose about 35-40 hours.

  3. Ralph,

    I'm pleased you took the time to read and respond to my thoughts on this issue. Your reply is thoughtful and I wouldn't want to argue over our different interpretations. Your views are "fair enough".

    I too am in favor of a voluntary working out of hours, although I am much more skeptical of the likelihood of this occurring satisfactorily under current "rigged market" conditions. This is not to say that rigged regulation would be any better, though.

    Even if I were to accept your contention that most people freely "choose" to work 35-40 hours a week, this still leaves the absurd conclusions that the unemployed choose to work 0 hours and the underemployed choose to work fewer hours than they desire to work. In my view, the reality of unemployment casts serious doubt on any presumption of choice.

    The conventions of wage labor place an inordinate amount of emphasis on the individual sale of labor and consequently on an illusion of individual achievement. The actual organization of work is highly collective and co-operative. The supposed individual contribution to this collective output is derivative and relies excessively on averages and conventions. But if twelve people can lift a heavy weight two feet, it doesn't follow that one person can lift the same weight even two inches!

    I want to reiterate that the claim about productivity leading in the long run to employment is not mine. It is a truism among economists. I would see it as an article of faith rather than a scientific statement. I only use it in the sense of "if you're going to insist on x, then you have to admit that y follows..."

    Whether the 40-hour level is an arbitrary plateau or some kind of enduring optimum can't be determined from the data or a trend projection. It is instead a matter of judgment. Ultimately, in my judgment, the number of hours I work is irrelevant. What really matters is the quality of those hours. If I'm doing work that I enjoy, that I feel is meaningful and that doesn't wear me out physically or mentally, 35 or 40 hours a week may well be TOO FEW!


Post a comment.