I was asked to explain something I asserted in How to mow the lawn, regarding the tendency of infinite investment growth to produce either unemployment or overwork or a mixture of the two. Now this is miles away from mainstream thought, and indeed I have not fully explored the speculation myself, but the basic concept is this: In the geometric growth model we have a population that grows at an ever-increasing (compounding) rate, and we attempt to overcome the ever-increasing needs of that populace through ever-increasing production. We take advantage of the compounding nature of economic growth to make sure it follows the same pattern as population, by continually investing a portion of our output in improving our ability to produce. This is really the basis of economic growth at all: Rather than putting 100% of our abilities towards making what we consume today, we dedicate say, 10% to improved machines and methods, which will make tomorrow’s production more abundant than today’s. By partially restraining our appetite for consumption and properly directing what we have saved into real improvement of means, we will enable ourselves to produce more. If we repeat this process each year or each generation, we can expect that next year (since we have increased our productive capacity) the 10% of our output which we invest is a bigger total investment than was this year’s. To take Keynes’s metaphor, we are eating a smaller portion of the pie, in order to make the pie grow larger, and by continuing in this practice, we can expect that the pie eventually becomes so large that the portion which we consume is actually larger than the entire pie with which we started. By investing a roughly fixed percentage of each year’s output in improving next year’s methods, we create an economy of compounding growth.
So once this has been achieved (a fixed practice of continuous, substantial investment) we have on the chart two geometric curves, continually growing ever steeper according to their rate of growth, and as long as our level of investment is sufficiently high, the curve on top will forever be production, while population is always a little shallower, a little lower, and always falling a little behind. In fact, the whole idea is that production must grow faster than does population so that quality of life may continuously improve. If the curves are actually parallel (the rate of growth in production and population are identical) quality of life must actually decline in the sense that the per-capita production is actually declining. This is reflected visually in the fact that the curves always stay the same vertical distance apart, but the that distance becomes more and more insignificant compared the enormity of the totals. What I’m getting at is that the growth of production must be on a trajectory to pull ahead of population, not track it, for the outcome to seem desirable. Per-capita production must increase (obviously) to achieve our goal of appreciable prosperity. If population is to grow on an infinite compounding course, production simply must compound even more aggressively to satisfy this.
[None of the above is inevitable. What I am saying is merely that given the goals of an infinite growth policy, the only apparent or real success of such a policy will consist of production growth accelerating ahead of population growth. This may not be possible, the investments may yield declining returns, production may be misallocated, production growth may hit absolute limits, etc. I merely claim that the infinite growth policy, if it ever succeeds or appears to succeed, will inevitably involve this: The rate of growth in production must exceed the rate of growth in population, because per-capita production must continually increase in order for economic policy to be of visible benefit to the consumer. Each generation must be richer, and not merely larger at the same level of wealth, for the public to view policy as a success.]
Along with the Austrians, I don’t approve of the use of mathematics to attempt to describe human action, but since I’m illustrating a concept of which I don’t approve, I’m going to include some graphs to illustrate what I’m getting at.Here we have the Malthusian view of the relationship between population and output (production), population being the red line, and projected output being the blue line. Their crossing represents the Malthusian “crisis point,” where population growth overtakes the limits of production, and scarcity and starvation take over. In this vision, the red line would not actually continue up as projected, but instead the population would be limited by the crisis, through famine and other destructive events resulting from the failure of the economy to provide for the vast numbers of people.
This represents the pattern of population and output where economic (output or production) growth is occurring at a rate identical to the rate of growth of population. This is what would happen if we tried to tune economic growth to exactly track population, with a little bit of a head start, ie perhaps by planners expanding production in response to population growth. You can see visually how the vertical distance between the curves, though always the same in absolute terms, is a smaller and smaller proportion of the height of the curves as we move further to the right. I used very simple formulas to graph this, y=x^2 for population, and y=x^2+c for output, with that +c representing production’s “head start” in this case. This allowed me to create the chart below…
This red line relates to the previous chart; it represents per-capita production, which, thanks to the simplicity of my example formulas was merely y=(x^2)/(x^2+c). I want to emphasize that it is merely the shape of the curves that are of interest; changing the +c to +1 or +100,000,000 does nothing to alter the fundamental character of the situation; if a statistician tuned these models to match real-life numbers, he would be finding out when or under what conditions population or output reached exact values, but he would not find out say, that per-capita output actually increases under this paradigm. What we learn from this is that if production grows at a rate equal to (or lower than) the rate of growth in the population, then per-capita production must continually fall and as a consequence people will have declining access to material goods.
This last chart represents what I think is the obvious implication of the expectation that infinite production growth will overcome infinite compounding population growth. I simply increased the rate of growth in production to exceed the rate of growth in population, and as you can see, the gap between the two lines grows continually larger. This represents a continuous increase (though not a compounding one) in per-capita production, and thus, in attainable quality of life. I strongly suspect that those economists who expect compounding economic growth to be the permanent solution to the problem of want are thinking of this, and not of the declining per-capita production we saw when production simply tracked above the population line.
So that’s point one of my claim: That the geometric growth model of investment, continuously stimulated to make sure we invest sufficiently to stay ahead of population growth, depends on the rate of growth in production exceeding the rate of growth in population, which implies continuous growth in per-capita production. Why is this important? Because of its relationship to employment. We’ve been pursuing the goal of increasing per-capita production in order to increase per-capita consumption, enjoyment, and quality of life. But in increasing per-capita production for a given set of methods of production, we are obviously increasing the per-capita work input required. All other things being equal, an increase in per-capita production is an increase in per-worker workload.
Yet we know this isn’t the case, obviously the first thing to be eaten up in this process would be the problem of unemployment; long before we begin to overwork existing workers we should have drawn in every available hand from the pool of unemployment. If the long term effect of this policy is to be a labor shortage, shouldn’t we enjoy a period of full employment before that hits? The reason this does not occur, nor does the rate of employment have any sort of general upward trend, is that the method by which per-capita production increases is as much by eliminating work inputs as consuming them. This should be obvious when we consider the actual production methods of today vs. those of the past: They are chiefly distinguished by an increase in what each man or group of men can do in a given time, yes, but because a far greater proportion of the lifting, shoving, cutting, breaking, calculating, measuring, etc. is done not by men, but by machines. Machines add to productivity in proportion to their ability to substitute for hands, as well as their ability to do what no hand can do. Now from a pure input/output standpoint, this is purely good news: Men’s bodies will be less strained, and better provided for. However, given our anomalous problem of unemployment in the modern age, there is more to consider.
The question of whether a machine will amplify a person’s effort or displace him completely seems like it should be a headline question of economics, as soon as it is considered. Marx, i think, tried to draw a distinction between tool and machine on this basis, and others have tried to argue that tools are good and machines bad, because of their effect on this question. The tool that must be operated by human effort can never fully displace its operator the way a robotic factory can. However, even if we decided we had the authority to ban “machines” on this basis, the most advanced of tools will merely displace every man except one, rather than every man. And depending on maximum scale…Anyhow, we gain nothing from this attempt to limit, for the benefit of employment, what we emphatically don’t want to limit, because of its immense benefits to production. The way to overcome poverty does not involve hamstringing the production of desired things (please see “How to mow the lawn” for more on that).
Now, it must be that there is, at any stage of technical development, some portion of the work that cannot be done by machine. If this portion remains relatively fixed during economic growth (by no means a certainty) then the demand for that class of work, and thus employment of that kind, must rise in line with the growth of the economy. But even still this does not necessarily imply a rise in employment, because the work involved may only be possible to a portion of the population. If it is merely a question of trainable skill, then a system of free prices can address this by increasing the relative wages of those capable of doing this unmechanizable work, and thus drawing a greater proportion of workers into that field (or those fields) with each passing generation. This would be a relatively good outcome, and one we seem to have missed, perhaps for the following reason: The big problem is that virtually all work that can be mechanized is being mechanized, and this is occurring more rapidly than the population is transitioning into the necessary trades. As to the reasons for this overly rapid mechanization, there are two: The cost of employing people is artificially increased by payroll taxes, regulatory requirements, minimum wage laws, and general favoritism that must be anticipated by any potential employer when considering whether to use man or machine to do some work. At the same time, the cost of employing machines is artificially reduced by enormous tax advantages, thanks to the superstition of scale prevalent among policymakers. For this reason we do not see the line representing demand for labor in a mechanized economy following a comprehensible course, but continually veering towards overmechanization and unemployment, even as the demand for those specific types of work that cannot be mechanized does, in fact, grow. Regrettibly, making sense of the behavior of employment during mechanization, given the two contending factors of increased worker productivity and decreased worker necessity, is not something about which i am willing to draw any conclusions at this time (beyond the above general observation). Perhaps after a few more nights sleep, or after the end of Lent (i gave up whiskey and it’s playing hell with my writing process) i will be satisfied with my grasp of the process. For now, let me say that my offhand assertion that over-automation leads to simultaneous overwork and unemployment was a gut-level suspicion that the general effect of overinvestment is to displace human hands, except those hands skilled in ways that machines cannot duplicate. Because such skilled hands represent a small, and not rapidly-growing portion of the population, the effect of these two tendencies must be to overwork the skilled hands, and lay off the unskilled hands. i say it as a hypothesis, for now.