The preceding picture is a far cry from the standard neoclassical U-shaped microeco- nomic cost curves.30It also undermined the notion of “fixed” production coefficients. 30As noted, standard neoclassical analysis added a normal profit per unit output to the ac curve to get what is in effect a price of production curve. It is also typically focused on net output, which
The latter camp therefore attempts to reinstate this idea in one of two ways. First, by taking production coefficients corresponding to the end of the first, second, and third shifts, operated at customary lengths and intensities of the working day, as repre- senting separate “technologies.” The long-run competitive combination under given wages and prices would then correspond to the one with the minimum average cost (Kurz and Salvadori 1995, 204–205, 474). But it must be said that this stretches things rather far, since the definition of a “technology” now encompasses not only socially determined working conditions but also all potential combinations of wage payment schemes and shift lengths, intensities, and premia. The second alternative goes in the opposite direction by assuming that there are no shift premia, that the labor coefficient is constant across all shifts, and that wages are paid per hour. These conditions ensure that unit labor cost31and hence average variable cost is constant across shifts (in the latter case because average material cost is also assumed to be constant). Since average fixed cost always declines as output rises, average total cost declines steadily until pro- duction hits engineering capacity. On the assumption that competition forces firms to operate in the long run at their minimum cost point, we may then characterize a “tech- nology” by its production coefficients at engineering capacity (Andrews 1949, 58–59, 61, 65, 80, diagram I).32Once again, any change in work conditions would change the magnitude of the production coefficients. Moreover, since both alternatives as- sume that competition forces firms to operate in the long run at their minimum cost point, the associated production coefficients cannot then also be used to characterize production in the short run.
As previously noted, the second form of the fixed coefficient hypothesis, in which material and labor coefficients as well as hourly wages are constant across all shifts, also plays a central role in the post-Keynesian tradition. These conditions ensure that average variable (prime) cost is constant across shifts. Then under conditions of oli- gopoly rather than competition, price is assumed to be set by adding a markup over prime costs in accordance with the particular monopoly power of a firm. Excess ca- pacity is assumed to be normal in this case, and the focus is generally on the short run (Sawyer 1985, 28; Lavoie 1996b, 122–123; Dutt 1997, 245–246; Lavoie 2003, 59; Shaikh 2009, sec. 9).
We will return to these contrasts in chapter 8 during the examination of theories of perfect and imperfect competition. But in the meantime three further points are
excludes any possible influence of changing material costs across shifts. And it typically assumes the same wage for all shifts, which excludes the possibility of wage premia across shifts.
31 Even if the labor coefficient was the same across all shifts, if wages were paid per worker, the daily wage bill and hence daily unit labor costs would rise stepwise at the beginning of each shift because productivity is lower (and the labor coefficient thereby higher) at the beginning of a shift than it is once the shift has gotten going. The shape of the average cost curve would then depend on the respective influences of declining average fixed costs and stepwise rising unit labor costs. Then the minimum could not be specified a priori.
32 Andrews (1949, 89) assumes that the minimum cost level of production is at the end of the first shift, so that one shift is normal. He assumes that avc is constant over the shift, so that ac declines with output due to the fact that afc does the same. As he points out, the fact that average variable costs are horizontal implies that marginal cost is also horizontal, which “makes nonsense of any idea that in a purely competitive market . . . equilibrium price would be that which equaled marginal prime costs.”
important. First of all, the notion of “excess” capacity has no standing unless one can specify what is meant by the normal capacity. The neoclassical and classical traditions differ sharply on how they characterize production. Nonetheless, they share the view that under competitive conditions the economically desirable utilization of plant and equipment is at the minimum point of average cost (Liebhafsky and Liebhafsky 1968, 277). Insofar as the minimum cost point occurs at the end of the first or second shifts, economic capacity will be substantially below engineering capacity. The difference be- tween the latter and the former is economically desired reserve capacity, which can be used to meeting short-run fluctuations in demand. From this point of view, true ex- cess capacity exists only when plants are running shorter-than-desirable shifts and/or are unable to operate all cost-efficient machines (Winston 1974, 1301). Persistent ex- cess capacity is then a signal to reduce new investment, while persistent utilization of reserve capacity is a signal to raise new investment. By conflating reserve capac- ity with excess capacity, post-Keynesian economics typically downplays supply-side considerations and exaggerates the influence of the demand side.
The second point has to do with the relation between micro processes and macro patterns. At any moment of time, depending on technology and on cost conditions, some types of plants will normally operate with one 8-hour shift, others with two, and still others with two-and-a-half. If the daily engineering limit to machine operation is 20 hours, and if we define the rate of capacity utilization as the ratio of actual shift length to the engineering limit, these shift patterns would correspond to normal rates of capacity utilization 40%, 80%, and 100%, respectively. Then sufficient changes in shift premia could induce sharp discrete changes in normal rates of capacity utiliza- tion at the level of individual plants, say from 100% to 40%, or from 40% to 80%, and so on. Yet at the aggregate level the average rate of capacity utilization might change quite smoothly as individual firms shift at different points, so that the aggregate func- tional relation between shift premia and capacity utilization might be very different from that at individual plant levels. The micro-level analysis is relevant to the behavior of individual firms. At an aggregate level the main significance of the microeconomic connection is that it identifies potentially key variables. But the functional forms which obtain between these variables at the micro level do not generally carry over to the behavior of aggregates. This is, of course, the point made previously in chapter 3: the statistical average agent is not representative of any single firm, precisely because aggregates have emergent properties.
The third point has to do with the observation that the rule p = mc is consistent with multiple production levels (figures 4.16 and 4.17), so that it is useless in identifying profit-maximizing output. The latter task would require direct calculation of profit, as in figure 4.18. We will see in the next section that this issue has been repeatedly raised for almost a century in light of the empirical evidence on cost curves. The reaction of neoclassical theory has been to admit the possibility (sotto voce), ignore it, ostracize attempts to build upon it, and when necessary to fall back on the argument that in any case the failure of the p = mc rule does not jeopardize the more general neoclassical claim that firms select output so as to maximize short-term profits (Machlup 1946; Bishop 1948; Lee 1984; Marcuzzo 1996, 7–15). We will see in chapter 7 that Harrod criticizes the logic of the neoclassical argument and constructs a pathway to the classi- cal notion that even in the short run the optimal point of production of a firm is at its lowest average cost of production.