The theory of technological unemployment is a stepchild of economic science. The facts seem to stand in such blatant contradiction to orthodox doctrine, according to which no "permanent" technological unemployment is possible, that most American textbooks prefer not to mention the problem itself. This attitude is of recent times. The analysis to which Ricardo subjected the displacement of labor by the machine in the last edition of the Principles had stimulated a lively discussion among the later classical economists, who, as we shall see instantaneously, followed two different lines of thought. With the rise of neoclassical equilibrium analysis, the discussion died down, at least in Anglo-Saxon literature' and only recently the oldest argument against technological unemployment, originally developed by McCulloch, was revised in a little more sophisticated form by two American economists, P. H. Douglas and A. Director in The Problem of Unemployment (New York, 1931). We can, therefore, distinguish three approaches:The Technological Unemployment and Structural Unemployment Debates, Gregory Ray Woirol (1996)
1. The "Law of Markets" approach, formulated at first by McCulloch in Principles of Political Economy (first edition, 1825) Part I, chapter VII, and, as pointed out above, revised by Douglas and Director, applies Say's Law of Markets to the Labor Market. As there cannot be a general over-production of commodities produced, so there cannot be a general over-supply of labor. We shall analyze this argument in the first section of this paper, with some supplementary remarks in Section V.
2. McCulloch's argument was not taken up by the other classical authors, because it is at variance with the classical theory of the demand for labor. As John Stuart Mill stated it most pointedly: "Demand for commodities is not demand for labor" (Principles, vol. I, p. 5, para. 9). The maintenance of the demand for commodities according to Say's Law, therefore, does not militate against an over-supply of labor. It is the volume of circulating capital, interpreted as wage fund, that governs the demand for labor. Following Ricardo's lead, the theory of "compensation" of technological displacement of laborers was worked out. In contrast to the Law of Markets approach, which does not allow any exceptions to the denial of "permanent" technological unemployment, the Wage Fund School maintains the occurrence of compensation only as the general rule, exceptions from which are deemed possible though unlikely. In Section III, we shall consider this argument.
3. The neo-classical equilibrium approach differs from the preceding ones by denying the possibility of technological unemployment only as to a state of long-run general equilibrium proper, in which complete adjustment of all the variables of the economic system is attained (size of firm, input, output, prices of goods produced, prices of productive services, interest rate). The difference of the neo-classical approach from the Law of Markets approach is concealed by the use of the terms "temporary" and "permanent" by the latter school. By "permanent," Douglas and Director do not refer to the state of long-run equilibrium proper. This is clear from their definition of "temporary" technological unemployment (op. cit., pp. 113 ff.), which refers only to such obstacles to the reabsorption of laborers as: slow working of competitive mechanism, slow transfer of expenditure from one good to the other, or of workers from one industry to the other. A state of affairs in which these obstacles are overcome (as we shall assume throughout the present paper) still might be in a merely "short-run equilibrium" in the neoclassical sense, which is based on the assumption that all equipment is "given" as to quality and quantity, while long-run equilibrium proper in the neo-classical sense requires, among other things, the adjustment of the size of the firm and of the quality of equipment in such a way that average costs equal price for all firms. Indeed, if the Law of Markets is valid at all, it must be applicable to periods of any length, provided only the period is long enough to overcome the temporary obstacles; and similar considerations apply, as we shall see, to the wage-fund argument.
While there is little merit in the two classical approaches, the neoclassical one stands on much firmer ground, on account of its lesser scope. However, even the neo-classical approach is far from giving the unambiguous answer its adherents ascribe to it. This will be shown in Section IV. On the other hand, while the unqualified denial of "permanent" technological unemployment in traditional theory is not justified, preliminary empirical investigations (which cannot be presented in the present paper) have convinced the present writer that popular opinion vastly exaggerates the amount of unemployment which properly could be called "technological." The relative small size of technological unemployment in history is attributable, partly, to the independent forces increasing employment, which briefly will be discussed in the last section. In no case would it be permissible to use simply the current unemployment statistics as a verification or a repudiation of the theories which affirm or deny the existence of technological progress that creates unemployment. Hitherto the discussion has been marred by a confusion of historical and theoretical statements. ... What would one think of an argument against the law of gravity based on the undeniable truth that only a very small number of people habitually fall against the center of the earth with an acceleration of 33 feet per second? And yet, the reference to unanalyzed observation is not worse than the reference to unanalyzed historical facts. In order to obtain a reliable answer to our question, it is necessary to keep constant the other factors as far as they are truly independent, i.e., not exclusively or almost exclusively governed by the volume of technological unemployment itself.
In this mid to late 1940s theoretical literature, a third period of consensus was achieved to add to the consensus periods of 1927—29 (the Say-Douglas purchasing power argument) and 1933—40 (the neoclassical price-flexibility argument). As illustrated by the work of Neisser, Hagen, Lange, Belfer, and Pu, this consensus — built on Keynesian arguments — was that no mechanism existed in the economic system to guarantee the automatic reabsorption of technologically unemployed labor.
This is where the technological unemployment debates ended. With the full employment of World War II and the spread of Keynesian policy ideas about how to achieve full employment, the pessimism about unemployment and productivity trends that had motivated the debates of the 1920s and 1930s disappeared.... In effect by the early 1950s the state of aggregate professional opinion about technological unemployment had returned to the confident views of the late 1920s.