The Economic Theory of Automation

Current economic theories of the effect of automation are very incomplete, giving no clear-cut conclusions of the over-all effect on the pace of automation and changes in total employment. A theory is sketched out here to suggest that the total effect is positive, that is that automation tends to increase total employment rather than decrease it. This is very important because many of the numerous current arguments are based on the assumption that automation in general decreases employment and that we may tolerate automation only because it increases productivity and economic welfare. Inherent in the misunderstanding is the assumption that labour's position is worsened by automation to the benefit of capital. It is certainly true that there is danger to specific jobs, and union organizations based upon them. But the demand for labour as a whole tends to increase, according to our new global theory, as the pace of automation quickens. During an economic recession we should increase automation, not decrease it.

The proof of this theory remains as yet sketchy, but what is offered here does provide substantial support for its belief, and very little support for the opposition. The argument leaves aside questions of changing costs and prices and elasticities of demand which explain the sharing of improving productivity. It uses a simple model of homogeneous man-hours of work, balancing the man hours lost through the installation of machines against the man hours required in the production and installation of the machines. It takes into account the fact that the machines must be produced and installed before they cause unemployment, that those rendered unemployed will tend to seek employment elsewhere and not remain indefinitely unemployed, and that the amount of employment in making machines is likely to be large indeed relative to the loss of employment during the few months succeeding the installation.

This theory throws in relief the very great importance of high mobility of labour, and the training and information services needed to assist in this mobility. The size of the employing organization is of some importance in helping with such adjustments, as in the basic schooling of the whole population.

More information is needed to assess these matters, and the theory suggests the kind of information that is needed for such assessment. We need to know for example who are actually thrown out of work, how long a period of unemployment is experienced by those who are rendered unemployed by automation.