We hear with great trepidation about Frey and Osborne’s (2013) study, which claims that 47% of jobs in the United States are going to be replaced by the rise of robots. But the really interesting question is not how many jobs might be displaced by robots based on job tasks that economists themselves determine, but whether capitalists want to invest in a given robot or not. Naturally, in a market-based economy it is rather difficult to assess the precise equilibrium outcomes of rising automation, and- in the absence of good data- philosophers and futurists (Bostrom 2017; Ford 2016) are trying to prime us on a future without work, when the robots have displaced most human workers.
Some economists have now ventured to make empirical investigations into how robots will affect productivity growth (Graetz and Michaels 2015) and overall employment figures (Acemoglu and Restrepo 2017). There is no doubt that we need more of these empirical investigations.
What I will do in this post is to theoretically lay out the logic of automation, which is summarized in the following graph:
In this model, the foundational assumption is that automation is already happening. The first effect is on the side of the employment channel, namely that it reduces the number of workers. Historically, we might argue that automation actually creates jobs, and in some modern tech sectors there is evidence of continuous automation among software engineers leading to more innovation and thus new job creation in that sector (Shestakofsky 2017). In the case of the retail sector the displacement of department store workers has thus far been offset by the growth in e-commerce, which creates jobs in warehouses and transport logistics.
But in my model, the potential growth of jobs is captured in the automation effects on the consumer channel. Lower prices of goods will increase savings of consumers, which can be channeled into increasing consumption in the same industry or in other areas of the economy. But we might also assume the opposite, namely that automation reinforces the monopoly sector’s profits (Google, Amazon and associates) and thus retains the high cost of living despite rising automation. Continued high product prices in the absence of competition implies less consumer purchasing power. Lowered consumer purchasing power also happens via the channel of displaced labor (as mentioned in the beginning), which has to compete for jobs in other areas of the labor market, which lowers the organizational clout and wage claims of workers.
The rescue for beleaguered consumer-workers happens via the consumer finance channel, as rising profits are deposited in financial institutions, who have to aggressively market those accumulated savings in the form of credit. Credit expansion in turn is made possible by lax government regulations, but it crucially buys time for the capitalists because they combine rising profits with declining wages but rising consumption. The problem with the consumer finance route is that it is inherently unstable as the lack of rising income among the masses means a high risk of default and financial crisis. Deleveraging households thus create the third way in which consumption is lowered. Lower consumption in turn predicts less profits, less investments and an economic crisis.
Automation decisions themselves are endogenous to wage trends. In the case that automation displaces workers, increases competition in the labor market, thus reduce worker bargaining power and lower wages, the pressure to automate might actually decrease, as cheaper workers compete with expensive robots. Developing countries upper middle class can still afford human butlers, which is the preserve of the upper class in the developed countries given the high cost of human labor. In the contrary case, a new consumer bonanza will allow workers to gain higher wages, which will increase the pressure to automate. This theoretical prediction has some empirical evidence as the Europeans have a higher robot penetration on average than the US, while the former also have better labor-protective legislation and higher wage costs with more generous social contributions to finance the pensions and other social welfare spending. In addition, China is the hottest market for robot imports, partly because of the manufacturing dependence of the country and partly because of the massively rising wages in the export-intensive manufacturing sectors.
Can we, thus, formulate a summarizing statement about the logic of automation? I would say, perhaps not, because the arrows are pointing to two possible outcomes: (1) a higher economic equilibrium with more productivity growth and economic growth with rising wages and more jobs, and (2) economic crisis, a depletion of middle class employment options, an entrenchment of monopoly capitalism and a further concentration of wealth in the hands of the few. For whatever reason, the more pessimistic perspective takes overhand in our contemporary experience with automation.