Many people are optimistic that for every job eliminated by automation, new jobs will be created so that people will always find work that pays them a reasonable wage. Is there sound economic reasoning behind this or is it just blind optimism?

By the way, I am taking about 50 or 100 years from now when we would expect technology to be able to automate 95% of today's jobs.

I have often seen the argument that loss of jobs in agriculture and later in manufacturing were replaced by jobs in service industries. The reasoning is that if something happens twice, it must continue to happen for ever.

Human labor is a resource. The laws of supply and demand say that a drop in demand will result in a drop in price. The decline in price leads to a reduction in quantity of labor hours supplied.

Is there a theory that says that for some resources, demand will always rise to meet supply? The only case I can imagine, is when there is a latent demand ( or shortage ) of a resource.


1 Answer 1


A short answer…

Say’s Law says “supply creates its own demand”. In the long term Say's Law will apply. In the short term a surprise will mean some suppliers are producing something they should not be and it takes time to adapt so there can be temporary unemployment. When it is widespread, that is a recession.

Link to Say's Law

This is not a satisfying answer because Say’s Law will apply whether there is less automation as in the middle of the 1800s or more automation as in the middle of the 2100s. This suggests that for a long term trend like technology improvement where presumably you are interested in an answer that applies in an equilibrium, it is not productive to ask if supply and demand are mismatched.

In a highly automated world if a person works what we now consider to be part-time, they still have 1 job so jobs per person might not be the best measure for showing the effects of automation. This suggests we should consider not only the unemployment rate. I consider leisure and absence from the labour force.

As technology driven productivity increases, people can accept a mix of more income and more leisure. These are substitutable so we would expect both to increase in certain proportions, which is what occurred in the past. In the future both will increase in unknown proportions.

A longer answer…

A historical change under the circumstance of a technology driven increase in economic productivity is more years young people spent not working while enrolled in an “education” system that increases subsequent productivity but also seems to wastes peoples’ time. Productivity increases can make this possible if parents and taxpayers transfer an increasing amount of their output to young people.

Non-work years early in life

This graph of the percentage of the population with college attainment indirectly suggests that years spent in education has increased since 1940.

Link to educational attainment data

educational attainment graph

Leisure years late in life

Consider also the historical increase in the number of leisure years between initial retirement and the end of life. At the following link is information on the number of years spent in retirement. It shows that in the U.S. it increased from 11.3 or 15.0 (depending upon gender) in 1970 to 16.4 or 19.8 in 2018.

Link to years in retirement data

Leisure hours per week

In the following graph we see that leisure hours per week has been increasing (or equivalently work hours per week decreasing). This is a non-farm measure. Farm and ranch workers are only 1.3% of workers so a non-farm measure is probably still useful. It might be harder to measure farm hours.

Reference for the 1.3% figure

work week graph

More income

Lastly, I offer this graph to justify my earlier claim that income increased. More leisure (early in life, late in life, in the week) was part of a mix that included more real income.

income graph

The U6 measure of unemployment includes people who are jobless and discouraged from job seeking. The main measure of unemployment is U3 and it is defined as jobless and actively seeking a job. The people engaged in various forms of leisure or non-work do not get classified as U3 unemployed so automation does not have to lead to a long term increase in the main measure of unemployment. An increasing level of technology has been the trend for a long time, yet we know that immediately before the 2020 pandemic there was not an equilibrium condition of "mass unemployment" and so I would not expect it after the pandemic.

  • $\begingroup$ Thanks for the answer. My main concern that demand for human labor must decrease as other better sources of this labor are created. Surely this must force down the price of human labor. For example, if a farmer buys a tractor that allows one person to do the work of 10, this could allow 10 workers to work on the farm for 1 tenth of the time. This is a big improvement in productivity. But why would the farmer pay each one a full wage rather than just hire a single worker? Given that most will end up unemployed, he should even be able to negotiate a lower salary for the one employee. $\endgroup$
    – Garincha
    Commented Dec 20, 2021 at 3:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.