Less than two years before Stephen Hawking’s death in December 2016, the renowned physicist has warned in The Guardian that the automation and artificial intelligence was destroying the middle class. He wrote that there was a widening gap between society and the intellectual elites he was a member of. According to Hawking, Brexit was the embodiment of this gap. He claimed that the poorer British people who were left behind used Brexit as a way of showing a red card to the government.

Hawking believed that globalization and the accelerating pace of technological change were responsible for social unrest. According to him, the rich could get richer and the poor were to get poorer primarily because of automation, a phenomenon consisting of limiting or replacing human labor with autonomous machines, or of performing work that was previously impossible.

He claimed that automation has already “decimated” employment in traditional manufacturing, and soon artificial intelligence was to “extend this destruction” into the middle-classes. Only positions where empathy, creativity and supervisory functions were important were supposed to remain. He painted a grim picture of the future: Growing inequalities all around the world, with small groups of people reaping huge profits on one side, and with ever-expanding masses of poor people without the possibility of survival on the other.

What was Hawking worried about? Imagine a future for Mark, a freshman studying in college. Mark has no professional experience, so he found a job as a clerk in a local grocery store as it did not require any specific skills. Many people like him perform simple and low-paid jobs that are easy to automate. The owner of the shop, Leon, will soon buy an automatic cash register that will make Mark’s job obsolete. Leon has already bought a self-learning computer program to replace accounting in his small business. Because of this he no longer requires the services of Joanne, who is an accountant working in a small accounting office.

It seems that only the owner of the store benefits from automation.

But the process is far from over. A Japanese online company Nozama will soon enter the market, delivering purchases via flying drones. Leon will lose his shop. Nozama’s enormous warehouses will be operated by a host of self-learning robots that will require supervision of only a few dozen people. Someday their jobs will be made obsolete as well.

On the surface it seems that automation will harm all of these people. The freshman Mark will not get the experience needed to find a job that requires creativity, empathy, and supervisory skills. He will wind up living on the streets. What’s more, according to Hawking, masses of the unemployed will be joined by immigrants seduced by an illusion of happiness straight from Instagram.

Hawking was not an economist — he wrote as a very intelligent layman. With no economic skill set nor economic knowledge he could offer little more than platitudes. And he was not the only one. Entrepreneur Elon Musk and independent candidate for US president entrepreneur and statistician Andrew Yang also warn against automation.

In a survey conducted by Pew Research Center in 2014, a wide group of experts in the field of technology (only four per thirty people mentioned by name were professional economists) indicated the following threats of automation:

  1. Technological unemployment: low-level and especially blue-collar employees are already being fired, and soon the same fate will fall upon white-collar workers as well.

  2. Few highly-skilled specialists will earn more, but others will be displaced at best into lower paying service industry jobs. This will lead to an increase in income inequality and will create a class of permanently unemployed people.

  3. The education system as well as political and economic institutions are unprepared for the challenges of automation.

Among the opportunities offered by automation, the respondents mentioned:

  1. Though some occupations will go away, in the past technological changes have always created even more new jobs. We will adapt to these changes by finding new job opportunities and by using uniquely human skills.

  2. Technology will free us from tedious, repetitive work, and will turn our relation with work into something more positive and socially beneficial.

  3. After all, we control our destiny and we choose our future.

In spite of strong arguments against it, the fears of “technological unemployment” linger since the industrial revolution. At its dawn economic conditions were difficult after the Napoleonic Wars. Some manual workers decided to destroy stocking frames, believing that such textile machinery deprived them of their work. After 1811 a wave of riots swept over Britain, with workers burning mills and sending death threats to force inventors into exile. The rebels called themselves Luddites in honor of the fictional character of Ned Ludd, who was supposed to be the first to destroy stocking frames. Perhaps they were not aware of a fact that Henry Hazlitt pointed out later in his Economics in One Lesson. He wrote that in 27 years after the introduction of cotton-spinning machinery in 1760, labor savings strengthened the weaving industry so that the number of employed weavers and spinners had risen from 7,900 to 320,000.

Automation was further examined by an American economist Murray N. Rothbard in his book Science, Technology, and Government.

According to Rothbard, instead of destroying jobs, automation merely shifts them. In order to understand this process, let us discuss price elasticity.

Price elasticity measures the ratio of the percentage change in demand to the percentage change in price of a particular good. Demand is what we call elastic when percentage change (e.g. growth) in demand is greater than percentage change (e.g. decline) in price. Suppose that the price of a cheap ham is $2.95 per pound. Lowering the price by 12 cents per pound may encourage us to buy more. Instead of eating 163 pounds of ham per year we may increase our consumption of ham to 172 pounds per year.

This means that the butcher’s revenue will increase. Let us see: At a price of $2.95/lbs a year, each customer pays to the butcher around $2.95/lbs x 163 lbs = $480. After lowering the price to $2.83/lbs, his annual revenue per consumer will increase to around $2.83/lbs x 172 lbs = $487.

Elastic demand means that people more easily adapt to price increases; it may be the case that they can substitute a good that is getting more expensive with an inexpensive alternative. In result, they may even buy more goods with the same amount of money.

On the other hand, inelastic demand means that a percentage change (e.g. decline) in price results in a smaller percentage change (e.g. growth) in demand.

Let us get back to the butcher. This time, let us assume that the price decline will result in an increase of average annual purchases to only 168 pounds of ham. At a price of $2.95/lbs, annual revenues previously amounted to $2.95/lbs x 163 lbs = around $480 per customer. At a lower price, the butcher’s revenue will fall to $2.83/lbs x 168 lbs = around $475.

For example, demand can be inelastic on a relatively monopolized fuel market where consumers cannot easily withdraw from paying increased prices, or on markets where consumer demand has already been saturated, so that they do want to increase purchases when prices decline.

Rothbard pointed out that there are many industries in the economy where demand is elastic and many where demand is inelastic. Market research1 confirms this finding.

We have seen why producers willing to increase their revenue will increase production whenever demand is elastic. You may feel that the explanation was a bit too abstract. What was the real perspective of the butcher? He lowered the employment of labor and increased the employment of machines. This enabled him to lower his prices. The sales increased. Given the new ratio of labor to machinery costs, the butcher has to keep a sufficient number of employees in order to maintain the increased production.

This is why the increased production will lead to less redundancies. But there is more. Production of these new machines also requires some labor. Some machines will be produced by yet other machines that will also require labor to be produced. This is why in some proportion the production will still require labor and other primary factors, or in other words natural resources. The cost of factors of production will in part shift to labor, and in part to the owners of metal ores and other raw materials.

For example: The butcher’s cost of each pound of ham is slightly below $2.83/lbs, split between machinery costs of $1.24/lbs and the same amount of labor costs. Suppose that the butcher bought the machines and laid off some of the employees, and pays machinery costs of $1.36/lbs or 12 cents per pound more, and labor costs of $1.06/lbs or 18 cents per pound less. Let us calculate the butcher’s annual revenue per client:

  1. Before the layoffs: $1.24/lbs x 163 lbs = around $202 equally for machinery and labor.

  2. After the layoffs:

    1. $1.06/lbs x 163 lbs = around $173 for labor,

    2. $1.36/lbs x 163 lbs = around $221 for machinery.

  3. After the layoffs and after increasing production:

    1. $1.06/lbs x 172 lbs = around $183 for labor,

    2. $1.36/lbs x 172 lbs = around $233 for machinery.

As we can see, even though the butcher’s ratio of layoffs to increase in machinery costs was 3:2, the eventual layoffs amounted only to $19 instead of $29, while machinery costs increased by $31, meaning a new ratio of 5:8.

Now for the essential point: The $31 of machinery costs will not simply vanish from the economy. Machine manufacturers will also have to employ labor and other factors of production. Employment in machine factories will increase, as no shift in the ratio of labor to machinery costs occurred there.

Let us now return to inelastic markets, where a decline in prices results in a lesser degree of increase in consumer demand. Given inelastic demand for a specific good total consumer spending on it will decrease, as consumers will rather spend money on other goods. This means that the employment will increase in the other sectors of the economy, for example in machinery manufacturing. Consumers will have more goods, and “technological unemployment” will not appear.

Let us make an important note here. Critics of automation offer dire visions of total and sudden displacement of human labor by machines. They sometimes even allude to the Terminator movies, in which cyborgs have violently eradicated humanity. In their visions employees do not have time for training because they lose their jobs immediately and in large groups.

There is nothing further from the truth. All serious changes in production processes, such as modifications in production lines, fine-tuning of supply chains, adjustments in production scheduling, or even thorough reorganization of the sales system, require time. Such changes need to be tested and implemented gradually and carefully. Some parts of the production process will be modified, while other parts will remain unchanged. The existing employees can be transferred to these departments (to catch up with the overall increase of production).

In reality this process is much more benevolent than it seems. According to research by Earl Tony Halsbury presented in the book The Push-Button World, there is a 2 percent “natural” turnover in industry per annum due to retirement of old and recruiting of young workers. In addition, in stable industries up to 10 percent of employees leave every year for other reasons, for example looking for a better job. Moreover, manufacturers do not want to dismiss even partially qualified personnel, because the costs of training new employees are higher than retraining costs. For these reasons, mass redundancies will not occur even in inelastic markets, because natural and voluntary turnover will protect the employment of workers who are willing to retain their positions.

As Rothbard pointed out, the demand for skilled workers will increase, while the demand for unskilled workers will decline. The latter will be able to retrain or change their profession. Employment will expand in sectors where automation is more difficult, specifically in labor-intensive sectors. The Services sector which includes education, health care, or legal services is labor-intensive; and indeed, empirical studies confirm a stable increase in the share of services in total employment in the last centuries.

Automation allows us not only to work more closely with other people, but also protects us from tasks that are hazardous to health. It frees us from routine and arduous jobs that destroy our natural creativity. The Modern developed world is a perfect example of this. Instead of digging a pit by shoving our hands repeatedly into the ground, we can use an excavator for the same purpose. Instead of painting each picture in this video manually, we can use animation software programmed by other people. We do not have to record each sound effect separately; instead, sound libraries allow us to create original content faster and more efficiently, so that you may benefit from it more easily.

Thanks to automation, everyone in the economy produces more goods that become more widespread and more comfortable in use. Our jobs become more interesting and more varied than in the past. And although the number of both people and machines on the planet continues to rise, there is no corresponding rise in unemployment, and opportunities for productive and ingenious use of human efforts are plentiful.

The author of the script is Jan Lewiński from University of Wrocław

1 The figure shows a distribution of price elasticity of demand. Demand is elastic for price elasticities in range of (-18.9, -1), and inelastic in range of (0, -1).