Let’s imagine a near future in which automation produces intractable unemployment on a massive scale—say, 25% of the potential workforce. Imagine further that this goes on for a long time, that the usual unemployment benefits are merely temporary, and that we have a social crisis on our hands. We will need to provide everyone who is displaced with incomes large enough for a decent standard of living. In an earlier blog post I calculated that providing this for 25% of the workforce would cost roughly 1.5 trillion dollars a year, or just over 5% of America’s total GDP.
Where should we go to find that money? The easy answer is simply to raise it through taxes, but that isn’t as straightforward as it sounds. Total federal spending in the United States in 2025 was just over 7 trillion dollars, or 23% of GDP. If we want to raise that by 5% of total GDP to support the displaced, we’re now talking about raising federal tax revenues by over 21%–a significant increase.
With that in mind, I want to suggest a more specific (though still expensive) answer. Automation-induced unemployment is a problem that contains its own solution. To see how, we need to understand the difference between automation and displacement. Automation is the substitution of machines and software for human labor. Automation eliminates jobs and always has (that’s the point), but throughout history new jobs have eventually arrived to replace those lost to automation. Displacement happens when new jobs do not arrive. This can be either temporary, when there’s a lag between old jobs disappearing and new jobs arriving, or permanent, if new jobs never arrive in sufficient numbers.
Displacement produces a windfall equal in size to the total incomes the displaced used to receive
When an economy becomes automated in ways that produce displacement (that is, new jobs are not arriving to employ the displaced), that economy will—all else being equal—be producing the same GDP with a smaller percentage of potential workers on the job. That’s just what it means to automate jobs out of existence: it now takes fewer people to produce the same total economic output. That, in turn, means that the total labor cost in that economy is now smaller. If, for example, it now takes only 75% of the potential workforce to generate the same total output, then the total payroll is (again, all else being equal) 25% less than before.
Now, this won’t happen if new jobs do arrive to employ the displaced; we are talking about a situation where jobs go away and no new jobs arrive to replace them. That’s displacement. There’s a difference between automation eliminating jobs but new jobs arrive to replace them, and automation eliminating jobs and new jobs do not arrive. The latter situation is displacement.
However, if GDP is undiminished, then the money that used to pay the 25% of people who are now displaced is still out there somewhere in the economy. Most or all of it will almost certainly land in the hands of capital owners, or some of them. Those who receive some of that money have received a displacement windfall. (I’ll say more about this in a future blog post, but this windfall won’t always appear in the form of a smaller payroll. We’ll also see that not all firms receive part of it, nor do they receive it in the same percentage relative to their size.)
Since we’re talking about the money that used to pay wages, salaries, and other compensation, the displacement windfall is equal in size to the total wages, salaries, etc. that would be paid to the unemployed 25% of the workforce if they had their old jobs.
Notice what this means. We could simply give that money to the unemployed 25% so that they have their old incomes and live as well as before. Of course, that isn’t quite fair to those who must still work, and in a future post I’ll talk about what we can do about that, but the point for now is that the money the unemployed have lost is still around.
And there’s a lot of it. The higher the rate of displacement, the more money we have to work with. If automation displaced everyone and the rate of displacement were 100%, the displacement windfall would be equal to all the incomes everyone received when they were working. When I say that, so far as supporting the unemployed is concerned, automation is a problem that contains it’s own solution, this is what I mean. When automation displaces people, it creates a displacement windfall, and that windfall is the solution that automation contains.
So far, simple, but there’s an obvious objection. Aren’t we overlooking the cost of all that automation? When the economy was automated to the point of displacing 25% of the potential workforce, there was a large increase in the amount of automation deployed in the economy, and all that automated equipment and software costs money. If a firm, for example, buys enough automation to lay off 25% of its workers, it must also pay for the equipment that replaces them. The labor savings that firm enjoys is the total incomes those 25% used to receive, minus the cost of all the automated systems.’
If this is true at the level of an individual firm, won’t it be true at the level of the entire economy? If so, then the displacement windfall isn’t equal in size to the total incomes of the displaced, it’s equal in size to the total incomes minus the cost of all that new automation. We seem to have much less money to work with when it comes to supporting the displaced.
The answer to this is that what’s true at the level of an individual business firm is not true at the level of the economy as a whole, or at least not in this instance. At the level of the economy as a whole, there is no overhead expense for automation that doesn’t ultimately go to people, either as incomes or as asset appreciation. When a firm buys some robots or AI software, that money leaves the firm and does to the firm that made the robots or provided the software. However, the revenues going into that firm ultimately go to pay incomes to that firm’s workers and management, other kinds of compensation to management and owners of the firm (stock appreciation, if nothing else), and all the other expenses of the firm: inventory purchases, whatever services the firm needs, additional capital for that firm, and so on. However, each time the robotics firm spends money on something other than compensating it’s human workers, managers, and owners, that money goes to other firms and the cycle begins again.
In the end, all overhead ultimately goes to people, so at the level of the economy as a whole, the displacement windfall is not reduced by the cost of the automated systems. The windfall is equal in size to all the incomes lost to displacement, with no deduction for overhead.
We can distribute the entire displacement windfall without reducing anyone’s profits or income!
Another thing follows from this. We can take the displacement windfall away from whoever has it without leaving them with less money than they would have in an economy where no displacement is going on. This is because the displacement windfall is an extra benefit for capital owners (they will get most of it), on top of the incomes, profits, rents, return on investments, share and capital appreciation, and all the other sources of money going to that group. Let me show you how this works. First, consider consider an economy with no displacement:
$100 Billion GDP with No Displacement
$10 billion: paid to capital owners and the remaining workers
$30 billion paid for overhead other than profits for the owners of capital
$60 billion paid in wages and salaries
Now consider an economy with 25% displacement:
$100 Billion GDP with 25% Displacement
$10 billion: paid to capital owners and the remaining workers
$30 billion paid for overhead other than profits for the owners of capital
$45 billion paid in wages and salaries to those who still have jobs
$15 billion that used to be paid to the displaced (25% of $60 billion)
Notice that the capital owners and remaining workers are still getting as much in the displacement economy as they got in the economy without displacement. The windfall comes from the smaller payroll, which used to be $60 billion but is now $45 billion because the workforce is 25% smaller.
We can take the displacement windfall while leaving capital owners and remaining workers as well compensated as they would be in the absence of displacement.
But what if the GDP per capita is growing; shouldn’t capital owners and those who still work get higher incomes than before, now that the economy is more productive per worker? Yes. They should, and they still can. Suppose, for example, that per capital GDP is now 10% higher, and that we give the owners and workers 10% more compensation each so that they can enjoy the benefits of increased productivity. This is fine, for the displacement windfall itself will also be 10% greater, for the displaced workers would be enjoying a 10% increase in incomes if they had their old jobs. Just run the math above with an initial figure of $110 billion instread of $100 billion, and increase everything below that line by 10%.
So capital owners and the still-employed are not locked into whatever incomes were when displacement began; their boats can rise with a rising tide of productivity brought on by the very automation that brought on displacement and the windfall.
The moral of this story
No matter how high the rate of displacement gets, there is a windfall equal in size to what the displaced would be earning if they had their old jobs, and we can distribute that windfall without reducing the profits the capitalists would get if displacement were not happening, and without reducing the incomes of anyone who still has work.
But that raises another question: If we’re not taxing in a way that reduces anyone’s income or leaves capitalists with less profit or return than they would get in a world without displacement, where, then do we go to get that money? This seems too good to be true. To answer that, we need to think about the right tax base for collecting the displacement windfall. I’ll talk about that in my next blog.