Tuesday, April 12, 2016

Luddites and All That

Scott Sumner responded to the attack on free trade by asking if there will now be an attack on technological advance.   Well, Scott, where have you been?   The attack on "automation" and "robots" has occurred in parallel with the attack on free trade.

I think it is for the exact same reason.   In 2008, there was a deep recession and a very significant increase in unemployment.   The recovery that started in 2009 was very gradual and has only recently resulted in a return to full employment.  Long periods of time with weak labor markets result in the growth of bad economic ideas.   We are now suffering from the consequences of competition of low wage foreign workers, or maybe it is some more reasonable version of "unfair trade practices."   Or it is the long awaited time when labor will be replaced by machines destroying all of our jobs.

A student at Friday's meeting of The Citadel Libertarian Society stated with complete confidence that robots will lead to mass unemployment in the next thirty or forty years.   He explicitly stated that there is only so much that can be produced.

I think the first thing to keep in mind about the introduction of robots, other sorts of automation, or more simply, of labor saving technology is that the total production of goods and services not only can increase, but almost certainly will increase.

When someone states that "soon" robots (or other machinery) will have taken 90% of the "jobs," the baseline assumption should be that the remaining 10% of "jobs" remaining to humans will be enough so that every human who wants a job will have one, and that total output will be approximately 10 times more than was being imagined.  

Of course, "numbers of jobs done by machines" versus "number of jobs available for humans" is probably not the best way to think about growing real income and output due to improvement in technology.    It is fundamentally based upon the fallacy of thinking of the number of jobs as being fixed and each job as providing a flow of income like manna from heaven.   If more of the jobs are done by machines, then there are fewer jobs left for humans.    It is the same fallacy applied when thinking about those low wage Chinese workers taking "our" jobs through imports or the immigrants from central Mexico taking "our" jobs.

If, instead, jobs are recognized as productive activity, then people working provide added output which is what generates their income.   Robots (or other machinery,) or Chinese in export industries, or Mexican immigrants do not stop other people from working to produce output too.

It might seem that the amount that can be produced is limited by demand.   Sure we could produce more shoes, but who would want to buy them?   There only a limited number of jobs needed to produce shoes.   While in truth there are billions of people in this world who would like to have more shoes, there is no doubt that the amount of shoes that can be sold for an amount covering the cost of production of them is limited at any particular time.

But the "cost of production" is an opportunity cost.   The reason "costs" do and should limit the production of shoes (and anything else) is that labor and other resources must be pulled from the production of other valuable goods and services.   The whole problem of costly production is due to a need to use labor and other resources to produce other goods and services.

Further, from a "macro" perspective, an increase in the output generates a matching increase in real income, which allows for an increase in the demand for shoes.   If we consider an increase in shoe output alone, the added income for the entire economy is relatively small and the impact of that on shoe demand is also likely to be small.    But when considering growing output throughout the economy and growing real incomes, the demands for all sorts of goods increases.   For the most part, it is the increase in the "supplies" of other goods and services that results in more demand for shoes.   And the increase in the supply of shoes mostly generates an increase in demand for other goods and services.   

Now, this process most certainly will make shifts of labor from producing some goods rather than other goods necessary if resources are to be allocated to produce what people want to buy most.   This is why technological progress is a source of unemployment.   Creative destruction does destroy some firms and some jobs and even whole industries and vocations.   But it generates expansions in sales and demand for other firms and jobs.  Sometimes new industries and vocations are created.

I think the most likely course over the next thirty or forty years will be much the same.   Higher real incomes but with a need to change jobs sometimes.   Further, those who do need to change jobs may take a temporary loss in real income.   But what they will be able to buy with their new job will gradually be more and better, surpassing where they were before.   And if not for them, their children or grandchildren.    

It is not impossible that the demand for human labor in general will fall off due to technological change.  However, it is easy to be confused.   For example, suppose using robots becomes cheaper than using human workers in some industry.  Simple supply and demand suggests that the reduced demand for human labor results in lower wages.   However, the increased supply of the products due to the fact they are now cheaper to produce implies that these lower wages now have added purchasing power.   If, as is conventional, this is all supposed to be done in real terms, this just shows that the impact of substituting machinery for labor has ambiguous effects on real wages--the goods and services that can be purchased with labor.

An alternative perspective is that real output and real income rises, but improvement in labor saving technology results in a reduction in the share of real income going to labor and an increase in the share of real income that goes to other sorts of resources.  Call that investment income for now.   Since real income has increased and investors get a larger share, then that is an unambiguous improvement for them.   But labor has a smaller share of the bigger pie.   That is ambiguous and could either result in higher or lower real labor incomes.

Further, the very same improved automation--best understood as cheaper robots (or other machinery) increases the demand for complementary labor.   That tends to raise the share of income going to labor and reduce the share going to investment income.   

 Finally, even if the net effect of innovation is to provide equipment that substitutes rather than complements labor, that is only a problem if the demand for human labor falls and the supply of human labor doesn't fall as much or more.   Suppose growing productivity and higher real incomes causes people to work fewer hours per day, days per week, weeks per year, or years during their life.     While the technological progress might allow machinery to be substituted for labor, if the reduction in labor supply is greater than the reduction in labor demand, the result would be an increase in real wages.

Consider a scenario where the demand for labor falls off more rapidly than the supply, and the result is that despite growing total output, the share of that output going to those working falls off so rapidly that real wages fall.   The only way to make a decent living, then, would be to save, accumulate assets, and earn investment income.   In other words, it would be necessary to own robots--perhaps a fraction of one or else several.   (Shares of stock in a robotic operation or bonds funding a robot operation is more realistic.)   If we imagine someone born into such a world with nothing but their own mind and hands, they might have difficulty earning enough to survive, much less save enough to "retire" on investment income.

Aside from there not being many productive tasks for humans in this scenario, it would seem that the robots would also need to be quite expensive compared to their output.   Otherwise, it would be relatively easy to purchase a share of a robot and have a good income.   Also, there would need to be plenty of demand for the output of the robots.   The robot owners would need to have a large demand for some kind of output rather than be nearly satiated for material goods and services.  If not, these robots would have a low opportunity cost for endeavors like producing consumer goods or even building additional robots for the poor still working for a living.

The most plausible "nightmare" scenario is where automation substitutes for human labor but scarce natural resources limit the expansion of aggregate output and income.   Suppose the world depends on depleting fossil fuels.   Those owning claims to these land resources would earn a growing share of income.   Claims to "capital" like the robots wouldn't generate much income nor would human labor.

Nick Rowe pointed out that the development of mechanization in the twentieth century didn't result in draft horses remaining fully employed along with the new fangled tractors.   The employment of horses in production fell off.   His point was that human labor could also become obsolete.

I think that he was correct.  However, it seems to me that the reason horses became unemployed is that the human labor needed to direct them was too expensive--the opportunity cost was too high.   However, another factor and more relevant to a possible world where human labor becomes obsolete would be that the opportunity cost of using the agricultural land with horse powered equipment was too high.

It is not the fabulously wealthy "post scarcity" world that generates the problem.  It is rather a world of scarcity where human labor is not very scarce.   If such a future develops, perhaps a quasi-Georgian approach of sharing out the rents from this natural resource would be the least bad option.

Further, in the happier future, where nuclear fusion or solar power or whatever creates plenty of energy, and nothing else of true significance creates a binding constraint, and if it really was true that there was so little use for human labor that investment income is necessary to thrive, then I would think that bringing children into such a world without providing them with a trust fund would be irresponsible.   Providing your child with a "good education" so that they will get a "good job" would no longer be responsible or reasonable.    As for those irresponsible or unlucky (which would include arriving into this world with insufficient net worth,) perhaps providing tax funded trust funds for their children would be the least bad option.   In our imaginary world of plenty, that should be a small expense.

I don't think that trying to switch to a system of sharing consumption is likely to be desirable, even in such a world.  Nor do I see much point in trying to  make consumption or incomes equal.   Further, replacing markets to allocate resources with central planning would not be necessary and probably not desirable.   Even if information processing technology makes managing an entire world economy manageable or even more efficient than competition, I am not sure that putting human survival in the hands of a single artificial intelligence would be wise.   Having multiple ones compete is probably better.

But most importantly, it would be foolish to make major economic reforms now based upon what might happen in  the distant future.   Human labor remains the most valuable resource, and there are billions of people who are desperately poor.   For many years, the most likely scenario is that technological improvement will raise real wages and bring the majority of people out of poverty and into a decent standard of living.

For generations, people in market economies have put up with the uncertainties of creative destruction with the result that those of us in the developed world are very well off materially by historical and world standards.   Sometimes, some people must start over, and when they do, they are still much better off materially than most people in the developing world or anyone from a century ago.   Continuing with the market system not only will improve the material well being of our descendants, but also will benefit those in the rest of the word who are in desperate need.