Arnold Kling  

Robots and Comparative Advantage

Consensus Macroeconomics?... Electricity Economics...

Using the theory of comparative advantage, James Miller explains why robots cannot replace humans completely.

Now, assume that in our simple wine/cake world robots begin large-scale manufacturing. They could easily change the relative prices of wine and cake. Perhaps if they just made cake, cake would become cheap and so you could buy 1000 cakes for just one bottle of wine. In such a cheap cake world, humans could profitably make wine. Alternatively, robots might just produce oceans of wine and therefore make cake relatively more valuable than wine, creating incentives for humans to bake cakes. It must always be profitable, however, for humans to make at least one of these goods.

For Discussion. Does re-writing the paragraph substituting "cheap Asian labor" for "robots" and "American workers" for "humans" change the analysis?

COMMENTS (6 to date)
Eric Krieg writes:

Question. Is Crusoe eating the cake and Friday drinking the wine, or vice versa?

Hi Arnold,

You know, this is a silly example -- and not because its a 'toy' world (only two goods). It's silly because it implies that robots can expand production without incurring higher costs. I don't mean anything as sophisticated as decreasing returns to scale but just opportunity costs.

For the two-actor/two-good example to work there are a couple of unstated conditions that don't (as far as I know) apply to robots. First, both actors in the economy (robots, people) must incur an opportunity cost in devoting themselves to the expanded production of just one good. The more of one good they produce the less they can produce of the other. But robots, as far as I can see from the example, can expand production of either good at will -- apparently without any opportunity cost to themselves.

This brings us to the second unstated condition of two-actor/two-good example. Both actors must DESIRE both goods. Its possible that robots incur no opportunity cost of producing a mountain of cakes because they really don't desire wine at all. Or they can produce a lake-full of wine because cakelessness is no problem for them. (Of course, I'm ignoring ENERGY costs etc. outside this toy example; that's likely to be a problem even for robots.)

If robots produce tons of cakes, the cake-price of a bottle of wine falls only because there is now more cake in the world than is required at the old cake/wine price. If robots don't desire wine, but only cakes, there won't be any cake/wine exchange in which to establish this exchange price because robots will never go to the market. The two-actor/two-price world collapses.

So where does this leave us? Well it leaves me wondering why we built these stupid robots in the first place. We would have been better to spend our money on a few nice lunches.

Best wishes,


Tom Dougherty writes:

A robot trading with humans is utterly absurd. The reason why one person exchanges with another person is because each values what the other has more highly. If I have one unit of cake and another person has one unit of wine, we will only trade if I value his wine more than my cake and he values my cake more than his wine. Humans value things; robots do not. Since robots do not value wine or cake, why would a robot engage in an exchange if it did not benefit from the exchange?

One could use robots as a metaphor for capital or machinery. In this case, humans would be using robots (machines) to produce cake in order to exchange it with other humans for wine. Instead of substituting “cheap Asian labor” for robots and “American workers” for humans, wouldn’t it be more reasonable to substitute “highly productive, capital intensive, American labor” for robots and “unproductive, labor intensive, Asian labor” for human? Or would it better to come up with a brand new article altogether?

Matt Young writes:

Whoops! Europe has a government that eats 55% of its economy, and they have lost 1/4 of their future population, and they are not done yet. The history of massive socialism may have wiped out some Eastern European nations.

So the answer is, it is sometimes more profitable for humans to disappear entirely, hence making anything with robots a useless endeavor.

Peter Gallagher writes:

Hi Arnold,

On reflection, I think I may not have been fair in my criticism of the Robot/Human story. I was thinking of it as the classic two-country/two-good illustration of comparative advantage in trade and the gains from trade. From that point of view it's easy to make fun of its limits.

If, however, you assume that the robots have an un-valued opportunity cost in the production of either good (making more cakes means they make less wine because they don't have the time to make both in the same proportions) the story does, after all, illustrate a point that James Miller made in his article.

Even if the robots desire neither cakes nor wine (and therefore put no value on their own output or costs), WE become much wealthier wine producers when the robots make a mountain of cakes because our OWN prices change (assuming we have access to the robots' output without trade: they being, in this example, simply uninterested in exchange).

Or, if the robots produce wine, we become much wealthier cake producers when the wine-price of our cake output rises dramatically in OUR estimation. In either case, as Miller points out, the example illustrates that robot production doesn't displace humans but potentially enriches humans.

I made the mistake of seeing Miller's example as an illustration of the gains from exchange when, in fact, there is a much simpler point to his story.

My apologies to Miller and best wishes to you.


Jason Ditz writes:

Doesn't this whole argument ignore the demand side of the equation? If these robots are capable of creating all the demanded goods in the economy at an excess of what the demand is, isn't the fact that some are more in excess than others kind of meaningless?

The human race can only drink so much wine and eat so much cake.

It seems like raw materials would be the desirable commodity to have in the event of a large scale robot production binge. Gold, silver and the like.

No matter how good the robots are, they still can't create gold out of thin air.

Comments for this entry have been closed
Return to top