Comparative advantage
Two people both gain from trading even when one of them is better at everything.
The claim
This is the least intuitive true thing in economics, and the one most often explained wrongly.
Take a surgeon who also happens to type faster than any secretary in the city. Should she type her own letters? No — and not because she is bad at typing. Because every hour she spends typing is an hour not spent in surgery. The secretary is worse at typing in absolute terms and still should do the typing, because what he gives up to type is far less.
What matters is not who is better. It is who gives up less. That is comparative advantage, and David Ricardo set it out in 1817 using English cloth and Portuguese wine.
The conclusion is sharp: a country that is worse at producing everything still has something to gain from trade, and still has something to offer. There is no such thing as being uncompetitive at everything.
The assumption underneath
Several, and they are load-bearing.
Resources must be able to move. The argument says the country as a whole gains — it says nothing about the textile worker whose factory closed. Gains are averages, and averages have people inside them.
Costs must stay roughly constant as you specialise, and the technology must be given rather than shaped by what you produce. Both are questionable. If making semiconductors teaches you how to make better semiconductors, then today’s comparative advantage is partly a choice about tomorrow’s, which is the serious argument behind industrial policy.
And trade must actually be two-way. The theorem compares trade with no trade. It does not promise anything about who captures the gains.
Where it stops working
The honest reading is that comparative advantage proves there are gains available from trade, and says almost nothing about how they get divided. Most arguments about trade policy are arguments about division. Quoting Ricardo at them answers a question nobody asked.
The one thing to keep
“We are worse at this than they are” is not an argument against doing it. The question is always: worse compared with what else we could be doing instead?