[After November emails with Avinash Dixit, I think I’ve found the flaw in my argument below: I assume we are starting with suboptimal income taxes. The argument does hold if we start with a labor income tax but not a capital tax, but it probably falls apart if we start by taxing both. Ivan Werning via X referred me to the 1986 “Gains from trade without lump-sum compensation” by Avinash Dixit and Victor Norman, which in turn is an application of Diamond and Mirrlees (1971). See also Robots, Trade, and Luddism: Sufficient Statistic Approach to Optimal Technology Regulation,”, Arnaud Constinot and Ivan Werning, Rev. Econ. Stud. (2023). ]
I thought of an argument that would justify import tariffs in certain situations. Like almost all economists, I am opposed to tariffs, so this has me a bit bothered. I’ll lay it out, though, and see if I get any comments. The idea is this.
Since America raises the prices of goods made in America by making the workers pay income tax, shouldn’t we also put a tax on imported goods, to equalize the costs and to raise revenue from foreigners?
In America, people pay income tax. This raises pre-tax wages (though not by as much as the tax costs the workers) since employers have to make working more palatable if some of the wages go to government.1 Employers charge higher prices to consumers, since their wage costs have risen. Imports, however, do not pay U.S. income tax. Thus, unless the foreign products pay a tariff, Americans will purchase foreign goods even if the pre-tax cost of American goods is lower. It would be as if imports didn’t have to pay sales tax— the imports would be unduly attractive to American consumers. Thus, America should impose an import tax, a tariff, equivalent to the effect of the income tax on domestic production.
I took Paul Krugman’s course in international trade when I was a student at MIT around 1983, but I don’t recall hearing this argument for tariffs. I expect it’s come up somewhere as an example of the “Theory of the Second Best”, the idea that if there is one distortion in the economy that we can’t remove, we can improve the situation by adding a second distortion to undo it.2 The first distortion here is the income tax; the second is the tariff.
The first thing an economist does with an idea is to try to construct an example in which the idea works. Very often, ideas fail at this first stage. They sound good in words, but when you try to put the words into numbers, you see the flaw. I didn’t see it here, but maybe you will. Let’s look at a small country, one whose exports and imports don’t affect world prices.3 The world prices of the two goods in the economy, widgets and wodgets, are both $10. If there weren’t any international trade or income tax, firms could produce widgets at a cost of $8 and wodgets at a price of $9. With the income tax in place, however, widgets would have a price of $11 and wodgets would have a price of $10.
Note the assumption that the income tax raises the price of widgets by $3 but the price of wodgets by just $1. That’s important. It could be because making widgets requires more labor than making wodgets, and labor is taxed more heavily than capital. Or it could just be that wodget labor is taxed less, perhaps because it is made by workers with lower wages, who are in a lower tax bracket.
Now let’s think what would happen if we open up trade. The country would import widgets, since it can import at the world price of $10 instead of the domestic price of $11. To pay for the imports, the country must export wodgets, which it does at the world price of $10, the same as the domestic price of wodgets.4 Since consumers are now saving $1 per widget compared to when there isn’t any trade, it looks like we have an improvement.
The improvement is imaginary, though. Widgets were paying more taxes than wodgets, so when the country starts to import widgets, income tax revenue falls $3 per widget, even though it rises by $1 per wodget. Consumers gain, but the government loses even more— and since the government has to replace the lost tax revenue, consumers will lose out too, in the end.
Now impose a tariff on widgets of $3. Imported widgets now cost $13, and domestic ones cost only $11, so everyone will switch to domestic widgets. Now that there are no imports, our country won’t need to export wodgets any more. The price of wodgets is $10, just as it was before the tariff, and the price of widgets is $11, since domestic ones are the cheapest. Consumers have lost $1 per widget because of the tariff, but the government has gained $3 per widget, which is bigger. Thus, overall our country is better off with the tariff.5 Also, the tariff could be applied to all imports, not just widgets, since wodgets aren’t imported anyway and having a tariff on them wouldn’t matter.
This argument for tariffs does not depend on what the rate of income tax is in foreign countries. The argument is not that our country should charge a tariff to balance out other countries’ income taxes. If the other countries have zero income tax, the argument works. If they have income taxes higher than our country, it still works. Their taxes will affect what the world prices are, but when we said that the world price of widgets is $10 and of wodgets is also $10, those world prices incorporate all the effect of foreign income taxes.
Nor does the argument depend on other countries not retaliating with tariffs of their own. Since our $3 tariff means we don’t need to export wodgets any more, it doesn’t matter if foreign countries tax our wodget exports.
What the argument does depend on is our income tax having an unequal effect on widgets and wodgets. Otherwise, the tariffs would just make things worse. In fact, if the income tax falls more heavily on goods that are still exported after the tax is imposed, we should probably subsidize exports, though I haven’t worked that case out.
The argument I have just made is a “second-best” argument, second-best in the sense that it starts with an economy that is already distorted, in this case by the existence of an income tax. To say that is second-best is actually misleading, because it is still optimal, because it’s better to have taxes than to not have them. The government may be too large now, but we don’t want to eliminate taxes and government completely.
Still, it’s worth saying what the “first-best” would be here. What would happen if our country had no income tax, but was open to international trade? The world price of widgets and wodgets would be $10 for each. The domestic prices would be $8 for widgets and $9 for wodgets. So it would seem as if our country should export both widgets and wodgets.
In fact, our country should export widgets and import wodgets. This is because of the idea of “comparative advantage”. Even though our country is absolutely better at producing both widgets and wodgets, it is comparatively better at widgets. Thus, it should produce only widgets, and it should export most of them and use some of the export revenue to import wodgets. To see this better, imagine that the country was producing some wodgets domestically, at a cost of $9. If, instead of using $9 to produce wodgets, imported that amount of wodgets for $10, it could do that by shifting the $9 to producing and exporting widgets with a cost only $8 and a price of $11, so the cost saving and the export profit could together allow the country to import more wodgets so it ended up with more wodget consumption that it started.6
This argument for tariffs to undo a distortion in the economy caused by an income tax is not an argument for blanket tariffs. It requires that the distortion created by the income tax have a greater effect on the imported good than on the exported one. Beware of using this as a justification for any particular policy, e.g. Trump’s tariff plan. Whether the tariff helps depends crucially on the relative impact of the income tax on the prices of import versus export goods. In practice, tariffs tend to be imposed not for theoretical reasons that raise national wealth, but for political reasons involving which industries win and which industries lose.
That a tax raises pre-tax wages and reduces after-tax wages is the result of supply and demand. This is basic economics, but take my word for it if you haven’t ever learned it.
"The General Theory of Second Best," Robert Lipsey & Kelvin Lancaster, Review of Economic Studies, 24 (1): 11–32 (1956) doi:10.2307/2296233. See also his reflections of 50 years later, “Reflections on the general theory of second best at its golden jubilee,” International Tax and Public Finance 14: 349–364 (2007) DOI 10.1007/s10797-007-9036- I took a class from Lipsey as an undergraduate while he was visiting for a semester at Yale in 1979. It’s interesting what a small world economics can be.
In trade theory, a “small” country is one which can import more and export more without changing the world price by increasing demand and supply. On the other hand, if a “large” country imports more, the price of the import rises. A large country has “market power”, like a monopoly does, and has to think about how its tariffs move the world price of goods. The income tax argument here would still apply, but with that extra complication.
It is crucial in thinking about import policy to think about its effect on exports too— an effect it will always have. Imports have to balance exports in the long run, so if you reduce imports with tariffs, exports will have to go down too.
To be sure, instead of exporting wodgets to pay for importing widgets, the country could a export ownership of its property or shares of stock. In the long run, though, foreigners would want to get a return on their investment, and that would have to be in the form of our country’s goods. This is one of the complexities of international economics. So let’s put exports of capital aside.
This does assume that government spending is not 100% wasteful. Or, more precisely, that government spending is not more than 67% wasteful, since the government gains $3 compared to the consumer’s loss of $1.
I am not sure that the way I presented the way comparative advantage works in this situation is correct. I’ve obscured the role of the domestic resources used to produce widgets and wodgets by using the dollar costs and by not making clear the amounts of each good produced. I think that is allowable, because we are using dollars to represent the relative cost of different resources used in production, but I wish I could write this more clearly.
For a while I have suspected that implementing tariffs instead of an income tax would be beneficial, so I quite appreciate this write-up. Good analysis, and plenty to think about.