Trump is back on doing tariffs, signing a 25% tariff on imported steel. It is worthwhile to ask, what happened the last time we tried that? In 2018, massive tariffs were imposed on many goods, including steel. Average tariffs increased from 3.7% to 25.8%. The fraction of GDP which was taxed — about 2.6% — was twice as large as under the notorious Smoot-Hawley tariffs. The results were extremely poor – Amiti, Redding, and Weinstein (2019) found that it cost us $1.4 billion per month in pure deadweight losses. The trade war did not lead to negotiating a “fair trade” with lower tariffs, but to a permanent elevation in tariff rates. The tariffs were regressive, and the impact on utility greater than it would appear due to falling on lower-income consumers. The cost of the tariff was passed through almost entirely onto the consumers, with the US bearing the majority of the cost. The tariffs were not targeted in any rational manner, and there is no credible argument that they were made to protect infant industries which might one day grow to be world-competitive. Instead, exemptions were made on the grounds of political connections, with firms which donated to Republicans being more likely to receive exemptions.
These tariffs are extremely regressive, although finding the full impact is controversial. This is because there is an impact on consumption, and an impact on the labor market, and these are negatively correlated. Poorer consumers spend considerably more of their income on traded goods, and their marginal utility from buying more goods is likely to be higher. (Making this break from a representative agent model is explored by Waugh (2023), and leads to gains from trade being three times larger, and four and a half times larger for poor households). Within narrowly defined categories, tariffs on lower-quality consumer goods tend to be substantially higher than on higher quality goods. Acosta and Cox (2024) estimate that a doubling in the value per unit among consumer goods is associated with a 1 percentage point reduction in the ad valorem tariff rate. (This pattern began in the 1930s with the Smoot-Hawley tariff, but have persisted since due to the US government generally bringing down tariffs on all varieties, rather than adjusting the relative rates within varieties.) Bringing down the tariffs for low-quality goods to be in line with high-quality goods in the same categories would substantially improve the welfare of the poor. Russ, Shambaugh, and Furman (2017) go through consumption by income deciles and match it up to the products which people buy. The lowest deciles pay far more as a fraction of their income than do the rich. (Note that this is ordered by pre-tax deciles, but is the fraction of post-tax income, so that there isn’t anything weird coming people with no earnings.
)
On the other hand, the changes in the labor market benefit poorer people more. Borusyak and Jaravel (2023) estimate that 99% of changes in income occur within income deciles, rather than across. I am willing to believe, however, that for temporary shocks the cost is primarily regressive. Employment adjusts slower than consumption. In addition, the losses from uncertainty are pure waste. Domestic producers who might benefit from the tariffs are also uncertain, and will avoid making investments for the same reasons as affected users of the input. People need to make big investments and plan for the future, and they can’t do this if they face large and unexpected taxes.
Similar tariffs in the past had similar results. In 2003, George Bush imposed tariffs on steel, which is an input into many other goods, rather than consumed on its own. For every 1 job in the steel industry, Cox and Russ (2020) estimate that there are 80 jobs which use steel as an input. Lydia Cox (2025) found that not only did the steel tariffs have a profoundly negative impact on employment, but that the negative impacts lingered due to uncertainty over future policy. It would be better to have a constant tariff, than to unexpectedly flip between different tariff levels.
These losses are compounded by them often falling on intermediate goods. Modern trade often takes the form of value chains, where the same product in different states of development might repeatedly cross boundaries. Goods are generally not “made in” any particular country at all, but are made out of lots of different pieces manufactured in different countries. The automotive industry is the strongest example of this. An engine might be assembled out of pistons made in one country and spark plugs made in another, and the whole thing shipped to Tennessee to be assembled into a car sold in Mexico. Placing a tariff on each intermediate step reduces the quantity made, and causes larger distortions than a tax on the final product alone. In many cases we don’t even really know how valuable a thing is at all, leaving inspectors guessing and the level of taxation unpredictable. How are you supposed to price a specialized variety of a good, manufactured by one part of a company to be used by another part?
The editorial throughline of this substack is pretty clear. I believe that tariffs are really bad, and that specialization is good. I believe that, in the long run, supply curves slope downward, and that an increase in the population trading in a market will increase variety and reduce prices. I also believe that we should learn from experience, and not try insane things over and over again. The arguments for tariffs advanced by the Trump administration are totally divorced from any rational economic argument for them.
What hope do we have? I do not think that I have any influence over the Trump administration. What I hope I might have is some influence over what is acceptable and not acceptable amongst us liberal types. The past administration was rather receptive to tariffs and to big labor, and continued the impoverishment of America started by Trump. We can do better than him. “Tariff” should become an infamous word, totally unthinkable. Better to give it up entirely, than leave the grenade in our hands.
I agree that most of Trump tariffs are deeply misguided, especially those on our neighbors and allies. That said, we do need to decouple from the Chinese economy now that Xi had gone full-dictator and is committed to both expanding China's borders and re-drawing global rules to favor oppressive dictatorships. What else can we do to decouple before the People's Republic decides that Bhutan has always been part of China, like Taiwan, and Mongolia, and next will be the Korean peninsula, and we know such regimes never stop until someone stops them.
So what are better tools than tariffs?
Tariffs, like sanctions, are economic means to political ends. A (trade) war is simply politics by other means. This is probably especially true of the ones directed towards our neighbors on the North American continent.