Do protective tariffs boost growth?

An empirical investigation

In the cauldron of World War II, American elites, led by Cordell Hull, came to believe that the geopolitical catastrophes of the 1930s and 1940s were due, at least in part, to the great inward turn of the capitalist powers. Specifically, the thesis—as correct now as it was back then—was that the tariff walls erected by the satisfied powers, America and England, during the Great Depression, forced the dissatisfied powers, the Soviet Union, Germany, Italy, and Japan, to seek economic security through conquest and autarky. A deep and stable consensus emerged from that diagnosis. From 1943 to 2016, the United States waged relentless global war against economic protectionism. In particular, the US, through its dominance of multilateral institutions, and through its bilateral foreign economic policies, pushed the rest of the world to lower trade barriers.

This US-led effort was extremely successful. Kick and scream they did, but eventually, most nations saw the light and increasingly gave up tariff and non-tariff protection. By 2004, the (GDP-weighted) global average effective tariff rate had fallen below one percent.

The working class revolution that Trump rode to power broke the seventy year elite consensus on free trade. Trump alleged that the decline of the American working class was due to unfair competition by China and other US trade partners. Deindustrialization due to Chinese import competition was blamed for the decline of the working class family, despite the fact that the decline of industrial employment largely preceded China’s rise as a trading nation; and the fact that industrial employment has declined across the advanced economies, none of whom show anything like the US’s morbid symptoms of working class decline.

A growing number of American elites have bought Trump’s hypothesis. At the informed end of the spectrum, this way of thinking was exacerbated by a widespread misreading of the Autor et al. so-called “China Shock” paper. The effect they document in the cross-section of US commuting zones does not add up to even a tenth of industrial jobs lost—nor do they add back the industrial jobs gained due to exports to China (once you add that back, the net job loss due to import-competition from China becomes trivial). At any rate, whether due to genuine belief in the false theory or because they think that the political price for defending free trade is now prohibitively high, both major political parties are now prepared to ramp up tariff protection for American industry.

The signature measure adopted by the Biden administration is the 100% tariffs imposed on Chinese EVs imports and 25% on batteries. The measures were justified by the fraudulent claim that Chinese cost competitiveness in these industries was due to massive state subsidies rather than genuine innovation-driven spurt of productivity growth. Just these two measures virtually guarantee that the energy transition in the US will be substantially delayed and cost far more than it would otherwise have. And, worryingly, it looks increasingly like we’re at the beginning of the revolution in US trade policy.

Even otherwise well-informed people seem to be suffering under the illusion that tariffs helped countries to industrialize, that they can help boost growth rates. When I suggested on Twitter that “There is no evidence that trade protectionism helped the US or anyone else to industrialize. The meager evidence that exists is against the thesis.” Yuen Yuen Ang, Alfred Chandler Chair Professor of Political Economy at Johns Hopkins, responded with “Vestiges from the last century. Relentless denialism.”

I can’t influence a lot of people. But perhaps I can influence some elites at the informed end of the spectrum. That a professor at Johns Hopkins, a named chair no less, thinks there is evidence for the protectionist case suggests to me that some sort of serious push-back is in order.

---

Ultimately, it is an empirical question whether tariffs boost growth rates. Here we investigate whether higher tariffs have historically been associated with higher growth rates. We obtain panel data from the CEPII’s TRADHIST dataset, that “includes about 1.9 million bilateral trade flows, 42,000 observations on total imports and exports, and 14,000 observations for GDP and exchange rates, for a 188-year period spanning 1827–2014.”

We construct two pairs of measures of tariff intensity. We construct average tariffs faced-by, and imposed-by a country. And the averaging is either weighted by partner GDP, or, in our preferred specification, by the inverse of sea distance to the partner. The second is more consistent with the standard gravity model of trade that posits that neighbors are likely to trade more than countries located far away. There are 1,429 observations in our panel that includes 183 years and 11 countries. Our response is per capita income growth. We exclude outliers from our analysis, but note that their inclusion does not change our results qualitatively or quantitatively. We standardize all variables so that the parameter estimates are in standard deviation units. The baseline model is estimated with OLS. We also report the results of panel OLS where we admit country fixed effects.

We estimate regressions that model growth as a linear function of (1) average tariff rate applied to partners by ego, (2) average tariff rate faced by ego, and (3) average growth rate of trade partners. The main parameter of interest is the slope of (1). If the protectionist thesis holds water, this coefficient should be large and significant. If the coefficient is small and insignificant, the protectionist hypothesis is refuted.

Here’s our estimate of the baseline model using GDP weights. We find no evidence that tariff protection boosts growth rates. The coefficient is not even marginally significant (b = 0.003, P = 0.64). However, we do find that countries facing high tariffs see slower growth. The coefficient is small but significant (b = -0.020, P < 0.001). Finally, we find very strong evidence that trade partners’ growth rates give a positive boost to home country growth. The coefficient is large and highly significant (b = 0.095, P < 0.001).

We now admit country fixed effects. This controls for all time-invariant country characteristics, so the parameter estimates are expected to be more reliable and precise. The estimates remain virtually unchanged. And the slope coefficient of interest—the gradient of average tariff rates applied to trade partners—falls further into insignificance (P = 0.803).

Using inverse sea distance weights yields a much better fit, consistent with well-known results from the gravity model. The slope coefficient estimates for the tariff intensity variables do not change much. In particular, the slope coefficient of interest still vanishes (P = 0.390). However, we now find a much stronger effect of trade partners’ growth rates on home country growth (b = 0.324, P < 0.001).

Finally, we find that admitting country-fixed effects does not change results. Average tariff rates applied on trade partners continue to have zero effect on home country growth; there’s a small but significant effect of trade partners’ tariff rates on home country growth rates; and there’s a very large and highly significant effect of trade partners’ growth rate on home country growth rates.

We have documented a very robust pattern in the historical data. There is no evidence that tariff protection helps boost growth, but there is some evidence that tariffs hurt other countries at the margin. So, as I said on Twitter, there’s no evidence for Chang’s hypothesis that protective tariffs have historically helped countries industrialize and grow (which amounts to the same thing for the period under investigation). What is clear is that growth has a strong tendency to spill over. We should like our trade partners to grow faster because that gives a big boost to our own economic performance.

---

As I’ve argued at length before, US China policy is fundamentally misguided. American political elites have blown the threat posed by China all out of proportion to justify hawkish policies that guarantee that China will be a committed adversary, without doing the math on whether we can prevail in a protracted struggle against the combined industrial might and geopolitical influence of the three power alliance we have foolishly forced into existence.

But the problem with the wrong turn on trade goes well beyond the geopolitical risks inherent in great power confrontation. By sheltering US firms behind a tariff wall, we are condemning them to being fat, lazy and incompetent; by imposing tariffs on crucial intermediates, we’re virtually making certain that they will never again be able to compete on the world market. Biden’s tariffs have already made it prohibitively expensive for Americans to ditch their gas guzzlers for EVs. And tariffs coming down the pipeline are equally certain to make Americans poorer and less well off. Last but not the least, by hunkering down behind a great big tariff wall, we’re quite literally ceding the world economy to China.

American elites need to come to terms with the fact that we no longer enjoy the vast margin for error we did back in the unipolar world. In fact, conditional on the confrontation with the triple alliance, there’s no margin for error at all—the cost of US policy errors has risen quite dramatically. We simply cannot afford this wrong turn on trade. Don’t complain later after we get our ass handed to us.