I won’t bury the lede. I’m very skeptical that tariffs spur innovation. That’s not my reading of the economic theory and empirical research. But it’s not impossible and the research isn’t conclusive. Let’s take the idea and the empirics seriously.

The idea often goes something like this. Necessity is the mother of invention. This folk wisdom has a powerful grip. When politicians propose tariffs on Chinese semiconductors or European steel, they often invoke a simple logic: make foreign goods expensive, and domestic companies will innovate their way to competitiveness. Cut off cheap imports, and necessity will spark the…. something something… needed to build better products and processes at home.

This isn’t just political rhetoric. The infant industry argument in economics has a long intellectual pedigree, suggesting that temporary protection protects nascent industries develop the capabilities needed to eventually compete globally.

But overall, I think we should be skeptical of the idea that tariffs will spur innovation.

The first issue is that it’s not clear why this “necessity” would dominate. Surely, increased competition from foreign companies introduces new necessities. Improve or go out of business seems like a necessity. Would that effect dominate? It’s an empirical question. Second, we need to disentangle genuine innovation—things that move out of the set of possible things made—from adaptation just to deal with a problem.

If I cut off my arm, I will probably learn lots of new ways to do household chores. I don’t think that’s what we mean by innovation. It’s certainly not the path to actual increased productivity.

The mixed historical record

The idea that making things hard increases innovation comes up all over the place. “Minimum wages are good because they increase automation” is one common one. While it may seem contradictory to “raise costs to lower costs,” there are ways it can work, and historical examples suggest it could be possible.

Without question, changes in relative prices can direct the path of innovation. In plain terms, if something becomes more expensive, people find ways to use less of it. If you start using less, through some mechanism such as learning by doing, maybe through time, you can use even less. That’s how economic growth occurs.

This induced innovation idea shows up all the time in discussions about growth and technology discourse. If carbon is taxed (raising the cost of dirty energy), firms innovate in clean tech. This has a different rationale related to externalities, but the pure innovation story makes sense. By analogy, a tariff that raises the cost of an input or makes a certain import scarce could encourage firms to develop new techniques or substitute products. History provides a famous illustration that isn’t directly about tariffs but illustrates the possibility that necessity is the mother of invention.

According to Robert Allen, England had expensive labor in the 1700s relative to capital and fuel. This high-wage economy created an incentive to invent and adopt labor-saving machines, which was a key reason the Industrial Revolution took off in England first. In Allen’s induced-innovation view, English entrepreneurs mechanized tasks like spinning and iron-making to save on costly workers, leveraging England’s cheap coal instead. Indeed, the British iron industry’s breakthrough – using coke (from coal) instead of charcoal to smelt iron – can be seen in this light. Switching to coke dramatically lowered the fuel cost of producing iron, an innovation driven by the scarcity (and rising price) of wood. The broader point is that when some necessity becomes acute or an input becomes dear, innovators might respond with new solutions. Other historians (I’m thinking of Anton Howes, in particular) stress different causes for England’s take-off related to ideas instead of wages, but Allen puts forward a plausible story that maybe applies more generally about necessity and invention.

There are other interesting examples that are closer to tariffs that speak to necessity and invention. During the Napoleonic Wars, Napoleon imposed the Continental Blockade (1806) to shut British goods out of Europe. This was essentially a forced trade embargo (let’s call it an infinite tariff) across much of Continental Europe against the era’s leading industrial exporter (Britain). While the blockade caused plenty of economic pain, it also created an unintended laboratory to test “necessity is the mother of invention.” A paper by Réka Juhász studies this episode. Regions of the French Empire that were more effectively cut off from British imports saw a surge in mechanized cotton spinning capacity relative to regions that remained more exposed.

In other words, when French markets couldn’t get cheap British textiles, local entrepreneurs responded by adopting the new technology of the Industrial Revolution – building water-powered spinning mills to replace British imports. These protected regions didn’t just temporarily boost output; they apparently established a lasting industrial base. Juhász finds that the areas with exogenously higher protection during the blockade had higher industrial value-added per capita for decades afterward (through the mid-19th century). The blockade, though lifted in 1815, gave French cotton textile producers a head-start to learn and improve such that some of them remained competitive even after trade resumed. This is a rare clear-cut case where temporary protection led to technological adoption that persisted.

Did the China shock lower innovation?

So far, we have a grand theory of the industrial revolution and a particular example related to war. Interesting, but it’s hard to say how much they tell us about any policy on the table today.

The best case that tariffs (a way of raising trade costs) today would increase innovation comes from looking at what happened when technological and policy trade barriers were lowered with China. Autor, Dorn, Hanson, Pisano, and Shu have a study looking at US manufacturing and the effects of rising import competition from the “China Shock.” Manufacturing is important to understand in this context. Oren Cass has argued, “Manufacturing drives innovation.” From an accounting perspective, that is definitely true. Manufacturing makes up more than three-quarters of US corporate patents, but let’s never reason from an accounting identity.

Autor et al find that increased competition reduced patent production in affected sectors. Firms facing greater import pressure cut R&D expenditure and filed fewer patent applications. The effects were concentrated among less profitable and less capital-intensive firms—exactly the companies that protection advocates claim would benefit most from breathing room. It’s not exactly a tariff experiment but it’s plausible that if increased trade with China hurt R&D, then decreasing trade with China in the future would help.

One thing to note is that the innovation decline was roughly proportional to the overall contraction these firms experienced. As Chinese competition intensified, US manufacturers cut sales, employment, capital investment, and R&D by similar amounts. This suggests the innovation response wasn’t strategic pivoting toward new technologies, but rather just a broad retreat as firms struggled to compete. That doesn’t mean it wouldn’t be good to reverse it and have more marginal American companies surviving, but that’s not the impression I get when people say tariffs will increase innovation.

Product competition misses half the story

The problem with focusing solely on whether import competition spurs innovation is that it only looks at one piece of the modern economy. As I explained in a different newsletter on tariffs, the clean textbook division between domestic producers (who benefit from protection) and consumers (who pay higher prices) doesn’t reflect how trade actually works today.

A large part of international trade occurs in intermediate inputs and components, not finished products ready for consumers. Companies import parts, materials, and unfinished goods to use in their own production processes. This means tariffs often act more like a tax on domestic manufacturers who use these imports, raising their costs instead of protecting them from foreign competition.

Autor et al. are careful to isolate product-market competition coming from the China shock; they deliberately do not try to parse out an “input-cost” channel. Everything in their empirical design—the definition of the shock, instrument, and the way they map it to firms—treats rising Chinese imports as an external rival that erodes sales and markups, not as a cheaper source of parts that could lower costs. They basically ask, When Chinese rivals lure away your customers, do you innovate more or less?

But that whole framing misses the innovation effects (positive or negative) when American manufacturers can’t access the intermediate goods, components, and machinery they need to innovate. Suppose necessity is the mother of invention in the way the blockade paper or Robert Allen’s work suggests. In that case, it is the companies that now can’t import things as easily that should be seeing the innovation. They should be the ones learning how to DIY and increasing their own productive capacity.

This limitation in the Autor et al. approach helps explain why other researchers, using different methodologies that capture both sides of trade, have reached strikingly different conclusions. Studies that examine the overall innovation effects of trade—rather than isolating just the product competition channel—consistently find positive effects on technological upgrading.

When researchers can observe both the competitive pressure from foreign products AND firms’ access to foreign inputs and export opportunities, a more complete picture emerges. Bloom, Draca, and Van Reenen’s 2016 study of publicly traded European firms is one example. It doesn’t just include manufacturing, which is great. Unfortunately, it only looks at publicly traded firms. Most important for our discussion is the improvement in that, rather than treating Chinese imports purely as competitive threats, they examine how exposure to Chinese trade affects innovation across multiple dimensions.

They measure innovation much more broadly. While Autor et al. focus primarily on patents, Bloom et al. examine patents, IT intensity, total factor productivity, R&D expenditure, and management practices. This matters because different types of technological upgrading might respond differently to trade pressure—a firm might not file more patents but could still become more productive through better IT systems or management.

Most importantly, their empirical design doesn’t separate input and output market effects. They explicitly measure offshoring to China using input-output tables to capture how much each industry relies on imports from China as production inputs. They can therefore distinguish between the competitive threat of Chinese imports and the productivity benefits of accessing Chinese inputs (or export) markets.

European firms facing greater Chinese import competition increased patenting, IT intensity, and total factor productivity. The study found that trade shocks accounted for 14-15% of European technology upgrading between 2000 and 2007. But here’s the crucial insight: the innovation response was strongest among firms that simultaneously had good access to international inputs and export markets. Competition alone wasn’t enough; firms needed access to the global economy to innovate effectively.

Another paper by Aghion et al. looks at the China shock and French firms. They look at customs and patenting data at the firm level. Firms that have to compete with Chinese goods in the output market see a drop in sales, employment, and innovation. This is especially large among low-productivity firms. That matches the Autor et al result. At the same time, firms that are primarily importers have the opposite outcome. Innovation increases from the China shock, especially for low-productivity firms.

Models can help us think through the effects

When thinking about international trade, these results shouldn’t be seen in isolation. There is a huge literature on international trade and its overall effects on things like competition, product variety, and productivity. I’ve talked about this literature before and why we need to understand it.

In modern trade, there are lots of moving parts. Akcigit and Melitz have a handbook chapter trying to connect trade and innovation through multiple channels. Their model has 1) the market size effect (access to export markets increases innovation incentives), 2) the business-stealing effect (import competition can reduce innovation by shortening monopoly duration), and 3) an “escape competition” effect where firms facing foreign competition may increase innovation to differentiate themselves.

Firms jointly choose whether to innovate and export, with more productive firms self-selecting into both activities. Trade liberalization doesn’t just affect today’s competition; it reshapes innovation incentives with long-lasting dynamic consequences. For export liberalization, the effect is clear: more innovation through expanded market access. But for import liberalization, the outcome depends on industry structure and competitive dynamics—sometimes spurring innovation through escape competition, sometimes dampening it through business-stealing.

By thinking through these mechanisms explicitly, we can sort different results into different bins. They summarize the literature saying: “Increased export market access—the impact on innovation is unambiguously positive across those studies,” while “the impact of increased competition from imports are a bit more ambiguous.”

Akcigit has a forthcoming paper with Ates and Impullitti that pulls even more pieces together. Think of two large economies locked in a Schumpeterian race: firms invest in R&D to leapfrog each other, trade costs shape who sells where, and governments can pull two levers—import tariffs or R&D subsidies. Because the model tracks the entire transition path, not just the long-run steady state, it lets them ask the politically relevant question: what happens next quarter, next decade, and beyond?

As they wrote in a summary of the paper, “import tariffs generate large dynamic productivity losses and may enhance welfare in the short run only and when trading partners do not retaliate.” Overall, tariffs generate “large welfare losses in the medium and long run.” The optimal unilateral import tariff is zero for all policy horizons, while direct innovation support policies consistently outperform protection.

The reason is straightforward: innovation in the modern economy requires access to global knowledge flows, specialized inputs, and competitive pressure. Trade protection disrupts all three. If policymakers want to promote innovation, they should focus on policies that enhance rather than distort these innovation incentives.

Looking more broadly suggests that easier trade improves innovation

So those are different models to think through the trade-offs. We don’t need to take them completely seriously. But they do suggest that again, there are different mechanisms, and which ones are in play depends on the exact economy and policy in question. Unfortunately, the main empirical papers discussed above (Autor et al., Bloom et al., and Aghion et al.) all just look at one episode, the China Shock, across different countries.

To get a better sense of the average effect across trade liberalizations and the opposite, we want to look more broadly. This will allow more of the different mechanisms to be in the data and will help us have a better sense of the role of trade on innovation in order to try to back out the role of reducing trade through tariffs on innovation in a particular case.

Coelli, Moxnes, and Ulltveit-Moe have a 2022 paper that exploits the “Great Liberalization” of the 1990s across 65 countries and finds that trade policy reforms explained approximately 7% of global knowledge creation increases during that period. Crucially, both improved market access and import liberalization contributed positively to innovation.

What makes their approach compelling is how they measure trade liberalization. Rather than relying on a single metric, they use multiple indicators, including tariff reductions, non-tariff barrier removals, and capital account liberalization. This comprehensive approach helps capture the multifaceted nature of trade reform that characterized the 1990s.

The results strongly support the pro-innovation effects of trade openness. When they decompose the channels, they find that both export market access and access to more imports increase innovation. Export opportunities increase innovation by expanding the potential market for new technologies, raising the returns to R&D investment. Import liberalization, meanwhile, spurs innovation through competitive pressure and access to foreign knowledge and intermediate inputs.

My bold policy advice: Don’t cut off your arm.

The folk wisdom about necessity and invention sounds compelling, but the rigorous empirical evidence tells a different story. As a good piece by Bradford Jensen and Scott Wallsten put it:

These examples of “constraint-driven innovation” are real. However, these examples raise important questions about opportunity costs. When brilliant engineers spend their time solving problems that would not exist without trade barriers, we lose the innovations they might have created instead. Such innovation is more like “defensive innovation”—adjusting to artificial constraints rather than pursuing the most valuable improvements.

Overall, we should be very skeptical of stories that say cutting off your arm will be good for you. It’s possible it happens sometimes. People overcome adversity, which motivates them to great things. But as a general approach to policy, the more straightforward idea that raising the cost of things means you can do fewer things is a helpful guide.