Do Tariffs Shrink Trade Deficits? Engel Curve Holds the Key, Say MIT Economists

MIT economists Arnaud Costinot and Iván Werning show that permanent tariffs can reduce trade deficits if imports behave like luxury goods, reflected in a convex Engel curve. However, they caution that while tariffs may work in theory, they are not advisable as policy tools.

In a bold and timely contribution to international economics, MIT economists Arnaud Costinot and Iván Werning, in collaboration with the National Bureau of Economic Research (NBER), have revisited the ever-controversial question: Can import tariffs reduce trade deficits? Their latest working paper, How Tariffs Affect Trade Deficits (April 2025), doesn’t just regurgitate the usual economic intuition, it retools it through a sophisticated two-period trade model with general preferences and technologies. Rather than debating whether tariffs should be used, the paper is concerned solely with whether they can work as intended under realistic economic conditions. To that end, the authors craft a general model that allows for a wide range of goods, consumption patterns, and production technologies, while revealing that the trade deficit’s response to a tariff depends crucially on one elegant mathematical insight: the shape of the Engel curve.

History Repeats, but Does Economics Agree?

The belief that tariffs can mend trade deficits is not new. During the Great Depression, countries raised trade barriers not merely to protect industries but to tackle imbalances in trade. Nearly a century later, similar motives resurfaced in U.S. policy under the Trump administration’s 2025 “reciprocal tariffs,” driven by the idea that curbing imports would naturally shrink the trade deficit. Yet mainstream economists have remained skeptical. They frequently cite Lerner symmetry, the notion that an import tariff acts as a tax on exports, and argue that trade balances are intertemporal by nature. Simply put, trade deficits reflect a nation’s saving and investment behavior over time, not just current-period buying and selling of goods. What makes Costinot and Werning’s analysis stand out is that it doesn’t merely reiterate this textbook reasoning; it dives deeper to explore whether and how a permanent tariff might alter consumption dynamics between the present and the future.

The Engel Curve: A Hidden Signal in Trade Imbalances

The core of the paper revolves around a surprising yet powerful sufficient statistic: the curvature of the Engel curve in the aggregate space of imports and exports. If this curve is convex, meaning that imports increase more than proportionally with income, then imports function as luxury goods. In that case, a tariff raises the relative cost of consuming now versus later, nudging the economy to save more today, consume less, and reduce the trade deficit. However, if the Engel curve is linear, tariffs are neutral, they neither help nor hurt the trade balance. This finding is central because it moves the discussion away from simplistic metrics like “import share in GDP” toward more nuanced, structural attributes of the economy. In short, it’s not just how much a country trades, but how its trading patterns respond to income and prices that determine the effectiveness of a tariff.

Intensive vs. Extensive Margins: Why Structure Matters

But what shapes the Engel curve in the first place? The authors identify two main channels: preferences and economic structure. On the preferences side, if consumers treat imported goods as superior or luxury items, the Engel curve is naturally convex. But perhaps more intriguingly, even when preferences are standard and homothetic (meaning consumption patterns scale proportionally with income), convexity can arise from the structure of trade itself, specifically from an active “extensive margin.” This refers to shifts in which goods are traded as the economy grows, such as certain products moving from non-traded to imported or exported categories. In contrast, if all trade occurs within a fixed set of goods, the “intensive margin”, the Engel curve remains linear, and the effect of tariffs on trade deficits disappears. This structural nuance proves to be pivotal: economies with dynamic trading patterns are far more susceptible to deficit reduction through tariffs than those with rigid trade structures.

Simulations and Warnings: Theory Meets Reality

To make their abstract findings more tangible, Costinot and Werning simulate outcomes using a constant elasticity of substitution (CES) model. Their example shows that, under realistic assumptions, a 60% permanent tariff could reduce the trade deficit from 6% to 4% of GDP. Yet the magnitude of this effect depends heavily on the responsiveness of trade patterns, particularly the extensive margin. A more flexible trade structure amplifies the tariff's impact, while a static one mutes it. They also revisit the theoretical contributions of Obstfeld and Rogoff (2000), whose work linked trade costs with interest rate differentials and intertemporal consumption decisions. Costinot and Werning extend this line of reasoning by showing that Engel curves capture all the relevant information needed for such interest rate channels. However, they caution that previous empirical models may have overstated the effect of trade costs on deficits by not adequately accounting for the shape of Engel curves or the balance between extensive and intensive margins.

While the paper is a theoretical tour de force that enriches our understanding of trade imbalances, it ends with a sober reminder: just because tariffs can work under certain conditions doesn’t mean they should be used. The authors stress that their inquiry is not a policy endorsement but an attempt to clarify the mechanics underlying a long-debated economic question. Through meticulous modeling, mathematical insight, and grounded simulations, Costinot and Werning provide a sophisticated framework for thinking about the real impact of tariffs, one that goes far beyond the political soundbites and into the heart of intertemporal economic behavior.