Gita Gopinath on Trade, Currencies, and China

Here is a puzzle that should bother anyone who took a semester of economics. The textbook says that a country buying more from the world than it sells should watch its currency slide until the imbalance corrects itself. Australia has run trade deficits for decades. Its dollar has not collapsed. Nothing about Australian demand looks broken. So what happened to the adjustment mechanism?

Gita Gopinath, the Harvard economist who served as first deputy director and formerly as chief economist at the International Monetary Fund, a tenure that ended in 2025, has an answer that reframes the whole question. A trade balance, she points out, depends on two things: relative prices and relative demand. "If in the US consumption is strong, while in China consumption is weak, even if relative prices favor US goods, just the overall level of demand being lower in China would reduce the overall level of exports" (01:30). The exchange rate is only half the story. The other half is which country's households and firms are actually spending.

What ties the two together, in her account, is not a market that clears every quarter but a constraint that only has to clear eventually. Economists call it a no-Ponzi condition: a country that borrows from the rest of the world has to pay it back at some point, on pain of default. Gopinath put it this way:

"A country that is running deficits and therefore accumulating a lot of liabilities to the rest of the world will have to repay those liabilities." (02:21)

The repayment can come from a weaker currency, or it can come from a recession that crushes imports, or from a lucky natural-resource discovery that boosts exports. The math does not care which. That is why the tight link between deficits and depreciation, taught in sophomore textbooks, keeps failing in the data.

A Currency With Habits

Gopinath's best-known research, the "dominant currency paradigm," pushes the same insight into the plumbing of global trade. The old assumption, going back to Milton Friedman, was that when the US sells to China, it prices in dollars, and when China sells to the US, it prices in its own currency, the renminbi. Move the exchange rate and both prices move with it. That is not how trade actually works. "Close to 90% of China's exports to the US are priced in dollars and is relatively stable in dollars" (08:11). China prices in the currency everyone else uses, largely because the parts and materials it imports to build those exports are themselves priced in dollars. There is no margin left to absorb a currency swing, so the dollar price barely moves (11:12). That single fact, Gopinath argues, explains why a Trump-era tariff shows up almost fully in the price American importers pay, while an equivalent swing in the dollar-renminbi exchange rate barely registers. Tariffs are bolted directly onto an already-dollarized price. Currency moves have to fight through someone else's sticky pricing decision first.

Argentina supplies the sharpest illustration of how hard that stickiness is to break once expectations come unglued. Two years before this conversation, inflation there ran around 150%; by the time of the interview it had fallen to about 30% (16:15). Gopinath credits Javier Milei's government for recognizing, more clearly than its predecessors, that the root problem was fiscal: years of deficits financed by printing money. Milei has run primary surpluses since taking office and is pushing for central bank independence so those deficits cannot be monetized again (17:06). But the last mile is the hardest. "Argentina's inflation was completely deanchored at the time when he took over. Any news of the possibility of the exchange rate depreciating even by 1% or 2% would immediately show up in prices" (18:27). A government can announce good policy; it takes years before people stop pricing in the worst case by reflex.

Gopinath's prescription is almost counterintuitive: let the peso fall further, not less. Argentina has moved from a rigid peg to a crawling peg to a wide crawling band, which in principle allows more currency movement (20:16). She thinks officials are still too hesitant to buy dollars and build reserves, especially with a national election due in October 2027 (19:30). A cheap, undervalued currency paired with credible fiscal and monetary policy would let the real exchange rate adjust without reigniting the inflation psychology that had proven so hard to break. What it should not do, in her view, is chase a fully dollarized economy, the path Milei once campaigned on. "The two main dollarized economies of the world, Ecuador, El Salvador, are all in programs with the IMF because ultimately if you don't have the right fiscal policies, you're going to end up needing a bailout" (25:32). Giving up a currency does not give up the need for discipline. It just removes a tool that a well-run country could otherwise use.

Surpluses as Symptoms

The same logic reframes how Gopinath reads China's enormous manufacturing surplus, and it is where she parts ways with economists like Scott Sumner, who argue that trade imbalances between nations deserve no more worry than imbalances between American states. She agrees with half of that: a surplus or deficit is not, by itself, a policy target. What she rejects is the idea that China's surplus reflects healthy comparative advantage. "China's surpluses are a reflection of things going wrong in China. It's not a reflection of strength in China. It's a reflection of weak consumption" (33:10). Households save too much and spend too little, a property-market bust from five years ago remains unresolved, and the government is now telling firms to raise prices to fight deflation rather than letting markets clear (33:10). A surplus, in other words, can be evidence of a country's growth engine running well, or evidence that something inside it is misallocated. The number alone cannot tell you which.

She applies a similarly clinical eye to newer monetary technology. Stablecoins, the private digital tokens backed one-to-one by a currency like the dollar, are often pitched as a way to fix slow, expensive cross-border payments. Gopinath is skeptical that the cost savings are real once on-ramp and off-ramp fees into ordinary currency are included (34:48). What she does like is the competitive pressure they put on incumbents:

"One thing I do like about stablecoins is that they're finally putting pressure on the banks, on the traditional banks to actually pay attention to the prices that they charge for their services, especially cross-border." (35:37)

A cheaper alternative does not have to win to be useful. It just has to exist.

The conversation's most consequential number may be the least dramatic: half a percentage point. That is the annual productivity boost some economists expect from artificial intelligence, roughly comparable to the internet-driven growth spurt of the late 1990s. If it materializes and holds, US debt-to-GDP could stabilize near 120% rather than spiral. Gopinath does not dismiss the scenario, but she is not ready to bank on it either, citing uncertain adoption, public pushback against AI tools, and the possibility of a financial-market correction along the way (40:13). More striking is her judgment that America's debt path is now a bigger worry than France's, Germany's, or Japan's, precisely because of trajectory rather than current level (39:09):

"It's the US debt trajectory going forward that is particularly concerning in addition to what the levels that we're seeing right now." (39:09)

She is now building a research project on exactly this kind of lag. Past waves of automation show that roughly 85% of the job losses they cause do not show up gradually. They show up all at once, concentrated in the first year of a recession, when companies that have been quietly automating finally use a downturn as cover to cut (54:10). If AI is doing that quietly now, official job numbers would not catch it until the next recession arrives.

By the end of the conversation, Gopinath is describing her own field in similar terms: not a hard science that produces clean, testable laws, but one entangled with politics, history, even inheritance customs. Asked why parts of Kerala, her family's home state in India, post higher literacy and stronger outcomes for women than much of South Asia, she points to matrilineal property traditions where inheritance passes through women rather than men (51:26), while admitting she has not formally studied the question. It is a small, honest moment in an interview built on hard models, a reminder that the discipline she has spent a career refining still runs, in the end, on judgment.