Brad Setser on Global Trade Imbalance and the Dollar

ELI5/TLDR

The dollar isn’t strong right now because everyone trusts it as the world’s safe-haven currency. It’s strong because foreign money is chasing big returns in US stocks and bonds — especially the AI boom. That makes it a much more fragile kind of strength: if the AI trade cracks, the dollar has a long way to fall. Underneath all of it sits one stubborn fact — a handful of Asian manufacturing giants, above all China, keep selling far more to the world than they buy, and the surplus dollars they pile up are what quietly funds America’s habit of importing more than it exports.

The Full Story

Brad Setser runs the Follow the Money blog at the Council on Foreign Relations and has spent his career — at the IMF, the Treasury, and the National Economic Council — doing forensic accounting on where the world’s money actually goes. This conversation is him connecting a lot of dots that usually live in separate boxes: the dollar, the AI capex boom, China’s export machine, oil, and the slow death of the postwar trade order.

Why the dollar is strong is not what you think

The popular story is that the dollar is strong because it’s the reserve currency — the thing central banks hoard and everyone reaches for in a crisis. Setser says that story is out of date.

Dollar strength right now is a function not of the dollar’s role as a reserve currency or demand for dollar safe assets. It’s much more a function of old-fashioned searching for yield and looking for exceptional returns.

Quick mechanics, because this is the spine of the whole talk. A reserve currency flow is defensive — foreign central banks parking savings in something safe. A yield-chasing flow is offensive — private investors buying US assets because they expect to make more money there than at home. The first is sticky and slow to leave. The second can sprint for the exits.

His history of the dollar runs in eras. In the early 1990s it was actually weak. Then the Asian crisis and the dot-com bubble pulled money in. After dot-com burst, something odd happened: US interest rates were lower than elsewhere, yet money still flooded in — not from yield-seekers but from Asian central banks (China above all) buying dollars to stop their own currencies from rising. Since roughly 2014–15, we’ve been in the current era: relatively high US rates, a strong dollar, and crucially, the inflows come from private investors, not central banks.

The US is not borrowing from the world at an especially low price.

That last line matters. America isn’t getting a discount for being special. It’s paying up to attract money, like anyone else.

The AI boom changed the plumbing

For most of the post-2014 stretch, the money coming in was bond money — Taiwanese and Japanese insurers, German insurers, Swedish pension funds, all reaching for higher yields on US corporate bonds than they could get at home. Boring, steady, bond flows.

About a year ago that broke. Suddenly two to three percentage points of US GDP started flowing into US equities — into the tech platforms riding the AI run-up.

Recently it’s become a very significant factor, probably financing… attracting flows equal to about half or more of the US external deficit.

There’s a second twist. The big US tech companies — Google, Meta and the rest — have flipped from being net savers (they generated more cash than they spent) to net borrowers, taking on debt to build data centers. That extra borrowing pulls in still more foreign capital. So the AI boom is now propping up the dollar from two directions at once: foreigners buying the stocks, and the companies themselves issuing debt.

Which sets up the obvious question — what if it pops?

If the bubble burst, I think the dollar actually has a long way it could fall… In some sense, the stronger you go, the higher you go, the more room there is to fall.

His logic: a burst stops new equity inflows cold, and existing foreign holders start hedging (selling dollars to protect their positions), which piles on downward pressure. And the dollar today, on an inflation-adjusted basis against most of Asia, is as strong as it was at the dot-com peak. Lots of altitude, lots of room to fall.

Oil: a coin flip with two ugly faces

The interview was recorded with the Strait of Hormuz crisis live, so oil looms large. Setser refuses a forecast and instead lays out a bimodal outcome — two very different worlds with little in between.

In one world the Strait reopens, inventories get rebuilt, and you return to a glut: the Gulf producers compete all-out, the UAE has quit OPEC, more Iranian supply comes back, and oil settles low. In that world there’s little “petrodollar” accumulation — at $70 oil, Gulf spending has risen so much that only Norway and the UAE run real surpluses anyway.

In the other world the Strait stays shut or the war wrecks Gulf export infrastructure, and oil spikes to $140–150. The strange part: even then the Gulf doesn’t win, because the Gulf is the very thing that’s broken.

The biggest winners from… super high oil prices will be the oil guys in Texas and the oil guys in Alberta, Canada… they’re going to accumulate dollars, but they’re not going to be offshore dollars.

That’s a subtle, important point. Texas and Alberta keep their windfall onshore. The traditional petrodollar flow — Gulf oil money recycled back into global markets — doesn’t happen. Even a giant oil spike no longer mints the offshore dollars it used to. The US shale revolution added a few million barrels a day, but it never replaced the Gulf’s ~20% of world supply. There is no alternative to the Gulf at scale.

The real engine: surplus countries

Now the central thesis. Setser splits the dollar’s “global role” into separate jobs that usually get blurred together: reserve asset, payment vehicle (when Brazil trades with Africa, both happily use dollars — the genuine network effect), and the question of who is actually adding to their dollar holdings and thereby financing America’s deficit.

The answer to that third one is the big manufacturing economies of Northeast Asia.

China’s running a $1.2 trillion trade surplus. Korea and Taiwan, with a run-up in chip prices, are going to run a surplus… of $400 to $500 [billion]. These are big, big numbers.

These surplus dollars have nicknames in his telling — “DRAM dollars, chip dollars, Chinese unwillingness to let your currency go up dollars.” And there are far more of them than petrodollars, and will be until oil clears $130. The plumbing differs by country: in Korea, chip-export dollars fund a wave of Korean retail traders buying US stocks; in Taiwan, life insurers funnel them into US bonds; in China, it moves through the banking system and is, in his words, “a bit more of a mystery.”

When the interviewer suggests India’s IT-services exports are the local equivalent, Setser cuts in dryly:

You guys spend them all on imports of Chinese goods. So you guys don’t have a surplus. You have a deficit.

India generates gross service-export dollars but not the net surplus that funds global markets, because it runs a goods deficit (largely with China).

Why nobody stops

If these surpluses are destabilizing, why don’t surplus countries fix them? Because, Setser says, the pain has never landed hard enough on them.

The upper limit is not likely to come from “we can’t accumulate more dollars.” It’s more likely to come from other countries saying “we can’t keep running these kinds of trade deficits. We can’t continue to deindustrialize.”

The incentives haven’t changed in decades: grow through exports, use export profits to climb the technology ladder, and let exports compensate for weak demand at home. The drivers differ by country — China suppresses household consumption through a thin safety net and a regressive tax system (its individual income tax raises just ~1% of GDP, less than it collects from tariffs); Korea and Japan have aging populations; Germany runs fiscal conservatism and wage restraint. These distinctions matter, he argues, precisely because some policies can change. Germany is now doing a (defense-flavored) fiscal expansion; Korea is rethinking how much its pension fund invests abroad. China, under Xi, has been the stubborn constant — preferring to subsidize investment over households.

His prescription for China is concrete: let the yuan strengthen (the market actually wants it stronger right now, and the central bank could just set the daily fix higher), use the strong balance sheet of the central government to clean up the property mess and put cash in households’ pockets, build real pensions and unemployment insurance, and tax high incomes instead of low-wage work and consumption.

The China squeeze, up and down the value chain

A theme runs through the back half: China has become a giant exporter of high-end goods — solar panels, batteries, EVs — without giving up its share of low-end labor-intensive goods. Citing economist Arvind Subramanian, Setser notes this leaves everyone squeezed: poor countries can’t break in at the bottom, rich ones get undercut at the top. Profits are thin everywhere, even in the high-end stuff, because China keeps flooding supply.

Chinese automotive exports have gone from nothing to two times more than Japan ever did at the peak of Japanese automotive success in like 5 years.

That’s the headline statistic of the talk. It connects to economist Dani Rodrik’s recent turn into a “manufacturing skeptic” — the worry that poor countries can no longer manufacture their way to prosperity. Setser thinks the jury’s still out on whether development is possible without manufacturing, but agrees the global environment is brutal: if China backs its industries with a weak currency, a small country would need an even weaker one to compete — meaning poorer living standards for its own people.

His framework for understanding these transformations: industrial policy and currency policy are separate tools that become lethal when combined. Industrial policy alone (build the battery sector) often just creates a domestic industry. Add an undervalued currency and you turn that domestic success into a global export tsunami. That mix is how China went from clean-tech importer to dominant exporter in a decade.

Trade wars, allies, and “unbalanced globalization”

On Trump’s trade policy, Setser is blunt:

There is no coherence in US policy right now.

You can’t shrink a trade deficit while running a 6%-of-GDP fiscal deficit and an import-hungry AI investment boom. Tariffs haven’t reduced the deficit or slowed global trade — in fact, lengthening supply chains (China ships to Vietnam, Vietnam assembles for the US) creates more trade, not less. He calls this unbalanced globalization, not deglobalization: China’s trade keeps growing two to three times faster than global trade, with exports outrunning imports, which mathematically means someone else’s trade is being crowded out.

He also flags a chunk of “trade” that’s really tax avoidance — US pharma profits booked in Ireland, IP shuffled offshore — that tariffs don’t touch.

On leverage: “Sadly, Xi has the upper hand right now.” China absorbed the tariff punch (its firms knew how to reroute through Vietnam), while the US got hurt where it’s genuinely dependent — Chinese retaliation on soybeans, beef, and aircraft, and above all China’s chokehold on rare earths and critical minerals. Trump discovered that 100–150% tariffs don’t pressure the other side if they can reroute; they mostly pressure your own importers who have to pay them.

And the idea of a US-Europe coalition against China? “It is dead for the Trump administration” — closed for at least the next two and a half years. The timing was tragically mismatched: when the US wanted to be tough on China (Trump 1, early Biden), Europe wasn’t interested; now that Europe has soured on China after its latest export wave, the US wants to fight everyone at once.

The detective work

The most delightful stretch is Setser on his actual craft — tracing money that’s been deliberately disguised.

Countries have gotten very creative about how to manage currencies without appearing to manage their currencies.

Taiwan quietly scrapped hedging requirements for life insurers — which looks like telling insurers to take more risk, but conveniently solved a currency problem. China shows almost no FX accumulation on its central bank’s balance sheet, yet clearly piles it up in state commercial banks, policy banks, and sovereign funds. And his favorite catch: China’s holdings of US agency mortgage bonds appeared to drop from $250bn to $150bn — until you notice Canada and Luxembourg (a custodial center with no natural appetite for US mortgages) suddenly holding a fortune in the same bonds. The money didn’t leave; it changed costume.

Key Takeaways

The dollar’s current strength is yield-driven (private investors chasing returns), not reserve-driven (central banks seeking safety). Yield flows are far more fugitive — they can reverse fast.
About a year ago, flows into the US flipped from bonds to equities; foreign equity inflows now finance roughly half of the US external deficit, chasing the AI/tech run-up.
US big tech flipped from net savers to net borrowers to fund data centers — a second channel pulling capital in and propping up the dollar.
An AI bust would crater the dollar twice over: new inflows stop, and existing holders hedge (sell dollars). The dollar is near dot-com-era highs vs most of Asia — lots of room to fall.
Oil is bimodal: reopen the Strait → glut → low prices → little petrodollar accumulation; keep it shut → $140-150 → but the winners are Texas/Alberta (onshore dollars), not the broken Gulf. Shale never replaced the Gulf’s ~20% of world supply.
The deficit financiers are Northeast Asian manufacturers: China (~$1.2T surplus), Korea + Taiwan ($400-500B on chips). These “chip dollars” dwarf petrodollars until oil tops ~$130.
India generates gross IT-export dollars but no net surplus — it runs a goods deficit, largely with China, so it doesn’t fund global markets the way East Asia does.
Surplus countries won’t self-correct because the pain lands on deficit countries (deindustrialization), not on the surplus ones. The binding limit is political, not financial.
China’s fix is available and cheap on the currency front (just set the fix higher; the market wants a stronger yuan) but politically hard on the fiscal front (build a safety net, tax high incomes not low-wage work — its income tax raises only ~1% of GDP).
China expanded at the high end (panels, EVs, batteries) without ceding the low end — squeezing competitors at both ends. Chinese auto exports hit 2x Japan’s all-time peak in ~5 years.
Mental model: industrial policy and currency policy are separate tools; combined (domestic capability + undervalued currency), they convert local industries into global export floods.
Tariffs created more trade, not less, by lengthening supply chains (China → Vietnam → US). Setser calls the regime “unbalanced globalization,” not deglobalization.
In the US-China standoff, Xi holds leverage: China rerouted around tariffs; the US got hurt on soybeans, aircraft, and especially rare earths. High tariffs don’t bite if the other side can reroute.
A US-Europe coalition against China is “dead” for this administration — the timing of who wanted to be tough on China never aligned.
Capital-flow forensics: governments disguise currency intervention (Taiwan dropping insurer hedging rules; China hiding FX in state banks; China’s US-mortgage holdings reappearing under Canada and Luxembourg).

Claude’s Take

This is a high-signal hour. Setser is the rare macro voice who works bottom-up from balance-of-payments data rather than top-down from a thesis, and it shows — the claims come with mechanisms and numbers attached, and he repeatedly refuses to forecast when he doesn’t have an edge. The framing that reframes dollar strength from “safe-haven demand” to “yield-chasing” is genuinely clarifying, and the corollary — that this kind of strength is fragile and tied to the AI trade — is the most important takeaway for anyone trying to understand current markets.

The BS filter finds little to flag. His priors are visible: he clearly thinks China should rebalance toward consumption and that the Trump trade strategy is incoherent, and he says so plainly rather than smuggling it in. Reasonable people can disagree on whether China’s model is unsustainable or simply different, and Setser is more confident than the data can fully bear that surplus countries are “doing something wrong.” But he’s intellectually honest about uncertainty (the oil distribution, the Rodrik hypothesis being “unproven”), and his disclosed positions don’t distort the analysis.

Score 8/10. It loses points only because the conversation is dense and assumes some prior fluency — a listener who doesn’t already know what a current account is will drift. But for anyone with a finance grounding, the dot-connecting between AI capex, the dollar, Asian surpluses, and oil is exactly the kind of synthesis you rarely get in one sitting. The detective stories at the end (chicken feet through Hong Kong, mortgages hiding in Luxembourg) are the cherry — they make the abstract machinery feel concrete and a little absurd, which is the whole point.

Brad Setser on Global Trade Imbalance and the Dollar | Subtext by Zerodha