The Carry Subsidy: EM Local-Currency Debt and the Repricing of Dollar Risk

A one-line headline crossed my feed this week and I almost scrolled past it. No numbers. No country names. No time window. Just a rotation: emerging-market investors are favoring local-currency bonds while dollar debt lags.

That is exactly the kind of headline I have learned to read twice, because the absence of detail is the detail. When a market signal arrives stripped of its supporting data, it usually means the signal is structural rather than episodic — nobody bothered to attach a chart because the move is being taken for granted by the people who trade it.

I have spent the last nine years watching two systems that look unrelated trade on the same underlying machine: sovereign credit and crypto liquidity. They rhyme more than either side admits. And this particular rotation — out of hard-currency EM debt, into local-currency EM debt — is the cleanest read we have had in eighteen months on where the marginal dollar is going, and what it costs to borrow one.

The rotation is not a sentiment story. It is a funding-cost story, and funding costs are now the single most important input into every crypto valuation model nobody wants to talk about.

Two Indexes, One Decision

To understand the signal, you have to understand that "emerging-market debt" is not one asset. It is two, and they behave nothing alike.

The hard-currency universe — tracked by the EMBI family of indices — is dollar-denominated sovereign and quasi-sovereign paper. When you buy it, you are buying a US Treasury plus a credit spread, and you are buying it with the US dollar as the unit of account. Your return decomposition is brutally simple: the risk-free rate, the spread compression or widening, and the occasional default. There is no currency line item. The sovereign carries the currency risk; the investor does not.

The local-currency universe — tracked by the GBI-EM family — is the opposite. These are bonds issued by emerging-market governments in their own money, denominated in pesos, real, rand, rupiah, zloty, pesos again, and a long tail of others. The universe is roughly three trillion dollars in market value, which makes it larger than most crypto people realize and smaller than it should be given the size of the economies behind it.

When you buy local-currency debt, your return has three engines: the local yield, the duration of the bond, and the currency itself. You are not buying a spread. You are buying a carry trade wearing a bond's clothing.

That distinction matters enormously, because the two universes respond to completely different shocks. A US Treasury selloff hurts hard-currency EM debt directly and immediately. It may barely touch local-currency debt, or even help it, if the underlying inflation story is improving and the local central bank has room.

For most of the last three years, dollar debt was the crowded trade and local-currency debt was the unloved one. The logic was straightforward. The Federal Reserve held policy rates at generational highs for longer than anyone modeled in 2022. For an emerging-market sovereign or corporation that had borrowed in dollars, that meant a refinancing wall at the worst possible moment: every maturity that came due had to be rolled at a spread plus a base rate that had tripled.

Economists have a name for the deeper problem. They call it original sin — the structural inability of an emerging economy to borrow abroad in its own currency. The consequence is a permanent mismatch: liabilities in dollars, revenues in local money. When the dollar strengthens, the debt burden expands without anyone issuing a single new bond. It is a balance sheet that gets weaker while standing still.

What the current rotation suggests is that the market is beginning to price the partial remission of that condition. Not its cure. Its remission.

Three things had to be true for that to happen. First, emerging-market central banks had to build enough credibility that investors believe they will defend real rates rather than inflate away the debt. Second, inflation had to genuinely roll over in enough of these economies that positive real yields became a coherent story rather than a wish. Third, the dollar had to stop being the only place to earn a real return.

That third condition is where the crypto reader should sit up. Because the same three conditions describe almost exactly what happened to on-chain dollar yields between 2022 and 2025.

The Mechanics Nobody Prices

Let me get specific, because this is where most commentary stops and where the actual analysis begins.

Total return in local-currency EM debt decomposes into: local nominal yield, minus duration loss if yields rise, plus or minus currency movement. That is it. Everything else is decoration.

The seductive part is the first term. Several emerging markets have offered nominal policy rates in the high single digits or low double digits while their inflation prints were decelerating. On paper, that is a real yield of four, five, sometimes six percent — in local currency, backed by a sovereign, in a market with a functioning central bank. Compared to a US T-bill, that looks like free money.

It is not free. The price of that carry is the currency line item, and the currency line item is the one thing no model predicts well.

Here is the mechanical detail that gets buried: most institutional investors hedge their currency exposure back to dollars, and the hedge is not free. The cost shows up in the cross-currency basis — the premium or discount for swapping one currency into another over a forward period. When the basis is expensive, a large chunk of that five-percent local yield evaporates on the way home. What looks like a five-percent carry to a headline reader can be a two-percent carry to a hedged fund, and a negative carry in a stress event when the basis blows out.

So when I read that "investors favor local-currency bonds," I ask a specific question: hedged or unhedged? Because those are two entirely different trades that share a headline.

Hedged local-currency debt is a rates trade. You are expressing a view that emerging-market real rates are attractive relative to developed-market real rates, with the currency risk neutralized. That is a respectable, boring, institutional allocation decision.

Unhedged local-currency debt is a currency trade. You are long a basket of emerging-market currencies against the dollar, with a coupon attached to make the wait tolerable. That is a directional macro bet, and it lives or dies on the dollar cycle.

The headline does not distinguish. Which means the market probably does not want to distinguish, which means the marginal buyer is likely unhedged, which means the trade is more reflexive than it appears.

The Reflexivity Nobody Wants to Name

I want to draw a comparison that will annoy people on both sides of this desk, because it is the most important thing in this article.

In the summer of 2020, I spent weeks inside the mechanics of yield farming. I wrote a primer that went further than I expected it to, mostly because the analogy I used was simple: yield farming is a project paying you to park capital in a pool, and the payment attracts more capital, and the larger pool attracts more attention, and the attention attracts more capital. For as long as the incentive program runs, the APY looks organic. The moment it stops, the liquidity walks out and you discover that the real users were a rounding error against the mercenary ones.

Liquidity mining APY is a subsidy dressed as a yield. It is real money for as long as someone else is paying it.

Now read the emerging-market local-currency story again with that lens and tell me it does not rhyme.

Foreign inflows into a local-currency bond market push the currency up. A stronger currency improves the dollar-denominated return of the position. An improved dollar-denominated return attracts more foreign inflows. On the way up, the carry looks better than the coupon because the currency is doing free work underneath it. Every allocator who enters is paid partly by the currency appreciation that the previous allocator helped create.

That is the subsidy. A meaningful slice of EM local-currency total return in an inflow cycle is not compensation for risk-bearing. It is compensation for being early to a flow.

This is not a reason to avoid the trade. It is a reason to size it correctly and to know exactly which term of the return you are underwriting. If you are underwriting the coupon, you are underwriting central bank credibility. If you are underwriting the currency, you are underwriting the flow. Those are different bets with different failure modes and different time horizons, and conflating them is how portfolios that look diversified turn out to be one position.

I learned this the hard way in a different market. In 2021, I interviewed thirty holders of a blue-chip NFT collection across Taipei and Tokyo for a piece on status and identity as an asset class. The financials looked impeccable — a floor price that had risen relentlessly, a community that felt permanent, a supply cap that guaranteed scarcity. What the financials could not show me was the composition of the bid. Almost everyone I interviewed had bought to hold, but almost everyone had bought because someone else was buying. The floor was not supported by utility. It was supported by a shared belief that the floor would keep rising. When that belief broke, the floor did not soften. It gapped. The market reaches for liquidity, and when there is none, it finds the next available price, which is always much lower than anyone's model.

Local-currency sovereign debt is not an NFT collection. But both are priced at the margin by a bid that can be described, and not always predicted, by flow.

Where the Fragility Actually Moved

The consensus reading of this rotation is optimistic, and I understand why. If emerging markets can borrow in their own currencies, they are less exposed to dollar strength. If they are less exposed to dollar strength, they are less likely to be forced into a crisis by a Fed decision made three thousand miles away. If they are less crisis-prone, their assets deserve lower risk premiums.

All of that is true at the level of the sovereign balance sheet. What it misses is that the fragility did not disappear. It changed address.

Dollar debt concentrates risk at the sovereign. The government carries the mismatch, and when the mismatch becomes unbearable, the government restructures, and the loss is shared among creditors through a negotiating process that is ugly but organized. There is a mechanism. There is a committee. There is a precedent.

Local-currency debt shifts that risk to the market. The sovereign no longer carries the currency mismatch — the investor does. And investor risk does not get restructured. It gets repriced, instantly, through the exit.

Here is the structural problem: local-currency bond markets are thinner than dollar markets at the exact moment liquidity matters most. Dealer inventories in most EM local rates are a fraction of the depth available in US Treasuries or even in dollar-denominated EM credit. Foreign investors who bought into an inflow cycle are the same investors who will need to leave in a risk-off cycle, and they will all be trying to use the same door.

So the honest assessment is this: the rotation reduces sovereign fragility and increases market-structure fragility. It moves the fault line from the balance sheet to the order book. Whether that is an improvement depends entirely on which failure you consider more survivable — a slow restructuring with lawyers, or a fast repricing with no bid.

I have a personal bias here. I audited a piece of code in 2016 that was, at the time, the largest collective investment vehicle anyone had assembled on-chain. I did not need the full picture to see the problem. I needed one function, one external call, and the realization that the collateral accounting updated after the value moved instead of before. The system was not fragile because it was small. It was fragile because its safety depended on state that was read after it had already changed.

That is the failure mode I look for everywhere now, in code and in markets. Not size. Sequencing. Who reads the state, and when, and what happens if everyone reads it at once.

A local-currency bond market that depends on foreign inflow to support its currency, which supports the return, which supports the inflow, is a system where the state is read after it has already changed. That does not make it a bubble. It makes it a reflexive structure, and reflexive structures do not fail gently.

What Crypto Is Getting Wrong

Now the part that matters for the readers of this particular publication.

Every macro rotation eventually gets tokenized into a narrative. The moment "de-dollarization" entered the mainstream vocabulary, I started seeing products that claimed to let you trade it. There will be a de-dollarization token. There will be a BRICS-settlement narrative. There will be a sovereign-digital-currency story dressed up as an investment thesis, with a whitepaper that explains monetary history in the first eight pages and the token distribution in the last two.

Most of these will have no coupon, no claim, and no mechanism by which the thing they describe flows into the thing you can buy. Searching for truth in the noise of the network means being willing to admit when the noise is not a signal, just a crowd with a word.

The honest observation is different, and more interesting: the emerging-market local-currency rotation has almost no on-chain expression at all.

Tokenized treasuries — the on-chain wrapper around US government debt — have become a real category with real assets and real distribution. Tokenized emerging-market local-currency sovereign debt essentially does not exist. Not because nobody wants it. Because the pieces are missing.

You need a legal wrapper that a local sovereign recognizes. You need a custodian that can hold in-country. You need an FX hedge that is executable at institutional size. You need a redemption mechanism that survives a local market holiday. You need a reporting layer that satisfies two regulatory regimes simultaneously. None of these are cryptographic problems. They are legal and operational problems, and they are the reason the category is empty.

Here is the information gain in that emptiness: the gap is not a missing product. It is a missing primitive, and the primitive is the on-chain FX hedge.

Without an on-chain cross-currency swap that an institution can trust, tokenized local-currency yield is unusable for anyone managing against a dollar benchmark. And most of the world manages against a dollar benchmark. That single missing leg is what separates a functioning on-chain asset class from a curiosity.

What exists today under the banner of "on-chain FX" is, with a few exceptions, stablecoin arbitrage. It is USDC and USDT trading at small premiums and discounts across venues, and the arbitrageurs who close that gap. That is a useful service. It is not an FX market. It has no forward curve, no basis, no tenor, and therefore no ability to let an institution hedge a three-year bond.

I have written before that the Cosmos interchain stack is technically elegant and structurally unable to capture the value that flows through it — sovereign zones, each competently engineered, collectively fragmented, with the value leaking to the rails that connect them rather than the zones that constitute them. Emerging-market local-currency bond markets have the same shape. Individually, several of them are sophisticated. Collectively, they leak. The connection layer — dollar funding, global custody, the basis — captures the spread, and the zones keep offering yield to whoever holds the pipe.

That parallel is uncomfortable. It is also instructive. Because the same question applies to both: who owns the connection layer, and does the yield ever accrue there?

One more piece of mispricing worth naming. A local-currency sovereign bond pays you to wait. A governance token asks you to hope. The bond has a coupon schedule, a legal claim, and a central bank reaction function you can model. The governance token has a treasury you do not control and a roadmap you cannot enforce. When the market narrative turns toward "real yield" and "cash flow," it is not rediscovering crypto. It is rediscovering the oldest idea in finance, and the assets that will benefit are the ones that actually distribute, not the ones that promise to.

The narrative is the asset. The code is the proof. In this rotation, the code is a coupon schedule, and that is precisely why the money is moving.

The Contrarian Case

Let me put the sharpest version of the bear case on the table, because the optimistic reading is already well represented.

First: conflating market-led de-dollarization with monetary-system de-dollarization is how people lose money. A rotation inside a fixed income allocation is not the end of dollar hegemony. Reserve currency composition moves on a decade timescale, governed by inertia, network effects, and the simple fact that there is no replacement asset with comparable depth. Trade invoicing is even stickier. What we are watching is investors choosing a different instrument within the same dollar-centric system, not a change in the system itself. The dollar's share of global reserves will not move meaningfully on the back of a local-currency bond rally, and anyone trading as if it will is trading a story, not a market.

Second: the "reduced reliance on foreign capital" framing is backwards. If the marginal buyer of local-currency debt is a global asset manager running a benchmark, then the sovereign has not reduced its dependence on foreign capital. It has changed the denomination of that dependence. The money still comes from outside. It still leaves when the global risk cycle turns. The only difference is that the exit now shows up in the currency instead of in the credit spread — and a currency move can be faster than a rating downgrade.

You did not remove the dependence. You re-denominated it.

Third: the stability narrative undersells the liquidity mismatch. Local-currency markets have fewer committed market makers, less repo financing, and thinner hedging infrastructure than dollar markets. In a disorderly unwind, foreign investors heading for the exit will find that liquidity is available at prices they will refuse to accept, which means they will wait, which means the price discovery will be delayed and then concentrated. That is a worse pattern than an immediate repricing. Concentrated exits produce gaps, and gaps produce margin calls, and margin calls produce forced sellers who do not care about the fundamentals at all.

Fourth: the same marginal dollar funds this trade and funds crypto. This is the part the crypto market consistently fails to internalize. August 2024 gave us a live demonstration. A carry trade in one currency unwound, and the liquidation did not stay in that currency. It propagated through leveraged positions across every venue that runs on the same funding currency, and crypto — which is the most leverage-sensitive market in the world — was the most violently repriced. If the EM local-currency carry unwinds, the first casualty is not the EM bond fund. It is the leveraged basis trade, and the second casualty is crypto liquidity, because both are borrowing from the same pool.

That is not a prediction. It is a structural relationship. The market that is most sensitive to dollar funding conditions is the one that trades 24/7 with the thinnest balance sheets, and that market is ours.

What to Watch, and What It Means

We are in a chop market. Chop is for positioning, not for conviction, and positioning in a chop market means watching signals that tell you the funding regime before it changes.

Here is the stack I am actually monitoring.

One: the relative performance of local-currency versus hard-currency EM debt indices. As long as the spread is widening in favor of local currency, the rotation is intact. When it stops widening and starts compressing, the flow has stopped, and the subsidy is gone. This is the single cleanest read on the thesis.

Two: the cross-currency basis for the major EM currencies. If hedging costs are rising, the trade is becoming more expensive for the disciplined institutional money, which means the marginal buyer is becoming less disciplined. That is the tell that the composition of the bid is deteriorating.

Three: stablecoin net issuance as a dollar-demand proxy. A stablecoin is a claim on dollar assets issued by a non-bank, which makes it the retail edition of the same instrument the rotation is moving away from. If demand for dollar liabilities is genuinely peaking at the margin, the growth rate of stablecoin float should show it — not the level, the growth rate. And watch the premium or discount of dollar stablecoins in emerging-market peer-to-peer venues. That is a real-time, high-frequency read on local dollar demand that no central bank publishes.

Four: perpetual funding rates and the cost of leverage. If funding on the major venues is persistently positive while the dollar is softening, the market is leaning long on cheap dollars. That is precisely the configuration that produces the violent unwind. Negative or flat funding during a dollar rally is a healthier signature than positive funding during a dollar decline.

Five: the tokenized treasury supply curve. This is the quiet one. Tokenized government debt is where the on-chain dollar complex is being built, and its growth is the best available proxy for whether institutions are actually moving balance sheet on-chain or just talking about it. Where code meets culture, the real value emerges — and in this category the code is a wrapper around a government's promise to pay, which is the least glamorous and most durable primitive in finance.

What I would not do is buy a token because it says "de-dollarization" on the label. The rotation is real. The narrative around it is mostly noise. The tradeable expression is not a coin — it is the infrastructure that lets a dollar-based investor hold a non-dollar coupon without giving up the hedge. That infrastructure does not fully exist yet. When it does, it will not be announced with a token sale. It will show up as a settlement layer that suddenly has institutional volume and no marketing.

The next cycle will be built for a world where the world's savings are denominated in more than one currency. The question worth sitting with is not whether that happens. It is who ends up owning the pipe between the currencies — because in every fragmented system, the yield accrues to the connection, not to the zone.

The rotation has begun. The rails have not been built. That gap is the entire opportunity, and it will close faster than the last one did.