In monetary hierarchy terms, stablecoins act as a tokenized US Treasury. The users of stablecoins thus have a measurable impact on the Treasury market. So, who are the users? Retail users in emerging markets have emerged as a primary driver of stablecoin adoption, using dollar-pegged tokens as a substitute for hard-currency bank deposits — a form of digital dollarization. For the first time in history, the behaviors of households in emerging markets are directly and structurally linked to the Treasury market through the stablecoin reserve mechanism. This terminal tracks macroeconomic stress in emerging markets as it propagates through stablecoin demand into US Treasury markets.
This operates through two channels. Forward: EM stress → stablecoin demand surge → Tether/Circle minting → T-bill demand pressure → yields compress. Reverse: a stablecoin run → T-bill fire sale → systemic shock transmitted into short-end rates.
A BIS study (Ahmed & Aldasoro, WP 1270) measured this: a $3.5B inflow over 5 days lowers the 3-month yield by up to 5 basis points within ~13 trading days — and roughly double that when the Treasury market is already under stress. With the market now ~$300B and forecasts pointing to $2 trillion by 2030, this channel deepens with every dollar of adoption. Screen 8 explains the full causal architecture.
Furthermore, the structural channel has strengthened with the maturation of the stablecoin sector: materially deeper price impact as the market has grown larger and more institutionally embedded — a function of both scale and the increasingly tight integration between stablecoin reserve management and the T-bill market. With the stablecoin market cap now ~$300 billion and a baseline forecast of $2 trillion by 2030, these stress-regime amplification effects will deepen with every dollar of adoption. The four state lights above determine the regime multiplier applied to the model's live yield prediction on screen 1.
Emerging markets are increasingly adopting stablecoins. This establishes a novel transmission channel between the dollar demand of emerging markets and the US Treasury market. When these markets enter a currency crisis, retail savers flee their local currency into dollars — increasingly by buying USDT on local crypto exchanges (BTCTurk, Binance P2P) rather than through the formal banking system. That demand bids up the USDT/TRY or USDT/ARS premium above the official exchange rate: a real-time signal of retail dollar flight.
Additionally, crises can force mass redemptions as firms and savers convert USDT back to local currency to meet liquidity needs. In this case, Tether must liquidate T-bill holdings to satisfy redemptions — transmitting EM liquidity stress directly into the short-end Treasury market.
The method. Exchange Market Pressure (EMP) is the stress gauge and trigger — a variance-weighted combination of FX depreciation, policy-rate moves and reserve drawdowns (Eichengreen-Rose-Wyplosz); above μ+1.5σ marks a crisis regime. The dollarization premium with placebo differencing is the descriptive core: the raw USDT/TRY premium shares an FX/microstructure component with all locally traded crypto, so we subtract the BTC/TRY premium — the USDT−BTC spread is what remains, the part attributable to dollar access specifically. For identification, the event study around crisis and political episodes is the primary causal-leaning evidence (it needs no instrument), and OLS local projections with a rich control set are reported as a conditional association, not a causal effect. Instrumental-variable attempts were tried and rejected as too weak (first-stage F < 10), so no IV estimate is relied upon here.
The findings. The raw USDT/TRY premium does rise around Turkish stress episodes — but the BTC/TRY placebo rises by a similar amount, so most of that move is common to all locally traded crypto, not USDT-specific. After differencing, the dollarization-specific response is small and within noise; the evidence is suggestive, not conclusive. The structural backdrop still makes the channel plausible: Turkey's standing bank dollarization (deposit/loan FC ratios and persistently negative net foreign assets, below) shows deep dollar reliance even where the high-frequency premium test cannot pin it down cleanly.
Event studies of crisis events in Turkey have shown weak correlation. I have not yet found an effective instrument variable to test causation. That being said, stablecoin adoption in emerging markets is still an emerging trend. It would not be expected to see significant correlation between economic crises and stablecoin flows in emerging markets just yet, because the local flows are still a small portion of the global flows. As stablecoin adoption in emerging markets continues, I expect the causation to become more clear.
Why it matters: it is this screen's trigger variable. Currency stress is what pushes retail savers out of lira and into dollars — increasingly via USDT — so EMP is the upstream signal for the whole emerging-market → stablecoin → Treasury channel.
How to read it:
• Formula: EMP = α·Δ%FX + β·Δpolicy_rate − γ·Δ%reserves (Eichengreen-Rose-Wyplosz, variance-equalised weights).
• Dashed line = crisis threshold (μ+1.5σ); readings above it mark a crisis regime.
• ▲ red = stress-increasing shock · ▼ green = stress-reducing (orthodox) shock.
The method. Exchange Market Pressure (EMP) is the stress gauge — a variance-weighted combination of FX depreciation, policy-rate moves and reserve drawdowns (Eichengreen-Rose-Wyplosz). It is computed twice, on the official and the blue rate, because the controls pin the official peso rate and real pressure escapes into the parallel market. The structural backdrop comes from the banking sector: loan dollarization (IMF FSI; the deposit ratio is not published for Argentina) and net foreign assets (positive — Argentine banks are net foreign-currency creditors, the opposite of Turkey). The ILY minimum-variance portfolio decomposes currency risk into inflation, FX and covariance pieces, using the IMF CPI and USD/ARS depreciation.
The findings. The blue-rate EMP currently sits below its crisis threshold — controls have been loosened and the blue premium has compressed, so official and blue pressure have largely re-converged. The ILY result is notable: Argentina's minimum-variance dollar share λ* sits roughly at its observed loan dollarization (~17%), meaning bank dollar reliance is close to what pure variance-hedging would justify — unlike Turkey, where observed dollarization runs far above λ* (excess, flight-driven dollarization).
Caveat. Argentina coverage is currently structural and descriptive — there is not yet a high-frequency causal test here (no exchange-level USDT premium series, placebo differencing or event study, as on the Turkey screen). As Argentine stablecoin data improves, more of the retail-flight channel will become directly measurable.
Two rates, because of the controls: Argentina's capital controls (the cepo) pin the official peso rate, so genuine depreciation pressure escapes into the parallel blue market. We therefore compute EMP twice — amber = blue rate (the primary, controls-aware signal, which sets the crisis threshold and regime) and cyan = official rate (suppressed during control episodes). The gap between them is itself a measure of how much stress the official rate is hiding.
Why it matters: currency stress pushes retail savers out of pesos and into dollars — increasingly via USDT — so EMP is the upstream signal for the whole emerging-market → stablecoin → Treasury channel.
How to read it:
• Formula: EMP = α·Δ%FX + β·Δpolicy_rate − γ·Δ%reserves (Eichengreen-Rose-Wyplosz, variance-equalised weights), computed on each rate.
• Dashed line = crisis threshold (μ+1.5σ, blue series); readings above it mark a crisis regime.
• Built from the ARS official & blue rates, BCRA policy rate and FX reserves.
Based on Ahmed & Aldasoro, BIS Working Paper No. 1270 (2025), with Observatory extensions
Stablecoin issuers are required by the recently passed GENIUS Act to hold 1:1 reserves in highly liquid assets. Major issuers Tether and Circle heavily rely on US Treasuries to back their coins. There is a clear theoretical link between demand for stablecoins and the US Treasury market: every newly minted coin requires the backing of a new liquid asset, and every burned coin means the sale of its equivalent in the issuer's reserves.
The intuition is straightforward: Tether takes in dollars, buys T-bills with them — increased demand for T-bills pushes their price up, yield falls. This is the forward channel. Simultaneously, there is a reverse channel: a stablecoin run forces issuers to liquidate T-bill holdings to satisfy redemptions, transmitting redemption pressure directly into the short end of the yield curve at exactly the moment when the Treasury market is already under stress.
Private issuers capture the exorbitant privilege — pocketing T-bill yield on reserves they hold. Public institutions bear the exorbitant duty — the Fed provides a de facto backstop in crisis. This asymmetry is the central policy concern. Every number on this terminal traces to a public source.
The intuition above is simple. But correlation does not equal causation. During normal market conditions, stablecoins are minted when crypto markets are booming and money is flowing freely. It is during times like this that investors are moving out of safe assets into riskier investments, pushing T-bill prices down and yields up. The macroeconomic causes of an increase in stablecoin flow are simultaneously the cause of upward pressure on T-bill yields.
A linear regression would tangle all these effects together, making it impossible to determine the extent to which stablecoin flows really affect the Treasury market. The question is whether Treasury market movements are driven by stablecoin flows or by broader market sentiment. The problem for statistical analysis is that stablecoin flows are endogenous — tangled up with the very thing you're trying to measure.
What you'd ideally want is episodes where stablecoin flows moved for reasons that had absolutely nothing to do with macro conditions, T-bill yields, risk sentiment, or anything else in the financial system. If you could isolate just those episodes, you could look at what yields did afterward and be confident it was the flows causing the yield movement — not anything else. An instrument variable provides exactly this: a variable that predicts stablecoin flows reliably but has no connection to Treasury yields except through those flows.
A BIS study (Ahmed & Aldasoro, WP 1270) built a Granular Instrument Variable (GIV) to measure the stablecoin–Treasury link. A GIV is a type of instrumental variable built from inside the market itself, using the concentration structure. Instead of hunting for some external variable that happens to work, the instrument is constructed mathematically from the cross-sectional variation between large units in a concentrated market.
The GIV is built from idiosyncratic shocks to large stablecoin-chain pairs. Stablecoins are issued not as standalone coins, but as coins present on a particular blockchain. In the stablecoin market, the top coins are heavily concentrated into particular chains: USDT on Tron is about a third of the entire market, and the top ten pairs (of 226 pairs in total) account for over 95% of market cap.
When something happens specific to one of those deployments — not because of macro conditions, but for internal operational reasons — it moves aggregate flows. But it has no reason to affect Treasury yields directly. These idiosyncratic shocks are exactly the kind of variation needed for causal analysis: they move flows while being disconnected from the financial-system conditions that also affect yields. This satisfies the two conditions for a valid instrument: relevance (the GIV predicts flows) and exclusion (the GIV affects yields only through flows, not directly).
Think of any pair's daily growth rate as being made up of two components: (1) the part everyone experienced — sector-wide conditions, crypto market sentiment, macro environment, whatever was pushing all stablecoins in the same direction that day; and (2) the part specific to that pair — something unique to that deployment, that chain, that issuer. The construction proceeds in three steps:
- Compute the daily growth rate of each pair's market cap.
- Subtract the weighted aggregate growth rate (the common factor). This measures the shared experience. When it is subtracted from a specific pair's growth rate, the shared component is removed.
- The residual is the idiosyncratic component — by elimination, it must be the pair-specific part, containing no macro variation.
The daily idiosyncratic residuals of all pairs are then combined into a single daily number, weighted by each pair's market cap share. Larger pairs get more weight — because larger pairs are the ones whose idiosyncratic shocks don't cancel out in aggregate, but are large enough to meaningfully move US Treasury demand. Five consecutive daily GIV values are summed to align with the 5-day stablecoin flow measure. The result: a continuous daily series capturing how much of aggregate market cap movement on any given day was driven by pair-specific factors rather than sector-wide common factors — by construction, containing no macro variation. Screen 3 plots this series live.
The first stage regression verifies that the GIV actually predicts stablecoin flows — the relevance condition. Here is the equation:
This asks: when the GIV is high — meaning large pairs had unusually strong idiosyncratic growth — did aggregate stablecoin flows actually increase? For the GIV to be a valid instrument, the answer must be yes, and the effect must be statistically strong.
The F-statistic on the first-stage coefficient is the standard diagnostic. An F-statistic above 10 is the conventional threshold in econometrics for a sufficiently strong instrument; below 10 indicates a weak instrument whose second-stage estimates cannot be trusted. Ahmed & Aldasoro's first-stage F-statistic was 54 — well above the threshold. The current live F-statistic calculated on this terminal from live data is shown on screen 3.
The second stage uses the GIV to instrument flows in the main regression. Rather than a single regression, a local projection is run — a series of regressions, one for each time horizon h from 0 to 40 trading days:
Where y is the 3-month T-bill yield. The βh coefficients trace out the path of the yield response over time — this is the impulse response function (IRF) shown on screen 4. Crucially, Flow(5d)t here is instrumented by GIV(5d)t — meaning only the variation in flows predicted by the GIV is used, not total observed flows. That's what makes the estimate causal rather than correlational.
A broad set of control variables absorbs remaining confounders: forward Treasury yield changes across the curve, crypto asset prices, money market fund flows, short-term Treasury ETF flows, T-bill auction supply, monetary policy surprises (3M OIS), financial conditions indices, and risk sentiment indicators. The standard errors are Newey-West HAC to account for serial correlation.
The analysis by Ahmed & Aldasoro determined that a $3.5B inflow of stablecoins — two standard deviations above the mean — compresses 3-month T-bill yields by approximately:
This is a seemingly small impact, but significant in the Treasury market — the deepest and most liquid fixed-income market in the world. Its significance grows materially as the stablecoin sector scales: a $2 trillion market producing the same percentage inflow generates six times the absolute dollar flow as today's $300 billion market. Screen 4 re-runs this estimation live on current data each week.
The stablecoin-to-yield effect is not constant across market conditions. The research finds it is "amplified during periods of Treasury-market intermediary stress, in particular during aggressive Treasury General Account (TGA) rebuilds and during episodes of impaired secondary-market liquidity in Treasuries." When the Treasury is rebuilding its cash balance, dealers have less capacity to absorb flows; when secondary market liquidity is impaired, price impact per dollar is mechanically larger. In those regimes, the same $3.5B inflow produces roughly double the yield impact. Screen 5 tracks these regime conditions live.
Furthermore, the structural channel has "strengthened with the maturation and scale of the stablecoin sector" — materially deeper price impact as the market has grown larger and more institutionally embedded. These two findings compound: the effect is larger in stress, and it has become structurally larger over time as the sector institutionalized.
The stablecoin market cap is now roughly $300 billion. Baseline forecasts project $2 trillion by 2030. At that scale, with the GENIUS Act formally integrating stablecoin issuers into short-end Treasury markets, a meaningful geopolitical shock in an emerging market — a central bank governor firing in Ankara, a capital control announcement in Buenos Aires — can, through the chain traced on this terminal, influence short-term US borrowing costs and Treasury market depth. This is a transmission mechanism with no formal policy oversight, and a legitimate cause for concern. The purpose of this terminal is to track it live as adoption increases.