A portfolio balance explanation of the 2025 dollar sell-off
President Trump’s April 2, 2025 tariff announcement produced a pattern that standard theory struggles with: the dollar depreciated while U.S. yields rose. This was the opposite of the usual flight-to-safety response, and inconsistent with the textbook effect of tariffs on a country’s terms of trade.
With commentators openly questioning the dollar’s reserve-currency status in the wake of that episode, Rohan Kekre and Moritz Lenel set out, in this paper, to test quantitatively, whether a plausible decline in foreign demand for dollar bonds, rather than the tariffs themselves or a generic flight to safety, could actually explain what happened.
Methodology
Kekre and Lenel build a general-equilibrium, multi-country portfolio-balance model in the Tobin-Kouri tradition:
- Households in each country trade their own local short-term bond freely, but hold largely inelastic and exogenously given positions in foreign bonds and long-term debt.
- Global arbitrageurs trade all asset and when foreign demand for a country’s bonds shifts, arbitrageurs must absorb the resulting flow. Because arbitrageurs are risk-averse, absorbing an unwanted flow requires currency and bond risk premia to move, which shows up as changes in exchange rates and yield curves.
- Importantly, the model treats a shift in demand for dollar bonds as coming from either private investors rebalancing or a government “managing its reserves” as official and private flows work through the identical channel.
- Price elasticities are disciplined using the observed market impact of the Federal Reserve’s 2008–09 QE1 announcements. The model is calibrated to the G10 economies plus 15 emerging markets, then used to run the April 2025 tariff shock.
Findings:
- A 3–12% decline in global demand for dollar bonds relative to annual U.S. GDP is enough to reproduce both the dollar’s depreciation and the rise in dollar yields.
- Because asset prices are forward-looking, this price impact appears even if the actual rebalancing is expected to unfold gradually rather than all at once.
- Currencies offering the highest average excess returns appreciated the least, and higher-yielding long-term bonds saw the largest yield increases. This is a pattern attributed to a simultaneous rise in arbitrageur risk aversion, not to tariffs themselves.
- Countries with the largest external surpluses and thus the steepest announced tariffs, appreciated the most against the dollar, precisely because they were perceived as “safer.” Tariffs alone cannot explain this.
- The same forces continued through the following year: the dollar kept weakening against G10 currencies as rate differentials narrowed, and against emerging markets as risk aversion eased.

What Do Asset Prices in April 2025 Say About Demand for the Dollar?
Authors: Rohan Kekre, Moritz Lenel
From: University of California, Berkeley, Princeton University
Can the Kekre and Lenel paper explain the recent yen intervention?
The Kekre-Lenel model’s core mechanism i.e. a shift in demand for dollar bonds that arbitrageurs must absorb, moving risk premia, exchange rates, and yields, is not tied to April 2025 tariff announcement specifically, nor to private investors alone. The model explicitly allows this demand shock to originate with a government managing its official reserves, making it directly applicable to the U.S.-Japan yen intervention of late July/early August 2026.
Japan holds more U.S. Treasuries than any other official foreign holder. Funding large-scale yen purchases by selling even a portion of that stockpile would have produced exactly the kind of sudden drop in foreign dollar-bond demand that the model identifies as the trigger for rising U.S. yields and dollar depreciation.
The two governments avoided this channel by design: the New York Fed sold euros, not dollars, to fund the yen purchases on Treasury’s behalf, and Japan borrowed dollars against its existing Treasury holdings through the Fed’s FIMA repo facility instead of liquidating them. Both moves were explicitly designed to avoid adding to the supply of dollar bonds hitting the market, exactly the kind of portfolio flow that Kekre and Lenel model as the driver of dollar weakness and rising yields.
Academic reactions worth reading:
- Barry Eichengreen (UC Berkeley), in a Financial Times commentary, argued the intervention’s design i.e. avoiding Treasury sales at almost any cost, is itself the story: “the dollar is not the attractive reserve currency it once was… when this message sinks in, other countries will redouble their search for more attractive, readily usable alternatives.” This is essentially a real-time, qualitative version of the Kekre-Lenel mechanism: a shift in the perceived elasticity of dollar-bond demand shows up in prices and policy behavior well before any actual portfolio rebalancing occurs.
- Robin Brooks (Brookings) in his Substack blog argues that the US Treasury market’s daily turnover is around $1 trillion. “So it’s not obvious that Japan’s intervention will have driven yields higher single-handedly…Inevitably, markets will now be asking if the EUR/JPY twist reflects US concern that Japan’s interventions are pushing US yields higher. If there’s something to this, markets will see this as a sign of weakness, which will further undercut the already embattled efficacy of Yen intervention.”