Why is the impact of tariffs on the exchange rate relevant?

Do tariffs strengthen or weaken the currency? The textbook answer is unambiguous: by raising the domestic price of imported goods, a tariff shifts demand toward home goods and appreciates the exchange rate. Tommaso Monacelli and Alfonso Merendino show that U.S. data routinely point the other way — unexpected tariff increases are followed by a persistent depreciation of the dollar. The currency response is not a sideshow: it shapes import prices, the external accounts, the stance of monetary policy, and the real burden of dollar-denominated debt, and the question has gained urgency as the United States returns tariffs to the centre of its economic strategy. The paper shows that the textbook prediction is conditional: whether a tariff appreciates or depreciates the currency depends on Structural Trade-Policy Uncertainty (S-TPU), the uncertainty agents face about how persistent tariffs are. Put differently, S-TPU is the uncertainty about whether a tariff increase is a one-off, transitory disturbance or the onset of a persistent change in the trade-policy regime. S-TPU enters as an exogenous state of the economy that shapes how tariff shocks transmit.

Reaction of exchange rates react on tariff measures

The authors find that unexpected increases in U.S. import tariffs tend to be followed by a persistent weakening of the dollar (both in nominal and real terms) while output and investment fall, inflation falls in the short run and rises in the medium run, the trade balance improves and later deteriorates, and monetary policy turns accommodative. The exchange-rate response is state-dependent in S-TPU. When S-TPU is low, tariffs deliver the textbook outcome — appreciation, tighter policy, higher activity and inflation; when S-TPU is high, the same shock depreciates the currency, contracts activity and inflation, and elicits a more accommodative stance. Conditioning instead on a more traditional (and broader) trade policy uncertainty index removes the asymmetry, so it is structural, not noise-driven, trade policy uncertainty that governs transmission. S-TPU rises around major, durable shifts in trade policy (the 2017 exit from the Trans-Pacific Partnership, the NAFTA–USMCA renegotiation) and stays muted when the uncertainty is merely transitory — i.e., announcement noise about the level of tariffs that will actually apply, which dissolves once the measures are implemented, as with the 2025 “Liberation Day” announcements. The shock also transmits to the euro area as an adverse demand shock.

Why depreciation can happen
A small open-economy model rationalizes these facts, and two ingredients are crucial. First, agents observe the tariff in force but not whether it is temporary or durable; when S-TPU is high, they attribute a larger share of any tariff surprise to a persistent change in tariffs and revise up expected future tariffs, and with them expected future trade surpluses. Second, international financial markets are incomplete: households borrow today against those anticipated surpluses, the net foreign asset position deteriorates, and through a modified interest-parity condition this is priced into a weaker currency today. The authors label this the persistence channel. In response to a transitory tariff, the persistence channel is absent under certainty, as the transitory measure is recognized as such and the standard expenditure-switching channel produces an appreciation. Under structural trade policy uncertainty, the impact response to a transitory tariff features an additional forward-looking persistence force, which pushes toward depreciation; S-TPU governs the relative weights of the static and the forward-looking forces, and above a threshold the (depreciation) persistence channel dominates. A low short-run trade elasticity reinforces the result. In short, misperceived persistence combined with incomplete markets makes depreciation an uncertainty state-contingent outcome, not a puzzle.

What data and which methods were applied?

The authors estimate a narrative Bayesian VAR on quarterly U.S. data for 1990–2025 (results are sharper on a 2002–2025 subsample) that includes output, investment, exports, imports, consumer prices, a trade-weighted dollar, the federal funds rate, a macroeconomic-uncertainty index, and the import tariff. Tariff shocks are identified by a narrative-dominance scheme anchored on clearly tariff-driven episodes (the 2018 Section 301 lists and the 2025 “Liberation Day” package). The scheme leaves the sign of every impulse response free — the exchange rate included — with the single exception that real imports are restricted to fall at horizons of one and two quarters – imports are not restricted on impact to accommodate the front-loading of shipments ahead of tariff hikes. The S-TPU index is estimated from a state-space stochastic-volatility model of tariff rates that splits the observed tariff into a persistent and a transitory component, with S-TPU defined as the volatility of innovations to the persistent component. State-dependent local projections, with S-TPU as an exogenous conditioning state, trace how the transmission of tariff shocks flips across high- and low-uncertainty regimes. A small open-economy New Keynesian model with incomplete financial markets and Bayesian learning about tariff persistence rationalizes the evidence, and the results are robust across tariff and exchange-rate measures, sample windows, and identification choices.

What policymakers should take away 

There are three practical lessons. First, tariffs are macroeconomic, not merely sectoral, interventions: they move the exchange rate, real activity, inflation, the external accounts, and monetary policy. Second, their effects on the exchange rate and the broader economy are not signed in advance: the same tariff shock can appreciate the currency when it is read as a one-off transitory shock, or depreciate it when read as the onset of a persistent trade-policy regime change — so the prevailing state of structural uncertainty governs the wider macroeconomic effects of tariffs. Third, policymakers should therefore monitor narrow, structural measures of trade-policy uncertainty, rather than broad ones, to gauge the expected costs of protection.

Tariffs, Uncertainty, and the Exchange Rate

Authors:

Alfonso Merendino (Bocconi University)

Tommaso Monacelli (Bocconi University, IGIER and CEPR)

(paper presented at the 2nd Economic Policy: Papers on European and Global Issues Conference)