Decoded, Economics & finance

Tariff Whiplash: What Five US Tariff Regimes Reveal About the Swiss Export Model

Tariff Whiplash as a Natural Experiment
The Swiss Economy in 2026 | Analytical Series, Paper 1

Executive Summary

Between April 2025 and August 2026 the additional duty that the United States levied on Swiss goods moved through five distinct legal regimes: a 10 percent baseline, a 39 percent "reciprocal" tariff, a 15 percent framework rate, a 10 percent temporary surcharge after the US Supreme Court voided the emergency-powers tariffs, and a 12.5 percent duty under Section 301. Few small open economies have ever been subjected to so many policy switches in so short a time. For the analyst, this sequence is a rare gift: a natural experiment in which the tariff rate rose, collapsed and rose again while most other drivers of Swiss exports stayed broadly constant.

This article exploits that experiment. Using monthly customs data published through the Swiss National Bank data portal (January 2014 to August 2026), we build three independent counterfactuals for Swiss goods exports to the United States, validate them with rolling out-of-sample placebo tests, and then ask a simple question: how much of the observed export loss is explained by the level of the tariff itself?

The answer overturns the conventional reading of the tariff episode. Our main findings are the following.

  1. The cumulative export balance with the United States turned negative in March 2026. The pre-tariff front-loading of the first quarter of 2025 created an excess of about CHF 5.3 billion relative to the counterfactual. That cushion was exhausted after exactly twelve months. By August 2026 the cumulative shortfall stood at roughly CHF 3.9 billion, and the January to August 2026 shortfall alone was about CHF 6.1 billion.
  2. The tariff rate does not explain the gap. Once pharmaceutical shipment pulses and the anticipation window are controlled for, the estimated semi-elasticity of exports to the tariff rate is statistically indistinguishable from zero (plus 0.05 percent per percentage point, standard error 0.11). Exports to the US were not lower when the tariff was 39 percent than when it was 10 percent. The gap was, in fact, deepest in 2026, when tariffs were at their lowest.
  3. What explains the gap is a steady drift. The export gap widens by about 1.5 percent per month (standard error 0.2), independently of the tariff in force. This is the statistical signature of relocation, not of price competition: the investment pledges made by Swiss firms to secure the November 2025 framework are being converted into US capacity, and US-bound volumes are being supplied from inside the United States.
  4. Volatility, not the tariff, is the real tax. The standard deviation of monthly growth in exports to the US more than doubled, from 13.8 percent before 2025 to 29.3 percent since (Levene test p-value below 0.001). A Markov-switching model identifies pharmaceutical exports as operating in a high-volatility "pulse" regime for 98 percent of the first half of 2025 and for 89 percent of May to August 2026. Two of these pulses (March 2025 and July 2026) were more than five standard deviations above anything observed in the decade before.
  5. A structural break appeared in Swiss exports to Belgium in February 2026: a level jump of 76 log points (standard error 5), equivalent to an extra CHF 3.1 billion in eight months, roughly half the size of the 2026 US shortfall. The trade geography of Swiss pharmaceuticals is being redrawn, and the US bilateral statistics no longer capture the full picture.
  6. For 2027, our Monte Carlo scenarios put the export shortfall relative to the counterfactual between CHF 7 billion (if a binding agreement partially reverses the drift) and CHF 14 billion (if relocation continues on its current path). The most important variable is not the tariff rate that Washington sets but the speed at which pledged US capacity comes online.

Key results at a glance

Indicator Value
Front-loading excess, Jan to Mar 2025 CHF 5.3 bn (+31% vs counterfactual)
Month in which cumulative balance turned negative March 2026
Shortfall vs counterfactual, Jan to Aug 2026 CHF 6.1 bn (about -18%, range -11% to -22%)
Tariff semi-elasticity of exports to the US +0.05% per pp (s.e. 0.11), not significant
Tariff-independent drift of the export gap -1.5% per month (s.e. 0.2)
Volatility of monthly US export growth 13.8% before 2025, 29.3% since
Largest pharma pulses (pre-2025 s.d. units) 6.2 (Mar 2025), 5.0 (Jul 2026); prior max 3.5
Belgium level break from Feb 2026 +76 log points; CHF 3.1 bn extra in 8 months
Projected 2027 shortfall, central relocation scenario CHF 14.4 bn (80% interval 12.3 to 16.3)

Source: author's calculations based on FOCBS and SECO data via the SNB data portal.

The policy conclusion is uncomfortable. Switzerland negotiated for sixteen months as if the problem were the tariff rate. The data say that the problem is the location of production. A lower tariff, on its own, buys very little; what matters is whether the Swiss export model can retain the high-value stages of production at home while the volume stages migrate.

1. Introduction: Sixteen Months as a Laboratory

Economists rarely get to watch a tariff move up and down repeatedly on the same trading relationship while the rest of the world stays roughly still. Most trade-policy evidence comes either from one-off shocks, whose effects are hard to separate from the business cycle, or from cross-country panels, where the tariff changes are small and the countries differ in a hundred unobserved ways. Switzerland between 2025 and 2026 offers something closer to a controlled experiment. The same exporter, selling broadly the same goods to the same destination, faced a tariff of zero, then 10 percent, then 39 percent, then 15 percent, then 10 percent again, then 12.5 percent, each change driven by legal and political events in Washington rather than by anything happening in the Swiss economy.

The public debate in Switzerland treated each of these changes as an independent event. The 39 percent tariff imposed in August 2025 was described as a shock to competitiveness. The November 2025 framework that brought the rate down to 15 percent was celebrated as a rescue, with the KOF institute estimating that it would add 0.3 to 0.5 percent to GDP and save between 7,500 and 15,000 manufacturing jobs. The Supreme Court ruling of February 2026 was greeted as a vindication and as a source of refunds, estimated by UBS economists at up to 2 billion dollars. The Section 301 duty of July 2026 was read as a new setback. In each case the implicit model was the same: a higher tariff means fewer exports, a lower tariff means more.

This article tests that implicit model against the data, and finds that it fails. The reason is not that tariffs are irrelevant. It is that the tariff episode triggered a second, slower process, the relocation of production capacity to the United States, which now dominates the export data and which does not respond to the tariff rate in the way that price-based trade models assume. Once that process is under way, a lower tariff does not bring the volumes back, because the volumes are no longer produced in Switzerland.

The article proceeds as follows. Section 2 reconstructs the chronology of the five tariff regimes and their legal basis. Section 3 describes the data and the construction of the counterfactual, which is the central methodological problem of any tariff evaluation. Sections 4 to 9 present the results: the size and timing of the export gap, its insensitivity to the tariff rate, the role of pharmaceutical shipment pulses, the emergence of a Belgian trade channel, the resilience of exports to the rest of the world, and the macroeconomic cost of volatility. Section 10 interprets these results through a relocation-hysteresis lens. Section 11 projects scenarios for 2027. Section 12 draws the implications for Swiss economic strategy, and Section 13 concludes. A methodological appendix documents the models.

2. The Chronology: Five Regimes in Sixteen Months

The sequence of US measures affecting Swiss goods is complex because it rests on four different legal instruments, each with its own logic and its own exemptions. Table 1 summarizes the regimes as they applied to the bulk of Swiss non-pharmaceutical goods, which is the relevant rate for our tariff variable.

Table 1. US tariff regimes applied to Swiss goods, April 2025 to August 2026

Regime Period Legal basis Additional duty, most goods Pharmaceuticals
R0 Threat window Jan to 4 Apr 2025 None (announcements) 0% Front-loading of shipments
R1 Baseline 5 Apr to 6 Aug 2025 IEEPA, universal baseline 10% Largely exempt
R2 Reciprocal 7 Aug to 13 Nov 2025 IEEPA, country rate 39% Largely exempt
R3 Framework 14 Nov 2025 to 23 Feb 2026 IEEPA, framework of 14 Nov 15% (ceiling) Future sectoral duties capped at 15%
R4 Post-ruling 24 Feb to 23 Jul 2026 Trade Act 1974, Sec. 122 10% (150 days max.) Sec. 232 duties announced 2 Apr 2026
R5 Forced-labour action From 24 Jul 2026 Trade Act 1974, Sec. 301 12.5% total incl. MFN Swiss cap from 31 Jul 2026 to 20 Jan 2029

Sources: SECO, USTR, US Federal Register, economiesuisse. IEEPA tariffs were declared unlawful by the US Supreme Court on 20 February 2026.

The first regime opened on 5 April 2025 with the 10 percent baseline tariff that Washington applied to almost all trading partners. The "reciprocal" rate announced for Switzerland on 2 April, 31 percent, was suspended a week later. During this phase Swiss exporters, and above all pharmaceutical exporters, had already moved: the first quarter of 2025 saw an unprecedented surge in shipments to the US, as importers built inventories ahead of the expected duties.

The second regime began on 7 August 2025, when the United States imposed a 39 percent tariff on most Swiss goods, one of the highest rates applied to any advanced economy. The measure took Bern by surprise. It was justified by Washington with reference to the bilateral trade deficit, a deficit that was itself heavily inflated by flows of refined gold, as the SNB argued at the time. The Swiss economy contracted in the third quarter of 2025, its first quarterly contraction since 2023.

The third regime followed the framework agreement of 14 November 2025. The United States cut the tariff to 15 percent, matching the rate applied to the European Union and Japan, and capped at 15 percent any future sectoral duties on Swiss pharmaceuticals and semiconductors. In exchange, Swiss companies pledged to invest 200 billion dollars in the United States by the end of 2028, with the White House stating that 67 billion dollars of that amount was expected in 2026. US Trade Representative Jamieson Greer was explicit about the aim: moving manufacturing of pharmaceuticals, gold smelting and railway equipment to the United States.

The fourth regime was the product of the US courts rather than of negotiation. On 20 February 2026 the Supreme Court ruled, by six votes to three, that the International Emergency Economic Powers Act did not authorize tariffs. From 24 February the administration replaced the voided duties with a temporary 10 percent surcharge under Section 122 of the Trade Act of 1974, a provision limited to 150 days. Paradoxically, Switzerland thus found itself facing a lower rate than the European Union, whose 15 percent rested on a negotiated arrangement.

The fifth regime started on 24 July 2026, when the Section 122 surcharge expired and the United States introduced a duty under Section 301, justified by an investigation into imports produced with forced labour. For Swiss goods the total duty, most-favoured-nation rate included, was set at 12.5 percent; where the MFN rate already exceeds 12.5 percent, only the MFN rate applies. In parallel, Section 232 duties on pharmaceutical products, announced on 2 April 2026, introduced an additional tariff of up to 15 percent on Swiss pharmaceuticals with numerous exemptions. A country-specific cap took effect on 31 July 2026 and runs until 20 January 2029. Several of the large Swiss and multinational producers had in any case signed confidential pricing agreements with the US administration that exempt them from tariffs for three years.

Figure 1. Five US tariff regimes for Swiss goods in sixteen months
Figure 1. Five US tariff regimes for Swiss goods in sixteen months. Source: author's reconstruction from SECO, USTR and US Federal Register notices. The monthly rate is a day-weighted average of the additional duty applicable to non-exempt Swiss goods.

Two features of this chronology matter for what follows. First, the tariff rate did not move in one direction. It rose sharply, fell back, fell again and then rose slightly, so that time and the tariff level are not confounded in a simple way. If exports were governed by the tariff rate, they should have recovered in the first half of 2026, when the rate was back at 10 percent. Second, the most consequential commitments made in this period were not tariff commitments at all, but investment commitments. The tariff rate was the visible variable; the location of production was the variable actually being negotiated.

3. Data and the Counterfactual Problem

3.1 Data

The analysis relies on official monthly foreign trade statistics compiled by the Federal Office for Customs and Border Security (FOCBS) and published on the SNB data portal. We use two tables. The first gives Swiss goods exports by goods category, in CHF millions, from 2012. The second gives exports by destination country. Both refer to the business-cycle total used by the SNB and SECO, which excludes non-monetary gold and other precious metals, precious stones and works of art. This exclusion is essential. Swiss gold flows to and from the United States are enormous and were themselves affected by the tariff episode, but they generate very little domestic value added and would swamp any analysis of real export performance. We complement the trade data with monthly average exchange rates from the same portal and with quarterly real GDP from SECO's national accounts. The sample runs from January 2014 to August 2026, the last month available at the time of writing.

Over this period, the United States became Switzerland's dominant export market outside Europe. Its share of total Swiss exports rose steadily from 12.4 percent in 2014 to 18.6 percent in 2024, overtaking Germany as the single largest destination by value from 2021. In 2025 that share held at 18.6 percent, flattered by the front-loading surge. In the first eight months of 2026 it fell to 15.6 percent, its lowest level since 2017.

3.2 Why the counterfactual is the whole problem

Measuring the effect of a tariff requires an answer to a hypothetical question: what would Swiss exports to the United States have been in the absence of the tariff episode? Every estimate in the public debate rests, explicitly or implicitly, on an answer to that question, and most of them use the crudest possible answer, namely last year's figure. A year-on-year comparison is badly suited to this episode, because the base period itself was distorted by front-loading. The 86 percent year-on-year jump in watch exports to the US in September 2026, for instance, says more about the collapse of September 2025 than about any recovery in 2026.

We therefore construct three independent counterfactuals, each with different strengths and weaknesses, and treat their dispersion as a measure of model uncertainty.

Model A is a time-series model of the share of the United States in Swiss exports. We model the log ratio of exports to the US over exports to all other destinations as a seasonal autoregressive process with a quadratic trend and the euro-dollar cross rate as an exogenous variable, estimated on January 2014 to December 2024 and projected forward. This model captures the long rise of the US as a destination and the effect of exchange rates on its relative attractiveness.

Model B is a deliberately simple benchmark: the US share of exports is frozen at its 2024 average. This model makes no assumption about the continuation of trends and therefore provides a conservative counterfactual if the pre-2025 rise of the US share was about to level off in any case.

Model C is a synthetic control. We construct a weighted combination of other destination markets (Germany, France, Italy, the United Kingdom, China, Hong Kong, Japan, Singapore, India, the United Arab Emirates, the Netherlands, Spain and Canada) that best reproduces the path of exports to the US over 2016 to 2024, allowing for a US-specific linear drift. The optimal weights fall on France (0.23), Germany (0.20), Japan (0.14), the Netherlands (0.06), Spain (0.04), China (0.04), Canada (0.04) and India (0.01). This model uses the actual post-2025 evolution of other markets, so it automatically absorbs common shocks such as the global slowdown or the energy price surge linked to the conflict in the Middle East.

3.3 Validating the counterfactuals

A counterfactual is only as credible as its out-of-sample record. We test Models A and B with rolling placebo exercises: we re-estimate each model with data ending at nine different dates between December 2018 and December 2022, project it twenty months forward, which is the length of our post-shock window, and compare the projection with what actually happened, in a period when no tariff shock occurred.

Table 2. Out-of-sample performance of the counterfactual models (rolling placebo tests, 20-month horizon)

Model Approach Mean error (log pts) Mean abs. error (log pts)
A Seasonal AR model of US share, quadratic trend, FX -1.2 2.8
B US share frozen at previous-year average 2.3 2.9
C Synthetic control from 13 donor markets with drift n.a. 9.5 (pre-period RMSE)

Nine rolling origins, December 2018 to December 2022. Source: author's calculations.

The results are reassuring. Model A has an average placebo bias of minus 1.2 log points and a mean absolute error of 2.8 log points over twenty-month horizons. Model B has a bias of plus 2.3 log points and a mean absolute error of 2.9 log points. Model C tracks the pre-period with a residual standard deviation of 9.5 log points, a larger month-to-month error, as expected from a cross-country construction, but with no systematic drift. In other words, in normal times our counterfactuals miss the average level of exports to the US by about 3 percent over a twenty-month window. Any gap materially larger than that is unlikely to be a modelling artifact.

The three models are combined into an equal-weighted ensemble, and the range spanned by the three is reported as the uncertainty band in all figures and tables. We prefer this transparent approach to a single "best" model, because the disagreement between reasonable counterfactuals is itself information: where the three agree, the result is robust; where they diverge, the reader should be cautious.

3.4 Why year-on-year comparisons mislead

The importance of a proper counterfactual can be illustrated with a simple example. In September 2026, Swiss watch exports to the United States rose by 86 percent year on year, and total watch exports by 11.7 percent. Read at face value, these figures describe a spectacular recovery. In fact the watch federation itself attributed the jump primarily to a favourable base effect: September 2025 was the first full month of the 39 percent tariff, when shipments to the US had collapsed by more than half. A comparison with a depressed base produces an apparent boom even if the underlying level remains below normal.

The same distortion runs in the opposite direction for 2026 as a whole. Comparisons of the first half of 2026 with the first half of 2025 suggested a fall of almost 15 percent in watch exports to the US, but the first half of 2025 had been inflated by front-loading, when exports to the US surged by around 150 percent in April 2025. Neither the 86 percent rise nor the 15 percent fall says anything about the level that Swiss exports would have reached without the tariff saga. Only a counterfactual can answer that question, which is why the rest of this article relies on one.

More generally, any series that has been subjected to front-loading and repayment will produce misleading year-on-year growth rates for at least two years: twelve months of distortion caused by the shock itself, and twelve further months of distortion caused by the base effect. Switzerland's foreign trade statistics are in the middle of this second phase, and the public interpretation of monthly releases is likely to remain confused until well into 2027.

4. Result One: The Size and Timing of the Export Gap

Figure 2 compares actual Swiss exports to the United States with the ensemble counterfactual. The picture is one of three distinct phases.

Figure 2. Swiss goods exports to the US versus a no-tariff-shock counterfactual
Figure 2. Swiss goods exports to the US versus a no-tariff-shock counterfactual. Source: FOCBS via SNB data portal; author's calculations. Counterfactual is the mean of Models A, B and C; the shaded band spans the three models.

The first phase, from January to April 2025, is one of front-loading. In March 2025 alone, exports to the US reached CHF 9.5 billion, more than double the counterfactual of about CHF 4.4 billion. Over the three months of the threat window, the excess over the counterfactual totals about CHF 5.3 billion, or 31 percent above normal (range across models: 26 to 37 percent). Importers, anticipating duties, pulled forward shipments that would otherwise have been spread over the following year.

The second phase, from May 2025 to February 2026, is one of repayment. Exports to the US fell below the counterfactual in most months, by 9 percent on average during the 10 percent baseline regime, by 9 percent during the 39 percent shock and by 14 percent under the 15 percent framework. Part of this shortfall is the mechanical counterpart of the front-loading: inventories built in early 2025 were drawn down, and new orders were deferred. The cumulative balance, positive at CHF 6.1 billion at its peak in April 2025, declined month after month.

The third phase began in March 2026, when the cumulative balance turned negative. From that point on, the shortfall can no longer be read as a repayment of earlier front-loading. It is a net loss. Between January and August 2026 exports to the US were about CHF 6.1 billion below the counterfactual, a gap of 18 percent on average (range 11 to 22 percent across models), and the gap in the most recent two months, under the Section 301 regime, reached 22 percent.

Table 3. Swiss exports to the US versus counterfactual, by tariff regime (CHF million)

Regime Mths Avg. tariff % Actual Counter-factual Gap Gap % Range %
R0 Threat window 3 0.0 18 650 13 373 5 277 31.2 26 to 37
R1 10% baseline 4 9.7 16 122 17 302 -1 180 -8.6 -13 to -3
R2 39% shock 3 37.1 11 838 12 991 -1 153 -9.4 -15 to -2
R3 15% framework 4 17.4 14 780 17 107 -2 327 -14.2 -20 to -6
R4 10% Section 122 4 10.0 15 371 17 844 -2 473 -14.5 -20 to -7
R5 12.5% Section 301 2 11.6 7 434 9 441 -2 007 -22.5 -27 to -17
Total Jan 2025 to Aug 2026 20 84 195 88 060 -3 865

Gap % is the exponential of the mean monthly ensemble log gap; range spans Models A, B and C. Source: author's calculations.

Table 3 and Figure 3 summarize the gap by regime. The ordering is striking. The gap is not largest under the 39 percent regime. It is largest under the two regimes with the lowest tariffs: 15 percent under the Section 122 surcharge of 10 percent, and 22 percent under the 12.5 percent Section 301 regime. If the tariff rate were the main driver, the bars in Figure 3 would track the tariff line in Figure 1. They do not.

Figure 3. Export gap versus counterfactual by tariff regime, % (range across models in brackets)
Figure 3. Export gap versus counterfactual by tariff regime, % (range across models in brackets). Source: author's calculations. Bars show the ensemble gap; the range across the three counterfactual models is shown in brackets.

5. Result Two: The Rate Does Not Matter

The visual impression from Figure 3 can be formalized. We regress the monthly log gap between actual and counterfactual exports on four variables: the average additional tariff in force during the month; a measure of pharmaceutical shipment pulses (the deviation of total pharmaceutical exports from their centered twelve-month median, which captures lumpy batch shipments that are unrelated to policy); a dummy for the pre-Liberation Day anticipation month of March 2025; and a linear time trend starting in January 2025. Standard errors are robust to heteroskedasticity and autocorrelation.

Table 4. What explains the monthly export gap? (dependent variable: log gap x 100)

Regressor Ensemble Model A Model B Model C
Tariff rate (per pp) +0.05 (0.11) +0.10 (0.11) +0.12 (0.12) -0.06 (0.10)
Pharma pulse (per log pt) +0.79 (0.21) +0.67 (0.19) +0.74 (0.21) +0.96 (0.22)
Anticipation, Mar 2025 +48.37 (8.07) +51.47 (7.21) +50.59 (8.21) +43.05 (9.10)
Monthly trend -1.49 (0.22) -1.22 (0.23) -1.51 (0.25) -1.73 (0.22)
Tariff, no-trend spec. -0.02 (0.28) +0.04 (0.23) +0.05 (0.29) -0.14 (0.31)
R-squared 0.83 0.79 0.81 0.88

January 2025 to August 2026, 20 observations. Newey-West standard errors (3 lags) in parentheses. Pharma pulse coefficient is an elasticity. Source: author's calculations.

Three results stand out.

First, the tariff coefficient is essentially zero. In the ensemble specification it is plus 0.05 percent per percentage point of tariff, with a standard error of 0.11. The 95 percent confidence interval runs from minus 0.16 to plus 0.26 percent per point. Taken at face value, this means that the data can rule out any effect of the tariff rate on exports to the US larger than about 1.6 percent for each 10 percentage points of tariff. For comparison, conventional trade-model calibrations imply that a 10 point tariff increase should cut bilateral imports by something on the order of 10 to 40 percent over one to two years. The result holds in each of the three counterfactuals separately, and it holds when the trend is dropped from the regression (the tariff coefficient then becomes minus 0.02 percent per point, standard error 0.28).

Second, the trend is large and very precisely estimated: minus 1.5 percent per month, with a standard error of 0.2. Across the three models it ranges from minus 1.2 to minus 1.7 percent per month. Over the twenty months of the sample, this cumulates to roughly 30 log points. Nothing in the tariff path explains this drift.

Third, pharmaceutical pulses and the anticipation window together account for most of the month-to-month volatility. A pharmaceutical shipment 10 percent above its trend raises exports to the US by about 8 percent relative to the counterfactual, confirming that pharmaceuticals are the transmission belt between Swiss production and the US market. The anticipation month adds 48 log points.

Figure 4 shows the same result in a single picture. After removing the effect of pharmaceutical pulses and of the anticipation month, the monthly export gap is plotted against the tariff in force. The fitted line is flat. The months of the 39 percent regime (August to October 2025) sit above the cloud of 2026 observations, which were recorded under tariffs of 10 to 15 percent.

Figure 4. Net of pharma pulses, the tariff rate does not explain the gap
Figure 4. Net of pharma pulses, the tariff rate does not explain the gap. Source: author's calculations. Partial residuals from the regression in Table 4, excluding the tariff and trend terms. The dashed line is the bivariate fit (slope minus 0.02 log points per tariff point, standard error 0.27).

Why would exports to the US be insensitive to a tariff rate that varied between 10 and 39 percent? Three mechanisms, which are not mutually exclusive, are consistent with the evidence.

The first is product composition. Pharmaceuticals, which account for roughly half or more of Swiss exports to the US, were largely exempt from the IEEPA tariffs, and the November 2025 framework capped future sectoral duties on them at 15 percent. For this dominant product group, the headline rate was simply not the operative rate. But this mechanism cannot explain the drift: if pharmaceuticals were unaffected by tariffs, they should have continued to grow along their pre-2025 path.

The second is pricing power. Swiss watches, precision instruments and specialty chemicals are sold in segments where price elasticities are low and where Swiss producers could absorb part of the tariff in their margins or pass it to US consumers. This mechanism explains why the 39 percent regime did less damage than feared to volumes. It does not explain the drift either.

The third mechanism, and the only one consistent with a steady trend independent of the tariff, is relocation. When a firm commits to building or expanding capacity in the United States, the US market is progressively supplied from that capacity rather than from Switzerland. This process is slow, it is not reversed when tariffs fall, and it shows up in the trade data precisely as a gradual, tariff-insensitive decline of the US share. We return to this interpretation in Section 10.

6. Result Three: Pharmaceuticals as the Transmission Belt

Pharmaceutical exports have always been lumpy. A single batch of a high-value biologic can be worth hundreds of millions of francs, and the timing of shipments depends on production campaigns, regulatory releases and the inventory strategies of the multinational groups that dominate the sector. Before attributing any monthly movement to tariffs, one must therefore know what normal lumpiness looks like.

To answer that question, we estimate a two-regime Markov-switching model on the monthly growth rate of total Swiss pharmaceutical exports from 2014 to 2026. The model lets the data sort months into a "calm" regime and a "pulse" regime with different volatilities, and estimates the probability of being in each regime at each point in time.

Figure 5. Pharmaceutical exports (CHF bn per month, top) and probability of a
Figure 5. Pharmaceutical exports (CHF bn per month, top) and probability of a "pulse" regime (bottom). Source: FOCBS via SNB data portal; author's Markov-switching estimates. The lower panel shows the smoothed probability of the high-volatility regime.

The model identifies two well-separated regimes. In the calm regime the monthly growth rate has a standard deviation of 7.0 percent; in the pulse regime it is 18.3 percent, more than two and a half times larger. Calm phases last on average sixteen months and pulse phases about six and a half months. Before 2025 the pulse regime was not rare: it was active about 20 percent of the time, notably in 2019, early 2022 and late 2023, all periods with no trade-policy shock. This is the baseline against which the tariff period must be judged.

Against that baseline, the tariff period stands out in two ways.

First, the pulse regime has become the norm rather than the exception. Its probability averaged 98 percent in the first half of 2025, fell to 21 percent during the relatively quiet period from July 2025 to March 2026, and rose again to 89 percent between May and August 2026, the window between the announcement of Section 232 pharmaceutical duties on 2 April 2026 and the entry into force of the Swiss-specific cap on 31 July. Overall, 55 percent of months since January 2025 are classified as pulse months, against 20 percent before.

Second, the size of the pulses has changed. We measure each month's pharmaceutical exports as a deviation from their centered twelve-month median and express it in units of the pre-2025 standard deviation. Before 2025, the largest pulse on record was 3.5 standard deviations (September 2023). In March 2025 the pulse reached 6.2 standard deviations, and in July 2026 it reached 5.0. The collapse that followed in August 2026, minus 2.2 standard deviations, is consistent with the drawdown of inventories built in July. These are not normal fluctuations: they lie far outside anything in the twelve-year historical record.

The July 2026 episode is particularly instructive because it did not go mainly to the US. Total pharmaceutical exports jumped by CHF 2.2 billion between June and July 2026, but exports to the US were broadly flat (CHF 4.4 billion in July against 4.5 billion in June), while exports to Italy and China rose by a combined CHF 0.8 billion. This suggests that the 2026 pulses reflect a reorganization of the multinational supply chain ahead of the Section 232 regime rather than a simple rush into US warehouses. Swiss production is increasingly routed through other hubs before reaching the United States, which brings us to the fourth result.

7. Result Four: The Belgian Anomaly

Among all Swiss export destinations, one series behaves in a way that has no precedent in the twelve-year sample. Exports to Belgium, which had fluctuated around CHF 400 million per month for years, jumped to CHF 782 million in February 2026 and have stayed at roughly double their previous level ever since, peaking at CHF 1.07 billion in May.

Figure 6. The Belgian anomaly: a step change in Swiss exports from February 2026
Figure 6. The Belgian anomaly: a step change in Swiss exports from February 2026. Source: FOCBS via SNB data portal. The dotted line marks the 2023 to 2024 monthly average for Belgium.

A break test on the log of monthly exports to Belgium, with a linear trend and a level shift from February 2026, estimates the jump at 76 log points, with a standard error of 5. The change is not shared by the Netherlands, the other large logistics hub of the European Union, whose share of Swiss exports was stable. Between January and August 2026 Swiss exports to Belgium exceeded their 2023 to 2024 average by a cumulative CHF 3.1 billion, a sum equal to about half of the shortfall in exports to the US over the same months. The correlation between Belgian exports and total pharmaceutical exports in 2026 is 0.48.

The data do not allow us to identify the precise goods or the firms involved, and we refrain from speculation on individual companies. But three readings are consistent with the timing and the structure of the break. The first is a reallocation of fill-finish, packaging or distribution stages of pharmaceutical supply chains to Belgium, which hosts one of the densest clusters of pharmaceutical logistics and contract manufacturing in Europe. The second is a change in the routing of goods destined for third markets, including possibly the United States, through EU entities. The third is a change in the legal ownership of goods at the point of export, which can shift the recorded destination without any change in the physical flow.

Whatever its exact cause, the Belgian anomaly carries an important lesson for the interpretation of tariff effects. The bilateral US figure, which is the one watched by both governments, understates the reorganization under way. Part of what looks like lost US exports has reappeared as additional exports to the European Union. Conversely, the bilateral surplus that Washington uses as its benchmark is becoming an increasingly poor guide to the actual dependence of the US market on Swiss production. Watch exports offer a smaller parallel: the Swiss watch federation reported in October 2026 that France had served as a logistics hub for watch manufacturers because of lower US tariffs on EU imports.

8. The Resilience Paradox: Global Exports Hold, the US Share Falls

If the tariff episode had damaged the competitiveness of Swiss industry as a whole, it would show up in total exports to all destinations. It does not. Table 6 compares the first eight months of 2026 with the same months of 2024, the last year untouched by front-loading, for the main product groups.

Table 6. Swiss exports by product group, all destinations, January to August (CHF million)

Product group 2024 2025 2026 2026 vs 2024 2026 vs 2025
Pharmaceuticals 75 140 81 485 75 703 +0.7% -7.1%
Chemicals 21 441 23 369 23 261 +8.5% -0.5%
Watches 17 113 17 134 17 366 +1.5% +1.4%
Instruments and electronics 25 608 25 785 26 024 +1.6% +0.9%
Machinery 14 066 13 518 13 462 -4.3% -0.4%
Metal products 3 954 4 006 4 105 +3.8% +2.5%

Source: FOCBS via SNB data portal; author's calculations.

Five of the six main product groups exported more in January to August 2026 than in the same period of 2024. Pharmaceuticals were 0.7 percent higher, chemicals 8.5 percent, watches 1.5 percent, instruments and electronics 1.6 percent and metal products 3.8 percent. Only machinery, the sector most exposed to the European industrial cycle and to the 39 percent tariff, was lower, by 4.3 percent, in line with the sharp fall in shipments of machine tools reported by the industry association Swissmem in late 2025. Compared with 2025, pharmaceuticals appear 7 percent lower, but this comparison is distorted by the front-loading surge of the first quarter of 2025 and should not be read as a contraction of the sector.

Figure 8. Swiss exports by product group, all destinations (index, 2024 average = 100, 3-month average)
Figure 8. Swiss exports by product group, all destinations (index, 2024 average = 100, 3-month average). Source: FOCBS via SNB data portal; author's calculations. Three-month moving averages, all destinations.

Figure 8 confirms the picture over time. By August 2026, the three-month average of exports was above its 2024 level for every major group except machinery, and chemicals were 15 percent above it.

This is the resilience paradox. Swiss industry is selling as much as before, or more, to the world as a whole, while selling markedly less to the United States, its largest single market, than its pre-2025 trajectory implied. The two facts can only be reconciled if Swiss output that would previously have gone to the US is being sold elsewhere, or if US demand formerly met from Switzerland is now met from production located outside Switzerland while Swiss plants serve other markets. Both mechanisms point in the same direction: the composition of Swiss production by destination is being reorganized, and the volume relationship with the US is being loosened.

The paradox has two implications for the interpretation of official statistics. First, aggregate indicators of export performance, such as total exports or the manufacturing purchasing managers' index, which stood at 55.3 in September 2026, will not signal the change under way, because the reorganization is a reallocation rather than a contraction. Second, the bilateral surplus with the United States, which was the trigger of the 39 percent tariff, is shrinking for reasons that have little to do with the tariff and much to do with the location of production. Washington may read the shrinking surplus as evidence that its policy works; Bern should read it as evidence that the geography of its industry is changing.

9. Result Five: Volatility as the Real Tax

The fifth result concerns not the level of exports but their variability. Before 2025, the monthly growth rate of Swiss exports to the United States had a standard deviation of 13.8 percent. Since January 2025 it has been 29.3 percent. A Levene test on the equality of variances rejects the null hypothesis with a p-value below 0.0001. The trade relationship with the US has become more than twice as volatile.

This volatility does not stay in the trade statistics. It feeds directly into the quarterly national accounts, through pharmaceutical value added, and it has made Swiss GDP figures in 2025 and 2026 extraordinarily difficult to read. Figure 7 compares real quarterly GDP growth with the quarterly change in the US export gap, expressed as a percentage of quarterly GDP.

Figure 7. Tariff-driven export swings and quarterly GDP
Figure 7. Tariff-driven export swings and quarterly GDP. Source: SECO national accounts via SNB data portal (real GDP adjusted for sporting events); author's calculations.

The orders of magnitude are large. The front-loading of the first quarter of 2025 added an export swing equivalent to 2.4 percent of quarterly GDP; its reversal in the second quarter subtracted 2.8 percent. In the second quarter of 2026, the narrowing of the gap added 0.7 percent of quarterly GDP. These are gross export swings, not value-added contributions, and the link to GDP is much weaker than these figures suggest. Our estimate of the pre-2025 relationship between real goods exports and real GDP growth gives an elasticity of only 0.07 (standard error 0.02), reflecting the low domestic content of the most volatile pharmaceutical flows, which are partly matched by imports of intermediate goods. Even so, SECO itself attributed nearly half of the exceptional 1.5 percent GDP growth of the second quarter of 2026 to the highly volatile value added of the chemical and pharmaceutical industry, and warned that it probably overstated the underlying momentum.

The cost of volatility is real even when the average is unchanged. It complicates monetary policy, since the SNB must form a view of the output gap from GDP figures dominated by shipment timing. It complicates fiscal planning, since tax receipts in the cantons that host pharmaceutical groups depend on the timing of their profits. And it raises the cost of capital for exporters, who must hold larger buffers against swings that are no longer driven by demand but by legal deadlines in Washington. In this sense, the tariff episode imposed a tax that does not appear in any tariff schedule: an uncertainty tax, paid in the form of noise.

9.1 The refund windfall and its limits

The Supreme Court ruling of February 2026 added a further twist. Duties collected under IEEPA between April 2025 and February 2026 became refundable, and the US Court of International Trade ordered refunds in certain cases from March 2026. UBS economists estimate that Swiss companies could receive up to 2 billion dollars, with the largest amounts going to precision instruments and watchmaking.

The macroeconomic significance of this windfall should not be overstated. Refunds are paid to the importer of record, which is often, but not always, a US subsidiary of the Swiss exporter. Where the importer is an independent US distributor, the refund accrues to the distributor, not to the Swiss producer. Where the importer is a subsidiary, the refund strengthens the balance sheet of the group but does not, by itself, raise Swiss production or employment. Even in the most favourable case, 2 billion dollars is less than 0.2 percent of Swiss GDP, spread over a period of procedural uncertainty that could last into 2027. The refunds compensate for a past cost; they do not change the location calculus that drives the export gap. If anything, the experience of having paid duties that were later declared unlawful reinforces the case for supplying the US market from US-based capacity, which is immune to the legal volatility of tariff instruments.

10. Interpretation: Relocation and Hysteresis

The five results fit together into a coherent story that differs sharply from the one told in the Swiss debate.

In the conventional story, the US tariff acts as a price wedge. Swiss goods become more expensive in the US, demand falls, and exports decline in proportion to the tariff and to the price elasticity of demand. When the tariff falls, the wedge narrows and exports recover. On this view the November 2025 framework, the Supreme Court ruling and the 12.5 percent Section 301 regime should all have brought relief, and the 2026 figures should show a recovery.

The data tell a different story. The tariff wedge mattered for timing: it created a front-loading surge before each deadline and a hangover after it. But it did not determine the level. What determined the level was a slow, steady drift that began in 2025 and accelerated in 2026, at a pace of about 1.5 percent per month relative to the counterfactual. That drift is the signature of a decision that is costly to reverse: the decision to supply the US market from US capacity.

Several facts support this reading. Swiss firms committed to invest 200 billion dollars in the United States by the end of 2028 as the price of the November 2025 framework, with 67 billion dollars expected in 2026 alone according to the White House. The two largest Swiss pharmaceutical groups had already announced US investment programmes of 50 billion dollars and 23 billion dollars in 2025, the latter with the explicit aim that its key medicines for US patients would be produced in the United States. In January 2026, one of the two groups indicated its intention to raise the US share of its global investment to 50 percent. These are not tariff-avoidance tactics that can be unwound when tariffs fall. They are multi-year capital commitments, and once a plant is built, the cheapest way to supply the US market is from that plant, regardless of the tariff on Swiss goods.

This is what economists call hysteresis: a temporary shock that produces a permanent change in the structure of the economy. The tariff regime of 2025 was temporary, and most of it has been voided by the US courts. But the investment commitments it extracted were not, and they are now reshaping trade flows independently of the tariff that triggered them.

The drift we estimate is consistent with the gradual commissioning of new capacity. If anything, it is likely to be underestimated in the short sample: new plants take two to four years to build and validate, and most of the capacity announced in 2025 and 2026 is not yet in production. The decline of the US share of Swiss exports observed so far reflects mainly the reallocation of existing global capacity, for instance the decision by a multinational group to supply the US from a plant in the United States, Ireland or Singapore rather than from Basel. The larger effect, from capacity that does not yet exist, lies ahead.

It is important to be precise about what this interpretation does and does not imply. It does not imply that Swiss industry is being hollowed out. Total Swiss goods exports in the first eight months of 2026 were only about 2 percent below the same months of 2025, a period itself inflated by front-loading, and pharmaceutical production in Switzerland expanded sharply in the second quarter of 2026. What it implies is a change in the geography of Swiss production: the high-value stages (research, development, clinical supply, complex active ingredients) remain in Switzerland, while the volume stages for the US market migrate. The question for Swiss economic policy is whether that division of labour is stable, or whether it is the first step of a more general migration.

11. Scenarios for 2027

To translate these findings into forward-looking magnitudes, we project the export gap from September 2026 to December 2027 under four scenarios. Each starts from the average gap of the last six months (minus 19 log points) and a monthly counterfactual level of about CHF 4.6 billion. Uncertainty enters through the estimated drift (drawn from its sampling distribution) and through a level shock calibrated on the placebo dispersion of the counterfactual models (3.5 log points). Each scenario is simulated 20,000 times.

Scenario 1, stabilization, assumes that the 12.5 percent regime holds and that no further relocation occurs: the gap stays at its recent level. Scenario 2, relocation drift, assumes that the estimated drift continues but decays as the pledged capacity comes online, with a half-life of twelve months. Scenario 3, binding agreement, assumes that a legally binding Swiss-US agreement restores enough certainty to close 40 percent of the gap by the end of 2027, for instance by securing Swiss-based supply for some product lines. Scenario 4, escalation, assumes that Section 232 pharmaceutical duties bite fully from the end of 2026 and that the headline rate is raised to 25 percent, adding a further 10 log points to the gap.

Table 5. Scenarios for Swiss exports to the US, shortfall vs counterfactual in 2027 (CHF billion)

Scenario Gap end-2027 Median 2027 80% interval
S1 Stabilisation (12.5% holds, no further relocation) -17% 9.5 7.4 to 11.5
S2 Relocation drift (pledged US capacity comes online, half-life 12m) -29% 14.4 12.3 to 16.3
S3 Binding agreement (gap closes 40% by end-2027) -11% 7.1 5.0 to 9.2
S4 Escalation (Sec.232 pharma bite plus 25% headline) -25% 13.8 11.9 to 15.6

20,000 Monte Carlo draws. Shortfalls are gross export values relative to the counterfactual. Source: author's calculations.

The results are shown in Table 5. In the central relocation scenario, the shortfall relative to the counterfactual reaches about CHF 14.4 billion in 2027 (80 percent interval: 12.3 to 16.3 billion), and the gap at the end of 2027 approaches 29 percent. Even the most favourable scenario, a binding agreement that partially reverses the drift, leaves a shortfall of around CHF 7 billion. The escalation scenario, which corresponds to the worst case in the public debate, produces a shortfall of about CHF 13.8 billion, smaller than the central relocation scenario.

This last comparison is the key message of the scenario analysis. The scenario that dominates the Swiss debate, a renewed escalation of US tariffs, is not the worst outcome for Swiss exports to the US. The worst outcome is the quiet continuation of a trend that is already under way and that does not depend on any further decision in Washington. To put the orders of magnitude in perspective, a shortfall of CHF 14 billion is about 1.6 percent of annual GDP in gross export terms, although, as discussed above, the value-added impact is considerably smaller.

12. Implications for Swiss Economic Strategy

The findings carry five implications for Swiss economic strategy. They are presented as analytical conclusions rather than as a list of recommendations, because the main point is a change of diagnosis rather than a change of instrument.

The first implication is that the object of negotiation has been misidentified. For sixteen months Swiss diplomacy concentrated on the tariff rate, and it achieved real results: from 39 percent to 15 percent, then, through the US courts, to 10 and 12.5 percent. But our estimates show that these reductions bought very little in terms of export volumes, because the binding constraint had shifted from price to location. A negotiation strategy for 2027 that aims at a further reduction of the headline rate would address the wrong variable. The variable that matters is the share of US demand that continues to be supplied from Switzerland, and that variable is determined by investment decisions in corporate headquarters, not by tariff schedules.

The second implication is that the volatility of the Swiss economy has risen structurally, and that its statistical apparatus has not adapted. Quarterly GDP figures are now dominated by the timing of pharmaceutical shipments, which in turn responds to legal deadlines in Washington. The SNB and SECO already adjust GDP for sporting events. A similar adjustment, or at least the systematic publication of a measure of GDP excluding the chemical and pharmaceutical industry, would make the underlying cycle readable again. The second article in this series will propose such a measure.

The third implication concerns the structure of the Swiss export model itself. The relocation of volume production to the US is not necessarily damaging to Swiss prosperity if the high-value stages, and the corporate functions that capture the profits, remain in Switzerland. The real risk is that relocation does not stop at volume production. The early evidence from 2026, including the Belgian anomaly, suggests that the multinational groups are redesigning their global supply chains as a whole, not just moving final assembly. Switzerland's long-term interest lies less in the tariff it pays at the US border than in the attractiveness of its location for the stages that will not move: research, high-complexity manufacturing and headquarters functions. This shifts the policy focus from trade diplomacy to the domestic conditions that determine location choices, including the regulatory environment, access to talent under the bilateral agreements with the European Union, energy security and the tax treatment of research.

A fourth implication reaches beyond Switzerland. The Swiss experience is a warning for every small, highly specialized open economy that negotiates with a large partner under tariff pressure. The standard negotiating currency in 2025 and 2026 was the investment pledge: Switzerland, the European Union, Japan and South Korea all secured lower tariffs by committing their firms to invest in the United States. Our results suggest that such pledges are far more costly than the tariff they replace, because they act on the location of production, which is persistent, rather than on the price of exports, which is reversible. A tariff can be lowered by a court ruling or by the next administration; a plant, once built, remains. Small economies that accept investment pledges as the price of market access should therefore evaluate them not as diplomatic concessions but as long-term transfers of productive capacity, and should insist on the time-profile and the sectoral composition of the commitments, which determine how quickly the trade relationship will be hollowed out.

A fifth and final implication concerns monetary policy. The SNB's assessment of the franc and of the output gap has so far treated the tariff episode as a temporary demand shock to exporters, to be cushioned by an accommodative stance. If the export gap reflects relocation rather than lost price competitiveness, the appropriate response is different. A weaker franc can offset a price wedge; it cannot reverse a plant location decision. The intervention threshold that the SNB defends against franc appreciation should therefore be calibrated on the parts of the export base that remain price-sensitive (watches, machinery, precision instruments), not on aggregate exports, which are increasingly dominated by intra-group flows whose volume is set by location rather than by exchange rates.

13. Conclusion

The tariff saga of 2025 and 2026 has been told in Switzerland as a story about rates: the shock of 39 percent, the relief of 15 percent, the legal victory of February and the new uncertainty of July. This article has shown that the data tell a different story. Once the counterfactual is constructed carefully and the pharmaceutical shipment cycle is accounted for, the level of the tariff explains almost nothing of the evolution of Swiss exports to the United States. What explains it is a steady, tariff-insensitive drift, consistent with the relocation of production capacity triggered by the investment commitments that Switzerland made to obtain lower tariffs.

The irony of the episode is therefore considerable. The commitments that secured the lower tariff are doing more damage to bilateral export volumes than the higher tariff ever did. The cumulative balance with the US turned negative in March 2026, the gap is widening, and the scenario that dominates the public debate, renewed escalation, is less damaging for exports than the trend already in motion.

None of this means that the Swiss export model is broken. Total exports have held up, pharmaceutical production in Switzerland remains strong, and the high-value stages of production have not moved. But it does mean that the question Switzerland faces is no longer how much it pays at the US border. It is which parts of its production will still be Swiss in five years. That question cannot be answered in Washington; it will be answered by the location decisions of a few dozen multinational groups, and by the conditions Switzerland offers them at home.

About this series. This paper is the first in a ten-part series on the Swiss economy in 2026 published by Global Strategy Files. Forthcoming papers examine underlying growth once pharmaceutical volatility is removed, US drug pricing and Swiss health costs, gold and the trade balance, the franc and the energy shock, the fiscal windfall from commodity trading, the capital rules for UBS, the debt brake and central bank distributions, housing after the end of the imputed rental value, and winter electricity security.

Methodological Appendix

A.1 Data

Monthly Swiss exports by goods category (table ausshawarm) and by destination country (table ausshalam), in CHF millions, business-cycle total excluding non-monetary gold, precious metals, precious stones and works of art (source: FOCBS, published on the SNB data portal, data as published on 21 October 2026). Monthly average exchange rates CHF per EUR and CHF per USD (table devkum). Quarterly real GDP adjusted for sporting events and real goods exports excluding non-monetary gold (table gdpap, source SECO). Sample: January 2014 to August 2026 for monthly data, first quarter 2014 to second quarter 2026 for quarterly data.

A.2 Tariff variable

The tariff variable is the day-weighted monthly average of the additional US duty applicable to non-exempt Swiss goods: 0 percent until 4 April 2025; 10 percent from 5 April 2025; 39 percent from 7 August 2025; 15 percent from 14 November 2025; 10 percent from 24 February 2026; 12.5 percent (total duty including MFN) from 24 July 2026. Sector-specific Section 232 duties and company-specific exemptions are not included in the headline variable and are captured indirectly through the pharmaceutical pulse variable.

A.3 Counterfactual models

Model A: log(X_US / X_nonUS) = c + b1 t + b2 t^2 + b3 log(USD/EUR cross rate) + u, with u following a seasonal AR(1)(1)12 process, estimated by maximum likelihood on January 2014 to December 2024. Model B: log(X_US / X_nonUS) frozen at its 2024 average. Model C: log X_US approximated by a non-negative combination of the demeaned log exports to thirteen donor countries plus a linear drift, with weights estimated by non-negative least squares on January 2016 to December 2024. The ensemble is the equal-weighted mean of the three log gaps; the reported range is the minimum and maximum across models.

A.4 Placebo tests

For Models A and B, nine rolling origins from December 2018 to December 2022 at six-month intervals; at each origin the model is re-estimated and projected for twenty months (truncated at December 2024). Reported statistics are the mean error and the mean absolute error of the average log gap over each window.

A.5 Elasticity regression

gap_t = a + b tau_t + c P_t + d A_t + e trend_t + v_t, for January 2025 to August 2026 (20 observations), where gap_t is the log gap, tau_t the tariff in percentage points, P_t the log deviation of total pharmaceutical exports from their centered thirteen-month rolling median, A_t a dummy for March 2025 and trend_t a linear trend. Newey-West standard errors with three lags. Given the short sample, coefficients should be read as indicating orders of magnitude; the key result (b close to zero) is robust across all three counterfactuals and with or without the trend term.

A.6 Markov-switching model

Two-regime Markov-switching model with regime-dependent mean and variance for the monthly log change in total pharmaceutical exports (February 2014 to August 2026), estimated by maximum likelihood; smoothed regime probabilities are reported. Pulse sizes are measured as the deviation from the centered thirteen-month median, standardized by the pre-2025 standard deviation of that deviation.

A.7 Scenarios

Monte Carlo with 20,000 draws, horizon September 2026 to December 2027. Starting gap: mean of March to August 2026 ensemble gap. Counterfactual monthly level: 2026 average counterfactual increased by 2 percent. Drift draws from N(-0.0149, 0.0022). Level noise N(0, 0.035). The relocation scenario applies a cumulative drift with a twelve-month half-life; the agreement scenario closes 40 percent of the starting gap linearly by December 2027; the escalation scenario adds minus 0.10 log points from December 2026. Reported losses are for calendar year 2027.

A.8 Formal statement of the relocation interpretation

The relocation interpretation can be written compactly. Let X_t be Swiss exports to the US and X*_t the counterfactual. Decompose the log gap as g_t = log X_t minus log X*_t = a(tau_t) + p_t + h_t + e_t, where a(tau_t) is the static price effect of the tariff, p_t the pharmaceutical shipment cycle including front-loading, h_t the cumulative effect of relocated capacity and e_t noise. In a price-driven world, h_t is zero and a(tau_t) is strongly negative in tau. In a relocation world, a(tau_t) is small and h_t evolves as h_t = h_(t-1) + d_t, with d_t negative and driven by the commissioning of new US capacity, itself a function of past investment commitments rather than of the current tariff. The regression in Table 4 is a reduced-form test of these two hypotheses: a near-zero coefficient on tau combined with a significant negative trend is the signature of the relocation world. A richer specification would replace the linear trend with a measure of commissioned capacity, which could be constructed from company disclosures as plants come online in 2027 and 2028; we leave this extension for future work.

A.9 Limitations

The analysis uses customs data by destination for total goods and by goods for all destinations; a full cross-tabulation of goods by destination would sharpen the decomposition between pharmaceuticals and other goods. The post-shock sample is short (twenty months), so all estimates should be read as indicative. The counterfactuals assume that the drivers of the US share before 2025 would have continued; a structural slowdown of US demand for Swiss goods unrelated to tariffs would bias the gap upward, but the placebo tests suggest that such errors are of the order of 3 percent, well below the estimated 2026 gap.

References

Federal Office for Customs and Border Security (FOCBS), foreign trade statistics, via SNB data portal (data.snb.ch), tables ausshawarm, ausshalam, devkum and gdpap, accessed November 2026.

State Secretariat for Economic Affairs (SECO), "Handelsbeziehung Schweiz USA" and detailed information on US additional tariffs, July and August 2026; GDP press releases of 14 August and 3 September 2026.

economiesuisse, "12.5 percent US tariffs: uncertainty persists for Swiss companies", July 2026; "Supreme Court halts IEEPA tariffs: uncertainty for Swiss companies remains", 21 February 2026; "US tariffs and the Swiss economy: questions and answers", 2026.

Swiss Medtech, "US Tariffs" information page, 2026.

swissinfo.ch (Keystone-SDA and Bloomberg reports): "What lies ahead for Switzerland: an economic outlook for 2026"; "Swiss economy grows surprisingly fast in Q2"; "Swiss firms may see 2 billion dollars in US tariff refunds, UBS says"; "Drugmakers reassess Europe as Trump overhauls drug-pricing rules"; "Timeline: How US President Trump upended the pharma industry in 2025"; "The art of the deal: what are US pharma announcements really worth?"; "Switzerland's gold export surge highlights its US trade dilemma".

Reuters and Al Jazeera reporting on the US-Switzerland framework agreement of 14 November 2025.

Fierce Pharma, "Trump relents on Swiss tariffs, knocking rate on pharmaceuticals and other goods down to 15%", November 2025.

blue News (Keystone-SDA): "Federal Government significantly raises growth forecast for Switzerland"; "Swiss watch exports rise again in September", October 2026.

Driss EFFINA, PhD
Independent Researcher, Global Strategy Files

Free preview
The Q4 2026 Global Outlook

The opening pages of our four country reports: United States, United Kingdom, Japan and China. A free PDF, sent to your inbox.

You will also receive occasional analysis from Global Strategy Files. Unsubscribe in one click, at any time.