Trade policy may be negotiated bilaterally, but its consequences spill over to the global economy. Tariffs, sanctions, export controls, and other trade restrictions can redirect demand and competition across the world economy (Freund et al. 2024, Alfaro and Chor 2025). When two large economies fight a trade war, firms in third countries may gain new market opportunities, face tougher competition, or both. Which firms gain, which firms lose, and why?

The answer is not obvious. Consider how the 2018-2019 trade war between the US and China may have affected firms in third countries. On the one hand, higher US tariffs on Chinese products may have created opportunities for third-country firms selling close substitutes in the US market. On the other, Chinese firms may have redirected exports to other destinations, intensifying competition there. Some third-country products substitute for US or Chinese goods, while others complement them. And if firms operate under non-constant returns to scale, these shocks can be amplified or dampened by changes in their production costs (Bartelme et al. 2025).

In a new paper (Conteduca et al. 2026), we disentangle these channels through the lens of the 2018–2019 US–China trade war, focusing on firms in a third country, namely, Italy. The US–China trade war provides an ideal setting because it sharply changed bilateral tariffs – and hence prices – between two large economies, with Italian firms not having been directly involved in the dispute. We use this setting to quantify how a bilateral trade shock propagates to bystander firms through demand diversion, foreign competition, and scale economies channels. Importantly, our framework is flexible enough to be applied to other countries or trade shocks.

A tractable model to identify winners and losers

We propose a tractable trade model with heterogeneous firms. Our framework accounts for the fact that the same bilateral trade shock can affect firms differently depending on where they sell and what they sell. We therefore allow trade elasticities to vary across destinations, rather than imposing a single elasticity across markets. We also allow products from different origins be either substitutes or complements, leaving cross-price elasticities unrestricted. Finally, firms may operate under non-constant external economies of scale, so that changes in aggregate export volumes can either lower or raise production costs.

The framework nests standard Armington-, Ricardian-, and Melitz-type models as special cases, while allowing for a richer quantification of how trade shocks affect individual firms across products and destinations. We estimate the model using administrative data on Italian firms’ export activity and product-level tariff changes during the US–China trade war.

The 2018–2019 US–China trade war generated net export gains, but the average hides many losers

Overall, we find short-run net export gains for Italian firms, with average export revenues rising by 2.5% over the period. Average gains are higher in the US, consistent with Italian exports substituting for Chinese products. Exports to the EU, China, and – to a lesser extent – the rest of the world also expanded, consistent with the presence of external economies of scale in Italian exports. These average effects, however, mask substantial heterogeneity: the standard deviation of pooled export revenue changes is 7.6%, with roughly one in five firms experiencing a decline in exports.

Figure 1 Distribution of firm-level export revenue changes

Note: The figure shows the distribution of model–predicted export revenue changes across Italian firms between 2017 and 2019, both pooled across destinations and disaggregated by destination.

Linking model-predicted export changes to firms’ pre-trade war characteristics, we find that export gains were concentrated among firms that were initially more productive, more skilled labour-intensive, more investment-intensive, and exporting a broader product range. After the shock, export revenue growth is associated with higher employment for both blue-collar and white-collar workers, higher wages, and more investment, with no evidence of crowding-out of domestic sales. Overall, these results suggest that tariff–induced reallocations worked, in the Italian case, in the direction of better allocative efficiency.

External scale economies were the main driver of export revenue changes

External scale economies account for roughly three-quarters of the total variation across deciles of firm-level export changes, making them the main driver of export revenue changes for both winners and losers. The remaining one-quarter is jointly explained by destination-specific, own-demand, and cross-demand effects, all of which pushed revenues up across deciles. Overall, our results suggest that Italian firms did not benefit only because US buyers substituted away from Chinese goods. The broader reallocation of trade flows triggered by the US–China trade war also increased Italian export volumes to other destinations, and these volume changes fed back into costs through external economies of scale. As a result, exports rose on average not only to the US, but also to China, other EU countries, and the rest of the world.

Figure 2 Decomposition of firm-level export revenue changes

Note: The figure decomposes firm-level export revenue changes into the four channels accounted for by our model: destination, own-demand, external economies of scale, and cross-demand effects. It shows the average contribution of each channel across deciles of total model-predicted export revenue changes.

Standard models get the average right but miss the distribution

A natural question is whether our richer structure matters quantitatively. To answer this, we benchmark our model against standard trade models abstracting from scale economies and cross-price demand effects (Arkolakis et al. 2012, Head and Mayer 2014). The restricted models perform well on average: they predict a similar average export gain to the full model. However, they miss the dispersion across firms. In our model, the standard deviation of export revenue changes is 7.6%. We find it falls to 1.6% when scale economies are removed, and to 1.1% when cross-price effects are also shut down. For comparison, the dispersion observed in the raw data is 17.9%.

Figure 3 Distribution of firm-level export revenue changes across models

Note: The figure shows the distribution of model–predicted export revenue changes across Italian firms between 2017 and 2019 pooled across destinations. The black line represents the distribution in the benchmark model; the blue line represents the distribution shutting down economies of scale; the red line represents the distribution shutting down both economies of scale and cross-demand effects.

Global fragmentation has sizeable distributional effects even for bystander countries, effects absent from most alternative models

Two broad lessons emerge from our work. First, trade wars may create short-term opportunities for bystander countries, but they also widen the gap between winners and losers. Aggregate gains can coexist with firm-level losses. Second, firm-level responses to trade tensions can be driven by channels – such as cross-price demand effects and external economies of scale – that conventional models do not capture. Policymakers concerned with the distributional consequences of trade tensions therefore need a framework that accounts for these effects. As Baldwin (2026) emphasises, the trading system that emerges from recent trade shocks is unlikely to resemble the one that preceded them. Understanding the firm-level responses is a precondition for designing the policy responses it will require.

Authors’ note: The views expressed here are those of the authors and do not necessarily represent the views of the Bank of Italy or the International Monetary Fund.

References

Alfaro, L and D Chor (2025), “An Anatomy of the Great Reallocation in US Supply Chain Trade”, Technical Report, National Bureau of Economic Research. 

Arkolakis, C, A Costinot, and A Rodrıguez-Clare (2012), “New Trade Models, Same Old Gains?” American Economic Review 102(1): 94–130. 

Baldwin, R (2026), World War Trade: Conflict, Containment, and the Emergent World Trading Order, CEPR Press.

Bartelme, D, A Costinot, D Donaldson, and A Rodriguez-Clare (2025), “The Textbook Case for Industrial Policy: Theory Meets Data”, Journal of Political Economy 133(5): 1527–1573. 

Conteduca, F P, M Errico, F Leone, L Panon and G Romanini (2026), “The Impact of Trade Wars on Firms in Third Countries”, CEPR Discussion Paper 21421.

Freund, C, A Mattoo, A Mulabdic, and M Ruta (2024), “Is US Trade Policy Reshaping Global Supply Chains?”, Journal of International Economics 152, 104011

Gopinath, G, P-O Gourinchas, A F Presbitero, and P Topalova (2025), “Changing Global Linkages: A New Cold War?” Journal of International Economics 153, 104042. 

Head, K and T Mayer (2014), “Gravity Equations: Workhorse, Toolkit, and Cookbook”, in G Gopinath, E Helpman and K Rogoff (eds), Handbook of International Economics, Vol. 4, Elsevier.