Data show that trade sanctions reduce commerce between the countries imposing them and their targets. But this bilateral success can conceal a wider failure if the sanctions induce the target to build stronger trade links elsewhere. This column uses sector-level trade data, including domestic sales, to estimate both the bilateral trade destruction and the third-country trade creation effects together. Across 22 sectors, complete sanctions usually promote trade between targets and third countries. Russia’s 2006 embargo on Georgia illustrates this fact: additional trade to third markets more than offset the estimated 57% decline in exports to Russia.

How can governments design trade sanctions that reduce a target’s access to markets rather than merely redirecting its commerce? A sharp decline in trade between sanctioning countries and the target is often interpreted as evidence that sanctions are effective. Standard trade theory predicts that such disruptions induce trade diversion towards third countries and that this reallocation generally does not offset the adverse welfare consequences. However, if sanctions prompt the target to reduce trade costs with third countries, the resulting expansion of alternative trade relationships may increase the target’s overall trade and even improve its welfare.

This is not merely a theoretical possibility. Early (2015) describes how third countries engage in “sanctions busting”, while Scheckenhofer et al. (2025) document evasion in military goods.

The empirical evidence on the overall third-country response is mixed. Caruso (2003) finds that unilateral US sanctions reduced trade between targets and other G7 countries. More recent work by Gammadigbe (2025) and Yalcin et al. (2025) finds positive third-country effects, including stronger Russian trade links with China, India, and Türkiye after the sanctions imposed in 2022 (summarised on VoxEU in Felbermayr et al. 2025).

In a new paper, we examine this broader response systematically across sectors (Felbermayr et al. 2026). We distinguish the direct effect of a complete trade sanction, which is the change in trade between senders and targets, from its third-country effect, which is the change in trade between targets and all other countries. The distinction matters because a sanction can succeed at closing one bilateral channel but fail to curb the target’s total trade.

Measuring trade redirection

We work with a state-of-the-art gravity model of bilateral trade flows to estimate the direct and the third-country effects of sanctions while controlling for the standard general equilibrium trade diversion effects that would arise with third-country trade costs held constant at the pre-sanction levels. We stress that both the bilateral and the third-country estimates are direct effects of the sanctions, not general-equilibrium outcomes: the general-equilibrium adjustments that any sanction sets off are absorbed by our fixed effects and therefore do not drive the results. Our analysis combines information on sanctions with the International Trade and Production Database for Estimation. The data cover more than 200 countries, 170 industries, and the period from 1986 to 2022. For presentation purposes and statistical precision, we organise the industries into 22 broad sectors, ranging from food and textiles to machinery, energy, and services.

An important feature of our work is the inclusion of domestic trade, which records the goods and services sold within the producing country. Domestic sales are what make it possible to estimate the two effects at the same time, because without them the direct and third-country sanction indicators are perfectly collinear with the country, industry, and time fixed effects. Our framework also accounts for persistent trading relationships, globalisation trends, regional trade agreements, WTO membership, and other types of sanctions. The underlying paper reports the full econometric specification, the data construction, and the sector-by-sector estimates.

We focus on complete trade sanctions because they are intended to shut down trade broadly and have stronger direct effects than partial restrictions. Partial trade, financial, arms, military, travel, and other sanctions remain in the analysis as controls.

Direct losses, but gains elsewhere

The conventional bilateral result is strong. When the third-country response is omitted, the estimated direct effect is negative in all 22 sectors and statistically distinguishable from zero in 20. Complete sanctions clearly disrupt trade between senders and targets.

But what happens to the target’s trade with everyone else? Figure 1 reports both effects for each of the 22 sectors, estimated jointly. The third-country effect is positive in 17 of them and is sizeable and statistically significant in 13.

The largest estimated increases occur in chemicals and metal products, where trade between targets and third countries more than doubles. Estimated gains are also large in mining and energy, apparel and footwear, textiles, and specialised machinery.

These increases are not the mechanical result of trade diversion. Our estimation absorbs the structural multilateral resistance terms of Anderson and van Wincoop (2003) through fixed effects that vary by exporter, industry and year and by importer, industry and year. Those terms already capture the general equilibrium reallocation that occurs when one market closes and others become relatively more attractive, so that ordinary trade diversion is controlled for rather than measured. What the third-country estimates isolate is therefore an active liberalisation of trade between the target and third countries, as if the target and those countries had newly signed a free trade agreement.

Figure 1 also shows why bilateral statistics can mislead. In chemicals, electronics, textiles, and transportation, the estimated direct effect is not statistically different from zero, while the third-country effect is positive and ranges from 32 to 129%. In metal products, the estimated contraction in trade between senders and targets is about 24%, while trade between targets and third countries rises by roughly 119%. Whether this translates into an increase in the target’s overall trade depends on its initial exposure to the sanctioning coalition relative to the rest of the world. Under plausible initial trade patterns, however, the net effect can readily be positive. 

Figure 1 Complete trade sanctions redirect trade towards third countries

Notes. Figure plots the estimated percentage change in trade for each of the 22 sectors. Red markers show the direct effect on trade between senders and targets. Blue markers show the third-country effect on trade between targets and third countries. Horizontal lines show 95% confidence intervals. Both effects are estimated jointly, and the estimation includes domestic trade flows and absorbs multilateral resistance terms. 
Source: Felbermayr et al. (2026), Table 1, panel B.

Georgia demonstrates that sanctions may have perverse effects

Russia’s 2006 embargo on Georgia provides a concrete illustration. Before the embargo, Russia bought about 14% of Georgia’s exports. We apply the sector-specific direct and third-country estimates to Georgia’s 2005 export pattern and compare predicted post-sanction exports with that baseline.

Figure 2 summarises the result. The bilateral channel behaves as intended, with Georgia’s exports to Russia falling by an estimated 56.9%, yet exports to the rest of the world rise by 79.5%. Because the rest of the world was already a much larger market than Russia before the embargo, the export gains to third countries accrue over a much larger base. Consequently, Georgia’s total exports are estimated to increase by 60.9%, with net gains in 16 of the 22 sectors.

We term this outcome a perverse trade effect, whereby a sanction reduces bilateral trade with the sanctioning coalition yet increases the target’s aggregate trade. This is not a claim that Georgia’s welfare necessarily improved, since finding alternative markets may require costly negotiations, export promotion, changes in logistics, or price concessions, and our partial equilibrium calculation cannot measure all of those costs. It nevertheless illustrates why policymakers should not equate a collapse in bilateral trade with economic isolation. 

Figure 2 Russia’s 2006 embargo redirected rather than reduced Georgia’s exports

Notes. Figure plots a partial equilibrium prediction obtained by applying the 22 sector estimates to Georgia’s 2005 export pattern. Russia accounted for about 14% of Georgia’s baseline exports in that year. The calculation describes trade flows and is not a welfare estimate. 
Source: Felbermayr et al. (2026).

Designing sanctions for a world of alternative partners

The findings suggest three policy priorities. First, success should be evaluated against the policy objective rather than bilateral trade, with monitoring focused on the target’s total trade in the sanctioned products. Second, substitution possibilities should be assessed ex ante, as sanctions are most effective where alternative suppliers, routes, and payment channels are limited. Third, the sanctioning coalition should be as broad as possible to reduce trade diversion to neutral countries. More generally, analyses should account explicitly for third-country effects, as ignoring them biases estimates of the sanctions’ direct impact.

The broader lesson is not that trade sanctions are ineffective: they can substantially reduce targeted bilateral trade. Rather, economic pressure operates through a global trade network, so sanctions are best evaluated by their effect on the target’s global economic opportunities rather than on bilateral trade alone.  

Source : VOXeu