This paper investigates the incidence of economic sanctions according to the type of sanction design. I document that sanctions have increasingly targeted specific firms rather than entire sectors in recent years, a shift explained neither by differential success rates nor by political objectives. This points to an economic rationale. I thus develop a two-country model in which heterogeneous firms source inputs in both locations and compete in oligopoly. Sanctions operate along two margins: exclusion from the sanctioning market (import ban) and constrained sourcing decisions (export control). The pricing decisions of firms internalises their sourcing decisions. A sanction will increase the prices in the foreign market and can modify the sourcing decision of all firms worldwide. Analytically, the damages borne by the targeted firm relative to foreign consumers is maximal when the largest firm of a sector is targeted. The simulation of the model shows that firm-level sanctions split the burden between the sanctioning and target economies, whereas sectoral sanctions place most of the burden on the target economy. The different incidences originate mainly from the import ban component of the sanction, as the export control has moderate impact on either economy's welfare. The results point toward a shift in sanction design moving away from destructive sanctions paid by the consumer, towards more parsimonious sanctions actually harming the firms they aim at coercing.

This paper presents the Firms Sanctioned DataBase (FSDB), a novel dataset documenting all firms and entities sanctioned by the European Union since 2001. The FSDB details the timeline of sanctions and the exact addresses of every targets inside of sanction programs (e.g., Iran, Russia). Using this granular dataset, I characterize anticipations and location of targets, and study how they affect the identification of the impact of sanctions on disaggregated bilateral trade flows. I find that extraterritorial sanctions decreased flows of firms' sanctioned products by 38%. The estimate is similar when solely accounting for the staggered enforcement, with an higher variance though. Ignoring the staggered enforcement of sanctions considerably biases the estimate.