This paper demonstrates that economic disruptions from civil conflicts can persistently relocate bilateral trade flows and endanger countries' chances of economic recovery. We derive a novel estimation approach based on the structural gravity model of international trade, and find that importers substitute away from exporters that are engaged in civil conflict. This effect persists after the resolution of a conflict, as countries cement their relocation decisions by decreasing bilateral trade costs with their substitute trade partners via Preferential Trade Agreements. The persistent trade relocation is most pronounced in manufacturing trade, but does not occur in the fuels sector. Our results suggest that supportive trade policies from high income countries are an important policy tool to help countries recover from episodes of political violence. Finally, our estimation approach can be adapted to investigate alternative unilateral shocks, e.g. natural disasters, or to analyze other bilateral dependent variables like migration.

We use a unique case study to estimate the effect of withdrawing from a free trade agreement on international trade. Lately, the political opposition to international economic cooperation has been on the rise, but little is known about how the withdrawal from a trade agreement affects trade. We analyze a quasi-natural experiment to provide first empirical evidence. In 2004, Estonia joined the European Union, which mandated that it withdraws from its FTA with Ukraine. Based on the gravity model of international trade, we provide evidence from triple Difference-in-Differences as well as PPML panel estimations that trade volumes between Estonia and Ukraine fell by more than 20%. We find that withdrawing an FTA revokes all benefits and that no institutional memory is left behind. General equilibrium estimates suggest that FTA withdrawal led to a noticeable loss in members' welfare. 

Revise & Resubmit at Scandinavian Journal of Economics.

This paper proposes a new way to account for subnational conflict exposure and provides new evidence on the longevity of conflict's detrimental effects. The literature disagrees whether civil conflicts can alter long-run economic growth paths. I trace this disagreement to different approaches in measuring conflict exposure. The common practice of using subnational aggregates of conflict intensity ignores essential intra-regional variation. As a more accurate measure, I propose the share of economic activity in proximity to conflict events. Estimating a Bartik-like IV model at the district level of 70 countries, I provide causal evidence that conflict exposure significantly decreases economic activity in the medium run. As potential channels, I identify persistent diversions of investments and human capital.

We investigate the role of World War I casualties on German economic and political development in the interwar period. We geocoded the birthplaces of 8.5 million wounded and killed German soldiers, and linked these casualties to newly digitized county-level census data from before and after the war. Our main results are based on continuous difference-in-differences estimations and illustrate that counties that suffered a higher human loss during the war specialized more in industrial production and moved employment out of self-sufficient farming. Counter-intuitively, we find that wages in the low-skilled sector decreased in more affected counties, which we interpret as a negative skill selection effect. Finally, we find that more affected counties were more likely to support radical right-wing parties in interwar elections.

PEDD Young Scholar Award (2023) & RGS Econ Best Paper Award (2023)

This study examines the effects of the violent repression of independence movements on ethnic politics and social cohesion. We exploit local variation in the intensity of repression to analyze the long-run impacts of British detention camps in 1950s colonial Kenya. Using a rich body of census and survey data and a triple-difference design, we show that exposure to a detention camp increases ethnic voting in the contested 2007 presidential election and erodes contemporary trust. In addition, we show that affected individuals accumulate less wealth, are less literate, and have poorer labor market outcomes three to five decades after the event.

This paper constructs a direct measure of military state capacity and studies its effect on the dynamics of civil wars. Using yearly data for more than 120 countries, we estimate the effect of the composition of the stock of military capital equipment (e.g. number of attack helicopters vis-à-vis tanks) on how a civil war escalates. Among broad categories of  military equipment, we find that only attack helicopters are associated with civil war de-escalation. We further probe the prominence of helicopters and find that it is indeed the only category associated with battle locations shifting from rural to urban areas and with significant increases of unintended civilian casualties. We also show that only partial democracies significantly favor using helicopters for a speedier termination of a civil war, even when this involves indiscriminately sacrificing civilian lives.