Single-authored

Do weapons imports increase or contain violence?
Surprisingly causal estimates remain missing, leaving this question unanswered. Except for a few studies constrained to the US. However, most of the suffering from armed violence is experienced by developing countries. Countries who most of the time lack the technology and production capacity to produce weapons, forcing them to imports weapons from developed countries. 
Using an instrumental variable and worldwide data from 1992 to 2011, I show that weapons imports increase the probability of internal conflict, the number of internal conflicts within a country, the number of battle-related deaths and the number of refugees.
A simple back of the envelope computation reveals that if Europe would stop for a year to send weapons to Africa it would reduce the number of refugees by half a million.

With Dominic Rohner

Globalization is routinely blamed for various ills, including stirring tensions and conflict in strategic locations for world trade. We investigate in the current article whether these accusations are well founded, building a novel dataset of any given location's strategic importance around the whole globe. In line with our game-theoretic model of strategic interaction, we find that overall fighting is more frequent in strategic locations close to maritime choke points (e.g. straits or capes), but that booming world trade openness reduces considerably the risks of conflict erupting in locations crucial for maritime transport. The impact is quantitatively very sizable, as moving one standard deviation (1,100 km, corresponding to the distance from New York to Chicago) closer to a choke point increases by 0.3 percentage points (20\% of the baseline risk) the conflict likelihood in periods of low globalization, while reducing it by 0.2 percentage points in periods of world trade booms. Our results have important policy implications for the role of supranational coordination.

with Jean-Philippe Bonardi, Dimitrija Kalanoski and Rafal Lalive

Countries across the world introduced various types of lockdown measures restricting their populations’ movements in order to reduce the morbidity and mortality rates related to the COVID-19 pandemic, caused by the novel coronavirus SARS-CoV-2 that first emerged in China in 2019. We analyze whether these measures helped reduce the spread of infections and the number of deaths. We also compare their effectiveness in developed countries against developing countries. Our data covers 184 countries in the period from 31st December 2019 to 4th May 2020 and identifies when lockdowns were adopted, along with confirmed cases of infections and deaths due to COVID-19. The panel data structure, by addressing inherent endogeneity issues, enables us to make some causal claims. We find that partial lockdowns were as effective in reducing the number of infections and deaths as stricter measures. We estimate that in developed countries, they reduced about 650,000 deaths, but we do not find such significant effects in developing countries. This suggests that lockdowns work only if the opportunity costs of staying at home are not too high. We also find that countries that acted fast fared better, but that closing borders has had no appreciable effect, even after fifty days.

With Massimo Morelli and Dominic Rohner

Natural resources are often argued as one of the major drivers of conflict. Primary energy consumption, today, are oil (34%), coal (27%) and natural gas (24%). While coal and oil relationship on conflict have been well studied, natural gas remains to be addressed. The cheapest and less technology intense way of sending natural gas is by pipeline. Thus, we study how central positions in the pipeline network drives military interstate disputes. The Caucasus, Central Asia, the Middle-East and Eastern-Europe are often experiencing tensions caused by this natural resource. Those regions are typically the one linking large suppliers to consumers. Interestingly they are not necessarily producer of gas but sometimes only central player on the pipeline network.