The fiscal cost of conflict: Evidence from Afghanistan 2005–2017

I use a novel monthly panel of provincially-collected central government revenues and conflict fatalities to estimate lost due in Afghanistan since 2005. Headline estimates are large, implying total revenue losses $3bn 2005 that gains from peace would have been about 6 percent GDP per year. That this is larger than cross-country studies reflects the uncommon intensity Afghanistan. The key challenge identification omitted variable bias, which address by extending Powell’s (2021) generalized synthetic control method dynamic setting. This allows estimation impulse response functions robust broad class variables bias.