How costly can a conflict be? While one can measure the expenses poured into a conflict and the physical destruction that ensued, there are important indirect costs that are much harder to get to. Think about the fear, the uncertainty, and the costs of migration. Taking the example of Northern Ireland, Timothy Besley and Hannes Mueller measure violency by conflict related death in eleven regions and look at their impact on house prices, which are supposed to capitalize all costs, and in particular incorporate both immediate suffering and risk as well as the likelihood of future violence. Indeed, a casualty is often a signal about future violence to come, and thus house prices capture also the learning process about conflict violence.
Besley and Mueller model this with a Hamilton regime switching process. Whenever an event occurs, it changes the likelihood that a region is in a violent regime. And this likelihood impacts hose prices. They find that the peace dividend corresponds to about 4-12% in Belfast and 0.8-2.5% over all regions. They also find a migration effect: when violence increases in Belfast, surrounding areas gain, and this effect is even stronger than the learning effect about the region's own violence process. Thus, it looks like every region reacts differently.
Are these effects big? I would say so given that the violence has directly impacted a very small fraction of the population. In fact, the cost of conflict can be even much larger. Indeed, the study only looked at differences across regions in Northern Ireland. There can be substantial costs for the region as a whole, costs that are washed out by trends and fixed effects. For example, the recent housing price boom in the area may have reflected the new peace (along with reduced conflict related expenses by authorities).