Wildfires are enormously costly events. Human causes make up 85% of wildfire ignitions in the US. This paper examines household location choice as a driver of wildfire risk. I estimate an inverted U-shaped relationship between population density and wildfires: locations with medium population density have the highest ignition rates. I incorporate this relationship into a quantitative spatial model of Los Angeles County. In the model, insurance prices respond to increases in population-driven wildfire probability and mitigate future wildfire risk. A counterfactual scenario which restricts new construction in high-risk locations induces a 6% improvement in welfare.

An indispensable feature of quantitative spatial models (QSM) are structural residuals that fit the distribution of residents across locations to the data. While they are often interpreted as amenities, do they actually represent observed amenities? We collect data on 41 amenities in Los Angeles County and build a QSM to study the relationship between the structural residuals and observed amenities. We find that 45% of the variation in the residuals is explained by observed amenities. This suggests that one should be judicious when interpreting these residuals as amenities. We also find that 14 percentage points of the explained variation is accounted for by natural amenities and 31 by man-made amenities, supporting modeling amenities as endogenous.

A growing literature seeks to evaluate the resilience of transportation infrastructure in developing country cities and to target investments in infrastructure improvements. Much of this literature focuses on improved design, materials, and costs of resilient infrastructure, but relatively little is known about the economic benefits associated with reducing the impacts of weather events in a transportation system. This study presents some of the first estimates of the economic benefits associated with travel delays due to flood events in a developing country city. Our empirical strategy makes use of repeat observations of trip durations from Google's traffic API to separately identify the general effects of precipitation events and the specific effects of delays that result specifically from infrastructure flooding.