This paper examines local economic vulnerability over time by comparing how county-level unemployment rates were affected by two major flooding events in the Midwest region, one in 1993 and the other in 2008. One of the challenges for studies of this kind is to separate out the impact of the hazard events from the effects of more general macroeconomic dynamics. A case–control design is used in the analysis for that purpose, applying an optimal matching algorithm to select a non-flooded control county for each flooded case county and then model the flood effects using ARIMA intervention models. The main research finding of this study is that the 1993 flood affected a larger area and caused more severe economic disturbances than the 2008 flood. Counties that were hit hard in 1993 seemed to fare well in 2008. It was also found that higher damage was associated with an increase in the county-level unemployment rate.