We combine real estate data with various types of crime data using time and geospatial information to detect discontinuities in transaction densities and pricing around crime events in Rochester, NY. Discontinuities in transaction densities invalidate causal inference for price responses implied by the regression discontinuity design (RDD) approach. However, these discontinuities also capture the liquidity response to crimes and, together with the commonly emphasized price response, provide a richer picture of how crime affects housing valuation. A calibrated match-and-bargain model reveals that house valuations decrease between 6% and 25% after a crime, depending on the type of crime. These predictions are manifolds of the estimated effect on prices documented in this paper and in the literature. The welfare effects of crime are not uniform across market participants and can elicit considerable disappointment to uninformed buyers that move into a high-crime neighborhood.