Association for Public Policy Analysis & Management

Although there is a growing literature on how mortgage lenders decide to extend credit to local areas affected by severe weather shocks, like hurricanes and floods, local disasters can take many other forms. In April 2014, the city of Flint, Michigan began suffering a major public health crisis following a switch in the city's water source led to increased lead contamination. Leveraging the fact that the city of Flint changed its water source while the rest of Genesee County did not, we employ standard difference-in-differences approach comparing census tracts within the city of Flint to those outside the city.

Our topline result is that individuals living in Flint experienced a reduction the supply of mortgage credit along both the intensive and extensive margins. We find a roughly 12 percentage point increase in the mortgage application denial rate for homes in Flint census tracts relative to border census tracts. At the same time, demand for credit in Flint remained close to pre-crisis levels. Using administrative credit bureau data, we show that this retraction in the supply of credit cannot be fully explained by a change in the riskiness of the borrower pool.

Instead, we present multiple pieces of evidence consistent with the denials being driven by a downward shift in lenders’ perceptions of property values (via appraisals) relative to agreed transaction prices. A low appraised value increases the likelihood that the loan is no longer conforming. This result may be of concern to policy makers as it implies that the private mortgage market is not consistently lending following local disasters and may require targeted interventions.

In ongoing analysis, we compare mortgage loan decisions of banks with a strong Flint deposit presence to national banks. Our goal is to determine whether the U.S.’s continued departure from a traditional savings-and-loan banking model exacerbates the rate of loan denials following a local disaster.