LAUSR

Abstract We illustrate for two U.S. real estate markets how spillover effects may contaminate a causal impact analysis that relies on a research design with a control unit. Our two markets (Chattanooga, TN and Spartanburg, SC) are linked via dominant local employers (VW and BMW) that compete with each other in the automotive sector. The spillovers originate with an unanticipated negative news event for VW (2015 emissions scandal), which made the company’s very survival uncertain for a limited time period. The troubling news for VW translated into favorable news for BMW’s competitive position. In VW’s location, the demand for homes and prices declined temporarily; in BMW’s location the reverse occurred. A standard impact analysis of the VW scandal on the local housing market in Chattanooga using BMW’s Spartanburg location as a control would have overestimated its impact as a result of the spillover effects.