Following the collapse of property values and an increasing rate of default on high-risk mortgages, the United States experienced a subprime lending crisis that led to massive financial losses for holders of mortgage-backed securities. The purpose of this correlational study was to examine if loan product and loan amount predict the likelihood of loan foreclosure. The theoretical framework grounding the study was Minsky's financial instability hypothesis, which describes the basis of capitalism as economic expansionism followed by financial crises. The population consisted of 473 loan cases from archival data of the Atlanta Sixth Federal Reserve District in Georgia. The method used to collect the data was a probabilistic simple random sample taken from the archival data. The use of binary logistic regression resulted in a finding that the variables of loan product and loan amount significantly predicted the likelihood of loan foreclosure, Ï?2(4) = 10.65, p = .031, Nagelkerke R2 = .09. The Nagelkerke R2 value indicated that the model explained 9% of the variability in foreclosure. The findings specifically showed that Federal Housing Authority and Veterans Administration loan products were significantly more likely than conventional loans to cause losses for mortgage lenders. The implications for positive social change include increased stakeholder knowledge of various factors that can contribute to foreclosure and sustainment of community value with fewer homeowners losing their home in foreclosure.