This paper studies the impact of a change in the resolution policy of the Federal Deposit Insurance Corporation (FDIC) on bank and thrift behavior in the midst of the banking crisis of the 1980's. This study examines whether the change in the policy stance of the FDIC towards more stringent resolution methods prompted banks and thrifts that were at the most risk of failure to improve their financial condition in order to avoid failure. The research questions will be addressed using Bayesian Nonparametric methods in two ways. First, the density of outcomes related to bank risk will be estimated using a Mixture of Polya Trees in order to test for the presence of low-risk and high-risk clusters. Secondly, the causal links between the change in FDIC policy and the banks' asset portfolio will be estimated using a two-step procedure. The probability of failure is estimated in the first step and the predicted probability of failure is incorporated into the second step, which models bank asset portfolio. Gaussian process priors are applied on the the mean function in order to ensure that the relationship between the FDIC's policy change and bank riskiness is estimated flexibly.