Abstract
We develop a non-parametric clustering model of spatial and/or spatio-temporal phenomena based on Kolmogorov entropy. The methodology will be tested using Federal Financing and Housing Administration (FHFA), formerly known as FHA, an agency of HUD (US Dept of Housing and Urban Development) quarterly HPI (Housing Price Index) data of 350 plus cities in the US. This multivariate data will be also analyzed with Principal Component Analysis (PCA) techniques to identify key regions involved in creating the housing bubble and its spread to the rest of cities. This provides evidence of the differential impact of the housing crisis across US regions and suggests an important role for state level as opposed to federal (centralized) regulations.
JEL Classification: A12, C01, C15, H30, R32, R38
