Neighborhood Stabilization Program Grants - Detailed Description of the Underlying Data

The estimated rate of foreclosure problems do not reflect "real" numbers of foreclosures but rather reflect neighborhood characteristics that are estimated to have a high level of risk for foreclosure, notably a concentration of loans made between 2004 and 2007 that were determined to have been “high cost” relative to other loans made at that same time, "highly leveraged" loans which are loans where the mortgage-to-income ratio is very high (see http://www.huduser.org/periodicals/ushmc/summer08/summary.pdf), falling home values in metropolitan or non-metropolitan balance of states, and both the average unemployment rate for the county in 2008 and the change in average unemployment rate between 2007 and 2008. The model is adjusted at the state level to match the rate of total foreclosure starts in 2007 and 2008 in combination with loans 90 day delinquent at the fourth quarter of 2008 from the Mortgage Bankers Association National Delinquency Survey. If county level data from private vendors are higher than the model score, county level model scores are adjusted upward to account for this higher rate.

The "vacancy" risk score is only calculated for Census Tracts which have a "foreclosure" risk score of 10 or higher and where the number of USPS residential addresses represent at least 75 percent of the total housing units in a Census Tract as of 2000. If a Tract does not meet this criteria, their “vacancy” risk score is based only on the "foreclosure risk" score. For those that do meet the criteria, they receive a score based on their ranking from multiplying the estimated foreclosure problem rate times the vacancy rate. The vacancy rate is calculated as the percent of addresses 90 days or more vacant as of March 2009 divided by the total USPS addresses in the Tract. Tracts are ranked from 1 to 20 on the result of this cross.

A more detailed methodology will be posted shortly.