�Targeting Federal Aid to Neighborhoods Distressed by the Subprime Mortgage Crisis�

On behalf of U.S. Department of Housing and Urban Development (HUD) Deputy Secretary Roy Bernardi, thank you Chairman Kucinich and Chairwoman Waters for the invitation to appear at this joint hearing. I am Todd Richardson, Director of the Program Evaluation Division of the Office of Policy Development and Research at HUD.

My testimony focuses on allocation formulas, what data are available related to increasing rates of vacancy and abandonment, and the analysis HUD recommends pursuing in order to develop a thoughtful formula for targeting funds to neighborhoods.

I have worked on issues related to allocation formulas at HUD since the mid-1990s, including reports in 1995 and 2005 on how the Community Development Block Grant (CDBG) formula targets to community development need. My experience with research and development of allocation formulas identifies two key ingredients for a “successful” formula:

• Clearly defined goals of the need that Congress intends to target; and

• Available data that is uniformly collected across all potential grantees.

Careful analysis is required to ensure the available data match the goals which have been set by Congress. For this hearing you have identified a specific goal: targeting federal funds to neighborhoods most affected by rising rates of vacant and abandoned properties.

While there are promising data sets that may be able to achieve this Congressional goal, further analysis of the data is required to ensure accuracy in targeting funds to all communities across the country. As you know, the United States is a very diverse nation. Where one method is effective at targeting need in Cleveland, it might not be effective in Los Angeles.

In general, for allocation formulas HUD prefers to use data collected uniformly across the nation by a public agency. This is preferred because we can be more confident the formula will produce a fair allocation. Proprietary data is not preferred for three reasons:

1. The firms that produce the data may have other customers whose agendas conflict with those Congress has given us;
2. Those firms might set unreasonable terms and conditions for the use of the data; and
3. The public could not review proprietary data and therefore HUD’s allocations would not be transparent.

We might, however, compare the public data to privately-collected proprietary data to serve as a check on what the public data is showing. When all or most data sources point to the same answer, we have much greater confidence in the validity of our findings.

Combining information from several public data sets has the greatest potential for accurately targeting funds to areas with vacant and abandoned properties. The public data sets I know of that could be analyzed for these purposes are:

• United States Postal Service data on active and vacant addresses as provided to HUD every quarter at the block level and that HUD makes available publicly at the Census Tract level. These administrative data have some anomalies that we have not yet fully sorted out. Nonetheless, they are a rare data set that tells us what is going on in neighborhoods across the country in real time with very current information on the trend that is the subject of today’s hearing: increasing vacancy rates.

• The Office of Federal Housing Enterprise Oversight (OFHEO) Housing Price Index for Metropolitan Areas. These data are also available quarterly. They are available at the Metropolitan Statistical Area (MSA) and non-metropolitan balance of state level. MSAs with falling home values means more property owners will have negative equity in their properties, thus increasing the risk of foreclosure.

• Bureau of Labor Statistics data at the county level on Labor Force Participation and Unemployment. These data are available monthly and represent good measures of economic decline. Job loss means both a loss of income and often a need to relocate. If this is occurring in an environment where it is difficult to sell a home, the risk for housing vacancy is increased.

• Home Mortgage Disclosure Act (HMDA) data from 2004 to 2006 on census tracts with high-cost loans and/or high loan-to-income ratios. Individuals who have high-cost loans or are highly leveraged as measured by a high loan-to-income ratio could easily become unable or unwilling to continue to pay the mortgage on a home, especially if mortgage payments rise above those for which borrowers were initially qualified, if values fall, or if there is an economic downturn.

• Census 2000 Census Tract level data on vacancy and home value and American Community Survey data on vacancy and home value at the city and county level from 2006. If home values are falling and/or vacancies are increasing at the city or county level, the neighborhood-level concentration of these problems is likely reflected by 2000 census tract concentrations of low house values and vacant units. There is a very good chance that homes in these neighborhoods are not only vacant, but also are being abandoned.

It is highly likely that a careful combination of the information in these data sets could achieve the Subcommittee’s goal of developing a formula that targets neighborhoods with increasing numbers of vacant and abandoned homes.

An important step in developing the formula would be verifying the precision of the formula’s targeting by testing it against other available data sets that might capture some portions of the vacancy problem. As noted above, it is preferable to have multiple sets of data lead to the same conclusion. For example, there are private data sets on foreclosures, lender-owned properties, and homes for sale. While these other data sets do not have full coverage for every community in the United States, nor do they represent all of the reasons a unit may become vacant, they represent some of the country and some of the reasons a unit may become vacant, and thus can be used as a check on the public data.

Those data sets include:

• The Mortgage Bankers Association (MBA) National Delinquency Survey, which provides information on approximately 80 percent of all loans being serviced and their delinquency rates by type. MBA collects this data from loan servicers and provides this information down to the state level every quarter. States with increasing delinquencies and foreclosures might be expected to also have increasing vacancy rates.

• Loan Performance (also known as True Standings) data on delinquent loans or loans in foreclosure (but with no information on foreclosure completions). Loan Performance has information from loan servicers on roughly 80 percent of active prime loans, but a smaller share (about 50 percent) of active subprime loans. These data are available at the MSA and zip code levels to identify foreclosure risks. Loan Performance also has compiled data from the private mortgage backed securities market that covers over 50 percent of outstanding subprime loans. The Loan Performance securities data does contain information on foreclosure completions.

• McDash Analytics, which has data on approximately 30 million of the 55 million active mortgages in the U.S., and has detailed information on loan characteristics at the MSA level, including default status and whether a loan is a Real Estate Owned (REO) property.

• Housing “agency” data from the Federal Housing Administration, Fannie Mae, and Freddie Mac on the number of REO properties – that is, houses which have gone through foreclosure, or other liquidation methods (such as deed-in-lieu of foreclosure), and are owned by these agencies. Fannie Mae and Freddie Mac may not wish to disclose detailed REO data because they may consider it proprietary and confidential. However, the combined agency data, if made available by the agencies, would contain REO counts for a large portion of the mortgage market. It is possible that some measure of the “time in REO” (time it takes to sell these properties) could be estimated to identify areas where foreclosed properties are sitting vacant for long periods of time.

• The National Association of Realtors has data on local housing market conditions it may be willing to share.

• The Case-Shiller home price index for 20 metropolitan areas. These data can be used to compare against the OFHEO Home Price Index noted earlier.

In summation, our available public data are the best data to be used for an allocation formula, but careful analysis needs to be done before a specific formula is established.

Thank you for the opportunity to appear before the Subcommittees today. The Department looks forward to working with Congress on this issue.