New data analysis helps identify future foreclosure trouble spots
An analysis conducted by the Minneapolis Fed and the Federal Reserve Board reveals which areas of the Twin Cities region are at greatest risk of becoming mortgage foreclosure hotspots.
Andreas Lehnert - Economist, Board of Governors
Michael Grover - Assistant Vice President, Community Development
Published May 1, 2008 | May 2008 issue
Nationally, foreclosure rates have steadily climbed to record levels. Recent research suggests that they may keep climbing for a year or more. According to information from the Board of Governors of the Federal Reserve System (Board), roughly 200,000 active subprime ARMs (adjustable rate mortgages) underwent their first rate reset in each quarter of 2007. Monthly payments for an average of 380,000 subprime mortgages are scheduled to undergo their first interest rate reset in each quarter of 2008. Weak housing prices are also contributing to foreclosures on these and other mortgages.
Concerned policymakers and financial regulators, including the Board, have encouraged mortgage lenders, servicers, and investors to increase efforts to mitigate foreclosures. Community organizations, counseling organizations, lenders, and loan servicers have formed local and national foreclosure-mitigation partnerships, such as Minnesota's Foreclosure Prevention Funders Council and the national HOPE NOW Alliance, to promote such efforts.
For the various stakeholders that make up these foreclosure-mitigation partnerships, the geographic distribution of mortgage-borrower distress is a matter of great concern. Timely and detailed information about borrowers who are in trouble and the number of vacant properties resulting from foreclosures is generally lacking. Useful information on future foreclosure patterns remains even more elusive.
To help fill this knowledge gap, the Federal Reserve Bank of Minneapolis and the Board recently analyzed state and ZIP Code-level reports on the aggregate performance of loans in Minnesota and the Twin Cities region that are securitized in subprime and alt-A pools. On a continuum that measures the likelihood or risk that a borrower may default on a mortgage over time, subprime refers to loans that are considered the riskiest, alt-A refers to loans that are less risky than subprime, and prime refers to loans that are considered least likely to enter default.
Our analysis uses data from First American LoanPerformance (LP), a firm that tracks and analyzes the performance of securitized mortgages. The aim of this and future analyses is to help foreclosure-mitigation partnerships better direct resources and bring public attention to neighborhoods and borrowers in the greatest need of some sort of assistance. Our analysis, described below, reveals which geographic areas are at greatest risk, by some criteria, as future mortgage trouble spots. Specifically, these potential trouble spots are the exurban and center-city areas of the metropolitan region.
Growth and fusion
First, some background on the data we used. This background discussion includes a brief overview of two recent and related changes in the mortgage market: namely, mortgage securitization and the use of risk-based pricing for mortgage loans.
In the past, lenders originated, serviced, and owned their mortgages. However, in recent years, it has become more common to separate these functions. Typically, mortgages are now pooled and sold to secondary market investors, while the rights to service the loans are sold to a servicer, a firm that specializes in conducting this activity for a fee. The share of U.S. residential mortgage debt in a mortgage pool or trust has grown in the past decade. As of the second quarter of 2007, it accounted for 57 percent of total mortgage debt.1/
As the use of securitization expanded, lenders found that investors had a ready appetite for securities backed by nonprime (that is, subprime and alt-A) loans. By 2006, the number of nonprime mortgage originations increased substantially, accounting for 40 percent of all newly securitized mortgages, compared to only 9 percent in 2001.2/
In retrospect, it appears as though the chain of securitization failed to align the interests of mortgage originators, who earned fees by making loans, and investors in mortgage-backed securities, who ultimately bore the credit risk of the loans. Why investors did not exert sufficient oversight to ensure the quality of securitized mortgages is beyond the scope of this article.
The data used in our analysis capture a good deal of information about the simultaneous growth in and fusion of nonprime lending and mortgage securitization. In particular, the data include a sizable proportion of all loans sold into subprime or alt-A securities. As noted above, alt-A and subprime loans are considered riskier than prime loans and more prone to default. The risk is due mainly to quality and size considerations that make these loans "nonconforming" in the eyes of Fannie Mae and Freddie Mac.3/
Our understanding is that LP captures about 70 percent of subprime securities and 95 percent of alt-A securities. Still, it is important to remember that these data do not include any loans held on a bank's books. The data used in this article are from October 2007.
For our analysis, we began with a review of statewide statistics for Minnesota. We selected Minnesota because, of the six states in the Ninth Federal Reserve District, it currently has the largest share of loans that LP tracks. We then examined the data at the lowest level of geographic identification available: individual five-digit ZIP Codes. At the ZIP Code level, we examined the ways in which loans in one area differed from the loans in another.
We focused our geographic analysis on the sequence of mortgage distress, beginning with loans that are current on payments and ending with those that are foreclosed and bank-owned. Specifically, we examined the number of loans characterized by the following variables:
- loans with a current payment;
- delinquent loans;
- variable rate loans set to reset in 2008;
- loans in foreclosure; and
- loans classified as "real estate owned" or "REO," meaning the borrower has lost the home to foreclosure and the home is now owned by the loan servicer.
Finally, we identified the geographic patterns of note from the resulting maps.
Case study: A statewide view
As of October 2007, LP tracked 39,200 alt-A and 54,300 subprime loans in Minnesota. These loans represent 2.7 percent and 4.5 percent of all mortgaged owner-occupied properties in the state, respectively. As the table below indicates, borrowers originated most of these loans after 2004. The newness of the loans is consistent with the view that the nonprime lending boom is a relatively recent phenomenon. Note also that alt-A loans, on average, had a higher balance and a lower interest rate than subprime loans; this is consistent with the view that alt-A pools largely contain loans that lenders perceived as less risky than subprime loans. Borrowers used almost half of the subprime loans to refinance an existing mortgage and pull cash out. This practice was less common for alt-A mortgages.
A snapshot of Minnesota's nonprime mortgagesFrom data tracked by First American LoanPerformance
As of October 2007
As of October 2007, 8.4 percent of the subprime and 2.5 percent of the alt-A owner-occupied loans in Minnesota were in foreclosure, and an additional 11 percent of subprime mortgages were REO. Some loans with variable rates (38 percent of subprime and 31 percent of alt-A mortgages) had already undergone their first rate reset; however, the majority of variable rate loans have yet to face their first rate reset. While the vast majority of variable rate subprime loans are scheduled to reset by the end of 2008, reset dates for six out of every ten alt-A loans will occur in 2009 and beyond.
While all of this statewide, aggregate information is useful intelligence to foreclosure-mitigation efforts, these groups have recently raised additional concerns about the last set of statistics—that is, future rate resets, which will create higher monthly payments that could limit borrowers' ability to repay their loans. To date, data on foreclosure patterns has focused, with a good deal of precision, on foreclosures that have already happened. While efforts to predict future foreclosure patterns using statistical models exist,4/ actual loan-based data may offer greater potential for targeting foreclosure-mitigation strategies.
Click on map to view larger image.
Case study: A ZIP Code-level view of nonprime loans
When mapped at the ZIP Code level using GIS (geographic information systems) software, information about the status of mortgage loans (if they are current, delinquent, will reset some time in the future, in foreclosure, or in REO) tell a more nuanced story.
For foreclosure-mitigation groups, each finding from the data presents a distinct problem that requires its own distinct strategy. In other words, groups focused on helping distressed borrowers will need to deploy different resources in different geographic areas, depending on whether the area has many properties that are already in foreclosure or has many borrowers that are current on their mortgages but face significant payment increases. For example, areas where there are many homeowners facing rate resets may require a different form of assistance, such as one-on-one financial counseling, than areas where the majority of the properties have already been through foreclosure and are now classified as REO.
Overall, the geographic pattern of borrower distress for the Twin Cities indicates that some areas are already hard hit by delinquencies and foreclosures, while others are at risk of future increases in delinquency rates.5/ According to the LP data, the proportion of nonprime loans that are current ranges from 35 percent to 85 percent across Minnesota ZIP Code areas. In Hennepin and Ramsey counties, which encompass the Twin Cities of Minneapolis-St. Paul and many of their suburbs, the suburban ZIP Codes tend to have a relatively high proportion of current loans. In contrast, ZIP Codes in the exurban fringe to the north, west, and southwest of the two core counties show a lower proportion of loans that are current. These parts of the region have experienced significant population growth and new housing development since the late 1990s.
Portions of the two central cities also exhibited lower proportions of loans that are current. For example, ZIP Code 55411 in North Minneapolis had the lowest rate of current nonprime loans in the state, at 34.9 percent. Previous analysis has shown that this area has already experienced high rates of foreclosure.6/
Loan delinquency patterns can often portend future difficulties. For example, four out of every ten loans in the U.S. that were 60 days delinquent in 2007 deteriorated further into delinquency and foreclosure. Our analysis of the Twin Cities area reveals that delinquency rates were highest in suburban ZIP Codes outside of the two central cities, in communities as geographically varied as Forest Lake, Oakdale, and Lakeville. In contrast, delinquency rates in the neighborhoods around downtown St. Paul and in portions of North Minneapolis were lower.
So, does this mean nonprime mortgage loans in the central cities are in better shape than those in the suburbs? Not exactly. It is important to remember that the data used here are only a snapshot of loan activity. Lower delinquency rates in these areas may be deceptive, since a higher proportion of the areas' loans are already in foreclosure or REO and, hence, are no longer classified as delinquent. For example, in St. Paul's North End neighborhood (ZIP Code 55101), 20 percent of nonprime loans are delinquent, but 33 percent are in foreclosure or REO.
In addition, ZIP Codes in some of the hardest-hit areas of the central cities, including those on the east side of St. Paul and in North Minneapolis, have many loans scheduled to undergo their first rate resets this year. Thus, some of the communities that have already seen high rates of borrower distress and foreclosure are at risk for continued deterioration of credit quality.
In suburban areas of Hennepin and Ramsey counties, the distribution of scheduled rate resets shows the same geographic pattern as the distribution of borrower distress. Some suburban communities in other counties have a large number of loans scheduled to undergo their first rate reset in 2008. Suburban communities such as Ham Lake, Apple Valley, Shakopee, and portions of Woodbury will have high concentrations of rate resets this year on variable rate loans.
As expected, properties classified as REO tend to be concentrated in areas where, by all accounts, foreclosure rates have been highest. Such areas include North Minneapolis and nearby suburbs, as well as the neighborhoods around downtown St. Paul. For example, close to one-third of all LP-tracked loans in ZIP Code 55411 in North Minneapolis were classified as REO. This same ZIP Code has only 12 percent of the loans in foreclosure, suggesting that the crest of the foreclosure wave may have already passed through these neighborhoods. Higher rates of loans in REO, as compared to foreclosure, were also evident in suburban communities like Jordan and Belle Plaine in rural Scott County and along the Interstate 94 corridor between Minneapolis and St. Cloud.
A useful, but limited, contribution
Our analysis of the LP data presents a considerable amount of information about the geographic pattern of borrower distress in Minnesota and the Twin Cities, albeit among a select and relatively risky group of loans. We find that delinquency and foreclosure rates are highest in inner-city and exurban neighborhoods. We also find that relatively high proportions of borrowers in some suburban communities are scheduled to undergo interest rate resets this year.
Each geographic concentration of borrower distress requires a distinct approach by interested community groups. Unfortunately, our analysis cannot provide all the information required to help community groups target their resources. In particular, we cannot identify individual loans and properties in distress, nor do the data cover the entire residential mortgage market. Nonetheless, we believe the trends revealed in our analysis will be a useful contribution to foreclosure-mitigation efforts.
Andreas Lehnert is an economist with the Board of Governors of the Federal Reserve System. Michael Grover is the Community Affairs Manager at the Federal Reserve Bank of Minneapolis.
For more information: Mortgage maps and data
The Federal Reserve System recently created a set of dynamic maps that illustrate subprime and alt-A mortgage loan conditions across the U.S. The maps are based on First American LoanPerformance data and will be updated monthly. To access the maps and related data, visit www.newyorkfed.org/regional/subprime.html.
1/ Source: 2007:Q3 flow of funds accounts of United States. Of this 57 percent, 37 percent were prime conforming mortgages guaranteed by a housing-related government-sponsored entity such as Fannie Mae, and the remaining 20 percent were in so-called "private label" securities.
2/ Danielle DiMartino and John Duca, "The Rise and Fall of Subprime Mortgages," Economic Letter, Federal Reserve Bank of Dallas, November 2007.
3/ Frank Raiter and Francis Parisi, "Mortgage Credit and the Evolution of Risk-Based Pricing," Working Paper Series, Joint Center for Housing Studies, Harvard University, February 2004.
4/ Michael Grover, Laura Smith, and Richard M. Todd, Targeting Foreclosure Interventions: An Analysis of Neighborhood Characteristics Associated with High Foreclosure Rates in Two Minnesota Counties, Federal Reserve Bank of Minneapolis, June 2007. Available at www.minneapolisfed.org/community/community-development/publications-and-papers/.
5/ The analysis excludes ZIP Codes with fewer than 25 owner-occupied loans.
6/ Jeff Crump, "Subprime Lending and Foreclosure in Hennepin and Ramsey Counties," CURA Reporter, June 2007, p. 14-18.