Written for state Housing Finance Agencies (HFAs), this report furthers the work of the Innovations in Manufactured Homes (I M HOME) initiative s explorations into manufactured home mortgage data. This report explores how HFAs have delivered this performance and how other HFAs can improve performance and increase manufactured home mortgage lending.
Introduction... 1 Why Manufactured Housing?... 1 Three factors can trump traditional loan underwriting standards... 4 High-Touch Servicing... 4 Manual Underwriting... 6 Loan Counseling... 8 Summary Conclusions... 8 Appendix A... 9
The Innovations in Manufactured Homes (I M HOME) data project began in 2011 to identify issues in manufactured housing (MH) lending that impede consumer access to this essential housing resource for low-income families. In February 2013, the project published a report on MH appraisal issues, titled Real Homes, Real Value, and in March 2013 it published an MH loan origination and performance analysis, titled Toward a Sustainable and Responsible Expansion of Affordable Mortgages for Manufactured Homes (the Mortgage Report ). The Mortgage Report revealed several best practices in MH lending of which originators, investors and policymakers ought to be aware in order to make informed decisions that benefit low-income households without compromising the bottom line. Since State Housing Finance Agencies (HFAs) prioritize supporting lower-income homeownership, they are a natural first audience for improved lending and loan servicing practices in MH. Further, the Mortgage Report identified that MH mortgages purchased or originated by state HFAs achieved overall high performance when compared to other originators and investors. Therefore, this report is, in large part, geared towards HFAs and exploring how state HFAs have delivered this performance and how other HFAs can improve performance and increase MH lending. This best practices summary distills some of the key findings of the MH Mortgage Report, along with follow-up data collection and interviews with selected high-performing HFAs identified in the initial report. Manufactured homes are owned by eight million Americans and house over seventeen million residents. The average income of an owner of a manufactured home is $26,000 (between 50-60% of average median income), as compared to slightly over $56,000 for owners of non-manufactured homes. More than one-fifth of MH residents have incomes at or below the federal poverty level. 1 homeownership of manufactured homes represents the largest unsubsidized homeownership option in the United States. MH is particularly important as an affordable housing resource and currently represents the largest supply of new affordable housing units in the US. MH is affordable under new Qualified Residential Mortgage requirements for over 63% of low-income households; in contrast, only 26% of site-built homes are affordable to the same low-income demographic. The productioncost savings between a manufactured home and a site-built home are equally striking: $42 per square foot for MH and $86 for site-built. Manufactured homes cost less than half per square foot to produce and are made in factories across the United States, providing well-paying jobs in typically rural communities. Manufacturers complete these homes in one-fifth the time it takes to build a sitebuilt home and with 30% less waste. Manufactured homes are constructed in factory conditions to the specifications of the HUD Code, a national standard first implemented by the U.S. Department of Housing and Urban Development in 1976 to ensure safety, quality and durability of manufactured homes. Manufactured homes are neither necessarily trailers, nor mobile. Manufactured homes are built in a wide range of sizes and 1 Income data calculated by Housing Assistance Council (HAC) from American Housing Survey Data 1
styles for many different market segments, from families with children to singles and retirees. Manufactured homes are well suited to rural locations, although they can be found in urban and suburban settings. They are marketed to a broad range of incomes, from low- and moderate-income to affluent households. Lenders and investors are often under the impression that loans secured by MH perform badly. Fannie Mae, for example, is so skeptical about MH mortgages that it places a rate premium on all MH loans, and precludes state HFAs from including them in loan sales or securitizations. The I M HOME Data Project has collected underwriting and performance data from originations totaling $1.82 billion (hereinafter MH Mortgage Dataset ) through 2013 from 21 organizations including 14 state HFAs, three credit unions, two banks, one community development financial institution (CDFI) and the U.S. Department of Agriculture (USDA) (see appendix A for a complete list of participants). Approximately $866 million in loan data were collected from the 14 participating state HFAs. With a non-delinquent performance rate of 89.5%, the MH Mortgage Dataset performs equally with the OCC Government Guaranteed loans 2 dataset (http://tinyurl.com/kd9gs86), which has an 89.2% performance rate during the same loan period for each dataset. Figure 1 illustrates these findings. The performance of the USDA Guaranteed loans from the MH Mortgage Dataset, by contrast, is not as strong: only 77.9% of the loans are performing. 3 This comparison indicates that some MH loans 2 The government guaranteed loans category for the OCC is the most similar in terms of loan type, underwriting characteristics and borrower profile to the loans in the MH Mortgage Dataset. In both loan sets, loan sizes (and thus property values) are low to moderate, the average loan-to-value ratio is relatively high, and the borrowers are middle- to low-income households with slightly higher than average DTI ratios and lower than average credit scores. 2
do, in fact, perform poorly. However, the performance of the HFA-purchased USDA (Guaranteed) MH loans, when compared to the entire USDA Guaranteed loan dataset, which is overwhelmingly comprised of mortgages secured by site-built homes, indicates that state HFAs deliver significantly improved loan performance compared to loans sold to Ginnie Mae. 4 A comparison of the loan performance of two categories of I M HOME MH data providers Originators and HFAs shows that traditional underwriting criteria (high FICO scores and/or low loan-to-value ratios) predictably correlate with performance. Originators, including banks, credit unions and CDFIs, reported a combined dataset with a significantly higher weighted average FICO score than that of all loans purchased by HFAs (738 vs. 691, respectively). Similarly, the average weighted loan-to-value (LTV) ratio reported by the Originators is significantly lower than that of all loans purchased by HFAs (76% to 94%, respectively). See Figures 2 and 3 below: Two HFAs that provide high-touch servicing and employ manual underwriting practices yielded loan performance better than all the other participating HFAs as a group. In every FICO and LTV band and for every single loan product type, these two HFAs demonstrate better performance. The charts below provide a breakdown of loan performance by the Idaho and Pennsylvania HFAs, two high-performing HFAs that participated in the I M HOME data project. 3 Performing equals 60 days or less late, including current, not including paid-in-full. All USDA Guaranteed and Direct loan data received from USDA under a FOIA request. 4 USDA Direct loans were not included in the comparison as their interest rates, which can be as low as 1%, are dependent upon the applicant s ability to pay. USDA Direct loans perform better than Guaranteed. 3
Several of the lenders and two of the state HFAs (Idaho and Pennsylvania) that participated in the Data Project retain servicing rights and provide high-touch protocols rather than relying upon thirdparty servicers using more traditional protocols. Not only do these approaches reduce foreclosures and keep more borrowers in their homes, they also reduce losses from homes that do proceed through foreclosure and real estate-owned (REO) proceedings. High-touch servicing involves compliant protocols that can include: Outreach to all borrowers in the first six months after boarding to introduce the servicing division s availability and ability to work through borrower issues. Establishing and maintaining borrower contact and trust delivers better long term performance. Initiating outreach to late-paying borrowers as early as 17 days after a late payment (as opposed to industry standard of approximately 30 days). Undertaking a wide range of compliant (pursuant to the Fair Debt Collection Act and other regulatory boundaries) approaches to establishing contact with the delinquent borrower, including: Rotating servicing staff work schedules to speak with borrowers when they are available. Dedicated staff assignments that support borrower continuity and trust and avoid different borrower interactions with different staff resulting in delays. Timely coordination of third-party vendors, including property inspectors. Referrals to nearby and qualified loan or credit counselors that can support the improvement of borrower back-end debt situations and establish borrower representatives to help the borrower organize necessary evidentiary materials. 4
Supervisory oversight of foreclosure notices and review of foreclosure attorney activities. Offering short- and long-term (up to 36 months) loan adjustments and loan modifications based upon affordable and sustainable approaches supported by evidence provided by homeowners. No one size fits all formulas. Second (or possibly third) attempts at loan adjustments and/or modifications. If the borrower is unwilling to work with the servicer, legal action is the last resort. High-touch loans perform better regardless of loan product type, and perform better even when specific underwriting metrics are considered. That specific approaches to loan servicing delivered such improved loan performance outcomes was an unexpected and significant finding from the loan data, as information about loan servicing was not a part of the initial data request. Initial reviews of loan performance indicated that some organizations had particularly strong performance; follow-up with these organizations suggested that their approach to servicing was driving superior loan performance. The data suggest that such high-touch serviced loans from the HFA Dataset portfolio perform significantly better than those serviced by unaffiliated third party servicers that use traditional loan servicing approaches. HFAs across the country are beginning to recognize their peers expertise in increasing loan performance through high-touch loan servicing. These and other HFAs with established internal servicing or sub-servicing capacities may be available for engagement by their brethren HFAs. Some HFAs have begun to contract with third-party servicers who provide some or many high-touch protocols at competitive pricing. Two of the HFAs that participate in the MH Data Project effort, the Pennsylvania Housing Finance Agency and the Idaho Housing and Finance Association, possess their own internal servicing divisions using high-touch protocols. Both require loans sold to them be done so servicingreleased, meaning that the HFA purchases both the loan asset and the mortgage servicing rights (MSRs) to collect payments from the borrowers, manage escrows and work with the borrowers if they become late on payments. The purchase of MSRs requires the HFA to pay the value of the MSR to the loan originator, called a servicing release premium (SRP). The SRPs can be structured to incent the originating lender to deliver all necessary documentation to allow quicker implementation of the servicing protocols. These two HFAs are working to structure their loan servicing pricing to allow their services to be offered on a competitive basis to other HFAs. 5
Manual, as opposed to automated, underwriting has long allowed loan products to reach lower scoring or no-fico applicants and low-income families who often can only afford low downpayments. The combination of the loss of the ability by HFAs to sell tax-exempt securities, tougher new underwriting standards for all loan products and the decline of the mortgage insurance (MI) industry have resulted in manual underwriting to be available only in government-insured or portfolio-held loans, which has led to attendant increases in weighted average FICOs for HFA portfolios as a whole. Previously, state HFAs were able to purchase or originate Self-Insured loans, which are loans over 80% LTV that do not possess either mortgage insurance or any government insurance or guarantee. In addition, several HFAs either have or have had affiliated MI companies with more flexible underwriting standards than national MI companies. The data from both of these loan product types are joined here and are called Self-Insured. The data for Self-Insured loans and other manually underwritten loans demonstrated superior performance to conventional loans with mortgage insurance, even though Self-Insured loans feature a relatively low-weighted average FICO score (675) and a relatively high-weighted average LTV (93%) compared to conventional loans with mortgage insurance. Self-Insured loans (SI in Figure 7 below), originated or purchased by nine organizations in the MH Mortgage Dataset, delivered better performance than all comparable products. In fact, the Self- Insured loans were the second-best performing loan product after Conventional loans, defined as those GSE-eligible loans with LTV < 80%. Self-Insured loans perform well even for borrowers in the 640-679 FICO band (a 97.9% performance rate) and, interestingly, very well in the less-than-600 FICO band (100%), but do not perform nearly as well for borrowers with middle-low credit scores (600-639). The performance discrepancy between the two lowest FICO band merit further exploration. 6
An important component of Self-Insured loans is that they require manual, as opposed to automated, underwriting. This includes the ability to apply alternative credit considerations and allow loan underwriters to approve applicants possessing FICO scores below generally accepted minimums. These loans charge a higher interest rate to qualified borrowers both to provide yield to the investor to include these loans in rated securitized offerings (previous tax-exempt offerings) and possibly to offset potential higher costs of manual underwriting (7.0% weighted average interest rates for Self Insured versus 5.9% for Conventional with MI). Additionally, some HFAs that included Self-Insured loans in their (prior) rated securitized tax-exempt offerings were required to apply additional equity supplements to the offerings for these loans, whose yield cost could be offset by the increased loan interest rates. The Data Study did not capture upfront MI fees to borrowers and so the increase in interest rates between Self-Insured and MI might in fact overstate the increase in interest rate charge when comparing the two products if the upfront fee were included in the calculations. The slight edge in loan performance of the Self-Insured product suggests that manual underwriting can pay for itself and even lead to better investor yields. At the same time, the Self-Insured product s relatively lower FICO scores and higher LTVs suggest that its marginally higher interest rates may not be a barrier to effectively meeting the home finance needs of some low- and moderate-income borrowers. In the HFA-only data set, average loan sizes for Self-Insured loans ($112,396) are similar to Conventional with MI ($107,961) and the weighted Debt-to-Income ratio (DTI) is lower for Self- Insured loans compared to Conventional with MI (40.4% versus 38.5%). Despite threshold underwriting metrics that are less stringent and based upon actual credit performance as opposed to FICO only, the Self-Insured loans in the MH (HFA-only) Mortgage Dataset perform better than the Conventional loans with MI, with a 92.7% performance rate versus an 88.1% performance rate. The performance results suggest that the Self-Insured product, with its manual underwriting of applicants, produces results that are highly competitive with Conventional mortgages with mortgage insurance, and allows nontraditional but creditworthy borrowers to access affordable financing. If Self-Insured products were still available, many potential manufactured homeowners would likely qualify for mortgages as buyers of this housing largely borrow outside of mainstream housing finance. The recent increase in FHA insurance and the newly extended loan insurance coverage terms by FHA make even government-insured loans more difficult to approve. These have resulted in significant access barriers for homeownership by low-income families. The relatively low cost of MH, however, is one reason why this housing type can help expand affordable, responsible homeownership, despite this and other such barriers. The performance of Self-Insured loans strongly suggests that a pilot program that allows these loan types to again be offered would increase homeownership opportunities for the lowest-income homebuyers and not compromise loan performance. Additional credit enhancements may be necessary for such a pilot as the secondary financial markets are only beginning to again consider tax-exempt or taxable offerings for HFAs. One such enhancement could be a requirement for credit or homeownership counseling for all Self-Insured applicants. 7
In addition to manual underwriting, homeownership counseling may contribute to the success of Self-Insured mortgages. While most state HFAs as well as Mortgage Data Project participants support counseling, no project participant provided data as to whether or not specific borrowers received loan or homeownership counseling. Some HFAs do possess a data field on which borrowers have received data and this suggests a follow-up that could provide performance comparisons for counseled and non-counseled borrowers. It is advised that, going forward, all HFAs capture or produce a data field that would allow the HFA and others to compare performance. A follow-up survey with several HFAs uncovered that homeownership counseling is required for homebuyers whose FICO scores are under 680 at or prior to origination. The weighted average FICO for Self-Insured loans was 672. It is therefore likely that many of these borrowers received loan and/or homeownership counseling prior to loan origination and that this counseling contributed to improved loan performance. High-touch servicing enhances the performance of manufactured home mortgages. Manual underwriting to higher-risk borrowers can lead to better performance than would automated underwriting. The design and implementation of credit enhancements supporting Self-Insured loans will increase the delivery of affordable and performing manufactured home mortgages. There are indications that homeownership counseling improves manufactured home mortgage performance. Self-Insured mortgages have historically high performance, but are no longer offered since the publication of the I M HOME data project report. There is a need for credit enhancements to re-engage lenders and investors in this high-performing product that reaches lower-income borrowers. 8
An alphabetical listing of project participants that have given us permission to list their names: BECU Bank2 Community Development Bank Colorado Housing Finance Authority Delaware State Housing Authority Hope Credit Union Idaho Housing & Finance Association MaineHousing Minnesota Housing Finance Agency Montana Board of Housing New Hampshire Community Loan Fund New Hampshire Housing Finance Authority New Mexico Community Development Authority Pennsylvania Housing Finance Agency Self-Help Credit Union State of New York Mortgage Agency Texas Department of Housing & Community Affairs U.S. Department of Agriculture Vermont Housing Finance Agency Washington State Housing Finance Commission Wyoming Community Development Authority 9