This is the accessible text file for GAO report number GAO-14-117 entitled 'Troubled Asset Relief Program: More Efforts Needed on Fair Lending Controls and Access for Non-English Speakers in Housing Programs' which was released on February 6, 2014. This text file was formatted by the U.S. Government Accountability Office (GAO) to be accessible to users with visual impairments, as part of a longer term project to improve GAO products' accessibility. Every attempt has been made to maintain the structural and data integrity of the original printed product. Accessibility features, such as text descriptions of tables, consecutively numbered footnotes placed at the end of the file, and the text of agency comment letters, are provided but may not exactly duplicate the presentation or format of the printed version. The portable document format (PDF) file is an exact electronic replica of the printed version. We welcome your feedback. Please E-mail your comments regarding the contents or accessibility features of this document to Webmaster@gao.gov. This is a work of the U.S. government and is not subject to copyright protection in the United States. It may be reproduced and distributed in its entirety without further permission from GAO. Because this work may contain copyrighted images or other material, permission from the copyright holder may be necessary if you wish to reproduce this material separately. United States Government Accountability Office: GAO: Report to Congressional Committees: February 2014: Troubled Asset Relief Program: More Efforts Needed on Fair Lending Controls and Access for Non- English Speakers in Housing Programs: GAO-14-117: GAO Highlights: Highlights of GAO-14-117, a report to congressional committees. Why GAO Did This Study: Treasury introduced MHA in February 2009 and indicated that up to $50 billion would be used to help 3 to 4 million struggling homeowners avoid potential foreclosure. Since then, questions have been raised about participation rates and the overall success of the program. The Emergency Economic Stabilization Act of 2008 requires GAO to report every 60 days on the Troubled Asset Relief Program (TARP) activities. This 60-day report examines (1) the status of MHA and steps Treasury has taken to increase program participation, (2) Treasury's oversight of the MHA-related fair lending internal controls of servicers, and (3) Treasury's and MHA servicers' policies and practices for ensuring that LEP borrowers have equal access to the program. For this work, GAO reviewed program documentation, analyzed HAMP loan-level data, and interviewed officials from Treasury, fair lending supervisory institutions, and the five largest MHA servicers. What GAO Found: Participation rates in the Home Affordable Modification Program (HAMP), a key component of the Making Home Affordable program (MHA), peaked in early 2010, generally declined during 2011, and remained relatively steady from 2012 through November 2013. As of November 2013, about 1.3 million borrowers had entered into a HAMP permanent modification. Treasury has made several efforts to increase participation, such as extending the program deadline through December 2015, expanding program eligibility requirements, and initiating the MHA Outreach and Borrower Intake Project. This project provides funding to counseling agencies to help borrowers complete and submit MHA application packages. The project was scheduled to end in December 2013 but was recently extended through September 2014. Treasury requires MHA servicers to develop internal control programs that monitor compliance with fair lending laws (the Fair Housing Act and Equal Credit Opportunity Act) but has not assessed the extent to which servicers are meeting this requirement. Treasury noted that it shares HAMP loan-level data with the federal agencies responsible for fair lending enforcement. GAO's analysis of HAMP loan-level data for four large MHA servicers identified some statistically significant differences in the rate of denials and cancellations of trial modifications and in the potential for redefault between populations protected by fair lending laws and other populations. Such analysis by itself cannot account for all factors that could explain these differences. Reviewing the fair lending internal controls of MHA servicers could give Treasury additional assurance that servicers are complying with fair lending laws. Despite an Executive Order issued in 2000 and a 2011 Attorney General's memorandum regarding improving access to federal programs for limited English proficiency (LEP) persons, Treasury only recently developed LEP-related written guidelines and procedures for the MHA programs. Treasury has taken measures to reach out to these borrowers and requires servicers to have a policy for “effective relationship management” with LEP borrowers. However, Treasury has not provided any clarifying guidance to servicers on what such a policy should contain or assessed servicer compliance with this requirement. Housing counselors have noted that LEP borrowers continue to encounter language-related barriers in obtaining access to MHA program benefits. Without a comprehensive strategy that includes guidance for servicers on engaging with LEP borrowers and monitoring of servicers, Treasury cannot ensure that all potential MHA participants have equal access to program benefits. Because the MHA program provides direct outlays of taxpayer dollars, it is important that Treasury take appropriate steps to ensure that all eligible borrowers, including those whose primary language is not English, have access to MHA program benefits. What GAO Recommends: Treasury should (1) assess the extent to which servicers have established internal control programs to monitor compliance with fair lending laws, (2) issue guidance to servicers on working effectively with LEP borrowers and (3) monitor servicers' compliance with the guidance. Treasury noted that it was considering GAO's recommendations and agreed that it should continue to strengthen its program. Treasury also provided technical comments that were incorporated into the report as appropriate. View [hyperlink, http://www.gao.gov/products/GAO-14-117]. For more information, contact Mathew J. Scirè at (202) 512-8678 or sciremj@gao.gov. [End of section] Contents: Letter: Background: HAMP Participation and Expenditures Have Remained below Expectations: Treasury Has Not Assessed Servicers' Internal Controls or Borrower Data Related to Fair Lending: Treasury Has Provided Servicers with Limited Guidance and Oversight on Interacting with Borrowers with Limited English Proficiency: Conclusions: Recommendations for Executive Action: Agency Comments and Our Evaluation: Appendix I: Objectives, Scope, and Methodology: Appendix II: Description of GAO's Econometric Analysis of HAMP Outcomes for Fair Lending Groups: Appendix III: Comments from the Department of the Treasury: Appendix IV: GAO Contact and Staff Acknowledgments: Selected Bibliography: Tables: Table 1: MHA Outreach and Borrower Intake Project Package Allocation to Housing Counseling Agencies: Table 1: HAMP Data Used in the Analysis by Phase of the HAMP Process, April 2009 to April 17, 2013: Table 2: Denial Rate and Percent Difference in Denial Rates between Selected Populations and their Comparison Populations for HAMP Applications Denied Due to a DTI Less Than 31 Percent: Table 3: Denial Rate and Percent Difference in Denial Rates between Selected Populations and their Comparison Populations for HAMP Applications Denied Due to an Incomplete Request: Table 4: Redefault of HAMP Permanent Modification and Percent Difference in Redefault Rate between Selected Populations and Their Comparison Populations: Figure: Figure 1: Monthly HAMP and Combined Tier 1 and Tier 2 Activity Levels through November 2013: Abbreviations: 2MP: Second Lien Modification Program: AIAN: American Indian or Alaska Native: CBO: Congressional Budget Office: CFPB: Bureau of Consumer Financial Protection: DOJ: Department of Justice: DTI: debt-to-income ratio: ECOA: Equal Credit Opportunity Act: EESA: Emergency Economic Stabilization Act of 2008: the enterprises: Fannie Mae and Freddie Mac: FDIC: Federal Deposit Insurance Corporation: FFETF: Financial Fraud Enforcement Task Force: FHA: Federal Housing Administration: FICO: Fair Isaac Corporation: FTC: Federal Trade Commission: HAFA: Home Affordable Foreclosure Alternatives: HAMP: Home Affordable Modification Program: HUD: Department of Housing and Urban Development: IR/2: Investor Reporting/2: LEP: limited English proficiency: LTV: loan-to-value: NFMC: National Foreclosure and Mitigation Counseling: MHA: Making Home Affordable: MHA-C: Making Home Affordable Compliance: NHPI: Native Hawaiian or Other Pacific Islander: NMS: National Mortgage Settlement: NPV: net present value: OCC: Office of the Comptroller of the Currency: OFS: Office of Financial Stability: PRA: Principal Reduction Alternative: RHS: Rural Housing Service: SIGTARP: Special Inspector General for the Troubled Asset Relief Program: SPA: Servicer Participation Agreement: TARP: Troubled Asset Relief Program: USDA: Department of Agriculture: VA: Veterans Affairs: [End of section] United States Government Accountability Office: GAO: 441 G St. N.W. Washington, DC 20548: February 6, 2014: Congressional Committees: The Making Home Affordable (MHA) program is one of the last active programs under the initially authorized $700 billion Troubled Asset Relief Program (TARP) created by the Emergency Economic Stabilization Act of 2008 (EESA).[Footnote 1] Among other things, EESA was intended to preserve homeownership and protect home values. The U.S. Department of the Treasury (Treasury) introduced MHA in February 2009, providing up to $50 billion to help as many as 4 million struggling homeowners avoid potential foreclosure. Treasury has subsequently reduced the amount obligated for housing programs under TARP to $38.5 billion, of which $29.9 billion is allocated to MHA programs. MHA's key component is the Home Affordable Modification Program (HAMP), which provides servicers and mortgage holders/investors with incentive payments for modifying first-lien mortgages. These modifications are intended to prevent foreclosures by reducing homeowners' monthly mortgage payments to affordable levels. Treasury has implemented additional programs that use TARP funds, including a program aimed at helping homeowners in areas hit hardest by the housing crisis and another to help underwater borrowers who owe more on their mortgages than their homes are worth.[Footnote 2] Entry into the MHA programs was originally scheduled to end in December 2012, but Treasury has extended the deadline twice, most recently until December 31, 2015.[Footnote 3] Only loans originated on or before January 1, 2009, and that meet other requirements, are eligible for assistance under the MHA program. Since its implementation, however, HAMP's participation rates have remained low relative to initial estimates of the number of homeowners the program was expected to assist. Treasury has made several efforts to increase participation rates, such as extending the deadline for participation in the MHA programs and working with a network of counseling agencies to increase borrower awareness of the programs. Some advocates have recently expressed concern about the extent to which groups that have been traditionally underserved by the mortgage markets have the same access to MHA programs as other borrowers. Specifically, these advocates point to racial and ethnic minorities and individuals with limited English language skills. In four prior reports, we looked at Treasury's design and implementation of HAMP and other MHA programs and made several recommendations to improve Treasury's oversight of the programs and the programs' effectiveness.[Footnote 4] This 60-day report examines (1) the status of MHA and steps Treasury is taking to increase participation in the program, (2) Treasury's oversight of the MHA- related fair lending internal controls of participating servicers, and (3) Treasury's and MHA servicers' policies and practices for ensuring that borrowers with limited English proficiency (LEP) have equal access to the program. To address these objectives, we reviewed MHA program documentation that Treasury issued, including supplemental directives for the recent MHA program changes, and interviewed officials from Treasury and NeighborWorks America (NeighborWorks).[Footnote 5] We spoke with the 5 largest MHA servicers, which, as of October 2013, collectively represented about 77 percent of the total trial HAMP loan modifications approved. We also spoke with various mortgage industry participants, including associations representing housing counselors and legal services attorneys. To understand Treasury's oversight of the MHA-related fair lending controls of servicers, we spoke with Treasury officials and reviewed MHA program policies. We also spoke with the 5 largest MHA servicers mentioned above and those federal agencies authorized to examine servicers for compliance with federal fair lending laws. We analyzed loan-level data from Treasury's HAMP database that included information reported by servicers on borrowers evaluated for HAMP participation through April 17, 2013. This analysis allowed us to identify the outcomes of borrowers at four large servicers at various stages of the HAMP process and to compare these outcomes across borrower groups, including populations protected by certain fair lending requirements. To understand how Treasury and MHA servicers work to ensure access to MHA for LEP borrowers, we obtained information from Treasury officials and spoke with the five servicers we contacted about their policies and practices. We also examined Treasury's draft LEP plan and MHA guidelines. For additional information on our scope and methodology, see appendixes I and II. We conducted this performance audit from February 2013 through February 2014 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on the audit objectives. Background: Making Home Affordable Programs: Treasury's Office of Homeownership Preservation within the Office of Financial Stability (OFS), is responsible for overseeing the TARP- funded programs that are intended to help prevent avoidable foreclosures and preserve homeownership. MHA is the primary TARP initiative for addressing these issues. Treasury allocated $29.9 billion in TARP funds to MHA, which consists of several programs designed to help struggling homeowners prevent avoidable foreclosures. [Footnote 6] * HAMP first-lien modifications. These loan modifications are available to qualified borrowers who took out loans on or before January 1, 2009. Only single-family properties (one to four units) with mortgages no greater than $729,750 for a one-unit property are eligible. HAMP uses a standardized net present value (NPV) model to compare expected cash flows from a modified loan to the same loan with no modification, using certain assumptions.[Footnote 7] Treasury also shares some of the costs of modifying mortgages with mortgage holders/investors and provides incentives of up to $1,600 to servicers for completing modifications.[Footnote 8] The Home Price Decline Protection Incentive provides investors with additional incentives to modify loans on properties located in areas where home prices have recently declined and where investors are concerned that price declines may persist. The original HAMP first-lien modification structure, or HAMP Tier 1, is generally available to qualified borrowers who occupy their properties as their primary residence and whose first-lien mortgage payment is more than 31 percent of their monthly gross income, calculated using the front-end debt-to-income (DTI) ratio.[Footnote 9] In June 2012, Treasury made a second type of first-lien loan modification available under HAMP. HAMP Tier 2 is available for either owner-occupied properties or rental properties, and borrowers' monthly mortgage payments prior to modification may be less than 31 percent DTI. Mortgages secured by owner-occupied properties must be in imminent default or be delinquent by two or more payments to be considered for either HAMP Tier 1 or HAMP Tier 2. For mortgages secured by rental properties, only those that are two or more payments delinquent are eligible for HAMP Tier 2. Under both HAMP Tier 1 and Tier 2, borrowers must demonstrate their ability to pay the modified amount by successfully completing a trial period of at least 3 months (or longer if necessary) before a loan is permanently modified and any government payments are made. * The Second Lien Modification Program (2MP). Under 2MP, when a borrower's first lien is modified under HAMP and the servicer of the second lien is a 2MP participant, that servicer must offer a modification and/or full or partial extinguishment of the second lien. A servicer need not service the related first lien in order to participate in 2MP. Treasury provides incentive payments to second- lien mortgage holders in the form of a percentage of each dollar of principal reduction on the second lien. Treasury has doubled the incentive payments offered to second-lien mortgage holders for 2MP permanent modifications that include principal reduction and have an effective date on or after June 1, 2012. * Principal Reduction Alternative (PRA Program). In October 2010, PRA took effect as a component of HAMP to give servicers more flexibility in offering relief to borrowers whose homes were worth significantly less than their mortgage balance. Under PRA, Treasury provides mortgage holders/investors with incentive payments in the form of a percentage of each dollar of principal reduction. Treasury has tripled the PRA incentive amounts offered to mortgage holders/investors for permanent modifications with trial periods effective on or after March 1, 2012. Servicers of nonenterprise loans must evaluate the benefit of principal reduction for mortgages with a loan-to-value (LTV) ratio that is greater than 115 percent when evaluating a homeowner for a HAMP first-lien modification.[Footnote 10] Servicers must adopt and follow PRA policies that treat all similarly situated loans in a consistent manner, but are not required to offer principal reductions, even when NPV calculations show that the expected value of the loan's cash flows would be higher with a principal reduction than without. When servicers include principal reductions in modifications under PRA, the principal reduction amount is initially treated as non- interest-bearing principal forbearance. If the borrower is in good standing on the first, second, and third anniversaries of the effective date of the modification's trial period, one-third of the principal reduction amount is forgiven on each anniversary. * Home Affordable Foreclosure Alternatives (HAFA) Program. Under this program, servicers offer foreclosure alternatives (short sales and deeds-in-lieu) to borrowers who meet the eligibility requirement for HAMP and cannot be approved for a HAMP trial modification, do not successfully complete a HAMP trial modification, default on a modification (miss two or more consecutive payments), or request a short sale or deed-in-lieu.[Footnote 11] The program provides incentive payments to investors, servicers, and borrowers for completing these foreclosure alternatives. * FHA and the Department of Agriculture's Rural Housing Service (RHS) modification programs. These programs are similar to HAMP Tier 1 and cover FHA-insured and RHS-guaranteed first-lien mortgage loans. If a modified FHA-insured or RHS-guaranteed mortgage loan meets Treasury's eligibility criteria, the borrower and servicer can receive TARP- funded incentive payments from Treasury. * Home Affordable Unemployment Program. This program offers assistance to borrowers who are suffering financial hardship due to unemployment. Borrowers can receive a 12-month forbearance period during which monthly mortgage payments are reduced or suspended. Servicers can extend the forbearance period at their discretion if the borrower is still unemployed. Borrowers should be considered for a HAMP loan modification following reemployment or expiration of the forbearance period or a foreclosure alternative, such as the HAFA program. No TARP funds are provided to servicers under this program. In 2009, Treasury entered into agreements with Fannie Mae and Freddie Mac to act as its financial agents for MHA. Fannie Mae serves as the MHA program administrator and is responsible for developing and administering program operations, including registering, executing participation agreements with, and collecting data from servicers and providing ongoing servicer training and support. Freddie Mac serves as Treasury's compliance agent and has designated an independent division, Making Home Affordable Compliance (MHA-C), which is responsible for assessing servicers' compliance with program guidelines, including conducting onsite and remote servicer loan file reviews and audits. Fair Lending: The Equal Credit Opportunity Act (ECOA) and the Fair Housing Act (collectively, fair lending laws) prohibit discrimination in making credit decisions.[Footnote 12] Specifically, ECOA prohibits creditors from discriminating against credit applicants on the basis of race, color, religion, national origin, sex, marital status, or age, or because an applicant receives income from a public assistance program or has in good faith exercised any right under the Consumer Credit Protection Act.[Footnote 13] The Fair Housing Act prohibits discrimination in connection with real estate-related transactions by direct providers of housing, as well as other entities whose discriminatory practices, among other things, make housing unavailable to persons because of race or color, religion, sex, national origin, familial status, or disability. Under one or both of the fair lending laws, a lender may not, because of a prohibited basis: * fail to provide information or services, or provide different information or services regarding any aspect of the lending process, including credit availability, application procedures, or lending standards; * discourage or selectively encourage applicants with respect to inquiries about or applications for credit; * refuse to extend credit or use different standards in determining whether to extend credit; * vary the terms of credit offered, including the amount, interest rate, duration, or type of loan; * use different standards to evaluate collateral; * treat a borrower differently in servicing a loan or invoking default remedies; * use different standards for pooling or packaging a loan in the secondary market or for purchasing loans; * use different standards in collecting indebtedness; or: * use different standards in modifying existing loans. Responsibility for federal oversight and enforcement of the fair lending laws is shared among eight agencies: the Department of Housing and Urban Development (HUD), Department of Justice (DOJ), the Federal Trade Commission (FTC), the Bureau of Consumer Financial Protection (CFPB), and the four prudential regulators, which supervise depository institutions. The four prudential regulators are the Federal Deposit Insurance Corporation (FDIC), the Board of Governors of the Federal Reserve System (Federal Reserve), the Office of the Comptroller of the Currency (OCC), and the National Credit Union Administration. Under the Fair Housing Act, HUD investigates all complaints alleging discrimination and may initiate investigations and file administrative complaints against any entity involved in residential real-estate related transactions, including independent mortgage lenders or any other lender, such as depository institutions that HUD believes may have violated the act.[Footnote 14] DOJ, which has enforcement authority for both ECOA and the Fair Housing Act, may initiate investigations of any creditor--whether a depository or nondepository lender--under its own independent authority or based on referrals from other agencies. CFPB has supervisory and primary enforcement authority under ECOA over mortgage servicers, all insured depository institutions with assets greater than $10 billion and their affiliates, and certain nondepository lenders (including independent mortgage originators).[Footnote 15] The four prudential regulators generally have ECOA oversight responsibilities for insured depository institutions with assets of $10 billion or less, as well as certain subsidiaries and affiliates of those institutions. Each of the prudential regulators has authority to examine the institutions it supervises for compliance with the Fair Housing Act.[Footnote 16] The prudential regulators conduct examinations of institutions they oversee to assess their compliance with fair lending laws and regulations. Under ECOA, prudential regulators and CFPB are required to refer lenders to DOJ if there is reason to believe that a lender has engaged in a pattern or practice of discouraging or denying applications for credit in violation of ECOA.[Footnote 17] A prudential regulator that identifies a possible ECOA violation, that may also be a Fair Housing Act violation, must notify HUD about the potential violation if the regulator does not refer the matter to DOJ. [Footnote 18] Under the Fair Housing Act, HUD is required to transmit information to DOJ regarding any complaint in which there is reason to believe that a pattern or practice of violations has occurred or that a group of persons has been denied rights under the Fair Housing Act and that the matter raises an issue of general public importance. [Footnote 19] Limited English Language Access: Title VI of the Civil Rights Act of 1964 provides that no person shall be subjected to discrimination on the basis of race, color, or national origin under any program or activity that receives federal financial assistance. In certain circumstances, failure to ensure that LEP persons can effectively participate in and benefit from federally assisted programs and activities may violate the prohibition under Title VI against national origin discrimination. Executive Order 13166, issued in 2000, addresses the application of Title VI's prohibition on national origin discrimination in connection with federally conducted and assisted programs and activities. The Executive Order requires that federal agencies examine the services they provide and develop and implement systems by which individuals with limited proficiency in English can access agency programs and services.[Footnote 20] While the Executive Order does not prescribe specific approaches to language access services, it does require federal agencies to prepare plans (referred to as LEP plans) outlining the steps they will take to ensure that eligible LEP persons can meaningfully access their programs and activities. With respect to recipients of federal financial assistance, DOJ issued guidance which states that recipients should provide LEP individuals with "meaningful access" to their programs, services, and activities. [Footnote 21] Rather than express uniform rules of compliance, the guidance suggests that agencies assess whether recipients have provided meaningful access through consideration of factors such as the number or proportion of LEP persons eligible to be served or likely to be encountered by the program or recipient; the frequency with which LEP persons come in contact with the program; the nature and importance to people's lives of the program, activity, or service provided by the recipient; and the resources available to the recipient and the costs of language access. The intent of DOJ's LEP guidance is to ensure meaningful access by LEP persons to critical programs, services, and activities. HAMP Participation and Expenditures Have Remained below Expectations: HAMP participation levels--number of new permanent modifications added on a monthly basis--have made it uncertain whether Treasury will disburse the nearly $30 billion it has obligated to help borrowers avoid foreclosure. Treasury has taken several steps to increase participation, such as extending the program deadline, expanding program eligibility criteria through HAMP Tier 2, and providing funding to counseling agencies to assist homeowners with completion and submission of application packages (intake project). Since the implementation of HAMP Tier 2 in June 2012, the number of new HAMP modifications started each month has been relatively stable through November 2013. Treasury has recently begun to assess the performance of counseling agencies participating in the intake project, which has been extended to provide funding for packages submitted through September 2014. HAMP Participation Has Slowed, Leaving Obligated Funds Unspent: Treasury has reported that about 1.3 million borrowers have received permanent loan modifications under HAMP as of November 30, 2013. However, as shown in figure 1, participation in HAMP, as measured by trial and permanent modifications started each month, peaked in early 2010, generally declined in 2011, and has remained relatively steady through November 2013. Treasury made several changes to HAMP to address barriers to borrower participation, such as extending the application deadline for new HAMP modifications to December 2015 and expanding eligibility criteria for program participation. In particular, Treasury expanded the pool of homeowners potentially eligible to be assisted through the launch of HAMP Tier 2 in June 2012.[Footnote 22] HAMP Tier 2 expanded eligibility to various borrowers previously ineligible for HAMP, including borrowers with mortgages secured by "rental property" and borrowers with a wider range of debt-to-income ratios.[Footnote 23] Figure 1: Monthly HAMP and Combined Tier 1 and Tier 2 Activity Levels through November 2013: [Refer to PDF for image: multiple line graph] Number of borrowers: 2010: January; Trial modifications started: 95,000; Permanent modifications started: 50,000. February; Trial modifications started: 88,000; Permanent modifications started: 53,000. March; Trial modifications started: 70,000; Permanent modifications started: 61,000. April; Trial modifications started: 49,000; Permanent modifications started: 68,000. May; Trial modifications started: 27,000; Permanent modifications started: 48,000. June; Trial modifications started: 22,000; Permanent modifications started: 51,000. July; Trial modifications started: 24,000; Permanent modifications started: 37,000. August; Trial modifications started: 24,000; Permanent modifications started: 33,000. September; Trial modifications started: 31,000; Permanent modifications started: 28,000. October; Trial modifications started: 30,000; Permanent modifications started: 24,000. November; Trial modifications started: 31,000; Permanent modifications started: 30,000. December; Trial modifications started: 35,000; Permanent modifications started: 30,000. 2011: January; Trial modifications started: 28,000; Permanent modifications started: 28,000. February; Trial modifications started: 35,000; Permanent modifications started: 26,000. March; Trial modifications started: 35,000; Permanent modifications started: 36,000. April; Trial modifications started: 33,000; Permanent modifications started: 29,000. May; Trial modifications started: 27,000; Permanent modifications started: 32,000. June; Trial modifications started: 20,000; Permanent modifications started: 32,000. July; Trial modifications started: 23,000; Permanent modifications started: 28,000. August; Trial modifications started: 22,000; Permanent modifications started: 26,000. September; Trial modifications started: 18,000; Permanent modifications started: 40,000. October; Trial modifications started: 18,000; Permanent modifications started: 26,000. November; Trial modifications started: 18,000; Permanent modifications started: 27,000. December; Trial modifications started: 18,000; Permanent modifications started: 23,000. 2012: January; Trial modifications started: 16,000; Permanent modifications started: 18,000. February; Trial modifications started: 17,000; Permanent modifications started: 23,000. March; Trial modifications started: 17,000; Permanent modifications started: 20,000. April; Trial modifications started: 17,000; Permanent modifications started: 15,000. May; Trial modifications started: 19,000; Permanent modifications started: 17,000. June; Trial modifications started: 18,000; Permanent modifications started: 17,000. July; Trial modifications started: 15,000; Permanent modifications started: 17,000. August; Trial modifications started: 17,000; Permanent modifications started: 17,000. September; Trial modifications started: 15,000; Permanent modifications started: 14,000. October; Trial modifications started: 17,000; Permanent modifications started: 16,000. HAMP Tier 1 and Tier 2 combined: November; Trial modifications started: 17,000; Permanent modifications started: 15,000. December; Trial modifications started: 14,000; Permanent modifications started: 14,000. 2013: January; Trial modifications started: 13,000; Permanent modifications started: 15,000. February; Trial modifications started: 13,000; Permanent modifications started: 16,000. March; Trial modifications started: 14,000; Permanent modifications started: 12,000. April; Trial modifications started: 17,000; Permanent modifications started: 12,000. May; Trial modifications started: 18,000; Permanent modifications started: 15,000. June; Trial modifications started: 20,000; Permanent modifications started: 17,000. July; Trial modifications started: 16,000; Permanent modifications started: 14,000. August; Trial modifications started: 16,000; Permanent modifications started: 19,000. September; Trial modifications started: 14,000; Permanent modifications started: 13,000. October; Trial modifications started: 13,000; Permanent modifications started: 16,000. November; Trial modifications started: 12,000; Permanent modifications started: 13,000. Source: GAO analysis based on monthly Making Home Affordable Program Performance Reports. [End of figure] HAMP Tier 2 appears to have helped stem the decline in the number of new HAMP modifications added on a monthly basis. More than one-fourth of the permanent modifications started in November were Tier 2 modifications (3,460 out of 12,996 modifications). Through November 2013, a cumulative total of 29,134 borrowers had entered into a HAMP Tier 2 permanent modification representing about 11 percent of all permanent modifications started since the implementation of Tier 2 in June 2012. Tier 2 trial modifications represented about 18 percent of all trial modifications started since June 2012. When HAMP was first announced in February 2009, Treasury had developed an internal projection that up to 3 million to 4 million borrowers who were at risk of default and foreclosure could be offered a loan modification under HAMP. However, we subsequently reported that because of the unsettled dynamics of the mortgage market and overall economic conditions, actual outcomes may well be different from the projection.[Footnote 24] Further, Treasury stated to us that the number of potentially eligible borrowers has shrunk steadily since the beginning of the program, as has the number of delinquent borrowers across the mortgage industry, generally. Extending the deadline for HAMP applications and expanding program eligibility may provide more borrowers the opportunity to participate in the programs. However, because the number of borrowers that have received permanent modifications as of November 30, 2013 (1.3 million) is well below Treasury's initial estimate of 3 million to 4 million and the pool of estimated HAMP eligible borrowers is declining, it is unclear whether Treasury will disburse all the funds it has obligated to MHA. As of November 30, 2013, $7.0 billion (23 percent) of the 29.9 billion set aside for MHA had been disbursed. According to Treasury, if all active modifications made as of November 30, 2013, in association with MHA were to remain current and receive incentives for the full 5 years, $13.6 billion in incentives will ultimately be disbursed. However, this estimate does not take into account modifications that borrowers enter into after November 2013 through the program's deadline of December 31, 2015 nor does it consider the impact of redefaults on projected outlays. The Congressional Budget Office (CBO) has estimated that Treasury will ultimately disburse much less than the $29.9 billion currently obligated for MHA. In its May 2013 TARP update report, CBO estimated that only $16 billion, (about 53 percent) for all of the TARP-funded housing programs (MHA, HHF, and FHA Short Refinance Program), would likely be disbursed over those programs' lifetimes.[Footnote 25] CBO stated to us that about $11 billion of their estimate was attributable to HAMP. CBO's estimate assumed that participation rates would continue at the current pace and that redefault rates on modifications would remain consistent regardless of the year in which the modification was started. However, CBO's May 2013 estimate did not consider the impact of the 2-year extension of MHA through 2015. Treasury officials told us that because of the uncertainty in uptake due to the constantly changing economic environment, potential program changes, and in order to be conservative in their forecasts, they continue to assume that the entire $29.87 billion currently allocated for MHA will be used. Impact of Treasury's MHA Outreach and Borrower Intake Project on Program Participation is Unclear: In May 2013, Treasury launched its MHA Outreach and Borrower Intake Project in "an effort to ensure that every potential borrower has a chance to be considered for foreclosure prevention assistance under MHA." Treasury entered into an agreement with NeighborWorks to launch a nationwide effort with housing counselors to increase the number of homeowners that successfully request assistance under MHA. The project's goal is to make more homeowners aware of the full range of options under MHA and to help eligible homeowners successfully complete an MHA assistance application for servicers to consider. Originally the project was scheduled to end in December of 2013, but Treasury extended the project through September of 2014. As a result, it is too early to determine the project's impact on HAMP participation. The project pays housing counseling agencies to conduct borrower outreach, assess borrowers for eligibility, help eligible homeowners prepare complete application packages, and deliver those packages to MHA servicers. The applications are to be submitted through the Hope LoanPort, an Internet-based document delivery portal that allows servicers to be notified when an application arrives. The Hope LoanPort uses an intake code to indicate whether the counseling agency is eligible for funding provided by Treasury under the project. Participating housing counseling agencies receive a document preparation and submission fee of $450 for each completed initial application package submitted to and accepted by a MHA servicer, even if the borrower does not receive a modification. Additionally, participating agencies receive funding to cover outreach and administrative costs. Initially, Treasury allocated $18.3 million in TARP funds for the MHA Outreach and Borrower Intake Project. Of this allocation, $12.6 million was to cover the costs of the document preparation and submission fee for 20,000 applications, outreach and certain administrative costs incurred by counseling agencies, and supplemental outreach funds to target specific populations that require specialized services.[Footnote 26] Treasury allocated the remaining $5.7 million to NeighborWorks' for outreach and administrative costs associated with the project.[Footnote 27] However, according to NeighborWorks, only two-thirds of the housing counseling agencies eligible to participate in the project have decided to participate and received an application package allocation, resulting in a total of 92 agencies participating with a production goal of 15,318 application packages to be submitted on behalf of borrowers.[Footnote 28] As a result, nearly 5,000 packages and $2.9 million remain unallocated to counseling agencies. As shown in table 1, NeighborWorks ultimately allocated about $9.2 million in funding to the 92 participating agencies to cover the cost of document preparation fees, outreach, and administration. Table 1: MHA Outreach and Borrower Intake Project Package Allocation to Housing Counseling Agencies: Participating agency type: State housing finance agencies (18); Number of packages allocated: 4,502; Initial funding allocation for participating agencies: $2,700,721; Supplemental outreach funds: $157,730; Total: $2,858,451. Participating agency type: HUD counseling intermediaries (14); Number of packages allocated: 8,504; Initial funding allocation for participating agencies: $5,061,111; Supplemental outreach funds: $225,525; Total: $5,286,636. Participating agency type: Community-based organizations (60); Number of packages allocated: 2,312; Initial funding allocation for participating agencies: $1,430,425; Supplemental outreach funds: $114,075; Total: $1,544,500. Participating agency type: Total allocations; Number of packages allocated: 15,318; Initial funding allocation for participating agencies: $9,192,257; Supplemental outreach funds: $497,330; Total: $9,689,587. Participating agency type: Unallocated packages and funding; Number of packages allocated: 4,682; Initial funding allocation for participating agencies: $2,909,743; Supplemental outreach funds: $2,670; Total: $2,912,413. Participating agency type: Total initial allocations; Number of packages allocated: 20,000; Initial funding allocation for participating agencies: $12,102,000; Supplemental outreach funds: $500,000; Total: $12,602,000. Source: Treasury. [End of table] The MHA Outreach and Borrower Intake Project became effective in May 2013. As of December 31, 2013, counseling agencies had submitted 2,253 initial packages that had been accepted as complete by servicers under the program with another 878 initial packages in the process of being reviewed by the servicers. Document preparation fees associated with these packages totaled about $1.0 million. As of December 2013, NeighborWorks reported to us that they had disbursed over $1.9 million to housing counseling agencies for outreach and NeighborWorks had expended about $779,121 in administration costs associated with the project. To assist agencies in meeting stated production goals, NeighborWorks generates a semimonthly Production Dashboard report for each housing counseling agency that is shared with the respective agency.[Footnote 29] The Production Dashboard summarizes historical information, such as how many initial packages have been accepted by servicers as complete, and the percentage of the agency's cumulative goal that has been reached. The Production Dashboard also includes intermediate goals and projections, such as how many initial packages the agency must submit each month to reach its cumulative goal, and how many initial packages are projected to be delivered by the end of the performance period (based on the agency's average submission rate). Treasury stated that they periodically review the Production Dashboard for individual agencies, as needed. An agency that does not meet its production goals would receive less compensation because document preparation fees are only paid for complete initial packages accepted by servicers. In addition, NeighborWorks may re-allocate funds from an underperforming agency to another agency if it reaches its allocation goal. However, Treasury officials noted that the funding for 20,000 applications had not been fully allocated and, thus, NeighborWorks would first allocate unallocated funds to any agency needing a higher allocation before reducing the allocation of an underperforming agency. In addition to the Production Dashboard report, NeighborWorks provides Treasury with quarterly reports describing what housing counseling agencies have characterized as successes and challenges to project implementation. For example, in September 2013, NeighborWorks reported that 55 counseling agencies identified internal programmatic changes such as streamlining processes, specialized staff, and direct engagement with borrowers as factors associated with success. Counseling agencies also reported challenges with servicers that did not subscribe to the Hope LoanPort, unresponsive servicers, and borrowers that did not engage with counselors. Fully understanding and analyzing the nature of these successes and challenges could be useful to Treasury in working with NeighborWorks to improve the performance of the project. For example, Treasury stated to us that the majority of large MHA servicers subscribe to the Hope LoanPort, representing over 80 percent of HAMP activity. Treasury stated to us that servicers may have chosen not to subscribe to the Hope LoanPort due to the related subscription costs, and that the servicers that did not subscribe were generally either smaller servicers or those with their own document collection system. Additionally, Treasury noted that its compliance agent has begun assessing servicers' processes associated with the MHA Outreach and Intake Project and has noted instances where certain servicers could enhance their design and execution of controls, but, the compliance agent's loan level testing indicated that in most instances the loans were processed accurately and timely. By extending the project from December 2013 through September 2014, it is more likely that Treasury will reach its goal of 20,000 HAMP application packages completed through the project. However, it is not clear if the project is in fact increasing access to the program given the challenge of determining whether a borrower would have applied successfully in the absence of the project. Treasury Has Not Assessed Servicers' Internal Controls or Borrower Data Related to Fair Lending: Treasury requires MHA servicers to develop an internal control program to monitor compliance with fair lending laws. However, Treasury has not evaluated the extent to which servicers have effective internal control programs for assessing compliance with fair lending laws. Additionally, Treasury requires servicers to collect and report data on the race, ethnicity, and gender of MHA applicants, but has not analyzed the data for potential differences in outcomes of groups protected under the laws. Our analysis of HAMP loan-level data, which focused on four large MHA servicers, identified some statistically significant differences within these servicers' portfolios for certain protected groups in denials and cancellations of trial modifications and in the potential for redefault of permanent modifications, which might indicate a need for further review. Treasury Requires Servicers to Develop and Maintain Systems to Monitor Compliance with Fair Lending Laws: The MHA Servicer Participation Agreement and MHA Handbook require that servicers have an internal control program to monitor compliance with relevant consumer protection laws, including ECOA and the Fair Housing Act, and that the servicers review the effectiveness of their internal control program quarterly. The internal control program must document the control objectives for MHA activities, the associated control techniques, and mechanisms for testing and validating the controls. Servicers are also required to provide Treasury's compliance agent with access to all internal control reviews related to MHA programs performed by the servicer and its independent auditing firm and to provide a copy of the reviews to the MHA program administrator. Although Treasury requires MHA servicers to certify that they have developed and implemented an internal control program to monitor compliance with applicable consumer protection and fair lending laws, Treasury has not monitored servicers to determine whether they have developed such internal control programs. Specifically, Treasury officials told us that it has not required its compliance agent to obtain information from servicers on such programs. The five MHA servicers we spoke with told us that they had not shared with Treasury details on their internal control programs for monitoring compliance with fair lending laws. However, four of the servicers said that they regularly shared the details of these programs, as well as the results of fair lending analyses, with their federal financial regulators. Treasury officials explained that Treasury does not examine servicer compliance with fair lending laws because other federal agencies-- CFPB, DOJ, FTC, HUD, and the banking regulators--have the sole responsibility for enforcement and supervision of federal fair lending laws. Therefore, only those agencies, and not Treasury, are charged with the responsibility for determining whether a servicer (subject to the jurisdiction of the appropriate agency) complies with the federal fair lending laws. According to representatives of the prudential regulators, their fair lending reviews have a broader overall focus, which include examining the servicers' overall servicing and loss mitigation activities.[Footnote 30] They added that, while the reviews may not specifically focus on MHA activities, HAMP modifications may be included in the loan portfolios of the MHA servicers examined. Officials from two prudential regulators said that their examinations of servicing portfolios had resulted in supervisory guidance to a few of the larger MHA servicers related to (1) potential disparities between certain fair lending protected classes and their comparison populations, (2) communication issues with non-English speaking borrowers, and (3) handling of loss mitigation and loan modification complaints. Additionally, one regulator, on behalf of the Financial Fraud Enforcement Task Force's (FFETF) Non-Discrimination Working Group, conducted exploratory analysis to characterize outcomes of the HAMP program and identify fair lending risks.[Footnote 31] According to officials from this regulator, the aggregate results of the exploratory analysis were shared with Treasury and other members of the Non-Discrimination Working Group in January 2012, and no fair lending issues of note were identified. Additionally, officials said that this regulator also shared the supervisory guidance discussed above and summaries of its fair lending reviews, which included statistical analysis of MHA servicers under its jurisdiction, with the working group. Officials from the prudential regulators noted that they consider complaints from consumers alleging discriminatory practices in their examinations of regulated banking institutions. According to the prudential regulators, results of their fair lending examinations are considered confidential supervisory information and are sensitive and privileged. The regulators explained that because of the nature of the information, they would not have shared the details of examination results with Treasury. Further, these regulators told us that they had not identified fair lending violations related to the MHA program specifically. Treasury officials told us that while they have not specifically examined servicers' controls for ensuring compliance with fair lending laws, the compliance agent did examine servicers' internal controls related to other HAMP requirements, such as soliciting borrowers who are 60 days delinquent and performing "second look" loan reviews which focused on determining whether HAMP denials were appropriate. Additionally, Treasury officials noted that the processes servicers use to solicit borrowers and determine the eligibility and terms of a modification were highly structured due to MHA requirements. These processes limit servicer discretion with respect to implementing the MHA requirements, and as a result, outcomes in HAMP modifications are less likely to result in fair lending compliance issues, according to Treasury officials. Despite the structured nature of HAMP, we have previously found instances where servicers varied their application within HAMP guidelines. For example, in 2010 we reported that servicers have inconsistent practices for evaluating borrowers for imminent default because Treasury has not provided specific guidance on how to evaluate nonenterprise borrowers for imminent default. [Footnote 32] Additionally, Treasury does not require servicers to apply principal reduction in connection with modifications; instead servicers are required to establish written policies detailing when principal reduction will be offered.[Footnote 33] While these policies must treat similarly situated borrowers in a consistent manner, there may be variations across servicers in the use of principal reduction and in some cases servicers may reasonably refuse to reduce principal. Also, servicers, and their employees, may make errors in applying HAMP policies to modifications. For example, in 2010 we reported that 5 of the 10 servicers we contacted reported at least a 20 percent error rate for income calculations. We noted that without accurate income calculations, which are key in determining borrowers' DTI, similarly situated borrowers applying for HAMP may be inequitably evaluated for the program and may be inappropriately deemed eligible or ineligible for HAMP.[Footnote 34] Treasury also assesses servicers on their income calculations and tracks the percentage of loans for which MHA- C's income calculation differs from the servicer's on a quarterly basis. In their July 2013 assessment results, Treasury noted an average income error rate of less than 2 percent, down from an average of about 7.5 percent in July 2011. Although the prudential regulators have not identified any fair lending violations of MHA servicers, they did share some fair lending- related concerns of some large MHA servicers with Treasury. Furthermore, opportunity for variations and errors within and across servicers can impact borrowers. By evaluating the extent to which servicers have developed and maintained internal controls to monitor compliance with fair lending laws, Treasury could gain additional assurance that servicers are implementing the MHA program in compliance with fair lending laws. Treasury Collects but Has Not Analyzed MHA Data Related to Fair Lending: Treasury requires that MHA servicers collect and report data on the race, ethnicity, and gender of borrowers seeking HAMP modifications or entering into a HAMP trial or permanent modification. While Treasury has not yet examined the data to understand if there are any potential fair lending risks in servicers' implementation of HAMP, and thus if servicers' fair lending internal controls are sufficient, Treasury does provide those data to federal agencies charged with oversight and enforcement of fair lending laws. As the federal agency with authority for enforcement of the Fair Housing Act, HUD requested that Treasury require servicers to collect this information, and servicers began collecting and reporting it beginning on December 1, 2009.[Footnote 35] Treasury shares loan-level HAMP data, including information on the servicer and the borrowers' race, ethnicity, and gender, with the federal agencies with fair lending oversight and enforcement authority.[Footnote 36] Treasury also makes a more limited public file available to the general public that excludes, among other things, information identifying the servicer and personal identifying information about the borrower (name, address, etc.). On first releasing the public file containing loan-level data in January 2011, Treasury stated that it intended to engage one or more independent, third-party research firms to conduct a more detailed analysis of fair lending in MHA and that it would make the results of this analysis available to the public. In March 2013, Treasury entered into an interagency agreement with HUD to engage a third-party contractor to conduct a fair lending analysis of HAMP loan modifications. As of September 30, 2013, HUD secured a contractor to conduct the analysis. Our analysis of Treasury's HAMP data through April 17, 2013, suggested that there may be some issues that warrant a closer look at servicers' fair lending internal control systems by Treasury and the pertinent fair lending regulatory agency.[Footnote 37] We examined the rate of denial or cancellation of HAMP modifications, and the rate of redefault of permanent HAMP modifications experienced by selected population groups and compared them to the same rates for their comparison populations at various stages of the HAMP process. We primarily focused on the outcomes for certain protected groups under federal fair lending laws plus low-income groups and groups in neighborhoods that consisted primarily of minority populations (substantially minority); we refer to these groups collectively as "selected populations."[Footnote 38] We used a multivariate econometric analysis to control for several observable characteristics of the borrower, servicer, loan, and property, allowing us to appropriately estimate the outcomes these populations experienced. [Footnote 39] In focusing our analysis on four large MHA servicers, we found some statistically significant differences in the outcomes experienced by our selected populations compared to their comparison population. For example, we found that for all four servicers non- Hispanic African-Americans had a statistically significant higher trial modification denial rate compared to non-Hispanic whites due to DTIs being less than 31 percent. When examining denials of trial modifications because borrowers had not provided complete information to the servicer, denial rates were significantly higher for Hispanics than for the comparison population of non-Hispanic whites for three of the four large servicers we analyzed. We also found that for all of the servicers we analyzed, non-Hispanic African-Americans had a statistically higher rate of redefault than non-Hispanic whites, regardless of whether or not the servicer applied capitalization, principal forbearance, or principal forgiveness to the loan modification, holding other key factors constant. For additional findings from our analysis, see appendix II. We are unable to determine from the available HAMP data whether the statistically significant differences between the selected populations and their comparison populations identified in our analysis were the result of servicer discretion, servicer errors in the application or interpretation of HAMP guidelines or servicing protocols, differences among servicers' policies, or the unintended consequences of HAMP guidelines or program design. Additional analysis is needed to determine the reasons for the differences and the extent to which servicer implementation of HAMP guidelines could be a potential cause as well as other potential causes for the differences in outcomes. As noted in appendix II, there are some limitations of our analysis. For instance, we could not control for all potential factors that affect these outcomes due to the lack of certain data, such as the wealth of the borrowers and their knowledge of the loan modification process. Also, our analysis cannot account for some important factors, such as whether equivalent borrowers in these populations apply to HAMP at different rates or are more or less likely to receive assistance outside of HAMP. Further, race and ethnicity data were not available for 54 percent of borrowers in the early stage of the HAMP process and 43 percent of borrowers in the later stage of HAMP. Although we took appropriate steps to minimize the impact of missing data our results should be interpreted with caution.[Footnote 40] Despite the limitations noted above, statistical differences in outcomes among population groups might suggest potential fair lending concerns that merit further examination. Officials from fair lending regulatory agencies told us that results of econometric analyses of fair lending populations were one of multiple sources of information they review when examining fair lending compliance of banking institutions and servicers. They noted that the existence of a statistical disparity alone would not necessarily result in the finding of a fair lending violation but could be a reason to further investigate an institution. Such analyses could be useful to Treasury as the agency considers whether servicers participating in the HAMP program have sufficient internal controls to assess compliance with fair lending laws. Treasury Has Provided Servicers with Limited Guidance and Oversight on Interacting with Borrowers with Limited English Proficiency: Treasury has taken various actions to increase access to the program for borrowers whose primary language is not English, but only recently has begun to systematically assess access to the MHA program, for these borrowers. Federal agencies, including Treasury, are required by Executive Order 13166, issued in August 2000, to "examine the services it provides and develop and implement a system by which LEP persons can meaningfully access those services." Under MHA, borrowers apply for and obtain mortgage modifications directly from their mortgage servicers. Although Treasury has not specified for servicers how they should meet the needs of LEP persons or assessed their efforts to do so, it has taken steps to provide information and support to LEP borrowers in connection with MHA through various sources and methods. For example, Treasury publishes a website with information about the MHA program [hyperlink, http://www.makinghomeaffordable.gov], which has a mirror Spanish website and critical content pages in Chinese, Vietnamese, Russian, Tagalog, and Korean. Treasury also has published advertisements and public service announcements in Spanish, and conducted outreach to Spanish-speaking media as well. Additionally, the MHA website, along with Treasury's outreach materials, directs interested homeowners to the Homeowners HOPE™ Hotline, which provides over-the-telephone support to LEP borrowers related to MHA programs. As part of the MHA escalations process, Treasury also provides a toll- free call center--MHA Help--where, according to Treasury, LEP borrowers can receive more specialized assistance over the phone. Treasury has also translated the MHA application form and certain outreach materials in other languages such as Spanish, Chinese, Korean, Russian, Vietnamese, and Tagalog. However, it does not require that servicers use the translated materials. The executive order also directs federal agencies to "prepare a plan to improve access to its federally conducted programs and activities by eligible LEP persons." The plans are to include the steps the agency will take to ensure that eligible LEP persons can meaningfully access the agency's programs and activities. In February 2011, the Attorney General issued a memorandum to the heads of federal agencies that renewed the federal government's commitment to language access obligations under the executive order and called on agencies to, among other things, review their programs and activities for language accessibility and submit an updated LEP plan within 6 months after the date of the memorandum.[Footnote 41] Treasury issued its last LEP plan in 2000 and, as such, it did not cover newer programs such as MHA that began in early 2009. As of November 2013, Treasury was working on finalizing an updated agency-wide LEP plan, which would address newer programs and activities, such as the MHA programs. The draft plan indicated that Treasury intended to publish the plan in the Federal Register and on the Treasury website for public comment. The draft plan included information related to Treasury's assessment of the language needs for the MHA programs. In addition, it described Treasury's current and planned steps to assist LEP borrowers in accessing the information and support provided by Treasury in connection with the MHA programs. Additionally, Treasury's Office of Financial Stability (OFS) has developed draft guidelines to assist OFS staff in providing access to LEP persons in connection with the MHA activities described above. Treasury's draft LEP plans and guidance are directed towards their own policies and procedures, and do not address how MHA servicers should interact with LEP borrowers. As noted in a LEP planning tool issued by the Department of Justice, it is important for agencies to examine the manner in which the agency interacts with the public and/or LEP individuals.[Footnote 42] In the case of MHA programs, LEP borrowers access the benefits of the MHA programs through MHA servicers that play a key role in reaching out to borrowers to inform them of their potential eligibility and assist them through the application process. In May 2011, Treasury required that servicers develop and implement a single point of contact policy and that the policy "include a provision for providing effective relationship management to borrowers whose primary language is a language other than English."[Footnote 43] Representatives of the five servicers we contacted said that Treasury had not provided any guidance on what a policy on effective relationship management should include, which Treasury confirmed. Treasury officials also told us that they had not required their compliance agent to review servicers' implementation of the requirement for effective relationship management for LEP borrowers. According to Treasury officials, concerns about LEP borrowers' lack of access to MHA were only recently raised as an issue by consumer advocates in May 2013. In response to this feedback, Treasury conducted a survey of the LEP-related policies of the 17 largest MHA servicers to better understand how these servicers worked with LEP borrowers. All of the 17 servicers Treasury surveyed reported that they had staff that spoke Spanish, and 15 servicers indicated that they had contracted with a vendor for real-time translation. Representatives of four MHA servicers we contacted confirmed this practice. The remaining servicer told us that they contract with a vendor for all non-English customer communication related to modifications. Additionally, representatives of three servicers told us that their firms had electronic systems that could note in the borrower's file if the borrower's primary language was Spanish. However, the systems contain no similar notation for other languages. Three servicers we spoke with told us that they also referred borrowers to the Homeowners HOPE™ Hotline to find a housing counseling agency that could assist with languages. Nonetheless, representatives of some advocacy groups we spoke to raised concerns about the sufficiency of the practices followed by servicers in meeting the needs of non-English-speaking borrowers. The advocacy groups represented housing counseling agencies whose counselors worked one-on-one with potential HAMP borrowers and legal services attorneys. These groups were concerned that servicers' current practices of using Spanish-speaking staff or contracting with a language interpretation service were insufficient. For example, one advocacy group said that some servicers used Spanish-speaking customer service agents who might be knowledgeable about banking and mortgages generally, but not about servicing, loss mitigation, or HAMP specifically. Similarly, representatives of three advocacy groups noted that staff from a language interpretation service might not be familiar enough with banking terminology or the details of HAMP to provide quality interpretation services. Another group pointed out the importance of translated documents and noted that it would be more beneficial for borrowers to have important documents, such as the trial modification offer letter, translated into their preferred language so that they could refer to it when needed. In fact, in a 2013 national survey conducted by the National Housing Resource Center and a similar survey conducted by a California-based research group, nearly half of the 296 housing counselors who responded said their LEP clients who were seeking mortgage servicing assistance "never" received translated foreclosure-related documents. [Footnote 44] Additionally, in both surveys, over 60 percent of the housing counselors said that their LEP clients were "never" or only "sometimes" able to speak to their servicer in their native language or through a translator provided by the servicer, while the rest said their clients were "always" or "often" able to do so. Furthermore, in the national survey, nearly half of the survey respondents said their LEP clients "always," "often," or "sometimes" received worse loss mitigation outcomes than their English-proficient clients, while the other half said their clients "never" received worse outcomes. Ultimately, the lack of clear guidance and expectations of servicers on what constitutes effective relationship management in serving LEP borrowers can potentially affect the effectiveness of servicers' ability to work with such borrowers and result in unequal access to the program by these borrowers. Treasury officials noted that MHA- eligible loans represent a small portion of participating servicers' overall servicing activity, and thus Treasury is cautious in imposing additional requirements on participating servicers that could lead to excessive costs and burdens. They added that participating servicers interact with borrowers from a number of communities that speak a variety of languages and are in a better position to ascertain how to best provide effective relationship management to the LEP borrowers they serve. According to Treasury, servicers have told Treasury that mandating the translation or use of certain documents, among other things, would be of little benefit given the overall low demand for such documents in languages other than English, the added legal risks, the potential for inaccurate translation, and increased costs associated with the translation of documents. Treasury also noted that it may not be appropriate to require servicers to conduct business in languages other than English, especially when other regulators have not done so. For example, they noted that CFPB's recent mortgage servicing rules do not require servicers to accept applications in other languages or provide specific translation services. Treasury officials stated that the issues faced by LEP borrowers extend beyond HAMP to the broader areas of loss mitigation and mortgage origination. Accordingly, Treasury believes that it is appropriate for such industry-wide issues to be addressed by those government entities that have broad jurisdiction over the financial institutions operating in these fields. However, the MHA program provides direct outlays of taxpayer dollars to servicers and is intended to provide benefits to eligible borrowers. As such, it is important that Treasury take appropriate steps to help ensure that all eligible borrowers, including those whose primary language is not English, have access to the MHA program benefits. Without guidance on effective relationship management for LEP borrowers, the policies that MHA servicers develop may vary, and LEP borrowers may be treated differently across servicers, depending on which company services their loan. Additionally, because Treasury has not provided guidance to servicers describing the essentials of a relationship management policy for LEP borrowers, Treasury is limited in what it can measure when assessing servicers' compliance with Treasury's requirement or the effectiveness of their current practices for interacting with LEP borrowers. Ultimately, the lack of LEP policies and procedures for the MHA programs and clear expectations for effective relationship management make it less likely that servicers may effectively meet borrowers' needs for language services and therefore limit their opportunity to benefit from MHA. Conclusions: While below initial expectations, over a million borrowers have had their mortgages modified under the program. However, with respect to MHA servicer compliance, Treasury could be taking additional steps to ensure that borrowers are being treated in accordance with fair lending laws. MHA servicers are required to develop an internal control program to monitor compliance with fair lending laws that prohibit discrimination. However, Treasury has not examined servicers' internal control programs or conducted any analysis of borrowers' outcomes in HAMP modifications to identify potential fair lending risks. Our analysis found some statistically significant differences in the outcomes of fair lending populations, when compared to others, and while these variations alone do not indicate that borrowers were treated differently, they suggest that further examination may be warranted. Conducting further analyses would permit Treasury to better identify where it might apply examination resources, such as those of its compliance agent, and ascertain whether these differences are due to servicers' discretion in the application of HAMP guidelines or other factors. By requiring its compliance agent to review the fair lending internal controls of loan servicers, or reviewing the data MHA servicers collect on the race, ethnicity, and gender of borrowers, Treasury could gain additional assurance that servicers are implementing the MHA program in compliance with fair lending laws as the servicers contracted to do. Finally, despite an executive order issued in 2000 to improve access to federal programs for people with limited English proficiency and a 2011 memorandum by the Attorney General renewing the federal government's commitment to that executive order, Treasury officials have only recently developed a written plan that covers the Making Home Affordable programs as of November 2013. While Treasury does take certain measures to raise awareness and outreach to LEP borrowers, it does not provide any clarifying guidance to servicers on its requirement to have a relationship management policy for their LEP borrowers. According to a Treasury survey of MHA servicers and our discussions with five large MHA servicers, these servicers had some processes in place to assist LEP borrowers, such as using an oral translation service. Housing counselors and housing advocacy groups that work with LEP borrowers have questioned the ability of servicers to assist LEP borrowers. Without additional guidance on providing meaningful language assistance, LEP borrowers may be treated differently across servicers and have unequal access to the MHA program. Moreover, Treasury has not assessed the effectiveness of its own or its servicers' LEP practices. Further, without more specific guidance on what it expects of servicers in ensuring LEP access, Treasury and its compliance agent are limited in their ability to assess servicers' compliance with those requirements. Recommendations for Executive Action: As part of Treasury's efforts to continue improving the transparency and accountability of MHA, we recommend that the Secretary of the Treasury take actions to: * require that its compliance agent take steps to assess the extent to which servicers have established internal control programs that effectively monitor compliance with fair lending laws that apply to MHA programs; * issue clarifying guidance to servicers on providing effective relationship management to limited English proficiency borrowers; and: * ensure that the compliance agent assess servicers' compliance with LEP relationship management guidance, once established. Agency Comments and Our Evaluation: We provided a draft of this report to CFPB, DOJ, FDIC, Federal Reserve, HUD, OCC, and Treasury for review and comment. We received a written comment letter from Treasury, which is presented in appendix III. We also received technical comments from CFPB, DOJ, Federal Reserve, HUD, and Treasury that are incorporated as appropriate in the report. FDIC and OCC did not provide any comments on the draft report. In its comment letter, Treasury noted that it was still considering our findings and recommendations, and agreed that it should continue to strengthen its program in order to help as many homeowners as possible avoid foreclosure. Treasury also noted that since MHA's launch in 2009, more than 1.9 million homeowner assistance actions had taken place under the program and that they continue to take action to maximize participation rates. In response to our recommendation that it take action to require that its compliance agent begin assessing the extent to which servicers had established internal control programs that effectively monitor compliance with fair lending laws, Treasury said that it remained committed to working to ensure that homeowners are treated fairly by servicers participating in MHA. Treasury stated that it had a robust compliance program to assess servicers' performance and that it published the results of its assessments to provide greater transparency and hold servicers accountable. However, as noted earlier, Treasury does not require its compliance agent to assess servicers' internal control programs for monitoring fair lending compliance. Treasury stated that it planned to continue to explore ways to promote fair lending policies, including through coordination with fair lending supervisions and enforcement agencies and improving access to data. We agree that continuing to improve the transparency and accountability of MHA is important. As part of this effort, it will be important that Treasury require its compliance agent to assess the extent to which servicers have established internal control programs that effectively monitor compliance with fair lending laws that apply to MHA programs. Treasury also provided comments related to our recommendations that Treasury issue clarifying guidance to servicers on providing effective relationship management to limited English proficiency borrowers and ensure that its compliance agent assess servicers' compliance with this guidance. Treasury noted that it recognized the challenges homeowners with limited English proficiency faced, and had made some program materials available in other languages and sponsored call centers that offer translation services. Treasury added that the challenges faced by these homeowners extend beyond MHA to industry- wide areas of loan servicing and mortgage lending. Treasury stated that it would continue to explore additional ways to assist LEP homeowners and work with federal regulators that have broad jurisdiction over these issues. While these challenges likely extend beyond the MHA program, the MHA program provides direct outlays of billions of taxpayer dollars in incentive payments to participating servicers and is intended to provide benefits to all eligible borrowers needing assistance to avoid foreclosure. Taking appropriate steps to help ensure that LEP borrowers have access to the MHA program benefits would place this federal program in the forefront of efforts to reach these borrowers and ensure that taxpayer dollars are put to the most effective use. In its technical comments, Treasury indicated that it disagreed with three statements in the draft report. Specifically, Treasury disagreed with our characterization of participation levels in the HAMP first- lien modification program as declining despite Treasury's efforts to increase participation. We modified the text to clarify that since the implementation of HAMP Tier 2 in June 2012, the number of modifications started each month has been relatively steady through November 2013. Treasury also questioned the accuracy of our statement that it lacked assurance that the MHA program, and servicers' implementation of it, were treating all borrowers fairly and consistently citing, among other things, the role of the prudential regulators in enforcing fair lending laws and its compliance program for assessing the performance of participating servicers. However, these mechanisms only provide limited assurance since, as noted previously in the report, the prudential regulators do not focus their fair lending reviews on MHA program activity and Treasury's compliance program does not look at the fair lending controls of participating servicers. As a result, we continue to believe that it is important that Treasury require its compliance agent to assess the internal control programs that servicers are required to put into place to monitor compliance with fair lending laws that apply to MHA programs. Lastly, Treasury noted in its technical comments that it disagreed with the statement that it has only recently begun to systematically assess and take measures to ensure access to the program for borrowers whose primary language is not English. We clarified the text to acknowledge the actions taken to raise awareness and outreach to LEP borrowers, but that Treasury has not provided guidance to servicers on its requirement to have a relationship management policy for their LEP borrowers or assessed the effectiveness of its own or its servicers' LEP practices. We are sending copies of this report to the appropriate congressional committees. This report will also be available at no charge on our website at [hyperlink, http://www.gao.gov]. If you or your staffs have any questions about this report, please contact me at (202) 512-8678 or sciremj@gao.gov. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made major contributions to this report are listed in appendix IV. Signed by: Mathew J. Scirè: Director: Financial Markets and Community Investment: List of Committees: The Honorable Barbara Mikulski: Chairwoman: The Honorable Richard C. Shelby: Vice Chairman: Committee on Appropriations: United States Senate: The Honorable Tim Johnson: Chairman: The Honorable Mike Crapo: Ranking Member: Committee on Banking, Housing, and Urban Affairs: United States Senate: The Honorable Patty Murray: Chairman: The Honorable Jeff Sessions: Ranking Member: Committee on the Budget: United States Senate: The Honorable Max Baucus: Chairman: The Honorable Orrin G. Hatch: Ranking Member: Committee on Finance: United States Senate: The Honorable Hal Rogers: Chairman: The Honorable Nita Lowey: Ranking Member: Committee on Appropriations: House of Representatives: The Honorable Paul Ryan: Chairman: The Honorable Chris Van Hollen: Ranking Member: Committee on the Budget: House of Representatives: The Honorable Jeb Hensarling: Chairman: The Honorable Maxine Waters: Ranking Member: Committee on Financial Services: House of Representatives: The Honorable Dave Camp: Chairman: The Honorable Sander Levin: Ranking Member: Committee on Ways and Means: House of Representatives: [End of section] Appendix I: Objectives, Scope, and Methodology: The objectives in this report were to examine (1) the status of Making Home Affordable (MHA) and steps Treasury is taking to increase participation in the program, (2) Treasury's oversight of the MHA- related fair lending internal controls of participating servicers, and (3) Treasury's and MHA servicers' policies and practices for ensuring that borrowers with limited English proficiency (LEP) have equal access to the program. To examine the status of the MHA programs, we reviewed and analyzed Treasury's Monthly MHA Performance Reports and MHA program and expense information in the quarterly reports to Congress issued by the Special Inspector General for the Troubled Asset Relief Program (SIGTARP). We also reviewed the Congressional Budget Office's (CBO) Report on the Troubled Asset Relief Program (TARP) and spoke to CBO officials about their costs estimates for the MHA program.[Footnote 45] We also spoke with Treasury officials to obtain their views on future MHA expenditures. To understand steps Treasury has taken to increase program participation, we reviewed Treasury's Supplemental Directive and spoke with Treasury officials about their MHA Outreach and Borrower Intake Project. We also spoke to and reviewed documentation from NeighborWorks America about their involvement in the project. To examine Treasury's oversight of the MHA-related fair lending internal controls of participating servicers, we reviewed MHA program documentation, such as the Servicer Participation Agreement, MHA Handbook, and associated Supplemental Directives, to understand servicers' fair lending obligations. We spoke with officials at Treasury to gather information on their oversight of MHA servicers' practices. We also spoke with other federal agencies with fair lending oversight authority to gather information on the results of their fair lending oversight of MHA servicers. Specifically, we spoke with officials from the Department of Housing and Urban Development (HUD), Department of Justice (DOJ), Bureau of Consumer Financial Protection (CFPB), and three depository institution prudential regulators (the Federal Deposit Insurance Corporation (FDIC), the Board of Governors of the Federal Reserve System (Federal Reserve), and the Office of the Comptroller of the Currency (OCC)). Further, we spoke with staff at the five largest MHA servicers, in terms of HAMP trial modifications approved, about their internal control programs and compliance with fair lending laws. The five servicers we selected--Bank of America, CitiMortgage, JPMorgan Chase Bank, Ocwen Loan Servicing, and Wells Fargo Bank--collectively represented about 77 percent of the total HAMP trial loan modifications approved, as of October 2013. In order to determine if any potential disparities exist in the outcomes of borrowers in protected classes and other groups, we compared their outcomes to that experienced by other borrowers, for four large servicers. We obtained and analyzed Treasury's HAMP data in its system of record, Investor Reporting/2 (IR/2), through April 17, 2013. For additional information on the data reliability and methodology for this analysis, see appendix II. We determined that the IR/2 data were sufficiently reliable for the purposes of our analysis. To understand how Treasury and MHA servicers ensure access to MHA for LEP borrowers, we examined Treasury's 2000 LEP plan and its updated LEP plan and MHA guidelines, which are still in draft form. We reviewed MHA program documentation to understand servicers' obligations regarding LEP borrowers and spoke with Treasury officials about their review of servicers' LEP policies and practices. We reviewed a recent survey Treasury conducted of 17 servicers to understand how servicers work with LEP borrowers. Further, we spoke with the 5 servicers we contacted about their current LEP policies and practices and Treasury's oversight of servicers' policies. Additionally, we spoke with various mortgage industry participants, such as associations representing housing counselors, including those who directly work with LEP borrowers, and legal services attorneys. We also reviewed a national survey conducted by the National Housing Resource Center and a similar survey conducted by the California Reinvestment Coalition about servicer compliance with the new servicing standards resulting from a settlement involving 5 of the largest mortgage servicers and the federal and most state governments.[Footnote 46] The national survey collected responses from 212 housing counselors representing 28 states and the District of Columbia and the California survey received responses from 84 counselors and legal service advocates.[Footnote 47] We collected information about the survey methodology used and determined it was reliable for the purposes of reporting housing counselors' views on the experiences of individuals they work with. We conducted this performance audit from February 2013 through February 2014 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on the audit objectives. [End of section] Appendix II: Description of GAO's Econometric Analysis of HAMP: Outcomes for Fair Lending Groups: This appendix provides a summary of our econometric analysis of the Home Affordable Modification Program (HAMP) outcomes for selected population groups--primarily focusing on the outcomes for protected groups under federal fair lending laws plus low-income groups and groups in substantially minority neighborhoods--at key stages of the HAMP process.[Footnote 48] The modification process under HAMP is highly structured, based on the standard waterfall in the HAMP guidelines, so servicers' compliance with the required terms of the program may not necessarily warrant further investigation of differences in HAMP outcomes across certain groups. However, servicers have some discretion over certain actions and outcomes in the process, such as calculating borrowers' income, determining the sufficiency of borrowers' documentation and whether a borrower is in imminent danger of default, and developing their own policies to determine whether to offer principal forgiveness as part of the modification.[Footnote 49] In addition, servicers may incorrectly apply program requirements. Further, even if standards were applied uniformly, it is possible that the application of HAMP and servicer-specific guidelines could be resulting in unintended adverse consequences for some population groups. We are unable to determine from the available HAMP data whether the statistically significant differences between the selected populations and their comparison populations identified in our analysis were the result of servicer discretion, servicer errors in the application or interpretation of HAMP guidelines or servicing protocols, differences among servicers' policies, or the unintended consequences of HAMP guidelines or program design. Also, our analysis cannot account for some important factors, such as whether equivalent borrowers in these populations apply to HAMP at different rates or are more or less likely to receive assistance outside of HAMP. To the extent possible, we have controlled for several characteristics of the borrower and loan, as well as other factors which could confound potential differences in the outcomes experienced by the fair lending and other selected populations and their comparison populations in our analysis. Nonetheless, further investigation would be warranted to identify the source of the statistically significant differences identified by our analysis and what action, if any, would be appropriate to consider. Borrowers applying for loan modifications under HAMP go through a selection process, according to the HAMP guidance. Key parts of the process include several steps at three key stages. * Pretrial or application stage: At the application stage the servicer determines if the borrower is eligible for HAMP, including the requirement that the borrower must be either 60 days or more past due on mortgage payments prior to the modification, or in imminent danger of default.[Footnote 50] The borrower's application may be denied because the application was ineligible or for reasons not related to eligibility, such as an incomplete request or if the modification would require excessive forbearance. Also, the borrower may decide not to accept an approved offer from the servicer.[Footnote 51] * Trial modification stage: Once the borrower accepts the offer, they must make three timely monthly payments of the modified loan or the trial modification may be canceled for nonpayment. Prior to June 2010, a borrower could begin a trial based on stated information with data verification as a condition for conversion to permanent modification. For these borrowers, the trial may also be canceled because the loan was subsequently determined to be ineligible. Trials may also be canceled for reasons not related to eligibility such as an incomplete request and a negative net present value (NPV) result if the loan were to be modified. The modification becomes permanent if the borrower successfully completes the trial modification. * Permanent modification stage: A permanent modification is canceled if a borrower is unable to sustain the modification by redefaulting (i.e., the loan becoming 90 days or more delinquent). If the borrower is eligible for HAMP, the servicer evaluates the loan using a standardized NPV test, which compares the net present value of cash flows with and without the modification. The HAMP Tier 1 modification must reduce the borrower's first-lien mortgage payment to as close to 31 percent housing debt-to-income (DTI) ratio using a sequence of steps--the so-called standard modification waterfall. The steps include capitalizing accrued interest, reducing the interest rate on the loan, and extending the term or amortization period of the loan. Principal forbearance could be used as needed and principal forgiveness could be offered at any stage of the modification process. Description of Data Used: The main data source for the study is Treasury's HAMP data in its system of record, Investor Reporting/2 (IR/2) made available to government agencies. The data used are restricted to first-lien loans in the 50 states of the United States and in the District of Columbia. We excluded loans owned or guaranteed by the Federal Housing Administration (FHA) or Veterans Affairs (VA).[Footnote 52] The time period analyzed is for HAMP applications and modifications from April 2009 through April 17, 2013. The HAMP data were supplemented with housing-and mortgage-related data from the 2010 Census from the Census Bureau: these data provided neighborhood-level characteristics such as the poverty rate, household education, mortgages with second liens, and ratio of rental values to home values (property risk) that could be associated with HAMP outcomes.[Footnote 53] The HAMP data include a variety of information on individual borrowers and other characteristics about the loan, property, investor, servicer, and loan modification terms, and the current status of the modification. Some of the data are specific to conditions before or after the modification, as well as at the loan origination. The data are generally available for the fair lending and our other selected populations and their comparison populations; however, data on borrower income are not available in the early stage of the HAMP process for borrowers whose applications were denied. In general, borrowers whose HAMP application did not advance to the NPV evaluation stage have more missing data because the information used for the NPV evaluation was not recorded in the database.[Footnote 54] The HAMP data are supplied by servicers that participate in the program. Servicers are required to report data when borrowers request a modification, during the trial period, when the trial is converted to a permanent modification, and to report the monthly performance of the permanent modification. The data used for the analysis consist of 4.7 million loans, representing 92 percent of the HAMP applications as of April 17, 2013. See table 1 for details of the HAMP data used in our analysis, by phase of the HAMP process. Table 2: HAMP Data Used in the Analysis by Phase of the HAMP Process, April 2009 to April 17, 2013: Pretrial or application stage: 4,703,140; Completed applications: 1,972,906; Denied applications: 2,547,380; Unavailable outcome: 182,854; Trial modification stage: 1,972,906; Successful trial modifications within 6 months: 1,090,881; Canceled trial modifications: 614,510; Unavailable outcome: 267,515; Permanent modification stage: 1,090,881; Ongoing or paid-off permanent modifications within 12 months[A]: 679,815; Canceled permanent modifications: 125,937; Unavailable outcome: 285,129. Source: GAO analysis of Treasury data. Notes: The data are for first-lien HAMP loans in the 50 states of the United States and in the District of Columbia. They exclude loans owned or guaranteed by FHA or VA, and loans owned or guaranteed by Fannie Mae, a government-sponsored enterprise, that were modified under a pilot program on principal forgiveness (about 240 loans). The data are based on HAMP loan activity from April 2009 to April 17, 2013. [A] Paid-off loans comprised about 1 percent of the loans. [End of table] For every loan, the data included a descriptor of whether or not the loan completed the respective stage of the HAMP process, and, if it had not, the reason for the loan not reaching the next stage. We separated the denials of applications and cancellations of trial modifications due to reasons we determined to be related to the HAMP eligibility guidelines from the other reasons to provide additional insight into differences in outcomes for the groups analyzed.[Footnote 55] * In the application stage, the top three reasons for denial of applications for modification that were related to eligibility guidelines, were (1) borrowers whose current DTI was less than 31 percent, (2) ineligible mortgage, and (3) borrower was found to not be in imminent danger of default. The three top reasons for denial of applications for modification not related to eligibility guidelines were (1) incomplete request, (2) approved offer not accepted by borrower or withdrawn request, and (3) loan modification that would require excessive forbearance. * In the trial modification stage, the three top reasons modifications were canceled due to reasons we determined were related to eligibility guidelines were (1) borrowers whose current DTI was less than 31 percent, (2) ineligible mortgage, and (3) the property was not owner- occupied. The top reasons for trial modification cancellations not related to eligibility or payment default were (1) incomplete request, (2) approved offer not accepted by borrower or withdrawn request, (3) loans with negative NPV. We also analyzed trial modification cancellations due to payment default (loans that became 30 days or more delinquent). * Lastly, in the permanent modification stage, there was only one reason that modifications could be canceled--redefault (loans that were 90 days or more delinquent). Using information on where a property is located, we also include variables to control for potential differences in state laws, regulations, and programs that could affect the cost of foreclosure and treatment of delinquent borrowers.[Footnote 56] We also constructed several mortgage-and housing-related variables of the neighborhoods for the loans in our analysis using data from the 2010 Census. Based on the census tract in which the property is located, we associated various variables to the HAMP loan-level data. The location- specific variables include characteristics such as minority concentration, poverty, age, foreign-born concentration, college education of household, and property risk characteristics. We examined the availability of HAMP data for the selected population groups and their comparison populations for our analysis. The data show that missing data for race/ethnicity reduced from 54 percent in the early stage of the HAMP process when fewer data were reported by servicers to 43 percent in the later stage. Although the race/ethnicity of some borrowers was unavailable, the data suggest that the properties of borrowers with unavailable race/ethnicity were disproportionately located in areas where racial minorities were predominant, particularly in the early stage of the HAMP process. The proportions of missing data were generally much lower for gender, [Footnote 57] income, and minority composition of the areas and their comparison populations.[Footnote 58] While previous studies that used the HAMP data acknowledge the limitations of missing data on the fair lending populations and their comparison populations, none of the studies indicated that the available data are not suitable for fair lending analysis.[Footnote 59] Nonetheless, we took several steps to reduce the potential bias of missing data of the selected populations on our analysis. First, we included observations representing the missing data as another category, where possible, particularly since previous studies have indicated that missing race/ethnicity data are not likely to be purely random.[Footnote 60] Second, as part of our robustness checks, we also conducted the analysis excluding the observations representing the missing data to learn about its potential impact on our results. The results were similar for our key findings. And, third, we restricted the data to the period since December 2009, when servicers were required to report fair lending- related data, and the proportions of missing data for the fair lending populations and their comparison populations decreased significantly. All the variables used are in categorical format (i.e., each variable is divided into sub-groups), except the modification types (capitalization, principal forbearance, and principal forgiveness), which are measured by the percent change of the loan balances. [Footnote 61] There are fewer missing data of the variables in the later stage of HAMP since the data availability generally improved as the borrower moved through the modification process. Details of the Regression Analysis: While our econometric analysis is consistent with previous studies that have analyzed differences between the certain selected populations and their comparison groups, it differs in several ways, including the data used, estimation technique, market participants, geographic coverage, and time period analyzed.[Footnote 62] We used a multivariate regression technique for all the HAMP outcomes, rather than tabular analysis, which allowed us to control for several potential confounding factors, including credit risk-related factors, for which loan, borrower, property, and neighborhood characteristics serve as proxies. In particular, we included several variables that are used to capture the creditworthiness of borrowers in mortgage markets, such as delinquency status of borrowers before the modification, FICO credit scores at modification, the debt-to-income (DTI) ratios for both the front-end before modification and back-end after modification, and the LTV of the property at origination and at the time of the modification. We also used income cohorts for borrowers relative to incomes in their geographic areas (metropolitan statistical areas--MSAs) instead of nationally. Also, we measured default and redefault by the age of the loan since modification, which is important since default rates generally vary over time. Some of these are important differences between our study and Mayer and Piven, which is the closest of all the previous studies to ours in terms of the data used and issues that were addressed. Mayer and Piven argued that, overall, the fair lending populations did not experience differential outcomes compared to their comparison populations. Thus, race, ethnicity, gender, or income has "very little" impact on borrowers' successful participation in HAMP as well as benefits of the program at every key stage of the program. Their results, nonetheless, suggest that the authors found some disparities for certain groups. For instance, non-Hispanic African-Americans, compared to whites, were more likely to redefault; Hispanics, compared to non-Hispanics, were more likely to have their trial or permanent modifications canceled; women were at least as successful as men with respect to HAMP outcomes analyzed; and low-income borrowers were less likely to redefault on their permanent modifications compared to higher income borrowers. Moreover, their study assessed overall HAMP program outcome results and did not analyze potential outcome differences and actions of individual servicers because the data set used for the analysis--the HAMP general public data file--did not contain variables that could be used to identify the servicer of the loan. A study by the California Reinvestment Coalition of HAMP trial modifications in four MSAs in California found racial and ethnic disparities in the experiences of borrowers, which they argued was supported by their survey of housing counselors. The analysis involved tabulation rather than multivariate regression analysis and did not consider the effects of servicers due to the same limitation that the Mayer and Piven study faced with the lack of servicer identifying information in the data set used for the analysis. The National Community Reinvestment Coalition conducted a study of distressed homeowners who sought assistance from NCRC's Housing Counseling Network. The data were collected over a 2-month period in 2010 from 29 organizations and 179 borrowers. The 179 respondents included both HAMP-eligible and noneligible borrowers. The findings related to fair lending included the following--servicers foreclosed on delinquent non-Hispanic African-American borrowers more quickly than on their counterpart whites or Hispanic borrowers, and HAMP- eligible white borrowers were almost 50 percent more likely to receive a modification than their non-Hispanic African-American counterparts. The study acknowledged the limitation that it did not use a nationally representative sample of distressed homeowners. Furthermore, similar to the CRC study, the analysis used tabulation rather than multivariate regression. Voicu et al. analyzed redefault rates using data for New York City for HAMP and proprietary (non-HAMP) loan modifications from January 2008 to November 2010. While they found that borrowers who received HAMP modifications were less likely to redefault compared to those that received proprietary modifications, the borrower's race or ethnicity was not significantly correlated with the odds of redefault. The analysis covered a limited geographic market and did not include outcomes in the early stage of HAMP. Model: Based on economic reasoning, data availability, and previous studies on loan modifications, we used a relatively flexible specification to estimate the outcome of a loan at certain stages of the HAMP process. The general regression specification for the models is: =X + Zä + å. * y is the HAMP outcome measure being assessed, such as whether a loan was eligible for trial modification compared to being denied due to DTI less than 31 percent, ineligible mortgage, or not in imminent default--a multinomial outcome; it could also be whether a loan remained current or redefaulted within 12 months of the permanent modification--a binomial outcome; * X represents the fair lending and other selected populations--the income-related variables could not be used in the equations for the early stage of the HAMP process due to lack of data; * Z represents a series of control variables, including other borrower characteristics, the loan, property, neighborhood, modification terms, geographic and time effects, as well as investor/lender and servicer effects; * â and ä ? are the parameters to be estimated; and: * å represents an error term. Econometric estimates: We estimated the regression models using the logistic technique for pooled data for all the servicers.[Footnote 63] We used probability weights that were based on the distribution of the outcomes at a stage of the HAMP process because the sample used for the regression, especially in the early stage of HAMP, differed from the full data due to missing observations for certain key variables, including the fair lending and other selected populations and their comparison populations.[Footnote 64] Overall, the models are statistically significant, and most of the control variables are also significant at the 5 percent level or better and their effects (the direction of their impacts) are generally consistent with our expectations. We present below the results of the HAMP outcomes we analyzed for four of the large servicers with significant HAMP activity, where there are statistically significant differences between the fair lending and other selected populations and their comparison populations. Although our results show adverse as well as favorable outcomes for the selected populations compared to their comparison populations, we focus on cases where the outcomes were unfavorable to the selected populations because they are underrepresented in housing and mortgage markets. This approach is generally consistent with the focus of fair lending analysis on adverse outcomes for protected groups. Also, we focus below on the effects where the predicted probability of an outcome for all borrowers is 10 percent or more. Although this threshold has no strong statistical, economic, or legal justification, it helps us to focus on the more important findings, and is therefore appropriate for the diagnostic purpose of our study. The complete estimated probabilities are presented in tables 2 through 4. Pretrial or Application Stage Results: Denial of Application Due to Ineligibility: We compared borrowers who are ineligible for trial modification--due to their debt-to-income (DTI) ratios being less than 31 percent, their mortgages were ineligible, or they were found to not be in imminent danger of default--to those eligible for trial modification. The estimates are based on a multinomial logistic regression of denial of application for these three reasons using pooled data for all servicers. The main results, from table 2, are: * Overall, the denial rate of borrowers because their DTI was determined to be less than 31 percent, was about 11 percent.[Footnote 65] * We found statistically significant differences in the denial rates of trial modification between fair lending populations and their comparison populations due to the servicer's determination that the borrower's DTI was less than 31 percent. - The difference in denial rates between the non-Hispanic African- American borrowers and their comparison group of non-Hispanic whites was at least 13 percent higher for all four large servicers. - The difference in denial rates between non-Hispanic American Indians, Alaska Natives, Native Hawaiians, and Other Pacific Islanders (collectively referred to as AIPI in this appendix) borrowers and their comparison group of non-Hispanic whites was at least 11 percent higher, for two of the large servicers we analyzed. - Borrowers with unavailable information on their race/ethnicity (these borrowers have properties in substantially minority areas) or gender had at least 22 and 15 percent higher denial rates, respectively, than their comparison populations for all four large servicers. - Borrowers in substantially minority areas had at least 3 percent higher denial rates than their comparison populations of borrowers in nonsubstantially minority areas for all four large servicers. - On the other hand, non-Hispanic Asians, Hispanics, and females had generally lower denial rates than their comparison populations. * Although we found some disparities between the selected populations and their comparison populations for denials due to servicers' determination that borrowers had ineligible mortgages or were not in imminent danger of default, the results are not discussed since the overall denial rates are small. Table 2: Denial Rate and Percent Difference in Denial Rates between Selected Populations and their Comparison Populations for HAMP Applications Denied Due to a DTI Less Than 31 Percent: DTI less than 31 percent; Average denial rate: 11%. Percent difference from comparison group: Race/ethnicity: Non-Hispanic White[A]; Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%. Race/ethnicity: Non-Hispanic Asian; Servicer A: -21%; Servicer B: -21%; Servicer C: -22%; Servicer D: -22%. Race/ethnicity: Non-Hispanic African-American; Servicer A: 13%; Servicer B: 13%; Servicer C: 14%; Servicer D: 14%. Race/ethnicity: Hispanic; Servicer A: 16%; Servicer B: -18%; Servicer C: -18%; Servicer D: -18%. Race/ethnicity: Non-Hispanic AIPI; Servicer A: NA; Servicer B: NA; Servicer C: 11%; Servicer D: 11%. Race/ethnicity: Not available[B]; Servicer A: 22%; Servicer B: 26%; Servicer C: 25%; Servicer D: 26%. Gender: Male[A]; Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%. Gender: Female; Servicer A: -10%; Servicer B: -12%; Servicer C: -11%; Servicer D: -11%. Gender: Not available[B]; Servicer A: 15%; Servicer B: 21%; Servicer C: 19%; Servicer D: 19%. Substantially minority area[C]: Nonsubstantially minority[A]; Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%. Substantially minority area[C]: Substantially minority; Servicer A: 3%; Servicer B: 3%; Servicer C: 4%; Servicer D: 4%. Source: GAO analysis of Treasury data. Notes: AIPI: American Indian or Alaska Native (AIAN) and Native Hawaiian or Other Pacific Islander (NHPI). NA means not available because of lack of sufficient data for some of the denial outcomes. The estimates are based on a probability-weighted multinomial logistic regression of denial of application due to DTI less than 31 percent, ineligible mortgage, or borrower found not to be in danger of imminent default compared to loans eligible for trial modification (the reference group). The predicted probabilities are based on the coefficient estimates from the multinomial logistic model and the mean values of the covariates, where the means are based on all the observations used in the estimation. All of the estimates in the table are significant at the 5 percent level or better, meaning that the estimated effects have a 5 percent or less chance of being random. [A] Comparison populations are shown as 0 since they are the base for percent change calculation. [B] Not available includes all other subgroups not listed in the category, including missing observations. [C] An area is substantially minority if the proportion of all minority households in a tract is at least 30 percent. [End of table] Denial of Application for Reasons Not Related to Eligibility: We compared borrowers who were found to be eligible for trial modification but had their application denied for a number of reasons-- such as the servicer deemed their request incomplete, the borrower did not accept the approved offer from the servicer or withdrew their request for a modification, or the servicer determined that the loan modification would require excessive forbearance--to those approved for trial modification.[Footnote 66] The estimates are based on a multinomial logistic regression of denial of application for these three reasons using pooled data for all servicers. There are no estimates for one of the four large servicers because of insufficient data for some of the outcomes. The main results, from table 3, are: * Overall, the denial rate of borrowers due to an incomplete request is about 23 percent.[Footnote 67] [Footnote 68] * We found disparities in denial rates where the reason was an incomplete request between fair lending populations and their comparison populations. - The difference in denial rates between Hispanics from their comparison population of non-Hispanic whites was about 8 percent higher for all three of the large servicers we analyzed. - The difference in denial rates between non-Hispanic Asians and African-Americans from their comparison population of non-Hispanic whites was at least 1 and 3 percent higher, respectively, for all three servicers we analyzed. - The difference in denial rates for borrowers with unavailable information on their race/ethnicity was about 7 percent higher compared to non-Hispanic whites for the three large servicers we analyzed. Furthermore, we found in our analysis that these borrowers' properties were located disproportionately in substantially minority areas. Also the difference in denial rates for borrowers with unavailable information on their gender was at least 24 percent higher for two of the three large servicers we analyzed. - Borrowers in substantially minority areas had at about 1 percent higher denial rates than their comparison populations of borrowers in nonsubstantially minority areas for two of the three large servicers we analyzed. - On the other hand, non-Hispanic AIPI and females were less likely to be denied than their comparison populations. * Although we found some disparities between the fair lending and other selected populations and their comparison populations for denials due to borrowers not accepting their approved offers or for excessive forbearance, the results are not discussed since the overall denial rates are small. Table 4: Denial Rate and Percent Difference in Denial Rates between Selected Populations and their Comparison Populations for HAMP Applications Denied Due to an Incomplete Request: Incomplete request: Average denial rate: 23%. Percent difference from comparison group: Race/ethnicity: Servicer: Non-Hispanic White[A]; Servicer A: 0%; Servicer B: NA; Servicer C: 0%; Servicer D: 0%. Race/ethnicity: Servicer: Non-Hispanic Asian; Servicer A: 2%; Servicer B: NA; Servicer C: 1%; Servicer D: 4%. Race/ethnicity: Servicer: Non-Hispanic African-American; Servicer A: 4%; Servicer B: NA; Servicer C: 4%; Servicer D: 3%. Race/ethnicity: Servicer: Hispanic; Servicer A: 8%; Servicer B: NA; Servicer C: 8%; Servicer D: 9%. Race/ethnicity: Servicer: Non-Hispanic AIPI; Servicer A: NA; Servicer B: NA; Servicer C: NA; Servicer D: -21%. Race/ethnicity: Servicer: Not available[B]; Servicer A: 8%; Servicer B: NA; Servicer C: 7%; Servicer D: 9%. Gender: Male[A]; Servicer A: 0%; Servicer B: NA; Servicer C: 0%; Servicer D: 0%. Gender: Female; Servicer A: -1%; Servicer B: NA; Servicer C: -12%; Servicer D: -15%. Gender: Not available[B]; Servicer A: 0%; Servicer B: NA; Servicer C: 24%; Servicer D: 36%. Substantially minority area[C]: Nonsubstantially minority[A]; Servicer A: 0%; Servicer B: NA; Servicer C: 0%; Servicer D: 0%. Substantially minority area[C]: Substantially minority; Servicer A: 0%; Servicer B: NA; Servicer C: 1%; Servicer D: 1%. Source: GAO analysis of Treasury data. Notes: AIPI: American Indian or Alaska Native (AIAN) and Native Hawaiian or Other Pacific Islander (NHPI). NA means not available because of lack of sufficient data for some of the denial outcomes. The estimates are based on a probability-weighted multinomial logistic regression of denial of application due to incomplete request (i.e., missing documentation), approved offer not accepted, or excessive forbearance compared to loans approved for trial modification (the reference group). The predicted probabilities are based on the coefficient estimates from the multinomial logistic model and the mean values of the covariates, where the means are based on all the observations used in the estimation. All the estimates in the table are significant at the 5 percent level or better, meaning that the estimated effects have a 5 percent or less chance of being random. [A] Comparison populations are shown as 0 since they are the base for percent change calculation. [B] Not available includes all other subgroups not listed in the category, including missing observations. [C] An area is substantially minority if the proportion of all minority households in a tract is at least 30 percent. [End of table] Trial Modification Stage Results: Cancellation of Trial Modification for Ineligibility: We compared borrowers whose trial modification was canceled because their DTI was less than 31 percent, mortgage was ineligible, or property was not owner-occupied to those borrowers that were eligible for permanent modification. The estimates are based on a multinomial logistic regression of cancellation of trial modification for these three reasons using pooled data for all servicers. Overall, the cancellation rate for borrowers with a DTI less than 31 percent was about 0.8 percent. The cancellation rate was about 0.3 percent for borrowers with ineligible mortgages and 0.3 percent for those with non owner-occupied properties. Although we found some differences between the fair lending and other selected populations and their comparison populations, the results are not discussed since the overall cancellation rates were small. Cancellation of Trial Modification for Reasons Not Related to Eligibility or Payment Default: We compared borrowers whose trial modification were canceled because the servicer determined their request was incomplete, the borrower did not accept the offer they received or withdrew their requests for trial modification, or the servicer determined that modifying the mortgage would result in a negative NPV. The estimates are based on a multinomial logistic regression of cancellation of trial modification for these three reasons using pooled data for all servicers. Overall, the cancellation rate for borrowers with incomplete requests was about 4 percent. The rate was about 3 percent for borrowers who did not accept their approved offer or withdrew their request and 0.3 percent for loans that would have a negative NPV if modified. Although we found some differences between the fair lending and other selected populations and their comparison populations, the results are not discussed since the overall cancellation rates were small. Cancellation of Trial Modification for Payment Default: We compared borrowers whose trial modification was canceled for payment default (i.e., their loans became 30 days or more delinquent) within 6 months of the modification to those borrowers who were approved for permanent modification. Overall, the estimated default rate of trial modifications was 4 percent. Although we found differences between the fair lending and other selected populations and their comparison populations, the results are not discussed since the overall cancellation rate was small. Permanent Modification Stage Results: Cancellation of Permanent Modification for Redefault: We compared borrowers who received permanent modification but redefaulted (i.e., their loans became 90 days or more delinquent) within 12 months of the modification to those borrowers who remained current on their payments (or paid off the loan). The estimates are based on a binary logistic regression of redefault using pooled data for all the servicers, and including the three modification types-- capitalization, principal forgiveness, and principal forbearance--as covariates. The main results, from table 4, are: * Overall, the estimated redefault rate of permanent modifications was 11 percent. * The redefault rates differ by the modification type--capitalization, principal forgiveness, or principal forbearance. The use of principal forgiveness results in lower redefault rates compared to capitalization or principal forbearance. Capitalization increases the redefault rate by 3 percentage points while it decreases by 3 percentage points with principal forgiveness; principal forbearance lowers the redefault rate by 1 percentage point. * We found disparities in redefault rates between certain fair lending populations and their comparison populations. - The difference in redefault rates between non-Hispanic African- Americans and non-Hispanic whites whose loans were serviced by any of the four large servicers was about 14 percent higher, irrespective of the modification type. -The difference in redefault rates between non-Hispanic AIPI and non- Hispanic whites whose loans are serviced by any of three large servicers we analyzed was about 7 percent higher, irrespective of the modification type. - The differences in redefault rates between borrowers with unavailable information on their race and income were about 5 and 18 percent higher, respectively, compared to their comparison populations, irrespective of the modification type. - However, non-Hispanic Asians, Hispanics, borrowers in substantially minority areas, and borrowers with low, moderate incomes, and middle- income, were less likely to redefault compared to their comparison populations. Table 5: Redefault of HAMP Permanent Modification and Percent Difference in Redefault Rate between Selected Populations and Their Comparison Populations: Estimated redefault rate of permanent modification: Average redefault rate: 11%. Modification type: Capitalization; Redefault rate: With capitalization: 11%; Without capitalization: 8%. Modification type: Principal forgiveness; Redefault rate: With forgiveness: 8%; Without forgiveness: 11%. Modification type: Principal forbearance; Redefault rate: With forbearance: 10%; Without forbearance: 11%. Percent difference from comparison group: Race/ethnicity: Non-Hispanic White[A]; Capitalization: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%; Principal forgiveness: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%; Principal forbearance: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%. Race/ethnicity: Non-Hispanic Asian; Capitalization: Servicer A: -12%; Servicer B: -13%; Servicer C: -13%; Servicer D: -12%; Principal forgiveness: Servicer A: -13%; Servicer B: -13%; Servicer C: -12%; Servicer D: -13%; Principal forbearance: Servicer A: -13%; Servicer B: -13%; Servicer C: -13%; Servicer D: -12%. Race/ethnicity: Non-Hispanic African-American; Capitalization: Servicer A: 15%; Servicer B: 15%; Servicer C: 15%; Servicer D: 14%; Principal forgiveness: Servicer A: 15%; Servicer B: 15%; Servicer C: 15%; Servicer D: 14%; Principal forbearance: Servicer A: 15%; Servicer B: 15%; Servicer C: 15%; Servicer D: 14%. Race/ethnicity: Hispanic; Capitalization: Servicer A: -15%; Servicer B: -15%; Servicer C: -15%; Servicer D: -14%; Principal forgiveness: Servicer A: -15%; Servicer B: -16%; Servicer C: -15%; Servicer D: -15%; Principal forbearance: Servicer A: -15%; Servicer B: -15%; Servicer C: -15%; Servicer D: -14%. Race/ethnicity: Non-Hispanic AIPI; Capitalization: Servicer A: NA; Servicer B: 7%; Servicer C: 7%; Servicer D: 7%; Principal forgiveness: Servicer A: NA; Servicer B: 8%; Servicer C: 7%; Servicer D: 7%; Principal forbearance: Servicer A: NA; Servicer B: 7%; Servicer C: 7%; Servicer D: 7%. Race/ethnicity: Not available[B]; Capitalization: Servicer A: 5%; Servicer B: 5%; Servicer C: 5%; Servicer D: 5%; Principal forgiveness: Servicer A: 5%; Servicer B: 8%; Servicer C: 5%; Servicer D: 5%; Principal forbearance: Servicer A: 5%; Servicer B: 5%; Servicer C: 5%; Servicer D: 5%. Gender: Male[A]; Capitalization: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%; Principal forgiveness: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%; Principal forbearance: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%. Gender: Female; Capitalization: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%; Principal forgiveness: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%; Principal forbearance: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%. Gender: Not available[B]; Capitalization: Servicer A: -2%; Servicer B: -2%; Servicer C: -2%; Servicer D: -2%; Principal forgiveness: Servicer A: -2%; Servicer B: -2%; Servicer C: -2%; Servicer D: -2%; Principal forbearance: Servicer A: -2%; Servicer B: -2%; Servicer C: -2%; Servicer D: -2%. Substantially minority area[C]: Nonsubstantially minority[A]; Capitalization: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%; Principal forgiveness: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%; Principal forbearance: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%. Substantially minority area[C]: Substantially minority; Capitalization: Servicer A: -5%; Servicer B: -6%; Servicer C: -5%; Servicer D: -5%; Principal forgiveness: Servicer A: -6%; Servicer B: -6%; Servicer C: -6%; Servicer D: -5%; Principal forbearance: Servicer A: -5%; Servicer B: -6%; Servicer C: -6%; Servicer D: -5%. Income[D]: High-income[A]; Capitalization: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%; Principal forgiveness: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%; Servicer D: 0%; Principal forbearance: Servicer A: 0%; Servicer B: 0%; Servicer C: 0%. Servicer D: 0%. Income[D]: Low-income; Capitalization: Servicer A: -11%; Servicer B: -12%; Servicer C: -11%; Servicer D: -11%; Principal forgiveness: Servicer A: -12%; Servicer B: -12%; Servicer C: -12%; Servicer D: -11%; Principal forbearance: Servicer A: -11%; Servicer B: -12%; Servicer C: -12%; Servicer D: -11%. Income[D]: Moderate-income; Capitalization: Servicer A: -6%; Servicer B: -7%; Servicer C: -6%; Servicer D: -6%; Principal forgiveness: Servicer A: -7%; Servicer B: -7%; Servicer C: -7%; Servicer D: -6%. Income[D]: Middle-income; Capitalization: Servicer A: -3%; Servicer B: -3%; Servicer C: -3%; Servicer D: -3%; Principal forgiveness: Servicer A: -3%; Servicer B: -3%; Servicer C: -3%; Servicer D: -3%; Principal forbearance: Servicer A: -3%; Servicer B: -3%; Servicer C: -3%; Servicer D: -3%. Income[D]: Not available[B]; Capitalization: Servicer A: 20%; Servicer B: 20%; Servicer C: 20%; Servicer D: 18%; Principal forgiveness: Servicer A: 20%; Servicer B: 21%; Servicer C: 21%; Servicer D: 19%; Principal forbearance: Servicer A: 20%; Servicer B: 21%; Servicer C: 20%; Servicer D: 19%. Source: GAO analysis of Treasury data. Notes: AIPI: American Indian or Alaska Native (AIAN) and Native Hawaiian or Other Pacific Islander (NHPI). NA means not available because of lack of sufficient data for some of the denial outcomes. The estimates for the modification types are based on a single probability-weighted logistic regression of redefault of permanent modifications (i.e., 90 or more days delinquent) within 12 months compared to loans that remained current or were paid off. The predicted probabilities are based on the coefficient estimates from the logistic model and the mean values based on all the observations used in the estimation. All the estimates in the table are significant at the 5 percent level or better, except the female variable, meaning that the estimated effects have a 5 percent or less chance of being random. [A] Comparison populations are shown as 0 since they are the base for percent change calculation. [B] Not available includes all other subgroups not listed in the category, including missing observations. [C] An area is substantially minority if the proportion of all minority households in a tract is at least 30 percent. [D] The income groups are identified as follows: low-income, if borrower's income is less than 50 percent of MSA median household income; moderate-income, if income equals to or is greater than 50 percent and less than 80 percent of MSA median household income; middle-income, if income equals or is greater than 80 percent and less than 120 percent of MSA median household income; and high-income, if income is greater than 120 percent of MSA median household income. [End of table] Results of Additional Analysis on Household Area Incomes: Furthermore, we included area incomes by groups.[Footnote 69] While the results were consistent with the key findings reported in Tables 2 to 4, we also found that households in low-moderate income areas were more likely to be denied for DTI less than 31 percent and those in middle-income areas were more likely to be denied for incomplete request compared to their comparison populations of households in high income areas. Also households in areas with unavailable median incomes were more likely to re-default on their permanent modifications compared to their comparison population. Robustness Checks of Results: We took several steps to check these regression models for robustness, particularly for the key findings on denials of HAMP applications due to a servicer's determination that the borrower's DTI was less than 31 percent, cancellation of trial modification due to the servicer determining the request was incomplete, and for redefaults of permanent modification. Specifically, we did the following: * estimated the outcomes separately for each of the four large servicers; * estimated the models excluding the observations for the missing fair lending populations; * estimated the model comparing Hispanics to non-Hispanics (the majority comparison group); * restricted the data to the period since December 2009 when servicers were required to collect fair lending related data; * for the estimation of redefault rates of permanent modifications we used loans that had aged 24 months since the permanent modification; and: * estimated the models without probability weights. These checks were consistent with the reported key findings of differences between the fair lending and other selected populations and their comparison populations. Nonetheless, there are limitations of this study, including limited data on the selected populations, the lack of certain variables that could help capture the credit risks of borrowers and the loans such as the wealth of the borrowers and their knowledge of the loan modification process, and, in particular, if borrowers have accessed housing counseling services. Also, as noted, this analysis concerns potential disparate outcomes for some populations and does not mean to imply disparate treatment by some servicers or that borrowers experienced disparate impact in violation of fair lending laws, so the findings in this study should be interpreted cautiously and further analysis may be appropriate. [End of section] Appendix III: Comments from the Department of the Treasury: Department of the Treasury: Washington, D.C. 20220: January 23, 2014: Mathew J. Scire: Director: Financial Markets and Community Investment: U.S. Government Accountability Office: 441 G Street, NW: Washington, D.C. 20548: Dear Mr. Scire: I write in response to the Government Accountability Office's (GAO) draft report, GAO-14-117, regarding the Making Home Affordable Program (MHA). The Department of the Treasury (Treasury) supports GAO's important oversight function. We appreciate not only the opportunity to provide comments to the draft report but also the constructive input and feedback from your team during the course of the audit. The draft report offers several findings on Treasury's efforts to increase participation in MHA, monitor servicers' compliance with fair lending laws, and assist borrowers with limited English proficiency. The draft report also makes three recommendations. Although we are still in the process of considering those findings and recommendations, we agree that we should continue to strengthen our program in order to help as many homeowners as possible avoid foreclosure. MHA has been a critical part of the Obama Administration's efforts to provide relief to families at risk of foreclosure and help the housing market recover from a historic crisis. Since its launch in 2009, more than 1.9 million homeowner assistance actions have taken place under MHA, including nearly 1.3 million permanent first lien mortgage modifications under the Home Affordable Modification Program (HAW)). In addition, MHA has indirectly helped millions more by creating new standards for mortgage assistance that led to more affordable and sustainable private modifications. We continue to take action to maximize participation rates. Despite the declining number of homeowners at risk of foreclosure, HAMP participation rates have remained relatively steady since early 2012 due in part to program changes such as expanding borrower eligibility through HAMP Tier 2. Treasury remains committed to working to ensure that homeowners are treated fairly by servicers participating in MHA. Our efforts in that regard include, first and foremost, a robust compliance program to assess the performance of participating servicers. We also publish the results of those assessments in order to provide greater transparency and hold servicers accountable. In addition, Treasury provides a toll- free call center—MHA Help—where homeowners can receive an independent review of their MHA application, as well as general support during the application process. To date, tens of thousands of homeowners have received assistance through MHA Help. In an effort to promote fair lending policies, Treasury shares extensive loan-level program data with the federal regulators that are responsible for the supervision and enforcement of fair lending laws. These regulators include MHA transactions in their examinations of MHA servicers, and we are pleased that GAO found that these regulators "had not identified fair lending violations related to the MHA program specifically." Additionally, we make a loan-level data file available to the public, including the independent research community. We will continue to explore ways to promote fair lending policies, including through coordination with the fair lending supervision and enforcement agencies and improving access to MHA data. Lastly, since the inception of the program, Treasury has recognized the challenges faced by homeowners who have limited English proficiency. Accordingly, we have made core program materials, online resources, and public service announcements available in multiple languages, and MHA Help and the other toll-free call centers we sponsor offer translation services in more than 170 different languages. It is important to note, however, that these same challenges extend beyond MHA to industry-wide areas of loan servicing and mortgage lending. Thus, as we continue to explore additional ways to assist homeowners with limited English proficiency, we will work with the federal regulators that have broad jurisdiction over these issues. Treasury values GAO's review of our efforts to provide mortgage relief to homeowners and prevent avoidable foreclosures, and we look forward to continuing work with you and your team. Sincerely, Signed by: Timothy J. Bowler: Acting Assistant Secretary for Financial Stability: [End of section] Appendix IV: GAO Contact and Staff Acknowledgments: GAO Contact: Mathew J. Scire, (202) 512-8678 or sciremj@gao.gov: Staff Acknowledgments: In addition to the contact named above, Harry Medina (Assistant Director), Bethany Benitez, Emily Chalmers, William R. Chatlos, Lynda Downing, John Karikari, Anar Ladhani, John Lord, Thomas J. McCool, Susan Offutt, Jena Sinkfield, Anne Y. Sit-Williams, Jim Vitarello, and Heneng Yu made key contributions to this report. [End of section] Selected Bibliography: Agarwal, S., G. Amromin, I. Ben-David, S. Chomsisengphet, T. Piskorski, A. Seru. "Policy Intervention in Debt Renegotiation: Evidence from the Home Affordable Modification Program." NBER Working Paper Series, No. 18311, August 2012. Berkovec, J., G. Canner, S. Gabriel, and T. Hannan. "Race, Redlining, and Residential Mortgage Loan Performance." Journal of Real Estate Finance and Economics, vol. 9, no. 3 (1994): 263-294. California Reinvestment Coalition (CRC), Race to the Bottom: An Analysis of HAMP Loan Modification Outcomes By Race and Ethnicity for California (July 2011). Cheng, P., Z. Lin, and Y. Liu. "Do Women Pay More for Mortgages?" Journal of Real Estate Finance and Economics, vol. 43, no. 4 (2011): 423-440. Collins, J., K. Lam, and C. Herbert. "State Mortgage Foreclosure Policies & Lender Interventions: Impacts on Borrower Behavior in Default." Journal of Policy Analysis and Management, vol. 30, no. 2 (2011): 216-232. Dietrich, J., Missing Race Data in HMDA and the Implications for the Monitoring of Fair Lending Compliance., Office of the Comptroller of the Currency, OCC Economics, Working Paper No. 2001-1 (March 2011) . GAO. Foreclosure Mitigation: Agencies Could Improve Effectiveness of Federal Efforts with Additional Data Collection and Analysis. GAO-12- 296. Washington, D.C.: June 28, 2012. GAO. Troubled Asset Relief Program: Further Actions Needed to Fully and Equitably Implement Foreclosure Mitigation Programs. GAO-10-634. Washington, D.C.: June 24, 2010. GAO. Troubled Asset Relief Program: Home Affordable Modification Program Continues to Face Implementation Challenges. GAO-10- 556T.Washington, D.C.: March 25, 2010. Huck, P. "Home mortgage lending by applicant race: Do HMDA figures provide a distorted picture?" Housing Policy Debate, vol. 12, no. 4 (2001): 719-736. Karikari, J. "Why Homeowners' documentation went missing under the Home Affordable Mortgage Program (HAMP)?: An analysis of strategic behavior of homeowners and servicers." Journal of Housing Economics, vol. 22 (2013a): 146-162. Karikari, J. "Race/Ethnicity, Gender and Redefault of HAMP Loan Modifications," Paper presented at the annual meeting of the American Real Estate Society, Kohala Coast, HI, April 2013. Mayer, N. and M. Piven, Experience of People of Color, Women, and Low- Income Homeowners in the Home Affordable Modification Program. The Urban Institute, Washington, D.C., June 2012. National Community Reinvestment Coalition. HAMP Mortgage Modification Survey 2010 (Washington, D.C: 2010). Orlando, James. "Comparison of State Laws on Mortgage Deficiencies and Redemption Periods." OLR Research Report, No. 2010-R-0327. Hartford, Conn.: Office of Legislative Research, December 9, 2011. Accessed December 22, 2012. [hyperlink, http://www.cga.ct.gov/2010/rpt/2010-R- 0327.htm]. Voicu, I., V. Been, M. Weselcouch, and A. Tschirart. "Performance of HAMP Versus Non-HAMP Loan Modifications--Evidence from New York City." New York University Law and Economics Working Papers, No. 288 (2012). Zhang, Y. "Fair Lending Analysis of Mortgage Pricing: Does Underwriting Matter?" Journal of Real Estate Finance and Economics, vol. 46, no. 1 (2013): 131-151. [End of section] Footnotes: [1] Pub. L. No. 110-343, 122 Stat. 3765 (2008) (codified at 12 U.S.C. §§ 5201-5261). The Dodd-Frank Wall Street Reform and Consumer Protection Act, Pub. L. No. 111-203, 124 Stat. 1376 (2010): (1) reduced Treasury's authority to purchase or insure troubled assets to a maximum of $475 billion and (2) prohibited Treasury, under EESA, from incurring any additional obligations for a program or initiative unless already introduced prior to June 25, 2010. § 1302. [2] The Housing Finance Agency Innovation Fund for the Hardest Hit Housing Markets ($7.6 billion) aims to help borrowers in the areas most affected by the housing crisis, and the Department of Housing and Urban Development's Federal Housing Administration (FHA) Refinance of Borrowers in Negative Equity Positions (FHA Short Refinance) aims to help underwater borrowers. In 2013, Treasury reduced the amount obligated to the FHA Short Refinance program from $8.1 billion to $1.0 billion due, in part, to participation levels being lower than originally projected. [3] On March 9, 2012, Treasury announced the extension of the MHA program for 1 year, until December 31, 2013. On June 13, 2013, Treasury announced the extension of the MHA program for 2 years, until December 31, 2015. [4] GAO is required to report at least every 60 days on findings resulting from the oversight of, among other things, TARP's performance in meeting the purposes of the act, the financial condition and internal controls of TARP, the characteristics of both asset purchases and the disposition of assets acquired, the efficiency of TARP's operations in using appropriated funds, and TARP's compliance with applicable laws and regulations. Pub. L. No. 110-343, § 116(a), 122 Stat. 3765, 3783-85 (codified at 12 U.S.C. § 5226(a)). Under this statutory mandate, we have reported on Treasury's use of TARP funds to preserve homeownership and protect home values. See GAO, Troubled Asset Relief Program: Treasury Actions Needed to Make the Home Affordable Modification Program More Transparent and Accountable, [hyperlink, http://www.gao.gov/products/GAO-09-837] (Washington, D.C.: July 23, 2009); Troubled Asset Relief Program: Further Actions Needed to Fully and Equitably Implement Foreclosure Mitigation Programs, [hyperlink, http://www.gao.gov/products/GAO-10-634] (Washington, D.C.: June 24, 2010); Troubled Asset Relief Program: Treasury Continues to Face Implementation Challenges and Data Weaknesses in Its Making Home Affordable Program, [hyperlink, http://www.gao.gov/products/GAO-11- 288] (Washington, D.C.: Mar. 17, 2011); and Troubled Asset Relief Program: Further Actions Needed to Enhance Assessments and Transparency of Housing Programs, [hyperlink, http://www.gao.gov/products/GAO-12-783] (Washington, D.C.: July 19, 2012). [5] NeighborWorks America was established under Title VI of the Housing and Community Development Act of 1978, Pub. L. No. 95-557, § 603, 92 Stat. 2080, 2115 (codified at 42 U.S.C. § 8102), as a congressionally chartered nonprofit organization dedicated to improving distressed communities. Treasury is working with NeighborWorks to leverage the nonprofit's network of local housing counseling agencies as part of a project intended to increase awareness of MHA and assist potentially eligible homeowners with successfully applying for assistance under MHA. [6] Only financial institutions that signed a Commitment to Purchase Financial Instrument and Servicer Participation Agreement (SPA) on or before October 3, 2010, are eligible to receive TARP financial incentives under the MHA program. Treasury pays the incentives for HAMP modifications for loans not owned or guaranteed by the government- sponsored enterprises Fannie Mae or Freddie Mac (the enterprises). The enterprises bear the cost of HAMP modifications for loans they own or guarantee. [7] The NPV model compares the expected cash flow from the loan if a modification were to be made using program guidelines against the expected cash flow from the loan if no modification were to be made. If the expected cash flow with a modification is "positive" (i.e., more than the estimated cash flow of the unmodified loan), the loan servicer is required to make the loan modification. That is, if the NPV of the expected cash flow with a modification is greater than the NPV of the expected cash flow without a modification, the loan servicer is required to modify the loan. [8] Additional incentive payments are available to servicers, borrowers, and mortgage holders/investors that meet certain conditions. [9] With respect to the borrower's first-lien mortgage, the front-end, DTI ratio under HAMP is the percentage of a borrower's gross monthly income required to pay the borrower's monthly housing expense (mortgage principal, interest, taxes, insurance, and, if applicable, condominium or cooperative fees or homeowners association dues). [10] An LTV ratio for a mortgage is the ratio of the mortgage amount to the value of the home. [11] Under a deed-in-lieu of foreclosure, the homeowner voluntarily conveys all ownership interest in the home to the lender as an alternative to foreclosure proceedings. In a short sale, a homeowner sells a house rather than going into foreclosure. Proceeds from short sales are generally less than the mortgage amount, so the homeowner must have the lender's permission for the sale. Under a HAFA short sale, a lender must forgive the shortfall between the loan balance and net sales proceeds and release the lien on the subject property. Under HAFA, a deed-in-lieu must satisfy the borrower's entire mortgage obligation in addition to releasing the lien on the subject property. [12] Respectively, Pub. L. No. 90-321, tit. VII, as added Pub. L. No. 93-495, tit. V, 88 Stat. 1521 (1974) (codified at 15 U.S.C. §§ 1691- 1691f) and Pub. L. No. 90-284, tit. VIII, 82 Stat. 81 (1968) (codified at 42 U.S.C. §§ 3601-3619). [13] Pub. L. No. 90-321, 82 Stat. 146 (1968) (codified at 15 U.S.C. §§ 1601-1667e, 1671-1693r; 18 U.S.C. §§ 891-896). [14] Residential real estate-related transactions are defined by the Fair Housing Act as the making or purchasing of loans or providing other financial assistance for purchasing, constructing, improving, repairing, or maintaining a dwelling; or secured by residential real estate. Under the Fair Housing Act, residential real estate-related transactions include the selling, brokering, or appraising of residential real property. [15] 12 U.S.C. §§ 5514-15. [16] See 12 U.S.C. 1818(b); 15 U.S.C. § 1691e(k). [17] 15 U.S.C. § 1691e(g). A prudential regulator also may refer an ECOA case to DOJ when it has reason to believe that one or more creditors have violated the nondiscrimination provisions of ECOA. Id. [18] 15 U.S.C. § 1691e(k). [19] 42 U.S.C. §§ 3610(e)(2), 3614(a). [20] Exec. Order No. 13166, 65 Fed. Reg. 50121 (Aug. 16, 2000). [21] Enforcement of Title VI of the Civil Rights Act of 1964--National Origin Discrimination Against Persons With Limited English Proficiency; Policy Guidance, 65 Fed. Reg. 50123 (Aug. 16, 2000). [22] Servicers must evaluate each loan secured by an owner-occupied property that meets the eligibility requirements for a HAMP Tier 1 modification before considering a borrower for a Tier 2 modification. [23] For the purposes of HAMP Tier 2 a "rental property" is defined as a property that is used by the borrower for rental purposes only and not occupied by the borrower, whether as a principal residence, second home, vacation home, or otherwise. Treasury has established various requirements, including a certification that the borrower intends to rent the property to a tenant or tenants for at least 5 years following the effective date of any permanent modification, to determine eligibility for a HAMP Tier 2 modification. Further, to be eligible for a HAMP Tier 2 modification the borrower may have a DTI below 31 percent, but no lower than 10 percent. Under HAMP Tier 2, the borrower's postmodification monthly mortgage payment ratio (DTI ratio) must be greater than or equal to 10 percent and less than or equal to 55 percent. [24] GAO, Troubled Asset Relief Program: Treasury Actions Needed to Make the Home Affordable Modification Program More Transparent and Accountable, [hyperlink, http://www.gao.gov/products/GAO-09-837] (Washington, D.C.: July 2009). [25] CBO, Report on the Troubled Asset Relief Program-May 2013 (Washington, D.C.: May 23, 2013). [26] On June 28, 2013, NeighborWorks announced supplemental allocations totaling $500,000 to 10 state housing finance agencies, 12 HUD-approved housing counseling intermediaries, and 31 NeighborWorks organizations. These allocations provide funding for participating agencies targeting populations that are potentially eligible for MHA, but require culturally and linguistically specialized services, are underrepresented by mortgage modifications, or are disproportionately impacted by predatory lending or unemployment. [27] According to NeighborWorks, it provided funding to agencies to support outreach activities designed to reach homeowners in need of mortgage assistance. The outreach funds allocated were the greater of $5,000 or 25 percent of the document preparation and submission fees. Further, HUD intermediaries and state housing finance agencies are provided with additional allocations to cover costs of quality control, oversight and management of funds, and training. [28] Counseling agencies participating in the National Foreclosure Mitigation Counseling Program were invited to apply to participate in the MHA Outreach and Borrower Intake Project. NeighborWorks stated that agencies that chose to apply to participate in the project were required to submit an outreach plan and budget. NeighborWorks accepted all but two of the counseling agencies that had applied. Participating agencies were then allocated packages based on their National Foreclosure and Mitigation Counseling (NFMC) grant award. [29] The first Production Dashboard was delivered to the agencies in October 2013. [30] Prudential regulators conduct examinations of the depository institutions they oversee to assess their fair lending compliance, including determining whether lenders have violated ECOA or the Fair Housing Act. Those examinations may include an assessment of lenders' fair lending compliance specifically in the mortgage servicing area. [31] In November 2009, the President created the FFETF, a multiagency effort to combat a range of financial crimes relating to the recent financial crisis and recovery efforts. FFETF is composed of more than 20 federal agencies, 94 U.S. Attorneys' Offices, and state and local partners. The task force established the Non-Discrimination Working Group, which focuses on discrimination in the housing and finance markets. The working group is co-chaired by senior officials of DOJ, HUD, the Federal Reserve, and the National Association of Attorneys General. [32] GAO, Troubled Asset Relief Program: Home Affordable Modification Program Continues to Face Implementation Challenges, [hyperlink, http://www.gao.gov/products/GAO-10-556T] (Washington, D.C.: Mar. 25, 2010). [33] GAO, Troubled Asset Relief Program: Further Actions Needed to Fully and Equitably Implement Foreclosure Mitigation Programs, [hyperlink, http://www.gao.gov/products/GAO-10-634] (Washington, D.C.: June 24, 2010). [34] [hyperlink, http://www.gao.gov/products/GAO-10-634]. One of the eligibility requirements for HAMP Tier 1 is that the borrower's first- lien mortgage payment must be more than 31 percent of the homeowner's gross monthly income. Therefore, if the servicer determines that the borrower's debt-to-income (DTI) ratio is already less than 31 percent, then the servicer would deem the borrower ineligible for HAMP Tier 1. [35] Supplemental Directive 09-02, Home Affordable Modification Program - Fair Housing Obligations under the Home Affordable Modification Program, April 21, 2009. [36] Treasury provides data to FRB, OCC, FDIC, HUD, and the Federal Housing Finance Agency. As of November 2013, Treasury officials told us they were working to provide the same data file to CFPB as they provide to other federal agencies and with the same frequency. [37] See appendix II for additional information on the description of the data we used, details of the regression analysis, previous relevant studies, and results of our analysis. [38] Our econometric analysis focused on certain protected classes as defined by the Equal Credit Opportunity Act and Fair Housing Act (which we refer throughout the report as "fair lending populations")-- women, non-Hispanic African-Americans, Hispanics, non-Hispanic Asians, and non-Hispanic American Indians or Alaska Natives, and Native Hawaiians or Other Pacific Islanders--as well as low-and moderate- income groups and persons living in substantially minority neighborhoods. We define substantially minority neighborhoods as census tracts where the proportion of all minority households (based on race and ethnicity) was at least 30 percent. [39] We also used several factors that are typically used to capture the creditworthiness of borrowers in mortgage markets, such as delinquency status of borrowers before the modification, Fair Isaac Corporation (FICO) credit scores at modification, DTI for both the front-end before modification and back-end after modification, and the LTV of the property at origination and at the time of the modification. [40] Our previous work on MHA found that Treasury's HAMP database was missing a significant amount of information on borrowers' race and ethnicity, resulting in an inability to assess whether HAMP is being fairly implemented across servicers. As of September 30, 2010, race and ethnicity information was not available for 65 percent of nonenterprise borrowers in active trial modifications. See GAO, Troubled Asset Relief Program: Treasury Continues to Face Implementation Challenges and Data Weaknesses in Its Making Home Affordable Program, [hyperlink, http://www.gao.gov/products/GAO-11-288] (Washington, D.C.: Mar. 17, 2011). For this analysis we used data through April 17, 2013, and fewer race and ethnicity data were missing. While servicers are required to request race and ethnicity information from borrowers, borrowers have the option of providing it. [41] Department of Justice, Office of the Attorney General, Federal Government's Renewed Commitment to Language Access Obligations Under Executive Order 13166, Memorandum to the Heads of Federal Agencies, General Counsels, and Civil Rights Heads (Washington, D.C.: Feb. 17, 2011). [42] Department of Justice, Federal Coordination and Compliance Section, Civil Rights Division, Language Access Assessment and Planning Tool for Federally Conducted and Federally Assisted Programs (May 2011). [43] Supplemental Directive 11-04, Making Home Affordable Program, Single Point of Contact for Borrower Assistance (May 18, 2011) and Section 4.2 of Chapter I of the MHA Hand book for Servicers of Non-GSE Mortgages, version 4.3 (Sept. 16, 2013). [44] National Housing Resource Center, National Mortgage Settlement Servicing Standards and Noncompliance: Results of a National Housing Counselor Survey (Philadelphia, Pa., 2013). California Reinvestment Coalition, Chasm Between Words and Deeds IX: Bank Violations Hurt Hardest Hit Communities (2013). [45] CBO, Report on the Troubled Asset Relief Program-May 2013, (Washington, D.C.: May 23, 2013). [46] The consent orders entered into in the National Mortgage Settlement require servicers to provide borrowers with a single point of contact as well as improved communication, cease charging improper fees, and have adequate staffing levels and training, among other things. United States v. Bank of America Corp., No. 1: 12-CV-00361 (D.D.C. Apr. 4, 2012). [47] There were two broad categories of questions in the surveys. First, with respect to each of the five servicers covered by the National Mortgage Settlement (NMS), survey respondents were asked how often they see each of a number of potential violations of NMS servicing standards. Second, respondents were asked how often they see various fair lending violations, such as disparate treatment of borrowers of color, borrowers with limited English proficiency, and borrowers with disabilities. See National Housing Resource Center, National Mortgage Settlement Servicing Standards and Noncompliance: Results of a National Housing Counselor Survey (Philadelphia, Pa, 2013) and California Reinvestment Coalition, Chasm Between Words and Deeds IX: Bank Violations Hurt Hardest Hit Communities (2013). [48] This econometric analysis focused on the protected classes covered by the Equal Credit Opportunity Act, 15 U.S.C. §§ 1691-1691f, and Fair Housing Act, 42 U.S.C. §§ 3601-3619, (which we refer to throughout the report as "fair lending populations")--women, non- Hispanic African-Americans, Hispanics, non-Hispanic Asians, non- Hispanic American Indians or Alaska Natives, and Native Hawaiians or Other Pacific Islanders (12 C.F.R. § 1002.13)--as well as low-and moderate-income groups, and persons living in substantially minority neighborhoods. We define substantially minority neighborhoods as census tracts where the proportion of all minority households (based on race and ethnicity) is at least 30 percent. [49] We have previously found instances where servicers have variations in their application of HAMP guidelines. For example, in 2010 we reported that servicers have inconsistent practices on evaluating borrowers for imminent default because Treasury has not provided specific guidance on how to evaluate nonenterprise borrowers for imminent default.See GAO, Troubled Asset Relief Program: Home Affordable Modification Program Continues to Face Implementation Challenges, [hyperlink, http://www.gao.gov/products/GAO-10-556T] (Washington, D.C.: Mar. 25, 2010). Servicers, and their employees, may also make errors in applying HAMP policies to modifications. In 2010 we reported that 5 of the 10 servicers we contacted reported at least a 20 percent error rate for income calculations. We noted that without accurate income calculations, which are key in determining the borrower's debt-to-income (DTI) ratio, similarly situated borrowers applying for HAMP may be inequitably evaluated for the program and may be inappropriately deemed eligible or ineligible for HAMP. See GAO, Troubled Asset Relief Program: Further Actions Needed to Fully and Equitably Implement Foreclosure Mitigation Programs, [hyperlink, http://www.gao.gov/products/GAO-10-634] (Washington, D.C.: June 24, 2010). Treasury also assesses servicers on their income calculations and tracks the percentage of loans for which MHA-C's income calculation differs from the servicer's on a quarterly basis. In their July 2013 assessment results, Treasury noted an average income error rate of less than 2 percent, down from an average of about 7.5 percent in July 2012. [50] The criteria also include the following: the mortgage is a first- lien loan originated prior to January 1, 2009, the borrower's front- end housing DTI ratio exceeds 31 percent, and the unpaid principal balance for a one-unit dwelling does not exceed $729,750. [51] Reasons for trial failures that occurred prior to December 1, 2009, were reported only at the discretion of the participating servicer, therefore reason codes for trial denials and cancellations may not be available prior to December 1, 2009. [52] We also excluded loans owned or guaranteed by the enterprises that were modified under a pilot program on principal forgiveness (about 240 loans) conducted by Fannie Mae. Servicers of loans not insured by FHA, VA, and U.S. Department of Agriculture (USDA) undertake modifications based on HAMP guidelines, while servicers of loans owned or guaranteed by the enterprises use guidelines from its regulator, the Federal Housing Finance Agency (FHFA). [53] The census data were obtained from Minnesota Population Center, National Historical Geographic Information System: Version 2.0 (University of Minnesota: Minneapolis, Minn.: 2011). [54] About 11 percent of the loans not approved/not accepted contain NPV data. [55] Mayer and Piven used a similar approach of separating the denials of applications and cancellations of trial modifications due to ineligibility from the other reasons for denials and cancellations. See N. Mayer, and M. Piven, Experience of People of Color, Women, and Low-Income Homeowners in the Home Affordable Modification Program (Washington, D.C.: The Urban Institute, June 2012). [56] Specifically, as done in previous studies, we include dummy variables for whether or not a state is a judicial foreclosure state, is a foreclosure nonrecourse state, has foreclosure rights of redemption, or has foreclosure prevention programs. We relied on previous studies to identify the classifications for each state. See J. Collins, K. Lam, and C. Herbert, "State Mortgage Foreclosure Policies & Lender Interventions: Impacts on Borrower Behavior in Default," Journal of Policy Analysis and Management, vol. 30, no. 2 (2011): 216-232, and James Orlando, "Comparison of State Laws on Mortgage Deficiencies and Redemption Periods," OLR Research Report, No. 2010-R-0327 (Office of Legislative Research, Hartford, Conn.: Dec. 9, 2011), accessed December 22, 2012, [hyperlink, http://www.cga.ct.gov/2010/rpt/2010-R-0327.htm]. [57] Gender is based on the (primary) borrower; co-borrower information is not used in this study because of data limitations. [58] We classified the income groups as follows: low-income, if borrower's income is less than 50 percent of Metropolitan Statistical Area (MSA) median household income; moderate-income, if income equals to or is greater than 50 percent and less than 80 percent of MSA median household income; middle-income, if income equals to or is greater than 80 percent and less than 120 percent of MSA median household income; and high-income, if income greater than 120 percent of MSA median household income. [59] See, for example, Mayer and Piven and CRC. Missing race/ethnicity data in HAMP are expected to be high, compared to the data collected under the Home Mortgage Disclosure Act of 1975 (HMDA)(12 U.S.C. §§ 2801-2810) and the act's implementing regulation, Regulation C, which requires banks to collect data at origination. Information on the ethnicity, race, and sex of the borrower or applicant must be requested by the lender, including applications made entirely by telephone, mail, or Internet. 12 C.F.R. § 1003, App. B.I and II.A. If the information is not provided by the applicant and if the application is submitted in person, the lender is required to note the information on the basis of visual observation or surname. Id. at II.B. California Reinvestment Coalition (CRC), Race to the Bottom: An Analysis of HAMP Loan Modification Outcomes By Race and Ethnicity for California (July 2011). [60] For example, Dietrich and Huck, who argued that missing race data in HMDA are systemically lost and introduce bias and efficiency problems into fair lending examinations. J. Dietrich, Missing Race Data in HMDA and the Implications for the Monitoring of Fair Lending Compliance, Office of the Comptroller of the Currency, OCC Economics, Working Paper No. 2001-1 (Washington, D.C.: March 2011) and Huck, P, "Home mortgage lending by applicant race: Do HMDA figures provide a distorted picture?" Housing Policy Debate, vol. 12, no. 4 (2001): 719- 736. [61] Using categorical format helps to avoid the exclusion of variables with missing observations. [62] Previous studies that are relevant to ours because they analyzed fair lending using HAMP data include the following: CRC, National Community Reinvestment Coalition, HAMP Mortgage Modification Survey 2010 (Washington, D.C.: 2010), J. Karikari, "Race/Ethnicity, Gender and Redefault of HAMP Loan Modifications," Paper presented at the annual meeting of the American Real Estate Society, Kohala Coast, HI, April 2013, Mayer and Piven, and, I. Voicu, V. Been, M. Weselcouch, and A. Tschirart, "Performance of HAMP Versus Non-HAMP Loan Modifications--Evidence from New York City," New York University Law and Economics Working Papers, No. 288 (2012). Other relevant studies on loan modifications include S. Agarwal, S., G. Amromin, I. Ben- David, S. Chomsisengphet, T. Piskorski, and A. Seru, "Policy Intervention in Debt Renegotiation: Evidence from the Home Affordable Modification Program," NBER Working Paper Series, No. 18311 (August 2012). [63] This technique is commonly used to estimate discrete choice models. The approach we use here is appropriate for our models of default of trial modifications and redefault of permanent modifications since we are primarily interested in the probability that a modified loan defaults or redefaults within a certain time period and not the hazard rate of the outcome (i.e., the probability that a loan defaults at a certain time if it has not already defaulted up to that time). See, for example, Agarwal et al., "Policy Intervention in Debt Renegotiation: Evidence from the Home Affordable Modification Program," [hyperlink, http://www.gao.gov/products/GAO-12-296], and Karikari, "Race/Ethnicity, Gender and Redefault of HAMP Loan Modifications" for similar approaches. Our results of redefault were unchanged when we extended the analysis from 12 to 24 months since the permanent modification. [64] In particular, the usable data for the regression analysis of denial of applications lacked several variables for applicants who were found to be ineligible and their loans were not evaluated for the NPV test. [65] The denial rates were about 2 percent for borrowers with ineligible mortgages or who were found to not be in imminent danger of default. [66] The data do not allow us to exclude those borrowers that withdrew their request for modification or did not accept an approved offer. It could be argued that not accepting an approved offer should not be regarded as a cancellation by the servicer. However, our purpose was to examine if there were disparities for this outcome since the nature of the offer could be a reason why the borrower did not accept it. According to HAMP guidelines, failure of a borrower to make the first trial period payment in a timely manner is considered nonacceptance of the trial period plan, which could imply a different motivation for this outcome. [67] A study, J. Karikari, "Why Homeowners' documentation went missing under the Home Affordable Mortgage Program (HAMP)?: An analysis of strategic behavior of homeowners and servicers," Journal of Housing Economics, vol. 22 (2013): 146-162, found that actions by both servicers and homeowners are consistent with missing documentation. Servicers have an incentive to "lose" the documentation of borrowers with low credit risks in order to "steer" them away from HAMP to their own (proprietary) less costly modification programs. At the same time, borrowers with high risks have less incentives or are unable to provide complete documentation to support the reason for their "hardships," as well as having difficulty in fulfilling the HAMP requirements. [68] The denial rates are 8 percent for borrowers not accepting their approved offers and 3 percent for excessive forbearance. [69] The area income groups are identified as follows: low-income area, if the tract-level household median income is less than 50 percent of MSA-level household median income; moderate-income area, if the tract-level household income equals to or is greater than 50 percent and less than 80 percent of MSA-level household median income; middle-income area, if tract-level household median income equals or is greater than 80 percent and less than 120 percent of MSA-level household median income; high-income area, if tract-level household median income is greater than 120 percent of MSA-level household median income, and not available, if tract-level or MSA-level household median income is unavailable. [End of section] GAO's Mission: The Government Accountability Office, the audit, evaluation, and investigative arm of Congress, exists to support Congress in meeting its constitutional responsibilities and to help improve the performance and accountability of the federal government for the American people. GAO examines the use of public funds; evaluates federal programs and policies; and provides analyses, recommendations, and other assistance to help Congress make informed oversight, policy, and funding decisions. GAO's commitment to good government is reflected in its core values of accountability, integrity, and reliability. Obtaining Copies of GAO Reports and Testimony: The fastest and easiest way to obtain copies of GAO documents at no cost is through GAO's website [hyperlink, http://www.gao.gov]. Each weekday afternoon, GAO posts on its website newly released reports, testimony, and correspondence. To have GAO e-mail you a list of newly posted products, go to [hyperlink, http://www.gao.gov] and select “E-mail Updates.” Order by Phone: The price of each GAO publication reflects GAO's actual cost of production and distribution and depends on the number of pages in the publication and whether the publication is printed in color or black and white. Pricing and ordering information is posted on GAO's website, [hyperlink, http://www.gao.gov/ordering.htm]. Place orders by calling (202) 512-6000, toll free (866) 801-7077, or TDD (202) 512-2537. Orders may be paid for using American Express, Discover Card, MasterCard, Visa, check, or money order. Call for additional information. Connect with GAO: Connect with GAO on facebook, flickr, twitter, and YouTube. Subscribe to our RSS Feeds or E mail Updates. Listen to our Podcasts. Visit GAO on the web at [hyperlink, http://www.gao.gov]. To Report Fraud, Waste, and Abuse in Federal Programs: Contact: Website: [hyperlink, http://www.gao.gov/fraudnet/fraudnet.htm]; E-mail: fraudnet@gao.gov; Automated answering system: (800) 424-5454 or (202) 512-7470. Congressional Relations: Katherine Siggerud, Managing Director, siggerudk@gao.gov: (202) 512-4400: U.S. Government Accountability Office: 441 G Street NW, Room 7125: Washington, DC 20548. Public Affairs: Chuck Young, Managing Director, youngc1@gao.gov: (202) 512-4800: U.S. Government Accountability Office: 441 G Street NW, Room 7149: Washington, DC 20548. [End of document]