|United States General Accounting Office|
|Subcommittee on Housing and Community Opportunity, Committee on Banking and Financial Services, House of Representatives|
2 p.m. EDT
April 30, 1998
|Weaknesses in HUD's Management and Oversight of the Title I Program|
of Stanley J. Czerwinski, Associate Director
Housing and Community Development Issues,
Resources, Community, and Economic
Mr. Chairman and Members of the Subcommittee:
We are pleased to be here today to discuss the preliminary results of our assessment of certain aspects of the Department of Housing and Urban Development's (HUD) management and oversight of its loan insurance program for home improvements under Title I of the National Housing Act. As you know, we are conducting this assessment at the request of this Subcommittee and Congressman Kenneth E. Bentsen, Jr., and plan to report the results of our assessment this summer.
Homeowners who have little equity in their home, at times obtain Title I property improvement loans to make alterations or repairs to their homes. These loans are made by banks and other private lenders from their own funds and are insured by HUD's Federal Housing Administration (FHA). If borrowers default on their loans, banks submit claims to HUD, which approves or denies them.
Concerned about how well this Title I program was being operated, you and Congressman Bentsen asked us to determine (1) the extent to which the information needed to manage the program was available to HUD, (2) the extent to which HUD was overseeing program lenders, and (3) whether HUD has any ongoing or planned efforts under way to strengthen its management and oversight. To meet these objectives, we reviewed, among other things, HUD's Title I program regulations and guidelines, 16,556 claims paid by HUD between January 1994 and August 1997 to identify if any multiple claims were submitted on the same borrower for the same property, documents from a random sample of 53 loan claims paid to lenders by HUD from fiscal years 1995 through July 1997,1 and interviewed HUD officials in Washington, D.C., and Albany, New York--the location of the programs' claims examination unit.
In summary, our preliminary analysis shows that HUD is not collecting the information needed for managing the program. Specifically, we found that HUD collects little information when loans are made on program borrowers, properties, and loan terms, such as the borrower's income and the address of the property being improved. Moreover, HUD does not maintain information on why it denies loan claims or why it subsequently approves some for payment.
HUD also provides limited oversight of lenders' compliance with program regulations, conducting only 2 on-site lender reviews in fiscal year 1997 of the approximately 3,700 program lenders. Regarding the need for oversight of lenders' compliance, we found that loan claim files submitted by lenders to HUD following loan defaults often do not contain required loan documents, including the original loan applications and certifications signed by the borrower that the property improvement work has been completed. In addition, some claims were paid by HUD even though there were indications that lenders did not comply with required underwriting standards when insuring the loan.
As a result of the management and oversight weaknesses we have observed, our preliminary work indicates that HUD does not know who the program is serving, if lenders are complying with program regulations, and whether certain potential program abuses are occurring, such as violations of the $25,000 limitation on the amount of Title I loan indebtedness for each property. HUD officials attributed these weaknesses to the program's being lender-operated, limited staff resources, and HUD's assignment of monitoring priorities.
Under the HUD 2020 Management Reform Plan and related efforts, HUD is making significant changes in all of its single-family housing programs, including the Title I property improvement program. Theses changes are motivated in part by HUD's goals to downsize the agency and address long-standing agencywide management weaknesses. We are assessing the extent to which these changes may affect the management and oversight weaknesses we identified.
Before I discuss these issues in greater detail, let me briefly explain how FHA's Title I property improvement insurance program operates.
The Title I property improvement program was established by the National Housing Act (12 U.S.C. 1703) to encourage lending institutions to finance property improvement projects that would preserve the nation's existing housing stock. Under the program, FHA insures 90 percent of a lender's claimable loss on an individual defaulted loan. The total amount of claims that can be paid to a lender is limited to 10 percent of the value of the total program loans held by each lender. Today, the value of Title I's outstanding loans is relatively small compared with other FHA housing insurance programs. As of September 30, 1997, the value of loans outstanding on the property improvement program totaled about $4.4 billion on 364,423 loans. By contrast, the value of outstanding FHA single-family loans in its Mutual Mortgage Insurance Fund totaled about $360 billion. Similarly, Title I's share of the owner-occupied, single-family remodeling market is small--estimated by the National Association of Home Builders to be about 1 percent in fiscal year 1997.
Approximately 3,700 lenders are approved by FHA to make Title I loans. Lenders are responsible for managing many aspects of the program, including making and servicing loans, monitoring the contractors, and dealing with borrowers' complaints. In conducting these activities, lenders are responsible for complying with FHA's underwriting standards and regulations and ensuring that home improvement work is inspected and completed. FHA is responsible for approving lenders, monitoring their operations, and reviewing the claims submitted for defaulted loans. Title I program officials consider lenders to have sole responsibility for program operations and HUD's role is primarily to oversee lenders and ensure that claims paid on defaulted loans are proper.
Homeowners obtain property improvement loans by applying directly to Title I lenders or by having a Title I lender-approved dealer--that is a contractor--prepare a credit application or otherwise assist the homeowner in obtaining the loan from the lender. During fiscal years 1986 through 1996, about 520,000 direct and 383,000 dealer loans were made under the program. By statute, the maximum size of property improvement loans is $25,000 for single-family loans and the maximum loan term is about 20 years. Title I regulations require borrowers to have an income adequate to meet the periodic payments required by a property improvement loan. Most borrowers have low- to moderate incomes, little equity in their homes, and/or poor credit histories.
HUD's expenses under the Title I program, such as claim payments made by FHA on defaulted loans, are financed from three sources of revenue: (1) insurance charges to lenders of 0.5 percent of the original loan amount for each year the loan is outstanding, (2) funds recovered from borrowers who defaulted on loans, and (3) appropriations. In an August 1997 report on the Title I program, Price Waterhouse concluded that the program was underfunded during fiscal years 1990 through 1996. Price Waterhouse estimated that a net funding deficit of about $150 million occurred during the period, with a net funding deficit in 1996 of $11 million.2 Data from the Price Waterhouse report on estimated projected termination rates for program loans made in fiscal year 1996 can be used to calculate an estimated cumulative claim rate of about 10 percent over the life of Title I loans insured by FHA in that fiscal year.
INFORMATION NEEDED TO MANAGE THE PROGRAM NOT COLLECTED BY HUD
When FHA-approved Title I lenders make program loans, they collect information on borrowers, such as age, income, and gender; the property, such as its address; and loan terms, such as interest rate. While lenders are required to report much of this information to their respective regulatory agencies by the Home Mortgage Disclosure Act, HUD collects little of this information when Title I loans are made. Using information that it requires lenders to provide, HUD records the lender's and borrower's names, state and county, as well as the size, term, and purpose of the loan. Other information collected by HUD on other single-family loan insurance programs, such as the borrower's address, Social Security number, income, and debt are not collected by HUD when Title I loans are made. HUD does collect all of the information available on borrowers, property, and loans when Title I loans default and lenders submit claims. Title I officials told us they collected little information when loans were made because they consider the program to be lender-operated.
As a result, HUD cannot identify the characteristics of borrowers and neighborhoods served by the program, nor can it identify certain potential abuses of the program. For example, HUD does not collect borrowers' Social Security numbers and property addresses when loans are made. Therefore, HUD would have difficulty determining if some borrowers are obtaining multiple Title I loans or if some borrowers are exceeding the maximum amount of Title I loans per property when loans are made. HUD regulations limit the total amount of indebtedness on Title I loans to $25,000 for each single-family property.
In this regard, our examination of HUD's Title I claims data found a number of instances in which the same Social Security number was used for multiple claims. As discussed previously, claims on about 10 percent of the program's loans can be expected over the life of program loans. Our examination of 16,556 claims paid by HUD between January 1994 and August 1997 revealed 247 instances in which the same Social Security number appeared on multiple claims. These cases totaled about $5.2 million in paid claims. In several instances, claims were paid on as many as five loans having the same Social Security number during the 3-1/2-year period. Our Office of Special Investigations, together with HUD's Office of the Inspector General, is inquiring further into the circumstances surrounding these loans. However, because these loans may have been for multiple properties, or multiple loans on the same property that totaled less than $25,000, they may not have violated program regulations. Allowing individual borrowers to accumulate large amounts of Title I HUD insured debt, however, exposes HUD to large losses in the case of financial stress on the part of such heavily indebted borrowers. In addition, while information available to HUD allows identification of potential abuses of the $25,000 indebtedness limit after loans have defaulted, control over the indebtedness limitation is not possible for 90 percent of the program's loans made that do not default because borrowers' Social Security numbers and property addresses are not collected when the loans are made.
Information on Types of Loans for Which Claims Submitted Not Always Accurate
While HUD collects more extensive information on program loans when they default, we found problems with the accuracy of some of the information recorded in its claims database. Our random sample of 53 loans on which a claim had been denied and subsequently paid by HUD, found that 7 loans, or 13 percent, had been miscoded as dealer loans when they were direct loans, or direct loans when they were dealer loans. This is important because HUD recently cited high default rates on dealer loans, among other reasons, for proposing regulations to eliminate the dealer loan portion of the program. Considering the miscoding on identifying loans as dealer or direct, we question HUD's ability to identify default experience by loan type.
Information on Why Claims Denied and Subsequently Approved Not Maintained
In addition, HUD's information on claims denied and subsequently approved was problematic. Although HUD can deny claims for property improvement loans for a number of reasons, HUD did not have a system in place to provide information on why claims are denied or approved for payment following a denial. HUD could not provide us with information on how many claims it denied because of poor underwriting or other program abuses or which lenders had a higher-than-average number of claims denied for specific program violations. In addition, we were unable to determine from HUD's data system why a denied claim was subsequently paid following an appeal by the lender or waiver by HUD. Such information is important in determining how well lenders are complying with program regulations, whether internal controls need to be strengthened, and which lenders should be targeted for review by HUD's Office of Quality Assurance.
We also found that files for claims that were initially denied by HUD and subsequently paid frequently did not contain the names of program officials who decided the denied claims should be paid and the reasons for their decisions. Of the 53 randomly selected loan claim files we examined, 50 contained no evidence of further review by a HUD official following the initial denial or provided any basis for eventually paying the claim. Unless information on who makes decisions to deny claims and the reasons for the denial and subsequent payments are documented, HUD has no basis for reviewing the reasonableness of those decisions.
HUD recently made changes to its claims database system to identify the reasons claims are denied. Program officials agreed that such information is important in determining how well program regulations are being complied with and in targeting lenders for quality assurance reviews. Claims examiners are now required to identify their reasons for denial, including the section of the regulation that was violated. However, the change does not address the problem of missing documentation in the claims file explaining the reasons for paying claims that were previously denied.
LIMITED OVERSIGHT BY HUD OF PROGRAM LENDERS
HUD's monitoring reviews of Title I lenders to identify compliance problems have declined substantially in recent years. Between fiscal years 1995 and 1997, HUD performed 33 Title I on-site quality assurance reviews of lenders. Most of these reviews (26) were performed in fiscal year 1995. During fiscal years 1996 and 1997, HUD performed five and two on-site lender reviews, respectively. According to HUD officials, prior to fiscal year 1997, HUD had a staff of 23 individuals to monitor the 3,700 lenders approved by FHA to make Title I loans and about 8,000 other FHA approved lenders making loans on other FHA insurance programs. Because of this limited monitoring resource, HUD decided to focus its lender monitoring on major high volume FHA programs, according to these HUD officials. Monitoring priorities have also led to few follow-up reviews by HUD. As a result, it is difficult to determine the impact of the quality assurance reviews that were performed on improving lenders' compliance.
Required Documents Missing from Loan Files
When making Title I loans, lenders are required to ensure that borrowers represent acceptable credit risks, with a reasonable ability to make payments on the loans, and to see that the property improvement work is completed. However, our examination of 53 loan claim files revealed that one or more required documents needed to ensure program compliance were missing from more than half (30) of the files.
In 12 cases, the required original loan application, signed by the borrower, was not in the loan file. The original loan application is important because it is used by the claims examiner to review the adequacy of the lender's underwriting and to ensure that the borrower's signature and Social Security number matches those on other documents, including the credit report. Furthermore, for 23 of the 53 claim files, we found that required completion certificates, certifying that the property improvement work had been completed, were missing or were signed but not dated by the borrowers. According to program guidelines, claims submitted for payment after defaults have occurred on dealer loans should not be paid unless a signed completion certificate is in the file. We found that completion certificates were missing from the files for 13 dealer loans and were not dated for another 4 dealer loans. Lastly, for 33 loans on which program regulations required that an inspection be conducted by the lender, 18 loan files did not contain the report.
Lenders Not Always Complying With Program Regulations
We also reviewed the 53 claim files to determine how well lenders were complying with underwriting standards. All documentation supporting the underwriting determination should be retained in the loan file, according to HUD regulations. HUD can deny a lender's claim if the lender has not followed HUD underwriting standards in making the loan. However, HUD does not examine the quality of a lender's loan underwriting during the claims process if 12 loan payments were made by the borrower before defaulting on the loan. Since 27 percent of the Title I loans that default do so within the first year, this practice, in effect, exempts the majority of defaulted loans from an examination of the quality of the lenders' underwriting. Of the 53 loans in our sample, 13 defaulted within 12 months of loan origination and were subject to an underwriting review by HUD. We focused our underwriting examination on these 13 loan claim files.
We found that for 4 of the 13 loans, on which HUD eventually paid claims, lenders made questionable underwriting decisions. Title I program regulations require that the credit application and review by the lender must establish that the borrower, is an acceptable credit risk, had 2 years of stable employment, and that his/her income will be adequate to meet the periodic payments required by the loan, as well as the borrower's other housing expenses and recurring charges. However, for four of these loans, information in the files indicated that the borrowers may not have had sufficient income to qualify for the loan or had poor credit. For example, on one loan, the lender used a pay stub covering the first 2 weeks of March to calculate the borrower's annual income. The pay stub showed that the borrower's year-to-date earnings were $6,700 by the middle of March, and this amount was used to calculate that his annual income was $34,000, or about $2,800 per month. However, the pay stub also showed that for the 2-week period in March, the borrower worked a full week with overtime and only earned $725, or about $1,600 per month. The file contained no other documentation, such as income tax returns, W-2 forms, or verification from the employer to support the higher monthly income. Program officials told us that it was acceptable to use one pay stub to calculate monthly income; however, the "yearly earnings to date" figure should not be used because it can at times inflate the actual income earned during a normal pay period. The borrower, with about $1,600 per month in corrected income, still met HUD's income requirements for the amount of the loan. However, HUD denied the original claim because its underwriting standards had not been followed in that the borrower had poor credit at the time the loan was made. In a letter responding to HUD's denial of its claim, the lender acknowledged that the borrower had limited credit at the time the loan was made, but pointed out the (mis-calculated) higher income of $2,800 per month to justify making the loan. This reasoning was apparently accepted by HUD as there was no evidence in the claim file that HUD questioned the error in calculating the borrower's monthly income. The borrower defaulted on the loan after making two payments, and HUD paid a claim of $14,000.
Similar problems with lenders' noncompliance with Title I program regulations have been identified by HUD. As noted previously, between fiscal years 1995 and 1997, HUD performed 33 Title I on-site quality assurance reviews of lenders. Among other things, HUD cited lenders for engaging in poor credit underwriting practices and having loan files with missing inspection reports or inspection reports that were not signed or dated. HUD sent the lenders letters detailing its findings and requested a written response addressing the findings. HUD, however, did not perform follow-up, on-site reviews on 32 lenders to ensure that they had taken corrective actions. For the 33 on-site reviews, nine lenders were referred to HUD's Mortgagee Review Board for further action. The Board assessed four of these lenders a total of $23,500 in civil penalties.
RECENT AND PROPOSED CHANGES TO THE TITLE I PROGRAM
Under its HUD 2020 Management Reform Plan and related efforts, HUD has been making changes to the Title I program operations. HUD has relocated its claims examination unit to the Albany (New York) Financial Operations Center and contracted with Price Waterhouse to develop claims examination guidelines. According to program officials in Albany, the new claims process will be more streamlined and automated and include lenders filing claims electronically. In addition, HUD is consolidating all single-family housing operations from 81 locations across the nation into four Single-Family Homeownership Centers. Each center has established a quality assurance division to (1) monitor lenders, (2) recommend sanctions against lenders and other program participants such as contractors and loan officers, (3) issue limited denials of program participation against program participants, and (4) refer lenders for audits/investigations. However, since HUD's quality assurance staff will monitor lenders involved in all FHA single-family programs, the impact of this change on improving HUD's oversight of Title 1 lenders is unclear. Overall, by the end of fiscal year 1998, the quality assurance staff will increase to 76, up from 43 in February 1998. HUD expects that the addition of more quality assurance staff will increase the number of reviews of lenders and allow more comprehensive reviews of lender operations.
_ _ _ _
In closing, Mr. Chairman, our preliminary analysis shows weaknesses in HUD's management of its Title I property improvement loan insurance program and oversight of program lenders. These weaknesses center on the absence of information needed to manage the program and HUD's oversight of lenders' compliance with program regulations. HUD officials attributed these weaknesses to the program's being lender-operated, limited staff resources, and HUD's assignment of monitoring priorities. Because of these weaknesses, we are concerned that HUD may have little assurance that the property improvement program is operating efficiently and free of abuse. The challenge faced by HUD in managing and overseeing this program centers on how to obtain the information needed to manage the program and to strengthen the oversight of lenders for this program, which is relatively small compared with other FHA housing insurance programs. Our report will include any recommendations or options we have to offer to strengthen HUD's management and oversight of the program.
Mr. Chairman, this concludes my statement. We would be pleased to respond to any questions that you or Members of the Subcommittee may have.
1We sampled from the 5,646 program claims that were originally denied and then paid by HUD during the 3-year period.
2Price Waterhouse defined the net funding position as the current value of the premiums collected minus the current value of the claims. Current value refers to past payments plus accumulated interest, plus expected future payments discounted by the interest rate on 5-year Treasury bonds. The estimated negative net funding deficit implies that premiums will be insufficient to pay the expected claims.