ROAD TO FHA REFORM: BACK TO BASICS

PRESENTED AT THE 2013 AMERICAN MORTGAGE CONFERENCE SPONSORED BY THE NORTH CAROLINA BANKERS ASSOCIATION

Edward Pinto, Resident Fellow American Enterprise Institute

September 10, 2013
The views expressed here are those of the author alone and do not necessarily represent those of the American Enterprise Institute.

nightmareatfha.com

|1

FHA LENDING RISKIER THAN EVER

VA AND FHA DELINQUENCY RATES
10%

1946-1967
1979-2000 2001-2012

the FHAs rate averaged 96%of the VAs the FHAs rate averaged 118% of the VAs the FHAs rate averaged 197% of the VAs

9% 8% 7% 6% 5% 4% 3% 2% 1% 0%

VA serious delinquency FHA serious delinquency

Sources: 1946-1967: John P. Herzog and James S. Earley, Home Mortgage Delinquency and Foreclosure (Cambridge, MA: National Bureau of Economic Research, 1970), www.nber.org/books/herz70-1 and 1979-2012: MBA National Delinquency Survey. All data year-end, except 2012 data, which is Q2:2012.

1946 1949 1952 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012 nightmareatfha.com

|2

FHA LENDING MAKING FORECLOSURES COMMONPLACE

WEIGHTED AVERAGE FAMILY FORECLOSURE CLAIM RATE OF 12.54% FOR 1975-2011--3.14 MILLION FORECLOSURES AND 1 IN 8 FAMILIES
1,800,000 1,600,000 1,400,000 25% 1,200,000 35% 30%

FHA projected cumulative claim ratenote: annual claim rates do not exclude streamline/FHA-toFHA refinances (right axis)
FHA adjusted loan count (excludes streamline/FHA-FHA refi from 1983 on) left axis

1,000,000
800,000 600,000

20%

Sources:
15%

Loan count: HUD PD&R historical data

10% 400,000 200,000 0 5%

Projected annual cumulative claim rate and streamline/FHA-to-FHA refinances: Annual FHA Actuarial Studies
Number of claims by year = loan count (includes streamline/FHA-to-FHA refinances) x claim rate

0%

nightmareatfha.com

|3

FHA LENDING MISSION FAILURE

COMMON CRITIQUES OF FHA OVER THE YEARS:

Excessive foreclosure rates Poor property management and disposition Excessive loan limits Imprudent appraisal practices 100% guarantee promotes misaligned incentives Poor underwriting practices impact first time, working class and minority borrowers and their communities
nightmareatfha.com

|4

VAS SUPERIOR PERFORMANCE

Comparison of FHA and VA serious delinquency rates across FICO bands for 2009:
30% 25%

20%

FHA @41-45% DTI

15% VA @41-45% DTI VA @51-55% DTI 10%

5%

0% 580-599 600-619 620-659 660-679

The first (FHA @41-45% DTI) and second bars (VA @41-45% DTI ) for each group show how VA underwritten loans perform relative to FHA loans in similar FICO bands and DTIs. The VAs performance across these bands averages about 67% of the FHAs. The third bar for each group shows the VAs performance for 51-55% DTI, which average about 80% of FHAs rate for 41-45% DTI loans.
nightmareatfha.com

|5

FHA VERSUS VA
AFRICAN AMERICAN LOAN GUARANTEE PERCENTAGES
The VA serves a greater percentage of African American families buying a home than FHA
15% FHA African American Percentage VA African American Percentage Source: Mortgage Bankers Association, derived from Home Mortgage Disclosure Act.

14%
13% 12%

11%
10% 9%

8%
7% 6%

2004

2005

2006

2007

2008

2009

2010

2011

nightmareatfha.com

|6

HARMING WORKING-CLASS FAMILIES AND COMMUNIITIES

FHA Loan Projected Foreclosure Rate Low-Income Zip Code vs. Middle-Income Zip Code Areas (21 cities plus 3 New York boroughs)
50% 45%
Low Income Rate

40%

Middl e Income Rate

Projected Foreclosure Rate

35% 30% 25% 20% 15% 10% 5% 0%

From a recent study where 19 of 22 cities had higher FHA projected foreclosure rates in low-income zip codes than in middleincome zip codes. Replicates the National Training and Information Centers 2002 study for the same 22 cities. That study found 21 of 22 cities had higher FHA loan default rates in low-income census tracts than in middleincome census tracts.
nightmareatfha.com

Richmond, VA

NYC, Queens

Philadelphia, PA

Washington, DC

Buffalo, NY

Memphis, TN

Baltimore, MD

Cincinnati, OH

Birmingham, AL

Los Angeles, CA

Columbus, OH

NYC, Brooklyn

Rochester, NY

Syracuse, NY

Newark, NJ

Chicago, IL

Detroit, MI

Miami, FL

Dallas, TX

Cleveland, OH

Indianapolis, IN

Trenton, NJ

Atlanta, GA

NYC, Bronx

|7

HARMING MINORITY FAMILIES AND COMMUNIITIES

45%

FHA Loan Projected Foreclosure Rate Minority Zip Code vs. White Zip Code Areas (21 cities plus 3 New York boroughs)

40% 35%

Mino rity Rate

White Rate

Projected Foreclosure Rate

30% 25% 20%

From a recent study where 19 of 22 cities had higher FHA projected foreclosure rates in minority zip codes than in white zip codes. Replicates the National Training and Information Centers 2002 study for the same 22 cities. That study found 19 of 22 cities had higher FHA loan default rates in minority census tracts than in white census tracts.
nightmareatfha.com

15%
10% 5% 0%

NYC, Brooklyn

Los Angeles, CA

Philadelphia, PA

Richmond, VA

Newark, NJ

NYC, Queens

Syracuse, NY

Buffalo, NY

Washington, DC

Birmingham, AL

Cleveland, OH

Memphis, TN

Chicago, IL

Detroit, MI

Miami, FL

Dallas, TX

Baltimore, MD

Indianapolis, IN

Cincinnati, OH

Columbus, OH

Rochester, NY

Trenton, NJ

Atlanta, GA

NYC, Bronx

|8

FORECLOSURE CRISIS: QUADRANT OF DOOM

In Chicago, the highest foreclosure rates and percentage of loans with FICOS greater than 660 are concentrated in working-class zips where incomes and home prices are below area median

nightmareatfha.com

|9

DISPROPORTIONATE IMPACT ON WORKING-CLASS COMMUNIITIES

From a study showing that an FHA borrower living in a zip code in the fourth quartile (those with a mean Equifax Risk Score of 519 to 673) had an average 45 percent higher likelihood of being 90+ days delinquent compared to a borrower with the same risk characteristics (measured across 245 risk buckets) living in a zip code in the first quartile (with mean Equifax Risk Scores of 721 to 826).
nightmareatfha.com | 10

DISPROPORTIONATE IMPACT ON WORKING-CLASS COMMUNIITIES

Neural Net Predicted 90+ Rate By Risk Bucket 1st & 4th Interthinx Risk Index Quartiles
45%
Neural Net Predicted 1st Quartile 90+ Rate Neural Net Predicted 4th Quartile 90+ Rate

40%
35%

30%
90+ Rate 25%

20%
15%

10%
5%

0%

Loans to FHA borrowers in zip codes with a higher fraud risk score (FRS) perform worse than loans in zips with a lower FRS, even when borrower risk factors such as FICO score, down payment, and DTI ratio are held constant. An FHA borrower in a zip code in the 4th quartile (the 25% of the zips with the highest FRS), on average, had a 31% higher likelihood of being 90+ days delinquent compared to a borrower with the same risk factors living in a zip in the 1st quartile (the 25% of the zips with the lowest FRS). nightmareatfha.com | 11

NEEDY FAMILIES NEED FHAS FULL ATTENTION

Step back from markets that can be better served by the private sector. Concentrate on low- and moderate-income homebuyers who truly need help purchasing their first home.

nightmareatfha.com | 12

STOP FINANCING FAILURE

FHAs average one-in-eight Family Claim Rate (FCR) masked much higher family failure rates in many working-class and minority neighborhoods. The FHAs current credit box for 580-679 FICO borrowers yields a weighted average FCR of 18%. FCRs range from 26% (580-599 FICO scores) to 13% (660-679 FICO scores). The deleterious effects resulting from such high failure rates can be avoided by a two-step process designed to achieve FCRs of 7.5% on 580-679 FICO credit score loans: Implement the general process improvements designed to align incentives. Balance down payment, loan term, FICO, and debt-to-income (DTI). These steps would drop FHAs overall rate more than in half to 5-6 percent.

nightmareatfha.com | 13

ALIGNMENT OF INCENTIVES
Contributes greatly to the VAs success:
Combines public sector mission of benefiting the veteran backed by public sector guarantee. Alignment of financial interests:
25% guaranty limit aligns originator, servicer, issuer and Ginnie with the VA and homeowner Doing it right from start to finish is in everyones interest Allows the VA to address issues within this rationalized structure, rather than responding with ad hoc solutions Origination, securitization, servicing

Utilizes processes to reduce risk and fraud The lack of properly aligned incentives helps explains why, notwithstanding the magnitude and persistency of FHAs problems, efforts at reform have failed.
nightmareatfha.com

14

VAS ABILITY TO PAY PRACTICES

VA requires underwriters to identify and verify income available to meet:
The mortgage payment
Other shelter expenses (includes utilities and maintenance)

Debts and obligations (includes job related expenses such as child care)
Family living expenses Residual income needs

The resulting debt-to-income ratio (DTI) is secondary to residual income as an underwriting factor.
nightmareatfha.com | 15

APPRAISER PANELS
Alignment of incentives:
Panel selected based on experience and geographical competence (VA) Size: 4500 (VA) vs. 55,000 (FHA). VA did 40% of the FHAs volume (2012)
Reopening of local panel based on need, additions based on competence (VA) 2012: goal of increasing VA panel size from 4600 to 5800 , now at 5200

Assignment based on rotation (VA) vs. lender selection (FHA) Quality control (VA)
VA staff appraisers or designated lenders Minimum of 10% of work is field reviewed

Two benefits of appraiser panels merit special mention:

Appraiser independence takes away a tool from unscrupulous parties. Appraiser independence results in greater identification of needed property repairs and shortcomings.

nightmareatfha.com | 16

VAs 25% COVERAGE

VA pays up to an average of 25%, averaging 25% while FHA pays up to 100% of the claim amount, averaging 63%.

FHAs loss rate is an estimated 5 times the VAs (2 times the incidence and 2.5 times the severity).
The VA charges 1/3 FHAs premium (present value basis). VA Issuers absorb 1.6 times the overall loss rate compared to FHA issuers for the same fee.

nightmareatfha.com | 17

VAS 25% COVERAGE

Did not deter ability to play a countercyclical role.
VAs volume tripled from 2007 to 2009. In 2009 the VAs median FICO score was 705, similar to the FHAs median of 694.

Did not deter ability to serve underserved borrowers in a sustainable manner.

In 2005, 43% of the VAs originations had a FICO between 600-679, virtually identical to FHAs 44%.

VA has experienced substantially lower serious delinquency rates than the FHA for decades and are currently half the FHA rate (2001-2012).
nightmareatfha.com | 18

REDUCE SELLER CONCESSIONS

In July 2010, FHA Commissioner Stevens proposed eliminating seller concessions >3%.
FHA allows up to a 6% seller concession vs. 3% for conventional market:
The incidence of concessions and the average concession is highest for loans <$180,000 (lowest loan size for which FHA provided data). When concession is >3%, default rate 1.9 times that of loans where 0% (1/3 of FHA loans below $180,000 have a 0% concession. When concession is >3%, default rate 1.3 times that of loans where >0% and <=3%.

Concessions of >3% subject working class families and neighborhoods to needless foreclosure risk.

nightmareatfha.com | 19

WORKING CLASS COMMUNITIES DESERVE COMMON SENSE REFORM

Reduce coverage below 100 percent (Section 234 of the Protecting American Taxpayers and Homeowners Act of 2013 (PATH)); Advise consumer as to foreclosure risk based applicants risk profile (Section 236 of PATH); Limit seller concessions to 3 percent (Section 263 of PATH); Utilize residual income test (Section 267 of PATH); Reinstitute vetted appraisal panels with rotational assignment; Introduce countercyclical stress test and LTV ratios; and Underwrite for risk.

nightmareatfha.com | 20