Briefing on National Mortgage Risk Index and Other Risk Measures - - PowerPoint PPT Presentation

Briefing on National Mortgage Risk Index and Other Risk Measures

Edward Pinto and Stephen Oliner AEI International Center on Housing Risk HousingRisk.org August 25, 2014

1

Key Takeaways from Today’s Briefing

• National Mortgage Risk Index (NMRI) for purchase loans

– Using additional data, the NMRI has been revised to

• Rate all loans in month of first loan payment rather than month of securitization
• Extend composite, FHA, and RHS series back to Nov. 2012 from Aug. 2013

– 196,000 loans added in July, bringing total in NMRI to 3.66 million – NMRI at 11.41% in July, was down ¼ percentage point from June (revised) but up more than ¾ percentage point from July 2013 (new data) – Little discernible volume impact from QM regulation: over past 3 months, 20.9% of loans had total DTI > 43%, little changed from 2013:H2 – FHA not compensating for riskiness of high DTI loans; Fannie and Freddie compensating only to a limited extent – FHA’s NMRI, at 23.8%, was unchanged from June (revised). But it’s up 2½ percentage points from July 2013 (new data), a level that risks fueling home price volatility, particularly in lower income and minority areas – The softness in mortgage lending not due to tight standards but to reduced affordability, loan put back risk, and sluggish economic recovery

2

Key Takeaways from Today’s Briefing

• State-level Mortgage Risk Indices (SMRIs)

– Wide range across states (lowest: Hawaii and Washington, DC; highest: Mississippi) – Recent trend (May-July vs. February-April):

• For FHA/RHS loans, SMRI continuing to rise in almost all states
• For Fannie/Freddie loans, SMRI continuing to rise in most states
• Metro-area Indices

– Focusing on California – Mortgage risk highest in Central Valley and lowest in San Francisco Bay area – Housing risk in most California metros is at a high level and has risen substantially over the last year

3

Background

• Financial crisis largely stemmed from a failure to understand

build-up of housing risk:

– Mortgage risk – House-price (collateral) risk – Capital adequacy

• AEI’s International Center on Housing Risk (HousingRisk.org)

is addressing this problem:

– Will provide objective and transparent risk measures – Mortgage risk indices published monthly – Monthly tracking of housing cycle risk for California metro areas and regions alongside mortgage risk indices – Indices of collateral risk and capital adequacy to be released later this year

4

Change to Timing Convention

• Usual timeline: Loan origination  securitization  first payment
• Loan origination date not known. Previously used month of securitization.
• However, monthly securitization volume can be lumpy compared to loan originations

(and associated first payment dates). Induced spurious volatility in monthly NMRI.

• Switched to first payment date. Revised NMRI is less volatile month to month. Also,

shifted a month or two later relative to previous series.

5

10.0% 10.5% 11.0% 11.5% 12.0% 10.0% 10.5% 11.0% 11.5% 12.0% Aug-13 Sep-13 Oct-13 Nov-13 Dec-13 Jan-14 Feb-14 Mar-14 Apr-14 May-14 Jun-14 Jul-14

Effect on Composite NMRI

Revised (1st payment month) Stressed default rate Previous (securitization month)

The NMRI

• Principles behind the NMRI

– Market stability depends on the preponderance of loans being low risk, defined as good performance under stress – NMRI provides a stress test, similar to crash tests for cars and structural stability ratings for buildings in hurricane zones

• Basics of index construction and coverage

– Places loans in risk buckets and assesses default risk based on performance

• f 2007 vintage loans with similar characteristics

– Currently covers nearly all gov’t-guaranteed mortgages for home purchases (about 75% of all purchase loans) – Plan to expand coverage later this year to include the VA, non-agency loans, refinances, and second mortgages

6

Stressed Default Rates

• Takeaway: Huge spread of default rates across risk buckets
• All 320 risk buckets for home purchase loans are shown at Periodic

Table on HousingRisk.org

• Additional loan risk factors applied to ARMs, investor loans, second

homes, 15 year terms, and 20 year terms

7

Risk Bucket FICO CLTV Total DTI Default Rate Very Low ≥ 770 61-70% ≤ 33% 0.8% Low 720-769 76-80% 34-38% 4.2% Medium 690-719 81-85% 39-43% 9.3% High 660-689 91-95% 44-50% 22.7% Very High 620-639 > 95% > 50% 45.8%

Note: Default rates represent cumulative defaults through year-end 2012 for Freddie Mac’s 2007 vintage of acquired loans. The loans included in the calculation are all primary owner-

• ccupied, 30-year fixed-rate, fully amortizing, fully documented, home purchase loans.

NMRI for Home Purchase Loans

* Change from November 2012 to July 2014. Source: AEI International Center on Housing Risk, www.HousingRisk.org. RHS is Rural Housing Service. VA guaranteed loans not included.

Composite fell ¼ percentage point, reflecting a decline at Freddie and a drop in the FHA/RHS share. RHS at a new series high, other 3 agencies at or near highs.

8

4% 8% 12% 16% 20% 24% 4% 8% 12% 16% 20% 24% Nov-12 Jan-13 Mar-13 May-13 Jul-13 Sep-13 Nov-13 Jan-14 Mar-14 May-14 Jul-14

Fannie/Freddie: +1.1 ppt, from 4.9% to 6.0%* Composite: +0.2 ppt, from 11.2% to 11.4%* Stressed default rate FHA/RHS: +2.5 ppt, from 20.6% to 23.1%* FHA/RHS share of purchase loans: Nov-12: 39.9%; Jun-14: 33.2%; Jul-14: 31.8%

Calibrating Mortgage Safety

• Composite index substantially above 1990 level, but not approaching 2007

level when underwriting was exceptionally lax

• Fannie/Freddie index somewhat above 1990 level
• FHA index is extremely high. Sharp contrast with safe underwriting during

1935-55.

9

NMRI – purchase loans Latest date Latest value 1935-1955 vintages (est.) 1990 vintage (est.) 2007 vintage (est.) Composite index July 11.4% NA 6% 19% Fannie and Freddie July 6.0% NA 4% 13% FHA July 23.8% 3% 15% 30% An index value of less than 6 is indicative of conditions conducive to a stable market.

Recent Headlines about Credit Availability

• Large Lenders Loosen Credit Standards, Small Lenders Tighten

Up—Fannie Mae survey (the MReport, July 25)

• Mortgage Credit Loosens in July Index—MBA survey (the

MReport, August 5)

• Lenders Waiting for FHFA to Act Before Boosting Credit (National

Mortgage News, August 7)

• Banks Making Fewer Mortgages Because of New CFPB Rules,

Fed Says (American Banker, August 4)

• [Ability to Repay], QM Aren’t Majorly Impacting Prime Mortgage

Market—Fed Senior Loan Officer Opinion Survey (the MReport, August 5)

10

Fed’s Senior Loan Officer Survey

11

• 40
• 20

20 40 60 80 100

• 40
• 20

20 40 60 80 100 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

Net percentage of respondents tightening standards on residential mortgage loans Tightening Loosening All residential mortgage loans Prime Subprime

Limitations of the survey:

• Can be unreliable. Showed no net loosening of standards over 1996-2006.
• Subjective. Responses are opinions of loan officers. Not based on hard data.
• Imprecise definitions. Where do banks draw the line between prime and subprime?
• Misses mix shifts. No information on share of lending that is subprime.
• Small sample. Only a handful of banks report on subprime lending.

In contrast, NMRI is objective, transparent, and based on millions of loans

Risk Shares for Home Purchase Loans

Source: AEI International Center on Housing Risk, www.HousingRisk.org.

Loan risk greater than level conducive to long-run market stability, with low-risk loans accounting for only about 42% of activity in July, down from 45% in 2013:H2.

12

15% 20% 25% 30% 35% 40% 45% 50% 15% 20% 25% 30% 35% 40% 45% 50% Nov-12 Jan-13 Mar-13 May-13 Jul-13 Sep-13 Nov-13 Jan-14 Mar-14 May-14 Jul-14

Low risk High risk Medium risk Low risk defined as stressed default rate of less than 6%, medium risk is 6% to 12%, and high risk is 12% or higher.

Renewed Pressure for Subprime Lending Succeeding

• “[L]ender overlays are damaging the recovery by limiting access to creditworthy

borrowers”. FHA Commissioner Carol Galante, March 2013

• “[Lender] credit overlays result in the rejection of many loans that would otherwise

meet [GSE] credit standards”. FHFA Director Mel Watt, May 2014

• Calls to more fully utilize existing the FHA and GSE credit boxes risk fueling home

price volatility, particularly in lower income and minority areas

13

10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Nov-12 Mar-13 Jul-13 Nov-13 Mar-14 Jul-14

% of top 25 FHA purchase-loan issuers with loans below credit-score cutoffs

Below 620 Below 600 Below 580 Top 25 issuers account for about ¾ of total FHA volume. Below 640 0% 5% 10% 15% 20% 25% 30% 35% Nov-12 Mar-13 Jul-13 Nov-13 Mar-14 Jul-14

% of FHA purchase loans at top 25 FHA issuers by credit score

640-659 620-639 <620 Range for July NMRI: 32% for 640-659 to 37% for <620 All loans below 660 (July NMRI: 33%)

GSE Low-Risk Origination Shares, Purchase Loans

Source: AEI International Center on Housing Risk, www.HousingRisk.org.

Low-risk shares rose slightly in July but remain well below year-ago levels.

14

58% 60% 62% 64% 66% 68% 70% 72% 74% 58% 60% 62% 64% 66% 68% 70% 72% 74% Nov-12 Jan-13 Mar-13 May-13 Jul-13 Sep-13 Nov-13 Jan-14 Mar-14 May-14 Jul-14

Freddie Mac Fannie Mae Combined Note: FHA/RHS low-risk share (not shown) averages about 3%.

Purchase Loans with Total DTI Greater Than 43%

Source: AEI International Center on Housing Risk, www.HousingRisk.org.

Little impact to date from QM regulation. Federally guaranteed loans are exempt from total DTI limit in QM, so share with total DTI > 43% likely to remain high.

15

10% 15% 20% 25% 30% 35% 40% 10% 15% 20% 25% 30% 35% 40% Aug-13 Sep-13 Oct-13 Nov-13 Dec-13 Jan-14 Feb-14 Mar-14 Apr-14 May-14 Jun-14 Jul-14

FHA/RHS Composite Fannie/Freddie

Additional Detail on Total DTIs

• 38% of Fannie/Freddie loans have total DTIs > 38%, up from only 14% in 1990 (based on

Fannie random sample)

• FHA is only agency with a sizable share of loans with total DTIs > 50%. This is an

extremely high pre-tax payment burden

• High total DTIs crowd out participation in defined contribution retirement plans such as

401Ks, most of which come with employer match. These provide a reliable and attractive means for private wealth accumulation, particularly for lower income families

16

July 2014 Total DTI > 38% Total DTI > 43% Total DTI > 50% Composite 45% 21% 4% Fannie/Freddie 38% 13% ≈ 0% FHA 65% 40% 14% RHS 43% 15% < 1%

Source: AEI International Center on Housing Risk, www.HousingRisk.org

Ratio of NMRI for Loans with Total DTIs > 43% to Loans with Lower Total DTIs

“FHFA will continue to permit these compensating factors in each company’s underwriting standards.” FHFA Director Mel Watt, May 13, 2014

*For FHA include: down payment ≥10%, previous credit history, minimal housing expense increase, accumulated savings/cash reserves, or income beyond effective income. For GSEs include: down payment, credit score, or cash reserves. Factors not disclosed at loan level by agencies.

17

Full use of compensating factors* would imply ratios of 100%. Instead, see risk layering.

90% 100% 110% 120% 130% Aug-13 Oct-13 Dec-13 Feb-14 Apr-14 Jun-14

NMRI Ratio: FHA

Ratio to loans with DTIs of 34-38% Ratio to loans with DTIs of 39-43% No use of compensating factors

90% 100% 110% 120% 130% Aug-13 Oct-13 Dec-13 Feb-14 Apr-14 Jun-14

NMRI Ratio: Fannie and Freddie

Ratio to loans with DTIs of 34-38% Ratio to loans with DTIs of 39-43% Only limited use of compensating factors

Purchase Loans with Down Payment of 5% or Less

Source: AEI International Center on Housing Risk, www.HousingRisk.org.

Half of all purchase loans and a quarter of Fannie/Freddie purchase loans have a minimal down payment. Fannie/Freddie share has risen since the start of the series.

18

20% 25% 30% 35% 40% 45% 50% 55% 20% 25% 30% 35% 40% 45% 50% 55% Nov-12 Jan-13 Mar-13 May-13 Jul-13 Sep-13 Nov-13 Jan-14 Mar-14 May-14 Jul-14

Fannie/Freddie Composite More than 90% of FHA/RHS purchase loans have a down payment of 5% or less.

Deleterious Effects of the 30-Year Loan with High Debt Ratios and Low Down Payment

• Equity accumulation is slow and uncertain

– 30-year loan amortizes very slowly and homeowner can’t count on price appreciation (“The housing market is a ‘crapshoot’” – Prof. Karl Case1) – Appreciation especially slow for middle- and low-income homeowners in most major cities2

• High total DTIs crowd out participation in DC retirement plans
• Total DTIs ignore the cost of commuting

– The median commute is 10 miles one-way

• One-worker household: totals $1440/year or 3% of median household income3
• Two-worker household: totals $2900/year or 6% of median household income

– Can result in very heavy payment burdens for lower income borrowers – Promotes sprawl

• Combining faster amortization, stronger underwriting, and a 401(k) provides a

broad, straight highway to building wealth

1 http://money.cnn.com/2014/07/07/investing/housing-market-case/ July 7, 2014 2 Based on Zillow data from April 1996 to May 2014 for appreciation by house-price tier in 28 major cities 3 Commuting expense based on 30 cents per mile

19

State and Metro Area Mortgage Risk Indices

• Comparison to NMRI

– Exactly the same stress-test methodology applied to loans at the state and metro-area levels – Some additional assumptions required to calculate metro-area indices because of limited geographic information in the data

• Metro area indices

– Continuing to focus on California – Coverage includes Fannie, Freddie, and FHA loans

20

SMRI for Home Purchase Loans All Agency Composite, May-July 2014

Source: AEI International Center on Housing Risk, www.HousingRisk.org.

21

Most states have a composite SMRI between 10% and 14%. Those at the extremes tend to have high or low concentrations of FHA/RHS loans.

6% 7% 8% 9% 10% 11% 12% 13% 14% 15% 16% MS LA WV KY AL IN NV RI WY OK AR TX NM OH TN AZ ID MD UT ME MI GA MO NH KS PA SC FL NE CA AK CT SD IA CO VA WA MN NJ IL DE MT NC NY WI MA OR ND VT HI DC

“Sand States”

SMRI for Home Purchase Loans Fannie/Freddie, May-July 2014

Source: AEI International Center on Housing Risk, www.HousingRisk.org.

22

Variation across states for Fannie/Freddie loans is less than for all- agency composite. Almost all states have an SMRI between 5% and 7%.

4.0% 4.5% 5.0% 5.5% 6.0% 6.5% 7.0% 7.5% WY ID LA RI UT MS WI NM NH AL OH MA MN IN IL KS NE WA TX IA SD MI NV GA KY VT CT OK CO PA MO MT ND TN ME FL MD WV AK AR AZ OR CA SC NJ VA NC NY HI DC DE

“Sand States”

SMRI for Home Purchase Loans FHA/RHS, May-July 2014

Source: AEI International Center on Housing Risk, www.HousingRisk.org.

23

The SMRI is very high – above 20% in 49 states. “Sand States” in the upper part of the range.

17% 18% 19% 20% 21% 22% 23% 24% 25% 26% TX CO RI NM CA WA MA NJ NV IL AZ DE FL MD UT SC LA GA ID MS AL IA IN OK OH NH WI CT WY ND KY MO MI TN KS VA PA NE NC NY AK MN WV VT OR AR ME MT HI SD DC

“Sand States”

Change in SMRI for Home Purchase Loans Fannie/Freddie

Source: AEI International Center on Housing Risk, www.HousingRisk.org.

24

SMRI for Fannie/Freddie loans increased in most states from Feb-Apr 2014 to May-July 2014.

4.5% 5.0% 5.5% 6.0% 6.5% 7.0% 7.5% 4.5% 5.0% 5.5% 6.0% 6.5% 7.0% 7.5%

May '14 - Jul '14 Feb '14 - Apr '14 Increase in MRI Decrease in MRI

Change in SMRI for Home Purchase Loans FHA/RHS

Source: AEI International Center on Housing Risk, www.HousingRisk.org.

25

SMRI for FHA/RHS loans increased in nearly all states from Feb-Apr 2014 to May-July 2014.

17% 18% 19% 20% 21% 22% 23% 24% 25% 26% 17% 18% 19% 20% 21% 22% 23% 24% 25% 26%

May '14 - Jul '14 Feb '14 - Apr '14 Increase in MRI Decrease in MRI

26

Fannie/Freddie/FHA MRI

Price Volatility

Demographics Metro Area/Region

• Nov. 2012

July 2014 (level and change from prior mo.)

Best year – worst year (1979-2014)

Area median income/ Black or Hispanic share

San Francisco Metro Area 7.4% 8.3% (+.0 ppt) 47 ppt $101,200 / 30% San Diego CBSA 8.4% 9.3% (+.5 ppt) 49 ppt $72,300 / 37% Los Angeles CBSA 9.2% 9.7% (+.1 ppt) 48 ppt $61,900 / 51% Sacramento CBSA 10.4% 11.7% (-.3 ppt) 44 ppt $70,900 / 27% Riverside-San Bernardino CBSA 13.3% 13.5% (+.0 ppt) 66 ppt $62,600 / 56% Central Valley region 14.6% 14.8% (+.2 ppt) NA $53,000 (approx.) / 49% California average 10.3% 11.0% (+.1 ppt) NA $69,600 / 44% National average 10.7% 11.0% (-.2 ppt) 34 ppt $64,400 / 29%

Note: Fannie Mae MRIs are estimated using the Fannie Mae California state MRI and the Freddie Mac ratio of the metro area or region MRI to the California state MRI; geographic distribution of Fannie Mae loans assumed to be the same as Freddie Mac. FHA MRI in each metro area/region equals the California state MRI; geographic distribution of FHA loans is as reported in FHA Neighborhood Watch for the year ending March 31, 2014. San Francisco Metro Area includes San Francisco CBSA plus Napa, San Benito, Santa Clara, Santa Cruz, Solano, and Sonoma counties. Central Valley includes Butte, Fresno, Kern, Kings, Madera, Merced, San Joaquin, Shasta, Stanislaus, Sutter, Tulare, and Yuba counties. Volatility Index for top 50 metros from Zillow.com. Area median income from HUD; Black or Hispanic share from Census Bureau, table DP05.

Home Purchase Mortgage Risk Index (MRI) in Major California Metro Areas

• Current level of MRI for California same as national average but has increased more since late 2012
• FHA’s California volume is concentrated in Riverside-San Bernardino and the Central Valley, pushing

the MRIs for both areas well above the national average. These are areas with relatively low income and large minority populations.

27

Indexes Housing Cycle Risk Fraud Bubble Watch Metro or Region

• Dec. 2012

June 2014 (change prior mo.) 2013:Q4 National: Q1:12 = 100 Change, 2013:Q2 to 2014:Q2 San Francisco Metro

A+ C+ (unch.) 145 +7 ppt

San Diego CBSA

A C (unch.) NA +10 ppt

Los Angeles CBSA

A C (unch.) 149 +10 ppt

Sacramento CBSA

A+ B- (unch.) NA +7 ppt

Riverside-San Bernardino CBSA

A+ C (unch.) 135 +16 ppt

Central Valley

A B-/C+ (C+) 145.5 +9 ppt

California average

A+/A C+ (unch.) 139 NA

National average

A/A- B- (unch.) 101 +3/+4 ppt

Note: All metros/regions are matched to the extent possible based on geographic coverage of source data; results for metro area/regions are unweighted averages of included counties. San Francisco Metro Area includes San Francisco CBSA plus Napa, San Benito, Santa Clara, Santa Cruz, Solano, and Sonoma counties. Central Valley includes Butte, Fresno, Kern, Kings, Madera, Merced, San Joaquin, Shasta, Stanislaus, Sutter, Tulare, and Yuba counties. Housing Cycle Risk index provided by John Burns Consulting. Mortgage Risk Fraud Risk Index from Interthinx. Bubble Watch shows change relative to fundamentals and is from Trulia.com.

Housing Risk in Major California Metro Areas

• House price risk in California is relatively high and rising.
• Combination of elevated mortgage and housing cycle risk in historically volatile California market cause

for concern, especially in lower income and minority areas such as Riverside-San Bernardino and Central Valley.

Release Dates for Rest of 2014

• At 10 AM on dates shown below
• No change from previous schedule
• Next briefing will be Monday, September 29

28

Monday September 29 Monday October 27 Monday November 24 Monday December 22