OMS Housing-Pinto PDF

AEI Housing Market Indicators (HMI)
Edward Pinto (Edward.Pinto@AEI.org) and

Tobias Peter(Tobias.Peter@AEI.org)
AEI Center on Housing Markets and Finance

AEI.org/housing
April 1, 2019
We grant permission to reuse this presentation, as long as you cite as the source:
AEI, Center on Housing Markets and Finance, www.aei.org/housing.
1
AEI Housing Market Indicators: An Introduction
• Provide accurate and timely metrics for the housing market. These include:
• Mortgage Risk/Leverage
– Particular focus is paid to agency first-time buyer volume and risk
• House prices and appreciation trends
• Housing sales
– New and existing sales whether institutionally financed, cash, and other-financed
• Months inventory
• The housing market is influenced by many different levers. All need to be
connected and considered to better understand market trends.
– AEI HMI adds geography and price points to the broad set of metrics:
• Geography: national, state, and selected metros
– House prices down to the census tract
• Price points: low, low-medium, medium-high, and high price tiers
– Price tiers are defined based on the availability of leverage for borrowers at certain geographies.
• Expanded Housing Market Indicators use and connect many different datasets:
• HMDA
• Public Records Data
• National Mortgage Risk Index (agency MBS data)
• CoreLogic’s LLMA and Black Knight’s McDash (servicer data)
• Fannie Mae’s Loan Performance data and Freddie Mac’s Loan-Level Data (acquisition data)
• FHA Snapshot data (endorsement data)
• Data from Zillow on existing home sales and unique listings
• Advantages of the AEI Housing Market Indicators:
– Most in-depth resource for key housing data and trends (select data available online for download)
– Accurate, timely, and in-depth coverage of purchase trends
– Connects the dots for many housing indicators, yielding the most comprehensive analysis of the
housing market and boom/bust cycles
• Detailed Methodologies are available after “Next Briefing Dates” slide. 2
HMI Key Takeaways: Tracking the Home Price Boom
• Given the interest rate drop since November, the rate of house price appreciation (HPA) is again rising.
– At our January 7, 2019 briefing we predicted: “Given the rate drop since November, we would expect a modest
pickup in the HPA rate”
– Preliminary numbers for February 2019 indicate national HPA of 4.3% (yoy) up from 3.5% in December 2018 (yoy)
– HPA remained strongly bifurcated. Prices in the low price tier appreciated at 6.8% (yoy), while house prices in the
high tier fell by 2.9% (yoy)
– Given the recent further drop in rate, we expect HPA to accelerate further, perhaps even extending to a return to
positive HPA for the high tier.
• Mortgage risk continued to increase. The composite Purchase National Mortgage Risk Index (NMRI) set a
series’ high in December 2018. The index was up 0.5 percentage point from December 2017.
• Rumors of the return of a buyer’s market are greatly exaggerated
– The national seller’s market continued and now stands at 75 consecutive months.
– Nationally, months' supply stood at 3.5 months for the 4 th quarter of 2018.
– Inventories remained exceptionally tight (2.5 months) for the entry-level home market (low and low-medium tiers)
and moderately tight at the move-up market (5.3 months)
• On the demand side, there was a pull back in the purchase transactions during the most recent quarter.
– For the four quarters ending in 2018:Q4, 6.0 million sales transactions were reported.
– Of those 6.0 million, 596,000 were new construction sales.
– Sales transactions decreased 1.5 percent in the 4th quarter compared to a year ago.
• December 2018 Refi volume also declined (down 50% from a year ago), but refi risk increased.
– This trend was driven by a share shift from higher-quality to lower-quality borrowers, from the GSEs to FHA and VA,
and from rate and term refis to cash-out refis.
• With access to plenty of leverage from the government agencies, the first-time buyer mortgage share index
(FBMSI) continued to increase in December 2018
– For December 2018 the index stood at 59.2%, up 0.8 ppt from a year ago.
– Compared to five years ago, the FBMSI is up 3.8 ppts. from 55.4%.
3
It’s a Buyer’s Market, It’s a Buyers’ Market!
• After finally admitting there is a house price boom, the media has quickly changed
its tune to: It’s a buyers’ market and this will help entry-level buyers
– “Real estate agents say housing market is favoring buyers” Housing Wire, 3.25.19
– “Out of the Seller’s Market, Into the Buyer’s Market” DSNews, 3.21.19
– “Homebuyers gain edge in this year's housing market” USA Today, 3.7.19
– “A Buyer’s Market? Hopes Rise With Falling Rates, More Homes for Sale” WSJ, 3.31.19
• Yet, the rumor of the return of a buyer’s market is greatly exaggerated:
– Unsustainable house price appreciation (HPA) continues for low price tier
• In February the low tier HPA accelerated to 6.8% (yoy), the largest increase since June 2018
– At under 2.5 months, months’ inventories remain exceptionally tight for the national
entry-level home market and at 4.6 months, moderately tight for the national med-high
bin (most of move-up market). For the high bin (8% of national sales) months’ inventory
is at 8.2 months, on the bubble for a buyers’ market (>8 months for the high bin)
• Our expectation is that house prices will continue to increase, with greatest
increases in the lower price tiers, thus worsening affordability
• The house price boom, now in its 7th year, has been driven by 2 punchbowls:
– Easy monetary policy which applied equally to all types of buyers, and
– Loosening mortgage underwriting standards by government agencies, which has almost
exclusively been aimed at first-time buyers.
• The Fed is now dovish (rates have plummeted 88 bps since November, 2018).
– This along, with expected continued job and wage growth and the FTB punchbowl,
virtually guarantees that the long running boom with not only continue, but strengthen.
This may well have dangerous consequences, especially for highly leveraged entry- 4
level buyers.
FHA’s March 2018 Action to Address Excessive Loan Risk Is a
Positive Step, but Too Early to Tell How Significant
We’ll be looking to confirm three positive outcomes, note however rates are down,
wages and jobs are up, and these will spur demand:
• Entry-level homes may be more affordable.
• FHA's excessively high level of loan leverage (example: DTIs >50% leads to increased entry level
homes house price appreciation due to the long-standing sellers' market. This change will slow
the rate of house price appreciation (all things equal)
• But impossible to say by how much, as definitive data until mid-June or mid-July
• A small reduction in imbalance between supply and demand*
• Nationally an extremely strong sellers' market continues in the entry-level price range comprising
57% of sales.
• 96 of the 100 largest metros are experiencing a moderate to extreme sellers’ market for the
low price tier (28% of sales);
• 98 of the 100 largest metros for the low-med tier (29% of sales).
• This high demand (and no change in supply) will result in about the same number of homes
selling as before (all things being equal)
• Those who remain as renters will avoid jumping into a 7 year old boom market.
•We do not know when the boom will end, just that it will.
•High risk borrowers entering the market late in the cycle tend to get foreclosed on at high rates.
The FHA loans these renters avoid have an extremely high risk of default under stress: >40%.
With the recent rate drop, this step may prove to little to avoid unsustainable entry-level price increases.
*Data are for 2018:Q4. Months' inventory is for 912 counties (weighted by volume) and represents well over 90% of all sales. Share by price
range is for the entire US. Source AEI Housing Center and Zillow
Update: NMRI for Agency Home Purchase Loans
Composite index has consistently been trending up since mid-2013, with FHA leading
the way. The Composite, FHA, and VA indices all set new series’ highs in December.
Unless household income accelerates, future support for the housing market is likely
to involve further increases in leverage from an already high level. As noted earlier, it
is too soon to tell the impact of FHA’s March 2018 Total Scorecard changes.
Stressed default rate
32% 32%
FHA share of purchase loans: FHA: +7.8 ppts, from 20.9 to 28.7%*
28% Oct-18: 22.4%; Nov-18 22.3%; Dec-18 23.1% 28%
24% 24%
20% RHS: -0.5 ppt, from 19.4 to 18.9%* 20%
16% 16%
Composite: +2.2 ppts, from 11.1% to 13.3%*
12% 12%
VA: +1.8 ppts, from 10.8% to 12.6%*
Fannie: +3.0 ppts, from 5.0% to 8.1%*
8% 8%
Freddie: +1.7 ppts, from 4.9% to 6.7%*

4% 4%
Sep-12 Sep-13 Sep-14 Sep-15 Sep-16 Sep-17 Sep-18
*Change from December 2012 to December 2018.

Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing. RHS is Rural Housing Service.
6
Credit Easing = Punchbowl Spiking Continues, Led by FHA
The Composite NMRI for purchase loans increased from already elevated levels a year
ago. For FHA, the index is rising at a rate of 1.7% year-over-year. First-time buyers have
consistently been taking on greater leverage and default risk, which has helped fuel
accelerating house price growth for entry-level homes. Higher default risk combined
with unsustainable home price increases will lead to unnecessarily high default rates
during the eventual market correction.
Change from 12 months earlier, in percentage points
3.0 3.0
2.5 2.5
FHA
2.0 2.0
1.5 1.5
1.0 1.0
First-time buyers
0.5 0.5
Easing
0.0 0.0
Composite
Repeat buyers
Tightening
-0.5 -0.5
-1.0 -1.0
Sep-13 Jan-14 May-14 Sep-14 Jan-15 May-15 Sep-15 Jan-16 May-16 Sep-16 Jan-17 May-17 Sep-17 Jan-18 May-18 Sep-18
Note: Includes all types of NMRI purchase loans (primary owner-occupied, second home, and investor loans).
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing.
7
Fannie’s Unhealthy Competition with FHA Is Moving
Both Out the Risk Curve
A closer look at the risk distribution of Fannie and Freddie reveals that while Freddie has
loosened underwriting moderately, Fannie has been more aggressive. The share pickup for
Fannie, Freddie, and FHA shows that Fannie is increasingly competing with FHA for loans with
a risk score between 8-24. This segment currently accounts for 39% and 35% of Fannie and
FHA’s business respectively. FHA is replacing this lost business with much higher risk loans,
those with an MRI of greater than 32%.
Percent of loans
50%
40% December 2018 Distribution Freddie Fannie FHA
30%
20%
10%
0%
0-4 4-8 8-12 12-16 16-20 20-24 24-28 28-32 >32
Mortgage Risk Index
Percentage points
15%
10%
Change in Distribution, December 2016 to December 2018
Freddie Fannie FHA
5%
0%
-5%
-10%
0-4 4-8 8-12 12-16 16-20 20-24 24-28 28-32 >32
Mortgage Risk Index
8
Unforgiving Home Price Cycles: Booms Fueled by Increasing
Leverage in a Seller’s Market, Followed by Mean Reversion
Fueled by growing loan leverage and tight supplies, real home prices have increased 29% since
the early 2012 trough. Contrary to prevailing view, post-crisis underwriting/regulatory changes
promote rather than constrain a boom. The pattern is similar to the initial years of the price
boom that began in 1998. If it continues, the risk of a serious house price correction increases.
Real House Price Index (1975:Q1 = 100)*, through 2018:Q4
Predominantly a buyer's market Entirely a buyer's market
Entirely a seller's market Predominantly a seller's market
220 220
Real average annual growth rate
200 1997:Q2-2003:Q2 -- 4.3% 200
1997:Q2-2006:Q2 -- 5.1%
180 2012:Q2-2018:Q2 -- 4.2% 180
160 GSE affordable housing goals take effect for 160

CY 1993 as mandated by the Housing
Enterprises Safety and Soundness Act of 1992
140 140
120 120
2012 to date: easing loan
standards, very loose Fed
100 1993-2006: period of credit easing policy, and historically 100
and generally falling mortgage rates low mortgage rates
80 80
1975:Q1 1979:Q2 1983:Q3 1987:Q4 1992:Q1 1996:Q2 2000:Q3 2004:Q4 2009:Q1 2013:Q2 2017:Q3
* Calculated as FHFA's all-transaction house price index divided by BEA's price index for personal consumption expenditures.
Note: National Association of Realtors (NAR) defines a seller's market as inventory that is less than or equal to 6 months of sales. NAR data pertain to existing homes; not available
before June 1982. Data from the Census Bureau for new home inventories used before June 1982.
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, FHFA, BEA, Census Bureau, and NAR.
9
Dovish Fed = Monetary Punchbowl Getting Spiked Again
As predicted in our last briefings, the year-over-year rate of house price appreciation (HPA)
has picked up again. The national rate of HPA for February was 4.3%. This is down from its
recent peak of 6.1% in March 2018, but up from the 3.5% in December 2018. This coincides
with recent movements in mortgage rates. After having increased by 116 basis points from
September 2017 to early November, rates have since declined by 88 basis points.
30-year Fixed Rate Mortgage Year-over-Year Rate of House Price Appreciation
5.50 9.0%
8.0%
4.94
5.00 7.0%
6.0%
6.0%
4.50
5.0%
4.3%
4.0%
4.00
4.06
3.0%
2.0%
3.50
Red markers show February HPA in each year.
1.0%
Red markers show late-March rate in each year.
3.00 0.0%
Apr-14
Apr-15
Apr-16
Apr-17
Apr-18
Oct-15
Oct-13
Oct-14
Oct-16
Oct-17
Oct-18
4/3/2014
4/3/2015
4/3/2016
4/3/2017
4/3/2018
4/3/2019
10/3/2013
10/3/2014
10/3/2015
10/3/2016
10/3/2017
10/3/2018
Note: Data are for the entire country. Data for January and February 2019 are preliminary.
Source: Freddie Mac Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing.
10
National House Price Appreciation (HPA) by Price Tier
In February, (left panel) the low price tier not only continued, but reaccelerated its
unsustainable trend. In February 2019 (right panel), house prices in the low price tier
appreciated at 6.8% (yoy) - the strongest rate of growth since 2016. In the low-medium and
medium-high tiers, they increased at 3.9% and 3.1%, respectively. House prices continued to
decline in the high price tier at a rate of 2.9%.
Home Price Appreciation by Tier Year-over-Year HPA - by Tier
Index: Jan-2012 = 100
160 12%
10%
150
8%
140
6%
130
4%
120
2%
Overall
Overall
110 Low 0% Low
Low-Med
Low-Med
Med-High -2%
100 Med-High
High
High
-4%
90
Jan-13
Jan-14
Jan-15
Jan-16
Jan-17
Jan-18
Jan-19
May-13
May-14
May-15
May-16
May-17
Sep-17
May-18
Sep-13
Sep-14
Sep-15
Sep-16
Sep-18
Aug-16
Jul-14
Jun-12
Nov-12
Apr-13
May-15
Jan-17
Jun-17
Nov-17
Apr-18
Jan-12
Sep-13
Feb-14
Dec-14
Mar-16
Sep-18
Feb-19
Oct-15
Note: Data for January and February 2019 are preliminary. Price tiers are set at the metro level and are defined as follows: Low: all sales at or below the 40th percentile
of FHA sales prices; Low-Medium: all sales at or below the 80th percentile of FHA sales prices; Medium-High: all sales at or below the 125% of the GSE loan limit; and
High: Rest. HPAs are smoothed around the times of FHFA loan limit changes.
11
Supply-Demand Imbalance in the Market Is Driving Prices Up
Given the strong relationship between the level of supply and price movements, today’s
housing market has too much highly leveraged demand chasing too little supply.
According to the NAR, monthly inventory for February was at 3.5 months, 0.1 months
higher than last year’s reading. While house price appreciation (HPA) at 5.3% has
slowed from a year ago, it is still much higher than the rates of wage and inflation
growth. Given the previously noted drop in rates, we expect supply to tighten and HPA
rate to increase.
National Month’s Inventory & Changes in Nominal House Prices*
12 -18.0%
YoY Price Change (scale inverted)

10 -12.0%
Month's Supply*
(left axis)
Month's Supply
8 -6.0%
Dashed line: Price

Buyer's Market, Prices Falling equilibrium point
6 0.0%
Seller's Market, Prices Rising
FHFA Price Index,
smoothed** (right axis)
4 6.0%
2 12.0%
Jan-99 Jan-01 Jan-03 Jan-05 Jan-07 Jan-09 Jan-11 Jan-13 Jan-15 Jan-17 Jan-19
*Month’s supply updated through February 2019; FHFA House Price Index updated through January 2019.
** The NAR defines a seller’s market to exist when the inventory of existing homes for sale would be exhausted in six months or less at the current sales pace. Conversely, a
buyer’s market exists when the inventory of existing homes for sale exceeds six months at the current sales pace. (http://www.realtor.org/news-releases/2013/04/march-
existing-home-sales-slip-due-to-limited-inventory-prices-maintain-uptrend).
*** FHFA Monthly Purchase-Only Seasonally Adjusted house price index. The series is a 6 month trailing average.
Source: National Association of Realtors, FHFA, and AEI, Center on Housing Markets and Finance, www.AEI.org/housing. 12
Housing Volatility Index
Today’s 23 quarters constitute an extended housing boom, only exceeded by the last
boom. Sustained periods with few price declines allow market excesses to build and
may lead to a Minsky Moment.** Unsustainable increases in entry-level home prices
result in speculation in land, the more volatile part of the structure/land package.
Distribution of Negative House Price Change from Four Quarters Earlier in 81 US MSAs*
100%
90% Quiescent period
80% Correction
70%
60%
23 Qtrs.
21 Qtrs. 10 Qtrs. 15 Qtrs. 36 Qtrs. 39 Qtrs. 25 Qtrs. (through
50%
2018:Q4)
40%
30%
20%
10%
0%
1992:Q3
2011:Q3
1976:Q3
1977:Q3
1978:Q3
1979:Q3
1980:Q3
1981:Q3
1982:Q3
1983:Q3
1984:Q3
1985:Q3
1986:Q3
1987:Q3
1988:Q3
1989:Q3
1990:Q3
1991:Q3
1993:Q3
1994:Q3
1995:Q3
1996:Q3
1997:Q3
1998:Q3
1999:Q3
2000:Q3
2001:Q3
2002:Q3
2003:Q3
2004:Q3
2005:Q3
2006:Q3
2007:Q3
2008:Q3
2009:Q3
2010:Q3
2012:Q3
2013:Q3
2014:Q3
2015:Q3
2016:Q3
2017:Q3
2018:Q3
*Only 30 metros included at beginning of series. This number grows until 1977:Q4, when 81 metros are consistently reported.
**A Minsky moment occurs when a market collapses following a prolonged period of growth due to speculation and borrowing.
Source: FHFA Quarterly House Price Index and AEI Center on Housing Markets and Finance
13
Supply-Demand Imbalance Is Greatest in the Low Price Tier
There is a growing bifurcation on months’ supply in the market by entry-level (low and
low-med) vs. move-up (med-high and high). From a year ago, the supply-imbalance
has improved at all price points, but most at the upper end of the market. Inventories
remain historically tight at the lower end, continuing the strong seller’s market, which
implies that house prices will continue to increase, thereby worsening affordability.
10 Months’ Supply by Price Tier 10
9 9
Grey bars show Q4 for
each year.
8 8
7 7
6 6
5 5
4 4
3 3
2 Low 2
Low-Med
Med-High
1 1
High
0 0
2013:Q1 2013:Q3 2014:Q1 2014:Q3 2015:Q1 2015:Q3 2016:Q1 2016:Q3 2017:Q1 2017:Q3 2018:Q1 2018:Q3
Note: Data are for 918 counties representing approximately 90% of sales.
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, and Zillow. 14
Comparing the Supply-Demand Imbalance: 100 Largest Metros
While supply-demand imbalance has improved slightly, it remains tight in most metros,
especially at lower price tiers. For low and low-med tiers, 96% and 98% of metros are
sellers’ markets. For high tier, 69% are buyers’ markets, up from 48% a year ago.*
Low Price Tier Low-Medium Price Tier
10 10
9 Less supply than 9 Less supply than
Months Supply: 2017:Q4

8 one year ago 8 one year ago
7 7
6 6
seller's market Two out of range seller's market Two out of range
5 5
4 4
3 3
2 2
1 More supply than More supply than
1
one year ago one year ago
0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Months Supply: 2018:Q4 Months Supply: 2018:Q4
Medium-High Price Tier High Price Tier (note different axes maxima)
10 20

7 14
6 12
seller's market
5 Two out of range 10
4 8
3 6 seller's market
2 More supply than 4 Thirteen out of range
1 one year ago 2 More supply than
one year ago
0 0
0 1 2 3 4 5 6 7 8 9 10 0 2 4 6 8 10 12 14 16 18 20
* The demarcation point between buyer’s and seller’s market likely varies by price point. We have estimated them at 5, 6, 7, and 8 months respectively.
Note: Data for largest 100 metros using Zillow’s existing home sales. 15
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, and Zillow.
Months’ Supply: A Tale of Two Markets
Levels of inventories remained strongly bifurcated. In the low price tier, 2.1 months’
supply on average, while it was 8.2 months in the high price tier. This relationship
holds pretty much across the country as the maps show.
Low Price Tier High Price Tier

Median of 271 metros : 2.3 months Median of 271 metros: 15.1 months
Overall*: 2.1 months Overall*: 8.2 months
Months’ Supply (2018:Q4) : 271 CBSAs
0.7 6 24+
* Both maps show 271 metros. The overall months’ supply takes 271 metros as one market, and is calculated by dividing the total number of
listings by total sales.
Source: Zillow, AEI Center on Housing Markets and Finance, www.AEI.org/housing. 16
Months’ Supply: A Tale of Two Markets (Cont.)
Levels of inventories are even tighter for the largest metros, 1.8 months vs. 7.3 months
on average overall.

Median of 50 metros: 1.6 months Median of 50 metros: 9.2 months
Months’ Supply (2018:Q4) : Top 50 CBSAs
0.7 6 24
* Both maps show the top 50 metros, ranked by 2017 HMDA loan originations. The overall months’ supply takes 50 metros as one market, and is calculated by dividing the
total number of listings by total sales.
Source: Zillow, AEI Center on Housing Markets and Finance, www.AEI.org/housing.
17
House Price Appreciation: A Tale of Two Markets
As levels of inventories have remained tight for the largest metros, additional leverage
that is mostly applied to the lower price tiers has stimulated even more demand. With
supply limited, this leverage has been capitalized into higher house price increases at
the lower end.

Median House Price Appreciation: 43% Median House Price Appreciation: 19%
House Price Appreciation (2012:Q4 vs. 2018: Q4):

Top 50 CBSAs
10% 25% 74%
Note: Both maps show the top 50 metros, ranked by 2017 HMDA loan originations.
18
Annualized Home Sales (New vs Existing)
The national housing market retreated slightly in 2018. In 2018, 6.0 million sales
transactions were reported, down 90,000 transactions, or 1.5 percent, from 2017. New
home sales accounted for 10.7% of sales in 2018:Q4, which is up 0.9 ppt from 2013:Q4.
Sales Transactions in millions

6.5 6.5
Data are for the 4 quarter period ending with respective quarter.
6.0 6.0
New construction
Existing
5.5 5.5
5.0 New construction share of all sales: 5.0

2013:Q4: 9.8%
2014:Q4: 10.0%
2015:Q4: 10.3%
4.5 2016:Q4: 10.6% 4.5
2017:Q4: 10.9%
2018:Q4: 10.7%
4.0 4.0
2012:Q4 2013:Q2 2013:Q4 2014:Q2 2014:Q4 2015:Q2 2015:Q4 2016:Q2 2016:Q4 2017:Q2 2017:Q4 2018:Q2 2018:Q4
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, and First American Data Tree (DataTree.com). 19
Quarterly Home Sales (New and Existing): by Type
Sales for 2018:Q4 were down (-7.9%) from 2017:Q4, with institutionally financed sales
down (-7.5%). Compared to the 2012:Q4, sales by count have grown 13%. As predicted,
over-all 2018 volume has come in about flat over 2017 volume. Over the next months, we
expect a rebound in sales due to lower mortgage rates and availability of leverage.
Home purchase transactions
2,000,000 2,000,000
Red markers show Q4 counts in each year.
1,800,000 1,800,000
All home sales
1,600,000 1,600,000
1,400,000 1,400,000
1,200,000 1,200,000
1,000,000 1,000,000
Institutionally
800,000 800,000
financed sales
600,000 600,000
Cash sales
400,000 400,000
200,000 200,000
Other financed sales

0 0
2012:Q4 2013:Q2 2013:Q4 2014:Q2 2014:Q4 2015:Q2 2015:Q4 2016:Q2 2016:Q4 2017:Q2 2017:Q4 2018:Q2 2018:Q4
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, and First American Data Tree (DataTree.com).
20
Origination Shares Based on Purchase Loan Counts
The GSEs, FHA, VA, and RHS continue to account for about 80% of institutionally
financed home sales. The GSEs accounted for nearly half of all mortgage lending in
2018:Q4. They have more than regained market share lost to FHA after its mortgage
insurance premium (MIP) cut in January 2015. FHA’s market share is now over 1 ppt.
below its pre-MIP cut level. FHA and the GSEs have all had substantial increases in
mortgage risk.
60% 60%
GSE
50% 50%
40% 40%
30% 30%
Private
20% 20%
FHA
10% VA 10%
RHS
0% 0%
2012:Q4 2013:Q2 2013:Q4 2014:Q2 2014:Q4 2015:Q2 2015:Q4 2016:Q2 2016:Q4 2017:Q2 2017:Q4 2018:Q2
Note: Data are for institutionally financed sales only.

Agency Refi and Purchase Loan Counts
Agency refi volume (by count) for December 2018 was down 46% from December 2017.
The majority of refi lending is now Cash-Out refis, which accounted for 70% of all refis
during the current month. No Cash-Out refi volume has declined sharply with the
increase in mortgage rates in 2017.
800,000 800,000
Cash-out share of refi
Total Red markers show December count in each year. Dec 2012 - 20%
700,000 Dec 2013 - 31% 700,000
Dec 2014- 34%
Dec 2015 - 40%
600,000 Dec 2016 - 41% 600,000
Refis Dec 2017 - 56%
Dec 2018 - 70%
500,000 500,000
400,000 400,000
300,000 No-Cash-Out Refis 300,000

Purchase loans
200,000 200,000
100,000 100,000
Cash-Out Refis
0 0
Note: Data are for Agency loan market only.
22
Cash Out Refi Volume by Agency
Agency cash out volume (by count) for December 2018 was down 46% from
December 2016. Cash-Out refi volume has declined sharply as rates have risen
(albeit declined less than no-cash out refis.) The decline has been particularly
sharp for the GSEs, with FHA and VA having more stable volumes.
140,000 140,000
FHA VA Freddie Fannie

120,000 120,000
Red markers show December counts in each year.

100,000 100,000
80,000 80,000
60,000 60,000
40,000 40,000
20,000 20,000
0 0
Sep-12 Mar-13 Sep-13 Mar-14 Sep-14 Mar-15 Sep-15 Mar-16 Sep-16 Mar-17 Sep-17 Mar-18 Sep-18
Note: Data are for Agency loan market only.
23
Compositional Change of Cash-Out Refis
A closer look at the risk distribution reveals that lower risk borrowers, mostly served by the
GSEs, have been exiting the market as rates have risen, while higher risk borrowers – mostly
served by FHA and VA - have been less sensitive to rate changes.
50%
Percent of loans December 2018 Distribution
Fannie Freddie FHA VA
40%
30%
20%
10%
0%
0-4 4-8 8-12 12-16 16-20 20-24 24-28 28-32 >32
Mortgage Risk Index
15% Percentage points Change in Distribution, December 2016 to December 2018

10%
5%
0%
-5%
-10%
Fannie Freddie FHA VA
-15%
-20%
0-4 4-8 8-12 12-16 16-20 20-24 24-28 28-32 >32
Mortgage Risk Index
24
Agency First-Time Buyer Mortgage Share
The Agency First-Time Buyer Mortgage Share Index (FBMSI) for December 2018 stood
at 59.2%, up 0.8 ppt and setting a new series’ high for the month of December.
Compared to five years ago, the FBMSI is up 3.8 ppts. from 55.4%. It appears that the
index has increased from its already high level due to repeat buyers’ greater sensitivity
to higher rates.
First-time buyer mortgage share
61% 61%
60% 60%
59% 59%
58% 58%
57% 57%
56% 56%
55% 55%
Red markers show December share in each year.
54% 54%
53% 53%
Feb-13 Feb-14 Feb-15 Feb-16 Feb-17 Feb-18

25
Ratio of Sales Price for First-time to Repeat Buyers
The trend upward is towards higher first-time buyer (FTB) prices relative to repeat
buyers (RBs). FTBs have access to the leverage punchbowl, thereby greatly
reducing the tendency to make downward quality adjustments to offset rapid home
price appreciation. RBs without access to this punchbowl, tend to make downward
quality adjustments to offset home price appreciation. This adds to demand at lower
price tiers.
Ratio of FTB to RB sale price
76% 76%
Red markers show ratio in December of each year
75% 75%
74% 74%
73% 73%
72% 72%
71% 71%
70% 70%

26
Median Sale Price by Risk Segment*, FTB Purchase Loans
Higher risk borrowers are being provided additional leverage which is fueling rapidly
increasing home prices. Market prices for subprime borrowers have increased 27
percent since Feb-2013, while market prices for prime borrowers have only increased
12 percent.
Index: Feb-13 == 100
130 130
Subprime
(4.0% avg. annual growth)
NMRI Dec-18: $221,000
125 125
Prime Subprime
Feb-13 3.0% 20.4%
120 Dec-18 3.2% 22.7% 120
Mean price Feb-13

115 Prime $ 270,000 115
Subprime $ 170,000
110 Prime 110

Dec-18: $307,000
105 105
100 100
95 95
Feb-13 Aug-13 Feb-14 Aug-14 Feb-15 Aug-15 Feb-16 Aug-16 Feb-17 Aug-17 Feb-18 Aug-18
* We define prime loans as low-risk (with a stressed default rate of less than 6%), and subprime as high risk (with a stressed
default rate of 12% or greater).
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing. 27
Try Out the NMRI Interactive Tool
Visit https://www.aei.org/housing/mortgage-risk-index/ to explore National Mortgage
Risk Index data using our new interactive tool.
Later this week we will be adding market condition data on the largest 60 metros.
Scroll over “Indexes & Indicators” and then

click on “Mortgage Risk Index” on the drop
down menu
Latest data will go live at 10 am

on the day of the briefing call.
28
Remaining Briefing Dates for 2019
• Next briefing on Monday, April 29.
• The remaining briefings for 2019 are listed below:
Monday April 29
Tuesday May 28
Monday July 1
Monday July 29
August – no briefing
Monday September 30
Monday October 28
Monday November 25
Monday January 6, 2020
• All briefings take place at 11 AM ET.
Please note that this month’s appendix

slides have not been updated.
29
National Mortgage Risk Index (NMRI): A Quick Primer
• Overall goal:
– Monitor market stability through accurate, real-time tracking of leverage that, if left unchecked,
would result in destructive housing booms/busts.
• Principles behind the NMRI

– NMRI is a stress test, similar to a car crash safety rating or hurricane rating for buildings.
– The NMRI’s stress event is the financial crisis from 2007.
• Basics of index construction

– The NMRI is a standardized quantitative index for mortgage risk (leverage)
– Places loans in risk buckets and assesses default risk based on the performance of the 2007
vintage loans with similar characteristics
• Advantages of the NMRI

– Near-complete census of gov’t-guaranteed loans,
– Accurate, timely, and in-depth coverage of purchase mortgage trends
– NMRI provides significant signals of market trends without the noise of other indices
• What does an increasing or decreasing NMRI mean?

– Increasing NMRI = increasing leverage = looser lending
– Decreasing NMRI = decreasing leverage = tighter lending
30
Stressed Default Rates, Home Purchase Loans
Risk Bucket Credit Score 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-
occupied, 30-year fixed-rate, fully amortizing, fully documented, home purchase loans.
• Takeaway: Huge spread of default rates across risk buckets
• All 320 risk buckets for home purchase loans are shown at Periodic Table – Purchase
• Analogous tables for cash-out and no-cash-out refi loans are at Periodic Tables –
Refinance
• Additional loan risk factors are applied to VA loans and to ARMs, investor loans, second
homes, 15 year terms, and 20 year terms
31
Home Sales Methodology
• Data Inputs
– Public Records (near-real time with latency and coverage problems).
– HMDA (annual dataset of institutionally financed sales (IFS); covers around 99% of loans; released with
lag).
– FHA Snapshot (monthly dataset of all FHA endorsements; released around mid-month with a one month
lag).
– National Mortgage Risk Index (NMRI) (covers 99% of Agency loans; two months lag).
• Assumptions
– Recorder offices process transactions in random order; latency in reporting applies equally across all sales
types.
– FHA loans are properly recorded (stamp on mortgage document).
– On average, the difference between loan origination and endorsement is one month. ( We have confirmed
this on aggregate by comparing monthly FHA Snapshot to NMRI counts.)
– Conventional loans have same seasonal pattern as GSE loans.
• Construction
– Aggregation from the county level up.
– Use FHA Snapshot for all FHA sales.
– When HMDA is available: Use HMDA for remaining IFS when available:
• Impute cash and other financed sales as a percent of IFS (assume state average for counties with latency
problems);
• Impute seasonal pattern from either public records or NMRI.
– When HMDA is not yet available: Use Public records with adjustments:
• Limited to ~ 700 counties that account for ~80% of sales (remove counties with insufficient FHA counts or
breaks in series);
• Gross up all sales in that county by the ratio of FHA Public Records loans to FHA Snapshot loans;
• Assume same rate of change for ~2400 counties with ~20% of sales -> still working on improving this
assumption.
• As a robustness check of this, we compare state VA and RHS totals to the NMRI and adjust totals.
32
House Price Appreciation (HPA) Index: A Quick Primer
• Overall goal:
– Monitor market stability through accurate, real-time tracking of house prices.
• Basics of index construction

– Most widely known HPA Indices are repeat sales (i.e. Case Shiller or FHFA) or hedonic (Zillow)
indices.
– AEI’s HPA is a “quasi” repeat sales index with a hedonic element.
– Index measures HPA by constructing an artificial sales pair consisting of one actual sale and one
“artificial” sale as measured by the property’s AVM.
– The AVM (Automated Valuation Model) approximates a property’s sale price at a given point in
time. The AVM used for AEI’s HPA Index is unbiased and accurate.
• Advantages of AEI’s HPA Index:

– Combines the best of repeat and hedonic models.
– Unlike a true repeat sales index, which is limited to repeat sales and may therefore be biased,
AEI’s index includes the entire universe of sales.
– Unlike a true hedonic index, which incorporates every property (even unsold ones), it reduces the
amount of errors since at least one sale of the transaction pair actually occurred.
– Allows for an index construction by price tier and fine geographic levels (down to census tract).
• Data for the HPA index

– National Public Records data and AVM for Dec-2017 come from First American via DataTree.com.
– Uses virtually all institutionally financed sales back to January 2012.
– Data are weighted at the county level to make them representative.
– HPAs for the medium-high and high price tiers are spliced around the time of loan limit changes.
33
New Construction Methodology
• Data Inputs
– Public Records (Deed & Assessor files)
– Zillow API and/or Listings data
• Identification of New Constructions
– Year Built in Assessor data
• We also just received Effective Year Built so going to analyze in future
– If Year Built is missing:
• Seller name (we have assembled a list of over 400 builders with their subsidiaries and key words to identify smaller builders.) If a
seller is a builder and the Year Built is missing, then it is most likely a new construction that has not yet been assessed.
• Ping Zillow API for Year Built and Use Code. Data not perfect, but even some information helps us determine status.
• Sellers with multiple sales that are not individuals/gov’t/lender/other corporation are most likely builders. (Relatively small number.)
– Count only first sale of home as a new construction.
– Still working on identifying owner-built homes without a long lag.
• Verification
– Random sampling and checking of new constructions and existing homes using Zillow data, Google street
view/satellite images.
– Find very few false positives and false negatives. (Random sample found <4% were false positive, <3% of remaining
88% found to be new construction.)
• Final dataset allows us to:
– Monitor new constructions at the property level with minimal lags,
– Accurately estimate new home sales at very fine geographic levels when combined with Home Sales numbers,
– Estimate additions to the existing housing stock when combined with Assessor data,
– Estimate sales by builder and track builder,
– Combine new construction numbers with Months’ Supply and house price appreciation,
– Much more.
34
List of Abbreviations
Term Description
MRI The Mortgage Risk Index (MRI) measures how the loans originated in a given month would perform if
subjected to the same stress as loans originated in 2007, which experienced the highest default rates as
a result of the Great Recession.
NMRI The National Mortgage Risk Index (NMRI) currently covers home purchase and refinance loans
(except for VA refinances) that have been (1) acquired and securitized by Fannie Mae or Freddie Mac or
(2) insured or guaranteed by the Federal Housing Administration (FHA), the Department of Veterans
Affairs (VA), or the Rural Housing Service (RHS).
SMRI The State-level Mortgage Risk Index (SMRI) measures mortgage risk on a state level. It employs
exactly the same stress-test methodology as the national index.
FBMSI The First-time Buyer Mortgage Share Index (FBMSI) equals the number of loans made to first-time
buyers divided by the number of all home purchase loans excluding those made to investors and second
home buyers for any given month (see first-time buyer (FTB) definition below). The agency FBMSI
covers government-guaranteed loans, while the combined FBMSI covers both government-guaranteed
and private-sector loans. The agency loans are from the same database used for the NMRI, while the
private-sector component of the combined FBMSI come from AEI’s National Housing Market Index
(NHMI) and assumptions believed to be reasonable.
FBMRI The First-time Buyer Mortgage Risk Index (FBMRI) is calculated using the same methodology as for
the NMRI. The only difference is that the set of included loans is restricted to first-time buyers.
FTB AEI uses the federal government’s definition of a first-time homebuyer (FTB). A FTB is an individual
borrower who (1) is purchasing the mortgaged property, (2) will reside in the mortgaged property as a
primary residence, and (3) had no ownership interest (sole or joint) in a residential property during the
three-year period preceding the date of the purchase of the mortgaged property. Investment properties,
second homes, and refinance transactions are not eligible to be considered first-time homebuyer
transactions. Other organizations such as the National Association of Realtors (NAR) use a different
definition of FTB based on self-identification.
RB Repeat Buyers (RB) are all home buyers that are not first-time buyers.
35
List of Abbreviations (cont’d)
Term Description
GSE A Government-Sponsored Enterprise (GSE) is an entity created by Congress that operates under a
government-defined mission and charter. There are two housing-related GSEs: Freddie Mac and Fannie
Mae. They purchase mortgages on the secondary market and subsequently pool them into mortgage-
backed securities (MBS), which are purchased by government and private investors.
Fannie Mae The Federal National Mortgage Association (FNMA), known as Fannie Mae, was founded in 1938 as
part of the New Deal legislation.
Freddie Mac The Federal Home Loan Mortgage Corporation (FHLMC), known as Freddie Mac, was created in
1970 to complement Fannie Mae.
Ginnie Mae The Government National Mortgage Association (Ginnie Mae) is a federal government corporation
that aims to promote homeownership for low- and moderate-income families. It ensures the timely
payment of principal and interest on mortgage-backed securities formed from mortgages that are
guaranteed or insured by FHA, VA, RHS, or smaller programs for Native Americans. Ginnie Mae was
created in 1968. Prior to 1968 its role was performed by Fannie Mae.
FHA The Federal Housing Administration (FHA), founded in 1934, is a federal agency that today provides
mortgage insurance for residential loans made to high-risk borrowers. The borrower pays an upfront
mortgage insurance premium as well as monthly insurance premiums for the service. In return, FHA
covers 100% of the lender’s loss in case of the borrower’s default.
RHS The Rural Housing Service (RHS) is a program within the U.S. Department of Agriculture that
guarantees mortgages in rural areas. The borrower pays an upfront annual fee for the service. In return,
RHS covers 100% of lender’s loss in case of the borrower’s default.
VA The Department of Veterans Affairs (VA) guarantees mortgages to eligible veterans and generally
pays 25% of lender’s loss in case of the borrower’s default. The borrower pays an upfront annual fee for
the service.
HUD FHA has been overseen by the Department of Housing and Urban Development (HUD) since its
creation in 1965.
36
Term Description
FICO® The FICO Credit Score is a statistical credit evaluation score developed by Fair, Isaac and Co. The
FICO score attempts to measure a borrower’s risk of default through his or her personal financial history.
FICO scores range from a high default-risk score of 300 to a low default-risk score of 850. The term
“credit score” is used to connote a generic score.
LTV / CLTV The Loan-to-Value Ratio (LTV) is the ratio of the 1st lien loan amount to the property’s value. Since the
down payment on a purchase transaction is the property’s value minus the loan amount, the LTV is
inversely related to the down payment. The Combined Loan-to-Value (CLTV) is the ratio of all loan
amounts at 1st lien origination to the property’s value. Both ratios are a measure of a borrower’s skin in
the game.
DTI The total Debt-to-Income Ratio (DTI) gauges the ability of a borrower to repay a mortgage by
measuring the amount of income consumed for repayment of all outstanding debts of the borrower.
ARM An Adjustable-Rate Mortgage (ARM) is a mortgage whose interest rate varies over the lifetime of the
loan based on market conditions. ARMs have on average a higher default risk than FRMs.
FRM A Fixed Rate Mortgage (FRM) maintains the interest rate at origination throughout the lifetime of the
loan.
MSA A Metropolitan Statistical Area (MSA) is a geographical region with a population of at least 50,000
inhabitants at its core and close economic ties throughout the region.
PCE price index The Personal Consumption Expenditure (PCE) price index measures the prices of goods and
services purchased by consumers in the U.S. economy. It is published monthly by the Bureau of
Economic Analysis in the Department of Commerce. The PCE price index is the measure of inflation
targeted by the Federal Reserve.
SLOOS The Senior Loan Officer Opinion Survey on Bank Lending Practices (SLOOS) is a survey of lending
conditions conducted quarterly by the Federal Reserve among roughly eighty large domestic banks and
twenty-five U.S. branches and agencies of foreign banks.
37
Term Description
QM/QRM The Qualified Mortgage (QM) and the Qualified Residential Mortgage (QRM) are mortgage terms
created under the Dodd-Frank Act. A mortgage that meets the QM requirements provides legal
protection for lenders against a claim that the loan was made without due consideration of the
borrower’s ability to repay. The QRM designation relates to the securitization of mortgages. If the loans
in a mortgage-backed security are QRMs, the securitizing agent is not required to retain any risk position
in the security. Although the initial proposed QRM definition was relatively strict, the final definition was
watered down to be equivalent to the looser QM definition. The five guarantee agencies (Fannie Mae,
Freddie Mac, FHA, VA, and RHS are exempt from substantial portions of the QM rules and entirely from
the QRM rules. (For Fannie and Freddie, this exemption applies only while they are in conservatorship).
MIP The Mortgage Insurance Premium (MIP) is a payment to compensate for the risk of default on the
mortgage. As noted above, FHA mortgages carry both upfront and monthly MIP payments. Fannie Mae
and Freddie Mac generally require mortgage insurance for loans they guarantee with LTVs above 80%;
borrowers with these GSE-guaranteed loans may make monthly MIP payments depending on the
premium plan.
TRID The TILA-RESPA Integrated Disclosure (TRID) rule – commonly also known as Know Before You
Owe – requires lenders to summarize and more prominently display the loan terms on the mortgage
form. It also institutes a three-day waiting period before closing to allow borrowers time to review the
contract. The form change is currently suppressing sales volume as it is delaying loan closings by
creating additional burdens on lenders. TRID was mandated by the Consumer Financial Protection
Bureau (CFPB) and applies to mortgage applications filed on or after October 3, 2015.
38
Appendix
Additional slides and those not included every month:
– National Mortgage Risk Index: Loan Totals
– Background: Financial Crisis and AEI’s Response
– Principles of Housing Finance over 125 years (1850-1975)
– Pinto’s Principles of Housing Finance
– Definition of Low-Risk Loans
– Current State of the Housing Market
– Recent Steps by the GSEs, the FHA, and Regulators Add Fresh Fuel to the Long Running and Accelerating House Price Boom
– Agency Origination Shares, Purchase Loans
– Changes in the >95% CLTV Purchase Loan Market
– Agency Origination Shares, Cash-Out Loans
– Agency Origination Shares, No Cash-Out Loans
– Agency Total, Refi, and Purchase Loan Counts
– DTI Distributions, Agency Primary Purchase Loans
– Prime, Subprime, and Nearprime, Purchase Loan Counts
– GSEs: Large Lender Market Share and Relative Risk Share, Purchase Loans
– FHA: Large Issuer Lender Type Market Share and Relative Risk Share, Purchase Loans
– Combined, Purchase, and Refi NMRIs
– No-Cash-Out Refi and Cash-Out Refi NMRIs
– FHA’s Pro-Cyclical Policies Continue to Fuel the Boom
– Which Risk Factors Have Driven Up the Purchase NMRI?
– FHA’s NMRI for Home Purchase and Refinance Loans
– What explains FHA’s riskiness?
– How wide is the FHA credit box?
– Risk Overlap between FHA and the GSEs
– FHA NMRI by Risk Decile, Home Purchase Loans
– FHA Median Downpayment and Sales Price
– FHA’s Should Begin Reining in Its Pro-Cyclical Policies
39
Appendix (cont’d)
– BCFP’s Pro-Cyclical QM Patch Continues to Fuel the Boom
– Pro-Cyclical Parallels to the Last Boom
– Purchase Loans with Total DTI Greater than 43%
– It Is Time for the BCFP to Announce that the Temporary GSE QM Patch Will Be Allowed to Sunset in January 2021
– The Role of Leverage
– Ratio of Sales Price for First-time to Repeat Buyers
– Median Sale Price by Market Segment, FTB Purchase Loans
– Measure Market Behavior in Four Leverage Based Price Tiers
– House Price Trends Impacted by Leverage
– Constant-quality Prices by Guarantor Type: Low Price Tier
– High risk home purchase lending is fueling home price appreciation
– Scatterplots: Introduction
– Strong Positive Correlation Between Mortgage Risk & Home Price Appreciation
– Strong Positive Correlation Between Mortgage Risk & Tract Income
– Measured Steps Now Would Moderate Unsustainable Home Price Increases, Not Lead to Home Price Declines
– DTI Distributions, GSE & FHA Purchase Loans
– Share of Fannie Purchase Loans by DTI Bucket
– RHS Reduced Borrower DTIs from 2013 to 2018, while the FHA Kept Increasing DTIs
– The FHA’s and the GSEs’ Rising DTIs Have Been Pro-Cyclically Fueling the House Price Boom
– Origination Shares Issuer Lender Type,
FHA and RHS Purchase Loans
– MRIs by Issuer Lender Type,
– Origination Shares and MRIs by Seller Lender Type,
GSE Refinance Loans
– Origination Shares and MRIs by Issuer Lender Type,
FHA Refinance Loans
– State NMRI and FHA Share, Purchase Loans
– State NMRI Change, Purchase Loans
40
Appendix (cont’d)
– Pricing Changes, Home Purchase Loans
– Stressed Default Rates by Loan Type
– Median Downpayments
– Volume Growth in Counts and Dollars, Purchase Loans
– A closer look at RHS’ October 2016 MIP cut
– No-Cash Out Refi Demand and 30-yr Mortgage Rate
– Low-Risk Origination Shares, Purchase Loans
– Calibrating Mortgage Safety
– Credit Conditions: 1990 to 2013-14
– Role of Income Leverage During Housing Boom
– Fed Tightening and Efforts to Maintain Buying Power
– Appraisals Should Be the Guard Rail Against Speculative Booms
– Cross-subsidies Return to the GSEs
– Change in Agency Purchase Loan Volume
– GSEs: Ratio of NMRI for Loans with Total CLTV > 95% to Loans with Lower Total CLTVs
– FICO® Score Distribution
– Median Credit Score on Primary Purchase Loans
– Aggregate Default Risk Surge for Home Purchase Loans Is Over Five Years Old
– The Effect of April 2016 PMI Price Change
– Credit Score Distribution & MRIs, Purchase Loans
– Purchase Loans with Down Payment of 5% or Less
– Composite Origination Shares and MRIs by Channel, Purchase Loans
– Large Bank Origination Shares and MRIs by Channel, Purchase Loans
– Fed’s Senior Loan Officer Survey is Badly Flawed
– Urban Myth: Tight Credit Keeping
“Creditworthy” Borrowers Out of Market
– FHA Perpetuates This Myth
41
Appendix (cont’d)
– FHA Is All about Moral Hazard
– While FHA’s Capital Reached Required 2% Statutory Level for 1st Time since 2008, It Is Insufficient
– Share of States with Increase in SMRI for Purchase Loans from Year-Earlier Period
– CBSA NHMI: Investor Type for Home Purchase Loans
– Riverside/San Bernardino: A Case in Point
– House Price Volatility, 51 Largest Metro Areas
– Median Values of Risk Factors by Loan Type
– Risk Shares for Home Purchase Loans
– DTI Distributions and MRIs, Primary Purchase Loans
– Cash-Out Share and Home Equity
– Agency Cash-Out Share and Defaults
– Nonbank Origination Shares and MRIs by Channel, Purchase Loans
– Greater House Price Volatility at the Lower End
– Tale of two markets: Low end and entry level and high end repeat buyers (high price tier)
– Tale of two markets: Low end and entry level and high end repeat buyers (low price tier)
– Comparison: Home Sale Transactions (New and Existing)
– Home Sales, by Metro Market Size
– Mortgage Risk Indices by Lender Type, Purchase Loans
– Jumbo portfolio-GSE spreads (in bps)
– Homeowners Can’t Count on House Price Gains to Build Wealth
– Evaluating the GSEs 2017 Business
– Evaluating the GSEs 2017 Business (cont.)
– Evaluating the GSEs 2017 Business (cont.)
– Update: John Burns Intrinsic Home Values
– GSEs: Large Lender Market Share and Relative Risk Share, Refinance Loans
– FHA: Large Issuer Lender Type Market Share and Relative Risk Share, Refinance Loans
– Leverage Fueled Housing Demand Continues to Climb
42
– Agency Origination Shares by Risk, Purchase Loans
Appendix (cont’d)
– What Does this Mean for the Broader Market?
– Borrowing at the Conforming Loan Limit, GSE Purchase Loans
– The CFPB’s Qualified Mortgage Policy and GSE QM Patch Allowed for Credit Easing While Supply Is Constrained, a Direct and
Continuing Cause of the Current House Price Boom
– Number of Investors Flipping Houses Creeping Up
– Average House Price Change by Zip (%, annual avg.)
– Raising the Conventional Loan Limit – A Prediction
– Raising the Conventional Loan Limit – A Good Idea?
– Cash-Outs and the Economy
– Cash-Out Share and Expected Defaults
– FTB slides begin
– Punchbowl 1: Mortgage Rate Changes Applicable to
FTBs and RBs
– Punchbowl 2: Eased Underwriting Standards
Only Available to Agency First-time Buyers
– Agency Purchase Loan Demand Remains Strong
– Market Segmentation:
Median Sales Price for First-time and Repeat Buyers
– Constant Quality Prices Outlook for the Bifurcated Market:
Slowing Price Appreciation for at the Higher End, Continued Robust Appreciation for at the Lower End
– Outlook for Bifurcation of Market – Quality Changes
– Outlook for a Bifurcated Market –
Transaction Prices by State & Changes in Transaction Volume
– Outlook for a Bifurcated Market – Transaction Prices by State & Changes in Median Transaction Prices
– While FHA’s Forward Program Capital Is at 3.9%, in Excess of Statutory Minimum of 2%, It Should be 7%
– FHA Cash Out Count and MRI
– Average Credit Score and DTI: FHA Purchase Loans
– FTB Purchase Loan NMRI: Credit Easing Continues
43
Appendix (cont’d)
– Which Risk Factors Have Driven Up the FTB NMRI?
– Share of GSE FTB Purchase Loans w. DTIs of 46-50%
– FTB Purchase Loans, by Level of Downpayment
– Agency First-time Buyer Purchase Loan Share
– Government Housing Policy Creates an Economics Free Zone
– Definition of Low-Risk / Prime Loans
– Agency First-Time Buyer Loan Count
– Agency Origination Shares, FTB Purchase Loans
– Origination Shares by Credit Score Bin, First-time Buyer Purchase Loans
– Agency Origination Shares, FTB Purchase Loans by Market Segment
– Originations by Market Segment, FTB Purchase Loans
– Combined First-Time Buyer Mortgage Share Index
– The NAR’s first time buyer series is fatally flawed. After removing seasonality, most of what remains is noise
– Changes in the >95% CLTV Purchase Loan Market
– Share of States with Rise in First-time Buyer Loan Volume and Share from Year-Earlier Period
– Profiles of GSE and FHA First-time Buyers with >95% CLTV
– Characteristics of Mortgages Taken Out by First-Time and Repeat Homebuyers
– Rising Prices Have Disparate Effects on Buyers
– Agency-Specific First-Time Buyer Mortgage
Share Indices
– DTI Distributions, Agency FTB Purchase Loans
– The Effect of FHA Mortgage Insurance Premium Cut
– Income and Debt Growth by Income Group: FHA Purchase Loans
– House Price Appreciation (HPA) by Price Tier
– House Price Appreciation (HPA) by Price Tier: 73 Metros
– Supply-Demand Imbalance in the Market Is Driving Prices Up
– Affordability Worsens in a Seller’s Market
44
Appendix (cont’d)
– Fannie vs. Freddie Risk Index, GSE Purchase Loans
– History Repeats Itself: the “Quiet” Battle for Subprime (High Risk >95 CLTV Purchase Loans) among Fannie, Freddie & FHA
– Leverage Fueled Housing Demand Pauses Due to Higher Rates
– Agency First-time and Repeat Buyer Mortgage Risk Indices
– Origination Shares and MRIs by Seller Lender Type, GSE Purchase Loans
– Origination Shares and MRIs by Issuer Lender Type, FHA Purchase Loans
– Mortgage Risk Indices by Lender Type, Refi Loans
45
National Mortgage Risk Index:
Loan Totals
• The November 2018 NMRI covers over 35.9 million agency loans dating
back to Sept. 2012. These data are used to construct the NMRI, First-Time
Homebuyer Indices, and the National Housing Market Indexes (NHMI).
• This total consists of nearly 18.2 million agency purchase loans and over
17.7 million agency refinance loans
• NMRI and other risk indices published for:

– Purchase loans, with separate indices for first-time and repeat buyers
– Refinance loans, with separate indices for no-cash-out and cash-out refis
– Composite of purchase and refinance loans
– Purchase loan NMRI is the primary measure for monitoring mortgage risk and the
impact of housing policy, particularly with respect to first-time buyers
– Refinance loan NMRI contributes to overall assessment of changes in leverage
46
Background: Financial Crisis and AEI’s Response
• Financial crisis largely stemmed from a failure to understand

buildup of housing risk:
– Mortgage risk
– House-price (collateral) risk
– Capital adequacy
• AEI’s Center on Housing Markets and Finance (AEI.org/housing)
addresses this problem by undertaking evidence-based research
that expands the body of knowledge concerning housing markets
and finance:
– Provides objective and transparent mortgage risk measures
• Risk indices published monthly
– Provides objective and transparent housing market indicators
• Market indicators published quarterly
– Provides objective and transparent house price appreciation measures
47
Principles of Housing Finance over 125 years (1850-1975)
•At all times, but especially in the last few years, people have dreamt of universalizing wealth by universalizing
credit.… Now, in no country is it possible to transfer from one hand to another more products than there are(1850)1
•Since value depends on location, & location on convenience, & convenience on nearness, the intermediate steps
may be eliminated & say that value depends on nearness. (1903)2
•If a new utility does not arise, [sales] prices may advance & recede, while intrinsic values do not change. If a new
utility arises, both [sales] prices & intrinsic values will alter their levels. (1903)2
•[s]peculative elements cannot be considered as enhancing the security of residential loans [rather they] enhance
the risk of loss to mortgagees [if] permit[ed] to creep into valuations….(1938)3
•Because situations of scarcity [seller's market] or over-supply [buyer's market] do not last indefinitely they cannot
be considered as phenomena the affect valuations for long-term use…. & not truly indicative of value for mortgage
insurance purposes. (1947)4
•The sequence of [market cycle] events is fairly predictable, though the period of the phases of the cycle & the
amplitude of the variations are not subject to dependable forecasting. (1949)5
•[I]nflationary construction costs, home purchase prices, & land prices not only loan disproportionate financial
burdens upon the owners at time of acquisition but also form the bloated base upon which the major costs of
occupancy [including property taxes] are determined for the entire term of ownership. (1949)5
•The essential nature of housing demand is changeability; the nature of housing supply is rigidity. (1949) 4
•[I]n a seller's market, when choice is restricted & the seller virtually dictates sales terms, more liberal credit is likely
to be [capitalized] in price with probably a reduction in housing standards. (1951)6
•[Transitioning] from buyer's to seller's market, maximum terms become so commonly used they tend to be
considered the minimum. (1951)6
•The parallel between the increases in the “costs” of new housing units & increases in the amount & percentage of
needed funds that could be obtained by lengthening their terms & [reducing] downpayments raises the radical
question of whether the disbursements made to assist purchasers & (renters) have not benefited others more than
those whom they were intended to relieve. The largest groups to whom it is sometimes suggested some of the
benefits may have flowed are the builders, building labor, the suppliers of building materials, & real estate brokers
& speculators. (1975)7
1 Frederic Bastiat, What Is Seen and What Is Not Seen, 1850, 2 Hurd, The Principles of City
3 FHA 1938 Underwriting Manual (main authors: Ernest Fisher and Frederick Babcock)
Land Values, 1903,
4 FHA 1947 Underwriting Manual
5 Ratcliffe, Urban Land Economics, 1949
6 Fisher, Financing Home Ownership, NBER, 1951 48
7 Fisher, Housing Markets and Congressional Goals, 1975
Pinto’s Principles of Housing Finance
1. Corollary to Fisher’s capitalization rule: capitalization is added to land price
2. Uncertainty Principle: Can’t simultaneously set an asset’s credit risk & risk weight
– A low risk designation and corresponding low capital weight (greater leverage) unleashes
demand pressures causing it to no longer be low risk (think GSEs, private MBS, Greek
sovereign debt)
3. Dual Underestimation Principle: Never underestimate the government’s

willingness & ability to (i) add leverage to stimulate the market & (ii) ignore its
impact on raising home prices and default risk under stress
– Housing debt & default risk have increased with over 60 years of housing policies focused
on increasing leverage
4. Law of the Marginal Buyer: Home prices will keep rising so long as the marginal
buyer, who sets the price for all, has access to higher leverage (see #3)
5. Corollary: Historically the government has endeavored to add leverage in both
buyer’s & seller’s markets; but the latter has potential for dangerous buildup of
risk (see #1)
– Result is an economics free zone promoting demand, while supply is restricted by regulation
• FHA neither prices nor underwrites for risk
• Government policies increase leverage regardless of rates going up or down
• Low capital entities (FHA and GSEs) compete with each other over loosening credit
49
• Affordable housing goals and duty to serve policies promote risky lending
Definition of Low-Risk Loans
• We define low-risk loans as those with a stressed default rate of less

than 6%. Why?
• Low-risk definition calibrated from two sources

– Original QRM proposal to implement Dodd-Frank
– FHA underwriting standards over 1935-55
– Both yield an average stressed default rate of ≈ 3%
• This is consistent with a maximum stressed default rate of ≈ 6% on

individual loans, assuming a uniform distribution starting near 0%
• Hence the use of 6% as the highest stressed default rate for a low-risk
loan
50
Current State of the Housing Market
• The current house price boom is about 6 years old and rate of house price
increases is accelerating
• “Home Values Climbing at Fastest Rate in 12 Years….The median U.S. home value rose 8.7 percent to
$215,600 in April, the fastest year-over-year climb since June 2006” Zillow, 5.24.17
• “Start of year sees strongest home price growth since 2005. … About 60% of all U.S. metros saw an
acceleration in the rate of price increases through February this year.” (Housing Wire, 5.7, 2018)
• “Housing confidence hits record high as home prices skyrocket. Consumer confidence in housing jumped to
its highest level on record in April, according to Fannie Mae. Those who think home prices will move even
higher rose the most, and those who think now is a good time to sell came in second.” (CNBC, 5.7.18)
• “Mortgage lenders are making it easier for you to buy a house. But are they repeating last decade's
mistakes? Dana Wade, the acting Federal Housing Administration commissioner, minced few words in
testimony last month before a U.S. House of Representatives committee. The FHA, the federal housing
agency that insures mortgages made to first-time and lower-income buyers, has seen “certain trends and
indicators of potential defaults.” Philadelphia Inquirer 5.4.18
• A house price boom is when prices rise faster than fundamentals
• The boom is driven by too much money chasing too few properties
– When the market is supply constricted (a seller's market), credit easing is likely to be capitalized in price.
– FHA, Fannie, Freddie, and the VA are all pro-cyclically fueling the boom
• The length and acceleration of the boom adds urgency to shrink the GSEs and FHA
by administrative action
51
Recent Steps by the GSEs, the FHA, and Regulators Add Fresh
Fuel to the Long Running and Accelerating House Price Boom
• “Freddie Mac takes aim at FHA with widespread expansion of 3% down mortgages…. But now, Freddie
Mac is about to supercharge its 3% down program and launch a widespread expansion of the offering.”
(Housing Wire, 4.26.18)
• “Credit scores may jump starting this month…. Because of improved standards [from regulators] for
utilizing new and existing public records, the three major credit reporting companies are now excluding all
tax liens from credit reports. That means some scores will head higher, for some by as much as 30 points.”
(CNBC, 4.12.18)
• “Manufactured housing giant endorses HUD's call for regulatory relief…. But the FHA has suffered major
losses from insuring manufactured loans in the past and is unlikely to increase its role in this sector.”
(National Mortgage News, 4.3.18)
• “Will The Gig Economy Change Mortgage Lending?...[r]ather than two years of iron-clad documentation,
[GSEs] now say as little as 12 months of self-employment are enough, as long as the applicant’s previous
employment is in the same field and his or her income remains steady.” (Mortgage Orb, 7.26.17)
• “If the lender obtains documentation to evidence the actual monthly payment is $0, the lender may
qualify the borrower with the $0 payment as long as the $0 payment is associated with an income-driven
repayment plan.” (Fannie Mae Selling Guide, 7.25.17)
• “Fannie Mae will ease financial standards for mortgage applicants next month… Fannie will be raising its
DTI ceiling from the current 45 percent to 50 percent as of July 29.” (Washington Post, 6.6.17)
52
Agency Origination Shares, Purchase Loans
Major market shifts are often related to pricing changes. The largest effect was from
FHA’s mortgage insurance premium (MIP) cut in January 2015, which boosted FHA’s
market share from 23% to 30%. Recently, FHA’s share has declined, returning FHA
back to its pre-MIP cut level. As Freddie’s MBS execution price has improved, it has
recently picked up share.
45% 45%
40% 40%
Fannie
35% 35%
30% 30%
Freddie
25% 25%
20% FHA 20%
15% 15%
VA
10% 10%
5% RHS 5%
0% 0%
Sep-12 Feb-13 Jul-13 Dec-13 May-14 Oct-14 Mar-15 Aug-15 Jan-16 Jun-16 Nov-16 Apr-17 Sep-17 Feb-18 Jul-18

53
Changes in the >95% CLTV Purchase Loan Market
While FHA continues to dominate this market segment with a 73% share, this is down
from a 92% share in Oct. 2015. Fannie continues its dominance over Freddie, coming in
at a 20% share, up from 7% in Oct. 2015, compared to Freddie’s 7%, up from 1% in Oct.
2015. This is contributing to Fannie’s Risk Index sprinting ahead of Freddie’s.
Number of loans with CLTV > 95% Text not updated

100,000 100,000
Total (FHA, Fannie, Freddie)
75,000 75,000
FHA
50,000 50,000
25,000 25,000
Fannie
Freddie
0 0
Sep-12 Apr-13 Nov-13 Jun-14 Jan-15 Aug-15 Mar-16 Oct-16 May-17 Dec-17 Jul-18
Note: Excludes loans made by VA and RHS.

54
Agency Origination Shares, Cash-Out Loans
Market share for cash-out refis has shifted from the GSEs to the FHA and VA. FHA
and VA accounted for less than 10% of market share in 2012. In October 2018, they
accounted for 33%, with FHA’s share surging recently. This increase has powered the
increase in the riskiness in the cash-out index.
70% 70%
Nov 2018 Cash-Out NMRI
Composite 15.8%
Fannie 11.4%
60% Freddie 10.3% 60%
FHA 27.1%
VA 23.4%
50% 50%
Fannie
40% 40%
Freddie
30% 30%
20% 20%
FHA
VA
10% 10%
RHS
0% 0%
55
Agency Origination Shares, No Cash-Out Loans
VA and FHA were both losing market share as early as 2018. However, the trend
between the two agencies diverges around May 2018, right around the time the VA
was subjected to a new statute designed to reign in predatory no cash-out refi
lending. The VA share is now near its series’ low dating back to October 2013.
70% 70%
Nov 2018 No Cash-Out NMRI
Composite 12.1%
Fannie 10.4%
60% Freddie 8.0% 60%
FHA 26.8% Fannie
VA 22.8%
50% 50%
40% 40%
Freddie
30% 30%
20% 20%
FHA
10% 10%
RHS VA
0% 0%

56
Agency Total, Refi, and Purchase Loan Counts
All agency volume was down 29% from a year ago. Refi volume has slowed since the end
of 2016. As interest rates have moved sharply higher since November 2016, refi volume,
and especially no-cash out refi volume, has contracted.
800,000 800,000
Cash-out share of refis
Total Nov-12: 19%
700,000 Nov-13: 30% 700,000
Nov-14: 33%
Nov-15: 40%
Red markers show November count in each year.
Nov-16: 37%
600,000 Nov-17: 55% 600,000
Refis Nov-18: 71%
500,000 500,000
400,000 400,000
No-Cash-
300,000 Out Refis 300,000
Purchase loans
200,000 200,000
100,000 100,000
Cash-Out Refis
0 0
Sep-12 May-13 Jan-14 Sep-14 May-15 Jan-16 Sep-16 May-17 Jan-18 Sep-18

57
DTI Distributions, Agency Primary Purchase Loans*
DTIs have been shifting higher as the rise in house prices has been outpacing
income gains. The share of DTIs below 34% has declined sharply, offset by a much
greater share of DTIs above 40%. While bullish for home prices in the near term,
this presents long-term sustainability problems for both homeowners and the FHA.
6%
DTI Distribution for Entire U.S. Shift in DTI Distribution
5% DTI Feb. 2013 Nov. 2018Difference
<34% 39% 25% -13 ppts
4%
34-40% 26% 22% -3 ppts
3% >40% 36% 53% 17 ppts
Feb. 2013
2%
GSE November 2018
1%
Ginnie November 2018
0%
<20 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65
California shows how the shift could intensify as affordability worsens.

8%
DTI Distribution, November 2018 DTI CA U.S. ex. CA Diff (CA - U.S. ex. CA)
<34% 14% 27% -13 ppts
6% 34-40% 19% 23% -4 ppts
>40% 66% 49% 16 ppts
4%
U.S. excluding CA
2%
GSE California
Ginnie CA
0%
<20 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65
*Data pertain to all agency purchase loans for primary owner-occupied properties. 58
Prime, Subprime, and Nearprime, Purchase Loan Counts
While growth in purchase lending volume has paused, it has not paused equally across
the risk spectrum. Historically, most of the growth in volume has come from the near-
prime and especially the subprime segment.
160,000 160,000
140,000 140,000
120,000 120,000
Prime
100,000 100,000
80,000 80,000
Subprime
60,000 60,000
40,000 40,000
Near-prime
20,000 20,000
0 0
Sep-12 May-13 Jan-14 Sep-14 May-15 Jan-16 Sep-16 May-17 Jan-18 Sep-18

Note: Prime loans are defined as having a stressed default rate less than 6%; near-prime loans are between 6 to 12%; subprime loans are greater than 12%. 59
GSEs: Large Lender Market Share and Relative Risk Share, Purchase Loans
Large banks Not updated Large nonbanks
2018/ 2018/ 2018/ 2018/
2013 2014 2015 2016 2017 Q1-Q3 Oct. 2013 2014 2015 2016 2017 Q1-Q3 Oct.
NMRI 5.4% 6.0% 6.2% 6.5% 6.8% 7.3% 7.4% 5.4% 6.0% 6.2% 6.5% 6.8% 7.3% 7.4%
Pennymac
Wells Fargo
Quicken Loans
JP Morgan Chase
United Shore
Caliber
U.S. Bank Home
Fairway
Independent
Flagstar Bank
Lakeview
Citizens Bank Nationstar
Freedom
Mortgage
USAA
Guild
BB&T
Amerihome
Franklin
Bank of America American
LoanDepot
30% 20% 10% 5% 1% Larger circle represents larger New Penn
market share. Lenders shown Financial
represent the 8 largest banks
and 15 largest nonbanks by Ditech
origination share in 2017.
25+% 15 to 25% 5 to 15% 5 to -5% -5 to -15% -15 to -25% -25+% Money Source
Higher GSE risk share Lower GSE risk share 60

(relative to market share) (relative to market share)
FHA: Large Issuer Lender Type Market Share and Relative Risk Share, Purchase Loans
2018/ 2018/ 2018/ 2018/
2013 2014 2015 2016 2017 Q1-Q3 Oct. 2013 2014 2015 2016 2017 Q1-Q3 Oct.
NMRI 22.0% 24.1% 23.8% 24.6% 26.0% 27.8% 28.2% 22.0% 24.1% 23.8% 24.6% 26.0% 27.8% 28.2%
PennyMac
Wells Fargo
Lakeview
US Bank
Amerihome
Freedom
Citizens
Mortgage
Caliber
Flagstar
Nationstar
JPMorgan
Chase Quicken Loans
Ditech
BB&T
United Shore
Bank of
LoanDepot
America
Money Source
Guild
30% 20% 10% 5% 1% Larger circle represents larger
market share. Lenders shown New Penn
and 15 largest nonbanks by
origination share in 2017. Fairway
Independent
25+% 15 to 25% 5 to 15% 5 to -5% -5 to -15% -15 to -25% -25+%
Higher FHA risk share Lower FHA risk share 61

Combined, Purchase, and Refi NMRIs
The Combined Purchase and Refi NMRI set a series’ high in November. There has been
a sharp trend reversal on refis, which tend to follow feast-and-famine cycles depending
on the mortgage rate. The Refi series is pulling away steeply from the Purchase one
after having converged at elevated levels.
15% 15%
Red markers show November stressed default rate in each year. Refi NMRI
14% Purchase and 14%

Refi NMRI
13% 13%
Purchase NMRI
12% 12%
11% 11%
10% 10%
9% 9%
8% 8%

62
No-Cash-Out Refi and Cash-Out Refi NMRIs
The Refi NMRI set a series’ high in November, powered by the rapidly increasing
Cash-Out index. The Cash-Out NMRI has more than doubled since the start of the
series, and now exceeds both No-Cash-Out by 3.8 ppts and Purchase NMRIs by 2.7
ppts. Cash-Out NMRI is largely driven by growth in volume and risk on FHA and VA
guaranteed loans. Stressed default rate
16% 16%
Red markers show November stressed default rate in each year. Cash-out NMRI
14% 14%
Refi NMRI
12% 12%
No-cash-out NMRI
10% 10%
8% 8%
6% 6%

63
FHA’s Pro-Cyclical Policies Continue to Fuel the Boom
• Pro-cyclical policies support the housing market when the market is going
up, and withdraw support when the market is going down. Therefore,
such policies push the market further away from its long-term mean,
which ends up prolonging booms and worsening busts.
• FHA’s mortgage risk indices jumped in Nov. setting new series’ highs
– Purchase index at 28.5%
– Refi index at a high for the month of November, with the Cash-Out Refi NMRI at a series’
high.
• Higher NMRI indicates FHA continues to increase leverage to maintain
levels of mortgage activity and to further their “affordable housing”
mission.
– FHA’s credit box is wide, therefore credit for entry-level buyers is not tight.
– FHA continues to loosen at a breath-taking pace.
– FHA is adding mostly high risk borrowers, whose risk index keeps climbing through the
effects of risk layering.
64
Which Risk Factors Have Driven Up the Purchase NMRI?
Not updated
Freezing FHA at its
Primary home purchase loans
Oct-12 shares
Risk factor Oct-12 Oct-14 Oct-16 Oct-18 Oct-18
Credit score < 660 13% 16% 16% 18% 13%
DTI > 43% 21% 24% 27% 38% 33%
CLTV ≥ 95% 57% 56% 58% 58% 58%
30-year term 94% 94% 95% 96% 95%
Risk layering 20% 26% 28% 34% 30%
*Risk layering is defined as having at least 3 of the 4 features presented in the table above present in a loan.
Note: Calculated for primary home purchase loans with a government guarantee and reported risk factor. Data for the last column
hold FHA’s shares for each risk factor constant at their Oct-2012 level, thereby assuming no credit easing for FHA.
• Since 2013, all the key risk factors have contributed, which has magnified the
effect on the NMRI through risk layering. An increasing share of loans have:
– Subprime credit scores
– High DTIs
– High CLTVs
– 30-year terms
• Major upward moves by each are in red font.
• Over the past two years, DTIs have moved higher, promoting risk layering. 65
FHA’s NMRI for Home Purchase and Refinance Loans
All of FHA’s indices have consistently been trending up since early-2013 (earliest
data available). For comparison purposes, Rural Housing Services’ Purchase
MRI has been flat. Unless FHA makes policy changes, its current credit box will
continue to lean into the current housing boom, thereby leading the way in the
promotion of unsustainable home price increases.
30%
Stressed default rate 30%
FTB
28% 28%
Refi CO
26% 26%
Repeat Buyer
Refi NCO
24% 24%
22% 22%
20% 20%
RHS FTB
18% 18%
16% 16%

What explains FHA’s riskiness?
Primary Home Purchase Loans

FHA Rest of Agency
Risk factor Nov-16 Nov-18 Nov-16 Nov-18
Credit score < 660 36% 46% 8% 9%
DTI > 43% 48% 60% 20% 32%
DTI > 50% 19% 30% 3% 4%
CLTV ≥ 95% 91% 90% 45% 47%
30-year term 99% 100% 93% 94%
% high risk loans 88% 93% 24% 29%
Note: Calculated for primary home purchase loans with a government guarantee and reported risk factor.
Source: AEI, Center on Housing Markets and Finance, www.aei.org/housing/.
• Across all risk factors FHA is more risky than the rest of the Agency Market.
• Over the past 2 years an increasing share of FHA loans has had higher DTIs
and lower credit scores.
• With term and CLTV basically maxed out, further FHA loosening will have to
come from subprime credit score borrowers (<660) or higher DTIs.
67
How wide is the FHA credit box?
FHA borrowers are the marginal borrowers. FHA’s credit box is wide and its riskiest
portions are being used more and more. It spans as low as a 580 credit score, as high
as a 57 DTI, and generally a 98.2 CLTV. In addition, about 1/3 of FHA borrowers make no
downpayment. Therefore, the credit box for the marginal buyer is not tight, it is loose.
FHA Primary Home Purchase Loans

Credit Score DTI (in %) CLTV
Percentile
Nov-12 Nov-18 Nov-12 Nov-18 Nov-12 Nov-18
5 631 600 54 57 99 103
10 642 614 52 56 99 99
25 659 636 47 52 99 99
50 / median 688 664 41 46 99 99
75 731 699 34 39 97 99
90 770 737 27 32 95 95
average 697 670 40 45 97 98
Share that received

25 32
downpayment assistance
*2018 data for downpayment assistance are from September.

Source: AEI Center on Housing Markets and Finance, www.aei.org/housing, and FHA Snapshot data.
68
Risk Overlap between FHA and the GSEs
While there is a clear separation between FHA and the GSEs at the high and low ends
of the risk spectrum, there is substantial competition for borrowers with MRIs between
8-20%. Over the past year, the GSEs have moved out the risk curve and therefore
gained market share from FHA. On the other hand, FHA has even further moved out its
risk curve and has therefore been picking up borrowers with MRIs > 32%.
Percent of loans November 2018 Distribution

50
40
GSEs FHA
30
20
10
0
0-4 4-8 8-12 12-16 16-20 20-24 24-28 28-32 >32
Percentage points Change in Distribution, November 2017 to November 2018

20
16
12
8 GSEs FHA
4
0
-4
-8
0-4 4-8 8-12 12-16 16-20 20-24 24-28 28-32 >32
Source: AEI Center on Housing Markets and Finance, www.aei.org/housing. 69

FHA NMRI by Risk Decile, Home Purchase Loans
All FHA loans are increasing in risk, but alarmingly, it is the riskiest FHA loans that
are getting even riskier. As house prices and leverage continue to rise, it will be
largely borrowers at the lower end of the market that will continue to add on risk
and drive up house prices for everyone.

50% 50%
Nov-2018
45% 45%
40% 40%
35% 35%
Nov-2012
30% 30%
25% 25%
20% 20%
15% 15%
10% 10%
5% 5%
0% 0%
1 2 3 4 5 6 7 8 9 10
Risk Decile (1 = lowest, 10 = highest)
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing.

70
FHA Median Downpayment and Sales Price
With both equity and income leverage increasing, the long running boom in home
prices is not only shows no signs of abating, but is rather accelerating. One sign of
growing equity leverage is the fact that for home buyers guaranteed by FHA, the dollars
of initial equity has stayed roughly the same since September 2012, while home prices
over the same period have increased by 27%.
$220,000 $10,000
Median Sales
$200,000
Price (left axis) $8,000
$180,000
$6,000
$160,000
$4,000
$140,000 % of FHA borrowers receiving
downpayment assistance: Median
Oct-12: 25% Downpayment
Oct-13: 30% (right axis) $2,000
$120,000 Oct-14: 33%
Oct-15: 35%
Oct-16: 33%
Oct-17: 32%
Sep-18: 31%
$100,000 $0
Note: In April 2017, Ginnie Mae started including the FHA upfront mortgage insurance premium in the LTV. Due to this switch and lagged reporting
of loans for March 2017, this month’s median downpayment is imputed by averaging the median downpayments for February and April 2017, which
are largely unaffected by Ginnie Mae’s reporting change.
Source: AEI Center on Housing Markets and Finance, www.aei.org/housing. 71
FHA’s Should Begin Reining in Its Pro-Cyclical Policies
• Start by taking immediate steps to reduce the risk posed to it and its
borrowers by an excessively risky credit box:
– Eliminate DTIs above 50% on 30-year term loans with credit scores below 660
– Reduce seller concessions to 3% on 30-year term loans
– Eliminate cash out refinances
– Crowd in 20-year term loans by lowering mortgage insurance premium and allowing
somewhat expanded DTIs and seller concessions
72
BCFP’s Pro-Cyclical QM Patch Continues to Fuel the Boom
• In 1.13, “Ability-to-Repay and Qualified Mortgage Standards” rule was issued, effective
1.10.14, pursuant to the Dodd-Frank Act’s calling for minimum mortgage standards
• The Bureau noted it will “protect consumers from irresponsible mortgage lending.”
• The rule effectively set a maximum debt-to-income (DTI) limit of 43% for the private sector.
• Temporary GSE QM Patch (the QM Patch exempted the GSEs and their automated underwriting systems from
this provision for seven years.
• Similarly, FHA, the VA and the Department of Agriculture’s Rural Housing Services (RHS) , were exempted for up
to seven years or until these agencies issued their own rules codifying their own lending practices (which all
subsequently did).
• It was to make sure “prime” loans will be made responsibly, yet it sets no minimum down payment, no minimum
standard for credit worthiness, and no maximum debt-to-income ratio (for government agencies)
• Under this definition of “prime”, a borrower can have no down payment, a credit score of 580, and a debt-to-
income ratio over 50% as long as they are approved by a government-sanctioned underwriting system.
• It was foreseeable that this rule would promote an unsustainable home price boom:
• In 2013: “Booms are fueled by excessive leverage” and “this rule does little to limit borrower leverage and lays
the foundation for the next bust.”* **
• In 1951: “[In transitioning] from a buyer's to a seller's market, maximum terms become so commonly used they
tend to be considered the minimum.”***
• Flaw 1: The QM Patch does not operate counter-cyclically so as to “take the punch bowl
away” during a leverage-fueled price boom.
• Flaw 2: The QM Patch has crowded out the private market, leaving it more risky scraps.
*Pinto, “CFPB’s new ‘qualified mortgage’ rule: The devil is in the details”, http://www.aei.org/publication/cfpbs-new-qualified-mortgage-rule-the-devil-is-
in-the-details/
**Wallison and Pinto, “New Qualified Mortgage rule setting us up for another meltdown”
https://www.washingtontimes.com/news/2013/mar/3/wallison-and-pinto-new-qualified-mortgage-rule-set/
***Fisher, Financing Home Ownership, NBER, 1951
**** WSJ, No Pay Stub? No Problem. Unconventional Mortgages Make a Comeback, 1.23.19 73
Pro-Cyclical Parallels to the Last Boom
In the last 20 years we have experienced two gigantic house price booms.* It is no
coincidence that rising debt-to-income ratios have provided the fuel for both the last
and current house price boom. Going back to at least the 1930s, there is no other time
when DTIs were so high (or interest rates so low).
50% GSE Loan Share with DTI ≥ 42% and Real House Prices 160
45% 150
40% 140
35% 130
GSE Loan Share with DTI ≥ 42%

30% 120
(blue, left axis)
25% 110
20% 100
Real house prices indexed to 100 in 2000
15% (red, right axis) 90
10% 80
5% 70
0% 60
* Shiller, The Housing Boom Is Already Gigantic. How Long Can It Last?, New York Times, 12.7.18
Note: Rate for 1988-1991 is conservatively estimated at 5 percent, and is likely well below that rate.
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, Fannie Mae, BCFP, and FHFA. 74
Purchase Loans with Total DTI Greater than 43%
As we have been predicting, the share of loans with DTI > 43% is now growing rapidly to
compensate for faster home price increases compared to incomes, a trend most
pronounced for Fannie (+3.9 ppts over past 12 months) and FHA (+6.7ppts). Despite
Fannie’s announcement in March to update its Desktop Underwriting, after it had first
raised the DTI limit to 50 in August 2017, there is little evidence that it has actually
reigned in this segment. The only exceptions to the trend are RHS and Portfolio lenders.
65% 65%
60%
FHA 60%
55% 55%
50% VA 50%
45% 45%
40% Agency composite 40%
35% 35%
Fannie
30% 30%
Freddie
25% 25%
20% 20%
15% 15%
RHS
10% 10%
5% Portfolio 5%
0% 0%
Feb-13 Jul-13 Dec-13 May-14 Oct-14 Mar-15 Aug-15 Jan-16 Jun-16 Nov-16 Apr-17 Sep-17 Feb-18 Jul-18
Note: Data pertain to purchase loans for primary owner-occupied properties. Data for the portfolio line come from LLMA and McDash after removing duplicative loans. The data are weighted
by loan amount buckets and origination year using HMDA weights (lag due to time needed to allow for sales to GSEs). Weights for 2018 are assumed to be identical to 2017.
* A seller’s market, defined by the National Association of Realtors (NAR) as a home inventory supply of 6 months or less, has been present since Sept. 2012.
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, CoreLogic, and Black Knight.
75
It Is Time for the BCFP to Announce that the Temporary
GSE QM Patch Will Be Allowed to Sunset in January 2021
• As is noted on Slide 19, the percentage of agency loans with DTIs greater
than 43% has exploded since the QM rule was announced in 2013, a period
that coincides with the current home price boom
• The BCFP, in its January 2019 report, found*:
– The continued prominence of Temporary GSE QM originations is contrary to the Bureau’s
expectations at the time of the rulemaking, and certain goals of the Rule have therefore
not been met.
– In accounting for the continued prominence of Temporary GSE QM originations, two
factors can be distinguished. First, the scope of GSE-eligible loans is broad, and it grew
even broader for a period of time after the Rule became effective as the GSEs loosened
their credit eligibility in various respects.
– In contrast, the underwriting guidelines and DTI limits for General QM loans have
remained static since they were issued.
• BCFP should immediately take steps to:
– Announce that the GSE patch will not be renewed.
– Provide guidance to GSEs that they should immediately begin reducing industry’s reliance on
patch in a measured manner, thereby reducing any market impacts between now and the
2021 expiration of the patch.
– Coordinate with HUD/FHA on reductions to its DTI policies as part of a broader effort to
counter-cyclically slow down the home price boom.
– Indicate it will be looking at changes to the QM rule so that, in the future, it has a counter-
cyclical component.
* BCFP, Ability-to-Repay and Qualified Mortgage Rule Assessment Report, January 10, 2019
76
The Role of Leverage
Despite worsening affordability, leverage is allowing lower price tier borrowers to forego
a quality adjustment. The same concept applies when mortgage rates rise.
Cumulative Constant-quality and Market Expenditure

House Price Appreciation Indices (Oct-2012 = 0%)

60% 60%
50% 50%
Quality
offset
House Price Appreciation

House Price Appreciation
40% 40%
Constant-quality
30% 30%
Constant-quality
20% 20%
Market Expenditure Quality
offset
10% 10%
Market Expenditure
0% 0%
Note: HPIs are smoothed around the times of FHFA loan limit changes.
77
Ratio of Sales Price for First-time to Repeat Buyers
The trend upward is towards higher first-time buyer (FTB) prices relative to repeat
buyers (RBs). FTBs have access to the leverage punchbowl, thereby greatly reducing
the tendency to make downward quality adjustments to offset rapid home price
appreciation. RBs without access to this punchbowl, tend to make downward quality
adjustments to offset home price appreciation. This adds to demand at lower price tiers.
75%
Ratio of FTB to RB sale price 75%
Red markers show ration in November in each year.
74% 74%
73% 73%
72% 72%
71% 71%
70% 70%

78
Median Sale Price by Market Segment*, FTB Purchase Loans
Higher risk borrowers are being provided additional leverage which is fueling rapidly
increasing home prices. Market prices for subprime borrowers have increased 25 percent
since Feb-2013, while market prices for prime borrowers have only increased 11 percent.
130 130
Feb-2013 = 100 Subprime

125 125
Nov-18: $218,000
NMRI Mean price Feb-13
Prime Subprime Prime $ 270,000
120 120
Feb-13 3.0% 20.4% Subprime $ 170,000
Nov-18 3.2% 22.7%
115 115
110 110
Prime
105 105
Nov-18: $304,000
100 100
95 95
* We define prime loans as low-risk (with a stressed default rate of less than 6%), and subprime as high risk (with a stressed
default rate of 12% or greater).
Measure Market Behavior in Four Leverage Based Price Tiers
One of AEI’s innovations to track home price appreciation is to use four price bins, because the
market behaves differently in each price bin.
• “Low” bin has all sales priced less than the bottom 40% of sales prices for FHA insured homes.
• The “Low-Medium” bin has all sales priced in the next 40% of sales prices for FHA insured
homes.
Most first time buyers (FTB) in the bottom two bins, and their mortgage loans are much riskier.
By contrast, the top two bins have relatively fewer FTBs, and buyers have much less risky loans.
Note: Dare are for largest 73 CBSAs and consist of 8.5 million sale transaction study covering 5-years of home price appreciation (HPA) for 41,000 census tracts. Weighting
based on HMDA. Shares based on count. Low & med-low price tiers defined respectively as <=40th & >40th to <=80th percentile of FHA sales prices & med-high & high price
tiers defined respectively as >80th percentile of FHA sales prices & <= 125% of GSE limit & > 125% of GSE limit, all at county-level. Mortgage Risk (Leverage) Loan Grades:
High risk = >12%, Medium risk = >6%-12%, Low risk = <=6%
80
House Price Trends Impacted by Leverage
On a constant-quality basis and market price basis, prices of low and low-medium
priced homes have increased much faster than medium-high and high priced homes.
With easy access to government-supplied leverage, buyers in low and low-medium
tiers have had to make little compromise on quality.
Cumulative Constant-Quality HPI, by Price Tier Cumulative Market Expenditure HPI, by Price Tier
(2012:Q4 = 0%) Not updated (2012:Q4 = 0%)
50% 50%
Low Low
45% 45%
Low-Med Low-Med
40% 40%
Med-High Med-High
35% High 35% High
30% 30%
25% 25%
20% 20%
15% 15%
10% 10%
5% 5%
0% 0%
2012:Q4 2013:Q4 2014:Q4 2015:Q4 2016:Q4 2017:Q4 2012:Q4 2013:Q4 2014:Q4 2015:Q4 2016:Q4 2017:Q4
Source: Dare are for largest 73 CBSAs and consist of 8.5 million sale transaction study covering 5-years of home price appreciation (HPA) for 41,000 census tracts. Weighting based on
HMDA. Shares based on count. Low & med-low price tiers defined respectively as <=40th & >40th to <=80th percentile of FHA sales prices & med-high & high price tiers defined
respectively as >80th percentile of FHA sales prices & <= 125% of GSE limit & > 125% of GSE limit, all at county-level. HPIs are smoothed around the times of FHFA loan limit changes.
81
Constant-quality Prices by Guarantor Type: Low Price Tier
FHA, GSE, & Private HPI for the low priced tier all went up about the same amount over 5 years—
45%. Buyers with high mortgage risk set the price in this and low-medium market segment. VA &
RHS had lower price gains, likely due to differing appraisal practices & DTI limitations.
Not updated Cumulative Constant-quality House Price Index, by Guarantor Type:

Low Price Tier (2012:Q4 = 0%)
50% 50%
45% 45%
FHA
40% GSE 40%
Portfolio
35% 35%
VA
30% RHS 30%
25% 25%
20% 20%
15% 15%
10% 10%
5% 5%
0% 0%
2012:Q4 2013:Q2 2013:Q4 2014:Q2 2014:Q4 2015:Q2 2015:Q4 2016:Q2 2016:Q4 2017:Q2 2017:Q4 2018:Q2
Data for RHS are not available in years for which HMDA data has not yet been published.
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing. 82
High risk home purchase lending is fueling home price appreciation
In the largest 73 metros, currently 41% of agency purchase lending is high risk. FHA accounts for 57%
of this high risk lending, which is down from 74% in 2012. Significantly, the GSEs account for nearly all
of this high risk share shift. Their high risk share has increased from 10% in 2012 to 30% in 2018.
Data for RHS are not available in years for which HMDA data has not yet been published.
83
Scatterplots: Introduction
There is a strong positive correlation between higher mortgage risk (higher expected
default rates under stress) and higher home price appreciation, lower home prices, and
lower income.
The scatter charts on the two slides that follow show correlations at the census tract level
relating to mortgage risk which measures expected default rates under stress (x-axis) and:
• The ratio of tract home price appreciation (HPA) to county HPA,
• Income as a percent of metro area income.
The scatterplots are binned to better show the trend. Instead of a standard scatterplot,
which plots all the data points, the binned scatterplot only plots the binned data points.
The scatter dots for each chart are color coded based on the percentage of high risk
purchase loans as a share of all purchase loans in the tract.
• Those from the green color palette have a high risk share of less than 30%.
• Those from the blue color palette have a high risk share of greater than or equal to 30%
Source: Dare are for largest 73 CBSAs and consist of 8.5 million sale transaction study covering 5-years of home price appreciation (HPA) for 41,000 census tracts.
84
Strong Positive Correlation Between Mortgage Risk &
Home Price Appreciation
House price appreciation increases with a census tract’s mortgage risk index:
• For the dark green dots (MRI < 15%), the median ratio of tract to county house
price appreciation is 0.86
• For the dark purple dots (MRI ≥ 60%), ratio is 1.19—a 38% higher level of price
appreciation
Together the blue color palette tracts (MRI ≥ 30%) represented about 50% of all sale
transactions.
Note: Instead of a standard scatterplot, which plots all the data points, the binned scatterplot only plots the binned data points. It first groups the x-axis variable into equal-sized bins and
then computes the mean of the x and y-axis variables within each bin thereby simplifying the plot while keeping the relationship between x and y variable intact. High risk loans are
defined as loans with a Mortgage Risk Index ≥12%. 85
Strong Positive Correlation Between Mortgage Risk & Tract Income
• Dark green dots on the right, with <15% high risk loans, had low average tract MRIs (about 3-6%)
and dark purple dots on the right, with >=60% high risk loans, had high tract MRIs (about 17-23%)
• For the dark green dots, the median tract income was 158% of metro area income, while for the
dark purple dots, the median tract income was 89% of metro area income
• 75% of the census tracks with median income below 120% of metro area income had average tract
MRIs of 9% or greater
Note: Instead of a standard scatterplot, which plots all the data points, the binned scatterplot only plots the binned data points. It first groups the x-axis variable into equal-sized bins and
then computes the mean of the x and y-axis variables within each bin thereby simplifying the plot while keeping the relationship between x and y variable intact. High risk loans are
defined as loans with a Mortgage Risk Index ≥12%.
86
Measured Steps Now Would Moderate Unsustainable Home
Price Increases, Not Lead to Home Price Declines
Unlike FHA, rural housing services (RHS) has not moved out risk curve during boom
2.0, keeping housing more affordable for RHS buyers. RHS’ stressed default rate is
unchanged over the last 5+ years, while FHA’s First-Time Buyer (FTB) risk index has
increased from 21.5% to 28.9%. (The same increases apply to other FHA risk indices.)
Median downpayment Median saleprice

November 2013 November 2018 Change from November 2013 to November 2018
RHS -$2,200 -$800 6%
FHA $3,700 $3,900 29%
30% 30%
FHA FTB
28% 28%
26% 26%
24% 24%
22% 22%
20% 20%
RHS FTB
18% 18%

DTI Distributions, GSE & FHA Purchase Loans
income gains. The credit easing race between the GSEs and FHA continues. After
Fannie (and Freddie) eliminated compensating factors in July 2017, virtually all GSE
borrowers, not just those around the previous DTI limit of 45 percent, have shifted to
higher DTIs. We expect FHA volume to continue to shift to higher DTIs.
GSE FHA
6%
6%
5%
5%
Nov 2018
Nov 2018
4%
4%
3% 3%
July 2017
2% 2%
July 2017
1% 1%
0% 0%
1 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 1 20 24 28 32 36 40 44 48 52 56
DTI DTI

Share of Fannie Purchase Loans by DTI Bucket
Despite Fannie’s announcement in March to update its Desktop Underwriting after it had
first raised the DTI limit to 50 in August 2017, there is little evidence that it has actually
reigned in this segment. Compared to Feb-2018, the pullback was minor and the share of
loans with a DTI in excess of 44 is still much greater than just a year ago.
Share of Fannie Purchase Loans

9% 9%
8% Fannie Purchase Loans by DTI Bucket Jul-17 8%
Fannie Purchase Loans by DTI Bucket Feb-18

7% 7%
Fannie Purchase Loans by DTI Bucket Nov-18
6% 6%
5% 5%
4% 4%
3% 3%
2% 2%
1% 1%
0% 0%
<20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
DTI
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing/.

89
RHS Reduced Borrower DTIs from 2013 to 2018, while the
FHA Kept Increasing DTIs
DTIs limits act as counter-cyclical friction to slow the increase of house prices when
supply is tight. Remove the friction and house prices increase, fueling a boom.
Not updated
Purchase Loans by DTI Bin: February 2013 Purchase Loans by DTI Bin: February 2018
9% 9%
However FHA was pro-
RHS also allowed DTIs up By 2018, RHS was acting
FHA also cyclically fueling the
to 48%, based on counter-cyclically against the
8% allowed higher 8% house price boom. Loans
compensating factors. house price boom by lowering its
DTIs, generally with DTIs greater than
DTIs above 48% were rare. semi-hard DTI limit from 43% to
up to 57%, with 43% increased from 37%
7% 41%.
compensating 7% in 2013 to 55% in 2018.
factors. DTIs More importantly, RHS's hard Those above 50% to 57%
above 57% were stop was reduced to 46% from more than doubled to
6% In 2013 RHS appears to have 6% 27%. Additionally there
rare. 48% As a result, since 2013 the
had a semi-hard stop at 43%. percentage of loans with DTIs is no evidence of the use
greater then 43% DECLINED of compensating factors.
5% 5%
from about 20% to about 10%.
FHA
4% RHS 4% RHS
FHA
3% 3%
2% 2%
In 2013 FHA appears have
1% had a semi-hard stop at 1%
50% DTI.
0% 0%
<20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 < 20 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57

The FHA’s and the GSEs’ Rising DTIs Have Been Pro-Cyclically
Fueling the House Price Boom
Under QM, their credit boxes allow for DTIs well above 43%. As a result, DTIs have
increased dramatically. It is the use of compensating factors that reduces risk
layering, which is an important policy during a boom. However, the use of
compensating factors has been reduced markedly.
Not updated
Purchase Loans by DTI Bin: February 2013 Purchase Loans by DTI Bin: February 2018
9% 9%
FHA was also acting pro-cyclically. DTIs >43%
<20 GSE bin has FHA also increased from 37% of loans in 2013 to 55% in
8% allowed many 8%
a value of 12% In 2013 FHA 2018. Those >50% up to 57% more than doubled
appears have DTIs up to 57% to 27%. There is no strong evidence indicating
7% had a semi- with limited 7% the use of compensating factors.
In 2013 the GSEs
hard stop at use of
had a semi-hard stop
50% DTI. compensating
6% at 45% DTI. Over 2013-2018 the GSEs were
factors. 6%
pro-cyclically fueling the boom.
DTIs >43% increased from 13%
5% 5% in 2013 to 27% in 2018.
FHA GSE
4% 4%
FHA GSE
3% 3%
2% 2% The GSEs' requirement for

In 2013 the GSEs also allowed
compensating factors was
relatively few DTIs up to a
1% 1% removed in 2017. As a result,
hard stop of 50% with
DTIs >45% up to 50% increased
compensating factors.
from 3.5% of loans to 19%.
0% 0%
<20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 < 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57
20
Origination Shares Issuer Lender Type,
Similar dramatic market shifts occurred from large banks to nonbanks for both FHA
and RHS loans. Today nonbanks account of 80% of FHA and RHS originations.
FHA Purchase Origination Shares* Not updatedRHS Purchase Origination Shares*

90%
90% Nonbanks
Nonbanks 80%
80%
70% 70%
60% 60%
50% 50%
40% 40%
30% 30%
20% 20%
Large banks
Large banks
10% 10%
Other banks
0% Other banks
0%
Sep-12 May-13 Jan-14 Sep-14 May-15 Jan-16 Sep-16 May-17 Jan-18
*Origination shares do not show shares for State Housing Finance Agencies and Credit Unions which account for about 4% of the FHA Purchase market and 1%
of the RHS Purchase market.
MRIs by Issuer Lender Type,
In the case of the FHA, migration to nonbanks has boosted overall risk level, as its
wide-open credit box encourages higher risk lending by nonbanks who originate what
they can sell and sell what they originate. This has not happened with RHS, apparently
due to a risk management approach that monitors risk so as not to lean into the
current house price boom. Counter-cyclical policies are key to not promoting a boom.
Not updated
FHA Purchase Mortgage Risk Indexes RHS Purchase Mortgage Risk Indexes
29% 22%
Nonbank MRI at Composite shown by blue line Nonbank MRI
28%, up from at 19%, down
21%
23% 5 years prior from 20% 5
27% years prior
Nonbanks 20%
Nonbanks
19%
25% Composite
18%
Other banks Other banks
23%
17%
Large banks
16%
Large banks
21%
15%
19% 14%
Sep-12 May-13 Jan-14 Sep-14 May-15 Jan-16 Sep-16 May-17 Jan-18 Sep-12 May-13 Jan-14 Sep-14 May-15 Jan-16 Sep-16 May-17 Jan-18
93
Origination Shares and MRIs by Seller Lender Type,
GSE Refinance Loans
Shift away from large banks in GSE refi market has mirrored that in GSE purchase
market. Banks (both large and other) have lower risk profile than nonbanks.
Refi Origination Shares* Refi Mortgage Risk Indexes

13%
70%
Composite shown by blue line
Large banks shown by black line
60% Nonbanks
11%
50% Nonbanks
40%
9%
30%
Large banks
Other banks
20%
7%
10% Other banks
0% 5%
Sep-12 Sep-13 Sep-14 Sep-15 Sep-16 Sep-17 Sep-18 Sep-12 Sep-13 Sep-14 Sep-15 Sep-16 Sep-17 Sep-18
Note: Data for most recent months may understate large-bank share by perhaps 2 percentage points, as large banks are slower to move recent originations
to the guarantee agencies for securitization and our market shares are based on securitized loans. MRI for state housing agencies not shown because loan
volume is nil.
*Origination shares do not show shares for State Housing Finance Agencies and Credit Unions which account for about 3% of the GSE Refi market. 94
Origination Shares and MRIs by Issuer Lender Type,
FHA Refinance Loans
Massive shift from large banks to nonbanks in FHA refi market. Nonbanks now
have 94% of the market, along with a higher risk profile than large banks.
Refi Origination Shares* Refi Mortgage Risk Indexes
100% 29%
Nonbanks
90%
27%
80%
25%
70%
Other banks
60% 23%
50%
21% Composite
40%
Non Large banks
19% banks
30%
20% 17%
Large banks
10%
Other banks
15%
0% Sep-12 Jun-13 Mar- Dec-14 Sep-15 Jun-16 Mar- Dec-17 Sep-18
Sep-12 Jun-13 Mar-14 Dec-14 Sep-15 Jun-16 Mar-17 Dec-17 Sep-18 14 17
Note: Data for most recent months may understate large-bank share by perhaps 2 percentage points, as large banks are slower to move recent originations
to the guarantee agencies for securitization and our market shares are based on securitized loans. MRI for state housing agencies and credit unions not shown
because loan volume is nil.
*Origination shares do not show shares for State Housing Finance Agencies and Credit Unions which account for about 1% of the FHA Refi market. 95
State NMRI and FHA Share, Purchase Loans
The share of FHA purchase loans in a state is heavily correlated with overall lending risk.
FHA, as the riskiest lender by far, is accounting for a significant portion of risk, but is
also moving the risk curve out for other agencies.
State NMRI and State FHA Share of Agency Loans, August-October 2018
16%
LA
RI
15% WV GA
KY NM MD
TX
14% FLAR
AL OK
IN
ME CT OH
IL MO
13% SC DE NJ PA
VA NHTN KS
WY NV
SD MA AZ UT
CA
12% IA AK
WA MI NY
State NMRI
NC NE
ID
WI CO
11%
VT MN
OR
10% MT
HI ND R² = 0.679
9% DC
8%
7%
6%
5% 10% 15% 20% 25% 30% 35% 40%
FHA Share of Agency Purchase Loans

96
State NMRI Change, Purchase Loans
The states with the largest FHA and greatest risk levels have experienced faster growth
in risk. All but three states have seen their risk levels increase over the past 5 years.
Not updated Change in State NMRI:
Jan. 2013 to Jan. 2018 (in ppts)

97
Pricing Changes, Home Purchase Loans
Agency Date Effect
RHS Oct. 2014 RHS raised monthly fee from 40 bps to 50 bps
FHA Jan. 2015 FHA lowered annual MIP from 135 bps to 85 bps
RHS Oct. 2015 RHS raised upfront fee from 200 bps to 275 bps
As a result of GSE-imposed private mortgage insurer (PMI) capital
GSEs Apr. 2016 requirements, industry revised premium structure to focus more on
borrower’s credit score
RHS lowered upfront MIP from 275 bps to 100 bps and lowered
RHS Oct. 2016
monthly fee from 50 bps to 35 bps
• The GSEs find themselves in a multi-faceted competitive situation

• At one end is the FHA which neither prices nor underwrites for risk
• At the other end, the GSEs have risk-based loan level fee adjustments and private
mortgage insurers are required to hold capital in a manner that more accurately reflects
risk
– The recently implemented PMI premium changes lowered cost for borrowers with higher credit
scores (>720) and increased cost for borrowers with lower credit scores (<700)
• To meet affordable housing goals in this difficult competitive environment, the GSEs are
resorting to heavy subsidies
– However, stiff competition from FHA and due to the new PMI premium structure, the GSEs have
been forced to fill affordable housing quotas with higher credit score loans (median of 737)
98
Stressed Default Rates by Loan Type
Compared to an identical purchase loan, refis have higher stressed default rates across
all CLTV buckets. Cash-out refis are even riskier than no-cash-out refis.
At its current level, the average CO is as risky as a >90% purchase loan and the average
NCO is as risky as a mix of 81-90% and >90% purchase loans.
Reasons: weakness of appraisal process and borrower self-selection.
Note: All stress default rates computed for credit score of 720-769 and DTI of 39-43%.
99
Median Downpayments
Not updated
Median downpayment on primary home purchase loans,
April 2018
All
Guarantee agency First-time buyers Repeat buyers
buyers
Composite 5.0% / $12,100 3.0% / $5,800 10.2% / $33,300
Fannie, Freddie 13.0% / $33,700 7.0% / $18,400 20.0% / $50,500
Ginnie (FHA, VA, RHS) 1.8% / $2,900 1.8% / $2,900 1.8% / $3,200
Note: Calculated for primary home purchase loans with a government guarantee.
Source: AEI Center on Housing Markets and Finance, www.aei.org/housing.
• For agency market as a whole, median downpayment is small (5%, $10,700)

• Median is even smaller for first-time buyer loans, especially for Ginnie loans (1.8%,
$2,800). Ginnie accounts for almost 60% of agency first-time buyer volume
• Traditional 20% downpayment is the norm only for Fannie/Freddie repeat buyers.
Ginnie repeat buyers typically put down barely more than first-time buyers
• Hence, in today’s market, little saving or accumulated equity is needed to buy a

home, particularly a first home
100
Volume Growth in Counts and Dollars, Purchase Loans
As prices have been rising, dollar volume has been outgrowing count volume.
Credit easing, particularly by the FHA, is fueling this trend. This creates a vicious
cycles of price appreciation and credit easing.
Solution: dial back flow of money into housing system.
YoY Change Growth Rates
40% 40%
35% 35%
30% 30%
Not updated
25% 25%
20% Dollar Volume 20%
15% 15%
10% 10%
Count
5% 5%
Difference
0% 0%
-5% -5%
-10% -10%
Sep-13 Jan-14 May-14 Sep-14 Jan-15 May-15 Sep-15 Jan-16 May-16 Sep-16 Jan-17
101
A closer look at RHS’ October 2016 MIP cut
As expected, RHS’ purchase volume jumped immediately after its MIP cut in October
2016. Since the cut, RHS has grown faster than FHA, its most direct competitor. In
January, its growth surpassed all other agencies.
30% 30%
RHS premium cut

25% 25%
Not updated
20% 20%
15% 15%
10% 10%
RHS
5% 5%
0% 0%
-5% -5%
FHA
-10% -10%
-15% -15%
-20% -20%
Feb-16 Apr-16 Jun-16 Aug-16 Oct-16 Dec-16 Feb-17 Apr-17 Jun-17

102
No-Cash Out Refi Demand and 30-yr Mortgage Rate
In our last NMRI briefing we wrote that in response to higher rates “refi volume could
drop by 40% to 150,000 per month.” In January refi volume was down 40% from its peak
in October. Refi demand, especially no-cash outs, and the mortgage rate are strongly
correlated.
500,000 3.0
Not updated
450,000 3.2
30-yr FRM, scale inverted (right axis)
400,000 3.4
350,000 3.6
300,000 3.8
250,000 4.0
200,000 4.2
150,000 4.4
100,000 4.6
No-Cash Refinance loans (left scale)
50,000 4.8
0 5.0
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing, and Freddie Mac.
103
Low-Risk Origination Shares, Purchase Loans
Fannie’s low-risk (prime) share has recently dropped below 50% for the first time in
the history of the series. The low risk percentage gap between Fannie and Freddie is
also the widest in series history. VA’s low-risk share is well below the GSEs’.
74% 40%
72%
38%
70%
36%
68%
66% 34%
64% Freddie Mac

32%
62%
Combined 30%
60%
58% 28%
VA
56% 26%
54%
24%
52%
Fannie Mae 22%
FHA/RHS low-risk share (not shown) averages about 2%
50%
48% 20%
Sep-12 May- Jan-14 Sep-14 May- Jan-16 Sep-16 May- Jan-18 Sep-18 Sep-12 May- Jan-14 Sep-14 May- Jan-16 Sep-16 May- Jan-18 Sep-18
13 15 17 13 15 17
104
Calibrating Mortgage Safety
Not updated
Latest Latest 1935-1955 1990 vintage 2007 vintage
NMRI – purchase loans
date Value vintages (est.) (est.) (est.)
Composite index Jul 12.8% NA 6% 19%
Fannie and Freddie Jul 7.3% NA 4% 13%
FHA Jul 28.0% 3% 15% 33%
VA Jul 11.9% NA NA 15%

An index value of less than 6 is indicative of conditions conducive to a stable market.
• NMRI captures the complex interplay of changes in three types of leverage: property (LTV
and term), income (DTI, ARM vs. FRM, and term), and credit score
• 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.
• VA index less than half the level of FHA, both recently and in 2007
105
Credit Conditions: 1990 to 2013-14
1990-92 2000-03 2005-07 2013-14
% Loans with DTI ≥ 42% 5-10% 28% (2003) 43% (2007) 28%
Median borrower credit score 735 701 705 735-740
% Loans with credit scores < 640 9.5% 25% NA 5%
% Loans with CLTV > 90% 26% 35% 40% 45%
% Loans with CLTV ≥ 97% 1% 11% 40% 30%
% Loans Low/No Doc Nil (1992) 6% (2000-02) 25% Nil
30-year fixed interest rate 10% (1990) 7% (2001) 6.5% (2006) 4% (2013)
First-time buyers as % of primary
38-42% (1990) NA NA 50%
home purchase mortgages
Perfect credit (no lates) as a % of
57-60% (1990) NA NA 60%
home purchase borrowers
NMRI 6% NA 19% 10%
Compiled by AEI. Sources: CoreLogic for DTI data for 2000-03 and 2005-07 and median credit score for 2005-07. Other data from AEI, Fannie
Mae, FHA, Equifax, Freddie Mac, FICO®, and miscellaneous other sources. In general, figures shown are for the entire purchase loan market
(conventional and government guaranteed). John Burns (John Burns Real Estate Consulting) collaborated on the presentation format.
• Clear buildup of risk from 1990-92 to 2005-07
• 2013-14 loan cohort less risky than 2005-07 cohort due to smaller percentage of loans with high DTIs,
higher median credit score, and very few low/no doc loans
• Still, 2013-14 cohort is substantially riskier than 1990-92 cohort. Main differences are sharp increase in
loans with high CLTVs and high debt ratios, notwithstanding much lower interest rates 106
Role of Income Leverage During Housing Boom
• Less attention paid to income leverage than to property and credit leverage
– Owes to data scarcity, as the FHA and GSEs published virtually no DTI data until 2013, when
FHFA released DTI trends for the period 1996 onward for both the GSEs and FHA
• Over 1996-2005, higher income leverage raised overall home purchase

buying power 46%, three-quarters of the 62% rise in real home prices*
– GSE median housing DTIs (purchase transactions): 23% in 1996, 27% in 2005 ― 17% boost
in buying power (based on 8% interest rate for both years)
– Median loan rates fell from 8% in 1996 to 6% in 2005 ― 20% added boost in buying power
– Low doc/no doc loan share was near 0% in 1996 with minimal income overstatement. By
2005, share was 15% with 25% income overstatement (source: CoreLogic). This increased
housing DTI in 2005 another percentage point to 28% (4% added boost in buyer power).
• Push/pull of increasing leverage at work today

– Home Prices Start to Heat Up: Double-digit growth arrives in more cities, but affordability
worries emerge amid thin supply (WSJ, May 12, 2015)
• NMRI tracks changes in income leverage

– Since Nov. 2012, median total DTI for all agency primary purchase loans increased from 36%
to 38%, leaving buyers more highly leveraged even as income volatility increases: Cash
Crunch Is, for Many, a Monthly Problem (WSJ, May 20, 2015)
* Does not take into account increases in income, home size or quality. Ex. the median new home size increased 14% from 1996 to 2005.
107
Fed Tightening and Efforts to Maintain Buying Power
• Historical precedent: end of the Fed’s interest rate peg in effect from World War II
– Long-term mortgage rate rose from 4.1% in 1953 to 6% in 1962
– 5 amendments to National Housing Act (1954-61) increased FHA’s LTV and loan term limits
– These changes, along with rising housing DTIs, kept buying power constant from 1953 to 1962
• Today: with the Fed now starting to tighten, long-term interest rates will rise
– All else equal, a rise in the 30-year mortgage rate from 4% to 6% would reduce buying power by
same amount as a 19% jump in home prices
• Two steps would keep buying power largely constant with no change in income-
to-house price ratio
– Reduce FHA’s annual premium an additional 35 basis points to 0.50% (requires action by FHA and would further
stress FHA’s capital level)
– Boost median total DTI from 41% today to 45% for FHA and from 34% today for GSEs to 38%. These changes
would be QM compliant due to the agency QM exemption.
• Very risky steps. Would result in median total DTI for FHA well above the peak
level in 2005-06 and for GSEs equal to the 2005-06 peak level.
• Wealth Building Home Loan provides a sustainable alternative
*FHFA’s 2014 fee report indicates that the GSEs were undercharging on high risk loans and overcharging on low risk ones and that
overall guarantee fees were lower than needed to meet capital return levels. This is equivalent to a hidden guarantee fee cut and 108
could be
repeated in the future.
Appraisals Should Be the Guard Rail Against Speculative Booms
An appraisal should provide an opinion as to the relationship between
market selling price and intrinsic or fundamental value
• Property valuations and appraisals should review and provide:
– A robust and transparent opinion of a property’s most likely market price based on a
systematic analysis of generally available information rather than 3 subjectively chosen
comparison properties
• Including a range around the most likely market price at a specified confidence level
– Trends in and nearness to key elements of utility such as employment, shopping,

transportation, other infrastructure and amenities, along with zoning, density restrictions,
and tax burden that impact intrinsic value and market price
– Market conditions and an assessment of whether a substantial differential between a
property’s intrinsic value and market price is substantiated by a change in utility:
• At least 10-year nominal and real home price trends and a determination as to current position in
market cycle relative to equilibrium
• At least a 5-year history of buyer’s market (inventory > 6 mo.) and/or seller’s market ( ≤ 6 mo.)
– Impact on buying power over last 5 years due to changes in loan leverage or prevailing
interest rates
– Current land value and land share, and trends in both
– Whether real price changes are due to leverage growth, improving utility or a combination
– A property’s overall condition and a recommendation as to any readily observable repairs
necessary to make it meet generally accepted minimum property requirements
109
Cross-subsidies Return to the GSEs
• FHFA’s Report on Single-Family Guarantee Fees in 2014 disclosed numerous

instances of mispricing and cross-subsidies, a significant deviation from 2013
report1,2
– High risk 30-year loans subsidized by low risk 15-years
– Borrowers with low credit scores scores subsidized by those with high credit scores
– High LTV loans subsidized by low LTV loans
– Overall guarantee fee levels were found insufficient to meet the estimated future cost of providing
the guarantee
• Mispricing promotes adverse selection and increases overall risk of mortgage

finance system
– Allows the GSEs to implement guarantee fee cuts in an opaque manner
– Ability to subsidize risky loans will cause progressive loosening of underwriting standards
– As was the case the last time around, this movement out the risk curve may take 5-10 years
• NMRI is designed to track these risks in real time
1 http://www.fhfa.gov/AboutUs/Reports/Pages/Fannie-Mae-and-Freddie-Mac-Single-Family-Guarantee-Fees-in-2014.aspx
2For a detailed analysis of the role robust risk-based pricing plays in promoting a fair and efficient mortgage market, see Board
of Governors of the
Federal Reserve System, Report to the Congress on Credit Scoring and Its Effects on the Availability and Affordability of Credit, August 2007,
www.federalreserve.gov/boarddocs/rptcongress/creditscore/creditscore.pdf
110
Change in Agency Purchase Loan Volume
Since April 2016 the GSEs have again overtaken FHA as the fastest growing agencies
indicating that FHA MIP cut effect from January 2015, which led to massive poaching and
some new homebuyers, has worn off.
Percent change from 12 months earlier
60% 60%
50% 50%
FHA
40% 40%
30% 30%
20% Composite 20%
10% 10%
0% 0%
Fannie/Freddie
-10% -10%
-20% -20%
-30% -30%
Sep-13 Feb-14 Jul-14 Dec-14 May-15 Oct-15 Mar-16 Aug-16 Jan-17 Jun-17 Nov-17 Apr-18 Sep-18

111
GSEs: Ratio of NMRI for Loans with Total
CLTV > 95% to Loans with Lower Total CLTVs
FHFA Director Mel Watt stated that with use of compensating factors “loans with a
3 percent down payment backed by GSEs are no riskier than those with a down
payment of 10 percent …” (Jan. 27, 2015). Based on NMRIs, this is not true.
220% 220%
NMRI ratio: >95% CLTV loans to 86-90% CLTV loans

200% 200%
180% 180%
160% Full use of compensating factors would imply ratios of 100% 160%
NMRI ratio: >95% CLTV loans to 91-95% CLTV loans

140% 140%
120% 120%
Fannie accounts for the vast majority

100% of GSE loans with CLTVs > 95% 100%
80% 80%

112
FICO® Score Distribution
FHA’s minimum scores are near the bottom of the FICO credit score distribution. An
FHA borrower with a 500 credit score has an NMRI of 50%, twice as risky as today’s
median FHA loan and eight times riskier than today’s median GSE loan.
100% 100%
90% 90%
80% 80%
70% 800 ≈ 80th percentile 70%

750 ≈ 62nd percentile
700 ≈ 46th percentile
60% 60%
580 ≈ 19th percentile (min for FHA, 3.5% down)
500 ≈ 5th percentile (min for FHA, 10% down)
50% 50%
Share of total population with score
below level shown on horizontal axis
40% 40%
30% 30%
20% 20%
10% 10%
0% 0%
500 520 540 560 580 600 620 640 660 680 700 720 740 760 780 800 820 840
FICO score
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing, from FICO 8 score distribution for October 2014. Distribution
for FICO scores of 300-800 and 850 directly from FICO; distribution between 800 and 850 interpolated by Edward Pinto. 113
Median Credit Score on Primary Purchase Loans*
Median scores about unchanged from January 2017. FHA’s all-buyer median at 34th
percentile of scored distribution, with room to drop given FHA’s minimum scores. In
current seller’s market, this will boost home prices faster than income.
780 780
760 760
Repeat buyers
740 740
All buyers
720 720
First-time buyers
700 700
680 680
FHA only, all buyers

660 660
*Data pertain to purchase loans for primary owner-occupied properties. Percentiles based on population of all scorable individuals.
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing 114
Aggregate Default Risk Surge for Home Purchase Loans
Is Over Five Years Old
Aggregate default risk (which measures the combined effect of loan-level risk and
volume) continues to rise. FHA continues to account for more than half of the
aggregate agency risk.
Number of expected defaults under stress
45,000 45,000
Composite
Red markers show October count in each year.
40,000 40,000
35,000 35,000
30,000 30,000
25,000 25,000
FHA
20,000 20,000
15,000 15,000
10,000 Other agencies 10,000
5,000 5,000
115
The Effect of April 2016 PMI Price Change
Pricing for risk matters. GSE pricing for higher credit scores are now competitive
with FHA, which is reflected in changes to market shares. It has also led to GSE
gaining a greater share of lower risk FHA borrowers.
GSEs’ Market Share in CLTV > 95%, by Credit Score Bins
New PI+PMI Monthly Payment Comparison
45%
Not updated Change in PMI Credit PMI Pricing PMI Pricing Risk-based required
40% pricing takes effect
Score Bin Change Compared to FHA asset amount factors
>= 760 ($101) ($5) 4.83%
35% Fannie and Freddie guarantee
mortgages with as little as 3% down 740-759 ($60) $35 7.60%
30% 720-739 ($32) $82 9.84%
700-719 ($30) $136 11.55%
25%
680-699 $18 $187 14.25%
20% 660-679 $119 $308 19.20%
640-659 $149 $353 19.20%
15% 620-639 $155 $414 19.20%
* assumes a 3.5% downpayment on a $250,000 home. Conforming mortgage
10% rate of 3.89% and a FHA mortgage rate of 3.50%.
5%
Average NMRI in CLTV > 95% Market
0% Apr-Oct 2015 Apr-Oct 2016 Change

Jan-15 Mar-15 May-15 Jul-15 Sep-15 Nov-15 Jan-16 Mar-16 May-16 Jul-16 Sep-16 Nov-16 Jan-17 GSEs 14.3% 13.9% -0.4 ppt
>=760 740-759 720-739 700-719 Total
FHA 24.4% 25.3% +0.8 ppt
680-699 660-679 640-659 620-639 <620
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing, and Fannie Mae.
116
Credit Score Distribution & MRIs, Purchase Loans*
Stark contrast between credit score distributions for FHA and GSE borrowers. FHA
accounts for over 80% of scores below 660, while GSEs account for nearly 90%
above 740.
FHA share GSE share Share of Total, by Credit Score Bucket, October 2018
100%
80%
60%
40%
20%
0%
580-589 600-609 620-629 640-649 660-669 680-689 700-709 720-729 740-749 760-769 780-789 800-809 820-829
MRIs rise as credit scores decline – evidence of risk layering rather than
compensation for risk. In a seller’s market, risk layering artificially pushes up
prices, resulting in a wealth transfer from buyers to sellers of entry-level homes.
50%
40% MRI by FICO score, October 2018

30%
FHA
20%
GSEs
10%
0%
580-589 600-609 620-629 640-649 660-669 680-689 700-709 720-729 740-749 760-769 780-789 800-809 820-829
*Data pertain to purchase loans for primary owner-occupied properties.

Purchase Loans with Down Payment of 5% or Less*
58% of all primary purchase loans and 36% of such Fannie/Freddie loans have a minimal
down payment. With QM silent on down payments, lots of room for these shares to rise.
In current seller’s market, this will drive up home prices more than income.
65% 65%
60% 60%
55% Composite 55%

The October 2018 share for first-time buyers was 72%.
50% 50%
45% 45%
40% In October 2018, 90% of FHA primary owner-occupied purchase 40%

loans had a down payment of 5% or less; the share for VA was 88%.
35% 35%
Fannie/Freddie
30% 30%
25% 25%
20% 20%

Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing
118
Composite Origination Shares and MRIs by Channel,
Purchase Loans
Retail and correspondent shares have stabilized at around 45% each. Broker share has
remained around 10%. Correspondent and broker composite MRIs tracking higher at
levels significantly above retail MRI.
Composite Origination Shares Composite Mortgage Risk Indexes

60% 16%
Broker
Retail
50%
14%
Correspondent
40%
Correspondent
12%
30%
10% Retail
20%
Broker 8%
10%
0% 6%
Feb-13 Dec-13 Oct-14 Aug-15 Jun-16 Apr-17 Feb-18 Feb-13 Dec-13 Oct-14 Aug-15 Jun-16 Apr-17 Feb-18

119
Large Bank Origination Shares and MRIs by Channel,
Purchase Loans
Nearly all large-bank volume comes through retail and correspondent channels; broker
volume has dropped to de minimis level. MRI shows that large banks are acting to limit
defaults among retail customers and reducing risk tolerance on correspondent loans.*
Large Bank Origination Shares Large Bank Mortgage Risk Indexes

80% 16%
Correspondent
70%
14%
60%
Correspondent
12%
50%
40% 10%
Retail Broker
30%
8%
20%
Retail
6%
10%
Broker
0% 4%
*Sharp drop in MRI for broker channel is due to greatly reduced volume of GNMA loans.
120
Fed’s Senior Loan Officer Survey is Badly Flawed
• Showed no systematic loosening in mortgage lending standards in the run-up to the
2007-08 financial crisis
• Survey design problems
– Only covers commercial bank lenders
– Based on opinions of a small number of loan officers
– Weights all responses equally
• Results over past year: some easing for GSE loans, little change for Ginnie loans.
Survey misses the caution prevailing at banks revealed by NMRI (see table).
• Mortgage lending standards have eased but this is due to mix shifts not captured by the
survey (from banks to nonbanks and from GSEs to FHA).
• Bottom line: don’t use the Fed survey
Change in NMRI, Change in standards,

Oct. 2016 to Oct. 2017 2016:Q4 to 2017:Q4 Not updated
Senior Loan Officer
Agency All banks
Survey
GSEs Some easing Some easing
FHA, VA, and RHS Some easing Little change
Note: “Easing” denotes a rise in the NMRI of 0.25 percentage point or more, “Tightening” denotes a
decline in the NMRI of 0.25 percentage point or more, and “Little change” denotes a change in the
NMRI of less than 0.25 percentage point in either direction.
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing, and Federal Reserve 121
Board, http://www.federalreserve.gov/boarddocs/snloansurvey/201608/fullreport.pdf)
Urban Myth: Tight Credit Keeping
“Creditworthy” Borrowers Out of Market
• Assertion: “Today’s lenders are simply not originating loans for borrowers with less
than perfect credit.” (Urban Institute, April 2015)1
• Fact: 40% of home purchase borrowers in 2013-14 had less than perfect credit
(perfect being no lates)
• Fact: Median credit score for FHA purchase loans was 674 in April 2015, well
below the median for all individuals in U.S. with a score
• Assertion: “Severe” 2013 standards caused 1.25 million purchase loans to be
missing relative to “cautious” 2001 standards1
• Fact: 70% of these “missing” borrowers had a credit score < 660; would have
an MRI above 25% due to extensive risk layering on FHA loans2
• Fact: Urban study is fatally flawed. Credit score distribution was the same in
2005 as in 2001, so the number of “missing” loans would be the same using
either year as the baseline. Because credit standards in 2005 were extremely
lax, this makes the notion of “missing loans” meaningless
• Fact: Credit standards in 2001 were much looser than in the early 1990s. Thus,
the early ’90s would be a more appropriate baseline for cautious standards.3
1Impact of Tight Credit Standards on 2009-2013 Lending, http://www.urban.org/publications/2000165.html.
2In addition to subprime credit score, initial equity of 3% or less, 30 year loan term, average total debt ratio of 41% without use of residual income.
3A 1999 Urban Institute study (http://www.urban.org/publications/1000205.html) documented the easing of standards by the GSEs through 1998 but also noted that
“The GSEs’ guidelines, designed to identify creditworthy applicants, are more likely to disqualify borrowers with low incomes, limited wealth, and poor credit histories;
applicants with these characteristics are disproportionately minorities.” HUD relied on this study when it greatly expanded the affordable housing goals in 2000.
122
FHA Perpetuates This Myth
• FHA promotes lending to very high-risk borrowers: credit score floors of
500 with 10% down and 580 with 3.5% down
• 2007 vintage of FHA loans indicative of performance under stress
FHA’s 2007 Loan Cohort: 90-day Delinquency Rate by Credit Score
≤ 620 620-650 650-700 700-750 >750
47% 35% 25% 14% 9%
Source: Urban Institute, VA Loans Outperform FHA Loans. Why? And What Can We Learn?, table 3, panel B
http://www.urban.org/research/publication/va-loans-outperform-fha-loans-why-and-what-can-we-learn
• FHA charges the same mortgage insurance premium regardless of borrower

credit risk. Lack of risk-based pricing:1
– Misleads high-risk borrowers into thinking they are creditworthy
– Exposes FHA to adverse selection
– Is inherently unfair
– Increases overall risk of mortgage finance system.
• AEI’s Wealth Building Home Loan offers a better solution for higher-risk
borrowers
1Fora detailed analysis of the value of credit scoring and risk-based pricing for promoting a fair and efficient mortgage market, see Board of Governors of the
Federal Reserve System, Report to the Congress on Credit Scoring and Its Effects on the Availability and Affordability of Credit, August 2007,
123
FHA Is All about Moral Hazard
Moral hazard: “A situation where one party gets involved in a risky event knowing that it is
protected against the risk and the other party will incur the cost.”1
• FHA insurance presents a classic case with multiple layers of moral hazard:
– FHA insures 100% of the loss for high-risk loans, has minimal capital, and is taxpayer backed
o It neither prices for risk nor underwrites for risk layering, which is inherently unfair to borrowers and exposes FHA to
adverse selection.2
o Exact opposite of the original FHA structure in the 1934 National Housing Act
– Ginnie Mae and nonbank lenders, both with minimal capital, are able to ignore borrower solvency
risk since they are protected by FHA
– High-risk borrowers, misled into thinking they are creditworthy, borrow more than they should.
Greater borrowing spurred by recent cut in mortgage insurance premium is a textbook example.
– Increases overall risk of mortgage finance system
o Effectively unconstrained by QM,3 increasing competition between Fannie and FHA, and eventually Freddie, will cause
progressive loosening of underwriting standards
o As was the case during the last boom/bust cycle, this movement out the risk curve may take 5-10 years
• NMRI is designed to track these risks in real time

1http://economictimes.indiatimes.com/definition/moral-hazard
2For a detailed analysis of the role risk-based pricing plays in promoting a fair and efficient mortgage market, see Board of Governors of the Federal
Reserve System, Report to the Congress on Credit Scoring and Its Effects on the Availability and Affordability of Credit, August 2007,
3QM as implemented does not constrain leverage (LTV/CLTV, credit score, or total DTI). It does constrain loan term, but at a highly levered 30 years.
The rest of QM is largely window dressing (except for the current 5 year fully indexed requirement on ARMs). For example, FHA has had full doc and
fully amortizing loans since its inception. This did not prevent 1 of 8 (3.4 million) of its borrowers going to claim from cohort years 1975-2013.
124
While FHA’s Capital Reached Required 2% Statutory
Level for 1st Time since 2008, It Is Insufficient
Mutual Mortgage Insurance Fund at 2.07% in FY 2015 compared to 0.41% in FY 2014.
A further reduction in insurance fee is unjustified and counter productive.
• Impact of premium cut was minimal as most of the gain since FY 2014
projection was due to FY 2015 volume gain which was offset dollar for dollar
by reductions in mortgage insurance premiums.
• Volume gain was largely due to poaching, mostly from Fannie and RHS, and an
improving economy.
• Most of the increase in buying power was capitalized into the purchase of higher priced
homes.
• Higher home price projection vs. FY 2014 projection also added to economic value.
– Home prices are assumed to continue to increase faster than incomes for foreseeable future.
• 36% of FY 2015 volume in CA, FL and AZ (traditionally volatile states) along with TX (has
high house-price risk), up from 28% in FY 2010.
• Premium cut substantially reduced FY2021 projected economic value.
• 2% capital level is insufficient.
– FY 2014 report indicated a 4% capital level more appropriate given that U.S. is already
in the 7th year of an economic expansion.
– FHA not projected to hit a 4% single-family forward loan capital level until the end of FY
2020, at which point the current expansion, were it to continue, would be the longest on
record.
• FHA’s MRI continues to hover near 25% and is 37% for loans with credit
scores < 660.
• Extraordinarily high default rate on loans with scores below 660 is an abusive lending practice.
• These borrowers are disproportionately low-income and minority. 125
Share of States with Increase in SMRI for Purchase
Loans from Year-Earlier Period*
Credit easing trend has stopped in majority of states – SMRI down in about two-
thirds of states for agency composite. Contrast between composite and individual
agencies has re-appeared as market shares have shifted back to the GSEs after
effects of FHA premium cut have worn off.
100% FHA 100%
90% 90%
Fannie/
80% 80%
Freddie
70% 70%
Percent changes calculated from Composite
year-earlier three-month average.
60% 60%
50% VA 50%
40% 40%
Not updated
30% 30%
20% 20%
10% 10%
0% 0%
-10% -10%
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49
*Final value for each series based on change in each state from Aug.-Oct. 2015 average to Aug.-Oct. 2016 average. Earlier values calculated analogously.
Note: SMRI applies exactly the same stress-test methodology from the NMRI to loans at the state level.
CBSA NHMI: Investor Type for Home Purchase Loans
FHA has greater presence in lower cost CBSAs, while the Conventional side of the
market has greater presence in higher cost CBSAs. FHA, due to its highest risk rating, is
driving up risk in these lower cost CBSAs.
Not updated Investor Share by Top 25 CBSA (count),
2017:Q3 - 2018:Q2
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Most volatile
FHA RHS VA Conventional CBSAs
Riverside/San Bernardino: A Case in Point
A metro with very volatile house prices, especially for bottom price tier. Since the
Jan. 2012 trough, bottom-tier prices up almost 70%, boosted by liberal credit
terms and low rates in a seller’s market.
Not updated Index = 100 in April 1996
450 450
400 400
350 350
Series show nominal house prices
300 300
Top tier
250 250
200 200
Bottom tier
150 150
100 100
50 50
Apr-96 Apr-98 Apr-00 Apr-02 Apr-04 Apr-06 Apr-08 Apr-10 Apr-12 Apr-14
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing, based on data from Zillow.
128
House Price Volatility, 51 Largest Metro Areas
House prices most volatile in California and Florida metros, moderately volatile in 16
other metros, with 25 metros having low volatility.
120% 120%
Not updated Florida metros
100% 100%
California metros
80% 80%
60% 60%
Other volatile
metros
40% 40%
More stable metros

20% 20%
0% 0%
1984 1987 1990 1993 1996 1999 2002 2005 2008 2011
-20% -20%
-40% -40%
-60% -60%
Note: Each series shows the percent change from 20 quarters (5 years) earlier. Volatile metros are defined as those for which the difference
between the highest and lowest annual percent changes is more than 30 percentage points. All other metros are in the “more stable” group.
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing, using data from Zillow.
129
Median Values of Risk Factors by Loan Type
Median value, October 2018
Risk Factor Purchase No-Cash-Out Refi Cash-Out Refi
Credit Score 731 733 710
Total DTI 40 38 41
CLTV 95 73 75
Note: Calculations based on loans with non-missing data for credit score, DTI, and CLTV.
• Greater riskiness of refi loans for a given credit score, total DTI, and
CLTV is offset by tighter lending standards. Refis have:
– Higher credit scores
– Lower total DTIs
– Much lower CLTVs
130
Risk Shares for Home Purchase Loans
Loan risk greater than level conducive to long-run market stability, as low-risk
loans accounted for only 37% of volume in October, far from comprising the
preponderance of loans, which is necessary for long-term market stability.
50% 50%
For first-time buyers, the October 2018 low-risk prime share was 21%.
45% 45%
High risk
40% 40%
Low risk
35% 35%
30% 30%
Low risk prime defined as stressed default rate of less than 6%,
medium risk near prime is 6% to 12%, and high risk subprime is 12% or higher.
25% 25%
Medium risk
20% 20%
Note. Risk shares pertain to the composite of all purchase loans.

131
DTI Distributions and MRIs, Primary Purchase Loans*
FHA has DTIs as high as 57% and GSEs have some as high as 50%. DTI limits should
operate to “take the punch bowl away” before a leverage fueled boom goes too far.
But the current DTIs maximums are so high as to present no such constraint.
8%
DTI Distribution, October 2018

6%
GSEs FHA
4%
2%
0%
<20 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 >57
For FHA, MRIs rise with DTIs – evidence of risk layering. The same is true for the GSEs
up to a DTI of 45%; they compensate for risk on only the very highest DTI loans.
40%
FHA
MRI by level of DTI, October 2018
30%
20%
10%
GSEs
0%
<20 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 >57
Cash-Out Share and Home Equity
Cash-outs accounted for 71 percent of total refis in October, more than triple the
share at start of the series, owing in part to greater home equity. Temporary spike
down early last year was due to a surge in no-cash-outs from FHA premium cut and a
drop in mortgage rates. Recent spike is due to a large decline in no-cash-outs from
higher mortgage rates while the demand for cash-outs has remained relatively stable.
$20.0 75%
$17.5
65%
$15.0
Available equity by households and nonprofits
in trillions (left axis) 55%
$12.5
$10.0 45%
$7.5
35%
Cash-out share of all agency refis
(right axis)
$5.0
25%
$2.5
$0.0 15%
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing, and Financial Accounts of the United States. 133
Agency Cash-Out Share and Defaults
As cash-out share has grown, its agency composition has also changed. Compared
to the series’ start, VA and FHA have tripled their share by loosening lending
standards faster than the GSEs. Today, they account more over half of the expected
defaults, up from just 20%.
Market Share Expected Defaults under Stress
100% 9,000
Not updated
90%
8,000
80%
GSE 7,000
70% GSE
6,000
MRI Oct. 2012 Oct. 2017
60% Composite 6.3% 13.3%
Fannie Mae 5.2% 9.9% 5,000
50% Freddie Mac 5.5% 9.8% Ginnie
FHA 19.3% 26.0% 4,000
40% VA 16.1% 21.4%
3,000
30%
20%
Ginnie 2,000
10% 1,000
0% 0
Sep-12 Apr-13 Nov-13 Jun-14 Jan-15 Aug-15 Mar-16 Oct-16 May-17 Sep-12 Apr-13 Nov-13 Jun-14 Jan-15 Aug-15 Mar-16 Oct-16 May-17

Nonbank Origination Shares and MRIs by Channel,
Purchase Loans
Nonbank’s correspondent share has been increasing at the expense of retail and
broker. While the MRIs of all three channels are increasing, the correspondent channel
has the highest MRI and has increased the most.
Nonbank Origination Shares
Nonbank Mortgage Risk Indexes
20%
65%
Correspondent
18%
55%
16%
Retail
45% Broker
14%
Correspondent 12% Retail

35%
10%
25%
8%
Broker
15%
6%
5% 4%

135
Greater House Price Volatility at the Lower End
In the past, increasing leverage has fueled unsustainable house price trends. Since the
advent of expanded “affordable housing” efforts, these trends have become stronger at
the lower end of the market, as indicated by the higher peaks and lower troughs. Since
2012, a similar boom pattern has emerged.
136
Housing Volatility Index
Today’s 21 quarters look to constitute the early part of an extended housing boom.
Sustained periods with few price declines allow market excesses to build and may lead
to a Minsky Moment.** Unsustainable increases in entry-level home prices result in
speculation in land, the more volatile part of the structure/land package.
Distribution of Negative House Price Change from Four Quarters Earlier in 81 US MSAs*
100%
90%
Quiescent period
80% Correction
70%
60%
21 Qtrs.
21 Qtrs. 10 Qtrs. 15 Qtrs. 36 Qtrs. 39 Qtrs. 25 Qtrs. (through
50%
2018:Q3)
40%
30%
20%
10%
0%
1976:Q3
1981:Q3
1986:Q3
1991:Q3
1996:Q3
1977:Q3
1978:Q3
1979:Q3
1980:Q3
1982:Q3
1983:Q3
1984:Q3
1985:Q3
1987:Q3
1988:Q3
1989:Q3
1990:Q3
1992:Q3
1993:Q3
1994:Q3
1995:Q3
1997:Q3
1998:Q3
1999:Q3
2000:Q3
2001:Q3
2002:Q3
2003:Q3
2004:Q3
2005:Q3
2006:Q3
2007:Q3
2008:Q3
2009:Q3
2010:Q3
2011:Q3
2012:Q3
2013:Q3
2014:Q3
2015:Q3
2016:Q3
2017:Q3
2018:Q3
*Only 30 metros included at beginning of series. This number grows until 1977Q4, when 81 metros are consistently reported.
**A Minsky moment is a sudden major collapse of asset values which is part of the credit cycle or business cycle. Such moments occur because long periods
of prosperity and increasing value of investments lead to increasing speculation using borrowed money. Wikipedia 137
Source: FHFA Quarterly House Price Index and AEI Center on Housing Markets and Finance
Unforgiving Home Price Cycles: Booms Fueled by Increasing
Leverage in a Seller’s Market, Followed by Mean Reversion
Fueled by growing loan leverage and tight supplies, real home prices have increased 29% since
the early 2012 trough. Contrary to prevailing view, post-crisis underwriting/regulatory changes
promote rather than constrain a boom. The pattern is similar to the initial years of the price
boom that began in 1998. If it continues, the risk of a serious house price correction increases.
Real House Price Index (1975:Q1 = 100)*, through 2018:Q3

Predominently a buyer's market Entirely a buyer's market
Entirely a seller's market Predominently a seller's market
220 220
Real average annual growth rate
200 1997:Q2-2003:Q2 -- 4.3% 200
1997:Q2-2006:Q2 -- 5.1%
180 2012:Q2-2018:Q2 -- 4.2% 180
GSE affordable housing goals take effect for

160 160
CY 1993 as mandated by the Housing
140 140
120 2012 to date: easing loan 120

standards, very loose Fed
1993-2006: period of credit easing
100 policy, and historically 100
and generally falling mortgage rates
low mortgage rates
80 80
1975:Q1 1979:Q2 1983:Q3 1987:Q4 1992:Q1 1996:Q2 2000:Q3 2004:Q4 2009:Q1 2013:Q2 2017:Q3
* Calculated as FHFA's all-transaction house price index divided by BEA's price index for personal consumption expenditures.
Note: National Association of Realtors (NAR) defines a seller's market as inventory that is less than or equal to 6 months of sales. NAR data pertain to existing homes; not available
before June 1982. Data from the Census Bureau for new home inventories used before June 1982.
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, FHFA, BEA, Census Bureau, and NAR.
138
Supply-Demand Imbalance Is Greatest in the Low Price Tier
There is also a greater bifurcation on months supply in the market by price point.
From a year ago, the supply-imbalance has improved most at the upper end of the
market, which is approaching a buyer’s market nationally. Inventories remain very
tight at the lower end, continuing the strong seller’s market, which implies that house
prices will continue to increase, thereby worsening affordability.
12 Months Supply by Price Tier: 73 Metros 12
Grey bars show Q3

10 for each year. 10
8 8
6 6
4 4
Low
2 Low-Med 2
Med-High
High
0 0
2013:Q1 2013:Q3 2014:Q1 2014:Q3 2015:Q1 2015:Q3 2016:Q1 2016:Q3 2017:Q1 2017:Q3 2018:Q1 2018:Q3
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, and Zillow.
139
Comparing the Supply-Demand Imbalance: 100 Largest Metros
While the supply-demand imbalance has generally improved slightly, it remains tight in
most metros, especially at lower price tiers. For high end homes, 40 of the 100 metros have
buyer’s market conditions, up from 31 a year ago (high tier buyer’s market ≥ 8 months).
Low Price Tier Low-Medium Price Tier
10 10

7 7
6 6
5 5 seller's market
4
seller's market
4
3 3
2 2
1
0 Home Sales, by Metro Market Size
one year ago
0
one year ago
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Medium-High Price Tier High Price Tier (note different axes maxima)
10 20

8 one year ago one year ago
16
7 14
6 seller's market 12
5 10
4
seller's market
8
3 6
2 4
one year ago 2
0 one year ago
0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Note: Data are for largest 100 metros using Zillow’s existing home sales. Urban Honolulu in the high price tier is outside of range shown.
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, and Zillow. 140
Tale of two markets: Low end entry level and high end repeat buyers
For high price tier properties, the median months’ supply for all 271 metro areas are far
higher than for the 50 largest metros. This is because the additional 221 metros have
substantially higher month’s inventory of about 16.2 months compared to 9.2 months
just for the 50 metros. On the other hand, when looked at overall, there is little
difference, since the 50 metros account for about 2/3 of the of the overall market
accounted for in the 271 metros.
High Price Tier High Price Tier
Median of 271 metros: 15.1 months Median of 50 metros: 9.2 months
Months’ Supply (2018:Q4) : 271 CBSAs Months’ Supply (2018:Q4) : Top 50 CBSAs
0.7 6 24+ 0.7 6 24 141
*The overall months’ supply takes metros on each map as one market, and is calculated by dividing the total number of listings by total sales.
Tale of two markets: Low end entry level and high end repeat buyers
For low price tier properties, the months’ supply of 221 smaller markets are only
somewhat higher at about 2.5 months, than for the top 50 metros (1.6 months),
resulting in a relatively small difference in the median.
Low Price Tier Low Price Tier

Median of 271 CBSAs : 2.3 months Median of 50 CBSAs: 1.6 months
Months’ Supply (2018:Q4) : 271 CBSAs Months’ Supply (2018:Q4) : Top 50 CBSAs
0.7 6 24+ 0.7 6 24
142
*The overall months’ supply takes metros on each map as one market, and is calculated by dividing the total number of listings by total sales.
Comparison: Home Sale Transactions (New and Existing)
Over the past 2 years, the combined total of the NAR’s EHS and the Census Bureau’s
NRS has become more accurate. In Q3, a wider gap has opened up again as the NAR and
Census Bureau have reported larger declines in lending volume. Given that the NAR’s
EHS and the Census Bureau’s NRS are based on surveys with large gross ups, which
tend to amplify the errors, the downturn in their data may be overstated.
2,000,000 Period Absolute Error* % Error* 2,000,000
2013:Q4 - 2014:Q3 172,084 3.2%
2014:Q4 - 2015:Q3 135,656 2.4% NAR & Census
2015:Q4 - 2016:Q3 12,814 0.2%
1,800,000 NHMI 1,800,000
2016:Q4 - 2017:Q3 (35,592) -0.6%
2017:Q4 - 2018:Q3 (239,075) -3.9%
1,600,000 1,600,000
1,400,000 1,400,000
1,200,000 1,200,000
NAR
1,000,000 Grey bars show Q3 1,000,000
for each year.
800,000 800,000
2012:Q4 2013:Q2 2013:Q4 2014:Q2 2014:Q4 2015:Q2 2015:Q4 2016:Q2 2016:Q4 2017:Q2 2017:Q4 2018:Q2
* Error refers to count and percent difference between the NAR’s Existing Homes Sales (EHS) plus the Census Bureau’s New Residential Sales (NRS) and NHMI.
Note: The Census Bureau in its November 2018 release (https://www.census.gov/construction/nrs/index.html) estimated a 90% confidence interval of 5.8%, which equals +/-
31,000 sales for all 532,000 sales from January through October 2018. The NAR does not publish a confidence interval with its Existing Home Sales
(https://www.nar.realtor/news-releases/2016/12/existing-home-sales-forge-ahead-in-November) but its numbers are based on a sample of about 40 percent of Multiple Listing
Services data each month. Sales outside of MLSs are not captured in the monthly series. The NAR states that it “rebenchmarks home sales periodically using other sources to
assess overall home sales trends, including sales not reported by MLSs.”
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, and First American Data Tree (DataTree.com), the NAR, and Census Bureau.
143
Home Sales, by Metro Market Size
In 2018:Q3, home sales started to decrease significantly in the largest 50 markets,
held steady in mid-size markets (top 51-100), and expanded in smaller metros/more
rural areas based on market size. This indicates further bifurcation of the market
with first-time homebuyers potentially shifting to typically less expensive markets.
NHMI by Metro Group (Count) (Index: 2013:Q3 = 100)
Years in the chart refer to Q3 of the labeled year.

125
123
120 120
120
116
116
114
115 113
112 112 113
112 112
110 110
110
108 108
107 107
105 105
105
100 100 100 100 100 100

100 99
98 98
97
95
90
Top 25 (41%) Top 26-50 (14%) Top 51-100 (13%) Rest/No CBSA (32%) All (100%)
2013 2014 2015 2016 2017 2018
* Simple average of FHFA’s CBSA Annual House Price Index for all CBSAs within group.
Note: Percentages in the chart refer to the respective CBSA market share by count in the most recent period. 144
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, First American Data Tree (DataTree.com), and FHFA.
Mortgage Risk Indices by Lender Type, Purchase Loans
A huge gap has opened up in the riskiness of purchase loans originated by banks and
nonbanks. Banks have reduced risk by shifting away from subprime borrowers and
low downpayment loans. Nonbanks have increased share by taking advantage of broad
Agency credit boxes and continued easing, thus making them the preferred risk
channel. While this share shift has stabilized, risk levels continue to diverge. The entire
year-over-year increase in risk is attributable to nonbanks.
16% 16%
Nonbank market share, purchase loans Nonbanks
Nov12 Nov 13 Nov 14 Nov 15 Nov 16 Nov 17 Nov 18
29% 41% 50% 55% 55% 56% 64%
14% 14%
Composite
12% 12%
10% 10%
Banks
8% 8%
Note: Composite includes credit unions and state housing agencies, which are not shown separately.
Jumbo portfolio-GSE spreads (in bps)
Portfolio jumbo rate has been below the GSE rate since 2014, reversing prior pattern.
The reasons are an increase in the GSE guarantee fees but also lenders may be bidding
aggressively for jumbo loans to obtain low-risk assets with cross-selling opportunities.
80
Period Spreads in bps

60 2001-2006 25
2007-2009 t 57
2010-2013 t 21
40 2014-2017 tt -26
20
GSEs less expensive

0
GSEs more expensive
-20 <--Underpricing by GSEs --> <--Financial Crisis-->
-40
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Note 1: Jumbo Portfolio minus GSE and Jumbo PMBS minus GSE spreads (in bps) between 90% and 110% of conforming limit.
Note 2: Chart omits PMBS-GSE spreads for years with less than 200 jumbo PMBS loans. Inset box uses loans for all years, except as indicated by line breaks.
Note 3: Data for 2017 are for January - September only.
Note 4: For loans between 90 percent and 110 percent of the applicable conforming loan limit
Source: AEI International Center on Housing Risk, www.aei.org/housing, and CoreLogic.

146
Homeowners Can’t Count on House Price Gains to Build Wealth
A better approach would be to focus on actual wealth building through
widespread adaptation of the Wealth-Building Home Loan (WBHL).
Using zip-level data for top 100 CBSAs to provide most complete analysis to date of risk by price tier
Share of zips with decline in house price index

Zip codes
1990-1995 1995-2000 2000-2005 2005-2010 2010-2015
Top price tier 28% 0% 0% 72% 15%
Middle price tier 37% 1% 0% 83% 30%
Bottom price tier 42% 2% 0% 84% 42%
Note: Top 100 CBSAs are defined by 2010 population. Analysis uses all five-digit zip codes in these CBSAs with a FHFA house price index
back to 1990 or earlier and a Zillow median house price in 2000. Zips with a median house price in 2000 in the bottom third, middle third,
and top third of all the zips in its CBSA are placed in the bottom, middle, and top price tiers, respectively.
• Prices rose almost everywhere from 1995 to 2005, but many zips saw declines in other periods,
especially 2005-2010
• Bottom price tier – where most first-time buyers locate – was worst-performing tier
• These results are for price indices, which average across many homes. Risk for individual homes
greater than shown here. WBHL mitigates this risk.
Source: AEI International Center on Housing Risk, www.aei.org/housing.

147
Evaluating the GSEs 2017 Business
Principle: the only plausible reason for government to back the housing market is to
help low- or moderate income families buy homes. An evaluation of the GSEs 2017
business shows, that the GSEs fail to meet this simple test.
Refi Cash Out

21% share
$300,000 median sales price (SP)
738 median FICO
Almost half of the GSEs’ 2017
Refi No Cash Out volume wasn’t even related to
19% share
$286,000 med. SP buying a primary residence.
746 med. FICO These borrowers could be served
2nd home & investor
by the private sector
7% share
$229,000 med. SP
774 med. FICO
Source: AEI Center for Housing Markets and Finance. All share percentages based on dollars (YTD Aug. 2017)
148
Evaluating the GSEs 2017 Business (cont.)
Another 41% went to help well-to-do buyers, of which 25 percentage points went to well-to-do repeat
buyers of primary residences and 16 percentage points went to well-to-do first-time buyers.
First-time buyer (FTB) w.>85% CLTV & loan>$250,000

8% share
$353,000 med. SP
746 med. FICO
FTB w.<85% CLTV

9% share
These buyers $280,000 med. SP
could be served 752 FICO Unrelated to buying
by the private a primary residence
Repeat buyer w. >85% CLTV & loan >$250,000
sector 8% share
$365,000 med. SP
755 FICO
Repeat buyer w. <=85% CLTV

18% share
$327,000 med. SP
774 med. FICO
149
Evaluating the GSEs 2017 Business (cont.)
Only 6.5% (1 in 16) GSE Dollars went to first-time buyers of more modest homes and only 3.7% (1 in 30)
GSE Dollars went to repeat buyers of more modest homes.
Repeat buyer w. >85% CLTV & First-time buyer w. >85% CLTV & loan<=$250,000
loan<=$250,000 6.6% share
3.7% share $168,000 median SP
$189,900 median SP 736 median FICO
755 median FICO
The private sector and a targeted and

reformed FHA could replace the GSEs
over time:
• The private sector could handle the 50% who are
not buying a primary residence and the 40%
well-to-do repeat & 1st time buyers of primary
residences
• The remaining 10% could be handled by the
FHA and the private sector
150
Update: John Burns Intrinsic Home Values
Over the past year the intrinsic over-valuation of the vast majority of metros has
increased – the most in the metros that were already highly valued . Almost 75% of
metros tracked by John Burns are overvalued today. These overvalued metros are
largely concentrated in CA, NV, FL, and AZ, (the Sand States—ground zero in last
boom/bust) and CO, TX, OR, and WA (states that largely sat out the last boom/bust).
Not updated National Intrinsic Home Value Index: +18%

50% 0.5
Reno, NV, 41%
0.45
Fairly Valued: 34 Metros. Over-Valued: 97 Metros
40% These metros account for 15% These metros account for nearly 53%
0.4
of the overall market* of the overall market*
Aug-18
0.35
30%
0.3
20% 0.25
Aug-17 0.2
10%
0.15
# of overvalued Metros
AZ, CA, FL, NV 47 (out of 49 metros) 0.1
0%
CO, TX, OR, WA 19 (out of 19 metros) 0.05
Hartford, CT, -6%
Rest 31 (out of 63 metros)
-10% 0
*Based on HMDA data for 2017.

Note: The Intrinsic Home Value Index shows current price versus intrinsic value assuming 6% mortgage rate. It tracks 131 metros in the U.S.
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing, and John Burns Real Estate Consulting.
151
GSEs: Large Lender Market Share and Relative Risk Share, Refinance Loans
2017/ 2017/ 2017/ 2017/ 2017/ 2017/
2013 2014 2015 2016 2013 2014 2015 2016
H1 Q3 Oct. H1 Q3 Oct.
NMRI 8.6% 9.7% 8.3% 7.8% 8.9% 9.1% 9.1% 8.6% 9.7% 8.3% 7.8% 8.9% 9.1% 9.1%
Quicken
Wells Fargo
Loan Depot
JP Morgan
United Shore
Flagstar Nationstar
Pennymac
US Bank
Caliber
SunTrust
Amerihome
Ditech
BB&T
Franklin
American
Fifth Third
Freedom
Mortgage
Bank of Fairway
America
Stearns
30% 20% 10% 5% 1% Larger circle represents larger Guild

market share. Lenders shown
and 15 largest nonbanks by Plaza
origination share in 2016:Q3.
25+% 15 to 25% 5 to 15% 5 to -5% -5 to -15% -15 to -25% -25+% Lakeview
Lower FHA risk share

15
Higher FHA risk share
(relative to market share) (relative to market share) 2
FHA: Large Issuer Lender Type Market Share and Relative Risk Share, Refinance Loans
2017/ 2017/ 2017/ 2017/ 2017/ 2017/
2013 2014 2015 2016 2013 2014 2015 2016
H1 Q3 Oct. H1 Q3 Oct.
NMRI 18.3% 21.3% 21.8% 22.4% 23.6% 24.6% 25.0% 18.3% 21.3% 21.8% 22.4% 23.6% 24.6% 25.0%
Quicken
Flagstar
Freedom
Mortgage
Wells Fargo
PennyMac
LoanDepot
US Bank
Nationstar
JP Morgan
Lakeview
SunTrust Amerihome
Caliber
Fifth Third
Ditech
BB&T Plaza
Stearns
Amerihome
United
Shore
30% 20% Larger circle represents larger
10% 5% 1% market share. Lenders shown Guild
represent the largest 8 banks
and 15 nonbanks by origination
share in 2016:Q3. Fairway
25+% 15 to 25% 5 to 15% 5 to -5% -5 to -15% -15 to -25% -25+% Franklin
Higher GSE risk share Lower GSE risk share 15

3
Leverage Fueled Housing Demand Continues to Climb
Even though the rate of increases have slowed over the past three years, volume is
still growing from a high base. Compared to 3 years prior, October 2017 volume by
count is up 24 percent; first-time buyer volume is up 30 percent.
50% 50%
Not updated
Change from 3 years ago
40% 40%
30% 30%
20% 20%
10% 10%
0% 0%
Year-over-year change
-10% -10%
Sep-13 Dec-13 Mar-14 Jun-14 Sep-14 Dec-14 Mar-15 Jun-15 Sep-15 Dec-15 Mar-16 Jun-16 Sep-16 Dec-16 Mar-17 Jun-17 Sep-17

154
Agency Origination Shares by Risk, Purchase Loans
Fannie and, to a lesser extent, Freddie have expanded their holdings of higher risk
subprime loans, now accounting for a combined 27 percent of such loans, up from 7.5
percent in Sept. 2012. While Freddie has offset this by adding safer, prime and near-
prime loans, Fannie has shed some of its business in these categories. Over the past 6
months, Fannie appears to be changing its positioning as talks of GSE reform heat up.
Prime Near Prime Subprime
80% 80%
70% 70%
FHA
60% 60%
Fannie
50% 50%
Fannie
40% Freddie 40%
30% 30%
Freddie
Purple: RHS
Fannie
20% 20%
VA RHS
VA
VA
10% FHA 10%
FHA RHS
Freddie
0% 0%
Feb-13 Dec-13 Oct-14 Aug-15 Jun-16 Apr-17 Feb-18 Feb-13Oct-13 Jun-14Feb-15Oct-15 Jun-16Feb-17Oct-17 Jun-18 Feb-13 Dec-13 Oct-14 Aug-15 Jun-16 Apr-17 Feb-18
* We define prime loans as low-risk (with a stressed default rate of less than 6%), near prime as medium risk (with a stressed default rate of 6% to
less than 12%), and subprime as high risk (with a stressed default rate of 12% or greater).
What Does this Mean for the Broader Market?
Due to FHA’s loose lending standards, historically high loan limits and market share,
and an appraisal process focused on market price, not market value, FHA borrowers
are setting the price for a large share of the market including conventional loan buyers.
Law of Marginal Buyer: home prices will keep rising so long as the marginal buyer, who
sets price for all, has access to higher leverage. Historically, the government, has been
the most willing provider of this leverage.
* Source: John Ligon, Heritage Foundation

156
Borrowing at the Conforming Loan Limit, GSE Purchase Loans
Current policy is driving loan balances higher during a very tight market. FHFA
first raised the conforming loan limit from $417,000 to $424,100 in Jan. 2017, then
to $453,100 in Jan. 2018.* Borrowers in non-high cost areas immediately borrowed
at the new maximum. The same holds for high-cost areas (not shown).
Share of Loans
6% 6%
Dashed lines mark the change in the Not updated

conforming loan limit from $417,000 to % of Loans with Loan Amount over
5% $424,100 in Jan. 2017 and the change in the 5%
$424,100 and at or below $453,100
conforming loan limit from $424,100 to
$453,100 in Jan. 2018.
4% 4%
% of Loans with Loan Amount over

$417,000 and at or below $424,100
3% 3%
% of Loans with Loan Amount
of $417,000
2% 2%
1% 1%
0% 0%
Jan-16 Apr-16 Jul-16 Oct-16 Jan-17 Apr-17 Jul-17 Oct-17 Jan-18
*FHA and VA also raised its maximum guaranty amount in line with FHFA and HUD. Data for February 2018 are partial.
157
The CFPB’s Qualified Mortgage Policy and GSE QM Patch
Allowed for Credit Easing While Supply Is Constrained, a Direct
and Continuing Cause of the Current House Price Boom
• In 1.13, “Ability-to-Repay and Qualified Mortgage Standards” rule issued, effective 1.10.14
• The Bureau noted it will “protect consumers from irresponsible mortgage lending.”
• The rule effectively set a maximum debt-to-income (DTI) limit of 43% for the private sector.
• GSEs and their automated underwriting systems were exempted from this provision for seven years.
• Similarly, FHA, the VA and the Department of Agriculture’s Rural Housing Services (RHS) , were
exempted for up to seven years or until these agencies issued their own rules codifying their own
lending practices (which all subsequently did).
• The QM rule was pursuant to the Dodd-Frank Act’s calling for minimum mortgage standards
• It was to make sure “prime” loans will be made responsibly
• Yet it sets no minimum down payment, no minimum standard for credit worthiness, and no maximum
debt-to-income ratio (for government agencies)
• Under this definition of “prime”, a borrower can have no down payment, a credit score of 580, and a
debt ratio over 50% as long as they are approved by a government-sanctioned underwriting system.
• That this would promote an unsustainable home price boom could be foreseen:
• In 2013: “Booms are fueled by excessive leverage” and “this rule does little to limit borrower leverage
and lays the foundation for the next bust.”*
• In 1951: “[In transitioning] from a buyer's to a seller's market, maximum terms become so commonly
used they tend to be considered the minimum.”**
• The QM Patch does not operate counter-cyclically so as to “take the punch bowl away” so as
to slow a leverage-fueled price boom.
*Pinto, “CFPB’s new ‘qualified mortgage’ rule: The devil is in the details”, http://www.aei.org/publication/cfpbs-new-qualified-mortgage-rule-the-devil-is-in-the-details/
Wallison and Pinto, “New Qualified Mortgage rule setting us up for another meltdown” https://www.washingtontimes.com/news/2013/mar/3/wallison-and-pinto-new-qualified-mortgage-rule-set/
**Fisher, Financing Home Ownership, NBER, 1951 158
Number of Investors Flipping Houses Creeping Up
Due to higher house prices and cash availability house flipping is making a
comeback. Levels today are back to the levels seen in 2003.
Not updated
100,000 100,000
90,000 90,000
80,000 80,000
70,000 70,000
60,000 60,000
50,000 50,000
40,000 40,000
30,000 30,000
20,000 20,000
10,000 10,000
0 0
2000:Q1 2002:Q1 2004:Q1 2006:Q1 2008:Q1 2010:Q1 2012:Q1 2014:Q1 2016:Q1
Entirely a buyer's market Entirely a seller's market
Note: National Association of Realtors (NAR) defines a seller's market as inventory that is less than or equal to 6 months of sales.
Source: ATTOM Data Solutions and the NAR.
159
Average House Price Change by Zip (%, annual avg.)
1990- 1995- 2000- 2005- 2010- 1990-
Top 100 CBSAs
1995 2000 2005 2010 2015 2015
All zips 1.3 4.9 9.3 -2.8 1.9 2.8
By price tier within CBSA
Top-tier zips 1.7 5.3 8.4 -1.7 2.2 3.1
Middle-tier zips 1.4 4.8 9.2 -2.8 1.9 2.8
Bottom-tier zips 0.8 4.6 10.3 -3.9 1.7 2.5
Note: Top 100 CBSAs based on 2010 population. House prices are measured using FHFA’s all-transactions HPI for five-digit zips and
are combined with Zillow’s all single-family residences median house price in 2000 for about 5,300 zips in total. Zip codes are assigned
to tiers based on the median house price in 2000. The price changes in the table are unweighted averages across the included zip
codes. For more, see https://www.aei.org/wp-content/uploads/2017/12/Wealth_Building_WP.pdf.
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing, FHFA, and Zillow.
• Over the past quarter century, the average rate of house price appreciation
has been slow and subject to substantial volatility.
• The outcomes for buyers in bottom-tier zips are worse than for buyers in top-
tier zips, with lower average appreciation and greater volatility.
• The difference in volatility was especially pronounced during the housing
boom (2000- 2005) and bust (2005-2010).
• The reasons:
– Perhaps widening of income equality
– cyclical swings in mortgage lending standards have a greater impact on FTBs than on RBs
160
Raising the Conventional Loan Limit – A Prediction
Raising the conforming loan limit during a seller’s market will drive up borrowing and
therefore likely increase house prices. A case in point is San Diego, CA.
San Diego, CA, MSA: Freddie Loan Distribution, 2013 San Diego, CA, MSA: Freddie Loan Distribution, 2015
250 250
200 200
Conforming loan limit Conforming loan limit
# of loans
150 in 2013: $546,250 150 in 2015: $562,350

100 100
50 50
0 0
$530 $535 $540 $545 $550 $555 $560 $565 $570 $575 $580 $585 $530 $535 $540 $545 $550 $555 $560 $565 $570 $575 $580 $585
San Diego, CA, MSA: Freddie Loan Distribution, 2014 San Diego, CA, MSA: Freddie Loan Distribution, 2016*
250 250
200 200
Conforming loan limit Conforming loan limit

# of loans
150 150
in 2014: $546,250 in 2016: $580,750
100 100
50 50
0 0
$530 $535 $540 $545 $550 $555 $560 $565 $570 $575 $580 $585 $530 $535 $540 $545 $550 $555 $560 $565 $570 $575 $580 $585
Loan Amount Bin (in $1,000 intervals) Loan Amount Bin (in $1,000 intervals)
* Through November 2016. Data point for $580,000 bin in 2016 is 315 loans.
Note: Data are for 1-unit properties only.
Raising the Conventional Loan Limit – A Good Idea?
Raising the conforming loan limit would be of no value to the vast majority of FTBs.
Furthermore, conforming loan limit acts as a constraint on house prices during a
seller’s market. Removing it will: 1) drive up borrowing, 2) increase house prices, and
3) shift market share away from the private lenders to the GSEs. Thus the impact on
affordability is largely overstated.
7%
Share of Borrowers at or above $417,000
Median
Agency First-time Repeat
6%
GSEs 6.3% 7.6%
September 2015 – August 2016 Fannie 6.2% 7.5%
5% Freddie 6.6% 7.7%
FHA 3.4% 5.2%
RHS 0.1% 0.2%
4%
VA 6.2% 12.3%
All 4.8% 7.6%
3%
$417,000
2%
$625,500
1%
0%
$0 $100,000 $200,000 $300,000 $400,000 $500,000 $600,000
Note: Bar chart refers to first-time buyer loans originated from September 2015 – August 2016.
162
Cash-Outs and the Economy
Cash-out refis (CO) have grown in share and absolute number. As equity is extracted
from real estate, it gets recycled into the economy driving up GDP. Problem: house
prices have risen faster than fundamentals can support (with no end in sight). In the
long-run, this makes a house price correction likely. However, as equity is extracted
from real estate, owners have less capital to protect themselves from house price
declines.
Cash-out counts and share

140,000 60% Back-of the envelop calculation:
Cash-out Refi Count (left axis)
120,000
Cash-out share of refis (right axis)
50% According to Black Knight, $31bn in equity
100,000
was extracted via COs in 2016:Q4.
40% Cash-out extraction was 50% higher y-o-y.
80,000
30% On an annualized rate, this amounts to
60,000
$120bn in 2016, or a $60bn increase over
40,000
20% 2015.
20,000
10% For an $18.5tr economy, this amounts to
an extra annualized stimulus of ~0.3% of
- 0%
Nov-12 Nov-13 Nov-14 Nov-15 Nov-16 Nov-17
GDP.
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing and Black Knight. 163
Cash-Out Share and Expected Defaults
As cash-out share has grown, its agency composition has also changed. Compared
to the series start, VA and FHA have tripled their share by loosening lending
standards faster than the GSEs. Today, they account more over half of the expected
defaults, up from just 20%.
Market Share Not updated Expected Defaults under Stress
100% 9,000
90%
8,000
GSE
80%
GSE 7,000
70%
6,000
60%
5,000
50% Ginnie
4,000
40%
Ginnie 3,000
30%
2,000
20%
10% 1,000
0% 0
Sep-12 May-13 Jan-14 Sep-14 May-15 Jan-16 Sep-16 May-17 Jan-18 Sep-12 May-13 Jan-14 Sep-14 May-15 Jan-16 Sep-16 May-17 Jan-18

Punchbowl 1: Mortgage Rate Changes Applicable to
FTBs and RBs
Mortgage rates, due to Fed easing, have been near all time lows. Rates for FTBs and
RBs have moved in lock-step with a small premium for RBs over FTBs. The boost
from lower rates has therefore applied equally to both buyer types – as has the
increase in rates since November 2016.
Average GSE Note Rate: Primary Owner-Occupied 30-yr Fixed-Rate Purchase Mortgages
5.00% 5.00%
4.75% 4.75%
First-time buyers Repeat buyers
4.50% 4.50%
4.25% 4.25%
4.00% 4.00%
3.75% 3.75%
3.50% 3.50%
Note: Data are for GSE primary owner-occupied 30-year fixed-rate purchase mortgages with credit scores of 720-769, CLTVs of 76-80, and DTIs of 39-43.
165
Punchbowl 2: Eased Underwriting Standards
Only Available to Agency First-time Buyers
The Agency First-time Buyer MRI (FBMRI) stood at 16.7% in August, up 0.1 ppt from a
year earlier and up 2.6 ppts. from 5 Agency RBMRI is virtually unchanged since August
2013 (down 0.1 pyears earlier. The pts.). The Agency FBMRI is 7.5 ppts higher than the
Agency RBMRI, 0.5 ppt. wider than the gap a year earlier. If the FBMRI trend continues,
it will reach almost 20% by August 2022.
36% 36%
Historical Projection
32% 32%
28% FHA FTB 28%
24% 24%
20% 20%
Agency FTB
16% 16%
12%
Agency RB 12%
8% 8%
Feb-13 Oct-13 Jun-14 Feb-15 Oct-15 Jun-16 Feb-17 Oct-17 Jun-18 Feb-19 Oct-19 Jun-20 Feb-21 Oct-21 Jun-22
Note: Calculated for primary owner-occupied home purchase mortgages.

166
Agency Purchase Loan Demand Remains Strong
These two punchbowls have largely driven strong growth in agency volume. However,
the market has plateaued at its current high level. Agency FTB volume was unchanged
compared to one year ago and up 38 percent compared to five years ago. RB volume
has pulled back slightly, but still up 21 percent from five years ago.
160 160
FTB vs RB Agency Transactions Index:

150 Feb-2013 to Jan-2014 = 100 150
FTB
140 140
130 130
RB
120 120
110 110
100 100
90 90
80 80
Jan-14 Jul-14 Jan-15 Jul-15 Jan-16 Jul-16 Jan-17 Jul-17 Jan-18 Jul-18
Note: First-time buyer volume not available before February 2013. The index is a 12 months rolling index.
167
Market Segmentation:
Median Sales Price for First-time and Repeat Buyers
The housing market is largely segmented by price. FTBs, or entry level buyers,
traditionally buy at lower price points than RBs, or move-up buyers. Lately, FTBs have
reduced the gap to RBs, an indication that recipients used added buying power from
looser lending to bid up FTB homes, ironically made more expensive by FTB leverage,
as RBs have had to make downward quality adjustments.
350,000 350,000
RB
300,000 300,000
250,000 250,000
FTB
200,000 200,000
150,000 150,000
FTB to RB median sales price ratio

100,000 Sept-12: 69% 100,000
Aug-18: 74%
50,000 50,000
0 0
Sep-12 Mar-13 Sep-13 Mar-14 Sep-14 Mar-15 Sep-15 Mar-16 Sep-16 Mar-17 Sep-17 Mar-18
Note: Data are for primary owner occupied properties only.

Constant Quality Prices Outlook for the Bifurcated Market:
Slowing Price Appreciation for at the Higher End, Continued
Robust Appreciation for at the Lower End
During the recent house price boom, the lower tiers of the market have experienced
faster house price appreciation (HPA) due to the interaction of greater availability of
leverage and extremely low inventory. The higher tiers have seen more restrained
HPA. Even as the rate punchbowl is further withdrawn, we expect lower tier HPA to be
robust as available leverage continues to power prices. Significantly, our research
shows that a concentration of about 30% highly-leveraged borrowers in a census tract
can raise prices for everyone in the tract. For the higher tiers, which mostly consist
of RBs, who are less reliant on leverage, the prediction is a slight moderation in HPA.
Cumulative Constant-Quality HPI, by Price Tier (2012:Q4 = 0%)
50% 50%
45% Low 45%
40% Low-Med 40%

Med-High
35% 35%
High
30% 30%
25% 25%
20% 20%
15% 15%
10% 10%
5% 5%
0% 0%
2012:Q4 2013:Q2 2013:Q4 2014:Q2 2014:Q4 2015:Q2 2015:Q4 2016:Q2 2016:Q4 2017:Q2 2017:Q4 2018:Q2
Note: HPIs are smoothed around times of FHFA loan limit changes. Date are for 73 largest CBSAs.
Outlook for Bifurcation of Market – Quality Changes
So far in the boom, borrowers in the higher price tiers have offset higher constant-
quality (CQ) prices by reducing the quality of their home purchases. This quality
adjustment has kept the market transaction price nearly flat. As the rate punchbowl is
withdrawn further, we expect additional quality offsets in higher price tiers to
compensate for higher rates. We expect borrowers in lower price tiers to continue to
use the leverage punchbowl to largely offset both higher CQ prices and interest rates.
50% 50%
45% 45%
Quality
40% 40%
offset
35% 35%
30% 30%
Constant-quality
25% 25%
20% 20%
Constant-quality
15% 15% Quality
Market Expenditure offset
10% 10%
5% 5% Market Expenditure
0% 0%
Note: HPIs are smoothed around times of FHFA loan limit changes. Date are for 73 largest CBSAs.
170
Outlook for a Bifurcated Market –
Transaction Prices by State & Changes in Transaction Volume
There is evidence for this market bifurcation at the state level. States with lower
median prices, which also tend to be states with higher risk scores, have seen their
volume flatten or increase, while higher priced states, which tend to be states with
lower risk scores (and which tend to be mostly featured in the media), have tended to
see declines. This trend will likely continue. Yet, there is a second component to this
story as the next slide shows.
15%
VT June-August 2018
10%
OK DE
AR
Change in Agency Purchase Volume
IN WY
5% LA
AL NM
WVMS PA ME
OH MO SC GANC FL
from One Year ago
KY TX HI
0% KS NE AZ
WISD TN CT VA
ID RI NH NJ
MI MN MT UTMD
IL NV
OR CO
-5% NY WA
IA AK MA
-10% ND
CA
R² = 0.3278
-15%
DC
-20%
$100,000 $150,000 $200,000 $250,000 $300,000 $350,000 $400,000 $450,000 $500,000 $550,000 $600,000
Median Agency Transaction Price
171
Outlook for a Bifurcated Market – Transaction Prices by State
& Changes in Median Transaction Prices
There is no correlation between the level of state transaction prices and changes in
said transaction prices. Transaction prices are still rising in virtually every state, with
the few exceptions having more or less flat transaction prices over the last year. This
is further evidence of the bifurcation of the market, where slowing house price
appreciation (HPA) at the top is offset by continuing rapid HPA at the bottom.
16%
ID
14%
Change in Median Agency Transaction Price
June-August 2018
12% UT
NV
10%
from One Year ago
OR WA CO
8%
TN ME AZ MT
6% MINE SD GA R² = 0.0071
SC NCMN FL NH
IN
AL NY
TX MA
4% MO WINMPA WY ND RI CA
OK KY
OH AR MS CT AK MD
NJ
IA DE HI
2% IL
KS LA
VT VA
0% WV
DC
-2%
$100,000 $150,000 $200,000 $250,000 $300,000 $350,000 $400,000 $450,000 $500,000 $550,000 $600,000
Median Agency Transaction Price
172
While FHA’s Forward Program Capital Is at 3.9%, in
Excess of Statutory Minimum of 2%, It Should be 7%
• FHA’s 2014 Annual Report to Congress provides a useful starting point and methodology for evaluating an
appropriate level of capital today for FHA’s forward program.
– The 2014 report concludes that a 8.5% capital buffer on outstanding insurance in force is needed
– Our analysis adjusts this ratio upward to 9% to account for the growing risk of FHA’s portfolio.
– Our analysis also excludes the substantial negative impact of the HECM program on the capital level of the
Mutual Mortgage Insurance Fund.
• Applying the 2014 framework to the 2018 book, we find that FHA had a Capital Resource shortfall of 1.1%
or about $13 billion
– At end of FY 2018, FHA had $1.2 trillion in outstanding insurance in force (IIF).
– FHA reported at 9.30.18, Capital Resources of 3.9% or $46.8 billion on $1.2 trillion outstanding
– Given the 9% assumptions from the 2014 report, this suggests that FHA would need $108 billion in NPV
Claims-Paying Capacity in the next crisis, similar to the Great Recession.
• This assumes that steady state MIP income would provide a stream of 4% or $48 billion
• Therefore, the Capital Resources portion would need to equal 5.0% or $60 billion, of which only $46.8 billion are currently covered
• Yet, FHA’s 2014 approach does not adequately address the current risks. While the 2014 Report noted
“capital [in the form of house price appreciation] disappears in times of stress,” it misses:
– The current home price appreciation (HPA) for entry level homes has again been inflated by excess leverage,
most of which has been provided by FHA.
– Entry-level homes’ faster HPA vs the slower HPA of non-highly leveraged higher priced homes
– FHA’s sizable market share, its geographic concentration, and how its underwriting policies are exacerbating
the current house price cycle.
• When taking these factors into account, FHA’s Capital Resources shortfall rises to 3.1% or about $37 billion
– We think a buffer of an additional 2% in Capital Resources should be provided for each 10% that home
prices in the low price tier have increased faster than prices in the med-high and high price tiers (currently a
difference of +16%)
• Today, this would require $24 billion in additional Capital Resources for a total of $84 billion to support $1.2 trillion in IIF
• Thus, Capital Resources would need to total 7% or $84 billion (forward program only) compared to the 3.9% or $46.8 billion at
9.30.18 for the forward program 173
FHA Cash Out Count and MRI
FHA commissioner Brian Montgomery stated that the agency was closely monitoring
“the exponential rise in cash-out refinance transactions.” FHA’s CO volume has
tripled from the beginning of the series and risk has increased from 19% to 27%. COs
do nothing to promote homeownership for lower-income and minority buyers.
FHA Cash Out Count FHA Cash Out MRI

15,000 30%
28%
12,500
26%
10,000
24%
7,500
22%
5,000
20%
2,500
18%
0 16%
Sep-12 Sep-13 Sep-14 Sep-15 Sep-16 Sep-17 Sep-12 Sep-13 Sep-14 Sep-15 Sep-16 Sep-17
174
Average Credit Score and DTI: FHA Purchase Loans
FHA commissioner Brian Montgomery also stated that the agency was closely
monitoring “a continuing increase in the average FHA-insured borrower’s debt-to-
income ratio, and declining average credit scores.” FHA’s average credit score for
purchase loans has dropped from 697 in September 2012 to 671 in August 2018, while
it’s average DTI has risen from 40 to 44.1 over the same time period. Lower credit
scores are often combined with higher DTIs, a process known as risk-layering.
Average Credit Score Average DTI

700 45
695 44
690
43
685
42
680
41
675
40
670
39
665
660 38
655 37
Sep-12 Sep-13 Sep-14 Sep-15 Sep-16 Sep-17 Sep-12 Sep-13 Sep-14 Sep-15 Sep-16 Sep-17
175
FTB Purchase Loan NMRI: Credit Easing Continues
The First-time Buyer MRI continued to increase. Setting a new series high, FHA’s First-
time Buyer MRI stood at 28.6% in February, up 2.1 ppts from a year earlier. With
individual agencies easing credit standards and continued home price escalation, we
expect higher FBMRIs in the coming months.
2.5%
Change from 12 months earlier, in percentage points 2.5%
FHA FTBs
2.0% 2.0%
1.5% 1.5%
All Agency FTBs
Fannie FTBs
1.0% 1.0%
0.5% 0.5%
Easing
Freddie FTBs
0.0% 0.0%
(blue)
Tightening
-0.5% -0.5%
-1.0% -1.0%
Feb-14 Jun-14 Oct-14 Feb-15 Jun-15 Oct-15 Feb-16 Jun-16 Oct-16 Feb-17 Jun-17 Oct-17 Feb-18 Jun-18
176
Which Risk Factors Have Driven Up the FTB NMRI?
Share of first-time buyer home purchase loans with
Aug Aug Aug Aug Aug Aug

Risk factor
2013 2014 2015 2016 2017 2018
Credit score < 660 15% 19% 21% 21% 21% 23%
DTI > 43% 24% 24% 27% 27% 31% 38%
CLTV ≥ 95% 64% 67% 71% 71% 71% 71%
30-year term 95% 96% 97% 97% 97% 97%
Risk Layering 27% 30% 34% 35% 37% 41%

Note: Calculated for primary owner-occupied home purchase loans with a government guarantee and reported risk factor.
Risk layering is defined as having at least 3 of the 4 risk features presented in the table above present in a loan.
• Since 2012, all the key risk factors have contributed, which has magnified the
effect on the NMRI through risk layering.
• Over the past 3 years DTIs have contributed the most to a higher FTB NMRI,
with about one-third of FTBs having a DTI in excess of the QM “limit” of 43
percent
177
Share of GSE FTB Purchase Loans w. DTIs of 46-50%
DTI limits should “take the punch bowl away” so as to slow a leverage-fueled price
boom. The GSEs had standard DTIs as high as 45%, but traditionally allowed DTIs up
to 50% with compensating factors. In this 46-50% DTI range, Freddie has historically
outpaced Fannie. Fannie responded in Aug. 2017 by eliminating the requirement for
compensating factors. The GSEs’ competition on income leverage, combined with
FHA’s even looser DTI standards, will help fuel the ongoing price boom.
20%
18%
16%
14%
12%
10%
8%
6%
4%
2%
0%

FTB Purchase Loans, by Level of Downpayment
It is often reported that down payments of 20% are an impediment to homeownership
today. The truth is that VA and RHS don’t require any downpayment at all. And down
payment or closing cost assistance is available through State Housing Finance
Agencies, with about 15% of FTBs taking advantage of these programs, which often
lowers their downpayment to $0. FHA purchasers have an average CLTV of 98%. Only
17% of FTBs put down 15% or more.
40% 40%
35% 35%
3-3.5%
30% 30%
No downpayment
25% 25%
20% 20%
15% or more
15% 5% 15%
10% 10% 10%
5% 5%
0% 0%
Note: Calculated for primary home purchase loans with a government guarantee and reported CLTV. Borrowers with downpayment assistance and
CLTVs of 95% or greater are assumed to have no downpayment. Terms and conditions of the downpayment assistance programs vary by program, but
in most cases they allow borrowers to offset the downpayment entirely. 179
Source: FHA and AEI Center on Housing Markets and Finance, www.AEI.org/housing.
Agency First-time Buyer Purchase Loan Share
Agency FTB share for August stood at 57.8%, up 0.3 ppt from a year ago. FTB share has
likely reached saturation with tight inventory holding back buyers. An expanding
economy and further credit easing will help maintain current levels as they offset higher
prices and higher mortgage rates.
61% 61%
60% 60%
59% 59%
58% 58%
57% 57%
56% 56%
Red markers show August share in each year.

55% 55%
54% 54%
53% 53%
Note: First-time buyer volume not available before February 2013.

180
Government Housing Policy Creates an Economics Free Zone
• Law of the Marginal Buyer: In a seller’s market, prices rise faster than
incomes as long as marginal buyer, who sets the price for all, has access
to higher leverage. Determines not only price level, but also degree of
stability, as price is not necessarily equal to value.
• Fisher’s Law: [I]n a seller's market, when choice is restricted and the seller
virtually dictates sales terms, more liberal credit is likely to be capitalized in
price.*
• Law of Ignorance: Policy makers ignore principles of supply, demand, and

housing finance, resulting in an economics free zone. Cross-subsidies and
expanded access to credit push up demand against a regulation-
constrained supply.
* Fisher, Financing Home Ownership, NBER, 1951 (FHA’s first chief

economist)
181
Definition of Low-Risk / Prime Loans
• We define low-risk / prime loans as those with a stressed default rate of

less than 6%. Why?
• Low-risk / prime definition calibrated from two sources

– Original QRM proposal to implement Dodd-Frank
– FHA underwriting standards over 1935-55
– Both yield an average stressed default rate of ≈ 3%
• This is consistent with a maximum stressed default rate of ≈ 6% on

individual loans, assuming a uniform distribution starting near 0%
• Hence the use of 6% as the highest stressed default rate for a low-risk /
prime loan
182
Agency First-Time Buyer Loan Count
Agency FTB volume remained unchanged and up 38 percent compared to one and
five years ago, respectively.
200,000 200,000
180,000 180,000
Red markers show August count in each year.
160,000 160,000
140,000 140,000
120,000 120,000
100,000 100,000
80,000 80,000
60,000 60,000
Note: For primary owner-occupied home purchase mortgages with a government guarantee. November 2017 count is a preliminary estimate.
183
Agency Origination Shares, FTB Purchase Loans
FHA FTB origination market share, which jumped after its cut in mortgage insurance
premium (MIP) in January 2015, has been gradually trending down over the last two
years. Since then, the GSEs started clawing back some of the market share they had lost.
In August, FHA’s FTB share was near its pre-MIP cut level.
45% 45%
40% 40%
FHA
35% 35%
30% 30%
Fannie
25% 25%
20% 20%
15% Freddie 15%

VA
10% 10%
RHS
5% 5%
0% 0%
184
Origination Shares by Credit Score Bin, First-time
Buyer Purchase Loans
Story in the media has been of too tight credit holding back first-time buyers.
Reality is the long-term trend has been towards looser credit and record setting
volume (especially FTB). Especially noteworthy is the influx of subprime
borrowers with credit score below 660.
30% 30%
25% 25%
660-699
700-739
<660
20% 20%
740-779
15% 15%
≥780
10% 10%

185
Agency Origination Shares, FTB Purchase Loans by
Market Segment
GSEs dominate prime segment accounting for 86% of that market. FHA has consolidated
most of the subprime segment. Competition is greatest in near-prime segment.
Prime Near Prime Subprime

80% 80%
70% 70%
FHA
60% 60%
Fannie
50% 50%
Fannie
40% 40%
Freddie
30% 30%
Freddie
Purple: RHS
20% Fannie 20%
VA RHS
VA
FHA VA
10% 10%
RHS
FHA
0%
Freddie 0%
Feb-13 Dec-13 Oct-14 Aug-15 Jun-16 Apr-17 Feb-18 Feb-13 Dec-13 Oct-14 Aug-15 Jun-16 Apr-17 Feb-18 Feb-13 Dec-13 Oct-14 Aug-15 Jun-16 Apr-17 Feb-18
* We define prime loans as low-risk (with a stressed default rate of less than 6%), near prime as medium risk (with a stressed default rate of 6% to
less than 12%), and subprime as high risk (with a stressed default rate of 12% or greater).
Originations by Market Segment, FTB Purchase Loans
The high-risk subprime market segment continues to outpace the growth in the lower-
risk segments. The near-prime segment now accounts for equal number of loans as low-
risk prime segment.
100,000 100,000
90,000 90,000
80,000 80,000
70,000 70,000
60,000 60,000
Subprime
50,000 50,000
40,000 40,000
Prime
30,000 30,000
20,000 Near-prime 20,000
10,000 10,000
0 0
Feb-13 Jun-13 Oct-13 Feb-14 Jun-14 Oct-14 Feb-15 Jun-15 Oct-15 Feb-16 Jun-16 Oct-16 Feb-17 Jun-17 Oct-17 Feb-18 Jun-18
* We define prime loans as low-risk (with a stressed default rate of less than 6%), near prime as medium risk (with a stressed default rate
of 6% to less than 12%), and subprime as high risk (with a stressed default rate of 12% or greater).
Combined First-Time Buyer Mortgage Share Index
Combined first-time buyer share at 54.4% in August, up 0.3 ppt from a year earlier.
The NAR’s monthly realtor survey is badly flawed, providing a much noisier picture,
and as of recently, perhaps the wrong trend. NAR Sep ‘18 down 3 ppt. from Sep ’17.
60% 35%
NAR realtor survey (moved
Red markers show August share in each year. ahead 2 months), right scale
58%
33%
56%
31%
54%
AEI measure, left scale

52%
29%
50%
27%
48%
46% 25%
Note: Calculated as a share of primary owner-occupied home purchase mortgages (both government guaranteed and private-sector mortgages). The
NAR’s monthly survey (http://www.realtor.org/reports/realtors-confidence-index) is sent to more than 50,000 realtors (out of a total of 1.3 million
members), but has a low response rate; only 7,605 responses were received for the March 2018 survey.
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing, and the NAR. 188
The NAR’s first time buyer series is fatally flawed. After
removing seasonality, most of what remains is noise
As a result, the NAR series yields little real trend information.
AEI’s First-time Buyer Market Share Index (FBMSI) conveys real trend information.
Bottom line: don’t use the NAR survey.
6% 6%
Y-o-Y Change in NAR FTB Share
Y-o-Y Change in AEI FBMSI
4%
4%
2% R² = 0.364
R² = 0.4996
0%
2%
-2%
-4%
0%
-6%
-8%
-2%
Based on monthly survey with around 2,000 Based on census of agency loans: April figure based
-10% responses for closed sales. on 275,000 loans, over 100 times more than the NAR.
-12% -4%
Feb-14 Oct-14 Jun-15 Feb-16 Oct-16 Jun-17 Feb-18 Feb-14 Oct-14 Jun-15 Feb-16 Oct-16 Jun-17 Feb-18
Note: Calculated as a share of primary owner-occupied home purchase mortgages (both government guaranteed and private-sector mortgages). The
NAR’s monthly survey (http://www.realtor.org/reports/realtors-confidence-index) is sent to more than 50,000 realtors (out of a total of 1.3 million
members), but has a low response rate; only 4,555 responses were received for the April 2018 survey.
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing, and the NAR. 189
Share of States with Rise in First-time Buyer Loan
Volume and Share from Year-Earlier Period*
First-time buyer loan volume is trending up in vast majority of states. First-time
buyer share is also trending up in two-thirds of states due to faster growth than
repeat buyers.
NOT UPDATED
100% 100%
Volume
90% 90%
80% 80%
70% 70%
Share
60% 60%
50% 50%
40% 40%
Percent changes calculated from
year-earlier three-month average.
30% 30%
20% 20%
Feb '14 - May '14 - Aug '14 - Nov '14 - Feb '15 - May '15 - Aug '15 - Nov '15 - Feb '16 - May '16 - Aug '16 - Nov '16 - Feb '17 -
Apr '14 Jul '14 Oct '14 Jan '15 Apr '15 Jul '15 Oct '15 Jan '16 Apr '16 Jul '16 Oct '16 Jan '17 Apr '17
*Final value for each series based on change in each state from December 2015-Febraruy 2016 average to December 2016-February 2017 average.
Earlier values calculated analogously.
190
Profiles of GSE and FHA First-time Buyers with >95% CLTV
The GSEs are primarily expanding their high CLTV business to higher credit score
borrowers. These borrowers mainly profited from lower PMI capital requirements which
resulted in lower insurance fees. FHA has expanded further down the credit distribution.
With high credit scores and relatively low DTIs, GSE risk index (14.8%) is about half of
FHA’s (29.1%).
Share of GSE and FHA FTB Loans with CLTV > 95 by Credit Score
Bin
20%
18% March 2016

August 2018
16%
14%
12%
10%
8%
6%
4%
2%
0%

191
Characteristics of Mortgages Taken Out by First-Time and
Repeat Homebuyers
August 2018
30-year CLTV ≥ Credit Risk
DTI > 43%
Term 95% Score < 660 Layering
First-time Buyers 97% 71% 23% 38% 41%
Repeat Buyers 93% 39% 10% 36% 22%
Note: Calculated for primary owner-occupied home purchase loans with a government guarantee and reported risk factor.
Risk layering is defined as having at least 3 of the 4 risk features presented in the table above present in a loan.
• The higher risk of the mortgages taken out by first-time buyers is largely due to risk layering.
• Given the combination of little money down and slow amortization, these buyers will have very
little home equity for a number of years unless their house appreciates substantially.
• The mortgages taken out by repeat buyers are less risky along two dimensions in particular:
– a much smaller share had a CLTV of 95 percent or higher and
– a smaller share had a credit score below 660.
• Bottom line: the supply of mortgage credit to first-time buyers is not tight.
192
Rising Prices Have Disparate Effects on Buyers
Repeat buyers profiting from higher prices have managed to lower their CLTVs, while
FTBs have to stretch further. Recently we are also seeing greater separation in DTIs.
Chart Title
Average CLTV Chart
AverageTitle
DTI
96 40
First-time
94 buyers First-time
39 buyers
92
38
90
37
88
Repeat buyers
36
86
Repeat buyers 35
84
82 34
Feb-13 Oct-13 Jun-14 Feb-15 Oct-15 Jun-16 Feb-17 Oct-17 Jun-18 Feb-13 Oct-13 Jun-14 Feb-15 Oct-15 Jun-16 Feb-17 Oct-17 Jun-18
193
Agency-Specific First-Time Buyer Mortgage
Share Indices
Share varies widely across agencies. FHA and RHS are at the high end with a share
of around 83 percent, while Freddie Mac is at the low end with a share around 45
percent.
90% 90%
RHS
FHA
80% 80%
70% 70%
60% 60%
VA
50% 50%
Fannie
Freddie
40% 40%
30% 30%
Note: Calculated as a share of primary owner-occupied home purchase mortgages. RHS is Rural Housing Service. FHA share is taken directly from FHA’s
monthly production report, due to concerns about the accuracy of the first-time buyer classification in the NMRI dataset.
Source: AEI Center on Housing Markets and Finance, www.AEI.org/housing and FHA. 194
DTI Distributions, Agency FTB Purchase Loans*
income gains. The share of DTIs below 34% has declined, offset by a greater
share of DTIs above 40%. While bullish for home prices in the near term, this
presents long-term sustainability problems for both homeowners and the FHA.
6%
FTB DTI Distribution for Entire U.S.
5%
4%
3% February 2013
2% GSE August 2018
1%
FHA August 2018
0%
<20 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58
California shows how the shift could intensify as affordability worsens.

10%
FTB DTI Distribution, August 2018
8%
6%
4%
U.S. excluding CA
2% GSE California
FHA CA
0%
<20 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 >57
*Data pertain to all first-time buyer agency purchase loans for primary owner-occupied properties. 195
The Effect of FHA Mortgage Insurance Premium Cut
• FHA’s Jan. 2015 MIP failed to live up to its billing because it was undertaken during a
seller’s market. FHA’s recently announced (and since suspended) MIP cut, during an
even stronger seller’s market, likely would have had a similar outcome.
• Our research addresses two questions:
• Question 1: How did FHA borrowers use the 6% in additional buying power?
– Have analyzed this question with data from ATTOM Data Solutions, which allowed a robust
comparison of FHA and Conventional buyers
– Because prices rose by 3% for FHA financed homes vis-à-vis conventionally financed, borrowers
only saved half of the MIP cut
– The other half was capitalized into higher prices:
o The median price paid by ALL FHA borrowers amounted to $1,300 more for the exact same
house
o The rest (around $3,600) was used to go up-market, as FHA buyers opted to purchase larger
or more expensively appointed homes or opted for more expensive neighborhoods
• Question 2: How accurate was FHA’s prediction that the cut would spur 250,000 first-time
buyer (FTB) home purchases over the coming 3 years (≈ 83,000/year)?
– FHA’s first-time buyer volume increased about 180,000 in 1st year after MIP cut. Using the NMRI
data, we estimate that roughly:
o 35,000 (20%) went to new entrant FTB brought in by the MIP cut, only 42% of projection
o 85,000 of these loans (nearly half) were poached from the other Agencies
o 60,000 (33%) represented market trend growth unrelated to the MIP cut
– Upshot: FHA fell far short of goal despite big rise in [largely poached] total FTB volume
196
Income and Debt Growth by Income Group: FHA
Purchase Loans
Over the last 6 years FHA-insured home buyers have seen a growth in housing debt
that has greatly outpaced income growth. Lower income borrowers have had the
largest increase in debt burden relative to incomes (+14 ppts. differential).
Cumulative change, 2013:Q1 – 2017:Q4

40%
Income Growth
35%
32%
Housing Debt Payment Growth
30%
25%
22%
20% 18%
15% 14%
11%
10%
4%
5%
0%
25th Percentile 50th Percentile / Median 75th Percentile
Household Income
Note: Data are for largest 73 metros.

197
House Price Appreciation (HPA) by Price Tier
House price appreciation (HPA) has been greater for entry-level homes (low and low-
medium price tiers) than for move-up homes (medium-high and high price tiers). Since
October 2012, house prices in the low tier have risen 49% but only 29% in the high tier.
This trend of divergent growth rates is continuing and even accelerating.
50% 50%
Cumulative HPA (Oct. 2012 - Jan. 2019): by Price Tier
40% 40%
Low
Low-Med
30%
Med-High 30%
High
20% 20%
10% 10%
0% 0%
Oct-12 Apr-13 Oct-13 Apr-14 Oct-14 Apr-15 Oct-15 Apr-16 Oct-16 Apr-17 Oct-17 Apr-18 Oct-18
Source: AEI, Center on Housing Markets and Finance www.AEI.org/housing.

198
House Price Appreciation (HPA) by Price Tier: 73 Metros
HPA remained strongly bifurcated. In January 2019, house prices in the low and low-medium
price tiers appreciated at a faster pace than in December, with the biggest rebound in prices
coming in the low price tier, where access to credit is most prevalent. Compared to a year ago,
house prices in the low price tier appreciated 5.3% and 4.0% in the low-medium tier, while
house prices in the medium-high tier appreciated 3.1% and only 1.5% in the high tier.
Low Low-Med Med-High High

12% 12% 12% 12%
10% 10% 10% 10%
8% 8% 8% 8%
5.3% 5.2%
6% 6% 6% 6%
5.0%
5.7%
4% 4% 4% 4%
4.0% 3.7%
2% 2% 2% 3.1% 2%
1.5%
0% 0% 0% 0%
Oct-13
Oct-14
Oct-15
Oct-16
Oct-17
Oct-18
Oct-13
Oct-14
Oct-15
Oct-16
Oct-17
Oct-18
Oct-13
Oct-14
Oct-15
Oct-16
Oct-17
Oct-18
Oct-13
Oct-14
Oct-15
Oct-16
Oct-17
Oct-18
Red markers show November HPA in each year.
Note: Data for October 2018 to January 2019 are preliminary. Price tiers are set at the metro level and are defined as follows: Low: all sales at or below the 40th
percentile of FHA sales prices; Low-Medium: all sales at or below the 80th percentile of FHA sales prices; Medium-High: all sales at or below the 125% of the GSE loan
limit; and High: Rest. HPAs are smoothed around the times of FHFA loan limit changes.
Supply-Demand Imbalance in the Market Is Driving Prices Up
The supply-demand imbalance persists. The NAR’s not-seasonally adjusted months (mo.)
inventory in January, which is traditionally the month with the greatest inventory and
lowest sales, stood at 5.6 mo., up 0.7 mo. from a year ago. While this metric has started
to increase over the past 5 mo., it is still averaging below 6 mo., the demarcation between
a buyer’s and seller’s market, and it will fall back with the beginning of the spring buying
season. Thus, it is too soon to project a return of a buyer’s market. Instead, we expect
the seller’s market to modestly strengthen. This means further credit easing will
continue to be capitalized into higher home prices. According to the FHFA, not-seasonally
adjusted home prices rose 5.8% in November year-over-year, down from 6.8% a year ago.
The chart below shows the strong inverse relationship between supply and prices.
16 National Month’s Inventory & Changes in Nominal House Prices* -40%
14 -32%
Months Supply* (left axis)
Y-o-Y Price Change (scale inverted)

12
Length of current seller’s -24%
FHFA House Price Index, smoothed** market: 77 months
(right axis)
Month's Supply
10 -16%
Buyer's Market, Prices Falling Dashed line: Price

8 -8%
equilibrium point
6 0%
4 8%
2 Seller's Market, Prices Rising 16%
0 24%
Jan-99 Jan-01 Jan-03 Jan-05 Jan-07 Jan-09 Jan-11 Jan-13 Jan-15 Jan-17 Jan-19
* National Association of Realtors (NAR) “Number of homes available for sale (NSA) divided by NAR’s “Existing Homes Sales (NSA)”. The NAR defines a seller’s market to exist when
the inventory of existing homes for sale would be exhausted in six months or less at the current sales pace. Conversely, a buyer’s market exists when the inventory of existing homes
for sale exceeds six months at the current sales pace. (http://www.realtor.org/news-releases/2013/04/march-existing-home-sales-slip-due-to-limited-inventory-prices-maintain-uptrend).
** FHFA Monthly Purchase-Only Not Seasonally Adjusted house price index. The series is a 6 month trailing average.
Source: National Association of Realtors, FHFA, and AEI, Center on Housing Markets and Finance, www.AEI.org/housing. 200
Affordability Worsens in a Seller’s Market
Nominal Price-to-Income Ratio* has retraced 53% of the drop from the 2006 peak to the
2012 trough. Combination of a continued highly accommodative monetary policy and
easier lending promotes further capital flows into real estate, increasing the potential for
economic damage as highly leveraged lending fuels a cyclically volatile housing sector.
Nominal Price-to-Income Ratio, through 2018:Q4*
Predominantly a buyer's market Entirely a buyer's market
Entirely a seller's market Predominantly a seller's market
4.4 4.4
4.2 4.2
4.0 4.0
3.8 3.8
3.6 3.6
GSE affordable housing goals take effect for
3.4 CY 1993 as mandated by the Housing 3.4
3.2 3.2
3.0 3.0
2.8 2012 to date: easing loan 2.8
2.6 standards, very loose Fed 2.6
1993-2006: period of credit easing policy, and historically
2.4 and generally falling mortgage rates low mortgage rates 2.4
2.2 2.2
1975:Q1 1979:Q2 1983:Q3 1987:Q4 1992:Q1 1996:Q2 2000:Q3 2004:Q4 2009:Q1 2013:Q2 2017:Q3
* Calculated as median house price divided by median household income.
Source: Zillow.
* Calculated as median house price divided by median household income.
Note: The National Association of Realtors (NAR) defines a seller’s market as inventory that is less than or equal to 6 months of sales. NAR data pertain to existing homes; not
available before June 1982. Data from the Census Bureau for new home inventories used before June 1982.
Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing, Zillow, Census Bureau, and the NAR.
201
Fannie vs. Freddie Risk Index, GSE Purchase Loans
Fannie Mae’s risk index continues to outpace Freddie Mac’s. Fannie’s purchase MRI in
Nov. 2018 was 1.5 ppts (or 22%) higher than Freddie’s. Interestingly, this risk pick-up
has not translated into any meaningful market share gains for Fannie. While its share
has been volatile, it has averaged around 58-61% for each November since 2013.
Mortgage Risk Index Fannie Share of GSEs

Stressed default rate GSE Market Share
9% 75%
Rough contribution to higher MRI
over last 2 years from higher… Fannie
8%
Credit 70%
Scores CLTVs DTIs
Freddie -24% 36% 88%
Fannie 20% 56% 24%
7% Freddie 65%
6% 60%
5% 55%
4% 50%

202
History Repeats Itself: the “Quiet” Battle for Subprime (High Risk
>95 CLTV Purchase Loans) among Fannie, Freddie & FHA
As this segment of the GSE’s business has grown, Fannie has been holding the clear
advantage over Freddie with a market share of 75-95% (Fannie’s share of all GSE
purchase business is currently less than 60%). Interestingly, with Fannie in the driver
seat, it has been charging higher loan rates on a risk-adjusted basis. As the prior slide
shows, Fannie is able to poach these loans from FHA, since FHA does not price for risk.
High Risk GSE Purchase loans Fannie-Freddie risk-adjusted note

with a CLTV > 95% rate spread (in bps.)
20,000 100% 35
18,000 90% 30
16,000 80%
25
14,000 70%
20
12,000 60%
15
10,000 # of Loans 50%
(left axis) 10
8,000 Fannie Share 40%
(right axis)
Fannie Share 5 Fannie w/
6,000 30%
(all purchase, right axis) higher rate
0
4,000 20%
Fannie w/
2,000 10% -5 lower rate
0 0% -10
Sep-13
Sep-12
Mar-13
Mar-14
Sep-14
Mar-15
Sep-15
Mar-16
Sep-16
Mar-17
Sep-17
Mar-18
Sep-18
Sep-12
Mar-13
Sep-13
Mar-14
Sep-14
Mar-15
Sep-15
Mar-16
Sep-16
Mar-17
Sep-17
Mar-18
Sep-18
Note: Data are for high risk primary owner-occupied home purchase loans with a CLTV > 95%. High risk are loans with a stressed default rate of 12% or greater.
Leverage Fueled Housing Demand Pauses Due to Higher Rates
While still being up 25 percent from 5 years ago, purchase volume in November 2018
declined 5.1 percent from a year earlier. First-time buyer volume was down 3.6 percent,
while repeat buyer volume was down 7.0%. Greater access to credit is allowing first-
time buyers to offset higher mortgage rates and higher house prices, while repeat
buyers, with less access to credit, are electing to drop out of the
market in larger numbers.
Purchase loans
360,000 360,000
Composite
Red markers show November count in each year.
310,000 310,000
260,000 260,000
210,000 210,000
First-time
buyers
160,000 160,000
110,000 Repeat 110,000

Buyers
60,000 60,000
Note: October 2018 count is a preliminary estimate. First-time buyer volume not available before February 2013.
204
Agency First-time and Repeat Buyer Mortgage Risk Indices
The November Agency FBMRI stood at 17.0%, up 0.6 ppt from a year earlier. It is
also 7.4 ppts. higher than the mortgage risk index for repeat buyers, which is 0.4
ppt. wider than the gap a year earlier. Given supply constraints and absent a
triggering event, we expect house prices and leverage to continue to rise for FTBs.
21% 21%
Historical Projection First-time buyer (FBMRI)

19% 19%
17% 17%
15% 15%
13% 13%
11% 11%
Repeat buyer MRI
9% 9%
7% 7%
Feb-13 Feb-14 Feb-15 Feb-16 Feb-17 Feb-18 Feb-19 Feb-20 Feb-21 Feb-22
Note: Calculated for primary owner-occupied home purchase mortgages.

205
Origination Shares and MRIs by Seller Lender Type,
GSE Purchase Loans
The shift in GSE market share from large banks to nonbanks appears to be
resuming. The large-bank share has dropped to a series’ low of around 25% in
November 2018, down from 34% a year ago. Other banks are also losing share.
Banks (both large and other) have a lower GSE risk profile than nonbanks.
Origination Shares* Mortgage Risk Indexes
70% 8.0%
Composite shown by blue line

60% 7.5%
Nonbanks
Nonbanks
50% 7.0%
40% 6.5%
Large banks
Other banks
30% 6.0%
20% 5.5%
Other banks
Large banks
10% 5.0%
0% 4.5%
Note: Data for most recent months may understate large-bank share by perhaps 2 percentage points, as large banks are slower to move recent originations to
the guarantee agencies for securitization and our market shares are based on securitized loans. MRIs for credit unions and state housing agencies are not shown
because of low loan volumes.
*Origination shares do not show shares for State Housing Finance Agencies or Credit Unions which account for about 5% of the GSE Purchase market. 206
Origination Shares and MRIs by Issuer Lender Type,
FHA Purchase Loans
The dramatic market shift from large banks to nonbanks for FHA loans appears to be
continuing. In November, the large bank share dropped below 10%, a new series’
low. Migration to nonbanks has boosted overall risk levels, as nonbanks are willing
to originate riskier FHA loans than large banks.
Origination Shares* Mortgage Risk Indexes
90% 30%
Nonbanks Nonbanks
80% Composite shown by blue line
28%
70%
60% 26%
Other banks
50%
24%
40%
30% 22% Large banks
20%
Large banks
20%
10%
Other banks
0% 18%
Note: Data for most recent months may understate large-bank share by perhaps 2 percentage points, as large banks are slower to move recent originations to
the guarantee agencies for securitization and our market shares are based on securitized loans. MRIs for credit unions and state housing agencies are not shown
because of low loan volumes.
*Origination shares do not show shares for State Housing Finance Agencies and Credit Unions which account for about 4% of the FHA Purchase market.
207
Mortgage Risk Indices by Lender Type, Refi Loans
The same share shift from banks to nonbanks applies to agency refi loans. The large
bank share dropped below 20% for the first time in the series. Similar to purchase
loans, the gap in riskiness between banks and nonbanks has also widened over time.
Refi Origination Shares Refi Mortgage Risk Index

70% 20%
Nonbanks
60% 18%
Nonbanks
16%
50%
14%
40%
12%
30% Composite
Large banks 10%
Large banks
20%
8%
Other banks
10% Other banks
6%
0% 4%
Note: Composite includes credit unions and state housing agencies, which are not shown separately.

OMS Housing-Pinto PDF

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

OMS Housing-Pinto PDF

Încărcat de

Drepturi de autor:

Formate disponibile

AEI Housing Market Indicators (HMI)

Edward Pinto (Edward.Pinto@AEI.org) and

AEI Center on Housing Markets and Finance

20% RHS: -0.5 ppt, from 19.4 to 18.9%* 20%

Freddie: +1.7 ppts, from 4.9% to 6.7%*

*Change from December 2012 to December 2018.

160 GSE affordable housing goals take effect for 160

YoY Price Change (scale inverted)

Dashed line: Price

90% Quiescent period

Months Supply: 2017:Q4

Months Supply: 2017:Q4

Low Price Tier High Price Tier

Months’ Supply (2018:Q4) : 271 CBSAs

Low Price Tier High Price Tier

Months’ Supply (2018:Q4) : Top 50 CBSAs

Low Price Tier High Price Tier

House Price Appreciation (2012:Q4 vs. 2018: Q4):

10% 25% 74%

Sales Transactions in millions

5.0 New construction share of all sales: 5.0

Other financed sales

Note: Data are for institutionally financed sales only.

300,000 No-Cash-Out Refis 300,000

FHA VA Freddie Fannie

Red markers show December counts in each year.

15% Percentage points Change in Distribution, December 2016 to December 2018

Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing.

Red markers show ratio in December of each year

Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing.

Mean price Feb-13

(2.0% avg. annual growth)

Scroll over “Indexes & Indicators” and then

Latest data will go live at 10 am

• All briefings take place at 11 AM ET.

Please note that this month’s appendix

• Principles behind the NMRI

• Basics of index construction

• Advantages of the NMRI

• What does an increasing or decreasing NMRI mean?

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%

• Takeaway: Huge spread of default rates across risk buckets

• Basics of index construction

• Advantages of AEI’s HPA Index:

• Data for the HPA index

• NMRI and other risk indices published for:

• Financial crisis largely stemmed from a failure to understand

3. Dual Underestimation Principle: Never underestimate the government’s

• We define low-risk loans as those with a stressed default rate of less

• Low-risk definition calibrated from two sources

– FHA underwriting standards over 1935-55

– Both yield an average stressed default rate of ≈ 3%

• This is consistent with a maximum stressed default rate of ≈ 6% on

20% FHA 20%

Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing.

Number of loans with CLTV > 95% Text not updated

Note: Excludes loans made by VA and RHS.

Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing.

Source: AEI, Center on Housing Markets and Finance, www.AEI.org/housing.