1. The Devolution of Appraisal and
Underwriting Theory and Practice
Edward Pinto
Codirector and Chief Risk Officer
AEI International Center on Housing Risk
Edward.Pinto@AEI.org
HousingRisk.org
Third International Conference on Housing Risk: New Risk
Measures and their Applications
September 17, 2014
The views expressed are those of the author alone and do not necessarily represent those of the
American Enterprise Institute.
1

2. INTRODUCTION
2

3. ‘Mortgage’ has become another word for
‘Mortgage’ has become another word for trouble
‘trouble’
• Long term home price appreciation rates are
modest and averages mask volatility
– No diversification when one’s wealth is in a single asset
– “The housing market is a ‘crapshoot’”—Professor Karl
Case1
• Excessive leverage promotes volatility
1 http://money.cnn.com/2014/07/07/investing/housing-market-case/ July 7, 2014 3

4. House Price Volatility, 50 Largest Metro Areas
House prices are highly and serially volatile in many metro areas, less so in others. But
even in the more stable metros, there can be substantial variation across neighborhoods
160%
140%
120%
100%
80%
60%
40%
20%
0%
-20%
-40%
-60%
-80%
160%
140%
120%
100%
80%
60%
40%
20%
0%
-20%
-40%
-60%
-80%
California metros
Other volatile metros
More stable metros
1984 1987 1990 1993 1996 1999 2002 2005 2008 2011
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 “more stable” group.
Source: AEI International Center on Housing Risk, www.HousingRisk.org, using data from Zillow.com 4

5. 28 cities: nominal annual percentage increase in home
5
prices from Apr. 1996-May 2012 (source: Zillow)

6. City of Atlanta: cumulative price change for lowest price
tier and constituent zips from Apr. 1996-May 2012*
6
*Source: Zillow

7. Volatility at the Level of the Individual Home
• Averages are deceiving since one is buying a single home, not the index
average for the U.S., the MSA, the zip, or the neighborhood.
– To test the impact of home price volatility on wealth building at the house level, 2 different
pro forma loan transactions 1 were simulated using over 112,000 unique properties in
Prince George’s County, MD.2
– Individual home price index data for 1997-20133 was used to create 11 cohort time
periods4resulting in over 1.2 million 30-year fixed rate scenarios and over 1.2 million 15-
year fixed rate Wealth Building Home Loan scenarios
– The wealth building capacity of the two loan transactions was evaluated as follows:
» For each scenario, the loan was amortized according to its terms, the property
experienced the price gain or loss over 84 months as indicated by the index, the
owner was assumed to have sold the home at the value indicated by the index at the
end of the 84 month period, and the owner was assumed to have incurred a selling
transaction cost of 10%.
» The set of properties was divided into 3 price ties: low, medium, and high. The wealth
building capacity of the two loan transactions was evaluated using these tiers.
1Loan transaction 1: 30-year, fixed rate loan with 5% down, 1.75% upfront FHA mortgage insurance premium which is
financed, and an assigned interest rate based on Freddie Mac’s Primary Mortgage Market Survey.
Loan transaction 2: 15-year, fixed rate loan with 0% down (down payment repurposed to fund a 1.25% permanent rate
buy down) and an assigned interest rate based on Freddie Mac’s Mortgage Rates Survey (net of buy down).
2Prince George’s median household income is approximately 80% of that for the Washington, DC area.
3Source: Weiss Residential. Since Index is quarterly, values interpolated as necessary.
4For example, cohort time period 1 started in January 1997 and ended 84 months later. Cohort time period 2 started in
January 1998 and ended 84 months later. This process was continued for a total of 11 cohorts.
7

8. Prince George’s County: Volatility and Loan Type
Impact Wealth Building at the Individual Home Level
Loan/Tier # of
transactions
Average
cumulative
$ gain1
Average
cumulative
% gain
% of
transactions
with a loss
Median $ loss
for
transactions
with a loss
Worst cohort: 2006
median cumulative
% gain
30-
year2/Low
413,732 $19,756 29% 38% -$54,945 -33%
15-year
WBHL3/
Low
413,732 $54,604 54% 26% -$13,353 -8%
30-year2/
High
413,743 $53,541 32% 36% -$124,232 -31%
15-year
WBHL3/
High
413,743 $136,124 57% 21% -$29,835 -6%
1Average gains calculated for 11 home purchase year cohorts (1997-2007). Each cohort assumed a sale at
the end of an 84 month holding period. Net gain equaled house price appreciation over period (net of 10%
sales costs) plus scheduled loan amortization minus initial 5% investment.
230-year, fixed rate loan with 5% down and an assigned interest rate based on Freddie Mac’s Mortgage
Rates Survey. House price change based on Weiss Residential’s individual house index.
315-year, fixed rate loan with 0% down (down payment repurposed to fund a 1.25% permanent rate buy
down) and an assigned interest rate based on Freddie Mac’s Mortgage Rates Survey (net of buy down). 8

9. Prince George’s County: Leveraged 30-Year
Loans Perform Poorly Under Stress
• The 100% LTV WBHL protects against negative equity compared to a
95% 30-year loan even with extreme home price volatility of the
recent boom/bust
9

10. Trouble Occurs When there Is Too Much of
the Wrong Kind of Debt
• Long sustained growth in housing leverage is unhealthy
– Jobs create demand for housing, not housing creates demand for jobs.
– Mortgage debt to smooth housing consumption is positive economically.
– Debt to finance consumption in excess of income is destabilizing.
• Bids up existing assets and the land they sit on, creating temporary
wealth effect.
• Crowds out capital investment financing real demand and job growth.
• Borrowers become over extended and susceptible to economic shocks.
• Debt overhang effect depresses any post-shock recovery.
– Today nominal and intrinsic housing debt both remain historically high.
The above is largely drawn from Adair Turner’s Too Much of the Wrong Sort of Capital Flow, January 13, 2014
10

11. Source: “Securitization in the 1920’s”, William Goetzmann , 2009, Securitization in the 1920's - Yale University
11
Trouble Occurs When Mortgage Debt
Grows Faster than Home Value

12. Trouble Occurs When Loan to Value Grows
Faster than Fundamentals
12

13. Leverage Leverage takes takes many forms
forms
• Three types of asset leverage:
– Reduced down payment on purchase loans: increases the asset price of the home financeable with
the same level of savings. This took the form of higher loan-to-values (LTVs) on first mortgage and
higher combined LTVs (CLTVs) on combination first and second mortgages.
– Longer loan term or use of interest only (IO) period: keeps asset leverage elevated due to the reduced
earned equity buildup from amortization during a loan’s early years
– Higher LTVs on rate and term and cash-out refinances: allows borrowers to take advantage of higher
home prices that result from higher leverage and the inherent weaknesses of the appraisal process,
which are enhanced due to the lack of an actual sales transaction..
• Five types of income leverage:
– Increase in total debt-to-income (DTI) ratio: increases the asset price of the home financeable with the
same level of income.
– Longer loan term or use of IO period: slower loan amortization reduces the monthly debt service
payment, thereby increasing the asset price of the home financeable with the same level of income.
– Use of adjustable rate mortgages (ARMs) or hybrid ARMs: these loans tended to start out at a low rate
and increase over time, thereby increasing the asset price of the home financeable with the same level
of income. Negatively amortizing ARMs (Pay Option ARMs) allowed the increase in monthly payments
to be added to the loan balance, resulting in negative amortization.
– Expand definition of eligible income to include less certain types: raises income thereby increasing the
asset price of the home financeable.
– Reduced documentation standards for income verification: “low doc” and “no doc” foster “liar loans”
creating phantom income which increases the asset price of the home financeable.
• One type of credit leverage:
– Reduce the level of acceptable credit score: increases the pool of eligible buyers. Credit risk increases
if borrower credit impairment is not offset by compensating with lower risk factors such increased
down payment or faster amortization. 13

14. Ignoring expenses adds to leverage
• With the exception of the VA , the FHA, Fannie and
Freddie underwriting approaches ignore many
expenses:
– DTIs are calculated based on pre-tax income and do not take into
account other normal living expenses such as income and payroll
taxes, food, clothing, home utilities, home repairs and
maintenance, auto and commute expenses, child care, retirement
savings, etc.
– Cost to commute: The value of improved land is determined by its
utility. A core feature is proximity to jobs and more particularly the
household member(s)’ jobs.
– Homes situated further out from jobs)generally sell for less than
homes closer in, largely due to differences in the cost of land.
• DTI limits generally serve to set the home buyer’s
upper price limit
• Ignoring these expenses increases income leverage, which was
exacerbated by increasing formal DTIs 14

15. EVOLUTION OF APPRAISAL AND
LENDING THEORY AND PRACTICE
15

16. 1890 to 1920s: Ely Advances Early US Research on
Land Economics and Appraisal Theory and Practice
• University of Wisconsin Professor Richard T. Ely ("Under all,
the land”) recognizes land use as the product of economics,
institutional forces, and physical constraints (1854-1943).
– Studied at the University of Heidelberg (1878-1879) with, Karl
Knies (1821-1898), a leading historical economist , who:
• Authors Money and Credit (1873).
• Advances concepts of “value in use” and “value in exchange (price)”
– 1892-1925: Professor of political economy and director of the
School of Economics, Political Science, and History, University of
Wisconsin
• “Father of Land Economics” and real estate studies.
• 1920: Establishes the Institute for Research in Land Economics and
Public Utilities 16

17. 1903: Hurd - Principles of City Land Values
• 1903: The Principles of City Land Values (Richard M. Hurd, 1865-
1941)--considered first United States’ treatise on city (non-farm)
land values.
– While intrinsic value is correctly derived by capitalizing ground rent, exchange value
may differ widely from it.
– Value in urban land, as in farm land, results from economic or ground rent
capitalized.
– Since value depends on economic rent, and rent on location, and location on
convenience, and convenience on nearness, the intermediate steps may be
eliminated and say that value depends on nearness.
– In cities, economic rent is based on the superiority of location only, the sole function
of urban land being to furnish area on which to erect buildings.
– Land prices on the outskirts are lower as area increases as the square of the distance
from any given point.
– If a new utility does not arise, exchange prices may advance and recede, while
intrinsic values do not change.
– If a new utility arises, both exchange prices and intrinsic values will alter their
levels.
– General financial and economic conditions enter so largely into exchange values, that
values are at times not based on income, or supply and demand, but represent
simply a condition of the public mind.
17

18. 1920s: Ely and Colleagues Continue Working on
Land Economics and Appraisal Theory and Practice
• Books by individuals that Ely collaborates with, trains,
or are within his circle of influence:
– 1923: Ernest Fisher (1893-1981), student of Ely and member of
Ely’s Institute, writes Principles of Real Estate Practice (edited
by Ely and published by the Institute).
– 1924: Frederick Babcock (1897-1983) writes The Appraisal of
Real Estate (1st generalized appraisal book and part of Ely’s Land
Economics Series)
18

19. 1920’s: Lending and Appraisal Practice
• The concept of intrinsic value was largely forgotten, as lending
became based on exchange value only..
• Property appraising largely the province of local real estate boards.
– Boards comprised of those realtors who also held themselves out as
appraisers.
– There were neither training nor certification requirements.
• Lending standards were loose.
– In 1927-28, first mortgages were available at 100% of exchange value.
– In 1927-28, second mortgages were available at 120% of exchange value.
– Commonplace for mortgage underwriting to be based on the mortgaged
collateral, ignoring income and credit.
– Fraudulent lending and appraisal practices were rampant.
– Mortgage backed bonds were relatively common-place (all failed in the
1930s). 19

20. 1930s: Ely’s Colleagues Lay Foundation for the FHA
and Sustainable Lending Practices
• Individuals Ely collaborates with, trains, or within circle of
influence:
– 1932: Frederick Babcock (1897-1983) writes The Valuation of Real Estate
(considered to be the most significant appraisal book since Hurd’s in 1903).
Ernest Fisher helped Babcock with this second book as both were at the
University of Michigan in the early 1930s.
– 1933: Homer Hoyt (1895-1984) receives PhD (U. of Chicago), publishes his
dissertation, 100 Years of Land Values in Chicago (influenced by Chicago-based
research done by Institute members). Writes The Structure and
Growth of Residential Neighborhoods in American Cities (1939) with Ernest
Fisher and Principles of Urban Real Estate with Arthur Weimar (1939).
– Early-1930s: Richard Ratcliff (1906-1980), student of Richard Ely and Ernest
Fisher. Writes Urban Land Economics (1949).
– 1934: Arthur Weimer (1909-1987) receives PhD (U. of Chicago). Writes
Principles of Urban Real Estate with Hoyt (1939).
– Fisher (Chief Economist), Babcock (Chief Underwriter), Hoyt (chief housing
economist), Ratcliff (economist), and Weimer (economist) are all at FHA
starting in 1934.
20

21. FHA and Sustainable Lending
• 1935: “‘Mortgage’ was just another word for trouble—an epitaph
on the tombstone of their aspirations for home ownership.”*
– Replaces loose and dangerous lending practices that had made foreclosures
commonplace with “a straight, broad highway to debt-free ownership.”*
• “[s]uccessful mortgage lending must be predicated upon a measurement of risk
factors in mortgage investment. Mortgage risk comes into existence in the
moment mortgage funds are disbursed to a borrower. The risk continues until
there has been a complete recapture of the money which has been lent. This risk
is greater in some loans than in others. It differs from time to time for each loan.
The best we can hope to do is establish a method by means of which to estimate
the degree of risk at the time the loan is submitted. If the measurement of risk
indicates hazards which are too great, the institution must necessarily refuse to
make the loan. In other cases it is highly important that the institution
determine not only that it is willing to make the loan but the intrinsic quality of
the loan as a portion of its mortgage portfolio.” Frederick M Babcock, FHA’s first
chief underwriter
– Sound lending practices include:
• A sizable down payments (a minimum of 20%) and a maximum 20-year term;
• Solid borrower credit histories and solid appraisals;
• Ability to pay (imposed on FHA by the1934 National Housing Act): proper
income documentation and sufficient income to make regular payments
(includes review of a borrower’s monthly expenses and residual income);
• Fully amortized loan with a ban on second mortgages.
* Federal Housing Administration, “How to Have the Home You Want,” 1936.
21

22. A Broad, Straight Highway to Debt-free Home
Ownership
• In 1935 FHA provided a “broad, straight highway to debt-free home ownership”
• The homeownership rate soared from 44% in 1940 to 62% in 1960
• The 30-year loan played a minimal role throughout 1940-1960
– Role limited to FHA and VA new construction in the latter part of the 1950’s
• The Debt-Free Highway enabled the Greatest Generation to burn their mortgages
*Source: John P. Herzog and James S. Earley, Home Mortgage Delinquency and Foreclosure.
22

23. FHA and Sustainable Lending
• FHA’s highway to debt-free ownership led to:
– An explosion in the homeownership rate
• From 43.6 percent in 1940 to 61.9 percent in 1960
– The virtual elimination of foreclosures
• Over its first 20 years, the FHA paid only 5,712 claims
out of 2.9 million insured mortgages, for a cumulative
claims rate of 0.2 percent.
• Claim loss severity was 9 percent of the original
insured mortgage balance, or a total of $3 million on
5,712 claims.*
*Thomas N. Herzog, A Brief History of Mortgage Finance with an Emphasis on Mortgage Insurance, Society of Actuaries, 2009,
www.soa.org/library/monographs/finance/housing-wealth/2009/september/mono-2009-mfi09-herzog-history-comments.pdf.
23

24. Basis for FHA’s Valuation Theory
• Babcock in his The Valuation of Real Estate (1932)
establishes the concept of warranted value.
– Fair market value would then be the price which a buyer were
warranted in paying in view of the potential utility of the property.
The fact that several hundred purchasers have been found who were
willing to buy certain undesirable subdivision lots at exorbitant prices
would in no way be presentable as evidence of market value.
– Value will be used to designate the concept in which the thoroughly
informed buyer is present and market price will be used to designate
the prices which properties actually do bring in the real estate
market.
• In 1934 Babcock becomes Chief Underwriter for FHA and
proceeds to implement his concept of warranted value.
24

25. Marked Similarities Between FHA’s Valuation for
Mortgage Loan Purposes (1930s) and Germany’s
Mortgage Lending Value (1990s)
• FHA Underwriting Manual and warranted value (1938)
– § 13 Methods of dwelling valuation—the character of value
• The word “value” refers to the price which a purchaser is warranted in
paying for a property for continued use or a long-term investment.
• The value to be estimated, therefore, is the probable price which typical
buyers are warranted in paying.
• This valuation is sometimes hypothetical in character, especially under
market conditions where abnormalities in price levels indicate the
presence of serious quantitative differentials the two value concepts
[warranted value and available market price].
• Marked differences between “available market prices” and “values” will
be evident under both boom and depression conditions of market.
• Attention is directed to the fact that speculative elements cannot be
considered as enhancing the security of residential loans. On the
contrary, such elements enhance the risk of loss to mortgagees who
permit them to creep into the valuations of properties upon which they
make loans.
25

26. Marked Similarities Between FHA’s Valuation for
Mortgage Loan Purposes (1930s) and Germany’s
Mortgage Lending Value (1990s)
• FHA Underwriting Manual and warranted value (1938)
– § 13 Methods of dwelling valuation—the character of value
• Value does not exist unless future benefits are in prospect. Its measure
is the present worth of expected benefits which may be realized only
upon the occurrence of future events.
• The first step in the basic valuation procedure, the study of future utility,
includes the selection of possible uses, the rejection of uses which are
obviously lower uses than others, and the determination of uses in
terms of alternative kinds of possible buyers and differing motives of
such buyers.
• No other definition is acceptable for mortgage loan purposes inasmuch
as one of the objectives of valuation in connection with mortgage
lending is to take into account dangerous aberrations of market price
levels. The observance of this precept tends to fix or set market prices
nearer to value.
26

27. Marked Similarities Between FHA’s Valuation for
Mortgage Loan Purposes (1930s) and Germany’s
Mortgage Lending Value (1990s)
• From Pfandbrief Act
– § 3 Principle of the determination of the mortgage lending value
(1) The value on which the lending is based (mortgage lending value) is the
value of the property which based on experience may throughout the life
of the lending be expected to be generated in the event of sale,
unattached by temporary, e.g. economically induced, fluctuations in
value on the relevant property market and excluding speculative
elements.
(2) To determine the mortgage lending value, the future marketability of
the property is to be taken as a basis within the scope of a prudent
valuation, by taking into account long-term sustainable aspects of the
property, the normal and local market conditions, the current use and
alternative appropriate uses of the property.
27

28. DEVOLUTION OF APPRAISAL AND
LENDING THEORY AND PRACTICE
28

29. 1930s to Today – Policy Pressures to Increase All
Forms of Leverage
• Late-1930s on: Congress raises FHA leverage limits
– FHA’s underwriting grid and valuation practices limit layering of risk for 20
years
• 1992 and on: Congress places Fannie and Freddie in competition
with FHA and private subprime
• The result is a ‘curious’ policy whereby low income home buyers
with volatile incomes are encouraged to buy homes in areas with
volatile prices using high leverage
• Policies based on a view that “[o]ne unique aspect of
homeownership is that it is one of the few leveraged
investments available to households with little wealth,
enabling homeowners with very little equity in their homes to
benefit from appreciation in the overall home value.”1
1Herbert and Belsky, 2008, The Homeownership Experience of Low-Income and Minority Households: A Review
and Synthesis of the Literature, Cityscape: A Journal of Policy Development and Research
• T
29

30. FHA abandons the ‘Debt-Free Highway’
• FHA’s underwriting grid and valuation practices successfully limit
layering of risk until the mid-1950s, but these are overwhelmed by
continual increases in FHA’s leverage limits by Congress
– “Until 1964 all loans offered for FHA insurance were subjected to an
underwriting "risk rating" based on a combination of mortgage,
property and borrower characteristics. The rating factors included the
maturity of the loan relative to the estimated economic life of the
residence, the loan-to-value ratio, locational and physical property
characteristics, mortgage payment and housing expense relationship
to estimated effective mortgagor income, and a credit rating of the
borrower. To be accepted for insurance, a loan was required to have a
‘rating pattern’ of at least 50 points out of a possible 100. Ratings
from 50 to 59 were considered "marginal," although acceptable.
Since 1964 no over-all rating pattern has been used and numerical
ratings have been dropped altogether. Now the underwriter must
rate the borrower, the property, and the location as ‘reject,’ ‘fair,’
‘good,’ or ‘excellent.’”
Source: John P. Herzog and James S. Earley, Home Mortgage Delinquency and Foreclosure, 1970 30

31. FHA mortgages becomes another word for
‘trouble’
Source: John P. Herzog and James S. Earley, Home Mortgage Delinquency and Foreclosure, 1970 31

32. FHA mortgages = ‘trouble’:
3.39 million foreclosures and 1 in 8 families
FHA’s WEIGHTED AVERAGE FORECLOSURE CLAIM RATE OF 12.8% FOR 1975-2013
| 32

33. 1930s to Today – Competing Valuation Theories
• FHA’s warranted value/German Mortgage Lending Value
• Exchange value/price , point-in-time value/price, or market
value/price.
– Over time this approach received support from the rational pricing
theory.
• The same asset must trade at the same price on all markets.
• Reconciliation of the three approaches: market, income, and
cost.
– These derived from classical economic theory based on the three
aspects of value.
33

34. 1992: CONGRESS MANDATES
FANNIE AND FREDDIE TO COMPETE
WITH FHA AND SUBPRIME
34

35. FHA and Fannie’s growing percentages of low
down payment loans
Sources: FHA 2009 Actuarial Report and HUD. Fannie percentages for 1994-1996 are estimated based on the fact that it first started acquiring 97%
LTV loans in 1994 and the percentage of such acquisitions in 1997 was 3.3%. Combined LTV percentages for 2004-2007 are based on Fannie’s
disclosure in its 2007 10-K that 9.9% of its credit book (home purchase and refinance loans) had an LTV>90% (the average LTV of these loans was
97.2%), 15% of its credit portfolio had an LTV or combined LTV >90%. This increased Fannie’s exposure in loans with downpayments of 5% or less by
50%. A common combination loan was an 80% first and a 20% second, yielding a combined LTV (CLTV) of 100%. Fannie purchased the first. 35

36. • The first panel on the next page demonstrates
that the percentage of verified DTIs >42%
increased dramatically from the late-
1980s/early-1990s when it was effectively 0% to
43% for 2007.
• The second panel presents the data in a slightly
different format. It demonstrates that the DTI
percentage at the 75th percentile increased from
a 36% DTI in the late-1980s/early-1990s to a
49% DTI in 2007.
36
DTIs in excess of traditional levels become
commonplace for the GSEs

37. GSEs ratchet up borrower total debt capacity
GSEs: % with DTI =>42%
50%
48%
46%
44%
42%
40%
38%
36%
GSEs: DTI % @75th Percentile
Sources:
1988-1992: debt-to-income (DTI) distribution interpolated and extrapolated from a random sample review of Fannie
Mae’s single-family acquisitions for the period October 1988-January 1992, dated March 10,1992. For this data set, the
maximum DTI grouping is “>38%” which constituted 13.5% of sampled loans. Document contained in the author’s files.
The “zero” or nil incidence at >42% DTI is an estimate based the data. This random sample is also the data source for
the serious delinquency (90-days or more and in foreclosure) data for 1988-1992 shown on the next two charts. Given
the observation date of early-1992, these data represent seriously delinquent loans with an average of 2 years
seasoning. While these delinquency rates are not directly comparable to the rates for 1997-2009, the relative
relationship among DTI buckets is valid .
1997-2009: DTI distributions derived using interpolation and extrapolation of data contained in the Consumer Financial
Protection Bureau’s request for further comment on Ability-to-Repay mortgage rule dated May 31, 2012. Dataset
consists of fully documented income loans that are fully amortizing with a loan term <=30 years. For this data set, the
maximum DTI grouping is “=>46%” which constituted 31% of sampled loans.
http://files.consumerfinance.gov/f/201205_cfpb_Ability_to_Repay.pdf This is also the data source for the delinquency
data for 1997-2009 shown on the next two charts. The observation date is 2012 and represents ever 60 days or more
delinquency rates.
37
50%
45%
40%
35%
30%
25%
20%
15%
10%
5%
0%
1988-1992
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
% >=42%
34%
1988-1992
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
DTI % @75th
percentile

38. • The graph on the next page demonstrates in the late
1980s through 1998, higher DTIs performed better than
lower DTIs.*
– This was due to the fact that a very high proportion of DTIs
were within well defined levels (28/36, 33/38, etc.).
– DTIs above these levels were the exception and generally
required strong compensating factors.
– This result may have led to a false conclusion that DTIs could
be increased without a substantial increase in risk.
– This turned out not to be the case.
– As DTI increased, so did delinquency rates.
• By 1999 loans with higher DTIs were acting differently.
• By 2003 the difference was quite substantial.
*Source: CFPB and document in author’s file
38
As acceptable DTI percentages grew, loan
performance worsened

39. 39
As acceptable DTI percentages grew, loan
0.54%
Graph 4: Delinquency Rate by DTI, 1988 – 2003
0.69% 0.69%
1.13%
0.80% 0.80%
8.00%
7.00%
6.00%
5.00%
4.00%
3.00%
2.00%
1.00%
0.00%
1997
DTI < 32 32 ≤ DTI < 3434 ≤ DTI < 3636 ≤ DTI < 3838 ≤ DTI < 4040 ≤ DTI < 4242 ≤ DTI < 4444 ≤ DTI < 46 46 ≤ DTI
2000-2002
1988-
1992*
1997
1998
1999
2000
2003
1999
1998
1988-1992
performance worsened

40. • The graph on the next page demonstrates that this
trend accelerated in 2004-2007.
– For the 2007 cohort, loans with DTIs>46% experienced
over 3 times the default rate of loans with DTIs <32%.
– For the 2007 cohort 31% and 43% of loans had a DTI>46%
and 42% respectively.
– This was dramatic change from earlier years when the
percentage of DTIs >42% was 15% or less.
• Clearly DTI mattered when it comes to a borrower
having a reasonable ability to repay.
40
As acceptable DTI percentages grew, loan
performance deteriorated markedly

41. 41
As acceptable DTI percentages grew, loan
performance deteriorated markedly
0.54% 0.69% 0.69%
Delinquency Rate by DTI
1.13% 0.80% 0.80%
35.00%
30.00%
25.00%
20.00%
15.00%
10.00%
5.00%
0.00%
DTI < 32 32 ≤ DTI < 34 34 ≤ DTI < 36 36 ≤ DTI < 38 38 ≤ DTI < 40 40 ≤ DTI < 42 42 ≤ DTI < 44 44 ≤ DTI < 46 46 ≤ DTI
1988-1992*
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2007
2006
2005
2004
2003
1997, 1999-
1998 2002
1988-1991

42. Growth in Alt-A - including low and no doc loans
• Myth: the GSEs’ Alt-A dollar volume was relatively minor relative to non-agency
Alt-A and came late in the cycle.
• Fact: the GSEs largely dominated the Alt-A market from 2002 onward
(note: limited or no GSE data is available before 2002, but they are known
to have been active pre-2002).
42
20.0%
18.0%
16.0%
14.0%
12.0%
10.0%
8.0%
6.0%
4.0%
2.0%
0.0%
Self-Denominated non-agency Alt-A
annual $ volume (net of Fannie/Freddie
PMBS Alt-A purchases)/Annual total
origination $ volume (source: Inside
Mortgage Finance).
Self-Denominated & SEC disclosed
Fannie/Freddie Alt-A Annual $ Volume
(includes acquisition of PMBS)/Annual
total origination $ volume (data missing
before 2001 and partial data for 2001).
PMBS never accounted for >18% (in
2005) of GSE Alt-A acquisitions

43. • Rampant inflation of income on low doc/no doc loans
compounded the increase in acceptable debt-to-income ratios
Source: Market Pulse, August 2011, http://www.corelogic.com/about-us/researchtrends/the-marketpulse.aspx
43
Result: a doubling down on income leveraging

44. Lower standards for credit histories expanded the
credit pool, further driving demand
• FICO distribution for all new mortgages (not just Fannie/Freddie)
100.0%
90.0%
80.0%
70.0%
60.0%
50.0%
40.0%
30.0%
20.0%
10.0%
0.0%
40.0%
35.0%
30.0%
25.0%
20.0%
15.0%
10.0%
5.0%
0.0%
1987 (left axis)
1992 (left axis)
2005 (left axis)
Freddie Mac: %
unacceptable credit
(1997) - right axis
• In 1987 and 1992, the percentage of new mortgages with a <660 FICO (subprime) was
13.3% and 14.5% respectively. By 2005 it was 32.7%.
• In 1997 borrowers in FICO bands <579, 580-619 and 620-659 (subprime bands which
also were goal-rich) had unacceptable credit 97.7%, 70.7% and 42.3% of the time,
respectively. The rate was only 13.2% for the 660-699 FICO band (low end of prime). 44

45. Impact of ignored expenses on leverage
• Consider a two-commuter household considering two locations:
– Location 1 has two 40-mile round trip commutes for a total of 80 miles per day or
16,000 miles based on 200 commute days, adding up to $8000 per year at 50 cents
per mile. This is 12% of pre-tax income based on a median first-time buyer income
of $67,400. The maximum priced home this household would be able to purchase
at a 28% housing debt ratio is $203,000. This household has a 41% total DTI
yielding a 53% combined DTI and commuting expense ratio.1
– Location 2 results in two 10-mile round trip commutes for a total of 20 miles per
day or 4,000 miles based on 200 commute days, adding up to $2000 per year at 50
cents per mile. This is 3% of pre-tax income based on the same median first-time
buyer income as above. As above the maximum priced home this household would
be able to purchase is $203,000. This household also has a 41% total DTI yielding a
44% combined DTI and commuting expense ratio.1 Using the same 53% combined
DTI and commuting expense ratio as for Location 1, this household would qualify
for a $269,000 home, 17% higher than when commuting costs are ignored.
– This underwriting flaw was compounded by rising fuel costs in the early-2000s, as
the cost of gasoline rose from $1.43/gallon in 2004 to $4.10/ gallon in 2008. This
impacted default rates in areas like Riverside-San Bernardino, CA.
• The traditional DTI methodology used by all extant underwriting approaches other
than the VA’s also ignore utility costs and the expected costs of maintenance and
repairs. Ignoring the expense variances that occur from home to home is yet another
way to increase leverage.
1Based on a 30 year fixed rate loan at 6% and a 28% housing debt ratio and a 41% DTI (FHA’s current averages). 45

46. Combined LTV and FICO Are Heavily Determinative of
Default Rate for Home Purchase Loans (2007 Vintage)
• t
46

47. 47
Increasing leverage drove the boom, keeping
markets from correcting until it was too late

48. What kind of mortgage product best
meets the needs of today’s borrowers?
• Is the thirty year fixed-rate mortgage what we need?
– While it is a proven “affordability product” of long standing, the
thirty-year fixed-rate mortgage does not build equity very quickly.
– Lots of things can happen to a borrower over those thirty years– job
loss, health problems, divorce.
– As Monitor of the National Mortgage Settlement, I have done a lot
of listening in the last two and a half years; including to distressed
borrowers, the people who represent them, and public officials who
deal with the fallout from increased foreclosures and bankruptcies.
– What I have heard confirms what I know from prior experience: that
one or two of those life issues – or, in many, many cases, the trifecta
– have resulted in real financial crisis on a large scale.
– Absent substantial home equity at the outset, the thirty-year fixed
rate mortgage increases the fragility of a borrower’s overall financial
position and puts the borrower at risk for a very long time.
Remarks by Joseph Smith at the American Mortgage Conference, September 11, 2014
48

49. PeArpipoednidcix T 1a: bPeleri oodfic M Tabolret ogfa Rgisek Risk