Executive Summary

Six months after the implementation of the QM/ATR rule, the market appears to
be shifting modestly. This survey serves as both a measure of change in the 2
nd

quarter and a bellwether for originators impressions of market dynamics in light
of regulatory changes under the QM/ATR rule as well as changes at the FHA,
FHFA, and new credit scoring models. The panel was expanded in the 2
nd
quarter
to include members of Community Mortgage Lenders of America.

Highlights of the Survey
The non-QM share of originations more than tripled in the 2
nd
quarter to an
originations-weighted 2.6% from 0.8% in the 1
st
quarter. Rebuttable presumption
expanded as well to 12.8% from 9.8% over this same time frame.
Respondents were less sanguine about their comfort with the QM/ATR rules in the 2
nd

quarter, with just 61.9% indicating that they had fully adapted compared to 73.7% in the
1
st
quarter.
The share of lenders offering rebuttable presumption and non-QM products in the 2
nd

quarter improved, but willingness to originate non-QM and rebuttable presumption
mortgages fell from the 1
st
quarter to the 2
nd
quarter. Lenders were more willing to
originate prime mortgages, though.
Over the next 6 months, nearly half of respondents expected improved access to credit
for prime borrowers with FICO scores between 620 and 720. However, the vast
majority expected no change for rebuttable presumption and non-QM borrowers.
Respondents expect improved investor demand for all mortgage types
Half of respondents indicated that the premium reductions under the FHAs HAWK
program were insufficient or the education fees were too high, while 55% indicated that
the program would not expand credit.
Only 15% of respondents felt that FHAs program of early reviews would help to
alleviate overlays.
However, 85% of respondents indicated that a reduction of LLPAs directed at high LTV
and low FICO borrowers would stimulate access to credit.
Finally, 60% of respondents indicated that the Fair Isaac Companys new FICO 9 scoring
model would help to stimulate access to credit. Only 35% expected no change as they
either defer to their investors or the GSEs scoring models.


The Qua
On Friday
went into
borrower
for varyin
lenders as
mortgage
which inc
fees, a ca
products t

NARs firs
environm
impact of

Responde
sharply fr
offering n
presumpt
while safe

1
For a mo
mortgage-
alified Mo
y, January 10
th
o effect. The D
s ability to re
g degrees of
s it allows the
e applications
ludes full doc
p of 43% on t
that qualify a
st Survey of M
ent and their
f the QM/ATR
ents indicated
om 0.8% in th
non-QMs, tho
tion loans also
e harbor QM

re in-depth dis
qm-rule
ortgage Ru
h
, 2014, the re
Dodd-Frank a
epay their mo
assumed com
em to minimiz
received on
cumentation o
the back-end
and prepayme
Mortgage Orig
r anticipated o
R on productio
d a production
he 1
st
quarter
ugh net willin
o increased a

scussion of the
le and Its I
equirements
act requires th
ortgage and im
mpliance with
ze and to bud
or after Janua
of income, as
debt-to-inco
ent penalties
ginators queri
operational c
on.
n-weighted s
r. The improv
ngness to len
s a share of t
new rules see
Impact
of the ability
hat originator
mposes stiff p
h the ability to
dget for poten
ary 10
th
are r
ssets and emp
me ratio, and
among other
ied originator
changes. This
hare of 2.6%
vement was d
d moderated
otal producti
e http://www.r
y to repay and
rs make a goo
penalties if th
o repay rule,
ntial penaltie
required to co
ployment, a m
d limitations o
r requiremen
rs about their
survey is the
for non-QM
driven by a la
d (discussed b
ion over this
realtor.org/art
d qualified mo
od faith effor
hey do not. T
which is adva
es and litigatio
omply with th
maximum of 3
on the type o
ts.
1

r expectation
e second that
loans in the 2
rger number
below). Rebut
period, from
icles/summary
ortgage (QM)
rt to verify a
The QM rule a
antageous to
on expenses.
he ATR/ QM r
3% for points
of mortgage
s for the QM
measures the
2
nd
quarter, u
of lenders
ttable
9.8% to 12.8
y-of-new-quali
) rule
allows
All
rule
s and
/ATR
e
p
%,

fied-
loans shra
reflect inc
In a set ba
61.9% ver
additiona
to fully im
potential

2
Average F
NARs RCI
increased s
3
The diffe
ank from 89.4
creased numb
ack from last
rsus 73.7% in
l 12 months o
mplement und
litigation cos

FICOs for FHA a
survey were re
share of non-Q
rence could als
4% to 84.6%.
bers of origin
quarters sur
the 2
nd
quart
or more to ad
derwriting cha
ts.
3


and GSE produ
elatively stable
QM and rebutta
so point to a ch
The rise in sh
ators offering
rvey, fewer re
ter. A surpris
dapt to the ru
anges, to ove
ction as measu
e over this peri
able presumpt
hange in the ex
hares of non-
g these produ
espondents in
sing 19.1% of
ules. This shif
ercome invest

ured by Ellie M
od, suggesting
tion lending.
xpanded samp
-QM and rebu
ucts or improv
ndicated havi
respondents
ft could point
tor confidenc
Mae data along
g that a looseni
ple not capture
uttable presu
ved demand
ng adapted to
s indicated tha
t to an altered
ce issues, or t
with first-time
ing of credit di
ed by the contr
mption may
from investo
o the regulat
at it would ta
d view on the
o identify
e buyer shares
id not drive the
rol measures.
ors.
2

ions,
ake an
e time
s from
e
The share
quarter. O
Specifical
between
7.0% and

Responde
quarter re
total diffe
product a
these wer
LTVs grea
e of originator
Originators o
ly, the share o
680 and 720,
9.2%, respec
ents were also
elative to the
erence betwe
re displayed
re focused on
ter than 90 a
rs offering reb
ffering non-Q
of originators
points and fe
ctively.
o asked to rat
1
st
quarter, c
en those who
above. In ge
n high DTI loa
nd interest o
buttable pres
QM loans incr
s offering mo
ees greater th
te their willin
conditional th
o were more
neral, only 20
ns just outsid
nly products.
sumption mo
reased as wel
rtgages to no
han 3.5% and
gness to orig
hat the firm o
willing versus
0% to 30% of
de the 43% th
. Low balance
rtgages incre
l, but at singl
on-QM borrow
d DTIs of 45%
ginate various
offers the prod
s less willing,
respondents
hreshold as w
e, high fees, a
eased by 16.0
e-digit impro
wers with cre
and higher in
s types of mo
duct. Net wi
and the shar
s offered non-
well as high FIC
and low FICO

% in the 2
nd

ovements.
edit scores
ncreased by 1
rtgages in the
illingness, the
re that offer t
-QM loans an
CO borrowers
O loans were t
1.8%,
e 2
nd

e
this
nd
s with
the
least freq
non-QMs
Nearly 80
net willing
quarter an
The vast m
to origina
90 and FIC
prime bo
willingnes
The share
mortgage
responde
uently offere
expanded th
0% of respond
gness to origi
nd 26.3% for
majority of or
te mortgages
COs greater t
orrowers was
ss to lend in t
e of responde
e rule (QM) ju
nts did not ha
d. Relative to
e most while
dents offered
inate rebutta
non-FHA reb
riginators wer
s of this type
han 720, resp
flat in the 2
n
he non-QM s
nts that had
mped from 4
ave an issue c
o last quarter
e high DTI and
rebuttable p
ble presumpt
uttable presu
re willing to o
increased ov
pectively. Ho
d
quarter rela
space of pens
issues closing
47.4% in the 1
closing their p
r, net willingn
d low FICOs m
resumption l
tion loans dec
umption loan
offer prime lo
er this period
owever, willin
ative to the fi
iveness acros
g mortgage(s)
1
st
quarter to
production du
ness to origina
moderated.
oans, both FH
clined 20.0%
s.
oans regardles
d by 13.0% an
gness to orig
rst, a sign cou
ss the spectru
) due to some
66.7% in the
ue to the QM
ate interest o
HA and non-F
from the 1
st
q
ss of FICO an
nd 17.4% for L
ginate lower F
upled with th
um.
e facet of the
2
nd
. A full th
M rule.
only and high
FHA. Howeve
quarter to th
d LTV. Willin
LTVs greater t
FICO or near
he pull back in
qualified
ird of
FICO
er,
e 2
nd

gness
than
n

Some lend
presumpt
end DTIs w
ahead of t
of QM. 9
adapted t
Looking to
borrower
rebutable
while 9.6%
ders have opt
tion or non-Q
with 22.7% of
the pricing de
9.5% of respo
to the QM rul
o the 2
nd
half
s with FICO s
e presumption
% expected it
ted for buffer
QM mortgage.
f respondent
elineation bet
ndents indica
es, a decline
ot 2014, nea
cores betwee
n loans and 3
t to weaken.
rs ahead of th
. The use of b
s employing o
tween rebutt
ated that they
from 22.2% l
rly half of res
en 620 and 72
3.4% expecte
he QM param
buffers was m
one. Only 9.1
table presum
y would not r
ast quarter.
spondents ex
20. However,
ed improvem
meters to prev
most common
1% and 4.5%
ption and saf
remove buffe
pected impro
, the vast maj
ent in the acc
vent producin
n on the 3% c
of responden
fe harbor stan
ers even once
oved access to
jority expecte
ces for non-Q
ng a rebuttab
ap and 43% b
nts used buffe
ndard definiti
they were
o credit for p
ed no change
QM borrowers

ble
back-
ers
ions
rime
e for
s,

More than 40% of respondents expected investor demand for both rebuttable presumption QM and
prime borrowers with credit scores between 620 and 720. Interest in non-QM lending was also
expected to improve, roughly in line with that of high FICO, prime lending.

Other In

Under the
repay a lo
asked wh
willingnes
the produ
space, wit
definitive
reluctance
In the sum
annual M
annual M
The educa
whether t
responde
20% felt t
indicated

ndustry an
e ATR rule, a r
oan, but it is n
ether the add
ss to originate
ucts indicated
th 53.0% indi
residual inco
e to originate
mmer of 2014
IP and 50 bps
IP would fall
ation program
the incentives
nts indicated
hat it would
that the upfr
nd Policy Is
residual incom
not a definitiv
dition of a br
e in the rebut
d that a residu
cating that th
ome test wou
e product of lo
4, the FHA an
s in the UFMI
by an additio
m is estimated
s are sufficien
that the ince
help, but cou
ront costs of t
ssues
me test can b
ve bright line
right line res
ttable presum
ual income te
hey would be
ld also help in
ower quality.
nounced its H
P to consume
onal 15 basis p
d to cost the
nt to attract c
entives would
ld be stronge
the education
be used as sup
e proof of co
sidual income
mption and/o
est would be m
either more
n the non-QM

HAWK progra
ers who comp
points if the b
borrower $40
consumers to
d result in incr
er and 40% fe
n program we
pporting evid
ompliance wit
e test to the Q
r non-QM spa
most helpful
likely or muc
M space, but e
m that will of
plete a buyer
borrower is n
00 upfront. R
o the new pro
reased dema
elt that they w
ere too high f
dence of a bor
th the rule. P
QM/ATR wou
ace. Respond
in the rebutta
ch more likely
even there, le
ffer a discoun
r education pr
ot delinquen
Respondents
ogram. Only 1
nd for FHA in
were not. 10%
for consumer
rrowers abili
Participants w
ld improve th
dents who off
able presump
y to originate.
enders displa
nt of 10 bps in
rogram. The
t after 18 mo
were asked
15% of
nsurance, whi
% of respond
rs.
ty to
were
heir
fered
ption
. A
yed a

n
onths.
ile
ents
In its pres
delinquen
counselin
program w
that there
somewha
The FHA i
reduce th
intent is t
were aske
ss release for
ncy rates of u
g. Survey pa
would affect
e would be no
at.
s also rolling
e risk to lend
o ameliorate
ed whether th
the HAWK pr
p to 30% for
rticipants we
lenders cred
o change in ac
out a program
ders of proble
concerns abo
he early revie
rogram, the F
counseled bo
re asked whe
it overlays on
ccess, while 2
m for earlier
ems in loan fil
out buy-back
ews would im
FHA cited rese
orrowers com
ether, given th
n loans origin
20% indicated
reviews of m
es that could
risk and to re
pact credit ov
earch that sh
mpared to sim
he evidence o
ated for the
d that it woul
ortgage files.
lead to subs
educe lender
verlays on loa
ows a reduct
milar borrower
of reduced ris
FHA. A 55% m
d ease credit
. The program
equent costly
r overlays in t
ans originated
ion in serious
rs without
sk, the HAWK
majority indic
t tightness
m is intended
y put-backs.
urn. Particip
d for the FHA

s
K
cated

d to
FHAs
ants
A.
60% indic
somewha
The FHFA
was appro
demand i
whether a
the credit
box, while
Finally, th
given to t
cated that the
at and signific
took comme
opriate. Seve
n general and
a reduction in
t box. A robu
e 20% felt tha
he Fair Isaac C
he impact of
ere would be
antly reduce
ents from the
eral of the que
d from high LT
n LLPAs target
st majority of
at there woul
Corporation re
unpaid medi
no change, w
overlays. 10
public in ear
estions posed
TV and low cr
ted at lower F
f 80% felt tha
d be no chan
ecently introd
cal bills and t
while 15% ind
% defer to th
ly September
d by the FHFA
redit borrowe
FICO and/or h
at a reduction
ge.
duced a new
the effect of m
icated that th
heir investors
r as to wheth
A dealt with t
ers. Survey p
higher LTV bo
n in rates wou
scoring mode
missed payme
he reviews wo
rules.
er the curren
he impact of
participants w
orrowers wou
uld help to ex
el. In "FICO 9
ents on debts
ould both
nt g-fee struct
the g-fees on
were asked
uld help to ex
xpand the cre
9", less empha
s that have

ture
n
xpand
dit

asis is
subseque
25 basis p
participan
A 60% ma
indicated
while 35%
GSEs mar
ntly been pai
points for the
nts were aske
ajority indicat
that it would
% deferred to
rket share an
id off are elim
former group
ed if this new
ted that the n
d not impact t
their investo
d their relian
minated. FICO
p and by as m
scoring mode
new scoring m
their decision
ors model or
ce on an earl
O estimates th
much as 100 p
el would incre
model would i
n to accept as
that of the G
ier FICO mod
hat the new m
points for the
ease accepte
increase acce
they use an
SEs. This res
del.
model could i
latter group.
d application
epted applicat
earlier versio
ult was a surp
improve scor
. Survey
ns at their firm
tions. 5%
on of the mod
prise given th

es by
m.

del,
he
Appendix A: About the Survey
In August of 2014, NAR Research sent out a survey to a panel of 135 different mortgage originating
entities. The panel was expanded in the second quarter from 65 to include members of Community
Mortgage Lenders of America. The survey instrument was sent by email on Tuesday the 19
th
of August
and closed on Tuesday, September 2
nd
. Questions in the survey instrument covered the characteristics
of the originators, a subset of questions focused on the qualified mortgage rule, and a set of questions
focused on the FHA, FHFA, and the new FICO 9 score. There were 24 unique responses to the survey for
a response rate of 17.8% and a margin of error of 6.4% at a 95% level of confidence.

Mortgage bankers dominated this sample, but this sample had more diversity in terms of size of annual
originations, a proxy for bank size. Originator profiles were also consistent with prior surveys in terms of
geographic distribution, purchase share, average annual production volume, and the distribution of
destinations/purchasers of the originators production.

Questions can be directed to:
Ken Fears
Senior Economist,
Director, Housing Finance and Regional Economics
The National Association of REALTORS
kfears@realtors.org
(202)383-1066

Kenneth R. Trepeta Esq.
Director Real Estate Services
National Association of REALTORS
500 New Jersey Ave, NW
Washington, DC 20001
(202) 383-1294