202 Global Securitisation and Structured Finance 2007
29
T his chapter updates Deutsche Bank research published in 2004 entitled
“Credit Resiliency of European Residential Mortgage-Backed Securities”.
At that time, there was no shortage of headlines in the press and other
publications about the residential housing bubble in many European
markets, with commentators speculating on whether a correction in house
prices was pending. Since 2004, residential house prices have continued to
climb across most European jurisdictions, the only exception being Germany.
The recent US subprime meltdown has brought into light similar deflation
risks in Europe, with the United Kingdom, Spain and Ireland particularly cited
as markets most exposed to risks of a property slump.
This chapter looks at the resiliency of European mortgage-backed
securities to a decline in house prices. Losses experienced by residential
mortgage-backed securities (RMBS) are a simultaneous function of default
rates and recoveries (the latter in turn determined primarily by house prices).
The chapter isolates the stresses on house prices in order to test for break-
even default rates for each hypothetical country RMBS. It observes that
default rates and house-price trends are often not independent, in that
higher mortgage default rates would in all likelihood depress house prices
and sometimes vice versa. House-price declines alone may not necessarily
result in higher defaults, particularly in countries where borrowers remain
willing to pay despite negative equity.
The chapter formulates hypothetical benchmark RMBS pools that most
closely resemble RMBS portfolios outstanding in selected European
mortgage-backed markets. It then calculates the default rates that each
tranche in typical European RMBS capital structures can withstand under a
Credit resiliency of European residentialmortgage-backed securities in 2007
Carole Bernard
Deutsche Bank
© Deutsche Bank 2007
Credit resiliency of European residential mortgage-backed securities in 2007 I Global Securitisation and Structured Finance 2007
Global Securitisation and Structured Finance 2007 203Deutsche Bank
severe house-price shock, which is assumed as a mean
reversion of house prices back to their average levels
since 1995 in each European country.
Defining the hypothetical European RMBS
Deutsche Bank has created 10 hypothetical RMBS deals
representative of mortgage-backed securitisations in
the most developed RMBS markets in Europe.
For each hypothetical RMBS portfolio, loan-to-value
distributions have been formulated using pools
information from benchmark deals outstanding. Loan-to-
value distributions are an important consideration, as two
mortgage pools with the same weighted average loan-to-
value but different loan-to-value distributions will behave
differently from house-price stresses. This study includes
for the first time high loan-to-value Spanish RMBS,
reflecting the recent proliferation of such deals.
For each country-specific RMBS benchmark portfolio,
a geographic distribution of borrower/property domiciles
has been assumed, reflecting the typical RMBS deal from
that jurisdiction. The house-price stress applied in the
study uses an average house-price trend weighted by the
regional significance of the hypothetical RMBS pool,
rather than the national average house-price trend. In
effect, the house-price decline assumptions are made
more meaningful by taking into account RMBS-pool
regional concentrations.
The seasoning assumed in this study is consistent
with the various RMBS markets (all mortgage loans are
assumed to share the same pool average seasoning). The
analysis gives credit to seasoned collateral by taking into
account house-price changes (if any) over the age of the
mortgage, thus reflecting current loan-to-value figures
rather than original loan-to-value figures. As no principal
repayment in seasoned mortgages is taken into account,
the bias is conservative.
For each jurisdiction, different recovery periods and
legal costs associated with repossessions are assumed.
Figure 1: Average price of residential properties in Europe, 1989 to 2007
Source: Department of Environment (Ireland), Nederlandse Vereniging van Makelarrs (Netherlands), Ministerio de
Vivienda (Spain), Nationwide (UK), Bulwien AG (Germany), Scenari Imobiliari (Italy), Bank of Greece (Greece),
OFHEO (USA), CBA (Australia)
400
300
200
150
Netherlands
250
Q291
USA
Italy
Spain
Index 1995 = 100
50
100
350
UK
Australia
West Germany
Ireland
Q292
Q294
Q295
Q296
Q297
Q298
Q299
Q200
Q201
Q202
Q203
Q204
Q205
Q206
Q290
Global Securitisation and Structured Finance 2007 I Credit resiliency of European residential mortgage-backed securities in 2007
204 Global Securitisation and Structured Finance 2007 Deutsche Bank
Such assumptions are based on information taken from
Fitch’s rating methodology reports (see Fitch’s RMBS
criteria reports for the relevant jurisdictions). The
recovery period represents the time lag between the first
day of the delinquency and the forced sale of the
repossessed property, and all legal costs associated with
the legal process include lawyer fees, taxes, eviction of
residents if necessary, refurbishment, valuation and
brokerage. Legal and accrued carry costs (current interest
rates have been assumed during the foreclosure period)
are deducted from gross proceeds for property
repossessions. Table 1 summarises the assumptions
made in costing the recovery process in each country.
All 10 hypothetical RMBS deals assume a static,
pass-through structure (with the exception of UK prime
RMBS). Hard-credit enhancement (ie, subordination
only or ignoring excess spread) at each rating or tranche
level is based on averages for benchmark deals in each
of the mortgage markets under study. For example, the
subordination level for AAA high loan-to-value (ie, 97
Figure 2: Impact of house price shocks on European RMBS – outline of study
Source: DB Global Markets Research
Assumptions Stress runs
ash Reserve 0
50
100
150 200 250 300 350
‘89
‘90 ‘91
‘92 ‘93
‘94 ‘95
‘96 ‘97
‘98 ‘99
‘00 ‘01
HypotheticalRMBS
credit structures
Hypothetical RMBS portfolio
Apply house price ‘shocks’
AAA94%
A4%
BBB2%
Cash reserve2%
weighted average LTV,
distribution of LTV
seasoning
geographic distribution
repayment type and
prepayments
LTV distribution
recovery timing and costs
Tranche loss = f (default rate, loss severity, credit enhancement)
Quantify the hypothetical default rate that the RMBS portfolio can withstand before the rated notes experience the first unit of
principal loss
Long termtrend line
Marketvalue
decline
Credit resiliency of European residential mortgage-backed securities in 2007 I Global Securitisation and Structured Finance 2007
Global Securitisation and Structured Finance 2007 205Deutsche Bank
per cent loan-to-value) Spanish RMBS is over 13 per
cent, while the comparable support for AAA 65 per cent
loan-to-value Spanish RMBS is 5 per cent. The assumed
average RMBS hard-credit support has increased across
the capital structure in most jurisdictions compared to
the 2004 study, with the exceptions of the Netherlands,
Italy and Portugal (German RMBS subordination has
remained generally stable). Higher or lower observed
credit enhancement in newer-generation European
RMBS reflects commensurate changes in the risk profile
of securitised portfolios (eg, loan-to-value figures and
seasoning) to include the market value decline
assumptions made by the rating agencies.
Tables 2 and 3 sets out the main characteristics of
the 10 mortgage portfolios for each country, as well as
the hard-credit support assumed at each tranche level.
It also shows other mortgage characteristics (eg, lien,
owner-occupied or investment mortgages and type of
interest rate), which are simply illustrative of the
different mortgage market characteristics and have not
been incorporated into the analysis.
Hypothetical impact of house price shocks on
European RMBS
The study calculates the hypothetical default rates that
can be withstood by the different country-specific
RMBS under the impact of an immediate shock decline
in house prices back to their long-term trend. The stress
applied to house prices in the respective countries is
based on a fall in prices back to the average level of the
period between 1995 and 2007. Based on this
calculation, the market value decline assumptions are
shown in Table 5. Shown separately are the market
value decline (MVD) assumptions based on current
house prices (at the end of 2006), as well as the
effective MVD taking into account any property price
appreciation since origination of the loan (ie, giving
benefit to seasoning of the mortgage portfolio). The
latter effective MVD assumption is used in the
following analysis.
Assuming that property prices fall on day one, the
extent of the shock for each mortgage market is the
effective MVDs shown in Table 5. Break-even default
rates given this shock are shown in Table 4.Table 4 shows
an average break-even default rate for each country
that, by definition, has a random frequency - that is,
defaults are assumed to occur randomly across the
RMBS pool irrespective of factors such as loan-to-value,
borrower or property type or loan interest rate structure.
The results in this updated study should not be
Spain Ireland UK Netherlands Germany Italy Portugal
Legal costs 10% 5.5% 5.0% 5.0% 2.0% of 7.0% 10.0%property value
Recovery 2.5 years 3.0 years 1.5 years 1.0 year 2.0 years 3.0 to 3.0 years
period years* (1 year for 10.0
buy-to-let) years
Current 5.1% 5.2% 5.8% prime 5.2% 5.0% 5.4% 5.3%
mortgage (floating (floating 6.8% non- (10-year fixed) (10-year (floating (floating
interest rate rate) rate) conforming fixed) rate) rate)
(floating rate)
Table 1: Recovery timing and cost assumptions for our hypothetical European RMBS
Source: Fitch, Council of Mortgage Lenders, Associcion Hipotecaria Espanola, European Mortgage Federation.
Note: *Fitch assumes three years.We have adjusted our assumptions to reflect recent changes in law and practice.
Global Securitisation and Structured Finance 2007 I Credit resiliency of European residential mortgage-backed securities in 2007
206 Global Securitisation and Structured Finance 2007 Deutsche Bank
SpainSpain (high
IrelandU
K prim
eU
K buy-to
-U
K no
n-N
etherlands*G
ermany***
ItalyPo
rtugal
LTV po
ol)
letco
nform
ing
Poo
ls characteristics
Original w
eighted65%
97%74%
75%79%
78%85%
55%**
65%72%
average LTV
Seasoning 1.5 years
1.7 years2.0 years
2.0 years6 m
onths6 m
onths1.0 year
3.0 years1.0 year
2.0 years
Mortgage typ
esA
ll floatingA
ll floating30%
fixed55%
fixed60%
fixed25%
fixed100%
fixed100%
fixedA
ll floatingA
ll floating
First lien100%
100%100%
100%100%
100%100%
50%100%
100%
Ow
ner-100%
100%100%
100%0%
95%100%
70%100%
100%
occupied
Repayment type
Annuity
100%100%
100%60%
10%
50%0%
30%100%
100%
Interest only0%
0%0%
10%90%
50%50%
5%0%
0%
Endowm
ent/life/0%
0%0%
30%0%
0%50%
65%0%
0%
savings
Geograp
hic Valencia 15%
,C
atalunya 12%,D
ublin40%
London &London &
London &Z
.Holland
Berlin 10%,
North 60%
Lisboa 35%
concentrationC
atalunyaM
adrid 65%SE 30%
SE 50%SE 30%
20%,
East Germ
anyCentral 20%
20%,M
adrid N
.Holland
15%South 20%
30%15%
,
N.Brabant
15%
Average p
ortfolio25
2827
2025
2530
2520
25
maturity (years)
Table 2:Euro
pean RM
BS hypothetical po
ols and capital structures
Source: D
B Global M
arkets Research
Note:*Loan to indexed foreclosure.**
Loan to appraised value.***A
ssumes cash structure
Credit resiliency of European residential mortgage-backed securities in 2007 I Global Securitisation and Structured Finance 2007
Global Securitisation and Structured Finance 2007 207Deutsche Bank
Spai
nSp
ain
(hig
hIr
elan
dU
K p
rim
eU
K b
uy-t
o-
UK
no
n-N
ethe
rlan
ds*
Ger
man
y***
Ital
yPo
rtug
alLT
V p
oo
l)le
tco
nfo
rmin
g
Rat
ing
AA
A5.
0%13
.5%
8.5%
11.0
%16
.0%
16.0
%6.
0%6.
0%8.
0%6.
5%
AA
3.5%
10.0
%5.
0%7.
5%9.
0%9.
0%4.
0%3.
0%4.
0%4.
5%
A
2.0%
6.2%
2.3%
5.0%
5.0%
5.0%
2.5%
2.0%
2.5%
3.0%
BBB
1.0%
4.5%
1.0%
1.6%
1.8%
1.7%
1.0%
1.0%
1.0%
1.0%
Sour
ce:D
B G
loba
l Mar
kets
Res
earc
h
Not
e:*L
oan
to in
dexe
d fo
recl
osur
e.**
Loan
to
appr
aise
d va
lue.
***
Ass
umes
cas
h st
ruct
ure
Tabl
e 3
:Ass
umed
har
d cr
edit
enh
ance
men
t
Spai
n (l
owSp
ain
(hig
hIr
elan
dU
K p
rim
eU
K b
uy-t
o-
UK
no
n-N
ethe
rlan
dsG
erm
any
Ital
yPo
rtug
alLT
Vpo
ol)
LTV
po
ol)
let
conf
orm
ing
Rat
ing
AA
A23
%25
%31
%38
%48
%46
%22
%10
0%45
%50
%
AA
16%
18%
18%
26%
27%
26%
14%
100%
22%
35%
A
9%11
%9%
17%
15%
14%
9%10
0%14
%23
%
BBB
5%8%
5%6%
5%5%
4%10
0%6%
8%
Sour
ce:D
B G
loba
l Mar
kets
Res
earc
h
Tabl
e 4
:Hyp
oth
etic
al b
reak
-eve
n de
faul
t ra
tes
assu
min
g ‘d
ay-o
ne’m
ean-
reve
rsio
n pr
ice
sho
ck
Global Securitisation and Structured Finance 2007 I Credit resiliency of European residential mortgage-backed securities in 2007
208 Global Securitisation and Structured Finance 2007 Deutsche Bank
surprising and are consistent with the findings in the
original study. The thrust of the analysis and the
assumptions used accentuate the influence of loan-to-
value coverage and historical house-price trends on
RMBS credit resiliency, taking into account the benefits
provided by hard-credit support. High loan-to-value
RMBS in a country that has experienced strong house-
price growth will, by definition of the analysis, have
lower break-even default rates compared to low loan-
to-value RMBS or RMBS from jurisdictions experiencing
more moderate house price inflation. The conclusions
are again intuitively defensible, but do not take into
account household default behaviour or affordability
factors. With the exception of Germany, all other RMBS
markets in this updated study look incrementally more
exposed to a day-one mean reversion in house prices
compared to the original study in 2004. Further
observations are set out below by country.
The Netherlands
Dutch RMBS exhibit a lower break-even default rate in the
updated study compared to the original analysis, despite
lower effective MVD assumptions. This is explained by
lower credit enhancement levels in recent benchmark
Dutch RMBS relative to the observations in 2004
Spain
Spanish RMBS pools are particularly penalised under this
scenario given the significantly greater market-value
stresses used today, without commensurate increases in
credit enhancement. However, Spanish high loan-to-
value pools exhibit incrementally higher break-even
default rates across the capital structure compared to
Spanish low loan-to-value pools in the study, suggesting
- academically at least - that the risks associated with
higher loan-to-value loans appear adequately covered by
higher credit support.
United Kingdom
Similar to Spain, UK prime RMBS look less resilient in
the updated study given the greater potential degree of
house-price correction today without corresponding
increases in credit support. A significant additional
factor for the UK prime market is that the 2004 study
gave some benefit to mortgage indemnity guarantees,
whereas the updated study does not take into account
any mortgage insurance, reflecting current market
practice. For example, in 2004 Deutsche Bank calculated
that a BBB prime UK RMBS tranche would have had a
break-even default rate of 30 per cent, compared to 6
per cent today. By contrast, senior UK non-conforming
RMBS look incrementally more resilient in the updated
study, although junior tranches are more exposed
compared to the 2004 study. This default break-even
outcome reflects greater observed credit enhancement
at the senior part of non-conforming RMBS capital
structures, which acts to more than offset the more
severe house-price decline stresses used in the updated
study. It is no coincidence that higher credit support
among senior bonds in certain UK non-conforming
RMBS stems from recent adjustments to rating agency
methodology in order to take into account pool loan-
to-value distributions.
Ireland
Irish RMBS also exhibit lower break-even default rates
compared to the results in the 2004 study, owing
mostly to the fact that mortgages securitised are no
longer covered by mortgage indemnity guarantees,
which used to benefit loans with a loan-to-value
greater than 75 per cent.
Germany
According to the analysis, German RMBS remain equally
well positioned to withstand this scenario three years
on, in that the entire pool could hypothetically default
under the assumptions without loss of principal to
investors. This resiliency is largely thanks to historically
stable house prices.
Italy
Italian RMBS are shown to have a lower break-even
default rate compared to the original study, given the
greater stresses used in the model coupled with lower
observable hard-credit support across recent capital
structures. Italian mortgage securitisations are also
Credit resiliency of European residential mortgage-backed securities in 2007 I Global Securitisation and Structured Finance 2007
Global Securitisation and Structured Finance 2007 209Deutsche Bank
penalised for the high legal and accrued carry costs
associated with the longer time to recovery
Portugal
Similar to Italian mortgages, Portuguese RMBS also
look more exposed to a day-one house-price
mean reversion compared to the analysis performed in
2004. This is explained by both the greater MVD stress
applied and lower assumed credit support compared to
the 2004 study
Comment
If RMBS were stressed for similar house price shocks
after two years, each of the hypothetical RMBS
markets would be able to withstand even higher default
rates before experiencing the first unit of loss of
principal. This is simply because a two-year lag
would effectively allow for some benefit to accrue
from pool seasoning and structural deleverage.
Countries characterised by high pre-payment speeds
will naturally benefit the most from a build-up in hard-
credit support as the deal delevers. The larger the
combined impact of greater credit enhancement and
lower loan-to-value figures, the higher the incremental
break-even default rate relative to a house price shock
assumed on day one.
Concluding remarks
The equation below solves the resiliency of default
rates, assuming differing timing of house-price decline
stresses.
As the equation highlights, the extent of resiliency
depends primarily on the loan-to-value figures and
subordination levels assumed, with the former taking
into account factors such as seasoning and loan pay-
down and adjusted for the costs normally associated
with default and repossession (the study assumes that
RMBS credit support is derived entirely from
subordination and ignores the benefits of additional
forms of credit support, such as excess spread, and
structural triggers). The analysis was overlaid by a
number of assumptions and caveats that compromise
to some extent the practical utilisation of the findings.
Country MVD assumption in this study
vs. current house price vs. price at time of mortgage origination*
Spain 46% 38% (1.5 years weighted average seasoning)
Ireland 46% 37% (2.0 years weighted average seasoning)
UK 43% 35% (2.0 years weighted average seasoning)
Netherlands 26% 24% (1.0 year weighted average seasoning)
Italy 23% 16% (1.0 year weighted average seasoning)
Portugal 16% 13% (2.0 years weighted average seasoning)
Germany 0% 0% (2.0 years weighted average seasoning)
Source: DB Global Markets Research
Note: * These effective MVD stresses are used in our study.
Table 5: Market value decline applied to property price
Tranchei credit support = B/even default ratei ∑Ó1-
House price post-shock%/Current LTVk%
k=loans where post-shock (adj.) LYVs>100%
Global Securitisation and Structured Finance 2007 I Credit resiliency of European residential mortgage-backed securities in 2007
210 Global Securitisation and Structured Finance 2007 Deutsche Bank
Figure 3: Default-loss combinations for our hypothetical European RMBS capital structures under a
‘day one’ house price shock
35
0
5
10
15
20
25
30
0 10080604020
UK prime
% Hypothetical capital structure loss
Scenario default rate %
Cash reserve
AA notes
AAA notes
35
0
5
10
15
20
25
30
0 80604020
UK non-conforming
% Hypothetical capital structure loss
Scenario default rate %
AA notes
AAA notes
100Cash reserve
A notes
BBB notes
0 10080604020
Spain (high LTV pool)
% Hypothetical capital structure loss
Cash reserve
AA notes
AAA notes
BBB notes
0 10080604020
Netherlands
AAA notes
0 10080604020
Spain (low LTV pool)
0 10080604020
Ireland
Cash reserve
A notes
AA notes
BBB notes
AAA notes
A notes
BBB notes
40
60
0
10
20
30
40
50
A notes
30
0
5
10
15
20
25
% Hypothetical capital structure loss
Scenario default rate % Scenario default rate %
AA notesA notes
BBB notes
Cash reserve
% Hypothetical capital structure loss25
0
5
10
15
20
Scenario default rate %
Cash reserve
AA notes
AAA notes
A notesBBB notes
% Hypothetical capital structure loss30
0
10
15
20
25
5
Scenario default rate %
Credit resiliency of European residential mortgage-backed securities in 2007 I Global Securitisation and Structured Finance 2007
Global Securitisation and Structured Finance 2007 211Deutsche Bank
This has been an academic exercise (for a full list of
other assumptions, please see the report published on
April 18 2007). However, Deutsche Bank believes that
the study will nonetheless serve as a useful litmus test
of European RMBS resiliency to any housing market
shocks. The break-even default rates, although
hypothetical, can also provide greater academic clarity
as to comparative strengths of different RMBS under
extreme scenarios.
Based on these findings, AAA RMBS tranches among
European RMBS continue to appear to be well insulated
against any house-price shock, although hypothetically
less so compared to the 2004 study. However, the risk of
principal loss on the hypothetical senior RMBS tranches
continues to appear very remote. Subordinated
mortgages look noticeably more vulnerable compared to
the 2004 results (Germany being the only exception), yet
the cushion to BBB default thresholds remains
appreciable considering current pool performance and
historical default trends.Take the example of Spanish and
Dutch BBB RMBS, which are among the weakest of the
RMBS tranches under study. However, assuming house
prices mean revert on day one, the risk of loss is limited
unless default rates reach 40 and 16 times the existing
late-stage arrears experienced by both markets
respectively. In the case of Spain, the likelihood of house
prices falling by 46 per cent looks remote, considering
historical house-price performance and housing market
fundamentals. Defaults in UK prime BBB RMBS could
hypothetically reach 75 times the current mortgage
market repossession rate before tranches experience the
first unit of principal loss. Although UK non-conforming
BBB RMBS appear to be the most vulnerable when
comparing hypothetical break-even defaults to current
late-stage arrears, the analysis ignores the benefit of
excess spread which provides material protection in
such deals.
In short, the findings of the analysis suggest that
any bursting of house-price bubbles, if taken in
isolation, appears unlikely to pose material credit risks
to European RMBS.
This chapter is taken from previously published Deutsche
Bank research.