Professional Documents
Culture Documents
Commercial Mortgage Backed Securities: 1 © 2014 Oncourse Learning. All Rights Reserved
Commercial Mortgage Backed Securities: 1 © 2014 Oncourse Learning. All Rights Reserved
Commercial Mortgage Backed Securities: 1 © 2014 Oncourse Learning. All Rights Reserved
© OnCourse Learning
© 2014 OnCourse Learning. All Rights Reserved. 5
CMBS - Servicers and Lingo …
Source: Based on data from Commercial Mortgage Alert and Clodfelter (2005).
In CMBS it is usually the default risk dimension that is most important (most commercial
mortgages have “prepayment protection”).
The opposite used to be true in RMBS, where duration was traditionally the prime concern,
due to the greater prepayment risk in residential loans (RMBS pools from FNMA FHLMC
had “default protection”).
The way the CMBS industry has worked, an “IO” class is “stripped” off of the other
securities (from the excess of pool loan coupon interest over the A-Tranche coupon
interest), providing shorter-duration securities that have been major source of profit to
CMBS issuers.
© 2014 OnCourse Learning. All Rights Reserved. 10
20.2.1 A Simplified Numerical Example of Tranching
© OnCourse Learning
© 2014 OnCourse Learning. All Rights Reserved. 11
20.2.2 Allocating the Credit Losses
Example bond pricing, for Class B, the 12% market yield implies bond
price (value) is $24.15, discounting contractual cash flows @ the yield:
2.50 27.50
B $24.15
1.12 1.122
Example realized yield for Class B (after credit losses) is the ex post IRR
based on realized cash flows & original market price:
2.50 19.00 2.50
$24.15 , IRR 0.34%
(1 IRR ) (1 IRR )2
© 2014 OnCourse Learning. All Rights Reserved. 15
20.2.1 (2e revised) A simple numerical example of tranching...
CMBS Structure of Securities in the Deal…
Three classes (tranches) are created based on the underlying pool, and sold into the
bond (CMBS) market:
20-7 (revised) Credit Value as Contract Cash Flows:
Class Par Value WAM Support Coupon YTM CMBS Yr 1 Yr 2
A $75 1.33 25% 8% 8% $75.00 56.00 27.00
X (IO) $100 1.25 N/A 1.5%, 1% 8% $1.82 1.50 0.50
B $25 2.00 0% 10% 12% $24.15 2.50 27.50
Pool $100 1.50 N/A 10%(WAC) 10% $100.97 60.00 55.00
CMBS IRR 9.27% <== Due to Added Value. ==> -$100.97 60.00 55.00
Pool IRR 10.00% -$100.00 60.00 55.00
A Tranche is “senior”, “investment grade” securities:
• Gets retired 1st (all five 1-yr loans liquidating pmts would go to A).
• 25% credit support 25% of pool par value will be assigned credit losses (par
value lost in default) before Tranche A receives any credit losses (any reduction in
par due to default). Effective LTV for A tranche = (1-0.25)70% = 52.5%.
(Underlying properties would have to lose 47.5% of their value before Tranche A
gets hit, since it is most senior tranche.)
• Shorter duration: WAM = (50/75)*1 + (25/75)*2 = 1.33 yrs.
The idea is that at issuance the sum of the values of all of the tranches
exceeds the value (cost) of the pool of all the individual mortgages:
A + B + X = $75.00 + $24.15 + $1.82 = $100.97 > $100.
Example, in the case of Bond B, its price (mkt value) is less than its
par value ($24.15 < $25). This reflects (implies) its YTM is greater
than its coupon interest rate (12% > 10%).
Why do you suppose this is so for Bond B?...
© 2014 OnCourse Learning. All Rights Reserved. 22
Now suppose all loans pay as contracted except one of the 2-yr loans
defaults in yr.2 paying no interest that year and recovering only $5
million in foreclosure sale proceeds (5/11 = 45% “recovery rate”,
55% loss “severity”). What will the ex post CMBS cash flows look like?...
20-7 (revised) Credit Realized Cash Flows: Realized
Class Par Value WAM Support Coupon YTM Yr 0 Yr 1 Yr 2 IRR
A $75 1.33 25% 8% 8% -$75.00 56.00 27.00 8.00%
X (IO) $100 1.25 N/A 1.5%, 1% 8% -$1.82 1.50 0.50 8.00%
B $25 2.00 0% 10% 12% -$24.15 2.50 21.50 -0.34%
Pool $100 1.50 N/A 10%(WAC) 10% -$100.97 60.00 49.00 5.44%
CMBS IRR 9.27% <== Due to Added Value. ==> 60.00 49.00
Pool IRR 10.00% -$100.00 60.00 49.00 6.16%
Example, in the case of Bond B, its price (mkt value) is less than its
par value ($24.15 < $25). This reflects (implies) its YTM is greater
than its coupon interest rate (12% > 10%).
Why do you suppose this is so for Bond B?...
© 2014 OnCourse Learning. All Rights Reserved. 25
Recall value of the deal at time of issuance…
20-7 (revised) Credit Value as Contract Cash Flows:
Class Par Value WAM Support Coupon YTM CMBS Yr 1 Yr 2
A $75 1.33 25% 8% 8% $75.00 56.00 27.00
X (IO) $100 1.25 N/A 1.5%, 1% 8% $1.82 1.50 0.50
B $25 2.00 0% 10% 12% $24.15 2.50 27.50
Pool $100 1.50 N/A 10%(WAC) 10% $100.97 60.00 55.00
CMBS IRR 9.27% <== Due to Added Value. ==> -$100.97 60.00 55.00
Pool IRR 10.00% -$100.00 60.00 55.00
The issuer’s track record is considered as well as the pool of loans & the
underlying property collateral.
Traditional underwriting measures such as LTV ratio and DCR are examined for
the pool as a whole.
Larger mortgages in the pool are examined individually.
Pool aggregate measures (weighted average) are considered.
Pool heterogeneity is also considered:
• Dispersion in LTV & DCR,
• Diversification of collateral (by property type, geographic location).
Diversity & heterogeneity of the mortgages within a pool can matter as much as
the average characteristics of the pool, esp. for lower-rated tranches:
• e.g., Diversification Reduced default risk for senior trances; Increased
default risk for lower tranches (esp. first-loss). Why?...
Rating agencies (and consultants working for them) develop & employ:
• Econometric models of commercial mortgage default probability (e.g.,
regression, proportional hazard models).
• Empirical estimates of conditional loss severity.
• Monte Carlo simulation of interest rates, property market, property
dispersion, and credit losses, to “stress test” the pool and the various
tranches that may be defined based on it. (Need to consider idiosyncratic
risk in individual properties.)
But ultimately, in a mass-production industry (CMBS by 2000s), standardization
occurs. Each rating agency produced their “model”, which they shared with
issuers, to facilitate the security production process. Also necessary to make
rating process fair and objective across all deals & issuers.
Variables that can be important in analyzing the credit quality of a mortgage pool
and the various tranches that can be carved out of it, in either quantitative or
qualitative analysis, include:
In the early days of the CMBS industry (1990s), they probably could not have gotten
away with this. The bond mkt needed to “believe” that the ratings meant the same thing
as what they were familiar with. But by now, the MBS industry may (may) have enough
history that it will work?...
© 2014 OnCourse Learning. All Rights Reserved. 35
20.3.2 Credit Rating, Market Yields, & CMBS
Structure
Rating agencies (Moody’s, Fitch, S&P, Morningstar, etc…) develop
“models” that they use (& make available to issuers) to determine
amount of subordination required for each rating in each issue
(considering mortg pool characteristics).
More lenient the model (lower subord requirements) More profit
for securities issuer. Hence, pressure on rating agency. However…
Model too lenient Investors lose confidence in rating, discount
price willing to pay for securities, drive up mkt yield on the issue,
reduces profitability and undercuts viability of entire industry. Result:
Trade-off. Balancing act.
Example, in the case of Bond B, its price (mkt value) is less than its
par value ($24.15 < $25). This reflects (implies) its YTM is greater
than its coupon interest rate (12% > 10%).
100
110
80
20 90
0
80
-20
-40
70
-60
-80 60
1978Q2
1980Q2
1982Q2
1984Q2
1986Q2
1988Q2
1990Q2
1992Q2
1994Q2
1996Q2
1998Q2
2000Q2
2002Q2
2004Q2
2006Q2
2008Q2
2010Q2
Mortg Capital Flow (Net Chg Bal) CRE Prices (TBI net infl, Right-Hand Axis)
100
80
60
40
$ Billions
20
-20
-40
-60
-80
1978Q2
1980Q2
1982Q2
1984Q2
1986Q2
1988Q2
1990Q2
1992Q2
1994Q2
1996Q2
1998Q2
2000Q2
2002Q2
2004Q2
2006Q2
2008Q2
2010Q2
All CMBS
CMBS accounted for up to half of all CRE credit (more than half of
permanent debt) during peak bubble period.
© 2014 OnCourse Learning. All Rights Reserved. 41
Exhibit 20-7: History of CMBS Subordination Levels,
1995-2008 (pre-crisis)
20-7: History of CMBS Subordination Levels, 1995-2008 (pre-crisis)
35
30
25 Super Sr AAA
20 Aaa
Aa
15
A
10 Baa
Ba
5
0
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
Source: Authors' correspondence with Moody's Investors
30
25 Super Sr AAA
20 Aaa
Aa
15
A
10 Baa
Ba
5
0
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
Source: Authors' correspondence with Moody's Investors
What should it have said (to the rating agencies) when starting in 2004 the
bond mkt began to demand (and “price”, i.e., pay for) “super-senior” tranches
above the AAA subordination cut-off?...
30
25 Super Sr AAA
20 Aaa
Aa
15
A
10 Baa
Ba
5
0
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
Source: Authors' correspondence with Moody's Investors
If 12% was the worst case ACLI credit loss experience, and AA should be
protected against that, then shouldn’t the AA protection have been a bit greater
than 12% subordination?... Problem 2004-2007.
© 2014 OnCourse Learning. All Rights Reserved. 45
The pre-crash history of Rating Agency CMBS subordination levels…
20-7: History of CMBS Subordination Levels, 1995-2008 (pre-crisis)
35
30
25 Super Sr AAA
20 Aaa
Aa
15
A
10 Baa
Ba
5
0
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
Source: Authors' correspondence with Moody's Investors
Furthermore…
How much sense does it make to think of a constant level of
“correct” subordination if real estate prices are highly cyclical?...
© 2014 OnCourse Learning. All Rights Reserved. 46
Cyclicality in U.S. Commercial Property Asset Market…
U.S. Commercial Real Estate Same-property Prices Superimposed on CPI less Real Depreciation:
2000 = 100
200
120
100
80
60
Dec-71
Dec-75
Dec-77
Dec-81
Dec-83
Dec-85
Dec-87
Dec-89
Dec-91
Dec-95
Dec-97
Dec-99
Dec-01
Dec-03
Dec-05
Dec-09
Dec-73
Dec-79
Dec-93
Dec-07
*(Sources: Moody's/REAL, TBI, Author's estimates.)
If loan origination underwriting is not counter-cyclical, i.e., not more conservative near
cycle peak (which it clearly is not), then mortgages are more risky near cycle peak.
47
Cyclicality in U.S. Commercial Property Asset Market…
U.S. Commercial & Multifamiliy Quarterly Mortgage Flows: 1978-2011
(Change in Balance Outstanding), & Commercial Property Prices
120 120
100
110
80
20 90
0
80
-20
-40
70
-60
-80 60
1978Q2
1980Q2
1982Q2
1984Q2
1986Q2
1988Q2
1990Q2
1992Q2
1994Q2
1996Q2
1998Q2
2000Q2
2002Q2
2004Q2
2006Q2
2008Q2
2010Q2
Mortg Capital Flow (Net Chg Bal) CRE Prices (TBI net infl, Right-Hand Axis)
If loan origination underwriting is not counter-cyclical, i.e., not more conservative near
cycle peak (which it clearly is not), then mortgages are more risky near cycle peak.
48
Conduit loan origination underwriting did not generally get more conservative as we approached the cycle
peak. Same LTV on peak property prices much more risky loan than same LTV on trough property prices.
(And these figures ignore possible loss of realism in the stated LTV & DCR numbers near the peak.)
Stated (pro-
forma)
underwriting
criteria of loan
originators
NOT more
strict during
bubble years.
120.0%
117.5%
CMBS 1.0 CMBS 2.0
110.0%
100.0% 95.2%
Loan-t 0-Value Rat io
90.0%
80.0%
73.3%
70.0% 61.9%
60.0%
50.0%
Source: Moody’s Investors Service, Trepp LLC, September 2011. © Moody’s Investors Service, Inc. &/or its
Affiliates. All rights reserved. Reprinted by permission.
Here note how the Moody’s haircut LTV rose alarmingly to very high levels in
2005-07. © 2014 OnCourse Learning. All Rights Reserved. 50
Even though credit rating agencies had their own “hair-cut” models of credit
risk in the mortgages, which showed deterioration in loan origination
underwriting during 2005-07, the agencies did not increase subordination
levels (credit protection) for the bond ratings…
Here note how the Moody’s haircut DSCR fell alarmingly to historically low
levels in 2005-07.
© 2014 OnCourse Learning. All Rights Reserved. 51
The pre-crash history of Rating Agency CMBS subordination levels…
20-7: History of CMBS Subordination Levels, 1995-2008 (pre-crisis)
35
30
25 Super Sr AAA
20 Aaa
Aa
15
A
10 Baa
Ba
5
0
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
Source: Authors' correspondence with Moody's Investors
Maybe subordination levels could have (correctly) been lower (less conservative) in the late-
1990s (perhaps even at 2007 levels then); but should have been higher by 2005 and later
as we entered the asset mkt bubble (maybe actual late-90s levels then). Could this have
been known at the time (without advantage of hindsight?). What pressures prevented a
more rational behavior?...
© 2014 OnCourse Learning. All Rights Reserved. 52
How the (bond) market prices the same amount of (rated) default risk (AAA) varies over time.
Downward pressure on spreads Upward pressure on rating agency subord. Requirments…
120
Exh.20-9: CMBS AAA Spreads To Swaps: 1996-2006
Prior to 2008 crisis,
Big event:
bond mkt yields
100 1998 fin crisis (prices) were not
(Russia, Asia, signaling fear, were
LTCM): putting downward
80
Liquidity shock, pressure on
quick recovery subordination
60 (except “Super-Srs”
demand was a hint).
40
20
0
7/5/1996
3/5/1997
7/5/1997
3/5/1998
7/5/1998
3/5/1999
7/5/1999
3/5/2000
7/5/2000
3/5/2001
7/5/2001
3/5/2002
7/5/2002
3/5/2003
7/5/2003
3/5/2004
7/5/2004
3/5/2005
7/5/2005
3/5/2006
7/5/2006
11/5/1996
11/5/1997
11/5/1998
11/5/1999
11/5/2000
11/5/2001
11/5/2002
11/5/2003
11/5/2004
11/5/2005
11/5/2006
Early history: Mkt Mature industry: RE boom after 2001 recession & stock
learning, spreads falling mkt tech bust: spreads falling (along with subord.)
1/5/2011
7/5/2010
1/5/2010
7/5/2009
1/5/2009
7/5/2008
1/5/2008
CMBS AAA Spreads To Swaps: 1996-2010
7/5/2007
1/5/2007
1/5/2005
compared to
7/5/2004
experience
1/5/2004
history &
2008-09
7/5/2003
nothing
was as
1/5/2003
7/5/2002
1/5/2002
7/5/2001
1/5/2001
7/5/2000
crisis
1998
1/5/2000
7/5/1999
1/5/1999
7/5/1998
Exh.20-10:
1/5/1998
7/5/1997
1/5/1997
7/5/1996
800
600
400
200
0
1800
1600
1400
1200
1000
“Moral Hazard”