A Model of CMBS Spreads

1
A Model of CMBS Spreads

• Joseph B Nichols and
• Amy Cunningham
• Federal Reserve Day Ahead Conference
• January 2, 2009

Note The analysis and conclusions expressed
herein are those of the author and do not
necessarily represent the views of the Board of
Governors of the Federal Reserve System and its
staff.
2
(No Transcript)
3
(No Transcript)
4
Why?

• CMBS has been an increasingly important source of
  financing for CRE.
• Market has seen several significant disruptions.
• 9/11 attacks.
• Subprime crisis.
• What has been the effect of these disruptions on
  pricing?

5
What

• CMBS market is still evolving and changing.
• CMBS pools are very heterogeneous.
• Pool composition and credit quality have changed.
• Some changes in response to shocks.
• Increase use of fusion pools after 9/11.
• Increase in credit support after subprime crisis.

6
(No Transcript)
7
How to Model Effects

• Changes in composition raises challenge for use
  of on-the-run spread models.
• Address that by using spreads on new issuance.
• Allows us to control for change in composition
  both across time and cross-sectionally.
• Measure premium paid for higher credit support or
  higher risk exposure.

8
9/11

• After 9/11 there was an significant jump in the
  perceived risk exposure for large commercial
  buildings.
• The existing levels of terrorism insurance was
  viewed as insufficient and too costly.
• GMAC and the Mall of America.
• Concerns about impact on construction.
• Increased use of fusion pools in CMBS.

9
The Terrorism Risk Insurance Act of 2002

• Passed November 11, 2002.
• Federal Government will act as reinsurer.
• Triggered by terrorism attack where the damage
  exceeds a given threshold (currently 100
  million).
• No premium payments prior to attack, payments
  assigned after attack.
• Designed to be temporary until market for
  terrorism insurance develops.
• Renewed in 2005 and in 2007 for seven years.
• Extended to cover domestic acts and nuclear,
  biological, chemical, and radiological attacks.

10
Subprime and CMBS Part 1

• CMBS spreads started to widen in response to
  subprime crisis in late February, 2007.
• Initial response was due to fall in demand from
  CDOs, which had been significant buyers of BBB
  CMBS.
• In April, Moodys announced higher levels of
  credit support to address ongoing concerns
  regarding higher leverage and looser
  underwriting.
• The market pushed back, eight of the first twelve
  CMBS pools issued in Q3 did not have Moodys
  ratings.
• Originations for CMBS peaked in June, and started
  to slow.
• Conditions worsened in August.
• Market largely shut down in Fall.
• Credit support rising and leverage falling for
  new issuance since fall.

11
Subprime and CMBS Part 2

• CMBS saw significant increases in leverage and
  looser underwriting in 2005-2007.
• Increased demand from private equity for LBOs of
  REITS.
• Increased use of pro forma instead of historical
  rents.
• Increase in interest-only loans.
• CMBS differs significantly from Subprime MBS.
• Investors can examine loan and property level
  data for entire pool.
• Collateral is income-generating properties.
• No oversupply of properties.

12
Literature

• Brown, Kroszner, and Jenn (2002)
• Impact of milestones leading to passage of TRIA
  on stock prices of affected industries.
• Results show largely negative impact.
• Deng, Gabriel, and Sanders (2008)
• Model of on-the-run CMBS spreads
• Impact of demand from CDOs
• Maris and Segal (2002)
• Model of spreads of new CMBS issuance.
• Used data through 1999.
• Starting point for our model.

13
Hypothesis

• BBB CMBS spreads for pools containing large loans
  should be responsive to changes in perceived
  terrorism risk.
• Investors differentiate between similarly rated
  tranches based on the levels of credit support
  and pool composition.
• The premium demanded for lower credit support and
  higher default risk started to rise before the
  onset of the subprime crisis.

14
Data

• Generate models using both on-the-run spreads and
  spreads on new issuance.
• On-the-run spreads from Morgan Stanley, weekly.
• Spreads on new issuance, and pool
  characteristics, CMAlert.
• Use junior AAA tranche.
• Convert spreads to be on 10-year treasury.
• Control for autocorrelation by including lagged
  term.

15
Independent Variables
16
(No Transcript)
17
Results from On-The-Run Models

• Sign on corporate credit spread opposite from
  Maris and Segal results.
• Our sample includes Enron scandal period.
• CMBS was used as substitute for corporate bonds
  during that period.
• Volatility measures act as risk proxies.
• Despite declines in nominal spreads, model shows
  increase in BBB spreads prior to onset of
  subprime crisis.

18
(No Transcript)
19
(No Transcript)
20
Results from AAA Spreads on New Issuance

• Results from macro variables consistent with
  on-the-run models.
• Signs for tranche and pool amount consistent with
  Maris and Segal.
• Larger tranche increased liquidity.
• Larger pool difficulty absorbing increased
  supply.
• Little results from pool characteristics.
• AAA tranche largely insulated from credit risks.

21
(No Transcript)
22
(No Transcript)
23
Results from BBB Spreads on New Issuance

• Results from macro variables consistent with
  on-the-run model.
• Higher LTV on loans or lower subordination rates
  results in higher spreads.
• Controlling for pool composition, BBB spreads
  fell 16 bps after TRIA passed.
• Premium for lower subordination rates started to
  rise prior to onset of subprime crisis.

24
Conclusions

• CMBS market responded to shocks by changing pool
  composition and levels of credit support..
• TRIA reduced premium paid on pools with large
  loans.
• Result of transfer of risk to government or
  widespread adoption of fusion pools.
• Premium paid for lower credit quality and credit
  support starter to rise prior to onset of
  subprime crisis.
• Reflecting pre-existing concern regarding
  weakening underwriting standards.