Professor Anthony Saunders

1
The Secondary Market for Bank Loans
Professor Anthony Saunders Department of
Finance Stern School of Business New York
University
2
Two Research Questions

• Do bank loan announcements contain private
  information even when loans are traded?
• Can secondary market loan prices predict
  financial distress and bankruptcy?

3
Motivation

• Banks are considered special for several reasons,
  including reducing the agency costs of monitoring
  borrowers.
• First study to examine the influence of a
  secondary market for bank loans on
• bank specialness, and
• co-existence of banks and (secondary loan)
  markets.

4
Prior Research

• Bank specialness Positive loan announcement CAR
  in contrast to insignificant or negative response
  for other forms of financing.
• James (1987).
• Lummer and McConnell (1989).
• Best and Zhang (1993).
• Billett, Flannery, and Garfinkel (1995).
• James and Smith (2000) survey source of Table 1.
• The above studies use data from 1970s and 1980s
  (a well-developed secondary market for bank loans
  did not exist back then).

5
Table 1
6
Prior Research (Continued)

• Co-existence of banks and other markets (e.g.,
  stock markets)
• Stiglitz (1985).
• Shleifer and Vishny (1986).
• Rajan (1992).
• Allen (1993), Allen and Gale (1995, 1999).
• Dow and Gorton (1997).
• Levine (2002) survey.
• Our focus is on co-existence of banks and
  secondary loan markets.

7
Research Questions

• How does a secondary market for bank loans
  influence bank specialness? Important issue
• Figure 1 Ten-fold increase in secondary market
  loan transactions between 1991-2002.
• Unique opportunity to address whether such a
  market enhances or erodes the specialness of bank
  lending.
• Whether banks and (secondary loan) markets can
  co-exist?
• Are information sets of banks and secondary loan
  markets complements or substitutes for each
  other?
• Are certain types of borrowers most adversely
  affected?

8
Figure 1
9
Sample

• Sample period
• 1987-2003 (Dealscan)
• 1999-2003 (Secondary market loan data).
• Primary data sources
• LPC Dealscan.
• Secondary market loan data (unique dataset).
• CRSP daily stock returns and index returns.
• Compustat.

10
Research Question 1

• How does a secondary market for bank loans
  influence bank specialness?
• Hypothesis 1 A secondary market for loans ? bank
  specialness.
• Potentially reduces incentives of banks to
  monitor.
• Secondary market loan prices provide an
  alternative source of information on borrower
  creditworthiness.
• Hypothesis 2 Any loss of bank specialness
  impacts low-rated borrowers more than high-rated
  borrowers.
• Billett, Flannery and Garfinkel (1995).

11
Research Question 2

• Whether banks and (secondary loan) markets can
  co-exist?
• Hypothesis 3 Secondary market trading in loans
  is valuable to the equity investors of a
  borrower.
• Secondary market loan prices provide an
  additional source of information on borrower
  creditworthiness that may substitute or
  complement information gathered by banks through
  loan monitoring.
• We control for any liquidity effect resulting
  from the secondary market trading in loans.

12
Test methodology

• Standard event-study methodology outlined in
  Mikkelson and Partch (1986).
• Loan announcement date (day 0) Earliest date in
  Dealscan (active, launch, signing, closed dates).
• Estimation period (150 days) -200,-51.
• Univariate results.
• Linear regressions.
• Heteroskedasticity adjusted standard errors.
• Robustness tests with clustering (to be
  footnoted).

13
Hypothesis H1 Regression results

• Hypothesis H1 A secondary market for loans ?
  bank specialness.
• Day 0 loan announcement date.
• Dependent variable loan announcement -1,0 CAR.
• Control variables SDPE, BETA, RUNUP, MATURITY,
  LN(AMOUNT), SENIORITY, SECURED.
• Inference variable POST TRADE indicator
  variable.
• Inference variable ve significant Table 5 6.
• Positive loan announcement effect even when a
  borrowers loans trade on the secondary market.
• Post-trade CARs higher than pre-trade CARs.

14
H1 Univariate results (Table 2)
15
Hypothesis H1 (Table 5)
16
Hypothesis H1 (Table 6)
17
Hypothesis H2 Regression results

• Hypothesis H2 Any loss of bank specialness
  impacts low-rated borrowers more than high-rated
  borrowers.
• Day 0 Trading of a borrowers loans for the
  first time.
• Dependent variable loan trading date -1,0 CAR.
• Control variables SDPE, BETA, RUNUP, MATURITY,
  LN(AMOUNT), SENIORITY, SECURED.
• Inference variable DISTRESSED and POST TRADE x
  DISTRESSED variables.
• Inference variable ve significant Tables 7 8.
• Distressed borrowers are not adversely affected.

18
H2 Univariate results (Table 4)
19
Hypothesis H2 (Table 7)
20
Hypothesis H2 (Table 8)
21
Hypothesis H3 Regression results

• Hypothesis H3 Secondary market trading in loans
  is valuable to the equity investors of a borrower
  (Table 9).
• Day 0 Trading of a borrowers loans for the
  first time.
• Dependent variable loan trading date -1,0 CAR.
• Control variables SDPE, BETA, RUNUP, MATURITY,
  LN(AMOUNT), SENIORITY, SECURED, ROLL SPREAD Roll
  (1994), Stoll (2003).
• Inference variable TRADING FREQUENCY Lesmond,
  Ogden, and Trczinka (1999).
• Inference variable ve significant secondary
  market is complement rather than a substitute.

22
Hypothesis H3 (Table 9)
23

• Liquidity versus information
• Additional control in Table 9 ROLL SPREAD Roll
  (1994), Stoll (2003) as proxy for
  non-informational spread (liquidity).
• Inference variable TRADING FREQUENCY.
• Lesmond, Ogden, and Trczinka (1999).
• Inference variable ve significant Complement
  rather than a substitute.

24
Summary of results

• New loan announcements are associated with a
  positive announcement effect even when a
  borrowers loans trade on the secondary market.
• Distressed borrowers (for whom bank monitoring is
  likely to be most valuable) are not adversely
  affected by the presence of a secondary market
  for bank loans.
• loans trading for first time in secondary market
  positive announcement effect on distressed
  borrowers stock price.
• Banks and secondary loan markets are
  complementary sources of information.
• Result robust to the effects of increased
  liquidity of loans as a result of secondary
  market for trading in such loans.

25
Conclusions

• Banks continue to be special even in the presence
  of a secondary market for loans.
• Banks and markets can co-exist as information
  producers bank monitoring function and the
  secondary market for bank loans are complementary
  sources of information about borrowers.

26
Research questions

• H1 Whether loan prices, adjusted for risk, fall
  more than bond prices of the same borrower prior
  to an event, such as
• Loan default date.
• Bond default date.
• Bankruptcy date.
• H2 Whether loan prices, adjusted for risk, fall
  less than bond prices of the same borrower in
  periods surrounding the same event.
• Should be less of a surprise for loan investors.

27
Figure 2

• Loan Returns vs. Bond/Equity Prices

Return
Bond/Equity Prices
Loan Prices
0
Distress
Bankruptcy time
28
Main results

• We find evidence consistent with a monitoring
  advantage of loans over bonds prior to and in
  periods directly surrounding events, such as
  corporate (loan or bond) defaults and
  bankruptcies.
• Risk-adjusted loan prices fall more than
  risk-adjusted bond prices prior to an event.
• Risk-adjusted loan prices fall less than
  risk-adjusted bond prices in the periods directly
  surrounding an event.
• Robust to several alternative explanations.

29
Main results (continued)

• Alternative methodology Loan returns Granger
  cause bond returns for firms during periods
  leading up to a default on debt (loans or bonds)
  or bankruptcy.
• Vector-Auto Regression (VAR).
• Alternative dataset TRACE
• Results hold during a more recent time-period
  with improved dissemination of bond price
  information.
• Results also extend to loans versus stocks.

30
Data and sample selection

• Sample period 11/1/99-6/30/05.
• Loan price dataset unique dataset of daily bid
  and ask price quotes aggregated daily across
  dealers by Loan Syndications Trading
  Association (LSTA).
• Bond price dataset (a) Salomon (now Citigroup)
  Yield Book, (b) Datastream, (c) TRACE.
• Stock price dataset CRSP daily stock files.
• Loan defaults dataset SP. Missed interest or
  principal (not technical defaults on covenants).
• Bond defaults dataset NYU Salomon Centers
  Altman Bond Default Database.
• Bankruptcy dataset Lexis/Nexis and
  bankruptcydata.com.
• Loan and Bond characteristics datasets LPC, SP,
  Fixed Income Securities Database (FISD).

31
Test hypotheses

• H1 Loan prices fall more than bond prices prior
  to an event date.
• H2 Loan prices fall less than bond prices
  surrounding an event date.
• Events
• Loan default dates (will be presenting these
  results today).
• Bond defaults dates (reported in the paper).
• Bankruptcy dates (reported in the paper).

32
Univariate results Table 2 Panel A
33
Alternative explanations

• Univariate results do not control for
  security-specific characteristics such as
  difference in maturity and difference in issue
  size.
• We also consider several other alternative
  explanations, such as differences in
• Seniority, collateral, and recovery rates.
• Liquidity
• Covenants
• Timing of defaults
• Lender forbearance
• Alternative empirical methodology
  Granger-causality.

34
H1 (prior to an event) Table 5
35
H2 (surrounding an event) Table 6
36
Granger-Causality Tests

• Vector-auto regression (VAR) Follow Hotchkiss
  and Ronen (2002 RFS).

37
Granger-causality tests Table 7
38
Summary Granger-causality tests

• Loans are more efficient than bonds in reflecting
  information into prices around events, such as
  corporate (loan and bond) defaults, and
  bankruptcies
• Loan returns Granger cause bond returns.
• Bond returns do not Granger cause loan returns.
• Next TRACE

39
TRACE

• Similar results obtain during July 2002-June
  2005) that was characterized by an improved
  dissemination of bond prices
• H1 Statistically significant at 10 level.
• H2 Statistically significant at 5 level.
• Regression results qualitatively unchanged when
  augmented with data from this period.
• Next Loans versus stocks

40
Loans versus stocks Table 8
41
Additional robustness checks

• CAR related
• Mikkelson and Partch (1986) rather than Brown and
  Warner (1985).
• Holding period ARs versus CARs.
• Six different CARs versus market-model adjusted
  CAR.
• Use parameters from first six months of
  estimation window to compute pre-event CARs.
• Regression related
• Combining loan default, bond default, and
  bankruptcy into a single regression.
• Controlling for fixed effects or correcting
  standard errors for clustering effects.

42
Summary of results

• Loan market informationally more efficient than
  bond market prior to and in periods directly
  surrounding events, such as corporate (loan or
  bond) defaults and bankruptcies.
• Results robust to several alternative
  explanations.
• Alternative methodology Loan returns Granger
  cause bond returns for firms that defaulted on
  debt (loans or bonds) or went bankrupt.
• Vector-Auto Regression (VAR).
• Results also extend to loans versus stocks.

43
Conclusions

• Evidence consistent with a monitoring advantage
  of loan market over the bond market prior to and
  surrounding default and bankruptcy.
• Results have important implications for
• Continued specialness of banks.
• Benefits of loan monitoring for other financial
  markets (stocks and bonds).
• Benefits of including loans as a separate asset
  class in an investment portfolio.