Transcript Sovereign Debt Rating and Stock Liquidity around the World

Sovereign debt rating
• Sovereign debt ratings are assessments of the
probability of default in government bonds.
– Rating agencies consider economic and political factors
and make a qualitative and quantitative evaluation.
• By doing so, sovereign debt rating changes deliver
new information about a country, as evidenced by
their significant effects on:
– Stock market returns:
• Kaminsky and Schmukler (2002), Brooks, Faff, Hiller, and Hiller (2004),
Martell (2005)
– Individual stock returns:
• Martell (2005), Lee, Sapriza, and Wu (2010)
– Bond yields/spreads:
• Cantor and Packer (1996), Larrain, Reisen, and von Maltzan (1997),
Gande and Parsley (2005)
2
This paper is about …
• However, there are no studies investigating the
effect of sovereign debt rating changes on stock
liquidity. Our paper fills in that gap.
– We study the impact of changes in sovereign credit ratings
on daily stock liquidity for 40 developed and emerging
markets from January 1990 to December 2009.
– Stock-level analysis: Cross-sectional variation of impact
– Cross-country analysis: differences in response over
countries
– Regression approach rather than event-study framework
(Gande and Parsley, 2005)
• The importance of liquidity has been growing
rapidly both for researchers and for practitioners.
3
Why is liquidity important?
“Whenever the market turns against you, you take the biggest losses in
illiquid securities,” says Richard Bookstaber, former head of risk
management at Salomon Bros. “Because there are so few buyers,
you’re forced to sell at a discount that is both huge and highly
unpredictable.”
(Fortune, November 26, 2007)
• Liquidity affects asset prices.
•
Amihud and Mendelson 1986, Brennan et al 1996, Amihud 2002, Pastor
and Stambaugh 2003, Acharya and Pedersen 2005, Lee 2011
• Liquidity is systematic, and its commonality may lead to
the fragility of financial markets.
•
Brunnermier and Pedersen 2009, Hameed, Kang, and Viswanathan 2010,
Karolyi, Lee, and van Dijk 2010
• Liquidity affects market efficiency.
•
Chordia et al 2008
4
How could rating changes be related to liquidity?
• Sovereign debt ratings impose a significant externality to
the country’s private sector, which can affect investors’
incentives to hold stocks. Especially, sovereign
downgrades may lead to:
– Capital outflows (Reinhart and Rogoff, 2004; Kim and Wu, 2008;
Gande and Parsley, 2010), which may be accompanied by fire
sales of assets, thus, liquidity constraint (Pastor and Stambaugh,
2003; Acharya and Pedersen, 2005)
• Low ratings are related to high adverse selection, thus to
low liquidity in equity (Odders-White and Ready, 2006).
– In times of market stress, bond investors chase liquidity (flight-toliquidity), not credit quality (Beber, Brandt, and Kavajecz, 2009)
– Low ratings are also related to low liquidity in bonds (Cantor and
Packer, 1996)
5
How rating changes could be related to liquidity?
• Sovereign downgrades may lead to limited access to
international capital markets (Dittmar and Yuan, 2008),
– which, in turn, limits liquidity due to limited diffusion of information
in the global capital markets (high information asymmetry).
• Sovereign ceiling lite (Ferri , Lui, and Majnoni, 2001; Ferri and Liu, 2002;
Kaminsky and Schmukler, 2002; Borensztein, Cowan, and Valenzuela, 2007)
• More information asymmetry at the time of earnings announcement (Kim and
Verrecchia, 1994)
– Funding constraints of traders (e.g., hedge funds, banks, marketmakers) affect, and are affected by, market liquidity:
• Intermediaries make markets by absorbing liquidity shocks subject to funding
constraints (via posted margins on collateral). When markets decline and if
funding constraints bind, they endure loss in collateral values, margin limits are
hit, thus forcing them to liquidate positions. One trader’s hitting his limit in one
security leads to falling prices and greater illiquidity in other securities and
makes other traders hit their respective limits; “liquidity spirals” or “liquidity
black holes” arise.
• Brunnermeier and Pedersen (2009), Gromb and Vayanos (2002), Kyle and Xiong
(2001), Morris and Shin (2004), Hameed, Kang, and Viswanathan (2010),
Karolyi, Lee, and van Dijk (2010)
6
Outline
• Data and sovereign debt ratings
• Liquidity measure
– Filtering seasonal effect of liquidity
• Empirical results
–
–
–
–
Rating changes and their impact on liquidity
Asymmetric impact of rating changes on liquidity
Nonlinear impact of rating changes on liquidity
Cross-sectional differences of effects on stock
liquidity
– Cross-country differences of effects on stock
liquidity
• Conclusion
7
Data
• Sovereign rating data from S&P
– S&P rating is more active and tends to precede other rating agencies
(Brooks et al., 2004; Gande and Parsley, 2005)
– Total 424 event days
• Datastream/Worldscope for 1990-2009 from 40 countries
(26 emerging and 14 developed markets)
– Including dead-stocks (free of survivorship bias)
– Require:
• only common shares listed in major exchanges
• available total return index, trading volume, market value of equity, and B/M
– Exclude:
• if daily return or trading volume is at 1% extreme in a given country
• non-trading days
• observations for non-event days
– Final Sample: 67,518 stock-event days
8
Numeric conversion of ratings
• Event=absolute value of changes in ratings
• E.g. Brazil: BB+/Positive (June 12, 2007), BBB-/Stable (Apr 30, 2008) 9
Table 1.
Number of
events by
country
Event+:
upgrade
Event-:
downgrade
Event Big+:
upgrade by more
than 2 notches
Event Big-:
downgrade by
more than 2
notches
10
Event clustering and regression framework
• Many previous studies perform event-study analyses.
– Cantor and Packer (1996), Brooks et al. (2004), Martell
(2005)
• However, clustering of events may contaminate eventwindows.
– E.g. Argentina: 2001/10/9 (CCC+/Neg), 2001/10/30
(CC/Neg), 2001/11/6 (SD/NM=Not meaningful)
• Hence, following Kaminsky and Schmukler (2002)
and Gande and Parsley (2005), we adopt a regression
framework:
Change in liquidityi,t = a + b*Eventi,t + ei,t
11
Liquidity measure and its seasonality
• Illiquidity measure from Amihud (2002)
– Popular price impact measure:
• Hasbrouck (2005), Goyenko et al. (2009)
– Can be computed at daily frequency.
– By taking log, we use:
• Illiquidity has day-of-the-week effect and seasonality
with calendar month.
– Chordia et al. (2005), Hameed et al. (2010)
– Hence, we run the following filtering regression:
By taking first-differences of residuals,
12
Table 2. Filtering regression
13
Table 3.
Descriptive
statistics
14
Hypotheses
A. If sovereign rating changes deliver new information to the
market, investors will repackage their portfolio, hence stock
liquidity will drop on event.
B. The effect of sovereign rating changes will be asymmetric:
downgrades will have a larger impact than upgrades.
• Reluctant to downgrade, hence downgrading delivers more information
-> larger effect on portfolio rebalancing needs
• Downgrades limit access to international financial market, increasing
information asymmetry
• Downgrades may lead to a higher level of funding constraints
C. The effect of sovereign rating changes will be non-linear:
large downgrades will have a larger impact than small
downgrades.
D. The effect of sovereign rating changes will be non-linear:
rating changes from investment grade to non-investable
grade (or vice versa) will have a larger impact than rating
changes not involving investment grade changes.
15
Rating changes and stock liquidity (Table 4)
Country dummy: Yes, Year dummy: Yes
Event=absolute value of changes in ratings
16
Asymmetric effect on stock liquidity (Table 4)
• Event(+)=positive changes, zero otherwise
• Event(-)=absolute value of downgrade, zero otherwise
Country dummy: Yes, Year dummy: Yes
17
Nonlinear effect on stock liquidity (Table 5)
• Big+ (-) Event=absolute value of changes in ratings by more than 2
notches upward (downward), zero otherwise
• Small+ (-) Event=absolute value of changes in ratings by less than 2
notches upward (downward), zero otherwise
Country dummy: Yes, Year dummy: Yes
18
Investable -> Non-investable or vice versa (Table 5)
• Event(Inv->Non-inv)=1 if the change is from investable to noninvestable rating, and zero otherwise
• Event(Non-inv->Inv)=1 if the change is from non-investable to
investable rating, and zero otherwise
19
Country dummy: Yes, Year dummy: Yes
Reinhart, Carmen, and Kenneth Rogoff, 2004. "Serial Default and the ‘Paradox’ of Rich-to-Poor Capital Flows," American Economic Review, Papers and Proceed
Cross-sectional differences in response
Firms with the following characteristics will have
smaller impact on events:
–
Politically-connected
•
–
Politically-connected firms are more likely to be bailed out (Faccio,
2006)
Large-cap; High transparency
•
–
Information effect by rating changes will be smaller
Low book-to-market, High profitability
•
–
Less likely to default, hence smaller incentive to rebalance
Not closely-held
•
–
More free-floating shares, hence easier to trade
Liquid
•
Less liquidity constraint
20
Crosssectional
differences
in response
(Table 6)
Controls:
Global market returns
log(MV)
log(B/M)
Volatility
Country dummies
Year dummies
21
Cross-country analyses
Stocks from the following type of country will experience a
smaller impact (ie., less commonality in liquidity) on events:
• High transparency
•
•
High investor protection
•
•
More incentive to trade based on firm-specific information (Jin and
Myers, 2006)
Well-developed stock market, high GDP
•
•
Revision in expectations from new information delivered by rating
changes will be smaller either due to smaller sensitivity of
expectations to new information in transparent setting or due to
smaller amount of new information delivered by events.
Less liquidity constraint or less information asymmetry
Low/High (?) foreign institutional ownership
•
•
Low – Less tendency to herd
High – Less closely-held shares (more free floating shares)
(Dahlquist, Pinkowitz, Stulz, and Williamson, 2003)
22
Cross-country analyses: Variables
• Investor protection
– LAW=1 (civil law), =0 (common law) (La Porta et al. 1998)
– ANTIDIRR = a higher numbers indicating stronger minority investor
protection (La Porta et al. 1998)
– EXPRISK: Higher values indicating smaller political risk (La Porta et al.
1998)
• Financial market development
– SCAP = Average of “stock market cap / GDP” for 1988-1999 (Stulz, 2005)
– GDP per capita = in US dollars, as of 2003 (world development indicator)
– Foreign institutional ownership (as % of market cap)
• Transparency
– ACCSTAND = Higher values indicating higher accounting standards (La
Porta et al. 1998)
– DISCL = Credibility of financial disclosure
– EARNMGT = Lower values indicating lower level of earnings management
(Leuz et al. 2003)
– NANALYSTS = No. of analysts who covered the largest 30 firms in a 23
country
Crosscountry
differences
in
response
(Table 7)
Controls:
Global market
returns
log(MV)
log(B/M)
Volatility
Country dummies
Year dummies
24
Conclusion
• We study the impact of sovereign rating changes on
stock liquidity for 40 countries for 1990-2009.
• Our findings show that sovereign rating changes have
a significant and robust impact on stock liquidity.
– The impact is nonlinear and asymmetric and varies across
stocks and countries.
– In a cross-section, firms with no political connection,
smaller size, higher book-to-market ratio, closely-held
ownership, less liquidity, less profitability, and less
transparency tend to experience more negative liquidity
effects from sovereign debt downgrades.
– Stocks from countries with higher investor protection, welldeveloped stock markets, more transparency, higher foreign
institutional ownership, and higher GDP tend to experience
milder drops in stock liquidity – less commonality in
liquidity.
25