Liquidity Effects in the U.S. Corporate Bond Market€¦ · - Required rate of return adjusts to...
Transcript of Liquidity Effects in the U.S. Corporate Bond Market€¦ · - Required rate of return adjusts to...
Liquidity Effects in the U.S. Corporate Bond Market
Marti G. SubrahmanyamStern School of Business
New York University
For presentation at the Centre for Advanced Financial Research and Learning
Reserve Bank of India January 17, 2014 1
Outline
• Why are liquidity effects important in asset pricing?
• Theory on liquidity effects in asset prices: The current literature.
• Empirical evidence on liquidity effects in asset prices: The current literature.
• Liquidity effects in illiquid over-the-counter markets: Problems of measurement.
• Special features of over-the-counter markets.• Characteristics of the U.S. corporate bond
market.
2
Outline (Contd.)• Liquidity effects in illiquid over-the-counter markets: The
US corporate bond market."Latent Liquidity: A New Measure of Liquidity with an Application to Corporate Bonds," (with G. Chacko, S. Mahanti, G. Mallik and A. Nashikkar), Journal of Financial Economics, May 2008.
"Limited arbitrage and liquidity in the market for credit risk," (with S. Mahanti and A. Nashikkar), Journal of Financial and Quantitative Analysis, June 2011.
"Price Dispersion in OTC Markets: A New Measure of Liquidity," (with R. Jankowitsch and A. Nashikkar), Journal of Banking and Finance, February 2011.
"Illiquidity or Credit Deterioration: A Study of Liquidity in the US Corporate Bond Market during Financial Crises,“ (with R. Jankowitsch and N. Friewald), Journal of Financial Economics, July 2012. 3
Outline (Contd.)
"Liquidity, Transparency and Disclosure in the Securitized Product Market," (with N. Friewald and R. Jankowitch).
"The Determinants of Recovery Rates in the US Corporate Bond Market," (with R. Jankowitsch and F. Nagler), Journal of Financial Economics, forthcoming.
• Conclusions and issues for discussion.
4
Why are liquidity effects important? • Assets with similar risk characteristics have
different expected returns.• One candidate for missing factor – liquidity.• Liquidity differentials may explain differences in
return performance for the same level of risk.• Liquidity raises trading volume and reduces the
cost of capital – even small improvements reduce the cost of capital substantially.
• Examples: Long term investors such as insurance companies buy illiquid assets. Hedge funds buy illiquid assets and sell liquid ones–“off-the-run” versus “on-the-run.” 5
Theory on liquidity effects: The current literature
• Reasons for costs of illiquidity (or liquidity premium or decrease in price):- Adjustment for the present value of future
transaction costs including, spreads, trading costs market impact and asymmetric information, e.g. Amihud and Mendelson (1986) and several others.
- Required rate of return adjusts to compensate investors for the level of illiquidity and the risk of illiquidity, e.g. Mayers (1972), Pastor and Stambaugh (2002) and Acharya and Pedersen (2005).
- Liquidity has an option value due to the possibility of selling the asset when necessary, e.g. Copeland and Galai (1983) and Longstaff (1995).
6
Theory on liquidity effects: The current literature (Contd.)
• Distinction between level and risk of illiquidity.• Are these effects priced? How much?• Notice that all these arguments assume that the
marginal trader has a long position.• What happens if investors are allowed to go
short and if the marginal trader has a short position?
• Implicit assumption that the asset is in positive net supply.
• No statement about the relationship between the asset and other positions of the agent. Is it a hedge?
7
Empirical evidence on liquidity effects in asset prices
• Some representative empirical results:- Amihud and Mendelson (1986): a 0.5% increase in
spread annual return to decrease by 1-2%. - Brennan, Chordia and Subrahmanyam (1998): a
one standard deviation increase in volume decrease in annual excess returns by 1.3-3.5%.
- Acharya and Pedersen (2005): annual risk premium for liquidity level is 3.5% and for liquidity risk is 1.1%.
- Silber (1991): mean discount of 34% for restricted stock relative to traded counterparts.
- Nashikkar, Subrahmanyam and Mahanti (2011): a one standard deviation change in liquidity changes yields and the basis by 10 bps. 8
Liquidity effects in highly illiquid markets
• Problem with most liquidity metrics is that they are transaction-based measures, based on prices or volumes.
• Applicable in the more liquid markets, e.g. equities, foreign exchange, some treasury bonds.
• Many asset markets, such as the U.S. corporate bond market, are too illiquid to permit measures such as bid-offer spread, depth, trading volume or even the Amihud measure of market impact (Kyle’s ).
• Classic case of “looking for lost keys under the lamp-post rather than where they were lost.”
9
Special features of over-the-counter markets
• Importance of over-the-counter (OTC) markets: Real estate, bond (Treasury and corporate), most new derivative markets etc.
• Microstructure of OTC markets is different from exchange-traded (ET) markets.
• Lack of a centralized trading platform: Trades are result of bilateral negotiations take place at different prices at the same time.
• Search costs for investors and inventory costs for broker-dealers (and information asymmetry).
• Challenges of assembling market-wide data.• Important issues of illiquidity, in crises such as
the present credit crisis.10
Characteristics of the U.S. corporate bond market.
• Large number of corporate bonds: over 20,000 with some activity and many more that are completely illiquid. (Over a million different fixed income instruments.)
• Poor liquidity in most of them, except for a few hundred bonds.
• Transaction data (TRACE) available from 2004 onwards with good coverage. Other data from Markit, S&P, Moody’s, Bloomberg etc.
• New data base on structured products to be made available from 2011 onwards from the TRACE platform.
• Better data than most other bond markets, internationally. 11
Latent Liquidity Paper• Liquidity effects in illiquid over-the-counter markets: The
US corporate bond market."Latent Liquidity: A New Measure of Liquidity with an Application to Corporate Bonds," (with G. Chacko, S. Mahanti, G. Mallik and A. Nashikkar), Journal of Financial Economics, May 2008.
"Limited arbitrage and liquidity in the market for credit risk," (with S. Mahanti and A. Nashikkar), Journal of Financial and Quantitative Analysis, June 2011.
"Price Dispersion in OTC Markets: A New Measure of Liquidity," (with R. Jankowitsch and A. Nashikkar), Journal of Banking and Finance, February 2011.
"Illiquidity or Credit Deterioration: A Study of Liquidity in the US Corporate Bond Market during Financial Crises,“ (with R. Jankowitsch and N. Friewald), Journal of Financial Economics, July 2012. 12
Latent Liquidity Paper (Contd.)
"Liquidity, Transparency and Disclosure in the Securitized Product Market," (with N. Friewald and R. Jankowitch), Working Paper, December 2012."The Determinants of Recovery Rates in the US Corporate Bond Market," (with R. Jankowitsch and F. Nagler), Journal of Financial Economics, forthcoming.
13
14
Theoretical basis for Latent Liquidity• Amihud and Mendelson (1986) – In equilibrium,
assets with the highest transaction costs are held by investors with the lowest trading frequencies (longest trading horizons) which enables them to amortize these costs.
• Vayanos and Wang (2005), Duffie, Garleanu and Pedersen (2003) – Conversely, higher turnover by agents of particular assets leads to lower search costs, and hence, greater demand to hold these assets.
• Turnover of agents weighted by holdings should be a natural measure of potential liquidity.
Latent Liquidity as a Measure of Liquidity
• US corporate bond market is very illiquid:Median bond trades less than once a year.Only a couple of hundred bonds trade every day. Micro-structure based measures infeasible.
• New measure proposed by Mahanti, Nashikkar, Subrahmanyam, Chacko and Mallik (2008) (MNSCM) based on accessibility of a bond.
• Latent liquidity is defined as the weighted average turnover of investors who hold a bond, where the weights are the fractional investor holdings.
15
The Database
• Custodial database of State Street Corporation, one of the world’s largest
• Database comparable to the universe (Tables 1, 2 and 3)
• Frequency of trading (Table 5)
16
Frequency of Trading
17
18
Definition of Latent Liquidity
• is the fractional holding of bond i by fund j at the end of month t.
• is the average portfolio turnover of fund j from month t to month
• We show that latent liquidity- has predictive power for transaction costs and
price impact over and above bondcharacteristics and trade-based measures.
- exhibits relationships to bond characteristics similar to trade-based measures.
Latent Liquidity Paper: Main Insights
• Latent liquidity does not rely on transaction information.
• It provides a good indication of liquidity: better predictor than volume or bond characteristics of transaction costs or price impact (Amihud).
• Difference between highest and lowest percentiles: 200 bps unconditionally and 91 basis points conditionally.
19
Latent Liquidity and Price Impact
20
Bond Basis Paper• Liquidity effects in illiquid over-the-counter
markets: The US corporate bond market."Latent Liquidity: A New Measure of Liquidity with an Application to Corporate Bonds," (with G. Chacko, S. Mahanti, G. Mallik and A. Nashikkar), Journal of Financial Economics, May 2008.
"Limited arbitrage and liquidity in the market for credit risk," (with S. Mahanti and A. Nashikkar), Journal of Financial and Quantitative Analysis, June 2011.
"Price Dispersion in OTC Markets: A New Measure of Liquidity," (with R. Jankowitsch and A. Nashikkar), Journal of Banking and Finance, February 2011.
"Illiquidity or Credit Deterioration: A Study of Liquidity in the US Corporate Bond Market during Financial Crises,“ (with R. Jankowitsch and N. Friewald), Journal of Financial Economics, July 2012.
21
Bond Basis Paper (Contd.)
"Liquidity, Transparency and Disclosure in the Securitized Product Market," (with N. Friewald and R. Jankowitch), Working Paper, December 2012."The Determinants of Recovery Rates in the US Corporate Bond Market," (with R. Jankowitsch and F. Nagler), Journal of Financial Economics, forthcoming.
22
23
Outline
• What factors affect the CDS-Bond basis? • Latent liquidity as a measure of liquidity• Computation of CDS-bond basis : the
methodology• Definition and computation of the basis• Data for the present study
24
Outline (Contd.)• The main features of the study• Determinants of the Credit Default Swap (CDS) -
corporate bond basis: – Non-default related
• Liquidity – bond and CDS market• Other bond characteristics, e.g.. tax status
– Default-related • Shorting costs, present and anticipated• Firm level variables, e.g.. financial variables, covenants• CDS contract definition, including cheapest-to-deliver option
25
Outline (Contd.)
• Univariate and multivariate results• Effects of bond covenants• Interactive effects• Conclusions and issues for further
research
26
Definition of the Basis
• Definition of the Basis
Basis = CDSactual - CDShypothetical
• CDSactual is the actual mid-price of the spread• CDShypothetical is the spread on a hypothetical
CDS contract at par that has the same default risk and recovery rate as implied by the price of a risky corporate bond issued by the same issuer.
27
Computation of CDS-Bond Basis: The Methodology
• Use corporate bond prices, under the assumption of a flat term structure of default, and constant recovery rate to compute default probability.
• Use default probability to compute spread of a hypothetical par-equivalent CDS contract.
• Basis is the difference between this theoretical spread and the actual CDS spread.
• Under ideal conditions (frictionless markets, no CDS optionalities, no covenants, etc.) the basis should be zero.
28
What Affects the CDS-Bond Basis?• “Non-default related” :
- Bond market liquidity (and bond characteristics)More liquid bonds - higher basis
- CDS market liquidityBonds with liquid CDS contracts - higher basis
- Other effects e.g.. taxesTax differential between benchmark and bonds
• “Default-related”: - Anticipated shorting costs
Higher default probability, shorting cost-higher basis- Covenant protection
Relative seniority, other options for issuer/investor- CDS contract definition
Protection level, cheapest-to-deliver
29
Computation of the CDShypothetical
• Assumptions- Defaults only on coupon dates- Constant default intensity, - Constant recovery of proportion of FV
• Solve for from bond price• Compute the par equivalent CDS price which is
CDShypothetical• Basis is the difference between this and the
actual CDS prices• The greater the liquidity, the more positive the
basis.
30
Data for the Study
• Corporate bond prices (Volume weighted average, from TRACE)
• Corporate bond volumes (TRACE)• Latent liquidity (Custodial holdings of SSC)• CDS prices (GFI/CMA Datavision)• Firm level financial variables (Compustat)• Bond covenants (Bloomberg)• Sample broadly representative of the market
Univariate Pooled Regressions
31
Multivariate Pooled Regressions
32
33
The Effect of Bond Covenants on the Basis
Covenant Number of bonds Definition Expected effect on basisCredit
Sensitive 38The bond’s coupon is linked to credit rating
Positive (Greater protection)
Defeased 3323
The bond has an option to be defeased in the future, i.e. the issuer may set aside funds for the bondholders and free himself from covenants
Positive/Negative (Issuer has the option to defease, defeasement itself increases basis)
Equity Clawback 39
The issuer has an option to redeem the security using equity proceeds
Positive (Greater protection)
Eusd Tax 3611The bond is subject to the E.U. Savings tax directive applying to bonds issued after July 2001
Negative (Bonds are older and attract higher tax rates)
Exch Listed 666 Is Exchange listed Positive (More liquidity and oversight)
Fund Terms 84The issuer has the obligation to redeem the bond in part before maturity
Positive (Greater protection)
Guarantee 626 Is Guaranteed Positive (Greater protection)
Poison Put 192Indicates the presence of a put option by investors or a call option by the issuer in case of ownership change
Positive (Greater protection)
Prin Idx Lnk 103Indicates if the principal amount fluctuates with a market index
Unclear
Prosp On File 5203 The prospectus is on file with the SEC Positive (Higher liquidity and oversight)Secured 253 The bond is secured by collateral Positive (Greater protection)
Spec Put 1623The presence of a special put option in case of a corporate event like a spinoff or an asset sale
Positive (Greater protection)
Tap Issue 1479 The bond may be reissued Unclear
Tax Call 700The security may be called if there are changes in tax laws
Positive/Negative (The option may or may not benefit the issuer at the expense of the investors)
Interaction between Liquidity and Credit Rating
34
35
Conclusions and Issues for Further
Research
• The CDS-bond basis includes default-related components, in addition to a non-default component
• Previous work by MNSCM shows that latent liquidity has explanatory power for transaction costs and price impact, over and above other bond characteristics.
• This paper shows that latent liquidity is priced –explains part of the CDS-bond basis over and above other bond characteristics e.g.., age, coupon, rating, and number of trades.
36
Conclusions and Issues for Further Research (Contd.)
• CDS market liquidity also affects the CDS-bondbasis.
• Shorting costs proxied by firm-specific variables such as leverage, as well as covenants explain part of the CDS-bond basis.
• The CDS contract definition including the level of protection and the cheapest-to-deliver option also explains part of the CDS-bond basis.
37
Conclusions and Issues for Further Research (Contd.)
• Bond covenants that reduce risk and hence increase bond price, also lead to a higher basis.
• There are interactive effects between credit risk and shorting costs that may make the relationship between liquidity and the basis dependent on these variables.
• It would be interesting to study these effects in periods of crisis as in the last six months.
Bond Basis Paper: Main Insights
• Liquidity effects explain the bond basis.• Latent liquidity is better than transaction
based measures of liquidity in explaining the basis.
• Asymmetric nature of liquidity effect due to difficulty in shorting the bond.
• The CDS contract does not fully take into account credit risk – residual effects.
• The basis is related to the existence of covenants.
38
Price Dispersion Paper• Liquidity effects in illiquid over-the-counter
markets: The US corporate bond market."Latent Liquidity: A New Measure of Liquidity with an Application to Corporate Bonds," (with G. Chacko, S. Mahanti, G. Mallik and A. Nashikkar), Journal of Financial Economics, May 2008.
"Limited arbitrage and liquidity in the market for credit risk," (with S. Mahanti and A. Nashikkar), Journal of Financial and Quantitative Analysis, June 2011.
"Price Dispersion in OTC Markets: A New Measure of Liquidity," (with R. Jankowitsch and A. Nashikkar), Journal of Banking and Finance, February 2011.
"Illiquidity or Credit Deterioration: A Study of Liquidity in the US Corporate Bond Market during Financial Crises,“ (with R. Jankowitsch and N. Friewald), Journal of Financial Economics, July 2012. 39
Price Dispersion Paper (Contd.)
"Liquidity, Transparency and Disclosure in the Securitized Product Market," (with N. Friewald and R. Jankowitch), Working Paper, December 2012."The Determinants of Recovery Rates in the US Corporate Bond Market," (with R. Jankowitsch and F. Nagler), Journal of Financial Economics, forthcoming.
40
41
Price Dispersion in OTC Markets: A New Measure of Liquidity
• In the presence of search costs for traders and inventory costs for dealers: how are prices determined in an OTC market?
• What determines price dispersion effects, i.e., deviations between the transaction prices and their relevant market-wide valuation?
• How does price dispersion capture illiquidity in such markets?
• How is the “hit rate” – the proportion of transactions within the average quoted bid-askspread – related to illiquidity?
41
42
Market Microstructure Model• i assets, i = 1,2…I, and a continuum of dealers,
j, = 1,2…J.• Competitive dealers face inventory costs and
quote bid-ask prices depending on their desired inventory levels.
• Investors, with exogenously given buying and selling needs, trade with the dealers.
• Investors directly contact dealers to obtain price quotes (“telephone market”).
• Investors face search costs every time they contact a dealer.
42
43
The Dealer’s Decision• Denote by si,j the inventory of asset i with dealer of type j.• Each dealer faces inventory holding costs H that are
convex in the absolute quantity held, given by H = H(s).Independent across assets.
• The marginal holding cost of adding a unit is approximated by h = H’(s).
• Each trade incurs a marginal transaction cost function faand fb
• The ask price of asset i quoted by dealer j is denoted as pa
i,j , the bid price pbi,j, for one additional unit.
• Since the dealership market is competitive: pai,j = mi,j +
fa(h(si,j))and pb
i,j = mi,j – fb(h(si,j)) .• The market’s expectation of the price of asset i is defined
by mi = E(mi,j). 43
44
The Investor’s Decision• An investor wishes to execute a buy-trade of one
(infinitesimal) unit.• The investor has contact with one dealer and is offered an
ask price pa,0.• The investor faces search cost c for contacting an
additional dealer; thus, she evaluates the marginal cost and benefit of doing so.
• Garbade and Silber (1976) show that the investor will buy the asset at pa,0 if this price is lower than his reservation price pa*.
• The reservation price solves:
where ga(.) is the density function for the ask price when contacting an arbitrary dealer.
*
0
* )()(ap aa dxxgxpc
44
45
Price Dispersion and “Hit Rate”• Assumption for inventory holding distribution:
– Uniformly distributed with mean zero (zero net supply)– Support from –S to +S– Independent across assets
• Assumption for the holding costs: H = s2 s/2
• Assumption for the transaction cost: h = S/2
• Solving for the reservation prices for a trader gives:pa* = m + (2c s)0.5
pb* = m - (2c s)0.5
• Ask and bid prices, when contacting a dealer are uniformly distributed with supports [m; m+ s] and [m; m- s] 45
Graphical depiction of solution –zero net inventory
mPbl= m – S/2 Pa
h= m + S/2
Pb*= m – 2 cS Pa*= m – 2 cS
Range of quotes
Range of transactedprices
E(Pb) E(Pa)
Average Bid-ask spread
46
Price Dispersion and “Hit Rate”• Based on this setup, the dispersion of transacted prices
pk from the market’s valuation, m, have a mean zero and variance equal to:
• Percentage of trades that fall within the median quote (hit-rate) can be derived:
47
2if)31(2if)32(
)( 222
ScS
ScScmpE k
S/8cif%100
S/2cS/8if2c2
S/2cif%50
SHR
47
48
Liquidity Measure
• Based on the model we propose the following new liquidity measure for bond i on day t:
where Ni,t … number of transactions, for bond i on day tpi,j,t … transaction price for j = 1 to Ni,t, for bond i on day tVi,j,t … trade volume j = 1 to Ni,t,for bond i, trade j, on day tmi,t … market-wide valuation, for bond i on day t
• Intuition behind the measure: Sample estimate of the price dispersion using all trades within a day.
t,j,iN
1j2
t,it,j,iN
1j t,j,i
t,i V)mp(V
1d t,i
t,i
48
49
Liquidity Measure
• Represents the root mean squared difference between the traded prices and the respective market-wide valuation.
• Is an estimate for the absolute deviation and, more importantly, has the interpretation as the volatility of the price dispersion distribution.
• Volume-weighting assumes that price dispersion is revealed more reliably in larger trades and eliminates potential erratic prices of particularly small trades.
49
50
Data for the Present Study• Time period: October 2004 to October 2006• US bond market data from three sources:
– TRACE: all transaction prices and volumes– Markit: average market-wide valuation each trading
day– Bloomberg: closing bid/ask quotes at the end of each
trading day– Bloomberg: bond characteristics
• Dollar denominated• Fixed coupon or floating rate• Bullet or callable repayment structure• Issue rating from S&P, Moody’s or Fitch• Traded on at least 20 days in this period 50
51
Data for the Present Study
• Selected bonds represent:– 7.98% of all US corporate bonds– 25.31% (i.e., $1.308 trillion) of the total amount
outstanding– 37.12% of the total trading volume
• Available bond characteristics:– Coupon, maturity, age, amount issued, issue rating,
and industry• Available trading activity variables:
– trade volume, number of trades, bid-ask spread and depth (i.e., number of major dealers providing information to Markit)
51
52
Data for the Present Study• Distribution of bonds across Bloomberg industry
categories:
BANK FINANCIAL INDUSTRIAL TRANS - NON RAIL UTILITY - ELEC
Industry
Num
ber o
f Bon
ds
020
040
060
080
0
Bank Gas Transm Telephone Trans Rail Utility-Gas Financial Industrial Trans-Non Rail Utility-Elec Industry
Num
ber o
f Bon
ds
52
53
Data for the Present Study• Distribution of bonds across ratings:
AAA AA A BBB BB B C/CCC
010
020
030
040
050
0
Rating Grade
Num
ber o
f Bon
ds
53
54
Data for the Present Study
• Distribution of other variables:Min Median Max _
Coupon 1.95% 6.125% 11.25%Maturity 0.07 4.80 31.61Age 0.04 3.34 16.23Amount Issued $100m $500m
$6.5bBid-Ask 1.55bp 32.17bp 90bpTrade Volume $127,925 $2.8m $61.3mTrades 1.21 4.14 121.6Depth 3 4.62 12.86
54
55
Data for the Present Study
• Trading frequencies:
Days per year 10/2004-10/2005 10/2005-10/2006> 200 411 392151 – 200 309 369101 – 150 236 32251 – 100 221 222
50 444 459 _Total # bonds 1621 1704
55
56
Empirical Results – Market Level Analysis
• Volume-weighted average difference between TRACE prices and respective Markit quotations is 4.88 bp with a standard deviation of 71.85 bp significant bias.
• Price dispersion measure (i.e. root mean squared difference) is 49.94 bp with a standard deviation of 63.36 bp.
• Market-wide average bid-ask spread is only 35.90 bp with a standard deviation of 23.73 bp.
• Overall, we find significant differences between TRACE prices and Markit composite that cannot be simply explained by bid-ask spreads or trade time effects.
56
57
Empirical Results – Bond Level Analysis
• At the individual bond level, we relate our liquidity measure to bond characteristics and trading activity variables to show its relation to liquidity.
• We employ cross-sectional linear regressions using time-weighted averages of all variables.
• We present results based on the whole time period, as well as based on each available quarter (2004 Q4 to 2006 Q3).
• To further validate the results, we analyze the explanatory power of our liquidity measure in predicting established estimators of liquidity measure.
57
58
Empirical Results – Bond Level Analysis
• Cross-sectional regressions with the new price dispersion measure as dependent variable:
2004 Q4 2006 Q3 Overall __Constant 231.732*** 167.760*** 187.648***
Maturity 2.576*** 1.453*** 1.840***Amount Issued-5.597*** -3.710*** -3.060***Age 3.849*** 1.242*** 2.064***Rating 2.090*** 1.096*** 1.254***Bid-Ask 0.237*** 0.544*** 0.568***Trade Volume -7.963*** -6.023*** -8.458*** _R2 44.9% 49.3% 61.5% _Observations 1270 1513 1800 58
59
Empirical Results – Bond Level Analysis: Maturity
59
60
Empirical Results – Bond LevelAnalysis: Bid-Ask Spread
60
61
Empirical Results – Bond LevelAnalysis :Trade Volume
61
62
Empirical Results – Bond Level Analysis :Age
62
63
Empirical Results – Bond Level Analysis : Rating Grade
63
64
Empirical Results – Bond Level Analysis:Amount Issued
64
65
Empirical Results – Bond Level Analysis
• To validate these results, we compare the new measure to established estimators of liquidity in the literature.
• One important approach to measure liquidity is through the price impact of trading.
• A popular (and intuitive) measure was introduced by Amihud:
where ri,t … return on the bond i on day tVi,t ... trade volume of the bond i on day t
T
tti
ti
V
r
Ti
1,
,1bondformeasureILLIQAmihud
65
66
Empirical Results – Bond Level Analysis
• Cross-sectional univariate regressions with the Amihud measure as dependent variable:
2004 Q4 2006 Q3 Overall __Constant -18.192*** -18.377*** -17.932***
Dispersion 0.021*** 0.027*** 0.025*** _R2 22.0% 27.3% 31.3% __Observations 1169 1426 1800
66
67
Empirical Results – Bond Level Analysis
• Cross-sectional multivariate regressions with the Amihud measure as dependent variable:
2004 Q4 2006 Q3 Overall Overall 2__ Constant -10.033*** -8.464*** -7.624*** 0.177Price Dispersion 0.015*** 0.021*** 0.018*** -Coupon 0.190*** 0.198*** 0.186*** 0.296***Amount Issued -0.156* -0.226*** -0.275*** -0.583***Trades -0.049*** -0.059*** -0.067*** -0.050***Trade Volume -0.366*** -0.375*** -0.336*** -0.436*** _R2 42.5% 59.2% 63.7% 52.8% _Observations 1169 1426 1800 1800
67
68
Empirical Results – Hit Rate Analysis
• Many studies use bid-ask quotations (or mid quotes) as proxies for traded prices. Our data set allows us to validate this assumption.
• The hit-rate for the TRACE price is 51.37% (i.e., in these cases, the traded price lies within the bid and ask quotation).
• Deviations are symmetric bid and 49.88% are higher than the ask.
• Even the hit rate of the Markit quotation (58.59%) is quite low.
• Overall, we find that deviations of traded prices from bid-ask quotations are far more frequent than assumed by most studies. 68
69
Price Dispersion Paper: Main Insights
• A new liquidity measure based on price dispersion effects is derived from a market microstructure model.
• The proposed measure is quantified in the context of the US corporate bond market.
• It is larger and more volatile than bid-ask spreads and shows a strong relation to bond characteristics, trading activity variables, as well as established liquidity proxies.
• A “hit-rate” analysis shows that bid-ask spreads can only be seen as a rough approximation of liquidity costs.
• The proposed measure can potentially explain and quantify the liquidity premia.
• These findings foster a better understanding of OTC markets and are relevant for many practical applications, e.g. pricing, risk management, and market regulation. 69
Liquidity during Financial Crises• Liquidity effects in illiquid over-the-counter
markets: The US corporate bond market."Latent Liquidity: A New Measure of Liquidity with an Application to Corporate Bonds," (with G. Chacko, S. Mahanti, G. Mallik and A. Nashikkar), Journal of Financial Economics, May 2008.
"Limited arbitrage and liquidity in the market for credit risk," (with S. Mahanti and A. Nashikkar), Journal of Financial and Quantitative Analysis, June 2011.
"Price Dispersion in OTC Markets: A New Measure of Liquidity," (with R. Jankowitsch and A. Nashikkar), Journal of Banking and Finance, February 2011.
"Illiquidity or Credit Deterioration: A Study of Liquidity in the US Corporate Bond Market during Financial Crises,“ (with R. Jankowitsch and N. Friewald), Journal of Financial Economics, July 2012. 70
Liquidity during Financial Crises (Contd.)
"Liquidity, Transparency and Disclosure in the Securitized Product Market," (with N. Friewald and R. Jankowitch), Working Paper, December 2012."The Determinants of Recovery Rates in the US Corporate Bond Market," (with R. Jankowitsch and F. Nagler), Journal of Financial Economics, forthcoming.
71
Liquidity during Financial Crises: Main Insights
• We employ a wide range of liquidity proxies (bond characteristics, trading activity variables and liquidity measures) to explain yield spread (changes) while controlling for credit risk.
• We examine three different regimes in our sample period which allows as to compare liquidity effects during two periods of crisis (GM/Ford crisis, sub-prime crisis) with a more normal period in between.
• We analyze investment vs. speculative grade bonds to provide evidence whether liquidity is priced differently in these sub-segments.
• We use panel regressions and Fama-MacBethregressions to analyze liquidity effects.
72
Liquidity during Financial Crises: Main Insights
• Liquidity effects explain about 14% of the explained market-wide corporate yield spread variation.
• During periods of crisis, the economic impact of the liquidity measures increases significantly (more than doubles in the sub-prime crisis.)
• More pronounced liquidity effects are seen in the speculative grade segment, particularly in the sub-prime crisis.
• Results are relevant for pricing, risk management, and regulatory policy.
73
Liquidity Effects in the Securitized Fixed Income Market
• Liquidity effects in illiquid over-the-counter markets: The US corporate bond market.
"Latent Liquidity: A New Measure of Liquidity with an Application to Corporate Bonds," (with G. Chacko, S. Mahanti, G. Mallik and A. Nashikkar), Journal of Financial Economics, May 2008."Limited arbitrage and liquidity in the market for credit risk," (with S. Mahanti and A. Nashikkar), Journal of Financial and Quantitative Analysis, June 2011.
"Price Dispersion in OTC Markets: A New Measure of Liquidity," (with R. Jankowitsch and A. Nashikkar), Journal of Banking and Finance, February 2011.
"Illiquidity or Credit Deterioration: A Study of Liquidity in the US Corporate Bond Market during Financial Crises,“ (with R. Jankowitsch and N. Friewald), Journal of Financial Economics, July 2012. 74
Liquidity Effects in the Securitized Fixed Income Market (Contd.)
"Liquidity, Transparency and Disclosure in the Securitized Product Market," (with N. Friewald and R. Jankowitch), Working Paper, December 2012."The Determinants of Recovery Rates in the US Corporate Bond Market," (with R. Jankowitsch and F. Nagler), Journal of Financial Economics, forthcoming.
75
Liquidity Effects in the Securitized Fixed Income Market: Main Insights • The average transaction cost for a round-trip trade is 66
bp.• ABS and MBS segments comparable to the US
corporate bond market, the CMO segment is much less liquid, and the TBA segment is much more liquid.
• Liquidity measures that use additional dealer-specific information can be efficiently proxied by measures using less information.
• Liquidity effects cover around 10 % of the explained variation in yield spreads: justification for dissemination of trade data.
76
Recovery Rates in the Corporate Bond Market
• Liquidity effects in illiquid over-the-counter markets: The US corporate bond market.
"Latent Liquidity: A New Measure of Liquidity with an Application to Corporate Bonds," (with G. Chacko, S. Mahanti, G. Mallik and A. Nashikkar), Journal of Financial Economics, May 2008."Limited arbitrage and liquidity in the market for credit risk," (with S. Mahanti and A. Nashikkar), Journal of Financial and Quantitative Analysis, June 2011.
"Price Dispersion in OTC Markets: A New Measure of Liquidity," (with R. Jankowitsch and A. Nashikkar), Journal of Banking and Finance, February 2011.
"Illiquidity or Credit Deterioration: A Study of Liquidity in the US Corporate Bond Market during Financial Crises,“ (with R. Jankowitsch and N. Friewald), Journal of Financial Economics, July 2012. 77
Recovery Rates in the Corporate Bond Market (Contd.)
"Liquidity, Transparency and Disclosure in the Securitized Product Market," (with N. Friewald and R. Jankowitch), Working Paper, December 2012."The Determinants of Recovery Rates in the US Corporate Bond Market," (with R. Jankowitsch and F. Nagler), Journal of Financial Economics, forthcoming.
78
Recovery Rates in the Corporate Bond Market: Main Insights
• Evidence for the market-based definition of realized recovery rates:
- Based on analysis of the microstructure of trading in a time window of 90 days before, on, and 90 days after default.
• Time-series and cross-sectional effects based on a wide range of variables:
- Quantifying effects of bond characteristics, firm fundamentals, macroeconomic indicators.
• Assessing the effects of individual bond liquidity onrecovery rates:
- Particularly interesting due to potential illiquidity in default.
79
Recovery Rates in the Corporate Bond Market: Main Insights
• Interesting results for bonds that are deliverable into CDS contracts:
- Exhibit higher recoveries of around 6.2 % of FV.
• Ratios motivated by structural credit risk models are important:
- A 10% point increase in equity (default barrier) increases (decreases) recoveries by around 1.3-2.2 % of FV.
• Overall macroeconomic conditions have a significant effect on recoveries:
- A 1 % point increase in market-wide and industry-specific default rates decreases recoveries by 3.3 % and 0.7% of FV.
• Illiquid bonds recover less:- A 1 % percentage point increase in transaction costs decreases recoveries by 7.5-8.8 % of face value. 80
Conclusions and Issues for Discussion
• Liquidity and liquidity risk are important aspects of asset pricing – a premium for assets in positive net supply.
• Corporate bond markets are highly illiquid, hence the premia could be quite large.
• Need to have a metric for liquidity – the case for a liquidity rating.
• The use of trading activity based and derived measures of liquidity.
• Need for further work on the liquidity of corporate bonds and CDS contracts.
81