the information in option volume for future stock prices
Post on 17-Jul-2015
210 Views
Preview:
TRANSCRIPT
The Information in Option Volume for Future Stock Prices
Jun PanMI T Sloan School of Management
Allen M. PoteshmanUniversity of Illinois at Urbana-Champaign
2
Why Do Investors Trade Derivatives?
Risk sharing, in an otherwise incomplete market [Ross 1976]: Multiple risk factors [Bates (2002), Liu and Pan (2002)] Dynamic trading not allowed [Haugh and Lo (2001)] Other market frictions: short-sale constraints [Basak
and Croitoru (2000)], non-hedgeable background risk [Franke, Stapleton, and Subrahmanyam (1998)]
Differences of Opinion [Kraus and Smith (1996), Bates (2002), Buraschi and Jiltsov (2002)]Information Trading [Black (1975), Back (1993), Easley, O’Hara, and Srinivas (1998)]
3
“Fact and Fantasy in the Use of Options” (Fischer Black, 1975)“Since an investor can usually get more action for a
given investment in options than he can by investing directly in the underlying stock, he may choose to deal in options when he feels he has an especially important piece of information.
Also, it is easier to take a short position by writing options than by shorting the underlying stock…
And many potential information traders will trade on the options market when they wouldn’t bother to trade at all if the options market did not exist.”
4
Return to 1 month ATM call as a function of return to stock
-30 -20 -10 0 10 20 30-100
0
100
200
300
400
500
600
700
800
Return to the Underlying Stock (%)
Ret
urn
to O
ne-M
ont
h At-th
e-Mone
y Call
Option
(%
)
5
Two Main Goals
Determine whether option volume is informative for future stock prices, (if so, in expected direction?)Provide evidence on key features of information-based theoretical models Do prices adjust more quickly to public than to
non-public information? Is volume more informative when the
concentration of informed investors is greater? Is there more information in volume of more
highly levered contracts?
6
Prev. Evidence: Opt. Vol. Inform. Before Events but not During “Normal” Times
Option volume contains information before important firm specific news Earnings announcements [Amin and Lee (1997)] Takeover announcements [Cao, Chen, and Griffin (2003)]
During “normal” times, stock volume not option volume predicts underlying stock returns Daily intervals [Cao, Chen, and Griffin (2003)] Intraday [Chan, Chung, and Fong (2002)]
Easley, O’Hara, and Srinivas (1998) find information in option volume for contemporaneous stock price movements. Weaker intraday evidence that option volume contains information for future stock prices … but the sign tends to be backwards
7
Empirical Finding: Is Option Volume Informative for Future Stock Prices?
Long-short stock portfolio formed from a put-call volume ratio get 40 basis points the next day and 100 basis points over the next week with very large t-statistics Clear evidence that during “normal” times option
volume contains information about future stock price movements
First evidence in support of expected directional relationship between option volume and future stock price movements
8
Why the Difference in Results?Our data allow us to identify “open buy” option volume – that is, volume which is buyer-initiated to open new option positions We construct our put-call ratios only from
open buy option volumeWe use all CBOE options from 1990-2001 EOS (1998): 50 firms for 44 trading days CCF (2002): 14 firms for 58 trading days
9
Empirical Findings on Aspects of Information-Based Models
Speed of price adjustment Non-public open buy volume is predictive for
several weeks into future (no reversion) Publicly observable volume is predictive for only 2
to 3 days into future (also reverts)
Concentration of informed traders Option volume is more informative for stocks with
higher concentrations of informed investors
Leverage Volume from more highly levered option contracts
is more informative
10
OutlineTheoryData setEmpirical results Main test Private versus public information Concentration of informed traders Leverage Other option volume types Index options
Conclusion
11
Information-Based Theory ModelsGlosten-Milgrom (1985) and Easley-O’Hara (1987): Sequential trade, risk-neutral competitive market-maker, fraction
µ of traders informed Knowing µ are informed, market-maker updates beliefs
conditional on buy or sell and sets prices so expected profit = 0
General Features: Volume is more informative when µ is higher Prices adjust immediately to public information revealed in
trading process but not to private information The price converges to full-information value exponentially at a
rate that increases monotonically in µ:
( ) ( )ln 1 / 1µ µ µ+ −
12
Theory Model with OptionsEasley, O’Hara, Srinivas (1998) (EOS) introduce options into the information-based model framework Informed traders choose to buy or sell stock, put,
or call to maximize expected profit Market-makers watch both markets Separating equilibrium: Informed traders transact
only in stock market Pooling equilibrium: Informed traders transact in
both the stock and option market Induced by high implicit leverage in options, low liquidity
in stock market, high fraction µ of informed traders
13
Empirical Questions
Are stock and option market in separating or pooling equilibrium?If latter, Do stock prices adjust more quickly to
public information? Are options more informative when
concentration of informed traders is higher? Are more levered options more informative?
14
Option Dataset
Daily records of CBOE trading volume for all CBOE listed options from January 1990 through December 2001Each option is identified by its underlying stock or index, as a put or a call, and by its strike price and time to expirationA unique feature of our dataset is that the daily trading volume for each option is broken down into 16 categories defined by 4 trade types and 4 investor classes
15
Trade TypesEach option transaction is identified as: Open buy: buyer initiated to open a new position Open sell: seller initiated to open a new position Close buy: buyer initiated to close an existing short position Close sell: seller initiated to close an existing long position
We know with certainty the “sign” of the trading volume. By contrast, previous studies infer the sign, with some error, from quote and trade data.We know whether the initiator of observed volume is opening a new position or closing an existing one. None of the previous studies have had this information available.
16
Investor ClassesThe volume data is further categorized according to which of the following four investor classes initiates the trades: Firm proprietary traders Public customers of discount brokers Public customers of full-service brokers Other public customers
Trades initiated by market makers are not recorded in our data.
17
Table 1: Option trading volume by trade type and investor class
Open buy Open sell Put Call Put Call Small Stocks Avg volume (Contracts) 16 53 18 49 % Firm Proprietary 7.48 4.46 5.42 4.09 % Discount 7.35 12.92 9.96 11.97 % Full-service 72.61 71.73 75.84 73.66 Medium Stocks Avg volume (Contracts) 38 96 36 89 % Firm Proprietary 10.87 8.81 9.89 7.62 % Discount 8.49 12.48 9.38 9.97 % Full-service 69.22 67.90 71.38 72.37 Large Stocks Avg volume (Contracts) 165 359 135 314 % Firm Proprietary 14.45 11.36 13.61 10.14 % Discount 9.77 13.18 7.83 8.02 % Full-service 63.60 64.70 69.68 71.98
18
Information in Open Buy VolumeTrades initiated by buyers to open new positionsBuild daily cross-sections of stocks with liquid option trading – at least 50 contracts of open buy volume (91 in 1990, 359 in 2001)On each day compute put-call ratio from open buy volume:
If investors with bad news buy new puts and investors with good news buy new calls, then high open buy put-call ratio stock portfolios should subsequently underperform low put/call ratio stock portfoliosPut-call ratio has an average value of 30%. Breaking stocks into quintiles by put-call ratios there is little variation in size, BM, momentum, or analyst coverage.
. .
. . . .
o b put volumeopen buy put call ratio
o b put volume o b call volume− ≡
+
19
Table 2: Option trading behavior of four investor classes (percentages)
Call Put Call Put Call Put Call Put
more than 10% OTM 14.3 22.8 26.8 29.6 20.9 24.6 22.2 25.53% to 10% OTM 24.4 24.9 31.2 32.3 27.9 27.3 27.5 26.1near-the-money 30.6 27.9 26.0 27.6 26.1 26.4 26.4 27.13% to 10% ITM 14.7 11.9 9.60 7.84 13.1 13.3 12.7 13.6
more than 10% ITM 16.0 12.4 6.41 2.75 12.0 8.42 11.3 7.70
under 30 days 35.5 39.6 40.2 52.5 37.3 44.4 38.4 46.830 to 59 days 28.6 25.2 27.6 26.6 29.4 29.9 29.1 27.560 to 89 days 7.82 7.01 7.66 6.25 7.64 6.71 7.36 6.3190 to 179 days 17.7 15.5 15.3 10.9 16.1 12.8 15.6 13.0
more than 179 days 10.3 12.7 9.22 3.74 9.60 6.14 9.53 6.35
Panel A: Moneyness
Panel B: Maturity
Firm Prop. Discount Full Service Other
20
Publicly observable option volume
Lee-Ready algorithm is used on trade and quote data from Berkeley option database over 1990-1996.All trades are classified as buyer- or seller-initiatedPublicly observable buyer- and seller-initiated put-call ratios are computed
21
Table 3: Non-public PC ratios regressed on public PC ratios
lee-ready lee-readydependent variable intercept buyer-initiated seller-initiated R2
open buy 0.08 0.74 0.45(90.37) (304.06)
close buy 0.16 0.42 0.13(111.52) (94.26)
open sell 0.11 0.60 0.35(124.35) (205.90)
close sell 0.05 0.79 0.40(34.43) (232.29)
22
Result 1: Is option volume informative about future stock prices?
Regression: τ-day ahead stock returns on open-buy put-call ratios:
β=-53 basis points (t-stat = -32.9 FM s.e.’s)
Portfolios: 15.7 bps next day for bottom PC quintile -26.6 bps next day for top PC quintile
Robust to removing trade days within +/- 5 days of earnings announcements.Slightly smaller for raw returns
four-factor adj. open-buy , 1, 2,...it it itR PCτ τα β ε τ+ += + + =
23
Temporary Price PressureSuppose market makers delta-hedge and non-market makers do not.Then a buyer-initiated call transaction will result in market maker selling a call and hedging by buying delta shares.So, open-buy call may result in upward price pressure on underlying stock even if the buyer of the call has no information (e.g., is trading for liquidity)Temporary price pressure is not a promising alternative explanation Price pressure would occur on day 0 There would be a price reversal…
24
Conclusion: No evidence of price pressure; markets are in pooling equilibrium
four-factor adj. open-buy , 1, 2,...it it itR PCτ τα β ε τ+ += + + =
25
Closing Time DifferencesThe CBOE closes after the underlying stock market.Part of day +1 return may reflect information released on day 0 after stock market closes but before option market closesOn June 23, 1997, CBOE changed the closing time for equity options from 4:10 PM (EST) to 4:02 PM (EST).If day +1 return results from closing time difference, it should decline after June 1997: Pre- 1997: β=-46 bps (t-stat = -22.3) Post-1997: β=-60 bps (t-stat = -20.8)
26
Table 4: Liquidity and short-term reversal controls
intercept put-call ratio turnover spread R-5,-1
13.04 -59.31(11.00) (-32.82)
2.12 -55.10 6.47 3.34(1.40) (-31.68) (6.82) (2.02)
13.73 -55.62 -0.028(12.25) (-31.56) (-23.53)
3.02 -51.23 6.60 3.56 -0.032(2.09) (-30.21) (6.86) (2.19) (-27.69)
27
Private versus public information
Theoretical models: Prices adjust quickly to public info but may take a while to incorporate non-public infoOver 1990-1996 investigate 3 specs.
four-factor adj. open-buyit it itR PCτ τα β ε+ += + +four-factor adj. Lee-Ready Buyer-Initiatedit it itR PCτ τα β ε+ += + +
four-factor adj. open-buy Lee-Ready B.I.it it it itR PC PCτ τα β γ ε+ += + + +
28
29
Table 5: Predictability conditioning on size and PIN
PIN is a variable derived from a microstructure model (Easley, Kiefer, and O’Hara (1997) and Easley, Hvidkjaer, and O’Hara (2002)) which measures the probability that a trade on a stock is information-basedTheory models predict higher concentration of informed traders leads both to more informative volume and quicker incorporation of information into security pricesDaily minima and maxima of PIN average 0.05 and 0.28; PIN is correlated –0.61 with size.
( )four-factor adj. open-buy open-buy open-buylnit it it i it i itR PC PC size PC PINτ τα β γ δ ε+ += + + × + × +
30
Table 5: Predictability conditioning on size and PIN
Adding PIN significantly reduced predictability from PCMoving from low to high PIN stocks (i.e., PIN = 0.05 to 0.28 adds 43 bps of predictability to PC
put-call ratio put-call ratiointercept put-call ratio x ln(size) x PIN
9.49 -34.56(11.94) (-22.17)
9.28 -152.83 5.27(11.56) (-6.50) (5.13)
9.42 -10.48 -189.29(11.79) (-2.29) (-5.05)
9.38 -91.51 3.18 -112.38(11.66) (-2.45) (2.22) (-2.14)
31
Table 6: Predictability by investor class
prop. avg. num.intercept traders discount full service other of stocks
-5.59 -1.52 53(-4.68) (-0.75)
4.91 -34.82 175(4.80) (-20.02)
9.10 -44.26 336(11.13) (-37.00)
3.41 -28.94 141(3.04) (-17.51)
8.87 4.72 -12.96 -30.39 -24.47 27(2.67) (0.99) (-1.71) (-3.53) (-4.35)
public customers
32
Table 7a: Predictability by moneyness
Liquidity (as measured by volume) is comparable across first there categories but informativeness of volume is not
avg. num.category intercept put-call ratio of stocks
more than 10% OTM 14.65 -44.67 207(13.06) (-29.57)
3% to 10% OTM 1.86 -21.15 181(2.19) (-16.71)
near-the-money -2.32 -11.74 152(-2.64) (-8.43)
3% to 10% ITM -4.79 -2.71 125(-5.07) (-1.85)
more than 10% ITM -6.21 7.95 134(-6.10) (3.52)
33
Table 7b: Predictability by maturity
avg. num.
category intercept put-call ratio of stocksunder 30 days 8.77 -34.83 382
(11.04) (-31.20)
30 to 59 days 7.71 -28.52 328(9.57) (-24.64)
60 to 89 days 6.50 -19.92 251(7.87) (-15.91)
90 to 179 days 6.25 -17.40 219(7.37) (-13.16)
more than 179 days 4.40 -6.91 106(4.38) (-3.63)
34
Table 8: Predictability of various option volume types
avg. num.volume type intercept put-call ratio stocks
open-buy 12.08 -52.55 242(12.50) (-32.92)
open-sell -11.00 20.03 253(-13.30) (12.39)
close-buy -5.31 -0.93 147(-5.07) (-0.49)
close-sell -17.72 27.45 175(-18.72) (14.63)
35
Table 9: Predictability from index option volume
prop.
Index traders discount full service otherSPX -8.50 10.20 -1.50 1.80
(-1.13) (1.08) (-0.14) (0.24)
OEX 7.30 43.70 64.50 -5.60(0.90) (3.12) (3.60) (-0.46)
NDX -3.20 46.50 12.10 36.00(-0.26) (3.11) (0.69) (3.09)
public customers
36
ConclusionWe produce strong evidence that option volume contains information about the future movements of underlying stock pricesPublicly inferrable information from option volume is quickly incorporated into stock prices but adjustment to non-public information takes up to several weeksOption volume is more predictive on stocks with higher concentrations of informed investorsThe volume of more levered options contains more information about future stock pricesOption volume of full-service customers is most informative while we find no information in option volume of firm proprietary tradersThis paper has focused on directional information, in future research it would be interesting to explore information about the future volatility of underlying stocks
top related