Upload
jeffrey-dixon
View
215
Download
0
Embed Size (px)
Citation preview
1
Transparency , Information Content and Order Placement Strategy
Tai Ma, Yaling Lin, Hsiu-Kuei Cheng
Department of Finance, National Sun Yat-sen University
2
Introduction
Increasing pre-trade transparency in Taiwan Stock Exchange
Before July 1st, 2002: only the best price is disclosed.
Since July 1st, 2002: the volume of best bid/ask is disclosed additionally.
Beginning in 2003: both the price and volume of the best five unexecuted orders are disclosed
3
Introduction
How much additional information is valuable after grater transparency?
How does additional information influence order strategies of individual and institutional investors?
4
Related Research
Traders’ choices between market and limit orders
limit orders are not as informative as market orders [e.g., Glosten, 1994 ; Angel, 1994 ; Rock, 1996; Harris, 1998 ].
informed traders would prefer to submit limit orders rather than market orders [e.g., Bloomfield, O’hara and Saar ( 2005 ) and Kaniel and Liu ( 2006 ) ].
5
Related ResearchLOB information affects traders’ behavior
the state of LOB will influences the forthcoming order flow [e.g., Handa and Schwartz ( 1996 ) , Parlour ( 1998 ) and Foucault ( 1999 ) , Coppejans and Domowitz (2002) , Biais et al.( 1995), Griffiths et al.( 2000), Hollifield et al. ( 2003 ) , and Ranaldo (2004) ].
the impact of LOB information beyond it’s first step on investors’ strategies [e.g., Cao, Hansh and Wang (2003) and Pascual and Veredas (2003) ].
6
Related Research
Comment:Comment:
none of these studies has measured the information content of the book beyond its first step for different types of investors .They do not connect the information content and order placement strategy with transparency enhancement.
7
Contribution
The first one to combine the issues of LOB information content, investors’ order strategies and transparency.
Exploring information content of the best quote and the quote beyond the best for different kinds of traders.
Investigating the transparency impact on different types of traders’ order placement strategies.
8
Sample Period1st stage :
least transparent : Feb-02 、 Jun-02
2nd stage :
partially transparent :Jul-02 、 Dec-02
3rd stage :
most transparent : Mar-03 、 Jun-03
9
Samples and subsamples
50 stocks listed in the Taiwan Stock Exchange.
buyer seller
informed group1 group2
uninformed group3 gruop4
10
Samples and sub-samples
Time Series for Analysis (Information content) :
In the second stage
In the third stage
11
Methodology
Information share of LOB VAR (vector autoregressive model)
Partial transparent period:
1 、 model1 2 、 model2 The most transparent period
1 、 model3 2 、 model4
12
The Information content of LOB
Vector Autoregressive Model Information content:
the percentage of variation in trading prices (volume)
due to posted bid/ask price (bid/ask volume).
Model 1 (2nd stage ) (back)Pt
p
1i
1iti
1it
p
1ii
p
1iiti
Pt εΔQδΔPγΔPβαΔP
0
11111
01
11
11
1 Qt
p
llt
Qllt
p
l
Ql
p
llt
Ql
Qt QPPQ
11111
01
11
11
1 Pt
p
jjt
pjjt
p
j
pj
p
jjt
pj
pt QPPP
13
The Information content of limit order book
Model 2 (2nd stage ) (back)
Qt
p
kktk
p
kktkkt
p
kk
Q
t QQPQ
1
1
1
1
10
11111
01
1
1
1
1
1 Pt
p
jjt
pj
p
jjt
pjjt
p
j
pj
pt QQPP
11111
01
1
1
1
1
1 Qt
p
llt
Ql
p
llt
Qllt
p
l
Ql
Qt QQPQ
14
The Information content of limit order book
Model 3 (3rd stage ) (back)
Pt
p
iitiit
p
ii
p
iiti
Pt PPPP
1
521
110
11
01
521
11
1 Pt
p
jjtjjt
p
jj
p
jjtj
pt PPPP
5252
1
521
110
52
Pt
p
mmtmmt
p
mm
p
mmtm
pt PPPP
15
The Information content of limit order book
Model 4 (3rd stage ) (back)
Qt
p
kktk
p
kktk
p
kktk
Qt QQQQ
1
52
1
1
10
11
01
52
1
1
1
1 Qt
p
lltl
p
lltl
p
lltl
Qt QQQQ
5252
1
52
1
1
10
52
Qt
p
nntn
p
nntn
p
nntn
Qt QQQQ
16
MethodologyOrder Strategies: (aggressiveness of buyers) C1: bid price = maximum price limit
C2: best ask < bid price < maximum price limit
C3: best bid < bid price <= best ask
C4: bid price = best bid
C5: best bid minus two ticks <= bid price < best bid
C6: minimum price limit <= bid price < best bid minus two ticks
17
Methodology
Order Strategies:(aggressiveness of sellers) C1: ask price = minimum price limit
C2: minimum price limit < ask price < best bid
C3: best bid <= ask price < best ask
C4: ask price = best ask
C5: best ask < ask price <= best ask add two ticks
C6: best ask add two ticks < ask price <= maximum price limit
18
Methodology Order Probit Model
Aggressiveness index: (C1 to C6) Explanatory variables:LOB first step information:
DS1 ( DO1 ) :The best quote depth on the same ( opposite ) side LOB information beyond the first step:
DS25(DO25): The accumulative quote depth from step2 to step5 on the same ( opposite ) side.
LS12(LO12):The distance between the best and second best quote on the same ( opposite ) side.
LS25(LO25):The distance between the second and fifth best quoteon on the same ( opposite ) side..
Other control variables: one-lagged order aggressiveness Margin trading spread
19
Empirical Results
1.Information contents1.1 Summary Statistics for Information
share Disclosing bid price and volume have
more information content than ask price and volume, especially in the most transparent period.
Table 2 (stage 2) Table 3 (stage 3)
20
Empirical Results (BACK)1.2 Test results of information
content In the partial transparent market , the best
quote price has more information content than the best quote volume when we decompose the trading price. Table4 (stage 2)
Institutional traders’ unmatched orders provide more information content than individual traders. Table4 (stage 2)
Table5(stage 3)
21
Empirical Results
The best quote price and the average step2 to step5 quote price• Individual trader’s best bid/ask of
unexecuted orders always has more information content than the average of step2 to step5 quote price Table5
• Institutional investor’s average of step2 to step5 quote price contains the same information as the best unexecuted orders at least. Table5
22
Empirical Results
2 Orderplacement strategies
2.1 Summary Statistics for Orderplacement strategies
orders are distributed evenly among all aggressive categories except the sub-aggressive category ( C2) Table6
23
Empirical Results2.2 Transparency and orderplacement strategies
stage 1 into stage 2 or 3: both institutional and individual investors reduce the most aggressive orders ( C1 ) , and C3 as well as C4 increase instead.
stage 2 into stage 3 : the increasing tendency in C3 is more striking for institutional investors’ buy orders .
Table7
24
Empirical Results2.3 Probit model: the determinants of order aggressiveness.
the order aggressiveness of all sub-samples performs significant positive first-order autocorrelation.
The sign of the margin trading dummy is positive ,especially for individual traders
When the spread widens, the order aggressiveness will decline , especially for individual traders and sell orders.
Table8 Table9
25
Empirical Results2.3 Probit model: the determinants of order aggressiveness.
The order submission depends mostly on the best quote and corresponding depth, the disclosures beyond step 1 still have explanatory power.
Quote depth
The aggressiveness is positive ( negative ) with the book depth on the same ( opposite ) side .This trend is more evident for the best quote than the quote beyond the first.
The accumulated depth at step2 to 5 has more significant effect on individual strategies.
The aggressiveness of buyers is affected more by opposite-side depth, while the sellers are more concerned about the same side depth
Table8 Table9
26
Empirical Results2.3 Probit model: the determinants of order aggressiveness.
Quote price
Investors become more aggressive with smaller quote distance on the same side or with larger quote distance on the opposite side.
This trend is more evident for the best quote than the
step2 to step5 quote price, institutional buy orders are only exception.
Order submissions are affected by the quote distance of the same side rather than the opposite side.
27
Conclusion--Informaiton content
The LOB information content of buy orders is more than that of sell orders.
the quote price is more informative than the quote volume.
Institutional traders’ unmatched orders provide more information content than individual traders.
28
Conclusion--Informaiton content
the individual traders’ best quote always has more information content than the average 2nd to 5th quote, while it is not true for institutional investors.
29
Conclusion—Order placement strategy
Higher transparency makes all investors reduce unnecessarily costly orders.
The competition among institutional traders is stronger than individual traders as transparency increases.
Investors not only utilize the best quote and depth to make decisions but also use the information beyond step 1 of the book.
30
Conclusion—Order placement strategy
Investors are more aggressive as the same side of the book is more crowded and the opposite side of the book is more disperse.
the depth of limit order book has quite diversified effects on order submission for the buyers, sellers, institutional and individual investors .
31
Table 2 Summary Statistics for Information share in the partially transparent market (stage 2) (BACK)
The number in the cell is the percentage of information share. We list the results for buy and sell orders
separately. Investors are classified into informed (institutional) and uninformed ( individual ) traders.
Moreover, "bid1" is the best bid, "bidq1" is the volume of best bid, "ask1" is the best ask and "askq1" is
the volume of best ask. Panel A: Information share of trading price
Identity of investors Identity of investorsSource of variation bid1 bidq1 bid1 bidq1 Source of variation ask1 askq1 ask1 askq1
mean 63.95 2.06 25.52 0.95 mean 49.90 2.59 23.55 1.34median 76.69 0.84 24.05 0.68 median 57.36 1.07 22.69 1.20maximum 96.08 9.58 77.07 3.14 maximum 85.09 26.61 54.92 3.34minimum 2.54 0.04 4.14 0.13 minimum 0.07 0.00 5.70 0.38variance 29.21 2.70 15.57 0.71 variance 29.50 5.75 11.95 0.89skewness -0.87 1.60 1.48 1.27 skewness -0.26 3.91 0.94 0.85kurtosis 2.64 4.57 6.00 4.64 kurtosis 1.51 16.87 3.57 2.52
Panel B: Information share of trading volume
Identity of investors Identity of investorsSource of variation bid1 bidq1 bid1 bidq1 Source of variation ask1 askq1 ask1 askq1
mean 4.01 8.45 0.87 1.45 mean 2.46 1.18 1.34 1.80median 1.09 1.91 0.61 0.92 median 0.67 0.60 0.81 0.89maximum 22.70 68.65 2.44 14.53 maximum 12.39 4.41 10.93 11.48minimum 0.02 0.29 0.04 0.00 minimum 0.00 0.02 0.00 0.00variance 6.72 16.88 0.74 2.85 variance 3.60 1.31 2.40 2.49skewness 1.93 2.78 0.96 4.06 skewness 1.69 1.36 3.01 2.75kurtosis 5.20 9.75 2.70 19.15 kurtosis 4.71 3.66 11.88 10.67
Buy Sell informed
Buy Sell
uninformed informed uninformed
informed uninformed informed uninformed
32
Table 3 Summary Statistics for Information share in the most transparent market (stage 3) (BACK)
"bid1" is the best bid, "bid25" is the average of step 2 to step 5 bid price, "bidq1" is the volume of best
bid, "bidq25" is the average of step 2 to step 5 bid volume, "ask1" is the best ask, "ask25" is the average
of step 2 to step 5 ask price, "askq1" is the volume of best ask and "askq25" is the average of step 2 to
step 5 ask volume.
Panel A: Information share of trading price
Identity of investors Identity of investorsSource of variation bid1 bid25 bid1 bid25 Source of variation ask1 ask25 ask1 ask25
mean 13.59 35.21 23.60 7.48 mean 12.67 11.59 11.32 3.59median 4.52 34.68 24.26 0.56 median 3.97 3.56 7.34 0.28maximum 71.16 89.90 71.82 44.11 maximum 74.66 62.95 49.51 25.94minimum 0.62 0.00 0.28 0.00 minimum 0.00 0.00 0.00 0.00variance 20.08 33.41 18.22 13.33 variance 20.22 20.96 14.14 7.06skewness 1.94 0.29 0.67 1.65 skewness 2.08 1.84 1.69 2.07kurtosis 5.91 1.59 3.25 4.24 kurtosis 6.48 4.66 4.65 6.09
Panel B: Information share of trading volume
Identity of investors Identity of investorsSource of variation bidq 1 bidq 25 bidq 1 bidq 25 Source of variation askq1 askq25 askq1 askq25
mean 2.48 6.59 2.68 1.77 mean 4.92 4.82 1.52 0.54median 0.22 0.43 0.27 0.03 median 0.51 0.04 0.37 0.04maximum 27.56 68.64 36.37 37.23 maximum 56.31 59.40 22.28 6.79minimum 0.00 0.00 0.01 0.00 minimum 0.00 0.00 0.00 0.00variance 6.54 16.78 7.44 7.58 variance 12.37 13.18 4.40 1.39skewness 3.29 3.04 4.11 4.54 skewness 3.45 3.42 4.45 3.90kurtosis 12.82 11.28 19.10 21.75 kurtosis 14.40 14.21 21.55 18.02
Buy Sell informed uninformed informed uninformed
Buy Sellinformed uninformed informed uninformed
33
Table 4 Difference in the information share of trading price and volume in
the partial transparent market (stage 2) (BACK)Panel A: Information share of trading price
bid1 - bidq1 61.89 *** 24.57 ***
ask1 - askq1 47.31 *** 22.21 ***
bid1 38.44 ***
bidq1 1.11 ***
ask1 26.35 ***
askq1 1.24
Panel B: Information share of trading volume
bid1 - bidq1 -4.44 -0.59ask1 - askq1 1.28 -0.47
bid1 3.14bidq1 7.00 *
ask1 1.13askq1 -0.62
* indicates significant at 10%, ** indicates significant at 5%, and *** indicates significant at 1%
informed uninformed informed - uninformed
informed uninformed informed - uninformed
34
Table 5 Difference in the information share of trading price and volume in
the most transparent market (stage 3) (BACK1) (BACk2)
Panel A: Information shares of trading price
bid1 - bid25 -21.62 * 16.12 **
ask1 - ask25 1.08 7.73 **
bid1 -10.01 *
ask1 1.34bid25 27.73 ***
ask25 7.99 **
Panel B: Information shares of trading volume
bidq1 - bidq25 -4.11 0.92 **
askq1 - askq25 0.10 0.98 ***
bidq1 -0.21askq1 3.40 *
bidq25 4.82 *
askq25 4.29* indicates significant at 10%, ** indicates significant at 5%, and *** indicates significant at 1%
informed uninformed informed - uninformed
informed uninformed informed - uninformed
35
Table 6 Distribution of order types by institutional and individual investorsBACK
I nsti tuti onal I ndi vi dual I nsti tuti onal I ndi vi dual
C1 (%) 23.82 22.12 18.54 20.07C2 (%) 8.60 6.32 7.35 8.64C3 (%) 13.12 13.85 16.13 12.78C4 (%) 22.14 21.64 21.90 19.07C5 (%) 15.17 17.46 17.10 16.98C6 (%) 17.15 18.61 18.98 22.46
Total 100.00 100.00 100.00 100.00
Proporti on (%) 3.83 48.05 4.62 43.50
I nsti tuti onal I ndi vi dual I nsti tuti onal I ndi vi dual
C1 (%) 18.89 19.12 14.10 16.14C2 (%) 8.92 4.37 5.08 8.58C3 (%) 18.52 17.06 23.31 17.43C4 (%) 24.13 25.03 25.71 22.39C5 (%) 14.77 17.68 15.52 16.53C6 (%) 14.77 16.74 16.28 18.93
Total 100.00 100.00 100.00 100.00
Proporti on (%) 4.48 47.22 4.62 43.68
I nsti tuti onal I ndi vi dual I nsti tuti onal I ndi vi dual
C1 (%) 16.04 19.48 13.09 15.76C2 (%) 4.37 6.77 6.77 9.04C3 (%) 21.64 15.62 23.28 16.78C4 (%) 27.43 24.29 25.17 22.73C5 (%) 15.95 17.19 16.07 17.08C6 (%) 14.57 16.65 15.62 18.61
Total 100.00 100.00 100.00 100.00
Proporti on (%) 4.59 43.39 4.68 47.34
Buyers Sel l ers
Stage 1 (l east transparency)
Stage 2 (parti al transparency)
Stage 3 (most transparency)
Buyers Sel l ers
Buyers Sel l ers
36
Table 7 Differences in order aggressiveness between stages of various transparency levels Back
stage1 - stage2 stage1 - stage3 stage2 - stage3
C1 (%) 4.93 7.78 ** 2.85
C2 (%) -0.32 4.23 * 4.55C3 (%) -5.40 *** -8.52 *** -3.12 ***
C4 (%) -1.99 -5.29 ** -3.30 **
C5 (%) 0.40 -0.78 -1.18C6 (%) 2.38 2.58 0.20
stage1 - stage2 stage1 - stage3 stage2 - stage3
C1 (%) 4.44 *** 5.45 *** 1.01
C2 (%) 2.28 0.58 -1.69C3 (%) -7.18 *** -7.14 *** 0.03C4 (%) -3.81 ** -3.28 ** 0.53C5 (%) 1.58 1.03 -0.56C6 (%) 2.69 3.36 ** 0.68
stage1 - stage2 stage1 - stage3 stage2 - stage3
C1 (%) 3.00 * 2.64 -0.35
C2 (%) 1.94 -0.46 -2.40C3 (%) -3.20 ** -1.77 1.44 *
C4 (%) -3.39 ** -2.65 0.74C5 (%) -0.22 0.27 0.49C6 (%) 1.87 1.96 0.09
stage1 - stage2 stage1 - stage3 stage2 - stage3C1 (%) 3.93 *** 4.31 *** 0.38C2 (%) 0.06 -0.40 -0.47C3 (%) -4.65 *** -4.00 *** 0.65C4 (%) -3.33 ** -3.66 *** -0.34C5 (%) 0.45 -0.10 -0.55C6 (%) 3.54 ** 3.86 ** 0.32
Nonparametri c Wi l coxon test. * i ndi cates si gni fi cant at 10%, ** i ndi cates si gni fi cant at the5%, and *** i ndi cates si gni fi cant at the1%
Insti tuti onal Buyers
Insti tuti onal Sel l ers
Individual Buyers
Individual Sel l ers
37
Table 8 Ordered Probit Model of Order Choice for Institutional Investors Back1 Back2
AGG(-1) TRUST SPR DS1 DO1
Mean 0.1364 0.2010 -5.2341 0.0001 -0.0004
Medi an 0.1288 0.2465 -5.9870 0.0001 -0.0004
Si gni fi cant(%) 94 34 78 82 86
Si g. Posi t i ve(%) 94 28 6 80 0
AGG(-1) TRUST SPR DS1 DO1 DS25 DO25 LS12 LO12 LS25 LO25
Mean 0.0841 -0.0480 -6.1837 0.0004 -0.0002 0.0002 -0.0002 -0.3668 0.6236 -0.3035 0.3550
Medi an 0.0995 -0.0990 -8.1858 0.0001 -0.0001 0.0002 -0.0001 -0.0019 0.0001 -0.0111 0.0019
Si gni fi cant(%) 92 44 80 84 92 76 76 62 64 78 60
Si g. Posi t i ve(%) 88 36 10 80 0 66 6 10 54 8 52
AGG(-1) TRUST SPR DS1 DO1
Mean 0.1181 0.1979 -4.1545 0.0003 -0.0002
Medi an 0.1097 0.2162 -3.2211 0.0003 -0.0001
Si gni fi cant(%) 94 36 84 100 88
Si g. Posi t i ve(%) 94 26 4 96 4
AGG(-1) TRUST SPR DS1 DO1 DS25 DO25 LS12 LO12 LS25 LO25
Mean 0.0977 -0.0616 -5.4512 0.0004 -0.0004 0.0001 -0.0001 -0.2177 0.2038 -0.2003 0.3855
Medi an 0.0916 0.1438 -4.1418 0.0003 -0.0003 0.0001 -0.0001 -0.1985 0.0011 -0.0301 0.0637
Si gni fi cant(%) 94 30 84 96 88 72 70 92 72 80 68
Si g. Posi t i ve(%) 94 12 6 96 2 64 12 0 72 10 64
Stage 2
Stage 3
Insti tutional Buyers
Insti tutional Sel l ers
Stage 2
Stage 3
38
Table 9 Order Probit Model of Order Choice for Individual Investors Back1 Bacl2
AGG(-1) TRUST SPR DS1 DO1
Mean 0.0871 0.1187 -3.3428 0.0002 -0.0004
Medi an 0.0919 0.1052 -4.7271 0.0002 -0.0001
Si gni fi cant(%) 98 94 98 90 100
Si g. Posi t i ve(%) 98 86 4 90 0
AGG(-1) TRUST SPR DS1 DO1 DS25 DO25 LS12 LO12 LS25 LO25
Mean 0.0952 0.0725 -5.7092 0.0001 -0.0004 0.0001 -0.0002 -0.4605 0.9923 -0.1693 0.2012
Medi an 0.1107 0.0730 -4.1437 0.0001 -0.0002 0.0001 -0.0001 -0.0414 0.0050 -0.0100 0.0009
Si gni fi cant(%) 96 88 96 92 100 82 84 90 80 88 76
Si g. Posi t i ve(%) 94 88 0 90 0 78 2 0 80 8 64
AGG(-1) TRUST SPR DS1 DO1
Mean 0.1083 0.2060 -8.9953 0.0001 -0.0002
Medi an 0.1011 0.1970 -9.0877 0.0001 -0.0001
Si gni fi cant(%) 100 92 96 96 92
Si g. Posi t i ve(%) 100 88 0 96 2
AGG(-1) TRUST SPR DS1 DO1 DS25 DO25 LS12 LO12 LS25 LO25
Mean 0.1158 0.2240 -8.4875 0.0003 -0.0001 0.0002 -0.0001 -0.1184 0.3632 -0.0292 0.5147
Medi an 0.1271 0.2830 -6.5474 0.0003 -0.0001 0.0001 -0.0001 -0.1019 0.0009 -0.0072 0.2385
Si gni fi cant(%) 100 96 100 100 90 80 72 92 78 92 76
Si g. Posi t i ve(%) 96 94 0 100 0 76 0 0 76 8 70
Stage 2
Stage 3
Individual Buyers
Individual Sel l ers
Stage 2
Stage 3