637
greater. If the information is related to the earnings or in accordance with the analysts recommendation, abnormal return during 20 days becomes greater in range of 3,00-
4,00 for good news, and the lowest of -2,25 for bad news. Grundy Kim 2002 stated that rank of information heterogeneity affect the increase of price variability equal
to 20-46 compared to the economics information homogeneity. This price variability means that private information positively contribute to price variability compared to public
information. Suhaibani Kryzanowski 2000 examined the information contents of new bids in Saudi Stock Market SSM. The new bids which are greater and more aggressive
are caused by information arrivals. The relative measurement of bids information implies that private information is dominant factors in stock trading decisions. Therefore, private
information also affects price volatility.
Bery Howe 1994 stated that investor‘s reaction against new information arrival is reflected in stock price change which indicates expected risks and acquired return.
Public information is responded longer in overnight periods than morning and afternoon trading session. Therefore, return volatility is hypothesized higher during nontrading
period than during trading periods. Inversely, Amihud Mendelson 1991 and Huang, Liu, Fu 2000 proved empirically that return volatility is higher during trading periods
caused by private information arrival. Private information is published during trading periods by the informed traders, and private information is hypothesized that return
during trading periods is higher than during nontrading periods.
A. B.
Examination Stage and Hypothesis
C. This study focuses on examining the existence of private
information based on U-Shaped curve formula. Until recently, this formula is trading model which believe to private information arrival. Essentially,
this formula explain corrected price variance during the early morning trading session Wood, McInish Ord, 1985; Harris, 1986; Andersen
Bollerslev, 1997; Admati Pfleiderer, 1988; Foster Vismanathan, 1990; Slezak, 1994. Private information refers to information that fill two criteria,
namely not in form of publicly known and always related to price Ito Lin, 1992; Ito, Lyons Melvin, 1998. Meanwhile, French Roll 1986 define
that private information is correctly identifiable because it is related to price momentarily or permanently. This study defines and emphasizes that
private information is related to price, so that temporary and permanent impacts are still qualified as private information.
D.
First stage, this study examines private information arrival based on theory shown by French Roll 1986 and Ito, Lyons Melvin
1998 who examined by focusing on the lowest line in U-Shaped curve. This examination is none other than lunch break. This examination uses
lunch break return volatility by comparing closing return variance and opening return variance which must greater than one. Inversely, if the
comparison value is equal to one, this can be considered as public information arrival, which means that return variance does not change from
opening price at morning trading session until closing price at lunch break session. Therefore, private information arrival can be hypothesized as
follows. E.
F. H1: Lunch break return volatility in IDX is caused by private information
arrivals. G.
638
H. Test:
1
C L
O L
V V
I. J.
where,
C L
V
and
O L
V
are closing and opening lunch break return variance. Notation: C close, O open and L lunch.
Second stage, this study analyse the differentiated private information arrival by
examination of return volatility change during morning and afternoon trading session. Specifically, from private information theory, this study predicts the behavior of intraday
return volatility which respond to opening price at morning trading session. By ignoring pricing error due to the inability of pricing error model to predict return variance,
predicted return volatility change is able to confirm private information arrivals.
The examination in this second stage is conducted by assigning model exposed by Admati Pfleiderer 1988. The research suggests that if a number of private
information did not change while the trading drives the change, private information should not cause price change whose return is not distributed during morning trading
session until afternoon. In fact, private information always drives price change which ends up in return distibution all day long. Therefore, this study deduces that return
distribution occurs due to private information captured during trading.
K. H2: Private information arrival is revealed during trading session marked by the decrease of lunch break return volatility bottom line of U-
Shaped curve
L. Test:
O M
O L
C M
C L
V V
V V
and
O A
O L
C A
C L
V V
V V
M. N.
With additional notes from previous test,
O M
V
and
C M
V
are return variance during the opening and closing of morning trading session, and
O A
V
and
C A
V
are return variance during the opening and closing of afternoon trading session, where M morning, and A afternoon.
O.
The prediction of private information arrival can be done by cutting off the trading during morning trading session for the first four
hours. In other words, the trading is limited until lunch break. This cutting off is based on logical framework recommended by Ito, Lyons Melvin
1998. This research
suggested that −if not limited during lunch break− bottom line of U-Shaped curve flattens, it means that U-Shaped during one
full day is not confirmed Slezak, 1994; Hong Wang, 1997. U-Shaped framework in morning trading session cutting off can be hypothesized as
follows.
P. H3: The private information arrival is uncovered during morning trading session marked by increasing return variance during early morning
trading session
−limited until lunch break session− that is able to form U-Shaped curve during one full day
Q. Test:
1
C EM
C MM
V V
and
1
C MM
C LM
V V
and
1
O MM
O LM
V V
R.
639
S. With additional notes from previous tests,
O EM
V
,
C EM
V
,
O MM
V
,
C MM
V
,
O LM
V
and
C LM
V
are return variance during the opening or closing price at early, mid, and late morning trading session. Additional notes: EM early
morning, MM mid morning, and LM late morning.
The examination of private information arrival continues by referring research concepts exposed by Kyle 1995. This research stated that private information is not
related to price in long term. On the contrary, private information should be related only to price in short term because informed traders always choose to do trading as long as
the information is reflected by the price Ito, Lyons Melvin, 1998. Return volatility during short term is suspected whether the opening return variance is higher than
closing return variance. It means that opening return variance during morning trading session affects the return variance during all morning trading session. Moreover,
opening return variance during afternoon trading session should also be determined by return variance during previous morning trading session, because traders are motivated
not to delay their transaction which enlarges return variance during morning trading session. Such characteristic refers to private information model Foster Viswanathan,
1990. This comparative condition can be used to develop hypothesis that private information always occurs during short term as follows.
T. H4: The private information arrival is revealed during trading session in short term when opening return variance ratio is greater than closing
return variance ratio outside lunch break forming the descending line of U-Shaped curve
U. Test:
1
C A
C M
O A
O M
V V
V V
V. W.
With additional notes from previous test,
C M
V
,
C A
V
,
O M
V
and
O A
V
are return variance during opening or closing morning trading session and
afternoon trading session.
3. Research Method X.
The sample in this research is companies listed in LQ45 index during either first or second semester of 2006-2007. LQ45 selection is
based on reasons that companies listed in LQ45 have high liquidity, so this study are able to minimize sleeping stocks during the trading day. The
sleeping stock can affect internal and conclusion validity of this study. This sample selection method is used because IDX is thin market marked by
lots of sleeping stocks. Y.
Z. Return
AA. Opening and closing price for return each 30 minutes interval lay in
trading day which acquired from intraday data. Return is calculated by natural logarithm of relative price R
i,30‟,t
=lnP
i,30‟,t
P
i,30‟-1,t
where i is firm and t is day for each firm. To calculate 30 minutes interval return,
640
companies‟ trading data is divided into 12 intervals, and the formulation is as follows.
Return interval γ0‘number -01 : R
i,16.00t-1-09.30t
= lnP
i,09.30t
P
i,16.00t-1
Return interval γ0‘number -02 : R
i,09.30t-10.00t
= lnP
i,10.00t
P
i,09.30t
Return interval γ0‘ number -03 : R
i,10.00t-10.30t
= lnP
i,10.30t
P
i,10.00t
Return interval γ0‘ number -04 : R
i,10.30t-11.00t
= lnP
i,11.00t
P
i,10.30t
Return interval γ0‘ number -05 : R
i,11.30t-11.00t
= lnP
i,11.30t
P
i,11.00t
Return interval γ0‘ number -06 : R
i,12.00t-11.30t
= lnP
i,12.00t
P
i,11.30t
Return interval γ0‘ number -07 : R
i,13.30t-13.00t
= lnP
i,13.30t
P
i,13.00t
Return interval γ0‘ number -08 : R
i,14.00t-13.30t
= lnP
i,14.00t
P
i,13.30t
Return interval γ0‘ number -09 : R
i,14.30t-14.00t
= lnP
i,14.30t
P
i,14.00t
Return interval γ0‘ number -10 : R
i,15.00t-14.30t
= lnP
i,15.00t
P
i,14.30t
Return interval γ0‘ number -11 : R
i,15.30t-15.00t
= lnP
i,15.30t
P
i,15.00t
Return interval γ0‘ number -12 : R
i,16.00t-15.30t
= lnP
i,16.00t
P
i,15.30t
Trading Session and Return Trading session is not equal during each day. Trading is opened at 09.00 every day, but
the first session is closed at 12.00 on Monday until Thursday, while on Friday the first session is closed at 11.30. the second session is opened at 13.30 on Monday until
Thursday, while on Friday the second session is opened at 14.00. the second session is closed at 16.00 every day. Picture 1 shows trading day and trading period along with
their relation with hypotheses examination in this research. -------------------------------------
Insert Picture 1 about here -------------------------------------
Data Analysis Data analysis was conducted analysis in the following procedural steps:
1. From intra-day data, 12 series of price was obtained that is price within 30 minutes
interval. This 30 minutes interval price was used to calculate return. 2. Calculating return by R
i,γ0‘,t
=lnP
i,γ0‘,t
P
i,γ0‘-1,t
, which is return within minute interval from the first until twelfth. Opening return was calculated by lnP
i,09.30t
P
i,16.00t-1
3. Forming 12 series of 30 minutes interval return from Monday until Friday to determine the sensitivity rate against noise and overreaction. The analysis in this
examination is only focused to differentiate the return in one 30 minutes interval from other returns of 30 minutes interval.
4. Eliminating the days around dividend announcement under the reason to eliminate high price fluctuation H
-3
and H
+3
, and to make adjustment against stock dividend, stock split, bonus share and stock reserve split.
5. Identifying points related to OM: open morning; OEM: open early morning; CEM: close early morning; OMM: open mid morning; CMM: close mid morning; OLM: open
late morning; CM: close morning; CLM: close late morning; OL: open lunch; CL: close lunch; OA: open afternoon; and CA: close afternoon.
6. Calculating
C L
V
,
O L
V
,
C M
V
,
O A
V
,
O EM
V
,
C EM
V
,
O MM
V
,
C MM
V
,
O LM
V
,
C LM
V
,
C M
V
,
C A
V
,
O M
V
and
O A
V
, which consecutively show variance during opening O: open, closing C: close, lunch break L: lunch, morning trading session M: morning, afternoon trading
session A: afternoon, and the two digits letter begining with E, M dan L which represent early morning trading session E: early, mid morning trading session M:
mid and late morning trading session L: late.
7. Examining all hypotheses according to applicable ratio
