Pacific-Basin Finance Journal 7 Ž1999.

495–521
www.elsevier.comrlocatereconbase

Do trading rules impact on market efficiency? A

comparison of opening procedures on the
Australian and Jakarta Stock Exchanges
)
Carole Comerton-Forde
Department of Finance, UniÕersity of Sydney, Sydney 2006, NSW, Australia

Abstract

This paper examines the impact of opening rules on stock market efficiency. In
particular, it contrasts the opening call on the Australian Stock Exchange ŽASX. and the
continuous open on the Jakarta Stock Exchange ŽJSX.. The results suggest that the use of a
call enhances market efficiency by increasing liquidity and lowering volatility at the open.
The results also indicate that some of the benefits associated with a call accrue even when
there is no trading at the call. These results suggest that the use of a call market at the open
may add to the efficiency of the JSX and other similar markets. q 1999 Elsevier Science
B.V. All rights reserved.

JEL classification: G10; G14

Keywords: Microstructure; Opening mechanism; Call markets

1. Introduction

Financial markets around the world utilise different rules and procedures to
open the market. Essentially, there are two mechanisms: call and continuous
openings. In a call market, orders are entered prior to the opening but remain
unexecuted until there is a simultaneous execution of matching orders. In contrast,
continuous trading executes orders as they come to the market, assuming they

)
Tel.: q61-2-8234-5261; fax: q61-2-8234-5450; e-mail: [email protected]

0927-538Xr99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 7 - 5 3 8 X Ž 9 9 . 0 0 0 2 0 - 7
496 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

match with an existing order. A number of markets, such as the London Stock
Exchange, the Nasdaq Stock Market, the Stock Exchange of Singapore ŽSES. and
the Jakarta Stock Exchange ŽJSX., have chosen to open their markets with
continuous trading, whereas the New York Stock Exchange ŽNYSE., the Tokyo
Stock Exchange ŽTSE., the Kuala Lumpur Stock Exchange ŽKLSE. and the
Australian Stock Exchange ŽASX. open trading with a call market.
A growing volume of empirical literature has analysed the opening mecha-
nisms, in a number of markets, to determine how they affect price discovery,
volatility and liquidity at the open. For example, Amihud and Mendelson Ž1987.
and Stoll and Whaley Ž1990. found that the volatility at the opening call on the
NYSE was greater than the volatility at the close. This may suggest that a call at
the open induces volatility. However, this difference is potentially attributable to
the role of the specialists or to the extended non-trading period prior to the call,
rather than the call itself. Amihud and Mendelson Ž1991. overcame this problem
by examining the opening of the TSE. Trading on the TSE is conducted by the
Saitori who, like the NYSE specialists, manage the order book, but may not trade
on their own account. This allowed the authors to remove the impact of the
specialist. The TSE also utilises a call at the opening of the morning session and at
the opening of the afternoon session, allowing the authors to control for the
influence of the preceding non-trading period. They found that the volatility in the
afternoon session was considerably less than the morning session, which suggests
that the volatility may be caused by the non-trading period rather than the call
mechanism.
Chang et al. Ž1995. examined the volatility at the open and the close on the JSX
prior to the automation of the market. They report that the volatility at the open is
greater than at the close. Surprisingly, the paper also finds that approximately 16%
of the total daily trading value is traded at the open.
A number of automated exchanges have also been examined in the literature.
Choe and Shin Ž1993. considered the Korea Stock Exchange ŽKSE. which trades
continuously, except for a call at the open, at the start of the afternoon session and
after the continuous close. Their results indicate that the morning and afternoon
opening calls both exhibit increases in volatility, but the afternoon closing call
displays lower volatility than the continuous morning close. Consistent with
Amihud and Mendelson Ž1991., these results suggest that a call preceding
continuous trading may decrease volatility at the open. More recently, Frino et al.
Ž1996. compared the opening mechanisms used by the ASX and the SES. They
reported that volatility and the bid–ask spread at the opening call on the ASX
were significantly less than the open of the SES which does not use a call to open
the market. This again suggests that a call at the open may enhance market
efficiency. However, this study is limited by the fact that only opening and closing
prices and daily trading volume data were available for the SES. Chang et al.
Ž1999. examined call market trading on the Taiwan Stock Exchange and reported
that this mechanism reduced volatility by approximately 50%.
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 497

Despite the indications from the literature that a call market can enhance price
discovery and liquidity and reduce volatility, particularly in an automated market,
a number of exchanges choose not to open their markets with a call. This suggests
that there is still a degree of uncertainty about the most appropriate way to open a
market. This paper, therefore, builds on the previous literature by considering the
alternative opening mechanisms used by the ASX and the JSX. With the exception
of the call at the open, these markets operate similar trading protocols. Differences
that do exist, such as differences in the tick sizes, are dealt with using a series of
controls. The different mechanisms used by the ASX and JSX provide an
opportunity to determine how a call opening procedure affects market efficiency at
the open.

2. Institutional detail

2.1. The ASX

The ASX operates using the Stock Exchange Automated Trading System
Ž SEATS .. SEATS is a competitive and transparent electronic order book which
trades continuously Žfrom 10:00 to 16:00. from Monday to Friday, except at the
opening and more recently after the continuous close. 1 During the trading day,
SEATS displays a complete order book of unexecuted orders subject to one
exception, namely undisclosed orders. 2 Matching orders are automatically exe-
cuted according to price and time priority. 3 Orders may be amended or with-
drawn, and both market and limit orders may be entered. The minimum price tick

1
In February 1997, the ASX began a series of trial extensions to normal trading hours. Three trials
were conducted. The first involved a closing single price auction market with normal SEATS order
disclosure rules. This was followed in August 1997 with a basic matching facility ŽASX Match. as an
adjunct to SEATS. This operated as an undisclosed fixed price matching process. The final trial
involved an undisclosed single price auction at the close. ŽAustralian Stock Exchange, Circular to
Member Organisations Number 278r97.. Currently, the ASX closes the market randomly between
16:05 and 16:06 with a single price auction. They now run ASX Match at 10:45 and 15:30; however, it
has not been used for some time.
2
ASX Business Rule 2.6Ž5. allows participants to enter hidden or ‘‘undisclosed’’ orders. Prior to 16
October 1996, undisclosed orders could be entered for values in excess of $25,000. However, after this
time, orders needed to be in excess of $100,000.
3
Again, there is an exception to this. ASX Business Rule 2.6Ž13. allows particular types of trades to
be executed off-market. These are special crossings, which are orders in excess of $1,000,000, and
portfolio special crossings, which must consist of at least 10 orders in different securities each with a
value of not less than $200,000. These are referred to collectively as ‘‘off-market trading’’ which
represents approximately 20% of total trading on the ASX during the period of the study.
498 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

varies depending on the price of the stock. Stocks with prices less than 10 cents
have a tick of one-tenth of a cent, stocks priced between 10 and 50 cents have a
tick of half a cent, stocks priced greater than 50 cents have a 1-cent tick and stocks
priced greater than $999 have a tick of $1.
The ASX uses a call market at the opening of trading each morning. The
pre-opening period, which takes place between 7:30 am and the open, allows
orders to be entered, amended and withdrawn but not executed. During the
pre-opening period, brokers can view the complete order book, except for the
volume for orders over $100,000 Žor $25,000 prior to 16 October 1996. which
need not be disclosed. This provides brokers with indicative prices and quantities
prior to the market open and hence assists in the price discovery process.
At the open, orders are executed at a price determined by the opening algorithm
which is explained in detail in Appendix A. 4 This algorithm sets the opening
price using a volume-weighted average price of the orders involved in the trade
resulting in the opening trades taking place at a series of prices. However, the
official opening price is the first trade in the series. The opening algorithm means
that participants can sometimes get a better price than their limit order price. The
algorithm is calculated to the nearest tenth of a cent, allowing for more accurate
price determination during the call.
There are two other distinctive features of the ASX opening procedure that
warrant attention. These are the batch opening and the fact that these batches open
randomly. Since March 1994, the ASX has opened the market in five tranches
based on alphabetical groupings of securities. These groups are shown in Table 1.
Frino and West Ž1997. report that participants ‘‘learn’’ from the trading in the
stocks which open in earlier batches, therefore reducing the uncertainty in the
stocks which open later. This suggests that it may be necessary to control for
differences which may arise for the stocks which open in later batches.
These five tranches are opened randomly up to 15 seconds before or after their
designated opening time. This mechanism is designed to reduce the incentive for
price manipulation at the open and hence, may reduce volatility.

2.2. The JSX

The JSX operates using the Jakarta Automated Trading System Ž JATS .. JATS
is also a transparent electronic order book which trades continuously between 9:30

4
The opening algorithm outlined in Appendix A was used by the ASX until 9 June 1998, at which
time they amended the algorithm in order to reduce the volatility during large openings. The new
algorithm calculates a single volume-weighted average opening price ŽAustralian Stock Exchange,
SEATS Circular to Member Organisations 135r98..
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 499

Table 1
Opening times for stocks on the ASX
First letter of stock code Time of opening
A to B 10:00:00
C to F 10:02:15
G to M 10:04:30
N to R 10:06:45
S to Z 10:09:00

to 12:00 and 13:30 to 16:00 on Monday to Thursday and between 9:30 to 11:30
and 14:00 to 16:00 on Friday. 5 During the period of the study, approximately
65% of trading was done on the Regular Board with the remainder being executed
on the Negotiated Boards. 6 The equivalent of the Negotiated Board on the ASX is
called ‘‘off-market trading’’.
Like the ASX, orders placed on the Regular Board are matched according to
price and time priority. Orders may be amended or withdrawn prior to execution,
but only limit orders may be entered. The minimum price tick for JSX stocks is
IDR 25 and there is also a maximum price movement of IDR 200 on any one
trade. Typically, orders expire at the end of each Exchange Day, although it is also
possible to enter orders which are valid only for one Exchange Session. This
means that there are no orders in JATS each morning at the opening of trading.
There is no pre-opening period and consequently no call opening procedure. 7
During the lunchtime close, the order book remains unchanged as orders may not

5
The longer trading break on Fridays allows the mainly Muslim population to comply with religious
commitments to prayer.
6
For the period of the study, there were five negotiated markets available to investors. These were
the Crossing Board, the Foreign Board, the Block Sales Board, the Odd Lot Board and the Cash Board.
Negotiated Board trades arise from negotiations between brokers and do not compete with the Regular
Board trades and are not automatically matched by the trading system. Trades take place on the
Crossing Board when the same broker represents both the buying and the selling client or is buying or
selling for himself. Trades between foreigners must take place on the Foreign Board when the foreign
ownership limit of 49% has been reached. All trades in excess of 200,000 units must be executed on
the Block Sales Board and trades, which are less than the minimum parcel of 500 units, must be
executed on the Odd Lot Board. Where parties have failed to settle their trades on T q4, they are
required to close out their position in the Cash Market.
7
Despite the fact that there is no pre-opening period, both the Regulations No. II Regarding
Securities Trading and the JATS Guidelines provide for the use of a 30 minute pre-opening period
prior to the opening of the first session. This regulation has never been put into practice. However, the
JSX has power to implement this regulation provided they give one day’s notice to Exchange
Members.
500 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

be amended or withdrawn until the market reopens for the afternoon trading
session.
The difference in the opening procedures used by these two markets provides
an opportunity to examine the impact of a call market on the efficiency of a
market at the open. Other differences in the trading protocols, such as the tick
sizes and opening times, are dealt with using a series of controls which is outlined
in Section 4. The differences in the existing ASX opening algorithm and the
proposed JSX algorithm will also provide a point of comparison if the JSX
chooses to implement these rules, provided everything else remains unchanged.

3. Theoretical considerations

The market opening gives rise to a period of considerable uncertainty due to the
extended non-trading period which precedes it. The purpose of a call market is to
encourage investors to enter the market at this time of uncertainty by enhancing
price discovery ŽEconomides, 1995.. Theoretically, this should lead to increases in
liquidity and a reduction in volatility. A call should facilitate this in a number of
ways. First, if the order book is open, investors are able to obtain indicative prices
prior to the market open. 8 This helps reduce the uncertainty caused by the
overnight non-trading period. Second, by batching orders at the call, the prices
reflect all orders entered into the market and hence the views of multiple
participants ŽMadhavan, 1992.. Batching orders prior to execution can also reduce
the market impact costs associated with large trades ŽEconomides and Schwartz,
1995.. Third, adverse selection costs may be reduced because limit orders cannot
be instantaneously ‘‘picked-off’’ by informed investors ŽSchnitzlein, 1996.. Fourth,
where there is an overlapping spread, bid–ask spread costs are eliminated. Trades
at the call may take place within the bid–ask spread which occurs during
continuous trading because the opening algorithm leads to a weighted average
price Žrounded to the nearest tenth of a cent.. As a result, investors may earn rather
than pay a spread.
If the call achieves the purpose of enhancing price discovery, then investors
will prefer to trade using the call facility and may be more willing to enter the
market early. As a result, the delay between the market open 9 and the first trade

8
Economides and Schwartz Ž1995. suggest that allowing orders to be amended or withdrawn during
the pre-opening period without imposing a penalty can lead to gaming strategies.
9
The market open is assumed to be the time when it becomes possible to execute trades on-market.
For the JSX, this is always 9:30, while for the ASX this time will vary as outlined in Section 2.1.
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 501

should be reduced and the liquidity at the open should increase. This leads to
Hypotheses 1 and 2.

H 1 : The time between the open and the first trade will be less for the ASX than for
the JSX, ceteris paribus.

and

H 2 : The Õolume traded at the open will be higher for the ASX than the JSX,
ceteris paribus.

The call is likely to be more important for illiquid stocks if it provides more
efficient price discovery. There are a number of reasons for this. First, liquid
stocks generally receive a great deal of attention from the large brokerage houses
ŽArbel and Strebel, 1982.. As a consequence, there is much more information
available about these stocks and hence less uncertainty about price. This also leads
to less adverse selection costs in liquid stocks. The converse will be true for
illiquid stocks. Second, illiquid stocks tend to trade less frequently and hence the
last traded price is less likely to be a good indicator of the true price. Therefore,
investors trading in the illiquid stocks will take advantage of the opportunity for
price discovery offered by the call by trading at the open. This suggests that the
call is more advantageous for illiquid stocks. 10 If this is the case, then the
difference between the delay in trading across liquid and illiquid stocks will be
smaller for the ASX than the JSX. This leads to Hypothesis 3.

H 3 : The difference in the delay in trading across liquid and illiquid stocks will be
less for the ASX than the JSX, ceteris paribus.

If the call facilitates price discovery, then the percentage of daily volume traded
in each stock at the open on the ASX will increase as stock liquidity decreases.
This gives rise to Hypothesis 4.

H 4 : The percentage of daily Õolume traded at the open will increase as stock
liquidity decreases for ASX stocks.

If the call assists price discovery by providing indicative prices prior to the
opening, then the uncertainty associated with the market opening would be

10
Recent work by Chang et al. Ž1999. on the Taiwan Stock Exchange found that the call was more
effective in reducing the volatility of liquid rather than illiquid stocks. However, they attribute this to
two unique features of the Taiwan Stock Exchange, namely, the dominance of individual investors who
are usually uninformed and a lack of transparency and disclosure by listed companies.
502 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

reduced and non-information-related volatility would fall. 11 As a result, a market

which uses a call will have lower non-information-related volatility at the open
than at other times of the day. This leads to Hypotheses 5 and 6.

H 5 : Non-information-related Õolatility will be less for the ASX than for the JSX,
ceteris paribus.

and

H 6 : Non-information-related Õolatility at the open will be less than at the close

for the ASX, ceteris paribus.

4. Data and research method

The data used in this study are taken from the SEATS and JATS databases
maintained by the Securities Industry Research Centre of Asia-Pacific Ž SIRCA..
The SEATS and JATS databases provide complete details of all order and trade
records placed on the ASX and JSX. These records provide details of price,
volume, date, time and broker for every order and trade.
Using the data from 14 November 1995 to 30 June 1997, a sample of 264 JSX
and 1368 ASX stocks is obtained. During this period, the JSX had turnover of
approximately IDR 143.88 trillion ŽAUD 79.93 billion. while the ASX had
turnover of approximately AUD 284.56 billion. This represents approximately
3,344,497 and 8,096,533 trades, respectively. This suggests that there may be a
number of other factors affecting liquidity which need to be controlled for.
To help control for such differences, a matched-pairs design is employed. The
matching is performed on the basis of the average number of trades executed in a
stock between 14:30 and 15:30. 12 This interval is chosen because it is the only
interval which is at least 30 minutes before or after a non-trading period in both
markets. The top 15 ASX stocks are removed from the sample as the trading
frequency in these stocks exceeds that of the top JSX stock. The top 100 JSX
stocks are then matched to the closest matched ASX stock. The final sample
consists of 100 stocks in each market. Stocks are also eliminated from the sample
if they were not listed on 14 November 1995 and did not trade for at least 1 year
after this date.

11
This has been supported by a number of papers including ŽAmihud and Mendelson, 1987, 1991.
and ŽEconomides and Schwartz, 1995..
12
The robustness of this matching is tested by using pairs matched on average daily trading volume
and average daily trading frequency. The results using this measure are generally consistent with the
14:30 to 15:30 trade frequency measure and are therefore not reported.
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 503

Only Regular Board trades are included from the JSX as these are the only
trades which are executed through order matching. For the same reason, only
on-market trades 13 are included in the ASX analysis.
The impact of a call market on market efficiency at the open is measured using
two market quality indicators: liquidity and non-information related volatility.
Efficiency is enhanced by increases in liquidity and reductions in transactions
costs and non-information related volatility. The previous literature has indicated
that the behaviour of these variables at the open is significantly different from the
behaviour at the close. This paper explores the issue further by examining liquidity
and non-information related volatility at the open of the JSX and ASX, using a
series of proxies outlined below.

4.1. Liquidity

Liquidity is operationalised using five metrics: the timing of opening trades,

volume as a percentage of total daily volume Žrelative volume., trade frequency as
a percentage of the total daily trade frequency Žrelative trading frequency., relative
and tick Žnumber of ticks. bid–ask spreads.

4.1.1. Timing of the opening trades

The delay between the market open and the first trade in each stock is
calculated and averaged across each stock. This is compared to the delay in its
matched pair. The nature of the ASX call means that when there is a trade at the
call, this will be at the instant the market opens. For this reason, the frequency
with which ASX stocks trade at the call is calculated and averaged across stocks.
Where there is no trade at the call, the difference between the market open and the
first trade after the call is examined.
In addition, the stocks in each market are divided into quartiles according to
liquidity, proxied by trading frequency between 14:30 and 15:30. 14 This allows
Hypothesis 3 to be tested. The average time between the open and the first trade is
then calculated for each quartile to determine whether the liquidity of a stock
affects the timing of the opening trade.

4.1.2. Intraday patterns in Õolumes and bid–ask spreads

To determine whether there is any systematic difference between the volume
traded at the open in either market, the volume as a percentage of total daily
volume Žrelative volume. is calculated during the first and last hours of trading

13
Like the Regular Board trades on the JSX, on-market trades on the ASX are executed according to
price and time priority, whereas off-market trades are negotiated between brokers.
14
Again, the robustness of the results is tested by dividing the stocks into quartiles based on the
average daily volume and average daily trading frequency with no significant change in the results.
504 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

using 1, 5, 10 and 30 minute intervals. This allows for a comparison of the impact
of the opening mechanism on the two markets. Trading frequency as a percentage
of total daily trading frequency Žrelative trading frequency. is also calculated to
determine whether patterns in volume are driven by changes in trade sizes or a
change in trading frequency. Due to the fact that the ASX is open for trading for 6
hours whereas the JSX is open for only 5 hours Žexcept on Fridays, when it is
open for 4 hours., the average percentage traded in each interval will be higher on
JSX than the ASX. This will bias against the hypothesised result of higher trading
volume at the open on the ASX than the JSX. This is not considered to be
problematic because each 30 minute interval represents about 8% of the trading
day on the ASX and 10% of the trading day on the JSX. Therefore, the fraction
traded in any given interval should be lower on the ASX than the JSX, all else
being equal.
Previous research examining bid–ask spreads at the open and close has relied
on Roll Ž1984. to calculate implied spreads. However, both the ASX and JSX data
allow actual spreads to be examined. 15 The relative time-weighted bid–ask spread
is calculated across 1, 5, 10 and 30 minute intervals during the first and last hours
of trading and in the hour after lunch for the JSX. Due to the fact that the two
markets have different minimum tick sizes, the tick spread Žnumber of ticks. is
also calculated across the same intervals.
Each of the intraday liquidity metrics is calculated for each stock. Stocks are
again divided into groups based on liquidity and the average results are generated.
This allows Hypothesis 4 to be tested.
Some difficulties arise when comparing the patterns on the ASX and JSX. First,
the markets open at different times. Second, different stocks on the ASX open at
different times and as a result grouping stocks together means that some stocks in
the sample may be trading while others are not. These problems may be addressed
by aligning the opening of the two markets at time zero. Each metric is examined
across each interval during the first hour of trading. Similarly, each metric is
examined across each stock during the last hour of trading. A short event window
is used to ensure that the impact of the opening is isolated.

4.2. Volatility

4.2.1. Open-to-open and close-to-close returns

Previous research dealing with the impact of the trading mechanism on price
volatility has examined open-to-open and close-to-close returns. 16 When exam-

15
Rhee and Wang Ž1998. suggest that the major components of bid–ask spreads in order-driven
markets without market makers are adverse selection costs and order processing costs. There are no
inventory holding costs.
16
For instance, Amihud and Mendelson Ž1987., Stoll and Whaley Ž1990. and Amihud and Mendel-
son Ž1991..
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 505

ined over the same time period, these measures allow the impact of public
information to be removed. This means that differences in the open-to-open and
close-to-close returns are more likely to be associated with the particular opening
procedure. Accordingly, the open-to-open and close-to-close returns are calculated
for the ASX and JSX stocks. 17
Consistent with the previous research, volatility at the open on the ASX and the
JSX is calculated using a variance ratio of the open-to-open and close-to-close
returns. A variance ratio greater than one indicates that the volatility at the open is
greater than the volatility at the close. If the call reduces the volatility at the open,
then the variance ratio for the ASX will be less than for the JSX.
Due to the fact that the ASX opening algorithm allows trades to occur inside
the bid–ask spread, bid–ask bounce-induced volatility is reduced at the call. This
may lead to a lower variance ratio on the ASX. On the other hand, the opening
algorithm induces volatility of opening prices as a result of the sequential order
matching procedure, as well as providing an incentive to delay orders until after
the market opening ŽBrown et al., 1997.. This may lead to greater volatility at the
open and a longer delay between the open and the first trade. Therefore, the
variance ratio of open-to-open and close-to-close midpoint returns is also calcu-
lated. This is designed to remove the impact of bid–ask spreads on the variance
ratio, therefore isolating the impact of the call on volatility.

4.3. SensitiÕity analysis

A series of sensitivity tests is conducted to control for factors, other than the
call, which may cause differences in liquidity and volatility at the open of the ASX
and JSX.

4.3.1. Controlling for the impact of the ASX batch opening

Frino and West Ž1997. report that the batch opening on the ASX reduces the
volatility and transactions costs of stocks which open in later batches. They
suggest that this occurs because participants are able to ‘‘learn’’ from trading in
the stocks which open earlier. Therefore, the inclusion of stocks in the ASX
sample which opens in later batches may bias the ASX results, leading to
artificially low volatility and high liquidity. To control for this, the ASX sample is
partitioned to identify stocks which open in the first batch. Each of the metrics
outlined above is re-examined for this sub-sample of stocks and their JSX matched
pair.

17
For the ASX ŽJSX., the opening price is defined as the first on market ŽRegular Board. trade of the
day and the closing price is the last on market ŽRegular Board. trade of the day.
506 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

4.3.2. Controlling for Õolatility

To control for the possibility that differences in the liquidity and volatility
metrics across the quartiles are driven by differences in the underlying volatility
across stocks, the two samples are sorted according to overnight returns and are
again divided into quartiles based on these returns. Again, each of the metrics is
recalculated for these samples.

4.3.3. Controlling for cross-listed stocks

Stocks, which are cross-listed on other markets such as the NYSE and Nasdaq,
are likely to exhibit lower volatility at the open than stocks which are not. This is
due to the fact that price discovery continues to take place in these stocks
overnight. For this reason, it is necessary to examine these stocks and in particular
to ensure that these stocks do not bias the results.

4.3.4. Trading at and immediately after the call

The relative volume and trading frequency metrics discussed above do not
isolate the trading which takes place at the call and the trading that takes place in
the first 10 minutes after the call. These two elements are therefore separated and
reported for the ASX across each quartile. An examination of the quartile results
provides a further opportunity for testing Hypothesis 4.

4.3.5. ASX stocks not trading at the call

Despite the hypothesised liquidity offered by the call, not all stocks trade at the
call on the ASX every day. Therefore, it is possible to isolate the days when there
is trading at the call, those when there is no trading at the call, and those when
there is no change in the order schedule prior to the call. If the call drives the
hypothesised changes in liquidity and volatility at the open, then the removal of
trading at the call should reduce or eliminate these changes. Therefore, each of the
intraday metrics discussed above is recalculated for the three new samples.
The ASX sample which contains the days when there is trading at the call
should exhibit a greater increase in liquidity and reduction in volatility than the
whole sample. Where there is no trading at the call, price discovery is reduced.
However, it is not eliminated as participants can obtain information about the
stock price from the change in the order schedule during the pre-opening period.
Therefore, the increase in liquidity and reduction in volatility at the open should be
less than the sample with trading at the call, but greater than the JSX trading.
Identifying instances where there is no trading at the call and no changes in the
order schedule prior to the pre-opening replicates the environment at the open on
the JSX. Where this occurs, there should be no price discovery prior to the open
and hence no increase in liquidity or reduction in volatility. One difficulty
associated with this measure, however, is that the ASX selection is biased on the
most illiquid stocks in the sample. It may also be biased toward stocks with no
overnight information announcements, which may lead to lower volatility.
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 507

5. Results

5.1. Sample statistics

The final sample of 100 stocks in each market represents approximately 80% of
the volume traded on the JSX and 40% of the volume traded on the ASX. Table 2
provides summary information about the trading frequency, volume and value Žin
AUD. for each sample and for each of the quartiles. This illustrates that the
matching process is reasonable. The frequency of trading stocks on the JSX is
slightly above the ASX, so any illiquidity in these stocks at the open on the JSX is
not likely to be caused by differences in the samples.

5.2. Liquidity

5.2.1. Timing of opening trades

Examination of the timing of opening trades for each matched pair reveals that
the delay between the open and the first trade is consistently greater on the JSX
than the ASX. Table 3 provides a summary of the average time between the
market opening and the first trade on the ASX and the JSX. Notice that the delay
is significantly smaller on the ASX. The average time to the first trade on the ASX
is approximately 9 minutes compared to 36 minutes on the JSX. 18 This supports
Hypothesis 1.
The results also indicate that the delay between the open and the first trade, on
both markets, falls as the liquidity of the stock rises. The differences in the delays
across quartiles are statistically significant. Consistent with Hypothesis 3, the
difference in the delay across liquid and illiquid stocks is less for the ASX than for
the JSX. The differences in the delays for the ASX sample appear to be partially
attributable to the frequency with which stocks trade at the call. Notice that the
Quartile 1 stocks trade at the call approximately 79% of the time compared to
approximately 49% for the Quartile 4 stocks.

5.2.2. Intraday patterns in Õolumes and bid–ask spreads

The intraday patterns in relative volumes, relative trading frequency, relative
bid–ask spreads and tick bid–ask spreads clearly indicate that there is more
liquidity at the open on the ASX than the JSX. These results are consistent across
all interval sizes and therefore only the results using 10 minute intervals are
reported. Fig. 1 shows that approximately 7.4% of the average daily volume on the

18
Even if 5 to 10 minutes are allowed for JSX brokers to re-enter orders from the previous day, the
delay between the open and the first trade is still greater on the JSX.
508 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

Table 2
Summary trading statistics
This table provides details of the average trading frequency, volume and value ŽAUD. between 14
November 1995 and 30 June 1997 for the sample of 100 JSX and ASX stocks. Summary trading
statistics is also provided across quartiles based on the average trading frequency between 14:30 and
15:30.
Trades Volume Value ŽAUD.
ASX all stocks 5980 65,200,481 168,696,004
ASX Quartile 1 3034 27,874,453 104,934,573
ASX Quartile 2 1398 12,749,169 37,194,763
ASX Quartile 3 942 13,036,803 19,007,841
ASX Quartile 4 605 11,540,056 7,558,827
JSX all stocks 6265 96,185,430 99,565,598
JSX Quartile 1 3346 60,489,894 62,503,751
JSX Quartile 2 1400 18,219,523 21,570,696
JSX Quartile 3 969 12,136,033 10,677,007
JSX Quartile 4 550 5,339,980 4,814,144

ASX is traded within the first 10 minutes. By comparison, the JSX only trades
2.4% of the average daily volume within the first 10 minutes, including the call.
T-statistics shows that this difference is statistically significant. The relative
volume traded during the first hour is consistently greater on the ASX; however,
the magnitude and significance of the difference between the markets fall after the
first 10 minute interval, although it increases again during the last hour. These
results support Hypothesis 2, that liquidity at the open on the ASX is higher than
the JSX. The results for the relative trading frequency display a similar pattern and
therefore are not shown.

Table 3
Time between market open and first trade on the ASX and the JSX
This table shows the average time, in minutes, between the market open and the first trade on the ASX
and the JSX. This information is provided for the entire sample and across quartiles based on liquidity.
For the ASX, the frequency with which there is trading at the call is also reported.
JSX ASX Difference between T-statistics Frequency of trading
JSX and ASX at the ASX call Ž%.
All stocks 35.47 8.95 26.52 18.72UU 60.94
Quartile 1 22.35 2.48 19.86 13.14UU 79.16
Quartile 2 33.56 6.56 27.00 18.32UU 59.84
Quartile 3 39.55 10.11 29.45 12.68UU 55.84
Quartile 4 46.41 16.64 29.77 12.39UU 48.92
UU
Significant at the 0.01 level.
U
Significant at the 0.05 level.
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 509

Fig. 1. Comparison of relative volumes across the JSX and the ASX. This figure provides details of the
relative volumes Žpercentages. traded on the JSX and the ASX during the first and last hour of trading
using 10 minute intervals. U Indicates significance at the 0.05 level. UU Indicates significance at the 0.01
level.

Fig. 1 also shows that the relative volume traded during the last hour is
consistently greater on the JSX; however, this difference is greatest during the last
10 minute interval. During this interval, 14.2% of daily volume on the JSX is
traded. This is almost twice the percentage traded on the ASX. These results
suggest that JSX investors prefer to trade at the end of the day. This is likely to be
due to the fact that orders build up during the day and therefore there is an
increased possibility of trading.
Fig. 2 displays the patterns in the relative bid–ask spreads for the ASX and
JSX. These results show that the relative bid–ask spread during the first 10
minutes is 5.3% on the JSX compared to 1.3% on the ASX. However, the results
also reveal that the relative bid–ask spread is consistently higher, throughout the
day, on the JSX. This may be driven by the minimum tick size on the JSX and
hence tick spreads Žthe number of ticks. are reported in Fig. 3. These results show
that the tick spreads on the JSX are significantly larger during the first hour of
trading. The greatest difference of approximately 1.5 ticks occurs during the first
10 minute interval. The difference declines in magnitude and significance through-
out the first hour. During the last hour of trading, T-statistics shows that there is no
significant difference in the tick spreads.
An analysis of the relative volumes and bid–ask spreads across the four
quartiles allows Hypothesis 4 to be tested. Fig. 4 shows that the relative volume
traded in ASX stocks during the first interval declines as the liquidity of the stock
rises. This indicates that investors take advantage of the liquidity provided by the
call by trading at this time, thus supporting Hypothesis 4. T-statistics show that the
differences between the quartiles are significant during the first 10 min interval.
510 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

Fig. 2. Comparison of relative bid–ask spreads across the JSX and the ASX. This figure provides
details of the relative bid–ask spreads Žpercentages. on the JSX and the ASX during the first and last
hour of trading using 10 minute intervals. U Indicates significance at the 0.05 level. UU Indicates
significance at the 0.01 level.

In contrast, patterns in relative trading frequency do not exhibit the same

results. Fig. 5 shows that there is no significant difference in the relative trading
frequency across the different quartiles. This suggests that there may be a
reduction in average size of trades in liquid stocks at the open on the ASX. This is
consistent with Brown et al. Ž1997., who report a reduction in the average trade

Fig. 3. Comparison of tick bid–ask spreads across the JSX and the ASX. This figure provides details of
the tick bid–ask spreads Žnumber of ticks. on the JSX and the ASX during the first and last hour of
trading using 10 minute intervals. U Indicates significance at the 0.05 level. UU Indicates significance at
the 0.01 level.
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 511

Fig. 4. Relative trading volumes on the ASX across quartiles. This figure exhibits the patterns in
relative trading volumes across the four quartiles for the ASX sample using transaction time, in 10
minute intervals across the first and last hour of trading.

and order value after the introduction of the current ASX opening algorithm. They
find that their results are stronger for liquid stocks on the ASX. They argue and
offer evidence that the ASX opening algorithm provided an incentive for investors
to split their pre-opening orders, particularly in liquid stocks.
An analysis of the relative bid–ask spreads across the ASX quartiles reveals
that on average bid–ask spreads increase as the stock liquidity declines. This is

Fig. 5. Relative trading frequency on the ASX across quartiles. This figure exhibits the patterns in
relative trade frequency across the four quartiles for the ASX sample, in 10 minute intervals across the
first and last hour of trading. Relative trade frequency is calculated by dividing the number of trades in
a given interval by the number of trades throughout the day.
512 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

Fig. 6. Relative bid–ask spreads on the ASX across quartiles. This figure exhibits the patterns in
relative bid–ask spreads across the four quartiles for the ASX sample, in 10 minute intervals across the
first and last hour of trading.

illustrated in Fig. 6. However, Table 4 shows that the percentage difference

between the bid–ask spread in the first 10 minute interval and the average bid–ask
spread during the other intervals declines as stock liquidity declines. In fact, the
difference is only significant for the first two quartiles. This also supports
Hypothesis 4, that the call is more beneficial for illiquid stocks.
A similar analysis of trading in JSX stocks shows that the relative volume
traded at the open is not affected by stock liquidity. This is shown in Fig. 7. The
results for relative trading frequency are similar and hence are not reported.
Similar to the ASX results, the relative bid–ask spreads on the JSX increase as
the stock liquidity declines. This is shown in Fig. 8. However, unlike the ASX,

Table 4
Difference in relative spreads on the ASX during the first interval and other intervals
This table displays the average relative spread during the first 10 min and the average during other 10
minute intervals during the first and last hours of trading on the ASX. The percentage difference is
calculated as the difference in the spreads divided by the average.
Relative Average Percentage T-statistics
spread in relative spread difference
morning in remaining in relative
interval 1 intervals spreads Ž%.
Quartile 1 0.570 0.437 30.53 4.1427
Quartile 2 0.986 0.727 35.65 4.0551
Quartile 3 1.422 1.173 21.25 1.6044
Quartile 4 2.133 1.908 11.79 0.8799
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 513

Fig. 7. Relative volumes on the JSX across quartiles. This figure exhibits the patterns in relative
volumes across the four quartiles for the JSX sample using transaction time, in 10 minute intervals
across the first and last hour of trading.

Table 5 shows that the difference in the relative bid–ask spread in the first 10
minute interval and the average bid–ask spread across the other intervals does not
decline with stock liquidity.
5.3. Volatility
5.3.1. Open-to-open and close-to-close returns
Table 6 shows the ratio of open-to-open and close-to close return variances for
the trade to trade returns. These results show that the trade-to-trade return variance

Fig. 8. Relative bid–ask spreads on the JSX across quartiles. This figure exhibits the patterns in relative
bid–ask spreads across the four quartiles for the JSX sample, in 10 minute intervals across the first and
last hour of trading.
514 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

Table 5
Difference in relative spreads on the JSX during the first interval and other intervals
This table displays the average relative spread during the first 10 minutes and the average during other
10 minute intervals during the first and last hours of trading on the JSX. The percentage difference is
calculated as the difference in the spreads divided by the average.
Relative Average Percentage T-statistics
spread in relative spread difference
morning in remaining in relative
interval 1 intervals spreads Ž%.
Quartile 1 3.758 2.580 45.64 2.1892
Quartile 2 4.846 3.351 44.62 2.1637
Quartile 3 5.866 3.738 56.94 4.2087
Quartile 4 6.629 4.408 50.38 5.4801

ratio is 1.19 for the JSX and 0.98 for the ASX. This indicates that the volatility at
the open on the JSX is 19% greater than the close. In contrast, volatility at the
open on the ASX is 2% less than the close. These results support Hypothesis 5,
that non-information-related volatility is less for the ASX than the JSX and
Hypothesis 6, that non-information-related volatility at the open is less than at the
close for the ASX. Also notice that the volatility at the open on the JSX increases
as the liquidity of the stocks declines while the converse is true for the ASX.
However, these differences are statistically insignificant in both markets. These
differences in variances are consistent when the mid-point returns rather than
actual returns are used.

5.4. SensitiÕity analysis

5.4.1. Controlling for the impact of the ASX batch opening

The results for the sub-sample of ASX stocks, which open in the first batch and
their matched pairs, are consistent with the whole sample and are therefore not

Table 6
Ratio of open-to-open and close-to-close return variances
This table shows the ratio of open-to-open and close-to-close return variance for the ASX and the JSX.
Both trade-to-trade and mid-point return variances are calculated.
All stocks Quartile 1 Quartile 2 Quartile 3 Quartile 4
Ratio of open-to-open and close-to-close trade return variances
JSX 1.19 1.13 1.16 1.18 1.28
ASX 0.98 1.02 0.99 0.97 0.96
Difference 0.21 0.11 0.18 0.21 0.33
T-statistics 7.11UU 2.67U 3.25UU 3.42UU 4.65UU
UU
Significant at the 0.01 level.
U
Significant at the 0.05 level.
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 515

exhibited. This suggests that the differences in liquidity and volatility at the open
of the ASX and JSX are not caused by the ‘‘learning’’ from trading which occurs
with the batch opening on the ASX.

5.4.2. Controlling for Õolatility

When stocks are partitioned according to volatility, there is no statistically
significant difference in any of the liquidity or volatility metrics across the
samples and therefore these results are not shown.

5.4.3. Controlling for cross-listed stocks

There are two cross-listed stocks in the ASX sample and three in the JSX
sample. Although these stocks exhibited slightly lower volatility than the whole
samples Ž0.97 for the ASX and 1.12 for the JSX., the JSX remains more volatile at
the open than the ASX.

5.4.4. Trading at and immediately after the call

Table 7 illustrates the relative volumes traded at the call and the 10 minutes
after the call for the ASX stocks. It shows that on average 2.4% of trading volume
takes place at the call. The remaining 5% of trading executed during the first 10
minutes remains double the relative volume traded on the JSX. This suggests that
the call not only attracts participants to the market for the call but also increases
liquidity immediately after the call. This may be due to the resolution of
uncertainty by the call.
The quartile results shown in Table 7 reveal that the difference in the relative
volume traded during the first 10 minutes of trading is caused by the volume
traded at the instant of the call. As the liquidity of the stock declines, the relative
volume traded at the call increases significantly, while there is only a relatively
small increase in the volume traded during the subsequent 10 min. This provides
further evidence that a call opening is more important for illiquid stocks.

Table 7
Relative trading volume at the call and interval 1
This table compares the relative volume Žpercentages. traded at the call and during the first 10 minutes
of trading after the call on the ASX.
Relative volume Relative volume Combined
at the call in interval 1 relative volume
All stocks 2.424 4.981 7.405
Quartile 1 1.277 4.587 5.864
Quartile 2 1.854 4.470 6.324
Quartile 3 2.938 5.192 8.130
Quartile 4 3.627 5.674 9.301
516 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

5.4.5. ASX stocks not trading at the call

Fig. 9 shows the difference in relative volumes traded on ASX in three samples
based on: stocks which traded at the call; those that did not trade at the call; and
those that had no change in the order book during the pre-opening period. An
analysis of the days when there is trading at the call reveals that approximately
9.4% of relative volume is traded during the first 10 minutes of trading. Even
when there was no trading at the call, an increase in relative trading volume to
approximately 4% of daily volume was observed. This suggests that the pre-open-
ing period provides an opportunity for price discovery even when there is no
trading at the call due to the fact that orders may be entered, amended or
withdrawn at this time. On the days when there was no trading at the call and no
change in the order schedule during the pre-opening period, there was a reduction
in the relative volume traded at the open to approximately 1.4% of daily volume.
The relative volume traded in the first 10 minutes for this sample is statistically
less than the comparable JSX sample of 2.6%. This difference is likely to be
caused by the difference in the number of trading hours on the two markets and by
the fact that the ASX sample is biased towards the most illiquid days in the
illiquid stocks.
An examination of the relative bid–ask spreads across the three samples shows
results consistent with those for relative volumes and therefore are not displayed.
The relative volumes are also examined across the quartiles for the sample of
stocks where there is no trading at the call and where there is no change in the

Fig. 9. Comparison of relative volumes on the ASX openings with and without calls. This figure
compares the relative volumes traded on the ASX across three samples. The ‘Call’ sample only
includes those days when there is a trade at the call. The ‘No call’ sample only includes stocks on the
days where there is no trading at the call. The ‘No change to order schedule’ sample only includes
stocks on the days when there is no trading at the call and no change in the order schedule during the
pre-opening period.
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 517

Fig. 10. Relative volume traded at the call when there is no trading at the call on the ASX. This figure
shows the relative volumes traded when there is no trading at the call on the ASX. These results are
partitioned across quartiles based on liquidity.

order book during the pre-opening period. Fig. 10 shows that the sample of stocks
where there is no trading at the call continues to exhibit the same patterns as the
entire sample. That is, the relative volume traded during the first 10 minutes
increases as the liquidity of the stock declines. This again supports Hypothesis 4 as
it suggests that the price discovery which takes place during the pre-opening
period is more beneficial for illiquid stocks.

Fig. 11. Relative volume traded at the call when there is no change in the order schedule during the
pre-opening period on the ASX. This figure shows the relative volumes traded when there is no trade at
the call and no change in the order schedule during the pre-opening period on the ASX. These results
are partitioned across quartiles based on liquidity.
518 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

In contrast, Fig. 11 shows that this pattern disappears when there is no trading
at the call and no change in the order schedule during the pre-opening period. This
pattern is consistent with that of the JSX stocks. 19

6. Conclusions

This paper compares opening rulesrprocedures used by the ASX and the JSX
and reports that the choice of opening procedure has significant implications for
liquidity and volatility at the opening of trading. The results reveal that the time
taken to trade after the open is on average 27 minutes shorter on the ASX than the
JSX. It also demonstrates that the relative volume and trading frequency during the
first 10 minutes is on average 5% greater on the ASX, which is more than three
times the relative volume and trading frequency on the JSX. The relative spreads
during the first 10 minutes are also considerably lower on the ASX. The extent to
which this is driven by the call market is difficult to quantify due to the fact that
bid–ask spreads on the ASX are systematically lower than the JSX. However, the
tick spreads reveal that the difference in the two markets is greatest during the first
10 minutes of trading and falls throughout the first hour of trading. These results
indicate that liquidity is enhanced by a call market at the open.
The ASX results reveal that the increase in liquidity at the open is greatest for
the illiquid stocks. They also show that there is an increase in liquidity at the open
even when there is no trading at the call, provided there are changes in the order
schedule prior to the call. This suggests that price discovery takes place during the
pre-opening period even without any trading at the call. This has important
implications for designing a call market opening procedure.
Consistent with Frino et al. Ž1996. and Chang et al. Ž1999., the results reveal
that volatility is reduced with a call market. Volatility at the open of the JSX is
19% greater than at the close. This compares to the ASX where volatility is 2%
less than the close. The lower volatility at the open on the ASX is partially caused
by the fact that trades at the call may occur within the best bid and ask. However,
even after adjusting for this, the volatility on the ASX remains significantly lower
than the JSX. These results also support the use of a call market at the open.
This paper illustrates that trading rules have a considerable impact on market
efficiency and therefore deserve careful attention from market regulators. It
suggests that the efficiency of the JSX at the open may be enhanced by the
introduction of a call opening mechanism and by carrying forward orders from the

19
An examination of the sample of the stocks, which do not trade at the call, reveals that they tend to
be the illiquid stocks. However, there is no evidence to indicate that these stocks are less likely to have
overnight information announcements. In fact, 9.5% of the days that had no trading at the call had
overnight information announcements compared to only 7% of the days which did trade at the call.
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 519

previous day. This increase in efficiency is likely to be greatest for the illiquid
stocks.

7. Future research

A number of areas of future research are identified by this paper. Differences in

the intraday patterns in volume on the ASX and JSX suggest that a more
comprehensive review of all intraday trading patterns on the JSX is required.
Differences in the trading rules used by the two markets appear to give rise to
different intraday patterns. Future research may consider the use of the Negotiated
Boards and the maximum price movement for any single trade. Further considera-
tion should also be given to the behaviour of informed and uniformed investors
and the factors driving the large increase in volume at the end of the trading day.
The bid–ask spreads on the JSX also appear to deserve further attention.
Bid–ask spreads for the sample of stocks used in this paper range between 2 and
7%. These spreads are consistently higher than the ASX, across all groups of
stocks, throughout the day. It is likely that these large spreads are driven by the
minimum tick size, of IDR 25. However, further research is required to determine
the extent of this influence.

Acknowledgements

The author is indebted to the JSX, the ASX and Securities Industry Research
Centre of Asia-Pacific ŽSIRCA. for data. The author is grateful to Michael Aitken,
Ghon Rhee, the Futures Research Centre and an anonymous referee for their

Table 8
Order book for an ASX stock prior to the market open
Buy Sell
Broker Quantity Price Žcents. Price Žcents. Quantity Broker
AAA 1000 500 497 1500 GGG
BBB 2000 499 498 2500 HHH
CCC 1500 499 499 1500 III
DDD 500 498 500 500 JJJ
EEE 500 496 500 1000 KKK
FFF 2000 492 1000 LLL
500 MMM
520 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521

Table 9
Calculation of trade price for opening trades
Trade number Buyer Seller Quantity Price calculation Trade price
1000=500q1500=497
1 AAA GGG 1000 498.2
2500
2000=499q500=497
2 BBB GGG 500 498.6
2500
1500=499q2500=498
3 BBB HHH 1500 498.375
4000
1500=499q1000=498
4 CCC HHH 1000 498.6
2500
5 CCC III 500 limit prices are equal 499

helpful comments and suggestions. The author wishes to thank SMARTS Pty
Limited, especially Thomas Jones for programming assistance.

Appendix A. Opening price calculation on the ASX prior to 9 June 1998

The pre-opening period prior to the call in the ASX may result in overlapping
bid and ask orders Ži.e., a negative bid–ask spread.. Therefore, when the market
opens, overlapping bids and orders are executed using an opening algorithm.
Since 9 June 1998, this algorithm has been a single volume weighted average
opening price, however, during the period covered by this paper, a multiple price
algorithm was used. This multiple price algorithm sequentially traded the highest
priority bid against the highest priority ask until all overlapping orders were
executed. These trades were executed at a weighted average price of the orders
involved in the trades. Subsequent to this, the market opened for continuous
trading.
This process may be explained through the use of an example. Assume that the
order book for an ASX stock prior to the market open is shown in Table 8.
When the market opened, the trades shown in Table 9 would have taken place.
The official opening price was the price of the first trade in the series, $4.982.
The unexecuted orders remained in the order book after the pre-opening period
and maintained their price and time priority.

References

Amihud, Y., Mendelson, H., 1987. Trading mechanisms and stock returns: An empirical investigation.
Journal of Finance 42, 533–553.
Amihud, Y., Mendelson, H., 1991. Volatility, efficiency, and trading: Evidence from the Japanese
stock market. Journal of Finance 46, 1765–1789.
 C. Comerton-Forder Pacific-Basin Finance Journal 7 (1999) 495–521 521

Arbel, A., Strebel, P., 1982. The neglected and small firm effects. The Financial Review 17 Ž4.,
201–218.
Brown, P., Harris, L., Wee, M., 1997. The incentive to split pre-opening orders on the Australian Stock
Exchange. Working Paper 97–95, Department of Accounting and Finance, University of Western
Australia.
Chang, R.P., Rhee, S.G., Soedigno, S., 1995. Price volatility of Indonesian stocks. Pacific-Basin
Finance Journal 3, 337–355.
Chang, R.P., Hsu, S.T., Huang, N.K., Rhee, G.S., 1999. The effects of trading methods on volatility
and liquidity: Evidence from the Taiwan Stock Exchange. Journal of Business Finance and
Accounting, forthcoming.
Choe, H., Shin, H.S., 1993. An analysis of interday and intraday return volatility – evidence from the
Korea Stock Exchange. Pacific Basin Finance Journal 1, 175–188.
Economides, N., 1995. How to enhance market liquidity. In: Schwartz, R. ŽEd.., Global Equities
Markets. Irwin Professional, New York.
Economides, N., Schwartz, R.A., 1995. Electronic call market trading. Journal of Portfolio Manage-
ment 21 Ž3., 10–18.
Frino, A., West, A., 1997. The SEATS opening procedure: among the best in the world, but can it be
improved? ASX Perspective 3rd Quarter, 61–64.
Frino, A., Lau, S.T., McInish, T., 1996. What is the best way of opening an electronic market? ASX
Perspective 1st Quarter, 48–51.
Madhavan, A., 1992. Trading mechanisms in securities markets. Journal of Finance 47 Ž2., 607–642.
Rhee, G.S., Wang, C.J., 1998. Quoted and effective spreads on the Taiwan stocks. Working paper,
Asian Development Bank and University of Wollongong.
Roll, R., 1984. A simple implicit measure of the bidrask spread in an efficient market. Journal of
Finance 39, 1127–1139.
Schnitzlein, C., 1996. Call and continuous trading mechanisms under asymmetric information: an
experimental investigation. Journal of Finance 51 Ž2., 613–636.
Stoll, H.R., Whaley, R.E., 1990. Stock market structure and volatility. Review of Financial Studies 3,
37–71.