You are on page 1of 30

Research Topic:

“SWICHING TO CONTINUOUS TRADING:


EMPIRICAL EVIDENCE ON CHI MINH STOCK EXCHANGE.”

FIRST DRAFT

COURSE BA 615: INDEPENDENT STUDY


STUDENT: CAO THANH HOANG
ID: 7492100545
CLASS: MBA IP 2006 ENTRANT

Advisor: Professor Dr. Andrew Criswell


Graduate School
Bangkok University
Abstract:
“This paper investigates stocks’ behaviors on HoSE when trading
mechanism switch from call auction to continuous trading. Event study
method is used to test market response to the announcement day and
implementation day. Market volatility and market liquidity are also
examined. The market volatility is measured by rolling standard
deviation and ARCH/GARCH method. Liquidity is tested by 10-day
rolling ILLIQ. The result shows that daily stock returns decrease around
2 percent after market enter into the new trading method. Volatility and
liquidity do not improve after market switch from call auction to
continuous trading.”

2
1. Introduction
The securities market is an important financial channel in attracting long-term investment
capital to economic regions, industries and enterprises, providing the chance for circulating
investment capital in the economy. The securities market appeals to different types of investors,
from financial institutions to individuals. Around the world, market officers always try to
improve the trading mechanism in order to satisfy the market participants. The London Stock
Exchange was implemented the system for the FTSE-100 stocks aiming to increase the liquidity
and reduce transaction cost. In the late 1980s, Paris stock exchange replaced the floor trading by
computerized limit order book. Toronto also adapted the computerized trading system in
1997(Stoll, 2003).
Vietnam stock exchange is one of the latest exchanges in The South East Asian region
compared with Singapore, Thailand, Malaysia or Philippine. Ho Chi Minh Stock Exchange
(HoSE) was established in 2000 and stocks were traded through the trading floor system at HoSE
trading center. HoSE gradually computerized the trading system during 2002 to 2005 when there
were more companies listed increased investors participation. From 2000 to 2006, HoSE had
only two trading sessions per day. The third trading session started from June 2006. Since July
2007, HoSE switched from three call auction sessions to a combination of one continuous
auction session and two call auction sessions. The HoSE authorities intend to improve market
liquidity and more orders in the market should be executed.
Improvement of the trading mechanism was the critical step of HoSE. This event raised
the question for the financial economists and practitioners about whether improving market the
structure would have any effects on stocks’ values. This kind of thinking is related to the market
microstructure topic which has been emerging during the past few years. Therefore, this paper
will examine the impact of switching to continuous trading on stocks behaviors in HoSE. This
paper will take the event study approach to test market response to new trading method. Market
volatility and liquidity will also be examined during the time before and after switching.
Section two will review the literatures of market microstructure regarding to trading
mechanism and switching effects. Section three introduce about Ho Chi Minh Stock Exchange
and its trading mechanism. The methodology and hypotheses will be developed in section four.
Section five will present the data collection empirical result analysis. Finally, discussion and
conclusion will summarize the paper.

3
2. Literature review:
2.1 Market design
Designing a securities market can be considered as framing the rules of a game for
market participants. The question that arises among authorities is how to design the market
efficiently, which can minimize any externality factors. The market can be run under a call
auction or continuous trading method. It depends on bundles of features involved such as
technology infrastructure, investors’ behaviors, and so on. A successful market must satisfy all
traders by minimizing processing costs and any related risks. Schreiber and Schwarts (1986)
argued that a securities market should be designed with regard to two objectives: reduction of the
costs of transaction for the participants in a trade, and enhancement of price discovery for the
broad market.
Most of securities exchanges start with call auction method. However, today most
markets are continuous trading. Many of them apply call auction for setting opening and closing
prices. In call auction, all selling and buying orders are aggregated in order to set the equilibrium
point. Statistically, in call auction trading, the orders are considered as batch arrival in contrast
with single arrival in continuous trading. The issue of call auction method was firstly founded by
Kyle, in 1985. Kyle modeled the trade between informed and uninformed traders. The clearing
price in call auction method will be affected by informed traders as they know the order flow
from uninformed traders and intervene prior the execution time. The problem with some
exchange market when they apply call auction method together with continuous trading is that
some investors wait until the last minutes. In order to eliminate this situation The Euronext and
some others exchanges use “random stop”.
In addition, Vayanos (1999) figured out that informed traders can take advantages of call
auction method by splitting their orders and concentrate on one specific transaction. This activity
will reduce the impact of market risk on their trading and improve the benefit. However, with
small interval of time in continuous trading, the informed traders must split orders into many
trades that will be not efficient in controlling market information. The result is similar to
Copeland and Galai, 1983, but in different approaching method. Copeland and Galai stated that if
the interval times of order arrival in continuous trading are greater than interval of information
arrival the informed traders will gain less benefit than what they can gain in call auction with
smaller interval of order arrival (Biais et al, 2004).

4
Another issue in call auction trading method is some bias in price determination. The
situation can be happened when there is a same direction of many orders entering the system
simultaneously. The problem usually occurs at the opening time when the stock market “sleeps”
for several hours. Amihud and Mendelson,1987, examined the difference between open-to-open
and close-to-close daily returns of stocks on NYSE. They found that the opening returns are
more volatile and auto-correlated pattern than closing returns. The explanation is that opening
returns may be impacted by the call auction method at the beginning of transaction day and
“sleeping time” of market before it opens. Mendelson and Murgia (1990) further investigated the
real reason for the volatility of opening return. They changed the call auction method to another
point of time instead of opening time and found that the impact of call auction trading is not
significantly. The conclusion is that closing market overnight effects on the volatility of the
opening returns.

One of the biggest differences between call auction and continuous trading is the market
immediacy. Traders, in securities market, likely want to execute their orders as soon as possible.
This involves in market liquidity factors. Clearly, continuous trading can meet this requirement
of the traders. However, this assumption firstly is questioned by Economides and Schwartz,
1995, with several respects. One of them is that some participants do not choose to pay the prices
for market immediacy when they have alternatives. These traders concern about the cost of
transaction more than immediacy. The cost of immediacy can be categorized as bid-ask spread
market impact, a higher commission and also less accurate price discovery (Economides and
Schwartz, 1995). This argument is supported by some following papers. Bronfeld, 2001,
investigated the adoption of firm’s shares in Tel-Aviv stock exchange when continuous trading
launched. Bronfeld found that the small stocks with less liquidity requirement seem to be traded
more efficiently in call auction than continuous trading. Recently, Foster, Gervais, and
Ramaswamy (2007) tested the effect of trading mechanism between volume-conditional order
crossing and continuous-auction market. The result showed that the volume-conditional trading
market is more attractive to traders who do not require much for immediacy in their trades.

2.2 Switching trading mechanism effect

Improvement trading mechanism is the trend of many stock exchanges around the world.
Some financial markets gradually applied new trading methods replacing the current ones. Some
others simultaneously operate bilateral trading systems that allow firms the alternatives to
5
choose. However, switching from one trading system to another may effect a change on the
stocks’ features. Mendelsen et al (1996) used the event study to investigate stocks’ behaviors in
Tel-Aviv Stock Exchange, Israel when the commissioner announced to start the continuous
trading method. They observed stocks during 5 days prior to the announcement and 30 days after
trading with new method. The results showed that market microstructure is valuable. When
changing to the new trading method, stocks values are positively increased together with market
liquidity. The abnormal return of individual stock skyrocketed on the day of switching to new
trading method. The market efficiency also improved when stocks prices are more fluctuant with
information related to the firms. Mendelsen’s et al (1996) research inspired other following
papers studied on stocks switching in the Tel-Aviv Stock Exchange.

Lauterbach and Ungar , 1997, also studied the impact on returns and trading volume of
stocks in Tel-Aviv Stock Exchange when they switch from daily auction to continuous trading.
They found that, upon entering the continuous trading, stocks prices volatilities decreased and
trading volume increased. More interestingly, their analysis result showed the variance of
unmatched orders reduced immediately when stocks enter the continuous trading. Furthermore,
Lauterbach, 1999, examined 97 stocks moving back from continuous trading to single daily
auction. The result is totally opposite with stocks’ status when they enter the continuous system.
Upon existing continuous date, the stock liquidity, price accuracy and value drop. More
importantly, Lauterbach uncovered that the continuous trading system is not efficient for all kind
of stocks. There are some stocks of small firms react positively when they move from continuous
to call auction trading.

The effect of a trading mechanism on stock prices is similar to Abad and Pascual (2005)
studying on Spanish Stock Exchange (SSE). Based on the model of Madhavan (1992), Abad and
Pascual (2005) tested the properties of switching from continuous trading to call auction aiming
to cool down the SSE market. They found that the solution is only appropriate for small-cap
stocks trading. Unfortunately, the high volatility, and illiquidity level persist after rule-based
auction. Abad and Pascual (2005) also argued that the information asymmetric risk is likely
higher in auction market because informed traders can anticipate the order flows in the market.
After switching to auction, the adverse selection cost is higher meaning that pre-auction cost of
risk from informed trader is higher in this case.

6
Recently, Suzuki and Yasuda (2006) investigated the difference between an order-driven
system (AC) and a market-making system (MM) in the JASDAQ securities exchange. They
found that stocks’ prices steadily appreciate when firms switch from AC to MM system. They
also pointed out the liquidity of stocks trading in market-making system significantly increase
and the transaction cost decrease as consequence. Suzuki and Yasuda (2005) further observed
few stocks switched back to AC from MM system. The result is liquidity did not improve
anymore. Suzuki and Yasuda concluded that the market-making system is more advantageous
than the order-driven system in terms of liquidity and transaction cost.
Financial research papers, regarding the Vietnam Stock Exchange, are very scarce. There
are only a few topics had been studied such as daily return volatility (Thanh Long, 2005) or
Policies impacts on Vietnam Stock Exchange (Faber, Van Nam, Hoang Vuong, 2006).Due to the
increasing demand from organizations for listing on stock exchange and booming traded
volumes in the market, authorities have to upgrade the current trading infrastructure. This paper
examines the stocks features when trading mechanism changes from three call-auction sessions
to a combination of two call-auction sessions and continuous auction session. The research is
trying to figure out any impacts of market structure on stock returns and volatility in prices as
well as trading volume. It is not same as previous works because two different trading methods
are implemented on the same day and in same exchange. The data collection will focus only on
firms’ stocks, not including fund certificates or bonds.

3. Vietnam Stock Market in overview


Ho Chi Minh Stock Exchange (HOSE) was established in 2000 with only 2 companies
listed for trading. Until March, 2008, there are 483 listed stocks on HOSE, of which 119 stocks
totaled a value of VND 300,934 billion; 2 funds certificates had volume of 150 million fund
units and 361 types of bond. Trading partners on the market include individual and institutional
investors, locally and oversea. There are 273,936 individual trading accounts, of which 265,540
accounts are Vietnamese individual investors.

3.1Trading process
All trading stocks activities are completely executed by the central computer system at
Ho Chi Minh Stocks Trading Center. Investors have to open trading accounts at Securities
Companies and place orders through these accounts. The orders will be transferred immediately
to the desk of Security Companies in Stocks Trading Center for matching. After that the traders
7
will be announced results of the transaction. At the beginning, stocks transactions were executed
only two sessions and switched to three sessions, from June 2006. But all these sessions applied
call auction methods that the orders would be matched in three times per day. Since July 30 th
2007, HOSE officially introduce the continuous auction method together with call auction
method. The transactions will be executed within three sessions including call auction from 8.30
– 9.00 am, continuous auction session from 9.00 to 10.00am and back to call auction from 10.00
to 10.30am.
Call auction is a trading method in which stocks are priced depending on the transaction
with highest volume. All sell and buy orders will be gathered and executed in a specific period of
time. In other words, the Trading Center will collect all supply and demand for individual stock
and match price at the equilibrium point. This price will be the benchmark for all transactions.
The advantage of call auction is that it can help inexperienced or new traders easy to participate
in the stock market. It is also suitable for some exchanges with low technology infrastructure and
lacking of talent people. However, call auction contains many underlying problems. Investors
tend to place orders at the time close to matching time that cause the selling orders and buying
orders increase suddenly. That is also the chance for some traders make the fake supply or
demand. Another issue for call auction method is that investors must wait for long time until
receiving transaction results. They may also buy or sell at unexpected prices.
Continuous auction is a trading method in which the transaction is continuously executed when
there are any sell and buy orders corresponding in the system. Under continuous auction, orders,
after inputting the system, will be matched and priced immediately. The advantages of
continuous auction can cover all weak points of call auction. But it does require running on the
high technology system and the investors also need to be careful when making decisions.
Switching from call auction to continuous auction, stocks trading center authorities aim
to increase numbers of successful transactions. That will not only help traders to satisfy with
their decisions but also create chances for Securities Companies to increase the revenue.
Moreover, during call auction session, some traders want to place the buy orders with ceiling
prices or sell orders at floor pries even they do know that these orders will not be matched. This
circumstance will not be restricted when continuous auction applied because such orders may be
matched by the prices priority.
3.2 Trading Orders
Presently, there are two types trading orders and negotiated transaction in Ho Chi Minh
Stock Exchange. Negotiated transactions are usually used by big organizations or institutions.
8
Limited orders can be placed during continuous or call auction. The At-The-Open (ATO) and At-
The-Close (ATC) orders are only applied during call auction sessions. ATO orders will determine
the opening prices whereas ATC orders determine the closing prices.
ATO/ATC is an order that the traders did not specify any selling or buying prices. All
orders including limited and AT are submitted to Trading Center where they will be matched for
opening price or closing price of the day. The Trading Center will match orders with prices based
on supply and demand of individual stock. Sometimes, referenced prices, the closing prices of
previous day, are used for matching orders. Some orders unmatched during the call auction will
be transferred to the next trading session.
The biggest advantage of ATO/ATC orders is harmonization in the stock market. Stock
prices are determined at the equilibrium point between supply and demand. If the traders due to
aiming to get orders matched may place the selling orders with low prices or buying orders at
high prices that can distort stock values. ATO/ATC orders will help them trade at good prices. In
other word, the trader can sell at high price or buy at low price depending on market trend.
However, ATC/ATO also has some disadvantages. As discussed before during call auction, stock
prices can be lead by some people and untrue demand and supply can happen.
Limit Order is an order that specified the limit for selling or buying prices. Under limited
orders, the traders require their agents to execute transactions with prices stated or better. For
example, “Buy 1,000 shares at $10.00 per share” or “Sell 1,500 shares at $12.00”. The traders
accepted to buy 1,000 shares at $10.00 or lower and sell 1,500 shares at $12.00 or higher price.
Limited orders are applied during continuous auction, and they jump into trading system
randomly. The new coming limited order will be compared with all orders already available in
the system for assuring a match. For example, a buying order comes to the system stated that
“Buy 1,500 shares at $10.00 per share”. And there are some selling orders available in the system
such as “Sell 500 shares at $9.00 and another 500 shares at $10.00”. The buying order will be
matched at $9.50 per share on average. The system will automatically cancel the surplus volume
unmatched. If there is a difference between limit of selling and buying prices, the order will be
unexecuted. The limited orders can be adjusted or cancelled by the traders at any time that make
the books are very active. In addition, the HOSE authorities require all information about orders
must be shown in the screen for every market participant.
Limited orders markets allow the buyers and the sellers to trade directly. Brokers
companies operate as the bridge for transaction and the traders must have their own strategies
based on information holding. According to many experts, limited orders traded in the
9
continuous auction help the investors have more opportunities to execute the orders. But they
also impose more risk on stock trading. It requires that investors must be more professional and
experienced.
There are some rules governing the sequence of orders executed. Best price is the first
priority. It means that the trading system will execute orders from highest buying price and
lowest selling price to the second runner. Time is the secondary priority. If there are two orders
with the same prices, the sooner entering order will be executed before the other. In other word,
the orders will be matched following the First in First out (FIFO) rule. Ho Chi Minh Stock
Exchange also applies some limits for the volume of order and range of prices. The volume of
each order is limited at 9,900 shares. Daily range of prices currently is 5%. The highest and
lowest price will be calculated based on reference price and range of prices. The highest price is
equal 105% of referenced price whereas the lowest is equal 95% of referenced price.

4. Methodology and hypotheses


4.1Event Study Model introduction
There are some papers in financial research defining the meaning of event study. Mc
William and Siegel (1997) defined that even study determines whether there is an “abnormal”
stock prices associated with an unexpected event. From this point the researcher can infer the
significance of the event. Dombrow et al (2000) stated event study will examine the direction,
magnitude and speed of price reactions to various phenomenons. The event study methodology is
designed to follow certain steps. McWilliam and Siegel (1997) introduced seven steps for
implementing event study method including: (1) identify event and the relevant event period, (2)
determine expected impact, (3) identify firm impact by event, (4) eliminate or adjust sample,
(5)compute Cumulative Abnormal Return (CAR), (6) analyze results, and (7)report data in
appendix. Event study is popularly used because of its advantages. The methodology is powerful
and easy to design. It is also help the researcher to explain the abnormal stocks performance with
less perfect assumptions. However, it holds some drawbacks too. The general issue is event study
must assume that the event is independent and unexpected. In addition, choosing sample and
determining abnormal return are sometime bias.
Ball and Brown (1968), Horsky and Swyndegouw (1985), Surbramani and Walden
(2001) used the event study to test stocks behavior under different circumstances. Some papers
studying on trading mechanism changes also used event study methodology. The authors
observed what happened with stocks behaviors when there is an announcement related with
10
trading method. Suzuki and Yasuda (2005) used the market model in event study to examine the
daily abnormal return in JADAQ market when companies’ stocks switch from orders-driven
system to market making system. They calculated the cumulative abnormal return (CAR) within
20 days before and 50 days after announcement. Suzuki and Yasuda’s (2005) analysis showed the
interesting impacts of announcement on spread and cumulative abnormal stocks return.
This paper will apply event study on examining how stocks return react with the announcement
of switching trading mechanism. The event is unexpected by investors and not many of them
familiar with the new method even when they got the information. We will observe market
behavior during the period of 10 days before and 10 days after the announcement released.

4.2 Measuring Daily return


Market return is the most popular feature being tested in current stock market research.
Switching to continuous trading system, the stocks are traded more frequently. In other words,
the market is more active that may bring companies’ stocks to their real values. Kyle (1985),
Madhavan (1992), and Vayanos (1999) agree that the big difference between call auction and
continuous trade is the level of risk sharing. They argue that informed traders can take advantage
of batch and periodic execution in call auction to maximize their benefit. The situation can be
reduced in continuous trade when numbers of trade significantly increase. That mean the risk on
market tend to be reduce as consequence. In the same way, this research tests the first hypothesis:

Hypothesis1: When switching from call auction to continuous trading, volatility of daily
stocks return will be reduced.

The daily returns will be calculated as the continuously compounded rate of return, in percent:
• Simple daily stock return
Pt  Pt 1
Rt  x100
Pt
where Rt is the rate of return of the stock on day t, and Pt is the closing price of the stock
on day t. Returns in the sample were corrected for splits, dividends, and other distributions.
Based on event study, this research also tests abnormal stock return during the period prior and
after the announcement of switching trading mechanism released. The new was officially
announced on the 15th July 2007 and would be applied from 30th of July. I will collect the daily

11
data within 30 days prior and before 15th July 2007 for testing the abnormal return. I will use the
formula that Subramani and Walden, 1999, used to investigate the abnormal return of stocks in
NYSE when the firms announced to apply ecommerce.

• Rit    i Rm,t  i

where Rit is the simple return of stock i at time t. The subscript t indicates time, the subscript i

indicates a specific stock, and the subscript m indicates the market. The i is a random error term

for stock s at time t, α and i are coefficients to be estimated. Rm ,t is the market return. In this
study, the VNindex is used as the market index.
• Abnormal Return
ARit = Rit – (   i Rm ,t )

 Cumulative Abnormal Return (CAR)

CARi =  AR
1
it

 Mean CAR
n

 CAR
1
i

CAR = n
4.3 Measuring liquidity
Measuring market liquidity is a popular topic in micro structure research. The authors try
to figure out what factors have the greatest impact on market liquidity. On top of that, they want
to find solutions for improving the market liquidity. Consistently, switching from call auction to
continuous auction, HoSE authorities want to increase the liquidity of the stock market in
Vietnam. Therefore, this research project challenges this argument with the second hypothesis:

Hypothesis 2: Switching to continuous auction causes market liquidity improve.

There are some methods to measure the stock market liquidity. Grossman and Miller
(1987), examine market liquidity by framing the supply and demand for the immediacy. They
observed the role of market makers in liquidity based on the friction between suppliers and
demanders during a specific period of time. Following, Pastor and Stambaugh (2000) measure
the market liquidity by using regression method to test the impact of prices with large trading
12
volume. Recently, Amihud (2002) measure market illiquidity by calculating the ratio between
absolute return to trading amount during period of time. Amihud’s method is generated from
Kyle’s paper (1985) which concerns the influence of noise traders and informed traders on prices
and volume of trading. This paper will test the market liquidity by measuring the illiquidity with
time series correlation. The method is motivated by ILLIQ introducing by Amihud (2002).
Di
ILLIQi  1/ Di  Rid / VOLDivd
t 1

ILLIQi : Illiquidity index of stock i

Di :The number of days

Rid : Daily return of stock i during the period of d days

VOLDivd : Volume of stock i trading in amount

4.4 Data
This research uses the daily VNindex provided by Ho Chi Minh Stock Exchange
presenting for the entire market performance. VNindex is the official index being widely used by
all media channels in Vietnam. It is calculated by weighted average of all stocks in the market.
These data are collected from HoSE website http://www.vse.org.vn/ during period from Jan 2005
to Feb 2008.
For individual stocks, ten biggest stocks on HoSE in term of their capitalization have
been selected. These stocks are traded during the call auction to continuous trading. Data on
stock prices, trading volume were obtained from HoSE and FPT Securities Company. We define
the announce date as the day on which HoSE announce to transfer to the continuous trading.
Stock daily returns are continuously calculated by daily closing price. Trading volume is
measured as turnover in 1,000VND unit.

5. Empirical findings

5.1 Event study result


This paper conducts the event study by measuring the abnormal return and cumulative
abnormal return (CAR) of ten stocks during the switching time. Abnormal returns are calculated
based on the market model during estimation period. The estimation period is 255 days before
announcement date. This estimation period is generated by Cowan (1996) and widely used. The
event window is 10 days before announcement (A-10) and 10 days after announcement (A+9).

13
Figure 1 show the data of abnormal return and cumulative return before and after the
announcement. The CAR fluctuates from day A-10 to A. CAR decrease sharply in day A-8 to A-
6 and increase back until A-4. More importantly, nothing happens between day A-1 and A. The
CAR is -1.105% at day A-1 and -1.178% on day A. The difference is even smaller than other
normal days. Continuously, the stock’s return drops on day A+2 and A+3. It slightly goes up on
day A+4 and A+5. However the negative daily return on average is not significantly different
with other period of time.
This result is totally different with many previous findings. Average excess returns on
event day have been found by several researches across some exchanges in the world. Most of
them come up with the positive outcome in excess average return: 5% (Amihud, Mendelson and
Lauterbach, 1997) or 10% (Lauterbach and Henke, 2003).

Figure 1:
This paper further observes the stock behaviors by moving the spot-time from the
CAR average

0
0

-8

-2

T
A

+4

+6

8
10

12
-2
-1

A-

A-

A+

A+

T+

T+

T+

T+
A

T+

T+
A

A
A

-4

-6

-8

-10

-12

-14

announcement date (A) to the application date (T). The event window now is (A+5; T+5). The
reason is that there is a doubt on the immediacy of investors’ response. They were not familiar
with the new method of continuous trading. Therefore, they may react only when the new trading
method was applied. Table 1 shows that daily average stock return dropped from -0.76% on day
T to around -1.96% on day T+1. The stocks return continuously drop more deeply in day T+2,
T+3. They drop on average around -2.00% per day. However the return grows positively on day
T+4 and T+5. CAR also declines sharply from day T to day T+ 4. But it increase back from day

14
T+5 and on. The declination and improvement back of stocks’ performance after witching to new
trading method reflects the gradual adoption of the market.
Figure 2 shows the significant difference of stocks’ return between two regimes in the
window. Stocks returns tend to slightly improve when they are traded in new method. Mean of
daily average return in call auction is -0.1969% where as mean in continuous trading is
-0.1296%. On the other hand, the daily returns are more volatile. Standard deviation of daily
stocks return in 20 days before switching is 1.15% that is less than 1.94% in 13 days after
switching.
Figure 2

Regime N Mean Std. Deviation Std. Error Mean


Abrtn_avrg Call auction 20 -.1969 1.15943 .25926
Continuous 13 -.1296 1.94323 .53895
CAR_avrg Call auction 20 -2.6931 1.42014 .31755
Continuous 13 -7.4717 2.31543 .64218
Abrtn_avrg denotes for the abnormal returns of ten stocks on average
CAR_avrg present for cumulative abnormal return on average.

More interestingly, not all stocks response to the event of applying new trading method.
The result shows that most of stocks drop on the day of switching to continuous trading. Many of
them still keep dropping and grow back in few days. However, there are few stocks that do not
react with this event. One stock decreases on day T and reserve to increase on next days. Another
stock does not response with the event at all. They are big companies and listed in the stock
exchange for long time, ITA and VNM.

5.2 Event study test


The reader may concern about any other factors effecting to the abnormal stock return
during the event window. There are several methods to test the validity of event study result.
Patell (1976), Brow and Warner (1980,1985), Musumeci and Poulsen (1991) introduce range of
parametric tests including standardized residual test or standardized cross-sectional test.
Simultaneously, some non-parametric methods are also used to test the event study. Sanger and
Mc Conell (1986), Cowen and Sergeant (1996) apply the generalized sign test or Corrodo (1989)
introduce the rank test. This paper applies the generalized sign test generated by Cowen (1996).
The concept of generalized sign test is the same as traditional sign test. This method
assumes that number of stocks with positive return on event date follow the binomial distribution

15
with parameter p. In this paper, I reverse the concept by assume the number of stock with
negative returns follow the binomial distribution. The null hypothesis for traditional sign test is p
= 0.5 or 50%. The generalized sign test in this paper use the parameter p to test the null
hypothesis by calculating as follows:
N
1 1
P=
N
 Tj   jt
j 1

P: the fraction of stocks with negative abnormal return expected during estimation time.
N: the number of stocks sample
Tj: the time of estimation period: 255 days
φjt = 1 if Ajt < 0 ( Abnormal return of stock j on day t) , = 0 otherwise.
After finding p value, I use this value to test the null hypothesis by binomial test. The
data for testing is all abnormal cross stocks during the event window. The null hypothesis
assumes that negative abnormal return during the even window have to binomial distribute with
probability at p. Otherwise we can reject the null hypothesis.
Figure 3

Binomial Test

Observed Asymp. Sig.


Category N Prop. Test Prop. (1-tailed)
VAR00001 Group 1 <= 0 70 0.636 0.531000 0.017
Group 2 > 0 40 0.364
Total 110 1.0

Figure 3 shows the result of binominal test. The fraction of negative abnormal return
during even window is 0.636. During the estimation period, the probability of negative daily
return is 0.531. The result shows that the distribution of negative during the event window is
significant. The p value is 0.017 that lower than 5% as a norm. Therefore we can reject the null
hypothesis and conclude that event has an impact on negative abnormal stock returns.

16
5.3 Market Volatility
Market volatility is one of the most important characteristics in any stock exchanges. The
methods for testing market volatility have been evolved for many years. It can be defined
volatility as the standard deviation of stock returns during a period of time. I firstly examine the
market volatility between two regimes by measuring the standard deviation of daily stock returns
from Jan 2005 to Feb 2008. Further, I measure the 5-day rolling standard deviations during two
regimes. Finally, the ARCH/GARCH method will be used to compare variance of daily market
returns before and after switching from call auction to continuous trading.
Table 2: Daily return volatility between two regimes of 10 sample stocks.
AGF FPT GMD ITA KDC
Call Continuous Call Continuous Call Continuous Call Continuous Call Continuous
Mean 0.194 -0.435 0.015 -0.430 0.188 -0.198 0.561 -0.194 0.400 -0.305
Median 0.000 -0.312 -0.221 -0.290 0.000 -0.238 0.218 -0.222 0.360 -0.260
Maximum 11.022 5.008 9.227 6.431 9.763 8.858 10.974 5.882 9.560 5.620
-16.31
Minimum 4 -5.872 -7.094 -6.652 -9.091 -7.052 -15.263 -9.160 -15.000 -7.490
Std. Dev. 2.463 1.928 3.401 2.472 2.660 2.385 3.238 2.187 3.145 2.535
Skewness -0.543 -0.081 0.208 0.327 -4.145 0.326 -0.767 -0.452 -0.711 -0.062
Kurtosis 9.258 3.537 2.471 3.829 64.802 4.545 6.578 6.435 5.162 2.960
Probability 0.000 0.386 0.238 0.035 0.000 0.000 0.000 0.000 0.000 0.950
Observations 638 145 152 145 638 145 170 145 402 145

PPC REE SAM STB VNM


Call Continuous Call Continuous Call Continuous Call Continuous Call Continuous
Mean -0.496 -0.145 0.353 -0.306 0.252 -0.564 0.016 -0.114 0.325 -0.288
Median -0.590 -0.170 0.000 -0.350 0.000 -0.460 0.000 -0.300 0.125 -0.340
Maximum 5.110 5.310 7.610 6.170 11.260 5.500 7.730 6.800 8.740 4.470
Minimum -6.580 -7.430 -8.180 -8.600 -21.320 -20.110 -10.370 -5.980 -7.030 -6.690
Std. Dev. 2.428 2.276 2.136 2.063 2.559 2.898 2.669 2.383 2.591 2.110
Skewness 0.232 -0.176 0.288 -0.284 -1.085 -2.117 -0.304 0.505 0.125 -0.292
Kurtosis 3.166 3.471 4.015 5.218 13.505 16.070 0.719 0.516 3.377 3.548
Probability 0.543 0.352 0.000 0.000 0.000 0.000 0.000 0.040 0.201 0.144
Observations 121 145 638 145 638 145 259 145 376 145

Table 2 shows that changes in market trading mechanism effect on individual stocks very
differently. Some stocks show significantly to reduce returns’ volatilities whereas some others
are not clearly. The kurtosis figures are various too. There are some stocks that their kurtosis
reduces from value greater than 3 to lower level. This means that the bell shape, normal
distribution, is likely thinner. In contrast, few stocks show the result in reversal. Very few stocks
show the skewness values changing from negative to positive or via versa. Most of them remain

17
the negative skewness value during two regimes, which mean the left tails are particular extreme
across all the stocks.
I further examine the volatility by measuring 5-day rolling standard deviation of daily
stock returns. Figure 3 and 5 show the results in consistence with previous descriptive analysis.
The effects on individual stocks are very different. Some stocks reduce the fluctuation in daily
return when they enter the continuous trading, whereas some are not. Volatility of the entire
market returns are not improved at all.
Figure 3: 5-Day Standard Deviation daily market return

5-day STDVE Market Return

6
5
4
3
2
1
0
05

06

07
5
5

8
/0
/0

/0

/0

/0

/0

/0

/0

/0

/0

/0

/0

/0

/0

/0

/0
0/

0/

0/
10

10

10

10

10

10

10

10

10
10

10

10

10

10

10

10
/1

/1

/1
1/

3/

5/

7/

9/

1/

3/

5/

7/

9/

1/

3/

5/

7/

9/

1/
11

11

11
Figure 4: Daily market returns comparison between two regimes

Daily market return


Call Continuous
Mean 0.232 -0.226
Median 0.000 -0.210
Maximum 6.940 4.600
Minimum -4.880 -4.630
Std. Dev. 1.658 1.743
Skewness 0.161 0.105
Kurtosis 4.192 3.368
Probability 0.000 0.581
Sum 148.190 -32.710
Observations 638 145

Figure 4 provides the closer view for daily market return between two regimes. Daily
market returns tend to be negative during the continuous trading with the mean at -0.2250
compare with 0.2322 during the call auction. The standard deviation in continuous period shows
slightly higher than that in the call auction time meaning market returns tend to be more volatile
when stocks enter in continuous trading. Both the kurtosis figures are greater than 3 meaning the

18
bell shapes of normal distribution are flat. However, the range of daily return during call auction
method from -4.88% to 6.94% is slightly higher than that in the continuous trading.

ARCH/GARCH model testing


ARCH (Autoregressive Conditional Heteroskedasticity) model can be considered as a
breakthrough in measuring and forecasting the volatility. Introduced by Engle (1982), ARCH
model deals with the persistence of variance in time series data. Bollerslev and Taylor (1986)
develop the GRACH (Generalized ARCH) from traditional ARCH. These models are widely
used in many financial research papers (Engle, 2001). GRACH model allows explaining the
correlation of time serial in daily market returns. This paper applies the basis model GRACH
(1,1) to test the volatility of market return during call auction and continuous trading on HoSE.
GARCH formulation is similar to Depken (2001)
Rt = t 1 +  t

 t / (  t 1 ,  t  2 ...)~ (N, ht2 )

ht2 = ω+ α  t21 + β ht21

Where Rt : Daily market returns at time t


t 1 : The conditional mean

 t : Residuals that normal distribution with variance ht2


2
The variance function shows regression relationship between ht and other parameters.
ω : Constant term
 t21 : News about volatility from previous period

ht21 : Last period forecast variance

Figure 5 & 6

Daily Market Return


Switching time
8.00

6.00

4.00

2.00

0.00
1/4/05

2/4/05
3/4/05

4/4/05

5/4/05

6/4/05

7/4/05

8/4/05

9/4/05

10/4/05

11/4/05

12/4/05

1/4/06

2/4/06
3/4/06

4/4/06

5/4/06

6/4/06

7/4/06

8/4/06

9/4/06

10/4/06

11/4/06
12/4/06

1/4/07

2/4/07
3/4/07

4/4/07
5/4/07

6/4/07
7/4/07

8/4/07

9/4/07

10/4/07

11/4/07

12/4/07

1/4/08

2/4/08

-2.00

-4.00

-6.00
19
Dependent Variable: MRT
Method: ML - ARCH (Marquardt) - Normal distribution

GARCH = C(2) + C(3)*RESID(-1)^2 + C(4)*GARCH(-1)


Call auction
Coefficient Std. Error z-Statistic Prob.
C 0.069527 0.034451 2.018109 0.0436
Variance Equation
C 0.037568 0.008598 4.369504 0
RESID(-1)^2 0.273309 0.037765 7.237195 0
GARCH(-1) 0.750284 0.023078 32.51102 0
Continuous
R-squared -0.009648 Mean
Coefficient Std.dependent
Error var z-Statistic 0.232273
Prob.
Adjusted R-squared -0.014425 S.D. dependent var 1.658226
C of regression -0.1452
S.E. 0.1315
1.670143 Akaike info criterion -1.1038 0.2697
3.34526
Sum squared resid Variance Equation
1768.466 Schwarz criterion 3.373212
C
Log likelihood 0.3251 0.2287
-1063.138 Durbin-Watson stat 1.4216 0.1551
1.585316
RESID(-1)^2 0.2877 0.1302 2.2094 0.0271
GARCH(-1) 0.6196 0.1474 4.2044 0.0000

R-squared -0.0021 Mean dependent var -0.2256


Adjusted R-squared -0.0235 S.D. dependent var 1.7431
S.E. of regression 1.7634 Akaike info criterion 3.8165
Sum squared resid 438.4558 Schwarz criterion 3.8986
Log likelihood -272.6930 Durbin-Watson stat 1.7161

The result shows that the sum of ARCH and GARCH coefficients (α+β) during both regimes are
very close to one, indicating that the volatility shocks are quite persistent. The impact of ARCH
and GRACH terms are slightly different between two regimes. During the call auction the
variances of daily market return depend more on the variance of past period, GRACH term equal
0.75028 compare with 0.6196 in continuous trading. In contrast, the residual variance effect in
call auction is slightly smaller than that in continuous trading. Figure 7 provides a closer for
variance before and after switching to new trading method. The band of residual in both regimes
are the similar around (-2;+2). The maximum and minimum values of residual during call
auction are larger than in continuous trading. That means the residual range is bigger. However
there is no much difference in distribution of residual that effect to the daily market return which
sending the message of no improvement in volatility.

Figure 7:

20
Call auction trading

Continuous trading

5.4. Market liquidity

21
There are some methods being used to measure market liquidity: market-wide (Pastor and
Stambaugh, 2001), bid-ask spread (Amihud and Mendelson, 1986; Chordia, Roll and
Surahmanyam, 2000), (Joel, 2007). This paper investigates the market liquidity by calculating
stocks illiquidity. The methodology is introduced Amihud (2002) and widely applied by several
papers. The stock illiquidity is computed on daily basis and 10-day rolling from Jan 2005 to Feb
2008. Regression model will be used to test the correlation between daily illiquidity indexes
across ten stocks and time series.
Figure 8
ILLIQ and Time Series Correlation Test
Stock Pearson Correlation Sig (1-tailed)
AGF -0.423 0.000
FPT -0.136 0.010
GMD -0.300 0.000
ITA 0.043 0.226
KDC -0.532 0.000
PPC 0.714 0.000
REE -0.717 0.000
SAM -0.419 0.000
STB -0.476 0.000
VNM -0.367 0.000

Figure 8 shows the Pearson correlation between ILLIQ and time series. The ILLIQ across
eight of ten stocks have the negative correlation with time series. That means the ILLIQ indexes
reduce during daily series from 2005 to 2008. In other words, the market liquidity is improved
along that time. Gosten and Harris (1988) figure out the positive strongly relation ship between
ILLIQ and market illiquidity. Amihud, 2002, argues that illiquidity and stock return are
positively relation. Consistently, Brenman and Subramayam (1996), Brenman,Chorida and
Subramayam (1998) found the negative relation ship between securities’ liquidity and their
return (Amihud, 2002). Therefore, the illiquidity indexes have negative relationship with market
liquidity.
In order to examine the market liquidity effects between call auction and continuous
trading, I compare the ILLIQ index between two regimes. Figure 9 shows the Pearson correlation

22
between 10-day rolling ILLIQ and time series of three different period of time. ILLIQ indexes
during the call auction are significantly different with continuous trading. ILLIQ of most stock in
call auction show strong negative relationship with time series, meaning market liquidity
improves during the time of call auction. In contrast, most of stocks show the positive
relationship between ILLIQ and time series during the continuous period. Few stocks keep
improving liquidity but with low confidence level. STB, Sacom Bank, is only one stock
positively improving its liquidity when entering the new trading method.
Figure 9
Period from 2005 to
2008 Call auction Continuous
Pearson Sig Pearson Sig Pearson Sig
Stock Correlation (1-tailed) Correlation (1-tailed) Correlation (1-tailed)
AGF -0.423 0.000 -0.450 0.000 0.684 0.000
FPT -0.136 0.010 -0.250 0.001 0.633 0.000
GMD -0.300 0.000 -0.322 0.000 0.170 0.021
ITA 0.043 0.226 -0.264 0.000 -0.002 0.492
KDC -0.532 0.000 -0.457 0.000 -0.219 0.004
PPC 0.714 0.000 0.591 0.000 0.121 0.074
REE -0.717 0.000 -0.721 0.000 0.445 0.000
SAM -0.419 0.000 -0.426 0.000 0.178 0.016
STB -0.476 0.000 -0.339 0.000 -0.557 0.000
VNM -0.367 0.000 -0.487 0.000 0.581 0.000

The result also shows that the speed of market liquidity improving during call auction
regime is slightly higher than entire period from 2005 to 2008. The reason can be seen from the
illiquidity during continuous trading time. In other words, the level of market depth reduces
when stocks enter to continuous trading. Importantly, the significance level (p-values) across
stocks shows result less than 0.005 (5%). That means there is a strong correlation between 10-
day rolling ILLID index and time series. Therefore, I have to accept the null hypothesis and
conclude that the market liquidity does not improve when stocks switch from call auction to
continuous trading.

23
6. Conclusion

This paper investigates the impact of trading mechanism on Ho Chi Minh Stock
Exchange. I employ ten biggest stocks from different business industries for samples test. Event
study is used to test stocks behaviors to the announcement of switching date. The result shows
that market does not response on the news of switching trading mechanism. However, market
does react on the day of implementing continuous trading. Daily market returns decrease around
2% on average few days after stocks enter the new trading method. The market returns resume
increasing back after that. This result can explain that the immediacy of market response is very
low. In other word, the market is not efficient in term of information value.
This paper also examines the volatility and liquidity of market affected by changing in
trading mechanism. Daily data is collected from Jan 2005 to Feb 2008. The result shows that
market volatility does not improve between call auction and continuous trading regimes. Market
liquidity tends to reduce during the new trading method. This result means that the continuous
trading is not well adapted in HoSE. The result also shows that the impact of trading mechanism
is different between stocks. There are few stock react controversially with the entire market. This
paper is the very first research on market microstructure of HoSE. The future papers can examine
on other aspects such as market transparency or information risk.

REFERENCE

24
1. Amihud,.Y. (2002). Illiquidity and Stock Returns: Cross-Section and Time Series Effects.
Journal of Financial Market 5:31-56.

2. Amihud, .Y and Mendelson, .H. Asset Price and The Bid Ask Spread. Journal of Financial
Economics, 2 (1986):223-249.

3. Amihud,Y, Mendelson,.H, Lauterbach,.B. (1997). Market Micro Structure And Securities


Values: evidence from Tel Aviv Stock Exchange, Journal of Financial Economics 45.

4. Amidud.Y. and Mendelson.H (1987). Trading Mechanism and Stock return: An Empirical
Investigation. Journal of Finance 42,356-390.

5. Biais, B., Glosten, L., Spatt, C.(2004). The Micro Structure of Stock Market: A survey of
Microfoundations, Empirical Results and Policy Implication. Journal of Financial
Markets.

6. Bessembinder .H and Rath .H. Does Market Structure Matter: Trading Costs and Return
Volatility around Exchange Listings.

7. Burt Porter.R. (2003). Measuring Market Liquidity. Warrington College of Business,


University of Florida.

8. Burt Porter .R. The Multiple Dimensions of Marketwide Liquidity, Implication for Asset
Pricing.

9. Campel, Lo, Mackinlay. (1997). The Econometrics of Financial Markets. Princeton


University Press.

10. Chordia .T. (2002) Liquidity and Return: The Impact of Inclusion into The S&P 500 Index.

11. Engle Robert (2001). GARCH 101: The Use of ARCH/GARCH Model in Applied
Econometrics. Journal of Economic Perspectives, 4 (2001):157-168.

12. Ghysels.E and Pereira. J. (2007). Liquidity and Conditional Portfolio Choice: A
Nonparametric Investigation. Journal of Empirical Finance 2008.

13. Hans R.Stoll. (2003). Market Structure. Financial Markets Research Center, Working
Paper, Nr 01-16, 2003.

14. Hasbrouck Joel. (2007). Empirical Market Microstructure, Oxford University Press.

15. Jun S.G, Marathe. A, Shwaky .A.H (2002). Liquidity and Stock Returns in Emerging
Equity Markets. Emerging Market Review 4(2003) 1-24.

16. Lauterbach.B and Ungar.M. (1997). Switching To Continuous Trading And Its Impact On
Return Behavior And Volume Of Trade. Journal of Financial Service Research, 12, 30-59.

25
17. Lauterbach.B. (1999). A Note On Trading Mechanism And Securities’ Value: The Analysis
Of Rejects From Continuous Trades. Journal of Banking and Finance, 25 (2001): 419-430.

18. Lauterbach.B and Henke.H. (2003). Firmed-Initiated versus Exchange Initiated Transfer to
Continuous trading: Evidence from Warsaw Stock Exchange.

19. Ryan.B, Scapens.RW, Theobald.M. (2002). Research Method & Methodology in Finance &
Accounting. Thomson Publishing.

20. Serge Lhabitant-Francois and Gregoriou Greg N. (2008). Stock Market Liquidity. John
Wiley & Sons publishing.

21. Suyuki.K and Yusada.Y.(2005). Market Structure And Stock Prices: Firm And Their
Selection of Trading Mechanism in JASDAQ Market, Tokyo University of Science Paper
series, MS-05-02.

22. Vayanos. D (1999). Strategic Trading and Welfare in Dynamic Market, Review of
Economic Studies, 1999,66,219-254.

23. Venkataraman .K and Waisburd .A.C (2007). The Value of the Designated Market Maker.
Journal of Financial and Quantitative Analysis 3 (2007): 735-758.

24. Vuong Thanh Long. (2006). Volatility With Regime Change: The case of Vietnam Stock
Market. KOBE University.

APENDIX
Figure 10.

26
Trading Method and Order in Summary

Session1 Session 2 Session1


Trading Order ( 8.30-9.00am) ( 9.00-10.00am) ( 10.00-10.30am)

ATO X

Limit X X X

ATC X

Trading method Call Auction Continuous Auction Call Auction

Trading Mechanism

Conclusion of
contract
Securities
Securities companies
companies
Sell order Buy order

Buy order
Sell order

Customers
Customers

HOSE Stock Trading System

27
Table 2:

Abnormal Abnormal
Date Regime Return CAR Date Regime Return CAR
A-10 Call -0.13923 -0.13923 A+7 Call 0.244483 -6.17086
A-9 Call -0.19385 -0.33308 A+8 Call 0.083612 -6.08725
A-8 Call -2.46811 -2.80119 A+9 Call -0.04798 -6.13523
A-7 Call -2.5965 -5.39769 T Continuous -0.7607 -6.89593
A-6 Call 2.176481 -3.22121 T+1 Continuous -1.96355 -8.85948
A-5 Call 0.283227 -2.93798 T+2 Continuous -2.28404 -11.1435
A-4 Call -1.08642 -4.0244 T+3 Continuous -1.89431 -13.0378
A-3 Call -0.27366 -4.29806 T+4 Continuous 0.587197 -12.4506
A-2 Call 1.474141 -2.82392 T+5 Continuous 2.424332 -10.0263
A-1 Call 0.731168 -2.09275 T+6 Continuous -2.72271 -12.749
A Call -0.23177 -2.32452 T+7 Continuous 0.118081 -12.6309
A+1 Call -0.62953 -2.95405 T+8 Continuous 0.23132 -12.3996
A+2 Call -1.32731 -4.28136 T+9 Continuous 1.038943 -11.3607
A+3 Call 0.308548 -3.97281 T+10 Continuous 3.456048 -7.90461
A+4 Call -0.3024 -4.27521 T+11 Continuous 0.849374 -7.05524
A+5 Call -0.98599 -5.26121 T+12 Continuous -2.09394 -9.14918
A+6 Call -1.15414 -6.41535

28
0
2
4
6
8
10
0
2
4
6
8
10
0
2
4
6
8
10

0
2
4
6
8
10
11/21/2006 12/18/2006 1/4/2005

12/5/2006 1/3/2007 2/15/2005

0
2
4
6
8
10
12/21/2006
1/10/2005
1/17/2007 3/22/2005

1/5/2007 2/23/2005 4/26/2005


1/31/2007
1/19/2007 4/1/2005
6/2/2005
2/14/2007
2/2/2007 5/12/2005 7/7/2005
3/8/2007
2/26/2007 6/20/2005 8/11/2005
3/22/2007
3/12/2007 7/27/2005 9/16/2005
4/5/2007
3/26/2007 9/5/2005 10/21/2005

12/16/2005 2/13/2006 3/31/2006 5/22/2006 7/7/2006


4/19/2007
4/9/2007 10/12/2005 11/25/2005
5/9/2007
4/23/2007 11/18/2005 12/30/2005
5/23/2007
5/11/2007 12/27/2005 2/14/2006
6/6/2007
5/25/2007 3/21/2006
2/13/2006
6/20/2007
6/8/2007 4/25/2006
3/22/2006
6/22/2007 7/4/2007 6/1/2006
4/28/2006
7/6/2007 7/18/2007 7/6/2006
6/8/2006
7/20/2007 8/1/2007 8/10/2006
7/17/2006
5-day ST DV FPT

8/3/2007

5-day ST DVE I T A
8/15/2007

5-days ST DE V KDC
9/15/2006
8/23/2006
8/17/2007 8/29/2007 10/20/2006
5- day Rolling S DVT AFG

5-day STDVE GMD

10/2/2006
8/31/2007 9/13/2007 11/24/2006
11/8/2006
9/17/2007 1/3/2007

8/24/2006 10/12/2006 11/29/2006 1/19/2007 3/16/2007 5/9/2007


9/28/2007
12/18/2006
10/2/2007 2/7/2007
10/12/2007
1/26/2007
10/16/2007 3/22/2007
10/26/2007
3/14/2007
10/30/2007 5/2/2007
11/9/2007
4/20/2007
11/13/2007 6/6/2007
11/23/2007
6/4/2007
11/27/2007 7/11/2007
12/7/2007
7/11/2007
12/11/2007 8/15/2007
12/21/2007
8/17/2007
12/25/2007 9/20/2007
1/8/2008
1/10/2008
9/26/2007 10/26/2007
1/22/2008
1/24/2008
11/5/2007 11/30/2007
2/13/2008

6/26/2007 8/13/2007 10/2/2007 11/19/2007 1/8/2008


2/15/2008 12/12/2007 1/8/2008
2/27/2008
2/29/2008 1/22/2008 2/20/2008

29
5-day STDVE PPC

10

0
2/1/2007 3/9/2007 4/6/2007 5/10/2007 6/7/2007 7/5/2007 8/2/2007 8/30/2007 10/1/2007 10/29/2007 11/26/2007 12/24/2007 1/23/2008 2/28/2008

5-day STDVE REE

10

0
1/10/2005 4/7/2005 6/30/2005 9/21/2005 12/12/2005 3/13/2006 6/5/2006 8/24/2006 11/15/2006 2/8/2007 5/15/2007 8/3/2007 10/26/2007 1/18/2008

5-day STDVE SAM

12

10
8

6
4
2

0
1/10/2005 4/6/2005 6/28/2005 9/16/2005 12/6/2005 3/6/2006 5/26/2006 8/15/2006 11/3/2006 1/26/2007 4/25/2007 7/19/2007 10/10/2007 12/28/2007

5-day STDVE STB

5
4

3
2
1

0
7/18/2006 8/30/2006 10/13/2006 11/27/2006 1/12/2007 3/6/2007 4/18/2007 6/6/2007 7/19/2007 8/31/2007 10/17/2007 11/29/2007 1/15/2008

5-day STDVE VNM

7
6
5
4
3
2
1
0
1/25/2006 3/29/2006 5/26/2006 7/21/2006 9/18/2006 11/13/2006 1/11/2007 3/16/2007 5/17/2007 7/12/2007 9/7/2007 11/5/2007 1/2/2008

30

You might also like