Content uploaded by Faroque Ahmed
Author content
All content in this area was uploaded by Faroque Ahmed on Jun 03, 2020
Content may be subject to copyright.
European Journal of Scientific Research
ISSN 1450-216X / 1450-202X Vol. 156 No 3 May, 2020, pp.262 - 274
http://www. europeanjournalofscientificresearch.com
Analyzing Volatility of Dhaka Stock Exchange (DSE) with
Historical Events around Breakpoints:
ICSS Algorithm Approach
Faroque Ahmed
Research Associate
Bangladesh Institute of Governance and Management (BIGM)
E-mail: faroque.ahmed@bigm.edu.bd
Md. Monirul Islam
Assistant Professor
Bangladesh Institute of Governance and Management (BIGM)
E-mail: monirul.islam@bigm.edu.bd
Md. Mazharul Islam
Research Associate
Bangladesh Institute of Governance and Management (BIGM)
E-mail: mazharul.islam@bigm.edu.bd
Abstract
This paper inspects the types of incidence that lead to the hefty changes in the
volatility of Dhaka Stock Exchange (DSE). We identify when the bulky shifts in the
volatility of DSE returns takes place and then determine the local events at the time of
change in volatility. An Iterated Cumulative Sum of Squares (ICSS) algorithm is used to
detect the points of abrupt quakes in the variance of returns. DSE General Index & DSE
Broad Index daily data are combinedly collected from 1st January, 2003 to 29 July, 2019.
Our contribution is to detect each point of changes in the volatility. Then the study
sightsees around the events and policy regime over those detected periods. The findings
help the policy makers to modify and execute discussed rules and regulations in order to
avoid such collapses in future.
Keywords: Dhaka Stock market, Volatility, ICSS algorithm, DSE, GARCH, Historical
Break Points
JEL Classification: C22, G14, G18, G41
1. Introduction
Like many other developing economies, Bangladesh has an emerging stock market; but it has been
witnessing a colossal shock from its dawn. (Ahmed, Uchida, and Islam 2012) Divulge that last two
decades, Bangladesh faced two major shocks in stock market in 1996 and in 2010. The index points of
Dhaka Stock Exchange (DSE) flew from around 800 points in June 1996 to about 3,600 points in
November 1996. After that, the index started falling down radically. From 2010, the market
experienced another catastrophe, demonstrating the fall of index 3,032 points in 2011 and then 1,038
points in 2012. In this circumstance, a good number of researches have been made in modeling the
Analyzing Volatility of Dhaka Stock Exchange (DSE) with Historical Events around
Breakpoints: ICSS Algorithm Approach 263
variability dynamics of stock market returns and its diverse features in the context of Bangladesh. The
amount of uncertainty or risk related to the size of changes in a stock market index can be termed as
market volatility. High volatility is known as a common feature of emerging stock markets (Aggarwal,
Inclan, and Leal 1999). Stock market volatility refers to the variance in stock prices for a time, which is
stipulated as a risk measure for investors. It affects directly the business investment ration, financial
market performance and economic growth (Hu 1995), (Mala and Reddy 2007). Knowing stock market
volatility can be worthwhile to determine the cost of capital and evaluate the decision of wealth
distribution (Abdullah et al. 2018). As an indicator, estimation of market volatility can help policy
makers to calculate the vulnerability of financial markets(Rufus Ayo Olowe 2009);(Rufus Ayodeji
Olowe 2009). So, exact volatility model specifications for detecting variance in asset returns has been
an important issue to the decision makers in the field of economy. In addition, the capacity to model
and predict the volatility of stock returns is indispensable for investors in making decision relating to
risk management and portfolio adjustment (Abdullah et al. 2018).
Our study is an attempt intending the same to scrutinize the shocks in the volatility of Dhaka
Stock Exchange (DSE) covering the period of January 2003—July 2019. Based on the daily data, the
paper uses an iterated cumulative sum of squares (ICSS) algorithm to detect the cutoff points of abrupt
changes in the variance of DSE stock market returns. Not only this, the existing study also determines
how long the shocks last and for what (political, social and economic causes occurred locally during
the break points), which differs our study from the earlier studies in this field.
1.2. Background of the DSE
The Dhaka Stock Exchange (DSE), is the largest monetary souk in Bangladesh (another one is the
Chittagong Stock Exchange (CSE)). Its official transaction initiated in 1956 after being unified in
1954. It was known as the East Pakistan Stock Exchange Association Ltd.; the title was crammed to:
East Pakistan Stock Exchange Ltd in 1962; two years later, the name once more reformed to the
present, Dhaka Stock Exchange Ltd. The exchange is reckoned as a Public Limited Company (PLC)
and is controlled by the Bangladesh Securities and Exchange Ordinance of 1969,the Securities and
Exchange Commission (SEC) Act of 1993 (Bangladesh) and the Companies Act of 1994 (Bangladesh).
2. Extant Literature Review
The large instability of emerging markets is characterized by recurrent, unexpected changes in variance
(Aggarwal, Inclan, and Leal 1999). (Baufays and Rasson 1985) measured the discrepancies and the
points of the shocks of maximum possibility.(Bailey and Chung 1995)found that significant political
proceedings are related to abrupt shocks in instability. The times of enlarged volatility have a tendency
to be usual to returns calculated in local coins and dollar-adjusted returns. By employing an iterated
cumulative sums of squares (ICSS) algorithm, (Aggarwal, Inclan, and Leal 1999) examined the points
of sudden changes in the variance of returns in 10 of the leading developing souks in Asia and Latin
America, along with Hong Kong, Singapore, Germany, Japan, the U.K., and the U.S and how long the
changes sustain. The study found that most events are predominantly local, including the Mexican peso
disaster, phases of hyperinflation in Latin America, the Marcos-Aquino clash in the Philippines, and
the stock souk scam in India.
Aside from global experiences of stock souk instability, there exists a lot of works that
examined the volatility in the variance of Dhaka Stock Exchange (DSE)—the major stock market in
Bangladesh.(A. R. Chowdhury 1994)investigated the behavior of stock return by employing
EGARCH-M model in view of GED allocation.
(S. Chowdhury, Mollik, and Akhter 2006)
examined how
forecasted macroeconomic instability is associated with the envisaged instability of stock souk in
Bangladesh. They applied only GARCH (1, 1) model to explore all the variables’ predicted instability
in the paper. Then they used VAR to detect the association between the variables within a dynamic
264 Faroque Ahmed, Md. Monirul Islam and Md. Mazharul Islam
framework. By utilizing ARMA (p,q), GARCH(1,1)-M models, (Basher, Hassan, and Islam
2007)inspected the topic of souk effectiveness and time-varying risk-return association. They found
that Gaussian error is relevant to distribution while it is not pertinent to the returns. They also showed a
significant association between restricted volatility and the stock yields in DSE, but the risk-return
factor is explored to be both bad and good. (Mollah 2009) utilized GARCH (1, 1), GARCH (2, 1) and
GARCH (2, 2) techniques to inspect the time varying risk return association and doggedness of shocks
to instability in the stock souk of Bangladesh. He found the positive skewness and excessive kurtosis,
demonstrating the non-normality of the DSE return series and the study does not point out their
considered allocation(s).
(A. Hossain and Nasser 2011) found that SVM model demonstrates long memory possessions
in predicting pecuniary returns moreover it outstrips both the finite combination of ARMA-GARCH
and BP models in aberration performance standards. (M. S. Hossain and Uddin 2011) investigated the
effectiveness and conditional volatility of DSE utilizing three indices of price e.g. DSEG, DSI and
DSE20. They found that risk return association is positive for DSI and DSE 20. Thus they stated
positive risk-return connection, which conforms to the portfolio theory. (Rayhan, Sarker, and Sayem
2011) examined the pattern and causes responsible for the volatility of DSE’s monthly stock returns.
They determined the AIC, SBC and LR, GARCH (1, 2) as the best instability specification models for
DSE’s stock return. Using ARCH model, (Alam, Siddikee, and Masukujjaman 2013)examined the
predicted volatility of the DSE20 and DSE general index. They employed first order
EGARCH,GARCH, TARCH, and PARCH models and considered ARCH and PARCH as the best
fitted model for DSE 20 and ARCH for DSE general index return. (Roni et al. 2017) found that for the
model precision the threshold generalized autoregressive conditional heteroskedasticity (TGARCH)
model is more precise. Moreover, GARCH model is treated to be more efficient than others and has
also more forecasting ability indicated by the statistic error measurements.
Above literature review delineates that utmost of the studies either covered volatility modeling
of market indices using different techniques in case of Dhaka Stock Exchange (DSE).The existing
study scrutinizes the DSE’s stock price volatility of 18 years along with their historical causes by
utilizing an iterated cumulative sum of squares (ICSS) algorithm based on GARCH modelling, which
make it unique from previous studies.
3. Methodology
On the actions of volatility of asset revenues, a lot of diverse works exists. The GARCH model and
dissimilarities of it are cast-off to investigate the instability in many dissimilar time series. In most of
the financial series, the restricted variance is time-varying whereas the unrestricted variance is
constant. Generalized Autoregressive Conditionally Heteroscedastic (GARCH) is such a model that
handles with this problem, first presented by (Engle 1982).
3.1. Data and Variables
DSE General Index (DGEN) for the period from 1
st
January, 2003 to 29January, 2013and DSE Broad
Index from 30 January, 2013 to 29 July, 2019 a total observation of 6054 is castoff in the study. DSE
General Index (DGEN) observed consecutive falls of index points in later periods after attainment of
the ever-highest point (8,918.5135) on 5
th
December, 2010. After the market collapse of 2011 (DGEN)
has replaced by a new index called DSE Broad Index and comes in effect from 30 January, 2013 till
now. We have combined these two indexes of DSE in our study. The data of (DGEN) have been
composed from diverse journals of DSE. Historical data of DSE Broad Index has been collected from
publicly available website “https://www.investing.com/indices/dhaka-stock-exchange-broad-historical-
data”. We have converted daily observations into weekly average dataset consisting 863 observations.
The analysis is done using the “R” and MATLAB software.
Analyzing Volatility of Dhaka Stock Exchange (DSE) with Historical Events around
Breakpoints: ICSS Algorithm Approach 265
3.2. Unit Root Test
A series having time varying mean and variance has a unit root. When mean and variance of a series
are independent of time, it is called stationarity of the series. By differencing k times a non-stationary
data can be changed into stationary, which is known as integrated of order k, denoted by I (k); thus, a
series is I (0) when it is already stationary. (Dickey and Fuller 1979) Unit root test is generally used to
test the stationarity properties of a time series.
3.3. GARCH/ARMA-GARCH Model
In time series heteroscedasticity means shifting or unequal variance across the series. The problem of
increasing or decreasing volatility can easily be taken care of using the ARCH method. A
generalization of this method is GARCH or Generalized Autoregressive Conditional
Heteroskedasticity that permits the method to care about the variations in the time dependent
instability.
The basic linear “pure” GARCH (1, 1) model is
t t t
y
σ ε
=
(1)
( )
2 2 2
1 1 1 1
~ 0,1
t t t
t
y
IID
σ ω β σ α
ε
− −
= + + (2)
Where
2
t
σ
is their restricted variance.
The straightforward Linear ARMA (p, q)-GARCH(1, 1) model
1 1 1 1 1
t t t p t q t q t
y y y u u u
µ φ φ θ θ
− − − −
= + + − − + + + − − + +
(3)
(
)
1 1 1 1 1
2 2 2
1 1 1 1
t t t t t t p t q t q t
t t t
u y y y u u u
u
σ ε µ φ φ θ θ
σ ω β σ α
− − − −
− −
= = − + + − − + + + − − + +
= + +
(4)
at what time p = 1 and q = 0, it turns into AR(1)- GARCH(1, 1) process. The central fact is that the
restricted variance of u
t
is specified by
2 2
1 1
ˆ
t t t t t
E u u
σ
− −
= =
. Consequently, the restricted variance of u
t
is
the ARMA process specified by the expression
2
t
σ
in the equation 1 and 2
(
)
2 2
1 1 1 1 1
2 2 2 2
1
ˆ
t t t t
t t t t t t
u w u w w
w u u u
α β β
σ
− −
−
= + + + −
≡ − = −
(3)
white noisy error is
t
w
. The constraints
ω
,
1
β
and
1
α
must satisfy, w>0,
1
0
β
≥
and
1
0
α
≥
in order to
the positiveness of restricted variance. Additionally, if
1 1
1
β α
+ <
then
2
t
u
is covariance stationary and
the speed for which the shock to volatility decays become slower as
1 1
β α
+
approaches to one. The
feeble GARCH parameter
1
α
controls the response to souk surprises, i.e. the vol-of-vol; the
doggedness in volatility is determined by the parameter β
1
after a shock and these, composed with the
parameter
ω
control the rapidity of mean weakening and the lengthy run GARCH instability.
3.4. Diagnostic Checking
Ljung-Box test statistics is generally used to assess the performance of the probable GARCH model.
These assessments check no autocorrelation and homoscedasticity as the null hypothesis of the
projected residuals and squared uniform residuals till a precise lag respectively. To test no enduring
ARCH effects till an accurate order as the null hypothesis ARCH LM test can also be used.
266 Faroque Ahmed, Md. Monirul Islam and Md. Mazharul Islam
3.5. ICSS Algorithm
The Iterated Cumulative Sums of Squares (ICSS) algorithm was initially offered by (Inclan and Tiao
1994). This method was established to study the detection of multiple changes of variance in a
sequence of independent observations. It uses cumulative sums of squares in order to detect cutoff
points in a series, by thoroughly penetrating diverse parts of the series. In general, for an autocorrelated
series AR (1), GARCH (1, 1) is generally assessed, and the extracted residuals used in the algorithm.
(Dubois and Bacmann 2002)stated that the ICSS algorithm is perfectly specified in case of a GARCH
(1, 1) model. Let a series of independent realizations from a normal distribution with mean zero and
unrestricted variance
2
t
σ
is denoted by
{
}
t
ε
. The discrepancy inside each break is denoted by
2
, 0,1,....,
j T
j N
τ
=
where
T
N
is the whole quantity of discrepancy fluctuations in T observations, and
1 2
1 ......
NT
k k k T
< < < <
denotes the set of break points.
2 2
0 1
2
1 1 2
2
1
.....
T T
t
N N
t k
k t k
k t T
σ τ
τ
τ
= < <
= < <
= < <
For estimating the number of deviations in discrepancy and each variance shifting point of
time, a cumulative sum of squares is utilized. Let
2
1
, 1, 2,.....,
k
k t
t
C k T
ε
=
= =
is the collective amount of
the squared (mean-centered) observations from the beginning of the series to the k
th
point of time. Now
the statistic can be written as
0
, 1, ......., 0
k
k t
t
Ck
D k T with D D
C T
= − = = =
.
The D
k
statistic vacillates round zero, if there are no fluctuations in discrepancy over the trial
period. In disparity, if there are one or more abrupt discrepancy changes in the series, the value of the
statistic will change either up or down from zero. The null hypothesis of no variations is disallowed when
the extreme of the absolute value of the statistic exceeds the critical value. Assume
max | |
k k
D
is achieved
for a value of k let k*. If
( )
max / 2 | |
k k
T D
beats the previously fixed borderline, then a supposition of the
variation point will be k*.For standardization of the distribution
( )
/ 2
T
factor is essential.
4. Results
Figure 1 presents weekly averaged Dhaka stock exchange General &broad index data from 1
st
January
2003 to 29
th
July 2019.The data has a long-term upward trend with short-term fluctuations.
Figure 1: Weekly averaged Dhaka stock exchange General & broad index
Analyzing Volatility of Dhaka Stock Exchange (DSE) with Historical Events around
Breakpoints: ICSS Algorithm Approach 267
We have tested autocorrelation and stationarity of data (table 1) using Ljung–Box test and
Augmented Dickey-Fuller test (ADF) test suggesting autocorrelation and non-stationarity in data.
Table1: Ljung–Box test and Augmented Dickey-Fuller test (ADF) of DSE data
Data Ljung–Box test ADF test
Test statistic P-value Test statistic P-value
DSE data (weekly) 851.18 <0.01 -1.55 0.123
We have found that the data series become stationary at first difference (table 2). The first
differenced stationary data were used to fit AR(1)-GARCH(1,1) model since ARMA-GARCH
processes are stochastic methods, it has to be confirmed that the return sequence is stationary (Abu
Hasan and Abdul Wadud 2016) and (table 3) presents estimation of the parameters.
Table 2: ADF test of first difference of data
Data Test statistic P-value
First differenced data (weekly) -18.81 <0.01
Table 3: AR (1)-GARCH(1,1) parameter estimation
Coefficients Estimate (standard error) P-value
µ -0.3618(2.2891) 0.874
Ar1 0.4484(0.0358) <0.01
Omega (ω) 157.12 (37.12) <0.01
Alpha (α
1
) 0.1609 (0.0212) <0.01
Beta (β
1
) 0.8667 (0.0094) <0.01
The ARCH and GARCH coefficients are significant in all periods. The null hypothesis of no
ARCH effect cannot be rejected. The null hypothesis of the Ljung-Box Test, H
0
, is that our model does
not show lack of fit which cannot be rejected in this case (table 4). The ARCH-LM test statistics
generally don’t show additional ARCH effect (Null hypothesis of no ARCH effect cannot be rejected).
This demonstrates that the specification of variance equations is appropriate. (table 4).
Table 4: Standardized Residuals Tests
Test Estimate P-value
Ljung-Box Test 3.872245 Q(15) 0.9981264
LM Arch Test 3.557768 TR
2
0.9901613
The impact of one-week lagged series on current series was estimated as 0.4484 which was
found significant (p value <0.01). The long run variance of current series,
ω
was estimated 157.12
which was also significant. The ARCH coefficient
α
1
and GARCH coefficient
β
1
were found 0.1609
and 0.8667 which implies the impact of lagged squared variance and lagged squared residuals of
current value respectively. The DSE General index has high persistence in volatility with
1 1
β α
+
=
1.0276 for student-t distribution. High persistence implies that average variance will endure high since
surge in restricted variance due to shudders will decline gradually (Rachev et al. 2010). Figure 2
presents AR(1)-GARCH(1,1) residuals.
268
study period. ICSS algorithm is applied to this series and breakpoints have been identified. ICSS
algorithm assumes stationarity of series. Stationarity has been tested using ADF test of AR
GARCH
Table
Residual
algorithm. X axis showing 863
software since daily obser
breakpoints in terms of date and corresponding
Table 6:
268
The residual series has two big fluctuations and some
study period. ICSS algorithm is applied to this series and breakpoints have been identified. ICSS
algorithm assumes stationarity of series. Stationarity has been tested using ADF test of AR
GARCH
(1,
1) resi
Table
5:
ADF test of GARCH residual
Residual
Figure 3 shows breakpoints in terms of week
algorithm. X axis showing 863
Corresponding dates of breakpoints are obtained from week numbers by “R” programming
software since daily obser
breakpoints in terms of date and corresponding
Table 6:
Periods of breakpoints
Period
1
2
3
4
5
6
Figure 2:
The residual series has two big fluctuations and some
study period. ICSS algorithm is applied to this series and breakpoints have been identified. ICSS
algorithm assumes stationarity of series. Stationarity has been tested using ADF test of AR
1) resi
duals (table 5) and found stationary.
ADF test of GARCH residual
Data
Figure 3 shows breakpoints in terms of week
algorithm. X axis showing 863
Figure 3:
Corresponding dates of breakpoints are obtained from week numbers by “R” programming
software since daily obser
breakpoints in terms of date and corresponding
Periods of breakpoints
Period
1
2
3
4
5
6
Figure 2:
Residual from fitted AR (1)
The residual series has two big fluctuations and some
study period. ICSS algorithm is applied to this series and breakpoints have been identified. ICSS
algorithm assumes stationarity of series. Stationarity has been tested using ADF test of AR
duals (table 5) and found stationary.
ADF test of GARCH residual
Figure 3 shows breakpoints in terms of week
algorithm. X axis showing 863
-we
ek numbers representing 6054 daily observations of original dataset.
Figure 3:
Breakpoints of volatility of DSE stock General & broad index
Corresponding dates of breakpoints are obtained from week numbers by “R” programming
software since daily obser
vations are there in the original dataset.
breakpoints in terms of date and corresponding
Periods of breakpoints
Start date (mm/dd/yy)
1/1/2003
11/25/2003
3/24/2004
4/14/2004
8/26/2005
11/4/2009
Faroque Ahmed, Md. Monirul Islam and Md. Mazharul Islam
Residual from fitted AR (1)
The residual series has two big fluctuations and some
study period. ICSS algorithm is applied to this series and breakpoints have been identified. ICSS
algorithm assumes stationarity of series. Stationarity has been tested using ADF test of AR
duals (table 5) and found stationary.
ADF test of GARCH residual
Test statistic
-
Figure 3 shows breakpoints in terms of week
ek numbers representing 6054 daily observations of original dataset.
Breakpoints of volatility of DSE stock General & broad index
Corresponding dates of breakpoints are obtained from week numbers by “R” programming
vations are there in the original dataset.
breakpoints in terms of date and corresponding
number
Start date (mm/dd/yy)
1/1/2003
11/25/2003
3/24/2004
4/14/2004
8/26/2005
11/4/2009
Faroque Ahmed, Md. Monirul Islam and Md. Mazharul Islam
Residual from fitted AR (1)
-
GARCH (1, 1) model
The residual series has two big fluctuations and some
study period. ICSS algorithm is applied to this series and breakpoints have been identified. ICSS
algorithm assumes stationarity of series. Stationarity has been tested using ADF test of AR
duals (table 5) and found stationary.
Test statistic
-
24.4771
Figure 3 shows breakpoints in terms of week
in the volatile error series obtained by ICSS
ek numbers representing 6054 daily observations of original dataset.
Breakpoints of volatility of DSE stock General & broad index
Corresponding dates of breakpoints are obtained from week numbers by “R” programming
vations are there in the original dataset.
number
of weeks.
Start date (mm/dd/yy)
End date (mm/dd/yy)
11/
3/23/2004
4/13/2004
8/19/2005
11/03/2009
2/03/2011
Faroque Ahmed, Md. Monirul Islam and Md. Mazharul Islam
GARCH (1, 1) model
moderate fluctuations at the middle of the
study period. ICSS algorithm is applied to this series and breakpoints have been identified. ICSS
algorithm assumes stationarity of series. Stationarity has been tested using ADF test of AR
in the volatile error series obtained by ICSS
ek numbers representing 6054 daily observations of original dataset.
Breakpoints of volatility of DSE stock General & broad index
Corresponding dates of breakpoints are obtained from week numbers by “R” programming
vations are there in the original dataset.
Table
of weeks.
End date (mm/dd/yy)
11/
24/2003
3/23/2004
4/13/2004
8/19/2005
11/03/2009
2/03/2011
Faroque Ahmed, Md. Monirul Islam and Md. Mazharul Islam
GARCH (1, 1) model
moderate fluctuations at the middle of the
study period. ICSS algorithm is applied to this series and breakpoints have been identified. ICSS
algorithm assumes stationarity of series. Stationarity has been tested using ADF test of AR
P
in the volatile error series obtained by ICSS
ek numbers representing 6054 daily observations of original dataset.
Breakpoints of volatility of DSE stock General & broad index
Corresponding dates of breakpoints are obtained from week numbers by “R” programming
Table
6
represents the identified
End date (mm/dd/yy)
Number of weeks
Faroque Ahmed, Md. Monirul Islam and Md. Mazharul Islam
moderate fluctuations at the middle of the
study period. ICSS algorithm is applied to this series and breakpoints have been identified. ICSS
algorithm assumes stationarity of series. Stationarity has been tested using ADF test of AR
P
-value
<0.01
in the volatile error series obtained by ICSS
ek numbers representing 6054 daily observations of original dataset.
Breakpoints of volatility of DSE stock General & broad index
Corresponding dates of breakpoints are obtained from week numbers by “R” programming
represents the identified
Number of weeks
47
17
3
69
217
64
Faroque Ahmed, Md. Monirul Islam and Md. Mazharul Islam
moderate fluctuations at the middle of the
study period. ICSS algorithm is applied to this series and breakpoints have been identified. ICSS
algorithm assumes stationarity of series. Stationarity has been tested using ADF test of AR
(1)-
in the volatile error series obtained by ICSS
ek numbers representing 6054 daily observations of original dataset.
Corresponding dates of breakpoints are obtained from week numbers by “R” programming
represents the identified
Number of weeks
Analyzing Volatility of Dhaka Stock Exchange (DSE) with Historical Events around
Breakpoints: ICSS Algorithm Approach 269
Period Start date (mm/dd/yy) End date (mm/dd/yy) Number of weeks
7 2/4/2011 9/28/2012 85
8 9/29/2012 2/10/2017 261
9 2/11/2017 7/29/2019 90
4.1. Empirical Results with Historical Events
Within our study period from 1
st
January 2003 to November 24, 2003 there were 47 weeks of stable
volatility in the DSE price index. November 25, 2003, DSE experienced a colossal shock in the index
of its prices. There appear two significant flows in the DSE market prices at the time of 24
th
March
2004 and 14
th
April 2004 consisting 3 weeks. Another change point in the DSE’s volatility in price
index took place on August 26, 2005 according to the study result. From this time point to November
3, 2009 comprising 217 weeks there was stability in the volatility of DSE price Index. On November,
2009, the DSE witnessed a major shock in its price volatility. Our study succeeded to identify the
major shock on February, 2011 in DSE’s price Index which continues till September, 2012. The study
detected a significant change point on September 2012 when decrease in the volatility of DSE started
which stabilizes till February, 2017. From February, 2017 the study detects relatively decreased
volatility in the DSE Broad Index. That is after the major market crash in 2011 necessary steps was
taken to get back the stability of market. We see that after 2012 volatility in DSE Broad Index
gradually decreases. Apart from one notable adjustment in April, 2017 the market index sustained a
predominantly upward flight. Now according to the findings of table 6, major events along with
corresponding policy regime are discussed.
Period no. 1 (01.01.2003 to 11.24.2003)
Major Events during the Period
After the first abstract gurgle and eruption in 1996 the souk started to alleviate. The souk slowdown
sustaining for followings even years till April 2004, when DGEN rarely traversed 1000 points (Sajid
Amit 2016).
Policy Regime
After the induction in 1998 the Central Depository Bangladesh Limited (CDBL) for the stock
exchanges was traditionally installed in 2003 for the automation of the trading system (Sajid Amit
2016). It was a part of the transformation and automation ingenuity for the capital souk as suggested by
the Asian Development Bank.
Period no. 2 & 3 (11.25.2003 to 04.13.2004)
Major Events during the Period
There appear two significant flows in the DSE market prices at the time of 24
th
March 2004 and 14
th
April 2004 consisting 3 weeks. Meanwhile interest percentage cut on static interest-bearing reserves
tools subsidized a lot to the rush on the stock souks with the remaining sale of the once profitable
government certificates dropping by 52% in the first quadrant of the then up-to-date fiscal year (Byron
2004).
Policy Regime
Amendments in the Listing Regulation-1997, Council &Administration Regulation-2000 took place.
The compulsory investments bound of investment companies in government securities and government
permitted securities was amended to only 30% of their entire investable capitals by the Section 27 of
Insurance (Amendment) Act 2000. Therefore, the insurance firms are allowed to invest up to 70% of
their capitals into other cases, together with securities souk products (Wahab, Faruq, and Bank 2012).
270 Faroque Ahmed, Md. Monirul Islam and Md. Mazharul Islam
Period No. 4 & 5 (04.14.2004 to 11.03.2009)
Major Events during the Period
Listing of lucrative Govt. entities, Multinational Companies (MNCs), bigger number of mutual funds,
lively participation of commercial banks, Foreign Institutional Investors (FIIs), and Non performing
companies were delisted. DSE Index was higher among last 10years in the year end of
2007(approximately 66 percent), ended it Asia's uppermost player after China. The maximum overseas
investment in 2007 were received by the banking, the power, pharmaceutical and cement sectors
respectively (Rahman, Hossain, and Habibullah 2017).The cumulative net inflow for 12-month of
foreign funds touched US $ 148 mn in October 2007.The utmost important rushing of the DSE index
caused from November 2009 about the time of IPO of Grameen phone. In a solo day on 16
th
November, 2009, which was the first swap day of GP in DSE, the DGEN flew to top heights of point
4148; a 23% surge of 765 points. Meanwhile late December 2009 souk capitalization in the DSE
amplified from Taka 1,903 billion to Taka 2,701 billion indicating a 42% rise (Sajid Amit 2016).
Policy Regime
On 23 January 2006, BSEC send letter to Registrar of Joint Stock Companies and Farms about release
of companies for violating company law 1994.In order to hold invasion of capitals into the over-heated
souk and aid it cool down SEC bounded on the equity-to-loan ratio at 1:1.5 and controlled such types
of loaning just for stocks with P/E ratio less than or equal to 75. Afterward, SEC supplementary
squeezed the norms by dropping the loan fraction to 1:1 in stocks consisting P/E ratios not greater than
50 (Ahsan Mansur 2010).
Period No. 6 & 7 (11.04.2009 to 09.28.2012)
Major Events during the Period
DGEN steadily climb abruptly attaining 5828 on February, 2010, recording a 71% surge from
November, 2009. Powered by funds by banks, the DGEN touched 8500 by the 2
nd
week of December.
Souk capitalization was about US$ 52 billion, which was approximately 60 times the souk
capitalization of the year 2000.The quantity of listed BO accounts with CDBL enlarged 1.79 million to
1.90 million from December, 2009 to January, 2010. It reached to 2.5 million by June 2010.The
Foreign Financial Investors (FFIs) started to wrench out their investments and the 12-month collective
net removal from DSE emaciated to US$ 486 mn by April, 2010. On 31
st
December, 2011, DGEN fell
by 41 percent; market capitalization dejected by 29 percent; and total trade cost of DSE grieved loss to
the tune of 83 percent from the top on 12
th
December, 2010, when DGEN reached 8918 points.
Price/Earnings (P/E) ratio had dropped to 13.68 which emblem to as high as 29.7 in November 2010
(Sajid Amit 2016).
Policy Regime
A probe committee was formed on 24
th
January, 2011 led by Ibrahim Khaled. According to the expert’s
Government swapped the preceding Administrative Committee of the SEC. Together the Probe
Committee and the MoF arose up with an action plan consisting 36
‐points which were to be
implemented in three stages (i.e. instant, quick and intermediate‐term). MoF founded an advisory
Committee consisting four‐members for getting advisorial provision on the road to a precisely
operational capital market. SEC new authority has also organized a work plan including 29‐points.
Among these, 8 actions were recognized as ‘topmost urgent’, 3 as ‘urgent’, 14 as ‘quick span tasks’
and 4 as ‘intermediate span tasks (Moazzem and Rahman 2012). According to the work strategy, SEC
has arranged a swatted standard on the book building method. Adjustments of a number of guidelines
(1969 Ordinance and 1993 Act, joint fund and commercial governance guideline, etc.) have been
conscripted which were sent on SEC’s website for receiving civic response on the proposals.
Immediately after the breakdown, the Bangladesh Bank prolonged the timeline for commercial banks
to diminish their asset in the interior of the approved limit (less than 10 percent of bank’s total
Analyzing Volatility of Dhaka Stock Exchange (DSE) with Historical Events around
Breakpoints: ICSS Algorithm Approach 271
responsibilities); prolonged the timeline for quite a few periods to diminish commercial bank’s credit
to a solitary gathering within the acquaintance limit (Wahab, Faruq, and Bank 2012).
Period 8 (09.29.2012 to 02.10.2017)
Major Events during the Period
Investigation software was mounted to sustain transparency and blameworthiness of the market
through nearer investigation of transactions. Foreign portfolio investment moved to all-time high of
US$ 526.85 mn in June, 2014 in contrast the record low of US$ -85.18 mn was in March, 2011
(“Bangladesh Foreign Portfolio Investment 1997 - 2018 | Quarterly | USD Mn | CEIC Data” 2018).
International Organization of Securities Commissions (IOSCO) agreed BSEC request concerning full
signatory to the MoU on 20
th
December, 2013(“Commission’s Achievement of IOSCO ‘A’ Category
Membership” 2013). DSE has accomplished complete affiliation of the World Federation of
Exchanges (WFE) on 06
th
June, 2017 (“Dhaka Stock Exchange Limited (DSE) Got Full Membership of
World Federation of Exchanges (WFE)” 2017).
Policy Regime
Three innovative indices were introduced named DSE Broad Index (DSEX), DSE 30 Index (DS30)
and DSE Shariah Index (DSES) on 28
th
January, 2013. The Exchange Demutualization Act 2013 was
approved by the Parliament on 29
th
April, 2013, gazette on 2
nd
May, 2013 (Islam and Islam 2011).
Financial Reporting Act consisting necessities in formation of an independent Financial Reporting
Council has been permitted by the Assembly in September 2015 (Wahab, Faruq, and Bank 2012).
Period 9 (02.11.2017 to 07.29.2019)
Major Events during the Period
The Shenzhen Stock Exchange (SZSE) and Shanghai Stock Exchange (SSE) permitted to cooperate in
some important areas like product expansion, bazaar cultivation, technology, and in a methodical style
to accomplish higher-quality progress of the market and the economy. Dhaka Stock Exchange Limited
(DSE) and Colombo Stock Exchange Limited (CSE) have contracted a Memorandum of
Understanding (MoU) to validate collaborative efforts directed at mutual development on March 28,
2019 in Colombo (“Memorandum of Understanding between Dhaka Stock Exchange and Colombo
Stock Exchange” 2019).
Policy Regime
A tactical partnership of DSE with the Shenzhen Stock Exchange (SZSE) and the Shanghai Stock
Exchange (SSE) took place on 04 September, 2018 (“The Consortium of SZSE and SSE Has Become
the Strategic Shareholder of DSE through Settlement of the Transaction as per the SPA” 2018). The
DSE apprise with SME platform for the development of SME sector on 30 April, 2019 (“Inauguration
Ceremony of DSE SME Platform” 2019).
5. Conclusion and Recommendations
The stock market collapse in Bangladesh affected the life of a massive number of unfortunate small
stakeholders who lost almost sum of their asset. Despite this, harsh legal movements against the
condemned people were not taken. Moreover, no appropriate corrective enticements to the affected
investors were provided. This state of affair concerning the market crash constrained in recovering the
breakable sureness of the stockholders about stock market.
In this situation, the study intended to look into the situation of stock market volatility and its
crash using the weekly dataset (863) covering 6054 daily observations within 18 years. The study finds
some points of volatility changes in the variance of DSE General & Broad Index using weekly dataset.
The overall volatility scenarios show that the period 2005-2009 represents lower volatility in the
272 Faroque Ahmed, Md. Monirul Islam and Md. Mazharul Islam
variance of stock returns than the period 2009-2012 when the DSE stock market faces sudden increase
in the volatility. The study identifies significant decrease in the volatility from 2012 till 2017 in
contrast to the unstable period of year 2009-2012. Then the study sightsees around the events and
policy regime over those detected periods. The findings will help the policy makers to modify and
execute discussed rules and regulations in order to avoid such collapses in future.
There is a limitation of this study. We have identified the breakpoints of unusual changes in the
volatility of DSE price index and discuss some positive as well as negative major events around those
periods with policies taken by Govt., DSE & BSEC. Unfortunately, we did not search for any causal
relationship between those events and policies taken within those periods. Here further research can be
done in order to find out those relationships.
References
[1] Abdullah, S. M., Mohammod Akbar Kabir, Kawsar Jahan, and Salina Siddiqua. 2018. “Which
Model Performs Better While Forecasting Stock Market Volatility? Answer for Dhaka Stock
Exchange (DSE).” Theoretical Economics Letters 08 (14): 3203–22.
https://doi.org/10.4236/tel.2018.814199.
[2] Abu Hasan, Md, and Md Abdul Wadud. 2016. “Modeling Stock Market Volatility Using
Univariate GARCH Models: Evidence from Bangladesh.” International Journal of Business
and Development Studies 8 (1): 61–76.
[3] Aggarwal, Reena, Carla Inclan, and Ricardo Leal. 1999. “Volatility in Emerging Stock
Markets.” Journal of Financial and Quantitative Analysis 34 (1): 33–55.
[4] Ahmed, Sarwar Uddin, Shigeru Uchida, and Zahidul Md Islam. 2012. “Stock Market Crashes:
Reasons and Implications.” KEIEI TO KEIZAI 91 (4).
[5] Ahsan Mansur. 2010. “Is the Bull Market Sustainable?” The Daily Star, Forum, Volume 3 Issue
4, April 2010. https://archive.thedailystar.net/forum/2010/april/bull.htm.
[6] Alam, Md. Zahangir, Md. Noman Siddikee, and Md. Masukujjaman. 2013. “Forecasting
Volatility of Stock Indices with ARCH Model.” International Journal of Financial Research 4
(2). https://doi.org/10.5430/ijfr.v4n2p126.
[7] Bailey, Warren, and Y Peter Chung. 1995. “Exchange Rate Fluctuations, Political Risk, and
Stock Returns: Some Evidence from an Emerging Market.” Journal of Financial and
Quantitative Analysis 30 (4): 541–61.
[8] “Bangladesh Foreign Portfolio Investment 1997 - 2018 | Quarterly | USD Mn | CEIC Data.”
2018. 2018. https://www.ceicdata.com/en/indicator/bangladesh/foreign-portfolio-investment.
[9] Basher, Syed A, M Kabir Hassan, and Anisul M Islam. 2007. “Time-Varying Volatility and
Equity Returns in Bangladesh Stock Market.” Applied Financial Economics 17 (17): 1393–
1407.
[10] Baufays, P, and J P Rasson. 1985. “Variance Changes in Autoregressive Models. Time Series
Analysis: Theory and Practice (Ed. 0. D. Anderson) 7, 119-127.” North-Holland, New York.
[11] Byron, R. K. 2004. “Stocks Sail as Savings Instrument Sales Sink.” The Daily Star, Business,
Vol. 05, No. 197, December, 2004.
http://archive.thedailystar.net/2004/12/11/d41211050150.htm.
[12] Chowdhury, Abdur R. 1994. “Statistical Properties of Daily Returns from the Dhaka Stock
Exchange.” The Bangladesh Development Studies 22 (4): 61–76.
[13] Chowdhury, Shah, Abu Taher Mollik, and M Selim Akhter. 2006. “Does Predicted
Macroeconomic Volatility Influence Stock Market Volatility? Evidence from the Bangladesh
Capital Market.” University of South Australia.
[14] “Commission’s Achievement of IOSCO ‘A’ Category Membership.” 2013. 2013.
https://www.sec.gov.bd/latest_news/Commission’s_achievement_of_IOSCO_‘A’_category_me
mbership_20122013.pdf.
Analyzing Volatility of Dhaka Stock Exchange (DSE) with Historical Events around
Breakpoints: ICSS Algorithm Approach 273
[15] “Dhaka Stock Exchange Limited (DSE) Got Full Membership of World Federation of
Exchanges (WFE).” 2017. 2017. https://www.dsebd.org/picture_gallery/WEF membership.pdf.
[16] Dickey, David A., and Wayne A. Fuller. 1979. “Distribution of the Estimators for
Autoregressive Time Series with a Unit Root.” Journal of the American Statistical Association
74 (366a): 427–31. https://doi.org/10.1080/01621459.1979.10482531.
[17] Dubois, Michel, and Jean-Francois Bacmann. 2002. “Volatility in Emerging Stock Markets
Revisited.”
[18] Engle, Robert F. 1982. “Autoregressive Conditional Heteroscedasticity with Estimates of the
Variance of United Kingdom Inflation.” Econometrica: Journal of the Econometric Society,
987–1007.
[19] Hossain, Altaf, and Mohammed Nasser. 2011. “Comparison of the Finite Mixture of ARMA-
GARCH, Back Propagation Neural Networks and Support-Vector Machines in Forecasting
Financial Returns.” Journal of Applied Statistics 38 (3): 533–51.
[20] Hossain, Md Sharif, and Md Gazi Salah Uddin. 2011. “Efficiency Analysis and Volatility
Effect of Bangladesh Stock Market.” In Cambridge Business & Economics Conference, 1–31.
[21] Hu, Zuliu. 1995. Stock Market Volatility and Corporate Investment. International Monetary
Fund.
[22] “Inauguration Ceremony of DSE SME Platform.” 2019. 2019.
https://www.dsebd.org/pdf/Inauguration ceremony of DSE SME platform.pdf.
[23] Inclan, Carla, and George C Tiao. 1994. “Use of Cumulative Sums of Squares for Retrospective
Detection of Changes of Variance.” Journal of the American Statistical Association 89 (427):
913–23.
[24] Islam, Md Shariful, and Mohammad Rezaul Islam. 2011. “Demutualization: Pros and Cons for
Dhaka Stock Exchange (DSE).” European Journal of Business and Management 3 (12): 24–33.
[25] Mala, Rajni, and Mahendra Reddy. 2007. “Measuring Stock Market Volatility in an Emerging
Economy.” International Research Journal of Finance and Economics 8 (5): 53–71.
[26] “Memorandum of Understanding between Dhaka Stock Exchange and Colombo Stock
Exchange.” 2019. 2019. https://www.dsebd.org/pdf/Memorandum of Understanding between
DSE&CSE.pdf.
[27] Moazzem, Khondaker Golam, and Md Tariqur Rahman. 2012. “Stabilising the Capital Market
of Bangladesh: Addressing the Structural, Institutional and Operational Issues.” Centre for
Policy Dialogue (CPD).
[28] Mollah, S A. 2009. “Stock Return & Volatility in the Emerging Stock Market of Bangladesh.”
Journal of the Academy of Business & Economics 43 (2): 29–78.
[29] Olowe, Rufus Ayo. 2009. “The Impact of the Announcement of the 2005 Capital Requirement
for Insurance Companies on the Nigerian Stock Market.” The Nigerian Journal of Risk and
Insurance 6 (1): 43–69.
[30] Olowe, Rufus Ayodeji. 2009. “Stock Return, Volatility and the Global Financial Crisis in an
Emerging Market: The Nigerian Case.” International Review of Business Research Papers 5
(4): 426–47.
[31] Rachev, Svetlozar T, Borjana Racheva-Iotova, Stoyan V Stoyanov, and Frank J Fabozzi. 2010.
“Risk Management and Portfolio Optimization for Volatile Markets.” In Handbook of Portfolio
Construction, 493–508. Springer.
[32] Rahman, T, S Hossain, and Md Habibullah. 2017. “Stock Market Crash in Bangladesh: The
Moneymaking Psychology of Domestic Investors.” American Journal of Theoretical and
Applied Business 3 (3): 43–53.
[33] Rayhan, Mohammed Abu, S A Sarker, and Sheikh Mohammad Sayem. 2011. “The Volatility of
Dhaka Stock Exchange (DSE) Returns: Evidence and Implications.” ASA University Review 5
(2): 87–99.
274 Faroque Ahmed, Md. Monirul Islam and Md. Mazharul Islam
[34] Roni, Bhowmik, Chao Wu, Roy Kumar Jewel, and Shouyang Wang. 2017. “A Study on the
Volatility of the Bangladesh Stock Market—Based on GARCH Type Models.” Journal of
Systems Science and Information 5 (3): 193–215.
[35] Sajid Amit. 2016. “Bangladesh Capital Markets: An Overview.” The Financial Express, April
23, 2016. https://thefinancialexpress.com.bd/views/bangladesh-capital-markets-an-overview.
[36] “The Consortium of SZSE and SSE Has Become the Strategic Shareholder of DSE through
Settlement of the Transaction as per the SPA.” 2018. 2018.
https://www.dsebd.org/pdf/consortium of SZSE and SSE.pdf.
[37] Wahab, Md Abdul, Md Omor Faruq, and Bangladesh Bank. 2012. “A Comprehensive Study on
Capital Market Developments in Bangladesh.” Dhaka: Research Department, Bangladesh
Bank.
