The Modeling of Jakarta Composite Index Data Before and During

COVID-19 Pandemic and its Alignment into Government Policy in

Energy Sector

FLORENTINA KURNIASARI

Technology Management Department, Universitas Multimedia Nusantara,

Tangerang, INDONESIA

EKO ENDARTO

Management Department, Universitas Multimedia Nusantara,

Tangerang, INDONESIA

HELENA DEWI

Management Department, Universitas Multimedia Nusantara,

Tangerang, INDONESIA

CYNTHIA SARI DEWI

Management Department, Universitas Multimedia Nusantara,

Tangerang, INDONESIA

NURHUDA NIZAR

Fakulti Pengurusan dan Perniagaan, Universiti Teknologi MARA

Shah Alam, MALAYSIA

Abstract: The COVID-19 pandemic brings significant effects to the global stock market, including Indonesia.

This study investigates the behavior and fluctuation of Jakarta Composite Index (JKSE) before the COVID-19

pandemic arises (2018–2019) and 2 years during the COVID-19 pandemic (2020–2021) and its alignment with

the government policy in the energy sector. This study will use the JKSE data before and during the Covid-19

pandemic. The study showed that before COVID-19 pandemic, the JKSE was in normal conditions and showed

an increasing trend. However, the study found anomalies in the JKSE volatility when COVID-19 pandemic was

officially announced in Indonesia during 1st quarter 2020. This study is able to find the forecasted next 30 days

best models that can describe the pattern of JKSE data are AR (2)–GARCH (1,1) models for the closing price

of JKSE data before the COVID-19 pandemic and AR (5)–GARCH (1,1) models for the closing price of JKSE

data during the COVID-19 pandemic. With the government economic recovery program related to the energy

sector, this study was able to forecast the next 30 days for the closing price of JKSE during COVID-19, which

showed the improvement of JKSE into the small increasing trend. These findings are expected to increase

public investor trust, especially foreign investors investing their money in the JKSE. The positive trend in JKSE

will ensure the government continues its economic policy recovery plan.

Key-Words: - Jakarta Composite Index, Auto regression, GARCH model, Forecasting

Received: July 29, 2022. Revised: February 9, 2023. Accepted: March 2, 2023. Published: March 22, 2023.

1 Introduction

The GDP of Indonesia was still dominated by the

commodity and power sectors. The government

royalties from these two sectors reached IDR 121.1

trillion. Budget deficit balance as the key indicator

showed significant improvement from −2.59% in

2015 to −1.82% in 2019. The Indonesia crude price

was growing at 30% yearly, which reached at

US$70 per barrel in the end year 2019. Meanwhile,

the rise of fuel and gas subsidies that has been

triggered by the upward trend of international oil

and gas prices reached 69.605 billion Rupiah for the

gas subsidy and 59.300 billion Rupiah for the

electricity subsidy. Coal as one source of

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.64

Florentina Kurniasari, Eko Endarto,

Helena Dewi, Cynthia Sari Dewi, Nurhuda Nizar

E-ISSN: 2224-2899

694

Volume 20, 2023

commodity export significantly contributes to

Indonesia's balance of trade. Coal exports before the

pandemic occurred reached 73.57 million tons, [1].

The strong performance of exports and the stability

of the Rupiah exchange rate gave the positive

contribution of macroeconomic performance and

were shown by the increasing trend of JKSE index.

Indonesia is ranked the highest investment

destination especially in the stock and bond market

and is the most promising country among the other

32 emerging market countries for 2019 based on the

survey conducted by Bloomberg on 57 global

investors, [2].

The condition changed when COVID-19 virus

spread all over the world. The first positive COVID-

19 case in Indonesia was officially announced on

March 2, 2020, and on March 31, 2020; the

Indonesian government declared large-scale social

distancing to minimize the spread of the virus. The

pandemic is having a major impact on economic

activity in Indonesia. COVID-19 pandemic caused

the Indonesia government to widen the deficit level

of −6.34% in 2020. In 2020, the government

revenue from two main energy sectors (oil and gas)

decreased to IDR 53.3 trillion. Subsidies for

transport fuel fell 41%, while the LPG subsidy fell

56%. Fuel sales and power sector dropped due to the

travel restrictions, physical distancing policies as

well as economic slowdown. Electricity usage for

the commercial sector in Jakarta has decreased by

11.38%, the industrial sector has decreased by

15.81%, and the residential sector has increased by

4.73%, [3]. On the contrary, there is an increase in

LPG subsidy since there is increasing consumption

of food and homemade cooking, [4]. This subsidy

burdens the country’s budget since 70% of

Indonesia’s LPG supply is imported, [5]. Coal

industries also experienced the same significant

drops in volumes and prices and affected

Indonesia’s exports due to lower coal demand in

some countries and coal oversupply globally.

During the pandemic, the coal exports have

decreased by 44.35% from the same period last

year, [1].

Stock market investors respond to this changing

situation and their reactions have an impact on the

volatility of stock markets in Indonesia, [6]. The

JKSE reached its lowest level and fell by 38% from

the year 2020 due to negative sentiment from the

major global financial market and COVID-19

outbreak in Indonesia, [7]. In addition, the fear of

COVID-19 outbreak had caused capital outflow of

foreign investors from the Jakarta Stock Market.

Some previous studies have examined the impact of

the COVID-19 pandemic on financial market

performance, in which the market had a negative

reaction in response to the virus spreading, [8]

which pushed higher volatility in the stock market,

[9].

However, this condition was only temporary and

showed improvement when the Indonesia

government started to take precautionary action to

stabilize the economic condition. [10] suggested that

government policy stimulus packages would restore

overall investor confidence by boosting stock prices.

The Indonesian government launched the National

Economic Recovery Program (PEN) containing

specific programs to revive the economy. The

government started to launch the National Economic

Recovery along with the healthcare program as a

response to minimize the impact of COVID-19

pandemic. The following economic stimulus

programs in macro and micro economics were

implemented: the restructuring program or

relaxation program for the SME’s customers, giving

capital assistance for SMEs, vaccination programs

all across the country, social distancing, working

from home activities, and various programs to the

most impacted industries such as tourism and

services.

In addition, some strategy initiated was lowering the

energy subsidy allocation for fiscal year 2020 from

IDR 31.044 billion to IDR 19,900 billion, [11].

Through the Ministry of Energy and Mineral

Resources Regulation No. 7/2020 in March 2020,

the Indonesia government expanded the coal mining

industries that are eligible to receive fiscal

incentives in terms of reduced royalties. Through

fiscal stimulus package II, the government offered

four types of incentives for the 6 months, including

free import tax, 30% deduction of income tax, and

personal income tax borne by the government, and

accelerated restitution with the limit raised to IDR 5

billion, [12]. The government also issued financial

regulation to stimulate the business sector by

allowing deferment of loan installments, a lower

increase rate for specific projects, and exemption

from penalties, [13].

This research used modeling time series data in

analyzing the stock market behavior before and

during COVID-19 pandemic and can be used for

forecasting and prediction of future values. The

autoregressive conditional heteroscedasticity

(ARCH) or generalized autoregressive conditional

heteroscedasticity (GARCH) model can be used to

forecast stock volatility as suggested by [14] in his

previous study. Meanwhile, the application of this

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.64

Florentina Kurniasari, Eko Endarto,

Helena Dewi, Cynthia Sari Dewi, Nurhuda Nizar

E-ISSN: 2224-2899

695

Volume 20, 2023

model in the Indonesia stock market can be found in

the research of, [15]. The study will use the Jakarta

Composite Index (JKSE) over 2 years before the

COVID-19 pandemic rises in Indonesia (2018–

2019) and during the COVID-19 pandemic (2020–

2021). This study aims to find the best model for

JKSE data before and during COVID-19 pandemic

and its alignment with the government policy in

energy sectors that was already covered in the

economic recovery plan.

2 Statistical Modeling

In this study, the closing price of Jakarta Composite

Index (JKSE) shares from January 2018 to October

2021 will be discussed using time series modeling.

The closing price of JKSE data will be divided into

two categories, namely, data before (BPC) and

during the COVID-19 pandemic (DPC), and each

data will be modeled using time series modeling. In

the JKSE data study, both data sets will be analyzed

using time series model analysis methods, parameter

estimation, hypothesis testing, and further analysis.

Before constructing the best model for both data sets

(BPC and DPC), the assumption of stationary data

will be checked. Stationary assumption checks will

be carried out by looking at data trend patterns in

the 2018–2019 (BPC) and 2020–2021 (DPC). In

addition, it will also be examined whether the data

has an ARCH effect. There are two approaches to

examining the stationary data used: to examine the

behavior of the data plot and to examine the results

of the stationary test using the augmented Dickey–

Fuller (ADF) test with the null hypothesis that the

data is not stationary, [16], [17], [18], [19]. If the

data is not stationary, then the data is transformed

using a differencing process, so that the data

becomes stationary, [17], [20]. To test the effect of

ARCH, the Portmanteau Q test and the Lagrange

multiplier (LM) test were used. If there is an ARCH

effect on the data before the COVID-19 pandemic

(BPC) and during the COVID-19 pandemic (DPC),

then the GARCH model will be used to model the

residual BPC and DPC data. To estimate the order

of the autoregressive (AR) model, the corrected

Akaike information criterion (AICC) will be used.

The AR process provides a class of models that are

very useful in univariate time series to describe the

dynamics of individual time series. The order pth

AR, AR(p), is formulated as follows: for example,

for JKSE closing price data before the COVID-19

pandemic (BPC):

tptp2t21t10t BPC...BPCBPCBPC  

, (1)

where

BPC

is a closing price of JKSE at time t

before COVID-19 pandemic,



is a constant,

p21 ,...,, 

are the coefficient parameters for

,BPC 1t

,BPC 2t

…,

BPC 

, respectively, and



is the white noise.

Similar for data closing price JKSE during

COVID-19 pandemic (DPC), the order qth AR,

AR(q), is:

tqtq2t21t10t DPC...DPCDPCDPC  

, (2)

where

DPC

is a closing price of JKSE at time t

during COVID-19 pandemic,



is a constant,

q21 ,...,, 

are the coefficient parameters for

,DPC 1t

,DPC 2t

…,

DPC 

, respectively,

and



is the white noise.

Augmented Dickey–Fuller (ADF) Test

The ADF test is used to check the stationary data

BPC and DPC with the null hypothesis that the data

BPC and DPC are non-stationary, [16], [17], [19].

The ADF test with lag-p is formulated as follows,

[21], [16], for data BPC and DPC can be conducted

the same way.

tit

1i i1ttt BPCBPCBPC 

 







, (3)

where αt is a constant function at time t, ΔBPCt =

BPCt−BPCt−1 is the difference of a series of BPCt,

and εt is the white noise. The ADF test (or tau test)

statistic is formulated as follows:

τ-test (ADF test) =

)

(std





(4)

and the null hypothesis is rejected if the P-value is

<0.05, [16], [22].

AICC

In time series modeling, several model selection

criteria are available, such as AICC, HQC, AIC, and

BSC. The best model is selected from several

models with the smallest criterion value. Basically,

AICC is an estimate of the quality of the statistical

model. In this study, the AICC will be used in an

effort to select the best model. The calculation

process is as follows: let a linear model with p

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.64

Florentina Kurniasari, Eko Endarto,

Helena Dewi, Cynthia Sari Dewi, Nurhuda Nizar

E-ISSN: 2224-2899

696

Volume 20, 2023

coefficient of parameters and let



be the

likelihood estimator of variance. Therefore,

RSS

ˆp

p

, (5)

where RSSp =

1t t)BPCBPC( 





, for data before

COVID-19 pandemic, is the residual sum of squares

under the model with p coefficients of parameters.

The AICC is defined as follows:

2pT

lnAICC 2

p





, (6)

where T is the sample size, [16], [23].

Testing for White Noise

To check whether errors (residuals) are white noise,

Q-statistic (or Box–Pierce test) or Ljung–Box test

will be used, [23]. The Q-statistic (QBP) tests the

null hypothesis that the errors (residuals) are white

noise. The Q-statistic is calculated as follows:





BP ˆ

, (7)

where



is the estimate of autocorrelation at lag j

and T is the sample size. Under the null hypothesis,

the QBP statistic is asymptotically the Chi-square

distribution with k degrees of freedom, χ2(p).

Test for Normality Distribution

There are some methods available to check the

normality of the errors (residuals). Some methods

are commonly used to check whether the errors

(residuals) are normally distributed: (1) check the

histogram of the residuals, (2) check the Q–Q plot

of the data or error (residuals), and (3) use the

statistical test, the Jarque–Bera (JB) test, with the

null hypothesis that the data are normally

distributed, [22], [16]. The JB test is calculated as

follows:













 4

)3K(

JB 2

, (8)

where T is the sample size, S is the expected

skewness, and K is the expected excess kurtosis.

Testing for ARCH Effect

The ARCH was introduced by [23] and later

developed further by [14] who expanded the ARCH

concept into GARCH. The GARCH model was

developed based on the assumption that the variance

is not constant or heteroscedastic. Before we apply

the ARCH or GARCH model to the JKSE BPC and

JKSE DPC data, it must be tested whether the data

has an ARCH effect on the residual (error). If there

is an ARCH effect, then the ARCH or GARCH

model will be used in modeling for the JKSE BPC

data and JKSE DPC data. To test the ARCH effect,

we used the LM test with the null hypothesis that

there is no ARCH effect. The null hypothesis is

rejected if the P-value <0.05. To perform the LM

test, the residual model is built as follows:

tu...  

(9)

From model (9), R-square (R2) value can be

calculated, and then calculate the LM test. The LM

test is defined as follows:

LM = T. R2, (10)

where T is the sample size and R2 is the R-square

computed from model (9). Under the null

hypothesis, the LM test approximately has a Chi-

square distribution with p degrees of freedom, χ2(p),

[17].

AR(p)–GARCH (k, l) Model

In the AR(p) model, it is assumed that error εt is

independent and identically distributed (white

noise). However, in practice, this assumption is

sometimes violated. Let the observations for data

JKSE before the pandemic COVID-19 are: BPC1,

BPC2, …, BPCn be generated by the AR(p) model,

and then the residuals are generated by the GARCH

(k, l) process as follows:

tit

1i i0t BPCBPC 

 



(11)

and

2it

1i i

2it

1i io

t













(12)

Models (11) and (12) are called the AR(p)–

GARCH (k, l) model. Under the GARCH (k, l)

model, the conditional variance depends on the

squares error (residual) in the previous k periods and

the conditional variance in the previous l periods.

Usually, a GARCH (1,1) model is adequate to

obtain a good model fit for share price time series

data, [21].

Forecasting

Forecasting will perform after obtaining the best

model for both data JKSE before COVID-19

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.64

Florentina Kurniasari, Eko Endarto,

Helena Dewi, Cynthia Sari Dewi, Nurhuda Nizar

E-ISSN: 2224-2899

697

Volume 20, 2023

pandemic and data closing price of JKSE during

COVID-19 pandemic. By using the best model that

fits the data, forecasting is carried out directly for

the next 30 periods.

3 Result and Discussion

The data used in this study is the closing price of the

JKSE from January 2018 to October 2021, where in

January 2018 to December 2019, there were 485

data before the COVID-19 pandemic, and the price

January 2020 to October 2021 is data at the time of

the COVID-19 pandemic as many as 448 data with

an average of 20 trading days in each month.

Fig. 3: Plot data for the closing price of the JKSE

before the COVID-19 pandemic (blue line) and

during the COVID-19 pandemic (red line)

Figure 3 shows that the closing price JKSE before

the COVID-19 pandemic (2018–2019), namely, the

blue line, shows that the fluctuations are relatively

not too extreme. From January 2018 to April 2018,

the trend decreased; from April 2018 to October

2018, the trend was horizontal; from October 2018

to January 2019, the trend was increased; and from

January 2019 to December 2019, the trend was flat

and fluctuating. During the COVID-19 pandemic,

the red line indicates that the stock price has fallen

quite deep, from around 6300 at the beginning of

January 2020 to around 4000 at the end of March

2020; from April 2020 to December 2020, the

closing price JKSE trend increased, and throughout

the year 2021, the closing price of JKSE is relatively

stable, but the price is still below the JKSE price in

2019 before the COVID-19 pandemic. Figure 3

shows a significant downward price change during

the COVID-19 pandemic throughout 2020. Figure 4

also shows that the JKSE data both before and after

COVID-19 showed non-stationary data.

Since the data is not stationary, data

differencing is carried out so that the data meets the

stationary assumption, [17]. To check the stationary

of the data, we can use analytical approaches such

as the ADF (augmented Dickey–Fuller) test, the

ACF test, and also other suitable tools to test the

stationary of the data, [18].

Table 2. ADF test of JKSE closing price before and

during the COVID-19 pandemic after differencing

with lag = 1 (d = 1)

Type

Data

ADF Test

P-value

Mean

Before the COVID-19

pandemic

−16.67

<0.0001

During the COVID-19

pandemic

−19.50

<0.0001

(a)

(b)

Fig. 4: ACF plotting after differencing with d = 1

for (a) closing price of JKSE before the COVID-19

pandemic and (b) during the COVID-19 pandemic

Due to the closing price of JKSE data before

and during the COVID-19 pandemic, it was not

stationary; the next step is to transform the data so

that the data becomes stationary, namely, by

differencing the data with lag = 1 (d = 1). After the

data is differencing, the closing price of JKSE data

before and during the COVID-19 pandemic became

stationary. By using the ADF test (Table 2), it was

significantly rejected because the P-value (<0.0001)

was smaller than the value of = 0.05. Furthermore,

this result is also reinforced by the ACF (Fig. 4 (a)

and (b)) for the closing price of JKSE data before

and during the COVID-19 pandemic; it decreases

very quickly, indicating that both data were

stationary. Thus, after the data is stationary, we need

to know whether this research requires

autocorrelation modeling using autoregression (AR)

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.64

Florentina Kurniasari, Eko Endarto,

Helena Dewi, Cynthia Sari Dewi, Nurhuda Nizar

E-ISSN: 2224-2899

698

Volume 20, 2023

or moving average or a combination of the two,

namely, the ARMA model using the AICC criteria.

Model Determination by Using AICC Criteria

AICC is one of the criteria used to determine the

optimum lag in this study, [17]. Based on the results

of the AICC analysis (Table 3), the AR(P) model

with the optimum lag for the closing price of JKSE

data before the COVID-19 pandemic is AR(2) and

the best AR(p) model for the closing price of JKSE

data during the COVID-19 pandemic is AR(5)

because it has the smallest value compared to other

values.

Table 3. Determination of candidate models based

on the minimum AICC for the closing price of

JKSE data before and during the COVID-19

pandemic

Minimum information criterion based on AICC for JKSE

Closing Price before the COVID-19 Pandemic

Lag

MA 0

MA 1

MA 2

MA 3

MA 4

MA 5

AR 0

7.957

7.960

7.958

7.961

7.965

AR 1

7.957

7.964

7.962

7.965

7.970

7.969

AR 2

7.954

7.961

7.964

7.969

7.966

AR 3

7.958

7.965

7.968

7.971

7.968

7.970

AR 4

7.963

7.969

7.968

7.967

7.972

7.974

AR 5

7.967

7.969

7.967

7.970

7.975

7.963

Test for ARCH Effect

The problem arises that causes an estimation

measurement in time series data to be inefficient due

to heteroscedasticity, so we need to apply an

adequate method such as the GARCH model.

Therefore, we need to confirm whether there is a

heteroscedasticity problem in the data using the

ARCH effect test.

Table 4. ARCH effect of the closing price of JKSE

data before and during the COVID-19 pandemic

Data

Test

P-value

The closing price of JKSE

data before the COVID-19

pandemic

ARCH effect

0.0022

Normality

<0.0001

The closing price of JKSE

data during the COVID-19

pandemic

ARCH effect

<0.0001

Normality

<0.0001

ARCH effect test with the null hypothesis there is

no ARCH effect and the alternative hypothesis is

there is an ARCH effect. Judging from the

information contained in Table 4, from the test

results for the closing price of JKSE data before the

COVID-19 pandemic, it can be concluded that the

null hypothesis is rejected so that there is an ARCH

effect that is indicated by a P-value of less than

0.05, which is 0.0022, so it is necessary to model the

variance, namely, GARCH modeling for the

residue. For data on the closing price of the JKSE

during the COVID-19 pandemic, the null hypothesis

was also rejected, so it is necessary to model

GARCH for the residual, which is indicated by a P-

value (<0.0001) less than 0.05.

Model AR (p)–GARCH (p, q) Data Closing Price

of JKSE Data Before and During the COVID-19

Pandemic

The following is the best model for parameter

estimation for AR(p) and GARCH (p, q) for JKSE

closing price data before and during the COVID-19

pandemic.

Table 5. Parameter estimation model AR(2)-

GARCH (1,1) for the closing price of JKSE data

before the COVID-19 pandemic

Model parameter estimates the closing price of JKSE data before

the COVID-19 pandemic

Paramete

Estimate

Std

error

value

P-

value

Variabl

CONST1

1.09601

2.35321

0.47

0.641

AR1_1_1

0.01269

0.05122

0.25

0.804

BPCt-1

AR2_1_1

−0.10055

0.04448

−2.26

0.024

BPCt-2

GCHC1_1

2386.51187

201.3741

11.85

0.000

ACH1_1_

0.39061

0.08135

4.80

0.000

GCH1_1_

−0.00006

0.48511

−0.00

0.999

Based on the results of the analysis in Table 5, the

best model for the closing price of JKSE data before

the COVID-19 pandemic is AR(2)–GARCH(1,1),

so the best model estimation is:

Mean Model AR(2):

  

 

  

and the variance model, GARCH(1,1):



   

 



where  is the closing price of JKSE data before

the COVID-19 pandemic,  is the residual at time t,

Minimum information criterion based on AICC for JKSE

Closing Price during the COVID-19 Pandemic

Lag

MA 0

MA 1

MA 2

MA 3

MA 4

MA 5

AR 0

8.958

8.965

8.938

8.933

8.893

8.888

AR 1

8.958

8.963

8.941

8.927

8.897

8.890

AR 2

8.910

8.921

8.914

8.909

8.898

8.889

AR 3

8.916

8.919

8.913

8.912

8.902

8.893

AR 4

8.895

8.897

8.902

8.901

8.905

8.897

AR 5

8.881

8.892

8.896

8.901

8.904

8.901

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.64

Florentina Kurniasari, Eko Endarto,

Helena Dewi, Cynthia Sari Dewi, Nurhuda Nizar

E-ISSN: 2224-2899

699

Volume 20, 2023

and 

 is variance at time t. The AR(2) model for the

closing price of JKSE data before the COVID-19

pandemic was chosen as the best model based on the

optimum AICC. AR(2) means that the current price

is influenced by the price of the previous 2 days. The

constant value of the JKSE closing price before the

COVID-19 pandemic was 1.09601, which means

that if the other variables are fixed, the closing price

of JKSE data before the COVID-19 pandemic is

1.09601. If the closing price of the JKSE before the

COVID-19 pandemic at time t-1 (BPCt-1) increases

by 1 unit and other variables are held constant, then

the closing price of the JKSE before the COVID-19

pandemic will increase by 0.01269. If the closing

price of the JKSE data before the COVID-19

pandemic at time t-2 (BPCt-2) increases by 1 unit and

other variables are held constant, then the closing

price of the JKSE before the COVID-19 pandemic

will decrease by 0.10055.

Table 6. Parameter estimation model AR(5)-

GARCH (1,1) for the closing price of JKSE data

during the COVID-19 pandemic

Model parameter estimates the closing price of JKSE data during

the COVID-19 pandemic

Parameter

Estimate

Std

error

value

P-

value

Variable

CONST1

4.0686

3.0234

1.35

0.1791

AR1_1_1

−0.2081

0.0553

−3.76

0.0002

DPCt-1

AR2_1_1

−0.1107

0.0531

−2.08

0.0377

DPCt-2

AR3_1_1

0.0195

0.0524

0.37

0.7106

DPCt-3

AR4_1_1

0.1160

0.0505

2.29

0.0222

DPCt-4

AR5_1_1

0.0662

0.0485

1.37

0.1726

DPCt-5

GCHC1_1

592.9424

174.357

3.40

0.0007

ACH1_1_1

0.4468

0.0565

7.91

0.0001

GCH1_1_1

−0.8378

0.0359

−23.33

0.0001

From Table 6, the model chosen for the closing

price of JKSE data during the COVID-19 pandemic

is AR(5)–GARCH(1,1), so the best estimated model

is:

Mean Model AR(5):

    

   

   



And the variance model, GARCH(1,1):



   





.

Here,  is the closing price of JKSE data during

the COVID-19 pandemic,  is residual at time t,

and 

 is the variance at time t. For the closing price

of JKSE data during the COVID-19 pandemic based

on the optimum AICC, the AR(5) model is the best

model. AR(5) means that the current price is

influenced by the price of the previous 5 days. The

constant value of the closing price of JKSE data

during the COVID-19 pandemic was 40,686, which

means if the other variable is zero (0), then the

closing price of JKSE data before the COVID-19

pandemic is 4,0686. If the closing price of JKSE

data during the COVID-19 pandemic at time t-1,

DPCt-1, increases by 1 unit and other variables are

held constant, then the closing price of JKSE data

during the COVID-19 pandemic will decrease by

0.0662. If the closing price of JKSE data during the

COVID-19 pandemic at time t-5, DPCt-5, increases

by 1 unit and other variables are held constant, then

the closing price of JKSE data during the COVID-

19 pandemic will decrease by 0.1005.

(a)

(b)

Fig. 5: Distribution of error predictions and Q-Q

plots for data (a) JKSE closing price before and (b)

during the COVID-19 pandemic

From the results of the normality test of the closing

price of JKSE data before and during the COVID-19

pandemic, the residual value of the data is not

normally distributed because the P-value is smaller

than 0.05. However, the error distribution and the Q-

Q plot for the closing price of JKSE before (Fig. 5a)

and during the COVID-19 pandemic in Fig. 5b

shows that there is a deviation that is not too far from

the normal distribution although statistically the null

hypothesis is rejected, but the error distribution

graph is close to normal.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.64

Florentina Kurniasari, Eko Endarto,

Helena Dewi, Cynthia Sari Dewi, Nurhuda Nizar

E-ISSN: 2224-2899

700

Volume 20, 2023

(a)

(b)

Fig. 6: Conditional variance (a) Model AR(2)-

GARCH(1,1) for the closing price of JKSE data

before the COVID-19 pandemic. (b) Model AR(5)-

GARCH(1,1) for the closing price of JKSE data

during the COVID-19 pandemic

Figure 6(a) shows the ARCH effect model

AR(2)-GARCH(1,1) for the closing price of

JKSE data before the COVID-19 pandemic,

from January 2018 to September 2018 the volatility

was high, this condition showed that prices were

very unstable and unpredictable. From October

2018 to December 2019, the volatility is relatively

small; this shows that data changes in this time

range are relatively stable in terms of not drastic ups

and downs. Figure 6(b) is the conditional variance

for the AR(5)-GARCH(1,1) model showing high

volatility from January 2020 to April 2020; this

shows that the closing price of JKSE data is very

unstable, and this in line with Figure 3, indicating a

very drastic decrease in the closing price and this

situation also indicating that the prices were

unpredictable in this time range. From May 2020 to

December 2021, the volatility is relatively small,

this shows that the price changes up and down are

not drastic.

Forecasting

Forecasting is carried out for the next 30 days.

(a)

(b)

Fig. 7: (a) Plot of data and forecasting for the next

30 days for the closing prices of JKSE data before

the COVID-19 pandemic and (b) data and

forecasting for the closing prices of JKSE data

during the COVID-19 pandemic

Table 7. Forecasting data for the next 30 days for

the closing price of JKSE data before and during the

COVID-19 pandemic

Forecasting the next 30 days for JKSE closing price data before

COVID-19

Obs

Forecast

Standard

error

95% confidence limits

486

6299.26

50.20

6200.87

6397.66

487

6303.35

73.18

6159.91

6446.79

488

6304.53

88.03

6131.99

6477.06

489

6305.23

100.44

6108.36

6502.09

490

6306.21

111.66

6087.37

6525.06

491

6307.25

121.86

6068.40

6546.10

492

6308.26

131.27

6050.98

6565.54

493

6309.27

140.04

6034.80

6583.73

494

6310.27

148.29

6019.63

6600.91

495

6311.28

156.11

6005.32

6617.24

496

6312.29

163.55

5991.73

6632.84

497

6313.30

170.67

5978.79

6647.81

498

6314.30

177.51

5966.40

6662.21

499

6315.31

184.09

5954.51

6676.11

500

6316.32

190.44

5943.06

6689.57

501

6317.33

196.59

5932.02

6702.63

502

6318.33

202.55

5921.34

6715.33

503

6319.34

208.34

5911.00

6727.68

504

6320.35

213.98

5900.96

6739.74

505

6321.36

219.47

5891.21

6751.50

506

6322.36

224.82

5881.72

6763.01

507

6323.37

230.05

5872.47

6774.27

508

6324.38

235.17

5863.45

6785.30

509

6325.39

240.18

5854.65

6796.12

510

6326.39

245.08

5846.05

6806.74

511

6327.40

249.89

5837.63

6817.17

512

6328.41

254.60

5829.39

6827.42

513

6329.42

259.24

5821.32

6837.51

514

6330.42

263.79

5813.41

6847.43

515

6331.43

268.26

5805.65

6857.21

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.64

Florentina Kurniasari, Eko Endarto,

Helena Dewi, Cynthia Sari Dewi, Nurhuda Nizar

E-ISSN: 2224-2899

701

Volume 20, 2023

Forecasting the next 30 days for JKSE closing price data during

COVID-19

Obs

Forecast

Standard

error

95% confidence limits

449

6559.03

66.44

6428.80

6689.25

450

6558.76

86.08

6390.04

6727.48

451

6553.24

100.19

6356.88

6749.60

452

6565.20

114.20

6341.38

6789.02

453

6568.76

129.77

6314.42

6823.09

454

6571.30

142.74

6291.54

6851.06

455

6574.62

154.65

6271.51

6877.73

456

6579.60

166.16

6253.93

6905.27

457

6583.29

177.24

6235.90

6930.68

458

6586.96

187.65

6219.17

6954.75

459

6590.87

197.62

6203.54

6978.20

460

6594.89

207.24

6188.70

7001.08

461

6598.73

216.52

6174.35

7023.11

462

6602.60

225.47

6160.68

7044.52

463

6606.50

234.15

6147.59

7065.42

464

6610.41

242.56

6134.99

7085.82

465

6614.29

250.74

6122.84

7105.73

466

6618.18

258.70

6111.14

7125.22

467

6622.07

266.45

6099.84

7144.31

468

6625.97

274.02

6088.90

7163.03

469

6629.86

281.40

6078.32

7181.40

470

6633.75

288.62

6068.06

7199.44

471

6637.64

295.68

6058.11

7217.17

472

6641.53

302.60

6048.45

7234.62

473

6645.42

309.37

6039.06

7251.79

474

6649.32

316.02

6029.93

7268.70

475

6653.21

322.54

6021.05

7285.37

476

6657.10

328.93

6012.40

7301.80

477

6660.99

335.22

6003.97

7318.01

478

6664.88

341.40

5995.76

7334.01

From Table 7 and Figure 7(a), predictions for the

next 30 days for the closing price of JKSE before

the COVID-19 pandemic are as follows: to

experience an increasing trend, assuming normal

conditions or no COVID-19 pandemic. However, if

we compare them with the closing price of JKSE

data plot at the beginning of COVID-19 (Figure

7(b)) in the month of January 2020 to April 2020,

the closing price of JKSE data has decreased

drastically; this shows that the COVID-19 pandemic

caused price chaos and the closing price of JKSE

decreased drastically and unpredictable. This means

that the COVID-19 pandemic has had a huge impact

on changes in the closing price of JKSE. For

forecasting when COVID-19 occurs (Table 7 and

Figure 7(b)), it shows an increasing trend. The

forecasting results obtained in this study can only be

used for short-term periods because we can see that

the risk for longer periods increases significantly

over time. This can be seen from the value of the

confidence interval for both the closing price of

JKSE data before and during the COVID-19

pandemic, which shows that for a long forecast

period, the predictive value is unstable (high

standard error).

4 Conclusion

This study discusses the behaviour of the closing

price of JKSE data 2 years before the COVID-19

pandemic (2018–2019) and 2 years during the

COVID-19 pandemic (2020–2021).

The best model for the closing price of JKSE

data before the COVID-19 pandemic is AR(2)-

GARCH (1,1), while the best model for the closing

price of JKSE data during the COVID-19 pandemic

is AR(5)-GARCH(1,1).

The results of data forecasting for models before

the COVID-19 pandemic describe an increasing

trend, assuming normal conditions and no COVID-

19 pandemic. However, in reality, there was a

COVID-19 pandemic and the closing price of JKSE

data experienced a very drastic decline in the period

January 2020 to April 2020; this shows that the

COVID-19 pandemic has had a very high impact on

price changes, which decreased drastically from the

closing price of the JKSE. This large and unstable

price is also indicated by the high value of the

volatility data in the range of January 2020 to April

2020. Forecasting for the next 30 days for the

closing price of JKSE data during COVID-19 shows

an increasing trend, although the increase is small.

The growing domestic demand in consumption

and investment and the improvement of global

economic conditions will speed up Indonesia’s

economic recovery into a better curve. The

Indonesian government also actively participates in

maintaining the financial market stability. There

was a stock market anomaly during the COVID-19

pandemic in Indonesia in that the pandemic

interacted positively with the Indonesian stock

market.

The increasing trend of JKSE stock price

showed the higher confidence and trust from

investors to the government's commitment to

achieve the target of the country's economic growth

at 4.8% in the year 2022.

There is an urgency to continue in implementing

new reform of energy subsidies as it has had an

important impact on GDP growth by spending

priorities in fuel consumption and reducing energy

subsidy to boost economic development across the

country.The non-tax state revenue from the oil and

gas sector was expected to rebound aligned with the

good performance of export commodities.

The results of this study have several practical

implications. First, the existence of a stock market

anomaly implies investors’ confidence in future

returns and in an eventual market recovery.

Therefore, financial market authorities should

implement strategies to maintain and even increase

investors’ confidence. In addition, the government

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.64

Florentina Kurniasari, Eko Endarto,

Helena Dewi, Cynthia Sari Dewi, Nurhuda Nizar

E-ISSN: 2224-2899

702

Volume 20, 2023

should intervene by implementing stimulus

packages to alleviate stock market panic and avoid

capital outflow from the country.

The future research should be able to consider

other variables that can influence the JKSE

performance growth such as: the exchange rate;

foreign capital outflow/inflow and other industries

which had major contribution in JKSE, including:

financial industries.

References:

[1] Petriella, Y., Ada Corona, Insentif Penurunan

Royalti Batu Bara Sangat Diperlukan [in

Indonesian], Ekonomi, (2020a, March 28).

https://ekonomi.bisnis.com/read/20200328/44/

1219082/ada-corona-insentif-penurunan-

royalti-batu-bara-sangat-diperlukan

[2] Utami, S. S., Akibat Covid-19, Pertamina

Setop Operasi Kilang Balikpapan [in

Indonesian], MSN, (2020b, April 18).

https://www.msn.com/id-

id/news/other/akibat-covid-19-pertamina-

setop-operasi-kilang-balikpapan/ar-BB12O7rf

[3] Arvirianty, A., Hingga Oktober, Impor LPG

RI Sentuh Rp 36 T, CNBC Indonesia [in

Indonesian], 2018.

https://www.cnbcindonesia.com/news/201811

30144755-4-44447/hingga-oktober-impor-

lpg-ri-sentuh-rp-36-t.

[4] Griffith, R., Levell, P., & Stroud, R., The

impact of COVID-19 on share prices in the

UK, Fiscal Studies, 41(2), 2020, 363–369.

https://doi.org/10.1111/1475-5890.12226

[5] Sugandi, E. A., Indonesia’s financial markets

and monetary policy dynamics amid the

Covid-19 pandemic, ADBI Working Paper

1198, Asian Development Bank Institute

2020.

https://www.adb.org/publications/indonesia-

financial-markets-monetary-policy-dynamics-

covid-19-pandemic.

[6] Topcu, M., & Gulal, O. S., The impact of

COVID-19 on emerging stock markets.

Finance Research Letters, 36, 101691, 2020.

https://doi.org/10.1016/j.frl.2020.101691

[7] Hong, H., Bian, Z., & Lee, C. C., COVID-19

and instability of stock market performance:

Evidence from the U.S, Financial Innovation,

2021.

[8] Kurniasari, F. et.al., The role of financial

technology to increase financial inclusion in

Indonesia, International Journal of Data and

Network Science ,5, 2021, 391–400.

[9] Sembiring, L. J., Subsidi Turun, Harga BBM

dan Tarif Listrik Naik, CNBC Indonesia,

2020.

https://www.cnbcindonesia.com/news/201908

20163558-4-93374/subsidi-turun-harga-bbm-

dan-tarif-listrik-naik-di-2020.

[10] Kurniati, D., Selain Manufaktur, Ini 11 Sektor

Usaha yang Bakal Dapat Insentif Pajak, 2020.

DDTC news. https://news.ddtc.co.id/selain-

manufaktur-ini-11-sektor-usaha-yang-bakal-

dapat-insentif-pajak-20345?page_y=0

[11] Ebtke, J., Dampak Covid-19 pada

Pengembangan Energi Terbarukan di

Indonesia, 2020, Institute for Essential

Services.http://iesr.or.id/wp-

content/uploads/2020/04/Bahan-Vidcon-

DJEBTKE-dengan-IESR-21-April-

2020dek.pdf.

[12] Bollerslev, T., Generalized autoregressive

conditional heteroscedasticity, Journal of

Econometrics, 31(3), 1986, 307–327.

https://doi.org/10.1016/0304-4076(86)90063-

[13] Kurniasari, et.al., The Effect of Perceived

Usefulness, Perceived Ease of Use, Trust,

Attitude and Satisfaction Into Continuance of

Intention in Using Alipay, Management &

Accounting Review, Vol. 19, No.2, 2020.

[14] Tsay, R. S. (2005), Analysis of financial time

series, John Wiley & Sons, Inc., 2005.

[15] Wei, W. W. S. , Time series analysis:

Univariate and multivariate methods, New

York: Pearson, 2006

[16] Virginia, E., Ginting, J., & Elfaki, F. A. M.,

Application of GARCH model to forecast data

and volatility of share price of energy (Study

on Adaro Energy Tbk, LQ45), International

Journal of Energy Economics and Policy,

8(3), 2018, 131–140.

[17] Warsono, W., Russel, E., Wamiliana, W.,

Widiarti, W., & Usman, M., Modeling and

forecasting by the vector autoregressive

moving average model for export of coal and

oil data (case study from Indonesia over the

years 2002–2017). International Journal of

Energy Economics and Policy, 9(4), 2019,

240–247. https://doi.org/10.32479/ijeep.7605

[18] Montgomery, D., Jennings, C., & Kulachi,

M., Introduction time series analysis and

forecasting. Hoboken, New Jersey: John

Wiley & Sons, Inc, 2008.

[19] Zivot, E., & Wang, J., Modeling financial

time series with S-plus, New York: Springer-

Verlag, 2006.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.64

Florentina Kurniasari, Eko Endarto,

Helena Dewi, Cynthia Sari Dewi, Nurhuda Nizar

E-ISSN: 2224-2899

703

Volume 20, 2023

[20] Brockwell, P. J., & Davis, R. A., Introduction

to time series and forecasting (2nd Ed.), New

York: Springer-Verlag, 2002.

[21] Shumway, R. H., & Stoffer, D. S. (2006),

Time series analysis and its applications with

R examples (2nd ed), Springer-Verlag, 2006.

[22] Ljung, G. M., & Box, G. E. P., On a measure

of lack of fit in Time Series Models.

Biometrika, 65(2), 1978, 297–303.

https://doi.org/10.1093/biomet/65.2.297

[23] Engle, R. F. (1982), Autoregressive

conditional heteroscedasticity with estimates

of the variance of United Kingdom inflation.

Econometrica, 50(4), 1982, 987–1007.

https://doi.org/10.2307/1912773.

Contribution of Individual Authors to the

Creation of a Scientific Article (Ghostwriting

Policy)

-Florentina Kurniasari and Nurhuda Nizar carried

out the modelling and statistical testing.

-Eko Endarto and Helena Dewi collecting the

secondary data (JKSE stock price).

-Cynthia Sari Dewi responsible for collecting

information of macro-economic especially in

Indonesia policy in energy sector.

Sources of Funding for Research Presented in a

Scientific Article or Scientific Article Itself

The research was conducted under the collaboration

between Faculty of Business, Universitas

Multimedia Nusantara and Fakulti Pengurusan dan

Perniagaan, Universiti Teknologi MARA with the

Reference Number: 500-FPP(PT.23/1).

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.64

Florentina Kurniasari, Eko Endarto,

Helena Dewi, Cynthia Sari Dewi, Nurhuda Nizar

E-ISSN: 2224-2899

704

Volume 20, 2023

Conflict of Interest

The authors have no conflicts of interest to declare

that are relevant to the content of this article.

Creative Commons Attribution License 4.0

(Attribution 4.0 International, CC BY 4.0)

This article is published under the terms of the

Creative Commons Attribution License 4.0

https://creativecommons.org/licenses/by/4.0/deed.en

_US