“Stock price volatility during the COVID-19 pandemic: The GARCH model”

This study examined the response of stock prices on the Indonesia Stock Exchange (IDX) to COVID-19 using an event study approach and the GARCH model. The research sample is the closing price of the Composite Stock Price Index (JCI) and companies that are members of LQ-45 in the 40-day period before the COVID-19 incident, 1 day during the COVID-19 incident (March 2, 2020) and 10 days after, January 6, 2020 – March 16, 2020. Empirical findings prove that abnormal returns react negatively to COVID-19, JCI volatility fluctuates widely during the COVID-19 event, and the GARCH(1,2) model can be used to assess volatility and predict stock abnormal returns in IDX in market conditions infected with COVID-19. The practical implication of the study’s findings for investors is that the COVID-19 event caused stock price volatility, which affects abnormal returns. Therefore, to face the conditions of uncertainty and increased volatility in the future, several lines of risk management are needed in managing a stock portfolio. In addition, it also opens up opportunities for speculators to profit in an inefficient market environment. This study is based on the empirical literature currently being developed to investigate the phenomenon of stock price volatility behavior during COVID-19 on the IDX. The GARCH model used proves that during the COVID-19 pandemic, stock price volatility increases and leads to a decrease in abnormal returns. The empirical findings also validate the efficient market hypothesis theory related to the study of events and the theory of financial behavior related to uncertainty.


INTRODUCTION
The COVID-19 pandemic has not only infected many people, but has also caused the world economy to go into recession.According to the International Monetary Fund's (IMF) report, the world economy has experienced a drastic decline and is facing a crisis after around 95 percent of countries in the world experienced negative economic growth contractions.The IMF also reported that COVID-19 also caused the global economy to suffer a loss of US$12 trillion (IMF, 2020).The world's financial markets, especially the stock exchanges, have also been infected by COVID-19, since the announcement of positive cases in early 2020 caused securities prices to fall and volatility to become high.Especially for the stock market, there was a drastic decline, where the market value lost about 30% in a few weeks.COVID-19 has caused severe shocks in most stock markets, including Indonesia.For Indonesia, COVID-19 has caused negative economic growth of 2.19 percent, and the Composite Stock Price Index (JCI) was corrected by 5.09 percent in 2020.High JCI volatility continues to this day due to the rapid spread of COVID-19.High volatility indicates that stock prices are increasingly volatile.
Many previous studies have tested the impact of the COVID-19 event on stock price volatility, but conflicting empirical evidence is still found.Most of the empirical research findings confirm that stock market move-ments decline due to cases of COVID-19 transmission (Rahman et al., 2021;Cao et al., 2020;Ahmar & del Val, 2020).Ashraf (2020) also found poor stock market performance in 64 countries due to being infected with COVID-19.The US stock market also corrected due to the increase in positive cases of COVID-19 (Alfaro et al., 2020).By using the event study method, several findings prove that during COVID-19, the global stock market experienced a decline (Heyden & Heyden, 2020;Liu et al., 2020a).Xu (2001) proved that the Canadian stock market experienced a decline as a result of an increase in confirmed COVID-19 cases.2001) revealed that the stock markets of European countries reacted differently to the COVID-19 events.For the most common risk measurements, volatility is usually used.Modeling uncertainty with volatility can evaluate a portfolio by considering two factors, namely investment opportunities and potential losses.Therefore, the measurement of volatility becomes an evaluation guide in determining investment decisions to assess assets and risks (Naik & Reddy, 2021).This study uses the event study method to measure the impact before and during the COVID-19 pandemic in Indonesia on volatility and abnormal returns on the IDX using the GARCH model.Abnormal returns in the event study method is calculated in the period of 30 days before to 30 days after.The test results show that the property and real estate, construction, and financial sectors experienced a decline in abnormal returns (AR), while the infrastructure, utilities, and transportation sectors with AR tended to be constant, and other sectors experienced an increase in AR.Judging from the cumulative abnormal return value, the sectors most affected are finance, trade, services, and investment.Mining and consumer goods sectors did not have much influence, while other sectors showed negative sentiment.Overall, COVID-19 has caused stock prices to drop drastically, where the cumulative value is negative from the average AR.Utomo and Hanggraeni (2021) studied the impact of COVID-19 on the daily stock returns of 272 companies listed on the IDX and concluded that stock prices declined during the pandemic.Ryandono et al. (2021) also prove the negative impact on Indonesian Islamic stocks.

LITERATURE REVIEW
Volatility is a barometer of measuring the risk of facing uncertainty that investors consider when purchasing financial assets.Risk considerations are the main parameter in making investment portfolio decisions (Suryadi et al., 2021).Stock price volatility is a systematic risk faced by investors holding market portfolios (Nurhayati et al., 2021a).Stock market volatility tends to increase drastically if there are important events that have a broad economic and financial impact, such as the 1997/98 Asian economic crisis, and most recently the health crisis due to COVID-19.Although the causes of stock market volatility are not well understood, higher stock market volatility can lead to pessimism about future economic developments (Razak et al., 2020).Schwert (1989) states that stock market volatility indicates that future cash flows and discount rates are uncertain.2021) apply various GARCH models, which consist of the GARCH, PGARCH, EGARCH, TGARCH and GARCH-M standards, to test return volatility on the Malaysian and Singapore stock exchanges.The event period is divided into two, namely before and during the COVID-19 pandemic, using daily closing price data from the stock market index between July 1, 2019 to August 31, 2020.Empirical findings prove that there is persistence in both stock markets, and the persistence has decreased during the pandemic.Empirical research evidence suggests that after COVID-19, volatility persistence increases across all series.

FRAMEWORK AND HYPOTHESES DEVELOPMENT
This paper introduces the GARCH model to forecast volatility and abnormal returns during COVID-19.The framework of this study can be stated in the diagram presented in Figure 1.

Population and sample
The population in this study is the daily closing price of LQ45 shares and the Composite Stock Price Index (JCI).The sample of this study is the closing price of the JCI and companies that are members of LQ45 in the period of 40 days before the COVID-19 incident, 1 day during the COVID-19 incident (March 2, 2020) and 10 days after that, January 6, 2020 -March 16, 2020.LQ45 shares selected as the research sample based on the following criteria: 1.The companies are consecutively incorporated in the LQ45 share list during the observation period, namely August -February 2020 and February -July 2020.There are 42 companies that meet these criteria.
2. Do not perform corporate actions during the window period.Of the 42 companies, four companies conduct corporate actions, so the sample for this study is 38 companies.

Analysis models and variables
Hypothesis testing in this study uses a quantitative approach using the event study method and the GARCH model.The stages of the research carried out started from data collection and data processing, calculation and analysis of abnormal returns and JCI volatility, as well as forecasting abnormal stock returns and JCI price volatility.The research findings can provide both theoretical and practical implications.The following is a formula for calculating stock price volatility, abnormal returns and the GARCH model.where PV -share price volatility, t H -the highest share price in period , t t L -the lowest stock price in period , t and n -the number of observations.

Event window and samples
The COVID-19 was announced on March 2, 2020.This was one of the variables that were chosen as the period event.The event window used in this study is 61 days, of which 30 days before the COVID-19 announcement, 30 days after the COVID-19 announcement and 1 event date on the COVID-19 announcement.
The observation period T 0 -T 1 is the estimation window, T 1 -T 2 is the event window, and T 2 -T 3 is the post-event period.

RESULTS
The Indonesian stock exchange's reaction to the COVID-19 event is indicated by significant abnormal returns during the event period.The results of measuring the average abnormal return are summarized in Table 1.The daily AAR stock return value can be positive or negative.A indicates that the average actual return of stocks during the event period is greater than the average return expectation predicted by investors (Suryanto, 2015).
The AAR during the event period shows 61.90% or 13 trading days, which are positive, and 38.10% or 8 trading days, which are negative.The estimation results on proving the effect of volatility on abnormal returns using the GARCH (1,2) model are written in the following equation: 0.0058 1.0577 .

Abnormal Return
The estimation results for the volatility measured by variance and residuals from the GARCH (1,2) model are shown in Table 4. From Table 4, the test estimation results from the GARCH (1,2) model can be written in the form of an equation as follows: From the GARCH(1,2) model, estimation equation produces a constant value of 2.5400.The ARCH coefficient is 1.7293, indicating a fluctuating change in volatility that causes stock market movements to experience an unstable tendency.The GARCH(1,2) coefficient value of 0.10876 indicates that for every movement that occurs in a variable, it shows that the variable quickly returns to a stable condition.
Testing the R 2 test to measure how much the independent variable's contribution can explain the variations that occur in the dependent variable changes is shown in the results of the GARCH model R 2 test (1,2) (Table 5).
The value of the R 2 termination coefficient of the GARCH model (1,2) shows the adjusted R 2 value of 0.0679, which indicates the ability of the independent variable; JCI volatility can explain the correla-tion of data variations of 67.9%, while the remaining 32.1% (100% -67.9% = 32.1%) is influenced by other factors outside the model.Statistical testing The t-test was conducted to prove the influence of the independent variables individually on the dependent variable.This test is based on a comparison between T-count with a 5% confidence degree.If the significance of T-count < 5%, then each explanatory variable affects the dependent variable, or vice versa, the GARCH model test (1,2) shows the dependent variable affects the independent variable.The t-test statistic test concluded that the volatility of the JCI on abnormal returns has a probability (0.0012) < 5% with a coefficient value of 3.380048, so that the volatility of the JCI with abnormal returns has a positive effect.

DISCUSSION
The results of the event study test found that negative abnormal returns did not appear on t0 or the event date.This means the capital market does not react directly to government announcements about positive COVID-19 patients.Abnormal returns begin to appear when t2 and t3 show investors who are waiting and see, namely there are actions taken by investors to temporarily move their assets to safe heaven (safer instruments such as deposits and others).Negative and significant abnormal returns reappear at t7 and t8 or the seventh and eighth day after the event.This is because investors are still worried about the COVID-19 pandemic, which continues to claim victims, in addition to affecting several businesses of issuers.
The volatility of the JCI has an impact on abnormal returns, with the increasing risk of higher stock prices causing abnormal returns in IDX to experience pressure.The study by Herwany et  The increased volatility of the JCI and negative abnormal returns were evident in the Indonesian stock market based on the observation period after the 11th day since the COVID-19 incident.The GARCH (1,2) model can be used for forecasting in this study (Endri et al., 2020).This finding supports the study by Chaudhary et al. (2020), which also applies the GARCH model to the stock exchanges of the top 10 countries based on GDP that experience negative average daily returns.The volatility of the stock market indices in response to the key events of COVID-19 has led to much higher volatility than normal, and also indicates a bearish trend in the market that occurred in the second quarter of the COVID-19 period.In GARCH modeling with COVID-19 as an exogenous variance regressor, it was found that all market indices were infected.The study by Bai et al. (2021) with the developed GARCH model proves that within 24 months, COVID-19

CONCLUSION
COVID-19 has infected most of the world's stock markets, including Indonesia, as evidenced by the high volatility that has an impact on the decline in stock prices, with many investors suffering losses.In the current state of uncertainty, research related to the volatility of stock price movements has begun to emerge and the debate about these results has become an interesting discussion.This study provides theoretical and empirical benefits of testing the efficient market hypothesis in the context of Indonesia, which is still an emerging stock market, to test an event originating from the COVID-19 outbreak by measuring stock price volatility using the GARCH model.The test results prove that COVID-19 has made stock prices drop drastically.Investors responded directly to COVID-19 by selling their shares so that abnormal returns became negative.JCI volatility has been shown to have a unidirectional relationship with abnormal returns during the COVID-19 period, meaning that if JCI volatility increases, abnormal returns will increase.The GARCH model can be used to assess volatility and predict abnormal returns on Indonesian stocks when the COVID-19 incident occurs.The model that can be used is GARCH (1,2).The empirical findings of the study have implications related to the possibility that the crisis in the stock market caused by the COVID-19 pandemic outbreak in the context of market inefficiency opens up opportunities for speculators to profit.
Empirical findings that show a positive response to stock prices due to the COVID-19 event are proven (among others, Liu et al., 2020b; Brueckner & Vespignani, 2021; Yong et al., 2011).Liu et al. (2020b) found that the United States stock exchange (return) responded positively to COVID-19.Brueckner and Vespignani (2021) point out that the performance of the Australian stock exchange has improved due to the COVID-19 infection.Shahzad et al. (2001) show evidence of a time-varying and substantially intense impact of asymmetric volatility during COVID-19.Duttilo et al. ( Mazur et al. (2020) found that extreme asymmetric volatility is inversely related to stock returns.Studies examining stock price volatility due to the impact of COVID-19, among others, were carried out by Baker et al. (2020), Zaremba et al. (2020), Albulescu (2020), Sharma (2020), and Engelhardt et al. (2020).Baker et al. (2020) proved that stock price volatility increased after being infected with COVID-19.This finding is also supported by Albulescu (2020) and Zaremba et al. (2020) who conclude that stock price volatility has increased quite significantly.Bai et al. (2020) prove the persistent effect of COVID-19 on stock volatility.Sharma (2020) also proves that stock price volatility is infected by COVID-19, but the effect is different for each country, where the stock market belonging to the high-income country group initially overreacts and bounces back faster than the low-income country group.Engelhardt et al. (2020) revealed that stock price volatility trends in response to COVID-19 depend on confidence: low volatility occurs in high confidence countries.Stock volatility testing uses the traditional GARCH model with indicators based on stock price behavior and past volatility (Endri et al., 2020; Gokcan, 2000; Emenogu et al., 2020).The GARCH model is also used by Fakhfekh et al. (2021) to analyze dynamics of the Tunisian sectoral stock market index volatility during COVID-19.Yong et al. ( the residual variant in the previous period(GARCH component).

Table 1 .
Calculation of the daily AAR of shares during the observation period

Table 2 .
Hypothesis testing results using the ttest

Table 3
Abnormal return results

Table 4
Volatility results al. (2021) also gives the same result that the COVID-19 pandemic has had a negative impact on the AR stocks in IDX.Nurhayati et al. (2021b) prove that COVID-19 has caused stocks to underperform.Khatatbeh et al. (2020) provide evidence of a lack of reaction to pandemic announcements, but significant CAR with a delayed stock market response.Baek et al. (2020) also prove that COVID-19 has had a significant impact on US stock market volatility.A different finding was revealed by Apergis and Apergis (2020) that COVID-19 had a positive effect on the volatility of stock market returns in China.

Table 5 .
Result of the determination coefficient test (R2 Test) of the GARCH (1,2) model

Table 6 .
T-statistical test results has consistently had an impact on increasing volatility in most global stock exchanges.Gherghina et al. (2021), using the GARCH (1,1) model, prove that the daily return volatility in the Romanian stock market increases.Bora and Basistha (2021) using the GARCH model also prove that during COVID-19, stock market volatility in India increases.