COVID-19 and Changes in Indonesia’s Sharia Stock Market
Capital markets play an important role in economic development by connecting companies that need funding with investors seeking opportunities to grow their wealth. In Indonesia, the Jakarta Islamic Index provides a benchmark for monitoring the performance of selected stocks that comply with Sharia principles.
Launched on July 3, 2000, the JII comprises 30 relatively liquid Sharia-compliant stocks listed on the IDX. Its constituents are selected according to established criteria, including market capitalization and trading liquidity, and are reviewed periodically.
The COVID-19 pandemic created substantial uncertainty in global financial markets. Economic disruptions, changing investor expectations, and shifts in risk perceptions affected stock prices across many countries, including Indonesia. As markets moved from the crisis period toward recovery, historical patterns of stock index movements could also change.
Azzakhy and Sumarto examined whether the statistical characteristics of the JII differed between two periods: the COVID-19 pandemic from 2020 to 2022 and the post-pandemic period from 2023 to 2025. Their research provides a quantitative comparison of the models that best described the index's historical movements in each period.
How Researchers Analyzed the Jakarta Islamic Index
The researchers used the Autoregressive Integrated Moving Average (ARIMA) method, a statistical approach designed to identify patterns in data collected over time. ARIMA uses previous observations and past prediction errors to model a time series and can support forecasting when its assumptions and predictive performance are properly evaluated.
The study used daily JII closing prices obtained from Investing.com. The dataset was divided into two groups:
- Pandemic period (2020–2022): 735 trading days.
- Post-pandemic period (2023–2025): 714 trading days.
The researchers processed the data using Microsoft Excel and EViews 12. They first checked whether the data were stationary, meaning their statistical properties were sufficiently stable for the modeling procedure. Because the original JII data were non-stationary, the researchers applied first-order differencing to analyze changes between consecutive observations.
They then examined autocorrelation patterns, estimated candidate ARIMA models, and assessed whether the remaining model errors showed significant autocorrelation. The Akaike Information Criterion (AIC) and Schwarz Information Criterion (SIC) were used to compare candidate models, with lower values indicating a better balance of model fit and complexity among the models being compared.
Two Different ARIMA Models Emerged
The findings revealed that different ARIMA specifications best represented JII movements during and after the pandemic.
For the pandemic period, the selected model was ARIMA([2,3],1,[9]). It incorporated autoregressive components at lags two and three, along with a moving-average component at lag nine.
For the post-pandemic period, the selected model was ARIMA([1],1,[13]). This model included an autoregressive component at lag one and a moving-average component at lag thirteen.
The model comparison produced the following results:
- Pandemic model: AIC of 6.946495 and SIC of 6.977820.
- Post-pandemic model: AIC of 6.265914 and SIC of 6.291549.
Both models used first-order differencing and met the study's residual white-noise criterion. Their Durbin–Watson statistics were approximately 2.02 for the pandemic model and 2.00 for the post-pandemic model, consistent with the reported conclusion that significant residual autocorrelation was not detected.
The lower AIC and SIC values for the post-pandemic model indicated a better fit according to these criteria. However, this does not establish that the post-pandemic market was necessarily less risky or that the model could predict future prices more accurately. The comparison concerns relative model fit within the analyzed periods, not independently verified forecasting performance.
What the Differences Mean for Market Analysis
The different model structures suggest that the historical patterns used to describe JII movements changed between the pandemic and post-pandemic periods.
During the pandemic, the selected model required two autoregressive components, indicating that observations at two different historical lags contributed to the model. The post-pandemic model retained one autoregressive component, while its moving-average component appeared at a longer lag.
These results are consistent with the possibility that market disruptions and changing investor expectations affected the statistical behavior of the index. Nevertheless, the ARIMA analysis alone cannot establish that COVID-19 was the sole cause of the differences. The model used historical JII closing prices as its only variable and did not directly incorporate external factors such as inflation, interest rates, exchange rates, government policies, or geopolitical developments.
In their discussion, Azzakhy and Sumarto emphasize that the most appropriate model depends on the characteristics of the data and the market conditions being analyzed. This insight is relevant because a model that describes a market during a crisis may not be the best choice once economic conditions change.
Implications for Investors and Future Research
The findings offer a practical lesson for analysts and participants in Indonesia's Sharia-compliant capital market: statistical models should be reviewed periodically rather than applied indefinitely without reassessment.
For investors, the study provides a framework for understanding how historical stock index movements can be analyzed under different market conditions. However, the selected models should not be treated as standalone tools for making buying or selling decisions. The research did not proceed to out-of-sample forecasting or evaluate prediction accuracy, so it does not establish how well either model would perform on future JII prices.
The authors recommend extending the analysis by evaluating forecasting accuracy through measures such as Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE). They also suggest comparing ARIMA with methods designed to capture changing volatility, including Generalized Autoregressive Conditional Heteroskedasticity (GARCH) and hybrid ARIMA-GARCH models, as well as machine-learning approaches such as Long Short-Term Memory (LSTM).
These extensions could help researchers determine whether more complex methods provide better forecasts during periods of financial uncertainty. For policymakers, understanding how market dynamics change during economic crises may also contribute to broader assessments of financial market resilience and the need for responsive stabilization policies.
Overall, the study by Azzakhy and Sumarto demonstrates that the statistical patterns of the Jakarta Islamic Index differed between the pandemic and post-pandemic periods. Its central contribution is not a guarantee of future market direction, but evidence that historical stock market data should be evaluated in the context of changing economic conditions.
Author Profiles
Muhammad Fadhil Azzakhy is affiliated with Universitas Mercu Buana, Indonesia. His contribution to this article focuses on quantitative analysis of the Jakarta Islamic Index using time-series modeling. The published article does not provide a detailed academic biography or specify his academic degree.
Agus Herta Sumarto is affiliated with Universitas Mercu Buana and is the corresponding author of the article. The paper acknowledges him as the researcher's thesis adviser and identifies his academic credentials as Dr. Agus Herta Sumarto, S.P., M.Si. His contribution to this work concerns the analysis of stock index movements using statistical methods.
Research Source
Article title: “Analysis of the Jakarta Islamic Index (JII) Using the Autoregressive Integrated Moving Average (ARIMA) Method During and After the COVID-19 Pandemic”
Authors: Muhammad Fadhil Azzakhy and Agus Herta Sumarto
Journal: International Journal of Sustainable Applied Sciences (IJSAS)
Publication year: 2026
Volume and pages: Volume 4, Issue 10, pp. 1381–1392
DOI: https://doi.org/10.59890/ijsas.v4i10.78
Official journal website: http://ijsasjournal.my.id/index.php/ijsas
0 Komentar