5.1. Analyses before COVID-19 Lockdown
Table 3 presents the results of the normality test before the COVID-19 lockdown. The Asymp. Sig. (two-tailed) result was 0.346, which is greater than 0.05. Therefore, it can be concluded that the data were normally distributed, or, in other words, the data passed the normality test.
Table 4 presents the results of the normality test after the COVID-19 lockdown. The table shows the value of Asymp. Sig (two-tailed) to be 0.349, which is greater than 0.05. Therefore, it can be concluded that the data were normally distributed, or, in other words, the data passed the normality test.
Table 5 presents the results of the paired sample
t-test before and after the COVID-19 lockdown. The table shows the results of the statistical tests for different tests before and after the lockdown. The results show that for Pair 1, the difference in foreign investment flows before and after the lockdown with Sig. (two-tailed) was 0.3485, which is greater than 0.05. Therefore, Pair 1 showed no significant difference for foreign investment flows before and after the lockdown. The absence of a difference between foreign investment flows in this study is not consistent with the findings of
Moosa and Merza (
2022), who suggested that there were differences between foreign investment flows before and after the lockdown. However, although there was no difference between foreign investment flows before and after the lockdown in this study, as reported on the news portal of
The Jakarta Post (
Rahman 2020), domestic investors dominated the Indonesian stock market during the pandemic, and foreign investors, according to data from the Ministry of Finance, diverted their investments to safe assets or government bonds. In addition, as reported by the news portal of
The Jakarta Post, many foreign investors withdrew their investment due to the uncontrolled situation, especially when the lockdown was announced to try to overcome the spread of the virus by closing several economic sectors.
In Pair 2, this study analyzed the stock turnover rate or overconfidence, resulting in a Sig. (two-tailed) value of 0.1960, which is greater than 0.05. Therefore, this study shows that Pair 2 had no significant difference for overconfidence before and after the lockdown. The absence of a difference for overconfidence in this study is not in accordance with the findings of
Sitinjak (
2020), who suggested an increase in investor confidence before and after the lockdown. This is consistent with the findings of
Parulian and Syahwildan (
2021), who explained that trading volume activity that assesses overconfidence has no difference before and after the lockdown due to the level of public confidence remaining stagnant.
For Pair 3, the stock price assessment before and after the lockdown had a Sig. (two-tailed) value of 0.0011, which is smaller than 0.05. Therefore, it was concluded that Pair 3 showed a significant difference in stock price movement before and after the lockdown. This is in accordance with the findings of
Anggraini (
2021), who explained that there was a difference between stock price movement before and after the announcement of government policies related to social distancing.
Table 6 presents the descriptive statistics before the COVID-19 lockdown. From the table, it can be seen that the average share price movement was 0.0004, which means that most of the stock price movement data were positive. The movement of the highest share price increase of 0.52 on Day 10, 9 April 2020, was probably due to the fact that when a publicly listed company, namely Barito Pacific Tbk (BRPT), recorded a decrease in profits and a decrease in expenses, it planned to pay off its debt by issuing bonds and decrease the minimum share price movement. The highest share price increase of 1.00 was for Sejahtera Bintang Abadi Textile (SBAT), a textile manufacturing company, on Day 2, 30 March 2020, which was probably because SBAT had just conducted an IPO and still needed time to adjust.
For foreign investment flow, the average foreign investment flow was around 2102.6093 with a maximum value of 623,100 for a publicly listed company known as Arwana Citramulia (ARNA) on Day 10, 9 April 2020. Because ARNA had increased profits for its business sector, it gave investors investment prospects. The minimum value was 0, spread over Days 1–10, because no income for investment by foreign investors was recorded in the data source.
For overconfidence, the average stock turnover rate from overconfident investors was around 0.0027 with a maximum value of 0.31, namely for Klasifikasi Baku Lapangan Usaha Indonesia (KBLI) on Day 10, 9 April 2020. This is because after KBLI obtained a loan from one of the conventional banks, it registered a profit increase, so it could increase investor confidence. The minimum value was 0.10, namely for Toba Pulp Lestari Tbk PT (INRU) on Day 4, 1 April 2020. This occurred because INRU recorded a decline in sales, which was predicted to run until the end of the year, so it was possible that investors lost their confidence in investing in INRU. In determining whether panel data are better estimated using the fixed effect model (FEM) or the common effect model (CEM), the Chow test was performed by testing the following hypotheses:
H1. Common Effect Model (CEM);
H2. Fixed Effect Model (FEM).
If the probability (cross-section F) is smaller than 0.05 (Prob. < 0.05 (a)), H1 is rejected and H2 is accepted. This is also explained by the panel data; it is better to estimate using the fixed effect model (FEM).
Table 7 presents the results of the Chow test and shows that the appropriate Chow test is the CEM due to the probability value of 0.5757, which is greater than 0.05.
In determining whether panel data are better estimated using the random effect model (REM) or the FEM, the Hausman test was carried out by testing the following hypotheses:
H3. Random Effect Model (REM);
H4. Fixed Effect Model (FEM).
If the resulting chi-square is smaller than 0.05, the hypothesis H3 is rejected and H4 is accepted, which can also explain that panel data are better estimated using the FEM.
Table 8 presents the results of the Hausman test. The results of the Hausman test explain that the REM is better than the FEM, as the results show the chi-square probability value of 1.0000, which is greater than 0.05.
In determining whether the panel data are better estimated using the CEM or the REM, the Lagrange test was carried out. If the resulting
p value is greater than 0.05, then the hypothesis Ho is accepted and Ha is rejected.
Table 9 presents the results of the Lagrange test. The results of the Lagrange test explain that the CEM is better than the FEM, as the results show that the
p value is 0.9335, which is greater than 0.05.
The multicollinearity test was used in this study to test the existence of a very strong relationship condition between several or all of the independent variables involved in the regression model. If there is multicollinearity, then there is a bias in the regression coefficient. In panel data regression analysis, the VIF value is seen to determine whether there is multicollinearity. Based on the results of the multicollinearity test for the post-pandemic data, all dependent variables showed a VIF value below 10.0 and a 1/VIF value above 0.1, indicating that the regression model did not experience multicollinearity problems.
Table 10 presents the results of the multicollinearity test.
The heteroscedasticity test was used to test whether the residual variance from one security datum to another observation remains. This test was performed using the Park test, which has the following criteria:
If prob. > 5% (0.05), then there is no symptom of heteroscedasticity;
If prob. < 5% (0.05), then heteroscedasticity occurs.
Table 11 presents the results of the Park test. From the results of the Park test, it can be seen that the probability of each variable is >5% (0.05). Hence, it can be concluded that there were no symptoms of heteroscedasticity.
Table 12 presents the results of the panel data regression before the COVID-19 lockdown. The results show that a constant (a) of −30.43313 is the predicted value of the stock price when the simultaneous or joint effect of foreign investment flow, overconfidence, and the exchange rate is 0 (zero). The regression coefficient for foreign investment flow was −0.0000543and had a negative value. This means that every time there is an increase in the value of foreign investment flow, with the other independent variables assumed to be constant, it is predicted that the stock price will decrease by −0.0000543.
The regression coefficient for overconfidence was −168.3404 and had a negative value. This means that every time there is an increase in the value of overconfidence, with the other independent variables assumed to be constant, it is predicted that the stock price will decrease by −168.3404. The regression coefficient for the exchange rate was 0.003314 and had a positive value. This shows that every time there is an increase in the exchange rate value, with the other independent variables assumed to be constant, it is predicted that the share price will increase by 0.003314. The regression equation formed is as follows:
5.2. Analyses after COVID-19 Lockdown
Table 13 presents the descriptive statistics after the COVID-19 lockdown. From the table, it can be seen that the average share price movement was 0.0003, which means that most of the stock price movement data were positive. The highest movement of the stock price was an increase of 0.13, which was for Semen Indonesia (Persero) Tbk PT (SMGR) on Day 10, 9 April 2020, and was likely achieved because SMGR still recorded a positive sales performance during that period. The highest stock price movement of −0.28, namely for MBTO, was on Day 10, 9 April 2020. This happened in the prediction because during sales growth, MBTO still recorded losses from the previous period.
The average foreign investment flow was around 313,778.3029 with a maximum value of 37,949,900, namely for Ashland Inc. (ASII), Jakarta, Indonesia on Day 5, 2 April 2020. This was probably due to investor interest related to the news about one bank owned by Astra that was acquired by a Thai bank. The minimum value was 0 because no income for investment by foreign investors was recorded in the data source. For overconfidence, the average stock turnover rate from overconfident investors was around 0.0013 with a maximum value of 0.11, namely for Buyung Poetra Sembada Tbk (HOKI) on Day 4, 1 April 2020. This was likely because, during that period, HOKI recorded a positive performance, resulting in an increase in sales and profits that triggered investors to invest. The minimum value was 0, which included 37 out of 72 companies, spread over Days 1–10.
The results of the Chow test in
Table 14 explain that the CEM is better than the FEM, as the results show that the probability value of 0.2622 is greater than 0.05.
The Hausman test was carried out to determine whether panel data are better estimated using the random effect model (REM) or the fixed effect model (FEM). The results are shown in
Table 15. The results of the Hausman test explain that the REM is better than the REM, as the results show that the probability value of the chi-square is 1.0000, which is greater than 0.05.
The Lagrange test was carried out to determine whether the panel data are better estimated using the CEM or the REM.
Table 16 shows that the REM is better than the CEM. This is because the results of the test show that the
p value is 0.0000, which is less than 0.05.
Table 17 shows the results of the multicollinearity test for the post-pandemic data. All dependent variables showed a VIF value below 10.0 and a 1/VIF value above 0.1, indicating that the regression model did not experience multicollinearity problems.
Table 18 shows the results of the heteroscedasticity test. From the results of the Park test, it can be seen that the probability of each variable is >5% (0.05). Hence, it can be concluded that there were no symptoms of heteroscedasticity.
Table 19 provides the results of the panel data regression before the lockdown. The results show a constant (a) of −30.43313 is the predictive value of the stock price when the simultaneous or joint effect of foreign investment flow, overconfidence, and the exchange rate is 0 (zero). The regression coefficient for foreign investment flow was −0.0000543, and had a negative value. This indicates that every time there is an increase in one foreign investment flow value, with the other independent variables assumed to be constant, it is predicted that the stock price will decrease by −0.0000543.
For overconfidence, the regression coefficient was −168.3404 and had a negative value. This indicates that every time there is an increase in one overconfidence value, with the other independent variables assumed to be constant, it is predicted that the share price will decrease by −168.3404. The regression coefficient for the exchange rate was 0.003314 and had a positive value. This shows that every time there is an increase in the exchange rate value, with the other independent variables assumed to be constant, it is predicted that the stock price will increase by 0.003314.
Table 20 provides the results of the panel data regression after the COVID-19 lockdown. The results show that the constant (a) of 0.113061 is the predictive value of the stock price when the simultaneous or joint effect of foreign investment flow, overconfidence, and the exchange rate is 0 (zero). The regression coefficient for foreign investment flow was −0.000000000424 and had a positive value. This means that every time there is an increase in the foreign investment flow value, with the other independent variables assumed to be constant, it is predicted that the stock price will increase by −0.000000000424.
The regression coefficient for overconfidence was −0.229183 and had a negative value. This means that every time there is an increase in the value of overconfidence, with the other independent variables assumed to be constant, it is predicted that the stock price will decrease by 0.139. The regression coefficient for the exchange rate was −0.00000736 and had a negative value. This shows that every time there is an increase in the exchange rate value, with the other independent variables assumed to be constant, it is predicted that the stock price will decrease by −0.00000736.
The results further show that foreign investment flow both before and after lockdown did not have a significant effect and showed a negative value on stock prices (−0.0000528 and −0.000000000424). However, the findings in this study are not consistent with the findings of previous studies such as those by
Kim and Jo (
2019) and
Gupta et al. (
2013) that suggested a positive and significant relationship between foreign investment flows and stock prices. Similarly,
Rujiravanich (
2015) found a positive relationship between foreign flows and stock prices from the stock market in Thailand after the Asian Financial Crisis.
Alawi (
2019) stated that there was no significant relationship between foreign investment flows and stock price movements. This was probably due to disruptions in the exchange rate association and global trading conditions in Saudi Arabia. In this regard,
Rahman (
2020) explained that domestic investors dominated the Indonesian stock market during the pandemic, and foreign investors, according to data from the Ministry of Finance, shifted their investments to safe assets or government bonds. This transfer was carried out because foreign investors had considered the risks that occurred during uncertain times, such as this pandemic.
The regression analysis in this study shows that overconfidence both before and after the lockdown did not have a significant effect and showed a negative value on stock prices (−1.636773 and −0.229183). The effect of overconfidence before and after the lockdown in this study is not consistent with the findings of previous studies such as those by
Gasteren (
2016) and
Machmuddah et al. (
2020) that suggested a positive and significant relationship between overconfidence and stock prices. However, although it has no effect, the tendency of a negative relationship between overconfidence and stock prices in this study is in accordance with previous findings (
Tsai et al. 2022) showing that stock prices and overconfidence had a negative relationship. This is based on the fact that if the stock price is low, there is a high level of overconfidence, and if the stock price rises, there is a low level of overconfidence. Similar findings were found by
Phan et al. (
2020), who stated that there was no evidence of the effect of overconfidence on the stock market in Thailand due to a lack of a solid stock market there and a weak market shape.
In terms of the interaction effect between foreign investment flow and overconfidence on stock price movement before the lockdown, the results showed R2 = 0.00049 before the lockdown and 0.002530 after the lockdown, supporting the earlier results in this study that showed foreign investment flow and overconfidence do not influence stock prices. The calculated F value of foreign investment flow and overconfidence before the lockdown was 0.011168 with a significance of 0.998384 where the calculated f value was smaller than the f table of 2.617, and a significance of 0.998384 is greater than 0.05. Hence, it can be concluded that foreign investment flow and overconfidence had no simultaneous effect on the stock prices before lockdown. The calculated F value of foreign investment flow and overconfidence on the stock price after the lockdown was 0.605304 with a significance of 0.611702 where the calculated f value was smaller than the f table of 2.617, and a significance of 0.611702 is greater than 0.05, so it can be concluded that foreign investment flow and overconfidence had no simultaneous effect on stock prices after the lockdown.