Descriptive statistics explains the distribution of data. There are several components of descriptive statistics, including the mean, median, maximum, minimum and normality test. Normality test can be assessed by the Jarque-Bera under the assumption that the probability value is not significant. It means that the data is normally distributed. A detailed explanation of the descriptive statistics of all study variables is shown in Table 2.

During 2000-2018 the mean of GDP growth for ASEAN countries was about 4.45%. Meanwhile, the median, maximum and minimum are 4.61%, 12.79% and -3.70%, respectively. The condition indicates that the level of economic growth in the ASEAN countries tends to vary and fluctuate, and the distribution of economic growth data is not normal.

The corruption variable shown by the CP) can be categorized into two measurement periods: 2000-2011 and 2012-2018. The descriptive statistics of CPI during 2000-2011 for the mean, median, maximum and minimum were 2.94, 2.12, 9.28 and 1.17, respectively. However, during 2012-2018 the mean, median, maximum, and minimum of CPI were 38.70, 35.29, 85.00 and 20.86, respectively. The condition delivers a signal that most ASEAN countries face a low level of control over corruption.

The institutional indicators consist of six indicators that reflect the quality of governance with a value between -2.5 (weak) to 2.5 (strong). It denotes that a country’s governance is of high quality if the institutional indicator has a higher value that is close to or equal to 2.5. Numerous empirical studies have outlined why institutions in

ASEAN nations typically have poor quality.

The descriptive statistics on institutional indicators reveal that the five institutional indicators have a mean and median negative (VA, PST, RQ, RL and CC) while the GE is positive but at a level of around 0. The condition illustrates that, in general, the quality of institutions in the ASEAN needs to be improved. This institutional quality improvement is expected to be able to encourage the improvement of domestic and regional economic conditions, both in the short and long run. Furthermore, institutional indicators that have a maximum value close to 2.5 covers Government Effectiveness (GE) and Control of Corruption (CC), reaching 2.44 and 2.33, respectively. It could be a positive signal that the governments of ASEAN countries have been able to promote the effectiveness of policies and state institutions for the better. In addition, they try to eliminate and minimize corrupt behaviour.

- Unit Root Test

(Gujarati, 2003) has explained that the stationarity test has been widely used in time series data analysis which is better known as unit root test. The stationarity test commonly employed is the Dickey-Fuller (DF) method. The DF test emphasizes that the error term, ut, is uncorrelated. However, if it is correlated, the stationarity test can utilize the Augmented Dickey-Fuller (ADF) method. Moreover, the DF test assumes that it is independent and normally distributed, while the ADF assumes that there is a possibility of serial correlation. An alternative method is Phillips and Perron (PP). The PP test with a nonparametric statistics approach will solve serial correlation problems.

The panel data stationarity test will emphasize the stationarity test for individual variables both at the level and at the 1st difference. There are three stationarity tests, namely: Im, Pesaran & Shin (IPS), Augmented Dickey-Fuller (ADF)-Fisher, and Phillips-Perron (PP)-Fisher. The results of the stationarity test for all variables are expressed in Table 3.

At the level, both intercept and intercept & trend indicate that GDP growth has been stationary. Another variable that is also stationary at the level is the ICT Development Index (IDI). Meanwhile, total foreign investment (TFI) is only stationary at a level with an intercept & trend under the PP-Fisher method. The stationary results also confirm that all institutional indicators (VA, PST, GE, RQ, RL and CC) are stationary at the level of both the intercept and intercept & trend. In contrast, the corruption perception index (CPI) is not stationary at level.

Furthermore, better results can be found in the 1st difference stationarity test. The results show that all variables are stationary. It means that the variables other than GDP growth already have stationarity at I(1) or are cointegrated. However, this condition still requires further testing under cointegration tests such as Johansen test.

The findings in the stationarity test can serve as a basis for consideration that the panel data estimation and VAR can be estimated. The benefit of stationarity is a robust estimation of both static panel data and Panel-VAR. Therefore, it is possible to generalize the estimation results to explain the study objectives, and the estimation results can provide unbiased estimation parameters.

- Statics Panel Estimation

This study estimates Equations (3a) - (3c) and (4a) - (4c). These equations are estimated using Pooled OLS, Fixed Effects Model (FEM) and Random Effects Model (REM). The three methods are standard techniques in static panel data analysis.Table 4 describes the estimation results of the static panel about the effect of corruption perception index and institutional indicators on economic growth in ASEAN during 2000-2018.

The findings of the Model 1 (Equation 3a until 3c) show that the Breusch-Pagan (BP) Lagrange Multiplier (LM) test is about 14.00 and significant at α of 1%. Further, the Hausman test is about 12.07 and significant at α of 5%. Thus, both FEM and REM can be used to explain the study objectives. The findings on the FEM estimation are divided into two dummy effects, namely cross-section and period (time series). The results of the FEM-cross section indicate that economic growth is determined by both domestic and foreign investment (LTDI and LTFI). LTDI has a negative impact on GDP growth. Besides, the results of the FEM period exhibit that economic growth is impacted by both LTDI and LTFI, whereas the ICT development index (LIDI) has a negative effect on GDP growth. The REM estimation describes that economic growth is driven by labour force (LTLF), LTDI, LTFI, and LIDI. Hence, statistically, the empirical findings of the Solow Growth Model have been expressed in ASEAN countries during the study period. Similarly, the finding is explained by Pooled OLS estimation. An increase in domestic investment will lead to a decline in economic growth. The governments in ASEAN countries should concern on the source of domestic investment. The negative impact of domestic investment on economic growth drives a risk signal that the domestic economy did not stimulate by local investors. In contrast, the findings of Model 1 are not able to provide significant results on the relationship between CPI and economic growth (GDPG) in ASEAN countries during the study period.

Model 2 (Equation 4a until 4c) provides the results of the effect of institutional indicators on economic growth. The BP LM test and Hausman test indicate values were about 24.60 and 32.58, respectively. Both values/tests are significant at α of 1%. It means that FEM or REM can express the second objective. The FEM-cross section finds that labour force (LTLF), LTDI, and LTFI have a significant effect on GDP growth. Economic growth will slow down as domestic investment, and the number of workers both rise, whereas institutional indicators have little bearing on economic expansion. In contrast, the FEM period informs that LTLF), LTDI, LTFI, LIDI, and institutional indicators (control of corruption, CC) have a significant effect on GDP growth. LTDI and LIDI have a negative effect, while LTLF, LTFI and CC have a positive effect on GDP growth. Then, REM provides the finding that GDP growth is determined by the labour force, domestic and foreign investment, and some institutional indicators. Voice and Accountability (VA) and Corruption Control (CC) are two institutional indicators that can accelerate ASEAN’s economic growth. Indeed, REM affirms that during the study period, institutional indicators significantly influenced ASEAN’s economic growth.

In general, the adjusted R-square provides a value range between 39%-67% in the estimation results of Models 1 and 2. It explains there are other variables that are considered explanatory variables to determine ASEAN’s economic growth. Furthermore, the F-statistics for all estimation models informs that simultaneously all independent variables have a significant effect on GDP growth during the study period.

(Dzhumashev, 2014) noted some critical issues regarding c o r r u pti o n - e c o n o mi c growth nexus. The issues cover there is no clear relationship between corruption level and economic growth by incorporating an institutional quality, the government size does not reflect the ability of government policy to restrain corruption level and stimulate economic growth, and the corruption level is connected to the level of economic development. (Malanski & Póvoa, 2021) added that the different levels of economic freedom provide a significant impact on corruption-growth nexus.

Some empirical studies found that corruption has a significant and positive effect on economic growth. For example, (Ahmad et al., 2012) analyzed the impact of corruption and institution on economic growth during 1984-2009 in 71 developed and developing countries. The findings demonstrate that while corruption and institutions have a significant negative impact on economic growth, capital has a significant and positive impact. The correlation between corruption and economic growth is inverted U-shaped. Furthermore, Huang (2015) estimated the effect of corruption on economic growth during 1997-2013 in thirteen Asia-Pacific countries. The results explain that corruption has a significant and positive impact on economic growth in South Korea, while in China, economic growth has a significant and positive impact on corruption. However, in the eleven countries of the Asia-Pacific region, there is no significant correlation between corruption and economic growth. Directly, corruption has a significant and positive impact on economic growth (Cieslik & Goczek, 2018). But on the other hand, economic growth generated through investment is also significantly and adversely impacted by corruption. It demonstrates how investment, particularly foreign investment, can be used to control corruption. Economic growth is significantly and favourably impacted by political indicators as well.

(Gründler & Potrafke, 2019) show that corruption has a significant and negative effect on per capita GDP. The results inform that increasing corrupt practices will undermine the domestic economy. Similarly, the finding made by (Saha & Ben Ali, 2017) that a non-linear estimation model can express the significant impress of per capita GDP and institutions on corruption. The non-linear estimation results show a negative effect on corruption. Furthermore, the estimated effect of corruption can also be analyzed from the threshold effect. (Alfada, 2019) found that provinces in Indonesia have a level of corruption below the threshold effect of 1,765 points will be able to promote progressive economic growth, while the level of corruption above the threshold effect will hamper the local economy.

- Panel-VAR Estimation

Vector Auto Regression (VAR) is a-theoretic estimation method. It is employed to explain the causality between corruption and institutional indicators on GDP growth in ASEAN countries during 2000-2018. There are some steps to utilize VAR. The first step is to test the stationarity of data and determine the lag length. The results of the stationarity test are described in Table 3. Meanwhile, the results of the lag length confirm that the Akaike Information Criterion (AIC) score is about 4.45 on lag 1. The Hannan-Quinn information criterion (HQ) and Final Prediction Error (FPE) also demonstrate that lag 1 is the optimal lag. Thus, the lag length in the VAR estimation is lag 1.

The empirical findings with GDP growth (GDPG) as the dependent variable inform that only two institutional indicators have a significant effect, namely: Regulatory Quality (RQ) and Control of Corruption (CC). It expresses those institutions play an important role in economic growth and corruption perception index (CPI) does not have a significant impact.

Similarly, when the CPI and institutional indicators become the dependent variables. The empirical findings show that these variables do not have a significant relationship with economic growth in ASEAN countries.

Hence, the hypothesis clarifies that corruption has a relationship with economic growth is rejected. The governments of ASEAN countries should strengthen efforts to eradicate the level of corruption and improve the quality of institutions in order to increase economic growth significantly. In particular, (Nairobi, 2021) is concerned with the impact of corruption on economic growth for 16 provinces in Indonesia during 2014-2018. The findings reveal that the higher corruption level leads the higher economic growth. In the case of Pakistan, the corruption level impedes economic growth during 1987-2009 (, 2013). Besides, (Spyromitros & Panagiotidis, 2022) found that in developing countries, the level of corruption stifles economic growth. It implies that corruption has a detrimental effect on economic expansion. On the other hand, corruption has a beneficial effect on economic growth in Latin American nations.

Hereafter, the response between economic growth, CPI and institutional indicators takes several years to have a significant response. In general, the response period for these variables is more than six years, with a total study period of 19 years. In contrast, only the rule of law (RL) has a response of less than 4 years. This indicates that law enforcement efforts can be carried out quickly compared to other institutional indicators.

The empirical findings of the BP LM test and Hausman test in Model 1 confirm that FEM or REM is the best model because the value of the two tests is 15.095 and 26.171, respectively, which are significant at α of 1%. The FEM estimation can be categorized into FEM-cross section and FEM-period. It is utilized to analyze precisely the effects of the dummy-cross section and dummy-time series. A detailed explanation of the estimation results can be seen in Table 6.

Model 1 provides an indication that during 2000-2011 the CPI did not have a significant contribution to economic growth in ASEAN countries. However, domestic investment, foreign investment and the ICT development index contributed significantly to GDP growth. Thus, robustness checking for the sample during 2000- 2011 cannot clarify the significant contribution of CPI to economic growth.

Furthermore, the BP LM and Hausman tests in Model 2 have values are 27.56 and 35.99, respectively, which are significant at α of 1%. It suggests that both FEM and REM estimations can be utilized to describe the second objective. Empirical findings show that the FEM-cross section found a significant impact of institutional indicators (political stability) on GDP growth. In addition, domestic and foreign investment contributed significantly to economic growth. The FEM-period results inform the significant contribution of domestic investment, foreign investment and the ICT development index to economic growth. The REM results can prove that labour force, domestic investment and foreign investment contribute significantly to economic growth.

The empirical findings explain that the BP LM test and Hausman test on Model 1 are 15.45 and 15.14, respectively, which are significant at α of 1%. Hence, both FEM and REM estimations can explain the first objectives. Further, the results of the BP LM test and Hausman test on Model 2 were 6.79, which is significant at α of 5% and 37.68, which is significant at α of 1%. It means that the FEM and REM estimations can also describe the second objective.Table 7 provides a detailed explanation of statics panel results.

The findings of Model 1 and Model 2 confirm that the CPI did not significantly contribute to GDP growth during 2012-2018. The robustness checking clarifies the hypothesis of the significant impact of corruption on economic growth is rejected. In contrast, the hypothesis of the significant effect of institutional indicators on economic growth is accepted. The institutional indicators cover political stability, government effectiveness, regulatory quality and rule of law.

Statistically, the results in Table 7 are more significant than in Table 6. It confirms that the differences in the sample period will express significantly different findings. For example, the pooled OLS, FEM-period and REM results found the effects of labour, investment and technology on economic growth. In addition, Pooled OLS and FEM-period have found that three institutional indicators contribute significantly to economic growth. The three indicators are Political Stability (PST), Government Effectiveness (GE) and Rule of Law (RL). In contrast, REM estimation informs there were only two significant institutional indicators, namely: PST and RL.

(Afonso & Sá Fortes Leitão Rodrigues, 2021) argued that corruption delivered a negative impact on economic growth for 48 countries during 2012-2019. Besides, the private investment can stimulate the corrupt practices. It means that the private sector can lead the higher bribery practices under uncertainty time to sustain their business (Afzali et al., 2021). The corrupt practices also come from the quality of institutions in the government system (Pulido et al., 2020). The lower quality of institutions produces the higher corrupt practices. (Saha & Ben Ali, 2017) found that the quality of economic freedom and economic growth can reduce the level of corrupt practices in Middle Eastern and North Africa (MENA) during 1984-2013. Moreover, (Ben Ali & Gasmi, 2017) found that a higher quality of information and communication technology (ICT) diffusion and institutions creates a lower level of corruption.