Can Energy Intensity Impede the CO2 Emissions in Indonesia? LMDI-Decomposition Index and ARDL: Comparison between Indonesia and ASEAN Countries

In several ways, the AEC has increased connectivity between the businesses, merging business activities, and funneling them to end customers. Moreover, it increased energy consumption and increased CO2 emissions in ASEAN countries. This study analyzed the driving factors of carbon emissions in ASEAN and identified the differences between member countries based on decomposing the extended Kaya identity via the logarithmic mean divisia index (LMDI) method. Since the energy intensity effect “EI-effect,” gross domestic effect “GDP-effect,” population effect “POP-effect” and CO2 emission effect “CO2-effect” were a mixture of I(0) and I(1), Johansen cointegration test cannot be applied. Hence the study deployed an autoregressive distributed lag (ARDL). This study’s ARDL model captured a long-run and short-run relation of the whole cointegrated variables in ASEAN countries. Based on a panel of cross-country and time-series observations, the study analyses that the ARDL model was used to cover a model of short-and long-run implications. Based on the result, we identified the root cause of significantly increasing CO2 emission in the past 36 years. This study’s result was that a positive long-run relationship interacted with a mostly negative short-run relationship between the energy intensity’ EI-effect,’ gross domestic effect’ GDP-effect,’ population effect’ POP-effect “and CO2 emission effect‘ CO2-effect.”


INTRODUCTION
Three scenarios were used for the projected demand and energy supply period in Indonesia, such as Business As Usual (BaU), the Continuous Development Scenario (PB), and the Low CARBON Scenario (RK) (Suharyati et al., 2019). The same basic gross domestic income (GDP) growth assumption is used in these three scenarios, with average GDP growth of 5.6% per year and population growth of 0.7% for the same population. Kebijakan Energy National (KEN) targeted a renewable energy mix in the primary energy mix at least 23% in 2025 and minimizes the petroleum use by <25% in 2025. Besides, energy efficiency was also targeted down 1% per year to encourage energy consumption savings in all sectors. The projected demand for the national final energy scenario BaU, PB, and RK will increase with the average growth per year. Industrial and transportation sectors will still dominate the final energy demand until 2050 as conditions in the year 2018. In the year 2050, the industrial sector will dominate more than the other sectors. Meantime the total projected Indonesia CO 2 emissions in 2030 would increase to 912 million tones CO 2 eq (BaU), 813 million tons CO 2 eq (PB), 667 million tones CO 2 eq (RK). Also, per capita emission indicators showed an increase of 1.7 tones of CO 2 /capita in 2018-6.4 tons of CO 2 /capita (BaU), 5.3 tons of CO 2 /capita (PB), 3.3 tones of This Journal is licensed under a Creative Commons Attribution 4.0 International License CO 2 /capita (RK) in 2050, in line with increasing emissions and population growth.
In 2016, the ASEAN Economic Community ("AEC") emerged. In 2018, with a total GDP of USD 3.0 trillion, AEC is actually the world's fifth largest economy (The ASEAN Secretariat, 2019). Indonesia was destined to play an important role. Within the AEC role, Indonesia has a crucial market. Indonesia will be all-eyed by undeniable trends like urbanization and consumerism. Indonesia is one of 10 ASEAN members and the AEC, i.e., Brunei, Cambodia, Indonesia, Malaysia, Myanmar, Singapore, Thailand, Laos, the Philippines, and Vietnam. Since then, a few Multilateral Free Trade Agreements have been ratified by Indonesia and the AEC, leading to a reduction in market barriers for neighboring countries. Together with six free trade agreement partners: India, China, Japan, Korea, Australia, and New Zealand, the 'Regional Comprehensive Economic Partnership ("RCEP") was formed. In many areas such as production and the supply chain, linking business activities and channeling to end customers, the AEC and RCEP have increased cooperation between the various business sectors. It will ultimately increase the consumption of energy and increase CO 2 emissions.
Many factors affect the relationship between energy and economic development. Detailed indicators between energy use and activity were necessary to give a reasonable interpretation of the aggregated indicators for the country as a whole. Increased activity and economic growth were the most important explanations of total energy use in the economy. This study was to analyse the driving factors of carbon emissions in ASEAN and identify the differences between member countries based on decomposing the extended Kaya identity via the Logarithmic Mean Divisia Index (LMDI) method during 1971-2017. It was to respond to the question on How has carbon abatement in ASEAN over the past 36 years? What will happen if the trend was going to continue? After assessing the decomposition values, we launch the measurement model's robustness through granger causality and vector error correction model (VECM). Furthermore, the performance evaluation model was also discussed. Based on the result, we identify the root cause of significantly increasing CO2 emission in the past 36 years.

LITERATURE REVIEW
Decomposition analysis is performed to decompose increases in CO 2 emissions into many predefined variables. Laspeyres index was used in the early 1990s. The Laspeyres index and the Divisia index are the overall index decomposition analysis (IDA) to date. The Laspeyres index calculates the percentage change over time in certain aspects of a category of items, using values-based weights for a particular base year. The Laspeyres index method results are measured in the same way as LMDI, but with a percentage change from the base year to year t. However, the Divisia index is a weighted sum of the logarithmic growth rates where the weights are the share of the total value of the components. IDA is more widely accepted as a decomposition method because of the simplicity of adoption, ease of use, and relatively low data requirement. In terms of their benefits and drawbacks, Ang (2015) summarizes the IDA processes then advocated, for general use, the logarithmic mean divisia index (LMDI). The IEA is pioneering LMDI, and then most of the researchers follow after. The addictive LMDI decomposition and extended KAYA identity are used in this paper to capture the various effects of energy consumption shifts. Ma and Stern, 2008;Zhang (2019) use LMDI as decomposition analysis in their studies coupled with the expanded Kaya identity. Integrating Kaya identity and LMDI for decomposition into total energy-related CO 2 also have been performed in the construction sector by Ma and Cai (2018). The research about the decoupling study between economic growth and CO 2 emissions to measure critical determinants of or energy use emerged when the OECD put it as its environmental strategy. Additionally, research such as Kojima andBacon (2009) andde Freitas andKaneko (2011) shows that popularity increases due to the combination of index decomposition. Some researchers have combined decomposition or decoupling analysis and LMDI methods with econometric methods like practical VECM (Jiang and Liu, 2014;Moutinho et al., 2015;Zhang and Su, 2016). Despite national reach, others decide to study in the different sectoral industry (Zhao et al., 2016). Toba and Seck (2016) clarified how different decomposition factors are interlinked. They thought that incorporating elements of the energy systems that contribute to the climate and community would promote energy policy. Meantime, Zhang and Su (2016) select ten indicators of rural household energy consumption, then placed all at four-dimension share factors: Social, economic, technological, and environmental. This paper used the same methodology as Pui and Othman (2019a). Pui and Othman (2019b) explored the economic, technological, and social aspects of the aggregate decomposition process.
The purpose is to gauge the relative strength of these four effects on emission changes. The paper decomposes four factors into four effects considering all the POP-effects, GDP-effects, EIeffects, and CO 2 -effects using the LMDI approach. Firstly, we analyze the change in energy consumption in four effects. Then we deployed the VECM to investigate the causality between POP-effects, GDP-effects, EI-effects, and CO 2 -effects relating to CO2 emissions growth in ASEAN countries, including Indonesia. It is used The pooled mean group estimator and causality analysis.

METHODOLOGY/MATERIALS
This study employed the addictive LMDI decomposition method. Under LMDI, one factor can be decomposed into different elements, and LMDI can measure the influence of those factors over one factor. In This study, authors can compose CO2 into POP-effect, GDP-effect, EI-effect and CO 2 -effect factors for Indonesia and ASEAN countries from 1971 to 2016. The Data of CO 2 gas emission, GDP, population, Primary energy consumption was taken from IEA. To capture the different effects of energy consumption changes, the decomposition addictive LMDI model was used to get four aspects: population effect, GDP effect, energy intensity effect, and CO 2 effect. Decomposition Effect equations as follows:   was CO2 effect After applying the LMDI KAYA analysis, the next step was used data panel analysis where the combination of time series and crosssection data. By accommodating in both the model of information related to cross-section variables and time series, the data panel was substantially able to reduce omitted variables' problem, the model that ignores the relevant variables. The long-run model panel data regression model in the study was as follows: We set the basis for understanding the contradicting effects of energy intensity on population, GDP and CO 2 intensity by concentrating on effects at varying time horizons. The findings are analyzed using the ARDL. We connect our short-and long-run effects to the notable predictive framework on the effects of energy intensity (Cansino et al., 2019). Our econometric method emphasizes us to estimate short-run effects relevant to the region. The framework can also be defined as a panel error-correction model (ECM), where short-and long-run effects from a panel ARDL model are mutually measured. When the data was strictly I(0) or purely I(1) or a mixture of both but not I(2), the ARDL model was sufficient. The entry of I(2) variables in the analysis should be avoided since the ARDL model only provides critical boundary values for the I(0) and I(1) series. Therefore, this research conducts ADF and PP tests to determine the order in which targeted variables are integrated. These two tests in econometric literature have been widely used. The results of both root unit tests have been included in Table 1.
All the variables were checked by both the unit root checks I(1).
By reformulating Eq.(2) above as an ARDL(p, q,…, q) model. ARDL model as forecasting model for energy intensity effect "EI-effect," gross domestic effect "GDP-effect," population effect "POP-effect" and CO2 emission effect "CO2-effect," can be written as follows: . 1 1 1 (4) In order to comply with the requirements, we embed a VECM into an ARDL (p, q) model. VECM was a model used to analyze multivariate time series data that was not a stationer. In other words, the VECM model was a VAR Model that has a linear cointegration relationship, which can be written: The α and β parameters have a dimension N x R, where N was the number of coefficients, and R was the cointegration). The degree of cointegration indicates several long-term relationships between the Yt and the model that we make, so that cointegration can be said was the main requirement of using VECM.

Data
There were 3128 total data observations on the original data among nine ASEAN countries taken from World Development Indicators (World Bank) and the International Energy Association (IEA). From 1971 to 2017, except for Brunei Darussalam. In Table 1, the descriptive statistical test results on CO2, EI-effects, GDP-effects, POP-effects, and CO 2 -effects values show a mean average with the data distribution having a maximum value, minimum value, and standard deviations for each decomposition variable.

Decomposition Analysis
As explained in the previous paragraph, to determine each emission reduction factor's significance, this study used the KAYA identity to decompose the CO 2 component into the population effect, GDP effect, energy intensity effect, and CO 2 intensity effect. The sum of all four of these factors was equal to that of CO 2 . The principal driving force of CO 2 emissions was the Kaya identity.

ASEAN energy situation
The energy intensity components first weakening happened throughout 1985 until 1990 due to the decline of oil price. Oil price was drastically declining in September 1985 from USD 69.97 per barrel to only USD 31.11 per barrel in February 1986. Indonesia's GDP growth has also decreased by about 2.1%, 7.3%, and 7. 8% in 1985, 1986, and 1987, respectively. The second decline occurred in the Asian crisis year-round 1997 up to 1998.

Population effect
Another factor that aggravates the increase in CO 2 emissions was the population effect, characterized by urbanization (Zhao et al., 2016). Based on the Figure 1, for almost 46 years from 1971 to 2017, Brunei's CO 2 emissions were generated solely based on the population effect. Unfortunately, if the decomposition was based on the proportion of the population over CO 2 , Malaysia, Singapore, and Indonesia were ranked 1 st , 2 nd , and 3 rd of the highest population effect than Brunei Darussalam last decade, Table 2. Urbanization was the correct indication of the outcome of decomposition. The majority of factors have contributed to CO 2 emissions due to the energy increases being consumed by households. Fortunately, over the last decade, the outcome shows that Brunei's GDP impact was taking off due to Brunei's government enhancing the private sector's growth to diversify outside the hydrocarbon economy. The completed decomposition on a yearly based for ASEAN countries can be seen in Appendix Table 1.

GDP effect
The results showed that the GDP effect was the most influential factor in the annual increase in CO 2 emissions (Zhao et al., 2016), followed by Indonesia's population impact, as shown in Figure 1 and most ASEAN countries. This study found that Malaysia was the most crucial GDP effect contributing to CO 2 emissions, based on the percentage of the GDP effect over CO 2 emissions in the last decade followed by Singapore and Thailand. Overall, over the 1971-2017 study period for the ASEAN countries, the GDP effect caused CO 2 emissions to increase by 2514.18 million tons. The effect of GDP, characterized by the share of GDP production, was in line with existing literature (Mitić et al., 2017). The completed decomposition on a yearly based for ASEAN countries can be seen in Appendix Table 2.

Energy intensity effect
Most of the energy intensity effect was due to the decrease in total CO 2 emissions. By improving the technical aspect intensity (Cansino et al., 2019;Mitić et al., 2017), energy intensity in most ASEAN countries has hampered their CO 2 emissions. Singapore was the only country able to tackle CO 2 emissions through energy efficiency from 1971 to 2017, Figure 1. The cornerstone of regulating rising CO 2 emissions has been energy efficiency. The energy intensity effect was linked to a decrease in CO 2 emissions over the period. Based on the percentage of the energy intensity effect on CO 2 emissions over the last decade, this study found that Malaysia, Brunei, and Indonesia were the ASEAN countries' champions, as can be seen from the Table 1. The completed decomposition on a yearly based for ASEAN countries can be seen in Appendix Table 3.

Carbon intensity effect
Carbon intensity was the emission rate of a given CO 2 relative to the primary energy consumption intensity (Mitić et al., 2017;Zhao et al., 2016). Only Singapore benefited from the carbon intensity effect on CO 2 emissions and impeded CO 2 emissions for almost 46 years from 1971 to 2017, as shown in Figure 1. This study found that, based on the percentage effect of carbon intensity   on CO 2 emissions over the last decade, CO 2 emissions have also been reduced in Malaysia, followed by Singapore and Indonesia, Table 2. The completed decomposition on a yearly based for ASEAN countries can be seen in Appendix Table 4.

Indonesia energy situation
Below is the Indonesia's CO 2 LMDI decomposition analysis: Indonesia's energy began when the rise in oil prices in the 1970s led to a windfall in Indonesia's export revenue. Exports contributed to high GDP rates, averaging more than 7% between 1968 and 1981, but then, due to falling oil prices, growth slowed to an average of 4.5% per year between 1981 and 1988. At the end of the 1980s, economic reforms took place, including the rupee's managed devaluation to improve exports' competitiveness and the deregulation of the financial sector. Foreign investment flowed to Indonesia, especially to the export-oriented manufacturing sector, and Indonesian GDP accounted for more than 7% on average from 1989 to 1997. In 1998, real GDP contracted 13.1%, and the economy reached its low point with real GDP growth of 0.8% in mid-1999. Indonesia's real GDP growth reached 6% in 2012, decreasing steadily to 5.1% in 2004 and 5.6% in 2005. After Joko Widodo succeeded Susilo Bambang Yudhoyono, the government avoided foreign direct investment control to improve the economy.
In 2016, Indonesia managed to increase its GDP growth by slightly above 5%-17. Indonesia's demand for energy reflects the size of the country's economy; Indonesia's consumption of primary energy has also increased rapidly, with an annual average growth rate of 5.157% during 1971-2017. The total supply of primary energy was more than 10,462.6 PJ. Since 1971-2017, Indonesia has experienced robust emissions growth of around 20,48 metric tons per year, driven only slightly by strong economic growth and moderately improved energy intensity. For the IEA, total CO 2 emissions from 1971 to 2017 in Indonesia were recorded at 941,40 metric tons for 46 years from 1971 to 2017. Indonesia has pledged to reduce by 29.41% its emission intensity by 2030. Due to the vigorous and required economic activity, Indonesia's emissions may continue to grow strongly in the next decade.
For Indonesian leaders, maintaining a stable and sufficient energy supply in Indonesia has become extremely challenging. Based on Figure 2, the primary energy supply folds more than 49 times. GDP components and components of population growth have been responsible for determining increased energy demand. Just in Figure 1 and the map. 1 Through decomposition analysis, the GDP components clearly show that they play a more critical role in promoting the growth of energy demand and energy intensity components to play their role in soaking energy demand over the year 1971-2017. Only in the Asian crisis, which started in 1997-1998, did GDP and energy-intensity components decrease. The Decomposition for ASEAN countries can be seen in Appenidx Figure 1.
This study currently proposes that it would be a good time for Indonesia and ASEAN countries to deploy energy renewable energy sources faster. For Indonesia, the Government must promote the introduction of the Energy Efficiency Saving Industry development initiatives in Indonesia (Nasip and Sudarmaji, 2018).
To mobilize alternative funding through retrofitting programs in Indonesia, the ESCO can unlock the possibilities and benefits of Energy Efficiency Saving Industry (Sudarmaji and Ardianto, 2020). The use of the national nudge program is another way for the government to make the energy efficiency program efficient. "Nudges" framings are acceptable not only for particular manufacturing sectors but also in many other fields. According to (Sudarmaji and Thalib, 2020), the impetus for reducing electricity use in rural areas has been impacted by social norms and curtailment by Nudge framing. Architectural solutions for energy conservation were included in the definition of "Nudge." In many nations, this notion is commonly used. The definition of "Nudge" can be used to frame rewards for other fields of industry. The 'Nudge' principle could result from decreased CO2 emissions   (Krstic and Krstic, 2015).

The Pooled Mean Group Estimator
The dependent variables in this study were energy Intensity, and there were three free variables, namely population-effect (X1), GDP-effect (X2), and CO 2 -effect (X3). The empirical framework of the analysis has the following components: 1. Panel unit root tests 2. Panel optimal lags selection 3. Panel cointegration tests 4. Panel VEC model estimations 5. Panel causality analysis tests 6. Innovative accounting approach

Unit root and cointegration
Unit root and cointegration factors can be seen from the Table 3.
The test results of Table 3 show that overall, variable EI-effect, GDP-effect, and CO2-effect indicate stationary at level. POPeffect indicates stationary at 1 st differences.
Based on the LR, FPE, AIC, SC, and HQ, obtaining the optimal lag length was two, Table 4. The authors select the max second lags for deploying the panel VEC model.
The integration study findings were summarized in Table 5. Pedroni's cointegration probability approach was a test based on HQIC with a max lag of 9 provided proof at a significance level of 0.05 to reject the null hypothesis for panel rho, panel PP and panel ADF.

Causality analysis result
The short-term causality model using VEC Granger causality/ block Exogeneity Wald tests and pairwise granger causality tests for robustness tests was estimated in this research. Statistically, based on Table 6, there was no short-run granger causality for EI-effect, GDP-effect, POP-effect, and CO 2 -effect individually and jointly in the first model. In the second model, economic aspects have the causality of short-run granger at 0.01 level of 0.01 and economic aspects with the causality of short-run granger at 0.05 level of technical aspects. At 0.10 level, both jointly have short-run granger causality. In the last model, all economic and technological aspects have short-term granger causality at 0.10 and 0.05 stages, both individually and jointly, with social aspects.
Based on Table 7, statistically, on the Pair-wise Granger Causality Tests, there was uni-direct granger causality between EI-effect, GDP-effect, POP-effect, and CO2-effect at 0.01 level.

Robustness check (ARDL)
In Table 6, respectively, long-term and short-term results were published. The long-term results show that POP-effect, and CO 2 -effect has a significant positive effect on EI-effect instead GDP-effect has negative effect. Table 8 also shows that the four models' approximate results show that the ECT coefficient are almost negative, −0.878, −0,712, −0,468 and −0,060 with longterm statistical causality. It has been shown that the long-term balance of EI-effect, GDP-effect, POP-effect, and CO 2 -effect is valid significant with 0.01% and 0.05%. It means that the previous period's imbalance shocks reconnected into a long-run equilibrium. In other words, there is a long-term causality between EI-effect, GDP-effect, POP-effect, and CO 2 -effect.

CONCLUSION
This study has broken down the driving factors for CO 2 emissions in Indonesia and the ASEAN countries on an aggregate basis. This study found that the rise in CO 2 emissions was mainly due to GDP or economic growth, (Saunders, 2015) accompanied by population expansion (urbanization). This study aims to improve energy intensity, particularly in the sense of another strategy to boost energy efficiency in the economic field, as effective emission control strategies. Energy intensity should not be confused with energy efficiency. Energy efficiency involves the use of technology to perform the same function, requiring less energy. More efficient use of energy at all stages of the supply/demand chain could reduce the negative impacts of energy usage while still enabling much economic growth. Improved energy efficiency at the national level means a decrease in fuel imports, thus decreasing foreign exchange pressures and increasing the availability of scarce energy resources to be used. Thus, it will enable increased energy-dependent behaviors to lead to the population's economic well-being as a whole. Increased energy usage also benefits society as a whole, mainly by decreased negative environmental impacts of energy consumption. Energy efficiency refers to the activity or service that can be generated with a given quantity of energy. Further analysis can also consider whether the low-carbon economic goal of Indonesia was technically competent. Our findings also illustrate how powerful the energy intensity was and could be a key component and driving force of economic growth in ASEAN countries. Whether future economic development can be restricted to climate-based policies leads to trade-offs.
The study analyses that the ARDL model was used to cover a model with short-and long-run consequences based on a panel of cross-country and time-series observations. Based on the outcome, the root cause of dramatically rising CO2 emissions over the past 36 years has been established. This analysis's outcome was that a positive long-run relationship interacted with a mainly negative short-run relationship between the "EI-effect," gross domestic effect "GDP-effect," population effect "POP-effect" and CO 2 emission effect "CO 2 -effect" of energy intensity. Authors focus on the groups instead of individual analysis, which means that the authors realized that information is lost by taking a panel perspective. However, using panel data rather than time series can increase the total number of observations and their variations and reduce the noise coming from the individual time series. Heteroscedasticity does not become an issue. The panel data also best suited from developing countries due to short periods for a variable were rampant, often sufficient for fitting time-series regressions. Meanwhile, heterogeneity (differences) among units in the panel, but the special panel data techniques can take this heterogeneity into account by allowing for subject-specific variables. The panel data also suits for studying dynamic changes due to repeated cross-section sectionals observations.

APPENDIX 1
First decomposition factors