Shadow Economy and the Ecological Footprint Nexus: The Implication of Foreign Direct Investment in ASEAN Countries

Abstract

This study examines the influence of economic growth, energy consumption, a shadow economy, and foreign direct investment (FDI) on the ecological footprint in ASEAN countries. The analysis covers a panel of nine member states—Brunei, Cambodia, Indonesia, Lao PDR, Malaysia, the Philippines, Singapore, Thailand, and Vietnam—over the period from 1993 to 2017 due to data availability. To ensure robustness, various panel econometric techniques were employed, including cross-sectional dependence, panel unit root, and cointegration tests, as well as estimation methods such as Driscoll–Kraay standard errors, feasible generalized least squares (FGLS), and panel-corrected standard errors (PCSE). The results do not support an inverted U-shaped Environmental Kuznets Curve (EKC) between economic growth and ecological footprint in the ASEAN countries. Moreover, the findings consistently show that energy consumption, the size of the shadow economy, and FDI exert a statistically significant and positive impact on the ecological footprint towards the Driscoll–Kraay standard errors, FGLSs, and PCSE estimators. For policy recommendations, a country’s pursuit of economic growth should be aligned with a higher degree of environmental sustainability by strategically reducing energy consumption, curbing the shadow economy, and managing foreign direct investment responsibly.

1. Introduction

More than one-third of the Gross Domestic Product (GDP) was from the shadow economy (also known as informal, unofficial, underground, hidden, gray, irregular, parallel, or black economy), especially in developing countries, and as a result, the shadow economy created more than 70% of their employment (Elgin et al., 2021; Ohnsorge & Yu, 2022; Schneider & Enste, 2000). In some developing countries, this number is far higher (Elgin et al., 2021; Ohnsorge & Yu, 2022; Schneider & Enste, 2000). The growth of the shadow economy frequently has unfavorable effects on development and macroeconomic conditions (Enste, 2015; Medina & Schneider, 2019; Ohnsorge & Yu, 2022; Schneider & Enste, 2000), making it one of the biggest issues in developing countries (Dada et al., 2021; Krstić & Schneider, 2015). In fact, the shadow economy, which is almost universal, consists of unreported economic activity that falls within the jurisdiction of national authorities, which is different from the dual economy of official record under the national accounting systems (NASs) (A. Baloch et al., 2022; Ngoc, 2020; Schneider & Enste, 2000). Along with this, hidden economic activities from official authorities of the shadow economy are varied reasons from monetary, regulatory, and institutional perspectives (Medina & Schneider, 2019). Additionally, the larger size of the shadow economy, on the one hand, attracts foreign investment by taking advantage of tax evasion opportunities in host countries (Ali & Bohara, 2017; Cuong et al., 2021). Conversely, the shadow economy obstructs foreign investment by means of institutional enhancement (Bayar et al., 2020; Huynh et al., 2020; Nikopour et al., 2009).

On top of that, one of the major sources of environmental deterioration along with economic growth has been recognized by either the shadow economy (Dada et al., 2021; Eren et al., 2022; Imamoglu, 2018) or foreign direct investment (FDI) (Bhujabal et al., 2021; Solarin & Al-Mulali, 2018; Zafar et al., 2019). Environmental quality heavily depends on the intensity of government regulations and the implementation of environmental standards (Imamoglu, 2018); however, informal economic activities hamper these legislations and standard enforcements. Therefore, a large shadow economy poses a severe challenge to implementing environmental regulations, significantly impacting environmental quality (Baksi & Bose, 2010; Imamoglu, 2018). Similarly, inflows of FDI—a substantial capital inflow from foreign investors—may exacerbate pollution in host countries, as posited by the “pollution haven hypothesis” (Balsalobre-Lorente et al., 2019), whereby firms relocate to jurisdictions with weaker environmental oversight to minimize compliance costs.

Since the Industrial Revolution, the global challenge has been to cope with apparent social and environmental deteriorations. The sustainable development from the World Commission on Environment and Development (WCED) was a significant turning point in the late 1980s (International Institute for Sustainable Development, 2012; Strange & Bayley, 2008; UN, 1987). Thus, it is the dynamic momentum of change. The world is trending toward sustainable development with the universal 2030 agenda of Sustainable Development Goals (SDGs) (UNDP, 2022). Along with this, each country progresses its economy while improving human quality and conserving the environment. After that, various bodies of knowledge greatly expanded the study of environmental issues (Sweileh, 2020; Waas et al., 2011). Moreover, for the study of environmental quality, the ecological footprint by Wackernagel and Rees (1996) has been gaining interest in the current decade, which is replacing carbon dioxide (CO₂) emission (Katircioglu et al., 2018; Solarin & Bello, 2018; Ulucak & Bilgili, 2018; Zafar et al., 2019). The ecological footprint serves as a holistic indicator of environmental degradation resulting from human activities, accounting for both direct and indirect impacts on natural ecosystems (Ulucak & Bilgili, 2018). Also, the CO₂ emission is a part of the ecological footprint (Destek & Sarkodie, 2019). Thus, a framework Environmental Kuznets Curve (EKC) hypothesis has been widely concerned in these decades in the pollution research (Ahmed & Wang, 2019; Altıntaş & Kassouri, 2020; Aşıcı & Acar, 2018; Dogan et al., 2020; Kongbuamai et al., 2020a; Sharif et al., 2020; Wang et al., 2020b). However, several studies on the shadow economy and CO₂ nexus have been conducted (A. Baloch et al., 2022; Camara, 2022; Canh et al., 2019; Dada et al., 2021). Still, the influence of the shadow economy on the ecological footprint under the EKC hypothesis remains insufficiently understood and presents notable limitations in existing research.

Among the developing countries, the Association of Southeast Asian Nations (ASEAN), founded in 1967 on the Southeast Asian peninsula, has grown to become one of the most influential economic blocs in the developing world. Its member states include Brunei Darussalam, Cambodia, Indonesia, Lao PDR, Malaysia, Myanmar, the Philippines, Singapore, Thailand, and Vietnam (ASEAN Secretariat, 2014). In 2021, over 650 million inhabitants resided in these countries (World Bank, 2021). Accordingly, their combined GDP in 2019 was approximately 3243.01 billion US dollars, with average growth above the world figures for more than two decades (World Bank, 2021), making it one of the areas with a significant driver of global growth and the fastest economic progression (ASEAN Secretariat, 2019). Coherently, the FDI of 182.46 billion US dollars was subjected to inflow into ASEAN countries in 2019; thus, its share of GDP was above the world average (World Bank, 2021). Furthermore, the shares of the shadow economy in 2017 were about 30%, 40%, 20%, 23%, and 28% of the GDP for Brunei Darussalam, Cambodia, Indonesia, Lao PDR, and Malaysia, respectively. Moreover, 34%, 10%, 41%, and 12% of the shadow economy were in the GDP of the Philippines, Singapore, Thailand, and Vietnam, accordingly (Medina & Schneider, 2019). Additionally, every ASEAN country has been a deficit ecological country for many years, with these deficits becoming increasingly substantial (Global Footprint Network, 2021; Yilanci & Korkut Pata, 2020).

This study primarily examines the relationship between the shadow economy and the ecological footprint nexus, thus incorporating energy consumption and foreign direct investment under the Environmental Kuznets Curve (EKC) hypothesis. This study makes several key contributions to the existing body of knowledge on this issue. First, along the trajectory of economic progression, FDI serves as a critical engine of growth and development for host countries (Ko et al., 2025). In this regard, both official and shadow economic activities, in tandem with investment, intensify the demand for energy as an indispensable input. This reinforces the causal nexus among economic growth, FDI, and energy consumption (Omri & Kahouli, 2014). Therefore, this study underlines how the ecological footprint of ASEAN countries is affected by energy consumption, economic growth, the shadow economy, and foreign direct investment within the framework of the EKC hypothesis. Additionally, this study employs the second-generation tests of unit root and cointegration, Driscoll–Kraay standard errors, the feasible generalized least squares (FGLS) estimator, and panel-corrected standard errors (PCSE) estimators to overcome the existence of the cross-sectional dependence issue, especially the data of ASEAN countries, which share economic, cultural, and geographic connections. Finally, the region of ASEAN countries represents a significant economic bloc with diverse aspects related to growth, shadow activities, investment, and environment issues.

To achieve the research objective, this study further comprises essential components, building upon a literature review (Section 2), data and methodology (Section 3), empirical results (Section 4), and conclusion and policy recommendations (Section 5).

2. Literature Review

The Kuznets Curve, introduced by Simon Kuznets in 1955, originally illustrated the relationship between a country’s income and its income inequality (Kuznets, 1955; Todaro & Smith, 2015). In the 1990s, this concept was adapted to the environmental context, giving rise to a term called the Environmental Kuznets Curve (EKC) (Grossman & Krueger, 1991; Panayotou, 1997; Stern et al., 1996). The EKC suggests an inverted U-shaped relationship between income per capita (X-axis) and environmental degradation (Y-axis). It posits that environmental deterioration initially rises with income growth but declines after reaching a certain income threshold (Grossman & Krueger, 1991; Panayotou, 1997; Stern et al., 1996). This implies that economic growth eventually improves environmental quality, validating the EKC hypothesis.

Within the framework of the Environmental Kuznets Curve (EKC) hypothesis, an inverted U-shaped relationship between economic growth and the ecological footprint has been widely examined across countries over the past two decades. This relationship has been identified in various contexts, such as in India using the Autoregressive Distributed Lag (ARDL) approach by Ahmed and Wang (2019), in Turkey through the Quantile ARDL method by Sharif et al. (2020), and in ASEAN countries using the Driscoll–Kraay panel regression model by Kongbuamai et al. (2020a). Likewise, Altıntaş and Kassouri (2020) and Wang et al. (2020b) reported that economic growth contributes to improved environmental quality—measured based on reductions in the ecological footprint—after surpassing a turning point of economic growth, as evidenced in 14 European Union countries using the Fixed Effects (FE) model (Altıntaş & Kassouri, 2020) and in G7 countries through the Dynamic Seemingly Unrelated Regression (DSUR) approach (Wang et al., 2020b). In contrast, several studies challenge the validity of the inverted U-shaped EKC hypothesis, suggesting that economic growth may instead accelerate ecological degradation. For example, Aşıcı and Acar (2018), who analyzed 87 countries with Fixed Effects (FE) and Random Effects (RE) models, and Dogan et al. (2020), who examined BRICST countries with Augmented Mean Group (AMG), Dynamic Ordinary Least Squares (DOLS), and Fully Modified Ordinary Least Squares (FMOLS), both reported an invalid EKC relationship. Similar evidence was found by Yilanci and Pata (2020) in China using the Bootstrap Fourier ARDL (FARDL) and by Kongbuamai et al. (2022) in Next-11 countries employing Driscoll–Kraay standard errors and feasible generalized least squares (FGLS). Beyond the EKC framework, other studies have also shown that economic growth directly intensifies environmental degradation. This has been documented in Turkey using DOLS, FMOLS, and the ARDL bounds testing approach (Imamoglu, 2018), in Pakistan with the ARDL bounds test (Danish et al., 2019a), in the United States using the ARDL bounds test (Zafar et al., 2019), in Indonesia through ARDL bounds testing (Nathaniel, 2021), and in BRICS countries using the DSUR estimator (Kongbuamai et al., 2021).

Along the country’s growth path, foreign direct investment (FDI) serves as a key driver of economic growth by facilitating capital inflows, particularly through inward foreign investment. However, the relationship between FDI and environmental degradation (ecological footprint) has been widely debated and remains contested. Zafar et al., using the ARDL bounds testing approach, reported that FDI significantly increased the ecological footprint in the United States (Zafar et al., 2019). Similarly, Doytch, applying the system GMM estimator across 117 countries, demonstrated that FDI in non-financial services sectors contributed to a rise in ecological footprints in both high- and low-income economies (Doytch, 2020). Consistent with these findings, Roy, employing the ARDL approach in India, identified a long-run positive relationship between FDI and ecological footprints (Roy, 2023). Similarly, Ponce et al., analyzing data from 100 countries through a panel ARDL estimator, found that FDI exerted a positive effect on environmental degradation (Ponce et al., 2023). By contrast, Arogundade et al. (2022) employed the Driscoll–Kraay estimator to investigate the nexus of FDI and the ecological footprint in 31 African countries. They found mixed results: (1) FDI reduced the ecological footprint in the initial stage, and (2) FDI increased the ecological footprint after the threshold. In addition, Zheng and Uprasen (2022) conducted a study investigating the impact of FDI on the ecological footprint in five ASEAN countries. The study employed the ARDL model for analysis and found a positive relationship in Indonesia and Malaysia, a negative relationship in Thailand, and no relationship in the Philippines and Singapore (Zheng & Uprasen, 2022). Moreover, Liu and Kim (2018) investigated the casualty using the PVAR Granger Causality Test in the Belt & Road Initiative (BRI) countries. They found a bidirectional relationship between the FDI and ecological footprint in the BRI countries (Liu & Kim, 2018).

In fostering economic growth along with attracting inward foreign direct investment (FDI), the energy sector functions as a core driver, underpinning growth, investment, and development. However, conventional energy consumption exerts harmful effects on the environment, exacerbating ecological degradation. The negative impact of conventional energy use on the ecological footprint has been widely documented. For example, Alola et al. found that energy consumption significantly increased the ecological footprint in EU countries using the Panel Pool Mean Group Autoregressive Distributed Lag (PMG-ARDL) approach (Alola et al., 2019). Similarly, Kongbuamai et al. reported a positive association between energy consumption and the ecological footprint in Thailand, employing the ARDL bounds testing methodology (Kongbuamai et al., 2020b). Supporting these findings, Ahmed et al. (2020) and Destek and Sinha (2020) demonstrated that higher energy use elevated ecological footprints in G7 countries via the CUP-FM and CUP-BC estimators and in OECD countries using DOLS and FMOLS, respectively.

Conversely, in certain contexts, energy use—particularly renewable energy consumption—has been shown to improve environmental quality in various countries. A positive relationship between renewable energy use and the ecological footprint was identified in BRICS countries by Danish et al. (2019b), MINT countries by Balsalobre-Lorente et al. (2019), and OECD countries by Destek and Sinha (2020) using DOLS and FMOLS estimators. Similarly, Altıntaş and Kassouri (2020) reported a beneficial effect of renewable energy consumption on the ecological footprint in EU countries using the Fixed Effects (FE) method.

The findings above suggest that economic activities, including foreign investment and energy consumption, have a significant impact on the environment. Meanwhile, the association between the shadow economy and the ecological footprint is not yet sufficiently understood, particularly in the ASEAN countries. However, some studies have demonstrated the significant impact of the shadow economy on CO₂ emissions (A. Baloch et al., 2022; Biswas et al., 2012; Camara, 2022; Dada et al., 2021). As a result, the relationship between the shadow economy and the ecological footprint requires further investigation, as the shadow economy contributes a significant amount of economic activity in comparison to the overall economy, particularly in ASEAN countries. Initially, Imamoglu (2018) investigated the impact of the informal economy on the ecological footprint using the DOLS, FMOLS, and ARDL approaches and found a positive effect of the informal economy on the ecological footprint in Turkey (Imamoglu, 2018). Similarly, Köksal et al. (2020) researched the nexus of the shadow economy and the ecological footprint using the Johansen cointegration test methodology and discovered that Turkey’s ecological footprint and shadow economy were positively correlated. Qayyum et al. (2021) also investigated the relationship between the shadow economy and the ecological footprint in South Asia countries by using the Panel ARDL estimator and revealed a significant positive impact of the shadow economy on the ecological footprint. Interestingly, they also found that the shadow economy associated with urbanization has a negative impact on the ecological footprint (Qayyum et al., 2021). In addition to this, Alvarado et al. (2022) studied the impact of the shadow economy on the ecological footprint in 95 countries using the panel Augmented Mean Group (AMG) and Common Correlated Effects Mean Group (CCE-MG) estimators and found that the informal economy contributed a positive (adverse) impact on the ecological footprint. For the CO₂ indicator, Biswas et al. (2012) utilized ordinary least squares (OLS) to find a positive relationship between the shadow economy and CO₂ emissions across over 100 countries (Biswas et al., 2012). Similarly, Dada et al. (2021) utilized African countries and employed OLS, Fixed Effects, and System Generalized Method of Moments (SGMM) and discovered that the shadow economy led to increased CO₂ emissions. A comprehensive study by A. Baloch et al. (2022) that investigated the influence of the shadow economy on CO₂ emissions in Pakistan using the ARDL bounding test, Fully Modified Least Squares (FMOLS), and Dynamic OLS (DOLS) revealed a significant long-term impact of the shadow economy on the escalation of CO₂ emissions. Conversely, a study demonstrating that the shadow economy had a negative effect on CO₂ conditions in 14 ECOWAS nations using estimators based on the System Generalized Method of Moment (SGMM) was performed by Camara (2022). Based on the literature review, this study proceeds to collect data from ASEAN countries and develops an analytical methodology to address the research objectives in the subsequent section.

3. Data and Methodology

3.1. Data Source

This current study examines the relationship among the ecological footprint, economic growth, energy consumption, shadow economy, and foreign direct investment in ASEAN countries. Given the availability of annual data and to avoid distortions caused by the COVID-19 pandemic, the data for analysis were collected over the period from 1993 to 2017. The analysis includes nine ASEAN countries: Brunei Darussalam, Cambodia, Indonesia, Lao PDR, Malaysia, the Philippines, Singapore, Thailand, and Vietnam. Apparently, the data on the ecological footprint (global hectare per person) were retrieved from the Global Footprint Network (2021). Furthermore, economic growth data (real GDP constant 2010 US$ per person), energy consumption data (kg of oil equivalent per capita), and foreign direct investment data (% of GDP) were obtained from the World Development Indicators from the World Bank (2021). In addition, data on the shadow economy (Index 0–100) were gathered from the CESifo Working Papers No. 7981 (Medina & Schneider, 2019).

3.2. Model Specification

From the research gap, the functional form of the EKC conceptualization in this study is drawn from a prior study of Grossman and Krueger (1991). This is illustrated in Equation (1) as follows:

$$EF=f(GDP,GD{P}^2,ENC,SHA,FDI),$$

(1)

In Equation (1), the EF is the ecological footprint, and GDP and GDP² are economic growth and its square. ENC, SHA, and FDI also represent energy use, the shadow economy, and foreign direct investment, respectively. Consistent with the study of Ozturk and Acaravci (2010), Katircioğlu (2014), and Zaman et al. (2016), we convert all variables into the natural logarithmic form to capture the growth effects and reduce heteroscedasticity. Hereafter, the empirical equation can be rewritten in Equation (2):

$$\ln E{F}_{it}={\beta }_0+{\beta }_1\ln GD{P}_{it}+{\beta }_2\ln GD{P^2}_{it}+{\beta }_3\ln EN{C}_{it}+{\beta }_4\ln SH{A}_{it}+{\beta }_5\ln FD{I}_{it}+{\epsilon }_{it},$$

(2)

Equation (2) ${\beta }_0$ is the slope of the coefficient. ${\beta }_1,{\beta }_2,{\beta }_3,{\beta }_4$ coefficients reflect the importance of each independent variable concerning the dependent variable in country i at time t. The error term, denoted by ${\epsilon }_{it}$, is assumed to have a normal distribution.

3.3. Econometric Methodologies

3.3.1. Test for Cross-Sectional Dependence

In panel data analysis, economic data often have cross-sectional dependence (CD) issues because those countries are interlinked in several ways, e.g., economic policy, trade agreements, technology and financial spillover, and shared borders, etc. (Phillips & Sul, 2003; Zaidi et al., 2019). Therefore, this study previously investigated the CD issue among variables to prevent biased and unreliable results (Paramati et al., 2017). To achieve this, the Breusch and Pagan LM test (Breusch & Pagan, 1980) and Pesaran CD test (Pesaran, 2004) were applied as shown in Equations (3) and (4):

$$LM=T{\displaystyle \sum _{i=1}^{N-1}{\displaystyle \sum _{j=i+1}^N{\widehat{\rho }}_{ij}^2}},$$

(3)

$$CD=\sqrt{\left({\displaystyle \frac{2T}{N\left(N-1\right)}}\right)}\left({\displaystyle \sum _{i=1}^{N-1}{\displaystyle \sum _{j=i+1}^N{\widehat{\rho }}_{ij}}}\right),$$

(4)

where T and N are the time period and sample size. ${\widehat{\rho }}_{ij}$ represents the pairwise correlation of error terms between countries i and j. The null hypothesis (H₀) for the LM and CD tests assumes no CD in the panel data.

3.3.2. Test for Panel Unit Root

After testing the CD issue in panel studies, the second-generation panel unit root tests are advised for analyzing data with the presence of CD and homogeneity in the data set (Dogan & Seker, 2016; Kongbuamai et al., 2022; Pesaran, 2004). To test this, the cross-sectionally augmented ADF (CADF) and the cross-sectionally augmented IPS (CIPS) unit root tests by Pesaran (2007) are applied to check their stationary properties. The model for the CADF test can be formulated in Equation (5):

$$\Delta y_{it}=a_i+{\rho }_i{y}_{it-1}+{\rho }_i{\overline{y}}_{it-1}+{\displaystyle \sum _{j=0}^k{\gamma }_{ij}\Delta {\overline{y}}_{it-1}}+{\displaystyle \sum _{j=0}^k{\delta }_{ij}{y}_{it-1}}+{\epsilon }_{it},$$

(5)

where $\Delta$$a_i$ are the differences between parameters and the intercept value. $\kappa$ and ${\overline{y}}_t$ denote the lag order and mean value of CD for time t.

From Equation (5), the CIPS is designed based on the mean of CADF values, and the estimation model can be reformed as in Equation (6):

$$CIPS=\left({\displaystyle \frac{1}{N}}\right){\displaystyle \sum _{i=1}^k{t}_i\left(N,T\right)},$$

(6)

The null hypothesis (H₀) for the panel unit root test assumes that the panel data contain a unit root.

3.3.3. Test for Panel Cointegration

After the panel unit root test, the next step is to analyze the cointegration properties of the ecological footprint, economic growth, energy consumption, shadow economy, and foreign direct investment. To analyze this, the LM bootstrap panel cointegration test by Westerlund and Edgerton (2007) was the second-generation panel cointegration test and was used due to issues of the existence of the CD and heterogeneity attributes across the panel (Dogan & Aslan, 2017; Dogan & Seker, 2016; Wang et al., 2020a). The LM bootstrap panel cointegration test can be calculated as follows:

$$L{M}_N^{+}=\left({\displaystyle \frac{1}{N{T}^2}}\right){\displaystyle \sum _{i=1}^N{\displaystyle \sum _{t=1}^T{\stackrel{⌢}{w}}_i^{-2}}{s}_{it}^2},$$

(7)

where $T$ and N denote the period and sample size. ${\stackrel{⌢}{w}}_i^{-2}$ and $s_{it}$ are the long-run variance of the residuals and the partial sum process of error terms. For this test, the null hypothesis (H₀) for the LM bootstrap panel cointegration test assumed that no cointegration exists in the panel data.

3.3.4. Test for Long-Run Elasticities

At this stage, the study assesses the long-run relationship among variables after the fundamental analyses, which include the cross-sectional dependence test, panel unit root test, and panel cointegration. Once there are cross-sectional dependence issues in panel data, the Driscoll–Kraay standard errors estimation by Driscoll and Kraay (1998) is suggested for this purpose (Hashemizadeh et al., 2021; Kongbuamai et al., 2020a). Also, this method confronts the problem of balanced and unbalanced panel data, cross-sectional dependence, heteroscedasticity, missing values, and autocorrelations, etc. (Hashemizadeh et al., 2021; Kongbuamai et al., 2020a; Sarkodie & Strezov, 2019). The Driscoll–Kraay standard error estimation for pooled ordinary least squares (OLS) can be formulated in the linear model form as follows:

$$y_{i,t}=x_{i,t}^{\prime }\beta +{\epsilon }_{i,t},i=1,\dots ,N,t=1,\dots ,T.$$

(8)

where i and t are the individual/cross-sectional data and time-series data. For this analysis, the null hypothesis (H₀) assumes no long-run relationship in the panel data.

To consolidate the robustness of the long-run results, this study further analyzes the long-run relationship by using the feasible generalized least squares (FGLS) model by Parks (1967) and panel-corrected standard errors (PCSE) estimator by Beck and Katz (1995). Furthermore, the FGLS and PCSE estimators can reveal the results to overcome the cross-sectional dependence and heteroskedasticity problems in panel data analysis (Hoechle, 2007; Khan et al., 2020; Reed & Ye, 2011; Shahbaz et al., 2019; Wu et al., 2021). Additionally, the FGLS can achieve reliable findings when the number of time periods (T) ≥ the number of cross-sections (N) (Hoechle, 2007; Kongbuamai et al., 2022; Reed & Ye, 2011; Wu et al., 2021). Thus, the PCSE can achieve reliable findings when the number of cross-sections (N) ≥ the number of time periods (T) (Hoechle, 2007; Khan et al., 2020; Reed & Ye, 2011). For the hypothesis, the null hypothesis (H₀) for the FGLS and PCSE estimators assumes that there is no long-run relationship in the panel data. After completing the discretionary methodological design, the results of these analyses are presented in the next section.

4. Empirical Results

4.1. Results of Descriptive and Correlation Tests

Table 1. Descriptive statistics.

Variables	lnEF	lnGDP	lnGDP2	lnENC	lnSHA	lnFDI
Mean	0.756	8.223	69.798	9.240	3.341	1.255
Median	0.507	7.922	62.761	8.931	3.407	1.299
Maximum	2.111	10.960	120.122	12.161	4.058	3.379
Minimum	−0.304	5.772	33.319	5.355	2.240	−2.870
Std. Dev.	0.692	1.479	25.299	1.620	0.463	1.048
Skewness	0.507	0.410	0.613	0.136	−0.575	−0.647
Kurtosis	1.825	1.982	2.086	2.178	2.581	4.292
Jarque-Bera	22.612	16.010	21.954	7.0268	14.081	31.372
Probability	0.000	0.000	0.000	0.029	0.000	0.000
Observation	225	225	225	225	225	225

Source: author’s compilation.

reports descriptive statistics, and

Table 2. Correlation matrix of variables.

Variables	lnEF	lnGDP	lnENC	lnSHA	lnFDI
lnEF	1.000
lnGDP	0.955 (0.000)	1.000
lnENC	0.956 (0.000)	0.970 (0.000)	1.000
lnSHA	−0.466 (0.000)	−0.450 (0.000)	−0.516 (0.000)	1.000
lnFDI	0.287 (0.000)	0.173 (0.009)	0.209 (0.001)	−0.408 (0.000)	1.000

Note: p-values are presented in parentheses. Source: author’s compilation.

demonstrates the correlation matrix of variables. It indicates that the data have a normal distribution according to the Jarque–Bera hypothesis, and variables are correlated according to Pearson’s correlation test. Thus, Figure 1 demonstrates the correlation of each pair of variables, including maximum, minimum, and mean values.

4.2. Results of the Cross-Sectional Dependence Test

Table 3. Results of cross-sectional dependence test.

Variable	LM Test		CD Test
	Statistic	p-Value	Statistic	p-Value
lnEF	44.53	0.155	9.05 ***	0.000
lnGDP	51.30 **	0.047	17.67 ***	0.000
lnENC	59.69 ***	0.007	24.62 ***	0.000
lnSHA	86.46 ***	0.000	23.23 ***	0.000
lnFDI	37.32	0.408	3.00 ***	0.003

Note: *** and ** denote statistical significance at the 1% and 5% levels, respectively. Source: author’s compilation.

shows the results of the cross-sectional dependence test. Based on the p-value, the null hypothesis is rejected, especially with the method of the CD test. It confirms the existing cross-sectional dependence in the panel data.

4.3. Result of Panel Unit Root Test

After the CD confirmation among variables,

Table 4. Results of CADF and CIPS panel unit root tests.

Test Statistics		lnEF	lnGDP	lnENC	lnSHA	lnFDI
CADF	Level	−1.553	−1.456	−1.942	−1.331	−2.680 ***
	First difference	−3.163 ***	−2.651 ***	−3.450 ***	−2.612 ***	−4.376 ***
CIPS	Level	−0.838	−0.324	−0.363	−0.969	−2.687 ***
	First difference	−4.213 ***	−3.826 ***	−4.310 ***	−4.802 ***	−5.735 ***

Note: *** denotes statistical significance at the 1% level. Source: author’s compilation.

illustrates the second generation of panel unit root tests using the CADF and CIPS methods. According to the p-value, the null hypothesis is rejected at the first difference for both methods, indicating that the variables are stationary at that range accordingly.

4.4. Results of Panel Cointegration Test

From the above analysis, the CD issue and stationary at the first difference are statistically confirmed. Obviously, the LM bootstrap panel cointegration test verifies that the null hypothesis cannot be rejected, as in

Table 5. LM bootstrap panel cointegration test.

Tests	Constant		Constant and Trend
	LM Statistic	Bootstrap p-Value	LM Statistic	Bootstrap p-Value
LM bootstrap	7.011	0.999	15.258	0.993

Source: author’s compilation.

. Likewise, the p-values are nearly one. It can be concluded that the ecological footprint, economic growth, energy consumption, shadow economy, and foreign direct investment are cointegrated in this case.

4.5. Results of the Long-Run Relationship Based on the Driscoll–Kraay Standard Errors Estimation

According to the long-run relationship based on the Driscol–-Kraay standard errors estimation, the empirical findings reveal that the inverted U-shaped Environmental Kuznets Curve (EKC) does not hold between economic growth and the ecological footprint across all estimated models—Model 1 (EKC and energy consumption), Model 2 (EKC, energy consumption, and shadow economy), and Model 3 (EKC, energy consumption, shadow economy, and foreign direct investment). Moreover, energy consumption consistently demonstrates a positive and statistically significant relationship with the ecological footprint in ASEAN countries across all models. The nexus between the shadow economy and ecological footprint (Model 2 and Model 3) also confirms a positive connection in these countries. In addition, foreign direct investment (Model 3) is positively related to the ecological footprint in ASEAN countries according to the Driscoll–Kraay standard error estimation. These results are illustrated in

Table 6. Results of regression with Driscoll–Kraay standard errors.

Variable	Model (1)		Model (2)		Model (3)
	Coefficient	p-Value	Coefficient	p-Value	Coefficient	p-Value
lnGDP	−0.484 ***	0.000	−0.657 ***	0.000	−0.408 ***	0.002
lnGDP2	0.037 ***	0.000	0.045 ***	0.000	0.032 ***	0.000
lnENC	0.271 ***	0.000	0.320 ***	0.000	0.286 ***	0.000
lnSHA	-	-	0.096 ***	0.000	0.123 ***	0.000
lnFDI	-	-	-	-	0.058 ***	0.001
Constant	−0.400 **	0.020	−1.876	0.120	−1.316 ***	0.001
F statistics	1624.82		1327.94		1285.82
p-value	0.000		0.000		0.000
R2	0.938		0.940		0.945
Root MSE	0.173		0.170		0.163
Observation	225		225		225
Number of groups	9		9		9

Note: *** and ** denote statistical significance at the 1% and 5% levels, respectively. Source: author’s compilation.

.

In Model 3, a 1% increase in economic growth is associated with a 0.408% reduction in the ecological footprint, whereas a 1% rise in the squared term of economic growth corresponds to a 0.032% increase in the ecological footprint across ASEAN countries, holding other factors constant. These findings indicate that the Environmental Kuznets Curve (EKC) hypothesis—which suggests an inverted U-shaped relationship between economic growth and environmental degradation—is not valid in the context of ASEAN nations. Instead, the results resemble a U-shaped trend, where economic growth is accompanied by an increase in environmental degradation. The ASEAN countries are often classified as middle-income and low-income nations. Therefore, their economic activities (industry, manufacturing, economic activities) still rely on traditional technology with cost-saving customs for their economically efficient processes. The results of this study are in line with the findings in 87 countries (Aşıcı & Acar, 2018) and BRICST countries (Dogan et al., 2020). However, it is in contrast with the study in ASEAN countries (Kongbuamai et al., 2020a), 14 European (EU) countries (Altıntaş & Kassouri, 2020), G7 countries (Wang et al., 2020b), and India (Ahmed & Wang, 2019).

For a further finding of the energy dimension, a 1% rise in energy consumption is associated with a 0.286% increase in the ecological footprint in ASEAN countries, assuming that all other factors remain constant. This result highlights energy consumption as the most significant driver of environmental impact, as evidenced by the magnitude of its estimated coefficient. The higher the energy consumption, the higher the environmental degradation (ecological footprint) in ASEAN countries. Thus, energy consumption has the strongest impact on environmental quality. The plausible reason is that petroleum, natural gas, and other conventional energies are massively used in ASEAN countries compared to renewable energy utilization. Additionally, the negative impacts of renewable energy production in ASEAN countries—particularly from sources such as biomass and biogas—remain substantial across their life cycles, largely due to low environmental efficiency, the reliance on traditional technologies, and unsustainable practices. The findings of this study align with previous research conducted in various regions, including EU countries (Alola et al., 2019), Thailand (Kongbuamai et al., 2020b), G7 countries (Ahmed et al., 2020), Organization for Economic Co-operation and Development (OECD) countries (Destek & Sinha, 2020), ASEAN countries (Kongbuamai et al., 2021), and Next-11 countries (Kongbuamai et al., 2022). However, contrasting evidence has been reported in studies focusing on the BRICS countries (Danish et al., 2019b), MINT countries (Balsalobre-Lorente et al., 2019), OECD (Destek & Sinha, 2020), and EU countries (Altıntaş & Kassouri, 2020), where differing patterns were observed.

For the shadow economy, a 1% increase in the size of the shadow economy is associated with a 0.123% rise in the ecological footprint in ASEAN countries, assuming that other factors remain constant. This finding suggests that a larger shadow economy contributes to greater environmental degradation across the region. It can be explained that much shadow money is being invested in non-environmentally friendly businesses and activities. Also, shadow money tends to be invested in short-term business, which is concerned with high returns as a core principle, with the limitation of monitoring and controls. As a result, the shadow economy undermines environmental sustainability in ASEAN countries for several reasons. First, informal firms often operate outside regulatory frameworks, allowing them to bypass environmental standards and avoid compliance with pollution control measures. This results in excessive emissions, unregulated waste disposal, and the unsustainable use of natural resources. Second, weak institutional oversight and limited fiscal capacity for environmental protection and mechanisms in many ASEAN nations make it difficult to monitor and enforce environmental laws, particularly when large segments of economic activity remain hidden from official statistics. Third, shadow economic activities frequently rely on outdated, low-efficiency technologies that maximize short-term profits but cause long-term ecological damage. In conjunction with our findings, this discovery aligns with previous research indicating a positive correlation between (1) the shadow economy (informal economy) and ecological footprint in Turkey (Imamoglu, 2018; Köksal et al., 2020), as well as in 95 countries (Alvarado et al., 2022), and (2) the shadow economy and CO₂ emissions in Pakistan (A. Baloch et al., 2022). However, our study’s results contradict those observed in South Asian countries regarding the relationship between the shadow economy and ecological footprint (Qayyum et al., 2021).

Lastly, a 1% increase in foreign direct investment increased the ecological footprint by 0.058% in ASEAN countries, with other things being equal. This result suggests that greater foreign direct investment inflows lead to more environmental deterioration in the region. This phenomenon is largely attributed to foreign investment in ASEAN countries, where firms often relocate from regions with stringent environmental standards to jurisdictions with weaker regulations in pursuit of cost savings, as outlined by the Pollution Haven Hypothesis. Moreover, much of the foreign investment in ASEAN flows into traditional manufacturing and resource-intensive industries, which are inherently associated with extractive practices and high levels of pollution. In many cases, these investments were established many decades ago, relying on outdated technologies and unsustainable production methods. Consequently, such inflows of foreign capital not only perpetuate obsolete industrial practices but also exacerbate environmental degradation. This effect is further amplified by the already elevated levels of emissions in developing countries (e.g., lower emission reduction commitments under the Kyoto Protocol and Paris Agreement, etc.), making foreign investment a significant contributor to environmental deterioration across the region. In response to this finding, there are some similar results in the study of developing countries (Solarin & Al-Mulali, 2018), BRI countries (M. A. Baloch et al., 2019), high-income and low-income countries (Doytch, 2020), 100 countries (Ponce et al., 2023), the United States (Zafar et al., 2019), and India (Roy, 2023). Nonetheless, it differs from the findings from developed countries (Solarin & Al-Mulali, 2018), 31 African countries (the initial stage), Thailand (Zheng & Uprasen, 2022), the Philippines, and Singapore (Zheng & Uprasen, 2022).

4.6. Results of Long-Run Relationship Based on FGLS and PCSE Estimations

The FGLS and PCSE estimations were used to further confirm the robustness of the long-run relationship. The FGLS results indicate that a 1% increase in economic growth leads to a 0.468% reduction in the ecological footprint, while a 1% increase in the squared term of economic growth results in a 0.047% rise in the ecological footprint, holding other factors constant. Similarly, the PCSE estimation reveals that a 1% rise in economic growth reduces the ecological footprint by 0.756%, whereas a 1% increase in the square of economic growth increases the ecological footprint by 0.051%. In conclusion, these results also confirm the absence of an inverted U-shaped Environmental Kuznets Curve (EKC) relationship between economic growth and the ecological footprint in ASEAN countries for both FGLS and PCSE estimations. It is similar to the result (invalidity of the EKC hypothesis) from the analysis of Driscoll–Kraay standard error estimation.

Furthermore, a 1% increase in energy consumption in ASEAN countries increased the ecological footprint by 0.102% and 0.317%, according to FGLS and PCSE estimations, respectively, keeping other things constant. Also, energy consumption has the most significant coefficient related to the ecological footprint. This is the same direction as the result from the Driscoll–Kraay standard error estimation. For the shadow economy, a 1% increase in the shadow economy in ASEAN countries increased the ecological footprint by 0.070% and 0.121% according to FGLS and PCSE estimations, respectively, ceteris paribus. This is the same direction as the result from the Driscoll–Kraay standard error estimation. Additionally, a 1% increase in foreign direct investment in ASEAN countries increased the ecological footprint by 0.020% and 0.041% according to FGLS and PCSE estimations, respectively, with other things being equal. This is the same direction as the result from the Driscoll–Kraay standard error estimation. These results are illustrated in

Table 7. Results of FGLS and PCSE estimations.

Variable	FGLS		PCSE
	Coefficient	p-Value	Coefficient	p-Value
lnGDP	−0.468 ***	0.000	−0.756 ***	0.000
lnGDP2	0.047 ***	0.000	0.051 ***	0.000
lnENC	0.102 ***	0.000	0.317 ***	0.000
lnSHA	0.070 *	0.053	0.121 ***	0.000
lnFDI	0.020 ***	0.001	0.041 ***	0.000
The Wald Chi2	3379.15		33,470.48
p-value	0.000		0.000
R-squared	-		0.974
Observation	225		225
Number of groups	9		9

Note: *** and * denote statistical significance at the 1% and 5% levels, respectively. Source: author’s compilation.

.

5. Conclusions and Policy Recommendations

The expansion of the shadow economy and foreign direct investment have been explored recently, notably in response to environmental sustainability. These aspects are, at the very least, theoretically, essential for a sustained process of a country’s development (for obtaining the SDGs). Therefore, the main objective of this study is to examine the impact of economic growth, energy consumption, shadow economy, and foreign direct investment on the ecological footprint in ASEAN countries. In doing this, macro-data from Brunei, Cambodia, Indonesia, Lao PDR, Malaysia, the Philippines, Singapore, Thailand, and Vietnam from 1993 to 2017 were gathered for analysis due to their availability. For accurate data analysis and to overcome the problems of cross-sectional dependence, a series of econometric methods was utilized in this study, namely the cross-sectional dependence test, panel unit root tests, panel cointegration test, Driscoll–Kraay standard errors, feasible generalized least squares (FGLS), and panel-corrected standard errors (PCSE) estimators. As a result, the inverted U-shaped Environmental Kuznets Curve (EKC) hypothesis does not hold in the context of ASEAN countries. Moreover, energy consumption, the shadow economy, and foreign direct investment exhibit a positive and statistically significant relationship with the ecological footprint, as confirmed based on the Driscoll–Kraay standard errors, feasible generalized least squares (FGLS), and panel-corrected standard errors (PCSE) estimations.

For policy implications, governments should promote economic growth in line with the 2030 Sustainable Development Goals (SDGs) by strengthening governance and regulatory frameworks to curb the shadow economy, while advancing sustainable energy use and foreign investment. ASEAN governments should establish a regional framework agreement aimed at reinforcing institutional capacity through enhanced monitoring systems, greater transparency, and digital platforms that make informal economic activities more visible and traceable.

Second, targeted regulatory reforms are necessary to encourage the transition of informal firms into the formal economy. These may include tax incentives, simplified business registration procedures, and access to credit contingent upon compliance with environmental standards. Such measures would expand the country’s fiscal budget for environmental monitoring mechanisms and national development initiatives.

Third, governments should design foreign investment promotion strategies that prioritize sustainable and responsible investments across key sectors, particularly in clean energy technologies. This requires the adoption of clear environmental, social, and governance (ESG) standards, as well as technology transfer policies for incoming investors. Enforcing strict environmental regulations, mandating transparent reporting, and linking investment incentives to measurable sustainability outcomes would help prevent “pollution haven” dynamics while fostering a “pollution halo” effect.

In addition to this, incentive structures must be redesigned to stimulate both domestic and foreign responsible investment, including tax breaks for green technologies, subsidies for low-carbon innovation, and preferential treatment for firms demonstrating strong sustainability commitments.

Finally, long-term economic growth in ASEAN countries will depend on the deeper integration of environmental sustainability, which can be achieved by reducing reliance on conventional energy and accelerating the adoption of high-technology solutions and investment (by citizens or foreigners) in sustainable energy production and consumption, supported by tax reductions and targeted subsidies.

Based on the limitation of data availability, future panel data research should aim to extend both the temporal coverage and the number of countries included, drawing on local data sources from each country. Moreover, additional explanatory variables could be incorporated to enrich the analysis and capture the complexity of the nexus under investigation. In this regard, advanced methodological approaches such as Machine Learning techniques, Quadratic Regression, and Bayesian modeling could be employed to enhance robustness and predictive accuracy. Furthermore, time-series analyses focusing on the shadow economy–environment relationship could be conducted within alternative theoretical frameworks and methodological paradigms, offering deeper insights into the dynamic interactions between informal economic activities and environmental sustainability.