The shadow economy in South Asia: dynamic effects on clean energy consumption and environmental pollution.

This study explores the symmetric and asymmetric effects of the shadow economy on clean energy and air pollution of South Asian countries over the period 1991-2019. The short-run ARDL findings for the clean energy model suggest that shadow economy increases clean energy consumption in Pakistan and Sri Lanka, whereas this effect is negative for India and insignificant for other countries. The long-run results indicate the adverse impact only for India and the effects of tax revenue on clean energy are positively significant in Sri Lanka while negatively signiicant in Nepal and Bangladesh. Institutional quality significantly increases clean energy in Pakistan, India, and Nepal. However, in the case of Pakistan and Nepal, institutional quality deteriorated the environmental quality. The results for the pollution model confer that shadow economy increases emissions in Pakistan, decreases in Bangladesh and Nepal, and has no effect in India and Sri Lanka. The nonlinear ARDL results reveal that the positive components of the shadow economy significantly increase clean energy consumption only in Pakistan; however, the negative components of the shadow economy are negatively significant in all countries except Sri Lanka and Nepal. However, the negative component of the informal sector of the economy reduces CO2 emissions in India and increases CO2 emissions in Bangladesh and Nepal. The results offer important policy implications for achieving clean energy and better environmental quality in South Asian countries.

Introduction

It is a well-known notion now that environmental economics has become one of the most emerging areas of interest for researchers from all over the world. Most of the time, the researchers use carbon dioxide emissions as a proxy for measuring environmental degradation, occasionally the researchers use few other proxies as well such as nitrous oxide and sulfur dioxide. On the other hand, economic activity is measured by the gross domestic product. We found immense literature on defining the linkage between economic activity and environmental degradation (Apergis and Ozturk 2015; Özokcu and Özdemir 2017; Churchill et al. 2018; He et al. 2018; Majeed and Tauqir 2020; Zulfa and Resha 2020; Sarkodie and Ozturk 2020). However, we could not find many studies on the role of informal or shadow economy in environmental degradation. This issue is even more important for developing countries where the size of the shadow economy is really large. Therefore, it is interesting to find out the linkage between the shadow economy and environmental degradation in the form of air pollution, whereas air pollution is considered as a harmful substance in the air and most industries and vehicle emissions are considered the main cause behind these emissions.

The existence of an informal or shadow economy is a vital issue in both developing and developed countries around the globe. In the economic literature, we found different names for the description of the shadow economy, for example, informal economy, irregular economy, underground economy, black economy, and unofficial economy. It is important to quantify the existence of the shadow economy as such the whole economic activities cannot be presented by the conventional approach of measuring formal GDP. Thus, reducing the size of the shadow economy is an important task for governments. Meanwhile, in the recent era, the research in environmental degradation has taken the central stage in the policy arena (Chen et al. 2018a, b; Lv et al. 2019, Sohail et al. 2019). Thus, we can say with certainty that any study on environmental degradation without incorporating the role of the shadow economy is not providing a complete picture.

Even though we cannot find any consistent definition of shadow economy in the literature, most of the researchers described shadow economy as the unaccented income activities in the official income accounts of an economy. Sometimes, it is referred to those income activities which are not part of the nation’s regulatory frameworks (Sassen 1994; Tanzi 1999; Sohail et al. 2014). Under the umbrella of the shadow economy, the firms take the advantage of tax evasion and lax environmental standards and keep on producing different goods without taking care of environmental quality. The governments around the world try their best to decrease the size of the shadow economy due to its deep-rooted economic impacts on different sectors of the economy. To combat environment-related issues, it is important to quantify the shadow economy.

To measure the shadow economy, the literature is dominated by three different approaches. The first approach is a direct approach to measure the shadow economy; it is also called a survey-based approach (Haigner et al. 2013). The second approach is called the indirect approach to measure the shadow economy and it uses different macroeconomic variables as proxies to measure the existence of the shadow economy, the most prominent proxies for the shadow economy are currency demand and income and expenditure gaps (Thomas 1999; Alm and Embaye 2013). The third approach is also very well known in the literature in measuring the shadow economy and it is based on the multiple indicators and multiple causes methods, which is usually known as a MIMIC method. This method has an advantage in measuring the shadow economy from multiple causes, for example, tax morality, tax burden, and self-employment. We found a rich literature which is based on this method to measure the shadow economy, for instance, Wang et al. (2006), Asiedu and Stengos (2014), Kaghazian et al. (2015), Schneider and Buehn (2016), and Franić (2019).

Few studies in the previous literature explored the linkage between shadow economy and environmental pollution. These studies found a positive association between shadow economy and environmental pollution, for example, Biswas et al. (2012), Elgin and Oztunali (2014), Hille (2018), Chen et al. (2018a, b), Köksal et al. (2020), and Pang et al. (2020). According to the previously mentioned literature, due to the existence of the shadow economy, the environmental standard is lax in these countries, and therefore the firms take the advantage of these lax environmental standards and produce different products without taking care of environmental standards, and therefore it brings an increase in air pollution in these countries. We also found relevant literature on the impacts of environmental regulation on air pollution and as we expect, the majority of the evidence supports the negative association between environmental regulations and air pollution (Chattopadhyay et al. 2010; Mazhar and Elgin 2013; Li et al. 2019; and Hashmi and Alam 2019).

South Asia is a developing country that is confronting many challenges at this point. It includes poverty, unemployment, inflation, and a huge gap between revenues and expenditures. South Asia has a lower tax base as compared to other developing countries in the world. To combat these challenges, there is a need for effective policy steps and solutions. However, in the presence of a shadow economy, it is not possible to get the desired results out of these policy solutions. Several studies have been conducted to measure the shadow economy in the case of Asia, for instance, Khan and Khalil (2017), Mughal and Schneider (2018), and Huynh (2020). All of these mentioned studies provide evidence of the existence of the shadow economy in the case of South Asia. Therefore, it is interesting to assess the impact of the shadow economy on air pollution for the South Asian economy. We choose the South Asia economies that are at a small level of the formal economy.

We have found many studies in the literature to elaborate the impact of shadow economy on environmental pollution in the case of different countries around the world, for example, Abid (2015) for Tunisia, Blackman et al. (2006) for Mexico, Chen et al. (2018a, b) for China, Benkraiem et al. (2019) for Bolivia, and similar studies for other developed and developing countries. The previous literature does not provide studies on this topic in South Asian countries. It adds to the existing literature in various ways. First, it is a first-ever attempt to analyze the impact of the shadow economy on air pollution in the case of South Asian countries. Second, it applied the latest econometric technique, i.e., nonlinear ARDL, to capture the impact of shadow economy indicators on air pollution. Nonlinear ARDL has the added advantage to any other conventional method of estimation that it is capturing the asymmetries in the said relationship in the case of South Asian countries. There are very few studies in the existing literature that used the nonlinear ARDL model to capture asymmetries in defining the role of the shadow economy in air pollution. This study will provide new insights in this regard. Third, it is also assessing the role of fiscal policy instruments on air pollution, and indirectly it is also giving an insight into the role of the shadow economy and its impacts on air pollution by incorporating asymmetries. By incorporating the role of asymmetries in our analysis, we can find better results as compared to the other conventional approaches, and we will be able to get a few significant and superior empirical results. Our study will provide a few very important policy suggestions for the policy practitioners, which may be generalized for many other developing countries in the world.

Literature review

It is a generally observed notion that government regulation and environmental standards are the two most important and detrimental factors of the level of air pollution in an economy (Elgin and Oztunali 2014). The problem of air pollution even becomes worse in low-income and developing countries where environmental standards are lax due to the presence of a large segment of the shadow economy. In these countries, firms in the informal sector take advantage of these lax environmental standards and use polluting intermediate goods to produce the final goods. Therefore, the shadow economy is playing an important part to raise air pollution levels. However, the negative impacts of the shadow economy can be mitigated through the control of corruption.

Most of the recent literature found a positive association between the shadow economy and the rise in air pollution, as it is mostly observed from the previous literature that the shadow economy mostly worsens the environmental problems (Chaudhuri and Mukhopadhyay 2006a, b; Elgin and Oztunali 2014; Imamoglu 2018). In this regard, Pang et al. (2019) explored the association between the shadow economy and air pollution for three Chinese provinces, and the empirical results confirm that the shadow economy is the most important source of an increase in air pollution. In the same perspective, Köksal et al. (2020) explored the association between shadow economy and ecological footprints for the case of Turkey during the time period from 1961 to 2014. The results of the study confirmed that in the long run, shadow economy plays a positive role in surging up the ecological footprints. Similarly, in another interesting study, Huynh (2020) explained the impact of the shadow economy on air pollution in 22 developing countries in the Asian region during the time period 2002 to 2015. By applying the GMM and fixed effects models, the study revealed that the shadow economy is positively associated with the higher level of air pollution in the selected region. Biswas et al. (2012) theoretically explained the underlying linkage among shadow economy, corruption, and environmental pollution. However, the study inferred that control of corruption is an important way to decrease the level of environmental pollution. In another interesting study of 22 selected countries of Sub-Saharan Africa, Nkengfack et al. (2020) explained the negative association between shadow economy and environmental quality. However, this relationship is statistically significant in the case of lower-middle-income countries included in the sample. Maddah (2017) described the relationship among corruption, shadow economy, and environmental quality and concluded that an increase in corruption is upsurging the incidence of the shadow economy, which in turn increases the environmental pollution.

Most of the studies take into account the role of environmental regulation when discussing the impact of the shadow economy on any form of environmental degradation. Gupta (2006) assessed the impact of environmental regulations on the shadow economy and the study concluded a positive association between them. Chaudhuri and Mukhopadhyay (2006a, b) are of the view that tax increase is one of the main reasons for an increase in the size of the informal economy. On the same lines, Mazhar and Elgin (2013) also described that strict environmental regulations are causing an upsurge in the size of the shadow economy. Besides, the previous literature also supports the positive impact of government expenditures on air pollution (Bernauer and Koubi 2013; Galinato and Galinato 2016; Sun et al. 2020). On the other hand, Halkos and Paizanos (2016) revealed that government spending is an important source to decrease the level of pollution. Similarly, many researchers are of the view that an increase in carbon tax may help decrease environmental pollution (Halkos and Paizanos 2013; Hafeez et al. 2019). Gerlagh et al. (2018) explained that a higher rate of the carbon tax is helping to reduce CO2 emissions in selected countries from the European Union.

Shadow or informal economy is considered to be the most important factor to bring an increase in air pollution for different countries in the world. Abid (2015) explored that the informal economy is bringing an upsurge in CO2 emissions. Therefore, it is important to limit the size of the informal economy in Tunisia. Similarly for Mexico, Blackman et al. (2006) found out that controlling the size of the informal economy is an important step to cut down the level of air pollution. However, Baksi and Bose (2010) revealed that there is a nonlinear relationship between the informal economy and the level of air pollution. According to the study, strict environmental regulations can help to curtail the level of air pollution.

The previous literature also provides insights into government size and levels of air pollution. In this context, Bernauer and Koubi (2013) explained that a higher level of government spending is a major cause of an increase in air pollution. The reason behind this result is that most of the time large governments have to suffer from bureaucratic inefficiencies. Galinato and Galinato (2016) explained that an increase in government spending coupled with social safety nets brings an increase in CO2 emissions as to ensure food security; the local authorities expand agricultural land and the resultant deforestation causes an upsurge in CO2 emissions. Galinato and Islam (2017) explained two types of consumption-generated pollution effects concerning the increase in government size and pollution. (1) Due to increase in income, the consumption-generated pollution also increases. (2) Due to strict environmental regulations, the consumption-generated pollution decreases. However, Farzanegan and Markwardt (2018) inferred that government size is negatively connected with CO2 emissions in selected Middle East and North African countries.

In the context of the South Asian economy, almost no attention has been paid to the existing literature to assess the impact of the shadow economy on environmental pollution. Therefore, the present study will provide an important insight into the existing literature on the impact of the shadow economy on environmental pollution. Besides, the study will apply the nonlinear ARDL model for estimating this important linkage which also captures asymmetries in the said relationship.

Model, methodology, and data

To observe the impact of the shadow economy on clean energy consumption and carbon emission in selected South Asian economies, we have created the following models (1 and 2) based on the information provided by previous studies.

Equations (1) and (2) are long-run models in which carbon (CO2) emission and clean energy (CE) consumption in South Asian economies depend on the shadow economy (SE), tax revenue (TAX), and control of corruption (CC). The model formulation and variable selection are based on Biswas et al. (2012) and Huynh (2020) studies. We use the annual time series data and data retrieved from the World Bank, global economic indicators, and world governance indicators. The detailed variable information is given in Table 1. Estimates of β1 and α1 could be negative and positive in both models. Next, Eqs. (1) and (2) are written in an error-correction format so that we can also measure the short-term impacts of the shadow economy. An econometric approach that permits us to estimate the long-term and short-run impacts in a single step is to estimate the following specification:

Table 1

Variable definition and sources

Variable	Symbol	Measure	Source
Carbon dioxide emissions	CO2	CO2 emissions from manufacturing industries and construction (% of total fuel combustion)	WDI
Clean energy	CE	Electric power consumption (kWh per capita)	WDI
Shadow economy	SE	The shadow economy a percentage of total annual GDP	GE
Tax revenue	TAX	Tax revenue (% of GDP)	WDI
Control of corruption	CC	Percentile rank among all countries (ranges from 0 (lowest) to 100 (highest) rank)	WGI

WDI means world development indicators, GE means global economy indicators, WGI means world governance indicators

Equations (3) and (4) are an error-correction model; they contain the linear combination of lagged level variables (Pesaran et al. 2001). Estimation of the OLS method is given the short-run and long-term effects of each variable on the clean energy and environmental pollution are reflected by the coefficient estimates attached to “first-differenced” and “lagged level” variables. To confirm the validity of long-run estimates, Pesaran et al. (2001) suggest two tests. One of them is the F-test to recommend and to determine the joint lagged level of significance for cointegration using the new tabulate, new critical values and the other is the t-test or ECM.

Equations (3) and (4) can only be used to measure the symmetric effects of shadow economy on clean energy consumption and environmental pollution. While the economic implication is that shadow economy changes have an asymmetric manner on clean energy consumption and environmental pollution, therefore, we modify Eqs. (3) and (4) in a new way so that we can assess the possibility of asymmetric effects of shadow economy on clean energy consumption and the environment. To do that, we follow Shin et al. (2014); (SEt) is decomposed into two new time series variables employing the partial sum concept as follows:

where the SE+t (SE−t) variable is the partial sum of positive (negative) changes in the shadow economy. Once SE in Eqs. (3) and (4) is replaced by the two new partial sum variables, the extended models are as follows:

Since the formation of the partial sum variables concept introduced into the models and most commonly referred to as nonlinear ARDL, Shin et al. (2014) revealed that both the symmetric and asymmetric ARDL models are subject to similar estimation techniques and diagnostic tests. After estimating the models, we can apply three asymmetry assumptions. First, short-run adjustment asymmetry will be established if ∆SE+t − k and ∆SE−t − k take a dissimilar lag order. Second, if the coefficient estimate of similar lags attached to ∆SE+t − k is different than the estimate of coefficients attached to ∆SE−t − k , short-run asymmetric effects of shadow economy changes will be established. However, in the short run, stronger asymmetric effects will be established, if we nullified the null of ∑δ2p = ∑δ3p and∑γ2p=∑γ3p. Finally, in the long run, stronger asymmetric effects will be established, if we nullified the null of = and = and these applications were confirmed through the Wald test (for some other econometric applications of NARDL, see Ullah et al. (2020), Ullah and Ozturk (2020), and Usman et al. (2020)).

Results and discussions

In this section, the first step is to decide the maximum number of lags we need to apply. As our data is annual following the literature, we have imposed a limit of a maximum of three lags. To choose a suitable number of lags out of the maximum, we have used the Akaike information criterion. Moreover, pre-unit testing is not a mandatory condition for applying ARDL, as the macroeconomic variables become stationary after taking the first difference. However, for our satisfaction, we have confirmed that all our variables are either I (0) or I (1) by relying on the augmented Dickey-Fuller (ADF) unit root test. Nevertheless, to avoid space, the results of structural breaks on the data are not provided and are available on the demand of readers.

Table 2 reports the results of both the linear models, i.e., clean energy model and environmental model. In the short run, the estimated coefficient of the shadow economy is significant in five out of three countries, Pakistan, India, and Sri Lanka. The signs of the estimates attached to the shadow economy confer that the informal economy increased clean energy consumption in Pakistan and Sri Lanka, whereas, clean energy consumption is negatively affected by the shadow economy in India. If we see the impact of the shadow economy in the CO2 model, we observe that in the case of India, Sri Lanka, and Nepal, underground economy reduces CO2 emissions, while, in the case of Pakistan, informal economy does not have a noticeable impact in polluting the environment at a current lag. However, at previous lag, the informal economy was contributing significantly to polluting the environment which suggests that over time, the informal economy’s contribution to environmental degradation has turned to minuscule and eventually insignificant.

Table 2

ARDL estimates

	Pakistan		India		Bangladesh		Sri Lanka		Nepal
	CE	CO2	CE	CO2	CE	CO2	CE	CO2	CE	CO2
Panel A: Short-run estimates
ΔSEt	0.006**	0.178	−0.012**	−0.490*	−0.005	−0.252	0.002**	−0.654**	0.001	−0.589**
	(0.002)	(0.276)	(0.004)	(0.295)	(0.005	(0.390)	(0.001)	(0.333)	(0.008)	(0.234)
ΔSEt − 1		0.639*							−0.002
		(0.353)							(0.007)
ΔSEt − 2		−0.952**							0.009
		(0.254)							(0.005)
ΔTAXt	−0.004	−0.249	−0.008	−0.758*	0.011	0.878	−0.007**	−1.200	−0.003	−0.079
	(0.003)	(0.262)	(0.006)	(0.422)	(0.016)	(1.707)	(0.003)	(0.950)	(0.005)	0.520)
ΔTAXt − 1								1.450	0.017
								(0.722)	(0.013)
ΔTAXt − 2									−0.034
									(0.016)
ΔCCt	0.002**	−0.235	−0.007*	−1.041**	−0.001	0.676	0.026	−0.061	0.001	0.989**
	(0.001)	(0.324)	(0.004)	0.266)	(0.006)	(0.591)	(0.018)	(0.617)	(0.008)	(0.425)
ΔCCt − 1				−0.678**		−1.444**			−0.003
				(0.284)		(0.756)			(0.010)
ΔCCt − 2						1.316**			−0.019**
						(0.624)			(0.009)
Panel B: Long-run estimates
SE	0.003	1.665**	−0.018*	1.226	0.026	−0.200	0.013	−0.667**	−0.010	−0.477**
	(0.012)	(0.786)	(0.011)	(1.384)	(0.023)	(0.333)	(0.012)	(0.325)	(0.008)	(0.253)
TAX	−0.004	0.476	0.016	1.778	0.177**	0.698	−0.047**	1.217**	0.037**	−0.063
	(0.003)	(0.883)	(0.030)	(1.988)	(0.056)	(1.258)	(0.019)	(0.546)	(0.012)	(0.419)
CC	0.002**	1.102**	0.010**	0.874	0.003	−0.039	0.026	−0.058	0.021**	0.801**
	(0.001)	(0.526)	(0.005)	1.079	(0.003)	(0.098)	(0.018)	(0.595)	(0.003)	(0.245)
C	2.800**	−44.67**	2.835**	−30.26	−0.216	18.05	2.351**	21.32	1.457**	37.51**
	(0.581)	(6.824)	(0.577)	(59.66)	(1.144)	(19.71)	(0.793)	(22.99)	(0.424)	(12.70)
Panel C: Diagnostic tests
Adj R2	0.986	0.921	0.985	0.976	0.943	0.99	0.998	0.994	0.93	0.954
F-test	1.217	7.419	3.216	6.879	11.696	6.251	2.247	3.615	2.747	4.045
ECMt-1	−0.178	−0.350**	−0.240**	−0.236	−0.436**	−1.258**	−0.155**	−1.045**	−0.608**	−1.235**
	(0.097)	(0.111)	(0.127)	0.156)	(0.156)	(0.156)	(0.051)	(0.275)	(0.199)	(0.305)
LM	0.291	0.018	0.269	1.155	0.736	1.964	0.278	5.296	0.050	0.218
RESET	1.415	1.082	0.285	1.411	0.322	2.249	0.095	1.802	0.040	1.104
CUSUM	S	S	S	S	S	S	S	S	S	S
CUSUMQ	S	US	US	S	S	S	S	US	US	S

Numbers inside the brackets are t-ratio and level of significance is 10% (5%) at the 1.64 (1.96). LM and RESET critical values are significant 10% (5%) at 2.71 (3.84)

* indicates significance at the 10% level and ** at the 5% level

In the long run, shadow economy does not have any visible impact on clean energy consumption in all countries except India. In India, clean energy consumption is negatively affected by the shadow economy which implies that the informal economy has either no access to green energy sources or the informal sector is not answerable to any environmental regulations or responsible enough and hence consuming conventional sources of energy. On the other side, the informal economic sector contributed significantly to pollute the environment of Pakistan. The probable reason could be the large size of the informal sector of Pakistan which is involved in various types of economic activities that contribute to the environmental degradation of Pakistan. In other South Asian economies like Bangladesh and Nepal, the shadow economy exerts a negative influence on CO2 emissions conferring that the informal sector is involved in the production of smart and green products which improve the environmental quality of these countries. Whereas, in the context of India and Sri Lanka, the impact of the shadow economy is insignificant in the long run.

Next, the long-run effects of tax revenue, in the clean energy model, appeared to be significant and positive in Sri Lanka and significant and negative in Nepal and Bangladesh. These findings suggest that in Sri Lanka, the tax revenue collection is spent on the production of clean energy projects which increases the clean energy consumption eventually, whereas the tax revenue collection in Bangladesh and Nepal does not contribute to the production of green energy resources which will reduce the consumption of renewable energy and increase the consumption of non-renewable energy consumption. In the CO2 model, the tax revenue collection applies a positive and significant impact on CO2 emissions only in Sri Lanka implying that the revenue collection increases due to a rise in the economic activities in the country which deteriorates the environmental quality.

Control of corruption is another control variable included in our analysis which represents the institutional quality in South Asian economies. The estimated coefficients of the control of corruption variable, in the clean energy model, are significantly positive in Pakistan, India, and Nepal. This result suggests that better institutional quality will pave the way for the consumption of clean energy in South Asian economies.

The long-run results are valid only if we can approve co-integration among them through any of the co-integration tests, i.e., F-test of joint significance of the long-run estimates or the error correction modelling (ECMt-1). From panel C, we collect that co-integration is confirmed in all the countries except Pakistan in the clean energy model. To further check the reliability of our estimates, we have executed some other diagnostic tests: Langrage multiplier (LM) test of serial correlation, Ramsey’s RESET test of misspecification, CUSUM and CUSUMSQ tests of the instability of the parameters1. From panel C of Table 2, we confirm that no serial correlation or misspecification is found in our models; besides, the parameters are stable.

In Table 3, we have provided the nonlinear estimates of both models. In the short run, in the clean energy model, the positive part of the shadow economy (∆SE+) exerts a positive and significant influence only in the case of Pakistan and in all other countries, the estimates attached to the positive component of the shadow economy appeared to be insignificant.

Table 3

NARDL estimates

	Pakistan		India		Bangladesh		Sri Lanka		Nepal
	CE	CO2	CE	CO2	CE	CO2	CE	CO2	CE	CO2
Panel A: Short-run estimates
ΔSEt+	0.028**	0.397**	0.004	0.103	0.007	0.699	−0.008	−0.634**	0.025	0.786**
	(0.011)	(0.209)	(0.009)	(0.937)	(0.012)	(1.272)	(0.007)	(0.334)	(0.026)	(0.262)
ΔSEt − 1+	0.027			−1.519	0.032		0.003
	(0.018)			(0.829)	(0.017)		(0.011)
ΔSEt − 2+	−0.038				−0.021		−0.010
	(0.013)				(0.015)		(0.006)
ΔSEt−	−0.023**	−0.476	−0.023**	−0.997*	−0.029**	−0.800**	0.017**	0.565**	0.012	−0.784**
	(0.007)	(0.503)	(0.007)	(0.562)	(0009)	(0.388)	(0.006)	(0.235)	(0.009)	(0.384)
ΔSEt − 1−		−0.194		0.132	−0.020**		−0.009
		(0.517)		(0.639)	(0.008)		(0.007)
ΔSEt − 2−		−0.384			0.011*		−0.006
		(0.498)			(0.006)		(0.005)
ΔTAXt	−0.001	−0.185	−0.013**	−0.329	−0.013	−1.793	0.007	0.816	−0.004	2.389**
	(0.007)	(0.287)	(0.006)	(0.656)	(0.020)	(1.650)	(0.006)	(0.690)	(0.007)	(0.833)
ΔTAXt − 1	0.018**	0.522**		−0.880	−0.052**		0.024**	−1.200**	0.017**	−1.946**
	(0.008)	(0.283)		(0.573)	(0.025)		(0.007)	(0.950)	(0.016)	(0.903)
ΔTAXt − 2	−0.008				0.021		−0.016	1.450**	−0.028**	−0.720
	(0.006)				(0.019)		(0.012)	(0.722)	(0.010)	(0.941)
ΔCCt	−0.010**	0.300**	−0.008	−1.095	0.006	0.238	0.015*	−0.061	0.003	−1.060*
	(0.005)	(0.150)	(0.004)	(0.321)	(0.005)	(0.246)	(0.008)	(0.617)	(0.008)	(0.566)
ΔCCt − 1	0.009*			−0.581*			−0.005	−0.879*	0.003
	(0.005)			(0.395)			(0.005)	(0.583)	(0.010)
ΔCCt − 2	−0.010*			−0.253			−0.007		−0.014**
	(0.005)			(1.944)			(0.006)		(0.008)
Panel B: Long-run estimates
SE+	0.024**	1.768*	0.012	0.002	0.029	−2.441**	0.134	−0.607*	0.043	0.328
	(0.011)	(1.034)	(0.022)	(0.744)	(0.010)	(1.020)	(0.137)	(0.325)	(0.047)	(0.280)
SE−	0.004	0.750	−0.018	1.089*	−0.012	−1.095**	0.083	0.922	0.003	−0.370**
	(0.007)	(0.768)	(0.006)	(0.682)	(0.006)	(0.597)	(0.098)	(0.987)	(0.010)	(0.169)
TAX	−0.001	1.159	−0.008	0.085	0.066**	−2.453	−0.010	1.217**	0.020	2.679**
	(0.014)	(1.449)	(0.016)	(0.530)	(0.026)	(2.424)	(0.063)	(0.546)	(0.025)	(0.346)
CC	0.077**	1.235**	0.002	1.006**	0.020**	0.326	0.095	−0.058	0.015	0.640**
	(0.005)	(0.708)	(0.005)	(0.033)	(0.002)	(0.316)	(0.123)	(0.595)	(0.013)	(0.260)
C	2.572	0.971	2.413	−15.27**	1.346	31.666	2.148	21.312	1.484	−41.737
	(0.200)	(21.383)	(0.138)	(4.850)	(0.173)	(15.891)	(1.511)	(22.995)	(0.117)	(7.383)
Panel C: Diagnostic tests
Adj R2	0.938	0.920	0.975	0.996	0.987	0.994	0.993	0.998	0.861	0.786
ECMt-1	−0.713**	−0.290**	−0.393**	−0.306	−0.314*	−0.730**	−0.175	−0.545**	−0.591**	−0.717**
	(0.225)	(0.110)	(0.179)	(0.211)	(0.178)	(0.193)	(0.239)	(0.275)	(0.195)	(0.416)
F-test	3.798	4.489	2.338	4.447	6.037	3.038	2.447	3.615	7.098	4.812
LM	0.383	0.298	0.208	1.030	1.887	0.600	1.727	5.296	0.409	0.963
RESET	0.022	0.635	0.0749	0.114	1.079	0.384	1.533	0.872	0.575	0.010
CUSUM	S	S	S	S	S	US	S	S	S	S
CUSUMQ	S	S	S	S	S	US	S	S	US	S
Wald-SR	8.098**	6.134**	2.654	9.654**	8.087**	5.022**	11.05**	4.002**	2.010	5.901**
Wald-LR	4.043**	9.021**	3.012*	8.103**	2.021	2.013	6.809**	7.020**	4.076**	9.044**

Numbers inside the brackets are t-ratio and level of significance is 10% (5%) at the 1.64 (1.96). Wald, LM, and RESET critical values are significant 10% (5%) at 2.71 (3.84)

* indicates significance at the 10% level and ** at the 5% level

These findings suggest that with the growth of the shadow economy, clean energy consumption rises in Pakistan, while remaining insignificant in all other countries. However, the estimates attached to a negative component of the shadow economy (∆SE−) appeared to be negatively significant in the case of all countries except Sri Lanka and Nepal. In the case of Sri Lanka and Nepal, the estimated coefficient of (∆SE−) is positively significant and insignificant, respectively. According to this finding, if there is a decrease in the volume of the shadow economy, the consumption of clean energy will rise in Pakistan, India, and Bangladesh, while the clean energy consumption will reduce in Sri Lanka. Likewise, the short-run impacts of ∆SE+ are found to be detrimental to the environmental status of Pakistan and Nepal but proved to be environment-friendly in the case of Sri Lanka. Conversely, the short-run effects of ∆SE− increase the CO2 emissions in India, Bangladesh, and Nepal and decrease CO2 emissions in Pakistan. Afterwards, we want to see whether the positive and negative components of the shadow economy follow the symmetric or asymmetric path. First, we check the lag length attached with positive and negative parts of the shadow economy in both models. In both models, this type of asymmetry, i.e., adjustment asymmetry, is confirmed in the context of Pakistan only as the lag length attached to both positive and negative parts is different. Moreover, from panel C of Table 3, we see that the Wald-SR statistic is significant in Pakistan, Sri Lanka, and Bangladesh, in the clean energy model. In the CO2 model, the same statistic is significant in all countries. Significant Wald-SR is a validation of impact asymmetry which is defined as if the sum of estimates attached to (∆SE+) is different from the sum of estimates attached to (∆SE−). After that, we need to see whether these short-run estimates remain significant in the long run or not.

From panel B, we see that the estimate attached to PSE is significant and positive in Pakistan in the clean energy model, conferring that the positive change in the informal economy increases clean energy consumption in Pakistan. Similarly, in the CO2 model, the positive change in the shadow economy increases CO2 emissions in Pakistan and reduces the emissions in Bangladesh and Sri Lanka. This finding is consistent with Huynh (2020), who noted that the shadow economy has a positive effect on air pollution in developing Asian countries. While Bangladesh and Sri Lanka economies results are nullified, the study of Biswas et al. (2012) noted that large informal sector may be attended by higher environmental pollution levels.

On the other hand, the negative change in the shadow economy does not have any noticeable impact on clean energy consumption in any country. However, the decreased volume of the informal sector of the economy reduces CO2 emissions in India and increases CO2 emissions in Bangladesh and Nepal. These findings imply that in Bangladesh and Nepal, the informal sector is more responsible with regard to environmental regulations or may involve in the economic activities that are conducive to the environment. However, in India, the scenario is the opposite and the reduction in the size of the underground economy improves the environmental position of the country. Other long-run variables like tax revenue and control of corruption behave in the same way as in our linear model, hence, do not require any further elaborations.

Once again, the nonlinear estimates are valid only if the cointegration is established between them. To that end, we check the estimates of F-test and ECMt-1 which confirm the co-integration in all countries except Sri Lanka, in the clean energy model. Then, the long-run asymmetry is observed through significant WALD-LR statistic in all the countries, in both the models, except Bangladesh. Few other diagnostic tests, just like linear models, are also reported in panel C, which confirms that nonlinear models are autocorrelation free, specified correctly, and stable.

Conclusion and policy implications

The objective of this study is to explore the dynamic effects of shadow economy on clean energy and air pollution for South Asian economies from 1991 to 2019. The empirical estimates are drawn using both linear and nonlinear ARDL approaches. The short-run ARDL results suggest that shadow economy increases clean energy consumption in Pakistan and Sri Lanka, whereas this effect is negative for India and insignificant for other countries. The long-run ARDL results do not show visible impact in all countries except India where shadow economy adversely affects clean energy consumption. This finding suggests that the informal sector relies on conventional energy sources and environmental regulations are not effective in this sector. The results for the second model confer that the informal sector increases emissions in Pakistan and has insignificant effects in India and Sri Lanka. Conversely, it decreases emissions in Bangladesh and Nepal suggesting that these economies utilize smart and green methods of production in the informal sector.

The long-run effects of tax revenue on clean energy are positively significant in Sri Lanka while negatively significant in Nepal and Bangladesh. These findings imply that tax collections in Sri Lanka are used to accommodate clean energy projects whereas in Bangladesh and Nepal, these collections are not diverted towards clean sources of energy. In the pollution model, tax revenue has a positive and significant effect on emission only in Sri Lanka suggesting that revenues coming from increased economic activities also deteriorate the environmental quality. Institutional quality significantly increases clean energy in Pakistan, India, and Nepal implying that better institutional setup can pave the way for the consumption of clean energy in South Asian economies. However, in the case of Pakistan and Nepal, institutional quality deteriorated the environmental quality.

The nonlinear estimates for the clean energy model suggest that the positive components of the shadow economy significantly increase clean energy consumption only in Pakistan while their impacts are insignificant in the remaining countries. However, the negative components of the shadow economy are negatively significant in all countries except Sri Lanka and Nepal. Similarly, the short-run effects of shadow economy are found to be detrimental to the environmental status of Pakistan and Nepal but proved to be environment-friendly in the case of Sri Lanka. Conversely, the short-run effects of shadow economy increase the CO2 emissions in India, Bangladesh, and Nepal and decrease CO2 emissions in Pakistan.

The long-run asymmetric estimates suggest that an increase in positive component increases clean energy consumption in Pakistan. Likewise, the positive change in the shadow economy increases CO2 emissions in Pakistan and reduces the emissions in Bangladesh and Sri Lanka, whereas the negative components of shadow economy do not have any noticeable impact on clean energy in any country. However, the negative component of the informal sector of the economy reduces CO2 emissions in India and increases CO2 emissions in Bangladesh and Nepal. These findings imply that in Bangladesh and Nepal, the informal sector is more responsible with regard to environmental regulations or may involve in the economic activities that are conducive to the environment. However, in India, the scenario is the opposite and the reduction in the size of the underground economy improves the environmental position of the country. “Especially, governments in developing countries should allocate more budgets on environmental projects in their fiscal reforms for the sake of moving to greener and more inclusive economies with low-carbon activities.”

Overall findings establish the dynamic relationships among shadow economy, clean energy, and air pollution and offer diverse policy implications. First, clean energy practices in the informal sector need to be encouraged by increasing awareness and facilitating the provision of renewable energy sources in South Asian countries in general, and in India, in particular, where this problem is more assertive. Fiscal policy instruments need to be aligned with environmental reforms as the results have established the importance of tax revenue collection for the environment. Particularly, governments in South Asian economies can spare more funds for eco-friendly projects in their fiscal reforms for sustainable development. Finally, institutional quality needs to be internalized in the macroeconomic policy framework to preserve the environment. Similar studies can be conducted for other economies by using the nonlinear ARDL approach.