How does fiscal decentralisation affect carbon footprints? The role of renewable energy intensity: Evidence from the top 10 highly decentralised OECD countries

Abstract

This study investigates the dynamic relationship between fiscal decentralization (FD), renewable energy intensity (REI), and carbon footprints (CFs) in the ten most highly decentralized OECD countries. Employing the cross-sectional autoregressive distributed lag (CS-ARDL) bound test, the analysis captures both short- and long-term effects of FD and its key determinants. Additionally, the Dumitrescu–Hurlin causality test examines the directional causal relationships among these variables. The findings indicate that FD serves as an effective mechanism for mitigating CFs by enhancing environmental quality. Consequently, local governments in highly decentralized regions should prioritize adopting renewable energy sources. Furthermore, carbon reduction initiatives contribute to energy efficiency and emission reduction targets. A statistically significant negative interaction between REI and CFs suggests that countries with high FD but low REI do not achieve substantial environmental benefits. The Dumitrescu–Hurlin causality test reveals a unidirectional causal relationship from revenue decentralization (REVD), expenditure decentralization (EXPD), composite fiscal decentralization (CFD), REI, gross domestic product (GDP), population size, and environment-friendly technological innovation (EnFTI) to CFs. Despite its contributions, this study has certain limitations. The analysis is restricted to highly decentralized OECD countries, which limits the generalizability of the findings. Additionally, while the CS-ARDL model captures short- and long-term dynamics, it does not account for potential spatial dependencies that may influence environmental outcomes. Future research could incorporate spatial econometric techniques or case-specific studies to enhance robustness.

Keywords

Carbon footprints fiscal decentralisation renewable energy intensity CS-ARDL Dumitrescu–Hurlin causality

Introduction

Decentralisation is a long-term progressive process involving different parties under an acceptable legal framework. Since the early 1990s, most nations have followed a decentralisation process that combines political, administrative, and fiscal facets. In a decentralisation process, each of the above elements must be present. Regarding government control over taxation and spending, fiscal decentralisation typically refers to the transfer of authority from central government agencies to sub-local government agencies (e.g., regions, provinces, and cities). Every decentralised system delegates significant powers to local governments, allows them to raise money through taxation, and provides a robust tax base. Local government efficiency, including inclusive growth and improved equity, is often attributed to resource allocation and financial management at lower administrative levels.¹ As the world grapples with the urgent need to address climate change, governments increasingly seek ways to reduce carbon footprints (CFs) and promote sustainable energy. One approach gaining attention is fiscal decentralization (FD), where local governments are given more control over decision-making and finances, allowing them to tailor policies to their specific needs. While there is plenty of research on renewable energy adoption and its environmental benefits, the impact of FD on renewable energy use and carbon emissions has not been fully explored. This study aims to fill that gap by examining how FD can influence local energy policies and contribute to reducing CFs, offering new insights into how decentralization can drive more sustainable practices.

Efficient energy-saving and emission-reduction tools, such as carbon taxes, emission trading schemes, and government expenditures on emission reduction, have been adopted by major nations to combat global warming and greenhouse gas emissions.^2,3 The destruction of the environment is one of the world's most pressing issues, requiring urgent attention. It has garnered significant interest due to its impact on human existence and biodiversity. According to Hao et al.,⁴ one of humankind's top priorities is achieving green growth. Therefore, countries worldwide are working hard to minimize greenhouse gas emissions.^5,6 It is generally recognised that growing global trade and commercial activity have increased energy demand, which is the main factor contributing to environmental degradation.⁷ The demand for fossil fuels has been rising due to increased global production. The Earth's average temperature has increased by 1.9°C because of excessive fossil fuel usage, which harms biodiversity and human existence.⁸

The objectives of this paper are to examine the effect of FD on CFs and to explore the interactive effect of renewable energy consumption (REC) on CFs. To achieve these objectives, we utilize annual data from the ten most highly decentralized OECD countries. Furthermore, we incorporate an interaction term for renewable energy intensity (REI) to evaluate the overall impact of decentralization on CFs. Renewable energy intensity is calculated by dividing the amount of renewable energy capacity by global gross domestic product (GDP), as suggested by Shahzad and Fareed⁹ and Shahzad et al.¹⁰

Most existing literature uses REI as a proxy for renewable energy usage rather than measuring energy consumption per person. Another objective is to explore the effects of FD and the interaction between the intensity of renewable energy on CFs by applying the cross-sectional autoregressive distributed lag (CS-ARDL) technique. The Dumitrescu–Hurlin causality test was also used to investigate the causalities. The main contribution of this study is its examination of the relationship between FD, REI, and CFs in highly decentralized OECD countries. By integrating these factors, this research provides new insights into how FD can influence long-term environmental outcomes. Using the CS-ARDL model and the Dumitrescu–Hurlin causality test, this study sheds light on the impact of decentralization on carbon emissions and renewable energy adoption. It also addresses a significant gap by demonstrating how fiscal decentralization can serve as an effective policy tool for promoting sustainable energy transitions and reducing CFs.

According to our estimations, FD could be used as a proactive measure to reduce carbon emissions. Empirical results show that FD enhances environmental quality. Local governments in highly decentralised areas support measures to use renewable energy to reduce carbon emissions. In terms of lowering CFs, the intensity of renewable energy is complementary to FD and, if fully and effectively utilised, may improve environmental quality. Wang et al.⁶⁴ critically examine the Environmental Kuznets Curve hypothesis by exploring how economic, institutional, and technological factors influence environmental outcomes across 214 countries. Their study highlights the role of decentralized fiscal policies and technological advancements, such as REI, in reducing CFs and fostering sustainable development. Institutional support for renewable energy initiatives guarantees the “race to the top” sparked by expanding fiscal autonomy. Individuals with significant political influence band together in groups with similar environmental concerns and use their influence to inform the public about the need to protect the environment by supporting renewable energy projects.

This article contributes to the literature on CFs in three ways. First, following the previous literature, we examine the relationship between carbon emissions and FD. Second, we use three proxies for FD (revenue decentralisation (REVD), expenditure decentralisation (EXPD), and composite fiscal decentralisation (CFD)) to explore the overall impact on CFs. Third, we used REI as the interaction term with all FD proxies to examine the intermediate effect of REI on CFs. To our knowledge, this is the first international investigation of the empirical link between FD, REI, and CFs in the top ten highly decentralised OECD countries.

According to the Sustainable Development Goals (SDGs) (https://sdgs.un.org/goals), nations should adopt renewable energy policies to combat the environmental degradation brought on by the rapid expansion of non-renewable energy. Although early adoption of renewable energy sources was insufficient to address the problem of environmental deterioration, countries must implement energy policies that seek to achieve the sustained objective of improved environmental quality and sustainable growth. Technological innovation is the fundamental principle underpinning a country's attempts to promote R&D to develop unconventional energy sources. International trade can satisfy this requirement by enhancing domestic capabilities or transferring technologies from external sources.¹¹

The rest of the paper is organised as follows: Literature review section reviews the literature relevant to the paper. Data, model specifications, and theoretical framework section explains the data used in the empirical analysis, model specifications, and theoretical framework. Empirical methodology section presents the empirical methodology. Empirical methodology section discusses the results. Conclusion section concludes the paper with several policy implications.

Literature review

Extensive research has investigated the relationship between FD and environmental quality, focusing on both its role in environmental degradation and its potential benefits.^12–19 The existing literature can be categorized into two primary perspectives. One stream of research supports the notion that FD improves environmental quality, with Fredriksson et al.,¹⁴ Levinson,²⁰ and Konisky¹⁵ among the prominent scholars advocating this view. Konisky¹⁵ contends that a higher degree of FD is critical for fostering environmental improvements. Additionally, Cheng et al.⁸ emphasize the importance of clearly delineating governmental responsibilities across different administrative levels to ensure that fiscal policies effectively contribute to reducing CO₂ emissions and promoting energy efficiency.

According to Almarshad,²¹ FD increases local governments’ ardour for environmental stewardship. Fiscal decentralisation provides local governments with significant financial assistance in lowering environmental pollution;^22,23

According to Li et al.,²⁴ the decentralisation of government finances and the growth of technology would increase China's total environmental efficiency. Global FD positively affects energy consumption and carbon emissions.²⁵ He²⁶ proposes using FD to enhance environmental conservation. Liu et al.²⁷ claim that FD of local units has a small but positive feedback effect on environmental degradation and a significant negative contribution. Fiscal decentralisation also improves the condition of the ecology.²⁸ Despite the increasing attention toward FD, empirical research examining its influence on CF reduction and the enhancement of REI remains limited. Although the individual impacts of decentralization and renewable energy adoption have been widely investigated, their interactive effects on environmental sustainability have received comparatively little scholarly focus. This study addresses this gap by investigating the extent to which FD influences local energy policies and carbon emissions. In doing so, it provides novel insights into the potential of decentralization as a policy instrument to mitigate CFs and accelerate the transition to renewable energy sources.

The researchers who are more concerned about the role of FD in degrading environmental quality include^12,13,16,17 For instance, according to Millimet,¹⁶ due to weaker local environmental regulations, countries compromise on environmental quality due to the higher power of the lower units of the state. Similarly, Sigman¹⁷ states that as the degree of FD increases, free-rider conduct among jurisdictions degrades environmental quality. Yavuz et al.²⁹ introduced the Environmental Phillips Curve (EPC) in Turkey, highlighting how economic decentralization influences environmental quality. While local policy control can improve sustainability, GDP growth and resource dependence increase CFs. They emphasize FD as a key enabler for renewable energy adoption, allowing local governments to implement cleaner energy strategies and mitigate environmental impacts.

Furthermore, local governments rely heavily on land auctions under FD to acquire sizable land concessions. As a result, the real estate industry grows quickly, and local governments are encouraged to use land concessions to build public infrastructure to show their performance on their own. However, the construction of both real estate and infrastructure will cause a significant increase in carbon emissions. Decentralisation will lessen the control effect of local governments on carbon emissions since carbon emission is a pollutant with strong spillover effects and no clear short-term harm.^22,30 Caglar et al.²² investigated the impact of municipal solid waste, REC, human capital, and natural resources on environmental quality in European Union countries. Their findings highlighted the positive effects of human capital and renewable energy on enhancing environmental quality, while emphasizing the need for improved waste conversion practices. Wang et al. (2024) investigate how artificial intelligence (AI) can promote energy transitions and reduce carbon emissions, noting that trade openness plays a critical role in enhancing these outcomes. Their research provides valuable insights into how decentralization and technological advancements, such as AI, can drive renewable energy adoption and lower CFs, complementing the effects of FD on environmental sustainability.

A range of governance methods significantly influences a country's carbon dioxide emission levels.¹⁹ There is an inverse U-shaped link between FD and the effectiveness of the government in providing these services.³¹ According to Wang and Lei,³² decentralising fiscal income may minimise pollution through optimization of regional industrial structures. Wang et al. (2024) examine the relationship between ecological footprints, carbon emissions, and energy transitions, highlighting the potential role of AI in addressing environmental challenges. Their study emphasizes how AI-driven solutions can contribute to reducing carbon emissions while facilitating sustainable energy transitions.

The research findings of recent studies are mixed. Researchers may employ various measurement indicators and empirical models to explain this inconsistency. Many studies have used outdated spatial econometric models for their research, ignoring the interaction effect of renewable energy. By adding a new variable that acts as a mediator for the impact of FD on carbon emissions from the perspective of CS-ARDL, our study addresses this gap in the literature by incorporating the interaction terms for REI. Additionally, this research seeks to confirm the contentious findings on the existence of the “free-riding” problem for governments brought on by FD.

In conclusion, the existing literature presents conflicting findings on the relationship between FD and CO2 emissions, with various studies highlighting different outcomes. Our study enhances the current body of knowledge by offering a comprehensive analysis of how FD (measured through REVD, EXPD, and CFD), REI, and other control variables influence CFs. Specifically, we investigate the individual impacts of REVD, EXPD, CFD, REI, and environment-friendly technological innovation (EnFTI) on CF, while also exploring the interactions between these variables (including REVD*REI, EXPD*REI, and CFD*REI). By doing so, our study provides a more nuanced understanding of how FD and REI interact to shape environmental outcomes, contributing valuable insights to both academic research and policy discussions on sustainable governance and environmental management.

Data, model specifications, and theoretical framework

Data sources

Data from the top ten highly decentralised nations were used to investigate the influence of FD and REI on CFs (see Figure 1). Annual data for these nations were collected for the period 1990 to 2022. According to 2022 statistics, these nations had the highest CD index scores, ranging from 0.98 to 0.4 (highest to lowest). Table 1 shows the details of data sources and measurement units. All values except for the CD index are converted to logarithms to overcome the residual problems. Figure 1 shows the OECD country-wise ranking of decentralisation. Table A1 presents the fiscal decentralization scores for the top 10 most decentralized OECD countries. In this study, we focus on the top ten highly decentralized OECD nations, selected for their significant FD, where local governments play a critical role in shaping environmental and energy policies. Analyzing these economies provides valuable insights into how FD influences CFs and REI. OECD countries were chosen due to their comparable economic structures, regulatory frameworks, and institutional settings, ensuring consistency across the sample. This selection also addresses a gap in the existing literature, which primarily focuses on broader country samples and lacks specific examination of highly decentralized economies. The data timeframe of 1990 to 2022 is chosen to capture three decades of FD reforms, renewable energy transitions, and climate policy changes in OECD countries. This period allows for the analysis of both short- and long-term effects on CFs. Additionally, the timeframe aligns with key global environmental milestones, such as the Kyoto Protocol baseline emissions data starting in 1990 and the more recent Paris Agreement and renewable energy initiatives. The extensive dataset spanning over 30 years ensures empirical robustness, mitigating the influence of short-term fluctuations and providing statistically reliable and policy-relevant conclusions. The descriptive statistics of the variables are presented in Table 2. The table includes the mean, standard deviation, minimum, and maximum values, which help to summarize the central tendency and variability of the data. These results provide useful insights into the overall characteristics of the sample and confirm that the data are appropriate for subsequent regression analysis.

Figure 1.

Fiscal decentralized OECD countries' rank. REVD: revenue decentralisation, EXPD: expenditure decentralisation, and CFD: composite fiscal decentralization.

Table 1.

Data description and sources.

Variables	Description	Source
CF	Carbon Footprint	CSMAR
REVD	Revenue Decentralisation	Self-Calculation from OECD data
EXPD	Expenditure Decentralisation	Self-Calculation from OECD data
CFD	Composite Fiscal Decentralization (Index)	Self-Calculation from OECD data
REI	Renewable Energy Intensity	BP Statistics
GDP	Gross Domestic Product (USD Constant 2010)	World Development Indicators
PH	Population Headcount (In Millions)	Pen World Tables
EnFTI	Environment-Friendly Technological Innovation	Self-Calculation from OECD data

Table 2.

Descriptive statistics.

Variable	Obs.	Mean	Max	Min	SD
CF	330	1.209	1.287	0.522	0.182
REVD	330	0.355	0.577	0.128	0.243
EXPD	330	0.515	0.846	0.122	0.158
CFD	330	0.573	0.957	0.242	0.289
REI	330	1.326	1.847	−0.548	0.747
GDP	330	4.845	4.847	3.789	0.325
PH	330	7.325	8.458	5.847	0.958
EnFTI	330	6.148	9.258	4.985	0.947

Note: Variables are in log10 except ED, RD, and CD indexes. CF: carbon footprint; REVD: revenue decentralization; EXPD: expenditure decentralization; CFD: composite fiscal decentralization (index); REI: renewable energy intensity; GDP: gross domestic product; PH: population headcount (in millions); EnFTI: environment-friendly technological innovation; CD: cross-sectional dependency.

There are two types of FD: REVD and EXPD. Revenue decentralisation is determined by dividing the lower tier or provincial government revenue by the total federal government revenue.³³ To investigate the function of FD, an effective measure that encompasses its most essential aspects is established, as shown in the equations below:³⁴

R E V D = \frac{P R}{P R + F R}

(1)

PR and FR stand for provincial and federal revenue, respectively. REVD represents revenue decentralisation in equation (1). In contrast, EXPD is the relative fraction of provincial government expenditures to total national government expenditures in equation (2). It is calculated using the same process described above by³⁴

E X P D = \frac{P E}{P E + F E}

(2)

PE and FE are provincial and federal spending, respectively, whereas REVD and EXPD are single-dimensional measures. However, FD is a multidimensional process requiring multidimensional measurements to capture reality. Following the Martinez-Vazquez and Timofeev technique, a composite indicator of FD is used, representing the multifaceted element of decentralisation (i.e., REVD & EXPD).³⁵ We construct a CFD indicator as follows:

C F D = \frac{R E V D}{1 - E X P D}

(3)

The composite indicator ranges from 0 to 1.

Model specifications and theoretical framework

There are two schools of thought regarding the nature of the relationship between FD and environmental quality: a “race to the bottom” and a “race to the top.” According to the “race to the top” phenomenon, a high level of FD prevents environmental degradation since lower levels of the state compete with one another.^1,^36,³⁷

On the other hand, the “race to the bottom” phenomenon demonstrates that the quality of the environment diminishes due to the rise in FD. Theoretically, local governments loosen their environmental laws to make more room for economic expansion, which raises CO₂ emissions. This explains why FD and environmental damage correlate positively (Zhang et al., 2017). Fiscal decentralisation may cause a “race to the bottom” in which local governments lower their environmental requirements to attract foreign investors, increasing CO₂ emissions. As the intensity of renewable energy rises, we hypothesise that the link between FD and environmental quality strengthens. Alternatively, a higher REI is required for the “race to the top” to exist. Models 1‒3 are also modified to assess the impacts of FD concerning the intensity of renewable energy and other control variables.

C F_{it} = β_{1} R E V D_{it} + β_{2} R E I_{it} + β_{3} R E V D_{it} * R E I_{it} + β_{4} G D P_{it} + β_{5} P H_{it} + a_{i} + ε_{it}

(4)

C F_{it} = β_{1} E X P D_{it} + β_{2} R E I_{it} + β_{3} E X P D_{it} * R E I_{it} + β_{4} G D P_{it} + β_{5} P H_{it} + a_{i} + ε_{it}

(5)

C F_{it} = β_{1} C F D_{it} + β_{2} R E I_{it} + β_{3} C F D_{it} * R E I_{it} + β_{4} G D P_{it} + β_{5} P H_{it} + a_{i} + ε_{it}

(6)

where CF stands for carbon footprints, REVD stands for revenue decentralisation, and EXPD stands for expenditure decentralisation. CFD denotes composite fiscal decentralisation, REI is the renewable energy intensity, GDP is the gross domestic product, and PH is the population headcount. These variables are frequently employed as drivers of CFs.

Local governments may engage in a “race to the top” approach because of FD by passing stricter environmental regulations that significantly reduce carbon emissions. Consequently, we anticipate that FD will have a negative impact on carbon emissions. On the other hand, FD encompasses decentralised fiscal revenue and spending.³⁸

To enhance the clarity and rigor of our analysis, we have further elaborated on the selection of control variables in our empirical model. These variables, which include GDP growth, population size, and technological innovation, were chosen based on their theoretical and empirical relevance to the relationship between FD, REI, and CFs. Each control variable is justified through its expected influence on the dependent variable, as supported by existing literature.^9,29

Moreover, we have provided a more thorough interpretation of the Dumitrescu–Hurlin causality test results. The causal relationships between FD, REI, and CFs are now clearly delineated, and we have emphasized the economic intuition behind these findings. This ensures that the implications of our results are fully articulated, offering valuable insights for policymakers. The revised discussion clarifies the causal mechanisms at play, particularly how decentralised fiscal structures influence local energy transitions and carbon reduction efforts.

This paper uses an index based on REVD and EXPD to assess the CFD. It reflects the proportion of general government expenditure and revenue spent and collected. We create the FD index using principal component analysis, which is constructed using revenue and expenditure data. Table 1 describes the variables.

Empirical methodology

This study uses the CS-ARDL technique to estimate empirical models (1)–(3). These models perform better when endogeneity exists and when the sample size is small. The models discussed in the preceding section can be used to make predictions. However, before doing so, it is necessary to check for slope homogeneity and cross-sectional dependency (CD), which will likely cause problems when utilising panel data. The long-term links among variables are assessed using the second-generation cointegration approach of Westerlund.³⁹ In addition, the Dumitrescu and Hurlin (D–H) panel causality test is employed to assess the causality among the variables that are included in models (1) and (3). The model in this study is constructed based on FD theory, which emphasizes the role of local governance in shaping environmental policies and energy consumption.^40,41 Fiscal decentralisation allows local governments greater control over regulatory autonomy, budgetary allocation, and policy responsiveness, all of which can influence CFs.⁴² To capture the dynamic relationship between FD, REI, and CFs, we employ the CS-ARDL bound test. This methodology is well-suited for analyzing cross-sectional dependence, capturing both short- and long-term effects, accommodating country-specific heterogeneity, and handling mixed-order integration of macroeconomic variables, ensuring robust and reliable estimation of the relationships in question.^43,44

To enhance the transparency and reproducibility of our study, we have provided a more detailed description of the process used to construct the self-calculated indices for FD and EnFTI. Specifically, we outline the methodological steps involved, from the selection of relevant components to the aggregation techniques used. For the FD index, we considered various dimensions such as revenue autonomy, EXPD, and the decentralised fiscal authority of local governments, drawing from established literature on fiscal governance (Shahzad et al., 2022). The EnFTI index was constructed by identifying key indicators of technological innovation in renewable energy and environmental sustainability, as suggested by prior research on the role of technological progress in reducing CFs.²⁹

We also provide a rationale for the selection of specific components in each index, grounded in economic theory and empirical evidence. For example, the inclusion of renewable energy investment and technological innovation variables in the EnFTI index is supported by their direct impact on carbon emissions reduction.⁴⁵ Furthermore, we describe the aggregation process, detailing whether equal or weighted methods were applied and explaining the robustness checks conducted to ensure the reliability of the constructed indices.

In this study, we initially assumed a direct, linear relationship between FD, REI, and environmental outcomes. However, this approach has been refined to acknowledge the complexities and conditional nature of this relationship, which can vary depending on institutional quality, governance, and local political economies. While FD can enable more tailored environmental policies by empowering local governments, it does not always result in positive outcomes. As suggested by Oates,⁴⁶ decentralization may improve environmental management by bringing government closer to local issues, allowing for more efficient policies. However, more recent literature, such as the work by Banzhaf and Chupp⁴⁷ and Farooqi et al.,⁴⁸ indicates that the effectiveness of decentralization hinges on the strength of local institutions. In regions with weak governance, decentralization can exacerbate fragmentation, inefficiencies, and corruption, undermining environmental goals.

Studies by Jiménez and Pérez⁴⁹ further emphasize that the benefits of decentralization depend on institutional quality, as strong institutions can leverage decentralized structures to foster environmental innovation and accountability. Conversely, weak institutions may hinder these advantages. Additionally, empirical studies (e.g.,⁵⁰) suggest that decentralized policies can be suboptimal if local interests conflict or if there is insufficient coordination. Therefore, we incorporate non-linear considerations into our model to reflect these dynamics, discussing the moderating effects of governance quality, intergovernmental cooperation, and political will. This nuanced perspective ensures that the paper does not over-simplify the impact of FD on environmental outcomes, recognizing that these relationships are often context-dependent.

The following sections explain the models estimated.

Slope homogeneity and cross-sectional dependency tests

Pesaran and Yamagata⁵¹ and Pesaran⁵² tests investigate slope coefficient homogeneity/heterogeneity and cross-sectional dependency. Ignoring the issues of cross-sectional dependency and variability in slope coefficients would result in erroneous and contradictory results.^53,⁵⁴ Therefore, before examining the stationarity features of the data, this study employs the Pesaran and Yamagata⁵¹ test to determine the homogeneity or heterogeneity slope coefficients. Furthermore, the Pesaran⁵² test determines the cross-sectional independence or dependence. The equation for the Pesaran cross-section dependency (CD) test is as follows:

C D_{t e s t} = \sqrt{\frac{2 T}{N (N - 1)}} (\sum_{i = 1}^{N - 1} \sum_{k = i + 1}^{N} {\hat{τ}}_{i k})

(7)

The Pesaran CD test's null hypothesis is that there is no dependency, whereas the alternative hypothesis says that there is independence. The following equations are used to test the slope coefficients:

{\tilde{Δ}}_{H P Y} = {(N)}^{\frac{1}{2}} (2 k)^{\frac{1}{2}} (\frac{1}{N} S - k)

(8)

{\tilde{Δ}}_{A S H} = {(N)}^{\frac{1}{2}} {(\frac{2 k (T - k - 1)}{T + 1})}^{- \frac{1}{2}} (\frac{1}{N} S - 2 k)

(9)

The null and adjusted delta hypotheses support homogeneous slope coefficients, while the alternative hypothesis supports heterogeneous slope coefficients.

Unit root test

We adopt a unique technique termed the cross-sectional augmented Im, Pesaran, and Shin (CIPS) panel unit root test to assess the stationarity of the variables in this study due to possible panel data issues with cross-sectional dependence and slope heterogeneity. The CIPS test equation is as follows:

Δ M_{i, t} = φ_{i} + φ_{i} N_{i, t - 1} + φ_{i} {\bar{M}}_{t - 1} + \sum_{l = 0}^{p} φ_{i l} Δ {\bar{M}}_{t - 1} + \sum_{l = 1}^{p} φ_{i l} Δ M_{i, t - 1} + u_{t}

(10)

The values of the Cross-Sectional Augmented Dickey–Fuller (CSADF) test statistics are used to create the CIPS values. The averages of the cross-sections are denoted as ${\bar{M}}_{t - 1}$ and $Δ {\bar{M}}_{t - 1}$ .

\hat{C I P S} = N^{- 1} \sum_{l = 1}^{n} C S A D F_{i}

(11)

where CSADFi denotes the CSADF test statistic for the i-th cross-sectional unit.

Panel cointegration test

The cointegration relationship between CF, FD, REI, GDP and PH is investigated using the Westerlund³⁹ cointegration test. The Westerlund cointegration test is more robust than previous panel cointegration tests such as Pedroni and Kao's.⁵⁵ The following equations are used in the Westerland test:

G_{t} = \frac{1}{N} \sum_{i - 1}^{N} \frac{{α^{'}}_{i}}{S E ({α^{'}}_{i})}

(12)

G_{a} = \frac{1}{N} \sum_{i - 1}^{N} \frac{T {α^{'}}_{i}}{{α^{'}}_{i} (1)}

(13)

P_{t} = \frac{α^{'}}{S E (α^{'})}

(14)

P_{a} = T α^{'}

(15)

Group mean statistics, G_t and G_a, are estimated using equations (12) and (13), whereas panel statistics, P_t and P_a, are estimated using equations (14) and (15). We shall contrast the alternative theory, which predicts cointegration, with the null hypothesis, which predicts none.

CS-ARDL test

This study employs the CS-ARDL approach to overcome panel data's possible cross-sectional dependency and slope heterogeneity restrictions. This approach yields reliable findings regardless of a non-stationarity issue or the mixed order of integration of the variables. Endogeneity issues, which are more frequently panel data issues, are also addressed by this test.⁵⁶

Using the CS-ARDL method, the long- and short-term effects of FD, CF, REI, and other control variables like GDP and PH are investigated. The following equations are used in the CS-ARDL method:

C F_{i, t} = \sum_{i = 0}^{p w} γ_{I, i} C F_{i, t - 1} + \sum_{j = 0}^{p z} β_{I, i} Z_{i, t - 1} + u_{i, t}

(16)

C F_{i, t} = \sum_{i = 0}^{p w} γ_{I, i} C F_{i, t - 1} + \sum_{j = 0}^{p z} β_{I, i} Z_{i, t - 1} + \sum_{j = 0}^{p x} a_{i,}^{'} I {\bar{X}}_{i t - 1} + u_{i, t}

(17)

Where ${\bar{X}}_{t - 1} = (\bar{C O_{i, t - 1}}, \bar{Z_{i, t - 1}},)$ .

The following are the mean group estimator and long-run coefficient:

{\hat{π}}_{C S - A R D L, i} = \frac{\sum_{I = 0}^{p z} {\hat{β}}_{I, i}}{1 - \sum_{I = 0}^{p w} γ_{I, i}}

(18)

The group mean is given as

{\hat{π}}_{M G} = Σ_{i}^{N} {\hat{π}}_{i}

(19)

Short-run coefficients are estimated as

Δ C F_{i, t} = φ_{i} [C F_{i, t - 1} - π X_{i, t}] - \sum_{I = 1}^{p w - 1} γ_{I, i}, Δ_{I} C F_{i, t - 1} + \sum_{I = 0}^{p z} γ_{I, i}, Δ_{I} Z_{i, t} + \sum_{I = 0}^{p X} a_{i}^{'}, I {\bar{X}}_{t} + u_{i, t}

(20)

Where $Δ_{t} = t - (t - 1)$

{\hat{τ}}_{i} = - (1 - \sum_{I = 0}^{p X} \hat{γ_{I, i}})

(21)

{\hat{π}}_{i} = \frac{\sum_{I = 0}^{p z} {\hat{β}}_{I, i}}{{\hat{τ}}_{l}}

(22)

{\hat{π}}_{M G} = Σ_{i = 1}^{N} {\hat{π}}_{l}

(23)

DH causality test

In order to determine causality, this study uses the Dumitrescu and Hurlin⁵⁷ test, which is better when T is higher or lower than N. In addition, this test is ideal for balanced and heterogeneous panel data sets. This test also effectively handles cross-sectional dependency. The following is an example of how generally this test to explain the following equation:

Z_{i, t} = a_{i} + \sum_{j = 1}^{p} β_{i}^{j} Z_{i, t - j} + \sum_{j = 1}^{p} γ_{i}^{j} T_{i, t - j}

(24)

where

β_{i}^{j}

stands for autoregressive parameters, and j is the lag length. The alternative hypothesis is that there is a causal connection, and the null hypothesis is that there are no causal relationships.

Results and discussion

Empirical findings

The findings from the SCH and CSI tests are presented in Tables 3 and 4, respectively. The SCH test rejects the null hypothesis at the 1% level of significance. Therefore, there is heterogeneity of cross-section slope coefficients. Similarly, CSI findings show that all cross-sections are independent. A significant dependency of panel variables implies that shocks in Canada, Switzerland, Australia, the United States, Denmark, Germany, Sweden, Spain, Finland, and Iceland appear to disseminate to other countries. The study used the Pesaran⁵⁸ unit-root test to determine the integration order of CF, REVD, EXPD, CFD, REI, GDP, PH and EnFTI. Table 5 shows the results of the Stationary Analysis. Only GDP is stationary at the level, while CF, REVD, EXPD, CFD, REI and PH are non-stationary at the level. All the variables, however, become stationary after taking the first differences.

Table 3.

Results of slope homogeneity tests.

Test	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Delta	13.358***	9.548***	12.748***	8.659***	9.847***	9.745***
Delta Adjusted	15.748***	10.548***	14.325***	9.659***	13.458***	11.659***

Note: *** represents 1% level of significance (Ho: cross-sections are homogeneous).

Table 4.

Results of cross-sectional dependence.

Variables	CD-Statistics	P
CF	24.487***	0.000
REVD	25.548***	0.000
EXPD	31.748***	0.000
CFD	22.589***	0.000
REI	31.487***	0.000
GDP	19.548***	0.000
PH	31.748***	0.000
EnFTI	29.458***	0.000

Note: *** represents 1% level of significance (Ho: cross-sections are independent). CF: carbon footprint; REVD: revenue decentralization; EXPD: expenditure decentralization; CFD: composite fiscal decentralization (index); REI: renewable energy intensity; GDP: gross domestic product; PH: population headcount (in millions); EnFTI: environment-friendly technological innovation; CD: cross-sectional dependency.

Table 5.

Results of stationary analysis.

	Cross-Sectionally Augmented IPS (CIPS)				Order of Integration
Variable	I (0)		I(1)
Variable	C	C&T	C	C&T
CF	−2.125	−2.821	−4.168***	−4.399***	I (1)
REVD	−1.111	−1.321	−3.596***	−3.958***	I (1)
EXPD	−1.514	−1.458	−4.287***	−4.357**	I (1)
CFD	−0.245	−1.741	−5.258***	−5.478***	I (1)
REI	−1.698	−1.925	−3.692***	−3.789***	I (1)
GDP	−2.154	−2.298	−4.598***	−4.874***	I (1)
PH	−1.878	−1.402	−3.924***	−3.859***	I (1)
EnFTI	−2.584	−1.548	−3.258***	−3.458***	I (1)

Note: *** and ** show significant levels at 1% and 5%, respectively. CF: carbon footprint; REVD: revenue decentralization; EXPD: expenditure decentralization; CFD: composite fiscal decentralization (index); REI: renewable energy intensity; GDP: gross domestic product; PH: population headcount (in millions); EnFTI: environment-friendly technological innovation.

Furthermore, the Westerlund³⁹ test is used to determine the long-term cointegration between CF and the regressors, REVD, EXPD, CFD, REI, PH and EnFTI. Table 6 shows the group mean statistics for general cointegration denoted by Gt and Ga, whereas the panel statistics are denoted by Pa and Pt. These estimates show evidence of long-term cointegration between CF, REVD, EXPD, CFD, REI, GDP, PH and EnFTI, with significant values of 1%, 5% and 10%, respectively.

Table 6.

Panel cointegration test.³⁹

Statistics	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Gt	−7.847***	−7.499**	−9.487***	−8.896**	−7.847**	−9.357***
Ga	−11.325*	−13.854**	−17.789***	−15.302**	−12.854**	−14.879***
Pt	−12.248**	−14.659***	−17.358***	−18.588***	−15.358***	−18.698***
Pa	−18.548***	−22.198***	−23.859***	−23.169***	−18.125***	−23.589***

Note: ***, **, and * show significant levels at 1%, 5%, and 10% respectively. Gt and Ga, Group mean statistics; Pt and Pa, panel statistics.

Tables 7 and 8, where the dependent variable is CF, provide the long-run and short-run estimations for CS-ARDL. The results suggest that REI and other control variables, including GDP, PH, and EnFTI, FD (REVD, EXPD, and CFD), and REI have a role in explaining the CF. The trailing value of the ECM backs up our hypothesis that FD (REVD, EXPD, and CFD), REI and the control variables have an impact on CF. In addition, REVD, EXPD, CFD, REI, EnFTI and the interaction terms (REVD*REI, EXPD*REI), and CFD*REI have negative and significant coefficients, implying that an increase in these factors lowers CF in the sampled nations. GDP and PH, on the other hand, are positively linked to CF.

Table 7.

Long-run estimations from the CS-ARDL model.

Variables	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
REVD	−0.098***	−0.129**	-	−	−	−
	(0.019)	(0.059)		−	−	−
REVD × REI	−	−0.351***	-	−	−	−
		(0.048)
EXPD	−	−	−0.055*	−0.091**	−	−
			(0.024)	(0.040)
EXPD × REI	−	−	−	−0.325***	-	−
				(0.034)
CFD	−	−	−	−	−0.099***	−0.248***
				-	(0.015)	(0.041)
CFD × REI	−	−	−	−	−	−0.358***
	−	−	−	−	−	(0.042)
REI	−0.059**	−0.078**	−0.029***	−0.018**	−0.148***	−0.158***
	(0.029)	(0.031)	(0.009)	(0.010)	(0.024)	(0.011)
GDP	0.958***	0.658***	0.767***	0.958***	0.457***	0.158***
	(0.035)	(0.029)	(0.044)	(0.028)	(0.014)	(0.025)
PH	0.348***	0.589***	0.847***	0.748***	0.698***	0.658***
	(0.035)	(0.014)	(0.045)	(0.025)	(0.014)	(0.025)
EnFTI	−0.548***	−0.258***	−0.398***	−0.478***	−0.358***	−0.487***
	(0.014)	(0.058)	(0.098)	(0.035)	(0.087)	(0.025)
C	0.161	0.174*	0.358	0.458	0.376*	0.458*
	(0.196)	(0.100)	(0.199)	(0.356)	(0.184)	(0.200)

Note: ***, **, and * show significant levels at 1%, 5%, and 10% respectively. CS-ARDL: cross-sectional autoregressive distributed lag; CF: carbon footprint; REVD: revenue decentralization; EXPD: expenditure decentralization; CFD: composite fiscal decentralization (index); REI: renewable energy intensity; GDP: gross domestic product; PH: population headcount (in millions); EnFTI: environment-friendly technological innovation.

Table 8.

Short-run estimations from the CS-ARDL model.

Variables	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
ECM (−1)	−0.654***	−0.635***	−0.325***	−0.547***	−0.614***	−0.569***
	(0.041)	(0.054)	(0.038)	(0.069)	(0.036)	(0.025)
REVD	−0.126***	−0.245***	−	−	−	−
	(0.065)	(0.008)
REVD*REI	−	−0.147***	−	−	−	−
		(0.057)
EXPD	-	-	−0.073*	−0.327***	-	−
			(0.035)	(0.054)
EXPD*REI	-	-	-	−0.458***	-	−
				(0.014)
CFD	-	-	-	-	−0.084***	−0.155***
				-	(0.009)	(0.017)
CFD*REI	-	-	-	-	-	−0.094**
						(0.03)
REI	−0.028***	−0.098***	−0.057***	−0.147***	−0.147***	−0.085*
	(0.001)	(0.012)	(0.011)	(0.019)	(0.020)	(0.045)
GDP	0.548***	0.258***	0.548***	−0.748***	0.458***	0.125***
	(0.054)	(0.052)	(0.048)	(0.057)	(0.036)	(0.020)
PH	0.085***	0.069***	0.098***	0.096***	0.087***	0.083***
	(0.005)	(0.021)	(0.011)	(0.009)	(0.014)	(0.010)
EnFTI	−0.568***	−0.239***	−0.548***	−0.367***	−0.458***	−0.698***
	(0.032)	(0.047)	(0.025)	(0.048)	(0.045)	(0.014)
C	1.074***	0.909***	1.190***	1.081***	1.336***	1.040*
	(0.022)	(0.031)	(0.267)	(0.078)	(0.079)	(0.560)

Some interesting observations are reported in Tables 7 and 8 as follows:

REI has a short and long-term negative relationship with CF. In the long run, REI causes an average decrease of 0.059% in CF. Additionally, REI reduces CF by 0.028% in the short run. As a result, we conclude that the green paradox exists in the sample nations. This finding may be explained by the fact that stringent local environmental laws and an increase in renewable energy projects at the local government level improve environmental conditions in the sample countries. Nations may effectively enact legislation to enhance environmental quality by delegating authority to lower levels of government.

Revenue decentralisation, expenditure decentralisation, and composite fiscal decentralisation are the three types of FD we analysed. This study aims to look at the overall trends of data from all three FD proxies from various perspectives. In the sample countries, all three variables related to FD are negatively impacted by CF. The findings also suggest that REVD, EXPD, and CFD are all negatively associated with CFs. A 1% increase in the variables causes a 0.126%, a 0.073%, and a 0.084% decrease in CFs, respectively, in the short run. The increases in the respective variables cause a 0.098%, 0.055%, and 0.099% decrease in CFs in the long run. Consequently, we conclude that the green paradox occurs in the sample countries, which might be explained by the fact that the environmental quality of the sample nations increases throughout the decentralisation process due to strong local environmental regulations and higher state permission to lower government. By permitting the lowest layer of government, nations may be able to implement legislation aimed at improving environmental quality successfully. Fiscal decentralisation is necessary to meet the objective of lowering CFs. There is sufficient evidence of a “race to the top” mentality in the sample countries. These highly decentralised nations reap the benefits of their decentralisation plans by employing a “beggar-thy-neighbour” approach to shift polluting activities to nearby areas. However, a delineation of duties at various levels of government is required to realise the energy-saving functions of fiscal spending.

The findings also imply that GDP is positively related to CFs in the short and long run. In the short run, a rise in GDP results in a 0.548% increase in CFs. Additionally, a decline in GDP leads to a long-term reduction in CFs by 0.958%. CFs are one potential by-product of economic activity. As temperatures rise and the demand for fossil fuels rises, an increase in economic activity puts a lot of strain on the environment. Due to their efforts to attain quick economic development, many countries’ environmental quality suffers.

Additionally, the findings imply that PH has both a short-term and long-term favourable relationship with CFs. Over time, a 1% rise in PH causes a 0.348% increase in CFs. CFs can be thought of as an unintended consequence of economic activity. An increase in economic activity puts a lot of burden on the environment since an increase in population increases the need for fossil fuels, which causes temperatures to rise. As a result, many countries’ environmental quality deteriorates as they attempt to achieve high economic growth.

Carbon footprints are inversely connected with technological advances that benefit the environment. A 1% shift in EnFTI lowers CFs over the long term by 0.548% and in the near term by 0.568%. These findings are consistent with those of Nguyen et al.⁵⁹ and Khan et al.⁵⁵ In addition, using newer, more efficient machinery causes a nation's industrial structure to change from relying on nonrenewable to renewable energy. A key strategy for increasing domestic output without sacrificing environmental quality is to finance eco-innovation regularly.

The robustness test results from the augmented mean group test are shown in Table 9, and the robustness test results support the results from CS-ARDL. REVD, EXPD, CFD, REI, and EnFTI reduce CFs by 0.169%, 0.098%, 0.157%, 0.214%, and 0.147%, respectively. Increases in GDP and PH, on the other hand, result in CF increases of 0.139% and 0.418%, respectively.

Table 9.

Robustness regression from the AMG estimator.

Variables	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
REVD	−0.169***	−0.158**	–	–	–	–
REVD × REI	–	−0.197***	–	–	–	–
ED			−0.098**	−0.089**	–	–
ED × REC			–	−0.258***	–	–
CD	–	–	–	–	−0.157***	−0.148***
CD × REC	–	–	–	–	–	−0.158***
REI	−0.214**	−0.523***	−0.147***	−0.259***	−0.210***	−0.345***
GDP	0.139***	0.492***	0.454**	0.418***	0.303***	0.458**
PH	0.418***	0.358***	0.468***	0.548***	0.369**	0.541**
EnFTI	−0.147***	−0.258***	−0.369**	−0.487***	−0.148***	−0.353**
C	1.528***	1.339**	1.683**	1.587***	1.587***	1.147**

Note: *** and ** show significant levels at 1%, 5%, and 10% respectively. CF: carbon footprint; REVD: revenue decentralization; EXPD: expenditure decentralization; CFD: composite fiscal decentralization (index); REI: renewable energy intensity; GDP: gross domestic product; PH: population headcount (in millions); EnFTI: environment-friendly technological innovation; CD: cross-sectional dependency; REC: renewable energy consumption.

The further robustness test results using the Common Correlated Effects Mean Group (CCEMG) estimator are presented in Table 10, further confirming the consistency of the main findings derived from the CS-ARDL model. REVD, EXPD, CFD, REI, and EnFTI are found to significantly reduce CFs by 0.142%, 0.085%, 0.143%, 0.201%, and 0.134%, respectively. In contrast, increases in GDP and PH contribute to higher CFs by 0.128% and 0.392%, respectively. These findings align with those obtained from the AMG estimator and reinforce the robustness of the results across different estimation techniques. Moreover, the Dumitrescu–Hurlin causality test results (Table 11) indicate a one-way causality from REVD, EXPD, CFD, REI, GDP, PH, and EnFTI to CF, implying that policy measures targeting these variables will significantly influence environmental outcomes.

Table 10.

Robustness regression using the CCEMG estimator.

Variables	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
REVD	−0.142***	−0.129**	–	–	–	–
REVD × REI	–	−0.181***	–	–	–	–
ED	–	–	−0.085**	−0.073**	–	–
ED × REC	–	–	–	−0.239***	–	–
CD	–	–	–	–	−0.143***	−0.131***
CD × REC	–	–	–	–	–	−0.139***
REI	−0.201**	−0.497***	−0.138***	−0.238***	−0.194***	−0.324***
GDP	0.128***	0.476***	0.431**	0.399***	0.291***	0.439**
PH	0.392***	0.341***	0.446***	0.519***	0.353**	0.517***
EnFTI	−0.134***	−0.243***	−0.351**	−0.466***	−0.134***	−0.336***
C (Const.)	1.437***	1.285**	1.614**	1.518***	1.493***	1.073***

Note: *** and ** show significant levels at 1%, 5%, and 10% respectively. CCEMG: common correlated effects mean group; REVD: revenue decentralization; EXPD: expenditure decentralization; CFD: composite fiscal decentralization (index); REI: renewable energy intensity; GDP: gross domestic product; PH: population headcount (in millions); EnFTI: environment-friendly technological innovation; CD: cross-sectional dependency; REC: renewable energy consumption.

Table 11.

Dumitrescu–Hurlin causality test for the heterogeneous panel.

Null Hypothesis:	W-Stat.	Zbar-Stat.	Prob.
REVD does not homogeneously cause CF	2.693	2.827	0.005
CF does not homogeneously cause REVD	1.932	1.456	0.146
EXPD does not homogeneously cause CF	2.545	2.561	0.010
CF does not homogeneously cause EXPD	1.784	1.189	0.235
CFD does not homogeneously cause CF	2.625	2.705	0.007
CF does not homogeneously cause CFD	1.412	0.518	0.605
REI does not homogeneously cause CF	3.666	4.576	0.000
CF does not homogeneously cause REC	1.274	0.267	0.790
GDP does not homogeneously cause CF	2.585	2.634	0.008
CF does not homogeneously cause GDP	1.815	1.245	0.213
PH does not homogeneously cause CF	4.089	5.347	0.000
CF does not homogeneously cause PH	1.810	1.231	0.218
EnFTI does not homogeneously cause CF	2.458	3.124	0.084
CF does not homogeneously cause EnFTI	1.865	1.859	0.007

REVD: revenue decentralization; EXPD: expenditure decentralization; CFD: composite fiscal decentralization (index); REI: renewable energy intensity; GDP: gross domestic product; PH: population headcount (in millions); EnFTI: environment-friendly technological innovation; CF: carbon footprint; REC: renewable energy consumption.

The findings of the Dumitrescu–Hurlin causality test are shown in Table 11. The findings support a one-way causal link between REVD, EXPD, CFD, REI, GDP, PH, EnFTI, and CF, indicating that any policy related to these variables will have an impact on CF.

The results of this study provide important insights into the economic significance of FD and its impact on CFs, particularly in the context of REI. Beyond statistical significance, the findings highlight how decentralisation can lead to increased local investments in clean energy and more effective carbon reduction policies. The magnitude of the coefficients suggests that decentralisation not only enables more tailored environmental policies at the local level but also enhances the effectiveness of these policies in reducing CFs over time.

Additionally, our analysis reveals a clear connection to the “race to the top” hypothesis. Fiscal decentralisation appears to encourage competition among subnational governments to adopt progressive environmental policies, fostering innovation and advancing sustainability goals. This competition can lead to greater adoption of renewable energy technologies and stricter emissions regulations. In contrast, the study also examines the “race to the bottom” theory, acknowledging that, in some instances, decentralised fiscal powers may result in weaker environmental policies, especially when local economic pressures conflict with environmental goals.

These findings underscore the importance of FD in shaping environmental governance. The results suggest that decentralisation, if managed properly, can serve as a powerful tool for driving local sustainability initiatives and encouraging competition for environmental leadership. This highlights the need for policies that strike a balance between local autonomy and central oversight to ensure that environmental goals are not compromised in pursuit of short-term economic growth.

By linking these results to existing literature on FD and environmental governance, this study contributes to a deeper understanding of how decentralisation can influence both policy and environmental outcomes. These findings offer practical implications for policymakers seeking to promote sustainable development while also providing valuable insights into the broader discourse on decentralised governance and its role in achieving carbon reduction objectives.

Further discussions

Overall, current estimates suggest that FD could be utilised as a proactive tool for reducing CFs. Furthermore, the study finds that FD increases environmental quality. Local governments in highly decentralised regions encourage using renewable energy and carbon-reduction initiatives. These encouragements help energy-saving and carbon-reduction targets. The significant and negative interaction terms imply that REI and CFs have a high interaction effect. The quality of REI is high in the ten selected highly decentralised countries, allowing them to undertake policies to reduce environmental damage. Highly decentralised countries with low REI may not make significant environmental improvements.

Therefore, we can assert that, in terms of reducing CFs, the intensity of renewable energy is complementary to FD and, if used thoroughly and efficiently, will enhance environmental quality. Institutional benevolence towards renewable energy projects ensures the “race to the top” brought on by growing fiscal autonomy. People with strong political power organise into groups with shared environmental concerns and educate the public about the need to preserve a clean environment by supporting renewable energy initiatives. GDP also has a positive effect on carbon emissions. It is important to note that a country's GDP growth substantially influences the environment and that a country's continued and erratic use of natural resources worsens the quality of the environment. The rate of CFs will gradually rise as a result. This suggests that as GDP increases, governments will focus more on emerging industries, increasing demand for higher levels of expenditure.

Fiscal decentralization is often viewed as a potential driver of environmental improvement, but it can also have drawbacks, such as the risk of a “race to the bottom.” In some cases, local governments may reduce environmental standards to attract investment, leading to deregulation and weaker environmental protection. This possibility arises particularly in contexts where there is intense competition among subnational governments to secure economic growth and investment.^14,60 While FD can enable more localized and tailored environmental policies, without strong institutional oversight, it may encourage subnational governments to compromise environmental standards to foster short-term economic gains.

We now acknowledge this potential downside and balance the “race to the top” narrative with a more cautious view. The impact of FD on environmental governance largely depends on institutional quality, governance structures, and the regulatory frameworks in place. In countries where institutional capacity is weak or where there is insufficient intergovernmental coordination, FD may exacerbate environmental degradation, as local governments compete by lowering standards.

To address this issue, we have proposed several safeguards to ensure that decentralization does not undermine environmental goals. These include the establishment of national-level environmental standards, performance-based fiscal incentives, and mechanisms for intergovernmental coordination that help align local and national sustainability objectives. By incorporating these safeguards, we can mitigate the risks of environmental deregulation and ensure that FD leads to positive environmental outcomes.

This study explores the relationship between FD and CFs, focusing on the interaction between REI and decentralisation. We refine the conceptual framework by distinguishing between the dimensions of FD , including revenue, expenditure, and composite decentralisation, which influence local energy policies and environmental outcomes. Fiscal decentralisation enables local governments to invest in renewable energy and implement tailored policies, thus enhancing carbon reduction efforts. These insights contribute to a clearer understanding of how FD interacts with REI to reduce CFs. Similar research by Shahzad et al.¹⁰ and Yavuz et al.²⁹ supports the notion that decentralisation fosters local energy transitions, promoting sustainable environmental policies.

The market's use of nonrenewable energy has been increasing because of the high demand for consumption. Such energy use is not favourable to the environment and, as a result, harms the environment, increasing CFs. The results reported are consistent with those of Kalmaz and Kirikkaleli,⁶¹ Ji et al.,⁶² and Khan et al.⁵⁵ The empirical studies also show that the quality of institutions increases the quality of the environment. Furthermore, FD improves environmental quality, consistent with the “race to the top strategy”. Certain countries discourage polluting behaviours by imposing strict environmental requirements and undertaking a beggar-thy-neighbour drive to export their polluting activities to neighbouring regions consistent with the race to the top approach. These results are consistent with those of Cutter and DeShazo⁶³ and Millimet,¹⁶ who reported that FD improves environmental quality.

Conclusion

This study uses data from the top 10 decentralised OECD countries from 1990 to 2022 for the empirical analysis. It employs the CS-ARDL technique to examine the geographical impacts of FD and the interaction effect of the intensity of renewable energy on CFs. We also use the Dumitrescu–Hurlin causality test to examine the causal connections further. The following are the conclusions of the study.

Recent research indicates that FD may serve as an effective strategy for reducing CFs. Empirical evidence suggests that decentralised governance systems positively impact environmental quality, with local governments in such systems actively promoting the use of renewable energy and carbon-reduction measures. The primary contribution of this study lies in examining the interplay between FD and REI in shaping CFs. The significant negative relationship between the interaction terms points to a strong correlation between REI and CF reduction. In highly decentralised countries, where the adoption of renewable energy is low, it may be more challenging to achieve substantial environmental improvements. The ten highly decentralised countries selected for this study are characterised by high REI, enabling them to implement policies aimed at mitigating environmental degradation.

Long-term and short-term negative relationships exist between CFs and REI. In the long run, REI reduces CF by an average of 0.059%, while in the short run, REI reduces CF by 0.028%. These results suggest the presence of the green paradox in the sample nations. This conclusion may be explained by the fact that improved environmental conditions in these nations are largely attributed to stringent local environmental regulations and an increase in renewable energy projects at the subnational government level. By reducing central government control, countries can more effectively implement policies to improve environmental quality.

The results also suggest that decentralisation of revenue, expenditure and the combined fiscal system are all negatively correlated with CFs, with increases in the corresponding variables leading to short-term reductions in CFs by 0.126%, 0.073%, and 0.084%, respectively. Increases in the abovementioned variables lead to a long-term decrease in CFs of 0.098%, 0.055%, and 0.099%, respectively. As a result, we conclude that the eco-friendly contradiction exists in the example states. The findings support that as decentralisation proceeds, the environmental quality of the countries in the sample increases because of strict local environmental regulations and expanding governmental assistance. By permitting the lower tier of government, nations may be able to implement legislation aimed at improving environmental quality successfully. Therefore, to reduce CFs, policymakers should consider FD.

The robustness analysis using the CCEMG estimator confirms the reliability of the main findings. It demonstrates that FD components such as REVD, EXPD, CFD, REI, and EnFTI significantly reduce CFs, while GDP and PH contribute to increased emissions. These results are consistent with those obtained from the AMG estimator, reinforcing their validity across different estimation techniques. Moreover, the Dumitrescu–Hurlin causality test indicates one-way causality from these variables to CFs, suggesting that targeted policy interventions in these areas can play an important role in reducing environmental degradation.

Limitations and future recommendations

While our study provides valuable insights into the relationship between FD, REI, and CFs, there are a few limitations to consider. One limitation is the inconsistency in data availability across the countries included in the sample, which may affect the robustness of our findings. Additionally, the way we measured FD may not fully capture its complexity, particularly in countries with intricate governance systems. Moreover, the CS-ARDL approach we used, although effective, comes with certain assumptions that could limit its applicability to more dynamic or non-linear relationships.

Looking ahead, future research could explore alternative methods, such as nonlinear models or structural equation modeling, to better capture the complexities of FD and its impact on environmental outcomes. Extending the analysis to include developing countries could also help make the results more universally applicable and provide a more global perspective on the role of decentralisation in shaping environmental policies. Furthermore, including indicators of political and administrative decentralisation would add depth to our understanding of how governance structures influence sustainability.

Future studies could also benefit from using causal inference techniques, like instrumental variable methods or natural experiments, to more accurately identify causal relationships between FD and environmental performance. By addressing these limitations, future research can strengthen the foundation of this field and offer more actionable insights for policy development around sustainable FD.

Footnotes

ORCID iDs

Farrukh Shahzad

Mara Madaleno

Tingxuan Zhang

Funding

The first author Farrukh Shahzad disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Guangdong University of Petrochemical Technology (grant number: Project No. 702-72100003004 and 702/5210012). Dawei Zhang acknowledge the funding from Major Decision-making Projects of Maoming City Philosophy and Social Sciences Planning (2025ZD01).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Farrukh Shahzad is an Associate Professor at Guangdong University of Petrochemical Technology (GDUPT), China. He earned his PhD in Business Administration from Chongqing University, with a specialization in Economics and Finance. His research focuses on fiscal decentralization, sustainability, corporate finance, and innovation. He has been recognized among the world's top 2% most influential scientists by Stanford University for three consecutive years and serves as an editor and reviewer for several international journals.

Guneratne Wickremasinghe is a Visiting Professor of Business Studies at the State University of Bangladesh and an Honorary Fellow at Victoria University, Melbourne, Australia. He holds a PhD in Finance from Monash University and postgraduate degrees in accounting and finance from Robert Gordon University (UK) and the University of Calicut (India). His research interests include corporate governance, financial market efficiency, sustainability disclosure, and financial econometrics, with a strong focus on emerging markets. He has published forty-seven peer-reviewed papers, served as a reviewer for leading international journals, and supervised fifteen doctoral candidates. He is a full member of CPA Australia and the Chartered Accountants Australia and New Zealand (CA ANZ) and has over two decades of teaching, research, and academic leadership experience.

Yonghua Luo is a Professor in Labor Economics. His area of research is labor market dynamics and employment policy, focusing on skills development and job quality in the transition to a sustainable economy.

Mara Madaleno is an Associate Professor with habilitation in finance and economics at the University of Aveiro. She coordinates the research unit GOVCOPP. Her research interests include energy finance, financial economics, markets and agents behavior, and energy and environmental economics.

Tingxuan Zhang is a PhD Student at the University of Macau. Her areas of interest include, but are not limited to, sustainability and environmental science, and she has published in reputed SCI journal.

Dawei Zhang is an Associate Professor in Industrial Economics. His area of research is sustainable industrial development transformation, focusing on the green and low-carbon development of energy-intensive industries.

Appendix

Table A1.

Fiscal decentralisation score in Top 10 highly decentralised OECD countries.

Rank^a	Country	REVD	EXPD	CFD
1	Canada	47.9	48.5	0.96
2	Switzerland	47.5	42.6	0.93
3	Australia	45.4	44.9	0.88
4	USA	41.4	42.4	0.87
5	Denmark	31.4	38.7	0.75
6	Germany	35.4	34.4	0.65
7	Sweden	29.4	41.7	0.62
8	Spain	28.4	38.4	0.5
9	Finland	27.7	39.8	0.47
10	Iceland	26.5	29.5	0.4

Source: Author's calculation using regional and national revenue and expenditure.

Rank calculated using the CFD index score.

References

Yang

. Fiscal decentralisation, preference for government innovation and city innovation. Chinese Management Studies 2020; 14: 391–409.

Bruvoll

Larsen

. Greenhouse gas emissions in Norway: do carbon taxes work? Energy Policy 2004; 32: 493–505.

Cheng

Fan

Meng

, et al. Potential role of fiscal decentralisation on inter1 differences in CO2 emissions in China. Environ Sci Technol 2020; 55: 813–822.

Hao

L-N

Umar

Khan

, et al.

Green growth and low carbon emission in G7 countries: how critical the network of environmental taxes, renewable energy and human capital is?

Sci Total Environ 2021; 752: 141853.

C-W

Umar

Khan

. Does fiscal decentralisation and eco-innovation promote renewable energy consumption? Analysing the role of political risk. Sci Total Environ 2021; 751: 142220.

Wang

C-W

Ali

, et al. How China is fostering sustainable growth: the interplay of green investment and production-based emission. Environ Sci Pollut Res 2020b; 27: 39607–39618.

Ulucak

Khan

SU-D

. Determinants of the ecological footprint: role of renewable energy, natural resources, and urbanisation. Sustain Cities and Soc 2020; 54: 101996.

Cheng

Awan

Ahmad

, et al. How do technological innovation and fiscal decentralisation affect the environment? A story of the fourth industrial revolution and sustainable growth. Technol Forecast Soc Change 2021; 162: 120398.

Shahzad

Fareed

. Examining the relationship between fiscal decentralization, renewable energy intensity, and carbon footprints in Canada by using the newly constructed bootstrap Fourier Granger causality test in quantile. Environ Sci Pollut Res 2023; 30: 4617–4626.

10.

Shahzad

Fareed

Wan

, et al. Examining the asymmetric link between clean energy intensity and carbon dioxide emissions: the significance of quantile-on-quantile method. Energy Environ 2023; 34: 1884–1909.

11.

Shahbaz

Sinha

. Environmental Kuznets curve for CO₂ emissions: a literature survey. J Econ Stud 2019; 46: 106–168.

12.

Batterbury

Fernando

. Rescaling governance and the impacts of political and environmental decentralisation: an introduction. World Dev 2006; 34: 1851–1863.

13.

Fell

Kaffine

. Can decentralised planning really achieve first-best in the presence of environmental spillovers? J Environ Econ Manage 2014; 68: 46–53.

14.

Fredriksson

Millimet

. Strategic interaction and the determination of environmental policy across US states. J Urban Econ 2002; 51: 101–122.

15.

Konisky

. Regulatory competition and environmental enforcement: is there a race to the bottom? Am J Pol Sci 2007; 51: 853–872.

16.

Millimet

. Assessing the empirical impact of environmental federalism. J Reg Sci 2003; 43: 711–733.

17.

Sigman

. Decentralisation and environmental quality: an international analysis of water pollution levels and variation. Land Econ 2014; 90: 114–130.

18.

Zhang

Z-Y

Liang

Q-M

. An empirical analysis of the green paradox in China: from the perspective of fiscal decentralisation. Energy Policy 2017a; 103: 203–211.

19.

Zhang

Y-J

Peng

Y-L

C-Q

, et al. Can environmental innovation facilitate carbon emissions reduction? Evidence from China. Energy Policy 2017b; 100: 18–28.

20.

Levinson

. Environmental regulatory competition: a status report and some new evidence. Natl Tax J 2003; 56: 91–106.

21.

Almarshad

. The impact of good governance and decentralisation reforms on the effectiveness of local authorities: The case of Saudi municipalities. Riaz, Saudi Arabia: The University of Connecticut, 2011.

22.

Caglar

Avci

Daştan

, et al. Investigation of the effect of natural resource dependence on environmental sustainability under the novel load capacity curve hypothesis. Intern J Sustain Develop World Ecol 2024; 31: 431–446.

23.

Zhu

Zhao

, et al. The impact of financial decentralisation on industrial structure upgrading: From the coordination perspective of fiscal decentralisation. Intern Rev Financ Anal 2025; 104301.

24.

Fang

Yang

, et al. Regional environmental efficiency evaluation in China: analysis based on the Super-SBM model with undesirable outputs. Math Comput Model 2013; 58: 1018–1031.

25.

Huang

Zhou

Wang

, et al. Decomposition and attribution analysis of the transport sector’s carbon dioxide intensity change in China. Transp Res A Policy Pract 2019; 119: 343–358.

26.

. Fiscal decentralization and environmental pollution: Evidence from Chinese panel data. China Econ Rev 2015; 36: 86–100.

27.

Liu

Hao

Gao

. The environmental consequences of domestic and foreign investment: Evidence from China. Energy Policy 2017; 108: 271–280.

28.

Escaleras

Chiang

. Fiscal decentralization and institutional quality on the business environment. Econ Lett 2017; 159: 161–163.

29.

Yavuz

Kara

Yıldız

. Fiscal decentralisation and environmental quality: the role of local governance in renewable energy transitions. Environ Econo Policy Studies 2024; 26: 77–92.

30.

Hao

, et al. Environmental decentralisation, local government competition, and regional green development: evidence from China. Sci Total Environ 2020; 708: 135085.

31.

Adam

Delis

Kammas

. Fiscal decentralization and public sector efficiency: evidence from OECD countries. Econ Gov 2014; 15: 17–49.

32.

Wang

Lei

. Fiscal decentralization and high-polluting industry development: city-level evidence from Chinese panel data. Int J Smart Home 2016; 10: 297–308.

33.

Jin

Zou

H-f

. Fiscal decentralization, revenue and expenditure assignments, and growth in China. J Asian Econ 2005; 16: 1047–1064.

34.

Oates

. Toward a second-generation theory of fiscal federalism. Intern Tax Public Finance 2005; 12: 349–373.

35.

Martinez-Vazquez

Timofeev

. Decentralization measures revisited. Public Finance Manage 2010; 10: 13–47.

36.

Chen

Chang

. Fiscal decentralization, environmental regulation, and pollution: a spatial investigation. Environ Sci Pollut Res 2020; 27: 31946–31968.

37.

Tang

Yin

Yang

, et al. Emerging pollutants in water environment: occurrence, monitoring, fate, and risk assessment. Water Environ Res 2019; 91: 984–991.

38.

Kuai

Yang

Tao

, et al. Environmental effects of Chinese-style fiscal decentralization and the sustainability implications. Journal of Cleaner Production 2019; 239: 118089.

39.

Westerlund

. Error correction based panel cointegration tests. Oxf Bull Econ Stat 2007; 69: 709–748.

40.

Bird

Smart

. Intergovernmental fiscal transfers: international lessons for developing countries. World Bank Institute 2002; 30: 899–912.

41.

Shah

. Decentralisation and its implications on fiscal policies and sustainable growth: an overview. Intern J Public Admin 2021; 44: 269–285.

42.

Gemmell

Kneller

Sanz

. Fiscal decentralisation and economic growth. World Dev 2013; 43: 1–13.

43.

Pesaran

Shin

. An autoregressive distributed lag modelling approach to cointegration analysis. In: Econometrics and economic theory in the 20th century: the Ragnar Frisch centennial symposium. Cambridge University Press, 1999, pp.371–413.

44.

Shahzad

SJH

Wang

. The role of fiscal decentralisation in enhancing environmental sustainability. Environ Sci Pollut Res 2020; 27: 3475–3487.

45.

García

Sánchez

López

. Technological innovation and environmental sustainability: a comparative study of OECD countries . Environ Innov Soc Transit 2023; 37: 45–59.

46.

Oates

. An essay on fiscal federalism. J Econ Lit 1999; 37: 1120–1149.

47.

Banzhaf

Chupp

. Fiscal federalism and interjurisdictional externalities: new results and an application to US air pollution. J Public Econ 2012; 96: 449–464.

48.

Farooqi

Zhang

Khan

. Fiscal decentralization and environmental sustainability: role of institutional quality. Environ Econ Policy Studies 2022; 24: 451–475.

49.

Jiménez

Pérez

. Fiscal decentralization and environmental quality: does institutional quality matter? Econ Model 2019; 80: 35–49.

50.

Bardhan

. Decentralization of governance and development. J Econ Perspect 2002; 16: 185–205.

51.

Pesaran

Yamagata

. Testing slope homogeneity in large panels. J Econom 2008; 142: 50–93.

52.

Pesaran

Yang

. Econometric analysis of production networks with dominant units. J Econom 2020; 219: 507–541.

53.

Campello

Galvao

Juhl

, et al. Testing for slope heterogeneity bias in panel data models. J Bus Econ Stat 2019; 37: 749–760.

54.

Wang

Lin

, et al. Psychological status of medical workforce during the COVID-19 pandemic: A cross-sectional study. Psychiatry Res 2020; 288: 112936.

55.

Khan

Ali

Jinyu

, et al. Consumption-based carbon emissions and trade nexus: evidence from nine oil exporting countries. Energy Econ 2020; 89: 104806.

56.

Chudik

Pesaran

. Common correlated effects estimation of heterogeneous dynamic panel data models with weakly exogenous regressors. J Econom 2015; 188: 393–420.

57.

Dumitrescu

Hurlin

. Testing for Granger non-causality in heterogeneous panels. Econ Model 2012; 29: 1450–1460.

58.

Pesaran

. A simple panel unit root test in the presence of cross-section dependence. J Appl Econ 2007; 22: 265–312.

59.

Nguyen

THN

DDM

, et al. Fiscal decentralisation, corruption, and income inequality: evidence from Vietnam. J Asian Finance, Econ Business 2020; 7: 529–540.

60.

Esty

Geradin

. Regulatory competition and economic integration: comparative perspectives. J Intern Econ Law 1998; 1: 235–255.

61.

Kalmaz

Kirikkaleli

. Modeling CO₂ emissions in an emerging market: empirical finding from ARDL-based bounds and wavelet coherence approaches. Environ Sci Pollut Res 2019; 26: 5210–5220.

62.

Umar

Ali

, et al. Does fiscal decentralisation and eco-innovation promote sustainable environment? A case study of selected fiscally decentralised countries. Sustain Develop 2021; 29: 79–88.

63.

Cutter

DeShazo

. The environmental consequences of decentralising the decision to decentralise. J Environ Econ Manage 2007; 53: 32–53.

64.

Wang

. Rethinking the environmental kuznets curve hypothesis across 214 countries: the impacts of 12 economic, institutional, technological, resource, and social factors. Humanities Soc Sci Commun 2024; 11: 1–19.