Environmental regulations,fiscal decentralization,and health outcomes

Abstract

A complete understanding of the interplay between environmental regulations and fiscal decentralization for the realization of health outcomes is crucial for policy formulation and decision-making. The aim of the study is to investigate both policy variables’ separate and combined impacts using data on four BRICS economies from 2000 to 2020. The study has employed the novel method of moments quantile regression to quantify the effect. The findings of the study show (i) a significant impact of regulations on health outcomes in higher quantiles. Environmental regulations have a strong positive impact on all three-health proxies, that is, health expenditures, life expectancy, and the number of infant deaths. Total revenue and expenditure decentralization affect health outcomes positively, while tax revenue decentralization negatively impacts them, with the effect being stronger in the lower quantiles; (ii) the combined impact of decentralization and environmental regulations turned out to be negative and significant in our study; and (iii) all variables have unidirectional causality. However, with tax revenue decentralization, health expenditures, life expectancy, and infant deaths have bidirectional causality. This finding has a strong policy implication for the policymakers. Although both policies positively impact health indicators, their interaction leads to deteriorating health outcomes. From a policy point of view, it is suggested to strike a balance between regulations and fiscal decentralization to realize the full potential of this policy mix to get better health outcomes. This study adds to previous research by incorporating the interconnected impact of environmental regulations and fiscal decentralization on health outcomes in BRICS economies.

Keywords

Environmental regulations health expenditures environmental degradation fiscal decentralization MMQREG JEL classifications E62 H23 H51 Q51

Introduction

The profound history of emerging air contaminants has led to human civilization experiencing the negative consequences of global climate change. Recent dramatic atmospheric shifts have hurt world economies and human populations.^1,2 In recent years, environmental pollution-related health effects have drawn much attention from the public and academia.^3,4 Relevant studies have demonstrated that environmental pollution considerably lowers life expectancy, raises health care costs, and causes more deaths.^5,6 Environmental control laws have sound environmental effects, but their public health effects are unknown. Environmental regulations (ER) can improve environmental quality, reduce health risks, and lower health care costs.⁷ Improving public health care, increasing government health spending, and promoting public health is crucial in achieving the 3rd goal of sustained development. To achieve this target, promoting effective public health care through efficient resource management is necessary. In this regard, fiscal decentralization (FD) schemes can ensure efficient management of public health resources. Local governments are responsible for most medical and health services under a FD system. However, a decentralized healthcare system's potential benefits and pitfalls are still debatable.⁸ The literature on decentralization is not always conclusive. On the whole, empirical attempts to analyze environmental degradation and FD's effects on health outcomes are scarce. This study aims to fill this gap by examining how environmental rules and FD may affect health indicators separately and jointly.

Health is key to human capital and well-being. Individual and national efficiency grows with good health. WHO defines public health as avoiding disease, extending life, and improving health via organized social action. It influences life expectancy, quality of life, and the economy. Improving public health care, increasing government health spending, and promoting public health are crucial in achieving the 3rd goal^a of sustained development. To achieve this target, promoting effective public health care through efficient resource management is necessary. Policymakers must understand the factors that affect health outcomes. Since environmental degradation is a potential cause of deteriorating health standards worldwide, policies aimed at controlling pollution and enhancing environmental performance may bring massive health-related benefits.⁹ ER are one such measure that may help combat environmental degradation,^10,11 especially in developing countries. Environmental laws lower medical costs by improving air quality.¹² Another potential policy-level measure to improve health indicators is the degree of FD of the economies. The decentralization concept dates back to Charles (1956)¹³ and Oates (1972),¹⁴ who argue that FD schemes can ensure efficient management of public health resources. FD is a global economic trend. The classic idea of FD states that FD, as an institutional structure of fiscal division between the central and local governments, helps local governments deliver public products according to local conditions. FD shifts government revenue and spending to lower levels.¹⁵ When the central government gives up political power (political and spending decentralization), subnational governments can make their own laws about health care and decide how to spend money depending on citizens’ preferences and regional needs. It can also give lower-level governments more control over their budgets by letting them use different sources of revenue to pay for services.

The literature on FD is context and country-specific. In other words, FD isn’t necessarily beneficial. Presumably, the situational effects of FD are caused by the fact that local governments in many developing countries and even some European countries don’t have enough administrative skills (like the ability to raise taxes and borrow money), which hurts their ability to protect fiscal freedom. FD might under-provide public goods due to spillovers. Local elites and interest groups can “capture” elected officials, creating corruption. Nevertheless, the influence of FD is dual on social welfare. According to first-generation federalism, local governments (agencies) serve as helpful agents for the general good, such as social welfare,^14,16 by allowing local control over spending (expenditure decentralization (ED) per se). Second-generation fiscal federalists argue that “Selfish public leaders with their agenda” are more likely to seek national interests.⁸ So, enabling local governments with the ability to generate a significant percentage of their money (revenue decentralization (RD)) would encourage them to act more responsibly toward the general public, come up with innovative market-enhancing public goods, and be less corrupt. In the preceding context, this study aims to examine the impact of ER and FD on health outcomes, such as health expenditures, life expectancy, and the number of infant deaths in the four BRICS nations: Brazil, Russia, China, and South Africa.

The BRICS, as developing countries, demand greater attention for their health potential than developed countries. Despite achieving economic growth, even in the BRICS, healthcare spending as a percentage of the national budget is still far lower than in the top Western and Asian OECD companies. With the growing burden of non-communicable diseases, low-middle-income nations account for 75% of the worldwide NCD burden in disability-adjusted life-years; this challenge is vast in width and scale, with ineffective financial approaches. According to the statistical review of world energy,¹⁷ China has the most carbon emissions among the BRICS. China accounted for 28.8% of global carbon emissions in 2020; India was third with 7.3%. Russia, Brazil, and South Africa emit 4.5%, 1.3%, and 1.4% of carbon. In 2020, BRICS countries emitted 43.3% of global carbon. Such challenges remain high on policymakers’ agendas to welcome effective ER, even more so than in traditional high-income nations. Alarming environmental and human health issues may threaten BRICS unity. With these health hazards in BRICS economies, it is essential to pay attention to this area. In this regard, BRIC nations need specific ER and effective FD for the future.

Figures 1 and 2 depict an interesting picture. Figure 1 shows the studied countries’ health expenditures as a percentage of GDP. Brazil is spending the highest share of its GDP in the health sector, while Russia and China spend a minimal share in the health sector compared to the other BRICS economies. Figure 2 shows that the countries sparing small amounts for health expenditures are less decentralized than those that spend more in the health sector. Henceforth, the higher degree of decentralization may be linked with higher health sector expenditures, as the figures suggest. Therefore, we suspect a positive link between decentralization and health improvements.

Figure 1.

Health expenditures in BRCS economies.

Figure 2.

Fiscal decentralization in BRCS economies.

This study contributes to prior research in two essential aspects ways. First, most research has examined the relationship between ER and FD separately. Few researches have examined how environmental degradation affects health. This study determines both variables’ separate and combined impact on health outcomes in BRICS countries. Second, earlier studies used life expectancy, number of infant deaths, or health expenditures as health outcome proxies. This study uses all three-health proxies to assess the link between ER, decentralization, and health indicators in BRICS economies. Thus, according to our understanding, this is the first research to evaluate the individual and combined effects of ER and FD on health outcomes in the BRICS nations.

The study has employed moment quantile regression using data from 2000–2020 to measure the individual and collective influence of ER and FD on health indicators. The results indicate that ER significantly affect the health expenditures. The impact gets larger in the higher quantiles. However, ED has a strong positive influence on health expenditures; yet, this impact is stronger in the lower quantiles. The total and tax revenue decentralization (TRD) and ER significantly reduce health expenditures in the median. In comparison, the impact is insignificant in the ED equation. In the case of life expectancy at birth, decentralization decreases life expectancy, but the impact reduces significantly in the higher quantiles. ER increase life expectancy in higher quantiles in the RD equation but decrease life expectancy in the ED equation. However, the impact reduces in the higher quantiles. Revenue and regulation interaction reduces life expectancy in higher quantiles. In contrast, expenditure and decentralization interaction increases life expectancy, and the impact is higher in lower quantiles. Infant deaths significantly reduce with higher regulations and more ED. The mutual impact of both policy variables is also positive (i.e. lowers infant deaths). While the effect is stronger in lower quantiles. In sum, ER and FD positively impact health outcomes; however, their combined impact is negative in most specifications. Moreover, TRD is bidirectional causality with health expenditures, life expectancy, and infant deaths, while the other variables are unidirectional. Therefore, policy decisions regarding more decentralized systems with higher ERs must consider their mutual impact very carefully.

Literature review

To better understand the independent and combined effects of environmental degradation and FD on public health, divide them into three parts and discuss each set of variables separately. As explained in the following subsections, the relationships between each pair has received extensive scholarly investigation.

Environmental degradation and public health

Today, academia considers ER to have multifaceted consequences and is vital for sustainable development.^7,18–22 Limited studies have been done on the health implications of expanding and deepening environmental and economic effects. Few researchers have discovered direct links between environmental management and health. However, empirical research on ER and health spending is inconsistent.¹⁸ Two arguments can be used to summarize the inconsistency. According to certain studies, environmental management can lower health costs by enhancing human health.^18,23,24 ER reduces health hazards by increasing environmental quality. Similarly, Do et al.¹⁸ found that ER reduces infant mortality in India, contrary to Greenstone and Hanna.²⁵ Tanaka²⁴ demonstrated that ER reduced newborn mortality in China. Moreover, Cesur et al.²³ observed that replacing coal with natural gas reduced mortality in Turkey. Yang and Chou⁷ evaluated environmental regulatory implications for prenatal health using New Jersey coal plant closure data. After the nearby coal-fired power plant was shut down, low birth weight and preterm deliveries were reduced by 15% and 28%, respectively. According to some studies, environmental control may not help human health and may even worsen it, increasing health care costs.²⁵ Besides the studies listed above, other researchers contend that ER may indirectly impact health due to the mediating effects of specific factors.²⁶ Based on the above discussion, this article offers hypothesis:

Hypothesis 1: Fiscal decentralization has a significant impact on health outcomes.

Financial decentralization and public health

Most research has demonstrated the benefits of FD for public health. Robalino, Picazo, and Voetberg²⁷ evaluated cross-country panel data from 1970 to 1995 and found that FD enhanced health outcomes. Schwartz, Guilkey, and Racelis²⁸ evaluated fiscal devolution's influence on per capita health spending. From 1995 to 1998, fiscal devolution boosted per capita health spending in the Philippines. Cantarero and Pascual²⁹ found a link between FD and public health in the developed countries. With a larger degree of FD, sub-governments understand the genuine requirements of local citizens and may focus on the most vital issues, such as healthcare.

From 1992 to 2003, FD was negatively associated with infant mortality and positively associated with life expectancy.²⁹ Samadi, Keshtkaran, Kavosi, and Vahedi³⁰ demonstrate FD reduces infant mortality in 19 OECD nations and Iran, respectively. Sanogo³¹ used local government revenue data from 2001 to 2011 for 115 municipalities to show that FD positively influences access to public services in Côte d'Ivoire municipalities. Xu and Lin³² employed a two-way fixed effect model, threshold regression, and intermediate effect models using 2008–2019 panel data from 30 mainland provinces and 4 municipalities and autonomous areas to evaluate FD's influence on public health. According to the study, FD increases public health spending.

On the other hand, scholars have found that FD is unsuitable for public health. Local governments’ actions after FD receive less attention. Spending on public healthcare will affect local public health performance. This spending component is sometimes a low priority for local governments. Thus, FD's impact on public health encompasses direct and indirect effects. Jin and Sun³³ analyzed data from 32 Chinese provinces and discovered that FD did not lower the infant mortality rate, a valuable public health indicator. Limited research has examined ER’ health effects.

Schwartz et al.²⁸ found that after decentralization in the Philippines, local governments reduced the share of funds given to public health. Ferrario and Zanardi³⁴ believe decentralization in Italy may worsen regional health expenditure disparities due to a lack of mechanisms to transfer resources from rich to poor jurisdictions. Due to an excellent fiscal equalization mechanism, decentralizing public health care in Spain has not increased per capita public expenditure inequality.²⁹ Based on the above analysis, this article offers hypothesis:

Hypothesis 2: Environmental regulation has a significant impact on health outcomes.

Environmental degradation, financial decentralization, and public health

Today, economic development and environmental conservation are vital to emerging countries’ sustainable development. Under FD, local environmental behavior is crucial to environmental governance. ER reduces carbon emissions, improves energy efficiency, and solves pollution externalities,¹² like health outcomes. Scholars believe decentralization is essential for economic development but disagree on environmental impacts.³⁵ Literature finds two generations explaining the impact of FD on environmental degradation. The supporters of the first generation of FD find some experts feel FD is effective for environmental pollution control.³⁵ FD may push developed and undeveloped regions to use unorthodox development methods.

Local governments must improve the economy and people's lives under FD. Moreover, the academics proposing the “race to the top” thesis argue that governments are inclined to improve public services and environmental quality through increased monetary autonomy after being threatened by “voting by foot.”³⁶ Second-generation FD theory argues local governments, spurred by tax revenues, degrade regional environments.³⁷ Many researchers believe FD will exacerbate haze pollution by weakening environmental supervision and capturing the market.³⁸ Local governments don’t enforce restrictions or levy modest fines on polluters, increasing pollution. Therefore, it is important to check the health outcomes of environmental degradation and FD and determine whether the local government's development orientation is a “race to the bottom” or a “race to the top.” Based on the discussion, this article offers hypothesis:

Hypothesis 3: fiscal decentralization and environmental regulation have a significant mutual/joint impact on health outcomes.

Research gap

As demonstrated by the overview above, it is clear that most studies have examined the bilateral relationships between FD and health outcomes.^28,29,31,32 However, little work is done to explain the impact of ER on health outcomes. In addition, no study has yet discovered the combined effect of environmental degradation and FD on health outcomes. This work incorporates BRICS as a case study to evaluate the separate and combined impact of FD and ER on health outcomes.

Data and methodology

Data description

This study used health outcomes as the dependent variable. Health is measured through three proxies: life expectancy, number of infant deaths, and health expenditures as a percent of GDP. The data on health proxies, that is, health expenditures, life expectancy, and the number of infant deaths, are taken from World Development Indicators (WDI). Life expectancy estimates the average predicted lifespan based on current mortality. Infant deaths, a globally recognized indicator of the socioeconomic level and population health, reflect regional health care, education, and economic growth. Similar to health status, health expenditures reflect socioeconomic development and quality of life in a population. The state of health reflects the standard of living and average lifespan, and the prospects for its economy. From a policy point of view, it is important to understand the factors that affect health outcomes.

There are two focus variables, ER and FD. ER is measured by the environmental policy stringency index, taken from OECD data. ER boost the health status of individuals. However, empirical research on ER and health spending is inconsistent.¹⁸ Two ways may help explain this inconsistency. First, environmental management can lower health costs by enhancing human health.^18,23,24 Second, some studies imply that ER may not benefit human health and may even worsen it, leading to increased health expenditures.²⁵

FD is measured by expenditures, revenue, and tax decentralization. The data on these variables are collected from the OECD dataset. FD involves collecting revenues and spending among several levels of government.³⁹ The influence of FD is dual on social welfare. According to first-generation federalism, local governments (agencies) behave as beneficent agents acting in the public interest, for example, social welfare,¹⁴ by allowing local control over spending (ED per se). Second-generation fiscal federalists argue that national governments are more inclined to pursue their goals.⁸ Hence, local governments can raise a lot of revenue (RD). It would make them more accountable to citizens, inventive with public goods, and less corrupt.

A “free-rider” mentality is common in regions with low fiscal revenues because of the negative externalities of air pollution and the positive externalities of atmospheric environmental governance. This makes environmental management less effective at boosting local economic growth, discourages investments in environmental governance, cuts spending on R&D for green technologies, relaxes local environmental rules, etc.⁴⁰ and vice versa. Henceforth, the combined impact of ER and FD on health outcome is unclear. Urban population growth, fixed capital formation, and CO₂ emissions are control variables. Data of control variables are taken from WDI. Urbanization is a two-fold phenomenon, with the first fold having beneficial effects such as easy access to jobs, healthcare, and education. In contrast, the second fold has adverse effects, such as social deprivation brought on by overcrowding. Further, gross fixed capital formation is the control variable.⁴¹ Energy generation hurts the environment because it involves chemical reactions with oxygen in the air. It pollutes and releases CO₂. Global warming results. Reduced snow and sea ice cause coastal flooding. Warmer temperatures can cause animal extinction. Rising global temperatures can cause health issues, including headaches, dizziness, restlessness, tingling, difficulty breathing, sweating, weariness, increased heart rate, high blood pressure, etc.

WDI, IMF, and the OECD provide 2000–2020 data for all variables. Data over the last 21 years covers four BRICS economies: Brazil, Russia, China, and South Africa; we exclude India due to the unavailability of FD data. We analyzed the BRICS economies for many reasons. First, these five countries account for 23% of the global GNI, 45% of the global population, and 40% of the global disease burden. By 2050, the world's two largest economies will be China and India, with Russia and Brazil coming in fifth and sixth, respectively. Second, due to economic progress, carbon dioxide emissions keep rising, endangering the health of BRICS citizens. Considering the inevitable economic growth of the BRICS countries and the need for fossil fuels, their CO₂ emissions may continue to be high, harming the ecosystem and climate. Also, it will affect medical and health spending. Thus, BRICS economies provide an intriguing case to measure the separate and combined influence of ER and FD on health outcomes.

Model specification

We have used FD and ER as the main explanatory variables in the estimated model. For FD, we used three proxies: RD, TRD, and ED. Canare (2021)⁴⁷ suggested that RD and fiscal independence have a very positive and significant impact with the performance outcomes of the governments, while ED has no clear trend. Therefore, rather than a composite indicator of FD, we have used separate indicators of FD to dig deeper into the impact of these indicators on health outcomes.

The general model form of this study is as follows:

H e a l t h o u t c o m e s_{i t} = f (E R_{i t}, F D_{i t}, E R * F D_{i t}, U R B_{i t}, G F C F_{i t}, C O 2_{i t})

(1)

The following standard model shows that ER, FD which is measured through RD, TRD, ED, the combined impact of ER and FD is introduced by interaction terms of ER with RD, TRD, and ED, respectively, urban population growth (URB), gross fixed capital formation (GFCF), and carbon dioxide emissions (CO₂) are functions of health outcomes namely health expenditures (HE), life expectancy (LE), and the number of infant deaths (ID). To test the concept empirically, it can be turned into a regression model in the direction of Tu, Feng, and Zhao (2022)⁴⁸ as follows:

Equations (2) to (4) are estimated using health expenditures (HE) as dependent variable.

H E_{i t} = α_{1} + α_{2} E R_{i t} + α_{3} R D_{i t} + α_{4} E R * R D_{i t} + α_{5} U R B_{i t} + α_{6} G F C F_{i t} + α_{7} l C O 2_{i t} + ε_{i t}

(2)

H E_{i t} = α_{1} + α_{2} E R_{i t} + α_{3} T R D_{i t} + α_{4} E R * T R D_{i t} + α_{5} U R B_{i t} + α_{6} G F C F_{i t} + α_{7} l C O 2_{i t} + ε_{i t}

(3)

H E_{i t} = α_{1} + α_{2} E R_{i t} + α_{3} E D_{i t} + α_{4} E R * E D_{i t} + α_{5} U R B_{i t} + α_{6} G F C F_{i t} + α_{7} l C O 2_{i t} + ε_{i t}

(4)

Equations (5) to (7) are estimated using life expectancy at birth (LE) as dependent variable.

L E_{i t} = β_{1} + β_{2} E R_{i t} + β_{3} R D_{i t} + β_{4} E R * R D_{i t} + β_{5} U R B_{i t} + β_{6} G F C F_{i t} + β_{7} l C O 2_{i t} + ε_{i t}

(5)

L E_{i t} = β_{1} + β_{2} E R_{i t} + β_{3} T R D_{i t} + β_{4} E R * T R D_{i t} + β_{5} U R B_{i t} + β_{6} G F C F_{i t} + β_{7} l C O 2_{i t} + ε_{i t}

(6)

L E_{i t} = β_{1} + β_{2} E R_{i t} + β_{3} E D_{i t} + β_{4} E R * E D_{i t} + β_{5} U R B_{i t} + β_{6} G F C F_{i t} + β_{7} l C O 2_{i t} + ε_{i t}

(7)

Further, equations (8) to (10) are estimated using infant death (ID) as dependent variable.

l I D_{i t} = γ_{1} + γ_{2} E R_{i t} + γ_{3} R D_{i t} + γ_{4} E R * R D_{i t} + γ_{5} U R B_{i t} + γ_{6} G F C F_{i t} + γ_{7} l C O 2_{i t} + ε_{i t}

(8)

l I D_{i t} = γ_{1} + γ_{2} E R_{i t} + γ_{3} T R D_{i t} + γ_{4} E R * T R D_{i t} + γ_{5} U R B_{i t} + γ_{6} G F C F_{i t} + γ_{7} l C O 2_{i t} + ε_{i t}

(9)

l I D_{i t} = γ_{1} + γ_{2} E R_{i t} + γ_{3} E D_{i t} + γ_{4} E R * E D_{i t} + γ_{5} U R B_{i t} + γ_{6} G F C F_{i t} + γ_{7} l C O 2_{i t} + ε_{i t}

(10)

Where α's, β's, and γ's are the coefficients. The subscripts “i denote cross-sections” and “t,” which are 4 and 21, respectively, l for log, t for time, and ε is also the regression model's error term.

Econometric technique

This study computed descriptive statistics for health expenditures, life expectancy, infant deaths, ER, FD (expenditures, revenue, and tax), urban population growth, gross fixed capital formation, and carbon dioxide emissions: More specifically, this study estimates the mean, median, and range for each variable. In addition, standard deviation values are calculated, which measure each observation's distance from the mean and duplicate variable volatility. As a logical first step, the study checks for data normality using skewness and Kurtosis. Jarque and Bera⁴⁹ developed a complete normality test. This test uses skewness and excess Kurtosis to determine if data is normally distributed. The following equation gives normality statistics.

J B = \frac{N}{6} (S^{2} + \frac{{(K - 3)}^{2}}{4})

(11)

Jarque-normality Bera's test considers the data are assumed normal distribution as the null hypothesis, but statistically significant estimates may challenge this assumption.

Slope heterogeneity and cross-sectional dependence

After examining variable regularity and irregularity, the current study used slope heterogeneity and panel cross-section dependence. ER are improvising after the massive CO₂ emissions. Various influences push economies to depend on others. Specifically, suppose a country implements ER or changes in decentralization methods in one country. If so, this affects other economies and assists in achieving technological, financial, social, and economic objectives. As a result, economies may have similarities and contrasts with others. Slope homogeneity and cross-section dependence in panel data could produce estimation issues in econometric analysis.⁵⁰ To determine whether a phenomenon is homogeneous or heterogeneous, this study uses the Pesaran and Yamagata⁵¹ slope coefficient homogeneity (SCH) test and the Pesaran⁵² cross-sectional dependence test. The SCH equation is:

Δ_{S C H} = \sqrt{N {(2 k)}^{- 1}} (N^{- 1} S - K)

(12)

Additionally, the above test provides estimated results for the adjusted SCH, which are as follows:

Δ_{S C H} = \sqrt{N} . \sqrt{\frac{T + 1}{2 K . (T - K - 1)}} . (N^{- 1} S - 2 K)

(13)

The null hypothesis of this test implies homogeneous slope coefficients, while the alternative hypothesis shows heterogeneous slope coefficients when significant estimates are established. After estimating slope coefficients, the recent study performed the Pesaran⁵² cross-section dependence test between BRICS economies. Neglecting this problem could result in estimation bias.⁵³ Standard form for evaluating cross-section dependency:

C D_{T e s t} = \frac{\sqrt{2 T}}{{[N . (N - 1)]}^{\frac{1}{2}}} \sum_{i = 1}^{N - 1} \sum_{K = 1 + i}^{N} T_{i k}

(14)

The test's null hypothesis suggested BRICS cross-sections are independent. Once significant estimates are established, alternate cross-section dependence hypotheses can be accepted.

Unit root tests

After confirming heterogeneous slope coefficients and cross-sectional dependency, this study applied the second-generation unit root test. We employed Pesaran⁵⁴ cross-sectionally enhanced IPS (CIPS) unit root test.

Pesaran⁵⁵ introduced a cross-sectional factor modeling method. Cross-sectional averages are unified as unobserved model components. Pesaran⁵⁴ expanded the Augmented Dickey-Fuller (ADF) regression model to include mean and first differenced cross-sectional lags. This method allows for panel data issues, such as cross-sectoral reliance, when the panel is unbalanced (N≠T). Standard form regression equation of cross-sectional ADF:

Δ y_{i, t} = θ_{i} + β *_{i} y_{i, t - 1} + d_{0} {\bar{y}}_{t - 1} + d_{1} {\bar{y}}_{t} + ε_{i t}

(15)

In the given equation,

y_{t}

is the average observation. To address serial correlation, equation (15) can be translated by adding initial difference lags of

y_{i t}

and

{\bar{y}}_{t}

t = θ_{i} + β *_{i} y_{i, t - 1} + d_{0} {\bar{y}}_{t - 1} + \sum_{j = 0}^{n} d_{j + 1} Δ {\bar{y}}_{t - j} + \sum_{k = 1}^{n} c_{k} Δ y_{i, t - k} + ε_{i t}

(16)

Pesaran⁵⁴ built the CIPS, which we use in BRICS nations, by averaging t-statistics for each cross-sectional unit, known as

C A D F_{i}

. It may be computed as shown in equation (17).

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

(17)

When the estimates are statistically significant, the CIPS unit root test's presumption that the time series is non-stationary or has a unit root can be rejected.

Method of moment quantile regression

Firstly, Koenker and Bassett Jr⁵⁶ proposed a panel quantile estimation approach that evaluates the dependent variance and conditional mean statistics. Quantile regression gives accurate results even with irregularly distributed variables. Following the property of quantile regression, the current study employed Machado and Silva⁵⁷ moment's quantile regression. This method evaluates distributional and heterogeneous quantile effects.⁵⁸ The typical form for location-scale $Q_{y} (τ | X)$ estimates are as follows:

Y_{i t} + ϑ X_{i t} + (δ_{i} + ρ {\overset{'}{Z}}_{i t}) . μ_{i t}

(18)

The preceding equation shows

P . (δ_{i} + ρ {Z^{'}}_{i t} > 0) = 1

, where p is the probability (.). Moreover,

θ, ϑ, δ, and ρ

are parameters to be estimated. The subscript i shows the fixed impact of

θ_{i} a n d δ_{i}

, i = 1, 2, 3…n, and Z exhibits k-vector of predictable X elements that are variation conversions

\sim

as follows:

Z_{\sim} = Z_{\sim} (X), \sim = 1, 2, 3 \dots . k

(19)

In equation (19), X is distributed independently for each i and t. It is orthogonal to X and may be distributed across fixed cross-sections and time,⁵⁷ stabilizing the remaining components and anticipating excessive exogenic behavior. Thus, equations (2) to (10) become:

Q_{y} (τ X_{i t}) = (θ_{i} + δ_{i} q (τ)) + ϑ X_{i t} + p \overset{'}{Z_{i t}} q (τ)

(20)

From equation (20), X is the independent variables’ vector, including ER and FD. Additionally,

Q_{y} (τ X_{i t})

in the equation above displays the quantile distribution. Health is the dependent variable, and its estimate will be conditional on the locations of the dependent variables. Moreover,

- Q_{y} (τ) \equiv θ_{i} + δ_{i} q . (τ)

is a scalar coefficient of quantile

τ

for each cross-section (i). Individual effects don’t control intercept shift, unlike least square fixed effects. Due to variables’ time-invariance, heterogeneous influence can shift across quantiles. q (

τ

) also shows the

τ - th

quantile sample, which is 25th, 50th, 75th, and 90th. Each quantile's equation is as follows:

m i n_{q} \sum_{i} \sum_{t} γ_{t} (R_{i t} - (δ_{i} + ρ {\overset{'}{Z}}_{i t}) q)

(21)

Where

γ_{t} (A) = (τ - 1) . A I {A \leq 0} + T A I {A > 0

(22)

specifies check function.

Panel causality test

The method of moment quantile regression (MMQREG) technique shows the long-run influence of each explanatory component on health outcomes at each quantile. This estimator doesn’t indicate causality between variables. The current study used Dumitrescu and Hurlin⁵⁹ Granger panel causality assumptions. This specification covers the T ≠ N panel issue. This method addresses slope heterogeneity and cross-section dependence (Table 1).⁶⁰

Table 1.

Description of variables.

Variables	Role	Measurements	Sources	References
Life expectancy	DP	Yearly	WDI	Mahalik, Le, Le, & Mallick; Nakatani, Zhang, & Garcia Valdes^42,43
Number of infant deaths (Log)	DP	Per 1000 live births	WDI	Bellinger; Nakatani et al.^43,44
Health expenditures in US dollars (Log)	DP	percent of GDP	WDI	Benali; Xu & Lin^32,41
Environmental policy stringency index	IND	Index	OECD	Yirong⁴⁵
Expenditure decentralization	IND	Ratio of own spending to general government spending	IMF	Xu & Lin³²
Revenue decentralization	IND	Ratio of own revenue to general government revenue	IMF	Xu & Lin³²
Tax revenue decentralization	IND	Share of general government	IMF	Tunio & Nabi⁴⁶
Urban population growth	CV	Percent	WDI	Xu & Lin³²
Gross fixed capital formation	CV	Percent of GDP	WDI	Benali⁴¹
CO₂ emissions	CV	Kilotons	WDI	Mahalik et al.⁴²

Note: DP, IND, and CV indicate dependent, independent, and control variables, respectively.

Results and discussion

Descriptive statistics

This section presents the study findings starting from descriptive statistics and data diagnostics to determine the correct type of estimation technique for our data. In descriptive statistics (Table 2), the mean values of all variables are positive. All factors are trending up. All variables, except urban population growth (minimum), have positive maximum–minimum ranges. Urban population growth readings ranged from negative 0.467% to positive 4.189%. The standard deviation, a measure of a variable's volatility across time, demonstrates that all variables are time varying or fluctuating.

Table 2.

Descriptive statistics.

Variable	Obs.	Mean	Std. Dev.	Min	Max
Health expenditures percent of GDP	84	6.039	2.129	2.936	9.591
Life expectancy at birth	84	68.634	6.307	53.444	77.097
Number of infant deaths (Log)	84	10.802	1.029	8.959	13.088
Environmental policy stringency index	84	0.898	0.524	0.375	2.163
Expenditure decentralization	63	0.455	0.107	0.306	0.625
Revenue decentralization	84	0.603	0.160	0.370	0.827
Tax revenue decentralization	84	0.655	0.160	0.480	0.947
Urban population growth	84	1.838	1.158	−0.467	4.198
Gross fixed capital formation percent of GDP	84	25.434	9.588	13.050	44.519
CO₂ emissions in kilotons (Log)	84	14.056	1.128	12.647	16.186

Source: Author's calculations.

Normality tests

This study additionally verified each variable's normality using skewness and Kurtosis tests. Table 3 presents the empirical results of the normality test. The joint test of skewness and kurtosis and Jarque and Bera⁴⁹ offer significant estimates for all variables. This test accounts for skewness and excess Kurtosis. The null hypothesis showed the variable might be normally distributed since chi (2) values for all variables except urban population growth are above 0.05. The null hypothesis can be rejected. Hence, the distribution of the variable seems to be abnormal. Since most empirical estimates are limited in managing irregular data, the current work employed an efficient estimator that empirically evaluates long-run results by addressing variable abnormality.

Table 3.

Normality tests.

Variable	Skewness and kurtosis Tests joint		Jarque-Bera test
Variable	Joint chi²(2)	Prob > chi²	chi²(2)	Prob > chi²
Health expenditures percent of GDP	4.012	0.1235	10.66	0.0048
Life expectancy at birth	9.61	0.0082	12.28	0.002
Number of infant deaths (Log)	2.87	0.23	2.55	0.27
Environmental policy stringency index	16.06	0.0003	25.66	0.000
Expenditure decentralization	33.96	0.000	6.78	0.033
Revenue decentralization	3.872	0.1518	8.71	0.013
Tax revenue decentralization	12.95	0.0015	15.75	0.000
Urban population growth	3.91	0.1418	3.14	0.21
Gross fixed capital formation	14.77	0.0006	12.34	0.0021
CO₂ emissions in kilotons (Log)	12.37	0.0021	7.66	0.021

Source: Author's calculations.

Slope heterogeneity test

As noted before, a country depends on other countries for economic and non-economic reasons, resulting in similarities and differences in specific ways. Table 4 shows the findings of Pesaran and Yamagata⁵¹ SCH test. Inefficient estimation may result from disregarding slope heterogeneity/homogeneity. It is necessary to analyze slope heterogeneity. Following the null hypothesis of homogeneous slopes, both SCH (Delta) and adjusted SCH (Delta Adjusted) are statistically significant. This shows that the null hypothesis can be rejected, and slope coefficients are heterogeneous.

Table 4.

Testing for slope heterogeneity.

Slope heterogeneity test	Statistics
Delta	23250.487***
Delta adjusted	32881.154***

Source: Author's calculations.

*** p < 0.01, ** p < 0.05, * p < 0.1.

Cross-sectional dependence test

Next, as suggested by Campello et al.,⁵³ cross-sectional dependency in panel data causes estimation bias. We have applied Pesaran⁵² CD test (Table 5). All variables, except environmental policy stringency index and ED, are highly statistically significant, rejecting the null hypothesis of cross-sectional independence in life expectancy, number of infant deaths, RD, TRD, population growth, health expenditures, and gross fixed capital formation. These variables are cross-sectional dependent, illustrating that these variables of one country have a spillover effect on the variables of another. The null hypothesis of cross-sectional dependence for environmental policy stringency and ED cannot be rejected. Thus, these three variables do not have a spillover effect on other economies.

Table 5.

Cross-sectional dependence.

Variable	Statistics
Health expenditures percent of GDP	7.454***
Life expectancy at birth	7.653***
Number of infant deaths (Log)	6.932***
Environmental policy stringency index	0.456
Expenditure decentralization	−0.361
Revenue decentralization	2.368***
Tax revenue decentralization	4.809***
Urban population growth	2.094*
Gross fixed capital formation percent of GDP	4.516***
CO₂ emissions in kilotons (Log)	6.776***

Source: Author's calculations.

*** p < 0.01, ** p < 0.05, * p < 0.1.

Unit root test

Since the variables revealed slope heterogeneity and cross-sectional dependence, this study used Pesaran⁵⁴ CIPS test, a second-generation unit root testing approach, and the empirical results are given in Table 6. Only life expectancy yields statistically significant estimates at I(0), rejecting the null hypothesis of a unit root in time variables. The remaining variables are non-stationary at level. This study evaluated stationarity at I(1), where these variables produce significant estimates that reject the unit root null hypothesis. All variables are stationary under mixed order of integration.

Table 6.

Unit root testing (Pesaran panel unit root test).

Variables	Intercept and trend
Variables	I(0)	I(1)
Health expenditures percent of GDP	−2.26	−3.98***
Life expectancy at birth	−3.797***	−2.135
Number of infant deaths (Log)	−1.24	−2.95*
Environmental policy stringency index	−2.31	−4.37***
Expenditure decentralization	−3.56	−5.078***
Revenue decentralization	−3.7	−5.71***
Tax revenue decentralization	−2.57	−4.72***
Urban population growth	−1.54	−3.52***
Gross fixed capital formation percent of GDP	−2.009	−3.14**
CO₂ emissions in kilotons (Log)	−1.79	−3.4***

Source: Author's calculations.

*** p < 0.01, ** p < 0.05, * p < 0.1.

The Jarque and Bera⁴⁹ test found that the variables are not normally distributed. Therefore, we have used the MMQREG, which handles non-normal variables. Table 7 shows the approach's estimated results.

Table 7.

Quantile regression estimates (revenue decentralization).

Dependent variable: Health expenditures (% of GDP)			Quantiles
Variables	Location	Scale	Qtile_25	Qtile_50	Qtile_75
Environmental policy stringency index	2.85***	0.84**	2.06***	2.48***	3.48***
Environmental policy stringency index	(−0.56)	(−0.37)	(−0.56)	(−0.55)	(−0.69)
Revenue decentralization	3.02	2.28*	0.88	2.03	4.72**
Revenue decentralization	(−1.86)	(−1.24)	(−1.91)	(−1.84)	(−2.27)
Regulation & decentralization interaction	−3.22***	−1.83***	−1.5	−2.42**	−4.58***
Regulation & decentralization interaction	(−0.95)	(−0.63)	(−0.93)	(−0.95)	(−1.19)
Urban population growth	0.16	0.05	0.11	0.14	0.19
Urban population growth	(−0.15)	(−0.1)	(−0.16)	(−0.14)	(−0.18)
Gross fixed capital formation percent of GDP	0.03	−0.03	0.05	0.04	0.01
Gross fixed capital formation percent of GDP	(−0.05)	(−0.03)	(−0.05)	(−0.05)	(−0.06)
CO₂ emissions in kilotons (Log)	−1.95***	0.16	−2.10***	−2.02***	−1.83***
CO₂ emissions in kilotons (Log)	(−0.38)	(−0.25)	(−0.41)	(−0.37)	(−0.46)
Constant	30.31***	−2.45	32.61***	31.38***	28.48***
Constant	(−5.49)	(−3.66)	(−5.85)	(−5.42)	(−6.6)
Observations	84	84	84	84	84

Note: Standard errors in parentheses.

Source: Author's calculations.

*** p < 0.01, ** p < 0.05, * p < 0.1.

Method of moment quantile regression results

According to our estimations, ER and RD positively affect health expenditures. One unit increase in ER causes an increase in health expenditures by 2.06% in lower quantiles to 3.48% in upper quantiles. Statistically, these findings are significant at 1%. However, the combined impact of ER and RD negatively affects health expenditures. The finding indicates that the impact of each variable is getting larger with a higher quantile. In control variables, CO₂ emissions’ impact on health expenditure is negative and significant: a 1% increase in CO₂ emissions causes a 32.61% to 28.48% reduction in health expenditures across three quantiles (Qtile_25 to Qtile_75).

Table 8 includes the second FD proxy, TRD. Again, ER is positively correlated with health expenditures in BRICS economies. MMQREG's empirical estimation reveals that ER potentially impacts health expenditures as a 1 unit increase in the ER increases health expenditures by 2.35 and 3.58 in the lower to upper quantiles (Qtile_25, Qtile_ 75). Similar results are presented by Do et al.¹⁸ and Greenstone and Hanna.²⁵ Statistically, these findings are significant at 1%. The combined impact of ER and TRD is statistically negative and significant in the lower and middle quantile. This situation presents the “race to the top” where governments are inclined to improve public services and environmental quality through increasing monetary autonomy after “voting by foot.”³⁶

Table 8.

Quantile regression estimates (tax revenue decentralization).

Dependent variable: Health expenditures (% of GDP)			Quantiles
Variables	Location	Scale	Qtile_25	Qtile_50	Qtile_75
Environmental policy stringency index	2.93***	0.66	2.35***	2.56***	3.58***
Environmental policy stringency index	(−0.78)	(−0.64)	(−0.41)	(−0.53)	(−1.37)
Tax revenue decentralization	−0.62	−0.74	0.02	−0.21	−1.35
Tax revenue decentralization	(−1.3)	(−1.07)	(−0.69)	(−0.87)	(−2.27)
Regulation & decentralization interaction	−2.27*	−0.69	−1.66***	−1.89**	−2.96
Regulation & decentralization interaction	(−1.2)	(−0.99)	(−0.64)	(−0.8)	(−2.11)
Urban population growth	0.35**	0.19	0.19**	0.25**	0.53*
Urban population growth	(−0.16)	(−0.13)	(−0.08)	(−0.11)	(−0.28)
Gross fixed capital formation percent of GDP	−0.01	−0.04	0.03	0.02	−0.05
Gross fixed capital formation percent of GDP	(−0.05)	(−0.04)	(−0.02)	(−0.03)	(−0.08)
CO₂ emissions in kilotons (Log)	−2.11***	−0.02	−2.09***	−2.10***	−2.13***
CO₂ emissions in kilotons (Log)	(−0.26)	(−0.22)	(−0.14)	(−0.17)	(−0.46)
Constant	34.89***	1.81	33.31***	33.89***	36.68***
Constant	(−3.16)	(−2.6)	(−1.68)	(−2.11)	(−5.54)
Observations	84	84	84	84	84

Note: Standard errors in parentheses.

Source: Author's calculations.

*** p < 0.01, ** p < 0.05, * p < 0.1.

Table 9 includes the third proxy of FD, that is, ED. The ED is positively related to health expenditures. Such findings are consistent with Schwartz et al.²⁸ A 1 unit increase in ED increases health expenditures by 15.64%, 11.49%, and 6.65% in Qtile_25, Qtile_50, and Qtile_75, respectively. Lower to middle quantile estimates are more statistically significant (s). However, the lower to middle quantile indicates a comparatively lower impact of FD on health expenditures; decentralization may lower or worsen regional health expenditure due to a lack of resource transfer channels.³⁴ The results are significant at 1%, but the third (Qtile_75) quantile is insignificant. The impact of the interaction of ER and ED reveals an insignificant impact on health expenditures. All the control variables yield the expected signs.

Table 9.

Quantile regression estimates (expenditure decentralization).

Dependent variable: Health expenditures (% of GDP)			Quantiles
Variables	Location	Scale	Qtile_25	Qtile_50	Qtile_75
Environmental policy stringency index	1.14	−3.17**	4.26	1.34	−2.07
Environmental policy stringency index	(−2.51)	(−1.45)	(−2.67)	(−2.56)	(−3.26)
Expenditure decentralization	11.21***	−4.49**	15.64***	11.49***	6.65
Expenditure decentralization	(−3.37)	(−1.95)	(−3.61)	(−3.46)	(−4.39)
Regulation & decentralization interaction	−1.41	7.43**	−8.73	−1.89	6.11
Regulation & decentralization interaction	(−6)	−3.47	(−6.39)	(−6.12)	(−7.78)
Urban population growth	1.28***	−0.35**	1.63***	1.31***	0.93**
Urban population growth	(−0.29)	(−0.17)	(−0.31)	(−0.29)	(−0.37)
Gross fixed capital formation percent of GDP	−0.18***	0.02	−0.20***	−0.18***	−0.16***
Gross fixed capital formation percent of GDP	(−0.03)	(−0.02)	(−0.03)	(−0.03)	(−0.04)
CO₂ emissions in kilotons (Log)	0.86	−0.46	1.31**	0.89	0.4
CO₂ emissions in kilotons (Log)	(−0.54)	(−0.31)	(−0.56)	(−0.54)	(−0.69)
Constant	−7.69	8.55*	−16.11*	−8.24	0.96
Constant	(−8.02)	(−4.64)	(−8.47)	(−8.13)	(−10.36)
Observations	84	84	84	84	84

Note: Standard errors in parentheses.

Source: Author's calculations.

*** p < 0.01, ** p < 0.05, * p < 0.1.

Tables 10 to 12 and 13 to 15 provide the estimated results of life expectancy and the number of infant deaths, respectively, under ER, three proxies of FD, and interaction terms of ER and specific proxy FD (tax revenue, revenue, and ED. In the life expectancy case, the impact of ER is higher with a higher quantile. The impact of RD is negative and significant but declining with higher quantile: 1 unit increase in RD reduces life expectancy by 69.40%, 51.39%, and 34.73% in Qtile_25, Qtile_50, and Qtile_75, respectively. The combined impact of ER and RD is negative and significant in the middle and upper quantile. Similarly, the impact of TRD is negative, significant, and declining from lower to upper quantile (50.97% to 39.50%).

Table 10.

Quantile regression estimates.

Dependent variable: Life expectancy at birth			Quantiles
Variables	Location	Scale	Qtile_25	Qtile_50	Qtile_75
Environmental policy stringency index	9.81*	0.06	9.77	9.82**	9.87***
Environmental policy stringency index	(−5.21)	(−4.86)	(−8.34)	(−4.47)	(−2.79)
Revenue decentralization	−55.11***	20.14	−69.40**	−51.39***	−34.73***
Revenue decentralization	(−19.68)	(−18.36)	(−30.51)	(−16.77)	(−10.55)
Regulation & decentralization interaction	−17.84**	−6.72	−13.07	−19.08**	−24.64***
Regulation & decentralization interaction	(−8.64)	(−8.06)	(−13.77)	(−7.45)	(−4.64)
Urban population growth	0.05	0.05	0.06	0.04	0.05**
Urban population growth	(−0.01)	(−0.01)	(−0.01)	(−0.01)	(−0.01)
Gross fixed capital formation percent of GDP	0.14	−0.31	0.36	0.08	−0.17
Gross fixed capital formation percent of GDP	(−0.54)	(−0.5)	(−0.85)	(−0.46)	(−0.29)
CO₂ emissions in kilotons (Log)	−8.98**	2.24	−10.57*	−8.56***	−6.70***
CO₂ emissions in kilotons (Log)	(−3.6)	(−3.36)	(−5.69)	(−3.08)	(−1.93)
Constant	222.32***	−31.99	245.02***	216.41***	189.94***
Constant	(−55.41)	(−51.69)	(−87.76)	(−47.43)	(−29.69)
Observations	84	84	84	84	84

Note: Standard errors in parentheses.

Source: Author's calculations.

*** p < 0.01, ** p < 0.05, * p < 0.1.

Table 11.

Quantile regression estimates.

Dependent variable: Life expectancy at birth
Variables	Location	Scale	Qtile_25	Qtile_50	Qtile_75
Environmental policy stringency index	3.07	1.92	1.21	3.13	5.13
Environmental policy stringency index	(−3.23)	(−1.57)	(−3.36)	(−3.27)	(−3.84)
Tax revenue decentralization	−45.52***	5.62**	−50.97***	−45.33***	−39.50***
Tax revenue decentralization	(−5.77)	(−2.81)	(−6)	(−5.96)	(−6.82)
Regulation & decentralization interaction	−0.96	−2.57	1.52	−1.05	−3.71
Regulation & decentralization interaction	(−5.52)	(−2.69)	(−5.75)	(−5.58)	(−6.58)
Urban population growth	0.71	0.64**	0.09	0.73	1.40**
Urban population growth	(−0.59)	(−0.29)	(−0.62)	(−0.61)	(−0.7)
Gross fixed capital formation percent of GDP	0.12	−0.21***	0.33*	0.11	−0.11
Gross fixed capital formation percent of GDP	(−0.17)	(−0.08)	(−0.18)	(−0.18)	(−0.2)
Co2 emissions in kilotons (Log)	−2.76***	1.14***	−3.87***	−2.72***	−1.54
Co2 emissions in kilotons (Log)	(−0.88)	(−0.43)	(−0.91)	(−0.92)	(−1.04)
Constant	131.44***	−13.62***	144.62***	130.98***	116.86***
Constant	(−9.91)	(−4.82)	(−10.3)	(−10.46)	(−11.64)
Observations	84	84	84	84	84

Note: Standard errors in parentheses.

Source: Author's Calculations

*** p < 0.01, ** p < 0.05, * p < 0.1.

Table 12.

Quantile regression estimates.

Dependent variable: Life expectancy at birth			Quantiles
Variables	Location	Scale	Qtile_25	Qtile_50	Qtile_75
Environmental policy stringency index	−4.20***	1.34***	−5.48***	−4.15***	−2.85***
Environmental policy stringency index	(−0.83)	(−0.49)	(−1.04)	(−0.88)	(−0.92)
Expenditure decentralization	−5.99***	1.51**	−7.44***	−5.93***	−4.46***
Expenditure decentralization	(−1.06)	(−0.62)	(−1.32)	(−1.11)	(−1.17)
Regulation & decentralization interaction	9.74***	−3.02***	12.64***	9.63***	6.70***
Regulation & decentralization interaction	(−1.93)	(−1.13)	(−2.41)	(−2.04)	(−2.13)
Urban population growth	−0.45***	−0.05	−0.40***	−0.45***	−0.50***
Urban population growth	(−0.08)	(−0.05)	(−0.1)	(−0.08)	(−0.09)
Gross fixed capital formation percent of GDP	0.05***	0.03	0.06***	0.05***	0.05***
Gross fixed capital formation percent of GDP	(−0.01)	(−0.01)	(−0.01)	(−0.01)	(−0.01)
Co2 emissions in kilotons(Log)	−1.60***	−0.08	−1.52***	−1.60***	−1.68***
Co2 emissions in kilotons(Log)	(−0.16)	(−0.1)	(−0.2)	(−0.16)	(−0.18)
Constant	33.93***	0.69	33.27***	33.95***	34.62***
Constant	(−2.34)	(−1.37)	(−2.84)	(−2.34)	(−2.54)
Observations	84	84	84	84	84

Note: Standard errors in parentheses.

Source: Author's Calculations

*** p < 0.01, ** p < 0.05, * p < 0.1.

Table 13.

Quantile regression estimates.

Dependent variable: Number of infant deaths (Log)			Quantiles
Variables	Location	Scale	Qtile_25	Qtile_50	Qtile_75
Environmental policy stringency index	−0.05	−0.26	0.14	−0.05	−0.3
Environmental policy stringency index	(−0.29)	(−0.20)	(−0.37)	(−0.29)	(−0.28)
Revenue decentralization	−1.38	−1.04	−0.65	−1.4	−2.36**
Revenue decentralization	(−1.02)	(−0.69)	(−1.30)	(−1.03)	(−0.97)
Regulation & decentralization interaction	−0.73	0.59	−1.14	−0.72	−0.17
Regulation & decentralization interaction	(−0.66)	(−0.45)	(−0.83)	(−0.66)	(−0.63)
Urban population growth	0.54***	−0.02	0.56***	0.54***	0.52***
Urban population growth	(−0.06)	(−0.04)	(−0.07)	(−0.06)	(−0.05)
Gross fixed capital formation percent of GDP	0.08***	−0.02	0.09***	0.08***	0.07***
Gross fixed capital formation percent of GDP	(−0.02)	(−0.01)	(−0.02)	(−0.02)	(−0.02)
Co2 emissions in kilotons (Log)	−0.57***	0.03	−0.59***	−0.57***	−0.54***
Co2 emissions in kilotons (Log)	(−0.17)	(−0.11)	(−0.21)	(−0.16)	(−0.16)
Constant	17.07***	0.82	16.49***	17.09***	17.85***
Constant	(−2.57)	(−1.74)	(−3.26)	(−2.56)	(−2.43)
Observations	84	84	84	84	84

Note: Standard errors in parentheses.

Source: Author's calculations.

*** p < 0.01, ** p < 0.05, * p < 0.1.

Table 14.

Quantile regression estimates.

Dependent variable: Number of infant deaths			Quantiles
Variables	Location	Scale	Qtile_25	Qtile_50	Qtile_75
Environmental policy stringency index	−0.15	−0.04	−0.11	−0.14	−0.18
Environmental policy stringency index	(−0.25)	(−0.16)	(−0.31)	(−0.25)	(−0.28)
Tax revenue decentralization	−1.98***	0.08	−2.06***	−1.99***	−1.91***
Tax revenue decentralization	(−0.38)	(−0.24)	(−0.47)	(−0.38)	(−0.42)
Regulation & decentralization interaction	−0.17	0.08	−0.24	−0.17	−0.09
Regulation & decentralization interaction	(−0.38)	(−0.24)	(−0.47)	(−0.38)	(−0.42)
Urban population growth	0.71***	−0.05	0.75***	0.71***	0.66***
Urban population growth	(−0.06)	(−0.04)	(−0.07)	(−0.06)	(−0.06)
Gross fixed capital formation percent of GDP	0.04**	−0.02*	0.06***	0.04**	0.02
Gross fixed capital formation percent of GDP	(−0.02)	(−0.01)	(−0.02)	(−0.02)	(−0.02)
CO₂ emissions in kilotons (Log)	−0.28***	0.14**	−0.41***	−0.28***	−0.14
CO₂ emissions in kilotons (Log)	(−0.09)	(−0.06)	(−0.12)	(−0.1)	(−0.1)
Constant	14.03***	−1.36**	15.26***	14.06***	12.73***
Constant	(−1.05)	(−0.67)	(−1.3)	(−1.06)	(−1.15)
Observations	84	84	84	84	84

Note: Standard errors in parentheses.

Source: Author's calculations.

*** p < 0.01, ** p < 0.05, * p < 0.1.

Table 15.

Quantile regression estimates.

Dependent variable: Number of infant deaths			Quantiles
Variables	Location	Scale	Qtile_25	Qtile_50	Qtile_75
Environmental policy stringency index	−4.20***	1.34***	−5.48***	−4.15***	−2.85***
Environmental policy stringency index	(−0.83)	(−0.49)	(−1.04)	(−0.88)	(−0.92)
Expenditure decentralization	−5.99***	1.51**	−7.44***	−5.93***	−4.46***
Expenditure decentralization	(−1.06)	(−0.62)	(−1.32)	(−1.11)	(−1.17)
Regulation & decentralization interaction	9.74***	−3.02***	12.64***	9.63***	6.70***
Regulation & decentralization interaction	(−1.93)	(−1.13)	(−2.41)	(−2.04)	(−2.13)
Urban population growth	−0.45***	−0.05	−0.40***	−0.45***	−0.50***
Urban population growth	(−0.08)	(−0.05)	(−0.1)	(−0.08)	(−0.09)
Gross fixed capital formation percent of GDP	0.05***	0.02	0.06***	0.05***	0.05***
Gross fixed capital formation percent of GDP	(−0.01)	(−0.01)	(−0.01)	(−0.01)	(−0.01)
CO₂ emissions in kilotons (Log)	−1.60***	−0.08	−1.52***	−1.60***	−1.68***
CO₂ emissions in kilotons (Log)	(−0.16)	(−0.1)	(−0.2)	(−0.16)	(−0.18)
Constant	33.93***	0.69	33.27***	33.95***	34.62***
Constant	(−2.34)	(−1.37)	(−2.84)	(−2.34)	(−2.54)
Observations	84	84	84	84	84

Note: Standard errors in parentheses.

Source: Author's calculations.

*** p < 0.01, ** p < 0.05, * p < 0.1.

The ED is negatively and significantly related to life expectancy; 1 unit increase in the expenditure's decentralization reduces life expectancy by 7.44%, 5.93%, and 4.46% in Qtile_25 Qtile_50, and Qtile_75, respectively. Lower to middle quantile estimates have a declining trend. Moreover, the combined impact of ER and expenditures decentralization is positive and significant. ER is connected negatively with life expectancy in the ED equation in BRICS economies. However, the results are promising if more regulations are placed with higher ED. The combined effect of regulations and ED is significantly positive. The finding implies that regulations or ED alone may worsen the health standards and policymakers need to strike a balance between the two.

Another finding from Tables 10 to 12 is that ED, along with better ER, is more encouraging than the total and TRD regarding health outcomes. Tables 10 and 11 show that the separate and joint impact of ER is insignificant, along with a strong negative impact of both types of RD on life expectancy. Furthermore, the separate and joint impact of both variables (FD and ER) tends to decrease, moving from lower to higher quantiles. A lower quantile indicates a country with a lower ER and FD level, whereas higher quantile represents the prevalence of a higher level of both indicators. Combined impact provides mixed results (positive and negative) in the case of life expectancy and indicates dual theories of FD.^36,38 Similarly, ER and FD reduce the number of infant deaths in the case of the number of infant deaths.^18,29,30

Pairwise Dumitrescu Hurlin panel causality tests

MMQREG gives empirical results for each scale, location, and quantile. However, estimating causality across variables is insufficient. Table 16 shows the empirical results of the Dumitrescu and Hurlin⁵⁹ Granger panel causality test. Bidirectional and unidirectional causal links exist between the research variables. In the health expenditure model, health expenditures do not substantially cause EPSI while EPSI granger causes health expenditures. There exists a unidirectional causality between health expenditures and EPSI. Similarly, unidirectional exists in the case of RD and ED. Only health expenditures substantially cause TRD and TRD granger causes health expenditures, which indicates a two-way causal link between these variables in BRCS nations. In the life expectancy model, all variables have unidirectional causality. Only life expectancy and TRD have bisectional causality. A similar line of causality is found in the case of the infant death model.

Table 16.

Causality check: pairwise Dumitrescu Hurlin panel causality tests.

Null hypothesis	W-Stat	$\bar{Z}$ -Stat	Prob.
With health expenditures
Health expenditures ↛ EPSI	0.538	−0.635	0.51
EPSI ↛ Health expenditures	2.307*	1.8484	0.06
Health expenditures ↛ Revenue FD	4.42***	4.83	0.00
Revenue FD ↛ Health expenditures	0.92	−0.11	0.91
Health expenditures ↛ Expenditure FD	2.49*	1.82	0.06
Expenditure FD ↛ Health expenditures	1.62	0.77	0.44
Health expenditures ↛ Tax revenue FD	2.34**	1.9	0.05
Tax revenue FD ↛ Health expenditures	3.33***	3.83	0.00
With life expectancy
Life expectancy ↛ EPSI	0.85	−0.203	0.83
EPSI ↛ Life expectancy	3.52***	3.56	0.00
Life expectancy ↛ Revenue FD	2.24**	1.76	0.05
Revenue FD ↛ Life expectancy	0.55	−0.63	0.52
Life expectancy ↛ Expenditure FD	3.22***	2.72	0.00
Expenditure FD ↛ Life expectancy	1.28	0.35	0.72
Life expectancy ↛ Tax revenue FD	2.94***	2.75	0.00
Tax revenue FD ↛ Life expectancy	4.82***	5.404	0.00
With infant deaths
infant deaths ↛ EPSI	0.308	−0.309	0.32
EPSI ↛ infant deaths	3.52***	3.56	0.00
Infant deaths ↛ Revenue FD	1.87	1.23	0.215
Revenue FD ↛ infant deaths	0.88	−0.15	0.87
Infant deaths ↛ Expenditure FD	2.09	1.33	0.18
Expenditure FD ↛ infant deaths	2.64**	2.01	0.04
Infant deaths ↛ Tax revenue FD	3.17***	3.07	0.0021
Tax revenue FD ↛ infant deaths	6.95***	8.42	0.00

Note: Significance level is denoted by *** for 1%, **for 5%, and *for 10%.

Source: Calculated by the authors.

Discussion

This study investigates the separate and combine impact of ER and FD in assessing health outcomes. In this study's initial regression analysis, the MMQREG model is considered. The statistical analysis is displayed in Tables 7 to 15. This study concludes based on the given results. (1) ER considerably affect health expenditures, increasing their impact in higher quantiles. ER increases life expectancy in higher quantiles for RD but decreases it for ED. Moreover, regulations reduce the number of infant deaths. High ER may increase health expenditures and improve access to health services, ultimately increasing life expectancy and reducing infant deaths. Such findings are consistent with the previous studies.^7,18,23 (2) FD has a considerable positive impact in lower quantiles. Decentralization reduces life expectancy at birth, but the influence lessens in higher quantiles. Moreover, expenditure decentralizing minimizes infant deaths. FD increases health expenditures and reduces infant deaths. However, FD has negative impact on life expectancy. This dual impact can be explained in the light of two generations of FD theories. The first generation of decentralization theories based on classical theory of FD claims that local governments provide public goods more efficiently, boosting public welfare.^15,61 These scholars felt that local governments understand local inhabitants’ preferences better than the central government, making them better positioned to provide public goods and services to their communities. The second generation of FD ideas, based on public choice theory, argue that the government does not maximize resident welfare^32,62 and may even decrease public welfare, such as life expectancy. (3) Decentralizing tax revenue and environmental controls minimizes median health expenditures while minor ED. Revenue and regulation reduce life expectancy. When expenditures and decentralization combine, life expectancy is higher at lower quantiles in upper quantiles. The two policies’ combined effect is positive (i.e. it reduces infant deaths). However, the impact is stronger in lower quantiles. In a nutshell, the effects of decentralizing revenue and spending are different, with expenditure decentralizing alone and in combination with ER showing the most promise for improving health outcomes. In this scenario, governments are motivated to enhance public services and environmental quality by gaining financial autonomy following “voting by foot.”³⁶

Conclusion

The study is designed to test the impact of ER and FD on health outcomes in BRICS economies. We have found both policy measures’ significant and positive impact on the health indicators. However, the combined effect of these measures is significantly negative.

In the last 20 years, several emerging countries have embraced ER and decentralization. Decentralization is a powerful way to improve efficiency and justice in providing public goods such as health care. Still, the impact of ER on health is limited, and the combined influence of both rising trends in developing countries is undetermined. This study examines the effects of ER and FD on health outcomes, health expenditures, life expectancy, and the number of infant deaths. This study examines the BRCS economies from 2000 to 2020 using MMQREG requirements.

Since BRICS economies comprise emerging economies, the panel contains variations in data. Regarding this, data normality is examined, showing that the majority of variables have non-normal distributions, which may encourage the development of an efficient estimator for non-normal data. The current study is able to use a second-generation panel unit root test, which offers mixed-order data integration, due to heterogeneous slope coefficients and cross-sectional dependency. The empirical estimations of the MMQREG show that ER positively affect health indicators. The study found that ER affect higher-quantile health outcomes. ER positively affect health costs, life expectancy, and infant mortality. In most circumstances, higher quantiles have more impact. Total revenue and ED positively improve health outcomes, while TRD negatively affects them. Our analysis found that decentralization and environmental laws had a negative influence. Moreover, the Dumitrescu-Hurlin panel causality tests indicate the bidirectional causality of TRD with health expenditures, life expectancy, and infant deaths, while the remaining variables have unidirectional causality.

Policy recommendations

Based on empirical findings, the study suggests the following policies for the decision-makers. (i) The policy makers need to implement ER in order to improve the health standards of the public at large. (ii) Powers should be delegated to lower levels and degree of FD should be increased to bring in positive health outcomes. The central government should increase local government autonomy on revenues and expenditures, so that local governments may design policies well suited to their local conditions and may improve health standards. (iii) Even though both measures (ER and FD) favor health indicators, their combination worsens health outcomes. Hence keeping in mind this finding, it is crucial to balance these two policy measures to maximize the effectiveness for better health standards.

The negative impact of regulations and FD seems to be the result of the race to the bottom hypothesis.^c^,11 In this regard, further research may focus on the specific limit and balance of ER and FD to improve health status.

Limitations of the study

The study suffered two important data limitations: (i) Due to the unavailability of FD data for India, the study was limited to four BRICS economies; (ii) Data on expenditures decentralization for China was unavailable. Hence, the estimations of expenditures decentralization do not include China. Therefore, further initiatives are needed to extend work by incorporating all BRICS economies on data availability and environmental policies.

Footnotes

Author contributions

Feng Wang: critical revision, incorporation of intellectual content. Seemab Gillani: literature search, data collection, data interpretation, drafting. Rabia Nazir: study design and concept, data analysis, data interpretation. Asif Razzaq: writing-review and editing.

Declaration of conflicting interests

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

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the General Project of National Philosophy and Social Sciences Foundation in 2022: Driving effects, potential impacts, and coordinating pathways of the forcing mechanism from China's “dual carbon” targets on its high-quality economic development (grant number: 22BJY127).

ORCID iDs

Feng Wang

Seemab Gillani

Notes

Feng Wang is Professor at School of Economics and Finance, Xi'an Jiaotong University. His research focuses on energy and environmental economics, low-carbon economics, and sustainable economic development. He has many SCI/SSCI publications in renowned journals.

Seemab Gillani is currently a PhD student at School of Economics and Finance, Xi'an Jiaotong University. Her research areas include health economics, environmental economics, and development economics.

Rabia Nazir is Assistant Professor at Department of Economics, The Islamia University of Bahawalpur. Her research focuses on energy & environmental economics, political economy, and sustainable economic development. She has many SCI/SSCI publications in renowned journals.

Asif Razzaq is a PhD in economics from Dalian University of Technology. Currently, he is working at CAREC Institute Urumqi, China and ILMA University, Karachi, Pakistan. His research area is environmental economics and sustainable economic development. He has many SCI/SSCI publications in renowned journals.

Appendix

Figure A.2.

Number of infant deaths in BRCS economies.

Figure A.1.

A.Life expectancy in BRCS economies.

References

Sarwar

, et al. Spatio-temporal evolution and driving effects of the ecological intensity of urban well-being in the Yangtze river Delta. Energy & Environment 2022; 33: 1181–1202.

Hung

. Effect of economic indicators, biomass energy on human development in China. Energy & Environment 2022; 33: 829–852.

Gatti

Di Paola

Monaco

, et al. The spatial association between environmental pollution and long-term cancer mortality in Italy. Sci Total Environ 2023; 855: 158439.

Hao

Gai

Yan

, et al. The spatial spillover effect and nonlinear relationship analysis between environmental decentralization, government corruption and air pollution: evidence from China. Sci Total Environ 2021; 763: 144183.

. Does pollution-intensive industrial agglomeration increase residents’ health expenditure? Sustainable Cities and Society 2020; 56: 102092.

Yang

Zhang

. Air pollution and healthcare expenditure: implication for the benefit of air pollution control in China. Environ Int 2018; 120: 443–455.

Yang

Chou

S-Y

. The impact of environmental regulation on fetal health: evidence from the shutdown of a coal-fired power plant located upwind of New Jersey. J Environ Econ Manage 2018; 90: 269–293.

Martinez-Vazquez

Lago-Penas

Sacchi

. The impact of fiscal decentralization: a survey. J Econ Surv 2017; 31: 1095–1129.

Alharthi

Hanif

. The role of energy types and environmental quality on human health in developing Asian countries. Energy & Environment 2021; 32: 1226–1242.

10.

Hao

Ren

. How do environmental regulation and environmental decentralization affect green total factor energy efficiency: evidence from China. Energy Economics 2020; 91: 104880.

11.

Zahra

Khan

Nouman

. Fiscal policy and environment: a long-run multivariate empirical analysis of ecological footprint in Pakistan. Environmental Science and Pollution Research 2022; 29: 2523–2538.

12.

Zhang

Uddin

, et al. Environmental regulation, foreign investment behavior, and carbon emissions for 30 provinces in China. J Cleaner Prod 2020; 248: 119208.

13.

Charles

. A pure theory of local expenditures. Journal of Political Economy 1956; 64: 416–424.

14.

Oates

. Fiscal Federalism. New York: Harcourt Brace Jovanovich, 1972.

15.

Hao

Chen

Y-F

Liao

, et al. China's fiscal decentralization and environmental quality: theory and an empirical study. Environ Dev Econ 2020; 25: 159–181.

16.

Oates

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

17.

Statistics

. Statistical review of world energy 2020. 2020. (69 ed.). London: Statistical Review of World Energy. Retrieved from https://www.bp.com/content/dam/bp/business-sites/en/global/corporate/pdfs/energy-economics/statistical-review/bp-stats-review-2020-full-report.pdf

18.

Q-T

Joshi

Stolper

. Can environmental policy reduce infant mortality? Evidence from the Ganga pollution cases. J Dev Econ 2018; 133: 306–325.

19.

Geng

Cui

Nazir

, et al.

How do CSR and perceived ethics enhance corporate reputation and product innovativeness?

Economic Research-Ekonomska Istraživanja 2022; 35: 5131–5149.

20.

Jinru

Changbiao

Ahmad

, et al. How do green financing and green logistics affect the circular economy in the pandemic situation: key mediating role of sustainable production. Economic Research-Ekonomska Istraživanja 2022; 35: 3836–3856.

21.

Wang

Laila

Nazir

, et al. Unleashing the influence of industrialization and trade openness on renewable energy intensity using path model analysis: a roadmap towards sustainable development. Renewable Energy 2023; 202: 280–288.

22.

Zhang

Liu

Wang

, et al. Influence of ecological innovation and green energy investment on unemployment in China: evidence from advanced quantile approach. Economic Research-Ekonomska Istraživanja 2022: 1–19.

23.

Cesur

Tekin

Ulker

. Can natural gas save lives? Evidence from the deployment of a fuel delivery system in a developing country. J Health Econ 2018; 59: 91–108.

24.

Tanaka

. Environmental regulations on air pollution in China and their impact on infant mortality. J Health Econ 2015; 42: 90–103.

25.

Greenstone

Hanna

. Environmental regulations, air and water pollution, and infant mortality in India. Am Econ Rev 2014; 104: 3038–3072.

26.

Tang

Jia

. Impact of environmental regulations on environmental quality and public health in China: empirical analysis with panel data approach. Sustainability 2020; 12: 23.

27.

Robalino

Picazo

Voetberg

. Does fiscal decentralization improve health outcomes? Evidence from a cross-country analysis. Available at SSRN 2001.

28.

Schwartz

Guilkey

Racelis

. Decentralization, allocative efficiency and health service outcomes in the Philippines. Carolina Population Center: University of North Carolina at Chapel Hill: MEASURE Evaluation, 2002.

29.

Cantarero

Pascual

. Analysing the impact of fiscal decentralization on health outcomes: empirical evidence from Spain. Appl Econ Lett 2008; 15: 109–111.

30.

Samadi

Keshtkaran

Kavosi

, et al. The effect of fiscal decentralization on under-five mortality in Iran: a panel data analysis. Int J Health Policy Manag 2013; 1: 301–306.

31.

Sanogo

. Does fiscal decentralization enhance citizens’ access to public services and reduce poverty? Evidence from Côte d’Ivoire municipalities in a conflict setting. World Dev 2019; 113: 204–221.

32.

Lin

. Fiscal decentralization, public health expenditure and public health–evidence from China. Front Public Health 2022; 10: 773728.

33.

Jin

Sun

. Does fiscal decentralization improve healthcare outcomes? Empirical evidence from China. Public Finance Manage 2011; 11: 243–261.

34.

Ferrario

Zanardi

. Fiscal decentralization in the Italian NHS: what happens to interregional redistribution? Health Policy 2011; 100: 71–80.

35.

Kuai

Yang

Tao

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

36.

Chen

Chang

C-P

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

37.

Qian

Roland

. Federalism and the soft budget constraint. Am Econ Rev 1998; 88: 1143–1162.

38.

Cheng

Zhu

. The spatial effect of fiscal decentralization on haze pollution in China. Environ Sci Pollut Res 2021; 28: 49774–49787.

39.

Von Braun

Grote

. Does decentralization serve the poor? In: Ahmad

Tanzi

(eds) Managing fiscal decentralization. London: Routledge, 2002, pp.84–112.

40.

Cheng

Awan

Ahmad

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

41.

Benali

. The dynamic links between natural disaster, health spending, and GDP growth: a case study for lower middle-income countries. J Knowl Econ 2022; 13: 1993–2006.

42.

Mahalik

T-H

H-C

, et al. How do sources of carbon dioxide emissions affect life expectancy? Insights from 68 developing and emerging economies. World Develop Sustain 2022; 1: 100003.

43.

Nakatani

Zhang

Garcia Valdes

. Fiscal Decentralization Improves Social Outcomes When Countries Have Good Governance. 2022.

44.

Bellinger

. Rural decentralization and infant mortality among select Indian states: a preliminary analysis. Contemp South Asia 2022; 30: 202–209.

45.

Yirong

. Does environmental policy stringency reduce CO2 emissions? Evidence from high-polluted economies. J Cleaner Prod 2022; 341: 130648.

46.

Tunio

Nabi

. Political decentralization, fiscal centralization, and its consequences in case of Pakistan. Cogent Soc Sci 2021; 7: 1924949.

47.

Canare

. Decentralization and development outcomes: what does the empirical literature really say? Hacienda Publica Espanola 2021; 237: 111–151.

48.

Feng

Zhao

. Revisiting energy efficiency and energy related CO2 emissions: evidence from RCEP economies. Econ Res-Ekonomska Istraživanja 2022; 35: 5858–5878.

49.

Jarque

Bera

. A test for normality of observations and regression residuals. Int Stat Rev/Revue Internationale de Statistique 1987; 55: 163–172.

50.

Bao

HHG

. Renewable and nonrenewable energy consumption, government expenditure, institution quality, financial development, trade openness, and sustainable development in Latin America and Caribbean emerging market and developing economies. Int J Energy Econ Policy 2020; 10: 242–248.

51.

Pesaran

Yamagata

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

52.

Pesaran

. General diagnostic tests for cross-sectional dependence in panels. Empir Econ 2021; 60: 13–50.

53.

Campello

Galvao

Juhl

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

54.

Pesaran

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

55.

Pesaran

. Estimation and inference in large heterogeneous panels with a multifactor error structure. Econometrica 2006; 74: 967–1012.

56.

Koenker

Bassett

. Regression quantiles. Econometrica: J Econom Soc 1978; 46: 33–50.

57.

Machado

Silva

. Quantiles via moments. J Econom 2019; 213: 145–173.

58.

Sarkodie

Strezov

. A review on environmental kuznets curve hypothesis using bibliometric and meta-analysis. Sci Total Environ 2019; 649: 128–145.

59.

Dumitrescu

E-I

Hurlin

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

60.

Banday

Aneja

. Renewable and non-renewable energy consumption, economic growth and carbon emission in BRICS: evidence from bootstrap panel causality. Int J Energy Sect Manage 2019; 14: 248–260.

61.

Oates

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

62.

Hao

Liu

Z-N

, et al. Impact of income inequality and fiscal decentralization on public health: evidence from China. Econ Model 2021; 94: 934–944.