Fiscal Policy, Growth, Financial Development and Renewable Energy in Romania: An Autoregressive Distributed Lag Model with Evidence for Growth Hypothesis

Abstract

This research aims to identify the influence of fiscal policy, financial development and economic growth on the increase of renewable consumption in Romania. To achieve our objective, we employ bivariate regressions through the Autoregressive Distributed Lag method, over the 2000–2020 period, to examine these influences. We find clear evidence that the variables observed (implicit tax rate on energy, external debt stocks, real GDP per capita, environmental tax revenues from energy taxes, and market capitalisation of listed domestic companies) have significant effects on the use of renewable energy. Four unidirectional causal relationships were identified in the long run: two from independent variables towards the dependent variable and two from the dependent variables towards two other independent variables. The importance of this study is that its results can contribute to the finding of the most suitable solutions to improve renewable energy consumption in Romania and mitigate the impact of climate change. Consequently, the results of this study reveal significant conclusions and policy recommendations for Romania moving towards sustainable and green economic growth, through a balanced set of policies and measures smartly applied, accompanied by a solid rate of absorption of green funds.

1. Introduction

Globally, the energy sector is considered of utmost importance because energy demand is growing, determined by population growth, economic development, lifestyle changes, and production growth. Most energy comes from fossil fuels: oil, gas and coal, representing 84.3% of the world’s primary energy mix [1]. If we look at the past, we notice that in each significant stage of economic development, there was a transition from one fuel source to another. Although fossil fuels are currently the primary source of energy in developing countries, we are on the verge of another transition: from fossil fuels to renewable energy sources, driven by both environmental concerns and limited fossil fuels, and changes in technologies. To reduce the harmful effects produced by human activities on the environment, sustained political efforts are needed, including the implementation of fiscal measures. Thus, many of the climate change initiatives at the European level can be directly linked to fiscal policy, mainly through public spending or taxation, and they can also have an indirect influence on macroeconomic and fiscal outcomes. Actions taken to overcome these climate change challenges have generated discussions about the introduction of environmental taxes and environmental technologies and their subsequent impact on the demand for renewable energy.

If we focus on Romania, we notice that this country has a share of renewable energy above the EU average [2] but is dependent on coal for electricity generation, which leads to an energy price of over 50% higher than the market energy. At the same time, according to an analysis of greenhouse gas emissions in Romania [3], the emissions related to the energy sector category represent approximately 70% of total national greenhouse gas emissions.

Although in 2020 Romania achieved a significant percentage of total energy consumption from renewable sources (24%), more recently, due to a lack of regulations and appropriate governmental support, there has been a decrease in the attractiveness of investments in renewable energy [4]. Renewable energy production requires significant investments that are depreciated over a reasonably long period [5], which makes the interest in such investments not very high. A key role in this could be played by the financial sector, together with the government sector, to stimulate the financing of projects leading to the creation of renewable energy. Thus, Wurgler [6] stated that countries with well-developed financial sectors support investments in growing industries and reduce those in declining industries to implicitly support investments in renewable energy creation.

Therefore, we consider that government should be aware of the need to update policies and legislation to stimulate public and private investment to promote the development of renewable energy production.

Considering the above, the purpose of this research was to highlight how fiscal policy and financial development influence renewable energy consumption, offering a series of policy proposals determining the sustainable use of renewable energy. In the process of achieving the purpose of this study, a series of results were obtained, which contain elements of scientific novelty. First, the methodology employed is innovative for studying the relationships observed in the case of Romania. Second, limited research has explored the nexus between external debt stocks and renewable energy consumption. Third, this research is one of the few to test the nexus between the selected variables through the Autoregressive Distributed Lag (ARDL) model.

Further material is divided into several parts. In the next section, we present a brief literature overview on the topic, Section 3 describes the applied materials and methods, Section 4 presents the results of our analysis, and Section 5 presents conclusions, practical implications and study limitations.

2. Literature Review

In recent years, special attention has been brought to renewable energy in the scientific literature because it is seen as a way to ensure energy use without generating adverse effects on the environment. We can thus classify research according to the variables employed to analyse the relationship with renewable energy.

2.1. The Nexus between Fiscal Policy and the Use of Renewable Energy

The study by Chien et al. [7] emphasised that green fiscal policies related to public support and tax cuts significantly reduce the energy poverty of various international countries by promoting energy efficiency. Furthermore, Ike et al. [8] identified a one-way relationship from fiscal policy towards energy consumption. Their study was conducted on data from Thailand and confirmed the importance of fiscal policy on energy consumption and environmental quality.

Moreover, the relationship between renewable and non-renewable energy consumption and the budget deficit was also studied, observing more than 30 countries, based on their net energy-import [9]. A two-way relationship between non-renewable energy consumption and the budget deficit was observed, as well as a one-way relationship from the budget deficit to renewable energy consumption.

Amoah et al. [10] analysed the role of economic well-being and economic freedom as factors of renewable energy in a study of 32 African countries. They found that reducing the tax burden will lead to an increase in the share of renewable energy consumption.

A study building an equilibrium model of the energy market under the influence of environmental taxes analysed whether they can help motivate renewable energy producers. They concluded that environmental taxes regulation could increase the share of renewable energy production in China’s energy market [11]. Similar research on the influence of regulations concurrently with environmental technologies has shown that they have different effects on the environment depending on the characteristics of each country [12,13,14]. There are also published works that have highlighted that imperfect regulation of the environment delays the implementation of policies and restrictive taxation on carbon emissions, which leads to a paradox of the increase of greenhouse gas emissions [15,16].

Therefore the following hypothesis is postulated:

Hypothesis 1:

There is a nexus between fiscal policy and the use of renewable energy in Romania.

2.2. The Nexus between Financial Development and the Use of Renewable Energy

Existing strands in the literature focusing on the effects of financial development on renewable energy use identified the relationship between financial development and renewable energy consumption, economic growth and energy prices. Studies focused on the neighbouring countries Azerbaijan (over the period of 1993–2015) and Turkey (data from 1980 to 2019), emphasised through empirical results a significant and direct influence on renewable energy consumption from both financial development and economic growth [17,18]. Furthermore, other authors [19] brought to attention the way financial development influences renewable energy consumption at the US level between 1975 and 2019, and they observed a long-term asymmetric effect of financial development on renewable energy consumption, while in the short term renewable energy consumption does not react to changes in financial development.

In a study conducted in China, Lei et al. [20] observed that a positive or negative shock of bank deposits and of currency generates a significantly increasing effect on long-term renewable energy consumption. Therefore, the empirical results prove a statistically significant relationship between financial development, measured through banking variables and renewable energy consumption. Wang et al. [21] also proved that in China the development of the renewable energy sector depends on economic growth and financial aspects. Although the influence is positive over the short term, over the long run financial development negatively influences renewable energy consumption.

Another strand in the literature considers that financial development generates an increase in the demand for renewable energy consumption [22]. Accordingly, governments should implement fiscal policies that direct businesses to renewable energy resources. Furthermore, Amin et al. [23] identified a one-way relationship from financial development to renewable energy consumption, and the influence exerted is negative, so increasing financial development generates a reduction in renewable energy consumption.

However, other research papers point out the importance of other variables influencing the use of renewable energy. For example, the relationship between financial development, renewable energy, trade, health and tourism was discussed by Ali et al. [24], and the results revealed a two-way causal relationship between financial development and renewable energy variables for low-, middle- and high-income countries in Asia. Furthermore, Eren et al. [25] discussed the sustainability of renewable energy consumption, financial sustainability and economic growth in India for the period of 1971–2015. They observed a one-way relationship between financial development, renewable energy consumption and gross domestic product (GDP). Moreover, Yazdi and Shakouri [26] explored the Iranian case over the period of 1992–2014, evidencing causal links between renewable energy consumption, economic development, financial progress, and globalisation. They employed Granger causality tests, and results revealed a bidirectional causality between renewable energy use, economic growth and financial progress.

Therefore, the following hypothesis is postulated:

Hypothesis 2:

There is a nexus between financial development and the use of renewable energy in Romania.

2.3. The Nexus between Economic Growth and the Use of Renewable Energy

Among researchers [27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56], another assumption is related to the relationship between the consumption of renewable energy and economic growth. In this view, four hypotheses were identified [27]. The first one, the hypothesis of growth, considers energy as a major source of growth in terms of one-way causality from energy consumption to economic growth. The second, the hypothesis of conservation, considers economic growth as a determinant of energy consumption. The third hypothesis—that of feedback—implies a two-way relationship between energy consumption and economic growth. The last, the hypothesis of neutrality, considers that energy consumption and economic growth are independent of each other. Therefore, in a paper performing a systematic analysis of the literature, Kassim and Isik [27] evidenced that 54% of the studies gathered confirm the feedback hypothesis, 23% of studies confirm the conservation hypothesis, 14% of studies the growth hypothesis, and 9% of studies confirm the neutrality hypothesis.

From the multitude of empirical studies analysing the causal relationship between renewable energy consumption and economic growth, we observed results supporting a bidirectional causality or the feedback hypothesis, unidirectional growth or the conservation hypothesis, or even no causality (neutrality hypothesis).

Pirlogea and Cicea [28], Salim et al. [29], Amri [30], Matei [31], and Badircea et al. [32] highlighted the one-way link from economic growth and renewable energy consumption. In addition, Liu [33] revealed the one-way relationship from renewable energy consumption to economic growth, using the Gray–Markov model for 34 countries over a period of 21 years. Numerous authors [34,35,36,37,38,39,40,41] reached similar conclusions, emphasising a one-way causality from the consumption of renewable sources to economic growth.

However, Saad and Taleb [42] identified a one-way relationship from economic growth to renewable energy consumption in the analysis conducted in the case of 12 EU countries in the short run. Still, in the long run they identified a two-way relationship. Also, Sadorsky [43] analysed the two-way link between economic growth and renewable energy consumption for 18 emerging economies, noting that a 1% increase in real per capita income leads to a 3.5% increase in renewable energy consumption. Nevertheless, Shahbaz et al. [44] examined the link between Pakistan’s renewable energy consumption, GDP, capital and labour, and validated the feedback hypothesis for the period of 1972–2011.

In the same view, Marinaș et al. [45] analysed the causal relationship between renewable energy consumption and economic growth in Central and Eastern European countries. They concluded that all analysed countries indicate a two-way relationship in the long run. Similarly, Al-Mulalli et al. [46] found that in 79% of the Latin American countries the long-term two-way relationship between renewable energy consumption and economic growth is validated, referring to the feedback hypothesis.

Also, in other papers, Apergis and Payne [47,48,49] analysed the causal relationship in OECD countries, Asian countries and six Central American countries, respectively, and reached the same conclusions on the bidirectional causal relationship between renewable energy consumption and sustainable and economic growth. Other studies, such as Paramati et al. [50], Neitzel [51] and Ohler and Fetters [52], also highlighted this two-way relationship.

However, other research papers could not evidence a relationship between economic growth and renewable energy consumption. This was the case for Payne [53] who analysed this relationship in the case of the United States for a period of 58 years. Similarly, studies conducted in European countries by Menegaki [54] and Smiech and Papiez [55] found evidence for the neutrality hypothesis, validating and explaining this situation by the unequal and insufficient exploitation of renewable energy sources in Europe. Ozcan and Ozturk [56] analysed the causal relationship, supporting the viability of the neutrality hypothesis for 16 of the 17 emerging countries observed, the exception being Poland. Also, Marinaș et al. [45] found that, over the short term, economic growth and renewable energy consumption are independent in Romania and Bulgaria.

Considering the conclusions of previous research, the following hypothesis is postulated:

Hypothesis 3:

There is a nexus between economic growth and the use of renewable energy in Romania.

Therefore, we can conclude that the results for the causal relationship between economic growth and renewable energy consumption may vary significantly from one country to another, especially due to the use of different ways of measuring energy consumption and production, different econometric methods, but also depending on the period of time overviewed.

From the three hypotheses postulated for this research, it is reasonable to investigate the factors influencing the use of renewable energy and to evaluate their long-term impact on consumption. Therefore, we raise the following research question: to what extent is the use of renewable energy influenced by different factors, and which are the most significant ones across Romania?

To achieve our objective, we employ bivariate regressions through the ARDL method for Romania, over the 2000–2020 period, examining the influence of fiscal policy, financial development and economic growth on renewable energy consumption. The following sections will bring to attention the clear evidence that these variables have significant effects on the use of renewable energy.

3. Materials and Methods

The main objective of this research is to identify the influence of fiscal policy, financial development and economic growth on the increase of renewable energy consumption in Romania. This goal can be achieved by placing the long-term causal relationships through the ARDL method. The importance of this study lies in finding the most suitable solutions to improve renewable energy consumption in Romania and mitigate the impact of climate change.

The variables considered representative for the purpose of this study are presented in

Table 1. Selected variables for the study.

Indicator	Acronym	Variable Type	Unit of Measure	Data Source
Implicit tax rate on energy	ITR	Independent	Euro per ton of oil equivalent (TOE)	Eurostat [57]
Environmental tax revenues from energy taxes	RET	Independent	Million euro	Eurostat [57]
Market capitalization of listed domestic companies	MC	Independent	% of GDP	World Development Indicators [58]
External debt stocks	EDS	Independent	% of GNI	World Development Indicators [58]
Real GDP per capita	GDP	Independent	Chain linked volumes (2010), euro per capita	World Development Indicators [58]
Renewable energy consumption	RENC	Dependent	Share of renewable energy in gross final energy consumption—Percentage	Eurostat [57]

.

Two variables were chosen to express the financial development: market capitalisation of listed domestic companies and external debt stocks. Energy taxation is represented by the following variables: implicit tax rate on energy and environmental tax revenues from energy taxes. Also, the economic growth proxy, namely the real gross domestic product per capita, was selected as the most representative indicator. The analysis was performed on an annual basis, for data series from 2000–2020, selected from the official databases of the European Commission (Eurostat) and World Bank.

Starting from the equation proposed by Jabari et al. [59], according to the purpose of this study, we developed the following equation:

lnRENC_t = β₁ + β₂lnGDP_t + β₃lnMC_t + β₄lnEDS_t + β₅lnITR_t + β₆lnRET_t + ε_t

(1)

where lnRENC is the natural logarithm of the renewable energy consumption, lnGDP is the natural logarithm of the real GDP per capita, lnMC is the natural logarithm of the market capitalisation of listed domestic companies, lnEDS is the natural logarithm of external debt stocks, lnITR is the natural logarithm of the implicit tax rate on energy, lnRET is the natural logarithm of the environmental tax revenues from energy taxes, ε is the residual term, and t is the period of time.

There are some requirements before applying the ARDL model: absence of autocorrelation between the variables, heteroscedasticity must not occur in the data, data should follow a normal distribution and it must be stationary, either on level or first difference or on both.

To obtain correct and rigorous results, we will go through the following stages in the econometric analysis: testing the stationarity of the data series (Phillips–Perron test), eliminating autocorrelation problems between variables through the Pearson correlation matrix, identifying cointegration relationships between variables through ARDL bound test, estimating the influence of explanatory variables on the dependent variable using the ARDL method, testing the stability of the proposed model by applying CUSUM and CUSUMsq and identifying causality directions by means of the Granger pairwise causality test (Figure 1).

To check the stationarity of the data, unit root tests were applied for every indicator employed in the analysis. According to Arltová and Fedorová [60] the appropriate unit root test for a short data series is the test proposed by Phillips and Perron (PP), whose equation is presented below [61]. In the unit root testing of time series generated by the process with the auto-correlated and heteroscedastic non-systematic component, there is often a problem of selection of lag in the regression model. Phillips and Perron (1988) dealt with this problem by employing the standard Dickey–Fuller test with non-parametrically modified test statistics.

$$Z\left(\tau \right)=\tau \left(\frac{{\widehat{\sigma }}^2}{{\widehat{\sigma }}_{sl}^2}\right)-\left(\frac{1}{2}\right)\left({\widehat{\sigma }}_{sl}^2-{\widehat{\sigma }}^2\right)T\sqrt{{\widehat{\sigma }}_{sl}^2{\displaystyle \sum }_{\mathrm{t}=2}^{\mathrm{T}}{\left(x_{t-1}-{\overline{x}}_{T-1}\right)}^2}$$

(2)

The ARDL presumes standard least squares regressions that include the lags of dependent and explanatory variables. Although the ARDL method has been performed for decades, it has recently gained more popularity as a method of examining cointegrating relationships between variables. This method was based on the research of Pesaran and Shin (1998)—PS (1998) and Pesaran, Shin and Smith (2001)—PSS (2001) [62], being superior to the traditional techniques because it does not consider the degree of integration of the model’s variables. Therefore, it is suitable for small samples, such as the one considered in our study, overviewing only one country.

As this study employs six variables and the ARDL method assumes a regression equation for each variable, it is evident that we will have a system consisting of six equations to test the cointegration between the variables and to express the long-run relationship between renewable energy consumption, energy taxation, and financial development. The system includes the following six Equations (3)–(8):

$$\begin{array}{l}\Delta \ln {\mathit{RENC}}_t={\beta }_1+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_1\Delta \ln REN{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_2\Delta \ln GD{P}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_3\Delta \ln M{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_4\Delta \ln ED{S}_{t-i}\hfill \\ +{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_5\Delta \ln IT{R}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_6\Delta \ln RE{T}_{t-i}+{\varnothing }_1\ln REN{C}_{t-i}+{\varnothing }_2\ln GD{P}_{t-i}+{\varnothing }_3\ln M{C}_{t-i}\hfill \\ +{\varnothing }_4\ln ED{S}_{t-i}+{\varnothing }_5\ln IT{R}_{t-i}+{\varnothing }_6\ln RE{T}_{t-i}+{\epsilon }_{1t}\hfill \end{array}$$

(3)

$$\begin{array}{l}\Delta \ln {\mathit{GDP}}_t={\beta }_2+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_1\Delta \ln REN{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_2\Delta \ln GD{P}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_3\Delta \ln M{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_4\Delta \ln ED{S}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_5\Delta \ln IT{R}_{t-i}\hfill \\ +{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_6\Delta \ln RE{T}_{t-i}+{\varnothing }_1\ln REN{C}_{t-i}+{\varnothing }_2\ln GD{P}_{t-i}+{\varnothing }_3\ln M{C}_{t-i}+{\varnothing }_4\ln ED{S}_{t-i}+{\varnothing }_5\ln IT{R}_{t-i}\hfill \\ +{\varnothing }_6\ln RE{T}_{t-i}+{\epsilon }_{2t}\hfill \end{array}$$

(4)

$$\begin{array}{l}\Delta \ln {\mathit{MC}}_t={\beta }_3+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_1\Delta \ln REN{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_2\Delta \ln GD{P}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_3\Delta \ln M{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_4\Delta \ln ED{S}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_5\Delta \ln IT{R}_{t-i}\hfill \\ +{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_6\Delta \ln RE{T}_{t-i}+{\varnothing }_1\ln REN{C}_{t-i}+{\varnothing }_2\ln GD{P}_{t-i}+{\varnothing }_3\ln M{C}_{t-i}+{\varnothing }_4\ln ED{S}_{t-i}+{\varnothing }_5\ln IT{R}_{t-i}\hfill \\ +{\varnothing }_6\ln RE{T}_{t-i}+{\epsilon }_{3t}\hfill \end{array}$$

(5)

$$\begin{array}{l}\Delta \ln {\mathit{EDS}}_t={\beta }_4+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_1\Delta \ln REN{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_2\Delta \ln GD{P}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_3\Delta \ln M{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_4\Delta \ln ED{S}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_5\Delta \ln IT{R}_{t-i}\hfill \\ +{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_6\Delta \ln RE{T}_{t-i}+{\varnothing }_1\ln REN{C}_{t-i}+{\varnothing }_2\ln GD{P}_{t-i}+{\varnothing }_3\ln M{C}_{t-i}+{\varnothing }_4\ln ED{S}_{t-i}+{\varnothing }_5\ln IT{R}_{t-i}\hfill \\ +{\varnothing }_6\ln RE{T}_{t-i}+{\epsilon }_{4t}\hfill \end{array}$$

(6)

$$\begin{array}{l}\Delta \ln {\mathit{ITR}}_t={\beta }_5+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_1\Delta \ln REN{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_2\Delta \ln GD{P}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_3\Delta \ln M{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_4\Delta \ln ED{S}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_5\Delta \ln IT{R}_{t-i}\hfill \\ +{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_6\Delta \ln RE{T}_{t-i}+{\varnothing }_1\ln REN{C}_{t-i}+{\varnothing }_2\ln GD{P}_{t-i}+{\varnothing }_3\ln M{C}_{t-i}+{\varnothing }_4\ln ED{S}_{t-i}+{\varnothing }_5\ln IT{R}_{t-i}\hfill \\ +{\varnothing }_6\ln RE{T}_{t-i}+{\epsilon }_{5t}\hfill \end{array}$$

(7)

$$\begin{array}{l}\Delta \ln {\mathit{RET}}_t={\beta }_6+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_1\Delta \ln REN{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_2\Delta \ln GD{P}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_3\Delta \ln M{C}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_4\Delta \ln ED{S}_{t-i}+{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_5\Delta \ln IT{R}_{t-i}\hfill \\ +{\displaystyle {\displaystyle \sum }_{i=1}^n}{\alpha }_6\Delta \ln RE{T}_{t-i}+{\varnothing }_1\ln REN{C}_{t-i}+{\varnothing }_2\ln GD{P}_{t-i}+{\varnothing }_3\ln M{C}_{t-i}+{\varnothing }_4\ln ED{S}_{t-i}+{\varnothing }_5\ln IT{R}_{t-i}\hfill \\ +{\varnothing }_6\ln RE{T}_{t-i}+{\epsilon }_{6t}\hfill \end{array}$$

(8)

The null hypothesis of each equation is given by ${\varnothing }_1={\varnothing }_2={\varnothing }_3={\varnothing }_4={\varnothing }_5={\varnothing }_6=0$, implying the lack of co-integration. The alternative hypothesis assumes the existence of at least one co-integration relationship between variables and is denoted by ${\varnothing }_1\ne {\varnothing }_2\ne {\varnothing }_3\ne {\varnothing }_4\ne {\varnothing }_5\ne {\varnothing }_6\ne 0$. If the F-statistic value is higher than the upper critical bound values, there is a co-integration between the model variables. If the F-statistic falls between the values of the bounds test, then we cannot determine the existence or non-existence of a co-integration relationship between variables. Ultimately, if the F-statistic value is lower than the lower critical bound, there is no co-integration relationship between the variable in the model [63].

Further, we applied the tests based on recursive residuals to observe the structural stability of the model. From these, the most relevant are the cumulative sum of recursive residuals (CUSUM) and the cumulative sum of squares of recursive residuals (CUSUMsq).

Correlation does not necessarily imply causality in the meaningful sense of the concept and might be spurious. Therefore, we employ the Granger causality method to explore the causal interaction among the variables investigated. The bivariate regressions take the following forms:

$$y_t={\alpha }_0+{\alpha }_1{y}_{t-1}+\cdots +{\alpha }_l{y}_{t-l}+{\beta }_1{x}_{t-1}+\cdots +{\beta }_l{x}_{t-l}+{\epsilon }_t$$

(9)

$$x_t={\alpha }_0+{\alpha }_1{x}_{t-1}+\cdots +{\alpha }_l{x}_{t-l}+{\beta }_1{y}_{t-1}+\cdots +{\beta }_l{y}_{t-l}+{\upsilon }_t$$

(10)

for all possible pairs of (x, y) series in the group. The null hypothesis in the first regression is that x does not Granger-cause y and in the second equation that y does not Granger-cause x.

Figure 1 presented above summarises the methodology structure for this research, listing each stage undertaken. They are similar to the stages undertaken by Badircea et al. [64], who described a comparison between Romania and Sweden in terms of the relationship between environmental performance, green taxation and economic growth.

4. Results and Discussions

Before undertaking the econometric analysis, a brief presentation of the evolution of renewable energy consumption in Romania may be observed in Figure 2, followed by the descriptive statistics of the variables included in the analysis (

Table 2. Descriptive statistics of the variables.

	lnRENC	lnGDP	lnMC	lnEDS	lnITR
Mean	3.138061	9.212004	2.295180	4.029817	4.751205
Median	3.173293	9.164149	2.328055	4.004501	4.842769
Maximum	3.220155	9.466164	2.865130	4.251577	5.003476
Minimum	2.900291	9.013605	1.955693	3.859577	4.350149
Std. Dev.	0.094210	0.159357	0.246839	0.137981	0.220113
Skewness	−1.409424	0.485746	0.702185	0.262965	−0.385972
Kurtosis	4.101024	1.941300	3.292244	1.703462	1.799881
Jarque-Bera	4.960667	1.118347	1.114567	1.060374	1.102933
Probability	0.083715	0.571681	0.572763	0.588495	0.576104
Sum	40.79479	119.7561	29.83734	52.38761	61.76566
Sum Sq. Dev.	0.106506	0.304735	0.731157	0.228466	0.581397

Note: lnRENC is the natural logarithm of renewable energy consumption, lnGDP is the natural logarithm of real gross domestic product per capita, lnMC is the natural logarithm of market capitalization of listed domestic companies, lnEDS is the natural logarithm of external debt stocks, lnITR is the natural logarithm of implicit tax rate on energy, and lnRET is the natural logarithm of environmental tax revenues from energy taxes.

).

In Figure 2 it can be observed that in Romania the consumption of renewable energy has had an increasing trend since the beginning of the century. This was mainly due to rising prices for energy produced from traditional sources and fiscal policy measures adopted in this regard. The implicit tax rate on energy and the revenues from energy taxes followed the same upward trend as renewable energy consumption over the time interval analysed (Figure 3).

The descriptive statistics included in Table 2 reflect the main characteristics of the variables included in the econometric model. Skewness, expressing the degree of asymmetry of the series distribution around its mean, indicates values between −1.5 and 1, underlining an approximate bell-shaped curve. Kurtosis, measuring the flatness of the series distribution, indicates values between +1 and +4, underlining a frequency distribution with a peaking tendency.

The Jarque–Bera [65] test statistic is used to check if the series has a normal distribution, comparing the skewness and kurtosis of the series with those appropriate for a normal distribution. The null hypothesis of the test supposes that there is a normal distribution. If the probability value is lower than 0.05, the null hypothesis is rejected. From the results obtained we conclude that our variables have a normal distribution, as the probability for every test indicates that the null hypothesis cannot be rejected.

To test the cointegration relationship between variables, all variables must be stationary in level or at the first difference order. To confirm it, the results of the PP unit root tests applied to the selected variables are presented in

Table 3. Unit root test results for the variables.

Variables	Intercept		Trend and Intercept		None
	t-Stat	p	t-Stat	p	t-Stat	p
lnRENC	−0.8035	0.7963	−2.1427	0.4930	1.4195	0.9557
dlnRENC	−5.8096	0.0002	−20.2482	0.0001	−4.4736	0.0001
lnGDP	−2.6093	0.1076	−1.2634	0.8675	2.2493	0.9915
dlnGDP	−3.4976	0.0315	−3.9874	0.0306	−1.9324	0.0529
lnMC	−4.6613	0.0020	−3.8647	0.0367	−0.2043	0.5986
dlnMC	−3.4542	0.0243	−3.4048	0.0860	−3.6160	0.0013
lnEDS	−2.0989	0.2469	−1.5515	0.7756	1.1004	0.9233
dlnEDS	−3.9114	0.0085	−4.1435	0.0210	−3.7373	0.0008
lnITR	−1.1255	0.6711	−1.8501	0.6223	1.1024	0.9195
dlnITR	−3.5082	0.0275	−6.4577	0.0014	−3.2071	0.0041
lnRET	−3.6676	0.0140	−0.8748	0.9383	3.5439	0.9995
dlnRET	−3.8714	0.0097	−5.7652	0.0011	−2.6383	0.0115

Note: lnRENC is the natural logarithm of renewable energy consumption, lnGDP is the natural logarithm of real gross domestic product per capita, lnMC is the natural logarithm of market capitalization of listed domestic companies, lnEDS is the natural logarithm of external debt stocks, lnITR is the natural logarithm of implicit tax rate on energy, lnRET is the natural logarithm of environmental tax revenues from energy taxes, t-stat is t-statistics, and p is p-value.

. It can be observed that most variables are not stationary at level. This is the case for models where exogenous regressors are included (intercept or trend and intercept), as well as in cases where they are not specified (none). The stationarity of the data is confirmed at the first difference for all the variables due to a p value lower than 0.5, which rejects the null hypothesis of a unit root.

After examining the stationarity properties of the data, the study undertakes the correlation analysis to investigate the multicollinearity between the variables employed.

Table 4. Correlation matrix.

	lnRENC	lnGDP	lnMC	lnEDS	lnITR	lnRET
lnRENC	1.0000
	-----
lnGDP	0.4367	1.0000
	(0.1356)	-----
lnMC	−0.1437	−0.0705	1.0000
	(0.6395)	(0.8187)	-----
lnEDS	0.0623	−0.5480	−0.1561	1.0000
	(0.8397)	(0.0525)	(0.6105)	-----
lnITR	0.8398	0.4659	0.2292	−0.2672	1.0000
	(0.0003)	(0.1085)	(0.4512)	(0.3773)	-----
lnRET	0.7575	0.7203	0.1913	−0.4501	0.9370	1.0000
	(0.0027)	(0.0055)	(0.5311)	(0.1227)	(0.0000)	-----

Note: lnRENC is the natural logarithm of renewable energy consumption, lnGDP is the natural logarithm of real gross domestic product per capita, lnMC is the natural logarithm of market capitalization of listed domestic companies, lnEDS is the natural logarithm of external debt stocks, lnITR is the natural logarithm of implicit tax rate on energy, lnRET is the natural logarithm of environmental tax revenues from energy taxes, and p-values are in parenthesis.

includes the correlation coefficients.

Although the ARDL test does not involve prior testing of the variables, we applied the unit-root tests for better results. In addition, our variables considered the first-order difference the most. Thus, based on the results obtained after applying unit root tests, ARDL models were established to test the causality relationship between fiscal policy, financial development and renewable energy consumption in Romania. Six ARDL models were formulated, one for every variable becoming the dependent one. It is extremely important to establish the optimal lag for each variable, and then to apply a bounds test to establish if there is a co-integration relationship between variables.

The results from testing lags and the bound tests are presented in

Table 5. ARDL Long Run Form and Bounds Test.

Estimated Models		LAG	F-Statistics
F(lnRENC/ lnGDP, lnMC, lnEDS, lnITR, lnRET)		2,0,0,0,0,0	7763.267
F(lnGDP/ lnMC, lnEDS, lnITR, lnRET, lnRENC)		1,0,0,0,0,0	23.488
F(lnMC/ lnEDS, lnITR, lnRET, lnRENC, lnGDP)		1,0,0,0,0,0	2.179
F(lnEDS/ lnITR, lnRET, lnRENC, lnGDP, lnMC)		1,0,0,0,0,0	540.126
F(lnITR/ lnRET, lnRENC, lnGDP, lnMC, lnEDS)		2,0,0,0,0,0	19286.56
F(lnRET/ lnRENC, lnGDP, lnMC, lnEDS, lnITR)		1,0,0,0,0,0	145.591
Critical Value Bounds
Significance (%)	I0 Bound	I1 Bound
10	2.08	3.00
5	2.39	3.38
2.5	2.7	3.73
1	3.06	4.15

Note: lnRENC is the natural logarithm of renewable energy consumption, lnGDP is the natural logarithm of real gross domestic product per capita, lnMC is the natural logarithm of market capitalization of listed domestic companies, lnEDS is the natural logarithm of external debt stocks, lnITR is the natural logarithm of implicit tax rate on energy, lnRET is the natural logarithm of environmental tax revenues from energy taxes, ARDL is Autoregressive Distributed Lag, I0 bound is lower critical value, and I1 bound is upper critical value.

. If the F-statistic associated to the bounds test is higher than the critical values at all levels of significance, then there is a co-integration relationship among the variables. In our research, all F-statistics values are above the critical values, except for the model with lnMC as the dependent variable. Therefore, we conclude that there is a long-run relationship between the variables and five co-integrating vectors in Romania.

After checking the existence of a co-integration relationship between renewable energy consumption, real GDP per capita, market capitalisation of listed domestic companies, external debt stocks, implicit tax rate on energy and environmental tax revenues from energy taxes, the ARDL estimation helps us evidence to what extent the fiscal policy, financial development and real GDP per capita explain the renewable energy consumption in the long run. Regarding the connection between variables, several aspects can be formulated after observing the results reported in

Table 6. ARDL estimation.

Dependent Variable: lnRENC
Model Selection Method: Akaike Info Criterion (AIC)
Selected Model: ARDL(2, 0, 0, 0, 0, 0)
Variable	Coefficient	Std. Error	t-Statistic	p-Value
lnGDP	1.861147	0.026800	69.44662	0.0092
lnMC	−0.198535	0.004616	−43.01251	0.0148
lnEDS	0.569229	0.008196	69.45454	0.0092
lnITR	1.870817	0.024101	77.62430	0.0082
lnRET	−1.549605	0.022401	−69.17689	0.0092
C	−9.661687	0.148523	−65.05192	0.0098

.

First, all variables exert a statistically significant influence on the dependent variable based on a p-value associated with each variable that is lower than 0.05. From the five independent variables, three have a positive influence on renewable energy consumption (lnGDP, lnEDS and lnITR), and two have a negative impact on the dependent variable (lnMC and lnRET). The positive influence of the external debt stock can be explained by the fact that part of the external loans is oriented towards investments in renewable energy sources. Also, the fiscal policy stimulates renewable energy consumption by increasing taxes on energy from traditional sources.

The CUSUM and CUSUMsq tests overview the stability of the model, based on the cumulative sum of the recursive residuals [56]. The CUSUM test plots the cumulative sum, identifying the parameter instability if the cumulative sum goes outside the area between the two critical lines. The graphical results of both tests are illustrated in Figure 4, indicating the parameters’ stability over the long- and short-run at a level of significance of 5%.

The existence of a co-integration relationship between the model’s variables implies at least a unidirectional causality relationship. To determine the causality relationship between the variables employed, we further apply the Granger test for every pair of variables. In Figure 5 we illustrate the direction of causality between variables in the long-run, after applying the pairwise Granger causality test.

In order to delimit the causal relationships between the variables, two types of differently coloured arrows were used. The directions of causality between the dependent variable and the explanatory variables were represented with green arrows and the directions of causality among the explanatory variables were represented with orange arrows.

Four unidirectional causal relationships were identified in the long run: two from independent variables towards the dependent variable (RENC) and two from the dependent variable towards another couple of independent variables. In the first situation, one can observe the causal direction from energy taxation to renewable energy consumption, but also the causality determined by the market capitalisation on the change of renewable energy consumption. The second situation captures the causal direction of renewable energy consumption towards the accumulation of income from energy taxation, but also towards the real GDP per capita.

Environmental taxes are based on the “polluter pays” principle, representing an important factor in the development and use of sustainable energy. These taxes were introduced to generate changes in the behaviour of consumers and producers towards sustainable consumption and production. Therefore, an increase in taxes on energy from traditional sources, with a negative impact on the environment, exerts a double influence: on the one hand, it influences the behaviour of consumers by orienting them towards the consumption of energy from renewable resources, and on the other hand, it increases the income to the state budget, confirming the double dividend theory [63].

This study confirms the hypotheses of this research. First, we find that all variables exert a statistically significant influence on the dependent variable (renewable energy consumption). Of the five independent variables, three have a positive influence on renewable energy consumption (lnGDP, lnEDS and lnITR), and two have a negative impact on the dependent variable (lnMC and lnRET). The results of the studies conducted by Chien et al. [7] and Amoah et al. [10] emphasised the important role of green fiscal policies in reducing the energy poverty of various countries by promoting energy efficiency and that reducing the tax burden will lead to an increase in the share of renewable energy consumption. Similarly, our study reveals the negative impact of environmental tax revenues from energy taxes on the renewable energy consumption. Accordingly, the higher the energy taxation from traditional sources, the more the consumers will be oriented towards renewable energy consumption. Furthermore, this situation will be followed by a reduction in tax revenues from energy taxes, thus stimulating the use of renewable energy.

A study that used a correlation analysis between the tax rate on energy and the renewable energy consumption from different renewable resources carried out in the Baltic region [66] over the period of 2005–2015 also identified significant positive impact of energy taxation on sustainable energy development. This confirms the fact that the harmonization of energy and environmental policies can provide win-win solutions for increasing renewable energy sources, energy efficiency and energy independence.

Similar to Ike et al. [8], who identified a one-way relationship from fiscal policy to energy consumption, confirming the importance of fiscal policy on energy consumption and environmental quality, our research supports the growth hypothesis which implies that energy consumption drives economic growth. Our results are consistent with several research papers [33,34,35,36,37,38,39,40,41] emphasising a one-way causality from the consumption of renewable sources to economic growth.

Our findings are in contradiction to the results of others studies presenting the introduction of environmental taxes (including taxation of energy use) as a negative influence for renewable energy consumption [67,68,69,70].

Regarding the nexus between financial development and renewable energy consumption, the results of our study are similar to most of the previous ones. Amin et al. [23] and Shahbaz et al. [22] identified a one-way relationship from financial development to renewable energy consumption, and the influence exerted is a negative one: increasing financial development generates a reduction in renewable energy consumption. Similarly, our study reveals that market capitalisation has a negative impact on renewable energy consumption. This emphasises the need for better environmentally friendly policies to control energy demand and reduce energy consumption. Such green policies should stimulate future financial development by promoting technological innovation and also increasing the number of job opportunities.

The results obtained in our analysis are in contradiction with the conclusions according to which financial development leads to the increase of industrialisation, to the consumption of fossil fuels, and implicitly to the decrease in the use of renewable energy sources [71] or those proving that financial development, by increasing FDI in developing countries, contributes to environmental degradation by increasing dependence on traditional energy sources [72].

However, Mukhtarov et al. [18], Lahiani et al. [19] and Lei et al. [20] obtained empirical results that revealed a positive and statistically significant influence of financial development on renewable energy consumption. In our study, the external debt stocks positively influence renewable energy sources. The positive influence of the external debt stocks can be explained by the fact that part of the external loans committed is oriented towards investments in renewable energy sources. Also, the fiscal policy is increasing the taxes on energy coming from traditional sources, which stimulates the consumption of renewable energy.

5. Conclusions, Practical Implications and Limitations

The purpose of this study was to identify the factors influencing the increase of renewable energy consumption and to evaluate its long-run impacts. After establishing our research hypotheses, we employed bivariate regressions to test the influence of different variables, namely the real GDP per capita, the capitalisation of listed domestic companies, the external debt stocks and the environmental tax revenues from energy taxes on the increase of renewable energy consumption. Our research was based on the ARDL model, which is adequate to test the nexus between fiscal policy, financial development, economic growth and renewable energy consumption, based on a small sample, reflecting the Romanian annual data series from 2000 until 2020.

The originality of our research comes from the methodology employed, which is innovative for studying the relationships observed in our research. The majority of the previous studies focus on the nexus between macroeconomic variables and renewable energy consumption. Many authors are testing various models with different selected variables, periods and countries. However, our study fills several gaps in the literature, from two different perspectives: first, limited research explores the nexus between external debt stocks and renewable energy consumption; second, this research is one of the few to test the nexus between the selected variables through the ARDL model. The ARDL test is recommended over traditional co-integration tests because it takes into account the degree of integration of the variables contained in the model. Moreover, it is adequate for small sample analyses.

We obtained specific findings for Romania, connected to the level of economic development. Nevertheless, the co-integration tests strongly support our main findings. Based on the fact that similar results were presented in the literature for other countries and different regions worldwide, our results are robust and validate the three hypotheses initially described. In addition, the results of this study support the growth hypothesis that considers energy as a significant source of growth in terms of the one-way causality from energy consumption to economic growth.

The results of this study indicate that all variables employed in the ARDL model (associated with the real gross domestic product per capita, the capitalisation of listed domestic companies, the external debt stocks and the environmental tax revenues from energy taxes) have significant impacts on the increase of renewable energy consumption. Four unidirectional causal relationships were identified in the long run: the direction of the causality from energy taxation to renewable energy consumption and the causality determined by the market capitalisation on the change of renewable energy consumption; then, the causal direction between the consumption of renewable energy towards the accumulation of income from energy taxation but also towards the real GDP per capita.

As for the policy implications and recommendations, we emphasise the impact and the utility of these results below. From a more practical point of view, this research provides the most suitable solutions to improve renewable energy consumption in Romania and mitigate climate change’s impact. The government has a huge role to play in enhancing investments in renewable energy sources and creating awareness about their benefits. A precise and more evident illustration that could be interpreted from the business viewpoint is that increasing the number of consumers using renewable sources has benefits for the business itself. They need to use various sources of renewable energy in order to protect the environment. It is highly recommended that proper awareness campaigns should be in place on the part of small- and medium-size enterprises so that consumers become more aware of the adoption of these renewable energy sources. Campaigns should directly emphasise the benefits they can give to both the business and the consumers. Hence, we recommend the development of market-specific mechanisms as some specific incentive marketing campaigns to emphasise the maximum benefits to the participants and to make them realise the full potential of integrating renewable energy sources.

Moreover, the health crisis has increased concerns about climate change and stressed the importance of increasing sustainability and protecting the world. The need for decarbonisation requires changes in governmental and corporate strategies in terms of energy efficiency and renewable energy sources. On a global level, China and US represent the major polluters, currently finding ways for decarbonisation, while in Europe, through the European Green Deal, member countries are committed to a cleaner and circular economy. We believe that the energy transition to low carbon emissions in Romania could be achieved only by synchronising private and public initiatives. In all EU member states, the decarbonisation of the energy sector is primarily based on the support provided by the European Green Deal. Developing a resilient and flexible infrastructure became the central element of integrating renewable energy sources. Also, a high degree of network stability can facilitate the energy transition. Another recommendation coming from this research is that adopting new technologies for energy storage, improving energy efficiency and increasing decentralised production would increase the share of clean energy that will be installed.

Besides tackling the imperative to decarbonise economies, renewable energy can design solutions for economic growth: stimulating the use of renewable energy sources will create new employment opportunities, increase human welfare, and contribute to a climate-safe environment. In this view, the policymakers in Romania should adopt green energy policies by increasing the investments and production in renewable energy, leading to improved energy efficiency, decreased energy costs, and sustainable energy development. Furthermore, the results of the study suggest that policymakers in Romania could use the development of the financial sector to mitigate environmental degradation by promoting investment in energy and production through renewable energy resources. Overall, the findings of this research provide useful conclusions and policy recommendations for Romania to head towards sustainable and green economic growth through a balanced set of policies and smart measures, accompanied by a solid rate of absorption of green funds. Romania has major potential to play a significant role in decarbonisation efforts, both at the EU and global levels.

This research focused on data series from Romania to study the specific impact of fiscal policy, financial development and economic growth on the constant increase of renewable energy consumption. However, some limitations of this study are mainly attributed to the lack of available data for specific variables. Regarding the selection criteria of the variables, other control variables may also be considered (for example, bank deposits to GDP ratio and loan to deposits ratio). Future empirical research may extend the geographic analysis, highlighting the similarities and differences between EU countries and the particularities of different sectors of activity.