Introduction

There has been extensive scholarly interest in the link between renewable energy consumption (henceforth: REC) and its drivers during the past few decades, perhaps due to growing knowledge of greenhouse gas emissions and their detrimental environmental implications. However, energy is critical for developing and advanced economies to attain sustainable development. Population growth, improved standard of living, production enhancements, and economic competition have resulted in a substantial increase in energy usage over the last decade [1]. However, according to [2], the proportion of fossil fuels to total energy is very high; although renewable energy demand grew by approximately 5% each year from 2009 to 2019, the share of fossil fuels remained at approximately 80% of total energy usage over the same period. Excessive fossil fuel combustion results in large carbon dioxide (CO2) releases into the atmosphere, leading to adverse environmental impacts such as climate change [3]. Since it threatens the natural equilibrium of life, the greenhouse effect is one of the most pressing issues of our era [4]. One option for preventing global warming and its dire consequences on the environment involves minimizing the use of fossil fuels in power generation and transitioning to renewable energy sources that do not generate greenhouse gases. In addition to reducing environmental externalities, renewable energy may safeguard economies that rely heavily on energy imports against exogenous shocks [5, 6]. Although the link between energy demand and economic growth has gotten much attention in energy-growth literature, it is still unclear how these variables influence one another due to differences in econometric methods, the measure of financial development investigated, the nation and the period under consideration.

The start-up costs and operating expenditures of renewable energy projects are more than those of fossil fuels [7,8,9]. Mukhtarov et al. [9] and Best [10] suggested that an advanced financial sector can accelerate renewable energy sector development by efficiently channelling funds towards the sector. To buttress this, Kim and Park [11] pointed out that renewable energy projects highly dependent on equity and debt financing advance quicker in economies with developed financial sectors than in underdeveloped ones because institutional and equity financing are more appropriate for capital-intensive investments [12]. Since Ghana embarked on the Financial Sector Adjustment Program (FINSAP) in 1988, financial liberalisation in 1990, the implementation of the universal banking system in the first quarter of 2003, and the financial sector clean-up, which reduced the number of universal banks from 30 at the start of 2018 to 23, the country’s financial sector is still intensifying reforms and liberalisation. Thus, its influence on the energy industry is in a continuous state of flux. Therefore, it is necessary to explore the relationship between REC and financial development (henceforth: FD), specifically how FD impacts REC.

Against this backdrop, this study empirically examines how financial development impacts renewable energy consumption. For this purpose, various econometrics techniques were utilised for Ghana. Ghana is a developing country with a population above 32 million. As the population continues to increase, the energy demand also increases. However, Ghana imports a more significant share of its energy needs. Due to this occurrence, Ghana continues to maintain a negative current account (% GDP), which increased from −2.73% in 2019 to −3.04% in 2020, mainly driven by sluggish oil production, cocoa production, and higher energy imports. The link between financial development and renewable energy usage has been studied in a number of countries, which are covered in subsequent sections. The current paper deviates from previous studies in various ways. First, to the best of the authors’ knowledge, this is perhaps the first study to explore the nexus between financial sector development and renewable energy usage in Ghana at the individual country level. Kwakwa [13] is the only study that attempted to link renewable energy usage to financial development in Ghana. He investigates the factors influencing renewable energy use in Ghana. From the outcomes of the ARDL and FMOLS models, he established that financial development has a deleterious effect on renewable energy usage. The current study focuses on the impact of financial development on renewable energy consumption over a relatively long period. In Addition, we attempt to fill this gap by proposing a model to investigate the impact of financial development, economic growth and energy prices on renewable energy usage in the case of Ghana, employing the ARDL, Gregory and Hansen cointegration, VECM, Bayer and Hanck cointegration, fully modified ordinary least squares (FMOLS), canonical cointegrating regression (CCR) and dynamic ordinary least squares (DOLS) techniques.

In lieu of relying on conventional measures of financial development, our study utilizes a comparatively new and detailed financial development index created by Svirydzenka [14]. This index measures the overall financial development capturing the development of both financial institutions and financial markets. We enrich the extant literature by examining the theoretical link between financial development and renewable energy in Ghana. In addition, the empirical findings derived by analysing the link between financial development, economic growth, energy prices, and renewable energy consumption can aid policymakers in Ghana in formulating policies to achieve the goal of renewable energy consumption; this is the pragmatic contribution of the paper. This paper’s findings demonstrate that financial sector development in Ghana predominantly promotes renewable energy usage. However, renewable energy investments are influenced by numerous factors. Therefore, policymakers face a colossal task in formulating particular responses to this circumstance. In this context, one of the paper’s contributions is to suggest what should be prioritized to accomplish this objective.

The following structure applies to the remaining sections: Sect. 2 offers an overview of the renewable energy profile of Ghana and relevant literature on the subject; Sect. 3 describes the method and data sources; the fourth section contains the results of the analyses and the discussion. Finally, the conclusion, which includes an overview of the results and policy ramifications, is discussed in the last part.

Ghana renewable energy profile and related literature review

Ghana renewable energy profile

Ghana is an energy-importing and energy-exporting country with an energy dependency problem. Ghana imports a substantial portion of its energy requirements from countries such as La Cote D’Ivoire and other parts of the world while exporting power to Togo, Benin and Burkina Faso. Ghana’s primary energy sources are hydro, thermal and solar. However, a significant portion of energy requirements is met by fossil fuels. Ghana has abundant renewable energy resources that have yet to be completely tapped. These comprise bioenergy, hydropower, wind and solar energy. In 2011, the Renewable Energy Act (832) was enacted to transform these renewable potentials into actual energy by fostering an environment conducive to attracting private investment in the renewable energy sector.

Figure 1 depicts the trend of REC (% of total energy usage) on a downward spiral. In 1990 the proportion of renewable energy in the total energy consumed was approximately 80%, which declined to approximately 42% in 2019. This raises the question of why REC (% of total energy consumption) declines despite growing income levels and the financial sector deepening. Is it that the increasing income levels and deepening of the financial sector are not promoting the generation and use of renewable energy in Ghana, or are some other variables accounting for this? These are the questions the paper attempts to find solutions to. Ghana is currently at the centre of urbanisation in West Africa, notably in education, technology, and the economy. Additionally, industrialization has contributed to an increase in energy consumption. This has compelled the government to pursue alternate renewable energy sources to supplement the predominant hydroelectric energy supply, such as the Akosombo Dam on the River Volta. However, the electricity producing company, Volta River Authority (VRA), cannot meet users’ demands. Therefore, the nation has been obliged to practice electricity rationing, sometimes known as “Dumsor,” over the past decade. Therefore, Ghana must increase its renewable energy production not just to fulfil the energy needs of its citizens but also to take the lead in lowering CO2 emissions and combating climate change [15]. In light of these energy problems concerning economic growth, it is crucial and feasible for Ghana to implement steps to deepen, diversify, and seek cheap alternative energy sources to enhance the energy supply to meet demand.

Fig. 1
figure 1

A plot of renewable energy consumption and financial development in Ghana (1990–2019)

Full size image

Relevant literature review

Literature on energy usage and economic growth has received significant interest in the past decades, albeit conflicting findings. The findings are focused on four key hypotheses. One is the conservative hypothesis provided by the pioneering study of Kraft and Kraft [16], which opines that there is unidirectional causation from growth to energy usage, e.g., [9, 17,18,19,20,21,22,23,24,25]. Consequently, as economies expand, the demand for energy to support greater quality of life will also increase. On the other hand, some studies have also claimed that energy is a crucial aspect of industrialisation, contributing to economic expansion [26,27,28,29,30,31,32,33]. These studies lend credence to the growth hypothesis. Some studies also found a two-way (bidirectional) causality between growth and renewable energy usage [34,35,36,37,38,39,40,41,42,43,44]. These studies’ findings are in sync with the feedback hypothesis, which incorporates the conservative and growth hypotheses. It postulates that a stable energy supply promotes economic growth, which drives the demand for energy consumption. Finally, other studies have validated the neutrality hypothesis, which asserts no causality between economic growth and renewable energy usage [45,46,47,48,49,50,51].

Recent years have witnessed the emergence of the finance-energy nexus as one of the most pressing matters in economics and finance. Consuming fossil fuels is known to increase the amount of atmospheric carbon dioxide [52], while it is acknowledged that using renewable energy decreases carbon dioxide emissions [53,54,55,56,57]. For example, some studies found that financial development drives energy demand [58,59,60,61,62,63,64,65,66,67,68,69,70]. Other studies, however, demonstrated a detrimental impact of FD on REC [22, 71,72,73]. However, a study by Denisova [63] demonstrated that FD does not affect REC. Other studies also argued that the FD-energy nexus is nonlinear; thus, they tested for asymmetric effects and nonlinearities in the link (see [23, 74,75,76,77,78], among others).

The study’s primary purpose is to investigate the impact of financial development on adopting renewable energy, which is a fascinating field of study. Renewable energy projects demand high initial outlays and extended payback periods [52, 79, 80]. This circumstance is a crucial obstacle to improving these projects [9]. Recent research on the FD-REC nexus has shown that advancements in financial intermediation operations drive down the costs of renewable energy investments and stimulate the usage of renewable energy [81,82,83]. Market-based financial development also promotes productivity by channelling capital towards green energy projects, stimulating the demand for green energy [11, 58, 59, 84, 85]. To buttress this, Brunnschweiler [86] argued that advanced financial systems could efficiently channel investments toward the green energy sector, while weak financial systems may inhibit the flow of capital toward the green energy sector. Therefore, participants in the renewable energy sector would favour nations with a developed financial system due to the high likelihood of obtaining access to funding quickly [87, 88].

Additionally, McFarlane et al. [74] emphasized that progress in the financial sector improves financial intermediation, making it easier for businesses and households to take loans (as well as increasing investment opportunities). This allows individuals to spend more money on luxury products and increase the demand for energy, largely when the goods in question are automobiles and machines [88]. However, until now, studies on the exact impact of FD on REC are limited. Babbar and Schuster [89] uncovered significant knowledge gaps in funding renewable energy projects and emphasized the need for private participation in developing nations.

Regarding 34 developing nations, Shahbaz et al. [52] assessed the effect of FD on REC from 1994 to 2015. The empirical evidence indicates that FD drives REC, whereas economic growth negatively impacts REC. Regarding the effect of energy prices on REC, the study found that it had no effect. Consequently, they asserted that REC would lower carbon dioxide emissions compared to fossil fuels. To fulfil sustainable development objectives, the government should thus introduce incentives and tax policies encouraging companies’ demand for renewable energy use. Mukhtarov et al. [9], utilising the VECM and ARDL techniques, studied the effect of FD on REC in Turkey from 1980 to 2019. They discovered that FD has a beneficial influence on the REC. In addition, the study indicates that economic growth favourably influences REC. However, their analyses demonstrated that price increases harm renewable energy consumption.

Khan et al. [90] applied a panel quantile regression technique for a worldwide panel of 192 nations from 1980 to 2018. The heterogeneity of REC, FD, and CO2 emission was investigated. They discovered that REC dampens CO2, but FD affected REC favourably. Employing annual data from the top 15 REC countries and a panel smooth transition model (PSTR), Raza et al. [77] found that a high (low) FD regime has a positive (dampening) effect on REC. In the case of India, Eren et al. [1] investigated the relationships between FD, economic growth, and REC. Their dynamic ordinary least square (DOLS) estimate results indicate that FD has a beneficial effect on REC. Concerning causation, they supported the feedback hypothesis and one-way causality from FD to economic growth and REC without feedback.

Anton and Nucu [91] based their study on the fixed effect model for a panel of 28 European Union (EU) nations from 1990 to 2015. Their findings show that FD generally has a positive effect on REC. However, the data indicate that the expansion of financial markets has little impact on REC in new EU member nations. Kim and Park [11] found that in countries with robust financial markets, renewable energy sectors more reliant on debt and equity funding expand rapidly. Using a panel of 20 emerging nations covering 1991–2012, Paramati et al. [85] concluded that stock mark development, foreign direct investment and growth positively affect REC. The findings lend credence to Wu and Broadstock [92], who found that FD positively influenced REC for 22 emerging economies from 1990 to 2010. Using the ARDL-PMG model, Wang et al. [6] showed that economic growth stimulates REC while FD harms it for both China and western China. Ali et al. [93] confirmed the feedback loop between FD and REC variables. Rasoulinezhad and Saboori [94] found a one-way causality from FD to REC. Hassine and Harrathi [27] investigated the FD-REC nexus for the Gulf Cooperation Council nations from 1980 to 2012. They established a unidirectional causality emanating from the REC to FD. The study further showed that the long-run coefficients from REC to FD are positive and significant.

The FD-REC nexus for each nation or regional bloc is a priori uncertain, based on the review of relevant literature. Therefore, the link must be empirically determined. Unfortunately, to the best of the authors’ knowledge, empirical research has not yet explored the nexus between Ghana’s financial sector development and renewable energy usage. Second, most studies concentrated on a particular indicator and renewable energy proxy. Nonetheless, additional variables may have a substantial influence on these projects. Hence, a study that evaluates how several variables affect REC concurrently is necessary to comprehend crucial challenges for enhancing renewable energy investments. In this context, we examine how financial development, economic growth and energy prices impact renewable energy usage. Third, we observed that existing studies on the FD-REC nexus have largely employed numerous quantity-based indicators of FD, such as domestic credit (% GDP), total value of stocks traded (% GDP), liquid liabilities (% GDP) and stock market turnover (% GDP) (e.g., [6, 9, 13, 65, 92, 95, 96]). Nevertheless, financial development is a multifaceted dimension; thus, measuring it with any of the enumerated variables might result in excluding other crucial dimensions. To address this problem, this study’s measure of FD is predicated on a comparatively recent and in-depth index constructed by Svirydzenka [14]. Analysing this comprehensive dimension within the REC framework will yield results that might pave the way for future investigations seeking to influence renewable energy and financial development policy. Despite all this empirical evidence, financial sector reforms, and developments in the renewable energy sector in Ghana, one needs to be reminded by Kwakwa [13] that there is little unanimity on how FD relates to REC.

Given that renewable energy is much more expensive than conventional energy sources [97], improved access to financial resources is essential for its advancement. That is to say, the development of the financial sector may facilitate investments in renewable and clean energy sources. Therefore, it is anticipated that there is a favourable relationship between FD and REC. Drawing on recent studies, the following hypotheses are examined:

H1o

Financial development is not positively related to renewable energy consumption in Ghana. Hence, increases in financial development do not promote renewable energy consumption in the long-run.

H1a

Financial development is positively related to renewable energy consumption in Ghana. Hence, increases in financial development promote renewable energy consumption in the long-run.

H2o

There is no cointegration between financial development and renewable energy consumption in Ghana. Hence, there is no long-run association between the two variables.

H2a

There is cointegration between financial development and renewable energy consumption in Ghana. Hence, there is a long-run association between the two variables.

Methods

Several authors [23, 52, 65, 78, 98] modelled energy consumption as a function of financial deepening, energy cost and economic growth. Using the same conceptual model, McFarlane et al. [74], Mukhtarov et al. [9], Shahbaz et al. [52], and Mukhtarov et al. [66] developed a model in which REC is a function of financial development, real GDP per capita and consumer price index. This model was further extended by Anton and Afloarei Nucu [91] by including foreign direct investment (FDI). However, some studies [23, 59, 99] utilised FDI to measure financial development. Considering these studies, we failed to capture FDI as a separate term in our model specification. Therefore, in sync with previous studies, we specify the model as follows:

$${lnREC}_{t}={\alpha }_{0}+{\alpha }_{1}{lnFD}_{t}+{\alpha }_{2}{lnEG}_{t}+ {\alpha }_{3}{lnCPI}_{t}+ {\varepsilon }_{t}$$
(1)

In Model 1, \({REC}_{t}\) represent renewable energy consumption; \({FD}_{t}\) denotes financial development; \({EG}_{t}\) denotes economic growth; and \({CPI}_{t}\) represents energy prices. \({\varepsilon }_{t}\) is the stochastic error term. Level or log forms of data can be used to estimate Model 1. A priori FD (\({\alpha }_{1}\)) and EG (\({\alpha }_{2}\)) are expected to increase ( +) REC, while CPI (\({\alpha }_{3}\)) is expected to drag (−) REC.

This study examines the impact of FD on REC while controlling for EG and CPI in Ghana. Given the data availability, the sample period spans from 1990 to 2019. The following data: renewable energy consumption (% of total energy consumption), economic growth (EG) measured by real GDP per capita (constant 2015 US$) and energy prices (CPI) measured by Consumer price index (2010 = 100) utilised in this study were all gleaned from World Development Indicators (WDI) [100]. However, due to the unavailability of data on energy prices for all nations, CPI, which has been used extensively to measure energy prices in earlier studies [9, 23, 59, 66, 74, 101, 102], was employed in this study. For our analysis, FD is measured by a financial development index introduced by Svirydzenka [14], which encompasses the overall depth, accessibility and efficiency of financial institutions and markets. They have since been updated and are available in the IMFs [103] financial development index database. Table 1 provides a brief descriptive analysis of the variables before their natural logarithm transformation. REC ranges from 41.480 to 82.928% of total energy consumed, FD ranges from 0.082 to 0.153, EG ranges from 891.305 to 2053.587, and CPI ranges from 1.881 to 278.452. The Jarque–Bera test demonstrates that all variables have a normal distribution, as shown by the corresponding p-value. Figure 2 depicts graphical representations of the variables.

Table 1 Descriptive statistics.
Full size table
Fig. 2
figure 2

Source: WDI and IMF Database

A plot of variables used in the study.

Full size image

We convert all variables to their natural algorithmic form for estimation and interpretation of results purposes. In this study, we evaluate the long-run impact of FD on REC while controlling for changes in EG and CPI using the ARDL bound test cointegration technique. This technique was chosen above conventional cointegration approaches because it yields accurate and reliable results for small and finite sample data. It is suitable when dealing with a mixture of I(0) and I(1) variables. Furthermore, the ARDL utilise dynamic specifications, which account for the effect of the lagged values of the outcome and the explanatory variables. Additionally, this flexible specification technique permits a simultaneous estimate of the long- and short-run associations. The paper employs the Zivot-Andrews [104] unit root test to explore the unit root characteristics and structural break(s) presence in the series. Traditional unit roots testing approaches such as Augmented Dickey-Fuller (ADF) and Phillips and Perron (PP) are unsuitable and may yield biased results in the presence of structural break(s) in the data [95, 105]. The ARDL technique necessitates the series is at most I(1). After estimating the ARDL model, we do the bounds test. The estimation of the ARDL model is given by Model 2:

$${\Delta lnREC}_{t}=\,{\Phi }_{0}+{\Phi }_{1}{lnREC}_{t-1}+{\Phi }_{2}{lnFD}_{t-1}+{\Phi }_{3}{lnEG}_{t-1}+{\Phi }_{4}{lnCPI}_{t-1}+\sum_{i=1}^{n}{\varkappa }_{i}{\Delta lnREC}_{t-1}+\sum_{i=1}^{n}{\beta }_{i}{\Delta lnFD}_{t-1}+ \sum_{i=1}^{n}{\sigma }_{i}{\Delta lnEG}_{t-1}+\sum_{i=1}^{n}{\psi }_{i}{\Delta lnCPI}_{t-1}+{\varepsilon }_{t}$$
(2)

For the ARDL model, the variables have already been specified, n signifies the maximum lag length, and ln denotes the variables converted into natural logs. The first component of model 2, with \({\Phi }_{1}\),\({\Phi }_{2},\) \({\Phi }_{3}\) and \({\Phi }_{4}\) represents the long-run parameters, and the second component with \(\varkappa ,\) \(\beta\), \(\sigma\) and \(\psi\) indicates the short-run parameters. The null hypothesis is that cointegration does not exist H0:\({\Phi }_{1}\)= \({\Phi }_{2}=\) \({\Phi }_{3}\)=\({\Phi }_{4}\)  = 0 and the alternate hypothesis H1: \({\Phi }_{1}\)≠ \({\Phi }_{2}\ne\) \({\Phi }_{3}\ne {\Phi }_{4}\)≠ 0 suggests cointegration among the series. The bounds test consists of the F-bounds test and the t-bounds test. The F- and t-statistic for the model are calculated using the parameter estimates given in Model 2, which are then compared to their corresponding I(0) and I(1) values at the selected significance level. The null hypothesis associated with each test is rejected if their estimated values exceed their corresponding I(1) values at the acceptable significance level. Then the analysis leads to the conclusion that the variables are cointegrated. The maximum lag (n) was determined using the Akaike information criterion (AIC) so that the estimated model’s residuals are free from serial correlation, heteroscedasticity, or non-normality. In addition, to check the stability of Model 2, we employ the cumulative sum (CUSUM) of recursive residuals and the cumulative sum (CUSUM) of recursive residuals. Lastly, we undertake the Ramsey RESET test to ensure the absence of any functional form misspecification in the ARDL model. If the model is stated accurately, no nonlinear function of the explanatory variables should be a significant predictor of the outcome variable.

However, if the time series variable includes unidentified structural breaks due to economic crises, exogenous shocks, policy shifts, or technological shocks, using conventional cointegration techniques to the residuals of regression will result in specification errors [106]. Gregory and Hansen [106] presented a cointegration test to account for the potential of regime shifts resulting from a single endogenous structural break. Therefore, to explore the robustness of ARDL cointegration outcomes in the existence of structural breaks, the regime-switching cointegration test proposed by Gregory and Hansen [106] and the combined cointegration test suggested by Bayer and Hanck [107] were implemented since all the series are I(1). The Bayer and Hanck [107] cointegration test focus on reducing superfluous multiple-test approaches to provide accurate estimates. After confirming cointegration among the variables, we use the ARDL model to evaluate their long-run relationships. To check the fortitude of the ARDL long-run estimates, the VECM, fully modified OLS (FMOLS) proposed by Phillips and Hansen [108], the canonical cointegrating regression (CCR) and dynamic OLS techniques developed by Stock and Watson [109] were utilized. FMOLS, DOLS, and CCR enable asymptotic coherence to be established by taking serial correlation into account.

The paper only estimates the long-run parameters of model 2. The estimation approaches employed in this study have been extensively utilised in comparable studies. In addition, several previous studies have provided detailed information [106, 107, 110,111,112,113,114], among others.

Results and discussion

Before testing for cointegration, we explored the order of integration of variables to confirm that none of the variables utilised in this paper is I(2). This is because the ARDL bound testing technique is inappropriate when variables I(2) are included. In addition, the Bayer-Hanck and Gregory and Hansen cointegration conditions demand that all variables be I(1). We explore the unit root characteristics and the presence of structural breaks in the variables using the Zivot and Andrews [104] unit root test. The outcome of the unit root tests is reported in Table 2. The results suggest that none of the variables is stationary at level. After taking the first difference, however, all variables were stationary, with structural breaks for lnREC, lnFD, lnEG, and lnCPI in 2014, 2011,2014, and 2000, respectively. Consequently, the ARDL, Gregory and Hansen, and Bayer and Hanck cointegration procedures are applicable.

Table 2 Zivot-Andrews unit root test outcomes.
Full size table

After determining via the Zivot and Andrews unit root tests that all the variables are I(1), the cointegration bound test grounded on the well-specified ARDL model is next conducted. To apply the ARDL bounds test method, the appropriate lag length must first be identified. Therefore, using a randomly selected lag length, we specify a VAR model where all the variables of interest are endogenous. Finally, the lag order selection criteria test was executed on the residuals to determine the optimum lag interval for the cointegration bounds test. Table 3 displays the results of the lag interval estimations. In this study, all the VAR lag order selection criterion confirms that a lag order of one is optimum.

Table 3 Lag selection test.
Full size table

The VAR has worthwhile qualities, according to the diagnostics tests reported in Table 4. The residuals of the estimated VAR do not exhibit heteroscedasticity, serial correlation and are normally distributed. The model is also stable, no root lies outside the unit circle.

Table 4 Diagnostics results for the VAR model.
Full size table

The cointegration association between the variables was evaluated using the ARDL bounds test for cointegration. Table 5 summarizes the findings of the cointegration analysis. The relationship represents REC as a function of REC, EG, FD and CPI. We demonstrate that the F-test statistic for Model 2 surpasses the I(1) value at the 1% significance level (7.92 against 5.61). The F-test indicates that we cannot accept the null hypothesis. The t-test result supports this since the estimated value in absolute terms surpasses the I(1) value at the 1% significance level (-5.10 against -4.37). Therefore, we deduce that the variables have a cointegrating relationship.

Table 5 ARDL bounds test.
Full size table

The Bayer-Hanck [107]combined cointegration, which estimated Fisher statistics for EG-JOH and EG-JOH-BO-BDM and the Gregory and Hansen [106] cointegration tests results are presented in Table 6. At a 5% significance level, the results of both tests suggest that the variables are cointegrated, indicating a long-run association among the variables.

Table 6 Cointegration tests outcomes.
Full size table

Given the cointegration between the outcome variable and the regressors, we estimate the long-run relationship between the variables using the ARDL technique. Then, we performed model diagnostics check using the residuals of the estimated ARDL model. We can conclude from Table 8 that the ARDL model is statistically appropriate. First, we fail to reject the null hypothesis for Breusch-Godfrey serial correlation LM Test, Breusch-Pagan-Godfrey heteroskedasticity test and Jarque–Bera normality test since their associated p-values are greater than the 10% significance threshold. Moreover, neither the t-statistic nor the F-statistic of the Ramsey RESET test is significant, indicating that the model is devoid of any functional form misspecification. Finally, as shown in Figs. 3 and 4, the CUSUM and CUSUM of squares are within their respective 5% confidence intervals, indicating no evidence of model instability at the 5% confidence level or below.

Fig. 3
figure 3

Plot of the cumulative sum of recursive residual

Full size image
Fig. 4
figure 4

Source: The author

A plot of the cumulative sum of squares of recursive residuals.

Full size image

Finally, we employ the VECM, FMOLS, DOLS and CCR techniques to ascertain the ARDL model’s robustness. The VECM residuals were also analysed, and Table 7 demonstrates that all the results met the acceptance criteria. As depicted in the Table, there are no problems of serial correlation, heteroscedasticity or non-normality.

Table 7 ARDL, VECM, DOLS, FMOLS and CCR long-run estimates.
Full size table

The long-run estimates of these five methods are significant and comparable in sign and magnitude. The outcomes of the long-run estimates reveal that, at a significance level of 5%, a rise in FD increases REC. From Table 7, a 1% rise in FD leads to 0.237, 0.534, 0.276%. 0.178% and 0.195% rise in REC as indicated by the ARDL, VECM, DOLS, FMOLS and CCR long-run estimators, accordingly. Our results accord with economic theory, Mukhtarov et al. [9] for Turkey and Shahbaz et al. [52] for 34 developing economies but contradicts the findings of Kwakwa [13]. They have empirically shown that FD has a deleterious impact on REC in Ghana. The discrepancies in findings may be attributable to the measures of FD and sample period. Our outcome is based on a more accurate measure of FD and a larger sample size. From a theoretical viewpoint, the beneficial impact of FD on REC suggests that advanced financial systems enable institutions and individuals to easily access inexpensive sources of funds to expand or start their businesses or acquire luxury items, increasing the demand for renewable energy.

In contrast, the results show that at the 1% significance level, EG has a deleterious effect on REC. The implication is that holding other variables constant, a 1% expansion in the economy will lead to a 0.642% decline in REC. A plausible explanation is that as Ghana’s per capita income increased between 1990 and 2019, REC decreased due to inadequate income levels, and consumer spending drifted to fossil fuels. The empirical analysis confirms the findings of Shahbaz et al. [52] for developing countries but contradicts the findings of Mukhtarov et al. [9] for Turkey and McFarlane et al. [74] for Jamaica which established that economic growth drives REC. We also established an inverse association between REC and CPI, which is significant at the 1% level. This finding concurs with economic theory and empirical findings of Mukhtarov et al. [9] and McFarlane et al. [74]. However, this empirical evidence differs from Shahbaz et al. [52], who confirmed the neutral effect of energy prices on REC.

Additionally, the model’s error correction term was estimated. The error correction term generated by the ARDL bounds testing approach was negative and significant at the 1% level, in accordance with theory. The implication is that there is a mean-reverting process at work and that, following a shock, the system’s disequilibrium tends to be corrected toward the long-run path (see Table 8).

Table 8 Diagnostic results for the ARDL model.
Full size table

Conclusions and policy implications

Despite its economic and environmental downsides, the use of fossil fuels as energy sources is nevertheless widespread. Due to global environmental deterioration and macroeconomic issues, policymakers and scholars are interested in renewable energy resources and their usage. As renewable energy sources become prevalent worldwide, we investigate the dynamic relationship between financial deepening and renewable energy usage. To the authors’ knowledge, the impact of financial development on renewable energy consumption while controlling for income levels and energy prices has not been explored comprehensively using newly developed econometric approaches in the case of Ghana. Using the ARDL, Bayer and Hanck, Gregory and Hansen cointegration, VECM, FMOLS, DOLS, and CCR tests, this paper attempts to address this lacuna in the literature on renewable energy. Our major discoveries are twofold. First, the outcome of the ARDL, Bayer and Hanck, Gregory and Hansen cointegration tests suggest a long-run linkage between renewable energy usage and its potential drivers (namely FD, EG and CPI). Second, the outcome of the ARDL, VECM, DOLS, FMOLS and CCR long-run estimators suggest that increasing levels of FD promote REC. However, increasing energy prices and economic growth negatively influence renewable energy usage.

The study has both practical and societal ramifications. The practical ramifications are the deepening of the financial sector, which is Ghana’s dominating strategy for sustainability and energy-related policymaking. Consequently, the beneficial causal effect of FD on REC implies that increasing cost of borrowing, higher financial risk, high cost of capital, lessening information transparency, and difficult access to funds are likely to have a deleterious impact on REC, thereby harming the environment. Concerning the social ramifications of our findings, we claim that a higher level of financial development in climate financing will positively affect society. Therefore, financial institutions should support environmental sustainability by providing affordable loans for developing the renewable energy sector. This would help Ghana achieve its sustainable development objectives and reduce its reliance on fossil fuels.

Furthermore, Ghana has embraced the Sustainable Development Goal (SDG) of the United Nations; Goal 7 of the Sustainable Development Goals is to ensure universal access to affordable, reliable, and reliable power services. However, this objective poses a difficult hurdle for Ghana because of its heavy reliance on non-renewable energy sources, despite the vast untapped potential for renewable energy. Therefore, it is suggested that policymakers establish the required stimulus and tax regulations to promote the demand for renewable resources among financial sector organizations. Furthermore, governments might give green energy investors interest-free financing facilities regarding incentives. Moreover, renewable energy investments are exempt from taxation. These factors have a favourable effect on cost reduction, enhancing the projects’ efficiency. A major limitation of this paper is that the analyses conducted are at an aggregate level. To the author’s best knowledge, the association between these variables have not been studied at the disaggregated level for the case of Ghana. Such an investigation is one direction for future study. Another avenue for future investigation could be to explore the asymmetric impact of financial development on renewable energy consumption in Ghana and other parts of the world.