Does domestic investment matter? A multivariate time series analysis of the energy-CO2 emission-growth nexus in Ghana

The economic cost of greenhouse gas (GHG) emissions to African economies have increased. Therefore, GHG emissions and their concomitant effect on the environment are fast becoming costly for emerging economies like Ghana. Hence, the justification for the growing literature on the subject. This study employed the Autoregressive Distributive Lag (ARDL) bounds test and Granger causality techniques with data from 1983 to 2014. The study examines the dynamic relationship between income growth, power consumption, and carbon dioxide (CO2) emissions in Ghana, capturing the role of domestic investment and foreign direct investment (FDI) in the nexus. All variables were found to be cointegrated in the long run based on the bounds test. The Granger causality test indicates a unidirectional causality from energy consumption to CO2 emissions and economic growth. Furthermore, a unidirectional causality from CO2 to economic growth was found in Ghana. Results from the error correction model and the bounds tests indicate that, while energy consumption increases carbon emissions by more than 44%, the lagged values of domestic investment were found to reduce CO2 emissions by more than 41% in both the short run and the long run. Due to the significant effect of domestic investments on the reduction of CO2 emissions, the study recommends policymakers to adopt policies that may increase domestic capital in place of FDI, which has been proven to exacerbate environmental degradation in host countries.


Introduction
Greenhouse gas (GHG) emissions are fast becoming a global concern with their simultaneous effect on climate change Adjei et al. 2022a, b;Khan et al. 2020). Experts predict that by 2030, global carbon emissions will be twice as much as the world would need to reach its target of 1.5 °C, which world leaders agreed upon in the 2015 Paris Agreement (IEA 2021). Also, approximately 90% of the world's population breathes polluted air, resulting in more than 5 million premature deaths yearly (IEA 2021). The economic cost of this menace to Africa was estimated at $450 billion in 2013 (Ayetor et al. 2021). Therefore, the cost of inaction to air pollution and its resultant climate change is immense and invaluable. Past studies on global carbon emissions (Acheampong 2018;Al-Ahmad et al. 2016;Ang 2007) have received much readership in the recent past due to their significance to the sustainability of the environment. To contain the globally rising temperatures, the net-zero emission policy, which aims at reducing global marginal carbon dioxide emissions to zero by 2030, would have to be supplemented by a concerted effort on the part of governments to cut down emissions drastically (IEA 2021).
In the past two decades, developing countries such as China and India have more than doubled their annual carbon emissions per capita through massive industrialization and economic transformation, with high-energy consumption Responsible Editor: Arshian Sharif * Bernard Boamah Bekoe bekoe92@hhu.edu.com . For instance, China's per capita carbon dioxide (CO 2 ) emissions of 2.6 Mt in 2000 almost quadrupled to 7.3 Mt in 2018 (Our World in Data 2022). Other developing countries like Ghana that seek to transition from lower-middle-income status to regional economic superpowers face the daunting challenge of balancing their energy consumption to boost economic activities without compromising the environment's safety (Appiah 2018).
Historically, Ghana has exhibited strong support for numerous global policies aimed at tackling climate change. The country ratified the Kyoto protocol in 2003 and remained a signatory to the Paris Accord of 2015, which aims at promoting comprehensive sustainable development for global economies (Abokyi et al. 2019). Albeit these prevalent efforts, Ghana's carbon dioxide emissions have doubled between 2000 and 2010 (Twerefou et al. 2016). Energy production and supply, predominantly through electricity, biofuels, and liquid and solid fuels, continue to escalate (see Fig. 1). The country's electricity from thermal power plants is projected to skyrocket from 3721.7 GWh (in 2008 figures) to 65,239.6 kWh by 2030 (Abokyi et al. 2021). Power demand by the industry sector alone is forecasted to rise from 3433.1 to 50,145.6 kWh within the same period.
A plausible explanation for these concerning projections may be the country's ambitious one-district-one-factory industrialization initiative that is set to bridge the income gap, improve living standards, minimize the alarming dependency ratio, and generate employment . Similar initiatives, such as planting for food and jobs and rural enterprise development programs, are notably all capital-energy-intensive programs that would, in turn, harm the environment through GHG emissions and land and forest degradation (Abokyi et al. 2021).
Ghana's desired economic transformation requires enormous domestic capital investments and a huge amount of the country's energy, which has been skewing towards fossil fuels in recent years (see Fig. 2). It is documented that Ghana's 2007 energy crisis cost the country about 1.8% of its GDP. This percentage point increased by 2% during the 2013-2014 energy crisis (Appiah 2018). With Ghana currently regarded as one of the fastest-growing economies in the sub-region in the past decade, the emissions are bound

Ktoe
Year to surge upward, exacerbating the already concerning 76% global GHG emitted by developing economies (IEA 2021). From a globalist's point of view, the recent ratification and commencement of the African Continental Free Trade Agreement (AfCFTA), headquartered in Ghana's capital-Accra, is set to cause an upward surge in domestic investment and foreign investment inflows (Adjei et al. 2022b;Uzuner et al. 2020), which comes with enormous energy requirements. Therefore, while these investments have the capacity to speed up the country's robust economic growth (Ahmad et al. 2020), there is the danger of their negative impact on the environment through carbon emissions (Fig. 3). Several studies have investigated the impact of foreign investments on the economy and environment in Ghana (Appiah 2018;Mensah et al. 2021;Musah et al. 2021). However, the impact of domestic investments on the economy and their effect on the environment have received less to no attention in Ghana. Capital formation, often referred to as domestic investment, is imperative for developing countries' economic growth (Kobayakawa 2019) as it increases their infrastructure needed for robust growth. However, these domestic or capital investments have been accused of resource depletion and environmental pollution, including climate change (Chen and Graedel 2015). This is due to the enormous amount of energy inputs and their resultant carbon dioxide and other greenhouse gas emissions. These investments are likely to skew toward Ghana's extractive industry due to the country's abundant mineral deposits and natural resources (Ben-Salha et al. 2018). This is backed by the factor endowment hypothesis (FEH), which postulates that economies are more likely to specialize in producing and exporting goods that stem from their abundant production factors in the presence of free trade. Thus, these investments are likely to increase economic activities in the country, which will ruin the environment through land degradation, silting of water bodies, GHG emissions, and deforestation (Busse 2004). On the contrary, Mahmood et al. (2019) posit that these domestic investments may be inimical to carbon dioxide emissions by reducing the direct impact of energy consumption on emissions. Several studies have found evidence supporting the invaluable role energy plays in economic growth (Alshehry and Belloumi 2015;Aïssa et al. 2014;Ahmad et al. 2019;Oztek and Simba 2020;Ozturk and Bilgili 2015). Therefore, emerging economies, predominantly from sub-Saharan Africa (SSA), are escalating their robust industrialization mechanism with high-energy consumption as a means of economic transformation (Khan et al. 2019) Mahmood et al. 2019. This capital-energy-intensive industrialization leads to emissions of greenhouse gases into the atmosphere, such as methane and CO 2 , exacerbating the economic costs to these economies (Ahmad et al. 2017;Sarkodie and Strezov 2018). Therefore, this study employed the ARDL bounds test and Granger causality techniques to assess the emission-energygrowth nexus while capturing the role of domestic investment in Ghana to ascertain whether it could serve as an alternative or complement to foreign direct investment (FDI).
The study seeks to contribute to knowledge in three ways: (i) contribute to the ongoing energy-growth-emission literature on Ghana by employing recent data; (ii) assess the role of domestic investments in Ghana's economic growth, energy consumption, and carbon emissions in both the short run and long run by incorporating gross fixed capital formation as domestic investment proxy. To the best of our knowledge, this study is the first to do so. (iii) Suggest policy measures based on findings that may be significant to Ghana's environmental safety and economic growth in view of the AfCFTA. Since Ghana shares major environmental similarities with many West African countries, the findings may be essential for countries' policymakers within the region.
The remainder of the study is organized as follows: "Literature review" entails the review and discussion of existing literature on the connection between CO 2 emissions and economic growth, energy and economic growth, investment and economic growth, and its effect on the environment. Residential Industry Transport other "Methodology" will encompass the data (materials) and methods employed in the study, while "Results" presents the results and discussion of the empirical simulations. "Conclusions and policy implications" entails the policy suggestions, conclusions, and recommendations.

Literature review
Extensive literature exists on the growth-energy-emission nexus (Al-Mulali et al. 2013;Alam et al. 2016;Andreoni and Galmarini 2016;Arouri et al. 2012). However, findings from these studies remain inconsistent due to the sensitivity of the parameters, country-specific effects, and time interval of the study. On the other hand, there is a lack of extensive research on the dynamic relationship between domestic investment, economic growth, and carbon emission in Ghana. However, some are existing literature on the correlation between domestic investment and economic growth or the former and environmental degradation, albeit their inconclusive outcomes.

Domestic investment effect on economic growth and CO 2 emissions
While many past studies have investigated the connection between domestic investment and economic growth globally, few studies have analyzed the linear effect of domestic investment and economic growth on the environment. Bouchoucha and Bakari (2021) empirically analyzed the individual impact of both domestic and foreign investment on Tunisia's economic growth by employing the ARDL and bounds test techniques. They concluded that, while domestic investment positively affects economic growth in the short run, both foreign and domestic investment have a negative effect on economic growth in the long run. An earlier study by Choe (2003), which adopted the panel VAR estimation method for 80 developing countries, also revealed a negative effect of domestic investment on economic growth. Adam (2009) employed the panel fixed effect and OLS techniques to investigate the relationship between domestic investment, foreign direct investment, and economic growth in SSA between 1990 and 2003. His results from both the fixed effect and the OLS estimation proved a positive effect of domestic investment on economic growth. Shabbir et al. (2020) investigated the individual effects of domestic investment and foreign private investment on the economic growth of Pakistan from 1980 to 2017. The results from the ARDL estimation showed a positive relationship between domestic investment and economic growth in the long run. However, in the short run, foreign capital and domestic investment negatively affected economic growth. Ogunjinmi (2022) also made similar findings when he employed the ARDL model to investigate the role of domestic investment in Nigeria from 1981 to 2019. His study revealed a short-run negative effect of domestic investment on economic growth. His results were corroborated by the findings of Musa et al. (2021), who also found a short-run negative effect of domestic investment on economic growth in Nigeria between 1981 and 2018. Evidence from Roy (2022) corroborates the positive effect of domestic investment on economic growth. His study, which employed panel fixed effect, system GMM, and OLS estimation for 40 European countries, revealed that domestic investment has the capacity to mitigate the adverse effect of aging on economic growth. On the other hand, Ajide and Ibrahim (2021) investigated the non-linear threshold effect of domestic investment on carbon emissions in G20 by adopting Hansen (1999) and the bootstrap method. Their study revealed a threshold of 3.086, after which domestic investment significantly increases carbon emissions. Similarly, Anwar and Elfaki (2021) investigated the effect of domestic investment, economic growth, and energy on carbon footprints in Indonesia between 1965 and 2018. Based on the FMOLS, DOLS, and canonical cointegration analyses, it was revealed that gross fixed capital formation reduced environmental pollution while economic growth and energy consumption degraded the environment. An extensive study by Kobayakawa (2019), which adopted the structural decomposition analysis, tried to ascertain the effect of capital formation on carbon footprint in developing countries. He concluded that developing countries that tend to increase investment in carbon-intensive capital formation achieved faster economic growth; however, it has an adverse effect on the environment. He opined that proper environmental space for carbon footprints should be left for low-income countries as a means to faster economic growth.

Energy consumption, carbon emission, and economic growth
Globally, studies on the energy-emission-growth nexus have been extensively investigated, albeit their mixed findings. Salahuddin and Gow (2019) tested the effect of energy consumption and economic growth on environmental quality by employing the ARDL and Toda-Yamamoto causality checks. Their results indicated a negative effect of energy consumption on all three indicators of environmental quality (CO 2 , energy intensity, and adjusted national savings) in Qatar. They also recorded a bidirectional causality between economic growth and environmental pollution. An earlier work by Al-Mulali et al. (2013) employed the canonical cointegration regression technique to investigate the energy-emission-growth nexus for all Latin American and Caribbean countries. Their study reported that while a bidirectional long-run causality exists between energy consumption, carbon emissions, and economic growth in about 60% of the sampled countries, the remaining 40% exhibited mixed results. Nketiah et al. (2022) investigated the role of energy, economic growth, and biocapacity on the ecological footprint in west Africa. The results from the FMOLS and DOLS estimation revealed a bidirectional causality between energy, economic growth, and ecological footprints. According to Huang et al. (2008), high-income group countries are likely to improve their environment through efficient energy use. Their study, which adopted the GMM-SYS approach, investigated the dynamic panel relationship between carbon emission and energy consumption for 82 countries. They revealed that while energy consumption drives economic growth in low-income countries, economic growth drives energy consumption in middle-and high-income countries. Alam et al. (2016) tested the prevalence of the EKC hypothesis in Brazil, China, India, and Indonesia by adopting the ARDL bounds cointegration technique. They demonstrated that while income growth in China, Indonesia, and Brazil mitigates CO 2 emissions, income growth in India significantly worsens its carbon emissions. This study is contextually different from our current study in the sense that while they adopted a panel technique that fails to ascertain the country-specific dynamics, our study adopts time series data to analyze the situation for one specific country. Andreoni and Galmarini (2016) found support for the work of Alam et al. (2016) in their study, which investigated the main drivers of CO 2 emissions for 33 world economies. They opined that economic growth is the main driving force behind global carbon emissions, especially in the case of China and India, which play an invaluable role in the global economic panorama. Munir and Riaz (2020) estimated the asymmetric impact of energy consumption on environmental pollution in Australia, China, and the USA by using the non-linear ARDL model. It was revealed that fossil energy consumption worsened environmental quality by increasing CO 2 emissions in Australia, China, and the USA. While Oztek and Simba (2020) in Tanzania found agreeing evidence for the negative impact of energy consumption on GHG emissions, Tamba et al. (2017) and Menyah and Wolde-Rufael (2010) found strong support for the energy-led-growth assumption in Nigeria and South Africa, respectively. Sarkodie and Owusu (2017) found similar evidence in Senegal. Their study, which employed the non-linear partial least squares technique, concluded that industrialization increased growth with its high-energy demand. However, urbanization and output decreased CO 2 emissions in Senegal. Nevertheless, their surprising outcome regarding the effect of urbanization on pollution may be attributed to the strong collinearity found in their explanatory variables.

Energy-growth-emission dynamics in Ghana
Studies on the energy-growth-emission nexus in Ghana have also produced inconclusive outcomes. For instance, by using ARDL and structural time series models, Acquah (2014) found that Ghana's productivity growth reduced carbon emission in the short run, while renewable energy had the same effect in the long run. Another observation from the study was the simultaneous effect of forest depletion on GHG emissions. Kwakwa et al. (2022) adopted the ARDL bounds testing technique to investigate the existing relationship between output, industrialization, carbon emission, and capital. They revealed that while industrialization and CO 2 emissions curtailed agricultural output, financial development and capital increased output. In assessing the situation of the EKC assumption in Ghana's manufacturing sector by utilizing time-series data from 1971 to 2014, Abokyi et al. (2021) found results contrary to the U-shape EKC assertion in Ghana's manufacturing sector. However, the ARDL bounds test validated energy consumption-led carbon emissions in the sector. Their findings were supported by a recent study by Kwakwa et al. (2022), which opined that between 1971 and 2018, Ghana's population and industrialization drive had been significant contributors to the country's carbon emissions. In an earlier study, however, Abokyi et al. (2019) also found no validation for the EKC assumption in Ghana's industrial sector. Their results from the ARDL bounds test and Granger causality checks revealed that fossil fuel, which is pivotal in Ghana's industrial output, contributed to Ghana's GHG emissions. Their findings were corroborated by Mensah et al. (2021), who analyzed the direct impact of Ghana's one-district-onefactory initiative on the environment by employing ARDL, FMOL, and Johansen cointegration techniques to test the EKC hypothesis. Although they found no validation for the EKC hypothesis in Ghana's industrial sector, they revealed that the ostentatious industrialization initiative would only benefit Ghana's environment and growth if cleaner energy were utilized. Their findings were supported by Minlah and Zhang (2021), who also failed to validate the EKC hypothesis in Ghana. However, they found a significant feedback effect between growth and emissions in Ghana between 1960 and 2014.
As evidenced in the reviewed literature, research on the role of energy or economic growth on carbon emission has been rigorous and extensive. However, results on the magnitude of effect or direction of causality remain an empirical issue that is mostly determined by choice of variable and sample region or country. Moreover, while extensive research has established divergent views on the effect of domestic investment on either environmental degradation or economic growth, few studies have analyzed the linear effect of domestic investment, economic growth, and energy on environmental degradation and the direction of causalities among them. The situation is even worse when it comes to Africa or Ghana. This study seeks to fill this gap by utilizing gross fixed capital formation as a proxy for domestic investment to investigate its effect on Ghana's energy, carbon emission, and economic growth. There are several studies that applied the ARDL bounds test (Sabir et al. 2020;Salahuddin and Gow 2019;Kwakwa 2021). However, to the best of our knowledge, this is the first study to utilize the ARDL bounds test to ascertain these dynamics in the context of Ghana. Based on our analysis, the discussed outcomes and policy recommendations would be a major contribution to the energy-emission-growth discourse in Ghana. Sarkodie and Strezov (2019) revealed that countries with a per capita income lower than $8910 are likely to damage their environment to promote economic growth. Since Ghana's per capita income falls below the threshold, the sign for GDP t is expected to be negative. Since most of the country's power comes from solid and liquid fuels (Abokyi et al. 2021), energy consumption is expected to increase carbon emissions, as evidenced by past research (Hanif 2018; Kwakwa and Alhassan 2018). The sign for the coefficient of E t is expected to be positive. Investment is expected to carry a positive sign. This is because investments drive industrialization, which requires energy consumption and results in CO 2 emissions. FDI is expected to positively influence economic growth while negatively affecting the environment due to its associated high-energy consumption, as evidenced by past research (Salahuddin et al. 2018).

Data and econometric model
This study adopts an augmented single equation aggregate growth model of Jorgenson and Wilcoxen (1993), which was later exploited by Ang (2008) and Shahbaz et al. (2013) within a multivariate framework to analyze the relationship between CO 2 , energy consumption domestic investment, and economic growth in Ghana. Following recent literature, we make use of the World Banks' World Development Indicators (WDI) data on CO 2 emissions, measured in kilotons of oil equivalent, as a proxy for environmental degradation, and per capita GDP growth as a proxy for economic growth. Fossil fuel consumption as a percentage of total Ghana's total energy consumption was implored as a proxy for energy consumption, as was gross fixed capital formation as an indicator for domestic investments in Ghana. The study employs World Bank data from 1993 to 2014 due to data availability (Table 1).
The estimated equation(s) with its natural logarithm transformation of the variables is modeled in Eqs. (1) and (2) below; Moreover, the study adopts an optimal approach by incorporating domestic investments (gross fixed capital formation) into the model. This strategy serves two primary purposes in the Ghana contest. Firstly, the West African country has long been regarded as an investment hub and gateway in the sub-region (IMF 2016). Moreover, the recent ratification and subsequent commencement of the AfCFTA in 2021, headquartered in Ghana's capital, set the stage for robust industrialization through investment inflows and capital reinvestments. Since this industrialization, spearheaded by capital mobilization, comes with energy consumption (Kwakwa et al. 2022), it is imperative to capture its role in the environmental quality of Ghana. Secondly, past research that has adopted the ARDL model for integrating the emission-growth, growth-energy, and energy-emission nexus has justified its framework due to its ability to deal with potential variable omission biases in the bivariate models (Ang 2007). However, works like that of Shahbaz et al. (2013) have opined that the bivariate independent variables of energy consumption and growth effect on carbon emissions may be exaggerated due to potential, influential variable omission bias. Hence, the inclusion of domestic investment and FDI in our model seeks to also resolve the potential variable omission since domestic investment is a driving force of industrialization and energy consumption, which subsequently affects carbon emissions. (1) (2) lnCO 2 = α 1 + β 1 lnE t + β 2 lnGDP t + β 3 lnFDI t + β 4 lnK t + ε t where lnCO 2 is CO 2 emissions as a proxy for environmental degradation, lnE is the percentage of fossil energy in Ghana's total energy consumption, lnGDP is the per capita GDP growth (%) proxy for economic growth, and lnK is the gross fixed capital formation proxy for domestic investment. lnFDI is represented by the US dollar value of net inflows of foreign direct investments captured in the balance of payments.

Estimation technique
To analyze the long-run relationships between the series, we employ the ARDL bounds testing procedure of vector error correction (Narayan 2005). Recent works concerning the energy-growth-emission nexus have favored the ARDL model over other traditional linear models due to its several advantages. Adopting the ARDL model is meritorious due to its ability to produce reliable results regardless of the sample size (Ang 2007;Cheung and Lai 1993) or order of integration (Tong et al. 2020). Thus, while other estimation techniques require the series to be integrated of the same order, I(1), ARDL can still be applied irrespective of the order of integration of the series, either I(O) or I(1).
Moreover, the ARDL model is compatible with other cointegration techniques. For instance, by a simple linear transformation, a dynamic error correction model (ECM) (3) lnCO 2 = α 1 + β 1 lnE t + β 2 lnGDP t + β 3 lnFDI t + β 4 lnK t + ε t (4) can be computed from the ARDL bounds specification in the general equation below: where β, , δ, ϑ, and Φ, are the short-run parameters to be determined, whereas λ 1 , λ 2 , λ 3 , λ 4 , and λ 5 refer to the long-run relations. The existence of cointegration can be inferred by the rejection of the null hypothesis H 0 : λ 1 ≠ λ 2 ≠ λ 3 ≠ λ 4 ≠ λ 5 ≠ 0. The decision to accept or reject the null hypothesis is based on the F-statistics (Narayan 2005). Inferences are made from both the test's upper and lower critical bounds. The lower critical bound (LCB) value presupposes that all the series are I(0), while the upper critical bound assumes that all the series are I(1). If an F-statistic is greater than the UCB, then the series are cointegrated. However, there is no cointegration if the F-statistics is lower than the LCB. An inconclusive cointegration occurs when the F-statistics is greater than the lower critical bound in absolute value terms but less than the critical upper bound. In this case, inferences of the long-run relationship can be examined from the lagged error correction term of the model. To ascertain the model fit, robustness, and the possibility of variable omission bias, a sensitivity analysis is conducted to check any potential problems associated with the model. Since the direction of causality cannot be inferred from the cointegration equation (Ang 2007), we conduct the vec Granger causality to check the direction of causality within the variables. In implementing the Granger causality framework, we adopt Shahbaz et al.'s (2013) augmented Granger causality, which is formulated as a bivariate kth-order vector error-correction model (vecm) as follows; (5) Including the error correction term helps determine the longrun relationship among the series. The ECM depicts the error-correction term, whereas C i (i = 1…5) is the intercept. The error terms are represented in the equation by μ i (i = 1…5). The direction of long-run Granger causality can be captured from the vecm by inference from the t-statistics of the lagged ECMs. The Granger-Wald or F-statistics also captures the short-run causalities.

Results
This study adopts the ARDL bounds testing cointegration to analyze the emission-energy-growth nexus in Ghana between 1983 and 2014. This particular period was chosen based on data availability. The Pairwise correlation matrix in Table 2 exhibits significant relationships between the series. Economic growth (0.410) exerts the weakest correlation with carbon emissions, whiles domestic investment (0.925) and FDI (0.925) shows the strongest correlation with carbon emission. Also, the J'arque-Bera test statistics satisfy all the series' normal distribution requirements.

ADF and PP unit root results
Before utilizing the ARDL bounds testing technique for long-run relationships, we first test for stationarity dynamics of the series by adopting the Phillips and Perron's (1988) and the augmented Dickey-Fuller unit root test. Although the ARDL model is still applicable even if the series is not integrated of the same order, according to Pesaran et al. (2001), the F-statistics of the ARDL bounds test would be useless if any of the variables are found to be integrated in order two. Therefore, the unit root test ensures that none of the series is integrated beyond I(1). Thus, none of the series is integrated beyond its first difference. The ADF and the Phillips and Perron's (1988) unit tests are presented in Table 3. The outcome of the ADF test indicates that CO 2 and GDP are integrated at the order I(0) when both trend and intercept are considered. All other variables are integrated in the order I(1). The PP results also agree with the ADF results. Thus, apart from GDP and CO 2 with I(0), all other variables are integrated in order 1 when both intercept and trend are included. This affirms that the ARDL bounds test of cointegration can be applied to capture the long-run relationship since the series comprises both I(0) and I(1) variables.

ARDL bounds test of cointegration
In applying the ARDL bounds testing, an appropriate lag length of the series can be chosen by either the AIC or the SBC selection criterion. Following the observation of Lütkepohl (2006) and Shahbaz et al. (2013) that the AIC shows more efficiency and consistency in capturing the dynamic relations between the variables, we adopt the AIC selection criterion for the lag order selection, which is reported in Table 4. Recent studies have adopted the Narayan (2005) critical bounds instead of Pesaran et al. (2001) due to the former's suitability with both larger and fewer samples. Since our sample size is 32, we employ the critical bounds developed by Narayan (2005).
The null hypothesis of no long-run cointegration is indicated by the coefficient of an F-statistic that is less than the 5% upper critical bound. As depicted in Table 4, the F-statistics for the CO 2 equation of 6.29 are higher  Table 5 shows both the long-run and ECM short-run results. The coefficient of the adjusted R-square indicates that more than 66% and 96% of the changes in carbon emission in Ghana are explained by the selected variables in the short run and long run, respectively. In the long run, the previous year's emission may either contribute negatively or positively to the current year's emission. However, the negative value of the previous 2 years (− 0.28) showed significance at the 10% level according to the corresponding t-statistic. Although statistically insignificant, GDP and its first lag exert a positive relationship with CO 2 by 68% and 0.05%, respectively. On the other hand, energy consumption significantly increases carbon emissions by 44.8%. The results validate the findings of Abokyi et al.(2019Abokyi et al.( , 2021 and Lin and Agyeman (2019), who found evidence of energy-led-carbon emissions in Ghana, while partially contradicting that of Kwakwa (2021), who opined that energy consumption was insignificant to carbon emissions in Ghana. The results, however, indicate that the second lag coefficient of energy significantly reduces carbon emissions by 50%. Thus, past energy consumption significantly reduces carbon emissions in the second year. However, it shows no significant effect on emissions in the first year. In the long run, FDI reduces carbon emissions; however, the coefficients are statistically insignificant in both current (− 1.8%) and previous (− 4.8%) years. lnK significantly increased carbon emissions by 15.9% in a given year, whereas previous domestic investments significantly reduced pollution by 41%.

Long-and short-run results
In essence, capital investment can significantly reduce Ghana's carbon dioxide emissions in the long run. The short-run results from the error correction model in Table 5 also exhibit some remarkable findings. Economic growth significantly reduced carbon emissions by 1.3% in a given period. While current energy consumption may not significantly impact current carbon emissions, the previous year's energy consumption may significantly increase carbon dioxide emissions by 51%. Although FDI increases emissions by 1.7%, the t-statistics showed no significance; hence, FDI may be regarded as having no significant impact on carbon emissions in the short run. Domestic investment, however, exhibited a significant negative impact on carbon emissions, causing more than 41% emission reduction in the short run. In addition, the coefficient is statistically significant at the 1% level. The error correction term is expected to carry a negative sign to ascertain a long-run return to equilibrium in the shortrun shocks to the series (Appiah 2018). Therefore, the ECT t-1 term suggests that variations in the CO 2 emission model are perfectly corrected each year in the occurrence of shocks. Thus, the adjustment process to return to equilibrium is completed once every year in Ghana when it comes to carbon emissions. The CUSUM and CUSUMsq tests of stability in Figs. 4a, b, respectively, also show the reliability and consistency of our ARDL model, as none of the graphs exceed the 5% critical bounds level. Moreover, none of the p values in the diagnostic tests in Table 5 is below 0.05. In essence, the model does not suffer from any serial correlation or autocorrelation.

Granger causality test
Results from the Granger causality test are presented in Table 6. According to Shahbaz et al. (2013) and Salahuddin and Gow (2019), ascertaining the causal links among variables gives a better understanding of the future policy implications of any empirical findings. The null hypothesis of no causality is rejected when the p value of the Chi-square is less than 0.1. According to the corresponding p values of the relational Chi-square coefficients, a unidirectional causality from energy to carbon emissions was confirmed within the variables. Similarly, a unidirectional causality running from CO 2 to GDP was established. Thus, while energy consumption in Granger causes carbon dioxide emissions, these emissions, in turn, cause economic growth in Ghana. Also, a unidirectional causality exists from FDI to carbon emissions, while a similar relationship is recorded from CO 2 to domestic investments. In essence, while foreign investment inflows cause the emission of GHG gases in Ghana, these emissions, however, cause a rise in domestic investments. The results further reveal a unidirectional causality from energy consumption to economic growth. A unidirectional causality can also be witnessed from GDP to domestic investment, while similar causality is recorded from domestic investment to FDI.

Discussion
Results from both the short-and long-run estimations and the Granger causality tests reveal a corroborative implication. Thus, it can be observed that a unit increase in energy consumption may result in a 45% increase in CO 2 emissions in both the short run and long run, thereby validating the findings of Abokyi et al. (2021) in Ghana; Munir and Riaz (2020) in the USA, China, and Australia; and those of Ehigiamusoe (2020), who disclosed the negative repercussions of fossil energy in 25 African countries. According to the ARDL output, while economic growth only materially increases pollution in the long run, it reduces emissions by 1.3% in the short run. Although this outcome sharply contradicts Adeleye et al. (2021) who found a significant positive effect of GDP on CO 2 emissions in both the short run and long run, our outcome partially validates the EKC hypothesis. The Granger causality checks further reveal that, while energy exhibits a unidirectional causality on both economic growth and carbon emissions, this carbon emission leads to robust economic growth in Ghana. This outcome is in tandem with previous studies of Minlah and Zhang (2021), who found similar unidirectional causality between the same variables. In essence, Ghana's energy consumption significantly harms its environment, yet the country needs this energy and its concomitant effect on pollution in order to ensure economic growth. This further validates the position of Appiah (2018) that any energy conservation policy not derived from energy efficiency may significantly derail Ghana's economy.
The regression estimates further reveal that, while FDI insignificantly reduces CO 2 emissions in the long run, it exhibits a unidirectional causality on pollution. In contrast, domestic investment shows no direct causal effect on CO 2 emissions. Precisely, FDI only materially increases carbon emissions by 1.7%. This outcome corroborates the results of Salahuddin et al. (2018) and Sarkodie (2018), who also found a unidirectional causality from FDI to carbon emissions in Kuwait and Ghana, respectively. According to Sarkodie (2018), environmental protection regulations in the least developed countries are usually weaker, and pollutionoriented foreign investments target such countries as Ghana, thereby validating the pollution haven hypothesis. Moreover, no causality was established between FDI and economic growth, which contradicts some past studies (Adam 2009;Nguyen et al. 2020;Bouchoucha and Bakari 2021), which found either a negative or positive influence of FDI on a country's economic growth. On the other hand, domestic investments in capital accumulation reduce CO 2 emission by 41% in both the short run and long run. Although it initially increases carbon emissions in the long run, it returns to its equilibrium negative effect on pollution as time goes by. Moreover, both GDP and CO 2 Granger cause domestic investment in Ghana, which also causes FDI inflows. Albeit partially corroborating Choe (2003), this reveals a complex dynamic in the country regarding capital investment and economic growth. Kobayakawa (2019) posited that developing countries that invest more in carbon-intensive capital formation achieve faster economic growth. This may be true in Ghana because the country's energy mix, which is gradually shifting toward fossil fuels, is a driving force of its economic growth and determines its domestic investments in both magnitude and direction. According to Nishan and Ashiq (2020), countries facing such a daunting phenomenon may consider shifting from fossil fuels to renewable energy to maintain growth momentum and domestic investments (in the case of Ghana) while not damaging the environment.

Conclusions and policy implications
Energy is the fulcrum around which all global economies revolve. However, these massive energy consumptions that propel economic growth have been found to be detrimental to the environment due to the emission of greenhouse gases. Developing countries that seek to achieve faster and more robust economic growth are faced with the daunting task of improving their economies without compromising on their environmental safety. Several studies have incorporated foreign direct investment into these complexities as an alternative solution to offsetting environmental pollution while maintaining growth. It is against this background that we sort to analyze the interconnectedness and the dynamic role of domestic investment on energy consumption, economic growth, and carbon emission in the context of Ghana. The data employed in the study were collected from 1983 to 2014 from the World Bank's World Development Indicators (WDI). The ARDL bounds testing and the error correction models were utilized to ascertain the short-and long-run relationships, respectively. Since the direction of causality within the series cannot be inferred from the sign of the respective coefficients of the variables, we adopted the Granger-Wald test to analyze the direction of causality within the series. The cointegration analysis reveals a long-run relationship between carbon emissions, energy consumption, FDI, and domestic investments in the Ghanaian economy. The short-and long-run estimates and the causality test reveal a detrimental effect of Ghana's energy consumption on the country's environment. However, this same energy mix is a driving force for the country's economic growth, which in turn reduces carbon emissions by 1.3% in the short run. Carbon emissions propel economic growth, which in turn drives domestic investments. These domestic investments have been found to significantly inhibit environmental pollution in both the short run and long run by more than 41%. Although FDI materially reduces carbon emissions in the long run, it similarly worsens pollution in the short run while not directly impacting Ghana's economic growth.
Based on these findings and given the growing energy demand for fossil fuels in Ghana, the country would have to pursue a robust energy efficiency policy rather than energysaving options. Ghana currently gets more than half of its energy from fossil fuels to sustain its economic growth, which in turn attracts the capital investment needed for total economic liberation. With the enactment and commencement of the AfCFTA and the country's flagship one-districtone-factory (1D1F) initiative, policies targeting domestic capital mobilization should be prioritized to reap maximum benefits. Since economic growth has a significant influence on these investments, Ghana should look at diversifying the economy towards a more service-based one by channeling investment toward services such as education, health, tourism, ICT, and R&D industries, which are environmentally friendly. Therefore, the Government of Ghana should rigorously promote its tourism-oriented initiatives like the "year of return," which boosted tourism and remittances significantly in 2019. To ensure energy efficiency, public and private partnerships in the energy sector should be implemented to offset some of the negative environmental consequences of the current energy mix. The country has massive solar energy potential, therefore, investments in these sectors are an alternative way to increase energy consumption without compromising environmental safety. Also, domestic investment-oriented policies such as easy access to loans and government subsidies to domestic firms should be prioritized. Ghana's one-district-one-factory initiative, which seeks to establish a locally owned factory in every district, for example, may have the potential to help the country industrialize without harming the environment if stringent environmental regulations accompany it. To reduce the impact of foreign direct investment on environmental pollution, the government may consider introducing incentives such as selective taxation for domestic firms while increasing taxes on foreign firms, especially those in the carbon-intensive sector. For example, they give tax credits to domestic firms instead of foreign firms. However, these initiatives should be implemented cautiously.

Limitations of the study
Although this study contributes significantly to the ongoing literature and makes some profound recommendations for Ghanaian policymakers, it suffers from a few limitations. Firstly, the study only considered CO 2 the standard environmental pollution measure, while other indicators are still non-exhaustive. Moreover, the utilization of time series data implies that the empirical findings are constrained to being country-specific. The results of this study are, moreover, not expected to be indifferent across the spectrum of other econometric methods. Future studies may address these issues by utilizing panel data from some selected countries instead of country-specific time series data to ascertain these dynamics in, for example, SSA countries. Moreover, including other domestic investment, economic growth, CO 2 , and energy consumption indicators in the model would provide more robust findings. Lastly, the utilization of other estimation techniques may also offer variant outcomes with alternate recommendations. These prospects are left for future research.
Author contribution All the authors contributed to the manuscript. Bekoe Bernard Boamah: conceptualization, methodology, formal analysis, software, and writing-original draft preparation. Junfei Chen: supervision, conceptualization, validation, and funding acquisition. Odette Tasinda Odette Tougem: formal analysis and writing-reviewing and editing. Kulena Sakuwunda: reviewing. Nketiah Emmanuel: data curation, formal analysis, and writing-reviewing and editing.

Declarations
Ethics approval All relevant references have been appropriately cited in the text and the reference section.

Consent to publish
We confirmed that this manuscript has not been published elsewhere, either in part or in full, and is not under consideration by any other journal. All authors consent to the publication of this research work in this journal.
Consent to participate Not applicable.

Competing interests
The authors declare no competing interests.