Modelling electricity consumption in Ghana: the role of financial development indicators

1.   Introduction

The importance of electricity to the socio-economic development of emerging economies including Ghana cannot be over-emphasized (Alam et al., 2016; Kwakwa and Solomon, 2014; Onisanwa and Adaji, 2020). This is because, in the last few decades, electricity has become a mainstay for technological development. Without a dependable electricity supply, poor individuals in developing countries will not be able to make ends meet by engaging in income making activities. This will make it difficult for the poor to absolve themselves from poverty and unable to contribute their quota to the economic development of the economy (Kwakwa et al., 2021; Onisanwa and Adaji, 2020). Similarly, governments in countries with scant or low electricity supply cannot fulfill their obligation of providing basic amenities and development to their citizenry (Bhagat, 2011; Kumar, 2015; Meles, 2020; Williams and Porter, 2006). Equally, the quest to industrialize an economy calls for the provision of electricity (Abbasi et al., 2020; Kwakwa, 2012). It is therefore necessary for governments of emerging economies to ensure adequate supply of electricity. However, developing countries and their governments are faced with the problem of inadequate electricity supply for development due to the numerous challenges (such as insufficient investment). This leads to load shedding most of the time (Sichone et al., 2016; Zalengera et al., 2014). In short, it is not out of place to conclude that while the demand for electricity in emerging economies is very high, supply has been low (Ansu-Mensah and Bein, 2019; Kwakwa et al., 2021). With the exception of South Africa which has close to 100% of citizens having access to electricity, several countries in Sub-Saharan Africa (SSA) are yet to achieve electrification rate of 100%.

Ghana currently has an electrification rate of 84% and average electricity consumption has been increasing by 5% annually (World Bank, 2021; Adom, 2016). That notwithstanding, electricity supply constraints in Ghana had led to a recurring power crisis associated with load shedding and power outages (Nduhuura et al., 2021; Boakye et al., 2016). As part of the process of resolving Ghana's electricity problems it will be easier, financially, to manage the demand side than the supply side. With an unreliable power supply, failure to manage the consumption side of electricity in the country may worsen the situation (Burgess et al., 2020; Gore et al., 2018; Nduhuura et al., 2020; Quartey and Ametorwotia, 2017; Amoah et al., 2017). Another reason to manage the country's electricity consumption is that its usage is found to be a contributory factor of carbon dioxide emissions and climate change problems (Churchill et al., 2021; Osobajo et al., 2020; Shahbaz et al., 2013). Empirical studies on Ghana by Kwakwa (2021), Lin and Agyeman (2019), Kwakwa and Alhassan (2018), Asumadu-Sarkodie and Owusu (2017) and elsewhere by Craig et al. (2018), Jiang et al. (2021), Karmellos et al. (2021), Liddle and Sadorsky (2017), Liu et al. (2021) and Rehman et al. (2019) have found that increasing electricity usage increases carbon dioxide emissions. Available data from the World Bank (2021) suggests that a rise in Ghana's carbon emissions follows an upward trend with the country's electricity consumption (Kwakwa, 2021; Liu et al., 2021). In this regard, once electricity consumption is managed, there may be greater chances of reducing carbon dioxide emissions which will help the country attain Sustainable Development Goals seven (affordable and clean energy), twelve (responsible consumption and production) and thirteen (climate action).

Studies on the determinants of electricity consumption abound and a number of factors including price, income, urbanization, population and financial development have been identified to influence its consumption by a number of researchers (Ozcan and Ozturk, 2019; Peñalvo-López et al., 2019; Abokyi et al., 2018). However, the argument and the empirical evidence regarding financial development have been mixed. While it is argued by some that through financial development, firms and households get access to credit at a lower cost which enables them to acquire electricity dependent equipment thereby increasing electricity consumption, it is also argued that the access to cheap credit enables firms and households to buy energy efficient equipment thereby reducing electricity consumption (Mahalik and Mallick, 2014). The empirical works of Matar (2020), Kwakwa (2017), Rafindadi and Ozturk (2016), Sbia et al. (2017), Sekantsi et al. (2016), Shahbaz and Lean (2012) reported the former effect, while Asuamah (2016), Matar (2020), Muibi and Omoju (2016), and Solarin et al. (2021) reported the later effect. A possible reason for the mixed evidence is that different proxies have been used to measure financial development in the empirical studies. Thus, it appears each study has its own measurement for financial development. Such conflicting outcome does not help in policy formulation.

To really appreciate the effect of financial development on electricity consumption, it will be appropriate to estimate the effect of each of the indicators in a study to inform policy direction. Against this background, since Ghana's financial sector has seen some vibrancy over the past few years including recapitalization as argued by the Bank of Ghana, it is important to assess how this has influenced electricity consumption in the country. Therefore, the main purpose of this paper is to model electricity consumption in Ghana, an emerging economy, by investigating the effect of financial development on electricity consumption using various financial sector development indicators.

This study contributes to the literature in the following ways. It is among the few studies to examine the role financial development plays in Ghana's electricity consumption. Ghana's energy situation has been of concern to many researchers (Adusah-Poku et al., 2021; Kwakwa et al., 2018; Kwakwa, 2018; Adom, 2011; Adom and Bekoe, 2013; Adom, 2013; Aboagye, 2017; Dramani and Tewari, 2014; Mensah and Adu, 2015). However, empirical studies on the drivers of electricity consumption (Dramani and Tewari, 2014; Adom and Bekoe, 2013; Adom, 2013) have barely analyzed the effect of financial development. The study's contribution to the literature in general is also seen in the area that it uses more than one proxy for financial development. A financial development index is also constructed to assess the effect of financial development on electricity consumption. This is to ensure the results obtained are robust.

The remainder of this paper is structured as follows. The empirical modeling comes in the next section. The results segment and the analysis of the long and short run regression results follow in section three, while the conclusion and policy recommendations are highlighted in section four.

2.   Methods

2.1.   Empirical modeling

The theoretical foundation for the study is the theory of demand which indicates that demand for goods and services is influenced by a number of factors key among them are price, income and population pressure. Electricity is a service and its consumption or demand is derived demand since one does not have direct benefit from its consumption but rather through the gadgets that it helps to power. Economic theory argues that the relationship between price and electricity consumption is negative. An increase in the price of electricity will reduce the purchasing power of consumers and as a result, individuals not willing to pay for more will reduce the extent that they use their electrical appliance leading to a reduction in electricity consumption. As a normal commodity, income is posited to positively influence the level of electricity consumption since an increase in income makes it relatively easier for individuals to afford electrical appliances and pay for the charges associated with its usage (Adom, 2013). The number of population is also argued to have a positive effect on electrical consumption. The reason is that a rise in population means demand for food will increase. Consequently, firms and households will have to use more energy including electricity to meet their food demand. This same idea can be translated to mean that population pressure is likely to increase electricity consumption.

Since financial development has become a force to reckon with in the energy literature and the objective of the study is to assess its effect on electricity consumption in Ghana. It is mentioned in the literature that an increase in financial development may have a positive or negative effect on electricity consumption. Financial development, according to Mahalik and Mallick (2014) promotes, among other things, trade and industrialization which requires electricity thereby leading to higher electricity consumption. Contrarily, financial development makes it easier and cheaper for firms and households to access credit which enables them to acquire energy efficient equipment. The effect is that financial development will reduce electricity consumption (Mahalik and Mallick, 2014). Empirically, the effect has been mixed and this could be attributed to the various ways financial development is measured. Many studies use a single proxy including domestic credit to private sector, domestic credit to private sector by banks, financial development and stock market development (Matar, 2020; Rafindadi and Ozturk, 2016; Sekantsi et al., 2016). Where data is available, it is important to have a single study that will incorporate the various proxies of financial development for effective policy.

where EC is the level of electricity consumption, YPC represents income, PR is price of electricity, POP is population density (POPDEN) and FD is financial development. Also, L denotes the natural logarithm form of the variable. Again, α is the intercept, t represents time, ε is the error term and i denotes the various proxies for financial development. Since there is no one proxy for financial development, five different variables were used to ascertain the effect of financial development on electricity consumption. These five variables are mentioned at the data description below.

2.2.   Data source, data description and estimation technique

It is important to ensure that the variables used for the time series analysis do not contain unit root and thus, they must be stationary to avoid getting spurious results. As a result, the study first of all tests the stationarity property of the variables by using the Augmented-Dicky Fuller, the Phillips-Perron and the Zivot-Andrews tests. After ensuring the variables do not contain unit roots, the autoregressive distributed lag (ARDL) cointegration technique is employed to ascertain the long-run relationship between electricity consumption and the explanatory variables. The long and short run estimation from the autoregressive distributed lag (ARDL) cointegration technique is then estimated. The theoretical details of the ARDL model can be read from Pesaran et al. (2001). The dynamic OLS (DOLS) was also estimated to check for the robustness of the regression results. Once the long-run relationship between the dependent and the explanatory variables have been established, the Granger causality analysis is done to ascertain the direction of causality among the variables. For cointegrated variables as found in this study, the vector error correction model of the granger causality is employed. Details of this approach can be found in Lütkepohl (2005), Toda and Yamamoto (1995), Maddala and Kim (1998).

To get data to meet the objective of the paper, we relied on the World Bank's 2021 World Development Indicators (1971-2019). Price was measured by using consumer price index, income was measured as GDP per capita (constant US$), population pressure was represented by population density and electricity consumption was measured as Electric power consumption (kWh per capita). Following previous works such as Adu et al. (2013) and Omisakin (2009), financial development was represented by the following five indicators: Domestic credit to private sector (% of GDP) (FD-CP), Domestic credit to private sector by banks (% of GDP) (FD-BP), Broad money (% of GDP) (FD-BRM), Foreign direct investment net inflows (% of GDP) (FD-FDI) and FDINDEX, an index constructed by using principal component analysis method from the four indicators.

The descriptive statistics of the variables and the correlation among the variables are reported in Table 1 and Table 2 respectively. The correlation results indicate that there is no perfect correlation among the explanatory variables that may cause the problem of multicorrelation. The financial development index constructed from the Principal Component Analysis reported in Table 3, shows that foreign direct investment and Domestic credit to private sector (% of GDP) (FD-CP) combined explains 95% of the changes in the original data.

3.   Results and discussion

The results of the unit root test, the cointegration test, and the long and short run regression are presented and discussed below.

3.1.   Unit root and cointegration tests results

The unit root test results are presented in Table 4 and it is seen that the variables used for the study are stationary at different levels. While ADF unit test result shows that at levels only electricity consumption is stationary and the other variables become stationary at first difference, the PP test shows that price and electricity consumption are stationary at levels with the rest attaining stationarity at first difference. The Zivot-Andrews test results reported in Table 5 also shows that in the presence of structural break, price and electricity are stationary at levels while the remaining variables become stationary at first difference. Thus, it can be concluded from the results that electricity consumption and price are stationary at levels (Integrated on the order 0), while income, financial development and population density are stationary at first difference (Integrated on the order 1). This outcome implies that the variables are suitable for regression analysis.

Once the variables are seen to be a mixture of those integrated of order 0 and 1, the ARDL cointegration method is employed to determine the long run relationship among the variables. Since financial development was measured by five different proxies, five different tests were done with electricity consumption as the dependent variable while income, price, population density and the different financial development indicators become the explanatory variables. The reported results in Table 6 show that at a 1% level of significance, there is a confirmation of a cointegeration between electricity consumption and the explanatory variables irrespective of the financial indicator used. This goes to suggest there is a long run relationship between electricity consumption and income, price, population density and financial development. Previous studies like Abosedra et al. (2015) has found that electricity consumption is cointegrated with price, income, population density and financial development.

3.2.   Analysis of the long and short run regression results

From the ARDL cointegration, the long and short run estimates of parameters are generated and reported in tables 7 and 8 respectively. In Table 7 one notices that similar evidence is provided by all the models regarding the effects of income, price and population density. Thus, the results indicate that income and population density have positive effects on electricity consumption while price has negative effect. The positive effect of income suggests that a 1% rise in the per capita income level of Ghanaians is expected to increase the level of electricity consumption by about 2.7-5.0%. This confirms that indeed energy is a normal good that people would want more of it whenever they witness an increase in their income level. The reason is that electricity like many other energy is derived demand. They are indirectly demanded to meet certain needs. As income of people increases and they purchase more electronic appliances the demand for electricity also rises. This allusion is common among Ghanaians whose appetite for electricity gadgets like television and air conditioner among others increase with income. Previous studies in Ghana including Adom (2013) also obtained positive effect of income on electricity consumption. Elsewhere, Kwakwa (2018) reported similar outcome for Benin, Mukhtarov et al. (2020) for Kazakhstan, Ali et al. (2021) for China, and Bohlmann and Inglesi-Lotz (2021) for South Africa. In the short run, however, income is seen not to exert any significant effect on electricity consumption. This suggests that people do not immediately respond to consume electricity whenever there is an increase in their income level but rather will wait for a longer period of time.

In the long run, price is seen to reduce electricity consumption. This is also in line with economic theory. The law of demand establishes an inverse relationship between price and demand for good/service. The rise in price of goods and service makes things generally expensive for people to afford. It also erodes part of their purchasing power and as such quantity demanded for goods and services reduces. Although the price element used in this study is not the price of electricity but rather CPI, the outcome is generally a reflection of the case in Ghana. Whenever utility charges are increased, citizens are compelled to reduce their consumption. An increase in the general price level which also imposes some hardships on the citizens makes electricity consumption lower. Previous works including Xia and Hu (2012) and Al-Bajjali and Shamayleh (2018) reported of a negative effect of price on electricity consumption. The effect of price in the short run remains negative and significant. Thus, in the short run price reduces electricity consumption. In this case, people hurriedly adjust their consumption level whenever there are price changes.

A rise in a country's population increases the number of people who consume electricity and as a result, it is expected to increase the level of electricity consumption. This is likely to be heightened when there is an increase in population density. Ghana's population has increased from 8 million in 1971 to about 30.42 million at the moment (Ansu-Mensah et al., 2021). The population dynamics in Ghana has seen urban population exceeding rural population since 2010. The effect is that not only has the consuming population of electricity increased but also the demand for electricity in the urban centers has increased. The combined effect is that electricity consumption also increases. This could explain the positive effect of population density on electricity consumption in the long run. In their work, Huang et al., (2018) found that an increase in population pressure increases energy consumption leading to lower energy efficiency in China. The work of Kwakwa (2018) also had similar outcome. In the short run, population density generally increases electricity consumption. It is not surprising since the increase in population will definitely increase the demand for electricity consumption.

On the effect of financial development indicators, the results show that in the long run, domestic credit to private sector (LFD-CP), domestic credit to private sector by banks (LFD-BP) and broad money (LFD-BRM) reduces electricity consumption. Also, while the overall financial development index constructed reduces electricity consumption, foreign direct investment increases electricity consumption. The results tell that domestic credit given to the private sector helps reduce electricity consumption. Access to credit is very essential for firms to expand their production via the acquisition of raw materials, repair or maintenance of their machines and acquisition of new machines. When it is difficult to access credit, firms are compelled to use obsolete equipment for production and this consumes more electricity. The opposite is the case when firms easily access credit. By increasing the credit given to the private sector in Ghana, firms are placed in a position to afford energy efficient equipment for their operations which make the sector more efficient thereby reducing electricity consumption. Similarly, credit by banks to the private sector also helps achieve same outcome. This suggests that over the years it appears banks in the country have also been concerned with the promotion of efficiency among firms in the private sector. Regarding the effect of broad money supply, it also implies that as money supply increases, citizens and firms easily access funds to afford efficient energy gadgets, hence, the reduction effect of broad money supply on electricity consumption. The above findings are in line with Matar (2020), Rafindadi and Ozturk (2016), Sbia et al. (2017), Sekantsi et al. (2016), Shahbaz and Lean (2012). In the short run, however, the effects of the financial indicators discussed above on electricity consumption is generally insignificant which means an increase in the level of the financial sector does not have immediate effect on electricity consumption.

The effect of foreign direct investment as an indicator of financial development is however observed to increase electricity consumption. Electricity is a critical factor that attracts foreign investors. However, there is the argument that foreign direct investment may through the pollution haven effect increases energy consumption in developing countries while through the pollution halo effect it reduces energy consumption. There has been conscious effort by authorities in Ghana to attract foreign investors. However, the finding suggests that the presence of such foreign firms in the country increases electricity consumption in the long run implying that they may not be energy efficient. In the long run, a 1% increase in foreign direct investment is associated with about 0.12% increase in electricity consumption which is similar to the work of Amoako and Insaidoo (2021).

Overall, the financial development index shows that Ghana's financial development in the long run helps to reduce electricity consumption. In the short run, it is rather seen that the overall financial development increases electricity consumption. This gives the impression that the positive effect of the financial development on electricity consumption is short-lived and that in the long run it becomes negative to the benefit of the country.

To ensure that the regression results are robust, the DOLS regression analysis was also employed. The regression results reported in Table 9 confirm what was reported from the ARDL. Thus, it also endorses that income and population density positively affects electricity consumption. Also, the results show that foreign direct investment increases electricity consumption while the other proxies for financial development have the opposite effect confirming what has been reported from the ARDL technique in Table 7.

3.3.   Diagnostic tests for regression results

The error correction term ECM (-1) for each model presented in Table 8 is negative and statistically significant. This is an indication that there is a long run relationship between electricity consumption and the explanatory variables in this study. Also, the results in Table 10 demonstrate that generally the models are free from the problems of instability, non-normality, heteroskedasticty and serial correction. It therefore means that the estimated results are robust. Similarly, the CUSUM test and CUSUM of squares shown in figure 1 and figure 2 respectively confirm that the models are stable.

Finally, a vector error correction granger causality is analyzed among the variables (electricity consumption, income, population density and financial development index, representing financial development) to ascertain their direction of causality (Table 11). In the short run electricity consumption is granger caused by income, price and population density. In the long run income, price, financial development and population density are also found to granger cause electricity consumption. The above gives weight to previous results discussed above. Also, a unidirectional causality from price, financial development and population density to income is observed in the short run. Financial development is granger caused by population density in the short run and long run; while a bidirectional causality runs between population density and price in the short run. The unidirectional causality from income to electricity consumption is similar to Kwakwa (2012). The above finding is an indication that policies to influence electricity consumption may have a feedback effect on population density without affecting the other variables.

4.   Conclusions and policy recommendations

This paper modelled electricity consumption in Ghana and assessed the effect of financial development on it. The study used the time series data from 1971 to 2019 to model electricity consumption in Ghana as a function of income, price, population density and financial development. Financial development was measured using five proxies namely domestic credit to private sector, domestic credit to private sector by banks, broad money supply, foreign direct investment net inflows and an index constructed from the four indicators. The ARDL cointegration method was employed to determine the long run relationship among the variables. The results indicated that there is a long run relationship between electricity consumption and income, price, population density and financial development indicators. The regression analysis showed that income and population density increase electricity consumption while price reduces its consumption. All the financial development indicators were found to reduce electricity consumption except foreign direct investment which had the opposite effect. A VECM approach to granger causality test confirms income, price and population density granger cause electricity consumption in the short run while in the long run income, price, financial development and population density granger cause electricity consumption. In addition, a bidirectional causality runs between population density and price in the short run.

The policy implications arising from the study include the need for authorities in the country especially the Bank of Ghana to impress upon financial institutions to further reduce the interest rate on credit taken from them. In addition, studies may be carried out to unearth why the lending rate from financial institutions in the country remain high. Although the Bank of Ghana's policy rate has gradually reduced from about 26% in 2016 to 13.5% presently, the lending rate from commercial banks and other financial institutions has seen little reduction (hovering around 20-23%), a situation which has received complaints from many business managers. Ability by the Central Bank to persuade financial institutions to reduce their lending rate or when the factors behind the interest rate spread are identified and dealt with, it may enable more firms and households access funds to acquire energy efficient equipment. Since foreign direct investment is found to increase electricity consumption, authorities in the country may need to strictly regulate the activities of foreign firms or persuade foreign firms to ensure they are energy efficient in their operations. As the population pressure grows, and the economy expands and income levels of citizens' increase, there will be need for intensive education to the growing population on the need to conserve electricity in their bid to consume more.

Conflict of interest

All authors declare no conflicts of interest in this paper.