Researches have been conducted to study the link between natural gas consumption (NGC) and economic growth in the contexts for a group of countries or a specific country. Many studies have used panel data while others have used time series data for different time periods. According to the economic model used, studies can be classified into four models. The first model is a bivariate model, the second model is multivariate based on the production function to include capital and labor, the third model is based on expanding the production function to include the indicator of trade and financial development, while the fourth model uses natural gas with several other classified categories of energy such as coal and oil Electricity, etc. Moreover, researchers used different econometric techniques by applying Granger causality, VECM, ARDL, and nonlinear ARDL techniques to examine the cointegration and causality relationship between the variables in both the short and the long run. The results of empirical studies on the relationship between NGC and economic growth also presented four hypotheses describing this relationship. Conservation Hypothesis, Feedback Hypothesis, growth hypothesis; and Neutrality Hypothesis (Ozturk, 2010).
The growth hypothesis states that energy consumption plays an important role in economic growth either directly or indirectly in the production process as a complement to labor and capital. Therefore, energy is considered a factor of production and thus economic growth, therefore shocks to the energy supply will have a negative impact on economic growth. literatures provided empirical evidence that NGC plays an important role in stimulating economic growth, as natural gas is an essential source of energy and a major determinant of economic growth. the results supported the growth hypothesis that reflects a unidirectional causal relationship from NGC to economic growth, which is known in the literatures the energy consumption-led growth hypothesis. Therefore, any attempt to adopt the Conservation policy to reduce gas utilization would be detrimental to such an economy. Rather, it is to improve the level of natural gas consumption in order to enhance economic growth. Among the studies that confirmed the growth hypothesis: Omisakin, and Olusegun, (2008) for Nigeria. Shaari, et al. (2013) for Malaysia, as well as Akhmat and Zaman (2013), for Bangladesh, Bhutan, India and Maldives. Farhani, et al (2014) for Tunisia. Lach, (2015) for Poland. ( Furuoka, 2016; Zhi-Guo, et al. 2018). in the case of China. Destek and Okumus, (2017) for Italy, Japan, UK, and USA. Solarin and Ozturk, (2016) for Iraq, Kuwait, Libya, Nigeria, and Saudi Arabia. (Muhammad, et al. 2011; Ali, et al. 2019; Sohail, et al. 2021) for Pakistan. Rahman, et al. (2020) for China. Foye and Benjamin, (2021) for the selected sub-Saharan African countries. Furthermore, other empirical literatures have demonstrated the positive relationship between NGC and economic growth, i.e., supporting the growth hypothesis, including: Shahbaz, et al. (2013). Hassan, et al. (2018) for Pakistan. Solarin and Shahbaz, (2015) for Malaysia. Galadima and Aminu (2018) for Nigeria. Aydin (2018) for the top 10 natural gas-consuming countries from 1994 to 2015. Wu, et al (2021) for China. Awodumi and Adewuyi (2020) for the consumption of both petroleum and natural gas in Africa’s largest oil-producing economies. Etokakpan, et al. (2020) found that a 1% increase in natural gas consumption leads to a 0.02% increase in production in the Malaysian economy.
The feedback hypothesis states that energy consumption and economic growth are jointly determined and affected by each other at the same time. That is, energy consumption promotes economic growth, and at the same time, economic growth stimulates energy consumption. empirical literatures have provided evidence of bidirectional causality between natural gas consumption and economic growth and supported the feedback hypothesis, including: Yang, (2000) for Taiwan. Kum, et al. (2012). for France, Germany, and United States. Lim and Yoo, (2012) for Korea. Heidari, et al. (2013) for Iran. Bildirici and Bakirtas (2014) for Brazil, Russia, and Turkey. As well, (Erdal, et al. 2008; Dogan, 2015) for Turkey. Furuoka, (2016) in the case of Japan. Ummalla and Samal (2019) for both China and India. Sinaga (2019) for Indonesia. Magazzino, et al. (2021) for Germany and Japan. Likewise, Apergis and Payne, (2010) supported the feedback hypothesis for a panel of 67 countries within the period 1992–2005, in both the short- and long-run. as well as Solarin and Ozturk (2016), for a panel of 12 OPEC countries over the period (1980–2012).
The conservation hypothesis holds if an increase in GDP leads to an increase in energy consumption. Therefore, energy conservation policy is implemented with little or no impact on economic growth. Some studies have supported the conservation hypothesis and the existence of a unidirectional causal relationship from economic growth to NGC, where economic growth drives the NGC. Such as (Ghosh & Basu, 2006; Behera, 2015) for India. Payne, (2011) for USA 1949–2006. Akhmat and Zaman (2013) for Nepal, Sri Lanka, and Pakistan. Das, et al. (2013) for Bangladesh. Alshehry and Belloumi (2014) for Saudi Arabia. Destek, and Okumus (2017) for Germany. Solarin and Ozturk (2016). for Algeria, Iran, United Arab Emirates and Venezuela.
The neutrality hypothesis means that neither conservative nor expansionary policies regarding energy consumption have any effect on economic growth. some studies have confirmed the neutrality causality relationship between natural gas consumption and economic growth in some countries. Chang, et al. (2016) for the panel of G7 countries 1965–2011), excluding the case of the UK. Likewise, (Güvenek, et al. 2017; Erdoğan, et al. 2019) for Turkey. As well as, Arora, (2016) for USA. Sharaf (2016) for Egypt. Destek and Okumus (2017) for Canada and France. Solarin and Ozturk (2016) for Angola and Qatar. Mallick (2009) concluded that no evidence that the various components of energy, including NGC, significantly affect the components of economic growth in India.
The impact of natural gas consumption on economic growth in the short or long run varies according to country, the empirical studies have proven this difference in the results obtained, like Işik, (2010) showed that natural gas consumption positively affects economic growth in the short run and negatively in the long run in Turkey )1977–2008(. Lash (2015) also concluded, based on quarterly data (2000–2009), that NGC caused short-run GDP growth in Poland. While the long-run causality was in the opposite direction. In the same context, Makala and Zongmin (2019) indicated that there is no long-run relationship between NGC and economic growth in Tanzania. Conversely, Destek (2016) revealed that NGC positively affects economic growth in the long-run in 26 Organization for Economic Co-operation and Development (OECD) countries (1991–2013). Dolgopolova, et al. (2014) found that there are long-run relationships between real GDP, labor force, real capital, oil consumption, electricity consumption, gas consumption, and coal consumption for 7 non-OPEC members. Likewise, Fadiran, et al. (2019) confirmed the long-run, rather than short-run, impact of NGC on economic growth in 12 European countries.
Some empirical literatures have relied on growth models to examine the relationship between natural gas consumption and economic growth by expanding the production function to include, in addition to labor and capital, other variables, such as energy consumption. Shahbaz, et al. (2014) concluded that the impact of NGC on economic growth is greater than other factor inputs suggesting that energy is a critical driver of production and growth in Pakistan. also, Dogan, (2015) discovered that for Turkey economy the coefficient estimation of the NGC became smaller in the long run and turned negative in the short run after the exhaustion of capital and labor. Farhani, et al (2014) added capital. Solarin, & Shahbaz, (2015) included foreign direct investment, capital and trade openness. Rafindadi and Ozturk, (2015) investigated the nexus between NGC, exports, capital, labor and economic growth in Malaysia, the results revealed that NGC has an indirect effect on the Malaysian economic growth. Destek (2016) revealed that natural gas consumption, GDP growth, gross fixed capital formation, and trade openness are cointegrated. Farhani, and Rahman. (2019) turned out to be that natural gas consumption, exports, capital and labor are the contributing factors to France's economic growth. Balitskiy, et al. (2016) amended the neoclassical growth model to include capital and labor as explanatory variables the results showed that the increased economic output in the European Union member states leads to increased NGC. Additional NGC requires more investment in infrastructure that would allow natural gas to be processed. However, the increase in natural gas consumption also leads to a decline in economic development. On the other hand, Li, et al. (2019) proven that the higher the level of economic development, the greater impact of natural gas consumption on economic growth.
Most previous studies examined the long- and short-run linear relationship between NGC and economic growth. some researchers measure the nonlinear correlation. Hu, and Lin, (2008) emphasized the long-run nonlinear equilibrium relationship between GDP and energy consumption in Taiwan, when energy consumption exceeds a certain threshold level in the energy-inefficient periods in which energy consumption grows faster than GDP. also, Galadima, and Aminu. (2018). estimated the value of the NGC threshold in Nigeria, and found that the level of consumption is less than the optimal level. The same authors Galadima and Aminu, (2019) examined the asymmetric effects of NGC on economic growth, the increase in NGC can stimulate long-run growth and the negative impact is minimal. Therefore, energy conservation policies do not lead to a decline in economic growth. In 2020, the same researchers did another study on Nigeria, Galadima, and Aminu (2020) concluded that NGC and economic growth follow a non-linear process. The positive change of NGC is consistent with the “feedback hypothesis” and the negative effect of NGC is consistent with the conservation hypothesis. Sohail, et al. (2021) found evidence that positive changes in NGC and financial development boost Pakistan's economic growth.
In the Saudi context. Despite the importance of natural gas consumption and production in the Saudi economy, the empirical evidence on the relationship between natural gas consumption and economic growth in Saudi Arabia is still less explored and limited, except for some studies that dealt with the Saudi issue with the OPEC group or the Gulf Cooperation Council, or in the context of carbon dioxide emissions Alshehry, and Belloumi, (2014) used the multivariate cointegration approach to study the causal relationships between fossil fuels consumption, CO2 emissions, and economic activity. The results supported the conservation hypothesis. The study suggested that energy conservation policies might be enforced without affecting economic growth. While in another study on 12 OPEC countries over the period (1980–2012), by Solarin, and Ozturk, (2016). showed evidence for the feedback hypothesis in a panel OPEC country and the growth hypothesis in Saudi Arabia. Ozturk, and Al-Mulali, (2015) added trade openness, labor force, and gross fixed capital formation as determinants of GDP growth in the Gulf Cooperation Council countries 1980–2012. The results supported the feedback hypothesis and the positive effects of the NGC on the GCC country’s economic growth in the long run. Akadiri, et al. (2019) Examineded the contribution of NGC and trade to the real GDP of Saudi Arabia over the period 1968–2016, using the ARDL method, found a long-run co-integration relationship between NGC, trade, and real GDP. The study also confirmed the Growth hypothesis. The study suggested that the policy of preserving natural gas will harm the demand for natural gas, impede total trade and thus delay domestic output.
Economic model specification and data source
Energy economists assert that energy is a necessary factor in the production process since it acquires all the characteristics of a factor of production, the result is that production is determined by energy, stock capital, and labor. Robert Solow in 1956 introduced a simplified model that relied on the production Cobb-Douglas function to develop a framework for the causes of growth (Solow, 1956). Then, in 1957, noting that the rate of growth occurs due to a set of growth rates in other factors of production beyond physical and human capital, which is technical progress (Solow, 1956). according to Solow model, the production function takes the following form:
The variable T for time appears in the function to allow for technological change, The usage of energy determines technological change, empirical works and theoretical concepts assume that energy can be taken as a part of technology (Balitskiy, 2016) with the following aggregate production function:
This study is Based on the neoclassical economic growth theory, following: (Furuoka, 2016; Hassan, et al. 2018; Luqman, et al. 2019; Fadiran, et al. 2019; Farhani, & Rahman, 2019; Li, et al. 2019; Awodumi & Adewuyi, 2020; Foye & Benjamin, 2021). Energy is an input in the production process, since the economy is driven by an increase in energy demand, The expanded production function according to which output growth is a function of capital stock, labor and energy is an input in the production process, and since the economy is driven by an increase in energy consumption, (Lee & Chang, 2008; Pirlogea & Cicea, 2012 ). Then modified production function.
Rewritten the function using the Cobb Douglas production function form as:
$$\:GDP=A{K}^{{\gamma\:}_{1}}{L}^{{\gamma\:}_{2}}{E}^{{\gamma\:}_{3}}{e}^{\mu\:}$$
Then extended the model with control variables. Flowing (Furuoka, 2016; Farhani, et al. 2014; Farhani & Rahman, 2019; Foye & Benjamin, 2021). Added trade openness TO, measured by aggregates of imports and exports. Following, Sohail, et al. (2021) adding bank credit (BC) to the private sector in the economic model to represents the financial sector development. Bank credit is the main source of investment financing and thus affects aggregate demand, which ultimately boosts economic growth (Alzyadat & Alwahibi, 2021). Therefore, the extended model is expressed as:
$$\:GDP=(K,L,\:NGC,TO,\:BC)$$
Where GDP represents the real gross domestic product and measures for economic growth, NGC is natural gas consumption in billion cubic feet, K denotes capital and it is measured by gross fixed capital formation, L represents labor, measured by the number of labor force. TO trade openness, (BC) bank credit to the private sector. The economic model has been converted to logarithmic form to facilitate the estimation and interpretation of the regression coefficients
$$\:LinGD{P}_{t}={\gamma\:}_{0}+{\gamma\:}_{1}LinK+{\gamma\:}_{2}LinL+{\gamma\:}_{3}LinNGC+{\gamma\:}_{4}LinTO+{\gamma\:}_{5}LinBC+{\epsilon\:}_{t}$$
To assess the linear short and long-run relationships between the NGC and economic growth, this study like: (Furuoka, 2016; Farhani, et al. 2014; Farhani, & Rahman, 2019; Foye & Benjamin, 2021: among others). applies the Autoregression Distributed lags approach (ARDL) suggested by Pesaran and Shin (1998) and Pesaran et al. (2001). ARDL is a linear time series models where both the dependent variable and the explanatory variables are related not only concurrently, but also with their lagging values. Hence, the general form of the symmetry and structural break effects in short-run and long run as proposed by Pesaran and Shin (1998) is written as:
$$\:\varDelta\:{Y}_{t}=\:{\delta\:}_{0i}+\:{\delta\:}_{1}{Y}_{t-i}+{\delta\:}_{2}{X}_{t-i}+\:\sum\:_{i=1}^{q}{\alpha\:}_{1}\varDelta\:{y}_{t-i}+\sum\:_{i=1}^{k}{\propto\:}_{2}\varDelta\:{X}_{t-i}+{\epsilon\:}_{it}$$
Where Xs are the explanatory variables, and Y is the dependent variable, `q and k are the numbers of maximum lag order in the ARDL model. The maximum lag lengths of q and k for the dependent and explanatory variables, respectively. The ARDL bound test approach is to redefine the economic model as error correction model:
$$\:\varDelta\:LinGD{P}_{t}={\gamma\:}_{0}+{\gamma\:}_{1}LinGD{P}_{t-1}+{\gamma\:}_{2}Lin{K}_{t-1}+{\gamma\:}_{3}Lin{L}_{t-1}+{\gamma\:}_{4}LinNG{C}_{t-1}+{\gamma\:}_{5}LinT{O}_{t-1}\:+{\gamma\:}_{6}LinB{C}_{t-1}+{\sum\:}_{i=1}^{p}{\delta\:}_{1}\varDelta\:LinGD{P}_{t-i}+{\sum\:}_{i=0}^{p}{\delta\:}_{2}\varLambda\:Lin{K}_{t-i}+\sum\:{\delta\:}_{2}\varDelta\:Lin{L}_{t-i}+{\sum\:}_{i=0}^{p}{\delta\:}_{3}\varDelta\:LinNG{C}_{t-i}+{\sum\:}_{i=0}^{p}{\delta\:}_{5}\varDelta\:LinT{O}_{t-i}+{\sum\:}_{i=0}^{p}{\delta\:}_{6}\varDelta\:LinB{C}_{t-i}+{\epsilon\:}_{t}$$
Where (δ1 – δ6) represent the coefficients of short run dynamics relationships of the underlying variables in the model. 𝜀 is the speed of short run adjustment of the model’s convergence to long run equilibrium, the error correction term (ECT). The short-run coefficients can then be derived from the following corresponding error correction model:
$$\:\varDelta\:LinGD{P}_{t}={\sum\:}_{i=1}^{p}{\delta\:}_{1}\varDelta\:LinGD{P}_{t-i}+{\sum\:}_{i=0}^{p}{\delta\:}_{2}\varLambda\:Lin{K}_{t-i}+\sum\:{\delta\:}_{3}\varDelta\:Lin{L}_{t-i}+{\sum\:}_{i=0}^{p}{\delta\:}_{4}\varDelta\:LinNG{C}_{t-i}+{\sum\:}_{i=0}^{p}{\delta\:}_{5}\varDelta\:LinT{O}_{t-i}+{\sum\:}_{i=0}^{p}{\delta\:}_{6}\varDelta\:LinB{C}_{t-i}+{\delta\:}_{7}EC+{\epsilon\:}_{t}$$
The null and alternative hypotheses are as follows: \(\:{H}_{0}:\:{\delta\:}_{1}={\delta\:}_{2}={\delta\:}_{3}={\delta\:}_{4}={\delta\:}_{5}={\delta\:}_{6}=0\:\:\:\:\text{n}\text{o}\:\text{s}\text{h}\text{o}\text{r}\text{t}\:\:\text{r}\text{u}\text{n}\:\text{r}\text{e}\text{l}\text{a}\text{t}\text{i}\text{o}\text{n}\text{s}\text{h}\text{i}\text{p}\:\text{e}\text{x}\text{i}\text{s}\text{t}\text{s}\). Against the alternative hypothesis \(\:{H}_{A}:\:{\delta\:}_{1}\ne\:{\delta\:}_{2}\ne\:{\delta\:}_{3}\ne\:{\delta\:}_{4}\ne\:{\delta\:}_{5}\ne\:{\delta\:}_{6}\ne\:0\:\:\text{t}\text{h}\text{e}\:\text{s}\text{h}\text{o}\text{r}\text{t}\:\text{r}\text{u}\text{n}\:\text{r}\text{e}\text{l}\text{a}\text{t}\text{i}\text{o}\text{n}\text{s}\text{h}\text{i}\text{p}\:\text{e}\text{x}\text{i}\text{s}\text{t}\text{s}.\:\:\)Where (\(\:\gamma\:\)1 – \(\:\gamma\:\)6) represent the coefficients of the long-run relationships, p is the lag order of the variables are defined as before. The bounds test for the absence of any level relationships between the dependent and independent variables is through the exclusion of the lagged level variables. That is, it involves the following null and alternative hypotheses
\(\:LinGD{P}_{t}={\gamma\:}_{0}+{\gamma\:}_{1}LinGD{P}_{t-1}+{\gamma\:}_{2}Lin{K}_{t-1}+{\gamma\:}_{3}Lin{L}_{t-1}+{\gamma\:}_{4}LinNG{C}_{t-1}+{\gamma\:}_{5}LinT{O}_{t-1}+{\gamma\:}_{6}LinB{C}_{t-1}+{\epsilon\:}_{t}\) T
The null and alternative hypotheses are as follows: \(\:{H}_{0}:\:{\gamma\:}_{1}={\gamma\:}_{2}={\gamma\:}_{3}={\gamma\:}_{4}={\gamma\:}_{5}={\gamma\:}_{6}=0\:\:\:\:\text{n}\text{o}\:\text{l}\text{o}\text{n}\text{g}\:\text{r}\text{u}\text{n}\:\text{r}\text{e}\text{l}\text{a}\text{t}\text{i}\text{o}\text{n}\text{s}\text{h}\text{i}\text{p}\:\text{e}\text{x}\text{i}\text{s}\text{t}\text{s}\). Against the alternative hypothesis \(\:{H}_{A}:{\gamma\:}_{1}\ne\:{\gamma\:}_{2}\ne\:{\gamma\:}_{3}\ne\:{\gamma\:}_{4}\ne\:{\gamma\:}_{5}{\:\:\ne\:\gamma\:}_{6}\ne\:0\:\:\text{t}\text{h}\text{e}\:\text{l}\text{o}\text{n}\text{g}\:\text{r}\text{u}\text{n}\:\text{r}\text{e}\text{l}\text{a}\text{t}\text{i}\text{o}\text{n}\text{s}\text{h}\text{i}\text{p}\:\text{e}\text{x}\text{i}\text{s}\text{t}\text{s}.\:\:\)
The study employs annual secondary data from different sources including Saudi Central Bank and World Development indicator (WDI). The dependent variable is economic growth measured by Gross Domestic Product (GDP). While the independent variables are NGC, gross-fixed capital formation, labor force, trade openness, and financial development proxy with the bank credit to private sector. The data cover a period from 1990 to 2021. The study adopts ARDL approach for data analysis.