The Effect of Financial Development on Energy Consumption: Evidence from Russia

 This paper explores the effect of financial development, economic growth, and energy prices represented by consumer price index (CPI) on energy consumption in Russia by performing VECM, CCR, DOLS and FMOLS analyses to the annual data from 1995 to 2019. The findings of this empirical analysis reveal that financial development and economic growth have positive impact on energy consumption in Russia. Furthermore, the effects of energy prices expressed by CPI is revealed to be negative, which is consistent with the theory and expectations in practice. Based on the findings of this study, the nexus and impacts of financial development on energy consumption are discussed, as well as plausible explanations and policy implications.


INTRODUCTION
Energy is very important for the continuation of vital activities. Increase in the human population and the widespread use of hitech products to meet basic needs such as education, health and housing require more energy consumption (EnCon thereafter). Thus, EnCon plays a vital role in shaping the economic policies of countries. Being either an energy-rich or an energy-poor country has a key role in shaping these policies. The primary goal of energy-poor countries is to provide the energy they need in a way that provides the maximum benefit. Energy-rich countries, however, have priorities such as providing a competitive advantage among themselves and having impact on the world energy market. In this regard, EnCon remains important for both groups of countries. The fact that energy is so critical makes determining the factors affecting EnCon a priority. financial development (FinDev thereafter) is a key factor affecting EnCon.
The financial sector is comprised of a set of institutions, instruments, and markets, as well as legal and regulatory frameworks, that allows transactions by lending. Basically, development in financial sector considers lowering the "costs" that the financial system is exposed to. This search, aimed at minimizing the transaction costs of obtaining information, implementing obligations, and trading, has yielded in the widespread of financial contracts, markets, and intermediaries (World Bank, 2021). The faster the level of FinDev of countries, the easier it becomes to meet the costs to be incurred as the results of increase in production and consumption. In order to meet the costs, a sustainable financial system must be found in developed and developing countries (Keskingöz and İnançlı, 2016).
The most important advantage of a sustainable and efficient financial system is that it ensures the continuity and stability of economic growth (Schumpeter, 1911;McKinnon, 1973). The growth hypothesis states that an increase in EnCon causes an increase in real GDP, given that the economy is dependent on energy (Apergis and Payne, 2009). Sadorsky (2010;2011), andMukhtarov et al. (2020b) mentioned three effects of FinDev that lead to an increase in EnCon. An initial effect is that with higher FinDev, consumers can easily borrow funds to purchase air conditioners, automobiles, etc., leading to an increase in EnCon. In the other hand, in a well-functioning financial system, businesses can easily access financing to invest their capital and current assets in their business. The third effect appears in the form of wealth growth. It is known that more wealth can stimulate economic activity and increase EnCon.
Considering the above facts, it is important to analyse the nexus between FinDev and energy use in Russia, the 11 th largest economy in the world according to an IMF report. Clearly Russia is a country with a very high potential of energy resources (Amelia, 2008;Bayraç, 2009;Mitrova and Melnikov, 2019). It produces the highest amount of crude oil, as well as condensate, and it is the second largest producer of dry natural gas. Russia also produces an important amount of coal. With these resources, it can be said that Russia has a significant influence on the global market (EIA, 2021;Kutcherov et al., 2020;Su et al., 2020). Considering that Russia produces just 3% of world GDP and has just 2% of the world population, it is impressive that it is the world's thirdbiggest producer and consumer of energy resources, after China and the US. With energy production of nearly 1470 mtoe, Russia exports a major portion of the primary energy it produces while supplying 16% of the global cross-regional energy trade, which makes Russia the world leader in energy exports (ERI RAS, 2019). Further, Russia's energy density is 2.19 times more than the world average density, and it exceeds the comparable index for European Union (EU) countries by 3.08 times. In 2007-2018 the Russian GDP grew by 14%, while it reduced EnCon by 12% (ACGRF, 2020). In Russia, several financial methods were developed to implement activities in the field of energy efficiency, though the main restriction regarding the implementation of an energy performance policy is the lack of present long-term borrowed funds. Accordingly, the most important function for the state is to create additional financing methods while maintaining the existing ones (Matraeva et al. 2019). Therefore, it is of great importance to research the extent and the long-term consequences of the relationship between the level of FinDev and the level of EnCon.
As far as we know, no research has been conducted on the nexus between EnCon and FinDev in Russia, employing different time-series methods, namely VECM, CCR, DOLS, and FMOLS that allow for the examination of country-specific aspects of this association. As a result, the goal of this research is filling this gap by investigating the influence of FinDev on EnCon for Russia, one of the largest crude oil producing economies, a unique case study for this study. The outputs of the research have significant implications for policymakers to formulate appropriate policies for sustainable energy use. Moreover, the outputs of this paper are also important for post-Soviet and developing oil-rich countries. The research's results are crucial for politicians as it will enable them to create appropriate policies in courtesy of sustainable EnCon. Furthermore, the findings of this paper are significant for other post-Soviet countries as well as developing oil-rich countries.
The remaineder of the study is laid out as following: Section 2 summarizes literature review. Section 3. Contains the method and data. Section 4 contains the empirical findings and the last, Section 5, contains the conclusion and suggestion.

LITERATURE REVIEW
A significance amount of empirical research and publications are focused on the impact of FinDev on EnCon in different countries with various funding models. Sadorsky (2010) studied the impact of FinDev on EnCon in 22 emerging economies using annual data for the period of 1990-2006. Using ARDL, VECM and Panel GMM models, he reported positive impact. Zhang et al. (2011) examined how the Chinese stock market influenced EnCon via the Granger causality test using 1992-2009 annual data, reporting a positive nexus. Many other studies such as Coban and Topcu (2013), Al-Mulali and Lee (2013), Islam et al. (2013), Chang (2014), Komal and Abbas (2015) and Mukhtarov et al. (2018) found positive effects of FinDev on EnCon in various countries. Also, Gaies et al. (2019) found the same result for MENA countries using the GMM model with 1996-2014 annual data.
However, Ali et al. (2015) reported a negative effect of FinDev on EnCon in Nigeria employing the ARDL model from 1972Q1 to 2011Q4.
Applying ARDL, Johansen cointegration, VECM and Granger causality test to annual data for the timeframe 1971-2008, Shahbaz and Lean (2012) reported a bidirectional causality between FinDev and EnCon in Tunisia. Kahouli (2017) and Bekhet et al. (2017) reported long run cointegration between FinDev and EnCon using ARDL models for the South Mediterranean and Gulf Cooperation Council countries respectively.
Since financial series exhibit nonlinear structures Aliyev (2019), we look at some studies that focus on nonlinear dynamics. Mahalik et al. (2017) reported a non-linear inverted U-shaped association between FinDev and EnCon for the period of 1971-2011 using cointegration test and ARDL model in Saudi Arabia.
In recent studies, Danish et al. (2018), Khan et al. (2019), andMukhtarov et al. (2020a) reported positive effect of FinDev on EnCon employing the DSUR, 3SLS, GMM, and VECM methods. Karacan et al. (2021) studied the impact of carbon dioxide emissions, income, and oil prices on renewable EnCon in Russia employing the Canonical Cointegrating Regression method and the VECM method for the period of 1990-2015. They found a negative relationship between oil prices and renewable EnCon, and a positive relationship between real GDP per capita and renewable EnCon. Table 1 summarizes this literature review. As seen from this review, the impact of FinDev on EnCon in Russia has not been intensively studied using VECM, CCR, DOLS and FMOLS techniques for a wider timespan.

Equational Specification and Data
Following Mukhtarov et al. (2018) and Mukhtarov et al. (2020a), the functional specifications in this paper are described as below: Growth (EG) is proxied by real GDP per capita (2010 US dollars). The consumer price index (CPI) (2010=100) is used to measure the energy prices. Because data on energy prices for all countries and all years is not readily available, energy prices are proxied by the consumer price index, as in previous studies by Sadorsky (2010;2011), Komal and Abbas (2015), Chang (2015), Mukhtarov et al. (2018), Mukhtarov (2020a), and Mukhtarov (2020b). The data for EnCon was compiled from Enerdata (Enerdata, 2021) while FinDev was obtained from the database of Federal Reserve Bank of St. Louis (FRED, 2021). The EG and CPI data were provided Where, EnCon t represents energy consumption, FinDev t is financial development, EG t denotes economic growth, CPI t denotes consumer price index as measure of energy prices, and ε t is an error term.
We utilized 1995-2019 annual data for the EnCon, FinDev, economic growth and energy prices. The dependent variable is EnCon, and is expressed by kilogram of oil equivalent. Our key independent variable is FinDev, which is expressed by domestic loans to the private sector as a percentage of GDP. Economic from the World Bank database (WB, 2021). We used logarithmic expressions of all variables for empirical estimation.

Methodology
We evaluated the effect of FinDev, economic growth, and energy prices expressed by CPI on EnCon using the VECM, CCR, DOLS, and FMOLS techniques. In the beginning step, the Augmented Dickey Fuller (ADF) unit root test is employed to define non-stationarity characteristics of variables under study.
As the next step, since the orders of integration of the variables are the same, Johansen cointegration test is used to define if they are cointegrated.
Ultimately, we applied the Vector Error Correction Model (VECM) to assess the long-term relationship between the variables. The VECM method is the first-best choice if there is just one cointegration link between the variables under study. In order to achieve more reliable findings, we also used the Canonical Cointegrating Regression (CCR), Fully Modified Ordinary Least Squares (FMOLS), and Dynamic Ordinary Least Squares (DOLS) tests.
To conserve space and avoid confusing readers with econometric complexities, we do not discuss the above-mentioned approaches here. In addition, these methods are extensively used and wellknown. Dickey and Fuller (1981), Johansen (1988), Phillips and Hansen (1990), Johansen and Juselius (1990), Hansen (1992a;b), Park (1992), and Stock and Watson (1993), have all published research that provide extensive information.

EMPIRICAL FINDINGS AND DISCUSSION
ADF unit root test verifies the stationarity characteristics of the variables and the findings of ADF are summarized in Table 2. As seen from the test results all variables are non-stationary at their level, though they are stationary at their first difference. Thus, the cointegration test may be applied.
To identify the optimal lag interval on the sample, a Vector Auto Regressive (VAR) model with the endogenous variables of EnCon, FinDev, EG, and CPI was initially specified through a random-selected lag interval, then defining test of lag interval was employed to the model residuals. Table 3 shows the results of the analysis. In this study, three lag selection criteria indicated that a lag of order two is optimum, which is naturally suitable regarding the less observations in the sample.
It's worth noting that, the VAR model with two lags successfully passes all residual diagnostic tests, as well as the stability test, as exhibited on Panels A-D of Table 4.
Panels E and F of the Having verified the cointegration among the variables, VECM, CCR, DOLS, and FMOLS techniques are utilized to assess coefficients of the long-run link among the variables. If there is just one cointegration link between the variables, the VECM technique is the first-best option. Furthermore, the VECM residuals were explored in diagnostic testing. Table 5 shows the outputs of the VECM, CCR, DOLS, and FMOLS techniques. Table 5 shows that VECM residuals carry no problems about serial correlation, instability or heteroscedasticity. As a consequence, the assessed specifications' residuals fulfill the requirements of residual diagnostics tests, confirming the estimation findings' robustness.
As seen from the outputs, the long-run coefficients of all approaches are statistically significant and remarkably near in significance value and sign. The outputs of the VECM model, which are exhibited on the top row of Table 5, are prioritized, as stated in the methodology. According to the VECM findings, FinDev has a statistically significant positive impact on EnCon. According to the findings, a 1% rise in FinDev caused a 0.02% rise in EnCon. This means that when FinDev increases, so does the demand for energy. Our findings are consistent with several studies conducted by Sadorsky (2011) Furthermore, we discovered that EG has a positive and statistically significant influence on EnCon. And this means that a 1% increase in EG corresponds to a 0.45% increase in EnCon. The results we obtained are in line with the traditional expectation. Furthermore, according to the findings, energy prices  Notes: At 10%, 5%, and 1% significance levels, accordingly, *, **, and *** imply null hypothesis rejection represented with CPI has a negative and statistically significant influence on EnCon, which is reflected in economic theory.

CONCLUDING REMARKS
This paper examines the influence of FinDev, economic growth, and energy prices denoted with CPI on EnCon in Russian Federation. ADF unit root test results show all variables have the same integration order, which is I(1). Therefore, the cointegration link between the variables may then be tested. Long-run comovement was tested using the Johansen cointegration test. The VECM, DOLS, CCR and FMOLS techniques were employed to estimate possible long-run relationships.
The empirical results stated that FinDev and economic growth exhibit positive effect on EnCon, whereas the energy prices expressed by CPI has a negative impact on EnCon in Russia. The positive influence of economic growth on EnCon shows that Russia uses its expanding revenues to increase energy sources.
The positive impact of FinDev, as measured by bank loans to the private sector as GDP, indicates that the Russian financial system leads to a reduction in both material and transaction costs in debt markets. This enables households and companies to find "easy" money. In this way, economic units that earn more income will be able to buy the goods and services they need.
Since an increase in FinDev results in higher EnCon and that FinDev is a favourable economic outcome, a policy recommendation from our findings is that the Russian government should exploit alternative energy resources such as hydropower, wind and biomass. Considering the positive impact of economic growth on EnCon, transition from fossil fuels to renewable energy resources is important for sustainable economic growth within Russia. Moreover, our findings suggest to policymakers as well as to researchers the need to envision the relationships between FinDev, economic growth, and EnCon for sustainable development goals and federal macroeconomic stability in Russia and similar oil-rich countries.  EnCon t shows dependent variable; ***, **, and * represent significance levels of 1%, 5%, and 10%, accordingly; P values are in brackets; LM SC represents Lagrange multiplier statistic for serial correlation test; ᵡ 2 HETE represents Chi-squared statistic for heteroscedasticity; JB N test represents Jarque-Bera statistic for normality test