Empirical analysis of renewable and non-renewable energy resources consumption impact on economic development in Uzbekistan

. In this paper the association among renewable, non-renewable energy consumption and GDP growth is analyzed by using the secondary date of The World Bank and International Energy Agency for the period 1990-2021 of Uzbekistan. The paper employs the Autoregressive Distributed Lags (ARDL) model to estimate the long-run and short-run dynamic multipliers of energy consumption variables. Empirical results show that hydropower energy consumption (renewable energy) has a positive effect on GDP growth in the long term. Also, consumption of non-renewable energy resources (coal, natural gas, oil) has a positive effect on GDP growth in the short and long term. In addition, the increase in the consumption of non-renewable energy resources has a positive effect on CO2 emissions, which in turn means that the government should take measures to increase the share of renewable energy resources.


Introduction
As the economy of the world is developing the demand for energy resources is increasing and Uzbekistan is also no exception in this condition.The country's GDP is growing around 5-6 % for the last years and it is estimated 80.4 billion US dollars in 2022 and the GDP deflator index was 113.8% compared to 2021 prices.According to the Statistics Agency of Uzbekistan GDP per capita in 2022 at current prices was 2,254.9US dollars.It is obvious that any kind of economical growth requires the use of energy resources.Investigating the country's economy it was found out that the share of industry in the GDP was 33.4% and over the years this sector is leading the main consumer of total energy with 40% of total in 2019 in the following places population use 23%, Agriculture 20%, Utility 13%, Transport 3% and Construction only 1%.To boost its energy sector, Uzbekistan has been pursuing extensive reforms recently.High levels of equipment wear and tear, a slow pace of infrastructure updates, improper equipment operations, poor installations, and both gas pipelines and electrical lines that have reached the end of their useful lives are all related with problems.Uzbekistan is one of the leading producers of natural gas in the world, producing more than 60 billion cubic meters (bcm) of gas annually, of which 35 to 40 bcm are supplied by the joint-stock company (JSC) "Uzbekneftegaz" In 2019, production totaled 60. 4

billion cubic meters (bcm).
Although the country has a sizable potential for solar energy, no large-scale solar power plants exist there.Furthermore, there are no industrial-scale wind farms because the potential of wind has not been sufficiently researched.However, Uzbekistan is taking action to create a legal framework for the expansion of this energy sector.The Law on the Use of Renewable Energy Sources, the Law on Public-Private Partnerships, and the Regulations for Connecting Businesses that Produce Electricity, Including from Renewable Energy Sources, to the Unified Electric Power System have all been passed.For the last years the government is focusing on energy consumption and improving it.By Presidential Decree No. PD-5646 of 1 February 2019 on Measures to Radically Improve the Management System of the Fuel and Energy Industry of the Republic of Uzbekistan, the Ministry of Energy was founded.It is responsible for overseeing the production, processing, distribution, sale, and use of coal, electric and thermal energy, as well as the production, transmission, and consumption of oil, gas, and their byproducts.Uzbekistan enacted Law No. ZRU-539 on the Use of Renewable Energy Sources on May 21, 2019.It specifies benefits and preferences for RES use, such as exemptions from paying fees.Additionally, the Republic of Uzbekistan's President issued a decree on February 16, 2023, number PD-57, titled "On Measures to Accelerate Implementation of Renewable Energy Sources and Energy Saving Technologies in 2023," which is regarded as a major source of reforms.

Literature review
Energy is one of the most significant sources as the world economy grows.Ummalla et al. examined the relationship between hydroelectric energy consumption and economic growth over the period 1965-2016 using the ARDL bounds testing approach to cointegration and found that renewable energy had a positive impact.Additionally, the Granger causality test demonstrates a unidirectional causal relationship between the consumption of hydroelectric energy and economic growth [1].In another study by Ching-Chi Hsu et al it is stated that high energy production is the global requirement that is the demand of high economic growth of the country and requires regulators and analyzed the impact of economic factors such as gross domestic product (GDP), national income, employment rate, foreign direct investment (FDI), and inflation and technological advancement on energy production in China with 1976 to 2020 secondary date with NARDL model and according to the results all the economic factors such as GDP, national income, employment rate, FDI, infation, and technological advancement have a significant and positive association with energy production in China [2].Mallesh et al. conducted research on the two largest rising market economies in the world, Between 1965 and 2016, China and India examined the effects of natural gas and renewable energy usage on economic growth using a multivariate framework.The results showed that natural gas consumption causes economic growth in China, but no causality is confirmed in India in the short run.The autoregressive distributed lag bounds testing approach to cointegration and vector error correction model (VECM) is used to explore the long-run and causal nexus among the consumption of natural gas, renewable energy, coal and petroleum, CO2 emissions, and economic growth, respectively [3].Massoud et al carried out research in the case of Vietnam with the secondary data have been extracted from 1983 to 2020 using World Development Indicators (WDI) database by NARDL model and the results revealed that GDP, exports, human capital, employment rate, international tourism receipts, and expenditures have a significant and positive relationship with REP in Vietnam [4].To determine the direct and indirect energy expenditures associated with the extraction and conversion of net energy at the point of use in China, Ke Zhao et al. utilized the term "energy expenditure".Using data ranging from 1987 to 2015, a multivariate linear regression model was used to determine the nation's maximum allowable level of energy consumption.According to the findings, China's economic system requires 3217 mtce net energy to maintain average annual GDP growth above 5% until the 2030s, which means that energy consumption cannot exceed 45.44% in 2030.[5].Charifa Haouraji et al maintained that the more economic growth, the more energy consumption forms [6].
Correlative models based on a link between REC and GDP and urbanization rate outperform MLR models in their research in the case of Morocco, where great economic growth occurred for the last years with secondary dates from 1990 to 2013, and according to the results, providing more reliable and accurate results in terms of prediction errors during the test period.Forecasting show that Moroccan residential energy needs will increase by 70%.It means that energy planning should be carried out in a comprehensive and precise manner.Energy consumption, CO2 emissions and GDP relationship was studied by Sulaiman et al in Nigeria condition [7].According to the results energy consumption shows significant negative impact on GDP in the short run and in this case renewable source of energy such as solar and wind could be explored and considered as an alternative source of energy since the country is well endowed with solar energy.Consequently, assist will appear in reducing CO2 emissions and at the same time sustaining long run growth in GDP.The correlation between renewable energy and GDP per capita also has been studied in several researches.Amarachi W. Konyeaso et al tested this relationship in 32 selected countries of African region with the period from 1996 to 2018 by categorizing them oil-rich and non-oil-rich as well as income levels and employed Pooled Mean Group, Augmented Mean Group, and Dynamic OLS model.It was found out that there is a significant positive renewable energyeconomic growth relationship in all the different groups [8].Another study by Muhammad et al. contributes to our understanding of the relationship between Pakistan's economic development and its energy supply.The study used linear and nonlinear ARDL models to examine the symmetry or asymmetry of Pakistan's GDP per capita's relationship to its consumption of conventional and renewable energy from 1980 to 2016 and vice versa.The study's findings lead to the conclusion that renewable energy has long-term, asymmetric effects on Pakistan's economic growth [9].Similarly, the asymmetric effects of GDP, in the long run, are confirmed in both energy models.In some studies results shows that there mainly two factors that can impact the use of renewable energy directly.First one is economic well-being that increases renewable energy consumption and second one is both property rights and tax burden decline the share of renewables in total energy consumption [10] Because data on established nations was easier to come by than on developing and impoverished nations, developed nations were the focus of the majority of the early studies on this subject.Now that it is possible to test almost all nations in the world, a paper by Selim et al. examined the factors that influenced the consumption of renewable energy in Africa from 1990 to 2013 and discovered that countries with higher GDP per capita and the Human Development Index have lower shares of renewable energy in their national grids, while an increase in foreign direct investment has been linked to higher renewable energy integration [11].
There is a high demand for energy due to the world's population growth, industrialization, technological advancements, rising standards of living, and consumer spending.Since fossil fuels are less costly, traditional fossil fuels (NREN resources) are predominantly preferred in energy production to meet the increasing demand.G7 countries accounted for almost half of the global GDP.In addition, these countries have consumed approximately one-third of the World's energy production.G7 economies are also one of the communities with the largest share in renewable energy production [12].Okumus et al. examined the G7 countries in this situation, and the results of the CS-ARDL estimation indicate that REN and NREN usage are both favorably correlated with both long-term and short-term economic growth.However, it is discovered that NREN use has a stronger and statistically more significant impact on economic growth when the coefficients of these two variables are compared [13].Maha Harbaoui Zrelli studied the relationship between various forms of energy consumption and economic growth in a few Mediterranean nations and discovered that a 1% increase in renewable electricity consumption lengthens economic growth by 0.032% while a 1% increase in non-renewable electricity consumption accelerates it by 0.155% [14].
Examining the connection between the energy-growth nexus and export is important because exports and export quality growths are important factors that contribute to a country's GDP expansion.These findings were made by Faik et al.Increased usage of fossil fuels is a direct result of GDP growth [15].Increases in real GDP have a positive and statistically significant impact on the use of renewable and non-renewable energy (and vice versa) only over the long run, according to Luliana Matei's examination of the GDP impact on energy consumption in OECD nations.Depending on whether the energy is produced using renewable or non-renewable resources, the immediate effects change.In order to make renewable energy sources competitive with fossil fuels, the paper recommends financing research and development as a crucial component of promising renewable technologies and related infrastructure networks.It also encourages regional cooperation and the development of clean energy efficiency between nations [16].

Data
We used data from 1990-2021 to quantify the relationship between annual GDP growth rate and consumption of non-renewable and renewable energy resources.GDP annual growth rate was used as a dependent variable, measured in percentage, non-renewable (gas, oil, coal) and renewable (hydropower) energy consumption was used as an independent variable, measured in terawatt-hours (TWh).The data are on an annual basis and are obtained from World Bank Open data (https://data.worldbank.org) and International Energy Agency (https://www.iea.org/).Figures 1, 2, 3, 4 and 5 show the values of these indicators in a graph by year.Table 1 describes these indicators.From Figures 2 and 3, it can be seen that the consumption of gas energy has a tendency to increase, while the consumption of oil energy has a tendency to decrease.Also, according to Figure 4, we can observe that the consumption of coal energy decreased significantly in the first 13 years, and then the trend increased again.Table 2 presents the statistical description of the variables.According to it, the average gas energy consumption in the Republic of Uzbekistan during 1990-2021 is 439.14 (TWh), oil energy consumption is 70.99 (TWh), coal energy consumption is 13.86 (TWh), and hydropower energy consumption is 17.83 (TWh).established In addition, during this period, the average annual growth rate of the country's GDP was equal to 4.25%.Also, in the table, the values of the asymmetry and kurtosis coefficient and the Jarque-Bera criterion, which determine whether the variables extend to the normal distribution or not, are given.Based on these values, it is possible to estimate that the variables are close to the normal distribution.Source: Authors' estimations

Model and methodology
To analyze the impact on the dependent variable, the analysis uses a linear model.The model yields: where _ represents annual percentage growth rate of GDP, while t represents the time period from 1990 to 2021. 0 denotes the constant, while  1 ,  2 ,  3 ,  4 are the coefficients of gas, oil, coal and hydropower energy consumption variables, respectively;   is the error term.
The ARDL model is used in the analysis [17,38].In order to determine which one should be employed, the Bound test [18] to be run given the existence of the short-and long-run model equations.Standard F-and t-statistics form the foundation of the Bound test.Regardless of whether the regressors are I(0) or I(1), the asymptotic distributions of these statistics are nonstandard under the null hypothesis that there is no level link.There are two sets of asymptotic critical values offered: one for just I(1) regressors and the other for exclusively I(0) regressors.The range of classifications of the regressors into purely I(0), solely I(1), or mutually cointegrated is covered by these two sets of critical values.The ARDL's null hypothesis where ∆ represents the first difference factor,  0 denotes intercept,  1 ,  2 ,  3 ,  4 ,  5 -shortrun dynamic multipliers,  -lag length, i -lag order. 1 ,  2 ,  3 ,  4 ,  5 -long-run dynamic multipliers. is the error term.Equation ( 2) denotes a linear ARDL model, which provides both short-and long-run estimates.Since the variables in Figures 1, 2, 3, 4, and 5 have different trends, the Ljung-Box test [19] is performed to determine whether there is autocorrelation in the data.Then, the the Augmented Dickey-Fuller (ADF) test [20], the Phillips-Perron (PP) test [21], and the Kwiatkowski-Phillips-Schmidt-Shin (KPSS) [22] unit root test are both applied to check the stationarity of the variables and their integrated order.Then, considering the presence of ARDL short-and long-run model equations, the Bound test should be run to specify which one must be used.Then, the study applies certain diagnostic and robustness tests for the developed ARDL model.

Results
We check the variables selected above as voluntary and non-voluntary variables by autocorrelation, i.e. "Ljung-Box" (Q) test.It is known that the hypothesis  0 (no , 050 (2024) BIO Web of Conferences MSNBAS2023 https://doi.org/10.1051/bioconf/2024820500202 82 autocorrelation in the time series) and the alternative hypothesis  1 (the presence of autocorrelation in the time series) of the "Ljung-Box" test are considered as the main hypotheses.It is known that in this test,  < 0.05 means that the  1 hypothesis is accepted, and  > 0.05 means that the  0 hypothesis is accepted.Source: Authors' estimations.
The presence of autocorrelation in time series requires testing these time series for stationarity.Therefore, in the next step, the variables are tested for stationarity tests.Table 4 below presents the results of testing variables for the Augmented Dickey-Fuller (ADF), the Phillips-Perron (PP) test and the Kwiatkowski-Phillips-Schmidt-Shin (KPSS) unit root test.According to the results, GDP growth, gas, oil and coal energy consumption variables are stationary at first difference, while hydropower energy consumption variable is stationary at both levels.It can be concluded that the integrated order of the variables is (1).

Discussion
For the ADF test, we report the p-values.Lags have been automatic selected using SIC and deterministic components based on their statistical significance.We remind that the ADF test's null hypothesis is the presence of a unit root.For the PP test, the p-values are reported which were obtained with a specification adopting an automatic selection of the bandwidth and including trend and intercept.For the KPSS, the test statistics are reported which were obtained with a specification adopting an automatic selection of the bandwidth and including trend and intercept.In the PP test, the null hypothesis is the presence of a unit root, while the null hypothesis in the KPSS test is the series stationarity.The null hypothesis is rejected for PP test when p-value is less than 0.05.The null hypothesis is accepted for KPSS test when p-value is higher than 0.05.
For the main empirical analysis, the ARDL bound testing method is used.This method has a number of advantages over other estimators.First, it can be applied when the variables are stationary or non-stationarity or both.Second, it is effective in the presence of small sample sizes, although the number of observations must be less than the time dimension.Third, it simultaneously estimates short-and long-run models.
To identify whether the ARDL model should be developed in the short or long run, the results of the ARDL bound test (Table 5) are checked.For the ARDL bound test, the alternative hypothesis is that there exists long-run relationship.The alternative hypothesis is accepted when the F-statistic is higher than lower and upper bound given the level of significance.The alternative hypothesis is that there exists long-run relationship.The alternative hypothesis is accepted when F-statistic is higher than lower and upper bound at the all significance level.Given the evidence that there is a longrun and short-run relationship among the variables as indicated in Table 5, the ARDL longand short-run model can be developed.The estimates are presented in Table 6.It can be seen from Table 6 that all coefficients representing long and short-term effects are statistically significant.Therefore, all coefficients can be interpreted.The assessment's findings indicate that, over the long term, an increase in gas energy consumption of 1 TWh will result in a 0.06 percent GDP growth increase, an increase in oil energy consumption of 1 TWh will result in a 0.14 percent GDP growth increase, and an increase in coal and hydropower consumption of 1 TWh will result in increases in GDP growth of 0.56 and 0.42 percent, respectively.Additionally, the results of the short-term impact show that an increase in gas energy consumption of 1 (TWh), an increase in oil energy consumption of 1 (TWh), and an increase in coal energy consumption of 0.08% all result in an increase in GDP growth of 0.03%, 0.08%, and 0.08%, respectively.An increase in electricity consumption of 1 (TWh) results in a 0.25 percent increase in GDP.It is justified by the ECM coefficient (-0.98)According to Table 6 the RESET (regression equation specification error test) coefficient (0.81) gives evidence that the model is adequate and there is no specification error.The Breusch-Godfrey Serial Correlation LM test coefficient (0.64) suggests that there is no auto correlation in the estimated ARDL.   6) and CUSUM of squares (Fig. 7) tests show that the cumulative sums of the standardized deviations (blue line) do not exceed a specified range (red lines).In conclusion, the estimated ARDL model passes all stability tests and it does not suffer from any misspecification error.

Conclusion
The link between economic development indicators and energy resource usage has been the subject of numerous empirical investigations.This study differs from others in that it took Uzbekistan's level of economic growth into account while analyzing the short-and long-term effects of using renewable and non-renewable energy resources.In the long term, coal energy consumption (0.56) and hydropower consumption (0.42) have a higher impact on the country's GDP growth than other energy resources.In the short term, gas, coal and oil energy consumption will be affected.It is worth noting that the positive impact of hydropower consumption on the country's GDP growth in the short term is very small.This situation means that in the short term, it is necessary to develop consistent measures by the government to further increase the efficiency of renewable energy resources.It is known that increasing the positive effect in the short term a multiplicative effect in the long term.About 8.0 percent of the country's renewable energy production is provided by hydropower energy.In addition, hydropower energy accounts for 0.9 percent of the total supply of renewable energy sources.The contribution of wind and solar energy is negligible .
According to the available public data, the potential of renewable energy resources in Uzbekistan is high.The high annual wind speed and sunlight that exist in the country are spread over huge areas of several thousand square kilometers that are not used for agriculture and other economic activities.In these areas, the combined production of wind and solar energy can produce several TWh of electricity.This will ensure the green energy potential, which is much higher than the current and future needs of the country, as well as of regional and international importance.

Fig. 7 .
Fig. 7. CUSUM of squares test for estimated ARDL model Next, CUSUM tests are carried out to test the stability of the model.Both CUSUM (Fig.6) and CUSUM of squares (Fig.7) tests show that the cumulative sums of the standardized deviations (blue line) do not exceed a specified range (red lines).In conclusion, the estimated ARDL model passes all stability tests and it does not suffer from any misspecification error.

Table 2 .
Statistical description of variables

Table 3 .
Serial correlation analysisFirst, we need to focus on the correlogram of the variables.Table3below shows the correlogram of variables, autocorrelation coefficients and their p-value.It can be seen from the table that there is autocorrelation in the data representing the volume of GDP growth, gas, oil and coal energy consumption, and there is no autocorrelation in the data of the volume of hydropower energy consumption.

Table 4 .
The results of unit root tests

Table 5 .
The results of ARDL bound test

Table 6 .
ARDL (1, 2, 1, 2, 0) model estimation results Authors' estimations.For RESET and Breusch-Godfrey Serial Correlation LM tests, p-values are reported.The null hypothesis in the RESET test is that there is no equation specification error, while for the Breusch-Godfrey Serial Correlation LM test, the null hypothesis is that there is no serial correlation.