Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

The Effect of the COVID-19 Outbreak on the Turkish Diesel Consumption Volatility Dynamics

The Effect of the COVID-19 Outbreak on the Turkish Diesel Consumption Volatility Dynamics Ertuğrul, H. M., Güngör, B. O., & Soytaş, U. (2020). The Effect of the COVID-19 Outbreak on the Turkish Diesel Consumption Volatility Dynamics. Energy RESEARCH LETTERS, 1(3). https://doi.org/10.46557/001c.17496 COVID-19 and Energy The Eff The Effec ect of the C t of the CO OVID-19 Outbreak on the T VID-19 Outbreak on the Turkish Diesel urkish Diesel C Consump onsumption V tion Volatilit olatility Dynamics y Dynamics a b c 1 2 3 H. Murat Ertuğrul , B. Oray Güngör , Uğur Soytaş 1 2 3 Ministry of Treasury and Finance, Turkey, Energy Market Regulatory Authority, Turkey, Department of Business Administration, Middle East Technical University, Turkey Keywords: covid-19 pandemic, arch family models, arima models, diesel consumption https://doi.org/10.46557/001c.17496 Energy RESEARCH LETTERS Vol. 1, Issue 3, 2020 We analyze the effect of the COVID-19 outbreak on volatility dynamics of the Turkish diesel market. We observe that a high volatility pattern starts around mid-April, 2020 and reaches its peak on 24/05/2020. This is due to the government imposed weekend curfews and bans on intercity travels. Two policy suggestions are provided. First is a temporary rearrangement of profit margins of dealers and liquid fuel distributors; and, second is a temporary tax regulation to compensate lost tax revenue. distributed lag based co-integration model in order to fore- 1. Introduc 1. Introduction tion cast gasoline and diesel demand for India. In a similar fash- ion, but with a simpler model, Bhutto et al. (2017) forecast- Energy is a necessity for societies and its security and ac- th ed annual gasoline consumption for Pakistan. Given this lit- cessibility is a priority for policy makers. Since the 19 cen- erature, one research gap is that there are no studies that tury, one of the main sources of energy has been oil and examine how energy consumption volatility changes in the oil products, and the importance of energy has grown with face of a global pandemic. This study aims to fill this gap time . Hence, movements in oil prices have been an im- by studying the Turkish diesel market during the recent portant subject of research. All major oil price shocks, such COVID-19 pandemic. In addition, we also show how the as the rise in oil prices of the 1970s and 2000s and the oil pandemic distorts the forecasting performance of models. price glut in the 1980s, have attracted research interests The fluctuations in oil prices started at the beginning with the literature attempting to explain fluctuations (and of 2020 with the spread of COVID-19. Volatility in energy hence volatility) in energy prices (Kocherlakota, 2009). markets increased. We add to the understanding of energy Energy consumption and production forecasts were markets by examining the effect of the COVID-19 pandemic made with special attention to oil crises and oil gluts over on diesel consumption. Our paper is the first to investigate the last 50 years. Many econometric applications were con- volatility dynamics of the diesel consumption in light of the ducted to analyze the demand dynamics of various oil prod- global pandemic. ucts and to forecast consumption. For example, for Chinese The purpose of this paper is to examine the effects of the electricity consumption forecasts, Zhu et al. (2011) pro- COVID-19 pandemic on the Turkish diesel market and as- posed an integration of the moving average procedure and sess the performance of simple forecasting tools. To achieve seasonal autoregressive integrated moving average model this aim, we employ variants of the generalized ARCH (SARIMA) with weight coefficients. J. Li et al. (2018), on (GARCH) models. These models allow us to investigate the the other hand, considered more than 20 combination mod- disruptive effects of the pandemic on the Turkish diesel els using traditional combination methods to forecast oil market. We find that volatility starts rising with the an- consumption in China. Rashid and Kocaaslan (2013) exam- nouncement of the first case and peaks at the end of May ined the relationship between energy consumption volatili- 2020. Even though purchases increase in the 2 days follow- ty and unpredictable variations in real GDP by estimating a ing the announcement, they then assume a steady down- Markov switching autoregressive conditional heteroskedas- ward trend. ticity (ARCH) model for the UK. In addition, Z. Li et al. The paper is organized as follows: Data and methodology (2010) provided an analysis of future gasoline demand in are presented in section 2 and results are presented in sec- Australia by using various models. They found that more tion 3. Section 4 concludes. sophisticated models do not always produce better forecast- ing results compared to simple models. They advised to de- 2. Data and Methodology 2. Data and Methodology termine the characteristics of time series data before mod- elling and forecasting them. Their suggestions were fol- In the empirical modeling, we use daily diesel consump- lowed by Agrawal (2015) who employed an autoregressive a E-mail: [email protected] b E-mail: [email protected] c E-mail: [email protected] 1 COVID-19 refers to the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2). The Effect of the COVID-19 Outbreak on the Turkish Diesel Consumption Volatility Dynamics tion. Data cover the 01/01/2014 - 15/06/2020 period. Diesel consumption data are obtained from the Energy Market Regulatory Authority database. Descriptive statistics of the daily diesel consumption of Turkey are presented in Table 1. The data suggest that the average and median daily diesel consumption are 48.8 and 49.2 million liters, respec- tively, while the maximum and minimum values are 22.3 and 6.2 million liters, respectively. The standard deviation stands at 10.01 million liters. This implies that daily diesel consumption varies a lot in our sample period. Moreover, the Jarque-Bera test rejects the null hypothesis of a normal distribution for daily diesel consumption. Negative skew- ness value for diesel consumption indicates left skewed dis- tribution, whereas the kurtosis value of 3.242 indicates the fat tail characteristic of diesel consumption. In order to ensure the stationarity condition, we employ Figure 1: V Figure 1: Volatilit olatility variable based on the T y variable based on the TGAR GARCH(1,1) CH(1,1) logarithmic difference (growth rate) of diesel consumption Model Model data. We first investigate the stationarity of diesel con- Notes: The figure presents the volatility dynamics from 2013 till 2020 for Turkish sumption growth rate using both the conventional Ng-Per- Diesel consumption. The volatility variable (VOL_TARCH) is obtained from ron unit root test and Zivot- Andrews (1992) structural TGARCH model. break unit root test and found stationarity of the diesel con- sumption data. For empirical analysis, in order to obtain the conditional the Turkish diesel consumption volatility. Volatility dynam- heteroscedasticity (volatility) of the daily diesel consump- ics over time is presented in Figure 1. tion data, we limit ourselves to simple univariate models As seen in Figure 1, the uncommonly high volatility that have comparable performances with more sophisticat- starts after 11/03/2020, which is the day after of the first ed models as shown in the literature. We first estimate the COVID-19 case was announced in Turkey. The volatility val- best fit ARIMA structure for the mean equation. Next, we ue reaches its peak in mid-April 2020 due to new sanctions estimate alternative GARCH type models in order to analyze imposed by the government, such as the weekend curfews volatility dynamics of the daily diesel consumption. These and intercity travel restrictions. As of 11/03/2020, when the models include ARCH, GARCH, Exponential GARCH pandemic started, there was a significant increase in diesel (EGARCH) and Threshold GARCH (TGARCH), and the best consumption volatility. Uncertainty driven initial purchases model is determined based on information criteria and fore- of diesel were followed by a steady decline in consumption. cast performances. Then, we obtain the conditional vari- Volatility reached highest levels during curfews, fell in re- ance from the selected model as daily diesel volatility. sponse to normalization policies in Turkey, and converged to 0 over time. 3. R 3. Results esults 4. C 4. Conclusion onclusion After confirming stationarity, SARMA(7,7)(1,1) model is found to be the best model for daily diesel consumption ac- In this note, we employ a SARMA(7,7)(1,1) model and cording to model selection and forecast performance crite- the TGARCH(1,1) model to represent and variance of the 3 4 ria . Turkish diesel consumption. We find that a phase of high After we define the mean equation, we analyze volatility volatility starts after mid-April 2020 and peaks on 24/05/ dynamics of the daily diesel consumption variable by em- 2020. A 29% decrease in diesel sales is observed between ploying alternative ARCH family models including ARCH, 10/03/2020 and 01/06/2020. Volatility declines after 01/06/ GARCH, EGARCH and TGARCH models . We define the best 2020 consistent with a return to normalcy. Shrinkage in the volatility model by comparing the forecast performances of diesel market has an important adverse effect on indirect the models . Table 1 presents the results of the alternative tax revenues. We make two policy suggestions to mitigate volatility models. the disruption in the market. First, we suggest a temporary Table 1 indicates that the TGARCH(1,1) model is the best rearrangement of profit margins of dealers and distributors. performing model according to both model selection and Second, we suggest enactment of a temporary tax regula- forecast performance criteria. The conditional het- tion to compensate for lost tax revenue. eroscedasticity of the selected model is taken as a proxy for 2 We kindly refer the reader to Gungor et al. (2020) for technical details of ARIMA models and volatility models. 3 We tried all alternative models from SARMA(0,0)(0,0) to SARMA(7,7)(1,1) that includes 256 models and defined the best 5 models accord- ing to the Log Likelihood (LogL), Akaike Information Criterion (AIC), and the Bayesian Information Criterion (BIC). The five estimated models are ARMA(7,1), ARMA(7,7), ARMA(7,6), SARMA(7,6)(1,1) and SARMA(7,7) (1,1). 4 Alternative ARMA model results are not presented in the text to save space. The results are available from authors upon request. 5 Student t distribution is used instead of a normal distribution as Jarqua-Bera test indicated non-normality. 6 Gungor et al. (2020) found that COVID-19 outbreak distorts the gasoline and diesel forecasts in Turkey. 10/03/2020 is the day on which the first COVID-19 case was announced. For this reason, we employ 01/01/2004 to 30/01/2020 period for model estimation and 31/01/ 2020 to 10/03/202 period for forecasting. Energy RESEARCH LETTERS 2 The Effect of the COVID-19 Outbreak on the Turkish Diesel Consumption Volatility Dynamics T Table 1: R able 1: Results from the v esults from the volatilit olatility model y model AR ARCH(1) CH(1) GAR GARCH(1,1) CH(1,1) T TGAR GARCH(1,1) CH(1,1) EGAR EGARCH(1,1) CH(1,1) Mean Equation Mean Equation AR(7) 0.998* 0.998* 0.998* 0.998* SAR(1) 0.629* 0.650* 0.676* 0.662* MA(7) -0.987* -0.863* -0.989* -0.923* SMA(1) -0.874* -0.906 -0.880* -0.881* V Variance Equation ariance Equation ε 0.833* 0.514* 0.385* t − 1 h 0.325* 0.423* t − 1 I 0.243* t − 1 -0.078 0.699* ln h 0.232* t − 1 C 0.001 0.001 0.001 0.705 Model Selection Criteria Results Model Selection Criteria Results AIC -3.144 -3.047 -3.164 -3.164 -3.104 SC -3.127 -3.028 -3.142 -3.142 -3.082 F Forecast P orecast Performance erformance RMSE 4729051 4597081 3360312 3360312 3944596 MEA 3699032 3594395 2698345 2698345 2919957 MAPE 7.056 6.867 5.228 5.228 5.812 Theil 0.047 0.046 0.033 0.033 0.038 No Notes tes: The estimation period is 01/01/2014-30/01/2020 and the forecast period is 31/01/2020-3/10/2020. * indicates significance at the 1% level. AIC and SC represent Akaike Informa- tion Criterion and Schwartz Criterion, respectively. RMSE is root mean square error, MEA is mean absolute error, MAPE is mean absolute percentage error, Theil stands for the Theil inequality index. This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CC- BY-SA-4.0). View this license’s legal deed at https://creativecommons.org/licenses/by-sa/4.0 and legal code at https://cre- ativecommons.org/licenses/by-sa/4.0/legalcode for more information. Energy RESEARCH LETTERS 3 The Effect of the COVID-19 Outbreak on the Turkish Diesel Consumption Volatility Dynamics REFERENCES Agrawal, P. (2015). India’s petroleum demand: Li, Z., Rose, J. M., & Hensher, D. A. (2010). Estimations and projections. Applied Economics, Forecasting automobile petrol demand in Australia: 47(12), 1199–1212. https://doi.org/10.1080/0003684 An evaluation of empirical models. Transportation 6.2014.993131 Research Part A: Policy and Practice, 44(1), 16–38. htt ps://doi.org/10.1016/j.tra.2009.09.003 Bhutto, A. W., Bazmi, A. A., Qureshi, K., Harijan, K., Karim, S., & Ahmad, M. S. (2017). Forecasting the Rashid, A., & Kocaaslan, O. K. (2013). Does Energy consumption of gasoline in transport sector in Consumption Volatility Affect Real GDP Volatility? pakistan based on ARIMA model. Environmental An Empirical Analysis for the UK. International Progress and Sustainable Energy, 36(5), 1490–1497. h Journal of Energy Economics and Policy, 384–394. ttps://doi.org/10.1002/ep.12593 Zhu, S., Wang, J., Zhao, W., & Wang, J. (2011). A Kocherlakota, N. (2009). Modern Macroeconomic seasonal hybrid procedure for electricity demand Models as Tools for Economic Policy. In N. forecasting in China. Applied Energy, 88(11), Kocherlakota (Ed.), The Region. 3807–3815. https://doi.org/10.1016/j.apenergy.2011.0 5.005 Li, J., Wang, R., Wang, J., & Li, Y. (2018). Analysis and forecasting of the oil consumption in China based on combination models optimized by artificial intelligence algorithms. Energy, 144, 243–264. http s://doi.org/10.1016/j.energy.2017.12.042 Energy RESEARCH LETTERS http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Energy RESEARCH LETTERS Unpaywall

The Effect of the COVID-19 Outbreak on the Turkish Diesel Consumption Volatility Dynamics

Ertuğrul, H. Murat; Güngör, B. Oray; Soytaş, Uğur
Energy RESEARCH LETTERS – Oct 4, 2020
Download PDF
4 pages

Loading next page...
 
/lp/unpaywall/the-effect-of-the-covid-19-outbreak-on-the-turkish-diesel-consumption-biSL4U0RAe

References (13)

Publisher
Unpaywall
ISSN
2652-6433
DOI
10.46557/001c.17496
Publisher site
See Article on Publisher Site

Abstract

Ertuğrul, H. M., Güngör, B. O., & Soytaş, U. (2020). The Effect of the COVID-19 Outbreak on the Turkish Diesel Consumption Volatility Dynamics. Energy RESEARCH LETTERS, 1(3). https://doi.org/10.46557/001c.17496 COVID-19 and Energy The Eff The Effec ect of the C t of the CO OVID-19 Outbreak on the T VID-19 Outbreak on the Turkish Diesel urkish Diesel C Consump onsumption V tion Volatilit olatility Dynamics y Dynamics a b c 1 2 3 H. Murat Ertuğrul , B. Oray Güngör , Uğur Soytaş 1 2 3 Ministry of Treasury and Finance, Turkey, Energy Market Regulatory Authority, Turkey, Department of Business Administration, Middle East Technical University, Turkey Keywords: covid-19 pandemic, arch family models, arima models, diesel consumption https://doi.org/10.46557/001c.17496 Energy RESEARCH LETTERS Vol. 1, Issue 3, 2020 We analyze the effect of the COVID-19 outbreak on volatility dynamics of the Turkish diesel market. We observe that a high volatility pattern starts around mid-April, 2020 and reaches its peak on 24/05/2020. This is due to the government imposed weekend curfews and bans on intercity travels. Two policy suggestions are provided. First is a temporary rearrangement of profit margins of dealers and liquid fuel distributors; and, second is a temporary tax regulation to compensate lost tax revenue. distributed lag based co-integration model in order to fore- 1. Introduc 1. Introduction tion cast gasoline and diesel demand for India. In a similar fash- ion, but with a simpler model, Bhutto et al. (2017) forecast- Energy is a necessity for societies and its security and ac- th ed annual gasoline consumption for Pakistan. Given this lit- cessibility is a priority for policy makers. Since the 19 cen- erature, one research gap is that there are no studies that tury, one of the main sources of energy has been oil and examine how energy consumption volatility changes in the oil products, and the importance of energy has grown with face of a global pandemic. This study aims to fill this gap time . Hence, movements in oil prices have been an im- by studying the Turkish diesel market during the recent portant subject of research. All major oil price shocks, such COVID-19 pandemic. In addition, we also show how the as the rise in oil prices of the 1970s and 2000s and the oil pandemic distorts the forecasting performance of models. price glut in the 1980s, have attracted research interests The fluctuations in oil prices started at the beginning with the literature attempting to explain fluctuations (and of 2020 with the spread of COVID-19. Volatility in energy hence volatility) in energy prices (Kocherlakota, 2009). markets increased. We add to the understanding of energy Energy consumption and production forecasts were markets by examining the effect of the COVID-19 pandemic made with special attention to oil crises and oil gluts over on diesel consumption. Our paper is the first to investigate the last 50 years. Many econometric applications were con- volatility dynamics of the diesel consumption in light of the ducted to analyze the demand dynamics of various oil prod- global pandemic. ucts and to forecast consumption. For example, for Chinese The purpose of this paper is to examine the effects of the electricity consumption forecasts, Zhu et al. (2011) pro- COVID-19 pandemic on the Turkish diesel market and as- posed an integration of the moving average procedure and sess the performance of simple forecasting tools. To achieve seasonal autoregressive integrated moving average model this aim, we employ variants of the generalized ARCH (SARIMA) with weight coefficients. J. Li et al. (2018), on (GARCH) models. These models allow us to investigate the the other hand, considered more than 20 combination mod- disruptive effects of the pandemic on the Turkish diesel els using traditional combination methods to forecast oil market. We find that volatility starts rising with the an- consumption in China. Rashid and Kocaaslan (2013) exam- nouncement of the first case and peaks at the end of May ined the relationship between energy consumption volatili- 2020. Even though purchases increase in the 2 days follow- ty and unpredictable variations in real GDP by estimating a ing the announcement, they then assume a steady down- Markov switching autoregressive conditional heteroskedas- ward trend. ticity (ARCH) model for the UK. In addition, Z. Li et al. The paper is organized as follows: Data and methodology (2010) provided an analysis of future gasoline demand in are presented in section 2 and results are presented in sec- Australia by using various models. They found that more tion 3. Section 4 concludes. sophisticated models do not always produce better forecast- ing results compared to simple models. They advised to de- 2. Data and Methodology 2. Data and Methodology termine the characteristics of time series data before mod- elling and forecasting them. Their suggestions were fol- In the empirical modeling, we use daily diesel consump- lowed by Agrawal (2015) who employed an autoregressive a E-mail: [email protected] b E-mail: [email protected] c E-mail: [email protected] 1 COVID-19 refers to the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2). The Effect of the COVID-19 Outbreak on the Turkish Diesel Consumption Volatility Dynamics tion. Data cover the 01/01/2014 - 15/06/2020 period. Diesel consumption data are obtained from the Energy Market Regulatory Authority database. Descriptive statistics of the daily diesel consumption of Turkey are presented in Table 1. The data suggest that the average and median daily diesel consumption are 48.8 and 49.2 million liters, respec- tively, while the maximum and minimum values are 22.3 and 6.2 million liters, respectively. The standard deviation stands at 10.01 million liters. This implies that daily diesel consumption varies a lot in our sample period. Moreover, the Jarque-Bera test rejects the null hypothesis of a normal distribution for daily diesel consumption. Negative skew- ness value for diesel consumption indicates left skewed dis- tribution, whereas the kurtosis value of 3.242 indicates the fat tail characteristic of diesel consumption. In order to ensure the stationarity condition, we employ Figure 1: V Figure 1: Volatilit olatility variable based on the T y variable based on the TGAR GARCH(1,1) CH(1,1) logarithmic difference (growth rate) of diesel consumption Model Model data. We first investigate the stationarity of diesel con- Notes: The figure presents the volatility dynamics from 2013 till 2020 for Turkish sumption growth rate using both the conventional Ng-Per- Diesel consumption. The volatility variable (VOL_TARCH) is obtained from ron unit root test and Zivot- Andrews (1992) structural TGARCH model. break unit root test and found stationarity of the diesel con- sumption data. For empirical analysis, in order to obtain the conditional the Turkish diesel consumption volatility. Volatility dynam- heteroscedasticity (volatility) of the daily diesel consump- ics over time is presented in Figure 1. tion data, we limit ourselves to simple univariate models As seen in Figure 1, the uncommonly high volatility that have comparable performances with more sophisticat- starts after 11/03/2020, which is the day after of the first ed models as shown in the literature. We first estimate the COVID-19 case was announced in Turkey. The volatility val- best fit ARIMA structure for the mean equation. Next, we ue reaches its peak in mid-April 2020 due to new sanctions estimate alternative GARCH type models in order to analyze imposed by the government, such as the weekend curfews volatility dynamics of the daily diesel consumption. These and intercity travel restrictions. As of 11/03/2020, when the models include ARCH, GARCH, Exponential GARCH pandemic started, there was a significant increase in diesel (EGARCH) and Threshold GARCH (TGARCH), and the best consumption volatility. Uncertainty driven initial purchases model is determined based on information criteria and fore- of diesel were followed by a steady decline in consumption. cast performances. Then, we obtain the conditional vari- Volatility reached highest levels during curfews, fell in re- ance from the selected model as daily diesel volatility. sponse to normalization policies in Turkey, and converged to 0 over time. 3. R 3. Results esults 4. C 4. Conclusion onclusion After confirming stationarity, SARMA(7,7)(1,1) model is found to be the best model for daily diesel consumption ac- In this note, we employ a SARMA(7,7)(1,1) model and cording to model selection and forecast performance crite- the TGARCH(1,1) model to represent and variance of the 3 4 ria . Turkish diesel consumption. We find that a phase of high After we define the mean equation, we analyze volatility volatility starts after mid-April 2020 and peaks on 24/05/ dynamics of the daily diesel consumption variable by em- 2020. A 29% decrease in diesel sales is observed between ploying alternative ARCH family models including ARCH, 10/03/2020 and 01/06/2020. Volatility declines after 01/06/ GARCH, EGARCH and TGARCH models . We define the best 2020 consistent with a return to normalcy. Shrinkage in the volatility model by comparing the forecast performances of diesel market has an important adverse effect on indirect the models . Table 1 presents the results of the alternative tax revenues. We make two policy suggestions to mitigate volatility models. the disruption in the market. First, we suggest a temporary Table 1 indicates that the TGARCH(1,1) model is the best rearrangement of profit margins of dealers and distributors. performing model according to both model selection and Second, we suggest enactment of a temporary tax regula- forecast performance criteria. The conditional het- tion to compensate for lost tax revenue. eroscedasticity of the selected model is taken as a proxy for 2 We kindly refer the reader to Gungor et al. (2020) for technical details of ARIMA models and volatility models. 3 We tried all alternative models from SARMA(0,0)(0,0) to SARMA(7,7)(1,1) that includes 256 models and defined the best 5 models accord- ing to the Log Likelihood (LogL), Akaike Information Criterion (AIC), and the Bayesian Information Criterion (BIC). The five estimated models are ARMA(7,1), ARMA(7,7), ARMA(7,6), SARMA(7,6)(1,1) and SARMA(7,7) (1,1). 4 Alternative ARMA model results are not presented in the text to save space. The results are available from authors upon request. 5 Student t distribution is used instead of a normal distribution as Jarqua-Bera test indicated non-normality. 6 Gungor et al. (2020) found that COVID-19 outbreak distorts the gasoline and diesel forecasts in Turkey. 10/03/2020 is the day on which the first COVID-19 case was announced. For this reason, we employ 01/01/2004 to 30/01/2020 period for model estimation and 31/01/ 2020 to 10/03/202 period for forecasting. Energy RESEARCH LETTERS 2 The Effect of the COVID-19 Outbreak on the Turkish Diesel Consumption Volatility Dynamics T Table 1: R able 1: Results from the v esults from the volatilit olatility model y model AR ARCH(1) CH(1) GAR GARCH(1,1) CH(1,1) T TGAR GARCH(1,1) CH(1,1) EGAR EGARCH(1,1) CH(1,1) Mean Equation Mean Equation AR(7) 0.998* 0.998* 0.998* 0.998* SAR(1) 0.629* 0.650* 0.676* 0.662* MA(7) -0.987* -0.863* -0.989* -0.923* SMA(1) -0.874* -0.906 -0.880* -0.881* V Variance Equation ariance Equation ε 0.833* 0.514* 0.385* t − 1 h 0.325* 0.423* t − 1 I 0.243* t − 1 -0.078 0.699* ln h 0.232* t − 1 C 0.001 0.001 0.001 0.705 Model Selection Criteria Results Model Selection Criteria Results AIC -3.144 -3.047 -3.164 -3.164 -3.104 SC -3.127 -3.028 -3.142 -3.142 -3.082 F Forecast P orecast Performance erformance RMSE 4729051 4597081 3360312 3360312 3944596 MEA 3699032 3594395 2698345 2698345 2919957 MAPE 7.056 6.867 5.228 5.228 5.812 Theil 0.047 0.046 0.033 0.033 0.038 No Notes tes: The estimation period is 01/01/2014-30/01/2020 and the forecast period is 31/01/2020-3/10/2020. * indicates significance at the 1% level. AIC and SC represent Akaike Informa- tion Criterion and Schwartz Criterion, respectively. RMSE is root mean square error, MEA is mean absolute error, MAPE is mean absolute percentage error, Theil stands for the Theil inequality index. This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CC- BY-SA-4.0). View this license’s legal deed at https://creativecommons.org/licenses/by-sa/4.0 and legal code at https://cre- ativecommons.org/licenses/by-sa/4.0/legalcode for more information. Energy RESEARCH LETTERS 3 The Effect of the COVID-19 Outbreak on the Turkish Diesel Consumption Volatility Dynamics REFERENCES Agrawal, P. (2015). India’s petroleum demand: Li, Z., Rose, J. M., & Hensher, D. A. (2010). Estimations and projections. Applied Economics, Forecasting automobile petrol demand in Australia: 47(12), 1199–1212. https://doi.org/10.1080/0003684 An evaluation of empirical models. Transportation 6.2014.993131 Research Part A: Policy and Practice, 44(1), 16–38. htt ps://doi.org/10.1016/j.tra.2009.09.003 Bhutto, A. W., Bazmi, A. A., Qureshi, K., Harijan, K., Karim, S., & Ahmad, M. S. (2017). Forecasting the Rashid, A., & Kocaaslan, O. K. (2013). Does Energy consumption of gasoline in transport sector in Consumption Volatility Affect Real GDP Volatility? pakistan based on ARIMA model. Environmental An Empirical Analysis for the UK. International Progress and Sustainable Energy, 36(5), 1490–1497. h Journal of Energy Economics and Policy, 384–394. ttps://doi.org/10.1002/ep.12593 Zhu, S., Wang, J., Zhao, W., & Wang, J. (2011). A Kocherlakota, N. (2009). Modern Macroeconomic seasonal hybrid procedure for electricity demand Models as Tools for Economic Policy. In N. forecasting in China. Applied Energy, 88(11), Kocherlakota (Ed.), The Region. 3807–3815. https://doi.org/10.1016/j.apenergy.2011.0 5.005 Li, J., Wang, R., Wang, J., & Li, Y. (2018). Analysis and forecasting of the oil consumption in China based on combination models optimized by artificial intelligence algorithms. Energy, 144, 243–264. http s://doi.org/10.1016/j.energy.2017.12.042 Energy RESEARCH LETTERS

Journal

Energy RESEARCH LETTERS – Unpaywall

Published: Oct 4, 2020

There are no references for this article.