SOLUTION: volatility of crude oil prices | My Assignment Tutor

volatility of crude oil prices Forecast using ARIMA Model [Date] [Company name] [Company address] Contents Introduction. 3 Research Objectives. 3 Data Description. 3 Model Description. 4 Empirical Analysis. 5 Discussion and Recommendation. 7 Conclusion. 7 References. 9 Appendix. 10 Introduction The trading of crude oil occurs in the global market. There are various streams of crude oil productions around the world, which tend to move together. A number of events affect the volatility of crude oil prices around the world. The events that affect the volatility of the crude oil prices can be divided into two major kinds – “geopolitical developments”, and “weather-related developments”. The location of most of the oil production facilities in countries with a history of political breakdowns is a major factor affecting the volatility of the crude oil prices around the world. The higher volatility in events is caused by the uncertainty of the future demand caused by the events (U.S Energy Information Administration, 2021). A forecast in crude can help in reducing the disadvantages resulting from the higher volatility of crude oil prices around the world. Statistical models can be employed to determine the future price trends which can reduce the disadvantages. Research Objectives The research aims to employ an ARIMA model to a set of time series data to determine the future price trends associated with crude oil prices. One of the major benefits of ARIMA model is that it can be helpful in calculating seasonality. The primary question that the project aims to answer is – “Can the trend in crude oil volatility be forecasted in the long run?” Data Description The primary requirement of the data set when using ARIMA models is that, there should be a minimum of 50 observations, but for more accurate results, it is essential to have a dataset of minimum 100 observations (Linden, 2015). The data set for the study includes the monthly crude oil price for a period of 100 months starting from November, 2012 and ending on February, 2021. In calculating the average monthly prices, the weighted average of the prices given by Brent, Dubai and Texas Intermediate has been used. Brent crude releases the price of the Atlantic Basin Crude Oils. The crude oil streams that Brent releases the price of are light crude. More than two-thirds of the globally traded crude oils follow the prices released by Brent Crude. As compared to other forms of crude oil, it is easier to make gasoline from the low-density crude oil which is priced by Brent (The Economist, 2018). The crude oil produced in the middle eastern countries is slightly of a lesser grade as compared to that of Brent. It is difficult to produce gasoline from middle eastern crude on account of higher sulfur content as compared to the crude oil extracted from the Atlantic basin. The price of middle eastern crude oil is priced in Dubai, which is the center point from where the oil is extracted (Fattouh, 2006). West Texas Intermediate, referred to as WTI in short is another price for the crude oil produced in the United States. Although the crude oil streams whose oil is priced under the WTI are light weight containing low sulfur, the fact that the steams are mostly located in landlocked areas makes it costly to produce gasoline from the WTI crude oil streams. The price for the WTI crude is released by the U.S Department of Energy (NASDAQ, 2021). In collecting the data for the research, the monthly weighted average value consists of all major areas from where the crude oil around the world is produced – Atlantic, Middle Eastern Countries and the United States. The data set has been collected in the above manner to render the results of the study more accurate. The weighted average monthly data is provided in Table 1 in the Appendix. Model Description A time-series in the simples of terms can be defined as a sequence, where the data points are recorded over constant time intervals, which combine together to form a data set. The frequency of the data set can be of any time frame such as daily, weekly, monthly etc (Prabhakaran, 2021). The present study makes use of monthly data, which forms a data set stretching over a period of 10 years. The ARIMA Model, also referred to as the “Auto Regressive Integrated Moving Average”, is a statistical time series model which can be used for the determination of future time points. The ARIMA model is based on the assumption that the current and the future time points can be determined by using the past time points. The forecast in ARIMA is predicted by using a number of lagged observations which are observed in the time series. If the usage of the ARIMA model shows that there is a trend existing in the data set, which is the past time set points, it will mean that the data set is not stationary and the existence of seasonality can be proved (Duke University, 2021). In the present study, the above principle will be applied to check if there is a seasonality in the crude oil prices. The present study makes use of an “Univariate Time Series Forecasting”, where predictors other than the time series are not used for the purpose of forecasting the future trend. One of the major benefits of ARIMA model is that, it makes use of only the time series for future forecasting. When there is a large number of data points which needs to be used for making the assessment, ARIMA model can be one of the best available models for making the forecast. The problems which are observed in multivariate forecasting are generally not observed in case ARIMA Model (Wang, Shen, & Jiang, 2019). Empirical Analysis The ARIMA set has been applied using Excel 2019 and NumXL 1.63. First the ARIMA model was applied to the data set using the ARIMA functionality available in NumXL 1.63. For the purpose of the study, the most basic form of ARIMA Model – ARIMA (1,1,1) has been used, where the values of the Auto-Regression, Integration and the Moving Averages is taken to be 1. The results of the ARIMA (1,1,1) test are provided below. The three tables display the results of the test, the goodness of fit of the model to the data set, and the residuals analysis. The reason 100 months were considered for the study is that NumXL allows a total of 100 iterations while creating an ARIMA model. The descriptions to the parameters in the model are provided in Table 2 in the Appendix. ARIMA(1,1,1) ParamValue μ-0.15 φ10.62 θ1-0.90 σ5.02 d1 Goodness-of-fit LLFAICCHECK -298.47604.951 Residuals (standardized) AnalysisAVGSTDEVSkewKurtosisNoise?Normal?ARCH?0.060.980.964.28FALSEFALSEFALSETarget0.001.000.000.00SIG?FALSEFALSETRUETRUE It can be seen from the Goodness of Fit table that the Model is a good fit for the data set at hand. Following the application of the ARIMA model, a forecasting has been carried out using the differentials in the original data set, the results of which are provided below. The forecasting below has been calculated for a period of 10 years. StepMeanSTDULLL1-5.639835.01834.195855-15.47552-4.813445.6491966.258782-15.88573-4.360066.0299767.458477-16.17864-4.137136.0823877.784127-16.05845-4.056586.1637748.024197-16.13746-4.063996.1647688.018738-16.14677-4.125736.1914778.009338-16.26088-4.221066.1933237.917634-16.35979-4.337126.2070587.82849-16.502710-4.4666.2121367.709566-16.6416 The graph below shows the increase in standard deviation over the period of years. The purpose of the study is to find out whether the crude oil prices can be forecasted in the long run. Discussion and Recommendation In statistics, standard deviation can be defined as the amount of variation observed in a given set of values. The greater the standard deviation, the greater is the conditional volatility that can be observed (Financial Forecasts Center, 2021). It can be seen from the above analysis that there is an increase in standard deviation over the period of time. If the horizon of the data set were to further increase, a further increase in standard deviation could have been observed in the data set. The above fact shows that, it is difficult to calculate the trend in crude oil prices in the very long-term period, but it might be calculated in the near to mid-term period. There are a number of factors affecting the volatility of crude oil prices. Some of the regions where the crude oil is produced are politically unstable locations. In certain cases, the prices of the crude oils knowingly inflated to increase the price and the profit, by limiting the supply. Thus, a varied number of events affecting the crude oil prices make it considerably difficult to assess the prices in the medium and the long run (U.S Energy Information Administration, 2021). The empirical research carried out also shows that the volatility increases with the timeframe making it difficult to calculate the crude oil prices in the long-term period. Conclusion There are a varied number of events which affects the price of global crude oil prices. It would be highly beneficial if the value of the crude oil prices could be calculated in the long-term period. To assess if the calculation of the prices is possible, an empirical analysis was carried out using a data set ranging over a period of 10 years. The data set consisted of the weighted average prices of Brent, WTI and Dubai, the three major indices for calculation of crude oil prices. On application of the ARIMA model and a forecast, it was observed that the standard deviation increased with the increased timeframe, which shows that calculation of the crude oil prices will be difficult in the long term. References Duke University. (2021). ARIMA models for time series forecasting. Retrieved from Duke University: https://people.duke.edu/~rnau/411arim.htm Fattouh, B. (2006). Middle East Crude Pricing and the Oman Crude. Oxford Institute of Energy Studies. Financial Forecasts Center. (2021). What is the Standard Deviation? Retrieved from Financial Forecasts Center: https://www.forecasts.org/stdev.htm Linden, A. (2015, March 11). What should be the minimum number of observations for a time series model? Retrieved from ResearchGate: https://www.researchgate.net/post/What-should-be-the-minimum-number-of-observations-for-a-time-series-model#:~:text=It%20depends%20on%20the%20modelling,(Box%20and%20Tiao%201975). NASDAQ. (2021). West Texas Intermediate. Retrieved from NASDAQ: https://www.nasdaq.com/glossary/w/west-texas-intermediate Prabhakaran, S. (2021). ARIMA Model – Complete Guide to Time Series Forecasting in Python. Retrieved from Machine Learning Plus: https://www.machinelearningplus.com/time-series/arima-model-time-series-forecasting-python/ The Economist. (2018, October 29). What is Brent crude? Retrieved from The Economist: https://www.economist.com/the-economist-explains/2018/10/29/what-is-brent-crude U.S Energy Information Administration. (2021). WHAT DRIVES CRUDE OIL PRICES? Retrieved from U.S Energy Information Administration: https://www.eia.gov/finance/markets/crudeoil/spot_prices.php#:~:text=These%20types%20of%20events%20may,changes%20in%20the%20short%20run. U.S Energy Information Administration. (2021). WHAT DRIVES CRUDE OIL PRICES? Retrieved from U.S Energy Information Administration: https://www.eia.gov/finance/markets/crudeoil/supply-opec.php Wang, Y., Shen, Z., & Jiang, Y. (2019). Comparison of autoregressive integrated moving average model and generalised regression neural network model for prediction of haemorrhagic fever with renal syndrome in China: a time-series study. BMJ Open Journal. Retrieved from https://bmjopen.bmj.com/content/9/6/e025773 Appendix Table 1: Monthly Crude Prices from March, 2011 to February, 2021 MonthPriceChangeMar-11108.65–Apr-11116.246.99%May-11108.07-7.03%Jun-11105.85-2.05%Jul-11107.921.96%Aug-11100.49-6.88%Sep-11100.820.33%Oct-1199.85-0.96%Nov-11105.415.57%Dec-11104.23-1.12%Jan-12107.072.72%Feb-12112.695.25%Mar-12117.794.53%Apr-12113.67-3.50%May-12104.09-8.43%Jun-1290.73-12.84%Jul-1296.756.64%Aug-12105.278.81%Sep-12106.280.96%Oct-12103.41-2.70%Nov-12101.17-2.17%Dec-12101.190.02%Jan-13105.13.86%Feb-13107.642.42%Mar-13102.52-4.76%Apr-1398.85-3.58%May-1399.370.53%Jun-1399.740.37%Jul-13105.265.53%Aug-13108.162.76%Sep-13108.760.55%Oct-13105.43-3.06%Nov-13102.63-2.66%Dec-13105.482.78%Jan-14102.1-3.20%Feb-14104.832.67%Mar-14104.04-0.75%Apr-14104.870.80%May-14105.710.80%Jun-14108.372.52%Jul-14105.23-2.90%Aug-14100.05-4.92%Sep-1495.85-4.20%Oct-1486.08-10.19%Nov-1476.99-10.56%Dec-1460.7-21.16%Jan-1547.11-22.39%Feb-1554.7916.30%Mar-1552.83-3.58%Apr-1557.548.92%May-1562.518.63%Jun-1561.31-1.92%Jul-1554.34-11.37%Aug-1545.69-15.92%Sep-1546.281.29%Oct-1546.961.47%Nov-1543.11-8.20%Dec-1536.57-15.17%Jan-1629.78-18.57%Feb-1631.034.20%Mar-1637.3420.34%Apr-1640.759.13%May-1645.9412.74%Jun-1647.693.81%Jul-1644.13-7.46%Aug-1644.881.70%Sep-1645.040.36%Oct-1649.299.44%Nov-1645.26-8.18%Dec-1652.6216.26%Jan-1753.591.84%Feb-1754.351.42%Mar-1750.9-6.35%Apr-1752.162.48%May-1749.89-4.35%Jun-1746.17-7.46%Jul-1747.663.23%Aug-1749.944.78%Sep-1752.956.03%Oct-1754.923.72%Nov-1759.939.12%Dec-1761.192.10%Jan-1866.238.24%Feb-1863.46-4.18%Mar-1864.171.12%Apr-1868.797.20%May-1873.436.75%Jun-1871.98-1.97%Jul-1872.670.96%Aug-1871.08-2.19%Sep-1875.366.02%Oct-1876.731.82%Nov-1862.32-18.78%Dec-1853.96-13.41%Jan-1956.584.86%Feb-1961.138.04%Mar-1963.794.35%Apr-1968.587.51%May-1966.83-2.55%Jun-1959.76-10.58%Jul-1961.482.88%Aug-1957.67-6.20%Sep-1960.044.11%Oct-1957.27-4.61%Nov-1960.45.47%Dec-1963.354.88%Jan-2061.63-2.72%Feb-2053.35-13.44%Mar-2032.2-39.64%Apr-2021.04-34.66%May-2030.3844.39%Jun-2039.4629.89%Jul-2042.076.61%Aug-2043.443.26%Sep-2040.6-6.54%Oct-2039.9-1.72%Nov-2042.36.02%Dec-2048.7315.20%Jan-2153.69.99%Feb-2160.4612.80% Table 2 ParametersMeaningμARMA Long Run Meanφ11st Coefficient of AR Componentθ11st Coefficient of MA ComponentσStandard Deviation of Residuals/Innovations dIntegration Order