Regression Analysis Of Oil Price Return Essay Example

of crude oil also has an effect on the average consumer through petrol and heating bills. Over the past ten years, the price of oil has seen wild fluctuations, going from about $30 a barrel in 2004 to $140 a barrel in 2008, before dropping below $40 a barrel in 2009. The reasons for this rapid rise and fall in prices re the subject of much debate. For the purpose of our study, we considered various multiple factors as possible regression for our y variable "The return on the spot price of crude oil". In a study by the European Central Bank, Dies, Gassiest, et al. 2008) looks at changes in downstream sectors (such as refineries), lack of sufficient spare production capacity, and expectations of futures shortages (quantified by futures market conditions).

Hang et al. (1996) finds that the broader stock market is largely unaffected by changes in crude oil price volatility, although it does affect the price of Oil companies' stocks. Other factors that have been linked with oil price changes include gross domestic product (Hamilton, 1 996), interest rates (Ferreter, 1996), and stock returns (Hang et al. , 1996).

Azores (2014) looks at the relationship between Oil prices and Natural Gas pence. In a report released in June 2014, the US Energy Information Administration established a link between oil prices and seven factors including changes in expectations of economic growth, geopolitical and economic events, changes in Saudi Arabian Oil production, and inventory builds, among others. Aliquots, Killing and Effusion (2011 ) studies the relationship between oil price and agronomic aggregates, and tests the ability of oil futures in forecasting the nominal price of

oil.

Burst, Bujumbura (2011) studies the relationship between oil prices and the CSS Dollar, citing possible reasons as declining real interest rates and the use of oil futures as a hedge against a falling US Dollar. Through group brainstorming, we came up with a number of variables that theoretically should affect the price of crude oil, and we used Bloomberg to find data on the same.

Our two main criteria for a "good" variable were statistical significance and RE. We conducted a regression analysis as well as ultimate regression analysis to double check the variables we selected on the Bloomberg terminal.

Moreover, so as to not omit any good variables, we broadened our search to the Oil commodity section to find relevant industry reports and prospective variables. Through a process of rough trial and error, and after eliminating several variables due to problems such as multimillionaires and heterosexuality, we finalized the three variables that are mentioned below. As crude oil is invoiced in USED, it is of interest to note how fluctuations in the value of the USED affect oil prices.

Another of our factors is the price of natural as, the closest substitute as a source of energy to oil.

Lastly, we seek to establish a relationship between returns in the S&P500 and oil prices. We used monthly time-series data over a period of ten years beginning from 2005 for the purpose of this study. To avoid issues of Nan-stationary data, we used oil price returns and S&P500 returns. 2.

0 Methodology Our y variable is the percentage monthly return on WIT oil spot prices. West Texas Intermediate Cushing crude oil price is typically

used as the reference spot price in the US, and therefore has been selected out of the different possible options.

For the explanatory variables, we have used the monthly return (percentage change) on the 500, monthly return on Natural gas (percentage change) and the monthly return on the USED index (percentage change). We have collected the time-series data from Bloomberg from 1/2005 to 12/2014 for the purpose of this study, resulting in 120 observations. This is a period of very high volatility and thus is of particular interest.

Our proposed equation is as follows: Return on Oil = ;0 + ;1*Anteaters + PA*Extenders + + E Where, - Coefficient of variables E - White noise error 2. . Descriptive Statistics The descriptive statistics for the data series is illustrated in the table above. This table describes the basic features of the dependent variable and the three explanatory variables, which represent the results for stationary and normality. The mean of the dependent variable (the return of crude oil price) is generally higher than each mean of each explanatory variable. However, the mean of the oil price return is lower than its median, because it is influenced by outliers and negative keenness.

The keenness is an indicator of asymmetry and deviation from a normal distribution (Mansfield et al. 2001). The return of S&P500 index suffers from the same issue of keenness. On the other hand, both the mean of dollar index return and the mean of return for natural gas price are higher than their medians because of positive keenness.

The standard deviation usually measures the deviation from the mean. It is a sensitive to keenness and

outlier. According to the TA able of summary statistics, the standard deviation of the return for natural gas price is higher than other variables.

It means the return for natural gas price is more sensitive to the outlier and keenness than other variables.

The standard aviation for dollar index return is less sensitive than the others. Kurtosis is used for checking whether the distribution is "peskiness" or "flattening', and the kurtosis of Gaussian distribution usually has O (Christensen et al. , 2007). According to the table, the distribution for all variables is more peaked than a Gaussian distribution. They are sharper than the normal distribution, with thicker tails.