Forecasting Gasoline Price with Time Series Models

Authors

Xin James He, Fairfield UniversityFollow

Abstract

This research forecasts the gasoline price in U.S. and analyzes its managerial implications by means of both univariate and multivariate time series forecasting models. Gasoline price forecast is among the most difficulty time series variables during its importance to the economy and extremely volatile nature. The average regular gasoline price in U.S. reached $5.06 per gallon in the month of June 2022, as opposed to $3.41 at the beginning of 2022, a 48% increase. While gasoline prices had been rising over the first half of 2022 due to supply chain disruptions as a result of global Covid 19 lockdowns and the Russia invasion of Ukraine since February 24, 2022, they then surprisingly came down in the second half of 2022 to $3.85 in November 2022, which has caught many consumers and business organizations off- guard. Both univariate time series forecasting models, such as exponential smoothing and autoregressive integrated moving average, and multivariate time series forecasting models, such as time series regression models, are used in this research with the data for the period January 2002 through November 2022. We find that the time series regression model with trend, season, GDP, CPI, and crude oil price turns out the be the best forecasting model both on the training data and the testing data, even with the testing data containing significant turning points. Managerial implications and future research directions are also discussed.

Recommended Citation

He, Xin James (2023) "Forecasting Gasoline Price with Time Series Models," Communications of the IIMA: Vol. 21: Iss. 1, Article 1.
DOI: https://doi.org/10.58729/1941-6687.1440
Available at: https://scholarworks.lib.csusb.edu/ciima/vol21/iss1/1