Application of ARIMA Model and Exponential Smoothing Method in GDP Prediction of the United States

Abstract

The GDP of the United States always attracts the global attention. It represents the strength of the world's largest economy and the per capita living standard of developed countries. Therefore, many scholars try to use various methods to predict it. On the basis of previous research, this paper collects the Fred’s quarterly U.S. GDP data, uses ARIMA model and exponential smoothing to model the US's GDP from 1950 to 2021. Then, the authors compare the actual values with the predicted values to select a better model. After that, the established model is used to forecast the US's GDP. The results show that the two methods generally maintain a high degree of fit, but the error generated increases with the widening of the expected range. Among these two methods, the average error value of the exponential smoothing method is slightly smaller than ARIMA, but ARIMA's forecasts are more accurate in the early. Therefore, ARIMA should be used preferentially for the forecast in the short term and the exponential smoothing method should be preferred for long-term forecasts. Furthermore, we use the exponential smoothing method to make a short-term forecast for the U.S. GDP by the fourth quarter of 2023. The results show that the U.S. GDP will increase at an average rate of 0.9344%, and the upward trend is stable and cyclical. Finally, the result of the above provides an effective basis for the US economic construction work and plays an inspirational role for China's economic policy formulation.