Modeling the Future of Saudi Arabia’s Economy: An Assessment of Time Series Forecasting Methods Up to 2030

Abstract

This paper compares three well-known time series forecasting models, ARIMA (autoregressive integrated moving average), NAR (nonlinear autoregressive), and RNN (recurrent neural network), to predict what Saudi Arabia's GDP will be by 2030. Given the nation's significant economic transformation under its Vision 2030 reform plan, accurate GDP forecasting becomes crucial for strategic planning and policy formulation. The study uses historical economic data, augmented by global economic conditions and national economic policies, to model and forecast future economic scenarios. We utilize the ARIMA model for its proficiency in handling linear time-series data with seasonal variations, and we explore the NAR and RNN models for their capacity to process nonlinear dynamics and complex temporal dependencies, respectively. The paper evaluates each model's predictive accuracy using root mean square error (RMSE) and coefficient of determination (R2), providing a comprehensive overview of their performance in forecasting Saudi Arabia's GDP. The findings aim to guide policymakers and economic stakeholders in selecting the most appropriate forecasting tool, given Saudi Arabia's dynamic, rapidly evolving economic landscape. The results show that ARIMA models are good at. Still, RNN models, especially those with LSTM architectures, are better at predicting long-term economic trends, which is important for making sureshort-term predictions. Still, RNN models, especially those with LSTM architectures, are better at predicting long-term economic trends, which is important for ensuring that the strategic goals of Vision 2030 are met.