Previsão de exportação do setor alimentício brasileiro para 2025-2026: uma abordagem através da análise de séries temporais

Abstract

The present study aims to forecast Brazilian food sector exports for the period 2025–2026 using statistical time series models. The research was conducted based on monthly export data for food products provided by IPEADATA (series FUNCEX12_XVAL2N12), covering the period from January 1996 to August 2025. Two models were applied for analysis and forecasting: the Exponential Smoothing Method (ESM) and the Box & Jenkins Method (ARIMA). From the comparison of their performances, both models showed satisfactory results, with mean absolute percentage errors (MAPE) below 10%. However, the ESM demonstrated better out-of-sample performance and was therefore selected as the most suitable model to generate the final forecasts. The results indicate a slight decline of 1.89% in exports in 2025, followed by a recovery of 1.41% in 2026, suggesting a stabilization phase after a period of strong growth. These findings reinforce the relevance of applying time series models as a tool to support economic analysis and the formulation of foreign trade strategies. By enabling the identification of trends, seasonal patterns, and cyclical fluctuations, such models contribute to strategic planning in the sector and consequently strengthen Brazil’s position as one of the world’s leading food exporters.