A Data-Driven Trade Forecasting Framework for Industrial Chemical Imports: Evidence from Vietnam's Sodium Hydroxide Market

This study develops an integrated Business Intelligence (BI) and econometric framework for forecasting industrial chemical imports, validated through the case of Vietnam's Sodium Hydroxide (NaOH) market (HS Code 28151200). Utilizing a high-frequency dataset of 1,993 customs transactions spanning December 2024 to April 2025 - capturing 113,466 metric tons of NaOH valued at USD 41.4 million across 146 importers and 108 exporters - the research synthesizes transaction-level trade records with macroeconomic drivers including Vietnam's Industrial Production Index (IPI), GDP growth, VND/USD exchange rate and global caustic soda price benchmarks. A hybrid forecasting architecture combining log-linear structural demand estimation, Seasonal ARIMA and Vector Autoregression (VAR) is developed and validated. The proposed framework achieves a Mean Absolute Percentage Error (MAPE) of 6.2%, substantially outperforming traditional ARIMA (12.4%) and OLS (10.8%) benchmarks. VAR impulse response analysis confirms a statistically significant negative response of import volumes to exchange rate shocks, with a trough at Month 3 and an IPI spillover effect persisting for six months. The study makes three original contributions: (1) the first application of transaction-level customs data to econometric NaOH import demand modeling in Vietnam; (2) a validated hybrid ARIMA-VAR forecasting architecture; and (3) a replicable BI operationalization framework translating econometric outputs into procurement decision support. Results demonstrate that IPI and downstream industrial output are the dominant demand drivers, with a short-run price elasticity of -0.41 and exchange rate elasticity of -0.32. The framework is scalable to other HS-coded industrial inputs across emerging markets.