External Energy Shocks, Competitiveness, and Manufacturing Forecasts in Peru: A Bayesian VAR Benchmarking Approach

This study examines how external energy shocks propagate through sustainable competitiveness and manufacturing activity in Peru while assessing the forecasting performance of a Bayesian Vector Autoregression (BVAR). Using quarterly data derived from monthly observations for 1997Q1–2025Q4, the analysis considers Brent crude oil inflation, fuel inflation, real depreciation, terms-of-trade growth, export volume growth, and non-primary manufacturing production growth. A BVAR(2) with Minnesota prior was estimated to capture dynamic interdependence while mitigating overparameterization, and forecast performance was evaluated through a pseudo-out-of-sample recursive exercise against an unrestricted VAR(2) and a random walk benchmark. The results indicate that Brent oil inflation behaves as the most exogenous disturbance in the system and significantly predicts domestic fuel inflation, whereas terms-of-trade growth emerges as a relevant predictor of export volume and manufacturing activity. Generalized impulse responses and forecast error variance decompositions further support the view that external energy shocks are transmitted through price and competitiveness channels rather than through a fully reciprocal feedback structure. From a predictive perspective, the BVAR outperforms the random walk at all forecast horizons, showing RMSE gains ranging from 8.03% to 38.95% while performing very similarly to the unrestricted VAR. These findings support Bayesian shrinkage as a useful tool for interpretable and operational macroeconomic forecasting in Peru.