Journal of Energy Research and Reviews

The relationship between energy efficiency and manufacturing productivity is often assumed to be linear. Yet theoretical and empirical evidence suggest significant asymmetries: firms may respond differently to efficiency gains than to efficiency losses. This study investigates the asymmetric impact of energy efficiency on manufacturing output in Nigeria using annual time-series data from 1981 to 2023. Estimating a Nonlinear Autoregressive Distributed Lag (NARDL) model, we decompose energy efficiency into positive (EE⁺) and negative (EE⁻) partial sums. The bounds test confirms cointegration (F = 9.036 > upper bound at 1%). Results reveal a striking asymmetry: in the long run, a 1% increase in energy efficiency raises output by 0.93%, while a 1% decrease reduces output by 2.24% — a loss aversion ratio of approximately 2.4:1. While short-run effects appear symmetric (F = 1.52, p = 0.229), long-run asymmetry is statistically significant (F = 11.61, p = 0.002). Capital positively affects output (elasticity 0.98), while labour exhibits a negative coefficient (-0.96), reflecting structural inefficiencies. The error correction term (-0.194, p < 0.01) indicates a 19% annual convergence to equilibrium. Diagnostic tests confirm the absence of serial correlation, heteroscedasticity, or specification error, and the CUSUM test indicates structural stability. These findings suggest that preventing declines in energy efficiency is more critical than promoting improvements — a fundamental departure from conventional policy wisdom.