Probabilistic Forecasting of Regulated Industrial Gas Tariffs: A Cost-Stack Framework with Application to Pakistan

Regulated industrial gas tariffs can differ substantially from international gas benchmarks because consumer prices reflect domestic and imported gas costs, exchange-rate movements and policy-mediated cost recovery. This study develops a probabilistic cost-stack framework for long-term forecasting of consumer-end industrial gas tariffs in mixed domestic-import gas systems. The model separately represents indigenous gas cost, imported-gas cost, import share, benchmark-linked LNG pricing, exchange-rate exposure and regulatory loading. Uncertainty in these drivers is propagated through Monte Carlo simulation to generate probabilistic tariff trajectories for industrial energy planning and technoeconomic analysis. Pakistan is used as the case study using semiannual data from 2015H1 to 2025H2. Model performance is assessed through terminal-informed historical reconstruction, rolling-origin out-of-sample validation and comparison with simple and reduced structural benchmarks. The historical reconstruction captured the broad tariff trajectory but showed that discrete regulatory adjustments are difficult to time. In rolling-origin validation, the structural model without regulatory markup achieved the lowest aggregate MAE of 207.58 PKR/MMBtu and CRPS of 153.39 PKR/MMBtu, indicating that the physical cost structure had predictive value while historical regulatory loading did not transfer reliably across periods. Under the baseline forward assumptions, the 2035H2 P10, P50 and P90 tariffs were 4944, 6214 and 7780 PKR/MMBtu, corresponding to annual growth rates of 8.39%, 11.03% and 13.69%. Terminal import share, exchange rate and regulatory loading were the dominant sensitivity drivers. The framework provides a transferable basis for long-term fuel-price exposure analysis in regulated gas markets.