Why did the historical energy forecasting succeed or fail? A case study on IEA's projection

• GDP is the leading source of IEA's historic energy demand forecast errors.
• Gas appears to be the most severely biased among all fuel types.
• China remains to be more severely underestimated than any other country groups.
• Underestimation of GDP leads to aggravated errors in demand forecasts.

Medium-to-long term energy prediction plays a widely-acknowledged role in guiding national energy strategy and policy but could also lead to serious economic and social chaos when poorly executed. A consequent issue may be the effectiveness of these predictions, and sources that errors can be traced back to. The International Energy Agency (IEA) has published its annual World Energy Outlook (WEO) concerning energy demand based on its long term world energy model (WEM) under specific assumptions towards uncertainties such as population, macroeconomy, energy price and technology. Unfortunately, some of its predictions succeeded while others failed. We in this paper attempt to decompose the leading source of these errors quantitatively. Results suggest that GDP acts as the leading source of demand forecasting errors while fuel price comes thereafter, which requires extra attention in forecasting. Gas, among all fuel types witness the most biased projections. Ignoring the catch-up effect of acquiring rapid economic growth in developing countries such as China will lead to huge mistake in predicting global energy demand. Finally, asymmetric cost of under- and over-estimation of GDP suggests a potentially less conservative stance in the future.