When will oil demand peak ? The very fact that the question focuses on demand rather than supply is in itself remarkable, given where conventional wisdom on the subject stood only a decade ago. Now there is a consensus that demand will peak first but there is no agreement on when that peak will come. Shell speculated a few weeks ago that it would be within five to 15 years. The Opec producers’ cartel suggested recently that the peak could come in about 2029. But the International Energy Agency in its latest World Energy Outlook predicts that oil demand will be rising up to 2040. have read and agree to the terms and conditions, cookie policy and privacy policy.

Even the best organisations have blind spots, and that is especially true when it comes to the unscientific art of forecasting. The IEA is probably the most effective of all the international institutions, with a record of excellent analytical work. But its latest report misses one of the most important trends in today’s energy market and, as a result, the quality of the long-term forecast suffers.

The issue at stake is the level of demand. The IEA is now predicting that oil demand will continue to rise from today’s figure of about 92.5m barrels a day, and will not peak until after 2040, by which time the world will be consuming 103.5m b/d.

Forecasts should never be relied on for precision, and readers should be very wary of any projection looking almost a quarter of a century ahead that includes a decimal point. As I am sure the IEA’s economists would agree, all forecasts are written on the basis of uncertainty and imperfect knowledge. The IEA itself publishes what it calls a “450 scenario”, which starts from the assumption that governments across the world will take the actions necessary to limit climate change. That scenario is useful because it shows what needs to be done. But it is the IEA’s central scenario that companies, governments and investors use to shape their actions. That is where a directional mistake becomes dangerous.

The problem is that the forecast of continuously rising oil consumption for another three decades does not give any weight to what has been happening in the recent past.

Total primary energy consumption in OECD countries was down 3 per cent in 2015 compared with 2005. Total OECD oil demand was down 9 per cent.

These are not random numbers. The global economy has its problems but we are not in a recession. Economic growth last year was about 2.8 per cent across the Group of 20 industrialised nations, which represents 80 per cent of world economic activity, including China and India; but total energy demand rose by only 0.5 per cent.

This comes down to a shift in the energy intensity of GDP and an accompanying shift in per capita use. The nature of economic activity is changing and each generation of the technology involved in energy consumption — from cars to washing machines to computer servers – is more efficient than the last.

There is still demand growth in the emerging economies, led by India — but there, too, new science and engineering will bring new solutions and the opportunity to leapfrog some existing technology.

Behind the IEA’s forecast is a projection based on the status quo rolled forward, especially in terms of technology. Some of the potential changes are recognised but the pace of change is not. If the IEA had been in the business of forecasting computer technology 25 years ago, it seems likely that it would have recognised that we would have mobile phones and similar devices — but have missed the extent to which mobile technology can disrupt existing business models and the patterns of energy use that go with them.

Of course, no serious forecast can be built on technical shifts that have as yet barely begun. But the weight of probabilities has to be taken into account. The forecast the IEA is offering implies a negative view of the chances of action to counter air pollution in cities (documented in an excellent analytical report published alongside the WEO) and an equally negative view of the prospects of falls in the costs of electric vehicles. The forecast also suggests that the growing dependence of oil-importing countries such as China and India on production from the Middle East will have little or no impact on their policy decisions.

One way of testing forecasts is to look back at past attempts and to examine how the actual outcomes varied from the projections.

If you go back to the WEO published in 2006 you will find that the IEA predicted (again on its main scenario) that oil demand would rise from 84m b/d in 2005 to 99m b/d in 2015 and to 116m b/d in 2030. The IEA will probably say that that, too. was just a projection — but projections from a trusted organisation are taken seriously, and companies, governments or investors that planned on the basis of the 2006 WEO will have made some big mistakes. The wide gap between forecast and outcome suggests that there is a systematic bias in the IEA’s models that overstates demand.

To maintain credibility the IEA should now look hard at the changing patterns of energy demand and should consider shifting its view of the future to a ranged forecast – setting out the factors that matter and allowing readers to make their own judgments. That would better reflect the genuine uncertainty that exists and would refocus attention on the excellent analytical work — for instance, in the chapter on Mexico in this year’s WEO — that is the organisation’s true strength. Meanwhile the rest of us should focus on the implications of the peak for investment, corporate valuations and geopolitics.