The IEA’s Latest Forecast Is Out: How Reliable Are The Price Scenarios?

The IEA’s latest World Energy Outlook has been released and at a ‘mere’ 464 pages, did not crash the internet. (When I was a boy, we would say its release meant the death of many trees for the printing.) As always, there will be any number of pundits gunning for it and I”ve got ol’ Betsy loaded with 10-gauge. To pick one tiny part of the report, the figure below shows the forecasts of prices under three scenarios (I know, Fatih, they are not predictions but scenarios.) I have interpolated between the points provided by the IEA; few bother to produce year-by-year price forecasts any more.

The 2020 Oil Price Scenarios from the IEA’s World Energy Outlook

The author from IEA data.

First, the IEA’s forecasters are to be commended for producing a timely, analysis-filled work and for subjecting themselves to the slings and arrows of outrageous pundits (like your humble narrator). Second, many observe it with a mix of naivete and ignorance, with some taking it as Holy Writ—if they agree with its findings. Others attack the Agency for presumed bias, pointing to their track record of price forecasts that (mostly) proved too high. The reality is that a) everyone is biased, including the computer models, because of human nature, b) the IEA’s forecasts are not significantly different from the consensus forecast, partly because others are cautious about differing from them and receiving the industry’s opprobrium. (I’ve had people suggest I was drunk or any idiot for not agreeing with the consensus of a $100 floor price for oil, for example.)

The tendency of people to say that oil markets can’t be forecast, and especially that I have shown that, is one minor irritant for me (and there are many, I’m a GOM, grumpy old man). I have many papers on both short- and long-term oil market forecasting which point to a) the bad track record, b) the inherent uncertainties, and c) the errors and bias that lead to much of the shortcomings. But it’s not the same as saying that better forecasts can’t be made compared to the industry’s track record.

It’s a bit like the old joke of the sculptor who, asked how he created such a beautiful statue of an elephant explained that you start with a block of stone and carve away everything that doesn’t look like an elephant. It’s the same with forecasting: get rid of the underlying errors and you won’t have an elephant but you’ll be getting a lot closer to one.

A glaring unelephantal aspect of oil market forecasting is the repeated insistence that prices must a) rise and b) return to ‘normal’. It is not wrong to say that oil prices are mean-reverting in general, it’s just that the open question is the level of the mean. From 1861 to 1971, the average price (in 2018 dollars) was $27/barrel. Since 1971, the average price has been $59 (to 2018), while from the year 2000, the average price was $74. Cherry-picking the time period can allow one to rationalize a variety of ‘means’ to which the price can revert. (Which also demonstrates the value of understanding the drivers of prices, rather than just curve-fitting.)

But most forecasters do not obey even this simple rule. When prices were high, as in 2014, the forecast consensus was for them to continue rising, as the figure below shows. (The typical exception being my forecast, labeled SEER, which is nearly always the lowest in surveys.) Even a superficial survey will show that when the price was above $100/barrel, prices below that level were all but unthinkable, highlighting the caution and consensus-seeking of most forecasters.

Oil Price Forecasts in 2014

The author from 2014 AEO price survey

In addition, it is more typical to adjust the long-term price forecast in response to current price levels. If prices soar, don’t expect them to mean-revert, but to keep rising. If prices are low, project them to be flat or to rise gradually. The figure below shows the IEA forecasts from different points: 2001, after the 1998 oil price collapse, the future price was seen to be low. (Again, this was the consensus, not the IEA being a flawed outlier.) In 2007, before the bubble in 2008, the expectation was that prices would rise slightly from then-current levels. After the 2008 bubble, the price forecast was raised sharply, basically doubling from the 2007 level in 2009, and still in 2014, even as U.S. shale production was surging.

Evolution of IEA Oil Price Forecasts (2019$/bbl)

The author from IEA data.

Now after the 2015 oil price collapse, the IEA has lowered its forecast so that it resembles that of 2007. This might be taken as evidence of a moderate outlook, and an implied recovery to mid-2000s price levels, but those levels were still nearly double the long-term historical mean. Which doesn’t mean it’s unreasonable, but that a degree of skepticism is warranted and a close examination of the reasons for prices to be higher than the long-term mean (and below the more recent experience).

via Energy News

Categories: Energy