Getting the right oil price forecast

Point estimates do not serve a business well when uncertainty is wide, and history shows a range of real price uncertainty that is a factor of two to three times the central figure. Ironically, it seems that most oil companies treat production uncertainty with more sophistication than price uncertainty. Perhaps because inaccurate precision is easier to tolerate than uncertainty.

(Real price for imported oil, 1986 to present, per EIA, color coded by decade. See next graphic for color key.)

There are a number of sources for opinions about what that price may be, and they are useful to understand. (You can read about many of the publicly available forecasts in part 1.) One problem lies in thinking that one of those forecasts is right, but the bigger problem lies in thinking that there is a "right" forecast.

My own study of the history of forecasting from various participants shows us to be mostly an industry of permabulls, but different sources are governed by different drivers. Interpreting each forecast helps then to interpret the group of forecasts, to understand the collective thinking of the industry.

Not that there is safety in consensus or central estimates because there is not. In fact, consensus can be a orange flag that the group behavior may be creating unforeseen consequences. Instead studying the collective thinking provides the perspective on the field of play. Like the second-order strategy of a Keynesian beauty contest, it helps to know how the rest of the participants are playing.

Not that there is safety in consensus or central estimates because there is not.

Not just over the range of opinions in the current context but also taking the higher level view over the historical context helps. Our (read "my") thinking can get so crowded with near term, high frequency data, that we (again, read "I") lose track of the low frequency but large amplitude signal that truly moves multi-year strategy. Three macro dynamics I have observed to be powerful and persistent:

  • Commodity prices move in cycles. They shift above and below a baseline.
  • Those cycles last for years.
  • Prices often overcompensate during adjustments.

The graph below shows a sample history of swings for oil price. The fact that this kind of dynamic is inherent to commodities suggests that it is also immutable for the future of both oil and gas. Since they are truisms, they should remain in sight as we peer into the future.

(Nominal price for imported oil, 1986 to present, per EIA, color coded by decade. Also showing a notional base trendline for comparison.)

Of course, we think of macroeconomic forces driving these swings, but the total supply and total demand are not monoliths to be balanced, and they are not static. To remain balanced, supply must grow at the same pace as demand. Imbalances develop from uneven growth, and the growth of each is complicated. Both are dynamic composites of loosely related parts, some parts more local and some more global.

Supply comes from basins and technologies of various stages of maturity and regulated by very different above-ground risks and costs. As for demand, natural gas can be burned for electricity or for heat, catalyzed into plastics, transformed into fertilizer or frozen to be shipped overseas, and each demand has its own dynamic. Oil demand is more hegemonic, dominated by fuel for cars, trucks, planes and even boats and trains. But still it also feeds industrial plants. It can burn, lubricate, pave and even metamorphasize into clothing. The larger moving parts and the correlation among the moving parts merit study. To understand the whole, understand the parts.

Interestingly, the causes of the cycles can be more than variable growth in supply and variable growth in demand. The largest and sharpest of the swings have historically pertained to behaviors of OPEC and governments. As much as the irrationality of broader markets (such as when they fear peak oil supply), the behavior of individual market movers pivots on psychology and personality more than quantitative economics. Even though such drivers cannot be precisely quantified, the forces can be understood. As in physics, the forces in the commodity markets have direction and magnitude even though they are also less well-defined and more probabilistic.

And those are the three measures of forces on the direction of commodity prices: direction, magnitude and probability. When each component of the changing market is estimated, then the aggregate pressure and probability emerges as the most likely direction of the market. This kind of directional thesis, though vague, can bend or even pivot the strategy for the portfolio.

Creating the evolving strategy does require cash flow analysis, and cash flow analysis does require precise inputs for future prices. And thus we come full circle. The need overrides the limitations.

Some have tried probabilistic forecasting of prices, but I've found them more than problematic to deploy. More importantly, I have yet to see a tool for creating probabilistic price forecasts which offered the same kind of insight as a probabilistic analysis of engineering forecasts. Instead, in my experience, scenario analysis offers the best utility for strategy-making.

Precision looks like accuracy, but it is not. Confidence looks like expertise, but it is not.

Precision looks like accuracy, but it is not. Confidence looks like expertise, but it is not. Imprecision and uncertainty can, in fact, signal mature and nuanced thinking about the issues, but it must arrive on the backside of detailed, thorough, circumspect analysis. Invoking uncertainty or even unknow-ability is often an expedient, ad hoc justification for lack of discipline. And while it may salve an ego after the fact, it certainly won't spare anyone the consequences of mistakes.

The best that can be done, as far as I can tell, is to watch and study the components of the market and track the magnitude, direction and probability of the forces in total, not in isolation. Then to place that understanding into the larger context of group thinking and historical behaviors. When the pressures are lopsided, then may be the time for scenario analysis and then for action.

How does this paradigm compare with your experience? Please comment below.

In part three of this series we will talk about my own successes and failures of this style of price forecasting over the last decade. You can watch for that next installment on my LinkedIN feed or in my newsletter.