Forecasting US Natural Gas Prices

Another challenge in building a model is to forecast natural gas prices.  Natural gas prices behave differently than oil prices because different factors drive consumption.  Although there is some substitution of crude oil (in the form of fuel oil) for natural gas, the markets are not closely connected.  Natural gas consumption is mostly seasonal, with consumption in winter while production is more or less constant throughout the year.  Thus, the largest factors driving US natural gas pricing are seasonality, storage inventories, winter temperatures, and production disruptions such as hurricanes.  In the US market, since gas is not imported or exported to a large degree, gas pricing is not as subject to the geopolitical disruptions that affect oil prices.

Not as much research has been published on forecasting natural gas prices as on oil prices.  But the work that has been done suggests that the futures prices are the best indicator, albeit only a weak one.

It is instructive to compare a graph of the historical Henry Hub price with the current futures curve.  Notice that there is a definite element of seasonality to the historical prices.  But the seasonality element has been overwhelmed by price shocks in many years.

 Source:  http://www.eia.gov/dnav/ng/hist/rngwhhdM.htm

The graph of the futures curve below shows that traders expect a dominant seasonal element in every year, except for the current 2014-2015 winter, when it has been overwhelmed by the growing production of tight gas and the drop in crude prices.  And notice how regular and beautiful the future curve is compared to the reality that we have experienced over the last two decades.  This comparison shows why the futures market has only a low predictive power.

Source:  Data from http://www.cmegroup.com/trading/energy/natural-gas/natural-gas.html on 9 Jan 2015.

Several comments on gas prices.  First, gas is bought and sold in the spot and futures markets in MMBtu (millions of British thermal units), but production is measured in Mcf (thousands of cubic feet), and producers are paid for Mcf.  An average conversion between the two is that 1 Mcf = 1 MMBtu, but this can vary greatly.  Before a field is producing, it can be difficult or impossible to get a reliable energy content factor.

Second, the US gas market still tends to be regional, so it is important to use the sales prices in the region where the gas will be sold.  There are many spot prices available, but the only widely available futures prices are Henry Hub, for gas to be delivered at the Henry distribution hub near Erath, La.  The Chicago Mercantile Exchange trades basis differential futures that can be used to forecast the price difference between Henry Hub and a number of other delivery points.

Finally, because gas is not as easily transportable as oil, the dynamics driving international markets are regional and can be very different than those driving US markets.  If your gas is not priced at the Henry Hub, YMMV.

Conclusion.  In contrast to best practice for oil prices, for natural gas we need to use the futures curve to get the best forecast.  But similar to oil prices, the best forecast that we have available does not have great predictive power.  So we need to present several forecasts using a range of options to guide decision-making.  See the discussion in the final paragraph of the post on oil prices for more details.

Here's an interesting post on the Economist blog about the four factors that are driving the current drop in oil prices.  

Forecasting Oil Prices

A little humility is in order when it comes to forecasting hydrocarbon prices.  Despite the confident talking heads on television, newsletters and e-mails that promise to forecast prices accurately, and prognostications from government agencies, no one knows what oil and gas prices are going to do.  And if anyone did know, he certainly would not be talking. 

As a result, prices are usually the least predictable element in the economic forecast; and therefore, they usually have the greatest impact on returns.  What is an analyst to do?

The academic literature gives some guidance on the subject.  First, it is important to note that oil and natural gas prices behave differently.  In fact, this should not be a surprise since, even though they occur together in nature, oil and natural gas have very different uses and are mostly not substitutes for one another.  So we will tackle oil prices first.

Second, forecasting methods that had the most predictive power in the 1980s and 1990s have not continued to be the best over the last ten years.  It appears that the factors that drive the price of oil change over time.  So even if we apply the findings of academic studies from the last decade, their conclusions may not be applicable over the next decade, which would be the time frame when our wells would be producing.

Recent studies have found that at time frames of 1-3 months, adjusted futures prices have the most predictive power.  But in new investments, this is almost always the preparation period when agreements are made with governments and landowners, engineering is done, and equipment is mobilized.  So futures markets are of little use in preparing price forecasts for project acquisitions, but they could be useful for analyses of producing fields.  And they should be more useful for fields that decline rapidly, such as shale.

For time frames of 12 months and longer, today’s spot price of oil has the greatest predictive power, although it is a fairly weak predictor.  If I had to guess the price of oil a year from now, I would guess today’s spot price, although it very likely could be something else.

How should we tackle a critical problem where the data we can observe provide weak guidance?  One common, although particularly bad practice, is to take today’s spot price and escalate it at some rate of inflation until it reaches a ceiling.  There is no evidence that oil prices tend to escalate in this manner.  Rather it represents the industry’s dream that oil will again reach its “natural level” of $140 and stay there.

A good practice is to present a base case with a constant-price forecast of today’s spot price.  (If a large portion of the production will come in the next 3-6 months, consider using an adjusted futures forecast for those months.)  I then present several alternative cases with a range of prices.  If a client is using debt-financing for the investment, I would focus attention on the worst-case scenarios to help them think about whether they can survive a sharp drop in oil prices.  And I compare the spot price to the range of historical prices to help the client visualize how much potential upside and downside there is.  The best you can do is to provide a number of price scenarios and think about the likely effects of each one on the investment.

Next time, gas prices.

Reserve Report Dirty Secrets

Do you ever read the introductory pages to a reserve report?  You should – because virtually every report contains some shocking wording.  Here is one from a report by a respected, certified reserves estimator:

“The reserve estimates were based on interpretations of factual data furnished by [Client]. Oil and gas prices, pricing differentials, expense data, capital investments, plug and abandonment costs, tax values were supplied by [Client]. Ownership interests were supplied by [Client] and were accepted as furnished.”

Then he adds, “To some extent, information from public records has been used to check and/or supplement this data.”  But this sentence just highlights the problem – even a conscientious reserves estimator has limited public data for a reasonableness check.  It would be cost-prohibitive to test most of the data supplied by the client.  Even worse, not every reserves estimator is conscientious.  The dirty secret – even independent reserves reports are of necessity built on a foundation that is sometimes nothing but sand.

The US Securities and Exchange Commission (SEC) has tried to alleviate this problem by regulating the parameters for a reserves report used by a publicly-listed company in the US.  To a certain extent, this has been helpful, especially by providing clarity and consistency on pricing and by establishing clear parameters for the physical limits of the reservoir.  But while providing clarity, sometimes these SEC-mandated assumptions are far from realistic.  And there is still a lot of wiggle room.  Even so, it is a good start if you see that a reserve report was based on SEC assumptions.

In reaction to the SEC requirements, the Society for Petroleum Engineers, in conjunction with several other groups, has released its Petroleum Resources Management System.  Conceptually they are a great improvement over the one-size-fits-all approach of the SEC because they have more flexibility.  But on the downside, they have more flexibility than the SEC mandates.  So the user of an SPE reserves report cannot know whether the report is really better or worse than an SEC reserves report unless he digs into the details.

Banks that lend on oil and gas reserves often provide their own assumptions in creating reserve reports for bank use.  But banks are notoriously reactive – when prices are low and assets are cheap, the banks usually use tight assumptions that make almost any project uneconomic.  When prices are high and times are good, banks loosen their assumptions and lend on anything.  At least that is the oil patch lore, and there seems to be a lot of truth to it.  So the impression is left that bank reserve reports are more geared to protecting the hide of the energy lending department rather than coming up with a real valuation.

What is a creator or user of reserves reports and economic analyses to do?  Fortunately, there is a lot that can be done to unpack the assumptions used and to determine if they are realistic.  In following blog posts, we will turn our spotlight on the more common problems.  Read on and you will become an expert user of reserves and economics reports.