Do Weekly Gas Storage Estimates Understate Actual Levels?

Abstract: Natural gas prices, rig counts, and blood pressure on Wall Street rise and fall each Thursday when the U.S. Department of Energy reports natural gas storage estimates. It now appears those weekly estimates may understate total storage levels.

Analysis: Oil and gas is an industry obsessed with numbers. There are rig counts, production numbers, and financial performance to list the most obvious. And it gets progressively more complex from there.

On closer examination, many of those numbers are never quite what they seem.

Things get interesting when those numbers are taken at face value, incorporated into analyses, which are then woven into various decision-making processes. Considering that some of those decisions involve millions of dollars, it behooves the industry to pay attention to what the available data really represents.

The latest numbers to come under question are U.S. Department of Energy estimates on weekly gas storage. Over the last few weeks, industry analysts, primarily in the financial community, have complained that the monthly sum of the reported weekly storage number does not equal the storage number later reported for the month in the U.S. DOE Natural Gas Monthly (NGM).

In other words, there is more gas in storage than appears to be the case from the weekly storage estimates.

And that discrepancy has repercussions for gas prices, industry cash flow, and ultimately reinvestment in field work.

The reasons for the disagreement are somewhat complex and can be found in the 14-plus-page appendices that accompany the DOE's Natural Gas Monthly. Further detail on methodology for weekly storage estimates is available from the DOE's Natural Gas Weekly Storage report .

Essentially, it boils down to the fact that weekly storage estimates are derived from a sampling of 45 representative companies. The monthly numbers, in contrast, are compiled from the comprehensive filings of all 120 companies involved in gas storage. The DOE has always been upfront about the nature of the weekly gas storage number. The agency bills it as a three-part estimate involving two consuming regions and one producing region. The agency shoots for a 95 percent certainty range in its methodology.

The financial industry and, to a lesser extent, the oil and gas industry often take these weekly estimates at face value because of a need for timely data. This is one repercussion from the fact that the natural gas storage number has evolved into the default proxy for the robustness of the natural gas industry.

A half a dozen years ago it was rare to find anyone beyond industry aficionados who could explain what the gas storage number was, or even where to find it. Now even the smallest contractors in remote areas of the United States can speak authoritatively on the natural gas storage number and its implications.

Hard to believe that the number that has become the proxy for the relative health of the natural gas industry has such short history. The first weekly storage estimates date to late 1993. While the U.S. government had been tracking gas storage on a monthly basis, there was a substantial lag time built into the reporting process. Interest grew in a more timely presentation of storage data following natural gas market restructuring in the early 1990s.

Initially, the American Gas Association began reporting weekly storage numbers in 1993 based on a sampling of member companies. The AGA abandoned the survey in 2002 because the program taxed the organization's resources. That left the DOE as the default compiler, and in May 2002 the U.S. DOE took over the weekly storage estimate from the AGA.

Historically, there was not a great deal of difference at the end of the refill season between AGA and DOE numbers. On the other hand, the AGA tended to show lower working gas in storage at the end of winter, but documented a greater refill rate during injection season when compared to DOE numbers.

And those trends have come into play now. After a series of comments in the energy trade press, the U.S. DOE took the unusual step last week of announcing it was not changing methodology for its weekly storage estimates, and had no plans to do so.

At the same time, a DOE spokesman acknowledged that storage is likely higher than its weekly estimates show.

What looked like a potential crisis in June to Federal Reserve Board chairman Alan Greenspan may now turn out in October to be adequate winter supplies, barring an unusually cold winter in energy-consuming regions.

This is the second time number revisions have altered the energy picture this year. In fact, last winter turned out to be colder than previously thought because of another statistical change involving Heating Degree Days (HDD) as calculated by the National Oceanic and Atmospheric Administration. NOAA this summer announced it had revised last winter's figure up five percent.

It was a quiet event news-wise, but the implications were important. What looked initially as though it would be a significant jump in gas consumption on an HDD basis turned out actually to be consumption at normal levels. As a result, theoretical winter storage levels remained essentially unchanged.

The DOE's acknowledgment last week that storage may be higher than recent estimates also casts doubts on the thesis the industry will come up short on natural gas this year.

Before the finger-pointing starts, it is important to remember that collecting oilpatch data is not an easy process. There are many ways to climb the data collection mountain. For example, the DOE annually collects natural gas data from more than 2,000 entities. The DOE's target is creating a body of accurate data. Given time, that can be done using the reporting process.

However, the natural gas industry--and now Wall Street--wants timely data. This necessitates sampling.

The DOE provides weekly estimates based on the sampling, then revises data upwards following the industry's monthly report. Furthermore, those numbers themselves can be revised upwards once again during the annual reporting process. The lag varies from three months for the monthly number into a year or better for the annual data. The annual data is the most accurate.

Meanwhile the industry--and now Wall Street--reacts in a reflexive, volatile manner to the weekly storage estimates, which are probably the least accurate of the data sets.

An example of how underreporting from the weekly sampling process can impact the overall gas market was evident 90 days ago when the prevailing consensus questioned whether the industry could win the race to replenish gas storage by October 31. The industry exited last winter with storage levels at historic lows. Futures prices climbed. There was talk of an impending natural gas crisis.

Now evidence grows weekly that the industry will reach normal storage levels barring some unusual event like a rush of hurricanes hurtling through the Gulf.

What does the current spat over the weekly number mean for the industry? It is important to remember that accuracy in oilpatch data sets improves given time. Secondly, an emphasis on timeliness produces data that is often less than definitive. At inflection points in the market, this can have significant repercussions for commodity pricing and field activity.

There is an analogy in classic literature. Plato, the Greek philosopher, wrote of humanity's plight in his vision of the future in his great work, The Republic. He pictured humanity in a deep, dark cave living from childhood as a mass of individuals chained by neck and leg so they cannot move, and seeing only what is in front of them. They face a wall. Behind them is a fire. And between their backs and the fire is a track with a parapet on it--a modern Plato would compare it to an underground "moving picture" show. The prisoners thereby see nothing of themselves or each other, except the shadows of the figures moving on the parapet, thrown by the firelight onto the wall. Thus, humanity sees only the shadow of reality and never reality itself.

So it can be with oilfield data. Caveat emptor.