Let the Buyer Beware When Benchmarking Oil & Gas Industry

Abstract: The oil and gas industry is obsessed with numbers for benchmarking purposes. Ironically, many of the numbers don't add up.

Analysis: The oil and gas industry is focused on numbers. It is good business management, and good science. It reflects the engineering background that dominates the industry. Numbers are used to benchmark individual company performance and they are used for insight into industry trends.

In an unpredictable and volatile industry, it is comforting to identify trend lines as markers for making sense out of the violent oscillations in drilling activity that seem to occur every few years.

There are a variety of data sets out there ranging from nonproprietary numbers published for free, such as rig counts and surveys of the drilling fleet, all the way up to proprietary services that specialize in tracking activity on a regional or national basis.

Customers pretty much get what they pay for. That explains the dichotomy where several industry data sets never quite add up when cross-checked against each other.

The most ubiquitous industry benchmark is the Baker Hughes survey of rig activity. This has been published continuously since 1944 and was expanded beyond its U.S. land base to include offshore and global activity over the last 20 years. The count, conducted by the company's sales force, is the most frequent benchmark employed in the oil and gas kingdom and represents the dean of industry statistics.

The Reed-Hycalog Rig Census was the second most frequently cited benchmark until its discontinuance as a cost-cutting move by Schlumberger. With its final presentation in 2001, this survey was the centerpiece of the International Association of Drilling Contractors annual meeting and represented a continuing history of changes in the U.S. land fleet dating back to 1952.

One topic the industry seldom discusses, however, is the discrepancy between these two measures. These discrepancies were most evident in 1997 and 2001.

In 1997, oil and gas operators were waiting six months for a land rig, contractors were experiencing drill pipe shortages, wages rose in an attempt to attract new employees, and the industry by all accounts was operating at full throttle. Comparing the Baker Hughes summary of rig activity at the peak in September 1997 against the Reed fleet census of 1,428 land units produced a utilization level well under 70 percent.

Then came 2001. Waits on U.S. land rigs extended nearly a year in some markets. Day rates for land rigs achieved replacement cost economics that summer for the first time in 20 years, indicating a demand-driven/supply-constrained market. Still, utilization in the U.S. land market, using the Baker Hughes/Reed Tool census parameters, was just 75 percent.

It did not compute.

To understand the reasons for the discrepancy, it is important to consider the disqualifiers each data set employs. For Baker Hughes, it means eliminating most of the shallower rigs in the market at a minimum. Since the company publishes only a lump-sum number, the rig count cannot be directly compared to the more complete proprietary counts out there and lacks transparency.

For Reed Tool, the disqualifiers involved capital investment hurdles and activity. If a rig had not worked in three years, or required $100,000 in capital expenditures to be made work-ready, it was not included in the census.

Those disqualifiers distort results. For example, if a rig needed half a string of drill pipe to be made work-ready, well, it was not part of the U.S. fleet under Reed disqualifiers. Nabors, Patterson (now Patterson UTI), and Grey Wolf together literally had more than 200 land rigs in yards in 2001 that did not exist according to the Reed survey.

Similarly, Baker Hughes disqualifiers suggest that a significant portion of coalbed methane activity in the U.S. is not represented in the rig count because the rigs do not meet the company's website definition of being significant consumers of oilfield-related products.

Coalbed methane has been the fastest expanding gas play in the U.S. land market over the last five years.

This is not to knock the nonproprietary counts. Rather, it is to point out that traditional benchmarks in the industry demonstrably understate activity levels in all phases of the oil and gas cycle, underestimate U.S. land fleet capacity, and provide an incomplete view of market dynamics.

Cross-checking data shows why. Last year, the Land Rig Newsletter picked up the telephone and worked through a list of U.S. land contractors. In two months, it was possible to itemize more than 1,940 individual rig descriptions in all classes, roughly 450 more units than showed up in the final Reed Tool fleet census in 2001. Not all the equipment is work-ready, but it is equipment that exists and can be made work-ready if the market permits.

Similarly, relying on proprietary activity counts such as Fort Worth-based RigData, it is possible to itemize land rig activity at levels more than 20 percent greater than found in the Baker Hughes counts. These are rigs on location--RigData provides driving directions on where to find the units--that are post-spud, pre-completion. They are earning a day rate. The data comes from permits and from drilling contractors.

Examining that data in detail, it is possible to identify more than 1,590 rigs that drilled one or more wells in 2001, 140 more land rigs than existed in the 2001 Reed Tool census.

Does this matter? Not if the casual observer takes things with a grain of salt. On a relative basis, the nonproprietary benchmarks indicate trend direction over time. Think of them as a weather vane showing the direction the wind is blowing. But the weather vane may not tell much else about the weather.

As a result, the oil and gas industry stumbles into flawed assumptions based on these incomplete benchmarks.

It is common in investor relations presentations to see service companies using indices based on Baker Hughes rig counts as measures of efficiency. The slides typically show higher revenues per rig for an individual company at cyclical peaks than is the case during other points of the cycle. On the other hand, those measures may be leaving out a fifth of the nation's rig activity. It is equivalent to turning in a great time for the 400-meter dash based on how the runner did on a 300-meter course.

Similarly, it is common to see inventories of manufactured goods compared to Baker Hughes rig counts as a way of measuring a sector's capacity utilization. Drill pipe per rig is a frequent example. Once again, if actual activity is higher than the Baker Hughes count, drill pipe inventory will be tighter than those indices indicate.

There are also issues of efficiency. Analysts measure efficiency per rig by comparing industry footage to the Baker Hughes count. Industry footage is the sum of all drilling; the Baker Hughes count is a summary of part of the market. This distorts the results. A similar argument can be applied to gas production versus Baker Hughes gas-directed rig count, which seems to be a popular analysis these days.

It is also possible to find situations where financial analysts take a publicly held company's published rig activity level and compare it to the Baker Hughes count, thus overstating that drilling company's share of the market.

When it comes to nonproprietary benchmarks in oil and gas, the motto is caveat emptor, regardless of history. To truly understand the oil and gas industry, it is important to look beyond the face value represented by the traditional number sets.