Responsible Reporting of Petroleum Reserves

Gary P. Citron1, James A. MacKay2, Mark A. McLane2, James Gouveia3, and Peter R. Rose4. (1) Rose & Associates LLP, Suite 320, 4203 Yoakum, Houston, TX 77006, phone: 7135288422, fax: 7135288428, garycitron@roseassoc.com, (2) Rose & Associates, LLP, Houston, TX 77006, (3) Rose & Associates, LLP, Calgary, AB, Canada, (4) Rose & Associates, LLP, 3405 Glenview Avenue, Austin, TX 78703

Since exploration is a repeated trials effort associated with many uncertain ventures, a statistical treatment of the associated undiscovered resources is appropriate. However, when we contemplate the required reporting of “Proved Reserves” after a specific discovery, we are asked to specify a volume of recoverable hydrocarbons that we are “reasonably certain” will be recovered from a well associated with that discovery. The phrase “reasonably certain” is a probability statement, except that no confidence-level is specified. Company appraisers may be influenced that larger estimates (if defendable) benefit the value or their company shares and perhaps their status within a company; while various negative consequences may ensue if the “reasonably certain” estimate turns out to be larger than the actual outcome. We view this clash of probabilistic methods versus determinism as an illogical professional conundrum. Since deterministic parameters are not mathematically specified, a professional's estimating ability can not be properly measured and calibrated. This approach encourages unrealistic thinking about uncertain resource values and thus can facilitate technical and financial unaccountability. In fact, ill-defined standards can actually encourage unethical behavior through confusion and manipulation, obscuring boundaries between professional objectivity and conflicting incentive systems.

One simple remedy is to set a unified standard within the E&P community that “Proved” equals a specified 90% confidence of that minor amount, or more, reinforced with probabilistic methods that help measure estimating accuracy of forecast ranges versus actual outcomes, facilitate reality checking against analogs and natural limits, and foster improvements in future estimating accuracy and efficiency.