Shell Petroleum Engineering Seminar Series
"The Value of Assessing Uncertainty in Petroleum Exploration and Production"
by Duane McVay, Ph.D.
Duane McVay, Ph.D. is the Rob L. Adams ’40 Professor in the Department of Petroleum Engineering at Texas A&M University, College Station, Texas. Distinguished Member of SPE.
Abstract: Despite the perception of lucrative earnings in the petroleum industry, industry performance has been routinely below expectations. The underperformance is generally attributed to poor project evaluation and selection due to chronic bias. While authors have investigated cognitive biases in oil and gas project evaluation, there have been few quantitative studies of the impact of biases on economic performance. Incomplete investigation and possible underestimation of the impact of biases in project evaluation and selection are likely at least partially responsible for persistence of these biases.
In this presentation I will first review the problem of assessing uncertainty and some of the consequences of underestimating uncertainty. I will then present a recently developed framework for assessing the monetary impact of overconfidence bias and directional bias (i.e., optimism or pessimism) on portfolio performance. For moderate amounts of overconfidence and optimism, expected disappointment was 30-35% of estimated NPV for the industry portfolios and optimization cases we analyzed. Greater degrees of overconfidence and optimism resulted in expected disappointments approaching 100% of estimated NPV. Comparison of modeling results with industry performance in the 1990s indicates that these greater degrees of overconfidence and optimism have been experienced in the industry.
The value of reliably quantifying uncertainty is reducing or eliminating expected disappointment (realized NPV substantially less than estimated NPV) and expected decision error (selecting the wrong projects). Expected disappointment and decision error can be reduced by focusing primarily on reduction of overconfidence, which can be achieved through lookbacks and calibration of probabilistic forecasts.