Managing scenario risk in offshore development projects
Offshore exploration and devel opment projects are inherently risky. Uncertainties of reservoir con ditions and performance, compounded by complexities of designing recovery processes, platforms, and equipment for offshore operations, focus attention on risks that arise from within the project.
"Scenario" risks arise from uncertainties outside individual projects. Oil and gas prices are major scenario uncertainties, as are changes in regulations, demand, technology, competitive conditions, and success on adjacent prospects. In addition to influencing individual projects, they tend to affect groups of projects (portfolios) in similar directions, compounding their impacts.
Many exploration and production (E&P) companies ignore scenario uncertainties as beyond their control. While these uncertainties cannot be controlled, selecting projects that offset one another can mitigate their effects. Investing in offsetting projects requires estimation of the influence of scenario uncertainties on individual projects and systematic combination of projects to exploit offsetting risks.
Risk spreading
Since the beginning of the oil industry, E&P managers and investors have diversified, "spreading their bets," by allocating capital across a variety of projects. The logic of diversification suggests that the odds of losing on any single project will be higher than the odds of losing on all of a number of projects.
In 1952, Harry Markowitz wrote a paper that revolutionized thinking about portfolios in the financial world (see Markowitz).
One of Markowitz's insights was that simply "spreading the bets" achieves only part of risk reduction. Additional reduction arises from understanding the interdependencies among the respective projects. If two projects are highly correlated (success in one depends on success on the other), the value of diversification is lost as both tend to move up or down together. But, if success on one project is associated with failure of the other, the odds of both failing are dramatically reduced - a "natural hedge" has been identified, reducing the risk of failure across the "portfolio" of two projects.
Effective risk management requires that projects selected be independent of one another or have offsetting responses to uncertain events. If independent, the full value of diversification is achieved. If negatively correlated, natural hedges further reduce the risk of loss.
In E&P, correlations arise from numerous conditions. For example, common source, migration, and/or seal, could affect all prospects in a given play. Or, depositional similarity among reservoirs could influence the effectiveness of certain technologies in all of them. The widely used notion of a "play" reveals sensitivity to this interdependence. Scenario uncertainties also contribute to positive or negative correlations. Such conditions include:
- Future prices and costs
- Changes in demand or the transportation/storage system
- Changes in technologies for exploration, production, and transportation
- Changes in regulations
- Changes in fiscal regime or pro duction-sharing terms
- Other public policies
Scenario uncertainties affect project viability in various ways that are distinctly different from prices, costs, recovery efficiency, timing, volume, or operating constraints. Scenario uncertainties often establish interdependencies that affect project and portfolio risk.
Changing regulations or fiscal regime could affect all projects in one jurisdiction, but not in others. A change in technology could benefit certain types of reservoirs relative to others. A new pipeline could radically shift demand from one region to another. As projects compete with one another, competition establishes interdependencies. The impact of these changes singly, and especially in combination, are difficult if not impossible to trace using conventional, single-reservoir analysis techniques. They are, however, the source of many of the inter dependencies that exacerbate or mitigate the risk.
Scenario risks and the inter dependences they pose have been ignored because the analytical methods for dealing with them are lacking. Combining scenario modeling with Markowitz portfolio analysis provides a way of selecting projects to manage these risks.
Scenario modeling
Since the mid-1970's, the US Department of Energy has sponsored a series of models to examine the response of US hydro carbon production to alternative federal oil and gas re search and development priorities.
To simulate industry's technology application decisions, production and economics are estimated by type-curves, stream-tubes, and money-forward cash flow models for nearly 25,000 US and Canadian prospects and reservoirs. These compete to meet demand through a network of pipelines and storage facilities under a variety of economic, policy, and technical constraints. The models have proved sufficiently credible to be used in National Petroleum Council studies on at least five occasions.
Competing portfolios
E&P decision-makers have long been advised to rank their project investment opportunities in decreasing order by the marginal return on capital (net present value divided by investment or NPV/I) and to fund the projects in descending order until available capital is committed. This is said to maximize the economic return. It does so - if, and only if, future conditions are as predicted. If not, disappointing results can follow. Markowitz-style portfolio evaluation based on scenario modeling of projects in their full context provides an alternative that explicitly recognizes the uncertainties in the scenario context and the interactions among projects.
For example, 14 projects were evaluated by the con ventional approach and by the Markowitz-scenario approach. The example assumes complete certainty about reservoir and recovery factors to isolate the effects of scenario uncertainties. Three simple scenarios were constructed using four major uncertainties (see table) to drive a range of wellhead gas prices:
- Whether or not new pipelines are built from Canada to the US
- Various stringency levels in US environmental regulations
- Various outcomes of the competition of coal and gas in power generation
- Variations in market penetration of advanced recovery technologies.
The three scenarios were run for the full US and Canadian natural gas resource base.
Cursory examination of the NPV/I for the 14 projects (see table) shows that seven of the projects responded as expected under the respective scenarios. The other seven show unexpected response to the scenarios.
Three scenarios were weighted equally in the project ranking. Following the conventional decision rule, those with highest NPV/I were selected until the capital budget was exhausted, resulting in partial funding of the last-selected project. The projects ranked 1 through 6 were fully funded and the seventh-ranked project was funded 66%. Because this selection accepts only the "best" projects, it is referred to as the "cherry-picking" solution.
A portfolio optimization was conducted for the same group of projects, resulting in an "efficient frontier," the locus of portfolios (project combinations) for which no lower risk combination is available with as great a return and no combination with a higher return is available without higher risk.
In this case, risk was indicated by variance. The efficient frontier is the curve of optimal risk-return trade-offs from which a decision-maker can select, according to tolerance for risk. Individual projects can be accepted in full or in part (as in taking on partners).
The conventional decision rule does maximize return as expected, but in ignoring uncertainty, it also maximizes risk (see accompanying figure). When full uncertainty is recognized, the conventionally selected "cherry pick" portfolio typically lies above and to the "northwest" of the efficient frontier, indicating that it included unnecessary and uncompensated risk. There are other combinations of projects with as high returns, but lower risks. If the decision-maker was unwilling to maximize risk, the curve defines all other optimal portfolios available at that budget level.
Recognizing interdependencies
The previous example assumed the respective projects were independent of one another, as is usual. But, by design, the uncertainties are caused exclusively by scenario factors that affect the full market. If a significant number of projects are being considered for the portfolio, only a scenario model can track the effects of these uncertainties, and the interdependencies they cause, across projects as they compete with one another and all the other projects in their markets.
Recognition of interdependency can shift the frontier up or down at a given level of return. The direction is determined by the correlation of the projects. To illustrate this, the portfolio example was re-calculated recognizing the previously ignored interdependencies among the projects. This shifts the efficient frontier upward, indicating higher risk for the same return, and larger positive than negative correlations. Positive correlations are more likely in an integrated market like North American natural gas.
Project selections
Recognizing the risks due to scenario uncertainties and the correlations they impose causes a direct and significant shift in the projects selected for the portfolio. If the Markowitz-style efficient frontier was used to reduce risk and return, but interdependencies were ignored, the resulting project selections shift.
Project investment (f) is reduced significantly to increase funding of Project (c) and to add portions of previously excluded Projects (d) and (m). The resulting portfolio has greater diversification, so lower risk in the face of uncertain scenarios. Most importantly, risk is explicitly considered in the decision and an optimal choice is made.
Recognizing interdependences further shifts the investment choices. If the decision-maker maintains the expected value selected, Project (d) is eliminated, but additional investments are made in Project (c), and new investments are added in Projects (a) and (b). Even though these last projects ranked 12th and 11th, respectively, their inclusion reduced the overall risk of the portfolio because they were inversely correlated with other selected projects.
Recognition of interdependencies among projects due to scenario risk then results in more accurate appraisal of risk and selection of a more robust portfolio. Explicit incorporation of these interdependencies suggests the strategy of actively searching for or designing negatively correlated investment projects to build financial or natural "hedges" into the final portfolio.
Conclusions
Explicit analysis of scenario uncertainties and interdependencies among projects in selecting portfolios of E&P projects improves the quality of investment decision-making. The example shows that:
(1) Conventional E&P evaluation processes can substantially increase (or underestimate) the risk of projects and portfolios by:
- Ignoring many scenario risks
- Assuming price variations affect projects uniformly when they may not
- Highgrading ("cherry-picking") projects without regard to risk, thus maximizing portfolio risk
- Ignoring interdependencies among assets due to scenario uncertainties, thus losing access to "natural hedges" of negative correlations, while underestimating risks due to positive correlations.
(2) Systematic analysis of scenario risks in the context of stochastic project evaluation and portfolio optimization:
- Quantifies the influence of uncertain scenario factors on specific projects
- Evaluates the factors that cause price risk to specific projects and on portfolios to guide project design and contingency management
- Incorporates these risks into estimates of portfolio risk in the risk-reward trade-off
- Identifies and exploits risk reduction due to "natural hedges" while avoiding unnecessary risks due to positive correlation.
(3) Significant competitive advantage is available to E&P companies that systematically recognize scenario risks and interdependencies in their investment decision-making.
Acknowledgment
The authors recognize the U.S. DOE (National Petroleum Technology Office, Federal Energy Technology Center, and Office of Natural Gas and Petroleum Technology) for sponsorship of the models described. Views expressed are the authors'. The example described was first reported in SPE 56574, SPE Annual Technical Conference, October, 1999, and is reprinted with permission.
References
Markowitz, H., "Portfolio Selection" Journal of Finance, Vol. VII, No. 1, 1952.
National Petroleum Council: Enhanced Oil Recovery, 1978; Enhanced Oil Recovery, 1984; Unconventional Gas Sources, 1980; Potential for Natural Gas in the United States, 1992; and Marginal Wells, 1994, Washington, D.C.