Portfolio Management in an Upstream Oil \u0026 Gas Organization

$Page 1: Portfolio Management in an Upstream Oil \u0026 Gas Organization$
Copyright � 1999 INFORMS0092-2102/99/2906/0084/$05.001526–551X electronic ISSN

ORGANIZATIONAL STUDIES—STRATEGYINDUSTRIES—PETROLEUM-NATURAL GAS

This paper was refereed.

INTERFACES 29: 6 November–December 1999 (pp. 84–104)

Portfolio Management in an Upstream Oil andGas Organization

Mazen A. Skaf Navigant Consulting, Inc.Strategy Consulting Practice (formerly StrategicDecisions Group)

2440 Sand Hill RoadMenlo Park, California 94025

Implementing portfolio management in an organization pres-ents organizational and analytical challenges. A successful im-plementation requires an integrated solution for both the peo-ple aspects and the analytical aspects of the problem. StrategicDecisions Group (now Navigant Consulting, Strategy Consult-ing Practice) developed and implemented a portfolio-management process and system for a client organization inthe upstream oil and gas industry. The tailored process pro-vides a forum for decision-focused dialogue between seniormanagement and asset teams. The system provides the analyti-cal support for the process and enables management to com-pare decisions and assets across the portfolio and determineoptimal allocation of resources. The system architecture buildson a rigorous model of the asset life cycle and the key deci-sions in the life cycle. To further integrate portfolio manage-ment into its business processes, the organization set up an in-ternal core team to facilitate the process and work with theasset teams on an ongoing basis. This has helped reduce thetime for developing regional and business-unit portfolio strate-gies. The value added from strategic alternatives developedusing the method is in the hundreds of millions of dollars.

PORTFOLIO MANAGEMENT

November–December 1999 85

Implementing portfolio management inan organization requires a process that

engages the expertise of the various stake-holders in the organization and a systemto provide the analytical support for theprocess. A well-thought-out process tai-lored for the organization takes care of thepeople issues and ensures buy-in for theselected portfolio strategy. A transparentand validated system that incorporates theknowledge of the various experts in theorganization and models the portfolio de-cisions and uncertainties solves the analyt-ical part of the problem. A successful im-plementation requires an integratedsolution for both the people aspects andthe analytical aspects of the problem.

For a client company in the upstream oiland gas industry, Strategic DecisionsGroup (now Navigant Consulting, Strat-egy Consulting Practice) developed andsuccessfully implemented a portfolio man-agement process and system. The overallportfolio management approach and keyelements of the system are easily transfer-able to other industries and organizations.Overview of the Portfolio ManagementApproach: Portfolios andInterdependencies

A portfolio is a collection of entitiesamong which there may be several inter-dependencies. An entity can be a singlephysical asset (such as a producing field inthe oil and gas industry or a manufactur-ing plant in another industry), a businessunit (which may include several assets), ora financial security. An interdependencymay exist between any two entities in theportfolio and may affect specific perfor-mance measures of the portfolio. Interde-pendencies among entities in a portfolio

can be informational, or they can be physi-cal or operational.

The concept of informational interde-pendency between two entities builds onthe concept of relevance between two un-certainties. An informational interdepen-dency exists between entities A and B inthe form of relevance when at least oneuncertainty about entity A is relevant to atleast one uncertainty about entity B (Fig-ure 1). This includes the special case inwhich the expert or decision maker be-lieves that entities A and B share a specificuncertainty. (An uncertainty Ai is relevantto an uncertainty Bj if knowing the valueor outcome of uncertainty Ai, we wouldassign a different probability distributionto uncertainty Bj [Howard 1990]). For ex-ample, consider two prospects A and B inan unproven play (a play is a group ofprospects and any related fields havingcommon hydrocarbon sources, migrationrelationships, reservoir formations, seals,and trap types [White 1992]). If the uncer-tainty about geologic success for prospectA is relevant to the uncertainty about geo-logic success for prospect B, an informa-tional interdependency exists betweenprospect A and prospect B. Many forms ofrisk correlation among entities would thusfall under this category of interdepen-dency. Other instances of informational in-terdependency are opportunities for learn-ing, process improvement, and the like.

The category of physical or operationalinterdependencies includes the use orsharing of the same physical resources,competencies, or skills. Examples of physi-cal interdependencies are a producingfield tied to a specific processing facilityfor another field or assets competing for

SKAF

INTERFACES 29:6 86

Figure 1: The concept of informational interdependency between two entities builds on theconcept of relevance between two uncertainties. An informational interdependency exists be-tween two entities, A and B, in the form of relevance, when at least one uncertainty about en-tity A is relevant to at least one uncertainty about entity B. (This includes the special case inwhich the expert or decision maker believes that entities A and B share a specific uncertainty.)The bold arrow in the upper part of the figure represents the informational interdependencybetween the two entities. In the lower part of the figure, relevance diagrams show the uncer-tainties driving the value (or any other measure) of each entity and possible relevances amongthem.

the same capital and human resources.Portfolio Management

In this approach to portfolio manage-ment, we focus on interdependenciesamong entities in the portfolio to developstrategies that improve the performance ofthe portfolio as a whole. Interdependen-cies bring relevance to the concept of port-folio management. Otherwise, in the ab-sence of interdependencies, we could treateach entity separately and not be con-cerned with portfolio management.

Portfolio Management at Two MainLevels in an Upstream Organization

Portfolio management can be applied attwo main levels of portfolio decisions inan upstream organization: the corporateportfolio of business units and the busi-ness unit portfolio of assets and projects(Figure 2). The corporate portfolio consistsof all the business units in the organiza-tion and any assets that do not fall in anyspecific business unit. Each business unitmay have its own portfolio of assets or



Figure 2: A corporate portfolio consists of business units and other entities that are not in anybusiness unit. A business unit’s portfolio of assets may include leased prospects, discoveries,and producing assets.

projects. An entity in a business unit port-folio is thus a single asset or project.Sometimes, an upstream oil and gas cor-poration may organize a business unitaround a single asset, in which case thebusiness unit portfolio consists of a singleasset.

The first level of portfolio managementdecisions in an organization is the corpo-rate portfolio strategy: What businessesdoes the organization want to be in, andhow much does it want to invest in eachbusiness? Examples of decisions at thislevel include the acquisition of new busi-nesses, investments in existing businessunits, mergers with other organizations,and divestitures.

Given a chosen corporate portfolio strat-egy, the next level of portfolio manage-ment is within each business unit: Whatopportunities or assets does the businessunit want in its portfolio, and how can itbest allocate resources across these assets?Examples of decisions at this level includethe acquisition of new assets, investmentsacross assets, and the timing of assetinvestments.

Portfolio management decisions fallwithin the hierarchy of decisions in an up-stream organization (Figure 3). At the topof the decision hierarchy are the corporatemission, vision, and values. They are whatdistinguish the organization from others,and they are taken as given when discuss-

SKAF

INTERFACES 29:6 88

Figure 3: The two main levels of portfolio management decisions fall within the context of thedecision hierarchy of an upstream organization.

ing portfolio management. At the nextlevel down are the corporate portfoliostrategy decisions. At the next level arestrategy decisions for the business unitportfolio. Each level of portfolio decisionsprovides scope and direction for lower de-cisions and is fully developed through thedecisions made at the next lower level. Forinstance, the corporate portfolio strategy isfully developed as the collection of strate-gies for the business units in that corpo-rate portfolio. Similarly, the portfolio strat-egy for a specific business unit is fullydeveloped as the collection of asset strate-gies (project strategies) for all the assets(projects) in that business unit’s portfolio.The Business Unit and Its Portfolio

Business units of major oil and gas com-panies operating in the Gulf of Mexico(Figure 4) oversee large portfolios of assetsand manage these assets through explora-tion, development, and production. For

simplicity and to protect the confidential-ity of my client, I describe the needs of atypical business unit in the Gulf of Mexicoand refer to it as the Gulf of Mexico Busi-ness Unit (GMBU). The GMBU formed as-set teams to manage the assets in its port-folio. Some teams were geographicallybased while others focused on assets thatshared a key characteristic.

The Gulf of Mexico has been the maingrowth area for exploration and develop-ment for several oil and gas companies inthe United States. Companies lease blocksfrom the Minerals Management Service(MMS) through participating in lease salesadministered by the MMS. Prospects mayextend over several blocks. Leaseholdersmust drill prospects before their leases ex-pire (most have 10-year terms); discoveriescan then be held for production beyondthe lease expiration date.

The GMBU portfolio consisted of many



Figure 4: Map of the Gulf of Mexico showing areas of operations for the Gulf of Mexico Busi-ness Unit.

assets distributed across different stages ofan asset life cycle (leads, leased prospects,risked prospects deemed ready to bedrilled, discoveries, and producing assets).Managing this growing portfolio of oppor-tunities while faced with resource con-straints (rigs, staff, capital, and so on) andlease expiries was becoming complex. TheGMBU management team engaged Strate-gic Decisions Group (SDG) to develop aportfolio management system and processthat would facilitate the allocation of re-sources across the portfolio, provide in-sight into portfolio-wide issues, and beeasy to use and update. The system was tohelp the management team and the assetteams compare decisions and assets acrossthe portfolio, determine optimum alloca-tions of resources, and evaluate current re-source levels and support resource plan-ning. Further, the system was to provideinsights into such portfolio-wide issues as

leasing strategy, portfolio balancing, andtechnology investments. The system andprocess were to be transferred to an inter-nal portfolio management team after train-ing and testing.

We embarked on a six-month effort todevelop the system and process, workingclosely with a core team of experts in theorganization. During the first part of theproject, we concentrated on developingthe system specifications and designingthe system architecture. Throughout theproject, we engaged the GMBU manage-ment team at key decision and progressreview points to ensure that we stayed ontrack to meet the project objectives.The System Architecture

The first step in developing the systemarchitecture was to establish what deci-sions and analyses the GMBU would ad-dress using the system. This determinesthe scope of the system and the level of

SKAF

INTERFACES 29:6 90

detail required. Working with the coreteam of experts and interacting with themanagement team, we established that thesystem should support the following deci-sions and analyses: (1) evaluation of port-folio strategic alternatives, (2) evaluationof various leasing alternatives and deci-sions on allocations to each year’s leasesale, (3) risking and exploratory drillingdecisions, such as deciding where to drill,establishing drilling priority, and deter-mining how many rigs to have (risking isan industry term referring to the processof seismic imaging, analysis, and interpre-tation for the purpose of deciding whethera prospect should progress to the explora-tory drilling stage), (4) evaluation of de-velopment options from a portfolio per-spective, evaluation of differentdevelopment configurations, and deter-mining the timing for developing pros-pects, and (5) determining needed levelsof resources, such as staff, rigs, andcapital.

After establishing what decisions thesystem should support, we delineated thestages in an asset life cycle and the asset-specific decisions the system should in-clude. We had several discussions on thework process in the organization and onasset stages. We decided to delineatestages in an asset life cycle according tothe following criteria: (1) there is uncer-tainty about whether the asset will pro-gress beyond that stage to a subsequentstage towards becoming a producing as-set, and (2) the stage precedes a phase inthe work process involving a major alloca-tion of resources.

Figure 5 shows the different steps in thework process (at the bottom), a stage-to-

stage asset transition diagram (in the mid-dle), and the set of key decisions in an as-set life cycle (at the top). An asset startsout as a lead, with the end-success stagebeing a producing asset. The stages arelead, viable scoped lead, leased prospect,drill-worthy prospect, commercial discov-ery, and producing asset. Each of thesestages, with the exception of the end-success stage, is the requisite stage for theasset to be in before we can exercise thenext decision. That is, if an asset is a lead,then if management decides to scope itand it successfully passes the scopingphase, it gets to be a viable scoped lead.Once an asset is deemed a viable scopedlead, management can decide to bid onthe asset, and if the lease is awarded, theasset becomes a leased prospect, and soon. At any stage, given that we decide theasset should progress, there is a probabil-ity for the asset to make a successful tran-sition to the next stage. We numbered thestages from lead (0) to commercial discov-ery (4). The probability that an asset com-pletes the scoping phase and becomes aviable scoped lead is thus P0-1 and soforth.Asset Evaluation and Calculation ofResource Requirements

The value of an asset is determined us-ing the summation of discounted projectedcash flows related to the asset across itslife cycle weighted with the respectiveprobabilities of occurrence for these cashflows. (The risk attitude of the organiza-tion with respect to the portfolio decisionsunder discussion is that of a risk-neutraldecision maker.) If the asset is a producingasset, then it is in its end-success stage andits value is a function of the projected pro-



Figure 5: In this mapping of the stages and decisions in the asset life cycle built into the systemto the organization’s work process, the top part of the figure shows the key decisions in theasset life cycle. The middle part shows a stage-to-stage asset-transition diagram. The assetstages marked with a bold border are the requisite stages for the asset to be in before manage-ment can exercise the next decision. The bottom part of the figure shows the organization’swork process.

duction volumes and the uncertainties re-lated to its cash flows, such as oil and gasprices, operating expense variables, andrequired future capital expenditures. Toevaluate an asset in an earlier stage, wetake into account the timing of the nextdecision and any subsequent decisions, theprobabilities of its making a successfultransition to subsequent stages, and thecash flow implications of resources allo-cated in each decision. Essentially, this isrolling back the tree corresponding to thestage-to-stage asset transition diagram. Re-garding resource allocations, if an asset isin a specific stage, such as Stage 2 (leasedprospect), then once the decision is madeto move the asset to the next stage, the re-sources required for risking are allocatedwith certainty. However, the probability

for the asset to successfully complete therisking phase is P2-3, and we use this prob-ability to calculate the resources requiredto move it beyond Stage 3.System Logic and Flow for SystemCalculations

We decided to build an open-loop deci-sion support system, as opposed to aclosed-loop optimizing system. That is, fora specific portfolio plan, the system calcu-lates the financial measures and the re-source requirements. The system also pro-vides the displays and reports to enablemanagement to ensure the plan’s feasibil-ity and to further optimize the company’sor business unit’s allocation of resourceswithin the portfolio. Thus, the system doesnot automatically solve for “the optimal”portfolio plan given a set of assets and a

SKAF

INTERFACES 29:6 92

specified level of available resources (aswould a closed-loop system). The mainreasons for opting for an open-loop deci-sion support system architecture over aclosed-loop self-optimizing system are thefollowing: (1) a decision support system ismore transparent and engages the exper-tise of the users and decision makers to agreater extent and (2) the calculation com-plexity in the case of a self-optimizing sys-tem would be significantly higher, thushaving serious implications on run timeand the overall process of portfolio analy-sis and management.

The core of the system consists of threeinput sheets (general inputs, plan inputs,and asset inputs), a sheet for processingthese inputs prior to evaluation, and anevaluation section consisting of a produc-tion sheet and an economic evaluationsheet (Figure 6).

The general inputs sheet contains all theparameters that have a portfolio-wide im-pact, such as cost of capital, forecasts of oiland gas prices, cycle times for activities,parameters for resource use, and parame-ters for operating costs. For each inputvariable, the user specifies the 10th percen-tile, 50th percentile, and 90th percentilevalues (or time series, for a time series in-put) to enable range sensitivity analysisand subsequently portfolio probabilisticanalysis.

The asset inputs sheet includes all asset-specific data and uncertainties, such inputsas water depth, depth to objective, leaseexpiry date, reserves (10th, 50th, and 90thpercentiles of reserves by asset), gas-to-oilratio, and stage-to-stage probabilities, en-tered by asset in a database type of struc-ture. This sheet contains all that the com-

pany knows about the assets (includingranges on uncertain parameters).

The plan inputs sheet contains the deci-sions for the assets in the portfolio de-scribed in the asset inputs sheet, such asworking interest, operatorship, asset activ-ity timing decisions (scoping, bidding,risking, exploratory drilling, developmentstart date, and so forth), and the develop-ment system type and configuration, en-tered by asset in a data structure parallelto that in the asset inputs sheet.

The calculated plan sheet processes thedecision inputs specified in the plan inputsheet using parameters from the generalinputs sheet (such as cycle time parame-ters) and the asset inputs sheet (such asstage-to-stage probabilities); mainly, it pre-pares the portfolio plan for calculating as-set yearly production and for economicevaluation.

The evaluation sheets are the produc-tion sheet and the economic evaluationsheet. The first calculates the constrainedproduction for oil and gas by asset afterdetermining the constraining fluid(whether oil or gas) at each hub and ac-counting for the production priority speci-fied in the plan input sheet. The secondperforms the economic evaluation and cal-culates resource requirements and finan-cial measures for each asset and for thewhole portfolio.Additional Notes on the SystemArchitecture

By including the key downstream deci-sions in the asset life cycle and accountingfor the interdependence between an assetand other assets in the portfolio, the sys-tem captures most of the strategic flexibil-ity for any specific asset. We have con-



Figure 6: The core of the system consists of three input sheets (general inputs, plan inputs, as-set inputs), a sheet for processing those inputs (calculated plan sheet), and an evaluation sec-tion consisting of a production sheet and an economic evaluation sheet. The system also in-cludes several modules designed for conducting tailored analyses. These modules access thedifferent sheets shown above to run specific analyses and produce reports.

sciously excluded the modeling of some ofthe additional operational flexibility avail-able in a single project that enables man-agers to later make or revise operationaldecisions within any phase. Trigeorgis andMason [1987] discuss the distinction be-tween strategic flexibility and operationalflexibility. For example, the system pro-vides the strategic flexibility of decidingwhen to start the development phase of aspecific asset, given that it successfullymakes the transition to stage 4, down tothe fraction of a year (Figure 5). The sys-tem then takes the specified number ofproduction wells for that asset and deter-mines their timing from the developmentstart date for the asset using an algorithm

assessed from experts in the organizationaccounting for drilling and completiontime cycles and the availability of capacityat the supporting hub. The system in-cludes an optimizing algorithm that de-lays the start of development based onwhen capacity becomes available at thesupporting hub and the asset’s priorityamong the assets tied to that hub. How-ever, the system does not [easily] allowthe user to specify the timing of each pro-duction well in an asset. Managers willeventually exercise this flexibility; how-ever, it is not modeled in the system. Thisadditional operational flexibility is valu-able and is better modeled and calculatedusing a dedicated system, which our cli-

SKAF

INTERFACES 29:6 94

ents have. Attempting to account for thefull operational flexibility in a portfoliomanagement system designed to handleportfolios of hundreds of assets would di-lute the focus of the system and the port-folio management process, increase runtime, and potentially lead to failure.

The issue of detailed budgeting is im-portant for assets that are in or close toproduction. For such assets, we added thecapability within the system to downloadthe line items for the different measuresfor such assets from a dedicated modelthat is used for detailed evaluation of oneasset at a time. The user still has to entersuch an asset, let’s call it Sierra, in theportfolio management system and specifya minimum number of parameters re-quired for evaluating other assets thatmight be tied to this asset Sierra. The de-tails behind all this are outside the in-tended scope of this article. The main ideais that we allowed the user the flexibilityof downloading and using the results of adedicated single-asset model for the caseswhere this was deemed necessary.The System Format

The system is a Microsoft Excel applica-tion with pull-down menus for accessingthe different parts of the system and forrunning various tailored portfolio analy-ses. To further integrate the system withexisting applications within the organiza-tion, we included in the system an inputmodule that interfaced with a database ap-plication used for storing asset-specificdata. Because of the system’s modularity,the user can tailor the size of the system tothe size of the portfolio being analyzed.Our clients have used it for analyzing abusiness unit portfolio of about 350 assets

and for analyzing a regional portfolio (orasset team portfolio) of less than 20 assets.The core part of the system contains anumber of detailed output modules thatshow such measures as resource require-ment and availability and by-asset bang-for-the-buck measures to facilitate portfo-lio management.

The system is easily linked to softwarepackages for conducting deterministic sen-sitivity analysis, probabilistic analysis, anddecision analysis. While the discussionhere has focused on the expected value ofa plan or the expected level of resource re-quirements, the user can easily obtain thefull probabilistic distribution of any valuemeasure or resource requirement.Using the System as a Decision-SupportTool: Main Benefits and Features

The system architecture provides theuser with the inputs and decision leversnecessary to manage a specific portfolio ofassets given the interdependencies that ex-ist among them. The set of decision leversavailable to the user, by asset, are asset-activity-timing decisions, ownership struc-ture and operatorship decisions, and de-velopment configuration decisions.Asset-activity-timing decisions includewhen to scope, bid on, lease, risk, drill, de-velop, or start producing a specific asset.Ownership-structure and operatorship de-cisions include what working interest tohave in each asset and whether to operatethe asset or hand it over to another opera-tor. Development-configuration decisionsinclude what type of development to usefor a specific asset; the user can specifywhether the asset is a hub (and chooseamong a set of hub configurations) or asubsea tied back to another hub (the term



subsea refers to a subsea development forthe production wells of a producing field).These decisions have significant implica-tions for the capital requirements andother resources needed as well as for theproduction profile of the asset.

The settings of these decisions across allassets in the portfolio constitute a specificportfolio plan. For each portfolio plan, thesystem calculates the financial measures(line items leading to yearly cash flows,net present value, and so on) and deter-mines resource requirements for executingthat specific plan. The interdependenciesamong assets include informational inter-dependencies as well as physical or opera-tional interdependencies. For assets in ma-ture areas, physical or operationalinterdependencies tend to be the impor-tant factors in improving the performanceof the portfolio. Examples are subsea de-velopments sharing the same hub andproduction-capacity considerations, andcompetition for and the sharing of re-sources (technical staff, capital and soforth). In frontier areas, informational in-terdependencies and asset-reserve uncer-tainties are important factors in managinga portfolio of prospects and leads. As abusiness-unit portfolio becomes populatedwith more and more mature assets, theimpact of physical and operational param-eters on the value of the portfolio in-creases accordingly.

The system supports a decision-focusediterative process for managing a portfolioof assets (Figure 7). The user starts byspecifying a portfolio plan and an initiallevel of available resources and then usesthe output of the system to increase thevalue of the portfolio by modifying the

plan or changing the levels of resources.Tracking the Use of Resources

At the first level, the portfolio manage-ment system tracks various financial mea-sures and expected resource requirementsfor each portfolio plan analyzed and com-pares the requirement levels to resourceavailability (Figure 8). The portfolio sys-tem provides displays for a number of re-sources, including rigs of different genera-tions (capabilities), technical staffresources (different categories), and pro-ject execution staff resources. The user canthen determine the value of relaxing theconstraints or the sequence of plan modifi-cations needed to meet the constraintswhile maximizing a specific measure. Thesystem provides several measures and dis-plays to help users modify a plan to meetspecific constraints in an optimal (or close-to-optimal) manner (Figure 9).Resource Allocation and Bang-for-the-Buck Measures

The portfolio-management system helpsmanagers faced with resource constraintsto compare investment opportunities andto allocate resources among assets in theportfolio. To improve the performance ofthe whole portfolio, they can use bang-for-the-buck measures to compare assetswithin a specific stage in the asset life cy-cle. Examples of bang-for-the-buck mea-sures are the ratio of expected net presentvalue (NPV) to expected developmentcapital expenditures required, the ratio ofexpected NPV to expected drilling re-sources required, and the ratio of expectedNPV to expected requirements for techni-cal staff in person-years. The objective is toensure that a certain resource is used mostefficiently across assets in the same stage;

SKAF

INTERFACES 29:6 96

Figure 7: The portfolio-management system supports an iterative portfolio management pro-cess. The user starts by specifying a portfolio plan and an initial level of available resourcesand then uses the output of the system to increase the value of the portfolio by modifying theplan or changing the levels of resources.

the system compares these assets using abang-for-the-buck ratio with the resourcerequirement by asset in the denominator.

In using bang-for-the-buck ratios, wemust compare assets that are in the samestage in the asset life cycle. These are onlyratios and not the ultimate value measuresthat we are seeking to maximize for thewhole portfolio. If we compared an assetearly in its life cycle (such as a lead) withone in a late stage (such as a discovery un-dergoing development), the total resourcerequirements for the first asset to reach theproducing stage (given that it successfullymoves through the requisite stages) wouldbe much greater than those needed for thesecond. In addition, most of the resources

required for an asset close to productionhave been spent, that is, are sunk costsand are not included in the analysis.

After determining bang-for-the-buckmeasures for the set of assets in a specificstage, we can plot them from highest tolowest (Figure 10). In some cases, becauseof a physical interdependency, the com-pany cannot pursue a high bang-for-the-buck asset (such as Gamma in this exam-ple) without pursuing another asset thathas a measure below the intended cutoffpoint (such as Zeus). In such cases, the de-cision maker would look at the bang-for-the-buck measure for the combination tomake a call. If the user inadvertently triesto exclude Zeus from the plan, the system



Figure 8: The portfolio management systemtracks expected yearly resource requirementsfor a specific team’s portfolio plan and identi-fies changes needed to ensure its feasibility.This shows the resource requirements of tworesources, X and Y, calculated for a specificunconstrained portfolio plan and comparesthe resource requirements to their projectedlevels of availability.

will warn her that asset Gamma is tied toZeus. The user can then look for anotherhub to support Gamma and exclude Zeusfrom the plan, analyze the alternative ofexcluding Gamma and Zeus from theplan, or analyze other plans that would in-clude both assets and then make adecision.

With multiple resource constraints, thedecision maker must consider displays forthe different resource constraints. Becauseuse of resources shows some correlation, adecision maker rarely needs to use morethan two such displays at a time.Identifying Key Value Drivers andQuantifying Their Effects at the PortfolioLevel

Managers often debate issues and spe-cific uncertainties without any systematicway to quantify their effects on their busi-ness and distinguish the key value drivers

requiring management’s attention. For ex-ample, in a given discussion, a manage-ment team may touch upon the increasingcompetition in the industry, the uncer-tainty on the reserves of a specific asset,the increase in a certain operating-cost pa-rameter, and other issues. In the absenceof a method for quantifying the effects ofsuch issues and uncertainties, these discus-sions usually end and resurface withoutcontributing any quality to the decision-making process. The purpose of a rangesensitivity analysis is to help managementidentify the key value drivers in the busi-ness and provide insight for further analy-sis and decision making.

The portfolio-management system en-ables the user to conduct range sensitivityanalyses (tornado analyses) on any set ofportfolio-wide and asset-specific variables.For each variable, the user specifies a basevalue (usually the 50th percentile), a lowvalue (the 10th percentile), and a highvalue (the 90th percentile). For time-seriesinputs, such as yearly forecast oil prices,the low (base, high) would be a series ofyearly prices that designated experts fore-cast as low (base, high) (the 10th, 50th,90th percentile scenarios). For a specificportfolio plan, the system calculates thebase-case value using user-specified basevalues for all the inputs entering the eval-uation of that plan. Then, taking one vari-able at a time, a system module changesits value from base to low, and then tohigh, and calculates the resulting swing inthe value of the plan. Sorting these swingsacross variables from highest to lowestand plotting them produces a tornado dia-gram (Figure 11).

The range-sensitivity analyses focuses

SKAF

INTERFACES 29:6 98

Figure 9: Using the system displays, the user can develop an alternative plan to meet certainconstraints on the availability of resources.

Figure 10: As an illustration of the use of bang-for-the-buck measures, in this plot, the systemranks assets within the same stage in the asset life cycle by the ratio of pretax value to requireddevelopment capital. For a budget constraint on development capital available for Area I pros-pects in Stage X, decision makers can use the system to decide which assets to fund.



Figure 11: A tornado diagram shows the range sensitivity analysis results for a specific portfolioplan. All variable names have been disguised and all absolute numeric values have been re-moved to maintain confidentiality. Dotted bars indicate swings resulting from ranges on re-serves of specific assets. The variations in plan value due to oil and gas price fluctuations havebeen omitted from the tornado for the purpose of this discussion.

managers’ attention on the few variablesthat drive most of the variation in portfo-lio value. The top six to eight variables ac-count for a great portion of the variationin that portfolio plan’s value (Figure 11).The variation in the portfolio plan’s valuefrom uncertainty on an asset’s reserves(such as asset �7, the fifth bar on the tor-nado in Figure 11) is generally smallerthan the variation from other operationaland cost-related variables. This becomesclearer as we realize that several of theseoperational or cost-related parametershave a portfolio-wide effect, whereas un-

certainties on asset reserves are predomi-nantly asset-specific. By focusing on someof the top variables in the tornado, whichare operational, the management team ofthat portfolio has an opportunity to createadditional value comparable to that cre-ated by adding a large discovery to theportfolio.Value of Technology Investments

To determine the value of investing in aparticular technology, the user or facilita-tor must first obtain assessments from adesignated expert regarding the differentdegrees of technical success of that tech-

SKAF

INTERFACES 29:6 100

nology. Then for each of these degrees, theuser or facilitator must determine whatsystem parameters will change and assesstheir new values. A breakthrough result-ing from an investment in technology maycause any of the following changes in sys-tem parameters: a reduction in the cycletime of an activity, an increase in the prob-ability that an asset will make the transi-tion from one stage to another (for exam-ple, due to improved imaging orinterpretation technology), a reduction in acapital-cost parameter, an improvement inthe production rates of a certain class ofstructures, and so forth. The user can thendetermine the value of the portfolio corre-sponding to each degree of technical suc-cess. For a technology that has a signifi-cant effect on the portfolio, the optimizedportfolio plan corresponding to a specificdegree of technical success may be differ-ent from that corresponding to another de-gree of success (one of these degreeswould represent the failure of the technol-ogy). Weighting these values by the prob-ability of each degree of technical success,the user is able to determine the value ofthe portfolio with the technology invest-ment. To compute the value of the tech-nology, the user would increase the cost ofthe technology up to the point of indiffer-ence between investing in the technologyand not investing.Developing and Evaluating DifferentBusiness Unit Portfolio Strategies

To a great extent, the portfolio manage-ment system has facilitated the process ofdeveloping and evaluating alternativebusiness-unit-portfolio strategies. A largebusiness unit would include several assetteams, each managing its own portfolio of

assets. Some teams may be regionallybased, others may focus on assets withcertain characteristics, and others may fo-cus on a single large asset and its sur-rounding area. From the business-unit-portfolio perspective, each teamconstitutes an aggregate decision area.Working with these asset teams, we facili-tate the development of team-strategy al-ternatives, which become the buildingblocks of alternative strategies for thewhole business-unit portfolio (Figure 12).

The set of all decision areas spanningthe scope of decisions the business unit isto address forms what we call a strategytable [Howard 1988]. These include team-specific and non-team-specific decision ar-eas. The latter include such decision areasas technology-investment decisions,portfolio-level acquisition decisions, andbusiness-unit-level supplier decisions. Un-der each decision area is a set of choices.Managers develop a business-unit strate-gic alternative by stringing a path acrossthe strategy table, selecting choices acrossthe decision areas that support a specificstrategic theme.

For asset-team decision areas, each stra-tegic choice corresponds to a specific assetplan that is entered into the portfolio man-agement system. A dedicated module inthe system then allows the user to storeand retrieve alternative plans under eachdecision area to form the whole business-unit portfolio. As previously discussed,the architecture of the system, through itsdatabase structure, allows this scaling upin the size of the portfolio plan. From theevaluation side, the system treats a largebusiness unit portfolio plan in the samemanner as a small team plan. Specific



Figure 12: A strategy table consists of the various decision areas in an organization, with differ-ent creative yet doable choices under each area. A specific strategy theme consists of a coherentset of choices across the decision areas. This is illustrated here using an example of a businessunit portfolio strategic alternative, under the theme “cost-effective growth,” with a specific pathacross the decision areas in the table.

modules for the analysis of a large busi-ness unit portfolio allow the user to minefor more insight by comparing the evalua-tion of different portfolio-strategyalternatives.The Portfolio Management Process

We designed an overall portfolio man-agement process that would provide a fo-rum for decision-focused dialogues be-tween senior management and assetteams. The portfolio management systemsupports the process and provides the an-alytical evaluation necessary to facilitatedecisions regarding resource allocationand asset-portfolio strategy.

The GMBU formed a portfolio manage-

ment core team under the leadership of aportfolio advisor to serve as the facilitatorand keeper of the process. The portfolioadvisor, a senior geophysicist with exten-sive experience in decision analysis, fi-nance, and overall project management,had a full-time commitment to the portfo-lio management process. The team in-cluded a reservoir engineer and a financialanalyst (both at half-time commitment)and a business and information analystwith a full-time commitment.

The portfolio advisor and the SDG teamjointly led the system design and develop-ment phase. During the initial designphase of the system, we developed an un-

SKAF

INTERFACES 29:6 102

Figure 13: The process of developing a portfolio strategy starts with a peer review of the cur-rent asset plan and an understanding of the business challenges to be addressed and ends withan alignment on the selected portfolio strategy. The arrows in the diagram represent the jointdecision-review meetings that bring together senior managers and asset team leaders to reviewthe progress made to a specific stage in the process.

derstanding of the existing portfolio man-agement process and worked on drafting anew process that would engage the vari-ous stakeholders in the organization whilebenefiting from the capabilities of the sys-tem. We further tailored the process to theneeds of the organization after running apilot version of the system.

In the new process, the portfolio man-agement core team, led by the portfolioadvisor, facilitates the decision-focused di-alogue between the decision team (mainlysenior managers) and the asset teams (Fig-ure 13). The process of developing a port-folio strategy starts with a peer review ofthe current asset plan and an understand-ing of the business challenges to be ad-dressed in developing a strategy and endswith an alignment on the selected portfo-lio strategy. In the joint decision-reviewmeetings that bring together senior man-

agers and asset team leaders to review theprogress made to a specific stage in theprocess, senior managers contribute to thequality of the portfolio decisions early inthe strategy process instead of waiting un-til the end and trying to inspect qualityinto the proposed plans.

After the asset teams perform a peer re-view of the asset base plans and assess-ments to ensure consistency in the assess-ments of volumes and probabilities, theportfolio-management core team conductsa preliminary diagnostic analysis of theportfolio base plan. A meeting is then heldto present the results of the diagnosticanalysis to the decision team and elicitchallenges to be addressed in the portfoliostrategy. The portfolio-management coreteam then works with the asset teams todevelop team-level alternatives and con-duct preliminary evaluations of these al-



ternatives. The decision team and theasset-team leaders meet to review the pre-liminary evaluation of team-level alterna-tives and develop portfolio strategic alter-natives. Thus, the expertise of seniormanagers on the decision team is engagedto a great extent through this developmentof portfolio strategic alternatives. Afterreaching agreement on the set of portfoliostrategy alternatives to be fully evaluated,the portfolio-management core team con-ducts the analyses while working againwith asset teams. Evaluating the portfolio-strategy alternatives may take several iter-ations. Often, insights gained in the evalu-ation process help teams to adjust theirplans (each plan being a choice in thatteam’s decision area, that is, a buildingblock of the portfolio alternatives). Finally,the decision-team members and the asset-team leaders meet to review the results ofthe final phase of evaluation and to decideon the portfolio strategy and resource allo-cations. Based on the alignment theyachieve, the asset teams work on refiningthe chosen plan and preparing forimplementation.

The company has implemented thisportfolio-management process and is cur-rently using it as part of the businessunit’s decision-making process. Theportfolio-management core team and SDGworked together to train users designatedby each asset team to use the system andto update asset data and plans. These peo-ple work on portfolio management whileinterfacing with the different stakeholdersin the organization, gaining experience inboth process and content that they retainand transfer back to the asset teams. Port-folio management is thus becoming a core

competence of the organization.Organizational Impact

The portfolio-management process hasenabled the organization to establish asystematic approach to developing itsbusiness-unit strategy. In the past, the adhoc process favored the asset-team leaderwith either the strongest arguments or themost optimistic asset plan. Now the orga-nization has a process that engages assetteams and senior management from dayone in identifying the best portfolio strat-egy for the business unit. Peer reviews en-sure consistency in asset-specific data(such as reserve distributions and stage-to-stage transition probabilities) early in theprocess. As a result, once the process getsto the review of evaluated business unitalternatives, people have confidence in thequality of the inputs and the consistencyacross team data, and can focus on deriv-ing insights from the analysis and improv-ing the set of alternatives. The shared un-derstanding of the value implications ofthe different alternatives builds commit-ment to action behind the selectedstrategy.

The portfolio-management process andsystem helped reduce the time for devel-oping regional strategies and business unitportfolio strategies. This is largely becausethe integration of the process and systemwith the organization’s business processeshas led to keeping the system up to datealmost continuously. For instance, theportfolio-management core team is nowable to work with a large asset team andfacilitate the development of a regionalstrategy in about two to three weeks (as-suming that most of the technical andasset-specific data is available within the

SKAF

INTERFACES 29:6 104

first week of the effort). Without theportfolio-management process and system,an effort of this size formerly took three tofive months. Without the portfolio processand system, the facilitator and analystteam would have conducted many moreassessments. They also would have had todesign, build, and debug a new dedicatedmodel for the effort. The power of a pro-cess that is integrated into the organiza-tion’s business processes cannot beoveremphasized.

The portfolio-management process andsystem have also enabled the organizationto examine any investment decision or ur-gent acquisition opportunity from an over-all portfolio perspective. This has affectedthe way the asset teams work and encour-aged asset-team leaders to consult withthe portfolio advisor (the leader of theportfolio management core team). Bytracking resource requirements for anyportfolio plan, the process and system alsohelp management ensure implementationsuccess for any strategic move the organi-zation undertakes.

The value added from strategic alterna-tives developed using the process and sys-tem is in the hundreds of millions of dol-lars. The process and system just supportthe decision dialogue among asset teams,senior management, and the portfoliomanagement core team. The insights thatlead to identifying sources of value andbuilding up to the selected business unitportfolio strategy come from the partici-pants in the process. The process and sys-tem are only enablers that reduce organi-zational barriers and analytical barriers tothis dialogue and provide the essentialperformance measures to guide the devel-

opment of the portfolio strategy.Acknowledgments

Developing the portfolio-managementprocess and system was a team effort thatbrought together several clients and col-leagues. I thank my client colleagues forthe many ideas we exchanged along theway and for their advice and input. FromNavigant Consulting, Strategy ConsultingPractice (formerly SDG), I thank WarwickBlyth, Jay Goldman, Tei Lin, DaveMacway, Jeff Robinson, Steve Uhl, andHannah Winter for the memorable mo-ments, and there were many, that wespent together on this effort. This paperhas benefited greatly from valuable inputfrom several people. I thank Sam Bodily,Steve Derby, Mary Haight, Leo Hopf,Ronald Howard, Jeannie Kahwajy, ElissaOzanne, Bill Poland, James Smith, and twoanonymous reviewers for their help.ReferencesHoward, R. A. 1988, “Decision analysis: Prac-

tice and promise,” Management Science, Vol.34, No. 6 (June) pp. 679–695.

Howard, R. A. 1990, “From influence to rele-vance to knowledge,” in Influence Diagrams,Belief Nets and Decision Analysis, eds. R. M.Oliver and J. Q. Smith, John Wiley and Sons,New York.

Trigeorgis, L. and Mason, S. P. 1987, “Valuingmanagerial flexibility,” Midland Corporate Fi-nance Journal, Vol. 5, No. 1, pp. 14–21.

White, D. A. 1992, “Selecting and assessingplays,” in The Business of Petroleum Explora-tion, ed. Richard Steinmetz, American Associ-ation of Petroleum Geologists, Tulsa,Oklahoma.

Documents

Portfolio Management in an Upstream Oil \u0026 Gas Organization