strategy under uncertainty.pdf

NOVEMBER-DECEMBIR 1997

Strategy UnderUncertainty

by Hugh Courtney, Jane Kirkland, and Patrick Viguerie

WHAT MAKES FOR A GOOD STRATEGY in highly Uncertain

business environments? Some executives seek to shape the

future with high-stakes bets. Eastman Kodak Company for exam-

ple, is spending $500 million per year to develop an array of digital

photography products that it hopes will fundamentally change the

way people create, store, and view pictures. Meanwhile, Hewlett-

Packard Company is investing $50 million per year to pursue a rival

vision centered around home-based photo printers. The business

press loves to hype such industry-shaping strategies because of their

potential to create enormous wealth, but the sober reality is that

most companies lack the industry position, assets, or appetite for

risk necessary to make such strategies work.

More risk-averse executives hedge their bets by making a number

of smaller investments. In pursuit of growth opportunities in emerg-

ing markets, for example, many consumer-product companies are

forging limited operational or distribution alliances. But it's often

HARVARD BUSINESS REVIEW November-December 1997 67

STRATEGY UNDER UNCERTAINTY

difficult to determine if such limited investmentstruly reserve the right to play in these countries orjust reserve the right to lose.

Alternatively, some executives favor invest-ments in flexibility that allow their companies toadapt quickly as markets evolve. But the costs of

Under uncertainty, traditiotialapproaches to strategic planningcan be downriglit dangerous.

establishing such flexibility can be high. Moreover,taking a wait-and-see strategy - postponing large in-vestments until the future becomes clear-can cre-ate a window of opportunity for competitors.

How should executives facing great uncertain-ty decide whether to het big, hedge, or wait andsee? Chances are, traditional strategic-planning pro-cesses won't help much. The standard practice is tolay out a vision of future events precise enough tobe captured in a discounted-cash-flow analysis. Ofcourse, managers can discuss alternative scenariosand test how sensitive their forecasts are to changesin key variables, but the goal of such analysis is of-ten to find the most likely outcome and create astrategy based on it. That approach serves compa-nies well in relatively stable business environ-ments. But when there is greater uncertainty aboutthe future, it is at best marginally helpful and atworst downright dangerous.

One danger is that this traditional approach leadsexecutives to view uncertainty in a hinary way-toassume that the world is either certain, and there-fore open to precise predictions about the future, oruncertain, and therefore completely unpredictable.Planning or capital-budgeting processes that re-quire point forecasts force managers to bury under-lying uncertainties in their cash flows. Such sys-tems clearly push managers to underestimateuncertainty in order to make a compelling case fortheir strategy.

Underestimating uncertainty can lead to strate-gies that neither defend against the threats nor takeadvantage of the opportunities that higher levels ofuncertainty may provide. In one of the most colossal

Hugh Courtney is a consultant in McKinsey &) Compa-ny's Washington. D.C, office: Jane Kirkland is a princi-pal with McKinsey in Pittsburgh, Pennsylvania-, andPatrick Viguerie is a principal with McKinsey in At-lanta, Georgia.

underestimations in business history, Kenneth H.Olsen, then president of Digital Equipment Corpo-ration, announced in 1977 that "there is no reasonfor any individual to have a computer in theirhome." The explosion in the personal computermarket was not inevitable in 1977, but it was cer-

tainly within the range of possibili-ties that industry experts were dis-cussing at the time.

At the other extreme, assumingthat the world is entirely unpredict-able can lead managers to abandonthe analytical rigor of their tradi-tional planning processes altogetherand base their strategic decisions pri-

marily on gut instinct. This "just do it" approach tostrategy can cause executives to place misinformedbets on emerging products or markets that result inrecord write-offs. Those who took the plunge andinvested in home banking in the early 1980s imme-diately come to mind.

Risk-averse managers who think they are in veryuncertain environments don't trust tbeir gut in-stincts and suffer from decision paralysis. Theyavoid making critical strategic decisions about theproducts, markets, and technologies they shoulddevelop. They focus instead on reengineering, qual-ity management, or internal cost-reduction pro-grams. Although valuable, those programs are notsubstitutes for strategy.

Making systematically sound strategic decisionsunder uncertainty requires a different approach-one that avoids this dangerous binary view. It is rarethat managers know absolutely nothing of strategicimportance, even in the most uncertain environ-ments. In fact, they usually can identify a range ofpotential outcomes or even a discrete set of scenar-ios. This simple insight is extremely powerful be-cause determining which strategy is best, and whatprocess should be used to develop it, depend vitallyon the level of uncertainty a company faces.

What follows, then, is a framework for determin-ing the level of uncertainty surrounding strategicdecisions and for tailoring strategy to that uncer-tainty. No approach can make the challenges of un-certainty go away, but this one offers practical guid-ance that will lead to more informed and confidentstrategic decisions.

Four Levels of Uncertainty

Even the most uncertain business environmentscontain a lot of strategically relevant information.First, it is often possible to identify clear trends,such as market demographics, that can help define

HARVARD BUSINESS REVIEW November-December 1997


potential demand for future products or services.Second, there is usually a host of factors that arecurrently unknown but that are in fact knowable-that could be known if the right analysis were done.Performance attributes for current technologies,elasticities of demand for certain stable categoriesof products, and competitors' capacity-expansionplans are variahies that are often unknown, but notentirely unknowable.

The uncertainty that remains after the best possi-ble analysis has been done is what we call residualuncertainty - ior example, theoutcome of an ongoing regulatorydebate or the performance attrib-utes of a technology still in devel-opment. But often, quite a bit canbe known about even those resid-ual uncertainties. In practice, wehave found that the residual un-certainty facing most strategic-decision makers falls into one offour broad levels:

Level 1: A Clear-Enough Future.At level 1, managers can developa single forecast of the future thatis precise enough for strategy de-velopment. Although it will beinexact to the degree that all busi-ness environments are inherentlyuncertain, the forecast will be suf-ficiently narrow to point to a sin-gle strategic direction. In otherwords, at level 1, the residual un-certainty is irrelevant to makingstrategic decisions.

Consider a major airline tryingto develop a strategic response tothe entry of a low-cost, no-frillscompetitor into one of its huh air-ports. Should it respond with a low-cost service ofits own? Should it cede the low-cost niche seg-ments to the new entrant? Or should it competeaggressively on price and service in an attempt todrive the entrant out of the market?

To make that strategic decision, the airline'sexeeutives need market research on the size of dif-ferent customer segments and the likely responseof each segment to different combinations of pric-ing and service. They also need to know how muchit costs the competitor to serve, and how much ca-pacity the competitor has for, every route in ques-tion. Finally, tbe executives need to know the newentrant's competitive objectives to anticipate howit would respond to any strategic moves their air-line migbt make. In today's U.S. airline industry^

such information is either known already or is pos-sible to know. It migbt not be easy to obtain-itmight require new market research, for example-but it is inherently knowable. And once that infor-mation is known, residual uncertainty would belimited, and tbe incumbent airline would be able tobuild a confident business case around its strategy.

Level 2: Alternate Futures. At level 2, the futurecan be described as one of a few alternate outcomes,or discrete scenarios. Analysis cannot identifywhich outcome will occur, although it may help

establish probabilities. Most important, some, ifnot all, elements of the strategy would change if theoutcome were predictable.

Many businesses facing major regulatory or leg-islative change confront level 2 uncertainty. Con-sider U.S. long-distance telephone providers in late1995, as they began developing strategies for enter-ing local telephone markets. By late 1995, legisla-tion that would fundamentally deregulate tbe in-dustry was pending in Congress, and the hroad formthat new regulations would take was fairly clear tomost industry observers. But whether or not thelegislation was going to pass and how quickly itwould be implemented in the event it did pass wereuncertain. No amount of analysis would allow thelong-distance carriers to predict those outcomes,

HARVARD BUSINESS REVIEW Niivcmlier-Decembcr 1997 69


How to Usethe FourLevels ofUncertointy 1 1

—

A Clear-Enough Future Alternate Futures

What CanBe Known?

Analytic Tools

Examples

A single forecasf precise enough fordetermining strategy

"Traditional" strategy tool kit

Strategy against low-cost airline entrant

A few discrete outcomes that definethe future

Decision analysisOption valuation modelsGame theory

Long-distance telephone carriers' strategyto enter deregulated local-service marketCapacity strategies for chemical plants

and the correct course of action-for example, thetiming of investments in network infrastructure-depended on which outcome occurred.

In another common level 2 situation, the value ofa strategy depends mainly on competitors' strate-gies, and those cannot yet be observed or predicted.For example, in oligopoly markets, such as those forpulp and paper, chemicals, and basic raw materials,the primary uncertainty is often competitors' plansfor expanding capacity; Will they build new plantsor not? Economies of scale often dictate that anyplant built would be quite large and would be likelyto have a significant impact on industry prices andprofitability. Therefore, any one company's deci-sion to build a plant is often contingent on competi-tors' decisions. This is a classic level 2 situation:The possihle outcomes are discrete and clear. It isdifficult to predict which one will occur. And thebest strategy depends on which one does occur.

Level 3: A Range of Futures. At level 3, a range ofpotential futures ean be identified. That range isdefined by a limited number of key variables, butthe actual outcome may lie anywhere along a con-tinuum bounded by that range. There are no naturaldiscrete scenarios. As in level 2, some, and possiblyall, elements of the strategy would change if theoutcome were predictable.

\

Companies in emerging industries or enteringnew geographic markets often face level 3 uncer-tainty. Consider a European consumer-goods com-pany deciding whether to introduce its products tothe Indian market. The best possible market re-search might identify only a broad range of poten-tial customer-penetration rates-say, from 10% to30%-and there would be no obvious scenarioswithin that range. Such a broad range of estimateswould be common when introducing completelynew products and services to a market, and there-fore determining the level of latent demand is verydifficult. The company entering India would belikely to follow a very different and more aggressiveentry strategy if it knew for certain that its cus-tomer penetration rates would be closer to 30%than to 10%.

Analogous problems exist for companies in fieldsdriven by technological innovation, such as thesemiconductor industry. When deciding whether toinvest in a new technology, producers can oftenestimate only a broad range of potential cost andperformance attributes for the technology, and theoverall profitability of the investment depends onthose attributes.

Level 4: True Ambiguity. At level 4, multipledimensions of uncertainty interact to create an

HARVARD BUSINESS REVIEW November-December 1997


/I

1

1/

A Range of Futures

• A ronge of possible outcomes, but nonatural scenarios

• Latent-demand research• Technology forecosting• Scenario planning

• Entering emerging markets, such as India• Developing or acquiring emerging

technologies in consumer electronics

True Ambiguity

Na basis ta forecast the future

Analogies and pattern recognitionNonlinear dynamic models

Entering the market for consumer multi-media applicationsEntering the Russian market in 1992

What CanBe Known?

Analytic Tools

Exomples

environment that is virtually impossible to predict.Unlike in level 3 situations, the range of potentialoutcomes cannot he identified, let alone scenarioswithin that range. It might not even be possihle toidentify, much less predict, all the relevant vari-ables that will define the future,

f̂ Level 4 situations are quite rare, and they tend to' migrate toward one of the other levels over time.Nevertheless, they do exist. Consider a telecommu-nications company deciding where and how tocompete in the emerging consumer-multimediamarket. It is confronting multiple uncertaintiesconcerning technology, demand, and relationshipsbetween hardware and content providers, all ofwhich may interact in ways so unpredictahle thatno plausible range of scenarios can be identified.

Companies considering making major entry in-vestments in post-Communist Russia in 1992 facedlevel 4 uncertainty. They could not outline the po-tential laws or regulations that would govern prop-erty rights and transactions. That central uncer-tainty was compounded hy additional uncertaintyover the viability of supply chains and the demandfor previously unavailable consumer goods and ser-vices. And shocks such as a political assassinationor a currency default could have spun the wholesystem toward completely unforeseen outcomes.

Those examples illustrate how difficult strategicdecisions can be at level 4, hut they also underscoretheir transitory nature. Greater political and regula-tory stability has turned decisions ahout whether toenter Russian markets into level 3 prohlems for themajority of industries today. Similarly, uncertaintyahout strategic decisions in the consumer multi-media market will migrate to level 3 or to level 2 asthe industry begins to take shape over the next sev-eral years.

Tailoring Strategic Analysis to theFour Levels of UncertaintyOur experience suggests that at least half of allstrategy prohlems fall into levels 2 or 3, while mostof the rest are level 1 problems. But executives whothink ahout uncertainty in a binary way tend totreat all strategy prohlems as if they fell into eitherlevel 1 or level 4. And when those executives basetheir strategies on rigorous analysis, they are mostlikely to apply the same set of analytic tools regard-less of the level of residual uncertainty they face.For example, they might attempt to use standard,quantitative market-research techniques to fore-cast demand for data traffic over wireless commu-nications networks as far out as ten years from now.



But, in fact, a different kind of analysis should bedone to identify and evaluate strategy options ateach level of uncertainty. All strategy makingbegins witb some form of situation analysis - tbatis, a picture of what the world will look like todayand wbat is likely to happen in tbe future. Identify-ing the levels of uncertainty thus helps define the

The old one-size-fits-all analyticapproach to evaluating strategyoptions is simply inadequate.

best such an analysis can do to describe each possi-ble future an industry faces.

To help generate level l's usefully precise predic-tion of tbe future, managers can use tbe standardstrategy tool kit - market research, analyses of com-petitors' costs and capacity, value chain analysis,Micbael Porter's five-forces framework, and so on.A discounted-eash-fiow model tbat incorporatestbose predictions can then be used to determine tbevalue of various alternative strategies. It's not sur-prising tbat most managers feel extremely comfort-able in level 1 situations-tbese are the tools andframeworks taugbt in every leading business pro-gram in the United States.

Level 2 situations are a bit more complex. First,managers must develop a set of diserete scenariosbased on tbeir understanding of how the key resid-ual uncertainties might play out-for example,whether deregulation occurs or not, a competitorbuilds a new plant or not. Eaeh sce-nario may require a different valua-tion model-general industry struc-ture and conduct will often befundamentally different dependingon which scenario occurs, so alter-native valuations can't be handledby performing sensitivity analysesaround a single baseline model. Get-ting information that helps establish the relativeprobabilities of the alternative outcomes should bea high priority.

After establisbing an appropriate valuation mod-el for eacb possible outcome and determining howprobable each is likely to be, a classic decision-analysis framework can be used to evaluate tberisks and returns inherent in alternative strategies.This process will identify the likely winners andlosers in alternative scenarios, and perhaps moreimportant, it will help quantify what's at stake for

companies that follow status quo strategies. Suchan analysis is often the key to making tbe case forstrategic change.

In level 2 situations, it is important not only toidentify the different possible future outcomes butalso to think through the likely paths tbe industrymight take to reacb tbose alternative futures. WiU

change occur in major steps at someparticular point in time, following,for example, a regulatory ruling or acompetitor's decision to enter tbemarket? Or will change occur in amore evolutionary fashion, as oftenhappens after a resolution of compet-ing tecbnology standards? This isvital information because it deter-

mines wbich market signals or trigger variablesshould be monitored closely. As events unfold andthe relative probabilities of alternative scenarioschange, it is likely tbat one's strategy will also needto be adapted to these changes.

At one level, the analysis in level 3 is very similarto tbat in level 2. A set of scenarios needs to he iden-tified tbat describes alternative future outcomes,and analysis should focus on the trigger events sig-naling that the market is moving toward one or an-otber scenario. Developing a meaningful set of sce-narios, however, is less straightforward in level 3.Scenarios tbat describe tbe extreme points in therange of possible outcomes are often relatively easyto develop, but tbese rarely provide much concreteguidance for current strategic decisions. Since tbereare no otber natural discrete scenarios in level 3, de-ciding which possible outcomes sbould be fully de-veloped into alternative scenarios is a real art. But

At level 4, it is critical to avoid theurge to throw up your hands and

act purely on gut instinct.

tbere are a few general rules. First, develop only alimited number of alternative scenarios-tbe eom-plexity of juggling more than four or five tends tobinder decision making. Second, avoid developingredundant scenarios that have no unique implica-tions for strategic decision making; make sure eachscenario offers a distinct picture of tbe industry'sstructure, conduct, and performance. Third, devel-op a set of scenarios that collectively account fortbe probable range of future outcomes and not nec-essarily tbe entire possible range.

72 HARVARD BUSINESS REVIEW Novcmber-DL-cember


Because it is impossible in level 3 to define acomplete list of scenarios and related probabilities,it is impossible to calculate the expected value ofdifferent strategies. However, establishing therange of scenarios should allow managers to deter-mine how robust tbeir strategy is, identify likelywinners and losers, and determine roughly the riskof following status quo strategies.

Situation analysis at level 4 is even more qualita-tive. Still, it is critical to avoid the urge to throwone's hands up and act purely on gut instinct. In-stead, managers need to catalog systematicallywhat they know and what is possible to know. Evenif it is impossible to develop a meaningful set ofprobable, or even possible, outcomes in level 4 situ-ations, managers can gain valuable strategic per-spective. Usually, they can identify at least a subsetof the variables tbat will determine ht)w the marketwill evolve over time - for example, customer pene-tration rates or technology performance attributes.And they can identify favorable and unfavorable in-dicators of these variables that will let them trackthe market's evolution over time and adapt theirstrategy as new information becomes available.

Managers can also identify patterns indicatingpossible ways tbe market may evolve by studyinghow analogous markets developed in other level 4situations, determining the key attributes of thewinners and losers in those situations and identify-ing the strategies they employed. Finally, althoughit will be impossible to quantify tbe risks and re-turns of different strategies, managers should beable to identify what information they would haveto believe about the future to justify the invest-ments they are considering. Early market indica-

tors and analogies from similar markets will helpsort out whether such beliefs are realistic or not.

Uncertainty demands a more flexible approach tosituation analysis. The old one-size-fits-all ap-proach is simply inadequate. Over time, companiesin most industries will face strategy problems thathave varying levels of residual uncertainty, and it isvitally important that the strategic analysis be tai-lored to the level of uncertainty.

Postures and Moves

Before we can talk about the dynamics of formulat-ing strategy at each level of uncertainty, we need tointroduce a basic vocabulary for talking about strat-egy. First, there are three strategic postures a com-pany can choose to take vis-a-vis uncertainty: shap-ing, adapting, or reserving the right to play. Second,there are three types of moves in the portfolio of ac-tions that can be used to implement that strategy:big bets, options, and no-regrets moves.

Strategic Posture. Any good strategy requires achoice about strategic posture. Fundamentally, pos-ture defines tbe intent of a strategy relative to thecurrent and future state of an industry. Shapers aimto drive tbeir industries toward a new structure oftheir own devising. Tbeir strategies are about creat-ing new opportunities in a market-eitber by shak-ing up relatively stable level 1 industries or by try-ing to control the direction of tbe market inindustries witb bigber levels of uncertainty. Kodak,for example, through its investment in digital pho-tography, is pursuing a shaping strategy in an effortto maintain its leadership position, as a new tech-nology supersedes the one currently generating

Shape the futurePlay a leadership role inestablishing how the industryoperates, for example:

- setting standards- creoling demand

Adapt to the futureWin through speed, agility, andflexibility in recognizing andcapturing opportunities inexisting markets

Reserve the right to playInvest sufficiently lo stay in thegame but avoid prematurecammitments

HARVARD BUSINESS REVIEW November-Dccembcr 1997 73


most of its earnings. Although its product technol-ogy is new, Kodak's strategy is still based on a tradi-tional model in which the company provides digitalcameras and film while photo-processing storesprovide many of the photo-printing and storagefunctions for the consumer. Hewlett-Packard alsoseeks to be a shaper in this market, hut it is pursu-ing a radically different model in which high-quality, low-cost photo printers shift photo process-ing from stores to the home.

In contrast, adapters take the current industrystructure and its future evolution as givens, andthey react to the opportunities the market offers. Inenvironments with little uncertainty, adapterschoose a strategic positiomng-that is, where andhow to compete-in the current industry. At higherlevels of uncertainty, their strategies are predicatedon the ability to recognize and respond quickly tomarket developments. In the highly volatiletelecommunications-service industry, for example,service resellers are adapters. They buy and resellthe latest products and services offered hy the majortelecom providers, relying on pricing and effectiveexecution rather than on product innovation astheix source of competitive advantage.

The third strategic posture, reserving the right toplay, is a special form of adapting. This posture isrelevant only in levels 2 through 4; it involves mak-

ing incremental investments today that put a com-pany in a privileged position, through either superi-or information, cost structures, or relationships he-tween customers and suppliers. That allows thecompany to wait until the environment becomesless uncertain hefore formulating a strategy. Manypharmaceutical companies are reserving the rightto play in the market for gene therapy applicationsby acquiring or allying with small biotech firmsthat have relevant expertise. Providing privilegedaccess to the latest industry developments, theseare low-cost investments compared with buildinga proprietary, internal gene-therapy R&X) program.

A Portfolio of Actions. A posture is not a com-plete strategy. It clarifies strategic intent but notthe actions required to fulfill that intent. Threetypes of moves are especially relevant to imple-menting strategy under conditions of uncertainty:big hets, options, and no-regrets moves.

Big bets are large commitments, such as majorcapital investments or acquisitions, that will resultin large payoffs in some scenarios and large lossesin others. Not surprisingly, shaping strategies usu-ally involve big bets, whereas adapting and reserv-ing the right to play do not.

Options are designed to secure the big payoffs ofthe best-case scenarios while minimizing losses inthe worst-case scenarios. This asymmetric payoff

structure makes them resemblefinancial options. Most optionsinvolve making modest initial in-vestments that will allow compa-nies to ramp up or scale hack theinvestment later as the marketevolves. Classic examples includeconducting pilot trials hefore thefull-scale introduction of a newproduct, entering into limitedjoint ventures for distribution tominimize the risk of hreaking intonew markets, and licensing analternative technology in case itproves to be superior to a currenttechnology. Those reserving theright to play rely heavily on op-tions, but shapers use them aswell, either to shape an emerginghut uncertain market as an earlymover or to hedge their big bets.

Finally, no-regrets moves arejust that-moves that will pay offno matter what happens. Man-agers often focus on ohviousno-regrets moves like initiativesaimed at reducing costs, gather-

74 HARVARD BUSINESS REVIEW November-December 1997


What's in a Portfolio of Actions?These building blocks are distinguished by three payoff profiles-that is, the amount of investmentrequired up front and the conditions under which the investment will yield a positive return.

Scenario

1.2.

3.4.

Value

• ••

• ••

•

No-regrefs movesStrategic decisions that havepositive payoffs in any scenario

OptionsDecisions that yield a significantpositive payoff in some outcomesand a (small) negative effect in others

Big betsFocused strategies with positivepayoffs in one or more scenariosbut a negative effect in others

ing competitive intelligence, or building skills.However, even in highly uncertain environments,strategic decisions like investing in capacity andentering certain markets can be no-regrets moves.Whether or not they put a name to them, mostmanagers understand intuitively that no-regretsmoves are an essential element of any strategy.

The choice of a strategic posture and an accompa-nying portfolio of actions sounds straightforward.But in practice, these decisions are highly depen-dent on the level of uncertainty facing a given busi-ness. Thus the four-level framework can helpclarify the practical implications implicit in anychoice of strategic posture and actions. The discus-sion that follows will demonstrate the different stra-tegic challenges that each level of uncertainty posesand how the portfolio of actions may he applied.

Strategy in Level l's Clear-Enough Future. In pre-dictahle husiness environments, most companiesare adapters. Analysis is designed to predict an in-dustry's future landscape, and strategy involvesmaking positioning choices about where and howto compete. When the underlying analysis is sound,such strategies are by definition made up of a seriesof no-regrets moves.

Adapter strategies in level 1 situations are notnecessarily incremental or boring. For example,Southwest Airlines Company's no-frills, point-to-point service is a highly innovative, value-creatingadapter strategy, as was Gateway 2O0O's low-costassembly and direct-mail distribution strategywhen it entered the personal computer market inthe late 1980s. In both cases, managers were ahle toidentify unexploited opportunities in relativelylow-uncertainty environments within the existing

market structure. The best level 1 adapters createvalue through innovations in their products or ser-vices or through improvements in their husinesssystems without otherwise fundamentally chang-ing the industry.

It is also possible to be a shaper in level 1 situa-tions, but that is risky and rare, since level 1shapers increase the amount of residual uncertain-ty in an otherwise predictable market-for them-selves and their competitors-in an attempt to fun-damentally alter long-standing industry structuresand conduct. Consider Federal Express Corpora-tion's overnight-delivery strategy. When it enteredthe mail-and-package delivery industry, a stablelevel 1 situation, FedEx's strategy in effect createdlevel 3 uncertainty for itself. That is, even thoughCEO Frederick W. Smith commissioned detailedconsulting reports that confirmed the feasibility ofhis business concept, only a broad range of poten-tial demand for overnight services could be identi-fied at the time. For the industry incumbents likeUnited Parcel Service, FedEx created level 2 uncer-tainty. FedEx's move raised two questions for UPS:Will the overnight-delivery strategy succeed ornot? and Will UPS have to offer a similar service toremain a viable competitor in the market?

Over time, the industry returned to level 1 sta-bility, but with a fundamentally new structure.FedEx's bet paid off, forcing the rest of the industryto adapt to the new demand for overnight services.

What portfolio of actions did it take to realizethat strategy? Like most shaper strategies, even inlevel 1 situations, this one required some big bets.That said, it often makes sense to build options intoa shaper strategy to hedge against bad bets. Smith



might have hedged his hets hy leasing existing car-go airplanes instead of purchasing and retrofittinghis original fleet of Falcon "minifrcighters," or hecould have outsourced ground pickup and deliveryservices. Such moves would have limited theamount of capital he would have needed to sinkinto his new strategy and facilitated a graceful exithad his concept failed. However, that kind of insur-ance doesn't always come cheap. In FedEx's case,had Smith leased standard-size cargo planes, hewould have come under the restrictive regulationsof the Civil Aeronautics Board. And outsourcinglocal pickups and deliveries would have dilutedFedEx's unique door-to-door value to customers.Thus Smith stuck mainly to big hets in implement-ing his strategy, which drove him to the brink ofhankruptcy in his first two years of operation hutultimately reshaped an entire industry.

Strategy in Level 2's Alternate Futures. If shapersin level 1 try to raise uncertainty, in levels 2through 4 they try to lower uncertainty and createorder out of chaos. In level 2, a shaping strategy isdesigned to increase the probability that a favoredindustry scenario will occur. A shaper in a capital-intensive industry like pulp and paper, for example,wants to prevent competitors from creating excesscapacity that would destroy the industry's prof-itahility. Consequently, shapers in such casesmight commit their companies to huilding new ca-pacity far in advance of an upturn in demand to pre-empt the competition, or they might consolidatethe industry through mergers and acquisitions.

Consider the Microsoft Network (MSN). A fewyears ago, one could identify a discrete set of possi-hle ways in which transactions would be conductedbetween networked computers. Either proprietarynetworks such as MSN would become the stan-dard, or open networks like the Internet would pre-

Shaping strategies can fail, so thebest companies supplement theirshaping bets with options that letthem change course quickly.

vail. Uncertainty in this situation was thus at level2, even though other related strategy issues-suchas determining the level of consumer demand fornetworked applications-were level 3 prohlems.

Microsoft could reasonably expect to shape theway markets for electronic commerce evolved if it

created the proprietary MSN network. It would, ineffect, he huilding a commerce hub that would linkhoth suppliers and consumers through the MSNgateway. The strategy was a big bet: the develop-ment costs were significant and, more important,involved an enormously high level of industry ex-posure and attention. In effect, for Microsoft, it con-stituted a big credibility bet. Microsoft's activitiesin other areas - such as including one-button accessto MSN from Windows95-were designed to in-crease the probability that this shaping bet wouldpay off.

But even the hest shapers must be prepared toadapt. In the battle between proprietary and opennetworks, certain trigger variables-growth in thenumber of Internet and MSN subscribers, for exam-ple, or the activity profiles of early MSN subscrib-ers-could provide valuable insight into how themarket was evolving . When it became clear thatopen networks would prevail, Microsoft refocusedthe MSN concept around the Internet. Microsoft'sshift illustrates that choices of strategic posture arenot carved in stone, and it underscores the valueof maintaining strategic flexibility under uncertain-ty. Shaping strategies can fail, so the best com-panies supplement their shaping bets with optionsthat allow them to change course quickly if neces-sary. Microsoft was able to do just that because itremained flexible by being willing to cut its losses,by building a cadre of engineers who had a widerange of general-programming and product-devel-opment skills, and by closely monitoring key trig-ger variables. In uncertain environments, it is amistake to let strategies run on autopilot, remain-ing content to update them only through standardyear-end strategy reviews.

Because trigger variables are often relatively sim-ple to monitor in level 2, it can be easy to adapt

or reserve the right to play. For in-stance, companies tbat generateelectricity - and others whose busi-ness depends on energy-intensiveproduction processes-often facelevel 2 uncertainty in determiningthe relative cost of different fuel al-ternatives. Discrete scenarios can of-ten he identified - for example, eithernatural gas or oil will be the low-costfuel. Many companies thus choose

an adapter strategy when building new plants: theyconstruct flexible manufacturing processes tbatcan switch easily between different fuels.

Chemical companies often choose to reserve theright to play when facing level 2 uncertainty in pre-dicting the performance of a new technology. If the

76 HARVARD BUSINESS REVIEW November - December li»97


a Regional Bank Canfronts the Uncertainties in Electronic Commerce

4. Actively manage the strategy

Monitor key trigger events such os adoption rates foremerging products and the behovior of nontraditionalcompetitors such os telephone componies.

Establish a shor^cycle review ofthe portfolio of options.

Participote in a number of industryconsortia to reduce uncertainty. Actively

maiKigethe strategy

3. Build a portfolioof actions

Near-term opportunities tooffer more innovativeproducts in specificareas where the bankis strong [for example, pro-curement cards, industry- O Build aspecific payment products) ^ portfoliorepresent no-regrets moves. of actions

Offering leoding-edgepayment products to high-valuecustomer segments that ore mostvulnerable to attackers is anotherno-regrets move.

Forming a small new-business unit is a grov^ option to:• Conduct R&D for new payment ideas* Monitor industry developments in the broad

area of retail electronic payments

1. Identify the nature andextent of residual urKertainties

Key areas of uncertainty include:

• How much electronic commerce willoccur on the Internet

• How quickly consumers will switch fromptaper-tased to electronic poyments

Which specific instruments will becomethe primary payment vehicles

[smart cards? E-cash?)• What structure will emerge for the

electronic commerce industryHow vertically integrated mostplayers will be• What roles bonks and

nonbanks will play

The bank is focing level 3urKertainty in some areas

ond level 4 in others

Identify thenature and

extent ofresidualuncertainties

Chaosea strategicposture 2. Choose a strategic posture

Objectives:• Defend current customer franchise from attack

by new technology-based competitors* Capture new business opportunities in

fast growing markets

Overall posture: reserve the right to play

technology performs well, companies will have toemploy it to remain competitive in the market. Butif it does not fulfill its promise, incumbents cancompete effectively with existing technologies.Most companies are reluctant to bet several hun-dred million dollars on building new capacity andretrofitting old plants around a new technology un-til it is proven. But if they don't make at least incre-mental investments in the short run, they riskfalling too far behind competitors should the tech-nology succeed. Thus many will purchase optionsto license the new technology within a specifiedtime frame or begin retrofitting a proportion of ex-isting capacity around the new technology. In ei-ther case, small, up-front commitments give thecompanies privileged positions, but not obliga-tions, to ramp up or discontinue development ofthe new technology as its performance attributesbecome clearer over time.

Strategy in Level 3's Range of Futures. Shapingtakes a different form in level 3. If at level 2, shap-

ers are trying to make a discrete outcome occur,at level 3, they are trying to move the market ina general direction because they can identify onlya range of possible outcomes. Consider the battleover standards for electronie cash transactions,currently a level 3 problem since one can define arange of potential products and services thatfall between purely paper-based and purely elec-tronic cash transactions, but it is unclear todaywhether there are any natural discrete scenarioswithin that range. Mondex International, a consor-tium of financial services providers and technol-ogy companies, is attempting to shape the futureby establishing what it hopes will become universalelectronic-cash standards. Its shaping posture isbacked by big-bet investments in product develop-ment, infrastructure, and pilot experiments tospeed customer acceptance.

In contrast, regional banks are mainly choosingadapter strategies. An adapter posture at uncertain-ty levels 3 or 4 is often achieved primarily through

HARVARD BUSINESS REVIEW November-DecembtT 1997 77


investments in organizational capabilities designedto keep options open. Because they must make andimplement strategy choices in real time, adaptersneed quick access to the hest market informationand the most flexible organizational structures.Many regional banks, for example, have put inplace steering committees focused on electronicpayments, R&D projects, and competitive-intelli-gence systems so that they can constantly monitordevelopments in electronic payment technologyand markets. In addition, many regional banks aremaking small investments in industry consortia asanother way to monitor events. This adapter ap-proach makes sense for most regional banks-theydon't have the deep pockets and skills necessary toset standards for the electronic payment market,yet it is essential that they be able to offer the latestelectronic services to their customers as such ser-vices become available.

Reserving tbe rigbt to play is a common posturein level 3. Consider a telecommunications com-pany trying to decide whether to make a $1 billioninvestment in broadband cable networks in theearly 1990s. The decision hinged on level 3 uncer-tainties such as demand for interactive TV service.No amount of solid market research could preciselyforecast consumer demand for services that didn'teven exist yet. However, making incremental in-vestments in broadband-network trials could pro-

vide useful information, and it would put the com-pany in a privileged position to expand the businessin tbe future should that prove attractive. By re-structuring the broadband-investment decisionfrom a big bet to a series of options, the companyreserved the right to play in a potentially lucrativemarket without having to bet the farm or risk beingpreempted by a competitor.

Strategy in Level 4's True Ambiguity. Paradoxi-cally, even though level 4 situations contain thegreatest uncertainty, they may offer higher retumsand involve lower risks for companies seeking toshape the market than situations in either level 2 or3. Recall that level 4 situations are transitional bynature, often occurring after a major technologi-cal, macroeconomic, or legislative shock. Since noplayer necessarily knows the best strategy in theseenvironments, the shaper's role is to provide a vi-sion of an industry structure and standards thatwill coordinate the strategies of other players anddrive the market toward a morc stable and favor-able outcome.

Mahathir bin Mobamad, Malaysia's prime minis-ter, is trying to shape the future of the multimediaindustry in the Asian Pacific Rim. This is truly alevel 4 strategy problem at this point. Potentialproducts are undefined, as are the players, tbe levelof customer demand, and the technology standards,among other factors. The government is trying to

Needed: A More Comprehensive Strategy Tool Kit

In order to perform the kinds of analyses appropriateto high levels of uncertainty, many companies willneed to supplement their standard strategy tool kit.Scenario-planning techniques are fundamental to de-termining strategy under conditions of uncertainty.Game theory will heJp managers understand uncer-tainties based on competitors' conduct. Systems dy-namics and agent-based simulation models can helpin understanding the complex interactions in the mar-ket. Real-options valuation models can help in cor-rectly valuing investments in learning and flexibility.The following sources will belp managers get started:• Scenario Plamung. Kees van der Heijden, Scenarios:The Alt of Strategic Conversation (New York; JobnWiley & Sons, 1996); Paul J.H. Scboemaker, "ScenarioPlanning: A New Tool for Strategic Thinking," SloanManagement Review, Winter 1995.

• Game iTieory. Avinash K. Dixit and Barry ]. Nale-buff. Thinking Strategically: The Competitive Edge inBusiness, Politics, and Everyday Life (New York: W.W.Norton, 1991); Adam M. Brandenburger and Barry |.Nalebuff, "Tbe Rigbt Game: Use Game Theory toSbapc Strategy," HBR July-August 1995.• System Dynamics. Peter N. Senge, Fifth Discipline:The Art and Practice of the Learning Organization(New York: Doubleday, 1990); Arie de Geus, "Plan-ning as Learning," HBR Marcb-April 1988.• Agent-Based Models. |obn L. Casti, Would-BeWorlds: How Simulation Is Changing the Frontiers ofScience (New York: John Wiley & Sons, 1997).• Real Options. Avinasb K. Dixit and Robert S.Pindyck, "Tbe Options Approacb to Capital Invest-ment," HBR May-June 1995; Timotby A. Luebrman,"Wbat'sItWortb?" HBR May-June 1997.

78 HARVARD BUSINESS REVIEW November-December 1997


create order out of this chaos by investing at least$15 hillion to create a so-called Multimedia SuperCorridor (MSC) in Malaysia. The MSC is a 750-square-kilometer zone south of Kuala Lumpur thatwill include state-of-the-art "smart" huildings forsoftware companies, regional headquarters formultinational corporations, a "Multimedia Univer-sity/' a paperless government center called Putra-jaya, and a new city called Cyberjaya. By leveragingincentives like a ten-year exemptionfrom the tax on profits, the MSC hasreceived commitments from morethan 40 Malaysian and foreign com-panies so far, including such power-houses as Intel, Microsoft, NipponTelegraph and Telephone, Oracle,and Sun Microsystems. Mahathir'sshaping strategy is predicated on thenotion that the MSC will create a weh of relation-ships hetween content and hardware providers thatwill result in clear industry standards and a set ofcomplementary multimedia products and services.Intel's Malaysia managing director, David B. Mars-ing, recognized Mahathir's shaping aspirationswhen he noted, "If you're an evolutionist, it'sstrange. They're |the Malaysian goverrmient] tryingto intervene instead of letting it evolve."

Shapers need not make enormous hets as theMalaysian government is doing to he successful inlevel 3 or 4 situations, however. All that is requiredis the credibility to coordinate the strategies ofdifferent players around tbe preferred outcome.Netscape Communications Corporation, for exam-ple, didn't rely on deep pockets to shape Internetbrowser standards. Instead, it leveraged the credi-bility of its leadership team in the industry so thatother industry players thought, "If these guys thinkthis is the way to go, they must be right."

Reserving the right to play is common, but po-tentially dangerous, in level 4 situations. Oil com-panies believed they were reserving the right tocompete in China by buying options to establishvarious beachheads there some 20 years ago. How-ever, in such level 4 situations, it is extremely diffi-cult to determine whether incremental invest-ments are truly reserving the right to play or simplythe right to lose. A few general rules apply. First,look for a high degree of leverage. If the choice ofheaebhead in China comes down to maintaining asmall, but expensive, local operation or developinga limited joint venture with a local distributor, allelse being equal, go for the low-cost option. Higher-cost options must be justified with explicit argu-ments for why they would put the company in abetter position to ramp up over time. Second, don't

get locked into one position through neglect. Op-tions should be rigorously reevaluated wheneverimportant uncertainties are clarified - at least everysix months. Remember, level 4 situations are tran-sitional, and most will quickly move toward levels3 and 2.

The difficulty of managing options in level 4 situ-ations often drives players toward adapter postures.As in level 3, an adapter posture in level 4 is

Netscape relied on its credibility,rather than deep pockets, to shape

Intemet browser standards.

frequently implemented hy making investments inorganizational capabilities. Most potential playersin the multimedia industry are adopting that pos-ture today but will soon he making bigger bets asthe industry moves into level 3 and 2 uncertaintyover time.

A New Approach to UncertaintyAt the heart of the traditional approach to strategylies the assumption that by applying a set of power-ful analytic tools, executives can predict the futureof any business accurately enough to allow them tochoose a clear strategic direction. In relatively sta-hle businesses, that approach continues to workwell. But it tends to break down when the environ-ment is so uncertain tbat no amount of good analy-sis will allow them to predict the future.

Levels of uncertainty regularly confronting man-agers today are so high that they need a new way tothink ahout strategy. The approach we've outlinedwill help executives avoid dangerous binary viewsof uncertainty. It offers a discipline for thinking rig-orously and systematically ahout uncertainty. Onone plane, it is a guide to judging which analytictools can help in making decisions at various levelsof uncertainty and which eannot. On a broaderplane, our framework is a way to tackle the mostchallenging decisions that executives have tomake, offering a more complete and sophisticatedunderstanding of the uncertainty they face and itsimplications for strategy.

This article is based on research sponsored by McKinsey's on-going Strategy Theory Initiative (STI). The authors would liketo thank theii STI colleagues for their significant contributionsto this article.

Reprint 97603 To order reprints, see the tast pige of this iasue.


strategy under uncertainty.pdf

Documents

good strategy

underestimating uncertainty

executives favor

costs ofunder uncertainty

companies toadapt

riskaverse executives

highstakes bets

strategic planningcan