Published on February 4, 2011
Accountability Challenges in Public Sector Contracting for Complex Products Trevor Brown John Glenn School of Public Affairs The Ohio State University 1810 College Road Columbus, OH 43210-1336 E-mail: firstname.lastname@example.org Matthew Potoski Department of Political Science Iowa State University 519 Ross Hall Ames, IA 50011 E-mail: email@example.com David Van Slyke Maxwell School of Citizenship and Public Affairs Syracuse University 320 Eggers Hall Syracuse, NY 13244 E-mail: firstname.lastname@example.org Manuscript prepared for the Kettering Symposium on Public Accountability, Dayton, OH, May 22-24, 2008
Accountability Challenges in Public Sector Contracting for Complex Products1 Governments buy lots of goods and services. The U.S. federal government spent over$419 billion in fiscal year 2006 for procurement, almost double 2001 procurement expenditures(Hutton, 2008). The rationale for buying is to lower costs through scale or market efficiencies,spark service delivery improvements or innovation through competition, and access expertise orcapacity unavailable in-house (Kelman, 2002). The risks are that contracting’s cost savings aresometimes illusory, quality suffers, and delivery delayed (Sclar, 2000).2 Some fear that suchlarge scale procurement undermines accountability: when government purchases rather thanproduces, the chain of accountability is extended yet further and perhaps even weakened. Peopleworking for government are more easily held accountable than people working for organizationsthat sell to the government. When contracting fails, contracting muddles responsibility: doesfault lie with the seller, the buyer, or unforeseen circumstances that nature delivered? Contracting produces low risk win-win outcomes when markets function well: marketsneed enough buyers and sellers, buyers and sellers need to be well informed about products andeach others’ preferences, and actors must be able to easily enter and exit the market andexchange resources at low costs.3 Market discipline provides its own accountability by revealingwho is responsible for deals gone bad and then punishing them for their transgressions. Buying1 This research is partially funded by the IBM Business of Government Foundation.2 There are other risks associated with contracted service delivery, like nepotism or cronyism. Depending on thecircumstances, these risks are also present for internal service delivery. Because so much is made of service quality,cost and timeliness in the comparison between internal versus contracted service delivery, we focus our discussionon these three.3 There are several ways to view the sources of market failures, including incomplete property rights, transactioncosts, and information asymmetries. Goods may be non-rivalrous or non-excludable so that transferable propertyrights cannot be established and enforced without transaction costs swamping gains from trade (Weimer and Vining,1999). Historical accident may inefficiently lock in path dependent technologies such as the QWERTY keyboard(David, 1985). Information asymmetries between buyers and sellers may create a “lemon market” where inferiorproducts keep good products off the market (Akerloff, 1970). A common thread in these cases is that the marketfailure is caused by the transaction costs stemming from limited information among participants, particularly buyers,or from goal incongruence between the buyers and sellers. 1
and selling is more complex when the market is thin (few buyers and sellers), buyers and sellersare uncertain about the terms of exchange, and the future is unpredictable. Complexcircumstances threaten the win-win gains from market exchanges and obfuscate responsibilityfor failed exchanges. Contracting in complex circumstances does not guarantee failure, but it does make itmore likely. Buyers and sellers can no longer rely on market discipline, but instead must managetheir relations to ensure the exchange bears its win-win fruits. In this paper we investigateaccountability issues raised when governments purchase complex products. Our inquiry uses thesimple but powerful heuristic of game theory to show the complexities of accountabilityquestions in public sector contracting. The outcome of a complex contract exchange is the resultof the buyer’s and seller’s efforts – whether they have made good faith efforts to maximize themutual gain the contract can generate – as well as the luck of nature – unforeseen forces makesome apparently difficult tasks turn out to be easier and some easy ones harder. Because theeffects of these factors are generally unknown, it is difficult to determine whether the outcome ofa complex contract was due to a well managed process or fortunate circumstances. Ourtheoretical inquiry highlights how buyers and sellers can improve the chances of win-winoutcomes and make the contracting process more accessible and accountable. This paper is divided into four sections beyond this introduction. In the first section wedistinguish between simple and complex products, and then highlight the risks and rewards ofcontracting for complex products. In the second section, we describe the basic bilateralcontracting game. We then use the contracting game to identify accountability challenges incontracting for complex products. Fundamentally we show the challenges of trying to determinewhether outcomes result from the behavior of the two parties to the exchange or exogenous 2
factors. In the third section we provide an illustration of these accountability challenges – theCoast Guard’s Project Deepwater, a multi-year, multi-billion dollar procurement. In the fourthand concluding section, we summarize our arguments by highlighting important implications forcontracting practice and accountability.Simple and Complex Contracting Successful contracting delivers a win-win exchange between buyer (in our case agovernment agency) and seller (a private firm, a nonprofit organization, or even anothergovernment). Buyers seek goods or services that meet their expectations for quality at a pricelower than the value they attribute to the product, while sellers want compensation that exceedstheir costs for producing the good or service. Much of what government buys are what we callsimple products. For simple products, buyers and sellers can easily define cost, quality andquantity parameters, and there is a vibrant market of buyers and sellers with many exchanges andclear price, quantity and quality signals. In a market exchange for simple products, the buyer candescribe with a high degree of detail what it wants the seller to do or produce. While a contractcan be specified in a variety of forms – such as inputs and tasks, the outputs, the outcomes, orsome combination of these (Bajari and Tadelis 2001; Heinrich 1999; Martin 2004; O’Looney,1998)4 – overall, contracts for simple products are relatively complete in the sense that theyunambiguously define each party’s obligations. If for some reason a buyer or seller fails to liveup to her contractual obligations, the transgression is quickly and easily recognized and a richlycompetitive market provides a replacement partner seeking similar terms. When contracting for4 Contract completeness is the degree to which these elements are contractually specified, most importantly bydefining task, outcomes, and compensation as the basic terms of exchange. These core elements structure thedegree to which the contract is complete. Around these core elements, contracts can also be more or less completebased on the nature of the governance features that specify how the exchange will take place and how the partieswill interact in the course of delivering the service (Brown, Potoski, Van Slyke, 2008). 3
simple products, buyers and sellers can be reasonably confident of achieving the mutual gainsfrom successful win-win market exchanges. Rosy win-win scenarios are less likely in contracting for complex products where neitherbuyers nor sellers know the products’ cost, quality and quantity parameters ahead of negotiationand production.5 Complex products are exchanged under incomplete contracts becauseuncertainty is so high that the costs of writing out all contract terms to account for all futurescenarios exceed the mutual gains from the trade and no contracting would occur. Theconsequence of an incomplete contract is that an uncertain future can create ambiguities aboutroles and responsibilities and produce circumstances that favor one party at the expense of theother. Consequently, the costs of contracting include the risks of renegotiations and of beingheld to contract obligations that unforeseen circumstances have made unfavorable. Theseunforeseen circumstances, what we call nature, can fundamentally change the payoffs forproducing, exchanging and consuming a product, making it more valuable to the buyer, raisingits production costs, and so on. Parties to complex contracts can work ex ante to reduce uncertainty so the contract can bemore specific, such as investing in research and development, or write a less specific contractexpecting to renegotiate ambiguities ex post as the product is produced and the exchangeexecuted. Such investments are asset specific to the extent they can not be put to alternative use.In an exchange with asset specific investments, incomplete contracts coupled with uncertaintyabout the future raise the specter of the classic hold-up problem (e.g. Williamson, 1996). Fromthe buyer’s perspective, the risk of making an asset specific investment is that once theinvestment has been made, there is no alternative seller for the product, which means the seller5 A lot has been written about lemons markets where buyers are uncertain about the product but sellers are not(Akerloff 1970). In this paper, we focus on contracts where buyers and sellers each are initially both uncertain. 4
can exploit contract ambiguities to “hold up” the buyer to extract more favorable terms, eventerms that the buyer would not have agreed to with foreknowledge. This fear is complicated bythe fact that for many complex goods and services, government is the only buyer (e.g. weaponssystems); once the buyer makes a decision about a particular product or solution the marketclears as other potential sellers have no purchasers for their proposed product or solution. Theseller faces a similar dilemma in that they devote significant resources to develop (and perhapsdesign) a specific solution for the buyer and face the uncertainty that at some point in the futurethe buyer may walk away from the contract for any number of reasons (e.g. the product doesn’tsuccessfully address the buyer’s needs, authorizers prohibit the buyer from continuing with theprocurement). Incomplete contracts for complex products are also vulnerable to the vagaries of anuncertain future: unforeseen circumstances may leave either party trapped in a contract it wouldnot have entered had it known what was to come. Producing the product reduces uncertainty assellers, and perhaps to a lesser extent buyers, learn about the product and its quality-costtradeoffs. Such learning can be captured as lower production costs and scale economies if thecontract is for multiple products of the same type. It is important to note that the fruits ofreduced uncertainty are not necessarily enjoyed equally by buyers and sellers. For example,governments may realize that the delivered service does not meet what is needed or vendors’production costs may turn out to be much higher than anticipated, with the government enjoyingcost savings and the vendor suffering losses. Incomplete foreknowledge also means that eitherside may find itself in a position to exploit unforeseen circumstances for its own advantage. Aseller, for example, may exploit contract ambiguities to lower service quality because thecontract may lack provisions for the buyer to enforce original terms. In the absence of clear and 5
specific tasks, outputs, or outcome terms, the seller may misrepresent the quality of its “product”.On the other hand, the buyer, because it may have delivered the service directly in the past, mayhave more information about the challenges the seller is likely to face as it delivers the service.The buyer can use this first-hand knowledge opportunistically to negotiate a contract that fails tospecify the known conditions under which the seller can claim compensation for costs. All in all, we can see that achieving win-win outcomes in more difficult in contracting forcomplex products. Complex contracting’s success is a function of managing the uncertaintyassociated with both the behavior of the other party and the unexpected impacts of exogenousnatural factors. Ex ante, both parties to the exchange can take a variety of steps to promotepositive cooperation, such as signaling cooperation and concern about future contracting,revealing information about production and profits, and so on. As nature reduces uncertainty,buyers and sellers can renegotiate contract terms, adjusting the existing contract midstream orwriting new terms for future transactions. This assumes, however, that the buyer and/or the sellercan identify nature’s impact on contracting’s outcomes. Both parties may have the advantage ofmore information than when they entered into the initial agreement, but perhaps not sufficientinformation to discern the causes of the outcome they and the other party received. The relativelevel of information at this ex post phase highlights the accountability challenges of contractingfor complex products. In the next section we turn to game theory as a tool to help analyze thecontracting process as it unfolds from the ex ante to ex post stage.The Contracting Game Contracting can be viewed as a bilateral game between a buyer and a seller. For eachparty, the risks and rewards, or payoffs, of a contract reflect the contract’s design (i.e. each 6
parties’ responsibilities), the behavior of each player relative to the behavior of the other player(i.e. the players’ “moves”), and exogenous factors (i.e. nature). Payoffs reflect the objectives ofeach player; buyers seek to optimize across cost and quality (inclusive of timeliness), whilesellers seek to maximize profits. Treating contracting as a simple bilateral game provides apowerful heuristic for examining the possibility of a win-win outcome for complex contracting,as well as for identifying key accountability issues. In this section we present a basic contractinggame with four discrete phases described below. In the first phase of the contracting game, the buyer and seller negotiate contract termsand decide whether to make any asset specific investments to produce the good or service (e.g.research and development). Higher asset specific investments mean higher hold up risks; lowerinvestments mean a less precise contract and a higher risk of renegotiation costs in the future. Inthe contracting game’s second phase, the buyer and seller simultaneously decide and execute theapproach they intend to take in fulfilling their responsibilities – however clearly or unclearlydefined – in implementing the contract. A party that chooses to “cooperate” executes hercontract obligations with the goal of maximizing win-win gains for both sides, even if she couldhave done better (and her partner done worse) through another strategy. Alternatively, a partythat chooses to “defect” makes minimal investments in achieving a win-win gain for both sides,and perhaps pursues a strategy of ensuring a win for herself even if it results in a loss for theother party. In the third phase, “nature” reveals itself as positive or negative exogenous factorsthat affect the product’s costs and qualities. In the fourth phase players’ strategies combine withnature to determine payoffs (e.g. win, win). At this point, buyer and seller enjoy moreinformation than at the outset and hence have the opportunity to renegotiate contract terms , 7
leave the contract in place, or exit the contract.6 Figure 1 models the game’s four phasesgraphically.7[INSERT FIGURE 1 HERE] In a classic prisoner’s dilemma game each player improves his own position by defecting,no matter what the other player’s strategy, even though mutual defection leaves the two playersworse off than cooperation.8 Assuming a one shot purchase, we can see how contracts forcomplex products mirror the basic prisoner’s dilemma game. Each party may find itselfstrategically advantaged in the contract to the extent the other side has made asset specificinvestments and the extent to which unforeseen circumstances allow it to exploit contractambiguities for its own gains and at the other side’s expense. A seller, for example, might findhimself with the opportunity to “gold plate” the product by adding costly features that increasehis profits but add little value and considerable expense for the buyer. Likewise, a buyer mayforce a seller to produce an expensive product even with knowledge that a much cheaper productwould meet her needs almost as well. The problem in these cases is that the winners’ gains aresmaller than the losers’ losses. In other words, the winner’s choice gives him a larger share of asmaller pie. Alternatively, the winner could make decisions to maximize her own gains onlywhen the gains are larger than the other side’s losses, and may even take on losses for herself if itmeans larger gains for her partner.6 In the event that the contract is finished at this point, but the buyer still demands more of the product, thealternatives are to draft a more complete contract for new purchases, utilize the same incomplete contract again, orforgo the exchange.7 Note that in the second column even though the buyer’s decision to cooperate or defect is presented first, buyer andseller make their strategy decision simultaneously without knowledge of the strategy decision of the other party.8 An appendix displays the basic payoff matrix of a simple prisoner’s dilemma game. 8
After parties have selected their contract strategy and begun to execute the contract,nature intervenes to affect the payoffs parties receive. The affect of nature is random in the sensethat the parties did not know the probabilities or events that could occur. Perhaps it turns out thatthe product is cheaper to produce than anticipated. Or perhaps the buyer learns that the productwill be even more valuable than she had anticipated. Nature can affect buyers and sellers, butbuyers and sellers do not necessarily know how nature affected them or the other party. Forexample, if a seller easily solves a problem thought to have been difficult was it because of luckor because the problem was easier to solve than anticipated? Payoffs in the contract are determined by the buyer’s and seller’s combined strategychoices, and the fortune nature has brought them. The parties’ strategic choices affect payoffs inthe manner proscribed by the prisoners’ dilemma: the combined payoffs are highest if bothpursue cooperative strategies, although each can do better by defecting when the othercooperates. If both defect, the combined payoffs are lowest. In our version of the complexcontracting game, each player knows their own payoff from the game, but not the other’s, andneither player knows why the payoffs turned out as they did. The problem is that the otherplayer’s strategy, payoffs and contribution of nature all occur under the shroud of uncertainty. Figure 2 depicts the second through fourth phases of the contracting game – strategyselection, the impact of nature, and payoffs. The figure maps the prisoner’s dilemma strategiesand payoffs with the outcome of nature in a simple tree-form of the game. When the buyer andseller both cooperate, the payoffs are three for each, and when they both defect, the payoffs aretwo. If one cooperates and the other defects, the cooperator receives one and the defectorreceives four. Figure 2 depicts the effect of nature as a plus or minus two, which to simplify ourillustration we show affecting equally buyer and seller payoffs. There are more complex and 9
realistic ways of depicting nature’s role in this game; nature’s payoff can be more variable, suchas by favoring one party while harming another, and by different amounts. We keep Figure 2simple to convey the model’s underlying logic. In the most favorable scenario in Figure 2, both the buyer and seller choose a cooperativestrategy and nature provides favorable conditions, giving each side a payoff of five: three eachfrom mutual cooperation plus nature’s positive contribution of two to each. Had nature not beenso fortunate, the payoff would have been lower, perhaps three or less for the buyer and seller hadnature’s contribution been zero or a negative value. If the buyer chose to cooperate and the sellerchose to defect, the buyer’s payoff would be lower – in Figure 2 it would be only one exclusiveof nature’s combination. If in this scenario nature’s contribution is positive, the buyer could endup with a payoff that is higher than she would have received under a mutual cooperation scenariowith a negative nature payoff. This theoretical depiction of complex contracting raises fundamental issues foraccountability in public sector procurement. The buyer only knows her own contract strategyand her own payoff; the seller’s strategy and payoff, and nature’s contribution are all unknown tothe buyer. Consequently, the buyer can not discern the extent to which the payoff she receivesresults from the interaction between her strategy with the seller’s or from nature’s fortune.Consider further the case of a buyer who has chosen a cooperative strategy: a moderate payofffor the buyer may be the result of a cooperative seller strategy coupled with a penalty fromnature, or may be the result of fortunate natural circumstances coupled with the buyer choosing adefect strategy. Such uncertainty is pervasive to the extent the buyer does not know nature’scontribution to the payoffs or the buyer’s strategy. 10
Illustration: Accountability Challenges in the Coast Guard’s Project Deepwater In this section we present an example – the Coast Guard’s Project Deepwater – toillustrate the accountability challenges inherent in contracting for complex contracts. We use thebasic logic from the contracting game described above to highlight the challenges of complexcontracts. Negotiations occurred in a context of high uncertainty and fear of lock-in. Theresulting contract was incomplete. The Coast Guard and the selected vendor both proclaimedthey were choosing cooperative contracting strategies, although these claims are difficult toverify. The contract’s initial payoffs were lower than anticipated for both the buyer and seller,although it is difficult to determine whether this was the result of contract strategies or anenvironment that turned out to be less favorable than anticipated. We conclude this section byhighlighting the accountability challenges faced by the Coast Guard and its political overseers asthe Deepwater procurement moves forward.9Project Deepwater Background In recent history the Coast Guard’s procurement practice was to separate purchases forindividual classes of vehicular assets – ships, cutters, planes and helicopters; when a class ofships was no longer sea worthy, the Coast Guard bought a new one to replace it, perhaps with amodified design better suited to the Coast Guard’s evolving mission.10 Because it bought fewerand smaller assets relative to other major naval buyers – notably the U.S. Navy – the CoastGuard largely made ad hoc purchases from a handful of small sellers (e.g. regional shipyards9 This section is based on a series of interviews with participants involved in Project Deepwater along withtestimony, oversight reports, and publicly available material produced by the vendor and the Coast Guard. TheDeepwater Program acquisition is continuing. At this stage we highlight early accountability challenges.10 The Coast Guard’s mission is that of a law enforcement and military organization and it is engaged in maritimesecurity (upholding the law), maritime safety (rescuing the distressed), the protection of natural resources(protecting the environment), maritime mobility (ensuring safe marine transportation), and national defense(operating in coordination with the U.S. Navy). 11
such as Bollinger). Without significant procurement experience or capacity, and purchasing onlyinfrequently and for small quantities, the Coast Guard sometimes even acquired assets as part oflarger U.S. Navy procurements. By the early 1990s it became clear that the Coast Guard needed a more targeted andstrategic approach to upgrade its rapidly aging asset fleets.11 Many of the Coast Guard’s assetswere reaching the end of their usable life-span and were not ideally suited to the modern CoastGuard’s missions. In response, the Coast Guard leadership lobbied for a long-term acquisitionstrategy that would upgrade and modernize the entire Coast Guard fleet. The Coast Guard’s goalwas to acquire a system of interoperable assets whose seamless communication and coordinationwould make the efficacy of the whole greater than the sum of its parts. In 1998, Congress andthe Clinton administration committed to a multi-year procurement at $500 million a year,significantly more than the Coast Guard’s historical acquisition expenditure.12 The result was theDeepwater program or Project Deepwater. The Coast Guard essentially sought a complex product for its Deepwater upgrade. TheCoast Guard and prospective sellers faced high levels of internal and external uncertainty. Interms of internal uncertainty, the challenge was to design and build an array of sophisticatedinteroperable assets – that is they all had to be able to communicate with each other andseamlessly coordinate their activity in pursuit of different targets (e.g. armed speedboats runningcontraband, sailors lost at sea, make-shift vessels porting illegal aliens) – with a hard cap onoverall costs (i.e. $500 million a year). The seller also had have the production capacity orpurchasing ability to deliver very different kinds of assets: ships, cutters, helicopters and planes.11 As of 2001, eight-six percent of the Coast Guard’s assets, deepwater and air, had reached or were expected toreach the end of their planned service life within five years. The Coast Guard’s fleet of assets was widely consideredto be one of the oldest in the world, ranking 37 out of 39 of the fleets worldwide (Acquisition Solutions, 2001, p.6).12 http://govinfo.library.unt.edu/npr/library/news/062999.html 12
In terms of external uncertainty, the distributed global reach of the Coast Guard meantthat the environmental conditions and operational missions varied dramatically from location tolocation. Coast Guard ice breakers in the Bering Strait needed ships that could do differentthings than Coast Guard personnel responsible for drug interdiction in the waters off Florida. Inaddition, while Coast Guard had secured the commitment of one administration and the sittingCongress to a $500 million-a-year procurement budget, there was no guarantee that subsequentadministrations and Congresses would adhere to that pledge. The combination of interoperability, dual aerial and naval production capacity, varyingenvironmental and mission requirements, and tenuous political commitments translated into anasset specific buy with a significant risk of lock-in. To meet the Coast Guard’s request the sellerwould have to make high up-front investments to design and build a product to meet the specificneeds of a single client with the real possibility that the client may be forced to walk away fromthe commitment at a later date. The seller would then be left with a product, and perhaps aproduction process, for which there were few, if any, other buyers. For the Coast Guard the riskwas that the seller that won the bid would eventually gain an information advantage over theCoast Guard as it began to design and build the product. At the same time, absent any otherbuyers of similar products, the market would clear. If things went wrong with the selected seller(e.g. the seller began to “gold plate” the product), the Coast Guard would have limited options onwhere to go to acquire alternative aerial and naval production capacity and systems engineeringexpertise. 13
Contract Negotiation In 1998, the Coast Guard issued an RFP describing the mission needs and performancegoals it sought for its upgraded fleet, including the interoperability of its assets, and invitingindustry to propose creative solutions. The Coast Guard evaluated the three industry proposals itreceived based on the degree to which they met mission requirements, lowered total ownershipcosts, and, to minimize risk, relied on proven, off-the-shelf technologies. The Coast Guardselected Integrated Coast Guard Systems’ (ICGS, a consortium between Lockheed Martin andNorthrop Grumman) proposal to upgrade or replace ten new asset classes of air and sea vesselsby 2027.13 Under the ICGS proposal, the modernized Deepwater Coast Guard force would notonly be technologically advanced, but each asset in the system would also be fully integrated in astate-of-the-art command, control, communications, computers and intelligence, surveillance,and reconnaissance system commonly referred to as C4ISR. In June 2002, the Coast Guard awarded to ICGS an initial contract for designing,building, integrating, and testing the assets in the system. At this early stage, most of the workwas for system and asset design and testing, including specifying performance standards for thesystem and each of the planned assets. Under the contract terms, ICGS had full technicalresponsibility for designing and constructing all Deepwater assets, and for deciding whethercontract components should be put out for competitive bids in second tier contracts. Given the uncertainty faced by both the Coast Guard and ICGS in producing a complexproduct, or system of products in this case, the two parties entered into an incomplete contract togovern the exchange – a performance-based indefinite delivery, indefinite quantity (IDIQ)contract. In the simplest terms, an IDIQ does not specify a firm quantity of products or the tasks13 ICGS proposal included five new sea vessels, two fixed-wing aircraft, two helicopters, and one unmanned aerialvehicle. The other assets were upgrades. 14
required to produce them. Instead, it must either specify a minimum or a maximum number ofproducts and some end-point for termination of the agreement.14 The Coast Guard and ICGSagreed to an initial IDIQ contract for five years, renewable in five year increments for a total oftwenty-five years. Within this contract architecture, ICGS and the Coast Guard agreed tonegotiate individual task orders for specific work to be performed (e.g. designing, building andtesting three National Security Cutters, the largest class of ships in the Coast Guard fleet). Thecontract specified broad goals in the form of the general assets categories, the performancespecifications for the assets, and a ceiling on the number of assets within each class. It did notspecify a minimum number of assets (a floor) or the actual design specifications of the assets.The latter was to be worked out through a variety of governance mechanisms, notably individualtask orders and Integrated Project Teams (IPT) for each of the asset acquisitions. The IPTs werecollaborative governance mechanisms that brought together ICGS personnel, relevantsubcontractors, and Coast Guard officials to negotiate and decide which technologies would beused to meet overall mission requirements. In short, it created vehicles and venues fornegotiation and renegotiation, but did not specify the rolls, responsibilities and authorities ofvarious actors in these processes.Strategy Selection, Nature and Payoffs With the IDIQ contract architecture, the individually negotiated task orders, and the IPTsto govern the details of the design and production process, both ICGS and the Coast Guardclaimed that they had cemented a “partnership” rather than entering into an arms length14 See e.g. FAR Subpart 16.5 – Indefinite-Delivery Contracts(http://www.arnet.gov/far/current/html/Subpart%2016_5.html#wp1093133) 15
transactional relationship. At they outset, they signaled to each other that they would“cooperate” in game theory terms. At this point unexpected external forces changed the dynamic of Project Deepwater. Theterrorist attacks of September 11, 2001 spurred the assignment of the Coast Guard to the newlycreated Department of Homeland Security (DHS). This added new mission requirements to theCoast Guard’s already extensive panoply of missions; the Coast Guard was now responsible forcollaborating with other government agencies to help prevent future terrorist attacks on U.S. soil.Not only did Coast Guard’s assignment to DHS create mission strain, but it also increased CoastGuard’s “operational tempo”, or the speed at which Coast Guard had to prepare and conduct itstasks and functions. 15 In response to these external events, Coast Guard acquisition personnelmade significant changes to the overall Deepwater program, both by adding new assets andchanging some of the performance requirements for planned asset acquisitions. For example,post 9/11 the National Security Cutter class of surface assets had to be able to withstand anuclear, biological, and chemical attack, a non-trivial change. Perhaps operating under the assumption that in their “partnership” with ICGS thesechanges could be made easily, Coast Guard elected not to engage ICGS in their discussionsabout modifying the Deepwater program. This was most apparent in the IPT meetings whereICGS personnel chaired and ran the coordinating and planning decision making process, andCoast Guard personnel in attendance did not actively participate.16 Ultimately, Coast Guardacquisition personnel simply delivered ICGS a new scope of work. ICGS agreed to make theadditions and changes, but indicated that production time-lines would lengthen and that there15 US GAO Testimony: “Homeland Security – Challenges Facing the Coast Guard as it Transitions to the NewDepartment”, February 2003. GAO-03-467T16 US GAO Testimony: “Coast Guard – Preliminary Observations on the Condition of Deepwater Legacy Assets andAcquisition Management Challenges”, June 2005. GAO-05-651T 16
would be additional costs. Faced with an increased “operational tempo” and new missionrequirements, Coast Guard had little choice but to agree. The “partnership” arrangementtransformed quickly into a transactional arrangement with ICGS billing Coast Guard for everychange or modification to the existing agreement, and considerable confusion about what ICGShad agreed to produce. One of the best examples of the failed promise of “partnership” was the 123-foot cutter,one of the first assets to be delivered. Prior to 9/11 the Coast Guard had planned to acquire anew fleet of fast response cutters to be delivered by 2018. After 9/11, the Coast Guard elected tomodify its existing 110-foot cutter, primarily by adding 13 feet of deck and hull. The new taskorder with ICGS called for all 49 existing cutters to be modified. The production and delivery ofthese assets quickly became plagued by a series of problems. For one, Coast Guard agreed to anICGS proposal to develop a composite hull, a novel, but untested approach to retrofitting theboat. Soon after the delivery of the first eight 123-foot cutters, testing and review of the patrolboat’s structure revealed hull buckling, potentially compromising the viability of the asset. Inaddition, the Coast Guard identified 22 of the 110-foot cutters that, due to unexpectedly severehull corrosion, required additional inspection and repair separate from the Deepwatermodification plans. Further, Coast Guard officials had four cutters in operation in the PersianGulf, which made them unavailable for modification.17 Both sides expressed confusion aboutthe status of the cutter modifications, the hull repair program, and the overall schedule. By 2006,the Coast Guard decided to dry-dock the 123-foot cutters, decommissioning them from the CoastGuard fleet.1817 US GAO Report. March 2004. “Contract Management: Coast Guard’s Deepwater Program Needs IncreasedAttention to Management and Contractor Oversight”. GAO-04-38018 U.S. Coast Guard Press Release. Nov. 30, 2006. “Coast Guard Suspends Converted Patrol Boat Operations”.http://www.piersystem.com/go/doc/786/138897/ 17
Both sides got less than anticipated from the first five-year installment of the IDIQcontract. The Coast Guard faced program delays, higher costs, and received a non-operationalasset. The overall acquisition schedule lengthened from 25 to 30 years and the plannedacquisition costs jumped from $17 billion to $24 billion.19 ICGS faced unexpected renegotiationcosts due to repeated changes to the original agreement, and the reputations of both NorthrupGrumman and Lockheed Martin suffered as a result of perceived poor performance. Thepromise of a win, win “partnership” deteriorated into a lose, lose transaction. What explains this outcome? Some evidence points to the behavior of the Coast Guard –repeated major changes to task orders and a failure to claim a role in the IPTs. Other evidencepoints to the behavior of ICGS – the performance test failure of the composite hull on the 123-foot cutters and its dominance of the IPTs. Finally, some evidence points to unexpected externalevents – 9/11 and its aftermath. In response, both parties lack trust in the other and fear“defection” moving forward, although neither party is able to determine whether currentoutcomes are a result of the strategy and behavior of the other party or exogenous forces outsideeither party’s control. Now other forces are at work, most notably more aggressive scrutiny froman array of overseers (e.g. DHS’s Inspector General, the Government Accountability Office, theCongressional Research Service). A changeover from a Republican to Democratic majority in2004 has intensified oversight; Congressional Democrats are particularly incensed by what theyperceive to be mismanagement by both parties, and as a result have pushed the Coast Guardaway from a reliance on ICGS as a systems integrator. At the moment, the Coast Guard is in theprocess of standing up its own system engineering and acquisition capacity for ProjectDeepwater in an internal entity called the Acquisition Directorate. The Coast Guard remains19 US GAO Report. April 2006. “Coast Guard: Changes to Deepwater Plan Appear Sound, and ProgramManagement Has Improved, but Continued Monitoring is Warranted.” GAO-06-546. 18
“locked in” with Lockheed Martin and Northrup Grumman to some extent, but they are movingaway from their integrated partnership back towards a piece-by-piece acquisition approach.Conclusion The game theory model and illustration of the Coast Guard’s Project Deepwater wepresented so far suggests complex contracting is a dire endeavor. Complex contracts are highlyuncertain, costly to negotiate and execute, and obfuscate accountability. Under the prisoners’dilemma, win-win cooperation is irrational and lose-lose outcomes most likely. Uncertaintyabout nature’s contributions to payoffs mean even the parties in the contract do not knowwhether contract outcomes stem from misfortune, the other party’s malfeasance, or their ownmismanagement. In the case of Deepwater the default response of both Coast Guard and ICGSis to assume that the other player “defected”. Overseers, notably Congressional oversightcommittees, have made a similar determination and have taken steps to push Coast Guard into a“defect” strategy in future rounds. As our analysis shows though, while there’s evidence tosuggest that one or both of the exchange parties are partially at fault, nature played a strong handin either producing the negative outcomes or pushing the exchange parties towards a lose, loseoutcome. In moving forward, the default response shouldn’t necessarily be to position CoastGuard and whatever sellers it engages into a rigid defect, defect posture almost. Because complex contracts are prone to renegotiation, the “shadow of the future” opens awealth of cooperative strategies, allowing contract parties to turn lose-lose conflict into win-wincooperation (e.g. Heide and Miner, 1992). Moreover, these cooperative strategies enhanceoverseer’s ability to hold people accountable in the contract process. In the case of ProjectDeepwater, the multi-stage architecture of the contracting process allows for this kind of 19
renegotiation. While the projection is that it will take 30 years to deliver all the componentDeepwater assets, the Coast Guard and ICGS did not formally commit to a 30-year agreement.Instead the contract is structured in five year increments. The asset specific nature of theproducts means that the Coast Guard faces a thin market of alternative suppliers, but at aminimum the overall contract arrangement allows for both exit and renegotiation. The shadowof the future is real here, and there are significant consequences to both parties in terms of costsand reputation from spiraling away from a partnership arrangement towards a more antagonisticarms length transactional relationship. In addition, Congress’ increasing oversight andinvolvement in the Deepwater program demonstrate how authorizers can take steps to change therules of the game. At present, Congress is pushing the Coast Guard towards more of a prisoner’sdilemma game, away from a cooperative game, but it doesn’t have to be so. Congress couldinstead take steps to foster cooperation albeit while increasing the level of information availableto all parties to insure accountability. For example, rather than forcing Coast Guard to moveaway from an integrated relationship with ICGS to a more differentiated purchasingarrangement, Congress instead encourage Coast Guard could build on recent efforts to enhanceCoast Guard’s role in the IPTs and rely more extensively on third party certification of productdesign and delivery. One of the primary reasons to take such steps is that there are significantopportunities to capture knowledge and information from the first round and apply it tosubsequent rounds of contracting. The practice of complex contracting need not be so dire. 20
ReferencesAkerloff, G. (1970). The Market for Lemons: Qualitative Uncertainty and the MarketMechanism. Quarterly Journal of Economics, 84(3): 488.Bajari, P., and Tadelis, S. (2001). Incentives versus transaction costs: A theory of procurementcontracts. Rand Journal of Economics, 32(3):387-407.Brown, T., Potoski, M., and Van Slyke, D. (2008). Trust and Contract Completeness in thePublic Sector. Local Government Studies, 33 (4): 607-623.David, P.A. (1985). Clio and the Economics of QWERTY. The American Economic Review, 75(2): 332-337.Heide, J., and Miner, A. (1992). The Shadow of the Future: Effects of Anticipated Interactionand Frequency of Contact on Buyer-Seller Cooperation. Academy of Management Journal 35(3): 265-291Heinrich, C.J. (1999). Do government bureaucrats make effective use of performancemanagement information? Journal of Public Administration Research and Theory, 9(3):363-394.Hutton, J. (2008). Presentation on Contracting to MPA students, Maxwell School of Citizenshipand Public Affairs, April 2. Data based on GAP Analysis of data from the Federal ProcurementData System.Kelman, S.J. (2002). Contracting. In Salamon, Lester M., The tools of government: A guide tothe new governance. New York, NY: Oxford University Press.Martin, L.L. (2004). Performance-based contracting for human services: Does it work?Administration in Social Work 29 (1): 63–77.O’Looney, J.A. (1998). Outsourcing state and local government services: Decision-makingstrategies and management methods (Westport, CT: Quorum Books).Sclar, E.D. (2000). You don’t always get what you pay for: The economics of privatization.Ithaca, NY: Cornell University Press.Weimer, D., and Vining, A. (1999). Policy Analysis: Concepts and Practice (Upper SaddleRiver, NJ: Prentice Hall). 21
Williamson, O. (1996). The Mechanisms of Governance. New York: Oxford University Press. 22
Figure 1: The Contracting Game Phase I: Phase II: Phase III: Phase IV: Negotiation Strategy Selection Nature Payoffs Buyer Seller Positive (#, #) Cooperate Negative (#, #) Cooperate Positive (#, #) Defect Negative (#, #) Incomplete Contract Positive (#, #) Cooperate Negative (#, #) Defect Positive (#, #) Defect Negative (#, #) 23
Figure 2: The Impact of Nature Phase II: Phase III: Phase IV: Strategy Selection Nature Payoffs Buyer Seller Buyer Seller Mutual Gain Positive 3+2=5 3+2=5 10 Cooperate Negative 3 + -2 = 1 3 + -2 = 1 2Cooperate Positive 1+2=3 4+2=6 9 Defect Negative 1 + -2 = -1 4 + -2 = 2 1 Positive 4+2=6 1+2=3 9 Cooperate Negative 4 + -2 = 2 1 + -2 = -1 1 Defect Positive 2+2=4 2+2=4 8 Defect Negative 2 + -2 = 0 2 + -2 = 0 0 24
Appendix I: The Basic Prisoner’s Dilemma Game Buyer Strategy Cooperate Defect Cooperate 3, 3 1, 4 Seller Strategy Defect 4, 1 2, 2 25
Contracting in the Public Sector ... price elements are important in public sector contracting. ... the actual procedure should only be used in complex ...
Complex contracting in the public sector: Managing relations and negotiating contracts in the absence of market discipline Posted by procureinsights on Feb ...
Energy Performance Contracting ... Public Sector Energy Saving Potential: ... (selection criteria, bidding procedures, complex projects, ...
Complex Contracting: ... "Our experience with public sector contracting for complex products and services suggests that we are still in the steep ...
This paper was was written by Trevor Brown (John Glenn School of Public Affairs, The Ohio State University), Matthew Potoski (Department of Political ...
Complex contracting in the public sector: Managing relations and negotiating contracts in the absence of market discipline
Contracting in Complex ... Developing a Contracting Framework for Security Sector ... Dr. Van Slyke’s research focuses on government contracting, public ...
The public sector is the part of the economy concerned ... The organization of the public sector (public ... Complete outsourcing or contracting ...
The public sector spends over £150 billion a year ... procurement policies to which contracting authorities are ... An Introduction to Public ...