Professional Documents
Culture Documents
Circle 2
Circle 2
com
Received 18 January 2008; received in revised form 3 April 2008; accepted 29 May 2008
Abstract
Government acquisitions requiring R&D efforts are fraught with uncertainty. The risks are often mitigated by employing a multi-
stage competition, with multiple vendors funded initially, until a single successful vendor is selected. While decision-makers recognize
they are using a real options approach, analytical tools are often unavailable to evaluate optimal decisions. We present an efficient sto-
chastic dynamic programming approach that public sector acquisition managers can use to determine optimal vendor selection strategies
in those competitions where Technology Readiness Levels (TRLs) are the measure of progress. We then use examples to demonstrate the
proposed approach and provide illustrative numerical results.
2008 Elsevier Ltd and IPMA. All rights reserved.
Keywords: Real options; Management of science/technology; Project management; Optimization; Public sector public acquisition; Dynamic program-
ming
Decision Milestones
A B C FOC
technologies are evaluated to determine which, if any, ven- ming is a standard method of evaluating decisions under
dors are selected to continue R&D and capability develop- uncertainty, our paper is unique in the kind of decisions
ment efforts. Milestone C decisions will typically evaluate that we consider. Real options models typically demon-
finished prototypes and result in a final down-select to a strate the increased benefits of managerial flexibility that
single winning vendor to commence a low rate of initial can be achieved through the inclusion of additional
production (LRIP) of the fielded capability. It is important options. This makes these models an ideal approach for
to note that the winning vendor may be selected for criteria evaluating the dynamic investment decisions in R&D port-
other than obtaining the highest or most robust technolog- folios, where the numbers of distinct options grow over
ical maturity, such as possessing the technology with the time as R&D projects progress. However, the vendor selec-
lowest expected total cost or development schedule, or hav- tion problem is quiet different in that the acquisition man-
ing the highest probability of successful implementation ager starts with many different options and then chooses to
conditional upon their current level of technological potentially reduce the number of options as the project
maturity. progresses. This decreasing options problem has been
While these multi-stage, multi-vendor competitions have ignored by the literature and the suggested solution meth-
proven useful for mitigating technical risk, acquisition odology constitutes a useful, practical approach for devis-
managers must address a number of key questions to effi- ing optimal vendor selection strategies. Moreover, using
ciently employ this strategy: How many vendors should our approach, acquisition managers may find optimal
be initially funded? How many stages? How should funding strategies that would not likely have been considered with-
be spread between stages? Which vendor should be funded out formally modeling the acquisition’s options.
after each decision point? The answers to these questions
present difficult tradeoffs that must be faced. For example, 2. Public sector R&D acquisitions
are more vendors, theoretically increasing the range of
technical alternatives, or fewer, better-funded vendors Unfortunately, while there exists a robust literature on
more likely to increase the probability of successfully the use of real options to manage uncertainty in R&D pro-
acquiring the desired capability on time and within budget? jects [6–14], this literature fails to account for the peculiar-
Should more funds be spent in the R&D phase, ensuring a ities of public sector R&D acquisitions. This is not to imply
more robust technological solution, or in the product that the technical risks in public sector R&D projects are
development phase, increasing the likelihood of a smoother somehow different than those encountered in private sector
implementation? Should the high-cost, mature technology R&D efforts. For example, the likelihood that a specific,
vendor be selected over the low-cost, less mature technol- scientific breakthrough occurs or whether developmental
ogy vendor as the winner? Of course, the answers to these subassemblies can be successfully integrated according to
questions depend upon the precise nature of the given the system’s initial design are common to both the public
acquisition program. However, a lack of formal models and private sectors. Rather, it is the relative rigidities of
to address the optimal design of these competitions typi- the public sector acquisition process that influence the
cally leads to ad hoc, qualitative solutions to these available approaches for mitigating the various technical
questions. risks that may occur during an R&D project. Commercial
Real options valuation techniques provide an analytical R&D projects are largely internal to the firm with direct
framework to find optimal solutions to these problems. In management oversight to guide and direct as technical
fact, acquisition managers often recognize that they are issues arise. While a portion of public sector R&D is per-
employing a real options approach by structuring such a formed in government facilities, a majority of the R&D
competition [1], [5]. The contribution of this paper is the required for new capabilities is either sourced to private
formulation of a stochastic dynamic program that public vendors or simply embedded within the contracts issued
sector acquisition managers can use to determine optimal for the completed capability. This significantly reduces
vendor selection strategies in such multi-stage, multi-ven- the public sector’s ability to directly mitigate technical risks
dor competitions. Though stochastic dynamic program- as they occur, subject to the provisions incorporated and
J.M. Eckhause et al. / International Journal of Project Management 27 (2009) 365–377 367
relative completeness of the vendor’s contracts. Therefore, processes allow flexibility through the use of continuation,
in a public sector context the question at hand is rarely how improvement, delay, or abandonment options at each stage
to mitigate specific technical risks that may occur in an of the development depending upon program value at each
R&D effort, but alternatively, how to mitigate the likeli- stage [6,7,14,18,19]. For the purposes of this paper, we
hood of technical risks preventing a successful project define a public R&D project as one that will deliver a
conclusion. non-market traded good or service upon completion and
In addition to the rigidities present in public sector R&D has been deemed sufficiently necessary that project comple-
efforts, there are often specific uncertainties that private tion will be funded. While all government R&D acquisi-
sector efforts do not face. Public sector acquisitions are fre- tions possess some cost or schedule ceiling through which
quently non-market traded goods which are often difficult program abandonment becomes an option, because there
to value. While techniques such as contingent valuation are typically no directly observed market payoffs, these
have been developed to meet this challenge [15], public ceilings are not easily definable. Further, it is not infre-
decision makers still must reconcile multiple, divergent val- quent for programs to continue in the face of tremendous
uations as both proponents and opponents of a given cost and schedule overruns compared to those in the pri-
acquisition submit their respective estimates. Regardless vate sector since government investment decisions are often
of the valuation method employed, the selection of an determined by political or other reasons [17]. For example,
appropriate discount rate, and whether this rate should Drezner et al. [20] find that major defense acquisitions
vary over the period of performance for lengthy acquisi- between 1960 and 1990 experienced an average of 20% cost
tions, continues to provide spirited debate among policy growth from their initial cost estimates and with a substan-
makers. This is not to imply that public sector R&D acqui- tial percentage of programs exceeding their initial estimates
sitions have been ignored by the real options literature. by as much as 50%. Therefore, we choose to evaluate the
Vonortas and Hertzfeld [16] use a real options approach likelihood of a given R&D program successfully develop-
to attribute social benefits to traditional net present value ing a desired capability subject to the total budget available
(NPV) calculations of public sector R&D investments. Post to the acquisition manager, while assuming that the man-
et al. [17] considers alternative implementation options for ager has no incentive to either conserve his budget or aban-
an FAA program. However, these models do not directly don development until the budget is exhausted. The
incorporate the technical risk inherent in such projects. abandonment option can be readily incorporated into
Our paper addresses two areas largely ignored in the real our model, but as it has been well studied by the real
options literature. First, we consider how real options can options literature, we find it adds no further qualitative
mitigate risk and uncertainty due to variability in project insights and focus instead on the multi-stage competition
performance and schedule for a non-traded public good. at hand.
Most studies of real options valuation techniques in
R&D projects have considered risk and uncertainty to 3. The multi-stage competition as a real options problem
occur in the project’s market payoff. Notable exceptions
are [9], [12,13]. Considering multi-stage development pro- A basic call option represents a right, but not an obliga-
jects where managers can consider continuing, improving, tion, to make a purchase at a future date. The price paid to
or abandoning development at each decision point, purchase this right, or option, is referred to as the option’s
Huchzermeier and Loch [9] evaluate changes in option val- strike price. The price paid to exercise a purchased option
ues in the presence of five types of operational uncertainty: is the option’s exercise price, but this price is only paid if
market payoff variability, budget variability, performance the option proves to be valuable at a later date, thus limit-
variability, market requirement variability, and schedule ing the buyer’s risk to the amount paid to initially purchase
variability. They conclude that the value of increased man- the option. Multi-stage, multi-vendor R&D competitions
agerial flexibility through the use of real options increases are similar in structure. The cost of issuing initial technol-
with increased variability in market payoffs and budgets ogy development contracts to a vendor represents the pur-
but may actually decrease in the presence of the other types chase price for that vendor option. A given vendor option
of uncertainty discussed. Building off of the same formula- is exercised upon the award of a subsequent contract to the
tion, Santiago and Vakili [12] find different results, with vendor to continue development of the actual capability.
uncertainties beyond market payoff providing ambiguous The exercise price of this option is the amount of funding
results for the value of increased managerial flexibility. each winning vendor receives at each subsequent stage. In
Santiago and Bifano [13] consider the application of a mul- the simplest two-stage problem, the competition reduces
tidimensional real options model that considers multiple to the selection of the optimal portfolio of simple call
types of operational uncertainty toward the development options to purchase and then exercise in the next stage.
of a specific product. Fig. 2 demonstrates such a two stage multi-vendor com-
Second, we consider the value of increased managerial petition. If the objective of the acquisition program man-
flexibility in a multi-stage, vendor source selection model ager is to maximize the likelihood of successfully
that does not permit a program abandonment option. Pre- developing a desired capability in time period t=2, the
vious option studies evaluating multi-stage development manager must determine how many and which of the ven-
368 J.M. Eckhause et al. / International Journal of Project Management 27 (2009) 365–377
t=1 Table 1
t=0 t=2
Technology readiness level (TRL) definitions
TRL Definition
1 Basic principles observed and reported
2 Technology concept and/or application formulated
3 Analytical and experimental critical function and/or characteristic
Vendor 1
proof-of-concept
4 Component and/or breadboard validation in laboratory
environment
5 Component and/or breadboard validation in relevant environment
6 System/subsystem model or prototype demonstration in a relevant
environment
7 System prototype demonstration in a space environment
Vendor 2
8 Actual system completed and ‘‘flight qualified” through test and
Uncertain demonstration
9 Actual system ‘‘flight proven” through successful mission
Outcomes
operations
Vendor 3
While the TRL metric has been used by NASA and the
Department of Defense, there are of course other metrics
Fig. 2. Two-stage multi-vendor competition.
that one could employ to gauge the completion level of a
project. These measures might include measures related
to earned value, number of successful prototypes devel-
dor options to purchase in period t=0 and then how many oped or deployed or the like. Moreover, it is also true that
and which of the purchased vendor options to exercise in the stochastic dynamic programming approach we propose
period t=1. If the competition is composed of several deci- can be used beyond these two application areas. Two other
sion stages before the winning vendor(s) are selected, each domains that lend themselves directly to such a methodol-
vendor represents a complex call option, as each subse- ogy include IT management and R&D efforts in low-car-
quent exercising prior to the last stage, also represents an bon technologies for the energy sector. In the first area,
additional purchase of the option. While this may create IT R&D managers may be concerned with fewer or differ-
potential computational problems as the state space ent completion levels (e.g., software system concept, proto-
expands, it does not change the formulation required to type development, alpha- and beta-level versions). In terms
evaluate such problems. Fortunately, current computing of funding R&D efforts for low-carbon technologies to
power is sufficient to address the state spaces required for produce power (e.g., tidal power, advanced solar or wind
many realistic acquisition applications. power), project managers also may have fewer or different
levels of completion. For example, these levels might
4. Technology readiness level include: initial concept (taking into account how related
to existing technology or novel), approval by a government
Before an optimal portfolio of vendor options to pur- regulatory agency, initial disbursement of funding to
chase and exercise can be identified, a metric must be research laboratories and universities, prototype develop-
employed to gauge the success of each vendor’s R&D ment, field deployment, market-ready product.
efforts. A common metric currently employed to assess To mitigate the risks of developing new capabilities,
the maturity of evolving technologies by many government many large-scale, expensive projects do not award a single
agencies, especially NASA and the Department of Defense, contract that will progress from TRL 1 to TRL 9. Rather,
is the Technology Readiness Level (TRL). NASA uses nine the observations and concepts, along with the proof-of-
TRLs to describe the maturity of an evolving technology.1 concept and exploratory research, are usually done first,
The Department of Defense employs a slightly different under smaller contract awards. If proven successful, or if
definition [21], but the essence of the level progression is sufficient progress is made, future contracts are awarded
the same. The general concept behind a TRL progression based on the preliminary success of the earlier TRL
is that at the beginning of technology development, general progression.2
concepts are observed; then, the concepts are developed; Although this multi-stage approach is sometimes used
the prototypes are designed and tested; and then the actual with a single vendor, it naturally leads to the multi-stage,
technology is tested and deployed. Table 1 provides a brief multi-vendor contracts usually employed. For example,
definition of each level, as defined by NASA. during the beginning stages of a project’s TRL progression,
1 2
http://research.hq.nasa.gov/code_y/nra/current/NRA-99-OES-07/ This strategy is adopted by NASA and the Department of Defense but
appendixf.html. applicable to other public sector areas.
J.M. Eckhause et al. / International Journal of Project Management 27 (2009) 365–377 369
the cost of concept-development may be relatively small the TRL progression from one stage to the next) at every
enough that the government agency can award several stage for every vendor, defining these pmfs can be challeng-
simultaneous contracts with a decision point for future ing for many applications. However, many R&D intensive
contracts occurring when vendors are expected to achieve public sector acquisitions, such as aerospace and defense
TRL 6. Each vendor is assumed to choose whichever tech- programs, already produce estimates of TRL success dur-
nology platform best suits its abilities to achieve its desired ing source selection and R&D portfolio funding decisions.
readiness level. Of course, it must be noted that TRL pro- Typically, these are discrete pmfs, such as the probability
gression alone is not a substitute for quality of the work that a program will achieve TRL 6 given a specific schedule
performed. Two competing developers or contractors or level of funding, that are obtained from subject matter
may claim to have ‘‘successfully” reached a certain TRL, experts and historical data [22]. NASA’s Strategic Assess-
but one of the two may be vastly superior to the other. ment of Risk and Technology (START) approach for eval-
We assume this type of judgment is considered in the tech- uating R&D investment decisions uses a peer review
nology readiness assessment [21]. Since TRLs are already process to assign cumulative probability values to different
commonly used for assessing technological maturity in performance range points as well as probabilities of project
multi-vendor competitions, we will also use this measure acceptance by the stakeholder once TRL 6 is achieved [23].
to assess competing vendors within our formulation of Recognizing the need for better estimates for the likelihood
multi-vendor, multi-stage acquisitions. that a technology development project successfully meets its
milestones, NASA Ames Research Center is currently devel-
5. Mathematical formulation oping a Technology Development Risk Assessment
(TDRA) tool to calculate TRL transitional probabilities as
The general framework for our multi-vendor, multi- a function of time and budget [24]. However, as current pub-
stage competition is as follows. We wish to potentially fund lic sector R&D funding decisions use some form of qualita-
a number of vendors, each with their own costs and prob- tive or Delphi approach [25] to evaluate the probabilities of
ability of success over various stages of an R&D acquisi- achieving a few specific program milestones,3 we employ the
tion project so that the probability of achieving a specific simple, discrete pmfs that modelers will most likely obtain
predetermined level of success for the overall R&D project from subject matter experts and historical data.
is maximized. The set of potential vendors is represented by We will develop our formulations for determining the
I. Using TRL as the measure of desired R&D success for optimal portfolios of real options to purchase and exercise
each vendor in each stage, we assume that we wish to max- in multi-vendor, multi-stage competitions by initially
imize the probability of achieving TRL 8 by the end of the examining a fairly restrictive version of the problem. We
acquisition process, as TRL 9 is usually reserved for pro- will then develop a formulation that relaxes many of the
ven, fielded systems, i.e., post initial acquisition. We should initial assumptions to better accommodate realistic acquisi-
note that many other objectives are possible within this tion programs. The pmf for each vendor is strictly deter-
framework, such as minimizing expected cost or expected mined by the funding decision for that vendor; we
development schedule. Furthermore, let us assume there assume the funding decisions for the other vendors do
are certain funding decision periods that allow us to assess not impact that pmf.
the level of maturity (success) of each funded vendor. There
Problem 1. In this version of the multi-vendor, multi-stage
are s time periods in which decisions are made and an addi-
competition, we assume that the total budget available to
tional final time period (s + 1) in which outcomes are real-
the acquisition manager at each stage is fixed and that the
ized. At each of these time periods, t, we can decide
potential funding level for each vendor at every stage is
whether or not to continue funding of the vendors cur-
also fixed at some predetermined level. The only decision
rently funded (or even, by how much we should fund them)
available to the acquisition manger is whether or not to
in the subsequent funding cycle.
fund any specific vendor(s) at each stage. We define the
We assume each vendor starts at a certain TRL, and can
following state variables and data for our formulation:
progress along the way according to a set of transitional
probabilities relating to funding. Thus, the state of any ven-
Let Cit 2 Si be the state of vendor (or contractor) i at
dor at the beginning of any time period is a value in the set
time period t; we assume that Si = S = {1,2,3,4,5,6,
S = {1,2,3,4,5,6,7,8,/}, where 1,. . .,8 correspond to the
7,8,;}"i.
current TRL achieved and / corresponds to no longer
Let Xit 2 {0,1} be the decision variable of whether to
being funded (or possibly having been never funded). We
fund vendor i at time period t.
allow for the possibility that the vendor may reach ‘‘suc-
Let ait represent the cost of funding vendor i at time per-
cess” (i.e., TRL 8) before the final stage (s + 1). Whether
iod t.
or not this is possible for a particular instance of this prob-
Let Bt represent the R&D budget available for time per-
lem can be specified by the probability mass functions
iod t.
(pmfs) for the transitions of each vendor. While we assume
that we know the transitional probabilities from each state
3
to every other state (i.e., the probability mass function of Often just the probability of achieving TRL 6 at some specific point.
370 J.M. Eckhause et al. / International Journal of Project Management 27 (2009) 365–377
As previously stated, we assume we are provided the This condition assumes no ‘‘consolation” prize for a ven-
state transition probabilities. In other words, given for dor reaching TRL 7, for instance. If the dynamic program
any state s1 and any state s2, we know the value of is solved optimally, the probability that the goal is accom-
P{Ci,t+1 = s2 jCit = s1 ,Xit = 1}. Many of the probabilities plished by the final time period is determined by the tran-
are obvious from the problem setup or simple assumptions. sitional probabilities and the R&D budgets for each time
For instance, "s1 2 Sn{8}, P{Ci,t+1 = /j Cit = s1,Xit = period (i.e., B1,B2,...Bs).
0} = 1. In other words, a vendor not funded at time period
t will necessarily be in state / in the next time stage (and all 5.1. Problem 1 Example
subsequent stages), unless that vendor has already attained
a TRL of 8. This also implies that P{Ci,t+1 = /j Cit = / Suppose that the National Reconnaissance Office
,Xit} = 1. If a vendor reaches TRL 8 (or ‘‘success”) before (NRO) decides to acquire a satellite with new sensing capa-
the final time period, then that vendor remains in the suc- bilities substantially out of the reach of current technology.
cess state, i.e., P{Ci,t+1 = 8j Cit = 8,Xit} = 1. Assuming the NRO is employing the DOD 5000 acquisi-
At time period t, the state of the system can be described tion framework presented in Fig. 1, they decide to pursue
as all the combinations of states. That is, the following acquisition strategy. The NRO will request
! ! proposals from four vendor teams that detail their techni-
Y Y
Ct 2 Si ¼ S ð1Þ cal approach, proposed schedule, and cost bids for per-
i2I i2I forming the R&D required to invent the new sensing
capability. Each vendor will also submit a similar proposal
For these combinations of states at time period t, we can
and bid for actually developing the satellite. At the Mile-
choose a set of feasible funding decisions:
( I P
stone A decision, the NRO will determine which vendors
X t 2 f0; 1g : ait X it 6 Bt will actually receive a technology maturation contract to
X ðC t Þ ¼ i2I ð2Þ invent the new capability. The NRO will purchase a simple
X it ¼ 0 ifC it ¼ / 8i 2 I
call option with each of these initial contracts it awards. At
The second constraint indicates that we do not fund a con- a predetermined Milestone B decision point, the NRO will
tractor that is already in state /. Explicitly adding the con- evaluate each of the selected vendors’ prototypes and exer-
straint Xit = 0ifCit = 8"i 2 I is unnecessary by an cise one or more of their previously purchased options by
optimality argument, since it is implicitly considered in awarded a follow-on contract to the winning vendor(s)
the objective function of maximizing overall project suc- selected to build the satellite. The NRO will decide whether
cess. That is, letting Xit = 1 when Cit = 8 does not increase to launch the satellite at the Milestone C decision point, at
the objective function, but rather decreases the available which time it is fielded. In essence, we are considering an
budget. Nevertheless, in order to preserve budget responsi- acquisition with two stages and four vendors. We’ll assume
bility, we can include such a constraint. that the acquisition has already reached a certain technical
If we wish to choose the optimal funding strategy to maturity, so each vendor begins at TRL 4, with the goal of
maximize the probability of at least one vendor reaching reaching TRL 8 by the end of the second stage. The budget
TRL 8, or success, then we can solve for the {0,1} decision available to the NRO for the first and second stages are
variables by formulating it as a dynamic program. We for- fixed at $10 million and $20 million, respectively (i.e.,
mulate the problem as B1 = $ 10 and B2 = $20), with decision makers facing the
V t ðC t Þ ¼ maxX t 2X ðCt Þ EfV tþ1 ðC tþ1 ÞjC t ; X t g t ¼ 1; . . . ; s ‘‘use it or lose it” constraint typical of government budgets.
Thus, with no budget flexibility or incentive to withhold
ð3Þ funds, the NRO’s acquisition managers will choose to
In this dynamic program, the calculation of the expectation exhaust their entire budget in each stage. Table 2 shows
depends on the distribution of Ct+1 conditioned on Ct and each vendor’s stated costs for each stage. The conditional
Xt, which we previously assumed as given. We note that transitional probabilities for each vendor are presented in
maximizing the above objective function will always have Tables A1, A2 in Appendix A
a solution, since we are considering a set of feasible funding Traditionally, the acquisition managers would construct
solutions over a discrete set of possibilities. Therefore, a capability or requirements matrix and assign appropriate
complete enumeration—while not always desirable in prac- qualitative and quantitative values to each of the vendors
tice—would guarantee an optimal solution.
Recall that the goal is to maximize the probability that
at least one vendor achieves TRL 8. We assume if all ven- Table 2
Vendor costs for each stage (millions)
dors fail to reach TRL 8, then we have failed to meet the
goals of the R&D acquisition. Thus, we can state the Vendors Stage 1 Stage 2
boundary condition of the dynamic program as 1 $3.5 $4.0
2 $3.7 $6.9
1 if C i;sþ1 ¼ 8 forsome i 2 I 3 $5.0 $10.4
V sþ1 ðC sþ1 Þ ¼ ð4Þ
0 otherwise 4 $2.3 $6.3
J.M. Eckhause et al. / International Journal of Project Management 27 (2009) 365–377 371
for comparison. Vendor selection in each stage would then t=2 t=3
typically be determined through either a weighted or un-
weighted Delphi approach. While this approach allows TRL 4
p = 0.0
TRL 8
acquisition managers the ability to carefully consider the
qualitative merits of each vendor, it fails to ensure that
the number and mix of vendors selected actually maximizes 3
0.
=
the probability of a successful acquisition given the NRO’s t=1 p
p = 0.1
p = 0.1 TRL 5 TRL 8
budget constraints. We determine the optimal portfolio of
TRL 4
vendor options to purchase and exercise by solving p=
0.4
maxX t 2X ðCt Þ EfV tþ1 ðC tþ1 Þ j C t ; X t g. The solution maximizes p
= p = 0.3
the expected value of the value function, which is the prob- 0.
15 TRL 6 TRL 8
ability that at least one vendor achieves TRL 8. The results
p
=
of the dynamic program for this two-stage, four-vendor
0.
05
p = 0.5
problem are that the acquisition manager purchases TRL 7 TRL 8
options, by awarding contracts, on both Vendor 3 and 4
in the first stage. As it turns out, both options would be p = 1.0
exercised in the second stage with the award of follow-on TRL 8 TRL 8
p = 0.7
TRL 7 TRL 8 ciently approximated discretely. In reality, transitional
probabilities for TRL progression would exist for only a
p = 1.0 few funding levels, since they rely heavily on subject matter
TRL 8
TR TRL 8
expertise and historical data. So, the number of funding
levels for each vendor and stage is limited to the number of
Fig. 3. Vendor 3 transition success probabilities. probability mass functions one is able to generate with
372 J.M. Eckhause et al. / International Journal of Project Management 27 (2009) 365–377
reasonable accuracy. We define the following state vari- 5.2. Problem 2 Example
ables and data for Problem 2:
We now consider the more robust problem outlined in
Let B1 denote the fixed budget available to the decision Problem 2. Reconsidering our hypothetical NRO satellite
maker at the beginning of the R&D acquisition process. procurement, we will assume that there are still two stages
Let Bt be the budget remaining at time period t. and four vendors, but instead of one funding level, the
Let aitl denote the cost of funding vendor i at time period NRO requests proposals from each vendor at different
t at level l. funding levels, to insulate the acquisition from pending
Let Xitl 2 {0,1} be the decision variable of whether to budget cuts. Of course, the degree of technological matu-
fund vendor i at time period t at level l. rity achieved will likely be reduced at lower levels of fund-
ing, but this will be reflected in the TRL transition
Extending the assumptions from Problem 1, we assume probabilities associated with each funding level. For our
given for any state s1 and any state s2, we know the value of example, we assume that the NRO receives as many as
P {Ci,t+1 = s2j Cit = s1, Xitl = 1}. The state of the system at three funding options (four, if one counts deciding not to
time period t is fund that vendor) for each vendor. We have retained the
! original four vendors, but assume that each of the vendors
Y can also be funded at some specific higher or lower level of
ðC t ; Bt Þ 2 S Rþ ð5Þ
i2I
funding. Other than incorporating the additional funding
levels, we will assume the NRO’s acquisition strategy
and the feasible decisions and budget transition at time per- remains unchanged. Table 3 shows the costs for funding
iod t can now be written as at the low, medium and high levels for each of the vendors
8 P in both time periods. Again, each vendor begins at TRL 4.
>
> ðX t ; Btþ1 Þ 2 f0; 1gIL Rþ : aitl X itl 6 Bt The conditional transitional probabilities for each vendor
>
> i2I;l2L
>
>
>
> P at each funding level are presented in Tables B1 and B2
>
> X itl 6 1 8i 2 I
>
> in Appendix B. However, we also assume that the NRO’s
< l2L
Table 3
Vendor costs at three funding levels in each stage (millions)
Vendors Stage 1 low Stage 1 medium Stage 1 high Stage 2 low Stage 2 medium Stage 2 high
1 $2.5 $3.5 $5.0 $3.0 $4.0 $5.0
2 $3.2 $3.7 $5.2 $6.9 $6.9 $7.9
3 $3.0 $5.0 $9.2 $10.4 $10.4 $10.4
4 $1.8 $2.3 $2.8 $5.3 $6.3 $7.3
program managers are able to retain unused funds, we pro- information arrives. As our results show, the ability to
duce an optimal strategy that would not likely have been include budget flexibility and multiple funding options in
discovered using the current Delphi based decision process. this example provides a significantly larger value for the
Lastly, with a more flexible budget as well as the allowance objective function in Problem 2.
for multiple funding levels, the success probability
increases from 56% to 71%. 6. Algorithm implementation
Another advantage of using this real options technique
is that the optimal portfolio of options to exercise in the The dynamic program employs the backward induction
second stage can be easily solved after incorporating the method in the standard manner [26]. It begins in the final
realized TRLs after the first stage. This provides additional time period (t = 2). For every C2 (i.e., all possible states
managerial flexibility since the acquisition manager can sig- for the four vendors at the beginning of time period 2)
nificantly alter his or her initial acquisition strategy as new and a given remaining budget, B2, we calculate the feasible
set of actions, X2, that maximizes the probability that at
Vendor 2 Low Vendor 3 Low
least one vendor reaches TRL 8 (i.e., we minimize the prob-
ability that all vendors fail). In other words, for each C2, we
1.0
0.9 solve for
0.8 Y
maxðX 2 ;B3 Þ2X ðC2 ;B2 Þ 1 ð1 PfC i;3 ¼ 8jC it ; X i2l gÞ ð8Þ
Probability
0.7
0.6 i2I
0.5
0.4 The optimal action’s probability of success, given a set of
0.3
0.2 outcomes and remaining budget, becomes the second-stage
0.1 value function. That is, for each feasible
0.0 Q (C2,B2), we calcu-
4 5 6 7 8 late V 2 ðC 2 ; B2 Þ ¼ maxðX 2 ;B3 Þ2X ðC2 ;B2 Þ 1 i2I ð1 PfC i;3 ¼ 8j
TRL C it ; X i2l gÞ.
0.7
0.6 First stage outcomes TRL Probabilities
0.5
0.4
Vendor 1 4 0.20
0.3 5 0.30
0.2 6 0.40
0.1 7 0.10
0.0
8 0.00
4 5 6 7 8
TRL Vendor 2 4 0.10
5 0.20
6 0.50
Vendor 2 High Vendor 3 High
7 0.20
1.0
8 0.00
0.9
0.8 Vendor 3 4 0.10
Probability
0.7
0.6
5 0.10
0.5 6 0.50
0.4 7 0.20
0.3 8 0.10
0.2
0.1 Vendor 4 4 0.30
0.0 5 0.10
4 5 6 7 8 6 0.40
TRL 7 0.15
Fig. B1. Cumulative distribution functions for vendors 2 and 3 8 0.05
374 J.M. Eckhause et al. / International Journal of Project Management 27 (2009) 365–377
In the first stage, for each set of funding actions, X1, and maxðX 1 ;B2 Þ2X ðC1 ;B1 Þ EfV 2 ðC 2 ; B2 ÞjC 1 ; X 1 g
its cost, we find the value that maximizes V1(C1,B1) by X
¼ maxðX 1 ;B2 Þ2X ðC1 ;B1 Þ V 2 ðC 2 ; B2 ÞPfC 2 jC 1 ; X 1 g ð9Þ
calculating C 2 2S
Table B1
Problem 2 first stage transition probabilities
Low first stage outcomes TRL Probabilities Middle first stage outcomes TRL Probabilities High first stage outcomes TRL Probabilities
Vendor 1 4 0.30 Vendor 1 4 0.20 Vendor 1 4 0.20
5 0.20 5 0.30 5 0.20
6 0.45 6 0.40 6 0.30
7 0.05 7 0.10 7 0.20
8 0.00 8 0.00 8 0.10
Vendor 2 4 0.10 Vendor 2 4 0.10 Vendor 2 4 0.10
5 0.20 5 0.20 5 0.20
6 0.50 6 0.50 6 0.40
7 0.20 7 0.20 7 0.25
8 0.00 8 0.00 8 0.05
Vendor 3 4 0.20 Vendor 3 4 0.10 Vendor 3 4 0.00
5 0.30 5 0.10 5 0.10
6 0.30 6 0.50 6 0.40
7 0.10 7 0.20 7 0.30
8 0.10 8 0.10 8 0.20
Vendor 4 4 0.30 Vendor 4 4 0.30 Vendor 4 4 0.20
5 0.20 5 0.10 5 0.20
6 0.30 6 0.40 6 0.40
7 0.20 7 0.15 7 0.15
8 0.00 8 0.05 8 0.05
Table B2
Problem 2 first stage transition probabilities
Low Stage TRL at Probabilities Middle Stage TRL at Probabilities High Stage TRL at Probabilities
secondstage 2 end of secondstage 2 end of secondstage 2 end of
outcomes TRL Stage 1 outcomes TRL Stage 1 outcomes TRL Stage 1
Vendor 1 4 4 0.40 Vendor 1 4 4 0.30 Vendor 1 4 4 0.20
5 4 0.30 5 4 0.40 5 4 0.30
6 4 0.20 6 4 0.20 6 4 0.30
7 4 0.10 7 4 0.10 7 4 0.20
8 4 0.00 8 4 0.00 8 4 0.00
5 5 0.50 5 5 0.40 5 5 0.40
6 5 0.40 6 5 0.35 6 5 0.30
7 5 0.10 7 5 0.25 7 5 0.20
8 5 0.00 8 5 0.00 8 5 0.10
6 6 0.40 6 6 0.30 6 6 0.25
7 6 0.50 7 6 0.50 7 6 0.50
8 6 0.10 8 6 0.20 8 6 0.25
7 7 0.50 7 7 0.40 7 7 0.50
8 7 0.50 8 7 0.60 8 7 0.50
8 8 1.00 8 8 1.00 8 8 1.00
Vendor 2 4 4 0.10 Vendor 2 4 4 0.10 Vendor 2 4 4 0.10
5 4 0.30 5 4 0.30 5 4 0.20
6 4 0.40 6 4 0.40 6 4 0.50
7 4 0.20 7 4 0.20 7 4 0.20
8 4 0.00 8 4 0.00 8 4 0.00
5 5 0.30 5 5 0.30 5 5 0.20
6 5 0.20 6 5 0.20 6 5 0.30
7 5 0.50 7 5 0.50 7 5 0.40
8 5 0.00 8 5 0.00 8 5 0.10
6 6 0.20 6 6 0.20 6 6 0.20
7 6 0.70 7 6 0.70 7 6 0.65
8 6 0.10 8 6 0.10 8 6 0.15
7 7 0.35 7 7 0.35 7 7 0.30
8 7 0.65 8 7 0.65 8 7 0.70
8 8 1.00 8 8 1.00 8 8 1.00
(continued on next page)
376 J.M. Eckhause et al. / International Journal of Project Management 27 (2009) 365–377
Table B2 (continued)
Low Stage TRL at Probabilities Middle Stage TRL at Probabilities High Stage TRL at Probabilities
secondstage 2 end of secondstage 2 end of secondstage 2 end of
outcomes TRL Stage 1 outcomes TRL Stage 1 outcomes TRL Stage 1
Vendor 3 4 4 0.20 Vendor 3 4 4 0.20 Vendor 3 4 4 0.20
5 4 0.40 5 4 0.40 5 4 0.40
6 4 0.20 6 4 0.20 6 4 0.20
7 4 0.10 7 4 0.10 7 4 0.10
8 4 0.10 8 4 0.10 8 4 0.10
5 5 0.40 5 5 0.40 5 5 0.40
6 5 0.35 6 5 0.35 6 5 0.35
7 5 0.15 7 5 0.15 7 5 0.15
8 5 0.10 8 5 0.10 8 5 0.10
6 6 0.30 6 6 0.30 6 6 0.30
7 6 0.40 7 6 0.40 7 6 0.40
8 6 0.30 8 6 0.30 8 6 0.30
7 7 0.30 7 7 0.30 7 7 0.30
8 7 0.70 8 7 0.70 8 7 0.70
8 8 1.00 8 8 1.00 8 8 1.00
Vendor 4 4 4 0.40 Vendor 4 4 4 0.40 Vendor 4 4 4 0.30
5 4 0.40 5 4 0.30 5 4 0.30
6 4 0.10 6 4 0.20 6 4 0.20
7 4 0.10 7 4 0.10 7 4 0.15
8 4 0.00 8 4 0.00 8 4 0.05
5 5 0.50 5 5 0.50 5 5 0.40
6 5 0.30 6 5 0.30 6 5 0.30
7 5 0.15 7 5 0.10 7 5 0.15
8 5 0.05 8 5 0.10 8 5 0.15
6 6 0.40 6 6 0.40 6 6 0.30
7 6 0.35 7 6 0.30 7 6 0.35
8 6 0.25 8 6 0.30 8 6 0.35
7 7 0.55 7 7 0.50 7 7 0.45
8 7 0.45 8 7 0.50 8 7 0.55
8 8 1.00 8 8 1.00 8 8 1.00
to determine the optimal portfolio of vendor options to [5] Glaros G. Real options for defense. Transform Trends 2003;9(June).
purchase and exercise. We find that allowing budget flexi- [6] Yeo KT, Qiu F. The value of management flexibility – a real option
approach to investment evaluation. Int J Project Manage
bility and multiple finding levels may increase the probabil- 2003;21(4):243–50.
ity of program success. While this framework is suitable for [7] Wang J, Hwang W. A fuzzy set approach for R&D portfolio selection
many practical applications, it also serves as a foundation using a real options valuation model. Omega 2008;36:122–30.
for further research, such as considering uncertainty in [8] Dixit A, Pindyck R. Investment under uncertainty. Princeton, NJ:
the budget actually available, in the hopes of addressing Princeton University Press; 1994.
[9] Huckzermeier A, Loch CH. Project management under risk: using the
larger classes of problems. real options approach to evaluate flexibility in R&D. Manage Sci
2001;47(1):85–101.
[10] Perdue R, McAllister W, King P, Berkey B. Valuation of R&D
Appendix A. See Tables A1–A3. projects using options pricing and decision analysis models. Interfaces
1999;29(6):57–74.
[11] Ford D, Sobek II D. Adapting real options to new product
Appendix B. See Table B1, B2. development by modeling the second Toyota paradox. IEEE Trans
Eng Manage 2005;52(2):175–85.
[12] Santiago L, Vakili P. On the value of flexibility in R&D projects.
Manage Sci 2005;51(8):1206–18.
References [13] Santiago LP, Bifano TG. Management of R&D projects under
uncertainty: A multidimensional approach to managerial flexibility.
[1] Ceylan B, Ford D. Using options to manage dynamic uncertainty in IEEE Trans Eng Manage 2005;52(2):269–80.
acquisition projects. Acquisit Rev Quart 2002;Fall:241–58. [14] Trigeorgis L. Real options: managerial flexibility and strategy in
[2] Department of Defense, Risk management guide for DoD acquisi- resource allocation. Cambridge, MA: MIT Press; 1996.
tion. Fort Belvoir, VA: Defense Acquisition University; 1996. [15] Carson R. Contingent valuation: a comprehensive bibliography and
[3] Department of Defense, Mandatory procedures for major defense history. Northampton, MA: Edward Elgar; 2007.
acquisition programs. Washington, DC: Department of Defense. [16] Vonortas N, Hertzfeld H. Research and development project
Rep. Regulation 5000. 2-R, 2001. selection in the public sector. J Policy Anal Manage 1998;17(4):
[4] Department of the Navy. Top eleven ways to manage technical risk. 639–57.
Washington, DC: Office of the Assistant Secretary of the Navy [17] Post J, Bennett M, Knorr D. Real options in capital investment
(RD&A). Rep. NAVSO P-3686; 1998. decision-making: data link case study. Presented at the AIAA 4th
J.M. Eckhause et al. / International Journal of Project Management 27 (2009) 365–377 377
Aviation Technology, Integration and Operations (ATIO) Forum, [22] Weisbin C, Rodriguez G, Elfes A, Smith J. Toward a systematic
2004. AIAA 2004-6238. approach for selection of NASA technology portfolios. Syst Eng
[18] Meier H, N.Christofides G.Salkin. Capital budgeting under uncer- 2004;7(4):285–302.
tainty – an integrated approach using contingent claims analysis and [23] Elfes A, Weisbin C, Manvi R, Adumitoaie V, Lincoln W, Shelton K.
integer programming. Operat Res 2001;49(2):196–206. Extending the START framework: computation of optimal capability
[19] Panayi S, Trigeorgis L. Multi-stage real options: the cases of development portfolios using a decision theory approach. Syst Eng
information technology infrastructure and international bank expan- 2006;9(4):331–57.
sion. The Quart Rev Econom Finance 1998;38:675–92. [24] Mathias D, Goodsell A, Go S. Technology development risk
[20] Drezner J, Jarvaise J, Hess R, Hough P, Norton D. An analysis of assessment for space transportation systems. In: Proceedings of 8th
weapon system cost growth. RAND Report: MR-291-AF, Probabalistic Safety Assessment and Management Conference, New
1993. Orleans, Louisiana, 2006.
[21] Deputy Under Secretary of Defense for Science and Technology. [25] Linstone HA, Turoff M, editors. The delphi method: techniques and
Technology readiness assessment (TRA) deskbook. Washington, DC: applications. Reading, MA: Addison-Wesley; 1975.
Department of Defense, 2005. [26] Puterman M. Markov decision processes. New York: Wiley; 1994.