Managing Cost Salience and Procrastination in Projects: Compensation and Team Composition

Managing Cost Salience and Procrastination inProjects: Compensation and Team Composition

Yaozhong WuNational University of Singapore, Business School, Singapore 119245, [email protected]

Karthik RamachandranScheller College of Business, Georgia Institute of Technology, Atlanta, Georgia 30308, USA, [email protected]

Vish KrishnanRady School of Management, University of California, San Diego, La Jolla, California 92093, USA, [email protected]

T he rising trend of projects with high-skilled and autonomous contributors increasingly exposes managers to the riskof idiosyncratic individual behaviors. In this article, we examine the effects of an important behavioral factor, an indi-

vidual’s cost salience. Cost salience leads individuals to perceive the cost of immediate effort to be larger than the cost offuture effort. This leads to procrastination in early stages and back-loaded effort over the course of the project. We modelthe problem confronting the manager of a project whose quality is adversely impacted by such distortion of individualeffort over time. Complementary to prior works focused on the planning and scheduling tasks of project management inthe absence of human behavior, we find that managers should reward contributions made in earlier stages of a project.Our analysis also yields interesting insights on the project team performance: teams with diverse levels of cost saliencewill perform better than homogeneous teams. We also address another important facet of team composition, namely, thechoice between stable and fluid teams, and find that the practice of creating fluid teams might have previously unrecog-nized benefits when behavioral aspects of projects are considered. We conclude with insights and organizational implica-tions for project managers.

Key words: project management; procrastination; cost salience; behavioral operations managementHistory: Received: July 2011; Accepted: February 2013 by Stylianos Kavadias, after 2 revisions.

1. Introduction

Project management (PM) is gaining importance inorganizations, as projects become vital vehicles forlaunching new offerings, integrating newly acquiredfirms, and implementing organizational change. Theability to manage projects for quality and on-timedelivery is increasingly recognized as a critical com-petence for firms in the knowledge economy (Peters1999). Even as PM becomes more important, the nat-ure of projects and the skills required to manage themare also evolving. Hierarchical industries of the pastallowed easier administration of project workers,where managers could specify, define, schedule, andcontrol project tasks clearly. Today, project managersface several new challenges that arise due to theincreasing autonomy of contributors. First, much ofthe information about activities and tasks is moreabstract, which makes it difficult for managers tospecify and require completion of certain amounts ofproject work by certain periods, unlike conventionalsettings like construction projects. Second, the

increasing complexity, specialization, and improvizedeffort required for projects increases the powerwielded by talented project professionals. This hasmade it harder for managers to dictate to individualsas easily as they could in a traditional hierarchicalorganization, and they must rely on nuanced manage-rial approaches and incentives. The greater flexibilityand discretion wielded by individual project teammembers makes managers more vulnerable to thebehavioral quirks and limitations of these workers.Project performance is particularly susceptible to

the procrastinating behavior of individual workers. Thisarticle focuses on one particular psychological causeof procrastination in which individuals attach agreater salience to the present moment, amplifying thecosts of immediate effort and increasing their ten-dency to defer work (Akerlof 1991, O’Donoghue andRabin 1999). Such behavior is particularly challengingin settings where the task requires skilled expertisebut is not sufficiently exciting to intrinsically motivateworkers. Examples include information technologyand business process outsourcing (BPO) projects that

1299

Vol. 23, No. 8, August 2014, pp. 1299–1311 DOI 10.1111/poms.12095ISSN 1059-1478|EISSN 1937-5956|14|2308|1299 © 2013 Production and Operations Management Society

lack the novelty of discovery-oriented projects,although they require contributions from highlyskilled workers. While hard deadlines may ensurethat ultimate completion itself is not delayed, procras-tinating workers tend to leave a greater amount ofwork closer to the deadline and cram to complete thetasks. This type of effort distortion can result in pooroutcomes for a project, as seen in prior work (Arielyand Wertenbroch 2002, Cadena et al. 2011) and in ourown studies of industrial projects discussed below.The challenge faced by contemporary project man-

agers is illustrated by a global provider of BPO ser-vices, which we will refer to as Omega. One of themain services provided by Omega’s BPO groupentails business plan and business case documentpreparation support for its clients. Omega’s keyobjective is to maximize client satisfaction by meet-ing promised deadlines and delivering high qualityoutput. When a new project arrives, the managerreviews the amount of content creation involved incompleting the project by the requested deadlineand assigns a small team of employees to work on it.Our interviews with a project manager at this com-pany indicate that a high frequency of errors inOmega’s deliveries has been associated with thetendency of workers to postpone exertion of effortduring the course of the project. The majority of pro-jects at Omega are performed by teams of youngprofessionals, who work around the clock to meetproject deadlines. While a significant portion of theirremuneration is tied to the total amount of workdone by an employee, the limited creativity requiredin document preparation projects results in cost sal-ience and, consequently, procrastination when thedeadline is farther away. Employees meet the dead-line by cramming during later stages of the projectpotentially resulting in mistakes (or insufficientlyvalidated business cases). Therefore, firms likeOmega cannot ignore the consequences of cost sal-ience, as they may incur damages to their reputationand the subsequent costs of poor quality. Not beingable to use command-and-control methods commonin more hierarchical settings, managers must refinemanagement skills and incentives to reduce theadverse impact of effort distortion on quality and cli-ent satisfaction.In this article, we seek to understand the implica-

tions of individual behavior in PM by modeling theinteractions between a project manager and workerswho are subject to different degrees of cost salience.Owing to the complexity of the setting, we mustextract some key features of the problem to obtainactionable managerial insights. To achieve this pur-pose, we consider a project under an externallyimposed deadline (of two periods). An employee’seffort during the project and the quality of output

received by the manager depend on the followingfactors: (a) the payment rate in each period, (b) theassignment of workers to projects, and (c) constitutionof the project team with respect to the behavioraldiversity of the team (as measured by their salience)or the fluidity of the team, which dictates the degreeto which workers are familiar with each other’sbehavioral tendencies.One of our main findings is that project payment

timing must be tuned to the cost salience of individualworkers. While previous research on project pay-ments suggests that more payments should be madein later stages in the absence of behavioral biases (e.g.,Dayanand and Padman 2001), we find that front-load-ing payments would be useful to motivate workers todistribute effort in a smooth manner, thereby mitigat-ing the effects of effort distortion. In other words,front-loading payments promotes urgency in earlierstages of a project, which is required if workers havegreater cost salience. The degree to which a manageruses wages is critically dependent on the combinationof the worker’s cost salience and the quality cost ofeffort distortion in a particular project, which wecharacterize using closed form results.Second, analysis of projects with multiple contribu-

tors leads to several interesting insights regardingteam constitution. Because workers in one projecthave to compete for a greater share of available work,their behavior is affected not only by their own costsalience but also other workers’ cost salience. We findthat a more diverse team (a team with greater separa-tion in the cost salience of workers) tends to outper-form a less diverse team with the same average levelof cost salience. Consequently, a manager woulddo well to assign a more diverse team (in terms oftendency to procrastinate) to more quality-sensitiveprojects in their portfolio.Third, project teams are often formed in a fluidman-

ner by bringing together skilled workers who arefamiliar with each other to varying degrees (Huck-man and Staats 2011). When paired with a co-workerwith whom they are not acquainted, workers wouldallocate their effort over the duration of the projectbased on their belief about the other worker’s cost sal-ience. Our results show that this belief could be avaluable strategic instrument for the project managerin certain scenarios. In short, the manager is able tochoose a fluid over a stable team or vice versa accord-ing to the beliefs of workers. We derive detailed con-ditions under which stable and fluid teams lead tobetter project outcomes.To the best of our knowledge, this is one of the early

efforts that explicitly incorporates behavioral biases inthe study of PM decisions. Our analysis leads not onlyto testable hypotheses but also to managerial insights,which are increasingly actionable due to recent

Wu, Ramachandran, and Krishnan: Managing Cost Salience and Procrastination in Projects1300 Production and Operations Management 23(8), pp. 1299–1311, © 2013 Production and Operations Management Society

technological advances. For example, the rise of free-lance project work has been accompanied by the ser-vices like elance and oDesk that help managersperiodically monitor the contributions of workers(Needleman 2010), providing guidance on how toimplement the actions we propose in this article.

2. Related Literature

This study adds a behavioral perspective to theplanning and management of projects and is closelylinked to both the PM and behavioral operationsmanagement literatures. Within the PM literature,the problem of scheduling projects for on time com-pletion and high quality execution is of huge practi-cal importance and has deservedly received themost theoretical attention (for a comprehensive sur-vey of this literature, see Herroelen 2005). Thisstream of research has largely focused on the tech-nical aspects of planning and scheduling projectswith rational contributors (Schonberger 1981). How-ever, new organizational challenges of PM, espe-cially the need for project managers to deal withthe behavioral biases of project workers, havereceived scant attention (Gino and Pisano 2008,Krishnan and Loch 2005). Such behavioral issueshave recently begun to attract attention in the oper-ations management literature (e.g., Loch and Wu2007), especially modeling and empirical work inthe domain of manufacturing and inventory man-agement (e.g., Bendoly et al. 2006), supply chainoperations (e.g., Croson and Donohue 2006), andcustomer behavior in service operations and reve-nue management (e.g., Shen and Su 2007).In the project scheduling literature, the relation-

ship between individual behavior and project out-comes was discussed by Goldratt (1997), whoproposed the usage of project completion buffers ascumulative project slack to minimize project tardi-ness. Subsequent papers have discussed the detailsof this approach, as seen in the work of Herroelenand Leus (2001), who study how to set buffer sizeand reschedule projects under the critical chainapproach. Even if project workers do not allow workto expand and fill time (Parkinson 1958), they mayprocrastinate and delay the start of their activitiesowing to the presence of shared project slack. Oncethe completion deadline is near, these project teammembers might have to accelerate their effortssignificantly (Brunnermeier et al. 2009), which couldlead them to commit errors and cut corners. The sub-jects of individual incentives and project selection/execution have also begun to receive increasingattention in the New Product Development litera-ture, but current studies do not yet focus on thebehavioral biases of contributors (Kavadias and Som-

mer 2009, Mihm 2010). We contribute to this growingliterature by studying the impact of individualbehavior on project execution.Beyond PM, there have been extensive studies on

procrastination in the psychology literature (see meta-analysis in Steel 2007). Procrastination has been foundto be not only widespread in the general population(Harriott and Ferrari 1996) but also negatively corre-lated with work performance, which has been consis-tently shown in several studies. For example, in anexperimental study, Ariely and Wertenbroch (2002)found that participants performed proofreading-liketasks better (detected more errors) when their effortswere smoothed by intermediate deadlines than thosewho were given only an end deadline. Also, Steel’s(2007) meta-analysis found consistent negative corre-lation between procrastination and performance inthe literature. Neither procrastination as a strongbehavioral regulator nor its negative impact on pro-ject output quality has, however, been explicitly takeninto formal modeling consideration in PM research.While there are various causes for procrastination, weparticularly focus on cost salience and analyze a pro-ject manager’s possible decisions that can influenceprocrastinating behavior and improve project perfor-mance.On the modeling side, we adopt the spirit of inter-

temporal time preference from economics developedat an individual unit of analysis (Akerlof 1991,O’Donoghue and Rabin 1999). We extend this to aproject context with multiple individuals, when delayis not an option, and focus on the quality concerns ofthe manager. Specifically, we consider settings whereproject quality is associated with the pattern of tem-poral effort allocation by project professionals. Whileincentives are one aspect of our analysis, we alsodevelop insights regarding the impact of workerassignment to projects, team composition, and infor-mation provided to project workers in a team.We develop a model of PM with biased workers in

section 3 and characterize the compensation structurefor a single-worker project in section 4. Managementof projects with multiple workers is explored in sec-tion 5. In section 5.2, we consider how the worker’slack of complete information about his peer’s salienceaffects his own behavior and identify conditionsunder which a project manager might use this toadvantage in section 5.3. We conclude with the impli-cations of our results in section 6.

3. Model of a Project with CostSalience

3.1. Cost Salience and ProcrastinationWe present a project setting in which a projectmanager employs team members who have a

Wu, Ramachandran, and Krishnan: Managing Cost Salience and Procrastination in ProjectsProduction and Operations Management 23(8), pp. 1299–1311, © 2013 Production and Operations Management Society 1301

tendency to procrastinate, as they attach greater sal-ience to immediate-term costs when allocating effortsover time. This aspect of cost salience can be simplyillustrated in the following example adapted fromAkerlof (1991). Suppose one must perform a task witha benefit of v at a cost of c and has the freedom tofinish it at an earlier time (“today”) or at a later time.Finishing the task today brings a net payoff of onlyv � hc because of the psychological salience of incur-ring immediate cost captured by h > 1, while the netpayoff of finishing at any later date is v � c. Thisleads to a preference for delaying work wheneverpossible.1

In the context of our project, in each period, theworker incurs a convex (marginally increasing) costin producing an effort level e, c(e) = e2/2. Further-more, the worker also overvalues any effort-relatedcost incurred in the first period relative to the sec-ond period by a cost-salience factor, h, also referredto as the worker’s type. The parameter h capturesthe degree of the worker’s cost salience such thatthe early cost has a higher impact on the worker’sutility when h > 1. The worker’s salience factor hcan also be construed as the degree to which theworker is behaviorally biased by valuing an imme-diate unit of leisure time over delayed unit ofleisure time.The project environment could be characterized by

heterogeneity in cost salience and uncertainty aboutthe worker’s type. Heterogeneity occurs when thereare workers with different levels of h, with lower val-ues of h representing greater diligence (or willingnessto work sooner) on the part of the worker. Uncer-tainty about worker behavior arises in organizationswhere workers are drawn from a shared, large poolof resources for each project; naturally, in settingswhere a manager repeatedly engages with a fewworkers who are dedicated to the manager, uncer-tainty about behavioral type will be a less importantfactor. In the interest of brevity, we present the moregeneral model with manager’s uncertainty aboutworker types in section 4 and a model with workers’mutual uncertainty about each other’s type in section5. The situation without manager’s uncertainty is pre-sented in more detail in Wu et al. (2012). It must benoted that to focus on the effect of cost salience onworker behavior, we normalize the time-consistentdiscounting factor to 1; therefore, effort levels in ouranalysis are purely the outcome of cost saliencerather than temporal discounting of costs andrewards.

3.2. Effort Distortion and Its ConsequencesThe project manager is responsible for the completionof a project with a fixed scope and a deadline. Thescope and deadline are assumed to be fixed to keep

the focus on the quality consequences of worker’s costsalience. In the base case, the manager hires a projectprofessional to complete the project and, without lossof generality, has a total of one unit work that must befinished by the worker in two periods. The managerfaces a high penalty (∞) if the project is not completedwithin two periods. The end of the first period mightrepresent a milestone planned by the project managerto track progress. The end of the second period corre-sponds to the completion date for the project commit-ted to by the manager. More importantly in ourmodel, the project manager also faces a quality costincurred by procrastinating project workers. Thequality of the project depends on the effort level ofworkers in the two periods. When more work is left tobe done in a later period than what is completed in anearlier period, the exertion of effort is distorted overthe duration of the project. Such a distortion lowersthe overall quality of the project. This connectionbetween procrastination and performance has beenobserved in a number of contexts in prior studies(Ariely and Wertenbroch 2002, Cadena et al. 2011,Steel 2007).The consequences of poor performance often

become observable only after the project is completedand the project manager is held to account for reme-dying quality issues that surface later. In our indus-trial project examples, for instance, software and ITservice delivery companies are asked to commit tofix all the quality problems that emerge from theirdelivery for a period of time beyond their deliverydate. In a simple and intuitive manner, we model theeffect of procrastination by the worker(s) as an indi-rect remedial cost paid by the manager that isincreasing in the amount of distortion. Specifically,we adapt a quadratic form, which is common in PMand product development to model the cost of effort(e.g., Bhaskaran and Krishnan 2009). If the amountsof completed work in the two periods are e1 and e2,the manager incurs a cost of quality loss ofj(max{e2 � e1,0})

2, where j ≥ 0 represents the man-ager’s cost of subsequent remedial actions to addressquality issues resulting from effort distortion. Notehere that the quality loss only occurs when the effortdistortion leads to more work to be finished by thedeadline, and more work in the later stage does notimprove the quality caused by the less work in theearly stage. In projects with multiple stages, the workin later stages is not just a repetition of that in earlierstages and therefore may not strengthen the qualitywith more work. For example, consider projects thatinvolve creation of presentations for overseas clients,which is an important component of Omega’s busi-ness. Earlier stages of the such projects involve care-ful understanding of a client’s requirements, whilelater stages involve execution to the template created.


Errors accumulated due to effort distortion in onepart of these projects, such as in requirements defini-tion, cannot be overcome easily with greater effort inother parts. Therefore, while time lost in one stagecan be reclaimed by crashing another stage, qualitylost in one stage cannot entirely be compensated withgains in another stage. In fact, cost of making thechanges after completion/launch are usually sub-stantially higher due to the greater coordination anddisruption entailed. This is also true in general ofmany projects in many industries such as softwaredevelopment, custom manufacturing (across tasks),and many creative services (e.g., advertisement cam-paigns).It is important to note that the firms in different

industries may face very different levels and kinds ofcosts owing to effort distortion. A BPO services firm,such as Omega discussed earlier, faces both internaland external costs if the outcomes are of poor quality:additional internal resources have to be redeployed tofix these quality issues after the project is delivered,and the firm also faces high costs of customer dissatis-faction and future customer acquisition if issues arediscovered after delivery due to effort distortion byits employees.2

3.3. Compensation SchemeAs in our motivating examples, we assume that theknowledge of specific activities is privately known tothe project professional, which offers the worker suf-ficient autonomy of managing the pace of execution.As a result, work division over the course of the pro-ject (i.e., how much work to complete in each period)is not contractible upfront by the manager. Even ifthe manager may have some idea of the job, she maynot want to contractually stipulate upfront the workdivision during the project. Research in psychology,termed self-determination research, shows thatoffering workers sufficient discretion over the plan-ning and execution of tasks serves to boost theirintrinsic motivation and performance (Deci and Ryan2002). The effort invested by a worker, however, isassumed to be observable ex post. For example, themanager in the BPO firm can check how much docu-mentation has been finished by the worker, eventhough she cannot demand the amount of work to befinished by a certain point of time. Such an ex antenon-contractible but ex post observable nature ofwork has been considered by several other research-ers in the economics literature dealing with contracttheory, as reviewed comprehensively by Tirole(1988).3

The manager offers workers incentives to induceimproved performance. A linear incentive (or wagerate) compensation scheme is chosen in our analysisfor its wide applications in businesses (Holmstrom

and Milgrom 1987). Wage rate compensation is com-monly used in managing knowledge-intensive pro-jects. Examples include hourly rate for programmersand for management consultants. While constantwage rate throughout the project is the common prac-tice, we are not restricted to such a scheme that isshown later in the analysis to be not always optimalfor the manager. It is also appropriate to note at thispoint that, while the effort is observable only at theend of a period, we do not consider any unobservableuncertainty in the process. While this sounds restric-tive, this is justifiable in a vast majority of execution-intensive projects that do not involve highly creativeor risky tasks. Furthermore, we make this assumptionto expressly focus on an important behavioral issuethat arises even when it is not compounded withexogenous uncertainties.The rigid deadline for the project requires the man-

ager to offer a wage rate plan over two periods (w1,w2), such that the worker participates and finishes theproject (e1 + e2 = 1). Here, we assume that one uniteffort translates into one unit finished work, andeffort and work are used interchangeably. Upon fini-shing the project, the worker receives total paymentof w1e1 + w2e2 from the manager.4 Since the totalamount of work is fixed, the manager’s objective is toinduce an execution plan (e1,e2) that minimizes thesum of direct (compensation) and indirect costs (qual-ity cost) of completing the project. The manager’s totalcost is given by the following:

CP ¼ jðmaxfe2 � e1; 0gÞ2 þ w1e1 þ w2e2: ð1Þ

In the rest of the article, we investigate how themanager can elicit the appropriate amount of effortfrom workers to minimize the total cost of executingthe project. Although our primary interest is in man-aging procrastination in an interpersonal context, webegin by analyzing a simpler setting with a singleagent. This allows us to develop some structuralinsights regarding the compensation that is requiredto minimize procrastination, which we build on insection 5.

4. Managing a Project with a SingleWorker

In this section, we analyze the optimal compensationscheme set by the manager who employs a singleworker to finish the project. The manager would liketo tailor a wage structure specifically for the hiredagent’s cost salience. Under many circumstances,however, the manager might not be able to tell aworker’s type with certainty. For example, the workeris a new hire and has not worked with the project


manager very often. In this section, we investigatehow optimal wages are affected by the uncertaintyregarding the worker’s type.5

We assume there are two types of workers: the hightype (H-type) with hH and the low type (L-type) withhL such that hH ≥ hL ≥ 1. While the type is known tothe individual worker, the manager only knows thatthe probability of an L-type or H-type worker is p and1 � p, 0 < p < 1. Depending on the salience parame-ter h and wage rates (w1,w2), the worker wouldallocate effort levels e1 and e2 over the course of theproject. Objectively, the total cost incurred by theworker is only ðe21 þ e22Þ=2. However, due to cost sal-ience, the worker perceives the overall utility at thebeginning of the project to be

Ui1 ¼ w1e1 � hie212þ w2e2 � e22

2; i ¼ L;H: ð2Þ

The manager offers a common wage scheme accept-able to both types.6 Her problem is given as follows:

minw1;w2�0

pðw1eL1þw2eL2þjðmaxfeL2� eL1;0gÞ2Þn

þð1�pÞðw1eH1þw2eH2þjðmaxfeH2� eH1;0gÞ2Þo

s.t.Ui1¼w1ei1�hie2i12þw2ei2� e2i2

2�0 i¼L;H ðIR1Þ

Ui2¼w2ei2� e2i22

�0 i¼L;H ðIR2Þfei1;ei2g :¼ argmax~ei1;~ei2Ui1ð~ei1;~ei2Þ ðICÞ

s.t.~ei1þ~ei2 ¼ 1

ð3Þ

The manager’s objective in Problem 3 is to mini-mize the sum of expected costs arising from directexpenses in wages and indirect expenses due to effortdistortion. Adhering to the individual rationality con-straint IR1 ensures that agents of both types have anincentive to accept the contract and start the project,whereas IR2 ensures that the project is completedregardless of the type of the agent. Finally, the incen-tive compatibility constraint IC accounts for the work-er’s self-interested effort allocation once they observethe wages set by the manager.7 The solution to theworker’s utility maximization problem is

e�i1 ¼1þ w1 � w2

1þ hi; e�i2 ¼

hi � w1 þ w2

1þ hi; i ¼ L;H:

ð4ÞIt is easy to observe that the effort in any given per-

iod increases with the wage rate for that period and

decreases with the wage rate for the other period.More importantly, for any given wage rate structure,the first period effort e�i1 decreases with hi, while thesecond period effort e�i2 increases with hi. As a specialcase consider a project where w1 = w2. Cost salienceresults in a difference between efforts in the two peri-ods given by e�i2 � e�i1 ¼ ðhi � 1Þ=ðhi þ 1Þ. This effortdistortion and its quality consequences are of centralimportance when the manager designs a contract forthe worker. In Lemma 1 below, we present the opti-mal wage scheme for the manager depending on thequality cost parameter j. All proofs (including theexpressions of x1 and x2 in the following lemma),unless otherwise mentioned, are provided in theAppendix S1.

LEMMA 1. OPTIMAL WAGE STRUCTURE FOR A SINGLE

WORKER PROJECT. For a project with a single workerwhose type is unknown to the manager, the optimal wagerates are as follows. There exist thresholds j1 and j2 suchthat

w�1 ¼

hH2ð1þ hHÞ ; w

�2 ¼

hH2ð1þ hHÞ ; if 0� j� j1

w�1 ¼ x1; w

�2 ¼

hH � x1

1þ 2hH; if j1\j� j2

w�1 ¼ x2; w

�2 ¼

hH � x2

1þ 2hH; if j[ j2:

Furthermore, the optimal wage rates are such thatw�

1 � w�2 for all j ≥ 0.

Lemma 1 suggests that the manager should offerdifferent wages to induce smooth effort allocationover time when the quality cost rate is high (j > j1), aregime where the manager controls effort distortionthrough a compensation scheme. When the qualitycost rate is low (j ≤ j1), however, the manager doesnot benefit by rewarding the worker to smooth hiseffort allocation over time, and the manager is willingto tolerate the minor costs of effort distortion. Whilethe unknown worker type results in non-linear opti-mal wage rates, the structural property is rather sim-ple, that is, the manager should in general offerhigher wage in the early stage. In most cases, the opti-mal wages still induce effort distortion (i.e., e�i1 \ e�i2).For the L-type worker when the quality cost is highj > j2, however, the optimal wages eliminate effortdistortion, that is, e�L1 [ e�L2. This is due to the highwage rates offered to ensure the participation of theH-type worker.In Proposition 1 below, we consider the impact of

uncertainty on the design of incentives. Let ðwL1 ;w

L2Þ

be the optimal wages for the L-type worker, and


ðwH1 ;w

H2 Þ be the optimal wages for the H-type worker

when the type is perfectly known to the manager.These wages can be directly derived by replacing hwith hL and hH in the optimal wage rates in Lemma1.

PROPOSITION 1. THE EFFECTS OF UNKNOWN WORKER TYPE.

(a) The first (second) period wage rate w�1 ðw�

2Þdecreases (increases) in p, the probability of aworker being L-type. The manager’s total costunder the optimal incentive wages is non-increas-ing in p when j ≤ j2. The utility functions of bothtypes under the optimal wages are non-increasingin p.

(b) Compared with the wage rates to the H-typeworker when the type is known, the optimal wagesfor a worker with unknown type are such that

w�1 \wH

1 , w�2 [ wH

2 . Compared with the wagerates to the L-type worker when the type is known,the optimal wages for a worker with unknown

type are such that w�1 [ wL

1 when j �max 2hLhHþhL�hH�2

8ð1þhHÞ ; j1n o

, and w�2 [ wL

2 when

j � max 3 hH�hLð Þ8hL�4 ; j2

n o.

Proposition 1(a) illustrates the impact of p on wagesand on costs. Intuitively, a higher chance of hiring anL-type worker leads to a lower first period wage rate.The project manager can also achieve a lower costwhen the chance is higher. When the quality cost isvery high (j > j2), however, the manager’s cost maygo up as p increases. The reason is that within thathigh quality cost region, the L-type worker, if hired,finishes more work in the first period, and, as a result,the manager pays too much for the L-type worker.When the hired worker is more likely to be an L-type(higher p), both types’ utilities decrease because themanager offers a lower first period wage, whichmainly determines the overall utility. Furthermore,compared with the case wherein the worker is aknown H-type,the first period wage becomes smallerand the second period wage becomes larger. This sys-tematic change is driven by the fact that the managercan achieve a lower cost in case the worker is an L-typewho is willing to accept a lower first period wage. Atthe same time, the manager needs to increase the sec-ond period wage to ensure the worker participates incase he is an H-type. Part (b) also suggests that theL-type worker benefits from the manager’s imperfect

knowledge. When max 2hLhHþhL�hH�28ð1þhHÞ ; j1

n o� j �

max 3ðhH�hLÞ8hL�4 ; j2

n o, the L-type worker receives higher

wage rates in both periods than when the managerknows his type.

5. Managing Projects with a Team ofWorkers

We now turn to the case in which the manager canenlist multiple workers, which can potentially lowerthe overall cost of the project. Our focus is tounderstand how the number of contributors and thecomposition of a team can influence effort distortionin execution (and thereby reduce the cost associatedwith managing it). We study two important aspectsof team composition: (i) behavioral diversity and (ii)team stability. The manager can control the behav-ioral diversity of the team by selecting the extent towhich workers differ from each other in terms oftheir cost salience. In section 5.1, we derive the opti-mal compensation package for teams as a functionof their diversity. Further, the manager can maketeams more stable by retaining prior project teamsor make them fluid by reconstituting them for aproject (Huckman and Staats 2011). In section 5.2,we consider whether teams should be fluid or sta-ble for projects that are susceptible to behavioralfactors.Projects with multiple contributors are subject to

individual behaviors of two kinds from which sin-gle contributor endeavors are exempt. Individualscould vie to complete a greater share of the projectif they are motivated monetarily or free-ride ontheir partner’s effort if their contributions cannot beobserved at an individual level. In our model, amanager cannot specify the work division betweenthe two periods upfront but can observe the exactamount of contributions made by a worker in eachperiod ex post, which allows them to manage andutilize the heterogeneity among workers. To obtainsome fundamental insights regarding the manage-ment of effort distortion in projects with multipleworkers, we focus on teams comprised of twoworkers.

5.1. Managing Stable TeamsThe manager sets the same wage rates w1,w2 for thetwo workers with salience levels hL and hH(hH ≥ hL ≥ 1), respectively. The two workers simulta-neously and independently set effort levels eit to max-imize their own overall net utilities subject to theavailable amount of work:

Ui1 ¼ w1ei1 � hie2i12þ w2ei2 � e2i2

2; i ¼ L;H: ð5Þ

The cost minimization problem faced by themanager engaging two workers is given below.


The overall problem is a two-stage sequential gamewith a simultaneous game between the workers in thesecond stage, unfolding as follows: (i) first, the man-ager sets common wage rates for the two periods, and(ii) the workers then set effort levels (ei1,ei2), i = L,H,over two periods, individually taking into consider-ation that the project has to be finished in order forthem to receive a payment in the second period.The strategic interaction between the two workers

in our model takes place exclusively through the ICconstraint in Problem 6 above. Yet, this is a crucialdeterminant of worker behavior and, as we shall see,the optimal compensation structure set by the man-ager. Since the total amount of work required tocomplete the project is limited, the workers alsoimplicitly compete with each other for the workavailable in the sense that each worker’s share ofwork is determined not only by his own cost saliencebut also by the other worker’s cost salience. In otherwords, a worker’s salience has a significant impacton the effort choice of the other worker and viceversa. Constraint IC insures that the combination ofei1 and ei2 chosen by worker i maximizes his net util-ity at the beginning of the project. It represents asimultaneous game between the two workers: whenchoosing their effort allocation over time (ei1,ei2), eachworker is aware that his effort choices are boundedby the total amount of work that remains after theother worker’s effort choices are made, that is,ei1 + ei2 = 1 � ej1 � ej2, i 6¼ j.8 The optimal wagesand effort levels are derived by backward inductionand provided in Lemma 2.

LEMMA 2. WAGES AND EFFORTS WITH TWO AGENTS. If

j\ hLþhHþ2hLhH16ð2hLhH�hL�hHÞ, the optimal wage rates are

w�1 ¼ w�

2 ¼ hLhH2ðhLþhHþ2hLhHÞ, and induced first period

efforts are e�L1 ¼ hHhLþhHþ2hLhH

; e�H1 ¼ hLhLþhHþ2hLhH

. If

j � hLþhHþ2hLhH16ð2hLhH�hL�hHÞ, the optimal wage rates are

w�1 ¼ ð4hLhH�hL�hHÞð1þ16jÞ

16ð2hLhHþð1þ8jÞðhLþhHÞÞ ;w�2 ¼ 8hLhHþð1þ16jÞðhLþhHÞ

16ð2hLhHþð1þ8jÞðhLþhHÞÞ,and the induced first period efforts are

e�L1 ¼ hHð3þ16jÞ4ð2hLhHþð1þ8jÞðhLþhHÞÞ ; e

�H1 ¼ hLð3þ16jÞ

4ð2hLhHþð1þ8jÞðhLþhHÞÞ.Furthermore, the wage rates are such that w�

1 � w�2.

Once again, we find that the manager does not offerdifferentiated wages to correct effort distortion unlessthe quality cost parameter is high. When j is less thanthe threshold, offering higher wages is less attractivethan absorbing the quality cost from the distortion ofworkers’ efforts. The manager’s decision to interveneby setting different wages can also be explained interms of the levels of salience parameters. As Figure 1shows, the manager sets w�

1 [ w�2 only if hL and hH

are significant. If they are small (e.g., hL = hH = 1), themanager is relatively unconcerned about the amountof effort distortion caused and sets w�

1 ¼ w�2. It is also

worth noting that for some intermediate values (e.g.,hL = hH = 1.5), the manager would differentiatewages only if j = 0.5 and not if j = 0.25.

Figure 1 Optimal Wages as a Function of Cost Salience

minfw1;w2 � 0g

CP ¼ j max eL2 þ eH2 � eL1 � eH1; 0f gð Þ2 þw1ðeL1 þ eH1Þ þ w2 eL2 þ eH2ð Þ

s.t. Ui1 ¼ w1ei1 � hie2i12þ w2ei2 � e2i2

2� 0 i ¼ L;H ðIR1Þ

Ui2 ¼ w2ei2 � e2i22

� 0 i ¼ L;H ðIR2Þ

ðei1; ei2Þ ¼ arg maxð~ei1;~ei2Þ

Ui1 ¼ w1~ei1 � hi~e2i12þ w2~ei2 �

~e2i22

i ¼ L;H ðICÞ

s.t. ~eL2 þ ~eH2 þ ~eL1 þ ~eH1 ¼ 1

ð6Þ


In the following proposition, we also examine theeffects of salience parameters on optimal wage struc-ture, effort allocation, and each actor’s welfare.

PROPOSITION 2. THE EFFECTS OF SALIENCE ON STABLE

TEAMS.

(a) The optimal wage rates ðw�1;w

�2Þ in both periods

increase in the salience factor of each worker, hL andhH. The first (second) period wage rate w�

1 ðw�2Þ

increases (decreases) in the quality cost parameter, j.(b) The first period effort level of a worker e�i1 decreases

in his own salience hi and increases in the salienceof the other worker hj.

(c) Worker i’s utility function under the optimalwages increases in j and in his own salience when

hi \hjð1þ2hjþ8jð3þ10hjÞþ128j2ð1þ2hjÞÞ

ð1þ8jþ2hjÞðð32j�2Þhj�1�16jÞ and decreases

otherwise; moreover, worker i’s utility functionunder the optimal incentive contract increases inthe other worker’s salience hj.

Proposition 2 provides some interesting insightsregarding the effects of one worker’s salience on theother’s welfare. A worker benefits by being pairedwith someone with a high level of cost salience (partc). This benefit comes from the following two com-pounding effects. First, a worker can get more workbecause the other worker with higher cost saliencetends to work less in the first period (part b). Second,the wage rates are increased by the manager to guar-antee the participation of the worker with higher costsalience. Combining these two positive effects, thehigher the other’s cost salience, the more a workerbenefits from it.

5.1.1. Is Diversity More Efficient? We now con-sider whether a manager should hire two behavior-ally similar workers or hire two workers with verydifferent salience levels. Suppose that the average costsalience level is �h, and the degree of diversity by d suchthat a two-worker team has salience parameters(�h � d, �h þ d). Thus, a larger d implies that the twoworkers are more diverse while preserving the aver-age salience level of the pair at �h. The question thenbecomes whether a manager prefers a more diverseteam of workers or a more homogeneous team.Should the manager choose a two-worker team withdifferent salience parameters (hL,hH) or with the samesalience (h,h) (i.e., two workers have the same type h)?The following proposition provides answers to thisquestion.

PROPOSITION 3. THE EFFECT OF DIVERSITY.

(a) The manager’s overall cost decreases in the diver-sity measure d. Moreover, a larger diversity(higher d) leads to lower effort distortion.

(b) The manager’s overall cost is lower for a diverseteam (hL,hH) than for the homogeneous team (h,h)if h [ 2hLhH

hLþhH; moreover, when h [ 2hLhH

hLþhH, the

diverse team creates smaller effort distortion.

Proposition 3(a) also implies that a diverse teamoutperforms a homogeneous team with averaged sal-ience (i.e., d = 0). When a team’s salience levelsdiverge, the cost saved from the worker with lowersalience (h � d) in terms of completing more work inthe first period outweighs the loss in terms of higherwage rates caused by the worker with higher salience(h + d). Therefore, overall performance improves asthe team becomes more diverse. Part(b) comparesdiversity and homogeneity from a more general per-spective. Note that the threshold above has the prop-

erty that hL � 2hLhHhLþhH

� hLþhH2 . This implies that the two

workers in the homogeneous team are more cost-effi-cient to the manager only when the salience is low,

h\ 2hLhHhLþhH

.

The analysis of diversity has implications for pro-ject-team assignment. When the manager has twoprojects (jH > jL), and two teams, a diverse team(�h � d; �h þ d) and a homogeneous team (�h; �h), sheshould assign the diverse team to the more qualitysensitive project (jH). Thus, behavioral diversity ismore valuable in teams where project outcomes aremore sensitive to effort distortion.

5.2. Managing Fluid TeamsQuite often, individuals with varying backgroundsand exposure are brought together to work on a pro-ject. In such fluid teams, workers might not have per-fect knowledge of each other’s salience (Huckmanet al. 2009). While this is more common in someindustrial settings than others, the wisdom of creatingnew groups, at the cost of breaking existing ones, isdebatable (Edmondson and Nembhard 2009). In thissection, we explore the usefulness of new group crea-tion through the lens of effort distortion. We make anatural and intuitive assumption that in new groups,individuals do not have perfect information about thebehavioral attributes of their cohorts. In our model,this is manifested as worker i’s lack of perfect infor-mation about hj (j 6¼ i). The manager of the project,however, is well aware of each worker’s salienceparameter because of repeated interactions with theemployee over several projects (Baker 2008).There are two types of workers in the workforce:

the high type (H-type) worker with salience factor hHand the low type (L-type) worker with salience factorhL such that hH ≥ hL ≥ 1. The probability that a workeris an L-type or H-type is p and 1 � p, respectively,and this probability is common knowledge to allworkers. Probability p can also be taken as the fraction


of workers that have salience hL. The manager, asbefore, sets wage rates w1 and w2 for the two periods,and these rates are common for the two workers. Aworker of type i knows that the total amount of workthat he can get is either 1 � eL1 � eL2 with probabilityp or 1 � eH1 � eH2 with probability 1 � p, where eLtand eHt (t = 1,2) are the effort levels of the otherworker depending on whether his salience is hL or hH,respectively. Note that workers rely exclusively ontheir own salience in making effort decisions and donot have more than the distributional informationabout their co-worker. Based on his own type, eachworker picks the effort level in the first period to max-imize his net expected utility from the project. The util-ity maximization problem faced by a type i worker,for given w1 and w2, is given below.

maxei1;ei2

E½Ui1� ¼ w1ei1 � hie2i12þ w2ðpeLi2 þ ð1� pÞeHi2Þ

� pðeLi2Þ2 þ ð1� pÞðeHi2Þ22

; ð7Þ

where eLi2 and eHi2 represent the amount of work theworker would do in the second period if his co-worker belongs to the L and H type respectively.Workers simultaneously decide their effort levels,and we determine effort levels chosen by eachworker type in equilibrium. The optimal wages andequilibrium efforts of workers in a fluid team arederived in the following lemma.

LEMMA 3. There exists j such that for j [ j, theoptimal wages for an (L,H) team when both workers areuninformed of each other’s type are given by the following:

The equilibrium efforts of the workers in the first periodare given in the following:

The lemma presents the optimal wages and equilib-rium efforts for a diverse team, but the expressions

can be obtained for a homogeneous team by settinghL = hH.

9 The expressions lead to some interestinginsights regarding the effects of information availabil-ity on effort distortion.

PROPOSITION 4. THE EFFECTS OF SALIENCE ON FLUID

TEAMS.

(a) The first period effort level of a worker decreases inhis own salience and increases in the salience ofthe other worker: e�L1ðe�H1Þ decreases with hL (hH)and increases in hH (hL).

(b) Regardless of their type, the effort level of a workerdecreases with p (the proportion of L-type work-ers):

@e�i1

@p \ 0 for both types i 2 {L,H}.(c) The optimal wage rates increase in p; the degree of

effort distortion also increases in p and therefore themanager incurs a higher cost if p becomes larger.

From Proposition 2 earlier, we know that in stableteams where workers have perfect information aboutthe participants, first period effort decreases with aworker’s own salience and increases in his counter-part’s salience. Does this insight extend to the fluidteams? Interestingly, Proposition 4(a) confirms thateach worker, regardless of his counterpart’s type,performs less work in the first period if his own costsalience is higher, and each worker can take morework in the first period if his co-worker’s salience ishigher. Note that this response to salience holds eventhough workers are not aware that they are membersof a diverse team and the value of probability p doesnot affect this property.Parts (b) and (c) reveal the effect the constitution of

the worker pool, represented by p, on the manager’s

cost. When a worker expects his co-worker to be morelikely an H-type (smaller p), we see that the worker

expects his co-worker to do less work in the first per-iod and increases the amount of work he completes in

w�1 ¼

ð1þ 16jÞð4hLhH þ 2ðphL þ ð1� pÞhHÞ � hH � hL � 1Þ16ð1þ 8jþ 2hLhH þ ð2þ 8jÞðhL þ hHÞ � phL � ð1� pÞhHÞ ;

w�2 ¼

ð1þ 16jÞðhL þ hH þ 1Þ þ 8hLhH þ 4ðphH þ ð1� pÞhLÞ16ð1þ 8jþ 2hLhH þ ð2þ 8jÞðhL þ hHÞ � phL � ð1� pÞhHÞ :

ð8Þ

e�L1 ¼ð3þ 16jÞð1þ 2hHÞ

8ð1þ 8jþ 2hLhH þ ð2þ 8jÞðhL þ hHÞ � phL � ð1� pÞhHÞ ;

e�H1 ¼ð3þ 16jÞð1þ 2hLÞ

8ð1þ 8jþ 2hLhH þ ð2þ 8jÞðhL þ hHÞ � phL � ð1� pÞhHÞ :ð9Þ


the first period. Indeed, this serves to compensate forhis co-worker’s increased salience in the first period.To understand why this occurs, let us suppose workeri does not alter his first period effort plan in responseto an increase in �hj, the expected cost salience of hispartner ð�hj ¼ phL þ ð1 � pÞhHÞ. Because of theincrease in �hj caused by a smaller p, the worker is leftwith more work than he would like ideally to performin the second period. This leads to a reduction in theoverall utility for worker i; to spread the additionalwork created over both periods, worker i advancessome of this to the first period.Counter-intuitively, note that optimal wages and

manager’s total cost increase with p, the proportion ofthe L-type workers (with low cost salience). This isbecause L-type workers are more productive in thefirst period. Therefore, if workers believe they havebeen paired with an L-type co-worker with higherprobability, their enthusiasm to compete for workdeclines in the first period. Thus, the presence ofmany L-type workers forces the manager to offerhigher wages and leads to higher overall costs.

5.3. Stable vs. Fluid TeamsIn this section, we use the results above to determinethe conditions under which fluid or stable team com-positions lead to efficient management of workerbehavior. As we show in Proposition 5, the answerdepends on the distribution of cost salience in thepopulation of workers.

PROPOSITION 5. OPTIMALITY OF STABLE AND FLUID TEAMS.When p\ hL

hLþhH, the manager’s overall cost is lower

under fluid team composition. Otherwise, the manager’soverall cost is lower under stable team composition.

A worker’s belief about his co-worker’s type, p, iscrucial to this question: when p is small, both work-ers expect that they are more likely to be pairedwith an H-type worker and are more motivated towork in the first period (Proposition 4(b)). As aresult, when p is small, the manager can lower costsby merely letting both workers continue in theirprior belief that their co-worker probably has ahigher cost salience. When, however, p is high (i.e.,the chance of being paired with an L-type worker ishigh), the L-type worker has less chance of takingmore work from his co-worker and he works less inthe first period. In this case, it is better to create sta-bility to induce more work from the L-type workerin the first period.More generally, fluid teams are more valuable for

the manager when a large fraction of workers experi-ence greater cost salience. Corollary 1 discussesthe relationship between the value of fluidity V as afunction of p.

COROLLARY 1. VALUE OF TEAM FLUIDITY.

(a) V(p) is decreasing in p, the fraction of hL workersin the population.

(b) Furthermore, V hLhLþhH

� �¼ 0.

6. Discussion and Conclusion

As projects migrate from the physical to the knowl-edge-intensive realm, contributors increasingly workwith greater flexibility and autonomy than before. Inaddition to the traditional focus on timeliness andcosts, managers are also concerned about the behav-ior of the contributors that could affect the quality ofthe outcome. In this article, we have developed amodel of PM with behavioral considerations inwhich workers are hired by the manager to finish atwo-period project. Workers have cost salience andtend to leave more work to the later period, whichleads to a distorted effort allocation over time andpotential quality problems for the project. The man-ager’s objective is to minimize the project cost includ-ing worker compensation and effort distortion costthat is explicitly incorporated as a function of theworker’s efforts.Our analysis shows that the cost salience not only

leads to effort distortion and lower product quality,but also higher compensation for the worker becauseof the manager’s motivation to meet the project dead-line. Furthermore, we identify conditions underwhich the manager should offer higher wages in theearly period to mitigate the adverse effects of cost sal-ience. In other situations, the manager is better off bytolerating effort distortion and offering the samewages in both periods. Extending the analysis toinclude multiple workers yields valuable insightsregarding the constitution and management of projectteams. First, we find that competition for workbetween team members could mitigate the individualcontributors’ tendency to procrastinate. The resultantreduction in effort distortion allows the manager toobtain greater project quality without escalatingincentives. Secondly, we show that a team of individ-uals with diverse preferences can outperform a teamof similar individuals (with the same average percep-tion of cost salience). Finally, we also investigate howthe emerging practice of assembling fluid teams con-tributes to the quality of project outcomes. In a fluidsetting, workers’ uncertainty about their co-worker’sbehavioral attribute is a valuable motivational toolbecause it accentuates the incentive to compete forwork with team members in the earlier period. Theseinsights complement recent empirical findingsregarding the behavior of fluid project teams (Huck-man and Staats 2011).


Our theoretical analysis, organized around thedegree of the cost salience and the sensitivity of qualityrelative to effort distortion, provides practical guide-lines for PM from a new perspective: a project managerfacing workers with hard-wired decision biases cantake different management policies listed as follows:

• Intervention and compensation: Managers shouldcontrol effort distortion based on the cost ofproject quality and worker salience. Accord-ingly, compensation should be structured aseither constant payment or front-loaded pay-ment, which is in contrast to prior findingsthat delaying payments may minimize projectcosts (Dayanand and Padman 2001).

• Team composition: In projects with multipleworkers, diverse teams should be given prefer-ence because of their positive effect on projectquality. In a diverse team, a fluid team structureis valuable when more employees have highcost salience.

This article offers a ground-breaking model of pro-crastination in the context of PM that explicitly incor-porates quality concerns and its behavioral causes.The model, though stylized, allows us to examine thetrade-offs of a particular decision bias in differentproject scenarios. In this regard, this article contribu-tes to a recent and growing area of research on inno-vation which looks at the incentives of the actorsinvolved in a more holistic manner by includingcareer concerns (e.g., Siemsen 2008) or by focusing oninnovation tournaments with multiple contributorsparticipating in contests (Terwiesch and Ulrich 2009).We believe the insights from this article are readily

applicable to a large class of projects in which asmooth and timely execution of tasks is crucial to thefinal quality of the project. The assumptions we makein this article, such as the absence of uncertainty aboutoutcomes, make it less applicable in highly risky envi-ronments like early stage ideation. Our analysis doesnot directly translate to projects that require substan-tial creativity to generate radical changes in existingproducts or services. For those truly innovative pro-jects, procrastination may not necessarily lead to lowerquality of project outcome. Future work should test—both theoretically and empirically—the robustness ofour findings. We hope that this effort stimulates addi-tional research on studying behavioral issues in theareas of project and operations management.

Acknowledgments

The authors thank the departmental editor Stylianos Kava-dias, the senior editor, and two anonymous reviewers fortheir constructive feedback on the article.

Notes

1Cost salience, as described here, is a special form ofdynamic inconsistency discussed in the economic litera-ture and is also similar to the notion of present biasedpreferences, in which an individual seeks immediate grati-fication by placing a stronger relative weight on rewardsreceived in an earlier moment of time (O’Donoghue andRabin 1999).2In our model, we assume that the quality cost is borne bythe manager and not shared by the workers. This can bejustified as follows: (i) the external cost in terms of cus-tomer satisfaction and acquisition cannot be directly speci-fied to the workers and (ii) the quality cost can also beseen as expected loss in the sense that the more effort dis-tortion, the more likely quality problems arise after deliv-ery, which also makes quality cost not contractible to theworkers.3In particular, in our model, the project manager delegatesthe whole project to the worker. However, the projectmanager is not able to contract on a detailed plan to carryout the project across stages. For example, in business casedocument preparation projects, while the project managerengages the worker to finish the whole project, the tasksin a project can vary substantially across projects and arenot standardized as in typical production lines. Therefore,it becomes too time/cost consuming for the project man-ager in each project to specify exactly how much to finishin each stage ex ante because the work required can bevery task specific. When the project is finished and outputis fully observable, however, it will become easy to evalu-ate how much has been done in each stage. The real chal-lenge in managing such projects is that the projectmanager can only contract the whole project to the workerto a degree beyond which the project manager is not abledemand more detailed execution.4The manager’s linear incentive contract may also includea fixed component, which is normalized to be zero. If thefixed component is included in the compensation scheme,the optimal value of it is always negative because thefixed component serves as a rent extraction instrument.The negative fixed component suggests that the projectmanager charges the worker for working on a project,which is not practical. Furthermore, the effort allocation isdetermined only by the wage rates in the two periods. Forsimplicity, we also assume that in the worker participationconstraint the outside option has zero value.5In the interest of brevity, analysis for the case in whichthe worker’s type is known is presented in the companionpaper Wu et al. (2012).6While we focus on this pooling contract, the manager mayalso consider a separating wage scheme that is only accept-able to the worker of hL (i.e., shutting down the H-typeworker). However, in our setting, the manager’s objectiveis to minimize the total cost of completing the project, andnot completing the project is not an option. Therefore,screening is not considered.7Inside the IC constraint, the worker faces a constrainedoptimization problem whereby the worker chooses effortallocations to finish, that is, ~ei1 þ ~ei2 ¼ 1. This restriction


can be interpreted in the following two different ways: (i)since the manager faces a sufficiently severe penalty fortardiness, she only offers the job to the worker who canfinish it. Thus, the manager imposes the restriction, and(ii) for an unfinished project, the worker also incurs a pen-alty proportional to the manager’s. When the penalty costis high enough, any effort allocation such that ~ei1 þ ~ei2 \ 1is not optimal for the worker. Hence, for simplicity, weuse an equality constraint equivalent to the heavy penaltysharing assumption for the analysis.8An alternative model in which two workers sequentiallycompete for work at the start of each period leads to thesame qualitative insights. Thus, the way the workers com-pete for the amount of work does not alter our resultsregarding the effects of cost salience. The analysis of thealternative model is available upon request from theauthors.9Note here j is the threshold beyond which both workers’IR constraints are satisfied (similar to the ones in Lemmas1 and 2). Here, we only present the results for j [ j inorder to keep our focus on the uncertainty between teammembers.

ReferencesAkerlof, G. A. 1991. Procrastination and obedience. Am. Econ. Rev.

81(2): 1–19.

Ariely, D., K. Wertenbroch. 2002. Procrastination, deadlines, andperformance: Self-control by precommitment. Psychol. Sci.13(3): 219–224.

Baker, S. 2008. The Numerati. Houghton Mifflin Harcourt, NewYork.

Bendoly, E., K. Donohue, K. Schulz. 2006. Behavior in operationsmanagement: Assessing recent findings and revisiting oldassumptions. J. Oper. Manag. 24(6): 737–752.

Bhaskaran, S., V. Krishnan. 2009. Effort, revenue, and cost sharingmechanisms for collaborative new product development.Manage. Sci. 55(7): 1152–1169.

Brunnermeier, M. K., F. Papakonstantinou, J. A. Parker. 2009.An economic model of the planning fallacy. NBER Work-ing paper 14228, National Bureau of Economic Research,Inc.

Cadena, X., A. Schoar, A. Cristea, H. M. Delgado-Medrano. 2011.Fighting procrastination in the workplace: An experiment.NBER Working paper 16944.

Croson, R., K. Donohue. 2006. Behavioral causes of the bullwhipeffect and the observed value of inventory information. Man-age. Sci. 52(3): 323–336.

Dayanand, N., R. Padman. 2001. Project contracts and paymentschedules: The client’s problem. Manage. Sci. 47: 1654–1667.

Deci, E., R. Ryan. 2002. Handbook of Self-Determination Research.University of Rochester Press, Rochester, NY.

Edmondson, A. C., I. M. Nembhard. 2009. Product developmentand learning in project teams: The challenges are the benefits.J. Prod. Innov. Manag. 26: 123–138.

Gino, F., G. Pisano. 2008. Towards a theory of behavioral opera-tions. Manuf. Serv. Oper. Manag. 10(4): 676–691.

Goldratt, E. M. 1997. Critical Chain. North River Press, Great Bar-rington, MA.

Harriott, J., J. R. Ferrari. 1996. Prevalence of procrastinationamong samples of adults. Psychol. Rep. 78(2): 611–616.

Herroelen, W. 2005. Project scheduling—Theory and practice.Prod. Oper. Manag. 14(4): 413–432.

Herroelen, W., R. Leus. 2001. On the merits and pitfalls of criticalchain scheduling. J. Oper. Manag. 19: 559–577.

Holmstrom, B., P. Milgrom. 1987. Aggregation and linearity in theprovision of intertemporal incentives. Econometrica 55(2):303–328.

Huckman, R. S., B. R. Staats. 2011. Fluid tasks and fluid teams:The impact of diversity in experience and team familiarity.Manuf. Serv. Oper. Manag. 13(3): 310–328.

Huckman, R. S., B. R. Staats, D. M. Upton. 2009. Team familiarity,role experience, and performance: Evidence from Indian soft-ware services. Manage. Sci. 55(1): 85–100.

Kavadias, S., S. C. Sommer. 2009. The effects of problem structureand team diversity on brainstorming effectiveness. Manage.Sci. 55(12): 1899–1913.

Krishnan, V., C. H. Loch. 2005. A retrospective look at productionand operations management articles on new product develop-ment. Prod. Oper. Manag. 14(4): 433–441.

Loch, C. H., Y. Wu. 2007. Behavioral operations management.Found. Trends Technol. Inform. Oper. Manag. 1(3): 121–232.

Mihm, J. 2010. Incentives in new product development projectsand the role of target costing. Manage. Sci. 56(8): 1324–1344.

Needleman, S. E. 2010. Managing at a distance. Wall Street J. June21, R.5. Available at http://online.wsj.com/article/SB10001424052748703862704575100221456493514.html (accessed dateJuly 14, 2010).

O’Donoghue, T., M. Rabin. 1999. Doing it now or later. Am. Econ.Rev. 89(1): 103–124.

Parkinson, C. N. 1958. Parkinson’s Law: The Pursuit of Progress.John Murray, London.

Peters, T. 1999. The wow project. Fast Company 44.

Schonberger, R. J. 1981. Why projects are “always” late: A ratio-nale based on manual simulation of a PERT/CPM network.Interfaces 11(5): 66–70.

Shen, Z.-J. M., X. Su. 2007. Customer behavior modeling in reve-nue management and auctions: A review and new researchopportunities. Prod. Oper. Manag. 16(6): 713–728.

Siemsen, E. 2008. The hidden perils of career concerns in R&Dorganizations. Manage. Sci. 54(5): 863–877.

Steel, P. 2007. The nature of procrastination: A meta-analytic andtheoretical review of quintessential self-regulatory failure.Psychol. Bull. 133(1): 65–94.

Terwiesch, C., K. T. Ulrich. 2009. Innovation Tournaments: Creatingand Selecting Exceptional Opportunities. Harvard BusinessSchool Press, Boston, MA.

Tirole, J., 1988. The Theory of Industrial Organization. The MITPress, Cambridge, MA.

Wu, Y., K. Ramachandran, V. Krishnan. 2012. Managing projectswith present-biased agents. Available at SSRN.

Supporting InformationAdditional Supporting Information may be found in theonline version of this article:

Appendix S1: Analysis of Sections 4 and 5.


Documents

Managing Cost Salience and Procrastination in Projects: Compensation and Team Composition