کسب اطلاعات و مدیریت پروژه مطلوب

This paper provides a rationale for why an organization often generates a bias in favor of a new project even after learning that its profitability will be certainly below more conventional ones. We analyze a principal–agent model with two alternative projects, one of which is to be chosen by the principal. In our model, the profitability of a project is determined by the cost of implementation. All parties are familiar with one of the projects (the known project), and thus the implementation cost of this project is common knowledge. Information on the other project (the new project), however, must be acquired by the agent. We find that the new project may be chosen in the optimal contract even when it turns out to be more costly to implement than the known project, if acquiring information is costly enough and the realized implementation cost of the new project is below a particular level. We also discuss distortion in the new project's output schedule when it is selected.

Project management occupies an important place in organizations. The top management within an organization often sets a policy on how projects are to be selected and implemented, and the projects are managed based on their direct and indirect contributions. One major difficulty in project management, as Cyert and March (1961) note, stems from the fact that organizations usually face a choice of alternative projects within a given sector. A new project that operates differently from the conventional ones may add to the cost of implementation, and practitioners point out that a frequent question in the process of project management is, “Why don't we just do it the old way?”1 Organizations sometimes hold onto a new project even after it turns out to be more costly to implement, thus less profitable than other alternatives. As Royer (2003) reports in Harvard Business Review, Esslier, a corrective lens supplier for eyeglasses, introduced a different material for its lenses in 1982 and manufactured the new lenses for nearly a decade even though production cost of the new lenses turned out to be much higher than the normal lenses. Similarly, in 1964 Sony Corporation launched its television sets with Chromatron tubes instead of the traditional shadow mask CRTs. Prior to manufacturing its new product, the development team warned that the assembly of the Chromatron would be much more complex for mass-production compared to the shadow mask. Despite its significantly higher manufacturing cost, the company chose to install Chromatron tubes for its television sets and stubbornly mass-produced them until 1968.2 Although there are a number of psychological explanations for these phenomena,3 few economic studies provide a rationale for why an organization generates a “bias” in favor of a new project even after learning that its profitability will be certainly below more conventional ones. Using a principal–agent model, we attempt to see why and under what circumstances a new project is favored by the principal, even after its profitability turns out to be lower than alternatives, as well as how distortion is generated in the outcome of such project. An innovative project usually involves gathering information, which is often uncontractible. Thus, organizations may necessitate an indirect way to provide incentives for information acquisition. As demonstrated in the studies by Lewis and Sappington, 1993 and Lewis and Sappington, 1997, when the principal cannot tell whether or not the agent is informed, distortion in the output schedule becomes larger compared to the case in which the agent is always privately informed. When the agent's private information is useful to the principal's decision making, the principal may want the agent to be informed even though the agent can command information rent. Moreover, when the agent must expend resources to be informed, the principal must deal with problems associated with both hidden action (information acquisition) and hidden information (truthful report of the acquired information). That is, the agent has an incentive to save the cost of information acquisition, and once information is acquired, he has an incentive to manipulate it to reap information rent. In our model, there are two alternative projects, one of which the principal will select. A project's profitability is determined by the cost of its implementation. Both the principal and the agent have experienced one of the projects (the known project), and hence the implementation cost of this project is public knowledge. The other project, however, employs a new technology (the new project) and neither the principal nor the agent knows its implementation cost. By gathering information, however, the agent can be privately informed of the new project's implementation cost, while the principal remains uninformed and must rely on the agent's report. Thus, the contract offered by the principal is contingent on the agent's report about the new project's implementation cost and specifies the project to be implemented, the selected project's output level, and the transfer to the agent. Our analysis reveals that the new project, even when it turns out to be more costly to implement than the known one, can be selected if the cost of the new project is below the expected level. Since information acquisition is a hidden action in our model, when the agent reports the new project's implementation cost, the principal cannot tell whether the agent in fact acquired the information or the agent is reporting it without being informed. Therefore, direct compensation for the information gathering cost, no matter how much the principal pays, cannot induce the agent to acquire information — the only way to induce information acquisition is to make the agent benefit from it. Since the agent obtains information rent only when the new project is selected (the implementation cost of the known project is public knowledge), the principal generates distortions not only in the output schedule for the new project, but also in selecting the project. Compared to the case where information acquisition is not a problem, the principal decreases the scope of selecting the new project for the range of implementation cost higher than the expected level, but increases the scope of selecting the new project for the cost range lower than the expected level. The reason is as follows. Due to the truth-telling mechanism, if not informed, the agent will report that the new project's implementation cost is at the expected level.4 To discourage the agent from reporting the new project's cost without being informed, the principal reduces the agent's rent when the agent reports that the cost is at the expected level. In doing so, the principal must also reduce the agent's rent for the entire range of the new project's implementation cost that is higher than the expected level. This is because the principal must also prevent the agent from exaggerating the implementation cost once he is informed of it. As a result, the region of selecting the new project is decreased for the range of the cost that is higher than the expected level. However, to make the agent benefit from information acquisition, the overall expected rent to the agent must be increased. Thus, for the range of implementation cost below the expected level, the principal provides more rent to the agent by increasing the region of selecting the new project. Consequently, if a large amount of rent must be provided to induce information acquisition, the principal sacrifices efficiency in project selection as long as the new project's cost is lower than the expected level — the new project is selected in the optimal contract even when it is more costly to implement than the known project. Our result also sheds light on some aspects of organizational behavior regarding commitment to research and development. For instance, in an interview with Economist, Livio Desimone (1995), Chairman and CEO of Minnesota Mining & Manufacturing Co. (3 M), reports that the company has a rule that 30% of annual sales must come from projects less than four years old and 10% of annual sales must come from projects introduced within a year. Our analysis suggests that although such rule may be associated with some inefficiency in selecting more profitable project, it may be determined to some extent by considerations of the R&D incentives. Although a number of studies look at the issue of selecting among alternative projects that have different profitability and incentive problems in information acquisition,5 few studies cope with trade-off between moral hazard and adverse selection. Lambert (1986), for example, shows that the principal may choose a risky project over a safe project to motivate the agent to acquire information on the risky project.6 His result requires risk-aversion of the agent since imposing risk upon the agent is the source of incentive provision in his paper. In our paper, the source of incentive provision is information rent, as the agent can benefit from the information he acquired. Campbell et al. (1993) study a case in which the agent acquires information about returns from two projects. In their model, information acquisition is a hidden action, but acquired information is observed publicly. They show that the sequence of the projects is chosen to reduce the agent's risk in information acquisition. In Levitt and Snyder (1997), the project can be either a good or a bad one, and the agent's effort increases the likelihood of a good project. The true state becomes public knowledge if the project is implemented. The authors show that the principal may choose to implement the project even after receiving an advice from the agent that the project is more likely to be a bad one.7 By doing so, the principal can learn more about whether or not the agent exerted high effort, which in turn mitigates the hidden action problem. Unlike ours, their result relies on the assumption that the true state becomes public knowledge if the project is implemented. None of the studies above associate information acquisition with the agent's rent by being informed. The current paper is also related to the literature on information structure in agency contracting. Lewis and Sappington (1993) analyze an optimal contract when the agent may or may not be informed. The authors show that there is a bunching in the optimal output schedule and the principal shuts down the operation for a range of the cost parameter. Sobel (1993) studies the principal's preference for the amount of information possessed by the agent. His paper only considers a truthful report by the agent because in their models, information acquisition is not the agent's choice variable. Lewis and Sappington (1994) study a situation in which a firm can choose whether to let its buyers be privately informed about their preferences for a product. In their model, there is no moral hazard problem in information acquisition because the buyer does not incur a cost to be informed. Lewis and Sappington (1997) and Cremer et al. (1998a) introduce costly and productive information acquisition to agency contracting with adverse selection.8Khalil et al. (2006) extend these studies to show that if the principal can use multiple agents, she partially separates information gathering and implementation tasks. Gromb and Martimort (2003) compare agency costs in information acquisition under different organizational forms. Iossa and Legros (2004) study a model in which information acquisition by the auditor makes the principal to ignore the report from the agent. In these papers, unlike in ours, the principal does not have an additional choice with no agency cost. Using an incomplete contract framework, Aghion and Tirole (1997) and Dewatripont and Tirole (1999) also study the incentive problems in information acquisition. The contract in the first paper specifies which party has the authority to choose whose information is to be used for the project. They argue that giving authority to the agent may increase his effective control at the principal's expense, but increases the agent's incentive to be informed. The second paper investigates whether and in what way creating advocates provides more incentive to acquire information. The authors show that using multiple agents does not always provide the best incentive for information acquisition.9 Neither of these studies looks at project selection or distortions in the output schedule. Finally, this paper is also related to the studies on the optimal contract that generates ex post inefficiencies for incentive provision ex ante. Rotemberg and Saloner (1994), for example, show that without commitment to implementing a project, the principal may restrict its business horizon to induce the agents to be more innovative, even if that means having to forgo profitable opportunities ex post. Hart and Moore (2004) have similar result that the optimal contract may limit the choices even though it may bring ex post inefficiencies. Unlike our paper, both studies rely on limited commitment for their results. The rest of the paper is organized as follows. In the following section we present the model. The constraints that the principal faces in our model are presented in Section 3. We present our analysis and results in Section 4, which is the main section of the paper. In Section 5, we conclude the paper with several extensions. All proofs are in the Appendix.

In this paper, we have discussed a situation that organizations often face when managing projects — a project must be selected from among alternatives, and the outcome of the selected project must be determined. In our model, the profitability of the new project is not known, and the agent must incur a cost to acquire the information. Because the agent commands information rent only when the new project is selected, to provide incentive to acquire information, the optimal project management generates a bias in favor of the new project. As a result, the new project, even after it turns out to be less profitable than the other reserved project, can be selected for implementation. Our analysis can be extended in several ways. We conclude this paper with discussion of the extensions.