In recent years, the agency paradigm has become central to theoretical research in managerial accounting. Empirical research in the area of executive compensation has tested the consistency of agency model predictions with observed compensation date.1 However, it seems difficult to trace specific instances where results and insights obtained from agency models have affected actual management practice .2 This article describes one such instance in the context of government contracting, showing how agency theory was used to design incentive contracts. The research reported here was initiated by the German Department of Defense (GDOD). The Department commissioned a study to examine the applicability af a class of incentive schemes subsequently referred to as budgeted-based schemes.3 To implement these schemes, the GDOD expressed interest in a constructive procedure that would derive suitable budget-based schemes for specific procurement projects. I describe such a procedure and discuss a number of institutional factors that affected the way the budget-based schemes were applied to two pilot projects in Germany. Traditionally, government contracts under sole-source conditions have been awarded either as fixed-price or cost-plus contracts. The use of fixed-price contracts has been confined to projects with relatively few technological and economic uncertainties. With such uncertainties, fixed-price contracts are unattractive from a risk-sharing perspective. Yet, even with a risk-neutral contractor (because of size and diversification) governments are typically reluctant to sign fixed-price contracts when there are major informational asymmetries. If the government is relatively ignorant about inputs and resources required for the project, it will have difficulties disputing the firm's ex ante cost calculation. As a consequence, the firm earns an informational rent; that is, the firm will extract a higher price than it would have if the government had shared the firm's knowledge and expertise. Cost-plus contracts avoid the problem of overpayment, but, as has been well documented, the government subjects itself to the problem of cost padding. To limit the negative incentives of cost-plus contracts, it has become common practice in the United States to replace standard cost-plus contracts with cost-plus-fixed-fee contracts, so that the firm's profit allowance is fixed rather than being proportional to actual project costs.4 In recent years, there has been an increasing trend in the United States to provide positive incentives for cost control by using cost-plus-incentive-fee contracts.5 At the outset of a project, the parties negotiate a cost target, and the firm's profit increases proportionally with cost underruns relative to the cost target. Conversely, the incentive profit decreases at the same rate with cost overruns. In effect, the firm thus bears a share of actual project costs. A recent survey by the U.S. General Accounting Office (GAO 1987) shows that for the period 1978-84 firms' cost-share parameters typically varied between 15 and 25 percent, but were as high as 50 percent in unusual cases. A major concern voiced repeatedly in connection with cost-plus-incentive-fee contracts is that the government is unable to formulate realistic cost targets for many projects. If the target is set unrealistically low, the firm is likely to suffer a financial penalty. Conversely, an unrealistically high cost target leads to additional "undeserved" profits. For this reason, the GAO states that cost-plus-incentive-fee contracts are confined to procurement projects where "the government has a sound basis to estimate contract costs, but where uncertainties exist that make a fixed-price contract impractical" (GAO 1987, 1). The relatively low cost-share parameters currently used (15 to 25 percent) may reflect the government's desire to mitigate the effects of unrealistic cost targets. The budget-based schemes considered in this study can be viewed as a refinement of cost-plus-incentive-fee contracts. In addition to actual cost, the incentive fee now depends as well on a cost estimate that the firm submits, typically at the start of the project. In effect, the firm selects a budget (target cost), and the incentive profit is proportional to the budget variance. Previous modeling analysis has shown that the budget-based schemes create desirable reporting and performance incentives (see Kirby et al. 1991). The government receives information that is useful for its budget planning process, since the contracting firm is induced to submit an unbiased cost estimate. Specifically, the firm has an incentive to reveal truthfully its own assessment of expected project costs. To some extent, the budget-based schemes therefore avoid the issue faced by cost-plus-incentive-fee contracts described above. Instead of having the government formulate a realistic cost target, this task is now left to the better informed firm. From a cost-control perspective, the budget-based schemes have been shown to be optimal incentive mechanisms. By offering a menu of contracts, the government can tailor performance incentives to the firm's privately observed cost information.6 Specifically, the firm chooses a high target profit in return for a high cost-share parameter, provided its cost information is relatively favorable. A high cost-share parameter will induce the firm to conduct the project in a more efficient way. The resulting cost savings are effectively split since the firm receives a large incentive profit. As a consequence, both sides will be better off.
