We investigate a transfer pricing problem between two divisions within a decentralized firm. An upstream division produces an intermediate good that is used by another division within the firm and is also sold in an external market, where the firm competes with a rival selling a differentiated substitute product. Assuming that headquarters has imperfect information about the upstream division's private information and that communication is restricted, we identify conditions under which the firm will prefer a market‐based transfer price based on the market price set by the firm's rival rather than on the market price set by the upstream division. The two transfer prices affect the price‐setting incentives of the upstream division and its rival differently, and convey different levels of private‐cost information to the downstream division, which impacts internal trade efficiency. The relative performance of the two transfer pricing regimes depends on the relative size of internal versus external demand for the upstream division's good and on the degree of uncertainty about the upstream division's costs. Overall, our analysis provides new insights about how alternative market‐based transfer prices can coordinate decentralized decision‐making in the absence of a perfectly competitive intermediate market.

We model relative performance evaluation ( RPE ) when a Chief Executive Officer ( CEO ) has the power to opportunistically influence the design of RPE by choosing the weight on an index‐based peer group or by customizing the selection of peers comprising a peer group. A powerful CEO compares the benefits of reducing common risk affecting his compensation with the benefits of receiving a higher bonus by economizing on expected peer‐group performance. As a consequence, the Board of Directors (BoD) is less likely to use RPE . Our analytical model yields hypotheses predicting that powerful CEO s choose to reduce common risk only partially and that BoDs choose to not implement RPE if expected peer performance is sufficiently high. Our model has further empirical implications in (i) providing new interpretations of tests for detecting strong‐form and weak‐form RPE in the presence of powerful CEO s, and (ii) suggesting a new empirical measure of CEO power with a focus on the delegation of RPE decision rights.