We study initiation of takeover auctions by potential buyers and the seller. A bidder's indication of interest reveals that she is optimistic about the target. If bidders' values have a substantial common component, as in takeover battles between financial bidders, this effect disincentivizes bidders from indicating interest, and auctions are seller‐initiated. Conversely, in private‐value auctions, such as battles between strategic bidders, equilibria can feature both seller‐ and bidder‐initiated auctions, with the likelihood of the latter decreasing in commonality of values and the probability of a forced sale by the seller. We also relate initiation to bids and auction outcomes.

Using data on auctions of companies, we estimate valuations (maximum willingness to pay) of strategic and financial bidders from their bids. We find that a typical target is valued higher by strategic bidders. However, 22.4% of targets in our sample are valued higher by financial bidders. These are mature, poorly performing companies. We also find that (i) valuations of different strategic bidders are more dispersed and (ii) valuations of financial bidders are correlated with aggregate economic conditions. Our results suggest that different targets appeal to different types of bidders, rather than that strategic bidders always value targets more because of synergies.

Traditional real options models demonstrate the importance of the “option to wait” due to uncertainty over future shocks to project cash flows. However, there is often another important source of uncertainty: uncertainty over the permanence of past shocks. Adding Bayesian uncertainty over the permanence of past shocks augments the traditional option to wait with an additional “option to learn.” The implied investment behavior differs significantly from that in standard models. For example, investment may occur at a time of stable or decreasing cash flows, respond sluggishly to cash flow shocks, and depend on the timing of project cash flows.

We analyze how a profit‐maximizing proxy advisor designs vote recommendations and research reports. The advisor benefits from producing informative, unbiased reports, but only partially informative recommendations, biased against the a priori likely alternative. Such recommendations induce close votes, increasing controversy and thereby the relevance and value of proxy advice. Our results suggest shifting from an exclusive emphasis on recommendations, highlighting the importance of both reports and recommendations in proxy advisors' information provision. They rationalize the one‐size‐fits‐all approach and help reinterpret empirical patterns of voting behavior, suggesting that proxy advisors' recommendations may not be a suitable benchmark for evaluating shareholders' votes.