We develop a new measure of underwriter bargaining power and a novel empirical approach, based on underwriters' comparative ability to place bonds. When an issuer has few "outside options" to take her bond to the market, the underwriter enjoys a stronger bargaining power over her. The key feature of our approach is that bargaining power varies for a given underwriter at a given point in time across different issuers, allowing us to separate the effects of bargaining power from those of reputation and certification with a fixed effects strategy. Using our measure, we document that powerful underwriters are able to extract rents at the expense of bond issuers. For issues with the highest underwriter bargaining power, fees and bond offering yields increase by a combined cost of USD 1.5 million, or about 7% of the average costs for the issuer. We rule out alternative mechanisms based on issuer-underwriter "loyalty". Our findings suggest that lack of competition increases underwriter bargaining power, resulting in material costs for corporate bond issuers.JEL codes: G32, G24 Keywords: Bargaining power; Corporate bonds; Underwriting 2 The global bond market is a major source of corporate financing. It has almost tripled in size since 2001, reaching nearly $50 trillion in outstanding value as of 2013 (Tendulkar andHancock (2014)). At the same time, the bond underwriting industry has become increasingly more concentrated, with a handful of very large banks underwriting the bulk of deals (Levinson (2014, p. 78)): in 2013, the ten largest underwriters had a combined market share of about 80%, up from 55% in 2000 and 30% in 1990. 1 Industry practitioners as well as the financial press have raised concerns that this may give disproportionate power over the issuance process to a few large underwriters, enabling them to extract rents to the detriment of corporate bond issuers. 2 In this paper, we address this concern and ask whether, and to what extent, underwriter power has an impact on corporate bond contracts.The main challenge for any study of bargaining power is identification, because the underwriter's power itself is unobservable and often overlaps with her reputation. For instance, at first glance, market share could seem a reasonable proxy for bargaining power. However, market share could also stand for the certification value of the underwriter's reputation as a signal of bond quality (Booth and Smith (1986)); and indeed, the literature has used it as a proxy for both power (Burch, Nanda, and Warther (2005)) and reputation/certification ability (Livingston and Miller (2000)). Similar considerations apply to alternative proxies such as past performance (Nanda and Yun (1997), Dunbar (2000)) or industry specialization (Dunbar (2000)). Disentangling the impact 1 Based on issuance data from Mergent FISD and SDC; the deal size (par value of bonds) is fully assigned to each lead underwriter. 3 of the "bargaining power" and "certification" channels, therefore, requires a measure of underwriter bargaining power th...