Why Does Size Matter For Bidder Announcement Returns?

In the standard regression of bidder announcement returns (ACARs) on bidder size in US data from 1981-2014, the coefficient on bidder size is positive and significant (0.5, t = 3.9) when the target is a public firm, where the average ACAR is negative (−1.4%); but it is negative and significant (−1.2, t = −11.5) when the target is a non-public firm, where average ACAR is positive (1.4%). We show this pattern of flipping signs is general and predictable. For example, using quantile regressions within the public deal subsample, the bidder size coefficient is positive (1.3, t = 11.5) when ACAR is negative (at the 20th percentile, where ACAR = −5.6%), but negative (−0.7, t = −6.4) when ACAR is positive (at the 80th percentile, where ACAR = 2.6%). These pervasive patterns in the data are important for understanding value creation in corporate takeovers: while bidder size is widely regarded to be a central determinant of bidder announcement returns, the patterns are at odds with all leading explanations in the recent M&A literature for why size matters. Because existing explanations assume size is a proxy for some underlying value driver (e.g., overconfidence, agency problems) with stable correlations (e.g., bigger firms have more overconfident managers who make worse deals), they do not predict flipping signs. So: why does size matter for ACARs? We offer a simple alternative model in which size is not a proxy. Instead size scales per-dollar value created (or destroyed) in a given deal. This simple framework is consistent with all the above patterns. It also creates predictions for target size, relative size and a sample of European deals, which we show are both, borne out by the data, and inconsistent with existing proxy explanations