“…While this method requires some initial cluster seeds, the large number of observations makes the cluster formation insensitive to both the initial seeds and to the order in which the observations are clustered (Hair, Black, Babin, Anderson and Tatham, 2006). This classification has proven to be robust in the literature and is used in scores of studies across a broad range of disciplines including accounting (e.g., Collins, Collins, Gong and Hribar, 2003;Ke and Petroni, 2004;Ke and Ramalingegowda, 2005;Ramalingegowda and Yu, 2012;Bentley, Omer and Sharp, 2013), finance (e.g., Gong, Louis and Sun, 2008;Cai, Garner and Walking, 2009;Field and Lowry, 2009;Joe, Louis and Robinson, 2009;Yan and Zhang, 2009;Burns, Kedia and Lipson, 2010;Cremers and Pareek, 2015;Yüksel, 2015), management (e.g., Connelly, Tihanyi, Certo and Hitt, 2010;Eccles, Ioannou and Serafeim, 2014), and marketing (e.g., Luo, Zhang, Zhang and Aspara, 2014). firm's equity is also related to market frictions and correlated trading, as stocks from large firms tend to be more liquid and belong to multiple indexes that are owned by ETFs. The average (median) firm in the full sample has capitalization of $3,068 ($293.29) million, while the average firm in the survivor sample is about twice that size, with average (median) capitalization of $6,279 ($865.2) million.…”