“…First, we examine the relation between PIPE issuances and lagged takeover probabilities to avoid confounding inferences through any reverse causality (Khan et al, 2012). Next, we estimate the relation between takeover likelihood and PIPE issuance using the recursive bivariate probit model (Greene, 1998;Byrd et al, 2012;Huang et al, 2014;Yermack, 2014), which assumes that the binary dependent and independent variables are each determined by latent linear models with jointly normal error terms (Evans and Schwab, 1995) that can produce an consistent estimator (Greene, 1998). 17 In this model, the probit equations on the takeover target dummy and the PIPE issuance dummy are estimated simultaneously using the maximum likelihood method, in which the instrumental variable, namely industry takeover (ITO), is used in the estimation model of takeover target.…”