Abstract (EN)

Evidence from quasi-random assignment (e.g. natural experiments, IV, and RDD) has been labeled ìthe most credible. We argue such causal evidence is often misleading in finance/economics,omitting a key component of the true empirical causal effect. Random assignment, in eliminating self-selection, simultaneously precludes signaling via treatment choice. However, outside experiments, agents enjoy discretion to signal, thereby causing changes in beliefs and outcomes.Therefore, if the goal is informing discretionary decisions, rather than predicting outcomes afterforced/mistaken actions, randomization is problematic. As shown, signaling amplifies, attenuates, or reverses signs of causal effects. Thus, traditional empirical finance methods, e.g. event studies, are often more credible/useful.