Discussion of "Feedback loops, fair value accounting and correlated investments"

My paper discusses Bloomfield, Nelson, and Smith's (BNS) model and experimental study of the price dynamics that arise when a firm's accounting reports are predictable from its stock returns. This phenomenon occurs when the firm takes a position in an asset that generates unrealized gains and losses (UGL's) that are correlated with the firm's own returns. My discussion of BNS focuses on three features that are often used to evaluate research-namely, potential for falsification, internal validity, and external validity. I view and evaluate the BNS paper in light of each of these features. I also briefly comment on how well the paper addresses issues related to fair value accounting.