Curtis Taylor, Duke University

"Selloffs, Meltdowns, and Bailouts: Can Asset Markets Inform Policy?"

Abstract

It is well-known that movements in asset markets often presage significant events in the real economy. For this reason policy makers often use financial selloffs to forecast economic downturns and to decide whether to intervene. Together, however, these facts give rise to a puzzle. If asset markets anticipate economic events, then they should anticipate the effects of remedial policies. Hence, if effective policy is forthcoming, and then no selloff in asset markets will occur, but if no selloff is observed, and then no policy intervention will be triggered. We study this dilemma in a game theoretic model with two players, an investor and a policy maker (PM).The investor either knows the underlying state and trades to exploit his valuable information or he is an uninformed noise trader.PM observes the investor’s trades and decides whether to undertake a costly intervention. The price of the asset embodies the likelihood that the investor is making an informed trade and it also anticipates the likelihood of intervention. If the cost of intervening is high or the probability that the investor is informed is low, then there is a unique PBE. In this equilibrium PM does not intervene for modest selloffs and intervenes randomly for large ones. The informed investor randomizes his orders so that PM never strictly benefits from the ability to intervene. The presence of PM causes the equilibrium asset price to be less Blackwell informative and lowers the expected payoff of the informed investor. In the remainder of the parameter space, PM intervenes after any selloff and benefits modestly. The informed investor’s expected equilibrium payoff is zero; i.e., his information has no value. We show that PM would benefit generally from committing to intervene less aggressively because this would induce the investor to reveal more information through his trades.

Contact person: Peter Norman Sørensen