Reducing Overreaction to Central Banks’ Disclosures: Theory and Experiment

Abstract

Financial markets are known for overreacting to public information. Central banks can reduce this overreaction either by disclosing information to a fraction of market participants only (partial publicity) or by disclosing information to all participants but with ambiguity (partial transparency). We show that, in theory, both communication strategies are strictly equivalent in the sense that overreaction can be indifferently mitigated by reducing the degree of publicity or by reducing the degree of transparency. We run a laboratory experiment to test whether theoretical predictions hold in a game played by human beings. In line with theory, the experiment does not allow the formulation of a clear preference in favor of either communication strategy. This paper, however, makes a case for partial transparency rather than partial publicity because the latter seems increasingly difficult to implement in the present information age and is associated with discrimination as well as fairness issues.

Limited Higher Order Beliefs and the Welfare Effects of Public Information

“…10 Since there is a continuum of identical agents, the fraction of agents who receive public information equals P almost certainly. Baeriswyl and Cornand (2013) show that limiting the degree of publicity has the same welfare effects as providing information to all agents with idiosyncratic noise. As Cornand and Heinemann (2008) have shown, the expected welfare for a degree of publicity P is given by:…”

“…This distribution of tasks achieves the first best. If the government 1 Examples are Nagel (1995), Kübler and Weizsäcker (2004), and Baeriswyl and Cornand (2013).…”

Experiments on Macroeconomics: Methods and Applications

“…The lower degree of publicity contains overreactions and leads to welfare improvements compared to entirely withholding imprecise signals. Baeriswyl and Cornand (2014) show that a limited degree of transparency (i.e. disclosing public information to all subjects as private signals with an added idiosyncratic noise) can achieve the same expected welfare as a limited degree of publicity.…”

Optimal Public Information Dissemination. Introducing Multiplier Effects into a Generalized Beauty Contest

“…Overall, the detrimental effect of public information is diminished when agents fail to form higher order beliefs. Baeriswyl and Cornand (2014) experimentally test the partial publicity argument introduced by CH. Comparing partial publicity -where only a fraction of all agents become informed -with partial transparency -where the public signal is disclosed to all agents, albeit with some ambiguity -they show that the two approaches are theoretically equivalent in their effectiveness of controlling overreaction.…”