Financial Markets Where Traders Neglect the Informational Content of Prices

• Author: MATTHEW RABIN,DIMITRI VAYANOS,ERIK EYSTER
• Date: 01 February 2019
• DOI: http://doi.org/10.1111/jofi.12729
• Published date: 01 February 2019

THE JOURNAL OF FINANCE •VOL. LXXIV, NO. 1 •FEBRUARY 2019

Financial Markets Where Traders Neglect the

Informational Content of Prices

ERIK EYSTER, MATTHEW RABIN, and DIMITRI VAYANOS∗

ABSTRACT

We model a financial market where some traders of a risky asset do not fully appreci-

ate what prices convey about others’ private information. Markets comprising solely

such “cursed” traders generate more trade than those comprising solely rationals. Be-

cause rationals arbitrage away distortions caused by cursed traders, mixed markets

can generate even more trade. Per-trader volume in cursed markets increases with

market size; volume may instead disappear when traders infer others’ information

from prices, even when they dismiss it as noisier than their own. Making private in-

formation public raises rational and “dismissive” volume, but reduces cursed volume

given moderate noninformational trading motives.

EVER SINCE MILGROM AND STOKEY (1982) and Tirole (1982), researchers have

understood that common knowledge of rationality combined with a common

prior precludes purely speculative trade. Of course, people might rationally

trade financial assets for a variety of nonspeculative motives, such as portfolio

rebalancing and liquidity. Yet even in settings where the presence of nonspec-

ulative motives allows for speculative trade, a rational understanding of the

adverse selection problem causes the overall volume of trade to be constrained

by nonspeculative motives. In many people’s estimation, trading volume in fi-

nancial markets greatly exceeds what can be plausibly explained by models

applying rational expectations equilibrium (REE).1

∗Erik Eyster is at the London School of Economics. Matthew Rabin is at Harvard University.

Dimitri Vayanos is at the London School of Economics, CEPR, and NBER. We thank Nick Bar-

beris; Bruno Biais (the Editor); John Campbell; Kent Daniel; Willie Fuchs; Xavier Gabaix; Tristan

Gagnon-Bartsch; Josh Schwartzstein; Andrei Shleifer; Jeremy Stein; Xavier Vives; Liyan Yang;

an anonymous Associate Editor and two referees; seminar participants at Cambridge, Columbia,

Copenhagen, Harvard/MIT, Humboldt, NYU Stern, Oxford, UC Berkeley, UC Davis, and Whar-

ton; and participants at the Barcelona Graduate School in Economics Summer Forum, Miami

Behavioral Finance Conference, NBER Behavioral Finance Conference, Paris School of Economics

Workshop in Bounded Rationality, Toulouse School of Economics Conference in honor of Jean Ti-

role, SAET (Rio), and Stanford Institute for Theoretical Economics for helpful comments. We also

thank Tristan Gagnon-Bartsch, TarsoMori Madeira, and Gianluca Rinaldi for research assistance.

Financial support from the Paul Woolley Centre at the London School of Economics is gratefully

acknowledged. We have read the Journal of Finance disclosure policy and have no conflicts of

interest to disclose.

1For example, in his presidential address to the American Finance Association, French (2008)

notes that, while the capitalized cost of trading exceeds 10% of market capitalization, turnover in

DOI: 10.1111/jofi.12729

371

372 The Journal of Finance R

Researchers have sought to explain excessive trading volume by relaxing the

common-prior assumption. Harrison and Kreps (1978) show that noncommon

priors about an asset’s payoff generate volume in a dynamic model where risk-

neutral traders cannot sell the asset short. Scheinkman and Xiong (2003)use

Harrison and Kreps’s framework to examine traders who are “overconfident.” In

their setting, all traders observe all signals about the payoff, but certain traders

overestimate the information content of certain signals.2In these models with-

out private information, trade derives from traders agreeing to disagree about

the relationship between payoff and public information—the lack of private

information disencumbers traders from the need to invert market prices.3A

second approach incorporates noncommon priors into incomplete-information

models by assuming that privately informed traders agree to disagree

about the precision of traders’ private information. Daniel, Hirshleifer, and

Subrahmanyam (1998) and Odean (1998), for example, show that traders’ over-

confidence about the precision of their private information can increase trad-

ing volume. Similarly, Odean (1998), Banerjee, Kaniel, and Kremer (2009),

Banerjee and Kremer (2010), and Banerjee (2011) show that when traders

downplay the precision of one another’s private signals—which we call

dismissiveness—volume increases.4In this second class of agreeing-to-disagree

models, the presence of private information infuses market prices with infor-

mation content, and traders are assumed to fully invert market prices to un-

cover others’ information. Both types of agreeing-to-disagree models consider

traders who recognize their disagreements in beliefs—and trade because of

these recognized disagreements.

In this paper, we propose a different conceptual approach to explaining spec-

ulative trade: people trade because they neglect disagreements in beliefs. We

capture this idea in a simple and tractable model where some or all traders,

when choosing their demands, do not fully invert prices to uncover others’ in-

formation. This approach builds on extensive evidence, reviewed in Section IV,

that people do not sufficiently heed the information content of others’ behavior,

even in the absence of intrinsic disagreements.

2007 was 215%, which creates a puzzle: “from the perspective of the negative-sum game, it is hard

to understand why equity investors pay to turn their aggregate portfolio over more than two times

in 2007” (p. 1552).

2Hong, Scheinkman, and Xiong (2006) model overconfidence similarly,also allowing for hetero-

geneous priors, in a model where the number of shares of a risky asset increases over time.

3Other models of trade deriving from differences in beliefs include Lintner (1969)andVarian

(1985), in which traders have different subjective priors; De Long et al. (1990), in which symmet-

rically informed traders disagree because some of them (“noise traders”) misperceive next-period

prices for exogenous reasons; Harris and Raviv (1993) and Kandel and Pearson (1995), in which

traders disagree about the informativeness of public signals; and Biais and Bossaerts (1998), in

which traders are uncertain about others’ belief hierarchy. Hong and Stein (2007) summarize this

literature. Eyster and Piccione (2013) model traders with incomplete theories of price formation,

also in a complete-information setting.

4Banerjee, Kaniel, and Kremer (2009), Banerjee and Kremer (2010), and Banerjee (2011)use

the term “differences of opinion” to describe the heterogeneity in beliefs that drives their mod-

els. We instead use the term dismissiveness, to distinguish it from overconfidence or from other

disagreements about signal structures that can create differences of opinion.