Costly Information Acquisition, Social Networks, and Asset Prices: Experimental Evidence

• Author: YOHANES E. RIYANTO,EDWARD HALIM,NILANJAN ROY
• Published date: 01 August 2019
• DOI: http://doi.org/10.1111/jofi.12768
• Date: 01 August 2019

THE JOURNAL OF FINANCE •VOL. LXXIV, NO. 4 •AUGUST 2019

Costly Information Acquisition, Social Networks,

and Asset Prices: Experimental Evidence

EDWARD HALIM, YOHANES E. RIYANTO, and NILANJAN ROY∗

ABSTRACT

We design an experiment to study the implications of information networks for incen-

tives to acquire costly information, market liquidity, investors’ earnings, and asset

price characteristics in a financial market. Social communication crowds out infor-

mation production as a result of an agent’s temptation to free ride on the signals

purchased by her neighbors. Although information exchange among traders increases

trading volume, improves liquidity, and enhances the ability of asset prices to reflect

the available information in the market, it fails to improve price informativeness.

Net earnings and social welfare are higher with information sharing due to reduced

acquisition of costly signals.

KNOWLEDGE ABOUT FUNDAMENTALS INFLUENCES SECURITY prices. The acquisition

of such costly information is one of the central topics in economics. A long line

of research initiated by Grossman and Stiglitz (1980) and Verrecchia (1982)

investigates the incentives to acquire costly information and its implications

for financial markets. Using the principle of rational expectations, this

literature shows that investors’ diverse information is reflected in asset prices

and that individuals incorporate the information content of prices into their

trading decisions. The information dissemination and aggregation properties

of market organization have been explored at length in the theoretical and

∗Nilanjan Roy is in the Department of Economics and Finance, College of Business, City Uni-

versity of Hong Kong. Edward Halim and YohanesE. Riyanto are from the Nanyang Technological

University. The authors have no conflicts of interest to disclose. The authors obtained IRB ap-

proval for data collection. We are grateful to Wei Xiong, an anonymous Associate Editor, and two

anonymous referees whose comments substantially improved the paper. We acknowledge finan-

cial support from the Singapore Ministry of Education Academic Research Fund (AcRF) Tier 2

grant (MOE2014-T2-1-105), Nanyang TechnologicalUniversity, and project number 7200511, City

University of Hong Kong. The paper benefited from discussions during presentations at the 2018

Asia-Pacific meeting of the Society for Experimental Finance at Brisbane, AEI5 Joint Conference

and SURE International Workshop at Seoul National University,ESSEC Business School (Singa-

pore campus), 2017 ESA World Meeting at San Diego, Symposium on the Frontier of Research in

Experimental Economics at Dalian, China, and the 2017 Behavioral and Experimental Analysis in

Macro-Finance (BEAM) International Conference at Kyoto, Japan. We are especially thankful to

Mikhail Anufriev, Yan Chen, Syngjoo Choi, Douglas Davis, Sascha F¨

ullbrunn, Nobuyuki Hanaki,

Chad Kendall, Alan Kirman, Yukio Koriyama, Wooyoung Lim, Charles Noussair, Jan Tuinstra,

Marc Willinger,and Songfa Zhong for helpful comments. Lim Chang Yi, Poh Huizhen Isabella, and

Quah Zong You provided excellent research assistance.

DOI: 10.1111/jofi.12768

1975

1976 The Journal of Finance R

experimental literatures. However, whether social communication occurs via

investors’ information networks is an open question.

Although the significance of the embeddedness of economic activity in social

settings has long been recognized in the sociology literature (Granovetter

(1985)), economists have been slow to acknowledge the role played by neighbors

and friends in influencing our beliefs, decisions, and behaviors. However, the

last two decades have seen an abundance of studies that demonstrate that the

effects of social networks on economic activity are large and pervasive, with

social networks playing a role in the transmission of information about jobs,

products, technologies, and political opinions (Jackson (2008,2010)). Several

research papers show that information-sharing with peers via social networks,

word-of-mouth communication among people with whom we interact on a

regular basis, and shared education networks play an important role for in-

vestment decision-making, including stock market participation and portfolio

choices.1It is now widely recognized that for many economic interactions,

social context is not a second-order consideration, but rather is a primary

driver of behaviors and outcomes.2

In this paper, we examine the effect of information exchange among

investors on an individual trader’s decision to invest in information production

and subsequently on market outcomes such as trading volume and asset price

characteristics. Specifically, we ask the following questions. How does social

communication influence the incentive to acquire costly information about

stock fundamentals? How does information-sharing via networks affect the

ability of market prices to reflect investors’ diverse information as well as

the propensity of prices to reveal the underlying state of nature? Moreover,

what are the implications for trading volume and trader profits? To address

these questions, we design an experimental asset market with endogenous

acquisition of costly information. We assume two equally likely states of

nature, Aand B, and a single asset, namely, an Arrow-Debreu security that

provides a payoff only in state A. Prior to trading, individuals can acquire

costly, imperfect signals about the state of nature. Signals are binary and are

independent and identically distributed (i.i.d.), conditional on the state.

While laboratory markets are much simpler in structure than actual asset

markets, they provide an invaluable controlled setting that allows causal

identification of the network structure. In particular, an exogenous network

of interactions can be imposed on a group of subjects, and several treatments

can be implemented to isolate the effect of the network structure on individual

behavior as well as on market outcomes. The novelty of our analysis stems

from the fact that we embed network structures within the framework of an

Arrow-Debreu security market.

1See, for example, Shiller (2000,2017), Kelly and ´

OGr

´

ada (2000), Duflo and Saez (2003), Hong,

Kubik, and Stein (2004,2005), Ivkovi´

c and Weisbenner (2007), Brown et al. (2008), and Cohen,

Frazzini, and Malloy (2008), among others.

2Several excellent surveys are available on networks in finance (Allen and Babus (2009)), social-

network applications for economic problems (Easley and Kleinberg (2010), Jackson (2010)), and

economic networks in the laboratory (Kosfeld (2004), Choi, Kariv, and Gallo (2016)).

Social Communication and Information Acquisition 1977

Distinct from previous studies on information acquisition, we consider the

existence of a network among traders. Before trading takes place, individuals

share the information that they have purchased with those connected to them

in the network. The network structure is assumed to be exogenous. As em-

phasized in Cohen, Frazzini, and Malloy (2008), a convenient aspect of social

networks is that they typically have been formed ex ante, sometimes years in

the past, and their formation is frequently independent of the information to

be transferred. We further assume that information exchange is perfect and

nonstrategic, such that any information acquired by one individual is automat-

ically exchanged to her connection and vice versa. We model a society in which

individuals are embedded in a social network of long-term relationships that

take time to form, express mutual trust, and are not easily undone (Granovet-

ter (1985)).3One can interpret networks as friendships, club memberships, and

social media connections, or more generally,being connected through a network

can be viewed as using common information sources, such as newsletters.4

On the one hand, social communication can reduce the risk of the asset by

enlarging each trader’s information set as well as increasing the informational

efficiency of prices. On the other hand, the expectation of learning from

informed connections and more informative market prices also gives rise to

a temptation to free ride on others’ acquired information. In our experiment,

we find that, on average, the likelihood of acquiring information and the

amount of signals purchased are both decreasing in the number of a trader’s

connections. Compared to the case of no information-sharing, the proportion

of investors not buying any signal rises by around 55% when information

exchange takes place on a complete network.

Despite reducing information disparity among investors, social com-

munication results in more trades and improves market liquidity. With

information-sharing among investors, a larger fraction of the information

available in the market is impounded into asset prices. However, while prices

reflect publicly available information, they fail to reflect all privately held

information, lending support to the semi-strong form of the efficient market

hypothesis (Fama (1970)). In addition, the extent of information aggregation

increases with the density of the information network.

Furthermore, we show that the ability of prices to correctly predict the

underlying state of nature is not improved with information-sharing. This is

because the strong free-riding incentive crowds out information production

to such an extent that the accuracy of the cumulative signals in the market

remains low. Thus, contrary to conventional wisdom, we provide evidence that

enhanced information exchange via social communication does not improve

the quality of prices as forecasting tools.

3In such a society, lying or withholding information is extremely costly. There could be severe

psychological costs associated with lying to a trusted friend.

4Although our study abstracts from imperfect (or noisy) communication of information as well

as from strategic information revelation, we stress that these are important topics that we leave

for future research.