Wisdom of Crowds: Cross‐Sectional Variation in the Informativeness of Third‐Party‐Generated Product Information on Twitter

• Author: VICKI WEI TANG
• DOI: http://doi.org/10.1111/1475-679X.12183
• Date: 01 June 2018
• Published date: 01 June 2018

DOI: 10.1111/1475-679X.12183

Journal of Accounting Research

Vol. 56 No. 3 June 2018

Printed in U.S.A.

Wisdom of Crowds: Cross-Sectional

Variation in the Informativeness

of Third-Party-Generated Product

Information on Twitter

VICKI WEI TANG

∗

Received 7 September 2016; accepted 7 July 2017

ABSTRACT

This paper examines whether third-party-generated product information on

Twitter, once aggregated at the firm level, is predictive of firm-level sales, and

if so, what factors determine the cross-sectional variation in the predictive

power. First, the predictive power of Twitter comments increases with the ex-

tent to which they fairly represent the broad customer response to products

and brands. The predictive power is greater for firms whose major customers

are consumers rather than businesses. Second, the word-of-mouth effect of

Twitter comments is greater when advertising is limited. Third, a detailed

analysis of the identity of the tweet handles provides the additional insights

that the predictive power of the volume of Twitter comments is dominated

by “the wisdom of crowds,” whereas the predictive power of the valence of

Twitter comments is largely attributable to expert comments. Furthermore,

∗McDonough School of Business, Georgetown University.

Accepted by Philip Berger. I would like to thank an anonymous referee for her/his insight-

ful comments and suggestions. This study benefits from the discussions with seminar partic-

ipants at Georgetown University, the University of Hong Kong, and the U.S. Securities and

Exchange Commission. This study also benefits from comments from William Baber, John

Core, Rohan Williamson, Prem Jain, Lee Pinkowitz, Allan Eberhart, James Sinclair,Jennie Bai,

and Jess Cornaggia. I would also like to thank the Center for Financial Markets and Policy at

McDonough School of Business for its financial support. LikeFolio.com provided the data on

Twitter comments.

989

Copyright C, University of Chicago on behalf of the Accounting Research Center,2017

990 V.W.TANG

Twitter comments not only reflect upcoming sales, but also capture an unex-

pected component of sales growth.

JEL codes: D83; G14; M41; O33

Keywords: wisdom of crowds; social media; product information; word of

mouth; Twitter; fundamental analysis

1. Introduction

One major shortcoming of the current corporate financial reporting

regulatory regime is that it does not require adequate disclosure by

listed firms of nonfinancial information, such as product information

and customer satisfaction, that would help investors and creditors make

informed decisions (Amir and Lev [1996]). This paper examines whether

third-party-generated comments about products and brands on Twitter,

once aggregated at the firm level, provide information that is useful in fore-

casting firm-level fundamentals. This study further explores what factors

determine the cross-sectional variation in the predictive power of product

information on Twitter. Conceptually, the research question extends the

scope of the investigation from the average predictive power in existing

“now-casting” studies to the cross-sectional variation in the predictive power

of online information. This study differentiates itself from most studies that

extract information about sentiment from electronic platforms by focusing

on firm fundamentals rather than stock prices. Furthermore, this study

distinguishes itself from earlier studies on social media by focusing on third-

party-generated information rather than company-initiated information.

This study chooses Twitter as the setting in examining the cross-sectional

variation in the information content primarily because of the level of aggre-

gation of product information. Though information about products and

brands is available at the product level from alternative sources,1the as-

signment of various products and brands to the businesses that own them

imposes a significant empirical challenge. The data provider uses its pro-

prietary information to achieve a reliable mapping between products and

brands and the entities that own them and, therefore, is able to aggregate

Twitter comments about products and brands at the firm level. The aggre-

gation of product information at the firm level provides a significant em-

pirical edge over other settings because the natural unit for fundamental

analysis is the firm. Twitter is also one of the two social media platforms the

Securities and Exchange Commission allows companies to use to commu-

nicate with investors. Accordingly, using Twitter as the setting has the ad-

ditional benefit of juxtaposing third-party-generated product information

with company-initiated disclosure on the same platform.

1Examples of alternative online sources include Google, Amazon, and Yelp.

WISDOM OF CROWDS 991

This study examines the cross-sectional variation in the predictive power

of product information on Twitter with respect to firm-level accounting

fundamentals. Accordingly, the target of Twitter comments is limited to

products and brands, and tweet handles (the holders of tweet comments)

are limited to third parties, excluding the company itself. The selected Twit-

ter comments are then aggregated at the firm level using the data provider’s

proprietary knowledge in mapping from products and brands to the enti-

ties that own them. Two statistics are used to summarize the volume and

valence of Twitter comments about products and brands. The first statis-

tic (PURCHASE) is defined as the total number of tweets that mention an

actual purchase of a product or brand in the past or a forward-looking in-

tent to purchase. PURCHASE maps Twitter comments directly into a recent

past sale or a potential sale in the future. The valence of each tweet is clas-

sified as positive, negative, or neutral. The second statistic (POSITIVE)is

defined as the ratio of the number of tweets that convey a positive assess-

ment of products and brands over the number of tweets that convey a non-

neutral (either positive or negative) assessment of products and brands.

POSITIVE summarizes the collective customer satisfaction or dissatisfaction

with a company’s products and brands.

Product information on Twitter could reflect firm-level sales through a

combination of two effects. First, the two statistics summarize Twitter users’

responses to products or brands and, therefore, provide easily accessible

signals of the broad customer response. This is labeled as the pure signal ef-

fect of Twitter comments. Second, Twitter comments could spur additional

sales through a word-of-mouth effect.

From a pure signal perspective, the ability of Twitter comments to re-

flect firm-level sales depends on whether those tweets are representative of

the broad customer response to the company’s products and brands. As

Twitter is largely a social platform for leisure rather than business activities,

individual consumers are more likely to share their product experiences

on Twitter than are businesses. Accordingly, Twitter comments are more

representative of the broad customer response for companies whose major

customers are consumers. Therefore, the predictive power of Twitter com-

ments is expected to be greater among those companies. Empirically, the

predictive power of PURCHASE with respect to upcoming sales is more pro-

nounced for companies whose major customers are consumers than oth-

erwise. The second summary statistic, POSITIVE, by construction, factors

in only nonneutral (either positive or negative) tweets. To the extent that

only extremely satisfied (dissatisfied) customers initiate positive (negative)

comments, POSITIVE is susceptible to a higher level of extremity bias. Not

surprisingly, the predictive power of POSITIVE with respect to upcoming

sales is rather limited.

From the word-of-mouth perspective, the ability of Twitter comments to

spur additional sales varies with advertising. Advertising targets a wide audi-

ence and seeks to increase sales by increasing brand awareness. The ability

of Twitter comments to spur more sales works through a mechanism similar