Journal of Business Research 64 (2011) 1361–1368

Contents lists available at ScienceDirect

Journal of Business Research

Value relevance of blog visibility

Nan Hu a,⁎, Ling Liu a, Arindam Tripathy b, Lee J. Yao c,d

a Department of Accounting and Finance, University of Wisconsin Eau Claire, Eau Claire, WI 54702, United Statesb Milgard School of Business, University of Washington Tacoma, Tacoma, WA 98402, United Statesc J.A. Butt College of Business, Loyola University New Orleans, New Orleans, LA 70118, United Statesd School of Business Administration, Southwestern University of Finance and Economics, China

⁎ Corresponding author.E-mail addresses: hun@uwec.edu (N. Hu), liul@uwec

(A. Tripathy), yao@loyno.edu (L.J. Yao).

0148-2963/$ – see front matter. Published by Elsevierdoi:10.1016/j.jbusres.2010.12.025

a b s t r a c t

a r t i c l e i n f o

Article history:Received 21 February 2010Accepted 12 December 2010Available online 5 February 2011

Keywords:Blog visibilityValue relevanceTrading VolumeWord of Mouth communication WOM

This study empirically examines the effect of a non-traditional information source, namely a firm's blogvisibility on the capital market valuation of firms. After controlling for earnings, book value of equity and othervalue relevant variables, such as traditional media exposure, R&D spending, and advertising expense, we finda positive association between a firm's blog visibility and its capital market valuation. In addition, we find blogvisibility Grange causes trading, not vice versa. Our findings indicate that non-traditional information sourcessuch as blogs help disseminate information and influence consumers' investment decisions by capturing theirattention.

.edu (L. Liu), tripathy@uw.edu

Inc.

Published by Elsevier Inc.

1. Introduction

Word of Mouth (WOM) is widely considered to be a major driverof consumer adoption and diffusion of new products and services(e.g., Bass, 1969; Brown and Reingen, 1987; Biyalogorsky et al., 2001;Chatterjee, 2001) as well as consumers' short-term and long-termproduct judgments (Bone, 1995). Prior research in this area, which isprimarily focused on examining the impact of online product reviews(a form ofWOM) on the sales of products, has found online reviews tobe a major information source, especially regarding product quality,for consumers (Gupta and Harris, 2010; Weinberg and Davis, 2005).More recent literature in marketing have studied consumer behaviorvisa vis electronic commerce (e.g., Hernández et al., 2010), use ofbrand netnography in evaluating visitors experience to places (e.g.,Hsu et al., 2009), and the impact of negative WOM on firm-idiosyncratic stock returns (e.g., Luo, 2007). However, there hasbeen little done as far as studying the effect WOM, regardless ofwhether it includes any genuine content or whether it deliverspositive or negative consumer sentiments, on the valuation of firms.Prior research (e.g., Barber and Odean, 2008) has shown that sinceattention is a scarce resource, stocks that grab investors' attention(e.g., by being mentioned in the news), are more likely to beconsidered for trading by investors and to enjoy higher returns. Given

the intrinsic difference between non-traditional information channelssuch as blogs and traditional information channels such as news, thisstudy seeks to examine the effect that blogs have in influencingconsumers' investment decisions. This study posits that, similar tonews, blogs (a form ofWOM) have a positive influence on the value ofa firm and investors' purchase behaviors.

As a form of WOM, blogs represent the fastest-growing medium ofpersonal publishing and a new method of individual expression andopinion on the Internet. According to blogpulse.com, there are around77 million blogs currently in the World Wide Web, with approx-imately 90,000 new blogs created and 700,000 blog posts on a dailybasis. The usefulness and importance of blogs are reflected in theiradoption bymany big companies as a tool to disseminate information,build relationships, co-create product design or customer service(Payne et al., 2009), and solicit feedback from stakeholders andinterested parties (i.e., potential customers and investors). Forexample, GM has adopted FastLane Blog (http://fastlane.gmBlogs.com/) to spread news, provide information, and create an onlinecommunity for customers to discuss information and views. Similarly,Microsoft uses blogs to communicate directly with customers tounderstand their response to its products and services (Wright, 2006).

Given the increasing importance of blogs in changing themarketingand daily operations of a firm, it is timely to examine whether the blogvisibility of a firm, as an information channel and amarketing tool, leadsto improved market valuation. Prior studies have investigated theimpacts of email spamandmessage board activity on short-term capitalmarkets (Wysocki, 1999; Das and Sisk, 2003). These studies show thatemail spam and message board posts are used to manipulate markets,resulting in temporary market reactions followed by price reversal.However, blogs are fundamentally different from message boards or

1362 N. Hu et al. / Journal of Business Research 64 (2011) 1361–1368

news. In fact, prior studies have found that blogs are more reliable andtrustworthy compared to both traditional television advertising andemail marketing (Craigie, 2006; Ho and Dempsey, 2010), and thatinformation released through blogs has better quality and a broaderaudience than traditional media. Although ex-ante blogs are free fromany formal quality control process (i.e., peer review or refereeing),comments fromblog readers and rankingbyblog search engines suchasTechnorati act as quality indicators of blogs. Blogs often discloseproprietary, in-depth, and more timely information than that coveredby traditional media.

In this study, we collect data on the blog visibility of firms andevaluate the relation between their blog visibility and their marketvaluation. We posit that securities enjoying higher visibility in theblog space will also have higher market valuations. Our study makesan important contribution to the existing literature on the importanceof WOM in the context of consumer adoption and diffusion ofnew products and services. This study brings to light the importanceof blogs in the valuation of firms. While prior literature has shownthat individual investors are net buyers of stocks that are discussedin the traditional information channels, such as newspapers (Barberand Odean, 2008), this study extends these findings to a non-traditional information channel-blogs. The positive association be-tween blog visibility of firms and their market valuation indicatesthat firms cannot ignore conversations about their products orservices in the blogspace. In fact, firms need to consistently monitorblogs for feedback on how the market views their products andservices. Firms could use blogs to communicate more efficiently withtheir shareholders and other stakeholders, including customers. Suchcommunication and resulting actions by the firms, especially firmswith more small investors, could translate into positive market valuefor the firms. This paper brings to the forefront the effect andimportance of non-verifiable information in the market valuation offirms, which has important implications for the SEC and otherregulatory organizations alike who are interested in regulating factorswhich act as important sources of information for investors in thecapital market.

In the next section we develop our hypothesis which examines therelation between blog visibility and the value of a firm. In Section 3wediscuss the formulation of the blog visibility measure and in Section 4,our empirical model. Section 5 discusses the empirical results andSection 6 our concluding remarks.

2. Hypotheses development

The impact of visibility on security valuation is founded ontwo psychological regularities, namely overconfidence and attributionbias (Daniel et al., 1998). Overconfidence is well-documented inthe psychology of judgment literature (DeBondt and Thaler, 1995)and is relevant for financial markets because the valuation ofsecurities requires “judgment about open-ended issues and feedbackis noisy and deferred” (Daniel et al., 1998). Overconfidence meansinvestors overestimate their abilities in evaluating securities. Andto make matters worse, investors have attribution bias. Theiroverconfidence does not flag even when their evaluation turns outto be wrong. In fact, their overconfidence grows when securityprices confirm their estimation. In other words, investors creditthemselves for past successes and blame others for past failures(Fischhoff, 1982; Taylor and Brown, 1988; Daniel et al., 1998).

Traditionally stock market participants evaluate a company byreading information released through conventional channels suchas newspapers, periodicals (e.g., BusinessWire and Newswire), 10Kor 10Q reports, and analysts' forecasts. Accordingly, the abovefindings in the prior literature were based on the visibility of a firmwithin traditional sources. Investors consider these informationchannels to be trustworthy and reliable because they believe thatthe information released through them has gone through some

form of quality control such as internal/external audit prior to beingreleased to the public. However, the increasing popularity andtrustworthiness of blogs have made them a valid informationchannel for stock market participants to consider when they makeinvestment decisions.

In the age of information explosion, in general, any publicity isgood publicity because even the most critical publicity will stimulatepeople to talk about the company (Chatterjee, 2001). One goodexample is book reviews, where researchers have documented thatany publicity is good publicity because even negative reviews lead toincreased sales. The main driver for such an impact on sales is thenumber of reviews rather than their content (Sorensen andRasmussen, 2004). Recent studies have also shown that luxury brandscreate superior value proposition by a verity of interactions betweenthe luxury brand owners, their customers and members of theirrespective networks (Tynan et al., 2010).

Today blogging has gained immense popularity and is experienc-ing rapid growth as a communication channel through the WorldWide Web. Blogging represents the fastest-growing medium ofpersonal publishing and one of the most recent forms of individualexpression and opinion on the internet. The goal of a blogger is toincrease the value of his or her blogs. Blog value can be measured byits authenticity, transparency, credibility, individualism, originality,relevance, and integrity (e.g., blogcorevalues.blogspot.com/2005/04/evaluating-blog-credibility.html). Bloggers can build the credibility oftheir blogs by disclosing personal information, discussing morerelevant information, and deciding which blogs or hypertext theirblogs should link to. Fundamentally, a blog is a form of a digital socialnetwork and often a source of social network (although not discussedextensively in the academic literature: e.g., Dann, 2010), comprising asocial structure made of nodes and links. Nodes are the individualswithin the network and links are the relationships (interlink,reciprocating citation, and online discussion) between the individuals(Sack, 2001). Owners/authors of blogs put substantial effort intoevaluating/cleaning/verifying the information they post on theirblogs, which in turn increases the quality of the blogs. Thus, blogscan gain their credibility through their subscribers and their peers(link and dis-link).

Wikipedia is an excellent example of how consumers access anduse information that has not been verified in a traditional sense.Unlike traditional resources such as Encyclopedia Britannica andColumbia Encyclopedia, Wikipedia is an online encyclopedia that iswritten by volunteers, and anyone with a web browser and aninternet connection can update Wikipedia at any time. By 2009,Wikipedia has more than three million articles, and adds morethan 1300 articles on a daily basis (http://en.wikipedia.org/wiki/Wikipedia). The “non-verifiable” information and the non-traditionalpublishing model do not hinder it being used by millions of usersworldwide. The popularity of Wikipedia reveals that sufficientnumber of people find free and “non-verifiable” information to bean acceptable substitute for the verified and edited information intraditional encyclopedias. To some extent, information on Wikipediacan achieve even higher quality compared to traditional mediabecause people can modify the content on Wikipedia if they do notthink the content is accurate.

Even in cases where blogs do not communicate any valuableinformation, but simply provide a reaction to existing corporate news,they might cause asset pricing to deviate from fundamental value bygrabbing investors' attention and influencing their purchasingbehavior. When choosing among thousands of common stocks,investors face a formidable search problem (Barber and Odean,2008) because they have limited cognitive processing power andcannot consider all possible scenarios (bounded rationality). Investorsaddress such an issue by limiting their choice sets (Barber and Odean,2008) to those stocks that have recently caught their attention(Odean, 1999). Most individual investors are unlikely to short-sell

Table 1Descriptive statistics.

Variable N Lowerquartile

Mean Median Upperquartile

Std dev

Market Value 315 0.5610 1.5955 1.1215 2.1074 1.7182Blog Visibility 341 0.0011 0.0358 0.0035 0.0121 0.1727Media Visibility 341 0.0846 1.7040 0.2650 0.5569 14.2871Book Value 319 0.2089 0.4200 0.4207 0.5656 0.3278Earnings 319 0.0256 0.0681 0.0607 0.1023 0.0824R&D Spending 334 0.0000 0.0300 0.0000 0.0284 0.0683Advertising Expense 339 0.0000 0.0135 0.0000 0.0146 0.0351Growth 319 0.0037 0.0190 0.0111 0.0250 0.0235Beta 343 0.8365 1.1079 1.0839 1.2986 0.3917

Notes in Table 1.Market Value is the market value of a firm.Blog Visibility is the average blog visibility of each firm in 2006 based on blogpulse.com.Media Visibility is the traditional firm media visibility based on Factiva in 2006.Book Value is the book value of a firm.Earnings are the earnings of a firm.R&D Spending is the research and development expense of a firm.Advertising Expense is the advertisement expense of a firm.Growth is the ratio of sales in 2006 to sales in 2005.Beta is the firm risk as compared to market risk measured by CAPM.All variables are deflated by the book value of the previous year.

1363N. Hu et al. / Journal of Business Research 64 (2011) 1361–1368

(Barber and Odean, 2008), and so are more likely to purchase stockswhich have generated more notice.

Based on the above discussion, we posit that there is a positiveimpact of blog visibility of a firm on its valuation. Formally stated:

H1. There is a positive association between the blog visibility of a firmand its market valuation.

3. Blog visibility measure

Themain variable of interest for this study is a firm's blog visibility,which is measured based on data from BlogPulse.com. BlogPulseprovides trend analysis capability, which is essential for estimatingthe average blog visibility of each firm. In Appendix 1 we detail astepwise process for collecting data from BlogPulse.com.

The first data collection step is to identify brand-names, products,or services associated with a company. The second is to record blog-space discussions about the company and its products using theconversation tracker tool provided by Blogpulse.com. A firm's blogvisibility is based on blog traffic, which is expressed as the percentageof the daily blogging for a certain company and its products as apercent of total daily blogging activity. To assure that the searchkeywords used in this study for each company are correct, thekeyword identification processes are redone by two additionalindependent researchers. The final sets of keywords comprise onlythose agreed upon by all the researchers. One limitation of this datacollection procedure is that the blogging conversation informationcollected is subject tomeasurement error as it might underestimate oroverestimate the true blog visibility information. Measurement isdependent on the final keyword specified for each company as well asthe text mining algorithm used by blogpulse.com to identify acompany. However, the results of such an estimate are more likelyto produce a downward bias in the regression slope coefficients. Inother words, the potential error associated with the blog visibilitymeasure will only bias against a finding of significance.

4. Empirical model

In this section, we develop an empirical model to evaluate ourresearch hypothesis regarding the value relevance of blog visibility.To examine the association between the blog visibility variable andmarket valuation of a firm, we use a market valuation frameworkproposed by Lev and Sougiannis (1996), which is based on Ohlson's(1995) model. As stated in Eq. (1), the market value of the firm isrelated to a combination of the current book value and earnings ofthe firm and other value-relevant variables (Lev and Sougiannis,1996; Collins, Maydew and Weiss, 1997). A variable is value-relevant if it provides incremental information about expectedfuture earnings beyond that conveyed by the book value and currentearnings. This valuation model incorporates the relevance ofaccounting data in the valuation of firms by following a three-stage process: (1) current earnings are useful for predicting futureearnings, (2) future earnings are an indicator of the future dividend-paying ability of firms, and (3) expected future dividends arediscounted to the present to infer equity value. Thus, this modelexpresses the value of a firm's equity as a function of its earnings,book value, and other value relevant variables such as blog visibility.To control for the size effect, variables are deflated by book value attime t−1.

Model 1: Market Valuei,t /Book Valuei,t−1

= α0 + β1 Book Valuei;t = Book Valuei;t−1 + β2 Earningst = Book Valuei;t−1

+ β � Other value relevant informationi;t = Book Valuei;t−1 + �i;t:

ð1Þ

We include a traditional media visibility variable as a controlvariable since blog visibility could be a proxy for traditional mediavisibility. We also include R&D spending as a control variable as it hasbeen shown to be an important variable in value-relevance studies(Lev and Sougiannis, 1996). Following Krishnan and Sriram (2000)and Yao et al. (2010), Eq. (2) includes several additional controlvariables, such as Growth and Beta. Growth captures the growth rateof a firm in terms of the increase in sales from a previous year, andBeta captures a firm's risk as compared to the market risk. Eq. (2)shows the modified model after including the control variables alongwith the blog visibility variable. Krishnan and Sriram (2000) use avery similar model in their study examining the effects of ITinvestments on firm value. While their model includes Y2K compli-ance cost, in the context of research objectives, the variable of interestin this study is blog visibility. A significant and positive β3 indicatesthat a firm with a high blog visibility is associated with a high marketvaluation.

Model 2: Market Valuei,t /Book Valuei,t−1

= α0 + β1 Book Valuei;t = Book Valuei;t−1 + β2 Earningsi;t = Book Valuei;t−1

+ β3 Blog Visibilityi;t=Book Valuei;t−1 + β4Media Visibilityi;t=Book Valuei;t−1

+ β5 R&D Spendingi;t = Book Valuei;t−1 + β6 Advertising Expensei;t

= Book Valuei;t−1 + β7Growthi;t + β8Betai;t þ �i;t:

ð2Þ

5. Empirical results and discussion

5.1. Data

The goal of this study is to examine the value relevance of blogs.This study focuses on the S&P 500 firms; hence it is less likely that thefinding is driven by firm size difference. We use Blogpulse.com tocapture the average daily blog visibility of each firm in 2006. The dataon a firm's traditional media visibility is collected from Factiva, byaveraging the daily non-redundant news items. Data on book value,earnings, R&D spending, advertising expenses, and sales are collectedfrom Compustat. Stock price details are collected from the CRSP(Center for Research in Security Prices) monthly files.

Table 2Pearson correlation.

Market Value Blog Visibility Media Visibility Book Value Earnings R&D Spending Advertising Expense Growth Beta

Market Value 1.00 0.47b .0001

0.220.0001

0.52b .0001

0.59b .0001

0.49b .0001

0.16180.0055

0.45b .0001

−0.010.9206

Blog Visibility 1.00 0.160.062

0.100.0987

0.160.0043

0.020.6989

0.020.7064

0.170.0035

−0.060.2677

Media Visibility 1.00 −0.040.5412

0.180.0020

−0.020.6717

0.0670.2466

0.210.0004

−0.050.4156

Book Value 1.00 0.36b .0001

0.37b .0001

0.040.4826

0.28b .0001

0.030.5688

Earnings 1.00 −0.060.3050

0.120.0365

0.25b .0001

0.060.3203

R&D Spending 1.00 0.020.6759

0.27b .0001

0.100.0701

Advertising Expense 1.00 0.140.0152

−0.080.1812

Growth 1.00 0.14 0.0137Beta 1.00

See Table 1 notes for variable definition.

1364 N. Hu et al. / Journal of Business Research 64 (2011) 1361–1368

5.2. Descriptive statistics and correlation analysis

Table 1 reports the descriptive statistics of all the variables used inthis study. Table 2 tabulates the Pearson correlations between themain variables. The correlation analysis offers some insights into therelationships between our variables of interest. Both Book Value andEarnings are positive and significantly correlated with the marketvalue of the firm (Pearson correlation=0.52; p-value=0.0001 andPearson correlation=0.59; p-value=0.0001 respectively), this isalong expected lines and is well established in the literature (e.g.,Ohlson, 1995; Collins et al., 1997). Media Visibility, which captures afirm's visibility through traditional media (e.g., theWall Street Journal,and The Financial Times), is also positively and significantly correlatedwith the market value of the firm (Pearson correlation=0.22;p-value=0.0001). Our variable of interest is Blog Visibility, which is

Table 3Blog Visibility and Market Valuation.Model 1: Market Valuei,t /Book Valuei,t−1

=α0+β1 Book Valuei,t /Book Valuei,t−1+β2 Earningsi,t /Book Valuei,t−1+ �i,t.

Model 1a: Market Valuei,t /Book Valuei,t−1.

=α0+β1 Book Valuei,t /Book Valuei,t−1+β2 Earningsi,t /Book Valuei,t−1+β3 Blog Visibilit

Model 2: Market Valuei,t /Book Valuei,t−1.

=α0+β1 Book Valuei,t /Book Valuei,t−1+β2 Earningsi,t /Book Valuei,t−1+β3 Blog VisibilitBook Valuei,t−1+β6 Advertising Expensei,t /Book Valuei,t−1+β7 Growthi,t+β8 Betai,t+

Model 1 Model 1a

Independent variable Coefficient T-value PrN |t| Coefficien

Book Value 0.3594 7.8000 b .0001 0.3368Earnings 0.4603 9.9900 b .0001 0.3926Blog Visibility 0.3287Media VisibilityR&D SpendingAdvertising ExpenseGrowthBetaAdjusted R-squared 0.4565 0.5578N 315 315

Notes on Table 3.See Table 1 for variable definition.All coefficients are standardized coefficients.

positively and significantly correlated with firm market value(Pearson correlation=0.47 and p-value=0.0001). Overall, the resultof the correlation analysis supports the hypothesis that Blog Visibilityis positively associated with the market valuation of firms.

5.3. Blog visibility and market valuation

We take an incremental approach by adding our variable ofinterests one at a time to our basic model (Eq. 1). We remove theinfluential observations with studentized residuals greater thanthree or Cook's D statistics greater than one (Belsley et al., 1980),when estimating the models. White's (1980) test did not reject ourhomoskedasticity assumption, indicating that heteroskedasticity wasnot a problem in our estimation. Belsley et al. (1980) diagnosticswere also applied to check for multicollinearity. All condition indices

yi,t /Book Valuei,t−1+ �i,t.

yi,t /Book Valuei,t−1+β4 Media Visibilityi,t /Book Valueit−1+β5 R&D Spendingi,t/�i,t.

Model 2

t T-value PrN |t| Coefficient T-Value PrN |t|

8.0800 b .0001 0.1403 4.1600 b .00019.2600 b .0001 0.4785 14.5300 b .00018.2200 b .0001 0.2774 8.8900 b .0001

0.0275 0.8800 0.37800.4252 13.0400 b .00010.0509 1.7400 0.08340.1279 3.9200 0.0001

−0.0898 −3.0500 0.00250.7673

302

Table 5Blog Visibility and Market Valuation using the predicted value of Blog Visibility.Market Valuei,t /Book Valuei,t−1

=α0+β1 Book Valuei,t /Book Valuei,t−1+β2 Earningsi,t /Book Valuei,t−1

+β3 Predicted_Blogi,t /Book Valuei,t−1

+β4 Media Visibilityi,t /Book Valuei,t−1+β5 R&D Spendingi,t /Book Valuei,t−1

+β6 Advertising Expensei,t /Book Valuei,t−1

+β7 Growthi,t+β8 Betai,t+εi,t.

Variable Coefficient T-value PrN |t|

Book Value 0.1261 3.32 0.001Earnings 0.5255 14.59 b .0001Predicted_blog 0.2505 3.92 0.0001Media Visibility 0.0882 2.65 0.0086R&D Spending 0.4375 12.15 b .0001Advertising Expense −0.1676 −2.62 0.0092Growth 0.1678 4.67 b .0001Beta −0.1081 −3.32 0.001Adjusted R-squared 0.7169N 302

Notes in Table 5.Predicted_Blog is the fitted value obtained by regressing exogenous variables on theBlog Visibility variable.All coefficients are standardized coefficients.

1365N. Hu et al. / Journal of Business Research 64 (2011) 1361–1368

were less than the suggested cutoff of values. These results,tabulated in table 3, suggest that multicollinearity was not aproblem in our estimation. Model 1 is the basic model in whichonly the Book Value of a firm and its Earnings are included as theindependent variables. The estimated coefficients on Book value andEarnings are consistent with those found in other studies that utilizethis model (e.g., Collins et al., 1997). The coefficient on Book Value is0.3594 (t-statistic=7.800 and p-valueb0.0001) and that on Earn-ings is 0.4603 (t-statistic=9.99 and p-valueb0.0001). In Model 1a,we present the results of estimating Eq. (1) including the blogvisibility variable, which is our main variable of interest. Thecoefficients on the Book Value and the Earnings are still positiveand significant as in the basic valuation model, and the coefficienton the blog visibility variable is also positive and significant(coefficient=0.3287 and t-statistics=8.22 and p-valueb0.0001),supporting the hypothesis that there is a positive associationbetween market valuation of a firm and its blog visibility.

Model 2 includes blog visibility, our variable of interest, and allthe other control variables, such as R&D spending, advertisingexpense, growth rate, Beta, and traditional media visibility of afirm, to control for the potential confounding effect. The estimatedcoefficients on Book Value (estimated coefficient=0.1403;t-statistic=4.16; P-valueb0.0001) and Earnings (estimated coeffi-cient=0.4785; t-statistic=14.53; P-valueb0.0001) are still posi-tive and significant. The coefficients on the blog visibility variableis also positive and significant (estimated coefficient=0.2774;t-statistics=8.89; P-valueb0.0001), with traditional media visibil-ity controlled, again supporting our hypotheses that there is apositive association between market valuation of a firm and its blogvisibility. We find a positive but statistically insignificant coefficientfor the traditional media visibility variable, Media Visibility (estimatedcoefficient=0.0275; t-statistics=0.8800; P-value=0.3780). Consistentwith prior studies, we also find R&D spending to be significant andpositively associated with the market value of a firm (estimatedcoefficient=0.4252; t-statistics=13.04; P-valueb0.0001). The coeffi-cient for the growth variable is significant and positive (estimatedcoefficient=0.1279; t-statistics=3.92; P-valueb0.0001). As expected,the coefficient for the market beta shows a negative sign (estimatedcoefficient=−0.0898; t-statistics=3.05; P-value=0.0025). Overall,the results show that blog visibility is positively associated with thevalue of a firm, after controlling for book value, earnings, traditionalmedia visibility, R&D spending, and other value-relevant variables.This finding suggests that, on average, market participants do take

Table 4Blog Visibility and Market Valuation using an alternative measure for Blog Visibility.Market Valuei,t /Book Valuei,t−1

=α0+β1 Book Valuei,t /Book Valuei,t−1+β2 Earningsi,t /Book Valuei,t−1+β3 Blog Visibilit/Book Valuei,t−1+β6 Advertising Expensei,t /Book Valuei,t−1+β7 Growthi,t+β8 Betai,t+

Independent variable Coefficient

Book Value 0.1455Earnings 0.5298Blog Visibility 0.1181Media Visibility 0.0909R&D Spending 0.4494Advertising Expense 0.0486Growth 0.1428Beta −0.0933Adjusted R-squared 0.7169N 302

Notes in Table 4.Blog Visibility is the average blog visibility of each firm in 2006 based on LexisNexis.All coefficients are standardized coefficients.

note of information released through blogs and consider it to be animportant source of information in valuing firms. Furthermore, thebreadth of information does influence market valuation, even if suchinformation is released through non-traditional information channels,such as blogs (Table 3).

5.4. Robustness check

5.4.1. Alternative measure for blog visibilityTo check the robustness of our blog visibility measure, we re-

collect blog visibility data from LexisNexis and re-estimate ourmodel with this alternative blog visibility measure. The result of there-estimation is presented in Table 4. We find that the coefficient onthe blog visibility variable is still positive and significant (coeffi-cient=0.1181 and p-value=0.0003) after controlling for othervalue-relevant information. The estimated coefficients on the othervariables are also qualitatively similar to our earlier results,

yi,t /Book Valuei,t−1+β4 Media Visibilityi,t /Book Valuei,t−1+β5 R&D Spendingi,t�i,t.

T-value PrN |t|

3.8800 0.000114.6300 b .00013.6600 0.00032.7200 0.0069

12.4300 b .00011.5000 0.13543.9500 0.0001

−2.8500 0.0047

1366 N. Hu et al. / Journal of Business Research 64 (2011) 1361–1368

indicating that our results are robust to alternative measure of blogvisibility.

5.4.2. Predicted value of blog visibility based on historical advertisingexpenses

Firms might selectively spend more on advertisement expense,and as a result, achieve a high visibility in blog spaces. Since firmswithmore advertisement expense might enjoy a higher market valuation,firms with higher blog visibility might enjoy a higher marketvaluation.

We evaluate this potential relationship between advertisingexpenses and the blog visibility of a firm by first creating a newvariable ‘Predicted_Blog’ and re-estimate our main model usingthis variable. This variable is created by regressing the blogvisibility variable on past advertising expenses and other exogenousvariables and then taking the fitted value as the value of the‘Predicted_Blog’ variable. This approach, known as the instrumentalvariable approach, has been used in prior research (Bascle, 2008;Heckman, 2008). Next, we replace the original Blog Visibility variablewith its predicted value Predicted_Blog, and estimate Eq. (3) asfollowing:

Market Valuei;t = Book Valuei;t−1

= α0 + β1Book Valuei;t = Book Valuei;t−1 + β2Earningsi;t = Book Valuei;t−1

+ β3 Predicted�Blogi;t = Book Valuei;t−1 + β4 Media Visibilityi;t = Book Valuei;t−1

+ β5 R&D Spendingi;t=Book Valuei;t−1 + β6 Advertising Expensei;t=Book Valuei;t−1

+ β7Growthi;t + β8Betai;t þ �i;t:

ð3Þ

Table 5 presents the results of relationship between the predicted blogvisibility and market valuation. The results indicate that blog visibility isstill significantly and positively associated with market valuation.

5.4.3. Granger causality between blog visibility and tradingSo farwe have documented that increased trading is associatedwith

increased blog visibility. However, the association between blogvisibility and trading could be subject to a potential issue of causality.For example high blog visibility might lead to high trading, and vice-versa. To address such an issue, we conduct a Granger causality test(Granger, 1969), to evaluate the direction of association between theblog visibility of a firm and its trading volume. We did not conduct theGranger causality test between the market valuation and blog visibilitybecause it is not reasonable to estimate the Ohlson valuation modelusing daily market value.

We proceed with the Granger test as follows: assume x representsa time-series trading volume; while y represents a time-series blogvisibility. Also assume that both x and y have a particularautoregressive lag length p. We use the autoregressive specificationof a bivariate vector autoregression to test the Granger causalitybetween blog visibility and trading volume. To be more specific, firstwe estimate the following unrestricted equation:

Xt = c1+ ∑p

i=1αixt−i + ∑

p

i=1βiyt−i + εt H0 : β1 = β2 = ::::::= βp= 0:

Next, we estimate the following restricted equation also by OLS:

Xt = c2 + ∑p

i=1λiyt−i + ϕt , then with their respective sum of

squared residuals, RSS1 = ∑T

i=1ε̂

2

t and RSS0 = ∑T

i=1ϕ̂t

2

estimated, if the

test statistic S = RSS0−RSS1ð Þ = pRSS1 = T−2p−1ð ÞeFp;T−2p−1 is greater than the specified

critical value, we reject the null hypothesis that Y does not Granger-causeX. An asymptotically equivalent test is given by S = T RSS0−RSS1ð Þ

RSS1 e

X2 pð Þ.Following the similar framework, we test whether X Granger-cause Y.

Our results indicate that past blog visibility indeed Granger causesfuture trading volumes (the p-value of the Granger-causality test is 0for both specifications); while that past trading volumes does notGranger cause future blog visibility. These findings indicate that higherblog visibility cause an increase in trading, supporting our hypothesis.We acknowledge that volume response is different from price response.Pricemeasures the “average” investor's response, while volume reflectsan individual's asymmetric perceptions of a firm's future prospects (Levand Ohlson, 1982). In other words, trading volume reflects individuals'asymmetric evaluations of the firm's future profitability, while pricecaptures, on average, the overall market evaluation or perception ofinformation (Lev and Ohlson, 1982). Nevertheless, our trading volumeresponse shows that it is the blog visibility that Granger causes stockmarket response, not the other way.

6. Conclusion

Prior literature has found WOM to be a major driver of consumeradoption and diffusion of new products and services and an importantsource of information on product quality for consumers. This studyextends prior literature to examine the role ofWOMas an informationsource in the valuation of companies. This study demonstrates that allpublicity is good publicity in terms of increasing a firm's valuation. Onaverage, market participants consider WOM discussions to be animportant information source when valuing firms and, as aninformation channel, blogs play a similar role to that of traditionalnewspapers.

This study shows that blog visibility has a positive impact on stockvaluation. This finding is important for managers in companies,especially those with a large pool of small investors (i.e., smallcompanies), which lack analyst and traditional media coverage. Thesefirms can stand out from their peers by creatively using blogs to getthe attention of potential investors.

Many blogs specialize in the collection, synthesis, and disseminationof online WOM communications relating to company products andservice. With the increasing popularity of blogs, there is a need for adeeper understanding of their impact on businesses. In an initial effort toprotect investors from potentially fraudulent spam, on March 8th, 2007the SEC suspended trading in the securities of 35 companies that hadbeen the subject of recent and repeated spam email campaigns. Thistrading suspension (known as Operation Spamalot) was due to concernsabout the adequacy and accuracy of company information. In light ofthese events and our findings, we encourage SEC to further investigatethe fundamental impact of blogs on the capital markets and regulate therelease of such information, which has important implications in themarket valuation of firms. This is especially the case for small and naïveinvestors who have limited channels to access and limited capabilities toprocess value-relevant information for companies.

This study also implies several interesting opportunities for futureresearch. A potential extension of this study would consider the textualcontent of blogs and evaluate the blog sentiment. A second extensionwould be to consider incorporating the reputation of a blogger and thecontent of a blog in evaluating its influence on the market pricing of thefirms. Studies incorporating the content and sentiments of blog postswould bring to light the specific information that acts as a catalyst inmarket valuation.

Appendix 1. Measurement of blog visibility

The source for the information of blog visibility is from www.blogpulse.com. The blog visibility measure is captured as follows,first this study uses keywords and a trend tool to collect the dailytrend information. For details, please refer the following link. http://www.blogpulse.com/trend?query1=cisco&label1=&query2=&label2=&query3=&label3=&days=180&x=40&y=6.

1367N. Hu et al. / Journal of Business Research 64 (2011) 1361–1368

Next this study uses the ‘view source’ option in Internet Explorer t

o collect the daily blog visibility information as below.

1368 N. Hu et al. / Journal of Business Research 64 (2011) 1361–1368

References

Barber BM, Odean T. All that glitters: the effect of attention and news on the buyingbehavior of individual and institutional investors. Review of Financial Studies2008;21(2):785–818.

Bascle G. Controlling for endogeneity with instrumental variables in strategicmanagement research. Strategic Organization 2008;6:3:285–327.

Bass FM. A new product growth model for consumer durables. Management Science1969;15(5):215–27.

Belsley DA, Kuh E, Welch RE. Regression Diagnostics: Identifying Influential Data andSource of Collinearity. New York: John Wiley; 1980.

Biyalogorsky E, Gerstner E, Libai B. Customer referral management: optimal rewardprograms. Marketing Science 2001;20(1):82–95.

Bone PF. Word-of-mouth effects on short-term and long-term product judgments.Journal of Business Research 1995;32(3):213–23.

Brown JJ, Reingen PH. Social ties and word-of-mouth referral behavior. Journal ofConsumer Research 1987;14(4):350–62.

Chatterjee P. Online reviews: do consumers use them? Advances in Consumer Research2001;28(1):129–33.

Collins D, Maydew E,Weiss I. Changes in the value relevance of earnings and book valuesover the past forty years. Journal of Accounting and Economics 1997;34:39–68.

Craigie B. Online marketing: Brits don't buy the brand blog. Marketing Week 2006:38.Daniel KD, Hirshleifer D, Subrahmanyam A. Investor psychology and security market

under- and overreactions. The Journal of Finance 1998;53(5):1839–85.Dann S. Redefining social marketing with contemporary commercial marketing

definitions. Journal of Business Research 2010;63:147–53.Das SR, Sisk J. Financial Communities. Santa Clara University: Working Paper; 2003.DeBondt WFM, Thaler RH, Voijslav Maksimovic, Ziemba William T. In: Jarrow Robert A,

editor. Financial Decision-making in Markets and Firms: a Behavioral Perspective,9. Finance, Handbooks in Operations Research and Management Science; 1995.p. 385–410. ~North Holland, Amsterdam.

Fischhoff B. For those condemned to study the past: heuristics and biases in hindsight.In: Daniel Kahneman, Paul Slovic, Amos Tversky, editors. Judgment underUncertainty: Heuristics and Biases. Cambridge: Cambridge University Press; 1982.

Granger CWJ. Investigating causal relations by econometric models and cross-spectralmethods. Econometrica 1969;37:424–38.

Gupta P, Harris J. How e-WOM recommendations influence product consideration andquality of choice: a motivation to process information perspective. Journal ofBusiness Research 2010;63:1041–9.

Heckman J. Econometric causality. National Bureau of Economic Research workingpaper #13934; 2008.

Hernández B, Jiménez J, Martín MJ. Customer behavior in electronic commerce: themoderating effect of e-purchasing experience. Journal of Business Research2010;63:964–71.

Ho JYC, Dempsey M. Viral marketing: Motivations to forward online content. Journal ofBusiness Research 2010;63(9–10):1000–6.

Hsu S, Dehuang N, Woodside AG. Storytelling research of consumers' self-reports ofurban tourism experiences in China. Journal of Business Research 2009;63:1223–54.

Krishnan GV, Sriram R. An examination of the effect of IT investments on firm value: thecase of Y2K-compliance costs. Journal of Information Systems 2000;14(2):95-108.

Lev B, Ohlson JA. Market-based empirical research in accounting: a review, interpretation,and extension. Journal of Accounting and Economics 1982;20:249–311.

Lev B, Sougiannis T. The capitalization, amortization, and value-relevance of R&D.Journal of Accounting and Economics 1996;21(1):107–38.

Luo X. Consumer negative voice and firm-idiosyncratic stock returns. Journal ofMarketing 2007;71:75–88.

Odean T. Do investors trade too much? The American Economic Review 1999;89:1279–98.

Ohlson J. Earnings, book values, and dividends in security valuation. ContemporaryAccounting Research 1995;11(2):661–87.

Payne A, Storbacka K, Frow P, Knox S. Co-creating brands: diagnosing and designing therelationship experience. Journal of Business Research 2009;62(3):379–89.

Sack W. Conversation map: an interface for very large-scale conversations. Journal ofManagement Information Systems 2001;17(3):73–92.

Sorensen AT, Rasmussen SJ. Is any publicity good publicity? A note on the impact ofbook reviews. Working paper; 2004.

Taylor SE, Brown JD. Illusion and well-being: a social psychological perspective onmental health. Psychological Bulletin 1988;103:193–210.

Tynan C, McKechnie S, Chhuon C. Co-creating value for luxury brands. Journal ofBusiness Research 2010;63:1156–63.

Weinberg BD, Davis L. Exploring the WOW in online-auction feedback. Journal ofBusiness Research 2005;58:1609–21.

White H. A heteroscedasticity-consistent covariance matrix estimator and a direct testfor heteroscedasticity. Econometrica 1980;48:817–38.

Wright J. Blog Marketing. McGraw-Hill; 2006.Wysocki PD. Cheap talk on the web: the determinants of postings on stock message

boards. University of Michigan: Working Paper; 1999.Yao LJ, Liu C, Chan S. The influence of firm specific context on realizing Information

Technology business value in manufacturing industry. International Journal ofAccounting Information Systems 2010;11(4):353–362.