Does geographic location matter to stock return predictability?

• Published date: 01 July 2019
• Date: 01 July 2019
• DOI: http://doi.org/10.1002/for.2556

SPECIAL ISSUE ARTICLE

Does geographic location matter to stock return

predictability?

Sabri Boubaker

1,2

| Imed Chkir

3

| Lamia Chourou

3

| Samir Saadi

3

1

South Champagne Business School,

Troyes, France

2

Institut de Recherche en Gestion (EA

2354), Université Paris Est, France

3

Telfer School of Management, University

of Ottawa, Ottawa, Ontario, Canada

Correspondence

Sabri Boubaker, South Champagne

Business School, Address: 217, Av. Pierre

Brossolette, BP 710, 10002 Troyes Cedex,

France

Email: sabri.boubaker@get‐mail.fr

Funding information

Telfer School of Management Research

Fund

Abstract

Building on recent and growing evidence that geographic location influences infor-

mation diffusion, this paper examines the relation between firm's location and the

predictability of stock returns. We hypothesize that returns on a portfolio com-

posed of firms located in central areas are more likely to follow a random walk than

returns on a portfolio composed of firms located in remoteareas. Using a battery of

variance ratio tests, we find strong and robust support for our prediction. In partic-

ular, we show that the returns on a portfolio composed of the 500 largest urban

firms follow a random walk; however, all variance ratio tests reject the random

walk hypothesis for a portfolio that includes the 500 largestrural firms. Our results

are robust to alternative definitions of firm's location and portfolio formation.

KEYWORDS

geographic location, information asymmetry, market efficiency, predictability of stock returns,

random walk, varianceratio

1|INTRODUCTION

We examine whether and how geographic location influ-

ences the predictability of stock returns. There is now

strong and consistent evidence associating geographic

proximity to (distance from) central locations with infor-

mation (dis)advantages.

1

For instance, Malloy (2005)

reported that US analysts located nearby companies' head-

quarters issued more precise earnings forecasts and had a

stronger impact on stock prices than distant analysts.

Christoffersen and Sarkissian (2009) showed that mutual

funds located in financial centers tended to outperform

other funds in terms of both gross and risk‐adjusted returns.

Anand, Gatchev, Madureira, Pirinsky, and Underwood

(2011) showed that geographic proximity shaped the ability

of capital markets to incorporate information into security

prices (i.e., price discovery). Other studies found that, hav-

ing a much smaller base of close‐by investors, fewer ana-

lysts, and less media coverage, rural firms suffered more

from information asymmetry problems than their urban

counterparts (e.g., Loughran, 2008; John, Knyazeva, &

Knyazeva, 2011; El Ghoul, Guedhami, Ni, Pittman, &

Saadi, 2013; Cai, Tian, & Xia, 2016). This evidence may

seem surprising given the advances in information technol-

ogy. It can be explained, however, by the difficulty and the

higher cost associated with obtaining soft information on

remote firms (e.g., Butler, 2008; Coval & Moskowitz, 2001;

Hau, 2001; Malloy, 2005). Indeed, contrary to hard informa-

tion, soft information is intangible and difficult to quantify,

transmit, and interpret (El Ghoul et al., 2013).

Financial institutions, investmentbankers, analysts and

major media outlets are predominantly located in central

areas and tendto neglect remotely located firms(Loughran,

2007, 2008; Loughran & Schultz, 2005; Malloy, 2005).

Loughran (2007) showed that information spread from

urban public firms to firms located in rural/small cities. In

1

See, among others, Coval and Moskowitz (2001), Feng and Seasholes

(2004), Malloy (2005), Ivkovic and Weisbenner (2005), Loughran and

Schulz (2005), Gaspar and Massa (2007), Loughran (2007, 2008), Butler

(2008), Kang and Kim (2008), Baik, Kang, and Kim (2010), Anand

et al. (2011), John, Knyazeva, and Knyazeva (2011), Arena and Dewally

(2012), El Ghoul, Guedhami, Ni, Pittman, and Saadi (2013), and

Branikas, Hong, and Xu (2017).

Received: 4 January 2018 Revised: 9 August 2018 Accepted: 8 September 2018

DOI: 10.1002/for.2556

Journal of Forecasting. 2019;38:311–326. © 2018 John Wiley & Sons, Ltd.wileyonlinelibrary.com/journal/for 311

particular, Loughran found that stocks of urban firms led

rural/small citystocks even after controlling for size,indus-

try, and analyst coverage. Loughran (2008) argued that,

with few adjacent investors and analysts and less media

coverage, nonurban firms had greater information asym-

metry between outside shareholders and insiders. This

may negatively affect information production about

remotely located firms, lower their monitoring, and lead

to a higher cost of gathering information fortheir investors.

Consistent with the argument that geographic distance

from central locations being associated with higher infor-

mation asymmetry, Arena and Dewally (2012) and El

Ghoul et al. (2013) found thatfirms located in remote areas

faced higher cost of capital than firms located in central

areas. More recently, Cai et al. (2016) reported that, com-

pared to nonurban firms, firms located in urbanareas were

more likely to become takeover targets and more likely to

complete the acquisition transaction. Cai et al. showedthat

their findings were consistent with urban firms facing less

information asymmetry than firms locatedin remote areas.

An important implication from the geographic loca-

tion literature is that the traditional assumption of infor-

mation spreading uniformly within a specific market does

not hold. This opens up new avenues of research on mar-

ket efficiency where such assumption on information dif-

fusion is critical. In fact, a natural yet unexplored

research question that we seek to address in this study

is whether geographic location is useful in predicting

stock returns. We hypothesize that returns on a portfolio

composed of firms located in central areas (urban or

financial centers) are more likely to follow random walk

than returns on a portfolio composed of firms located in

remote areas (rural or nonfinancial centers). To test our

hypothesis we employ a battery of advanced as well as

conventional versions of variance ratio (VR) tests to dif-

ferent sorts of portfolios composed of centrally located

firms and portfolios composed of remotely located firms.

The present study contributes to the literature in sev-

eral ways. First, to the best of our knowledge, this is the

first study to examine the role of geographic location in

stock market efficiency. Adhering to the seminal paper

of Fama (1970), efficient stock market prices should obey

a random walk model and always fully reflect all avail-

able and relevant information. Successive share price

changes are therefore independent and identically distrib-

uted (henceforth i.i.d.) and, as a result, future share prices

are unpredictable and fluctuate only in response to the

random flow of news. There is an extensive literature that

has appeared since the early 1970s testing the efficiency

of financial markets.

2

There are three forms of market

efficiency: (1) the weak form implies that the stock mar-

ket is efficient when current price of a stock reflects all

historical price information about a company; (2) the

semi‐strong form argues that current price of a stock

reflects all historical price information as well as public

information about the company; and (3) the strong form

implies that, in addition to the historical price data (i.e.,

weak‐form information set) and public information

about a company (i.e., semi‐strong‐form information

set), even nonpublic insider knowledge is reflected in

the current stock price. The vast majority of early studies

on market efficiency find support for the weak form of

efficient market hypothesis in most developed markets

while rejecting it in most developing markets. More

recent studies, however, find evidence of stock returns

predictability in developing as well as developed stock

markets (e.g., Kim & Shamsuddin, 2015; Kim,

Shamsuddin, & Lim, 2011; Kosfeld & Robé, 2001; Lim,

Brooks, & Hinich, 2008; Serletis & Shintani, 2003). By

showing that geographic location is useful in predicting

stock returns, the present study introduces a novel ele-

ment to the debate on stock market efficiency. Moreover,

virtually all studies examining stock market efficiency

employ stock market indices. However, stock market

indices are composed of both centrally located and

remotely located firms. Evidence of stock prices predict-

ability could, partially, be due to the use of aggregate

indices which naturally do not take into consideration

firms' geographic location.

Second, we contribute to the emerging literature on

the economic importance of geographic proximity, and

in particular the stream of studies investigating the rela-

tion between firm location and stock returns.

3

Pirinsky

2

For a thorough review of the market efficiency literature readers may

refer to Fama (1970, 1991, 2001) and a more recent survey paper by

Lim and Brooks (2011).

3

Evidence supporting a link between geographic proximity and infor-

mation flow are documented in other contexts as well (Agarwal &

Hauswald, 2010; Bae, Stulz, & Tan, 2008; Boubaker, Derouiche, &

Lasfer, 2015; Butler, 2008; Francis, Hasan, John, & Waisman, 2016;

John et al., 2011; O'Brien & Tan, 2015; Rajan et al., 2010). For instance,

Bae et al. (2008) showed that local analysts had significant advantages

over foreign analysts, which was reflected in higher earnings forecast

accuracy. Agarwal and Hauswald (2010) and Rajan et al. (2010) showed

that the physical distance between borrower and lender was inversely

related to lender's informational advantage stemmed in the ability to

acquire soft information, which, as pointed out by Petersen and Rajan

(2002), was necessary for credit decisions. O'Brien and Tan (2015) found

that geographic proximity affected an analyst's decision to cover IPO

firms in the USA. In particular, they showed that analysts were 80%

more likely to cover local newly listed firms than nonlocal ones. Using

a large sample of newly privatized firms, Boubakri, Guedhami, and

Saffar (2016) argued that a firm's distance from domestic financial cen-

ters was associated with lower foreign investor participation and owner-

ship. Francis et al. (2016) examined the link between geographic

location and CEO pay, and found that pay–performance sensitivity as

well as equity‐based compensation were significantly lower for rural

firms.

312 BOUBAKER ET AL.