1-s2.0-S1062976909001094-main

7/28/2019 1-s2.0-S1062976909001094-main

http://slidepdf.com/reader/full/1-s20-s1062976909001094-main 1/12

The Quarterly Review of Economics and Finance 50 (2010) 202–213

Contents lists available at ScienceDirect

The Quarterly Review of Economics and Finance

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / q r e f

Why do firms cross-list? International evidence from the US market

Abed Al-Nasser Abdallah a, Christos Ioannidis b,∗

a School of Business and Management, American University of Sharjah, P.O. Box: 26666, Sharjah, United Arab Emiratesb Department of Economics, University of Bath, Bath, United Kingdom

a r t i c l e i n f o

Article history:

Received 10 June 2007

Received in revised form 24 August 2009

Accepted 30 September 2009

Available online 1 December 2009

JEL classification:

G30

G15

G14

Keywords:

Cross-listing

Segmentation

Investor protection

CAPM

Event studies

a b s t r a c t

Using a modified international asset-pricing model we find strong evidence that publicly quoted firms

cross-list whenexhibiting strong performance in theirdomesticmarket andwish to takeadvantage of thissituation. After cross-listing, this advantage disappears. Our sample consists of daily data for 1165 firms

from 47 countriesthat have cross-listed on theUS equitymarkets over theperiod 1976–2007. Withinthe

context of this model we provide tests of thevalidityof themain hypotheses of capital marketsegmenta-

tion and investor protection, whichprovide explanations for equity cross-listing and investigate whether

the natureof the market(regulated or unregulated) and the accompanying legal framework (commonor

civil law) canaccount for theimpact of cross-listingon returns. Supporting thesegmentation hypothesis,

we report a decrease in local market risk after cross-listing. However, we find that themagnitude of such

a decrease is diminishing over time as international markets becomemore integrated. On theother hand,

we do not find any change in the global market risk after cross-listing, except for firms that cross-listed

between 2001 and 2007, where their exposure to international market risk decreases. Furthermore, we

find no evidence to support the investor protection hypothesis.

© 2009 The Board of Trustees of the University of Illinois. Published by Elsevier B.V. All rights reserved.

1. Introduction

The cross-listing phenomenon has attracted a considerable

amount of research into the area investigating the underlying

motives and related benefits. Early research in this area postu-

lated the reduction in capital market segmentation as a strong

motive for cross-listing (e.g. ownership restriction). Market seg-

mentation raises the firm’s cost of capital, whereas cross-listing

reduces segmentation, and hence, the risk associated with interna-

tional investment barriers and exposes the firm to global market.

This reduces the cost of capital, which in turn allows firms to raise

external capital and improve the liquidity of its shares (Errunza &

Losq, 1985; Stapleton & Subrahmanyam, 1977).

Another explanation that can account for cross-listing stems

from the structure of the firm’s ownership. The private benefits

of control that large shareholders enjoy, normally at the expense

of minority shareholders, are seen as the most challenging obsta-

cle facing corporations in their effort to raise capital at a lower

cost. Hence small investors may require a higher expected return

which in turn affects the ability of firms to grow. Stulz (1999) and

Coffee (1999, 2002) provided a new and more advanced expla-

∗ Corresponding author. Tel.: +44 1225 383225; fax: +44 1225 386474.

E-mail address: [email protected](C. Ioannidis).

nation for cross-listing when they argued that firms are able to

reduce their cost of capital by signalling their commitment to pro-

tect the interests of minority shareholders through listing on the

US regulated exchanges, where minority investors enjoy better

protection given the regulatory framework (the bonding hypothe-

sis).

After two decades of empirical research, there is no conclu-

sive evidence regarding the true economic benefits of cross-listing

under the hypothesis of capital market segmentation. Some stud-

ies found that cross-listing increases prices reducing the expected

returns (by reducing the cost of equity capital) and risk (e.g.

Alexander, Eun, & Janakiramanan, 1988; Foerster & Karolyi, 1993,

1999; Jayaraman, Shastri, & Tandon, 1993; Miller, 1999; Ramchand

& Sethapakdi, 2000), while other studies have reported results

that do not support the validity of the hypothesis (e.g. Howe &

Kelm, 1987; Howe, Madura, & Tucker, 1993; Lau, Diltz, & Apilado,

1994).

Regarding the bonding/signalling hypothesis, recent studies

conducted by Reese and Weisbach (2002) and Doidge, Karolyi, and

Stulz(2003) f ound preliminary supportive evidence for this theory.

Reese andWeisbach(2002)useda logistic model andprovidedevi-

dence that showedfirms that have poor investor protection in their

home markets tend to list on the US regulated exchanges, in order

to issue new capital. Doidge et al. (2003) computed the Tobin’s-

Q and reported a higher valuation for foreign firms listed on the

1062-9769/$ – see front matter© 2009 The Board of Trustees of the University of Illinois. Published by Elsevier B.V. All rights reserved.

doi:10.1016/j.qref.2009.09.009

http://www.sciencedirect.com/science/journal/10629769

http://www.elsevier.com/locate/qref

mailto:[email protected]

http://localhost/var/www/apps/conversion/tmp/scratch_4/dx.doi.org/10.1016/j.qref.2009.09.009

http://localhost/var/www/apps/conversion/tmp/scratch_4/dx.doi.org/10.1016/j.qref.2009.09.009

mailto:[email protected]

http://www.elsevier.com/locate/qref

http://www.sciencedirect.com/science/journal/10629769

7/28/2019 1-s2.0-S1062976909001094-main


A.A.-N. Abdallah, C. Ioannidis / The Quarterly Review of Economics and Finance 50 (2010) 202–213 203

US regulated exchanges compared to those listed on OTC. Karolyi

(2004) presents a comprehensive survey of the research effort that

explains the reasons for cross-listing in an international context

where he provides significant challenges to the ‘conventional wis-

dom’ regarding the rationale for cross-listing.

In this study we seek to re-examine the work of previous stud-

ies such as Miller (1999) and Foerster and Karolyi (1999), among

others, using an extended cross-listing sample and daily data. We

aim to add to the growing literature in the area of cross-listing by

providing new evidence using a larger sample that uses daily data

for 1165 firms from 47 countries which have been listed on the US

regulated and unregulated stock exchanges. The use of high fre-

quency data in conjunction with the breath of coverage, in terms

of the number of firms and stock exchanges, enhances the validity

of the results.

The analysis clearly shows that there is a “timing issue” asso-

ciated with cross-listing, but finds no evidence on the relation

between listing on regulated exchanges and signalling investor

protection. Abnormalreturns (AR)exhibit a significantdecline after

cross-listing, and the higher the pre-cross-listing AR (as in the case

of firms with IPOs) the higher is the decline in AR after cross-

listing. In addition, we report a positive relationship between the

pre-cross-listing valuation as measured by Tobin’s-Q, and perfor-

manceas measuredby ROAand thepost-cross-listing decline in AR.

Our study is the first to report such results. Moreover, we report a

decrease in local market risk, or beta, which is consistent with the

findings of Foerster and Karolyi (1999). However, we are the first

to provide evidence on themagnitude of such a decrease over time.

Theresults of differentcross-listing periods show that thedecrease

in local beta is diminishing overtime. We also find that the decline

in the post-cross-listing AR (beta) is higher (lower) for firms that

issued capital through cross-listing on the US regulated exchanges

compared to firms that did not issue, which is inconsistent with

Miller (1999) and Foerster and Karolyi (1999). On the other hand,

we did not find any change in the global market risk after cross-

listing, except for firms that cross-listed between 2001 and 2007,

where it decreases significantly. We report evidence that is notin favour of investor protection hypothesis. The decrease in AR is

present in both regulatedand unregulated exchanges,and civil and

common law countries, although the magnitude of the decrease is

higher for firms from common law countries.

This paper is organized as follows. Section 2 develops the

hypotheses of reducing segmentation and signalling investor pro-

tection through cross-listing. Section 3 explains the methodology,

sample, and data collection. The econometric and other statistical

evidence, along with robustness checks, are presentedin Section 4.

Finally, Section 5 concludes.

2. Hypotheses development

2.1. Market segmentation hypothesis

The most extensively examined reason for cross-listing is the

segmentation hypothesis. The theoretical models by Stapleton and

Subrahmanyam (1977), Errunza and Losq (1985) and Alexander

et al. (1988), suggest that under partial or complete segmenta-

tion, domestic investors require a higher rate of return on foreign

security compared to their home securities. When a firm cross-

lists on a foreign market, the risks that are due to the existence of

international investment barriers (segmentation between domes-

tic and foreign markets) are reduced. For example, the lifting of

restrictions on foreign investment, the regulation that governs the

trades in foreign securities, will be consistent with that of domestic

securities. The exchange rate risk will be discounted because the

foreign sharesand theirdividends will bepaidin thecurrencyof the

host country in which the foreign firm is listed.1 Furthermore, the

costs and risks of financial information are likely to decrease due

to the reduction in language barriers and diminishing differences

in accounting standards across countries.2 Cross-listing allows the

cross-listed firms to diversify away from the home market risk of

its shares by exposing them to the international asset markets.3

Therefore, under segmentation, the influence of the home market

risk on stock returns of the cross-listed firm is likely to decrease.

The influence of the foreign market, as measured by the foreign

market’s “beta” is expected to increase (Howe & Madura, 1990).

Hence, as long as the reduction in the domestic beta is lower than

the increase in foreign beta, cross-listing results in improved risk

diversification.4 Based on these arguments, we formulate the fol-

lowing hypotheses:

H1. Cross-listing will reduce the home market risk of the CL firms

after the cross-listing.

H2. Cross-listing will increase the foreign market risk of the CL

firms after the cross-listing.

2.2. Investor protection hypotheses

Another reason for cross-listing that has emerged recently in

the literature is the commitment to increase the level of investor

protection through cross-listing on an exchange with better reg-

ulations in order to issue capital domestically or internationally

(e.g. Coffee, 1999, 2002; Stulz, 1999). This is known as the bonding

hypothesis, which assumes that the cost of external financing for

firms with poor investor protection is higherthan that of firms with

good investor protection.

The hypothesis suggests that the private benefit of control

increases the risk to outsiders (i.e. minority investors) and subse-

quently the required return on the firm’s equity. This prevents the

insiders (the controlling shareholders/managers) from raising the

required capital and limits their ability to finance future growth

opportunities. The insiders will decide to cross-list on a foreign

exchange with higher investor protection regulations if the size of

the increase in the public value of shares is relatively larger than

the fall in the private benefit. This lowers the risk of expropriation

by the insiders and increases the public value of the firm’s shares,

which enables the firm to issue equity at a lower cost of capital.

Previous empirical evidence by La Porta, Lopes-de-Silanes,

Shleifer, & Vishny (1997, 1998) shows that the US has the high-

est level of investor protection compared to other countries. For

example, the anti-director rights index and accounting standards

(measures of investor protection) for the US is 5 and 71, respec-

tively, compared to an index that is below 5 and 70 for other

countries. This suggests that non-US firms that have listed in the

US committo increase the level of investor protection expecting to

reduce the required return on their shares.When dividing the sample into civil and common law coun-

tries, La Porta et al. (1997, 1998) shows that investor protection

in civil law countries (French, German and Scandinavian origin) is

1 The shares and dividends of all foreign firms listed in the US are in US dollars.2 Forexample, whena non-Englishspeaking firmcross-listsin theUS it isrequired

to report its accounting information in English and reconcile – partially or fully –

this information to US GAAP.3 Before cross-listing, the stock market risk (the covariance between the security

and market) is influenced only by home market.4 Other motivations for cross-listing are (1) increasingsales revenues, and hence,

profitability, by promoting the firm’s brands internationally (Pagano, Roell, &

Zechner, 2002), and (2) decreasing the level of the firm’s leverage by issuing more

equity (Davis-Friday, Frecka, & Rivera, 2005; Pagano et al., 2002).

7/28/2019 1-s2.0-S1062976909001094-main


204 A.A.-N. Abdallah, C. Ioannidis / The Quarterly Review of Economics and Finance 50 (2010) 202–213

lower than that of common law countries.5 Based on the above

arguments, we formulate the following hypotheses:

H3. Cross-listing on the US regulated exchanges (AMEX, NAS-

DAQ, and NYSE) will signal the firm’s commitment to increase the

level of investor protection as it lowers the risk of expropriation to

investors. This will reducethe expected rate of returnon theshares

of the cross-listed firm after the cross-listing.

H4. Given that common law countries provide better shareholder

protection than civil law countries, thedecrease in expected rate of

returnshould be higherfor firms from civil lawcountries compared

to firms from common law countries (mainly of English speaking

origin).

If signalling investor protection affects the share price of the

cross-listed firm, one can expect a negative relation between the

post-cross-listing abnormal return and investor protection mea-

sures (IPM). This implies that the lower the level of investor

protection, the higher the abnormal return. Our hypothesis there-

fore is:

H5. If the expected increase in prices of the cross-listed firms

comes in part from the insider’s commitment to increase the level

of investor protection through cross-listing, then there should be

a negative relation between the post-listing abnormal return and

investor protection’s measures.

2.3. Cross-listing, overvaluation, and timing hypotheses

Previous research hasshown a dramatic decrease in post-listing

AR without providing a testable explanation for such evolution

(see Alexander et al., 1988; Howe & Kelm, 1987; Lau et al., 1994;

Lee, 1991; Martell, Rodriguez, & Webb, 1999; Foerster and Karolyi,

1993, 1999). Such a pattern issimilarto oneobservedin IPO studies,where theARs show a drastic fall afterthe original floatation(Ritter,

1991). These patterns suggest that the decision to cross-list has a

strong timing motivation in addition to the ones already explored.

Our information set takes into account a measure of the firm’s per-

formance and market valuation to explore the timing issues that

are involved in the decision to cross-list. Firms with ‘good domes-

tic performance’ and market valuation have strong incentives to

cross-list as they take advantage of the possible overvaluation that

it is associated with their reported financial results. The higher the

overvaluation of thefirm, thelowerthe AR will be,sincethe market

adjusts to the ‘fundamental’ level of valuation. Thus we expect the

following:

H6. As a result of the pre-cross-listing price overvaluation and

timing of cross-listing, the post-cross-listing abnormal return is

expected to be negative.

It follows that:

H7. The relationship between the post-cross-listing abnormal

return (which is generally negative) and the chosen measure of

firm performance and valuation is positive.

5 For example, the average values of the anti-director rights index for French,

German and Scandinavian countries is 2.33, 2.33, and 3, respectively, compared to

4 for that of common law countries. See La Porta et al. (1997, 1998), Tables, 2, 3,

and 5.

3. Methodology, sample and data

3.1. Methodology

3.1.1. Event study analysis

Following Foerster and Karolyi (1999) and Miller (1999), we

measure the effects of cross-listing on security prices by estimat-

ing abnormal return (AR) and cumulative abnormal return (CAR).

To observe thechange in ARover thenumber ofdaysincludedin the

a-priori chosen time window, a standard event study is employed.

The ARfor each firm using the one-factormarketmodel is given as:

ARit = Rit − ( ˆ̨ + ˆ̌ Li

RLmt ) (t = −300,+250) (1)

where ARit denotes the abnormal return for firm i at day t over the

prediction period (−100, +250) relative to day 0, the cross-listing

date. The estimates of the parameters are generated by regressing

firm i’s actual return, Rit , over its local market returns (RLmt ) during

the estimation period (−300, −101).

Theabnormal returns arethen accumulated foreach firmacross

the prediction period days (−100, +250) as:

CARi,t =

t 1+1t =t 1

ARi,t (t = −100,+250) (2)

The CARs are then averaged for each day across all securities to

obtain the cumulative average abnormal return as follows:

CAARt =

N

i=1CARit

N (t = −100,+250) (3)

where N denotes the number of firms in the sample at each day

(this number is the same across all days). If cross-listing were to

prove beneficial to the firm, one would expect that the cumula-

tive abnormal returns will be positive (at least within the event

window).

3.1.2. Cross-sectional and time series analysisHowe and Madura (1990), Jayaraman et al. (1993) and Foerster

and Karolyi (1999) argue that after cross-listing, expected returns

are generated by two factors, namely domestic and foreign market

returns or the global index, whereas before cross-listing it is only

generated by one factor; the domestic market return.

Following the methodology of Foerster and Karolyi (1999), we

estimate a cross-sectional and time series two-factor IAPM model

as follows:

Rit = ˛PRE i +ˇPRE

iL RLmt + ˇPRE

iW RW mt + ˛LIST

i DLIST it + ˛POST

i DPOST it

+ˇPOST iL

RLmt D

POST it + ˇPOST

iW RW

mt DPOST it + εit (t = −250,+250)

(4)

where Rit is the daily return for firm i at time t ; ˛i’s are constantsthat are interpreted as the AR; ˇi,L’s are the coefficients (the local

market risk) associated with the daily local/home market return,

RLmt ; ˇiW ’s are the coefficients (the foreign market risk) on the

daily DataStream world market return index, RW mt , which isa value-

weighted index; DLIST it

is a dummy variable that takes the value

one in the three days around the cross-listing date (−1, 0, +1) and

zero otherwise; DPOST it

is a dummy variable that equals one in the

post-cross-listing period (+2, +250) and zero otherwise.

We link theresults from theeventstudy tothe parameters of the

model as follows: the benefits of cross-listing can be captured by

the coefficients ˛LIST i

and ˛POST i

. These are expected to be positive,

implyinga higherAR inthe threedaysaround andafter cross-listing

compared to the pre-cross-listing period. Hypotheses H1 and H2

can be tested from the sign of the parameter ˇPOST iL that is expected

7/28/2019 1-s2.0-S1062976909001094-main



to be negative implying lower market risk in the home market,

whereas ˇPOST iW

is expected to be positive indicating a higherforeign

risk after cross-listing.

3.1.3. Testing the timing of cross-listing, and the relation between

the post-cross-listing abnormal returns and the level of investor

protection

To testthe market signalling/bonding and timing issuehypothe-ses we examine the evolution of ARs. Doidge et al. (2003) provide

some heuristic evidence that firms that cross-list in the regulated

US market exhibit higher value growth compared to firms that

cross-list in the unregulated US market. They attribute this dif-

ference to the existence of investor protection, without providing

direct econometric evidence that links firm growth to the metrics

of investor protection, which has been proposed by La Porta et al.

(1997, 1998). We develop the following regressions:

˛POST i = o +1Tobin’s-Q i,PRE +2IPM +3i +4i

×Exchange Dummy+5LNVOi,PRE +6ROAi,PRE

+7LNMV i,PRE + EDM i + ε (5)

The dependent variable in Eq. (5) is the estimated AR’s, the

coefficient ˛POST i

from Eq. (4). LNMV i,PRE is the natural log of the

pre-cross-listing market value. The inclusion of the size variable is

motivated by the reporting of evidence on size anomalies, which

negatively links the size of the firm to AR after the realization of a

major event (Fama & French, 1992). The variable LNVOi denotes the

natural log of the trading volume. Once the firm is cross-listed, the

pool of potential investors increases as foreign investors can now

trade in the firm’s shares. Foerster and Karolyi (1998) report that

the increased trading volume for cross-listed firms has a positive

impact on the firms’ liquidity (see also Kyle, 1985), thus exerting

a positive influence on the firms’ ARs. EDM i is a dummy variable

that equals one if the firm is from a developed country and zero

otherwise. Miller (1999) reports a higher AR for firms from emerg-ing markets compared to firms from developed markets during the

three days around cross-listing.

For investor protection, we use three measures—the account-

ing standards rating index, anti-director rights index, and whether

the firm is from a civil or common law country.6 As the measures

of investor protection are highly collinear we do not include all of

them in a singleequation. Given ourhypothesis, 2 shouldbe nega-

tive, which means that the effect of cross-listing on shares prices of

firms from a poor investorprotectionsystem shouldbe highercom-

paredto firms originating from countries where minority investors

are strongly protected. In addition, we include Merton’s (1987)

cost of incomplete information, the shadow cost (k) of security

k, which plays a role in determining the expected return on secu-

rity k. Empirically, Foerster and Karolyi (1999) find the change inthe shadow cost, i, to be significantly related to a cross-listed

firm’s abnormal returns and the change in its local and global beta.

The shadow cost, i is calculated as:

i = ( 2i,εSIZE i)

1

SHRi,t +1−

1

SHRi,t

(6)

Size is the market capitalization of the cross-listed firm, and

SHR is its number of outstanding shares. Finally, our information

6 Theuse of theaccounting standard index as a credible measure of investor pro-

tectionis supported furtherby recent research,Bradshaw,Bushee,and Miller (2004),

which finds the adoption of US-GAAP by foreign firms listed in the US increases the

shareholding by US institutional investors.

set takes into account a measure of the firm’s performance and

market valuation to explore the timing issue that is involved in

the decision to cross-list. Firms with ‘good domestic performance’

and market valuation have strong incentives to cross-list as they

take advantage of the possible overvaluation that is associated

with their reported financial results. The higher the firm overval-

uation, the lower the AR will be, since the market adjusts to the

‘fundamental’ level of valuation. Thus we expect a positive rela-

tionship between the post-listing evolution of abnormal returns

(whichare generallynegative) andthe chosenmeasure of firmper-

formance. We use the average pre-cross-listing three years return

on assets, ROAi,PRE as a measure of a firm’s performance, and the

pre-cross-listing Tobin’s-Q; as a measure of the firm’s valuation in

the pre-cross-listing period, and is calculated as:

Tobin’s-Q i =BVTAi − BVE i +MVE i

BVTAi

where BVTA, BE , and MVE stand for the book value of total assets

(DS #WC02999), the book value of equity (DS #WC03501), and

the market value of equity (MV), respectively. Doidge et al. (2003)

argue that firms with growth opportunities have an incentive to

list in the US in order to raise capital. They find that these firmsare valued more highly than other firms in the sample. In the light

of our discussion on the time of cross-listing, we expect a positive

relationship between the pre-cross-listing Tobin’s-Q and pre-AR,

indicating that firms with higher valuation in the pre-cross-listing

period will experience higher pre-cross-listing ARs. On the other

hand, we expect a negative relationship between the pre-cross-

listing Tobin’s-Q and the post-cross-listing ARs, meaning that firms

with higher Tobin’s-Q in the pre-cross-listing period will experi-

ence lower post-cross-listingARs.This is an indication ofthe timing

issue.

3.2. Sample and data

The initial sample consisted of 2689 firms from 47 countriesthat have cross-listed on the US exchanges (AMEX, NASDAQ, NYSE,

OTC, and PORTAL) between 1976 and 2007. To avoid the survival

bias, the sample includes de-listed firms. The sample was collected

from the various stock exchanges websites, the research depart-

ment of the NYSE for foreign firmslisted on the NYSE, and the Bank

of New York. Firms with missing market and return data for the

period (−250, 0, +250) days relative to the cross-listing day (day 0)

were eliminated. The resulting sample consists of 1165 firms, each

spanning the same time-period around the listing date.7,8

7 Daily return data for the range of countries that we are examining is not fre-

quently available as monthly or weekly return data.Even studies thatemploy event

study methodology and used weekly data were able to obtain complete data setsforproportions of their original sample. Forexample, out of 317foreignfirms listed

in the US between 1976 and 1992, Foerster and Karolyi (1999) were only able to

get weekly data for 153 firms. Price data in Datastream is only available from 1974,

hence, as the data is required for 250 days before and 250 days after the date of

cross-listing, the 1165 firmsare those firmsthat havecross-listedduring the period

1976-2007.8 Using daily data instead of weekly or other lower frequency data improves the

efficiency of the covariance matrices thus increasing the power of the tests, ren-

dering our inferences more robust than otherwise. Schotman and Zalewska (2006)

report increased estimatesof standard errors using weekly data,comparedto those

obtained using daily. The use of lower frequency data, such as weekly, cannot be

seen as a solution to the asynchronous trading phenomenon, as it simply ‘spreads’

the ‘missing day’ information across each day along the week. Finally, daily data

allows one to examine the cross-listing day and a short period around it (−1, +1). A

reasonable solution to the problem of asynchronous trading is the appropriate lag-

ging of the foreign index depending upon the geographical location of the host and

home countries. Our sample is much larger than previous studies on cross-listing.

7/28/2019 1-s2.0-S1062976909001094-main



Table 1

Descriptive statistics for 1165 firms which have cross-listed on the US regu-

lated (AMEX, NASDAQ, and NYSE) and unregulated (OTC, and PORTAL) exchanges,

between 1976 and 2007.

Category Regulated Unregulated Total

By exchange

AMEX 13 13

NASDAQ 138 138

NYSE 218 218OTC 606 606

PORTAL 190 190

By region

Asia 51 405 456

Australia and New Zealand (NZ) 19 75 94

Canada 134 134

Europe (excluding UK) 62 155 217

Latin America 36 47 83

Middle East/N. Africa (NA) 6 13 19

South Africa 4 31 35

UK 57 70 127

By industry

Consumer good 55 136 191

Financials 50 117 167

Industrial 83 221 304

IT 32 64 96Resources 56 81 137

Services 77 153 230

Utilities 16 24 40

By level

Capital raising 106 46 152

No capital raising 263 750 1013

By years

1976–1983 19 112 131

1984–1990 40 27 67

1991–1995 73 195 268

1996–2000 160 190 350

2001–2007 78 271 349

We followed Foerster and Karolyi (1999) and group firms by

exchange, region, industry,levels, and cross-listing periods. Table 1shows that across exchanges, OTC attracts the highest number of

foreign firms (606) due to its lowest listing fees and requirements

compared to other US exchanges. This is followed by NYSE (218),

PORTAL(190),NASDAQ (138), and AMEX (13). The highest number

of firms in the sample is from Asia (456), followed by Europe (217),

Canada (134), and UK (127). The lowest group is firms from the

Middle East and North Africa (19). Splitting the sample by industry

shows that general industrial is the largest group of firms (304),

followed by service firms (230), financials (167), consumer goods

(191), resources (137), information technology (96), and utilities

(40). Furthermore, the table shows that out of 1165 cross-listed

firms, 152 firms raised capital through OTC (46), and ADR level III

(106). Finally, grouping firms by years results in 131 firms for the

period 1976–1983, 67 firms between 1984 and 1990, 268 firmsbetween 1991 and 1995, 350 firms between 1996 and 2000, and

finally 349 firms between 2001 and 2007, all of which have com-

plete daily market data for the period (−250, 0, +250).

4. Empirical results

4.1. Data descriptive statistics

Table 2 presents descriptive statistics for the daily return data,

which were obtained from DataStream International. For the entire

sample the average dailyreturnis positive andsignificant(0.00113)

in the pre-cross-listing period (−250,−2), positive but insignificant

(0.00101) during the three days around cross-listing (−1, 0, +1),

and declines dramatically to 0.00024 after cross-listing, by about

91.10%. This decline in returnaftercross-listing is consistent across

different regions. The highest decline in return after cross-listing is

for South African firms (307.29%), followed by Australian and New

Zealand firms listedon unregulated exchanges (198.37%), and then

Asian firms listed on regulated exchanges (145.55). The pattern of

the decline in firms’ returns after cross-listing is similar to that

of IPO companies after the IPO, and thus, Table 2 provides an ini-

tial evidence to suggest a timing issues where firms cross-list in a

period of good performance.

4.2. Results of event study

Table 3 presents the results for the average daily abnormal

return (AR) and cumulative average abnormal return (CAR) for the

event period (−100, +250) for both US regulated and unregulated

exchanges. For firms that are cross-listed on regulate exchanges,

we observe a positive andstatistically significant AR (AR= 0.00474)

on day −1, positive but insignificant on day 0 (AR = 0.00147), and

negative throughoutthe entire post-cross-listing period (+1, +250),

exhibiting a dramatic decline. However, the pattern of the CAR

shows a positive short period reaction to cross-listing. We observe

a positive CAR throughout the period (−100, +10), which is statis-

tically significant for most of the days, particularly between days−40 and +8, and reaching its peak value at day −1 (+0.0473), and

day 0 (+0.0487). This is consistent with the market segmentation

hypothesis, which implies a decrease in the cost of capital and an

increase in share value due to the removal of international invest-

ment barriers through cross-listing (e.g. Errunza and Losq, 1985).

Our evidence of the positive CAR is consistent with Miller (1999)

whoreportsa positive ARfor day0 and+1, andpositive butinsignif-

icant CAR between days −20, and +25. Nonetheless, Table 3 shows

that the benefits from cross-listing do not last after day +8; CAR

remains positive but insignificant between days +9 and +54, and

turns negative, which is statistically significant, throughout the

period(+55, +250). These negative CARresults arein harmony with

previous studies such as Howe and Kelm (1987), Lau et al. (1994),

Martell et al. (1999), and Foerster and Karolyi (1993, 1999) thatreport CAR subsequent to cross-listing in the US. In addition, the

pattern of CAR observed in our study is similar to that reported in

the IPO literature, suggesting thatthe market revaluate cross-listed

shares after being overvalued during the pre-cross-listing period.

A similar pattern of AR/CAR can be also realized from cross-listing

on the US unregulated exchanges (OTC and PORTAL). However, for

these firms, ARis only significantat days +55, +160,and +230 where

itispositive,andatday+210whereitisnegativeandtheCARispos-

itive andonly significant at days−90,−80,−75,−74,−70,and−60.

No significant CAR is being observed subsequent to cross-listing.

4.3. Results of the cross-sectional and times-series analyses

4.3.1. Results of testing segmentation hypothesesTable 4 presents the results of cross-sectional times-series anal-

yses (model 4) by foreign exchanges. In all regressions, the mean

AR (˛PRE i

) in the pre-cross-listing period (−250, −2) is positive and

significant. This is a daily average of 0.053% for the overall sample,

0.083% for AMEX, 0.141% for NASDAQ, 0.078%for NYSE, 0.01% for

OTC, 0.083% for PORTAL, These numbers are similar in magnitude

to that reported by previous studies.9 By contrast, no significant

9 Forinstance, while Jayaraman etal. (1993)reports a dailyaverageof 0.07%in the

pre-cross-listing period, Martell et al. (1999) show an average of 0.09% at day −2.

Furthermore, this pre-listing AR amounts to 0.38% per week or 1.617% per month

thatisalsoinsimilarrangetothatreportedbypreviousstudies.Forexample Foerster

and Karolyi (1999) report an average of 0.31% per week, whereas Alexander et al.’s

(1988) pre-listing ARs range between 1% and 4.37% per month.

7/28/2019 1-s2.0-S1062976909001094-main



Table 2

Descriptive statistics for the daily returns of cross-listed firms. The daily return is computed as RI t +1/RIt , where RI is the firm’s return index obtained from DataStream

International. The percentage decline in return is calculated as [(post return−pre return)]/pre return]×100.

Pre-cross-listing (−250, −2) Cross-listing (−1, 0, +1) Post-cross-listing (+2, +250)

NOBS Mean pv-t NOBS Mean pv-t NOBS Mean pv-t %Decline

All firms 290,199 0.00113 (0.000) 1165 0.00101 (0.399) 289,562 0.00024 (0.000) −91.10

Regulated

All 92,250 0.00168 (0.000) 369 0.00302 (0.212) 92,250 0.00015 (0.139) −91.10Asia 12,750 0.00144 (0.000) 51 0.00020 (0.973) 12,750 −0.00065 (0.018) −145.55

Australia and NZ 4750 0.00139 (0.002) 19 0.00127 (0.910) 4750 −0.00036 (0.471) −125.96

Canada 33,500 0.00225 (0.000) 134 0.00085 (0.871) 33,500 0.00058 (0.002) −74.37

Europe 15,500 0.00134 (0.000) 62 0.00192 (0.571) 15,500 −0.00019 (0.400) −114.53

Latin America 9000 0.00145 (0.000) 36 0.00455 (0.583) 9000 0.00043 (0.103) −70.35

Middle East/NA 1500 0.00201 (0.002) 6 0.02030 (0.247) 1500 0.00082 (0.226) −59.03

South Africa 1000 0.00076 (0.343) 4 0.02254 (0.148) 1000 −0.00158 (0.052) −307.29

UK 14,250 0.00120 (0.000) 57 0.00827 (0.021) 14,250 0.00028 (0.194) −76.42

Unregulated

All 199,000 0.00087 (0.000) 796 0.00007 (0.957) 199,000 0.00028 (0.000) −67.47

Asia 101,250 0.00090 (0.000) 405 0.00207 (0.156) 101,250 0.00038 (0.000) −57.61

Australia and N Z 18,750 0.00087 (0.035) 75 0.00582 (0.232) 18,750 −0.00086 (0.007) −198.37

Europe 38,750 0.00058 (0.001) 155 −0.00707 (0.093) 38,750 0.00037 (0.156) −36.72

Latin America 11,750 0.00157 (0.000) 47 0.00373 (0.253) 11,750 0.00102 (0.000) −34.72

Middle East/NA 3250 0.00381 (0.000) 13 −0.01202 (0.198) 3250 0.00191 (0.007) −49.82

South Africa 7750 0.00136 (0.000) 31 0.00175 (0.729) 7750 0.00017 (0.517) −87.24

UK 17,500 0.00016 (0.462) 70 −0.00278 (0.632) 17,500 0.00001 (0.956) −92.33

Table 3

Event study results for 1165 firms which have cross-listed on the US regulated and unregulated exchanges between 1976 and 2007. The p-values of the t -statistic are

presented in parentheses. Daily abnormal returns (AR) and cumulative abnormal returns (CAR) are calculated using the one-factor market model with an estimation period

of (−300, −101). ARit = Rit − ( ˆ̨ + ˆ̌ Li

RLmt ) (t = −250,+250) Eq. (1).

Event date Regulated (N = 369) Unregulated (N =796)

AR pv-t CAR pv-t AR pv-t CAR pv-t

−100 0.00150 (0.337) 0.00150 (0.337) 0.00017 (0.860) 0.00017 (0.860)

−90 0.00152 (0.170) 0.00350 (0.467) 0.00024 (0.812) 0.00724* (0.063)

−80 −0.00053 (0.690) 0.00809 (0.231) 0.00163 (0.132) 0.01241* (0.066)

−75 −0.00023 (0.867) 0.01247* (0.092) 0.00090 (0.450) 0.01445* (0.082)

−74 −0.00068 (0.617) 0.01179 (0.126) 0.00054 (0.556) 0.01499* (0.078)

−70 −0.00070 (0.563) 0.01188 (0.156) 0.00104 (0.285) 0.01785* (0.058)

−60 −0.00139 (0.351) 0.01037 (0.305) 0.00016 (0.885) 0.01805 (0.145)

−50 0.00063 (0.654) 0.01858 (0.121) −0.00075 (0.435) 0.01929 (0.188)

−40 0.00121 (0.400) 0.02807** (0.049) 0.00072 (0.521) 0.02260 (0.221)

−30 −0.00027 (0.851) 0.03008** (0.047) 0.00065 (0.481) 0.02392 (0.253)

−20 0.00019 (0.889) 0.03654** (0.028) −0.00013 (0.895) 0.02485 (0.298)

−10 0.00234 (0.153) 0.03938** (0.037) 0.00017 (0.855) 0.02299 (0.389)

−1 0.00474*** (0.007) 0.04726** (0.023) 0.00015 (0.905) 0.02150 (0.476)

0 0.00147 (0.523) 0.04873** (0.020) −0.00128 (0.320) 0.02022 (0.507)

+1 −0.00442** (0.041) 0.04430** (0.032) −0.00264 (0.152) 0.01758 (0.570)

+2 −0.00305 (0.263) 0.04126** (0.046) −0.00167 (0.105) 0.01591 (0.610)

+3 0.00058 (0.758) 0.04184** (0.047) −0.00160 (0.559) 0.01431 (0.655)

+4 0.00133 (0.634) 0.04317** (0.047) 0.00387 (0.197) 0.01818 (0.566)

+5 −0.00038 (0.817) 0.04279* (0.052) 0.00065 (0.499) 0.01883 (0.551)

+6 −0.00225 (0.111) 0.04054* (0.067) −0.00045 (0.628) 0.01838 (0.564)

+7 −0.00282 (0.037) 0.03771* (0.088) −0.00026 (0.827) 0.01812 (0.576)

+8 −0.00035 (0.819) 0.03736* (0.094) 0.00058 (0.514) 0.01870 (0.568)

+10 −0.00450*** (0.001) 0.03124 (0.162) 0.00082 (0.349) 0.01900 (0.567)

+55 −0.00035*** (0.005) −0.00222 (0.935) 0.00282** (0.019) 0.01343 (0.766)

+100 −0.00206 (0.107) −0.05504 (0.108) 0.00058 (0.627) −0.01930 (0.710)+110 −0.00353** (0.025) −0.06903* (0.052) 0.00014 (0.914) −0.02062 (0.713)

+120 0.00044 (0.724) −0.07659** (0.038) −0.00055 (0.611) −0.02080 (0.725)

+130 −0.00039 (0.762) −0.09221** (0.016) −0.00054 (0.553) −0.02570 (0.676)

+140 −0.00086 (0.538) −0.10809*** (0.006) 0.00033 (0.754) −0.03076 (0.629)

+150 −0.00385*** (0.002) −0.11768*** (0.004) 0.00024 (0.839) −0.03571 (0.595)

+160 0.00215 (0.090) −0.12038*** (0.004) 0.00187* (0.078) −0.03637 (0.604)

+170 −0.00290* (0.051) −0.13445*** (0.002) −0.00040 (0.672) −0.03792 (0.602)

+180 −0.00132 (0.328) −0.14221*** (0.002) −0.00011 (0.936) −0.04490 (0.552)

+190 −0.00215* (0.096) −0.16153*** (0.001) −0.00150 (0.117) −0.04202 (0.594)

+200 0.00056 (0.675) −0.17443*** (0.001) −0.00109 (0.486) −0.04023 (0.618)

+210 −0.00313** (0.013) −0.17730*** (0.000) −0.00230* (0.098) −0.04376 (0.591)

+220 −0.00174 (0.186) −0.18737*** (0.000) −0.00022 (0.823) −0.04249 (0.615)

+230 −0.00188 (0.153) −0.19566*** (0.000) 0.00168* (0.093) −0.04045 (0.649)

+240 −0.00274* (0.077) −0.22036*** (0.000) 0.00072 (0.476) −0.04579 (0.616)

+250 −0.00411*** (0.001) −0.23812*** (0.000) 0.00089 (0.466) −0.04600 (0.623)

***, **, and * indicate significance at 1%, 5%, and 10% levels.

7/28/2019 1-s2.0-S1062976909001094-main



Table 4

IAPM market model for 369 firms, which have cross-listed on the US regulated exchanges. Rit = ˛PRE i

+ ˇPRE iL

RLmt +

ˇPRE iW

RW mt +

˛LIST i

DLIST it

+ ˛POST i

DPOST it

+ˇPOST iL

RLmt

DPOST it

+

ˇPOST iW

RW mt D

POST it

+ εit (t = −250,+250) (Eq. (4)). The regression is run for each firm to obtain the regression parameters that are then averaged across groups. Rit and RLmt are

thedailyreturnsfor firm i andits weightedaveragelocal marketindex at time t , respectively. ˛i ’s are constantsthat are interpreted as abnormal returns. ˇiL and ˇiW ’sarethe

local and foreign market risk, respectively. RW mt is the world daily weighted average return index. DLIST

it is a dummy variable that takes the value one in the three days around

cross-listing and zero otherwise. DPOST it

is a dummy variable that equals one in the post-cross-listing period (+2, +250) and zero otherwise. is the difference between the

post- and pre-cross-listingperiods(post-pre).The cross-listing period (CL)refersto three daysaround cross-listing (−1,0, +1).KW -2 is thetwo-tailed Wilcoxon signed-rank

test for the difference in the regression parameters across groups. Returns data are obtained from DataStream.

Category ˛PRE

i ˇPRE

iL ˇPRE

iW ˛CL

i ˛POST

i ˇPOST

iL ˇPOST

iW Adj. R2

% ˛ ˇiL ˇiW

All firms 0.00053*** 0.8107*** 0.0681***−0.0008 −0.0007***

−0.011 1 0.031 0* 21.07 −0.0012***−0.8218***

−0.0371***

(0.000) (0.000) (0.000) (0.217) (0.000) (0.546) (0.069) (0.000) (0.000) (0.209)

AMEX 0.00083 0.6380**−0.0461 −0.0077 −0.0011 −0.2583* 0.2619 5.64 −0.0019 −0.8963** 0.3079

(0.229) (0.012) (0.725) (0.436) (0.172) (0.086) (0.102) (0.157) (0.015) (0.224)

NASDAQ 0.00141*** 0.7984*** 0.1570*** 0.0005 −0.0016*** 0.0273 0.0625 19.41 −0.0030***−0.7711***

−0.0945

(0.000) (0.000) (0.000) (0.826) (0.000) (0.525) (0.189) (0.000) (0.000) (0.216)

NYSE 0.00078*** 0.8682*** 0.0897** 0.0010 −0.0009***−0.039 6 0.095 3** 22.68 −0.0017***

−0.9078*** 0.0056

(0.000) (0.000) (0.031) (0.392) (0.000) (0.364) (0.020) (0.000) (0.000) (0.941)

OTC 0.0001*** 0.7759*** 0.0466 −0.0010 −0.0003**−0.0102 0.0026 18.74 −0.0004**

−0.7861***−0.0440

(0.173) (0.000) (0.004)*** (0.295) (0.018) (0.725) (0.907) (0.041) (0.000) (0.224)

PORTAL 0.00083*** 0.8769*** 0.0551 −0.0028**−0.0011*** 0.0078 0.0090 28.91 −0.0019***

−0.8691***−0.0461

(0.000) (0.000) (0.243) (0.017) (0.000) (0.750) (0.854) (0.000) (0.000) (0.621)

KW -2 41.40*** 9.01* 14.02*** 10.51** 35.46*** 4.36 6.18 42.59*** 2.76 6.43

(0.000) (0.061) (0.007) (0.033) (0.000) (0.360) (0.186) (0.000) (0.599) (0.169)

*** , **, and * denote significance at 1%, 5%, and 10% level, respectively.

AR is detected during the three days around cross-listing (˛LIST i

is

not significant),10 but a significant drop in the post-listing period

(+2, +250), ˛POST i

is negative and significant. The only exception is

PORTAL, where ˛LIST i

is negative and significant, which is inconsis-

tent with Miller (1999) who reports negative but insignificant AR.

Thedropin the AR following cross-listing hits a daily peak of 0.07%

for the overall sample, 0.11% for AMEX, 0.16% for NASDAQ, 0.09%

for NYSE, 0.03% for OTC, 0.11% for PORTAL.11 Except for AMEX, all

other post-AR coefficients are significant at 1% level. The highest

decline in AR is for NASDAQ foreign firms, which experience the

highest pre-cross-listing AR followed by PORTAL, OTC, and NYSE.Reported Chi-squares show that the mean and median AR signifi-

cantly different across stock exchanges and between the pre- and

post-cross-listing periods.

As for the market risk, the table shows that local beta decreases

after cross-listing for all firms in the sample, which is consistent

with the prediction of H1. The highest decrease in local beta is for

NYSE (−0.9078), followed by AMEX (−0.8963), PORTAL (−0.8691),

OTC(−0.7861), and NASDAQ (−0.7711), all of whichare statistically

significant at 1% level. On the other hand, we report insignificant

change in the world market risk, suggesting no change in global

betas of cross-listed firms following the US listing.

The econometric evidence shows an overall reduction in equity

pricesfor thesample of 1165 cross-listedfirms andsuch resultcon-

tradicts prior empirical studies on cross-listing that report priceincreases following such events (Alexander et al., 1988; Foerster

& Karolyi, 1993; Jayaraman et al., 1993; Miller, 1999; Ramchand

& Sethapakdi, 2000; Torabzadeh, Bertin, & Zivney, 1992). The

decrease in local beta is consistent with the results reported by

Foerster and Karolyi (1999) and Ramchand and Sethapakdi (2000).

Nonetheless, the results are consistent with those reported by

Howe et al. (1993), Lau et al. (1994), and Lee (1991), Foerster

10 Theresultsdo notchangewhenonlycontrollingfor theday ofcross-listing(day

0), or for 14 days around day 0.11 This isalso ina similar rangeto that reportedin previousstudies (e.g. Alexander

et al., 1988, Jayaraman et al., 1993, Foerster and Karolyi, 1999; and Martell et al.,

1999).

and Karolyi (1999), who report a negative AR in the post-listing

period.

It is possible that the difference between the results of prior

research and this study is due to the size of the sample. This is

an extended sample that covers more countries, employs higher

frequency daily data and evaluates the performance over a longer

window. Theotherexplanation forthe decline in thepost-listing AR

is based on the timing of the cross-listing where managers/or the

control group of the firm (who take the cross-listing decision) take

advantage of the firm’s temporary accomplishments and cross-list

only during in periods of exceptional performance.

4.3.2. Testing the change in AR and betas across groups

Following previous studies on cross-listing, we control for the

effect of differences across countries, between firms that issued

shares and firms that did not issue, and over time. Table 5

reports results for foreign firms that have cross-listed on regulated

exchanges. Consistent with previous studies, there is a wide vari-

ation of the cross-listing effects across groups. The results of the

region analysis show that AR and local market risk decline sig-

nificantly after cross-listing for all firms in different regions, but

global beta did not change. The highest decline in AR (−0.0031)

is for Canadian firms, which experience the highest positive AR

(0.00166) prior to cross-listing. This evidence lends more support

to the timing and overvaluation argument discussed before. On theother hand, South African firms enjoy the highest decline in their

exposure to local market risk, local beta declines by −1.2380, with

no change in their exposure to global risk following cross-listing.

None of thefirms in thesample experience a statistically significant

change in their exposure to the global market risk.

Similarly, firms that raised capital through issuing shares have

a pre-cross-listing AR of 0.00106, which is higher than 0.00100

for firms that do not issue capital. Nonetheless, the decline in AR

for the former group is 0.0026, which is higher than the Latter

(AR=−0.0020) by about 0.0006. This is consistent with the IPO

and overvaluation literature stated earlier (see Ritter, 1991), which

documented that firms with the highest average adjusted initial

return tend to have the worst post-listing IPO decline in AR. Our

results are inconsistent with Foerster and Karolyi (1999) and Miller

7/28/2019 1-s2.0-S1062976909001094-main



Table 5

IAPM market model for 369 firms, which have cross-listed on the US regulated exchanges. Rit = ˛PRE i

+ˇPRE iL

RLmt +

ˇPRE iW

RW mt +

˛LIST i

DLIST it

+˛POST i

DPOST it

+ ˇPOST iL

RLmt

DPOST it

+

ˇPOST iW

RW mt D

POST it

+ εit (t = −250,+250) (Eq. (4)). The regression is run for each firm to obtain the regression parameters that are then averaged across groups. Rit and RLmt are

thedailyreturnsfor firm i andits weightedaveragelocalmarket index at time t , respectively. ˛i ’s are constantsthat are interpreted as abnormal returns.ˇiL and ˇiW ’sarethe

local and foreign market risk respectively. RW mt is the world daily weighted average return index. DLIST




post- andpre-cross-listingperiods(post–pre).The cross-listing period (CL)refersto three daysaround cross-listing(−1,0, +1).KW -2 is thetwo-tailed Wilcoxonsigned-rank


Category ˛PRE

i ˇPRE

iL ˇPRE

iW ˛CL

i ˛POST

i ˇPOST

iL ˇPOST

iW Adj. R2

% ˛ ˇiL ˇiW

All firms 0.00102*** 0.8340*** 0.1101*** 0.0006 −0.0012***−0.0223 0.0889*** 20.86 −0.0022***

−0.8563***−0.0212

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.469) (0.004) (0.000) (0.000) (0.694)

By region

Asia 0.00085** 1.1181*** 0.0419 −0.0013 −0.0015***−0.0229 0.0126 29.83 −0.0023***

−1.1410***−0.0294

(0.019) (0.000) (0.298) (0.609) (0.006) (0.612) (0.797) (0.008) (0.000) (0.725)

Australia and NZ 0.00071* 0.9374 −0.0366 0.0038 −0.0013*−0.1010 0.0154 23.15 −0.0020**

−1.0384*** 0.0520

(0.070) (0.000) (0.613) (0.360) (0.057) (0.341) (0.907) (0.036) (0.000) (0.782)

Canada 0.00166*** 0.6212*** 0.2327*** 0.0003 −0.0014***−0.0548 0.1276* 8.68 −0.0031***

−0.6760***−0.1051

(0.000) (0.000) (0.002) (0.903) (0.000) (0.458) (0.062) (0.000) (0.000) (0.421)

Europe 0.00044** 0.9734*** 0.0724**−0.0035*

−0.0008***−0.0028 0.0976 32.09 −0.0013***

−0.9762*** 0.0252

(0.031) (0.000) (0.032) (0.072) (0.006) (0.947) (0.107) (0.009) (0.000) (0.738)

Lat in Amer ica 0.00065*** 0.8915*** 0.0147 0.0001 −0.0009**−0.0251 0.1498* 30.67 −0.0016***

−0.9165*** 0.1352

(0.008) (0.000) (0.772) (0.967) (0.023) (0.678) (0.053) (0.009) (0.000) (0.270)

Middle East/ NA 0.001 23 0.8 12 5*** 0.0058 0.0055 −0.0009 0.1016 0.0644 27.43 −0.0021 −0.7109** 0.0586

(0.260) (0.001) (0.938) (0.709) (0.514) (0.479) (0.446) (0.378) (0.023) (0.668)

South Africa 0.00032 0.8054** 0.1187 0.0095 −0.0023 −0.4326 −0.3 45 0 1 4. 51 −0.0026 −1.2380**−0.4637

(0.191) (0.012) (0.673) (0.279) (0.195) (0.098) (0.392) (0.154) (0.030) (0.495)

UK 0.00067*** 0.8616*** 0.0432 0.0051**−0.0007** 0.0770 0.0757 22.04 −0.0013***

−0.7846*** 0.0325

(0.001) (0.000) (0.128) (0.011) (0.039) (0.161) (0.105) (0.004) (0.000) (0.607)

KW -2 18.01** 44.51*** 13.32* 10.95 6.63 6.78 4.71 10.66 8.01 24.80***

(0.012) (0.000) (0.065) (0.141) (0.469) (0.452) (0.695) (0.154) (0.332) (0.001)

By level

Capital raising 0.00106*** 0.7401*** 0.2601***−0.0008 −0.0016*** 0.0081 0.0887** 23.34 −0.0026***

−0.7319***−0.1714*

(0.000) (0.000) (0.000) (0.671) (0.000) (0.799) (0.036) (0.000) (0.000) (0.050)

No capital raising 0.00100*** 0.8718*** 0.0496 0.0011 −0.0010***−0.0346 0.0890** 19.86 −0.0020***

−0.9064*** 0.0394

(0.000) (0.000) (0.136) (0.468) (0.000) (0.402) (0.023) (0.000) (0.000) (0.555)

KW -2 0.20 3.18* 5.44** 0.32 3.16* 0.46 1.32 1.19 0.72 2.24

(0.658) (0.074) (0.020) (0.570) (0.075) (0.497) (0.250) (0.275) (0.398) (0.135)

By years

1976–1983 0.00071** 1.5647***−0.2484 0.0025 −0.0012***

−0.4221 0.3530 23.98 −0.0019***−1.9867** 0.6014

(0.011) (0.002) (0.432) (0.369) (0.005) (0.306) (0.289) (0.004) (0.025) (0.352)

1984–1990 0.00044** 0.9991*** 0.0523* 0.0018 −0.0006*** 0.0570 −0.016 5 3 6. 29 −0.0011***−0.9421***

−0.0689

(0.013) (0.000) (0.058) (0.356) (0.009) (0.235) (0.709) (0.004) (0.000) (0.280)

1991–1995 0.00122*** 0.8808*** 0.0112 −0.0004 −0.0009*** 0.0306 0.1763*** 22.70 −0.0021***−0.8502*** 0.1651

(0.000) (0.000) (0.806) (0.817) (0.007) (0.653) (0.006) (0.001) (0.000) (0.105)

1996–2000 0.00148*** 0.7626*** 0.0904** 0.0016 −0.0016***−0.0004 0.0803* 16.26 −0.0031***

−0.7631***−0.0101

(0.000) (0.000) (0.011) (0.497) (0.000) (0.991) (0.066) (0.000) (0.000) (0.883)

2001–2007 0.00023 0.6687*** 0.3604***−0.0021 −0.0007***

−0.0606 0.0187 19.63 −0.0009**−0.7292***

−0.34***

(0.241) (0.000) (0.000) (0.295) (0.006) (0.109) (0.708) (0.027) (0.000) (0.001)

KW -2 18.50*** 21.04*** 15.41*** 3.61 6.58 2.87 6.56 13.09** 11.74** 7.56

(0.001) (0.000) (0.004) (0.461) (0.160) (0.579) (0.161) (0.011) (0.019) (0.109)

***, **, and * represent significance at 1%, 5%, and 10% level, respectively.

(1999) who report a positive AR following cross-listing for firms

that issued capital compared to firms that did not issue.

Because the effects of cross-listing vary over time, we grouped

into five cross-listing period, which are 1976–1983, 1984–1990,

1991–1995, 1996–2000, and 2001–2007. The results as pre-

sented in Table 5 are consistent with those presented before.

Across all periods, the post-cross-listing AR and the change in AR

(˛) is negative and significant, supporting our previous argu-

ment on cross-listing and timing issue. The table also shows

that across all years, firms’ local market risk decreases after

cross-listing, and the highest decrease is for firms that have cross-

listed between 1976 and 1983 (ˇ =−1.9867), followed by the

periods 1984–1990 (ˇ =−0.9421), 1991–1995 (ˇ =−0.8502),

1996–2000(ˇ =−0.7631), and 2001–2007 (ˇ =−0.7292). These

results provide interesting evidence which has not been shown

in previous studies that document a decrease in local beta (e.g.

Foerster & Karolyi, 1999). This evidence suggests that the mag-

nitude of the decline in local market risk is decreasing overtime.

No change in global beta is documented except for firms that have

cross-listed between 2001 and 2007 where global beta decrease

7/28/2019 1-s2.0-S1062976909001094-main



by 0.34, which is inconsistent with results of Foerster and Karolyi

(1999).

As for firms listed on unregulated exchange, Table 6 reports dif-

ferent results. For example, the post-AR and the change in AR (˛)

is negative across all groups, but statistically significant for Asia,

Australia and New Zealand, and Latin America only. On the other

hand, except for firms from Australia and New Zealand, all other

foreign firms on unregulatedexchange experience a decline in their

local market risk after cross-listing with no change in their expo-

sure to the global market risk. The only exception is for UK firms

where global market risk declines by −0.1716 (significant at 10%

level). In addition, Table 6 shows that the magnitude of the decline

in AR and local market risk is smaller for unregulated exchange

compared to regulated exchange, suggesting the cross-listing on

regulated exchanges is associated with a premium.

As for firms that issue capital through unregulated exchanges,

the AR does not change after cross-listing, but declines by−0.0008

for firms that do not raise capital. This is consistent with Miller

(1999), who also reports negative but insignificant post-cross-

listing AR for firms that issue shares through PORTAL. Although

local market risk changes for both groups, the change is higher for

thelatter group (no capital raising),and globalmarket risk does not

seem tochange; none of theglobal marketriskis statistically signif-

icant. Dividing the sample by cross-listing periods, Table 6 reveals.

Table 6

IAPM market model for 839 firms, which have cross-listed on the US unregulated exchanges (OTC and PORTAL). Rit = ˛PRE i

+ˇPRE iL

RLmt +ˇPRE

iF RW

mt +˛LIST i

DLIST it

+˛POST i

DPOST it

+

ˇPOST iL

RLmt

DPOST it

+ˇPOST iW

RF mt

DPOST it

+ εit (t = −250,+250). The regression is run for each firm to obtain the regression parameters that are then averaged across groups. Rit and

RLmt are the daily returns for firm i and its weighted average local market index at time t , respectively. ˛i ’s are constants that are interpreted as abnormal returns. ˇiL and

ˇiW ’s are the local and foreign market risk respectively. RW mt is the world daily weighted average return index. DLIST

it is a dummy variable that takes the value one in the three

days around cross-listing and zero otherwise. DPOST it

is a dummy variable that equals one in the post-cross-listing period (+2, +250) and zero otherwise. is the difference

between thepost- andpre-cross-listingperiods(post–pre). Thecross-listingperiod(CL) refers to three daysaroundcross-listing(−1,0, +1). KW -2 is thetwo-tailed Wilcoxon

signed-rank test for the difference in the regression parameters across groups. Returns data are obtained from DataStream.

Category ˛PRE i

ˇPRE iL

ˇPRE iW

˛CLi

˛POST i

ˇPOST iL

ˇPOST iW

Adj. R2% ˛ ˇiL ˇiW

All firms 0.0003*** 0.8000*** 0.0486***−0.0014*

−0.0005***−0.0059 0.0041 21.17 −0.0008***

−0.8059***−0.0445

(0.001) (000) (0.003) (0.066) (0.000) (0.796) (0.841) (0.000) (0.000) (0.208)

By region

Asia 0.00039*** 0.8917***−0.0031 −0.0014 −0.0004***

−0.0447** 0.0225 23.06 −0.0008**−0.9365*** 0.0256

(0.000) (0.000) (0.841) (0.173) (0.000) (0.024) (0.288) (0.000) (0.000) (0.457)

Australia and NZ 0.0003 8 0.5454*** 0.0905 0.0009 −0.0017*** 0.2032 0.0159 8.83 −0.0021**−0.3422 −0.0746

(0.395) (0.004) (0.155) (0.997) (0.001) (0.287) (0.850) (0.023) (0.351) (0.588)

Europe −0.00002 0.7818*** 0.1382***−0.0016 −0.0004 −0.0275 −0.0706 23.09 0.0000 −0.8093***

−0.2088

(0.929) (0.000) (0.025) (0.410) (0.896) (0.349) (0.313) (0.979) (0.000) (0.103)

Latin America 0.00088** 0.6722*** 0.0116 0.0027 −0.0005 0.0882 0.1165 19.89 −0.0014*−0.5840*** 0.1049

(0.016) (0.000) (0.850) (0.243) (0.226) (0.172) (0.187) (0.067) (0.000) (0.457)

Middle East/N. Africa 0.00084 0.9118***−0.0065 −0.0080 −0.0008 −0.0488 0.2470 44.20 −0.0017 −0.9606*** 0.2535*

(0.208) (0.000) (0.954) (0.124) (0.338) (0.410) (0.028) (0.267) (0.000) (0.093)

South Africa 0.00043 0.7414*** 0.0750 −0.0021 −0.0006 0.0303 −0.0541 20.50 −0.0010 −0.7111*** −0.1291

(0.127) (0.000) (0.108) (0.414) (0.158) (0.683) (0.468) (0.116) (0.000) (0.242)

UK −0.0001 0.6 73 3*** 0.1279***−0.0038 −0.0002 −0.0287 −0.0437 15.97 −0.0001 −0.7020***

−0.1716*

(0.719) (0.000) (0.004) (0.186) (0.525) (0.705) (0.418) (0.824) (0.000) (0.061)

KW -2 9.74 24.06*** 13.08** 7.22 18.09*** 10.11 11.03* 11.14* 11.38 23.64***

(0.136) (0.001) (0.042) (0.301) (0.006) (0.120) (0.087) (0.084) (0.077) (0.001)

By level

Capital raising −0.00007 0.7910*** 0.2156 −0.0091 −0.0001 0.0098 −0.1657 23.42 −0.00002 −0.7812***−0.3813

(0.914) (0.000) (0.225) (0.176) (0.896) (0.857) (0.349) (0.986) (0.000) (0.277)

No capital raising 0.00033*** 0.8005*** 0.0384***−0.0009 −0.0005***

−0.0069 0.0146 21.03 −0.0008***−0.8074***

−0.0238

(0.000) (0.000) (0.006) (0.181) (0.000) (0.775) (0.443) (0.000) (0.000) (0.441)

KW -2 0.63 0.03 0.03 1.25 0.41 0.03 0.03 0.51 0.01 0.01

(0.428) (0.871) (0.868) (0.264) (0.520) (0.853) (0.856) (0.475) (0.915) (0.943)

By years

1976–1983 0.00021** 0.8986*** 0.0105 −0.0001 0.0001 −0.0562 −0.0002 17.72 −0.0001 −0.9547***−0.0107

(0.012) (0.000) (0.499) (0.941) (0.295) (0.177) (0.993) (0.718) (0.000) (0.770)

1984–1990 0.00046** 0.9574*** 0.0192 −0.0006 −0.0001 −0.0267 0.0766 33.05 −0.0005 −0.9841*** 0.0574

(0.010) (0.000) (0.646) (0.741) (0.761) (0.659) (0.246) (0.127) (0.000) (0.554)

1991–1995 0.00035*** 0.8566*** 0.0093 0.0003 −0.0005***−0.0688*** 0.0331 26.69 −0.0009***

−0.9254*** 0.0239

(0.001) (0.000) (0.684) (0.751) (0.000) (0.001) (0.257) (0.000) (0.000) (0.617)

1996–2000 0.00088*** 0.7756*** 0.0060 −0.0016 −0.0009*** 0.0750** 0.0401 22.26 −0.0018***−0.7006*** 0.0341

(0.000) (0.000) (0.821) (0.255) (0.000) (0.027) (0.273) (0.000) (0.000) (0.560)

2001–2007 −0.00009 0.7212*** 0.1251***−0.0031 −0.0004* 0.0057 −0.0489 16.72 −0.0003 −0.7155***

−0.1740**

(0.646) (0.000) (0.002) (0.105) (0.073) (0.922) (0.317) (0.428) (0.000) (0.046)

KW -2 6.58 8.73* 6.37 1.66 15.34*** 14.07*** 1.27 12.07** 2.16 13.81***

(0.160) (0.068) (0.173) (0.798) (0.004) (0.007) (0.866) (0.017) (0.707) (0.008)

***

,**

, and*

represent significance at 1%, 5%, and 10% level, respectively.

7/28/2019 1-s2.0-S1062976909001094-main



Table 7

IAPM market model for 1165 firms, which have cross-listed on the US unregulated exchanges. Rit = ˛PRE i

+ˇPRE iL

RLmt +

ˇPRE iW

RW mt +

˛LIST i

DLIST it

+˛POST i

DPOST it

+ ˇPOST iL

RLmt

DPOST it

+

ˇPOST iW

RW mt D

POST it

+ εit (t = −250,+250). The regression is run for each firm to obtain the regression parameters that are then averaged across groups. Rit and RLmt are the daily

returns for firm i and its weighted average local market index at time t , respectively. ˛i ’s are constants that are interpreted as abnormal returns. ˇiL and ˇiW ’s are the local

and foreign market risk, respectively. RW mt is the world daily weighted average return index. DLIST




post- andpre-cross-listingperiods(post–pre).The cross-listing period (CL)refersto three daysaround cross-listing(−1,0, +1).KW -2 is thetwo-tailed Wilcoxonsigned-rank


Exch ange Or igin ˛PRE

i ˇPRE

iL ˇPRE

iW ˛CL

i ˛POST

i ˇPOST

iL ˇPOST

iW Adj. R2

% ˛ ˇiL ˇiW

Regulated English 0.00133*** 0.7483*** 0.1439*** 0.0019 −0.0013***−0.0200 0.0873** 15.34% −0.0026***

−0.7683***−0.0566

(0.000) (0.000) (0.001) (0.258) (0.000) (0.659) (0.040) (0.000) (0.000) (0.469)

French 0.00035* 0.8853*** 0.0094 −0.0016 −0.0006**−0.0050 0.1227** 31.78% −0.0010**

−0.8903*** 0.1133

(0.064) (0.000) (0.775) (0.496) (0.030) (0.905) (0.020) (0.029) (0.000) (0.161)

Germany 0.0005 5** 1.2090*** 0.0505 −0.0017 −0.0013***−0.074 9 0.071 7 3 0. 88 % −0.0018***

−1.2839*** 0.0212

(0.022) (0.000) (0.177) (0.441) (0.001) (0.123) (0.283) (0.002) (0.000) (0.810)

Scandinavian 0.00072*** 0.7902*** 0.1250 −0.0040 −0.0008* 0.0658 −0.0593 31.50% −0.0015**−0.7244***

−0.1844

(0.003) (0.000) (0.130) (0.258) (0.051) (0.394) (0.462) (0.014) (0.000) (0.234)

KW -2 11.81** 30.78*** 5.99 7.68 1.42 2.42 4.88 4.87 19.19 4.15

(0.019) (0.000) (0.200) (0.104) (0.841) (0.660) (0.300) (0.301) (0.001) (0.387)

Unr eg ulat ed E nglis h 0.00033** 0.7590*** 0.0434**−0.0011 −0.0008*** 0.0042 −0.0039 18.74% −0.0011***

−0.7548***−0.0474

(0.016) (0.000) (0.031) (0.253) (0.000) (0.921) (0.879) (0.000) (0.000) (0.271)

French 0.00046** 0.7667*** 0.0549 −0.0006 −0.0003 0.0329 0.0577 25.27% −0.0008* −0.7338*** 0.0028(0.032) (0.000) (0.219) (0.788) (0.218) (0.313) (0.417) (0.082) (0.000) (0.980)

Germany 0.00019 0.9206*** 0.0193 −0.0029 −0.0001 −0.053 3 0.0107 2 2. 97 % −0.0003 −0.9739***−0.0087

(0.112) (0.000) (0.218) (0.063) (0.294) (0.042) (0.625) (0.147) (0.000) (0.802)

Scandinavian 0.00006 0.8779***−0.0045 −0.0068 0.0002 −0.0711 0.0846 20.73% 0.0001 −0.9490*** 0.0891

(0.915) (0.000) (0.954) (0.035) (0.640) (0.520) (0.300) (0.863) (0.001) (0.540)

KW -2 2.32 21.38*** 0.38 6.56 15.33*** 4.74 6.81 9.09* 14.97 2.74

(0.678) (0.000) (0.984) (0.161) (0.004) (0.315) (0.146) (0.059) (0.005) (0.603)

That post ARs for firms that cross-listed between, 1991–1995, and

1996–2000, 2001–2007 are negative and statistically significant.

When testing the change in betas, Table 6 shows that irrespective

of when the firm cross-lists, local market risk declines for all firms

in the sample, and the change in global market risk is negative and

significant for firms that have cross-listed between 2001 and 2007

only. This is actually similar to the results of regulated exchanges’

cross-listed firms,suggesting thatin recentyears,firms can manage

to reduce global risk through cross-listing.

4.4. Results of testing investor protection hypotheses

The results reported in Table 7 show that for most groups in

both regulated and unregulated exchanges, the AR falls signifi-

cantly after cross-listing. This statistically significant decline in the

post-listing AR is as low as −0.0003 for German origin firms cross-

listed on OTC/PORTAL, and as high as −0.0026 for English origin

firms cross-listed on regulated exchanges.12 The evidence is not

supportive to H3 nor is supportive to H4. However, it can be notedfrom Table 7 that for both groups, the decrease in AR following

cross-listing is higher for firms from common law countries, i.e.

English origin, compared to firms from civil law countries. More

importantly, the higher the pre-cross-listing AR, as in the case for

English origin firms listed on regulated exchanges (AR= +0.00133),

the higher the decline in AR after cross-listing (−0.0026), suggest-

ing again both overvaluation and timing issues.

In relation to reducing segmentation, Table 7 reveals that all

firms from common and civil law countries experience a decrease

12 The classification of countries between civil and common-law is based on La

Porta et al. (1997, 1998), where there are five groups of legal frameworks: English

law origin, French law origin, German law origin, Scandinavian law origin.

in their local market risk, beta, after cross-listing with no change

in global market risk, supporting H1.

Because investor protection is only associated with regulated

exchanges, unregulated exchanges should provide a benchmark

in this context. Table 8 shows that in all regressions and for both

regulated and unregulated exchanges, the coefficients on investor

protection measures are statistically insignificant, implying no

relation between cross-listing and signalling the protection of

minority shareholders. This is inconsistent with the third hypoth-

esis on cross-listing and investor protection, H5. All in all, the

evidence above suggests that signalling the firm’s commitment to

protect the interest of minority shareholders through cross-listing

in the US has no effect on share price.

4.5. Evidence on timing issues, testing H 6 and H 7

Linking the preceding results with studies on IPOs (e.g. Ritter,

1991) suggests a timing issue. Ritter (1991) finds significantdecline

in the monthly average AR (CAR) from 0.38% in month +1 (onemonth afterequityoffering)to 1.67% (−29.13%)bytheendofmonth

+36. Furthermore, in comparing firms that offer IPOs and a sample

of matching firms, he finds that the post-floating performance of

IPO firms is lower than the matching group. This post-floatation

underperformance, according to Ritter, supports the argument that

managers lower the cost of issuing equity by going public during

theperiod when their companies are overvalued by the market. He

argues, based on evidence by Ritter (1987) and Barry, Muscarella,

and Vetsuypens (1991), and Lee, Shleifer, and Thaler (1991), that

thehigh transactions costs of raising external finance will be partly

offset by the low realized long-run returns during the period of

overvaluation.

Following these arguments, the evidence of this research backs

the notion that managers may cross-list during periods when

7/28/2019 1-s2.0-S1062976909001094-main



Table 8

˛POST i

= o +1Tobins Q i,,PRE +2IPM +3 i +4i ∗ Exc ha ng e D ummy +5LNVOi,PRE ++6ROAi,PRE +7LNMV i,PRE + EDM i+ε (Eq. (5))

˛POST i

is the post-cross-listing AR from eq. (4). LNMV i,PRE is the natural logarithm of the pre-cross-listing market value. LNVOi denotes the natural

logarithm of the trading volume. EDM i is a dummy variable that equals one if the firm is from a developed country and zero otherwise. IPM refers to three difference

measures of investor protection, which are accounting standards, anti-director rights, and rule of law. ROA is the pre-cross-listing return on assets. is calculated as:

= ( 2ε SIZE ).(1/SHRt +1 − 1/SHRt ). 2ε is IAPM residual variance for the pre-cross-listing period (−250, −2), Size is the market capitalization of the cross-listed firm, and

SHR is its number of outstanding shares. Tobin’s-q is calculated as: Tobin’s-q = (BVTA− BVE +MVE )/BVTA. BVTA, BE, and MVE stand for the book value of total assets, the

book value of equity, and the market value of equity, respectively. The p-values of the T -test statistics are presented in parentheses.

Variable Regulated (NASDAQ and NYSE) Unregulated (OTC and PORTAL)Accountingstandards Anti-directorrights Rule of law Accountingstandards Anti-directorrights Rule of law

Interc ept 0.00142 (0.575) −0.00075 (0.337 ) 0.00111 (0.614) 0.00125 (0.377) −0.00004 (0.942) 0.00083 (0.475)

pre-LNMV 0.00001 (0.281) 0.00001 (0.231) 0.00001 (0.233) 0.00000 (0.127) 0.00001 (0.153) 0.00001 (0.117)

pre-TQ 0.00012***(0.007) 0.00012***(0.008) 0.00013***(0.005) −0.00019**(0.019) −0.00019**(0.016) −0.00019**(0.017)

0.00013 (0.592) 0.00012 (0.619 ) 0.00013 (0.598) −0.00002 (0.745) −0.00003 (0.679) −0.00003 (0.701)

(NYSE) −0.00007 (0.779) −0.00007 (0.792) −0.00008 (0.765) 0.00008 (0.501) 0.00008 (0.487) 0.00009 (0.43 1)

pre-ROA 0.00002***(0.000) 0.00002***(0.000) 0.00002***(0.000) 0.00005***(0.002) 0.00005***(0.002) 0.00005***(0.003)

pre-LNVO 0.00002 (0.788) 0.00001 (0.928) −0.00001 (0.943) −0.00002 (0.661) −0.00002 (0.679) −0.00003 (0.619)

EDM −0.00061 (0.431) −0.00008 (0.871) −0.00075 (0.408) −0.00026 (0.475) −0.00011 (0.742) −0.00049 (0.347)

IPM −0.00003 (0.372) −0.00001 (0.939) −0.00019 (0.375) −0.00002 (0.243) −0.00009 (0.400) −0.00013 (0.287)

Adj. R2% 16.52 16.16 16.52 3.03 2.88 2.98

F 5.70 (0.000) 5.58 (0.000) 5.70 (0.000) 2.60 (0.009) 2.51 (0.011) 2.57 (0.010)

NOBS 191 191 191 410 410 410

*** , **, * denote significance at 1%, 5%, and 10% level, respectively.

investors are overoptimistic about the future growth of firms and

cross-list to take advantage of the increase in the value of share

prices.13 The severedecline in thepost-listing AR is consistent with

H6 and reveals that the market corrects the pre-listing overval-

uation in share prices of cross-listed firms. In most of the cases

presented before, the higher the pre-cross-listing AR, the higher

the decline in AR following cross-listing, and in particular for firms

that issued shares, which is consistent with the argument of Ritter

(1991) discussed before.

Furthermore, supporting H7, Table8 shows thatthe pre-Tobin’s-

Q andROA arepositive andsignificant in allregressions, suggesting

that the higher the pre-cross-listing performance as measured by

ROA and the valuation as measured by Tobin’s-Q, the higher thedecline in the post-cross-listing AR (where it is negative for all

firms).This provides a direct indicationof overvaluation and timing

of cross-listing. As for the shadow costs, the coefficient is statis-

tically insignificant in all regressions, suggesting that the cost of

incomplete information does not seem to explain the variation in

ARacross firms.This isinconsistentwiththe results ofboth Foerster

and Karolyi (1999) who find it negative andstatistically significant,

and with Merton (1987).

4.6. Robustness checks

The previous results are robust to the use of one-factor market

model. We also check and find that the results are not driven by

the presence of outliers, by deleting observations that are below orabove 5 standard deviations. The results also do not change when

accounting for the foreign exchange rate (US dollars for US foreign

firms) or when using allreturnsin thelocalcurrency. Moreover,the

findings do not change when using pooled regression and are not

sensitive to the use of a short window (−100, +100) instead of long

window (−250, +25). Furthermore, the results remain the same

when controlling for infrequent trading in some stocks, which we

13 Over-optimism might be due to increasing the level of disclosure, or improving

firm’s performance or both. Nanda (1991) contends that equity carve-outs (a sub-

sample of IPOs) may be offered when the manager of the parent company learns

that subsidiary shares are overvalued. Such an overvaluation in Nanda’s point of

viewis consistent with Mikkelson,Partch,and Shah’s (1997)evidence showingthat

the performance of firms that go public declines following a carve-out.

account for using Dimson (1979) method. We also conductedevent

study by years and we grouped our sample into periods as follows:

1976–1983, 1984–1990, 1991–1995, 1996–2000, and 2001–2007.

The results are consistent with those presented before: post-AR is

negative and significant. Similar results are found when grouping

firms by industry, based on DataStream level 3 industry (consumer

goods, Financials, Industrial, Information Technology, Resources,

Services, and Utilities). We also run regressions to explain the

change in local and global beta and find that firms from good

accounting standards environment have a higher decline in local

beta after cross-listing, and that both Tobin’s-Q and trading vol-

ume explain the variation in the magnitude of the decrease in local

beta across firms. Moreover, we run Eq. (5) without the inclusionof Tobin’s-Q in order to test whether itsinclusion affects the signif-

icance of the size (MV) coefficient, butthe results remain the same,

the coefficient on MV is small and insignificant. For the post-AR

regression, we also used the post-ROA and post-cross-listing VO in

addition to the difference in ROA and VO instead of the pre-ROA

and pre-VO, in order to check whether the results are subject to

the inclusion of the pre-cross-listing control variables. Results are

in the same direction to those presented before and lead to the

same interpretation about the timing of cross-listing.

5. Conclusions

This paper re-examined the work of previous studies such as

Miller (1999) and Foerster and Karolyi (1999), among others. Themain result that we are reporting, and which constitutes a novel

addition to the existing evidence, is that we find strong evidence

to indicate that firms cross-list in a period of good performance to

take advantage of the overvaluation of share prices in their ‘local’

market. Abnormal return (AR) exhibits a significant decline after

cross-listing, and the higher the pre-cross-listing AR (as in the

case of firms with IPOs), the higher the decline in AR after cross-

listing. In addition, we find that the higher the pre-cross-listing

Tobin’s-Q, a measure of a firm’s valuation, and ROA, a measure of

performance, the higher the decline in AR following cross-listing.

This is consistent with the IPOs overvaluation hypothesis of Ritter

(1991). Moreover, we find that all firms in the sample experience

a decrease in their local betas, which is consistent with the find-

ings of Foerster and Karolyi (1999). However, the results of period

7/28/2019 1-s2.0-S1062976909001094-main



analysis show that the decrease in local beta is diminishing over

time. Also, we report no change in the global beta, except for the

period 2001–2007 where it decreases after cross-listing. Further-

more, we find that the decline in the post-cross-listing AR (beta) is

higher (lower) for firms that issuedcapital through cross-listing on

the US regulated exchanges compared to firms that did not issue,

which is inconsistent with Miller (1999) and Foerster and Karolyi

(1999).

As for investor protection, the evidence does not support the

hypothesis that cross-listing in the US regulated exchange will sig-

nal the firm’s commitment to protect minority investors and thus

increase the firm’s value by reducing the required rate of return.

Our results show a decline in the post-listing AR for both regu-

lated and unregulated exchanges. The decrease in AR is common

across firms from civil and common law countries, regardless of

the location of cross-listing (regulated or unregulated exchanges),

which adds further support to the lack of validity of the prediction.

Furthermore, the cross-sectional regressions show no statistical

association between thechangein thepost-listing AR andthe three

measures of investor protection (accounting standards index, anti-

director rights index, and rule of law index). Various robustness

checks ensure that these results are not affected by the pres-

ence of outliers, nor do they change when accounting for foreign

currency, short window analysis (−100, +100), or infrequent trad-

ing.

References

Alexander, G. J.,Eun,C. S.,& Janakiramanan, S. (1988). International listings of stockreturns: Someempiricalevidence. Journal of Financial and Quantitative Analysis,

23(2), 134–151.Barry, C., Muscarella, C. J., & Vetsuypens, M. R. (1991). Underwriter warrants,

underwriter compensation, and the costs of going public. Journal of FinancialEconomics, 29(1), 113–135.

Bradshaw,M.T., Bushee,B. J.,& Miller,G. S.(2004).Accountingchoice,homebias,andUS investment in non-US firms. Journal of Accounting research, 42(5), 795–841.

Coffee,J. (1999). Thefutureas history:The prospectsof global convergencein corpo-rategovernanceand its implications. Northwestern UniversityLaw Review, 93(3),

641–708.Coffee, J. (2002). Racing towards the top? The impact of cross-listings and stock

market competition on international corporate governance. Working paper no. 205. Columbia Law School, The Centre for Law and Economics.

Dimson,E. (1979). Riskmeasurementwhen shares aresubjectto infrequent trading. Journal of Financial Economics, 7 , 197–226.

Davis-Friday, P. Y., Frecka, T. J., & Rivera, J. M. (2005). The financial performance,capital constraints and financial reporting environment of cross-listed firms:Evidence from Mexico. The International Journal of Accounting , 40(1), 1–30.

Doidge, G, Karolyi, A., & Stulz, R. M. (2003). Why are foreign firms listed in the U.S.worth more? Journal of Financial Economics, 1–34.

Errunza,V. R.,& Losq,E. (1985). Internationalassetpricingundermildsegmentation:Theory and test. Journal of Finance, 40, 105–124.

Fama,E. F.,& French,K. R.(1992).The cross-sectionof expectedstockreturns. Journalof Finance, 67 (2), 427–465.

Foerster, S. R., & Karolyi, G. A. (1993). International listings of stocks: The case of Canada and the U.S. Journal of International Business Studies, (Fourth quarter).

Foerster, S. R., & Karolyi, G. A. (1998). Multimarket trading and liquidity: A trans-action data analysis of Canada-US interlistings. Journal of International FinancialMarkets, Institutions and Money, 8(3–4), 393–412.

Foerster, S. R., & Karolyi, G. A. (1999). The effects of market segmentation andinvestor recognition on asset prices: Evidence from foreign stocks listing in theUnited States. Journal of Finance, 54(3), 981–1013.

Howe, J., & Kelm, K. (1987). The stock price impacts of overseas listings. FinancialManagement , 51–56.

Howe, J.,& Madura,J. (1990). Theimpactof internationallistingson risk:Implicationfor capital market integration. Journal of Banking and Finance, 14, 133–142.

Howe, J., Madura, J., & Tucker, A. (1993). International listings and risk. Journal of International Money and Finance, 12, 99–110.

Jayaraman, N., Shastri, K., & Tandon, K. (1993). The i mpact of international cross-listings on risk and return. Journal of Banking and Finance, 17 , 91–103.

Karolyi, A. (2004). The world of cross-listings and the cross-listings of the world:Challenging conventional wisdom. Dice Center Working Paper No. 2004-14.

Kyle, A. (1985). Continuous auction and insider trading. Econometrica, 53,1315–1335.

La Porta, R., Lopes-de-Silanes,F., Shleifer,A., & Vishny, R. (1997). Legaldeterminantsof external finance. Journal of Finance, 52, 1131–1150.

La Porta, R., Lopes-de-Silanes, F., Shleifer, A., & Vishny, R. (1998). Law and finance. Journal of Political Economy, 106, 1113–1155.

Lau, S., Diltz, D., & Apilado, V. (1994). Valuation effects of international stockexchange listings. Journal of Banking and Finance, 18, 743–755.

Lee, I.(1991). Theimpact of overseaslistings onshareholderwealth:The case of theLondon and Toronto Stock Exchanges. Journal of Business and Accounting , 18(4),

583–592.Lee, C., Shleifer, A., & Thaler, R. (1991). Investor sentiment and the closed-end fundpuzzle. Journal of Finance, 46, 75–109.

Martell, T. F., Rodriguez, L., Jr., & Webb, G. P. (1999). The impact of listing LatinAmerican ADRs on therisks and returns of theunderlying shares. Global Finance

Journal, 10(2), 147–160.Merton,RobertC. (1987). A simple modelof capital market equilibrium withincom-

plete information. Journal of Finance, 42(3), 483–510.Mikkelson, W., Partch, M. M., & Shah, K. (1997). Ownership and operating perfor-

mance of companies that go public. Journal of Financial Economics, 44, 281–307.Miller, D. (1999). The market reaction to the international cross-listings: Evidence

from depositary receipts. Journal of Financial Economics, 51, 103–123.Nanda, V. (1991). On the good news in equity carve-outs. Journal of Finance, 46,

1717–1736.Pagano, M., Roell, A. A., & Zechner, J. (2002). The geography of equity listing: Why

do companies list abroad? Journal of Finance, 57.Ramchand, L., & Sethapakdi, P. (2000). Changes in systematic risk following global

equity issuance. Journal of Banking and Finance, 24, 1491–1513.Reese, W. A., & Weisbach,M. S. (2002). Protection of minority shareholder interests,

cross-listings in the United States, and subsequent equity offerings. Journal of Financial Economics, 66, 65–104.

Ritter, J. R. (1987). The costs of going public. Journal of Financial Economics, 19,269–281.

Ritter, J. R. (1991). The long-run performance of initial public offerings. Journal of Finance, 46(1), 3–27.

Schotman, P. C., & Zalewska, A. (2006). Non-synchronous trading and testing formarket integration in Central European Emerging Markets. Journal of EmpiricalFinance.

Stapleton, R. C., & Subrahmanyam, M. G. (1977). Market imperfections,capital mar-ket equilibrium and corporation finance. Journal of Finance, 32(2), 307–319.

Stulz,R. M.(1999,February).Globalizationof equitymarkets andthe cost of capital.NYSE working paper 99-02. www.nyse.com.

Torabzadeh, K., Bertin, W., & Zivney, T. (1992). Valuation effects of internationallistings. Global Finance Journal, 3, 159–170.

http://www.nyse.com/

http://www.nyse.com/

1-s2.0-S1062976909001094-main

Documents