The Efffect of Market Segmentaion On Invetsor

American Finance Association
The Effects of Market Segmentation and Investor Recognition on Asset Prices: Evidence
from Foreign Stocks Listing in the United States
Author(s): Stephen R. Foerster and G. Andrew Karolyi
Source: The Journal of Finance, Vol. 54, No. 3 (Jun., 1999), pp. 981-1013
Published by: Wiley for the American Finance Association
Stable URL: https://www.jstor.org/stable/222432
Accessed: 23-04-2020 13:57 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
American Finance Association, Wiley are collaborating with JSTOR to digitize, preserve and
extend access to The Journal of Finance
This content downloaded from 103.140.30.110 on Thu, 23 Apr 2020 13:57:18 UTC
All use subject to https://about.jstor.org/terms
THE JOURNAL OF FINANCE * VOL. LIV, NO. 3 * JUNE 1999
The Effects of Market Segmentation and

Investor Recognition on Asset Prices:
Evidence from Foreign Stocks
Listing in the United States
STEPHEN R. FOERSTER and G. ANDREW KAROLYI*
ABSTRACT
Non-U.S. firms cross-listing shares on U.S. exchanges as American Depositary

Receipts earn cumulative abnormal returns of 19 percent during the year before
listing, and an additional 1.20 percent during the listing week, but incur a loss of
14 percent during the year following listing. We show how these unusual share
price changes are robust to changing market risk exposures and are related to an
expansion of the shareholder base and to the amount of capital raised at the time
of listing. Our tests provide support for the market segmentation hypothesis and
Merton's (1987) investor recognition hypothesis.
THE GLOBALIZATION OF U.S. CAPITAL MARKETS has accelerated dramatically in the

past decade. Increasing numbers of companies from overseas have chosen to
either raise capital through global equity issues or prepare for future capital
raising by way of cross-listings on U.S. exchanges. As of 1997, about 1,300
non-U.S. companies have listed their shares for trading on the New York
Stock Exchange (NYSE), the American Exchange (AMEX), the National As-
sociation of Securities Dealers' Automation Quotation (Nasdaq) system, or
*Foerster is with the Richard Ivey School of Business, University of Western Ontario, and
Karolyi is with the Fisher College of Business at Ohio State University. We are grateful for data
assistance from John Griffin, Rick Johnston, and Sonali Chalishazar; for background informa-
tion from Jim Shapiro (NYSE), Mike Shokhouhi (NASD), Vince Fitzpatrick and Joe Velli (Bank
of New York), Mark Bach (Citibank), and Rene Vanguestaine (JP Morgan); and for comments
from Yakov Amihud, John McConnell, Darius Miller, Ren6 Stulz (editor), and an anonymous
referee. Comments of conference participants at the 1997 NBER Market Microstructure Con-
ference, the 1997 AFA, 1997 FMA International, 1997 Berkeley Program in Finance, the 1996
Vanderbilt Conference on Investing Internationally, and the 1996 Northern Finance Association
meetings, and workshops at HKUST, Laval University, Queen's University, the University of
Toronto, and the University of Waterloo greatly improved the paper. Nelson Mark kindly pro-
vided access to the Harris Bank interest rate data. We thank the Social Sciences and Human-
ities Research Council of Canada and the Richard Ivey School of Business Plan for Excellence
for financial support, as well as Ohio State University's Dice Center and Summer Fellowship
program. All remaining errors are our own.
981
982 The Journal of Finance
over-the-counter (OTC), which represents a 75 percent increase since 1991.

In 1996 alone, 162 new programs were created, raising $20 billion in new
equity financing and generating exchange trading of over 7 billion shares.'
The goal of this paper is to study the stock price performance and changes
in risk exposure associated with the cross-listing of non-U.S. stocks in U.S.
markets. Our sample comprises first-time U.S. listings by 153 firms from
Canada, Europe, and the Asia-Pacific Basin region from 1976 to 1992. We
are motivated to study this phenomenon because important inferences per-
taining to the issue of capital market integration and segmentation can be
drawn from the reaction of stock prices to international listings.2 Segmen-
tation of markets due to investment barriers (e.g., regulatory barriers, taxes,
information constraints) creates an incentive for firms to adopt financial
policies to reduce their negative effects. Theory suggests that stock prices
for firms that cross-list from segmented markets are expected to rise and
their subsequent expected returns should fall as an additional built-in risk
premium compensating for these barriers dissipates. Our overall evidence is
consistent with this hypothesis. We also explore how these results extend
earlier studies of cross-border listing in the United States, such as those by
Alexander, Eun, and Janakiramanan (1988), Jayaraman, Shastri, and Tan-
don (1993), and Foerster and Karolyi (1993).3
We are also drawn to this question by a second branch of the finance
literature that identifies significant changes in share prices for firms that
choose to change the location for their traded shares. A number of studies
have shown how share prices increase for firms that list on the NYSE from
the Nasdaq OTC market and have attributed this outcome to increased in-
vestor recognition or superior liquidity.4 Merton (1987), for example, pro-
vides a rationale for the effects of greater investor recognition in an extension
to the Sharpe-Lintner Capital Asset Pricing Model (CAPM) which relaxes
the assumption of equal information for investors. He shows that expected
1 See "The Rise of ADRs," Fortune (March 6, 1995); "Four-year Surge in ADR and GDR
Issues," Financial Times (November 10, 1994); "The Return of ADRs," Euromoney (December
1995); and Cochrane, Shapiro, and Tobin (1996).
2 Survey papers by Adler and Dumas (1983) and more recently Stulz (1995) argue that an
understanding of the extent of international capital market segmentation is a key challenge for
research in international finance. Important studies with evidence of growing integration of
financial markets include Jorion and Schwartz (1986), Gultekin, Gultekin, and Penati (1989),
Campbell and Hamao (1992), Mittoo (1992), Chan, Karolyi, and Stulz (1992), Bailey and Chung
(1995), and Bekaert and Harvey (1995).
3 Karolyi (1998) surveys the literature on global exchange listings. He cites a number of
studies that have examined stock price reactions for U.S. firms listing their shares abroad,
including Howe and Kelm (1987), Howe and Madura (1990), Barclay, Litzenberger, and Warner
(1990), Varela and Lee (1993), and Lau, Diltz, and Apilado (1994).
4 Baker and Meeks (1991) and McConnell et al. (1996) survey the literature on domestic
exchange listings and delistings. Important contributions include those of Sanger and McCon-
nell (1986) and Dharan and Ikenberry (1995), who focus on the postlisting decline in stock
prices following NYSE and AMEX listing, and that of Christie and Huang (1993), who examine
the liquidity effects of Nasdaq and AMEX listings to the NYSE using transactions data.
Effects of Market Segmentation and Investor Recognition 983
returns decrease with the size of the firm's investor base, which he charac-
terizes as "the degree of investor recognition." Amihud and Mendelson (1986)
develop the liquidity hypothesis in the context of an asset pricing model in
which gross returns are an increasing and concave function of liquidity mea-
sured by the bid-ask spread. Kadlec and McConnell (1994) show that these
two hypotheses can in part explain the abnormal returns to NYSE listings
from Nasdaq.
We propose that U.S. exchange listing by non-U.S. firms could also be
associated with share price changes that are not due to the effects of inter-
national investment barriers but rather to investor recognition and liquidity
factors, as experienced by purely domestic exchange listings. We test this
hypothesis using information available on changes in shareholder base and
capitalization changes due to new issues of equity around the listing period
for these firms and find that the abnormal returns before, around, and fol-
lowing listing are significantly related to these variables. We interpret this
finding as evidence consistent with Merton's investor recognition hypoth-
esis. Finally, we indirectly test Amihud and Mendelson's (1986) liquidity hy-
pothesis and show that the sensitivity of the abnormal returns, as well as
changing risk exposures, to changes in shareholder base is different for non-
U.S. stocks listing on the NYSE versus those listing on the AMEX and Nas-
daq. The finding that stock returns around cross-border listings is related to
changes in the investor base and liquidity factors is new.
A primer on the cross-border listing process is presented in Section I. Sec-
tion II provides a description of our methodology and data. Outlines of the
various hypotheses about stock price effects on international listings and the
main empirical results are presented in Sections III and IV. Conclusions
follow in Section V.
I. A Primer on ADRs
Almost all non-U.S. companies that list their shares on U.S. exchanges do
so by creating American Depositary Receipts (ADRs). ADRs were developed
by JP Morgan in 1927 as a vehicle for investors to register and earn divi-
dends on non-U.S. stock without direct access to the overseas market itself.
U.S. depositary banks hold the overseas securities in custody in the country
of origin and convert all dividends and other payments into U.S. dollars to
receipt holders in the United States. Investors, therefore, bear all currency
risk and indirectly pay fees to the depositary bank. Each depositary receipt
denotes shares that represent a specific number of underlying shares in the
home market, and new receipts can be created by the bank for investors
when the requisite number of shares are deposited in their custodial account
in the home market. Cancellations or redemptions of ADRs simply reverse
the process.
There are a number of advantages to ADRs for issuers, including an en-
larged investor base, enhanced local market for shares, opportunity to raise
new capital, and a liquid secondary market in the United States. At the
same time, non-U.S. companies must satisfy two requirements to be listed in

the United States. First, they have to arrange with a transfer agent and
registrar for an exact replication of settlement facilities as for domestic se-
curities. Second, to register with the U.S. Securities and Exchange Commis-
sion (SEC), the non-U.S. company must file a registration statement and
furnish an annual report on a Form 20-F with a reconciliation of financial
accounts with U.S. Generally Accepted Accounting Principles (GAAP). Sev-
eral options are available to issuers to balance these advantages with the
costs associated with increased scrutiny by the SEC and reconciliation with
GAAP reporting requirements. ADRs must now be sponsored by a non-U.S.
company seeking access to U.S. markets, but some pre-1983 programs were
initially unsponsored, as an initiative by a U.S. securities broker with the
depositary bank. The ADR issues can also be associated with new capital
raised through the program, though usually this is not the case. Table I
outlines the different options with listing, reporting, and GAAP require-
ments. Level I ADRs trade over-the-counter as Pink Sheet issues with lim-
ited liquidity and they require only minimal SEC disclosure and no GAAP
compliance. These firms are exempt from SEC filing Form 20-F under Rule
12g3-2(b) allowing home country accounting statements with adequate En-
glish translation, if necessary. Level II ADRs are exchange-listed securities,
but without a capital-raising element. Level III ADRs, the most prestigious
and costly type of listing, require full SEC disclosure with Form 20-F and
compliance with the exchange's own listing rules. Finally, Rule 144A, known
as RADRs, are capital-raising issues in which the securities are privately
placed to qualified institutional buyers (QIBs) and, as a result, do not re-
quire compliance with GAAP or SEC disclosure rules. These securities trade
over the counter among QIBs with very limited liquidity. As of June 1995,
Level I programs comprised 55 percent of new ADRs, 23 percent were pri-
vate placements (RADRs), and 22 percent were exchange-listed on the NYSE,
AMEX, or Nasdaq (Levels II and III).5
An alternative option is a direct "ordinary" listing. These ordinary listings
must take place on a U.S. exchange. They require an exact replication of
settlement facilities as for U.S. securities but have somewhat different GAAP
reporting and SEC registration requirements. With rare exceptions, Cana-
dian firms are the only firms that maintain ordinary listings.6 For example,
U.S. and Canadian companies file Form 10-K due within 90 days of their
fiscal year-ends and Form 10-Q quarterly, whereas non-Canadian overseas
issuers trading as ADRs file an annual Form 20-F within 180 days of fiscal
year-end which is less extensive (e.g., fewer footnotes on income taxes, leases,
5 More details are available from Bank of New York's Global Offerings of Depositary Receipts;
A Transaction Guide (1996) and Citibank's An Information Guide to Depositary Receipts (1995).
6 See Cochrane et al. (1996) for a discussion of registration and trading of foreign securities
in the United States including exemptions and key SEC accounting accommodations made.
Biddle and Saudagaran (1992) and Frost and Kinney (1996) study disclosure choices among
foreign registrants in the United States.
oferings
TableI
mationGudeDpsryRcbCk'Sv(195).
DepositaryRcPgmbT
NoCapitlRsngwheIu
tradingoNsqmkeuPORTALxch
requidfonacls,LvI
reconilatfsqudR4A,w usedonlyfrbqti,aLvI
requidRl12g3-(b)fanyshotmF4A,w Form-6isue,aLvlI StaemnilpubcofrgR4A,w
TradingloctOCPkSheNYE,AMXsqU.pvm-
buyers(QIB)maktofi
ItemLvl-Ru14A(D)GobaOfring
GAPrequimntNoOlypaFcv,s U.SrepotingExmudF20-flay;Pvc,s SECregistaonRFm-1d6fNPvplc,
U.Sexchangmjortqulifdsk;b DescriptonUldLmajOfPv.Sw theSEC(curisAof193),andpgqmx4.MlvbI- FourdifentlvsAmcaDpyRgbwh,q
pensions, industry, and geographic segment information) and can be based

on home country accounting practices. Since 1991, Canadian companies have
been permitted to meet U.S. annual SEC reporting requirements with Ca-
nadian disclosure documents (under the auspices of the Ontario and Quebec
Securities Commissions) by way of the Multi-Jurisdictional Disclosure System.7
II. Data
The initial sample of new listings in the United States includes all of the
317 exchange-listing Level II and III ADR applications that were successful
for the period from 1976 to 1992. The sample and listing dates were ob-
tained directly from the NYSE, AMEX, and Nasdaq Economic Research de-
partments and were verified in publications such as the NYSE Fact Book,
the AMEX Fact Book, Moody's International Manuals, and Standard and
Poor's Stock Reports. To be included in the sample, the ADR program has to
be the first U.S. listing for the firm (i.e., not a transition from RADR private
placement to exchange-listed ADR) and has to have weekly (Friday closing)
home-market stock price, exchange, and stock index data available.8 Our
primary data source for prices and exchange rates was Reuter's Exshare
International Securities Database, accessed using ReuterlinkTM, its on-line
data service. Returns are calculated as price changes without dividends. We
cross-checked stock prices using various English-language sources, includ-
ing The Japan Times, Asian Wall Street Journal, Financial Times, Wall Street
Journal, and Datastream International. Canadian stock price information
was obtained directly from the Toronto Stock Exchange/University of West-
ern Ontario database. Listings before 1976 are excluded because of Reuter's
Exshare database limitations.
The final sample consists of 153 listings from 11 countries in four regions
of the world, including Europe, Canada, Asia, and Australia. Appendix A
lists the firms by country with the associated listing date. The largest con-
tingent is comprised of the 67 Canadian firms, all of which are ordinary
listings, followed by the 36 U.K. issues, 26 from Europe (excluding the U.K.),
13 Australian issues, and 11 Asian listings (all but one from Japan). Table II
provides summary statistics for the sample organized by home region, list-
ing exchange, industry group, listing year, and type of ADR issue (i.e., Level
III capital-raising or Level II non-capital-raising). Several facts are notewor-
thy. First, the accelerating trend for listings is visible with more than 54
new listings in the four-year period 1989 to 1992 in contrast to 11 new list-
ings in 1976 to 1980 and 26 in 1981 to 1984. Most of the listings (82) occur
on Nasdaq and this is dominated by many smaller, resource-based Canadian
firms. Overall, all major sectors are represented. Cross-listed firms tend to
be very large with an average capitalization of $2.5 billion, although the dis-
7See Multi-Jurisdictional Disclosure and Modifications to the Current Registration and Re-
porting System for Canadian Issues, Securities Act Release No. 6902 (July 1, 1991).
8 Exchange rates are based on Friday, 10 a.m. midpoint quotes from Reuter's Exshare.
Table II
Descriptive Statistics for Global Firms Listing

on U.S. Exchanges, 1976-1992
Firms listing American Depositary Receipt programs as well as direct listings in the United
States with associated listing dates are obtained from the New York Stock Exchange Fact Book,
American Stock Exchange Fact Book, and American Stock Exchange and Nasdaq Economic
Research departments directly. New share issues are computed as a fraction of total shares
outstanding for 31 capital-raising Level-III ADR issues. Market capitalization values, informa-
tion about type and sponsor institution of ADR program, and size of new issue for Level-III
ADRs are obtained from Moody's International Manuals (various issues). Data on shareholder
base are obtained from Moody's Manuals and supplemented with Standard and Poor's Stock
Reports (various issues). Sample of 153 firms screened for data on home-market weekly stock
price, local stock index values, and New York Friday, 10 a.m. midpoint quotes for home-market
exchange rate for 52 weeks before and after the listing date, all obtained from Reuter's Exshare
International Securities Database using Reuterlink on-line data service and supplemented with
various local periodicals. Data availability for certain variables is indicated in parentheses.
Home Region Listing Exchange Industry Group
Canada 67 New York 60 Industrial 34

Australia 13 American 11 Resource 48
Europe ex U.K. 26 Nasdaq 82 Consumer 37
Asia 11 Financial 14
United Kingdom 36 Technology 12
Utilities 8
Listing Years Capital Raising Capitalization
1976-1980 11 Level-III ADR 31 Mean ($mills) 2,515.3

1981-1984 26 Non-capital raising 122 Maximum 53,221.0
1985-1988 62 Median 717.0
1989-1992 54 Minimum 3.5
New Share Issue Shareholder Base Percent Change in Base

(% of shares outstanding) (000s before listing) (before/after listing)
Mean 3.0% Mean 52.9 Mean 28.8%

Maximum 12.6% Maximum 740.3 Maximum 479.%
Median 2.1% Median 10.0 Median 11.1%
Minimum 0.3% Minimum 0.1 Minimum -86.7%
Number 31 Number 145 Number 145
tribution is positively skewed with a median of only $717 million. By studying

the descriptions in Moody's International Manuals in the year immediately fol-
lowing listing, we also document that 31 of the issues were capital raising.
To proxy for changes in the shareholder base of each firm, data on the
number of registered shareholders for each security are collected prior to the
listing announcement and subsequent to listing. These data are obtained
directly from Moody's International Manuals for the year immediately pre-
ceding and the year following the recorded listing date, as available. Missing
observations are supplemented by Standard and Poor's Stock Reports.
Table II shows that for 145 of the firms for which data are available, the
mean number of registered shareholders is 52,900, although the sample is
strongly positively skewed with a median of only 10,000. The maximum of

740,000 shareholders belongs to Telefonica de Espana. On average, the change
in shareholder base is positive at 28.8 percent with a median of 11.1 percent.
The extreme outliers at 479 percent (Northern Telecom, Canada) and -86.7
percent (Petromet Resources, Canada) arise for firms with very low initial
bases (fewer than 1,500 shareholders).
III. Market Segmentation Tests
A. Global Market Segmentation, Cross-Listings, and Share Value
Global cross-listings have interested the international finance literature

because these securities overcome many of the regulatory restrictions, costs,
and information problems that comprise barriers to investing in overseas
securities. To the extent that these barriers influence how securities are
priced in their respective markets, empirical researchers can evaluate the
degree to which international capital markets are segmented or integrated.
That is, in the context of an equilibrium model of expected returns, they ask
whether market risk is "priced" differently in the two capital markets (see
Black (1974), Stapleton and Subrahmanyam (1977), and Stulz (1981)). If
markets are segmented, firms have an incentive to adopt policies to mitigate
the negative effects of investment barriers and promote the positive effects
of international diversification by direct foreign investment, mergers with
non-U.S. firms, or dually listing their shares for trading on a non-U.S. cap-
ital market. Errunza and Losq (1985) and Alexander, Eun, and Janakira-
manan (1987) refine the equilibrium models above for pricing shares of firms
cross-listing abroad. These models predict that cross-listing shares between
two segmented markets leads to a higher equilibrium market price for a
given stock and a lower expected return. For example, consider the Errunza
and Losq two-country model of "partial" segmentation in which investment
barriers are asymmetric: country l's investors can invest in country 2's se-
curities, but country 2's investors are prohibited from investing in country
l's securities. They show that country 2's (eligible) securities are priced as if
markets were completely integrated, but that country l's (ineligible) secu-
rities command a "super" risk premium. If a company from country 1 cross-lists
its shares in country 2, comparative statics show that the super risk premium
disappears, the share price increases, and the expected return decreases.
Alexander et al. (1988) are the first to have studied price reactions for 34
firms from six different countries that were listed on either the NYSE, AMEX,
or Nasdaq between 1962 and 1982. They demonstrate that the cumulative
abnormal returns (CARs) for their non-Canadian benchmark sample of firms
increase by an annualized 17 percent in the two years before listing and fall
by an annualized 33 percent over the three years following listing. The CARs
for the Canadian sample are considerably smaller in both periods, which
they interpret as evidence consistent with the market integration between
Canada and the United States. Foerster and Karolyi (1993) investigate a
larger sample of 52 Canadian firms during the 1980s and find a much more
dramatic reaction on the order of an annualized 21 percent run-up in the
100-day prelisting period and a 22 percent postlisting decline. They inter-
pret this finding as evidence of segmentation between Canadian and U.S.
markets, consistent with other findings by Booth and Johnston (1984),
Jorion and Schwartz (1986), and Mittoo (1992).
In summary, previous research suggests that global cross-listing of shares
can lead to a reduction in expected return on a security if the capital mar-
kets from which they originate are segmented completely or partially. If the
segmentation hypothesis is correct, we should observe several patterns re-
lated to non-U.S. firms listing in the United States. First, we predict abnor-
mal returns around interlisting should be positive. Second, abnormal returns
around interlisting should vary across stocks by home market in ways re-
lated to differences in degrees of market segmentation. That is, firms from
emerging markets are likely to experience larger abnormal returns than
firms from developed markets.
B. Cumulative Abnormal Returns around Announcement

versus Listing Dates
International asset pricing models suggest that when investors realize that
barriers to investments are to be removed, expected returns should decrease
as prices are bid up on the expectation of the removal of these barriers. Thus,
in order to properly examine market segmentation hypotheses, we should
examine price effects around interlisting announcements. A small number of
studies have examined announcement effects in an international listing con-
text. Lau at al. (1994) examine market reactions around the announcement
of U.S. firms listing on overseas markets. Miller (1998) examines market
reactions around announcements of international firms (primarily from Eu-
rope and emerging market countries) that interlisted on U.S. exchanges be-
tween 1985 and 1995, and Switzer (1997) examines announcements of
Canadian firms that interlisted over the 1985 to 1996 period.
Although announcement dates are theoretically more appropriate than list-
ing dates in order to test segmentation hypotheses, data collection presents
some challenges, particularly for our sample of interlistings dating back to
1976. For example, one challenge relates to data sources. The most common
data source for announcements is Lexis/Nexis, which includes hundreds of
information sources. However, there are few relevant business data sources
that precede 1980. For example, one of the main sources of information,
Reuters Financial Service, is only available since January 1987. Another
challenge relates to the determination of the announcement date, even if
Lexis/Nexis accurately captures what is known in the market. For example,
for some firms, markets have expected for years that a firm will eventually
list in the United States; in some cases, a company spokesperson indicates
that a firm is contemplating interlisting but the firm has not received board
approval; and in other cases a firm has received board approval but has not
received approval from the relevant exchange or securities commission. It is

possible to have firms announce an intent to interlist without actually in-
terlisting. For example, Grand Metropolitan announced, on November 1, 1989,
that it was applying to list on the NYSE; on June 2, 1990, it announced a
postponement in the listing; then on February 2, 1991, it announced again
that it was seeking a NYSE listing-which finally occurred on March 13,
1991. All of these factors suggest that announcement dates may provide a
noisy signal to the market.
Nonetheless, we gather as much data as possible related to announcement
effects. We use Lexis/Nexis to obtain announcement dates for the 153 firms
in our sample. Generally, the announcement date represents the earliest
press release related to the eventual interlisting. We are able to identify 45
announcements,9 the earliest of which is in 1982. Our sample of firms with
announcements includes firms from Australia (6 firms), Canada (7), France
(1), Italy (2), Japan (3), Netherlands (3), Norway (1), Spain (1), and the U.K.
(21). Of the 45 firms, 32 listed on the NYSE, three on AMEX, and 10 on
Nasdaq. For our sample, the mean (median) difference between the announce-
ment date and the listing date is 70 (44) days.10 The median across ex-
changes is 40 days for firms listing on the NYSE, 60 days for AMEX, and 65
for Nasdaq. Only nine announcement dates are more than 100 days prior to
the listing date.
We compare results for both announcement and listing dates. In order to
measure abnormal returns, we first estimate the (local) market model a and
,8 for each firm (i.e., each firm's return relative to a local stock market index
from Datastream International estimated during the 150-day prelisting pe-
riod from day -250 to day -101). Abnormal returns are then calculated
from days -100 to +250 as
it = R it- [ai + fiLRrLt], (1)
where Rit is firm i's local currency return on day 0'1

Cumulative abnormal returns from -100 days to +250 around both an-
nouncement and listing dates are presented in Figure 1, and statistical re-
sults are presented in Table III. Panel A of Table III examines announcement
effects. There is strong evidence of a preannouncement run-up in prices. For
example, in the preannouncement period between days -100 and -2, aver-
age daily abnormal returns are 0.11 percent and significant (t-statistic of
4.19). Around the announcement period (days -1 and 0), average daily ab-
9 Details of dates along with the rationale cited for listing are available in an Appendix from
the authors.
10 Citibank (1995) estimates a 9-week horizon for a Level I ADR between establishing a
program launch (U.S. counsel, depositary bank) and the start of Pink Sheet trading, a 14-week
period for Level II or III ADRs. The RADR programs require only a 7-week period.
11 We also examine tests using a two-factor IAPM and the Schipper and Thompson (1983)
methodology, which we employ below. Results are quantitatively and qualitatively similar. We
present the results above because they are most closely related to the methodology in other
"announcement effect" studies.
18%
Announcement
| Listing X
16% - -
14%-
12% - -
10%
8%
6%
4%
2%-
0% -
-2%-
Days Relative to Announcement or Listing Date
Figure 1. Cumulative abnormal returns around announcement/listing in the United

States. Abnormal returns are computed for each firm based on market model risk adjustments
using a local market index. Estimates are computed over days -250 to -101 relative to the
identified announcement date for the U.S. listing. Daily abnormal returns are averaged across
firms and cumulated. The sample includes 45 firms with identifiable announcement dates and
stock and index prices are obtained from Datastream International.
normal returns jump to 0.21 percent. Although 53 percent of the firms ex-
perience positive abnormal returns around the announcement on day -1
(and 58 percent on day 0), there is a fairly large amount of variability. Con-
sequently, the announcement effect results are not significant in our sample
(t-statistic of 0.85). Subsequent to the announcement, average abnormal re-
turns are not significantly different from zero.
It is possible that, because this analysis is based on event time, we may
not capture time variation effects. For example, if markets have become
more integrated over time, we might witness a decrease across time in ab-
normal returns around the announcement date. There is a need to develop
new time-varying event methodology, but this is beyond the scope of the
current paper.12
12 There are other methodological issues as well. For example, event complexities including
firms that signal intent to cross-list but do not actually cross-list could be incorporated in the
event methodology (see Prabhala (1997)).
Table III
Abnormal Returns for 45 Firms around Announcement

and Listing on U.S. Exchanges, 1981-1992
The market model is estimated for each firm relative to a local market index (Appendix B)
during the prelisting period from day -250 to day - 101. Abnormal returns are then computed
for each firm based on market model risk adjustments. Daily abnormal returns are averaged
across firms and cumulated. The sample includes 45 firms with identifiable announcement
dates.
Average Daily
Event Period (days) Abnormal Return (%) t-Statistic
Panel A: Announcements
(-100, -2) 0.1082 4.19**

(-100, -50) 0.1065 3.07**
(-49, -10) 0.1265 3.02**
(-9, -2) 0.0278 0.28
(-1, 0) 0.2057 0.85
(1, 10) 0.0388 0.55
(11, 50) 0.0291 0.63
(51, 100) -0.0041 -0.11
(1, 100) 0.0135 0.49
(1, 250) 0.0010 0.06
Panel B: Listings
(-100, -2) 0.0711 2.74**

(-100, -50) 0.0388 1.09
(-49, -10) 0.0950 2.28**
(-9, -2) 0.1568 1.83*
(-1, 0) 0.3494 1.96*
(1, 10) -0.0371 -0.41
(11, 50) 0.1169 2.44**
(51, 100) -0.0546 -1.32
(1, 100) 0.0157 0.53
(1, 250) -0.0157 -0.87
* indicate significa
Panel B of Table III examines listing effects. As in the preannouncement

results, there is strong evidence of a prelisting run-up in prices. For exam-
ple, in the prelisting period between days -100 and -2, average daily ab-
normal returns are 0.07 percent and significant (t-statistic of 2.74). However,
around the listing period (days -1 and 0), average daily abnormal returns
jump to 0.35 percent, almost twice as large as the announcement day effects.
As well, 69 percent of the firms experience positive abnormal returns around
the listing on day -1 (and 58 percent on day 0). Unlike the announcement
effect results, these listing effect results are statistically significant (t-
statistic of 1.96). Subsequent to the listing, average abnormal returns are
-0.02 percent between days +1 and +250 but not significantly different
from zero (t-statistic of -0.87).
Overall, results from Table III suggest there appears to be some informa-
tion surrounding both the announcement of interlistings and actual listings.
However, in addition to the recognition of the data collection challenges de-
scribed above, these short-term results should be interpreted cautiously for
other reasons. There may well be information dissemination (or leakage)
prior to the announcement, between the announcement and the listing, at
the listing date (with the removal of any remaining uncertainty about the
listing), or even in the postlisting period when the information is actually
disseminated to market participants. Thus, unlike the existing literature,
which examines short-window results around announcement dates, we focus
on the overall picture that emerges by examining longer periods around the
listing. Consequently, the remainder of this paper focuses on results based
on our larger sample of firms with listing dates using weekly returns for one
year prior to listing and one year subsequent to listing.
C. Pre- and Postlisting Returns Performance
Summary statistics for weekly returns of the 153 firms around the listing
dates are presented in Table IV. We report both mean excess returns-that
is, in excess of risk-free rates and the associated Newey and West (1987)
t-statistics in local currency and U.S. dollar-denominated terms for all firms
and separately by region. The returns are computed in excess of weekly one-
month Eurodollar quotes, which are obtained directly from Harris Bank's For-
eign Exchange Weekly.13 We compute average weekly returns before listing
(weeks - 52 to - 1), around listing (week 0), and after listing (weeks + 1 to + 52).
The general pattern of excess returns is similar to that in other studies of
interlistings, such as Alexander et al. (1988) and Foerster and Karolyi (1993).
The returns increase by 0.38 percent per week in local currency (0.44 per-
cent per week in dollars) during the period before listing, which is statisti-
cally significant. The cumulative return over the 12-month period corresponds
to about 22 percent (26 percent). The excess returns are significantly posi-
tive during the listing week period, with an average return of 1.20 percent
(1.24 percent) per week. After listing, the cumulative returns dissipate to a
significant extent with an average weekly postlisting decline of -0.27 per-
cent (-0.30 percent). This corresponds to an annualized cumulative return
of - 13 percent (- 15 percent), which yields a net cumulative excess return of
7.5 percent (8.7 percent) over the entire two-year period. Given the similar-
ity of our results using local currency and dollar returns, we only report
local currency returns in subsequent analyses.14
To explain this returns pattern around the listing decision, Alexander
et al. (1988), Foerster and Karolyi (1993), and many others propose the
market segmentation hypothesis. These studies posit that the listing week
13 We are grateful to Nelson Mark for providing us with these data.

14 In an earlier version, we report all results using both local currency and U.S. dollar-
denominated returns and find that the inferences from the two sets of results are qualitatively
similar.
Table IV
Summary Statistics for Weekly Excess Returns

of Global Listings around Listing Dates
Weekly (Friday close) returns for 153 listings are denominated in U.S. dollars and local cur-
rency terms using foreign exchange rates based on 10 a.m. (New York) midpoint quotes, all
obtained from Reuter's Exshare International Securities Database using Reuterlink on-line
data service and supplemented with various local periodicals. The returns are computed in
excess of weekly one-month Treasury bill yields obtained from Harris Bank quotes. Means and
standard deviations are computed separately across weeks (-52, -1) before listing, (0) during
listing, and (+ 1, +52) after listing.
Local Currency U.S. Dollar

No. of
Stocks Observations Mean (%) t-Statistic Mean (%) t-Statistic
Panel A: Before U.S. Listing (weeks -52 to -1)
All 7062 0.3815 4.65** 0.4409 4.96**

Australia 659 0.8746 3.26** 0.8118 3.03**
Canada 3033 0.4152 2.60** 0.5359 3.00**
Europe ex. U.K. 1103 0.3966 3.04** 0.3649 2.82**
Asia 525 0.1524 0.81 0.0917 0.49
United Kingdom 1742 0.1956 1.75* 0.2883 2.58**
Panel B: During U.S. Listing Week (week 0)
All 153 1.2026 1.65* 1.2404 1.71*

Australia 13 -0.1275 -0.08 -0.2071 -0.14
Canada 67 1.1463 0.75 1.2207 0.80
Europe ex. U.K. 26 2.5481 2.49** 2.4757 2.45**
Asia 11 0.7904 0.95 0.7325 0.89
United Kingdom 36 0.9419 1.18 1.0629 1.34
Panel C: After U.S. Listing (weeks +1 to +52)
All 7681 -0.2651 -2.92** -0.3035 -3.31**

Australia 674 -0.8594 -2.46** -0.9500 -2.72**
Canada 3368 -0.4764 -2.76** -0.5173 -2.99**
Europe ex. U.K. 1274 0.2051 1.63 0.1637 1.29
Asia 571 0.0026 0.01 -0.0649 -0.35
United Kingdom 1794 -0.0642 -0.48 -0.0670 -0.48
**,* indicate significance at the 5 and 10 perc
returns should be positive overall and greater for firms for which the domi-
cile market is more likely to be segmented from the U.S. market (e.g., emerg-
ing markets) and smaller for firms for which the domicile market is more
integrated with that of the United States (e.g., Canada). Unlike the early
Alexander et al. (1988) findings, we show in comparisons across different
regions that the stock price reactions for the Canadialn firms are at least as
dramatic as most of the others. For example, the Canadian firms on average
achieve an average prelisting stock price rise of 0.42 percent per week, equal
to that of the entire sample. This contrasts with the largest average weekly
rise for the Australian firms of 0.87 percent and with the insignificant run-
ups for Asian firms. The large positive excess returns during the listing
week for the Canadian firms is similar to that experienced by the U.K. firms
but lower than that for the European (excluding U.K.) firms. Finally, the
postlisting declines also exhibit considerable variation by region. The largest
declines are for the Australian firms (-0.86 percent per week) and Canadian
firms (-0.48 percent per week), yet for the Asian and European firms, the
postlisting returns are positive.
D. Risk Adjusted Returns
In order to isolate patterns in expected returns around interlistings, ex-

isting studies, such as those by Alexander et al. (1988), Howe and Kelm
(1987), Foerster and Karolyi (1993), and Lau et al. (1994), employ standard
event study methodology in which expected returns are derived from the
Sharpe-Lintner CAPM. This is a limiting approach for several reasons. First,
the expected returns are proportional to a market covariance risk (beta) and
the excess return on a benchmark portfolio, both defined relative to a do-
mestic stock market index. In an international context, it seems appropriate
for a firm listing in an overseas market to calibrate its exposure to overseas
market risks in addition to domestic risks.15 Second, these local and over-
seas market risks could be changing over time due to the interlisting itself
or due to other firm-specific factors (e.g., changes in capitalization, new eq-
uity issues for Level III ADRs) that occur around interlistings. Third, using
an International Asset Pricing Model (IAPM), such as that of Solnik (1974),
where covariance risks are defined relative only to the world market port-
folio, ignores local market risks that may impact prices of interlisting stocks
from markets that are not integrated. Our objective is then to specify a
returns generating model for these interlisting firms that captures both do-
mestic and global risks and their changes over time.
To generate abnormal excess returns around the interlisting, we estimate
a modified IAPM that captures both domestic and global market risk, where
the former is computed relative to a local market index (Appendix B lists
different indexes by local market) and the latter is computed relative to the
weekly excess return on the Datastream International World Index.16 We
draw on the methodology of Schipper and Thompson (1983) in which we pool
the cross-section and time series of returns to estimate our two-factor IAPM:
= PRE + PRE DL +pPRE DW LISTDLIST

I arO+ tSiL m+ iW DMt + .I t
+ aPOST DPOST + .PLOST RLtD PtOST +W8PwOSTRmwD POT + eit, (2)
15 Empirical studies by Jorion and Schwartz (1986), Chan et al. (1992), and Dumas and
Solnik (1995) focus on the pricing of domestic and foreign market risks as well as currency
risks for stocks. See the survey by Stulz (1995).
16 We chose the Datastream International World Index, denominated in U.S. dollars, because
other well-known global indices, such as Morgan Stanley's Capital International and Financial
Times Actuaries/Goldman Sachs indices do not extend back to 1976 when our sample begins.
where ari's are constants (which we interpret as abnormal excess returns);

'siL5 are the coefficients on the local market excess return, RLt; aiw's are t
coefficients on the global market index excess return, R w; DIST is a dummy
variable that equals one if observations are from the listing week (week 0)
and zero otherwise; and D/POST is a dummy variable that equals one if ob-
servations are from the postlisting period (weeks + 1 to + 52) and zero other-
wise. The returns are denominated in local currency and are defined in excess
of the weekly yield of the one-month Eurodollar rate for U.S. dollar returns
and in excess of weekly yields of one-month Euromarket rates for the Ca-
nadian dollar, U.K. sterling, German mark, Japanese yen, and French franc,
where appropriate.17 The advantage of this methodology is that we can mea-
sure the pre- and postlisting returns after adjusting for market covariance
risks in addition to just the event-period abnormal returns using conven-
tional event study methods. A distinct disadvantage is that by pooling cross-
sectionally across the firms, we average the beta risk measures. Furthermore,
the estimation and test periods are identical, which may not capture chang-
ing betas around the event period. However, in the next section, we compute
these results separately by firm for the univariate and cross-sectional re-
gressions and identify no major distortions or outliers that should generate
concern about the pooled estimation.18
Table V presents results for the IAPM model regressions. We report the
adjusted R2 and robust t-statistics that are computed using Newey and West
(1987) standard errors correcting for heteroskedasticity and serial correla-
tion (up to six weekly lags). The final column reports a robust Wald test
X2-statistic for the Chow test of structural break for the pre- and
coefficients on the local and global returns.
Overall, we find in the prelisting period that the average beta on the local
market excess return is close to one (1.03) and that of the global market beta
is much smaller (0.22) but still significantly different from zero. By contrast,
the postlisting local market beta drops considerably and significantly to 0.74
(a decrease of approximately 0.28) and the global market beta decreases
from 0.22 to 0.12, but this change is not significant. The p-value for the
Chow test for structural break indicates that the overall change is statisti-
cally significant. Our key finding here is that important changes in risk
exposure result for firms interlisting their shares in the U.S. market: Ex-
posure to the local market risk is diminished and exposure to global market
risk has not significantly changed.
17 Data were not available on Euromarket rates for each country from which samp
originate. For local currency excess returns, we make arbitrary assignments, such as Euromark
rates for Europe, except U.K. and French firms, Euroyen rates for all Asian and Australian
firms, and Euro-Canadian dollar rates for Canadian firms.
18 In an earlier version of the paper, we extend the model in two other ways: (a) to allow for
currency risk with a separate factor related to foreign exchange returns and (b) to allow market
risk factors (betas) to change with information variables related to macroeconomic conditions.
See Shanken (1990) and Ferson and Schadt (1996). The results do not change qualitatively.
p-valuebowthsim.
theon-mU.SEurdlayi),Rw:
*indcatesgfh510prlv,y.
TableV
usingWeklyLocaCrExRt
BeforU.SListng(wk-52,1)WA+
MarketodlRgsinfGbLuD
RPE+LtaISTDO3m76i-
1.64*8293-07 UnitedKgom0.16987-524%3 0.1798*6-25 Asia0.312945-87% 1.67*492-08
Europe(x.UK)0187964-%35
3.17*0592-46 Cand0.4916237-%8 2.59*7306-18 Australi0.5613278-%49 4.1*298503-6 Alfirms0.312596-84%7
asocitedfnrwhupPRElg,LISTO.Fx WestimaIAPMrkodlgnfxcu,(pB)
ai~PEOSjTR(-vle
PREISTOAdj2X(p-value)
tesofhrucalbking(w-521)d0+,px serialyuncotd(g=6)NwW1987pbhfm.AC dolartesfCnim,Eukh.Rb-cpgy wemployEurnatsfAi,k(xU.K)dg-C Themodlistawuyvrbnxg(DLISfk0)pPO+152
ofalc/regin-mthEukyd),RLbx(DsIW
Even after we adjust for these changes in risk, however, interlisting firms
still generate statistically significant abnormal returns, as measured by the
ai coefficients. In the prelisting period, firms earn 0.31 percent per week; in
the postlisting period, firms lose 0.22 percent per week (calculated as the
sum of aPRE, 0.31 percent per week, and the a.'OST, -0.53 percent per week).
This suggests that the abnormal returns performance around interlistings is
robust to changes in expected returns that are captured by shifts in risk
exposure.
Table V also reports similar results by groups of firms according to home
region. We find some important differences. First, the dramatic shift in local
market betas is evident for the Australian stocks. AL falls by 0.52 from 1.36
during the prelisting period to 0.84 during the post-listing period; /aw in-
creases by 0.22 from 0.08 to 0.30, although the prelisting global beta is not
significantly different from zero, nor is the change in the global beta. For
the Canadian firms, the significant positive abnormal prelisting returns
(0.49 percent per week) and negative abnormal postlisting returns (0.92 per-
cent lower or -0.43 percent per week) obtain, as in Foerster and Karolyi
(1993). Unlike other markets, however, neither the local market beta nor the
global market beta change is significant during the postlisting period. Firms
from the European and Asian subsample of ADRs retain the familiar in-
crease in prelisting abnormal returns, but do not yield a significant decline
in postlisting abnormal returns. The local market betas do drop signifi-
cantly, by 0.08 from 0.87 to 0.79 for Europe and by 0.90 from 1.23 to 0.33 for
Asia, and the global beta increases slightly for Europe but decreases for
Asia, although neither change is significant. Finally, the U.K. ADRs gener-
ate negative postlisting abnormal returns, and significant declines in both
local betas (by 0.52 from 1.00 to 0.48) and global betas (by 0.25 from 0.17 to
-0.08). One possible implication of the significant decrease in local betas
combined with the lack of significant increase in global betas suggests these
firms appear to be successful in lowering their cost of equity, and hence cost
of capital, by interlisting their stock on U.S. exchanges.
In order to observe the time series patterns in abnormal returns in event
time, we compute estimates of the two-factor IAPM model using Jbbotson's
(1975) Regression across Time and Securities (RATS) model. That is, we
reestimate equation (2) pooled across all securities (and by regional groups),
but in event-time on a week-by-week basis. We obtain estimates of the co-
efficients for each week T, where T runs from -52 to +52, and where the
listing week corresponds to r equal to zero. The cumulative values for the ai's
are computed. Figure 2 exhibits the results overall and by region. The cumu-
lative abnormal returns pattern for the Australian and Canadian firms fol-
low most clearly the prelisting run-up and postlisting decline. The drop for
the Australian stocks occurs during week 5, whereas that for the Canadian
stocks is more gradual. The cumulative abnormal returns for the European
and U.K. firms follow a general increase for the entire pre- and postlisting
period. The Asian ADRs demonstrate no discernible pattern at all.
30% -utai
Canada
-- - -Europe
25% -~~~~~ ~~~~~~~~~~~~~~~~~~~~~~Asia I
/ a ~~~~~~~~~- - -UK _
S20y%Az
15%
w .
o o5%
-5%
-1 0%
uz It I? CN CNe
Weeks Relative to Listing Date
Figure 2. Cumulative abnormal excess returns for global listings in the United States
by region. Cumulative abnormal excess returns are computed weekly using Ibbotson (1975)
RATS (Regression Across Time and Securities) in event time using a two-index IAPM model for
local currency excess returns with a local market index (Appendix B) and Datastream Inter-
national's World Index. The intercept coefficients from each weekly regression are cumulated
over the 52 weeks before and after listing. Data are obtained from Reuter's Exshare Inter-
national Securities Database using Reuterlink on-line data service and supplemented with var-
ious local periodicals. The returns are computed in excess of weekly one-month Euromarket
yields obtained from Harris bank quote sheets.
In summary, we compute measures of abnormal returns for a cross-

section of global listings before, around, and following their U.S. listing.
The pattern of a prelisting price run-up, listing week increase, and postlist-
ing decline is robust even after accounting for statistically important risk
changes around the listing. These abnormal return patterns also differ by
region, though caution should be exercised as these regional subsamples
are smaller. In the next section, we attempt to shed further light on these
results by examining whether there are important differences in these ab-
normal returns that are related to firm-specific variables, such as the list-
ing location, the industry group to which the firm belongs, or the type of
listing.
E. Univariate Cross-Sectional Tests
In this section, we recompute weekly abnormal excess returns and betas

for each firm based on 153 univariate time-series regressions as in equation
(2). We cumulate across event time a's, as appropriate, to compute listing
and postlisting abnormal returns and perform a series of cross-sectional tests
based on the univariate abnormal returns and changing betas. Table VI
reports dummy variable regression results of the abnormal returns and chang-
ing betas as the dependent variables with dummy variables as the indepen-
dent variables based on different regions, exchanges, industries, and listing
type. Average weekly abnormal excess returns are 0.15 percent in the prelist-
ing period, 0.12 percent in the listing week, and -0.14 percent in the post-
listing period. The fraction of firms with positive abnormal returns in the
prelisting period is 61 percent versus 52 percent around listing and 50 per-
cent in the postlisting period. The differences across regions indicate similar
patterns to those uncovered above. When we test whether the regional dif-
ferences are significant, we do not reject the null in the prelisting and postlist-
ing period but reject the null in the listing period itself. When comparing the
abnormal returns for global listings by exchange location, we find no signif-
icant differences on average. The NYSE listings seem to generate higher
positive abnormal returns, not only in the prelisting period but also in the
listing period as well as smaller declines than AMEX and Nasdaq firms
following listing. However, the X2-statistics suggest the differences are not
significant.
Industry differences are not an important factor in general. For the prelist-
ing period, the abnormal returns increase consistently but differences range
from as low as -0.27 percent per week for the technology stocks to 0.42
percent per week for industrial stocks. Similarly, the postlisting decline for
the technology stocks is the highest, and contrasts with the postlisting in-
crease observed for the utilities. The x2 test suggests the postlisting results
are measurably different by industry at the 10 percent significance level.
These postlisting returns results are consistent with earlier findings of dif-
ferences across industry sectors in Foerster and Karolyi (1993) and comple-
ment those studies examining the important role of industrial structure in
international diversification strategies.19
We also investigate whether there are possible confounding effects be-
tween industries and exchanges. For example, if technology stocks tend to
list on Nasdaq, then any apparent differences across exchanges may actually
be the result of differences across industries. To examine this, we rerun
cross-sectional regressions including both industry and exchange dummies
(to preserve space, results are not presented). We then test for differences
across industries (after controlling for exchanges) and for differences across
exchanges (after controlling for industries). In the prelisting period, abnor-
19 Roll (1992), Heston and Rouwenhorst (1994), and Griffin and Karolyi (1998) debate
importance of industry and country factors in international stock returns. Recent evidence is
also found in Karolyi and Stulz (1996).
Table VI
Mean Abnormal Returns and Changing Local and Global Beta by

Variables Related to Home Region, Choice of U.S. Exchange,
Industry Group, and Capital Raising
Abnormal returns are computed using regression model estimates of excess returns from before
listing (weeks -52 to -1), listing week (week 0), and postlisting periods (weeks + 1 to +52) for
an IAPM model using excess returns of local market index and global index. Distributions of
firms by region, U.S. exchange location industry group, and capital raising are provided in
Table II. A,8LO and A'WO are the differences between the prelisting and postlist
ative to the local market and global market, respectively. Abnormal returns are computed sep-
arately for each firm using local currency excess returns. x2 denotes a robust Wald test of the
difference between the mean abnormal returns across different groupings. Tests for signifi-
cance use Newey and West (1987) robust t-statistics. Wald tests and binomial Z-tests for per-
centage of positive/negative abnormal returns are indicated for significance at the 10 percent
* and 5 percent ** levels, respectively.
Abnormal Returns Beta Changes
Prelisting Listing Postlisting Change in Change in

a PRE aLIST aPOST Local Beta Global Beta
Category (-52,-i) (week 0) (+1,+52) AA3L APiW
Overall
Mean 0.0015* 0.0012 -0.0014 -0.3209** 0.1349
Percentage Positive 60.78%** 51.63% 50.32% 39.86%** 53.59%
By region
Australia 0.0052 -0.0276 -0.0016 -0.4904 0.1388
Canada 0.0034 0.0060 -0.0038 -0.1388 0.1910
Europe -0.0055 -0.0179 0.0007 -0.3674 0.4536
Asia 0.0016 0.0099 0.0011 -1.0679 -0.0739
United Kingdom 0.0019 0.0135 0.0007 -0.3367 -0.1372
Adjusted R2 3.25% 2.61% 2.98% 3.66% 2.21%
X2 5.52 12.29** 6.83 7.94* 5.886
By exchange
NYSE 0.0027 0.0100 -0.0011 -0.4075 -0.0790
AMEX 0.0071 -0.0047 -0.0037 0.0268 0.0563
NASD -0.0002 -0.0045 -0.0014 -0.3042 0.3021
Adjusted R2 1.30% 0.71% 0.27% 0.75% 1.88%
X2 3.58 1.62 0.35 1.64 3.75
By industry
Industrial 0.0042 -0.0029 -0.0001 -0.2508 -0.0990
Resource 0.0001 0.0026 -0.0034 -0.2974 0.2369
Consumer 0.0021 0.0079 -0.0007 -0.3407 -0.0037
Financial 0.0015 -0.0004 -0.0002 -0.4739 0.1097
Technology -0.0027 -0.0145 -0.0042 -0.2866 0.8240
Utility 0.0011 0.0055 0.0031 -0.4510 0.1695
Adjusted R2 1.08% 0.51% 2.22% 0.29% 3.23%
X2 5.85 1.21 10.47* 1.35 4.64
By level
Capital Level (III) 0.0013 0.0025 0.0017 -0.3922 -0.0765
No capital Level (II) 0.0016 0.0009 -0.0023 -0.3028 0.1886
Adjusted R2 0.04% 0.01% 1.62% 0.01% 0.63%
X2 0.02 0.02 7.22** 0.20 3.05*
**, indicate significance at the
mal returns are not significant across industries, but are marginally signif-
icant across exchanges (with a x2 p-value of 0.103). During the listing week,
abnormal return differences are not significant across either group. In the
postlisting period, abnormal returns are significantly different across indus-
tries (with a x2 p-value of 0.004), but are not across exchanges.
Capital raising differences provide surprising results. One hypothesis sug-
gested by Dharan and Ikenberry (1995) is that postlisting negative stock
returns following Nasdaq and AMEX listings on the NYSE can be explained
by managers of firms acting opportunistically in timing their listing. These
managers, typically of smaller firms, strategically apply for listing just be-
fore a decline in performance. Dharan and Ikenberry refer to this as the
opportunism hypothesis. They also offer this explanation for the equity is-
suance postlisting decline for U.S. initial public offerings (IPOs) and sea-
soned equity offerings (SEOs) uncovered by other studies, such as that by
Loughran and Ritter (1995). One possible explanation for the postlisting neg-
ative abnormal returns for the ADRs around their listing in the United States
is that the managers of these firms are timing their listing and equity is-
suance precisely when poor fundamental performance will follow. Our re-
sults in Table VI show that in the prelisting and listing periods both firms
that raised capital through ADRs and those that did not experience positive
abnormal excess returns and these returns are not significantly different
across the two groups. However, in the postlisting period, capital-raising
firms experience positive abnormal returns (weekly 0.17 percent, or annu-
alized 8.84 percent) but non-capital-raising firms experience negative abnor-
mal returns (weekly, -0.23 percent, or annualized -11.96 percent). The
differences across the two groups are statistically significant. These results
are contrary to findings in the IPO/SEO literature which show that firms
that raise capital tend to experience subsequent negative abnormal returns.
For example, Loughran and Ritter find a post-issue abnormal decline of 4.5 per-
cent for IPOs in the first year and a decline of 6.3 percent for SEOs. Our
results suggest global equity secondary offerings may differ from domestic
equity offerings. A more extensive investigation of this issue is warranted,
but is beyond the scope of this paper.
Table VI also presents results related to cross-sectional differences of chang-
ing local betas and global betas. Overall, firms experience a significant de-
cline in local betas. On average, postlisting local betas are 0.32 below the
level of prelisting betas. Furthermore, more than 60 percent of the firms
experienced a decline. In contrast, world betas tend to increase by 0.13, on
average, but only 54 percent of the firms experience an increase, and the
postlisting beta is not significantly different from the prelisting beta. Changes
in local betas are significantly different across regions. Canadian firms ex-
perience the smallest decrease in local betas, followed by U.K. firms; Asian
firms experience the largest declines. Firms from Australia, Canada, and
Europe experience increases in global betas; firms from Asia and the U.K.
experience declines, although differences in global beta changes are not sig-
nificant (X2 p-value of 0.210). There are no significant differences in either
local or global beta changes across exchanges or industries. In fact, local

beta changes are consistently negative across all industries and two of three
exchanges. Both capital-raising firms and non-capital-raising firms experi-
ence local beta declines of similar magnitudes, but significant differences
appear for the changing global betas. Capital-raising firms experience a de-
cline in global betas, non-capital-raising firms experience an increase.
IV. Abnormal Returns, Betas, Liquidity, and

Changes in the Shareholder Base
Although our findings on share price patterns around global cross-listings

are generally consistent with the market segmentation hypothesis, we pro-
pose in this section two alternative hypotheses that associate the share price
effects to changes in the underlying liquidity in the market for the shares
and to changes in the shareholder base. Foreign firms seeking access to U.S.
markets often cite these as two key factors in their decision to list abroad
(see Karolyi (1998)) but they are just as likely to influence U.S. firms that
may simply seek to change trading location (i.e., from Nasdaq to the NYSE).
In fact, existing research on domestic listings, such as Christie and Huang
(1993) and Kadlec and McConnell (1994), links the effects of listing choices
on share prices, liquidity, and changes in the shareholder base to theoretical
models developed by Amihud and Mendelson (1986) and Merton (1987). In
this section, we follow these studies on domestic listings and describe the
Amihud-Mendelson and Merton models, outline the new tests for global cross-
listings, and present key findings.
A. Merton's (1987) Investor Recognition Hypothesis
Merton's (1987) capital market equilibrium is different from that of the

Sharpe-Lintner CAPM in that investors consider only securities of which
they are aware, an assumption about incomplete information. With this as-
sumption, Merton shows that expected returns depend on factors other than
just market risk. Specifically, the shadow cost of incomplete information for
stock i is given as
Ai = 8O-i2 xi (Iq -)/qi, (3)
where 6 is the coefficient of aggregate risk aversion, oi2 is the firm-specific

component of the stock's return variance, xi is the relative market value of
the firm, and qi is the size of the firm's investor base relative to the total
number of investors. The relationship between the actual expected excess
return of the stock, E(Ri), and the expected excess return for the complete
information case (where qi equals 1), E(R*), is
E(Ri) - E(R*) = AiE(R*)/RO, (4)
where Ro is the return on the zero-beta asset. The intuition behind the re-
sult is that since investors consider only a part of the opportunity set, full
diversification is not possible and firm-specific risk is priced in equilibrium.
Moreover, this firm-specific risk is weighted by the relative market value of
the firm and its shareholder base. For firms with a relatively small share-
holder base, these factors are likely to be very significant on average. Indeed,
firms would have incentives to incorporate policies that actively expand the
investor base of the firm's shares. For non-U.S. firms cross-listing their shares
in the United States, we know that this is one of the primary motivations.20
We construct an empirical proxy for the shadow cost of incomplete infor-
mation, Ai, for each firm, following Kadlec and McConnell (1994). Specifi-
cally, we measure the change in A around the interlisting by
AAi = (,2ioSIZEi(1/SHRt+1 - 1/SHRt), (5)
where o-2i is the residual variance from our two-factor IAPM benchmark
model regressions, as in equation (2); SIZEi is the U.S.-dollar market capi-
talization of the firm as measured by the price of the stock in week 0 and the
number of shares outstanding, and is normalized by the level of the Data-
stream International World Stock index value in the listing week;21 and
SHRt is the number of shareholders of record in the year before (t) a
(t + 1) listing. The data on the number of shares outstanding and the share-
holders of record are obtained from various publications of Moody's Inter-
national Manuals in the year before and after listing. Due to data availability,
our sample is pared down slightly to 145 firms.
In Section II and Table II, we demonstrate that firms experience an in-
crease in their shareholder base by about 28.8 percent. The investor recog-
nition hypothesis of Merton (1987) suggests that the abnormal returns
experienced by firms during the pre- and postlisting period may be due to
changes in the shareholder base, adjusted by the stock's residual variance
and relative size. We follow Kadlec and McConnell (1994) and use cross-
sectional regressions, such as
Ti = yo + y1 AAi + eit, (6)
where ai is the pre- or postlisting or listing week abnormal returns.22 Given

that AAi is negative for most firms with an increase in shareholder base, we
expect that yi will be significant and negative in the cross-sectional regressions.
20 See Fanto and Karmal (1997) for a recent survey of corporate financial officers from
overseas companies that listed their shares in the U.S. for the first time.
21 We regress the capitalization value of each firm cross-sectionally on a constant and the
value of the Datastream International World Index (where January 1, 1975, equals 100). The
relative market value is computed as the residuals from this regression.
22 Note that this estimation suffers from an error-in-variables problem. Due to the estima-
tion error in AAi, this would imply that the coefficient yi is likely biased to zero.
We also investigate a cross-sectional relationship between changing betas

(both local and global) and changing shareholder base; that is, in separate
regressions, we replace the dependent variable, ai, of equation (6) with /XIL
and 4BVw, respectively. Changing betas reflect changes in the assessments
by the marginal investor of the domestic and global risk exposures of the
firm and, as a result, changes in the market's expectation of its future re-
turns. The Merton hypothesis would then predict a positive coefficient on
AAi for these beta changes because an increase in AAi due to a larger inves-
tor base should be associated with a decrease in covariance risk and thus
the cost of capital.
B. Amihud and Mendelson's (1986) Liquidity Hypothesis
Amihud and Mendelson (1986) develop an equilibrium asset pricing model

in which gross returns are shown to be an increasing and concave function
of liquidity. They proxy for liquidity using the bid-ask spread. In notational
form, the gross return required by investor i on asset j is
E(RJ)= R + iSj, (7)
where Rg is the required spr

liquidation cost (i.e., the inve
asset's relative spread, Sj). If
pected returns required by investors should give rise to an increase in share
value.
Unlike studies of the share price and liquidity impact of Nasdaq firms
listing on the AMEX or the NYSE, no data on bid-ask spreads in home mar-
kets are systematically available for the cross-section of non-U.S. firms. In
a domestic setting, Christie and Huang (1993) and Kadlec and McConnell
(1994) show that Nasdaq listings on the NYSE enjoy uniformly lower spreads
and that these changes are associated with abnormal returns for these firms
around listing. We propose to test this liquidity hypothesis of Amihud and
Mendelson (1986) indirectly by allowing the dependence of the abnormal
returns in the prelisting, postlisting, or listing weeks themselves in the re-
gression model of equation (6) to be differentially sensitive for NYSE, AMEX,
and Nasdaq listings. We employ a series of controls associated with the size
of the firm, whether it is a Level III capital-raising ADR or not, and, if so,
the size of the new issue. We also investigate changing betas in a similar
manner.
C. Multivariate Cross-Sectional Tests
Table VII presents the cross-sectional regressions of the abnormal re-

turns. Four different models are examined using the abnormal returns as
measured by the two-factor IAPM. We present regressions for the abnormal
returns from the pre- and postlisting periods and for the listing weeks in
erosctdfhkaiynl.
*indcatesgfh510prlv,y.
TableVI
A=(oE-2SIZ)*l/HRt+?,
ShareoldBs,nizfNwIu
asinTbleI.Mrktcompf(A)1987dKC4:
RegrsionfAbmaltuGLVdSz,
ICAP-0.1*76928 SIZE(10-9).826534
ISUE(10-6).435289*
PrelistngAbomaRuLd
Varible(1)234
R20.4%7598136 A(NSD)-2.3490716* A(MEX)-3.70265*49 A\-0.241*67385 Consta0.15*243-976 curenyadU.SolmitsR2hjf-NwW(1987)nr whero-c2istdualvnfmIAPMgbp.y capitlzon(SIZE),hgesrdbHRf umyCAPaniz(ISUE)re thesamodlfrbingp(wk-521)+.Vv Abnormalexcstupdfigwk(0)IPM nromhetc
three panels. In each regression, we report the coefficient estimates, de-

noted as statistically significant based on robust t-statistics from Newey and
West (1987) standard errors, and adjusted R2.
The first regression shows that the abnormal returns around the listings
are not significantly related to the market value of the firm, but are nega-
tively related to Merton's market incompleteness factor. The coefficient on
AXA is -0.24 for the prelisting period. During the listing weeks, we find that
the relationship is still negative on AXA at -1.27, but the relationship is not
significant. Finally, in the postlisting period, the coefficient on AXA is of the
expected sign, -0.45, and statistically significant at the 5 percent level.
These results are supportive of the Merton hypothesis and consistent with
Kadlec and McConnell (1994).
When the size and Merton incompleteness factors are introduced into the
regressions of abnormal returns, the intercept term is statistically indistin-
guishable from zero in the listing and postlisting periods. However, with R2
measures at approximately one percent, it appears that these two variables
cannot explain much of the cross-sectional variation in abnormal returns
across the 145 firms. The explanatory power is likely low due to measure-
ment errors in the incompleteness factor. Nevertheless, we interpret these
multivariate cross-sectional regression results as consistent with the inves-
tor recognition hypothesis because AXA is generally significant and is consis-
tently of the expected negative sign.
To gauge the robustness of these tests, we consider confounding influences
of other firm-specific variables, such as the type of ADR issue. For example,
we control for whether the ADR listing is Level III (capital-raising) and mea-
sure the size of the new issue. Regression (2) in Table VII demonstrates that
the dummy variable for a capital-raising ADR (ICAP) is statistically signif-
icant and positive in the post-listing period. The positive ICAP coefficient in
the postlisting period confirms our previous analysis presented in Table VI
and suggests that the general postlisting negative abnormal return pattern
is mitigated for firms that raise capital. Furthermore, the positive and sig-
nificant ISSUE coefficient in the postlisting period suggests the postlisting
decline is even more offset for larger issues. We combine the SIZE, AXA, ICAP,
and new issue variables in regression (3) and find that the statistical sig-
nificance of the negative AXA and positive ICAP coefficient in the postlisting
period is similar in the joint regression. As discussed earlier, this finding
that capital-raising ADR issues yield positive abnormal postlisting returns
is surprising given the U.S. IPO/SEO findings of Loughran and Ritter (1995).
To test the Amihud and Mendelson (1986) liquidity hypothesis, we allow
the coefficient on the Merton incompleteness factor to interact with a dummy
variable for the NYSE, AMEX, and Nasdaq. We know from Table VI that the
average abnormal returns in any of the subperiods around interlisting are
not different by exchange listing location. However, studies by Christie and
Huang (1993) and Kadlec and McConnell (1994) show important liquidity
effects for Nasdaq stocks listing on the NYSE. Regression model (4) indi-
cates that the impact of changes in the shareholder base around listings for
these firms does matter. In both the prelisting and postlisting periods, none
of the coefficients are significant. However, in the listing period, the AMEX
and Nasdaq AA coefficients are significantly negative and the NYSE coeffi-
cient is significantly positive. If the NYSE is viewed as a more liquid market
than the AMEX or Nasdaq, then these results are inconsistent with the li-
quidity hypothesis of Amihud and Mendelson (1986). More likely, these re-
sults represent only an indirect test of the liquidity hypothesis at best and
should be interpreted with caution.
Finally, we examine the relationship between changing betas and share-
holder base, as captured by AXA. We repeat the four regressions described
above using either the change in local betas or the change in global betas as
the dependent variable. Results are presented in Table VIII. Coefficients are
generally not significant, but the signs on AA are positive, which implies
that the decline in local and global betas is positively associated with an
increase in shareholder base. To the extent that the decline in betas can be
interpreted as a lowering of the firm's cost of equity, we find evidence once
again consistent with Merton's investor recognition hypothesis. The only sta-
tistically significant results obtain for regression (4), which allows AXA to
vary by exchange. In the case of both local betas and global betas, the NYSE
coefficient is positive and the AMEX and Nasdaq coefficients are negative.
This finding suggests that the overall statistical insignificance may stem
from an averaging of the effect across exchanges: the lower betas, and po-
tentially lower cost of equity, derive primarily from the NYSE cross-border
listings. Again, the explanatory power of these cross-sectional tests is lim-
ited and results should be interpreted with caution.
V. Conclusions
We document the effect on share value and global risk exposures of a

non-U.S. firm listing on U.S. exchanges. Our sample consists of 153 firms
from 11 countries that listed their shares for the first time in the United
States directly or as ADRs during the period from 1976 to 1992. We find that
these stocks earned a significant average excess return of 19 percent during
the year before listing, an additional 1.20 percent during the listing week,
but incurred a significant average decline of 14 percent during the year
following listing. We also find that a stock's market beta relative to its home
market index declines dramatically from 1.03 to 0.74 on average, but its
global beta relative to the world market index does not change significantly.
Existing studies have interpreted the dramatic patterns in share values around
cross-border listings as evidence of market segmentation due to direct or in-
direct investment barriers. To the extent that a higher risk premium is built
into the expected returns of such stocks as compensation for these investment
restrictions, the cross-border listings in the United States overcome these bar-
riers and their stock prices adjust accordingly. We find the evidence generally
consistent with the market segmentation hypothesis. However, we uncover two
other possible explanations for the abnormal return patterns for cross-border
onKadlecMC(194):
erosctdfhkaiynl.
*,indcatesgfh510prlvy.
TableVI
A=(f2SIZE)1/HRt+-,
ShareoldBs,nizfNwIu
ISUE(10-5).698732 ICAP-0.149327 SIZE(10-8).42936* ChangesiLoclBtGb
Varible(1)234 RegrsionfChaLcldGbBtVSz,
A0.241936*57
R20.38%4697
Consta-0.327*968415
A(MEX)-15.934*720
A(NSD)-13.879*52
curenyadU.SolmitsR2hjf-NwW(1987) whero-,2istdualvncfmIAPMgbp.y listng),caprdumy(ICAPezSUETb.Mkof1987 period(wks+1t52).VablvmcznSIZE,hgHRf one-mthU.Syild)Cagscbrfp(wk521 ofalc/regin-mthEukyd),bx(DsIW WestimanIAPMrkodlgfxcu,(pB)
listings which stem from the larger shareholder base and the greater liquidity
that these firms achieve upon listing in the United States. We show that a proxy
for a market incompleteness factor, which captures the impact of a heightened
level of investor recognition, is significantly related to the abnormal patterns
in the pre- and postlisting periods. This market incompleteness factor derives
from the incomplete information asset pricing model of Merton (1987). We also
offer evidence that the statistical importance of this increased investor rec-
ognition is sensitive to the listing location of the non-U.S. firm. Finally, a sur-
prising finding is that the postlisting decline of abnormal returns for these cross-
border listings is mitigated for those firms that raise capital at the same time.
This contrasts with the existing evidence on the long-run underperformance
of equity issuances via IPOs and SEOs in the United States, and poses an in-
teresting challenge for future research.
Appendix A. Sample Firms and Listing Dates
Date Date Date

(yr/mo/day) (yr/mo/day) (yr/mo/day)
Australia: Canada (continued): Canada (continued):

Boral 920319 Giant Bay Res 870121 Scintilore Res 831007
Broken Hill Prop 870528 Gold Knight Res 841024 Sonora 860731
Central Pacific 810312 Goldex 890112 Sceptre Res 820402
Coles Myer 881031 Greenstone Res 890112 Softkey Softward 901001
FAI Insurance 880928 Granges 860618 Tudor 880304
Natl Australia Bank 880624 Healthcare 881110 Tee Comm 920828
News Corp 860520 Hemlo Gold 920115 Total Energold 880923
Orbital Engine 911204 Horsham 900115 Transcda Pipelines 850530
Pacific Dunlop 870702 Highwood Res 820226 Westburne Intl. 780510
Santos 810321 Corona 880729 Wharfe Res 830201
Southern Pacific 810312 Intl Platinum 890526
Western Mining 900102 Intercity Products 780901 France:
Western Pacific 890317 IPSCO 911217 Alcatel Alsthom 920520
Intl Colin Energy 910515 LVMH Motet 871023
Canada: Lac Minerals 850731 Societe Elf Aquitaine 910614
American Barrick 850125 Laidlaw 831021 Thomson 860723
Aber Res 890320 Loewen 900515 Total 911025
BCE Inc 760818 Magna Intl 840823 Hong Kong:
BII Enterprises 920804 Mitel 810518 Hong Kong Teleph 881223
Belmoral Res 880104 Mirtronics 861202
Bunker Hill 900530 MSR Explorations 830913 Italy:
Central Fund Cda 860610 Minven Gold 891207 Benetton 890609
Centurian Gold 881003 Norcen Energy 890328 Fiat 890214
Consolidated Merc 860822 NHI Nelson 870626 Montedison 870716
Cornucopia Res 881003 Northern Telecom 751110
Consolidated Prof 820127 Nova Corp 880613 Japan:
Cineplex Odeon 870514 Newfield Mines 880314 CSK 830830
Cognos 870701 Nowsco Wells 770314 Hitachi 820414
Davidson Tisdale 850225 Petromet 880616 Honda 860523
Deprenyl 890808 Pegasus 820823 Kubota 761109
Dickenson Mines 810114 Pop Shoppes 770603 Makita 770125
Dreco Energy 900618 Quadra Logic 881025 Mitsubishi 890919
ECO Corp 900126 Quebec Sturgeon 830125 Pioneer 761213
Energex 870803 Rea Gold 850328 Sanyo 770401
Eastmaque Gold 870904 Repap Res 880202 TDK 820515
Fahnestock Vine 860828 SHL Systemhouse 850614 Wacoal 890428
Appendix A-Continued
Date Date Date

(yr/mo/day) (yr/mo/day) (yr/mo/day)
Netherlands Sweden: United Kingdom (continued):

Aegon 911105 AB Electrolux 870701 Grand Metropolitan 910313
Ahold 910408 AB SKF 851025 Hanson 861103
Akzo 890508 ASEA AB 830815 Huntingdon Intl 890216
Oce Van Grinten 841101 AB Volvo 841203 Imperial Chemical 831101
Philips 870414 LEP Group 880930
United Kingdom: London Intl Group 901002
Norway: Allied Irish Banks 890912 Medeva 910930

Hafslund Nycomed 920624 Attwoods 910412 Micro Focus Group 920526
Norsk Data 830210 Auto Security 920714 Natl Westminister 861022
Norsk Hydro 860625 Barclays 860909 Ratners Group 880713
Bass 900216 Royal Bank of Scot 891016
Beazer 870605 RTZ Corp 900628
Portugal: BET 870806 Saatchi & Saatchi 871208
Banco Comercial 920612 Cable & Wireless 890927 Smithkline Beecham 890727
Cadbury Schwepp 840911 Tiphook 911001

Spain: Carlton Commun. 870130 Tomkins 881107
Banco Bilbao Vizca 881214 CRH 890710 United Newspapers 870828
Banco Central 830720 Danka Business 921217 Waterford Wedgwood 870128
Banco De Santader 870730 ECC Group 920430 Wellcome 920727
Empresa Electridad 880601 ELAN Corp 910103 Willis Corroon 901009
Telefonica Espana 870612 Glaxo Holdings 870610 WPP Group 871229
Appendix B. Local Market Indexes
Australia All-Ordinaries (275) Norway FT-Actuaries-Norway

Canada TSE-Western Index Portugal Lisbon BPA
France CAC Broad Index Spain Madrid Gesant Index
Hong Kong Hang Seng Index Sweden J-P Index (40)
Italy BCI-Milan 260 Index United Kingdom FT-Actuaries UK
Japan Topix Index United States Standard & Poor 500
Netherlands Morgan Stanley Capital
International
REFERENCES
Adler, Michael, and Bernard Dumas, 1983, International portfolio choice and corporation fi-
nance: A synthesis, Journal of Finance 38, 925-984.
Alexander, Gordon, Cheol Eun, and S. Janakiramanan, 1987, Asset pricing and dual listing on
foreign capital markets: A note, Journal of Finance 42, 151-158.
Alexander, Gordon, Cheol Eun, and S. Janakiramanan, 1988, International listings and stock
returns: Some empirical evidence, Journal of Financial and Quantitative Analysis 23, 135-
151.
Amihud, Yakov, and Haim Mendelson, 1986, Asset pricing and the bid-ask spread, Journal of
Financial Economics 17, 223-249.
Bailey, Warren, and Peter Chung, 1995, Exchange rate fluctuations, political risk and stock
returns: Some evidence from an emerging market, Journal of Financial and Quantitative
Analysis 30, 541-562.
Baker, Kent, and Sue Meeks, 1991, Research on exchange listings and delistings: A review and
synthesis, Financial Practice and Education 1, 57-71.
Bank of New York, 1996, Global Offerings of Depositary Receipts: A Transaction Guide (Bank of
New York American Depositary Receipt Division, New York, N.Y).
Barclay, Michael, Robert Litzenberger, and Jerold Warner, 1990, Private information, trading
volume and stock-return variances, Review of Financial Studies 3, 233-253.
Bekaert, Geert, and Campbell Harvey, 1995, Time varying world market integration, Journal
of Finance 50, 403-444.
Biddle, Gary, and Shahrokh Saudagaran, 1992, Financial disclosure levels and foreign stock
exchange listing decisions, Journal of International Financial Management and Accounting
4, 106-148.
Black, Fischer, 1974, International capital market equilibrium with investment barriers. Jour-
nal of Financial Economics 1, 337-352.
Booth, Laurence, and David Johnston, 1984, The ex-dividend day behavior of Canadian stock
prices: Tax changes and clientele effects, Journal of Finance 39, 457-476.
Campbell, John, and Yasushi Hamao, 1992, Predictable stock returns in the United States and
Japan, Journal of Finance 47, 43-69.
Chan, K. C., G. Andrew Karolyi, and Ren6 Stulz, 1992, Global financial markets and the risk
premium on U.S. equity, Journal of Financial Economics 32, 137-167.
Christie, William, and Roger Huang, 1993, Dissimilar market structures and market liquidity:
A transaction data study of exchange listing, Journal of Financial Intermediation 3, 300-
326.
Citibank Securities Services, 1995, An Information Guide to Depositary Receipts (Citibank NA,
New York, N.Y).
Cochrane, James, James Shapiro, and Jean Tobin, 1996, Foreign equities and U.S. investors:
Breaking down barriers separating supply and demand, Stanford Journal of Law, Business
and Finance 241, 19-32.
Dharan, Bala, and David Ikenberry, 1995, The long-run negative drift of post-listing stock
returns, Journal of Finance 50, 1547-1574.
Dumas, Bernard, and Bruno Solnik, 1995, The world price of foreign exchange risk, Journal of
Finance 50, 445-479.
Errunza, Vihang, and Etienne Losq, 1985, International asset pricing under mild segmentation:
Theory and test, Journal of Finance 40, 105-124.
Fanto, James, and Roberta Karmel, 1997, Report on the attitudes of foreign companies regard-
ing a U.S. listing, Stanford Journal of Law, Business and Finance 3, 37-58.
Ferson, Wayne, and Rudi Schadt, 1996, Measuring fund strategy and performance in changing
economic conditions, Journal of Finance 51, 425-461.
Foerster, Stephen, and G. Andrew Karolyi, 1993, International listings of stocks: The case of
Canada and the U.S., Journal of International Business Studies 24, 763-784.
Frost, Carol,and William Kinney, 1996, Disclosure choices of foreign registrants in the United
States, Journal of Accounting Research 34, 34-58.
Griffin, John, and G. Andrew Karolyi, 1998, Another look at the role of the industrial structure
of markets for global diversification strategies, Journal of Financial Economics 50, 351-
373.
Gultekin, N. Bulent, Mustafa Gultekin, and Alessandro Penati, 1989, Capital controls and in-
ternational capital market segmentation: Evidence from Japanese and American stock mar-
kets, Journal of Finance 44, 849-869.
Heston, Steven, and Geert Rouwenhorst, 1994, Does industrial structure explain the benefits of
international diversification? Journal of Financial Economics 36, 3-27.
Howe, John, and Katherine Kelm, 1987, The stock price impact of overseas listings, Financial
Management 16, 51-56.
Howe, John, and Jeff Madura, 1990, The impact of international listings on risk: Implications
for capital market integration, Journal of Banking and Finance 14, 1133-1142.
Ibbotson, Roger, 1975, Price performance of common stock new issues, Journal of Financial
Economics 2, 235-272.
Jayaraman, Narayanan, Kuldeep Shastri and Kishore Tandon, 1993, The impact of inter-
national cross listings on risk and return: Evidence from American Depositary Receipts,
Journal of Banking and Finance 17, 91-103.
Jorion, Phillippe and Eduardo Schwartz, 1986, Integration versus segmentation in the Cana-
dian stock market, Journal of Finance 41, 603-616.
Kadlec, Gregory and John McConnell, 1994, The effect of market segmentation and illiquidity
on asset prices: Evidence from exchange listings, Journal of Finance 49, 611-636.
Karolyi, G. Andrew, 1998, Why Do Companies List Shares Abroad? A Survey of the Evidence
and its Managerial Implications (New York University Salomon Bros. Center Monograph,
Vol. 7, no. 1, New York, N.Y.).
Karolyi, G. Andrew, and Ren6 Stulz, 1996, Why do markets move together? An examination of
US-Japan stock return comovements, Journal of Finance 51, 951-986.
Lau, Sheila, David Diltz, and Vincent Apilado, 1994, Valuation effects of international stock
exchange listings, Journal of Banking and Finance 18, 743-755.
Loughran, Timothy, and Jay Ritter, 1995, The new issues puzzle, Journal of Finance 50, 23-51.
McConnell, John, Hank Dybevik, David Haushalter, and Edward Lee, 1996, A survey of evi-
dence on domestic and international stock exchange listings with implications for markets
and managers, Pacific Basin Finance Journal 4, 147-176.
Merton, Robert, 1987, Presidential address: A simple model of capital market equilibrium with
incomplete information, Journal of Finance 42, 483-510.
Miller, Darius P., 1999, The market reaction to international cross-listing: Evidence from de-
pository receipts, Journal of Financial Economics 51, 103-123.
Mittoo, Usha, 1992, Additional evidence on integration in the Canadian stock market, Journal
of Finance 47, 2035-2054.
Moody Investment Service, International Manuals, various issues, 1975-1993.
Newey, Whitney, and Kenneth West, 1987, A simple positive definite heteroscedasticity and
autocorrelation consistent covariance matrix, Econometrica 55, 703-708.
Prabhala, N. R., 1997, Conditional methods in event studies and an equilibrium justification for
standard event-study procedures, Review of Financial Studies 10, 1-38.
Roll, Richard, 1992, Industrial structure and the comparative behavior of international stock
market indices, Journal of Finance 47, 3-42.
Sanger, Gary, and John McConnell, 1986, Stock exchange listing, firm value and security mar-
ket efficiency: The impact of the Nasdaq, Journal of Financial and Quantitative Analysis
21, 1-25.
Schipper, Katherine, and Rex Thompson, 1983, The impact of merger-related regulations on
shareholders of acquiring firms, Journal of Accounting Research 22, 184-221.
Shanken, Jay, 1990, Intertemporal asset pricing: An empirical investigation, Journal of Econo-
metrics 45, 341-360.
Solnik, Bruno, 1974, Equilibrium in international asset markets under uncertainty, Journal of
Economic Theory 8, 500-524.
Standard and Poor Corporation, Stock Reports, various issues, 1976-1993.
Stapleton, Richard, and Marti Subrahmanyam, 1977, Market imperfections, capital market
equilibrium and corporate finance, Journal of Finance 32, 307-319.
Stulz, Ren6, 1981, On the effects of barriers to international asset pricing, Journal of Finance
25, 783-794.
Stulz, Ren6, 1995, International portfolio choice and asset pricing: An integrative survey; in
Robert Jarrow, Vojislav Maksimovic, and William Ziemba, eds.: Finance, Handbooks in
Operations Research and Management Science, Vol. 9 (Elsevier-North Holland, Amster-
dam).
Switzer, Lorne, 1997, Shareholder wealth effects of international listings: New evidence for
Canadian stocks listed on the NYSE, Amex and Nasdaq 1985-96, Unpublished working
paper, Concordia University.
Varela Oscar, and Sang Bin Lee, 1993, International listings, the security market line and
capital market integration: The case of U.S. listings on the London Stock Exchange, Jour-
nal of Business Finance and Accounting 20, 843-863.

The Efffect of Market Segmentaion On Invetsor

Uploaded by

Copyright:

Available Formats

The Efffect of Market Segmentaion On Invetsor

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Efffect of Market Segmentaion On Invetsor

Uploaded by

Copyright:

Available Formats

American Finance Association

The Effects of Market Segmentation and

STEPHEN R. FOERSTER and G. ANDREW KAROLYI*

Non-U.S. firms cross-listing shares on U.S. exchanges as American Depositary

THE GLOBALIZATION OF U.S. CAPITAL MARKETS has accelerated dramatically in the

over-the-counter (OTC), which represents a 75 percent increase since 1991.

same time, non-U.S. companies must satisfy two requirements to be listed in

pensions, industry, and geographic segment information) and can be based

Descriptive Statistics for Global Firms Listing

Home Region Listing Exchange Industry Group

Canada 67 New York 60 Industrial 34

Listing Years Capital Raising Capitalization

1976-1980 11 Level-III ADR 31 Mean ($mills) 2,515.3

New Share Issue Shareholder Base Percent Change in Base

Mean 3.0% Mean 52.9 Mean 28.8%

tribution is positively skewed with a median of only $717 million. By studying

strongly positively skewed with a median of only 10,000. The maximum of

III. Market Segmentation Tests

A. Global Market Segmentation, Cross-Listings, and Share Value

Global cross-listings have interested the international finance literature

B. Cumulative Abnormal Returns around Announcement

received approval from the relevant exchange or securities commission. It is

it = R it- [ai + fiLRrLt], (1)

where Rit is firm i's local currency return on day 0'1

Days Relative to Announcement or Listing Date

Figure 1. Cumulative abnormal returns around announcement/listing in the United

Abnormal Returns for 45 Firms around Announcement

(-100, -2) 0.1082 4.19**

(-100, -2) 0.0711 2.74**

Panel B of Table III examines listing effects. As in the preannouncement

C. Pre- and Postlisting Returns Performance

13 We are grateful to Nelson Mark for providing us with these data.

Summary Statistics for Weekly Excess Returns

Local Currency U.S. Dollar

Panel A: Before U.S. Listing (weeks -52 to -1)

All 7062 0.3815 4.65** 0.4409 4.96**

Panel B: During U.S. Listing Week (week 0)

All 153 1.2026 1.65* 1.2404 1.71*

Panel C: After U.S. Listing (weeks +1 to +52)

All 7681 -0.2651 -2.92** -0.3035 -3.31**

**,* indicate significance at the 5 and 10 perc

D. Risk Adjusted Returns

In order to isolate patterns in expected returns around interlistings, ex-

= PRE + PRE DL +pPRE DW LISTDLIST

+ aPOST DPOST + .PLOST RLtD PtOST +W8PwOSTRmwD POT + eit, (2)

where ari's are constants (which we interpret as abnormal excess returns);

1.64*8293-07 UnitedKgom0.16987-524%3 0.1798*6-25 Asia0.312945-87% 1.67*492-08

In summary, we compute measures of abnormal returns for a cross-

E. Univariate Cross-Sectional Tests

In this section, we recompute weekly abnormal excess returns and betas

Mean Abnormal Returns and Changing Local and Global Beta by

Abnormal Returns Beta Changes

Prelisting Listing Postlisting Change in Change in

**, indicate significance at the

local or global beta changes across exchanges or industries. In fact, local

IV. Abnormal Returns, Betas, Liquidity, and

Although our findings on share price patterns around global cross-listings

A. Merton's (1987) Investor Recognition Hypothesis

Merton's (1987) capital market equilibrium is different from that of the

All 7062 0.3815 4.65 0.4409 4.96

All 7681 -0.2651 -2.92 -0.3035 -3.31

1.648293-07 UnitedKgom0.16987-524%3 0.17986-25 Asia0.312945-87% 1.67*492-08

E(Ri) - E(R) = AiE(R)/RO, (4)