The Geography of Equity Listing:

Why Do European Companies List Abroad?

Marco PaganoCSEF, University of Salerno and CEPR

Ailsa A. RöellPrinceton University and CEPR

Josef ZechnerUniversity of Vienna and CEPR

First draft: 30 April 1999. This version: 11 December 1999

Abstract

This paper documents the aggregate trends in the foreign listings of companies and analyzesboth their distinctive pre-listing characteristics and their post-listing performance relative toother companies. In the 1986-97 interval, many European companies listed abroad, but did somainly on US exchanges. At the same time, the number of US companies listed in Europedecreased. The cross-listings of European companies appear to have sharply differentmotivations and consequences depending on whether they cross-list in the United States orwithin Europe. In the first case, companies pursue a strategy of rapid expansion fuelled by highleverage before the listing and large equity issues after the listing. They rely increasingly onexport markets both before and after the listing, and tend to belong to high-tech industries. Inthe second case, companies do not grow more than the control group, and increase theirleverage after the cross-listing. Also, they fail to increase their foreign sales in the wake of thecross-listing. The only common features of the two groups are their large size, high foreign salesbefore cross-listing and high R&D spending after cross-listing.

Keywords: cross-listings, going public, initial public offerings, geographyJEL classification: G15, G30, G39

Authors’ addresses: Marco Pagano, Dipartimento di Scienze Economiche, University ofSalerno, Via Ponte Don Melillo, 84084 Fisciano (Salerno), Italy; ph. +39 081 575-2508, fax +39081 575-2243, e-mail: mrpagano@tin.it. Ailsa A. Röell, Department of Economics, PrincetonUniversity, Fisher Hall, Princeton NJ 08544-1021, U.S.A.; ph. +1-609-258-4467, fax +1-609-258-6419, e-mail: aroell@princeton.edu. Josef Zechner, University of Vienna, Brünnerstrasse72, A-1210 Wien, Austria; ph. +43 1 4277-38071, fax +43 1 4277-38074, e-mail:zechner@finance1.bwl.univie.ac.at.

Acknowledgements: We thank Asher Blass, Yishay Yafeh, Gilles Chemla and seminarparticipants at Aachen, Copenhagen, Konstanz, Tel-Aviv, and the Barcelona CEPR-IAEWorkshop on Banking and Financial Markets for helpful comments. Larissa Lube and OttoRandl have provided outstanding research assistance. This research has been supported bygrants awarded by INQUIRE, the Fondation Banque de France, and the Italian Ministry ofUniversity and Scientific and Technological Research (MURST). This paper is produced as partof a CEPR research network on The Industrial Organization of Banking and Financial Marketsin Europe, funded by the European Commission under the TMR Programme (contract NoERBFMRXCT980222).

1

1. Introduction

Foreign listings are becoming an increasingly important strategic issue for companies

and stock exchanges alike. As companies become global in their product market and

investment strategies, direct access to foreign capital markets via an equity listing can

yield important benefits. At the same time, the international integration of capital

markets has led to unprecedented levels of competition among stock exchanges. In this

competitive struggle, the winners are the exchanges that manage to attract more foreign

listings and the attendant trading volume and business opportunities.

Despite the importance of these issues, still little is known about which exchanges

succeed in capturing more listings from abroad and why. This question is intimately

related with a second issue, namely which advantages companies expect to get from a

foreign listing: securing cheap equity capital for new investment, allowing controlling

shareholders to divest on a liquid market, preparing for foreign acquisitions, or simply

enhancing the company’s reputation. The evidence presented in this paper is relevant

for both issues − the determinants of exchanges’ success and the microeconomic

motives for listing abroad.

We start by providing a broad picture of the geography of cross-listings by European

and US companies, and of its changes in recent years. This aggregate picture shows that

European companies have become more “footloose” in recent years, and that most of

their cross-listings have been directed towards the US exchanges, while US companies

have reduced their cross-listings in Europe. Correspondingly, the ability of European

exchanges to attract listings from the rest of the world has declined, while the reverse

has happened to US exchanges. Interestingly, the European markets with the highest

trading costs, lowest accounting standards and worst shareholder protection have also

fared worst in attracting or retaining foreign listings, and companies from those

countries have been comparatively eager in seeking foreign listings.

We then turn to microeconomic data to gain a better understanding of these shifts in

the geography of cross-listings, by linking companies’ decision to list abroad to their ex

ante characteristics (e.g., size or foreign sales) and their ex post behavior (e.g., their

growth rate after listing abroad). We investigate these relationships by using company-

2

level data for non-financial European companies in 1986-97, drawn from the Global

Vantage and Worldscope databases.

We find that the European companies that list in other European exchanges and those

which list in the US have few common features: they all tend to be larger and more

export-oriented than the companies which not list abroad, and all tend to increase their

R&D spending after cross-listing. The differences between the two groups are far more

numerous and striking. European companies that cross-list in the US pursue a strategy

of rapid expansion fuelled by high leverage before the listing and large equity issues

after the listing. They rely heavily on export markets both before and after the listing,

and tend to belong to high-tech industries. Companies that cross-list elsewhere in

Europe, instead, do not grow more than the control group, and increase their leverage

after the cross-listing. Also, they do not rely on foreign sales to the same extent as firms

cross-listing in the US, and generally do not belong to high-tech sectors.

Thus, cross-listing in the US appears to be driven by the need to raise equity to fund

growth and foreign sales expansion, generally in high-tech sectors. These motives are

less common for European companies that cross-list in other European exchanges.

Therefore, the changing geography of cross-listings across the Atlantic is associated

with a difference in the type of companies that cross-list in the two continents. US

exchanges appear to be especially suited to the needs of high-growth, export-oriented

and high-tech European companies.

In Section 2 we outline the main reasons why companies may wish to list abroad and

draw testable predictions from each hypothesis. In Section 3 we analyze the overall

pattern of cross-listings, studying the geographical origin and destination of firms that

went public on the world’s major equity exchanges in 1986-97. In Section 4 we perform

a first exploration of company-level data using descriptive statistics centered on the year

of cross-listing. Section 5 presents an econometric analysis of the variables that affect

the choice to list abroad for the first time, as well as the choice between listing in the US

or in Europe. In Section 6 we try to gauge if listing abroad affects the subsequent

performance of companies relative to our control sample, and how this differential

performance hinges on cross-listing in the US as opposed to Europe. Finally, Section 7

summarizes the results of the paper, compares them with those of related studies, and

discusses their implications for the comparison between US and European exchanges.

3

2. Hypotheses and Related Literature

In this section we outline the reasons why companies may want to list on an

exchange outside their country of incorporation, either as their first port of entry into the

public equity market or after having already listed on their domestic exchange.1

First of all, companies may list abroad because funding abroad may be cheaper or

more easily available. This can happen for various reasons, detailed below in Section

2.1 jointly with their empirical implications. Second, the publicity that comes with a

listing can enhance a company’s reputation with its suppliers and customers, as

explained in Section 2.2. On the other side of the ledger, the costs of listing abroad may

deter certain companies, as discussed in Section 2.3. Table 1 summarizes the testable

implications of the various motives of the decision to list abroad, relating it both to (i)

the pre-listing company characteristics and to (ii) its likely effect on subsequent

performance.

2.1 Selling Securities on Better Terms

By listing abroad firms may improve the terms on which they can raise capital or on

which their shareholders can sell existing securities. This motive is strongest if the firm

or its existing shareholders need to raise capital and if financial constraints in the home

market are significant. Some of the empirical predictions relevant for this motive have

to do with the reason why capital is needed, and others have to do with why cross-

listing makes it cheaper.

The salient reason why a company may need funding is to carry out new investment

programs. The required funding is likely to be especially large for fast-growing

companies, and especially expensive if the company is already highly levered.

Therefore, companies which cross-list in order to raise capital cheaply should have a

1 The decision to list or cross-list on a foreign exchange is related to the more general issue of why firmsgo public, which has recently been explored in the finance literature: see Bolton and von Thadden (1998),Pagano and Röell (1998), Chemmanur and Fulghieri (1999), Mello and Parsons (1996), Pagano (1993),Röell (1996), Stoughton, Wong and Zechner (1996), and Subrahmanyam and Titman (1999). Thesepapers emphasize various advantages of going public from the standpoint of controlling shareholders,such as portfolio diversification, liquidity, information dissemination, ownership dispersion, and productmarket spillovers. Some of the insights of this literature can be brought to bear also on the decision to listabroad.

4

higher leverage, investment and growth rate before cross-listing and engage in a

primary equity offering at the time of the cross-listing or shortly afterwards. Moreover,

such companies would be more likely to cross-list on a deep stock market such as the

US or the UK than on shallow markets such as most exchanges in continental Europe.

Since higher expected growth should translate into higher price-earning ratios (P/E),

one would also expect them to have higher P/E ratios than comparable domestic

companies.

Rather than via organic growth, a company may choose to expand by a merger or

acquisition involving a foreign company. The acquisition of a target company is

facilitated by using the bidder’s shares as a medium of exchange, but the latter are an

acceptable “currency” only if the two companies are listed on the same exchange.2

Even if the firm has no need to finance new investment, its current shareholders may

want to sell out, and listing abroad can increase the market value of their stake.

Privatizations are an important special case, where the government is the divesting

shareholder. Therefore, we would expect newly privatized companies to be more likely

to cross-list than other comparable companies. A more direct test would look at

whether, in general, the main shareholders sell out at the time of cross-listing or shortly

afterwards. An imperfect proxy for such divestment can be an abnormally high

turnover.

We now turn to the reasons why listing abroad can raise a company’s stock market

value.

a) Reducing barriers for foreign investors

Widening the clientele for a firm’s shares improves risk sharing and thus lowers the

cost of capital, as shown by Stulz (1999), Martin and Rey (1999) and Lombardo and

Pagano (1999). The empirical evidence summarized in Karolyi (1998) shows that

indeed cross-listing reduces the cost of capital for firms, whether this is measured by

secondary market returns or by accounting variables such as the price-earnings ratio or

the market to book ratio. The reduction in the cost of capital appears to reflect, at least

2 Listing abroad may also relieve constraints on future growth options by creating the necessary contactsand reputation in the local financial community (e.g., with investment bankers, regulators, etc.), and byfacilitating the identification of potential target companies.

5

partly, the reduction of the company’s home market beta. This gain should be larger for

riskier firms, which therefore should have greater inducement to cross-list.3

Listing abroad can mitigate market segmentation by reducing barriers to foreign

investors, such as

(i) regulatory barriers: for instance, pension funds may face a ceiling on the proportion

of their assets invested in stocks not listed on their domestic market;

(ii) transaction costs: for instance, dividends of non-US shares held by US residents

must be converted into dollars at the expense of individual investors;

(iii) informational frictions: these range from total ignorance of foreign investment

opportunities as in Merton’s “awareness hypothesis”,4 to an informational

disadvantage when trading foreign stocks, as in Gehrig (1993), Kang and Stulz

(1994) and Brennan and Cao (1997).5 A foreign listing may reduce such frictions,

supplying local investors with more abundant, timely and transparent information.6

Foerster and Karolyi (1999) provide the most direct evidence connecting Merton’s

“awareness hypothesis” to the drop in the cost of capital at the time of cross-listing: they

show that the price of cross-listing companies rise more when they are accompanied by

a greater expansion of the shareholder base.

Related evidence is reported by Kadlec and McConnell (1994) for over-the-counter

shares which listed in the New York Stock Exchange (NYSE): they find that the listing

is accompanied by a 5 percent abnormal return, by an increase in the number of

shareholders and a reduction of the bid-ask spread. Also consistently with the

“awareness hypothesis”, Baker, Nofsinger and Weaver (1999) document that cross-

listing on the New York and London exchanges is associated with increased analyst

coverage and media attention.

3 No study so far has examined if cross-listing companies have lower betas with the destination marketand higher home market betas than comparable domestic companies. Consistently with Merton's (1987)model, such firms would reap the highest risk sharing gains from listing abroad.4 Merton (1987) derives a simple model of market equilibrium with incomplete information. Listing in aforeign market can be easily incorporated in this framework by assuming that it involves a cost butbroadens the firm's investor base. In such a framework, risk characteristics should determine which firmsare most likely to incur the cost of broadening its shareholder base by listing in a foreign market.5 The “home bias” induced by informational frictions may take the form of overconfidence aboutdomestic shares relative to foreign ones. Kilka and Weber (1997) produces experimental evidence thatpotential investors are more optimistic about the expected performance of domestic shares than that offoreign shares. The publicity associated with a cross-listing could change this perception.

6

Finally, Miller (1999) shows that the stock price reaction to a cross-listing is

positively correlated both to the increase in the shareholder base and to the barriers to

capital flows: abnormal returns are largest when companies list on major US exchanges

such as the NYSE or Nasdaq and when they originate from emerging markets.

b) Capitalizing on product market reputation

Companies which sell popular brands abroad may find it easier to place their shares

in foreign markets because local investors already trust them as consumers. A simple

strategy to test this hypothesis is to look at indicators of the degree of sales

internationalization for companies which cross-list. One would expect a larger fraction

of revenue coming from abroad to encourage eventual cross-listing. Saudagaran (1988)

showed that 104 companies from 9 countries that were already listed abroad in 1981

had a higher proportion of foreign sales than a control sample. This however begs the

question of which came first: the cross-listing or the outward orientation of these

companies. Only the latter would be consistent with the idea that these companies cross-

listed to capitalize on their product market reputation.

c) Relying on foreign expertise

The exchange where a company lists may be determined by the location of analysts

with superior technological knowledge of the industry. Especially in high-tech sectors,

the availability of such skills may substantially affect the availability of equity finance

and the terms at which it is available, by reducing informational asymmetries in the

primary market. This hypothesis predicts, for example, that high-tech companies may

be more likely to list in the US where the corresponding industries are well developed.

Evidence in this direction is already provided by Blass and Yafeh (1999), who show

that Israeli and Dutch firms which list in the US (bypassing their respective home

markets) are relatively high-tech and fast growing.

6 Access to cheaper capital not only through the stock market but also through the bond market and tradecredit through suppliers. This may be due to a reduction in adverse selection since strictly moreinformation is available about the company.

7

d) Committing to disclosure and corporate governance standards

The listing location may also be affected by differences in regulatory regimes,

especially concerning shareholder protection. By selecting a more tightly regulated

foreign exchange, a firm precommits to adhere to higher standards of corporate

governance and/or disclosure. Exchanges compete to attract cross-listings by designing

a regulatory environment that is expected to lower the cost of capital of their companies.

Huddart, Hughes and Brunnermeier (1998) present a model in which exchanges

competing for trading volume engage in a “race to the top” regarding disclosure

requirements.7 Ashbaugh (1997) documents firms’ decisions to voluntarily adopt the

tighter US accounting standards. Fuerst (1998) argues that firms signal quality by listing

on strictly regulated markets.8 Similarly, Stulz (1999) argues that companies from

countries with poor legal standards can secure a lower cost of capital by subjecting

themselves to the tighter regulatory standards of other markets and thereby reducing the

agency costs of external finance.

At the empirical level, this suggests that companies located in countries with

particularly inadequate supervision and disclosure standards would be more likely to

cross-list abroad. However, so far the evidence on this point is mixed. On one hand,

Biddle and Saudagaran (1989) and Saudagaran and Biddle (1992) find that stringent

disclosure requirements deter the listing of foreign companies. On the other, Reese and

Weisbach (1999) report that firms from countries with French Civil Law systems, which

give weak protection to minority shareholders, list abroad more frequently than firms

from other countries, and cross-list more often on organized exchanges such as NYSE

or Nasdaq rather than on the Over the Counter (OTC) market. Of course, to the extent

that exchanges compete for new listings by adjusting their regulatory standards, this

motive for cross-listing may diminish over time. For example, Fanto and Karmel (1997)

suggest that current improvements in European regulatory standards are attracting US

institutional investors to stocks exclusively listed in Europe.

If foreign accounting standards and disclosure requirements act as a better guarantee

for minority shareholders than corresponding home-market regulations, companies

which cross-list can also be expected to feature a larger decrease in the concentration of

ownership relative to their domestic counterparts.

7 On this point see also Chemmanur and Fulghieri (1998).8 See also Cantale (1996).

8

In addition, the signaling hypothesis predicts that the post-listing profitability of

companies which cross-list on a more demanding exchange should be better than that of

companies which cross-list on other exchanges. This should also be reflected in a

positive stock price reaction to the cross-listing announcement. This prediction is

supported by several studies surveyed in Karolyi (1998), which show that the price

reaction is significant for non-US companies listing in the US, which has the tightest

disclosure standards, whereas it is negligible otherwise.

e) Liquidity

Some markets may be better positioned than others in the production of liquidity, for

instance because of a better design of their microstructure or because of a larger number

of competing market makers. The beneficial effects of competitive pressure from

another exchange can also narrow the spreads on the domestic market and thereby raise

its trading activity, as found by Kadlec and McConnell (1994), Noronha, Sarin and

Saudagaran (1996), Foerster and Karolyi (1996) and Smith and Sofianos (1997).

However, cross-listing may not always lead to a greater liquidity for one’s shares,

due to the potentially offsetting impact of market fragmentation. Domowitz, Glen and

Madhavan (1998) show that opening trade on the foreign market may reduce liquidity in

both the domestic and the foreign market, if intermarket information linkages are poor,

and support this point with evidence concerning Mexican companies issuing American

Depository Receipts (ADRs).

To test if liquidity might be a reason for cross-listing, one can compare indicators of

trading for cross-listed shares, such as turnover ratios or bid-ask spreads. For instance,

the liquidity of the US market is unlikely to be an important motive for cross-listing for

European cross-listed companies if their turnover in the US turns out to be small

relative to their domestic turnover.9

f) Relative mispricing

Firms may decide to list abroad to take advantage of a temporarily high price for

their shares abroad relative to their home market. This could be due either to an

overvaluation in the foreign market or to an undervaluation in the domestic market. This

9 A related issue is whether foreign trading volume of cross-listed stocks tends to remain permanentlyhigh after the foreign listing or gravitates back towards the home market over time.

9

can be tested by including the price indices of the two exchanges (or the relevant

sectoral indices) in regressions explaining the probability of a foreign listing.

2.2. Product Market Spillovers

Firms’ decisions about where to list may be affected by the location of their product

market, because the cross-listing may generate a positive feedback on the company’s

demand from consumers and its relationships with suppliers, owing to its perceived

reliability, managerial and product quality.10 Stoughton, Wong and Zechner (1998)

provide a model where a company lists in order to signal to consumers its high product

quality, and as a result captures a larger share of the market and increases its profits. In

this model, a listing is not associated with the need to raise capital or with the intention

to sell out by existing shareholders. The product market spillover hypothesis predicts

that cross-listed companies increase their fraction of foreign sales. It is also consistent

with higher overall sales growth and profits after the cross-listing. Furthermore, it is

only potentially relevant for industries in which the firm’s product market reputation is

of particular significance.

As far as profitability is concerned, caution is in order because this effect is also

consistent with other hypotheses. Furthermore, the corresponding test suffers from the

problem that the timing of cross-listing may be endogenous. Companies may choose to

list when their earnings performance is abnormally good, so mean reversion may induce

a drop in profitability after listing. This bias could be partially corrected by including

the change in profitability before listing as one of the explanatory variables in a

regression context. This would still, however, fail to control for accounting

manipulation, as found, for instance, by Teoh, Welch and Wong (1997a, 1997b). The

latter may be corrected only by using as control sample a set of companies that go

public in the domestic market, since these would presumably attempt the same kind of

accounting manipulation.11

10 It may also improve the quality of its managerial decisions since, after the foreign listing, its stock priceincorporates information which otherwise managers may have overlooked.11 A post-listing improvement in profitability may also show up in the stock price performance, but weshall not analyze this aspect, which has already been thoroughly researched. For instance, Foerster andKarolyi (1999) analyse companies that went public in their domestic markets and subsequently listed inone of the main US markets, and document abnormal positive performance before but abnormal negativeperformance subsequent to the foreign listing.

10

2.3 Cost of Listing Abroad

Listing abroad also involves a variety of costs. There are direct costs, such as listing

charges and fees for professional advice. But the main costs cited in survey evidence

regarding potential cross-listings in the US (see Fanto and Karmel (1997)) are the cost

of complying with US GAAP accounting standards and the risk of lawsuits.

Presumably, shareholders’ power to interfere in managerial decisions increases with a

US listing. This survey evidence agrees with the results of the above-quoted studies by

Biddle and Saudagaran (1989) and Saudagaran and Biddle (1992).

Since the costs of cross-listing include a large fixed cost element, they bear most

heavily on small companies. Thus, we expect larger companies to be more likely to

cross-list. This prediction is borne out by Saudagaran’s (1988) study.

3. The Changing Geography of Equity Listings

This section provides a description of the cross-listing behavior of European and US

companies in the last decade. First, we document the “geography” of cross-listings, by

gauging regional clusters in cross-listing behavior. Second, we inquire if these patterns

have changed over time, and how. In particular, we investigate if there have been

substantial changes in “transatlantic listings”, that is, in the tendency of European

companies to list in the US and of US corporations to list in Europe. Third, we try to

relate these changes to the characteristics of the European and US exchanges concerned.

The sources of the cross-listing data used in the tables and figures of this section are

described in Appendix A.

3.1 Geographical Pattern of Cross-Listings

Table 2 summarizes the pattern of foreign listings in 1986-97 on the following stock

exchanges: Amsterdam, Brussels, Frankfurt, Milan, London, Madrid, Paris, Stockholm,

Vienna, Easdaq, Amex, Nasdaq, and NYSE.12 Panel A displays a matrix of foreign

12 The figures in the table refer to the stock of foreign listings on a given market, not the flow of newlistings in a given year.

11

listings, with the country of incorporation appearing in the columns and the destination

stock exchange along the rows. Each cell of the table contains three values: the top one

refers to 1986, the middle to 1991, and the bottom one to 1997. For each stock

exchange, the table displays only the foreign listings originating in the countries of our

sample: Netherlands, Belgium, Germany, Italy, UK, Spain, France, Sweden, Austria

(henceforth shortened to EU9 countries) and US. For instance, Japanese, Australian or

Canadian companies are excluded (evidence on these is deferred to Panel C of Table 2).

Looking at the column of a given country, one learns where the companies

originating from that country tend to cross-list. The column EU9 indicates in which

exchanges European (EU9) companies as a whole tend to cross-list. The last column

shows how the total number of cross-listing companies from the EU9 and US area have

distributed themselves within the area. The table indicates that European companies list

abroad more frequently within Europe than in the US. For instance, only 4 out of 32

foreign listings by Belgian companies were on the three US exchanges. This is true for

all EU9 countries’ companies (with the exception of the UK in 1997).

The table also suggests that common language and similar institutions foster cross-

listings. For example, the Vienna stock exchange is the single largest destination for

German companies and vice versa. The same is true for the US and the UK.

This “clustering” indicates that companies tend to cross-list in countries

geographically or culturally close to their country of incorporation. This parallels the

findings by Portes and Rey (1999) and by Tesar and Werner (1995) that geographical

proximity and cultural homogeneity (especially language) greatly enhance cross-border

securities transaction flows. Geography and culture seem to matter also in cross-listing

behavior, presumably for the same reason: as a proxy for informational asymmetries.

For a US investor, for instance, the accounting data and the behavioral patterns of a

British company are easier to decipher than those of a French or Spanish company.

Looking at each row across columns, we learn what has been the contribution of each

country to the total number of foreign listings in a given market. Comparing the figures

in the US column with the others, American companies turn out to be the single largest

source of foreign listings on European exchanges. This is particularly evident for

Amsterdam and London, where in 1997 American corporations still are over two thirds

of the total number of foreign listings, despite the considerable decline in their number.

12

3.2 Changes in the Geography of Cross-Listings

The information in Panel A of Table 2 also gives a picture of how the geography of

European and US cross-listings has changed between 1986 and 1997. The two bottom

lines give an overall view of the change in the cross-listings pattern. The row “Total

Listings” displays the number of listings that companies from a given country have in

the foreign exchanges included in our sample. The bottom row “Total Companies”

eliminates double counting by reporting the number of companies from a given country

with at least one foreign listing. The number of foreign listings originated by a given

country is greater (or at most equal) to the corresponding number of companies listed

abroad, because the same company can be listed in several foreign exchanges.

The numbers in these two rows reveal that European companies have become more

outward looking in their search for investors: the number of EU9 companies listed

abroad doubled (from 177 to 337) and the total number of their foreign listings

increased by 61 percent (from 320 to 516).

In contrast with European companies, European stock exchanges do not appear to

have become equally outward oriented. Foreign listings on most European exchanges

exhibit an inverse U-shaped time pattern over time. In the European exchanges as a

whole, the total number of foreign listings increased very slightly from 732 in 1986 to

757 in 1991, and then declined to 625 in 1997 (see the last cell in the row “European

exchanges”). So these exchanges lost over one hundred foreign listings in a decade.

The opposite picture emerges when one considers American companies and

exchanges. US companies have become less eager to list in Europe, with their number

decreasing from 284 to 184. In contrast, US exchanges (especially the Nasdaq and the

NYSE) have captured an increasing share of foreign listings by European companies:

the listings of EU9 companies in the US went from 53 in 1986 to 207 in 1997, while in

the same interval their listings within Europe went from 267 to 309.

The contrast between these two opposite flows of “transatlantic listings” emerges

very clearly in panel B of Table 2. While European listings in the US almost

quadrupled, the number of US companies listed in Europe fell by over a third. In 1986

the US firms listed in Europe were more than five times as many as the European firms

13

listed in the US. In 1997, the latter outnumber the former. This suggests that the relative

attractiveness of European equity markets declined in this time window.

Panels A and B of Table 2 do not provide a full account of the outward orientation of

each exchange, because they neglect the listings originating outside our sample of

countries. Panel C completes the picture, by reporting cross-listings originating from the

rest of the world. Canadian, Latin American and Israeli companies are major sources of

listings in US exchanges, while they list much less frequently in Europe. In contrast,

South African and Asian companies list predominantly in London - with the exception

of Japanese corporations, which gravitate primarily towards Frankfurt. Considering

instead how the overall pattern changed over time, one sees again that the US exchanges

have captured the lion’s share of the increase in cross-listings from the rest of the world,

especially those from Australia, Canada, Latin America and Israel. In contrast, in most

cases, European exchanges have lost cross-listings originating from these regions.

The data in Table 2 raise two questions. First, is the decline of foreign listings on

European exchanges part of a more general decline in their ability to attract new

listings, including domestic ones? Second, are the three data points reported in Table 2

representative of the overall history of cross-listings between 1986 and 1997? Figures

1a and 1b address both questions.

Figure 1a (on the left) displays the time pattern of domestic and foreign companies

listed on each exchange, as well as their total number. The general impression is that the

European exchanges’ inability to attract new listings is not confined to foreign listings

alone. Most of them have not attracted a large number of new domestic listings either,

especially in the 1990s, with the exception of Frankfurt and, to some extent, of London.

The opposite is true of US exchanges, where both domestic and foreign listings

increased over the sample period: domestic listings rose from 6,168 in 1986 to 7,950 in

1997 (a 29% increase), while foreign listings increased from 350 to 873 (a staggering

150% increment, mostly accounted for by the NYSE).

Figure 1b (on the right) shows how cross-listings from our EU9 countries and the US

evolved in each exchange. It is based on the same data as Table 2, except that it reports

figures for all the years of our sample. The dotted line is the number of foreign

companies (from the rest of EU9 and the US) listed on a given domestic exchange,

whereas the solid line is the number of domestic companies listed in other EU9 and US

14

exchanges. For almost all the European exchanges, the dotted lines are declining and the

solid lines are rising, especially toward the end of the sample period, whereas the

opposite is true for US exchanges. This confirms the finding of Table 2.

A possible measure of the “outward orientation” of these markets is the ratio between

its foreign listings and its total listings. This measure is shown in Figure 2. The ratio

declined in Amsterdam, Frankfurt, London, and Vienna, was roughly stable in the other

European exchanges, Amex and Nasdaq, and increased substantially in the NYSE.

Conversely, Figure 3 shows a measure of the tendency of domestic companies to

cross-list in foreign exchanges, by reporting the ratio of cross-listed companies to the

total of domestically listed ones. We call this the “diaspora” index. For most European

countries, the diaspora index has increased substantially, with the smaller countries

recording the strongest diaspora.

3.3 Relationship with Characteristics of Stock Exchanges

The changes in the geography of equity listings documented in Table 2 and in

Figures 1 to 3 raise the question if they are related to some characteristics of the

exchanges and countries concerned. Since our data include only twelve exchanges and

ten countries, we cannot hope to draw reliable inferences from the data about such

correlations. Nevertheless, it is intriguing to see if the markets whose companies are

more eager to list abroad and which attracted less foreign listings differ systematically

from those with the opposite experience. In Table 3 we provide some information on the

market characteristics that may be related to the cross-listing decisions of companies,

based on the hypotheses outlined in Section 2 and summarized in the last column of

Table 1: accounting standards, degree of investor protection, size of the market (as

measured both by market capitalization and turnover) and trading costs.

In the first three columns we report information on the gross and net change in cross-

listings of each exchange, drawn from the same data used for Table 2, Panel A.

Consistently with the results so far illustrated, most EU9 markets are net losers of

listings, Sweden being the only exception, while the US market experiences a net gain.

The normalized net change in the fourth column of the table indicates that the net loss

has been particularly large in the Netherlands, followed by Great Britain, Belgium and

Austria (in this order).

15

With the glaring exception of Great Britain, the normalized net change in cross-

listings appears to correlate positively with accounting standards and investor

protection, in agreement with the evidence produced by Reese and Weinbach (1999).

The Netherlands, Belgium and Austria have relatively low values for both variables,

even within Europe. Sweden and the US have instead the highest rating on accounting

standards; Sweden fares a fair degree of investor protection and the US scores very

highly on this front. In general, Europe − again except for the UK − compares

negatively with the US on both accounts.

Market capitalization and turnover hold less promise as a potentially explanatory

variable. Some of the markets that lost relatively more listings are small (the

Netherlands, Belgium and Austria), while the NYSE and NASDAQ are obviously the

largest. However, in this case the exceptions are not confined to the UK: relatively small

markets like Sweden are doing relatively well, and others like Italy and Spain are not

losing comparatively as much as larger markets like France, Germany and the UK.

Of all five variables, trading costs is the indicator which appears to have the closest

correlation with the normalized net change in cross-listings. The four markets with the

highest trading costs − Great Britain, Austria, the Netherlands and Belgium, in that

order − are precisely those featuring the largest net outflow of cross-listings. The

NYSE, which attracts most cross-listings, has the lowest trading costs (together with

Spain, and immediately followed by Germany, France, Italy, Nasdaq and Sweden).

In conclusion, the data in Table 3 are consistent with the view that low trading costs,

good accounting standards and strong investor protection may play a role in the cross-

listing decisions of European and US companies. They also help to explain the one-way

net flow of transatlantic cross-listings in the 1986-97 interval (with the partial exception

of the UK, which has high trading costs but excellent accounting and investor protection

standards). Of course, the paucity of the data points in Table 3 is such that this

conclusion should be regarded as purely tentative. To make further progress in our

understanding of the cross-listing decision, we now leave the big picture and turn to

microeconomic evidence.

16

4. Company-Level Data: Descriptive Statistics

In the rest of this paper, we investigate the characteristics and performance of the

companies that cross-list, using companies that do not as our control sample. The

sample includes all the companies incorporated in our nine European countries for

which balance sheet information is available in the Global Vantage data base for the

1986-1997 interval. We exclude from the sample financial companies and investment

funds, as well as companies not listed in their country of incorporation.13

Summary statistics for the entire sample are provided in the Panel A of Table 4. The

total number of companies is 1,823. The median company has assets of US $ 372

million, sales of US $ 400 million and 3,022 employees. The median growth rate of

assets is 7.6 percent, property plant and equipment 6.7 percent, sales 8.2 percent, and

employees 1.2 percent. The median company has leverage of 19 percent, market to book

ratio of 1.81 and earns one third of its revenue from foreign sales.

There is huge variation in the values of some of these variables, even though we

eliminated economically meaningless outliers, such as negative sales figures (see the

appendix for details). For instance, total assets range from US $ 170 thousand to 113

billion, and the growth rate of plant property and equipment ranges from –100 percent

to over 1 million percent. This points to the need for robust statistical analysis in our

hypothesis testing.

R&D data is only provided for a very small proportion of the companies in the

sample; these spend 1.72 percent of revenue on R&D on average. To remedy the

paucity of observations on R&D, we construct an alternative “high-tech intensity”

indicator, based on the company’s SIC 4-digit classification code (see the appendix for

details). This dummy classifies 10.15 percent of the sample as high-tech companies.

Panel B of Table 4 illustrates the composition of the sample in terms of country of

incorporation and proportion of companies cross-listed, distinguishing those that were

already cross-listed in 1986 from those that cross-listed during the sample period. For

all countries of origin only a small proportion of sample companies, about 12.5 per cent,

list abroad at all. In terms of the country composition of our sample, the United

17

Kingdom is very heavily represented: nearly half of all companies studied, and over half

of the companies that first list abroad during our sample period, are British.

We now turn to a first comparison of the companies that list abroad with those that

do not, mainly focusing on balance sheet variables (such as total assets and sales) and

ratios (such as leverage and market to book value). Panel A of Table 5 reports the

difference between the median values of these variables for the cross-listed companies

and the companies listed only domestically, controlling for calendar year and country of

incorporation. More precisely, the values reported in the table are obtained by fitting a

median regression on a constant, a cross-listing dummy variable as well as control

dummies for calendar year and country. There are eight cross-listing dummy variables:

each one represents a particular year relative to the year of cross-listing, ranging from

year –3 (three years before) to year 3 (three years after) and a “permanent” dummy (4

or more years after).

The results concerning company size show that cross-listing companies are

significantly larger than companies that are only listed domestically. This is the case for

all the years relative to the listing period and for every size measure considered: total

assets, market value of common stock, revenue and number of employees. The

coefficient of the dummy variable for the permanent difference in median size is by far

the largest. This reflects the fact that this dummy captures mostly companies that were

historically the first ones to cross-list, and these are the largest companies in their home

country.14 The relatively large size of cross-listed companies agrees with the view that

there are economies of scale in cross-listing, reflecting fixed costs combined with

benefits that increase with company size.

Turning to the relationship between cross-listing and company growth, the table

displays growth in total assets, employment, sales and plant-and-equipment. For all

these variables, there is a marked peak in growth in the four years surrounding the

cross-listing date. In that period, the growth rates for cross-listing firms exceed the

growth rates of the control sample by about three to six percent, peaking in year zero

and reverting to normal two years later. That cross-listing is associated with a period of

13 Within the data base, we select only the companies in the sections “industrial active” or “industrialresearch” and listed on their home-country stock exchange. To exclude financial companies, allcompanies with SIC-codes starting with 6 have been dropped.14 This finding does not imply that companies which cross-list grow faster in the long run.

18

exceptional growth, is consistent with the notion that new capital needs to be raised.

However, it is striking that the growth differential is not sustained in the long run.

As an indicator of international orientation, the table includes foreign sales as a

proportion of total sales. This variable is significantly greater for the cross-listing

companies in all the years considered, but particularly so after the cross-listing date. So

the data suggest that a foreign listing is more likely to be pursued by export-oriented

companies and at the same time is part of a strategy of expansion on foreign markets.

Surprisingly, the relatively high leverage of cross-listing firms increases upon cross-

listing. Before, leverage is about seven percent above that of the control group, rising to

about twelve percent in the three years after listing. This is mirrored by the marked and

permanent decline of the capitalization ratio (stockholders’ equity divided by total

assets) after the cross-listing. Thus, reducing leverage does not seem to be the main

motive for cross-listing.

There is also some weak evidence that cross-listing firms are R&D-intensive (the

ratio of R&D expense to total sales is larger in all the years around the cross-listing but

not always significantly so). They also pay significantly higher average wages in the

years before the cross-listing date and up to two years after. Thus, they seem to be skill-

intensive firms.

Trading activity on the home exchange – as measured by the number of common

shares traded divided by their total number outstanding – is larger for companies which

cross-list, both before and after the time of cross-listing. This is consistent with the

already mentioned fact that these are large companies in their home market, with

accordingly high turnover ratios. So far, the data do not appear to support the hypothesis

that cross-listing enhances home market liquidity, but it does not seem to divert trading

away from the home market either.

Finally, the market-to-book ratio of cross-listing firms is not significantly different

from that of other firms, suggesting that market timing holds little promise as a potential

explanation for cross-listing. Also the return on assets (ROA) does not differ

significantly from that of the control group around the cross-listing date, except for a

drop three years after the cross-listing and thereafter.

In Panel B of Table 5 we repeat the comparison separately for companies which

cross-list for the first time in the US and for those that do so within Europe. Compared

19

with the control group, the companies that cross-list in Europe tend to be larger than

those that cross-list in the US in terms of total assets and number of employees, both

before and after the cross-listing date. But the most visible differences between the two

groups concern their R&D intensity before cross-listing and their long-run profitability

relative to the control group. First, the companies which cross-list in the US spend more

on R&D than the control sample, using the three measures of Table 5, whereas this is

not true of the companies that cross-list within Europe. The high-tech nature of the

companies listing in the US is also mirrored by their higher labor cost per employee.

Second, the companies that cross-list in the US appear to have a larger ROA in the long-

run, compared to those that cross-list in Europe. This may reflect a different quality of

the two groups − a finding consistent with a signalling model, in the presence of higher

listing costs in the US.

5. Predicting Cross-Listing from Company Characteristics

The descriptive statistics discussed in the previous section provide some exploratory

evidence concerning the reasons why European companies list abroad. However, to

compare the explanatory power of the competing hypotheses and filter out spurious

correlations, we must turn to regression analysis. In the next subsection, we use logit

and duration analysis to investigate which company characteristics predict listing

abroad, and multinomial logit analysis to predict where they cross-list.

In Table 6 we analyze the determinants of the probability of listing abroad for the

first time in any given year. The set of determinants includes beginning-of-year values

of the logarithm of total assets, leverage, the domestic exchange’s market-to-book ratio,

the average of the three highest foreign market-to-book ratios, the previous year’s total

asset growth, proportion of sales abroad, and ROA.15 The regression also includes a

15 The foreign sales variable in our data set is missing for roughly one third of all companies. We imputethese missing values via regressions which generate predicted values of the percentage of foreign salesbased on the following regressors: the company mean value of the fraction of foreign sales (for thecompanies where at least one data point is available), the logarithm of total assets, the growth rates oftotal assets and sales, dummies for SIC codes at the 1-digit level, country of incorporation, calendar year,and the high-tech dummy. The regression results reported in Tables 6 and 7 use the data obtained withthis imputation method. Since a regional breakdown of sales may be missing more frequently incompanies with no foreign sales, we perform a robustness check via an alternative imputation methodwhereby the percentage of foreign sales is set equal to zero wherever it is missing. The estimates of thecoefficients in Table 6 and 7 are practically unaffected, and so are their estimated standard errors.

20

privatization dummy16, the “high-tech” dummy defined above, calendar year dummies

and regional origin dummies for each company: South (France, Italy and Spain), East

(Austria and Germany), North (Sweden, Belgium and Netherlands) and the default

(United Kingdom).

Since we are using repeated observations on individual companies to estimate a

logistic regression, we must take into account that the errors can be correlated across

observations referring to the same company. For this reason we use a Huber-White-

sandwich estimator of the variance-covariance matrix, which allows for dependence

within clusters of data concerning the same company.

The two most significant predictors of the decision to list abroad are the proportion

of sales abroad and the size of the company (as measured by the log of total assets). To

interpret the magnitude of their effect, observe that the estimates in the table are the

multiplicative impact of a unit increase in the independent variable on the odds ratio.

Therefore, a 1 percent point increase in the proportion of sales abroad increases the odds

ratio by a factor of 1.02, i.e. by 2 percent. A possible interpretation is that listing abroad

is partly a means of capitalizing on the reputation acquired through a presence on

foreign output markets. Conversely, companies that depend on foreign sales value the

positive publicity associated with a foreign listing – as suggested by Stoughton, Wong

and Zechner (1998). Size also raises the probability of listing abroad: the elasticity of

the odds of listing abroad with respect to total assets is 0.71 (the logarithm of 2.041).

The fact that the probability of listing abroad increases with company size suggests that

there are substantial fixed costs involved and that benefits are increasing in size: for

instance, a large company places larger demands on equity markets, thus benefiting

more from a wider shareholder base.

Several other variables are significant at the 1 percent level: the high-tech dummy

variable, the asset growth rate and the leverage ratio.

The odds of listing abroad are 2.70 times as large for high-tech companies as for

traditional companies. This agrees with the idea that high-tech companies turn to

foreign equity markets for capital because foreign investors and intermediaries know

more about the company’s business than their domestic counterparts, and thus can better

evaluate the stock.

16 This dummy equals 1 when the government makes a public offering of shares in the company.

21

There is also support for the view that companies list abroad after experiencing a

spurt in growth and investment, as found for domestic Italian initial public offerings

(IPOs) by Pagano, Panetta and Zingales (1998). Past growth of assets plays a significant

role in the regression: a 1 percent point increase in the growth rate is associated with a

0.1 percent increase in the odds of listing abroad. In addition, companies seem to list

abroad after accumulating debt, possibly the financial consequence of rapid asset

growth: a 1 percent point increase in leverage raises the odds ratio by 2 percent.

The privatization dummy is statistically significant at the 5 percent level. The odds of

listing abroad for a privatization issue are 4.41 times as large as for non-privatization

issues. Privatization issues tend to be very large, so that the depth of the international

equity market is likely to be needed to obtain a good price.

The market-to-book ratios on both the domestic and the foreign markets and the

ROA have no significant effects on the probability of listing abroad. So we do not find

evidence of companies trying to exploit “windows of opportunity” in the pricing of their

country’s stock or in the pricing on foreign stock markets, or to time their cross-listing

after an exceptional profit performance. This contrasts with the findings of the literature

on the timing of domestic IPOs (see Loughran, Ritter and Rydqvist, 1994, and Pagano,

Panetta and Zingales, 1998).

Lastly, there is evidence (at the 5% significance level) that companies from the

Southern region (France, Spain and Italy) are less likely to seek a listing abroad than

British companies: the odds of listing for a Southern company are only 0.33 times as

high as those for a British company.

Table 7 repeats the estimates using a Cox regression. This method measures the

impact of the various determinants on the hazard ratio, which in our case is the

probability of listing abroad for the first time. This method is particularly suited to the

prediction of discrete events in a panel setting. The results are very similar to those of

the logit regression of Table 6, both in magnitude and statistical significance.

We next investigate where companies cross-list. We wish to predict whether a

company is more likely to cross-list in Europe, in the US (possibly at the same time as

in Europe), or not at all. This is done in the multinomial regression shown in Table 8.17

17 The sample used to estimate this regression is about 10 percent larger than in the previous two tables,because now we also use the observations for companies which were already listed abroad in 1986 (eitherin a foreign European country or in the US). Another difference with the logit and Cox regression

22

All the regressors are lagged except for the High Tech dummy, for which no

endogeneity problems arise. As in the logit regression, standard errors are adjusted to

allow for dependence within clusters of data concerning the same company. The

estimates confirm that large size and high proportion of foreign sales are good

predictors of cross-listing – be it in the US or in Europe. High leverage and high-tech

industry classification are significant predictors of a cross-listing in US, but not in

Europe. Instead, the privatization dummy and the ROA are significant predictors only

for Europe.

Therefore, the overall picture is that a US listing is a more natural choice for

companies that have embarked on ambitious expansion programs and therefore have

accumulated a substantial amount of debt, and for high-tech companies. European stock

exchanges have instead been chosen more often by governments in the implementation

of their privatization programs. They have also been chosen by companies with a

stronger record of past profitability, but this may reflect the tighter listing requirements

of European exchanges (where only companies with some years of accounting profits

are eligible for listing) compared to US ones.

Interestingly, the companies which cross-list in the US are very unlikely to seek a

subsequent listing elsewhere in Europe, while a previous listing in Europe has no

significant effect on listing in the US.18 A previous listing in the US implies a sixfold

reduction in the relative odds ratio of cross-listing within Europe. The decision to access

US equity markets appears to be a one-way trip, which accords with the growing

imbalance in transatlantic cross-listings noted in Section 3.

The choice of cross-listing location also differs considerably by country of origin,

other factors being equal. British companies (the default regional dummy) are more

likely to list in the US than German and Austrian companies. Again, this agrees with the

greater tendency of UK companies to list in the US noted in the aggregate statistics of

Section 3.

estimates is that the observations for companies which already cross-listed within Europe are used inestimating the probability of cross-listing in the US and vice versa, since these two events are notmutually exclusive. In fact, the list of regressors used in Tables 6 and 7 is now expanded to include thenumber of previous foreign listings of each company in the other continent.18 In our sample, two thirds of the companies that cross-listed in the US had not previously cross-listedelsewhere in Europe. 88 firms that cross-listed in the US had no prior EU cross-listing, 17 had one, 9 hadtwo, 7 had three, 4 had four and 7 had five cross-listings in Europe prior to listing in the US.

23

6. Ex Post Evidence on Cross-Listed Companies

In this section we assess the effects of listing abroad on the subsequent performance

of companies. In the model to be estimated each variable is modeled as depending on

the dummies (first introduced in Table 5) to indicate the year of listing, the three years

after the listing and the subsequent years:

itp

itititiit dddfy εααααα +++++= −4

313

0210 ,

where ity is the dependent variable (e.g., the logarithm of total assets of company i at

time t), if denotes a company fixed effect, 0itd is a dummy intended to capture the

impact effect of the first cross-listing of company i, 31−itd is a dummy corresponding to

the three subsequent years and pitd captures the permanent effect of the cross-listing. To

limit the effect of influential observations, we use median regression estimation, and to

eliminate fixed effects we difference both sides of the equation, so that the specification

becomes:

itp

itititit dddy ηααα +∆+∆+∆=∆ −4

313

02 ,

where itit εη ∆≡ . In Table 9 we report the results of the estimation of this differenced

model. Very few coefficients appear to be precisely estimated. The turnover rate on the

domestic market appears to be negatively affected by the foreign listing, both upon

impact and in the three subsequent years, in contrast with the findings of Noronha, Sarin

and Saudagaran (1996) and Foerster and Karolyi (1996). The fraction of foreign sales

appears to increase upon cross-listing. Research per employee rises strongly both upon

impact and keeps increasing afterwards, witnessing the rising R&D intensity of these

companies, while the labor cost per employee changes in the opposite direction.

A possible reason why most of the estimated coefficients in Table 9 are small and

imprecise is that this specification unduly restricts the effects of cross-listing on the US

and European exchanges to be the same. This is confirmed by Table 10, where we

introduce separate dummy variables for the listings in the two continents.19

19 In Table 10 there are two cross-listing variables: one captures the first cross-listing in Europe and theother captures the first cross-listing in the US Thus, unlike in Table 9 (where only the first cross-listinganywhere was considered), in Table 10 the same company can go through two cross-listing events.

24

What happens after a cross-listing could hardly differ more sharply in the two cases.

The companies that cross-list in the US experience a 6.5 percent permanent increase in

total assets and a 4.62 percent long-run increase in the growth rate of employees. In

contrast, companies that cross-list within Europe end up with a 3.8 percent permanent

reduction of total assets and a 5 percent long-run decrease in the growth rate of sales,

relative to the control sample. The estimates for the capitalization and leverage ratios

show that the expansion of total assets for the companies which cross-list in the US is

funded by an increased amount of equity and no significant leveraging (actually some

deleveraging in the cross-listing year). The opposite is true for companies cross-listing

within Europe. The opposite changes in the capital structure of the two types of

companies are also mirrored in the opposing time patterns of the market value of their

outstanding stock after the cross-listing.20

In short, the overall picture is that cross-listings in the US are prompted by the need

to fuel rapid expansion via new equity issues, while those within Europe are at best used

to increase the debt capacity of the company and are hardly followed by rapid growth.

This striking difference is consistent with the results of Table 8, where cross-listings in

the US – but not in Europe – are shown to follow rapid expansion of the asset base.

Another difference concerns the degree of export orientation of the two group’s sales

strategies: companies cross-listed in the US experience a small (0.2 percent) increase in

the fraction of foreign sales upon impact, leveling off to 0.18 percent in subsequent

years, relative to the control group. Companies cross-listed within Europe temporarily

lose ground on foreign markets upon impact and in the subsequent three years, although

eventually they experience a foreign sales expansion comparable to the other group.

The turnover ratio on the home market also behaves differently in the two cases.

After a listing in the US the change in turnover is not statistically significant, whereas

after a cross-listing in Europe there is an 8 percent drop in the domestic turnover ratio

(significant at the 10 percent level) upon impact, followed by comparable reductions in

later years, though imprecisely estimated. This is consistent with the “time zone”

hypothesis proposed by Pulatkonak and Sofianos (1999), who show that NYSE trading

in non-US stocks tends to decrease with the time zone difference.21

20 It is less clear why the market-to-book ratio does not change appreciably for the companies that cross-list in the US but rises considerably (though not permanently) after a listing in Europe.21 In Table 9 (p. 47) of their study they show that for European, non-UK stocks the share of NYSE tradingis considerably lower than that of London trading: for the cross-listed stocks of the Netherlands, Spain,

25

R&D intensity is the only variable that appears to move in the same direction in both

cases, rising throughout the post-listing period. On a per employee basis and a percent

of sales, it increases more after European cross-listings, whereas as a percentage of total

labor expenses it rises more after US listings. But in any event the increasing reliance

on R&D that already emerges in Table 9 is confirmed by the results of Table 10.

7. Conclusions

We can now bring together the results in the two parts of this paper: the account of

the aggregate trends in the geography of listings in Europe and the US in 1986-97 and

the analysis of a panel of European companies in the same time interval. In particular, it

is worthwhile asking if our findings about the individual cross-listing decisions help us

explain the changes in the geography of equity listings.

Our aggregate figures show that the number of European companies cross-listing

their shares increased considerably, but most of the increase went to US exchanges (of

which the NYSE absorbed more than half). At the same time, the number of US

companies cross-listing in Europe fell by a third. The end result has been a decline of

foreign listings in Europe and a large increase in European listings in the US.

The decline of foreign listings on European exchanges appears to be part of a more

general decline in their ability to attract new listings. Most of them have not attracted a

large number of new domestic listings either, especially in the 1990s, with the exception

of Frankfurt and, to some extent, of London. The opposite is true of US exchanges,

where both domestic and foreign listings increased over the sample period.

Interestingly, the European countries whose companies have been more eager in

seeking foreign listings and whose exchanges have been least able to attract or retain

foreign listings are those with the highest trading costs and - with the exception of the

UK - with the lowest accounting standards and worst shareholder protection. Viceversa,

the US feature lower trading costs, tighter accounting standards and better shareholder

protection than most European countries.

Germany, Italy, France, and Sweden, the NYSE share of total trading is on average 12 percent, while theUK share is 28 percent. Being effected on a closer marketplace, a cross-listing within Europe tends to“eat” into domestic turnover much more than a listing effected in the US.

26

The microeconomic analysis of the characteristics and behavior of European

companies helps to understand better the motives of their cross-listing decisions, and

thus the reasons behind the one-way flow of cross-listings from Europe to the US. Apart

from a few common features, European companies that cross-list in Europe and in the

U.S. appear to have sharply different characteristics and performances.

The few common features are size, high foreign sales and high R&D spending before

listing abroad. The importance of size suggests that the cross-listing decision involves

non-negligible fixed costs and economies of scale, consistently with the findings of

studies of the decision to list in domestic market such as Pagano, Panetta and Zingales

(1998). The role of the fraction of foreign sales underscores the importance of

international orientation. Firms are able to take advantage of foreign investors’

familiarity with the firm and its products in order to find a market for their shares; or

conversely, the purpose of the cross-listing may be to enhance the firm’s profile and

reputation abroad so as to aid its product marketing efforts. The role of R&D spending

may indicate that firms seek a foreign listing to access a pool of investors who are more

knowledgeable about their prospects than domestic investors.

Apart from these common features, European companies that cross-list in the United

States differ considerably from those which do so within Europe. In the first case,

companies pursue a strategy of rapid expansion fuelled by high leverage before the

listing and large equity issues after the listing. They feature increasing reliance on

export markets not only before but also immediately after the listing, and tend to belong

to high-tech industries. Companies which cross-list in Europe, instead, do not grow

more than the control group and increase their leverage after the foreign listing.

Moreover, they do not rely on foreign sales to the same extent as firms cross-listing in

the US, and generally do not belong to high-tech sectors. Finally, they are more likely to

be newly privatized firms.22

Therefore, on the whole the motivation for a US listing appears to be the need of an

equity infusion by rapidly expanding, highly levered companies that plan to expand

their sales internationally and/or belong to high-tech industries. The latter finding is

consistent with Blass and Yafeh (1999), who report that Israeli and Dutch firms which

22 In addition, transatlantic listing seems to be a one-way street for European companies. A Europeancompany that is already listed abroad within Europe is not discouraged from seeking a US listing; butonce a company is cross-listed in the US, it is significantly less likely to cross-list abroad within Europe.

27

choose Nasdaq for the first listing are overwhelmingly high-tech oriented. The

motivations for cross-listing within Europe are not equally clear, but the companies that

take this route are definitely less dynamic, less outward-oriented and in more mature

sectors than those of the other group.

The contrast between these two groups is reminiscent of the contrast between

European and US companies’ domestic IPOs, documented by Pagano, Panetta and

Zingales (1998), Planell (1995), Rydqvist and Högholm (1995) and Mikkelson, Partch

and Shah (1997). These studies, respectively conducted on Italian, Spanish, Swedish

and US panel data, investigate the characteristics and behavior that distinguish

companies listing for the first time (on their domestic market) from those that decide to

stay private. In Italy, Spain and Sweden, domestic IPOs do not appear to finance

subsequent investment and growth while in the US they feature phenomenal growth.

Moreover, European IPOs are on average much older then their US counterparts.

These studies on domestic IPOs therefore suggest that in European countries the

stock market mainly caters to large, mature companies with little need to finance

investment, while the opposite is true of the United States. In the present paper we find

that this applies equally to cross-listing decisions: when it comes to cross-listing, the

most dynamic and outward-oriented European companies self-select in US exchanges.

The main remaining puzzle is why European exchanges are judged to be less attractive

by this group of companies. Probably the answer has several pieces to itself.

First, the high-tech nature of the European companies listing in the US suggests that

a key advantage of the US market is the presence of skilled analysts and institutional

investors specializing in evaluating these companies. This agrees with the finding by

Baker, Nofsinger and Weaver (1999) that listing on the NYSE induces higher analyst

coverage than listing in London. This comparative advantage of the US market may

partly reflect its sheer size, combined with the fixed costs of expertise in high-tech

industries. The costly investments in human capital required to evaluate high-tech

companies in their respective industries are worthwhile only if many such companies

are already listed, and this is true of a large continental market such as the US, but not

of European markets.

Second, as already stated, American exchanges are more liquid than most European

exchanges, and the US feature better accounting standards and shareholder rights’

28

protection than most European countries. Insofar as these comparative advantages

translate in a lower cost of equity capital, they may be particularly important to

companies who need to raise large amounts of fresh equity.

Last, but not least, the US economy has not only a large capital market but also a

huge product market − and one which has grown at a consistently higher pace than

European markets in the last decade. Therefore, it has been the natural springboard for

foreign companies with a strong export orientation, since it has allowed them to

capitalize on their product market reputation and expand their foreign sales rapidly,

possibly via acquisitions in the US.

If these are the main factors of comparative advantage of US exchanges relative to

European ones, they may attenuate gradually as the process of integration of European

capital markets proceeds. The removal of capital controls and the more homogeneous

regulatory framework of European directives is likely to lead to the birth of a truly

continental equity market and to increasing integration of markets for good and services

in Europe. If many of the factors of comparative advantage discussed above depend on

sheer market size, European companies may become less interested in cross-listing on

US exchanges. But this will not apply to companies from many non-European

countries, for which the US market is likely to retain its attraction.

29

Appendix: Data Sources and Definitions

(A) Market segments used and data sources

Stock exchange market segment(foreign companies)

market segmentcontrol group

Data sources used

Amex Foreign and CanadianIssues

- Stock exchange

Amsterdam Aandelen Buitenland Aandelen Binnenland,(excl. parallel market)

Het Financieele Dagblad;Officiele Prijscourant;Stock exchange

Brussels Premier Marche Premier Marché Stock exchangeEasdaq EASDAQ market - Financial Times 27. 11. 1997

and FT InformationFrankfurt Amtlicher Handel Amtlicher Handel Amtliches Kursblatt der

Frankfurter Wertpapier-börse,1986-1997

Milan Stock Exchange’s EquityMarket and MercatoRistretto

Stock Exchange’s EquityMarket and MercatoRistretto

Stock exchange

London Overseas Listings(excl. Ireland)[Official List]

Companies of theF.T. All Shares Index

Official price list, factbooksfrom the Financial TimesBusiness Research Centre,LSE Quarterly, RiskMeasurement Service, 1986-1997 from the LBS

Madrid Continuous and Floor Primero Mercado Stock exchangeNasdaq International Listings - Stock exchangeNYSE Non-US corporate issuers - Stock exchangeParis Premier, Second and

Nouveau MarchePremier, Second andNouveau Marche

Stock exchange

Stockholm A, O und OTC-list A, O und OTC-list Stock exchangeVienna Amtlicher Handel and

Geregelter FreiverkehrAmtlicher Handel andGeregelter Freiverkehr

Stock exchange

Note: The number of domestic companies used for Figures 1 to 3 are obtained by adjusting FIBV data onmain and parallel markets in various ways. First, since the FIBV 1986-88 figures include investmentfunds, 1986-88 figures are adjusted by the proportion of investment funds over companies in 1989.Second, we had to make a number of market-specific adjustments.For the Paris Stock Exchange the FIBV numbers before 1997 do not include the Second Marché. Wetherefore use FIBV data only for 1997, and before 1997 drawn our data from the SBF 1997 factbook.For the Frankfurt Stock Exchange we restrict ourselves to the Amtlicher Handel. We did not includeforeign listings in the Freiverkehr. This segment features an inflated number of foreign listings, since ittrades foreign firm’s shares even without the firm’s application. We were could not obtain data on theGeregelter Markt, but only very few companies in this segment would qualify to be in our sample.For the Nasdaq data before 1997, the number of domestic firms was provided by Nasdaq and the totalnumber of listings was obtained from the Nasdaq Factbook 1997. The number of foreign firms wascalculated as the difference between total and domestic listings.For the Stockholm Stock Exchange, we obtained the total number of listings and the number of foreignlistings from the factbooks 1997 and 1998. The number of domestic listings was obtained by calculatingthe difference between the two.For the Amsterdam stock exchange, the FIBV data for 1993 and 1994 are the number of shares, notcompanies. To obtain a proxy for the number of domestic companies, we calculated the proportion ofdomestic companies relative to domestic shares for 1993 by the ratio of the number of domesticcompanies (OM) to the number of domestic shares (OM) reported in the factbook "The Amsterdam StockExchange in 1993". We then multiplied the FIBV numbers by this ratio.

30

(B) Variable Definitions and Sources

Variable Source / formula Strategy used to purge thedata from measurement

errorCalendar YearDummies

The calendar year dummy for 1986 equals 1 in1986 and 0 in all other years. Analogous dummiesare defined for all calendar years from 1986 to1997.

Capitalization Ratio Global Vantage; the average of the most currenttwo years’ of stockholder’s equity divided by totalassets, multiplied by 100.

Set not available wheneversmaller than or equal tozero;Set not available ifshareholders’ equity issmaller than zero.

Common SharesOutstanding

Global Vantage; issue item; represents the netnumber of common / ordinary shares outstandingas of the company’s fiscal year-end.

Set not available wheneversmaller than or equal to zero

Common SharesTraded

Global Vantage; issue item; represents the numberof shares traded in the calendar month for an issue;December values were used.

Set not available wheneversmaller than zero

Common SharesTraded / Outstanding

Calculated from common shares traded divided bycommon shares outstanding

Countries’ PBV Morgan Stanley Capital International; yearendprice to book ratios for the countries investigated

Country Dummies Takes on the value of one for the country wherethe company is incorporated or legally registered.The information on the country of incorporation isobained from Global Vantage

Employees(in 1000)

Global Vantage; number of company workers asreported to shareholders. It is reported as anaverage number of employees by some companiesand as the number of employees at yearend byothers. These different bases of reporting are notdifferentiated.

Negative sign of number ofemployees has been changedinto a positive sign for thecompanies (name, year):Greenall Whitley, 1994Rugby Cement, 1991Spring Ram Corp PLC,1991Bluebird Toys, 1996Additionally set notavailable whenever smallerthan zero or employeesgrowth rate smaller than –99%

Employees Growth(in percent)

Calculated from employees – employees(-1)divided by employees(-1) multiplied by 100

Foreign SalesProportion(in percent)

Worldscope Set not available wheneversmaller than zero or largerthan 100

31

Variable Source / formula Strategy used to purge thedata from measurement

errorHightech SectorDummy

The dummy equals 1 for the following list of SIC-codes, and 0 otherwise:2830 drugs2833 medicinal chemicals, botanical products2834 pharmaceutical preparations2835 in vitro, in vivo diagnostics2836 biological products, ex diagnostics3570 computer and office equipment3571 electronic computers3572 computer storage devices3575 computer terminals3576 computer communication equipment3577 computer peripheral equipment3651 household audio and video equipment3660 communication equipment3661 telephone and telegraph apparatus3663 radio, tv broadcast, communication

equipment3669 communiations equipment3670 electronic components and accessories3671 electron tubes3672 printed circuit boards3674 semiconductor and related device3760 guided missiles, space vehicles3761 guided missiles, space vehicles3764 guided missiles, space vehicles propulsion3769 guided missiles, space vehicles parts3810 search, detection, naval, guided, aero

systems3812 search, detection, naval, guided, aero

systems3820 laboratory apparatus, optical, measure,

control instruments3821 laboratory apparatus and furniture3822 automatic regulating controls3823 industrial measurement instruments3826 laboratory analytical instruments3840 surgical, medical, dental instruments3841 surgical, medical instruments, apparatus4800 communications4810 telephone communications4812 radiotelephone communications4813 phone comm ex radiotelephone4820 telegraph and other mess. communication4822 telegraph and other mess. communication4830 radio, tv broadcasting stations4832 radio broadcasting stations4833 television broadcasting stations4840 cable and other pay tv services4841 cable and other pay tv services4890 communication services4899 communication services7370 cmp programming, data processing7371 computer programming service7372 prepackaged software7373 component integrated system design

32

Variable Source / formula Strategy used to purge thedata from measurement

errorIssue Market to BookRatio

Global Vantage; issue item; represents price-monthly-close divided by the result ofcommon/ordinary equity divided by commonshares outstanding (converted to the correspondingmonth). If the most recent observation forcommon shares outstanding is not available, theprior year’s value is used

Set not available wheneversmaller than or equal tozero;Set not available if totalshareholders’ equity issmaller than zero.

Issue Market Value(in billion USD)

Global Vantage; issue item; is the price-monthly-close mulitplied by common shares outstanding. Ifthe most recent observation for common sharesoutstanding (converted to the correspondingmonth) is not available, the prior year’s value isused.

Set not available wheneversmaller than or equal to zero

Labor & RelatedExpense(in million USD)

Global Vantage; represents direct payments to,and indirect payments on behalf of, all employees.

Set not available wheneversmaller than zero

Labor Cost / Employee(in 1000 USD)

Calculated as labor and related expense divided byemployees

Leverage(in percent)

Calculated from total debt divided by total assets,multiplied by 100

Set not available if largerthan 100

Privatisation Dummy Dummy is set equal to one in the year of aprivatisation (or seasoned offering) and zero inother years; the database used for assigning thevalues has been kindly provided by BernardoBortolotti, Fondazione ENI Enrico Mattei

Property PlantEquipment Growth

Calculated as property plant equipment net –property plant equipment net (-1) divided byproperty plant equipment net (-1), multiplied by100.

Property PlantEquipment Net

Global Vantage; net cost or valuation of tangiblefixed property used in the production of revenue.Calculated by Global Vantage as: Fixed Assets -Total (gross) less Depreciation, Depletion,Amortisation (Accumulated), less InvestmentGrants and Other Deductions

Set not available wheneversmaller than zero

Regional DummyNorth

Takes on the value of one if country ofincorporation is Netherlands, Sweden or Belgium,zero elsewhere

Regional DummyEast

Takes on the value of one if country ofincorporation is Germany or Austria, zeroelsewhere

Regional DummySouth

Takes on the value of one if country ofincorporation is France, Italy or Spain, zeroelsewhere

Regional Dummy UK Takes on the value of one if country ofincorporation is Great Britain, zero elsewhere

Research / LaborExpense

Calculated as research and development expensesdivided by labor and related expense

Research / Revenue(in percent)

Calculated as research and development expensesdivided by (total revenue*10)

Research andDevelopment Expenses(in million USD)

Global Vantage; represents all costs incurredrelating to development of new products orservices

Set not available wheneversmaller than zero

33

Variable Source / formula Strategy used to purge thedata from measurement

errorResearch per Employee(in 1000 USD)

Calculated by research and development expensesdivided by employees

Return on Assets(in percent)

Global Vantage; income before extraordinaryitems divided by the average of the most recenttwo years of assets-total multiplied by 100

Set not available wheneverreturn on assets is less than –100%.

SIC Codes Global Vantage; 4 digit Standard IndustryClassification code

Total Assets(in billion USD)

Global Vantage; total value of assets reported onthe Balance Sheet

Set not available whenevertotal assets is smaller than orequal to zero or total assetsgrowth is smaller than –95%.

Total Assets Growth(in percent)

Calculated from total assets – total assets(-1)divided by total assets(-1) multiplied by 100

Total Debt(in billion USD)

Global Vantage; sum of long term debt – total plusdebt in current liabilities

Set not available wheneversmaller than zero

Total Revenue(in billion USD)

Global Vantage; represents Sales/Turnover (net) Set not available if smallerthan zero

Total Revenue Growth(in percent)

Global Vantage; represents (total revenue dividedby total revenue(-1))-1, multiplied by 100.

Set not available if smallerthan–99 %

Total ShareholdersEquity

Global Vantage; represents common/ordinary andpreferred/preference shareholders’ interest in thecompany and any reserves reported in theShareholders’ Equity section

variables concerningForeign Listings

Data constructed using data sources summarizedin the appendix on market segments.The variables are defined in different ways for thedifferent methodologies used and explained in thenotes preceding the tables.

Note: Variables are always calculated using the “purged” data. The variables marked as “issue item”variables concern a selected issue of the company only. Where available, we have selected common /ordinary shares; otherwise, we have selected an issue as close as possible to common shares.

Additional adjustments:

- Several variables of Fiat 1988, DAF 1989, 1992, Heidelberger 1989 and ENI 1986-88 have been setnot available due to unrealistic magnitudes of the data in those years.

- If total revenue is actually zero or the ratio of total revenue to total assets is smaller than 0.01, thisfact is used as indicator of a holding company. All accounting variables are set not available for thesecompanies.

34

References

Ashbaugh, H., 1997. “Non-US Firms’ Accounting Standard Choices in AccessingForeign Markets,” working paper, University of Iowa.

Baker, H. Kent, John R. Nofsinger and Daniel G. Weaver, 1999. “International Cross-Listing and Visibility,” NYSE Working paper no. 99-01.

Biddle, G.C. and S.M. Saudagaran, 1989. “The Effects of Financial Disclosure Levelson Firms’ Choices among Alternative Foreign Stock Exchange Listings,” Journal ofInternational Financial Management and Accounting 1, 55-87.

Blass, A. and Y. Yafeh, 1999. “Vagabond Shoes Longing to Stray: Why Foreign FirmsList in the United States,” Hebrew University, unpublished manuscript.

Bolton, P. and E. von Thadden, 1998. “Blocks, Liquidity, and Corporate Control,”Journal of Finance 53, 1-25.

Brennan, M.J. and H.H. Cao, 1997. “International Portfolio Investment Flows,” Journalof Finance 52, 1851-1880.

Cantale, S. 1996. “The Choice of a Foreign Market as a Signal,” INSEAD workingpaper.

Chemmanur, Thomas J., and Paolo Fulghieri, 1999. “A Theory of the Going-PublicDecision,” Review of Financial Studies 12(2), 249-279.

Chemmanur, T.J. and P. Fulghieri, 1998. “Choosing an Exchange to List Equity: ATheory of Dual Listing, Listing Requirements, and Competition among Exchanges,”INSEAD, unpublished manuscript.

Domowitz, I., J. Glen and A. Madhavan, 1998. “International Cross-Listing and OrderFlow Migration: Evidence from an Emerging Market,” Journal of Finance 53, 2001-27.

Fanto, J.A. and R.S. Karmel, 1997. “A Report on the Attitudes of Foreign CompaniesRegarding a U.S. Listing,” NYSE Working Paper 97-01, forthcoming Stanford Journalof Law, Business and Finance.

Foerster, S.R. and G.A. Karolyi, 1996. “Multimarket Trading and Liquidity,” IveyBusiness School working paper.

Foerster, S.R. and G.A. Karolyi, 1999. “The Effects of Market Segmentation andInvestor Recognition on Asset Prices: Evidence from Foreign Stocks Listing in theUnited States,” Journal of Finance 54, 981-1013.

Fuerst, O., 1998. “A Theoretical Analysis of the Investor Protection RegulationsArgument for Global Listing of Stocks,” Yale School of Management, unpublishedmanuscript.

35

Gehrig, T., 1993. “An Information Based Explanation of the Domestic Bias inInternational Equity Investment,” Scandinavian Journal of Economics, 97-109.

Huddart, S., J.S. Hughes, and M. Brunnermeier, 1998. “Disclosure Requirements andStock Exchange Listing Choice in an International Context,” Working Paper, DukeUniversity.

Kadlec, G.B., and J.J. McConnell, 1994. “The Effect of Market Segmentation andIlliquidity on Asset Prices: Evidence from Exchange Listings,” Journal of Finance 49,611-636.

Karolyi, G.A., 1998. “Why Do Companies List Shares Abroad? A Survey of theEvidence and its Managerial Implications,” Financial Markets, Institutions &Instruments 7, New York University Salomon Center.

Kilka, M. and M. Weber, 1997. “Home Bias in International Stock Return Expectation,”unpublished manuscript, University of Mannheim.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., and R.W. Vishny, 1998. “Law andFinance,” Journal of Political Economy 106, 1113-1155.

Lombardo, D., and M. Pagano (1999), “Law and Equity Markets: A Simple Model,”Università di Salerno, CSEF Working Paper no. 25, and CEPR Discussion Paper no.2276, November.

Loughran, T., Ritter, J.F. and K. Rydqvist, 1994. “Initial Public Offerings: InternationalInsights,” Pacific Basin Finance Journal 2, 165--199.

Martin, Philippe and Hélène Rey (1999), “Financial Integration and Asset Returns,”CEPR Discussion Paper no. 2282, November.

Mello, A.S. and J.E. Parsons, 1996. “Going Public and the Ownership Structure of theFirm,” University of Wisconsin-Madison working paper.

Merton, R., 1987. “A Simple Model of Capital Market Equilibrium with IncompleteInformation,” Journal of Finance 42, 483-510.

Mikkelson, W.H., M. Partch and K. Shah, 1997. “Ownership and OperatingPerformance of Companies that Go Public,” Journal of Financial Economics 44, 281-308.

Miller, Darius (1999), “The Market Reaction to International Cross-listings: Evidencefrom Depository Receipts,” Journal of Financial Economics 51, 103-123.

Noronha, G., A. Sarin, and S. M. Saudagaran, 1996. “Testing for Micro-StructureEffects of International Dual Listings Using Intraday Data,” Journal of Banking andFinance, 20, 965-983.

Pagano, M., 1993. “The Flotation of Companies on the Stock Market: A CoordinationFailure Model,” European economic Review 37, 1101-1125.

36

Pagano, M., Panetta, F. and L. Zingales, 1998. “Why Do Companies Go Public? AnEmpirical Analysis,” Journal of Finance 53, 27-64.

Pagano, M. and A. Röell, 1998. “The Choice of Stock Ownership Structure: AgencyCosts, Monitoring and the Decision to Go Public,” Quarterly Journal of Economics113, 187-225.

Planell, S.B., 1995. “Determinantes y Efectos de la Salida a Bolsa en España: UnAnalisi Empirico,” working paper, Centro de Estudios Monetarios y Financieros.

Portes, Richard and Hélène Rey, 1999. “The Determinants of Cross-Border EquityFlows,” CEPR Discussion Paper no. 2225, September.

Pulatkonak, Melek and George Sofianos, 1999. “The Distribution of Global Trading inNYSE-Listed Non-U.S. Stocks,” NYSE Working Paper 99-03.

Reese, William A., Jr. and Michael S. Weinbach, 1999. “Protection of MinorityShareholder Interests, Cross-listing in the United States, and Subsequent EquityOfferings,” unpublished manuscript, University of Illinois.

Röell, A. A., 1996. “The Decision to Go Public - an Overview,” European EconomicReview, 40, 1071-1081.

Rydqvist, K. and K. Högholm, 1995. “Going Public in the 1980’s: Evidence fromSweden,” European Financial Management, 1, 287-316.

Saudagaran, S.M., 1988. “An Empirical Study of Selected Factors Influencing theDecision to List on Foreign Stock Exchanges,” Journal of International BusinessStudies, 101-127.

Saudagaran, S.M. and G.C. Biddle, 1992. “Financial Disclosure Levels and ForeignStock Exchange Listing Decisions,” Journal of International Financial Managementand Accounting 4, 106-147.

Smith, K. and G. Sofianos, 1997. “The Impact of an NYSE Listing on the GlobalTrading of Non-US Stocks,” NYSE working paper 97-02.

Stoughton, N., K.P. Wong and J. Zechner, 1998. “IPO’s and Product Quality,” workingpaper, University of Vienna.

Stulz, R.M., 1999. “Globalization of Equity Markets and the Cost of Capital,” NBERworking paper no. 7021.

Subrahmanyam, A. and S. Titman, 1999. “The Going-Public Decision and theDevelopment of Financial Markets,” Journal of Finance 54, 1045-1082.

Teoh, S.H., I. Welch and T.J. Wong, 1997a. “Earnings Management and theUnderperformance of Seasoned Equity Offerings,” unpublished manuscript, UCLA.

37

Teoh, S.H., I. Welch and T.J. Wong, 1997b. “Earnings Management and Long-RunMarket Performance of Initial Public Offerings,” unpublished manuscript, UCLA.

Tesar, L. and I. Werner, 1995. “Home Bias and High Turnover,” Journal ofInternational Money and Finance 15, 467-92.

38

Table 1

Motives for Cross-Listing and Their Empirical Implications

Hypothesis aboutmotive for cross-listing

Predicting cross-listing (ex anteevidence)

Consequences ofcross-listing (ex postevidence)

Stock Marketcharacteristics

1. Raising capitalfor investment

High leverage

High growth, P/E andreal investment

High growth, P/E andreal investment

Deep and liquid stockmarket

2. Stock sales byexistingshareholders

Newly privatizedfirms High share turnover

Deep and liquid stockmarket

3. Broadeningshareholders’base

High risk firms More foreigninvestors and highforeign turnover

Large stock market

4. Capitalizing onproduct marketreputation

High fraction offoreign sales

Stock market locatedwhere company’sforeign sales are high

5. Foreignexpertise

High-tech sector,large R&D spending

Knowledgeableinvestors and analysts

6. Commitment todisclosure &governancestandards

Low domesticregulatory standards

Decrease inownershipconcentration,no worsening inprofitability

High regulatorystandards

7. Liquidity Higher share turnover Stock market withlow spreads, lowbrokerage fees andhigh volume

8. Relativemispricing

Low domestic E/Pratio relative toforeign E/P ratio

9. Strengthen thecompany’soutput market

Higher foreign salesand profits, withoutnecessarily raisingmore capital

Market located wherecompany’s foreignsales have largegrowth potential

10. Listing costs arelow relative tobenefits

Large size Low listing feesand disclosurerequirements

39

Table 2. Number of Cross-Listings in 1986, 1991 and 1997 (End-of-Year Values)

Panel A: EU9-USA Cross-Listings Matrix

Country of origin

Stock Exchange

Nether-lands

Belgium Germany Italy UK Spain France Sweden Austria EU9 USA TotalComp.

7 12 3 14 2 38 129 1678 11 3 20 2 44 108 152

AmsterdamStock Exchange

7 10 1 11 2 31 83 11415 10 5 14 8 2 54 36 9015 9 4 17 13 1 1 60 36 96

BrusselsStock Exchange

14 8 2 11 12 1 1 49 34 8312 2 4 14 6 5 3 2 48 51 9916 4 6 21 4 10 4 9 74 58 132

FrankfurtStock Exchange

19 4 5 13 4 8 4 8 65 42 107

2 2 2ItalianStock Exchange

3 1 4 47 2 8 1 4 4 15 41 193 23410 1 11 1 4 7 13 47 159 206

LondonStock Exchange

11 2 11 4 5 14 47 111 158

3 3 3MadridStock Exchange

3 1 4 410 12 12 6 14 5 5 64 52 1169 11 15 6 24 5 5 1 76 52 128

ParisStock Exchange

8 9 13 3 17 4 5 2 61 37 980 1 1

1 2 3 1 4StockholmStock Exchange

1 1 2 4 5 94 17 1 22 3 255 21 3 1 30 4 34

ViennaStock Exchange

5 20 1 26 2 28Easdaq

6 2 3 5 2 18 2 2048 23 59 19 57 15 17 27 2 267 465 73255 24 73 23 83 13 34 23 11 339 418 757

Europeanexchanges(EU9 & Easdaq) 58 28 69 14 55 12 36 24 13 309 316 625

3 3 34 1 5 5

Amex

4 4 46 1 18 2 7 34 345 1 25 2 6 39 39

Nasdaq

17 3 1 2 55 8 10 96 964 11 1 16 166 4 26 7 3 46 46

NYSE

16 1 7 11 46 9 14 3 107 10758 23 60 19 89 16 19 34 2 320 465 78566 24 74 27 138 20 40 29 11 429 418 847

Total Listings

91 32 77 27 160 21 58 37 13 516 316 832

27 17 26 10 54 8 15 18 2 177 284 46132 15 29 11 89 9 22 15 9 231 234 465

TotalCompanies

48 24 31 19 130 10 43 21 11 337 184 521

40

[Table 2, continued]

Panel B: Summary of Transatlantic Listings

Country of originEU9-Countries U.S.

Stock Exchange Foreign Listings Foreign Companies Foreign Listings Foreign Companies267 147 465 284339 182 418 234EU9

Exchanges309 180 316 18453 5290 89U.S.

Exchanges207 206

41

[Table 2, continued]

Panel C: Listings on EU9-USA Exchanges from the Rest of World, by Country or Region of Origin

Country of origin

Stock Exchange

Australia,New Guinea,

NewZealand,Vanuatu

Canada

Centraland

EasternEurope

Centraland SouthAmerica

Israel JapanRest ofAfrica

Rest ofEurope

Rest ofAsia

SouthAfrica

WestIndies

2 13 23 1 4 1 25 8 24 1 5 2 1

AmsterdamStock Exchange

4 21 3 2 19 6 6 7 16

1 11 1 6 6 9 16 1BrusselsStock Exchange

9 1 5 4 6 182 57 12 5

6 4 1 60 23 5FrankfurtStock Exchange

3 2 56 18 6ItalianStock Exchange

18 25 1 7 3 8 8 14 22 90 419 29 1 16 3 27 7 24 15 94 15

LondonStock Exchange

14 22 14 19 2 29 6 18 50 55 5MadridStock Exchange

1 15 1 3 16 11 8 22 13 13 1 4 37 12 8 2 22 1

ParisStock Exchange

1 7 3 32 10 8 1 17 16

1 7 1Stockholm StockExchange

1 41

2 5ViennaStock Exchange

2 2Easdaq

1 234 1 5 3 1 144 1 5 2 8 1 2

Amex

1 40 4 5 4 1 312 119 11 16 16 8 2 17 910 125 8 23 15 6 1 17 9

Nasdaq

22 165 26 71 16 23 14 15 81 21 3 1 8 2 2 1 49 27 4 1 9 3 3 1 4

NYSE

15 65 2 93 6 11 21 32 1 1334 238 2 25 25 134 26 62 30 154 1953 262 4 35 32 178 26 92 32 157 32

Total Listings

56 316 18 146 84 170 20 105 103 114 30

32 198 1 21 23 81 24 44 30 102 1636 221 3 26 31 99 24 63 32 105 29

Total Companies

47 285 18 139 84 100 19 78 101 67 28

42

Table 3. Foreign Listings, Market and Country Characteristics

This table merges information on cross-listings within the EU9 and US area with market and country characteristics.Change in Cross-Listings into Market is the difference between the number of cross-listings of EU9 and UScompanies into a given market 1997 and 1986. Change in Cross-Listings Out of Market is the difference between thenumber of listings by domestic companies in other EU9 and US markets between 1997 and 1986. Net Change isdifference between Change in Cross-Listings into Market and Change in Cross Listings Out of Market. NormalizedNet Change is the ratio of Net Change to the total number of EU9 and US companies listed in 1991, multiplied by100. Accounting Standards is the rating reported by La Porta et al. (1998) on the basis of 1990 accountinginformation. Investor Protection is the Antidirector Rights Index from LaPorta et al. (1998). Market Capitalization ismeasured in billions of US dollars as of 1991 (source: International Federation of Stock Exchanges). Turnover is thevalue of share trading (main and parallel markets) as of 1991, measured in billions of US dollars (source:International Federation of Stock Exchanges). Note that comparison between exchanges is problematic due todifferent definitions and calculation methods used by different stock exchanges. T figures only count transactionsthat pass through the trading systems of an exchange, while R figures include all transactions, without making adifference between on- and off-market. Trading Cost is measured in basis points as of the 3rd quarter of 1998. It is theaverage sum of commission, fees and market impact based on trade data on all global trades done by 135institutional investors in a given market (source: Elkins/McSherry Co., Inc.).

Market

Change inCross-

Listingsinto

Market

Change inCross-

Listingsout of

Market

NetChange

Normal-izedNet

Change

Account-ing Stand-

ards

InvestorProtection

MarketCapital-ization

Turnover

TradingCosts(BasisPoints)

-53 +33 -86 -24.2 64 2 135.98 38.9 R 34.56Netherlands

-7 +9 -16 -5.7 61 0 71.11 8.2 T 33.21Belgium

+8 +17 -9 -2.2 62 1 392.47 404.6 R 29.70Germany

+4 +8 -4 -1.6 62 1 158.81 23.4 T 29.84Italy

-76 +71 -147 -6.6 78 5 986.11 553.9 R 51.88Great Britain

+4 +5 -1 -0.2 64 4 127.30 35.3 T 24.57Spain

-18 +39 -21 -2.2 69 3 373.36 116.6 T 27.63France

+8 +3 +5 +2.3 83 3 97.06 20.6 R 32.26Sweden

+3 +11 -8 -5.8 54 2 26.04 7.2 T 51.29Austria

+1 124.45 40.9 T N.A.Amex

+62 -149 +303 +4.6 71 5 490.68 693.9 R 30.64Nasdaq

+91 3484.34 1520.2 T 24.57NYSE

43

Table 4. The Global Vantage Sample

Panel A: Summary Statistics

Mean Std.Dev. Min Max Median No. of

Obs.No. ofcomp.

Total Assets 2.06 6.05 0.00 113.12 0.37 14929 1813Total Assets Growth 14.56 59.29 -92.24 3638.01 7.61 13047 1797Capitalization Ratio 40.01 17.32 0.40 99.68 38.80 12867 1787Com. Shares Traded / Outst. 322.87 6144.64 0.00 212610.40 24.13 5413 1370Employees Growth 38.75 1727.22 -98.24 136633.30 1.16 11607 1706Employees 12.39 34.41 0.00 1851.00 3.02 13533 1751Foreign Sales Percentage 34.78 29.20 0.00 100.00 33.02 11066 1273Leverage 20.58 15.07 0.00 100.00 18.95 14886 1813Issue Market to Book Ratio 385.59 4338.11 0.00 398001.30 181.48 11737 1719Issue Market Value 1.41 5.73 0.00 297.77 0.26 11926 1729Prop. Plant Equipm. Growth 113.14 9953.55 -100.00 1134913.00 6.67 13022 1793Research per Employee 14.70 141.38 0.01 3432.89 2.60 2635 483Research / Revenue 5.92 60.49 0.00 2961.89 1.72 2695 491Research / Labor Expense 16.96 240.18 0.00 3724.52 0.10 1692 313Total Revenue 1.93 5.06 0.00 71.96 0.40 14918 1812Total Revenue Growth 22.20 328.78 -98.57 31883.44 8.21 13029 1797Labor Cost / Employee 87.96 1321.30 0.00 33975.84 22.15 6458 878High Tech Dummy 0.10 0.30 0.00 1.00 0.00 21876 1823Return on Assets 4.61 8.77 -91.92 391.82 4.62 13073 1796

Panel B: Number of Companies by Country of Incorporation

Country of incorporation Total no. ofcompanies

No. ofcompaniescross-listedalready in

1986

No. ofcompaniesthat cross-list during1987-1997

Proportionof sample

with cross-listings

Austria 57 2 7 0.158Belgium 67 8 2 0.149Germany 204 16 11 0.132Spain 80 1 2 0.038France 333 15 13 0.084United Kingdom 812 28 68 0.118Italy 79 6 5 0.139Netherlands 111 16 9 0.225Sweden 80 15 4 0.238Total 1823 107 121 0.125

44

Table 5. Descriptive Statistics

Panel A: Difference in Medians between Cross-Listing Companies and Companiesthat Do Not Cross-List, by Year Relative to First Cross-Listing Date

This table reports the differences in medians between companies that do cross-list and those that do not.Columns give the differences in medians in the years -3, -2, etc. relative to the event of cross-listing. Thecontrol sample consists of companies that are not cross-listed at all during the whole sample period from1986 to 1997. Computation is implemented as median regression, where the variable of interest (e.g. totalassets) is the dependent variable, and a relative-listing-year dummy and dummies controlling for calendaryear and country of incorporation effects serve as explanatory variables; the relative-listing-year dummyfor year +n (-n) takes the value 1 for observations taken n years after (before) the year in which thecompany is first cross-listed abroad. A separate median regression is run for each cell in the table. Thevalue reported is the coefficient of the relative-listing-year dummy. Standard computation formulas (themethod of Koenker and Basett (1982) and Rogers (1993) are used for obtaining standard errors and hencelevels of significance, where *** reports significance at the 1 % level, ** 5 %, * 10 %. This computationdoes not take into account clustering effects (errors might be dependent within companies observed forsubsequent years). However, computation of the standard errors by bootstrapping does not change thesignificance levels substantially (results not reported).

-3 -2 -1 0 1 2 3 >3Total Assets 1.887*** 1.560*** 1.307*** 1.851*** 1.749*** 2.379*** 1.794*** 4.839***

Total Assets Growth 2.864 1.260 5.070** 5.231*** 4.878** 1.641 -0.771 -0.882Capitalization Ratio -2.038 -2.500 -3.458 -4.027* -5.530*** -7.404*** -8.391*** -3.890***

Com. Shares Traded / Outst. 19.150*** 11.415* 19.130*** 12.825*** 9.916** 13.586** 9.931** 16.467***

Employees Growth -1.799 2.983** -0.370 3.304*** 2.551** -0.508 -1.689 -1.734***

Employees 7.463*** 7.642*** 7.449*** 9.027*** 10.302*** 10.944*** 10.683*** 24.468***

Foreign Sales Percentage 22.140*** 20.580*** 23.400*** 22.550*** 26.810*** 36.530*** 35.700*** 29.260***

Leverage 7.035*** 7.572*** 5.797*** 9.710*** 9.289*** 12.794*** 12.786*** 4.636***

Issue Market to Book Ratio 22.081 27.253 21.025 15.178 9.899 -0.173 8.388 -6.115Issue Market Value 2.352*** 1.308*** 1.322*** 1.120*** 1.138*** 1.105*** 1.229*** 2.705***

Prop. Plant Equipm. Growth 1.070 0.713 3.683 6.244*** 4.016* 3.188 -0.490 -0.699Research per Employee 1.834*** 1.977*** 1.527*** 2.285*** 1.289*** 1.734*** 3.131*** 1.846***

Research / Revenue 0.308 1.039** 0.146 0.130 0.010 0.394 0.618 0.299***

Research / Labor Expense 0.238*** 0.029 0.104*** 0.022 0.281*** 0.299*** 0.174*** 0.106***

Total Revenue 1.485*** 1.194*** 1.364*** 1.683*** 1.784*** 2.268*** 2.461*** 4.403***

Total Revenue Growth 0.935 2.578 5.429*** 3.851** 3.030 -1.402 3.267 -1.622***

Labor Cost / Employee 3.585** 6.405*** 5.332*** 4.043*** 2.922** 4.212*** 2.014 0.272Return on Assets 0.531 -0.143 0.142 0.542 0.360 -0.834 -1.032* -0.391**

45

[Table 5, continued]

Panel B: Differences in Medians between Cross-Listing Companies and Companiesthat Do Not Cross-List, by Year Relative to First Foreign Listing Date and by

Continent of First Foreign Listing

This table reports differences in medians of companies that cross-list in the US with the control sampleand companies that cross-list within EU9 with the control sample. Only first foreign listings within thetotal geographical area (EU9 and US) are considered, then the sample is divided into sub-groupsdepending on where this first foreign listing takes place. Companies that cross-list in the same calendaryear in the US and in EU9 are excluded. For columns -3 to +3, subsequent cross-listings in the othergeographical area are ignored. The permanent effect column excludes observations for companies that arelisted abroad for more than three years in both continents. The sample period is from 1986 to 1997.Calculation is in the form of a median regression. The dependent variable is regressed on a US-relative-listing-year dummy and a EU9-relative-listing-year dummy, and dummies controlling for country ofincorporation and calendar year effects. The coefficients of the relative-listing-year dummies are thedifferences in medians. The setup of this calculation ensures that the effect of controlling for country andcalendar year effects is the same for the US and the EU9 subsample and allows a simple test for equalityof the coefficients. The P-values of these equality tests are also reported. Note that observations thatcannot be assigned to a specific subsample are not used for estimation in the relevant regression. Becauseof different samples, this table cannot be compared directly to Panel A of Table 4; in particular the >3column, where a relatively large number of observations is excluded.

US -3 -2 -1 0 1 2 3 >3Total Assets 1.770*** 1.560*** 1.307*** 1.850*** 1.993*** 2.284*** 1.480*** 3.318***

Total Assets Growth 2.042 1.944 2.316 6.184** 5.730** -1.555 -1.295 3.074**

Capitalization Ratio -2.670 -4.550 -8.056** -8.372*** -7.526** -9.123*** -10.194*** -4.643***

Com. Shares Traded / Outst. 19.283** 11.396 19.340*** 14.282*** 13.146** 8.430 8.177 18.000***

Employees Growth -1.729 0.595 -1.457 4.281*** 3.243* -0.688 -3.163 0.506Employees 5.560*** 4.348*** 3.670*** 6.287*** 8.749*** 10.466*** 10.599*** 18.288***

Foreign Sales Percentage 22.410** 20.690** 19.190*** 24.320*** 35.100*** 40.860*** 42.520*** 32.930***

Leverage 5.820* 7.572*** 7.725*** 10.955*** 13.834*** 13.812*** 12.466*** 5.541***

Issue Market to Book Ratio 1.555 48.438* 60.536** 74.191*** 33.967 88.925*** 69.582** 92.784***

Issue Market Value 1.398*** 1.269*** 1.088*** 0.960*** 1.739*** 2.210*** 1.562*** 4.600***

Prop. Plant Equipm. Growth 0.632 0.627 2.387 6.776** 6.206** 2.222 -3.556 2.074Research per Employee 2.614*** 2.790*** 1.930*** 3.398*** 2.067*** 1.754** 1.921*** 1.871***

Research / Revenue 1.562*** 2.015*** 0.317 0.295 0.321 0.394 0.593 0.588***

Research / Labor Expense 0.238*** 0.114*** 0.149*** 0.109*** 0.281*** 0.297*** 0.174*** 0.035*

Total Revenue 1.753*** 1.160*** 1.334*** 1.372*** 2.098*** 2.678*** 2.461*** 4.299***

Total Revenue Growth 2.550 -1.823 5.340** 2.055 6.666*** -1.831 4.170 0.100Labor Cost / Employee 3.585* 6.396*** 6.723*** 3.379** 0.170 5.885*** 7.676*** 6.752***

Return on Assets 0.719 -0.419 -0.095 -0.202 0.680 -1.080 -0.267 0.675*

EU9 -3 -2 -1 0 1 2 3 >3Total Assets 2.707*** 1.543*** 1.331*** 1.890*** 1.749*** 2.380*** 2.552*** 4.441***

Total Assets Growth 5.539 1.038 6.494* 5.665* 1.164 3.472 -0.351 -1.413*

Capitalization Ratio 0.134 -2.356 0.274 -1.225 -2.816 -5.464* -6.363** -2.621***

Com. Shares Traded / Outst. 30.022** 11.913 19.027** 12.179 7.028 18.763** 9.931 15.842***

Employees Growth -2.699 3.140 0.220 0.390 0.920 0.324 -1.218 -1.949***

Employees 14.157*** 12.140*** 14.310*** 14.828*** 10.302*** 10.368*** 10.819*** 22.450***

Foreign Sales Percentage 19.660* 13.900 32.650*** 19.610** 20.780** 26.630*** 29.810*** 26.660***

Leverage 8.865** 8.228** 5.704** 7.297** 7.395*** 12.097*** 12.780*** 3.366***

Issue Market to Book Ratio 27.968 -17.571 -17.382 -8.435 -27.296 -28.097 -25.659 -16.165***

Issue Market Value 2.696*** 1.950*** 1.581*** 1.147*** 0.694*** 0.713*** 1.066*** 1.888***

Prop. Plant Equipm. Growth 1.654 0.908 5.166 3.247 2.707 5.183 1.102 -1.105Research per Employee -0.100 -0.543 -0.083 -0.351 -0.683 0.450 0.098 1.086***

Research / Revenue -0.370*** -0.714 -0.672 -0.722 -0.801 -0.584 2.110*** 0.093Research / Labor Expense -0.028*** -0.039 -0.034 -0.043 -0.042 -0.047 no obs. 0.101***

Total Revenue 1.334*** 1.194*** 1.299*** 1.716*** 1.742*** 1.894*** 2.430*** 4.012***

Total Revenue Growth -7.301* 4.579 6.401** 3.861 0.843 -1.402 1.820 -2.103***

Labor Cost / Employee 2.069 4.704 -0.487 3.135 2.842 3.113 0.000 0.000Return on Assets -0.021 -0.055 0.282 0.584 -0.084 -0.759 -1.233 -0.345*

46

[Table 5, Panel B, continued]

Difference US - EU9 -3 -2 -1 0 1 2 3 >3Total Assets -0.937*** 0.017 -0.024 -0.040 0.244*** -0.096 -1.072*** -1.123***

Total Assets Growth -3.497 0.906 -4.178 0.519 4.566 -5.027 -0.944 4.487***

Capitalization Ratio -2.804 -2.194 -8.330* -7.147* -4.710 -3.659 -3.831 -2.022Com. Shares Traded / Outst. -10.739 -0.517 0.313 2.103 6.118 -10.333 -1.754 2.158Employees Growth 0.970 -2.545 -1.677 3.891 2.323 -1.012 -1.945 2.455***

Employees -8.597*** -7.792*** -10.640*** -8.541*** -1.553* 0.098 -0.220 -4.162***

Foreign Sales Percentage 2.750 6.790 -13.460 4.710 14.320 14.230 12.710 6.270Leverage -3.045 -0.656 2.021 3.658 6.439* 1.715 -0.314 2.175Issue Market to Book Ratio -26.413 66.009 77.918* 82.626** 61.263* 117.022*** 95.241** 108.949***

Issue Market Value -1.298*** -0.681*** -0.493*** -0.187** 1.045*** 1.497*** 0.496*** 2.712***

Prop. Plant Equipm. Growth -1.022 -0.281 -2.779 3.529 3.499 -2.961 -4.658 3.179*

Research per Employee 2.714** 3.333*** 2.013* 3.749*** 2.750*** 1.304 1.823* 0.785*

Research / Revenue 1.932*** 2.729** 0.989 1.017 1.122 0.978 -1.517* 0.495**

Research / Labor Expense 0.266*** 0.153** 0.183** 0.152** 0.323*** 0.344*** no obs. -0.066***

Total Revenue 0.419*** -0.034 0.035 -0.344*** 0.356*** 0.784*** 0.031 0.287***

Total Revenue Growth 9.851** -6.402 -1.061 -1.806 5.823 -0.429 2.350 2.203Labor Cost / Employee 1.516 1.692 7.210** 0.244 -2.672 2.772 7.676*** 6.752***

Return on Assets 0.740 -0.364 -0.377 -0.786 0.764 -0.321 0.966 1.020**

47

Table 6. Predicting Cross-Listing: Logit Analysis

This table reports the odds ratios of the logit estimation of the probability of a foreign listing. The listingyear dummy takes the value one in the year of the first foreign listing in the EU9 countries or in the USand zero otherwise. Subsequently to the event of cross-listing, a company is excluded from estimation.Standard errors and resulting p-values are adjusted for clustering on companies. All explanatory variablesare lagged, with the exception of the high tech dummy, where there is no danger of endogeneity. TheMean of 3 Highest Countries’ PBV is the arithmetic mean of the three highest values of the Price/Bookratio in each year within the countries of our sample. Calendar year dummies have been used in theestimation, but their coefficients are not reported for brevity.

Number of obs. 10786

χ²(20) 269.34Prob > χ² 0.000Pseudo R² 0.2029

Log Likelihood -332.685

Listing year dummy (adjusted) Odds Ratio z P>|z|Leverage (lagged) 1.022 2.290 0.022Foreign Sales Percentage (imputed, lagged) 1.020 3.449 0.001Total Assets Growth (lagged) 1.001 2.602 0.009Privatization dummy (lagged) 4.414 2.268 0.023Log of Total Assets (lagged) 2.041 6.392 0.000Return on Assets (lagged) 0.997 -0.067 0.947High Tech Dummy 2.697 2.614 0.009Mean of 3 Highest Countries’ PBV (lagged) 0.997 -0.513 0.608Domestic PBV (lagged) 1.001 0.260 0.795Regional dummy (North) 0.811 -0.562 0.574Regional dummy (South) 0.332 -2.236 0.025Regional dummy (East) 0.566 -1.236 0.216

48

Table 7. Cox Regression (Survival Time Analysis)

This table reports the hazard ratios of the Cox estimation of a foreign listing. The dependent variable isthe foreign listing dummy, which takes the value one in the year of the first foreign listing in the EU9countries or in the US and zero otherwise. Observations are excluded from the estimation sample after theevent of a foreign listing took place. Standard errors and resulting p-values are adjusted for clustering oncompanies. All explanatory variables are lagged, with the exception of the high tech dummy, where thereis no danger of endogeneity.

No. of subjects: 1642 Log likelihood -409.530No. of failures: 69 χ² (11) 238.05Time at risk: 10786 Prob > χ² 0.000

HazardRatio

z P>|z|

Leverage (lagged) 1.021 2.356 0.018Foreign Sales Percentage (imp., lagged) 1.019 3.603 0.000Total Assets Growth (lagged) 1.001 2.793 0.005Privatization dummy (lagged) 3.741 2.404 0.016Log of Total Assets (lagged) 1.993 6.575 0.000Return on Assets (lagged) 0.997 -0.077 0.939High Tech Dummy 2.480 2.543 0.011Domestic Price/Book Value (lagged) 1.001 0.271 0.786Regional dummy (North) 0.825 -0.548 0.584Regional dummy (South) 0.343 -2.285 0.022Regional dummy (East) 0.571 -1.288 0.198

49

Table 8. Predicting the Location of Cross-Listing by Multinomial Logit

This table reports the results of the multinomial logit estimation. The dependent variable is the region ofthe foreign listing. Considered are all within-region new foreign listings. Companies that have foreignlistings in both regions in a given year are excluded from the sample in the subsequent year. Twocompanies have new foreign listings in both the US and EU9 in a given year; these observations are usedin the US sample only. We control for whether a company is already cross-listed in the other continent byusing the lagged number of foreign listings as regressor. All explanatory variables are lagged, with theexception of the High Tech dummy. The mean of 3 highest EU – Domestic PBV is the differencebetween the arithmetic mean of the 3 highest EU9 Price/Book Values and the country of incorporationsPBV. US – Domestic PBV is the difference between the US and the domestic Price/Book ratio. Calendaryear dummies have been used in the estimation, but their coefficients are not reported for brevity.Standard errors and resulting p-values are adjusted for clustering on companies.

Number of obs. 11894χ² (42) 435.60Prob > χ² 0.000

Log Likelihood: -450.539 Pseudo R² 0.2112

Region of foreign listing relativerisk ratio

z P>|z|

U.S.A.Number of foreign listings (lagged) 1.117 0.931 0.352Leverage (lagged) 1.020 1.763 0.078Foreign Sales Percentage (lagged) 1.014 2.455 0.014Total Assets Growth (lagged) 1.001 2.721 0.007Privatization dummy (lagged) 1.723 0.633 0.527Log of Total Assets (lagged) 1.858 5.419 0.000Return on Assets (lagged) 0.991 -0.193 0.847High Tech Dummy 3.670 3.945 0.000US - Domestic Price/Book Value (lagged) 1.004 0.701 0.483Mean of 3 highest EU - Domestic PBV (lagged) 0.999 -0.131 0.895Domestic PBV value (lagged) 1.006 1.189 0.234Regional dummy (North) 0.734 -0.707 0.480Regional dummy (South) 0.447 -1.509 0.131Regional dummy (East) 0.255 -2.719 0.007

EuropeNumber of foreign listings (lagged) 0.160 -2.628 0.009Leverage (lagged) 1.018 1.548 0.122Foreign Sales Percentage (lagged) 1.028 3.130 0.002Total Assets Growth (lagged) 1.002 2.690 0.007Privatization dummy (lagged) 5.249 2.118 0.034Log of Total Assets (lagged) 2.505 5.284 0.000Return on Assets (lagged) 1.084 3.192 0.001High Tech Dummy 0.218 -1.303 0.192US - Domestic Price/Book Value (lagged) 0.992 -0.986 0.324Mean of 3 highest EU - Domestic PBV (lagged) 1.015 1.179 0.238Domestic PBV value (lagged) 1.002 0.157 0.875Regional dummy (North) 1.735 1.146 0.252Regional dummy (South) 0.434 -1.045 0.296Regional dummy (East) 1.931 1.186 0.236

(Outcome no cross-listing is the comparison group)

50

Table 9. Effects of Cross-Listing: Ex-Post Median Regressions

This table reports estimates of the ex-post effects of listing. Each row in the table gives the results of amedian regression, for a dependent variable (e.g. Total Assets). The explanatory variables are an impactdummy (1 in the year of cross-listing and 0 elsewhere), a three-year impact dummy (1 in the three yearssubsequent to cross-listing) and a permanent effect dummy (1 in year 4 from cross-listing and later). Wetake first differences of all variables in order to eliminate fixed effects. The following dependent variableshave been used in logarithmic form: total assets, employees, issue market value and total revenue. Aconstant and additional control dummies are included in non-differenced form: calendar year dummies inall regressions; country of incorporation dummies only in the regressions explaining total assets,employees, issue market value and total revenue. The coefficients of these variables are not reported forbrevity. Only data points from 1990 or later are used, so as to use the observations for companies alreadycross-listed in 1986 (For these companies, the permanent effect dummy is 1).

Impact effect Three yeareffect

Permanenteffect

Pseudo-R²

No. ofobs.

Total Assets 0.003 0.006 -0.016 0.139 10936

Total Assets Growth 3.673 -0.579 0.509 0.101 10089

Capitalization Ratio -0.059 -0.467 -0.527 0.001 9901

Common Shares Traded / Outstanding -4.785 -4.102* -6.256* 0.002 3941

Employees Growth 0.575 0.633 0.055 0.001 9031

Employees -0.001 0.010 -0.009 0.011 10004

Foreign Sales Percentage 0.130*** 0.110*** 0.110*** 0.001 8102

Leverage -0.129 0.239 0.004 0.003 10888

Issue Market to Book Ratio 10.820 -1.397 3.772 0.016 9147

Issue Market Value 0.078* 0.065 0.053 0.050 9331

Property Plant Equipment Growth 1.280 1.315 0.643 0.014 10074

Research per Employee 0.491*** 0.383*** 0.560*** 0.008 1956

Research / Revenue 0.051 0.066 0.096*** 0.001 2003

Research / Labor Expense 0.003 0.003 0.003 0.001 1285

Total Revenue 0.006 0.005 -0.016 0.116 10913

Total Revenue Growth -2.824 -3.281 -3.842* 0.100 10057

Labor Cost / Employee -0.760* -1.493*** -0.825** 0.022 4819

Return on Assets 0.388 0.429 0.300 0.014 10094

51

Table 10. Effect of Where Companies List: Ex-Post Median RegressionsDistinguishing Cross-Listings in Europe and US

This table reports estimates of the ex-post effects of cross-listing, distinguishing US cross-listings andEU9 cross-listings. The methodology is the same as for the estimates reported in Table 9, with thedifference that the impact, three year effect and permanent effect dummies are defined for first regionalcross-listings. Note that therefore these dummies do not sum up exactly to the dummies of Table 9, as inthat table only the first cross-listing anywhere (either in Europe or the US) is considered, while here firstcross-listings both in the US and in Europe are counted. The additional dummies used are the same as inTable 9; their coefficients are not reported for brevity.

USimpact

USthree year

effect

USpermanent

effect

EU9impact

EU9three year

effect

EU9permanent

effect

PseudoR²

No.of obs.

Total Assets 0.007 0.040 0.065** 0.009 0.001 -0.038** 0.139 10936

Total Assets Growth 0.944 -0.506 3.826 3.868 0.258 -1.785 0.102 10089

Capitalization Ratio 0.755 1.438 * 1.577* -0.167 -0.910 -0.954* 0.001 9901

Common Shares Traded / Outst. 4.774 1.520 0.099 -8.740 * -7.303 -9.413 0.002 3941

Employees Growth 0.962 2.672 4.622* -0.185 -0.994 -2.014 0.001 9031

Employees -0.002 0.028 * 0.009 0.000 0.011 -0.016 0.011 10004

Foreign Sales Percentage 0.200*** 0.180 *** 0.180*** -0.090 *** -0.140 *** 0.160*** 0.001 8102

Leverage -0.306* -0.015 0.112 0.475 ** 0.638 *** 0.010 0.003 10888

Issue Market to Book Ratio 2.013 -13.487 6.153 24.754 ** 19.029 * 2.601 0.016 9147

Issue Market Value 0.116** 0.110 * 0.126* 0.079 -0.020 -0.082* 0.050 9331

Property Plant Equipm. Growth -0.988 1.787 6.963 1.777 2.021 -2.507 0.014 10074

Research per Employee 0.056 0.201 ** 0.202** 0.659 *** 0.509 *** 0.632*** 0.008 1956

Research / Revenue 0.038 0.077 0.094* 0.057 0.082 0.112*** 0.002 2003

Research / Labor Expense 0.000 0.017 *** 0.015*** 0.002 ** -0.004 0.003 0.001 1285

Total Revenue 0.014 0.032 0.014 -0.006 -0.010 -0.030** 0.116 10913

Total Revenue Growth -2.453 0.365 3.183 -2.815 -4.665 -5.111** 0.100 10057

Labor Cost / Employee -0.605 -1.599 *** -0.613 -2.184 *** -0.064 -0.825* 0.022 4819

Return on Assets 0.521* -0.106 -0.030 0.111 0.152 0.296 0.014 10094

52

Figure 1a. History of companies Figure 1b. History of cross-listingslisted on each exchange: among sample countriesdomestic, foreign and total (domestic companies listed abroad in

EU9-US exchanges, and EU9-UScompanies on domestic exchanges)

Frankfurt Stock Exchange

0

100

200

300

400

500

600

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Total Domestic Companies Foreign Companies

Frankfurt Stock Exchange

0

25

50

75

100

125

150

175

200

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Foreign Companies (EU9+US)domestic companies listed abroad (EU9+US)

Brussels Stock Exchange

0

50

100

150

200

250

300

350

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Total Domestic Companies Foreign Companies

Brussels Stock Exchange

0

20

40

60

80

100

120

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Foreign Companies (EU9+US)domestic companies listed abroad (EU9+US)

Amsterdam Stock Exchange

0

50

100

150

200

250

300

350

400

450

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Total Domestic Companies Foreign Companies

Amsterdam Stock Exchange

0

25

50

75

100

125

150

175

200

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Foreign Companies (EU9+US)domestic companies listed abroad (EU9+US)

Italian Stock Exchange

0

50

100

150

200

250

300

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Total Domestic Companies Foreign Companies

Italian Stock Exchange

0

10

20

30

40

50

60

70

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Foreign Companies (EU9+US)domestic companies listed abroad (EU9+US)

53

Figure 1a (continued). Figure 1b (continued).

London Stock Exchange

0

500

1000

1500

2000

2500

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Total Domestic Companies Foreign Companies

London Stock Exchange

0

50

100

150

200

250

300

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Foreign Companies (EU9+US)domestic companies listed abroad (EU9+US)

Madrid Stock Exchange

0

100

200

300

400

500

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Total Domestic Companies Foreign Companies

Madrid Stock Exchange

0

10

20

30

40

50

60

70

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Foreign Companies (EU9+US)domestic companies listed abroad (EU9+US)

Paris Stock Exchange

0

250

500

750

1000

1250

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Total Domestic Companies Foreign Companies

Paris Stock Exchange

0

25

50

75

100

125

150

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Foreign Companies (EU9+US)domestic companies listed abroad (EU9+US)

Stockholm Stock Exchange

0

50

100

150

200

250

300

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Total Domestic Companies Foreign Companies

Stockholm Stock Exchange

0

10

20

30

40

50

60

70

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Foreign Companies (EU9+US)domestic companies listed abroad (EU9+US)

54

Figure 1a (continued). Figure 1b (continued).

Vienna Stock Exchange

0

50

100

150

200

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Total Domestic Companies Foreign Companies

Vienna Stock Exchange

0

10

20

30

40

50

60

70

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997Foreign Companies (EU9+US)domestic companies listed abroad (EU9+US)

Amex, Nasdaq, Nyse

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

Total Domestic Companies Foreign Companies

Amex, Nasdaq, Nyse

0

100

200

300

400

500

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997Foreign Companies (EU9+US)domestic companies listed abroad (EU9+US)

55

Figure 2. Proportion of Foreign Companies(No. of foreign companies listed on domestic exchange / total no. of companies

listed on domestic exchange)

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

Madrid Stock Exchange

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

Amsterdam Stock Exchange

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

Brussels Stock Exchange

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

Frankfurt Stock Exchange

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

Italian Stock Exchange

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

London Stock Exchange

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

Paris Stock Exchange

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

Stockholm Stock Exchange

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

Vienna Stock Exchange

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

NYSE

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

Nasdaq

0.00

0.10

0.20

0.30

0.40

0.50

0.60

86 87 88 89 90 91 92 93 94 95 96 97

Amex

56

Figure 3. Diaspora Indices(No. of domestic companies listed abroad / no. of domestic companies listed on

domestic exchange)

0.00

0.05

0.10

0.15

0.20

0.25

0.30

86 87 88 89 90 91 92 93 94 95 96 97

Madrid Stock Exchange

0.00

0.05

0.10

0.15

0.20

0.25

0.30

86 87 88 89 90 91 92 93 94 95 96 97

Amsterdam Stock Exchange

0.00

0.05

0.10

0.15

0.20

0.25

0.30

86 87 88 89 90 91 92 93 94 95 96 97

Brussels Stock Exchange

0.00

0.05

0.10

0.15

0.20

0.25

0.30

86 87 88 89 90 91 92 93 94 95 96 97

Frankfurt Stock Exchange

0.00

0.05

0.10

0.15

0.20

0.25

0.30

86 87 88 89 90 91 92 93 94 95 96 97

Italian Stock Exchange

0.00

0.05

0.10

0.15

0.20

0.25

0.30

86 87 88 89 90 91 92 93 94 95 96 97

London Stock Exchange

0.00

0.05

0.10

0.15

0.20

0.25

0.30

86 87 88 89 90 91 92 93 94 95 96 97

Paris Stock Exchange

0.00

0.05

0.10

0.15

0.20

0.25

0.30

86 87 88 89 90 91 92 93 94 95 96 97

Stockholm Stock Exchange

0.00

0.05

0.10

0.15

0.20

0.25

0.30

86 87 88 89 90 91 92 93 94 95 96 97

Vienna Stock Exchange

0.00

0.05

0.10

0.15

0.20

0.25

0.30

86 87 88 89 90 91 92 93 94 95 96 97

Amex, Nasdaq, NYSE