The Colors of Investors Money: Which Firms Attract Institutional Investors From Around the World?

Transcription

1 The Colors of Investors Money: Which Firms Attract Institutional Investors From Around the World? Miguel A. Ferreira ISCTE Business School-Lisbon CEMAF Pedro P. Matos Marshall School of Business University of Southern California This Version: February 2006 Abstract We study institutional investors stock holdings around the world using a comprehensive data set from 27 countries. Three groups of institutions based on their geographic origin (U.S., non-u.s. foreign, and domestic managers) have equal importance in the shareholder base of non-u.s. corporations. Thus, we offer a global (non-u.s. centric) view on what firm- and country-level characteristics attract investment by institutional investors. We find that all institutions reveal a strong preference for large and liquid stocks with good governance practices. There is, however, a substantial diversity between domestic and foreign institutions with respect to other firm characteristics. Foreign investors overweight stocks that are cross-listed in the U.S., members of the MSCI indexes, and globally visible through high foreign sales or analyst coverage. Domestic institutions, in contrast, seem to underweight these same stocks. The crosslisting effect is not concentrated in the holdings of ADRs as a significant increase in the holdings of local shares by foreigners is found, which sheds some light on multi-market trading and the flow-back phenomena. Our results show an important characteristic of modern international capital markets as firm and investor actions take place in inter-connected markets. Finally, we find that foreign institutional ownership has real effects as it is positively associated with higher firm valuation. JEL classification: G15, G20, G32 Keywords: Institutional Investors, Cross-listing, International Capital Markets We thank Andrew Karolyi, Michael Schill, and seminar participants at the University of Southern California for helpful comments. This research is supported by FCT/POCI Address: Complexo INDEG/ISCTE, Av. Prof. Anibal Bettencourt, Lisboa, Portugal. Phone: Fax: miguel.ferreira@iscte.pt. Address: Hoffman Hall-701, MC-1427, 701 Exposition Blvd., Ste. 701, Los Angeles, CA , USA. Phone: Fax: pmatos@marshall.usc.edu.

2 1. Introduction A key element in modern capital markets is the interplay between firms that increasingly raise capital internationally, and institutional investors that manage growing pools of assets. Many individual investors around the world are selecting mutual funds, pension funds, or retirement products offered by insurance companies and banks, as their main investment vehicle. Institutional investors managed over US$ 35 trillion of assets (stocks and bonds) in Organization for Economic Co-operation and Development (OECD) countries in 2001 (OECD (2003)), which represents a substantial share of these countries retirement savings and other wealth. The importance of institutional investors is also rapidly increasing in emerging market countries (e.g., International Monetary Fund (2004), and Khorana, Servaes, and Tufano (2005)). These professional money managers allocate a significant share of their assets in the stock market. They have become major players in their domestic stock markets, and they are more likely to invest abroad than individual investors. Most publicly-traded corporations in many countries have now institutional investors as their largest (minority) shareholders. These institutional investors could be a U.S.-based mutual fund manager, a domestic pension fund, or a global professional money manager. This paper uses a novel database on institutional investor holdings around the world to study the stock preferences of institutional investors from 27 countries. What firm characteristics attract institutional investors including domestic institutions as well as foreign (U.S. and non-u.s.) institutions? Our ultimate goal is to study the impact of institutional investors in the shareholder base of corporations. The data set contains stock-level holdings from more than 3,000 institutions (more than 22,000 funds) from 27 countries, with positions totaling US$ 6.8 trillion in more than 25,000 stocks as of December This research focus on the preferences of institutional investors when investing in non-u.s. stocks, which account for US$ 2.6 trillion of the holdings in December U.S. institutions hold, on aggregate, over US$ 0.9 trillion overseas in non-u.s. stocks. This is matched by non-u.s. institutional investors that hold US$ 0.9 trillion overseas in non-u.s. stocks, and an addi- 1

3 tional US$ 0.8 trillion in their domestic markets in local stocks. Thus, while most previous academic research has looked at U.S. investors as the primary source of capital, modern corporations worldwide are finding that three pools of professional investors (U.S., non-u.s., and domestic) have approximately equal pocket sizes. We study the revealed stock preferences of these three different institutional investor clienteles, and investigate what firm- and country-level characteristics attract these institutions as shareholders. First, we find that all institutional investors, irrespective of their geographic origin, share a preference for large, liquid, and widely-held stocks (i.e., without large controlling blockholders). Second, all institutions reveal a preference for stocks of countries with strong disclosure standards and geographically close to their home market. Third, foreign institutional investors have a strong bias for firms that are members of the Morgan Stanley Capital International (MSCI) All Country World Index and that are crosslisted in the U.S. market by the way of an American Depositary Receipt (ADR). Domestic institutions, in contrast, underweight these same stocks. Foreign and domestic institutional investors display other divergences in their stock preferences. Foreign institutions tend to avoid high dividend-paying firms, while these same firms are favored by domestic institutions. Foreign asset managers exhibit higher demand for firms with name value and foreign visibility (i.e., high foreign sales and analyst coverage). Finally, U.S. and non-u.s. foreign investors disagree on holding value versus growth stocks and on what are their favorite target markets. U.S. institutions show a clear preference for English-speaking countries and less developed markets when they decide to go abroad, while non-u.s. investors hold relatively more stocks in non-english-speaking countries and more developed markets. We conduct several robustness checks on these findings and examine in more detail stock preferences by investors from different geographical regions. Overall, there are both similarities and diversity in the revealed stock preferences of the various groups of institutional investors based on their geographic origin. The analysis is conducted on a sample of worldwide institutional investors individual stock holdings over 2

4 the period. Thus, our findings are not hampered by studying: foreign ownership in firms from a single country (Japan as in Kang and Stulz (1997) or Sweden as in Dahlquist and Robertsson (2001)); foreign ownership by investors from a single origin country (U.S. investors as in Aggarwal, Klapper, and Wysocki (2005), Ammer, Holland, Smith, and Warnock (2005), and Leuz, Lins, and Warnock (2005)); institutional holdings using country-level portfolio allocations (Chan, Covrig, and Ng (2005)); institutional holdings from just one class of institutions (mutual funds as in Chan et al. (2005) and Covrig, Lau, and Ng (2005)); institutional holdings using a single year of observations (as in Ammer et al. (2005), Chan et al. (2005), and Covrig et al. (2005)). In addition, our data set also contains U.S. stock holdings, which allows us to investigate U.S. institutions behavior at home, and match many of the stylized facts documented in Gompers and Metrick (2001) in their study of U.S. domestic institutional holdings. We find evidence of diversity between foreign and domestic investors preferences (and also between U.S. and non-u.s. investors when they invest abroad). Thus, corporations can attract foreign capital more effectively by cross-listing their shares in major financial centers in which institutional investors operate. One of the most prominent mechanisms in recent years has been to list shares in the U.S. market by the way of an ADR program. The argument is that the firm can tap into the pool of assets managed by U.S. investors who do not venture abroad because of transaction costs or unfamiliar practices, or even into the pool of assets managed by non-u.s. investors who prefer to trade in the U.S. market. The shares of cross-listed firms can potentially become more liquid and the firm can have access to external funds at a lower cost in the future. A countervailing argument, however, is that there may be very few U.S. institutional investors who cannot invest directly overseas (Financial Times (2004)). We test whether there is indeed a cross-listing effect, i.e., whether a firm is able to expand its foreign shareholder base when it cross-lists in the U.S. We find evidence of a significant increase in U.S. institutional holdings when a firm launches an ADR program. Moreover, firms are also able to capture a higher fraction of non-u.s. 3

5 foreign institutional investors. In total, foreign investors hold 5 percentage points more of the market capitalization of cross-listed firms than they would hold otherwise. We account for the potential selection bias as firmsthatdecidetocross-listcanbethetypeoffirms that foreign investors would tend to hold regardless of the cross-listing. To further investigate this endogeneity issue, we isolate the 101 firms that launched an ADR during our sample period ( ). We find evidence that these firmsexperiencean increase in their shareholder base around the time of the cross-listing in the U.S. Although foreign investors already have holdings in local shares of those firms prior to the ADR, they substantially increase their position at the time of the cross-listing in the U.S. Interestingly, however, the increase mainly accrues in local shares holdings rather than in the ADR shares directly. This finding sheds light on the ADRs flow-back phenomenon, i.e., after an initial blip in U.S. trading of ADR shares, trading moves back to the more liquid domestic exchange (Karolyi (2003) and Halling, Pagano, Randl, and Zechner (2004)). Even though trading is not retained by the U.S. exchanges, cross-listed firmsattractextraforeigninvestorsonboard (both U.S. and non-u.s.) who are making the trip to the firm s home market and invest in local shares. In a concluding analysis, we investigate whether the presence of foreign and domestic institutional ownership drives up firm s valuation and, consequently, reduces its cost of capital. Following the related literature (e.g., Lins (2003), Doidge, Karolyi, and Stulz (2004), and Durnev and Kim (2005)), we regress firm s Tobin s Q ratio on firm-, industry-, and countrylevel variables, including the fraction of shares held by foreign or domestic institutional investors as a potential determinant of firm valuation. We find that foreign institutional ownership has a significant positive impact on firm valuation, unlike domestic institutional ownership. Because institutional ownership is likely to be jointly determined with firm s Tobin s Q ratio and driven by other firm characteristics, we re-estimate the Tobin s Q and institutional ownership equations as a system of simultaneous equations. We find that the effect prevails and accordingly there exists a strong positive relation between institutional 4

6 foreign ownership and firm valuation. Our empirical results provide evidence of the real effects of a firm s shareholder base in an international capital markets setting. Our findings are consistent with market segmentation theories (Merton (1987)) that firms held by a larger investor base have lower expected returns demanded on their stocks. Other papers have analyzed how investor recognition reduces the cost of capital, whether by cross-listing or firm s foreign visibility (Foerster and Karolyi (1999) and Doidge et al. (2004)). Our results, however, go one step further and offer a direct link between foreign (institutional) shareholder presence and firms valuations. Our findings support the idea that the expansion of the foreign institutional ownership base is one of the channels by which cross-listing in the U.S. market reduces firms cost of capital. An alternative interpretation of our findings, however, is that firms that attract foreign investorscan becomeovervalued. Theriseinfirm valuations can also be evidence of price pressure effects when a foreign investor clientele buys into a stock. It is empirically difficult to distinguish between these two interpretations. Recent literature on cross-listing argues that firm valuation benefits of internationalization are transitory and dynamic (Levine and Schmukler (2005) and Sarkissian and Schill (2005)). The results here also give additional insights to the issue of whether country-level governance and firm-level governance are substitute or complementary mechanisms. Recent research (Doidge, Karolyi, and Stulz (2005) and Stulz (2005)) finds evidence that firm-level governance levels are in large part driven by country characteristics. We find, however, that foreign and domestic investors decisions are largely driven by firm characteristics (and more so, than by country-level characteristics). Institutions stocks preferences show that investors engage in firm-level analysis, and they dedicate particular attention to several firm-level governance indicators. In addition, we find that institutions are more sensitive to firm-level governance mechanisms in countries with weak country-level investor protection and quality of institutions. These findings support a substitute role between firm and country-level mechanisms, rather than a complementary role. Thus, investors track firm-level governance 5

7 indicators (besides country-level governance), and there is hope for a good firm in a bad country. The remainder the paper is organized as follows. Section 2 presents the institutional holdings data, the sample of firms, and the determinants of institutional ownership. In section 3, we conduct our tests on what firm and country characteristics attract institutional investors. We also present a detailed investigation of the cross-listing effect on institutional ownership, in particular foreign ownership. In section 4, we study the valuation effects of foreign and domestic institutional ownership. Section 5 concludes and discusses the implications of our work. 2. Data Description 2.1. Institutional Investors Holdings Data The institutional investors holdings data are drawn from the FactSet/LionShares ownership database, which is the leading information source for global institutional share ownership. FactSet/LionShares data feeds leading financial information providers such as Reuters, MSN Money, and brokers like E-Trade. Additionally, major information providers of ADRs, such as The Bank of New York ( and J.P. Morgan ( also rely on FactSet/LionShares as the source for institutional holdings of ADRs. FactSet/LionShares data sources are public filings by investors, companies, and security agencies around the world. Institutions have discretionary control over assets under management and are frequently required to publicly disclose their holdings. For securities traded on major U.S. exchanges, FactSet/LionShares gathers institutional ownership information via the mandatory 13F filings with the Securities and Exchange Commission (SEC) as well as by rolling up the sum of shares held by the individual mutual funds (N-30D filings with the SEC) managed by a particular fund management company. FactSet/LionShares also uses the rolling up method to gather ownership data for securities that are traded outside the 6

8 U.S. (i.e., shares traded in local markets). Additionally, a data collection center in France facilitates the collection of ownership data from non-u.s.-based institutions such as European and offshore mutual funds, from sources such as stock exchange announcements, data feeds, proxies, and annual reports. Finally, it uses data from the mutual fund industry directories (e.g., European Fund Industry Directory) and country-specific regulatory agencies, and a variety of other property resources. We use the historical filings of the FactSet/LionShares database. The historical coverage extends from January 2000 to December We consider all stock holdings (ordinary shares, preferred shares, ADR, GDR, and dual listings) and handle the issue of different report frequency by institutions from different countries by getting the latest holdings update at each quarter-end. The data comprises institutions located in 27 different countries (K) and stock holdings from 48 destination markets (J). 1 This data set offers a unique worldwide K J panel data (when aggregated at the country-level) for each quarter of the last five years. As of December 2004, FactSet/LionShares has holdings data on each firm 25,502 stocks, of which 18,474 are issued outside the U.S., for a total market value of US$ 6.8 trillion. The holdings are for each of 22,111 individual funds run by a total of 3,031 different institutions (such as mutual fund companies, pension funds, bank management divisions, and insurance companies). To our knowledge, this is the most comprehensive data set on institutional holdings available. While the holdings data is at the each security, each fund level, we compare the total holdings per country and year with other country-level aggregate data sources. Some official statistics aggregate all institutional investors categories (OECD (2003b)) or just the mutual fund industry segment (European Fund and Asset Management Association (2005) statistics, Chan et al. (2005), and Khorana et al. (2005)). For most countries, our holdings exceed the 1 For a group of 21 other countries (such as Argentina, Brazil, China, and Czech Republic) Lion- Shares/Factset does not have institutional holdings coverage but contains stock holdings from foreign institutions on local stocks. We keep these foreign stock positions in our tests, but the main results of the paper do not change if we restrict the sample to the 27 countries for which both institutions and stocks coverage is available. Results are available from the authors upon request. 7

9 values reported in those sources for just the mutual fund / Sicav industry segment, and are close to the OECD aggregate numbers for all institutional investor stock holdings. A detailed comparison is available from the authors. To summarize the coverage of the institutional holdings data, Table A.1 in Appendix A presents the total equity assets held by institutions domiciled in each country at the end of each of the sample years from 2000 to U.S.-based institutions are by far the largest group of professional managers of equity assets. When we detail the top five institutions by equity assets under management (i.e., the largest 13F entities) in December 2004, we find leading mutual fund families such as Fidelity, Capital Research and Management, Vanguard, and Wellington, but also the largest U.S. pension fund manager, TIAA-CREF. The Fact- Set/LionShares data also details stock holdings at the individual fund level and we find that TIAA-CREF s Stock Fund exceeds in assets the largest stock mutual fund (Vanguard s 500 Index Fund). Other countries with large institutional investors are the U.K., Germany, Canada, and France followed by other nations for a total of 27 countries for which FactSet/LionShares gathers institutional stock holdings data. Most of the leading managers in each country are well-known financial institutions. While some of countries leading fund managers are divisions of banks like Deutsche Bank s DWS for Germany, or CDC IXIS and BNP Paribas for France, in other countries the largest equity managers are public pension funds, like in Canada where the largest entity is the Canada Pension Plan or Ontario Teacher s Pension Plan, and Norway where the largest entity is the State s Petroleum Fund managed by Norges Bank. Domicile of the managing institution and of the individual fund can differ as shown by the large number of international funds, and relatively less institutions, that are domiciled in Luxembourg. FactSet/LionShares comprises holdings of domestic and foreign securities for institutions across all countries listed in Table A.1. To summarize the stock allocations by country of origin of the institution (in row) and country of destination (in column), Panel A of Table 8

10 A.2 presents the holdings data, as of December 2004, in matrix form. Institutions as a whole managed a total of US$ 6.8 trillion of equity assets, with almost US$ 2.2 trillion invested cross-border (the sum of the off-diagonal elements in the matrix), or US$ 1.9 trillion of non-u.s. foreign stock holdings (i.e., excluding the U.S. as destination market). Focusing on all non-u.s. destination markets, we find that domestic institutional investors with US$ 829 billion in market value are on equal footing to U.S. foreign institutions (US$ 944 billion) and non-u.s. foreign institutions (US$ 927 billion) see last column of Panel B of Table A.2. Thus, on aggregate, non-u.s. firmsacrosstheworldattractmoneyfromthree investor clienteles with roughly equal pocket sizes. For example, French firms (column FR in Panel A of Table A.2) attract a total of US$ 261 billion investment from institutional investors, led by French-based institutions (US$ 77 billion row FR ), followed by U.S.- based asset managers (US$ 66 billion row US ) and institutions from Germany and the U.K. that add up to another US$ 69 billion. This example illustrates how the three type of institutional investors (domestic, U.S., and non-u.s. foreign investors) are equally important as shareholders for corporations around the world. Panel C of Table A.2 shows the fraction of each country s stock market capitalization that is held by institutions. FactSet/LionShares institutional stock ownership is the greatest, as expected, in the U.S. stock market, but global institutional portfolio managers hold large fractions of stock market capitalization in countries such as Canada (25%), Sweden (25%), and Denmark (18%). However, not all shares issued by corporations can be held by institutions, as a significant fraction is closely-held by large shareholders in some countries. Correcting for the market-level percentage of closely-held shares (available from WorldScope), we compute in Panel D the investable market float per country. If we consider the percentage of market float held by institutional investors, countries such as Norway (34%), U.S. (32%), Sweden (32%), Canada (31%), Germany (19%), and France (18%) present the highest institutional ownership. Thepresenceofdomesticrelativetoforeigninstitutionsvariesacrosscountries,forexam- 9

11 ple foreigners matter more in France (5% of market float is in hands of domestic managers versus 13% in foreigners) than in the Sweden (21% in domestic versus 11% in foreigners). And when we breakdown into U.S. institutions versus non-u.s. foreign institutions, U.S. investors are relatively more present in France (5%) than in Sweden (4%). To provide a feel of the data, we take the specific cases of the largest French company (Total SA) and the largest Swedish (Ericsson Telefon AB) as of December 2004 and list their top five institutional investors: Total SA Market capitalization = US$ 138 billion Ericsson Telefon AB Market capitalization = US$ 48 billion Total institutional ownership = 30% Total institutional ownership = 33% Top five institutions (country, % held): Top five institutions (country, % held):. CDC IXIS Asset Mgt (FR, 2.2% local). Robur Fonder (SE, 2.7% local). Fidelity Mgt (U.S., 1.2% ADR). Fidelity Mgt (U.S., 1.7% ADR). Capital Research & Mgt (US, 0.9% local). Alecta Pensionsforsaking (SE, 1.7% local). Norges Bank (NO, 0.7% local). Nordea (SE, 1.5% local). Wellington (U.S., 0.7% ADR). SEB Fonder (SE, 1.3% local) This example illustrates how these companies have domestic, U.S., and foreign non-u.s. institutions among their leading shareholders. Also investors opt differently to have their holdings through local shares or ADRs Determinants of Institutional Ownership The main focus of the paper is to examine the determinants of the level of institutional investors participation in international firm s ownership. We consider both firm-level and country-level characteristics that attract or deter institutional investment as suggested by different branches of literature. 10

12 Firm-Level Characteristics Institutional Trading and Investing Strategies: A first factor determining institutional investment is institutional preferences. While this seems somewhat circular, what we mean is that there are some stylized investment policies documented in the context of U.S. markets followed by institutional investors that distinguish them from other types of investors (such as individual investors). There are reasons to expect that some of these preferences are also revealed internationally by U.S. institutions. However, when an institution invests outside of its domestic stock market it can behave differently. Furthermore, non-u.s. institutions can act differently from U.S. institutions as they operate in distinct environments. The major institutional preferences previously documented are the following: Firm Size (SIZE): Invest in large stocks. The preference for large firms is documented in Falkenstein (1996) and Gompers and Metrick (2001) who find stock market capitalization to be a major driver of the level of institutional (and mutual fund) ownership in the U.S. Dahlquist and Robertsson (2001) find similar preferences for Swedish firms. Size may be even more of a factor in international investment, because of greater concerns investors have over liquidity and transaction costs. Book-to-Market (BM): Invest in value stocks. Gompers and Metrick (2001) suggest that more sophisticated investors such as institutions can exploit the value anomaly. Thus, institutions should invest in high book-to-market stocks to earn higher excess returns. Gompers and Metrick (2001) find weak evidence of institutions favoring value stocks, however. We explore whether this investment strategy is pursued internationally. Investment Opportunities (INV OP): Invest in growing firms. The higher the real investment prospects of a firm, the more likely it is to attract institutional investors attention. A measure used in previous literature is the annual sales growth rate as in Doidge et al. (2004) and Durnev and Kim (2005). We investigate whether global 11

13 institutional investors are attracted by firms with stronger investment opportunities (and probably more need for external financing). Past Return (RET ): Chase recent outperforming stocks. Institutions are potentially momentum investors. First, there is some evidence and industry knowledge of the momentum effect in return patterns, i.e., abnormal returns can be obtained by holding stocks that have performed well in recent times (Gompers and Metrick (2001)). In addition, many observers have depicted foreign institutions as hot money chasing hot markets, while domestic investors tend to be contrarian (Grinblatt and Keloharju (2000)). We entertain this possibility by seeing whether a stock with high recent stock returns (past 12-month) attracts higher investment from foreign versus domestic institutions. Stock Market Turnover (TURN): Invest in liquid stocks. Institutions demand more liquidity in their investments than other investors because of being delegated portfolio managers. They prefer stocks that have a deeper market, where they can enter and exit easily (Gompers and Metrick (2001)). Prudent-Man Rules: Due to the fiduciary responsibility fund managers have to the ultimate owners of the assets they manage, many managers are constrained by prudent man rules that are designed to limit the risk of their investments. These can be just a set of best practice rules or actual formal investment restrictions written in management mandates. Del Guercio (1996) studies this in the U.S. context and documents that prudence considerations are more important for bank-managed funds than mutual funds. Even though not uniformly across types of funds and, certainly, across different markets (e.g., U.S. versus German funds), all funds investment policies are potentially constrained. The following are some of the rules that are expected to direct fund managers investment decisions: Dividend Yield (DY ): Invest in dividend-paying stocks. For example, many endowments in the U.S. have explicit policies to spend only their investment income which 12

14 is many times the money generated by dividends paid by stocks in fund s portfolio, instead of partial liquidations of assets. Gompers and Metrick (2001) find that, in contrast, U.S. mutual funds seem to be averse to stocks with high dividend yields. In the international context, foreign investors can have a particular dislike for high dividend-paying stocks for home-country dividend tax withholding issues (Dahlquist and Robertsson (2001) and Ammer et al. (2005)). Return-on-equity (ROE): Invest in profitable firms. Because of the oversight by endinvestors, many asset managers are under pressure to justify their choice of stock investments. One common indicator to justify an investment choice to ultimate owners offundsisastock spastprofitability. Stock Price Idiosyncratic Volatility (SIGMA): Invest in low risk stocks. Fiduciary motives can make money managers prudently avoid very risky stocks. Gompers and Metrick (2001) find conflicting evidence in the U.S. as institutions tend to prefer stocks with high volatility. Moreover, there is a literature that interprets idiosyncratic volatility as a measure of stock price efficiency (Roll (1988) and Morck, Yeung, and Yu (2000)). Moreover, high-levels of idiosyncratic volatility have been linked to good corporate investment decision-making (Durnev, Morck, and Yeung (2004)) and to a lower probability of expropriation of outside investors by insiders (Jin and Myers (2006)). In this sense, institutional investors could reveal a preference for more efficient stocks, i.e., with higher idiosyncratic volatility. MSCI Membership (MSCI): Invest in MSCI member stocks. The asset management industry is characterized by indexation, whether explicit (as in index funds) or implicit (in that funds performance is benchmarked by a market index). Thus, institutions can have a tendency to invest more in index-member stocks. The Morgan Stanley Capital International (MSCI) All Country World Index is the leading index used in international asset management. Foreign investors are likely to overweight stocks that 13

15 are members of the MSCI index. We investigate whether domestic institutions, in contrast, fill the void in non-msci stocks and thus, contrary to foreign institutions, underweight the MSCI index members. Current Firm Governance Indicators: Institutions can be particularly responsive to firm-level governance indicators. Several aspects of the control ownership structure and financial structure of firms can attract or repel institutional investors: Leverage (LEV ): Institutions invest in firms with less debt. Firms with more outstanding debt have less need for outsider oversight. As large investors, institutional investors are potentially outside monitors of managers actions (Gillan and Starks (2003)). Thus, we expect to find lower institutional ownership in firms that currently have high levels of debt, and where managers actions are monitored by major debtholders. Cash (CASH): Institutions invest in cash-rich firms. Firms with more cash are potentially more liable to value-reducing activities by managers (the agency costs of free cash flow in the Jensen (1986) sense). We explore whether institutional investors avoid or invest in firms with high levels of cash. Fraction of Closely-held Shares (CLOSE): Institutional investors avoid firms with dominant shareholders. Institutional investors tend to hold less shares of firms where one insider or a group of insiders own a large block of shares. Moreover, institutional investors can actively avoid firms with concentrated ownership because their interests, as minority shareholders, can be seconded to those of main blockholders. Leuz et al. (2005) find that U.S. investors invest less in poorly governed foreign firms (i.e., with concentrated ownership). Corporate Governance Quotient (CGQ): Institutional investors avoid firms with weak internal governance. Firms with weak governance structures and practices are more 14

16 likely to expropriate outside investors. In our empirical tests we use the comprehensive ranking from Institutional Shareholder Services (ISS), which assists institutional investors in evaluating the quality of corporate boards and of governance practices (e.g., board of directors, audit, charter and bylaw provisions, laws of the state of incorporation, executive and director compensation, qualitative factors, ownership, and director education). Presence and Visibility of the Firm in World Markets: One particular determinant of (foreign) institutional investment is investor recognition, as suggested by market segmentation theories (Merton (1987)) investors have limited information about just a subset of stocks and direct their investments to these stocks. American Depositary Receipt (ADR): Institutional investors are attracted to invest in cross-listed firms. The U.S. cross-listing decision is potentially motivated by firms efforts to increase information on their stock, spur analyst following (Lang, Lins, and Miller (2003)), and a means of attracting investment by U.S. investors (Foerster and Karolyi (1999)). We only consider exchange-listed ADRs as only these firms are required to follow U.S. GAAP and face corresponding stricter disclosure requirements. Cross-listing can be a magnet for U.S. investors but can also attract non-u.s. institutions. In contrast, we expect this cross-listing effect not to hold for domestic institutional investors which are familiar with local firms. Foreign Sales (FXSALES): Institutional investors invest in firms that sell abroad their products. A firm that conducts business abroad is more likely to have its name known among foreign investors and thus induce these investors to consider investing in the firm s stock. This argument follows from empirical evidence on the effects of familiarity in investment decisions. Analyst coverage (ANALY ST S): Institutional investors are attracted to invest in firms with high analyst coverage. The number of analysts following a stock is commonly 15

17 considered as an indicator of the firm s visibility in the market. Analyst coverage can also related to the extent to which information is incorporated into stock prices, though whether analysts contribute with firm-specific or just market-wide information is a controversial issue (e.g., Piotroski and Roulstone (2004)) Country-Level Characteristics The attractiveness of the destination country where the firm is located is likely to be a substantial factor in foreign investors decisions. We explore country-level characteristics in addition to the firm-level characteristics as determinants of institutional ownership. Several country factors also drive the volume of assets managed by domestic institutions in each market. Investor Protection: Foreign institutional investors are likely to prefer to invest in countries where their minority shareholder interests are protected (La Porta, Lopez-de-Silanes, and Shleifer (2005), and Leuz et al. (2005)). Stronger laws and regulations are also a major driver of the overall level of domestic capital markets development (La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1998)), and of the importance of domestic institutional investors. Also, Khorana et al. (2005) find that legal factors are important determinants of the size of mutual fund industry around the world. Legal (LEGAL): Institutional investors prefer to invest in firms from country with good legal conditions. We measure the strength of the country s legal environment using the product of the anti-director rights index(laportaetal.(1998))andtheruleof-law index following the literature (e.g., Durnev and Kim (2005)). Investors could avoid stocks of firms located in countries with weak legal environment and investor protection as they face a higher probability of expropriation by insiders. Common Law (COMMON): Institutional investors prefer countries with a common law legal origin. The legal tradition of the country (common versus civil law) has been 16

18 widely considered an indicator of the level of shareholder protection. Countries with common law are frequently though to provide shareholders a high degree of protection (La Porta et al. (1998)). Distance/Familiarity: Markets that are closer in terms of geographical distance or share a common language are likely to be overweighted by investors in their international investments,asshownbymutualfundcountry-levelallocationsinchanetal.(2005). Geographic Distance (DISTANCE): Institutional investors prefer to invest in firms located in countries that are geographically close. More remote countries are likely to draw less foreign investment. English Language (ENGLISH): Institutional investors prefer to invest in Englishspeaking countries. One barrier to international investment is language. English is the international language of business. Investors from English-speaking countries, such as U.K. and U.S., can be more attracted by this feature relative to other foreign investors. Size and Development of a Country s Capital Market: The size and level of development of a country s capital market can be a magnet for investment in stocks of that country (Chan et al. (2005)). Alternatively, however, investors from developed markets may prefer to invest abroad in emerging markets because of diversification benefits and growth opportunities. We consider the following proxies for the level of economic and financial development (e.g., Doidge et al. (2005)): GDP Per Capita (GDP ): Institutional investment is higher in more developed countries. Stock Market Capitalization as % of GDP (MCAP): Institutional investment is higher into countries with larger stock markets. 17

19 Table B.1 in Appendix B details the definitions and data sources for each of the variables introduced in this section Sample of Firms and Summary Statistics The firm-level financial data are drawn from Datastream (DS, stock market data) and World- Scope (WS, accounting data) for the years of 1999 through The initial sample includes all firms in the WS database excluding financial firms (SIC code between ). We merge this sample of firms with the institutional holdings data from FactSet/LionShares at the end of each calendar year using alternatively SEDOL codes (for non-u.s. firms), CUSIP codes (for U.S. firms), or ISIN codes. We sum the holdings of all institutions in a firm s stock at the end of each calendar year and divide it by the end-of-year market capitalization. Thus, our variable of interest is the fraction of shares held by institutional investors with a breakdown by domestic institutions (i.e., institutions domiciled in the same country in which the stock is issued), and foreign institutions (i.e., institutions domiciled in a country different from the one where the stock was issued). We further breakdown foreign holdings into U.S. and non-u.s. domiciled asset managers. We aggregate local shares and ADR positions per firm to have the total stock ownership regardless of the share type. If a stock in WS is not held by any institution in FactSet/LionShares then we set institutional ownership variables to zero, following Gompers and Metrick (2001). We also examine institutional positions in local and ADR positions separately in subsection 3.3 below. Our final sample includes 15,656 unique firms, for a total of 46,249 firm-years observations for which we have data for the main variables of interest (we winsorize financial ratios, such as return-on-equity and leverage, at the bottom and top 1% levels). The sample is split into non-u.s. firms, the main focus of this paper, and U.S. firms. Thesampleofnon-U.S.firms includes 10,951 unique firms, for a total of 31,382 firm-year observations. Table 1 provides summary statistics of institutional ownership variables, and firm- and country-level control variables. The average non-u.s. firm (see Panel A) in the sample has a market capitalization 18

20 of US$ 146 million, 14.6% are MSCI index members, and 3.8% are cross-listed in an U.S. exchange. About 38% of the observations are from English-speaking countries. The sample of U.S. firms (see Panel B) includes 4,705 firms, for a total of 14,867 firm-year observations. Comparing with non-u.s. firms characteristics, we can see that U.S. firms are larger, have lower book-to-market ratios, and higher trading activity. Average institutional ownership by domestic institutions is much higher for U.S. firms (15.3%) than it is for non-u.s. firms (2.7%). To recall, this summary statistics are equally-weighted averages and we know from previous studies that institutional ownership is higher for larger firms. Indeed, in the last column of Panel C of Table A.2, we can see that total institutional ownership for the sample of non-u.s. firms is 10.6%. Furthermore, as discussed in subsection 2.1, when we correct for shares that are closely-held, total institutional ownership is 15.4% (see Panel D of Table A.2). 3. Institutional Ownership and Firm and Country Characteristics This section reports the results of the cross-sectional determinants of the level of institutional ownership worldwide. The first subsection presents our main tests with respect to what firm and country-level characteristics attract institutional investors. The second subsection contains several robustness checks and extensions of our main results. The final subsection contains an analysis of the effect of U.S. cross-listing in attracting a foreign investor clientele What Attracts Foreign and Domestic Institutions? Table 2 presents the main tests on which firm- and country-level characteristics matter the most in attracting different types of institutional investors, as we discussed in Subsection 2.2. In our regressions, we focus on non-u.s. firms (Panel A of Table 2). Panel A has four subpanels for each group of investors whose fraction of total ownership we explain: all foreign 19

21 institutions (U.S. plus non-u.s. money managers), only U.S.-based institutions, only non- U.S.-based foreign investors, and domestic institutions (i.e. managers domiciled in the same country where the firm s stock is listed). For each subpanel we estimate three specifications: (1) with just firm-level variables; (2) with both firm and country-level variables; and (3) with firm-level variables and country fixed effects. The main results of Panel A (non-u.s. firms) in general support the conjectured effects on institutional preferences. We find that, on average, institutional investors around the world, whether foreign or domestic, have a preference for large firms (SIZE) with liquid shares (TURN). But the three different groups of managers also display asymmetric investment behavior in terms of other stock characteristics. U.S. institutions prefer value stocks (high BM), while non-u.s. foreign and domestic institutions prefer growth stocks. U.S. institutions are also more prone to chase stocks with recent positive stock return performance (RET ) than are non-u.s. institutions, and domestic investors seem to have a contrarian behavior. All institutions seem to load on stocks with strong profitability indicators (ROE and INV OP), reflecting some of the prudent man rules they are subject to in their investment decisions. Against these rules, however, foreign institutional investors seem to dislike highdividend paying stocks, in contrast to domestic institutions, perhaps for the taxation issues mentioned in subsection 2.1. Foreigners do not shy away from high idiosyncratic volatility stocks (SIGMA). Institutional investors seem to react to firm-level governance practices when they decide their level of stock ownership in a firm. Institutions hold smaller fractions of firms that are closely-held or with concentrated control rights (CLOSE), with high levels of debt (LEV ), andwithhighlevelsofcash(cash). Some of these results are consistent with the role of institutional investors as outside monitors advocated by the literature on investor activism (e.g., Gillan and Starks (2003)). Both U.S. and other foreign investors have a bias for companies that are members of the MSCI index (MSCI), and that have cross-listed their shares in an U.S. exchange (ADR). 20

22 The positive MSCI coefficient indicates the importance of this international benchmark for foreign investors. Thus, there is evidence that international institutional investors load on index members. The negative MSCI coefficient for domestic institutions (see Panel A.3) indicates that this investor group fills the void in non-msci stocks in their home market. The positive ADR coefficient for both U.S. and non-u.s. foreign institutional investors illustrates the positive effect exerted by the cross-listing on their investment decisions. Because of selection bias issues (i.e., firms with higher foreign ownership are more likely to cross-list), we analyze in more depth this cross-listing effect in Subsection 3.3 below. In contrast, when investing domestically (see Panel A.3), institutions do not seem to prefer firms with ADRs, and there is even evidence that they underweight these stocks. In terms of country-level variables, all institutions reveal a preference for stocks of countries with good disclosure standards (DISC) and that are geographically closer to their local market (DISTANCE). U.S. institutions show a clear preference for English-speaking countries (ENGLISH), common-lawcountries (COMMON), and less developed markets (MCAP), while non-u.s. investors make relatively more investments in non-englishspeaking countries and more developed markets when they invest abroad. These findings illustrate that these groups of institutional investors have different reasons for investing abroad. Good disclosure standards (DISC) and legal environment (LEGAL) are found to increase the presence of domestic institutional investors, but we find opposite results for investor protection with respect to attracting foreign institutions. This is inconsistent to what the law and finance literature would suggest (La Porta et al. (1998)). Our interpretation is that investors decision to go abroad balances lower shareholder protection against better investment prospects or diversification benefits. This argument explains why U.S. investors prefer emerging markets to European markets when investing overseas. OneimportantobservationfromTable2isthatourresultsshowthat firm-level characteristics have substantial explanatory power over country-level variables for foreign institutional ownership. As we can be seen by comparing the first specification (including only firm-level 21

23 variables) with the second specification (including firm- and country-level variables), the increase in R-squares of adding country-level variables is marginal in the first three subpanels (Panels A.1-A.3 for foreign ownership). Institutional investors do more than just countrylevel portfolio allocations: they engage in specific stock picking based on firm characteristics. In Panel A.4 (domestic institutional ownership in non-u.s. firms), however, country factors are particularly important to explain the cross-sectional variation. This finding is consistent with the idea that the size and development of the domestic institutional investor segment is related to the country s overall quality of institutions (Khorana et al. (2005)). Panel B of Table 2 considers the level of U.S. institutional investment in U.S. stocks. This is not at the core of our investigation but is more of a benchmarking exercise. Our results replicate previous findings by Gompers and Metrick (2001). Like these authors, we find a preference for large and liquid stocks with value orientation. As we have seen before, U.S. institutions have similar stock preferences when investing abroad. Table 3 extends the previous results on institutional investors preferences in terms of firm s foreign visibility and the quality of governance mechanisms. We address in more detail theroleoffirm s visibility by adding as explanatory variables the percentage of foreign sales (FXSALES) and the number of analysts covering a firm (ANALY ST S). We add these variables separately in the first and second specification in each panel of Table 3 because of data availability that reduces substantially the sample size. The following findings are not affected by including these variables simultaneously. We find that firm s name and visibility abroad entices more foreigners to hold more shares as shown by the positive and significant coefficients in Panels A.1-A.3. Panel A.4 for domestic institutional ownership in non-u.s. firms and Panel B for U.S. institutional investment in U.S. stocks show a differential investment behavior. There is evidence that firm s visibility indicators are not as important when institutional investors invest at home, i.e., in familiar stocks. There is even some evidence that domestic investors tend to underweight highly-visible firms. The third specification in each panel of Table 3 extends the previous results in terms 22