PORTFOLIO PREFERENCES ACROSS MARKETS: EVIDENCE FROM MUTUAL FUND OWNERSHIP

This paper is about evaluating and comparing the portfolio preferences of domestic and foreign mutual funds in developed and emerging markets over the period 1998–2007. We find that foreign and domestic mutual funds have some different preferences toward firm characteristics and firms’ information environments, and economic development affects the preferences for both types of funds. A country’s characteristics and institutions also influence mutual fund investment decisions when fund managers form their portfolio holdings. Results further show that foreign and domestic mutual funds play a monitoring role in their portfolio firms, but foreign mutual funds cannot monitor firms effectively in emerging markets.


Introduction to Literature on Ownership
Capital markets have become even more integrated and globalized over the last two decades as revealed by the mutual dependence of markets during and after the global financial crisis. This globalization reduces investment costs and constraints, which allows investors to reallocate assets easily across markets. It also enables companies to finance their projects through world capital markets. This pool of foreign capital is especially important for emerging markets because foreign capital not only increases market value, it also improves liquidity while reducing cost of capital (Ferreira and Matos, 2008;Aggarwal et al., 2011). Despite the rapid liberalization of capital markets in both developed and emerging economies, foreign financing and access to foreign capital still vary considerably across firms and countries (Ferreira and Matos, 2008;Leuz and Warnock, 2009).
Previous studies in general focused on one developed market or some emerging markets as a group to reveal portfolio preferences of foreign investors; some studies focused on all national markets (for example, see Dahlquist and Robertsson, 2001;Aggarwal, Klapper and Wysocki, 2005;Ferreira and Matos, 2008). It is still not clear whether foreign investors have different preferences for firms in developed and firms in emerging markets. Fan, Wei, and Xu (2011) argue that firms in emerging markets are structurally different from firms in developed markets.
Therefore, some factors discovered in the previous literature may have different effects on investor's choice of firms in developed versus emerging markets when attracting foreign investors. It is essential to understand what the differences are so firms in emerging markets may grow faster and more efficiently with the competition provided by foreign capital in order to catch up and compete with firms from developed markets.
This study focuses on one type of institutional investor namely mutual funds to investigate whether foreign investors have different preferences for firms in developed and emerging markets. It could document whether factors in previous studies show different effects on foreign and domestic ownership in developed and emerging markets. Mutual fund holdings are grouped into foreign and domestic portfolios. Mutual funds from locations domiciled in countries other than the country where the firm is located are classified as foreign investors. If mutual funds and firms are both located in the same country, the mutual fund ownership is classified as domestic ownership.
Second, we explore how firms' information environments affect mutual fund holding decisions. Previous studies usually document that foreign investors suffer from more information disadvantages than domestic investors (Bohn and Tesar, 1996). Thus, foreign investors in Japan and Sweden like to hold large stocks because large stocks are more visible and have more information than small stocks (Kang and Stulz, 1997;Dahlquist and Robertsson, 2001). We investigate whether foreign and domestic mutual fund ownership varies across different firmlevel information measures. Three dimensions of measure for a firm's information environment are used to proxy for the level of information asymmetry in different aspects. They are accounting standards (GAAP and IAS adoption), earnings management (accruals, earnings smoothing, and earnings correlation) and analyst variables (number of analysts covering a firm, analyst forecast errors, and analyst forecast dispersion). Results generally confirm the conjectures but with some variations.
Accounting standards and analysts play a major role on both foreign and domestic mutual fund ownership. Foreign ownership increases with good accounting standards, better analyst coverage, and lower analyst forecast dispersion. However, the effects of good accounting standards and lower analyst forecast dispersion reduce when foreign mutual funds invest in emerging markets but remain strong for domestic mutual funds in both developed and emerging markets. Earnings management seems to have little effect on the investment decisions for both foreign and domestic mutual fund holdings.
We also explore country characteristics as to whether these have an asymmetric impact on foreign and domestic mutual fund ownership. Foreign fund preference toward country characteristics is similar in both emerging and developed countries. Higher GDP and larger size of stock market have a negative effect on foreign ownership. These results are consistent with previous literature that international investors prefer small markets to diversify their portfolio holdings (Ferreira and Matos, 2008). Foreign trades, however, have a large positive impact on foreign mutual funds. This effect is indicative of the importance of visibility and familiarity for a country to attract foreign capital (Chan, Covrig and Ng, 2005). Contrary to foreign investors, a larger stock market has a positive effect but greater foreign trades have a negative impact on domestic mutual fund ownership. This negative effect of international trades is driven by emerging markets. In the developed markets, domestic fund ownership is positively associated with foreign trades.
Finally, legal environment plays important role on foreign mutual fund investment decisions.
Foreign mutual funds can only effectively monitor their holding firms in developed markets.
Previous literature documents that country's laws and disclosure quality affect foreign investors' intention to enter a market or to hold a firm's stock (Gelos and Wei, 2005;Ferreira and Matos, 2008;Leuz et al., 2009). Here, security regulations, disclosure requirements, accounting standard, and anti-self-dealing are used to proxy for the enforcement of security laws, a country's disclosure quality, the firms' financial disclosure quality, and the quality of corporate governance in the country, respectively. These factors are supposed to be positively related to foreign mutual fund ownership.
Results show that security regulations, disclosure requirements, accounting standard, and antiself-dealing are all positively related to foreign ownership in both developed and emerging markets. However, domestic mutual funds in general care more about disclosure requirements and anti-self-dealing but pay less attention to security regulations. When domestic mutual funds invest in emerging markets, all of these laws become important.
With regard to firm characteristics, existing results show that foreign mutual funds in general prefer firms with large size, lower dividend yields, higher operating performance, more cash holding, and lower closely-held ownership. They also prefer younger firms, growth stocks, member firms of MSCI index, and firms with American Depositary Receipt (ADR) issuance.
These preferences are driven by developed markets because a relatively large number of listed companies worldwide are from developed markets. Contrary to developed markets, foreign mutual funds in emerging markets focus more on firms' operating performance, such as ROE and sales growth. They also do not care about dividend yields. Although foreign mutual funds still prefer MSCI members, they have less of a preference for firms with ADR issuance. Aggarwal et al. (2005) explain that this may be due to low cross-listing in emerging markets. When pooling developed and emerging markets together, domestic mutual funds generally have preferences similar to foreign mutual funds, but they do not care about firm size and dividend yields. In addition, MSCI membership and ADR issuance actually are not important to domestic mutual fund ownership. The portfolio preferences are similar in developed markets. When comparing portfolio preferences in developed and emerging markets for domestic mutual funds, domestic mutual funds in emerging markets place more of an emphasis on dividend yields, ROE, and sales growth. Thus, the investment strategy that domestic investors adopt in emerging markets is generally firm-performance oriented.
This study therefore aims to make several contributions to the debate on ownership preferences across the world. The empirical results that link firm characteristics, information environments, and country institutions to both foreign and domestic investors expand the limited evidence on international portfolio allocation. In the literature, this study is most closely related to Dahlquist and Robertsson (2001), Aggarwal et al. (2005), and Ferreira and Matos (2008), which are examples of similar line of inquiries but with single aims. Dahlquist  This study takes a first look at foreign and domestic mutual funds' holdings across 28 countries from 1998 to 2007 by comparing portfolio preferences of foreign and domestic mutual funds in both developed and emerging markets. In addition, data contain mutual funds from 62 countries, not just from the U.S. 1 Second, this study adds to a growing body of papers that examine international portfolio allocation. Foreign investors have different portfolio preferences from domestic investors. In addition, both foreign and domestic mutual funds have different preferences toward firms in developed and emerging markets. Results also provide some additional information for researchers to solve the puzzle of home or foreign bias. The country's institutions play important role on mutual fund holding decisions. Finally, this study is one of the few studies that investigate the effective monitoring role of institutional investors in the international markets. Only domestic mutual funds play an effective monitoring role on firms in both developed and emerging markets. The reason may be that foreign investors face more information disadvantages in emerging markets than in developed markets.
The rest of the paper is organized as follows. Section 2 describes the sample and variables employed in the empirical tests, along with summary statistics. Section 3 presents the firm-level empirical results while in section 4 we evaluate the impact of country-level institutions on foreign and domestic mutual fund portfolio decisions. The final section concludes the paper.

Data and Sample Description
In this section, we briefly describe the construction of sample and report summary statistics of key variables. The main sample comprises the following data sources: worldwide mutual fund holdings from Thomson Reuters; stock returns and financial variables from Datastream and Worldscope; accounting standards from COMPUSTAT Global; and analyst information from The Thomson Reuters database contains mutual fund equity holdings information at the firm level for over 32,031 mutual funds worldwide. The portfolio holdings data include the identifications of funds and securities, and the number of shares held in a firm by each fund. 2 We require sample firms to have an assigned SEDOL code from COMPUSTAT Global from 1998 to 2007. There are 92,747 firm-year observations. However, the sample sizes in the multivariate analyses vary and are lower due to missing values for some variables. To classify fund holdings into domestic and foreign holdings, we check whether the funds and the holding firms are from the same countries. If both fund and firm are from the same country, the fund holdings are classified as domestic ownership; otherwise, foreign ownership applies.
We further split national markets into developed and emerging markets. There are 28 countries, 28,602 foreign mutual funds, 23,373 domestic mutual funds, and hence 92,747 firmyears in our sample over 1998 to 2007. The large number of worldwide mutual funds offers a common platform to analyze the investment decisions of a homogenous group of institutional investors who share the same investment savvy. Their investment decisions should offer the subtle yet important attitude that mutual funds have towards firms in developed and emerging markets. We first investigate the overall portfolio preference of domestic and foreign mutual funds, followed by the portfolio preference of domestic and foreign mutual funds for each of developed and emerging markets. Portfolio preferences include firm characteristics, firms' information environments, and country-level institutions. We also test whether foreign and domestic investors play an effective monitoring role on firms. Analyses provide several new and interesting results.
Drawing from the existing literature, firm characteristics in our regressions include: firm size (Size), book-to-market ratio (BM), stock return (Ret 12 ), stock return volatility (σ Ret ), dividend yield (DY), return on equity (ROE), a dummy variable (MSCI) that takes the value of one if the firm is included in the MSCI index for a given year, debt to asset ratio (DA), cash holdings (CASH), closely-held ownership (CHold), sales growth (gSales), a dummy indicator (ADR) that equals one if a firm has issued American Depositary Receipts, and firm age (Age). Size is the log of market capitalization at year-end. BM is the log of book-to-market equity ratio in June of year t. Ret 12 is the average of monthly stock returns over year t-1. σ Ret is the standard deviation of monthly stock returns over year t-1. DA is the ratio of long-term debt to total assets. CASH is the ratio of cash and short-term investments to total assets. CHold is the fraction of shares closely held by insiders and controlling shareholders in year t. gSales is the average of sales growth for years t-2 and t-1.
To assess the effect of firm-level information environment on foreign mutual fund ownership, firms' information measures include accounting standards (AccStd), analyst coverage Accrual is the magnitude of accruals. We follow Lai, Ng and Zhang (2012) to calculate the variable of Accrual. Smooth is the smoothness of the accounting reports and is the ratio of the standard deviation of operating income to the standard deviation of cash flows over the last five years. Corr is the correlation coefficient between change of accruals and change of cash flows over the last five years. When calculating Smooth and Corr, a minimum of three years is required.
Country characteristics included in the analyses are GDP per capita (GDP), stock market capitalization to GDP (MCAP), GDP growth (gGDP), trade to GDP ratio (TGDP), GDP growth volatility (σGDP), and stock market volatility (σMKT). All variables except stock market volatility are taken from the World Development Indicators. To access country-level information on disclosure or institutions, we implement indexes from La Porta et al. (1998Porta et al. ( , 2006. These indicators include security regulations, the accounting standard index, and the anti-self-dealing index. The index for disclosure requirements is from World Economic Forum. Median of the index scores across all countries is used to classify countries. Countries with score above the median are considered strongly regulated countries; otherwise they are considered weakly regulated countries. Table 1: Country-level sample distribution and descriptive statistics This table reports the time-series averages of country characteristics for 28 countries from 1998 to 2007, together with the number of firm-year observations (N(FYear)) covered in the sample and the fraction of shares held by foreign mutual funds (MF) in %. Country characteristics include GDP per capita (GDP), stock market capitalization to GDP (MCAP), GDP growth (gGDP) in year t-1, trade to GDP ratio (TGDP) in year t, GDP growth volatility (σGDP) over the last five years, and daily stock market volatility (σMKT) over year t-1.   FYear)) covered in the sample. Firm characteristics include the fraction of shares held by foreign mutual funds (MF) in %, log of market capitalization (Size), log of book-to-market ratio (BM), average of past one-year monthly stock returns (Ret 12 ), standard deviation of past one year monthly stock returns (σ Ret ), dividend yield (DY), return on equity ratio (ROE), MSCI membership (MSCI), debt to asset ratio (DA), cash holdings (CASH), closely-held ownership (CHold), sales growth (gSales), ADR issuance (ADR), and firm age (AGE).  They also pay dividends and have a sales growth rate of 15 per cent. In an unreported table, the correlations between variables in the sample are moderately low, suggesting that the problem of multicollinearity should be minor.

Determinants of Mutual Fund Holdings at the Firm Level
Previous literature shows that U.S.-based foreign investors usually prefer firms with better information disclosure or firms that are more visible to foreigners such as large firms, firms with ADR issuance, and MSCI member countries. One explanation is that foreign investors suffer from an information disadvantage. Therefore, foreign and domestic investors should have some different portfolio preferences if foreign and domestic investors possess different levels of private information about local firms. In this section, we report findings on firm-level determinants of portfolio preferences across all countries for foreign and domestic mutual funds.
Following this are the firm characteristics for each developed and emerging market for both foreign and domestic mutual funds. Determinants include firm characteristics and information environment.

What Attracts Foreign and Domestic Mutual Funds?
We first present results on which firm characteristics mutual fund investors care about the most. In this test, we control for country fixed effects and year fixed effects, and all associated tstatistics are calculated based on standard errors adjusted for heteroskedasticity and firm-level clustering. Therefore, firm characteristics across countries in the regressions should be comparable. Table 3 is a summary of results. In the first column of left panel, the foreign mutual fund that pay more dividends. It is interesting to note that ADR issuance has almost no effect on foreign fund holdings in the emerging markets. There are two possible explanations: (i) the number of ADR listings may be low in the emerging markets (Aggarwal et al., 2005), and (ii) the data include all mutual funds worldwide, and ADRs have less effect on foreign mutual funds from countries other than the U.S.

Information Disclosure Quality and Mutual Fund Holding Preference
Next set of results are on disclosure affects in both foreign and domestic mutual fund holdings in developed and emerging markets. In this test, three dimensions of information disclosure are used -accounting standards, analyst variables, and earnings management -to measure a firm's information environment.
To conduct the test, a dummy is set to one if the firm uses IAS or GAAP as its accounting standard. In the accounting literature, both accounting standards and auditor quality matter most for the quality of financial reports and foreign investors (see Bradshaw, Bushee, and Miller, 2004;Francis, 2004;Covrig, DeFond, and Hung, 2007). However, as Francis and Wang (2008) document, COMPUSTAT Global does not give proper auditor identification for Japan, South Korea, India, and Pakistan. In addition, there is no precise way to determine the auditors' identification in these four countries. Therefore, this study focuses on accounting standards to measure the reliability, consistency and comparability of accounting information for a firm. To  Table 4 reports the results. It is obvious that accounting standards and information provided by analysts both play a major role on mutual funds investment decisions. When a firm adopts better accounting standards, foreign and domestic mutual funds will increase their holdings in the firm by 0.76 per cent and 0.54 per cent, respectively. This is economically significant. From Table 2, the average foreign mutual fund holding is 2.57 per cent. Thus, an increase of 0.76 per cent for foreign shares is representative of 30 per cent increase in ownership. The effect of better accounting standards on foreign mutual fund ownership is stronger in developed markets but reduces in emerging markets. This indicates that emerging markets lack enforcement, which weakens the link between better accounting standards and foreign holdings (Francis and Wang, 2008). However, better accounting standards have a more significant impact on domestic mutual fund investment decisions in emerging markets than in developed markets. This may imply that domestic investors have an information advantage over foreign investors. Therefore, the lack of enforcement does not affect domestic mutual funds.
Among analyst variables, the number of analysts and analyst forecast dispersion are positively and negatively associated with the investment decisions of both foreign and domestic mutual funds, respectively. These two variables are significant at conventional levels. However, the significance of analyst forecast dispersion in emerging markets reduces for foreign investors.
Analysts' forecast errors have the least effect on investment decisions for domestic and foreign mutual funds. This may show that visibility is more important than information quality for mutual funds because higher analyst coverage indicates that the firm is more visible and provides more information to investors. It is also possible that investors have difficulty judging the accuracy of information so the correlation between fund ownership and forecast errors is not significant. Among all three earnings management measures, none of them has a significant effect on foreign or domestic holdings. These results are consistent between foreign and domestic mutual funds.

Determinants of Mutual Fund Holdings at the Country Level
This section is on how country characteristics affect foreign and domestic mutual fund ownership.
Previous literature documents that a country's information disclosure quality and its institutions both affect foreign investors' intention to enter the stock market. For example, Leuz, Nanda, and Wysocki (2003) show that information quality is poor for firms in countries with weak investor protection. Gelos and Wei (2005) and Leuz et al. (2009) show that low country information transparency or information disclosure quality reduces foreign investment and is likely to exacerbate the degree of information asymmetry faced by foreign investors. Therefore, foreign mutual funds' ownership varies with the countries' institutions and information disclosure quality.
However, it is not clear whether such country characteristics affect domestic mutual funds' investment decisions. This study first examines how a country's characteristics affect both foreign and domestic mutual fund ownership. Then, we test how mutual funds' ownership varies with a country's information disclosure. Lastly, we test whether foreign or domestic mutual funds play an effective role on monitoring.

Country Characteristics and Mutual Fund Holdings
Here, country characteristics included in the test are a dummy variable to indicate emerging market, GDP, stock market capitalization to GDP, GDP growth, ratio of trade (exports plus imports) to GDP, standard deviation of annual GDP growth over the past five years, and the annualized standard deviation of daily stock market returns over the past one year. The dependent variable is mutual fund ownership in percent at the firm level. Control variables include all firm-level characteristics used in Table 3. t-statistics are reported with clustered firmlevel heteroskedasticity-adjusted standard errors. Table 5 reports the results.
From the left panel of Table 5, it is clear that foreign mutual funds invest less in emerging markets. If all conditions are equal, foreign investors reduce their ownership by an average of 1.40% in emerging markets when compared with developed markets. All variables except trade to GDP and stock market volatility are negatively related to foreign ownership. Thus, foreign investors actually invest less in big countries and this is caused by the diversification needs for foreign investors (Ferreira and Matos, 2008). International trading seems to be an important factor for attracting foreign capital. If a country has a stronger connection with other countries, foreign investors tend to invest more in this country. The portfolio preference is similar for foreign investors in both developed and emerging markets except when considering GDP, GDP growth volatility, and stock market volatility. The coefficients of GDP, GDP growth volatility, and stock market volatility become insignificant in emerging markets for foreign investors.
For domestic mutual funds (right panel of Table 5), there is no difference in fund ownership between developed and emerging markets. Contrary to foreign investors, domestic mutual funds' ownership is positively related to market size and negatively related to trade to This table reports regression results of foreign and domestic mutual fund ownership on various country characteristics. Country variables include EmeD that is equal to one if the country is an emerging country, GDP per capita (GDP), stock market capitalization to GDP (MCAP), GDP growth, ratio of exports plus imports to GDP (TGDP), standard deviation of annual GDP growth over past five years (σGDP), and the annualized standard deviation of daily stock market returns over year t-1 (σMKT). Control variables include firm-level variables used in Table 3 GDP. It is reasonable that the size of equity market is positively related to domestic fund ownership because market size in general is associated with local people's wealth. Since foreign ownership is positively related to trade to GDP, the negative correlation between domestic ownership and trade to GDP indicates that well-known countries attract more foreign investors and this causes domestic investors to have a small fraction of ownership. The finding from emerging markets supports this argument because firms in emerging markets on average are smaller than firms in developed markets. An increased foreign ownership in emerging markets will lead to a significant drop in domestic ownership.

Country Institutions and Mutual Fund Holdings
This subsection provides results on how country-level institutions affect investors' decision making when they form their portfolio holdings. Previous literature documents that countries' laws and disclosure quality affect foreign investors' intention to enter a market or to hold a firm's stock (Gelos and Wei, 2005;Ferreira and Matos, 2008;Leuz et al., 2009). Here, we use security regulation to proxy for the enforcement of security laws, disclosure requirement to proxy for a country's disclosure quality, accounting standard to proxy for the countries' financial disclosure quality, and anti-self-dealing to proxy for the quality of corporate governance in the country.
These factors are supposed to be positively related to foreign mutual fund ownership. However, the effect of these factors on mutual fund ownership is still not clear when markets are further split into developed and emerging markets, and funds are grouped into foreign and domestic mutual funds. requirements and anti-self-dealing are still important for domestic mutual funds. Since domestic mutual funds know local firms better, information asymmetry is less of a problem when they This table reports regression results of mutual fund ownership on various measures for country-level information disclosure. The dependent variable is the foreign or domestic mutual fund ownership in all markets, developed markets, and emerging markets. The measures for information disclosure quality at the country level include dummies for security regulations, disclosure requirements, accounting standards index, and anti-self-dealing index. See La Porta et al. (2006) for the description of the indexes. The dummy for each country disclosure measure is equal to one if the country has an index score above the median of all countries. Control variables include country variables used in Table 5 and firm variables used in Table 3  They also pay less attention to anti-self-dealing. It is worth noting that all institutions play a significant role on the domestic investors' decision making in the emerging markets. Although domestic investors possess more information of local firms, they still value the quality of laws and information disclosure at the country level. Therefore, good institutions not only attract more foreign capital, but also stimulate local people to invest in emerging market.

Governance Role of Mutual Funds
Ferreira and Matos (2008)    from shareholders and to deter shareholders from playing monitoring role. It is also widely documented that good corporate governance is enforced through lower cash holdings (see e.g., Yun, 2009). If this is the case, cash holdings should be negatively related to firm performance. As shown in Table 7, the coefficient of cash holdings is negative, except emerging markets. In the emerging markets, cash holdings have a positive effect on firm performance, implying that cash holdings may play a different role in developed and emerging markets.

Conclusions
This study investigates the portfolio preferences of foreign and domestic mutual funds in developed and emerging markets. The sample covers 28 countries, 28,602 foreign mutual funds, 23,373 domestic mutual funds, and 92,747 firm-years from 1998 to 2007. Previous studies in general only use the sample from developed or emerging markets with one year of data. They also focus on foreign investors to explore the preferences of asset allocation. It is still unclear whether previous findings apply to both domestic and foreign investors and remain evident in either developed or emerging markets. The maintained hypothesis is that firms in different markets have different structures. Since foreign investors have information disadvantage in local markets, foreign and domestic investors should have different preferences when they form their portfolio holdings. This study uses both domestic and foreign mutual funds in developed and emerging markets to first investigate the overall portfolio preferences and then their portfolio preferences in each market.
Results show that foreign and domestic investors have some different portfolio preferences, and these preferences vary between developed and emerging markets. Foreign mutual funds in the developed markets prefer firms with large size, lower dividend yields, higher ROE, more cash holdings, and fewer closely-held shares. They also like young growth stocks, firms listed in the MSCI index, and firms that have ADR issuance. In emerging markets, they focus more on firms' operating performance, such as ROE and sales growth, but less on ADR issuance.
Contrary to foreign mutual funds, domestic mutual funds overall do not care about MSCI membership and ADR issuance. Firm size and dividend yields also play a small role in domestic mutual funds' portfolio holdings. Such findings are consistent in developed markets. In emerging markets, domestic mutual funds place a greater emphasis on dividend yields, ROE, and sales growth.
Firms' information environments also affect both foreign and domestic mutual funds' portfolio preferences. Three dimensions of measure -accounting standards, earnings management, and analyst variables -are utilized to proxy for the level of the firms' information asymmetry. Results indicate that accounting standards and analyst-related measures are significantly related to fund ownership for both foreign and domestic mutual funds. However, the positive effect of strong accounting standards and lower analyst forecast dispersion reduces for foreign mutual funds in emerging markets but remains significant for domestic mutual funds.
Among all measures, the number of analysts covering a firm is the most important.
Results also would suggests that foreign mutual funds obviously invest less in emerging markets. Higher GDP, larger equity market size, and higher GDP growth exhibit negative effects