Mutual Fund Industry Competition and Concentration: International Evidence

Transcription

1 Mutual Fund Industry Competition and Concentration: International Evidence Miguel A. Ferreira Universidade Nova de Lisboa Faculdade de Economia Av. Marques da Fronteira, Lisbon, Portugal Sofia B. Ramos ISCTE Business School Av. Forças Armadas Edifício ISCTE Lisbon, Portugal Abstract: This paper examines mutual fund industry competition and concentration in 27 countries using a sample of almost 50,000 mutual funds. The mutual fund industry is concentrated worldwide and some industries present large fund complexes. Our results indicate that countries with common law and higher stock market turnover are associated with low level industry concentration. There is more industry contestability in countries with better quality of institutions and where regulation is more open. Bank concentration and simultaneous restrictions to engage new activities in the financial industry tends to decrease firm entry in the mutual fund industry. The launch of new funds is positively related with openness of regulation and negatively related with industry age. Fees tend to be higher in countries with low stock market turnover, where industry size is smaller and where foreign mutual fund companies have a larger market share. Moreover, fund proliferation seems to be an important aspect of competition as it is negatively related with mutual fund fees. Overall, the results do not indicate a clear relation between competition and concentration, similar to the findings for the banking industry. Nevertheless, our evidence validates the contention the degree of competition is important for the variety of products, as we find a larger fund offer in more competitive industries. JEL: G15; G23; L11. Key Words: Mutual Funds; Industry Competition; Industry Concentration, Mutual Fund Fees. Financial support from Fundação para a Ciência e Tecnologia is greatly acknowledged (POCI/EGE/57002/2004).

2 Mutual Fund Industry Competition and Concentration: International Evidence January 2009 Abstract This paper examines mutual fund industry competition and concentration in 27 countries using a sample of almost 50,000 mutual funds. The mutual fund industry is concentrated worldwide and some industries present large fund complexes. Our results indicate that countries with common law and higher stock market turnover are associated with low level industry concentration. There is more industry contestability in countries with better quality of institutions and where regulation is more open. Bank concentration and simultaneous restrictions to engage new activities in the financial industry tends to decrease firm entry in the mutual fund industry. The launch of new funds is positively related with openness of regulation and negatively related with industry age. Fees tend to be higher in countries with low stock market turnover, where industry size is smaller and where foreign mutual fund companies have a larger market share. Moreover, fund proliferation seems to be an important aspect of competition as it is negatively related with mutual fund fees. Overall, the results do not indicate a clear relation between competition and concentration, similar to the findings for the banking industry. Nevertheless, our evidence validates the contention the degree of competition is important for the variety of products, as we find a larger fund offer in more competitive industries. JEL: G15; G23; L11. Key Words: Mutual Funds; Industry Competition; Industry Concentration, Mutual Fund Fees. 2

3 1. INTRODUCTION The dramatic development of the mutual fund industry worldwide has been documented and studied in several academic studies (e.g. Otten and Schweitzer (2002), Klapper, Sullas and Vittas (2004), Khorana, Servaes and Tufano (2005) and Ramos (2009)). An important aspect left out has been the concentration and the competition of the mutual fund industry worldwide. Studying concentration and competition in the mutual fund industry is of paramount importance for several reasons. Similarly to other industries, the degree of competition is important for the quality, variety and costs of products. In other words, the lack of competition can create inefficiencies in fund diversity, performance, and fees. For instance, studies have documented about economies of scale not passing to investors or that on general the performance of funds is too poor for the level of fees charged (e.g. Gruber (1996), Korkeamaki and Smythe (2004) and Gil-Bazo and Ruiz-Verdú (2007)). Second, there is evidence of distortions caused by competition inside categories and families. Since mutual funds inflows are sensible to past performance, portfolio managers have incentives to manipulate their position in the category ranking by changing the fund volatility (e.g. Brown, Harlow and Starks (1996) and Chevalier and Ellison (1997)). Moreover, fund families are tempted to coordinate actions across funds in the family complex in order to enhance performance of the funds that are most valuable to the family, even if that comes at expenses of other funds. Gaspar, Massa and Matos (2006) conclude that favouritism generates distortions in delegated asset management. Therefore, competition in the mutual fund industry can exacerbate conflicts of interest between investors and fund families. Third, this is an issue relevant for financial regulators. In the U.S., fees are under the close inspection of regulators. The SEC and the U.S. General Accounting Office are among the authorities that produce regular reports on the issue 1. Fee disclosure is considered crucial for investor-driven competition as investors can only exert fee pressure if they are fully aware of their level. Also, the European Union (EU) authorities have showed interest in measures to foster competition. For example, the EU has tried to foster competition in financial markets through the Investment Services Directive or the Undertakings for Collective Investment in Tradable Securities (UCITS) Directive that regulates collective funds that can be sold across national boundaries within the EU. These funds can be marketed within all countries that are a part of the EU, provided that the fund and fund managers are registered within an EU country. Ramos (2009) report that the majority of funds domiciled in EU countries are also UCITS accounting for 77% in 2005 of the value of the EU funds. 1 See and 3

4 The market structure of the financial industry worldwide has been mainly studied for the banking industry 2. For the mutual fund industry the evidence reported is only for the U.S.. Khorana and Servaes (2007) study competition over the period They document that over that period, industry assets increased by a factor of twenty, the number of active fund families tripled, and the average market share of a family declined by two thirds. Other reported features of the market structure of the mutual fund industry are fund proliferation, fee dispersion and increased segmentation (see e.g. Massa (2003) and Hortaçsu and Syverson (2004)). Mutual fund fees have nevertheless attracted more the attention of researchers and have been studied for Spain (Gil-Bazo and Martinez (2003)), Finland (Korkeamaki and Smythe (2004)), Italy (Geranio and Zanotti (2005)) and worldwide (Khorana, Servaes and Tufano (2006)). This paper contributes to the literature by being the first to analyze mutual fund industry concentration and competition around the world using a novel data base of 27 developed and emerging market countries from almost 50,000 open-end mutual funds for the year of We document the following results. The mutual fund industry tends to be quite concentrated worldwide. There are few large companies that dominate the national industries, and several small companies. On average, the five largest administration companies account for 60% and portfolio management for 54% of market share. The figure is higher for bond or equity mutual funds segments. Large fund complexes are present in countries such as Australia, Austria, Canada, Germany, the U.K. and the U.S. suggesting a more strong and developed mutual fund industry in those countries. The results indicate that countries with common law and higher stock market turnover are associated with low levels of industry concentration. A common feature of the organization of the mutual fund industry across countries seems to be the concentration of administrative tasks while portfolio management structure is more fragmented. In the large majority of the countries, the number of portfolio management companies is larger than the number of administrator companies. Note that Chen, Hong, Huang and Kubik (2004) contend that larger mutual fund companies suffer from a form of diseconomies of scale, hierarchy costs that erode performance. 3 Industry contestability is higher, i.e. more firm competing and more entry of new firms, in countries with better quality of institutions and where regulation is more open. All the indicators related with regulation like economic freedom, fund regulation and the level of restrictions faced by financial institutions are related with industry constestability. Bank concentration with simultaneous restrictions to engage new activities in the 2 See Claessens, Demirguç-Kunt and Huizinga (2001), Barth, Caprio and Levine (2004), Claessens and Laeven (2004). 3 According to Stein (2002) small companies are likely to achieve better performance because they process better soft information. In opposite, in large organizations with hierarchies, the process for agents fighting for having their ideas implemented will affect agents ex ante decisions of what ideas they want to work on. This happens because if the 4

5 financial industry tends to decrease firm entry in the industry. Also, more developed financial systems have less entry of new firms. In the same line, the launch of new funds is also associated with regulation. Industries with more open regulation and but also younger mutual fund industries tend to launch more new funds, a result similar to Claessens and Laeven (2004) for the banking industry. The level of segmentation is quite high in European industry. European industries show a large variety of mutual fund styles, while the Asian industry, with the exception of Japan and Singapore, shows the opposite. The evidence suggests bond and specially equity mutual funds as the most competitive mutual fund segments as the industries where more new firms enter have a larger offer of bond and equity funds per capita. Fees differ substantially around the world even within asset classes and similar styles similar to the findings of Khorana, Servaes and Tufano (2006). The U.S. stands out by the low level of annual fees. Fees tend to be higher in industries with low stock market turnover, and where industry size is smaller, which corroborates validates the hypothesis of economies of scale on the industry. Fees are higher in countries where foreign mutual fund companies have a larger market share. One of the possible explanations is that foreign companies can be attracted to enter in markets where they can charge higher fees and have higher revenues. On the other hand, foreign mutual fund companies can face higher costs when domiciling funds in a foreign market and that cost is reflected on fees. Fund proliferation seems to be also an important aspect of the competition in the industry. The rapid growing of mutual funds described by Gruber (1996) has been interpreted as a competitive strategy to limit competition and increase market coverage of fund families (Massa (1998, 2003)). Our results show that fund proliferation is negatively associated with the level of mutual fund fees. Fee proliferation has been reported for the U.S. (Massa (2003) and Hortaçsu and Syverson (2004)) but we also find it in countries where companies manage and introduce a large number of funds, like the ones with large fund complexes. Overall, the evidence suggests a great diversity of cases that challenge the stylised fact that a high competitive market is characterized by a large number of firms competing and firm entry, lower concentration and lower prices. We do not find that concentration and competition in the mutual fund industry present a direct relation, similarly to the findings in Demirguc-Kunt and Levine (2000) and Jansen and Haan (2003) for the banking industry. The paper has the following structure. Section 2 presents the data and discusses the determinants. Section 3 presents competition indicators, focusing on the following aspects of market structure: constestability, industry segmentation, fund proliferation and fees. Section 4 analyzes concentration in the mutual fund information is soft, then agents have a hard time convincing others of their ideas and it is more difficult to pass this 5

6 industry. Section 5 analyzes the determinants of concentration and competition in the mutual fund industry. Section 6 relates the concentration and competition. Section 7 describes robustness checks and Section 8 concludes. 2. DATA AND EXPLANATORY VARIABLES 2.1. Data The mutual fund industry data are drawn from the Lipper Hindsight database. This database contains mutual funds characteristics from 27 developed and emerging market countries. We collect information about almost 50,000 open-end mutual funds for the year of The net value asset of individual funds is aggregated by domicile country in order to give the size of the national mutual fund industry. Some caveats are in order on this issue. First, the mutual fund industry has global scope and many management companies sell funds in several countries. It is also quite common that mutual fund companies have funds domiciled abroad as off-shores. Second, investors' holdings in a country can be larger than the size of the national mutual fund industry since investors can hold mutual funds domiciled in other countries. Therefore, the definition of size of the mutual fund industry adopted in this study refers to funds domiciled in the country, not to the assets managed by firms in the country or the holdings of country investors in mutual funds 4. Table 1 describes our mutual fund sample by country. The first column reports the size of the mutual fund industry in USD million. These figures correspond to the sum of the individual funds domiciled in each country. We cross the coverage of funds by Lipper Hindsight with the statistics on mutual funds provided by the Investment Company Institute (2006). The total net asset value in our sample is $16.4 trillion which compares with $17.7 trillion worldwide. The breakdown by countries shows that the mutual fund industry is larger (in terms of net asset value) in the U.S. followed by France and the UK. The mutual fund industry is smaller in Indonesia and Malaysia. Table 1 also presents the number of funds (only primary) by countries. In absolute terms the U.S. presents more than 8,000 funds followed by Korea and France. Poland and China, two countries with young industries, have a small number of funds. (Table 1 around here) information up the organisation. 4 See also Khorana, Servaes and Tufano (2005) for a similar approach. 6

7 2.2. Explaining Mutual Fund Industry Concentration and Competition This section examines the determinants of the level of concentration and competition in the mutual fund industry. Concentration and competition in the financial industry have been mainly analyzed for the banking industry. Demirguc-Kunt and Levine (2000) find that bank concentration is negatively associated with restrictions on banking activity. Countries characterized by banks that engage in a wide range of financial activities tend to have relatively high levels of bank concentration. Claessens and Laeven (2004) analyze the role of entry and activity regulations in the competition environment of the banking industry. They find that openness to new entry is the most important competitive pressure. Following the previous literature, we assume that concentration and competition are endogenous variables determined by a set of exogenous variables that are presented and discussed next Quality of Legal Institutions and legal system To measure the quality of institutions we use the KKZ_Composite Index (KKZ_INDEX) an index constructed by Kaufman, Kraay and Zoido-Lobatón (1999) that has been used in the literature (e.g. Beck, Demirgüç-Kunt and Levine (2006)) to measure the overall level of institutional development. It is composed of the following items: voice and accountability, government effectiveness, political instability, regulatory quality, rule of law and control of corruption. In the line of previous work, we expect that countries with better institutions have less concentration and more competition and contestability. The quality of the legal environment as proxied by legal families is a determinant of a country's level of financial development. La Porta, Lopez de Silanes, Shleifer and Vishny (1997, 1998) advocate that the quality of the legal system is important for the enforcement of contracts capturing also the government's general attitude towards business. To gauge the impact of the legal system on the market structure of mutual fund industry we use a common law dummy variable (COMMON). Accordingly, we expect to find a positive relation between quality of institutions variables and industry competition (see La Porta, Lopez de Silanes, Shleifer and Vishny (1997, 1998), Levine (1999) and Levine, Loyaza and Beck (1999)) Regulation and Economic Freedom Regulation and the level of economic freedom can be potential determinants of mutual fund industry concentration and competition. Regulation can be a mechanism by which entry of new competitors in an industry is restricted, increasing concentration and decreasing competition. Zingales (2004) states that in modern democracies, regulation is the result of political pressures exerted by different lobbies. In particular, incumbents tend to be more organized and thus more political powerful and regulation can be biased against entry and competition. To measure barriers to entry we use a set of variables described above. The main 7

8 hypothesis is that barriers to competition intensify concentration and diminish competition in the mutual fund industry. To measure the overall level of economic freedom we use the index of economic freedom provided by the Heritage Foundation (ECON_FREEDOM). The index uses 10 specific freedoms such as business freedom, trade freedom, monetary freedom, freedom from government, fiscal freedom, property rights, investment freedom, financial freedom, freedom from corruption and is graded using a scale from 0 to 100, where 100 represent the maximum freedom. A score of 100 signifies an economic environment or set of policies that is most conducive to economic freedom. This index has been used in the work of Beck, Demirgüç-Kunt and Levine (2006) and Claessens and Laeven (2004) to capture the general level of barriers to entry by new companies. Given the close relation between banks and the mutual fund industry, we are interested in the relation between banking regulation and mutual industry concentration and competition. To investigate that we consider the level of restrictions faced by banks to enter in other financial activities such as securities activities and compete with mutual fund management companies. BANK_RESTRICT is a variable taken from Barth, Caprio and Levine (2001) and is a summary index of overall regulatory restrictiveness in items like Securities, Insurance, Real State and Non-financial firm ownership. It takes the values between 1 (least restrictive) and 4 (most restrictive). This index has been used by Beck, Demirgüç-Kunt and Levine (2006)) and Claessens and Laeven (2004) to measure the level of economic freedom and institutional development. Fund specific regulation can also be a barrier to competition, therefore we use variables related with fund regulation like custodians independence (CUSTODIAN). Custodians are the parties that hold the securities of the fund, and their independence insures that the fund s assets are not expropriated. Following Khorana, Servaes, and Tufano (2005, 2006) we create a dummy variable that assumes the value one if custodians are required to be independent from the mutual fund family. The data is from KPMG. Khorana, Servaes and Tufano (2006) use concentration in the banking systems as a proxy to barrier to competition when analysing the determinants of cross-country differences of fees. Similarly, we use BANKCONC the market share of the five largest banks from Beck, Demirgüç-Kunt and Levine (2000) Other Variables and Controls We also analyze the structure and development of the financial system by using the variable TURNOVER from the World Development Indicators. Like Ferreira, Miguel, and Ramos (2006) we assume that investor sophistication and the development of the financial system tends to increase with stock market turnover. We also control for other variables that are likely affect the development of the financial system. We should expect to see more companies in more wealthy countries, where investors are richer and where the industry is 8

9 older. Hence we consider several indicators of the country level of economic development: the logarithm of GDP per capita (GDP), and the level of internet users (INTERNET) from World Development Indicators. The influence of industry age on competition and concentration measures can be dubious. Countries where the mutual fund industry is older and mature should have more developed and competitive markets. However, more mature industries are usually related with less new entries. On the other hand, due to considerable economies of scale in the financial industry there is a natural trend to concentration. Therefore, one can expect to see more industry concentration as the industry grows and becomes older, but also that older industries have more firms competing. We consider the logarithm of the year the industry started in a country (IND_AGE). Data is from Khorana, Servaes and Tufano (2005). Table 2 displays all the referred variables (Insert Table 2) 3. MUTUAL FUND INDUSTRY COMPETITION In this section we discuss some aspects of the competition in the mutual fund industry. Like in any industry, mutual fund firm profits are maximized by increasing revenues and decreasing costs. Firms aim to increase assets under management since revenues are usually a percentage (i.e. in gross terms, the fees) of assets under management. Firm goals are naturally to gain market share and increase fees. The key driver of the mutual fund industry competition seems to be that the fact abnormal past performance originates a substantial amount of fund inflows. Early studies like Spitz (1970), Chevalier and Ellison (1997) and Sirri and Tufano (1998) have documented that abnormal positive returns generate disproportionately more inflows than abnormal negative returns would generate outflows. In addition, fund performance is also important because of the spillover" effect that the performance of a fund provides to all the other funds belonging to the same family. Nanda, Wang and Zheng (2004) show that star performance results in greater cash inflow to the fund and to other funds in its family. Moreover, families with higher variation in investment strategies across funds are shown to be more likely to generate star performance. Massa (2003) argues that the proliferation of funds is originated by the spillover effect that the performance of a fund provides to all other funds belonging to the same family Contestability A main element of economic models of competition is the number of firms competing. The usual claim is that the larger the number of rival firms, the more choices available to consumers and the greater the likelihood of competitive pricing. Thus the first element of competition we analyze is the number of firms in the mutual fund industry. We study the mutual fund industry competition worldwide by focusing on the fund administrator and portfolio management companies. The administrator company oversees the performance of other companies 9

10 that provide services to the fund s operations and is responsible for calculating NAV and supplying the price and dividend information. The portfolio management company is responsible for managing the portfolio of the fund. Let us introduce the following notation that we use through the paper: N is the number of management companies in each country; Nf is the total number of mutual funds in each country and Pop is the total population of a country in millions. We consider several country-level measures of the mutual fund industry competition: Number of management companies per million inhabitants (N/Pop). This variable indicates the number of available companies per capita (i.e., standardized by the country population). The larger the number of companies per capita, the higher is the expected the level of competition. Market share of foreign firms in a country: (Foreign_mf_sh). The presence of foreign competitors is usually related with fewer barriers to entry in the industry. Therefore Foreign_mf_sh is a measure of the openness of domestic mutual fund industry to foreign competition and can be indicative of the level of competition. Coates and Hubbard (2006) refer that the most direct indicator of barriers to entry and expansion is the extent of actual firm entry and existing firm expansion. We next analyze the entry of new companies computing the number of new companies in every year from 2003 through We then compute the annual average of new companies denoted New_comp. These figures are then used to compute the following ratios: - The average number of new companies to the total number of new management companies (New_comp/N). This ratio will express how much of the existing companies are recent companies, being indicative of industry turnover. - The average number of new management companies per inhabitants (New_comp/Pop). It will be indicative of the level of constestability of the industry. A priori in an industry where many companies exist and there is a rapid pace of firm entry is a competitive market. However, it must take into account that younger industries are usually related with a rapid pace of new entries as well as that the ratios are largely determined by the number of companies and population. Table 3 reports industry competition measures by country. Panel A reports data on administrator companies and Panel B on portfolio management companies. The fist two columns report the total number of companies and the number of companies per million inhabitants. The first column of each panel reports the total number of companies by country. In Panel A (data was not available for the U.S.), France has the largest number of administrator companies, while Poland has the lowest. Several other European countries also present a large absolute number of companies such as the UK, Germany and Spain. 10

11 Singapore presents the largest number of administrator companies, 9.46 per million, and Austria has the largest number of portfolio management companies per million around 21. India and China present the lowest value (near zero) for that indicator, as they are overpopulated countries. Panel B of Table 3 reports competition measures by country using portfolio management companies (now including the U.S.). The average number of portfolio management companies per country is 118 (97 excluding the U.S.), larger than the average number of administrator companies 63. This seems to be one of the main features of the organization of the mutual fund industry. Administrative tasks are more concentrated while the portfolio management structure is more fragmented. In countries like Austria, Belgium, Indonesia, Finland, Switzerland and the UK, the number of portfolio management companies is more than the double of administrator companies. There are also countries that do not present differences in both such as China, India, Japan, Portugal and Taiwan. (Insert Table 3 around here) Lastly, we analyze the presence of foreign portfolio management companies. Foreign companies have a large market share in countries such as Poland, Singapore, Indonesia, and the industry is dominated by national companies in Belgium, Switzerland and Sweden. Note that we cannot conclude for the absence of contestability in these countries and that foreign companies face barriers in entering in the market, as what we observe is the outcome of competition. Table 4 exhibits indicators on market entry, more precisely new companies. Australia and France present in absolute terms the largest number of new administrator companies. The U.S., Australia and the U.K. present the higher number of portfolio management companies in the period When we focus on new companies per million inhabitants, Australia and Singapore present the higher values for administrator companies and Switzerland and Austria for portfolio management companies, while overpopulated countries such as India, China and Indonesia present the lowest scores. Note that Belgium did not have any new administrator company in the period analyzed. New firms are a large percentage of the actual ones in countries like China and Australia, for administrator companies, and in China and India for portfolio management companies. Firm entry is low in Belgium for administrator companies and in Norway and Spain for portfolio management companies. (Insert Table 4) Table 5 analyzes competition indicators for the bond and equity mutual fund sector using portfolio management companies. Panel A reports data on bond funds and Panel B on equity funds. In both asset segments, the countries with higher number of firms competing are the U.S., the U.K. and France. Austria and Switzerland present the highest number of administrator companies per million inhabitants both for bond and equity funds, while again over populated countries present the lowest value. The number of portfolio 11

12 management companies is larger in Austria and Switzerland both for bond and equity funds. Other countries with a larger number of firms are Denmark, Finland and Singapore. The results must be taken with caution because the indicators are influenced by the standardization. For instance, Canada, Spain, France, Germany and the U.S. are countries with a large number of firm competing in absolute terms, but when standardized by the population this number becomes small. Nevertheless, Austria and Switzerland are two countries where there is a large number of firms competing in absolute terms and per capita. In the other extreme, Portugal, Poland and Thailand are countries with a small absolute and per capita number of firms competing as well as firm entry. Belgium and Indonesia are countries with a fair number of firms competing but with a low pace of firm entry. (Insert Table 5) 3.2. Performance versus non Performance Competition As referred above, firms try to induce performance in funds in order to gain market share. Gaspar, Massa and Matos (2006) find some evidence that mutual fund families strategically allocate performance across their member funds favoring those more likely to generate higher fee income or future inflows. These funds will not only generate inflows for the fund itself but for the all family. Guedj and Papastaikoudi (2005) find that the better performing funds in a family have a higher probability of getting more managers, one of the main resources available to firms. However, performance is not the only competitive variable explored by mutual fund firms. Many studies have documented investor heterogeneity 5 in mutual funds. Investors have different profiles and investment needs that can be explored by fund complexes. Cristoffersen and Musto (2002) advocate that one reason why funds charge different prices to their investors is that they face different demand curves. Gil-Bazu and Ruiz- Verdu (2007) document strategic fee-setting by mutual funds in the presence of investors with different degrees of sensitivity to performance. Massa (1998) advocate that fund and category proliferation phenomena can be seen as marketing strategies used by the managing companies to exploit investors' heterogeneity. These aspects are discussed in more detail below Category proliferation The creation of several styles or categories of mutual funds has been well documented (e.g. Massa (2003) and Teo and Woo (2004)). Given that funds do not seem to compete across categories (Navone (2003)), by launching funds in new categories, firms can avoid cannibalism within funds of the same family and gain market share. Khorana and Servaes (2007) report that product differentiation strategies are effective in 5 See for instance Hortaçsu and Syverson (2004) and O Neal (2004). 12

13 obtaining market share. Families that perform better, and start more differentiated funds from existing offerings relative to the competition (a measure of innovation 6 ) have a larger market share. Therefore more competition at the firm level should induce more category variety. Note that if the strategy is successful, i.e. if fund complexes that offer more styles gain market share, then concentration in the industry can increase. To gauge a sense of style proliferation worldwide, we analyse mutual fund style variety using Lipper Global Style Classification. Table 6 shows the results presenting style varieties per country differentiating the two main asset styles, bond and equity funds. The first columns show the total number of style variety in the industry. The average number is 57 styles which also coincides with the median. Thus half of the sample of the industries has more than 57 categories of funds, almost all European with the exception of Canada and Singapore. The next two columns show the number of styles for equity and bond funds. There is a larger variety of bond styles than equity styles explainable by the type of categorisation used by Lipper. The standard deviation in equity styles in quite high. There is a group of countries with a large equity variety like Austria, Finland, France, Germany, Italy, Japan, Spain and Switzerland. On the other extreme, a group of countries like China, India, Indonesia, Korea, Malaysia, Taiwan and Thailand where industry has a small number of styles. For bond mutual funds, China, Indonesia, Thailand and Poland show low variety, while countries like France, Germany, Singapore, Switzerland and the U.K.have a large variety of bond styles. (Insert Table 6 around here) In sum, Asian countries present a lower number of style varieties for bond and equity funds maybe due to the recent introduction of the industry. European countries like Austria, Belgium, France, Germany, Switzerland and the U.K. have a large variety of styles Fund proliferation Early studies like Gruber (1996) report a growing number of mutual funds despite the apparent managers inability for achieving superior performance. Massa (1998) argues that fund proliferation phenomena can be seen as marketing strategies used by the managing companies to exploit investors' heterogeneity. He explains category and fund proliferation providing an industry-specific micro foundation on the basis of the "spillover" that the performance of a fund provides to all the other funds belonging to the same family. Massa (2003) adds that fund proliferation becomes an additional tool that can be used to limit competition and increase market coverage. On the empirical side, Khorana and Servaes (1999) studying mutual funds starts find that large families and families that have more experience in opening funds in the past are more likely to 6 Style variety can also be understood as a sign of industry and investor sophistication. 13

14 open new funds. Note that an additional reason that can make larger firms generate more funds, is that the cost of generating a new fund can be smaller for large families as they can benefit from economies of scale and scope. Hence, it is likely that more competition among mutual fund firms should lead to a larger offer of mutual funds and to larger families to have a higher number of funds competing. Note again that this also implies that if the competing strategy is effective, i.e. launching more funds is effective in gaining market share, it can lead to an increase of concentration in the industry. Evidence from Khorana and Servaes (2007) shows that families that perform better and start more funds to the competition have higher market share. Therefore we compute the average number of funds per company (Nf/N). This measure gives an indication of fund diversity, and is likely to be related with fund competition in a country. In a more competitive industry it is likely to exist a larger number of funds per family. Note that industries in an early stage of development are likely to offer a smaller number of funds. We also analyze the number of new funds from 2003 through We then compute the annual average of new funds (New_funds 7 ). The following ratios are then analyzed: Average number of new funds to the total number of funds (New_funds/Nf). This ratio will express how much of the existing funds are recent funds, being indicative of industry turnover. Average number of new funds per management company (New_funds/N). It expresses the degree of innovation of companies, which might be indicative of the level of fund competition. The average number of new funds per million inhabitants (New_funds/Pop). It is more indicative of the level of contestability of the industry. If an industry has a small number of funds and/or launches few new funds, it can be read as sign of barriers to firms expansion and concomitantly, few industry competition. Yet, it can also be related with two other cases. One is the industry onset, where there is a small number of existing funds and a high number of new ones. The other is industry maturity characterized by the opposite: a large number of existing funds and a small number of entry of new funds. Table 7 reports the results. Panel A reports general indicators showing the results of the number of funds per capita (Nf/Pop), the ratio of new funds in the total funds (New_funds/Nf) and the number of new funds per capita (New_funds/Pop). Australia, Austria, Belgium, Finland, France, Korea, Singapore and Switzerland have a high number of funds per capita while China, India, Indonesia and Poland a small number. Note that new funds are a large 14

15 percentage of total existing funds in Asian countries like Korea, China, India, Indonesia and Thailand, while in European countries like Belgium, Netherlands, Norway and Sweden they are a very small percentage. Countries that present a large number of funds per capita are Korea, Australia, Austria, Finland, Singapore and Switzerland, while China, India, Indonesia and Poland present a small number of new funds per capita. We next analyze the number of funds as well the number of new funds managed by firms. First for administrator companies (Panel B) and then for portfolio management companies (Panel C). In Korea, Thailand, Australia, Belgium and Japan, administrator firms manage on average a large number of funds while in the Netherlands, Sweden, Germany, China, Poland, Malaysia and Taiwan they administrate a small number of funds. Administrator firms in Korea, Thailand and Australia launch a high number of new funds while in the Netherlands, Sweden, Germany, Norway and the UK, they tend to not introduce many new funds. Korea, Thailand and Japan present a high number of funds and new funds per portfolio management company while in the Netherlands, Germany and the U.K. we find the inverse. Portfolio management companies offer on average a small number of funds and launch few new funds. Then we analyze the number of funds owned by the five largest companies (Nf_CR5), and compared with the average number of fund owned by a company in a country. Table 7 shows the average number of funds owned by the five largest companies versus the average number of funds owned by company. We find that the five largest companies own six times more funds than the average company. These figures are higher for countries like Australia (15 more), Austria (14 more), Canada (14 more), Germany (13 more), the U.K. (11 more) and the U.S. (12 more). With low asymmetry in the size of families, we find countries such as China, India, Poland, Portugal, Taiwan and Thailand, where the largest five companies own at maximum 2 times more funds than the average fund company. (Insert Table 7) The existing number of funds and the launch of new funds seems much determined by the age of the industry. Asian countries tend to launch many new funds maybe because the industry is recent in that geographic region. On general, European countries have a very slow pace of new funds, although they have a large number of existing competing funds. We also find cases like Poland where there is a small number of funds as well of entry of new funds. The size of firms is quite asymmetric in countries like Australia, Germany, the U.K. and the U.S., where we find big fund families but many small firms too. 7 We use information on the launching date of the fund and for the companies when the first fund was launched. We also computed New_funds using only data from 2004 to 2006 and results were the same. 15

16 Fee level and differentiation One of the features of a competitive structure is low prices, which in the mutual fund industry case substantiates in lower fees. This contrasts with the profit maximization logic wheres by construction companies aim to increase fees in order to maximize revenues. Fees around the world Fund companies charge investors fees in order to compensate for the management costs of the fund (annual charges). They compensate the fund s manager for the expenses she incurs for providing its services, including security research and analysis. Fees also pay other administrative expenses resulting from the provision of record keeping and transactions services to shareholders like providing statements and reports, disbursing dividends, custodial services, taxes, auditing and legal costs. In addition, mutual funds often charge a sale surcharge (loads) when the investors purchase (initial charges or front-end load) or sell shares (redemption charges back-end load. The U.S. mutual fee structure is also characterized by class shares differentiated in the load structure and the 12b-1 distribution fee spent on advertising, marketing, and distribution services or on commissions to sales representatives. The key question for firms is to know how fund flows vary with fees, or using a more economic language to know the flow fee elasticity. Empirical evidence shows that investors seem more sensible to loads than expensive ratios (see Barber, Odean and Zheng (2005) and Wilcox (2003)). Chordia (1996) argues that investors who redeem fund shares impose externalities on those who do not, from liquidation of securities; this obliges the fund to have a higher cash position reducing performance. Thus, by making redemptions expensive, a mutual fund dissuades investors from redeeming shares and it is able to invest in a more risky portfolio to obtain higher performance. Greene, Hodges and Rakowski (2007) find that the redemption fee is an effective tool in controlling the volatility of fund flows. Khorana and Servaes (2007) find that price competition is important in the industry. Families that charge lower fees than the competition gain market share, but only if these fees are above average to begin with. In addition, fees charged explicitly for marketing and distribution (12b-1 fees) have a positive impact on market share. Table 8 show the level of annual fees in the mutual fund industry in our sample. The first column shows the average fee across all funds of the industry. Fees are high in Poland, Spain 8, Italy, India and Canada and lower in Korea, Belgium, Taiwan, the U.S. and Thailand. 8 Gil-Bazo and Martinez (2003) already reported that Spain ranks first in terms of average mutual fund fees among similar countries. 16

17 The other columns report average fees according to asset categories: bond, equity, mixed asset, money market. Fees are systematically high in all asset categories in Poland, Spain and Italy and systematic low across countries in Belgium, Taiwan, Thailand and the U.S.. In fact the U.S. appears with a favorable fee gap with all other countries of the sample. The trend of decreasing fees in the U.S. is confirmed in a recent analysis reporting that mutual fund fees and expenses fell to their lowest levels in more than a quarter century during 2005 (ICI, 2006). In equity funds, fees are high in Poland, Spain, Korea, Portugal and Italy, while in Taiwan, U.S., Belgium, India, the Netherlands and Switzerland are below the sample average. The classical competition hypothesis states that in a competitive industry prices are lower. However the global average level of fees is not sufficient to infer about the competitiveness of the market as the level of fees depend also in other factors like economies of scale, distribution costs, regulation and investment culture (Khorana, Servaes and Tufano (2006) 9 ) or investment objectives (Mack (1993), Keim and Madhavan (1997)). Therefore it is not possible to conclude for the absence of competition just because in an industry fees are higher. (Insert table 8) Fee proliferation A striking feature reported for the mutual fund industry has been fee proliferation. Hortaçsu and Syverson (2004) find a large variation in fees for an homogenous class of funds like index funds on S&P 500. Cristoffersen and Musto (2002) advocate that one reason why funds charge different prices to their investors is that they face different demand curves. Different investment needs of the investors like long term versus short term investors, or informed versus non informed investors can originate a differentiated fee structure. Evidence shows that this is a successful competition tool. Nanda, Wang and Zheng (2003) find that multipleclass of funds attracts significantly new money than single class of funds. According to Coates and Hubbard (2006) price dispersion in highly competitive markets is well documented by economists. It is possible to find different prices for identical items across types of outlets, such as full service department stores versus mass merchandiser price discount stores. Price dispersion in homogeneous good markets is partly attributed to search costs. Given that consumers lack perfect information, they search up to the point where search costs just exceed the expected lower price. Thus, search costs, including the opportunity costs of an investor s time, provide a basis for price dispersion in competitive markets. 9 Khorana, Servaes and Tufano (2006) report that larger funds and fund complexes charge lower fees and that fund fees are lower in countries with higher investor protection. Mack (1993) report that funds on small caps or international companies are more expensive. Keim and Madhavan (1997) find a large dispersion of transaction costs for different investment styles. 17

18 To measure fee proliferation we compute the dispersion of fees across funds following Massa (2003) and Hortaçsu and Syverson (2004). To see how industry competition is related with fee proliferation we select fund styles with the largest number of funds competing. Table 9 shows for each country the category according to Lipper Global Style with more funds competing. We report the number of funds, the average fee and standard deviation of fees. The first thing to notice is the diversity of preferences of investors, revealing heterogeneity in investors from different countries. In some countries, the category/style where most funds compete is the domestic equity market. In others, funds on the domestic bond market and in others, the protected and guaranteed funds. Given that different styles will have different fees by construction (e.g. bond fund fees are generally smaller than equity funds fees), fees are not comparable. Therefore we decide to follow another approach and compare more similar (competitive) styles. For each country we select the money market, bond and global style with the largest number of funds competing. We also select equity funds on the domestic equity market due to home bias preferences (see Lewis (1999) for survey on home bias literature). Table 10 reports the number of funds competing, average fee and fee dispersion in columns for the selected categories. Panel A shows the money market style with more funds competing in each country. A priori one could think that is a rather homogeneous segment but we find great disparity in fees. Fees in the segment are also in line with the evidence on Table 8. Panel B shows that fee dispersion on the bond mutual fund style with more funds competing according to Lipper categorization. It seems that bond funds that invest in the domestic bond market are the most chosen in most countries, except for Spain and Taiwan. Again we find great disparity of fees in countries. Similarly to the money market, Poland presents higher fees. Then Panel C shows funds competing on domestic equity market. Note that in some countries it is not only the doemstic market. Fees are lower in the U.S. and higher in Poland. Finally, Panel D shows the number of funds for Lipper Equity Global Style. Strikingly, investing abroad has an average lower fee than for instance investing in domestic equity market. This is strikingly given that for instance transaction costs are normally higher when investing abroad (Mack (1993)), and that fund size is smaller usually when investing outside (Ferreira, Miguel and Ramos (2006)), therefore they are likely to benefit from lower economies of scale. (Insert table 10) To analyze dispersion, we look at standard deviation of fees but in relative terms since the level of fees in countries is also quite different. Countries that tend to show low dispersion across all fund categories are China, India, Italy, Malaysia, and Taiwan, mainly countries where the industry is at an early stage. Poland shows low dispersion in equity funds. Countries that show systematic high dispersion in the selected styles are Canada, Korea, Switzerland and Singapore. The U.S. shows high dispersion in equity domestic funds, while Australia, Denmark and the Netherlands present for bond funds. Correlation results show that fee 18