DIFFERENCES IN PERFORMANCE OF INDEPENDENT AND FINANCE-AFFILIATED VENTURE CAPITAL FIRMS. Abstract

Transcription

1 The Journal of Financial Research Vol. XXV, No. 1 Pages Spring 2002 DIFFERENCES IN PERFORMANCE OF INDEPENDENT AND FINANCE-AFFILIATED VENTURE CAPITAL FIRMS Kangmao Wang Zheijiang University, China Clement K. Wang and Qing Lu National University of Singapore Abstract We examine the differences in the investment behavior of independent and finance-affiliated venture capital firms (VCFs). We find that differences in internal management mechanisms and staff backgrounds lead to external performance differences. Using VC-backed companies listed in Singapore as our sample, we find significant differences between these two types of VCFs in industry preference, investment duration, VCF syndication, number of board seats, initial underpricing, and long-term market returns. Independent VCFs add more value to their portfolios. Thus, we conclude that the participation of independent VCFs is an important corporate-level factor for the success of the venture capital market. JEL Classifications: G12, G15, G23, G32 I. Introduction Venture capital (VC) is an intermediate external source of financing for small- and medium-sized enterprises. A VC market consists of a supply side (the capital and other service providers) and a demand side (entrepreneurs seeking VC help). Venture capital firms (VCFs) have been successful in the United States and have contributed significantly to the growth of high-technology sectors, particularly in the information technology industry. The success of the VC market there helps maintain the dominant position of the United States in the world economy. However, many countries have failed to build dynamic VC markets to support their high-tech industries. In this article we explore the reasons behind the success of the VC market in the United States and for the failure of the VC The authors gratefully acknowledge the financial support from the National University of Singapore (R ) and thank the anonymous reviewers for helpful suggestions. 59

2 60 The Journal of Financial Research market in other countries. We investigate the micro-mechanism of the VC market and examine the effect of VCF characteristics on its success. Since the 1980s, inspired by the U.S. success, many countries, both developing and developed, created VC industries. Though these countries have tried to replicate the U.S. experience, their attempts have not been successful. Murray and Marriott (1998) find that European VCFs fail to invest much in high-tech start-ups because of poor returns in the early stages. Murray and Marriott question whether the successful investment in high-tech start-ups is an American phenomenon. Japan and other fast-growing Asian countries have had a similar experience (Hamao, Packer, and Ritter 2000). Most of these countries started their VC industries by providing high-tech start-ups with capital from sources such as governments, banks, and other financial institutions. However, after much initial enthusiasm, these high-tech funds soon diverted to low-tech and late-stage investments, or even to speculation in stock markets or property markets. Several studies examine the path to the success seen in the U.S. VC market. Gompers and Lerner (1998) study U.S. VC fundraising over twenty-five years and attribute the U.S. success to the demand side of the VC market, which they see as the real force driving the growth of the VC market. The negative policy implication of the Gompers and Lerner study is to caution against methods that increase the VC supply. Yet, these methods are the strategies most countries use to develop the VC market, including creating government-backed VC funds and giving tax incentives to existing VC funds. These strategies can increase the supply of VC by attracting more capital to VC investment but have no influence on the demand for VC funding. The positive policy implication of Gompers and Lerner (1998) is to increase research and development (R&D) spending. However, taking this route is a long-term solution to an urgent, short-term need. Because the purpose of VC is to encourage technological innovation and commercialization, increasing R&D spending to develop a VC market is an unrealistic solution for many countries. Black and Gilson (1998) study the VC market from the angle of the market structure and attribute the VC market success in the United States to the VC exit mechanism. They find that a strong stock-market-centered capital market is essential to the success of the VC market, because it enables VCFs to exit earlier from the companies in which they have invested and earn more by taking their portfolio firms public. However, because the capital markets of most countries are bank centered, Black and Gilson s conclusion gives little help. A VC market requires an active stock market to support its exit, but a strong stock market requires a supply of successful companies and entrepreneurs, which in turn requires a strong VC market. In other words, this is a classic chicken-and-egg problem. To break the cycle, Black and Gilson suggest that one country could piggyback on another country s institutions, and they mention Israel and Ireland as two successful examples.

3 Differences in Performance 61 Nevertheless, the Israel model might not be applicable to most countries eager to develop their own VC industries. Because of cultural, language, and geographical factors, most high-tech companies face too many difficulties in listing on foreign stock markets. For example, the success of the Israeli VC market could partly relate to cultural and heritage links with the United States. But for companies from other countries it can be difficult to find and engage top-ranked underwriters and analysts, the two essential factors for a successful listing. Therefore, for a country with a bank-centered capital market, the success of its VC market could be problematic unless the country can find a geographic or cultural neighbor with a strong stock market, such as the United Kingdom with Ireland or the United States with Israel. The studies cited here provide valuable insights into the VC market. However, their practical value is limited because macro-level changes are not easy to implement in a short period. To provide more useful implications for countries that wish to develop VC markets, we investigate the micro-mechanism of the VC market. We focus especially on the nature of VCFs. We classify VCFs into two types, independent VCFs and finance-affiliated VCFs, and hypothesize that the way independent VCFs function affects the success of the VC market. II. Rationale and Background When we focus on the micro level of the VC market, the difference between successful VC markets and failed VC markets becomes obvious. In the United States, most VCFs are independent. According to Barry et al. (1990), among VCbacked initial public offerings (IPOs) in the United States between 1983 and 1987, 82% of VCFs, or 74% of VC deals, were independent. But in countries whose capital markets are bank centered, most VCFs are affiliated with financial institutions such as securities companies and banks. In Japan, 81% of VC-backed IPOs listed on the Tokyo over-the-counter (OTC) market between 1989 and 1995 were supported by banks or securities firms VC subsidiaries (Hamao, Packer, and Ritter 2000). Independent VCFs made up only 8% of the market share. The situation in Germany is similar. Black and Gilson (1998) report that banks and insurance companies are the main source of VC funding. However, if we observe only the differences between the characteristics of VCFs in the United States and VCFs in other nations, these differences are not enough to prove they are related to VC market success. There are many differences between the United States and other nations in culture as well as in the capital market structure. These factors make it difficult to test the hypothesis empirically. To eliminate macro-level influences, we must test differences between independent and finance-affiliated VC-backed companies in one market. Because of the private

4 62 The Journal of Financial Research and sensitive nature of the VC industry, data are scarce. Thus, we use VC-backed public companies for comparison, because public companies must disclose information on their performance and their valuations are measured publicly by market prices. To eliminate macro-level influences, we must select a neutral market for testing. If we choose Japan or Germany to test the differences, the results would not be convincing because finance-affiliated VCFs dominate these markets and most of the independent VCFs are foreign firms. The investment preferences and expertise of foreign VCFs might be different from the local VCFs. Thus, the performance differences between companies in which they invest might not prove the effect of the VCFs characteristics. On the other hand, if we test the U.S. market, a market dominated by independent VCFs, the investment behavior of finance-affiliated VCFs might change. For example, finance-affiliated VCFs in the United States often co-invest with independent VCFs to avoid potential conflicts of interest (Gompers and Lerner 1999). A neutral market in which independent and finance-affiliated VCFs are roughly equally weighted would better suit our study. Because of these considerations, we choose Singapore as the test ground. Singapore enjoys equal weighting for both independent and finance-affiliated VCFs through its open economy, which brings in VC from many countries and generates a market mix of VCFs. As a small city state in Southeast Asia, Singapore is a major regional financial center with a bank-centered capital market. Four local banks control most of the business. The Singapore stock market has grown rapidly during the last two decades with the support of double-digit gross domestic product growth. The recent merger of the Stock Exchange and the Future Exchange further accelerates stock market growth. As a result, Singapore is a hybrid of a bank- and stock-marketcentered capital market. The Singapore VC industry started in 1983 when the first VCF, SEAVI, was established with a capitalization of S$48 million. By 1999, with strong government support, the size of the VC pool had reached S$10.2 billion (US$6 billion). Because of the limited human and capital resources in Singapore, there is no restriction on foreign talents or capital movement. Thus, the Singapore VC market is a mixture of both American-style independent VCFs and Japanese-style finance-affiliated VCFs. Unlike the United States, where the concept of VC excludes late-stage and buy-out financing, the term is used in a wider sense in Singapore. It includes all formal private equity financing, which is the same as in Europe.We classify VCFs that participate in the Singapore IPO market into four groups: independent, finance affiliated, corporate, and government funded. Typically, an independent VCF is an independent partnership management firm and manages several VC funds or partnerships. Within each VC fund, the VCF often participates in the form of a limited partnership. The VCF provides

5 Differences in Performance 63 professional equity managers as the general partners and institutional investors as the limited partners. 1 There are four large independent VCFs in Singapore: SEAVI, Transpac, Hambrecht & Quist (H&Q), and Walden. These four VCFs support around 40% of the VC-backed IPOs. Based on their reputation, the Big Four are also invited to manage some governmental strategic VC funds. Of the Big Four, H&Q and Walden are subsidiaries of VCFs in the United States. SEAVI is a joint venture between an American VCF (Advent) and local partners. Transpac is the only firm that does not have direct link with American VCFs. However, Transpac s predecessor, Transtech, was affiliated with an American VCF, Advent International. Hence, independent VCFs in Singapore are heavily influenced by the American VC model. The second type is the finance-affiliated VCF. The term finance refers to various types of financial institutions, such as insurance companies, commercial banks, investment banks, and securities brokerages. Some finance-affiliated VCFs are structured almost the same as independent firms except that they receive some capital injection from parent institutions. Others are governed as investment branches of their holding corporations. Many U.S. finance-affiliated VCFs such as H&Q behave like independent firms, and thus we classify them as independent VCFs. However, this is not the case in other countries. For example, finance-affiliated VCFs in Japan seldom use the limited partnership structure. Most of their staff members come from parent companies, have no technical background, and will be transferred back in a few years time. The investment objectives of finance-affiliated VCFs in Japan are not purely financial return. Some VCFs invest mainly for strategic reasons. Some finance-affiliated VCFs take the lead underwriter position for their parent companies as the principal investment motive (Hamao, Packer, and Ritter 2000). Their funding sources mainly are their parent companies and related parties. There are few unrelated public institutional investors. The Singapore VC market has several groups of finance-affiliated VCFs. The first group comprises subsidiaries of banks, local commercial banks, or foreign investment banks. Commercial banks in Singapore are allowed to conduct investment business; thus, IPO underwriting can be carried out by any bank or securities firm. The second group comprises securities firms, mainly Japanese VCFs such 1 In limited partnership structure, limited partners provide most of the capital, but management power is almost completely in the hands of the general partners. Any conflict of interest is either reduced or avoided by the short life span of the partnership (normally ten years) and the significant amount of compensation from the partnership profits. In the limited partnership structure, limited partners are public institutional investors seeking high returns. General partners are professionals with backgrounds such as entrepreneurs or senior managers in industrial corporations. They are experienced in moving start-up companies along the development path using their expertise and market knowledge from other investments in the start-up s industry or related industries.

6 64 The Journal of Financial Research as Nomura. The third group comprises insurance companies, many of which are primarily engaged in the buy-out business. The third type is the corporate VCF. In Singapore, corporate VCFs tend to be smaller than independent VCFs. Their parent companies provide most funding, and rarely does funding come from the public. However, corporate VCF staffs have technical backgrounds or industrial experience and can provide nonfinancial advice in their own industries but not in other industries. Their approach is between independent and finance-affiliated VCFs. Because they lack experience in the VC market, they seldom play the role of leading VC but often syndicate with other VCFs. It is possible, however, for a corporate VCF to become the leading VC of an IPO. In such a case, the VCF would be experienced in the VC market and would behave similar to independent VCFs. Consequently, we classify such corporate VCFs as independent VCFs. The fourth type is government-funded VCFs, which are active in the Singapore VC market. Their primary investment purpose is not to acquire financial returns, but rather to build up the local high-tech industries and accelerate the growth of promising local enterprises. They function much like finance-affiliated VCFs except that they are more willing to invest in high-tech, less established companies for strategic purposes. Because they lack expertise in selecting and monitoring their investment portfolios, most government-funded VCFs syndicate with independent VCFs when they invest in early-stage companies. When they are the sole investors, government-funded VCFs work in the same way as finance-affiliated VCFs. Therefore, we classify them as finance-affiliated VCFs. The distinction between an independent VCF and the subsidiary VCF of an investment bank is sometimes unclear. Here, we assign VCF type by whether the parent investment bank engages in underwriting business. Thus, H&Q is classified as an independent VCF, but Rothschild Venture, the subsidiary of Rothschild & Sons Bank, is classified as a finance-affiliated VCF. This is because Rothschild & Sons Bank is active in the Singapore underwriting market, even though the funding of Rothschild Venture comes from institutional public investors. III. Hypotheses Based on the definition of the two types of VCFs, we provide descriptive analysis for independent and finance-affiliated VCFs and derive hypotheses for empirical testing. Table 1 summarizes the differences between independent and finance-affiliated VCFs. Although the table describes four aspects, we note that if a finance-affiliated VCF has much the same attributes of an independent VCF, we could reclassify it as an independent VCF. These differences have significant implications for the operational behavior of VCFs. For VCFs that rely on investments by the public, reputation is essential. If

7 Differences in Performance 65 TABLE 1. Comparison of Characteristics of Independent and Finance-Affiliated VCFs. Independent VCFs Finance-Affiliated VCFs Source of funding Mainly public institutional investors Parent companies and associates Investment objective Investment return Not purely for return, must also bring in business for parental company or have other strategic purposes Organizational structure Limited partnership Corporate Staff background Experienced professionals, Staffs from parents without entrepreneurs, or industry managers technical background Note: The table describes the characteristic differences between independent and finance-affiliated venture capital firms (VCFs). Differences reflect in the funding source, investment objective, firm structure, and staff background. Typically, independent VCFs manage several VC funds. Each VC fund is organized in the form of the limited partnership with professional managers from the VCF as the general partners who have management control and public institutional investors as the limited partners who provide the funding. On the other hand, finance-affiliated VCFs normally are structured as corporate subsidiaries of financial institutions with funding and staff from parents or related parties. a VCF fails to generate high returns for its current investors, it will face difficulties in attracting future investments. Its own existence will also be in question because of the limited life of each partnership. The reputation factor constrains much of the conflict of interests between VCFs and their portfolio companies. However, for a VCF that relies on internal funding, reputation is not as important because its future funding is not related to reputation. The investment motive strongly influences the VCF s performance. If capital return is the only goal, a VCF will seek investment opportunities in high-tech and early-stage companies because there is great potential for higher returns even though the risk is high. However, if a VCF has other motives that might restrict the degree of risk it can take, it will more likely invest in low-tech, late-stage, well-established companies, thus reducing the risk. The effect of organizational structure is also significant. In the limited partnership structure, the hefty share of capital returns for general partners is an important factor in stimulating VC fund managers to work hard and take risks. However, for a VCF managed as a normal investment institution, managers will not have any incentive to take risks. Financial institutions are traditionally conservative and their high debt ratios reduce their capacity to take risks. Thus, their VC subsidiaries are also unwilling to invest in early-stage, high-tech companies. Staff background also directly influences a VCF s investment behavior. Staffs with technical and industrial backgrounds are better able to understand early-stage companies and to evaluate the feasibility of a start-up s business plan. Knowledgeable staffs can also better assess the value of a start-up s intangible assets, such as patents and software. However, evaluations by staffs without technical

8 66 The Journal of Financial Research backgrounds are similar to credit analysis in commercial banks, where only the visible profits and tangible assets matter. Hence, such staffs are more conservative and lack the knowledge to invest in high-tech start-ups. These differences are reflected in the IPO process. Because parent companies of finance-affiliated VCFs often participate in the underwriting business, independent and finance-affiliated VCFs differ in their IPO strategies. Independent VCFs that seek high returns view an IPO in their portfolio as a success. Because VCFs seldom sell at the IPO (Barry et al. 1990), the initial return after the IPO is not as important for independent VCFs. They use their reputation to certify the IPO and find good underwriters (Megginson and Weiss 1991), but they seldom hype the issue. According to studies in the U.S. VC market, the initial returns of VC-backed IPOs are less than non-vc-backed IPOs (Megginson and Weiss 1991). 2 However, the long-term market performance of VC- backed IPOs (Brav and Gompers 1997) is better than that of non-vc-backed IPOs because of the value adding and monitoring of VCFs after the IPO. However, finance-affiliated VCFs may work hard during the IPO process as their value added with the help of their parents. According to Hamao, Packer, and Ritter (2000), in the Japanese IPO market, underpricing of VC-backed IPOs is higher than that of non-vc-backed IPOs. Kutsuna, Cowling, and Westhead (2000) attribute this phenomenon to good marketing rather than to lower pricing, and they find an inferior short-run (three- and six-month) market performance of IPO firms with high levels of VC investment (more than 3% stake). Given this evidence, we can state that the effect of VCF characteristics is reflected in the performance difference between their portfolio companies in the following ways: 1. Because independent VCFs seek high returns, they are more likely to invest in high-tech industries. 2. Independent VCFs prefer to invest in start-up companies, and thus the length of time between their first investment and the IPO could be longer than that of finance-affiliated VCFs. 3. Because independent VCFs are more likely to invest in their portfolio companies at an early stage, they may engage other VCFs in syndications, thus spreading the high risks associated with their investments (Lerner 1994). 4. Because professionals in independent VCFs are more experienced and can provide more nonfinancial services to portfolio companies, they are 2 Recent studies on U.S. IPOs show a reverse pattern on underpricing. It may be caused by environmental and legal changes.

9 Differences in Performance 67 more likely to participate in the management. One important measure of this participation is the proportion of their seats on the board of directors. We expect to observe a higher number and percentage of board seats held by independent VCFs. 5. The difference between independent and finance-affiliated VCFs is reflected in the IPO process and leads to differences in the IPO underpricing and long-term market returns. Based on these statements, we propose six hypotheses and test them in the Singapore market where both independent and finance-affiliated VCFs are equally weighted. H 1 (Industry): When we compare independent with finance-affiliated VC-backed IPOs, there will be more independent VC-backed companies in high-tech industries. H 2 (Investment Duration): When we compare independent with finance-affiliated VC-backed IPOs, the investment duration (time between the first investment and the IPO) of independent VC-backed companies will be longer. H 3 (Syndication): When we compare independent with finance-affiliated VC-backed IPOs, VC syndication will be more likely if the leading VC is independent. H 4 (Number of Board Seats): When we compare independent with finance-affiliated VC-backed IPOs, independent VCFs will hold more board seats in their portfolio companies, both in number and in percentage. H 5 (IPO Underpricing): When we compare independent with finance-affiliated VCbacked IPOs, the initial returns of independent VC-backed IPOs will be lower because of less marketing activity. H 6 (Long-Term Return): When we compare independent with finance-affiliated VCbacked IPOs, the long-term after-market returns of independent VC-backed IPOs will be higher due to more value adding of independent VCFs after the IPO. IV. Data Sample To study VC-backed IPOs, we collect information on companies listed on the Stock Exchange of Singapore (SES) between 1987 (the year the first VC-backed company, Amtek Engineering, was listed) and We identify venture capitalists through IPO prospectuses, which disclose the major pre-ipo shareholders of companies. We use annual reports and other public documents to verify the shareholder information found in the prospectuses. After examining all sources in the public domain, we identify sixty-four VC-backed companies. Our sample includes all VC-backed companies listed on the SES before July Because the listing requirements of the SES have been relaxed significantly since mid-1999 and these changes could affect the market

10 68 The Journal of Financial Research behavior, 3 we use July 1999 as a cut-off date to avoid the possible influence of the new listing rules. Because many IPO companies are backed by more than one VCF, we classify IPO companies by the type of the leading VC. Here, we use the same definition of leading VC as Gompers (1996), that is, the longest serving board member who is a venture capitalist. When this criterion is the same for two or more venture capitalists, we consider the equity stake and strategy of these venture capitalists. The equity stake usually decides the leading position. However, a VCF with an active monitoring strategy (usually an independent VCF) will be deemed the leading VC even if its equity stake is smaller than other VCFs that are passive investors. We use this guideline only once in our sample. We find that most of the time, the leading VC is independent if two or more VCFs are shareholders. Figure I shows the distribution of VC-backed listed companies in Singapore by type of VCF. As mentioned in section II, we combine corporate with independent VCFs and government-funded and finance-affiliated VCFs. We divide the sixty-four VC-backed listed companies into two roughly equal groups, thirty-three independent and thirty-one finance-affiliated VCFs, or 52% and 48%. V. Empirical Test We divide the six hypotheses into three groups. The first four hypotheses (industry, investment duration, syndication, board seat) describe the backgrounds of VC-backed companies. H 5 projects the IPO initial return whereas H 6 projects the long-term after-market return. Backgrounds of the Two Types of VC-Backed Companies We define the variables used to test H 1 to H 4 as follows. H 1. Industry Preference (Industry). Because all our sample companies are listed on the SES, we use the SES industry classification to distinguish high- from low-technology companies. The high-tech classification includes the information technology services and whole manufacturing sectors except for the printing subsector and the other manufacturing subsector. In Singapore, most of the manufacturing subsectors such as electronics and plastics are computer related. We classify all other sectors as low-tech, such as wholesale and transportation. We code Industry as 1 if an IPO company is in a high-tech industry and 0 otherwise. H 2. Investment Duration (Duration). We define investment duration as the time between a VCF s first investment in a company and the company s IPO. Most 3 The formal announcement of SES s new listing requirement was on September 20, 1999, but three IPOs had already been approved under the new criterion before that date.

11 Differences in Performance 69 Note: This figure describes the distribution of venture-capital- (VC) backed companies listed on the Stock Exchange of Singapore between August 1987 and July Data sources are initial public offering (IPO) prospectuses and first-year annual reports after the IPO. The total number of VC-backed companies is sixty-four. We identify the VC type according to the type of leading VC. The independent VC firm (VCF) is an independent partnership management firm and manages several VC partnerships. The finance-affiliated VCF is a subsidiary of financial institutions, such as insurance companies and commercial banks. The corporate VCF is a subsidiary of industrial corporations. The government-funded VCF is a firm funded by government with policy priorities. Figure I. Distribution of VC-Backed Companies Listed in Singapore by VC Type: (August 1987 July 1999). VC-backed companies in Singapore are in computer-related or low-tech industries, so their growth cycles are similar. Because the sample does not include biotech companies that require long-term R&D, investment duration is a good measure of a VCF s investment stage. The earlier the investment stage, the longer is the investment duration. Duration is measured in months. H 3. Syndication (VC Number). We use the pre-ipo number of VCFs listed as company shareholders to measure the syndication level of VCFs. If several VC funds are managed by one VCF, we count this as one shareholder. H 4. Board Seats (Board Number and Board Percentage). Because company boards vary greatly in size, we use two variables Board Number, the number of board seats held by venture capitalists, and Board Percentage, the percentage of VC board seats to measure the influence of VCFs in the board of their portfolios. Because the number of board seats is closely related to the equity holding, we introduce two controlling variables: Total VC and Leading VC. We define Total VC as the total VC equity holding percentage of the company just before the IPO and Leading VC as the equity holding percentage of the leading VC just before the IPO. In addition to these variables, we use Company Age, VC Age, and Proceeds (offering proceeds) to give a complete overview of VC-backed IPOs in Singapore.

12 70 The Journal of Financial Research We define Company Age as the time between the incorporation year of the company itself or its predecessor company (some companies are restructured and incorporated just before going public, so we base their year of incorporation on their predecessor) and the IPO. We define VC Age as the time between the year of the leading VCF s incorporation and the company s IPO. We define Proceeds as the amount of the total offering to the public. Table 2 presents a comparison of these variables for independent and finance-affiliated VC-backed IPOs. To measure the significance level of the TABLE 2. Descriptive Statistics on Difference Between IPOs Backed by Independent and Finance-Affiliated VCFs. Independent Finance Affiliated Sample Size VC Backed VC Backed Z-scores by (Independent/ Mean Mean t-statistics Mann-Whitney- Finance (Median) (Median) by t-test Wilcoxon Test Affiliated) Industry /31 (1.00) (0.00) Duration (month) /27 (20.5) (11.0) VC Number /31 (3.00) (1.00) Board Number /31 (1.0) (0.0) Board Percentage 19.0% 10.6% /31 (16.7%) (0.0%) Total VC 17.6% 18.1% /31 (17.5%) (12.9%) Leading VC 13.7% 16.0% /31 (15.0%) (12.0%) Company Age (year) /31 (14.0) (14.0) VC Age (year) /31 (7.00) (7.00) Proceeds /31 (S$ million) (17.8) (10.34) Note: This table describes and compares independent and finance-affiliated venture-capital- (VC) backed initial public offerings (IPOs). The sample comprises sixty-four VC-backed IPOs listed on the the Stock Exchange of Singapore between August 1987 and July We define Industry as equal to 1 if the IPO company is in a high-tech industry and 0 if in a low-tech industry. Duration is the time between a VCF s first investment in a company and the company s IPO. Board Number is the number of board seats held by venture capitalists. Board Percentage is the percentage of VC board seats. Total VC and Leading VC are total VC shareholdings and leading VC shareholdings, respectively, based on pre-ipo data. Proceeds is the amount of the total IPO to the public. The t-statistics and Z-scores compare the mean and median differences between independent and finance-affiliated VC-backed IPOs, respectively. We calculate the two-tailed t-statistics based on an F-test for equal variance assumption. Significant at the 1% level. Significant at the 5% level. Significant at the 10% level.

13 Differences in Performance 71 two-type difference, we use a two-tailed t-test (t-statistics) and a nonparametric Mann-Whitney-Wilcoxon test (Z-scores). The table shows several similarities between independent and financeaffiliated VC-backed IPOs. Proceeds, Total VC, Leading VC, Company Age, and VC Age of both types are close. The similar Company Age and VC Age could exclude the grandstanding effect (Gompers 1996) in our analysis. 4 Table 2 also supports H 1 to H 4. We see that independent VCFs invest significantly more in high-tech industries than do finance-affiliated VCFs. The percentage difference between the two types is more than 20% (67% compared with 42%). Thus, H 1 is proven. Investment duration is shown to be different between independent and finance-affiliated VCFs. Duration for IPOs backed by independent VCFs is longer, though the significance level is statistically weak (between 5% and 10%). The median Duration of finance-affiliated VCFs is 11 months, compared with 20.5 months for independent VCFs. For 52% of the finance-affiliated VCFs (34% of the independent VCFs) sample, Duration is less than one year. If we remove the four portfolio companies of Rothschild Venture, whose investment behavior is a hybrid of independent and finance-affiliated VCFs, the mean Duration of finance-affiliated VCFs decreases to 15 months with and is statistically significant (t-statistic = 2.06, Z-score = 2.11) on the two-type difference. Thus, we can conclude that H 2 is valid; that is, there is a significant investment duration difference between independent and finance-affiliated VC-backed IPOs. Finance-affiliated VCFs tend to invest in the later stage. Table 2 shows there is a significant difference in VC Number in an IPO company. On average, IPOs backed by independent VCFs are supported by three VC investors, but IPOs backed by finance-affiliated VCFs are supported by fewer than two VCFs. This finding strongly supports H 3 ; that is, independent VCFs use syndication more frequently to spread the risk of their investments. Financeaffiliated VCFs tend to invest at a later stage with lower risks, and thus their motivation to syndicate is lower. The mean differences on Board Number (1.3 compared with 0.6) and Board Percentage (19% compared with 11%) between the two types are significant. The differences support H 4 ; that is, independent VCFs take a more active role in company management by holding more board seats compared with finance-affiliated VCFs. 4 Gompers (1996) defines grandstanding as a VCF s behavior in prematurely launching its portfolio companies to the public. The motive of the VCF is self-interest (the success of follow-on fund raising). It happens mostly among young VCFs. IPOs backed by young VCFs have deeper underpricing, lower investment duration, smaller VC stake, and closer timing between the IPO and the VCF s next fund raising. Furthermore, grandstanding effects are likely to be stronger among independent VCFs because of the reputation effect. Thus, the exclusion would cause a negative bias (finance-affiliated VCFs better than independent VCFs) in this study.

14 72 The Journal of Financial Research Between the two types, VC equity holdings are almost the same. Thus, we can exclude the effect of equity holdings on differences between the two types. The VCF s board seat differences can only attribute to the quality difference between the finance-affiliated and independent VCF. The low number of board seats held by finance-affiliated VCFs (the mean of 0.6 indicates many of them do not sit on the board at all) also shows that their investment behavior is different from traditional VC practices. Finance-affiliated VCFs might treat their investments with the same hands-off approach their parent companies use for institutional investments. Differences in the IPO Process To study the effect of VC type on the IPO process, we examine IPO underpricing and related factors, such as underwriter quality, auditor quality, hot issue period, price-to-earnings ratio, and book-to-market ratio. We define these factors as follows. Underwriter Quality (Underwriter). Following Megginson and Weiss (1991), we use the relative market share of underwriters to measure underwriter quality because it is continuous and relatively easy to construct. We define underwriter quality as the total dollar amounts for IPOs by a certain underwriter divided by the total IPO amount on SES between 1987 and The two largest underwriters, Development Bank of Singapore (DBS) and United Overseas Bank (UOB), account for 43% and 12% of the market share, respectively. In our sample, DBS underwrites exactly half the IPOs and has a larger market share among VC-backed IPOs. Auditor Quality (Auditor). We follow the approach of Feltham, Hughes, and Simunic (1991) and differentiate between the Big Six 5 (Arthur Andersen, Coopers & Lybrand, Deloitte & Touche, Ernst & Young, KPMG, and Price Waterhouse) and local small auditors. Companies audited by one of the Big Six are coded 1, and 0 otherwise. The Big Six audit 87.5% of the sixty-four companies in our sample. Hot Issue Period (Hot). This is the period when IPO initial returns are very high. As Ritter (1984) observes, hot issue periods occur in the stock market when the average IPO initial return is significantly higher than the long-term average. There were two hot issue periods between 1987 and July 1999: December 1992 to April 1994 (due to the global capital inflow to East Asia) and April 1999 to July 1999 (due to the recovery from the Asian economic crisis). Hot is a binary variable. Shares listed in these two periods are coded 1, and 0 otherwise. During these hot periods, IPO underpricing was high for both independent and finance-affiliated VC-backed IPOs. If we exclude IPOs in these periods, we can 5 Since this article was completed, Coopers & Lybrand and Price Waterhouse merged, and the Big Six have become the Big Five.

15 Differences in Performance 73 reduce data variation and show the VC type effect more clearly. In the underpricing comparison, we present the underpricing both for the whole period and for the nonhot issue period. Pricet-to-Earnings Ratio (P/E Ratio) and Book-to-Market Ratio (B/M Ratio). We define P/E Ratio as the ratio of the offering price to earnings per share in the last fiscal year before the IPO. We define B/M Ratio as the ratio of net tangible assets per share to the offering price in the same year. These two ratios measure the valuation of an IPO share at its offering price. IPO Underpricing ( Underpricing). We define Underpricing as the first trading day closing price minus the offering price, divided by the offering price. Table 3 compares Underwriter, Auditor, Hot, P/E Ratio, B/M Ratio, and Underpricing between the two types. We again use t-tests and Mann-Whitney- Wilcoxon tests to measure type differences. TABLE 3. Comparison of the IPO Process Between IPOs Backed by Independent and Finance- Affiliated VCFs. Independent Finance Affiliated Sample Size VC Backed VC Backed Z-scores by (Independent/ Mean Mean t-statistics Mann-Whitney- Finance (Median) (Median) by t-test Wilcoxon Test Affiliated) Underpricing 20.7% 33.4% /31 (10.5%) (16.7%) Hot 7.8% 22.9% /25 (2.4%) (15.5%) Underwriter 26.6% 22.3% /31 (43.0%) (12.0%) Auditor /31 (1.00) (1.00) P/E Ratio /31 (13.0) (11.3) B/M Ratio /31 (0.43) (0.46) Note: This table compares the initial public offering (IPO) process for independent and finance-affiliated venture-capital- (VC) backed IPOs. The sample comprises sixty-four VC-backed IPOs listed on the Stock Exchange of Singapore (SES) between August 1987 and July We define Hot as equal to 1 if the IPO date is in one of the hot issue periods, either December 1992 to April 1994 or April 1999 to July We define Underwriter as underwriter quality, determined by the relative market share of the underwriter on the SES. Auditor is a binary variable used to measure auditor quality; companies audited by one of the Big Six are coded 1 and 0 otherwise. P/E Ratio and B/M Ratio are the price-to-earnings ratio and book-to-market ratio, respectively, based on the IPO price. The t-statistics and Z-scores compare the mean and median differences, respectively, between independent and finance-affiliated VC-backed IPOs. We calculate the two-tailed t-statistics based on an F-test for equal variance assumption. Significant at the 1% level. Significant at the 5% level. Significant at the 10% level.

16 74 The Journal of Financial Research The table shows that companies backed by the two types of VCFs are similar in underwriter quality and audit quality, and in their valuation at the IPO, as measured by P/E Ratio and B/M Ratio. However, the mean difference in IPO underpricing is significant when we exclude the hot issue period (7.8% for independent VC-backed IPOs and 22.9% for finance-affiliated VC-backed IPOs). The median difference is weakly significant for the whole period. These findings support H 5, which projects deep underpricing for finance-affiliated VC-backed IPOs. To further understand the VC type effect, we use ordinary least squares to estimate our cross-sectional model on the IPO underpricing and show the results in Table 4. We follow Megginson and Weiss (1991) in choosing control variables (Proceeds, Company Age, and Underwriter). We add a binary hot issue factor from Ritter (1984) and the leading VCF s pre-ipo holding percentage from Lin (1996). Thus, Underpricing is the dependent variable, and VC Type (a binary variable, TABLE 4. IPO Underpricing by VC Type, Using Control Variables. Variables Coefficients t-statistics Constant (0.113) VC Type (0.056) Proceeds (0.033) Company Age (0.003) Underwriter (0.148) Hot (0.067) Leading VC (0.239) No. of observations 64 Adjusted R Note: This table presents the cross-sectional linear regression results for the initial public offering (IPO) underpricing for venture-capital- (VC) backed companies listed on the Stock Exchange of Singapore (SES) between August 1987 and July We classify VCFs into two types, independent and finance-affiliated, as measured by the dummy variable VC Type. VC Type is equal to 1 if the IPO is backed by an independent VCF and 0 if backed by a finance-affiliated VCF. Proceeds and Company Age are the natural logarithms of offering proceeds and company age, respectively. Underwriter is underwriter quality as determined by the relative market share of the underwriter on the SES. Hot is equal to 1 if the IPO date is in one of the hot issue periods, either December 1992 to April 1994 or April 1999 to July Standard errors appear in parentheses. Significant at the 1% level. Significant at the 5% level. Significant at the 10% level.

17 Differences in Performance 75 coded 1 for an independent VC-backed IPO and 0 for a finance-affiliated VCbacked IPO), Proceeds, Company Age, Underwriter, Hot, and Leading VC are the independent variables. Two variables, Proceeds and Company Age, are the natural logarithm of Proceeds and Company Age, respectively. The regression equation is as follows. IPO Underpricing = a 0 + a 1 VC Type + a 2 Proceeds + a 3 Company Age + a 4 Underwriter + a 5 Hot + a 6 Leading VC + e 1 (1) As expected, Table 4 shows that all control variables are related to IPO underpricing except for Underwriter. However, even controlling for these factors, the VCF type is still significantly related to underpricing. The negative sign of the coefficient ( 0.126) indicates that the underpricing of finance-affiliated VCbacked IPOs is more severe and that it is significant for the whole period. Hence, the regression analysis provides further support for H 5 ; that is, IPOs backed by financeaffiliated VCFs have higher initial returns compared with independent VCFs. As shown in Table 3, independent and finance-affiliated VCFs have similar levels of P/E Ratio and B/M Ratio. We could exclude the IPO pricing difference as the reason of the underpricing difference; however, further study on IPO short-term returns is needed to determine whether underpricing is the result of better IPO marketing by finance-affiliated VCFs. Differences in the Short- and Long-Term After-Market Performance In this subsection we study the influence of VC type on the short- and long-term after-market performance of IPOs. We examine short-term one- and three-month, and long-term one-, two-, and three-year buy-and-hold market returns. We choose October 1999 as the latest trading month, because of data availability and because of the possible effect of the trading currency change of foreign shares (from U.S. dollars to Singapore dollars). 6 We define the buy-and-hold market return for a given period as the monthly closing price (adjusted for dividend and bonus) after this period from the IPO minus the first-day closing price divided by the first-day closing price. For example, for a company listed in August 1994, monthly closing prices for one and three months, and one, two, and three years are the closing prices in September 1994, November 1994, August 1995, August 1996, and August 1997, respectively. Because of general market movement, raw market returns must be adjusted to reflect companies excess returns over the general market. 6 There are two kinds of shares, local shares and foreign shares, listed on the SES. Foreign companies are traded in U.S. dollars because they are not allowed to trade in the local currencies. Thus, their market price movement is different from local shares. However, this rule was changed in November 1999, and foreign companies have been allowed to trade in Singapore dollars since then.

18 76 The Journal of Financial Research Here, we use a market index to adjust the market return. Two market indexes, the UOB-SESDAQ index and the SES-Foreign index, provide measures for local and foreign shares in Singapore, respectively. Though the UOB-SESDAQ index is an index for shares listed on SESDAQ, the second board of Singapore stock market, the UOB-SESDAQ index is a good measure for small-capital shares on both SESDAQ and the SES Main Board. Because most local VC-backed shares are small capital shares, we calculate their market returns and adjust the market returns using the UOB-SESDAQ index return for the same period. For foreign shares, we use the SES-Foreign index to derive excess returns. Table 5 compares the one- and three-month, and one-, two-, and threeyear excess returns between independent and finance-affiliated VCFs. Because the data collection ends in October 1999, the long-term market returns for some new IPOs are not available, reducing our sample size. As in Tables 2 and 3, in Table 5 we measure two type differences by t-tests and Mann-Whitney-Wilcoxon tests. TABLE 5. Comparison of Short- and Long-Term Market Returns of Independent and Finance- Affiliated VC-Backed IPOs. Independent Finance Affiliated Sample Size VC Backed VC Backed Z-scores by (Independent/ Mean Mean t-statistics Mann-Whitney- Finance (Median) (Median) by t-test Wilcoxon Test Affiliated) One-month 2.5% 8.8% /31 excess return (2.1%) ( 10.1%) Three-month 12.6% 7.8% /31 excess return (8.6%) ( 8.5%) One-year 42.4% 1.7% /28 excess return (18.0%) ( 5.5%) Two-year 47.4% 3.5% /24 excess return ( 10.2%) ( 24.6%) Three-year 80.3% 11.7% /18 excess return ( 20.3%) ( 26.7%) Note: This table compares the short- and long-term market returns of independent and finance-affiliated venture-capital- (VC) backed initial public offerings (IPOs). The sample comprises sixty-four VC-backed IPOs listed on the Stock Exchange of Singapore (SES) between August 1987 and July Our data sources are public database Estimates Direct on share price and SES Fact Books in various years on market indexes. The market return is the buy-and-hold return. We define this return as the monthly closing price for a certain period after the IPO, minus the first-day closing price, divided by the first-day closing price. The excess return is the market return adjusted by the market index, the UOB-SESDAQ index for local shares and the SES-Foreign index for foreign shares. The t-statistics and Z-scores compare the mean and median differences between independent and finance-affiliated VC-backed IPOs, respectively. We calculate the two-tailed t-statistics based on an F-test for equal variance assumption. Significant at the 1% level. Significant at the 5% level. Significant at the 10% level.

19 Differences in Performance 77 The table shows significant short- and long-term return differences among IPO companies. There are significant differences in both one- and three-month market returns. Short-term returns of independent VCFs are much higher than those of finance-affiliated VCFs. When we compare these results with those in Table 3, we see that finance-affiliated VC-backed IPOs show higher first-day returns but demonstrate poorer market returns one to three months after the IPO compared with independent VC-backed IPOs. If we calculate the one-month market-adjusted return based on the IPO offering price rather than on the first-day closing price, we find that market returns of the two groups are similar. The mean one-month return of independent VCbacked IPOs is 22.9% and that of finance-affiliated VC-backed IPOs is 21.2%. The result of our t-test (t-statistic is 0.18) shows no significant mean difference between the two types. Because IPO shares are priced similarly at the IPO, we attribute the deeper underpricing of finance-affiliated VC-backed companies to the better marketing of VCFs or their underwriter associates. For the long-term market return, we see that the difference between the two types decreases as firms age. There is a significant difference in the one-year return, a weakly significant difference in the two-year return, but no significant difference in the three-year return. However, the market return of independent VC-backed IPOs is consistently higher than that of finance-affiliated VC-backed IPOs. Even for the three-year return, the independent VC-backed IPO shows an 80.3% in mean return, but the finance-affiliated VC-backed IPO shows an 11.7% in mean loss. Further analysis shows similar results if we change the market return basis from the first-day closing price to the IPO offering price. Thus, the results support H 6, which states that independent VC-backed IPOs will have better long-term market returns compared with finance-affiliated VC-backed IPOs. To further understand the effect of VC type, we estimate a cross-sectional model on long-term market returns, using ordinary least squares. We follow Brav and Gompers (1997), choosing the logarithm of B/M Ratio and the logarithm of Firm Size as our control variables. Our dependent variables are the logarithms of the excess one-, two-, and three-year market returns. Thus, we have three regression equations. Table 6 presents the regression results. The table shows that the regression equations gradually lose their explanatory power with time. When IPO firms age, factors that were influential at the IPO gradually reduce their effect. Two controlling factors, B/M Ratio and Firm Size, are not related to all the long-term returns, perhaps because of the small size of the capital market in Singapore, where all of the shares are closely followed by market investors. Furthermore, Table 6 shows that VC type is significantly related to the one-year return, weakly significant for the two-year return, and insignificant for the three-year return. These results are the same as those of the two-group comparison in Table 5. One possible explanation is that independent VCFs add more value