Ownership Concentration and Initial Public Offering Performance: Evidence from Thailand

Ownership Concentration and Initial Public Offering Performance: Evidence from Thailand Abstract This study examines the relation between ownership concentration and performance of initial public offerings (IPOs) in Thailand during 1989-1993. Ownership concentration plays a crucial role in emerging capital markets such as Thailand and can significantly affect IPO performance. Several testable hypotheses are constructed from the IPO literature to examine short-term and long-term IPO performance and their relation with ownership concentration. The empirical evidence reveals a positive relation between ownership concentration and IPO initial returns. This finding indicates that large, informed investors have the ability to identify underpriced IPOs, leaving uninformed investors with less underpriced issues. The findings are significant because the degree of information asymmetry is typically very high in emerging markets. However, evidence on the relation between initial ownership concentration and IPO long-run performance is not conclusive. C

Ownership Concentration and Initial Public Offering Performance: Evidence from Thailand Abstract This study examines the relation between ownership concentration and performance of initial public offerings (IPOs) in Thailand during 1989-1993. Ownership concentration plays a crucial role in emerging capital markets such as Thailand and can significantly affect IPO performance. Several testable hypotheses are constructed from the IPO literature to examine short-term and long-term IPO performance and their relation with ownership concentration. The empirical evidence reveals a positive relation between ownership concentration and IPO initial returns. This finding indicates that large, informed investors have the ability to identify underpriced IPOs, leaving uninformed investors with less underpriced issues. The findings are significant because the degree of information asymmetry is typically very high in emerging markets. However, evidence on the relation between initial ownership concentration and IPO long-run performance is not conclusive.

I. Introduction Ownership Concentration and Initial Public Offering Performance: Evidence from Thailand In emerging capital markets, initial public offerings (IPOs) perform one of the most crucial roles in resource allocation. In fact, an understanding of IPOs is an important first step in understanding emerging financial markets in general. One of the most peculiar aspects of IPOs is that new shares are being offered at prices significantly lower than the closing prices on the first trading day. This IPO underpricing is an international phenomenon documented in financial markets around the world (Loughran, Ritter and Rydqvist, 1994). Although these puzzling occurrences have prompted financial economists to investigate the causes underlying IPO underpricing, no single hypothesis has received overwhelming support as an explanation of this perplexing phenomenon. Moreover, financial economists are puzzled by poor long-run performance of IPOs in secondary markets. Understanding long-run performance of IPOs in emerging markets is also very important because it relates to the efficiency of the secondary market for IPOs and the capital market in general. The purpose of this study is to examine the short-run and long-run performance of initial public offerings in Thailand. Thailand provides a unique setting to test the relation because of the highly concentrated ownership structure. La Porta, Lopez-de-Silanes and Shleifer (1999) posit that, in most East Asian countries, corporate control is enhanced through pyramidal structures and cross-holdings among family-controlled firms. Further, the corporate governance structure in Thailand is relatively weak compared with other developed economies. Low transparency and the lack of disclosure are two major problems in the Thai market (Zhuang, 1

Edwards, Webb, and Capulong, 2000). Claessens, Djankov, Fan, and Lang (2002) report that concentrated control and the divergence between ownership and control in public corporations in eight East Asian economies diminish firm value, indicating the economic significance of the agency problem associated with ownership structure. Other studies in East Asia have also found that corporate governance factors affect firm valuation (Mitton, 2002; Lins, 2003). In this study, testable hypotheses on the relation between ownership concentration and IPO performance will be developed and examined using data from Thailand during 1989-1993. Previous IPO literature has suggested a number of possible explanations for IPO underpricing. Most of the rational expectation models suggest that information asymmetry among various participants is responsible for IPO underpricing 1. Baron (1982), for example, contends that underwriters have informational advantages over other participants in the markets and, therefore, tend to offer shares at discount to minimize share distribution efforts. Rock (1986), on the other hand, argues that underpricing is a result of information asymmetry between informed investors and uninformed investors. Subsequent empirical studies have been focusing on testing the possible explanatory power of the theoretical models. Consequently, several explanations of IPO underpricing are related to information asymmetry variables, such as the level of ex-ante uncertainty surrounding the intrinsic value of an issue (Beatty and Ritter, 1986), the quality of underwriters (Booth and Smith, 1986; Carter and 1 There are also other alternative explanations, such as fads (Aggarwal and Rivoli, 1990), hot issue markets (Ritter, 1984), industry effects (Ibbotson and Jaffe, 1975; Ritter, 1984), legal insurance (Tinic, 1988), monopoly power (Welch, 1992), and presale information gathering (Benveniste and Spindt, 1989). 2

Manaster, 1990), the quality of auditors (Titman and Trueman, 1986), retained ownership (Grinblatt and Hwang, 1989), and subscription levels (Koh and Walter, 1989). The rational expectation models (e.g., Rock, 1986; Beatty and Ritter, 1986; Carter and Manaster, 1990) yield rich predictions and explanations for IPO underpricing. Financial economists rely on these models to generate proxies of ex-ante uncertainty and test their effectiveness in predicting IPO underpricing. However, the literature has largely ignored the empirical implications of these models with regard to the ownership structure of IPO firms. In this paper, it is proposed that the rational expectation models also provide useful predictions with regard to ownership structure of IPO firms. For example, the adverse selection model by Rock (1986) implies a positive relation between ownership concentration and IPO underpricing. Specifically, the model assumes that informed investors have superior information to uninformed investors. Therefore, informed investors will subscribe to only underpriced issues and let uninformed investors subscribe to overpriced issues. If we assume that, especially in emerging markets, large investors are also informed investors, we can predict that action by large, informed investors should lead to highly concentrated ownership structures for underpriced IPOs. On the other hand, overpriced IPOs will have many small, uninformed owners which, in turn, leads to low ownership concentration. From this example, it is plausible that the existing theoretical models also provide predictions with regard to the relation between ownership structure and IPOs performance. However, the literature on IPOs has not thoroughly examined this relationship. Not until recently have financial economists begun to examine the role of ownership structure in IPOs. In a recent paper that examines the 3

IPOs mechanism, for instance, Stoughton and Zechner (1998) put forth the notion that ownership structure can also affect IPO underpricing. Consequently, the key research question is: What is the relation between ownership concentration and the short-run and long-run performance of IPOs? This study attempts to answer this question by deriving the empirical implications from the existing theoretical models (e.g., Rock, 1986; Carter and Manaster, 1990). Based on the implications from the models, testable hypotheses regarding the relation between ownership concentration and IPO performance are developed and tested using IPO data from the Stock Exchange of Thailand (SET). In an emerging capital market like Thailand, ownership structure plays a very important role due to a relatively undeveloped market structure. Further, the degrees of information asymmetry among participants in the Stock Exchange of Thailand should be much higher than among participants in developed countries. Consequently, large investors are generally more informed than small investors. In Thailand, further, top-five ownership concentration is approximately 55 percent, compared to 25.5 percent in the US (Demsetz and Lehn, 1985) and 33.1 percent in Japan (Prowse, 1992). Altogether, these factors allow a better distinction between informed and uninformed investors in the market by using holding size as the main indicator. Testing Rock s (1986) hypothesis in Singapore, Koh and Walter (1989) also use the subscription level (number of shares in a lot) as a proxy for informed demand. Because of the importance of IPOs in the economic development process, moreover, there is a need to understand the performance of IPOs in Thailand. This study also simultaneously examines explanatory power of various information asymmetry models in comparison with that of the ownership concentration variables, e.g., top-five ownership concentration. The results should 4

represent an important first step in understanding the IPO market in Thailand and yield additional insights regarding determinants of IPO underpricing and IPOs long-run performance in emerging capital markets. Overall, it is documented that the IPO market in Thailand during 1989-1993 can be characterized as a hot issue market. Yearly average initial returns on the first day range from 127% in 1989 to 43% in 1992. Top five shareholders account for 55.7% of outstanding shares. Most importantly, the empirical results show that there is a positive relation between top-five ownership concentration and initial underpricing. Institutional investor ownership also positively affects initial underpricing. These findings provide empirical support to the empirical implications derived from the adverse selection models (Rock, 1986; Carter and Manaster, 1900) and the agency framework (Stoughton and Zechner, 1998). The results are robust even after including other variables controlling for ex-ante uncertainty about the intrinsic value of IPOs. Issue size, an ex-ante uncertainty proxy (Beatty and Ritter, 1986), is negatively related to initial returns. In support of the certifying hypothesis by Booth and Smith (1986), further, underwriter prestige is negatively related to underpricing. In summary, it is concluded that initial ownership concentration significantly affects the short-run performance of IPOs in Thailand. The overall empirical evidence on the aftermarket performance (24 months) yields mixed results. IPO stocks exhibit a pattern of long-run performance somewhat similar to that documented in other capital markets. In addition, the relation between initial ownership concentration and IPO long-run performance is examined. The purpose is to see whether information on ownership concentration at the time immediately after an IPO can be used to predict future performance. It is hypothesized that informed investors affect long-run 5

performance of IPO stocks. The empirical results reveal that IPO stocks with high initial ownership concentration have lower cumulative adjusted returns (-11.33%) than IPO stocks with low initial ownership concentration (1.48%). Further, the ratio of IPO holding period returns to market holding period returns for IPO stocks with high ownership concentration is 0.868 while the same ratio for IPO stocks with low ownership concentration is 1.021. Only stocks with high ownership concentration exhibit underperformance. In fact, IPOs with low ownership concentration exhibit performance similar to the overall market. Regression results show a relation between change in ownership concentration and long-run IPO performance. Overall, the conclusion is that there is empirical evidence in support of the relation between ownership concentration and IPO long-run performance. This paper will be organized along the following lines. The next section provides a review of the existing theoretical models and formulates hypotheses to be tested along with other explanations of IPOs performance. The third section describes the Stock Exchange of Thailand (SET) and briefly explains the listing procedures required by the Stock Exchange of Thailand and the Securities Exchange Commission of Thailand (SECT). The fourth section describes the data and methodology used in this study. The fifth section presents the empirical results. Finally, the sixth section concludes the study. II. Ownership Concentration and IPOs Performance Recently, the effect of ownership concentration on firm value has received considerable attention from financial economists. The recent finance literature on the effect of ownership concentration on firm performance suggests that share ownership by large US investors can 6

influence firm behavior and market value (e.g., Bathala, Moon and Rao, 1994; Bhide, 1993; Brickley, Lease and Smith, 1988; Holderness and Sheehan, 1988; Holthausen, Leftwich and Mayers, 1990; Jarrell and Poulson, 1987; Jensen, Solberg and Zorn, 1992; McConnell and Servaes, 1990; Morck, Shleifer and Vishny, 1988; Wruck, 1989). Overall, previous studies conclude that ownership concentration can significantly affect firm value and performance. However, existing literature focuses mostly on seasoned stocks. The relation between ownership concentration and firm performance during and after initial public offerings has been largely unexplored. Although several of the existing theoretical models (e.g., Rock, 1986; Carter and Manaster, 1990) indirectly provide predictions regarding the relation between ownership concentration and IPO performance, the literature has focused mainly on proxies of information asymmetry and ex-ante uncertainty. The main purpose of this section is to develop, from the existing literature, testable hypotheses regarding the effect of ownership concentration on the value of the IPO stocks and to examine the hypotheses empirically in Thailand using IPO data during 1989-1993. As mentioned earlier, Thailand makes an interesting test setting because it is feasible to distinguish informed investors based on holding size. Although the use of balloting ensures fairness among investors who subscribe for the same number of shares, this allocation procedure favors large, informed investors who subscribe to a large number of IPO shares. Therefore, large investors should have a higher probability of being drawn from the pool. This section begins with a brief review of the existing theoretical models by Rock (1986), Carter and Manaster (1990) and Stoughton and Zechner (1998). 7

2.1 Theoretical Models of IPOs In one of the most well-known theoretical models explaining IPO underpricing, Rock (1986) suggests that underpricing is a consequence of rational behavior by issuing firms. This is due to the information asymmetry between two major groups of investors. The first group of investors has perfect information regarding the prospects of the issues and, therefore, is considered informed investors. The second group of investors is considered uninformed investors because they have less knowledge regarding the intrinsic value of issues than the informed investors do. As a result of this information asymmetry, informed investors only compete for good, underpriced issues and leave inferior, overpriced issues to small, uninformed investors. Consequently, uninformed investors receive a disproportionately larger numbers of overpriced issues creating an adverse selection mechanism. To alleviate this adverse selection problem, Rock (1986) argues that issuing firms have to underprice IPOs in order to induce participation by uninformed investors. If the model is correct, weighting the returns by the probabilities of obtaining an allocation should leave uninformed investors earning the riskless rate (Rock, 1986, p. 205). In a theoretical model emphasizing the role of investment banker reputation, Carter and Manaster (1990) also focus on information gathering activities by investors. Specifically, Carter and Manaster (1990) argue that initial public offerings are often associated with high uncertainty which, in turn, requires information acquisition efforts by investors. However, investors will specialize in acquiring information because they have scarce resources to invest in the information gathering process. Incidentally, these information gathering activities increase the number of informed investors and worsen the adverse selection bias against uninformed 8

investors. In addition, Carter and Manaster (1990) posit that investment bankers can provide a positive signal of IPO quality. Specifically, prestigious underwriters will select only less speculative offerings. At the same time, high quality firms will try to send positive signals by choosing prestigious investment bankers which, in turn, leads to low IPO underpricing. In a recent paper, Stoughton and Zechner (1998) develop a theoretical model explaining the effect of different initial public offering mechanisms on the resulting structure of share ownership. The major conclusion of their paper is that the offering mechanism directly affects the ownership concentration of the firm at its initial public offering. Subsequently, the value of a firm s IPO is also determined by the ownership concentration. In their model, one of the most important assumptions is that investors do not have equal ability to monitor the management of new public companies. Stoughton and Zechner (1998) argue that only large, informed investors have resources and ability to perform such tasks and, therefore, hold an advantage over small, uninformed investors. However, there are agency problems which can creates a conflict for risk-sharing and information production. The reason for the conflict is that monitoring activities are costly and difficult to observe among the two classes of investors. Therefore, small investors can enjoy a free ride and receive the benefits, thus creating an agency conflict. Knowing that a free-rider problem exists, large investors demand additional compensation for their services in the form of underpricing. Based on this agency framework, Stoughton and Zechner (1998) suggest that the optimal offering process will, within regulatory boundaries, give favorable treatment to large investors. Hanley and Wilhelm (1995) find empirical evidence consistent with this prediction in the United States. 9

2.2 Empirical Implications This section develops a number of empirical implications with regard to the relation between ownership concentration and IPO performance from the theoretical models by Rock (1986), Carter and Manaster (1990) and Stoughton and Zechner (1998). First of all, it is posited that the adverse selection model by Rock (1986) not only explains why underpricing initially occurs, but also indirectly provides predictions with regard to the relation between ownership concentration and IPO underpricing. According to Rock (1986), for example, the equilibrium underpricing is determined by the proportion of informed investors. Previous studies generally assume that large and established customers are more likely to be informed investors (e.g., Michaely and Shaw, 1994). This is usually an accurate assumption in emerging capital markets, where the degree of information asymmetry is very high. Large investors have resources to acquire information and become informed with regard to the prospects of IPOs. In a test of the adverse selection model (Rock, 1986), Koh and Walter (1989) also use the subscription level (number of shares in a lot) as a proxy for informed demand. In effect, they assume that there is a positive relation between the lot size and the informed level of investors. Lee, Taylor and Walter (1996b) also use the same variable as a measure of informed demand in their study of short-run and long-run performance of IPOs in Singapore. With this assumption, a prediction with regard to initial ownership concentration of IPOs in the aftermarket can be formulated. If the assumption is not accurate, there should be no significant relation. Although Rock s (1986) model suggests that IPO underpricing compensates small, uninformed investors for the risk of trading against large, informed investors with superior 10

information, the model also implies that the proportion of large, informed investors can increase the equilibrium underpricing. According to Rock (1986), large investors will subscribe to only underpriced issues and let small investors subscribe to overpriced issues. Therefore, it is hypothesized that the action by large, informed investors should lead to highly concentrated ownership structures for underpriced IPOs. On the other hand, overpriced IPOs will have many small, uninformed owners, which will in turn lead to low ownership concentration. Consequently, the expected results would be a positive relation between the number of large shareholders and initial underpricing and a negative relation between the number of small shareholders and initial underpricing. Koh and Walter (1989) find that the unique public rationing process in Singapore results in a riskless rate of return for uninformed investors, providing empirical support for Rock s (1986) model. Although Koh and Walter (1989) focus on the probability of obtaining an allocation and return to uninformed investors, their empirical results also indirectly provide support for the hypothesis described above. Specifically, their results show that Singaporean IPOs with high proportion of the issue allocated to small investors tend to have lower returns than IPOs with low proportion of the issue allocated to small investors. Further, it is posited that the theoretical model by Carter and Manaster (1990) indirectly provides a similar prediction to that from Rock s (1986) model. According to Carter and Manaster (1990), IPOs are highly risky and, therefore, require information gathering efforts by investors. The information acquisition in turn creates a group of large, informed investors who have information superior to other investors. Based on the information acquired, large, informed investors can focus on IPOs with high initial returns and leave IPOs with low initial 11

returns to small, uninformed investors. This information asymmetry attenuates the adverse selection problems faced by small, uninformed investors. Again, a similar pattern emerges. It is expected that large, informed investors should be concentrated in IPOs with high underpricing, whereas small, uninformed investors should be presented among IPOs with low underpricing. Based on Carter and Manaster (1990), a positive relation between ownership concentration and initial underpricing is expected. In other words, IPOs with more large, informed investors are associated with higher returns than those with more small, uninformed investors. A similar prediction with regard to the relation between ownership concentration and IPO underpricing can also be derived from an agency framework. With heterogeneity of investor groups and inherent agency problems, Stoughton and Zechner (1998) predict that the ability to ration in favor of large shareholders should be positively related to initial underpricing. Although the paper by Stoughton and Zechner (1998) analyzes the impact of the IPO mechanism and the role of underpricing and rationing, the assumptions underlying their model also indirectly provide predictions with regard to the relation between ownership concentration and IPO underpricing. Specifically, Stoughton and Zechner (1998) assume that large, informed investors have superior information to small, uninformed investors, who can freeride on the monitoring efforts by large investors. Consequently, large informed investors require additional compensation for their monitoring services and demand high initial underpricing. If the mechanism described by Stoughton and Zechner is accurate, a positive relation between ownership concentration and initial underpricing would be observed. 12

It is also proposed that the existing theoretical models indirectly imply the relation between ownership concentration and IPO long-run performance. Specifically, the theoretical models by Rock (1986) and Stoughton and Zechner (1998) imply that large shareholders may have some ability to predict the long-run performance of IPOs. The reason is that large investors have mechanisms to gather relevant information and, therefore, are assumed to be informed investors (Michaely and Shaw, 1994). As a result, it is predicted that IPOs with a high number of large shareholders (high concentration) will perform better than IPOs with a few large shareholders (low concentration) in the aftermarket. This prediction provides a significant contribution to the IPO literature, which suggests that IPOs generally underperform the overall market in the long-run (Ritter, 1991). The behavior of large shareholders of IPOs in emerging markets has not been widely examined. Kim, Kitsabunnarat, and Nofsinger (2002) examine the performance of Thai firms after their public offerings and note declining performance. Both the low and high levels of managerial ownership are positively related to the change in performance, lending support to the alignment-of-interest hypothesis. However, the authors also find support for the entrenchment hypothesis by discovering a negative relation between performance and intermediate levels of managerial ownership. Further, it is believed that large shareholders, as a group, may have superior information to other investors. Therefore, a positive relation between ownership concentration and the long-run aftermarket performance is predicted. With regard to the long-run performance of IPOs, a competing prediction is also offered. In the equilibrium model by Carter and Manaster (1990), they argue that the IPO risk, as measured by the dispersion in aftermarket prices, is positive related to the proportion of 13

large, informed investor participation in IPOs. Consequently, Carter and Manaster (1990) suggest that IPOs with high proportion of large, informed investors are highly risky. Although IPOs with many large owners generate high initial returns, these IPOs may not exhibit performance superior to other IPOs with fewer large, informed investors. There is also a possibility that large, informed investors fail to perform their monitoring duties as assumed by Stoughton and Zechner (1998). This, along with the fact that these IPOs may be highly risky, can lead to poor long-run performance of IPOs with high proportion of large, informed investors. As a result, a competing hypothesis suggests that high ownership concentration will be related to low performance in the long-run. In summary, the existing theoretical models are used to develop several important predictions with regard to the relation between ownership concentration and IPO performance. Although the theoretical models by Rock (1986) and Carter and Manaster (1990) emphasize information asymmetry among participants, these models also rely on several crucial assumptions that also have significant implications with regard to the ownership concentration of IPO firms. However, these implications have never been examined empirically. By testing the implications derived from the existing theoretical models, we should gain additional insight into the puzzling issue of IPO performance. III. The Stock Exchange of Thailand (SET) Established as the country s only stock market in 1975, the Stock Exchange of Thailand (SET) has become one of the fastest growing capital markets in the world. During the first few years of operation, market activity was quite low because of the lack of public interest 14

(Wethyavivorn and Koo-smith, 1991). After a long period of stagnation during 1979-1985, the market, benefiting from rapid economic and industrial growth, entered its first boom period in 1986. Despite the worldwide market crash in 1987, the SET revived rapidly due to the strong economy. Stock prices tripled over the next three years until the market experienced its first crash as a result of the Gulf Crisis in 1990. Since then, the market has risen steadily and reached its record high in the fourth quarter of 1993. In 1994, the SET Index closed at 1,360.09 points. Total turnover value reached 2,113,860.65 million Baht. Figure 1 shows the performance of SET since its inception in 1975. The number of listed companies quadrupled during 1981 to 1993. In 1994, the exchange listed about 430 companies with a total market capitalization of $142 million, making it the second largest stock market in Southeast Asia and representing 5% of the emerging markets. The SET is organized into 32 sectors representing major industries such as agribusiness, building materials, chemicals, commerce, communication, electronics, energy, insurance, machinery, property and development and textiles. Three major sectors, banking, finance and property, account for a majority of the trading volume and offer investors the greatest liquidity. [Insert Figure 1 Here] 3.1 Listing Procedures in Thailand Originally, the SET acted as both the marketplace and the regulatory agency. Sidney Robbins, a former officer of the US Securities and Exchange Commission, suggested the structure of the SET during the early years. Since its inception, the SET was operating under the SET Act of 1974. Following the enactment of the Securities Exchange Act of 1992 (SEA), 15

which replaces the SET Act of 1974, the Securities and Exchange Commission of Thailand (SECT) was established in 1992 to regulate the market and to oversee the public offerings process. Under the new Act, the Securities and Exchange of Thailand has the authority to supervise the public offerings process by using the Approval and Filing System indicated in the Act. One of the main objectives of the SEA of 1992 is to protect investors by requiring better disclosure of information and providing a more solid regulatory framework. Before 1993, there were two kinds of companies in the Stock Exchange of Thailand. In fact, listing requirements are different for listed and authorized companies. Listed companies are those that have (1) paid-up capital of at least Baht 20 million, (2) a minimum of 300 shareholders each holding no more than 0.5% and collectively owning at least 30% of paid-up capital. Authorized companies are those that have a minimum of 200 shareholders owning collectively at least 25% of the paid-up capital. 2 Any proposed changes in capital must be submitted for approval to the SET accompanied by a detailed memorandum outlining the use of proceeds together with financial projections and pro-forma balance sheet and income statements. As of July 1990, the SET established new requirements for all initial public offerings. A company wishing to qualify for listing on the Stock Exchange of Thailand must file an information booklet (prospectus) with the Securities and Exchange Commission of Thailand (SECT) and the SET. After the enactment of the SEA in 1992, the SECT takes the responsibility of regulating the primary market and assumes the supervision of the filing process. In order to comply with 2 As of July 1993, authorized companies were required to convert to listed companies within three years. 16

the SECT requirements, the company must provide the following data: (i) name and location of the company, (ii) proposes of the offer, (iii) clarification on whether the shares offered are existing shares or new shares, (iv) a detailed description of the offering including type and number of shares, par value and offer price, conditions of the offering, subscription procedures, allotment procedure, and underwriter terms, (v) average cost per share, (vi) background information on the issuer including past capital increases over the past three years, (vii) dividend policy, (viii) detailed financial statements, (xi) data on directors, officers and major shareholders, (x) name of auditor, (xi) Articles of Association and Memorandum, and (xii) date of publication of the information book, time and place of issue, and subscription documents. According to the listing requirements and procedures established by the SET, companies seeking listing on the SET must have an established operating record under substantially the same management and must demonstrate that their main business is beneficial to the Thai economy and society. After the company submits the listing application and other documents to the SECT, the company should also appoint a financial advisor, approved by the SET, to assist the company in the IPO process. Concurrently, the company must also submit the listing application to the SET. Once the paperwork is completed, the listing committee will consider the application. After reviewing the applications, the committee forwards its recommendation to the Board of Directors of the SET. The consideration can usually be completed within 30 days. If approved by the SECT and the SET Board of Governors, the stocks can commence trading within 5 days. The allocation procedure is non-discretionary. If the issue is oversubscribed, the authority will use a ballot procedure to ensure fairness. 17

IV. Data and Methodology The sample consists of 173 companies going public in Thailand during 1989-1993. The offering prices and listing dates are obtained from the New Listed Securities Report compiled by the Investor Services and Information Section, Public Relations Department of the Stock Exchange of Thailand (SET). Additional information (e.g., company age, underwriters, etc.) is also gleaned from prospectuses filed with the Security Exchange Commission of Thailand (SECT). Companies with incomplete records are deleted from the sample. Finally, financial statement and stock market data are drawn from the PACAP-Thailand Database provided by the Sandra-Ann Morsilli Pacific-Basin Capital Markets (PACAP) Research Center at the University of Rhode Island. In Thailand, all listed companies are required to report the proportion of shareholdings of the top ten largest shareholders at the end of each quarter to the Stock Exchange of Thailand (SET) and, after 1992, the Security Exchange Commission of Thailand (SECT). Therefore, the data on ownership structure and concentration for IPO firms will be available at the end of the first quarter after the IPO date. 3 The reason for using the end of quarter ownership data is that the investigation will be limited to large owners that maintain their holdings in the aftermarket. Obviously, large shareholders who purchase IPOs only for the initial returns (i.e., sell shares on the first few trading days) will not be able to provide monitoring services for other small 3 The following schedule is used to determine the appropriate date to collect ownership data: IPO Date December 16, year t-1 to March 15, year t March 16 to June 15, year t June 16 to September 15, year 15 September 16 to December 15, year t Quarter Date March 31, year t June 30, year t September 30, year t December 31, year t 18

investors as suggested by Stoughton and Zechner (1998) and, therefore, are eliminated from the analyses. Further, informed investors should want to hold shares for a long period of time to reap profits from good performance. Consequently, investors who subscribe to a large number of shares for a short period of time may not be considered informed investors. Therefore, the data availability constraint should not distort the empirical test of the hypotheses. Field (1995) also uses this procedure to collect institutional ownership data in the US sample. Quarterly ownership concentration data are obtained from the CD-ROM database compiled by the Stock Exchange of Thailand. 4 4.1 IPOs Initial Underpricing To measure the initial returns on the first trading day, raw returns (r i ) are calculated using the following formula: IR i = ( Pi S i ), (1) S i where P i is the closing price on the first trading day and S i is the subscription price. Marketadjusted returns are calculated using the following formula: AR it = R it - R mt, (2) 4 The database contains data on the proportions (number of shares and percentage) of outstanding shares held by each of the top ten shareholders at the end of the company s fiscal year. 19

where R it is the daily raw returns for IPO in month t and R mt is the market return on day t. Both the PACAP equally-weighted and value-weighted daily market returns are used to check for the robustness of results as suggested by Ritter (1991). Descriptive statistics for average market adjusted returns (AR t ) and cumulative market adjusted returns (CAR t ) across different firm attribute categories (e.g., industry, year of issuance, ownership concentration) are provided. Further, using a regression framework, this study examines other determinants of IPO underpricing along with ownership concentration. A number of common control variables are also included in the model. The explanatory power is then compared with that of ownership concentration variables. The following regression model will be performed on the IPOs sample. IR i = β 0 + β 1 FIVE i + β 2 Size i + β 3 Age i + β 4 SD i + β 5 DIV i + β 6 UW i + β j Ind j + β j Year j + e i, (3) where IR i is the measure of underpricing; FIVE i is the ownership concentration variable measured by the proportion of outstanding shares held by the top five shareholders reported at the end of the IPO quarter; Size i is the equity issue size (inflation adjusted to the 1993 value); Age i is the natural logarithm of one plus the number of years in operation before going public; SD i is the standard deviation of monthly returns for the twenty-five days after listing: DIV i is the first annual cash dividend yield after the IPO; UW denotes the underwriter quality dummy variable (1=High quality; 0=Low quality) as measured by the underwriter s frequency of IPO involvement during the study period; Ind i is a series of industry dummy variables; and Year i is a 20

series of dummy variable indicating the IPO year. Other control variables and alternative proxy variables in other regression analyses are included to examine the robustness of the results. The top five ownership concentration (FIVE) is included to test the hypothesis developed in this paper. 5 If large, informed investors are more likely to subscribe to underpriced issues (Rock, 1986; Carter and Manaster, 1990), there should be a positive relation between ownership concentration and IPO underpricing. A number of variables are also employed as proxies for the ex-ante uncertainty surrounding the IPOs (Beatty and Ritter, 1986). Following Ritter (1984), the risk proxy is the standard deviation of daily stock returns between the close of the first trading day and the close of the twenty-fifth day of trading. According to Beatty and Ritter (1986), there should be a positive relation between this proxy and initial underpricing. McGuinness (1992) finds a positive relation between 15-day standard deviation of daily stock returns and initial underpricing for IPOs in Hong Kong during 1980-1990. However, Lee, Taylor, and Walter (1996a) find an insignificant relation among Australian IPOs during 1976-1989. Adjusted net proceeds raised from the issue is used as a proxy for ex-ante uncertainty. Ritter (1984) argues that small, less established firms are related to a high degree of ex-ante uncertainty and, therefore, should require more underpricing than large firms. Kim, Krinsky and Lee (1993) obtain a negative relation between gross proceeds and initial underpricing for Korean IPOs during 1988-1990. McGuinness (1992) observes a similar relationship among IPOs in Hong Kong. However, Lee et al. (1996b) find an 5 Due to the number of missing values in the data, the proportion of top five shareholdings is selected as a proxy for ownership concentration. Demsetz and Lehn (1985) and Prowse (1992) also use top five shareholding as their proxy for ownership concentration in the U.S. and Japan, respectively. 21

insignificant relationship among Singaporean IPOs during 1973-1992. Also, firm age may be associated with ex-ante uncertainty and the difficulty in valuing a firm (Ritter, 1984). A negative relation between firm age and initial underpricing is expected because investors should be able to acquire pertinent information on established firms which, in turn, results in low ex-ante uncertainty. While Lee et al. (1996a) find supporting evidence for the negative relation in Australia, they observe an insignificant relation in Singapore (Lee et al., 1996b). Underwriter quality (UW) is included in the regression model to test for the effect of underwriter prestige on IPO underpricing. Baron (1982) posits that IPO underpricing is a result of information asymmetry between investment bankers and issuing firms. In general, investment bankers have superior information in terms of market conditions, investor contacts, or industry knowledge. Acting as the agents, therefore, investment bankers try to minimize their selling efforts by discounting offer prices. However, Beatty and Ritter (1986) argue that investment bankers have their reputation at stake in the process and risk losing market share if they set improper offering prices. Booth and Smith (1986) suggest that the reputation can serve as a bonding mechanism and an indication of underwriter quality. As a result, there should be a negative relation between underwriter quality and initial underpricing (Carter and Manaster, 1990). While Kim et al. (1993) find a negative relation between underwriter prestige and initial underpricing among Korean firms issuing new shares, McGuinness (1992) observe an insignificant relationship among IPOs in Hong Kong. Following Michaely and Shaw (1994), the first cash dividend yield (DIV) after the IPO is also included to test the signaling model. Allen and Faulhaber (1989) argue that, based on the signaling hypothesis, firms that underprice more are more likely to have higher dividends. To 22

control for the industry effects (Ibbotson and Jaffe, 1975; Ritter, 1984), the regressions use a series of industry dummy variables based on the SET Industry Code. Finally, yearly dummy variables are used to account for market conditions. Descriptions of variables are provided in Table 1. [Insert Table 1 Here] 4.2 Long-Run Analysis The long-run performance of IPOs will be measured using two different methods. The first method uses an equally-weighted (with monthly rebalancing) portfolio of IPOs and compares the performance to the market portfolio. The second method involves calculating the buy-and-hold returns for each IPO and constructing the wealth relative, the ratio of IPO holding period returns to the market holding period returns. Ritter (1991) suggests that the results may be sensitive to the choice of benchmark. Therefore, the performance of the IPOs portfolio will be compared with the following benchmarks: (1) the PACAP monthly equally-weighted returns and (2) the PACAP monthly value-weighted returns. Due to data availability, the analyses can be for only two years after each IPO. Following Ritter (1991), the wealth relatives are calculated using the formula shown below: Wealth Relatives = 24 ( 1 + r ) t = 1 24 ( 1 + r ) t = 1 ipo market, (4) where the numerator is the holding period returns for the IPOs and the denominator is the holding period returns of the market portfolios using the two benchmarks. A wealth relative 23

greater than one will indicate that IPOs outperform the market while a wealth relative smaller than one will indicate an underperformance. A regression analysis using the model in Fields (1995) and Ritter (1991) will examine the determinants of IPOs long-run performance. Adding ownership concentration variables modifies the original model. The regression model is shown below: R i = β 0 + β 1 FIVE i + β 2 IR i + β 3 Age i + β 4 Market + β 5 Vol i + β j Ind j + e i, (5) where R i is the raw two-year cumulative monthly return measured from the first aftermarket closing price, FIVE i is the proportion of outstanding shares held by top five owners, IR i is the market-adjusted initial return using the PACAP equally-weighted or value-weighted monthly market returns, Age i is the natural logarithm of one plus the number of years in operation before going public, Vol i is the annual volume of IPOs in the year of issuance, and Ind j is a series of industry dummy variables. To test the effect of ownership concentration on long-run performance, the model includes the top-five ownership concentration variable. The ownership concentration variable (FIVE) is used to examine the effect of ownership concentration on IPO long-run performance. If Rock (1986) and Stoughton and Zechner (1998) are correct, this variable should have a positive coefficient. On the other hand, the model by Carter and Manaster (1990) implies a negative coefficient for this variable. The initial underpricing (IR) is included to test the overreaction by investors (Ritter, 1991). A negative coefficient for this variable will indicate that stocks with high IPO underpricing tend to have poor long-run performance. Ritter (1991) suggests that a negative relation between initial returns and 24

aftermarket performance may also yield support for the lawsuit avoidance explanation (Tinic, 1988). Firm age is used as a proxy for ex-ante uncertainty (Beatty and Ritter, 1986). Following Ritter (1991), the overall market movement is also accounted for by including the market returns from the period corresponding to each IPOs. To control for industry effect (Ritter, 1991; Field, 1995), a series of industry dummy variables based on the combined SET Industry Code are included. Other ownership variables are later included in the regression models. V. Empirical Results 5.1 Short-Run Analyses Table 2 presents summary statistics for initial underpricing of initial public offerings in Thailand during 1989-1993. For the total sample, the average initial underpricing is 54%, much higher than that documented in the United States. This high degree of underpricing in Thailand is in line with underpricing documented in other regional emerging markets, such as Malaysia, Taiwan (Loughran et al., 1994) and Singapore (Lee et al., 1996b). Previous IPO studies in Thailand also document very high initial day returns in excess of 60% (Allen, Morkel-Kingsbury and Piboonthanakiat, 1999; Wethyavivorn and Koo-smith, 1991). High initial underpricing can be explained by the contractual arrangements in the market (Loughran et al., 1994). In Thailand, offer prices are set before information acquisition and the allocation process is nondiscretionary. These two factors can result in high underpricing. While the number of IPOs increases, the degrees of underpricing declines during the study period. In fact, the average initial return in 1989 was 127%, whereas initial return in 1993 was 45%. Overall, the empirical 25

results and those from other studies suggest that IPO initial returns in Thailand are declining since the late 1980s. This may coincide with the establishment of the Securities Exchange Commission of Thailand (SECT) in 1992 and provides partial support to the assumption underlying the lawsuit avoidance hypothesis (Tinic, 1988). [Insert Table 2 Here] Table 3 reports initial underpricing by SET industry classification codes. Initial returns vary widely from 109% in the vehicles and parts industry to -23% in the pharmaceutical and cosmetic industry. Notably, IPOs are concentrated in certain industries, such as agribusiness (26), building and furnishing materials (16), finance and securities (16), and property and development (23). As a group, companies in these industries account for almost half of the IPOs during the study period. Apparently, the growing stock market in Thailand contributes to the high number of IPOs in the finance industry. Recent financial deregulation by the Bank of Thailand also reinforces this trend. At the same time, the real estate and property development businesses were growing dramatically in Thailand. High land prices and strong demand for construction lead to a high number of IPOs in these industries. In fact, stocks in these industries also have the highest liquidity in the Stock Exchange of Thailand. Finally, the agricultural business has been the backbone of the Thai economy for a long time. These companies may have taken advantage of the favorable market for IPOs. [Insert Table 3 Here] 26

Table 4 provides IPO characteristics in Thailand. It should be noted that, after conversion (Baht 25 $1), gross proceeds in Thailand are, on average, much smaller than those in the United States. Like those firms in Singapore (Lee et al., 1996b), newly listed companies in Thailand are usually well-established firms with an average operating history of 15 years. Dividend yields are also quite low. The total sample is split into two groups based on top-five ownership concentration. The two groups have similar characteristics with respect to firm age, dividend yield and stock return volatility. However, IPO stocks with high ownership concentration have higher initial adjusted returns (61%) than IPO stocks with low ownership concentration (47%). Although this initial finding provides some support for the hypothesis described earlier, the difference is not statistically significant at conventional levels (t = 1.48, p- value = 0.13). On average, high ownership concentration IPOs are smaller issues than low ownership concentration IPOs. The average top-five ownership concentration among Thai IPOs in the sample is quite high (55.7%). In fact, the average top-five ownership concentration for all companies listed on the SET is 45%. [Insert Table 4 Here] Table 5 presents multiple regression results. In the first model (a), the initial return is regressed on the control variables related to information asymmetry. Issue size and underwriter prestige are negatively related to initial underpricing. Further, the coefficients for the yearly dummy variables are all negative and statistically significant, except for 1990. These findings indicate that the level of IPO underpricing in Thailand has been declining through the study period (1989-1993). The variable for top-five ownership concentration is then added into the 27