IPO Underpricing and Signalling Effects for Detection of Speculative Stocks. Abstract

Transcription

1 IPO Underpricing and Signalling Effects for Detection of Speculative Stocks Kulabutr Komenkul a,*, Janusz Brzeszczynski b, Mohamed Sherif a and Bing Xu a a Department of Accountancy, Economics and Finance, Heriot-Watt University, Edinburgh, EH14 4AS, United Kingdom b Department of Accounting and Financial Management, Newcastle Business School (NBS), Northumbria University, Newcastle upon Tyne, NE1 8ST, United Kingdom Abstract Using data from the Thai market about the, so called, Turnover List of speculative stocks spanning the period , we investigate the relationship between their IPOs pricing effects and their subsequent classification as speculative investments. Furthermore, we examine the signalling effects for the detection of speculative stocks by the degree of their prior IPOs underpricing. We employ the market-feedback hypothesis to investigate this signalling process. We found a significant positive relationship between the magnitude of the IPO underpricing and the probability of an IPO firm being classified officially as speculative on the Turnover List. We also found that a six-month abnormal return after going public increases the probability of speculative dealing in the SET IPOs. Interestingly, firms from the SET market that experience relatively insignificant underpricing and abnormal return are likely to be on the Turnover List more quickly after their IPOs. However, we found that in case of the firms from the MAI market there was no statistically significant relationship between the aftermarket return and the probability of Turnover List participation. Keywords: IPO underpricing; Signalling; Speculative stocks; Turnover List Corresponding author. Tel.: (+44) ; Fax: 44 (0) address: kk218@hw.ac.uk. (K. Komenkul)

2 1. Introduction The Stock Exchange of Thailand (SET) is unique in that, unlike other exchanges, it is comparatively new. When compared to other exchanges, the SET is moderately small and it has its own particular rules and regulations. Interestingly, investors in the Thai stock market do not pay taxes when making profits from trading securities (Capital Gains Tax: CGT) 1. On the other hand, they will pay 10% tax if they receive dividends. As a result, most of the individual investors in the Thai stock market prefer to purchase and speculate in common stocks rather than invest in them for a long horizon period to receive dividends. In Thailand there is also only a mild penalty in cases of stock price manipulation. As a result, the Thai stock market is in fact one of best targets for various manipulators (stock-price fixers) and speculators. Consequently, there are now many speculative stocks being traded on the SET. Broadly speaking, considerable speculative stock is today available in the stock markets around the world today but generally it may vary slightly in characteristics and criteria from country to country. For example, a penny stock in the U.S.A. is one of best-known speculative stocks. According to the U.S. Securities and Exchange Commission (SEC) website, a penny stock is defined as a stock that is traded for less than $5.00 per share. Penny stocks are highrisk and low-priced and are usually issued by new or growing companies with limited cash or resources. They are called penny stocks because originally in the past their price was less than a dollar per share; the SEC later modified the penny stock definition to include all shares trading below $5.00. However, at the present time penny stocks are often traded for under $1.00 per share, although they may be priced at up to $5.00 per share. A general definition of speculative stock is stock with a high risk relative to any potential positive returns. Speculative stocks are often purchased by risk-seekers or investors who ignore detailed analysis and believe the stock will appreciate in value 2. In the U.K. at present, the term a penny share refers to stock in a company with a small market 1 Capital gains tax (CGT) is a tax on capital gains. In general, most of capital gains come from sales of common stocks, bonds, precious metals and property. Several countries impose capital gains tax but have different rates of taxation for individuals and corporations. For example, there is a 25% CGT in Austria. In Brazil, the CGTs are normally 15%, except for day-trading (Net-settlement) when it is 20%. Canadian CGT is currently up to 50% calculated by capital gains (profit) 50% marginal tax rate. The CGTs in Ireland, Mexico, Norway, Philippines, Poland and Romania are 20%, 10%, 28%, 6%, 19% and 16% respectively. However, there are some countries such as Ecuador, Iran, Jamaica, Kenya, India (for long-term capital gain from common stocks), New Zealand, Singapore, and Thailand that do not impose CGT. 2 For the Toronto stock exchange (TSX), speculative stocks are identified as low-priced speculative issues of stock, selling at less than $1.00 per share. Penny stocks in Spain are shares. In the Australian Securities Exchange (ASX) penny stocks are usually low-priced stocks that are below $1.00 and have a relatively low market capitalisation. These are often highly speculative, high-risk stocks that are only for the most advanced investors. It is easy to manipulate stocks through using the Internet. For instance, a manipulator may buy up big on a penny stock, then hype up the value on the websites or release faulty information to drive up interest in the stock. Consequently, as many investors bid and force the price higher, the manipulator begins selling their shares and making a profit. 2

3 capitalization (small-cap), or less than 100 million and/or with a stock price of less than The nature of such a company is that it has a small amount of tangible assets and a short performing history. As a result, the most favourable place for trading penny shares is the Alternative Investment Market (AIM) 3. Interestingly, on the MoneyWeek (the U.K. s best-selling financial magazine) website (2011), it was reported that more speculative stocks are traded in the mid-cap FTSE 250 and AIM. Another market where speculative stocks are traded is the OFF Exchange (OFEX), which focuses on small and medium enterprises, including firms from outside the U.K. This is not a regulated market, and firms on it are officially unlisted. Like long shots at the racetrack, speculative stocks offer investors the probability of large profits at the risk of suffering very substantial losses (Scott 2005, p. 104). Scott defined speculative stocks as the stocks of companies that are highly in debt or rely on a single product or business to generate the majority of their income. In the same vein, Sindell (2005) claims it is not difficult to identify speculative stocks because they frequently have a price/earnings (P/E) ratio twice as high as that of other stocks. Other features of speculative stocks are that they have low prices, small capitalizations and less liquidity. Such characteristics may assist stock manipulators to pump and dump these particular share prices more easily. However, in recent decades the Stock Exchange of Thailand (SET) has also been suffering from a number of speculative stocks. As a result, many investors have lost considerable sums of money by rashly dabbling in speculative stocks. As a result, in 2004 the SEC (Thailand) laid down a new regulation, namely, the Turnover List to signal to all investors the risks associated with trading in such securities. SEC also intervened in the trading of Turnover List stocks by controlling the trading volume. Turnover List stock (TOL) is common stock that has a high speculation but a poor performance. It is mostly common stock from a lowquality company. However, blue-chip stocks, for example, have never been on the TOL. The SET and SEC clearly see that the TOL as a means of forestalling speculative gains by Thai investors. This study will make use of the signalling model proposed by Allen and Faulhaber (1989), Grinblatt and Hwang (1989) and Welch (1989). We shall examine the relationship between the initial returns of Thai IPOs listed on the Stock Exchange of Thailand (SET) and the Market for Alternative Investment (mai) between 2004 and 2012 and the probability of such stocks being on the Turnover List. This investigation is particularly interesting because, firstly, the Turnover List has not yet been closely studied. Secondly, our study will also contribute to IPO 3 AIM was launched in 1995 by the LSE and it is also the London Stock Exchange s international market for smaller growing companies as well as for young and recovering companies. The objective of AIM is to offer small firms, from any country and any industry sector, the opportunity to raise capital in the market. 3

4 underpricing and signalling hypothesis studies that through analyzing exactly how underpricing can be a good indication of a stock s likely future inclusion on the Turnover List. Finally, it will be invaluable for all investors and financial institutions as well as the SET and the SEC to signal at an early stage new issues might have a tendency towards becoming speculative stocks. The organization of this paper is as follows: Section 2 presents a Turnover List overview; Section 3 introduces the hypotheses; Section 4 discusses the data; Section 5 examines the relationship between the level of IPO underpricing and the Turnover List; Section 6 comprises various concluding remarks. 2. Turnover List Overview The Securities and Exchange Commission (SEC) was established by Congress in 1934 was to eliminate stock market manipulation. However, there is considerable evidence to show that speculative stocks still exist in present-day stock markets, including the Thai stock market. According to the SEC of Thailand (2008) database, there are about 400,000 brokerage accounts or only 0.6% of individual investors in the Thai stock market out of the whole population. The majority of these investors have little knowledge about stock market investment. Furthermore, investors who are trading securities on the Thai stock market are exempt from CGT when they gain profits from their investments but they must nevertheless pay a tax once they receive dividends. As a result, most of the individual investors in the Thai stock market prefer to speculate with share prices over a short-term period rather than buy-and-hold stock for a longer period in order to receive dividends. Consequently, many individual investors have suffered great losses from trading in speculative stocks. The SEC, Thailand, defines speculative stocks as those that have a high turnover ratio compared with other equities in the same stock market. For example, a company has common stocks, totalling 100 million shares, of which only 30 million are being traded in the secondary market (Free-float = 30%). If the company s trading volume amounts to 30 million shares a day, this implies that investors are quickly trading these stocks or preferring to buy-and-sell them on the same day (Net-settlement). In order to protect investors by reducing the amount of speculative stocks, the SET and SEC issued new regulations, establishing the Turnover List. 2.1 Turnover List calculation and criteria 4

5 According to the SEC Thailand database, the Turnover List regulations were established in 2004 and applied particularly to the Thai stock market. The purpose of the Turnover List is to disclose the list of securities that have a high turnover ratio and might lead to abnormal trading and settlement risk. The Turnover List is intended to warn both investors and brokers against the risks associated with trading in such securities. We can therefore assume that the common stocks on the Turnover List, published by the SEC (Thailand), are speculative stocks. The calculations and criteria of the Turnover List are as follows: [ ] (1) Where, %1W-Turnover is turnover in a one-week period and %FF is the percentage of the free-float. The daily average market capitalisation during the week is DAMC, and DAT is calculated from the average daily trading value during the week. [ ] (2) and [ ] (3) Whereas, TOTAL is the total number of common shares in each company and NUMSS is the number of shares held by strategic shareholders. DMC and t are defined as the daily market capitalisation and the trading day during the week. NUMDAY is the number of the trading day during a period of a week, excluding any day on which a suspend sign (SP) is imposed for the whole day. In each week the Thai stocks are calculated by Equations (1) at one time, when any common stocks are in line with the Turnover List criteria shown in Table 1. It will thus be declared to be Turnover List stock (TOL) by the SEC (Thailand). Currently there is a total of 579 listed-companies on the Thai stock market. However, the stocks of 218 firms (37.65%) have been placed on the Turnover List. We can also note that blue chip stocks have never been on the Turnover List. 5

6 Table 1 The criteria of the Turnover List for common stocks in SET and mai Common stocks in SET Criteria Stocks in Turnover list - Weekly turnover ratio (1W-Turnover) 30% and - Average daily trading value in 1 week 100% million Baht - No more than the first 50 stocks Stocks which are obliged to be reported to the SEC - Stocks in the Turnover List with a P/E ratio over 100 times or a net loss or in the REHABCO sector Common stocks in mai Criteria Stocks in the Turnover List - Weekly turnover ratio (1W-Turnover) 30% and - Average daily trading value in 1 week 20% million Baht - No more than the first 5 stocks Note: Newly listed securities (IPO) with trading of less than 4 weeks are excluded from the analysis Figure 1 The proportion of Turnover List and Non-Turnover List stocks on the Thai stock market 37.65% 62.35% Non-Turnover list Turnover List 2.2 Turnover List announcement and enforcement In general, the Turnover List is announced every Friday by the SEC (Thailand). The data used work out the Turnover list, is drawn from a period of five trading days (See Figure2). Importantly, under the SEC Turnover List regulations, the investors who wish to trade on any high trading stock on the SEC Turnover List and SET criteria, have to open a Cash Balance Account instead of using a Net settlement and Margin trading 4 and they are also obliged to deposit the full amount of cash (100%) before their orders can be executed. 4 According to Fred Tomczyk, CEO of online brokerage firm TD Ameritrade ( defined margin trading as that buying on margin or borrowing money to purchase stock. He also noted that buying on margin is a game of high risk and high reward, as gains as well as losses are amplified. 6

7 Figure 2 The time line indicates the data used in the Turnover List calculation and announcement The SET specifies that securities companies (brokers) must monitor their customers who are trading TOL stocks. The TOL regulation becomes effective the day after the TOL announcement and for at least the following 3 weeks. In cases when the stock is on the TOL, brokers will not allow it to be traded by net-settlement and margin trading until the stock is released from the Turnover List. For example (See Figure 3), stocks A, B and C were announced in the TOL on March 28 th, The investors willing to buy these stocks are obliged to open Cash Balance Accounts with their brokers. They then have to deposit cash into the account before trading the TOL stocks. In general, investors must pay 100% of the TOL stock price but they have a 3-day credit after trading (T+3) for paying cash for non-tol stock (by direct debit from their bank account). The Stocks A, B and C must be traded by Cash Balance until April 18 th, 2014 (a 3-week period after the TOL announcement). However, on April 18 th, the SEC announced that stocks A and D were TOL stocks. As a result, stock D was unable to be traded by netsettlement and margin trading until May 9 th 2014 but stock A could nevertheless be traded as if it were a non-tol stock after April 25 th Figure 3 An example of Turnover List announcement and enforcement periods 7

8 Table 2 An example of a Turnover List announcement and enforcement periods Date of Turnover List announcement Turnover List stocks Trading period by Cash Balance (no Net-settlement) Friday 28 th March 2014 A,B,C A,B,C 31 th March th April 2014 (3 trading weeks) Friday 4 th April A,B,C 31 th March th April 2014 Friday 11 th April A,B,C 31 th March th April 2014 Friday 18 th April 2014 A,D A 21 th April th April 2014 (1 trading week) D 21 th April th May 2014 (3 trading weeks) Friday 25 th April D 21 th April th May Theories and Hypotheses 3.1 Signalling model The signaling model of IPOs was derived from the concept of asymmetric information indicating that there is an informational asymmetry between issuing firms and investors. Suppose there are 2 types of companies, denoted high-quality and low-quality, which look indistinguishable to investors. Companies will use underpricing to signal the firm s intrinsic high-value when they have better information such as the present value and risk of their future cash flows than investors do. Ibbotson (1975) originally stated that issuers underprice in order to leave a good taste in investors mouths. Allen and Faulhaber (1989), Grinblatt and Hwang (1989) and Welch (1989) used the following features to develop their own signalling theories. They hypothesized that underpricing allows good firms to distinguish themselves from bad ones and to improve their external financing in the future. It is clear that, if successful, high-quality firms will return to the market by issuing a seasoned equity offering (SEO) on better terms after their IPOs. The best a low-value issuer can do is to take the money and run when its stock is initially offered (Su and Fleisher 1999, p. 181). A high-quality company will find it advantageous to signal through underpricing its IPO while a low-quality company will not find it worthwhile to underprice. Ljungqvist (2004) suggested the possibility of distinguishing between good and bad firms, claiming that a company s true type is revealed to investors before the post-ipo financing stage. This exposes bad issuers to the risk that any cheating on their part will be detected before they can reap the benefit from imitating high-quality issuers signals. There are a number of empirical works have supported the signaling models of IPOs (Jegadeesh et al. 1993; Su and Fleisher 1999; Harjoto and Garen 2003; Yu and Tse 2006; Francis et al. 2010). Their results claimed that IPO underpricing is explainable as a means of bribing bureaucrats and can be explained in terms of a separating equilibrium under asymmetric information, in which underpricing is a strategy for companies to signal their value to investors. In addition to the underpricing signalling, there is a variety of proposed signalling devices such as the reputation of the 8

9 underwriter and auditor and the quality of the board directors. Such a signalling model leads to the following empirical predictions: H 1-1. Firms with lower underpriced IPOs are more likely to become speculative stocks (Turnover List) than firms with more underpriced IPOs. H 1-2. Firms with lower underpriced IPOs are likely to become speculative stocks (Turnover List) more promptly than firms with more underpriced IPOs. Generally, on being placed being on the Turnover List, stocks experience negative announcement-date returns due to trading limitation (controlled and monitored by the SET and SEC) and to being exposed as being seen as bad firms (a poor performance but a high stock price and a high trading volume). In the signalling framework, we would also expect a less negative stock reaction in response to TOL announcements by high-quality companies, which, under separation, means that such companies underprice their IPOs even more. We can thus further hypothesize: H 1-3. Firms with higher underpriced IPOs are likely to experience a less unfavourable price reaction to a Turnover List announcement than firms with less underpriced IPOs. More interestingly, the government plays an important role for individual investors making investment decisions concerning the Thai stock market. The government mostly invests in high-quality companies or holds only blue-chip stocks 5. Under Rock s assumption (1986) 6, we can also assume that the Thai government (an informed investor) has better information than uninformed investors because they can perceive which companies will generate huge profits from any proposed mega-projects in the future. Thus, we hypothesize that: H 1-4. Firms with a lower proportion of IPOs subscribed by the government are likely to become speculative stocks more promptly than firms with a larger proportion of IPOs subscribed by the government. 3.2 Market-feedback hypothesis 5 Blue-chip stocks, for example, are good fundamental stocks or the shares of a company in which there is financial strength, a record of profit growth and a good reputation. 6 The winner s curse model by Rock (1986) is probably well known as an asymmetric information model. The assumption is that informed investors such as underwriters and issuing firms are better informed about the intrinsic value of the stocks on offer than general and individual investors. Informed investors invest only in attractively priced IPOs while uninformed investors invest randomly. He suggested that shares must on average be underpriced in order to attract less informed investors to preempt the new IPO. The uninformed compete with the informed, and the issuer must ultimately compensate them for their disadvantage (Rock, 1986, p. 207). 9

10 In addition to the signalling models, we consider what we call the market-feedback hypothesis (namely, the pooling equilibrium where IPO prices are a weighted average of the perceived present values of high-value and lowvalue issuers). This hypothesis presumes that the market is better informed than the issuer and then a high return in the IPO period implies that the issuer has underestimated the marginal return on the project. Under the pooling hypothesis, the IPO price per se is not informative but the initial return of IPOs provides a measure of the extent to which the market is able to discriminate high-value companies from low-value ones (Jegadeesh et al. 1993, p. 156). Therefore, the low-quality companies could appear on the Turnover List if the market discovers their true quality or when information is revealed after-market trading starts and price differentiation occurs. We also test the alternative market-feedback hypothesis as a potential explanation of Turnover List participation and compare TOL prediction to the signalling hypotheses. To investigate whether the observed relations between IPO abnormal returns after going public and the Turnover List can be explained by market-feedback or by the pooling hypotheses, we test the following hypotheses: H 2-1. IPO underpricing is a better predictor than after-market returns (returns in the period immediately after going public) of IPOs that become speculative stocks (Turnover List). H 2-2. IPO underpricing is a better predictor than after-market returns of the time between the IPO and the first Turnover List. H 2-3. IPO underpricing is a better predictor than after-market returns of the stock-price reaction at the announcement of the first Turnover List. 4. Data and Summary Statistics 4.1 Data Construction We obtained our sample of IPOs and Turnover List (TOL) in the Thai markets from the Stock Exchange of Thailand (SET) and Securities and Exchange Commission (SEC), Thailand database. Information and the number of IPOs issued are from the official prospectus filing form (Form 69-1) available on the IPO filing database. The IPO sample totals 187 IPOs, covering the period 2004 to 2012 and the TOL within one year of the IPO date. This period was selected because the Turnover List regulations were first established in We only include the first TOL for the firms in our IPO sample. The stock prices, the SET and the mai indices were obtained from the Thomson Reuter database. 10

11 Table 3 Sample size of Thai IPOs Panel A: sample size disaggregated by exchange and by IPO offering year Year Stock Exchange of Thailand (SET) Market for Alternative Investment (mai) Total Number % Number % Number % Total Panel B: sample size disaggregated by exchange and by industry group Industry Stock Exchange of Thailand (SET) Market for Alternative Investment (mai) Total Number % Number % Number % Agro & Food Consumer Products Financial Industrial Resources Services Technology Property & Construction Total Table 3 shows details of the sample by year and exchange of listing. Approximately 60% (40%) of IPO listings are on the SET (mai). Most of the firms in our sample issued their IPOs in 2004 (26.2%) and 2005 (24.1%). Interestingly, the IPO firms for the entire sample mostly came from the service and property & construction sectors. Similarly, the majority of IPOs in the mai sample were in the property & construction industry (29.8%). For SET IPOs, 27.5% of the sample came from the service sector, 16.8% from the industrial and property & construction sectors, 11.5% from the financial sector, 10.6% from the technology sector, and the rest of the industries each had less than 10% representation. For mai IPOs, 16.2% came from the industrial sector, 13.5% from the services and technology sectors, 10.8% from the resources industry, and the rest of the industries each had less than 10% representation. 4.2 Summary statistics Table 4 presents the proportions of Turnover List and non-turnover List for the sample of IPO firms that we use for our empirical study. It can be seen that 64 IPOs (34.2%) from the entire sample were on the Turnover List (TOL) within one year following their IPOs. When we consider the proportion of TOL:Non-TOL by listed 11

12 exchange separately, it shows a higher proportion on the TOL of stocks in the SET IPOs (46.9%). However, there are 11 IPOs (14.9%) that were TOL stocks for the mai sample. This implies that Thai speculative stocks are likely to come from large-sized companies. This is in contrast with the UK stock market and other markets where speculative stocks are likely to be traded in the mid-cap FTSE 250, AIM and small-cap stock markets. Interestingly, a high density of TOL occurred between 2011 and 2012 because of a bull-market period and high trading volume. In addition to this, it can be seen in Table 4, Panel B shows that the TOL stocks were mostly from the service sector. There are no TOL stocks from the Agro & Food and financial sectors for the mai sample. Table 4 The proportion of Turnover List stock (ToL) and Non-Turnover List stock (Non-ToL) for Thai IPOs issued during the period Panel A: sample size disaggregated by exchange and by IPO offering year Year The entire sample The Stock Exchange of Thailand (SET) Market for Alternative Investment (mai) TOL % Non- % TOL % Non- % ToL % Non- % ToL TOL TOL Total Panel B: sample size disaggregated by exchange and by industry group Industry The entire sample The Stock Exchange of Thailand (SET) Market for Alternative Investment (mai) TOL % Non- % TOL % Non- % TOL % Non- % TOL TOL TOL Agro & Food Consumer Products Financial Industrial Resources Service Technology Property & Construction Total Table 5 shows a comparison of our variables, which are taken into consideration when investigating the relationship between the Turnover List and IPO underpricing. The market-adjusted initial return or the IPO underpricing (MAIR) are evident across both stock markets. MAIR i is calculated by [(P i,1 -P i,0 )/P i,0 ] - R i,m or the percentage change between the offer price and the IPO closing price on the first trading day. P i,1 is the closing price on the first day of trading, P i,0 is the IPO offering price, and R i,m is the stock market (either the SET or the mai 12

13 depending on which is the listed exchange) index return from the IPO date to the first trading date. The average of IPO underpricing for Thai IPOs going to the public between 2004 and 2012 is 22.67%. We also found that the average market-adjusted initial return in the mai IPOs (34.46%) was larger than in the SET IPOs (14.95%) and showed a statistically significant difference at 0.01 level. BHAR i represents 6-month post abnormal returns for the IPOs calculated as. The averages of 6-month BHARs in the entire, SET and mai samples are 2.31%, 2.18% and 2.50% respectively. However, there is no significance in the difference in mean of BHAR between the SET and mai samples. The issue size (SIZE) is measured by the number of shares offered at the IPO times the IPO offer price. The mean of SIZE for the Thai IPOs is about 1,152 million Baht. It worth noting that that the average issue size in the SET is significantly (1%) larger than those in the mai. For AGE i which is the age of a firm in years from the establishment date to the date of the IPO, we found that the results associated with SET and mai are similar; there is also an insignificant difference for AGE between SET- and maimarket firms. The subscription details for IPOs for foreign and institutional investors, are also reported in Table 5 as PFS and INS. However, from the average foreigner and institution participation figures, it can be clearly seen that such investors like to preempt the Thai IPOs in the SET rather than in the mai. The average foreign and institution investors subscriptions for SET IPOs are 10.9% and 21.21%, but only 4.6% and 8.47% in the mai market respectively, when the number of days between an IPO and first Turnover List (TIMETOL)is included as an explanatory variable, and the mean durations of the SET and mai IPOs taken from IPO to the first TOL being 76 days and 41 days, respectively. We also found that the difference between these two average durations is significant at 1% level. To examine the excess return around the date when the firm was announced on the TOL, we estimate the excess return, REACT, over the event days -1 through +4, where day 0 is the TOL announcement date. REACT equals to [(P i,4 -P i,-1 )/P i,4 ] [(I i,4 -I i,-1 )/I i,4 ], where day 0 is the TOL announcement date, P i, 4 is the fourth day closing price of the stock and I i, 4 is the fourth day closing price of the corresponding market index after the TOL announcement. P i,-1 and I i,-1 are the stock price and index price 1 day before the TOL announcement, respectively. We obtain 6-day (-1, 4) abnormal returns using standard event study methodology (Brown and Warner, 1985). We use the SET and mai equal-weighted index returns as the market index in the event study. The average IPO react prices for the TOL announcement in the entire and in the SET samples show positive returns of 2.72% and 4.4% respectively but -4.6% in the mai. Notably, there is a statistically significant difference in the mean of 5-day abnormal returns between the SET and the mai IPOs. 13

14 Table 5 Descriptive Statistics of the SET and the mai IPOs The descriptive statistics are for the187 Thai IPOs issued during The SET sample refers to IPOs from the Stock Exchange of Thailand. The mai sample denotes IPOs from the Market for Alternative Investment. MAIR is the market-adjusted initial return (MAIR) or the IPO underpricing calculated by [(P i,1-p i,0)/p i,0] R i,m or percentage change between offer price and IPO closing price on the first trading day. P i,1 is the closing price on the first day of trading, P i,0 is the IPO offering price, and R i,m is the stock market index return from the IPO date to the first trading date. BHAR represents 100-day post abnormal returns for the IPOs calculated as. SIZE is the number of shares offered at the IPO times the IPO offer price, and AGE is the age of a firm in years from the establishment date to the date of the IPO. GOV i which is the proportion owned by the government is considered in the model as an explanatory variable. The percentage of foreign and institution investors subscribing for the IPOs are PFS and INS respectively. TIMETOL is the number of days between an IPO and its first Turnover List (TOL). REACT is the abnormal TOL 6-day announcement price reaction (over the event days -1 through +4) calculated by [(P i,4- P i,-1)/p i,4] [(I i,4-i i,-1)/i i,4], where day 0 is the TOL announcement date, P i, 4 is the fourth day closing price of the stock and I i, 4 is the fourth day closing price of the corresponding market index after the TOL announcement. P i,-1 and I i,-1 are the stock price and the index price 1 day before the TOL announcement, respectively. The significance of the difference in the mean of variables between SET and mai sample measures is computed using the independent-sample t-test (Levene's Test for Equality of Variances and t-test for Equality of Means). Variables The entire sample The SET sample The mai sample Diff. N Mean SD. N Mean SD N Mean SD. t-statistics MAIR (%) *** BHAR (%) SIZE (million Baht) 187 1, , , , *** AGE (year) GOV (%) ** PFS (%) *** INS (%) *** TIMETOL (day) *** REACT (%) ** ***, **, * Statistically significant at the 0.01, 0.05, and 0.10 levels respectively. 5. Relationship between IPO underpricing and Turnover List 5.1 The Probability of Turnover List and IPO Underpricing We test the hypothesis that the probability of a Thai IPO becoming a Turnover List stock is related to the signalling model of IPOs and to the market-feedback hypothesis, by estimating the following logit model: ln[ ] (4) where MAIR is the market-adjusted initial return (MAIR) or the IPO underpricing. BHAR represents 6-month postabnormal returns for the IPOs. We used the abnormal returns over the period from trading day 1 to trading month 6 following the IPO date because it corresponds to the average silent period for strategic shareholders in the Thai stock market. We expected that they would sell their IPOs or reduce their share proportion immediately after the silent period, particularly in cases of bad issuers (low-value companies) going to the public because they know the true value of their companies. lnsize is the natural logarithm of offer size of IPO firms. lnage is the natural 14

15 logarithm of the age of the firm in years from the establishment date to the year of IPO. GOV is the proportion owned by the government and is considered in the model as an explanatory variable. The percentages of foreign and institution investors subscribing for the IPOs are PFS and INS respectively. Other variables include YEAR and INDUSTRY for controlling each year s conditions and industry effects. i is the regression error term. Alternatively, as noted in equation 4, this model can be written in the following non-linear-in-the-parameters form:, (5) where, P i is the probability that the i th firm becomes a TOL item and x i is the column vector of explanatory variables. Again, the three independent variables of primary interest are the IPO underpricing (MAIR), the abnormal-returns in 6-month periods after going public (BHAR) and the proportion of IPOs owned by the Thai government (GOV). Moreover, in our logit models we can examine the effect of a one-unit change in an explanatory variable that TOL = 1 by considering the derivative, which is generally called the marginal effect 7 : (6) Table 6 shows the results of the logit regression estimations. 8 We first consider the relationship between IPO underpricing and the probability of these being speculative stocks. For the entire sample, the results can be seen in Models (1) to (4). On average, there is a positive and significant relationship between the MAIR and the probability of a TOL. The marginal effects are reported in brackets below the p-value where they indicate that the effect is economically important. For the aftermarket return variable BHAR, it is positively related to the likelihood of TOL and is statistically significant. For H 1-1, the applied signalling hypothesis implies a negative and significant role for the IPO initial return in explaining the likelihood of these being speculative stocks. However, we find a positive and significant coefficient for the initial returns. A possible explanation of this finding is that for the first 7 For logit models, this interpretation would be incorrect because the form of the function is not, for example, but rather, where F represents the non-linear logistic function. To obtain the required relationship between change in and, we would need to differentiate F with respect to and it turns out that this derivative is. Then in fact, a 1-unit increase in will cause a increase in probability. In general, these impacts of incremental changes, being an explanatory variables are evaluated by setting each of them to their mean values. 8 For the sake of brevity, we do not report the estimates of the coefficients of year and industry dummy variables. 15

16 trading day of IPOs, there was no ceiling price in the past in the Thai stock market. However, in 2012 the SET adjusted the ceiling price to 300% from the offering price. This implies that if stock price-fixers would like to pump up the stock price, it is a good opportunity to do this on the first trading day. The slope coefficient (t-statistic) on the variable MAIR is (0.0008) and the slope coefficient (t-statistic) on the aftermarket return variable BHAR is (0.0783). These point estimates suggest a stronger relation between the IPO underpricing appreciation and the probability of being on the Turnover List than between the aftermarket abnormal return and the latter. This supports our hypothesis (H 2-1 ) that the IPO underpricing is a better predictor than the after-market returns of IPOs that become speculative stocks (Turnover List). In addition, the positive coefficient of BHAR suggests that the higher the abnormal aftermarket return, the more likely the listed firms are to be speculative stocks. We also included a proportion of IPOs owned by the government (GOV) in our logit models (3) and (4). For these we found a slightly different regression result. In Model (4), when we added industry and year control variables, the GOV variable was negatively related to the probability of TOL participation. This finding is in line with hypothesis H 1-4. It is true that the government does not like to invest on low-quality companies that have a higher chance of being on the Turnover List. Interestingly, Models 3 and 4 reported a positive relationship between the IPO offered size and the probability of TOL participation. This is consistent with the proportion of TOL/non-TOL stocks in Table 4, Panel A, where the Thai speculative stocks mostly come from the SET rather than the mai. To contrast this with the UK stock market, Money Week magazine (2011) reported that more speculative stocks are traded in the mid-cap FTSE 250 and AIM. However, Table 6 shows the logit estimates with unexplained lnage, PFS and INS as our explanatory variables in place of the age of firms, and the proportion of foreign and institution investors subscribing for IPOs. On separating the entire sample into SET and mai groups, we gain significant insights into the determinations of TOL participants following IPOs. Using the same vector of explanatory variables as that used in the previous logit regression for the entire sample, we obtain a coefficient (p-value) for MAIR of (0.0118) for the SET IPOs, while for the mai firms we obtain a coefficient of (0.0081) (See Models 5 and 9). The marginal effects indicate that a 1% increase in underpricing their IPOs increases the likelihood of their being speculative stocks by 73.18% and 6.19% for SET and mai IPOs respectively. We found that a 6-month abnormal return after going public also positively affects the likelihood of TOL participation for the entire sample and for the SET sample and is statistically significant. Nevertheless, for mai IPOs, there is no evidence of a relationship 16

17 between the aftermarket abnormal return (BHAR) and the probability of their being speculative stocks. Again, as is apparent from the coefficients, p-values, and marginal effects, the strong positive relationship between the likelihood of TOL participation and the level of 6-month abnormal return shown by the IPOs from the SET sample does not exist for the group of mai IPOs. In addition to this, we found from the logit estimates that lnage, GOV, PFS and INS have insignificant effects on the likelihood of being listed on TOL or being speculative stocks for the SET and mai samples. 17

18 Table 6 Logit regression estimates of the probability of TOL The logit regression estimates of the relation between the initial returns of IPOs and the probability of a Turnover List (TOL) for the SET and mai IPOs in the period. The dependent variable is a dummy variable taking the value 1 if a firm is announced in the TOL by the SEC (Thailand) within 1 year after its IPOs, and 0 otherwise. The independent variables are MAIR, which is the marketadjusted initial return or the IPO underpricing calculated by [(P i,1-p i,0)/p i,0] R i,m or the percentage change between the offer price and the IPO closing price on the first trading day. P i,1 is the closing price on the first day of trading, P i,0 is the IPO offering price, and R i,m is the stock market index return from the IPO date to the first trading date. BHAR represents 3-month post abnormal returns for the IPOs calculated as. lnsize is the natural logarithm of the number of shares offered at the IPO times the IPO offer price, and lnage is the natural logarithm of the age of a firm in years from the establishment date to the date of the IPO. GOV which is the proportion owned by the government is considered in the model as an explanatory variable. The percentage of foreign and institution investors subscribing for the IPOs are PFS and INS respectively. P-values are reported in parentheses, and marginal effects are reported within brackets. is calculated by, where is means of explanatory variables. Thus, the marginal effect is in fact calculated by ( ). Variables The entire sample The SET sample The mai sample Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10 Model 11 Model 12 Constant *** *** *** *** * * *** *** *** (0.0000) (0.0000) (0.0000) (0.0001) (0.0784) (0.4530) (0.1747) (0.0561) (0.0000) (0.0000) (0.0054) (0.1723) MAIR ** *** * ** *** ** *** ** (0.0154) (0.0008) (0.0793) (0.0118) (0.0073) (0.0167) (0.0081) (0.0103) (0.1086) [0.3269] [0.4174] [0.2482] [0.7318] [0.8368] [0.7745] [0.0619] [0.1921] [0.0348] BHAR ** * * *** ** *** (0.0337) (0.0783) (0.0543) (0.0088) (0.0104) (0.0022) (0.5951) (0.3544) (0.1332) [0.2736] [0.2717] [0.5595] [0.5843] [0.8213] [ ] [ ] [ ] lnsize *** *** ** ** (0.0000) (0.0001) (0.1479) (0.0368) (0.0107) (0.2505) [0.2240] [0.1210] [0.2056] [0.1710] [0.0197] lnage (0.2427) (0.2732) (0.1760) (0.5010) (0.3390) (0.9332) [ ] [ ] [ ] [ ] [0.0676] [ ] GOV * (0.1200) (0.0768) (0.3216) (0.1564) [ ] [ ] [ ] [ ] PFS (0.6885) (0.8721) (0.4489) (0.8454) (0.1825) (0.2308) [0.2318] [0.0891] [0.6828] [0.1807] [0.7047] [0.1259] INS (0.4833) (0.5770) (0.5325) (0.6427) (0.3006) (0.7484) [ ] [ ] [0.3423] [ ] [ ] [ ] Year dummy No No No Yes No No No Yes No No No Yes Ind. Dummy No No No Yes No No No Yes No No No No n McFadden R p-value ***, **, * Statistically significant at the 0.01, 0.05, and 0.10 levels respectively. 18

19 5.2 Time-lag between Thai IPOs and the first Turnover List We also examined the relationship between IPO underpricing and the performance of IPOs after going public and the time before a firm is on the Turnover List (TOL) the first time. We presumed that if companies do not voluntarily leave more money on the table (low-quality firms) because they know the true value of such companies, then in cases of poorer value firms the time-lag between the IPO date and the first TOL should be shorter than that for other companies. For the firms that have TOL participation within one year of their IPOs, we regress the log of time between the TOL (lntime) and the IPO on underpricing (MAIR) and the aftermarket abnormal returns (BHAR) and the control variables used in the logit regressions. We have considered the timing of TOL as exogenous in our analysis because not all IPOs in our sample have been on the TOL. We, thus, apply tobit regression analysis in studying the time elapsed between IPO and the first TOL using the following tobit model: { (7) where lntimetol is the natural logarithm of the number of days elapsed between the IPO date and the first Turnover list date (the day when the TOL announcement is reported on the SEC, the Thailand database). The vector of explanatory variables (x i ) is the same as that used in the previous logit model (6). The tobit regression estimates are shown in Table 7. For the entire sample, the slope coefficient estimate (pvalue) of MAIR is (0.0496), indicating that firms that underprice more at their IPOs tend to be on TOL more slowly than other firms with a lower degree of IPO underpricing are. This finding contrasts with our expectation in H 1-2 which indicated that companies with lower underpriced IPOs (low-quality firms) are likely to become TOL more promptly. In addition to this, we found from the tobit regression models that BHAR and lnsize are positively related to the time-lag between IPO and the first TOL and are statistically significant. This result indicates that, among the companies that are TOL, those that experience a lower aftermarket return and have a smaller offer size tend to become speculative stocks earlier than the others. When we separate the sample into companies from SET and mai markets, a positive and significant relationship between IPO underpricing and the time lag between the IPO and the first TOL is found in both markets. However, we found only in SET IPOs (Models 18-20) that the coefficients of BHAR are positive and significant ( (0.0113), (0.0098) and (0.002)). Interestingly, we found in the SET IPOs that there is a negative relationship between the age of the firm (lnage) and the time-lag (lntimetol). This implies that the SET companies with an older age appear on the Turnover List earlier than other firms do. 19

20 Table 7 Tobit regression estimates of the time between IPO and the first TOL The tobit regression analysis of the relation between the IPO returns and the time between the IPO date and the first Turnover List (TOL) date in the period from 2004 to The dependent variable is the natural logarithm of time between the IPO date and the first TOL date (lntimetol). The independent variables are MAIR, which is the market-adjusted initial return or the IPO underpricing calculated by [(P i,1-p i,0)/p i,0] R i,m or the percentage change between the offer price and the IPO closing price on the first trading day. P i,1 is the closing price on the first day of trading, P i,0 is the IPO offering price, and R i,m is the stock market index return from the IPO date to the first trading date. BHAR represents 3-month post abnormal returns for the IPOs calculated as. lnsize is the natural logarithm of the number of shares offered at the IPO times the IPO offer price, and lnage is the natural logarithm of age of a firm in years from the establishment date to the date of the IPO. GOV which is the proportion owned by the government is considered in the model as an explanatory variable. The percentage of foreign and institution investors subscribing for the IPOs are PFS and INS respectively. P-values are reported in parentheses. Variables The entire sample The SET sample The mai sample Model 13 Model 14 Model 15 Model 16 Model 17 Model 18 Model 19 Model 20 Model 21 Model 22 Model 23 Model 24 Constant *** *** *** *** *** *** (0.0009) (0.0022) (0.0000) (0.0003) (0.6180) (0.4068) (0.3334) (0.0034) (0.0055) (0.0117) (0.182) MAIR ** *** ** ** ** ** ** ** * (0.0496) (0.0018) (0.0276) (0.0403) (0.0245) (0.0608) (0.0152) (0.0165) (0.0527) BHAR ** ** ** ** *** *** (0.0394) (0.0578) (0.0276) (0.0113) (0.0098) (0.002) (0.5842) (0.3687) (0.1446) lnsize *** *** ** (0.0000) (0.0001) (0.3113) (0.2525) (0.0184) (0.2618) lnage * (0.1164) (0.2041) (0.099) (0.2961) (0.9556) (0.5178) GOV * (0.0898) (0.0694) (0.3221) (0.2732) PFS (0.6047) (0.9228) (0.4272) (0.8151) (0.2009) (0.1682) INS (0.7594) (0.8665) (0.7926) (0.7094) (0.2899) ( Year dummy No No No Yes No No No Yes No No No Yes Ind. Dummy No No No Yes No No No Yes No No No No n Left censored obs Uncensored obs ***, **, * Statistically significant at the 0.01, 0.05, and 0.10 levels respectively. 20