ACADEMIA ECONOMIC PAPERS 32 : 1 (March 2004), 53 81 Underwriter Price Stabilization of Seasoned Equity Offerings: The Evidence from Transactions Data James F. Cotter Wake Forest University Wayne Calloway School of Business and Accountancy North Carolina, USA Anlin Chen Department of Business Management National Sun Yat-Sen University Lanfeng Kao Department of Finance National University of Kaohsiung Keywords: Seasoned equity offerings, Underwriting, Price stabilization, Transaction data JEL classification: G24, G32 Correspondence: Anlin Chen, Department of Business Management, National Sun Yat-Sen University, Kaohsiung 804, Taiwan. Tel: (07) 525-2000 ext. 4656; Fax: (07) 525-4698; E-mail: anlin @mail.nsysu.edu.tw. We thank Doug Foster, Melanie McNeil, J. Sa-Aadu and two anonymous referees for their thoughtful comments and encouragement.
10:52 AM page:54 ABSTRACT Underwriters of initial public offerings (IPO) and seasoned equity offerings (SEO) often engage in price stabilization to prevent significant declines in the market after a new security is issued. This paper examines the aftermarket price dynamics of companies offering SEOs to understand the motivations for stabilization and the characteristics of firms that have their securities stabilized. We find that the likelihood of price stabilization is negatively related to the magnitude of the offer price, the amount of trading and the volatility of transaction returns, and positively related to the holding period return between thetimewhentheofferisregisteredandwhentheofferispriced.
10:52 AM page:55 Underwriter Price Stabilization of Seasoned Equity Offerings (Cotter, Chen, and Kao) 1. INTRODUCTION In the ordinary course of market making, Section 9(a)(6) of the Securities and Exchange Act of 1934 (1934 Act) prohibits any transactions that are made for the purpose of pegging, fixing, stabilizing, or otherwise manipulating the price of a security. When a security is being underwritten, however, Rules 10b-6, 10b-7 and 10b-8 of the 1934 Act form the legal basis for an underwriter placing a bid... for the purpose of pegging, fixing or stabilizing the price of any security when such stabilization is employed to facilitate the offering of any security. 1 When securities are being sold, price stabilization is often employed to prevent a significant decline in price of the offered securities in the days following issuance. Recent academic work has focused exclusively on price stabilization for initial public offerings (IPOs). 2 However, no one has ever examined the existence of price stabilization in seasoned equity offerings (SEOs). Is it possible that there exists price stabilization in SEOs? Since the SEC does not rule out SEO stabilization, the possibility of SEO stabilization exists. Benveniste and Spindt (1989) and Ibbotson and Ritter (1995) argue that underwriters have to purchase the unsold shares of a firmcommitment offering and in turn accept the financial risks of reselling the shares in the open market after issuance. That is, underwriters bear the proceeds risk for their firm-commitment offerings. The reason why the market might not be able to complete the whole distribution of an offering is that the offer price is too high or the demand for the offering is lower than the supply. Obviously, there exists price history prior to SEO issuance. The offer price of an SEO is typically set at the market price right before issuance. Anecdotal evidence shows that the issuer or the underwriter has an incentive to buy shares days before issuance to raise the market price and thus to raise the offer price of the SEO. The stock manipulation of SEOs before issuance leads to the possibility of too high an offer price and to the need for SEO stabilization. Moreover, the dilution effect of SEOs is well documented. The dilution effect of SEOs results from the increase of shares outstanding after issuance. Of course, SEOs cause the number of shares outstanding of an offering to increase by the number of shares offered. The 1 Federal Securities Laws Selected Statutes, Rules and Forms, 1995 Edition, by Richard W. Jennings, Harold Marsh, Jr., John C. Coffee, Jr., and Joel Seligman, p.623, Foundation Press. 2 For example see Aggarwal (2000), Benveniste et al. (1996), Chowdhry and Nanda (1996), Ellis et al. (2000), Hanley et al. (1993), Ruud (1993), Schultz and Zaman (1994), and Cotter and Chen (2002). 55
10:52 AM page:56 Academia Economic Papers 32:1 (2004 ) increase of shares supply after issuance would result in the supply of SEO shares being larger than the market demand, leading to downward price pressure. Therefore, we argue that SEOs may also be stabilized to facilitate their distribution to protect their proceeds risk. 3 Furthermore, the availability of transaction level data has helped in our understanding of the relations between market-making activities and investors. In this paper, we examine price stabilization of SEOs using a methodology developed by Cotter and Chen (2002) that focuses on the trading patterns of an IPO in the days after issuance. We find that the magnitude of the offer price, the number of trades over a three-day period and the standard deviation of the transaction return are negatively related to the likelihood of SEO stabilization. This suggests that the larger the offer price, the more trades that occur, and that greater volatility of transaction returns leads to a lower likelihood of aftermarket price stabilization. We also find that prestigious underwriters are more likely to stabilize SEOs when needed. Interestingly, we fail to find a relation between likelihood of SEO price stabilization and the initial return after issuance. Initial return has been shown to be an important factor relating to the price stabilization of initial public offerings (see Aggarwal (2000), Ellis et al. (2000), Schultz and Zaman (1994), Cotter and Chen (2002) and others). The remainder of this paper is organized as follows. We briefly discuss previous empirical research relating to the issuance of seasoned equity in section 2. In section 3 we discuss relevant theoretical and empirical research relating to IPO stabilization. In section 4, we outline the control variables used in the empirical model specification. Section 5 reviews the data and descriptive statistics. In section 6, we examine the multivariate testing methodology and discuss our major results. Section 7 concludes the paper. 2. RESEARCH ON SEASONED EQUITY ISSUANCE Much of the work on SEOs focuses on the effect of an SEO announcement on firm value and on the pricing of the offer relative to transaction prices just prior to the final offering. Asquith and Mullins (1986) examine the effect on the market value of eq- 3 The US SEC regulates the price stabilization activities in securities offerings under the Securities Exchange Act of 1934. The US SEC does not rule out the price stabilization in SEOs. In fact, Taiwan securities trading rules allow the underwriters to stabilize SEO shares but not IPO shares. From the evidence of securities trading, we hypothesize that price stabilization exists in SEO markets. 56
10:52 AM page:57 Underwriter Price Stabilization of Seasoned Equity Offerings (Cotter, Chen, and Kao) uity due to the SEO announcement and show that the announcement leads to a decline in the value of the firm equaling 31% of the funds raised. This result is consistent with the general lemons argument offered by Ackerlof (1970) and with that focusing on SEOs proposed by Myers and Majluf (1984), who suggest that asymmetrically informed managers will offer securities when they believe the firm is over-valued and that investors will react to the announced offering with a decline in price. Mikkelson and Partch (1986) further examine the stock price reactions to announcements of seasoned equity offerings by looking at factors such as the amount of new capital received, credit quality of the issuer, and the use of the funds. Consistent with the prediction of Myers and Majluf (1984), Mikkelson and Partch (1986) conclude that more negative abnormal returns after SEOannouncement are related to larger issues and issues where the offer is being used to refinance debt. In addition to abnormal returns SEO announcement, Loderer et al. (1991) examine the pricing of seasoned equity offerings relative to the price in existence just prior to issuance. They find a small negative holding period return from the day before the offer to the offering day. Loderer, Sheehan and Kadlec further show that returns over a smaller period, from the pre-pricing offer to the close of trading on the day of offering, are not different from zero. These results suggest that declines in firm value occur upon the announcement of an SEO, but that the offer price of an SEO is nearly equal to the pre-offer stock price. 3. RESEARCH RELATING TO IPO PRICE STABILIZATION The effects of underwriter stabilization on the offer price dynamics and subsequent trading in the secondary market have largely been unexplored by theoretical models of initial public offerings. In one exception, Chowdhry and Nanda (1996) present a model that includes the impact of price stabilization and examines its effect on the IPO process. Chowdhry and Nanda argue that syndicates are formed to increase this loss capacity for an IPO under the existence of uninformed investors and further conjecture that the lead underwriter will be compelled to intervene in the aftermarket to defend its reputation, while additional underwriters are added to the IPO process to enhance the loss capacity and share the financial risks for the banks. 4 4 In SEO markets, there still exist uninformed investors. Therefore, Chowdhry and Nanda s model can be applied to support the necessity of SEO stabilization to facilitate the distribution of SEO shares. 57
10:52 AM page:58 Academia Economic Papers 32:1 (2004 ) Another theoretical examination of the impact of price stabilization in IPOs is Benveniste et al. (1996). Underwriters of IPOs have a continuing relationship with their investing public, yet have a single or limited number of contacts with the issuing firm. This continuing relationship with the investors leads to an important role for contracting mechanisms, such as the credible commitment to stabilize, and to set the terms and conditions of the offer to provide investors with adequate returns and issuers with a fair offer price. In their model, Benveniste et al. conclude that stabilization can be thought of as a bonding mechanism between the investors and the underwriter. Stabilizing bids may have the effect of rewarding investors who participate in an IPO that appears overpriced. Several recent studies provide indirect evidence of underwriter price stabilization for short periods after IPO issuance. Ruud (1993) reassesses the underpricing evidence by examining not only the mean of the initial return but the whole distribution of the initial return for a sample of IPOs. Ruud shows that the distribution of initial returns is positively skewed and peaked around zero. She conjectures that underwriter price stabilization may be responsible for this phenomenon, suggesting that some IPOs that should decline at issuance do not because stabilizing bids are preventing it. Ruud shows that after stabilization activities cease, stabilized offers decline in price within a few weeks of the offering. Finally, Ruud (1993) also shows that the holding period return of likely stabilized offers migrates to negative tail in the weeks after issuance, suggesting that a large proportion of the stabilized offers decline in price within a period of four weeks after the offer. Several studies examine the determinationofthebid-askspreadandstabilization activities for IPOs. Smith (1986) suggests that IPO price stabilization may be important in the relation between the underwriter and the investor because of the underwriter s possible offer to repurchase overpriced offers. In this context, investors may view offers with a credible promise of stabilization as less risky, since declines in price, to some extent, will be prevented by repurchases by the underwriter at or near the offer price. Hanley et al. (1993) use this idea to estimate the value of a protective put option and examine the relation between the value of the put and the bid-ask spread on a daily basis after issuance. Hanley, Kumar and Seguin show that price stabilization activity reduces the bid-ask spread for IPOs after issuance and that this effect dissipates approximately ten days after issuance, possibly due to the cessation of stabilization activity. Hedge and Miller (1989) find that bid-ask spreads of IPOs are three-fourths as large as those of seasoned securities. However, their finding may not 58
10:52 AM page:59 Underwriter Price Stabilization of Seasoned Equity Offerings (Cotter, Chen, and Kao) be due to price stabilization because the narrower spreads last for eight weeks, much longer than would be expected in typical stabilized IPO. Schultz and Zaman (1994) examine the process by which bids and asks are set, and specifically the inside bid and ask, for a sample of 72 IPOs. Since only the lead underwriters may engage in price stabilization, they will offer stabilizing bids higher than bids offered by other market makers when stabilization occurs. Schultz and Zaman (1994) find that underwriters are very likely to set the inside bid for cold offers, defined by Schultz and Zaman as IPOs that trade in the secondary market at a price below the offer price. By contrast, non-underwriters are four times more likely to set the inside ask than the inside bid for cold offers. This evidence offers further support regarding the impact of underwriter price stabilization on the bid and ask. Ellis et al. (2000) document the fact that lead underwriter becomes the main market maker for the stock issued and keeps a large inventory for the IPO. Keeping inventory for IPO shares is related to the stabilization activities. Ellis et al. (2000) find that the risk of keeping a high position of inventory of IPO shares is reduced by the use of over-allotment options. They also indicate that the time dimension for IPO stabilization appears to be 21 days. Aggarwal (2000) also supports the occurrence of underwriter price support for IPOs. Aggarwal argues that underwriters managepricestabilization activitiesby using a combination of aftermarket short covering, penalty bids and over-allotment options. Basically, the price behavior of IPOs differs from that of SEOs. Nevertheless, we are curious as to whether the price behavior of IPOs in the early aftermarket is different from that of SEOs in the early stage after issuance. 4. PRICE STABILIZATION AND THE CONTROL VARIABLES To disclose the possibility of price stabilization, issuers in combination with their lead underwriter file registration documents that include the following language to inform investors regarding the practice of stabilization: In connection with this offering, the underwriters may over-allot or effect transactions which stabilize or maintain the market price of the common stock offered hereby at a level above that which might otherwise prevail in the open market. Such transactions may be effected on the new york stock 59
10:52 AM page:60 Academia Economic Papers 32:1 (2004 ) exchange, in the over-the-counter market or otherwise. Such stabilizing, if commenced, may be discontinued at any time. Beyond this, no formal disclosures are made when stabilization actually occurs in the market place. This lack of disclosure has led researchers to look for numerous proxies for the existence of stabilization. A number of proxies have been suggested to detect the existence of stabilization. Ruud (1993) conjectures that issues with a zero one-day return at issuance are likely stabilized issues. Hanley et al. (1993) present two proxies to detect the presence of price stabilization. First, Hanley et al. form a proxy that equals the ln(the first closing bid price / Floor offer price) that attempts to capture the proximity of the market price to the floor price. Stabilized offers would have values close to zero for this measure. Hanley et al. also argue that stabilization provides investors with a put option and develop a measure of stabilization equal to the value of the put option, using a modified form of the Black-Scholes option pricing formula. Using transaction level data, Schultz and Zaman (1994) use inside bids (the highest bids offered by the market makers) to detect the evidence of stabilization. Schultz and Zaman apply the frequency of inside bids offered by the underwriters at the first 3 trading days to measure the occurrence of stabilization. Cotter and Chen (2002) further develop two new techniques to detect the presence of stabilization based on the dynamics of trading activity in the days after issuance. For their first proxy of IPO price stabilization, they argue that if an underwriter chooses to stabilize an offer at a single price, the stabilizing bids will lead to an inordinate number of transactions at that price. Using this insight, they argue that when the proportion of trade occurring at the mode transaction price (MTP) is large and when the MTP is at or just below the offer price, stabilization activities are likely. 5 A second proxy for IPO price stabilization focuses on the time series properties of successive trades as a measure of stabilization. For this measure, they incorporate three factors to detect IPO stabilization: (1) stabilizing transactions will occur at the prevailing bid price; (2) by law, stabilizing transactions willoccuratorbelowtheofferprice; and(3) stabilizing transactions will likely not lead to changes in subsequent transaction prices. Cotter and Chen (2002) combine these three measures into a single multi-factor measure of IPO price stabilization. 5 MTP means the most frequent transaction price. For this proxy, Cotter and Chen (2002) make the assumption that IPO price stabilization occurs at a single price. It is possible that an underwriter will attempt to stabilize an IPO at successively lower prices and gradually let the price decline with limited intervention or price stabilization. Using their first measure, it is possible to detect all but the single line of defense price stabilization activities of the underwriter. 60
10:52 AM page:61 Underwriter Price Stabilization of Seasoned Equity Offerings (Cotter, Chen, and Kao) As we can see, Ruud (1993) and Hanley et al. (1993) employ the first closing price to detect the existence of stabilization. Only one price (the first closing price) is used by Ruud and Hanley, Kumar and Seguin to detect stabilization. Schultz and Zaman (1994) use all the inside bids in the first 3 trading days to detect stabilization. Cotter and Chen (2002) develop a multi-factor consisting of transaction price, bids offered by the market makers, and the volatility of transaction returns at the first 5 trading days to detect stabilization. Obviously, the Schultz and Zaman measure or Cotter and Chen measure is better than the measures used by Ruud (1993) and Hanley et al. (1993), since Schultzand Zaman (1994) andcotter and Chen (2002) take the transaction level data into account to detect stabilization. Suppose the first closing price of an offer happens to be the offer price. The Ruud measure or Hanley, Kumar and Seguin measure would treat that offer as a very likely candidate for a stabilized offer. Furthermore, the Cotter and Chen measure should be even better than the Schultz and Zaman measure because the Cotter and Chen measure takes both transaction price and quotes into account while the Schultz and Zaman measure only takes quotes into account. Therefore, this paper applies the Cotter and Chen measure to detect the existence of SEO stabilization. As with IPOs, several factors may affect the likelihood of SEO price stabilization after issuance. The initial return, here defined as the return from the offer price to the first trade, can affect the lead underwriter s willingness and ability to undertake stabilization activities. SEOs with large positive or large negative initial returns are unlikely candidates for stabilization, whereas offers with initial returns near zero are more likely to be stabilized. While SEOs are less likely than IPOs to have a large positive initial return, SEOs with a large positive initial return cannotbe stabilized due to SEC restrictions when the trading price isabovetheofferprice.whenafirmhasa large positive initial return, stabilization could occur only if the positive initial return is then followed by a price decline to a level at or below the offer price, thus leading to possible stabilization activities by the underwriter. On the other hand, SEOs with a large negative initial return are not likely to be stabilized due to the costs associated with stabilization. Hence, we expect that SEOs with initial returns close to zero are most likely to be stabilized. Second, trading intensity may affect the lead underwriter s ability to stabilize SEOs. The larger the number of transactions occurring in the first few days, the less likely that an underwriter will actively attempt to stabilize an offer due to limitations in the financial resources necessary to stabilize. Capital availability will limit the un- 61
10:52 AM page:62 Academia Economic Papers 32:1 (2004 ) derwriter s willingness to place stabilizing bids leading to purchases of stock recently issued. Likewise, the return of the offered stock exposes underwriters to the risk of additional losses from the stabilizing purchases if subsequent liquidation sales occur below the offer price. Price stabilization may also be affected by the magnitude of the offer price, primarily due to the greater number of possible trading prices available for higher priced stocks than lower priced stocks. This problem arises principally due to share price discreteness, allowing low offer prices to trade more frequently at each discrete price level (given formal and informal minimum tick sizes rules). To control for this factor we include the offer price as a variable in each multivariate specification. Finally, the volatility of trading after issuance might affect the extent to which a firm s stock trades at a single price. Increased volatility of the stock price in the aftermarket will likely decrease the extent to which the stock will trade at a single price. Finally, Carter and Manaster (1990) examine the relation between underwriter quality, the initial return, and the variance of trading in the after market. Carter and Manaster find that the initial return for more prestigious underwriters is lower than for less prestigious underwriters, suggesting higher quality underwriters are better able to underwrite offerings and get maximum proceeds for the issuing firm. This may also be due to more prestigious underwriters attracting higher quality companies to offer to the market. In this paper, we apply this hypothesis to SEOs and examine the potential importance of underwriter quality in the stabilization of SEOs, and its effect on the bid-ask spread in trading after issuance. 5. DATA AND DESCRIPTIVE STATISTICS Data for this study consist of a sample of 879 SEOs issued in 1997 and 1998, as disclosed on Bloomberg Financial Markets. When a firm files its registration statement for an SEO with the SEC, it must disclose a number of factors to the investing public, as well as provide a detailed description of the business and recent financial performance of the firm. Data from Bloomberg include the preliminary prospectus price (typically thepriceonthedaythedocumentisfiledwith the SEC), the number of securities that are expected to be issued, the final offer price, the lead underwriter, and the comanaging underwriters of the SEO. SEOs with any missing data are deleted. As opposed to IPOs, SEOs have a financial history and a transaction history for 62
10:52 AM page:63 Underwriter Price Stabilization of Seasoned Equity Offerings (Cotter, Chen, and Kao) their common stock. The transaction prices and bid-askquotes for each SEO inits first thirty trading days are found using the Trade and Quote (TAQ) database established by the New York Stock Exchange, Inc. The TAQ database collects all trades and the inside quotes for the New York Stock Exchange (NYSE), American Stock Exchange (AMEX), and National Association of Securities Dealers Automated Quotation (NAS- DAQ) National Market System (NMS) securities. For our study, approximately 3.6 million individual transactions and 3.4 million quote revisions were obtained from the TAQ database. We also use the TAQ database to collect the first transaction price on the first day of trading after the SEO is priced to calculate the initial return, which is used as a control variable in our multivariate tests. Using the price data for each SEO on the first trading day, we calculate the initial return, R 1,asfollows:R 1 =(p 1 p 0 )/p 0 where p 0 is the offer price and p 1 is the first traded price after issuance. Prior research documents an important role for the general level of quality and prestige of the lead underwriter, especially when the value of the offered securities is uncertain. In this paper, we use a proxy to measure the quality of the lead underwriter that is equal to the total amount of issuer proceeds over the sample period. We conjecture that the larger the sum of the proceeds, the higher the quality of the underwriters. Therefore, we measure underwriter quality by calculating the sum of the underwriters SEO proceeds of their offerings in 1997 and 1998. We form a measure that is equal to: Underwriter Quality = n j S i=1 Proceeds i where Proceeds i is the offer price times the number of shares offered for SEO i,and n j is the number of SEOs in the sample underwritten by underwriter j. Using total proceeds as a proxy for underwriter quality has important advantages over using the number of SEOs underwritten. Our measure attempts to capture the ability and willingness of the underwriter to stabilize an offer if necessary. Several investment banks underwrite many SEOs in a single year, yet their total proceeds are significantly lower than those of better known investment banks underwriting a few large SEOs each year. Ideally, we would like to use the total capital available to each firm as our proxy for underwriter quality. Unfortunately, these data are not available for many investment banks in our sample. In this paper, we follow Cotter and Chen (2002) and form a measure of the frequency of trading that occurs at the mode transaction price. They argue that the per- 63
10:52 AM page:64 Academia Economic Papers 32:1 (2004 ) centage of trades that occur at the MTP in the first few trading days after the offering date, combined with the relationship of the MTP to the offer price, will provide an effective proxy for aftermarket price stabilization. Following Cotter and Chen (2002), we define a variable called percent as the proportion of trades at MTP to the total number of trades in the first three trading days. We select three days because the price stabilization activities are shorter lived for SEOs than for IPOs. The number of days used is somewhat arbitrary; however, sensitivities to the number of days does not lead to qualitatively different results. Anecdotal evidence suggests that stabilization is more likely for a longer period of time for IPOs than SEOs. For logistic regressions, we form an indicator variable equal to one for an SEO meeting two criteria: (1) the percent of transactions at the MTP is higher than 50%, 6 and (2) the ratio of MTP to the final offer price greater than or equal to 0.95 and less than or equal to 1.0. The indicator variable is zero otherwise. 7 First, this measure uses transaction data to proxy for an unobservable activity by underwriters in the after-market. Additionally, this measure of price stabilization incorporates all trading activity over a pre-specified period after issuance, allowing for pricestabilizationattheofferpriceaswellaspricesbelowtheofferprice.anecdotal evidence suggests that while stabilizationis quite common, it does not always occur at the offer price. In fact, the underwriter has the incentive to stabilize at a price lower than the offer price, earning more profits from the offer if the underwriter sells at the offer price and repurchases in the market at a price less than the offer price. Table 1 presents descriptive statistics on the characteristics of the 789 firm-commitment SEOs in our sample. In addition to examining the entire sample, we divide it into sub-samples: (1) SEOs with negative initial returns, (2) SEOs with non-negative initial returns. In our sample, 164 (32.86%) and 335 (67.14%) SEOs have initial returns less than 0.0% and higher than or equal to 0.0%, respectively. The mean (median) offer price and offer proceeds for the SEOs in the full sample is equal to $28.95 ($19.75) and $79.68 million ($46.53 million), respectively. In contrast to prior work on IPOs and SEOs, the initial return for our sample is a positive 2.79% (median 0.0%) on the first trade after securities are offered. As 6 While our choice of 50% is somewhat arbitrary, Cotter and Chen (2002) suggest that this assumption does not significantly alter the results when the variable percent is varied using values ranging from 25% to 75%. We test this sensitivity and find similar results. 7 Our choice of 0.95 is consistent with other cut-offs in this paper for IPOs that have initial returns equal to 5.0%. Again, we test alternate choices of this cut-off, ranging from.90 to 1.00 to.99 to 1.00 with similar empirical results. 64
Table 1 Descriptive Statistics of SEO Characteristics Mean and median descriptive statistics on the offer price, proceeds, transaction return volatility, initial return, and underwriter prestige for the full sample of SEOs with initial return less than or equal to 0.0%, and higher than 0.0% and for stabilized offers and non-stabilized offers. The data for the SEOs are collected from Bloomberg in 1997 and 1998. Initial return Presence of stabilization Full Reduced Less than or Less than or Non-stabilized Stabilized sample sample equal to 0.0% equal to 0.0% offers offers a Variable Mean Mean Mean Mean Mean Mean [Median] [Median] [Median] [Median] [Median] [Median] Sample size 789 499 164 335 434 65 Offer price b $28.95 $22.38 $22.74 $22.20 $22.95 $18.54 [$19.75] [$20.75] [$21.75] [$20.25] [$21.63] [$17.63] Proceeds (in million) $79.68 $79.62 $95.52 $71.83 $82.34 $61.46 [$46.53] [$50.80] [$59.28] [$46.90] [$51.15] [$45.00] Standard deviation of transaction return 2.02% 1.90% 2.26% 1.73% 2.02% 1.11% from pricing day to 2 days later c [0.00%] [1.31%] [1.12%] [1.40%] [1.40%] [0.86%] 2.79% 2.69% 5.79% 6.85% 2.41% 4.60% Initial return d [0.00%] [1.52%] [ 1.13%] [3.33%] [1.48%] [1.72%] Holding period return prior to issuance e 2.24% 2.02% 0.64% 2.70% 1.54% 5.22% [0.00%] [0.00%] [0.00%] [0.00%] [0.00%] [ 5.50%] Underwriter prestige f $3,306.67 $3,385.55 $4,115.92 $3,028.00 $3,344.02 $3,662.88 [$1,788.17] [$1,993.79] [$1,993.79] [$1,788.17] [$1,993.79] [$1,788.17] a SEOs with percent of MTP higher than 50% and ratio of MTP to the offer price between 0.95 and 1 are defined as stabilized offers. b Offer price is the price outlined by the offering prospectus and reported by Bloomberg. c The standard deviation of the transaction return is the standard deviation for returns calculated using the ratio of successive transaction prices minus one for each SEO. d Initial return is defined as the first trading price on the offer day divided by the offer price and then minus one. e The holding period return of the offering from 2 days before the filing date to 2 days before the offer date. f The total proceeds of each underwriter is used as the proxy for the prestige of the underwriter. The total proceeds variable is calculated based on the proceeds of individual offerings in the sample. 65
10:52 AM page:66 Academia Economic Papers 32:1 (2004 ) expected, the initial return is 5.79% and 6.55% for SEOs whose initial returns are less than0.0%andhigher thanor equal to0.0%, respectively. Interestingly, the holding period return from the filing date to the offer date is 2.24% (median 0.0%) for the full sample while the return is equal to 0.64% (median 0.0%) and 2.70% (median 0.0%) for SEOs whose initial returns are lower than 0.0% and higher than or equal to 0.0%, respectively. Following the definition of SEO price stabilization outlined in section 3 of this paper, Table 1 also presents univariate results for the sample, broken down by firms that are categorized as stabilized and non-stabilized SEOs. Interestingly, SEOs that are classified as stabilized tend to have smaller proceeds and a higher initial return around the issuance of the SEO. Further, stabilized offers tend to have a more negative return from the filing date to the offer date. This suggests that offers whose stock price declines from the filing date to the offer date are more likely to receive underwriter support in the aftermarket. Finally, SEOs classified as stabilized tend to be associated with more prestigious underwriters than non-stabilized offers. Table 2 presents data on the stock price performance for SEOs in the sample. Panel A and panel B suggest that the stock price of firms that issue equity in an SEO increases rather dramatically in the period prior to filing the registration statement. For the period from two years prior to the filing date to the filing date, the SEOs in this sample increased on average 188.20%, while they increased 67.39% over the one year period prior to issuance. These results are consistent with Loughran (1997) who documents stock price run-ups prior to the announcement of an SEO. Panel D, however, suggests that the stock price declines from the filing to the offer date. For the full sample, there is a 2.65% (median 5.04%) holding period return from five days before the filing date to five days after the offer date. Consistent with Loderer, Sheehan and Kadlec (1991), panel E outlines a small negative holding period return of 2.12% (median 2.77%) for the period from two days prior to the offer date to two days after the offer date. Overall, the results in Tables 1 and 2 suggest that there is a significant increase in the stock price prior to the announcement of an SEO (the filing date) and a negative announcement effect on the stock price. 66
10:52 AM page:67 Underwriter Price Stabilization of Seasoned Equity Offerings (Cotter, Chen, and Kao) Table 2 Stock Price Performance of Issuing Firms The stock price performance of issuing firms prior to the filing date, around filing date, and around pricing date. The performance is measured by the holding period return. Data sample consists of the issuers issuing secondary equity offerings in 1997 and 1998. Dates about issuance are collected from Bloomberg. Performance data are downloaded from CRSP. Variable N Mean STD-DEV Minimum Median Maximum Panel A 500 days before filing to 5 days before filing a Holding period return 494 188.20% 288.52% 59.88% 119.97% 2932.65% CAR 253 48.10% 130.94% 574.97% 43.17% 531.00% Beta 253 0.73867 0.633191 1.5641 0.6179 2.205 T -statistic 253 0.660922 2.145368 6.2356 0.6328 22.1695 Panel B 250 days before filing to 5 days before filing b Holding period return 495 67.39% 95.74% 46.59% 42.93% 1126.59% CAR 335 19.31% 68.73% 174.25% 16.17% 260.96% Beta 335 0.781258 0.699249 1.7007 0.7238 3.0396 T -statistic 335 0.324989 1.342571 3.9645 0.3637 3.9365 Panel C 2 days before filing to 2 days after filing c Holding period return 492 1.98% 5.42% 29.03% 1.19% 25.37% CAR 335 2.85% 6.16% 34.61% 2.33% 19.81% Beta 335 0.781258 0.699249 1.7007 0.7238 3.0396 T -statistic 335 0.37782 0.924033 3.1814 0.3927 4.8129 Panel D 5 days before filing to 5 days after pricing c Holding period return 468 2.65% 15.48% 40.00% 5.04% 89.33% CAR 335 3.59% 20.08% 96.65% 3.64% 147.06% Beta 335 0.781258 0.699249 1.7007 0.7238 3.0396 T -statistic 335 0.19835 0.985577 3.1609 0.2466 5.3152 Panel E 2 days before pricing to 2 days after pricing c Holding period return 492 2.12% 7.76% 25.00% 2.77% 34.29% CAR 335 1.45% 8.00% 30.30% 1.39% 30.25% Beta 335 0.781258 0.699249 1.7007 0.7238 3.0396 T -statistic 335 0.25664 1.171545 5.488 0.2188 2.9814 a Beta is estimated based on the market model for the period from 750 days before the filing date to 501 days before the filing date. b Beta is estimated based on the market model for the period from 500 days before the filing date to 251 days before the filing date. c Beta is estimated based on the market model for the period from 259 days before the filing date to 60 days before the filing date. 67
10:52 AM page:68 Academia Economic Papers 32:1 (2004 ) Table 3 presents univariate results for the trading behaviorof SEOs around four significant time periods: before and after the filing date, and before and after the offer date. For the full sample, panel A shows that the percent of trades that occur at the mode transaction price is equal to 31.49%, with SEOs that have been classified as stabilized having a slightly higher percent MTP than SEOs not classified as stabilized. Panel A of Table 3 also shows that the bid-ask spread is equal to $0.4850, or 2.01% of the most frequently traded price prior to the announcement of the SEO. Interestingly, the patterns of stabilization are very similar in the post-filing period (panel B), with the percent of trading at the MTP equal to 27.44% for the full sample and 34.04% and 26.46% for firms that are classified as stabilized and not, respectively. Likewise, the bid-ask spread is similar with a spread of $0.4852 for the full sample and $0.4862 and $0.4850 for firms that are classified as stabilized and not, respectively. The trading patterns for firms that issue SEOs differ dramatically for firms that are classified as stabilized and not for the three day period after the SEO is issued. The percent of trades at the MTP is 34.95% for the full sample yet is 65.76% for stabilized offers and 30.34% for non-stabilized offers. Consistent with evidence from Schultz and Zaman (1994), who show that stabilization leads to reduced bid-ask spreads, the bid-ask spread for the full sample declines to approximately $0.3890, while the bid-ask spread is $0.3548 for stabilized offers and $0.3943 for non-stabilized offers. 6. MULTIVARIATE RESULTS FOR SEO STABILIZATION The results presented thus far show numerous differences between the characteristics of and trading patterns of offers that are suspected to be stabilized offers. In this section, we report results regarding the differences between stabilized and non-stabilized offers, incorporating the possible correlationsamong the independent and controlvariables in the model. In the logistic regression, we define an SEO as stabilized if the percent of the MTP in the first three trading days is larger than 50% and the ratio of the MTP to the offer price is between 0.95 and 1. In Table 4, we present evidence using a logistic regression in the following form. Therefore, the logistic model specification is as follows: P (Stabilized offer) =f (Offer price, Number of trades, Absolute value of the initial return, Transaction return volatility, Pre-offer holding period return, Underwriter prestige) 68
10:52 AM page:69 Underwriter Price Stabilization of Seasoned Equity Offerings (Cotter, Chen, and Kao) Table 3 Descriptive Statistics of SEO Trading Pattern Mean and median descriptive statistics on MTP, percent of MTP, and bid-ask spread for full sample and for offers designated as stabilized and non-stabilized. An offer is a stabilized offer if the percentage of trades occurring at the mode transaction price (MTP) is greater than 50% of the total trades during the period of offer day to 2 days after offer day and the MTP is greater than 95% of the offer price and less than the offer price. The data for the SEOs are collected from Bloomberg in 1997 and 1998. The trading price, bid price, ask price, and number of trades are from TAQ. Full sample Non-stabilized offer Stabilized offer Mean Mean Mean Variable Median Median Median [N] [N] [N] Panel A: Before filing period (6 days before to 4 days before the filing day) $22.74 $22.95 $18.54 Offer price $21.75 $21.63 $17.63 [499] [434] [65] $24.04 $24.60 $20.26 MTP $22.75 $22.94 $19.19 [440] [384] [56] 31.49% 30.67% 37.15% Percent of MTP 29.40% 28.57% 35.64% [440] [384] [56] 329.04 356.34 141.84 Number of trades 144.50 156.00 86.00 [440] [384] [56] 48.50 cents 48.38 cents 49.72 cents Ask-bid 45.46 45.75 42.70 [433] [375] [58] 102.68% 102.60% 103.21% Ask/bid 102.08% 102.04% 102.37% [433] [375] [58] 69
10:52 AM page:70 Academia Economic Papers 32:1 (2004 ) Table 3 Descriptive Statistics of SEO Trading Pattern (continued) Full sample Non-stabilized offer Stabilized offer Mean Mean Mean Variable Median Median Median [N] [N] [N] Panel B: Filing period (Filing day to 2 days after the filing day) $22.74 $22.95 $18.54 Offer price $21.75 $21.63 $17.63 [499] [434] [65] $23.37 $23.93 $19.61 MTP $22.00 $22.25 $18.50 [455] [396] [59] 27.44% 26.46% 34.04% Percent of MTP 24.30% 23.54% 30.77 [455] [396] [59] 396.68 423.81 214.54 Number of trades 182.00 199.00 94.00 [455] [396] [59] 48.52 cents 48.50 cents 48.62 cents Ask-bid 45.64 45.78 44.23 [448] [388] [60] 102.78% 102.72% 103.20% Ask/bid 102.17% 102.16% 102.30% [448] [388] [60] Panel C: Before offer period (3 days before 1 day before the offer day) $22.74 $22.95 $18.54 Offer price $21.75 $21.63 $17.63 [499] [434] [65] $23.12 $23.72 $19.10 MTP $21.63 $22.25 $18.13 [497] [432] [65] 28.38% 27.15% 36.55% Percent of MTP 25.88% 25.15% 34.88% [497] [432] [65] 335.39 364.69 140.69 Number of trades 146.00 157.50 84.00 [497] [432] [65] 49.18 cents 49.28 cents 48.52 cents Ask-bid 45.42 45.87 39.29 [485] [420] [65] 102.90% 102.83% 103.35% Ask/bid 102.21% 102.21% 102.22% [485] [420] [65] 70
10:52 AM page:71 Underwriter Price Stabilization of Seasoned Equity Offerings (Cotter, Chen, and Kao) Table 3 Descriptive Statistics of SEO Trading Pattern (continued) Full sample Non-stabilized offer Stabilized offer Mean Mean Mean Variable Median Median Median [N] [N] [N] Panel D: Offer period (Offer day to 2 days after the offer day) $22.74 $22.95 $18.54 Offer price $21.75 $21.63 $17.63 [499] [434] [65] $22.72 $23.35 $18.53 MTP $21.25 $22.00 $17.63 [499] [434] [65] 34.95% 30.34% 65.76% Percent of MTP 31.75% 29.66% 63.73% [499] [434] [65] 842.67 894.02 499.80 Number of trades 596.00 643.00 368.00 [499] [434] [65] 38.90 cents 39.43 cents 35.48 cents Ask-bid 37.17 38.09 32.81 [485] [420] [65] 102.28% 102.25% 102.45% Ask/bid 101.87 101.87% 101.87% [485] [420] [65] 71
10:52 AM page:72 Academia Economic Papers 32:1 (2004 ) where P (Stabilized offer): Offer price: Number of trades: Absolute value of the initial return: Transaction return volatility: Pre-offer holding period return: Underwriter prestige: Probability of stabilized offer. Offer price at issuance. Number of trades in the first three trading days. Absolute value of initial return. Standard deviation of the transaction price returns. Holding period return from the filing date to the pricing date. Total proceeds of the lead underwriter. The results presented in Table 4 suggest that, as expected, the offer price, the number of trades and the volatility of transaction returns are negatively related to the probability of an SEO being stabilized. These results are consistent with prior work on IPO stabilization by Cotter and Chen (2002), and Schultz and Zaman (1994). Interestingly, the greater the decrease in the stock price from the filing date to the offer date, the greater the likelihood of a stabilizedoffer. Finally, and unlike the results from IPOs, the prestige of the underwriter does not appear to play a role in the likelihood of stabilization. The results that find a positive relation between the prestige of the underwriter and the likelihood of stabilization may suggest that underwriters play an important role in certifying a firm whose equity value is very uncertain prior to issuance. SEOs, on the other hand, do not have this same level of uncertainty about the value of the equity, given that it is actively traded prior to issuance. In addition to the logistic results presented above, Tables 5 and 6 relate the independent variables described earlier to the actual percentage of trades occurring at the MTP in an ordinary least squares (OLS) regression framework. In these Tables, the following model is tested: 72
10:52 AM page:73 Underwriter Price Stabilization of Seasoned Equity Offerings (Cotter, Chen, and Kao) Table 4 Logistic Regression of Stabilization Estimate of a logistic regression of the probability of stabilization on proceeds, number of trades, initial return, volatility of transaction return, underwriter prestige, and prior performance. In the logistic regression, SEOs are designated as stabilized offerings based on Ruud specification, MTP specification, and multi-factor specification. In the parentheses are the p-values. ***, **, * represent significance at the 1%, 5%, and 10% level for a test of whether the parameter estimate is equal to zero. model: P (stabilization) = f (proceeds, number of trades, initial return, transaction volatility, prior performance, underwriter prestige) Ruud MTP Multi-factor specification a specification b specification c 1.422 0.076 0.170 Intercept (0.000)*** (0.848) (0.530) Proceeds d 0.005 0.003 0.004 (0.070)* (0.342) (0.054)* Number of trades e 0.000 0.001 0.000 (0.300) (0.001)*** (0.086)* Absolute value of initial 1.115 0.814 return f (0.354) (0.485) Standard deviation of the 35.069 79.122 26.690 transaction return g (0.021)** (0.000)*** (0.013)** Holdingperiodreturnprior 0.199 3.368 3.323 to issuance h (0.864) (0.015)** (0.001)*** Underwriter prestige i 0.000 0.000 0.000 (0.176) (0.273) (0.938) Number of stabilized offerings 61 65 128 Pseudo R-square 3.18% 13.69% 6.50% Pr >F 0.038** 0.0001*** 0.0001*** Sample size 499 499 499 a SEOs with day one initial return equal to zero are classified as stabilized offerings. b Stabilization is defined as SEO with percentage of the most frequently traded price larger than 50% and the ratio of the most frequently traded price to the final offer price between 0.95 and 1.0. c SEOs with 25% of transactions during the first three trading days matching the following criteria are designated as stabilized offerings: (1) trading price less than offer price; (2) transaction return equal to zero; (3) trading price at bid. d Proceeds is the product of offer price and the number of shares offered. e Number of trades during the offer day and two days later. f Initial return is defined as the first trading price on the offer day divided by the offer price and then minus one. g The standard deviation of the transaction return is the standard deviation for returns calculated using the ratio of successive transaction prices minus one for each SEO. h Holding period return is calculated from 2 days before filing day to 2 days before the offer day. i The total proceeds of each underwriter is used as the proxy for the prestige of the underwriter. The total proceeds variable is calculated based on the proceeds of individual offerings in the sample. 73
10:52 AM page:74 Academia Economic Papers 32:1 (2004 ) Table 5 Regression Analyses of Stabilization Regression analyses of the proxy of stabilization on the proceeds, number of trades, magnitude of MTP to offer price, initial return, volatility of transaction return, underwriter prestige, and prior performance for the entire sample and for the offers with initial return less than or equal to 0.0%, and higher than 0.0%. The percentage of the trades occurring at the MTP is employed as the proxy for stabilization. In the parentheses are the p-values. ***, **, * represent significance at the 1%, 5%, and 10% level for a test of whether the parameter estimate is equal to zero. Model: stabilization = f (offer price, number of trades, magnitude of MTP to offer price, transaction volatility, prior performance, underwriter prestige,) Initial return a Full Less than or More than 0.0% sample equal to 0.0% Intercept 45.208 39.335 58.505 (0.000)*** (0.000)*** (0.000)*** Offer price b 0.272 0.151 0.460 (0.000)*** (0.197) (0.000)*** Number of trades c 0.006 0.005 0.006 (0.000)*** (0.000)*** (0.000)*** Magnitude of MTP to offer 32.509 15.187 47.638 price d (0.000)*** (0.273) (0.000)*** Standard deviation of the 4.623 11.100 451.975 transaction return e (0.789) (0.590) (0.000)*** Holdingperiodreturnprior 6.271 6.414 6.107 to issuance f (0.234) (0.545) (0.272) Underwriter prestige g 0.001 0.001 0.000 (0.007)*** (0.005)*** (0.361) R-square 16.54% 11.80% 32.24% Pr >F 0.0001*** 0.0002*** 0.0001*** Sample size 499 164 335 a Initial return is defined as the first trading price on the offer day divided by the offer price and then minus one. b Offer price is the price outlined by the offering prospectus and reported by Bloomberg. c Number of trades during the offer day and two days later. d The absolute value of MTP minus offer price divided by the offer price, i.e. (MTP offer price)/offer price. e The standard deviation of the transaction return is the standard deviation for returns calculated using the ratio of successive transaction prices minus one for each SEO. f Holding period return is calculated from 2 days before filing day to 2 days before the offer day. g The total proceeds of each underwriter is used as the proxy for the prestige of the underwriter. The total proceeds variable is calculated based on the proceeds of individual offerings in the sample. 74
10:52 AM page:75 Underwriter Price Stabilization of Seasoned Equity Offerings (Cotter, Chen, and Kao) Table 6 Regression Analyses of Stabilization Regression analyses of the proxy of stabilization on the proceeds, number of trades, magnitude of MTP to offer price, initial return, volatility of transaction return, underwriter prestige, and prior performance for the entire sample and for the offers with the ratio of MTP to offer price less than or equal to 1, and higher than 1. The percentage of the trades occurring at the MTP is employed as the proxy for stabilization. In the parentheses are the p-values. ***, **, * represent significance at the 1%, 5%, and 10% level for a test of whether the parameter estimate is equal to zero. model: stabilization = f (offer price, number of trades, magnitude of MTP to offer price, transaction volatility, prior performance, underwriter prestige) Full MTP/Offer price lower MTP/Offer price sample than or equal to 1 higher than 1 Intercept 45.208 55.310 47.973 (0.000)*** (0.000)*** (0.000)*** Offer price a 0.272 0.327 0.377 (0.000)*** (0.002)*** (0.000)*** 0.006 0.009 0.002 Number of trades b (0.000)*** (0.000)*** (0.000)*** Magnitude of MTP to offer 32.509 30.413 11.768 price c (0.000)*** (0.034)** (0.127) Standard deviation of the 4.623 9.969 525.875 transaction return d (0.789) (0.645) (0.000)*** Holdingperiodreturnprior 6.271 2.155 1.804 to issuance e (0.234) (0.835) (0.631) Underwriter prestige f 0.001 0.001 0.000 (0.007)*** (0.116) (0.153) R-square 16.54% 14.01% 39.63% Pr >F 0.0001*** 0.0001*** 0.0001*** Sample size 499 223 276 a Offer price is the price outlined by the offering prospectus and reported by Bloomberg. b Number of trades during the offer day and two days later. c The absolute value of MTP minus offer price divided by the offer price, i.e. (MTP offer price)/offer price. d The standard deviation of the transaction return is the standard deviation for returns calculated using the ratio of successive transaction prices minus one for each SEO. e Holding period return is calculated from 2 days before filing day to 2 days before the offer day. f The total proceeds of each underwriter is used as the proxy for the prestige of the underwriter. The total proceeds variable is calculated based on the proceeds of individual offerings in the sample. 75