Citation
The Marketing of Seasoned Equity Offerings

Material Information

Title:
The Marketing of Seasoned Equity Offerings
Creator:
Gao, Xiaohui
Place of Publication:
[Gainesville, Fla.]
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (72 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Business Administration
Finance, Insurance and Real Estate
Committee Chair:
Ritter, Jay R.
Committee Members:
Houston, Joel F.
Nimalendran, Mahendrara
Hamilton, Jonathan H.
Graduation Date:
8/9/2008

Subjects

Subjects / Keywords:
Demand curves ( jstor )
Discounts ( jstor )
Elasticity of demand ( jstor )
Investors ( jstor )
Market prices ( jstor )
Marketing ( jstor )
Price elasticity ( jstor )
Prices ( jstor )
Standard and Poors 500 Index ( jstor )
Stock prices ( jstor )
Finance, Insurance and Real Estate -- Dissertations, Academic -- UF
marketing, seos
Genre:
Electronic Thesis or Dissertation
born-digital ( sobekcm )
Business Administration thesis, Ph.D.

Notes

Abstract:
Accelerated seasoned equity offerings (SEOs), which include bought deals and accelerated bookbuilt offers, have increased dramatically in the U.S. and globally recently. Accelerated offers are cheaper than traditional fully marketed offers in terms of direct issue costs. To explain why some issuing firms choose a fully marketed offer instead of an accelerated offer, we develop a model in which marketing flattens the issuer's demand curve. Empirical analysis shows that the pre-issue elasticity of the issuing firm's demand curve and the relative offer size are important determinants of the offer method. In our analysis, the elasticity of demand at the time of issuance is endogenous. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2008.
Local:
Adviser: Ritter, Jay R.
Statement of Responsibility:
by Xiaohui Gao.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Gao, Xiaohui. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Classification:
LD1780 2008 ( lcc )

Downloads

This item has the following downloads:


Full Text





often announced and underwritten on the same day. So the CMAR includes both the

announcement effect and price pressure caused by an inelastic demand curve. Fully marketed

deals do tend to have a larger relative size.

Denoting the closing price on the previous trading day of the offer by MPt-1, the offer price

by OP, and the closing price on the offer day by MPt, the discount is the percentage price change

from the closing price on the previous trading day of the offer to the offer price,

OP -MP 1 x 100%

MI -1 Underpricing is the percentage price change from the offer price to the

MP OP 0%
closing price on the offer day, OP The absolute value of the discount and

underpricing are similar in magnitude on average. The average SEO in our sample is sold at a

2.76% price discount and experiences 3.12% underpricing on the offer day. This is generally

consistent with the numbers reported in Altinkilic and Hansen (2003), Corwin (2003), and Mola

and Loughran (2004). Both the discount and underpricing levels remain stable during the sample

period.

These averages, however, hide important differences across offer methods. On average,

bought deals have the largest price discount of-3.96% and the smallest underpricing of 1.09%.

this large difference suggests that the price discount for bought deals is capturing the negative

announcement effect. Accelerated bookbuilt offers experience an average discount of -2.43%

and underpricing of 2.1%. If an accelerated offer is priced and offered within 24 hours after the

announcement of the offer, the two-day announcement effect, [-1,0], can be decomposed into the









Shapiro (1980). For simplicity, we assume that new investors with the same demand curve

become aware of the issue and decide to participate. The new aggregate demand curve is

(M+N)X(p), represented by the dashed line in Figure 3-1. The new demand curve is now flatter

than the old curve NX(p). Therefore, the new post-issue stock price will be higher due to the

flatter demand curve. Our model is also similar to the one in Merton (1987). In Merton's model,

a larger investor base increases the equilibrium price. But Merton does not explicitly model the

demand curve in his study.

The new demand curve results in the issuer facing a different elasticity of demand and a

higher offer price P*. With the increase in the offer price from P2 to P*, the shaded rectangular

area represents the issuer's gross gain from marketing. For a given supply increase, the gain

from marketing will be larger the more inelastic is the pre-offer demand curve. If this gain is

sufficiently large and exceeds the direct cost saving from switching to an accelerated offer, the

issuing firm would prefer the fully marketed offer.

Also, note that the size of the shaded rectangle will be larger when there is a larger supply

increase. A larger relative offer size represents a larger move along the short-run demand curve,

which implies a larger price drop if the demand curve remains the same. If the issue

significantly increases the public float and will suffer from greater price pressure, a fully

marketed offer is preferred.

Model Predictions and Testable Hypotheses

Given that the costs of conducting a road show are largely fixed costs, our model predicts

that the benefits of a fully marketed offer exceed the costs if:

* The ex ante demand curve of the issuing firm's stock is relatively inelastic, and

* The issuing firm is offering a large number of shares relative to the shares outstanding
prior to the offer









medians to test for a distribution difference among the three subsamples with different offer

methods. Our analysis focuses on mean values but results based upon medians are similar. 12

After normalizing the market capitalization and proceeds by the nominal S&P 500 Index

level with 1996 being the base year, the average issuing firm in our sample has a market

capitalization of $1.2 billion before the offer and raises $113 million in proceeds.13 On average,

the issuer is increasing the shares outstanding by 22.54%, as measured by the relative offer size,

which is defined as the ratio of the offered shares to the total shares outstanding prior to the

issue. The average percentage increase is much larger than the average proceeds as a percentage

of the average pre-issue market capitalization because larger companies typically issue a much

smaller fraction of shares than smaller companies do. Primary shares, which are shares offered

by the issuing firm, are on average 85% of the total number of shares offered, with most SEOs

having 100% of the issue coming from the company rather than existing shareholders.

As revealed in columns 2, 3, and 4 in Table 4-3, bought deals, accelerated bookbuilt offers,

and fully marketed offers differ substantially from each other. Test statistics indicate that the

difference in most of the offer characteristics is statistically significant between the three groups.

Furthermore, there is usually a monotonic relation in each mean offer characteristic among the

three groups with accelerated bookbuilt offers in between. This indicates that accelerated

bookbuilding may be an alternative method for marginal issuers in the bought deals and fully

marketed deals groups.



12 We use the non-parametric t-test (Kruskal-Wallis test) to test for difference in means among the three groups and
the non-parametric median Chi-squared test (Brown-Wood test) on medians. The Kruskal-Wallis (KW) test allows
more than two groups and does not require the dependent variables to be normally distributed.
13 The market capitalization is calculated on the last trading day before the announcement of the offer and is adjusted
by the annual nominal S&P 500 Index level in the same fashion as the adjustment on the deal proceeds. For firms
with dual class shares, the market cap is restricted to the publicly traded shares reported on the CRSP tapes.










Table 4-2. Shelf Takedown SEOs versus Non-shelf Takedown SEOs, 1996 to 2007

The sample includes 3,281 seasoned equity offerings in Dealogic's ECM Analytics Database during
1996 to 2007. Table II lists the number of shelf takedown and non-shelf takedown offers that are
bought deals, accelerated offers, and fully marketed offers. We exclude 35 non-shelf takedown offers
that are identified by Dealogic as bought deals or accelerated bookbuilt offers.


Bought Deals


Accelerated
Bookbuilt SEOs


Fully Marketed
SEOs


Shelf-takedowns

Non-shelf Offers


1,335

1,946


1,946


All SEOs









on Jan 1st, 2004. Among the 331 fully marketed deals, four offers have a time span exceeding

150 calendar days. Two of these four offers' filing dates are consistent with Factiva. Dealogic is

apparently wrong with one offer's filing date and it is reclassified as a bought deal. Another

offer has an amended file date 10 days before the offer so it is still classified as a fully marketed

deal. Therefore, we feel comfortable to rely mainly on the length of time from filing to the

offering, supplemented by Dealogic's classification of the offer method.

Next, we focus on discrepancies between SDC's classification and Dealogic's

classification of the offer method. Among the 519 offers, 416 issues' offer methods are

classified consistently and 103 are inconsistent. Among the 103 inconsistent offers, 86 offers

have consistent offering dates and 17 offers have inconsistent offering dates. If the offering date

is consistent, we find that Dealogic's offer method classification is more accurate because all

accelerated offers are completed within 3 days and all fully marketed offers take more than 3

days to complete. For the 17 offers for which the offering dates are inconsistent, Dealogic's

offering dates are correct for 14 out of the 17, where SDC is correct for only 3 of these cases.

Overall, we conclude that Dealogic is a more reliable database for our study.
































To my dad, Dechen Gao, and my mom, Jiahui Shen













Table 5-4. Offering Methods and Pre-offer Stock Price


Table 5-4 presents the results from a binomial logistic regression similar to the one in Table VI. To
test the preference of speed alternative hypothesis, we include the BHR, the buy-and-hold stock
return and the return variance during the 250 trading days prior to the SEO announcement. The rest
of the variables are defined the same as in Tables VI. The marginal effects, in percentages, are
calculated at sample means and medians and are reported to the right of the estimates. The chi-
squared statistics are reported in parentheses below. *, **, and *** indicate significance at the 10, 5,
and 1 percent levels, respectively. The R-square is the likelihood-based pseudo r-square measure.

Dependent Variable

Pr (Accelerated Deal =1)

Estimate Marginal Effect at Marginal Effect at
Means (%) Medians (%)

Pre-offer stock return -0.16 *** -0.58 -2.03
( 9.60 )
Ln(Return Variance) -0.12 ** -0.43 -1.50
( 3.74 )
Ln(A1) -0.70 *** -2.54 -8.95
( 80.78 )
Ln(MV) -0.37 *** -1.34 -4.74
( 28.78 )
Relative size (%) -0.14 *** -0.52 -1.85
( 215.19 )
Fraction of Primary Shares (%) 0.05 *** 0.17 0.61
( 50.67 )
Ln(1 + Analysts) 0.52 *** 1.88 6.65
( 23.12 )
Detrended Ln(Bid-ask Spread) 0.37 1.33 4.70
( 12.90 )
Number of SEOs 3,275

R-square 0.28









and 46 accelerated bookbuilt offers. Bought deals and accelerated bookbuilt offers also have

gained market share in terms of proceeds raised over time.8 In 2007, these accelerated offers

account for 25% of the total proceeds raised in seasoned equity offerings.

The offer proceeds fluctuate with the overall market valuation. In our sample, an average

SEO raises $94 million in 1996, $276 million in 2000, and $182 million in 2007. Therefore, we

adjust the nominal proceeds via scaling by the nominal year-end value of Standard & Poor 500

Index level. The normalized proceeds, expressed in terms of 1996 valuations, are given by

S. S&P 500 Index
Nominal Proceeds x 1S&P996
S&P 500 IndexSEOYear

In unreported results, using two-digit standard industry classification (SIC) codes, 68

industries are represented, with 75% by industrial firms (2,460 out of 3,281) and 25% by

regulated and financial firms.9 For bought deals and accelerated bookbuilt offers, around 60%

(340 out of 569) are industrial firms. For fully marketed offers, 78% of issuers (2,120 out of

2,712) are industrial firms. Over the sample period, the industry representation in each year

exhibits no particular pattern.

During our twelve-year sample period, 1,573 firms conduct only one follow-on offering

and 660 firms conduct more than one SEO. Among the 660 repeated SEO issuers, 24 conduct

more than 5 offers with the most frequent issuers being financial firms.10 We identify 265 SEOs

where the offer method is different from the most recent SEO by the same issuer. Among the 58

SEOs that switched from an accelerated offer to a fully marketed offer, 48 have a larger relative

8 The total proceeds include the exercised overallotment value. Unlike the SDC New Issues Database, Dealogic
reliably reports the exercised amount.
9 Utility and other historically regulated firms, including transportation and telecom, have two-digit SIC codes
between 40 and 49. Financial firms have SIC codes between 60 and 69. All other firms are industrial firms.

10 Two firms, Allied Capital and American Capital Strategies (ACAS), did more than 10 offers. ACAS, which
invests in buyouts, had 26 SEOs in 1999-2007.










Table 4-3. Mean and Median Offer Characteristics of SEOs, 1996 to 2007
The sample is restricted to 3,281 issuing firms that have more than $75 million market capitalization before the offer. Table III lists means (medians in
brackets) of offer characteristics. Market capitalization is the total market capitalization of equity on the last day prior to the announcement of the offer.
Proceeds is the total amount of dollars (in millions) raised globally including exercised overallotment shares in all trenches. Normalized market capitalization
and proceeds are adjusted by the annual nominal S&P 500 Index level, normalized by the ratio between the 1996 S&P 500 index level and the SEO year's S&P
500 index level. Relative offer size equals offered shares divided by total shares outstanding prior to the issue. Fraction of primary shares equals primary (new)
shares divided by total number of shares offered. Number of bookrunners is the number of bookrunners on the issue. Bookrunner reputation is the
bookrunner's Carter-Manaster ranking obtained from Jay Ritter's webpage. If there are multiple bookrunners, we use the maximum ranking among all the
bookrunners. Number of days from filing to offer is the number of days between the filing date and the offer date. Gross spread is the disclosed gross fee paid
as percentage of the offer price. Pre-offer return is the buy-and-hold stock return during the 250 trading days, [-250, -1], ending with the announcement date
(trading day 0). Announcement effect is the cumulative market-adjusted return during the two-day window, [-1,0], ending with the announcement date .
Discount is the percentage decrease from the closing price day before the offer to the offer price. Underpricing is the percentage increase from the offer price to
the closing price on the offer day. The number of analysts following the issuer's stock is obtained from the I/B/E/S database. We include analysts who post a
recommendation within 12 months before the offer. Bid-ask spread (%o) is the average daily bid ask spread, scaled by the stock price, over 250 trading days
prior to the announcement date. The test for the difference in means among the three groups is the non-parametric t-test (Kruskal-Wallis test), which allows
more than two groups and does not require the dependent variables to be normally distributed. The test for the medians is the non-parametric median Chi-
squared test (Brown-Mood test). P-values from the KW and the median Chi-squared statistics are reported in the last column.


All SEOs

3,281


Bought
Deals

293


Accelerated
Bookbuilt
SEOs
276


Fully
Marketed
SEOs
2,712


P-values from
KW/Median
Statistics


Nominal Market Capitalization ($M)


Normalized Market Capitalization ($M)


1,886
[ 550

1,182
[ 360

176
[ 97

113
[ 64


Nominal Proceeds ($M)


Normalized Proceeds ($M)


Relative Offer Size (%) 22.54
[ 18.98

Fraction of Primary Shares (%) 85.00
[ 100


4,552
] [ 1,627

2,772
] [ 961

219
] [ 129

136
] [ 81

9.36
] [ 7.53

98.83
] [ 100


4,272
] [ 1,067

2,718
] [ 674

208
] [ 88

134
] [ 54

11.37
] [ 9.87


I [


98.58
100


Number


1,356
] [ 463

854
] [ 315

168
] [ 95

108
] [ 64

25.10
] [ 21.48

82.12
] [ 100


0.00
] [ 0.00

0.00
S[ 0.00

0.00
S[ 0.00

0.00
] [ 0.18

0.00
] [ 0.00

0.00
S[ 0.00









portfolio managers that are benchmarked to the index is difficult to quantify. When shares are

issued by the company, however, it is possible to estimate an elasticity. Kandel, Sarig, and Wohl

(1999) analyze the full demand schedules of 27 Israeli IPO auctions and estimate that the

average elasticity is -37, indicating that a 37% increase in shares issued is associated with a 1%

fall in price.

Seasoned equity offerings result in an increase in the supply of shares to the public. In the

U.S., but not all countries, the announcements are associated with stock price falls. The fall has

been interpreted in the literature as due to a combination of information (Myers and Majluf,

1984) and supply effects. Loderer, Cooney, and Van Drunen (1991) study the announcement

effect of seasoned offerings by regulated firms. Their event study is a direct test on price

elasticity of common stocks because regulated firms are less likely to suffer from information

asymmetry. Their results support finite price elasticities.

Share repurchases represent a decrease in the supply of shares to the public. A positively

sloped supply curve for buyers is analogous to a negatively sloped demand curve for sellers.

Bagwell (1992) documents that firms face upward-sloping supply curves when they repurchase

shares in a Dutch auction. Hodrick (1999) confirms the existence of an inelastic demand curve

and concludes that the expected stock elasticity is an important determinant of the tender offer

choice. She measures the inverse elasticity as the price premium above the market price before

the announcement divided by the fraction of shares tendered. She finds an average inverse

elasticity of 1.27 for Dutch auction firms and 0.99 for fixed price tender offer firms. Her study is

similar to ours in that she focuses on differences in elasticities determining corporate choices.

Unlike our analysis, however, she takes the elasticity as exogenous.









turnover is 0.02%, corresponding to an annual turnover of 5%, before its SEO on November 3rd,

2003. This small turnover rate generates an A1 of 840, almost 100 times the unconditional

median of 9.02. Consequently, we use a natural log transformation of A, and A2 in our empirical

work. The last three rows in Panels A and B report the log transformed values of A, and A2,

Ln(Ai) and Ln(A2). As expected, the log transformation significantly reduces the skewness in A1.

The log transformation has a smaller impact on the level ofA2.

As a robustness check, we substitute the stock's raw return in by its market adjusted

return using the CRSP value-weighted market index. The results remain the same. Our results

are also robust to the specification of the time window. We estimate and over one-month,

three-month, and six-month windows prior to the announcement date and all estimates are

qualitatively similar across the different windows.

Determinants of Offer Method

Next, we examine the determinants of the choice of offer method in a multivariate

framework.

Our univariate analysis shows that the choice of offer method is related to firm and offer

characteristics. In the multivariate analysis, we include six firm and offer characteristic variables

as the explanatory variables. The four firm characteristics are our two proxies for the elasticity

of the short-run demand curve (the estimated inverse elasticity or the estimated arbitrage risk);

the market capitalization before the announcement; the average bid-ask spread, scaled by the

stock price, over the 250 trading days prior to the announcement; and the number of analyst

recommendations within 12 months before the offer, obtained from the I/B/E/S database. We









probability of using an accelerated offer is 51.6 %. The fraction of primary shares and the

amount of analyst coverage also have significant impacts on the choice of the SEO offer method,

suggesting that firms with high information asymmetry tend to choose fully marketed offers.









L IST O F R E F E R E N C E S .............................................................................. ...........................67

B IO G R A PH IC A L SK E T C H .............................................................................. .....................72





















































6









from the fact that smaller firms tend to have higher returns and larger variance regression results

while they also tend to choose fully marketed offers. We want to point out that the unlogged

stock return variance is very highly correlated (0.99) with our second elasticity proxy, A2, the

residual variance. In addition, during our sample period, there is a monotonic downtrend in the

median time between the announcement and completed fully marketed deals. In 1996, the

median time is 31 days and it decreases to only 9 days in 2007. Therefore, the speed advantage

of an accelerated deal may have become less important in recent years.









offer size compared to the previous accelerated offer and the average increase in relative offer

size is 7.1%. Among the 207 SEOs that switched from a fully marketed offer to an accelerated

offer, 172 have a smaller relative offer size and the average decrease is 16.9%. Among all the

multiple issuers, the average time between two issues is 34 months.

The mean number of years from the issuing firm's IPO to the SEO is 10 years with a

median of 5 years. Firms conducting accelerated SEOs are usually more seasoned than those

conducting fully marketed offers, with medians of 10 years and 4 years respectively. More

NYSE-listed companies (27%) do accelerated offers compared to NASDAQ-listed companies

(12%). This is hardly surprising because large firms are more likely to do an accelerated offer,

as we will see in Section 4.4.

Shelf-takedown Offers

The deregulated streamlined shelf registration process allows issuing firms to file a single

all-encompassing registration statement once every two years rather than filing individual

registration statements for every security offering. Once its shelf registration statement is

approved by the SEC, a firm can issue securities without further disclosure requirements or

regulatory delays. In practice, even for bought and accelerated bookbuilt deals, firms normally

issue a prospectus.

Shelf registration eligibility may lead to a potential selection bias in our analysis on the

determinants of the offer method. An issuer needs to file a shelf registration prior to conducting

an accelerated SEO. This requires the issuer to have at least $75 million in market

capitalization. Therefore, some small firms choose a fully marketed offer because they are not

eligible for shelf registration and therefore cannot conduct an accelerated offer. These firms also


1 There are other requirements other than the market capitalization requirement. For details, see Bethel and
Krigman (2006).









patterns. Section 5 estimates two demand elasticity proxies and analyzes the determinants of

offer method with a binomial logistic model. Section 6 concludes.









Loderer, Claudio F., John W. Cooney, and Leonard D. Van Drunen, 1991, "The Price Elasticity
of Demand for Common Stock," Journal ofFinance 46, 621-651.

Lynch, Anthony, and Richard Mendenhall, 1997, "New Evidence On Stock Price Effects
Associated with Changes in the S&P 500 Index," Journal of Business 70, 351-383.

Madhavan, Ananth, 2003, "The Russell Reconstitution Effect," Financial Analyst Journal 59,
51-64.

Masulis, Ronald W., and Ashok N. Korwar, 1986, "Seasoned Equity Offerings: An Empirical
Investigation," Journal ofFinancial Economics 15, 91-118.

Meidan, Danny, 2004, "A Re-examination of Price Pressure around Seasoned Equity Offerings,"
Northwestern University working paper.

Merton, Robert C., 1987, "A Simple Model of Capital Market Equilibrium with Incomplete
Information," Journal ofFinance 42, 483-510.

Mola, Simona, and Tim Loughran, 2004, "Discounting and Clustering in Seasoned Equity
Offering Prices," Journal ofFinancial and Quantitative Analysis 39, 1-23.

Morck, Randall, and Fan Yang, 2002, "The Mysterious Growing Value of S&P 500
Membership", University of Alberta working paper.

Myers, Stewart C., and Nicholas S. Majluf, 1984, "Corporate Financing and Investment
Decisions When Firms Have Information that Investors Do Not Have," Journal of
Financial Economics 13, 187-221.

O'Hara, Maureen, Market Microstructure Theory, 1995, Basil Blackwell, Cambridge, MA.

Richard Carter and Steven Manaster, 1990, "Initial Public Offerings and Underwriter
Reputation", Journal ofFinance 45, 1045 1067.

Rinne, Kalle and Matti J. Suominen, 2007, "Value Uncertainty, Price Impact and the Choice of
the Equity Issuing Method", Helsinki School of Economics working paper.

Ritter, Jay R., 2003, "Investment Banking and Securities Issuance," Handbook of the Economics
ofFinance (North-Holland: Amsterdam), Ch 5.

Scholes, Myron S., 1972, "The Market for Securities: Substitution versus Price Pressure and the
Effects of Information on Share prices," Journal of Business 45, 179-211.

Schultz, Paul H., 2008, "Downward Sloping Demand Curves, the Supply of Shares, and the
Collapse of Internet Stock Prices," Journal ofFinance 63, 351-378.

Shapiro, Carl, 1980, "Advertising and Welfare: Comment," The Bell Journal of Economics 11,
749-752.










Shleifer, Andrei, 1986, "Do Demand Curves Slope Down?" Journal ofFinance 41, 579-590.

Smith, Clifford W. Jr., 1977, "Alternative Methods for Raising Capital: Rights vs. Underwritten
Offers," Journal of Financial Economics 5, 273-307.

Securities and Exchange Commission release 2005-99, "SEC Votes To Adopt Securities Act
Rule Reform and Shell Company Regulations; Considers Matters Remanded by Court of
Appeals", http://www.sec.gov/news/press/2005-99.htm, July 1st 2005.
Warn, Ken, "Salomon Places $lbn YPF Shares with Institutions," Financial Times (July 17th,
1997), downloaded from Factiva.

Wurgler, Jeffrey, and Ekaterina Zhuravskaya, 2002, "Does Arbitrage Flatten Demand Curves for
Stocks?" Journal of Business 75, 582-608.

Yuan, Yu, 2008, "Attention and Trading," University of Pennsylvania Working Paper.

Zhang, Donghang, 2004, "Why Do IPO Underwriters Allocate Extra Shares When They Expect
to Buy Them Back?" Journal ofFinancial and Quantitative Analysis 39, 571-594.










BIOGRAPHICAL SKETCH

Xiaohui Gao earned her Bachelor of Science in mathematics with concentration in Control

Science from Fudan University, China, in 2001. She earned her Master of Science in

mathematics with concentration in probability and statistics from the University of Florida in

2003. Requirements for the degree of Doctor of Philosophy in finance were completed at the

University of Florida during the summer of 2008. Upon graduation, Gao will join the University

of Hong Kong as an assistant professor in the Department of Economics and Finance. Gao's

primary research interests include security offerings and mutual funds.









however. The issuing firm needs to balance the gain from a higher stock price with the extra

underwriting expenses associated with greater marketing effort.

Downward Sloping Demand Curves

In a classical efficient market framework, demand curves for stocks are horizontal. At any

point in time, market prices reflect all available information. Therefore, a shock in the demand

or supply should not move the stock price if it does not change the expectation of the underlying

value. Empirical evidence from many sources, however, suggests that demand curves for

common stocks are actually downward sloping.

Previous studies reveal that both short-run and long-run demand curves are not perfectly

elastic. Scholes (1972) uses 'price pressure' to describe the downward sloping short-run demand

curves that he documents. He studies block sales and secondary offerings and documents a price

decline in such distributions of securities. However, Scholes interprets his findings as indicating

that information effects are the main cause of the price changes. Current studies most often

attribute the short-run effect on price to market liquidity constraints. Keim and Madhavan

(1996) document significant temporary price impacts from large block trades.

A number of studies that look at price changes when stocks are added to or deleted from an

index distinguish between short-run and long-run demand curves. Index additions and deletions

represent a clean test of demand curve slopes because in most cases they are per se

informationless. Shleifer (1986), Harris and Gurel (1986), Dhillon and Johnson (1991), Lynch

and Mendenhall (1997), and Wurgler and Zhuravskaya (2002) all document positive abnormal

returns when stocks are added to the Standard and Poor's 500 Index, with the positive abnormal

returns only partly reversed in subsequent trading. Furthermore, these studies show that the

magnitude of the announcement effects has increased over time, paralleling the growth in the

assets of funds that are indexed to the S&P 500. Lynch and Mendenhall (1997) find evidence of









SEO and then sets the offer price. More syndicate members, including co-managers and non-

managing underwriters, can reach out to more investors and provide more information about

market interest in the offer, which adds to the marketing efforts.

Corwin and Schultz (2005) examine the role of syndicate members in IPOs and find strong

evidence of information production. Huang and Zhang (2008) find that more managing

underwriters, including bookrunners and co-managers, result in a smaller offer price discount,

defined as the percentage price change from the previous market price to the offer price. They

attribute this to the managing underwriters' pre- and post-issue marketing efforts. Consistent

with these two studies, we find that fully marketed offers tend to have more co-managers and

larger syndicate groups compared to accelerated offers. In our sample, 66% of the accelerated

SEOs do not hire a co-manager. In contrast, only 6.1% of the fully marketed offers do not have a

co-manager.

Our study supports the argument that demand curve elasticity has a significant impact on

corporate financial decisions. Hodrick (1999) studies tender offer choice in share repurchases

and finds that firms that choose Dutch auctions instead of the fixed price tender offers tend to

face more elastic demand curves. Baker, Coval, and Stein (2007) argue that for acquiring firms

that face downward-sloping demand curves, investor's inertial behavior (i.e. the tendency to

react passively) mitigates the negative merger announcement effect and increases the benefit of a

stock-for-stock merger.

The rest of the paper is organized as follows: In section 2, we review the related literature

on marketing and stock demand elasticities. Section 3 develops a model of the role of marketing

in seasoned equity offerings. Section 4 discusses the sample construction and presents univariate









covering the stock lower the probability of an accelerated offer. The analyst coverage can be

viewed as a proxy for information asymmetry and the elasticity of demand. The smaller fraction

of primary shares is associated with a highly probability of overvalued stock, e.g. Lee (1997),

which is also related to the information asymmetry. Therefore, an issuing firm with high

information asymmetry is more likely to choose a fully marketed offer. The detrended log bid-

ask spread is positively correlated with the likelihood of an accelerated offer in the first model.

This is a surprising result, since we would expect that stocks with a high bid-ask spread would be

more likely to use a fully marketed offer.

One issue is whether the relative offer size is determined jointly with the choice of the

offer method, or whether this fraction is predetermined. For a fully marketed offer, if the stock

price drops too much after the announcement, or there is little demand for the issue, the issuing

firm may reduce the offer size or cancel the offer. On the other hand, it sometimes increases the

offer size if the demand is strong. For an accelerated offer, the underwriter has little time to

assess investors' demand but may still increase or decrease the offer size according to the

demand information collected in a very short time period. Around 75% of the bought deals (220

out of 293) and 44% of the accelerated bookbuilt offers (122 out of 276) reduce the deal value

and 36% of the fully marketed offers (975 out of 2,712) reduce the deal size.

For our empirical analysis, the question is whether this potential endogeneity problem

biases our results. To examine this, we also try another measure of the relative offer size. It is

defined as the ratio between the initial deal size, announced at the filing, and the issuing firm's

market capitalization prior to the announcement. Our results remain virtually identical, so this

potential endogeneity problem does not affect our conclusions.









Brav, Alon, and Paul A. Gompers, 2003, "The Role of Lock-ups in Initial Public Offerings,"
Review ofFinancial Studies 15, 1-29.

Busse, Jeffrey A., and T. Clinton Green, 2002, "Market Efficiency in Real Time", Journal of
Financial Economics 65, 413 437.

Chanine, Salim, Alexander Ljungqvist, and Roni Michaely, 2008, "Marketing Financial Claims",
working paper.

Chen, Hsuan-Chi, and Jay R. Ritter, 2000, "The Seven Percent Solution," Journal ofFinance 55,
1105-1131.

Comanor, William S., and Tomas A. Wilson, 1967, "Advertising Market Structure and
Performance," Review of Economics and Statistics 49, 423-440.

Corwin, Shane A., 2003, "The Determinants of Underpricing for Seasoned Equity Offers,"
Journal ofFinance 58, 2249-2279.

Corwin, Shane A., and Paul Schultz, 2005, "The Role of IPO Underwriting Syndicates: Pricing,
Information Production, and Underwriter Competition," Journal ofFinance 60, 443-486.

Dhillon, Upinder, and Herb Johnson, 1991, "Changes in the Standard and Poor's 500 List,"
Journal ofBusiness 64, 75-85.

Dixit, Avinash, and Victor Norman, 1978, "Advertising and Welfare," The Bell Journal of
Economics 9, 1-17.

Eckbo, B. Espen, and Ronald W. Masulis, 1992, "Adverse Selection and the Rights Offer
Paradox," Journal ofFinancial Economics 32, 293-332.

Eckbo, B. Espen, Masulis, Ronald W. and Norli, Oyvind, 2007, "Security Offerings", Handbook
of Corporate Finance: Empirical Corporate Finance (North-Holland: Amsterdam),
Chl3.

Eckbo, B. Espen and Norli, Oyvind, "The Choice of Seasoned-equity Selling Mechanism:
Theory and Evidence" (November 2004). Tuck School of Business Working Paper.

Field, Laura Casares, and Gordon Hanka, 2001, "The Expiration of IPO Shares Lockups,"
Journal ofFinance 56, 471-500.

Greenwood, Robin, 2005, "Short- and long-term Demand Curves for Stocks: Theory and
Evidence on the Dynamics of Arbitrage," Journal ofFinancial Economics 75, 607 649.

Hansen, Robert S., and John M. Pinkerton, 1982, "Direct Equity Financing: A Resolution of a
Paradox." Journal ofFinance 37, 651-665.











Table 5-1. Summary Statistics for Demand Elasticity Proxies


The sample includes 3,279 seasoned equity offerings during 1996 to 2007. Two offers are dropped
because their stock price information is unavailable on the University of Chicago Center for Research
in Security Prices (CRSP) database over the 250 trading day window [-250, -1] before the
announcement date. The average daily order flow inverse demand elasticity, Al, is defined as the
daily raw return divided by the daily turnover, averaged over 250 trading days before the
announcement date. The turnover is the trading volume divided by the number of shares outstanding.
Nasdaq-listed stocks' trading volumes are divided by two to eliminate double counting. The arbitrage
risk measure, A2, is constructed similar to that in Wurgler and Zhuravskaya (2002). A2 is the
residual variance, expressed as a squared percentage of a market model OLS regression estimated
over 250 trading days before the announcement date. We report means, medians, and standard
deviations of the raw values and natural log transformed values of Al and A2. In the last column,
the KW test and median test p-values for the means and the medians are the same as in Table III.


Panel A: Order Flow Inverse Demand Elasticity Al


Number


Raw Value Mean
Raw Value
Median
Raw Value STD

Ln Value Mean

Ln Value Median

Ln Value STD


All
SEOs

3,279

28.22

8.52

93.08

2.31

2.14

1.26


Bought
Deals

293

10.01

3.70


56.67

1.39

1.31

0.86


Accelerated
Bookbuilt
SEOs

276

10.17

4.03


27.22

1.59

1.39

1.00


Fully Marketed
SEOs


P-values from
KW/Median
Statistics


2,710

32.03

10.17

99.89


Panel B: Arbitrage Risk Measure A2 (%^2)


Number

Raw Value Mean
Raw Value
Median
Raw Value STD

Ln Value Mean

LnValue Median

Ln Value STD


All
SEOs

3,279

16.92

10.56

20.81

2.31

2.36

1.06


Bought
Deals

293

11.19


6.85

13.96

1.82

1.92

1.14


Accelerated
Bookbuilt
SEOs

276

10.72


4.99

15.35

1.71

1.61

1.15


Fully Marketed
SEOs


P-values from
KW/Median
Statistics


2,710

18.17

11.64

21.68










million in proceeds for example, a 2% gross spread represents $4 million in revenue for just a

few days work.

In Table 4-3, we also report the pre-offer stock return, announcement effect, price

discount, and underpricing. The pre-offer stock return is the buy-and-hold stock return (BHR)

estimated over the 250 trading days before the announcement. The announcement effect is the
0
cumulative market-adjusted return (CMAR) estimated over the two-day windi(3-ilQ~t, )
t= 1
,ending with the announcement date (day 0).17 On average, our sample offers experience a

significant run-up of 107% and -1.72% cumulative market adjusted return around the

announcement of the offer. If we expand the CMAR to a three-day window, [-1,1], the average

return is -2.1%, which is comparable to the numbers presented in previous studies.18 Our results

remain qualitatively the same if we use the three-day window.

The pre-offer stock run-up and announcement effects vary among the three groups. Fully

marketed offers experience the largest pre-offer BHR, an average of 118%. Bought deals and

accelerated bookbuilt offers experience smaller stock price increase, 72% and 45% respectively.

But note that accelerated issuers tend to be big firms so they are less likely to have a huge price

run-up.

Bought deals suffer the least amount of negative market reaction, an average of -1.49%.

Fully marketed offers experience an average of -1.66% and accelerated bookbuilt offers

experience an average of -2.55%. But note that bought deals and accelerated bookbuilt offers are



17 Dealogic reports the number of days from the announcement to the offer. We checked several offers and find that
if the announcement is made after the market closes and the offer takes place on the next day, Dealogic reports 0
between the announcement date and the offer date. So the announcement day should be the actual day that the stock
price reacts to the announcement. Therefore, unlike some previous studies on the announcement effect, we exclude
the return on the day after the announcement.

18 Interested readers can also refer to Table 5 in Ritter (2003) and Table 13 in Eckbo, Masulis, and Norli (2007),
which summarize the announcement effects from different studies.









CHAPTER 1
INTRODUCTION

Seasoned equity offerings (SEOs), also known as follow-on offerings, can be categorized

into three major types by their offer methods: fully marketed offers, accelerated offers, and rights

offers.1 The academic literature on the issue method for SEOs has focused on rights versus

underwritten (bookbuilt) offers, although rights offers are virtually nonexistent in the U.S.2 In

contrast, there is no theoretical or empirical treatment of the choice between accelerated offers

and traditional bookbuilt offers, also known as fully marketed offers.

Fully marketed SEOs are issued in much the same way as bookbuilt initial public offerings

(IPOs). Firms wishing to raise cash issue primary shares and current shareholders wishing to sell

existing shares issue secondary shares, and a prospectus is printed. The issuer negotiates with

one or more investment banks to market the offer and then set the price. The lead underwriter or

underwriters conduct a due diligence investigation and "certify" the quality of the company. As

part of the marketing effort to develop interest in the offer, the bookrunner usually conducts a

road show. In a typical road show, the issuer's management meets with selected institutional

investors, analysts, and securities sales personnel over a two week period. At the same time, the

bookrunner assesses investors' demand and builds an order book, which is used to help

determine the offer price. A syndicate of underwriters, led by the bookrunner, then distributes

the shares, although for some of the underwriters, their only involvement may be in providing

subsequent research coverage, or even less.





1 Throughout this paper, the following terms: fully marketed, fully underwritten, regular, firm commitment,
bookbuilt, and conventional offers, are used interchangeably unless noted otherwise.
2 Previous studies on rights offers versus bookbuilt offers include Smith (1977), Heinkel and Schwartz (1986),
Eckbo and Masulis (1992), Bohen, Eckbo, and Michalsen (1997), and Eckbo and Oyvind (2" '14).









LIST OF TABLES


Table page

4-1 Number of SEOs and Aggregate Proceeds by Year and Offer Method. .........................43

4-2 Shelf Takedown SEOs versus Non-shelf Takedown SEOs, 1996 to 2007........................44

4-3 Mean and Median Offer Characteristics of SEOs, 1996 to 2007 ...................................45

5-1 Summary Statistics for Demand Elasticity Proxies ................................. ............... 57

5-2 Correlation among Demand Elasticity Proxies and Firm and Offer Characteristics .........58

5-3 Determinants of Offering Methods for SEOs, 1996 to 2007 .....................................59

5-4 Offering M methods and Pre-offer Stock Price ........................................ .....................60









In each of the first five years of our 1996-2007 sample period, less than 10% of SEOs are

an accelerated offer, where as in each of the last seven years, at least 20% of SEOs do so. Thus,

we rerun the binomial regression for the sub-period 2001 to 2007 and present the results in the

second regression of Table 5-3. The sub-period results are stronger in the later sample period as

most of the marginal effects are larger. Ln(A2) becomes insignificant due to a higher correlation

(0.37) with Ln(A) If we exclude Ln(A) from the regression, Ln(A2) becomes significant and

the rest of the coefficients remain unchanged. The bid-ask spread also becomes insignificant in

the later sample period.

Does Speed Rather than Marketing Determine the Issue Method?

Finally, we discuss an alternative explanation on the choice between accelerated and fully

marketed offers. The choice of an accelerated offer may indicate the issuing firm's preference

for speed in the raising of equity capital. Firms usually conduct SEOs after a substantial price

run-up, as we have seen in Table 4-3. If the issuing firm thinks that the high stock price may not

be sustained during a fully marketed SEO process, it might prefer an accelerated offer to raise

the money quickly. To test this alternative hypothesis, we use two proxies to measure the issuing

firm's preference for speed. The first measure is the BHR, the buy-and-hold stock return during

the 250 trading days (one year) prior to the SEO announcement. The second measure is the

stock return variance during the same time period. Under the alternative hypothesis, firms that

have experienced an enormous price appreciation would have a stronger incentive to issue

quickly. If the stock price is volatile, the issuing firm might prefer an accelerated offer to avoid a

possible sharp decline in the stock price.

Table 5-4 presents the regression results of a binomial model similar to that in Table 5-3.

The alternative hypothesis is not supported by the results. The BHR and the return variance are

both negatively correlated with the probability of doing an accelerated offer. This may result









Extension of the Model

Two assumptions in this model can be relaxed to make it more general.

First, investors can have heterogeneous beliefs. Even if the new investors are less

optimistic than current shareholders, their participation will still flatten the demand curve, as

pointed out in Shapiro (1980). In fact, Bagwell (1992) and Kandel, Sarig, and Wohl (1999) have

demonstrated that investors have heterogeneous beliefs, or at least heterogeneous reservation

prices due to differential tax status.

Second, the marketing of SEO can shift the demand curve even if it does not change the

elasticity. The announcement of the SEO may convey information to the market. Myers and

Majluf (1984) show that under certain conditions, the market interprets the announcement of a

seasoned offer as a negative signal that the issuing firm is overvalued. The demand curve is

shifted downward in parallel and results in the price drop upon the announcement of the offer.

After the announcement, the underwriter's marketing efforts may reduce the information

asymmetry so the demand curve will be flattened and shifted upward in parallel. In our

empirical study, we do not attempt to disentangle the information asymmetry and the price

pressure. However, we predict that if a firm faces high information asymmetry prior to the issue,

it may benefit more from fully marketing the offer.









CHAPTER 6
CONCLUSION

The recent rise in accelerated SEOs in the U.S. offers us a chance to study the determinants

and economics associated with alternative offer methods. We study three major SEO offer

methods: bought deals, accelerated bookbuilt offers, and fully marketed offers.

The main difference between an accelerated offer, including bought deals and accelerated

bookbuilt offers, and a fully marketed offer is that there is no marketing for an accelerated offer.

Therefore, we focus on the role of marketing in SEOs.

We first develop a model of the issuing firm's stock demand and supply curves. The

issuance represents an increase in the supply. With a downward sloping demand curve, the stock

price decreases due to this increase in supply. Marketing flattens the demand curve and helps to

achieve a higher price after the offer. Therefore, the elasticity of demand curve is endogenous

and is affected by the chosen offer method. The model predicts that a fully marketed SEO has

greater benefits to the issuing firm when it ex ante faces a highly inelastic demand curve and

when the relative issue size is large.

Our empirical analysis supports the model predictions. We use two measures, the order

flow inverse demand elasticity A1 and the arbitrage risk measure A2, to proxy for the demand

elasticity. Regression results show that the ex ante elasticity of the issuing firm's demand curve

and the relative offer size are important determinants of the offer method. For an issuing firm

that is average in other ways, if it has an above average relative issue size of 30% of the pre-issue

shares outstanding and ex ante has a relatively inelastic demand curve, with an Ln(Ai) value of

4.06 (the 90th percentile of the order flow inverse demand elasticity), the probability of using an

accelerated offer is 0.4%. If the issuer has a relative offer size of 10% and is in the 10th

percentile of the order flow inverse demand elasticity (with an Ln(Ai) value of 0.84), the















Bought Deal Process

Investment banks submit bids after the close of trading


Accelerated Bookbuilt SEO Process

Select the bookrunner


Bank that offers the highest net price wins


Registration





Re-sellthe shares 4


Announcement of the filing


S Usually overnight


Trade begins


Registrationkbu



Accelerated bookbuilding
-----


Announcement ofthefilii



I Usually 1-2 days


Pricing and allocation Trade begins


Fully Marketed SEO Process


Select bookrunner and co-managers


Prepare preliminary prospectus


Registration Announcement ofthefiling


Marketing, road show, and bookbuilding 2 -3 weeks


Issue final prospectus, pricing, and allocation Trade begins


Figure 2-1. Description of the SEO Process Associated with the Three Offering Methods









LIST OF FIGURES


Figure page

2-1 Description of the SEO Process Associated with the Three Offering Methods ................23

3-1 Comparison of Demand Curves with and without Marketing................ ..................28

4-1 Scatter Diagram Relating Relative Offer Size to Log(Market Capitalization) ................42









THE MARKETING OF SEASONED EQUITY OFFERINGS


By

XIAOHUI GAO






















A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2008









or Ln(A2) decreases the probability of an accelerated offer by 2.56% and 0.79% respectively.

The marginal probability at medians is calculated in a similar way at the sample median level.

Error! Not a valid link. We start with a firm at the bottom 25th percentile of the order

flow inverse demand elasticity, which has a value ofLn(A1) equal to 1.43. This firm has a more

elastic demand curve than 75 percent of the sample. We hold the residual variance Ln(A2) the

market capitalization, and the relative size at the 25th percentile and the primary fraction, the

bid-ask spread, and the number of analysts at the 75th percentile. If we increase Ln(A1) by one

standard deviation (1.24), the probability of an accelerated offer decreases from 66.9% to 45.4%,

a marginal probability of 17%. Similarly, a one standard deviation increase in Ln(A2) (1.04)

decreases the probability of an accelerated offer from 66.9% to 61.7%, a marginal probability of

5.0%. For these values of the explanatory variables, the relative offer size has an even larger

impact on the offer method choice. For the same values of the variables, a one standard

deviation increase (18.84%) in relative offer size decreases the probability of an accelerated offer

from 66.9% to 11.7%.

The multivariate regression results support our first and second hypotheses. Both the pre-

issue demand elasticity and the relative offer size are important determinants of the offer method.

Firms facing a more inelastic demand curve are more likely to fully market the offer. A larger

relative offer size increases the probability of a fully marketed offer.

The results in Table 5-3 reveal other important determinants of the offer method. A larger

firm size decrease the probability of an accelerated offer. This is a rather counterintuitive result.

It is probably because the firm size is in the denominator of the relative offer size. And this is

also consistent with the pattern in Figure 4-1. Holding the relative size constant, larger firms

seem to prefer a fully marketed offer. A smaller fraction of primary shares and fewer analysts









"accelerated" in the sense that no road show is conducted and the underwriting procedure is

typically completed within 48 hours.6

Before the late 1990s, the US equity market was dominated by fully marketed SEOs. In

contrast, many Asian, European, and Australian SEOs used rights offers. In a rights offer,

current shareholders are given short-term warrants to purchase newly issued shares on a pro rata

basis at a discount relative to the current market price of the stock.

The major difference between a fully marketed offer and an accelerated offer, aside from

the speed of completing the deal, is the extent of the underwriter's marketing of the issue. We

develop a model of the role of marketing in seasoned equity offerings to explain the issuing

firm's choice between a fully marketed offer and an accelerated offer. In the model, marketing

flattens the demand curve of the issuer's stock. The underwriter's marketing effort, i.e. a road

show, can change the issuing firm's stock demand elasticity.

To the best of our knowledge, this is the first time this marketing paradigm has been used

in the finance literature. The bookbuilding literature has focused on the acquisition of

information by underwriters about the state of the demand, e.g., Benveniste and Spindt (1989).

In contrast, we posit that an important feature of traditional bookbuilding is the marketing of the

issue. While many studies have assumed that a negatively sloped demand curve exists, these

papers have assumed that the elasticity of demand is exogenous. This paper not only explicitly

adopts a supply and demand framework to analyze SEOs, but posits that the demand curve that

an issuing firm faces is affected by the choice of issue methods. In other words, the short-run

demand curve that an issuing firm faces is an endogenous choice variable.



6 This is confirmed in our sample where 96% of accelerated offers, including bought deals and accelerated bookbuilt
offers, are completed within two days.









rank bookrunners' reputation using the Carter-Manaster ranking obtained from Jay Ritter's

website. The highest Carter-Manaster ranking is 9 and the lower the ranking is, the less

prestigious the bookrunner is. If there are multiple bookrunners, we use the maximum ranking

among the bookrunners. The mean bookrunner ranking is 8.25 and the median is 9, so most

SEOs are underwritten by high reputation banks. Bought deals have slightly more prestigious

bookrunners but the difference is not significant.

Bought deals and accelerated bookbuilt offers, as suggested by the offer method, have a

very short underwriting process.15 Not surprisingly, Table 4-3 shows that accelerated offers

have a significantly shorter underwriting period. On average, bought deals and accelerated

bookbuilt SEOs are offered within one day of the filing. In contrast, fully marketed offers spend

31 days on average to complete after the filing, although there is considerable dispersion.16 A

longer underwriting process requires more resources and input from both the issuer and the lead

manager, which also explains why fully marketed offers are the most expensive. During our

sample period, there is a monotonic downtrend in the median time between the announcement

and completed fully marketed deals. In 1996, the median time is 31 days and it decreases to 9

days in 2007.







15 We dropped 10 accelerated SEOs that spend more than 3 days from the filing to the offer. Inclusion of these 10
deals has little impact on our analysis results. Five bought deals spend more than 100 days from the filing to the
offer. For example, Rowan Companies filed the registration on Oct 13th, 1999 and completed the offer on Feb 16th
2000 so there are 126 days between the filing and the offer. This bought deal has one underwriter, Lehman
Brothers.
16 Note that the average is affected by several extreme cases. 14 fully marketed offers spend more than 180 days.
The longest case is the offer by Carmike Cinemas Inc in 2004. This offer is announced on Jun 7th, 2002 with an
expected pricing date of August 5th, 2002, revised on July 17th, 2002, revised on Nov 18th, 2003, revised on Dec 16th
2003, and revised on Jan 14th, 2004, and eventually took place on Jan 29th, 2004. So there are 601 days between the
announcement and the offer.










Table 4-1. Number of SEOs and Aggregate Proceeds by Year and Offer Method.

The sample includes 3,615 seasoned equity offerings in Dealogic's ECM Analytics Database during 1996 to 2007. The issuer must be a US-based company
listed on NASDAQ, the American Stock Exchange (AMEX), or the New York Stock Exchange (NYSE). ADRs and ADSs, private placements, rights
offers, best efforts, non SEC registered offers, Rule 144A offers, units, closed-end funds, REITs, and pure secondary offerings are excluded. Non-shelf
registered bought deals and accelerated bookbuilt offers are also excluded. The issuing firm must be present on the University of Chicago Center for
Research in Security Prices (CRSP) database on the first trading day after the issue. Total proceeds is the total amount of dollars (in billions) raised globally
including exercised overallotment shares in all trenches. Total proceeds are adjusted by the annual nominal S&P 500 Index level, normalized by the ratio
between the 1996 S&P 500 index level and the SEO year's S&P 500 index level.

Sample SEOs Bought Deals Accelerated Bookbuilt SEOs Fully Marketed SEOs


Total Proceeds Total Proceeds Total Proceeds Total Proceeds
Year Number Number Number Number
($ billion) ($ billion) ($ billion) ($ billion)

1996 487 45.80 1 0.06 0 0.00 486 45.74

1997 418 33.34 2 0.45 8 1.40 408 31.49

1998 285 25.91 8 1.40 2 0.03 275 24.44

1999 321 33.00 20 2.54 2 0.32 299 30.14

2000 323 49.60 29 3.21 2 0.07 292 46.32

2001 244 32.51 35 6.34 18 1.47 191 24.71

2002 229 38.06 25 5.48 41 10.36 163 22.22

2003 267 28.47 36 6.09 42 5.15 189 17.10

2004 303 28.52 36 4.52 51 6.58 216 17.35

2005 249 22.73 32 3.91 29 3.17 188 15.65

2006 255 19.91 43 3.73 38 6.58 174 12.40


27 1.99 46


2007 234 21.44


3.17 161 15.05









tend to have inelastic demand curves. Including these small firms biases us towards finding a

positive relationship between inelastic demand curves and conducting fully marketed offers. To

avoid this bias, we restrict our sample to companies with a $75 million market capitalization

prior to the offer. This reduces our sample size from 3,615 SEOs to 3,281 SEOs for our

empirical analysis.

Table 4-2 compares shelf-takedown and non-shelf takedown offers within each offer

method. Over the twelve-year sample period, there are 1,335 shelf-takedown SEOs. 766 of

those shelf-takedown SEOs are fully marketed. 293 shelf-takedowns are bought deals and 276

are accelerated bookbuilt offers. All of the 1,946 non-shelf takedown offers are fully marketed.

Shelf-registered issuers do not seem to have any dominating preference between

accelerated offers and fully marketed offers. Almost sixty percent of the shelf-takedown SEOs

are fully marketed offers. Autore, Hutton, and Kovacs (2008) also conclude that neither method

dominates among shelf issuing firms. Therefore, we do not consider the issuer's endogenous

decision on shelf registration as an explanatory variable in our empirical tests.

Offer Characteristics

Table 4-3 compares offer characteristics among the three offer methods. The first column

lists results for the entire sample of SEOs, while the second, third, and fourth columns compare

among bought deals, accelerated bookbuilt SEOs, and fully marketed SEOs. For each offer

characteristic variable, Table 4-3 reports mean and median values. The last column reports the

p-values from the Kruskal-Wallis (KW) test on the means and the Chi-squared test on the









Harris, Lawrence, and Eitan Gurel, 1986, "Price and Volume Effects Associated with Changes in
the S&P List: New Evidence for the Existence of Price Pressures,' Journal of Finance 41,
815-829.

Heinkel, Robert, and Eduardo S. Schwartz, 1986, "Rights versus Underwritten offerings: An
Asymmetric Information Approach," Journal ofFinance 41, 1-18.

Heron, Randall A., and Erik Lie, 2004, "A Comparison of the Motivations for and the
Information Content of Different Types of Equity Offerings," Journal of Business 77,
605-632.

Hodrick, Laurie S., 1999, "Does Stock Price Elasticity Affect Corporate Financial Decisions?"
Journal ofFinancial Economics 52, 225-256.

Huang, Rongbing, and Donghang, Zhang, 2008, "Managing Underwriters and the Marketing of
Seasoned Equity Offerings," University of South Carolina working paper.

Kalay, Avner, Orly Sade, and Avi Wohl, 2004, "Measuring Stock Illiquidity: An Investigation of
the Demand and Supply Schedules at the TASE," Journal ofFinancial Economics 74,
461 486.

Kandel, Shmuel, Oded Sarig, and Avi Wohl, 1999, "The Demand for Stocks: An Analysis of
IPO Auctions," Review ofFinancial Studies 12, 227-247.

Kaul, Aditya, Vikas Mehrotra, and Randall Morck, 2000, "Demand Curves for Stocks Do Slope
Down: New Evidence from an Index Weights Adjustment," Journal ofFinance 55, 893-
912.

Keim, Donald, and Ananth Madhavan, 1996, "The Upstairs Market for Large-Block
Transactions: Analysis and Measurement of Price Effects," Review ofFinancial Studies
9, 1-36.

Korajczyk, Robert A., Deborah J. Lucas, and Robert L. McDonald, 1991, "The Effect of
Information Releases on the Pricing and Timing of Equity Issues," Review ofFinancial
Studies 4, 685-708.

Lee, Inmoo, 1997, "Do Firms Knowingly Sell Overvalued Equity?" Journal ofFinance 52,
1439-1466.

Liu, Shinhua, 2000, "Changes in the Nikkei 500: New Evidence for Downward Sloping Demand
Curves for Stocks," International Review ofFinance 1:4, 2000, 245-267.

Lucas, Deborah J., and Robert L. McDonald, 1990, "Equity Issues and Stock Price Dynamics,"
Journal ofFinance 45, 1019-1043.









Studies on initial public offerings (IPO) also support downward-sloping demand curves.

Braun and Larrain (2008) find that IPOs in emerging markets have negative impacts on the

prices of stocks that highly covary with the IPOs, particularly for large IPOs and in less

internationally integrated markets. They conclude that shocks to asset supply have a significant

impact on asset prices.

Field and Hanka (2001) examine returns around IPO lockup expirations. Because the date

of the lockup expiration is known well in advance, in an efficient market, there should be no

abnormal returns. However, they find a statistically significant three-day abnormal return of-

1.5. Many VC-backed funds distribute shares to limited partners at the lockup expiration date,

and many of the limited partners immediately sell their shares. VC backing, therefore, may be a

proxy for the magnitude of the increase in the flow supply at the time of the lockup expiration. If

this is the case, then VC-backed IPOs should have larger price effects if demand curves are

negatively sloped. Consistent with this prediction, Bradley, Jordan, Roten, and Yi (2001) and

Brav and Gompers (2003) find a larger price drop at the lockup expiration for VC- backed IPOs.

In the market microstructure literature, the depth of the limit order book determines the

short-run elasticity of demand. Larger sell orders get executed at progressively lower prices

because investors assume that larger sell orders are more likely to be motivated by negative

information (O'Hara, 1995). Trades that are not motivated by private information, however, face

a larger price impact the larger the trade. In equilibrium, the short-run elasticity reflects the

market's expectations of the mix of informed and uninformed trades.

To summarize, there is widespread evidence that both the short-run and long-run demand

curves for stocks are negatively sloped and. Only Bagwell (1992) and Hodrick (1999), however,

link the stock demand elasticity to corporate financial decisions.











Table 4-3. (continued)


Bought Deals


Accelerated
Bookbuilt SEOs


Fully Marketed
SEOs


P-values from
KW/Median
Statistics


Number of Bookrunners


Bookrunner Reputation


Number of Days from Filing to Offer


Gross Spread (%)


Pre-Offer Return (o%)


Announcement Effect (%)


Discount (%)


Underpricing (%)


Number of Analysts


Bid-ask Spread (%)


1.19
S1

8.25
S 9

26
S 21

4.82
5.00

107.70
57.45

-1.72
-1.53

-2.76
-1.93

3.12
1.51


1.06
] [ 1

8.36
] [ 9

0
] [ 0

2.28
] [ 1.70

72.34
] [ 26.45

-1.49
] [ -1.48

-3.96
] [ -3.47

1.09
] [ 0.24


1.49
] [ 1

7.98
] [ 9

1
] [ 1

4.23
] [ 4.20

45.21
] [ 20.75

-2.55
] [ -1.99

-2.43
] [ -0.90

2.10
] [ 0.94


1.17
S[ 1

8.27
S[ 9

31
S[ 25

5.10
] [ 5.15

117.89
] [ 65.66

-1.66
] [ -1.48

-2.66
] [ -1.89

3.44
] [ 1.87


0.00
] [ 0.00

0.08
] [ 0.56

0.00
] [ 0.00

0.00
S[ 0.00

0.00
S[ 0.00

0.00
S[ 0.00

0.00
S[ 0.00

0.00
S[ 0.00


5 9 7 4 0.00
[ 4 i [ 7 i [ 6 i [ 4 [ 0.00


1.39
[ 1.00


0.68
] [ 0.35


0.59
] [ 0.40


1.55
] [ 1.26


0.00
] [ 0.00


All SEOs









CHAPTER 3
A MODEL OF THE MARKETING OF SEOS

The Model

In this section, we develop a model to study the economics of the underwriter's marketing

effort.

Our model starts with a downward-sloping demand curve. A downward-sloping demand

curve can arise from asymmetric information, differences of opinions among investors as to the

value of the firm, and short-sale constraints. In Wurgler and Zhuravskaya's (2002) model, the

demand curve's slope is more negative when (i) the stock does not have close substitutes (more

arbitrage risk), (ii) the risk aversion of arbitrageurs is high, (iii) the heterogeneity of non-

arbitrageurs' beliefs is high, and (iv) the number of arbitrageurs is small. Here, we focus on the

issuer's short-term demand curve around the seasoned offering.

Figure 3-1 illustrates the model. Suppose that there is a group of N investors with

homogeneous beliefs about the issuing firm's fundamental value. Each investor has a downward

sloping demand curve X(p) Then the aggregate short-run demand curve, the solid line in Figure

3-1, is NX(p). Before the offer, the market equilibrium price is P1, which is jointly determined

by the aggregate demand curve and the supply of shares. P1 also represents the price that the

stock will converge to in the long-run as the demand curve rotates from the steep short-run

relation to the more elastic long-run relation. The offering represents an increase in supply (from

X1 toX2 in Figure 3-1). If there is no marketing, the short-run demand curve remains the same

throughout the offering and as a result of the movement along the demand curve, the stock price

drops from P1 to P2.

If the issuing firm decides to market the offering, the marketing efforts will attract new

investors into the issue, as modeled in a product market context by Dixit and Norman (1978) and









JPMorgan, announced on May 10th, 2007 and offered within one day, on May 11th. The filing

procedure was similar to the first bought deal. Consolidated Edison did not disclose the offer

price until late in the afternoon on Monday, May 14th. Unfortunately, we cannot find any news

report on either deal.

Nasdaq Stock Market Inc conducted two fully marketed SEOs in 2006, raising $639

million and $691 million respectively. The first SEO was announced on Jan 30th, 2006 and

offered 10 days later, on Feb 9th. Of the 13 million shares offered, almost half were sold by

existing shareholders. The second SEO was announced on April 21st, 2006 and offered 7 days

later, on April 27th.










return on the last trading day before the offer, the discount, and the underpricing.19 Fully

marketed offers experience an average discount of -2.66% and the largest underpricing of 3.44%.

As we explained in Section 3.3, we do not attempt to empirically estimate the gain from

marketing. Neither the price discount nor the underpricing is a good measure of the gain from

marketing. Bought deals and accelerated bookbuilt offers may be unknown to the market in

advance. So the observed price discount may incorporate both the discount and an

announcement effect.20

The last two rows in Table 4-3 examine the number of analysts who are following the

issuer's stock and the stock's average bid-ask spread. An analyst is included if he or she posts at

least one recommendation within 12 months prior to the offer. The analyst recommendations are

obtained from the Institutional Brokers' Estimate System (IBES) database. During the year

before each SEO, the average number of analysts that follow an SEO is 5, which is comparable

to that reported in Huang and Zhang (2008).21 Accelerated issuers, who tend to be larger

companies, receive significantly more analyst coverage, with an average of 7 to 9 analysts

following the stock while fully marketed issuers only have 4 analysts on average. If an issuing

firm receives little analyst coverage prior to the offer, it may decide to do a road show to reach

more investors and promote the stock.


19 244 of the 293 bought deals and 137 of the 276 accelerated offers are offered within 24 hours after the
announcement. It is interesting to notice that on average, bought deals experience a positive return of 1.6% on the
previous day while accelerated bookbuilt offers experience a negative return of -2.6%. One potential explanation is
that issuers may wait for a 'good' day, i.e. with positive stock return, to conduct a bought deal. For accelerated
bookbuilt offers, there may be some information leakage prior to the announcement of the offer.
20 In unreported results, we first estimate the announcement effect with the offer and firm characteristics for an
accelerated offer if the accelerated offer is announced and offered within 24 hours. Then we subtract the predicted
announcement effect from the realized price discount. After this adjustment, bought deals still have the largest
discount of -3.93%. The R-squared is around 0.04.

21 Huang and Zhang (2008) look at the number of recommendations issued prior to an SEO. We find comparable
numbers for SEOs in our sample, e.g. on average, there are 7 recommendations issued within 12 months prior to an
SEO.









investors are net buyers of attention-grabbing stocks, described as those that are in the news, or

have high trading volume, have extreme one-day returns. In addition, Yuan (2008) finds that

high attention also influences both the aggregate market price level and the trading behavior of

institutional investors.

To formalize the impact of marketing on prices, we follow the model in Dixit and Norman

(1978). In their model marketing changes consumer tastes and flattens aggregate demand

curves. In Shapiro's (1980) model, advertising flattens each individual's demand curve.1 But

our focus is different from theirs. These economic theories focus on the equilibrium level of

advertising while we focus on the decision of whether or not to market the offer in the context of

the demand for the shares of a company that is issuing stock. One could interpret our analysis as

modeling the discrete decision as to advertise or not advertise.

We also make a distinction between short-run and long-run demand curves. We assume

that in the very long run, the price of the stock is determined by its cash flow fundamentals and

the discount rate that investors apply, which in turn may be affected by liquidity characteristics.

The long-run stock demand curve is essentially perfectly elastic. In the short-run, however, this

is not the case. The short-run demand curve is assumed to be negatively sloped. We assume that

the short-run demand elasticity is determined by how many investors are paying attention to the

stock, and how heterogeneous are their expectations.

The underwriter's marketing effort helps to flatten the short-run demand curve, which is

reflected in a higher stock price after the issue. This marketing effort does not come for free,




1 Comanor and Wilson (1967) argue that the demand curve may become more inelastic with advertising. In their
analysis, advertising creates product differentiation and market power, both of which increase entry barriers. Their
framework is less relevant to our study, since investors own stocks primarily for the cash flows that are generated
(i.e. in the language of bidding theory, stocks are common value goods rather than private value goods.)









both temporary and permanent return components, which supports the notion of different short-

run and long-run elasticities. Wurgler and Zhuravskaya (2002) document that stocks without

close substitutes experience a larger price jump when added to the S&P 500 Index, implying less

elastic short-run demand curves for these stocks. Morck and Yang (2002) find that the valuation

premium associated with being in the S&P 500 Index has increased substantially over time and

grows in step with the growth of indexing. They argue that indexing demand for these stocks

results in a higher price because the long-run supply curve is positively sloped. A positively

sloped supply curve for buyers is analogous to a negatively sloped demand curve for sellers.

Similar results are reported for the stocks that are added to or deleted from other indices.

Madhavan (2003) and Biktimirov, Cowan, and Jordan (2004) report that stocks added to or

deleted from the Russell 2000 in its annual reconstitution experience significant returns. Liu

(2000) finds positive price changes when stocks are added to the Nikkei 500 Index with no

significant post-event reversals. In addition, Kaul, Mehrotra, and Morck (2000) examine stock

price changes in response to the re-weighting of the Toronto Stock Exchange 300 Index in 1996,

finding strong support for downward sloping demand curves for common stocks.

Greenwood (2005) studies the redefinition of the Nikkei 225 index in April 2000. He finds

that not only do additions and deletions experience significant price changes, but securities that

are not directly affected by the substitution of securities also experience abnormal returns due to

their correlation with securities undergoing changes in demand. Furthermore, he shows that the

short-run effects are partly reversed over the following 20 weeks, implying a more elastic long-

run demand curve than in the short run.

Although the studies examining index changes universally find price effects, they rarely

estimate the elasticity because the change in quantity demanded by index funds and other









CHAPTER 5
PRICE ELASTICITY PROXIES AND DETERMINANTS OF OFFER METHOD

Price Elasticity Proxy Estimate

The definition of elasticity of demand is the percentage change in quantity demanded

iAqj
( q

corresponding to a percentage change in price, formally, (/ In general, we cannot

precisely construct the demand and supply schedules for individual stocks and directly measure

the price elasticity. Due to data limitations that prevent us from identifying supply shifts along a

given demand curve or vice versa, we adopt two measures to proxy for the elasticity of demand

for the issuing firm's stock before the offer.

The first measure of price elasticity is an average daily order flow inverse price elasticity

measure, A1. The daily order flow inverse price elasticity on day T is defined as the ratio

between the absolute value of the stock's raw return and its turnover, with turnover defined as

the trading volume divided by the number of shares outstanding. This is called an inverse

elasticity because for an elasticity, quantity is in the numerator rather than the denominator.1 If

the stock is listed on NASDAQ, we divide the trading volume in half to control for the double

counting on NASDAQ relative to the AMEX and NYSE. A1 is the average daily inverse

elasticity over the 250 trading day window, [-250,-1], prior to the announcement date:2


1 250 Stock Raw Return,
250 ~ Number of Shares Traded,
/Number of Shares Outstanding,



1 We put the stock return in the numerator to avoid zeroes in the denominator.
2 If the company went public within 250 trading days prior to the SEO, we start with the first date available on
CRSP.









0. The reference group is fully marketed offers. The binomial logistic model is estimated as

follows:

1
Pr(Method=Accelerated Deal) =

where u = y0 oi xElasticity Proxy+y, xLn(Market Cap)+ y3 xRelative Size
+y4 xPrimary Fraction+y, xLn(1+Number of Analysts)
+76 xLn (Bid-ask Spread) +

Table 5-3 presents the binomial logistic regression results. The results are very similarly if

we include only one elasticity proxy, Ln(A1), order flow inverse demand elasticity, or Ln(A2),

residual variance. As predicted, the coefficients on the two demand elasticity measures are

negative, and they are statistically significant at the 1% level. Larger order flow inverse

elasticity or arbitrage risk indicates a more inelastic demand curve, and this inelasticity

encourages issuers to choose a fully marketed offer instead of an accelerated offer.

The relative offer size has a significantly negative impact on the probability of an

accelerated offer. As our model predicts, ceteris paribus, a large relative offer size increases the

gain from marketing so the issuing firm is more likely to choose a fully marketed offer. The

likelihood-based pseudo R-square is 28%, which indicate that the model fits well and can explain

at least 28% of the total variance. The model correctly predicts almost 90% of the sample offer

method. It should be noted, however, that 86% of the offers are fully marketed.

The binomial logistic model calculates the log ratio between probabilities from the

reference group to the alternative group, so the coefficients are hard to interpret. In Table 5-3,

we report two marginal probabilities, calculated at the sample means and medians, associated

with each independent variable. The marginal probability at means is the average change in the

probability given a one unit change in the independent variable at the sample mean level. For

example, holding all the independent variables at their mean values, a one unit increase in Ln(A1)









ACKNOWLEDGMENTS

I would like to sincerely thank Jay Ritter (my supervisory committee chair) for his

guidance, encouragement, and help. I would also like to thank Jonathan H. Hamilton, Joel

Houston, and Mahen Nimalendran, for their invaluable advice. I would like to thank seminar

participants at the Helsinki School of Economics, Norwegian School of Management (BI),

Singapore Management University, University of Florida, and University of Hong Kong for

helpful comments. Finally, I greatly appreciate all the support that I received from my current

and former colleagues, Tom Barkley, Alice Bonaime, and Jennifer Itzkowitz.









APPENDIX A
CLASSIFICATION OF SEO OFFER METHOD

Our main database is the Dealogic Equity Capital Markets (ECM) database. Dealogic

identifies three major SEO offer methods: accelerated bookbuilt, bought deal, and fully marketed

offers. The Thomson Financial Securities Data Company's (SDC) new issues database is more

commonly used in academic studies. As pointed out in Bortolotti, Megginson, and Smart (2007),

SDC's method of the classifying offering technique is sometimes confusing because it gives

multiple designations to the same offer. For example, some offers are classified as "block

trade/negotiated sale", "accelerated bookbuilt/firm commitment", "firm commitment/auction".

Dealogic gives a single designation to each offer, so we think its classification is less ambiguous.

Compared to SDC, Dealogic is more accurate with the filing date and is more consistent

with its classification of the offer method. We investigate 519 US seasoned equity offerings

during Jan 1st, 2004 to Dec 31st, 2005 listed by both Dealogic and SDC. We first hand-checked

35 random offers' filing dates in Dealogic with Factiva and all of them are correct. Dealogic's

classification of the offer method is mostly consistent with the time length from filing to

offering. Accelerated bookbuilt offers and bought deals are almost always completed within 3

calendar days from filing with the SEC. Fully marketed offers take a longer time, ranging from

3 calendar days to more than 150 calendar days. All 68 accelerated bookbuilt offers are

completed within 3 calendar days from filing with the SEC. Among the 119 bought deals, only 2

offers have a time span longer than 3 calendar days. One is the offer by Trinity Industries Inc,

filed on Dec 1st, 2004, which began trading on Dec 9th, 2004. The Dealogic filing date is

consistent with what we find in Factiva. This is a pure secondary offer so it is not included in

our study. The other is by Monster Worldwide Inc, offered on Jan 5th, 2004. Monster

Worldwide filed several S-3/As during the three months prior to the offer and the last one is filed









SEOs and Marketing

The vast majority of academic studies on the announcement effects associated with SEOs

have focused on information releases either explicitly or implicitly. One of the few exceptions is

Huang and Zhang (2008), who posit that the use of multiple managing underwriters is motivated

by marketing considerations. Huang and Zhang (2008) report that if one more managing

underwriter is hired, the offer price discount is reduced by 0.26%. They argue that the smaller

price discount results from the additional managing underwriter's marketing efforts, which

reduce the price pressure and the downside price risk. Huang and Zhang (2008) focus on the

gross spread and offer price discount to examine the impact of hiring additional mangers on

direct and indirect issue costs. Our study has a different focus. We examine the issuing firm's

pre-issue demand elasticity and analyze the determinants of the offer method.



















Supply




Demand




With Marketing


X, X2


Quantity


Demand Curve without Marketing

Demand Curve with Marketing -----------------


Figure 3-1. Comparison of Demand Curves with and without Marketing.

Note: The solid line is the issuing firm's demand curve before the offer. The dashed line
represents the issuing firm's demand curve if the offer is fully marketed. It becomes more elastic
than the ex ante demand curve because marketing flattens the demand curve. Therefore, the
post-issue price is higher at instead of if there is no marketing. The shaded rectangular area
represents the issuer's gross gain from marketing the offer.









The marketing effort can change the elasticity and helps the issuer to achieve a higher offer

price and post-issue market price. The benefits of marketing are larger if the demand curve is

more inelastic or if the relative offer size is higher. Therefore, our model predicts that an issuing

firm is more likely to pay the higher fees and choose the fully marketed offer method if it faces

an inelastic demand curve prior to the offer. The fully marketed offer method is also preferred if

the relative offer size is large.

Our empirical results support the model's predictions. We use two measures to proxy for

the demand elasticity. The first measure is an order flow inverse demand elasticity, the ratio of

the average daily absolute returns to the average daily turnover. The second measure is the

arbitrage risk measure used in Wurgler and Zhuravskaya (2002), the variance of market model

residual returns. We document that the pre-issue elasticity of the issuing firm's demand curve is

an important determinant of the offer method. The relative offer size is also an important

determinant. For example, of the 211 SEOs conducted in 2007, 91% of firms issuing more than

15% of their equity (72 out of 79) use fully marketed offers, whereas only 27% of firms issuing

less than 10% (13 out of 49) do so. We also find that a smaller fraction of primary shares and

less analyst coverage all increase the probability of using a fully marketed offer, which indicates

that firms with high information asymmetry favor fully marketed offers.

Our model is not mutually exclusive from previous studies on information asymmetry and

information production. The downward-sloping demand curve can arise from asymmetric

information, differences of opinion among investors (Miller 1977), or how many investors are

paying attention to the stock. If the issuing firm faces high information asymmetry, it has a

relatively inelastic demand curve and it can benefit from fully marketing the offer. During the

underwriting process, the book-runner produces information about the market demand for the









APPENDIX B
EXAMPLES OF ACCELERATED AND FULLY MARKETED SEOS

Google announced its first SEO on August 15th, 2005 and conducted the offer a month

later, on September 14th. This is a fully marketed offer. Google offered 14,159,265 shares at

$295 per share, raising $4.3 billion. Morgan Stanley and CSFB are the bookrunners. On March

29th, 2006, Google announced its second SEO and conducted the offer through accelerated

bookbuilding two days later, on March 31st. It offered 5.3 million shares at $389.75, raising

42.1 billion. Goldman Sachs is the sole bookrunner.

Sallie Mae announced its plan to raise $2.5 billion through an accelerated bookbuilt SEO

after the stock market closes on Dec 26th, 2007. It had an urgent need for capital because it was

facing a potential $1 billion loss on its equity forward contracts and a possible downgrade on its

debt outstanding by Moody's. The equity forward contracts could force Sallie Mae to purchase

44,039,890 shares at $45 per share while the stock price was around $20. Citibank could

demand immediate payment because the stock price had fallen well below the repurchase price.

After negotiation, Citibank allowed Sallie Mae to postpone the repurchase for two months and,

together with UBS, underwrote the SEO. Sallie Mae completed the offer on Dec 27th, 2007, less

than 24 hours after the offer announcement. It raised $2 billion by selling 101,781,170 common

shares and another $1 billion by selling preferred stock.

Consolidated Edison conducted two bought deals during 2006 to 2007, raising $447

million and $559 million respectively. The first one was solely underwritten by Citibank and

was offered on Sept 21st, 2006. Consolidated Edison first filed a preliminary prospectus with

the SEC on Sept 20th, 2006. Early in the morning on Sept 21st, it filed a prospectus supplement

without specifying the offer price. The offer price was disclosed in a following supplement filed

late in the afternoon on Sept 22nd, 2006. The second bought deal was underwritten by









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

THE MARKETING OF SEASONED EQUITY OFFERINGS

By

Xiaohui Gao

August 2008

Chair: Jay R. Ritter
Major: Business Administration

Accelerated seasoned equity offerings (SEOs), which include bought deals and accelerated

bookbuilt offers, have increased dramatically in the U.S. and globally recently. Accelerated

offers are cheaper than traditional fully marketed offers in terms of direct issue costs. To explain

why some issuing firms choose a fully marketed offer instead of an accelerated offer, we develop

a model in which marketing flattens the issuer's demand curve. Empirical analysis shows that

the pre-issue elasticity of the issuing firm's demand curve and the relative offer size are

important determinants of the offer method. In our analysis, the elasticity of demand at the time

of issuance is endogenous.










Table 5-3. Determinants of Offering Methods for SEOs, 1996 to 2007


Table 5-3 presents the results from the binomial logistic regressions for the full sample period and the sub-period of 2001 to 2007. The dependent
variable, offering method, is a dichotomous variable for which accelerated deals, including bought deals and accelerated bookbuilt offers, equal 1, and
fully marketed offers equal 0. All the independent variables are defined the same as in Tables III and V. The marginal effects, in percentages, are
calculated at the sample means and medians, and are reported to the right of the estimates. The chi-squared statistics are reported in parentheses below. *
**, and *** indicate significance at the 10, 5, and 1 percent levels, respectively. The R-square is the likelihood-based pseudo r-square measure.

Full Period: 1996 To 2007 Sub-period: 2001 To 2007
Dependent Variable Dependent Variable

Pr (Accelerated Deal = 1) Pr (Accelerated Deal = 1)

Estimate Marginal Effect Marginal Effect Estimate Marginal Effect Marginal Effect
at Means(%) at Medians (%) at Means (%) at Medians (%)
Ln(A1) -0.70 *** -2.56 -8.55 -0.51 *** -6.14 -11.33
( 80.27 ) ( 29.40 )
Ln(A2) -0.22 *** -0.79 -2.65 -0.03 -0.34 -0.63
( 15.99 ) ( 0.17 )
Ln(MV) -0.42 *** -1.51 -5.06 -0.34 *** -4.05 -7.46
( 36.74 ) ( 16.60 )
Relative size (%) -0.14 *** -0.53 -1.76 -0.15 *** -1.75 -3.23
( 215.73 ) ( 164.60 )
Fraction of Primary Shares (%) 0.05 *** 0.18 0.59 0.04 *** 0.45 0.84
( 50.32 ) ( 33.69 )
Ln(1 + Analysts) 0.59 *** 2.14 7.16 0.46 *** 5.47 10.09
( 30.54 ) ( 13.06 )
Detrended Ln(Bid-ask Spread) 0.38 *** 1.39 4.64 0.18 2.10 3.87
( 13.77 ) ( 2.02 )
Number of SEOs 3,275 1,655

R-square 0.28 0.32









Accelerated offers include bought deals and accelerated bookbuilt offers. They are

usually shelf registered offers.3 These accelerated deals, especially accelerated bookbuilt offers,

have become common during the last decade. Bortolotti, Megginson, and Smart (2007) provide

an excellent review of the history and development of accelerated offers.4 Bought deals started

in the mid-1980s. With a bought deal, the issuing firm announces the amount of stock it wishes

to sell and investment banks bid for these shares, usually by submitting bids shortly after the

market's close. The bank that offers the highest net price wins the deal. The winning bank then

re-sells the shares on the open market or to its investors, usually within 24 hours. Because of this

timing, bought deals are also known as overnight deals. Bought deals are essentially auctions to

underwriters followed by open market sales.

Accelerated bookbuilt offers emerged in the late 1990s.5 During the last decade,

accelerated offers have gained market share quickly both in the US and around the world. In

2004, more than a third of issues in the rest of the world were accelerated SEOs, according to

Table 2 in Bortolotti, Megginson, and Smart (2007). Unlike bought deals, in accelerated

bookbuilt offers, banks do not initially purchase the whole issue from the issuing company.

They submit proposals, usually specifying a gross spread but not necessarily an offer price, for

the right to underwrite the sale of the shares. The winning bank then usually forms a small

underwriting syndicate and begins marketing the shares to institutional investors. The offer price

is then negotiated between the issuing firm and the bank. The bookbuilding procedure is


3 Some previous studies exclude shelf registered offers (Altinkilic and Hansen, 2003, 2006) and some pool shelf and
non-shelf offers together (Corwin, 2003).
4 Bortolotti, Megginson, and Smart's (2007) terminology of bought deals is slightly different from ours. They refer
to bought deals in Canada as bought deals and bought deals in the US as block trades. We define US block trades as
bought deals.
5 Dealogic's ECM Analytics Database identifies the first accelerated bookbuilt offer in 1997, which is consistent
with the first usage of the term 'accelerated book-building' in July 1997 (Warn (1997)).









The two predictions yield two testable hypotheses:

* The issuing firm's ex ante demand elasticity is an important determinant of the offer
method.

* The relative offer size is an important determinant of the offer method.

We want to point out that our empirical study does not attempt to estimate the dollar value

of the gain from marketing because it is difficult to find the post-issue price with marketing, P*,

and the post-issue market price without marketing, P2. For a fully marketed offer, the ex ante

demand curve is unobservable so it is hard for us to estimate the counterfactual price P2 had

there been no marketing. To find P* is also a challenge. On the day of the SEO announcement,

market efficiency suggests that investors should anticipate any price pressure effects associated

with the increase in supply and the market price should adjust accordingly at the announcement

rather than waiting for the issue day. At the same time, the market price should also incorporate

the information released by the announcement. Therefore, the observed closing price on the

announcement day is the combination of the information effect and the anticipated price pressure

effect. To find P* requires isolating the information effect from the price pressure effect, which

is hard to implement empirically. Furthermore, the observed announcement day closing price

may still deviate from the expected post-issue price due to transaction costs and arbitrage

limitations, as well as uncertainty about whether the deal will be completed.

In our model, the demand curve should become more elastic after the offer due to the

marketing effort. However, to empirically compare the pre-issue and post-issue demand

elasticities is difficult. Our estimated proxies for elasticity correlate with other firm and market

characteristics, which cannot be held constant after the offer. In our empirical analysis, we find

that fully marketed offers experience the largest increase in demand elasticity. But an alternative

explanation is that a larger market capitalization after the offer may has improved the liquidity.









Accelerated bookbuilt SEOs tend to be chosen by larger firms and bought deals are chosen

by the largest firms. Compared to an average issuer of a fully marketed offer, the average

accelerated bookbuilt issuer is three times larger and the average bought deal issuer is almost

four times larger.

The offer proceeds, however, show only modest differences across offer methods. As a

result, the relative offer size for fully marketed deals is on average much larger. Figure 4-1 is a

scatter diagram of the relation between the issuing firm's market capitalization prior to the offer

and the relative offer size across offer methods for SEOs in 2007.14 Among the 211 SEOs in

2007, 91% of firms issuing more than 15% of their equity (72 out of 79) use fully marketed

offers, whereas only 27% of firms issuing less than 10% (13 out of 49) do so.

Figure 4-1 shows that firms choosing fully marketed offers tend to be small firms selling a

relatively large amount of shares. The larger relative offer size for fully marketed deals is

consistent with our model prediction: Firms that may suffer from a larger price decline due to a

larger move along the pre-issue demand curve prefer to fully market the offer because the

marketing effort flattens the demand curve and helps to achieve a higher offer price. The smaller

market capitalization for fully marketed deals is also consistent with our model's prediction:

Small firms tend to have less elastic demand curves because of lower institutional ownership,

and thus receive greater benefits from marketing.

Accelerated bookbuilt offers tend to hire slightly more bookrunners. In our sample,

however, almost 90% of all SEOs have a single bookrunner and the higher incidence of multiple

bookrunners for accelerated bookbuilt SEOs may partly reflect the higher frequency of these

deals in the later half of our sample period, when multiple bookrunners were more common. We

14 Other years show the same pattern. Only one year's data is used to avoid obscuring the pattern by plotting too
many points.









apply a natural log transformation on all of these variables, which have skewed distributions.5

To control for the strong declining trend in bid-ask spread, we detrend the log transformed bid-

ask spread by subtracting the mean value of this variable from SEOs in the same calendar year,

(bid-ask spread (bid-ask spread)
Detrended Ln bid-ask spread = Ln bid-ask spread year t average value
S price \ price The two offer

characteristics variables are offer size and fraction of the offer that is primary shares.

Table 5-2 reports the pairwise Pearson correlation coefficients among the two demand

elasticity proxies, the issuing firm's Ln(market capitalization) prior to the announcement,

Ln(total proceeds), the relative offer size, the fraction of primary shares, the detrended Ln(bid-

ask spread), and Ln(l+the number of analysts who cover the stock prior to the SEO). The two

elasticity proxies are moderately correlated (0.23) with each other. Both proxies are strongly

correlated with the issuing firm's market capitalization, the bid-ask spread, and the number of

analysts, indicating that small firms and firms with relatively high information asymmetry have

relatively inelastic demand curves. The relative offer size is highly negatively correlated (-0.47)

with firm size because of the tendency of large firms to not raise enormous amounts of money.

We do not include the offer proceeds in our regression because the relative offer size is the ratio

of the offer proceeds and the market capitalization.6

We use a binomial logistic model to investigate the determinants of the SEO offer method.

The dependent variable, offer method, is a dichotomous variable for which accelerated offers,

including bought deals and accelerated bookbuilt offers, equal 1, and fully marketed offers equal

5 If we use the raw values of A1, the regression coefficient is still negative but becomes insignificant. In unreported
results, we winsorized A,at the 1% and 99% level or at the 2% and 98% level, and our empirical results remain
qualitatively the same as with the log transformation. If we use the raw values of A, or the raw values of the
number of analyst recommendations, the results remain qualitatively the same.
6 Our results are qualitatively unchanged if we include the proceeds.










7) Pure secondary offerings are excluded.5

8) The issuing firm must be present on the University of Chicago Center for

Research in Security Prices (CRSP) database on the last trading day prior to the offer and the

first trading day after the issue.

9) Accelerated offers that are identified as non-shelf offers are excluded.6

Pure secondary SEOs are offerings in which all of the shares are being sold by existing

shareholders. We exclude pure secondary offerings because they are similar to large sales (block

trades) in the open market. Our empirical results are robust to including these pure secondary

SEOs, and tables are available upon request.

Offer Methods

Our sample includes 3,615 US SEOs during January 1st, 1996 to December 31st, 2007.7

In Appendix B, we list a few recent large accelerated and fully marketed sample SEOs with

some offer details. Table 4-1 reports the number of offers with each offer method by year. The

number of SEOs fluctuates over the twelve year sample period, although the fluctuation in

volume from year to year is much smaller than for IPOs. The number of bought deals and

accelerated bookbuilt offers has increased substantially since 2000. In 1996, there was only 1

bought deal and there were no accelerated bookbuilt offers. In 2007, there were 27 bought deals



5 846 pure secondary offerings, in which all of the shares are being sold by existing shareholders, are excluded.
Among the 846 pure secondary offerings, 262 are bought deals, 46 are accelerated bookbuilt offers, and 538 are
fully marketed offers.
6 Generally speaking, only companies that are eligible for shelf registrations are eligible for accelerated SEO offers.
There are exceptions and possible data mistakes. For example, the offer by First Republic Bank on August 15th
2005, is exempted from registration under Section 3(a)(2) of the Securities Act of 1933. MBNA filed a shelf
registration on March 29th, 1999 and conduct a bought deal on August 14th, 2000. But Dealogic reports this as a
non-shelf takedown. Most of these deals are small offers with less than $100 million in proceeds so we exclude all
35 non-shelf accelerated offers.

7 Our sample is smaller than Bortolotti, Megginson, and Smart's (2007) because we exclude ADRs, private
placements, and pure secondary offerings.










Table 5-2. Correlation among Demand Elasticity Proxies and Firm and Offer Characteristics
Table 5-2 reports pairwise Pearson correlation coefficients (p-values in parentheses) among the two demand elasticity measures, Ln(A1) and Ln(A2), and other
firm and offer characteristics. We apply log transformation on the market cap, proceeds, bid-ask spread, and number of analysts to control for extreme values
in these variables. We detrend the log bid-ask spreads by subtracting the sample average log bid-ask spread within the same calendar year.


Ln(A1)


Ln(A1)



Ln(A2)



Ln(MV)


Ln(Proceeds)



Relative Size



Primary Fraction


Ln(A2) Ln(MV) Ln(Proceeds Relative
) Size


-0.52


-0.39


0.26


Detrended
Primary Ln(Bid-ask
Fraction Spread)
Spread)


-0.14


-0.04


( 0.0 ) ( 0.0 ) ( 0.00 ) ( 0.0 ) ( 0.0 ) ( 0.04
0 0 0 0


-0.24


-0.15


0.09


-0.16


-0.03


( 0.0 ) ( 0.00 ) ( 0.0 ) ( 0.0 ) ( 0.05
0 0 0


0.77


-0.47


0.06


0.12


( 0.00 ) ( 0.0 ) ( 0.0 ) ( 0.00
0 0
1.00 0.01 -0.11 0.11
( 0.5 ) ( 0.0 ) ( 0.00
7 0
1.00 -0.15 -0.02
( 0.0 ) ( 0.20
0
1.00 0.07
( 0.00


Ln(1 +
Number
of
Analysts)
-0.04


) ( 0.00 )

-0.14
) ( 0.00 )

0.59
) ( 0.00

0.46
) ( 0.00

-0.28
) ( 0.00 )

0.07
) ( 0.00 )


Detrended Ln(Bid-ask Spread)


( 0.00 )









The bid-ask spread is the average daily bid-ask spread, scaled by the closing stock price on

that day, over the 250 trading days prior to the announcement of the SEO. Accelerated issuers

tend to have a smaller proportional bid-ask spread, an average of 0.64%, compared to fully

marketed offers with an average of 1.55%. However, the bid-ask spread exhibits a sharp

declining trend in the sample period, from an average of 2.69% in 1996 to an average of 0.31%

in 2007. This coincides with the increase in accelerated SEOs during our sample period and may

lead to spurious results between the bid-ask spread and the SEO method selection. Therefore,

we detrend the bid-ask spread in our multivariate regression in Section 5.

The univariate results show that accelerated SEOs tend to be large firms and the relative

offer sizes are smaller than fully marketed offers. Next, we proxy for the issuer's pre-offer

demand elasticity and examine the offer method choice in a multivariate setting in Section 5.









CHAPTER 2
RELATED LITERATURE

Marketing

Figure 2-1 illustrates and compares the SEO process associated with the three offer

methods: bought deals, accelerated bookbuilt, and fully marketed SEOs.

Aside from the speed of issue, the major difference between a fully marketed SEO and an

accelerated SEO is the marketing effort by the underwriter. In a fully marketed offer, a

preliminary prospectus is distributed and lead underwriters together with the issuing firm's

executives hold a traditional road show. They visit large institutional investors to create interest

and build the demand schedule of the offer. During the road shows, institutional investors

acquire more information regarding the issue. The road show process usually lasts for two

weeks, so investors have plenty of time to investigate and make their investment decisions.

In a bought deal, in contrast, the winning bank buys all the shares and re-sells them

immediately, usually overnight. In an accelerated bookbuilt SEO, the winning bank assembles at

most a small underwriting syndicate and usually sells the whole issue within 48 hours.

Therefore, there is little marketing effort involved in an accelerated underwriting and investors

have little time to evaluate the offer.

Marketing has several effects. Marketing increases the number of investors paying

attention to a given stock. Both theoretical and empirical research shows that attention plays an

important role in investor's trading behavior and the stock market price. In Merton (1987),

investors have a positive demand for only those stocks with which they are familiar. Busse and

Green (2002) report that trading volume for Nasdaq-listed stocks increased by an average of

300,000 shares in the four minutes after an analyst mentioned a stock favorably on CNBC's

Midday Call segment. Barber and Odean (2008) test and confirm the hypothesis that individual


































2008 Xiaohui Gao









Fully marketed offers pay the highest gross spreads, 5.10% on average. Bought deals pay

2.28% and accelerated bookbuilt offers pay 4.23% on average. These results are comparable to

those reported in previous studies, including Bortolotti, Megginson, and Smart (2007).

The difference in gross spread among the three offer methods is significant and important.

In a bought deal, the underwriter commits to purchase all of the shares for resale to the secondary

market. There is no book-building nor road show involved. The underwriter of a bought deal

faces greater uncertainty about the price at which the shares can be resold because in general the

market has not had a prior opportunity to react to the offer announcement. In an accelerated

bookbuilt offer, the lead manager collects price/quantity pairs from institutional investors and

underwriters, and then sets the price of the shares in agreement with the issuer using the order

book. In fully marketed offers, the lead manager conducts a road show while building the order

book so the bookbuilding process is longer. In accelerated bookbuilt offers and fully marketed

offers, the offer price is not set until after the market knows about the issue and has reset the

stock price, so the underwriter does not assume as much price risk as in bought deals. The risk is

smaller in fully marketed offers because the market has more time to value the deal and the

underwriter has more time to build the order book.

The evidence in Table 4-3 suggests that a higher gross spread is associated with

bookbuilding and expensive marketing efforts, even though the underwriter is exposed to more

resale price risk on bought deals. Unreported multivariate analysis confirms that after

controlling for various offer and firm characteristics, fully marketed offers pay an average 3%

higher gross spread than accelerated SEOs. Despite the lower gross spread, an accelerated SEO

can still be very lucrative for the underwriter. Take an average accelerated deal with $200









perfect substitutes, the demand curve is downward sloping because the arbitrage risk is nonzero

and arbitrageurs are risk averse. The larger the arbitrage risk, A2, the more inelastic the demand

curve is. Their empirical analysis documents a positive relation between arbitrage risk and

returns on the announcement day of S&P 500 additions, which suggests that stocks with greater

arbitrage risk have less elastic demand.

Our estimated order flow inverse demand elasticity and arbitrage risk measures are

qualitative proxies for the individual stock's demand elasticity. Both measures are negatively

related to the issuing firm's stock demand elasticity prior to the offer with a high value indicating

inelastic demand. We are interested in the cross-sectional difference in A1 and A2 among the

issuing firms.

Panel A and B in Table 5-1 reports means, medians, and standard deviations of the two

measures in the two panels. Our estimates of A2 are comparable to what Wurgler and

Zhuravskaya (2002) report. Comparing columns 2, 3, and 4 in Panel A and Panel B, we find the

same pattern for both measures. On average, fully marketed SEOs have inverse elasticity and

arbitrage risk at least twice as high as for bought deals and accelerated bookbuilt offers. This

shows that fully marketed offers are dominated by issuers with relatively inelastic pre-

announcement demand curves. The univariate results are consistent with our model's prediction:

if the demand curve is more inelastic, the issuing firm is more likely to choose a fully marketed

offer because the benefits of marketing are larger. Marketing can flatten the demand curve and

achieve a higher post-issue price.

The inverse demand elasticity, A1, has a highly skewed distribution, ranging from 0.03 to

1000. For A roughly 1% of the sample takes an extreme value because the annualized turnover

of the stock is less than 5%. For example, Centennial Communications Corp's average daily


















Market Capitalization and Relative Offer Size


Relative Offer Size (X)
70



60



50



40



30 -



20 -



10-



0


S. X


S U

C


Xx %x< x x:* x
X XX
YX X X
X X X x x ^


X
x
x
x


X.
X Xx
N


Log(Market Cap)

Fully Marketed Offers X X X Accelerated Offers


Figure 4-1. Scatter Diagram Relating Relative Offer Size to Log(Market Capitalization).


The sample consists of 211 SEOs in 2007. The S&P 500 Index level adjusted market capitalization ranges from $40 million, to $22
billion, The relative offer size is the number of shares offered divided by the number of shares outstanding prior to the offer and is

measured in percentages. The farest lower right dot represents Schering-Plough Corp's fully marketed SEO in August, 2007. It had

an actual market cap of $42 billion prior to the offer and a relative offer size of 3.86%.


X
0 XXX X X
X X
X





Ip I*


X ,


* A









TABLE OF CONTENTS

page

A CK N O W LED G M EN T S ................................................................. ........... ............. .....

L IST O F T A B L E S .......... .... ......................................................................................... 7

LIST OF FIGURES .................................. .. ..... ..... ................. .8

A B S T R A C T ......... ....................... .................. .......................... ................ .. 9

CHAPTER

1 INTRODUCTION ............... .............................. ............................ 10

2 R E L A TE D L ITE R A TU R E ........................................................................... ...................16

M marketing ...................................................................................... 16
D ow nw ard Sloping D em and Curves .......................................................................18
SEO s and M marketing ............... .................. ............ .................... ..... 22

3 A MODEL OF THE MARKETING OF SEOS .......................................... ............... 24

T h e M o d el ..................................... ................... ....................... ................2 4
M odel Predictions and Testable H ypotheses................................... .....................................25
Extension of the M odel ................................................... ............. ........... 27

4 SAMPLE AND DESCRIPTIVE STATISTICS .......................................... ............... 29

S a m p le S e le ctio n ....................................................................................................................2 9
O ffe r M eth o d s ................................................................................................................... 3 0
Shelf-takedow n Offers ..................................... ................ ............... .... ...... 32
Offer Characteristics....... ........ ............................... .. ... ..................33

5 PRICE ELASTICITY PROXIES AND DETERMINANTS OF OFFER METHOD............47

P rice E lasticity P roxy E stim ate ................................................................... .....................47
D term inants of O offer M ethod ........................................ ................................... ...............50
Does Speed Rather than Marketing Determine the Issue Method?................... ..............55

6 C O N C L U SIO N ......... .................................................................... ...................................... 6 1

APPENDIX

A CLASSIFICATION OF SEO OFFER METHOD .........................................................63

B EXAMPLES OF ACCELERATED AND FULLY MARKETED SEOS..............................65









LIST OF REFERENCES


Altinkilic, Oya, and Robert S. Hansen, 2003, "Discounting and Underpricing in Seasoned Equity
Offers," Journal of Financial Economics 69, 285-324.

Altinkilic, Oya, and Robert S. Hansen, 2006, "The Puzzling Wealth Loss around Equity
Offerings: Is it Merely Price Pressure or Is There More Information?" Working paper,
University of Pittsburgh.

Asquith, Paul, and David W. Mullins, 1986, "Equity Issues and Offering Dilution", Journal of
Financial Economics 15, 61 89.

Autore, Don, Raman Kumar, and Kilip Shome, 2008, "The Revival of Shelf-registered Corporate
Equity Offerings," Forthcoming, Journal of Corporate Finance.

Bagwell, Laurie S., 1992, "Dutch Auction Repurchases: An Analysis of Shareholder
Heterogeneity", Journal ofFinance 47, 71 105.

Baker, Malcolm, Joshua Coval, and Jeremy C. Stein, 2007, "Corporate Financing Decisions
When Investors Take the Path of Least Resistance," Journal ofFinancial Economics 84,
266 298.

Barber, Brad M., and Terrance Odean, 2008, "All that Glitters: The Effect of Attention and News
on the Buying Behavior of Individual and Institutional Investors," Review ofFinancial
Studies 21, 785 818.

Benveniste, Lawrence M., and Paul A. Spindt, 1989, "How Investment Bankers Determine Offer
Price and Allocation of New Issues", Journal ofFinancial Economics 24, 343 361.

Bethel, Jennifer, and Laurie Krigman, 2006, "Managing the Costs of Issuing Common Equity:
The Role of Registration Choice", Babson College working paper.

Biktimirov, N. Ernest, Arnold R. Cowan, and Bradford D. Jordan, 2004, "Do Demand Curves for
Small Stocks Slope Down?" Journal ofFinancial Research 27, 161 178.

Bohren, Oyvind, B. Espen Eckbo, and Dag Michalsen, 1997, "Why Underwrite Rights
Offerings? Some New Evidence," Journal ofFinancial Economics 46, 223-261.

Bortolotti, Bernardo, William L. Megginson, and Scott B. Smart, 2007, "The Rise of Accelerated
Seasoned Equity Underwritings". University of Oklahoma working paper.

Bradley, Daniel J., Bradford D. Jordan, Ivan C. Roten, and Ha-Chin Yi, 2001, "Venture Capital
and IPO Lockup Expiration: An Empirical Analysis," Journal ofFinancial Research 24,
465-492.

Braun, Matias, and Borja Larrain, 2008, "Do IPOs Affect the Prices of Other Stocks? Evidence
from Emerging Markets," Review of Financial Studies, forthcoming.









CHAPTER 4
SAMPLE AND DESCRIPTIVE STATISTICS

Sample Selection

We select all US common stock seasoned equity offerings in the Dealogic Equity Capital

Markets (ECM) Analytics Database between January 1st, 1996 and December 31st, 2007. To

identify the offer method, we mainly rely on Dealogic's classification, supplemented by the

length of time between the filing day and the trade day. In Appendix A, we compare Dealogic's

classification with the Thomson Financial Securities Data Company's (SDC) new issues

database's classification of the offer method and discuss some details regarding the accuracy of

the classification of the offer method. We find that Dealogic's classifications are much more

accurate.

We apply the following data restrictions:

1) The issuer must be a US-based public company. (ADRs and ADSs are excluded).

2) The issuer's stock must be listed on the NYSE, AMEX, or NASDAQ.

3) Offers on a best efforts basis are excluded.

4) Non SEC registered offers and offers under SEC Rule 144A are excluded.2

5) Private placements, rights offers, and unit offers are excluded.3

6) SEOs of closed-end funds and REITs are excluded.4



1 We excluded 1,213 offers issued on a best efforts basis. The majority of these best efforts are private placements,
usually by small firms. Dealogic does not report the final proceeds or fees for these deals and the prospectuses
usually only list the maximum proceeds and placement agent fees.
2 630 non SEC registered offers and one SEC registered Rule 144A offer are excluded. SEC Rule 144A applies to
private sales of securities to qualified institutional investors.

3 Dealogic-identified private placements include Private Investment in Public Equity (PIPEs), PIPOs and 'registered
direct' offers. The best efforts screening catches most PIPEs and after that, 1 PIPO, 1 rights offer and 223 unit
offerings are excluded.
4 879 closed-end funds and REITs are excluded.









If a stock has a -2% return on a day when 0.5% of its shares outstanding are traded, it

would have an inverse elasticity of 4 on that day. A1 is the average daily inverse elasticity over

the 250 trading days before the announcement.3

A1 is not a precise measure of an individual stock's demand elasticity because the trading

volume includes both buyer-initiated and seller-initiated trades. However, Kalay, Sade, and

Wohl (2004)'s study on all orders that are placed at the Tel Aviv Stock Exchange (TASE)

empirically documents a positive relationship between the flow demand for the stock and its

daily turnover. Therefore, using the total volume instead of just buyer-initiated volume in the

denominator of A, still produces a proxy for the demand elasticity. Our estimated A] is

negatively correlated with the demand elasticity since the price change is in the numerator,

which is why it is termed an inverse price elasticity. A large A, reflects a large change in price if

there is a big demand or supply shock, which implies an inelastic demand curve.

The second measure of price elasticity is an arbitrage risk measure, A2, as used in Wurgler

and Zhuravskaya (2002).4 A2 is the residual variance of a market model OLS regression over the

250 trading days prior to the announcement date:

(,- Rf,)=a+/x(RM, -Rf)+, t =1,2,...,250

In Wurgler and Zhuravskaya (2002)'s model, the demand elasticity for a stock is

determined by the arbitrage risk. Arbitrageurs keep the demand curves flat if the asset has

perfect substitutes and the arbitrage risk is zero. On the other hand, if the asset does not have


3 For the 15 fully marketed offers where there are more than 180 calendar days from the announcement to the offer,
we estimate A,and A2 over the 250 trading days ending 180 calendar days prior to the offer so that the estimation
window is not too distant from the offer date. This treatment has no impact on our results.
4 Wurgler and Zhuravskaya (2002) use two arbitrage risk measures. The other arbitrage risk measure is the residual
variance of a zero-net-investment portfolio with three substitute stocks. The three stocks are matched on industry
and as closely as possible on market capitalization and book-to-market ratio. The two measures are highly
correlated (.97) and all our results remain qualitatively unchanged with the other arbitrage risk measure.




Full Text

PAGE 1

THE MARKETING OF SEASON ED EQUITY OFFERINGS By XIAOHUI GAO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008 1

PAGE 2

2008 Xiaohui Gao 2

PAGE 3

To my dad, Dechen Gao, and my mom, Jiahui Shen 3

PAGE 4

ACKNOWLEDGMENTS I would like to sincerely th ank Jay Ritter (my supervisory committee chair) for his guidance, encouragement, and help. I would al so like to thank Jonathan H. Hamilton, Joel Houston, and Mahen Nimalendran, for their invalu able advice. I would like to thank seminar participants at the Helsinki School of Economics, Norwegian School of Management (BI), Singapore Management University, University of Florida, and University of Hong Kong for helpful comments. Finally, I greatly appreciate all the support that I r eceived from my current and former colleagues, Tom Barkley, Alice Bonaime, and Jennifer Itzkowitz. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT.....................................................................................................................................9 CHAPTER 1 INTRODUCTION................................................................................................................. .10 2 RELATED LITERATURE....................................................................................................16 Marketing................................................................................................................................16 Downward Sloping Demand Curves......................................................................................18 SEOs and Marketing...............................................................................................................22 3 A MODEL OF THE MARKETING OF SEOS.....................................................................24 The Model...............................................................................................................................24 Model Predictions and Testable Hypotheses..........................................................................25 Extension of the Model......................................................................................................... ..27 4 SAMPLE AND DESCRIPTIVE STATISTICS.....................................................................29 Sample Selection....................................................................................................................29 Offer Methods.........................................................................................................................30 Shelf-takedown Offers............................................................................................................32 Offer Characteristics.......................................................................................................... .....33 5 PRICE ELASTICITY PROXIES A ND DETERMINANTS OF OFFER METHOD............47 Price Elasticity Proxy Estimate..............................................................................................47 Determinants of Offer Method...............................................................................................50 Does Speed Rather than Marketing Determine the Issue Method?........................................55 6 CONCLUSION................................................................................................................... ....61 APPENDIX A CLASSIFICATION OF SEO OFFER METHOD..................................................................63 B EXAMPLES OF ACCELERATED AND FULLY MARKETED SEOS..............................65 5

PAGE 6

LIST OF REFERENCES...............................................................................................................67 BIOGRAPHICAL SKETCH.........................................................................................................72 6

PAGE 7

LIST OF TABLES Table page 4-1 Number of SEOs and Aggregate Proceeds by Year and Offer Method............................43 4-2 Shelf Takedown SEOs versus N on-shelf Takedown SEOs, 1996 to 2007........................44 4-3 Mean and Median Offer Char acteristics of SEOs, 1996 to 2007......................................45 5-1 Summary Statistics for Demand Elasticity Proxies...........................................................57 5-2 Correlation among Demand Elasticity Proxi es and Firm and Offer Characteristics.........58 5-3 Determinants of Offering Methods for SEOs, 1996 to 2007.............................................59 5-4 Offering Methods and Pre-offer Stock Price.....................................................................60 7

PAGE 8

LIST OF FIGURES Figure page 2-1 Description of the SEO Process Asso ciated with the Three Offering Methods................23 3-1 Comparison of Demand Curves with and without Marketing...........................................28 4-1 Scatter Diagram Relating Relative Offe r Size to Log(Market Capitalization)..................42 8

PAGE 9

9 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE MARKETING OF SEASON ED EQUITY OFFERINGS By Xiaohui Gao August 2008 Chair: Jay R. Ritter Major: Business Administration Accelerated seasoned equity offerings (SEOs) which include bought deals and accelerated bookbuilt offers, have increased dramatically in the U.S. and globally recently. Accelerated offers are cheaper than traditional fully marketed o ffers in terms of direct issue costs. To explain why some issuing firms choose a fully marketed offer instead of an accelerated offer, we develop a model in which marketing flattens the issuers demand curve. Empirical analysis shows that the pre-issue elasticity of the issuing firms demand curve and the relative offer size are important determinants of the offer method. In ou r analysis, the el asticity of demand at the time of issuance is endogenous.

PAGE 10

CHAPTER 1 INTRODUCTION Seasoned equity offerings (SEOs), also known as follow-on offerings, can be categorized into three major types by their offer methods: fully marketed offers, accelerated offers, and rights offers.1 The academic literature on the issue method for SEOs has focused on rights versus underwritten (bookbuilt) offers, a lthough rights offers are virtually nonexistent in the U.S.2 In contrast, there is no theoretical or empirical treatment of the c hoice between accelerated offers and traditional bookbuilt offers, also known as fully marketed offers. Fully marketed SEOs are issued in much th e same way as bookbuilt initial public offerings (IPOs). Firms wishing to raise cash issue primary shares and current shareholders wishing to sell existing shares issue secondary sh ares, and a prospectus is printe d. The issuer negotiates with one or more investment banks to market the offer and then set the price. The lead underwriter or underwriters conduct a due diligence investigatio n and certify the quality of the company. As part of the marketing effort to develop interest in the offer, the bookrunner usually conducts a road show. In a typical road show, the issue rs management meets with selected institutional investors, analysts, and securitie s sales personnel over a two week pe riod. At the same time, the bookrunner assesses investors demand and builds an order book, which is used to help determine the offer price. A syndicate of underw riters, led by the bookrunne r, then distributes the shares, although for some of the underwriters their only involvement may be in providing subsequent research coverage, or even less. 1 Throughout this paper, the following terms: fully marketed, fully underwritten, regular, firm commitment, bookbuilt, and conventional offers, are used interchangeably unless noted otherwise. 2 Previous studies on rights offers versus bookbuilt offe rs include Smith (1977), Heinkel and Schwartz (1986), Eckbo and Masulis (1992), Bohen, Eckbo, and Michalsen (1997), and Eckbo and Oyvind (2004). 10

PAGE 11

Accelerated offers include bought deals a nd accelerated bookbuilt offers. They are usually shelf registered offers.3 These accelerated deals, especially accelerated bookbuilt offers, have become common during the last decade. Bortolotti, Megginson, and Smart (2007) provide an excellent review of the history and development of accelerated offers.4 Bought deals started in the mid-1980s. With a bought deal, the issuing firm announces the amount of stock it wishes to sell and investment banks bid for these shares, usually by submitting bids shortly after the markets close. The bank that offers the highe st net price wins the deal. The winning bank then re-sells the shares on the open market or to its in vestors, usually within 24 hours. Because of this timing, bought deals are also known as overnight deals. Bought deals are essentially auctions to underwriters followed by open market sales. Accelerated bookbuilt offers emerged in the late 1990s.5 During the last decade, accelerated offers have gained market share qui ckly both in the US and around the world. In 2004, more than a third of issues in the rest of the world were accelerated SEOs, according to Table 2 in Bortolotti, Megginson, and Smart (2007). Unlike bought d eals, in accelerated bookbuilt offers, banks do not initially purchase the whole issue from the issuing company. They submit proposals, usually specifying a gross sp read but not necessarily an offer price, for the right to underwrite the sale of the shares The winning bank then usually forms a small underwriting syndicate and begins marketing the shar es to institutional investors. The offer price is then negotiated between the issuing firm and the bank. The bookbuilding procedure is 3 Some previous studies exclude shelf registered offers (Altinkilic and Hansen, 2003, 2006) and some pool shelf and non-shelf offers together (Corwin, 2003). 4 Bortolotti, Megginson, and Smarts (2007) terminology of bou ght deals is slightly different from ours. They refer to bought deals in Canada as bought deals and bought deals in the US as block trades. We define US block trades as bought deals. 5 Dealogics ECM Analytics Database identifies the first accelerated bookbuilt offer in 1997, which is consistent with the first usage of the te rm accelerated book-b uilding in July 1997 (Warn (1997)). 11

PAGE 12

accelerated in the sense that no road show is conducted and the underwriting procedure is typically completed within 48 hours.6 Before the late 1990s, the US equity market was dominated by fully marketed SEOs. In contrast, many Asian, European, and Australian SE Os used rights offers. In a rights offer, current shareholders are given s hort-term warrants to purchase newl y issued shares on a pro rata basis at a discount relative to the current market price of the stock. The major difference between a fully marketed offer and an accelerated offer, aside from the speed of completing the deal, is the extent of the underwriters marketing of the issue. We develop a model of the role of marketing in s easoned equity offerings to explain the issuing firms choice between a fully marketed offer and an accelerated offer. In the model, marketing flattens the demand curve of the issuers stock. The underwriters marketi ng effort, i.e. a road show, can change the issuing firms stock demand elasticity. To the best of our knowledge, this is the firs t time this marketing paradigm has been used in the finance literature. The bookbuilding l iterature has focused on the acquisition of information by underwriters about the state of the demand, e.g., Benveniste and Spindt (1989). In contrast, we posit that an important feature of traditional bookbuilding is the marketing of the issue. While many studies have assumed that a negatively sloped demand curve exists, these papers have assumed that the elasticity of dema nd is exogenous. This paper not only explicitly adopts a supply and demand framework to analy ze SEOs, but posits that the demand curve that an issuing firm faces is affected by the choice of issue methods. In othe r words, the short-run demand curve that an issuing firm f aces is an endogenous choice variable. 6 This is confirmed in our sample where 96% of accelerat ed offers, including bought deals and accelerated bookbuilt offers, are completed within two days. 12

PAGE 13

The marketing effort can change the elasticity and helps the issuer to achieve a higher offer price and post-issue market price. The benefits of marketing are larger if the demand curve is more inelastic or if the relative offer size is higher. Therefore, our model predicts that an issuing firm is more likely to pay the higher fees and ch oose the fully marketed offer method if it faces an inelastic demand curve prior to the offer. The fully marketed offer method is also preferred if the relative offer size is large. Our empirical results support th e models predictions. We use two measures to proxy for the demand elasticity. The first measure is an or der flow inverse demand elasticity, the ratio of the average daily absolute returns to the aver age daily turnover. The second measure is the arbitrage risk measure used in Wurgler and Zh uravskaya (2002), the variance of market model residual returns. We document that the pre-issue elasticity of the issuing firms demand curve is an important determinant of the offer method. The relative offer size is also an important determinant. For example, of the 211 SEOs conducted in 2007, 91% of firms issuing more than 15% of their equity (72 out of 79) use fully mark eted offers, whereas only 27% of firms issuing less than 10% (13 out of 49) do so. We also fi nd that a smaller fraction of primary shares and less analyst coverage all increase the probability of using a fully marketed offer, which indicates that firms with high information asymmetry favor fully marketed offers. Our model is not mutually exclusive from pr evious studies on information asymmetry and information production. The dow nward-sloping demand curve can arise from asymmetric information, differences of opinion among invest ors (Miller 1977), or how many investors are paying attention to the stock. If the issuing firm faces high information asymmetry, it has a relatively inelastic demand curve and it can benefit from fully marketing the offer. During the underwriting process, the book-r unner produces information about the market demand for the 13

PAGE 14

SEO and then sets the offer price. More s yndicate members, including co-managers and nonmanaging underwriters, can reach out to more investors and provide more information about market interest in the offer, which adds to the marketing efforts. Corwin and Schultz (2005) examine the role of syndicate members in IPOs and find strong evidence of information production. Huang and Zhang (2008) find that more managing underwriters, including bookrunners and co-managers, result in a smaller offer price discount, defined as the percentage price ch ange from the previous market price to the offer price. They attribute this to the managing underwriters preand post-issue marketing efforts. Consistent with these two studies, we find that fully market ed offers tend to have more co-managers and larger syndicate groups compared to accelerated o ffers. In our sample, 66% of the accelerated SEOs do not hire a co-manager. In contrast, only 6.1% of the fully marketed offers do not have a co-manager. Our study supports the argument that demand cu rve elasticity has a si gnificant impact on corporate financial decisions. Hodrick (1999) studies tender offe r choice in share repurchases and finds that firms that choose Dutch auctions instead of the fixed price tender offers tend to face more elastic demand curves. Baker, Coval, and Stein (2007) argue that for acquiring firms that face downward-sloping demand curves, invest ors inertial behavior (i.e. the tendency to react passively) mitigates the negative merger a nnouncement effect and increases the benefit of a stock-for-stock merger. The rest of the paper is organized as follows: In section 2, we review the related literature on marketing and stock demand elastic ities. Section 3 develops a m odel of the role of marketing in seasoned equity offerings. Section 4 discusse s the sample construction and presents univariate 14

PAGE 15

15 patterns. Section 5 estimates two demand elastic ity proxies and analyzes the determinants of offer method with a binomial logist ic model. Section 6 concludes.

PAGE 16

CHAPTER 2 RELATED LITERATURE Marketing Figure 2-1 illustrates and compares the SEO process associated with the three offer methods: bought deals, accelerated bookbuilt, and fully marketed SEOs. Aside from the speed of issue, the major di fference between a fully marketed SEO and an accelerated SEO is the marketing effort by the underwriter. In a fully marketed offer, a preliminary prospectus is distributed and lead underwriters together wi th the issuing firms executives hold a traditional road show. They vis it large institutional investors to create interest and build the demand schedule of the offer. During the road shows, institutional investors acquire more information regarding the issue. The road show proce ss usually lasts for two weeks, so investors have plenty of time to investigate and make their investment decisions. In a bought deal, in contrast the winning bank buys all th e shares and re-sells them immediately, usually overnight. In an acceler ated bookbuilt SEO, the winning bank assembles at most a small underwriting syndicate and usually sells the whole issue within 48 hours. Therefore, there is little marketing effort i nvolved in an accelerated underwriting and investors have little time to evaluate the offer. Marketing has several effects. Marketi ng increases the number of investors paying attention to a given stock. Both theoretical and empirical research shows that attention plays an important role in investors tr ading behavior and the stock ma rket price. In Merton (1987), investors have a positive demand for only those stocks with which they are familiar. Busse and Green (2002) report that trading volume for Nasdaq-listed st ocks increased by an average of 300,000 shares in the four minutes after an analyst mentioned a stock favorably on CNBCs Midday Call segment. Barber and Odean (2008) test and confirm the hypot hesis that individual 16

PAGE 17

investors are net buyers of attent ion-grabbing stocks, described as those that are in the news, or have high trading volume, have extreme one-day returns. In addition, Yuan (2008) finds that high attention also influences both the aggregate market price level and the tradi ng behavior of institutional investors. To formalize the impact of marketing on pri ces, we follow the model in Dixit and Norman (1978). In their model marketing changes cons umer tastes and flattens aggregate demand curves. In Shapiros (1980) model, advert ising flattens each individuals demand curve.1 But our focus is different from theirs. These econom ic theories focus on the equilibrium level of advertising while we focus on the decision of whethe r or not to market the offer in the context of the demand for the shares of a company that is i ssuing stock. One could in terpret our analysis as modeling the discrete decision as to advertise or not advertise. We also make a distinction between shortrun and long-run demand curves. We assume that in the very long run, the price of the stoc k is determined by its cash flow fundamentals and the discount rate that investors a pply, which in turn may be affect ed by liquidity characteristics. The long-run stock demand curve is essentially perf ectly elastic. In the s hort-run, however, this is not the case. The short-run demand curve is as sumed to be negatively sloped. We assume that the short-run demand elasticity is determined by how many investors are paying attention to the stock, and how heterogeneous are their expectations. The underwriters marketing effort helps to fl atten the short-run demand curve, which is reflected in a higher stock price after the issue. This marketin g effort does not come for free, 1 Comanor and Wilson (1967) argue that the demand curve may become more inel astic with advertising. In their analysis, advertising creates product di fferentiation and market power, both of which increase entry barriers. Their framework is less relevant to our study, since investors own stocks primarily for the cash flows that are generated (i.e. in the language of bidding theory, stocks are common value goods rather than private value goods.) 17

PAGE 18

however. The issuing firm needs to balance the gain from a higher stock price with the extra underwriting expenses associated with greater marketing effort. Downward Sloping Demand Curves In a classical efficient market framework, dema nd curves for stocks ar e horizontal. At any point in time, market prices reflect all availabl e information. Therefore, a shock in the demand or supply should not move the stock price if it does not change the expectation of the underlying value. Empirical evidence from many sources however, suggests that demand curves for common stocks are actua lly downward sloping. Previous studies reveal that both short-run and long-run demand curves are not perfectly elastic. Scholes (1972) uses price pressure to describe the downward sloping short-run demand curves that he documents. He studies block sa les and secondary offerings and documents a price decline in such distributi ons of securities. However, Scholes interprets his findi ngs as indicating that information effects are the main cause of the price changes. Curre nt studies most often attribute the short-run effect on price to market liquidity c onstraints. Keim and Madhavan (1996) document significant temporary price impacts from large block trades. A number of studies that look at price changes when stocks are added to or deleted from an index distinguish between short -run and long-run demand curves. Index additions and deletions represent a clean test of demand curve slope s because in most cases they are per se informationless. Shleifer (1986), Harris a nd Gurel (1986), Dhillon and Johnson (1991), Lynch and Mendenhall (1997), and Wurgler and Zhurav skaya (2002) all document positive abnormal returns when stocks are added to the Standard and Poors 500 Index, with the positive abnormal returns only partly revers ed in subsequent trading. Furthe rmore, these studies show that the magnitude of the announcement effects has increased over time, paralleling the growth in the assets of funds that are inde xed to the S&P 500. Lynch and Me ndenhall (1997) find evidence of 18

PAGE 19

both temporary and permanent return components, which supports the notion of different shortrun and long-run elasticities. Wurgler and Zhur avskaya (2002) document that stocks without close substitutes experience a la rger price jump when added to the S&P 500 Index, implying less elastic short-run demand curves for these stocks Morck and Yang (2002) find that the valuation premium associated with being in the S&P 500 I ndex has increased substantially over time and grows in step with the growth of indexing. Th ey argue that indexing demand for these stocks results in a higher price because the long-run supply curve is positively sloped. A positively sloped supply curve for buyers is analogous to a negatively sloped demand curve for sellers. Similar results are reported for the stocks that are added to or deleted from other indices. Madhavan (2003) and Biktimirov, Cowan, and Jordan (2004) report that st ocks added to or deleted from the Russell 2000 in its annual recons titution experience significant returns. Liu (2000) finds positive price changes when stocks are added to the Nikkei 500 Index with no significant post-event reversals. In addition, Kaul, Mehrotra, and Morck (2000) examine stock price changes in response to the re-weighting of the Toronto Stock Exchange 300 Index in 1996, finding strong support for downward sloping demand curves for common stocks. Greenwood (2005) studies the redefinition of the Nikkei 225 index in April 2000. He finds that not only do additions and deletions experience significant pri ce changes, but securities that are not directly affected by the substitution of securities also experience abnormal returns due to their correlation with securities undergoing change s in demand. Furthermore, he shows that the short-run effects are partly reversed over the following 20 weeks, implying a more elastic longrun demand curve than in the short run. Although the studies examining index changes un iversally find price e ffects, they rarely estimate the elasticity because the change in quantity demanded by index funds and other 19

PAGE 20

portfolio managers that are benchmarked to the in dex is difficult to quantify. When shares are issued by the company, however, it is possible to estimate an elastic ity. Kandel, Sarig, and Wohl (1999) analyze the full demand schedules of 27 Israeli IPO auctions and estimate that the average elasticity is -37, indicating that a 37% increase in shares i ssued is associated with a 1% fall in price. Seasoned equity offerings result in an increase in the supply of shares to the public. In the U.S., but not all countries, the a nnouncements are associated with stock price falls. The fall has been interpreted in the literature as due to a combination of information (Myers and Majluf, 1984) and supply effects. Loderer, Cooney, and Van Drunen (1991) study the announcement effect of seasoned offerings by regulated firms. Their event study is a direct test on price elasticity of common stocks because regulated fi rms are less likely to suffer from information asymmetry. Their results suppor t finite pri ce elasticities. Share repurchases represent a decrease in the supply of shares to the public. A positively sloped supply curve for buyers is analogous to a negatively sloped demand curve for sellers. Bagwell (1992) documents that firms face upwar d-sloping supply curves when they repurchase shares in a Dutch auction. Hodrick (1999) confir ms the existence of an inelastic demand curve and concludes that the expected stock elasticity is an important determinant of the tender offer choice. She measures the inverse elasticity as the price premium above the market price before the announcement divided by the fraction of shar es tendered. She finds an average inverse elasticity of 1.27 for Dutch auction firms and 0.99 fo r fixed price tender offer firms. Her study is similar to ours in that she focuses on differences in elasticities determin ing corporate choices. Unlike our analysis, however, she takes the elasticity as exogenous. 20

PAGE 21

Studies on initial public offerings (IPO) al so support downward-sloping demand curves. Braun and Larrain (2008) find that IPOs in em erging markets have negative impacts on the prices of stocks that highly covary with th e IPOs, particularly for large IPOs and in less internationally integrated markets. They conclude that shocks to asset supply have a significant impact on asset prices. Field and Hanka (2001) examine returns around IPO lockup expirations. Because the date of the lockup expiration is known well in advanc e, in an efficient market, there should be no abnormal returns. However, they find a statistic ally significant three-da y abnormal return of 1.5. Many VC-backed funds distribute shares to limited partners at the lockup expiration date, and many of the limited partners immediately sell their shares. VC backing, therefore, may be a proxy for the magnitude of the increase in the flow supply at the time of the lockup expiration. If this is the case, then VC-backed IPOs should ha ve larger price effects if demand curves are negatively sloped. Consistent with this prediction, Bradley, Jordan, Roten, and Yi (2001) and Brav and Gompers (2003) find a larg er price drop at the lockup expiration for VCbacked IPOs. In the market microstructure literature, th e depth of the limit order book determines the short-run elasticity of demand. Larger sell orde rs get executed at prog ressively lower prices because investors assume that larger sell orde rs are more likely to be motivated by negative information (OHara, 1995). Trades that are not motivated by private information, however, face a larger price impact the larger the trade. In equilibrium, the short-run elasticity reflects the markets expectations of the mix of informed and uninformed trades. To summarize, there is widespread eviden ce that both the shortrun and long-run demand curves for stocks are negatively sloped and. Only Bagwell (1992) and Hodrick (1999), however, link the stock demand elasticity to corporate financial decisions. 21

PAGE 22

22 SEOs and Marketing The vast majority of academic studies on the announcement effects associated with SEOs have focused on information releases either explicitly or implicitly. One of the few exceptions is Huang and Zhang (2008), who posit that the use of multiple managing underwriters is motivated by marketing considerations. Huang and Zha ng (2008) report that if one more managing underwriter is hired, the offer pr ice discount is reduced by 0.26%. They argue that the smaller price discount results from the additional ma naging underwriters marketing efforts, which reduce the price pressure and the downside pr ice risk. Huang and Zhang (2008) focus on the gross spread and offer price discount to exam ine the impact of hiring additional mangers on direct and indirect issue costs. Our study has a different focus. We examine the issuing firms pre-issue demand elasticity and analyze the determinants of the offer method.

PAGE 23

Bought Deal Process Accelerated Bookbuilt SEO Process Fully Marketed SEO Process Investment banks submit bids after the close of trading Bank that offers the highest net price wins Registration Re sell the shares Announcement of the filing Trade begins Usually overnight Select the bookrunner Registration Accelerated bookbuilding Pricing and allocation Announcement of the fili n Trade begins Usually 1 2 days Select bookrunner and co managers Prepare preliminary prospectus Registration Marketing, road show, and bookbuilding Issue final prospectus, pricing, and allocation Announcement of the filing 2 3 weeks Trade begins 23 Figure 2-1. Description of the SEO Process Associated with the Three Offering Methods

PAGE 24

CHAPTER 3 A MODEL OF THE MARKETING OF SEOS The Model In this section, we develop a model to study the economics of the underwriters marketing effort. Our model starts with a downward-sloping demand curve. A downward-sloping demand curve can arise from asymmetric information, di fferences of opinions among investors as to the value of the firm, and short-sale constraints. In Wurgler and Zhuravskayas (2002) model, the demand curves slope is more nega tive when (i) the stock does not have close substitutes (more arbitrage risk), (ii) the risk aversion of arbitrageurs is hi gh, (iii) the heterogeneity of nonarbitrageurs beliefs is high, and (iv) the number of arbitrageurs is small. Here, we focus on the issuers short-term demand curve around the seasoned offering. Figure 3-1 illustrates the model. Suppose that there is a group of N investors with homogeneous beliefs about the issu ing firms fundamental value. Each investor has a downward sloping demand curve X(p) Then the aggregate short-run demand curve, the solid line in Figure 3-1, is NX(p) Before the offer, the market equilibrium price is P1, which is jointly determined by the aggregate demand curve and the supply of shares. P1 also represents the price that the stock will converge to in the long-run as the demand curve rotates from the steep short-run relation to the more elastic long -run relation. The offering represen ts an increase in supply (from X1 to X2 in Figure 3-1). If there is no marketing, the short-run demand curve remains the same throughout the offering and as a result of the movement along the demand curve, the stock price drops from P1 to P2. If the issuing firm decides to market the offering, the marketing efforts will attract new investors into the issue, as modeled in a produc t market context by Dixit and Norman (1978) and 24

PAGE 25

Shapiro (1980). For simplicity, we assume that new investors with the same demand curve become aware of the issue and decide to par ticipate. The new aggregate demand curve is (M+N)X(p) represented by the dashed line in Figure 3-1. The new demand curve is now flatter than the old curve NX(p). Therefore, the new post-issue stock price will be higher due to the flatter demand curve. Our model is also similar to the one in Merton (19 87). In Mertons model, a larger investor base increases the equilibrium price. But Merton does not explicitly model the demand curve in his study. The new demand curve results in the issuer fa cing a different elasticity of demand and a higher offer price P*. With the increase in the offer price from P2 to P*, the shaded rectangular area represents the issuers gro ss gain from marketing. For a given supply increase, the gain from marketing will be larger the more inelastic is the pre-offer demand curve. If this gain is sufficiently large and exceeds the direct cost savi ng from switching to an accelerated offer, the issuing firm would prefer the fully marketed offer. Also, note that the size of the shaded rectangle will be larger when there is a larger supply increase. A larger relative offer size represents a larger move along the short-run demand curve, which implies a larger price drop if the demand curve remains the same. If the issue significantly increases the public float and will suffer from greater price pressure, a fully marketed offer is preferred. Model Predictions and Testable Hypotheses Given that the costs of conducting a road show are largely fixed costs, our model predicts that the benefits of a fully ma rketed offer exceed the costs if: The ex ante demand curve of the issuing firm s stock is relatively inelastic, and The issuing firm is offering a large number of shares relative to the shares outstanding prior to the offer 25

PAGE 26

The two predictions yield two testable hypotheses: The issuing firms ex ante demand elasticity is an important determinant of the offer method. The relative offer size is an importa nt determinant of the offer method. We want to point out that our empirical study does not attemp t to estimate the dollar value of the gain from marketing because it is diffi cult to find the post-issu e price with marketing, P*, and the post-issue market price without marketing, P2. For a fully marketed offer, the ex ante demand curve is unobservable so it is hard fo r us to estimate th e counterfactual price P2 had there been no marketing. To find P* is also a challenge. On the day of the SEO announcement, market efficiency suggests that investors should anticipate any price pressure effects associated with the increase in supply and the market pr ice should adjust accordingly at the announcement rather than waiting for the issue day. At the sa me time, the market price should also incorporate the information released by the announcement. Therefore, the observed closing price on the announcement day is the combination of the informa tion effect and the anti cipated price pressure effect. To find P* requires isolating the information effect from the price pressure effect, which is hard to implement empirically. Furtherm ore, the observed announcement day closing price may still deviate from the expected post-issue price due to transaction costs and arbitrage limitations, as well as uncertainty about whether the deal will be completed. In our model, the demand curve should beco me more elastic after the offer due to the marketing effort. However, to empirically compare the pre-issue and post-issue demand elasticities is difficult. Our estimated proxies for elasticity correlate with other firm and market characteristics, which cannot be he ld constant after the offer. In our empirical analysis, we find that fully marketed offers experience the largest increase in demand elasticit y. But an alternative explanation is that a larger market capitaliza tion after the offer may has improved the liquidity. 26

PAGE 27

Extension of the Model Two assumptions in this model can be relaxed to make it more general. First, investors can have heterogeneous beli efs. Even if the new investors are less optimistic than current shareholders, their partic ipation will still flatten the demand curve, as pointed out in Shapiro (1980). In fact, Bagwell (1992) and Kandel, Sarig, and Wohl (1999) have demonstrated that investors have heterogeneous beliefs, or at least heterogeneous reservation prices due to diffe rential tax status. Second, the marketing of SEO can shift the de mand curve even if it does not change the elasticity. The announcement of the SEO may co nvey information to the market. Myers and Majluf (1984) show that under certain conditions the market interprets the announcement of a seasoned offer as a negative signal that the i ssuing firm is overvalued. The demand curve is shifted downward in parallel and results in the price drop upon the announcement of the offer. After the announcement, the underwriters mark eting efforts may reduce the information asymmetry so the demand curve will be flatte ned and shifted upward in parallel. In our empirical study, we do not attempt to disentangle the information asymmetry and the price pressure. However, we predict that if a firm f aces high information asymmetry prior to the issue, it may benefit more from fully marketing the offer. 27

PAGE 28

Supply Demand Quantity Price With Marketing Without Marketing 1P 2P 1X2X *P Demand Curve without Marketing Demand Curve with Marketing Figure 3-1. Comparison of Demand Curv es with and without Marketing. Note: The solid line is the issuing firms dema nd curve before the offer. The dashed line represents the issuing firms dema nd curve if the offer is fully ma rketed. It becomes more elastic than the ex ante demand curve because marketing flattens the demand curve. Therefore, the post-issue price is higher at instead of if there is no marketing. The shaded rectangular area represents the issuers gross gain from marketing the offer. 28

PAGE 29

CHAPTER 4 SAMPLE AND DESCRIPTIVE STATISTICS Sample Selection We select all US common stock seasoned equity offerings in the Dealogic Equity Capital Markets (ECM) Analytics Database between Ja nuary 1st, 1996 and December 31st, 2007. To identify the offer method, we mainly rely on Dealogics classification, supplemented by the length of time between the fili ng day and the trade day. In A ppendix A, we compare Dealogics classification with the Thomson Financial S ecurities Data Companys (SDC) new issues databases classification of the offer method and discuss some details rega rding the accuracy of the classification of the offer me thod. We find that Dealogics classifications are much more accurate. We apply the following data restrictions: 1) The issuer must be a US-based public company. (ADRs and ADSs are excluded). 2) The issuers stock must be listed on the NYSE, AMEX, or NASDAQ. 3) Offers on a best e fforts basis are excluded.1 4) Non SEC registered offers and offers under SEC Rule 144A are excluded.2 5) Private placements, rights offers, and unit offers are excluded.3 6) SEOs of closed-end funds and REITs are excluded.4 1 We excluded 1,213 offers issued on a best efforts basis. The majority of these best efforts are private placements, usually by small firms. Dealogic does not report the fi nal proceeds or fees for these deals and the prospectuses usually only list the maximum pro ceeds and placement agent fees. 2 630 non SEC registered offers and one SEC registered Rule 144A offer are excluded. SEC Rule 144A applies to private sales of securities to qualified institutional investors. 3 Dealogic-identified private placements include Private Investment in Public Equity (PIPEs), PIPOs and registered direct offers. The best efforts screening catches most PIPEs and after that, 1 PIPO, 1 rights offer and 223 unit offerings are excluded. 4 879 closed-end funds and REITs are excluded. 29

PAGE 30

7) Pure secondary offerings are excluded.5 8) The issuing firm must be present on the University of Chicago Center for Research in Security Prices (CRSP) database on the last trading day prior to the offer and the first trading day after the issue. 9) Accelerated offers that are identi fied as non-shelf offers are excluded.6 Pure secondary SEOs are offerings in which all of the shares are being sold by existing shareholders. We exclude pure secondary offerings because they are similar to large sales (block trades) in the open market. Ou r empirical results are robust to including these pure secondary SEOs, and tables are available upon request. Offer Methods Our sample includes 3,615 US SEOs during January 1st, 1996 to December 31st, 2007.7 In Appendix B, we list a few recent large acceler ated and fully marketed sample SEOs with some offer details. Table 4-1 reports the numbe r of offers with each offer method by year. The number of SEOs fluctuates over the twelve year sample period, although the fluctuation in volume from year to year is much smaller than for IPOs. The number of bought deals and accelerated bookbuilt offers has increased substa ntially since 2000. In 1996, there was only 1 bought deal and there were no accelerated bookbu ilt offers. In 2007, there were 27 bought deals 5 846 pure secondary offerings, in which all of the shares are being sold by existing shareholders, are excluded. Among the 846 pure secondary offerings, 262 are bought d eals, 46 are accelerated b ookbuilt offers, and 538 are fully marketed offers. 6 Generally speaking, only companies that are eligible for shelf registrations ar e eligible for accelerated SEO offers. There are exceptions and possible data mistakes. For example, the offer by First Republic Bank on August 15th, 2005, is exempted from registration under Section 3(a)(2) of the Securities Act of 1933. MBNA filed a shelf registration on March 29th, 1999 and conduct a bought deal on August 14th, 2000. But Dealogic reports this as a non-shelf takedown. Most of these deal s are small offers with less than $100 million in proceeds so we exclude all 35 non-shelf accelerated offers. 7 Our sample is smaller than Bortolotti, Megginson, and Smarts (2007) because we exclude ADRs, private placements, and pure se condary offerings. 30

PAGE 31

and 46 accelerated bookbuilt offers. Bought deal s and accelerated bookbuilt offers also have gained market share in terms of proceeds raised over time.8 In 2007, these accelerated offers account for 25% of the total proceeds ra ised in seasoned equity offerings. The offer proceeds fluctuate with the overall ma rket valuation. In our sample, an average SEO raises $94 million in 1996, $276 million in 2000, and $182 million in 2007. Therefore, we adjust the nominal proceeds via scaling by the nominal year-end value of Standard & Poor 500 Index level. The normalized proceeds, expresse d in terms of 1996 valuations, are given by 1996 SEO YearS&P 500 Index Nominal Proceeds S&P 500 Index In unreported results, using two-digit standa rd industry classifica tion (SIC) codes, 68 industries are represented, w ith 75% by industrial firms ( 2,460 out of 3,281) and 25% by regulated and financial firms.9 For bought deals and accelerated bookbuilt offers, around 60% (340 out of 569) are industrial firms. For fully marketed offers, 78% of issuers (2,120 out of 2,712) are industrial firms. Over the sample pe riod, the industry representation in each year exhibits no particular pattern. During our twelve-year sample period, 1,573 firms conduct only one follow-on offering and 660 firms conduct more than one SEO. Among the 660 repeated SEO issuers, 24 conduct more than 5 offers with the most fr equent issuers being financial firms.10 We identify 265 SEOs where the offer method is different from the most recent SEO by the same issuer. Among the 58 SEOs that switched from an accelerated offer to a fully marketed offer, 48 have a larger relative 8 The total proceeds include the exer cised overallotment value. Unlike th e SDC New Issues Database, Dealogic reliably reports the exercised amount. 9 Utility and other historically regulated firms, includin g transportation and telecom, have two-digit SIC codes between 40 and 49. Financial firms have SIC codes between 60 and 69. All other firms are industrial firms. 10 Two firms, Allied Capital and American Capital Strate gies (ACAS), did more than 10 offers. ACAS, which invests in buyouts, had 26 SEOs in 1999-2007. 31

PAGE 32

offer size compared to the previous accelerated offer and the average increase in relative offer size is 7.1%. Among the 207 SEOs that switched fr om a fully marketed offer to an accelerated offer, 172 have a smaller relative offer size a nd the average decrease is 16.9%. Among all the multiple issuers, the average time between two issues is 34 months. The mean number of years from the issuing firms IPO to the SEO is 10 years with a median of 5 years. Firms conducting accelerated SEOs are usually more seasoned than those conducting fully marketed offers, with medians of 10 years and 4 years respectively. More NYSE-listed companies (27%) do accelerated offers compared to NASDAQ-listed companies (12%). This is hardly surprising because large firms are more likely to do an accelerated offer, as we will see in Section 4.4. Shelf-takedown Offers The deregulated streamlined shelf registration pr ocess allows issuing firms to file a single all-encompassing registration statement once ev ery two years rather than filing individual registration statements for every security offe ring. Once its shelf registration statement is approved by the SEC, a firm can issue securities without further disclosure requirements or regulatory delays. In practice, even for bought and accelerated bookbuilt deals, firms normally issue a prospectus. Shelf registration eligibility may lead to a pot ential selection bias in our analysis on the determinants of the offer method. An issuer need s to file a shelf regist ration prior to conducting an accelerated SEO. This requires the issu er to have at least $75 million in market capitalization.11 Therefore, some small firms choose a fully marketed offer because they are not eligible for shelf registration and therefore cannot conduct an accelerated offer. These firms also 11 There are other requirements other than the market capitalization requirement. For details, see Bethel and Krigman (2006). 32

PAGE 33

tend to have inelastic demand curves. Includi ng these small firms biases us towards finding a positive relationship between inelastic demand curves and conducting fully marketed offers. To avoid this bias, we restrict our sample to companies with a $75 million market capitalization prior to the offer. This reduces our sa mple size from 3,615 SEOs to 3,281 SEOs for our empirical analysis. Table 4-2 compares shelf-takedown and non-shelf takedown offers within each offer method. Over the twelve-year sample period, there are 1,335 shelf-takedown SEOs. 766 of those shelf-takedown SEOs are fully markete d. 293 shelf-takedowns are bought deals and 276 are accelerated bookbuilt offers. All of the 1,946 non-shelf takedown offers are fully marketed. Shelf-registered issuers do not seem to have any dominating preference between accelerated offers and fully marketed offers. Al most sixty percent of the shelf-takedown SEOs are fully marketed offers. Autore, Hutton, and K ovacs (2008) also conclude that neither method dominates among shelf issuing firms. Therefore, we do not consider the issuers endogenous decision on shelf registration as an explanatory variable in our empirical tests. Offer Characteristics Table 4-3 compares offer characteristics among the three offer methods. The first column lists results for the enti re sample of SEOs, while the second, third, and fourth columns compare among bought deals, accelerated bookbuilt SEOs, a nd fully marketed SEOs. For each offer characteristic variable, Table 4-3 reports mean a nd median values. The last column reports the p-values from the Kruskal-Wallis (KW) test on the means and the Chi-squared test on the 33

PAGE 34

medians to test for a distribution difference am ong the three subsamples with different offer methods. Our analysis focuses on mean valu es but results based upon medians are similar.12 After normalizing the market capitalization and proceeds by the nominal S&P 500 Index level with 1996 being the base year, the averag e issuing firm in our sample has a market capitalization of $1.2 billion before the offer and raises $113 million in proceeds.13 On average, the issuer is increasing the shar es outstanding by 22.54%, as measur ed by the relative offer size, which is defined as the ratio of the offered shar es to the total shares outstanding prior to the issue. The average percentage increase is much larger than the average proceeds as a percentage of the average pre-issue market capitalization because larger companies typically issue a much smaller fraction of shares than smaller companie s do. Primary shares, which are shares offered by the issuing firm, are on average 85% of the to tal number of shares offered, with most SEOs having 100% of the issue coming from the co mpany rather than existing shareholders. As revealed in columns 2, 3, and 4 in Table 4-3, bought deals, accelerated bookbuilt offers, and fully marketed offers differ substantially from each other. Test statistics indicate that the difference in most of the offer ch aracteristics is statistically sign ificant between the three groups. Furthermore, there is usually a monotonic relation in each mean offer characteristic among the three groups with accelerated bookbuilt offers in between. This indicates that accelerated bookbuilding may be an alternativ e method for marginal issuers in the bought deals and fully marketed deals groups. 12 We use the non-parametric t-test (Kru skal-Wallis test) to test for differ ence in means among the three groups and the non-parametric median Chi-squared test (Brown-Wood te st) on medians. The Kruskal-Wallis (KW) test allows more than two groups and does not require the dependent variables to be normally distributed. 13 The market capitalization is calculated on the last trading day before the announcement of the offer and is adjusted by the annual nominal S&P 500 Index level in the same fash ion as the adjustment on the deal proceeds. For firms with dual class shares, the market cap is restricted to the publicly traded shares reported on the CRSP tapes. 34

PAGE 35

Accelerated bookbuilt SEOs tend to be chosen by larger firms and bought deals are chosen by the largest firms. Compared to an average issuer of a fully marketed offer, the average accelerated bookbuilt issuer is three times larger and the average bought deal issuer is almost four times larger. The offer proceeds, however, show only mode st differences across offer methods. As a result, the relative offer size for fully marketed d eals is on average much larger. Figure 4-1 is a scatter diagram of the relation between the issuin g firms market capitalization prior to the offer and the relative offer size across offer methods for SEOs in 2007.14 Among the 211 SEOs in 2007, 91% of firms issuing more than 15% of thei r equity (72 out of 79) use fully marketed offers, whereas only 27% of firms issui ng less than 10% (13 out of 49) do so. Figure 4-1 shows that firms choosing fully marketed offers tend to be small firms selling a relatively large amount of shares. The larger re lative offer size for fully marketed deals is consistent with our model prediction: Firms that may suffer from a larger price decline due to a larger move along the pre-issue demand curve pr efer to fully market the offer because the marketing effort flattens the demand curve and help s to achieve a higher offer price. The smaller market capitalization for fully ma rketed deals is also consistent with our models prediction: Small firms tend to have less elastic demand cu rves because of lower institutional ownership, and thus receive greater be nefits from marketing. Accelerated bookbuilt offers tend to hire s lightly more bookrunners. In our sample, however, almost 90% of all SEOs have a single bookrunner and the higher incidence of multiple bookrunners for accelerated bookbuilt SEOs may part ly reflect the higher frequency of these deals in the later half of our sample period, when multiple bookrunners were more common. We 14 Other years show the same pattern. Only one years data is used to avoid obscuring the pattern by plotting too many points. 35

PAGE 36

rank bookrunners reputation using the Carter-M anaster ranking obtained from Jay Ritters website. The highest Carter-Manaster ranking is 9 and the lower the ranking is, the less prestigious the bookrunner is. If there are multiple bookrunners, we use the maximum ranking among the bookrunners. The mean bookrunner ranki ng is 8.25 and the median is 9, so most SEOs are underwritten by high reputation banks. Bought deals have slightly more prestigious bookrunners but the differen ce is not significant. Bought deals and accelerated bookbuilt offers, as suggested by the offer method, have a very short underwriting process.15 Not surprisingly, Table 4-3 shows that accelerated offers have a significantly shorter underwriting period. On average, bought deals and accelerated bookbuilt SEOs are offered within one day of the fili ng. In contrast, fully marketed offers spend 31 days on average to complete after the filing, although there is considerable dispersion. e lead 16 A longer underwriting process requires more resources and input from both the issuer and th manager, which also explains why fully marketed offers are the most expensive. During our sample period, there is a monotonic downtrend in the median time between the announcement and completed fully marketed deals. In 1996, th e median time is 31 days and it decreases to 9 days in 2007. 15 We dropped 10 accelerated SEOs that sp end more than 3 days from the filing to the offer. Inclusion of these 10 deals has little impact on our analysis results. Five bought deals spend more than 100 days from the filing to the offer. For example, Rowan Companies filed the registration on Oct 13th, 1999 and completed the offer on Feb 16th, 2000 so there are 126 days between the filing and the o ffer. This bought deal has one underwriter, Lehman Brothers. 16 Note that the average is affected by several extreme cases 14 fully marketed offers spend more than 180 days. The longest case is the offer by Carmike Cinemas Inc in 2004. This offer is announced on Jun 7th, 2002 with an expected pricing date of August 5th, 2002, revised on July 17th, 2002, revised on Nov 18th, 2003, revised on Dec 16th, 2003, and revised on Jan 14th, 2004, and eventually took place on Jan 29th, 2004. So there are 601 days between the announcement and the offer. 36

PAGE 37

Fully marketed offers pay the highest gross spreads, 5.10% on average. Bought deals pay 2.28% and accelerated bookbuilt offers pay 4.23% on average. These results are comparable to those reported in previous studies, including Bortolotti, Megginson, and Smart (2007). The difference in gross spread among the three offer methods is signif icant and important. In a bought deal, the underwriter co mmits to purchase all of the shares for resale to the secondary market. There is no book-building nor road sh ow involved. The underwriter of a bought deal faces greater uncertainty about the price at which the shares can be resold because in general the market has not had a prior opportunity to react to the offer announcement. In an accelerated bookbuilt offer, the lead manager collects price/qu antity pairs from institutional investors and underwriters, and then sets the pr ice of the shares in agreement with the issuer using the order book. In fully marketed offers, the lead manage r conducts a road show while building the order book so the bookbuilding process is longer. In a ccelerated bookbuilt offers and fully marketed offers, the offer price is not se t until after the mark et knows about the issu e and has reset the stock price, so the underwriter does not assume as much price risk as in bought deals. The risk is smaller in fully marketed offers because the market has more time to value the deal and the underwriter has more time to build the order book. The evidence in Table 4-3 suggests that a higher gross spread is associated with bookbuilding and expensive marketing efforts, even though the underwriter is exposed to more resale price risk on bought deals. Unreporte d multivariate analysis confirms that after controlling for various offer and firm characteristics, fully marketed offers pay an average 3% higher gross spread than accelerated SEOs. Desp ite the lower gross spread, an accelerated SEO can still be very lucrative for the underwriter. Take an average accelerated deal with $200 37

PAGE 38

million in proceeds for example, a 2% gross spread represents $4 million in revenue for just a few days work. In Table 4-3, we also report the pre-o ffer stock return, announcement effect, price discount, and underpricing. The pr e-offer stock return is the buy-and-hold stock return (BHR) estimated over the 250 trading days before th e announcement. The announcement effect is the cumulative market-adjusted return (CMAR) estimated over the two-day window [-1, 0], ending with the announcement date (day 0).17 On average, our sample offers experience a significant run-up of 107% and -1.72% cumula tive market adjusted return around the announcement of the offer. If we expand the CM AR to a three-day window, [-1,1], the average return is -2.1%, which is comparable to the numbers presented in previous studies.18 Our results remain qualitatively the same if we use the three-day window. The pre-offer stock run-up and announcement effects vary among th e three groups. Fully marketed offers experience the largest pre-offe r BHR, an average of 118%. Bought deals and accelerated bookbuilt offers experience smaller stock price increase, 72% and 45% respectively. But note that accelerated issuers tend to be big firm s so they are less likely to have a huge price run-up. Bought deals suffer the least amount of negativ e market reaction, an average of -1.49%. Fully marketed offers experience an averag e of -1.66% and accelerated bookbuilt offers experience an average of -2.55%. But note that bought deals and accelerated bookbuilt offers are 0 ,, 1 itmt trr17 Dealogic reports the number of days from the announcement to the offer. We checked several offers and find that if the announcement is made after the market closes and the offer takes place on the next day, Dealogic reports 0 between the announcement date and the offer date. So the announcement day should be th e actual day that the stock price reacts to the announcement. Ther efore, unlike some previous studies on the announcement effect, we exclude the return on the day after the announcement. 18 Interested readers can also refer to Table 5 in Ritter (2003) and Table 13 in Eckbo, Masulis, and Norli (2007), which summarize the announcement effects from different studies. 38

PAGE 39

often announced and underwritten on the sa me day. So the CMAR includes both the announcement effect and price pressure caused by an inelastic demand curve. Fully marketed deals do tend to have a larger relative size. Denoting the closing price on the prev ious trading day of the offer by MPt-1, the offer price by OP and the closing price on the offer day by MPt, the discount is the pe rcentage price change from the closing price on the previous trad ing day of the offer to the offer price, 1 1100%t tOPMP MP Underpricing is the percentage pr ice change from the offer price to the closing price on the offer day, 100%tMPOP OP The absolute value of the discount and underpricing are similar in magnit ude on average. The average SEO in our sample is sold at a 2.76% price discount and experien ces 3.12% underpricing on the offer day. This is generally consistent with the numbers reported in Altin kilic and Hansen (2003), Corwin (2003), and Mola and Loughran (2004). Both the discount and underpr icing levels remain stable during the sample period. These averages, however, hide important diffe rences across offer methods. On average, bought deals have the largest price discount of -3 .96% and the smallest underpricing of 1.09%. this large difference suggests that the price discount for bought deals is capturing the negative announcement effect. Accelerated bookbuilt offers experience an average discount of -2.43% and underpricing of 2.1%. If an accelerated offer is priced and offered w ithin 24 hours after the announcement of the offer, the two-day announcement effect, [-1,0], can be decomposed into the 39

PAGE 40

return on the last trading day before the offer, the discount, and the underpricing.19 Fully marketed offers experience an average discount of -2.66% and the largest underpricing of 3.44%. As we explained in Section 3.3, we do not at tempt to empirically estimate the gain from marketing. Neither the price di scount nor the underp ricing is a good measure of the gain from marketing. Bought deals and accelerated bookbu ilt offers may be unknown to the market in advance. So the observed price discount may incorporate both the discount and an announcement effect.20 The last two rows in Table 4-3 examine th e number of analysts who are following the issuers stock and the stocks averag e bid-ask spread. An analyst is included if he or she posts at least one recommendation within 12 months prior to the offer. The analyst recommendations are obtained from the Institutional Brokers Estimate System (IBES) database. During the year before each SEO, the average number of analysts th at follow an SEO is 5, which is comparable to that reported in Huang and Zhang (2008).21 Accelerated issuers, who tend to be larger companies, receive significantly more analyst co verage, with an average of 7 to 9 analysts following the stock while fully marketed issuers onl y have 4 analysts on aver age. If an issuing firm receives little analyst coverage prior to the offer, it may decide to do a road show to reach more investors and promote the stock. 19 244 of the 293 bough t deals and 137 of the 276 accelerated offers are offered within 24 hours after the announcement. It is interesting to notice that on average, bought deals experience a positive return of 1.6% on the previous day while accelerated bookbuilt offers experience a negative return of -2.6%. One potential explanation is that issuers may wait for a good day, i.e. with positiv e stock return, to conduct a bought deal. For accelerated bookbuilt offers, there may be some information leakage prior to the announcement of the offer. 20 In unreported results, we first estimate the announcemen t effect with the offer and firm characteristics for an accelerated offer if the accelerated offer is announced and of fered within 24 hours. Then we subtract the predicted announcement effect from the realized price discount. Af ter this adjustment, bought deals still have the largest discount of -3.93%. The R-squared is around 0.04. 21 Huang and Zhang (2008) look at the number of recommendations issued prior to an SEO. We find comparable numbers for SEOs in our sample, e.g. on average, there are 7 recommendations issued within 12 months prior to an SEO. 40

PAGE 41

41 The bid-ask spread is the average daily bid-ask spread, scaled by the closing stock price on that day, over the 250 trading days prior to th e announcement of the SEO. Accelerated issuers tend to have a smaller proportional bid-ask spre ad, an average of 0.64%, compared to fully marketed offers with an average of 1.55%. However, the bid-ask sp read exhibits a sharp declining trend in the sample pe riod, from an average of 2.69% in 1996 to an average of 0.31% in 2007. This coincides with the increase in accelerated SEOs during our sample period and may lead to spurious results between the bid-ask sp read and the SEO method se lection. Therefore, we detrend the bid-ask spread in our multivariate regression in Section 5. The univariate results show that accelerated SEOs tend to be large firms and the relative offer sizes are smaller than fully marketed offe rs. Next, we proxy for the issuers pre-offer demand elasticity and examine the offer method choice in a multivariate setting in Section 5.

PAGE 42

42Figure 4-1. Scatter Diagram Relating Relativ e Offer Size to Log(Market Capitalization). The sample consists of 211 SEOs in 2007. The S&P 500 Index level adjusted market capitalization ranges from $40 million, to $22 billion, The relative offer size is the number of shares o ffered divided by the number of sh ares outstanding prior to the of fer and is measured in percentages. The farest lower right dot represen ts Schering-Plough Corps fully marketed SEO in August, 2007. It had an actual market cap of $42 billion prior to the offer and a relative offer size of 3.86%.

PAGE 43

43Table 4-1. Number of SEOs and Aggregate Proceeds by Year and Offer Method. The sample includes 3,615 seasoned equity offerings in Dealogic's ECM Analytics Database during 1996 to 2007. The issuer must be a US-based company listed on NASDAQ, the American Stock Exchan ge (AMEX), or the New York Stock Exchange (NYSE). ADRs and ADSs, private placements rights offers, best efforts, non SEC registered offers, Rule 144A offers, units, closed-end funds, REITs, and pure secondary offerings are excluded. Non-shelf registered bought deals and accelerated book built offers are also excluded. The issuin g firm must be present on the University of Chicago Center for Research in Security Prices (CRSP) database on the first trading day after the issue. Total proceeds is the total amount of do llars (in billions) raised globally including exercised overallotment shares in all tranches. Total proceeds are adjusted by the annual nominal S&P 500 Index leve l, normalized by the ratio between the 1996 S&P 500 index level and the SEO year's S&P 500 index level. Sample SEOs Bought Deals Accelerat ed Bookbuilt SEOs Fully Marketed SEOs Year Number Total Proceeds ($ billion) Number Total Proceeds ($ billion) Number Total Proceeds ($ billion) Number Total Proceeds ($ billion) 1996 487 45.80 1 0.06 0 0.00 486 45.74 1997 418 33.34 2 0.45 8 1.40 408 31.49 1998 285 25.91 8 1.40 2 0.03 275 24.44 1999 321 33.00 20 2.54 2 0.32 299 30.14 2000 323 49.60 29 3.21 2 0.07 292 46.32 2001 244 32.51 35 6.34 18 1.47 191 24.71 2002 229 38.06 25 5.48 41 10.36 163 22.22 2003 267 28.47 36 6.09 42 5.15 189 17.10 2004 303 28.52 36 4.52 51 6.58 216 17.35 2005 249 22.73 32 3.91 29 3.17 188 15.65 2006 255 19.91 43 3.73 38 6.58 174 12.40 2007 234 21.44 27 1.99 46 3.17 161 15.05

PAGE 44

Table 4-2. Shelf Takedown SEOs versus Non-shelf Takedown SEOs, 1996 to 2007 The sample includes 3,281 seasoned equity offeri ngs in Dealogic's ECM Analytics Database during 1996 to 2007. Table II lists the number of shelf takedown and non-shelf takedown offers that are bought deals, accelerated offers, an d fully marketed offers. We excl ude 35 non-shelf takedown offers that are identified by Dealogic as bou ght deals or accelerated bookbuilt offers. All SEOs Bought Deals Accelerated Bookbuilt SEOs Fully Marketed SEOs Shelf-takedowns 1,335 293 276 766 Non-shelf Offers 1,946 0 0 1,946 44

PAGE 45

Table 4-3. Mean and Median Offer Characteristics of SEOs, 1996 to 2007 The sample is restricted to 3,281 issuing firms that have more than $75 million market capitalization before the offer. Table III lists means (medians in brackets) of offer characteristics. Market capitalization is the total market capitalization of equity on the last day prior t o the announcement of the offer. Proceeds is the total amount of dollars (in millions) raised globa lly including exercised overallo tment shares in all tranches. Normalized market capitalization and proceeds are adjusted by the annual nomin al S&P 500 Index level, normalized by th e ratio between the 1996 S&P 500 index lev el and the SEO year's S&P 500 index level. Relative offer size equals offered shares divide d by total shares outstanding prior to the issue. Fraction of primary shares e quals primary (new) shares divided by total number of shares offered. Number of bookrunners is the number of bookrunners on the issue. Bookrunner reputation is the bookrunner's Carter-Manaster ranking obtained from Jay Ritter's webpage. If there are multiple bookrunners, we use the maximum ranking among all the bookrunners. Number of days from filing to offer is the number of days between the filing date and the offer date. Gross spre ad is the disclosed gross fee paid as percentage of the offer price. Pre-offer return is the buy-and-hold stock return during the 250 trading days, [-250, -1], e nding with the announcement date (trading day 0). Announcement effect is the cumulative market-adjusted return during the two-day window, [-1,0], ending with t he announcement date Discount is the percentage decrease from the closing price day before the offer to the offer price. Underpricing is the percen tage increase from the offer price to the closing price on the offer day. The number of analysts following the issuer's stock is obtained from the I/B/E/S database. We include analysts who post a recommendation within 12 months before the offer. Bid-ask spread (%) is the average daily bid ask spread, scaled by the stock price, over 250 trading days prior to the announcement date. The test for the difference in means among the three groups is the non-parametric t-test (Kru skal-Wallis test), which allows more than two groups and does not require the dependent variables to be normally distributed. The test for the medians is the non-parametric median Chisquared test (Brown-Mood test). P-values from the KW and the median Chi-squared statistics are reported in the last column. All SEOs Bought Deals Accelerated Bookbuilt SEOs Fully Marketed SEOs P-values from KW/Median Statistics Number 3,281 293 276 2,712 Nominal Market Capitalization ($M) 1,886 4,552 4,272 1,356 0.00 [ 550 ] [ 1,627 ] [ 1,067 ] [ 463 ] [ 0.00 ] Normalized Market Capitalization ($M) 1,182 2,772 2,718 854 0.00 [ 360 ] [ 961 ] [ 674 ] [ 315 ] [ 0.00 ] Nominal Proceeds ($M) 176 219 208 168 0.00 [ 97 ] [ 129 ] [ 88 ] [ 95 ] [ 0.00 ] Normalized Proceeds ($M) 113 136 134 108 0.00 [ 64 ] [ 81 ] [ 54 ] [ 64 ] [ 0.18 ] Relative Offer Size (%) 22.54 9.36 11.37 25.10 0.00 [ 18.98 ] [ 7.53 ] [ 9.87 ] [ 21.48 ] [ 0.00 ] Fraction of Primary Shares (%) 85.00 98.83 98.58 82.12 0.00 [ 100 ] [ 100 ] [ 100 ] [ 100 ] [ 0.00 ] 45

PAGE 46

46Table 4-3. (continued) All SEOs Bought Deals Accelerated Bookbuilt SEOs Fully Marketed SEOs P-values from KW/Median Statistics Number of Bookrunners 1.19 1.06 1.49 1.17 0.00 [ 1 ] [ 1 ] [ 1 ] [ 1 ] [ 0.00 ] Bookrunner Reputation 8.25 8.36 7.98 8.27 0.08 [ 9 ] [ 9 ] [ 9 ] [ 9 ] [ 0.56 ] Number of Days from Filing to Offer 26 0 1 31 0.00 [ 21 ] [ 0 ] [ 1 ] [ 25 ] [ 0.00 ] Gross Spread (%) 4.82 2.28 4.23 5.10 0.00 [ 5.00 ] [ 1.70 ] [ 4.20 ] [ 5.15 ] [ 0.00 ] Pre-Offer Return (%) 107.70 72.34 45.21 117.89 0.00 [ 57.45 ] [ 26.45 ] [ 20.75 ] [ 65.66 ] [ 0.00 ] Announcement Effect (%) -1.72 -1.49 -2.55 -1.66 0.00 [ -1.53 ] [ -1.48 ] [ -1.99 ] [ -1.48 ] [ 0.00 ] Discount (%) -2.76 -3.96 -2.43 -2.66 0.00 [ -1.93 ] [ -3.47 ] [ -0.90 ] [ -1.89 ] [ 0.00 ] Underpricing (%) 3.12 1.09 2.10 3.44 0.00 [ 1.51 ] [ 0.24 ] [ 0.94 ] [ 1.87 ] [ 0.00 ] Number of Analysts 5 9 7 4 0.00 [ 4 ] [ 7 ] [ 6 ] [ 4 ] [ 0.00 ] Bid-ask Spread (%) 1.39 0.68 0.59 1.55 0.00 [ 1.00 ] [ 0.35 ] [ 0.40 ] [ 1.26 ] [ 0.00 ]

PAGE 47

CHAPTER 5 PRICE ELASTICITY PROXIES AND DETERMINANTS OF OFFER METHOD Price Elasticity Proxy Estimate The definition of elasticity of demand is the percentage change in quantity demanded corresponding to a percentage change in price, formally, q q p p In general, we cannot precisely construct the demand and supply schedule s for individual stocks and directly measure the price elasticity. Due to data limitations that prevent us from identifying supply shifts along a given demand curve or vice versa, we adopt two measures to pr oxy for the elasticity of demand for the issuing firms stock before the offer. The first measure of price elasticity is an av erage daily order flow inverse price elasticity measure, A1. The daily order flow inverse price elas ticity on day T is defined as the ratio between the absolute value of the stocks raw re turn and its turnover, with turnover defined as the trading volume divided by the number of sh ares outstanding. This is called an inverse elasticity because for an elasticity, quantity is in the numerator rather than the denominator.1 If the stock is listed on NASDAQ, we divide the trading volume in half to control for the double counting on NASDAQ relative to the AMEX and NYSE. A1 is the average daily inverse elasticity over the 250 trading day window, [-250,-1], prior to the announcement date:2 250 t 1 t 1 tStock Raw Return 1 Number of Shares Traded 250 Number of Shares OutstandingtA 1 We put the stock return in the numerator to avoid zeroes in the denominator. 2 If the company went public within 250 trading days prio r to the SEO, we start with the first date available on CRSP. 47

PAGE 48

If a stock has a -2% return on a day when 0.5% of its shares outstanding are traded, it would have an inverse elasti city of 4 on that day. A1 is the average daily inverse elasticity over the 250 trading days before the announcement.3 A1 is not a precise measure of an individual stocks demand elasticity because the trading volume includes both buyer-initiated and seller-initiated trades. However, Kalay, Sade, and Wohl (2004)s study on all orders that are placed at the Tel Aviv Stock Exchange (TASE) empirically documents a positive relationship between the flow demand for the stock and its daily turnover. Therefore, usi ng the total volume inst ead of just buyer-initiated volume in the denominator of A1 still produces a proxy for the demand elasticity. Our estimated A1 is negatively correlated with the de mand elasticity since the price change is in the numerator, which is why it is termed an inverse price elasticity. A large A1 reflects a large cha nge in price if there is a big demand or supply shock, which implies an inelastic de mand curve. The second measure of price elasticity is an ar bitrage risk measure, A2, as used in Wurgler and Zhuravskaya (2002).4 A2 is the residual variance of a market model OLS regression over the 250 trading days prior to the announcement date: ,, 1,2,...,250itft MtfttRRRRt In Wurgler and Zhuravskaya (2002)s model, the demand elasticity for a stock is determined by the arbitrage risk. Arbitrageurs keep the demand curves flat if the asset has perfect substitutes and the arbitrage risk is zero On the other hand, if the asset does not have 3 For the 15 fully marketed offers wher e there are more than 180 calendar days from the announcement to the offer, we estimate and over the 250 trading days ending 180 calendar days prior to the offer so that the estimation window is not too distant from the offer date. This treatment has no impact on our results. 1A2A4 Wurgler and Zhuravskaya (2002) use two arbitrage risk measures. The other ar bitrage risk measure is the residual variance of a zero-net-investment portfolio with three substitute stocks. The three stocks are matched on industry and as closely as possible on market capitalization and book-to-market ratio. The two measures are highly correlated (.97) and all our results remain qualitativel y unchanged with the other arbitrage risk measure. 48

PAGE 49

perfect substitutes, the demand curve is downward sloping because the arbitrage risk is nonzero and arbitrageurs are risk averse. The larger the arbitrage risk, A2, the more inelastic the demand curve is. Their empirical analysis document s a positive relation between arbitrage risk and returns on the announcement day of S&P 500 additions which suggests that stocks with greater arbitrage risk have less elastic demand. Our estimated order flow inverse demand el asticity and arbitrage risk measures are qualitative proxies for the indi vidual stocks demand elasticity. Both measures are negatively related to the issuing firms stock demand elasticity prior to the offer with a high value indicating inelastic demand. We are interested in the cross-sectional difference in A1 and A2 among the issuing firms. Panel A and B in Table 5-1 reports means, me dians, and standard deviations of the two measures in the two panels. Our estimates of A2 are comparable to what Wurgler and Zhuravskaya (2002) report. Comparing columns 2, 3, and 4 in Panel A and Panel B, we find the same pattern for both measures. On average, fu lly marketed SEOs have inverse elasticity and arbitrage risk at least twice as high as for bought deals and ac celerated bookbuilt offers. This shows that fully marketed offers are dominated by issuers with relatively inelastic preannouncement demand curves. The univariate results are consistent with our models prediction: if the demand curve is more inelastic, the issui ng firm is more likely to choose a fully marketed offer because the benefits of marketing are larg er. Marketing can flatten the demand curve and achieve a higher post-issue price. The inverse demand elasticity, A1, has a highly skewed distribution, ranging from 0.03 to 1000. For A1, roughly 1% of the sample takes an extreme value because the annualized turnover of the stock is less than 5%. For example, Centennial Communications Corps average daily 49

PAGE 50

turnover is 0.02%, corresponding to an annual tur nover of 5%, before its SEO on November 3rd, 2003. This small turnover rate generates an A1 of 840, almost 100 times the unconditional median of 9.02. Consequently, we use a natural log transformation of A1 and A2 in our empirical work. The last three rows in Panels A and B report the log transformed values of A1 and A2, Ln(A1) and Ln(A2) As expected, the log transformation significantly reduces the skewness in A1. The log transformation has a smaller impact on the level of A2. As a robustness check, we substitute the stock s raw return in by its market adjusted return using the CRSP value-weighted market inde x. The results remain the same. Our results are also robust to the specification of the time window. We estimate and over one-month, three-month, and six-month windows prior to the announcement date a nd all estimates are qualitatively similar across the different windows. Determinants of Offer Method Next, we examine the determinants of th e choice of offer method in a multivariate framework. Our univariate analysis shows th at the choice of offer method is related to firm and offer characteristics. In the multivariate analysis, we include six firm and offer characteristic variables as the explanatory variables. The four firm ch aracteristics are our two pr oxies for the elasticity of the short-run demand curve (the estimated inverse elasticity or the estimated arbitrage risk); the market capitalization before the announcemen t; the average bid-ask spread, scaled by the stock price, over the 250 trading days prior to the announcement; and th e number of analyst recommendations within 12 months before the o ffer, obtained from the I/B/E/S database. We 50

PAGE 51

apply a natural log tran sformation on all of thes e variables, which have skewed distributions.5 To control for the strong declining trend in bi d-ask spread, we detrend the log transformed bidask spread by subtracting the mean value of this variable from SEOs in the same calendar year, bid-ask spreadbid-ask spread Detrended Ln Ln year t average value price price The two offer characteristics variables are offer size and fr action of the offer that is primary shares. Table 5-2 reports the pairwi se Pearson correlation coe fficients among the two demand elasticity proxies, the issuing firms Ln(market capitalizat ion) prior to the announcement, Ln(total proceeds), the relative offer size, the fraction of primar y shares, the detrended Ln(bidask spread), and Ln(1+the number of analysts w ho cover the stock prior to the SEO). The two elasticity proxies are moderately correlated (0.23) with each other. Both proxies are strongly correlated with the issuing firms market capitalization, the bid-ask spread, and the number of analysts, indicating that small firms and firms with relatively high info rmation asymmetry have relatively inelastic demand curves. The relative o ffer size is highly negati vely correlated (-0.47) with firm size because of the tendency of large firms to not raise enormous amounts of money. We do not include the offer proceeds in our regression because the relative offer size is the ratio of the offer proceeds and the market capitalization.6 We use a binomial logistic model to investigat e the determinants of the SEO offer method. The dependent variable, offer method, is a dic hotomous variable for which accelerated offers, including bought deals and accelerated bookbuilt offers equal 1, and fully marketed offers equal 5 If we use the raw values of1 A the regression coefficient is still negative but becomes insignificant. In unreported results, we winsorized at the 1% and 99% level or at the 2% and 98% level, and our empirical results remain qualitatively the same as with the log transformation. If we use the raw values of or the raw values of the number of analyst recommendations, the results remain qualitatively the same. 1A2A6 Our results are qualitatively unchang ed if we include the proceeds. 51

PAGE 52

0. The reference group is fully marketed offers The binomial logistic model is estimated as follows: 01 2 3 45 6+ Relative Size Primary Fraction + Bid-ask Spread1 PrMethod=Accelerated Deal 1 where + Elasticity Proxy+ Ln(Market Cap) + + Ln(1+Number of Analysts) Ln + ue u Table 5-3 presents the binomial logistic regression results. Th e results are very similarly if we include only one elasticity proxy, Ln(A1) order flow inverse demand elasticity, or Ln(A2) residual variance. As predicte d, the coefficients on the two demand elasticity measures are negative, and they are statistically significant at the 1% level. Larger order flow inverse elasticity or arbitrage risk i ndicates a more inelastic demand curve, and this inelasticity encourages issuers to choose a fully market ed offer instead of an accelerated offer. The relative offer size has a significantly negative impact on the probability of an accelerated offer. As our model predicts, ceteris paribus, a large relative offer size increases the gain from marketing so the issuing firm is more likely to choose a fully marketed offer. The likelihood-based pseudo R-square is 28%, which indicate that the model fits well and can explain at least 28% of the total variance. The model correctly predicts almost 90% of the sample offer method. It should be noted, however, that 86% of the offers are fully marketed. The binomial logistic model calculates th e log ratio between probabilities from the reference group to the alternative gr oup, so the coefficients are hard to interpret. In Table 5-3, we report two marginal probabilities, calculated at the sample means and medians, associated with each independent variable. The marginal probability at means is the average change in the probability given a one unit change in the independe nt variable at the sample mean level. For example, holding all the independe nt variables at their mean va lues, a one unit increase in Ln(A1) 52

PAGE 53

or Ln(A2) decreases the probability of an accelerated offer by 2.56% and 0.79% respectively. The marginal probability at medians is calculated in a similar way at the sample median level. Error! Not a valid link. We start with a firm at the bo ttom 25th percentile of the order flow inverse demand elastic ity, which has a value of Ln(A1) equal to 1.43. This firm has a more elastic demand curve than 75 percent of th e sample. We hold the residual variance Ln(A2) the market capitalization, and the re lative size at the 25th percentile and the primary fraction, the bid-ask spread, and the number of analysts at the 75th percent ile. If we increase Ln(A1) by one standard deviation (1.24), the probability of an accelerated offer decreases from 66.9% to 45.4%, a marginal probability of 17%. Similarly, a one standard deviation increase in Ln(A2) (1.04) decreases the probability of an accelerated offer from 66.9% to 61.7%, a marginal probability of 5.0%. For these values of the explanatory variab les, the relative offer size has an even larger impact on the offer method choice. For the same values of the variables, a one standard deviation increase (18.84%) in relative offer size decreases the probability of an accelerated offer from 66.9% to 11.7%. The multivariate regression results support our first and second hypotheses. Both the preissue demand elasticity and the relative offer size are important determinants of the offer method. Firms facing a more inelastic demand curve are more likely to fully market the offer. A larger relative offer size increases the probab ility of a fully marketed offer. The results in Table 5-3 reveal other important determinants of the offer method. A larger firm size decrease the probability of an accelerated offer. This is a rather counterintuitive result. It is probably because the firm size is in the deno minator of the relative offer size. And this is also consistent with the pattern in Figure 4-1. Holding the relative size constant, larger firms seem to prefer a fully marketed offer. A sma ller fraction of primary shares and fewer analysts 53

PAGE 54

covering the stock lower the probability of an acc elerated offer. The analyst coverage can be viewed as a proxy for information asymmetry and th e elasticity of demand. The smaller fraction of primary shares is associated with a highly probability of overvalued stock, e.g. Lee (1997), which is also related to the information asymmetry. Therefore, an issuing firm with high information asymmetry is more likely to choose a fully marketed offer. The detrended log bidask spread is positively correlated with the likelihood of an accelerated offer in the first model. This is a surprising result, since we would expect that stocks with a high bidask spread would be more likely to use a fully marketed offer. One issue is whether the relativ e offer size is determined jointly with the choice of the offer method, or whether this frac tion is predetermined. For a fu lly marketed offer, if the stock price drops too much after the a nnouncement, or there is little demand for the issue, the issuing firm may reduce the offer size or cancel the offer. On the other hand, it sometimes increases the offer size if the demand is strong. For an acceler ated offer, the underwriter has little time to assess investors demand but may still increase or decrease the offer size according to the demand information collected in a very short time period. Around 75% of the bought deals (220 out of 293) and 44% of the accelerated bookbuilt offers (122 out of 276) reduce the deal value and 36% of the fully marketed offers (975 out of 2,712) reduce the deal size. For our empirical analysis, th e question is whether this potential endogeneity problem biases our results. To examine this, we also try a nother measure of the relative offer size. It is defined as the ratio between the initial deal size announced at the filing, and the issuing firms market capitalization prior to the announcement. Our results remain virtually identical, so this potential endogeneity problem does not affect our conclusions. 54

PAGE 55

In each of the first five years of our 19962007 sample period, less than 10% of SEOs are an accelerated offer, where as in each of the last seven years, at least 20% of SEOs do so. Thus, we rerun the binomial regression for the sub-pe riod 2001 to 2007 and presen t the results in the second regression of Table 5-3. The sub-period results are stronger in the later sample period as most of the marginal effects are larger. Ln(A2) becomes insignificant due to a higher correlation (0.37) with Ln(A1) If we exclude Ln(A1) from the regression, Ln(A2) becomes significant and the rest of the coefficients remain unchanged. The bid-ask spread also becomes insignificant in the later sample period. Does Speed Rather than Marketin g Determine the Issue Method? Finally, we discuss an alterna tive explanation on the choice between accelerated and fully marketed offers. The choice of an accelerated offer may indicate the issuing firms preference for speed in the raising of equity capital. Firm s usually conduct SEOs after a substantial price run-up, as we have seen in Table 4-3. If the i ssuing firm thinks that th e high stock price may not be sustained during a fully marketed SEO process, it might prefer an accelerated offer to raise the money quickly. To test this alternative hypo thesis, we use two proxies to measure the issuing firms preference for speed. The first measure is the BHR, the buy-and-ho ld stock return during the 250 trading days (one year) prior to the SEO announcement. The second measure is the stock return variance during the sa me time period. Under the alte rnative hypothesis, firms that have experienced an enormous price apprecia tion would have a str onger incentive to issue quickly. If the stock price is vol atile, the issuing firm might pref er an accelerated offer to avoid a possible sharp decline in the stock price. Table 5-4 presents the regression results of a binomial model similar to that in Table 5-3. The alternative hypothesis is not supported by the results. The BHR and the return variance are both negatively correlated with the probability of doing an accelerated offer. This may result 55

PAGE 56

from the fact that smaller firms tend to have hi gher returns and larger variance regression results while they also tend to choose fully marketed of fers. We want to point out that the unlogged stock return variance is very highly correlated (0.99) with our second elasticity proxy, A2, the residual variance. In addition, during our sample period, ther e is a monotonic downtrend in the median time between the announcement and comple ted fully marketed deals. In 1996, the median time is 31 days and it decreases to only 9 days in 2007. Therefore, the speed advantage of an accelerated deal may have become less important in recent years. 56

PAGE 57

57 Table 5-1. Summary Statistics for Demand Elasticity Proxies The sample includes 3,279 seasoned equity offerings during 1996 to 2007. Two offers are dropped because their stock price informati on is unavailable on the University of Chicago Center for Research in Security Prices (CRSP) database over the 250 trading day window [-250, -1] before the announcement date. The average daily order flow inverse demand elasticity, A1, is defined as the daily raw return divided by the daily turnover, averaged over 250 trading days before the announcement date. The turnover is the trading volume divided by the number of shares outstanding. Nasdaq-listed stocks' trading volumes are divided by two to eliminate double counting. The arbitrage risk measure, A2, is constructed similar to that in Wurgler and Zhuravskaya (2002). A2 is the residual variance, expressed as a squared percentage of a market model OLS regression estimated over 250 trading days before the announcement date. We report means, medians, and standard deviations of the raw values and natural log transformed values of A1 and A2. In the last column, the KW test and median test p-values for the mean s and the medians are the same as in Table III. Panel A: Order Flow Inverse Demand Elasticity A1 All SEOs Bought Deals Accelerated Bookbuilt SEOs Fully Marketed SEOs P-values from KW/Median Statistics Number 3,279 293 276 2,710 Raw Value Mean 28.22 10.01 10.17 32.03 0.00 Raw Value Median 8.52 3.70 4.03 10.17 0.00 Raw Value STD 93.08 56.67 27.22 99.89 Ln Value Mean 2.31 1.39 1.59 2.48 0.00 Ln Value Median 2.14 1.31 1.39 2.32 0.00 Ln Value STD 1.26 0.86 1.00 1.26 Panel B: Arbitrage Risk Measure A2 (%^2) All SEOs Bought Deals Accelerated Bookbuilt SEOs Fully Marketed SEOs P-values from KW/Median Statistics Number 3,279 293 276 2,710 Raw Value Mean 16.92 11.19 10.72 18.17 0.00 Raw Value Median 10.56 6.85 4.99 11.64 0.00 Raw Value STD 20.81 13.96 15.35 21.68 Ln Value Mean 2.31 1.82 1.71 2.42 0.00 LnValue Median 2.36 1.92 1.61 2.45 0.00 Ln Value STD 1.06 1.14 1.15 1.00

PAGE 58

Table 5-2. Correlation among Dema nd Elasticity Proxies and Firm and Offer Characteristics Table 5-2 reports pairwise Pearson correlation coefficients (p-v alues in parentheses) among the two demand elasticity measures, Ln(A1) and Ln(A2), and other firm and offer characteristics. We apply log transformation on the market cap, proceeds, bid-as k spread, and number of analyst s to control for extreme values in these variables. We detrend the log bid-ask spreads by subt racting the sample average log bid-ask spread within the same calendar year. Ln(A1) Ln(A2) Ln(MV) Ln(Proceeds ) Relative Size Primary Fraction Detrended Ln(Bid-ask Spread) Ln(1 + Number of Analysts) Ln(A1) 1.00 0.23 -0.52 -0.39 0.26 -0.14 -0.04 -0.04 ( 0.0 0 ) ( 0.0 0 ) ( 0.00 ) ( 0.0 0 ) ( 0.0 0 ) ( 0.04 ) ( 0.00 ) Ln(A2) 1.00 -0.24 -0.15 0.09 -0.16 -0.03 -0.14 ( 0.0 0 ) ( 0.00 ) ( 0.0 0 ) ( 0.0 0 ) ( 0.05 ) ( 0.00 ) Ln(MV) 1.00 0.77 -0.47 0.06 0.12 0.59 ( 0.00 ) ( 0.0 0 ) ( 0.0 0 ) ( 0.00 ) ( 0.00 ) Ln(Proceeds) 1.00 0.01 -0.11 0.11 0.46 ( 0.5 7 ) ( 0.0 0 ) ( 0.00 ) ( 0.00 ) Relative Size 1.00 -0.15 -0.02 -0.28 58(0 0. 0 ) ( 0.20 ) ( 0.00 ) Primary Fraction 1.00 0.07 0.07 ( 0.00 ) ( 0.00 ) Detrended Ln(Bid-ask Spread) 1.00 0.10 ( 0.00 )

PAGE 59

Table 5-3. Determinants of Of fering Methods for SEOs, 1996 to 2007 Estimate Marginal Effect at Means(%) Marginal Effect at Medians (%) Estimate Marginal Effect at Means (%) Marginal Effect at Medians (%) Ln(A1) ***-2.56 -8.55 ***-6.14-11.33 (80.27) (29.40) Ln(A2) ***-0.79 -2.65 -0.34 -0.63 (15.99) (0.17) Ln(MV) ***-1.51 -5.06 ***-4.05 -7.46 (36.74) (16.60) Relative size (%) ***-0.53 -1.76 ***-1.75 -3.23 (215.73) (164.60) Fraction of Primary Shares (%) ***0.18 0.59 ***0.45 0.84 (50.32) (33.69) Ln(1 + Analysts) ***2.14 7.16 ***5.47 10.09 (30.54) (13.06) Detrended Ln(Bid-ask Spread) ***1.39 4.64 2.10 3.87 (13.77) (2.02) Number of SEOs R-square 0.59 -0.42 0.46 0.05 -0.14 -0.70 -0.22 Table 5-3 presents the results from the binomial logistic regressions for the full sample period and the sub-period of 2001 to 2007. The dependent variable, offering method, is a dichotomous variable for which accelerated deals, including bought deals and accelerated bookbu ilt offers, equal 1, and fully marketed offers equal 0. All the independent variables are defined the same as in Tables III and V. The marginal effects, in percentages, are calculated at the sample means and medians, and are reported to the right of the estimates. The chi-squared statistics are rep orted in parentheses below. *, **, and *** indicate significance at the 10, 5, and 1 percent levels, respectively. The R-square is the likelihood-based pseudo r-square measure. Full Period: 1996 To 2007 Sub-period: 2001 To 2007 Dependent Variable Dependent Variable -0.51 -0.03 -0.34 -0.15 0.04 0.38 0.18 3,275 1,655 0.28 0.32 Pr 1 AcceleratedDeal Pr 1 AcceleratedDeal 59

PAGE 60

Table 5-4. Offering Methods and Pre-offer Stock Price Marginal Effect at Means (%) Marginal Effect at Medians (%) Pre-offer stock return ***-0.58 -2.03 (9.60) Ln(Return Variance) **-0.43 -1.50 (3.74) Ln(A1) ***-2.54 -8.95 (80.78) Ln(MV) *** -1.34 -4.74 (28.78) Relative size (%) ***-0.52 -1.85 (215.19) Fraction of Primary Shares (%) ***0.17 0.61 (50.67) Ln(1 + Analysts) ***1.88 6.65 (23.12) Detrended Ln(Bid-ask Spread) 1.33 4.70 (12.90) Number of SEOs R-square 0.28 -0.16 -0.12 Table 5-4 presents the results from a binomial logistic regression similar to the one in Table VI. To test the preference of speed alternative hypothesis, we include the BHR, the buy-and-hold stock return and the return variance during the 250 trading days prior to the SEO announcement. The rest of the variables are defined the same as in Tables VI. The marginal effects, in percentages, are calculated at sample means and medians and are re ported to the right of the estimates. The chisquared statistics are reported in parentheses below. *, **, and *** indicate significance at the 10, 5, and 1 percent levels, respectively. The R-squa re is the likelihood-based pseudo r-square measure. -0.37 0.52 0.37 -0.70 Dependent Variable Estimate 3,275 -0.14 0.05 Pr 1 AcceleratedDeal 60

PAGE 61

CHAPTER 6 CONCLUSION The recent rise in accelerated SEOs in the U.S. offers us a chance to study the determinants and economics associated with alternative offer methods. We study three major SEO offer methods: bought deals, accelerated bookbuilt offers, and fully marketed offers. The main difference between an accelerated offer, including bought deals and accelerated bookbuilt offers, and a fully marketed offer is that there is no marketing for an accelerated offer. Therefore, we focus on the role of marketing in SEOs. We first develop a model of the issuing fi rms stock demand and supply curves. The issuance represents an increase in the supply. With a downward sloping demand curve, the stock price decreases due to this incr ease in supply. Marketing flatte ns the demand curve and helps to achieve a higher price after the offer. Therefor e, the elasticity of de mand curve is endogenous and is affected by the chosen offer method. Th e model predicts that a fully marketed SEO has greater benefits to the issuing firm when it ex ante faces a highly inelastic demand curve and when the relative i ssue size is large. Our empirical analysis suppor ts the model predictions. We use two measures, the order flow inverse demand elasticity A1 and the arbitrage risk measure A2, to proxy for the demand elasticity. Regression results show that the ex ante elasticity of the issuing firms demand curve and the relative offer size are important determin ants of the offer method. For an issuing firm that is average in other ways, if it has an above average relative issue size of 30% of the pre-issue shares outstanding and ex ante has a rela tively inelastic demand curve, with an Ln(A1) value of 4.06 (the 90th percentile of the order flow inve rse demand elasticity), the probability of using an accelerated offer is 0.4%. If the issuer has a relative offer size of 10% and is in the 10th percentile of the orde r flow inverse demand elasticity (with an Ln(A1) value of 0.84), the 61

PAGE 62

62 probability of using an accelerated offer is 51.6 %. The fraction of primary shares and the amount of analyst coverage also have significant impacts on the choice of the SEO offer method, suggesting that firms with high information asy mmetry tend to choose fully marketed offers.

PAGE 63

APPENDIX A CLASSIFICATION OF SEO OFFER METHOD Our main database is the Dealogic Equity Capital Markets (ECM) database. Dealogic identifies three major SEO offer methods: accel erated bookbuilt, bought deal, and fully marketed offers. The Thomson Financial Securities Data Companys (SDC) new issues database is more commonly used in academic studies. As pointed out in Bortolotti, Megginson, and Smart (2007), SDCs method of the classifying offering technique is sometimes confusing because it gives multiple designations to the same offer. For ex ample, some offers are classified as block trade/negotiated sale, acceler ated bookbuilt/firm commitment, firm commitment/auction. Dealogic gives a single designation to each offer, so we think its classification is less ambiguous. Compared to SDC, Dealogic is more accurate with the filing date and is more consistent with its classification of the offer method. We investigate 519 US seasoned equity offerings during Jan 1st, 2004 to Dec 31st, 2005 listed by bot h Dealogic and SDC. We first hand-checked 35 random offers filing dates in Dealogic with F activa and all of them are correct. Dealogics classification of the offer method is mostly c onsistent with the time length from filing to offering. Accelerated bookbuilt offers and bought deals are almost always completed within 3 calendar days from filing with the SEC. Fully marketed offers take a longer time, ranging from 3 calendar days to more than 150 calendar days. All 68 accelerated bookbuilt offers are completed within 3 calendar days from filing w ith the SEC. Among the 119 bought deals, only 2 offers have a time span longer than 3 calendar days. One is the offer by Trinity Industries Inc, filed on Dec 1st, 2004, which be gan trading on Dec 9th, 2004. The Dealogic filing date is consistent with what we find in Factiva. This is a pure secondary offer so it is not included in our study. The other is by Monster Worldwid e Inc, offered on Jan 5th, 2004. Monster Worldwide filed several S-3/As duri ng the three months prio r to the offer and the last one is filed 63

PAGE 64

64 on Jan 1st, 2004. Among the 331 fully marketed deals, four offers have a time span exceeding 150 calendar days. Two of these four offers filing dates are consistent with Factiva. Dealogic is apparently wrong with one offers filing date and it is reclassified as a bought deal. Another offer has an amended file date 10 days before the o ffer so it is still classified as a fully marketed deal. Therefore, we feel comfortable to rely mainly on the length of time from filing to the offering, supplemented by Dealogics cl assification of the offer method. Next, we focus on discrepancies between SDCs classification and Dealogics classification of the offer method. Among the 519 offers, 416 issues offer methods are classified consistently and 103 are inconsistent. Among the 103 inconsistent offers, 86 offers have consistent offering dates and 17 offers have inconsistent offe ring dates. If the offering date is consistent, we find that Dealogics offer me thod classification is more accurate because all accelerated offers are completed within 3 days a nd all fully marketed offers take more than 3 days to complete. For the 17 offers for whic h the offering dates are inconsistent, Dealogics offering dates are correct for 14 out of the 17, where SDC is correct for only 3 of these cases. Overall, we conclude that Dealogic is a more reliabl e database for our study.

PAGE 65

APPENDIX B EXAMPLES OF ACCELERATED AND FULLY MARKETED SEOS Google announced its first SEO on August 15t h, 2005 and conducted the offer a month later, on September 14th. This is a fully mark eted offer. Google offered 14,159,265 shares at $295 per share, raising $4.3 billion. Morgan Stanley and CSFB are the bookrunners. On March 29th, 2006, Google announced its second SEO a nd conducted the offer through accelerated bookbuilding two days later, on March 31st. It offered 5.3 million shar es at $389.75, raising 42.1 billion. Goldman Sach s is the sole bookrunner. Sallie Mae announced its plan to raise $2. 5 billion through an accelerated bookbuilt SEO after the stock market closes on Dec 26th, 2007. It had an urgent need for capital because it was facing a potential $1 billion loss on its equity forward contracts and a possible downgrade on its debt outstanding by Moody's. The equity forward contracts could force Sallie Mae to purchase 44,039,890 shares at $45 per share while the stock price was around $20. Citibank could demand immediate payment because the stock price had fallen well below the repurchase price. After negotiation, Citibank allowe d Sallie Mae to postpone the re purchase for two months and, together with UBS, underwrote the SEO. Sallie Mae completed the offer on Dec 27th, 2007, less than 24 hours after the offer announcement. It raised $2 billion by selling 101,781,170 common shares and another $1 billion by selling preferred stock. Consolidated Edison conducted two bought deals during 2006 to 2007, raising $447 million and $559 million respectively. The first one was solely underwritten by Citibank and was offered on Sept 21st, 2006. Consolidated Ed ison first filed a prelim inary prospectus with the SEC on Sept 20th, 2006. Early in the morni ng on Sept 21st, it filed a prospectus supplement without specifying the offer price. The offer pr ice was disclosed in a fo llowing supplement filed late in the afternoon on Sept 22nd, 2006. The second bought deal was underwritten by 65

PAGE 66

66 JPMorgan, announced on May 10th, 2007 and offered within one day, on May 11th. The filing procedure was similar to the firs t bought deal. Consolidated Ed ison did not disc lose the offer price until late in th e afternoon on Monday, May 14th. Unfo rtunately, we cannot find any news report on either deal. Nasdaq Stock Market Inc conducted two fu lly marketed SEOs in 2006, raising $639 million and $691 million respectively. The first SEO was announced on Jan 30th, 2006 and offered 10 days later, on Feb 9th. Of the 13 million shares offered, almost half were sold by existing shareholders. The second SEO was a nnounced on April 21st, 2006 and offered 7 days later, on April 27th.

PAGE 67

LIST OF REFERENCES Altinkilic, Oya, and Robert S. Hansen, 2003, Dis counting and Underpricing in Seasoned Equity Offers, Journal of Financial Economics 69, 285-324. Altinkilic, Oya, and Robert S. Hansen, 2006, The Puzzling Wealth Loss around Equity Offerings: Is it Merely Price Pressure or Is There More Information? Working paper, University of Pittsburgh. Asquith, Paul, and David W. Mullins, 1986, Equity Issues and Offering Dilution, Journal of Financial Economics 15, 61 89. Autore, Don, Raman Kumar, and Kilip Shome, 2008, The Revival of Shelf-registered Corporate Equity Offerings, Forthcoming, Journal of Corporate Finance Bagwell, Laurie S., 1992, Dutch Auction Repurchases: An Analysis of Shareholder Heterogeneity, Journal of Finance 47, 71 105. Baker, Malcolm, Joshua Coval, and Jeremy C. Stein, 2007, Corporat e Financing Decisions When Investors Take the Path of Least Resistance, Journal of Financial Economics 84, 266 298. Barber, Brad M., and Terrance Odean, 2008, All that Glitters: The Effect of Attention and News on the Buying Behavior of Individu al and Institutional Investors, Review of Financial Studies 21, 785 818. Benveniste, Lawrence M., and Paul A. Spindt, 1989, How Invest ment Bankers Determine Offer Price and Allocation of New Issues, Journal of Financial Economics 24, 343 361. Bethel, Jennifer, and Laurie Krigman, 2006, M anaging the Costs of Issuing Common Equity: The Role of Registration Choice, Babson College working paper. Biktimirov, N. Ernest, Arnold R. Cowan, and Br adford D. Jordan, 2004, Do Demand Curves for Small Stocks Slope Down? Journal of Financial Research 27, 161 178. Bohren, Oyvind, B. Espen Eckbo, and Dag Michalsen, 1997, Why Underwrite Rights Offerings? Some New Evidence, Journal of Financial Economics 46, 223-261. Bortolotti, Bernardo, William L. Megginson, and Sc ott B. Smart, 2007, "The Rise of Accelerated Seasoned Equity Underwritings". Univ ersity of Oklahoma working paper. Bradley, Daniel J., Bradford D. Jordan, Ivan C. Roten and Ha-Chin Yi, 2001, Venture Capital and IPO Lockup Expiration: An Empirical Analysis, Journal of Financial Research 24, 465-492. Braun, Matias, and Borja Larrain, 2008, Do IPOs Affect the Prices of Other Stocks? Evidence from Emerging Markets, Review of Financial Studies forthcoming. 67

PAGE 68

Brav, Alon, and Paul A. Gompers, 2003, The Role of Lock-ups in Initial Public Offerings, Review of Financial Studies 15, 1-29. Busse, Jeffrey A., and T. Clinton Green, 2002, Market Efficiency in Real Time, Journal of Financial Economics 65, 413 437. Chanine, Salim, Alexander Ljungqvi st, and Roni Michaely, 2008, Marketing Financial Claims, working paper. Chen, Hsuan-Chi, and Jay R. Ritter, 2000, The Seven Percent Solution, Journal of Finance 55, 1105-1131. Comanor, William S., and Tomas A. Wilson, 1967, Advertising Market Structure and Performance, Review of Economics and Statistics 49, 423. Corwin, Shane A., 2003, The Determinants of Underpricing for Seasoned Equity Offers, Journal of Finance 58, 2249-2279. Corwin, Shane A., and Paul Schultz, 2005, The Role of IPO Underwritin g Syndicates: Pricing, Information Production, and Underwriter Competition, Journal of Finance 60, 443-486. Dhillon, Upinder, and Herb Johnson, 1991, Change s in the Standard an d Poors 500 List, Journal of Business 64, 75-85. Dixit, Avinash, and Victor Norm an, 1978, Advertising and Welfare, The Bell Journal of Economics 9, 1-17. Eckbo, B. Espen, and Ronald W. Masulis, 1992, Adverse Selection and the Rights Offer Paradox, Journal of Financial Economics 32, 293-332. Eckbo, B. Espen, Masulis, Ronald W. and Norli, Oyvind, 2007, "Security Offerings", Handbook of Corporate Finance: Empirical Corporate Finance (North-Holland: Amsterdam), Ch13. Eckbo, B. Espen and Norli, Oyvind, "The Choi ce of Seasoned-equity Selling Mechanism: Theory and Evidence" (November 2004). Tu ck School of Business Working Paper. Field, Laura Casares, and Gordon Hanka, 2001, The Expiration of IPO Shares Lockups, Journal of Finance 56, 471-500. Greenwood, Robin, 2005, Shortand long-term Demand Curves for Stocks: Theory and Evidence on the Dynamics of Arbitrage, Journal of Financial Economics 75, 607 649. Hansen, Robert S., and John M. Pinkerton, 1982, Direct Equity Financing: A Resolution of a Paradox. Journal of Finance 37, 651-665. 68

PAGE 69

Harris, Lawrence, and Eitan Gurel, 1986, Price a nd Volume Effects Associated with Changes in the S&P List: New Evidence for the Existence of Price Pressures, Journal of Finance 41, 815-829. Heinkel, Robert, and Eduardo S. Schwartz, 1986, Rights versus Underw ritten offerings: An Asymmetric Information Approach, Journal of Finance 41, 1-18. Heron, Randall A., and Erik Lie, 2004, A Comparison of the Motivations for and the Information Content of Different Types of Equity Offerings, Journal of Business 77, 605-632. Hodrick, Laurie S., 1999, Does Stock Price Elasticity Affect Corporate Financial Decisions? Journal of Financial Economics 52, 225-256. Huang, Rongbing, and Donghang, Zhang, 2008, M anaging Underwriters and the Marketing of Seasoned Equity Offerings, University of South Carolina working paper. Kalay, Avner, Orly Sade, and Avi Wohl, 2004, M easuring Stock Illiquidity: An Investigation of the Demand and Supply Schedules at the TASE, Journal of Financial Economics 74, 461 486. Kandel, Shmuel, Oded Sarig, and Avi Wohl, 1999, The Demand for Stocks: An Analysis of IPO Auctions, Review of Financial Studies 12, 227-247. Kaul, Aditya, Vikas Mehrotra, and Randall Morck, 2000, Demand Curves for Stocks Do Slope Down: New Evidence from an Index Weights Adjustment, Journal of Finance 55, 893912. Keim, Donald, and Ananth Madhavan, 1996, The Upstairs Mark et for Large-Block Transactions: Analysis and M easurement of Price Effects, Review of Financial Studies 9, 1-36. Korajczyk, Robert A., Deborah J. Lucas, and Robert L. Mc Donald, 1991, The Effect of Information Releases on the Prici ng and Timing of Equity Issues, Review of Financial Studies 4, 685-708. Lee, Inmoo, 1997, Do Firms Know ingly Sell Overvalued Equity? Journal of Finance 52, 1439-1466. Liu, Shinhua, 2000, Changes in the Nikkei 500: New Evidence for Downward Sloping Demand Curves for Stocks, International Review of Finance 1:4, 2000, 245-267. Lucas, Deborah J., and Robert L. McDonald, 1990, Equity Issues and Stock Price Dynamics, Journal of Finance 45, 1019-1043. 69

PAGE 70

Loderer, Claudio F., John W. Cooney, and Leonard D. Van Drunen, 1991, The Price Elasticity of Demand for Common Stock, Journal of Finance 46, 621-651. Lynch, Anthony, and Richard Mendenhall, 1997, New Evidence On Stock Price Effects Associated with Changes in the S&P 500 Index, Journal of Business 70, 351-383. Madhavan, Ananth, 2003, The Russell Reconstitution Effect, Financial Analyst Journal 59, 51-64. Masulis, Ronald W., and Ashok N. Korwar, 1986, Seasoned Equity Offerings: An Empirical Investigation, Journal of Financial Economics 15, 91-118. Meidan, Danny, 2004, A Re-examination of Pri ce Pressure around Seasone d Equity Offerings, Northwestern University working paper. Merton, Robert C., 1987, A Simple Model of Capital Market Equilibrium with Incomplete Information, Journal of Finance 42, 483-510. Mola, Simona, and Tim Loughran, 2004, Disco unting and Clustering in Seasoned Equity Offering Prices, Journal of Financial and Quantitative Analysis 39, 1-23. Morck, Randall, and Fan Yang, 2002, The Mysterious Growing Value of S&P 500 Membership, University of Alberta working paper. Myers, Stewart C., and Nicholas S. Majluf, 1984, Corporate Fina ncing and Investment Decisions When Firms Have Informa tion that Investors Do Not Have, Journal of Financial Economics 13, 187-221. OHara, Maureen, Market Microstructure Theory, 1995, Basil Blackwell, Cambridge, MA. Richard Carter and Steven Manaster, 1990, Initial Public Offeri ngs and Underwriter Reputation, Journal of Finance 45, 1045 1067. Rinne, Kalle and Matti J. Suominen, 2007, "Value Uncertainty, Price Imp act and the Choice of the Equity Issuing Method, Helsinki School of Economics working paper. Ritter, Jay R., 2003, "Investment Banking and Securities Issuance," Handbook of the Economics of Finance (North-Holland: Amsterdam), Ch 5. Scholes, Myron S., 1972, The Market for Securitie s: Substitution versus Price Pressure and the Effects of Information on Share prices, Journal of Business 45, 179-211. Schultz, Paul H., 2008, "Downw ard Sloping Demand Curves, the Supply of Shares, and the Collapse of Internet Stock Prices, Journal of Finance 63, 351-378. Shapiro, Carl, 1980, Advertis ing and Welfare: Comment, The Bell Journal of Economics 11, 749-752. 70

PAGE 71

Shleifer, Andrei, 1986, Do Demand Curves Slope Down? Journal of Finance 41, 579-590. Smith, Clifford W. Jr., 1977, A lternative Methods for Raising Ca pital: Rights vs. Underwritten Offers, Journal of Financial Economics 5, 273-307. Securities and Exchange Commission release 2 005-99, SEC Votes To Adopt Securities Act Rule Reform and Shell Company Regulations ; Considers Matters Remanded by Court of Appeals, http://www.sec.gov/news/press/2005-99.htm July 1st 2005. Warn, Ken, Salomon Places $1bn YPF Shares with Institutions, Financial Times (July 17th, 1997), downloaded from Factiva. Wurgler, Jeffrey, and Ekaterina Zhuravskaya, 2002, Does Arbitrage Flatten Demand Curves for Stocks? Journal of Business 75, 582-608. Yuan, Yu, 2008, Attention and Trading, Un iversity of Pennsylvania Working Paper. Zhang, Donghang, 2004, Why Do IPO Underwriters Allocate Extra Shares When They Expect to Buy Them Back? Journal of Financial and Quantitative Analysis 39, 571. 71

PAGE 72

72 BIOGRAPHICAL SKETCH Xiaohui Gao earned her Bachelor of Science in mathematics with con centration in Control Science from Fudan University, China, in 2001. She earned her Master of Science in mathematics with concentration in probability an d statistics from the University of Florida in 2003. Requirements for the degree of Doctor of Philosophy in finance were completed at the University of Florida during the summer of 2008. Upon graduation, Gao w ill join the University of Hong Kong as an assistant professor in the Department of Economics and Finance. Gaos primary research interest s include security offerings and mutual funds.