USF Libraries
USF Digital Collections

Two essays on security offerings

MISSING IMAGE

Material Information

Title:
Two essays on security offerings information production, investor perception and the types of external financing, and a unified analysis on financing choices and offering costs
Physical Description:
Book
Language:
English
Creator:
Yi, Bingsheng
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
Equity financing
Debt financing
Analyst coverage
Information asymmetry
Security offerings
Investor optimism
Dissertations, Academic -- Business Administration -- Doctoral -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: I investigate the impacts that information production, information asymmetry have on firms financing choices 3/4 equity financing or debt financing. I find that equity issue announcements encourage more information production than debt issue announcements, which in turn raises the probability of equity financing. In addition, the post-issue stock market performance is positively associated with information production. The results are robust after controlling for investor optimism. I also apply the Heckmans two-step procedure to jointly investigate firms financing choices and offering costs. I find that security-issuing firms choose the less-costly financing type.
Thesis:
Thesis (Ph.D.)--University of South Florida, 2005.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Bingsheng Yi.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 132 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001680949
oclc - 62391931
usfldc doi - E14-SFE0001173
usfldc handle - e14.1173
System ID:
SFS0025494:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001680949
003 fts
005 20060215071022.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 051201s2005 flu sbm s000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001173
035
(OCoLC)62391931
SFE0001173
040
FHM
c FHM
049
FHMM
090
HF71 (Online)
1 100
Yi, Bingsheng.
0 245
Two essays on security offerings
h [electronic resource] :
b information production, investor perception and the types of external financing, and a unified analysis on financing choices and offering costs /
by Bingsheng Yi.
260
[Tampa, Fla.] :
University of South Florida,
2005.
502
Thesis (Ph.D.)--University of South Florida, 2005.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 132 pages.
Includes vita.
3 520
ABSTRACT: I investigate the impacts that information production, information asymmetry have on firms financing choices 3/4 equity financing or debt financing. I find that equity issue announcements encourage more information production than debt issue announcements, which in turn raises the probability of equity financing. In addition, the post-issue stock market performance is positively associated with information production. The results are robust after controlling for investor optimism. I also apply the Heckmans two-step procedure to jointly investigate firms financing choices and offering costs. I find that security-issuing firms choose the less-costly financing type.
590
Adviser: Scott Besley.
653
Equity financing.
Debt financing.
Analyst coverage.
Information asymmetry.
Security offerings.
Investor optimism.
690
Dissertations, Academic
z USF
x Business Administration
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1173



PAGE 1

Two Essays on Security Offerings: Inform ation Production, Investor Perception and The Types of External Financing, and A Unified Analysis on Financing Choices and Offering Costs by Bingsheng Yi A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Finance College of Business Administration University of South Florida Major Professor: Scott Besley, D.B.A. Barry Lin, Ph.D. Christos Pantzalis, Ph.D. Ninon Sutton, Ph.D. Date of Approval: March 11, 2005 Key Words: Equity Financing, Debt Fi nancing, Analyst Coverage, Information Asymmetry, Security Offerings, Investor Optimism Copyright 2005, Bingsheng Yi

PAGE 2

Table of Contents List of Tables iii List of Figures v Abstracts vi Essay 1 Information Production, Investor Perception and the Types of External Financing 1 I. Introduction 1 II. Literature Review 8 A. Theories on Financing Choices Ignoring Information Production 9 B. Theories on Financing Choices Considering Information Production 12 C. Empirical Evidence on Firms Financing Choices 14 D. Financial Analysts, Information Production and Information Asymmetry 18 E. Objectives 21 III. Hypotheses 22 IV. Data Selection, Variable Descriptions and Methodology 25 A. Data Selection 25 B. Variable Descriptions and Methodology 26 V. Empirical Results and Discussions 30 A. Univariate Test Results 30 B. Multiple Regression Results 34 C. Information Production and Post-issue Stock Performance 38 D. Robustness Tests 40 VI. Summary and Conclusions 45 Essay 2 A Unified Analysis on Financing Choices and Offering Costs 48 I. Introduction 48 II. Literature Review 52 A. Theories on Market Reaction to Security Issue Announcements 52 B. Empirical Findings on Market Reaction to Security Issue Announcements 53 C. Literature on Self-selection Bias 56 III. Data Construction 58 IV. Model Building, Hypotheses and Variable Descriptions 59 A. Model Building 59 i B. Hypotheses 62

PAGE 3

C. Variable Descriptions 63 V. Empirical Results and Discussions 66 A. Summary Statistics and Characteristics Comparisons 66 B. Determinants of Financing Choices 67 C. Determinants of Offering Costs and Market Reaction 68 C.1. Results on Direct Offering Costs 68 C.2. Event Study Results 70 C.3. Results on Three-day Cumulative Abnormal Returns 71 C.4. Results on Indirect Offering Costs 72 C.5. Results on Total Offering Costs 73 VI. Summary and Conclusions 73 References 102 Appendices 116 Appendix A: Reexamine Firm quality and information production via different sample classifications 117 Appendix B: Reexamine information asymmetry and information production via different sample classifications 118 Appendix C: Information production and financing choices, two-step procedure results 119 Appendix D: Reexamine change in analyst coverage and post-issue stock market performance via different sample classifications 121 Appendix E: Construction of Predicted Change in Analyst Coverage 123 About the Author End Page ii

PAGE 4

List of Tables Table 1 Sample Distribution of Security Offerings over 1984-2002 Period 77 Table 2 Summary Statistics 78 Table 3 Firm quality and information production 79 Table 4 Information asymmetry and information production 80 Table 5 Firm Quality, Information Asymmetry and Information Production 81 Table 6 Regressions Without Controlling Endogeneity of Information Production 82 Table 7 Regressions Controlling Endogeneity of Information Production for the Whole Sample Firms 84 Table 8 Change in Analyst Coverage and Long-run Post-issue Buy-and-hold Abnormal Return 86 Table 9 Regressions Controlling Endogeneity of Information Production Based on Randomly Selected Half Sample Firms 88 Table 10 Investor Optimism, Post-issue Buy-and-hold Abnormal Return, and Financing Choices 90 Table 11 Sample Distribution 92 Table 12 Characteristic Comparisons between Equity Issues and Straight Debt Issues 93 iii

PAGE 5

Table 13 Heckman Two-Step-Procedure: The First-Step Probit Regression Results 95 Table 14 Results on Direct Offering Costs 96 Table 15 Event Study Results 98 Table 16 Results on Three-Day Cumulative Abnormal Returns 99 Table 17 Results on Indirect Offering Costs 100 Table 18 Results on Total Offering Costs 101 iv

PAGE 6

List of Figures Figure 1. Analyst Coverage and Time 76 v

PAGE 7

Two Essays on Security Offerings: Information Production, Investor Perception and the Types of External Financing and A Unified Analysis on Financing Choices and Offering Costs Bingsheng Yi ABSTRACT I investigate the impacts that information production, information asymmetry have on firms financing choices equity financing or debt financing. I find that equity issue announcements encourage more information production than debt issue announcements, which in turn raises the probability of equity financing. In addition, the post-issue stock market performance is positively associated with information production. The results are robust after controlling for investor optimism. I also apply the Heckmans two-step procedure to jointly investigate firms financing choices and offering costs. I find that security-issuing firms choose the less-costly financing type. vi

PAGE 8

Essay 1 Information Production, Investor Perception, and the Types of External Financing I. Introduction Every year U.S. companies raise large amount of external capital through the capital markets. On average, non-financial U.S. companies raise $86.4 billion through long-term non-convertible debt offerings, and $13.5 billion through primary equity offerings over the 1984 to 2003 period. 1 Rothschild and Stiglitz (1976) suggest that asymmetric and incomplete information play an important role in financial markets. Leland and Pyle (1977) remark that information asymmetries are particularly pronounced in financial markets. However, it is still an unresolved question whether and how information asymmetry affects a firms external financing choice. As Klein, OBrien and Petyers (2002) conclude, there is no definitive empirical support for specific information explanations of capital structure and financing decisions. The primary objective of this study is to investigate the impacts of information asymmetry and information production on firms financing choices. I examine whether equity financing encourages different information production than debt financing, and whether the difference in impact on information-production between equity financing and debt financing affects firms financing choices. The results provide new evidence on the relationships between information production, information asymmetry and firms external financing choices. 1 These numbers are calculated based on information recorded in Security Data Companys Global New Issues Database. 1

PAGE 9

Theoretical research does not provide consensus on the effect that information asymmetry has on firms financing choices in terms of debt versus equity. On one hand, some theories argue that debt should be used when asymmetric information is substantial. Myers (1984) and Myers & Majluf (1984) present the pecking-order theory, which argues that if firm managers possess more information about assets-in-place than outside investors and that they act in the best interests of current shareholders, managers would rather use debt instead of equity. Narayanan (1988) also argues that when information asymmetry about firm quality is considerable, debt would be preferred to equity because debt financing can keep unprofitable firms out of the capital market, thereby improving the average quality of firms in the capital market. Other theories argue against debt use when significant information asymmetry exists. Stiglitz and Weiss (1981) show that when information about project risks is asymmetric and all projects have the same expected positive returns, information asymmetry about project risks results in credit rationing in the loan market and consequently underinvestment. Using the assumptions in Stiglitz and Weiss (1981), DeMeza and Webb (1987) show that equity dominates debt as the financing method at equilibrium and results in efficient investment. In their model, because all projects have the same expected positive returns, they are equally attractive to equity investors. Therefore equity financing enables all firms to raise needed funds and undertake all the projects with positive net present values. Daniel and Tittman (1995) demonstrate that when information on cash flow variances is asymmetric and firms differ only in cash flow variance, equity is preferred to debt at equilibrium because equity financing does not have an adverse selection problem whereas debt financing does. 2

PAGE 10

The studies just mentioned implicitly assume that information asymmetry is exogenous. Fulghieri and Lukin (2001) model firms financing choices in terms of equity and debt when information asymmetry is endogenous. They assume that firm managers have private information about firm quality and choose a security type that is more likely to be issued successfully, because a successful security issue enables managers to undertake projects and consequently enjoy uncontracTable private benefits. The success of the issue depends on whether there is sufficient demand for the new security issued. Some outside investors (specialized investors) have access to information-production technology and can become informed at some cost. Because equity is more information sensitive than debt, the specialized investors can receive greater expected profits from information production when equity is issued than when debt is issued. As a result, a firm of good quality but facing high levels of information asymmetry can issue equity to induce more information production and informed demand, thus enhancing the chance that the issue will succeed. Sunder (2003) presents a theoretical model, as well as empirical evidence, that suggests firms facing high levels of information asymmetry might issue equity to induce information production to make the stock price more informative, which consequently reduces the borrowing costs in future. Empirical studies that examine the impact of information asymmetry on corporate financing choices are limited and have mixed results (Klein and Belt (1994), Helwege and Liang (1995), Jung, King, and Stulz (1996)). The mixed evidence might be, to some extent, attributed to the information asymmetry variables used and the failure to control for the endogeneity of information production. 2 For example, firm size, market-to-book 2 Helwege and Liang (1995) use variables such as R&D expenses, venture capital financing, output growth in the two-digit industry the issuing firm belongs to, firm age, tangible assets, size, and the number of non3

PAGE 11

ratio, research and development (R&D) expenditures are used as proxy for information asymmetry in some empirical studies. But these variables also measure other characteristics that affect the debt/equity decision. Big firms have greater debt capacity and tend to use more debt. Firms with higher market-to-book ratio have more growth opportunities and consequently might use less debt. R&D expenditures might reflect intangible assets. Firms with greater R&D expenditures have more intangibles assets, thereby use less debt. Therefore, it is difficult to decide whether the research results should be attributed to information asymmetry impact or other factors. This study uses the number of financial analysts following firms to measure information production and information asymmetry. The more analysts that follow a firm, the more information that will be produced and the lower levels of information asymmetry for the firm. I also construct an information asymmetry index based on analyst coverage and other variables that were used to measure information asymmetry in previous studies. Jensen and Meckling (1976) argue that financial analysts reduce information asymmetry and mitigate managerial non-value maximization behavior. Financial analysts earnings forecasts and recommendations are extensively distributed and used by individuals, businesses, and researchers. 3 Merton (1987, p485) remarks that the financial equity offerings in each year as proxies for information asymmetry. Klein and Belt (1994) use the log of sales level, the number of shareholders of a firms stock, or liquidity (the ratio of total shares traded over shares outstanding) as proxy for information asymmetry in their Logistic regressions. Although they share the same sign as sales, the number of common stock shareholders and liquidity never significantly affect the probability of debt issue. Jung, King and Stulz (1996) use total assets to proxy for information asymmetry. Sales and total assets might also be indicators of firms ability to access the debt market, or the size of tangible assets. In such cases, even if the pecking-order theory is correct, the results will be inconsistent with it. 3 Hong, Lim and Stein (2000) report that there are about 3,000 financial analysts (not including junior analysts) working for over 300 investment banks or brokerage houses. More than 63 percent of traded firms receive analyst coverage. Best and Zhang (1993) document that banks rely on financial analysts forecasts and recommendations to screen out loan applications and determine their evaluation and monitoring efforts. When information from analysts is noisy and signals declining prospects, banks have incentive to investigate the borrowers, and consequently such bank loans convey new information. DeGraw (2001) 4

PAGE 12

acquisition of information and its dissemination to other economic units are central activities in all areas of finance, and especially so in capital markets. Financial analysts activities are closely related to the production and distribution of information. Brennan and Hughes (1991) argue that investors tend to trade only securities that they know about. Brennan and Hughes (1991) and Womack (1996) find that analyst coverage expands investors cognizance of a firms securities. A large number of empirical studies have adopted analyst coverage and forecast properties as measures of information asymmetry (DMello and Ferris, 2000), information production (Liu and Qi, 2001), and investor optimism (Doukas, Kim and Pantzalis, 2002). However, no studies have investigated the effects that analyst coverage and forecast properties have on corporate financing choice decisions. Seasoned security issuance is an important event to companies and investors. Companies need to raise money from investors to finance their projects. Investors need to find good-quality firms to invest. When firms issue either equity or debt, the information provided to the SEC is essentially the same. Investors lack expertise in security analysis and desire for more information about security-issuing companies and the securities issued. I conjecture that analyst coverage should increase after security issue announcements due to increase in demand for new information, and the increase in analyst coverage should be greater after equity issue announcement than after debt issue announcements because equity provides greater potential profits from information production than debt. This study investigates some areas that have not been examined in previous studies, including: (1) whether equity financing encourages more information production mentions that lack of analyst coverage sometimes is the reason institutional investors avoid investing in some stocks. 5

PAGE 13

than debt financing, whether such difference can explain the types of external finanicng, and how information asymmetry affects firms financing choices when the impact of information production on financing choices is controlled for. My investigation provides the first empirical evidence on the argument in Fulghiery and Lukin (2001) and Sunder (2002) that firms might choose to issue equity rather than debt when information production is endogenous, (2) whether analysts forecasts change that is, become more optimistic or pessimistic prior to and subsequent to issue announcements, how analysts optimism toward debt-issuing firms differs from the optimism toward equity-issuing firms, and whether the post-issue underperformance of security issuers reflects investors pre-issue overoptimism; (3) whether and how investor optimism affects firms financing choices. Loughran and Ritter (1995) argue that the long-run underperformance of equity issuers is evidence of investor overoptimism and managers opportunism to take advantage of such sentiments. Healy and Palepu (1993, 1995) hypothesize that investors perceptions of a firm are important to corporate managers expecting to issue public debt or equity. My investigation thus provides direct evidence on such arguments. I use the change in analyst coverage the post-issue 12-month mean analyst coverage minus the pre-issue 12-month mean analyst coverage to proxy for information production following security offering announcements. The greater the increases in analyst coverage are, the more information is produced. I find that the number of analysts following a firm increases significantly by 1.43 after equity offerings, but there is only a slight increase of 0.13 after debt offerings. In addition, good-quality firms and firms facing high-levels of information asymmetry tend to have greater increases in analyst coverage. Using ordinary least square (OLS) and probit model 6

PAGE 14

regressions respectively, I find that equity financing leads to greater increases in analyst coverage, and that increases in analyst coverage raise the probability of equity financing. I employ a three-step procedure to control for the endogeneity of information production. I first run a probit regression using a set of explanatory variables that is commonly used in the literature that relates to firms financing choices to get the predicted probability of equity offerings versus debt offerings. I then use the predicted probability of equity issues along with other variables about firm quality, size, and information asymmetry to explain the change in analyst coverage. I find that the predicted probability of equity issue results in significant increase in analyst coverage. I then use predicted change in analyst coverage to explain the firms financing choices. The predicted change in analyst coverage is the predicted value of the model that explains the changes in analyst coverage in the second step. It measures the expected information production by outside investors in anticipation of security offerings. I find that information asymmetry has a negative effect on probability of equity issues whereas the expected information production raises the probability of equity financing versus debt financing. Moreover, there is also a significantly positive effect of the interaction term between information asymmetry and information production on probability of equity issues, indicating that the negative impact of information asymmetry on equity financing is mitigated by the positive effect of expected information production. This study not only provides new evidence that information asymmetry reduces the probability of equity financing, but also confirms the theoretical argument of Fulghier and Lukin (2001) that when information production is endogenous, some firms might issue equity rather than debt. I also examine why firms 7

PAGE 15

desire greater increases in analyst coverage and consequently greater reduction in information asymmetry. Prior studies have documented long-run underperformance of security offerings and that equity issuers have poorer performance than debt issuers (see Loughran and Ritter (1995), Spiess and Affleck-Graves (1995, 1998)). I find that the post-issue buy-and-hold abnormal returns increase with increases in analyst coverage. I also find that equity issuers with high analyst coverage increases neither underperform market nor the debt issuers. Finally I investigate the change in investor optimism during the 12 months surrounding security offers to determine whether differences exist between equity issuers and debt issuers. 4 The evidence indicates that investors are more optimistic about equity issuers than about debt issuers. Investor optimism is strengthened after equity issues but is weakened after debt issues. For both equity and debt issues, the post-issue buy and hold abnormal returns are negatively associated with the pre-issue investor optimism. However investor optimism does not affect firms financing choices. The remainder of this paper is organized as follows. Section II reviews relevant literature. Section III presents the hypotheses. Section IV describes the data and variables. Section V reports and discusses the empirical results, and finally section VI concludes. II. Literature Review 4 I use analysts forecast errors, long-term growth rate forecasts, the ratio of difference between number of analysts making upward revision and number of analysts making downward revision to the total number of analyst following, and the investor optimism index to measure investor optimism. Investor optimism index is the equal-weighted average of the ranks of the aforementioned variables. The construction of our information asymmetry index and investor optimism index follow the method in Butler, Grullon and Weston (2004). 8

PAGE 16

Modigliani and Miller (1958) argue that if capital markets are perfect, the market value of any firm is independent of its capital structure. In reality, the capital markets are not perfect due to the existence of taxes, bankruptcy risk, agency problems, asymmetric information, and so forth. Consequently, capital structure and financing choices are relevant to a firms value. In this section, I first review theoretical and empirical studies relating information asymmetry to firms financing choices in terms of debt versus equity, and then I review literature on analyst coverage. 5 Similar to Klein, OBrien and Petyers (2002), throughout this paper information asymmetry refers to the fact that firm insiders, especially managers, have more and better information than outside investors on the value and risk of their firms assets and investment opportunities. A. Theories on Financing Choice Ignoring Information Production Some researchers argue that debt should dominate equity as a means of external financing in the presence of asymmetric information about assets-in-place or future cash flows. Myers (1984) and Myers and Majluf (1984) assume that firm managers possess more information than outside investors and act in the best interests of current shareholders. If the firms stock is undervalued, managers would rather give up positive NPV projects than issue equity to finance the projects, because the benefits to the old shareholders from undertaking the projects are less than the dilution costs from issuing undervalued equity. If a firm issues equity, outside investors infer that the firms stock is overvalued and correspondingly discount the stock price. Therefore an equity-issue announcement would lead to negative market reaction. 5 The reader can refer to Klein, OBrien and Petyers (2002) for more complete literature review in this area. 9

PAGE 17

The pecking-order theory argues that firms prefer internal funding to external financing. If external financing is required, firms issue the straight debt first, then hybrid securities such as convertible debt or preferred stock are issued, whereas equity is only issued as the last resort. Narayanan (1988) assumes that information asymmetry exists about the mean of the distribution of the cash flows from the firms investment opportunities, but the variance of the distribution of cash flow is common knowledge. The pecking order still applies under such a setting, in that firms will always issue risky debt rather than equity because debt financing can keep unprofitable firms out of the market, thereby improving the average quality of firms in the market and increasing average valuation of firms. On the other hand, other theories argue against debt use under information asymmetry. Stiglitz and Weiss (1981) argue that due to information asymmetry between firm insiders and outside investors (banks) about project risks, high interest rates might induce firms to undertake riskier projects. As a result, banks do not increase interest rates to clear the loan market but rather leave some loan applications unsatisfied (credit rationing), hence debt financing results in underinvestment. To the other extreme, DeMeza and Webb (1987), Giammarino and Neave (1982) show that when firm managers have private information about project risks, issuing equity dominates issuing debt at equilibrium, because it does not incur the adverse selection problem, a debt issue does. In Giammarino and Neaves (1982) model, managers issue debt when their projects are riskier than investors believe. Investors will expect managers opportunistic behavior and refuse to buy debt. As a result, at equilibrium the pecking order is reversed: equity and convertible securities are preferred to debt. Daniel and Titman (1995) demonstrate 10

PAGE 18

that when information asymmetry is about project risks, firms will choose equity rather than debt. A number of models allowing for asymmetric information argue that capital structure or financing choices might act as a signal about firm or project quality. In the signaling model of Ross (1977), managers know the true quality of the firm, but investors do not. High (low) quality firms have high (low) profitability. Firm managers benefit if the firms securities are valued high, and they are penalized if the firm goes bankrupt. Bankruptcy risk rises as the amount of debt issued by the firm increases. High-quality firms issue more debt to signal their good quality, but low-quality firms can not mimic the high-quality firms because they do not have enough cash flow to back up the debt. Consequently, the leverage level is an unambiguous signal about the quality of a firm, and the value of a firm is positively related to its debt-equity ratio. Blazenko (1987) investigates the firms incentive to use debt to signal firm quality. He assumes that firm performance affects managerial compensation and reputation. Because leverage increases the total risk of share ownership, firm managers have preference of using equity financing. With symmetric information, managers exclusively use equity financing. If managers know more than outsiders and are sufficiently risk-averse, they will issue debt to signal the high quality of the projects they consider. Miller and Rock (1985) argue that investors might derive information about a firms future earnings from managers financing decisions. In their model, if a firm uses more (less) external financing than expected, outside investors conclude that firms cannot generate enough cash flows to meet their capital expenditure needs. As a result, the 11

PAGE 19

market responds negatively when firms announce large amounts of unexpected external financing. Noe (1988) models a firms financing decisions as a sequential signaling game. Noe shows that when firm insiders have perfect private information on firms future cash flows, debt dominates equity as the equilibrium financing method for all firms. However, when firm insiders have imperfect private information about the firms cash flows, at equilibrium, both high(H) and low-quality (L) firms issue debt, whereas medium-quality (M) firms issue equity. The market can correctly infer the value of M, but not that of H or L. This equilibrium rejects the pecking-order theory because a type M firm strictly prefers equity financing to debt financing. In addition, Lucas and McDonald (1990) argue that managers with private information will time their equity offerings. On one hand, undervalued firms delay issuing equity until the undervaluation is corrected. These firms, therefore, have above-average performance before the issue announcement. On the other hand, overvalued firms issue equity immediately upon identifying a new project, because waiting might lead to the loss of the project. These firms therefore experience below-average performance before equity issue announcement. Lucas and McDonalds model predicts that (1) equity issues on average are preceded by a positive abnormal return on the stock, while some issues may be preceded by a loss; (2) equity issues on average are preceded by an abnormal rise in the market; (3) stock price drops upon announcement of equity issue. B. Theories on Financing Choice Considering Information Production 12

PAGE 20

The aforementioned models ignore the information production problem and implicitly assume that information asymmetry between insiders and outsiders is exogenous and constant. This is quite unrealistic, because a firm disclosure behavior and the activities of financial analysts that follow the firm bring new information to the market, thereby affecting the extent of information asymmetry between firm insiders and outside investors. Boot and Thakor (1993) first show that information sensitive securities encourage endogenous information production. Fughieri and Lukin (2001) and Sunder (2002) argue that public equity provides a stronger incentive for information production than public debt. In the Fulghieri and Lukin model, managers have private information on their firms quality, which can be either good or bad, while the distribution of firm quality is common knowledge. There are two types of investors and a group of competitive market makers: Uninformed investors exert an exogenous and inelastic demand for firms securities, whereas specialized investors have access to costly information-production technology, and thus they choose whether to produce the information and thereby become informed or not. If specialized investors find that the firm is good, they initiate a purchase order on the security the firm issues to the market makers; if they find the quality of the firm is poor, they short the securities the firm issues. 6 The market makers set the price equal to the expected value of the security conditional on the observed demand. If the total demand from the informed and uninformed investors is not sufficient enough, the perceived quality of the issuing firm and, consequently, the issue price will be so low that the firm cannot raise the desired funds, which means the issue fails. Firm managers want to maximize the probability of a 6 Fulghieri and Lukin (2001) claim the main results in their paper will hold also when short sales are prohibited. 13

PAGE 21

successful issue and the stock values of original shareholders. They choose the security type to affect specialized investors information-production activities and their subsequent informed trading. Given the security issued by the firm, specialized investors will choose to become informed so long as they expect to earn profits from information acquisition. An increase in the volume of informed trading increases total demand and induces a higher issuing price, thereby raising the chance that the issue will succeed. Because equity is a junior claim and is more information sensitive than debt, debt does not always dominate equity when firms seek external financing. If the cost to acquire the information is sufficiently low, issuing equity increases informed trading and thereby decreases the minimum level of demand from uninformed investors that is necessary for the issue to succeed. Therefore, in this case, firms of good quality prefer to stimulate information production to enhance the chance of a successful issue by issuing equity rather than debt, whereas poor-quality firms might try to mimic the financing behavior of good-quality firms. Sunder (2002) presents both theoretical and empirical evidence on the impact that information spillovers from public securities have on firms capital structures and long-run financing costs. Sunder argues that firms might bear the full lemon costs of issuing equity today to induce information production, which is reflected in stock prices and spills over into prices of other tradable securities, leading to reduced information asymmetry and financing costs in the future. She finds that borrowing costs of firms decrease with measures of information production in stock markets. C. Empirical Evidence on Firms Financing Choices 14

PAGE 22

One branch of studies investigates stock market reaction to security offering announcements. Asquith and Mullins (1986), Dann and Mikkelson (1984), Eckbo (1986), Smith (1986)) and others document that on average, both equity and convertible debt issues result in significantly negative market reactions. 7 However, the evidence on market reaction to debt issues is mixed. 8 In general, empirical findings support the pecking order theory and the signaling models mentioned before (except Rock and Miller (1985)) in that market reaction to equity issues is significantly more negative than the reaction to debt issues. For example, Smith (1986) reports that the cumulative two-day announcement-period abnormal return is .15 percent for equity issues, -2.07 percent for convertible debt offers, and .26 percent for straight debt issues. The other branch of empirical studies investigates the determinants of security issue choice. Helwege and Liang (1996) track the financing behavior during the1984-1992 period of firms that went public in 1983. They apply a logit model to predict external financing and a multinomial logit model to predict the type of financing. Their results show that the probability of raising external funds is unrelated to the shortfall in internal funds and that firms do not follow the pecking order when choosing the type of security to offer. Helwege and Liang include variables such as R&D expenses, venture capital financing, output growth in the two-digit industry to which the issuing firm belongs, the firms age, tangible assets, size, and the number of non-financial equity offerings in each year as measures of information asymmetry. They find that information 7 The market reaction to equity issue is more negative than to convertible debt issue. 8 Dann and Mikkelson (1984), Ecobo(1986), Mikkelson and Partch (1986), Shyan-Sunder (1991) find that market response to straight debt issue announcements are insignificantly different from zero. Johnson (1995) finds significantly positive stock price reactions to debt issue announcements for low-dividend payout firms. Manuel, Brooks and Schadler (1993) document significantly negative market reactions to debt-issue announcements closely preceding dividend and earnings announcements. Howton, Howton and Perfect (1998) find a significant negative market reaction of .387 on the announcement date of straight debt issue without conditioning on dividend or earnings announcements. 15

PAGE 23

asymmetry variables increase both the probability of public bond issuance relative to the private debt issuance and the probability of equity issuance relative to private debt issuance in the same way, which is inconsistent with pecking-order theory. Klein and Belt (1994) choose firms with actual sales growth during the 1984-1988 period that exceeded a sustainable growth rate computed based on 1983 sales level, and apply the contingent table analysis as well as logit analysis. They separately use the natural logarithm of sales level, the number of shareholders of a firms stock, or liquidity (the ratio of total shares traded over shares outstanding) as proxy for information asymmetry. 9 In their logistic regressions of internal financing versus external financing, only sales and the number of common stock shareholders are significant. In their regressions of debt versus equity, although the three measures of information asymmetry have the same positive sign, only sales level significantly increases the probability of a debt issue. A possible explanation is that sales level might proxy for the ability to access the debt market, or proxy for the value of tangible assets. Also, the number of common stock shareholders and liquidity may not proxy for information asymmetry as well as analyst coverage. 10 Jung, Kim and Stulz (1996) apply logistic regressions on a sample of 192 equity issues and 276 bond issues from 1977 through 1984. They also examine stock price reactions to issue announcements, and post-issue corporate actions. They find strong support for the agency model, which argues that firm managers pursue their own objective (such as growth) at the expense of shareholders. The results show that firms 9 They believe that larger values for these variables represent lower levels of information asymmetry. 10 Since many common stock shareholders do not care about the firms they own, lack access to information sources and the expertise to analyze firms financial statements, the number of common stock shareholders may not be a good proxy for information asymmetry. Information reflected through liquidity may be limited. High-liquidity stocks may reflect information more quickly as a result of high analyst coverage on these stocks. Financial analysts have comparative advantages in generating and releasing information, as well as monitoring firm management. Analyst coverage increase investor cognizance (Brennan and Hughes (1991)) and stock liquidity (Brennan and Subrahmanyan (1995), Brennan and Tamarowski (2000). 16

PAGE 24

without valuable investment opportunities have a more negative stock price reaction to equity issues than firms with better investment opportunities. The worst stock price reactions occur to firms that have characteristics similar to debt-issuers but issue equity to finance capital expenditures. Jung et al. (1996) use past cumulative excess return, six-month leading indicators of economic activities, slack, and total assets to proxy for information asymmetry, and find that past cumulative excess return and six-month leading indicators of economic activities raise the probability of equity issue, indicating firms time equity issue to coincide with periods of low information asymmetry. However, the post-issue cumulative excess returns fail to explain the debt-equity choice, which rejects the hypothesis that firms time equity issues when stock prices are overvalued. Large firms are found to be less likely to issue equity and slack (measured by ratio of cash and liquid assets over total assets) does not affect debt-equity choice; both results are inconsistent with the pecking-order theory. In general, most studies on the determinants of security choice focus on choice between debt and equity, and they find that information asymmetry, agency problems, bankruptcy costs, and tax considerations affect a firms debt-equity issue choice (see Baxter and Cragg 1972, Marsh 1982, Titman and Wessels (1988), Makie-Mason (1990), Bayless (1994), Jung, King and Stulz (1996), Hovakimian, Opler and Titman (2001), among others). Particularly, factors such as growth opportunities, R&D expenditure, pre-issue price run-up induce equity issues, and large firms prefer to issuing debt rather than equity. However, the evidence is mixed on how information asymmetry affects a firms security issue choice. No empirical studies have ever investigated whether equity and debt have different impacts on information production and such differences, if existing, 17

PAGE 25

affect firms financing choices. This study first jointly examines the effects that information asymmetry, information production and their interaction have on firms financing choices. It also fills the gap by providing direct link between pre-issue investor optimism and firms financing methods and pre-issue investor optimism and post-issue stock market performance. D. Financial Analysts, Information Production and Information Asymmetry Jensen and Meckling (1976) contend that security analysts generate and disseminate information about firms they cover and they monitor managers agency behavior. Merton (1987) presents a model showing that firm values are positively associated with the breadth of investor cognizance. He remarks that The acquisition of information and its dissemination to other economic units are, as we all know, central activities in all areas of finance, and especially so in capital markets. There are numerous studies supporting that financial analysts play key roles in producing information and reducing information asymmetry. Brennan and Hughes (1991), and Womack (1996) find evidence supporting the notion that analyst coverage expands investors cognizance of a firms securities. Fried and Givoly (1982), Gordon, Gordon and Gould (1989) show that analysts earnings forecasts are superior to other earnings forecast methods. Brennan and Subrahmanyam (1995) and Roulstone (2001) report that firms with greater analyst coverage have smaller bid-ask spreads and greater liquidity than firms with less coverage. Brennan and Subrahmanyam (1995) also find that greater analyst coverage tends to reduce adverse selection costs measured by the inverse of market depth. Because analysts have access to specialized resources and possess the 18

PAGE 26

knowledge and skills needed to analyze corporate financial data, they have a comparative advantage over typical investors in monitoring corporate managers and generating economically relevant information, which is released to investors through public reports and announcements. Consequently, analyst coverage reduces information asymmetry between firm insiders and outside investors, and helps improve firm performance. Irvine (2003), Branson, Guffey, and Pagach (1998) find that the market responds positively to initiation of analyst coverage, indicating that the market expects analyst coverage to reduce information asymmetry. Consistent with this hypothesis, DMello and Ferris (2000) find that equity issue announcement period returns are significantly less negative for firms with higher analysts coverage, which also positively (negatively) affects a firms long-run stock return after issue. Since Bhushan (1989) and Moyler, Chatfield and Sisneros (1989), more and more studies use analyst coverage as measures on information asymmetry, information production or stock price informativeness. For example, Barth and Hutton (2004) find that stock prices for firms with more analyst coverage incorporate information on cash flows and accruals more quickly than stock prices for firms with less analyst coverage. Liu and Qi (2001) use the analyst coverage as the proxy for information production and find that analyst converge explains a significant portion of the cross-sectional variation in diversification discount during the period 1985-1999. Liu and Qi (2002) present a model predicting that cross-sectionally investment-cash flow sensitivity will be stronger for firms with more informative stock prices. They find firms with higher analysts coverage have greater investment-cash flow sensitivity than firms with lower analyst coverage. 19

PAGE 27

However, there is evidence (Trueman 1994; Welch 2000) suggesting that analysts may engage in herding, which casts doubts on analysts information production function. Some studies find that due to conflicts of interest, financial analysts may generate forecasts that are overly optimistic. Easterwood and Nutt (1999) show that analysts optimistically respond to information. Other studies like Stickle (1992), Abarbanell (1991), Dreman and Berry (1995), and Chopra (1998) also find that analyst forecasts are on average optimistically biased. Financial analysts make optimistic forecasts on purpose to cultivate management access for private information about the firm and to generate underwriting businesses and trading commissions for their brokerage houses. Analysts are rewarded for such behavior. For example, Baker (1996) reports that analysts earn large bonuses for bringing investment banking clients to their firms. 11 Lin and McNichols (1998), Michaely and Womack (1999) and Dechow, Hutton and Sloan (2000) document that analysts are overoptimistic about a firms future prospects around equity issues. Therefore, they make overoptimistic long-term growth forecasts about equity-offering firms earnings. Teoh and Wong (2002) find that the issue-year excess accruals have persistent impact on errors in analysts annual earnings forecasts after the issue. Analysts generally are credulous about the issuers total and excess accruals no matter whether they are affiliated or unaffiliated. Their results indicate that analysts contribute to investor optimism about new issues and thereby question financial analysts abilities in reducing 11 In recent years regulation has changed a lot. The Securities and Exchange Commission imposed the Regulation Fair Disclosure Rule, which prevents analysts from obtaining information exclusively from company insiders. The Sarbarnes-Oxley Act 2002 prohibits research analysts from participating in the solicitation of investment-banking business. To protect analyst independence, this act also prohibit firms from retaliating against research analyst who publish reports or express public opinions that may adversely the firms investment-banking business. See Brown (2005). 20

PAGE 28

information asymmetry. 12 More recent studies (see Leone and Wu (2002), Hong and Kubik (2003), Jackson (2004) among others) find that analysts compensations are tied with their forecast precision and reputation, and that analysts reputations are positively affected by their forecast precision. Therefore due to concerns for their reputations and compensations, analysts have incentives to generate precise information. In On Wall Street (Mar 1st, 2001), DeGraw writes, Without analyst coverage, a firm with limited recognition has little hope for attaining the investor attention needed for full valuation. The absence of any major analyst coverage is a sentence to trading obscurity. Lack of analyst coverage sometimes is the reason why institutional investors avoid investing in some stocks. Krigman, Shaw and Womack (2001) find that seasoned equity offering (SEO) firms choose underwriters different from those completing their initial public offerings in order to receive better research coverage from analysts, indicating that firms care very much about analyst coverage. E. Objectives In addition to the fact that theoretical studies do not agree with impact of information asymmetry on firms financing methods, empirical studies are limited and also do not provide consistent results on the effect that information asymmetry has on seasoned financing choices. Even though they are commonly used to measure information asymmetry or to proxy for investors perception, analyst coverage and forecast properties have never been used to examine firms incremental financing 12 Hansan and Sarin (1998) examine analysts forecast behavior around SEOs. They document and adjust the bias that equity issuers are among the high growth IBES firms which typically have much higher forecast errors. Their findings suggest that either the short-run or long-run abnormal forecasts observed around equity issuance are not unique but is instead a common phenomenon to high growth firms. They also find that managerial earnings manipulation does not affect analysts earning forecasts. 21

PAGE 29

choices, and no studies have ever examined whether equity financing will induce more information production than debt financing, thus raising the likelihood that firms will issue equity rather than debt. As a result, my primary objective is to investigate the relationship between financing choices and information production, and to provide new information-oriented evidence on seasoned corporate financing choices in terms of equity versus straight debt. My second objective is to examine some areas that have not been investigated in previous studies. For example, whether financial analysts also make optimistic earnings forecasts about debt-issuing companies before offerings, and whether analysts optimism toward equity issuers differs from their optimism towards debt issuers. Furthermore, I aim at investigating whether such difference may help explain the difference in long-run performance between equity issuers and debt issuers. My investigation thereby aims at providing direct evidence on some hypotheses found in previous studies. For examples, Loughran and Ritter (1995) argue that the long-run underperformance of equity issuers is evidence of investor overoptimism and managers opportunism to take advantage of such sentiments. Healy and Palepu (1993, 1995) hypothesize that investors perceptions of a firm are important to corporate managers expecting to issue public debt or equity. III. Hypotheses Theoretical studies show that information asymmetry should be related to firms financing choices. Because equity is more severely undervalued than debt when substantial information asymmetry exists about assets-in-place or future cash flows, Myers (1984), Myers and Majluf (1984), and Narayanan (1988) argue that firms prefer 22

PAGE 30

debt to equity financing when they seek external financing. Blazenko (1987) suggests that firm managers might exclusively use equity financing if information is symmetric. As information asymmetry intensifies managers will issue debt to signal firms high quality. These studies imply a negative relationship between probability of equity financing and extent of information asymmetry. Alternatively, Giammarino and Neave (1982), DeMeza and Webb (1987) show when firms differ in project risks and information asymmetry is about project risks, firms might issue equity rather than debt because debt financing incurs the adverse selection problem and equity financing does not. Debt issued by firms with riskier projects will be overvalued and firms issue debt when debt is overvalued. Expecting managerial opportunism of issuing overvalued debt, investors might refuse to purchase debt unless debt price is discounted. So at equilibrium equity financing dominates debt financing, which suggests that probability of equity financing will be positively related to extent of information asymmetry. The information asymmetry hypothesis predicts that the information asymmetry should affect firms financing choices in terms of equity and debt. But the sign is uncertain. Prior studies have found mixed evidence on impact of information asymmetry in firms financing choices without considering the information production problem (Klein and Belt (1994), Helwege and Liang (1996), Jung, Kim and Stulz (1996)). However, it is not known what the impact information asymmetry will have when controlling for endogeneity of information production. The Information production hypothesis argues that equity issuance induces more information production than debt issuance, and such difference in turn raises the probability of equity financing over debt financing. 23

PAGE 31

Fulghieri and Lukin (2001) assume that there are specialized outside investors who can generate information on a firms quality. Equity issues provide greater expected profits from information acquisition than debt issues because equity is a junior claim and is more information-sensitive than debt. As a result, equity issue announcements should encourage more information production than debt issue announcements. Managers of high-quality firms might issue equity rather than debt to raise the probability that the issue will succeed. Similarly, because public equity provides greater incentive to information production than public debt, Sunder (2002) argues that firms might choose to issue equity and assume the adverse selection cost today, expecting that more information will be produced, and therefore information asymmetry and financing costs will be reduced in the future. Seasoned public security issuance is a very important event in a companys history. So the demand for information should increase around security issue announcements. Naturally such events might attract more financial analysts to produce more information. I use change in analyst coverage post-issue analyst coverage minus pre-issue analyst coverage to proxy for the information production induced by security offerings. The greater the increase in analyst coverage after a security offering, the more information is produced. Because private information is more valuable when information asymmetry is high, firms with high levels of information asymmetry might attract more analysts. That is, information asymmetry and information production interact with each other. Consequently, the impact of information asymmetry or information production on financing choices might also depend on the interaction term between information production and information asymmetry. As a result, I also include such an interaction term to explain firms equity/debt choices in the probit model regression. 24

PAGE 32

IV. Data Selection and Variable Descriptions A. Data Selection Firms that issued straight debt or primary equity are selected from the Security Data Corporations (SDC) Global Financing Database for the period from January 1984 to December 2002. Convertible debt offerings, debt rollovers and debt offerings with maturity less than 1 year are excluded from the sample. Initial public equity offerings, preferred stock offerings, combined equity issues, secondary equity issues, rights offerings, unit offerings, withdrawn issues are also excluded. The initial sample includes 3,074 seasoned primary equity offerings and 12,289 straight debt offerings by U.S. companies. The final sample is formed using the following restrictions: 1 There are no security offerings before and during the one year period subsequent to the current issue. 2 12 month pre-issue daily stock return data is available from The Center of Research for Stock Prices (CRSP) using TSPRINT. 3 The number of analysts following the issuers (analyst coverage) must be available for the 12-month period before security offering and the 12-month period after the security offering from the Institutional Brokers' Estimate System (IBES) U.S. Summary History database. 4 Firms issuing primary equity and straight debt in the same year are excluded. Duplicate issues on the same day are excluded. 5 Firms in financial industry (SIC 6000-6999) and utility industry (SIC 4900-4949) are excluded. 25

PAGE 33

6 Accounting information on the fiscal year end before security offerings must be available from COMPUSTAT. The final sample includes 1,831 straight debt offerings and 1,375 primary equity offerings over 1984 to 2002 period. Insert Table 1 here B. Variable Descriptions and Methodology To measure the information asymmetry of security issuers, I use the reciprocal of analyst coverage, and the information asymmetry index. Unlike analyst coverage, which is inversely related to extent of information asymmetry, the reciprocal of analyst coverage and information asymmetry index directly reflects the extent of information asymmetry, thereby makes it more convenient to report and compare results on information asymmetry. 13 The information asymmetry index is the weighted average of the ranks of variables that have been used as information asymmetry proxies in prior empirical studies, including analyst coverage (ANN), log of total assets (LNTA), trading volume turnover (TURNOVER), market-to-book ratio (MBKR), research and development expenditures to total assets ratio (XRDTA), and the standard deviation of the residual returns (RESIDSD). 14 That is: ikKkkXRankKNASYIND111 The information asymmetry index ranges from 0 to 1. The greater the information asymmetry index, the greater the information asymmetry an issuer faces. Because ANN, 13 I also use the natural logarithm of 1 plus analyst coverage following prior literature, and the results are qualitatively the same. 14 The standard deviation of residual returns estimated by regressing a firms daily returns on value-weighted market index returns from 255 days before offering date until 46 days before the offering date. 26

PAGE 34

LNTA and TURNOVER are inversely associated with the extent of information asymmetry, their negative values are used when constructing the information asymmetry index. The information asymmetry index is constructed in a similar fashion as the liquidity index constructed in Butler, Grullon and Weston (2004). Fulghieri and Lukin (2001) and Sunder (2002) argue that when information production is endogenous firms might issue equity to encourage information production, which consequently reduces information asymmetry. For that reason, the change in analyst coverage is measured as the difference between the mean analyst coverage during the 12-month period before security offerings and the mean analyst coverage during the 12-month period after security offerings. I hypothesize that equity issues will lead to greater increases in analyst coverage than debt issues, and this in turn raises the probability of equity issues. A typical way to control for endogeneity of a variable is using a two-stage procedure. Following Pagan (1984), a two-step model is set as: eXzy (1) Wz (2) In the first stage, z is regressed against a matrix of instruments, W, to get the predicted value of z, In the second stage, y is regressed on and an independent variable matrix X. The two-step procedure has been widely used in empirical studies. Titman, Hovakimian, and Opler (2001) apply the two-step procedure to test whether firms make debt/equity choice to move toward their target leverage ratios or not. z z 27

PAGE 35

Ljungqvist, Marston and Wilhelm (2004) use the two-step procedure to examine whether analysts behavior influenced banks underwriting business. 15 Chaplinsky and Bayless (1991), Bayless (1994), Jung, Kim and Stulz (1996) and among others show that investors can predict the type of security offerings, and the predicted probability of security offerings affect market reactions to security issue announcements. I conjecture that outside investors can anticipate the type of external financing a firm will make, and different anticipated financing types have different impact on information production, consequently, such difference will affect firms financing choices. I use a three-step procedure to control for the predictability of security offerings and endogeneity of information production, because a change in analyst coverage might result from anticipation of the types of security to be issued. I first determine the probability of equity issues compared to debt issues using a probit model regression with the set of commonly used independent variables related to equity/debt choice. I then use the predicted probability of equity issues together with firm quality, size, and information asymmetry measures in an OLS model to explain the changes in analyst coverage. The predicted values (change in analyst coverage) and the residuals of the OLS model in the second step are used in the probit model to explain the equity/debt financing choice in the third step. The predicted change in analyst coverage proxies for the expected information production due to security offerings. The residual change in analyst coverage is the observed change in analyst coverage not explained by the financing choices decisions and firm quality and information asymmetry measures. 15 For econometric issues involved in two-step procedure, besides Pagan (1984), the reader may refer to Murphy and Topel (2002), Davidson, R. and J. G. MacKinnon (1993), and Greene (2000). 28

PAGE 36

I also use the typical two-stage procedure in the literature to control for the endogeneity of information production, ignoring the predictability of security offerings. I first use the following OLS model to explain change in analyst coverage respectively for debt and equity offerings. RESIDSDABCUMRETLNTAROAMBKRANNANNCH6543210 Then I use the predicted values of the above model to explain firms financing choices. Dechow, Hutton and Sloan (1996) find that analysts tend to make optimistic forecasts before seasoned equity offerings, and the optimism is displayed in analysts long-term EPS growth rate forecasts. Assuming that financial analysts opinions represent those of investors, investors optimism can be measured by analyst forecast errors (FE), the IBES revision ratio (REVR), and the long-term EPS forecast growth rate (LGTH). The greater the forecast errors, revision ratios and/or long-term EPS forecast growth rates, the stronger investors optimism is toward security issuers. Analysts forecasts errors are calculated as the difference between analysts consensus annual mean EPS forecasts ( itwF ) and the actual annual EPS (A it ) scaled by the absolute value of consensus mean EPS forecast ( itwititwFAFFE/)( ). Following Gibson, Safieddine, and Sonti (2003), the IBES revision ratio is calculated as the difference between the number of analysts making upward forecast revisions (#UPREV) and the number of analysts making downward forecast revisions (#DNREV) divided by the total number of analysts following the firm (#ANN). That is, Where i represents firm i, t stands for fiscal year t, and w denotes the month in which earnings forecasts are reported. Similar to )/(###itwitwitwitwANNDNREVUPREVREVR 29

PAGE 37

the construction of information asymmetry index, I construct the investor optimism index (OPTIM), which is an equal-weighted average of the ranks of analysts forecast error, the long-term growth forecast rat, and the IBES revision ratio. ikKkkiZRankKNOPTIM111 A probit model is used to investigate the impact of information production, information asymmetry, and investor optimism on a firms financing choice. Control variables are adopted according to previous studies, which find that information asymmetry, agency problems, bankruptcy costs, and tax considerations affect a firms debt-equity issue choice (see Baxter and Cragg 1972, Marsh 1982, Titman and Wessels (1988), Makie-Mason (1990), Bayless (1994), Jung, King and Stulz (1996), Hovakimian, Opler and Titman (2001), among others). Particularly, factors such as growth opportunities, R&D expenditure, pre-issue stock price run-up and stock market liquidity positively affect probability of equity issues compared with debt issues, whereas firm size, return on assets and tax payments are inversely associated with equity financing. Table 2 describes the variables used and reports the summary statistics for the sample of firms with complete information from SDC, IBES, CRSP and COMPUSTAT. Insert Table 2 here V. Empirical Results and Discussions A. Univariate Test Results As discussed previously, financial analysts play important roles in information generation and dissemination (Womack (1996), Branson, Guffey, and Pagach (1998), Barth and Hutton (2004). among others). The greater the number of analysts that follow a 30

PAGE 38

firm, the more information is produced and the lower the level of information asymmetry the firm faces. Security offerings are major events in corporate history. Firms offer new securities for different purposes. Some firms raise capital to undertake profitable investment opportunities, whereas some other firms issue securities to get money for empire building even though they do not have good projects to undertake. In the presence of information asymmetry, it is difficult for ordinary outside investors to distinguish the quality of different securities. As a result, the demand for information becomes stronger when new securities are offered. Financial analysts are experts in analyzing securities. Given the increase in demand for information on new securities, there should be an increase in analyst coverage after security offerings. Because equity is a junior and residual claim, it is more information sensitive than debts. Given the extent of information asymmetry, equity has greater gap between the market value and fundamental value than debt does. As a result, the potential payoffs of information production should be greater in the case of equity issues. Therefore, I expect to see greater increases in analyst coverage after equity issues than after straight debt issues. And this is shown in Figure 1. Insert Figure 1 here Figure 1 compares the average monthly number of analysts following equity issues with the average number of analysts following straight debt issuers from 36 months before the security-offering month until 36 months after the security-offering month. I use the change in analyst coverage to proxy for information production induced by security offerings. The change in analyst coverage is defined as the 12-month average analyst coverage after security offerings month minus the 12-month average analyst 31

PAGE 39

coverage before security offerings month. Table 3 compares the change in analyst coverage between debt-issuing and equity-issuing firms within different firm-quality quartiles. 16 I first divide sample firms into quartiles based on measures of firm quality. Firms within each quality quartile are further divided into debt group and equity group. Panel A of Table 3 reports the results by classifying the whole sample firms into quartiles based on pre-issue abnormal cumulative returns (ABCUMRET). For every ABCUMRET quartile, analyst coverage increases significantly more after equity issues than after debt issues. This supports the hypothesis that equity issues encourage more information production than debt issues. Furthermore, a change in analyst coverage tends to increase as the pre-issue abnormal return rises, indicating that information production is greater in the case of good-quality firms. This tendency is more obvious among equity-issuers than among debt-issuers. For example, in the lowest ABCUMRET quartile, analyst coverage slightly decreases after debt issues, whereas it increases by 0.7 after equity issues. And in the highest ABCUMRET quartile, analyst coverage increases by 0.6 after debt issues, which still is significantly much smaller than the 1.7 increases after equity issues. Panels B and C report results by classifying the sample into quartiles based on market-to-book ratio and return on assets, respectively. The results are similar to the results shown in Panel A, which suggest that equity issues encourage more information production than debt issues and information production tend to increase as firm quality improves. The latter conclusion is confirmed by results reported in Appendix A using different sample classifications. Sample firms are classified into debt and equity group first, and then firms in each financing group are further divided into quartiles based on firm quality measures. 16 There are eight groups after the division. Such classification displays significant changes in firm quality and allows enough sample firms in each group to compare the mean difference. 32

PAGE 40

Insert Table 3 here Prior studies find that firms facing high levels of information asymmetry are valued less and have higher costs of raising capital than firms with low levels of information asymmetry (DMello and Ferris 2000, Bowen, Chen and Cheng 2004, among others). Consequently, firms have incentives to reduce the extent of information asymmetry they face. To examine whether a) security issue announcements encourage more information generation among firms suffering from higher levels of information asymmetry than firms with lower information asymmetry, and b) equity issues also encourage more information production than debt issues, I classify firms into quartiles respectively according to the reciprocal of pre-issue 12-month average analyst coverage and the information asymmetry index. Panel A of Table 4 presents results using the inverse analyst coverage as a measure of information asymmetry. For debt-issuing firms, the change in analyst coverage increases significantly from .37 in quartile 1 to 0.94 in quartile 4. For equity-issuing firms, change in analyst coverage increases significantly from 0.71 in quartile 1 to 1.56 in quartile 4. These results indicate that information production increases as the extent of information asymmetry rises. Furthermore, in every quartile, the increases in analyst coverage after equity issues are significantly much more than the increases after debt issues, which supports the hypothesis that equity issues encourage more information asymmetry than debt issues. Panel B reports similar results when information asymmetry is measured by the information asymmetry index, show that equity financing encourages more information production than debt financing and information production tends to increases as information asymmetry intensifies. The 33

PAGE 41

latter conclusion is confirmed by results reported in Appendix B, which classifies the whole sample into debt and equity group first, and then firms in each financing group are further divided into quartiles based on information asymmetry measures. Insert Table 4 here Table 5 shows and compares the change in analyst coverage among eight groups classified by sorting on median values of firm quality measures (return on assets and pre-issue cumulative abnormal returns), median values of information asymmetry measures (inverse analyst coverage and information asymmetry index) and type of financing (debt versus equity). 17 These results suggest that better quality firms that face higher levels of information asymmetry might encourage greater amount of information production, and equity issues induce more information production than debt issues. Insert Table 5 here B. Multiple Regression Results Panel A of Table 6 reports the results of OLS regressions where the dependent variable is change in analyst coverage. EDDUM is the financing type indicator. EDDUM equals one if equity is issued and zero if debt is issued instead. Consistent with the univariate test results, all the coefficients of EDDUM are significantly positive at the 1 percent level, indicating that equity financing is related to a greater increase in information production than debt financing. The coefficient of EDDUM is 0.779, indicating that without controlling for anticipation of security issue type and holding other factors constant, equity financing raises analyst coverage by 0.779 more than debt 17 Such classification again is in order to show variations in the variable of interest across groups and obtain enough sample firms in each group for the statistic test. 34

PAGE 42

financing. Firm quality measures such as market-to-book ratio, return on assets, and pre-issue abnormal cumulative return all positively affect the change in analyst coverage. These results suggest that good-quality firms encourage more information production, which is also consistent with the univariate test results. Because Model 1 appears to be the best fit, I use it in subsequent analysis to explain change in analyst coverage. In the next step of my analysis I assume that firm managers have perfect foresight about the impact that a security offering has on analyst coverage. I use the change in analyst coverage to explain a firms financing choice. 18 Panel B of Table 6 reports the probit model regression results. The change in analyst coverage increases the probability of equity issues, and the effect is significant at the 1 percent level. This suggests that the greater information production associated with equity issues compared to debt issues increases the likelihood that equity financing will be used rather than debt financing. The significant negative impact of the inverse analyst coverage on probability of issuing equity suggests that probability of equity offering is reduced by the extent of information asymmetry. However, the coefficients of information asymmetry index are positively insignificant. Insert Table 6 here The assumption that mangers have perfect foresights as to the changes in analysts coverage that result from security offerings is quite unrealistic. As a result, I use a three-step procedure to control for the endogeneity of information production and firms financing choices. First, I use Model 1 in panel B of Table 6 to get EHAT, the predicted 18 In similar fashion, Jung, Kim and Stulz (1996) assume firm managers have perfect foresight on the post-issue stock market performance and use post-issue abnormal return to test the market timing hypothesis. 35

PAGE 43

probability of equity issues versus straight debt issues. 19 Second, I use the predicted probability obtained in the first-step, along with measures of firm quality, size, and information asymmetry to explain changes in analyst coverage, and to obtain the predicted change in analyst coverage (PCHANN) and the residual change in analyst coverage (RCHANN). 20 Third, I use the predicted change in analyst coverage and the residual change in analyst coverage estimated in Step 2, to explain firms financing choices considering the set of control variables used in step 1. In panel A of Table 7 the coefficient of EHAT is 1.82, which is significant at the 1 percent level. The result suggests that expectations that equity will be issued induce more information production than expectations that debt will be issued. The predicted change in analyst coverage is the predicted value of the OLS regression in Panel A of Table 7. It reflects the expected information production in anticipation of the type of securities to be issued. The residual change in analyst coverage is the residual of the OLS regression in Panel A of Table 7. The residual change in analyst coverage is the change in analyst coverage that is not explained by the predicted probability of equity issues and measures of firm quality and information asymmetry. It might proxy the real impact on information production that security offerings may have. The average predicted (residual) change in analyst coverage for 1,831 debt issues is 0.211 (.081), which is significantly lower than the level of 1.326 (0.108) for 1,375 equity issues. This difference is significant at the 1 percent level. 19 I also use Model 2 in panel B of Table 6 to get the predicted probability of equity issues versus debt issues, and the results are qualitatively the same. 20 The construction of residual change in analyst coverage is inspired by the construction of residual analyst coverage in Hong, Lim and Stein (2000). 36

PAGE 44

Panel B of Table 7 reports the results using the predicted change in analyst coverage as a proxy of information production. The results suggest that information asymmetry might reduce the probability of equity offerings, which support the pecking-order theory. However, the positive impact of an expected equity issue on information production also increases the likelihood of equity offerings over debt offerings. In Model 1 through Model 4 the coefficients of PCHANN exceed 0.22 and are significant at the 1 percent level. Even though the coefficient of PCHANN in Model 5 is .2156, it is non-significant. And the coefficient of the interaction term between ASYIND and PCHANN is 1.2225 and is significant at 1 percent level, suggesting that information production still has positive impact on the probability of equity issues over debt issues. The evidence indicates that the negative effect that information asymmetry has on probability of equity issues is mitigated by the anticipated information production. Model 4 in panel B of Table 7 ignores the interaction effect of information production and information asymmetry on financing choices. In this case the information asymmetry index has an insignificantly negative coefficient. After introducing the interaction term, the coefficient of information asymmetry index becomes significantly negative. Panel C of Table 7 shows the results using the residual change of analyst coverage as a measure of information production. The results are quite similar to those reported in Panel B. Insert Table 7 here The results using the two-stage procedure to control for the endogeneity of information production are reported in Appendix C. They are similar to the results using the three-step procedure, indicating that equity financing has greater impact on information production than debt financing, which thereby increases the likelihood of 37

PAGE 45

equity financing over debt financing. Information asymmetry reduces the probability of equity financing, but its impact is mitigated by information production. C. Information Production and Post-issue Stock Performance I have shown that raw analyst coverage increases much more after equity issue announcements than after debt issue announcements, and such a difference increases the likelihood of equity financing. I argue that such evidence supports my information production hypothesis that equity issues induce more information production than debt issues. Due to this effect some firms may choose to issue equity rather than debt to encourage information production and thereby reduce information asymmetry. If information production leads to a reduction information asymmetry, on would expect that firm valuation improves and that this effect is more significant among equity issuers than among debt issuers, because equity is more information sensitive than debt. Table 8 compares the 1-year, 2-year, and 3-year buy-and-hold abnormal returns between debt issuers and equity issuers within different quartiles of change in analyst coverage. The buy-and-hold abnormal return is calculated as: 100*]11[22tMttjtjtRRabcumretny Where n=1, 2 or 3, =252 if n=1, =504 if n=2, and =756 if n=3. R jt is firm js daily return since the second day of security offerings. R Mt is the markets daily return, which is measured by the NYSE/AMEX/NASDAQ value-weighted index return. Prior studies document long-run stock return underperformance of equity issuers (Spiess and Affleck-Graves 1995, Loughran and Ritter 1995, and among others), straight debt issuers (Spiess and Affleck-Graves 1999) and equity issuers have even worse post-issue stock 38

PAGE 46

market performance than debt issuers. Panel A, B and C of Table 8 respectively compare the post-issue one-year, two-year and three year buy-and-hold abnormal returns between equity issues and debt issues and between different analyst coverage change quartiles. The whole sample firms are first divided into quartiles based on change in analyst coverage. Then firms in each analyst coverage quartile are further classified as debt or equity group. The results indicate that post-issue stock market performance tends to improve as change in analyst coverage increases for both debt issuers and equity issuers. 21 Equity issuers with the highest increase in analyst coverage even beat the market by nearly 13 percent for the first year after issuance. For issuers with the highest increase in analyst coverage, equity issuers do not significantly underperform the market nor underperform debt issuers. For example, within the highest change in analyst coverage quartile, the two-year average abnormal return for equity issues is .93 percent, in contrast to -1.58 percent for debt issues. Both are insignificantly different from zero, meaning both equity issuers and debt issuers do not underperform the market. And the difference between debt issues and equity issues is insignificant. Panel D reports the multiple regression results of post-issue buy-and-hold 1-year, 2-year and 3-year abnormal returns on change in analyst coverage respectively for debt issues and equity issues. The coefficients of change in analyst coverage for equity issues all exceed 3.5 and are significant at 5 percent, indicating that holding other factors constant, one additional increase in analyst coverage after equity issues would increase the post-issue abnormal return by more than 3.5 percent. The coefficients of change in analyst coverage are also all positive for debt issues, but they are smaller in magnitude than those for equity issues. 21 Appendix D reports similar results using different sample classifications. Sample firms are first divided into equity and debt groups. Firms within each financing group are further divided into quartiles based on change in analyst coverage. 39

PAGE 47

Insert Table 8 here D. Robustness Tests D.1. Constructing and Using an Ex-ante Measure of Information Production The predicted change in analyst coverage used in the probit regression of Table 7, Panel A uses the whole sample firms and is based on the change in analyst coverage. Since the change in analyst coverage is defined as the 12-month post issue average analyst coverage minus the 12-month pre-issue average analyst coverage, it is an ex-post measure. In testing the impact of information production on firms financing choices, an ex-ante measure would be more appropriate. To get an ex-ante measure of information production, I first divide the whole 3,206 security offerings firms randomly into two groups with the same sample size: the base group and the test group. The base group contains 907 straight debt issues and 696 primary equity issues. The test group includes 924 straight debt offerings and 679 equity offerings. I then run the following probit model on the base group and test group respectively to get the predicted probability of equity issues versus straight debt issues respectively for the two groups. 22 )()1(Pr1211109876543210TURNOVERRESIDSDBETAABCUMRETROACFLOWTALTDTAXRDTATXTATFIXTALNTAMBKReddumob Panel A of Table 9 shows the probit regression results. These results are similar to each other and are also similar to Model 1 probit regression in Panel B of Table 7. The results show that the probability of equity offerings are positively associated with market-to-book ratio, R&D expenditures, pre-issue abnormal returns, stock market turnover as 22 I divide the whole sample randomly into two equal-sized group: base group and testgroup. I use parameter estimates on change in analyst coverage based on base group to predict change in analyst coverage for the test group, which is then used to explain the types of security offerings for the test group. 40

PAGE 48

well as standard deviation of residual returns. In contrast, the probability of equity issues is negatively related to firm size, tax payments, return on assets. These results are consistent with previous findings (see Titman and Wessles (1988), Junk, Kim and Stulz (1996), and among others). Next I regress the change in analyst coverage on the predicted probability of equity issues versus debt issues and other control variables for the base group as the following equation shows: RESIDSDABCUMRETLNTAROAMBKRANNEHATANNCH76543210 The OLS regression results for base group are reported in Panel B of Table 9. These results are also similar to results reported in Panel A of Table 7. The coefficient of EHAT is significantly positive at the1 percent level, indicating that information production is positively related to expect equity issues. I then multiply the coefficients estimated using the base group by the corresponding values of the independent variables of the test group. The product is the predicted change in analyst coverage in the test group, which is an ex-ante measure of information production anticipated before security offerings for firms in the test group, because it is constructed based on information before security offerings. Eventually I use this ex-ante measure on expected information production to explain the financing choices for the test group. The results are reported in panel C of Table 8. The results are very similar to the results based on the whole sample firms, indicating that expected information production raises the probability of equity offerings. However, although information asymmetry reduces the probability of equity offerings versus debt offerings, and the negative impact of information asymmetry on equity issues is mitigated by the positive impact of expected information production. Insert Table 9 here 41

PAGE 49

D.2. Investor Optimism, Post-issue stock performance, and Firms Financing Choices Previous studies (Lin and McNichols (1998), Michaely and Womack (1999), Dechow, Hutton and Sloan (2000) and Teoh and Wong (2002)) find that analysts are overoptimistic about a firms future prospects around equity issues. Therefore, they make overoptimistic long-term growth forecasts about equity-offering firms earnings, and analysts optimism may be one reason that SEO firms under-perform the market after issue. Spiess and Affleck-Graves (1995) and Loughran and Ritter (1995), among others document poor long-term stock price performance of equity issuers. Loughran and Ritter, in particular, argue that the long-run underperformance of equity issuers is evidence of investor overoptimism and managers opportunism to take advantage of such sentiments. McLaughlin, Safieddine, and Vasudevan (1998), and Spiess and Affleck-Graves (1999) document substantial long-run post-issue underperformance following bond issues. Healy and Palepu (1993, 1995) hypothesize that investors perceptions of a firm are important to corporate managers expecting to issue public debt or equity. It is therefore imperative to examine whether my prior results are reflecting analyst/investor optimism. In the next set of tests I analyze whether my results are robust after controlling for analyst optimism. Insert Table 10 Panel A of Table 10 compares analysts forecast errors (FE), analysts long-term growth forecast rates (LGTH), IBES revision ratio (REVR) and investor optimism (OPTIM), as well as changes in these variables between equity issues and debt issues. Investor optimism index (OPTIM) is an equal-weighted average of the rank of forecast error, long-term growth forecast rates and IBES revision ratio. OPTIM ranges from 0 to 42

PAGE 50

1. The greater the optimism index is, the more optimistic investors are toward security issuers. Before security offerings, equity issuers have significantly higher EPS growth forecast rates and higher IBES revision ratio than debt issuers have. Analyst forecast errors for debt issuers are insignificantly greater than those for equity issuers. The investor optimism index for equity issuers is 0.5704, which is significantly greater than the investor optimism index of 0.4558 for debt issuers. After security offerings, analyst forecast errors are increased for equity issuers but are reduced for debt issuers, and the differences in analyst forecast errors between equity issuers and debt issuers remain insignificant after security offerings. The post-issue long-term growth forecast for equity issuers increases a little whereas it decreases a little for debt issuers. For both types of issuers, the forecast revision ratios are significantly reduced after the security offerings. However, the investor optimism index for equity issuers increases from 0.5704 before offerings to 0.5912 after security offerings. In contrast, for debt issuers the investor optimism index decreases from 0.4588 to 0.4460. The changes in investor optimism for equity and debt issuers are both significant at the1 percent level. The results indicate that while overall investors are more optimistic about equity issuers than about debt issuers, investors become less optimistic after debt offerings but become more optimistic after equity offerings. Panel B of Table 10 reports the OLS regression results of post-issue buy-and-hold abnormal returns on pre-issue investor optimism index and other control variables. All the coefficients of the investor optimism index are significantly negative and have absolute values greater than 20. The results indicate that holding other factors constant, for every one unit increase (0.01) in pre-issue investor optimism index, the post-issue 43

PAGE 51

abnormal return would decrease by at least 0.20 percent. The evidence confirms arguments in Spiess and Affleck-Graves (1995), Loughran and Ritter (1995), Spiess and Affleck-Graves (1999) that the post-issue underperformance of security issuers may be attributed to the pre-issue investor optimism. Most importantly, consistent with my results in panel D of Table 9, the change in analyst coverage still increases the post-issue abnormal return significantly after controlling for the effect of investor optimism. Holding other factors constant, every additional analyst will enable equity issuers to outperform the market by at least 3 percent. Moreover, pre-issue analyst coverage is also positively associated with the post-issue abnormal return. Loughran and Ritter (1995) suggest that firm managers may take advantage of investors overoptimism on equity issuer. As a result, the pre-issue investor optimism index should raise the likelihood of equity issues over debt issues. Panel C of Table 10 shows the probit regression results using the pre-issue investor optimism to explain firms financing choices. However, all the coefficients of investor optimism are insignificantly negative, rejecting the conjecture that firm managers take advantage of investors overoptimism on equity issuers to issue equity. Consistent with my prior results, the predicted information production still increase the probability of equity issues and information asymmetry reduces the likelihood of equity issues, but its impact is mitigated by expected information production. In addition to the above robustness tests, I do the same analyses using the 6-month average analyst coverage instead of the 12-month analyst coverage, or the relative change in analyst coverage the difference between post-issue average analyst coverage and pre-issue average divided by the pre-issue average analyst coverage, and controlling for 44

PAGE 52

the impacts that IPO and underwriter switching may have on analyst coverage (see Rajan, and Servaes (1997), Krigman and Womack (2001)). The results are qualitatively very similar, indicating that equity financing encourages more information production than debt financing and the greater impact on information production of equity financing in turn raises the probability of equity financing over debt financing. VI. Summary and Conclusions Most studies on firms financing choices assume that information asymmetry is exogenous and constant (see Myers and Majluf (1984), Blazenko (1987), Narayanan (1988) and among others), thereby conclude that firms would prefer debt financing to equity financing when information asymmetry exists between firm insiders and outsiders. Fulghiery and Lukin (2001) argue that because equity is more information sensitive than debt, issuing equity might provide greater payoffs to outside specialized investors information production activities. As a result, good-quality firms that face severe information asymmetry might issue equity rather than debt to induce information production, thereby increasing the demand for their equity and raising the probability that the issue will be successful. Sunder (2002) argues that firms that face high levels of information asymmetry might choose to issue equity first to encourage information production and increase the informativeness of stock prices, expecting that in the future the information contained in stock prices will spill over into prices of other traded securities. As a result, firms future borrowing costs will decrease with increases in information production in the stock market. 45

PAGE 53

In this study, I use changes in analyst coverage to proxy for information production induced by security issue announcements, and the inverse of analyst coverage and an information asymmetry index to measure information asymmetry. I investigate whether equity issue announcements encourage more information production than debt issue announcements, and the impacts that information production, information asymmetry and investor optimism have on firms financing choices. I find that during the 12-month period after security offerings, firms having better quality and facing higher levels of information asymmetry tend to have greater increases in analyst coverage relative to the pre-issue 12-month period. Furthermore, the increases are more significant among equity issues than among debt issues. I also find that firms that exhibit greater increases in analyst coverage are more likely to issue equity. The latter results are robust when I control for the endogeneity of information production. By using the alternative measures on information asymmetry and controlling for endogeneity of information production, I provide evidence that information asymmetry reduces the probability of equity issues, which confirms the pecking order theory. The impact of information asymmetry also depends on its interaction with expected information production and is reduced consequently. My results provide the first empirical evidence on the argument in Fulghiery and Lukin (2001) and Sunder (2002) that some firms facing information asymmetry may choose to issue equity rather than debt when information production is endogenous. In addition, I find that firms within the greatest increase in analyst coverage (thereby the greatest reduction in information asymmetry) beat the market within one year after offerings, and they do not significantly underperform the market three years 46

PAGE 54

after the offerings. Further, within this group of firms, equity issuers do not significantly underperform debt issuers. I also examine how investor optimism varies surrounding security offerings and whether the previous results remain robust controlling for investor optimism. Investor optimism is measured by the analyst forecast error, the long-term earnings growth forecast rate, the forecast revision ratio and investor optimism index. I find that investors become less (more) optimistic after debt (equity) offerings and investors are more optimistic about equity issuers than about debt issuers both before and after security offerings. Additionally, I present direct evidence that the post-issue long-run underperformance of security offerings can be attributed to pre-issue investor optimism to some extent. However, pre-issue investor optimism fails to explain firms financing choices. If analyst coverage improves firm valuation, one would observe that the offer price of seasoned equity offerings (SEOs) should be positively related to analyst coverage. Theories provide no consensus on the effect of divergence of opinion on security prices (Miller (1977), and Varian (1985)). No empirical studies investigate what determine the offering price in SEOs. It will be interesting to examine whether and how analyst coverage and analysts forecast properties affect the offering prices. 47

PAGE 55

Essay 2 A Unified Analysis on Financing Choices and Offering Costs I. Introduction Previous studies have found that stock market prices drop significantly from two to three percent when firms announce a seasoned equity issue (see Asquith and Mullins (1986), Masulis (1986), Smith (1986), Jung, Kim and Stulz (1996) and among others), but the event study evidence is mixed on stock market reaction to debt issue announcements (Dann and Mikkelson (1984), Johnson (1995), Manuel, Brooks and Schadler (1993), Howton, Howton and Perfect (1998) and among others). 23 Nevertheless, it is a stylized fact that the markets reaction to equity issue is more negative than to debt-issue announcements. For example, Smith (1986) reports that the announcement day stock market reaction to equity issuance is about 2.88 percent more negative than the reaction to debt issuance. Bayless (1994) finds that the issue costs for equity would be 35.4 to 48.6 percent greater than those for a similar debt issue using the Asquith-Mullin (1986) measure. 24 Lee, Lochhead, Ritter and Zhao (1996) report that the total direct costs of seasoned equity issues amount to 7.11 percent of total proceeds on average, whereas the total direct costs of debt issues represent 2.24 percent of total proceeds. These 23 Dann and Mikkelson (1984), Ecobo(1986), Mikkelson and Partch (1986), Shyan-Sunder (1991) find that market response to straight debt issue announcements are insignificantly different from zero. Johnson (1995) finds significantly positive stock price reactions to debt issue announcements for low-dividend payout firms. Manuel, Brooks and Schadler (1993) document significantly negative market reactions to debt-issue announcements closely preceding dividend and earnings announcements. Howton, Howton and Perfect (1998) find a significant negative market reaction of .387 on the announcement date of straight debt issue without conditioning on dividend or earnings announcements. 24 Asquith and Mullins (1986) present a measure on how much the market value of a security-issuing firm is lost for every dollar obtained through security offerings. The Asquith-Mullin measure equals the product of market value of equity and the two-day cumulative abnormal return divided by the gross proceeds. 48

PAGE 56

empirical findings show that in general equity financing is much more costly than debt financing. It is puzzling that firms still use equity financing even though equity financing costs are much more than debt financing. The primary goal of this study is to examine whether firms choose a security type that minimizes their total offering costs under the presence of factors that may affect firms financing choices and offering costs such as access to capital markets, information asymmetry, firm and issue characteristics. Total offering costs are measured as a percentage of total proceeds raised. It is equal to the direct offering costs plus the indirect offering costs. Direct offering costs are the sum of investment banking fees, accounting, legal and other expenses as a percentage of total proceeds raised. Indirect offering costs are equal to the product of minus cumulative abnormal return and firm value on the day before security issue announcement divided by the total proceeds raised. If the market reacts positively to a firms security issue announcement, firm value will increase, which is equivalent to reduction in the offerings costs. Smith (1986) states,Maximizing behavior by firms implies that in voluntary transactions such as security sales, the firm should structure the transaction to yield the highest possible value of the firm (page 6). Because security issues are voluntary actions at the discretion of firm managers, they are subject to self-selection bias. As a result, the observed sample of firms that have chosen one type of financing over the other are not random, consequently the OLS estimates are biased and inconsistent, because the expected mean value of the error term is no longer zero. 25 This study adopts Heckmans 25 Past studies in finance have found that self-selection exists when: (1) IPO companies choose whether to include warrants in their offerings or not (Dunbar, 1995), (2) when firms choose their issuance procedure (Smith 1987), (3) when IPO firms choose their contract format (Francis et al. 1999). For example, Dunbar (1995) controls for the self-selection bias and finds that underpricing and total offering costs are reduced 49

PAGE 57

(1979) two-step procedure in order to jointly investigate the debt-equity choice and offerings costs. This framework allows estimation of what the offering costs would have been if the alternative security had been issued and thereby enables an examination of whether firms choose a less costly financing method. Another goal of this study is to shed new light on the determinants of both the market reaction and the costs of raising external capital. 26 A number of studies have examined the determinants of the (cumulative) abnormal returns on (around) the announcement date (period) but have produced mixed findings cross-sectionally (Dierkens (1991), DMello and Ferris (2000), Jung, King and Stulz (1996), among others). In addition, the empirical findings are limited to the determinants of offering costs. This study uses seasoned security offerings by U.S. publicly-listed firms from 1984 to 2002 in order to investigate: 1) whether self-selection exists in firms security offering decisions, and, 2) whether issuing firms choose the security type that minimizes their offering costs. 27 This paper first controls the selection bias in security offerings and considers both direct and indirect offering costs together, aiming at explaining the puzzle why firms issue equity even though equity offering costs are much higher than debt issuing costs. The study shows that the reason is that equity issuers would have not been for firms including warrants as underwriter compensation, thus resolving the puzzle documented in (Barry, Muscarella, and Vetsuypens (1991) who report that warrants are commonly used even though the total costs including warrants as underwriter compensation in IPOs are higher than the costs without warrants. 26 Following Lee, Lochhead, Ritter and Zhao (1996), total direct costs are measured as the sum of gross spreads (investment banking fees) and other direct expenses (registration fee, and printing, legal and auditing costs) as percentage of total proceeds. Lee et al. report that the total direct costs for SEOs average 7.11 percent (ranging from 3.15 to 13.28 depending on proceeds, gross spread averages 5.44 percent of total proceeds raised); the total direct costs for straight debt issues average 2.24 percent (ranging from 1.32 to 4.39 percent, gross spread averages 1.62 percent of total proceeds). Dunbar (1995) examines determinants of total direct costs among IPO firms. The empirical studies on determinants of total direct costs of seasoned security offerings, especially for straight debt issues are quite limited. 27 Since some firms issue equity only because they dont have access to debt market, to avoid such bias, I following Denis and Mihov (2003), in assuming that a publicly-listed firm will have accesses to debt market if the issuers book value of total assets exceeds $100 million, or the minimum issue size exceeds $50 million, and use these criteria to screen out firms without access to debt market. 50

PAGE 58

able to save offering costs if they had offered debt. In addition, this study throws new light on the determinants of offering costs. Finally, the study adds new evidence to the literature on market reaction to straight debt issues. I find that self-selection bias exists when firms choose the external financing method. On average, the market reaction to straight debt issues is insignificantly positive. If debt-issuers had issued equity instead, the direct offering costs would have increased by 2.45 percent and the market reaction would have become negative. Indirect offerings costs would have increased by 9.711 percent and the total offering costs would have risen by 12.65 percent. In contrast, the market reaction to equity issues is significantly negative. If equity-issuers had offered straight debt, the direct offering costs would have decreased by 3.37 percent and the market reaction would have become less negative. However, indirect offering costs would have gone up by 3.58 percent, and there would have been no significant decrease both economically and statistically in total offerings costs if equity issuers had offered straight debt. The results indicate that both straight debt issuers and equity issuers choose a security type that costs less. The remainder of this paper is organized as follows. Section II reviews related literature. Section III describes data construction. Section IV presents the methodology and hypotheses. Section V reports and discusses empirical results. Finally section VI concludes. 51

PAGE 59

II. Literature Review A. Theories on Market Reaction to Security Issue Announcements In the signaling model of Ross (1977), managers know the true quality of a firm, while investors do not. High (low) quality firms have high (low) profitability. Firm managers benefit if the firms securities are valued more and are penalized if the firm goes bankrupt. Bankruptcy risk rises as the amount of debt issued by the firm increases. In equilibrium, high quality firms issue more debt to signal their good quality, while low-quality firms cannot mimic high-quality firms because low-quality firms do not have enough cash flow to back their debt. Consequently, the market reacts positively (negatively) to leverage increasing (decreasing) security issue announcements. Myers and Majluf (1984) show that managers with superior information issue equity only when their firms stock is overvalued, otherwise they would rather give up a positive NPV project. The market reacts negatively to equity issue announcements, but less negatively to debt issue announcements. Cooney and Kalay (1993) extend Myers and Majlufs (1984) analysis by allowing for existence of negative NPV projects and information asymmetry on growth opportunities in addition to information asymmetry on assets-in-place. They conclude that not all equity issues convey bad news about firm value as firms with highly positive NPV projects might still issue undervalued equity to avoid giving up the projects. Thereby market reaction to equity issue can be positive if investors anticipate that the issuing firm is considering a net present value project. In the Miller and Rock (1985) model, changes in outside financing are a signal to investors of opposite changes in firms current earnings. The market reacts negatively to 52

PAGE 60

large unexpected security issue announcements since investors infer that the security-issuing firm cannot generate enough earnings to finance its planned investments. B. Empirical Findings on Market Reactions to Security Issue Announcements In general, empirical findings of studies examining market reactions to security issue announcements are consistent with the pecking-order model and the signaling models (e.g., Ross (1977)), in that the market is found to react more negatively to equity issues than to debt issues. Asquith and Mullins (1986), Masulis (1986), Smith (1986), Jung, Kim and Stulz (1996), among others, document a significantly negative market response to equity issue announcements. However, the findings on the markets reaction to debt issue announcements are mixed. Dann and Mikkelson (1984), Ecobo(1986) and, Mikkelson and Partch (1986) find that market responses to straight debt issue announcements are insignificantly different from zero. Johnson (1995) finds significantly positive stock price reactions to debt issue announcements for low-dividend payout firms. Manuel, Brooks and Schadler (1993) document significantly negative market reactions to debt-issue announcements closely preceding dividend and earnings announcements. Howton, Howton and Perfect (1998) find a significant negative market reaction of .387 percent on the announcement date of straight debt issues without conditioning on dividend or earnings announcements. Overall market reaction to equity issues is more negative than to debt-issue announcements. For example, Smith (1986) reports that the announcement day stock market reaction to equity issuance news is about 2.88 percent more negative than the reaction to debt issuance. Bayless (1994) controls for the predictability of security type and finds that the reaction to a first-time seasoned equity 53

PAGE 61

issue is more than 4.15 percent more negative than the reaction to debt issue. He adjusts the pre-issue market value and gross proceeds using the Asquith-Mullin (1986) measure and finds that the issuing costs for equity issue are 35.4 to 48.6 percent greater than those for a similar debt issue for his hypothesized average firm with a market equity value of $943.26 million and issue size of $80.54 million. However, Bayless (1994) considers the indirect offerings costs only, and does not control for the selection bias and estimate what the indirect offering costs would have been if the same firm had issued the alternative security type. A number of studies apply multiple regression analyses to investigate the determinants of stock market responses to equity or debt issue announcement and produce mixed results. Masulis and Korwar (1986) find that the two-day announcement period return is negatively (positively) associated with the cumulative stock (market) return over the 60 days prior to the offering announcement date for a sample of 301 seasoned equity issues of industrial firms. However, all the other independent variables such as proportional change in outstanding shares of common stock, offering induced leverage change, and stock return variance over the 60 days preceding the offering announcement lack explanatory power. Dierkens (1991) finds the market-adjusted abnormal return at the equity issue announcements is negatively related to measures of information asymmetry such as the residual standard deviation of daily stock returns and the number of pre-issue announcements listed in the Wall Street Journal Index (WSJI). The relative number of shares to be issued has no impact on the abnormal return, whereas the ratio of market value of equity to book value of equity has a significantly positive effect. Similar to Dierkens (1991), DMello and Ferris (2000) find that the three-day 54

PAGE 62

announcement-period returns of SEO firms are related to measures of information asymmetry between firm insiders and outsiders. Specifically, it is positively (negatively) related to analyst coverage (analyst forecast dispersion). However, variables such as the 11-month pre-issue excess return, the fraction of shares issued relative to the number of shares outstanding, the firm size, the market-to-book ratio, as well as the hot period dummy have no impact on the announcement period returns of SEO firms. Jung, King and Stulz (1996) show that firms with the most valuable investment opportunities (reflected by market-to-book ratio, the ratio of firm market value to total assets) do not experience adverse stock returns when they issue equity. However, firms, which issue equity against their type, register an extremely significant drop in their stock price when they issue. 28 Besides cash flow to asset ratio, they find that market to-book ratio is the only variable that significantly affects the two-day cumulative abnormal return on equity issuers; other variables such as stock return volatility, past cumulative excess return, cash, tax payments, long-term debt to total assets ratios, total assets, leading indicators, post-issue cumulative return, proceeds to market value of common stock, as well as log of proceeds all have no impact. Jung, King and Stulz (1996) also run similar regressions for debt-issuers, and find that the only variable that is significant is the ratio of proceeds to pre-issue market value of common stock. Howton et al. (1998) find that the standardized abnormal returns on the announcement date of straight debt issues is negatively related to firms cash flow to book value of total assets, and positively related to the ratio of firms total debt to market value of assets, and there is no impact of the ratio of gross proceeds from debt issue divided by market value of assets. Mikkelson and 28 Jung, Kim and Stulz (1996) consider an equity-issuing company as against its type if the company is estimated to have greater probability to issue debt than to issue equity. 55

PAGE 63

Partch (1986) find that the type of security issued is the only determinant of the price response. Offerings characteristics such as the net amount of new financing, relative size, and the quality rating of debt issues have no impact on the stock market response. These studies do not control for the selection bias. Some studies control for the impact that investors anticipation of a security issue type has on the cumulative abnormal returns upon security issue announcements. Bayless and Chaplinsky (1991) investigate how the market reactions to debt or equity offers are influenced by investors expectation as to the type of security to be issued. They document a significantly positive 1 percent announcement day return for debt issues undertaken by firms that are expected to issue equity. In contrast, the market reacts negatively to equity issue announcements of firms that are expected to issue debt. For both the equity and debt issues, the two-day cumulative abnormal returns are negatively correlated with the probability of issue as perceived by investors before the offering announcements. However, they find that issue size has no impact on market reactions. Bayless (1994) finds a more significant effect of security type on the two-day cumulative abnormal return after controlling for the predictability of security type, as relative to not controlling for the predictability of the issue type. C. Literature on Self-selection Bias Heckman (1979) discusses the bias that results from using non-randomly selected samples to estimate behavioral relationships, such as migration, manpower training or unionism. He developed the two-stage estimator that enables analysts to utilize simple regression methods to estimate behavioral functions by least square methods. 56

PAGE 64

Maddala (1991) presents an in-depth review and discussion on the methodology of models involving qualitative and limited-dependent variables, including the self-selection model. Lee (1978) applies the self-selection model to explain relationship between wage rates and unionism. The self-selection model has been applied in several finance studies. Smith (1987) investigates the choice of issuance procedure and the cost of competitive underwriting and negotiated underwriting. He shows that failure to correct for a selectivity bias in the choice of issuance procedure in studies prior to his partly leads to findings that firms choose the negotiated issuance procedure over the competitive one despite its apparently higher net interest cost. After controlling for the selectivity bias, Smith (1987) finds that firms in fact choose the issuance procedure that minimizes the net interest cost, and therefore, their choices are consistent with shareholder wealth maximization hypothesis. Using models to correct for self-selection bias, Dunbar (1995) and Ng and Smith (1996) investigate the use of warrants to compensate underwriters in IPO firms and SEO firms respectively. Their results suggest that issuers choose compensation contracts to minimize the total offering costs, in contrast to findings ignoring selection-bias prior to theirs, which show that warrant use increased the issuers offering costs. Fang (2004) examines the relationship between investment bank reputation and the price and quality of bond underwriting service. He finds that after controlling for the selection bias, more repuTable banks actually enjoy a reputation premium, not a reputation discount, which was found in previous studies without controlling for the selection bias. 57

PAGE 65

III. Data Construction U.S. firms offering straight debt or primary equity for the period from January 1984 to December 2002 are selected from the Security Data Company (SDC) global new issues database. Specifically, the final sample is formed using the following restrictions: 1 Debt offerings must mature more than one year after issue, while convertible debt offerings and debt rollovers are excluded. 2 Initial public offerings (IPOs), preferred stock offerings, combined issues of debt and equity, rights offerings, best-efforts offerings, secondary issues, unit offerings, and REIT issues are excluded. 3 Firms issuing primary equities and straight debts in the same year are excluded. Duplicate issues on the same day are excluded. 4 (-255, -31) pre-issue daily stock return data are available from the Center for Research in Security Prices (CRSP). 5 Firms in the financial (sic 6000-6999) and utility industry (sic 4900-4999) are excluded. 6 Accounting data at fiscal year end prior to security issue announcements are available from research insight. 7 The number of analysts following the issuers (analyst coverage) must be available for the 12-month period before security offering and the 12-month period after the security offering from the Institutional Brokers' Estimate System (IBES) U.S. Summary History database. 8 Firms with proceeds raised below $50 million are excluded. 29 29 Following Denis and Mihov (2003), firms with issue size below $50 Millions are assumed to have no access to debt market. 58

PAGE 66

9 Firms with indirect offering costs in the top (bottom) two percentiles are excluded. 30 The final sample includes 1,012 primary seasoned equity issues and 1,533 straight debt issues. Table 11 reports the sample distribution over 1984-2002 period. Insert Table 11 here IV. Model Building, Hypotheses and Variable Descriptions Smith (1986) remarks, Maximizing behavior by firms implies that in voluntary transactions such as security sales, the firm should structure the transaction to yield the highest possible value of the firm (page 6). This study provides a test of two alternative hypotheses. The managerial agency hypothesis predicts that managers choose a security issue that benefits themselves, but is at the expenses of shareholders. The cost minimization hypothesis, oppositely, predicts that managers choose the security issue that minimizes the offering costs. A. Model Building The following model is built in similar fashion to Lee (1978) and Fang (2004). iiiXED* (1) Where ED i is a latent variable about the financing choice, which is unobservable. Suppose equity is issued if ED i >0, and straight debt is issued if ED i 0. We can only observe whether a firm issues equity or straight debt. Let ED=1 if equity is issued and 30 The indirect costs for firms in these percentiles are quite unreasonable. 59

PAGE 67

ED=0 if straight debt is issued instead. Then ED i = 1 if ED i >0 and ED i = 0 if ED i 0. The offering costs are the function of a vector of independent variables as the following: eiieeiZC (2) diiddiZC (3) Where C ei is the offering costs when firm i issues equity, C di is the offering costs if the same firm i issues straight debt instead. The offering costs are measured by direct costs, indirect costs as well as total offering costs. Direct offering costs are the sum of investment banking fees, accounting, legal, printing and miscellaneous fees as percentage of principal amount raised). 31 Following Asquith and Mullins (1986) and Bayless (1994), I compute the indirect offering costs as the impact of the market reaction on firm value relative to the issue size. It measures how much the market value of the firm is lost for every dollar obtained through the security issue, thereby is a better measure of the indirect cost of security issue and is comparable with direct offering costs since they all denoted as percentage of gross proceeds raised ( iiideiPROCEEDSCARMVEINDCP*/, ). Where INDC i,e INDC i,d are indirect cost of firm is equity issue or debt issue announcement. MVE i is firm is market value of equity the day before issue announcement, CAR i is the cumulative abnormal return during the announcement period, and PROCEEDS i is the gross proceeds from the issue. For firms with positive (negative) cumulative abnormal returns, firm value increases (decrease) upon security issue announcement, which reduces (increases) the offering costs. That is why I use -CAR i in calculating indirect offering costs. The total offering costs are calculated as direct offering costs plus indirect offering 31 Because SDC does not report values on accounting, legal, printing and other expenses for many sample firms, I assume these values are zero if they are missing. 60

PAGE 68

costs. For each issuer we only observe either C ei or C di depending on the value of ED i If there is correlation between i ei and di then the expected values of the error terms will no longer be zero. The expected value of the error term for equity issuers and debt-issuers are respectively as the following: iieiieiiieiieiXXXXXEEDE 1*1),cov(|0| iidiidiiidiidiXXXXXEEDE1),cov(|0|1* I assume that the errors ( i ei di ) have a trivariate normal distribution with mean vector zero and covariance matrix as: 1 e1 d1 COV (, e d ) = e ed d Var( i ) = 1 because ED i is observed only as a dichotomous indicator. We cannot estimate ed the covariance between ei and di because we do not observe C ei and C di simultaneously for the same firm at a point of time. z and z are the density function and cumulative distribution function respectively of the standard normal distribution. iiXX is the Inverse Mills Ratio (IMR) for firms issuing equity. For firms issuing straight debt, The IMR is iiXX1 The OLS estimates from equation (2), (3) are inconsistent because they miss an important term, the Inverse Mills Ratio (IMR). Consistent estimates on parameters in (2) and (3) are obtained through 61

PAGE 69

Heckman (1979) two-stage method: First, equation (1) is estimated by a probit regression to obtain the Inverse Mills Ratio (IMR) terms. Second, the IMRs are added into equations (2) and (3) to get equations (4) and (5). After running OLS on equations (4) and (5) respectively, consistent estimates on the determinants of offering costs are obtained. eiiieieeiXXZC1 (4) diiididdiXXZC11 (5) Where e d are the selectivity bias adjusted residual errors, E[ e | ED i = 1] = 0, and E[ d | ED i = 0] = 0. The offering costs are estimated as the following if the alternative security type had been issued. For example, if an equity-issuer had issued straight debt instead, the offering costs of the debt issue would have been: 0|0|*iidiididiXZEEDCE iidiidXEZ | iiidiidXXZ),cov( (6) B. Hypotheses I test two hypotheses: Offering costs minimization hypothesis: firms choose to issue a security that minimizes the offering costs, which is to their shareholders best interests. 62

PAGE 70

If this hypothesis is true, then for equity (debt) issuing firms, the offering costs would have been higher if they had issued the alternative security type. Alternatively, the Managerial agency hypothesis argues that firm managers choose to issue a security type to maximize their own benefits, even this increases the offering costs. On the one hand, managers might be unwilling to issue new debt, because issuing new debt raises the total risk of share ownership (Blazenko (1987)), increases bankruptcy risk and consequential dismissal of managers (Zwiebel (1996)), or reduces resources under management control (Jensen (1986)). Zwiebel (1996) shows that managers target leverage level is lower than the optimal leverage level that maximizes a firms value. On the other hand managers might be unwilling to issue equity because equity issuance will dilute managerial ownership of the firms; or because issuing equity will reduce earnings-per-share. The latter might result in lower managerial compensation, because managerial compensation is often positively related to earnings per share (Bartov, Givoly and Hayn (2002), among others). Thus, agency reason may also explain why managers may choose to issue a particular security even though such action would incur greater offerings costs. C. Variable Descriptions I use the inverse of average analyst coverage over the 12-month period before security offerings to measure information asymmetry, and construct the predicted change in analyst coverage to measure expected information production in anticipation of security offerings. The inverse analyst coverage, predicted change of analyst coverage, and the interaction term between these two variables are used to explain firms financing 63

PAGE 71

choices. Following previous studies Jung, Kim and Stulz (1996), Opler and Tittman (2002) and among others, I also include other independent variables to explain the financing choice in equation (1) where the dependent variable equals one if equity is issued and equals zero otherwise: PCHANN the predicted change in analyst coverage, a measure on expected information production induced by security offerings. The construction of PCHANN is described in Appendix E. PCHANN is the predicted value of the following OLS model: RESIDSDABCUMRETLNTAROAMBKRANNEHATANNCH76543210 Where EHAT is the predicted value of the following probit model: )()1(Pr1211109876543210TURNOVERRESIDSDBETAABCUMRETROACFLOWTALTDTAXRDTATXTTAFIXTALNTAMBKReddumob Equity financing may induce more information production than debt financing, which in turn raise the probability of equity financing. INVANN the inverse of the pre-issue 12-month average analyst coverage. A measure on information asymmetry. Firms facing higher levels of information asymmetry may be less likely to issue equity. PCHANN*INVANN the product of PCHANN and INVANN. The negative impact of information asymmetry on probability of equity financing may be mitigated by the expected information production. LNTA the logarithm of book value of total assets. Large firms will be less likely to issue equity. 64

PAGE 72

FIXTA the ratio of net property, plant and equipment to book value of total assets. Firms with greater fixed asset ratio have more collateral to borrow and thereby may issue more debt. Alternatively, such firms face lower levels of information asymmetry and will therefore use more equity. LTDTA the ratio of long-term debt to total assets. The higher the long-term debt ratio, the higher the bankruptcy risks, therefore the greater probability to issue equity. TXTTA the ratio of tax payments to total assets. The more tax payments, the better the debts tax shield function and the smaller the probability to issue equity. CFLTA the ratio of cash flow to total assets. Jensen (1986) shows that leverage reduces resources at mangers discretion. Managers in firms with greater cash flows may issue debt to show their willingness to be subject to monitoring from their creditors, or they are unwilling to do so but issue equity to have more resources under their disposal. So the sign of CFLTA on probability of equity issue is uncertain. XRDTA the ratio of R&D expenses to total assets. XRDTA may reflect a firms intangible assets or growth opportunities. It is expected to enhance the chance of equity financing. MBKR the market-to-book ratio, defined as (market value of equity + book value of debt)/book value of total assets. Firms with higher market to book ratio may have more growth opportunities, therefore may be more likely to issue equity. ROA return on assets, firms have higher ROA are more profitable and have higher debt capacity, thus lower probability to issue equity. 65

PAGE 73

ABCUMRET the difference between a firms cumulative daily returns and the markets cumulative returns over (-255, -46) period. The market return is CRSP value-weighted NYSE/AMEX/NASDOQ index return. Firms with better pre-issue stock market performance are more likely to be overvalued and thus more likely to issue equity. BETA the coefficient of market returns by regressing firms daily stock returns on market returns over (-255, -46) period. RESIDSD the standard deviation of the residual returns from the market model, which regresses firms daily stock returns on market returns over (-255, -46) period. TURNOVER the average ratio of total trading volume to shares outstanding over [-252, -1] period. The determinants on offering costs and cumulative abnormal returns are adopted according to previous studies such as Dunbar (1995), Butler, Grullon and Weston (2004), DMello and Ferris (2000) and among others, many of them are described above, so I do not specify any more. V. Empirical Results and Discussions A. Summary Statistics and Characteristics Comparison Table 12 compares the offering costs and firm characteristics between straight debt issues and common equity issues. It shows that debt issues have much lower offering costs than equity issues. For example, the average total offering cost for debt issues is only 5.63 percent of the total proceeds raised, while it averages 20.99 percent of total proceeds raised for equity issues. Consistent with prior studies, debt-issuers are 66

PAGE 74

much larger than equity issuers, raise greater amount of capital and have higher fixed assets ratio than equity issuers. Equity issuers have higher market-to-book ratio, greater expenses on research and development, greater stock return volatility and better pre-issue stock market performance than debt-issuers. In contrast, straight-debt issuers have higher tax payment, more cash flows, greater amount of long-term debts and better profitability than equity issuers. These differences are all significant at 1 percent level. Insert Table 12 here B. Determinants of Financing Choice Table 13 reports the probit regression results of the first-step in Heckman (1979) two-step procedure. Consistent with Jung, Kim and Stulz (1996), firms with smaller asset size, less tax payment, greater standard deviation of residual returns, and better stock market performance are more likely to issue equity than debt. The coefficient of PCHANN is 0.364, significant at the 1 percent level and the coefficient of INVANN is 2.414, still significant at the 1 percent level. And the interaction term between them is 0.861, significant at the 5 percent level. The results indicate that information asymmetry reduces the probability of equity issues over debt issues. But the negative impact on equity issue probability is mitigated by the positive impact of information production. The evidence not only supports Myers and Majluf (1984) that information asymmetry might reduce the probability of equity financing compared with debt financing, but also supports the argument in Fulghiery and Lukin(2001) that when information production is endogenous, firms facing high levels of information asymmetry may issue equity rather than debt. 67

PAGE 75

Insert Table 13 here C. Determinants on Offering Costs and Market Reaction C.1. Results on Direct Offering Costs Panel A of Table 14 reports the OLS regression results with robust variance, for models where the dependent variable is direct offering costs. The existence of selection bias can be examined in two ways. First, compare the coefficients of the OLS model adjusting for self-selection bias with OLS model without correcting the selection-bias. There will be substantial differences if selection bias exists. In deed, Panel A shows substantial differences in determinants of direct offering costs between results obtained from models controlling for selection bias and from models that do not. For example, the coefficient of ABCUMRET, the pre-issue cumulative abnormal return, is .026 for straight debt issues without controlling for selection bias, and is both economically and statistically insignificant. It becomes .145 and is significant at the 10 percent level after controlling for selectivity. Second, self-selection can be detected by examining whether the inverse mills ratio is significant or not. For both straight debt issues and equity issues, the inverse mills ratio is significant at the 1 percent level. These results suggest that selection-bias exists when firms choose their financing types. Therefore controlling for the selection bias is important to obtain consistent estimates on determinants of the offering costs or market reactions. Consistent with Butler, Grullon and Weston (2004) who examine determinants of investment banking fees among SEO firms, I find that stock market liquidity (measured by TURNOVER) reduces direct offering costs for 68

PAGE 76

seasoned equity issues. 32 Straight debt issues by multiple book runners pay less direct offerings costs. For both straight debt and equity issues, direct offering costs are reduced for firms with more analyst coverage whereas they are increased for firms with greater relative issue size, or greater residual return standard deviation. If analyst coverage and residual return standard deviation measure information asymmetry, the results indicate that firms with higher levels of information asymmetry also pay higher direct offering costs. Based on the estimates of parameters controlling for selectivity, I estimate the direct offering costs if the alternative security had been issued. Panel B compares the mean difference between the actual direct offering costs and the hypothetical direct offering costs if the same firms had issued the alternative security type. For straight debt issues the actual mean direct offering costs are 1.093 percent of proceeds, 2.447 percent less than the average hypothetical direct offering costs if these debt-issuers had issued equity. On average the proceeds debt-issuers raised are $210.05 million, therefore it means that straight debt issuers could have saved $4.968 million than if they had issued equity instead. In contrast, equity issuers would have paid 3.374 percent less direct costs and saved $2.733 million if they had issued straight-debt. However, I cannot conclude that equity issuers did not choose the security type to minimize the offering costs because I have not considered the indirect offering costs yet. Only when we combine the direct and indirect offering costs together, we can determine whether or not firms choose the security type that is less costly for their shareholders benefits. The results are 32 Investment banks fees represent the lions share of direct offering costs. Direct offerings costs are the sum of investment banking fees, accounting fees, legal fees, printing and other expenses. However, the information on expenses except investment banking fees is missing for many firms in SDC data. I assume these fees are equal to zero if they are missing when we calculate the direct offering costs. Thats why the direct offering costs are close to gross spreads. The results are qualitatively the same when we use gross spread to proxy for the direct offering costs. 69

PAGE 77

qualitatively very similar when I use gross spreads as the dependent variable. They are not reported for the sake of brevity. Insert Table 14 here C.2. Event Study Results Panel A of Table 15 reports cumulative abnormal returns (CARs) for different event windows and compares CARs of straight debt issues with CARs of equity issues. Consistent with previous studies (see Asquith and Mullins (1986), Masulis (1986), Smith (1986), Jung, Kim and Stulz (1996) and among others), the stock market reactions to seasoned equity issues are significantly negative. The three-day CAR is .759 percent. On the equity-issue announcement dates, the stock prices drop 0.735 percent on average. In contrast, the stock market reactions to straight debt issues are insignificantly different from zero for all four-event windows. The results show that there is insignificant market reaction of .003 percent on the announcement dates of straight debt issues. CAR(-1,+1), the three-day cumulative abnormal for SEOs in my study are greater than what are documented in prior studies, which is often below minus 2.5 percent. One reason may be that all the sample firms in my study are followed by at least one financial analyst, which may help improve firm valuation. In deed, originally I have 2,106 equity issuers with the CAR(-1, +1) available, with an average CAR(-1, +1) of .60 percent. Panel B of Table 15 compares the CAR(-1, +1) for firms without analyst coverage and firms with analyst coverage. For equity issuers with analyst coverage the average CAR(-1, +1) is .145 percent, whereas it is .341 for equity issuers without analyst coverage. The difference is significant at 1 percent level. The average CAR(-1, +1) for straight debt issuers with 70

PAGE 78

analyst coverage is also greater than the average CAR(-1, +1) for straight debt issuers without analyst coverage. Panel C of Table 15 reports the CAR(-1, +1) for the ten groups generated by dividing all the 2,545 sample firms with complete information into quintiles according to analyst coverage and further classifying each analyst coverage quintile into debt and equity groups. The stock market reacts most negatively to equity-issuers with the least analyst coverage, whereas stock market reacts most positively to debt issuers with the least analyst coverage. However, the CAR(-1, +1) does not change monotonically with increase in analyst coverage for both equity and debt issuers. Insert Table 15 here C.3. Results on Three-day Cumulative Abnormal Return Panel A of Table 16 reports the determinants on three-day CARs for straight debt issues and equity issues. The inverse mills ratio (IMR) in the regression of equity issues is -1.19 and is significant at 5 percent level. After controlling for the selection bias, for debt issues the pre-issue cumulative abnormal returns is the only variable that significantly affects CAR(-1,+1); for equity issues, the standard deviation of residual returns is the other variable besides IMR that has significant impact on CAR(-1, +1). My results are different from prior studies that ignore the selection bias. For example, Jung et. al. (1996) find that relative issue size is the only significant determinant on market reaction to debt issues whereas market-to-book ratio and cash flow to asset ratio affect market reaction to equity issues. Panel B of Table 6 shows that market reaction to straight debt issuers would have changed from 0.034 percent to .965 percent if they had issued equity. In contrast, market reaction to equity-issuers would have become .012 percent from 71

PAGE 79

1.759 percent if they had issued straight-debt. These changes are significant at the 1 percent level. Insert Table 16 here C.4. Results on Indirect Offering Costs Following Asquith-Mullin (1986) and Bayless (1994), I assess the impact of security issue announcements on market values and consider it as the indirect costs for security offerings, INDCP. INDCP is denoted as percentage of principal amount raised, thereby is comparable with direct offering cost, DCP. Panel A of Table 17 shows that inverse mills ratios are significant and the coefficients differ substantially between OLS model without controlling for selection-bias and OLS model correcting for it, indicating self-selection bias exists when I use indirect offering costs as dependent variable. Bayless (1994) shows that the offering costs of an equity issue are 35.4 to 48.6 percent greater than those of a straight debt issue by firms with similar size and gross proceeds. Bayless does not estimate what the indirect cost would have been if the same firm had issued the alternative security type. Correcting for the selection bias, I find that the indirect offering costs would have risen significantly by 9.77 percent if debt-issuers had offered equity, which is equivalent to an increase of $19.653 million assuming a debt-issuer raises the average proceeds of $201.14 million for debt issues. In contrast, if an equity issuer had offered straight debt, their indirect offering costs would have increased significantly by 3.154 percent, which is equivalent to $3.956 million assuming that the equity issuing company realizes the average proceeds of $110.51 million for equity issues. I have shown that on average if an equity issuer had offered debt, its direct offering cost would have 72

PAGE 80

increased by 3.374 percent. The results suggest that there is a tradeoff between direct and indirect offering costs for equity issuers in case they had offered debt. As a result, it is necessary to examine the total offering costs to decide whether equity issuers choose the cheaper financing type. Insert Table 17 here C.5. Results on Total Offering Costs Table 18 compares the mean actual total offering costs and the hypothetical total offerings costs if a firm had issued the alternative security type. For straight debt offering firms, their total offering costs would have amounted to 17.85 percent if they had offered equity. The 11.22 percent increase is significant at the 1 percent level, and is equivalent to $24.58 million. In contrast, the actual total offering costs for equity issuers are 20.99 percent of proceeds, which would have risen slightly by 0.206 percent (equivalent to $0.23 million increase) if these equity-issuing companies had offered debt. The increase is both statistically and economically insignificant. The results support the offering cost minimization hypothesis that firms choose the financing type that minimizes the offering costs because choosing the alternative security type will not enable firms to save offering costs. Insert Table 18 here VI. Summary and Conclusions This paper investigates whether selectivity exists when firms choose to issue equity or debt, and whether managers choose to issue the security type that minimizes the offering costs thereby acting in shareholders interests or choose the security type for 73

PAGE 81

their own benefits. I apply Heckman s (1979) two-stage method to jointly examine the security choice and the offering costs. This procedure corrects the selection bias and enables the estimation of the offering costs that would have occurred if the same firm had issued the alternative security type. I consider the direct offering costs, indirect offering cost and total offering costs and control for many factors that may affect firms financing choices and offering costs, which include access to capital markets, expected information production, information asymmetry, firms growth opportunities, tax payments, tangible and intangible assets, pre-issue stock market performance as well as absolute and relative issue size, etc. My results show that self-selection bias exists when firms choose to issue securities. Debt-issuers would have had a significantly increase in offerings costs if they had issued equity. Equity-issuers would have had an insignificant increase in total offering costs if they had offered debt. These results indicate that managers in security offering firms choose the less expensive financing type when they seek external financing. I also find that stock market reaction to straight debt issue announcements is insignificantly negative on the announcement dates and other event windows, which is inconsistent with the finding in Howton, Howton and Perfect (1998) that the market reaction on the announcement dates of straight debt financing is significantly negative. This paper provides new evidence on the determinants of market reactions to security issue announcements and on the determinants of direct offering costs, and indirect offering costs with the selection bias controlled for. For the same set of independent variables, many of their impacts on market reaction and offering costs differ between straight debt issuers and equity issuers. For example, for straight debt issuers, pre-issue 74

PAGE 82

cumulative abnormal return is the only significant variable that affect the market reaction to security issue announcements; for equity issuers, besides the inverse mills ratio, the standard deviation of residual returns is the only significant variable to explain market reactions. Consistent with Butler et al (2004), I find that stock market liquidity reduces the direct offering costs. I also find that analyst coverage reduces investment banking fees and direct offering costs for both equity issues and straight debt offerings. It will be interesting in future to investigate the relationship between analyst coverage, stock price informativeness and corporate bonds yield, because empirical studies on analyst coverage and the cost of raising capital are limited, the extant studies focus on analyst behavior, and on how analyst forecast properties affect stock returns. 75

PAGE 83

Figure 1 Analyst coverage and time The vertical axis stands for the average number of analysts following security issuers (ANA), and the horizontal axis stands for the month relative to the offering month, 0. Negative (positive) month number denotes the number of month before (after) the security offering month. 76

PAGE 84

Table 1 Sample distribution of security offerings over 1984-2002 period The sample firms do not belong to financial (SIC 6000-6999) nor utility industries (SIC 4900-4949). These firms have complete information from SDCs Global New issues database, I/B/E/S historical Summary database, CRSP and COMPUSTAT. Year Number of straight debt issues Number of primary equity issues Total 1984 36 31 67 1985 73 43 116 1986 100 48 148 1987 69 54 123 1988 63 17 80 1989 59 36 95 1990 54 29 83 1991 111 102 213 1992 119 81 200 1993 153 105 258 1994 85 72 157 1995 126 113 239 1996 132 119 251 1997 132 85 217 1998 149 61 210 1999 114 94 208 2000 70 119 189 2001 103 80 183 2002 83 86 169 Total 1831 1375 3206 77

PAGE 85

Table 2 Summary statistics This Table reports the summary statistics for the principal variables. ANNCH is the change in analyst coverage, calculated as the post-issue 12-month mean analyst coverage minus the pre-issue 12-month mean analyst coverage. ANN is the pre-issue 12-month mean analyst coverage. MBKR is Market-to-book ratio, calculated as market value at prior fiscal year end plus book value of total assets minus book value of equity, then divided by the book value of assets. LNTA is natural logarithm of book value of total assets. FIXTA is the ratio of net value of property, plant and equipment to the book value of total assets (TA). TXTTA is the ratio of tax payments to the book value of total assets. XRDTA is the ratio of expenditure in research & development to book value of total assets. I assume a firms R&D expenditure is zero if it is missing in COMPUSTAT. LTDTA is the ratio of long-term debt to book value of total assets. CFLTA is the ratio of cash flow to book value of total assets. ROA is return on assets, calculated as earnings before interest, taxes, depreciation and amortization divided by book value of total assets. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. BETA is the beta coefficient of the market return in the market model, obtained by regressing a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. RESIDSD is the standard deviation of the residual returns of the market model. TURNOVER is the average daily stock trading turnover over [-252, -1] period. Daily trading turnover the daily trading volume divided by the number of outstanding shares. All information from COMPUSTAT is at the fiscal year end prior to the security offering year. The numbers in the parentheses are t-statistic under the null hypothesis that the mean difference between debt group and equity group is equal to zero. ASYIND is information asymmetry index using the following equation based on Butler, Grullon and Weston (2004): ikKkkXRankKNASYIND111 Where X ik is the k th variable used as information asymmetry measure, which include -ANN, -LNTA, -TURNOVER, XRDTA, MBKR and RESIDSD. Variables Total Sample mean (3206 obs) Debt sample mean (1831 obs) Equity sample mean (1375 obs) Mean difference T-stat H 0 : dif = 0 H a : dif 0 ANNCH 0.6894 0.1300 1.4344 -1.3045 a -16.08 ANN 12.0039 16.5111 6.0020 10.5092 a 39.79 ASYIND 0.4658 0.3813 0.5783 -0.1970 a -42.19 MBKR 2.2339 1.6176 3.0546 -1.4370 a -13.75 LNTA 6.8700 8.1634 5.1478 3.0156 a 54.47 FIXTA 0.1184 0.4414 0.3231 0.1183 a 13.76 TXTTA 0.0257 0.0310 0.0188 0.0122 a 11.83 XRDTA 0.0510 0.0147 0.0993 -0.0846 a -16.27 LTDTA 0.2491 0.2726 0.2177 0.0550 a 7.86 CFLTA 0.0506 0.1009 -0.0165 0.1174 a 15.57 ROA 0.0990 0.1570 0.0218 0.1352 a 17.05 ABCUMRET 0.3519 0.0087 0.8089 -0.8002 a -17.48 BETA 0.9519 0.8956 1.0270 -0.1314 a -6.46 RESIDSD 0.0265 0.0191 0.0363 -0.0172 a -36.34 TURNOVER 0.5443 0.3625 0.7863 -0.4238 a -20.60 a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 78

PAGE 86

Table 3 Firm quality and information production This Tables reports the univariate test results on the relationship between firm quality and information production. Firm quality is measured by ABCUMRET, MBKR, and ROA. Information production is measured by ANNCH. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. MBKR is Market-to-book ratio, calculated as market value at prior fiscal year end plus book value of total assets minus book value of equity, then divided by the book value of assets. ROA is return on assets, calculated as earnings before interest, taxes, depreciation and amortization divided by book value of total assets. Change in analyst coverage is calculated as the post-issue 12-month mean analyst coverage minus the pre-issue 12-month mean analyst coverage. Panel A. Pre-issue cumulative abnormal return and change in analyst coverage ABCUMRET quartiles Straight Debt (# observation ) Equity (# observation ) Mean difference (T-stat, H 0 : D E 0 H a : D E < 0) Quartile 1 (low) -0.1155 (631) 0.6919 (171) -0.8074 a (-4.407) Quartile 2 0.1029 (588) 1.1392 (275) -1.0363 a (-6.343) Quartile 3 0.3497 (466) 1.4153 (443) -1.0656 a (-6.638) Quartile 4 (High) 0.5985 (146) 1.7341 (655) -1.1356 a (-4.853) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -0.7140 a (-2.972) -1.0421 a (-5.956) Panel B. Market-to-book ratio and change in analyst coverage MBKR quartiles Straight Debt (# observation ) Equity (# observation ) Mean difference (T-stat, H 0 : D E 0 H a : D E < 0) Quartile 1 (Low) -0.0091 (570) 1.0236 (232) -1.0327 a (-6.552) Quartile 2 0.1530 (539) 1.0843 (262) -0.9313 a (-6.038) Quartile 3 0.2045 (484) 1.5024 (329) -1.2979 a (-8.379) Quartile 4 (High) 0.2593 (238) 1.7283 (563) -1.4690 a (-6.500) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -0.2684 (-1.191) -0.7047 a (-4.444) Panel C. Profitability and change in analyst coverage ROA quartiles Straight Debt (# observation ) Equity (# observation ) Mean difference (T-stat, H 0 : D E 0 H a : D E < 0) Quartile 1 (low) 0.2744 (210) 1.3021 (592) -1.0277 a (-5.052) Quartile 2 0.0061 (516) 1.2575 (285) -1.2514 a (-7.959) Quartile 3 0.0893 (550) 1.6006 (252) -1.5114 a (-8.595) Quartile 4 (high) 0.2308 (555) 1.7877 (246) -1.5568 a (-9.143) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) 0.0435 (0.205) -0.4856 a (-3.044) a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 79

PAGE 87

Table 4 Information asymmetry and information production This Tables reports the univariate test results on the relationship between information asymmetry and information asymmetry. Information asymmetry is measured by INVANN and ASYIND. Information production is measured by ANNCH. INVANN is inverse analyst coverage, the reciprocal of the pre-issue 12-month average analyst coverage. ANNCH is the change in analyst coverage, calculated as the post-issue 12-month mean analyst coverage minus the pre-issue 12-month mean analyst coverage (ANN). ASYIND is information asymmetry index using the following equation based on Butler, Grullon and Weston (2004): ikKkkXRankKNASYIND111 Where X ik is the k th variables used as information asymmetry measures, which include -ANN, -LNTA, -TURNOVER, XRDTA, MBKR and RESIDSD. LNTA is natural logarithm of book value of total assets. TURNOVER is the average daily stock trading turnover over [-252, -1] period. Daily trading turnover the daily trading volume divided by the number of outstanding shares. XRDTA is the ratio of expenditure in research & development to book value of total assets. I assume a firms R&D expenditure is zero if it is missing in COMPUSTAT. MBKR is Market-to-book ratio, calculated as market value at prior fiscal year end plus book value of total assets minus book value of equity, then divided by the book value of assets. RESIDSD is the standard deviation of the residual returns of the market model, which regresses a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. Panel A. Inverse analyst coverage and information production Inverse Analyst coverage Quartiles Straight Debt (# observation ) Equity (# observation ) Mean difference (T-stat, H 0 : D E 0 H a : D E < 0) Quartile 1 (Low) -0.3682 (748) 0.7080 (55) -1.0763 a (-2.524) Quartile 2 0.2202 (610) 1.0927 (199) -0.8725 a (-4.375) Quartile 3 0.7331 (315) 1.4941 (479) -0.7610 a (-4.657) Quartile 4 (high) 0.9378 (158) 1.5581 (642) -0.6202 a (-4.487) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -1.3061 a (-8.144) -0.8500 b (-2.031) Panel B. Information asymmetry index and information production Information asymmetry index quartiles Straight Debt (# observation ) Equity (# observation ) Mean difference (T-stat, H 0 : D E 0 H a : D E < 0) Quartile 1 (low) -0.1467 (721) 0.8644 (675) -1.0111 a (-3.357) Quartile 2 0.3454 (612) 1.1489 (189) -0.8035 a (-4.436) Quartile 3 0.1894 (427) 1.4696 (375) -1.2802 a (-8.228) Quartile 4 (high) 0.7257 (71) 1.5536 (730) -0.8278 a (-4.163) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -0.8725 a (-4.199) -0.6892 b (-2.335) a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 80

PAGE 88

Table 5 Firm quality, information asymmetry and information production The whole sample are first divided into two groups according to whether the measure on information asymmetry (inverse analyst coverage or information asymmetry index) is greater than the corresponding median value of the the whole sample or not; within each information asymmetry group, I further divide the sample into two groups depending on whether an observation has measures on firm quality (return on assets, pre-issue cumulative abnormal return respectively) that are greater than the corresponding median value for the whole sample or not. Eventually I divide the four groups classified by information asymmetry and firm quality measures into eight groups based on the type of financing: equity or debt. Information production is measured by the change in analyst coverage. T-stat is the t-statistic on H 0 : Difference=0 and H a : Difference 0. Panel A. Return on assets, inverse analyst coverage and change in analyst coverage Return on assets: median Return on assets: > median Inverse mean analyst coverage Debt (# obs) Equity (# obs) Difference (T-stat) Debt (# obs) Equity (# obs) Difference (T-stat) : median -0.3356 (483) 0.5340 (148) -0.8696 a (-3.830) 0.0239 (875) 1.6731 (106) -1.6492 a (-5.810) : > median 0.9170 (243) 1.4406 (807) -0.5236 a (-3.418) 0.6795 (230) 1.6984 (392) -1.0189 a (-6.597) Panel B. Pre-issue abnormal return, inverse analyst coverage and change in analyst coverage Pre-issue abnormal return: median Pre-issue abnormal return: > median Inverse analyst coverage Debt (# obs) Equity (# obs) Difference (T-stat) Debt (# obs) Equity (# obs) Difference (T-stat) : median -0.1980 (919) 0.6710 (92) -0.8690 a (-3.277) 0.0930 (439) 1.2015 (1066) -1.1086 a (-4.533) : > median 0.5652 (300) 1.0248 (292) -0.4595 a (-3.089) 1.2112 (173) 1.7090 (829) -0.4978 a (-2.870) Panel C. Return on assets, information asymmetry index and change in analyst coverage Return on assets: median Return on assets: > median information asymmetry index Debt (# obs) Equity (# obs) Difference (T-stat) Debt (# obs) Equity (# obs) Difference (T-stat) : median -0.0676 (572) 0.9920 (189) -1.0596 a (-5.466) 0.1895 (761) 1.2304 (81) -1.0409 a (-3.868) :> median 0.6455 (154) 1.3688 (688) -0.7233 a (-3.633) 0.0959 (344) 1.7829 (417) -1.6870 a (-10.472) Panel D. Pre-issue abnormal return, information asymmetry index and change in analyst coverage Pre-issue abnormal return: median Pre-issue abnormal return: > median Information asymmetry index Debt (# obs) Equity (# obs) Difference (T-stat) Debt (# obs) Equity (# obs) Difference (T-stat) : median -0.0768 (909) 0.8753 (106) -0.9521 a (-4.229) 0.4136 (424) 1.1852 (164) -0.7715 a (-3.582) : > median 0.1852 (310) 0.9647 (278) -0.7795 a (-4.783) 0.3987 (188) 1.7134 (827) -1.3147 a (-6.537) a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 81

PAGE 89

Table 6 Regressions without controlling endogeneity of information production Endogeneity here means that the financing choices may affect information production and in the mean time information production may affect the financing choices. This Table shows results not considering such endogeneity. EDDUM is a dummy variable, which equals one if equity is issued, and zero if debt is issued. ANN is the pre-issue 12-month mean analyst coverage. INVANN is the reciprocal of ANN. ASYIND is information asymmetry index. MBKR is Market-to-book ratio, calculated as market value at prior fiscal year end plus book value of total assets minus book value of equity, then divided by the book value of assets. LNTA is natural logarithm of book value of total assets. FIXTA is the ratio of net value of property, plant and equipment to the book value of total assets (TA). TXTTA is the ratio of tax payments to the book value of total assets. XRDTA is the ratio of expenditure in research & development to book value of total assets. We assume a firms R&D expenditure is zero if it is missing in COMPUSTAT. LTDTA is the ratio of long-term debt to book value of total assets. CFLTA is the ratio of cash flow to book value of total assets. ROA is return on assets, calculated as earnings before interest, taxes, depreciation and amortization divided by book value of total assets. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. BETA is the beta coefficient of the market return in the market model, obtained by regressing a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. RESIDSD is the standard deviation of the residual returns of the market model. TURNOVER is the average daily stock trading turnover over [-252, -1] period. Daily trading turnover is the daily trading volume divided by the number of outstanding shares. # OBS is the number of observations. Panel A. OLS regression with robust error The dependent variable is change in analyst coverage Independent variable Model 1 Model 2 Model 3 EDDUM 0.7792 a 0.8152 a 0.8005 a ANN -0.0640 a INVANN 0.4986 a ASYIND -0.6053 MBKR 0.0927 a 0.0627 b 0.0647 b ROA 1.0696 a 1.1531 a 1.1041 a LNTA 0.0285 -0.1462 a -0.2075 a ABCUMRET 0.1463 a 0.1400 a 0.1454 a RESIDSD -0.0968 b -0.0602 -0.0361 CONSTANT 0.8196 a 1.0966 a 1.8442 R 2 0.1199 0.0958 0.0949 # OBS 3206 3206 3206 82

PAGE 90

Table 6 (Continued) Panel B. Probit regression with robust error EDDUM is dependent variable. EDDUM equals 1 if equity is issued and 0 otherwise. Independent variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 ANNCH 0.0648 a 0.0721 a 0.0647 a INVANN -0.7363 a -0.8371 a ASYIND 0.2740 0.0338 MBKR 0.1775 a 0.1737 a 0.1704 a 0.1531 a 0.1457 a 0.1523 a LNTA -0.5295 a -0.5788 a -0.5181 a -0.5155 a -0.5709 a -0.5141 a FIXTA 0.3643 b 0.2885 c 0.3878 b 0.3325 b 0.2423 0.3354 b TXTTA -4.1049 b -4.5064 a -4.0488 b -4.2719 b -4.7649 a -4.2650 b XRDTA 3.0526 a 3.0455 a 2.7912 b 3.5757 a 3.6179 a 3.5422 a LTDTA -0.1443 -0.0484 -0.1489 -0.1894 -0.0856 -0.1900 CFLTA 0.9072 1.0502 0.8830 0.7671 0.9164 0.7644 ROA -3.5261 a -3.7185 a -3.5336 a -3.3022 a -3.4917 a -3.3034 a ABCUMRET 0.8124 a 0.8367 a 0.8114 a 0.7806 a 0.8035 a 0.7805 a BETA -0.0518 -0.0976 -0.0486 -0.0965 -0.1535 b -0.0961 RESIDSD 0.1550 a 0.1512 a 0.1459 a 0.1616 a 0.2024 a 0.1605 a TURNOVER 0.4242 a 0.4265 a 0.4514 a 0.3939 a 0.3060 a 0.3972 a CONSTANT 2.7548 a 2.6957 a 2.5705 a 2.6892 a 3.2677 a 2.6665 a Pseudo R 2 0.5613 0.5614 0.5614 0.5657 0.5702 0.5657 a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 83

PAGE 91

Table 7 Regressions controlling endogeneity of information production for the whole sample firms I first run the following probit model for the whole sample firms: )()1(Pr1211109876543210TURNOVERRESIDSDBETAABCUMRETROACFLOWTALTDTAXRDTATXTTAFIXTALNTAMBKReddumob Then I use the predicted value of the above probit model, EHAT, the predicted probability of equity issues versus debt issues, to explain change in analyst coverage in the following OLS model: RESIDSDABCUMRETLNTAROAMBKRANNEHATANNCH76543210 PCHANN (RCHANN) is the predicted (residual) value of the OLS model above. PCHANN is the predicted change in analyst coverage. RCHANN is the residual change of analyst coverage. Both are used to measure expected information production due to security offerings. INVANN is the reciprocal of ANN. ASYIND is the information asymmetry index. PCHANN*INVANN is the product of PCHANN and INVANN. PCHANN)ASYIND is the product of PCHANN and ASYIND. MBKR is Market-to-book ratio, calculated as market value at prior fiscal year end plus book value of total assets minus book value of equity, then divided by the book value of assets. LNTA is natural logarithm of book value of total assets. FIXTA is the ratio of net value of property, plant and equipment to the book value of total assets (TA). TXTTA is the ratio of tax payments to the book value of total assets. XRDTA is the ratio of expenditure in research & development to book value of total assets. I assume a firms R&D expenditure is zero if it is missing in COMPUSTAT. LTDTA is the ratio of long-term debt to book value of total assets. CFLTA is the ratio of cash flow to book value of total assets. ROA is return on assets, calculated as earnings before interest, taxes, depreciation and amortization divided by book value of total assets. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. BETA is the beta coefficient of the market return in the market model, obtained by regressing a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. RESIDSD is the standard deviation of the residual returns of the market model. TURNOVER is the average daily stock trading turnover over [-252, -1] period. Daily trading turnover s the daily trading volume divided by the number of outstanding shares. # OBS is the number of observations. Panel A. OLS regression with robust error The dependent variable is ANNCH, change in analyst coverage. Independent variable Coefficient t P >|t| EHAT 1.8182 a 6.73 0.000 ANN -0.0641 a -8.26 0.000 MBKR 0.0946 a 2.86 0.002 ROA 1.0231 a 4.71 0.000 LNTA 0.1616 a 2.96 0.000 ABCUMRET 0.1211 a 3.84 0.003 RESIDSD -0.1707 a -3.92 0.000 CONSTANT -0.3321 -0.73 0.464 R 2 0.1206 # OBS 3206 84

PAGE 92

Table 7 (Continued) Panel B. Probit regression with robust errors Independent variable Model 1 Model 2 Model 3 Model 4 Model 5 PCHANN 0.2220 b 0.4153 a 0.4185 a 0.2538 b -0.2156 INVANN -1.0214 a -2.2446 a PCHANN*INVANN 0.7894 c ASYIND -0.4085 -1.3287 c PCHANN*ASYIND 1.2225 a MBKR 0.1517 a 0.1233 a 0.1003 b 0.1586 a 0.1436 a LNTA -0.4605 a -0.4703 a -0.4536 a -0.4677 a -0.4484 a FIXTA 0.4216 a 0.3655 b 0.3272 b 0.3948 b 0.3568 b TXTTA -3.2680 c -3.0931 c -2.7668 -3.2310 b -2.9662 c XRDTA 3.1591 a 3.2150 a 3.1777 a 3.5711 a 3.4808 a LTDTA -0.2135 -0.1423 -0.0962 -0.2161 -0.1974 CFLTA 0.6931 0.7042 0.5694 0.6997 0.5321 ROA -3.2901 a -3.3533 a -3.1868 a -3.2470 a -3.0288 a ABCUMRET 0.6840 a 0.6059 a 0.5135 a 0.6672 a 0.6320 a BETA -0.0163 -0.0502 -0.0491 -0.0163 -0.0153 RESIDSD 0.1590 a 0.2154 a 0.2408 a 0.1731 a 0.1724 a TURNOVER 0.3858 a 0.2435 b 0.1960 c 0.3401 b 0.3373 b CONSTANT 2.1135 a 2.2692 a 2.2043 a 2.2970 a 2.4805 a Pseudo R 2 0.5625 0.5685 0.5692 0.5626 0.5643 Panel C. Probit regression with robust errors Independent variable Model 1 Model 2 Model 3 Model 4 Model 5 RCHANN 0.0622 a 0.0658 a 0.0154 0.0620 a 0.0596 a INVANN -0.7827 a -0.7151 a RCHANN*INVANN 0.3760 a ASYIND 0.2110 -1.1718 c RCHANN*ASYIND 0.7678 a MBKR 0.1612 a 0.1561 a 0.1524 a 0.1559 a 0.1333 a LNTA -0.5354 a -0.5888 a -0.5950 a -0.5266 a -0.4560 a FIXTA 0.3175 b 0.2339 0.2800 c 0.3356 b 0.3237 b TXTTA -4.5012 a -4.9651 a -4.9465 a -4.4587 a -3.2942 c XRDTA 3.5275 a 3.5421 a 3.4995 a 3.3228 a 4.1583 a LTDTA -0.1679 -0.0672 -0.1276 -0.1716 -0.2366 CFLTA 0.8343 0.9846 0.7777 0.8165 0.4868 ROA -3.3783 a -3.5714 a -3.3870 a -3.3843 a -2.9187 a ABCUMRET 0.8177 a 0.8430 a 0.8445 a 0.8169 0.6277 a BETA -0.1048 -0.1564 b -0.1465 c -0.1022 -0.0618 RESIDSD 0.1602 a 0.1980 a 0.1890 a 0.1533 a 0.1847 a TURNOVER 0.4062 a 0.3261 a 0.2881 a 0.4269 a 0.2947 b CONSTANT 2.8715 a 3.4242 a 3.4670 a 2.7291 a 2.5279 a Pseudo R 2 0.5652 0.5691 0.5725 0.5652 0.5676 a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 85

PAGE 93

Table 8 Change in analyst coverage and long-run post-issue buy and hold cumulative abnormal return Change in analyst coverage is the difference between the 12-month average analyst coverage after security offerings and the 12-month average analyst coverage before security offerings. The post-issue buy-and-hold abnormal return is calculated according to the following equation: 100*]11[22tMttjtjtRRabcumretny Where n=1, 2 or 3, =252 if n=1, =504 if n=2, and =756 if n=3. R jt is firm js daily return since the second day of security offerings. R Mt is the markets daily return, which is measured by the NYSE/AMEX/NASDAQ value-weighted index return. MBKR is Market-to-book ratio. LNTA is natural logarithm of book value of total assets. FIXTA is the ratio of net value of property, plant and equipment to the book value of total assets (TA). ROA is return on assets. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. BETA is the beta coefficient of the market return in the market model, obtained by regressing a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. RESIDSD is the standard deviation of the residual returns of the market model. Panel A. Change in analyst coverage and post-issue one-year abnormal return Change in Analyst coverage quartiles Straight Debt (# observation ) Equity (# observation ) Mean difference (T-stat: H 0 : D E 0 H a : D E > 0) Quartile 1 (low) -3.9582 a (616) -6.5283 (124) 2.5701 (0.476) Quartile 2 -4.9808 a (452) -12.7662 a (310) 7.7854 b (2.272) Quartile 3 0.1904 (323) -9.1483 a (428) 9.3387 a (2.644) Quartile 4 (high) 3.0094 (342) 13.3123 c (416) -10.3028 (-1.377) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -6.9676 a (-2.830) -19.8406 b (-2.231) Mean difference (T-stat, H 0 : Q2-Q4 0 H a : Q2-Q4 < 0) -7.9902 a (-3.042) -26.0784 a (-3.343) Panel B. Change in analyst coverage and post-issue two-year abnormal return Change in Analyst coverage quartiles Straight Debt (# observation ) Equity (# observation ) Mean difference (T-stat: H 0 : D E 0 H a : D E > 0) Quartile 1 (low) -6.4199 a (527) -21.5867 b (105) 15.1668 b (1.755) Quartile 2 -7.4578 b (418) -29.4845 a (270) 22.0267 a (4.034) Quartile 3 -6.4585 b (301) -21.8634 a (382) 15.4049 a (2.718) Quartile 4 (high) -1.5814 (308) -9.9316 (355) 8.3502 (1.191) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -4.8385 (-1.232) -11.6551 (-1.119) Mean difference (T-stat, H 0 : Q2-Q4 0 H a : Q2-Q4 < 0) -5.8763 c (-1.364) -19.5528 a (-2.516) 86

PAGE 94

Table 8 (Continued) Panel C. Change in analyst coverage and post-issue three-year abnormal return Change in Analyst coverage quartiles Straight Debt (# observation ) Equity (# observation ) Mean difference (T-stat: H 0 : D E 0 H a : D E > 0) Quartile 1 (low) -8.1482 b (473) -24.1738 b (87) 16.0256 c (1.355) Quartile 2 -15.2857 a (380) -47.0627 a (233) 31.7770 a (4.142) Quartile 3 -12.3249 a (279) -33.6480 a (323) 21.3231 a (2.891) Quartile 4 (high) -5.9669 (284) -16.2861 (293) 10.3192 (0.843) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -2.1814 (-0.3624) -7.8877 (-0.4953) Mean difference (T-stat, H 0 : Q2-Q4 0 H a : Q2-Q4 < 0) -9.3188 c (-1.537) -30.7766 a (-2.347) Panel D. Regressions of post-issue buy and hold abnormal return on change in analyst coverage Dependent variable: ABCUMRET1Y Dependent variable: ABCUMRET2Y Dependent variable: ABCUMRET3Y Independent variable debt equity debt equity debt equity ANNCH 1.031 a 5.256 a 1.248 c 3.645 b 0.526 4.542 a MBKR -2.876 b -0.171 -4.460 b -0.491 2.414 0.317 ABCUMRET 6.840 c -0.815 -8.890 -1.983 -6.116 -8.909 b ROA 17.139 9.242 43.304 2.215 23.502 -28.556 BETA -4.797 c -5.061 0.292 -7.595 c 0.608 -11.774 LNTA 2.244 a -0.367 5.200 a 5.076 a 8.728 a 8.889 a CONSTANT -14.188 c -2.161 -48.284 a -39.734 a -89.456 a -63.458 a R 2 0.017 0.015 0.022 0.020 0.027 0.023 # OBS 1733 1278 1554 1112 1416 936 a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 87

PAGE 95

Table 9 Robust tests: Regressions controlling endogeneity of information production based on randomly selected half sample firms The whole 3206 sample firms are randomly divided into two sample groups with equal sample size: the base group and the test group. I first run the following probit model on the base group and test group respectively to get EHAT for each group, EHAT is the predicted probability of equity issues over debt issues: )()1(Pr1211109876543210TURNOVERRESIDSDBETAABCUMRETROACFLOWTALTDTAXRDTATXTTAFIXTALNTAMBKReddumob Then I run the following OLS model for the base group: RESIDSDABCUMRETLNTAROAMBKRANNEHATANNCH76543210 Next we multiply the coefficients of the independent variables in the above OLS regression by the corresponding values of independent variables for the test group. The resultant product, PCHANN, now is an ex-ante measure on expected information production for firms in the test group, because it is constructed based on all the information before security offerings. INVANN is the reciprocal of ANN, the pre-issue 12-month average analyst coverage. ASYIND is the information asymmetry index. PCHANN*INVANN is the product of PCHANN and INVANN. PCHANN*ASYIND is the product of PCHANN and ASYIND. MBKR is Market-to-book ratio. LNTA is natural logarithm of book value of total assets. FIXTA is the ratio of net value of property, plant and equipment to the book value of total assets (TA). TXTTA is the ratio of tax payments to the book value of total assets. XRDTA is the ratio of expenditure in research & development to book value of total assets. LTDTA is the ratio of long-term debt to book value of total assets. CFLTA is the ratio of cash flow to book value of total assets. ROA is return on assets. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. BETA is the beta coefficient of the market return in the market model. RESIDSD is the standard deviation of the residual returns of the market model. TURNOVER is the average daily stock trading turnover over [-252, -1] period. Panel A. Probit regressions on base group and control group Independent variable Base group (907 straight debt issues, 696 primary equity issues) Test group (924 straight debt issues, 679 primary equity issues MBKR 0.1523 b 0.2266 a LNTA -0.5279 a -0.5338 a FIXTA 0.4089 c 0.3549 TXTTA -2.6041 -5.1444 a XRDTA 1.5552 4.8237 a LTDTA -0.0821 -0.1732 CFLTA 0.8674 0.9867 ROA -3.2591 b -4.0254 a ABCUMRET 0.8497 a 0.8025 a BETA 0.0114 -0.1182 RESIDSD 0.1477 a 0.1653 b TURNOVER 0.5319 a 0.3048 c CONSTANT 2.6158 a 2.8634 a Pseudo R 2 0.5598 0.5658 # OBS 1603 1603 88

PAGE 96

Table 9 (Continued) Panel B. OLS regression with robust error for the base group The dependent variable is change in analyst coverage. Independent variable Coefficient t P >|t| EHAT 2.0301 a 5.59 0.000 ANN -0.0722 a -6.89 0.000 MBKR 0.2006 a 6.16 0.000 ROA 1.4651 a 5.05 0.000 LNTA 0.2188 a 3.10 0.002 ABCUMRET 0.0583 1.04 0.296 RESIDSD -0.1423 b -2.51 0.012 CONSTANT -1.1178 c -1.92 0.055 # obs 1603 R 2 0.1487 Panel C. Probit regression with robust errors for the test group The dependent variable is EDDUM, which equals one if equity is issued, and 0 if straight debt is issued. Independent variable Model 1 Model 2 Model 3 Model 4 PCHANN 0.4963 a 0.4843 a 0.4564 a -0.1509 INVANN -0.7125 b -2.4928 a PCHANN*INVANN 1.2083 b ASYIND -1.1561 -2.1216 b PCHANN*ASYIND 1.5121 MBKR 0.1013 0.0374 0.1594 b 0.1332 c LNTA -0.4323 a -0.4146 a -0.4416 a -0.4228 a FIXTA 0.4229 c 0.3732 c 0.3917 c 0.3430 TXTTA -3.0765 -2.3792 -3.2581 -2.9207 XRDTA 4.7082 a 4.5496 a 5.9528 a 5.6690 a LTDTA -0.2409 -0.1572 -0.3148 -0.2910 CFLTA 0.6486 0.4512 0.4917 0.2069 c ROA -3.7275 a -3.4994 a -3.4400 a -3.1325 b ABCUMRET 0.5465 a 0.4280 a 0.5549 a 0.5346 a BETA -0.0457 -0.0398 -0.0269 -0.0230 RESIDSD 0.1804 a 0.2048 a 0.1885 a 0.1847 a TURNOVER 0.1639 0.1091 0.1186 0.1182 CONSTANT 2.1283 a 2.1535 a 2.4712 a 2.6894 a Pseudo R 2 0.5728 0.5745 0.5708 0.5736 # OBS 1603 1603 1603 1603 a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 89

PAGE 97

Table 10 Investor optimism, post-issue buy-and-hold abnormal return, and financing choice Investor optimism is measured by FE, LGTH, REVR and OPTIM. FER is the pre-issue 12-month mean forecast error. Forecast error is the mean annual consensus EPS forecast minus the actual annual EPS divided by the absolute value of the mean annual consensus EPS. LGTH is the pre-issue 12-month mean analysts long-term EPS growth rate forecast. REVIR is the pre-issue 12-month mean IBES revision ratio. The revision ratio is the number of analysts making upward revision minus the number of analysts making downward revision divided by the total number of analysts following a firm. OPTIM is investor optimism index based on the ranking of FER, LGTH and REVR. It is constructed using the following equation based on Butler, Grullon and Weston (2004): ikKkkiZRankKNOPTIM111 Where Z ik represents the k th variable used in constructing the investor optimism index. T-stat is t-statistic value on H 0 : Dif=0 and H a : Dif0. Panel A. Comparison of investor optimism between debt issues and equity issues Pre-issue 12 month mean (t-stat) Post-issue 12 month mean (t-stat) Mean Change (post pre) (t-stat) Variable debt equity dif debt equity dif debt equity dif FE 0.3584 0.2187 0.140 (1.19) 0.3161 0.4226 -0.106 (-0.997) -0.042 (0.820) 0.204 b (2.57) -0.246 c (-1.69) LGTH 0.1326 0.2399 -0.107 a (-23.2) 0.1303 0.2423 -0.112 a (-26.8) -0.002 a (-6.39) 0.002 (0.986) -0.005 c (-1.88) REVR -0.0614 0.0212 -0.083 a (-11.3) -0.0718 -0.0050 -0.067 a (-8.52) -0.010 c (-1.84) -0.026 a (-3.62) 0.016 c (1.72) OPTIM 0.4558 0.5704 -0.115 a (-22.6) 0.4460 0.5912 -0.145 a (-30.3) -0.010 a (-3.72) 0.021 a (5.45) -0.003 a (-6.60) Panel B. OLS Regression results Dependent variable: ABCUMRET1Y Dependent variable: ABCUMRET2Y Dependent variable: ABCUMRET3y Independent variable debt equity debt equity debt equity ANNCH 1.277 a 5.116 a 1.760 b 3.694 b 1.282 3.068 c ANN 0.200 0.661 0.400 c 1.013 c 0.579 c 0.963 OPTIM -20.168 b -56.826 a -30.153 b -74.659 a -43.122 b -149.010 a MBKR -2.757 b 0.771 -2.926 -0.494 3.876 -1.732 ABCUMRET 8.048 b 1.117 -10.134 -1.010 -9.401 -8.228 b ROA 18.800 28.669 b 28.105 30.304 b 7.904 22.682 BETA -4.743 c -3.300 -1.344 -5.819 -0.370 -1.203 LNTA 1.764 b -3.085 3.721 b -1.900 5.604 b -4.062 a CONSTANT -5.460 33.404 c -28.818 c 28.494 -54.095 b 74.355 b R 2 0.031 0.041 0.022 0.036 0.027 0.036 # OBS 1671 935 1495 810 1360 687 90

PAGE 98

Table 10 (Continued) Panel C. Probit model regression results Independent variable Model 1 Model 2 Model 3 Model 4 OPTIM -0.1702 -0.3327 -0.1192 -0.1884 ANNCH 0.0601a PCHANN 0.4146 a -0.3994 c INVANN -2.5936 b PCHANN*INVANN 0.9613 ASYIND -1.0626 PCHANN*ASYIND 1.6583 a MBKR 0.2058 a 0.1918 a 0.0826 0.1255 b LNTA -0.5521 a -0.5422 a -0.4369 a -0.4394 a FIXTA 0.4428 a 0.4056 b 0.4077 b 0.4650 a TXTTA -2.6719 -2.8329 -1.6954 -1.6154 XRDTA 3.0924 a 3.5429 a 3.2007 a 5.6690 a LTDTA -0.1116 -0.1361 -0.1138 -0.1849 CFLTA 0.3624 0.3159 0.0988 0.0127 ROA -4.4446 a -4.3327 a -4.1556 a -4.1103 b ABCUMRET 0.8123 a 0.7821 a 0.5184 a 0.6399 a BETA -0.0539 -0.0927 -0.0217 -0.0089 RESIDSD 0.2031 a 0.2120 a 0.2781 a 0.2167 a TURNOVER 0.3317 a 0.3107 a 0.1986 c 0.3000 b CONSTANT 3.0175 a 3.0375 a 2.2009 a 2.4544 a Pseudo R 2 0.5282 0.5324 0.5341 0.5340 # OBS 2784 2784 2784 2784 a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 91

PAGE 99

Table 11 Sample distribution Year Straight debt issues Primary equity issues Total 1984 36 22 58 1985 66 25 91 1986 96 37 133 1987 65 43 108 1988 60 14 74 1989 57 23 80 1990 50 21 71 1991 81 67 148 1992 107 58 165 1993 136 83 219 1994 65 50 115 1995 104 82 186 1996 106 63 169 1997 90 61 151 1998 114 50 164 1999 89 68 157 2000 50 91 141 2001 90 72 162 2002 70 83 153 Total 1532 1013 2545 92

PAGE 100

Table 12 Characteristic comparisons between equity issues and straight debt issues This Table reports summary statistics of the principal variables. DCP is direct offering cost, calculated as the sum of gross spread, accounting, printing, legal as well as miscellaneous fees as percentage of principal amount. If information on accounting, printing, legal or miscellaneous fees is missing we assume it is zero. CAR(-1,+1) is cumulative abnormal returns over the three-day event window. INDCP is indirect offering cost, defined as [(-1)*CAR(-1,+1)*market value on the day before issue announcement)/total proceeds raised]. TCP is total offering cost, equal to the sum of direct offerings cost and indirect offering cost. ANN is the pre-issue 12-month mean analyst coverage. LNPRDS is log of total proceeds raised. MBKR is Market-to-book ratio, calculated as market value at prior fiscal year end plus book value of total assets minus book value of equity, then divided by the book value of assets. LNTA is log of book value of total assets. FIXTA is the ratio of net value of property, plant and equipment to the book value of total assets. TXTTA is the ratio of tax payments to the book value of total assets. XRDTA is the ratio of expenditure in research & development to book value of total assets. I assume a firms R&D expenditure is zero if it is missing in COMPUSTAT. LTDTA is the ratio of long-term debt to book value of total assets. CFLTA is the ratio of cash flow to book value of total assets. ROA is return on assets, calculated as earnings before interest, taxes, depreciation and amortization divided by book value of total assets. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. BETA is the beta coefficient of the market return in the market model, obtained by regressing a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. RESIDSD is the standard deviation of the residual returns of the market model. TURNOVER is the average daily stock trading turnover over [-252, -1] period. Daily trading turnover the daily trading volume divided by the number of outstanding shares. RISIZE is relative issue size, defined as the ratio of proceeds raised to the market value of equity at the fiscal year end prior to the issue year. UWREPU is the lead underwriters reputation, measured as the market share of the lead manger. For equity issuers it is based on the entire SDC seasoned equity offerings database; for debt-issuers issuers, it is based on the entire SDC non-convertible debt offerings with maturity beyond one year. NUMBK is the number of book runners. Variables Straight debt (1532 observations) Equity (1013 observations) Mean Difference (T test, H 0 : Mean diff=0 H a : Mean diff0) GSP (%) 1.077 4.822 -3.745 a (-91.099) DCP (%) 1.094 5.311 -4.217 a (-74.463) CAR(-1,+1) (%) 0.034 -1.759 1.793 a (7.618) IDCP (%) 4.540 15.682 -11.142 a (3.710) TCP (%) 5.634 20.993 -15.359 a (-5.116) ANN 15.774 7.049 8.725 a (29.605) ANNCH 0.211 1.461 -1.250 a (-13.193) MBKR 1.532 2.339 -0.807 a (-11.120) ROA 0.153 0.089 0.065 a (12.997) LNTA 8.027 5.766 2.261 a (41.245) FIXTA 0.445 0.357 0.088 a (8.824) LNPRDS 5.053 4.169 24.642 a (13.195) LTDTA 0.279 0.258 0.021 b (2.577) TXTTA 0.030 0.020 0.010 a (8.693) CFLTA 0.098 0.042 0.056 a (11.268) XRDTA 0.013 0.056 -0.043 a (-11.246) BETA 0.894 1.020 -0.126 a (-5.426) RESIDSD 0.019 0.033 -0.014 a (-27.860) 93

PAGE 101

Table 12 (Continued) Straight debt (1532 observations) Equity (1013 observations) Mean Difference (T test, H 0 : Mean diff=0 H a : Mean diff0) ABCUMRET 0.007 0.659 -0.652 a (-15.946) RISIZE 0.178 0.238 -0.060 a (-4.908) UWREPU 0.098 0.062 0.036 a (14.751) NUMBK 1.103 1.043 0.060 a (5.371) TURNOVER 0.363 0.764 -0.401 a (-17.693) a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 94

PAGE 102

Table 13 Heckman two-step procedure: The first-step-probit regression results PCHANN is the predicted value of the following OLS model: RESIDSDABCUMRETLNTAROAMBKRANNEHATANNCH76543210 Where EHAT is the predicted value of the following probit model: )()1(Pr1211109876543210TURNOVERRESIDSDBETAABCUMRETROACFLOWTALTDTAXRDTATXTTAFIXTALNTAMBKReddumob PCHANN is the predicted change in analyst coverage. It is used to measure expected information production due to security offerings. LNTA is the logarithm of book value of total assets. INVANN is the reciprocal of ANN, the pre-issue 12-month average analyst coverage. MBKR is Market-to-book ratio. FIXTA is fixed assets to total assets ratio. TXTTA is the ratio of tax payments to the book value of total assets. XRDTA is the ratio of expenditure in research & development to book value of total assets. We assume a firms R&D expenditure is zero if it is missing in COMPUSTAT. LTDTA is the ratio of long-term debt to book value of total assets. CFLTA is the ratio of cash flow to book value of total assets. ROA is the return on assets. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. BETA is the beta coefficient of the market return in the market model, obtained by regressing a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. RESIDSD is the standard deviation of the residual returns of the market model. TURNOVER is the average daily stock trading turnover over [-252, -1] period. # obs is the number of observations. The dependent variable is EDDUM, which equals 1 if primary equity is issued, otherwise equals 0 if straight debt is issued. Independent Variable Coefficient Z-stat (P-value) PCHANN 0.363 a 3.59 (0.002) INVANN -2.419 a -3.39 (0.001) PCHANN*INVANN 0.861 b 1.99 (0.047) MBKR 0.049 0.77 (0.444) LNTA -0.438 a -9.39 (0.000) FIXTA 0.312 c 1.92 (0.055) TXTAT -2.823 -1.53 (0.126) XRDTA 3.429 a 2.94 (0.003) LTDTA -0.039 -0.18 (0.855) CFLTA 0.706 0.74 (0.449) ROA -4.033 a -4.61 (0.000) ABCUMRET 0.583 a 3.86 (0.000) BETA -0.056 -0.69 (0.488) RESIDSD 0.265 a 5.18 (0.000) TURNOVER 0.203 c 1.69 (0.095) CONSTANT 2.275 a 6.31 (0.000) Pseudo R 2 0.493 # OBS 2545 a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 95

PAGE 103

Table 14 Results on direct offering costs IMR is inverse mills ratio obtained through Heckmans two-step procedure by running probit model in Table 2 first. ANNLN is natural logarithm value of (ANN+1). UWREPU is reputation of the lead underwriter. LNPRDS is natural logarithm value of total proceeds raised, PRDS. MBKDUM is a dummy variable, which equals one if a firm has multiple book runners, and zero otherwise. MBKR is market-to-book value ratio. ABCUMRET is pre-issue abnormal cumulative return. RESIDSD is the standard deviation of the residual returns of the market model. TURNOVER is the average daily stock trading turnover over [-252, -1] period. ROA is return on assets. RISZFY is relative issue size. The values in the parentheses are t-statistic values. Panel A. OLS regressions with robust variance. DCP is the dependent variable Straight debt Equity Independent Variables OLS OLS in Coeff OLS OLS in Coeff ANNLN -0.380 a (-6.76) -0.285 a (-4.92) 33.33% -0.576 a (-8.07) -0.543 a (-7.53) 5.73% UWREPU -0.240 (0.96) -0.246 (-0.98) -2.50% -1.791 a (-3.06) -1.613 a (2.67) 9.94% LNPRDS -0.059 c (-1.84) 0.001 (0.03) 101.69% -0.724 a (-12.99) -0.665 a (-10.52) 8.15% MBKR 0.007 (0.26) -0.030 (-0.98) -528.57% 0.014 (0.75) -0.002 (-0.10) -114.29% MBKDUM -0.314 a (-4.04) -0.290 a (-3.83) 7.64% 0.104 (0.73) 0.138 (0.96) 32.69% TURNOVER 0.202 b (2.23) 0.107 (1.19) -47.03% -0.122 c (-1.88) -0.188 a (-2.73) -54.10% ROA -0.951 b (-2.58) -0.570 (-1.53) 40.06% -0.076 (-0.28) -0.004 (-0.01) 94.74% RISZFY 0.575 a (2.60) 0.575 a (2.71) 0.00% 0.635 a (2.76) 0.500 b (2.42) -21.26% AMCUMRET -0.025 (-0.35) -0.145 c (-1.92) -480.00% -0.010 (-0.40) -0.028 (-1.16) -180.00% RESIDSD 0.279 a (6.42) 0.228 a (5.25) -18.28% 0.305 a (7.32) 0.287 a (6.78) -5.90% IMR -0.231 a (-3.77) -0.305 a (-2.74) CONSTANT 1.864 a (8.52) 1.840 a (8.58) 8.423 a (27.77) 8.428 a (27.85) R-square 0.484 0.492 0.493 0.497 # OBS 1532 1532 1013 1013 96

PAGE 104

Table 14 (Continued) Panel B. Comparisons between mean DCP, and hypothetical DCP if a firm issued the alternative security instead Straight debt (1532 observations) Equity (1013 observations) Actual DCP 1.094 5.311 Hypothetical DCP if the other type of security issued 3.540 1.937 Difference (T-stat) -2.447 a (-110.14) 3.374 a (74.97) Difference in dollars if mean PAMT raised -$4.968 million $4.133 million a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 97

PAGE 105

Table 15 Event study results CAR(-1,+1) is three-day cumulative abnormal returns. Day 0 is the security offerings date. Panel A. Comparisons of cumulative abnormal returns (CAR) among different event windows and between straight debt issues and common stock issues Event window 1533 Straight debt issues (T-test: mean CAR=0) 1012 Common stock issues (T-test: mean CAR=0) CAR difference (T-stat: CAR difference=0) CAR(-1,+1) 0.034 (0.394) -1.759 a (-8.04) 1.793 a (7.62) CAR(-1,0) -0.005 (-0.068) -1.708 a (-9.35) 1.703 a (8.70) CAR(0,0), -0.003 (-0.057) -0.735 a (-5.65) 0.732 a (5.23) CAR(0,1) 0.036 (0.496) -0.786 a (4.33) 0.822 a (4.24) Panel B. Comparisons of CAR(-1,+1) between firms followed by analysts, and firms not followed by analysts Straight debt (obs) Equity (obs) Firms with analyst coverage 0.106 (2492) 2.145 (1806) Firms without analyst coverage -0.361 (80) -5.341 (300) Mean difference (t-stat:differencde=0) 0.467 (0.788) 3.196 a (4.777) Mean of Total 0.091 (2572) -2.600 (2106) Panel C. Comparisons of CAR(-1,+1) among analyst coverage quintiles and between straight debt issues and common stock issues Analyst coverage quintiles Straight debt issues (T-test: mean CAR=0) Common stock issues (T-test: mean CAR= 0) Mean difference: D-E (t-stat, H a: difference>0) Quintile 1 (low) 0.549 (1.475) -2.320 a (-5.345) 2.869 a (5.016) Quintile 2 0.055 (0.190) -1.523 a (-3.952) 1.578 a (3.279) Quintile 3 0.284 (1.293) -1.601 a (-4.038) 1.885 a (4.159) Quintile 4 -0.219 (-1.453) -0.962 c (-1.913) 0.743 c (1.415) Quintile 5 (high) -0.083 (-0.706) -1.207 b (-2.172) 1.124 b (1.997) Mean (Q1-Q5) (t-stat, H a: Q1-Q5<0) 0.633 (1.619) -1.113 c (-1.578) ------a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 98

PAGE 106

Table 16 Results on three-day cumulative abnormal return, CAR(-1,+1) IMR is inverse mills ratio obtained through Heckmans two-step procedure by running Probit model (4) in Table 12 first. LNANN is natural logarithm of (ANN+1), where ANN is the 12-month average analyst coverage before security-offering month. LNPRDS is natural logarithm of total proceeds raised, PRDS. MBKR is market-to-book value ratio. CFLTA is the ratio of cash flow to book value of total assets. ROA is return on assets. RISZFY is relative issue size. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. RESIDSD is the standard deviation of the residual returns of the market model which regresses a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. in Coeff is the change in coefficient. # OBS is the number of observations. The values in the parentheses are t-statistic values. Panel A. OLS regressions with CAR(-1,+1) as the dependent variable Straight debt Equity Independent Variables OLS OLS in Coeff OLS OLS in Coeff LNAAN -0.139 (-0.77) -0.152 (-0.67) -9.35% 0.272 (0.64) 0.364 (0.87) 33.82% LNPRDS 0.110 (0.87) 0.103 (0.68) -6.36% 0.166 (0.65) 0.382 (1.27) 130.12% MBKR -0.104 (-0.67) -0.010 (-0.62) 90.38% 0.122 (0.99) 0.050 (0.39) -59.02% CFLTA -2.592 (-0.83) -2.660 (-0.81) -2.62% -1.297 (-0.38) -1.974 (-0.57) -52.20% ROA 0.730 (0.26) 0.656 (0.23) -10.14% 3.946 (1.11) 4.791 (1.32) 21.41% RISZFY 0.038 (0.10) 0.038 (0.10) 0.00% 0.430 (0.24) -0.064 (-0.04) -114.88% ABCUMRET -0.833 b (-2.40) -0.817 b (-2.16) 1.92% -0.478 (-1.30) -0.562 (-1.47) -17.57% RESIDSD 0.254 (1.43) 0.264 (1.43) 3.94% -0.490 b (-2.25) -0.626 a (-2.76) -27.76% IMR 0.034 (0.12) -1.193 b (-2.10) CONSTANT -0.345 (-0.45) -0.345 (-0.45) -1.693 (-1.30) -1.444 (-1.14) R-square 0.016 0.016 0.034 0.038 # OBS 1532 1532 1013 1013 Panel B. Comparisons among mean actual CAR(-1,+1), and hypothetical CAR(-1,+1) if a firm issued the alternative security type instead Straight debt Equity Actual CAR(-1,+1) 0.034 -1.759 Hypothetical CAR if the other type of security issued -0.990 -0.012 Difference (T-stat) 1.024 a (11.51 ) -1.748 a (-7.99) a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 99

PAGE 107

Table 17 Results on indirect offering costs, INDCP LNANN is natural logarithm of (ANN+1), where ANN is the 12-month average analyst coverage before security-offering month. LNPRDS is natural logarithm of total proceeds raised, PRDS. MBKR is market-to-book value ratio. CFLTA is the ratio of cash flow to book value of total assets. ROA is return on assets. RISZFY is relative issue size. ABCUMRET is pre-issue abnormal cumulative return. RESIDSD is the standard deviation of the residual returns of the market model which regresses a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. in Coeff is the change in coefficient. # obs is the number of observations. The values in the parentheses are t-statistic values. Panel A. OLS regressions with robust standard errors. INDCP is the dependent variable Straight debt Equity Independent Variables OLS OLS in Coeff OLS OLS in Coeff LNAAN 7.305 b (1.97) 2.209 (0.63) 74.24% 1.145 (0.29) 0.259 (0.07) -73.38% LNPRDS -8.290 b (-2.00) -11.789 b (-2.39) 40.97% -2.799 (-1.02) -5.113 c (-1.65) -82.67% MBKR 8.630 (1.49) 10.730 c (1.77) -20.95% -0.210 (-0.15) 0.577 (0.39) 374.76% CFLTA 73.452 (1.07) 86.105 (1.25) -14.04% -18.352 (-0.48) -11.082 (-0.29) 39.61% ROA -43.382 (-0.71) -74.370 (-1.12) 61.74% -9.362 (-0.25) 18.580 (-0.48) 298.46% RISZFY 5.297 (1.57) 5.240 (1.51) -1.33% -26.656 b (-2.38) -21.309 b (-2.04) 20.06% ABCUMRET 9.059 (1.62) 15.795 b (2.31) -73.97% 6.631 c (1.84) 7.548 b (2.00) 13.83% RESIDSD -1.567 (-0.56) 2.500 (0.70) -157.03% 3.386 (1.60) 4.846 b (2.22) 43.12% IMR 13.751 c (1.67) 12.825 b (2.19) CONSTANT 15.478 (0.75) 15.516 (0.76) 17.979 (1.38) 15.651 (1.16) R-square 0.014 0.017 0.040 0.044 # OBS 1532 1532 1013 1013 Panel B. Comparisons among mean actual INDCP, and hypothetical INDCP if a firm issued the alternative security type instead Straight debt Equity Actual INDCP 4.450 15.682 Hypothetical INDCP if the other type of security issued (2) 14.312 19.262 Difference (T-stat) -9.771 a (-4.49) 3.580 c (1.65) Difference in dollars if mean PRDS raised -$19.654 million $3.956 million a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 100

PAGE 108

Table 18 Comparison of total offerings costs TCP is total offering costs, which is the sum of direct offering costs and indirect offering costs. The hypothetical TCP are the sum of hypothetical direct offering costs and the hypothetical indirect offering costs. Straight debt (1532 observations) Equity (1013 observations) Actual TCP 5.634 20.993 Hypothetical TCP if the other type of security issued 17.852 21.199 Difference (T-stat) -12.218 a (-5.79) -0.206 (-0.095) Difference in dollars if mean PAMT raised -$24.575 million $0.228 million a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 101

PAGE 109

References Abarbanell, Jeffrey S., 1991, Do analysts earnings forecasts incorporate information in prior stock price changes? Journal of Accounting and Economics 14, 147-165. Allen, Franklin, and Douglas Gale, 1998, Diversity of opinion and financing of new technologies, Journal of Financial Intermediation 8, 1999, 68-89. Arbel, A., & Strebel, P., 1982, The neglected and small firm effects, The Financial Review, 17, 201-218. Asquith, P. and D. Mullins, 1986, Equity issues and offerings dilution, Journal of Financial Economics 15, 61-89. Baker, M. and J. Wurgler, 2000, The equity share in new issues and aggregate stock returns. The Journal of Finance 55, 2219-2257. Baker, M. and J. Wurgler, 2002, Market timing and capital structure, Journal of Finance 57, 1-32. Baker, Molly, 1996, Heard on the street, Wall Street Journal July 2, C1. Barry, Christopher B., and Robert H. Jennings, 1992, Information and diversity of analyst opinion, Journal of Financial and Quantitative Analysis 27 (2), 169-183. Barry, Christopher B., Chris J. Muscarella and Michael R. Vetsuypens, 1991, Underwriter warrants, underwriter compensation, and the costs of going public, Journal of Financial Economics 29 (1), 113-136. Barth M. and A. Hutton, 2004, Analyst Earnings Forecast Revisions and the pricing of accruals, Working paper, Stanford University, Stanford, California, 1-59. Barth, Mary E., Ron Kasznik, and Maureen McNichols,1998, Analyst coverage and intangible assets, working paper, Stanford University. Bartov, E., D. Givoly, and C. Hayn, 2002, The rewards to meeting or beating earnings expectations. Journal of Accounting & Economics 33 173-204. Baxter, Nicholas and John Cragg, 1970, Corporate choice among long-term financing Instruments, Review of Economics and Statistics 52, 225-235. 102

PAGE 110

Bayless, Mark and Susan Chaplinsky, 1991, Expectations of security type and the information content of debt and equity offers, Journal of Financial Intermediation 1, 195-214. Bayless, Mark and Susan Chaplinsky, 1996, Is there a window of opportunity for seasoned equity issuance? The Journal of Finance 51 (1), 253-278. Bayless, Mark E., 1994, The influence of predictability on differences in the market reaction to debt and equity issue announcements, The Journal of Financial Research 17, 117-131. Bayless, Mark E., and Diltz, 1991, The empirical relevance of asymmetric information for the securities offering process, Journal of Business, Finance and Accounting 18 (3), 31-344. Berger, P., and E. Ofek, 1995, Diversifications effect on firm value, Journal of Financial Economics 37, 39-65. Best, J., W. Best and J. Young, 1998. An examination of proxies of information asymmetry, Advances in financial planning and Forecasting, Vol. 8, 17-34. Best, F., and M. Zhang, 1993, Alternative Information Sources and the Information Content of Bank Loans, Journal of Finance 48, 1507-1522. Best, Ronald W., 1994, Relative information asymmetry as a determinant of the market reaction to corporate financing announcements, Journal of Economics and Finance 18, 1994, 231-240. Bhushan, Ravi 1989, Firm characteristics and analyst following, Journal of Accounting and Economics 11,255-274. Blazenko, G. 1987, Managerial Preference, Asymmetric Information, and Financial Structure, Journal of Finance 42 (4), 1987, 839-862. Bowen, Robert M., Chen, Xia and Cheng, Qiang, "Analyst Coverage and the Cost of Raising Equity Capital: Evidence from Underpricing of Seasoned Equity Offerings" (April 2004). http://ssrn.com/abstract=417860 Bradford, W., 1987, The Issue Decision of Manager-Owners Under Information Asymmetry, Journal of Finance 42 (5), 1987, 1245-1260. Branson, B.C., Guffey, D.M., and D.P. Pagach 1998, Information conveyed in announcements of analyst coverage, Contemporary Accounting Research, Summer 1998. Brennan, M. and A. Subrahmanyam, 1995, Investment analysis and the price formation in securities markets, Journal of Financial Economics 38, 361-381. 103

PAGE 111

Brennan, M., & Hughes, P., 1991, Stock Prices and the Supply of Information, Journal of Finance, 46, 1665-1691. Brennan, Michael and Alan Kraus, 1987, Efficient financing under asymmetric information, Journal of Finance 42, 1225-1243. Brennan, Michael.J., Narasimhan Jegadeesh, and Bhaskaran Swaminathan, 1993, Investment analysis and the adjustment of stock prices to common information. The Review of Financial Studies Vol. 6 (4), 799-824. Bricker, Robert, Julia Grant, Timothy Fogarty, and Gary Previts, 1999, Determinants of analyst following, Journal of Corporate Communications, August 1999. Brown, Stephen, 2005, The post-traditional analyst, CFA Magazine 2005, January/February, 38-39 Brown, Stephen J. And Jerold B. Warner, 1985, Using daily stock returns; the case of event studies, Journal of Financial Economics 14 (1), 3-31. Byoun, Soky, and Jong C. Rhim, 2002. Tests of the pecking order theory and the tradeoff theory of optimal capital structure. Financial Management Association 2002 Conference accepted paper. Butler, W. Alexander, Gustavo Grullon, and James P. Weston, 2004, Stock market liquidity and the cost of issuing equity, Journal of Quantitative Analysis, Forthcoming Chang, C., 1992, Capital structure as an optimal contract between employees and investors, Journal of Finance 47, 1141-58. Chang, Xin, Dasgupta, Sudipto and Hilary, Gilles, "Analyst Coverage and Capital Structure Decisions". Tuck Contemporary Corporate Finance Issues III Conference Paper; HKUST Finance Working Paper. http://ssrn.com/abstract=571065 Chirinko, Robert S., and Anuja R. Singha, 2000. Testing static tradeoff against pecking order models of capital structure: a critical comment. Journal of Financial Economics 58, 417-425. Choe, Hyuk, Ronald W. Masulis, and VikramNanda, 1993. Common stock offerings across the business cycle: theory and evidence. Journal of Empirical Finance 1, 3-31. Chopra, Vijay K., 1998, Why so much error in analysts earnings forecasts? Financial Analyst Journal 54, 30-37. 104

PAGE 112

Chung, K., & Jo, H., 1996, The Impact of Security Analysts Monitoring and Marketing Functions on the Market Value of Firms, Journal of Financial and Quantitative Analysis 31, 493-512. Chung, Kee H., 2000, Marketing of stocks by brokerage firms: the role of financial analysts, Financial Management 29, 35-54. Clark, J.C. Dunbar and K. Kahle, 2001, Long-run performance and insider trading in completed vs. cancelled seasoned equity offerings, Journal of Financial and Quantitative Analysis 36, 415-430. Clarke, Jonathan and Shastri, Kuldeep, 2001, On Information Asymmetry Metrics, http://ssrn.com/abstract=251938 Cooney, J., and A. Kalay, 1993, Positive information from equity issue announcements, Journal of Financial Economics 33, 149-172. Dann, L. and W. Mikkelson, 1984, Convertible debt issuance, capital structure change, and financing-related information, Journal of Financial Economics 13, 157-186. Davidson, R. and J. G. MacKinnon. 1993, Estimation and inference in econometrics. New York: Oxford University Press. Dechow, Patricia M., Hutton, Amy P. and Sloan, Richard G.,1999, The relation between analysts' forecasts of long-term earnings growth and stock price performance following equity offerings, http://ssrn.com/abstract=168488 DeMeza, D., and C. Webb, 1987. Too much investment: A problem of asymmetric information, The Quarterly Journal of Economics 102, No. 2, 281-292. Denis,D.J., and V.T. Mihov,2003, the choice among bank debt, non-bank private debt and public debt: evidence from new corporate borrowings, Journal of Financial Economics 70 (1), 3-28. Diamond, W.D., and R.E. Verrecchia (1991), Disclosure, liquidity and the cost of capital, Journal of Finance 46, 1235-1259. Dierkens, Nathalie, 1991, Information asymmetry and equity issues, Journal of Financial and Quantitative Analysis 13 (2), 157-186. Dissanaike, G., B.M. Lambrecht, and A. Saragga-Seabra, 2001, Differentiating Debt Target from Non-target Firms: An Empirical Study on Corporate Capital Structure. University of Cambridge Working Paper. 105

PAGE 113

DMello, R., F.P. Schlingemann, and V. Subramaniam, 2002, The monitoring versus informational role of institutional investors: Evidence from seasoned equity offerings, Working Paper. DMello, Ranjan, and Ferris Stephen P., 2000, The Information effects of analyst activity at the announcement of new equity issues, Financial Management, Spring 2000, 78-95. Doukas, John A., Chansog (Francis) Kim and Christos Pantzalis, 2002a, security analysis, investor uncertainty, and capital allocation, Working Paper. Doukas, John S., Chansog (Francis) Kim and Christos Pantzalis, 2002b, The double faces of analyst coverage, Working Paper, 2002. Doukas, John, Chansog Kim and Christos Pantzalis, 2000, Security Analysis, Agency Costs, and Company Characteristics, Financial Analysts Journal 56 (6), 2000, 54-63. Dowling, James J., 1990, The equity and information content of a security: implications for issuance strategy, Massachusetts Institute of Technology, Ph.D. Dissertation 1990. Dreman, David, and Michael Berry, 1995, Analyst forecasting errors and their implications for security analysis, Financial Analyst Journal 51, 30-42. Dunbar, Craig G. 1995, The use of warrants as underwriter compensation in initial public offerings. Journal of Financial Economics, 38, 1995, 59-78. Durnev, Artyom, Morck, Randall and Bernard Yeung, 2001, Does firm-specific information in stock prices guide capital allocation? NBER WP 8093, January 2001. Dybvig, H. Philip, Jaime Zender, 1991, Capital structure and dividend irrelevance with asymmetric information, Review of Financial Studies 4, 201-219 Easterbrook, Frank H., 1984, Two agency-cost explanations of dividends, The American Economic Review 74 (4), 650-659, The American Economic Review 46 (2), 1956, 97-113. Eckbo, B. Espen, 1986, Valuation effects of corporate debt offerings, Journal of Financial Economics 15, 119-151. Eckbo, B. Espen, Ronald A. Masulis, and Oyvind Norli, 2000, Seasoned public offerings: resolution of the new issues puzzle, Journal of Financial Economics 56, 252-292. Eckbo, B.E. and R.W. Masulis, Seasoned equity offerings: a survey, in: Finance (Series of Handbooks in Operations Research and Management Science), Jarrow, R., V. Maksimovic and W.T. Ziemba (eds), Elsevier-North Holland, 1995, 1017-1072. Eckbo, B.E., and R.W. Masulis, 1992, Adverse selection and the rights offer paradox. Journal of Financial Economics 32, 293-332. 106

PAGE 114

Elliott, William B., Johanna Koeter-Kant, and Richard S. Warr, 2002. Valuation errors at the time of security issuance and the market timing theory of capital structure. Financial Management Association 2002 conference accepted paper. Fama, Eugene F., 1978, The effects of a firms investment and financing decisions on the welfare of its security holders, The American Economic Review 66 (3), 1956, 272-284. Fama, Eugene F., and Kenneth R. French, 2002, Testing tradeoff and pecking order predictions about dividends and debt, Review of Financial Studies 15, 2002, 1-33. Fang, Lily H., 2004, Investment bank reputation, the price and quality of underwriting services, Journal of Finance forthcoming. Fischer, Edwin O., Robert Heinkel, and Josef Zechner, 1989, Dynamic capital structure choice: theory and tests, Journal of Finance 44 (1), 1989, 19-40. Francis, Bill B., Hasan, Iftekhar 1999, An empirical investigation of the firms contract choice and level of underpricing of initial public offerings. Working paper, University of South Florida. Frank, Murray Z., and Vidhan K. Goyal, 2002, Testing the pecking order theory of capital structure, Journal of Financial Economics 67(2), 217-248. Frankel, R., M. McNichols, and G.P. Wilson, 1995, Discretionary disclosure and external financing, The Accounting Review (January), 135-150. Fried, D., and D. Givoly, 1982, Financial analysts forecast of earnings: A better surrogate for market expectations. Journal of Accounting and Economics 4 (2), 85-107. Fulghieri, P. and D. Lukin, 2001. Information production, dilution costs, and optimal security design, Journal of Financial Economics 61, 3-42. Gebhardt, William R., Charles M. Lee and Bhaskaran Swaminathan, 2001, Toward an implied cost of capital, Journal of Accounting Research June, 135-176. Giammarino, R.M., and E.H. Neave 1982, The failure of financial contracts and the relevance of financial policy, Queens University Working Paper No. 82-3. Gode, Dan and Partha Mohanram, 2001, What affects the implied cost of equity capital? Stern School of Business Working Paper. http://ssrn.com/abstract=258702 Gordon, D. A., M. J. Gordon and L. I. Gould, 1989, Choice among methods of estimating share yield, Journal of Portfolio Management 15, 50. 107

PAGE 115

Goswami, Gautam, T. Noe and M. Rebello, 1995, Debt financing under asymmetric information. Journal of Finance 50 (2), 633-659. Graham, J.R., 1996, Debt and the marginal tax rate. Journal of Financial Economics 41, 41-74. Graham, J.R., and C.R. Harvey, 2001, The theory and practice of corporate finance: Evidence from the field, Journal of Financial Economics 60, 187-243. Greene, William H., 2000, Econometric analysis, Prentice-Hall, Inc. Hansen, Robert S., and Atulya Sarin, 1998. Are analyst overoptimistic around seasoned equity offerings? http://ssrn.com/abstract=140435 Harris, Milton and Artur Raviv, 1991, The theory of capital structure, Journal of Finance 46, 297-356. Heckman, J., 1976, Simultaneous equations model with continuous and discrete endogenous variables and structural shifts, in S. Goldfeld and R. Quant eds., Studies in Non-linear Estimation, Cambridge: Ballinger. Heckman, J., 1979. Sample selection bias as a specification error, Econometrica 47, 153-161. Heinkel, R., 1982, A Theory of Capital Structure Relevance Under Imperfect Information, Journal of Finance 37 (5), 1982, 1141-1159. Helwege, Jearn, and Nellie Liang, 1996. Is there a pecking order? Evidence from a panel of IPO firms, Journal of Financial Economics 40, 429-458. Hogan, Chris E. 1997, Costs and benefits of audit quality in the IPO market: a self-selection analysis. Accounting Review, 72,67-86. Hong, Harrison, and Jeffrey D. kubik, 2003, Analyzing the analysts: career concerns and biased earnings forecast, Journal of Finance 58, 313-351. Hong, Harrison, Terrence Lim, and Jeremy Stein, 2000, Bad news travels slowly: Size, analyst coverage, and the profitability of momentum strategies, Journal of Finance 55, 265-295. Houge, Todd; Loughran, Tim; Suchanek, Gerry and Yan, Xuemin, 2001, Divergence of opinion, uncertainty, and the quality of initial public offerings, University of Notre Dame Working Paper. http://ssrn.com/abstract=273531 Hovakimian, Armen, 2004. The role of target leverage in security issues and repurchases, Journal of Business, 77, 1-31. 108

PAGE 116

Howton, Shawn D., Howton, Shelly W., and Steven B. Perfect, 1998. The market reaction to straight debt issues: The effects of free cash flow. The Journal of Financial Research Vol. XXI No. 2, 219-228. Hull, Robert M., and Robert Kerchner, 1996, Issue costs and common stock offerings. Financial Management 25 (4), 54-66. Ikenberry, D., J. Lakonishok and T. Vermaelen, 1995, Market underreaction to open market share repurchases, Journal of Financial Economics 39, 181-208. Irvine, Paul J. 2001, Do Analysts Generate Trade for Their Firms? Evidence from the Toronto Stock Exchange? Journal of Accounting and Economics 30, 209-226. Irvine, Paul J., 2003, The incremental impact of analyst initiation of coverage, Journal of. Corporate Finance 9, 431-451. Jackson, R. Andrew, 2004, Trade generation, reputation and sell-side analysts, Journal of Finance 60 (2),673-717. Jallivand, A., and R. Harris, 1984, Corporate behavior in adjusting to capital structure and dividend targets: A econometric study. Journal of Finance 39 (2), 127-145. Jegadeesh, N., M. Weinstein, and I. Welch, 1993, An empirical investigation of IPO returns and subsequent equity offerings, Journal of Financial Economics 34, 153-175. Jensen, M.C., 1986, Agency Costs of Free Cash Flow, Corporate Finance and Takeovers, American Economic Review (May), 654-665. Jensen, M.C., 2001, Value maximization, stakeholder theory, and the corporate objective function, Journal of Applied Corporate Finance 14 (3), Fall 2001, 8-21. Jensen, Michael C. and William H. Meckling, 1976, Theory of the firm: managerial behavior, agency costs and ownership structure, Journal of Financial Economics 3, 305-360. Jong, Abe de, and Chris Veld, 2001, An empirical analysis of incremental capital structure decisions under managerial entrenchment, Journal of Banking and Finance 25, 1857-1895. Jung, Kooyul, Yong-Cheol Kim, and Rene Stulz, 1996, Timing, investment opportunities, managerial discretion, and the security issue decision, Journal of Financial Economics 42, 157-185. Kamath, R., 1997, Long-term financing decisions: Views and practices of financial managers of NYSE Firms, The Financial Review, May 1997, 350-356. 109

PAGE 117

Kim, S., Lin, J., & Slovin, M., 1997, Market structure, informed trading and analysts recommendations, Journal of Financial and Quantitative Analysis 32, 507-524. Klein, Daniel P., and Brian Belt, 1994. Sustainable growth and choice of financing: a test of pecking order hypothesis, Review of Financial Studies 3, 143-154. Klein, S., J. OBrien and R. Peters, 2002. Debt vs. equity and asymmetric information: A review, The Financial Review 37, No. 3. Koch, Paul D. and Catherline Shenoy, 1999, The information content of dividend and capital structure policies, Financial Management 28 (4), 16-35. Koraczyk, R., D. Lucas, and R. McDonald, 1992, Equity Issues with Time Varying Asymmetric Information, Journal of Financial and Quantitative Analysis, September 1992, 397-417. Korajczyk, R.A., D.J. Lucas, and R.L. McDonald, 1991, The Effect of Information Releases on the Pricing and the Timing of Equity Issues, Review of Financial Studies 4, 397-417. Krasker, W.S., 1986, Stock Price Movements in Response to Stock Issue under Asymmetric Information, Journal of Finance 41 (1), 93-105. Krigman, L., W.H. Shaw and K.L. Womack, 2001, Why do firms switch underwriters? Journal of Financial Economics 60, 245-284. Lang, M. and R. Lundholm, 2002, Voluntary disclosure and equity offerings: Reducing information asymmetry or hyping the stock? Contemporary Accounting Research 17 (4), 623-662. Lee, Inmoo, Scott Lochhead, Jay Ritter, and Quanshui Zhao, 1996, The costs of raising capital, The Journal of Financial Research XIX (1), 59-74. Lee, Lung-Fei, 1978, Unionism and wage rates: A simultaneous equations model with qualitative and limited dependent variables, International Economic Review 19 (2), 415-433. Leone, Andrew J., and Joanna Wu, 2002, What does it to become a superstar? Evidence from institutional investor rankings of financial analysts, Working paper, University of Rochester. Lewis, Craig M., Richard J. Rogalski and James K. Seward, 1999, Is convertible debt a substitute for straight debt or for common equity? Financial Management, Autumn 1999. 110

PAGE 118

Lin, Hsiou-wei, and Maureen F. McNichols, 1998, Underwriting relationships, analysts' earnings forecasts and investment recommendations, Journal of Accounting and Economics 25, 101-127. Liu, Qiao and Rong Qi, 2001, Information production, analyst following and diversification discount: a market microstructure approach, http://ssrn.com/abstract=279250 Liu, Qiao and Qi, Rong, 2002, Information production, cash flow and corporate investment, AFA 2003 Washington, DC Meetings. http://ssrn.com/abstract=292844 Ljungqvist, Alexander, Felicia Marston and William J. Wilhelm, 2004, Competing for securities underwriting mandates: Banking relationships and analyst recommendations, Journal of Finance, forthcoming. Ljungqvist, Alexander P.,1999, IPO underpricing, wealth losses and the curious role of venture capitalists in the creation of public companies, Working Paper, Said Business School, Oxford University and CEPR. Loughram, Tim and Jay R. Ritter, 1995, The new issues puzzle, Journal of Finance 50, 23-51. Lucas, D. and R. McDonald, 1990, Equity issues and stock price dynamics, Journal of Finance 45, 1019-1044. Lucas, D. and R. McDonald, 1998, Shareholder Heterogeneity, Adverse Selection, and Payout Policy, Journal of Financial and Quantitative Analysis, June 1998, 233-253. Mackie-Mason, Jeffrey K., 1990, Do Firms Care Who Provides Their Financing?, in R. Glenn Hubbard, editor, Asymmetric Information, Corporate Finance, and Investment. Chicago: University of Chicago Press for National Bureau of Economic Research, pp. 63-104. Maddala, G.S., 1991, A perspective on the use of limited-dependent and qualitative variables models in accounting research, The Accounting Review 66 (4), 788-807. Marsh, Paul, 1982, The choice between equity and debt: an empirical study, Journal of Finance 37 (1), 1982, 121-144. Masulis, R.W., and A.N. Korwar, 1986. Seasoned equity offerings: an empirical investigation, Journal of Financial Economics 15, 91-118. McConnell, John J. and Henri Servaes, 1990. Additional evidence on equity ownership and corporate value, Journal of Financial Economics 27, 595-612. 111

PAGE 119

McLaughlin, Robyn, Safieddine, Assem, and Vasudevan, Gopala K., 1998, The information content of corporate offerings of seasoned securities: an empirical analysis, Financial Management 27 (2), 31-45. Merton, R. C., 1987, A simple model of capital market equilibrium with incomplete information, Journal of Finance 42, 483-510. Michaely, Roni, and Kent L. Womack, 1999, Conflict of interest and the credibility of underwriter analyst recommendations, Review of Financial Studies 12, 653-686. Mikkelson, W., and M Partch, 1985, Valuation effects of security offerings and the issuance process, Journal of Financial Economics 15, 31-60. Miller, Edward M., 1977, Risk, uncertainty, and divergence of opinion, Journal of Finance 32, 1151-1168. Miller, M. and K. Rock, 1985, Dividend policy under asymmetric information, Journal of Finance 40, 1031-1051. Modigliani, F. and M.H. Miller, 1958, Corporate income taxes and the cost of capital: A correction, American Economic Review 53, 433-443. Modigliani, F. and M.H. Miller, 1963, The cost of capital, corporation finance and the theory of investment, American Economic Review 48, 261-297. Modigliani, F., and M. Miller, 1958, The cost of capital, corporate finance and the theory of investment, American Economic Review 48, 261-97. Morck, R., A. Schleifer, and R. Vishny, 1990, Do managerial objective drive bad acquisitions? Journal of Finance 45, 31-48. Moyer, C., Chatfield, R., & Sisneros, P., 1989, Security Analyst Monitoring Activity: Agency Costs and Information Demands, Journal of Financial and Quantitative Analysis 24, 503-512. Murphy K. M.; Topel R. H., 2002, Estimation and inference in two-step econometric models, Journal of Business & Economic Statistics 20 (1), 88-97. Myers, Stewart C. and Nicholas Majluf, 1984, Corporate financing and investment decisions when firms have information that investors do not have, Journal of Financial Economics 13, 187-221. Myers, Stewart C., 1977, Determinants of corporate borrowing, Journal of Financial Economics 5, 147-175. Myers, Stewart C., 1984, The capital structure puzzle, Journal of Finance 39, 572-592. 112

PAGE 120

Myers, Stewart C., 2001, Capital Structure, Journal of Economic Perspectives 15 (2), Spring 2001, 81-102. Narayanan, M.P., 1988, Debt versus equity under asymmetric information, The Journal of Financial and Quantitative analysis 23 (1), 39-51. Ng, Chee K., and Richard L. Smith, 1996, Determinants of contract choice: The use of warrants to compensate underwriters of seasoned equity issues, Journal of Finance 51 (1), 363-380. Noe, T.H., 1988. Capital structure and signaling game equilibria, Review of Financial Studies 1, 331-355. OBrien, Thomas J., and Linda Schmid Klein, 2001, A review of debt/equity choice when securities are mispriced, University of Connecticut, Working paper. Pagan, Adrian, 1984, Econometric Issues in the Analysis of Regressions with Generated Regressors, International Economic Review 25 (1.) 221-247. Park, C., and E. Stice, Analyst forecasting ability and the stock price reaction to forecast revisions, Review of Accounting Studies, 5 (3), 259-272. Pilotte, Eugene, 1992, Growth opportunities and the stock price response to new financing, Journal of Business 65, 371-395. Piotroski, Joseph and Roulstone, Darren T.,2004, The influence of analysts, institutional investors, and insiders on the incorporation of market, industry, and firm-specific information into stock prices, University of Chicago GSB Working Paper. http://ssrn.com/abstract=299931 Pugel, Thomas A., White, Lawrence J. 1988, Am empirical analysis of the underwriting spreads on initial public offerings, Quarterly Review of Economics and Business 28, 6-20. Rajan, Raghuram and Henri Servaes, 1997, Analyst following of initial public offerings, Journal of Finance 52 (4), 507-529. Rajan, Raghuram G. and Luigi Zingales, 1995, What do we know about capital structure? Some evidence from international data, Journal of Finance 50, 1421-1460. Ritter, Jay R. 1987, the costs of going public. Journal of Financial Economics, 19, 269-281. Rothschild, M. and J. Stiglitz, 1976, Equilibrium in competitive insurance markets: An essay on the economics of imperfect information, Quarterly Journal of Economics 90, 629-650. 113

PAGE 121

Roulstone, D., 2001, Analyst following and market liquidity, Working Paper, The University of Chicago Graduate School of Business. Rudd, Judish S. (1993), Underwriter price support and the IPO underpricing puzzle, Journal of Financial Economics, 34, 135-151. Schartwz, E., and R. Aronson, 1967, Some surrogate evidence in support of a concept of optimal capital structure, Journal of Finance 22, 10-18. Shyam-Sunder, L., and S.C. Myers, 1999. Testing static tradeoff against pecking order models of capital structure. Journal of Financial Economics 51, 219-244. Smith, C., 1986, Investment banking and the capital acquisition process, Journal of Financial Economics 15, 3-29. Smith, Richard L., 1987, The choice of issuance procedure and the cost of competitive and negotiated underwriting: an examination of the impact of Rule 50. Journal of Finance 43 (3), 703-720. Spiess, D.K. and J. Affleck-Graves, 1995, Underperformance in long-run stock returns following seasoned equity offerings, Journal of Financial Economics 38, 243-267. Spiess, D.K. and J. Affleck-Graves, 1999, The long-run performance of stock returns following debt offerings, Journal of Financial Economics 54, 45-73. Stickel, S., 1992, Reputation and Performance Among Securities Analysts, Journal of Finance, 47, 1811-1836. Stiglitz, J., and A. Weiss, 1981. Credit rationing in markets with imperfect information, American Economic Review 71 (3), 393-410. Stulz, Rene, 1990, Managerial discretion and optimal financing policies, Journal of Financial Economics 26, 3-27. Sunder, Jayanthi, 2002, Information spillovers and capital structure: Theory and Evidence, Dissertation, New York University. Taggart, R., 1977, Model of Corporate Financing Decisions, Journal of Finance 32 (5), 1977, 1467-1484. Teoh, S.W., I. Welch and T.J. Wong, 1998, Earnings management and underperformance of seasoned equity offerings, Journal of Financial Economics 1935-1974. 114

PAGE 122

Teoh, Siew Hong, and T.J. Wong, 2002. Why new issues and high-accural firms underperform: the role of analyst credulity. The Review of financial Studies 15 (3), 869-900. Thomas, Shawn 2002, Firm diversification and asymmetry information: evidence from analysts forecasts and earnings announcements, Journal of Financial Economics 64 (3), 373-396. Titman, Sheridan and Roberto Wessels, 1988, The Determinants of Capital Structure Choice, Journal of Finance 43, March 1988, 1-19. Titman, Sheridan, Hovakimian, Armen G., and Opler, Tim C., 2001, The Debt-Equity Choice, Journal of Financial and Quantitative Analysis, 36 (1), 1-24. Tuna, A. Irem, 2000, External financing and voluntary disclosure: Is there a difference between equity-issuing firms and debt-issuing firms, Working Paper, University of Pennsylvania The Wharton School. Viswanathan, P.V., 1993, Strategic considerations, the pecking order hypothesis, and market reactions to equity financing, Journal of Financial and Quantitative Analysis 28, 213-234. Vogt, S.C., 1994, The role of internal financial sources in firm financing and investment decisions, Review of Financial Studies 4, 1-24. Womack, Kent L., 1996, Do brokerage analysts recommendations have investment value? Journal of Finance 51 (1), 1996, 137-167. Zwiebel, Jeffrey, 1996, Dynamic capital structure and managerial entrenchment, American Economic Review 86, 1197-1215. 115

PAGE 123

Appendices 116

PAGE 124

Appendix A Reexamine Firm quality and information production via different sample classifications The whole sample firms are first classified into debt and equity groups, then firms within each financing group are further divided into quartiles based on firm quality measures. Firm quality is measured by ABCUMRET, MBKR, and ROA. Information production is measured by ANNCH. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. MBKR is Market-to-book ratio, calculated as market value at prior fiscal year end plus book value of total assets minus book value of equity, then divided by the book value of assets. ROA is return on assets, calculated as earnings before interest, taxes, depreciation and amortization divided by book value of total assets. Change in analyst coverage is calculated as the post-issue 12-month mean analyst coverage minus the pre-issue 12-month mean analyst coverage. Panel A: Pre-issue cumulative abnormal return and change in analyst coverage ABCUMRET quartiles Straight Debt (# observation ) Equity (# observation ) Quartile 1 (low) -0.2202 (458) 0.8481 (344) Quartile 2 0.1365 (458) 1.4934 (344) Quartile 3 0.1884 (458) 1.5714 (344) Quartile 4 (High) 0.4158 (457) 1.8259 (343) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -0.6360 a (-3.699) -0.9777 a (-6.298) Panel B. Issuers market-to-book ratio and change in analyst coverage MBKR quartiles Straight Debt (# observation ) Equity (# observation ) Quartile 1 (Low) -0.0624 (458) 1.0165 (344) Quartile 2 0.1756 (458) 1.3365 (344) Quartile 3 0.1337 (458) 1.5178 (344) Quartile 4 (High) 0.2733 (457) 1.8683 (343) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -0.3357 b (-1.894) -0.8518 a (-5.116) Panel C. Issuers profitability and change in analyst coverage ROA quartiles Straight Debt (# observation ) Equity (# observation ) Quartile 1 (low) 0.1661 (458) 1.1841 (344) Quartile 2 0.0237 (458) 1.4417 (344) Quartile 3 0.1472 (458) 1.3425 (344) Quartile 4 (high) 0.1830 (457) 1.7704 (343) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -0.0169 (-0.010) -0.5863 a (-3.783) a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 117

PAGE 125

Appendix B Reexamine information asymmetry and information production via different sample classifications The whole sample firms are first classified into debt and equity groups, then firms within each financing group are further divided into quartiles based on information asymmetry measures. Information asymmetry is measured by INVANN and ASYIND. Information production is measured by ANNCH. INVANN is inverse analyst coverage, the reciprocal of the pre-issue 12-month average analyst coverage. ANNCH is the change in analyst coverage, calculated as the post-issue 12-month mean analyst coverage minus the pre-issue 12-month mean analyst coverage (ANN). ASYIND is information asymmetry index using the following equation based on Butler, Grullon and Weston (2004): ikKkkXRankKNASYIND111 Where X ik is the k th variables used as information asymmetry measures, which include -ANN, -LNTA, -TURNOVER, XRDTA, MBKR and RESIDSD. LNTA is natural logarithm of book value of total assets. TURNOVER is the average daily stock trading turnover over [-252, -1] period. Daily trading turnover the daily trading volume divided by the number of outstanding shares. XRDTA is the ratio of expenditure in research & development to book value of total assets. I assume a firms R&D expenditure is zero if it is missing in COMPUSTAT. MBKR is Market-to-book ratio, calculated as market value at prior fiscal year end plus book value of total assets minus book value of equity, then divided by the book value of assets. RESIDSD is the standard deviation of the residual returns of the market model, which regresses a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. Panel A. Inverse analyst coverage and information production Inverse Analyst coverage Quartiles Straight Debt (# observation ) Equity (# observation ) Quartile 1 (Low) -0.4879 (461) 1.0574 (344) Quartile 2 -0.0606 (465) 1.5933 (347) Quartile 3 0.2924 (449) 1.4829 (343) Quartile 4 (high) 0.7891 (455) 1.6043 (341) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -1.277 a (-7.950) -0.5469 a (-3.456) Panel B. Information asymmetry index and information production Information asymmetry index quartiles Straight Debt (# observation ) Equity (# observation ) Quartile 1 (low) -0.1938 (458) 1.1844 (344) Quartile 2 0.1009 (458) 1.4381 (344) Quartile 3 0.3402 (458) 1.7660 (344) Quartile 4 (high) 0.2728 (457) 1.3490 (343) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -0.4666 a (-2.921) -0.1646 (-1.146) a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 118

PAGE 126

Appendix C Information production and financing choices, two-step procedure results LNTA is logarithm of book value of total assets. INVANN is the reciprocal of ANN, the pre-issue 12-month average analyst coverage. MBKR is Market-to-book ratio. FIXTA is fixed assets to total assets ratio. TXTTA is the ratio of tax payments to the book value of total assets. XRDTA is the ratio of expenditure in research & development to book value of total assets. We assume a firms R&D expenditure is zero if it is missing in COMPUSTAT. LTDTA is the ratio of long-term debt to book value of total assets. CFLTA is the ratio of cash flow to book value of total assets. ROA is return on assets. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. BETA is the beta coefficient of the market return in the market model, obtained by regressing a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. RESIDSD is the standard deviation of the residual returns of the market model. TURNOVER is the average daily stock trading turnover over [-252, -1] period. # obs is the number of observations. The dependent variable is EDDUM, which equals 1 if primary equity is issued, otherwise equals 0 if straight debt is issued. Panel A. Step 1: construction of predicted change in analyst coverage The dependent variable is change in analyst coverage. Independent variable Coefficient t P >|t| ANN -0.0646 a -8.28 0.000 MBKR 0.0913 a 2.85 0.004 ROA 1.1142 a 5.26 0.000 LNTA -0.0718 c -1.85 0.065 ABCUMRET 0.1672 a 4.98 0.000 RESIDSD -0.0437 -1.04 0.298 CONSTANT 1.7012 a 5.74 0.000 # OBS 3206 R 2 0.1079 Panel B. Comparing predicted change in analyst coverage between debt and equity offerings PCHAN is predicted change in analyst coverage, the predicted value the OLS model is Panel A. Predicted change in analyst Straight debt (1532 observations) Equity (1013 observations) Test H 0 : Difference=0 H a : Difference0 Mean 0.2885 1.2233 t-statistic: -40.516 a Median 0.3339 1.2331 z-statistic: -35.051 a 119

PAGE 127

Appendix C (continued) Panel C. Step 2: Expected information production on equity/debt choices Independent variable Model 1 Model 2 PCHANN 0.2666 b -0.4689 b INVANN -4.9240 a PCHANN*INVANN 2.7192 a ASYIND -1.1113 PCHANN*ASYIND 1.4405 a MBKR 0.1411 a 0.1739 a LNTA -0.4768 a -0.4913 a FIXTA 0.3745 b 0.3872 b TXTTA -3.7487 b -3.9178 b XRDTA 3.7352 a 3.4079 a LTDTA -0.0850 -0.1815 CFLTA 0.9047 0.7671 ROA -4.7516 a -3.6518 a ABCUMRET 0.6826 a 0.7845 a BETA -0.0765 -0.0375 RESIDSD 0.2332 a 0.1446 a TURNOVER 0.3117 c 0.4457 a CONSTANT 2.6550 a 2.8276 a Pseudo R 2 0.5685 0.5634 # OBS 3206 3206 a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 120

PAGE 128

Appendix D Reexamine change in analyst coverage and post-issue stock market performance via different sample classifications The whole sample firms are first classified into debt and equity groups, then firms within each financing group are further divided into quartiles based on change in analyst coverage Change in analyst coverage is the difference between the 12-month average analyst coverage after security offerings and the 12-month average analyst coverage before security offerings. The post-issue buy-and-hold abnormal return is calculated according to the following equation: 100*]11[22tMttjtjtRRabcumretny Where n=1, 2 or 3, =252 if n=1, =504 if n=2, and =756 if n=3. R jt is firm js daily return since the second day of security offerings. R Mt is the markets daily return, which is measured by the NYSE/AMEX/NASDAQ value-weighted index return. Panel A. Change in analyst coverage and post-issue one-year abnormal return Change in Analyst coverage quartiles Straight Debt (# observation ) Equity (# observation ) Quartile 1 (low) -1.6699 (437) -10.3018 a (306) Quartile 2 -8.0689 a (426) -9.8301 a (328) Quartile 3 -1.0999 (426) -10.5978 a (324) Quartile 4 (high) 2.3351 (444) 20.8304 b (320) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) -4.0050 b (-1.6619) -31.1322 a (-3.2156) Mean difference (T-stat, H 0 : Q2-Q4 0 H a : Q2-Q4 < 0) -10.4040 a (-4.2951) -30.6606 a (-3.1526) Panel B. Change in analyst coverage and post-issue two-year abnormal return Change in Analyst coverage quartiles Straight Debt (# observation ) Equity (# observation ) Quartile 1 (low) -1.7681 (373) -25.3936 a (261) Quartile 2 -13.1615 a (380) -22.6730 a (298) Quartile 3 -5.0522 c (396) -28.9909 a (279) Quartile 4 (high) -3.1363 (405) -2.3073 (274) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) 1.3682 (0.3448) -23.0864 a (-2.4954) Mean difference (T-stat, H 0 : Q2-Q4 0 H a : Q2-Q4 < 0) -10.0252 a (-2.5239) -20.3658 b (-2.1311) 121

PAGE 129

Appendix D (continued) Panel C. Change in analyst coverage and post-issue three-year abnormal return Change in Analyst coverage quartiles Straight Debt (# observation ) Equity (# observation ) Quartile 1 (low) -4.1682 (330) -38.4364 a (218) Quartile 2 -17.2955 a (350) -37.7807 a (265) Quartile 3 -13.0370 a (362) -33.8992 a (233) Quartile 4 (high) -7.0790 c (374) -10.9968 (220) Mean difference (T-stat, H 0 : Q1-Q4 0 H a : Q1-Q4 < 0) 2.9108 (-0.4865) -27.4396 b (-1.7460) Mean difference (T-stat, H 0 : Q2-Q4 0 H a : Q2-Q4 < 0) -10.2165 b (-1.7337) -26.7839 b (-1.6984) a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 122

PAGE 130

Appendix E Construction of PCHANN PCHANN is the predicted value of the following OLS model: RESIDSDABCUMRETLNTAROAMBKRANNEHATANNCH76543210 Where EHAT is the predicted value of the following probit model: TURNOVERRESIDSDBETAABCUMRETROACFLOWTALTDTAXRDTATXTTAFIXTALNTAMBKReddumob1211109876543210)1(Pr PCHANN is predicted change in analyst coverage. It is used to measure expected information production due to security offerings. LNTA is logarithm of book value of total assets. INVANN is the reciprocal of ANN, the pre-issue 12-month average analyst coverage. MBKR is Market-to-book ratio. FIXTA is fixed assets to total assets ratio. TXTTA is the ratio of tax payments to the book value of total assets. XRDTA is the ratio of expenditure in research & development to book value of total assets. We assume a firms R&D expenditure is zero if it is missing in COMPUSTAT. LTDTA is the ratio of long-term debt to book value of total assets. CFLTA is the ratio of cash flow to book value of total assets. ROA is return on assets. ABCUMRET is pre-issue abnormal cumulative return, calculated as a firms pre-issue cumulative daily returns over [-252, -30] period minus the NYSE-AMEX-NASDAQ value-weighted return over the same period. BETA is the beta coefficient of the market return in the market model, obtained by regressing a firms daily return on the NYSE-AMEX-NASDAQ value-weighted return over [-255, -46] period. RESIDSD is the standard deviation of the residual returns of the market model. TURNOVER is the average daily stock trading turnover over [-252, -1] period. # obs is the number of observations. The dependent variable is EDDUM, which equals 1 if primary equity is issued, otherwise equals 0 if straight debt is issued. Panel A: probit regression with robust errors Independent variable Coefficient Z-stat P > |Z| MBKR 0.197 a 3.93 0.000 LNTA -0.494 a -15.99 0.000 FIXTA 0.344 b 2.14 0.032 TXTAT -4.206 b -2.41 0.016 XRDTA 3.098 a 2.87 0.004 LTDTA -0.086 -0.39 0.704 CFLOWTA 1.068 1.18 0.241 ROA -3.747 a -4.31 0.000 ABCUMRET 0.880 a 7.56 0.000 BETA -0.043 -0.56 0.562 RESIDSD 0.170 a 3.61 0.000 TURNOVER 0.441 a 3.74 0.000 CONSTANT 2.457 a 8.52 0.000 Pseudo R 2 0.4828 # Observation 2545 123

PAGE 131

Appendix E (continued) Panel B: OLS regression with robust errors Independent variable Coefficient t-stat P > |t| EHAT 1.858 a 5.71 0.000 ANN -0.067 a -8.17 0.000 MBKR 0.239 a 5.81 0.000 LNTA 0.187 a 2.71 0.007 RESIDSD -0.199 a -3.64 0.000 ROA 2.172 a 5.33 0.000 ABCUMRET 0.105 a 1.67 0.095 CONSTANT -0.735 -1.27 0.206 R 2 0.1296 # Observation 2545 a indicates significance at 1%, b indicates significance at 5%, c indicates significance at 10%. 124

PAGE 132

About the Author Bingsheng Yi received a Bachelors Degree in Management Engineering in 1993 and a M.S. in Economics in 1996. Both degrees were from Beijing Materials College. After getting his master degree, Mr. Yi worked one year as a full time college instructor. Thereafter he spent two years in the Ph.D. program in Economics at University of Ottawa, finishing all courses and passing the comprehensive exams. In August 1999 Mr. Yi entered the Ph.D. program in Finance at the University of South Florida. While in the Ph.D. program at the University of South Florida, Mr.Yi passed all three-level examinations in the Chartered Financial Analyst Program and received the CFA designation in 2004. He has also coauthored two publications in Global Business and Finance Review and made several paper presentations at Financial Management Association Annual Meetings and the 11 th Conference of Global Finance Association.