Electronic copy available at: http://ssrn.com/abstract=1933995Electronic copy available at: http://ssrn.com/abstract=1933995

Lockup Agreements in Seasoned Equity Offerings: Evidence of Optimal Contracting

Jonathan M. Karpoff Foster School of Business University of Washington

206-685-4954 karpoff@uw.edu

Gemma Lee Stillman School of Business

Seton Hall University 973- 275-2218

gemma.lee@shu.edu

Ronald W. Masulis School of Banking and Finance University of New South Wales

612-9385-5860 ron.masulis@unsw.edu.au

First draft: September 21, 2011 Current draft: February 18, 2012

We wish to thank Oya Altinkilic, James Brau, Grant McQueen, and seminar participants at Louisiana State University and the University of Colorado for helpful comments and suggestions.

Electronic copy available at: http://ssrn.com/abstract=1933995Electronic copy available at: http://ssrn.com/abstract=1933995

Lockup Agreements in Seasoned Equity Offerings: Evidence of Optimal Contracting

Abstract: We document the frequent use of lockup agreements in seasoned equity

offerings (SEOs), and examine the determinants of their use, duration, and early release.

From 1996 through 2006, 93.8% of all SEOs included lockups, which is comparable to

the 96.6% lockup rate for IPOs during the same period. The likelihood of an SEO lockup

and its duration both are positively related to the degree of information asymmetry

between insiders and outside investors. Lockups tend to be released early when share

prices increase after the SEO. These results indicate that lockups help to guarantee the

SEO’s quality by guarding against opportunistic selling by insiders, particularly when the

opportunity for mispricing is large. That is, lockups represent a contracting solution to

economize on the asymmetric information and agency problems that plague equity issues.

[Note: This paper is accompanied by a lengthy Internet Appendix, which reports on robustness tests that are referenced but not fully reported in the main body of the paper. The Internet Appendix is intended to be detached from the main paper.]

1

1. Introduction

A lockup agreement is a legal contract under which insiders of a stock-issuing

firm are prohibited from selling shares for a limited period after the stock issue. It is well

known that lockups are common among initial public offerings (IPOs).1 Less well

known, however, is the fact that lockups are also common in seasoned equity offerings

(SEOs). In this paper we document the frequent use of lockups in SEOs, examine the

reasons for their use, and estimate the determinants of their duration. We find that lockup

agreements are more likely to be used, and span longer periods, when insiders have

strong informational advantages over outside investors. Lockups are released early when

share prices rise substantially following SEOs, i.e., when the probability that new

investors will suffer losses turns out to be small. These results indicate that lockups and

their duration help to guarantee an SEO’s quality by guarding against opportunistic

trading by insiders, particularly when they have information that other investors do not.

This implies that lockups are a contracting solution for the asymmetric information and

agency problems that plague equity offerings.

In our sample from 1996 through 2006, 93.8% of all SEOs include lockups, which

is comparable to the 96.6% rate for lockups in IPOs over the same period. The length of

a typical SEO lockup, however, is 90 days, which is only half that of a typical IPO

lockup. The average SEO lockup period also has become shorter and more standardized

over time.

An additional feature of SEO lockups is that many lockups are released early,

which means that the investment bank allows company insiders to sell shares before the

1 See Field and Hanka (2001), Bradley, Jordan, Roten, and Yi (2001), Brav and Gompers (2003), Cao, Field, and Hanka (2004), Brau, Lambson, and McQueen (2005), and Goergen, Khurshed, and Renneboog (2006).

2

lockup period officially expires. We find that early releases occur in 39.5% of SEOs with

lockup agreements. As a result, many actual, or ex post, lockup periods are considerably

shorter than their initial ex ante periods. Moreover, the trend is for early releases to

become more common over time. A total of 34.8% of the lockups in 1996 were released

early, compared to 53.5% in 2006. This has led to an increasing divergence between the

ex ante and ex post lockup periods.

In this study, we examine three questions related to the use of SEO lockups. First,

what firm and managerial characteristics determine whether a firm uses a lockup?

Second, what determines the length of the ex ante lockup period? And third, what

determines whether and when an early release is granted? Empirical evidence about

these questions can shed light on the reasons firms and underwriters use lockup

agreements in equity offers.

Prior research on lockup agreements is almost exclusively based on IPOs.2 Yet,

SEOs provide an important advantage over IPOs when examining the motives for

employing lockup agreements. This is because SEO lockups and their durations are more

variable than IPO lockups. As reported in Table 1, 81.5% of IPO lockups are for the

same duration – 180 days – implying that IPO lockups tend to reflect boilerplate

arrangements. In contrast, only 64.4% of SEO lockups are for their modal period of 90

days. Since many SEO lockups are released early, the ex post lockup durations are even

more variable. Also, the frequency of lockups and their ex ante and ex post lengths

2 An exception is Brau, Lambson, and McQueen (2005), who examine a pooled sample of IPO and SEO lockups. Brau et al. develop a model of lockup choice and report that lockup length is positively related to transparency measures such as firm size and industry type. These empirical results are consistent with our finding that lockup likelihood and length are positively related to our information asymmetry measures. However, contrary to our findings, Brau et al. predict and find that lockup length is negatively related to idiosyncratic risk.

3

change noticeably over the sample period. This greater variation in the features of SEO

contracts provides a greater opportunity to gain insight into the cross-sectional

determinants of lockup use and duration. The existence of an active secondary market

for shares also enables us to investigate a broader range of asymmetric information

measures for these firms.

We find that the primary determinant of lockup agreements and ex ante lockup

periods is the degree of information asymmetry between insiders and outside investors.

Previous research has identified many proxies for information asymmetry. We use factor

analysis to summarize many of these proxies in a single information asymmetry factor.

Holding other control variables at their medians, a one standard deviation increase in the

information asymmetry factor corresponds to a 5.8% increase in the likelihood of a

lockup and a 15-day increase in the lockup period. The likelihood and duration of lockup

agreements are smaller among SEOs using an accelerated underwriting mechanism

relative to conventionally registered and underwritten SEOs. This is consistent with the

notion that information asymmetry problems are smaller among firms that are able and

willing to use an accelerated underwriting procedure (e.g., see Bortolotti et al. 2008; Gao

and Ritter 2010). Nevertheless, the effect of information asymmetry on lockup likelihood

and duration is pronounced for both accelerated and traditional SEOs.

Among SEOs that have lockups, the probability of an early release increases with

post-SEO stock price performance. Holding other control variables at their median

values, a one standard deviation increase in the cumulative return over days 0 through +5

relative to the SEO issue date corresponds to a 2.8% increase in the likelihood of an early

release and 2.4 day decrease in the ex post lockup period. This indicates that strong stock

4

price performance helps to alleviate the information asymmetry problem, decreasing the

value of the lockup to outside investors.

These results are consistent with the view that lockups guarantee the quality of a

new equity issue and assure investors that insiders will not trade opportunistically shortly

after the issue date. The guarantee is particularly important when the opportunity for

opportunistic behavior is greatest, i.e., when information asymmetry is large. If the post-

SEO stock price increases above the offer price, however, the marginal value of the

lockup guarantee decreases and the likelihood of an early release rises. These results

highlight the importance of information asymmetry in new security issues and the role of

lockups as a contracting solution for the problems that arise with such asymmetry.

The paper proceeds as follows. Section 2 discusses the determinants of lockup

agreements and describes the measures used in our empirical tests. Section 3 presents

results regarding the determinants of lockups and lockup periods. Section 4 examines

early releases and the ex post lockup period. Section 5 concludes.

2. Data and empirical method

The initial SEO sample is obtained from the Securities Data Company (SDC)

New Issue database for the 1996 to 2006 period.3 We require SEOs to be underwritten

public offers of common stock by U.S. corporations, listed on NYSE, NASDAQ, or

AMEX. The sample includes primary and secondary offers, firm commitments, and

accelerated offers, including shelf issues. We exclude rights issues, best effort

underwritings, ADRs, closed-end funds, unit investment trusts, REITs, limited

3 We begin our sample period at the start of 1996 since this is the earliest year that electronic prospectuses are available. We manually examined all SEO prospectuses in our sample period and corrected approximately 18% of the SDC lockup information, which was incorrect.

5

partnerships, unit offerings and SEOs with offer prices less than $5. Rights and standby

offers, combined offers of common stock and other securities are also excluded. These

exclusions yield a sample of 2,579 SEOs.

Many studies, beginning with Lease, Masulis, and Page (1991), adjust for the fact

that SEO issue dates in the SDC database do not take into account that many offerings are

launched after the close of daily exchange trading. In these cases, the date when the

market is affected by the offering is actually the next trading day. We follow the

Safieddine and Wilhelm (1996) approach to determining the actual SEO event date. Their

method uses trading volume surges to detect the effective issue date. If the trading day

following the issue date has more than twice the trade volume of the issue date, then the

next trade day is treated as the SEO issue date. Corwin (2003) also uses this approach and

Altinkilic and Hansen (2003) document that this approach yields almost identical results

to manually searching newspaper articles for the times of day of the offerings.

To conduct some empirical tests we require firms to have daily returns data

available from CRSP from 90 trading days before through 30 trading days after the public

offering dates, and annual financial statement data available from COMPUSTAT, which

reduces the sample to 2,424 SEOs. Due to the additional data requirements associated

with the information asymmetry factors and inside ownership and trading information,

our final sample consists of 1,597 SEOs. For comparison purposes, we also collect data

from SDC on 1,926 IPOs during the same 1996-2006 period. The IPO comparison

sample has the same requirements as the IPO sample, with the additional requirement that

we exclude equity carve-outs and reverse LBOs.

6

Table 1 reports on the use of lockups in the SEO and IPO samples. Over the 11-

year period, 93.8% of all SEOs included lockups, which is comparable to the 96.6% IPO

lockup rate. The data in Table 1 also indicate that the typical SEO lockup period is

shorter than its IPO counterpart. Throughout the 1996-2006 period, the modal SEO

lockup duration is 90 days, compared to 180 days for IPO lockups. Over the sample

period, the length of the lockup period has become both shorter and more standardized.

In 1996, 53.7% of all SEO lockups were for 90 days and 19.9% were for 180 days, while

by 2006, 61.2% of all lockups were for 90 days and only 3.5% were for 180 days. In

contrast, lockup periods in IPOs stayed relatively constant over the same period. In 1996,

82.4% of our sample IPOs had lockup periods of 180 days, while in 2006, 86.4% of our

sample IPOs had 180-day lockup periods.

Table 2 highlights another important feature of SEO lockups. Many lockup

agreements have early releases, meaning that the underwriter allows issuer insiders to sell

their shares before a lockup’s official expiration date. Panel A of Table 2 reports that

early releases occur in 1,057, or 39.5%, of the 2,419 SEOs with lockup agreements. We

determine the date of an early release using insider sale records from Thomson

Financial’s Inside Filing database. Most early releases occur shortly after the issue date –

on average, 18.2 days after the issue date and 81.5 days before the stated expiration date

in the lockup agreement. As a result, many actual, or ex post, lockup periods are

substantially shorter than their initial ex ante periods. Averaging across our full sample,

Panel B of Table 2 shows that the median ex ante lockup period is 90 days (mean = 90.9

days). In Panel C, however, we see that the median ex post lockup period is only 70 days

and the average ex post lockup period is 60.8 days. In contrast, Brav and Gompers (2003)

7

report that early releases occur in only 15% of their sample of IPOs, and tend to occur

much closer to the official expiration date than we observe for SEO lockups.

2.1. Lockups and information asymmetry

Previous research using IPO data focuses on how lockups can assuage investor

concerns about the quality of an IPO. Given the lack of secondary market prices,

uncertain future liquidity, and the paucity of analysts following the stock, information

asymmetry between insiders and outsiders is particularly severe. Field and Hanka

(2001) argue that lockups help managers to convey favorable inside information to

investors by committing to hold undiversified equity positions during the lockup period.

This idea is captured in models by Courteau (1995) and Brau et al. (2005), in which

insiders’ information is signaled by their commitment to hold shares for a specified

period of time. Brav and Gompers (2003) argue that lockups limit opportunistic trading

by insiders during the lockup period. These arguments imply that lockups convey

benefits particularly when there is large information asymmetry between managers (and

other insider shareholders) and outside investors. This is because the commitments of

managers and other insiders to continue to hold shares, and not to trade opportunistically,

have value when managers and other insiders have proprietary information that

shareholders lack. Thus, to the extent that information asymmetry explains the use of

lockup agreements, the likelihood and duration of a lockup will be positively related to

the information asymmetry between insiders and outside investors. We call this the

information asymmetry explanation of lockup agreements.

8

The finance literature has proposed numerous ways to measure information

asymmetry. Yet there is little agreement to date on what is the best measure. Thus, we

conduct empirical tests using eight common measures of information asymmetry.

Additional analysis using several alternative measures yield similar results to those

reported here.4

Previous research indicates that the following five variables are negatively related

to information asymmetry between managers/insiders and outside investors:5

• Firm size, measured as the natural log of the book value of assets at the year-end

prior to the SEO issue date;

• Firm age, measured by the number of years since a firm’s IPO;

• Number of analysts providing firm earnings forecasts in the year before the SEO,

taken from the I/B/E/S database;

• Tangible assets, measured as the ratio of property, plant, and equipment (PPE) to

total assets at the year-end prior to the SEO; and

• Number of prior stock offers since going public.

The rationale for these measures is that investors tend to have relatively good information

about firms that are larger, older, and followed by a greater number of analysts, which

decreases the information asymmetry between managers and outside investors. Firms

with substantial physical assets tend to be relatively easy to value compared to firms

whose values depend more heavily on growth opportunities, so there generally should be

4 Other proxies for information asymmetry include analyst disagreement, Tobin’s q, the amount of underpricing at the SEO, and industry-adjusted return volatility. As reported in the Internet Appendix, our main results are qualitatively similar using these proxy variables as well. 5 For firm size, see Bhushan (1989), Barth, Kasznik, and McNichols (2001), and Frankel and Li (2004); for firm age, see Clarke and Shastri (2001), Cai, Liu, and Qian (2009), Leary and Roberts (2010), and Lowry, Officer, and Schwert (2010); for the number of analysts, see Brown et al. (2009) and Hong et al. (2000); and for tangible assets, see Barth, Kasznik, and McNichols (2001) and Leary and Roberts (2010).

9

smaller discrepancies between insiders’ and outsiders’ knowledge. Firms that previously

raised capital through prior stock offers are likely to be closely examined and vetted by

underwriters and other sophisticated investors. Such prior vettings are likely to decrease

any informational advantages insiders might have.

Three additional variables have a positive relation to information asymmetry:

• Average bid-ask spread is the average bid-ask spread, scaled by price, calculated

from CRSP closing daily stock quotes and prices during trading days -90 through -

11 relative to the issue date.

• Return volatility is the standard deviation of the daily stock return over trading days

-90 through -11 relative to the issue date.

• Abnormal accruals is the absolute value of the residual from the Kothari, Leone,

and Wasley (2005) model of performance matched discretionary accruals. Details

of the calculation are provided in the Appendix.

Corwin (2003) and others argue that information environments in which insiders know

more than outside investors are associated with larger bid-ask spreads, to compensate

uninformed investors and market makers for bearing greater risk of trading with informed

investors at stale prices.6 Similarly, Coles, Daniel, and Naveen (2006), Drucker and Puri

(1999), Altinkilic and Hansen (2003), and Corwin (2003) argue that high return volatility

reflects a noisy information environment in which the cost to outside investors of

obtaining current information is relatively high, also leading to greater information

asymmetry between insiders and outside investors.7

6 See also Treynor (1971), Copeland and Galai (1983), Glosten and Milgrom (1985), Easley and O’Hara (1987), Corwin (2003), and Duarte-Silva (2010). 7 In the Internet Appendix, we report results using industry-adjusted return volatility. The results are similar to those reported in the main tables.

10

Francis et al. (2004, 2005) argue that Abnormal accruals is a systematic priced

risk factor and find a positive relation between it and the cost of equity and debt. In a

similar context, Lee and Masulis (2009) argue that firms’ financial statements are less

informative when they contain large idiosyncratic components, as reflected in Abnormal

accruals. This increases the degree of information asymmetry between managers and

outside investors, who tend to rely heavily on financial statements to assess a firm’s

market value.

One approach to estimating the importance of information asymmetry is to

include all eight proxies in tests that examine the presence and duration of lockup

agreements. However, a problem with this approach is that these proxies are correlated

to varying degrees. For example, firm size is positively correlated with firm age

(coefficient = .40) and the number of previous stock offers (coefficient = .38), and

negatively correlated with return volatility (coefficient = –.59). Including all proxies at

once creates a multicollinearity problem (e.g., see Maddala 1983) and also induces an

attenuation bias for the estimated coefficients (e.g., see Lubotsky and Wittenberg 2006;

Boone et al. 2007). Another alternative is to construct an equally weighted index from a

group of these individual measures. This approach, however, presents several problems.

Most importantly, it arbitrarily assigns equal weight to each measure and is affected by

the units of measurement for each measure.

As an alternative approach, we use factor analysis to construct a single

information asymmetry measure from the eight primitive measures. Factor analysis

aggregates the correlated information from the eight measures without inducing

multicollinearity or attenuation bias problems. In the Internet Appendix, we report

11

sensitivity tests that use the eight information asymmetry variables individually, as well

as several other measures of information asymmetry. Using these various measures of

information asymmetry, we obtain similar results to those reported in our main tests.

Table 3 reports the results of the factor analysis based on the eight above

mentioned information asymmetry measures. The eigenvalue for the first factor is 1.93,

indicating that it summarizes a significant amount of the variation in the eight original

measures. Importantly, the signs of all the factor loadings are consistent with intuition,

although the signs are reversed. Firm size, Firm age, Number of analysts, Tangible

assets, and Number of prior stock offers all load negatively, whereas Average spread,

Return volatility, and Abnormal accruals all load positively. Note that these signs are the

opposite of the predicted correlations between the proxy variables and information

asymmetry, indicating that the factor picks up what might be called “information

symmetry” characteristics of the firm. In addition, the Kaiser-Meyer-Olkin (KMO)

measures are sufficiently high for all of the variables used to construct the factor.8 To

transform this factor into a composite measure of information asymmetry (rather than

information symmetry), we multiply it by –1 before using it in subsequent empirical tests.

Combined with the factor’s high eigenvalue, the factor loadings indicate that the first

factor is a reasonable composite measure of information asymmetry.

As illustrated in the scree plot in Table 3, only the first factor captures a

significant amount of variation in the eight original variables. The eigenvalue for the

second factor is 0.220 and for the third factor it is 0.182 – far below the levels that are

commonly considered acceptable for reliable factors. Thus, we conclude that the first

8 Historically, the following labels are given to KMO values (Kaiser 1974): 0.00 to 0.49: unacceptable, 0.50 to 0.59: miserable, 0.60 to 0.69: mediocre, 0.70 to 0.79: middling, 0.80 to 0.89: meritorious, 0.90 to 1.00: marvelous.

12

factor is the only statistically reliable measure of the common information contained in

the eight primitive measures. In our subsequent empirical tests of lockup agreements, we

use only the first factor to measure information asymmetry.

2.2. Accelerated SEOs

One potentially important reflection of information asymmetry is the SEO’s

registration and marketing method. In our empirical tests we identify SEOs that are

executed using accelerated marketing procedures, including accelerated bookbuilt, block

trade, and bought deals. Bortolotti, Megginson, and Smart (2008) report that firms using

accelerated SEOs tend to be relatively large and well known, and Ritter and Gao (2010)

report that most accelerated SEOs use shelf registration. To qualify to use a shelf

registration, a firm must meet certain minimum size and reporting requirements (e.g., see

http://www.sec.gov/rules/final/2007/33-8878.pdf), and Bhagat, Marr, and Thompson

(1985), Denis (1991), and Lee and Masulis (2009) find that shelf issues have relatively

low underwriter spreads. Bethel and Krigman (2009) find that shelf registration is

particularly costly for equity issuers with high information asymmetry. Together, these

studies imply that firms with high information asymmetry tend to not qualify for a shelf

offer, or find an alternative registration method more preferable.

These results indicate that accelerated SEO issuers tend to be characterized by

less information asymmetry than conventional SEO issuers, implying that lockups in

accelerated SEOs should be less likely and of shorter duration. We test this inference by

including a separate indicator variable for accelerated SEOs in the empirical tests. In

robustness tests reported in the Internet Appendix, we also include an accelerated SEO

13

indicator in constructing the information asymmetry factor. This approach yields results

for the information asymmetry factor that are similar to those reported in our main tables.

The fact that firms must first qualify to use an accelerated SEO underwriting procedure,

and then select to use it, introduces potential selection biases into our tests. To address

this issue we conduct separate tests on the accelerated and conventional SEO subsamples.

These tests examine the hypothesis that, conditional upon the choice of registration and

flotation method, lockup agreement use and duration increase with the information

asymmetry associated with the stock at the SEO issue date.

2.3. Control variables

2.3.1. Insider demand to trade

Lockups are not costless, since they impose liquidity constraints on managers and

blockholders who are prohibited from selling for the life of the lockup agreement. To

measure insider demand to sell stock, we use insider pre-SEO ownership and trading

activities drawn from Thomson Financial’s Insider Filing (TFIF) database.9 Insider

ownership % is defined as the percent of outstanding shares owned by all officers,

directors, and 10% blockholders one month before the SEO filing date.10

9 These data are derived primarily from SEC Forms 3, 4, and 5, and proxy statements. The TFIF database has the advantage of having ownership and trading activity information just prior to the SEO. This is important because an SEO can substantially change insider ownership. Proxy statements often report post-SEO information (or pre-SEO data that are stale). We exclude information that does not meet Thomson’s confidence level concerning the accuracy of a particular data record. The TFIF database provides nine data cleansing indicators. Among those, we eliminate cases marked ‘S’ (no cleansing attempted; security does not meet the collection requirement) and ‘A’ (Numerous data elements were missing or invalid; reasonable assumption could not be made). 10 Blockholders include beneficial owners whose securities are registered in the name of a broker or nominee. “Insider reports” collects information on the beneficial ownership of equity securities for (1) any director or officer of an issuer with a class of equity securities registered pursuant to Section 12 of the Securities Exchange Act of 1934; (2) any beneficial owner of greater than 10% of a class of equity securities registered under Section 12 of the 1934 Act, as determined by voting or investment control over

14

Insider demand to sell stock can also vary across managers and other insiders.

We construct three measures of insiders’ demand to trade. Insider ownership % captures

insiders’ potential demand to sell shares and the degree to which new shareholders are

vulnerable to having insiders sell overvalued shares immediately after an SEO. Net

selling frequency is the number of insider sales minus the number of insider purchases of

stock over the six months before the SEO filing date, scaled by shares outstanding before

the SEO. Positive Net selling frequency implies that an insider is a net seller, while a

negative value indicates that an insider is a net buyer. Net selling value is the aggregate

dollar value of all insider sales minus their purchases over the six months before the SEO

filing date, scaled by the firm’s market capitalization at the year-end prior to the offer

filing date.

Whether these measures are positively or negatively related to the likelihood of a

lockup depends on insider motives for selling. If insiders sell shares to rebalance their

portfolios or to obtain liquidity, then a lockup agreement imposes a cost on insiders

without creating a benefit for outside investors. We would expect both measures to be

negatively related to the likelihood and duration of lockup agreements. In contrast, if

insider trading is frequently motivated by proprietary information, then Net selling

frequency and Net selling value could reflect the likelihood of informed trading. It is

precisely when insiders could trade on information that lockup agreements are valuable to

outside investors. Hence, if Net selling frequency and Net selling value reflect the degree

of information asymmetry, they should be positively related to the likelihood and

duration of a lockup.

the securities pursuant to Rule 16a-1(a)1; and (3) any trust, trustee, beneficiary, or settler required pursuant to Rule 16a-8.

15

In calculating Net selling frequency and Net selling value, we eliminate share

acquisitions or sales triggered by stock grants, awards, ongoing acquisition plans,

employee benefit plans, expirations of derivative positions, gifts, transfers pursuant to

qualified domestic relations orders, and “value added transactions”, which are generally

miscoded transaction records. We include three types of insider trades: open market

trades and private purchases and sales of shares; exercises of derivative securities; and

beneficial interest transactions. This definition of insider trades is broader than that used

in some previous studies (e.g., Lee, 1997; Karpoff and Lee, 1991). This is because

lockup agreements typically restrict not only open market sales, but also shares

transferred, assigned, pledged, or hypothecated, or subject to any hedging, short sale,

derivative, put, or call transactions.11 Insider ownership % is measured one month before

the SEO filing date. As reported in the Internet Appendix, however, the results are robust

to using other measurement periods for these three variables.

2.3.2. Other control variables

In the empirical tests, we include a number of control variables that capture issuer

and SEO characteristics. These control variables are defined in the Appendix. Several

control variables capture issuer characteristics such as ROA, Leverage, and Share

turnover. Other control variables reflect SEO characteristics such as Underwriter rank,

11 For example, the following statement is the lockup agreement example on the prospectus for Aeroflex Inc. on May 12, 2000: “We, as well as our executive officers and directors, have agreed to a 90-day ‘lock up’ with respect to approximately 2,400,000 shares of common stock, and certain other of our securities that they beneficially own, including securities that are convertible into shares of common stock and securities that are exchangeable or exercisable for shares of common stock. This means that, subject to certain exceptions, for a period of 90 days following the date of this prospectus, we and such persons may not offer, sell, pledge or otherwise dispose of these securities without the prior written consent of CIBC World MarketsCorp.”

16

the natural log of SEO proceeds (Log(proceeds)), and the percentage of the stock offering

sold by existing shareholders rather than the issuing firm (Secondary shares %).

Some control variables can reflect outside investor perceptions of firm risk and

the quality of the SEO, and therefore can be related to the marginal benefits of a lockup.

The direction of effect, however, is ambiguous. For example, the likelihood and duration

of a lockup could be either positively or negatively related to an underwriter’s reputation.

On one hand, underwriter reputation can signal issue quality, decreasing the incremental

benefit of a lockup agreement. This follows since more reputable underwriters have

stronger incentives to conduct more thorough due-diligence investigations and employ

stricter origination standards for the SEOs they underwrite. This would make the

likelihood and duration of a lockup negatively related to Underwriter rank. On the other

hand, reputable underwriters may have a greater ability to extract lockup agreements

from managers as part of the compensation they receive for their higher quality

certification services, and have greater incentives to require them. In this case, the

likelihood and duration of a lockup would be positively related to Underwriter rank.

Likewise, the relation between Secondary shares % and the use and duration of a

lockup is ambiguous. On one hand, larger sales by insiders can signal large information

asymmetry if it reveals that managers are attempting to take advantage of temporary

overpricing in the firm’s stock. This would imply a positive relation between Secondary

shares % and the likelihood and duration of a lockup. On the other hand, the risk that

managers might sell opportunistically could cause underwriters to oppose a large

secondary offering by insiders unless the information asymmetry is small, since the

underwriters must agree to the sale. Indeed, underwriters might allow secondary shares

17

only when they are confident that the issue is not overpriced based on non-public firm

specific information. In such a case the incremental bonding benefits of a lockup

agreement would be small, and Secondary shares % would be negatively related to the

likelihood and duration of a lockup.

In robustness tests, we include additional control variables, including the firm’s

number of business segments and capital expenditures, and the underwriter’s

shareholdings, all measured before the offer. Including these additional controls does not

alter our main conclusions. Most importantly, the likelihood and length of a lockup

period is positively and significantly related to the information asymmetry variable in all

models tested.

3. The determinants of lockups and lockup periods: Empirical tests

3.1. Univariate comparisons

Table 4 compares the mean and median values of the information asymmetry and

control variables for firms with and without lockup agreements. Panel A reports results

for the information asymmetry variables, and Panel B contains the control variables. The

mean value of the information asymmetry factor is 0.033 for firms with lockups

and -0.172 for firms without lockups, indicating that information asymmetry is larger for

SEO issuers using lockups. Among firms with lockups, the information asymmetry

factor increases with the length of the lockup period. The mean value is 0.394 for firms

with lockup durations greater than 90 days, compared to 0.007 for firms with lockups of

exactly 90 days and –0.414 for firms with lockups shorter than 90 days. These

differences are statistically significant, and are consistent with the information

18

asymmetry hypothesis explanation for lockup agreements. Reinforcing this conclusion,

IPO lockup periods are typically longer than SEO lockups as shown in Table 1, which is

consistent with IPOs having greater information asymmetry than SEOs due to the lack of

market prices before the offering.

Panel A of Table 4 also reports on the eight measures used to construct the

information asymmetry factor. A large majority of these individual comparisons also are

consistent with the information asymmetry hypothesis. The individual results for five of

the eight measures indicate that firms with lockups have significantly more information

asymmetry than firms without lockups. That is, firms with lockups are smaller and

younger, and exhibit fewer prior stock offers, larger return volatility, and larger abnormal

accruals, than firms without lockups. Firms with lockups also are followed by fewer

analysts, although this difference is not statistically significant.

Among firms with lockups, the results for all eight individual measures indicate

that the degree of information asymmetry is positively related to the length of the lockup

period, and the differences are statistically significant for six of the eight measures.

These results indicate that, in general, the individual components of the information

asymmetry factor are related to lockup periods in the manner that is predicted by the

information asymmetry hypothesis.

Sample comparisons for two individual measures – Tangible assets and Average

spread – are not consistent with the information asymmetry hypothesis. In particular, the

fraction of tangible assets is relatively high for firms with lockups, and the bid-ask spread

is relatively low. We note, however, that among firms with lockups, tangible assets are

relatively low for firms with longer duration lockups, and the bid-ask spread is relatively

19

high for firms with longer duration lockups. So among firms that have lockups, even

these two measures are consistent with the information asymmetry hypothesis.

In sensitivity tests we consider different combinations of variables to include in

the information asymmetry factor. As one example, in one test we exclude Tangible

assets and Average spread. As reported in the Internet Appendix, the results are similar

to those reported here, and consistently point to a link between information asymmetry

and the use and duration of lockup agreements in SEOs.

Panel B of Table 4 reports the univariate comparisons for the control variables.

All three measures of insiders’ demand to trade – Insider ownership, Net selling

frequency, and Net selling value – are larger, on average, for firms with lockups than

firms without lockups. Among firms that have lockups, lockup duration is positively

related to Insider ownership, and negatively related to Net selling frequency and Net

selling value. This suggests that greater insider ownership is associated with greater

information asymmetry, increasing the bonding benefits of a lockup agreement and

prompting the use of a lockup. The results for the two selling activity measures, in

contrast, are mixed. Across all firms, insider selling activity is associated with a greater

likelihood of a lockup. Among firms with SEO lockups, however, prior selling activity is

associated with shorter lockup periods, possibly reflecting greater insider resistance to

signing lockups when insiders often sell shares.

Several of the control variables are significantly related to the existence or length

of a lockup, including Log(proceeds), Leverage, and Underwriter rank. One noteworthy

result is that accelerated SEOs are more common among short duration lockups (< 90

days) than for long duration lockups (> 90 days), which is consistent with lower

20

information asymmetry in accelerated SEOs. Among firms with lockups, the mean

fraction of the issue that consists of secondary shares also is higher for SEOs with

relatively shorter lockup periods.

3.2. Regression analysis of the determinants of lockup agreements

Table 5 reports the results of nine logistic regressions in which the

(untransformed) dependent variable equals one for SEOs with lockup agreements and

zero for SEOs without lockups. We use the rare event logit model (see King and Zeng

2001) because only 6.2% of all SEOs exclude lockup agreements. Models 1-3 include all

SEOs with sufficient data.

Model 1 of Table 5 shows that the likelihood of a lockup is positively and

significantly related to the information asymmetry factor (p-value < .01). It remains

positive and significant (p-value < .01) in Model 2 when we include controls for insider

ownership and prior trading activity measures. In Model 3 we include all control

variables, and again, the coefficient for the information asymmetry factor is positive and

significant (p-value < .01).

As discussed above, the choice of flotation method can reflect differences in

information asymmetry across firms. In Model 3, the coefficient for the accelerated SEO

indicator is negative with a p-value of .105, suggesting that the likelihood of a lockup

agreement is relatively small in accelerated SEOs. This is consistent with the argument

that the net benefit of a lockup agreement is relatively small for firms that use accelerated

SEOs because they have a low degree of information asymmetry. A binary control for

accelerated SEOs, however, may be insufficient to account for the effects of differences

21

in information asymmetry across the two types of SEOs. As an alternative treatment, in

Models 4 – 9 we drop the indicator variable for accelerated offers and instead partition

the SEO sample into conventional and accelerated registration methods.

Models 4 – 6 report coefficient estimates for the subsample of conventional SEOs

and Models 7 – 9 report coefficient estimates for the subsample of accelerated SEOs.

Within each subsample, we consistently find that the information asymmetry coefficient

is positive and statistically significant in all three model specifications. Thus, even

conditional on the registration and flotation method, the likelihood of a lockup agreement

is positively related to the degree of information asymmetry. These findings are

consistent with the information asymmetry explanation of lockup agreements.12

In the full sample and the conventional SEO subsample, the probability of a

lockup is positively related to insider Net selling frequency (e.g., with a p-value of .019 in

Model 3). For the accelerated SEO subsample, the Net selling value coefficient is

positive and significant. These positive coefficients indicate that lockups are more likely

when insiders have an active trading history. This, in turn, suggests that investors view

frequent trading by insiders as motivated by private information, increasing the likelihood

of opportunistic informed trading and increasing the net benefit of a lockup agreement.

Among the other control variables in Model 3 of Table 5, the likelihood of a

lockup is positively related to offer size, measured by Log(proceeds), and is negatively

related to Share turnover. These results are similar to those found for the conventional

SEO subsample. Among accelerated SEOs, however, other control variables have

statistically significant coefficients. In particular, the likelihood of a lockup in an

12 The marginal effect of information asymmetry on the likelihood of a lockup is not significantly different between the two subsamples (Chi-squared value of 0.81 with p-value of 0.368).

22

accelerated SEO is positively related to offer proceeds and is negatively related to

underwriter rank and the fraction of secondary shares.

In the Internet Appendix, we report on tests in which we replace the information

asymmetry factor with the eight direct measures of information asymmetry, including

them individually and simultaneously in the regressions. In every model tested, the

coefficients for all eight measures have the predicted signs. The coefficients are

persistently statistically significant for four of the information asymmetry measures: Firm

size, Number of prior stock offers, Return volatility, and Abnormal accruals. The

coefficients on the other four measures are not always statistically significant. This,

however, is consistent with the attenuation bias that arises from including multiple

measures of the same underlying economic variable (information asymmetry).

3.3. Determinants of the ex ante lockup period

Table 6 reports on tests of the determinants of the length of the ex ante lockup

period. The tests parallel those in Table 5, except that we estimate Tobit models in which

the dependent variable is the ex ante length of the lockup period. The length is set to zero

for firms with no lockups.

The coefficients on the information asymmetry factor are positive and significant

in all nine models, indicating that information asymmetry affects not only the decision to

employ a lockup, but also the length of the lockup period. In Model 3, the coefficient on

the accelerated SEO indicator variable is negative and significant, indicating that the ex

ante lockup period tends to be shorter for accelerated SEOs than for conventional SEOs.

We infer that the degree of information asymmetry tends to be lower among accelerated

23

SEOs. Nonetheless, information asymmetry is associated with a longer lockup period

even within the subsample of accelerated SEOs, as the information asymmetry

coefficients in Models 7-9 are positive and statistically significant.

Models 2-3 and 5-6 of Table 6 indicate that the ex ante length of the lockup

period is positively related to Net selling frequency in the overall sample and the

subsample of conventional SEOs. This indicates that frequent selling by insiders

corresponds to longer lockup periods. Again, this is consistent with the view that such

selling tends to be motivated by private information – or at least that outside investors

think that this is the case – increasing the value of a lockup agreement to reassure

investors concerned about buying overpriced equity.

As in Table 5, several of the control variables are significantly related to the

length of the lockup period. For the overall sample and among conventional SEOs, the

length of the lockup period is positively related to firm leverage and negatively related to

share turnover. The positive leverage association could reflect the fact that higher

leverage makes a stock price more sensitive to information about firm value, which can

make investors more concerned with asymmetric information. Among accelerated SEOs,

the length of the lockup is positively related to offer size and negatively related to share

turnover and the fraction of secondary shares. Our most persistent finding, however, is

that the likelihood of a lockup agreement (Table 5) and the ex ante lockup period length

(Table 6) are positively related to the degree of information asymmetry between insiders

and outside investors. These results are consistent with the information asymmetry

explanation for lockups.

24

4. Early releases and the ex post lockup period

4.1. Early releases

Early releases occur when the lead underwriter agrees to let an insider sell shares

before the end of the formal (ex ante) lockup period. As documented in Table 2, nearly

40% of SEO lockups are released early. The conditions under which lockups are released

early provide an additional opportunity to shed light on the importance of information

asymmetry in explaining lockup agreements.

The information asymmetry hypothesis implies that the net benefit of a lockup

agreement increases with the degree of information asymmetry. This is consistent with

the results in Tables 4 – 6, which show that information asymmetry is a primary

determinant of lockups and lockup durations. Information asymmetry is important

because outside investors are concerned about issue quality and insiders’ ability to trade

opportunistically. It follows that increases in share price after an SEO can mitigate

investor concerns about information asymmetry, decreasing the net benefit of a lockup

and increasing the likelihood of an early release. Stated differently, an increase in the

secondary market price assures investors that they did not overpay when they participated

in the SEO, which is a primary concern when insiders have an informational advantage.

In addition, an increase in share price decreases the likelihood that underwriters will find

it necessary to undertake expensive price support activities, thereby also encouraging an

early release of the lockup.

Table 7 reports on tests of the proposition that positive share returns prompt early

lockup releases. Each of the nine models reported are logistic regressions in which the

sample is restricted to SEOs with lockups and the dependent variable equals one if there

25

is an early release and zero otherwise. CR(0,5) is the share return cumulated over trading

days 0 through +5 relative to the SEO’s issue date.13 We include controls for the degree

of information asymmetry and the ex ante number of lockup days in the initial agreement.

In Model 1 of Table 7, the coefficient for CR(0,5) is positive and significant (p =

.045). This is consistent with the view that investor concern about information

asymmetry is mitigated by strong stock price performance immediately following the

SEO. This result is observed in all of the models, including the subsamples of

conventional and accelerated SEOs.

In Models 1 and 4 the coefficient on the information asymmetry factor is positive

and significant at the 10% level or better, suggesting that the probability of an early

release increases with the degree of information asymmetry. To investigate this further,

in Model 2, 5, and 8 we include the interaction of the information asymmetry factor and

CR(0,5). Our thinking is that the marginal effect of a share price increase on the

probability of an early release could increase with the prior degree of information

asymmetry. This would imply that a rising share price after the SEO has a large effect on

the early release decision particularly when investors are at an informational

disadvantage. In general, however, the coefficient on the interaction term is not

statistically significant.

In Model 3 of Table 7, the probability of an early release is positively related to

Insider ownership, insider Net selling frequency, and insider Net selling value. Insider

ownership is significant also in the subsample of conventional SEOs (Model 6), whereas

Net selling frequency and Net selling value are positive and significant in both of the

13 We find similar results when we compute returns over a shorter window, such as (0,3), or a longer window such as (0,10).

26

subsamples (Models 6 and 9). These results indicate that early releases are more likely

when insiders place a high value on the ability to trade the company’s stock. This

implies that investment bankers are more willing to release insiders from the lockups

both when the share price increases and when the early release is important to insiders.

In Models 3, 6, and 9, the probability of an early release is positively related to

offer size (Log(proceeds)), the fraction of secondary shares relative to total shares offered

(Secondary shares %), and the firm’s prior ROA. The inclusion of these control

variables, however, does not affect our main finding, which is that the probability of an

early release is positively related to the change in share price after the offering.14

4.2. The ex post lockup period

The fact that some lockups are released early suggests that the determinants of the

ex post lockup periods may differ from those of the ex ante lockup periods. We examine

this issue in Table 8, which reports Tobit model estimates in which the dependent

variable is the ex post lockup period. As in Table 6, the dependent variable is set to zero

for SEOs without lockups.

Models 1 and 2 in Table 8 use data from all 1,597 SEOs in the sample that have

sufficient data. Model 1 includes only CR(0,5), the information asymmetry factor, and

Ex ante lockup days, whereas Model 2 includes our additional control variables. The ex

post lockup period is negatively and significantly related to CR(0,5), the cumulated return

14 In untabulated additional analysis, we find that the fractions of secondary shares sold, particularly by insiders or CEOs, are positively related to the probability of an early release. This relationship is consistent with the views that insiders’ demand to trade post-SEO is positively related to their SEO secondary share sales, that underwriters accept insider secondary share sales when there is less investor concern about opportunistic trading by insiders, or that investors’ concern about opportunistic trading is sufficiently reduced when insiders pre-announce their SEO secondary sale plans.

27

during the post-SEO period. This is consistent with the evidence in Table 7 that the

probability of an early release increases with the firm’s post-SEO share price

performance. Together, these results indicate that ex post lockup periods are decreasing

in the level of share price performance in the week after the SEO. This is consistent with

the information asymmetry hypothesis, which holds that strong post-SEO share price

performance undermines the reason for having a lockup, which is to assure investors that

they will not overpay at the SEO by buying shares at the offer price.

The results in Models 1 and 2 indicate that the length of the ex post lockup period

is positively related to the ex ante lockup period. The coefficient on Ex ante lockup days

of 0.699 in Model 1 indicates that each additional day in the ex ante lockup period

translates into an average increase in the ex post lockup period of 0.699 days. This result,

however, is affected by the fact that only 40% of the lockups are released early. This

means that 60% of the ex post and ex ante lockup periods are identical, biasing the

coefficient on Ex ante lockup days toward 1.

To remove this mechanical influence of Ex ante lockup days, in Models 3 and 4

we include only observations in which the lockup was released early. In this subsample,

the ex post lockup period remains negatively and significantly related to CR(0,5).

However, the coefficient on Ex ante lockup days decreases to 0.057 in Model 3 and 0.055

in Model 4. Thus, among firms that release their lockups early, the length of the ex post

lockup period is positively related to the ex ante period, but the size of the relation is

quite small.

In Models 3 and 4, the information asymmetry factor is negatively related to the

length of the ex post lockup period. This indicates that, among firms that release their

28

lockups early, information asymmetry is associated with shorter lockup periods. This in

turn implies that early releases concentrate among firms that have high information

asymmetry. It is particularly when the degree of information asymmetry is high that a

strong stock performance resolves investors’ concerns that the SEO was overpriced.

The results in Models 2 and 4 indicate that the ex post lockup period is negatively

related to Insider ownership, offer proceeds, secondary shares as a percentage of shares

offered, and the firm’s prior ROA. As indicated in Table 6, these control variables

generally are unrelated to the length of the ex ante lockup period. Rather, they are

negatively related to the ex post lockup period because, as reported in Table 7, they are

related to the probability of an early release.

5. Conclusion

The corporate form of organization offers many advantages, including risk

sharing, specialized risk-bearing, and the separation of the risk-bearing and management

functions (e.g., see Fama and Jensen 1983). For corporations to survive as a viable form

of organization, however, investors must be willing to invest in corporate securities

despite the risk that they are at an informational disadvantage compared to corporate

insiders. Nowhere is the risk to outside investors more apparent than when a firm issues

new stock. The risk that investors could overpay for newly issued shares creates a

serious obstacle in a firm’s effort to raise capital to finance profitable investment

opportunities (e.g., see Myers and Majluf 1984).

In this study we hypothesize that lockup agreements are a valuable contracting

mechanism that helps to mitigate the risks investors bear when purchasing newly issued

29

shares. The benefits of a lockup hold not only for IPOs, but also for seasoned equity

issues. We test this hypothesis with a sample of 2,579 seasoned equity offerings from

1996 – 2006. Lockups are very common in SEOs and occur in 94% of our sample, which

is comparable to the frequency of lockups in IPOs. Lockup periods can range upwards of

180 days, but the median period is 90 days. To test the proposition that lockups are a

contracting solution to the problem of asymmetric information between insiders and

outside investors, we construct an information asymmetry factor using data from eight

common proxies for information asymmetry. The probability that an SEO is

accompanied by a lockup agreement, and the duration of a lockup, are both positively

related to measures of information asymmetry.

We further find that 39.5% of all lockups are released early, allowing insiders to

sell shares before the specified lockup period is over. The probability of an early release

is positively related to the firm’s cumulated stock return over the offer date and the first

five days following the new issue. This finding provides further support for the

information asymmetry explanation of lockups, because it indicates that lockups become

less valuable as investors learn that they did not overpay for the newly issued shares.

These results indicate that lockups tend to be used when the degree of information

asymmetry between insiders and outside investors is large. They are released when such

asymmetry is resolved, or at least becomes less risky for outside investors, as share prices

increase following the SEO. This supports the view that lockup agreements do in fact

serve to guarantee the quality of the issue to new investors. They do so by committing

insiders to hold shares during the post-SEO period, and by constraining insiders’ abilities

to exploit their informational advantage by selling overpriced shares immediately after

30

outside investors participate in the SEO. That is, by committing insiders to hold their

shares for a specified period, lockups mitigate the problem of information asymmetry in

SEOs, thereby facilitating the capital formation process.

31

References

Altinkilic, O., and R. S. Hansen, 2003. Discounting and underpricing in seasoned equity offers

Journal of Financial Economics 69, 285-323.

Barth, M. E., R. Kasznik, M. F. McNichols, 2001. Analyst coverage and intangible assets,

Journal of Accounting Research 39, 1- 34.

Bethel, Jennifer E. and L. Krigman, 2009. Managing the costs of issuing common equity: The

role of registration choice, Quarterly Journal of Finance and Accounting 47, 57-85.

Bhagat, S. M., W. Marr, and G. R. Thompson, 1985. The Rule 415 experiment: Equity markets.

Journal of Finance 60, 1385-1401.

Bhushan, R., 1989. Firm characteristics and analyst following, Journal of Accounting and

Economics 11, 255-274.

Boone, A. L., L.C. Field, J.M. Karpoff, J. M., and C. G. Raheja, 2007. The determinants of

corporate board size and composition: Empirical evidence, Journal of Financial

Economics 85, 66-101 Bortolotti, B., S. Smart, W. L. Meggison, 2008. The rise of accelerated seasoned equity

underwritings, Journal of Applied Corporate Finance 20, 35-57.

Bradley, D., B. Jordan, I. Roten, and H. Yi, 2001. Venture capital and IPO lockup expiration: An

empirical analysis, Journal of Financial Research 24, 465-493.

Brau, J.C., V. E. Lambson, and G. McQueen, 2005. Lockups revisited, Journal of Financial and

Quantitative Analysis 40, 519-530.

Brav, A. amd P. A. Gompers, 2003. The role of lockups in initial public offerings, Review of

Financial Studies 16, 1-29.

Brown, S., S. Hillegeist, and K. Lo, 2009. The effect of earnings surprises on information

asymmetry, Journal of Accounting and Economics, 47, 208–225

Cai, J. , Y. Liu, and Y. Qian, 2009. Information asymmetry and corporate governance, Working

paper, Drexel University.

Clarke, J. and K. Shastri, 2001. On information asymmetry metrics, Working paper, Georgia

Institute of Technology.

Cao, C., L. C. Field, and G. Hanka, 2004. Liqudity consequences of lockup expirations, Journal

of Financial and Quantitative Analysis 39, 25-46.

Coles, J. L., N. D. Daniel, and L. N. Naveen, 2006. Managerial incentives and risk-taking,

Journal of Financial Economics 79, 431-468.

32

Copeland, T. E., and D. Galai, 1983. Information effects and the bid-ask spread, Journal of

Finance 38, 1457-1469.

Corwin, S.A., 2003. The determinants of underpricing for seasoned equity offers, Journal of

Finance 58, 2249–2279.

Courteau, L., 1995. Under-diversification and retention commitments in IPOs, Journal of

Financial and Quantitative Analysis 30, 487-517.

Denis, D. J., 1991. Shelf registration and the market for seasoned equity offerings, Journal of

Business 64, 189-212.

Drucker, S. and M. Puri, 1999. On the benefits of current lending and underwriting, Journal of

Finance 60, 2762-2799.

Duarte-Silva, T., 2010. The market for certification by external parties: Evidence from

underwriting and banking relationships, Journal of Financial Economics 98, 568-582

Easley, D., M. O’Hara, 1987. Price, trade size, and information in securities markets, Journal of

Financial Economics 19, 69–90.

Fama, E. F., M.C. Jensen, 1983. Separation of ownership and control, Journal of Law and

Economics 26, 301-325.

Field, L.C., and G. Hanka, 2001. The expiration of IPO share lockups, Journal of Finance 56,

471-500.

Francis, J., R. Lafond, P. Olsson and K. Schipper, 2004. Costs of equity and earnings attributes,

The Accounting Review 79, 967–1010.

Francis, J., R. Lafond, P. Olsson and K. Schipper, 2005. The market pricing of accruals quality,

Journal of Accounting and Economics 39, 295-327.

Frankel, R. M. and X. Li, 2004. Characteristics of a firm's information environment and the

information asymmetry between insiders and outsiders, Journal of Accounting and

Economics 37 (2), 229-259.

Gao, X. and J.R. Ritter, 2010. The marketing of seasoned equity offerings, Journal of Financial

Economics 97, 33-52.

Glosten, L. R., and P. R.Milgrom, 1985. Bid, ask and transaction prices in a specialist market

with heterogeneously informed traders, Journal of Financial Economics 14, 71-100.

Goergen, M., A. Khurshed, A., and L. Renneboog, 2006. Explaining the Diversity in Shareholder

Lockup Agreements, Journal of Financial Intermediation 15, 254-280.

Hong, H., T. Lim, J. Stein, 2000. Bad news travels slowly: size, analyst coverage, and the

profitability of momentum strategies, Review of Financial Studies 55, 265–295.

33

Karpoff, J., and D. Lee, 1991. Insider trading before new issue announcements, Financial

Management 20, 18-26

King, G. and L. Zeng, 2001. Logistic regression in rare events data, Political Analysis 9, 137-

163.

Kothari, S.P., A. J. Leone, and C. E. Wasley, 2005. Performance matched discretionary accrual

measures, Journal of Accounting and Economics 39, 163-197.

Leary, M. T., and M. R. Roberts, 2010. The pecking order, debt capacity, and information

asymmetry, Journal of Financial Economics 95, 332-355.

Lease, Ronald C., Ronald W. Masulis, and John R. Page, 1991, An investigation of market

microstructure impacts on event study returns, Journal of Finance 46, 1523-1536.

Lee, G., and R. W. Masulis, 2009. Seasoned equity offerings: Quality of accounting information

and expected flotation costs, Journal of Financial Economics 92, 443-469.

Lee, I., 1997. Do firms knowingly sell overvalued equity? Journal of Finance 52, 1439-66.

Lowry, M., M.S. Officer, G. W. Schwert, 2010. The variability of IPO initial returns, Journal of

Finance 65 425-465.

Myers, S. and N. Majluf, 1984. Corporate financing and investment decisions when firms have

information that investors do not have, Journal of Financial Economics 13, 187– 221.

Lubotsky, D. and M. Wittenberg, 2006. Interpretation of regressions with multiple proxies,

Review of Economics and Statistics 88(3), 549-562

Maddala, G. S., 1983. Limited-Dependent and Qualitative Variables in Economics, New York:

Cambridge University Press.

Safieddine, Assem, and William J. Wilhelm, 1996, An empirical investigation of short-selling

activity prior to seasoned equity offerings, Journal of Finance 51, 729-749.

Sherman, A.E., 1999. Underwriter certification and the effect of shelf registration on due

diligence, Financial Management 28(1), 5-19.

Treynor, J., 1971. The only game in town, Financial Analysts Journal 27, 12–14.

34

Appendix: Variable Definitions Variable Definition Abnormal accruals

To obtain our proxy of accruals quality based on a performance-matched abnormal accruals model based on Kotahri et al. (2005), we first estimate the modified Jones model:

ACit

TAi,t−1= α0

1TAi,t−1

+α1ΔSalesitTAi,t−1

+α2PPEit

TAi,t−1+ εit .

Here, TAit-1 = firm i’s total assets (Compustat item 6, using the old Compustat variable identification) at the beginning of year t , ACit =total accruals in year t=earnings before extraordinary items and discontinued operations (item #123) and operating cash flow (from continuing operations) taken directly from the statement of cash flows (Compustat #308 – Compustat #124), Salesit = sales (iterm 12) in year t, PPEit = property, plant, and equipment (item 7) in year t. For each year and each industry based on two-digit SIC code, we create five portfolios with at least four firms each by sorting the data into quintiles based on ROA from the previous year. The performance matched abnormal accruals for a sample firm is the absolute value of the firm-specific abnormal accruals minus the median abnormal accruals for its respective industry-performance-matched portfolio. (We considered alternate measures of abnormal accruals, e.g., the measure from Dichow and Dichev (2002). As Lee and Masulis (2009) note, however, this measure requires seven years of financial accounting data before the SEO, which decreases the sample size and creates sample selection problems.)

Accelerated SEO

Indicator variable is one if the SEO is an accelerated SEO (accelerated bookbuilt, block trade, or bought deal) and zero otherwise, taken from the Thomson Financial New Issues database. These are primarily shelf registered issues.

Average bid-ask spread The average bid-ask spread of daily stock price scaled by its price during the trading period (-90, -11) prior to the filing date, taken from the CRSP database.

CR (0,5) The cumulative raw stock returns from the SEO issue date to the end of the 5 trading days following. Day 0 is the issue date if the issue occurs before the close of trading; otherwise it is the following day.

35

Firm age Number of full years from issuer’s IPO to the current SEO filing date.

Firm size The natural log of the book value of assets (Compustat item 6) at the year-end prior to the SEO filing date.

Information asymmetry (IA) This is the first factor obtained by the factor analysis using 8 information asymmetry variables, measured at the year-end prior to the SEO fling date, shown in the panel B of Table IV, then we multiply by (-1).

Insiders Officers, directors and blockholders with 5% or more of shares. Insider ownership (%) The percent ownership is obtained by each insider's ownership level from his or her most recent

shareholdings or trading report filed with the SEC prior to the SEO filing date, taken from Thomson Financial 's Insider Filing database. All insiders include officers, directors, and beneficial owners of more than 10% of a class of security of the company.

Insider ownership ($MM) The amount ownership is obtained by each insider's shareholdings report filed with the SEC prior to the SEO filing date, taken from Thomson Financial 's Insider Filing database, date multiplied by the SEO offer price. All insiders include officers, directors, and beneficial owners of more than 10% of a class of security of the company.

Leverage The ratio of book value of short and long term debt (Compustat item 9 + item 34) over book value of total assets (Compustat item 6) in the year prior to SEO filing date.

Log (proceeds) The log of SEO proceeds excluding overallotments taken from the Thomson Financial New Issue.

Market value of equity ($MM) Common shares outstanding (Compustat item 199) multiplied by the closing stock price (Compustat item 25), all measured at the year-end prior to the SEO filing date.

Net selling frequency Net selling frequency is calculated by the number of trades of disposing stocks minus the number of trades of acquiring stocks over the six months before the SEO filing date, scaled by shares outstanding at the filing date. Therefore, positive (negative) number of net selling frequency implies that SEO firm insiders are net seller (buyer).

36

Net selling value Net selling value is calculated by a difference between dollar value of sales of stocks and dollar value of purchases of stocks by insiders over the six months before the SEO filing date, scaled by market capitalization at the filing date, then multiply by 1,000,000.

No. of Analysts The number of analysts forecasting firm earnings in the year before the SEO filing date, taken from I/B/E/S database.

No. of prior stock offers The number of prior IPO and SEOs by the issuing firm.

Relative volatility The ratio of idiosyncratic volatility based of the residual standard deviation based on a market model over the trading period (-90,-11) prior to the SEO filing date to the market return standard deviation over the same trading period, taken from the CRSP database.

Return on assets (ROA) Earnings (Compustat item 18) divided by total assets (Compustat item 6) in the year before the SEO filing date.

Return volatility The standard deviation of daily stock return during the trading period (-90, -11) prior to the SEO filing date, taken from the CRSP database.

Share turnover The ratio of average daily share trading volume during the trading period (-90,-11) prior to the SEO filing date, divided by SEO total shares outstanding at the SEO filing date, all taken from the CRSP database.

Secondary shares Percentage of SEO shares being sold by existing shareholders to total SEO shares offered, excluding overallotments, which is taken from the Thomson Financial New Issues database.

Tangible assets The ratio of property, plant & equipment to total assets at the year-end prior to the SEO filing date (Compustat item7/item6).

Underwriter rank

The Carter-Manaster reputation measure in the year prior to SEO filing date, taken from Jay Ritter’s website.

37

Table 1. Distribution of Lockup Periods by Year of IPOs and SEOs Panel A reports the frequencies of SEOs (in %’s) with ex ante lockup periods of varying lengths. The sample consists of 2,579 SEOs from 1996-2006 by U.S. issuers of common stock listed on the NYSE, Amex or Nasdaq. SEOs with no lockup agreement are coded as having a 0 lockup period. For comparison, Panel B reports on lockup agreements for 1,926 IPOs during the same period. These samples exclude: (1) SEOs lacking CRSP daily stock returns and prices for the SEO announcement period and the prior 90 trading days, (2) firms lacking COMPUSTAT annual financial statement data for the six years before the SEO filing date, the offer year and the following year, which are needed to estimate firm abnormal accruals, (3) completed SEOs with offer prices less than $5 and withdrawn SEOs with filing range midpoints less than $5, (4) spin-offs, (5) reverse LBOs, (6) closed-end fund, unit investment trusts, REITs and limited partnerships, (7) rights and standby issues, (8) simultaneous or combined offers of several classes of securities such as unit offers of stock and warrants and (9) non-domestic and simultaneous domestic-international offers. Panel A: SEOs

Lockup period in days: Number 0 <90 90 91-179 180 >180

1996 227 1.73 2.6 53.68 19.91 19.91 2.16 1997 359 6.99 3.11 63.73 9.59 15.8 0.78 1998 194 22.4 4.8 49.2 10 11.6 2 1999 165 6.25 6.82 60.23 7.39 19.32 0 2000 149 3.25 11.69 74.68 3.25 7.14 0 2001 144 5.88 15.69 68.63 2.61 7.19 0 2002 210 1.87 11.21 75.7 1.4 9.81 0 2003 253 5.24 19.85 68.54 1.87 4.12 0.37 2004 280 3.8 18.48 70.63 1.77 5.32 0 2005 299 4.47 22.68 64.54 3.51 4.47 0.32 2006 299 5.68 29.34 61.2 0.32 3.47 0

Total 2,579 6.2 13.94 64.39 5.5 9.45 0.53

Panel B: IPOs

Lockup period in days: Number 0 <90 90 91-179 180 >180

1996 476 4.83 0 1.89 0.63 82.34 10.29 1997 310 4.19 0.32 5.81 0.65 76.77 12.26 1998 157 5.1 0 3.18 1.91 78.34 11.46 1999 216 5.09 0.46 7.41 0.46 82.41 4.17 2000 141 1.42 0 4.96 0 90.07 3.55 2001 29 0 0 13.79 0 79.31 6.9 2002 43 2.33 0 2.33 2.33 83.72 9.3 2003 64 0 0 3.13 1.56 82.81 6.25 2004 179 1.68 1.12 8.94 0.56 84.36 3.35 2005 157 1.91 3.18 14.65 1.91 73.89 4.46 2006 154 1.3 1.3 5.19 1.95 86.36 3.9

Total 1,926 3.43 0.67 5.76 0.93 81.52 7.68

38

Table 2. Early Releases and the Difference Between Ex Ante and Ex Post Lockup Periods

Panel A reports on the number of SEOs with lockups that were released early by the underwriter, allowing one or more insider to sell shares before the end of the official lockup period. Panel B reports on the resulting length of the ex post lockup period. The SEO sample consists 2,579 SEOs over the 1996-2006 period by U.S. issuers of common stock listed on the NYSE, Amex or Nasdaq, of which 2,419 (93.8%) had lockup agreements.

Panel A: Early releases from SEO lockups

Early

Release #days from the issue date to the

lockup release date #days from the lockup release date

to the ex ante expiration date Freq % Q1 Median Q3 Mean Q1 Median Q3 Mean 1996 79 34.8 5 7 41 27.7 71 85 117 95.3 1997 127 35.4 5 6 15 15.8 76 85 98 97.5 1998 76 39.2 5 7 40.5 34.4 54.5 84 113.5 105.9 1999 64 38.8 5 6.5 38.5 25.0 54 84 88 76.5 2000 69 46.3 4 6 20 16.1 64 84 86 75.9 2001 48 33.3 4 6 20 16.0 61.5 84 87 79.7 2002 57 27.1 5 6 20 18.0 70 84 89 86.7 2003 92 36.4 5 7 29.5 18.1 56.5 82 86 76.6 2004 149 39.2 5 6 17 15.0 57 84 86 76.4 2005 136 45.5 5 6 17 15.0 50 83 84 69.6 2006 160 53.5 5 6 10 10.5 54 83 84.5 71.4

All 1,057 39.5 5 6 21 18.2 58 84 87 81.5

Panel B: Comparison of ex ante and ex post lockup periods

Ex ante lockup periods, in days Ex post lockup periods, in days (after early releases)

Q1 Median Q3 Mean Q1 Median Q3 Mean 1996 90 90 120 117.0 31 90 120 84.4 1997 90 90 120 102.7 6 90 90 70.6 1998 45 90 120 92.3 3 60 90 60.1 1999 90 90 120 100.9 7 90 90 73.1 2000 90 90 90 89.7 6 60 90 55.7 2001 90 90 90 85.0 7 90 90 60.3 2002 90 90 90 92.5 30 90 90 69.4 2003 60 90 90 82.5 10 60 90 56.1 2004 90 90 90 84.6 6 60 90 55.8 2005 60 90 90 83.0 6 48 90 52.8 2006 60 90 90 76.4 6 22 90 40.3

All 90 90 90 90.9 6 70 90 60.8

39

Table 3. Information Asymmetry Factor

This table reports the factor loadings for the first three factors from a factor analysis used to reduce the number of variables used to measure information asymmetry. The primitives used in the analysis are eight correlated measures of information asymmetry. The scree plot of eigenvalues illustrates that Factor 1 captures a substantial amount of the variation in the eight primitives, but that additional factors do not. The Factor 1 loadings have the opposite signs as predicted, indicating that it is a measure of information symmetry. To transform it into a measure of information asymmetry, we multiply Factor 1 by –1 in the empirical tests reported in subsequent tables, labeling it the information asymmetry factor.

Variable

Predicted correlation with info

asymmetry Factor 1 Factor 2 Factor 3 KMO Firm size – 0.8062 -0.0243 -0.0813 0.6909 Firm age – 0.4992 0.0170 0.2191 0.8108 No. of Analysts – 0.1240 -0.0853 -0.0461 0.6790 Tangible assets – 0.2964 0.2224 0.1585 0.7745 No. of prior stock offers – 0.5277 -0.1137 0.1945 0.8082 Average bid-ask spread + -0.4212 0.3237 0.0709 0.6740 Return volatility + -0.5453 -0.1877 0.1476 0.7397 Abnormal accruals + -0.4335 -0.0952 0.1891 0.8110 KMO overall 0.7429 Eigenvalue 1.9288 0.2196 0.1823

40

Table 4. Univariate Comparisons

This table describes the sample mean and median values for the variables used in the empirical analysis. Panel A reports on the information asymmetry variables and Panel B reports on the control variables. The sample consists of 2,579 SEOs from 1996 through 2006. The SEOs are for U.S. common stocks listed on the NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, closed-end fund, unit investment trusts, REITs, limited partnerships, unit offerings, and SEOs with offer prices less than $5. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. p-values are calculated from parametric t-tests assuming unequal variances (reported in the “Mean” rows) and Wilcoxon signed rank tests (reported in the “Median” rows). Variable definitions are described in the Appendix.

Panel A: Information asymmetry measures

(a) (b) (c) SEOs with lockups

All SEOs

SEOs with

SEOs with Variable

no lockups lockups

p-value (d)

Lockups vs. no

lockups

(e) Duration<90

days

(f) Duration=90

days

(g) Duration>90

days

p-value (h)

Long lockups vs. short lockups

Mean 0 -0.172 0.033 0 -0.414 0.007 0.394 0 Information asymmetry factor Median 0.058 -0.117 0.097 0 -0.467 0.059 0.51 0 Measures used to construct the information asymmetry factor: Firm size Mean 5.724 6.406 5.615 0 6.635 5.614 4.945 0 Median 5.623 6.315 5.494 0 6.725 5.451 4.728 0 Firm age Mean 9.228 10.598 8.919 0.008 10.746 9.327 6.9 0 Median 4.875 5.75 4.667 0.004 6.25 4.917 3.333 0 No. analysts Mean 1.465 1.514 1.451 0.622 1.552 1.442 1.422 0.227 Median 0 0 0 0.214 0 0 0 0.858 Tangible assets Mean 0.392 0.381 0.394 0.612 0.401 0.398 0.363 0.102 Median 0.265 0.23 0.269 0.293 0.255 0.275 0.251 0.921 No. of prior stock offers Mean 2.129 2.737 2.023 0 2.98 1.959 1.533 0 Median 1 2 1 0 2 1 1 0 Average bid-ask spread Mean 1.2 1.497 1.172 0 0.987 1 2.133 0 Median 0.71 1.155 0.602 0 0.595 0.519 1.797 0 Return volatility Mean 0.032 0.029 0.032 0 0.026 0.033 0.036 0 Median 0.028 0.025 0.029 0.06 0.022 0.03 0.033 0 Abnormal accruals Mean 0.13 0.111 0.133 0.114 0.104 0.12 0.189 0 Median 0.071 0.062 0.073 0.082 0.06 0.069 0.096 0

41

Panel B: Control variables

(a) (b) (c) SEOs with lockups

All SEOs

SEOs with

SEOs with

Variable no lockups lockups

p-value (d)

Lockups vs. no

lockups

(e) Duration<90

days

(f) Duration<90

days

(g) Duration<90

days

p-value (h)

Long lockups vs. short lockups

Insiders’ demand to trade:

Insider ownership (%) Mean 5.173 2.146 5.654 0 3.197 6.04 5.139 0

Median 0.518 0.005 0.609 0 0.199 0.598 0.64 0 Net selling frequency Mean 0.597 0.191 0.623 0 0.588 0.645 0.443 0.013 Median 0.139 0.082 0.142 0 0.126 0.157 0.087 0.001 Net selling value ($MM) Mean 18.94 12.754 18.453 0 18.402 22.85 17.684 0.434

Median 0.737 0.563 0.732 0.003 0.23 1.551 0.778 0 Other control variables:

Underwriter holdings Mean 0.325 0.263 0.336 0.122 0.342 0.323 0.307 0.591 Median 0.014 0 0.016 0.003 0.021 0.018 0 0 Leverage Mean 0.267 0.311 0.26 0.001 0.326 0.247 0.254 0 Median 0.233 0.303 0.225 0.002 0.332 0.2 0.212 0 Share turnover Mean 0.872 0.706 0.895 0 0.759 0.938 0.817 0.351 Median 0.625 0.493 0.645 0 0.546 0.672 0.586 0.401 Log(Proceeds) Mean 4.345 4.182 4.37 0 4.425 4.463 3.889 0 Median 4.309 4.11 4.317 0.038 4.389 4.402 3.785 0 Underwriter rank Mean 7.969 8.029 7.958 0.33 8.069 7.983 7.799 0 Median 8.1 8.1 8.1 0.012 8.1 8.433 8.1 0.032 Secondary shares (%) Mean 28.19 25.343 28.002 0.192 30.916 27.101 27.921 0.006 Median 0 0 0 0 0 0 9.09 0 Accelerated SEOs Mean 0.225 0.215 0.214 0.964 0.419 0.184 0.094 0 Median 0 0 0 0.964 0 0 0 0 ROA Mean -0.038 -0.013 -0.042 0.243 -0.002 -0.054 -0.035 0.153 Median 0.026 0.027 0.026 0.703 0.027 0.024 0.034 0.125

42

Table 5. Determinants of SEO Lockup Agreements

This table presents the estimates from nine specifications of a logit model using data from SEOs from 1996 through 2006. The (untransformed) dependent variable equals one when an issuer adopts a lockup agreement. The sample includes SEOs of U.S. common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are reported in brackets, and ***, **, and * represent significance at the 1%, 5%, and 10% levels.

Full Sample Conventional SEOs Accelerated SEOs

Variables: (1) (2) (3) (4) (5) (6) (7) (8) (9)

Information asymmetry 0.312*** 0.355*** 0.550*** 0.200*** 0.304*** 0.556*** 0.148** 0.296** 0.583** [0.000] [0.000] [0.000] [0.000] [0.001] [0.000] [0.027] [0.034] [0.020] Insider ownership % -0.002 -0.001 -0.002 -0.001 -0.009 -0.019**

[0.692] [0.601] [0.685] [0.577] [0.610] [0.047] Net selling frequency 0.150** 0.142** 0.152** 0.143** 0.031 0.083 [0.020] [0.019] [0.020] [0.018] [0.923] [0.785] Net selling value -0.001 -0.001 -0.001 -0.001* 0.058** 0.003** [0.255] [0.102] [0.169] [0.067] [0.039] [0.011] Accelerated SEOs -0.294

[0.105] Leverage 0.340 0.457 0.095 [0.264] [0.187] [0.870] Share turnover -0.091** -0.088* 0.083 [0.049] [0.068] [0.587] Log (proceeds) 0.264*** 0.245** 0.402***

[0.003] [0.012] [0.009] Underwriter rank -0.027 -0.041 -0.231* [0.535] [0.385] [0.057] Secondary shares (%) -0.002 -0.000 -0.008** [0.259] [0.840] [0.017] ROA 0.013 0.082 0.177

[0.945] [0.637] [0.746] Constant 1.121*** 0.366** 0.293** 1.187*** 0.516** 0.432** 0.743 0.544 1.145 [0.000] [0.042] [0.035] [0.000] [0.038] [0.041] [0.148] [0.351] [0.276] Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 1,597 1,597 1,597 1,096 1,096 1,096 501 501 501

Pseudo R2 0.203 0.212 0.253 0.123 0.132 0.168 0.349 0.372 0.442

43

Table 6. Determinants of the Lockup Period

This table presents the estimates from nine specifications of a tobit model using data from SEOs from 1996 through 2006. The dependent variable equals the ex ante lockup period in calendar days. The sample criteria require the SEOs to be US common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are reported in brackets, and ***, **, and * represent significance at the 1%, 5%, and 10% levels. Full Sample Conventional SEOs Accelerated SEOs

Variables: (1) (2) (3) (4) (5) (6) (7) (8) (9)

Information asymmetry 13.053*** 13.744*** 17.445*** 12.653*** 14.020*** 18.617*** 3.122** 5.342*** 12.120***

[0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.016] [0.003] [0.000] Insider ownership % -0.018 0.017 0.035 0.029 -0.266* -0.098 [0.895] [0.792] [0.838] [0.705] [0.059] [0.182] Net selling frequency 1.813** 1.277** 2.203** 1.638** 1.822 0.149 [0.015] [0.027] [0.013] [0.046] [0.259] [0.870] Net selling value -0.003 -0.002 -0.001 -0.002 -0.007 -0.004

[0.741] [0.643] [0.865] [0.642] [0.520] [0.711] Accelerated SEOs -10.226** [0.014] Leverage 18.856*** 23.036*** 5.277 [0.000] [0.000] [0.349] Share turnover -3.250** -2.523** -3.020*

[0.027] [0.028] [0.087] Log (proceeds) 2.693 -0.412 8.844*** [0.170] [0.862] [0.000] Underwriter rank -0.243 -0.429 -1.286 [0.809] [0.711] [0.235] Secondary shares (%) -0.047 0.022 -0.112***

[0.215] [0.642] [0.002] ROA 8.303 9.471 4.066 [0.182] [0.180] [0.488] Constant 103.516*** 102.941*** 92.566*** 104.419*** 103.579*** 102.190*** 41.647** 58.867*** 43.343*** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.013] [0.000] [0.007] Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes

Industry fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 1,597 1,597 1,597 1,096 1,096 1,096 501 501 501 Pseudo R2 0.203 0.212 0.253 0.123 0.132 0.168 0.349 0.372 0.442

44

Table 7. Determinants of Early Releases of SEO Lockups

This table presents the estimates from nine specifications of a logit model using data from SEOs with lockup agreements from 1996 through 2006. The (untransformed) dependent variable equals one when an issuer released the lockup before the end of the ex ante lockup period. The sample criteria require the SEOs to be US common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are reported in brackets, and ***, **, and * represent significance at the 1%, 5%, and 10% levels.

Full Sample Conventional SEOs Accelerated SEOs

Variables: (1) (2) (3) (4) (5) (6) (7) (8) (9)

CR(0,5) 0.631** 0.951** 0.705** 0.673** 1.151** 0.730** 1.222* 0.593** 0.753**

[0.045] [0.026] [0.018] [0.043] [0.019] [0.017] [0.059] [0.043] [0.035]

Information asymmetry 0.009* 0.043 0.250 0.019* 0.061 0.315 0.066 0.142* 0.177

[0.085] [0.392] [0.431] [0.097] [0.317] [0.231] [0.347] [0.076] [0.159]

Ex ante lockup days 0.000 0.001 0.002 0.000 0.001 0.001 -0.002 0.002 0.002

[0.814] [0.241] [0.106] [0.840] [0.460] [0.284] [0.345] [0.418] [0.454]

Info asymmetry*CR(0,5) 0.120 0.177 0.035 0.354 0.630 -0.756

[0.110] [0.736] [0.151] [0.549] [0.541] [0.519]

Insider ownership % 0.004** 0.009** 0.003

[0.043] [0.033] [0.413]

Net selling frequency 0.081*** 0.078** 0.148**

[0.010] [0.015] [0.048]

Net selling value 0.001* 0.001* 0.001*

[0.091] [0.075] [0.067]

Accelerated SEOs -0.047

[0.687]

Leverage -0.215 -0.110 -0.312

[0.148] [0.462] [0.209]

Share turnover -0.002 -0.024 0.048

[0.952] [0.563] [0.600]

Log (proceeds) 0.266*** 0.255*** 0.347***

[0.000] [0.000] [0.001]

Underwriter rank 0.003 0.014 -0.033

[0.902] [0.636] [0.492]

Secondary shares (%) 0.008*** 0.009*** 0.006***

[0.000] [0.000] [0.001]

ROA 0.382** 0.399** 0.538*

[0.016] [0.016] [0.061]

Constant -0.058 -0.075 -1.406*** -0.056 -0.037 -1.458*** 0.028 0.165 -0.699

[0.491] [0.618] [0.000] [0.546] [0.814] [0.000] [0.899] [0.836] [0.385]

Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes

Industry fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes

Observations 1,485 1,485 1,485 1,226 1,226 1,226 259 259 259

Pseudo R2 0.031 0.033 0.121 0.030 0.036 0.130 0.047 0.048 0.174

45

Table 8. Determinants of Ex Post Lockup Period This table presents the estimates from four specifications of a tobit model using data from SEOs from 1996 through 2006. The dependent variable equals the ex-post lockup period in calendar days. The sample criteria require the SEOs to be U.S. common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. The first two columns include the full SEO sample and the last two columns include only SEOs with early releases. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are reported in brackets, and ***, **, and * represent significance at the 1%, 5%, and 10% levels.

All SEOs SEOs with lockups and early releases

Variables: (1) (2) (3) (4)

CR(0,5) -24.505** -30.974** -21.444** -18.551* [0.047] [0.043] [0.036] [0.096] Information asymmetry -1.673 -7.871 -6.334*** -9.010*** [0.263] [0.121] [0.000] [0.000] Ex ante lockup days 0.699*** 0.594*** 0.057** 0.055** [0.000] [0.000] [0.011] [0.013] Insider ownership % -0.302** -0.013* [0.019] [0.078] Net selling frequency -1.003 -0.308 [0.295] [0.707] Net selling value -0.006* -0.002* [0.059] [0.055] Accelerated SEOs 0.980 2.068 [0.801] [0.509] Leverage 6.051 -6.573 [0.318] [0.157] Share turnover 0.398 -0.321 [0.813] [0.790] Log (proceeds) -6.676*** -2.978** [0.000] [0.044] Underwriter rank -1.372 0.275 [0.150] [0.704] Secondary shares (%) -0.290*** -0.089*** [0.000] [0.001] ROA -14.611*** -2.983 [0.001] [0.339] Constant -13.614*** 37.227*** 17.373*** 34.297*** [0.000] [0.001] [0.000] [0.000] Year fixed effects Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Observations 1,597 1,597 579 579 Pseudo R2 0.456 0.476 0.106 0.178

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 1

INTERNET APPENDIX for

“Lockup Agreements in Seasoned Equity Offerings: Evidence of Optimal Contracting,” February 18, 2012

This internet appendix reports on robustness tests for the results reported in the main body of the paper. There are four sections in this appendix. Section I reports on three alternative specifications of the information asymmetry factor. Section II reports estimates from the tests reported in Tables 5 – 8 of the paper using the eight individual asymmetry information measures (firm size, firm age, number of analysts, ratio of tangible assets, number of prior stock offers, average spreads, return volatility and accounting quality) used in the factor analysis as well as several additional information asymmetry measures (tobin’s q, underpricing, industry-adjusted return volatility, analysts disagreement). Section III reports estimates from the tests reported in Tables 5 – 8 of the paper, using all eight asymmetry information measures separately (firm size, firm age, number of analysts, ratio of tangible assets, number of prior stock offers, average spreads, return volatility, accounting quality) used in the analysis instead of the single information asymmetry factor. Section IV reports estimates from the tests reported in Tables 5 – 8 of the paper, using Net selling frequency and Net selling value measured over one year before the SEO filing date.

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 2

Internet Appendix Section I

Figure A1. Information asymmetry factor analysis adding additional possible information asymmetry variables such as Tobin’s q, underpricing, industry-adjusted return volatility, and analyst’s disagreement.

Variable

Predicted correlation with info asymmetry

Factor 1 Factor 2 Factor 3 KMO

Firm size – -0.8244 0.1283 0.0201 0.6654 Firm age – -0.4860 0.1793 0.0707 0.8040 No. of Analysts – -0.1855 0.1680 0.0600 0.4720 Tangible assets – -0.2736 -0.0433 0.2793 0.7334 No. of prior stock offers – -0.4907 0.2087 -0.0754 0.7937 Average spreads + 0.3684 -0.2739 0.2953 0.5969 Return volatility + 0.6463 0.3471 0.0631 0.6813 Discretionary accruals + 0.3898 -0.0055 -0.0515 0.7581 Tobin's q + 0.3045 0.2004 -0.2702 0.3254 Underpricing + 0.1780 0.0452 -0.0288 0.3371 industry adjusted return volatility + 0.2184 0.3900 0.2696 0.4533 Analysts disagreement + 0.0287 -0.0777 -0.1464 0.2130 KMO overall 0.5864 Eigenvalue 2.1433 0.5181 0.1823

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 3

Figure A2. Information asymmetry factor analysis including an accelerated SEO indicator variable.

Variable

Predicted correlation with info asymmetry

Factor 1 Factor 2 Factor 3 KMO

Firm size – 0.8077 -0.0511 -0.0930 0.7305 Firm age – 0.4930 -0.0311 0.2111 0.8328 No. of Analysts – 0.1262 0.0587 -0.0650 0.5048 Tangible assets – 0.2846 -0.2020 0.1994 0.7877 No. of prior stock offers – 0.5518 0.1389 0.1773 0.8321 Accelerated SEO – 0.4781 0.2270 0.0252 0.8359 Average spreads + -0.4526 -0.3000 0/1476 0.7288 Return volatility + -0.5347 0.2350 0.1044 0.7640 Discretionary accruals + -0.4089 0.2024 0.1744 0.8017 KMO overall 0.7782 Eigenvalue 2.1680 0.3048 0.1923

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 4

Figure A3. Information asymmetry factor analysis excluding tangible assets and the bid-ask spreads.

Variable

Predicted correlation with info asymmetry

Factor 1 Factor 2 Factor 3 KMO

Firm size – 0.7658 -0.0410 -0.0037 0.6735 Firm age – 0.4968 0.2029 0.0051 0.7783 No. of Analysts – 0.1820 -0.0104 0.0672 0.6514 No. of prior stock offers – 0.5127 0.2284 -0.0005 0.7391 Return volatility + -0.5667 0.1758 0.0093 0.7266 Discretionary accruals + -0.4470 0.1940 -0.0103 0.7694 KMO overall 0.7227 Eigenvalue 1.6174 0.1636 0.0047

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 5

Internet Appendix Section 2: Tables A5 – A8 report estimates from the tests reported in Tables 5 – 8 of the paper using the eight individual asymmetry information measures (firm size, firm age, number of analysts, ratio of tangible assets, number of prior stock offers, average spreads, return volatility and accounting quality) used in the factor analysis as well as several additional information asymmetry measures (tobin’s q, underpricing, industry-adjusted return volatility, analysts disagreement). To conserve space, the other control variables in model (3) are not shown.

Table A1. Determinants of SEO Lockup Agreements This table presents the estimates from nine specifications of a logit model using data from SEOs from 1996 through 2006. The (untransformed) dependent variable equals one when an issuer adopts a lockup agreement. The sample includes SEOs of U.S. common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are in brackets, and ***, **, and * represent significance at the 1%, 5%, and 10% levels.

Full Sample Conventional SEOs Accelerated SEOs VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Firm size -0.204*** -0.201*** -0.286*** -0.181*** -0.175*** -0.261*** -0.079** -0.172* -0.596*** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.031] [0.060] [0.002] Insider ownership % 0.004 0.005 0.005 0.007 -0.007 -0.012* [0.379] [0.310] [0.362] [0.252] [0.350] [0.070] Net selling frequency 0.111** 0.098** 0.109** 0.096* 0.152 0.158 [0.016] [0.048] [0.019] [0.053] [0.341] [0.334] Net selling value 0.001* 0.001 0.001* 0.001 0.043** 0.0031** [0.074] [0.274] [0.072] [0.237] [0.042] [0.036] Accelerated SEOs -0.298* [0.090] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Firm age -0.008** -0.007* -0.007* -0.009** -0.008* -0.009* -0.009** -0.008 -0.001* [0.013] [0.075] [0.096] [0.018] [0.087] [0.085] [0.019] [0.347] [0.094] Insider ownership % 0.006 0.005 0.006 0.006 -0.006 -0.006* [0.190] [0.303] [0.262] [0.265] [0.311] [0.075] Net selling frequency 0.074* 0.096** 0.074* 0.092* 0.109 0.08 [0.081] [0.044] [0.097] [0.053] [0.308] [0.488] Net selling value -0.001 0.001 -0.001 0.001 0.032 0.003* [0.169] [0.432] [0.159] [0.403] [0.636] [0.095] Accelerated SEOs -0.620***

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 6

[0.001] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) No. of analysts -0.006* -0.015* -0.031* -0.010* -0.021* -0.032 -0.031* -0.001 -0.090* [0.075] [0.090] [0.090] [0.052] [0.086] [0.108] [0.047] [0.977] [0.-94] Insider ownership % 0.006 0.005 0.005 0.006 -0.006 -0.007* [0.180] [0.286] [0.266] [0.257] [0.292] [0.091] Net selling frequency 0.075* 0.095* 0.074 0.091* 0.103 0.084* [0.081] [0.051] [0.102] [0.062] [0.339] [0.092] Net selling value -0.000 0.001 -0.001 0.001 0.053* 0.003* [0.169] [0.498] [0.163] [0.474] [0.098] [0.097] Accelerated SEOs -0.663*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Tangible assets -0.026* -0.033* -0.026* -0.034* -0.154 -0.125* -0.660** -1.150*** -1.214*** [0.081] [0.099] [0.056] [0.072] [0.259] [0.084] [0.016] [0.002] [0.003] Insider ownership % 0.007 0.005 0.006 0.006 -0.005** -0.008* [0.167] [0.293] [0.280] [0.281] [0.033] [0.072] Net selling frequency 0.097** 0.097** 0.096** 0.093* 0.131 0.102 [0.023] [0.041] [0.030] [0.052] [0.142] [0.288] Net selling value -0.001 0.001 -0.000 0.001 0.062** 0.003* [0.175] [0.460] [0.172] [0.430] [0.038] [0.054] Accelerated SEOs -0.650*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) No. of prior stock offers -0.157*** -0.115*** -0.072** -0.191*** -0.148*** -0.119*** -0.01 -0.013 -0.044 [0.000] [0.000] [0.043] [0.000] [0.000] [0.004] [0.821] [0.812] [0.641] Insider ownership % 0.005 0.005 0.005 0.006 -0.006 -0.011 [0.273] [0.321] [0.352] [0.294] [0.297] [0.500] Net selling frequency 0.073* 0.093** 0.073* 0.088* 0.100 0.097 [0.081] [0.047] [0.097] [0.062] [0.332] [0.419] Net selling value -0.001 0.001 -0.000 0.001 0.012 0.003* [0.172] [0.347] [0.161] [0.278] [0.599] [0.064] Accelerated SEOs -0.608*** [0.001] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Average spreads 0.062* 0.028* 0.04 0.040* 0.013* 0.031* 0.014 0.022 0.035 [0.067] [0.055] [0.431] [0.052] [0.060] [0.071] [0.912] [0.882] [0.836] Insider ownership % 0.005 0.005 0.005 0.006 -0.006 -0.008**

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 7

[0.237] [0.310] [0.328] [0.272] [0.308] [0.031] Net selling frequency 0.079* 0.098** 0.078* 0.094* 0.100 0.083 [0.064] [0.040] [0.083] [0.050] [0.344] [0.472] Net selling value -0.000 0.001 -0.000 0.001 -0.002 0.003* [0.178] [0.417] [0.164] [0.400] [0.615] [0.069] Accelerated SEOs -0.646*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Return volatility 18.214*** 12.781*** 10.787** 17.640*** 9.684*** 8.606* 3.070* 16.167* 13.952* [0.000] [0.000] [0.016] [0.000] [0.007] [0.072] [0.062] [0.052] [0.095] Insider ownership % 0.006 0.005 0.005 0.006 -0.008* -0.006* [0.226] [0.293] [0.305] [0.258] [0.090] [0.084] Net selling frequency 0.067 0.091* 0.070 0.088* 0.080 0.063 [0.135] [0.055] [0.137] [0.064] [0.362] [0.551] Net selling value -0.000 0.001 -0.000 0.001 0.009** 0.003** [0.286] [0.390] [0.385] [0.365] [0.049] [0.032] Accelerated SEOs -0.612*** [0.001] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Discretionary accruals 0.445** 0.450* 0.181** 0.274** 0.237** 0.121** 0.141* 5.502* 0.979* [0.020] [0.052] [0.053] [0.039] [0.037] [0.033] [0.054] [0.051] [0.085] Insider ownership % 0.002 0.002 -0.001 -0.001 -0.006* -0.003** [0.970] [0.947] [0.863] [0.912] [0.060] [0.020] Net selling frequency 0.120** 0.142** 0.126** 0.142** -0.074 -0.329 [0.029] [0.025] [0.027] [0.026] [0.623] [0.418] Net selling value -0.000** -0.001 -0.001* 0.001* 0.010 0.060** [0.030] [0.100] [0.058] [0.057] [0.709] [0.042] Accelerated SEOs -0.623*** [0.006] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Tobin's q 0.062*** 0.067*** 0.039* 0.047** 0.049** 0.031 0.094 0.315** 0.487*** [0.001] [0.004] [0.084] [0.015] [0.025] [0.156] [0.134] [0.028] [0.006] Insider ownership % -0.008 -0.006 0.007 0.007 -0.005 -0.003 [0.103] [0.232] [0.185] [0.218] [0.528] [0.713] Net selling frequency 0.095** 0.095** 0.094** 0.092* 0.066 0.028 [0.029] [0.042] [0.034] [0.051] [0.541] [0.830] Net selling value -0.000 0.001 -0.000 0.001 0.000 0.001 [0.122] [0.465] [0.125] [0.430] [0.620] [0.525] -0.639***

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 8

Accelerated SEOs [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Industry adjusted return volatility 6.953** 9.732** 9.603** 17.379** 1.600* 1.565* 64.851* 7.664*** 11.316** [0.019] [0.027] [0.022] [0.045] [0.056] [0.083] [0.060] [0.007] [0.029] Insider ownership % 0.006 0.005 0.005 0.006 -0.006 -0.002 [0.196] [0.298] [0.292] [0.270] [0.301] [0.810] Net selling frequency 0.075* 0.095** 0.076* 0.092* 0.102 0.068 [0.084] [0.046] [0.094] [0.054] [0.392] [0.632] Net selling value -0.000 0.001 -0.000 0.001 -0.001 0.001 [0.166] [0.467] [0.155] [0.440] [0.567] [0.801] Accelerated SEOs -0.650*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Underpricing 0.583 7.325 18.335 0.813** 8.702 1.42 0.07 82.531* 85.375 [0.115] [0.709] [0.409] [0.038] [0.688] [0.952] [0.923] [0.065] [0.133] Insider ownership % 0.006 0.005 0.005 0.006 -0.005 -0.004 [0.229] [0.337] [0.298] [0.268] [0.372] [0.677] Net selling frequency 0.077* 0.099** 0.076* 0.094* 0.074 0.057 [0.077] [0.038] [0.097] [0.051] [0.431] [0.618] Net selling value -0.000 0.001 -0.000 0.001 -0.000 0.001 [0.349] [0.431] [0.456] [0.428] [0.419] [0.951] Accelerated SEOs -0.656*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Analysts disagreement 0.136 0.228 0.236 0.344 0.368 0.331 0.129 -0.012 0.33 [0.676] [0.591] [0.569] [0.499] [0.563] [0.577] [0.659] [0.963] [0.900] Insider ownership % 0.013* 0.012* 0.012* 0.012* 0.002 0.019 [0.054] [0.068] [0.088] [0.086] [0.842] [0.651] Net selling frequency 0.039 0.026 0.035 0.019 0.030 0.274 [0.413] [0.607] [0.495] [0.728] [0.500] [0.172] Net selling value -0.001 0.001 -0.000 0.001 -0.000 0.001 [0.544] [0.510] [0.526] [0.460] [0.367] [0.879] Accelerated SEOs -0.641**

[0.013]

Table A2. Determinants of the Lockup Period

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 9

This table presents the estimates from nine specifications of a tobit model using data from SEOs from 1996 through 2006. The dependent variable equals the ex ante lockup period in calendar days. The sample criteria require the SEOs to be US common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. P-values are in brackets, and the symbols ***, **, and * represent significance at the 1%, 5%, and 10% levels.

Full Sample Conventional SEOs Accelerated SEOs VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Firm size -7.610*** -7.345*** -7.720*** -7.000*** -6.965*** -7.754*** -1.634* -2.271*** -5.924*** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.051] [0.010] [0.000] Insider ownership % 0.187* 0.138 0.217* 0.159 -0.023 -0.052 [0.077] [0.193] [0.090] [0.224] [0.867] [0.695] Net selling frequency 0.186 0.079 0.037 0.309 1.600 2.05 [0.818] [0.922] [0.968] [0.736] [0.299] [0.173] Net selling value -0.003 0.002 -0.001 0.002 0.014 0.015* [0.940] [0.746] [0.939] [0.756] [0.114] [0.067] Accelerated SEOs -16.617*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Firm age -0.336*** -0.221** -0.060* -0.479*** -0.338** -0.263* -0.338*** -0.271*** -0.107* [0.000] [0.028] [0.072] [0.000] [0.021] [0.084] [0.000] [0.002] [0.067] Insider ownership % 0.305*** 0.129 0.282** 0.135 0.069 0.001 [0.004] [0.229] [0.027] [0.306] [0.591] [0.996] Net selling frequency 0.220 0.274 0.210 0.501 1.400 1.725 [0.447] [0.736] [0.504] [0.589] [0.351] [0.261] Net selling value 0.002 0.002 0.001 0.001 0.013 0.013 [0.878] [0.993] [0.888] [0.992] [0.127] [0.123] Accelerated SEOs -22.836*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) No. of analysts -0.101* -0.050 -0.141 -0.065* -0.257 -0.285 -0.309 -0.345 -0.112 [0.082] [0.920] [0.778] [0.090] [0.678] [0.645] [0.588] [0.553] [0.852] Insider ownership % 0.304*** 0.128 0.271** 0.126 0.066 0 [0.003] [0.234] [0.031] [0.339] [0.614] [0.999] Net selling frequency 0.220** 0.261** 0.210 0.428 1.609 1.803 [0.048] [0.048] [0.505] [0.645] [0.288] [0.241] Net selling value 0.003 0.004 0.001 0.001 0.014 0.014 [0.882] [0.993] [0.891] [0.989] [0.110] [0.112] Accelerated SEOs -23.109***

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 10

[0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Tangible assets -3.556** -4.811 -0.278* -5.658 -8.353** -6.374 -8.422*** -9.953*** -10.669*** [0.029] [0.126] [0.093] [0.110] [0.041] [0.128] [0.006] [0.002] [0.001] Insider ownership % 0.284*** 0.127 0.258** 0.117 0.007 -0.018 [0.008] [0.237] [0.047] [0.376] [0.958] [0.895] Net selling frequency 0.054** 0.267** 0.023 0.485 1.186 1.701 [0.048] [0.042] [0.980] [0.601] [0.439] [0.262] Net selling value 0.004 0.005 0.001 0.001 0.012 0.012 [0.875] [0.993] [0.879] [0.985] [0.146] [0.152] Accelerated SEOs -23.095*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) No. of prior stock offers -8.351*** -6.133*** -3.482*** -10.888*** -7.790*** -5.748*** -1.716*** -2.069*** -1.799** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.004] [0.001] [0.011] Insider ownership % 0.254** 0.131 0.253** 0.137 0.043 -0.006 [0.013] [0.221] [0.042] [0.295] [0.736] [0.966] Net selling frequency 0.185** 0.251* 0.180 0.418 -1.469 -1.754 [0.017] [0.056] [0.562] [0.650] [0.327] [0.250] Net selling value 0.005 0.002 0.001 0.001 0.012 0.013 [0.918] [0.952] [0.929] [0.937] [0.143] [0.120] Accelerated SEOs -19.434*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Average spreads 11.071*** 10.921*** 7.622*** 10.210*** 10.637*** 6.892*** 0.339 0.317 4.378 [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.912] [0.918] [0.283] Insider ownership % 0.179* 0.102 0.132 0.107 0.061 -0.007 [0.093] [0.342] [0.312] [0.413] [0.642] [0.957] Net selling frequency 0.123 0.341 0.110 0.469 1.669 1.7 [0.669] [0.672] [0.724] [0.611] [0.273] [0.269] Net selling value 0.002 0.001 0.001 0.001 0.013 0.013 [0.647] [0.863] [0.688] [0.902] [0.126] [0.113] Accelerated SEOs -22.096*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Return volatility 636.535*** 395.074*** 389.296*** 515.692*** 225.891** 313.512** 338.817*** 442.448*** 600.290*** [0.000] [0.000] [0.000] [0.000] [0.029] [0.014] [0.002] [0.000] [0.000] Insider ownership % 0.262** 0.129 0.241* 0.133 0.017 -0.034 [0.013] [0.225] [0.064] [0.308] [0.897] [0.795] Net selling frequency 0.244** 0.223* 0.219 0.401 1.723 1.653 [0.032] [0.082] [0.483] [0.662] [0.252] [0.276]

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 11

Net selling value -0.001 0.001 -0.001 0.001 0.012 0.013 [0.847] [0.975] [0.862] [0.974] [0.138] [0.108] Accelerated SEOs -21.699*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Discretionary accruals 36.161*** 36.775*** 31.907*** 33.423*** 34.780*** 30.282*** 16.968** 16.021** 11.136** [0.000] [0.000] [0.000] [0.000] [0.000] [0.003] [0.029] [0.034] [0.048] Insider ownership % 0.090 -0.023 0.109 -0.016 -0.158 -0.193 [0.530] [0.870] [0.529] [0.924] [0.343] [0.247] Net selling frequency 0.615* 1.036** 0.599 1.317 -1.332 -2.222 [0.078] [0.028] [0.628] [0.264] [0.510] [0.268] Net selling value -0.001 -0.001 -0.000 -0.001 0.003 0.014 [0.954] [0.807] [0.971] [0.806] [0.886] [0.478] Accelerated SEOs -18.320*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Tobin's q 0.527** 0.805*** 0.788 0.389 0.573* 0.732 0.339 2.121*** 1.741* [0.032] [0.008] [0.122] [0.177] [0.088] [0.209] [0.296] [0.009] [0.085] Insider ownership % 0.298*** 0.134 0.280** 0.134 -0.016 -0.016 [0.006] [0.214] [0.031] [0.311] [0.903] [0.904] Net selling frequency 0.119 0.241 0.080 0.418 1.634 -1.874 [0.886] [0.767] [0.932] [0.652] [0.289] [0.221] Net selling value 0.003 0.003 0.001 0.001 0.013 0.014 [0.915] [0.970] [0.930] [0.968] [0.120] [0.111] Accelerated SEOs -22.998*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Industry adjusted return volatility 417.993** 66.649*** 62.517*** 101.458 55.664*** 63.567*** 3,073.646*** 166.666*** 145.147***

[0.045] [0.000] [0.000] [0.877] [0.001] [0.000] [0.000] [0.000] [0.000] Insider ownership % 0.283*** 0.121 0.252* 0.119 0.032 -0.016 [0.008] [0.258] [0.056] [0.366] [0.804] [0.905] Net selling frequency 0.267 0.261 0.250 0.43 -1.568 -1.597 [0.357] [0.748] [0.429] [0.641] [0.294] [0.292] Net selling value 0.002 0.002 0.001 0.001 0.013 0.013 [0.978] [0.953] [0.977] [0.952] [0.134] [0.127] Accelerated SEOs -22.702*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9)

Underpricing 44.463*** 469.679** 984.276 47.445*** 777.928 378.808 27.661 639.120*** 178.042***

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 12

[0.000] [0.024] [0.130] [0.000] [0.313] [0.623] [0.176] [0.000] [0.001] Insider ownership % 0.267** 0.122 0.231* 0.122 0.093 0.022 [0.013] [0.257] [0.079] [0.356] [0.475] [0.868] Net selling frequency 0.197 0.299 0.187 0.451 -1.682 -1.799 [0.494] [0.713] [0.552] [0.627] [0.263] [0.236] Net selling value -0.001 0.001 -0.001 0.001 0.011 0.012 [0.858] [0.979] [0.870] [0.982] [0.181] [0.165] Accelerated SEOs -23.009*** [0.000] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Analysts disagreement 1.600 0.625 3.145 6.311 3.112 3.395 -2.646 -2.703 -9.382 [0.873] [0.954] [0.775] [0.637] [0.847] [0.838] [0.771] [0.465] [0.325] Insider ownership % 0.298** 0.143 0.237 0.113 0.035 0.068 [0.029] [0.329] [0.161] [0.533] [0.876] [0.683] Net selling frequency 1.297* 0.079 -1.159 0.587 0.023 0.043 [0.095] [0.960] [0.543] [0.774] [0.897] [0.445] Net selling value 0.001 0.001 0.001 0.001 -0.053 0.063 [0.330] [0.635] [0.405] [0.662] [0.761] [0.897] Accelerated SEOs -23.503*** [0.000]

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 13

Table A3. Determinants of Early Releases of SEO Lockups This table presents the estimates from nine specifications of a logit model using data from SEOs with lockup arrangements from 1996 through 2006. The (untransformed) dependent variable equals one when an issuer released the lockup before the end of the ex ante lockup period. The sample criteria require the SEOs to be US common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. P-values are in brackets, and the symbols ***, **, and * represent significance at the 1%, 5%, and 10% levels.

Full Sample Conventional SEOs Accelerated SEOs VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Firm size 0.001 0.031 -0.080 -0.018 0.008 -0.095* 0.024 0.102 0.011 [0.434] [0.212] [0.113] [0.488] [0.808] [0.090] [0.607] [0.115] [0.929] CR(0,5) 0.606** 0.536*** 0.874*** 0.635** 1.135*** 1.068*** 0.439** 0.649** 0.602** [0.022] [0.004] [0.003] [0.020] [0.000] [0.005] [0.039] [0.021] [0.037] Ex ante lockup days 0.001 -0.000 0.001 0.000 0.001 0.001 0.003 0.004 0.007 [0.434] [0.936] [0.318] [0.817] [0.585] [0.657] [0.315] [0.316] [0.165] Info asymmetry*CR(0,5) -0.508 -0.775 -0.954 -0.821 -0.204 -0.910 [0.114] [0.114] [0.105] [0.128] [0.537] [0.369] Accelerated SEOs -0.040 [0.783] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Firm age 0.007 0.012 0.014 0.013 0.017 0.022 0.005 0.003 -0.004 [0.118] [0.107] [0.104] [0.143] [0.101] [0.101] [0.308] [0.703] [0.585] CR(0,5) 0.570** 0.723*** 0.893*** 0.629** 1.067*** 0.984*** 0.823** 0.563** 0.740* [0.034] [0.002] [0.003] [0.012] [0.001] [0.008] [0.038] [0.030] [0.077] Ex ante lockup days 0.001 0.001 0.002 0.000 0.001 0.001 0.004 0.003 0.007 [0.112] [0.364] [0.184] [0.551] [0.557] [0.454] [0.151] [0.488] [0.144] Info asymmetry*CR(0,5) 0.532 0.857 0.835 1.775 -0.629 -0.900 [0.113] [0.111] [0.109] [0.136] [0.736] [0.354] Accelerated SEOs -0.056 [0.692] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) No. of analysts 0.024* 0.033* 0.068*** 0.010 0.017 0.051** 0.079*** 0.098** 0.141*** [0.091] [0.057] [0.001] [0.539] [0.382] [0.032] [0.008] [0.011] [0.005] CR(0,5) 0.596** 0.835*** 1.142*** 0.650** 1.172*** 1.264*** 0.550* 0.908** 0.507** [0.019] [0.001] [0.001] [0.013] [0.000] [0.002] [0.068] [0.014] [0.041]

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 14

Ex ante lockup days 0.001 0.001 0.001 0.001 0.000 0.001 0.005 0.002 0.006 [0.110] [0.447] [0.338] [0.477] [0.628] [0.627] [0.105] [0.561] [0.235] Info asymmetry*CR(0,5) -0.664*** -1.087*** -0.969*** -1.094** -0.915 -1.243 [0.006] [0.004] [0.005] [0.012] [0.623] [0.279] Accelerated SEOs -0.108 [0.438] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Tangible assets -0.071 -0.208 -0.160 -0.021 -0.175 -0.144 -0.076 -0.228 -0.262 [0.418] [0.146] [0.212] [0.838] [0.147] [0.332] [0.650] [0.302] [0.331] CR(0,5) 0.602** 1.774*** 0.907*** 0.670** 1.153*** 1.121*** 0.387** 0.503* 0.635** [0.0 25] [0.001] [0.003] [0.016] [0.000] [0.005] [0.046] [0.058] [0.022] Ex ante lockup days 0.001 0.001 0.001 -0.000 0.000 0.001 0.003 0.003 0.006 [0.382] [0.444] [0.249] [0.998] [0.650] [0.534] [0.328] [0.510] [0.201] Info asymmetry*CR(0,5) -0.584 -0.892 -0.913 -0.972 -0.751 -0.918 [0.109] [0.109] [0.106] [0.117] [0.688] [0.357] Accelerated SEOs -0.074 [0.595] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) No. of prior stock offers -0.081*** -0.048 -0.050 -0.081** -0.078* -0.088* -0.051 0.031 0.007 [0.000] [0.148] [0.228] [0.015] [0.073] [0.097] [0.112] [0.605] [0.918] CR(0,5) 0.541** 1.021*** 0.898*** 0.629** 1.121*** 1.116*** 0.810** 0.572** 0.608** [0.015] [0.002] [0.003] [0.014] [0.000] [0.004] [0.035] [0.027] [0.032] Ex ante lockup days 0.001 0.001 0.001 0.000 0.000 0.001 0.004 0.003 0.007 [0.310] [0.490] [0.287] [0.687] [0.714] [0.642] [0.168] [0.446] [0.159] Info asymmetry*CR(0,5) -1.552 -1.890 -1.929 -1.978 -0.752 -1.899 [0.111] [0.109] [0.105] [0.116] [0.691] [0.362] Accelerated SEOs -0.041 [0.774] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Average spreads -0.161*** -0.195*** -0.113 -0.156*** -0.185*** -0.117* -1.135*** -1.436 -0.728 [0.000] [0.000] [0.170] [0.000] [0.000] [0.070] [0.006] [0.109] [0.287] CR(0,5) 0.491** 1.631*** 0.810*** 0.557** 1.033*** 1.048*** 0.957** 0.952* 0.246** [0.027] [0.003] [0.005] [0.017] [0.001] [0.006] [0.042] [0.056] [0.020] Ex ante lockup days 0.002** 0.002 0.002 0.001* 0.001 0.001 0.005 0.003 0.007 [0.010] [0.108] [0.156] [0.098] [0.205] [0.365] [0.100] [0.415] [0.167] Info asymmetry*CR(0,5) -0.813 -0.969 -1.171 -1.085 -1.156 -0.914 [0.104] [0.107] [0.103] [0.113] [0.540] [0.358] Accelerated SEOs -0.091 [0.514]

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 15

VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Return volatility 8.551*** 4.141 7.496 7.012*** 3.150 6.881 11.704* 3.887 12.547 [0.000] [0.149] [0.169] [0.005] [0.312] [0.120] [0.067] [0.693] [0.368] CR(0,5) 0.971** 1.096*** 0.998*** 1.052** 1.175*** 1.184*** 0.474** 0.376** 1.042** [0.015] [0.001] [0.002] [0.012] [0.000] [0.004] [0.017] [0.020] [0.024] Ex ante lockup days 0.001* 0.001 0.001 0.001 0.000 0.001 0.004 0.003 0.007 [0.085] [0.448] [0.324] [0.336] [0.636] [0.651] [0.164] [0.479] [0.167] Info asymmetry*CR(0,5) -0.527* -0.851* -0.903 -0.938 -0.493 -0.535 [0.072] [0.090] [0.106] [0.119] [0.793] [0.470] Accelerated SEOs -0.072 [0.603] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Discretionary accruals 0.461** 0.449** 0.660** 0.468** 0.460* 0.644** 0.343 0.390 1.046 [0.036] [0.050] [0.024] [0.041] [0.054] [0.035] [0.660] [0.623] [0.301] CR(0,5) 0.715** 0.760*** 0.893*** 0.792** 1.155*** 1.104*** 0.437* 0.591** 0.245** [0.048] [0.001] [0.003] [0.031] [0.000] [0.005] [0.064] [0.015] [0.021] Ex ante lockup days 0.001 0.001 0.001 0.001 0.000 0.001 0.004 0.003 0.007 [0.136] [0.483] [0.319] [0.307] [0.697] [0.653] [0.280] [0.487] [0.169] Info asymmetry*CR(0,5) -0.510** -0.783** -0.856*** -0.842** -0.675 -1.080 [0.013] [0.014] [0.008] [0.028] [0.713] [0.312] Accelerated SEOs -0.079 [0.572] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Tobin's q 0.019 0.026* 0.026 0.015 0.024* 0.030* 0.080* 0.038 0.052 [0.186] [0.072] [0.139] [0.273] [0.100] [0.093] [0.086] [0.460] [0.496] CR(0,5) 0.595** 1.158*** 0.889*** 0.626** 1.157*** 1.115*** 0.627* 0.628* 0.383** [0.028] [0.001] [0.003] [0.024] [0.000] [0.005] [0.058] [0.069] [0.033] Ex ante lockup days 0.001 0.001 0.001 0.000 0.000 0.001 0.003 0.003 0.007 [0.343] [0.426] [0.262] [0.738] [0.631] [0.588] [0.348] [0.464] [0.151] Info asymmetry*CR(0,5) -0.550 -0.882 -0.907 -0.952 -0.733 -0.985 [0.111] [0.109] [0.106] [0.119] [0.694] [0.346] Accelerated SEOs -0.085 [0.541] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Industry adjusted return volatility 0.577 -0.557 0.563 0.507 -0.536 0.465 0.978 -2.206 1.655 [0.333] [0.503] [0.571] [0.416] [0.543] [0.662] [0.651] [0.428] [0.619] CR(0,5) 0.395** 0.966*** 0.681** 0.483** 1.349*** 0.929** 0.094** 0.228** 0.062* [0.037] [0.002] [0.027] [0.022] [0.001] [0.027] [0.047] [0.025] [0.077]

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 16

Ex ante lockup days 0.001 0.001 0.001 0.001 0.001 0.001 0.005 0.003 0.007 [0.123] [0.364] [0.275] [0.516] [0.556] [0.581] [0.118] [0.410] [0.168] Info asymmetry*CR(0,5) -0.590 -0.931 -0.942 -1.013 -0.837 -0.816 [0.109] [0.108] [0.105] [0.115] [0.658] [0.383] Accelerated SEOs -0.088 [0.530] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Underpricing 15.278** 5.937* 34.502** 7.979* 4.214* 34.118** 83.044 50.361** 91.958** [0.045] [0.098] [0.021] [0.083] [0.064] [0.023] [0.111] [0.021] [0.037] CR(0,5) 0.828** 0.742*** 0.891*** 0.936** 1.134*** 1.093*** 0.575* 0.482* 0.689* [0.039] [0.001] [0.003] [0.026] [0.000] [0.005] [0.066] [0.067] [0.069] Ex ante lockup days 0.001* 0.001 0.001 0.001 0.001 0.001 0.004 0.003 0.007 [0.053] [0.395] [0.238] [0.295] [0.595] [0.521] [0.154] [0.484] [0.158] Info asymmetry*CR(0,5) -0.590 -0.913 -0.944 -0.985 -0.700 -0.994 [0.109] [0.108] [0.105] [0.117] [0.707] [0.335] Accelerated SEOs -0.081 [0.563] VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) Analysts disagreement -0.175 -0.216 -0.682 -0.134 -0.043 -1.203 -0.252 -0.603** -0.195 [0.522] [0.430] [0.273] [0.758] [0.923] [0.106] [0.433] [0.048] [0.577] CR(0,5) 0.380** 1.389* 1.384** 0.703** 1.416** 1.561** 1.147** 1.948* 1.276** [0.049] [0.098] [0.041] [0.023] [0.013] [0.023] [0.038] [0.052] [0.019] Ex ante lockup days 0.001 0.000 0.000 0.001 0.000 0.000 0.006 -0.003 0.005 [0.155] [0.868] [0.775] [0.479] [0.807] [0.904] [0.163] [0.539] [0.572] Info asymmetry*CR(0,5) -1.149 -0.782* -1.115** -1.686** 0.329 -2.529 [0.210] [0.089] [0.048] [0.028] [0.907] [0.392] Accelerated SEOs 0.022 [0.915]

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 17

Table A4. Determinants of Ex Post Lockup Period

This table presents the estimates from nine specifications of a tobit model using data from SEOs from 1996 through 2006. The dependent variable equals the ex-post lockup period in calendar days. The sample criteria require the SEOs to be U.S. common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are in brackets, and ***, **, and * represent significance at the 1%, 5%, and 10% levels.

VARIABLES (1) (2) (3) (4) Firm size 2.893 3.00 3.018 4.433 [0.167] [0.102] [0.121] [0.107] CR(0,5) -21.048** -27.891** -32.274*** -12.525** [0.046] [0.022] [0.001] [0.049] Ex ante lockup days 0.761*** 0.649*** 0.072*** 0.059*** [0.000] [0.000] [0.001] [0.005] VARIABLES (1) (2) (3) (4) Firm age 0.300 0.207 0.560 0.438 [0.120] [0.137] [0.121] [0.132] CR(0,5) -20.774** -26.791** -27.790*** -10.572** [0.039] [0.029] [0.005] [0.031] Ex ante lockup days 0.756*** 0.641*** 0.121*** 0.052** [0.000] [0.000] [0.000] [0.012] VARIABLES (1) (2) (3) (4) No. of analysts -0.542 -1.063** 0.174 0.209 [0.196] [0.026] [0.652] [0.581] CR(0,5) -22.289** -28.445** -27.301*** -10.425** [0.027] [0.020] [0.008] [0.049] Ex ante lockup days 0.753*** 0.642*** 0.121*** 0.053** [0.000] [0.000] [0.000] [0.012] VARIABLES (1) (2) (3) (4) Tangible assets 3.389 0.629 4.685* 5.670 [0.210] [0.833] [0.055] [0.123]

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 18

CR(0,5) -21.140** -27.445** -28.936*** -10.616** [0.045] [0.025] [0.004] [0.038] Ex ante lockup days 0.757*** 0.641*** 0.062*** 0.053** [0.000] [0.000] [0.004] [0.012] VARIABLES (1) (2) (3) (4) No. of prior stock offers 1.830 1.014 3.077 1.516* [0.103] [0.218] [0.132] [0.097] CR(0,5) -20.901** -27.247** -27.131*** -10.669** [0.038] [0.026] [0.008] [0.036] Ex ante lockup days 0.764*** 0.644*** 0.131*** 0.059*** [0.000] [0.000] [0.000] [0.006] VARIABLES (1) (2) (3) (4) Average spreads 2.352** 1.098 -1.203 -1.965 [0.026] [0.443] [0.337] [0.149] CR(0,5) -20.367** -26.984** -21.163** -11.690** [0.045] [0.028] [0.043] [0.023] Ex ante lockup days 0.743*** 0.639*** 0.125*** 0.061*** [0.000] [0.000] [0.000] [0.005] VARIABLES (1) (2) (3) (4) Return volatility -130.385* -213.432** -248.656*** -195.564** [0.055] [0.035] [0.000] [0.015] CR(0,5) -25.896** -29.612** -19.566** -14.649** [0.015] [0.015] [0.024] [0.019] Ex ante lockup days 0.752*** 0.658*** 0.116*** 0.055*** [0.000] [0.000] [0.000] [0.009] VARIABLES (1) (2) (3) (4) Discretionary accruals -8.268 -17.593 -10.117* -11.831 [0.246] [0.133] [0.060] [0.147] CR(0,5) -17.016** -25.267* -32.009*** -15.194** [0.014] [0.059] [0.002] [0.018] Ex ante lockup days 0.661*** 0.569*** 0.128*** 0.044** [0.000] [0.000] [0.000] [0.046]

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 19

VARIABLES (1) (2) (3) (4) Tobin's q -0.150 -0.768* -0.444 -0.486 [0.541] [0.096] [0.125] [0.114] CR(0,5) -20.898** -28.076** -28.260*** -11.010** [0.048] [0.022] [0.005] [0.021] Ex ante lockup days 0.756*** 0.642*** 0.062*** 0.056*** [0.000] [0.000] [0.004] [0.009] VARIABLES (1) (2) (3) (4) Industry adjusted return volatility -1.067 -1.933 -3.419 -5.507 [0.913] [0.890] [0.850] [0.780] CR(0,5) -21.111** -26.811** -26.674** -8.674** [0.044] [0.039] [0.020] [0.021] Ex ante lockup days 0.753*** 0.642*** 0.122*** 0.055** [0.000] [0.000] [0.000] [0.010] VARIABLES (1) (2) (3) (4) Underpricing -214.108 667.900 -548.285*** -232.288** [0.684] [0.276] [0.004] [0.025] CR(0,5) -24.263** -27.588** -13.967 -10.763 [0.024] [0.024] [0.215] [0.331] Ex ante lockup days 0.737*** 0.640*** 0.114*** 0.055*** [0.000] [0.000] [0.000] [0.009] VARIABLES (1) (2) (3) (4) Analysts disagreement 4.327 4.508 -5.384 -6.671 [0.669] [0.698] [0.582] [0.712] CR(0,5) -12.103** -3.870** -27.999* -12.104** [0.035] [0.027] [0.069] [0.021] Ex ante lockup days 0.746*** 0.611*** 0.086*** 0.079*** [0.000] [0.000] [0.003] [0.005]

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 20

Internet Appendix Section 3: Tables A5 – A8 report estimates from the tests reported in Tables 5 – 8 of the paper, using all eight asymmetry information measures separately (firm size, firm age, number of analysts, ratio of tangible assets, number of prior stock offers, average spreads, return volatility, accounting quality) used in the analysis instead of the single information asymmetry factor.

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 21

Table A5. Determinants of SEO Lockup Agreements Estimates are from the nine specifications of a logit model using data from SEOs from 1996 through 2006. The (untransformed) dependent variable equals one when an issuer adopts a lockup agreement. The sample includes SEOs of U.S. common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include year and industry fixed effects. Z-statistics, based on White robust standard errors with adjustment for SEO clustering by issuer, are reported in brackets. ***, **, and * represent 1%, 5%, and 10% significance, respectively.

Full Sample Conventional SEOs Accelerated SEOs

VARIABLES (1) (2) (3)

Firm size -0.521*** -0.519*** -0.563***

[0.000] [0.000] [0.003]

Firm age -0.005 -0.006 -0.004

[0.452] [0.433] [0.739]

No. of analysts -0.032 -0.028 -0.034

[0.198] [0.282] [0.558]

Tangible assets -0.021 -0.195 -0.712*

[0.907] [0.286] [0.054]

No. of prior stock offers -0.013** -0.012* -0.027*

[0.024] [0.039] [0.086]

Average spreads 0.068 0.084 0.022*

[0.300] [0.234] [0.086]

Return volatility 0.085** 2.034** 13.076*

[0.018] [0.025] [0.052]

Discretionary accruals 0.857** 0.884** 1.160

[0.033] [0.031] [0.544]

Insider ownership % -0.005 -0.005 -0.036**

[0.440] [0.509] [0.048]

Net selling frequency 0.166** 0.170** 0.205

[0.017] [0.016] [0.479]

Net selling value 0.001* 0.001* 0.005***

[0.095] [0.079] [0.006]

Accelerated SEOs -0.021

[0.929]

Leverage 1.029*** 1.251*** 0.811

[0.002] [0.001] [0.260]

Share turnover -0.185** -0.174* 0.005

[0.034] [0.067] [0.980]

Log (proceeds) 0.431*** 0.389*** 0.767***

[0.000] [0.000] [0.000]

Underwriter rank -0.012 -0.024 -0.333**

[0.795] [0.623] [0.015]

Secondary shares (%) -0.001 0.000 -0.005

[0.547] [0.923] [0.204]

ROA 0.001 0.054 0.384

[0.997] [0.794] [0.595]

Constant 2.324*** 2.723*** 1.758

[0.000] [0.000] [0.228]

Year fixed effects Yes Yes Yes

Industry fixed effects Yes Yes Yes

Observations 1,597 1,096 501

Pseudo R2 0 .267 0.218 0.491

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 22

Table A6. Determinants of the Lockup Period This table presents the estimates from nine specifications of a tobit model using data from SEOs from 1996 through 2006. The dependent variable equals the ex ante lockup period in calendar days. The sample criteria require the SEOs to be US common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are in brackets, and ***, **, and * represent significance at the 1%, 5%, and 10% levels.

Full Sample Conventional SEOs Accelerated SEOs

VARIABLES (1) (2) (3)

Firm size -11.797*** -11.720*** -9.399***

[0.000] [0.000] [0.000]

Firm age -0.229 -0.223 -0.145

[0.110] [0.242] [0.199]

No. of analysts 0.438 0.399 0.433

[0.458] [0.574] [0.448]

Tangible assets -1.018 -0.387 -3.622

[0.802] [0.939] [0.335]

No. of prior stock offers -1.238 -2.536 -0.383

[0.354] [0.171] [0.698]

Average spreads 5.535*** 4.805** 2.127*

[0.003] [0.019] [0.051]

Return volatility 58.637** 44.080** 165.311*

[0.023] [0.037] [0.083]

Discretionary accruals 3.338** 3.531** 13.227*

[0.020] [0.037] [0.092]

Insider ownership % -0.095 -0.067 -0.277**

[0.492] [0.690] [0.041]

Net selling frequency 1.214** 1.632* 0.298**

[0.037] [0.056] [0.038]

Net selling value -0.003 -0.003 -0.003

[0.554] [0.551] [0.757]

Accelerated SEOs -8.159*

[0.052]

Leverage 22.696*** 28.300*** 13.743**

[0.001] [0.001] [0.030]

Share turnover -3.933* -2.682 -5.578***

[0.053] [0.253] [0.007]

Log (proceeds) 8.320*** 5.042* 12.609***

[0.000] [0.073] [0.000]

Underwriter rank 0.515 0.313 -0.757

[0.614] [0.793] [0.480]

Secondary shares (%) -0.024 0.033 -0.087**

[0.541] [0.496] [0.018]

ROA 7.087 5.879 11.694*

[0.142] [0.280] [0.069]

Constant 110.176*** 127.903*** 71.980***

[0.000] [0.000] [0.000]

Year fixed effects Yes Yes Yes

Industry fixed effects Yes Yes Yes

Observations 1,597 1,096 501

Pseudo R2 0 .184 0.166 0.464

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 23

Table A7. Determinants of Early Releases of SEO Lockups This table presents the estimates from nine specifications of a logit model using data from SEOs with lockup arrangements from 1996 through 2006. The (untransformed) dependent variable equals one when an issuer released the lockup before the end of the ex ante lockup period. The sample criteria require the SEOs to be US common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are in brackets, and ***, **, and * represent significance at the 1%, 5%, and 10% levels.

Full Sample Conventional SEOs Accelerated SEOs

VARIABLES (1) (2) (3)

Firm size 0.032 0.031 0.064

[0.617] [0.659] [0.731]

Firm age -0.013** -0.020*** -0.003

[0.019] [0.004] [0.715]

No. of analysts 0.069 0.058 0.144

[0.101] [0.119] [0.104]

Tangible assets -0.051 -0.010 -0.162

[0.699] [0.950] [0.557]

No. of prior stock offers -0.003 -0.033 0.068

[0.948] [0.579] [0.429]

Average spreads -0.124** -0.134** -0.812

[0.048] [0.039] [0.259]

Return volatility 5.927 4.563 19.378

[0.193] [0.352] [0.212]

Discretionary accruals 0.675 0.612* 1.381

[0.130] [0.061] [0.225]

CR(0,5) 1.110*** 1.117*** 0.258**

[0.001] [0.005] [0.023]

Ex ate lockup days 0.001 0.001 0.005

[0.254] [0.505] [0.260]

Info asymmetry*CR(0,5) -0.927 -0.814 -0.371

[0.109] [0.134] [0.278]

Insider ownership % 0.011** 0.011** 0.011

[0.020] [0.039] [0.271]

Net selling frequency 0.097** 0.105* 0.050

[0.048] [0.055] [0.748]

Net selling value 0.001* 0.001* -0.001

[0.082] [0.080] [0.208]

Accelerated SEOs -0.088

[0.548]

Leverage -0.434* -0.235 -1.087*

[0.058] [0.381] [0.071]

Share turnover 0.001 0.009 -0.126

[0.992] [0.905] [0.554]

Log (proceeds) 0.189** 0.137 0.197

[0.032] [0.159] [0.442]

Underwriter rank 0.023 0.023 0.006

[0.516] [0.552] [0.949]

Secondary shares (%) 0.008*** 0.009*** 0.005

[0.000] [0.000] [0.114]

ROA 0.350** 0.325* 0.863

[0.032] [0.052] [0.123]

Constant -1.254** -0.915 -1.840

[0.013] [0.102] [0.436]

Year fixed effects Yes Yes Yes

Industry fixed effects Yes Yes Yes

Observations 1,485 1,226 259

Pseudo R2 0.138 0.144 0.175

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 24

Table A8. Determinants of Ex Post Lockup Period This table presents the estimates from nine specifications of a tobit model using data from SEOs from 1996 through 2006. The dependent variable equals the ex-post lockup period in calendar days. The sample criteria require the SEOs to be U.S. common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are in brackets, and ***, **, and * represent significance at the 1%, 5%, and 10% levels.

VARIABLES (1) (2) (3) (4)

Firm size -1.921* 0.936 1.366 4.771*

[0.091] [0.569] [0.135] [0.080]

Firm age 0.345 0.308 0.427*** 0.391***

[0.109] [0.123] [0.000] [0.001]

No. of analysts -0.128 -1.043* 0.761** 0.405

[0.811] [0.069] [0.044] [0.295]

Tangible assets 5.809 0.914 0.719 2.197

[0.196] [0.811] [0.787] [0.426]

No. of prior stock offers 2.634 0.941 0.798 -0.737

[0.127] [0.456] [0.391] [0.456]

Average spreads 0.671 -0.310 -0.988 -2.017

[0.156] [0.860] [0.452] [0.163]

Return volatility 65.893 -37.822 53.624 129.497

[0.528] [0.766] [0.522] [0.169]

Discretionary accruals -14.190 -11.494 -4.913 -0.651

[0.171] [0.188] [0.378] [0.916]

CR(0,5) -21.485* -27.266** -16.160** -15.312**

[0.081] [0.041] [0.031] [0.017]

Ex ate lockup days 0.669*** 0.591*** 0.055** 0.055**

[0.000] [0.000] [0.015] [0.014]

Insider ownership % -0.310** -0.022*

[0.017] [0.079]

Net selling frequency -1.092 -0.337

[0.253] [0.678]

Net selling value -0.006* -0.002*

[0.066] [0.058]

Accelerated SEOs 1.777 1.933

[0.654] [0.545]

Leverage 7.882 -6.957

[0.222] [0.155]

Share turnover 0.166 -1.652

[0.931] [0.217]

Log (proceeds) -5.488** -4.639***

[0.013] [0.008]

Underwriter rank -1.314 0.336

[0.172] [0.641]

Secondary shares (%) -0.284*** -0.083***

[0.000] [0.003]

ROA -14.396*** -2.574

[0.001] [0.427]

Constant -14.718 25.957* 5.494 14.574

[0.145] [0.053] [0.493] [0.154]

Year fixed effects Yes Yes Yes Yes

Industry fixed effects Yes Yes Yes Yes

Observations 1,597 1,597 579 579

Pseudo R2 0.417 0.427 0.144 0.214

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 25

Internet Appendix Section 4: Tables A9 – A12 report estimates from the tests reported in Tables 5 – 8 of the paper, using Net selling frequency and Net selling value measured over one year before the SEO filing date.

Table A9. Determinants of SEO Lockup Agreements This table presents the estimates from nine specifications of a logit model using data from SEOs from 1996 through 2006. The (untransformed) dependent variable equals one when an issuer adopts a lockup agreement. The sample includes SEOs of U.S. common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. P-values are shown in brackets and the symbols ***, **, and * represent significance at the 1%, 5%, and 10% levels.

Full Sample Conventional SEOs Accelerated SEOs VARIABLES (1) (2) (3) Information asymmetry 0.468*** 0.463*** 0.476** [0.000] [0.000] [0.024] Insider ownership % -0.003 -0.003 -0.085* [0.644] [0.700] [0.094] Net selling frequency 0.194*** 0.203*** 0.765* [0.005] [0.004] [0.084] Net selling value 0.002* 0.003** 0.021 [0.091] [0.015] [0.125] Accelerated SEOs -0.244 [0.381] Leverage 0.522 0.608 0.728 [0.144] [0.115] [0.443] Share turnover -0.153* -0.152* -0.072 [0.072] [0.081] [0.811] Log (proceeds) 0.219** 0.232** 0.239** [0.018] [0.019] [0.030] Underwriter rank -0.036 -0.049 -0.276 [0.481] [0.349] [0.132] Secondary shares (%) -0.001 -0.001 -0.012* [0.448] [0.690] [0.080] ROA -0.039 -0.036 -1.058 [0.896] [0.907] [0.451] Constant 0.525 0.554 1.779 [0.302] [0.296] [0.327] Year fixed effects Yes Yes Yes Industry fixed effects Yes Yes Yes Observations 1,597 1,096 501 Pseudo R2 0 .254 0.202 0.329

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 26

Table A10. Determinants of the Lockup Period

This table presents the estimates from nine specifications of a tobit model using data from SEOs from 1996 through 2006. The dependent variable equals the ex ante lockup period in calendar days. The sample criteria require the SEOs to be US common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are in brackets, and ***, **, and * represent significance at the 1%, 5%, and 10% levels.

Full Sample

Conventional SEOs

Accelerated SEOs

VARIABLES (1) (2) (3) Information asymmetry 14.408*** 15.748*** 10.187*** [0.000] [0.000] [0.000] Insider ownership % 0.027 0.069 -0.210 [0.852] [0.705] [0.130] Net selling frequency 1.503* 2.038* -0.161 [0.060] [0.093] [0.851] Net selling value 0.001 0.024 0.008 [0.923] [0.244] [0.383] Accelerated SEOs -10.767** [0.022] Leverage 16.330** 21.136** 8.790 [0.024] [0.017] [0.156] Share turnover -5.616*** -5.398** -2.986 [0.005] [0.018] [0.126] Log (proceeds) 2.181 0.472 6.801*** [0.342] [0.862] [0.003] Underwriter rank -0.161 -0.005 -1.489 [0.885] [0.997] [0.177] Secondary shares (%) -0.066 -0.019 -0.123*** [0.129] [0.724] [0.002] ROA 7.710 8.591 -1.504 [0.149] [0.156] [0.822] Constant 91.077*** 92.979*** 33.682** [0.000] [0.000] [0.042] Year fixed effects Yes Yes Yes Industry fixed effects Yes Yes Yes Observations 1,597 1,096 501 Pseudo R2 0 .223 0.134 0.424

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 27

Table A11. Determinants of Early Releases of SEO Lockups

This table presents the estimates from nine specifications of a logit model using data from SEOs with lockup arrangements from 1996 through 2006. The (untransformed) dependent variable equals one when an issuer released the lockup before the end of the ex ante lockup period. The sample criteria require the SEOs to be US common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are in brackets, and ***, **, and * represent significance at the 1%, 5%, and 10% levels.

Full Sample Conventional SEOs Accelerated SEOs Variables: (1) (2) (3) CR(0,5) 0.851** 0.972** 0.720** [0.019] [0.026] [0.016] Information asymmetry 0.346* 0.452 0.018 [0.087] [0.120] [0.125] Ex ante lockup days 0.002 0.001 0.001 [0.173] [0.282] [0.143] Info asymmetry*CR(0,5) 0.119 0.042 2.591 [0.844] [0.952] [0.146] Insider ownership % 0.003** 0.005** 0.001 [0.017] [0.026] [0.950] Net selling frequency 0.074** 0.088** 0.104** [0.021] [0.011] [0.031] Net selling value 0.001 0.002** 0.001 [0.149] [0.040] [0.680] Accelerated SEOs 0.110 [0.420] Leverage -0.069 0.123 -0.425 [0.727] [0.597] [0.310] Share turnover -0.011 -0.040 0.392** [0.844] [0.530] [0.018] Log (proceeds) 0.312*** 0.299*** 0.239* [0.000] [0.000] [0.056] Underwriter rank 0.024 0.035 0.005 [0.440] [0.302] [0.954] Secondary shares (%) 0.009*** 0.009*** 0.009*** [0.000] [0.000] [0.003] ROA 0.384** 0.395** 0.804* [0.021] [0.019] [0.073] Constant -1.930*** -1.937*** -1.486 [0.000] [0.000] [0.230] Year fixed effects Yes Yes Yes Industry fixed effects Yes Yes Yes Observations 1,485 1,226 259 Pseudo R2 0 .101 0.114 0.132

Internet Appendix for “Lockup agreements in seasoned equity offerings: Evidence of optimal contracting,” page 28

Table A12. Determinants of Ex Post Lockup Period

This table presents the estimates from nine specifications of a tobit model using data from SEOs from 1996 through 2006. The dependent variable equals the ex-post lockup period in calendar days. The sample criteria require the SEOs to be U.S. common stocks listed on NYSE, NASDAQ, or AMEX, and exclude rights issues, spin-offs, ADRs, reverse LBOs, the SEOs with offer prices less than $5, closed-end fund, unit investment trusts, REITs, limited partnerships, and unit offerings. In addition, firms are required to have available CRSP daily returns data and COMPUSTAT quarterly financial data. Variable definitions are in the Appendix. All regressions include a constant plus year and industry fixed effects. All tests use White heteroskedasticity robust standard errors with adjustment for SEO clustering by issuers. p-values are in brackets, and the symbols ***, **, and * represent significance at the 1%, 5%, and 10% levels.

VARIABLES (1) (2) CR(0,5) -27.733** -18.490** [0.029] [0.031] Information asymmetry -11.046 -6.372 [0.132] [0.163] Ex ante lockup days 0.527*** 0.085*** [0.000] [0.000] Insider ownership % -0.056** -0.010** [0.046] [0.013] Net selling frequency -0.955 0.339 [0.322] [0.568] Net selling value -0.014** -0.012** [0.022] [0.039] Accelerated SEOs -2.916 -0.541 [0.498] [0.856] Leverage -0.232 -3.248 [0.972] [0.457] Share turnover 0.719 -0.290 [0.693] [0.809] Log (proceeds) -7.181*** -1.966** [0.001] [0.018] Underwriter rank -1.783 -0.352 [0.179] [0.599] Secondary shares (%) -0.308*** -0.142*** [0.000] [0.000] ROA -12.639*** -2.023 [0.009] [0.533] Constant 52.354*** 35.688*** [0.000] [0.000] Year fixed effects Yes Yes Industry fixed effects Yes Yes Observations 1,597 579 Pseudo R2 0.377 0.195