seasoned equity offerings and agency problems: evidence ... · seo is associated with a number of...

1

Seasoned Equity Offerings and Agency Problems: Evidence from a

Quasi-Natural Experiment in China

E. Han Kim, Heuijung Kim, Yuan Li, Yao Lu, and Xinzheng Shi†

Abstract

We find that agency problems become more severe following seasoned equity offerings.

We examine publicly-listed Chinese firms over the period 2000 to 2012, which contains

exogenous regulatory shocks on the eligibility of SEOs. The data reveal that during the

year of SEO and the following year, overinvestments increase, acquisitions yield lower

shareholder returns, director and officer compensation increases with lower sensitivity to

performance, and tunneling increases. Cross-sectional differences in SEO announcement

returns suggest that investors partially anticipate some of the post-SEO changes. The

agency costs stemming from SEOs are negatively related to ownership concentration and

growth opportunities, but are unrelated to higher percentage of independent directors on

the board or to closer monitoring by regulators.

This Draft: April 19, 2015

Keywords: Equity Issuance, Agency Problems, Corporate Investment, Managerial

Compensation, Tunneling.

JEL Classifications: G32 G34

†E. Han Kim is Everett E. Berg Professor of Finance at the University of Michigan, Ross School of

Business, Ann Arbor, Michigan 48109: [email protected]. Heuijung Kim is a doctoral candidate at

Sungkyunkwan University, SKK Business School, Seoul, Korea: [email protected]. Yuan Li is a graduate

student at University of South California, US: [email protected]. Yao Lu is Associate

Professor of Finance at Tsinghua University School of Economics and Management, Beijing, China:

[email protected]. Xinzheng Shi is Associate Professor of Economics at Tsinghua University

School of Economics and Management, Beijing, China: [email protected]. We are grateful for

helpful comments and suggestions by Hongbin Li, Chen Lin and Gordon Philips and seminar participants

at Tsinghua University, Sungkyunkwan University, and participants at 2014 China Finance Review

International Conference, 2014 China Finance and Accounting Conference, the 2014 Conference on Asia-

Pacific Financial Markets, 2014 World Banking and Finance Symposium, Singapore, the 2nd IFMA

International Conference on Finance, 2014 Corporate Governance Conference at Renmin University, and

2014 Finance Conference at University of International Business and Economics. This project received

generous financial support from Mitsui Life Financial Research Center at the University of Michigan. Yao

Lu acknowledges support from Project 71202020 of National Natural Science Foundation of China.

mailto:[email protected]

mailto:[email protected]

2

I. INTRODUCTION

Seasoned equity offerings are an important source of external financing, bringing in a

large amount of potential free cash flows. Relying on Jensen‘s (1986) hypothesis that free

cash exacerbates agency problems, Jung, Kim, and Stulz (1996) argue that investors‘

concern with unproductive use of SEO proceeds is an important reason for the well-

documented negative stock market reaction to the announcement of SEOs. They provide

evidence of less negative market reaction to SEO announcements by firms with higher

market to book ratios, arguing high growth firms are less likely to waste newly raised

funds. Kim and Purnanandam (2014) go a step farther: They argue that misuse of SEO

proceeds is due to weak governance, providing evidence that the previously documented

negative investor reactions to the announcement of primary SEOs are limited to firms

with weak governance.1

Although this link between SEO announcement returns and agency problems is

informative, there is little direct evidence on how firms‘ real activities and agency costs

are jointly affected by SEOs, leaving several questions unanswered. Are SEO proceeds

indeed used less productively? If so, what are the specific channels through which

shareholder value gets damaged? What can be done to reduce the damages?

We investigate these issues by examining how SEOs affect corporate investments,

managerial compensation, and tunneling. We also relate the post-SEO changes in these

variables to stock market reaction at the time of SEO announcement. We focus on

1 Primary offerings are distinct from secondary offerings. Proceeds of shares sold through primary offerings

go to the firm, making them susceptible to misuse by the management. Secondary offerings, by contrast,

are sales of shares owned by corporate insiders and block-holders, so the proceeds do not go to the firm.

Kim and Purnanandam (2014) show that investors react negatively to the announcement of secondary

offerings because of the negative signal transmitted by better informed investors (Leland and Pyle, 1977).

They also show that the market does not react negatively to the announcement of primary offerings unless

the issuer has weak governance. Their evidence is based on difference-in-differences in the market reaction

to an external shock weakening corporate governance.

3

investments, managerial compensation, and tunneling because they are discretionary and

susceptible to managerial self-serving behavior.

Our investigation is based on Chinese SEOs. The main motivation for studying

the Chinese case is endogeneity issues in the choice of SEOs. The decision to issue an

SEO is associated with a number of firm level factors such as internal funds, debt

issuance, the market-to-book ratio, stock returns, and firm age and size (Alti and

Sulaeman, 2012; Baker and Wurgler, 2002; Jung et al., 1996; DeAngelo et al., 2010;

Hovakimian, Opler, and Titman, 2001), as well as other unaccounted time varying factors

affecting corporate activities and performance. These factors cannot be controlled by firm

fixed effects. China's Securities Regulatory Commission (CSRC) enacted two regulations

that became effective in 2006 and 2008, each imposing greater restrictions and higher

standards on the eligibility to issue SEOs. These regulatory changes provide exogenous

shocks that can be used to construct instruments to study causal effects.

In addition, China is the world‘s second largest economy attracting much

attention from practitioners and scholars in recent years. SEOs in China have grown over

time, making them one of the main sources of external financing. Chinese firms rely

more heavily on SEOs relative to US firms. Over the period 2010 through 2012, for

example, the ratio of capital raised through SEOs by non-financial Chinese firms to their

market capitalization was about 2.18%; the same ratio for US counterparts was about

0.6%. (Source: http://data.worldbank.org and SDC)

China has relatively weak corporate governance and legal system (e.g., Allen,

Qian and Qian, 2003, Aharony, Lee and Wong, 2000), providing more latitude for

managers and controlling shareholders to derive private benefits of control. Hence,

agency problems associated with SEOs, if any, would be more noticeable in China. The

http://data.worldbank.org/

4

type of SEOs available and the underwriting practices in China follow the international

standard, which allows generalization of findings based on Chinese SEOs.

Finally, in contrast to SEOs in the US, secondary offerings—sales of shares held

by insiders and block holders—are virtually non-existent in China.2

Proceeds of

secondary offerings do not go to the firm; hence, the proceeds are not subject to

managerial discretion or self-serving. But they transmit negative signals from better

informed insiders and block holders, causing negative investor reaction (Leland and Pyle,

1977). Hence, studying Chinese SEOs helps avoid confounding effects associated with

secondary offerings that are unrelated to agency costs.

We define the year of an SEO and the year after as SEO years. We observe more

over-investments during SEO years relative to non-SEO years. Corporate acquisitions

also yield substantially lower shareholder value. In addition, D&O (director and officer)

compensation increases with lower pay for performance sensitivity.

An illegal and yet pervasive form of private benefits in emerging Asian

economies is tunneling by controlling shareholders and managers (Johnson, La Porta, and

Lopez-de-Silanes, 2000; Bertrand, Mehta, and Mullainathan, 2002; Lemmon and Lin,

2003). Obviously those engaged in tunneling have every incentive to hide it; hence, it is

difficult to accurately quantify the magnitude. However, money has to change hands in

tunneling SEO proceeds, and for financial reporting, the funds siphoned off have to be

hidden in the balance sheet under asset accounts that appear less culpable. Our private,

2 There were three mixed offerings containing secondary offerings of state-owned shares during June 2001

and October 2001 when China Securities Regulatory Commission (CSRC) required that if a firm plans to

issue N new shares through an underwritten offering and the firm has state-owned shares (which were non-

tradable at the time), then the offering must contain 10% of N state-owned shares. This means the firm will

issue 1.1N shares in total, with 0.1N shares being state-owned shares. Such secondary offerings of state-

owned shares are unlikely to transmit the type of negative signals associated with secondary offerings in

the US. The regulation was effective for only four months, and only three SEO cases were completed

during that time.

5

confidential conversations with top Chinese executives and other informed sources

suggest that the siphoned funds usually end up with debit to account receivables or

prepaid expenses. If they are debited to accounts receivable, our sources say, the accounts

receivable soon become uncollectable. Thus, we employ three proxies for tunneling:

account receivables over total assets, prepaid expenses over total assets, and the fraction

of account receivables unlikely to be collected. The first two proxies represent levels of

accounts that could contain tunneling transactions; the third measures suspicious account

receivables. We find that all three proxies significantly increase during SEO years, with

surprisingly large increases in prepaid expenses and uncollectable accounts receivable.

These unproductive uses of SEO proceeds seem to be partially anticipated by

investors at the time of SEO announcements. The three-day returns surrounding the

announcement of SEOs in our sample averages -0.73%. Though statistically insignificant,

the magnitude of market reaction to SEO announcements is more negative when post-

SEO investments are less productive, compensation is more favorable to D&Os, and

tunneling is greater.

Do the negative impacts of SEOs vary across governance and firm characteristics?

We observe firms with high ownership concentration are associated with lower post-SEO

over-investment and managerial compensation, and less tunneling. However, there is no

indication that monitoring by the board—as measured by the fraction of independent

directors—and by regulators helps mitigate agency costs associated with SEOs. In

addition, firms with higher growth opportunities (high P/B ratio, young and small firms)

tend to make better use of SEO proceeds, consistent with the earlier findings by Jung et al.

(1996).

6

This paper contributes to the literature in several ways. Much of the SEO

literature focuses on information asymmetry, adverse selection, and market timing to

explain negative investor reaction to SEO announcements (e.g., Leland and Pyle, 1977;

Myers and Majluf, 1984; Choe, Masulis and Nanda, 1993). However, Jensen‘s (1986)

free cash flow argument suggests that SEO proceeds are susceptible to unproductive use

due to agency problems, leading to a number of studies yielding insights into the use of

SEO proceeds (e.g., Walker and Yost, 2008; Autore, Bray, and Peterson, 2009;

DeAngelo et al., 2010; McLean, 2011). We add to this literature by providing direct

evidence that SEOs are followed by significant reduction in the efficiency of investments

and managerial compensation, and by greater tunneling.

We also identify that managerial compensation and tunneling can both be

important channels through which managers and controlling shareholders can help

themselves with the SEO proceeds. Although the tunneling may not be generalizable to

economies with stronger governance systems, these negative behaviors have received

little attention in the SEO literature.

Finally, Chinese firms‘ reliance on SEOs as a source of external financing has

been rising sharply in recent years. How SEOs affect corporate investments and D&O

compensation in the second largest economy in the world with a rapid growth of financial

markets should be of interest on its own right. More generally, our findings raise

important issues about external financing in emerging markets, highlighting the need for

effective governance mechanisms that can help ensure productive use of externally raised

equity capital.

The next section provides general background and SEO regulations in China.

Section III describes data and empirical strategy. Section IV, V and VI estimates changes

7

in the productivity of investments, D&O compensation and the sensitivity to performance,

and tunneling following SEOs. Section VII relates SEO announcement returns and the

post-SEO changes. Section VIII analyzes cross-sectional differences across governance

and firm characteristics. Section IX concludes.

II. SEASONED EQUITY OFFERINGS IN CHINA

II.1 General Background

The Chinese financial market has several attractive features for studying SEOs.

China has a large SEO market relative to the size of its securities markets. Over the

period 2010 through 2012, the total capital raised through SEOs to the average market

capitalization for nonfinancial firms is 2.18% in China, whereas it is about 0.6% in US.3

Since China opened the Shanghai Stock Exchange (SHSE) and the Shenzhen Stock

Exchange (SZSE) in 1990 and 1991, equity markets have become an important source of

external financing, playing a much more important role than bond markets.4 Corporate

bond markets have been developing at a much slower pace than stock markets.5

Perhaps the most attractive aspect of studying Chinese SEOs is the unique

exogenous regulatory regime changes on the eligibility to issue SEOs, allowing us to

construct instruments to study the causal effects of SEOs. The next section describes the

regulatory changes.

3 Over the period 2010 through 2012, the average market capitalization of Chinese stock market is 3949.77

billion USD and Chinese listed firms raised 86.09 billion USD through SEOs. During the same period, the

average market capitalization of US stock market is 17149.34 billion USD and US listed firms raised

102.75 billion USD through SEOs. (These numbers are based on SEO activities by non-financial Chinese

and US firms. Stock market cap excludes financial firms. Capital raised through SEOs are taken from SDC

Platinum. The market cap data are taken from data in the World Bank website (http://data.worldbank.org/).

Capital raised through SEOs includes only proceeds from primary offerings.) 4 Over the period 2010 through 2012, Chinese listed firms raised 429.5 billion RMB through bond markets,

while they raised 2,147.5 billion RMB through equity markets (including IPOs). 5 A regulated bond market for enterprises began in 1996; however, the strict approval process required for

issuing bonds has led to a situation where only very large and stable companies can issue bonds.

8

Another attractive feature of Chinese SEOs is that virtually all of them are

primary shares. This is in sharp contrast to US underwritten offerings, which often

include secondary offerings, sale of shares held by insiders and block holders. Proceeds

of secondary offerings do not go to the firm; hence, they cannot be misused by the

management. Instead, secondary offerings transmit negative signals from better informed

insiders and block holders (Leland and Pyle, 1977). Because Chinese SEOs rarely contain

secondary offerings,6 they helps us avoid their confounding effects.

In addition, Chinese corporate governance system has been considerably weaker

than the global standard (e.g., Allen, Qian and Qian, 2003, Aharony, Lee and Wong,

2000). To the extent that agency problems affect how productively SEO proceeds are

used, the effects might be more noticeable in Chinese data.

The type of SEOs available and the underwriting practices in China follow the

international standard. There are three types: (1) rights offerings in which current

shareholders are given rights to purchase new shares at a discount such that a current

shareholder is given the opportunity to maintain a proportionate share in the company

before the shares are offered to the public; (2) underwritten offerings in which new shares

can be purchased by any investors; and (3) private placement in which new shares can be

purchased by no more than ten qualified and specific investors. Our analyses exclude

private offerings and focus only on rights and underwritten offerings, because the

external regulatory shocks used to construct instruments apply only to public offerings.

6 There were three mixed offerings containing secondary offerings of state-owned shares during June 2001

and October 2001 when China Securities Regulatory Commission (CSRC) required that if a firm plans to

issue N new shares through an underwritten offering and the firm has state-owned shares (which were non-

tradable at the time), then the offering must contain 10% of N state-owned shares. This means the firm will

issue 1.1N shares in total, with 0.1N shares being state-owned shares. Such secondary offerings of state-

owned shares are unlikely to transmit the type of negative signals associated with secondary offerings in

the US. The regulation was effective for only four months, and only three SEO cases were completed

during that time.

9

Chinese regulators require that firms hire an underwriter to issue new shares for rights

and underwritten offerings. As in the US, two types of underwriting contracts, best efforts

and firm commitment, are practiced in China. These similarities allow generalization of

findings based on Chinese SEOs.

II.2 Regulatory Changes on Chinese SEOs

Chinese regulatory changes on the eligibility for SEOs provide exogenous shocks,

which allow us to construct instruments to address endogeneity issues. Prior to 2006, a

listed firm could issue equities as long as it issued a dividend in the past three years. On

May 6, 2006, China's Securities Regulatory Commission (CSRC, equivalent to the US

SEC) issued Decree No.30, Measures for the Administration of the Issuance of Securities

by Listed Companies. The 2006 regulation requires that if a firm wants to conduct a

public SEO, the cumulative distributed profits of the firm in cash or stocks in the

immediate past three years shall not be less than 20% of the average annual distributable

profits realized over the same period.

CSRC strengthened the requirement further on October 9, 2008, when it issued

Decree 57, Notice on Amendment in Regulations for Listed Companies' Cash Dividend.

The 2008 regulation increases the dividend requirement; the cumulative distributed profit

in cash in the past three years shall not be less than 30% of the average annual

distributable profits realized in the same period. The 2008 regulation not only raises the

required dividend level, but also counts only cash dividends toward the 30% requirement.

III. DATA

III.1 Sample Description

10

Our sample is constructed with all A-share firms7 listed on the Shanghai Stock

Exchange and Shenzhen Stock Exchange. The sample includes listed firms from all three

boards (i.e., the main board, the small and medium enterprise board, and the growth

enterprises board).8 Our data are taken from several sources. Financial accounting data,

corporate governance data, and director and executive compensation data are taken from

Resset.9 SEO data are taken from CSMAR.

10 The dividend ratio required by the 2006 and

2008 regulations (the cumulative distributed profits in the past three years over the

average annual distributable profits realized over the same period) are taken from Wind

Information. 11

Financial firms as defined by the CSRC (e.g., banks, insurance firms, and

brokerage firms) are excluded. We also exclude ST (special treatment) and *ST

companies. Firms are classified as such if they have two (ST) and three (*ST)

consecutive years of negative net profit. Because these companies are not allowed to

issue SEOs, they are unaffected by the 2006 and 2008 regulations.

These selection criteria lead to 18,459 firm-year observations associated with

2,290 unique firms over the period 2000-2012. The sample period starts in 2000 because

underwritten offerings were first allowed in 2000. Board information also is available

7 In mainland China there are two types of stocks: A-share and B-share. Originally, the A-share market

was designed for domestic investors to trade with RMB, and the B-share market was designed for foreign

investors to trade with US dollars. The B-share market was opened up to domestic investors in 2001, and

qualified foreign institutional investors (QFII) were also allowed to trade in the A-share market beginning

in 2006. A firm can issue both A-shares and B-shares, and these shares have identical rights. We restrict

our sample to the A-share market because there are currently 106 firms listed in the B-share market, and 84

of them are also listed in the A-share market. The total market capitalization of the A-share market is about

122 times that of the B-share market as of the end of 2013. 8 The Shenzhen Stock Exchange has three boards: the main board, established in 1991; the small and

medium enterprise board, established in 2004; and the growth enterprises board, established in 2009. The

Shanghai Stock Exchange has only the main board. 9

Resset is a financial data provider in China, equivalent to Compustat in the US. Website:

http://www.resset.cn/en/ 10

CSMAR is another financial data provider in China. Its database for seasoned equity offering is more

detailed than to Resset‘s. Website: http://www.gtadata.com/ 11

Website: http://www.wind.com.cn/En/Default.aspx

11

only after 2000. For compensation analyses, the sample period starts in 2001 because

compensation data are not available until 2001. All accounting variables are winsorized

at the 1% level. All monetary variables are normalized to 2000.

Table I lists the sample distribution by year. Column (1) reports the number of

firms in the full sample by year. The table shows the number of public SEOs by two dates,

the announcement date and the offering date. The announcement date is when the

decision to issue an SEO is announced; the offering date is when the firm receives the

SEO proceeds. Because our analyses are about how firm behavior changes after SEOs,

we use the offering date to define SEO years—the year of SEO and the following year.

We focus on these two years because the impact of the newly-raised capital on the firm‘s

investments, compensation and tunneling, if any, should be most noticeable during those

years.

In total, 481 SEOs are announced, and 557 SEOs are made during 2000 to 2012.

The difference between the number of announcements and offerings is due to 76 SEOs

announced in 1999. The table shows a steady decline in the number of SEOs until 2007,

when a big jump in the number of announcements occurred. The very small number of

announcements in 2005 is due to the Split Share Structure Reform initiated in April, 2005,

when the CSRC stopped approving any IPO or SEO proposals until May, 2006.12

The

12

Prior to the Split Share Structure Reform, approximately two-thirds of domestically listed A-shares were

not tradable (Li, Wang, Cheung and Jiang, 2011), yet these non-tradable shares enjoyed the same rights as

tradable shares. Split Share Structure Reform was designed to convert these non-tradable shares into

tradable shares. The reform was initiated in April, 2005, and CSRC stopped approving SEO and IPO

proposals until the reform was completed. To account for the impact of Split Share Structure Reform, we

include the percentage of non-tradable shares as a control variable in all regressions.

12

sharp increase in the announcement of SEOs in 2007 reflects the release of suppressed

SEOs during 2005 and 2006. Chinese stock market also reached its peak in 2007.13

III.2.Definition of Key Variables

III. 2.1. The SEO Variable

The key explanatory variable is an SEO indicator, SEO, equal to one in SEO years

including the year of SEO and the year after. The coefficient on SEO reflects the two-

year average effect of an SEO. Using the two-year average reduces noise arising from

uneven timing of SEOs within a year (e.g., SEOs are in February vs. November). The

results are qualitatively similar when we define SEO equal to one only in the year after an

SEO, as in Kim and Weisbach (2008).

III. 2.2 Instrumental Variables

We use the 2006 and 2008 SEO regulations to construct instrumental variables.

The past dividend payout ratio requirements in these regulations alter the eligibility to

conduct SEOs for firms that did not pay sufficient dividends, while leaving those that

paid sufficient dividends unaffected. 14

The validity of instruments requires two conditions. The relevancy condition

requires that the instrument must be correlated to the endogenous variable (SEO). This

condition will not be satisfied if low dividend-paying firms can circumvent the

regulations without costs. For example, some firms may anticipate the regulatory changes,

increase cash dividends prior to the regulation, and gross up the size of SEO to make up

13

Shanghai Stock Exchange Composite Index reached its peak of 6124.04 on October 16, 2007 and has

declined since then. The index was 2115.98 on December 31, 2013.

14 One may consider the regression discontinuity design as an alternative identification strategy; i.e.,

comparing firms having dividend ratios just above 20% (30%) before 2006 (2008) with those having

dividend ratios just below 20% (30%) to identify the potential effects of SEOs. However, this approach is

problematic because firms choose dividends, which undermines its validity.

13

for the cash used to pay the additional dividends prior to the SEO. However, such

maneuvers impose several types of costs. For one, firms wishing to issue SEOs tend to be

cash constrained. Paying out extra cash dividends may lead to foregoing value-enhancing

investments. If extra cash dividends require more borrowing, it may lead to a higher than

optimal level of financial leverage. Perhaps more important, anticipation is subject to

uncertainty, making the benefits from dividend maneuvers subject to uncertainty, which

reduces the present value of the benefits. The uncertainty is not just about the future

regulations. There is also approval uncertainty. SEOs in China and the amount that can be

raised require the CSRC‘s approval, which adds further uncertainty over whether and

how much capital can be raised through an SEO.

The 2006 regulation counts stock dividends towards meeting the dividend

requirement. If low dividend-paying firms anticipated this aspect of forthcoming

regulation, they could have satisfied the dividend requirement by issuing sufficient stock

dividends during 2003 - 2005. Data show otherwise. Stock dividend is not popular in

China. Among 600 dividend cases in 2005, only 41 included stock dividends. And of all

the dividend cases over the period 2003-2005, 94% did not issue any stock dividends.

Furthermore, the 2008 regulation excludes stock dividends in defining the dividend

requirement. Considering all these, it seems safe to assume that the relevancy condition is

satisfied.

The second condition is the exclusion restriction; the instrument should not be

correlated with the error term of the second-stage regression. In other words, the

instrument should not be correlated with the dependent variable after controlling for

relevant variables. One source of concern is that higher dividends may reduce free cash

flows, discouraging firms from misusing their capital (Jensen, 1986). However, the

14

regulation is about past three years‘ dividend payout ratios, not current or future dividend

payout ratios. Even if some firms temporarily increase dividends prior to the regulation to

meet the regulatory requirement, such maneuver is unlikely to reduce free cash flows

after the SEO, because such a firm will gross up the size of SEO by the amount of

additional dividends paid prior to the regulation. Thus, the regulation is unlikely to

directly affect how the proceeds of SEOs are used.

Another concern is that dividend payout ratios may be serially correlated due to

financial policy persistency (Lemmon, Roberts, Zender, 2008) and current dividend

payout ratio is likely to be correlated to current investment or other corporate policies. To

control for persistency in corporate policies, we control for firm fixed effects. We also

estimate the correlation between past dividend payout ratios and current investment,

executive compensation, or tunneling variables and find insignificant correlations.

[Report this in online appendix.] Nevertheless, we re-estimate all regressions with the

dividend payout ratio in each year as an additional control and find the results are robust.

The instruments may be indirectly related to corporate behavior through its

relation with the strength of corporate governance. For example, better governed firms

tend to suffer less from misuse of SEO proceeds (Kim and Purnanandam, 2014). One

might argue that firms with better governance have higher dividend payout ratios and

hence are less likely to be affected by the regulation. For this reason, we control for a

number of proxies for the strength of corporate governance in all regressions.

Our construction of instrument variables involves three steps. First, we define

dummy variables, AFFECT_06 (AFFECT_08), based on the 2006 (2008) regulation.

AFFECT_06 (08) is one if the cumulatively distributed profit over the average annual

distributable profits over the past three years is smaller than 20% (30%); and zero

15

otherwise. Using these dummy variables, we define an indicator for firms affected by the

regulation, AFFECT_REG. It is equal to AFFECT_06 through 2009, and AFFECT_08

from 2010. This is because the impact of the 2008 regulation on the use of SEO proceeds,

if any, will be noticeable mostly from 2010. This two-year lag reflects the elapsed time

from SEO approvals to the use of the proceeds. Our sample shows that, on average, an

SEO process takes about 242 days from the initial announcement to the receipt of the

proceeds. Because our SEO years include the year of SEO and the year after, an SEO

equal to one in year 2009 (2007) could be an SEO issued in 2008 (2006) that was

approved in 2007 (2005).

Our instrumental variable, IV_SEO, is AFFECT_REG x POST_REG. The post

regulation indicator, POST_REG, is equal to one when the year of observation is 2008 or

2010 to ensure that IV_SEO is equal to one only when the SEO is affected by the

regulation in 2006 and 2008, respectively. Again, the two-year lag is to capture the

elapsed time from an SEO approval to the usage of SEO proceeds.

III.3. Summary Statistics

Table II provides summary statistics for all key variables. Variable definitions are

provided in Appendix I. Panel A shows the statistics for the full sample. The mean of

AFFECT_06 and AFFECT_08 are 0.35 and 0.37, indicating 35% of firms are affected by

the 2006 regulation, while 37% are affected by the 2008 regulation. Panel B compares

the mean of each variable for the SEO and non-SEO samples. The SEO sample shows

higher levels of capital expenditures and over-investment, and higher frequency of

acquisitions. In addition, SEO firms tend to have higher dividend payout ratios, leverage,

tangible asset ratios, ROE, non-tradable shares, and state-owned shares. SEO firms also

16

tend to have been listed for shorter periods and to have a lower percentage of independent

directors on their boards.

IV. CORPORATE INVESTMENT

In this section, we document that firms invest more, i.e., more capital

expenditures and acquisitions, following SEOs. Then we investigate the degree of

overinvestment in capital expenditures and shareholder value impact of acquisitions.

IV.1 Capital Expenditure and Overinvestment

To examine how capital expenditures changes following SEOs, we regress

ln(CAPEX), log of capital expenditures to SEO, the indicator for SEO years (the year of

SEO and the year after). CAPEX is defined as cash paid to acquire fixed assets, intangible

assets, and other long-term assets. The regression controls for firm- and year fixed effects.

Control variables include firm age, as measured by log of the number of years a firm has

been listed, ln(Listing_Years); non-linear firm size effects with sales (SALES) and its

square term (SALES2); return on equity, ROE; Leverage, the sum of short- and long-term

debt over total assets; PPE/TA, property, plants, and equipment over total assets;

SALES_GR, sales growth rate. We also control for governance characteristics and factors

unique to Chinese financial markets (e.g., Li et al., 2011): %_IND_DIR, the percentage of

independent directors on the board; %_STATE_OWN, the percentage of shares held by

the government; %_LARGEST_HOLD, the percentage of shares held by the largest

shareholder; and NONTRDPCT, the percentage of non-tradable shares, to control for

potential impacts of the Split Share Structure Reform in China. Standard errors of the

second stage IV regressions are corrected by bootstrapping, and standard errors of the

OLS results are clustered at the firm level.

17

Table III reports the second-stage IV regression result in the first column. All first-

stage estimation results are presented in Appendix II. The second-stage result shows

significantly positive coefficient on SEO. The point estimate suggests capital expenditure

increases, on average, by 79% during SEO years.

Are these increases in capital expenditures following SEO years ‗normal‖

increases reflecting changes in firm, industry, and year characteristics, or do some of the

increases represent what Jensen‘s (1986) free cash flow hypothesis predicts: The infusion

of free cash flows from SEOs leads to more overinvestments. To investigate this issue,

we follow Richardson (2006) and estimate the following model.

Invi,t = γ0 + γ1Tobin‘s Qi,t-1 + γ2Leveragei,t-1 + γ3Cashi,t-1 + γ4Firm_Agei,t-1

+ γ5Ln(TA)i,t-1 + γ6YRRETi,t-1 + γ7Invi,t-1 + at + aj + εi,t (1)

Invi,t is net investments firm i makes in year t, defined as the ratio of (CAPEX – cash

received from disposals of fixed assets, intangible assets, and other assets) to total assets

at the beginning of the year. Following Richardson (2006), we predict the normal

investment level using the previous year‘s Tobin’s Q, Leverage (the sum of short- and

long-term debt over total assets), Cash (cash and cash equivalent over total assets),

Firm_Age, (firm age), Ln(TA) (log of total assets), YRRET (stock return), Invi,t-1 (the one-

year lagged net investments), and year- and industry fixed effects. For industry

classification, we use CSRC‘s definition, which contains 12 industry sectors.

The residuals of Model (1) are measures of abnormal investments. We measure

overinvestments by AB_INV+, which equals the residual, or zero if negative. Then, we

estimate how SEOs affect AB_INV+t. The regression contains the same variables as

before and controls for firm- and year fixed effects. The estimation result shows a

significant increase in overinvestments during SEO years.

18

IV.2 Acquisitions and Shareholder Value

Because corporate investments are also made externally via acquisitions, we also

estimate how the likelihood of acquisitions changes when SEO proceeds become

available. The dependent variable is an indicator for whether an acquisition is made by

firm i in year t. Same control variables are included. Because the dependent variable is an

indicator variable, the regression is estimated by the firm level conditional logistic

regression. Year dummies are also included.

Column (2) reports the second-stage estimation result, which shows a

significantly positive coefficient on SEO. The point estimate suggests the likelihood of

making an acquisition increases by 9.3% after an SEO.

How do these increased acquisitions affect shareholder value? One way to

answer this question is to measure investor reactions to the announcement of acquisitions.

We measure the market reaction by abnormal stock returns over a three-event day

window (-1, 1) surrounding the announcement date (day 0). Abnormal returns are

estimated using the market model with the A-share value-weighted index over -270 to -2

event days. Following Moeller, Schlingemann, and Stulz (2004), we consider only

completed acquisitions and exclude acquisitions with a transaction value below 1 million

RMB (the 5th

percentile in the acquisition sample). When a firm has more than one

acquisition announcements in a year, we use the deal with the largest transaction value.

We regress ACQ_CAR, the acquirer‘s cumulative abnormal returns surrounding the

announcement date on SEO. The control variables are the same as before.

The last column of Table III reports the estimation result. The coefficient on SEO

is significantly negative. The point estimate suggests acquisition announcement returns

are, on average, 10.1% lower during SEO years relative to other years. Acquisitions made

19

during SEO years are substantially less value-enhancing, or more value-destroying,

relative to those made during non-SEO years. As a robustness check, we use an

alternative event window (-2, +2) to measure acquisition announcement returns. The

conclusions do not change.

In sum, firms increase capital expenditures and make more acquisitions

following SEOs. Much of these increases seem to be counter-productive. Significant

portions of the capital expenditure increases represent overinvestments, and acquisitions

made during SEO years are substantially less value-enhancing, or more value-destroying,

relative to those made during non-SEO years.

V. DIRECTOR AND EXECUTIVE COMPENSATION

How does the infusion of funds from SEOs affect compensation of those in

control, directors and officers? Jensen‘s (1986) free cash flow hypothesis suggests they

will become more generous to themselves. To investigate this possibility, we estimate

changes in the level of D&O compensation and changes in the sensitivity of the

compensation to performance.

V.1. D&O Compensation

Chinese regulators have been pushing public firms to provide more detailed

disclosure of managerial compensation. Starting 2001, the CSRC requires publicly listed

firms to disclose salaries and bonuses of directors and senior managers. In contrast to the

US, where stock grants and options constitute an important component of managerial

compensation, most compensation in China takes the form of cash payment. For example,

our database (Wind database) shows that only 1.6% (31 firms) of exchange-listed

companies in 2010 granted stocks or stock options. Thus, our analysis focuses on cash

compensation. The total D&O compensation, TOTYRPAY, is the sum of annual cash

20

salaries and bonuses to all directors and senior managers. This variable is obtained from

Resset database. The dependent variable is the logged value of TOTYRPAY. The control

variables are same as before, except that we add PAY_SIZE, the number of directors and

officers included in computing TOTYRPAY.

Table IV contains the estimation result in the first column, which shows a

significantly positive coefficient on SEO. The point estimate suggests an average increase

of 12% in D&O compensation during SEO years. Many control variables show

coefficients consistent with our intuition: total compensation increase with firm size at a

decreasing rate, ROE, and the number of directors and officers included in the calculation

of the total pay.

V.2. Pay-for-Performance Sensitivity

Does the higher D&O compensation represent rewards for better performance? To

answer this question, we estimate the sensitivity of compensation changes to changes in

profitability. Since SEOs change firms‘ equity base, total assets and market capitalization,

we use operating profit margin, EBITDA to sales ratio, as the performance measure.

When cash bonuses are based on profitability, in China the profitability is mostly

measured by accounting numbers. We estimate the following specification:

ΔLn(TOTYRPAYi,t) = β0 + β1Δ(EBITDA/SALES)i,t + β2SEOi,t

+ β3SEOi,t*Δ (EBITDA/SALES)i,t + β4Xi,t + at + ai + εi,t (2)

ΔLn(TOTYRPAY) is the yearly change in the logged value of total D&O compensation.

The coefficient of interest is β3, the coefficient of the interaction of SEO and Δ

(EBITDA/SALES), the change in (EBITDA/SALES) from year t-1 to year t, where

(EBITDA/SALES) is the ratio of EBITDA to total sales revenue. X is the same control

21

variables as in the regressions on the level of total compensation, except that PAY_SIZE

is replaced by ΔPAY_SIZE.

The estimation result shows a significantly negative coefficient on the interaction

of SEO and Δ(EBITDA/SALES), implying D&O pay becomes less sensitive to

performance during SEO years.

In sum, those in control get paid more following SEOs, with their pay becoming

less sensitive to performance. However, the statistical significance of the decrease in pay-

for-performance sensitivity is relatively weak. This may be due to the fact that

managerial compensation of China‘s publicly listed companies is relatively low to begin

with. Perhaps publicly disclosed compensation, which excludes private benefits, is not

the only channel through which corporate insiders get compensated.

VI. TUNNELING

An unethical, mostly illegal and yet quite prevalent channel through which

controlling shareholders and managers help themselves is tunneling (Johnson, La Porta,

and Lopez-de-Silanes, 2000). It is especially pervasive in emerging Asian economies

(Bertrand, Mehta, and Mullainathan, 2002; Lemmon and Lin, 2003). Jiang, Lee and Yue

(2009) empirically demonstrate the severity of tunneling problem in China. Assets

siphoned off through tunneling can be intellectual properties, business plans, tangible

assets including cash, and so on. The proceeds from SEOs are also subject to tunneling.

In 2006, CSRC and seven other government ministries have put into force a regulation

that threatens to put the top management into jail if tunneling activities are detected.

Since people weigh the costs and benefits in committing illegal activities, tunneling is

unlikely to be eliminated by the regulatory threats.

22

Accurate identification and measurement of tunneling is virtually impossible

because those who engage in tunneling have every incentive to hide it. To tunnel SEO

proceeds, however, money has to change hands, which will leave trace in accounting

numbers. So we have conducted private interviews with practitioners for hints on how

tunneling takes place. A typical tunneling scenario is as follows: Privately-owned entity

A, the controlling entity of a public firm B, has B issue an SEO and tunnel parts of the

proceeds to A. B is unlikely to transfer the money directly to A. To make it difficult to

detect the tunneling, the fund transfer is likely to go through a third entity, say C, which

takes money from B and gives it to A. (To make it more difficult to trace the trail of

money, the transfer may go through more entities than one.) Since Firm B has to cover up

the missing funds with another form of assets in its balance sheet, our sources tell us the

missing funds are likely to be recorded as either accounts receivable or pre-paid expenses.

If recorded as accounts receivable, they say, the accounts receivable will soon be

classified as unlikely to be collected.

Thus, we construct three proxies for tunneling: accounts receivable over total

assets, ACCV/TA, prepaid expenses over total assets, PREPAY/TA, and the percentage of

account receivables classified as unlikely to be collected, %_ACCV_BAD. All three

proxies represent the level of accounts that could contain tunneling. We estimate the

relation of each of these proxies to SEO. Control variables are the same as in Table III.

Table V reports the estimation results. The estimation results show significant

increase in all three proxies for tunneling. During SEO years, the ratio of accounts

receivable to total assets increases by 0.013; the prepaid expense ratio by 0.009; and the

fraction of uncollectible accounts receivable by 0.041. When compared to the sample

medians (0.07, 0.02. and 0.07, respectively), these increases are substantial, especially

23

prepaid expenses and uncollectable accounts receivables. These increases may not be all

due to tunneling. Some may arise from the overinvestments and/or unprofitable

acquisitions mentioned earlier, but both accounts receivable and prepaid expenses are

normalized by total assets. Particularly noteworthy, the fraction of uncollectible accounts

receivable increased by 59% relative to its median, while the fraction of accounts

receivable itself increased by only 19% relative to its median. These results suggest that

in China a significant portion of SEO proceeds are siphoned off through tunneling.

VII. SEO ANNOUNCEMENT RETURNS

Does the market reaction at the time of SEO announcements reflect the

unproductive uses of SEO proceeds? If investors are capable of anticipating how SEO

proceeds will be used, their reaction will be more negative to SEOs followed by an

increase in overinvestment, lower pay-for-performance sensitivity, and an increase in

tunneling. In this section we calculate the average SEO announcement returns in China

during our sample period and examine cross-sectional differences in the market reaction.

SEO announcement returns are calculated as cumulative abnormal returns over

the three-day window (-1, 1) surrounding the announcement date (0), using the market

model with the A-share value-weighted index. The estimation window for the market

model is 270 trading days prior to the event window. The filing date is used as the

announcement date. Table VI reports the results, which show a significant average

announcement return of -0.73%.

To examine cross-sectional differences, we divide the sample according to

changes during SEO years in: (1) the level of overinvestment, AB_INV, (2) D&O

compensation to performance sensitivity, defined as Δlog(TOTYRPAY) over

Δ(EBITDA/SALES), and (3) the fraction of uncollectible accounts

24

receivable, %_ACCV_BAD. We calculate changes in each of these measures during SEO

years by ΔX = (Xt+1 + Xt)/2 –Xt-1, where t is the year of SEO and X refers to the above

three measures. Then, we divide the sample by whether the change is positive or negative

and estimate the announcement returns separately for each subsample for each of the

three categories.

The results are reported in Panel B of Table VI. The sample size is reduced

considerably because data for compensation and uncollectible accounts receivable are

available only from 2001 and 2004, respectively, and measuring changes in variables like

ΔAB_INV involves one period lagged value and two period forward values. The

differences in the magnitude of negative announcement returns, albeit statistically

insignificant, are consistent with the notion that the unproductive usages of SEO proceeds

are partly anticipated by investors at the time of announcement.

Surprisingly, the possibility of anticipation seems stronger for tunneling. The

average announcement return is insignificantly different from zero for SEOs not followed

by an increase in the fraction of bad accounts receivable. In contrast, when there are

subsequent increases in the fraction of bad accounts receivable, the average

announcement return is -2.20% and significant. Although the difference between the

subsamples is statistically insignificant (t = -1.25), the power of test is quite weak due to

a very small sample size. We cannot rule out that investors somehow can tell at the time

of SEO announcements which SEOs have a higher risk of tunneling than others.

Taken together, these results suggest that the negative market reaction to SEOs is

partially attributable to investor expectations of the productivity of the usage of SEO

proceeds. The linkage between the ex-ante investor reaction and the ex-post usage of

25

funds further buttresses our argument that an important reason for the negative SEO

announcement returns is the agency costs associate with free cash flows.

VIII. GOVERNANCE AND FIRM CHARACTERISTICS

Our final inquiry is whether governance and firm characteristics matter in how

productively SEO proceeds are used. We assess the strength of governance by three

governance mechanisms: ownership concentration as measured by the percentage of all

outstanding shares held by the largest shareholder; board independence as measured by

the percentage of independent directors on the board; and monitoring by the regulator.

The CSRC states it will closely monitor the use of the funds if firms state SEO proceeds

are for specific projects when they file for the CSRC approval. But the CSRC‘s

monitoring is relatively lax if firms state that SEO proceeds will be used to supplement

operating capital. Thus, we label the former as ―for project‖ and the latter ―cash holding‖

to indicate strong and weak regulator monitoring. We also examine three firm

characteristics: firm age, growth opportunities as proxied by P/B ratio, and firm size as

measured by the book value of total assets.

This part of analysis is conducted at the SEO deal level. We divide the sample

into two subsamples based on the sample median for each governance mechanism and

firm characteristic. We also form subsamples by ―for project‖ and ―cash holding‖ to

separate the strength of regulatory monitoring. We then plot the mean overinvestment,

total D&O compensation, and fraction of uncollectible accounts receivable for each

subsample for the year of SEO and two years thereafter.

Figure 1 presents 18 graphs comparing the three outcome variables for

subsamples divided by three governance and three firm characteristics. These graphs are

meant to show only suggestive correlations, as the governance and firm characteristics

26

are mostly endogenous. Among the three governance mechanisms, only ownership

concentration seems to help control the agency problem associated with free cash

generated by SEOs. Firms with high ownership concentration are associated with lower

post-SEO over-investment, managerial compensation and uncollectible accounts

receivable. But signs are in the opposite direction for subsamples separated by board

independence and regulator monitoring. We do not infer from these that having more

independent directors in the board or stronger monitoring by the regulator makes matters

worse, because both governance characteristics are largely endogenous. However, it is

clear that having a higher fraction of independent directors or stronger monitoring by the

regulator does not help mitigate misuse of SEO proceeds. As for firm characteristics, our

results are consistent with previous literature that for firms with greater growth

opportunities, i.e., high P/B ratio, young and small firms, are less likely to misuse of SEO

proceeds (Jung et al., 1996)

Taken together, these findings suggest that the misuse of SEO proceeds can be

alleviated by high ownership concentration and growth opportunities, but not by

monitoring from the gatekeepers in the boardroom or in regulatory agencies.

IX. ROBUSTNESS

In this section, we reestimate baseline regressions with alternative definitions of

key variables and with inclusion of past dividend payout ratios as an additional control.

Table VII reports the reestimation results without showing control variables for brevity.

IX.1. Alternative Definitions of SEO

In our baseline regressions, the SEO indicator is turned on for all completed

underwritten offerings and rights offerings. We experiment with several alternative

definitions for the SEO indicator. First, we exclude small SEOs with proceeds in the 10th

27

percentile. These small SEOs are often made by small market cap firms with highly

volatile performance. The estimation results, reported in Panel A, are robust.

Second, following Kim and Weisbach (2008) we set the SEO indicator equal to

one only in the year following the year of SEO. Although this approach avoids noise due

to different timing within the year of SEO (e.g., early vs. late in the year), it

underestimates the total effects by omitting the year of SEO. The re-estimation results are

reported in Panel B. All coefficients on key variables show same signs as before, but

three of the nine coefficients become insignificant.

IX.2. Alternative Instruments

In constructing the instruments we assume a two-year elapsed time from the

beginning of an SEO process to usage of the proceeds, defining the post-regulation year

as 2008 and 2010. Since there are variations in how long this process takes, we re-

estimate the IV regressions by re-defining the post-regulation year as 2007 and 2009. The

re-estimation results, reported in Panel C, are robust.

IX.3. Alternative Dependent Variables

For acquisition announcement returns, we increase the event window from (-1, 1)

to (-2, 2). We use top 3 executives‘ compensation as the alternative measure of

managerial compensation. The re-estimation results, reported in Panel D, are robust.

IX.4. Additional Control Variable

A possible concern with our instruments is their correlation with the current

dividend payout ratio, which in turn may be related to the dependent variables. If so, the

exclusion restriction will be violated. Thus, we re-estimate all regressions with the

current dividend payout ratio as an additional control. The results, reported in Panel E,

are robust.

28

X. CONCLUSION

Using a sample of Chinese publicly-listed firms, we find robust evidence that

during SEO years—the year of SEO and the year after—corporate investment and

compensation policies become less shareholder friendly and more tunneling takes place.

Specifically, investment increases with higher levels of overinvestment in capital

expenditures and lower returns from acquisitions; D&O compensation increases with

lower sensitivity to performance; and the accounts containing possible tunneling cover-up

transactions increase substantially. These results are based on instruments constructed

with exogenous regulatory shocks on the eligibility to issue SEOs.

These results imply that agency costs associated with free cash flows (Jensen,

1986) are a severe problem in SEOs. In so far as shareholders are concerned, SEO

proceeds are often invested unproductively. D&Os seem to indulge in self-serving

behavior with SEO proceeds, as they increase their own compensation with lower

sensitivity to performance during SEO years. Controlling shareholders and managers

appear stealing more from shareholders when SEO proceeds become available.

These findings call for more efficient corporate governance mechanisms to

safeguard shareholders against the unproductive use of SEO proceeds and insiders‘ self-

serving behavior. SEOs are already heavily regulated in China, and we find no evidence

strong monitoring by the regulatory agency helps mitigate the negative effects of SEOs.

Thus, we do not call for more regulation. Regulations often lead to unintended

consequences with worse outcomes. What we are searching for is a market-based

governance mechanism that provides greater transparency to shareholders so they can

better mitigate agency problems arising from free cash flows raised through SEOs.

29

References

Alti, A., and J. Sulaeman. 2012. When Do High Stock Returns Trigger Equity Issues?

Journal of Financial Economics 103:61-87.

Autore, D. M., D. E. Bray, and D. R. Peterson. 2009. Intended Use of Proceeds and the

Long-run Performance of Seasoned Equity Issuers. Journal of Corporate Finance

15:358-367.

Baker, M., and J. Wurgler. 2002. Market Timing and Capital Structure. Journal of

Finance 57:1-32.

Bertrand, Marianne, Paras Mehta, and Sendhil Mullainathan, 2002. Ferreting Out

Tunneling: An Application to Indian Business Groups. Quarterly Journal of

Economics 117:1, 121-148.

Choe, H., R. Masulis, and V. Nanda. 1993. Common Stock Offerings across the Business

Cycle: Theory and Evidence. Journal of Empirical Finance 1:3-31.

DeAngelo, H., L. DeAngelo, and R. M. Stulz, 2010, Seasoned Equity Offerings, Market

Timing, and the Corporate Lifecycle. Journal of Financial Economics 95:275-295.

Eckbo, B. E., and R. W. Masulis. 1992. Adverse Selection and the Rights Offer Paradox.

Journal of Financial Economics 32:293-332.

. 1995. Seasoned Equity Offerings: A survey. In: Jarrow, R.A., Maksimovic, V.,

Ziemba, W.T. (Eds.), Finance: Handbooks in Operations Research and

Management Science Series. Elsevier/North-Holland, Amsterdam, Chapter 31,

1017–1072.

Heinkel, R., and E. Schwartz. 1986. Rights versus Underwritten Offerings: An

Asymmetric Information Approach. Journal of Finance 41:1-18.

Hovakimian, A., T. Opler, and S. Titman. 2001. The Debt-equity Choice. Journal of

Financial Quantitative Analysis 36:1-24.

Jensen, M. C. 1986. Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers.

American Economic Review 76:323-329.

Johnson, S., R. La Porta, A. Shleifer and F. Lopez-de-Silanes. 2000. Tunneling,

American Economic Review Papers and Proceedings 90, 22-27.

Guohua Jiang, Charles M.C. Lee, and Heng Yue, 2010. Tunneling through Inter-

corporate Loans: The China Experience, Journal of Financial Economics 98: 1-20.

30

Jung, K., Y. Kim, and R. M. Stulz. 1996. Timing, Investment Opportunities, Managerial

Discretion, and the Security Issue Decision. Journal of Financial Economics 42:

159-185.

Kim, E. H., and A. Purnanandam. 2013. Seasoned Equity Offerings, Corporate

Governance, and Investments. Review of Finance.

Kim, W., and M. S. Weisbach. 2008. Motivations for Public Equity Offers: An

International Perspective. Journal of Financial Economics 87:281-307.

Korajczyk, R. A., D. J. Lucas, and R. L. McDonald. 1991. The Effect of Information

Releases on the Pricing and Timing of Equity Issues. Review of Financial Studies

44:685-708.

Krasker, W. S. 1986. Stock Price Movements in Response to Stock Issues under

Asymmetric Information. Journal of Finance 41:93-105.

Leland, H. E., and D. H. Pyle. 1977. Informational Asymmetries, Financial Structure, and

Financial Intermediation. Journal of Finance 32:371-387.

Lemmon, Michael and Karl V. Lins. 2003. ―Ownership Structure, Corporate Governance,

and Firm Value: Evidence from the East Asian Financial Crisis‖. Journal of

Finance 58, 1445-1468.

Leone, A. J., J. S. Wu, and J. L. Zimmerman. 2006. Asymmetric Sensitivity of CEO Cash

Compensation to Stock Returns. Journal of Accounting and Economics 42:167-192.

Li, K., T. Wang, Y. L. Cheung, and P. Jiang. 2011. Privatization and Risk Sharing:

Evidence from the Split Share Structure Reform in China. Review of Financial

Studies 24:2499-2525.

Loughran, T., and J. R. Ritter. 1995. The New Issues Puzzle. Journal of Finance 50:23-

51.

Loughran, T., and J. R. Ritter. 1997. The Operating Performance of Firms Conducting

Seasoned Equity Offerings. Journal of Finance 52:1823-1850

Masulis, R. W., C. Wang, and F. Xie. 2007. Corporate Governance and Acquirer Returns.

Journal of Finance 62:1851-1890.

McLean, R. D. 2011. Share Issuance and Cash Savings. Journal of Financial Economics

99:693-715.

Moeller, S. B., F. P. Schlingemann, and R. M. Stulz. 2004. Firm Size and the Gains from

31

Acquisitions. Journal of Financial Economics 73:201-228.

Myers, S. C., and N. S. Majluf. 1984. Corporate Financing and Investment Decisions

when Firms Have Information that Investors Do Not Have. Journal of Financial

Economics 13:187-221.

Richardson, S. 2006. Over-investment of Free Cash Flow. Review of Accounting Studies

11:159-189.

Walker, M. D. and K. Yost. 2008. Seasoned Equity Offerings: What Firms Say, Do, and

How the Market Reacts. Journal of Corporate Finance 14:376-386.

32

TableI: Sample description.

This table reports,by year, the number of firms in our sample and seasoned equity offerings. The sample includes

Chinese firms listed on the Shanghai Stock Exchange and the Shenzhen Stock Exchange from 2000-2012. Financial

firms, ST (special treatment), and *ST firms are excluded. Firms are classified as ST and *ST if they have two (ST)

and three (*ST) consecutive years of negative net profit. Column (1) shows the number of firms in the full sampleby

year. Column (2) shows the number of public offerings (underwritten offerings and rights offerings) by

announcement yeat. Column(3) shows the number of public offerings by offeringyear.

Year Full By announcement year By offering year

(1) (2) (3)

2000 908 185 154

2001 982 80 131

2002 1,046 38 44

2003 1,110 23 38

2004 1,206 8 32

2005 1,218 1 7

2006 1,250 7 7

2007 1,378 64 28

2008 1,454 22 43

2009 1,549 18 18

2010 1,896 20 20

2011 2,172 13 23

2012 2,290 2 12

Total 18,459 481 557

33

Table II: Summary statistics.

This table reports summary statistics of key variables. Panel A reports the statistics for the full sample. Panel B reports

the means of pre- and post-SEO samples. Column (4) reports means for observations of SEO years and the following

year; Column (5), for non-SEO years. Column (6) reports the difference between SEO and non-SEO sample.

Coefficients marked with *, **, and *** are significant at 10%, 5%, and 1%, respectively. Variable definitions are

provided in AppendixI.

Panel A: Full Sample Panel B: SEO vs. Non-SEO sample

Mean Median Std. Dev SEO Non-SEO (4) - (5)

(1) (2) (3) (4) (5) (6)

Variables of interest

AFFECT_06 0.35 0.00 0.48 - - -

AFFECT_08 0.37 0.00 0.48 - - -

POST_REG 0.18 0.00 0.39 - - -

LOG(CAPEX) 4.12 4.24 1.90 4.89 4.07 0.82***

AB_INV+ 0.03 0.00 0.16 0.05 0.03 0.02***

ACQ 0.32 0.00 0.46 0.35 0.31 0.03**

ACQ_CAR(-1,1) 0.00 0.00 0.08 -0.00 0.00 0.00

TOTYRPAY 2.26 1.62 2.18 1.99 2.27 -0.28***

EBITDA/SALES 0.15 0.13 0.21 0.19 0.15 0.04***

ACCV/TA 0.09 0.07 0.10 0.08 0.09 -0.01***

PREPAY/TA 0.04 0.02 0.04 0.04 0.04 0.00**

%_ACCV_BAD 0.13 0.07 0.16 0.09 0.13 -0.04***

SEO_CAR(-1,1) -0.01 -0.01 0.05 - - -

Control variables

Listing_Years 7.13 7.00 5.03 6.21 7.18 -0.97***

FIRM_AGE 11.55 11.00 5.83 9.66 11.67 -2.01***

SALES 2.88 0.87 6.97 3.37 2.85 0.52**

ROE 0.06 0.07 0.15 0.07 0.06 0.01***

LEVERAGE 0.25 0.24 0.18 0.27 0.25 0.02***

SALES_GR 0.25 0.15 0.55 0.22 0.23 -0.01

PPE/TA 0.32 0.29 0.20 0.36 0.31 0.04***

%_IND_DIR 0.30 0.32 0.13 0.22 0.31 -0.09***

%_STATE_OWN 0.19 0.00 0.25 0.26 0.19 0.07***

%_LARGEST_OWN 0.39 0.37 0.16 0.41 0.39 0.02***

NONTRDPCT 0.21 0.00 0.30 0.41 0.20 0.21***

DIVPRT 0.26 0.17 0.31 0.35 0.25 0.10***

34

Table III: Corporate Investments, Overinvestments, and Acquisition Announcement Returns.

This table estimates changes in thelevelof capital expenditures, acquisitions, overinvestments, and acquisition announcement

returnsafter SEOs. The dependent variables are:LOG(CAPEX), the log of capital expenditures; ACQ, a dummy variable equal

to one if a firm makes an acquisition in year t; AB_INV+, a measure of the level of overinvestment; and ACQ_CAR(-1,1), the

three-day cumulative abnormal return surrounding acquisition announcements. Columns (1) – (4) report second-stage

estimation results of IV regressions.The first-stage regression results are reported in Appendix II, panel A. The sample period

covers 2000-2012. All regressions include firm- and year-fixed effects. Bootstrapped standard errors are reported in

parentheses for columns (1) – (3), and robust standard errors are reported for columns (4). Coefficients marked with *, **,

and *** are significant at 10%, 5%, and 1%, respectively. Variable definitions are provided in Appendix I.

Dependent Variable LOG(CAPEX) ACQ AB_INV+ ACQ_CAR(-1,1)

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

SEO 0.584*** 0.661** 0.043** -0.101**

(0.11) (0.33) (0.02) (0.04)

ln(NYEAR_LISTED) -0.356*** 0.393*** -0.050*** 0.017*

(0.04) (0.11) (0.02) (0.01)

SALES 0.182*** 0.007 0.005 -0.001

(0.01) (0.02) (0.00) (0.00)

SALES2 -0.003*** -0.000 -0.000 0.000

(0.00) (0.00) (0.00) (0.00)

ROE 1.228*** 1.096*** 0.053*** -0.009

(0.09) (0.17) (0.01) (0.01)

LEVERAGE 1.115*** 0.329* 0.045* -0.019

(0.10) (0.20) (0.02) (0.01)

PPE/TA 2.663*** -0.262 0.093*** 0.020

(0.12) (0.24) (0.02) (0.01)

SALES_GR 0.097*** 0.210*** 0.003 -0.002

(0.02) (0.04) (0.00) (0.00)

%_IND_DIR 0.178 0.181 -0.014 0.023

(0.11) (0.28) (0.02) (0.02)

%_STATE_OWN 0.087 -0.121 0.002 -0.003

(0.06) (0.13) (0.01) (0.01)

%_LARGEST_HOLD 0.753*** 0.452 -0.016 -0.029

(0.15) (0.28) (0.01) (0.02)

NONTRDPCT -0.008*** 0.004* -0.000 -0.000

(0.00) (0.00) (0.00) (0.00)

CONSTANT 3.177***

0.104*** 0.022

(0.13)

(0.03) (0.02)

Firm FE Y

Y Y

Year FE & Dummies Y Y Y Y

Observations 18,269 16,636 16,626 5,264

Adjusted R-squared 0.696

-0.017 0.093

Pseudo R-squared 0.0415

35

TableIV:D&O Compensation and the Sensitivity to Performance.

This table estimates changes in D&O compensation and itssensitivity to firm performance after SEOs. The dependent

variables are: ln(TOTYRPAY), the level of total cash compensation, andΔln(TOTYRPAY), the change inthe logged value of

total D&O cash compensation. Columns (1) and (2) report second-stage results of IV regressions. The first-stage regression

result for column (1) is shown in Appendix II Panel A, Column (2) and the first-stage results for column (2) are shown in

Appendix II, Panel B. The sample period covers 2001-2012. All regressions include firm- and year fixed effects.

Bootstrapped standard errors are reported in parentheses. Coefficients marked with *, **, and *** are significant at 10%, 5%,

and 1%, respectively. Variable definitions are provided in Appendix I.

Dependent Variable ln(TOTYRPAY) Δln(TOTYRPAY)

(1) (2)

SEO x Δ(EBITDA/SALES) - -15.387*

- (8.61)

Δ(EBITDA/SALES) - -0.042

- (0.09)

SEO 0.114** 0.000

(0.05) (0.00)

ln(NYEAR_LISTED) -0.054** 0.004

(0.02) (0.04)

SALES 0.042*** 0.004

(0.00) (0.00)

SALES2 -0.001*** -0.000

(0.00) (0.00)

ROE 0.336*** 0.284***

(0.03) (0.05)

LEVERAGE 0.014 -0.128***

(0.04) (0.04)

PPE/TA -0.176*** -0.000

(0.05) (0.05)

SALES_GR 0.003 0.066***

(0.01) (0.01)

%_IND_DIR 0.044 0.049

(0.05) (0.07)

%_STATE_OWN 0.029 -0.001

(0.02) (0.03)

%_LARGEST_HOLD 0.324*** 0.161**

(0.07) (0.08)

NONTRDPCT -0.002*** -0.001***

(0.00) (0.00)

PAY_SIZE 0.046*** -

(0.00) -

ΔPAY_SIZE - 0.046***

- (0.00)

Constant 0.186*** -0.014

(0.06) (0.11)

Firm & Year FE Y Y

Observations 15,068 12,578

Adjusted R-squared 0.806 0.058

36

TableV: Tunneling.

This table estimates changes in tunneling activities after SEOs. Three accounts likely to be used to cover up tunneling are

employed as proxies for tunneling:ACCV/TA, accounts receivable over total asset; PREPAY/TA, prepaid expenses over total

assets; and%_ACCV_BAD, the fraction of accounts receivable unlikely to be collected. Columns (1) – (3) show the second

stage IV regression results. The first-stage regression results are reported in Appendix II, panel A. The sample period is

2000-2012 for columns (1) – (2), and 2004-2012 for column (3). All regressions include firm- and year-fixed effects.

Bootstrapped standard errors clustered at the firm level are reported in parentheses. Coefficients marked with *, **, and ***

are significant at 10%, 5%, and 1%, respectively. Variable definitions are provided in Appendix I.

Dependent Variable ACCV/TA PREPAY/TA %_ACCV_BAD

(1) (2) (3)

SEO 0.013*** 0.009** 0.041***

(0.00) (0.00) (0.01)

ln(NYEAR_LISTED) 0.014*** 0.002 -0.005

(0.00) (0.00) (0.01)

SALES 0.001*** 0.001*** -0.007***

(0.00) (0.00) (0.00)

SALES2 -0.000*** -0.000*** 0.000***

(0.00) (0.00) (0.00)

ROE 0.004 0.008*** -0.043***

(0.00) (0.00) (0.01)

LEVERAGE 0.012*** 0.023*** -0.028

(0.00) (0.00) (0.02)

PPE/TA -0.049*** -0.039*** 0.056***

(0.00) (0.00) (0.02)

SALES_GR 0.002* 0.002*** -0.012***

(0.00) (0.00) (0.00)

%_IND_DIR 0.003 -0.003 -0.018

(0.00) (0.00) (0.02)

%_STATE_OWN -0.002 -0.002 -0.001

(0.00) (0.00) (0.01)

%_LARGEST_HOLD -0.010* -0.008 -0.103***

(0.01) (0.00) (0.02)

NONTRDPCT 0.000** -0.000*** 0.001***

(0.00) (0.00) (0.00)

Constant 0.065*** 0.035*** 0.190***

(0.01) (0.00) (0.02)

Firm & Year FE Y Y Y

Observations 18,365 18,365 13,809

Adjusted R-squared 0.739 0.433 0.596

37

Table VI: SEO Announcement Returns.

This table shows the average cumulative abnormal returns (CARs) surrounding the announcement date of seasoned equity

offerings from year 2000 to 2012. CARs are calculated based on the market model, with an estimation window of 270

trading days prior to event window. SEO_CAR(-1, 1) is the mean cumulative abnormal return from day -1 to day 1

surrounding the announcement date, where the announcement date is day 0.T-test of whether the mean CARis equal to 0 is

reported in column (4). Panel A is based on the full sample. In panel B, we divide the sample by the sign of post-SEO

changes in level of overinvestment; D&O compensation-for-performance sensitivity,defined as Δlog(TOTYRPAY)over

Δ(EBITDA/SALES); and the fractionof accounts receivable unlikely to be collected. Column (1) shows the grouping criteria.

Coefficients marked with *, **, and *** are significant at 10%, 5%, and 1%, respectively. Variable definitions are provided

in Appendix I.

Grouping Criteria N SEO_CAR (-1, 1) t-value

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

Panel A: Full-Sample

Full Sample

557 -0.73%*** -3.28

Panel B: By subsample

ΔAB_INV + 220 -1.19%*** -3.81

- 162 -0.85%*** -2.11

ΔPAY_SENSI + 89 -1.34%** -2.16

- 87 -1.97%*** -3.54

Δ%_ACCV_BAD + 56 -2.20%*** -3.15

- 58 -0.86% -1.06

39

TableVII: Robustness Tests.

This table reports re-estimation results of Tables III through VI with alternative SEO definitions, an alternative instrument, and alternative definitions of dependent variables, andthe current dividend payout ratio as an additional

control variable. All reported results are second-stage IV regression results. Panel A shows re-estimation results while excluding small SEOswith proceeds in the 10th percentile. Panel B shows re-estimation results while

excluding the year of SEO in the definition of SEO years. Panel C shows re-estimation results with a different instrumental variable construction, where the original two-year lag is changed to one-year lag. Panel D re-estimates

Tables III and IV with alternative dependent variables:ACQ_CAR(-1, 1) is replaced with CAR calculated over (-2, 2) event window, and the compensation measure is replaced with compensation to the top three highest paid

managers. Panel E shows re-estimation results with the dividend payout ratio added as a control variable. Definitions of all variables are provided in Appendix I. Robust standards errors reported in parentheses are clustered at

the firm level, except when acquisition announcement returns are the dependent variable, in which case standard errors are clustered at the industry level. Coefficients marked with *, **, and *** are significant at 10%, 5%, 1%,

respectively. Variable definitions are provided in Appendix I.

Panel A: Alternative SEO definition; excluding small SEOs

Dependent Variable LOG(CAPEX) ACQ AB_INV+ ln(TOTYRPAY) Δln(TOTYRPAY) ACCV/TA PREPAY/TA %_ACCV_BAD ACQ_CAR(-1,1)

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

SEO 0.630*** 0.704** 0.038** 0.125** -0.040 0.014*** 0.010** 0.041*** -0.102*

(0.13) (0.31) (0.02) (0.06) (0.09) (0.01) (0.00) (0.01) (0.05)

SEO x

Δ(EBITDA/SALES) -15.348*

(8.90)

Firm FE Y Y Y Y Y Y Y Y

Industry FE

Y

Year FE Y Y Y Y Y Y Y Y Y

Observations 18,269 16,636 16,626 15,068 12,578 18,365 18,365 13,809 5,264


-0.017 0.806 0.058 0.739 0.433 0.596 0.093


Panel B: Alternative SEO definition; excluding the year of SEO


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

SEO 0.544*** 0.303 0.068*** 0.018 -0.057 0.016*** 0.006* 0.036*** -0.098

(0.10) (0.26) (0.02) (0.05) (0.08) (0.00) (0.00) (0.01) (0.07)

SEO x


(21.34)


Industry FE

Y


Observations 18,269 16,636 16,626 15,068 12,577 18,365 18,365 13,809 5,264


-0.016 0.806 0.058 0.739 0.433 0.596 0.093


40

Panel C: Alternative SEO definition: one-year lag in defining the IVs


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

SEO 0.589*** 0.653* 0.042** 0.124** -0.178** 0.012** 0.009** 0.042*** -0.045

(0.11) (0.34) (0.02) (0.05) (0.08) (0.00) (0.00) (0.01) (0.03)

SEO x

Δ(EBITDA/SALES) 3.281

(15.14)


Industry FE

Y


Observations 18,269 16,636 16,626 15,068 12,578 18,365 18,365 13,809 5,264


-0.017 0.806 0.058 0.739 0.433 0.596 0.091

Pseudo R-squared

0.0415

Panel D: Alternative Definitions of Dependent Variables

Dependent Variable

ln(PAY3EXE) Δln(PAY3EXE) ACQ_CAR(-2,2)

(1) (2) (3)

SEO

0.114** 0.148 -0.054*

(0.05) (0.14) (0.03)

SEO x

Δ(EBITDA/SALES) -35.400***

(11.55)

Firm FE

Y Y

Industry FE

Y

Year FE

Y Y Y

Observations

15,068 9,326 5,264

Adjusted R-squared 0.806 -0.050 0.020

Panel E: With the current dividend payout ratio as an additional control variable


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

SEO 0.560*** 0.683* 0.038** 0.120** -0.038 0.013*** 0.009** 0.041*** -0.093*

(0.11) (0.35) (0.02) (0.05) (0.08) (0.01) (0.00) (0.01) (0.05)

SEO x


(9.06)


41

Industry FE

Y


Observations 18,269 16,636 16,626 15,068 12,577 18,365 18,365 13,809 5,264


-0.017 0.806 0.057 0.739 0.433 0.596 0.093


42

Figure I: Governance and Firm characteristics.

Each graph plots the mean overinvestment, total D&O compensation, or the fraction of uncollectible accounts receivable

over the year of SEO and two subsequent years for each subsample divided by the sample median of a specificgovernance

mechanism or firm characteristic. Three governance mechanisms are considered: ownership concentration as measured by

the percentage of all outstanding shares held by the largest shareholder, %_LARGEST_HOLD; board independence as

measured by the percentage of independent directors on the board,INDDIRPCT; and “for project” vs. “cash holding” to

indicate strong vs. weak regulator monitoring. Three firm characteristics are considered: firm Age,AGE;growth: P/B Ratio;

andfirm size: TOTASS. Variable definitions are provided in Appendix I.

Share Ownership Concentration

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

SEO Year SEO Year + 1 SEO Year + 2

AB_INV+

High Concentration Low Concentration

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45


TOTYRPAY


0

0.5

1

1.5

2

2.5

3

3.5

4

4.5


%_ACCV_BAD


1.

Board Independnece

0

0.01

0.02

0.03

0.04

0.05

0.06


AB_INV+

High % Indp. Director Low % Indp. Director

-0.4

-0.2

0

0.2

0.4

0.6

0.8


TOTYRPAY


0

1

2

3

4

5

6

7


%_ACCV_BAD


Regulator Monitoring

0

0.01

0.02

0.03

0.04

0.05

0.06


AB_INV+

For Cash Holding For Projects

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35


TOTYRPAY


0

1

2

3

4

5


%_ACCV_BAD


43

Firm Age

0

0.01

0.02

0.03

0.04

0.05

0.06


AB_INV+

Old Young

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6


TOTYRPAY

Old Young

0

1

2

3

4

5

6

7


%_ACCV_BAD

Old Young

Firm Growth

0

0.01

0.02

0.03

0.04

0.05

0.06


AB_INV+

High P/B Ratio Low P/B Ratio

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4


TOTYRPAY


0

1

2

3

4

5

6

7


%_ACCV_BAD


Firm Size

0

0.01

0.02

0.03

0.04

0.05

0.06


AB_INV+

Large Firm (by total asset) Small Firm

-0.4

-0.2

0

0.2

0.4

0.6

0.8


TOTYRPAY


0

1

2

3

4

5

6


%_ACCV_BAD


44

Appendices

Appendix I: Variable definitions.

Variable Name Description

Key variables

SEO Anindicator variable equal to one in SEO years (the year of SEO offerings and the year

after), and zero otherwise.

SEO_CAR(-1,1) Cumulative abnormal returns over a three-day event window surrounding the

announcement date of seasoned equity offerings.

AFFECT_REG Anindicator variable equal toone if the firm does not satisfy the dividend requirement in

year t. To satisfy the requirement, the two-year lagged dividend over profit ratio should be

larger than 20% before 2010, and larger than 30% from 2010 and thereafter.

IV_SEO Instrumental variable constructed based on the 2006 and 2008 regulation: IV_SEO =

AFFECT_REG * POST_REG, where POST_REG is equal to one when the yearis 2008 or

2010, to capture the effect of the 2006 or 2008 regulation.

CAPEX Capital expenditures to acquire fixed assets, intangible assets, and other long-term assets,

measured in millions of RMB, year 2000 price level.

AB_INV The residual of the following investment model in Richardson (2006): Invi,t = β0 + β1Tobin’s

Qi,t-1+ β2Leveragei,t-1 + β3Cashi,t-1 + β4Firm_Agei,t-1 + β5Ln(TA)i,t-1 + β6YRRETi,t-1 + β7Invi,t-1

+ at + aj + εi,t,, where Invi,t is the net investment firm i makes in year t, defined as the ratio of

(CAPEX – proceeds from disposal of fixed assets, intangible assets, and other assets) to

total assets at the beginning of the year. In our regression, we use AB_INV+, which is

AB_INV with negative values replaced by zero.

ACQ Acquisition dummy, equal to one if firm i makes an acquisition in year t.

ACQ_CAR(-1,1) Cumulative abnormal returnsover the three-day event window surrounding acquisition

announcements.

TOTYRPAY Total D&O cash compensation: the sum of cash salaries and bonusespaid to board chair,

CEO, vice president, board members, and key management members, measured in millions

of RMB, year 2000 price level.

PAY3EXE Total cash compensation paid to three highest paid executives, measured in millions of

RMB, year 2000 price level

EBITDA Earnings before interest, taxes, depreciation and amortization, measured in millions of

RMB, year 2000 price level.

ACCV / TA Accounts receivable over total assets

PREPAY / TA Prepaid expenses over total assets

%_ACCV_BAD The fraction of accounts receivable classified as unlikely to be collected

Control variables

Listing_Years Number of years since a firm's IPO

FIRM_AGE Number of years since a firm’s establishment

SALES Total sales, measured in billions of RMB, year 2000 price level

SALES2 The square of SALES.

ROE Return on equity: the ratio of net profit to owner's equity.

LEVERAGE The ratio of total debts (short term debt + long term debt) to total assets.

45

Control variables (contd.)

PPE/TA The ratio of tangible asset (properties, plants, and equipment) to total assets.

SALES_GR SALES growth rate from year t-1 to year t.

%_IND_DIR Percentage of independent directors on the board.

%_STATE_OWN Percentage of shares held by the government through a designated government agency.

%_LARGEST_HOLD Percentage of shares held by the largest shareholder.

NONTRDPCT Percentage of non-tradable shares.

PAYSIZE Number of managers included in the total D&O cash compensation, TOTYRPAY

46

Appendix II: First-stage Regression Results.

Panel A. First-stage results for Tables III, IV and V

Column (1) is for Tables III and V, and Column (2) is only for Table IV, Column (1).

Dependent Variable SEO SEO

(1) (2)

IV_SEO -1.996*** -1.569***

(0.55) (0.54)

Controls Y Y

Firm & Year FE Y Y


F-test (IVs) 13.08 8.36

Prob> F (IVs) 0.0003 0.0038

Panel B. First-stage results for Table IV, Column (2).

Dependent Variable SEO SEO xΔ(EBITDA/SALES)

(1) (2)

IV_SEO -1.415** -0.002***

(0.56) (0.00)

IV_SEO x Δ(EBITDA/SALES) -0.791 -0.000

(1.41) (0.00)

Controls Y Y

Firm & Year FE Y Y


F-test (IVs) 7.88 3.93

Prob> F (IVs) 0.0194 0.0196

seasoned equity offerings and agency problems: evidence ... · seo is associated with a number of...

Documents