Financial Research Reports on Corporate Finance

Description
Corporate finance is the area of finance dealing with the sources of funding and the capital structure of corporations and the actions that managers take to increase the value of the firm to the shareholders, as well as the tools and analysis used to allocate financial resources. The primary goal of corporate finance is to maximize shareholder value.

FINANCIAL RESEARCH REPORTS ON CORPORATE
FINANCE

ABSTRACT

Over the past twenty years, write-ofs have grown in popularity. With the increased usage

of write-ofs, it is becoming more important to understand the mechanisms behind why companies

take write-ofs and how write-ofs afect company performance. In this paper, I examine the

cross-sectional determinants of the decision to take write-ofs. I use a hand-collected dataset on

write-ofs that is much more comprehensive than existing write-of datasets. Contrary to much

hype and scandals surrounding a few write-ofs, I find that quality of governance is positively

related to write-of decisions in the cross-section. My results also suggest that poor governance

companies wait to take write-ofs until it becomes inevitable, while well-monitored companies take

write-ofs sooner. As a result, the charge is substantially larger than the average write-of charge.

When these poor governance companies announce write-ofs, the announcement generates negative

abnormal returns. However, when good corporate governance companies announce write-ofs, the

charge is substantially smaller than the average charge. These well-monitored companies take

write-ofs immediately following a problem. Following the write-of announcements of these types

of companies, average announcement day efects exceed a positive six percent. These results suggest

that companies with quality monitoring mechanisms use write-ofs in a manner that is consistent

with enhancing shareholder value.

In my second essay I examine the efect of write-of announcements on the stock market liquidity

of firms taking write-ofs from 1980 to 2000. I find that there are substantial improvements

in stock market liquidity following corporate write-ofs. Spreads decrease and turnover volume

increases after write-of announcements, which indicates an improvement in liquidity. The liquidity
improvement is greater for better governed companies. I decompose bid-ask spreads and show

that adverse selection costs decrease substantially as market participants respond to the write-of

announcement. The evidence suggests a liquidity benefit of write-ofs that must be weighed against

any other perceived cost of write-ofs. Such a liquidity benefit may validate that write-ofs convey

favorable information about the firm.
TABLE OF CONTENTS



1 Write-ofs and Corporate Governance 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.4 Characteristics of Write-of Companies . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.4.1

1.4.2

1.4.3

1.4.4

1.4.5

1.4.6

Corporate Cleanup Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . .

Executive Compensation and the Write-of Decision . . . . . . . . . . . . .

Monitoring Mechanism Hypothesis . . . . . . . . . . . . . . . . . . . . . . .

Multivariate Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Weak Shareholder Protection and Write-ofs . . . . . . . . . . . . . . . . . .

Governance and Size of Write-ofs . . . . . . . . . . . . . . . . . . . . . . .

14

18

20

23

25

26

1.5 Market Reaction and Write-of Announcements . . . . . . . . . . . . . . . . . . . . 28

1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

1.7 Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32


2 Write-ofs and Liquidity 43

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

2.2 Sample and Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

2.2.1 Liquidity Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

2.3

2.4

2.5

2.6

2.7

2.8

Liquidity Efects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Governance and Liquidity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Adverse Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Robustness Checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

47

48

49

51

52

53


Bibliography 65






iv
Chapter 1

Write-ofs and Corporate Governance

1.1 Introduction


Write-ofs have become increasingly common in the past two decades. The consumer-manufacturing

sector alone had write-of charges totaling over $2 billion in 2000 compared to $800 million in 1980,

an increase of 250 percent.



There are three potential explanations for why companies take write-ofs. First, write-ofs

are a consequence of poor managerial decision-making. Write-ofs become inevitable actions for

companies that are sufering from a chain of management errors. Second, write-ofs can be a re-

sponse by management proactively responding to negative shocks to the company. Even quality

managers that take calculated risks can have problems ex-post. Management consciously decides

to amend these problems by taking a write-of that also provides private information to the market

concerning the quality of the firm. Third, in companies with CEO turnover, write-ofs can act as

a tool that allows the new CEO to get rid of bad accounts left by the previous CEO.



In this paper, I examine why companies take write-ofs and how the market reacts to write-

of announcements; for this analysis, I use a carefully collected dataset of consumer manufacturing

companies, focusing on asset and lay-of based write-ofs. I characterize what defines good and

bad write-ofs, and analyze the characteristics of companies that take diferent types of write-ofs.

Finally, I examine the shareholder wealth efects of write-ofs, and whether firm specific factors

in?uence the market's reaction to the write-of announcement.



I find that the write-of decision is linked to industry shocks. Governance mechanisms also




1
afect the write-of decision. Companies with high pay-performance sensitivity, desirable board

composition, strong shareholder protection measures, and CEO turnover resulting in an external

replacement are all significantly correlated to a tendency to take write-ofs. I find a negative

relationship between governance quality and the size of write-ofs, which suggests that poorly

monitored companies wait to take write-ofs and continue to accumulate problems. Eventually

the problems become so large that a write-of is inevitable. Conversely, well-monitored companies

take write-ofs sooner. Since these companies act quickly, there is comparatively less that they can

write-of, so the charges of well-monitored companies are less than the charges of poorly monitored

companies. I also find that these well-monitored companies exhibit significant positive announce-

ment efects (upwards of six percent for write-of companies with small boards, strong shareholder

protection, and large percentage of outside directors. I conclude from these results that firms with

efective monitoring mechanisms take value-enhancing write-ofs.



The paper is constructed as follows. Section II reviews the relevant literature. I describe

my sample in Section III. Section IV examines the link between CEO turnover, pay-performance

sensitivity, corporate governance, and the write-of decision. Section V looks at how the market

reacts to write-of announcements. Section VII concludes. Appendix A discusses the tax issues

related to write-ofs.





1.2 Literature Review


Other studies look at write-ofs, but from diferent perspectives and with results that are not di-

rectly comparable to mine. The write-of literature focuses on three main areas, the efects of

write-ofs on returns, the relationship of earnings and write-ofs, and the impact of SFAS 121 on

write-of announcements. However, my primary focus is to explore the relations between the gov-

ernance of a firm and the motivation to take write-ofs as a business related decision.





2
The first branch of literature, which looks at efects from write-of announcements, has

mixed results. Some papers find that write-ofs generate no abnormal returns, while others find

that write-of announcements generate both positive and negative abnormal returns depending on

the segmentation of the sample. Francis, Hanna, and Vincent (1997) collect and analyze write-ofs

from 1989 to 1992. Their analysis shows that on average the market views write-ofs as negative

news, although it is possible to explain some of the dispersion in market reactions by identifying

diferent types of write-ofs, such as inventory or restructuring. Their study provides evidence that

both earnings management and asset impairment drive a write-of decision. Although Francis et

al. use the same collection technique as I use, their sample spans fewer years, and contains fewer

announcements. Their results motivate me to examine what role earnings management and asset

impairment have in write-of decisions of good governance companies.



Meyer and Strong (1987) identify a sample of 78 write-of firms from the Wall Street Jour-

nal Index during 1981 - 1985. They construct a picture of a typical write-of firm; it has weak

prior performance, changes in top management, and is highly leveraged. They also analyze an-

nouncement efects and report negative and insignificant abnormal returns, although the returns

are widely dispersed. This paper relates to my study in that we both consider the impact of CEO

turnover on a write-of decision, and it leads me to the yet untested hypothesis that the type of

governance structure and CEO turnover afect the write-of decision.



Bartov, Lindahl, and Ricks (1998) use a key word search from Dow Jones News to compose

a sample of write-of firms. They attempt to explain why the stock price changes around write-of

announcements are so small relative to the average write-of amount. They suggest that the mar-

ket under-reacts to the write-of announcement and find that abnormal returns are negative by as

much as 21 percent after the announcement. Brickley and Van Drunen (1990) find a positive and

significant average abnormal return around the announcement of restructuring charges. Kross,

Park, and Ro (1996) also find a positive market reaction to the announcement of an initial re-



3
structuring charge, as well as increases in trading volume and market return variability. Alciatore,

Easton, and Spear (2000) examine the timeliness of write-ofs for oil and gas firms under the SEC's

full-cost ceiling test. These authors find that write-ofs have a significant negative association with

contemporaneous quarterly returns and an even more negative association with prior quarter re-

turns. They conclude that such impairments are not timely insofar as they are re?ected in returns

before the announcement of a write-of. Zucca and Campbell (1992) find no significant diference

in stock performance from 60 days prior to 60 days after a write-of. He?in and Warfield (1995)

find that the returns for write-of firms during the write-of year are negatively correlated to the

amount of the charge. These papers all focus on the abnormal returns associated with write-ofs.

They do not consider what motivates firms to take write-ofs, and the relation between governance

and write-ofs, a main purpose of this paper.



Another branch of the write-of literature looks at the relation between write-ofs, earnings,

and performance. Kinney and Trezevant (1997) examine a large sample of Compustat data span-

ning the ten-year period 1981 through 1991 and find that write-ofs are consistent with earnings

management. They report that firms with large changes in reported earnings recognize significantly

negative income from special items. This finding is consistent with dampening large increases to

produce a smooth, upward trend in earnings. Elliot and Hanna (1996) study the information

content of earnings conditional on the presence of write-ofs. They also look at the incremental

information content of these write-ofs. Their main finding is a significant decline in the weight

attached to unexpected earnings in quarters following write-ofs. They conclude that this shows

evidence that write-ofs create noise in the information environment. These papers concentrate on

how write-ofs afect the information environment. They do not consider how endogenous factors

such as corporate governance might afect the value of the information contained in the write-of

announcements. My study adds to this branch of literature by examining how cross sectional

characteristics (such as CEO turnover, governance provisions, and pay performance sensitivity)

in?uence the information environment surrounding a write-of.



4
The last branch of literature is tied to the impact of FASB's 1995 issuance of SFAS 121,

Accounting for the Impairment of Long Lived Assets. SFAS 121 was intended to reduce manage-

rial ?exibility and enhance the reporting of long-lived asset write-downs.
1
Kim and Kwon (2001)

examine the diference in market reaction for early versus late adapters of the new FASB standard.

They find that early adapters have a positive market reaction, but late adapters have a negative

market reaction to a write-down announcement. Riedl (2002) compares the types of write-downs

taken before and after SFAS 121. He finds that write-downs reported prior to the standard have a

greater association to economic factors than do write-downs reported after the standard. Lindbeck,

Rezaee, and Smith (1996) find that write-downs increase in magnitude following the adoption of

SFAS 121. These papers all focus on one type of write-of and one main event, whereas my research

covers a broader period, as well as a more extensive array of write-of types.
2





1.3 Data


To generate my sample, I collect write-of information, focusing on announcement behavior be-

tween 1980 and 2000 made by companies in the 2000-2999 SIC code, which are primarily consumer

manufacturing companies. I focus on this particular industry because it is a mature industry that

is asset intensive, and therefore might have greater incentives to take write-ofs compared to other

industries. Using a CRSP generated perm and SIC code list, I search Lexis-Nexis and Dow Jones

Retrieval services for specific key words. For each company, I search for articles that match key

words. The key words I use are write down, write-of, restructure, charge against earnings, layofs,

and severance. When the query results in a match, I take the first article in the series of articles

that refers to a current write-of that the company is announcing. I use the date of the article as

1
This standard addresses (1) the criteria for when to test for the existence of an impairment, (2) the level at which to group assets in
testing for impairment, (3) the measurement basis for determining the existence of an
impairment, (4) the measurement of the impairment, and (5) the presentation of the recognized amount. The
standard only applies to write-downs.
2
I test the robustness of my results for the impact of SFAS 121 in two ways. First, I segment my sample across
time and do not find any significant changes in my analysis. Second, I remove the subset of write-downs from my
sample, and find no significant changes to my analysis.


5
the announcement date of the write-of. I obtain the following information from the article: the

amount of the write-of; whether the write-of was generated by an asset write-down, employee

layofs, or both; the purpose of the write-of (restructure, write-down, plant closing, etc.); the

justification cited by the company; and whether the write-of amount is stated on a before-tax or

after-tax basis. The sample contains asset-based and layof-based write-ofs. I find 2,429 companies

within the consumer manufacturing industry. From this sector, 803 companies (33 percent) had a

write-of, giving a combined 3,738 write-of announcements.



Write-ofs represent either a write-down of assets, charge due to corporate restructuring, or

charge due to lay-of events. SFAS 5 requires a firm to write down or expense asset values that will

not be recoverable from future operations. SFAS 121 clarifies these circumstances for write-downs.

SFAS 5 and APB Opinion 30 require firms to report restructuring charges, including charges from

the sale or acquisition of a business in the year incurred. The disposition of a complete business

segment must be reported as a separate line item called discontinued operations. Other write-ofs

can appear in the footnotes of financial statements. Appendix A provides a more complete expla-

nation of the way write-ofs are handled in financial reporting. In this study, I focus on write-ofs

that include the partial disposition of a segment, discontinued operations, restructuring charges,

plant closings, costs of employee terminations, and other special charges that are either unusual

or infrequent, but not considered an extraordinary item.



I do not include write-ofs due to litigation costs, bankruptcy, goodwill, or capital structure

refinancing in this data set. By including only write-ofs that are related to operational decisions,

I can examine the impact on future performance and avoid the legal and accounting peculiarities

that are associated with other types of write-ofs. Only write-ofs that are announced singularly are

included in the dataset, so that I can attempt to isolate both the reasons companies take write-ofs

and the market's reaction to the announcement.





6
Although COMPUSTAT has data on write-ofs, I opt to use the hand-collected data set

for the following reasons. Information on write-ofs can be found in Compustat data item #17.

Compustat does not report charges that it deems inconsequential, but these can be important to

establish a history of write-ofs. Compustat also understates most write-ofs. I compare the charge

amounts listed in the write-of announcements to the charges recorded in Compustat. Overall, the

public announcement of the write-of charge averages $3.41 million more than the COMPUSTAT

write-of charges. All of the write-ofs identified in COMPUSTAT are also listed in my sample,

but there are 352 write-ofs from my sample that are not listed in the COMPUSTAT sample. The

diferences in my sample versus COMPUSTAT are similar to the diferences reported in an earlier

study by Fried, Schif, and Sondhi (1989).



To ensure that write-ofs in my sample are not extensions of earlier events, I set an arbitrary

standard under which I assume that any write-of announcements occurring within six months of

earlier write-of announcements are related. This exercise is also performed for break of points

of one month, three months, four months, eight months, and twelve months. Although doing so

afects the sample size, it does not afect the analysis or findings. Therefore, I only describe results

using the 6-month break point.



It is important to determine which write-of is a first-time event or a subsequent event.

To define multiple write-ofs, I need to establish an arbitrary time interval. The standard most

researchers use defines multiple write-ofs as any write-of event that occurs within 16 quarters of

a prior write-of event.
3
To identify a company's first write-of, I look at all write-ofs that occur

during the first five years of the sample: 1980-1985. I require an initial period of 16 fiscal quarters

with no write-ofs before I add a firm to the sample. I denote the write-of following this break

as a first time write-of. Because the original sample begins in 1980, the first reported write-of in

the sample occurs in the first quarter of 1985. To test the sensitivity of this break point, I also

3
See Elliot and Hanna (1996).




7
use five other quarter break points to define first time write-ofs, (8, 12, 18, and 20 quarters) to

separate consecutive write-ofs. My conclusions become more robust with the longer measures and

weaken slightly with the short-term definitions. Since the inference changes only marginally, I use

16 quarters. This procedure leaves me with 767 firms and 1,798 write-of events to evaluate. After

I identify the first time write-of for a company, write-ofs that follow are labeled as second, third,

fourth write-ofs, etc. These subsequent write-ofs must occur within 16 quarters after the prior

write-of. If the write-of occurs after 16 quarters, I label it as another first time write-of.



To compare write-of company characteristics, I construct a sample of non-write-of firms.

Out of the 2,429 firms in the 2000-2999 SIC codes for 1980-2000, there are 1,626 firms that do

not have a write-of. I sort these non-write-of firms into their primary 4-digit SIC code. Each

write-of firm in the sample is matched to at least two firms in the non-write-of group. This

match is based on 4-digit SIC codes and similar total assets. If there are no firms that match

the 4-digit SIC code of the identified write-of, I use the 3-digit or 2-digit SIC code. To match by

size, I also pair write-of firms to non-write-of firms of similar total assets.
4
The matching results

in a sample size of 2,037 write-of events composed of 767 write-of firms and 995 non-write-of firms.



Using the PERM numbers for my write-of sample, I merge COMPUSTAT and CRSP data

into my sample. I measure the abnormal return measure as the market-adjusted returns. The

reported results use the diference between daily CRSP returns and daily returns on the CRSP

equally weighted market portfolio. I also use the value-weighted, beta-weighted, and market-

capitalized CRSP market portfolios, and the results remain similar. I use the quarter prior to

the write-of announcement to match COMPUSTAT data, such as book value, earnings per share,

sales, shares outstanding, and total assets, for the write-of sample.



To calculate abnormal returns, I define the event window as the day of the write-of an-

4
I

note that the non write-down firms have a total asset value that is at most 10 percent greater than the write-of
companies are, or at most 10 percent less than the write-of companies.



8
nouncement in the financial press, which acts as the date in which the information concerning the

write-of becomes public. I use the three-day horizon surrounding the announcement date (t = ÷1

to t = 1) to calculate the announcement efects. The abnormal return is the actual ex post return

of the security over the event window minus the normal return of the firm over the event window

(Brown and Warner, 1985). For any company i in month t,



AR
it
= R
it
÷ E(R
it
), (1.0)


where R
it
is the realized return on day t, and E is the expectations operator. I estimate the ex-

pected return E(R
it
) for each firm as the return on equal-weighted size portfolio model. I estimate

the average abnormal return (A
it
) for each day in the sample as follows: ¯


N
A
it
= 1/N
¯


i=÷1
AR
it
, (1.1)

where N is the number of securities. A
it
is a cross sectional average.
5
¯



Table 1, Panel A shows the distribution of write-ofs over the 16-year sample period. The

sample contains 604 pure asset related transactions, 1,175 write-ofs that combine both assets and

layofs, and 258 write-ofs related to lay-ofs. The number of write-ofs per year triples from the

beginning of the sample (49 in 1985 to 212 in 2000). Panel B of Table 1 shows the frequency of

write-ofs by firm from 1985-2000. Out of the 767 firms that take one write-of, 61 percent take an

additional write-of, and 42 percent have at least two additional write-ofs.



Table 2 describes the write-of sample. Restructures are by far the most common type of

event, occurring in more than 56 percent of write-of incidents. Discontinued operations are the sec-

ond most common write-of event, occurring 14.28 percent of the time. The table also displays the

average charge for each type of write-of. Partial asset write-downs, and restructuring charges have

5
I also calculate returns using the market model, with an estimation period of -90 to -61 days before the write-of announcement. I
choose the estimation period to minimize the problems associated with estimating parameters
with data in?uenced by the write-of event. The results are comparable to the reported results.


9
the greatest magnitudes, averaging $78 million and $72 million, respectively. Write-of amounts

range from $56 thousand to $2.1 billion, with a mean of $45 million. The distribution is posi -

tively skewed; the median write-of is $22 million. This skewness is also observed for each write-of

category. The table also shows the average write-of charge total assets (TA). When adjusted by

BV and TA, discontinued operations-based write-ofs are the largest, followed by restructure-based

write-ofs. On average, discontinued operations charges are over 13 percent of a write-of firm's

total assets. Restructuring charges were second with charges over 6 percent of a write-of firm's

assets. The total amount of firm value written of over the sample amounts to over $98 billion

dollars. Due to the potential shareholder welfare implications from a loss this large, it is important

to understand the cross sectional characteristics of write-of companies, and the resultant impact

on shareholder value.



Table 3 describes the mean and median of firm specific variables used in subsequent probit

models. I also report univariate significance tests to determine whether there is any diference

between the write-of values and the non write-of values. The variables shown in the table include:


• MV = the size of the firm, measured as the log of market value one quarter before the

write-of was announced.


• SHROWNPC = the percent of the new (if replacement) or old (if not replaced) stock holdings

in the company for the CEO.


• SALARY = the dollar salary of the CEO in the year of the write-of, in 100k.


• BONUS = the dollar bonus for the CEO in the year of the write-of, in 100k.


• OPTIONS = the aggregate dollar value of all options granted to the executive during the

year as valued by the company, in 100k.


• SIZEBD = the number of directors, both inside and external.


• PER OUT = the percentage of directors who have no relationship with the company.


10
• DIROPT = the number of options, which each non-employee director received during the

year in thousands.


• DIRSTK = the number of shares, which each non-employee director received during the year

in thousands.


• GOV INDEX = Governance index, the lower the number the better the shareholder protec-

tion, the higher the number, the worse the managerial entrenchment. This index is based on

charter provisions listed by IRRC publications


• DOLLAR SENSITIVITY = the dollar sensitivity of compensation to performance.


• RETURN SENSITIVITY = the return sensitivity of compensation to performance.


• ROA = the return of assets for the current quarter.


Firm size is related to the likelihood of write-ofs occurring, as discussed in Meyer and

Strong (1989). The larger the firm, the more assets it can divest. Table 3, Panel A, shows that

non-write-of firms have the lowest market value, with an average MV of $191 million. One-time

write-of firms are on average $15,367 million, and are significantly larger than the benchmark

(significant at 5 percent), while multiple write-of firms are the largest with a market value of

$ 265,667 million (significant at 5 percent).
6
One-time write-of executives own four percent of

their company's stock, followed by multiple write-of firms at three percent, and then non-write-of

firms at two percent. Only one time write-of company CEOS have significantly diferent stock

ownership as compared to the benchmark (significant at 10 percent). I find that CEOS of the

one-time write-of firms and the multiple write-of firms are paid $666 thousand, and $629,000

respectively, which is less than non-write-of CEOs pay, $657,000. Likewise, non-write-of firms

receive larger dollar bonuses than write-of firms ($848,000, $534,000, and $548,000 on average for

non write-ofs, one-time write-ofs, and multiple write-ofs, respectively). First-time write-of firms

have option grants of $2,385,000, which are the highest value of option grants, followed multiple at

6
All of the t-values presented test for whether there is a significant diference between the write-of firms and the benchmarks.



11
$2,039,000, and non write-of firms,$1,346,000. First-time write-of firms have the smallest boards,

with 10.17 members, followed by multiple, and non write-of firms (11 members, and 12 mem-

bers respectively). Indeed, not only are the boards smaller for write-of firms, but they are also

dominated by outsiders (76 percent, for multiple write-of firms, 73 percent for first-time write-of

firms, and 69 percent for non-write-of firms. These results suggest that the boards of write-of

firms are better monitors than are the boards of non-write-of firms. GOV INDEX, a measure of

the level of shareholder protection measures from the IRRC database, show moderately stronger

protection measures for write-ofs as compared to non write-of firms. The first time and multiple

write-of companies show an average ROA of 4.25 percent and 4.7 percent, respectively and are

both significantly less than the benchmark firm's ROA of 7.3 percent.



I measure pay-performance sensitivity in two ways. The first measure is the dollar sensitiv-

ity of CEO compensation, defined as the change in the dollar value of the CEO's stock and option

holdings for a dollar change in firm equity value. The second is the change in the dollar value of the

CEO's stock and option holdings for a one percent change in a company's stock price (the return

sensitivity). Baker and Hall (1998) argue that the return sensitivity measure is the appropriate

one to use when CEO actions afect firm percentage returns through their control of firm strategy.

I calculate these measures using the Core and Guay (1999a, 1999b) method, which allows me to

compute incentives using the one-year data on a CEO's stock option portfolio contained in the

annual proxy statements. I estimate the regressions using both measures of incentives and find

similar results. Pay-performance sensitivity is defines as follows, where W denotes CEO wealth in

options and stocks held, and V denotes firm value,





ReturnSensitivity = r = 0.01 - dW/(dV /V )


and

DollarSensitivity = (dW/dV ) = 100 - r/V. (1.2)


12
These ratios act as proxies for the degree of pay-performance sensitivity of a CEO. This

sensitivity shows the percentage by which pay increases (decreases) when company performance

increases (decreases) by one percent. Table 3 shows the summary of these two sensitivity measures.

First-time write-ofs show the greatest sensitivity, followed by multiple and then non-write-of firms.



In addition, I calculate the TLCF, the tax-loss-carry-forwards of the company, in the write-

of quarter.
7
Two possible relationships between TLCFs and write-ofs are if a company has had

prior poor performance, it is possible that they have had a tax-loss-carry-forward, and the TLCF

would act as a proxy for poor performance, which would lead to a positive relationship with write-

ofs. Second, if the company already has a TLCF, there are fewer tax incentives to take a write-of,

and so one would expect a negative relationship between write-ofs and the TLCF. In this paper,

TLCF is a dummy variable that is one if the company has an unused tax-loss-carry-forward, and

zero otherwise. I also calculate the debt ratio, DEBT RATIO, defined as total debt over total

assets. I expect that due to tax incentives, there will be a negative relationship between write-ofs

and debt. If a company has more debt, it has the tax shelter from the interest expense, which

would ofset a tax advantage from taking write-ofs.



Table 3, Panel B, shows the governance characteristics broken up by year. This table in-

cludes all firms in the sample over all years of the sample, regardless of whether the firm took a

write-of in that particular year. There is very little change in the governance variables from year

to year. Since the governance variables are sticky, they act like firm specific efects in the following

regressions. Since there is very little change in board size, percent of outside directors, and the

IRRC index over time, these variables can be considered given; that is, they are determined outside

the write-of decision and are not impacted by short term changes in performance. However, CEO

turnover does present a self-selection issue in relation to the write-of decision and performance

changes.

7
See Plesko (1999) for a description of the calculation of the TLCF.




13
1.4 Characteristics of Write-of Companies


Companies with efective monitoring mechanisms are not immune to problems. Certain circum-

stances, such as negative economic shocks, or increased product market competition can negatively

afect the company's performance. However, these well monitored companies quickly recognize the

problems and take actions to fix the problem areas.
8
This argument suggests that good gover-

nance firms have smaller multiple write-ofs, while poorly governed companies have fewer, but

much larger write-ofs. One would then expect to see a positive correlation between governance

quality and the probability of taking write-ofs, and an improvement in future earnings for the

write-of company.



In this paper, I look at how four factors might afect the write-of decision, and in?uence

short run consequences or long run benefits. These factors are CEO turnover, pay-performance

sensitivity, board composition, and managerial entrenchment. I first test each hypothesis individ-

ually to see how the firm characteristic is related to the write-of decision and then examine the

hypotheses jointly to see how the characteristics interact in relation to the write-of decision.




1.4.1 Corporate Cleanup Hypothesis


Borokovich, Parrino, and Trapani (1996) show that a turnover announcement is normally suc-

ceeded by a positive market reaction for outside replacement, and a negative reaction for insider

replacement, especially when the replacement is not voluntary. Prior corporate governance research

also emphasizes that a critical element of corporate governance mechanisms is an ability to identify

and terminate poorly performing executives (Kaplan (1994); Cofee (1999); Murphy (1999); Volpin

(2002); Berger, Ofek, and Yermack (1997)). For example, Macey (1997) observes that a necessary

8
See, e.g., Paul (2003), which describes how all managers make mistakes, but that companies with good gover- nance are the first to
correct these mistakes.



14
condition for competent corporate governance systems is the removal of poorly performing man-

agers. Gibson (1999) asserts that a primary purpose of corporate governance mechanisms is to

ensure that poorly performing managers are removed. The importance of replacing unfit CEOs

is also consistent with Shleifer and Vishny (1989, 1997), who speculate that the most important

form of managers expropriating shareholder wealth are unqualified managers who remain with the

company. Jensen and Ruback (1983) also support this position by arguing that poorly performing

managers who resist removal might be the costliest manifestation of the agency problem.



Whether CEO replacement is external or internal could also afect the write-of decision. An

external replacement would be more likely to result in a write-of than an internal replacement. An

external replacement typically indicates diferent agenda for the company, which can result in the

new CEO cleaning up the internal problems prior to pursuing new goals. An internal replacement

is a part of the earlier CEO's agenda and has less incentive to take a write-of. In addition, if the

new CEO expects to remain in place for a longer horizon, then she has an incentive to see the

company's profitability to improve in the long term. Formally stated:



H1Corporate Cleanup Hypothesis) The probability of a write-of is greater for

firms that have recently had CEO turnover, especially if the replacement is external .



Table 4 Panel A shows the characteristics of executive turnover associated with a write-of

announcement. I obtain my turnover data from Execucomp. After matching the two datasets,

my combined sample comprises of 886 write-of events for the 1992 to 2000 period. I label CEO

replacement that occurs within a year prior to the write-of announcement as a related event. Ex-

ecucomp lists the reason the CEO leaves the company. If applicable, I mark one of the following

options: resigned, retired, deceased, or unknown. The table shows two types of turnovers, those

with external replacements, and those with internal replacements. I define a replacement as ex-

ternal or internal by comparing the date the CEO entered ofce and the date the CEO joined the



15
company. If these coincide, the CEO is external, otherwise internal.



Table 4 Panel A describes the turnover statistics for non write-of, first-time write-of, and

multiple write-of firms. Non-write-of firms have less CEO turnover than do one-time write-of

firms, or multiple write-of firms. There are 130 CEO turnovers for multiple write-of companies,

59 for one-time write-of companies, and only 35 for the benchmark companies over the eight years

of the sample. The majority of replacements came from outside the company (65 percent, and

81 percent, and 77 percent for multiple, first, and non write-of firms respectively). These results

suggest that there is a link between write-ofs and turnover.



As discussed above, the firm that decides to terminate its CEO may do so because of

unobserved information that is potentially concealed with the information that leads to a write-

of. This leads to self-selection in the CEO turnover decision. In order to control for this source of

self-selection bias, I run a two-stage sample selection model. In the first stage, I run the following

probit estimate:



pr(EXT U RN OV ER) =|
1
+|
2
LOGM V
i
+|
3
ROA
i
+|
4
R
i
+
i
, (1.3)



where EXTURNOVER is a dummy variable that is equal to one if there was CEO turnover with

an external replacement, and zero otherwise. R , the unadjusted cumulative monthly stock return

for the firm over the past 12 months, is the independent variable, which measures CEO perfor-

mance . The specification follows Parrino 1997. Table 4 Panel B gives the results of the estimate.

Consistent with prior work, I find that the probability of forced CEO turnover is estimated to be

negatively and significantly related to the prior stock return. ROA shows the accounting prof -

itability of the firm one year prior. ROA, and the log of market value are also negatively related

to EXTURNOVER. The results are similar to those represented by Parrino (1997).



In the second stage, I model the decision to take a write-of using a probit model. A firm


16
takes a write-of if latent variable WO> 0 and no write-of if WO s 0. WO
i
is empirically specified

as:



pr(W O) =|
1
+|
2
SIZE
i
+|
3
SHROW N P C
i
+|
4
CEO OU T +|
5
ROA+


|
6
T LCF +|
6
DEBT RAT IO +¸
i
, (1.4)


where¸
i
is standard normal. CEO OUT is the inverse mills ratio from the CEO turnover probit

equation and it corrects for the self-selection in the decision to replace the CEO. ROA is return

on assets, TLCF is a dummy variable that is one if a company has tax-loss-carry-forwards, and

DEBT RATIO is the debt ratio. and¸ is a standard normal error term.



Table 5, Model 1, shows the probit results for Equation (1.4). I find that the probability of

a write-of occurring increases if a CEO turnover with an outside replacement occurs within the

one-year period prior to the write-of (significant at 5 percent). The control variables in Equation

(1.4) have the right signs. The results confirm a significant positive relation between the probabil -

ity of a write-of and the size of a firm. There is a negative correlation between performance and

the write-of decision. ROA is negatively related to write-ofs, while TLCF is positively related to

write-ofs. I also find that CEO shareholdings are related to the write-of decision, despite the fact

that the CEO turnover and the percent ownership are strongly negatively related. In addition,

debt is negatively related to the write-of decision. Table 6, Model 1, shows the marginal efects for

the probit model. The share ownership, the size, and CEO replacement from outside the company

have the largest impact on the write-of decision, respectively. In another words, the larger the

firm, the more likely it is that it will take a write-of. The impact of firm size on the write-of

decision is similar to the results in Meyer and Strong (1989). In addition, the results are consistent

with companies undergoing a period of poor performance. These results are robust to time period

specific random efects.





17
1.4.2 Executive Compensation and the Write-of Decision


It has become common practice for executive compensation to be tied to the company's perfor-

mance. Coughlan and Schmidt (1985), Murphy (1985, 1986), Abowd (1990), Jensen and Murphy

(1990) and Leonard (1990) study the relation between executive compensation contracts, incen-

tives and firm performance. These papers show that firm performance is largely positively related

to pay-performance sensitivity, after controlling for the risk, i.e., the variance of performance (Ag-

garwal and Samwick, 1999). Audt, Cready, and Lopez (2003) find that after controlling for the

growth in annual in?ation adjusted CEO cash compensation, CEOs are not protected from the

adverse efects of charges on earnings on their own utility.



If a CEO's actions are closely tied to firm performance, then the CEO will hesitate to

take unnecessary actions that afect his compensation. Therefore, it is plausible the CEOs with

high pay performance sensitivity will not take a write-of unless it is necessary to improve future

performance. There is a trade-of between short-term and long-term utility for the CEO. In the

short term, write-ofs can reduce stock price, which can reduce compensation. In the long term,

write-ofs can improve future performance, which can increase compensation. The future benefits

would dominate if there were a longer horizon for the CEO (e.g. for a new CEO), or compensation

is more dependent on future performance (e.g. stock options).



Formally stated:

H2Executive Compensation Hypothesis)The probability of taking a write-of is pos-

itively related to the pay-performance sensitivity of a CEO. The probability of taking

a write-of is also positively related to the actual compensation package.



To test this possibility, I use two diferent measures of compensation: actual compensation,

and pay-performance sensitivity. I use the following probit model to test hypothesis H2. I observe

a write-of is one if latent variable WO>0 and no write-of if WO s 0. WO
i
is empirically specified


18
as:



pr(W O) =|
1
+|
2
SIZE
i
+|
3
SALARY
i
+|
4
BON U S
i
+|
5
SHROW N P C
i
+|
6
OP T ION S
i



+|
7
RET Y RS
i
+|
8
ROA
i
+|
9
T LCF
i
+|
10
DEBT RAT IO
i
+
i
, (1.5)


where i is standard normal. I control for the size of the company and the tenure of the CEO

(RETYRS), as well as performance. I expect that higher compensation will be positively related

to write-ofs. I also expect that the tenure of the CEO will be negatively related to write-ofs. This

could be either because the CEO is entrenched, or because the CEO has not made any mistakes

and has no need for write-ofs.



Table 5, Model 2 (A), shows the estimates of Equation (1.5). CEOs with lower salaries are

more likely to take write-ofs than are CEOs with higher salaries (coefcient = -0.001, significant

at 5 percent), and CEOs with a greater percentage of shares are more likely to take write-ofs (co-

efcient = 4.23, significant at 5 percent). These results suggest that compensation packages, which

tie CEO incentives to performance, are related to write-ofs. The control variables in Equation

(1.5) have the right signs. The market value of a firm is positively related to the write-of decision,

while the tenure of a CEO is negatively related to the write-of decision. ROA is negatively related

to write-ofs and TLCF is positively related to write-ofs. In addition, the debt ratio is negatively

related to the write-of decision. These results suggest that CEOs who are less entrenched are

more likely to take write-ofs.



As discussed above, the pay-performance sensitivity might have implications in a write-of

decision. By using the following probit model, I test whether the pay-performance sensitivity of

managers and the level of entrenchment afect a company's write-of decision. I observe a write-of

is one if latent variable WO>0 and no write-of if WO s 0. WO
i
is empirically specified as:





19
pr(W O) =|
1
+|
2
SIZE
i
+|
3
IN T ERLOCK
i
+|
4
RET Y RS
i
+|
5
P P S
i
+|
6
ROA
i



+|
7
T LCF
i
+|
8
DEBT RAT IO +
i
, (1.6)


where i is standard normal and PPS is the degree of pay-performance sensitivity. I run

the regression using both dollar and return sensitivity as a measure of PPS, as discussed in Core

and Guay (1999a, 1999b). INTERLOCK is a dummy variable equal to one if the management is

entrenched and zero if it is not, as defined by Execucomp. Entrenchment generally involves one of

the following situations: the ofcer serves on the board committee that makes his compensation

decisions, or serves on the board (and possibly compensation committee) of another company that

has an executive ofcer serving on the compensation committee of the indicated ofcer's company.

Hallock (1997) describes the use of this variable as a proxy for managerial entrenchment. RETYRS

is the number of years the CEO has been in ofce. I expect that INTERLOCK and RETYRS will

be negatively related to write-ofs.



Table 5, Model 2 (B) shows the estimate of Equation (1.6) for the period 1990-2000. The

results indicate a positive relation between the probability of a write-of and the pay-performance

sensitivity of the CEO(t-value = 1.95, significant at 10 percent).
9
As before, the control variables

in Equation (1.6) have the right signs. Market value, and TLCF are positively and significantly

related to write-ofs, while the entrenchment variable, debt ratio, and ROA are negatively related

to write-ofs. Table 6, Model 2 (B), shows the results of the marginal efect of the probit estimation.




1.4.3 Monitoring Mechanism Hypothesis


The board of directors decides on both CEO compensation packages, and CEO turnover replace-

ments. In addition, the board of directors acts as a monitoring mechanism for CEOs. If the board

9
Table

5 shows the estimation for the return sensitivity measure. Results for the dollar sensitivity measure were
comparable.



20
is a proficient monitor, then there are fewer agency issues with management and the CEO has bet-

ter incentives to take actions that benefit the company and shareholders. When non-performing

assets afect a company, then a write-of is a tool that management can use to alleviate these op-

erational problems. If a link exists between performance and monitoring mechanism quality, then

it is plausible that companies with boards that are good monitors will have a higher probability

of taking a write-of.



The governance literature finds strong evidence that board composition (size of board and

percentage of insiders on the board) is related to the degree of agency problems (Byrd and Hickman

(1992); Wasatch (1988); Borokovich, Parrino, and Trapani (1996); Bhagat and Black (1999); Core,

Larcker, and Holthausen (1999); Hermalin and Weisbach (1991); and Yermack (1996). Larger

boards with more inside directors tend to have more agency problems. Conversely, firms with

small boards and a high percentage of outsiders will be more concerned about shareholder welfare

and firm performance.



Formally stated:

H3: The probability of a write-of increases when there are quality governance mech-

anisms in place (Monitoring Mechanism Hypothesis).



The size and percentage of outsiders on the board act as proxies for monitoring quality. I

also include the number of options and the percentage of equity that the directors own. If the

board's compensation is attached to the performance of the company, the incentive to be quality

monitors increases, as discussed in Mayers, Shivdasani, and Smith (1994). I also include the gov-

ernance index, GOV INDEX. Gompers, Ishii, and Metrick (2003) create a governance database

drawn from Investor Responsibility Research Center (IRRC) publications, an organization that

has tracked the provisions for about 1,500 firms per year since 1990. I merge the write-of sample

to the governance database using ticker symbols and year. G A higher GOV INDEX indicates a



21
firm with less shareholder rights. GOV INDEX was available for 756 write-of events.



Following Hypothesis 3A, I estimate the following probit specification:



pr(W O) =|
1
+|
2
SIZE
i
+|
3
SIZEBD
i
+|
4
DIROT P
i
+|
5
GOV IN DEX
i
+|
6
P ERC OU T
i



+|
7
DIROSK
i
+ beta
8
ROA
i
+|
9
T LCF
i
+|
9
RET Y RS
i
+|
10
DEBT RAT IO +
i
, (1.7)


where SIZEBD is the size of the board of directors, PERC OUT is the percent of outside directors,

GOV INDEX is the degree of shareholder protection, DIROPT is the value of options owned by

the directors, and DIROSK is the percent stock ownership of the directors. Since



Table 5, Model 3, shows the results of the test for whether better monitoring boards are

more likely to take write-ofs. The results indicate that boards that are more independent have

a greater likelihood of taking a write-of. SIZEBD is negatively related to the probability of a

write-of (significant at 5%), and the percentage of outsiders is positively related to the probability

of a (write-of significant at 5%). GOV INDEX is negatively related to the probability of taking a

write-of (significant at 5%). The percent of directors' option ownership is positively related to the

likelihood of a write-of, while the number of shares is not significantly linked to the tendency to

take write-ofs. These results suggest that firms with smaller boards, more outside directors, and

shareholder protection are more likely to take a write-of. The coefcients of the control variables

are consistent with the expected signs. The market value of the company (SIZE), and TLCF are

positively related to the write-of decision, while ROA, debt ratio, and RETYRS are negatively

related to the write-of decision. The signs and significance of ROA and TLCF are consistent with

companies having poor performance both in the write-of quarter, and in recent past quarters. The

options owned are positively related to the tendency to take write-ofs, while the percent stock

ownership is negatively related to the tendency to take write-ofs. Table 6, Model 3, looks at the

marginal efects of the independent variables on the write-of decision. Overall, these results con-



22
firm that companies with desirable board composition and strong shareholder protection measures

have a tendency to take write-ofs.




1.4.4 Multivariate Analysis


In the previous sections, I test the one-on-one relations between CEO turnover and write-ofs,

pay-performance sensitivity and write-ofs, board composition, shareholder protection, and write-

ofs. In this section, to examine the interaction of the independent variables, and to test which

characteristics are of the most importance, I combine these separate specifications.



In addition to firm characteristics described above, I include an industry shock variable.

Industry shocks and recessions are two possible factors that can afect a write-of decision. By

including these variables in the probit estimation, I can test whether these factors are related to

the write-of decision.



Using the Bartelsman and Gray (2002) dataset on the NBER website, I create a measure

of industry demand. I use the industry shipments at the 4-digit SIC code level de?ated by a 1987

industry price de?ator, and then aggregate this data at the 3-digit SIC code level. I detrend the

data by regressing the actual value of industry shipments on a yearly time trend variable. I then

calculate the industry shock is then calculated as the diference between the predicted and the ac-

tual value of shipments. I use the detrended real industry shipments for the same reasons cited in

Maksimovic and Phillips (2002). The reasons include the growth of an industry, which afects the

value of the capital in the industry, and firms' cash constraints can depend on industry conditions.



To test whether the market's reaction to the write-of is in?uenced by the relationship

between productivity and segment growth, I create dummy variables for recessionary and expan-

sionary periods. I also classify years as recession or expansion years for an industry, and determine



23
the recession and expansion years by looking at the relationship between the aggregate, and the

aggregate detrended production in each industry. If these two variables are negative, then the year

is a recession year. If they are both positive, it is an expansion year. Detrended production is the

actual production minus the predicted production, where the predicted production is calculated

as the production regressed on a time variable.



Table 5, Model 4, looks at the results of the following two regressions, which examine the

combined impact of the firm quality characteristics while controlling for industry efects.



P (W O) =|
1
+|
2
SIZE
i
+|
3
SHOCK
i
+|
4
÷
8
GOV V ARS
i
+|
9
÷
12
CON T ROLV ARS
i
+
i
, (1.8)





P (W O) =|
1
+|
2
SIZE
i
+|
3
RECESSION
i
+|
4
EXP AN SION
i
+|
5
÷
9
GOV V ARS
i



+|
10
÷
13
CON T ROLV ARS
i
+
i
, (1.9)


where GOVVARS are CEO turnover with external replacement(using IMR to control for endogene-

ity), pay-performance sensitivity, board size, percent of outsiders on the board, and shareholder

protection. I also include several control variables, such as percent of shares owned, RETYRS,

CEO tenure, ROA, TLCF, and an interlocking relationship dummy.



The results in Table 5, Model 4 show that if a negative shock afects the firm, the probability

of a write-of increases, especially when I control for governance quality (significant at 5 percent).

Likewise, a recession year for the company increases the probability of a write-of occurring (signifi-

cant at 5 percent), while an expansion year is negatively related to the write-of decision. Equations

(10) and (1.9) permit me to simultaneously examine the impact of the firm characteristics on the

write-of decision. Shareholder protection, board size, and percent of outside directors continue

to remain significant. CEO turnover, and pay-performance sensitivity do not have as significant

a role when combined with the other governance factors. One reason pay-performance sensitivity


24
may not be significant is because GOV INDEX encompasses compensation plans.



These results suggest that well governed companies that operate in industries impacted by

negative shocks or recessions have a tendency to take write-ofs. Table 6 shows the marginal ef-

fects for the combined probit estimation of Model 4 (A) and (B). I find that for the independent

variables conditional on the write-of decision, the number of outsiders on the board and the size

of the company have the greatest impact on the write-of decision. The size of the board and the

level of shareholder protection also have a highly significant efect on the probability of a write-of.
10




1.4.5 Weak Shareholder Protection and Write-ofs


So far, I have found evidence that companies with strong monitoring mechanisms have a tendency

to take write-ofs. However, these results do not answer any questions about weakly monitored

companies. It seems plausible that there is also a link between weak monitoring mechanisms and

write-ofs. Companies with less efective governance structures continually collect problems and

only take write-ofs when there is no other alternative. An example of a poor governance company

and write-ofs is Tyco Corporation. In an efort to hide slowing growth in its core divisions, Tyco

kept on diversifying into new areas. These diversification strategies were not successful, and so it

would diversify into yet another area. Eventually the problem became so huge that Tyco was left

with little alternative other than to take a write-of (Symonds, 2002).



I first determine firms with weak shareholder protection measures that take write-ofs. I

break the sample into three segments: weakly monitored governance companies, neutral gover-

nance companies and strong governance companies. I break the sample into three categories based

on their GOV INDEX. I sort the sample by GOV INDEX and label the lowest 10 percent of the

sample well monitored, and the top 10 percent weakly monitored companies. Table 7, Panel A

10
I

also include the industry efects in this estimate, but do not find any significant results. This is because I only
focus on one main industry - the consumer-manufacturing sector.



25
shows the results of this segmentation. The t-values test for whether there is a diference between

the average sizes of good versus bad governance characteristics. It becomes evident that there

is wide dispersion between the write-of firms in the sample, based on governance. The weakly

monitored companies have an average board of 18 people versus the good governance companies

with an average board size of six people (significant at five %). In addition, the poorly monitored

companies have significantly fewer outsiders on the board, and have significantly worse shareholder

protection measures (t-value=2.98 and 2.31, respectively.)



To formally test whether there is a relationship between poor governance and write-ofs,

I isolate worst 50 percent governance firms in my sample, both write-of firms and benchmarks,

based on GOV INDEX. To do so, I sort the sample based on the GOV INDEX, and then drop the

top 50 percent governance firms in the sample. I then re-estimate Equation (10) with only the bad

governance firms. Table 7, Panel B, shows the results. The most important factors in determining

whether bad governance companies take write-ofs are shareholder protection measures, and board

size. Both GOV INDEX and BDSIZE are positively and significantly related to the write-of deci-

sions. The other governance variables show the predicted signs but are not significant. The control

variables show the predicted signs discussed in earlier sections.




1.4.6 Governance and Size of Write-ofs


I have found evidence that suggests both well and poorly monitored companies are subject to

write-ofs. Even the best corporations are not immune from mistakes. However, these good gov-

ernance companies quickly recognize the mistake, and take actions to repair the problem. If this

is true, then it is expected that write-ofs would be relatively small for these quality companies.

Poorly monitored companies, on the other hand are slow to admit to mistakes, and even slower

to take actions to improve the situation. They collect mistakes until it becomes inevitable that a

write-of should occur. Following this argument, it is plausible that poorly monitored companies



26
will have relatively large write-ofs.



In this section, I test whether the size of a write-of is in?uenced by the quality of the

governance. If this is the case, then it supports the story that well governed companies are first to

repair problems, whereas poorly monitored companies are reluctant to repair problem areas. Table

8, Panel A, shows the univariate results of write-of size, segmented by write-of quality. I segment

the sample into three diferent groups: companies with weak monitoring structures, companies

with average governance, and companies with strong governance. I sort the companies on the

following items: GOV INDEX, board size, and percent of outsiders. I denote weak as the bottom

10 percent of write-ofs, and strong as the top 10 percent of write-ofs. I adjust write-ofs by the

total assets of a company. The average size of all write-ofs is 0.03. The average adjusted size of

well-monitored companies' write-ofs is 0.02, versus 0.06 for poorly monitored companies. I regress

the size of the write-of on the following Equation to test whether governance afects write-of size:



W O/T A =|
1
+|
2
÷
5
CON T ROLV ARS
i
+|
5
÷
10
GOV V AR
i
+
i
, (1.10)



where GOVVARS are CEO turnover with external replacement, pay-performance sensitivity, board

size, percent of outsiders on the board, and GOV INDEX. CONTROLVARS are MV, ROA, and

TLCF.



Table 8, Panel B, shows the results of this regression, which controls for heteroscedascity

in standard errors. The results suggest that companies with larger boards take larger write-ofs

(t-value = 2.76). Likewise, companies with more outside directors also take larger charges (t-value

= 1.98). Although not significant, the results also suggest that companies with worse shareholder

protection measures and lower PPS also take larger write-ofs. These results show that compa-

nies with worse governance take larger write-ofs, while good governance companies take small

write-ofs. The control variables show the expected relationship to write-ofs. These results are

consistent with the story that good governance companies are first to act when problems arise,


27
hence the size of the write-of is smaller. Bad governance companies wait to take write-ofs and

collect problems over an extended period, hence the size of the write-of is comparably larger.





1.5 Market Reaction and Write-of Announcements


Having shown that corporate governance impacts in what manner write-ofs are used, I now exam-

ine how investors react to write-ofs, taking into account the quality of governance of the announcing

company.



Table 9 looks at the abnormal returns surrounding the write-of announcement for the one-

time and multiple events, and for the combined sample. For the full sample, the average market

reaction to write-of announcements is -1.10 percent, and is not significant. These findings are

comparable with those in earlier studies, e.g. Meyer and Strong (1987), Elliot and Shaw (1988),

Bartov, Lindahl, and Ricks (1998). When looking at the pooled average of 15 years, only one-time

write-ofs show any significant abnormal returns. The one-time write-ofs display a significant -1.82

percent return around the announcement day (significant at 5%). I interpret this result as meaning

that the market only considers first-time write-ofs to be significant to the company's performance.



I look at the combined impact of CEO turnover, pay-performance sensitivity, and board

composition on the announcement efects of write-of firms:



AR
i
=|
1
+|
2
SIZE
i
+|
3
SIZEBD
i
+|
4
P ERC OU T
i
+|
5
CEO OU T
i




+|
6
RECESSION
i
+|
7
P P S
i
+|
8
ROA
i
+|
9
W O T A
i


+|
10
W O #
i
+|
11
T Y P E
i
+|
12
DEBT
i
+|
13
GOV IN DEX
i
+
i
. (1.11)


In addition to the governance variables I have used for the probit estimates, I include the



28
size of the write-of (W O T A), the type of write-of and the number of write-ofs a firm has taken

(W O #), which includes the current write-of (a first time write-of would be equal to one, etc.).

I would expect that larger write-ofs would have a more negative impact on returns. In addition,

I expect that companies with less write-of history will see a greater market reaction and stronger

monitoring mechanisms will lead to higher returns.



Table 10 shows the results of these estimates. I find that when controlling for negative

industry shocks such as recession, companies with strong governance measures actually experience

more positive abnormal return. Larger firms have a 0.2 percent increase in abnormal returns.

Strong shareholder protection leads to a 2.4 percent increase in abnormal returns. Small boards

lead to a 1.5 percent increase in abnormal returns. A larger percentage of outside directors leads

to a two percent increase in abnormal returns, while high PPS leads to a 2.6 increase in abnormal

returns. CEO turnover leads to a 0.2 percent increase in abnormal returns. However, larger write-

ofs lead to a one percent drop in returns. The debt ratio, the type of write-of, and the number

of write-ofs. In aggregate, companies with strong monitoring mechanisms have over 6 percent

abnormal returns following a write-of announcement.



Next, I consider companies with a write-of following CEO turnover. As Borokovich, Par-

rino, and Trapani (1996) show, the turnover announcement is normally succeeded by a positive

market reaction for outside replacement, and a negative reaction for insider replacement when the

replacement is not voluntary. These results give evidence that the market diferentiates between

inside and outside replacement, and that outside replacement are good for the future of the firm.

Hence, I test whether write-ofs generate similar reactions. Are write-ofs from CEO turnover

where the replacement is external associated with positive announcement day efects, or is the

write-of anticipated following the turnover? Table 11 segments the write-of announcement efects

by CEO turnover, and GOV INDEX. I find that there is more than a 6 percent positive write-of

announcement return for good governance companies that have had a recent CEO turnover using



29
an external replacement. These results suggest that an investor who owns a write-of company,

which has good governance and recently had a CEO turnover could make a six percent return over

the market.





1.6 Conclusion


This paper examines the relation between write-ofs and corporate governance measures. Com-

panies that have smaller boards with a higher percent of outside directors, stronger shareholder

protection measures, high pay-performance sensitivity, or CEO turnover are positively related to

the write-of decision. The write-ofs these good governance companies take are linked to industry

specific factors, such as industry shocks, or recessions. By segmenting out the good governance

companies from the sample, I am able to test whether there is a link between poor governance and

the write-of decision. I find that companies with poor shareholder protection measures, and large

boards are also positively related to the tendency to take write-ofs. In addition, lower quality

governance leads to larger write-ofs. Well-monitored companies are the first to act when they

realize that a problem has arisen and write-ofs are one tool that management can use to clean up

the problem area. Conversely, poorly monitored companies wait to amend the companies' prob-

lems, until the magnitude of the problem cannot be ignored. This explains why the size of the

write-ofs from poorly monitored companies is significantly larger than the size of write-ofs from

well-monitored companies.



I also look at the impact of write-ofs on investors. By segmenting the write-ofs based on

the governance quality, I determine whether investors diferentiate between the diferent compa-

nies taking write-ofs and the types of write-ofs. It becomes evident that companies with quality

monitoring mechanisms take write-ofs that result in a positive stock market reaction, while com-

panies with poor monitoring mechanisms take write-ofs that result in a negative stock market

reaction. The findings suggest that investors may understand the information content in write-



30
ofs, and are able to diferentiate between write-ofs that will improve performance and write-ofs

that will not. The stock market recognizes the quality of management and the board of directors

and an investor's reaction is based on this knowledge. By looking at the cross-sectional dispersion

of governance measures and based on the quality of the monitoring mechanisms, an announcement

day efect could have more than a six percent positive return in the short term. The evidence in

this paper indicates that companies with good governance use write-ofs in a way that is consistent

with enhancing shareholder value. In addition, I find evidence that is consistent with the idea that

governance matters.









































31
1.7 Tables

Table I Sample Information

Panel A shows, by year, the number of write-of announcements for layof based, asset based, and combined write-
ofs. Panel B shows the number of firms in the sample that take a first time write-of and
then breaks into the percent of these firms that go on to take another write-of. For instance, 61 percent
of first time write-of firms take a second write of, and 42 percent of first time write-of firms take two more write-ofs, etc.
Panel B also shows the breakdown of the types of write-ofs, whether they are layof
based, asset based, or a combination of the two.

Panel A: Number of Write-ofs by Year and Type
Year Asset Layof Both All
1985 20 4 25 49
1986 34 7 45 86 1987
48 6 48 102
1988 34 18 40 92
1989 34 10 45 89 1990
27 7 58 92
1991 40 18 51 109
1992 28 8 63 99 1993 43
20 72 135
1994 35 14 99 148
1995 36 18 91 145 1996
39 17 100 156
1997 42 27 99 168
1998 41 31 98 170
1999 42 22 121 185
2000 61 31 120 212
All 604 258 1175 2037
Panel B: Frequency of Write-ofs
Write-ofs # Firms Percent Layofs Assets Both
1 767 100 17% 37% 46%
2 468 61 14% 39% 47%
3 319 42 13% 34% 53% 4
216 28 9% 41% 50%
5 150 20 14% 33% 53%
6 116 15 11% 27% 62% 7
79 10 17% 15% 68%
8 61 8 10% 18% 72%
9 45 6 8% 28% 64% 10
31 4 7% 35% 58%
11 20 3 10% 35% 55%
12 16 2 23% 40% 38% 13
9 1 13% 20% 67%
14 7 1 14% 43% 43%
15 3 0.5 0% 67% 33% 16
3 0.4 0% 33% 67%
17 2 0.26 0% 0% 100%











32
Table II Write-of Characteristics of Sample

This table summarizes the diferent types of write-ofs that firms report. There are several diferent ways that a firm
can write employees and assets of the books. I discuss these methods in Appendix A. The average charge is the average
write-of the firm reported for each of the diferent types of write-downs.
Write-of/Book Value is the total charge divided by book value of shareholders equity one month prior to
the announcement. Write-of/Total Assets is the total charge divided by total assets one quarter prior to the write-of
announcement.


Write-of/Book Value Write-of/Total Assets
Type Percent Average Charge Mean Median Mean Median
Asset impairment charge 7.61 $67,100,000 0.02 0.01 0.09 0.02
Discontinued operations 14.28 $18,900,000 0.13 0.01 0.41 0.02
Layof charge 8.86 $69,700,000 0.03 0.01 0.08 0.01
Restructure(asset and layof based) 56.35 $72,100,000 0.06 0.01 0.23 0.03
Severance 4.09 $38,200,000 0.02 0.01 0.04 0.02
Partial write down 3.6 $78,800,000 0.05 0.01 0.17 0.03
Write-of of assets 5.22 $23,400,000 0.04 0.01 0.09 0.03
Total 2472 $59,400,000 0.04 0.01 0.13 0.02






































33
Table III Summary of Estimation Variables

This table shows the summary mean and medians of the independent variables used in the probit estimations.
Panel A is across the whole sample. MV is the size of the firm, defined as the log of the market value. SHROWNPC is the
CEO's percentage ownership in the company, shown as a percent. SALARY is
the dollar annual CEO salary in 100ks, BONUS is the dollar annual CEO bonus in 100ks, and OPTIONS
are the aggregate dollar value of all options granted to the executive during the year as valued by the company in 100ks.
DIRSTK is the number of shares, which each non-employee director received during
the year in thousands. SIZEBD is the number of directors, both inside and external. PER OUT is the
percentage of directors who have no relationship with the company. . GOV INDEX is the Governance index, the lower
the number the better the shareholder protection, and the higher the number, the higher
the level of managerial entrenchment. GOV INDEX is an index number formed from charter provisions
listed in the IRRC publications. It does not replace board composition. DOLLAR SENSITIVITY = is the dollar sensitivity
of compensation to performance. RETURN SENSITIVITY = is the return sensitivity
of compensation to performance. The significance test uses a two-sided test to determine whether there
is a statistical diference between the non write-of benchmark firms, and the write-of firms. Panel B shows the summary of
governance variables on a per year basis for all firms in the sample. It also shows
the percent of firms that decreased, remained unchanged , and increased their governance variables. A *
denotes significance at the 5 percent level, and ** denotes significance at the 10 percent level.

Panel A
Non-Write-of One-Time Write-of Multiple Write-of
Mean Median Mean Median t-value Mean Median t-value
MV 5.25 5.64 9.64 10.53 13.92* 12.490 13.45 5.62*
SHROWNPC 2.00 0.00 4.00 0.00 1.94** 3.00 0.00 0.22
SALARY 666.21 650.00 629.55 599.07 1.130 657.95 667.51 0.32
BONUS 848.11 413.27 534.5 351.00 2.24* 548.31 439.46 3.67*
OPTIONS 1,346.08 353.44 2,385.19 451.26 1.99** 2,039.56 663.81 0.89
DIRSTK 0.06 0.00 0.05 0.00 0.350 0.13 0.00 1.07
SIZEBD 12.05 12.00 10.17 10.00 -11.97* 11.05 11.00 -3.67*
PERC OUT 68.00 70.00 73.00 75.00 -2.87* 76.00 78.00 -7.11*
GOV INDEX 9.06 10.00 8.10 9.00 1.99** 9.06 10.00 2.62*
RETURN SENSITIVITY 3,286.00 61 20,948.00 57.00 -2.98* 8,842.00 68.00 -1.91**
DOLLAR SENSITIVITY 231.00 43.00 1,972.00 34.00 -2.33* 864.00 55.00 -1.98**
ROA 7.3 7.33 4.25 4.38 3.85* 4.7 5.83 3.71*
Panel B

GOV INDEX Board Size Percent Outsiders
Year Mean -1 0 1 Mean -1 0 1 Mean -1 0 1
1990 9.66 2% 93% 5% 11 9% 68% 23% 0.73 7% 66% 26%
1991 9.6 3% 94% 3% 11.86 11% 74% 15% 0.76 10% 65% 26%
1992 9.62 2% 94% 4% 12.51 7% 86% 7% 0.74 5% 64% 30%
1993 9.52 7% 78% 15% 11.52 10% 81% 9% 0.76 7% 64% 29%
1994 9.6 2% 97% 1% 11.76 8% 75% 16% 0.76 6% 64% 30%
1995 9.46 7% 76% 18% 12.24 10% 75% 14% 0.77 9% 57% 34%
1996 9.24 2% 96% 2% 11.58 12% 71% 17% 0.75 10% 57% 34%
1997 9.63 2% 95% 2% 11.83 14% 64% 22% 0.75 16% 62% 22%
1998 9.13 5% 83% 12% 11.83 18% 68% 14% 0.75 11% 64% 26%
1999 8.95 2% 96% 2% 10.61 13% 74% 12% 0.76 10% 69% 21%












34
Table IV CEO turnover and the Write-of Decision

Panel A shows the characteristics of executive turnover associated with a write-of announcement. The turnover
data is from EXECUCOMP. If the write-of occurs up to one year after the CEO turnover, I include the turnover in the
sample. There are two types of turnovers shown; those that occur with
the replacement coming from inside the firm; and those with replacements coming from outside the firm.
The table shows the percentage of non write-of, one-time write-of, and multiple write-of firms that have experienced CEO
turnover. The t-value tests the hypothesis that write-ofs have statistically significant
diferent number of CEO turnovers to the industry benchmark. Panel B shows the probit estimation of
CEO turnover with external replacement controlling for ROA, market value, and ROE. The results of this estimate are used
to calculate the Inverse Mill's Ratio.

Panel A
Non-Write-of One-Time Write-of Multiple Write-of
Reasons Inside Outside Total Inside Outside Total Inside Outside Total
Retires 0% 0% 0% 2% 7% 9% 2% 8% 10%
Resigns 0% 0% 0% 0% 2% 2% 2% 0% 2%
Dies 0% 0% 0% 0% 2% 2% 0% 0% 0%
Not Listed 23% 77% 100% 17% 70% 87% 31% 57% 88%
Total Number of Firms 8 27 35 11 48 59 46 85 130
t -value -1.99** -2.07*
Panel B

Coef. T-value
Constant -1.07 -5.65 * MV
-.15 -2.01*
R -0.03 -3.66 *
ROA -0.02 -2.96*































35
Table V Probit Model Results

Panel A shows the results of the probit estimations for various models. Model 1 tests for a relation between CEO
turnover and the write-of decision as shown in Equation (5). Model 2 (A) tests the probit estimate shown in Equation (6).
Model 2 (B) tests the probit estimation discussed in Equation (7). Model
3 shows the results for the estimation of Equation (8), which tests for the relationship between the board's
independence and the probability of a write-of occurring. Model 4 (A) tests the probit estimate of the impact of industry
shocks on the write-of decision, while controlling for the quality of the company, as
described in Equations (9) and (10). MV is the size of the firm; SHROWNPC is the percentage ownership
in the company. CEO OUT is a dummy variable, zero for firms without turnover, and one for firms with turnover and
replacement from outside the firm. ROA is the return on assets, and TLCF is a dummy
variable that is one if a firm has tax loss carry forwards. SALARY is the annual CEO salary, BONUS
is the annual CEO bonus, and OPTIONS are the aggregate value of all options granted to the executive during the year as
valued by the company. INTERLOCK is one if the management is entrenched and zero
if it is not, as defined by Execucomp. Entrenchment generally involves one of the following situations:
the ofcer serves on the board committee that makes his compensation decisions, or serves on the board (and possibly
compensation committee) of another company that has an executive ofcer serving on the
compensation committee of the indicated ofcer's company. Hallock (1997) describes the use of this
variable as a proxy for managerial entrenchment. RETYRS is the number of years the CEO has been in ofce. PPS is
either the return sensitivity or the dollar sensitivity. The results shown are for the
dollar sensitivity, although I perform the analysis for both measures. SIZEBD is the number of directors,
both inside and external. PER OUT is the percentage of directors who have no relationship with the company. DIROPT
is the number of options, which each non-employee director received during the year
in thousands. DIRSTK is the number of shares, which each non-employee director received during the
year in thousands. GOV INDEX = Governance index, the lower the number the better the shareholder
protection. A * denotes significance at the 5% level, and ** denotes significance at the 10 percent level.

Independent Model 1 Model 2(A) Model 2 (B) Model 3 Model 4 (A) Model 4 (B)
Constant -2.54 -3.49 * -3.46 * -5.65 * -7.12 * -7.20 *
MV 0.41 * 7.37 * 0.59 * 0.58 * 0.81 * 0.78 *
CEO OUT 0.04 * 1.13 * -1.36 * DEBT RAT -1.77 * -2.71 *
-2.36 * -0.44 * -5.98 * -5.01 *
SHROWNPC -0.63 ** 4.23* 3.21 1.62
ROA -0.14 * -0.07* -0.08 * -0.06 * -0.96 * -0.09 *
TLCF 0.99 * 0.08* 1.38 * 1.07 * 3.57 * 3.49 *
SALARY -0.01 *
BONUS 0.00
OPTIONS -0.02
PPS 0.01 * 0.91 0.01
RETYRS -0.02 * -0.29 -0.03
INTRLOCK -1.53 * -4.83 * -4.74 *
SIZEBD -0.08 * -0.27 * -0.24 *
PERC OUT 1.84 * 4.42 * 4.49 *
DIROTP 0.10 * 0.33 * -0.02 *
DIRSTK -0.08 -0.01 -0.01
GOV INDEX -0.03 * -0.21 * -0.16 *
SHOCK 0.17 *
RECESSION 0.58 *
EXPANSION -0.12 _
2
(d.f.) 451.90 417.00 370.00 629.00 438.00
385.00










36
Table VI Marginal Efects of Probit Estimates

This table shows the marginal efects of the probit estimations for various models. Model 1 tests for a relation
between CEO turnover and the write-of decision as shown in Equation (5). Model 2 (A) tests the probit estimate shown
in Equation (6). Model 2 (B) tests the probit estimation discussed in Equation
(7). Model 3 shows the results for the estimation of Equation (8), which tests for the relationship between
the board's independence and the probability of a write-of occurring. Model 4 (A) tests the probit estimate of the impact
of industry shocks on the write-of decision, while controlling for the quality of
the company, as described in Equations (9) and (10). MV is the size of the firm; SHROWNPC is the
percentage ownership in the company. CEO OUT is a dummy variable, zero for firms without turnover, and one for firms
with turnover and replacement from outside the firm. ROA is the return on assets, and
TLCF is a dummy variable that is one if a firm has tax loss carry forwards. SALARY is the annual CEO
salary, BONUS is the annual CEO bonus, and OPTIONS are the aggregate value of all options granted to the executive
during the year as valued by the company. INTERLOCK is one if the management is
entrenched and zero if it is not, as defined by Execucomp. Entrenchment generally involves one of the
following situations: the ofcer serves on the board committee that makes his compensation decisions, or serves on the
board (and possibly compensation committee) of another company that has an executive
ofcer serving on the compensation committee of the indicated ofcer's company. Hallock (1997) describes
the use of this variable as a proxy for managerial entrenchment. RETYRS is the number of years the CEO has been in
ofce. PPS is either the return sensitivity or the dollar sensitivity. The results shown
are for the dollar sensitivity, although I perform the analysis for both measures. SIZEBD is the number of
directors, both inside and external. PER OUT is the percentage of directors who have no relationship with the company.
DIROPT is the number of options, which each non-employee director received during the
year in thousands. DIRSTK is the number of shares, which each non-employee director received during the
year in thousands. GOV INDEX = Governance index, the lower the number the better the shareholder
protection. A * denotes significance at the 5% level, and ** denotes significance at the 10 percent level.

Independent Model 1 Model 2(A) Model 2 (B) Model 3 Model 4 (A) Model 4 (B)
MV 0.05 * 0.06 * 0.01 * 0.05 * 0.07 * 0.07 *
DEBT RAT -0.20 * -0.03 * -0.04 * -0.01 * -0.02 * -0.03 *
CEO OUT 4.49 * 0.02 * 0.03 ** SHROWNPC
0.07 * 0.57 * 0.47 0.35
ROA -0.02 * -0.07 * -0.08 * -0.06 * -0.08 * -0.09 *
TLCF 0.11 * 0.08 * 0.89 * 1.07 * 1.15 * 1.52 *
SALARY 0.01 *
BONUS -0.03
OPTIONS -0.02
PPS 0.09 * 0.01 0.00
RETYRS 0.00 * 0.00 0.00
INTERLOCK -0.13 -0.19 0.61 * SIZEBD -0.01 *
-0.01 * -0.01 *
PERC OUT 0.22 * 0.29 * 0.30 *
DIROTP 0.01 ** 0.00 ** 0.00 **
DIRSTK -0.01 0.00 0.00
GOV INDEX -0.01 * -0.01 * -0.01 *
SHOCK -0.01 *
RECESSION 0.01 **
EXPANSION -0.02












37
Table VII Corporate Governance Measures and Write-ofs

Panel A shows the univariate results for the governance quality of the weakest and strongest gov- ernance write-of
firms. The significance tests whether good governance variables are diferent from bad governance variables. Panel B
shows the estimation of Equation (9) for the 50 percent worst governance
firms. A * denotes significance at the 5% level, and ** denotes significance at the 10 percent level.

Panel A: Strong Monitors vs. Weak Monitors
Weak Monitors Strong Monitors t-value
Mean Median Mean Median t-value
BDSIZE 18.00 18.00 6.05 6.00 36.01*
PERC OUT 0.46 0.48 0.80 0.75 2.98*
GOV INDEX 10.12 11.00 5.94 6.00 2.31*

Panel B: Multivariate Analysis of Governance
Independent Model 4 (A) t-values
MV
CEO OUT
SHROWNPC
ROA
TLCF
PPS
RETYRS
INTERLOCK
SIZEBD
PERC OUT
DIRSTK
GOV INDEX
SHOCK
1.35
0.02
0.47
-0.05
2.63
0.00
-0.04
-0.19
2.35
-0.03
0.38
0.47
0.00
2.88*
0.06 0.35
-0.65
1.91**
-0.37
-0.98
0.61
2.01*
-0.99
1.38
2.69*
0.09






























38
Table VIII Corporate Governance Measures and Write-ofs

Panel A shows the univariate results for the size of the write-of charges based on governance quality. The
significance tests whether good governance charges are diferent from bad governance charges. Write-
of charges are adjusted by the total assets. Panel B shows the estimation of Equation (12),

W O/T A =|
1
+|
2
÷
7
GOV V ARS
i
+
i
. (12)

This robust regression tests whether governance afects the size of the write-of. A * denotes significance at the 5% level,
and ** denotes significance at the 10 percent level.

Panel A: Univariate Estimate
Mean Median t-value Median test
Well-monitored 0.02 0.01
Poorly-monitored 0.06 0.03
1.53 0.09**
Panel B: Robust Regression
Coefcient t-value
MV 0.01 2.21*
ROA -0.03 -0.9
TLCF 0.04 1.92**
SIZEBD 0.02 2.76**
PER OUT
CEO OUT
GOV INDEX
PPS
CONSTANT
R
2
-0.18
-0.03
0.02
0.01
0.12
0.09
-1.98*
-0.40
0.25
0.18
0.91































39
Table IX Abnormal Return Breakdown, by Year and Firm Type

This table reports the breakup of the type of firm year by year, based on write-ofs in the period
of 1985-2000. I compute abnormal returns as AR
i
=
T +1
R
i,t
÷ R
s,i,t
, where R
i,t
is the return on date T ÷1
t for firm i, and R
s,i,t
is the return on date t, of the equally weighted index of the size portfolio s to which
firm i belongs. AR is reported in percentage format. t is the announcement date. A * denotes significance
at the 5% level, and ** denotes significance at the 10 percent level.

One-time Write-of Multiple Write-of All Firms
Year -1.00 0.00 1.00 Sum -1 0 1 Sum -1 0 1 Sum
1985 -0.29 -0.65 -0.36 -1.30 -0.95 -2.26 -1.74 -4.95 -0.62 -1.46 -1.05 -3.13
-.62 -2.28* -1.81
1986 -0.09 1.20 -0.81 0.30 -0.48 0.05 1.07 0.64 -0.26 0.29 0.05 0.08
-1.34 -0.56 -1.12
1987 -0.12 0.09 -0.70 -0.73 -0.47 -0.03 0.82 0.32 -0.59 0.15 0.28 -0.16
-1.95** -0.89 -0.56
1988 -0.19 -0.98 2.20 1.03 -0.09 -1.10 0.85 -0.34 -0.33 -0.72 0.57 -0.48
-1.69 -.65 -1.49
1989 -0.25 0.69 0.72 1.16 0.21 -0.28 -0.54 -0.61 -0.02 0.21 0.09 0.28
-0.51 -1.47 -0.41
1990 -1.04 -0.70 -1.93 -3.67 0.25 0.71 0.47 1.43 0.19 -0.55 0.08 -0.28
-1.69 1.95** -0.14
1991 0.74 -0.10 0.11 0.75 -0.41 -0.23 0.07 -0.57 -0.11 -0.07 -0.24 -0.42
-0.79 -1.10 -1.53
1992 1.46 -1.65 1.04 0.85 0.71 -0.26 0.52 0.97 -0.74 -0.50 0.45 -0.79
-1.29 -2.03 -0.48
1993 -0.09 -0.25 -0.29 -0.63 0.70 0.51 0.16 1.37 -0.59 0.24 0.11 -0.24
-1.06 -0.08 -1.53
1994 -0.05 -0.63 -0.55 -1.23 0.20 0.37 0.16 0.73 0.07 -0.13 -0.18 -0.24
0.00 -2.16* -1.05 -0.54
1995 -0.09 -2.55 0.66 -1.98 0.14 0.04 -0.30 -0.12 0.25 -0.14 -0.29 -0.18
-1.29 -.23 -0.79
1996 0.18 1.77 0.84 1.11 - 0.43 0.24 0.19 0.10 0.45 0.33 0.88
-.69 -.41 -1.72
1997 0.20 -1.04 0.14 -0.70 -0.19 -0.41 -0.44 -1.04 -0.13 -0.42 -0.15 -0.70
-1.27 -1.11 -0.56
1998 0.48 0.45 0.43 1.36 -0.22 0.35 0.45 0.58 -0.43 -0.20 -0.19 -0.82
-1.95** -0.99 -0.79
1999 1.52 -1.74 -1.09 -1.31 -0.13 -0.78 -0.02 -0.93 0.40 0.02 -0.50 -0.08
-1.82 -.84 -0.73
2000 0.15 0.58 -2.38 -1.65 0.15 -0.03 0.38 0.50 0.13 0.54 -0.50 0.17
-1.95** -0.49 -0.34
ALL -0.36 -1.06 -0.40 -1.82 -0.01 -0.19 -0.17 -0.37 -0.19 -0.63 -0.29 -1.10
-2.02* -0.49 -1.08
















40
Table X Market Reaction to Write-ofs

This table shows the combined impact of CEO turnover, pay-performance sensitivity, and board
composition on the announcement efects for Equation (13), which uses only the write-of firms:

AR
i
=|
1
+|
2
SIZE
i
+|
3
SIZEBD
i
+|
4
P ERC OU T
i
+|
5
CEO OU T
i

+|
6
RECESSION
i
+|
7
P P S
i
+|
8
ROA
i
+|
9
W O T A
i
+|
10
W O #
i
+|
11
T Y P E
i
+|
12
DEBT
i
+|
13
GOV IN DEX
i
+
i
. (13)
T +1
I compute abnormal returns as AR
i
=
T ÷1
R
i,t
÷ R
s,i,t
, where R
i,t
is the return on date t for
firm i, and R
s,i,t
is the return on date t, of the equally weighted index of the size portfolio s to which firm
i belongs. A * denotes significance at the 10 percent level, and ** denotes significance at the 10 percent
level.

Independent Variable
MV
ROA
DEBT RATIO
GOV INDEX
SIZEBD
PERC OUT
TYPE
WO NUM
RECESSION
WO TA
PPS
CEO OUT
CONSTANT

Coefcient
0.002
0.001
-0.030
-0.024
-0.016
0.020
-0.001
0.001
0.010
-0.011
0.026
0.002
-0.026

t-Value
2.51* -
0.16
-1.65
-2.63*
-1.97**
2.56 *
-0.16
1.39
1.99**
-3.41 *
11.49*
2.42*
-2.15*































41
Table XI Announcement Day Returns Sorted by Governance Measures

This table breaks up the announcement day efects into portfolios based on the governance quality, and CEO
turnover. There are ten portfolios of governance qualities, with 1 housing the best governance companies, and 10
housing the worst. The governance data is from IRRC, and the turnover data from
Execucomp. In parentheses is the t-values that test whether the returns are significantly diferent from
0. I compute abnormal returns as AR
i
=
T +1
R
i,t
÷ R
s,i,t
, where R
i,t
is the return on date t for firm T ÷1
I, and R
s,i,t
is the return on date t, of the equally weighted index of the size portfolio s to which firm I
belongs. A * denotes significance at the 5% level, and ** denotes significance at the 10 percent level.

One-Time Write-of Multiple Write-of
GOV Port No CEO Turnover CEO Turnover No CEO Turnover CEO Turnover
AR t-value AR t-value AR t-value AR t-value
1 1.98 2.65* 6.10 2.70* 1.40 2.41* 1.70 1.55 2
1.90 4.90* 1.50 2.91* 1.10 2.11* 2.20 1.99**
3 0.20 0.02 1.60 1.03 0.70 0.46 1.00 1.30
4 1.50 -0.02 1.90 1.80 -0.80 -1.11 0.40 0.30
5 -1.70 -0.83 -2.10 -1.79 1.10 2.66* -0.80 -1.18
6 1.70 1.18 2.00 0.61 0.00 0.01 0.70 0.78
7 -3.10 -2.21* 7.00 0.94 -0.30 -0.41 0.90 0.72
8 -0.50 -0.43 -0.60 1.03 0.30 0.46 0.70 0.62
9 0.30 0.26 1.50 0.01 0.80 0.95 0.40 0.33
10 0.00 0.09 -1.40 -1.22 -1.60 -1.95** 0.40 0.15



































42
Chapter 2

Write-ofs and Liquidity

















































43
2.1 Introduction

Write-ofs are fast becoming a prominent event in U.S. financial markets. The number of write-ofs for consumer
manufacturing firms increased from 1980 to 2000 by 140 percent. With this increased usage of
write-ofs, it has become increasingly important to understand what, if any, impact write-ofs have.
In this paper, I analyze the efect of write-of announcements on stock market liquidity. If there
is a high level of asymmetric information, spreads will increase to re?ect this knowledge gap. Likewise, if
there is a decrease in asymmetric information, spreads will decrease to re?ect this improved information environment.

Information asymmetry between investors and management can hurt a firm's market value (My- ers and Majluf,
1984). Informed traders thrive in a less transparent environment and profit more from
their private information, which creates an adverse selection problem for investors. O'Hara (2003) argues
that reducing the amount of hidden private information can favorably afect asset prices due to improved price-discovery
process and liquidity. Prior research provides a framework for this study of investigating
the relationship between write-of announcements and secondary market liquidity.

Write-ofs may convey specific information about operating performance and strategies. When firms announce
write-ofs, two types of information might be disclosed to the public. The write-of announce-
ment could uncover a problem not known to exist before the announcement. The announcement can also show how the
firm is taking actions to repair the problem area. This voluntary disclosure of information
has the potential to reduce information asymmetry by making private information acquisition more readily
available to potential traders. The spread could therefore be afected by a decrease in perceived adverse selection risk that
is not re?ected by an observable decrease in volumes.

I use several tests to determine what impact write-ofs have on secondary market liquidity. Using univariate
analysis, I compare the absolute and relative spreads before a write-of announcement days to
the write-of window and find a significant improvement in liquidity following a write-of announcement. I
run the same analysis for trading volume and total number of transactions and find that both increase fol- lowing the write-
of announcement. In addition, I test whether the liquidity impact of write-ofs is diferent
from any other announcement. I find that write-of announcements show a greater liquidity improvement
than earnings announcements. I also use multivariate analysis to test whether the liquidity efect (as shown by absolute
spread, relative spread, and total number of transactions) is robust to the inclusion of price,
volume, and volatility control variables. Both absolute and relative spreads decrease following a write-of
announcement. The number of transactions improves following a write-of announcement. Taken as a whole, the findings
demonstrate that write-of announcements generate a benefit to investors in the form
of improved liquidity.

Next, looking only at write-of firms, I test whether the liquidity benefit of write-ofs is greater for companies with
good corporate governance versus companies with bad corporate governance. Minnick
(2004) show that the market reacts diferently to write-of announcements, based on the quality of the company's
governance. If a company has efective monitoring mechanisms, then traders may trust the
quality of the information to a greater extent than the information from a poor governance firm, leading to
a greater reduction in the asymmetric information. I find that governance does afect the liquidity efects of write-ofs. I find
the number of transactions increases and spreads decrease more for high governance
firms versus poor governance firms consistent with a larger reduction in asymmetric information for high
governance firms.

Lastly, I decompose the bid-ask spread in order to measure changes in the adverse selection com- ponent resulting
from write-ofs. A reduction in information asymmetry, resulting from the write-of announcement should generate a
decrease in the adverse selection component of the spread. In fact, it is
the decrease in this component that is expected to produce both narrower spread and greater transactions
volume. The positive relation between adverse selection and bid-ask spreads is well documented in the literature. See
Brockman and Chung (1999) and He?in and Shaw (2000) for evidence of the inverse rela-
tion between adverse selection and secondary market liquidity. I find that adverse selection costs decrease
following a write-of, and that this decrease is greater for companies with stronger monitoring mechanisms.

These findings paint an economically intuitive picture of managerial and investor behavior in the secondary
market. Write-ofs convey private information that managers possess, but that outside market


44
participants do not observe. Market participants understand that write-ofs convey some information.
When the write-ofs occur, they enter the market, thereby decreasing the bid-ask spreads and increasing
number of transactions. This is especially important for companies with good monitoring mechanisms.
This liquidity provision dynamic is important because higher liquidity can lead to lower costs of capital and higher firm
values (see Amihud and Mendelson (1986), Barclay and Smith (1988), and Jacoby, Fowler,
and Gottesman (2000)). Secondary-market investors adjust spreads, adverse selection costs, and number
of transactions in a manner consistent with a reduction in information asymmetry.

The remainder of this paper is structured as follows. In section II, the methodology and data selection methods
are discussed. In Section III, the liquidity efects of write-ofs are tested by comparing average bid-ask spreads, trading
volumes, and non-trading days before and after write-ofs while control-
ling for the behavior of non -write-of firms. Section IV looks at corporate governance and write-ofs, while
Section V looks at the adverse selection costs, and Section VI concludes the paper.



2.2 Sample and Methodology

To generate my sample, I collect write-of information, focusing on NYSE listed companies from 1980 to 2000 in the
2000-2999 SIC code. Using the original company list, I search Lexis Nexis and Dow Jones News
Retrieval services for specific key words. For each company, I search for articles that match key words for
my sample. The key words I used are write down, write-of, restructure, charge against earnings, layofs, and severance.
When the query results in a match, I take the first article in the series of articles that refers
to a current write-of that the company is announcing. I use the date of the article as the announcement
date of the write-of. I obtain the following information from the article: the amount of the write-of; whether the write-of
was generated by assets, layofs, or both; the purpose of the write-of (restructure,
write-down, plant closing, etc.); the justification cited by the company; and whether the write-of amount
is stated on a before-tax or after-tax basis. The sample contains asset-based and layof-based write-ofs (see Minnick
(2004) for a more in depth description of the data collection process).

To ensure that write-ofs in my sample are not extensions of earlier events, I set an arbitrary standard under
which I assume that any write-of announcements occurring within six months of earlier
write-of announcements are related. This exercise is also performed for break of points of one month,
three months, four months, eight months, and twelve months. Although doing so afects the sample size, it
does not afect the analysis or findings. Therefore, I only describe results using the 6-month break point.

I determine which write-of is a first-time event or a subsequent event. To define multiple write-ofs, I need to
establish an arbitrary time interval. The standard most researchers use defines multiple write-ofs
as any write-of event that occurs within 16 quarters of a prior write-of event. To identify a company's first write-of, I look
at all write-ofs that occur during the first five years of the sample: 1980-1985. I
require an initial period of 16 fiscal quarters with no write-ofs before I add a firm to the sample. I denote
the write-of following this break as a first time write-of. Because the original sample begins in 1980, the first reported
write-of in the sample occurs in the first quarter of 1985. To test the sensitivity of this break
point, I also use five other quarter break points to define first time write-ofs, (8, 12, 18, and 20 quarters) to
separate consecutive write-ofs. My conclusions become more robust with the longer measures and weaken slightly with
the short-term definitions, and since the inference changes only marginally, I use 16 quarters.
After I identify the first time write-of for a company, write-ofs that follow are labeled as second, third,
fourth write-ofs, etc. These subsequent write-ofs must occur within 16 quarters after the prior write-of. If the write-of
occurs after 16 quarters, I label it as another first time write-of.
1
The data collection and
cleansing process leaves me with 230 companies that announced 1,075 write-ofs from 1985-2000.

To examine write-of company characteristics, it is important to have a benchmark to compare the write-of firms.
Out of the 390 NYSE listed firms in the 2000-2999 SIC codes for 1985-2000, there are 160
firms that have never had a write-of. I match the announcement date of each write-of to the 160 non write-of firms. This
results in 172,000 non write-of matched firms. I then average across these firms to
create a benchmark measure for every write-of event.


1
See Minnick (2004) for more details on the data collection.


45
The distribution of write-ofs over the 15-year sample period appears in Table 1. The number of
write-ofs more than doubles from 1985 to 2000 (25 versus 90 write-ofs). Write-ofs that combine both
assets and lay-ofs, such as restructuring, have the largest charges ($100.2 million on average), followed by
layofs ($68.7 million on average), and assets ($63.5 million on average). There is no clear trend in the charge amounts
over time. When adjusting write-ofs by total assets, no ratio is greater than five percent.

Table 2 provides summary statistics for all of the NYSE listed consumer manufacturing companies from 1993 -
2000 in the consumer manufacturing sector a month prior to the write-of. Approximately 25
percent of all the NYSE consumer-manufacturing firms took a write-of over my sample. Although write-
ofs are representative of the population, they tend to have lower price levels, trading volumes, and daily returns than the
non-write-of firms. Market capitalization is larger for write-of firms, as compared to
non-write-of firms. Twenty-one percent of the write-of companies in my sample have taken two write-ofs,
and 14 percent have taken three or more write-ofs.



2.2.1 Liquidity Data
Raw trading data is collected from the New York Stock Exchange Trade and Quote database (TAQ).
This database reports every round lot trade and every quote from 1993 onwards on the New York Stock exchange.
2
I
match the TAQ data to my write-of dataset using perm numbers. This leaves 594 remaining
write-ofs. The loss in data comes from excluding NASDAQ and AMEX firms from the sample and limiting
the write-ofs to 1993 to 2000.

I begin my analysis by looking at the liquidity trends surrounding write-of announcements, both in the short and
long terms. Using two separate measures for liquidity, I look at the trend from 500 trading days before the write-of to 500
trading days after the write-of. The variables for liquidity include relative
bid-ask spread, absolute bid-ask spread, turnover, and number of transactions. Relative spread is defined
as follows,

RSP
i,t
= 0.5 APAP÷ BP
i,t
) , i,t
- (
i,t
+ BP
i,t

(2.1)

where AP
i,t
is the closing ask price on day t for firm i, BP
i,t
is the closing bid price on day t for firm i,
and RSP
i,t
is the relative spread on day t for firm i. Absolute spread is defined as follows,


ASP
i,t
= AP
i,t
÷ BP
i,t
, (2.2)


where ASP
i,t
is the absolute spread on day t firm i. I filter out quotations for which the ask is smaller than or equal to the
bid price (crossed markets), as well as all spreads greater than $5.00 and spreads that
represent more than 20% of the quote midpoint (outliers). These filters afect less than one percent of the observations.

Turnover is defined as the total monthly volume divided by number of shares outstanding. The daily average
trading volume is from TAQ, and the shares outstanding are from Compustat. The number
of transactions is defined as the number of round lots(100 shares)available to trade at the bid price plus
the number of shares available to trade at the ask price as follows:

T ransaction = BidT ransactions + AskT ransactions (2.3)

Market makers reduce the number of transactions when they are wary of the informational environ- ment. Lower
transaction numbers give market makers an opportunity to adjust prices quickly. Bacidore,
Battalio, and Jennings (2002) suggest that each measure of liquidity is deficient in properly assessing
the level of liquidity. Having a composite measure is especially helpful in empirical analysis, especially if spreads and
transactions point in diferent directions. To alleviate this issue, I calculate a composite
measure of liquidity, called transaction/spread ratio as described below. This measure is similar to both

2
See Hvidkjaer (2004) for an explanation of the database, and the technique used to aggregate the data to daily data.


46
Bacidore et. al. (2002) and Jain, Kim and Rezaee (2004). It depicts market liquidity as a function of both
higher number of transactions and lower quoted spreads:

T /S = T ransactionN umber/AbsoluteSpread (2.4)

I analyze changes in liquidity using two tests. The first is a t-test comparing the cross sectional mean from the
pre-announcement period to the cross sectional mean after the write-of announcement.
The second, more powerful test calculates for each stock, the diference between the mean before the
announcement date and the mean after the announcement date. I then compare the frequencies of the increases and
decreases between the write-of and non-write-of firms using a chi-square test under the null
hypothesis that the relative frequencies are the same (Gibbons (1976)).



2.3 Liquidity Efects

Table 3 shows the summary statistics for the diferent liquidity measures for the write-of firms six months before and after
the write-of announcement, compared to 25 days following the write-of announcement. I
define the write-of period as 25 days following the write-of announcement, so the write-of window is t=0
to t=25. The non write-of window is defined as any time 120 days before the write-of announcement, or 120 days after the
write-of announcement. The period after one write-of is not mutually exclusive with
respect to the period before another write-of. Because there is no clear interpretation of before and after
periods, I rely only on the surrounding non write-of period as my benchmark.

I calculate the means and medians for various measures across write-of periods and surrounding non-write-of
periods for each sample firm. Table 3 provides summary statistics for the write-of and sur- rounding non-write-of periods,
along with paired t-test and sign test results. Volume is the total trading
volume per day. Price is the average daily price transaction, and Returns is the average daily return.
Volatility measures the variance of returns. Absolute Spread, Relative Spread, Total number of transac- tions, Ask number of
transactions and Bid number of transactions are daily averages for the absolute
dollar spread, relative spread, total number of transactions, ask-side number of transactions, and bid-side
number of transactions, respectively.

The univariate test results show that average daily trading volumes are 9 percent higher during the write-of
period, as compared to the surrounding non write-of period. The average stock price is 12 percent lower during the write-
of period as compared to the surrounding non-write-of period. The average
absolute (relative) spread of the write-of firms is $0.28 (1%) while the average absolute (relative) spread
of the surrounding non write-of period is $0.37(2%). The average number of transactions of write-of firms is slightly
higher than the surrounding non-write-of period. The liquidity measures yield strong evidence.
Spreads tend to decrease, thereby increasing liquidity, while turnover and number of transactions tend to
increase, which also increase liquidity. Overall, the write-ofs appear to have improved liquidity as com- pared to the
surrounding non write-of periods.

I next test whether this liquidity improvement is unique to write-ofs or if it occurs for any quarterly
announcement. To test for this unique reaction, I compare the liquidity changes the write-of announce-
ment to the liquidity changes of earnings announcements in the write-of quarter. Using a univariate
analysis, I compare the mean diference in the liquidity changes. The results are shown in Table 4. I find that the liquidity
improvement for write-ofs is significantly diferent than it is for earnings announcements.
These findings suggest that write-ofs have a greater impact on liquidity than other types of announcements.

I have established that actual write-of periods are associated with significant changes in price, volume, and
volatility. The univariate tests have also shown that liquidity improves in two ways, number
of transactions, and spread. I next focus on measuring the impact of write-ofs on liquidity after controlling for changes in
price, volume, and volatility. These three independent variables are widely used in the market
microstructure literature to control for the trading efects on firm liquidity. Tinic and West (1974), Benston
and Hagerman (1974), and Weston (2000) have shown that spread is positively related to share price. We include the
return volatility measure since the risk of the security is a component of dealer risk and dealer
inventory carrying costs. Several theoretical studies include risk as a factor that positively afects the


47
spread, including Garman(1976), Stoll (1989), and Ho and Stoll (1981).
Table 5 presents the results from the following regression model:


Liquidity
i
=o +|
1
W O
i
+|
2
V olume
i
+|
3
P rice
i
+|
4
V olatility
i
+
i
, (2.5)


where Liquidity
i
, the dependent variable, represents three liquidity measures: absolute spread, relative spread, and total
number of transactions. V olume
i
, P rice
i
, and V olatility
i
are the independent control variables. W O
i
is a dummy variable
that is one if the day falls in the write-of window and zero otherwise.
All variables represent daily averages, and all but the dummy variable are transformed by taking the log.
I adjust the t-statistics for heteroscedasticity, serial correlation, and arbitrary cross-correlations by using the Newey and
West (1987) procedure.

Table 5 provides the results from estimating equation (2.5) for each of the four liquidity measures, absolute
spread, relative spread, total number of transactions, and the transaction/spread ratio. The co-
efcients for all of the control variables are significant at 5 percent or less. The signs of the coefcients
are consistent with microstructure theory. Higher volumes and prices are associated with higher liquidity, while higher
volatility levels are associated with lower liquidity. The estimated volume coefcients are
negatively related to absolute and relative spreads, while positively related to total number of transac-
tions. Increased stock prices are related to wider absolute spreads, narrower relative spreads, and increased
number of transactions. Higher volatility is positively related to spreads and negatively related to number
of transaction.

The most important result of this estimate is the coefcients for WO, the dummy variable for the write-of period.
The negative and highly significant write-of period coefcients for both the absolute and relative spread regression
demonstrate that bid-ask spreads decrease following a write-of announcement,
even after controlling for changes in price, volume, and volatility. The write-of coefcient is positive and
significant for the transaction regression, which shows the write-of activity increases firm transactions.
3


I interpret these results as evidence of the asymmetric-information hypothesis. When traders are afected by a
decrease in the asymmetric information, they increase liquidity by reducing bid-ask spreads and increasing number of
transactions. I show that spread decrease significantly in both univariate and
multivariate testing.



2.4 Governance and Liquidity

Minnick (2004) shows that the market reacts diferently to the information content of write-ofs based on the quality of the
announcing firm's governance. This relationship between governance and write-ofs can also have implications on the
liquidity efect of write-of announcements. If the information ?owing from
good governance companies' write-ofs were more transparent than the information from bad governance
write-ofs, then one would expect to see a greater improvement in liquidity for good governance firms, as
compared to bad governance write-of firms. To test whether this is true, I run the following model,


Liquidity
i
=o +|
1
÷
5
1GOV V ARS
i
+|
6
V olume
i
+|
7
P rice
i
+|
8
V olatility
i
+
i
, (2.6)


where Liquidity
i
, the dependent variable, represents the change in three liquidity measures: absolute spread, relative
spread, and total number of transactions. V olume
i
, P rice
i
, and V olatility
i
are the inde-
pendent control variables. All variables represent daily averages, and all are transformed by taking the log.
GOV V ARS are the various governance variables used in Minnick (2004), including CEO turnover, board size, percent of
outsiders on board, and shareholder protection index. NEWCEO is a dummy variable that
is one if there was CEO turnover, and zero otherwise. BDSIZE is the number of members on the board,

3
I also run the analysis with three sub samples: companies with strong monitoring mechanisms, companies with mediocre
monitoring mechanisms, and companies with weak monitoring mechanisms, based on the IRRC database
from Gompers, Ishii, and Metrick (2003). I find that the above results are driven by the strong and mediocre
monitored companies. The poor weakly monitored companies do not show any significant liquidity improvement.


48
PERCTOUT is the percent of uniquely independent outside board members, as discussed in Yermack
(1999). GOV INDEX is the IRRC metric, where the higher the index, the worse the level of shareholder
protection. Conversely, the lower the number, the stronger the shareholder protection. This measure is
discussed in detail by Gompers, Ishii, and Metric (2002). I adjust the t-statistics for heteroscedasticity by using a robust
regression. The analysis is run only on the write-of firm 25 day window.

Table 6 shows the results of the estimate of model (2.6). As before, the control variables in Table 6 are all highly
significant and exhibit the expected signs. More important are the results of the gov-
ernance variables. The results suggest that the better the governance, the higher the improvement in
liquidity. The board composition variables, as well as the shareholder protection variable are all signifi- cant determinants
for both the spread and number of transactions at the five percent significance level.
BDSIZE and GOV INDEX are positively related to spreads, and negatively related to number of transac-
tions. PERCTOUT is negatively related to spread and positively related to number of transactions. CEO turnover is
negatively related to spreads and positively related to number of transactions. I interpret these
results as evidence that companies with good board composition, or strong shareholder protection mea-
sures show a greater reduction in asymmetric information from a write-of announcement than companies with poor
governance.

The empirical evidence in Tables 5 and 6 supports the following conclusions. When companies divulge private
information, such as a write-of, absolute and relative spreads decrease significantly, and
total number of transactions increase substantially. The quality of the information released is better, and
less noisy for good governance firms, as compared to poor governance firms.

To test the robustness of my results, I run the above analysis for both first time and multiple write- ofs. Table 7
shows the results of these estimates. Panel A shows the results of model (2.6) for a company's first time write-of. Panel B
shows the estimate for multiple write-ofs. There is not much diference for
segmenting the impact of liquidity by first versus multiple write-ofs. I use a Hausman test to see if there
is any significant diference in liquidity for first and multiple write-ofs. I find that the liquidity benefit of write-ofs exists
regardless of how many write-ofs a company has taken in the past.

In addition, I examine whether the size of the write-of afects the changes in liquidity. The results are shown in
Table 8. I find that the larger the size of the write-of, the smaller the impact on liquidity.
I find that larger write-ofs lead to smaller changes in spreads, and absolute spreads. The number of
transactions is also negatively impacted by write-of size.



2.5 Adverse Selection

A significant recent advance in the market microstructure literature is the development of models that decompose the
bid-ask spread into various components. In these models, the spread generally has three
components: order processing, inventory holding, and adverse selection (asymmetric information). Dem- setz (1968) and
Tinic (1972) identify an order processing cost that is made up of exchange and clearing fees,
bookkeeping and back ofce costs, the market maker's time and efort, and other random business costs.
Since a large part of this cost is fixed, order-processing costs are lower for more heavily traded securities. Inventory
holding costs are due to order ?ow imbalances that cause the market maker's inventory positions
to deviate from optimal levels (Stoll (1978) and Ho and Stoll (1983)). Wider bid-ask spreads, and larger
inventory holding costs result from increased deviation. Copeland and Galai (1983), Glosten and Milgrom (1985) and
Easley and O'Hara (1987) suggest that asymmetric information and its consequent informed
trading is a third spread component. Adverse selection costs are included in the spread to cover market
participants' expected losses to informed traders.

In prior sections, I show that the write-of announcement is associated with narrower spreads, and increased
number of transactions, particularly after controlling for changes in price, volume, and volatility, and governance
measures. In this section, I focus on the adverse selection component of the spread in order
to isolate the potential cause of these improvements in liquidity. According to the information-asymmetry
hypothesis, increased information in the market will decrease adverse selection costs. The validity of my



49
empirical results, however is dependent on the accuracy of the component estimation technique.
4
To
control for co-tangential specification errors, I estimate the adverse selection components using several
decomposition models, which help to address issues of robustness and accuracy.

I intend to measure changes in adverse selection caused by a write-of announcement. In each of the
decomposition models, I introduce the interaction term WO, which takes the value of one for trades
associated with write-of announcements, and zero otherwise. I define WO to include the actual announce- ment day, and
25 subsequent trading days.
5
Alternative definitions of the period provide similar results
as those reported here within. Positive and significant coefcients on the WO term would confirm the
hypothesis that write-ofs induce higher adverse selection costs. Measures such as bid-ask midpoints and transaction
process are transformed by taking natural logarithms, as in Lin, Sanger, and Booth (1995).
Each decomposition model is estimated on a pooled cross sectional basis.
6


Lin Sanger and Booth (1995) develop a method of estimating empirical components of the efective spread that
follows Huang and Stoll (1994), Lin (1993), and Stoll (1989). Lin et al. (1995) define the
signed efective half spread, z
t
, as the transaction price at time t, P
t
, minus the spread midpoint, M
t
.
The signed efective half spread is negative for sell orders and positive for buy orders. To re?ect possible
adverse selection information revealed by the trade at time t, Lin et. al (1995) addì, which is the adverse
selection component of the bid-ask spread.

I follow the LSB (1995) decomposition technique. In the model, the parameters are estimated
through the following regression equation:


oM
t+1
=ìz
t
+ t
+1
, (2.7)

where,
M
t
= log quote midpoint at time t
ì = parameter of regression which estimates the adverse selection component of the spread
o = Change in relative variable from t to t+1
z
t
= P
t
- M
t

P
t
= log trade price at time t, and
t
+1
= random error term with zero mean and constant variance. I follow Lin et al. (1995) by using a
robust OLS to estimate the following equation:


?M
t+1
=ìz
t
+ì
WO
(z
t
- W O
t
) + t
+1
, (2.8)


whereì
WO
is the incremental adverse selection component during the write-of period.

Table 9 shows the estimated adverse selection component,ì of 0.15 (t-value = 23.55) is significant at five percent. The
result can be interpreted as 15 percent of the bid-ask spread is attributable to in-
formation costs. The more important result is the estimated interaction term, ì
WO
of -0.30 (t-value =
-23.44), which is significant at five percent. The interaction term confirms that write-of announcements
decrease adverse selection costs. During write-of periods, adverse selection decreases by 15 percent of the
bid-ask spread.

I also decompose the spread using the empirical model of Huang and Stoll (1997), and implemented by Weston
(2000). They derive a simple model that allows a one-step decomposition of the information
component as a percentage of the spread. The remaining spread stems from order-processing costs and market maker
rents.

The model identifies these components by measuring how the midpoint of the spread, M
t
, changes as a function of
the direction of trades. They define an indication variable Q
t
, which takes on the values,

4
See Van Ness, Van Ness, and Warr (2001) for a discussion of the various benefits of diferent adverse selection models.
5
See Brockman and Chung (2001) for a description of this interaction technique.
6
I also run the estimations on a firm-by-firm basis, and find that it does not alter the results.


50
{-1,0,1} based on the direction of trade. If P
t
¡ M
t
, then Q
t
= -1 (sell order), if P
t
= M
t
, then Q
t
=0, and
if P
t
. M
t
, then Q
t
=1 (buy order). The model is specified as


?M
t
=o(S
t
÷
1
/2)Q
t
÷
1
+ +
t
, (2.9)


whereo measures the proportion of the half spread S
t
÷
1
/2, that stems from information costs. I follow
Huang and Stoll (1997) by using a robust OLS to estimate the following equation:

?M
t
=o(S
t
÷
1
/2)Q
t
÷
1
+o
WO
(S
t
÷
1
/2)Q
t
÷
1
- W O
t
) + t
+1
, (2.10)

whereo
WO
is the incremental adverse selection component during the write-of period.

Table 8 shows that the estimated adverse selection component of the model,o has a coefcient of 1.43. (t-value = 5.01).
This can be interpreted as 143 percent of the bid-ask spread is attributable to
information costs. The interaction variable,o
WO
, is -1.5, (t-value = -4.61). This can be interpreted as
evidence that write-ofs decrease adverse selection costs by 7 percent.

Overall, the decomposition results are consistent with the liquidity results. Write-ofs reduce the information
asymmetry in the market. This results in a situation where market participants react by
reducing adverse selection costs, and increasing liquidity.



2.6 Robustness Checks

In addition to the microstructure literature concerning information asymmetry, another strand of literature considers
analysts estimates as a good indicator of information asymmetry, where an increase in information
about a firm tends to lead to a convergence of opinions regarding the firms expected future earnings.
These papers typically use proxies for asymmetric information derived from consensus analysts forecasts of earnings per
share. Krishnaswami and Subramaniam (1998), for example, use the analysts forecast
errors to examine the change in the information environment before and after the completion of a spin-of.
In this section, I use analyst forecasts as an additional measure of the level of information asymmetry.
I use IBES and First Call data to analyze the relationship between write-ofs and earnings. Table
10, Panel A, shows the summary statistics of the First Call analyst data. The estimates are for the first
quarter following the write-of announcement. Instead of using the average of all of the analyst forecasts for a particular
firm in a particular quarter, I look at each one individually. Doing so allows me to see if
there is less dispersion in forecasts after the write-of. The results are consistent with our earlier findings.
Overall, write-ofs lead to a significant reduction in information asymmetry which is re?ected in the analyst
estimates. These results are driven by the better goverened firms. The weakly monitored firms do not
show a significant change in earnings transparency(via a change is surprise). However, both the mediocre and strongly
monitored firms do show an improvement.

In Table 10, Panel B, I formally test the relationship between forecast error, and firm specific variables. Using
the following regression, I estimate what impact write-ofs, governance, and earnings
management have on the analyst forecast error:


F ORECAST ERROR =|
1
+|
2
SIZE
i
+|
3
W O
i
+|
4
÷
9
GOV V ARS
i
+|
10
ACCRU ALS
i
+|
11
GROW T H
t+1
,
(2.11)
where FORECASTERROR is defined as the absolute value of the surprise, GROWTH is defined as the change in sales
from the same quarter in the previous year. I also control for ACCRUALS, defined as
GAAP earnings less cash from operations. I include growth because growing firms have predictably lower
cash ?ows due to higher working capital and long-term capital investments, but less predictable earnings. I include
ACCRUAULS as a proxy for earnings management, as a company that utilizes earnings man-
agement techniques has more easily predicted future earnings. The results support the above conclusions,
where write-ofs lead to a significant decline in forecast error. In addition, firms with smaller boards, and a higher percent
of outside directors see a significant decline in the forecast error. Overall, the forecast error
evidence suggests that write-ofs, especially write-ofs from well monitored companies lead to a reduction


51
in information asymmetry, and an improvement in earnings transparency.



2.7 Conclusion

This study is motivated by the rise in the use of write-ofs over the past two decades. With the increase usage of write-ofs,
it has become unclear as to whether write-ofs improve the information environment, or
just create more noise. The problem is exacerbated by unclear disclosure policies for the write-of events.
Management has the discretion to decide when to take a write-of and for how much the write-of amount should be. This
paper attempts to shed some light on the impact of the announcements on the information
environment, by analyzing write-of announcements from 1990 to 2000.

The goal of this paper is to measure the impact of write-ofs on liquidity. Using three separate liquidity measures,
I find that liquidity increases substantially following a write-of announcement. In
univariate analysis, I find that bid-ask spreads, both relative and absolute, decline, and that the number of transactions
increase following write-ofs. Using a multivariate framework, after controlling for changes in
price, volume, and volatility, I still find that liquidity improves following a write-of announcement. These
results provide overwhelming evidence in favor of the information-asymmetry hypothesis.

Following the test of the information asymmetry hypothesis, it is important to see if there is any diference in the
liquidity benefit from good governance write-of firms versus bad governance write-of firms. To test whether governance
in?uences write-ofs, I use a multivariate analysis that controls for the
impact of price, volume, and volatility on spreads and number of transactions. I find that even when
controlling for systematic changes, write-ofs from companies with small boards, larger percent of outside directors, and
strong shareholder protection will lead to a greater improvement in liquidity than companies
with poor governance. As a robustness check, I also use analyst forecast error as a measure of information
asymmetry, and find that write-ofs, especially write-ofs from well-monitored companies are related to a reduction in
forecast error.

Finally, I decompose bid-ask spreads in order to measure the efect of write-ofs on adverse se- lection costs. The
adverse selection results confirm that write-ofs improve the information asymmetry,
and improve liquidity as a response. Adverse selection costs decrease significantly during the write-of
announcement period in all three decomposition models. Overall, spread, number of transactions, and
decomposition results suggest a picture where write-ofs, especially those from good governance firms, im-
prove the information environment and lead to a liquidity benefit for investors.























52
2.8 Tables

Table I Sample Information

This table shows, by year, the number of write-of announcements for layof based, asset based, and combined write-ofs. It
also shows the mean and median write-of charge by write-of type. The prior
quarter's total assets to create a ratio adjust the write-of charge amounts. The mean and median of this
ratio, Charge/TA, is shown. The charge amounts are in millions of dollars.


Asset Write-ofs Layof Write-ofs Combination Write-ofs
Charge Charge/TA Charge Charge/TA Charge Charge/TA
Year # Mean Med. Mean Median Mean Med. Mean Med. Mean Med. Mean Med.
1985 25 10.4 2.7 0.009 0.003 62.5 62.5 0.010 0.013 175.0 44.0 0.019 0.007
1986 34 44.8 6.7 0.024 0.012 3.5 3.5 0.034 0.013 44.7 13.2 0.013 0.005
1987 45 65.0 12.0 0.041 0.012 61.3 51.0 0.032 0.026 234.0 43.5 0.045 0.006
1988 44 11.2 5.7 0.009 0.002 67.4 34.2 0.030 0.009 122.0 12.2 0.033 0.007
1989 55 15.0 2.6 0.010 0.003 92.5 51.6 0.021 0.013 51.7 16.5 0.012 0.004
1990 53 41.7 13.7 0.010 0.006 147.0 139.0 0.032 0.013 88.8 35.0 0.026 0.005
1991 75 79.2 7.0 0.019 0.016 81.2 10.5 0.011 0.002 57.8 24.3 0.018 0.007
1992 68 36.5 25.0 0.029 0.006 82.6 48.0 0.021 0.015 108.0 49.5 0.022 0.010
1993 82 99.9 21.7 0.029 0.004 88.0 33.0 0.021 0.008 132.0 43.4 0.018 0.011
1994 82 134.0 13.6 0.021 0.017 35.1 18.0 0.011 0.005 129.0 49.5 0.022 0.012
1995 78 63.3 15.7 0.023 0.021 104.0 21.6 0.029 0.001 64.8 16.3 0.015 0.005
1996 85 69.3 42.9 0.058 0.007 40.5 1.9 0.049 0.002 109.0 29.8 0.016 0.008
1997 86 191.0 9.0 0.032 0.006 96.7 62.0 0.010 0.005 101.0 27.5 0.017 0.008
1998 86 28.6 16.2 0.016 0.005 30.6 15.0 0.049 0.002 100.0 30.6 0.019 0.007
1999 87 46.5 15.7 0.007 0.007 5.9 5.8 0.018 0.009 111.0 36.2 0.061 0.007
2000 90 79.6 16.5 0.014 0.004 63.2 62.4 0.006 0.007 66.2 27.9 0.022 0.011
All 1075 63.5 9.0 0.022 0.007 68.7 20.0 0.012 0.002 100.2 29.2 0.023 0.007





























53
Table II Summary Statistics

This table shows summary statistics on market and write-of activity for NYSE listed companies in the consumer
manufacturing industry. The sample period spans 1993-2000. Comparative market statistics over the same period are
given for the population of all non write-of firms on the NYSE.


Average market capitalization
Average trading Volume
Average daily closing price
Average daily returns (with dividends)
Average size of write-ofs, adjusted by total assets
Percentage of companies with one write-of
Percentage of companies with second write-ofs
Percentage of companies with third + write-ofs

Write-of Companies
11,200,000
118,319
43.98
0.02
0.01
25%
21%
14%

Non Write-of Companies
10,600,000
107,331
37.19
0.01











































54
Table III Liquidity ummary Statistics

This table shows summary statistics on liquidity measures for write-of periods versus surrounding non write-of periods.
The write-of window is t=0 to t=25. The non write-of window ends 120 days before an announcement, and begins 120
after a write-of announcement. Volume is the total trading volume on
a trading day. Returns is the returns over each trading day. Volatility measures the variance of returns.
Price is the average daily transaction price. Absolute spread, relative spread, total number of transactions, number of ask
transactions, and bid transactions are the averages for the absolute dollar bid-ask spread,
the relative percent bid-ask spread, total number of transactions, number of ask transactions, and bid
transactions. The t statistics are from the paired t-test for the diferences in means between the write-of window, and the
surrounding non write-of window. The sign test statistics are from the non-parametric
sign test for the diferences in the median measures between the write-of window, and the surrounding
non write-of window. All p-values are reported based on two tailed significance. Significance is indicated at the 0.05 and
0.01 levels by one and two asterisks respectively.


WO Period Non WO Period Diference Significance Tests
Mean Median Mean Median Mean Median Paired t-test Sign test
Volume 248.13 160.8 229.27 134.82 18.86 25.98 -2.8** 0.00**
Price 33.74 31.54 35.91 33.09 -2.17 -1.55 6.41** 0.00**
Returns 0 0 0 0 0 0 -0.83 0.21
Volatility 0.4 0.49 0.39 0.46 0.02 0.03 -7.39** 0.83
Absolute Spread 0.28 0.19 0.37 0.19 -0.09 0 12.96 ** 0.56
Relative Spread 0.01 0.01 0.02 0.01 -0.01 0 6.3 ** 0**
Total Transactions 349.26 279.99 280.05 212.68 69.22 67.31 -15.15** 0**
Ask Transactions 164.92 131.96 133.97 100.15 30.95 31.81 -13.81** 0**
Bid Transactions 158.8 146.83 128.86 113 29.94 33.83 -16.11** 0 **
































55
Table IV Univariate Analysis of Write-of Liquidity Changes to Earnings Liquidity Changes

This table looks at the univariate statistics for the change in liquidity when a company announces a write- of and when a
company announces its quarterly earnings. Change is calculated as the diference between
liquidity for a write-of firm, and all other firms on the NYSE on the announcement date. he t statistics
are from the paired t-test for the diferences in means between the write-of window, and the surrounding non write-of
window. I compare the average values of changes in absolute spreads, relative spreads, and
total volume for both the earnings and write-of announcement date. Significance is indicated at the 0.05
and 0.01 levels by one and two asterisks respectively.

WO change Earnings change t-value
Absolute Spread -0.08 -0.04 -8.69 **
Relative Spread -0.10 -0.006 -5.97 **
Turnover 52 30 8.67 **











































56
Table V Multivariate Analysis of Liquidity

This table shows the results of a regression of liquidity measures across write-of and non write-of periods, controlling for
the efects of price, volume, and volatility.

Liquidity
i
=o +|W O
i
+¸
1
V olume
i
+¸
2
P rice
i
+¸
3
V olatility
i
+ ,

where Liquidity
i
is the dependent variable and stands for either the log of absolute spread, relative spread ,total number of
transactions, or total number of transactions/absolute spread. Absolute spread is a
measure of the average absolute dollar bid-ask spread of a sample firm. Similarly, relative spread and total number of
transactions are the daily averages for the relative bid-ask spread, and total number of
transactions. WO is coded with a one if the day is within 25 days following a write-of event, otherwise 0.
Volume is the total trading volume during the trading day. Price is the average of all transaction prices recoded on the
trading day. Volatility is the variance of returns over the trading day. All non-dummy
variables are calculated by taking the natural logarithm. Significance is indicated at the 0.05 and 0.01
levels by one and two asterisks respectively and all results are presented based on two-tail significance.


Absolute Spread Relative Spread Total Trans. Total Trans/Abs. Spread
Coefcient t-value Coefcient t-value Coefcient t-value Coefcient t-value
WO -0.098 -5.92** -0.075 -2.05 ** 0.194 11.97 ** 0.129 3.32 **
Price -0.029 -2.83 0.152 12.43 ** -0.014 -12.98 ** -0.203 -7.23 **
Volume 0.004 1.43 ** -0.008 -2.43 ** 0.005 107.21 ** 0.621 65.74 **
Volatility -0.081 -23.95 ** -0.227 -55.38 ** 0.598 2.73 ** 0.028 3.89 **
Constant -1.168 -31.63 ** -4.755 -109.35 ** 2.521 55.91 ** 4.727 48.43 **
F(4,7592) 165.42 777.96 6989.84 1157.98

































57
Table VI Multivariate Analysis of Liquidity and Governance

This table shows the results of a regression of liquidity measures across write-ofs and governance measures, controlling for
the efects of price, volume, and volatility.

Liquidity
i
=o +|
1
÷
5
1GOV V ARS
i
+|
6
V olume
i
+|
7
P rice
i
+|
8
V olatility
i
+ i

where Liquidity
i
is the dependent variable and stands for either the percent change in absolute spread, relative spread , total
transactions, or total transactions/absolute spread. from the non write-of period
to the write-of period . Absolute spread is a measure of the average absolute dollar bid-ask spread of a sample firm.
Similarly, relative spread and total transactions are the daily averages for the relative bid-ask
spread, and total number of transactions. Volume is the total trading volume during the trading day. Price
is the average of all transaction prices recoded on the trading day. Volatility is the variance of returns over the trading day.
All liquidity variables are calculated by taking the natural logarithm. BDSIZE is the size
of the board of directors. PERCTOUT is the percent of outside directors on the board. GOV INDEX is
an index that ranks the level of shareholder protection for the shareholders, where the higher the number, the worse the
protection. It is calculated using the Gompers, Ishii, and Metrick (2002) index. NEWCEO
is a dummy variable that is one if there was a CEO turnover in the days following. Significance is indicated
at the 0.05 and 0.01 levels by one and two asterisks respectively and all results are presented based on two-tail
significance.


Absolute Spread Relative Spread Total Trans. Trans./Abs. Spread
Coefcient t-value Coefcient t-value Coefcient t-value Coefcient t-value
Price -0.24 -6.23 ** -2.64 -18.06 ** 0.27 10.97 ** 0.08 5.64 **
Volume 0.04 4.21 ** 0.12 3.39 ** -0.03 -13.41 ** 0.03 3.41 **
Volatility -0.04 -2.62 ** -0.27 -5.26 ** 0.55 9.67 ** 1.05 104.61 **
BDSIZE 0.03 3.5 ** 0.11 3.86 ** -0.12 -5.54 ** -0.02 -8.16 **
PERCTOUT 0.08 0.51 0.62 1.03 1.66 1.21 0.049 0.9
NEWCEO -0.16 -3.48 ** -0.43 -2.4 ** 0.28 2.99 ** 0.05
3.02 **
GOV INDEX 0.03 3.41 ** 0.17 5.22 ** -0.07 -3.15 ** -0.03 -10.9 **
Constant 1.02 5.09 ** 11.12 14.59 ** -4.96 -26.66 ** 1.56 22.49 **
F( 7, 3610) 101.29 129.41 1218.5 847.67



























58
Table VII Liquidity and Number of Write-ofs

This table shows the results of a regression of liquidity measures across write-ofs and governance measures, controlling for
the efects of price, volume, and volatility.

Liquidity
i
=o +|
1
÷
5
1GOV V ARS
i
+|
6
V olume
i
+|
7
P rice
i
+|
8
V olatility
i
+ i

where Liquidity
i
is the dependent variable and stands for either the percent change in absolute spread, relative spread , or
total Transactions from the non write-of period to the write-of period . Panel A
shows the results for first time write-ofs, while Panel B is for multiple write-ofs. Absolute spread is a measure of the
average absolute dollar bid-ask spread of a sample firm. Similarly, relative spread and total
Transactions are the daily averages for the relative bid-ask spread, and total number of transactions. I also
run the estimate for transactions divided by absolute spread. Volume is the total trading volume during the trading day.
Price is the average of all transaction prices recoded on the trading day. Volatility is the
variance of returns over the trading day. All liquidity variables are calculated by taking percent change
in liquidity write-ofs compared to the rest of the NYSE on that day. BDSIZE is the size of the board of directors.
PERCTOUT is the percent of outside directors on the board. GOV INDEX is an index that
ranks the level of shareholder protection for the shareholders, where the higher the number, the worse the
protection. It is calculated using the Gompers, Ishii, and Metrick (2002) index. NEWCEO is a dummy variable that is one
if there was a CEO turnover in the days following. Significance is indicated at the
0.05 and 0.01 levels by one and two asterisks respectively and all results are presented based on two-tail
significance.

Panel A


Absolute Spread Relative Spread Total Trans. Trans./Abs. Spread
Coefcient t-value Coefcient t-value Coefcient t-value Coefcient t-value
Price -0.24 -6.23 ** -2.64 -18.06 ** 0.03 1.23 0.12 4.80 **
Volume 0.04 4.21 ** 0.12 3.39 ** 0.17 16.47 ** 0.03 2.08 **
Volatility -0.04 -2.62 ** -0.27 -5.26 ** 0.03 4.62 ** 0.96 47.87 **
BDSIZE 0.03 3.50 ** 0.11 3.86 ** -0.05 -8.78 ** -0.02 -2.37 **
PERCTOUT 0.08 0.51 0.62 1.03 1.27 10.41 ** -0.46 -3.82 **
NEWCEO -0.16 -3.48 ** -0.43 -2.40 ** 0.22 5.88 ** -0.02 -0.56
GOV INDEX 0.03 3.41 ** 0.17 5.22 ** -0.01 -0.71 0.02 2.86
**
Constant 1.02 5.09 ** 11.12 14.59 ** -2.50 -17.16 ** 1.62 11.78 **
F( 7, 3610) 101.29 129.41 155.36 819.9

Panel B


Absolute Spread Relative Spread Total Trans. Trans/Abs. Spread
Coefcient t-value Coefcient t-value Coefcient t-value Coefcient t-value
Price -0.03 -2.03 ** -0.58 -27.62 ** 0.09 3.82 ** 0.05 3.37 **
Volume -0.05 -7.74 ** -0.05 -7.19 ** 0.42 49.88 ** 0.02 2.21**
Volatility 0.01 2.91 ** 0.01 1.82 -0.01 -2.15 ** 1.06 93.97 **
BDSIZE 0.01 3.09 ** 0.01 1.81 0.05 10.55 ** -0.02 -6.69 **
PERCTOUT 0.16 2.47 ** 0.08 2.47 ** 0.73 7.62 ** 0.21 3.49 **
NEWCEO -0.05 -2.54 ** -0.09 -3.47 ** 0.02 0.52 0.05 2.30 **
GOV INDEX 0.01 0.49 0.01 1.84 -0.04 -7.82 ** -0.05 -14.05 **
Constant -0.19 -2.16 ** 1.90 17.45 ** -2.80 -22.62 ** 1.59 19.97 **
F( 7, 3610) 18.94 169.58 628.53 4107.66










59
Table VIII Liquidity and Size of Write-ofs

This table shows the results of a regression of liquidity measures across write-ofs and governance measures, controlling for
the efects of price, volume, and volatility.

Liquidity
i
=o +|
1
÷
5
1GOV V ARS
i
+|
6
V olume
i
+|
7
P rice
i
+|
8
V olatility
i
+|
9
W O T A
i
+|
1
0W O#
i
i

where Liquidity
i
is the dependent variable and stands for either the percent change in absolute spread, relative spread , or
total transactions from the non write-of period to the write-of period. Absolute spread
is a measure of the average absolute dollar bid-ask spread of a sample firm. Similarly, relative spread and total transactions
are the daily averages for the relative bid-ask spread, and total number of transactions.
I also run the estimate for transactions divided by absolute spread. Volume is the total trading volume
during the trading day. Price is the average of all transaction prices recoded on the trading day. Volatility is the variance of
returns over the trading day. All liquidity variables are calculated by taking percent
change in liquidity write-ofs compared to the rest of the NYSE on that day. BDSIZE is the size of the
board of directors. PERCTOUT is the percent of outside directors on the board. GOV INDEX is an index that ranks the
level of shareholder protection for the shareholders, where the higher the number, the
worse the protection. It is calculated using the Gompers, Ishii, and Metrick (2002) index. NEWCEO is a
dummy variable that is one if there was a CEO turnover in the days following. WO TA is the size of the write-of divided
by total assets. WO# is the number of write-ofs that the firm has taken from 0 to 26.
Significance is indicated at the 0.05 and 0.01 levels by one and two asterisks respectively and all results
are presented based on two-tail significance.


Absolute Spread Relative Spread Total Trans. Trans./Abs. Spread
Coefcient t-value Coefcient t-value Coefcient t-value Coefcient t-value
Price -0.04 -3.04 ** -0.64 -33.33 ** 0.01 0.45 0.082 5.93 **
Volume -0.03 -6.37 ** -0.05 -6.54 ** 0.29 49.71 ** 0.03 4.08 **
Volatility 0.02 4.70 ** 0.02 3.71 ** 0.01 1.22 1.02 99.75 **
BDSIZE 0.01 5.05 ** 0.01 3.68 ** 0.04 12.86 ** -0.03 -9.86 **
PERCTOUT 0.11 1.82 0.04 0.52 0.32 4.70 ** 0.18 3.10 **
NEWCEO -0.06 -3.32 ** -0.07 -3.07 ** -0.01 -0.28 0.05 3.11 **
GOV INDEX 0.04 0.98 0.03 0.99 0.01 -0.84 -0.04 -11.02 **
WO TA 3.06 5.27 ** 3.75 4.99 ** -0.49 -0.78 -3.56 -6.43 **
WO# -0.01 -4.36 ** -0.02 -5.32 ** 0.01 5.12 ** 0.00 -0.82
Constant -0.11 -1.42 2.13 21.32 ** -2.11 -25.44 ** 1.614 23.11 **
F( 7, 3610) 193.28 26.04 533.57
3839.25






















60
Table IX Adverse Selection

This table shows the results of the adverse selection tests to see if write-ofs reduce adverse selection. Lin Sanger and
Booth (1995) develop a method of estimating empirical components of the efective spread,
where the signed efective half spread, z
t
, is defined as the transaction price at time t, P
t
, minus the spread
midpoint, M
t
. The signed efective half spread is negative for sell orders and positive for buy orders. To
re?ect possible adverse selection information revealed by the trade at time t, Lin et. al (1995) addì, which
is the adverse selection component of the bid-ask spread. I follow Lin et al. (1995) by using a robust OLS
to estimate the following equation:

?M
t+1
=ìz
t
+ì
WO
(z
t
- W O
t
) + t
+1
,

whereì
WO
is the incremental adverse selection component during the write-of period. The results are shown in Model 1. I also decompose the
spread using the empirical model of Huang and Stoll (1997), and
implemented by Weston (2000). The midpoint of the spread is defined as, M
t
, and changes as a function
of the direction of trades. An indication variable Q
t
takes on the values, {-1,0,1} based on the direction of trade. If P
t
¡ M
t
,
then Q
t
= -1 (sell order), if P
t
= M
t
, then Q
t
=0, and if P
t
. M
t
, then Q
t
=1 (buy order).
I follow Huang and Stoll (1997) by using a robust OLS to estimate the following equation:

?M
t
=o(S
t
÷
1
/2)Q
t
÷
1
+o
WO
(S
t
÷
1
/2)Q
t
÷
1
- W O
t
) + t
+1
,

whereo measures the proportion of the half spread S
t
÷
1
/2, that stems from information costs ando
W
Ois the incremental adverse selection
component during the write-of period. Significance is indicated at the
0.05 and 0.01 levels by one and two asterisks respectively and all results are presented based on two-tail
significance.

?M
t+1
- Model 1 ?M
t
- Model 2
Coefcient t-value Coefcient t-value
Z 0.146 23.55
Z*WO -0.302 -23.34
(S
t
÷
1
/2)Q
t
÷1 1.43 5.01
(S
t
÷
1
/2)Q
t
÷
1
*WO -1.45 -4.61
R-squared 0.114 0.009
F( 2, 7623) 277.3 5.35


























61
Table X Earnings and Write-ofs

This table examines the earnings in the periods surrounding the write-of announcement. Panel A shows the earnings
surprises, and Panel B shows the regression of the absolute value of forecast error on firm specific variabl es, where
forecast error is defined as the absolute value of the median analyst estimate
for the first earnings quarter following the write-of. Earnings data is from the First Call database. A *
denotes significance at the 5 percent level, and ** denotes significance at the 10 percent level.

Panel A: Earnings Surprises
Earnings Surprise
Mean Median t-value Sign-rank test
Weak Monitors -0.07 -0.02 -1.03 -1.30
Mediocre Monitors -0.02 0.00 -1.97* -2.26 **
Strong Monitors -0.03 -0.01 -2.54** -2.35 ** All
Write-ofs -0.03 -0.01 -2.89** -3.48 **
Panel B - OLSQ Results for Forecast Error
Coefcient t-value
WO -0.018 -3.40 **
SIZE 0.033 6.79 **
BDSIZE 0.026 5.16 **
PERCTOUT -0.015 -5.35 **
GOV INDEX
ACCRUALS
ROA
DEBT RATIO
NEWCEO
GROWTH
CONSTANT
-0.001
-0.001
0.001
0.001
-0.002
0.004
0.042
-1.87
-1.57
3.52 **
0.87
-0.51
0.92
4.22 **































62
Tax Laws and Write-ofs
Restructuring charges have become a popular topic with FASB. These charges are based on the big
bath practice where firms take one-time charges to clean up their balance sheet. These charges include
employee benefits, costs associated with discontinued operations, closed plants, product line elimination, and losses
incurred from asset disposal and impairment. Many of these charges are not currently addressed
by accounting standards, although some such as severance are. From a tax viewpoint, the main issue in-
volved with these write-ofs is the uncertainty of when these costs should be incurred.

Employee termination benefits or severance, are covered by accounting standards. EITF 94-3 applies to benefits
to be provided to employees afected by layofs. This liability is recognized in the period
management approves the layofs if the following criteria hold:
• Prior to the date of the financial statements, management approves of the termination benefits and
specifies the amount to be paid out.
• Prior to the date of the financial statements, the details of the layofs is communicated to the
employees in sufcient detail.
• The termination plan specifies the number of layofs, the job classification of the layofs, and the
specific departments.
• Changes in the plan are unlikely to occur.

Termination benefits that fall under the following criteria are not allowable as a write-of:
• Included with a disposal of a segment, which is also charged against earnings.
• Paid pursuant to the terms of an ongoing employee benefit plan.
• Paid under the terms of an individual deferred compensation plan.

Costs to exit an activity also have stated guidelines. When management commits to exiting an
activity, exit costs are incurred. Exit costs include:
• Costs that are a direct result of the exit plan and that the firm would not incur without the plan.
• Costs that existed through a contractual obligation prior to the exit plan, like the penalty to break
a lease.

The discontinued operations segment of an income statement consists of two parts. The first part is income (loss)
from operations, and the second is gain (loss) on disposal of assets. Income (loss) from
operations is disclosed for the current year only if the decision to discontinue operations is made after the beginning of
the fiscal year. Gain (loss) from sale of assets is a combination of income (loss) from
operations during the phase-out period and the gain (loss) from disposal of a segment. The gain (loss)
on disposal includes costs arising from the decision, such as severance, additional pension costs, employee relocation
expenses, and future rentals or leases.

If a loss is expected from the proposed sale or abandonment of a segment, the firm should provide for the loss it
makes at the time the decision to dispose. If there is a gain, it should be recognized when
the gain occurs. The results of discontinued operations appear as an independent line item on the income
statement, before extraordinary items.

In addition to being disclosed in the financial statements, the footnotes should disclose the following:
• The segment of business that has been afected.
• The expected disposal date.
• A description of the remaining assets and liabilities of the segment.
• The expected manner of disposal.
• The income or loss from operations and any proceeds from the disposal of the segment.

Extraordinary items are sometimes used to disclose write-ofs. An item is considered extraordinary if it is both
unusual in nature and infrequent in its occurrence. An item is considered unusual in nature if
it is unrelated to the line of business. To identify such items one needs to consider:


63
• Type and scope of operations.
• Line of business.
• Operating Policies.
• Industry.
• Geographic locale.
• Nature and extent of government regulations.

Accounting standards specifically note some items that can or cannot be included in extraordinary
items. The following items are considered extraordinary:
• Gains or losses from extinguishments of debt, except for sinking fund requirements.
• Profits or losses resulting from the disposal of a significant part of the assets of previously separate
companies.
• Write-of of operating rights of motor carriers.
• The investors share of an investees extraordinary item when the investor uses the equity method of
accounting.
• Gains of a debtor due to a troubled debt restructuring.


Items that cannot be considered extraordinary are:
• Write-down or write-of of receivables, inventory, equipment leased, or intangible assets. • Foreign
currency gains or losses.
• Gains or losses from the disposal of a business segment.
• Gains or losses from sale or abandonment of property, plant, or equipment. • Efects of a
strike.
• Adjustments on accruals on long-term contracts.
Extraordinary items are shown independently of ongoing operations and are shown net of taxes in
a separate section of the income statement.

In 2002, FASB created new guidelines in its Statement 144 for restructuring charges. Unless the disposal activity
involves a discontinued operation, costs associated with a disposal activity should be reported in continuing operations
before income taxes in the income statement. If the disposal activity
does involve a discontinued operation, then those costs should be included in discontinued operations.

The total amount of costs incurred and charged to expense should be reported by reportable seg- ment. The
accounting model for long-lived assets to be disposed of by sale applies to all long-lived assets,
including discontinued operations. Statement 144 requires that those long-lived assets be measured at the lower of either
carrying amount or fair value less cost to sell, whether reported in continuing operations or
in discontinued operations. Therefore, discontinued operations can no longer be measured at net realizable
value or include amounts for operating losses that have not yet occurred.

Statement 144 also broadens the reporting of discontinued operations to include all components of an entity with
operations that can be distinguished from the rest of the entity and that will be eliminated
from the ongoing operations of the entity in a disposal transaction.











64
BIBLIOGRAPHY

[1] Abowd, J., 1990, Does performance-based managerial compensation afect corporate performance?,
Industrial and Labor Relations Review, 43, 52-73.
[2] Aggarwal R. and A. Samwick, 1999, Executive compensation evaluation: Theory and evidence, Jour-
nal of Finance, 54, 1999-2043.
[3] Alciatore, M., P. Easton, and N. Spear, 2000, Accounting for the impairment of long-lived assets:
Evidence from the petroleum industry, Journal of Accounting and Economics, 29, 151-172.
[4] Amihud, Y. and H. Mendelson, 1986, Asset Pricing and the bid-ask spread, Journal of Financial
Economics 17, 223-49.
[5] Audt, D., W. Cready, and T. Lopez, 2003, Restructuring charges and CEO cash compensation: A
reexamination, Accounting Review, 78(1), 169-192.
[6] Bacidore, J., R. Battalio, and R. Jennings, 2002. Depth improvement and adjusted price improvement
on the New York Stock Exchange, The Journal of Financial Mar kets , 5, 169-195.
[7] Barclay, M. and C. Smith, 1988, Corporate payout policy: Cash dividends versus open market repur-
chases, Journal of Financial Economics 22, 61-82.
[8] Baker, G. and B. Hall, 1998, CEO incentives and firm size, Working paper-6868, NBER.
[9] Bartelsman, E. and W. Gray, 2002, The NBER Manufacturing Productivity Database, NBER Working
Paper.
[10] Bartov, E., F. Lindahl, and W. Ricks, 1998, Stock price behavior around announcements of write-ofs,
Review of Accounting Studies, 3/4, 327-346.
[11] Benston , G. and R. Hagerman, 1974, Determinations of bid-ask spreads in the over-the-counter
market, Journal of Financial Economics 1, 353-64.
[12] Berger, P., E. Ofek and D. Yermack, 1997, Managerial entrenchment and capital structure Decisions,
Journal of Finance, 52(4), 1411-1446.
[13] Bhagat, S. and B. Black, 1999, The uncertain relationship between board composition and firm
performance, Business Lawyer, 55, 921-963.
[14] Borokovich, K., R. Parrino, and T. Trapani, 1996, Outside directors and CEO selection, Journal of
Financial and Quantitative Analysis, 31, 337-355.
[15] Bradshaw, M and R. Sloan, 2002, GAAP versus the street: An empirical assessment of two alternative
definitions of earnings, Journal of Accounting Research, 4 0, 4 1-65.
[16] Brickley, J. and L. Van Drunen, 1990, Internal corporate restructuring: An empirical analysis, Journal
of Accounting and Economics, 12, 251-280.
[17] Brockman, P., Chung, D.Y., 1999, An analysis of depth behavior in an electronic, order-driven envi-
ronment, Journal of Banking and Finance 23, 1861-1886.
[18] Brockman, P., Chung, D.Y., 2001, Managerial Timing and Corporate Liquidity: Evidence from actual
share repurchases, Journal of Financial Economics, 61, 417-4
[19] Brown, L. and K. Sivakumar, 2001, Comparing the quality of three earnings measures, Working Paper,
Georgia State University.
[20] Brown, S. and J. Warner, 1985, Using daily stock returns: The case of event studies, Journal of
Financial Economics, 14, 3-31.
[21] Byrd, J. and K. Hickman, 1992, Do outside directors monitor managers? Evidence from tender ofer
bids, Journal of Financial Economics, 32, 95-221.
[22] Cofee, J., 1999, The future as history: The prospects for global convergence in corporate governance
and its implications, Northwestern University Law Revie w, 9 3, 641-708.
[23] Copeland, T.E., Galai, D., 1983, Information efects on the bid-ask spread,Journal of Finance 38,
1457-1470.
[24] Core, J. and W. Guay, 1999a, Estimating the value of stock option portfolios and their sensitivities
to price and volatility, Working Paper, University of Pen nsyl vania.
[25] Core, J. and W. Guay, 1999b, The use of equity grants to manage optimal equity incentive levels,
Journal of Accounting and Economics, 28, 151-184.


65
[26] Core, J., R. Holthausen, and D. Larcker, 1999, Corporate governance, chief executive ofcer compen-
sation, and firm performance, Journal of Financial Economi cs, 51, 371-406.
[27] Coughlan, A. and R. Schmidt, 1985, Executive compensation, management turnover, and firm per-
formance: An empirical investigation, Journal of Accounting and E cono mics 7(1-3), 43-66.
[28] Demsetz, H., 1968, The cost of transacting,Quarterly Journal of Economics 82, 33-53.
[29] Doyle J. and M. Soliman, 2002, Do managers use pro forma earnings to exceed analysts forecasts?,
Working Paper, University of Michigan.
[30] Doyle, J., R., Lundholm, and M. Soliman, 2003, The predictive value of expense excluded from pro
forma earnings, Review of Accounting Studies, 8, 145-74.
[31] Easley, D. and M. OHara (1987): Price, Trade Size and Information in Securities Markets, Journal of
Financial Economics, 19, pp 69-90.
[32] Easley, D., O'Hara, M., 1987, Price, trade size, and information in securities markets, Journal of
Financial Economics 19, 69-90.
[33] Elliot, J., and J. Hanna, 1996, Repeated accounting write-Ofs and the information content of earnings,
Journal of Accounting Research, 34, 135-156.
[34] Elliot, J., and W. Shaw, 1988, Write-ofs as accounting procedures to manage perceptions, Journal of
Accounting Research, 26, 91-119.
[35] Francis, J., D. Hanna, and L. Vincent, 1997, Causes and efects of discretionary asset write-ofs,
Journal of Accounting Research, 34, 117-134.
[36] Fried, D., M. Schif, and A. Sondhi, 1989, Impairments and write-ofs of long-lived assets, Management
Accounting, 71, 48-50.
[37] Garman, M., 1976, Market Micro-Structure, Journal of Financial Economics, pp. 257-275.
[38] Gibbons, J., Nonparametric Methods for Quantitative Analysis. 1976, Columbus, OH: American
Sciences Press.
[39] Glosten, L., Milgrom, P., 1985, Bid, ask, and transaction prices in a specialist market with heteroge-
neously informed traders, Journal of Financial Econom ics 14, 71-100.
[40] Gibson, M., 1999, Is corporate governance efective in emerging markets?, Working Paper, Federal
Reserve Board .
[41] Gompers, P, J. Ishii and A. Metrick, 2003, Corporate governance and equity prices, Quarterly Journal
of Economics, forthcoming.
[42] Hallock, K., 1997, Reciprocally interlocking boards of directors and executive compensation, Journal
of Financial and Quantitative Analysis, 32, 331-343.
[43] He?in, F. and K.W. Shaw, 2001, Adverse selection, inventory holding costs, and depth, Journal of
Financial Research 24.
[44] He?in, F., and T. Warfield, 1995, Earnings, firm value, and unusual charges of impaired assets,
Working paper, University of Wisconsin-Madison.
[45] Hermalin, B. and M. Weisbach, 1988, The determinants of board composition, The RAND Journal
of Economics, 19, 589-606.
[46] Hermalin, B. and M. Weisbach, 1991, The efects of board composition and direct incentives on firm
performance, Financial Management, 20, 101-112.
[47] Ho, T., Stoll, H.R., 1981, Optimal Dealer Pricing Under Transaction and Return Uncertainty, Journal
of Financial Economics, pp. 47-73.
[48] Ho, T., Stoll, H.R., 1983, The dynamics of dealer markets under competition, Journal of Finance38,
1053-1074.
[49] Huang, R.D., Stoll, H.R., 1997, The components of the bid-ask spread: a general approach, Review
of Financial Studies 10, 995-1034.
[50] Hvidkjaer, S., 2003, A trade-based analysis of momentum, Working paper, University of Maryland.
[51] Jacoby, G., Fowler, D., and A. Gottesman, 2000, The capital asset pricing model and the liquidity
efect: a theoretical approach, Journal of Financial mar kets 3, 69-81.



66
[52] Jain, P. K., and Z. Rezaee. 2004. The Sarbanes Oxlet Act of 2002 and Market Liquidity. Working
Paper, University of Memphis.
[53] Jensen, M. and K. Murphy, 1990, Performance pay and top-management incentives, Journal of Polit-
ical Economy, 98, 225-64.
[54] Jensen, M. and R. Ruback, 1983, The market for corporate control, Journal of Financial Economics,
11, 5-50.
[55] Johnson, W., and W. Schwartz, 2001, Are investors misled by pro forma earnings, Working Paper,
University of Iowa.
[56] Kaplan, S., 1994, Top executives, turnover, and firm performance in Germany, Journal of Law, Eco-
nomics & Organization, 10, 142-159.
[57] Kim, S. and S. Kwon, 2001, SFAS 121 asset write-down: Early versus late firms, Working paper,
Rutgers University - Camden.
[58] Kinney, M. and R. Trezevant, 1997, The use of special items to manage earnings and perceptions,
Journal of Financial Statement Analysis, 3, 45-53.
[59] Kross, W., T. Park, and B. Ro, 1996, The impact of operational restructuring announcements on
stock price, risk, and trading volume, Working Paper, Purdue Uni vers ity .
[60] Leonard, J., 1990, Executive pay and firm performance, Industrial and Labor Relations Review 43,
13S-29S.
[61] Lin, J., 1993, Order Persistence, Adverse Selection, and Gross Profits Earned by NYSE Specialists,
Journal of Finance Proceedings (Abstract)(July), 1108- 9
[62] Lin, J.-C., Sanger, G., Booth, G.G., 1995. Trade size and components of the bid-ask spread, Review
of Financial Studies 8, 1153-1183.
[63] Lindbeck, R., Z. Rezaee, and J. Smith, 1996, An examination of long lived asset impairments under
SFAS No. 121, International Advances in Economic Research, 2, 86-91.
[64] Macey, J., 1997, Institutional investors and corporate monitoring: A demand-side perspective, Man-
agerial and Decision Economics, 18, 601-610.
[65] May, D., 1995, Do managerial motives in?uence firm risk reduction strategies?, The Journal of Fi-
nance, 50 , 1291-1308.
[66] Mayers, D., A. Shivdasani, and C. Smith, Jr., 1994, Board composition and corporate control: Evi -
dence from the insurance industry, Working Paper, Ohio State Un iver sity .
[67] Meyer, J. and J. Strong, 1987, Asset write-downs: Managerial incentives and security returns, The
Journal of Finance, 42, 643-661.
[68] Minnick, K, 2004, Write-ofs and Corporate Governance, Working paper - University of Maryland.
[69] Murphy, K., 1985, Corporate performance and managerial remuneration: An empirical analysis, Jour-
nal of Accounting and Economic, 7, 11-42.
[70] Murphy, K., 1986, Incentives, learning, and compensation: A theoretical and empirical investigation
of managerial labor contracts, Rand Journal of Economics, 17, 59- 76.
[71] Murphy, K., 1999, Executive Compensation, in Orley Ashenfelter and David Card, Handbook of
Labor Economics, Vol. 3, North Holland, Amsterdam.
[72] Myers, S., and N. Majluf, 1984, Corporate financing and investment decisions when firms have infor-
mation that investors do not have, Journal of Financial Econo mics , 13, 187-221.
[73] Newey, W. and K. West, 1987, A simple positive semi-definite, heteroscedasticity and autocorrelation
consistent covariance matrix, Econometrica 55, 703-708.
[74] OHara, M., 2003. Presidential address: Liquidity and price discovery, Journal of Finance, 59, 1335-
1354.
[75] Parrino, Robert, 1997, CEO turnover and outsider succession: A cross-sectional analysis, Journal of
Financial Economics 46, 165-197.
[76] Paul, D., 2003, Board composition and corrective action: Evidence from corporate responses to bad
acquisition bids, Working Paper, Babson College.
[77] Phillips, G. and V. Maksimovic, 2002, Do conglomerate firms allocate resources inefciently across
industries? Theory and evidence, Journal of Finance, 57 , 72 1-767.


67
[78] Riedl, E., 2002, An examination of long-lived asset impairments, Working Paper, Pennsylvania State
University.
[79] Shleifer, A. and R. Vishny, 1989, Management entrenchment: The case of manager-specific investment,
Journal of Financial Economics, 25, 123-140.
[80] Shleifer, A. and R. Vishny, 1997, A survey of corporate governance, Journal of Finance, 52, 737-783. [81] Stoll, H.,
1989, Inferring the components of the bid-ask spread: Theory and empirical evidence,
Journal of Finance 44, 115-134.
[82] Stoll, H.R., 1978, The pricing of security dealer services: an empirical study of NASDAQ stocks,
Journal of Finance 33, 1153-1172.
[83] Symonds, B., The Long, Hard Road Ahead for Tyco, Business Week, July 2002.
[84] Tinic, S., 1972, The economics of liquidity services, Quarterly Journal of Economics 86, 79-93.
[85] Tinic, S., West, R., 1974, Marketability of common stocks in Canada and the U.S.A.: a comparison
of agent versus dealer dominated markets, Journal of Fina nce 29, 729-746.
[86] Van Ness, R. Van Ness, and R. Warr, 2001, "How well do adverse selection components measure
adverse selection?" Financial Management, 30, 3, 77-98.
[87] Volpin, P., 2002, Governance with poor investment protection: Evidence from top executive turnover
in Italy, Journal of Financial Economics, 64, 61-91.
[88] Weisbach, M., 1988, Outside directors and CEO turnover, Journal of Financial Economics, 20, 431-
460.
[89] Weston, J.P., 2000, Competition on the Nasdaq and the impact of recent market reforms, Journal of
Finance 55, 2565-2598.
[90] Yermack, D., 1996, Higher valuation of companies with a small board of directors, Journal of Financial
Economics, 40, 185-212.
[91] Yermack, D., 1997, Good timing: CEO stock option awards and company news announcements,
Journal of Finance, 52, 449-476.
[92] Zucca, L. and D. Campbell, 1992, A closer look at discretionary write-downs of impaired assets,
Accounting Horizons, 6, 30-41.



























68

doc_615807554.docx