Dissertation Report on Financial Policy and Ownership Stability

Description
Economic policy refers to the actions that governments take in the economic field. It covers the systems for setting interest rates and government budget as well as the labor market, national ownership, and many other areas of government interventions into the economy.

ABSTRACT
Title of dissertation: FINANCIAL POLICY AND
OWNERSHIP STABILITY
Matthew Lee Kozora
Doctor of Philosophy, 2011
Dissertation directed by: Associate Professor Nagpurnanand Prabhala
R.H. Smith School of Business
Department of Finance
I investigate the relationship between corporate …nancial policy and the ownership
stability of a …rm’s institutional shareholders. In each chapter of my dissertation I
empirically investigate this relationship in a di¤erent setting: the …rst chapter with
respect to earnings management, the second chapter with respect to corporate spin-
o¤s, and the third chapter with respect to payout policy. Unique to my research I
utilize the complete ownership history of each institutional stock position to create
measures of ownership stability including fund investment horizon and ownership
length. Overall, I …nd signi…cant relationships between each one of the three …nancial
policies and measures of ownership stability.
FINANCIAL POLICY AND OWNERSHIP STABILITY
by
Matthew Lee Kozora
Dissertation submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial ful…llment
of the requirements for the degree of
Doctor of Philosophy
2011
Advisory Committee:
Associate Professor Nagpurnanand Prabhala, Chair/Advisor
Associate Professor Russell Wermers
Assistant Professor Gerard Hoberg
Professor Marc Nerlove
Professor Lemma Senbet
c _ Copyright by
Matthew Lee Kozora
2011
Acknowledgments
I thank everyone who have made this thesis possible. First and foremost I thank
my parents, John and Joan Kozora, for giving me their love, support, and encour-
agement throughout my years of schooling. I thank my siblings Karen, John, and
Shaun for encouraging me and giving me a roadmap for success with their own ac-
tions. Lastly, I thank Dr. Jessica Hennessey for her companionship from math-camp
to graduation.
I thank Associate Professor Nagpurnanand Prabhala, my advisor, along with the
other members of my dissertation committee including Associate Professor Russell
Wermers, Professor Lemma Senbet, Assistant Professor Gerard Hoberg, and Professor
Marc Nerlove. My advisor and the members of the committee did an exceptional job
of guiding me when necessary but also allowing me to trail my own path.
ii
Table of Contents
List of Tables v
Chapter 1 - Earnings Management 1
1.1 Introduction 1
1.2 Earnings Management Literature 7
1.3 Firm Data 11
1.3.1 Firm Variables 11
1.4 Fund Investment Horizon 14
1.4.1 Measuring Ownership Length 14
1.4.2 Fund Investment Horizon, Mutual Fund Inclusion, & Sum-
mary
16
1.4.3 Comparison to other Measures of Investment Horizon 19
1.4.4 Ownership Stability Measures 22
1.5 Earnings Announcements within One Penny of Analyst Forecasts 24
1.5.1 Level of Earnings Surprise 24
1.5.2 Changes in Analyst Forecasts 27
1.6 Discretionary Accruals 29
1.6.1 Level Regressions 29
1.6.2 Di¤erence Regressions 34
1.7 Chapter Conclusion 36
Chapter 2 - Corporate Spin-o¤s 37
2.1 Introduction 37
2.2 Spin-o¤ Literature 43
2.3 Data 45
2.3.1 Spin-o¤ Event Data 46
2.3.2 Firm-Level Variables 47
2.4 Changes in Fund Ownership 49
2.4.1 Fund Ownership Variables 49
2.4.2 Univariate Tests of Fund Ownership Changes 51
2.4.3 Multivariate Regressions Explaining Adjusted Ownership
Changes
53
2.4.4 Relation to Post-Event Returns 57
2.4.5 Chapter 2.4 Summary 60
2.5 The Case of Pre-Existing Shareholders 61
2.5.1 Ownership Patterns 61
2.5.2 Multivariate Regressions 62
2.6 Ownership Stability Before & After Spin-o¤ Events 64
2.7 Chapter Conclusion 70
Chapter 3 - Payout Policy 71
3.1 Introduction 71
3.2 Payout Literature 78
3.3 Firm Data 81
iii
3.3.1 Sample 81
3.3.2 Payout Event Speci…cations & Event Speci…cations 82
3.3.3 Control Variables 83
3.4 Fund Ownership Characteristics 85
3.4.1 Determinants of Fund Ownership 87
3.4.2 Determinants of Ownership Length 90
3.5 Ownership Changes Around Payout Events 93
3.5.1 Changes in Shareholder Investment Horizon 94
3.5.2 Changes in Ownership Length 97
3.5.3 Explaining Changes in Shareholder Investment Horizon &
Current Ownership Length with Fund Investment Horizon
Tercile Ownership Changes
98
3.6 The E¤ect of The JGTRRA 99
3.6.1 Ownership Characteristics 100
3.6.2 Ownership Changes Around Payout Events 102
3.7 The Efect of Ownership Stability on Payout Choice 104
3.7.1 Pre-Event Ownership Comparison & Change in Fund Own-
ership
104
3.7.2 Tests of Pre-Event Fund Ownership 106
3.8 Chapter Conclusion 109
Tables 111
List of References 161
iv
List of Tables
1.1 Fund Investment Horizon Summary Statistics 111
1.2 Average Change in Fund Investment Horizon from Initial Measure 112
1.3 Comparison Between Fund Investment Horizon & Other Measures
of Portfolio Turnover
113
1.4 Firm and Ownership Variable Correlation Matrix 114
1.5 Summary of Earnings Announcements within One Penny of Ana-
lyst Forecasts
115
1.6 Ordered Probit Regressions Describing Earnings Surprises 116
1.7 Linear Regressions Describing Earnings Surprises 117
1.8 Linear Regressions Describing Changes in Analyst Forecasts 118
1.9 Level Regressions Describing Discretionary Accruals - Shareholder
Composition
119
1.10 Level Regressions Describing Discretionary Accruals - Ownership
Length
120
1.11 Level Regressions Describing Discretionary Accruals - Shareholder
Composition & Ownership Length
121
1.12 Di¤erence Regressions Describing Changes in Discretionary Ac-
cruals
122
2.1 Spin-o¤ Events By Announcement Year 123
2.2 Percentage of Shares Held Before & After Spin-o¤ Event 124
2.3 Fund Ownership Changes 125
2.4 Panel Regressions Describing Changes in Adjusted Ownership 127
2.5 Tests of Mean and Median Abnormal Returns 129
2.6 Panel Regressions Describing Changes in Adjusted Returns 130
2.7 Pre-Existing Fund Shareholder Ownership Patterns 132
2.8 Panel Regressions Describing Pre-Existing Fund Shareholder
Ownership Patterns
133
2.9 Panel Regressions Describing Changes in Fund Ownership Before
& After Spin-o¤ Events
134
3.1 Fund Ownership by Firm Type - Size, Market-to-Book, & Payout
Policy
135
3.2 Determinants of Ownership Proportion by Fund Investment Hori-
zon Tercile
136
3.3 Determinants of Relative Ownership Length by Fund Investment
Horizon Tercile
138
3.4 Shareholder Investment Horizon Changes Around Payout Events 140
3.5 Current Ownership Length Changes Around Payout Events 142
3.6 Adjusted Change in Ownership Percentage Around Payout Events 144
3.7 The E¤ect of the JGTRRA on Fund Ownership Characteristics 147
3.8 The Di¤erence in Shareholder Investment Horizon Changes Before
& After the JGTRRA
149
3.9 The Di¤erence in Current Ownership Length Changes Before &
After the JGTRRA
151
v
3.10 The Di¤erence in Ownership Percentage Changes Before & After
JGTRRA (Before - After) by FIH Tercile
153
3.11 Fund Ownership Comparisons Around Dividend Increases &
Share Repurchases
155
3.12 Bivariate Probit Models Describing Payout Choice 157
3.13 Fund Ownership Comparisons Prior to Dividend Increases &
Share Repurchases
159
vi
Chapter 1: Earnings Management
1.1 Introduction
In the 2004 Keynote Lecture at the Financial Management Association meet-
ings, Michael Jensen discussed the agency costs associated with overvalued equity
(Jensen(2005)). Jensen argued that although a high stock price seems ideal, justify-
ing overvalued equity ultimately leads to investment policy distortion and account-
ing manipulation. Graham, Harvey, and Rajgopal (2005) …nds support for some of
Jensen’s arguments after conducting a CFO survey concerning their views of earnings
management. 80% of surveyed CFOs admit they would forego discretionary spending
on real items like research and development, maintenance, and advertising to meet
earnings targets.
Jensen places the blame on the emphasis compensation structures and capital mar-
kets place on meeting performance targets. However, the emphasis placed on meeting
earnings targets may be related to the characteristics of institutional ownership. Firm
managers, as hired agents, will match the investment horizon of the …rm with the in-
vestment horizon of their shareholders for fear of removal. Even outside the fear of
removal, there are other reasons to believe why …rm managers should consider share-
holder investment horizon composition around earnings announcements. For instance,
systematic selling by short horizon investors can place signi…cant price pressure on
equity value, causing declines in liquidity and ultimately share value. Decreases in
share-value, even over the short-term, can impact multiple corporate policies includ-
ing managerial compensation, supplier contracts, and capital costs (Graham, Harvey,
and Rajgopal (2005)).
On the other hand, …rms held by longer-term institutional shareholders will take a
longer-term approach to the management of the …rm and thus place less emphasis on
meeting performance benchmarks. Although liquidity contraints may be a concern
for some long horizon investors and thus may not be agnostic toward short-term price
1
movements, generally these shareholders will be more concerned with the long-term
fundamental value of the …rm and less concerned with short-term price movements.
In this chapter, I empirically study the relationship between earnings management
and the ownership stability of a …rm’s institutional shareholders. I measure earnings
management with signed and unsigned discretionary accruals as well as the di¤erence
between earnings-per-share announcements and analyst forecasts. I control for own-
ership stability with the investment horizon composition and the ownership length
of a …rm’s institutional shareholders. Overall, I …nd strong evidence indicating fund
ownership, particularly by funds with shorter investment horizons, is important in de-
scribing earnings management. I also …nd some evidence indicating longer ownership
by funds with shorter investment horizons is positively related to upwards earnings
management, but longer ownership by funds with longer investment horizons is neg-
atively related to its overall level (unsigned discretionary accruals).
I take institutional stock positions at the fund level from the Thomson Reuters
(S12) Mutual Fund dataset. The dataset consists of positions from most domestic
mutual funds and some global funds that participate in US and Canadian equity
markets. The primary source for the dataset is SEC N-30D …lings. Although for the
majority of the time period the SEC required mutual funds to …le this form semi-
annually, Thomson Reuters supplements the …lings by examining fund prospectuses
and contacting mutual funds directly. The other approach is to use the Thomson
Reuters (13f) Investment Company dataset consisting of aggregate holdings of banks,
insurance companies, parents of mutual funds, pensions, and endowments. The pri-
mary source for this dataset is quarterly SEC 13f …lings required by all institutional
investment managers that exercise investment discretion over $100 million.
A fund’s investment horizon is equal to the average length of time (in months)
each share of every stock positions is held from the date of initial stock investment
to the date of measurement. Using the full ownership history of each stock position
2
to classify a fund’s investment horizon is a departure from past literature which uses
either a range of portfolio characteristics (e.g. Bushee (1998), and Hotchkiss and
Strickland (2003)) or portfolio turnover (e.g. Gaspar, Matos, and Massa (2005), and
Yan and Zhang (2009)). By using mutual fund holding data and measuring an in-
stitution’s investment horizon in this manner, I am able to increase the number of
institutional shareholders in my dataset and create a more precise measure of share-
holder investment horizon that is more directly related to the corporate governance
aspects of institutional ownership.
I …rst investigate the relationship between ownership stability and earnings man-
agement with the di¤erence between earnings-per-share announcements and analyst
forecasts. Speci…cally, I investigate whether ownership stability is related to the ten-
dency of …rms to just beat (by one penny), meet, or just miss (by one penny) analyst
earnings forecasts. These tests have several advantages on past work. First, the level
of earnings surprise directly relates to changes in share value. Past research …nds the
marginal e¤ect of an additional penny of announced earnings on share value is great-
est when earnings announcements are within one penny of analyst forecasts, and a
relative greater decrease in share value as a result of just missing earnings compared
to the increase in share value as a result of just beating (i.e., the torpedo e¤ect).
Second, earnings announcements that just beat analyst forecasts are likely to come
from changes in discretionary spending than economic surprises. Earnings surprises
that are economic based are more likely to be away from analyst forecasts. Third,
analysts presumably account for many of the …rm-level characteristics that could in-
directly relate ownership stability to the tendency of …rms to manage earnings thus
making for a more robust test. Lastly, although the use of discretionary accruals to
manage earnings may be strongly related to …rm type and thus a basis for greater
ownership by some fund types (i.e., those with shorter investment horizons), it is un-
likely funds take ownership in …rms with an explicit expectation announced earnings
3
will beat analyst forecasts.
I estimate ordered probit models using panel data from 1990 to 2007 explaining the
level of earnings surprise rounded to the nearest cent. I again estimate separate models
…rst controlling for ownership stability with measures of shareholder composition,
then measures of ownership length, and …nally both. Similar to the results with
respect to signed discretionary accruals, I …nd greater ownership by funds with shorter
investment horizons are more likely to have a positive earnings announcement. I also
…nd some evidence indicating longer ownership by short horizon funds is positively
related to the level of earnings surprise.
I extend the tests of earnings surprise in two ways. First, I re-model the level of
earnings surprise with linear regressions partitioning the sample to …rms with earn-
ings surprise strictly greater than one penny, strictly less than minus one penny, and
both. I estimate these regressions to better distinguish between two alternative expla-
nations of the initial results. Although greater ownership by short horizon funds may
in‡uence fund managers to act myopically, short horizon funds may instead simply
be using their information advantage to take positions in …rms with positive earnings
announcements. By investigating earnings surprises that are more likely to be caused
by economic changes, I can better di¤erentiate between these two explanations. I …nd
no indication the level short horizon fund ownership is signi…cantly related to the level
of earnings surprise for …rms with earnings surprises strictly greater than one penny.
However, I do continue to …nd evidence indicating short horizon fund ownership is
signi…cantly related to the leve of earnings surprise for …rms with earnings surprises
strictly greater and strictly less than minus one penny. However, the signi…cance of
short horizon ownership does decrease. These results indicate that although short
horizon funds take greater ownership with positive earnings surprises, the relation is
stronger for those …rms where earnings manipulation is more likely.
Second, I investigate whether fund ownership stability is related to changes in
4
analyst forecasts over the …nal month of the …scal year and whether this relationship
can explain the di¤erence between announced and forecasted earnings. Although
greater shorter-term institutional ownership may lead to greater earnings forecast
management, analysts may use characteristics of institutional ownership when up-
dating forecasts. Continuing to use only …rms announcing earnings within one penny
of analyst forecasts, I …nd a dichotomous relationship between analysts updates and
the two characteristics of ownership stability. First, analysts are more likely to in-
crease earnings forecasts when a …rm has greater ownership by funds with shorter
investment horizons. This result suggests analysts anticipate upwards earnings man-
agement when there is a greater focus on short-term returns. Second, analysts are
more likely to increase earnings forecasts when …rms are held longer by long horizon
funds, suggesting a decrease in earnings management when the …rm has long-term
dedicated investors. Additional evidence indicates the relationship between share-
holder composition and the change in median analyst forecasts is not as strong for
…rms eventually having a positive earnings surprise. This last result indicates either
greater expectations management by …rm executives or a failure by analysts to fully
account for short horizon fund ownership.
I next model both signed and unsigned discretionary accruals as a function of the
investment horizon composition and the ownership length of a …rm’s fund sharehold-
ers. I estimate signed and unsigned discretionary accruals using a modi…ed version of
the Jones (1991) model. I use signed discretionary accruals to test for the direction
of earnings management, and I use unsigned discretionary accruals to test for the
overall shifting of revenues and expenses between periods to smooth earnings (or sim-
ply the overall level of earnings management). To classify shareholder composition, I
use either average fund shareholder investment horizon or the percentage of common
shares held by funds classi…ed into one of three groups (short, medium, and long)
based on annual investment horizon tercile breakpoints. I measure ownership length
5
with the average percentage of other stock positions held within fund portfolios for a
strictly shorter period of time. This measure, with a range from zero to one, relates
to di¤erences between …rms in the length of time they have been held by the same
fund shareholder. I also create similar measures of ownership length for each fund
investment horizon tercile.
I …rst estimate regressions describing the level of signed and unsigned discretionary
accruals using …rm data from 1990 to 2007. I estimate separate regressions …rst con-
trolling for ownership stability with just measures of shareholder composition, then
ownership length, and …nally both. I …nd …rms with greater ownership by funds with
shorter investment horizons have greater levels of signed and unsigned discretionary
accruals. This result is driven primarily by greater ownership by short horizon funds
than less ownership by long horizon funds. Interestingly, I …nd greater ownership
by medium horizon funds is positively related to signed discretionary accruals but
negatively related to unsigned discretionary accruals, suggesting a push-and-pull be-
tween short term gains and long term value when the investment horizon of the fund
is neither short nor long. I also …nd ownership length to be a signi…cant determi-
nant. Speci…cally, …rms held longer by funds with shorter investment horizons are
more likely to manage earnings upward, but longer ownership by funds with longer
investment horizons are less likely to manage earnings overall.
In other tests, I …nd total fund ownership is positively related to signed discre-
tionary accruals. Thus, mutual fund ownership in general is positively related to the
emphasis placed on short term performance. On the other hand, I do not …nd evi-
dence indicating overall mutual fund ownership is either signi…cant by itself or alters
the signi…cance of shareholder composition with respect to unsigned discretionary
accruals.
The results from the level regressions demonstrate a strong negative relationship
between ownership stability and the use of discretionary accruals to manage earn-
6
ings. However, the results could stem from …rm-level …xed e¤ects that explains both
the stability of institutional shareholders and the propensity to manage earnings. To
determine if my initial results are robust to this potential explanation, I estimate dif-
ference regressions explaining changes in signed and unsigned discretionary accruals.
I continue to …nd strong evidence indicating a positive relationship between ownership
by funds with shorter investment horizons and signed discretionary accruals. How-
ever, I no longer …nd consistent evidence indicating a signi…cant relationship between
either the level of short horizon fund ownership and unsigned discretionary accruals
or measures of ownership length and the two discretionary accrual variables. Thus,
some of the evidence especially pertaining to ownership length in the level regressions
may be driven partially by …rm type.
Two broad conclusions can be taken from the …ndings. First, institutional own-
ership stability negatively relates to the level of earnings management. Second, both
shareholder composition and ownership length may be important when controlling for
the presence of institutional shareholders. Although this chapter does estimate sev-
eral tests attempting to distinguish between possible explanations, it does not provide
evidence of causality. In order to determine causality, tests using exogenous shocks
are needed to determine whether ownership is a signi…cant factor in determining man-
agerial behavior. This chapter instead provides strong evidence indicating di¤erent
aspects of institutional ownership is signi…cantly related to earnings management,
providing a necessary basis for future tests investigating causality.
1.2 Earnings Management Literature
In this section, I discuss in further detail the contributions to the earnings man-
agement literature made in this work.
Rajgopal and Venkatachalam (1997); Rajgopal, Venkatachalam, and Jiambalvo
(1999); Koh (2003); and Burns, Kedia, and Lipson (2006) use the levels of either
signed or unsigned discretionary accruals to investigate the corporate governance as-
7
pects of institutional ownership on corporate governance as it relates to earnings
management. All papers provide evidence indicating managers are less likely to man-
age earnings when institutional ownership is high.
1
Among these papers, only Burns
et al. (2006) classify institutions by type. They use the classi…cation scheme devel-
oped by Bushee (1998), who places institutions at the management company level
into one of three groups, "transient," "quasi-indexer," and "dedicated," based on
portfolio characteristics including position size, portfolio turnover, and trading sen-
sitivity to earnings news. Portfolios of transient institutions exhibit a high degree of
diversi…cation, high portfolio turnover, and are sensitive to …rm earnings. Conversely,
portfolios of "dedicated" institutions have a low degree of concentration, low portfolio
turnover, and low sensitivity to current earnings. Portfolios of "quasi-indexer" insti-
tutions exhibit a high degree of diversi…cation but low turnover. Burns et al. (2006)
…nd greater ownership by transient institutions and less ownership by dedicated and
quasi-indexer institutions is related to higher unsigned discretionary accruals. They
conclude the results are evidence of less monitoring performed by "non-dedicated"
institutions resulting in poorer earnings quality.
2
In Chapter 1.6 of this chapter, I also investigate the e¤ect of institutional own-
ership on the level of discretionary accruals. I extend the previous work in three
primary respects. First, I investigate the relationship between ownership stability
1
Yu (2008), in tests investigating the governance aspect of analyst coverage, …nds institutional
ownership to be an insigni…cant determinant of unsigned discretionary accruals.
2
The reliability and frequency of earnings announcements are two other characteristics poten-
tially in‡uenced by the level of institutional ownership. Velury and Jenkins (2006) …nd greater
institutional ownership is positively related to higher earnings quality as measured by its predic-
tive nature, neutrality, timeliness, and representational informativeness. Anjinkya, Bhojraj, and
Sengupta (2005) …nd evidence in both level and change regressions indicating …rms with greater
institutional ownership report earnings forecasts with more frequency, greater speci…city, and with
less bias. Consistent with the less dedicated institutional investors preferring to invest in …rms with
greater transparency, Bushee and Noe (2001) …nd a positive relationship between changes in corpo-
rate disclosure practices and ownership by transient institutions. However, Burns et al. (2006) …nd
…rms are more likely to have …nancial restatements and more severe restatements when transient
institutional ownership is high. These authors also …nd evidence indicating transient institutions
are more likely to sell their shareholdings at the announcement of the restatement. Hribar, Jenkins,
and Wang (2004) …nd similar evidence.
8
and the level of both signed and unsigned discretionary accruals, making compar-
isons between the two measures of earnings management. Second, I use measures of
both shareholder composition and ownership length. Lastly, I also estimate di¤erence
regressions modeling changes in discretionary accruals to account for …rm-level …xed
e¤ects.
Past research …nds strong evidence indicating …rm managers manipulate earn-
ings announcements to meet or beat benchmarks, and positive announcements have
a positive e¤ect on share value. Burgstahler and Dichev (1997) …nd a dispropor-
tionately low frequency of …rms reporting small decreases in earnings and income
compared to the number of …rms reporting small increases. The authors also …nd
changes in cash ‡ow from operations and working capital are used to achieve the
small gains. Degeorge, Patel, and Zeckhauser (1999) …nd evidence indicating …rm
executives manage earnings to report positive pro…ts, sustain recent performance,
and meet analyst expectations. Burgstahler and Eames (2006) …nd evidence indicat-
ing both upward earnings management with operating cash ‡ows and discretionary
accruals, and downward management of analyst forecasts to achieve positive or zero
earnings surprise.
Bartov, Givoly, and Hayn (2002) and Skinner and Sloan (2002) document a dis-
proportionate decrease in share price as a result of just missing analyst forecasts
compared to just beating analyst forecasts (i.e. the torpedo e¤ect), as well as a
greater e¤ect on stock price associated with reporting one additional penny when
announced earnings are closer to analyst expectations. This relation holds even when
analyst forecasts were lowered prior to the earnings announcement date (Bartov et
al. (2002)), and is stronger for growth …rms than value …rms (Skinner and Sloan
(2002)). Other evidence from Kinney, Burgstahler, and Martin (2002) and Kasznik
and McNichols (2002) …nds a stronger reaction to earnings surprises when forecast
dispersion is low, and an overall greater market premium only for …rms consistently
9
meeting earnings forecasts.
Several authors investigate the role of institutional ownership in earnings an-
nouncements. Cheng and Reitenga (2001), Chung, Firth, and Kim (2002), and Hsu
and Koh (2005) …nd greater overall ownership by institutions reduces the use of dis-
cretionary accruals especially for …rms more likely to manage earnings to meet or beat
benchmarks. Distinguishing institutional ownership using Bushee’s (1998) classi…ca-
tion scheme, Matsumoto (2002) and Koh (2007) …nd a positive relationship between
transient institutional ownership and the use of discretionary accruals to meet or
beat benchmarks. Bushee (1998) and Roychowdhury (2006) investigate changes in
real economic activities as a means to meet earnings benchmarks. Bushee (1998)
…nds …rms are less likely to cut research and development expenses to reverse an
earnings decline when they have greater overall institutional ownership, but are more
likely to reverse an earnings decline when transient institutional ownership is high.
Roychowdhury (2006) also …nds little supporting evidence indicating greater overall
institutional ownership leads to either reductions in price discounts, discretionary ex-
penditures, or the overproduction of goods to reduce costs to avoid annual income
losses.
In Chapter 1.5 of this chapter, I also investigate benchmark-related earnings man-
agement. I use …rms that announce earnings-per-share results within one penny of
median analyst forecasts, and distinguish between …rms that either just beat, meet,
or just miss analyst forecasts. A similar regression is estimated by Matsumoto (2002).
However, Matsumoto uses all …rm observations regardless of the level of earnings sur-
prise and classi…es …rms into only one of two groups depending on whether analyst
forecasts were at least met. Like the rest of this chapter, I also incorporate mea-
sures of shareholder composition and ownership lengths. Furthermore, I extend the
research by also investigating the relationship between institutional ownership and
analyst forecast updates.
10
1.3 Firm Data
I extract …rm observations from the Compustat Fundamentals Annual data …le.
Earnings report data are taken from I/B/E/S. Return and other share information
are taken from the Center for Research in Securities Prices (CRSP) monthly stock
return …le. I use all …rm observations from 1990 to 2007 that have 3 consecutive
years of …nancial data, 36 consecutive months of return data, and ordinary common
stock (CRSP share code 10 or 11) listed on the NYSE, AMEX, or NASDAQ (CRSP
header exchange code 1, 2, or 3). I do not exclude utilities (SIC codes 4949 to 4999)
or …nancial companies (SIC codes 6000 to 6999) from the analysis. The results do
not change if I instead exclude these …rms.
1.3.1 Firm Variables
For control variables in the tests below, I use …scal year stock return (FYRet),
stock return over the …nal three months of a …rm’s …scal year (3MRet), …rm size
(Size), market-to-book ratio (MB), debt (Debt), analyst forecast standard deviation
(FcstSD), and analyst number (AnNum). I derive the return variables from CRSP;
Size, MB, and Debt from Compustat; and FcstSD and AnNum from I/B/E/S.
« FYRet
t
= Compounded monthly returns over …scal year t
__

m2[m1;m12]
(1 + :ct
m;t
)
_
÷1
_
, where :1 designates the …rst month of the
…scal year, and :12 designates the last.
« 3MRet
t
= Compounded monthly returns over the …nal three months in …scal
year t
__

m2[m10;m12]
(1 + :ct
m;t
)
_
÷1
_
.
« Size
t
= The natural log of total assets (data6 or at).
« MB
t
= Fiscal year end market value divided by book value (MV
t
,BV
t
). Book
value (BV)is equal to the sum of total assets, deferred tax and investment credit
(data35 or txditc), and convertible debt (data79 or dcvt), minus preferred stock
(data10 or pstkl) and total liabilities (data181 or lt).
11
« Debt
t
=Total long termdebt (data9 or dltt) divided by total assets (data9
t
,data6
t
).
« FcstSD
t
= Standard deviation of analyst forecasts (stdev).
« AnNum
t
= The number of analysts covering the …rm (numest).
I use stock return variables to control for recent …rm performance, and Size and
MB control for …rm type. Growth …rms and smaller …rms may engage in greater
earnings management to enhance their reputation with stakeholders (Graham et al.
(2005)). Debt controls for the likelihood of debt covenant violation. FcstSD and
AnNum controls for the informational environment surrounding the …rm.
1.3.2 Earnings Management Measures
I use three measures of earnings management. The …rst measure of earnings
management, discretionary accruals (DA), is computed using a modi…ed version of
the Jones (1991) model.
3
DA is equal to the di¤erence between total accruals (TA)
and non-discretionary accruals. In year t, TA is equal to the sum of the changes in
current assets (data4 or act) and debt and current liabilities (data34 or dlc), minus
the change in current liabilities (data5 or lct), the change in cash and short term
investments (data1 or che), and depreciation (data14 or dp). All changes occur from
year t ÷1 to year t. In equation form, TA for …rm i is equal to
TA
i;t
= (act
i;t
÷act
i;t1
) + (dlc
i;t
÷dlc
i;t1
)
÷(lct
i;t
÷lct
i;t1
) ÷(che
i;t
÷che
i;t1
) ÷dp
i;t
(1)
Non-discretionary accruals are equal to the …tted values from annual linear re-
gressions describing TA. I regress total accruals on property, plant, equipment (data7
or ppegt), operating income before depreciation (data13 or oibdp), the di¤erence be-
tween changes in total receivables (data12 or rect) and common equity (data2 or ceqt)
3
Dechow, Sloan, and Sweeney (1995).
12
from year t ÷ 1 to year t, and a constant. I winsorize total accruals annually at the
5
th
and 95
th
percentiles and scale all variables with lagged total assets. In equation
form, the linear model can be written as
TA
i;t
at
i;t1
= ,
0;t
1
at
i;t1
+ ,
1;t
oibpd
i;t
at
i;t1
+ ,
2;t
(rect
i;t
÷ rect
i;t1
) ÷
_
ceqt
i;t
÷ ceqt
i;t1
_
at
i;t1
+ c
i;t
(2)
where ,
0
, ,
1
, and ,
2
are model parameters, and c is model error. I estimate separate
regressions for each Fama-French 48 Industry Classi…cation subject to at least 10
…rms having full information. In equation form, discretionary accruals is equal to
DA
i;t
=
TA
i;t
at
i;t1
÷
´
,
0;t
1
at
i;t1
÷
´
,
1;t
oibpd
i;t
at
i;t1
÷
´
,
2;t
(rect
i;t
÷ rect
i;t1
) ÷
_
ceqt
i;t
÷ ceqt
i;t1
_
at
i;t1
(3)
where ´ represents model estimates. The second measure of earnings management,
unsigned discretionary accruals (UnsDA), is equal to the absolute value of DA ([DA[).
The third measure of earnings management, the level of earnings surprise (ES),
is equal to the di¤erence between announced earnings and median analyst forecasts.
I take annual forecast data and announced earnings-per-share for all U.S. …rms from
the I/B/E/S database.
I use median analyst forecasts the …nal month of the …rm’s …scal year-end as
the earnings benchmark. I use I/B/E/S summary statistics taken from the statsum
data…le. ES is equal to actual announced earnings (actual) minus the median of
analyst forecasts (medest). I/B/E/S does adjust share-based summary statistics for
corporate actions such as stock splits. I round ES to the nearest penny due to its
widespread usage in the business press and evidence presented by Das and Zhang
(2003) indicating …rm managers round in order to report additional cents. I also
13
conduct tests using analyst forecasts the month prior to the end of the …scal year
because of its potential use by …rm managers as a benchmark when manipulating
accruals (Bhojraj, Hribar, Picconi, and McInnis (2009)). However, because there is
little di¤erence in the results, I do not report them.
1.4 Fund Investment Horizon
In this section, I describe the methodology used to measure the length of time
funds hold stock positions. I also de…ne my measure of fund investment horizon and
provide annual summary statistics. I then compare the measure of fund investment
horizon in this chapter with two other turnover-based measures.
4
1.4.1 Measuring Ownership Length
I extract the following information from the S12 database for all fund positions
to measure ownership length. I index mutual funds with i, and their individual stock
positions with ,.
« rdate
i
(current report date) = The date at which institutional holdings are valid.
I index fund report dates with t.
« S
i;j
(shares held) = The number of …rm shares held by the institution as of the
current report date.
From these two variables I create …ve additional variables.
« prdate
i
(prior report date) = The fund’s most recent report date prior to the
current report date.
« PS
i;j
(shares held at the prior report date) = The number of …rm shares held
by the fund as of the prior report date.
4
I do not repeat this section in Chapters 2 or 3. However, I do rede…ne measures of shareholder
investment horizon and ownership length when necessary.
14
« S
i;j
(change in shares held) = The change in the number of …rm shares held
by the fund from the prior report date to the current report date. I assume that
all portfolio changes from one report date to the next occur on the later date.
« bdate
i;j
(position begin date) = The fund’s most recent report date which sat-
is…es PS
i;j
= 0 and S
i;j
0.
« cdate
i;j
(position closure date) = The fund’s next report date which satis…es
PS
i;j
0 and S
i;j
= 0.
Each year, I calculate the average length of time a fund invests in a stock position
as the average number of months each share of stock is held from the position begin
date to the date of measurement. The date of measurement is equal to the fund’s last
report date in a given year. I measure average ownership length for all stock positions
held for at least one month from the beginning of the year to the fund’s last report
date in the year. This includes stock positions closed prior to the fund’s last report
date or opened over the course of the year. Also, I take as separate positions of the
same stock held at two or more disjoint periods of time within the same year. Stock
positions opened on the date of measurement have no ownership length and are not
used until the following year.
I employ the last-in-…rst-out queueing method to measure the length of time each
share in a stock position is held. That is, I assume the next set of stock , shares sold
are the ones currently held for the shortest period of time. The purchase date for
shares s
i;j
, t
p
, is equal to the fund report date such that S
i;j
0 and s
i;j
¸ S
i;j
.
The sale date, t
s
, is the next report date that satis…es the following equation.

s

=
p
+1
¸
¸
(S
i;j;
)

¸
¸
_ PS
i;j;
p + s
i;j;
p. (4)
The left hand side of the inequality represents the aggregate number of stock , shares
sold from the purchase date to the sale date, whereas the right hand side represents
15
the sum of shares eligible for sale as of t
p
. The total number of shares purchased on t
p
that I designate as being sold on t
s
is equal to the maximum number of shares, s

i;j;
p,
satisfying Equation (??) with equality. I designate the remaining shares purchased
on t
p
not sold on t
s
(s
0
i;j;
p ¸ S
i;j;
p, s
0
i;j;
p ,=s

i;j;
p) as held until a future report
date.
The length of time (LT) a share is held is equal to the number of months from
its purchase date to either the funds last report date in the year if the share remains
held, or the share’s sale date if the share was sold. The average length of time fund
i holds share / of stock , at the end of year t is equal to
LT
i;j;t
=
N
k

k=1
LT
i;j;k;t
N
k;t
(5)
where N
k
is equal to the number of purchased shares from the position begin date
until the date of measurement.
1.4.2 Fund Investment Horizon, Mutual Fund Inclusion, & Summary
Fund investment horizon (FIH) is equal to the value-weighted average length of
time a fund invests in a stock position. In equation form, FIH is written as
FIH
i;t
=

j2J
MV
i;j;t
+ LT
i;j;t

j2J
MV
i;j;t
(6)
where MV represents a stock position’s market value, and J represents the set of
eligible stock positions. If the position remains open as of the fund’s last report date
in the year, then MV is equal to the equity price times the number of shares held
as of this date. Again, I do not include purchased shares on this date as part of the
calculation. In case the position closes prior to the date of measurement, MV is equal
to the equity price times the number of shares sold on the closure date.
16
There are two primary requirements for fund inclusion. First, a fund must be
present in the dataset and meet SEC …ling requirements for the previous three years,
along with at least one …ling in the previous fourth. After this startup period, if in
any year the fund does not meet the minimum SEC …ling requirements, I drop the
fund and its holdings from the …nal sample until an additional startup period can
be completed. A gap of more than one year between report dates for the same fund
identi…cation number typically indicates a "di¤erent and unrelated" fund.
5
Assuming
unrelated funds hold di¤erent stock positions, investment horizon calculation may
measure position changes from the …nal holdings of the original fund to the …rst set
of holdings of the new fund, biasing FIH downward. Second, for each year I require
a fund to have at least twenty eligible stock positions for FIH calculation to ensure
meaningful investment horizon calculation. Lastly, I drop all index-related funds from
my sample.
The total number of fund-year combinations from 1990 to 2007 on the S12 dataset
is equal to 188,796. Among these …rm-year observations 52,538 funds do not meet
SEC …ling requirements during the calendar year. An additional 92,247 funds do not
meet …ling requirements over the previous three years. Although the number of fund
loss is large, it underscores the transitory nature of mutual funds in the dataset. For
instance, 40,017 funds did not meet SEC …ling requirements over two consecutive
years. Out of the remaining 44,011 funds, 18,106 are non-index related and hold
twenty or more sample …rms.
Table 1.1 presents fund investment horizon summary statistics for each year from
1990 to 2007. Columns 2 through 6 report the number of funds as well as annual
summary statistics for FIH including the mean, standard deviation, minimum, and
maximum. The total number of sample funds increases each year starting from 215 in
1990 to 2,537 in 2007, with the greatest increase occurring after 1996. From 1990 to
5
See the User’s Guide to Thomson Financial Mutual Fund and Investment Company Common
Stock Holdings Databases on WRDS.
17
1996, the number of funds in the dataset increased by 211. Between 1996 and 2007,
an additional 2,111 fund observations enter the dataset. Mean FIH ranges from a
minimum of 18.8 months in 2001 to a maximum of 24.8 months in 1994 and 1995.
The annual minimum of FIH ranges from 2.4 months to 5.8 months. Typically, the
most number of fund report dates in a given year is four. Thus, funds that hold shares
for very short periods of time will have an average investment horizon in this range.
Minimum investment horizons less than 3 months were the result of abnormally close
report dates over the measurement period. The annual maximum of FIH ranges from
79.4 months in 1990 to 223.1 months in 2005.
Columns 7 and 8 presents tercile breakpoints distinguishing between short, medium,
and long investment horizon funds. Over the sample period, depending on the year
short horizon funds have investment horizons less than 13.3 months to 16.6 months,
and long investment horizon funds have investment horizons greater than 19.3 months
to 26.7 months.
The last six columns present the mean ownership percentage for each fund stock
position and number of stock positions by investment horizon tercile. In thirteen out
of the eighteen years, long horizon funds take larger positions in terms of percentage
of shares held than short and medium horizon funds. Furthermore, in all but one
year, long horizon funds take more positions than either of the two other investment
horizon terciles. I also …nd medium horizon funds take greater positions and larger
positions than short horizon funds.
Although I require a three-year start-up period, it may not be enough time to
measure fund investment horizon. I next investigate the e¤ect of fund age on invest-
ment horizon measurement by averaging the change in FIH from its initial measure
to all subsequent updates. Table 1.2 presents the results. For all funds, the average
change in FIH from its initial measure to its update the following year is equal to
-0.56 months. However, from year 3 to year 10 (when only 4.3% of funds remain in
18
the sample) the average change in FIH is positive and varies between 0.14 to 0.74
months. Across fund investment horizon terciles, I …nd the initial negative change
in FIH from year 1 to year 2 primarily stems from long horizon funds. Whereas the
initial change in FIH for long horizon funds is equal to -4.63 months, the same change
is equal to 0.28 months for medium horizon funds and 2.56 months for short horizon
funds. Although the change in FIH from its initial measurement has a tendency to
be positive for both short and medium horizon funds and negative for long horizon
funds, the magnitude is less than 1 month for the majority of years funds remain in
the sample. Thus, although there is a initial drift toward the sample mean, FIH is a
relatively stable measure over the life of most funds.
1.4.3 Comparison to other Measures of Investment Horizon
6
Previous work typically classi…es institutional investment horizon with portfolio
based measures. Bushee (1998) classi…es institutions based on trading strategy. Wa-
hal and McConnell (2000); Hotchkiss and Strickland (2003); Gaspar, Matos, and
Massa (2005); Hotchkiss and Lawrence (2007); and Yan and Zhang (2009) use mea-
sures of portfolio turnover. Other measures of institutional ownership stability more
closely related to FIH can be found in Bøhren, Priestley, and Ødegaard (2005, 2008),
and Elyasiani, Jia, and Mao (2006). Bøhren et al. (2005, 2008) measure investment
horizon as the number of years an investor holds at least their initial stake. Their
main data source is the Norwegian Securities Registry. Elyasiani et al. (2006) measure
an institution’s ownership stability with the average standard deviation of ownership
percentages for all stocks held over a …ve year period for at least one quarter.
Although FIH should be highly correlated with the above alternatives, it has
several advantages. First, FIH is more related to the corporate governance aspects of
institutional ownership because it directly measures ownership length and is not based
6
I do not repeat this analysis in Chapters 2 or 3. The time periods of study di¤er only slightly.
There is little di¤erence in overall analysis.
19
on portfolio characteristics. Second, it is a more informative measure of investment
horizon because it utilizes the panel data nature of institutional shareholding datasets.
Lastly, I am able to summarize institutional ownership at the …rm level with several
measures controlling for di¤erent aspects of ownership stability. Past work instead
typically relies on classifying fund ownership with the total percentage of shares held
by institution type.
In the rest of this subsection, I compare my measure of investment horizon with
two recent annual turnover-based measures. Gaspar et al. (2005) measures a fund’s
turnover rate (TOT) from one report date to the next as the sum of aggregate port-
folio changes, divided by average portfolio market value. TOT between report date
t ÷1 and report date t for fund i with equity positions , is equal to
TOT
i;
=

jJ
[S
i;j;
P
i;j;
÷S
i;j;1
P
i;j;1
÷S
i;j;1
P
i;j;
[

jJ
S
i;j;
P
i;j;
+S
i;j;1
P
i;j;1
2
where P represents share price, and P

represents the change in share price between
report dates. A fund’s annual turnover rate is equal to the average turnover using all
report dates within the year. Yan and Zhang (2009) measure an institution’s churn
rate similarly, but instead use the absolute minimum of either aggregate purchases
or sales to account for the impact of investor cash ‡ows. Their measure of portfolio
turnover (TOM), is equal to
TOM
i;t;
=
min
_
TO_buy
i;t;
, TO_sell
i;t;
_

jJ
S
i;j;t;d
P
i;j;t;
+S
i;j;t;1
P
i;j;t;1
2
where
TO_buy
i;
=

jJ
S
i;j;
> S
i;j;1
[S
i;j;t;
P
i;j;t;
÷S
i;j;t;1
P
i;j;t;1
÷S
i;j;t;
P
i;j;t;
[
TO_sell
i;
=

jJ
S
i;j;
S
i;j;1
[S
i;j;t;
P
i;j;t;
÷S
i;j;t;1
P
i;j;t;1
÷S
i;j;t;d
P
i;j;t;
[
20
Because much of the analysis below groups funds by short, medium, and long
investment horizons, I compare fund investment horizon terciles between FIH, TOT,
and TOM from 1990 to 2007. I classify funds within the largest TOT and TOM
tercile as having short investment horizons, and funds within the smallest TOT and
TOM tercile as having long investment horizons. Panel A of Table 1.3 presents
tests of correlation between investment horizon terciles. Not surprisingly, FIH tercile
classi…cations are highly correlated with both measures of portfolio turnover. FIH
has a correlation coe¢cient with TOT equal to 0.45, and a correlation coe¢cient
with TOM equal to 0.44. Both correlations are signi…cant at the 1% level.
7
Panel B of Table 1.3 presents the proportion of funds by FIH tercile that have
short, medium, and long investment horizon classi…cations with TOT and TOM. By
FIH tercile, 56.5% of short horizon funds, 41.1% of medium term funds, and 58.5% of
long horizon funds have the same classi…cation with TOT. Also, 56.3%of short horizon
funds, 41.9% of medium horizon funds, and 58.7% of long horizon funds have the same
classi…cation with TOM. However, a substantial number of funds with short and
long FIH classi…cations have the complete opposite classi…cation with the two other
measures. I …nd 10.0% of funds with short investment horizons and 16.0% of funds
with long investment horizons have the opposite classi…cation with TOT, and 12.4%
of short horizon funds and 14.3% of long horizon funds have the opposite classi…cation
with TOM. Thus, although similarities exist, there are substantial di¤erences between
the two measures.
Lastly, I compare the stability of the three measures by computing the probability
a fund in an investment horizon tercile one year will either keep the same tercile
classi…cation the following year or switch to one of the other two. Panel C of Table
1.3 reports the results. Overall, I …nd FIH terciles to be slightly more stable one year
to the next with 68.3% of short horizon funds, 51.3% of medium horizon funds, and
7
The correlation between TOT and TOM is equal to 0.78, also signi…cant at the 1% level.
21
68.3% of long horizon funds retaining their classi…cations. TOT terciles are the next
most stable with 68.1% of short horizon funds, 50.7% of medium horizon funds, and
67.3% of long horizon funds with no change in investment horizon classi…cation. TOM
tercile classi…cations were the least stable with the least number of funds keeping the
same classi…cation one year to the next. Between the three measures, …rms with short
and long FIH classi…cations also have the smallest likelihood of having the opposite
classi…cation the following year.
1.4.4 Ownership Stability Measures
I …rst measure fund ownership stability with average fund shareholder investment
horizon (SIH). SIH is equal to the average investment horizon of funds holding …rm
, at …scal year-end t, weighted by the number of shares held. In equation form, SIH
is equal to
SIH
j;t
=

iI
S
i;j;t
+ Log (FIH
i;t
)

iI
S
i;j;t
(7)
where i indexes the set 1 of all fund shareholders. I take the average with respect to
the natural log of FIH to reduce the in‡uence of fund age on the statistic.
I also measure ownership composition with the proportion of common shares out-
standing held by funds at …scal year-end. In equation form, the ownership percentage
for …rm , in year t held by fund i is equal to
Own%
j;t
=
S
i;j;t
ShrOut
j;t
(8)
where ShrOut is the monthly CRSP measure of shares outstanding (shrout). I ag-
gregate ownership percentage at …scal year end across all funds (TotOwn%) as well
as by fund investment horizon terciles. I distinguish aggregate short horizon fund
ownership with Own%S, medium horizon fund ownership with Own%M, and long
horizon fund ownership with Own%L.
22
I also measure ownership stability with average relative ownership length (AROL).
AROL is equal to the average percentage of stock positions held for a strictly shorter
period of time within fund shareholder portfolios at the …rm’s …scal year end. The
percentage of positions held for a strictly shorter period of time within the same fund
portfolio than stock ,
0
is equal to
ROL
i;j
0
;t
=

jJ
1
_
LT
i;j
0
;t
LT
i;j;t
_
N
i;t
(9)
where , indexes the set of all fund positions J, N
i;t
represents the number of fund
positions, and 1
_
LT
i;j
0
;t
LT
i;j;t
_
is equal to 1 if …rm ,
0
has been held strictly longer
than …rm ,, 0 otherwise. ROL, with a range from [0, 1) , can be thought of as a
cumulative distribution function of average ownership length for each fund portfolio.
Average relative ownership length at the …rm level at report date t is equal to
AROL
j;t
=

iI
S
i;j;t
+ ROL
i;j;t

iI
S
i;j;t
(10)
I also estimate AROL by fund investment horizon tercile. Average relative ownership
length for short horizon funds is distinguished with AROLS, medium horizon funds
with AROLM, and long horizon funds with AROLL. AROLS, AROLM, and AROLL
are set to 0 if the …rm is not held by that particular fund type.
Table 1.4 presents correlations between …rm variables. I …nd a …rm’s market-to-
book ratio is positively correlated with short horizon fund ownership, but negatively
correlated with long horizon fund ownership. Ownership by all fund types decreases
with …rm size, but has a more negative correlation with respect to short and medium
horizon funds. Short horizon fund ownership is also more positively correlated with
past stock returns than the other two FIH terciles. With respect to ownership length,
I …nd …rms with higher market-to-book ratios have longer ownership by short horizon
fund shareholders but shorter ownership by long horizon fund shareholders. Larger
23
…rms and …rms with smaller past stock returns are held longer by their fund share-
holders. I also …nd shareholder composition and ownership length variables to be
positively correlated, both overall (SIH and AROL) and by FIH tercile.
Interestingly, I …nd a …rm’s market-to-book ratio and size are both negatively
correlated with the level of signed discretionary accruals, but positively related to the
level of earnings surprise. Thus, growth …rms and larger …rms are more likely to beat
earnings estimates but less likely to manage earnings upward using discretionary
accruals. This dichotomy is especially odd considering one would expect certain
fund types should be correlated with upwards earnings management regardless of the
measure. Growth …rms and smaller …rms have higher unsigned discretionary accruals.
1.5 Earnings Announcements within One Penny of Analyst Forecasts
In this section, I use the level of earnings surprise for …rms that report within
one penny of analyst forecast consensus (ES ¸ ¦÷0.01. 0.00. 0.01¦) to investigate the
relationship between ownership stability and earnings management. I also investigate
the relationship between ownership stability and changes to analyst forecasts.
1.5.1 Level of Earnings Surprise
I begin the analysis by comparing the percentage of …rms from 1990 to 2007 that
either just beat by one penny, meet, or just miss by one penny analyst forecasts. I
compare using all …rms, as well as by SIH and AROL annual tercile groupings. I
exclude …rms held by less than 5 mutual funds prior to the …scal year-end.
Table 1.5 presents the results. Panel A presents the results using all …rms; Panel
B presents the results when I divide …rms into SIH terciles; and Panel C presents the
results when I divide …rms into AROL terciles. Consistent with past research, I …nd
a greater disproportionate number of …rms either beat (2,732 or 38.8%) or meet (2,
779 or 39.5%) analyst forecasts than just miss (1,534 or 21.8%). I …nd …rms in the
lowest SIH tercile (thus having greater ownership by funds with shorter investment
24
horizons) have a greater tendency to just beat analyst forecasts (41.8%) than meet
(39.7%) or just miss (18.6%). Conversely, …rms in the highest SIH tercile are more
likely to meet analyst forecasts (40.6%) instead of just beat (35.2%). I also …nd some
evidence indicating …rms with shorter fund ownership lengths have a greater tendency
to just beat analyst forecasts. However, the di¤erences between the AROL terciles is
not as prominent as the di¤erences between the SIH terciles.
I formally test the relationship between ownership stability and the likelihood of a
positive earnings surprise by estimating ordered probit models using panel data from
1990 to 2007. Explanatory variables include the following: measures of ownership
stability, MB, FcstSD, AnNum, Debt, Size, industry …xed e¤ects, and year …xed
e¤ects.
8
I …rst control for ownership stability with measures of ownership composition,
then ownership length, and …nally both. I winsorize continuous variables at the 1
st
and 99
th
percentiles. I require …rms to be held by at least 5 mutual funds prior to
the earnings announcement date for all regressions. I cluster standard errors at the
…rm level.
Table 1.6 presents the results. The …rst three columns present regression results
when I control for ownership stability with measures of shareholder composition. In
the …rst two regressions, I control for shareholder composition with either SIH or
Own%S, Own%M, and Own%L, and in the third regression I use TotOwn%.
I …nd greater ownership by funds with shorter investment horizons is positively
related to the likelihood …rms just beat analyst forecasts. In either speci…cation, SIH
is negative and statistically signi…cant at the 1% level with t-statistics ranging from
3.52 to 3.56. Again, the sign and signi…cance of SIH stems from greater ownership
by funds with shorter investment horizons. When ownership is separated by fund
investment horizon tercile, Own%S is positive and statistically signi…cant with a
coe¢cient equal to 2.768. I also …nd greater overall fund ownership diminishes the
8
I classify …rms only at the Fama-French 12 Industry Classi…cation to aid in the convergence of
the ordered probit models.
25
importance of fund shareholder investment horizon composition with respect to the
direction of earnings management; TotOwn% is positive and signi…cant at the 1%
level.
Columns (4) and (5) presents regression results when I control for ownership sta-
bility only with measures of ownership length (either AROL or AROLS, AROLM,
and AROLL), and columns (6) and (7) present results when I control for institu-
tional ownership with both measures of ownership stability (…rst overall and then by
FIH tercile). In the regressions controlling for just ownership length, although I …nd
average ownership length using all fund shareholders is an insigni…cant determinant
in the level of earnings surprise, longer ownership by short horizon funds is a pos-
itive and signi…cant determinant of ES at the 5% level. When I include both sets
of institutional ownership variables, I …nd greater ownership by funds with shorter
investment horizons is the primary determinant of the level of earnings surprise. SIH
is a positive and signi…cant determinant at the 1% level, and Own%S and Own%M
are both positive with corresponding signi…cances at the 1% and 10% levels. AROLS
is no longer signi…cant with a t-statistic equal to 0.81. I also estimate models in-
cluding interaction terms between the percentage of ownership and its corresponding
ownership lengths at the FIH tercile level. However, I do not include the results from
this regression in the table because all interaction terms are statistically insigni…cant.
Going against the idea that the ability of …rms to beat analyst forecasts increases
when there is greater information asymmetry between the …rm and market partici-
pants, forecast dispersion is negatively related to ES. Interestingly, greater debtholder
presence decreases the likelihood of a positive earnings surprise, having the opposite
e¤ect than with signed discretionary accruals. Stock returns are positively related to
the level of earnings surprise. Market-to-book ratio, …rm size, and analyst number
are all insigni…cant.
I re-estimate the regressions instead using …rms with either earnings surprises
26
strictly greater than one penny and strictly less than minus one penny or earnings
surprises strictly greater than one penny. I continue using the same general economet-
ric methodology but estimate linear regressions instead of ordered probit regressions.
Table 1.7 presents the results. Columns (1) through (3) present regression results
explaining earnings surprises strictly greater than one penny and strictly less than
one penny, and columns (4) through (6) present regression results explaining earn-
ings surprises strictly greater than one penny. In both sets of regressions, I control for
mutual fund ownership with SIH, percentage ownership by fund investment horizon
tercile, and total mutual fund ownership percentage. Because I …nd only inconsistent
evidence with respect to ownership length in Table 1.6, I do not control for ownership
length in these regressions.
I …nd short horizon fund ownership continues to be signi…cantly related to the level
of earnings surprise for …rms with positive and negative earnings surprise greater than
one penny. However, short horizon fund ownership is not as signi…cant as before: in
column (1) SIH is signi…cant and negative at the 10% level, and in column (2) Own%S
is positive and signi…cant at the 10% level. Taken together, although short horizon
funds continue to take advantageous positions over a broader range of earnings sur-
prises, the relationship is strongest among those …rms announcing within one penny
of analyst forecasts and thus most likely to be engaging in earnings manipulation. I
…nd no indication total fund ownership is related to the level of earnings surprise in
column (3), and no indication fund ownership is related to earnings surprises greater
than one penny in columns (4) through (6).
1.5.2 Changes in Analyst Forecasts
I extend the analysis by investigating whether changes in median analyst forecasts
over the …nal quarter of the …scal year is related to a …rm’s ownership stability. For the
same reasons why ownership stability may be related to the management of earnings,
…rms with less ownership stability may engage in greater expectations management.
27
At the same time, analysts may use characteristics of institutional ownership when
updating forecasts.
I estimate linear regression models describing the change in median analyst fore-
casts (FcstMed). The dependent variable is equal to the di¤erence between median
analyst forecasts from the …rst month (10) to last month (12) in the …nal quarter of
the …rm’s …scal year (medest
12
÷medest
10
). Explanatory variables include measures of
ownership stability, MB, Debt, Size, 3MRet, industry …xed e¤ects, and year …xed ef-
fects. I control for ownership stability using either overall measures of fund ownership
or by FIH tercile. Standard errors are cluster-robust at the …rm level. I continue to
use …rms announcing earnings-per-share results within one penny of analyst forecasts
at …scal year end.
9
Table 1.8 presents the results in four columns. The …rst column presents regression
results when I control for fund ownership with SIH and AROL, and the second
column presents regression results when I control for ownership stability with Own%S,
Own%M, Own%L, AROLS, AROLM, and AROLL. Interestingly, I …nd both SIH
and AROL are signi…cantly related to changes in median forecasts but in opposing
directions. For instance, the results of the …rst regression presented in column (1)
detail how SIH is a negative predictor of FcstMed at the 1% con…dence level with
a t-statistic equal to 5.16, but AROL is negative and signi…cant at the 5% level.
Thus, although the presence of fund shareholders with longer investment horizons is
negatively associated with analyst forecast updates, the longer funds hold …rm shares
the more likely analysts will increase earnings forecasts upward. When I distinguish
ownership by FIH tercile, I …nd the negative relationship between SIH and FcstMed
stems from both greater ownership by short horizon funds and less ownership by
long horizon funds. Own%S is positive and signi…cant at the 1% level (t-statistic
9
An alternative test is to regress MedFcst on changes in fund ownership over the same time
period. However, because mutual funds are only required to …le the N-30D form semiannually and
not quarterly, there is not enough frequency in mutual fund shareholding data to perform these
tests.
28
= 5.39), and Own%L is negative and signi…cant at the 5% level (t-statistic = 2.25).
Furthermore, I …nd longer ownership by funds with long investment horizons is driving
the sign and signi…cance of AROL; …rms held for longer by long horizon funds have
more positive changes in analyst forecasts. This last result is signi…cant at the 1%
level.
These results support evidence found by Burgstahler and Eames (2003) indicat-
ing analysts anticipate greater earnings management to avoid small losses and small
earnings decreases. In this chapter, I …nd that analysts’ actions are related to charac-
teristics of institutional ownership. It also suggests managers engage in less earnings
forecast management when the …rm already has established long term investors.
In the last two regressions, I again control for fund ownership …rst with the over-
all measures and then measures by FIH tercile, but also include interaction terms
between all ownership variables and the level of earnings surprise. The signi…cance
of the interaction term indicates whether the relationship between institutional own-
ership and the change in earnings forecasts systematically di¤ers between …rms that
either beat, meet, or missed analyst forecasts. Between the two regressions, I …nd ES
interacts signi…cantly only with SIH. The positive coe¢cient of the interaction term
indicates analysts do not fully account for shareholder investment horizon for those
…rms that end up beating forecasts.
1.6 Discretionary Accruals
In this section, I use signed and unsigned discretionary accruals to investigate
the relationship between ownership stability and earnings management. I estimate
level regressions explaining discretionary accruals …rst with measures of ownership
composition, then with measures of ownership length, and …nally both. I further
these tests by estimating di¤erence regressions.
1.6.1 Level Regressions
29
I start by estimating least-squares regressions explaining levels of signed and un-
signed discretionary accruals controlling for ownership stability using measures of
shareholder composition. Other …rm-level control variables include market-to-book
ratio, debt, size, annual return, standard deviation of analyst forecasts, …scal year
…xed e¤ects, and industry …xed-e¤ects based on the Fama-French 48 Industry Clas-
si…cation.
10
I winsorize all continuous variables at the 1
st
and 99
th
percentiles. I
use …rms held by at least 5 mutual funds prior to the end of the …scal year to ensure
ownership measures are not driven by a small number of fund shareholders. Following
Petersen (2009), I estimate panel regressions clustering standard errors at the …rm
level. I use …rm data from 1990 to 2007.
11
Table 1.9 presents the results. There are eight columns of estimates. The …rst
three columns correspond to regressions explaining the level of signed discretionary
accruals, and the last three columns correspond to regressions explaining the level
of unsigned discretionary accruals. For either measure of earnings management, I
…rst control for shareholder composition with average shareholder investment horizon
(SIH), then with the percentage of shares held by fund investment horizon tercile
(Own%S, Own%M, and Own%L), and …nally with the total percentage of shares held
by all mutual fund shareholders (TotOwn%).
I …nd greater ownership by short horizon funds is positively related to the level
of earnings management. In regressions describing either DA or UnsDA, SIH is neg-
ative and statistically signi…cant at the 1% con…dence level indicating …rms with
ownership weighted toward funds with shorter investment horizons are more likely
to manage earnings upward and overall. When I distinguish ownership by fund in-
vestment horizon tercile, I …nd the sign and signi…cance of SIH is primarily due to
10
The Fama-French 48 Industry Classi…cation can be found on Ken French’s website. All industry
…xed-e¤ects employed in the tests below are based on this level of classi…cation.
11
In unreported tests I also estimate Fama-MacBeth (1973) time-series average coe¢cients and
t-statistics from annual cross-sectional regressions. I adjust coe¢cient standard errors for autocor-
relation using a Newey-West adjustment to two lags. The results do not change.
30
greater ownership by funds with short investment horizons than an absence of long
horizon funds. Own%S is positive and statistically signi…cant at the 1% con…dence
level regardless of the discretionary accrual measure. Although always a negative
determinant, Own%L is insigni…cantly related to signed discretionary accruals and
only signi…cantly related to unsigned discretionary accruals at the 10% con…dence
level. Interestingly, ownership by medium horizon fund shareholders (Own%M) is a
positive and signi…cant determinant of DA (t-statistic = 3.34), but a negative and
signi…cant determinant of UnsDA (t-statistic = 1.71).
The sign and signi…cance of ownership by each investment horizon tercile explains
the di¤erences in the overall importance of fund ownership between the two discre-
tionary accrual measures. In the regression describing signed discretionary accruals
TotOwn% is positive and statistically signi…cant at the 1% level (t-statistic = 3.67).
Thus, total fund ownership is important in describing the direction of earnings man-
agement. However, overall fund ownership is an insigni…cant predictor of unsigned
discretionary accruals.
In general, the explanatory variables are consistent with the motivation of earnings
management as a means to improve stakeholder relations. For instance, smaller …rms
are more likely to have greater unsigned discretionary accruals. In addition, …rms with
a greater probability of violating debt covenants are more likely to manage earnings
upward but not engage in greater earnings management overall. Lastly, when there is
greater information asymmetry between the …rm and market participants as measured
by MB and FcstSD, …rms have less of a tendency to manage earnings upward but
instead are more likely to smooth earnings between periods.
12
Some evidence is
found indicating annual return is a negative and signi…cant determinant of unsigned
discretionary accruals, but not signed discretionary accruals. Thus, …rms with better
12
Rajgopal et al. (1997, 1999), Burns et al. (2006), and Yu (2008) also …nd growth …rms and …rms
smaller in size to have lower levels of unsigned discretionary accruals.
31
recent stock performance are less likely to manage earnings overall.
I next estimate level regressions describing DA and UnsDA with measures of own-
ership length. I follow the test methodology above again estimating panel regressions
clustering standard errors at the …rm level. Table 1.10 presents the results. There
are four columns of results. The …rst two columns correspond to regressions ex-
plaining DA, and the second two correspond to regressions explaining UnsDA. For
each measure of discretionary accruals I …rst control for ownership length of all fund
shareholders (AROL) and then ownership length by fund investment horizon tercile
(AROLS, AROLM, and AROLL).
Although overall relative ownership length is not a signi…cant determinant of
signed discretionary accruals, I do …nd …rms with longer ownership by funds with
short and medium investment horizons are more likely to manage earnings upwards.
AROLS is a positive and signi…cant determinant of DA at the 5% con…dence level
(t-statistic = 2.38), and AROLM is a positive and signi…cant determinant of DA at
the 1% level (t-statistic = 2.75). With respect to unsigned discretionary accruals, I
do …nd overall ownership length to be a signi…cant factor. AROL is a negative and
signi…cant predictor of unsigned discretionary accruals at the 1% con…dence level.
When I separate ownership length by FIH tercile, the determinacy of AROL is pri-
marily driven by the ownership length of medium and long horizon funds. AROLM
and AROLL are both negative and signi…cant at the 1% con…dence level, with the co-
e¢cient of AROLL (-0.006) having a greater magnitude than AROLM (-0.004). The
sign and signi…cance of the other explanatory variables remain primarily the same.
Two interesting points can be made when comparing the results from Tables 1.9
and 1.10. First, greater ownership and longer ownership by short horizon funds is
related to more upwards earnings management, and greater ownership and longer
ownership by long horizon funds is negatively related to its overall level. Thus, the
relationship between fund ownership characteristics and earnings management is not
32
only dependent on the type of fund owner but also how one measures earnings man-
agement. Second, although greater "ownership stability" by medium horizon funds
is a positive predictor of signed discretionary accruals, it is a negative predictor of
unsigned discretionary accruals. This result is suggestive of a push-and-pull between
short term gains and long term value when the investment horizon of a fund is neither
short nor long.
Ultimately, the correlation in the results between the two measures of institutional
ownership (especially by FIH tercile) may stem from their positive correlation. To
determine if shareholder composition and ownership length can both be signi…cant
predictors of earnings management or if one characteristic is more important than
the other, I next estimate level regressions describing DA and UnsDA using both
sets of variables. I follow the same test methodology as above, estimating three
regressions for each measure of discretionary accruals. In the …rst regression, I control
for ownership stability with SIH and AROL, and in the second, I control for ownership
stability with the percentage of shares held and relative ownership length by FIH
tercile. In the third regression, I again control for ownership with measures at the
FIH tercile level but include interaction terms between the percentage of shares held
and its corresponding ownership lengths.
Table 1.11 reports the results. In general, I …nd the same variables signi…cant in
the previous two regression remain signi…cant here. This indicates that the two char-
acteristics of ownership stability control for di¤erent aspects of fund ownership and
can both be important determinants. There is one primary exception when I break
fund ownership by FIH tercile. Both Own%M and Own%L lose their signi…cance in
describing unsigned discretionary accruals when I regress them with measures of own-
ership length. This indicates one characteristic of ownership can dominate the other
depending on the measure of earnings management and fund type. Among interac-
tion terms, only Own%MAROLS is a signi…cant determinant of signed discretionary
33
accruals (no interaction terms are signi…cant with respect to unsigned discretionary
accruals). The negative sign of the coe¢cient indicates longer ownership and greater
ownership by short horizon funds are substitutes with respect to their determinacy
of DA.
1.6.2 Di¤erence Regressions
The results from Tables 1.9, 1.10, and 1.11 indicate ownership stability is nega-
tively related to the level of earnings management. However, the results could stem
from …rm-level characteristics that explains both ownership stability and the propen-
sity to manage earnings. To account for this potential explanation, I next estimate
di¤erence regressions explaining changes in signed and unsigned discretionary accru-
als. I di¤erence variables from year t ÷ 1 to year t + 1 to account for mechanical
changes in discretionary accruals as a result of transitioning revenues or expenses
between consecutive periods. I straight-di¤erence all variables except for …rm debt,
equal to long term debt in year t + 1 minus long term debt in year t ÷ 1, divided
by total assets in year t ÷ 1. I pre…x di¤erence variables with . I winsorize all
continuous variables at the 1
st
and 99
th
percentiles. I use …rms held by at least 5
mutual funds prior to the end of …scal year t ÷1 and …scal year t + 1 to ensure that
changes in ownership measures are not driven by a small number of funds. I report
coe¢cients and t-statistics from panel regressions clustering standard errors at the
…rm level. I use …rm data from 1991 to 2006.
13
Table 1.12 presents the results. There are eight columns. The …rst four columns
present regression results explaining changes in signed discretionary accruals, and
the next four columns present regression results with respect to changes in unsigned
discretionary accruals. I …rst control for changes in ownership stability with SIH
and AROL.
14
In the next two regressions, I control for changes in ownership stability
13
I also estimate Fama-Macbeth (1973) style regressions. The results do not change.
14
I also estimate separate regressions controlling for changes in ownership with SIH and AROL
independently. The sign and signi…cance of the variables remain the same.
34
using measures at the FIH tercile level, …rst with respect to the percentage of shares
held and then by ownership length. In the …nal regression, I include changes in
both the percentage of shares held and relative ownership lengths by fund investment
horizon tercile.
I continue to …nd strong evidence indicating ownership by funds with shorter
investment horizons is positively related to signed discretionary accruals. SIH is
negative and signi…cantly related to DA at the 1% con…dence level with a t-statistic
equal to 3.75. I again …nd the sign and signi…cance of SIH stems from funds
with short and medium investment horizons; both Own%S and Own%M are
positive and signi…cant regardless of whether I control for changes in ownership length.
Although I …nd some evidence indicating short horizon fund ownership length is
a positive determinant of DA, the result is not robust to inclusion of ownership
percentage change variables in the regression. I …nd all other measures of ownership
length to be insigni…cant. With respect to unsigned discretionary accruals, I …nd little
evidence indicating changes in any measure of fund ownership is signi…cant. The lone
exception is the change in medium horizon fund ownership. Consistent with the level
regressions found in Table 1.9, Own%M is negatively related to UnsDA at the
10% level.
In all di¤erence regressions, …rms transitioning from value to growth are more
likely to manage earnings. MB is positive and statistically signi…cant at the 1%
con…dence level. Although …rms increasing in size are found to engage in more positive
earnings management, the overall use of discretionary accruals decreases. Consistent
with the level regressions, an increase in analyst forecast dispersion is negatively
related to changes in signed discretionary accruals but positively related to changes
in unsigned discretionary accruals. Firms with a greater decline in stock price are also
more likely to manage earnings upward and overall than before. I also …nd evidence
indicating changes in debt is positively related to changes in unsigned discretionary
35
accruals.
Overall, the results in this section demonstrate a strong positive relationship be-
tween signed discretionary accruals and greater ownership by funds with shorter in-
vestment horizons. I also …nd consistent evidence indicating greater ownership by
medium horizon funds is positively related to the direction of earnings management
but not to its overall level. These two results relate to the di¤erences in the way
the two fund types interact with …rm managers. Interestingly, I …nd little indication
ownership by funds with long investment horizons is related to earnings manage-
ment, suggesting that funds with shorter investment horizons are the more important
shareholders in this case.
1.7 Chapter Conclusion
This chapter presents evidence indicating ownership stability by a …rm’s mutual
fund shareholders is an important determinant in the direction and overall emphasis
placed on earnings management. The results in this chapter also underscore how
di¤erent aspects of ownership stability can be important in describing …rm behavior. I
…nd the strongest relationship between ownership stability and earnings management
is the positive correlation between short horizon fund ownership and the direction of
earnings management. Also important is ownership by medium horizon funds relating
to more upwards earnings management but less earnings management overall.
36
Chapter 2: Corporate Spin-o¤s
2.1 Introduction
A justi…cation often given by …rm managers to spin-o¤ one or more subsidiaries
is to obtain greater and more specialized analyst coverage. Spin-o¤s increase analyst
number by increasing investor demand for coverage, …rm demand for investment
services, and analyst ability by allowing for a more perfect match with analysts and
their particular expertise.
15
The increase in business focus that improves analyst coverage may also alter the
level and type of institutional ownership. By increasing company focus, …rms en-
gaging in corporate spin-o¤s become more attractive to institutions that generally
hold shares for longer periods of time (or have longer investment horizons) where
portfolio composition is of greater importance. On the other hand, with the increase
in business focus and transparency spin-o¤ corporations may become less attractive
to institutions with shorter investment horizons thought to hold an informational
advantage (Wermers (2000), and Yan and Zhang (2007)).
The composition of institutional shareholders is important to …rm management.
In general, …rm managers prefer institutional shareholders that hold …rm stock for
longer periods of time. Longer-term investors not only allow companies to pursue
long-term strategies, but also are more likely to aid …rm managers by communicating
both their private outlooks as well as the opinions of sell-side analysts. Conversely,
institutions that hold shares for shorter periods of time are more likely to exert
greater pressure on …rm managers to act myopically, oftentimes on threat of removal
or company takeover (Useem(1996)).
In this chapter I investigate the relationship between business focus and institu-
tional ownership stability by comparing positions before and after corporate spin-o¤s.
15
For empirical evidence see Krishnaswami and Subramaniam (1999) and Gilson, Healy, Noe, and
Palepu (2001).
37
I investigate not only the role of business focus in portfolio composition between in-
stitutions with di¤ering levels of investment horizon, but also how changes in business
focus relates to the level and length of institutional shareholder stability.
I take institutional stock positions at the fund level from the Thomson Reuters
(S12) Mutual Fund dataset. The dataset consists of positions from most domestic
mutual funds and some global funds that participate in US and Canadian equity
markets. The primary source for the dataset is SEC N-30D …lings. Although for the
majority of the time period the SEC required mutual funds to …le this form semi-
annually, Thomson Reuters supplements the …lings by examining fund prospectuses
and contacting mutual funds directly. The other approach is to use the Thomson
Reuters (13f) Investment Company dataset consisting of aggregate holdings of banks,
insurance companies, parents of mutual funds, pensions, and endowments. The pri-
mary source for this dataset is quarterly SEC 13f …lings required by all institutional
investment managers that exercise investment discretion over $100 million.
A fund’s investment horizon is equal to the average length of time (in months)
each share of every stock positions is held from the date of initial stock investment
to the date of measurement. Using the full ownership history of each stock position
to classify a fund’s investment horizon is a departure from past literature which uses
either a range of portfolio characteristics (e.g. Bushee (1998), and Hotchkiss and
Strickland (2003)) or portfolio turnover (e.g. Gaspar, Matos, and Massa (2005), and
Yan and Zhang (2009)). By using mutual fund holding data and measuring an in-
stitution’s investment horizon in this manner, I am able to increase the number of
institutional shareholders in my dataset and create a more precise measure of share-
holder investment horizon that is more directly related to the corporate governance
aspects of institutional ownership.
I begin by investigating changes in the level of fund ownership around corporate
spin-o¤s. Prior to the spin-o¤, I measure fund ownership of the parent company (the
38
original conglomerate) with the total percentage of shares held by all fund share-
holders, and with the average fund shareholder investment horizon. I use the same
ownership measures following the spin-o¤ but create two sets. The …rst set measures
the "overall" fund ownership of all …rms originating from the same parent company
with market-value weighted averages. The second set measures fund ownership of
just the parent company. Initial results suggest a positive and signi…cant increase in
overall and parent company fund ownership not only by all funds but especially those
with longer investment horizons. However, consistent with evidence found by Abar-
banell et. al (2003), after controlling for similar changes in ownership by a control
…rm I …nd no indication ownership changes (both overall and in the parent company)
are signi…cant.
I explain the two sets of adjusted ownership changes with multivariate regressions.
I …rst explain overall changes in fund ownership with measures of business focus with
variables describing di¤erences between parent companies and subsidiaries based on
growth opportunities, size, operating performance, and industry classi…cation. I also
use the change in analyst coverage to measure the overall importance of the spin-o¤
event. Consistent with spin-o¤s as a means to separate businesses of the original
parent company with di¤ering levels of performance, I …nd di¤erences in operating
performance between spin-o¤ related …rms and greater overall mean changes in oper-
ating performance predict greater overall change in total fund ownership. Second, I
explain changes in parent company ownership with explanatory variables measuring
changes in …rm characteristics. Variables include changes in size, operating perfor-
mance, market-to-book, capital expenditures, debt, and payout yields. I also control
for the type of spin-o¤ with changes in analyst coverage and di¤erences in industry
classi…cation. I …nd funds increase ownership when growth opportunities increase and
leverage ratios decrease. The change in growth opportunities also positively predicts
a long-term increase in shareholder investment horizon.
39
Also of importance is the relationship between changes in adjusted measures of
fund ownership with abnormal returns following the e¤ective date. Consistent with
past researchers, I …nd overall (parent companies and all spun-o¤ subsidiaries) and
parent company abnormal returns are positive and statistically signi…cant 12 months,
24 months, and 36 months after the spin-o¤ event. Although I …nd no indication the
change in total adjusted fund ownership percentage is signi…cant, I do …nd evidence
indicating greater ownership by funds with shorter investment horizons is positively
related to the distribution of abnormal returns. This evidence supports past work
indicating the level of institutional ownership is not related to abnormal returns
(Abarbanell et. al (2001)) but ownership by institutions with shorter investment
horizons does (Wermers (2000), and Yan and Zhang (2009)).
I next investigate the ownership patterns of fund shareholders that hold …rm stock
prior to the spin-o¤ announcement date. I use two measures at the fund-…rm level
to describe changes in ownership. The …rst measure is equal to a discrete variable
that distinguishes between funds that holds no …rms, a proportion of …rms, and all
…rms originating from the pre-spin-o¤ conglomerate. The second measure is equal to
the change in ownership percentage using only those observations where the fund still
holds a positive stake at the later date. The purpose of these tests is to determine not
only the level at which pre-existing fund shareholders remain invested in the original
company, but also whether fund shareholders increase their stake in those …rms they
do continue to hold.
In univariate tests I …nd roughly 40% of pre-existing shareholders hold at least one
…rm from the original parent following the spin-o¤ event and that on average funds
increase the positions in the stocks they continue to hold. I again estimate multi-
variate regressions explaining the proportion of spun-o¤ …rms held and the change in
ownership percentage with variables describing di¤erences between …rms originating
from the same parent company. pre-existing shareholders are more likely to hold onto
40
a greater proportion of …rms following spin-o¤s when the di¤erence between market-
to-book ratios is greater and the di¤erence between operating performances is lower.
This is suggestive of funds holding onto more shares when the spin-o¤ splits …rms
with varying levels of growth opportunities and the divestment of poorly performing
businesses was not the motivation for the event. Interestingly, a positive relationship
is also found with respect to changes in analyst coverage. This is suggestive of funds,
even those holding the original parent company prior to the spin-o¤ event, preferring
companies that have separate and distinct businesses. Interestingly, the signi…cance of
the determinants decreases with investment horizon indicating the same determinants
associated with longer ownership length also decreases the sensitivity to changes in
…rm characteristics. I …nd little evidence changes in fund ownership percentage are
dependent on …rm di¤erences.
Instead of investigating changes in fund ownership around spin-o¤ events, I lastly
compare fund ownership patterns strictly before and strictly after spin-o¤ events. In
this instance, corporate spin-o¤s provide a unique environment to investigate whether
fund ownership patterns di¤er between more stand alone entities and multibusiness
corporations.
I estimate three sets of regressions with di¤erent fund ownership measures taken at
the fund-…rm level. The …rst dependent variable is equal to the change in ownership
percentage, the second dependent variable is equal to the di¤erence in the percentage
of other …rms held within the same fund portfolio but for a strictly shorter period of
time, and the third is equal to the likelihood a fund closes an equity position. For the
…rst two variables I di¤erence the ownership measures in two distinct and consecutive
time periods before and after the spin-o¤ event using only those observations where
the fund has a positive stake at the earlier date. For the last dependent variable I
use fund data three years before the announcement date and three years following
the e¤ective date to determine dates of position close.
41
There are three sets of explanatory variables of interest. The …rst set controls for
overall changes in fund behavior before and after the spin-o¤, testing for di¤erences
in fund behavior between conglomerates and more focused entities. The second set
of variables controls for changes in operating performance. If there is a di¤erence in
ownership stability between …rms based on focus, then it may be represented in the
sensitivity fund positions have toward changes in …rm pro…tability. The third set of
explanatory variables of interest measures the magnitude of the spin-o¤ with changes
in analyst coverage from before to after the spin-o¤ event. Fund positions may di¤er
between …rms with spin-o¤s of di¤ering signi…cance.
Although I …nd little indication changes in ownership percentage are signi…cantly
di¤erent following spin-o¤s, I do …nd strong evidence indicating funds, especially those
with longer investment horizons, have more positive changes in ownership and hold
shares relatively longer after spin-o¤s than before. Thus, one justi…cation to spin-o¤
businesses, especially those of a di¤erent industry, is to obtain not only more stable
ownership by all fund shareholders but especially fund shareholders that typically
hold onto shares for longer periods of time. This may be especially important for
diversi…ed companies that engage in spin-o¤s. Past research (Thomas (2002), and
Denis, Denis, and Sarin (1997)) note decreases in diversi…cation are associated with
managerial turnover and …nancial distress. Consistent with evidence of Chemmanur
and He (2008), I also …nd funds with shorter investment horizons seem to exhibit
more informed trading after the e¤ective date.
This chapter looks at speci…c holdings of mutual funds and relates it to the level
and length of institutional ownership. Overall, this chapter …nds evidence indicat-
ing corporate events, in this case spin-o¤s, have a signi…cant e¤ect on institutional
ownership. The results have two sets of implications. First, the comparison in insti-
tutional ownership provides evidence indicating how di¤erences between institutional
shareholders based on ownership length can manifest itself itself in changes before
42
and after spin-o¤ events. Second, the results indicate that although parent compa-
nies cannot signi…cantly alter shareholder composition (with respect to shareholder
investment horizon) by spinning o¤ subsidiaries, they can attract longer ownership
by shareholders which tend to have more of a positive role in …rm management.
2.2 Spin-o¤ Literature
In this section, I review related literature and discuss in further detail the contri-
butions of this work.
Habib, Johnsen, and Naik (1997) and Nanda and Narayanan (1999) theorize …rms
engage in corporate divestitures to improve the information environment surrounding
the …rm. In both models, …rms are able to improve the valuation of the …rm by mak-
ing cash ‡ows more observable to market participants, thus increasing share value
and improving investment decision quality. Krishnaswami and Subramaniam (1999)
and Gilson, Healy, Noe, and Palepu (2001) …nd evidence of a decrease in information
asymmetry by testing changes in analyst forecast errors and analyst coverage be-
fore and after spin-o¤ events. Thomas (2002), however, …nds no evidence indicating
conglomerates in general su¤er from information problems.
Another motivation for …rms to engage in corporate spin-o¤s is to improve invest-
ment e¢ciency. Rajan, Servaes, and Zingales (2000) and Scharfstein and Stein (2000)
both argue internal rent seeking can distort the investment e¢ciency of internal cap-
ital markets. Several researchers have used spin-o¤s as a environment to test changes
in investment quality. Gertner, Powers, and Scharfstein (2002) …nd subsidiaries re-
cently spun-o¤ from their parent company increase investment in high Q (Tobin’s)
industries but decrease investment in low Q industries. This result is primarily found
in subsidiaries with unrelated businesses to the parent company’s or that had a pos-
itive market reaction to the spin-o¤ event. Dittmar and Shivdasani (2003) …nds
the parent company improves investment e¢ciency following spin-o¤s compared to
stand-alone entities. Burch and Nanda (2003) and Ahn and Denis (2004) …nd similar
43
evidence using the investment decisions of the combined …rms (parent company and
spun-o¤ subsidiaries). However, after controlling for the decision to divest, Çolak and
Whited (2006) …nds no evidence of investment e¢ciency improvements. With a sim-
ilar argument, Chemmanur and Yan (2004) theorize spin-o¤s improve the e¢ciency
of management contests.
Evidence suggests corporate spin-o¤s are positively related to changes in …rmvalue
and operating performance. Hite and Owers (1983); Miles and Rosenfeld (1983);
Schipper and Smith (1983); Vijh (1994); and Cusatis, Miles, and Rosenfeld (1993)
all …nd positive abnormal returns following spin-o¤ events. Using a comprehensive
36 year sample, McConnell and Ovtchinnikov (2004) …nd positive long-run abnormal
returns for subsidiaries but not for parent companies. Daley, Mehrotra, and Sivakumar
(1997) and Desai and Jain (1999) both …nd post-event abnormal returns and changes
in operating performance are larger for …rms engaging in focus-increasing spin-o¤s.
Desai and Jain (1999) also …nd changes in operating performance are positively related
to changes in business focus.
More recent work links institutional ownership to changes in …rm value and infor-
mation production around spin-o¤ events. Abarbanell, Bushee, and Raedy (2003) …nd
institutional ownership patterns are consistent with previous investment styles and
…duciary restrictions but no indication institutional ownership changes are related to
short-term abnormal returns. Greenwood (2006) …nds similar evidence; di¤erences
between parent companies and subsidiaries, especially in size and growth opportu-
nities, induce predictable selling and buying between institutional investors and re-
late to short-term abnormal returns. Using proprietary institutional trading data,
Chemmanur and He (2008) investigate the role of institutional trading in information
production after corporate spin-o¤s. They …nd the trading imbalance between the
parent companies and subsidiaries increases with the level of information asymmetry,
di¤erence in beta risk, and the di¤erence in long-term growth prospects suggesting
44
information production, risk management, and investment in a particular business are
all motivations of trade. They also …nd evidence indicating a positive relationship
between institutional trading and abnormal returns. Although further evidence of in-
formed trading after spin-o¤s is also found by Huson and MacKinnon (2003), Brown
and Brooke (2003) …nd "uninformed" rebalancing by institutional investors placing
downward price pressure. Patro (2008) examines changes in subsidiary block owner-
ship after spin-o¤ events. He …nds overall block ownership increases after spin-o¤s
and is positively related to monitoring needs.
This chapter is most similar to the work of Abarbanell et al. (2003) and Greenwood
(2006). However, the focus of this chapter is not just institutional ownership patterns
around spin-o¤s, but institutional ownership as it relates to ownership stability. This
chapter also extends past work on informational advantages held by institutions with
short investment horizons. Wermers (2000) …nds evidence indicating value of active
mutual fund management. As part of the evidence, he …nds high-turnover …rms beat
the Vanguard Index 500 Fund on a net return basis. Yan and Zhang (2007) …nd a
positive relationship between short-term institutional ownership and future returns
with particularly strong evidence with respect to small …rms and growth …rms. To my
knowledge, past literature does not use spin-o¤s as a testing environment to address
the information advantage held by short horizon funds.
2.3 Data
In this section I describe the methodology used to create the spin-o¤ event dataset
and de…ne base …rm-level variables.
45
2.3.1 Spin-o¤ Event Data
I use the Securities Data Company (SDC) Platinum U.S. Mergers and Acquisi-
tions Database to obtain corporate spin-o¤ data. I extract the following variables:
announcement date, e¤ective date, percent acquired, target ultimate parent cusip,
target immediate parent cusip, target cusip, target ultimate parent name, target im-
mediate parent name, target name, and spin-o¤ status.
I match SDC dataset company names to the CRSP Monthly Stock Event Name
History data…le to obtain current cusip identi…ers as well as name history dates. I
cross-reference each observation with the Dow Jones Factiva database, con…rming
the spin-o¤ announcement and e¤ective dates, classifying the form of stock distri-
bution (e.g. tracking stock), and determining whether the spin-o¤ relates to other
restructuring events.
I take all completed spin-o¤ transactions with announcement dates between Jan-
uary 1, 1990 to December 31, 2007 for a total of 641 observations. I …rst re…ne the
dataset by combining concurrent spin-o¤ observations stemming from the same orig-
inal parent company into one. For these combined observations (33), I rede…ne the
announcement date as the earliest date the original conglomerate announced a spin-
o¤, and the e¤ective date as the last date the parent company distributed shares.
To be consistent with Abarbanell et al. (2003), I exclude 83 observations because
the stock distribution is less than eighty percent of the equity in the subsidiary (and
therefore does not qualify for tax-free treatment) or no stock distribution informa-
tion is available. Of the remaining 525 observations, I eliminate 86 that relate to
other corporate events such as mergers and acquisitions, reorganization as the result
of bankruptcy, and corporate liquidations; 70 because the parent company prior to
the announcement date is not listed on a major stock exchange (NYSE, AMEX, or
NASDAQ); and 44 because either the parent company or one of spun-o¤ subsidiaries
following the e¤ective date is not traded on one of these three exchanges. To more
46
accurately measure fund ownership and to ensure ownership changes are directly re-
lated to the spin-o¤ event, I discard an additional 13 spin-o¤ events because the
parent company (or its newly formed subsidiaries) lists multiple share classes and 42
spin-o¤ events because at least one spun-o¤ subsidiary was traded one month prior
to the announcement date. Lastly, to ensure share changes are not the result of past
spin-o¤ events, I exclude 24 spin-o¤ events with an announcement date within three
years of a previous spin-o¤ related stock distribution. The remaining number of spin-
o¤ event observations is 246. The number of spin-o¤ observations is similar to and
greater than the total number of spin-o¤ observations in past work. Table 2.1 reports
the …nal number of spin-o¤s each year from 1990 to 2007. The number of observations
in the tests below ‡uctuates based on the availability of other information.
2.3.2 Firm-Level Variables
I extract …rm data from the Compustat Fundamentals Annual data …le, the Center
for Research in Securities Prices (CRSP) monthly stock return …le, and the I/B/E/S
detail …le. I derive the following variables from Compustat data for each …scal year t.
« ROA
t
(return-on-assets) = Operating income before depreciation divided by
total assets (data13
t
, data6
t
).
« ROACA
t
(return on cash adjusted assets) = Operating income before depreci-
ation divided by the di¤erence between total assets and cash and short-term
investments (data13
t
, (data6
t
÷data1
t
))
« ROS
t
(return on sales) = Operating income before depreciation divided by net
sales (data13
t
, data12
t
).
« Size
t
(…rm size) = The natural log of total assets (data6 or at).
« MB
t
(market-to-book ratio) = Fiscal year end market value divided by book
value (MV
t
,BV
t
). Book value (BV)is equal to the sum of total assets, deferred
47
tax and investment credit (data35 or txditc), and convertible debt (data79 or
dcvt), minus preferred stock (data10 or pstkl) and total liabilities (data181 or
lt).
« CapEx
t
(capital expenditures) = Capital expenditures (data128 or capx) di-
vided by total assets (data128
t
,data6
t
).
« Debt
t
(debt) = Total long term debt (data9 or dltt) divided by total assets
(data9
t
,data6
t
).
« DivYld
t
(dividend yield) = Dividends paid per share as of the ex-dividend date
times …scal year end shares outstanding, divided by total assets (data6 or at)
(data26
t
data25
t
, data6
t
).
« RepYld
t
(repurchase yield) = The purchase of common and preferred stock
divided by total assets (data115
t
, data6
t
).
I derive the following two variables from CRSP data.
« AnnRet
t
(annual return) = Annual compounded monthly return percentage
__

m2[Jan, Dec]
(1 + :ct
m;t
)
_
÷1
_
in year t, using all 12 months : of data.
« ICD (industry classi…cation di¤erence) = 3 if at least one …rm stemming from
the same parent company (either the parent company or subsidiary) has a dif-
ferent Fama-French 12 industry classi…cation, 2 if at least one …rm has the same
Fama-French 12 industry classi…cation but di¤erent Fama-French 30 industry
classi…cation, 1 if at least one …rm has the same Fama-French 12 and 30 indus-
try classi…cations but di¤erent Fama-French 48 industry classi…cation, and 0 if
all …rms stemming from the same parent company have the same Fama-French
48 industry classi…cation.
I derive the last variable, the change in analyst number, from I/B/E/S.
48
« AnNum (change in analyst number) = the number of analysts covering the
…rm following the e¤ective date minus the number of analysts covering the …rm
six months before the announcement date divided by 100. I measure the number
of analysts 12, 24, and 36 months following the e¤ective date depending on the
time period of the test.
2.4 Changes in Fund Ownership
16
In this section I investigate changes in fund ownership and equity value following
spin-o¤ events. I explain changes with multivariate regressions.
2.4.1 Fund Ownership Variables
In this subsection I summarize fund ownership levels three years prior to the
announcement date and three years following the e¤ective date. I subdivide each
time period into 6 month intervals, using the most recent …lings at each breakpoint.
I …rst summarize fund ownership with the total percentage of shares held. Prior
to the announcement date, the percentage of shares held by fund i in parent company
(the original conglomerate) , at date t is equal to
Own%
Bef,P
i;j;t
=
S
i;j;t
ShrOut
j;t
(11)
where S is the number of shares held and ShrOut is the number of common shares
outstanding taken from the CRSP database (shrout). The total percentage of shares
held in the parent company (TotOwn%
Bef,P
) is equal to
TotOwn%
Bef,P
j;t
=

iI
Own%
Bef,P
i;j;t
(12)
16
See Chapter 1.4 for a full discussion of the mutual fund sample and measurement of fund
investment horizon.
49
where i indexes the set of fund shareholders 1. Following the e¤ective date, I cal-
culate the overall percentage of shares held in the parent company and all spun-o¤
subsidiaries as well as the percentage of shares held in just the parent company. To
calculate the overall percentage of shares held (TotOwn%
Aft,O
), I …rst sum and then
weight the total percentage of shares held in each …rm by market value. In equation
form, the overall percentage of shares held is equal to
TotOwn%
Aft,O
S;t
=

sS
MV
s;t
+ TotOwn%
Aft
s;t

sS
MV
s;t
(13)
where TotOwn%
Aft
represents the total percentage of shares held after the e¤ective
date, MV represents market value, and : indexes the set o of …rms stemming from
the same original conglomerate. I de…ne the percentage of shares held in just the
parent company after the e¤ective date (TotOwn%
Aft,P
) similar to Equation (13). I
de…ne a …rm after the spin-o¤ event as being the parent company if it retains the
same cusip as the original conglomerate.
The number of observations at each date is dependent on whether the parent
company or all …rms originating from the same parent company trade for at least
six months prior to measurement.
17
I exclude observations of overall ownership if
either the parent company or one of the spun-o¤ subsidiaries does not meet trading
requirements, and parent company ownership observations if it alone does not meet
trading requirements.
Table 2.2 presents sample mean ownership percentages three years before and
three years after the spin-o¤ event. In general, I …nd …rms engaging in corporate
spin-o¤s experience an increase in fund ownership before and after the event. Overall
fund ownership (in the parent company and all spun-o¤ subsidiaries) increases from
17
Funds are required to …le shareholdings semiannually. If the …rm is not traded for at least over
a six month interval, some fund observations may be absent from the measurement.
50
5.6% to 6.8% over the three years prior to the announcement date, and from 7.7%
to 8.6% over the three years following the e¤ective date (7.4% to 8.3% for just the
parent company). This increase in fund ownership is driven primarily by long horizon
funds. Long horizon funds increase their total ownership from 2.3% to 3.1% over the
three years prior to the announcement date, and from 3.7% to 4.3% over the three
years following the e¤ective date (3.6% to 4.3% for the parent company). On the
other hand, short and medium horizon funds either have no or just small increases in
ownership over the same time periods.
2.4.2 Univariate Tests of Fund Ownership Changes
The results above indicate an increase in ownership following spin-o¤s especially
by funds with longer investment horizons. In this subsection, I test whether the
ownership increases are signi…cant.
I test unadjusted changes in fund ownership 12 months, 24 months, and 36 months
following the e¤ective date using fund ownership levels 6 months prior to the an-
nouncement date as a baseline. I …rst test changes in total overall (TotOwn%
O
)
fund ownership percentage and in the parent company (TotOwn%
P
). I use all ob-
servations regardless of how many funds hold …rm stock before or after the spin-o¤
event.
To investigate changes in ownership composition, I use average fund shareholder
investment horizon (SIH). SIH is equal to the average investment horizon of funds
holding …rm , at date t, weighted by the number of shares held. In equation form,
SIH is equal to
SIH
j;t
=

iI
S
i;j;t
+ Log (FIH
i;t
)

iI
S
i;j;t
(14)
I take the average with respect to the natural log of FIH to reduce the in‡uence
of fund age. Similar to total ownership percentage, I calculate SIH for the parent
company 6 months prior to the announcement date (SIH
Bef,P
), and overall (SIH
Aft,O
)
51
and for the parent company (SIH
Aft,P
) following the e¤ective date. To ensure changes
in shareholder investment horizon are not driven by a small number of funds, I require
the parent company to be held by at least 5 mutual funds prior to the announcement
date and 5 mutual funds following the e¤ective date (either overall or just the parent
company depending on the statistic). I designate the overall change in shareholder
investment horizon change with SIH
O
, and the change in the parent company with
SIH
P
.
For all measures I calculate both mean and median changes, using two-sided t-
statistics to test for a di¤erence in mean change and two-tailed .-statistics from
Wilcoxon rank-sum tests to test for a di¤erence in median change. Panel A of Table
2.3 presents results with respect to overall fund changes, and Panel B presents results
with respect to fund changes in the parent company.
Consistent with the fund ownership levels in Table 2.2, mean and median changes
in total fund ownership percentage is positive and statistically signi…cant with con…-
dence levels greater than 5%. This is true regardless of the whether I measure overall
ownership or in just the parent company and the length of the time period (12, 24,
or 36 months following the e¤ective date). Furthermore, consistent with the relative
increase in long horizon fund ownership, the mean and median changes in SIH are
positive and statistically signi…cant 24 months and 36 months following the e¤ective
date with con…dence levels greater than 5%.
The strong results in Panels A and B, however, may instead stem from changes in
mutual fund ownership patterns over time. To account for this potential explanation,
I match each event …rm with a control …rm and di¤erence the change in ownership
in the e vent …rm with a similar change in the match …rm. Following Lie (2001)
and Grullon and Michaely (2004), I narrow the number of potential control …rms by
requiring the same 48 Fama-French Industry Classi…cation, and measures of MB
t1
,
ROA
t1
, and ROA
t1
within 80% to 120% of the event …rm’s value. For each event
52
…rm c, I then choose the control …rm c which minimizes
[ROA
e;t1
÷ROA
c;t1
[ +[ROA
e;t1
ROA
c;t1
[ +[MB
e;t1
÷MB
c;t1
[ (15)
If I cannot …nd a match from this sample, I repeat the procedure but loosen the indus-
try restriction to all …rms within the same Fama-French 12 Industry Classi…cation. If
I still cannot …nd a match, I use the control …rm which minimizes Equation (15) with
no regard to industry classi…cation. If still no match is found, I choose the control
…rm which minimizes Equation (15) with no restrictions. Like with event …rms, when
testing changes in SIH I require control …rms to be held by at least 5 funds at the
beginning of the measurement period and 5 funds at the end of the measurement
period. I again use tests of mean and median change to test for signi…cance.
Panels C and D in Table 2.3 presents tests of adjusted ownership changes similar
to Panels A and B. After adjusting for similar changes in ownership, I …nd no signif-
icant evidence indicating …rms attract greater fund ownership or longer horizon fund
ownership following spin-o¤ events. However, I do …nd mean and median changes in
total ownership percentage and shareholder investment horizon both overall and in
the parent company are positive 24 months and 36 months after the e¤ective date.
Thus, there is at least some indication total fund ownership and fund ownership by
long horizon funds increases.
2.4.3 Multivariate Regressions Explaining Adjusted Ownership Changes
In this subsection I explain the adjusted changes in fund ownership with multivari-
ate regressions. I estimate two sets of regressions, the …rst explaining overall changes
in fund ownership and the second explaining changes in just the parent company. I
estimate regressions explaining adjusted ownership changes 12 months, 24 months,
and 36 months after the e¤ective date.
In the …rst set of regressions, I explain overall changes in fund ownership with
53
measures describing di¤erences between …rms stemming from the same parent com-
pany. I estimate linear panel regressions with one observation per spin-o¤ event.
Explanatory variables include
« SDSize (standard deviation of …rm size) = the standard deviation of …rm sizes
for all …rms originating from the same parent company.
« SDMB (standard deviation of market-to-book ratios) = the standard deviation
of market-to-book ratios for all …rms originating from the same parent company.
« SDROA (standard deviation return-on-asset ratios) = the standard deviation of
return-on-asset ratios for all …rms originating from the same parent company.
« ROA (the mean change of return-on-asset ratios) = the mean return-on-asset
ratio for all …rms originating from the same parent company following the ef-
fective date minus the return-on-asset ratio of the parent company taken six
months prior to the announcement date.
Past work of Greenwood (2006) and Gompers and Metrick (2001) …nd evidence
linking …rm size and growth opportunities to institutional demand curves. Measures
of operating performance account for the motivation of a spin-o¤ as a means to sep-
arate underperforming businesses or improve investment allocations. I use standard
deviations instead of absolute di¤erences to account for spin-o¤s of more than one
subsidiary. I also use ICD and AnNum to control for di¤erences in industry classi-
…cation and the overall importance of the spin-o¤ as it relates to analyst demand. I
include an interaction term between ICD and SDSize (ICDSDSize) to di¤erentiate
between small and large spin-o¤s of di¤ering industries.
Past authors have used revenue-based and asset-based Her…ndahl indices, segment
number, and industry dummy variables to control for business focus; and forecast
errors, analyst number, and volatility of stock returns to control for information
54
asymmetry. ICD improves on past industry indicator variables because it controls for
the degree of industry di¤erence between spin-o¤ related …rms. I use AnNum as the
measure of information change because it relates to not only to the importance of the
spun-o¤ businesses based on future investor and investment service demand, but also
the change in transparency based on the cost of information creation and ultimately
analyst supply. Measures of forecast error, on the other hand, can be clouded by
earnings volatility and manipulation.
18
Lastly, I include year …xed-e¤ects based on
the e¤ective date. I do not include industry …xed e¤ects because match …rms were
chosen based on industry classi…cation. Standard errors are heteroscedastic-robust.
Panel A of Table 2.4 presents the regression results. Columns (1) through (3)
presents regression results describing overall changes in total ownership percentage 12
months, 24 months, and 36 months after the e¤ective date, and columns (4) through
(6) presents regression results explaining overall changes in shareholder investment
horizon. No factor is a signi…cant determinant in all six regressions. However, I do
…nd consistent evidence indicating di¤erences in industry classi…cation and operat-
ing performance as well as the overall mean change in operating performance are
signi…cant in describing changes in total ownership percentage.
Both the standard deviation and overall mean change in return-on-assets are posi-
tive and statistically signi…cant with con…dence levels greater than 10% in describing
total ownership changes 12 months and 36 months after the e¤ective date (both
measures are positive but insigni…cant in describing total ownership changes at 24
months). Thus, di¤erences and the subsequent gains in operating performance be-
tween the di¤erent businesses within the original parent company make all or parts of
the company more attractive overall to institutional shareholders. This evidence is in-
dicative of a positive reaction to conglomerates of either spinning o¤ under-performing
businesses or improving the management of the …rm by decreasing its scope.
18
Earnings manipulation is extensively researched in the accounting literature. See Graham,
Harvey, and Rajgopal (2005) for CFO survey evidence.
55
Interestingly, industry classi…cation di¤erences is negatively related to the change
in total ownership percentage in each regression, and signi…cant at 24 and 36 months.
Comparing these coe¢cient estimates to regressions describing changes in shareholder
investment horizon, the negative relationship with respect to ICD seems to stem pri-
marily from a decrease in short horizon fund ownership. The coe¢cients multiplying
ICD is positive in each regression describing SIH, and signi…cant at the 10% level
at 24 months. Consistent with the idea that di¤erences in industry classi…cation is
more important the greater the relative size of the spin-o¤, the interaction between
ICD and SDSize is a negative determinant in each regression describing changes in
SIH with signi…cance at 24 months. No other explanatory variable is signi…cant in
regressions describing changes in shareholder investment horizon.
In the second set of regressions, I explain fund ownership changes in just the
parent company with measures describing changes in characteristics. Explanatory
measures include changes in return-on-assets, market-to-book ratios, …rm size, capital
expenditures, debt ratio, dividend yield, and repurchase yield. Each variable is equal
to its post-spin-o¤ minus pre-spin-o¤ value. I also include ICD, AnNum, and year
…xed-e¤ects. Standard errors are heteroscedastic-robust.
Panel B of Table 2.4 presents the regression results similar to Panel A. The …rst
three columns present regression results explaining changes in total fund percentage
ownership 12 months, 24 months, and 36 months after the e¤ective date. The next
three columns present regression results describing SIH. Important to the change
in total fund ownership is the change in debt and market-to-book ratio. I …nd an
increase in leverage decreases total fund ownership, whereas an increase in growth
opportunities increases total fund ownership. Both determinants are signi…cant in
describing ownership changes at 12 months and ownership changes at 36 months and
have the same sign regardless of the time period. In addition, the increase in busi-
ness focus as it relates to industry classi…cation (ICD) is negatively and signi…cantly
56
related to the change in total fund ownership 24 months and 36 months after the
e¤ective date at con…dence levels greater than 5%.
Again, the relationship between ICD and the change in total ownership percentage
seems to primarily stem from a decrease in short horizon funds. Although insigni…-
cant, ICD is a positive predictor of SIH in each regression. Overall, little evidence is
found indicating any explanatory variable is an important determinant in describing
changes in shareholder investment horizon. However, the change in market-to-book
ratio is again positive and statistically signi…cant at the 1% con…dence level at 36
months in describing SIH
P
. Thus, …rms with an increase in growth opportunities
attract not only more fund shareholders but funds which typically hold shares for
longer periods of time.
These results are consistent with past evidence including Krishnawami and Sub-
ramaniam (1999), and Gilson et al. (2001) indicating spin-o¤s as a means to reduce
information asymmetry and potentially increase ownership. In addition, the results
especially with respect to change in market-to-book ratio is indicative of institutional
shareholders increasing ownership of parent companies when internal capital markets
improve (Gertner, Powers, and Scharfstein (2002), and Ahn and Denis (2004)).
2.4.4 Relation to Post-Event Returns
Past researchers (see Chapter 2.2) document positive abnormal returns after spin-
o¤ events. Here, I investigate changes in abnormal returns with respect to changes in
fund ownership as well as other explanatory variables describing the spin-o¤ event.
I measure post-event stock returns 12 months, 24 months, and 36 months after the
e¤ective date. I do not investigate short-term returns because of the low frequency of
ownership data. I measure both overall abnormal returns (of the parent company and
all subsidiaries) and parent company abnormal returns. Overall post-event returns
(RET
O
) t months following the e¤ective date is equal to
57
RET
O
S;
=
_
_
_
_

t=1
_
_

sS
MV
s;t1
(1 + ret
s;t
)

sS
MV
s;t1
_
_
_
_
÷1
_
_
(16)
where ret represents monthly returns taken from the CRSP monthly stock return …le,
MV represents market value, and : indexes the set o of …rms stemming from the same
parent company. When t is equal to 1, MV
t1
is equal to share price times shares
outstanding on the …rst day of trading in the month. I use only spin-o¤ observations
where all …rms stemming from the original parent company begin trading either the
month or two months after the e¤ective date and have continuous returns until month
t.
Post-event returns for the parent company , is equal to
RET
P
j;
=
__

t=1
(1 + ret
s;t
)
_
÷1
_
(17)
Adjusted post-event returns (AdjRET) is equal to the returns of spin-o¤ …rms
(both overall and the parent company) minus the returns of match …rms designated
in Chapter 2.4.2. I …ll missing return data for match …rms with the Fama-French 5x5
size and book-to-market sorted portfolios. I classify match …rms using lagged data at
the time of the match (see Equation (15)). I calculate two sets of adjusted returns,
the …rst when I place no restrictions on the number of fund shareholders at the latter
date (in tests of ownership percentage) and the second when I place restrictions (in
tests of SIH). Following Barber, Lyon and Tsai (1999), I test that the mean adjusted
returns are greater than zero using bootstrapped skewness adjusted t-statistics (Hall
(1992)). Bootstrapping involves 1,000 drawings of half the sample. I also test for
median signi…cance with Wilcoxon rank-sum tests. Table 2.5 presents the results.
Panel A presents overall abnormal returns and Panel B presents abnormal returns
for the parent company. When I place no ownership restrictions on the match al-
gorithm, I …nd overall and parent company mean abnormal returns are positive and
58
signi…cant from zero 12, 24, and 36 months after the e¤ective date. Positive abnormal
returns at 36 months are con…rmed with tests of median signi…cance. Interestingly,
when I place restrictions on fund ownership (number of fund owners has to be greater
than 5 prior to the announcement date and 5 after the e¤ective date) I only …nd
mean short-term returns both overall and for the parent company are positive and
statistically signi…cant. Tests of median returns are largely insigni…cant. The di¤er-
ences in results when I place no restrictions and some restrictions on fund ownership
is indicative of sample selection e¤ects.
I explain adjusted post-event returns with multivariate regressions using changes
in fund ownership and measures describing the spin-o¤ event as explanatory variables.
I estimate 12 separate regressions distinguished by return length (12, 24, and 36
months), ownership measure (total ownership percentage and SIH), and …rm set (all
…rms and parent company). The regressions follow the same speci…cations describing
changes in fund ownership in Chapter 2.4.3. Table 2.6 presents the results.
Panel A presents the regression results for overall abnormal returns. Columns
(1) and (2) present regression results describing abnormal returns 12 months after
the e¤ective date …rst using TotOwn%
O
and then SIH
O
to control for changes in
fund ownership. In a similar fashion, columns (3) and (4) presents regression results
describing abnormal returns 24 months after the e¤ective date, and columns (5) and
(6) presents regression results describing abnormal returns 36 months after the e¤ec-
tive date. Consistent with past evidence indicating institutions with short investment
horizons in‡uence stock returns (Wermers (2000), and Yan and Zhang (2009)), be-
tween the three time periods and the two measures of ownership change I …nd SIH
O
to be negative and signi…cantly related to the abnormal overall returns 12 months
after the e¤ective date. The signi…cance at 12 months is at the 5% level. SIH
O
is
negative but statistically insigni…cant at 24 months, and positive and insigni…cant at
36 months. I do not …nd any evidence TotOwn%
O
is a signi…cant predictor over
59
any time period. Among other explanatory variables, again ROA is consistently
signi…cant at 12 and 24 months indicating spin-o¤s that improve overall operating
performance is associated with more positive abnormal overall returns. No consistent
evidence is found with respect to all other explanatory variables.
Panel B presents the regression results for parent company abnormal returns anal-
ogous to Panel A. Similar to the regression results describing overall abnormal returns,
I …nd SIH
P
is a negative and signi…cant predictor of abnormal parent company re-
turns 12 months and 24 months after the e¤ective date. Importantly, in these regres-
sions signi…cance is at the 1% con…dence level. SIH
P
is a negative but insigni…cant
predictor at 36 months, and TotOwn%
P
is an insigni…cant predictor of abnormal
returns over all time periods. Between other explanatory variables, I …nd the overall
change in analyst number and the change in size are both positive and signi…cant
predictors of abnormal returns. Thus, returns are more positive for parent compa-
nies that engage in spin-o¤s of more important businesses and parent companies that
spin-o¤ smaller companies or have smaller decreases in size.
2.4.5 Chapter 2.4 Summary
Overall, this section provides evidence indicating spin-o¤s does not have a signif-
icant e¤ect on the level of overall fund ownership or fund ownership of the parent
company. However, operating performance, both between spin-o¤ entities and over-
all changes from before to after the spin-o¤ event, is the most important factor in
describing the cross-section of institutional ownership changes. Thus, it is the abil-
ity to properly manage various business segments, not other factors such as growth
opportunities and di¤erences in industry classi…cation, which initially limits invest-
ment. I also …nd evidence indicating short horizon funds opportunistically invest in
companies with more positive abnormal returns following spin-o¤ events, even after
controlling for measures relating to the level of pre-spin-o¤ transparency.
60
2.5 The Case of Pre-Existing Shareholders
In this section I detail the ownership patterns of funds that hold parent company
…rm stock six months prior to the announcement date.
2.5.1 Ownership Patterns
In this subsection I describe changes in shareholdings of funds holding original
parent company stock before the announcement date. I take the most recent fund
holdings 6 months prior to the announcement date and 12 months, 24 months, and 36
months following the e¤ective date. I use all funds that have an investment horizon
measure the year prior to the pre-spin-o¤ report date and classify a fund’s investment
horizon at this date. I discard fund observations following the e¤ective date if the
number of consecutive years of meeting regulatory reporting requirements ends.
I use two fund-level measures to describe ownership patterns following the spin-o¤.
The …rst measure, the proportion of …rms held (Prop), is a count variable describing
how many of the total number of …rms stemming from the same parent company are
still held. Prop is equal to 0 if the fund closes its position in all …rms (parent company
and all spun-o¤ subsidiaries), 1 if the fund remains invested in at least one …rm, and
2 if the fund remains invested in all …rms. The second measure is the change in
ownership percentage (Own%). Because mutual funds may close positions in …rms
following the spin-o¤ that are simply unrelated to the portfolio’s investment strategy
or the spin-o¤ itself was an attempt to discard a poorly performing business, I use
only those changes in ownership percentage where the fund retains a positive stake
at the latter date. As a result, the second measure thus relates to whether funds
increase or decrease their investment in …rms they continue to hold following spin-o¤
events.
Table 2.7 summarizes Prop and Own% by fund investment horizon tercile 12
months, 24, months, and 36 months following the e¤ective date. Panel A reports the
percentage of funds remaining in the sample by Prop (¸ ¦0, 1, 2¦). 20.2%, 16.8%, and
61
19.6% of long horizon funds remaining in the sample hold all …rms (parent company
and all spun-o¤ subsidiaries) 12 months, 24 months, and 36 months following the
e¤ective date. An additional 34.1%, 30.1%, and 23.1%of long horizon funds remaining
in the same hold at least one …rm. Thus, a large proportion of these funds retain
at least some ownership in the original parent company following a spin-o¤. The
proportion of funds retaining ownership in all or just one …rm following spin-o¤s
decreases with investment horizon tercile. For instance, 42.8%, 37.9%, and 32.3% of
medium horizon funds and 35.5%, 30.7%, and 27.4% of short horizon funds retain
ownership in at least one …rm 12 months, 24 months, and 36 months after the e¤ective
date.
Panel B reports summary statistics for Own%. Summary statistics includes
the mean change in ownership percentage, the percentage of positions exhibiting
an increase or non-decrease, and the percentage of positions exhibiting a decrease.
Overall, I …nd funds increase the size of the positions in …rms still held following the
e¤ective date. Long horizon funds generally increase their investment in …rms which
they still hold following corporate spin-o¤s. 62.1%, 65.2%, and 66.2% of long horizon
funds increase their ownership 12 months, 24 months, and 36 months following the
e¤ective date. Although these percentages are smaller for medium and short horizon
funds, greater than 50% of the positions of these two fund types increase in size.
2.5.2 Multivariate Regressions
In this subsection I describe Prop and Own% with multivariate regressions. I
describe both dependent variables using measures describing di¤erences between …rms
describing the same parent company (SDMB, SDSize, SDROA, ROA, AnNum,
and ICD). I use all shareholdings in the same regression, investigating di¤erences
between funds by including FIH and interaction terms between FIH and all other
explanatory variables. I model Prop with an ordered probit regression and Own%
with a linear regression, estimating separate models for each time period following the
62
e¤ective date (12, 24, and 36 months). I also include announcement year and industry
…xed-e¤ects. I de…ne year …xed-e¤ects using the year of the e¤ective date and industry
…xed-e¤ects using the Fama-French 12 Industry Classi…cation of the original parent
company. Following Petersen (2009), I estimate panel regressions clustering standard
errors at the fund level.
Table 2.8 presents the results. Columns (1) through (3) present regression esti-
mates explaining Prop. I …nd several interesting relationships between the proportion
of …rms held and SDMB, SDROA, and AnNum. First, funds are more likely to hold
more …rms following the spin-o¤ when the di¤erence between growth opportunities is
greater. Thus, funds are either willing to choose one …rm or the other, or hold onto all
…rms when the investment opportunities of the conglomerate are separated. Second,
contrary to the regressions explaining percentage changes in fund ownership in the
previous section, here I …nd pre-existing shareholders are less likely to hold more po-
sitions following the spin-o¤ when the di¤erences in operating performance between
…rms is greater indicating pre-exisiting shareholders may choose one business over the
other when di¤erences between operating performances is high. Lastly, the change
in analyst coverage is positively related to Prop. Thus, even funds holding parent
company stock prior to the spin-o¤ event react positively to an improved information
environment surrounding the …rms.
For AnNum, SDROA, and SDMB, the coe¢cient multiplying the respective
interaction term with FIH has the opposite sign than the coe¢cients multiplying
the stand alone variables. This suggests even after controlling for fund investment
horizon …rms that hold shares for longer periods of time are less sensitive to these
factors. The results with respect to the interaction terms with AnNum and SDMB
are especially interesting considering funds with longer investment horizons should
hold more of the original parent company the more "important" the spin-o¤ and the
more investment opportunities are distinguishable. These result are instead indicative
63
of long-horizon funds preferring to hold a portion of the original conglomerate and
divest the remainder. At the same time however, long horizon funds hold more parts
of an original conglomerate when operating performances di¤er than short horizon
funds.
Columns (4) through (6) present regression estimates explaining Own%. Over-
all, I …nd little indication the change in ownership percentage di¤ers systematically
between any explanatory variable or by FIH.
2.6 Ownership Stability Before & After Spin-o¤ Events
In this section, I compare changes to fund ownership sensitivity with respect
to business focus and …rm operating performance before and after spin-o¤ events.
I estimate three regressions using one of three fund level measures of ownership:
percentage of shares held, a within portfolio measure of ownership length, and the
likelihood of position close.
I …rst estimate least-squares regressions explaining changes in ownership percent-
age before and after spin-o¤ events. I di¤erence fund ownership in two distinct time
periods prior to the announcement date and in two distinct time periods following the
e¤ective date. Prior to the announcement date, I measure changes in fund ownership
from 30 months to 18 months, and from 18 months to 6 months. I take the most
recent fund holdings at each date, requiring positive fund holdings at 30 months for
the …rst time period and at 18 months for the second time period. I de…ne ownership
changes following the e¤ective date similarly, taking di¤erences from 12 months to
24 months and from 24 months to 36 months. I do not take ownership changes from
before the announcement date to after the e¤ective date to avoid ownership changes
speci…cally as a result of the event. I use holdings of funds that have investment
horizon information (FIH and tercile classi…cation) the year prior to earlier date.
The explanatory variables of primary interest include
« PSOInd (post spin-o¤ indicator) = 1 if the observation occurs following the
64
e¤ective date, 0 otherwise.
« FIH
« PSOIndFIH
« ROA (change in ROA) = the change in return-on-assets from the earlier date
to the later date.
« ROAPSOInd = an interaction variable between ROA and PSOInd.
« ROAFIH = an interaction variable between ROA and FIH.
« ROAPSOIndFIH = an interaction variable between ROA, PSOInd, and
FIH.
« AnNum
« AnNumPSOInd = an interaction variable between AnNum and PSOInd.
« AnNumFIH = an interaction variable between AnNum and FIH.
« AnNumPSOIndFIH =an interaction variable between AnNum, PSOInd,
and FIH.
Consistent with the work of Barber and Lyon (1996), I use ROA as the primary
measure of operating performance. I estimate alternate regressions instead using
either ROACA or ROS as the measure of operating performance. For the sake of
brevity I do not present these additional tests because the results remain primarily
the same. I use only AnNum as the sole measure of business focus because to
measures the added demand by investors in analyst coverage. I measure analyst
number after the spin-o¤ e¤ective date at 12 months.
65
Other …rm-level control variables include changes in size, market-to-book ratio,
capital expenditures, debt, dividend yield, repurchase yield, and annual return. Pre-
vious work including Falkenstein (1996), Gompers and Metrick (2001), Grinstein and
Michaely (2005), and Yan and Zhang (2009) have found these variables as important
in describing fund ownership. Changes in return-on-assets and other …rm-level control
variables are coincide with changes in ownership percentage. I also include interac-
tion terms between each variable and FIH to control for di¤erences between funds.
I include two indicator variables controlling for whether the mutual fund held the
parent company prior to the spin-o¤ (HldBef, equal to 1 if the fund held the parent
company prior to the announcement date, 0 otherwise) and if the parent company re-
tains its cusip following the spin-o¤ (PrntCo, equal to 1 if the parent company retains
its cusip, 0 otherwise). I also include announcement year …xed-e¤ects and industry
…xed-e¤ects based on the Fama-French 48 Industry Classi…cation. Following Petersen
(2009), I estimate panel regressions clustering standard errors at the fund level.
Column (1) in Table 2.9 presents the results. In general, changes in fund ownership
are more negative following the spin-o¤ than before as PSOInd is negative following
and statistically signi…cant at the 1%level (t-statistic =3.67). Importantly, ownership
percentage changes are more positive for funds with longer investment horizons after
the spin-o¤ then before. The coe¢cient multiplying the interaction term between
PSOInd and FIH is positive and statistically signi…cant at the 1% con…dence level
with a t-statistic equal to 2.81.
Although I …nd no indication of a signi…cant relationship between the change
in return-on-assets and the change in ownership percentage, I do …nd the change
in analyst coverage is an important determinant especially after the e¤ective date.
Those …rms associated with a greater overall change in analyst number (and thus
have a more important spin-o¤) experience greater increases in fund ownership after
the spin-o¤ event. However, the increases in ownership signi…cantly decline with fund
66
investment horizon.
Instead of changes in ownership percentage, I next estimate a least-squares regres-
sion explaining changes in a within-fund measure of ownership length strictly before
and strictly after the spin-o¤ event similar to changes in ownership percentage.
19
Rel-
ative ownership length (ROL) is equal to the average percentage of stock positions
held for a strictly shorter period of time within fund shareholder portfolios at the
…rm’s …scal year end. The percentage of positions held for a strictly shorter period
of time than stock ,
0
within the same fund portfolio is equal to
ROL
i;j
0
;t
=

jJ
1
_
LT
i;j
0
;t
LT
i;j;t
_
N
i;t
(18)
where , indexes the set of all fund positions J, N
i;t
represents the number of fund
positions, and 1
_
LT
i;j
0
;t
LT
i;j;t
_
is equal to 1 if …rm ,
0
has been held strictly longer
than …rm ,, 0 otherwise.
20
ROL, with a range from (0, 1] , can be thought of as a
cumulative distribution function of average ownership length for each fund portfolio.
ROL
i;j;t2
= ROL
i;j;t2
÷ROL
i;j;t1
(19)
I use only fund positions held at the earlier date. If the fund closes its position at the
later date, I set ROL
t2
= 0.
ROL has two primary advantages over Own%. First, the change in ownership
length is unitless and is independent of fund size. Second, ROL accounts for other
portfolio changes thus controlling for general fund behavior. The panel regression
follows the empirical approach above. Column (2) in Table 2.9 presents the results.
I …nd no evidence indicating changes in relative ownership length are signi…cantly
larger after the spin-o¤ than before. However, I do …nd signi…cant evidence indicating
19
See Chapter 1.4 for full discussion of mutual fund ownership length measurement.
20
See Chapter 1.4 for a full de…nition of LT.
67
funds with longer investment horizons hold subsidiaries and parent companies longer
after the e¤ective date than before. The interaction term is positive and signi…cant at
the 1% con…dence level with a t-statistic equal to 3.91. Changes in return-on-assets
overall has an overall positive and signi…cant relationship with respect to changes in
relative ownership length, but interestingly decreases following the spin-o¤. Thus,
funds are less sensitive to changes in operating performance after spin-o¤s than be-
fore. I …nd no evidence indicating di¤erences between funds by investment horizon.
Alternatively, I …nd the change in analyst number is positive and signi…cantly related
to changes in ownership length and decreases with fund investment horizon. I …nd
no evidence of a di¤erence post-e¤ective date. Thus, in general long horizon funds do
not hold parent companies and subsidiaries a¢liated with more meaningful spin-o¤s.
Instead of changes in fund ownership I model the likelihood a fund closes its
position. I use the semiparametric Cox proportional hazards model. The Cox model
assumes the instantaneous rate of position close (or hazard rate) after elapsed time
t given …rm characteristics X, /(t[X), takes the form
/(t[X) = /
0
(t) exp(X) (20)
where /
0
(t) represents the baseline hazard rate and represents the vector of model
coe¢cients. The range of the hazard rate is from 0 (no risk of position close) to in…nity
(de…nite position close). Importantly, the model makes no assumptions about the
baseline hazard rate or the shape of the hazard rate over time. Coe¢cient estimates
maximize the product of the conditional probability of position close by comparing
…rm characteristics for positions that close to positions that remain open. Conditional
probabilities are calculated at each point in time at least one fund position closes. I
employ the Breslow (1974) approximation in case of tied failures.
21
21
Cleves, Gould, Gutierrez, and Marchenko (2008) provides a good overview of the Cox model.
68
To create the sample, I …rst measure the length of time in months each fund posi-
tion is held from the date of initial investment (t = 0) to the date of position closure
(t = t
c
). The two dates correspond to bdate and cdate used in the computation of
fund investment horizon in Chapter 1.4. If either the fund or stock drops from the
dataset, I censor the fund position at the last known report date of ownership.
I next merge …rm-level variables with the most recent data prior to the date of
initial investment. If stock positions are held through the end of the …scal year, I
update …rm variables by partitioning the full interval of fund ownership ((0. t
c
]) at
each …scal year-end and merge to each new partition the most recent …rm data. For
example, if a …scal year end occurs after t
0
months of ownership, I partition the
interval of fund ownership into two segments, (0. t
0
) and (t
0
+ 1. t
c
], with the …rst
partition continuing to have the most recent …rm-level data prior to the date of initial
investment and the second partition having …rm-level data from the most recent …scal
year-end.
I use all ownership intervals of only those …rms associated with spin-o¤s (parent
companies and all spun-o¤ subsidiaries) and that falls within either three years prior to
the spin-o¤ announcement date or three years following the e¤ective date. I truncate
ownership intervals that either begins or ends outside of either time period. For
ownership intervals that begin before either time period, I rede…ne the beginning
of the ownership interval to coincide with the beginning of the time period. For
ownership intervals that end following the time period, I censor the observation by
rede…ning the end of the ownership interval to coincide with the end of the time
period.
I again follow the same empirical approach as above using the same set of ex-
planatory variables including announcement year and industry …xed e¤ects. I cluster
standard errors at the fund level. Column (3) in Table 2.9 presents the results.
PSOInd is positive and statistically signi…cant at the 10% level indicating funds are
69
more likely to close positions following spin-o¤s. The likelihood of position close
after spin-o¤s also increases with ROA and is more likely for funds with shorter
investment horizons. That is, short horizon funds are more likely to close equity po-
sitions after larger changes in operating performance after spin-o¤s. Potentially, the
spin-o¤ event allows short horizon funds to better use their informational advantage
and better time changes in …rm performance similar to the evidence of Chemmanur
and He (2008). Like with the change in relative ownership length, I …nd positions of
parent companies and subsidiaries associated with spin-o¤s with greater increases in
analyst number have a higher overall likelihood of close. I …nd no evidence indicating
di¤erences between time periods and between fund investment horizons.
Overall, the results in this section indicate long horizon funds increase ownership
level and length after spin-o¤s than before. Furthermore, evidence indicates funds
are more likely to shorten ownership length after increases in operating performance,
but this result primarily stems from funds with shorter investment horizons. The
importance of the spin-o¤ is not a factor in changes of ownership length after the
e¤ective date, but does factor in changes in ownership percentage especially for short
horizon funds.
2.7 Chapter Conclusion
This chapter …nds evidence indicating signi…cant changes in institutional own-
ership patterns surrounding corporate spin-o¤s. They include not only changes in
the percentage of shares held but the relationship between …rm performance and
ownership length. The interpretation of the results can be taken from the context
of di¤erences between conglomerates and …rms with greater business focus. Insti-
tutional shareholders, especially those with longer investment horizons, prefer …rms
with greater business focus. If …rm managers are concerned with the make-up of
institutional investors, then …nancial policy would be set to avoid investment into
unrelated businesses or where growth prospects and pro…tability di¤er.
70
Chapter 3: Payout Policy
3.1 Introduction
In general, …rm managers prefer greater ownership stability by their institutional
shareholders. Longer-term investors not only allow companies to pursue long-term
strategies, but also are more likely to aid …rm managers by communicating both their
private outlooks as well as the opinions of sell-side analysts. Conversely, institutions
that hold shares for shorter periods of time are more likely to exert greater pressure on
…rm managers to act myopically, oftentimes on threat of removal or company takeover
(Useem(1996)).
Although publicly traded companies cannot ultimately control the identity of its
institutional shareholders, …nancial policy can impact its composition. Anecdotally,
this can be seen in two examples. First, in 1989 Sealed Air Corporation engaged in a
leveraged recapitalization by borrowing most of the market value of its common stock
and distributing the funds in a special dividend. The event caused not only internal
change, but also a turnover from an investor base interested in consistent growth to
an investor base seeking large gains in pro…tability (Wruck (1994)). Another example
is the high price of Berkshire Hathaway class A shares. Warren Bu¤ett claims he is
able to retain a "slightly more long-term-oriented group of investors" by not initiating
a stock split.
22
In this chapter, I empirically investigate the relationship between a …rm’s payout
policy and the stability of its institutional shareholders. Institutions are typically
linked to dividend paying …rms because of their relative tax-advantage on dividend
income compared to other investors. Shleifer and Vishny (1986) and Allen, Bernardo,
and Welch (2000) use institutional tax-clienteles as the basis for their models of divi-
dend payout. They theorize …rms pay dividends as a means to attract tax-advantaged
22
The quote is from a Brent Schlender interview with Warren Bu¤ett and Bill Gates, printed in
the July 20, 1998 edition of Fortune magazine.
71
institutions in exchange for greater corporate oversight. Even though a segment of the
market faces higher tax rates on dividend income than capital gains, the combination
of dividend payout and greater institutional ownership (and oversight) maximizes eq-
uity value. Share repurchases have also been linked to greater institutional ownership.
Barclay and Smith (1988) and Brennan and Thakor (1990) use the informational ad-
vantage held by institutions over individual investors to predict greater institutional
ownership for share repurchasing …rms. The ability to pro…t at the expense of un-
informed (individual) investors by tendering over-valued shares during buyback pro-
grams motivates institutional (informed) investors to own share repurchasing …rms.
I create several variables to measure ownership stability describing the investment
horizon and longevity of a …rm’s institutional shareholders. Unique to this study,
all measures of ownership stability has as its basis the full ownership history of all
current stock positions held by institutions. For this chapter, I take stock positions
at the mutual fund level rather than at the investment company level. By focusing
on mutual funds, I am able to increase the number of institutional shareholders in
my dataset than what is typically utilized in related work.
I take fund positions from the Thomson Reuters (S12) Mutual Fund dataset from
1980 to 2007. The S12 database consists of positions from most domestic mutual
funds and some global funds that participate in US and Canadian equity markets.
The primary source for this dataset is SEC N-30D …lings. For the majority of the
time period, the SEC required mutual funds to …le this form semiannually. Thomson
Reuters supplements the N-30D …lings by examining fund prospectuses and by con-
tacting mutual funds directly. The other possible approach is to use the Thomson
Reuters (13f) Investment Company dataset. The 13f database consists of holdings
of banks, insurance companies, parents of mutual funds, pensions, and endowments.
The primary source for this dataset is quarterly SEC 13f …lings, required by all insti-
tutional investment managers that exercise investment discretion over $100 million.
72
Important to this study, the 13f dataset aggregates all holdings under a manager’s
control clouding any di¤erences in investment styles between funds within the same
institution.
The time period of study is from 1988 to 2007. I choose 1988 as the beginning
of the time period to avoid the historically large di¤erential tax rate between capital
gains and ordinary income prior to the Tax Reform Act (TRA) of 1986. By 1988, the
tax rate on both ordinary income and capital gains were set at 28% for individuals
in the highest tax bracket. The beginning of the sample period also avoids the early
1980s when few funds are observed in the data.
In the …rst part of the analysis, I investigate the correlation between …rm payout
policy and fund ownership. I start by estimating a tobit model explaining aggregate
ownership percentage by funds with short, medium, and long investment horizons. A
fund’s investment horizon is equal to the average number of months it holds each stock
position from the date of initial investment to the date of measurement. I account for
all position changes using the …rst-in-…rst-out queueing method. I calculate a fund’s
investment horizon annually, using only stock positions held for at least one month
during the year of measurement. I classify funds as either having short, medium, or
long investment horizons using annual tercile breakpoints.
For each investment horizon tercile I estimate four separate models using one of the
following four sets of payout variables to control for a …rm’s payout policy: dividend
and repurchase yields, total payout yield, dividend and repurchase indicator variables,
and a payout indicator variable. I …nd mutual funds with longer investment horizons
take greater ownership in dividend paying …rms and share repurchasing …rms than
mutual funds with shorter investment horizons. However, whereas share repurchases
attract greater ownership by long-horizon funds, dividends repel ownership by short-
horizon funds. I …nd this result whether I use yields or indicator variables to control
for …rm payout policy.
73
The results fromthe tobit regressions con…rmthe …ndings of Grinstein and Michaely
(2005) for mutual fund investors. The authors …nd greater institutional ownership
for dividend paying and share repurchasing …rms in general, but a greater attraction
to …rms with higher repurchase yields than dividend yields. The results here demon-
strate the di¤erences in ownership are not uniform across institutions and depend on
the fund’s investment horizon.
I next investigate the e¤ect payout policy has on the length of time funds with
short, medium, and long investment horizons hold stock positions. I estimate trun-
cated regressions similar to the tobit regressions mentioned above, but instead use
relative ownership length as the dependent variable. Relative ownership length is a
within-fund measure, equal to the proportion of other stock positions held on average
for a strictly shorter period of time. With a range from 0 to 1, this variable can be
thought of as the output of a fund-speci…c cumulative distribution function of average
ownership length. Similar to the regressions of aggregate fund ownership, I …nd fund
shareholders with longer investment horizons hold dividend paying or share repur-
chasing …rms for longer periods of time. However, I again …nd share repurchasing
…rms are more attractive to a broader range of funds than dividend paying …rms. For
instance, although long-horizon funds hold both dividend paying and share repur-
chasing …rms longer, short-horizon funds hold dividend paying …rms for signi…cantly
shorter periods of time.
The results in the …rst part of the chapter demonstrate how both dividends and
repurchases are positively related to greater ownership stability by attracting more
ownership and longer ownership by funds with longer investment horizons. Although
dividend paying …rms have less long-horizon fund ownership than share repurchasing
…rms, they are also held by fewer short-horizon funds.
In the second part of the analysis, I investigate whether payout events have sig-
ni…cant e¤ects on the composition and ownership length of fund shareholders. I use
74
as payout events dividend initiations, increases, decreases, and omissions, and share
repurchases. I distinguish share repurchases by whether it is an initiation or a contin-
uation of a repurchase program. I also distinguish share repurchases by non-dividend
and dividend paying …rms to account for ownership di¤erences related to a …rm’s
dividend policy. For the tests in this section, I measure fund ownership at the …rm
level with two primary measures. The …rst variable, shareholder investment horizon,
is equal to the average investment horizon of a …rm’s fund shareholders. The sec-
ond variable, current ownership length, is equal to the average length of time fund
shareholders have held …rm stock. I also quantify the changes in average shareholder
investment horizon and current ownership length by investigating changes in owner-
ship percentage by funds with short, medium, and long investment horizons.
Among all payout events, only dividend increases have a long-term signi…cant
change on average shareholder investment horizon. Dividend increases increase aver-
age shareholder investment horizon, with the change in fund composition primarily
stemming from a decrease in short-horizon fund ownership. However, only sizable
share repurchases, not dividend events, change fund ownership length. Interestingly,
share repurchases can increase or decrease ownership length depending on a …rm’s
dividend policy. For instance, although share repurchases of non-dividend paying
…rms decrease current ownership length, share repurchases of dividend paying …rms
increase current ownership length. The di¤erence stems from a greater turnover in
the shareholder base for non-dividend paying …rms than for dividend paying …rms.
The results to this point indicate that payout policy has a signi…cant e¤ect on
shareholder composition and the length of fund ownership. I extend the analysis in
two directions. First, I investigate the e¤ect of dividend taxes on the relationship
between payout policy and ownership stability by comparing fund ownership before
and after the Jobs and Growth Tax Relief Reconciliation Act (JGTRRA) of 2003.
The JGTRRA equated the tax rate on dividend income and capital gains at 15%
75
for individuals in the highest tax bracket. The tax-reform was made retroactive to
January 1, 2003.
To begin, I test for di¤erences in ownership percentage and relative ownership
length by investment horizon tercile before and after the JGTRRA. I estimate tobit
and truncated panel regression models similar to the ones above with …rm-year ob-
servations from 2002 and 2004. I control for the tax-reform by including a tax-period
indicator variable and interaction terms between the tax-period indicator variable and
payout variables. Signi…cance of the interaction terms represents a change in fund
ownership as the result of the tax-reform. I …nd the JGTRRA had no signi…cant
e¤ect on the percentage of fund ownership for all fund investment horizon classi…-
cations. However, short- and medium-horizon funds held dividend paying …rms and
…rms with higher dividend yields longer after the tax-reform than before. I …nd no
evidence indicating a di¤erence in the relationship between share repurchases and
fund ownership as a result of the JGTRRA.
Next, I test for di¤erences in the change in fund ownership as the result of payout
events between 1999 to 2001 and 2004 to 2006. I do not use payout events from
2002 and 2003 because changes in ownership may be related to the tax-reform, not a
payout event. I again measure ownership stability with shareholder investment hori-
zon, current ownership length, and the ownership percentage of funds by investment
horizon tercile. I …nd little indication of a signi…cant di¤erence in ownership change
as the result of dividend events or repurchase events between the two time periods.
Overall, the results in this section suggest the tax-reform did not have an overall
e¤ect.
In the second extension, I investigate whether shareholder stability is a signi…cant
factor in payout choice. Speci…cally, I compare fund ownership composition and fund
ownership length for dividend paying …rms that either increase dividends or repur-
chase shares. I study dividend paying …rms because of their relative homogeneity and
76
their already established long-term commitment to regularly pay dividends. I …rst
study fund shareholder investment horizon characteristics with average shareholder
investment horizon, current ownership length, and the ownership percentage of funds
by investment horizon tercile before and after the payout events. I …nd dividend pay-
ing …rms that increase dividends have lower average shareholder investment horizon,
are held for shorter periods of time, and have greater short-horizon fund ownership
and less long-horizon fund ownership prior to the event year than dividend paying
…rms that repurchase shares. As the result of the distributions, dividend increases
raise shareholder investment horizon more so than share repurchases. Examination
of ownership changes by funds with short, medium, and long investment horizons
indicates the increase in shareholder composition stems from a signi…cant decrease in
short-horizon fund ownership and a signi…cant increase in long-horizon fund owner-
ship.
To investigate the e¤ect of pre-event ownership stability more thoroughly, I es-
timate a bivariate probit model explaining a dividend paying …rm’s choice to either
increase dividends or repurchase shares. I use pre-event shareholder investment hori-
zon, current ownership length, and the ownership percentage by funds with short,
medium, and long investment horizons in the regressions to control for the stabil-
ity of fund shareholders. I initially …nd the probability a …rm increases dividends is
negatively related to shareholder investment horizon. That is, the longer the invest-
ment horizons of a …rm’s fund shareholders the less likely it will increase dividends.
However, this negative relationship with respect to average shareholder investment
horizon stems primarily from a positive relationship with short-horizon fund owner-
ship. I …nd no evidence indicating the length of time fund shareholders hold …rm
shares is a determinant in payout choice. Lastly, I investigate whether changes in
pre-event fund ownership di¤ers between dividend paying …rms that choose one pay-
out form over the other. Prior to the event year, I …nd dividend paying …rms that
77
increase dividends experience a larger increase in short-horizon fund ownership and
a smaller increase in long-horizon fund ownership than dividend paying …rms that
repurchase shares. Overall, the results in this section suggest that investor clientele
can be a signi…cant determinant in …rm payout choice.
3.2 Payout Literature
In this ection, I review past work relating to payout policy and institutional share-
holder clienteles as well as adjacent literature that classi…es institutional ownership
with investment horizon measures.
The role of dividends as means to attract institutional investors is supported by
two theoretical works. Shleifer and Vishny (1986) and Allen, Bernardo, and Welch
(2000) theorize the combination of lower tax rates faced by institutions and their abil-
ity to monitor …rm actions result in larger institutional ownership of dividend paying
…rms and greater …rm value. In Section 5 of their paper, Shleifer and Vishny (1986)
model dividends as a side payment from many small tax-disadvantaged shareholders
to one large tax-advantaged investor to retain its stake in the …rm. Allen et al. (2000)
model dividends similarly; however, the authors provide a clientele tilting argument
where the level of institutional ownership and thus oversight is directly related to the
level of dividends paid.
The existence of dividend tax-clienteles among large or important shareholders
has been investigated by several researchers. Pérez-González (2002) examines the ef-
fect tax reforms have on a …rm’s dividend policy with respect to the tax classi…cation
of their large shareholders. He …nds …rm dividend policy to be much more responsive
to personal income tax changes when the …rm’s large shareholders are individuals.
Desai and Jin (2007) test for dividend tax-clientele e¤ects within institutions. The
authors classify institutions based on the tax-sensitivity of its shareholder base. They
…nd a negative relationship between a …rm’s dividend payout ratio and the percentage
of its institutional shareholders with tax-sensitive clients, and a positive relationship
78
between changes in dividend policy and the tax-sensitivity of their institutional share-
holders. Hotchkiss and Lawrence (2007) test for institutional dividend clienteles by
relating dividend increases and decreases to changes in institutional shareholdings.
They classify institutions by either portfolio dividend yield or the percentage of stock
positions with high dividend yields. They con…rm the existence of dividend clienteles
by …nding a positive relationship between an institution’s demand for dividends and
ownership changes surrounding dividend increases and dividend decreases. Hotchkiss
and Lawrence do not examine shareholder horizons, nor do they investigate the e¤ect
repurchase activity has on institutional ownership. Barclay, Holderness, and Sheehan
(2009) …nd no evidence of tax-clienteles among corporate blockholders.
International evidence of dividend clienteles is found by Desai and Dharmapala
(2009) and Ferriera, Massa, and Matos (2009). Desai and Dharmapala (2009) use
the JGTRRA of 2003 as an exogenous event to analyze domestic portfolio holdings
of international …rms. They …nd evidence indicating a greater increase in foreign
portfolio investment in tax-favored countries than countries without a tax treaty with
the United States. Ferriera, Massa, and Matos (2009) …nd international institutional
ownership to be greater for …rms that do not pay dividends. They attribute the
relationship to transaction costs associated with dividend repatriation or dividend
reinvestment.
In general, however, dividends do not attract greater institutional ownership. In
an extensive analysis, Grinstein and Michaely (2005) investigate overall institutional
ownership and …rm payout policy. Although the authors …nd greater institutional
ownership for dividend paying …rms, ownership does not increase with dividend yield.
Instead, they …nd some evidence indicating institutional ownership decreases with
dividend yield.
Although the results of Grinstein and Michaely (2005) do not support Allen et al.
(2000), the greater institutional ownership of share repurchasing …rms and the positive
79
relationship between ownership and repurchase yield does support the theories of
Barclay and Smith (1988) and Brennan and Thakor (1990). Both models predict
…rms will choose between dividends and repurchases based on its mix of informed
and uninformed shareholders. Following repurchases, informed shareholders will own
relatively more of an undervalued …rm and relatively less of an overvalued …rm than
uninformed shareholders. Informed shareholders such as institutions will prefer share
repurchasing …rms, and uninformed shareholders like individuals will prefer dividends
to avoid the adverse selection.
Although institutions may not be as attracted to dividends than repurchases,
increases in institutional ownership of dividend paying …rms has had an a¤ect on
equity returns following dividend events. Amihud and Li (2006) …nd institutional
ownership to be partly responsible for the decline in short-run cumulative abnormal
returns surrounding dividend increases and decreases. They conclude that as the
level of institutional ownership has increased, the signaling function of dividends
has decreased. Gompers and Metrick (2001) …nd evidence indicating the increase in
institutional ownership has led to the disappearance of small stock premiums.
Past research has found evidence indicating the type of institution is important
in explaining future returns. Hotchkiss and Strickland (2003) …nd the stock price
response around negative earnings’ announcements is more negative for …rms held
more widely by growth, momentum, and high turnover investors. Yan and Zhang
(2009) …nd short-horizon institutional ownership positively relates to future returns
and earnings surprises. Bushee (2001) …nds investors less content to buy and hold
stock positions are more likely to overweight near-term expected earnings.
The investment horizon of institutional shareholders is also related to corporate
governance. Bushee (1998) …nds …rms with more transient institutional shareholders
are more likely to cut research and development costs to meet earnings. On the
other hand, Wahal and McConnell (2000) …nd no evidence indicating institutions,
80
short-horizon or otherwise, adversely a¤ect corporate investment. Bøhren, Priestley,
and Ødegaard (2005, 2008) …nd a negative relation between long-horizon institutional
ownership and …rm performance. They argue the decrease in …rm performance the
direct result of delegated monitoring by institutions more concerned with short-term
gain, as opposed to direct monitoring by individual investors. Elyasiani, Jia, and Mao
(2006) …nd greater ownership stability by institutional shareholders to be negatively
related to the cost of debt.
Papers by Gaspar, Matos, and Massa (2005) and Chen, Harford, and Li (2007)
investigate the e¤ect of ownership by shareholders with di¤erent investment hori-
zon around acquisitions. Consistent with short-horizon shareholders having a weaker
bargaining position, Gaspar, Matos, and Massa (2005) discover short-horizon share-
holders increase the probability of a takeover and lowers the acquisition premium for
target …rms. Furthermore, the returns of bidding …rms post-merger announcement
are negatively related to the proportion of short-horizon shareholders indicating an
absence of strong outside governance. Similarly, Chen, Harford, and Li (2007) …nd
independent long-horizon investors positively relates to post-merger performance.
3.3 Firm Data
In this section I describe the …rm-level data I use in the remainder of the chapter.
3.3.1 Sample
I extract annual data from the Center for Research in Securities Prices (CRSP)
monthly stock …le and the Compustat Fundamentals Annual data …le for each De-
cember from 1980 to 2008. For all tests, a …rm must have ordinary common stock
(CRSP share code 10 or 11) listed on the NYSE, AMEX, or NASDAQ (CRSP header
exchange code 1, 2, or 3), have return data for 36 months and …nancial data for 3
years. I include utilities (SIC codes 4949 to 4999) and …nancials (SIC codes 6000 to
6999) due to their propensity to pay dividends and repurchase shares.
81
3.3.2 Payout Event Speci…cations & Measures
I use annual Compustat variables to measure …rm payout activity. To measure
dividend payout, I use common dividends paid per share by the ex-dividend date
(data26 or dvpsx_f). Unlike common dividends (data21 or dvc), this variable ex-
cludes payments in preferred stock in lieu of cash and other non-cash payments.
I classify a …rm as initiating dividends if the …rm did not pay dividends in year
t ÷1 (data26
t1
= 0), but did so in year t (data26
t
0). Likewise, I classify a …rm as
omitting dividends if the …rm did not pay dividends in year t but did so in year t ÷1.
For dividend paying …rms, I identify a dividend change increase when the percentage
change in dividends from year t ÷ 1 to year t ((data26
t
÷data26
t1
) , data26
t
) is
greater than 10%. If the percentage change in dividends is less than -10%, then I
classify a …rm as decreasing dividends. This speci…cation is similar to Denis, Denis,
and Sarin (1995) and Yoon and Starks (1995), who distinguish quarterly dividend
changes greater than 10%.
I use the purchase of common and preferred stock (data115 or prstkc) to measure
a …rm’s repurchasing activity. Other authors, notably Grullon and Michaely (2002),
adjust this measure with reductions in preferred stock redemption value (data56 or
pstkrv). However, because repurchases of preferred stock have been found to consti-
tute only a small portion of repurchase activity (Stephens and Weisbach (1998)), I
make no adjustment.
Following Stephens and Weisbach (1998), I classify a …rm as repurchasing shares
if repurchases in year t are greater than 1% of the previous year’s …scal year end
market value (data115
t
, MV
t1
). Market value (MV) is equal to the …scal year end
share price (data199 or prcc_f) times the …scal year end number of common shares
outstanding (data25 or csho). If a …rm repurchases greater than 1% of the previous
year’s …scal year end market value in either year t, t ÷1, t ÷2, then I classify the …rm
as having an active repurchase program. Grinstein and Michaely (2005) use a similar
82
speci…cation. Lastly, I classify a …rm that repurchases shares in year t but not in year
t ÷1 or year t ÷2 as initiating a share repurchase program.
I use the following variables to control for a …rm’s payout activity. The divisor
for payout yields is …rm size, similar to Grinstein and Michaely (2005).
« DivYld
t
(dividend yield) = Dividends paid per share as of the ex-dividend date
times …scal year end shares outstanding, divided by total assets (data6 or at)
(data26
t
data25
t
, data6
t
).
« RepYld
t
(repurchase yield) = The purchase of common and preferred stock
divided by total assets (data115
t
, data6
t
).
« TotYld
t
(total payout yield) = Dividends paid per share times shares out-
standing plus share repurchases, all divided by total assets ((data26
t
data25
t
+data115
t
) , data6
t
).
« DivInd
t
(dividend indicator variable) = A dummy variable equal to 1 if the …rm
paid dividends in year t, 0 otherwise.
« RepInd
t
(repurchase indicator variable) = A dummy variable equal to 1 if the
…rm has an active repurchase program in year t, 0 otherwise.
« PayInd
t
(payout indicator variable) = A dummy variable equal to 1 if the …rm
either pays dividends or has active repurchase program in year t, 0 otherwise.
3.3.3 Control Variables
I use the following …rm-level control variables in the tests below. Past authors
have found the following variables to be signi…cantly related to either institutional
ownership or …rm payout policy. I derive the following variables from CRSP data.
« AnnRet
t
(annual return) = Annual compounded monthly return percentage
__

m2[Jan, Dec]
(1 + :ct
m;t
)
_
÷1
_
in year t, using all 12 months : of data.
83
« SDRet
t
(return standard deviation) = Standard deviation of daily returns
_
:t.dc·.
d2[Jan 1, Dec 31]
ct
d;t
)
_
in year t, using all daily returns.
« SP500
t
(S&P 500 dummy) = A dummy variable equal to 1 if the …rm is a
member of the S&P 500 as of December of year t, 0 otherwise.
« Beta
t
(beta) = The sum of coe¢cients (,
1
+ ,
2
) from OLS regressions monthly
returns on current and lagged market returns ct
t
= c+,
1
`/t
t
+,
2
`/t
t1
+
c
t
). Regressions use up to 60 months of past monthly return data ending in
December of year t. I use monthly returns previous to the 36 required months
until the …rst month of a missing return. I use NYSE/AMEX value weighted
returns for market returns. Falkenstein (1996), and Bennett, Sias, and Starks
(2003) use a similar speci…cation.
« Volm
t
(trading volume) - The average monthly ratio of trading volume to
shares outstanding
_
:cc:
m2[Jan, Dec]
(Trad. Vol.
m;t
, ShrOut
m;t
)
_
in year t, using all
12 : months of data.
« Age
t
(…rm age) - The number of months a …rm has its stock listed on a public
exchange since the …rst list date at the end of calendar year t. I use the CRSP
begin exchange date variable to de…ne the start date.
I derive the following variables from Compustat data. I de…ne the …rm’s current …scal
year as of December of year t.
« Size
t
(…rm size) = The natural log of total assets (data6 or at).
« MB
t
(market-to-book ratio) = Fiscal year end market value divided by book
value (MV
t
,BV
t
). Book value (BV)is equal to the sum of total assets, deferred
tax and investment credit (data35 or txditc), and convertible debt (data79 or
dcvt), minus preferred stock (data10 or pstkl) and total liabilities (data181 or
lt).
84
« ROA
t
(return-on-assets) = Operating income before depreciation (data13 or
oibdp) divided by total assets (data13
t
,data6
t
).
« NonOp
t
(non-operating income) = non-operating income (data61 or nopi) di-
vided by total assets (data61
t
,data6
t
).
« CapEx
t
(capital expenditures) = Capital expenditures (data128 or capx) di-
vided by total assets (data128
t
,data6
t
).
« Debt
t
(debt) = Total long term debt (data9 or dltt) divided by total assets
(data9
t
,data6
t
).
I also use the di¤erence in annual return, standard deviation of returns, beta, trading
volume, …rm size, return-on-assets, non-operating income, capital expenditures, and
debt from year t ÷ 1 to year t. I straight di¤erence all variables except for return-
on-assets, non-operating income, capital expenditures, and …rm debt which are equal
to their respective Compustat measure at year t minus the measure at year t ÷1, all
divided by total assets in year t. I also create a measure of the abnormal change in
return-on-assets (AbROA
t
) equal to the ratio at year t minus the mean ratio between
years t ÷1 and t ÷2.
3.4 Fund Ownership Characteristics
In this section, I investigate fund ownership characteristics by investment horizon
tercile. I start by summarizing fund ownership by …rm size quintile, market-to-book
quintile, and general payout policy. I then investigate the determinants of aggregate
ownership and relative ownership length.
Table 3.1 presents fund ownership and the number of fund shareholders per …rm
by size quintile, market-to-book quintile, and general payout policy using observations
from 1988 to 2007. I aggregate ownership using all funds and by fund investment hori-
zon tercile. I de…ne fund ownership as the percentage of common shares outstanding
85
by funds at year end (Own%). In equation form, Own% for …rm , in year t held by
funds indexed by i is equal to
Own%
j;t
=

iI
S
i;j;t
ShrOut
j;t
(21)
where S is the number of shares held, 1 is the set of fund shareholders, and ShrOut is
the monthly CRSP measure of shares outstanding (shrout). I also calculate aggregate
ownership each year by fund investment horizon tercile. I distinguish aggregate short-
horizon fund ownership with Own%S, medium-horizon fund ownership with Own%M,
and long-horizon fund ownership with Own%L.
With respect to …rm size, I …nd aggregate fund ownership to be weighted more
towards large stocks than small stocks. Aggregate fund ownership is equal to 1.0%
for …rms in the lowest size quintile, and increases monotonically to 13.2% for …rms
in the highest size quintile. The general preference for large stocks stems primarily
from long-horizon funds. Short- and medium-horizon funds had ownership patterns
more evenly distributed between size quintiles. Sample funds also had a tendency to
invest more into …rms with higher market-to-book ratios. For …rms in the highest
two market-to-book quintiles, average fund ownership is equal to 9.3% and 8.8%.
On the other hand, …rms in the lowest two market-to-book quintile had ownership
percentages of 4.4% and 6.6%. The ownership patterns of short, medium, and long
investment horizon funds re‡ect the aggregate statistics.
Lastly, I separate …rms into one of four following categories: non-dividend paying
and non-share repurchasing, non-dividend paying and share repurchasing, dividend
paying and non-share repurchasing, and dividend paying and share repurchasing.
Similar to Grinstein and Michaely (2005), I …nd average mutual fund ownership to be
greater for …rms that either pay dividends or repurchase shares than …rms that do not.
Furthermore, …rms that pay dividends and repurchase shares or just repurchase shares
86
are held more widely than …rms that just pay dividends. Across investment horizon
terciles, I …nd the greater ownership of divided paying stocks stemming primarily from
long-horizon funds. Medium-horizon funds also have greater ownership in …rms that
distribute excess capital. Short-horizon funds exhibit no ownership patterns across
payout groups.
The following analysis investigates the determinants of ownership percentage and
ownership length for short, medium, and long investment horizon funds. Overall, I
…nd evidence indicating dividend paying and share repurchasing …rms have a more
stable shareholder base than non-paying …rms.
3.4.1 Determinants of Fund Ownership
I …rst investigate the role of a …rm’s payout policy in explaining fund ownership.
I estimate a tobit model for each fund investment horizon tercile explaining aggre-
gate ownership using data from 1988 to 2007. The dependent variable is equal to
Own%S, Own%M, or Own%L for funds in the …rst (short), second (medium), and
third (long) investment horizon terciles. I estimate four separate models using one
of the following four sets of payout variables to control for a …rm’s payout policy:
dividend and repurchase yields, total payout yield, dividend and repurchase indica-
tor variables, and a payout indicator variable. Other explanatory variables include
operating income, non-operating income, capital expenditures, debt, size, market-to-
book, annual returns, standard deviation of returns, beta, trading volume, …rm age,
S&P 500 inclusion, and industry …xed-e¤ects based on the Fama-French 48 Industry
Classi…cation.
23
I report Fama-MacBeth (1973) time-series average coe¢cients and
t-statistics from annual cross-sectional regressions. I adjust coe¢cient standard errors
for autocorrelation using a Newey-West adjustment to two lags.
Columns (1) through (3) of Panel A of Table 3.2 present the Fama-MacBeth
estimates for short-horizon fund ownership, medium-horizon fund ownership, and
23
Fama-French 12 and 48 Industry Classi…cations can be found on Ken French’s website.
87
long-horizon fund ownership. For short- and medium-horizon funds, I …nd dividend
yield to be a negative and signi…cant determinant of aggregate ownership. Dividend
yield is negative and signi…cant at the 1% for short-horizon funds, and at the 10%
level for medium-horizon funds. Dividend yield is not a signi…cant determinant for
long-horizon fund ownership. Conversely, I …nd share repurchase yield to be a positive
predictor of fund ownership for each investment horizon tercile. Coe¢cient estimates
increase from short-horizon funds to long-horizon funds, with short-horizon funds
showing indi¤erence to a …rm’s repurchase yield, and medium- and long-horizon funds
showing a signi…cant preference for …rms with higher repurchase yields. Results for
medium- and long-horizon funds are signi…cant at the 1% con…dence level. I …nd
similar results when I replace dividend yield and repurchase yield with the dividend
indicator variable and the repurchase indicator variable. Columns (4) through (6)
present the results. Although short-horizon funds still prefer non-dividend paying
…rms to dividend paying …rms, a …rm’s dividend status is not a signi…cant predictor of
medium- and long-horizon fund ownership. Conversely, although short- and medium-
horizon funds show no preference for share repurchasing …rms, long-horizon funds
have greater ownership in share repurchasing …rms.
The results in Panel A of Table 3.2 add to the institutional ownership patterns
found by Grinstein and Michaely (2005). I …nd all mutual funds are either indi¤erent
or completely adverse to holding dividend paying …rms. On the hand, no mutual fund
type has an aversion to share repurchasing …rms. Stated di¤erently, share repurchases
attract a broader range of mutual fund investors than dividends, consistent with the
payout theories of Barclay and Smith (1988) and Brennan and Thakor (1990).
Regressions describing the determinants of institutional ownership can also be
found in Falkenstein (1996), Gompers and Metrick (2001), Grinstein and Michaely
(2005), and Yan and Zhang (2009). Falkenstein (1996) uses mutual fund data from
1991 and 1992. He does not include payout measures in his regressions. Gompers and
88
Metrick (2001) use institutional (13f) data and control for dividend yield only. They
…nd dividend yield to be negatively related to institutional ownership. Grinstein and
Michaely (2005) use institutional (13f) data and control for dividend and repurchase
activity with both yields and indicator variables. They do not sort institutions by
investment horizon. Lastly, Yan and Zhang (2009) sort institutions (13f) by their
measure of investment horizon (TOM), but do not control for a …rm’s repurchase
activity. Similar to the results in this chapter, they …nd dividend yield to be negative
and signi…cant determinant for short-horizon institutional ownership, and a negative
but insigni…cant determinant for long-horizon institutional ownership.
With respect to other …rm characteristics, I …nd short- and medium-horizon funds
are more likely to take on greater market risk and follow growth strategies than long-
horizon funds. For instance, …rm beta and market-to-book ratio are both positive
and signi…cantly related at the 1% level for short-horizon and medium-horizon fund
ownership, but are insigni…cant determinants of long-horizon fund ownership. The
relationship between fund ownership and annual return decreases with investment
horizon, with short-horizon funds more likely to be momentum investors, and long-
horizon funds less likely to follow returns. Other di¤erences between fund investment
horizon terciles include inclusion in the S&P 500 and …rm age. Long-horizon funds
are more likely to invest in older …rms and …rms in the S&P 500 than short- and
medium-horizon funds. For both short- and medium-horizon funds, I …nd S&P 500
inclusion and …rm age to be negatively related to fund ownership at the 1% con…dence
level. Similarities across funds include their preference for …rms with higher operating
income, capital expenditures, …rm size, and volume, and their dislike for …rms with
higher debt and return standard deviation.
Past research has reached similar conclusions with respect to volume, size, age,
and S&P 500 inclusion. Di¤erences with past work can be found with respect to the
market-to-book, annual return, and return standard deviation. Gompers and Metrick
89
(2001) and Yan and Zhang (2009) …nd past returns to be a negative and signi…cant
determinant of aggregate ownership. However, both papers separate annual returns
into three-month and nine-month intervals. These authors also …nd book-to-market
(not market-to-book) to be a positive determinant of aggregate institutional owner-
ship. The positive relationship I …nd with market-to-book, however, agrees with a
similar result in Grinstein and Michaely (2005). Lastly, Falkenstein (1996), Gompers
and Metrick (2001), Yan and Zhang (2009) …nd mixed evidence with respect to return
volatility, whereas I …nd this variable to be negative and highly signi…cant. However,
their measures of return standard deviation use at least two years of monthly return
data, while I use daily returns over a 12 month period.
Panel B of Table 3.2 presents the results when I replace the dividend and re-
purchase yield variables with the total payout yield variable, and the dividend and
repurchase indicator variables with the payout indicator variable. Otherwise, the test
speci…cation remains the same. Columns (1) through (3) present the Fama-MacBeth
estimates for short-horizon fund ownership, medium-horizon fund ownership, and
long-horizon fund ownership. I …nd total payout yield to be a negative and insignif-
icant determinant of short-horizon fund ownership, but a positive and signi…cant
determinant of medium- and long-horizon fund ownership at the 5% level. Columns
(4) through (6) present the results when I replace the total payout yield variable with
the payout indicator variable. I …nd payout …rms are held signi…cantly less than non-
payout …rms by short-horizon funds, and are held signi…cantly more by long-horizon
funds. The implications of other …rm variables remain the same.
3.4.2 Determinants of Ownership Length
The results of Table 3.2 demonstrate the signi…cant e¤ects a …rm’s payout policy
can have on fund ownership across investment horizon terciles. I next investigate the
e¤ect a …rm’s payout policy has on ownership length. Each year, I calculate a stock
position’s relative ownership length (ROL) as the percentage of other positions held
90
within the same fund portfolio but for a strictly shorter period of time. For stock ,
0
held by fund i at the end of year t, ROL is equal to
ROL
i;j
0
;t
=

jJ
1
_
LT
i;j
0
;t
LT
i;j;t
_
N
i;t
(22)
where , indexes the set of all fund positions J, N
i;t
represents the number of fund
positions, and 1
_
LT
i;j
0
;t
LT
i;j;t
_
is equal to 1 if …rm ,
0
has been held strictly longer
than …rm ,, 0 otherwise.
24
ROL, with a range from [0, 1) , can be thought of as a
cumulative distribution function of average ownership length for each fund portfolio.
I only use open positions as of the fund’s last report date in the year of measurement.
This variable can be considered a measure of ownership stability; the longer a fund
buys and does not adjust its stock position, the greater the value of ROL.
I estimate a truncated regression model to explain ROL for each fund investment
horizon tercile using data from 1988 to 2007. Other than the regression model and
dependent variable, the test methodology remains the same as the tobit regressions
of aggregate fund ownership. This implies estimating four separate models using
the four sets of payout variables to control for a …rm’s payout policy, including the
same explanatory variables, reporting Fama-MacBeth (1973) time-series average co-
e¢cients and t-statistics, and adjusting coe¢cient standard errors for autocorrelation
using a Newey-West adjustment. To my knowledge, these tests are unique.
Columns (1) through (3) of Panel A of Table 3.3 presents results for short-,
medium-, and long-horizon funds, with dividend yield and repurchase yield controlling
for …rm payout activity. I …nd dividend yield to be a negative and signi…cant determi-
nant of relative ownership length within short-horizon fund portfolios at the 5% level.
I do not …nd a signi…cant relationship between dividend yield and relative ownership
length within medium- and long-horizon fund portfolios. Thus, although dividend
24
See Chapter 1.4 for a full de…nition of LT.
91
yield does not lengthen the amount of time …rms are held by funds with longer in-
vestment horizons, it does shorten the time its equity is held by short-horizon funds.
Like aggregate ownership, repurchase yield has a positive e¤ect on relative ownership
length for each fund investment horizon tercile. While I …nd a …rm’s repurchase yield
to be positive but insigni…cantly related to short-horizon fund ownership, this variable
is a positive and signi…cant predictor for the relative length of time medium- and long-
horizon funds hold …rm equity. This result is signi…cant at the 5% con…dence level
for long-horizon funds, and at the 10% con…dence level for medium-horizon funds.
Columns (4) through (6) present the results when I replace dividend yield and
repurchase yield with the dividend indicator variable and the repurchase indicator
variable. The results remain primarily the same. However, with this speci…cation,
I …nd dividend paying …rms are held signi…cantly longer by long-horizon funds than
non-dividend paying …rms. The coe¢cient estimate is equal to 0.023 and is signi…cant
at the 1% con…dence level (t-statistic = 5.59). Although the coe¢cient estimate itself
is not large (…rms that pay dividends are held relatively longer than 2.3% of other
positions within long-horizon fund portfolios), it does indicate that dividends lengthen
the amount of time …rms are held by their long-term fund investors. On the other
hand, I now …nd medium-horizon funds hold dividend paying …rms for signi…cantly
shorter periods of time. The results for share repurchases remain the same.
The results from Panel A of Table 3.3 indicate that …rms paying dividends or
repurchasing shares are held signi…cantly longer than non-payout …rms. Important
to increasing the ownership stability of a …rm’s fund shareholders, both dividends
and repurchases attract longer ownership by funds with longer investment horizons. I
again …nd dividends reducing the ownership by funds with short investment horizons,
this time with respect to ownership length.
With respect to other …rm variables, factors found to increase or decrease the
level of fund ownership typically have the same e¤ect on the length of fund ownership
92
regardless of investment horizon tercile. This is true for debt, market-to-book, …rm
size, annual return, return standard deviation, age, and S&P 500 inclusion. However,
trading volume, which was a positive and signi…cant determinant of ownership level,
is a negative and signi…cant determinant of ownership length. Also, although …rms
with high beta are found to have higher ownership by short- and medium-horizon
funds, they are held for a shorter period of time by all funds.
Panel B of Table 3.3 presents the results when I replace the dividend and re-
purchase yield variables with the total payout yield variable and the dividend and
repurchase indicator variables with the payout indicator variable. Otherwise the test
speci…cation remains the same. Columns (1) through (3) present the Fama-MacBeth
estimates for short-, medium-, and long-horizon funds. I …nd total payout yield to be
signi…cant determinant of relative ownership length for long-horizon funds only. For
short-horizon funds, total payout yield is a negative but insigni…cant determinant.
Columns (4) through (6) present the results when I use the payout indicator variable
to control for …rm payout policy. I …nd …rms that distribute excess capital are held for
signi…cantly shorter periods of time by short-horizon funds, but signi…cantly longer
by long-horizon funds. The signi…cance of the payout indicator variables for short-
and long-horizon funds is at the 1% con…dence level.
3.5 Ownership Changes Around Payout Events
In this section, I conduct event studies investigating the e¤ect payout changes have
on fund ownership. I use as payout events dividend increases, decreases, initiations,
and omissions, as well as share repurchase initiations and non-initiations. I distinguish
between share repurchases by non-dividend paying …rms and dividend paying …rms to
account for di¤erences in ownership related to a …rm’s dividend policy. I exclude …rms
with multiple payout events in the same year. I measure fund ownership with average
fund shareholder investment horizon and average current ownership length of fund
investors. I then quantify the changes in shareholder investment horizon and current
93
ownership length by investigating changes in ownership percentage by fund investment
horizon tercile. Overall, I …nd evidence indicating payout events can cause changes to
the ownership composition and ownership length of fund shareholders. Furthermore,
the changes in ownership is dependent on the type of payout event and the dividend
policy of the …rm.
For each payout event in year t, I measure an initial ownership changes from t ÷1
to t +1 , a subsequent ownership changes from t +1 to t +2 , and an overall ownership
changes from t÷1 to t+2. I use payout events from 1988 to 2006 for tests of ownership
change from t ÷1 to t +1, and payout events from 1988 to 2005 for tests of ownership
change from t + 1 to t + 2, and from t ÷1 to t + 2.
3.5.1 Changes in Shareholder Investment Horizon
I …rst measure changes in fund ownership around payout events with average fund
shareholder investment horizon. I measure average shareholder investment horizon
(SIH) as the average investment horizon of funds holding …rm , at year end. In year
t, SIH is equal to
SIH
j;t
=

iI
S
i;j;t
+ Log (FIH
i;t
)

iI
S
i;j;t
(23)
where S represents the number of shares held and i indexes the set 1 of all fund
shareholders. Because measurement of fund investment horizon is in‡uenced by fund
age, I take the average with respect to the natural log of fund investment horizon
(FIH).
25
The change in shareholder investment horizon between dates is equal to
SIH
j;t
0 = SIH
j;t
0 ÷ SIH
j;t
.
To ensure changes in shareholder investment horizon are not in‡uenced by changes
in the fund sample, I use funds present in the sample in year t ÷1 only, and use their
measure of investment horizon in year t ÷ 1 for year t + 1 and year t + 2. I require
event …rms to be held by at least 10 mutual funds at the start of each measurement
25
See Chapter 1.4 for a full de…nition of FIH.
94
period. This requirement is to ensure that changes in fund composition are not driven
by the trading behavior of a small number of funds. The results remain primarily the
same when I require just 5 fund shareholders.
I calculate both unadjusted and adjusted ownership changes. I de…ne unadjusted
changes as the di¤erence in fund ownership between dates. Adjusted ownership
change is equal to the ownership change of the payout event …rm minus the own-
ership change of a control …rm. I initially match an event …rm with a control …rm
based on similar payout policies at the start of the event year. I then exclude from
this sample all control …rms that had a payout event regardless of payout type. Fol-
lowing Lie (2001) and Grullon and Michaely (2004), I then narrow the number of
potential control …rms by requiring the same 48 Fama-French Industry Classi…cation,
and measures of MB
t1
, ROA
t1
, and ROA
t1
within 80% to 120% of the event
…rm’s value. For each event …rm c, I then choose the control …rm c which minimizes
[ROA
e;t1
÷ROA
c;t1
[ +[ROA
e;t1
ROA
c;t1
[ +[MB
e;t1
÷MB
c;t1
[ (24)
If I cannot …nd a match from this sample, I repeat the procedure but loosen the indus-
try restriction to all …rms within the same Fama-French 12 Industry Classi…cation. If
I still cannot …nd a match, I use the control …rm which minimizes Equation (24) with
no regard to industry classi…cation. The last iteration chooses the control …rm which
minimizes Equation (24) with no restrictions. Like event …rms, I require control …rms
to be held by at least 10 funds at the start of the measurement period. Tests of signif-
icance are based on the mean and median changes of both unadjusted and adjusted
changes. I use a two-tailed t-statistic to determine signi…cance of mean change, and
a two-tailed .-statistic from Wilcoxon rank tests to determine signi…cance of median
change.
Panel A of Table 3.4 reports changes in the average shareholder investment horizon
95
for …rms that either increase, decrease, initiate, or omit dividends. Panel A reports
both unadjusted and adjusted changes. I …nd dividend increasing …rms have a posi-
tive and signi…cant increase in unadjusted SIH over each time interval. All tests of
mean and median unadjusted changes are signi…cant at the 1% level. The mean SIH
change from t ÷ 1 to t + 2 is equal to 0.055. Considering average SIH for dividend
increasing …rms in year t ÷1 is equal to 3.029 and average SIH in year t + 2 is equal
to 3.084, the change in SIH is equal to 1.16 months (exp (3.084) ÷ exp (3.029)), or
a 5.63% increase. Dividend initiations, like dividend increases, increases unadjusted
shareholder investment horizon over each measurement period and are signi…cant at
the 1% level. Opposite results hold for …rms that either decrease or omit dividends.
I …nd a mean decrease in unadjusted SIH for …rms decreasing dividends equal to
-0.028, signi…cant at the 5% level, and median decrease equal to -0.43, signi…cant at
the 1% level. Lastly, I …nd dividend omissions have a negative e¤ect on unadjusted
SIH from t ÷ 1 to t + 1 , but no evidence of an overall unadjusted change is found
from t ÷1 to t + 2.
Except for dividend increases, changes in SIH as the result of dividend events
are not robust to similar ownership changes in control …rms. For dividend increases,
both mean and median adjusted changes over each measurement period are again
signi…cant at the 1% level.
Panel B of Table 3.4 mirrors Panel A, but reports changes in SIH around share
repurchase events. I …nd …rms that repurchase shares have signi…cant unadjusted
increases in shareholder investment horizon. This is true for both non-dividend paying
…rms and dividend paying …rms, and for repurchase initiations and non-initiations.
However, I …nd much more positive and signi…cant results with respect to share
repurchases of non-dividend paying …rms than dividend-paying …rms. For instance,
the overall mean unadjusted change in SIH for all repurchases of non-dividend paying
…rms is equal to 0.070, and is signi…cant at the 1% level with a t-statistic equal to
96
12.05. The overall mean unadjusted SIH change for share repurchases of dividend
paying …rms is equal to 0.013, with a t-statistic equal to 2.67.
I …nd little evidence indicating changes in ownership composition are robust to
similar changes in control …rms. However, I do …nd an abnormal initial increase in
SIH as the result of share repurchases of dividend paying …rms. This is especially
true for non-initiation share repurchases with a mean adjusted change equal to 0.013,
signi…cant at the 5% level.
3.5.2 Changes in Ownership Length
I next measure changes in fund ownership around payout events with current
ownership length (COL) of fund shareholders. COL for …rm , in year t is equal to
COL
j;t
=

iI
MV
i;j;t
+ Log
_
LT
i;j;t
_

iI
MV
i;j;t
(25)
where i indexes the set 1 of all funds holding …rm ,. The change in current ownership
length between dates is equal to COL
j;t
0 = COL
j;t
0 ÷ COL
j;t
. I include all open fund
positions held as of the fund’s last report date in a given year as long as it is not a new
position (since these positions have no ownership length). I use funds present in the
sample the year before the payout event only, and require …rms to be held by at least
10 funds at the start of the measurement period. Because I am excluding new fund
shareholders from this statistic, I have fewer event …rms that meet data requirements
than with tests of SIH change. I follow the same test methodology as with changes
in SIH, calculating both mean and median unadjusted and adjusted changes.
Panel A of Table 3.5 presents the unadjusted and adjusted changes in COL as
the result of dividend events. I …nd evidence indicating an (unadjusted) increase in
ownership length for all dividend events. However, these results may be the result of
a natural increase in COL from one period to the next. After controlling for a similar
97
change in ownership length in a control …rm, I …nd no signi…cant di¤erence in the
length of time fund shareholders hold …rm stock as the result of any dividend event.
Panel B presents tests of current ownership length change as the result of share
repurchases. I again …nd for all share repurchases a positive and signi…cant increase
in unadjusted current ownership length. However, after I subtract a similar change
in ownership length by a control …rm, I …nd the e¤ect of share repurchases to be
dependent on a …rm’s dividend policy. For non-dividend paying …rms, I …nd the
adjusted overall change in COL from t ÷1 to t +2 to decrease by roughly one month.
These results are signi…cant at the 1% level, and hold regardless of repurchase type.
For dividend-paying …rms, however, share repurchases lengthen current ownership.
The overall mean adjusted change for all share repurchases of dividend-paying …rms
is positive (0.029) and signi…cant at the 5% level.
3.5.3 Explaining Changes in Shareholder Investment Horizon & Current Ownership
Length with Fund Investment Horizon Tercile Ownership Changes
The …rst two parts of this section demonstrate that payout events can cause signif-
icant changes to the composition and ownership length of a …rm’s fund shareholders.
Here, I attempt to quantify the changes in SIH and COL by examining changes in
ownership percentage by funds in each investment horizon tercile (Own%S, Own%M,
and Own%L). Tests follow the same methodology as above. For the sake of brevity,
I report adjusted ownership changes only.
Panel A of Table 3.6 presents the results with respect to dividend events. Dividend
increasing …rms have signi…cant decreases in short-horizon fund ownership over each
time period. Tests of both mean and median change for each time period are negative
and signi…cant at the 1% level. This evidence implies the positive increase in SIH
around dividend increases is the result of a decrease in short-horizon fund ownership.
I also …nd a similar overall signi…cant (10% level) decrease in short-horizon fund
ownership for …rms that initiate dividends. Dividend decreases, conversely, cause a
98
decrease in long-horizon fund ownership. Like dividend initiations, the results are
signi…cant at the 10% level. I …nd no evidence indicating dividend omissions have a
signi…cant e¤ect on the ownership patterns of any fund type.
Panel B presents changes in ownership percentage as the result of share re-
purchases. For non-dividend paying …rms, share repurchases create much greater
turnover but no signi…cant overall change in shareholder composition. For instance,
short-horizon funds initially decrease ownership from t ÷1 to t +1, but subsequently
increase ownership the following year. For dividend paying …rms that repurchase
shares, I again …nd an initial signi…cant decrease in all fund investment horizon ter-
ciles. Unlike non-dividend paying …rms, I do not …nd reversals in fund ownership
after share repurchases of dividend paying …rms.
The changes in ownership by each fund investment horizon tercile explain the
changes in SIH and COL around share repurchases. Share repurchases of non-
dividend paying …rms do not change the overall composition of share investment
horizon but do cause greater turnover in fund shareholders. This explains the absence
of change in shareholder investment horizon but the decrease in average ownership
length of its fund investors. On the other hand, for dividend paying …rms the overall
decrease in fund ownership and lack of new shareholders has no e¤ect on shareholder
composition but increases the average length of time the …rm equity is held.
3.6 The E¤ect of the JGTRRA
Noted in the introduction, one part of the Jobs and Growth Tax Relief Reconcil-
iation Act (JGTRRA) of 2003 equated the tax rate on dividend income and capital
gains at 15% for individuals in the highest tax bracket. One e¤ect of the tax-reform
was a reverse in the dramatic decline in the propensity for …rms to pay dividends from
the previous decades (Fama and French (2001)). Chetty and Saez (2005) …nd a 20%
increase in dividend payout over the following six quarters starting in the beginning
of 2003, with a large number of …rms initiating or increasing dividends. The authors
99
also …nd a positive relationship after the tax-reform between a change in ownership
by taxable institutions and changes in dividend payout.
In this section, I investigate whether tax-e¤ects have an impact on the investment
horizon and length of fund shareholders holding dividend paying or share repurchasing
…rms. Overall, the results from this section indicate little change in fund ownership
as the result of the tax-reform. However, I …nd evidence suggesting an intricate
relationship between the disproportionate tax rates on dividend income and capital
gains and fund ownership length.
3.6.1 Ownership Characteristics
I …rst investigate the change in ownership percentage and relative ownership length
as a consequence of the tax-reform. I repeat the regressions of Chapter 3.4, but instead
use observations just before and just after passage of the JGTRRA. First, I estimate
a tobit regression with all …rm-year observations from 2002 and 2004 explaining the
ownership percentage by fund investment horizon tercile. I again estimate four re-
gressions depending on the set of payout control variables while including all other
explanatory variables and industry …xed-e¤ects. Within each speci…cation, I also
include variables controlling for an overall change in fund ownership and ownership
change relating to …rm payout. To control for overall changes in fund ownership
between the two periods, I include a tax-period (TaxPd) indicator variable. TaxPd
is equal to 1 if the …rm-year observation is from 2004, 0 otherwise. To control for
changes in fund ownership percentage speci…cally relating to …rm payout, I include
interaction terms between TaxPd and each payout variable in the regression. Tests
of fund ownership change relating to the JGTRRA are based on the signi…cance of
the interaction terms. I cluster standard errors at the …rm level.
Columns (1) through (3) of Table 3.7 present the regression estimates for TaxPd,
the payout control variables, and the interaction terms between TaxPd and the payout
control variables, for short, medium, and long investment horizon funds. To conserve
100
space, the table does not report the coe¢cient estimates of the other explanatory
variables. Table 3.7 is separated into four panels, one for each of the four sets of payout
control variables. Overall, I …nd no evidence indicating the JGTRRA had a signi…cant
e¤ect on the relationship between fund ownership and payout policy. Regardless of
payout speci…cation or fund investment horizon tercile, all payout variable and TaxPd
interaction terms were insigni…cant.
Second, I estimate a truncated regression model by fund investment horizon ex-
plaining ROL with all stock positions taken from 2002 and 2004. Other than the
regression model and dependent variable, the test methodology remains the same
as the tobit regressions above. Columns (4) though (6) present the results. Unlike
the regressions with ownership percentage, I …nd a signi…cant increase in the relative
length of time short- and medium-horizon funds hold stock positions. Panel A reports
regression estimates when payout yields are included in the regression. I …nd stocks
with high dividend yields are held longer after 2003 by short- and medium-horizon
funds with signi…cance at the 5% level. Interestingly, I …nd long-horizon funds hold
stocks with higher dividend yields for relatively shorter periods of time after 2003.
I do not …nd any evidence indicating funds own …rms with higher repurchase yields
after the tax-reform.
The results with dividend and repurchase yields for short- and medium-horizon
funds are con…rmed when I instead use dividend and repurchase indicator variables.
Panel B presents evidence indicating short- and medium-horizon funds hold dividend
paying …rms signi…cantly longer after 2003. Both results are signi…cant at the 1%
level and indicate dividend paying stocks are held 3% longer than other fund posi-
tions within short-horizon fund portfolios, and 2.3% longer within medium-horizon
fund portfolios. Unlike the regressions using dividend yield, I …nd no evidence indicat-
ing long-horizon funds hold dividend-paying …rms di¤erently after 2003 than before.
Little evidence is again found regarding a change in ownership length for share re-
101
purchasing …rms. Panels C and D report coe¢cient estimates with respect to total
payout yield and the total payout indicator variable. Across both regressions, I …nd
a …rm’s payout status to a¤ect short-horizon funds only, as they hold payout …rms
relatively longer after 2003 than before. Overall, the results from this section indi-
cate that although investor clienteles did not change as the result of the tax-reform,
the length of time funds with shorter investment horizons hold the stock of dividend
paying …rms has increased.
3.6.2 Ownership Changes Around Payout Events
Next, I compare changes in ownership as the result of payout events before and
after the JGTRRA. To measure the di¤erence in ownership change, I compare mean-
and median-adjusted changes before and after the 2003 tax-reform. I only study
adjusted changes to account for di¤erences in the fund sample between periods.
I use a similar methodology to compare ownership change here as in Chapter 3.5.
I measure ownership changes around event year t with SIH, COL, Own%S, Own%M,
and Own%L from t ÷ 1 to t + 1, from t + 1 to t + 2, and from t ÷ 1 to t + 2. Tests
of ownership change from t ÷ 1 to t + 1 use payout events from 1999 to 2001 and
2004 to 2006, and tests of ownership change from t + 1 to t + 2 and t ÷ 1 to t + 2
use payout events from 1999 to 2000 and 2004 to 2005. I do not use payout events
from 2002 and 2003 to avoid ownership changes related to the tax-reform. All other
methodology remains the same including using only funds present in sample at time
t ÷ 1, measuring a fund’s investment horizon at time t ÷ 1, excluding …rms with
less than 10 fund shareholders at the start of each time period, and determining the
control …rms. I use the two-sided t-statistic to test for a di¤erence in mean change, and
the two-tailed .-statistic from Wilcoxon rank-sum to test for a di¤erence in median
change.
Table 3.8 presents tests of di¤erences in SIH change, and Table 3.9 presents tests
of di¤erences in COL change. In both tables, the third and fourth columns present
102
the number of …rm events before and after 2003 and the …fth column through the
eigth column presents the di¤erence in mean and median change before and after the
tax-reform (change before minus change after) as well as test statistics. Overall, I
…nd little evidence indicating a di¤erence in changes in shareholder composition or
ownership length as the result of payout events between the two time periods.
Similar to Tables 3.8 and 3.9, Panel A of Table 3.10 presents the tests of di¤er-
ences in ownership percentage change by fund investment horizon tercile for …rms
with dividend events. I …nd a more positive change in short-horizon fund ownership
percentage from t ÷ 1 to t + 1 after the JGTRRA than before. The di¤erence in
initial mean (and median) adjusted change from before 2003 to after 2003 is equal to
-0.005, and is signi…cant at the 5% level. This provides further evidence indicating a
lengthening in the ownership of short-horizon funds as the result of the tax-reform.
Although, I do not …nd a signi…cant overall change (from t ÷ 1 to t + 2) by short-
horizon funds, I do …nd an overall change by medium horizon funds The di¤erence
in mean adjusted ownership change from t ÷ 1 to t + 2 for medium-horizon funds is
equal to 0.015, and is signi…cant at the 1% level. Interestingly, I …nd a more positive
change in long-horizon fund ownership after 2003 for dividend decreasing …rms, but
a more negative change in ownership for dividend initiating …rms.
Panel B presents the tests of di¤erences in ownership change around share re-
purchases. Although I …nd no evidence with respect to dividend paying …rms, I do
…nd evidence of a di¤erence in ownership change with respect to non-dividend pay-
ing …rms. Short-, medium-, and long-horizon funds have more positive ownership
changes before the JGTRRA than after for non-dividend paying …rms that repur-
chase shares. This is true for changes from t ÷ 1 to t + 1 (short- and long-horizon
funds), changes from t + 1 to t + 2 (medium-horizon funds), and changes from t ÷1
to t + 2 (short- and medium-horizon funds). Overall, this subsection presents some
evidence indicating the JGTRRA had some e¤ect on fund ownership changes around
103
payout events.
3.7 The E¤ect of Ownership Stability on Payout Choice
In this section, I investigate whether shareholder stability is a signi…cant fac-
tor in …rm payout choice. I compare fund ownership characteristics surrounding a
dividend paying …rms choice to either increase dividends or repurchase shares. I in-
vestigate dividend paying …rms because of their relative homogeneity compared to
non-dividend paying …rms and their already established long-term commitment to
regularly pay dividends. I …nd evidence suggesting that short-horizon funds, not
long-horizon funds, are important in explaining the payout choice of dividend paying
…rms. Whether short-horizon funds compel managers to increase dividends or …rm
managers increase dividends to attract more long-term shareholders, I …nd dividend
paying …rms experience a greater shift in investor clientele from short- to long-horizon
funds with dividend increases than share repurchases.
3.7.1 Pre-Event Comparison & Change in Fund Ownership
I begin by comparing pre-event and unadjusted changes in fund ownership between
dividend paying …rms that either increase dividends or repurchase shares. I employ
the same methodology used in Chapter 3.5 to compute unadjusted ownership changes
from t ÷1 to t +1, from t +1 to t +2, and from t ÷1 to t +2 around event year t. This
includes measuring fund ownership with SIH, COL, Own%S, Own%M, and Own%L,
using only funds available in the dataset in t ÷1, measuring fund investment horizon
at t ÷1, and requiring …rms to be held by at least 10 funds at the start of each time
period. I employ tests of mean and median ownership di¤erences between the payout
choices. The results are presented in Table 3.11.
Panel A presents tests of pre-event and ownership change di¤erences in SIH be-
tween dividend paying …rms that either increase dividends or repurchase shares. In
the year prior the payout event, I …nd dividend increasing …rms have lower share-
104
holder stability than share repurchasing …rms. Mean SIH for dividend paying …rms
that increase dividends is equal to 3.041, and mean SIH for dividend paying …rms
that repurchase shares is equal to 3.114. The di¤erence in means, -0.073, is signi…cant
at the 1% level with a t-statistic equal to 10.24. A similar result holds for tests of
median di¤erence. However, as the result of the payout events, dividend increases in-
crease average shareholder investment horizon more so than share repurchases. Firms
that increase dividends observe a change in SIH from t ÷ 1 to t + 2 equal to 0.056.
The increase is signi…cantly greater than the change in SIH as the result of share
repurchases (0.014) at the 1% level. Similar results are found from t ÷1 to t + 1 and
from t + 1 to t + 2.
Panel B presents the results with respect to COL. Prior to the event year, I
…nd dividend paying …rms that repurchase shares are held signi…cantly longer than
dividend paying …rms that increase dividends. The mean di¤erence is equal to -0.065,
and is signi…cant at the 1% level. However, unlike shareholder investment horizon,
share repurchases of dividend paying …rms increases current ownership length more so
than dividend increases. The overall change from t ÷1 to t +2 for dividend increases
is equal to 0.115, and for share repurchases is equal to 0.145. The mean di¤erence,
-0.30, is signi…cant at the 10% level with a t-statistic equal to 1.74.
Panel C presents comparisons in pre-event and changes in ownership percentage
of funds by investment horizon tercile. Prior to the event year, I …nd dividend paying
…rms that increase dividends have signi…cantly higher ownership by short-horizon
funds and signi…cantly lower ownership by medium- and long-horizon funds. As the
result of the payout event, however, dividend paying …rms that increase dividends
have a signi…cantly greater decrease in short-horizon fund ownership and a greater
increase in long-horizon fund ownership. These results relate directly to the changes
in SIH and COL found in Panels A and B. For instance, the increase in new long-
horizon fund ownership for dividend increasing …rms increases average shareholder
105
investment horizon but at the same time causes a relative decrease in the average
length of time fund shareholders hold …rm stock.
3.7.2 Tests of Pre-Event Fund Ownership
The evidence above indicates dividends lengthen shareholder investment horizon
more so than share repurchases. The results also indicate dividend paying …rms that
increase dividends instead of repurchase shares have less long-horizon shareholders
and more short-horizon shareholders prior to the event year. To determine whether
or not shareholder composition and ownership length is a determinate in payout
choice, I estimate a bivariate probit model explaining the choice of a dividend paying
…rm to either increase dividends or repurchase shares. The dependent variable for the
…rst equation is equal to 1 if the …rm increases dividends, 0 otherwise. The dependent
variable for the second equation is equal to 1 if the …rm repurchases shares, 0 oth-
erwise. Similar to a seemingly unrelated regression, the model allows for correlated
disturbances between the two equations. Speci…cally, for indicator variables y
1
and
y
2
, and the corresponding sets of dependent variables x
1
and x
2
, the bivariate probit
model can be written as
y
1
= x
1
,
1
+ c
1
. y
1
= 1 if dividend increase, 0 otherwise,
y
2
= x
2
,
2
+ c
2
. y
2
= 1 if share repurchase, 0 otherwise,
1 [c
1
[ x
1
. x
2
] = 1 [c
2
[ x
1
. x
2
] = 0.
\ c: [c
1
[ x
1
. x
2
] = \ c: [c
2
[ x
1
. x
2
] = 0 . and
Co· [c
1
. c
2
[ x
1
. x
2
] = j (26)
where c
1
and c
2
are regression error terms for the …rst and second equations, and j
is a constant.
I estimate three sets of regressions depending on the measure of fund ownership
106
stability. In the …rst set, I use as explanatory variables pre-event SIH, operating in-
come, non-operating income, abnormal operating income, capital expenditures, debt,
size, market-to-book, annual returns, standard deviation of returns, beta, and trad-
ing volume. In the second set of regressions, I replace pre-event SIH with pre-event
COL. In the third set of regressions, I use pre-event Own%S, Own%M, and Own%L
as measures of fund ownership. I again employ the Fama-MacBeth methodology, esti-
mating annual bivariate models from 1988 to 2006, and adjusting coe¢cient standard
errors with a Newey-West adjustment to two lags. I use dividend paying …rms that
either increase dividends and/or repurchase shares, and are held by at least 10 mutual
funds prior to the event year. Due to the relatively small number of …rms each year,
I include …xed-e¤ects based on the Fama-French 12 Industry Classi…cation, not 48.
Lastly, for both equations I report average annual marginal e¤ects of the independent
variables at the respective means.
Columns (1) through (4) of Table 3.12 presents the results of the model when I use
SIH as the measure of pre-event ownership. SIH is a negative and signi…cant predictor
of a dividend paying …rm’s choice to increase dividends. The coe¢cient, equal to -
0.365 (marginal e¤ect = -0.129), is signi…cant at the 5% level with a t-statistic equal
to 2.25. On the other hand, I do not …nd average shareholder investment horizon to
be a signi…cant predictor of the …rm’s choice to repurchase shares.
Similar to the results found by Stephens, Jagannathan, and Weisbach (2000), I
…nd dividend paying …rms with higher returns, beta, standard deviation of returns,
capital expenditures, and market-to-book ratio are more likely to increase dividends
than repurchase shares. Conversely, …rms with greater volume, operating income,
non-operating income, and size are more likely to repurchase shares. Stephens et al.
(2000) also include institutional ownership in their regressions, however, the authors
…nd no evidence indicating ownership by institutions is a signi…cant predictor of
payout choice.
107
Columns (5) through (8) presents regression estimates when COL is the measure
of pre-event ownership. I do not …nd evidence indicating current ownership length
is a signi…cant predictor of either payout choice. Columns (9) through (12) present
the results when I classify pre-event ownership by the percentage of shares held by
short-, medium-, and long-horizon funds. I …nd greater ownership by short- and
medium-horizon funds increases the probability a …rm will increase dividends, but
greater ownership by long-horizon funds decreases the probability a …rm will increase
dividends. The result is especially strong for Own%S, with a marginal e¤ect equal to
1.232 and signi…cance at the 1% level (t-statistic = 3.91). The results for medium-
and long-horizon fund ownership are signi…cant at the 10% level. Again, no evidence
is found indicating a signi…cant relationship between fund ownership and a …rm’s
choice to repurchase shares. Implications of other control variables remain the same.
The results of the bivariate probit model indicate that overall shareholder invest-
ment horizon is important, but ownership by short-horizon funds may be the most
signi…cant factor in a …rm’s choice to increase dividends. For the last part of this
analysis, I investigate whether changes in fund ownership prior to the payout event
are related to a dividend paying …rms choice in payout. For dividend paying …rms
that either increase dividends or repurchase shares in year t, I compare fund owner-
ship in year t ÷3 and fund ownership changes from t ÷3 to t ÷2, from t ÷2 to t ÷1,
and from t ÷3 to t ÷1. I again measure fund ownership with respect to SIH, COL,
and the ownership percentage by funds in each investment horizon tercile. The test
methodology remains the same as before.
Panel A of Table 3.13 presents the results with respect to SIH. I …nd dividend in-
creasing …rms have signi…cantly lower average shareholder investment horizon in year
t ÷3. Furthermore, whereas dividend paying …rms that repurchase shares experience
an increase in SIH prior to the event year, dividend increasing …rms experience a
decline. The di¤erence in mean and median changes are signi…cant at the 1% level.
108
The di¤erence in mean SIH change from t ÷3 to t ÷1 is equal to -0.029, equivalent
to 0.631 months.
26
Panel B presents changes in current ownership length. Although
both payout …rms have similar ownership lengths at t ÷ 3, the ownership length of
share repurchasing …rms increases prior to the event year more so than dividend in-
creasing …rms. The mean di¤erence in COL is equal to -0.021 and is signi…cant at
the 5% level.
Panel C presents comparisons of ownership percentage by fund investment hori-
zon tercile. Overall, I …nd dividend increasing …rms have more short-horizon and
less long-horizon fund ownership three years prior to the event year. Furthermore,
dividend increasing …rms experience a greater increase in short-horizon fund owner-
ship and a smaller increase in long-horizon fund ownership than share repurchasing
…rms prior to the event year. However, the changes in short-horizon fund ownership
is larger and more signi…cant than ownership changes by long-horizon funds. Taken
together, the results in ownership change before and around dividend increases and
share repurchases of dividend paying …rms indicate dividend increases are not just
associated with a greater change in shareholder composition but also with a greater
reversal in short-horizon fund ownership.
3.8 Chapter Conclusion
The results of this chapter demonstrate how payout policy is related to the owner-
ship stability of a …rm’s fund shareholders. Both dividend paying and share repurchas-
ing …rms have greater ownership and are held relatively longer by funds with longer
investment horizons. Furthermore, payout events can alter the identity and longevity
of fund shareholders. Whether fund investors dictate payout policy to match their
preferences or …rm managers match payout policy to the preferences of their fund in-
vestors, what matters is that payout policy is positively related to ownership stability.
26
0.631 months is equal to the di¤erence between exp(SIH
t1
) ÷ exp(SIH
t3
) for dividend in-
creasing and share repurchasing …rms.
109
This implies a bene…t to distributing excess capital that stems from the attraction
of greater and longer ownership by shareholders more focused on long-term growth.
Empirically, the heterogeneity I …nd between mutual funds provides a clear example
in the importance of controlling for institution type beyond general classi…cations
(pension fund, endowment, etc.). Seen here, a wide range of investment horizons
exist even within mutual funds.
110
T
a
b
l
e
1
.
1
:
F
u
n
d
I
n
v
e
s
t
m
e
n
t
H
o
r
i
z
o
n
S
u
m
m
a
r
y
S
t
a
t
i
s
t
i
c
s
T
e
r
c
i
l
e
F
I
H
S
u
m
m
a
r
y
S
t
a
t
i
s
t
i
c
s
B
r
e
a
k
p
o
i
n
t
s
S
h
o
r
t
F
I
H
M
e
d
i
u
m
F
I
H
L
o
n
g
F
I
H
Y
e
a
r
N
M
e
a
n
S
t
d
.
D
e
v
.
M
i
n
M
a
x
1
-
2
2
-
3
%
#
%
#
%
#
1
9
9
0
2
1
5
2
3
.
8
1
4
.
1
5
.
1
7
9
.
4
1
5
.
5
2
5
.
5
0
.
5
2
%
6
9
.
4
0
.
5
7
%
8
7
.
4
0
.
7
2
%
1
2
7
.
9
1
9
9
1
2
4
5
2
2
.
5
1
5
.
2
3
.
2
1
0
7
.
3
1
4
.
9
2
3
.
7
0
.
4
0
%
7
4
.
1
0
.
6
8
%
1
1
0
.
4
0
.
7
6
%
1
0
9
.
4
1
9
9
2
2
5
1
2
1
.
9
1
3
.
6
3
.
0
9
0
.
4
1
4
.
0
2
3
.
7
0
.
5
1
%
8
5
.
0
0
.
6
7
%
9
4
.
7
0
.
6
2
%
1
3
5
.
0
1
9
9
3
3
0
4
2
4
.
8
1
7
.
1
5
.
0
1
3
4
.
1
1
6
.
0
2
5
.
4
0
.
4
9
%
8
6
.
3
0
.
6
5
%
1
0
9
.
8
0
.
5
9
%
1
7
5
.
7
1
9
9
4
3
0
9
2
4
.
8
1
4
.
9
3
.
0
9
9
.
0
1
6
.
6
2
6
.
7
0
.
6
7
%
1
0
1
.
3
0
.
5
3
%
1
0
5
.
2
0
.
6
5
%
1
4
3
.
4
1
9
9
5
3
5
3
2
4
.
0
1
4
.
1
5
.
4
1
0
0
.
5
1
5
.
9
2
6
.
7
0
.
5
5
%
9
1
.
9
0
.
8
6
%
1
1
0
.
8
0
.
6
1
%
1
6
7
.
2
1
9
9
6
4
2
6
2
2
.
8
1
6
.
4
4
.
7
1
9
7
.
8
1
5
.
5
2
3
.
2
0
.
5
0
%
1
1
1
.
3
0
.
8
8
%
1
2
5
.
4
0
.
6
0
%
1
4
9
.
0
1
9
9
7
5
4
9
2
1
.
5
1
1
.
9
5
.
8
9
6
.
4
1
5
.
4
2
2
.
9
0
.
5
4
%
1
1
2
.
3
0
.
6
3
%
1
2
4
.
7
0
.
7
6
%
1
5
0
.
1
1
9
9
8
7
4
0
2
1
.
7
1
3
.
7
4
.
2
1
6
7
.
4
1
5
.
4
2
2
.
7
0
.
3
2
%
1
0
8
.
5
0
.
4
5
%
1
2
3
.
3
0
.
6
1
%
1
4
9
.
8
1
9
9
9
8
8
2
2
1
.
0
1
1
.
7
4
.
7
1
0
6
.
9
1
5
.
3
2
2
.
2
0
.
2
9
%
1
0
2
.
4
0
.
4
3
%
1
2
8
.
4
0
.
5
0
%
1
6
4
.
4
2
0
0
0
1
0
2
7
2
1
.
9
1
5
.
9
3
.
9
1
6
9
.
5
1
4
.
3
2
2
.
3
0
.
2
6
%
9
7
.
5
0
.
2
8
%
1
0
5
.
9
0
.
4
8
%
1
5
0
.
9
2
0
0
1
1
0
1
5
1
8
.
8
1
3
.
6
4
.
3
1
9
0
.
2
1
3
.
3
1
9
.
3
0
.
2
2
%
1
0
4
.
0
0
.
2
9
%
1
1
3
.
7
0
.
3
1
%
1
5
9
.
7
2
0
0
2
1
3
3
4
2
0
.
2
1
3
.
2
3
.
6
1
8
4
.
8
1
4
.
2
2
1
.
0
0
.
1
7
%
1
0
8
.
4
0
.
2
5
%
1
1
2
.
8
0
.
3
3
%
1
4
5
.
1
2
0
0
3
1
4
6
4
2
0
.
1
1
2
.
7
3
.
0
2
0
8
.
1
1
3
.
9
2
1
.
4
0
.
1
6
%
1
0
0
.
4
0
.
2
0
%
1
2
1
.
4
0
.
3
0
%
1
6
9
.
3
2
0
0
4
1
8
8
0
2
0
.
8
1
3
.
8
2
.
4
1
8
9
.
7
1
4
.
5
2
1
.
9
0
.
1
2
%
9
8
.
7
0
.
1
9
%
1
3
2
.
5
0
.
2
7
%
1
8
2
.
8
2
0
0
5
2
1
8
9
2
1
.
8
1
5
.
4
2
.
8
2
2
3
.
1
1
4
.
7
2
2
.
4
0
.
1
2
%
1
1
7
.
5
0
.
1
8
%
1
3
3
.
6
0
.
2
7
%
1
8
7
.
1
2
0
0
6
2
3
8
6
2
2
.
0
1
3
.
0
3
.
3
1
1
0
.
7
1
5
.
2
2
3
.
3
0
.
1
2
%
1
2
5
.
4
0
.
1
6
%
1
4
2
.
1
0
.
2
5
%
1
8
5
.
2
2
0
0
7
2
5
3
7
2
1
.
2
1
3
.
0
3
.
4
1
0
2
.
3
1
4
.
2
2
2
.
7
0
.
1
0
%
1
2
8
.
7
0
.
1
4
%
1
4
5
.
5
0
.
2
1
%
2
1
7
.
0
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
s
u
m
m
a
r
y
s
t
a
t
i
s
t
i
c
s
f
o
r
f
u
n
d
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
(
F
I
H
)
f
r
o
m
1
9
9
0
t
o
2
0
0
7
.
M
e
a
n
,
m
a
x
i
m
u
m
,
m
i
n
i
m
u
m
,
a
n
d
s
t
a
n
d
a
r
d
d
e
v
a
t
i
o
n
,
a
n
d
t
e
r
c
i
l
e
b
r
e
a
k
p
o
i
n
t
s
a
r
e
i
n
m
o
n
t
h
s
.
T
e
r
c
i
l
e
b
r
e
a
k
p
o
i
n
t
s
a
r
e
c
o
m
p
u
t
e
d
a
n
n
u
a
l
l
y
u
s
i
n
g
s
a
m
p
l
e
m
u
t
u
a
l
f
u
n
d
s
.
1
-
2
r
e
p
r
e
s
e
n
t
s
t
h
e
b
r
e
a
k
p
o
i
n
t
b
e
t
w
e
e
n
s
h
o
r
t
a
n
d
m
e
d
i
u
m
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
f
u
n
d
s
,
a
n
d
2
-
3
r
e
p
r
e
s
e
n
t
s
t
h
e
b
r
e
a
k
p
o
i
n
t
b
e
t
w
e
e
n
m
e
d
i
u
m
a
n
d
l
o
n
g
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
f
u
n
d
s
.
T
h
e
l
a
s
t
s
i
x
c
o
l
u
m
n
s
c
o
m
p
a
r
e
s
s
t
o
c
k
p
o
s
i
t
i
o
n
s
a
t
y
e
a
r
-
e
n
d
b
e
t
w
e
e
n
f
u
n
d
s
w
i
t
h
s
h
o
r
t
,
m
e
d
i
u
m
,
a
n
d
l
o
n
g
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
s
.
%
i
s
t
h
e
a
v
e
r
a
g
e
o
w
n
e
r
s
h
i
p
p
e
r
c
e
n
t
a
g
e
f
o
r
e
a
c
h
s
t
o
c
k
p
o
s
i
t
i
o
n
,
a
n
d
#
i
s
t
h
e
a
v
e
r
a
g
e
n
u
m
b
e
r
o
f
s
t
o
c
k
p
o
s
i
t
i
o
n
s
p
e
r
f
u
n
d
.
F
I
H
i
s
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
4
.
111
Table 1.2: Average Change in Fund Investment Horizon from Initial Measure
Years
in Sample All Short Medium Long
# FIH # FIH # FIH # FIH
1 6189 - 1804 - 1681 - 2704 -
2 3642 -0.56 1295 2.56 1120 0.28 1227 -4.63
3 2431 0.29 947 1.25 767 0.48 717 -1.17
4 1687 0.67 666 0.78 523 0.89 498 0.30
5 1085 0.14 433 0.61 345 0.21 307 -0.59
6 796 0.57 336 1.09 249 0.59 211 -0.30
7 595 0.22 266 -0.05 179 0.29 150 0.60
8 467 0.55 211 0.68 144 1.03 112 -0.33
9 345 0.24 158 0.10 104 0.32 83 0.41
10 271 0.74 125 0.87 81 -0.27 65 1.75
11 192 -0.55 84 0.21 57 -1.27 51 -0.99
12 132 0.96 58 1.17 43 0.91 31 0.64
13 87 1.48 38 0.01 26 4.57 23 0.42
14 58 1.26 24 1.74 17 0.74 17 1.10
15 45 1.13 20 -2.19 13 3.65 12 3.94
16 33 -0.79 16 0.65 9 -1.59 8 -2.76
17 28 1.97 13 2.14 7 0.29 8 3.17
18 23 -0.37 11 -0.01 6 -3.13 6 1.72
This table reports the number of funds and the average change in fund investment horizon
(FIH) from the fund’s …rst year in the sample to all subsequent years. Averages are com-
puted for all funds and by initial FIH tercile classi…cation. The average change in FIH is
in months. FIH is de…ned in Chapter 1.4.
112
Table 1.3: Comparison Between Fund Investment Horizon & Other Measures of
Portfolio Turnover
Panel A: Correlation Between FIH, TOT, and TOM Terciles*
FIH TOT TOM
FIH 1.00
TOT 0.45 1.00
TOM 0.44 0.78 1.00
Panel B: FIH Tercile Classi…cation as a Percentage of TOT and TOM Terciles
Short FIH Med. FIH Long FIH
Short Med. Long Short Med. Long Short Med. Long
TOT 0.565 0.335 0.100 0.276 0.411 0.314 0.160 0.255 0.585
TOM 0.563 0.313 0.124 0.294 0.419 0.287 0.143 0.270 0.587
Panel C: Between Year Percentage Change in Tercile Classi…cations
Short
t1
Med.
t1
Long
t1
FIH TOT TOM FIH TOT TOM FIH TOT TOM
Short
t1
0.683 0.681 0.677 0.280 0.261 0.274 0.047 0.058 0.075
Med.
t1
0.244 0.252 0.252 0.513 0.507 0.500 0.270 0.269 0.273
Long
t1
0.073 0.068 0.071 0.207 0.232 0.226 0.683 0.673 0.652
This table compares the investment horizon tercile classi…cations between fund investment hori-
zon (FIH) and two turnover based measures (TOT and TOT). Investment horizon terciles are
calculated annually. All sample funds from 1990 to 2007 are used. Panel A reports correlation
coe¢cients between classi…cations. Panel B reports the proportion of funds with short, medium,
and long investment horizons based on FIH, that have short, medium, and long investment hori-
zons based on either TOT or TOM. Panel C reports the proportion of funds within an investment
horizon tercile one year that either keep the same tercile classi…cation the following year, or switch
to one of the other two. FIH, TOT, and TOM are de…ned in Chapter 1.4.

All correlations sign…cant at the 1% level.
113
T
a
b
l
e
1
.
4
:
F
i
r
m
a
n
d
O
w
n
e
r
s
h
i
p
V
a
r
i
a
b
l
e
C
o
r
r
e
l
a
t
i
o
n
M
a
t
r
i
x
O
w
n
-
A
R
O
L
-
U
n
s
-
F
Y
-
3
M
-
F
c
s
t
-
S
I
H
S
M
L
A
R
O
L
S
M
L
D
A
D
A
E
S
M
B
D
e
b
t
S
i
z
e
R
e
t
R
e
t
S
D
O
w
n
%
S
-
0
.
5
1
O
w
n
%
M
-
0
.
2
8
0
.
3
7
O
w
n
%
L
0
.
3
7
0
.
1
6
0
.
2
4
A
R
O
L
0
.
1
7
-
0
.
0
5
0
.
0
6
0
.
1
6
A
R
O
L
S
-
0
.
1
6
0
.
2
3
0
.
1
0
0
.
0
8
0
.
3
6
A
R
O
L
M
-
0
.
0
2
0
.
0
0
0
.
2
4
0
.
1
1
0
.
5
3
0
.
1
7
A
R
O
L
L
0
.
1
8
-
0
.
0
8
-
0
.
0
3
0
.
1
5
0
.
6
6
0
.
1
0
0
.
1
7
D
A
-
0
.
0
3
0
.
0
2
0
.
0
3
-
0
.
0
1
0
.
0
1
0
.
0
1
0
.
0
2
0
.
0
0
U
n
s
D
A
-
0
.
0
7
0
.
0
6
0
.
0
1
-
0
.
0
4
-
0
.
0
7
-
0
.
0
5
-
0
.
0
8
-
0
.
0
6
-
0
.
0
4
E
S
-
0
.
0
4
0
.
0
5
0
.
0
4
0
.
0
1
0
.
0
0
0
.
0
3
0
.
0
2
0
.
0
2
0
.
0
1
-
0
.
0
5
M
B
-
0
.
1
4
0
.
1
8
0
.
1
1
-
0
.
0
3
-
0
.
0
4
0
.
0
5
0
.
0
1
-
0
.
0
4
-
0
.
0
5
0
.
0
9
0
.
0
4
D
e
b
t
0
.
0
2
-
0
.
0
9
-
0
.
0
6
-
0
.
0
6
0
.
0
3
0
.
0
2
0
.
0
3
0
.
0
3
0
.
0
3
-
0
.
1
1
-
0
.
0
5
-
0
.
0
6
S
i
z
e
0
.
0
0
-
0
.
0
8
-
0
.
0
5
-
0
.
0
3
0
.
0
4
0
.
1
5
0
.
1
4
0
.
0
5
-
0
.
0
2
-
0
.
2
1
0
.
0
4
-
0
.
0
9
0
.
3
4
F
Y
R
e
t
-
0
.
2
1
0
.
1
8
0
.
0
8
-
0
.
0
9
-
0
.
1
4
-
0
.
0
6
-
0
.
1
0
-
0
.
0
7
-
0
.
0
1
0
.
0
2
0
.
1
1
0
.
3
1
-
0
.
0
6
-
0
.
0
8
3
M
R
e
t
-
0
.
0
9
0
.
0
3
-
0
.
0
1
-
0
.
0
4
-
0
.
0
7
-
0
.
0
5
-
0
.
0
5
-
0
.
0
3
-
0
.
0
5
0
.
0
2
0
.
1
2
0
.
1
8
-
0
.
0
2
0
.
0
0
0
.
4
3
F
c
s
t
S
D
0
.
0
3
-
0
.
0
8
-
0
.
0
9
-
0
.
1
0
-
0
.
0
6
-
0
.
0
6
-
0
.
0
6
-
0
.
0
6
-
0
.
0
3
0
.
0
3
-
0
.
2
7
-
0
.
0
8
0
.
1
3
0
.
1
2
-
0
.
0
5
-
0
.
0
4
A
n
N
u
m
-
0
.
1
0
0
.
0
3
0
.
0
3
-
0
.
0
9
-
0
.
0
2
0
.
1
2
0
.
1
0
-
0
.
0
2
-
0
.
0
4
-
0
.
1
1
0
.
0
5
0
.
1
2
0
.
0
5
0
.
6
0
-
0
.
0
4
0
.
0
0
0
.
0
0
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
c
o
r
r
e
l
a
t
i
o
n
c
o
e
¢
c
i
e
n
t
s
b
e
t
w
e
e
n
…
r
m
-
l
e
v
e
l
v
a
r
i
a
b
l
e
s
.
O
w
n
e
r
s
h
i
p
v
a
r
i
a
b
l
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
4
.
A
l
l
o
t
h
e
r
v
a
r
i
a
b
l
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
3
.
114
Table 1.5: Summary of Earnings Announcements within One Penny of Analyst
Forecasts
Panel A: Overall
ES Overall
Obs. % Cum. %
-$0.01 1,534 21.8 21.8
$0.00 2,779 39.5 61.2
$0.01 2,732 38.8 100.0
Total 7,045 100.0
Panel B: SIH Tercile
ES Short Medium Long
Obs. % Cum. % Obs. % Cum. % Obs. % Cum. %
-$0.01 480 18.6 18.6 556 23.2 23.2 498 24.2 24.2
$0.00 1,026 39.7 58.2 915 38.2 61.4 838 40.6 64.8
$0.01 1,081 41.8 100.0 925 38.6 100.0 726 35.2 100.0
Total 2,587 100.0 2,396 100.0 2,062 100.0
Panel C: AROL Tercile
ES Short Medium Long
Obs. % Cum. % Obs. % Cum. % Obs. % Cum. %
-$0.01 470 21.6 21.6 519 21.5 21.5 545 22.3 22.3
$0.00 831 38.1 59.7 941 39.0 60.4 1,007 41.1 63.4
$0.01 880 40.4 100.0 956 39.6 100.0 896 36.6 100.0
Total 2,181 100.0 2,416 100.0 2,448 100.0
This table reports the percentage of …rms overall and by average shareholder investment horizon
(SIH) and average relative ownership length (AROL) terciles that either just beat (ES = $0.01),
meet (ES = $0.00), or just miss (ES = -$0.01) analyst earnings forecasts. I use …rm observations
from 1990 to 2007. Firms are classi…ed into SIH and AROL terciles annually. SIH and AROL are
de…ned in Chapter 1.4. ES is de…ned in Chapter 1.3.
115
T
a
b
l
e
1
.
6
:
O
r
d
e
r
e
d
P
r
o
b
i
t
R
e
g
r
e
s
s
i
o
n
s
D
e
s
c
r
i
b
i
n
g
E
a
r
n
i
n
g
s
S
u
r
p
r
i
s
e
s
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)
(
7
)
S
I
H
t
-
0
.
2
0
6
a
(
3
.
5
6
)
-
0
.
2
0
2
a
(
3
.
4
9
)
O
w
n
%
S
t
2
.
7
6
8
a
(
4
.
3
1
)
2
.
5
9
4
a
(
3
.
8
8
)
O
w
n
%
M
t
0
.
6
8
5
(
1
.
5
0
)
0
.
7
8
5
c
(
1
.
6
8
)
O
w
n
%
L
t
-
0
.
0
8
5
(
0
.
2
4
)
-
0
.
0
5
0
(
0
.
1
4
)
T
o
t
O
w
n
%
t
0
.
7
7
6
a
(
3
.
5
3
)
A
R
O
L
t
-
0
.
0
8
2
(
0
.
9
8
)
-
0
.
0
4
8
(
0
.
5
7
)
A
R
O
L
S
t
0
.
1
0
8
b
(
2
.
0
1
)
0
.
0
4
5
(
0
.
8
1
)
A
R
O
L
M
t
-
0
.
0
3
2
(
0
.
5
5
)
-
0
.
0
5
5
(
0
.
9
3
)
A
R
O
L
L
t
-
0
.
0
5
2
(
0
.
7
5
)
-
0
.
0
2
4
(
0
.
3
4
)
M
B
t
0
.
0
0
4
(
0
.
6
8
)
0
.
0
0
3
(
0
.
5
8
)
0
.
0
0
5
(
0
.
8
4
)
0
.
0
0
5
(
0
.
8
5
)
0
.
0
0
4
(
0
.
6
8
)
0
.
0
0
4
(
0
.
7
1
)
0
.
0
0
3
(
0
.
5
9
)
D
e
b
t
t
-
0
.
2
3
7
b
(
2
.
1
8
)
-
0
.
2
4
5
b
(
2
.
2
6
)
-
0
.
2
3
5
b
(
2
.
1
6
)
-
0
.
2
1
9
b
(
2
.
0
1
)
-
0
.
2
1
4
b
(
1
.
9
7
)
-
0
.
2
3
8
b
(
2
.
2
0
)
-
0
.
2
4
6
b
(
2
.
2
7
)
S
i
z
e
t
0
.
0
0
3
(
0
.
2
2
)
0
.
0
0
8
(
0
.
5
6
)
0
.
0
0
6
(
0
.
3
9
)
0
.
0
0
3
(
0
.
2
0
)
0
.
0
0
0
(
0
.
0
1
)
0
.
0
0
4
(
0
.
2
7
)
0
.
0
0
9
(
0
.
5
8
)
F
Y
R
e
t
t
0
.
1
1
9
a
(
4
.
1
0
)
0
.
1
1
3
a
(
3
.
9
6
)
0
.
1
3
2
a
(
4
.
6
2
)
0
.
1
3
4
a
(
4
.
5
9
)
0
.
1
3
9
a
(
4
.
7
7
)
0
.
1
1
7
a
(
3
.
9
9
)
0
.
1
1
2
a
(
3
.
8
7
)
F
c
s
t
S
D
t
-
1
.
8
2
9
a
(
4
.
3
5
)
-
1
.
7
8
3
a
(
4
.
2
6
)
-
1
.
7
8
1
a
(
4
.
2
3
)
-
1
.
9
0
2
a
(
4
.
4
4
)
-
1
.
8
5
6
a
(
4
.
4
1
)
-
1
.
8
5
2
a
(
4
.
3
6
)
-
1
.
7
9
9
a
(
4
.
2
8
)
A
n
N
u
m
t
0
.
0
0
3
(
0
.
9
4
)
0
.
0
0
2
(
0
.
6
9
)
0
.
0
0
3
(
0
.
8
3
)
0
.
0
0
3
(
1
.
0
6
)
0
.
0
0
3
(
1
.
1
3
)
0
.
0
0
3
(
0
.
8
8
)
0
.
0
0
2
(
0
.
6
7
)
O
b
s
.
7
0
4
5
7
0
4
5
7
0
4
5
7
0
4
5
7
0
4
5
7
0
4
5
7
0
4
5
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
e
s
t
i
m
a
t
e
s
f
r
o
m
o
r
d
e
r
e
d
p
r
o
b
i
t
r
e
g
r
e
s
s
i
o
n
s
e
x
p
l
a
i
n
i
n
g
t
h
e
d
i
¤
e
r
e
n
c
e
b
e
t
w
e
e
n
a
n
n
o
u
n
c
e
d
a
n
d
a
n
a
l
y
s
t
f
o
r
e
-
c
a
s
t
e
d
e
a
r
n
i
n
g
s
-
p
e
r
-
s
h
a
r
e
(
E
S
)
.
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
i
n
p
a
r
e
n
t
h
e
s
e
s
.
S
t
a
n
d
a
r
d
e
r
r
o
r
s
a
r
e
c
l
u
s
t
e
r
e
d
a
t
t
h
e
…
r
m
-
l
e
v
e
l
t
o
t
h
e
r
i
g
h
t
o
f
c
o
e
¢
c
i
e
n
t
s
.
E
a
c
h
r
e
g
r
e
s
s
i
o
n
u
s
e
s
…
r
m
s
t
h
a
t
a
n
n
o
u
n
c
e
e
a
r
n
i
n
g
s
w
i
t
h
i
n
o
n
e
p
e
n
n
y
o
f
m
e
d
i
a
n
a
n
a
l
y
s
t
e
s
t
i
m
a
t
e
s
f
r
o
m
1
9
9
0
t
o
2
0
0
7
.
I
n
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
i
n
c
l
u
d
e
…
r
m
-
l
e
v
e
l
m
e
a
s
u
r
e
s
o
f
o
w
n
e
r
s
h
i
p
s
t
a
b
i
l
i
t
y
,
m
a
r
k
e
t
-
t
o
-
b
o
o
k
r
a
t
i
o
,
d
e
b
t
,
s
i
z
e
,
a
n
n
u
a
l
r
e
t
u
r
n
,
s
t
a
n
d
a
r
d
d
e
v
i
a
t
i
o
n
o
f
a
n
a
l
y
s
t
f
o
r
e
c
a
s
t
s
,
n
u
m
b
e
r
o
f
a
n
a
l
y
s
t
s
c
o
v
e
r
i
n
g
t
h
e
…
r
m
,
y
e
a
r
…
x
e
d
-
e
¤
e
c
t
s
a
n
d
i
n
d
u
s
t
r
y
…
x
e
d
-
e
¤
e
c
t
s
b
a
s
e
d
o
n
t
h
e
F
a
m
a
-
F
r
e
n
c
h
1
2
I
n
d
u
s
t
r
y
C
l
a
s
s
i
…
c
a
t
i
o
n
.
O
w
n
e
r
s
h
i
p
m
e
a
s
u
r
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
4
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
a
n
d
a
l
l
o
t
h
e
r
e
x
p
l
a
n
a
t
o
r
y
v
a
r
i
a
b
l
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
3
.
S
i
g
n
i
…
c
a
n
c
e
a
t
t
h
e
1
%
l
e
v
e
l
i
s
d
e
s
i
g
n
a
t
e
d
w
i
t
h
a
,
t
h
e
5
%
l
e
v
e
l
w
i
t
h
b
,
a
n
d
t
h
e
1
0
%
l
e
v
e
l
w
i
t
h
c
.
116
T
a
b
l
e
1
.
7
:
L
i
n
e
a
r
R
e
g
r
e
s
s
i
o
n
s
D
e
s
c
r
i
b
i
n
g
E
a
r
n
i
n
g
s
S
u
r
p
r
i
s
e
s
E
S
>
$
0
.
0
1
o
r
E
S
<
-
$
0
.
0
1
E
S
>
$
0
.
0
1
S
I
H
-
0
.
0
1
3
c
(
1
.
7
0
)
-
0
.
0
1
0
(
1
.
1
1
)
O
w
n
%
S
t
0
.
1
6
8
c
(
1
.
8
7
)
0
.
0
4
6
(
0
.
3
8
)
O
w
n
%
M
t
0
.
0
2
7
(
0
.
4
3
)
0
.
1
5
8
(
1
.
3
0
)
O
w
n
%
L
t
-
0
.
0
9
3
b
(
2
.
2
8
)
0
.
0
3
6
(
0
.
4
8
)
T
o
t
O
w
n
%
t
-
0
.
0
0
5
(
0
.
1
7
)
0
.
0
7
6
(
0
.
8
8
)
M
B
t
-
0
.
0
0
1
(
1
.
1
2
)
-
0
.
0
0
1
(
1
.
2
7
)
-
0
.
0
0
1
(
0
.
9
9
)
0
.
0
0
2
(
1
.
2
1
)
0
.
0
0
2
(
1
.
2
3
)
0
.
0
0
2
(
1
.
2
4
)
D
e
b
t
t
-
0
.
0
6
1
a
(
3
.
8
9
)
-
0
.
0
6
0
a
(
3
.
8
8
)
-
0
.
0
6
0
a
(
3
.
8
7
)
0
.
0
4
1
c
(
1
.
7
2
)
0
.
0
3
9
c
(
1
.
7
4
)
0
.
0
4
0
c
(
1
.
7
4
)
S
i
z
e
t
0
.
0
1
2
a
(
5
.
8
4
)
0
.
0
1
3
a
(
5
.
8
8
)
0
.
0
1
2
a
(
5
.
8
4
)
0
.
0
0
3
(
0
.
7
4
)
0
.
0
0
4
(
0
.
9
3
)
0
.
0
0
4
(
0
.
9
3
)
F
Y
R
e
t
t
0
.
0
4
1
a
(
9
.
4
1
)
0
.
0
4
1
a
(
9
.
2
9
)
0
.
0
4
2
a
(
9
.
8
5
)
-
0
.
0
0
2
(
0
.
3
9
)
-
0
.
0
0
2
(
0
.
3
4
)
-
0
.
0
0
1
(
0
.
3
0
)
F
c
s
t
S
D
t
-
0
.
6
0
1
a
(
1
0
.
4
2
)
-
0
.
6
0
1
a
(
1
0
.
3
9
)
-
0
.
6
0
0
a
(
1
0
.
3
7
)
1
.
2
0
0
a
(
3
.
1
0
)
1
.
2
0
4
a
(
3
.
0
7
)
1
.
2
0
4
a
(
3
.
0
7
)
A
n
N
u
m
t
0
.
0
0
0
(
0
.
4
2
)
0
.
0
0
0
(
0
.
4
1
)
0
.
0
0
0
(
0
.
5
2
)
-
0
.
0
0
3
a
(
3
.
7
7
)
-
0
.
0
0
3
a
(
3
.
6
7
)
-
0
.
0
0
3
a
(
3
.
6
9
)
O
b
s
.
1
4
6
4
4
1
4
6
4
4
1
4
6
4
4
8
4
8
6
8
4
8
6
8
4
8
6
A
d
j
.
R
2
0
.
1
1
0
.
1
1
0
.
1
1
0
.
1
2
0
.
1
2
0
.
1
2
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
e
s
t
i
m
a
t
e
s
f
r
o
m
l
i
n
e
a
r
r
e
g
r
e
s
s
i
o
n
s
e
x
p
l
a
i
n
i
n
g
t
h
e
d
i
¤
e
r
e
n
c
e
b
e
t
w
e
e
n
a
n
n
o
u
n
c
e
d
a
n
d
a
n
a
l
y
s
t
f
o
r
e
-
c
a
s
t
e
d
e
a
r
n
i
n
g
s
-
p
e
r
-
s
h
a
r
e
(
E
S
)
.
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
i
n
p
a
r
e
n
t
h
e
s
e
s
.
S
t
a
n
d
a
r
d
e
r
r
o
r
s
a
r
e
c
l
u
s
t
e
r
e
d
a
t
t
h
e
…
r
m
-
l
e
v
e
l
t
o
t
h
e
r
i
g
h
t
o
f
c
o
e
¢
c
i
e
n
t
s
.
T
h
e
r
e
g
r
e
s
s
i
o
n
s
u
s
e
…
r
m
s
t
h
a
t
a
n
n
o
u
n
c
e
e
a
r
n
i
n
g
s
e
i
t
h
e
r
o
u
t
s
i
d
e
o
f
o
n
e
p
e
n
n
y
o
r
j
u
s
t
g
r
e
a
t
e
r
t
h
a
n
o
n
e
p
e
n
n
y
o
f
m
e
d
i
a
n
a
n
a
l
y
s
t
e
s
t
i
m
a
t
e
s
f
r
o
m
1
9
9
0
t
o
2
0
0
7
.
I
n
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
i
n
c
l
u
d
e
…
r
m
-
l
e
v
e
l
m
e
a
s
u
r
e
s
o
f
o
w
n
e
r
s
h
i
p
s
t
a
b
i
l
i
t
y
,
m
a
r
k
e
t
-
t
o
-
b
o
o
k
r
a
t
i
o
,
d
e
b
t
,
s
i
z
e
,
a
n
n
u
a
l
r
e
t
u
r
n
,
s
t
a
n
d
a
r
d
d
e
v
i
a
t
i
o
n
o
f
a
n
a
l
y
s
t
f
o
r
e
c
a
s
t
s
,
n
u
m
b
e
r
o
f
a
n
a
l
y
s
t
s
c
o
v
e
r
i
n
g
t
h
e
…
r
m
,
y
e
a
r
…
x
e
d
-
e
¤
e
c
t
s
a
n
d
i
n
d
u
s
t
r
y
…
x
e
d
-
e
¤
e
c
t
s
b
a
s
e
d
o
n
t
h
e
F
a
m
a
-
F
r
e
n
c
h
1
2
I
n
d
u
s
t
r
y
C
l
a
s
s
i
…
c
a
t
i
o
n
.
O
w
n
e
r
s
h
i
p
m
e
a
s
u
r
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
4
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
a
n
d
a
l
l
o
t
h
e
r
e
x
p
l
a
n
a
t
o
r
y
v
a
r
i
a
b
l
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
3
.
S
i
g
n
i
…
c
a
n
c
e
a
t
t
h
e
1
%
l
e
v
e
l
i
s
d
e
s
i
g
n
a
t
e
d
w
i
t
h
a
,
t
h
e
5
%
l
e
v
e
l
w
i
t
h
b
,
a
n
d
t
h
e
1
0
%
l
e
v
e
l
w
i
t
h
c
.
117
Table 1.8: Linear Regressions Describing Changes in Analyst Forecasts
(1) (2) (3) (4)
SIH
t
-0.032
a
(5.16) -0.032
a
(5.08)
AROL
t
0.016
b
(2.19) 0.018
b
(2.33)
SIH
t
ES
t
0.423
a
(3.16)
AROL
t
ES
t
-1.133 (1.57)
Own%S
t
0.246
a
(5.39) 0.223
a
(4.24)
Own%M
t
0.040 (1.28) 0.035 (1.01)
Own%L
t
-0.061
b
(2.25) -0.067
b
(2.40)
AROLS
t
-0.004 (0.87) -0.004 (0.80)
AROLM
t
0.003 (0.71) 0.003 (0.54)
AROLL
t
0.015
a
(3.02) 0.015
a
(2.93)
Own%S
t
ES
t
5.614 (0.94)
Own%M
t
ES
t
0.061 (0.01)
Own%L
t
ES
t
3.576 (1.44)
AROLS
t
ES
t
-0.236 (0.45)
AROLM
t
ES
t
0.743 (1.44)
AROLL
t
ES
t
0.049 (0.11)
MB
t
0.001
a
(5.37) 0.001
a
(5.50) 0.001
a
(5.29) 0.001
a
(5.40)
Debt
t
-0.035
a
(4.56) -0.034
a
(4.32) -0.034
a
(4.41) -0.033
a
(4.23)
Size
t
0.000 (0.23) 0.000 (0.42) 0.000 (0.17) 0.000 (0.33)
3MRet
t
0.042
a
(6.31) 0.042
a
(6.26) 0.040
a
(6.04) 0.041
a
(6.05)
Constant 0.039 (1.63) -0.065
a
(3.10) 0.035 (1.43) -0.066
a
(3.08)
Adjusted R
2
0.03 0.03 0.04 0.03
Obs. 7675 7675 7675 7675
This table reports estimates from linear regressions explaining changes in median analyst
forecasts from the …rst month to the last month in the …nal quarter of the …scal year. One
panel regression is estimated for each model using …rms that announce earnings within one
penny of median analyst estimates between 1990 and 2007. Standard errors are clustered
at the …rm-level. t-statistics are in parentheses. Independent variables include measures
of ownership stability, market-to-book ratio, debt, size, …scal year stock return, year-…xed
e¤ects, industry …xed-e¤ects based on the Fama-French 48 Industry Classi…cation, and
a constant. The dependent variable is de…ned in Chapter 1.5. Ownership measures are
de…ned in Chapter 1.4. All other explanatory variables are de…ned in Chapter 1.3. Signif-
icance at the 1% level is designated with a, the 5% level with b, and the 10% level with
c.
118
T
a
b
l
e
1
.
9
:
L
e
v
e
l
R
e
g
r
e
s
s
i
o
n
s
D
e
s
c
r
i
b
i
n
g
D
i
s
c
r
e
t
i
o
n
a
r
y
A
c
c
r
u
a
l
s
-
S
h
a
r
e
h
o
l
d
e
r
C
o
m
p
o
s
i
t
i
o
n
D
A
t
U
n
s
D
A
t
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)
S
I
H
t
-
0
.
0
0
7
a
(
4
.
3
2
)
-
0
.
0
0
4
a
(
3
.
7
9
)
O
w
n
%
S
t
0
.
0
5
5
a
(
2
.
9
8
)
0
.
0
3
9
a
(
3
.
3
0
)
O
w
n
%
M
t
0
.
0
4
2
a
(
3
.
3
4
)
-
0
.
0
1
4
c
(
1
.
7
1
)
O
w
n
%
L
t
-
0
.
0
0
5
(
0
.
5
7
)
-
0
.
0
1
1
c
(
1
.
9
1
)
T
o
t
O
w
n
%
t
0
.
0
2
2
a
(
3
.
6
7
)
-
0
.
0
0
3
(
0
.
6
4
)
M
B
t
-
0
.
0
0
1
a
(
4
.
7
6
)
-
0
.
0
0
1
a
(
4
.
8
3
)
-
0
.
0
0
1
a
(
4
.
6
0
)
0
.
0
0
1
a
(
6
.
6
9
)
0
.
0
0
1
a
(
6
.
6
4
)
0
.
0
0
1
a
(
6
.
9
2
)
D
e
b
t
t
0
.
0
1
5
a
(
4
.
8
1
)
0
.
0
1
5
a
(
4
.
6
9
)
0
.
0
1
5
a
(
4
.
7
6
)
-
0
.
0
0
6
a
(
3
.
0
2
)
-
0
.
0
0
6
a
(
2
.
9
1
)
-
0
.
0
0
6
a
(
2
.
9
1
)
S
i
z
e
t
0
.
0
0
0
(
0
.
9
2
)
0
.
0
0
1
(
1
.
2
6
)
0
.
0
0
0
(
1
.
1
5
)
-
0
.
0
0
3
a
(
1
0
.
8
6
)
-
0
.
0
0
3
a
(
1
0
.
7
7
)
-
0
.
0
0
3
a
(
1
0
.
8
7
)
F
Y
R
e
t
t
-
0
.
0
0
1
(
0
.
9
4
)
-
0
.
0
0
1
(
0
.
9
0
)
0
.
0
0
0
(
0
.
3
6
)
-
0
.
0
0
1
b
(
2
.
1
8
)
-
0
.
0
0
1
b
(
2
.
0
2
)
-
0
.
0
0
1
(
1
.
4
5
)
F
c
s
t
S
D
t
-
0
.
0
2
7
a
(
4
.
5
0
)
-
0
.
0
2
6
a
(
4
.
3
2
)
-
0
.
0
2
6
a
(
4
.
3
0
)
0
.
0
2
9
a
(
6
.
9
2
)
0
.
0
2
9
a
(
6
.
8
8
)
0
.
0
2
9
a
(
6
.
9
0
)
C
o
n
s
t
a
n
t
0
.
0
2
9
b
(
2
.
1
9
)
0
.
0
0
6
(
0
.
5
1
)
0
.
0
0
6
(
0
.
4
9
)
0
.
0
7
2
a
(
1
2
.
6
6
)
0
.
0
6
0
a
(
1
2
.
5
7
)
0
.
0
6
0
a
(
1
2
.
6
5
)
A
d
j
u
s
t
e
d
R
2
0
.
0
1
0
.
0
1
0
.
0
1
0
.
1
0
0
.
1
0
0
.
1
0
O
b
s
.
2
1
6
6
9
2
1
6
6
9
2
1
6
6
9
2
1
6
6
9
2
1
6
6
9
2
1
6
6
9
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
p
a
n
e
l
r
e
g
r
e
s
s
i
o
n
s
r
e
s
u
l
t
s
d
e
s
c
r
i
b
i
n
g
s
i
g
n
e
d
(
D
A
)
a
n
d
u
n
s
i
g
n
e
d
(
U
n
s
D
A
)
d
i
s
c
r
e
t
i
o
n
a
r
y
a
c
c
r
u
a
l
s
c
o
n
t
r
o
l
l
i
n
g
f
o
r
o
w
n
e
r
s
h
i
p
s
t
a
b
i
l
i
t
y
w
i
t
h
m
e
a
s
u
r
e
s
o
f
s
h
a
r
e
h
o
l
d
e
r
c
o
m
p
o
s
i
t
i
o
n
.
S
t
a
n
d
a
r
d
e
r
r
o
r
s
a
r
e
c
l
u
s
t
e
r
e
d
a
t
t
h
e
…
r
m
l
e
v
e
l
.
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
l
o
c
a
t
e
d
t
o
t
h
e
r
i
g
h
t
o
f
c
o
e
¢
c
i
e
n
t
e
s
t
i
m
a
t
e
s
i
n
p
a
r
e
n
t
h
e
s
e
s
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
i
s
e
q
u
a
l
t
o
e
i
t
h
e
r
s
i
g
n
e
d
o
r
a
b
s
o
l
u
t
e
d
i
s
c
r
e
t
i
o
n
a
r
y
a
c
c
r
u
a
l
s
c
a
l
c
u
l
a
t
e
d
u
s
i
n
g
a
m
o
d
i
…
e
d
J
o
n
e
s
(
1
9
9
1
)
m
o
d
e
l
.
I
n
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
i
n
c
l
u
d
e
…
r
m
-
l
e
v
e
l
m
e
a
s
u
r
e
s
o
f
f
u
n
d
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
c
o
m
p
o
s
i
t
i
o
n
,
m
a
r
k
e
t
-
t
o
-
b
o
o
k
,
s
t
a
n
d
a
r
d
d
e
v
i
a
t
i
o
n
o
f
a
n
a
l
y
s
t
f
o
r
e
c
a
s
t
s
,
d
e
b
t
,
s
i
z
e
,
a
n
n
u
a
l
r
e
t
u
r
n
,
y
e
a
r
…
x
e
d
-
e
¤
e
c
t
s
,
a
n
d
i
n
d
u
s
t
r
y
…
x
e
d
-
e
¤
e
c
t
s
b
a
s
e
d
o
n
t
h
e
F
a
m
a
-
F
r
e
n
c
h
4
8
I
n
d
u
s
t
r
y
C
l
a
s
s
i
…
c
a
t
i
o
n
.
O
w
n
e
r
s
h
i
p
m
e
a
s
u
r
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
4
.
B
o
t
h
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
a
n
d
a
l
l
o
t
h
e
r
e
x
p
l
a
n
a
t
o
r
y
v
a
r
i
a
b
l
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
3
.
S
i
g
n
i
…
c
a
n
c
e
a
t
t
h
e
1
%
l
e
v
e
l
i
s
d
e
n
o
t
e
d
w
i
t
h
a
,
t
h
e
5
%
l
e
v
e
l
w
i
t
h
b
,
a
n
d
t
h
e
1
0
%
l
e
v
e
l
w
i
t
h
c
.
119
Table 1.10: Level Regressions Describing Discretionary Accruals - Own-
ership Length
DA
t
UnsDA
t
(1) (2) (3) (4)
AROL
t
0.002 (0.85) -0.009
a
(6.20)
AROLS
t
0.003
b
(2.38) 0.000 (0.51)
AROLM
t
0.004
a
(2.75) -0.004
a
(4.32)
AROLL
t
-0.001 (0.29) -0.006
a
(4.85)
MB
t
-0.001
a
(4.58) -0.001
a
(4.76) 0.001
a
(7.05) 0.001
a
(7.04)
Debt
t
0.015
a
(4.91) 0.016
a
(4.97) -0.006
a
(3.06) -0.006
a
(3.12)
Size
t
0.000 (0.87) 0.000 (0.52) -0.003
a
(10.55) -0.003
a
(10.17)
FYRet
t
0.000 (0.08) 0.000 (0.19) -0.001
b
(2.34) -0.001
b
(2.23)
FcstSD
t
-0.026
a
(4.43) -0.025
a
(4.21) 0.028
a
(6.54) 0.027
a
(6.38)
Constant 0.006 (0.53) 0.006 (0.46) 0.064
a
(13.62) 0.064
a
(13.34)
Adjusted R
2
0.01 0.01 0.10 0.10
Obs. 21669 21669 21669 21669
This table reports panel regressions results describing signed (DA) and unsigned
(UnsDA) discretionary accruals controlling for ownership stability with measures of
ownership length. Standard errors are clustered at the …rm level. t-statistics are
located to the right of coe¢cient estimates in parentheses. The dependent variable is
equal to either signed or absolute discretionary accruals calculated using a modi…ed
Jones (1991) model. Independent variables include …rm-level measures of fund invest-
ment horizon composition, market-to-book, standard deviation of analyst forecasts,
debt, size, annual return, year …xed-e¤ects, and industry …xed-e¤ects based on the
Fama-French 48 Industry Classi…cation. Ownership measures are de…ned in Chap-
ter 1.4. Both dependent variables and all other explanatory variables are de…ned in
Chapter 1.3. Signi…cance at the 1% level is denoted with a, the 5% level with b, and
the 10% level with c.
120
T
a
b
l
e
1
.
1
1
:
L
e
v
e
l
R
e
g
r
e
s
s
i
o
n
s
D
e
s
c
r
i
b
i
n
g
D
i
s
c
r
e
t
i
o
n
a
r
y
A
c
c
r
u
a
l
s
-
S
h
a
r
e
h
o
l
d
e
r
C
o
m
p
o
s
i
t
i
o
n
&
O
w
n
e
r
s
h
i
p
L
e
n
g
t
h
D
A
t
U
n
s
D
A
t
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)
S
I
H
t
-
0
.
0
0
7
a
(
4
.
5
0
)
-
0
.
0
0
3
a
(
3
.
0
2
)
O
w
n
%
S
t
0
.
0
5
2
a
(
2
.
7
8
)
0
.
1
0
9
a
(
2
.
7
0
)
0
.
0
3
5
a
(
2
.
9
8
)
0
.
0
3
5
(
1
.
4
1
)
O
w
n
%
M
t
0
.
0
3
5
a
(
2
.
6
9
)
0
.
0
6
6
b
(
2
.
2
1
)
-
0
.
0
0
6
(
0
.
7
0
)
-
0
.
0
0
6
(
0
.
3
0
)
O
w
n
%
L
t
-
0
.
0
0
6
(
0
.
7
1
)
-
0
.
0
1
9
(
0
.
8
2
)
-
0
.
0
0
5
(
0
.
9
2
)
0
.
0
1
4
(
0
.
9
2
)
A
R
O
L
t
0
.
0
0
3
(
1
.
4
7
)
-
0
.
0
0
8
a
(
5
.
7
4
)
A
R
O
L
S
t
0
.
0
0
2
(
1
.
4
4
)
0
.
0
0
4
b
(
2
.
2
2
)
-
0
.
0
0
1
(
1
.
2
5
)
-
0
.
0
0
1
(
1
.
0
2
)
A
R
O
L
M
t
0
.
0
0
3
b
(
2
.
0
4
)
0
.
0
0
5
b
(
2
.
4
5
)
-
0
.
0
0
4
a
(
3
.
9
5
)
-
0
.
0
0
4
a
(
3
.
1
3
)
A
R
O
L
L
t
0
.
0
0
0
(
0
.
2
0
)
0
.
0
0
0
(
0
.
1
0
)
-
0
.
0
0
5
a
(
4
.
4
6
)
-
0
.
0
0
4
b
(
2
.
5
4
)
O
w
n
%
S
t

A
R
O
L
S
t
-
0
.
1
2
0
c
(
1
.
6
6
)
0
.
0
0
0
(
0
.
0
0
)
O
w
n
%
M
t

A
R
O
L
M
t
-
0
.
0
6
2
(
1
.
2
3
)
-
0
.
0
0
1
(
0
.
0
2
)
O
w
n
%
L
t

A
R
O
L
L
t
0
.
0
2
3
(
0
.
5
9
)
-
0
.
0
3
6
(
1
.
4
0
)
M
B
t
-
0
.
0
0
1
a
(
4
.
8
1
)
-
0
.
0
0
1
a
(
4
.
9
5
)
-
0
.
0
0
1
a
(
4
.
9
1
)
0
.
0
0
1
a
(
6
.
8
5
)
0
.
0
0
1
a
(
6
.
8
3
)
0
.
0
0
1
a
(
6
.
8
4
)
D
e
b
t
t
0
.
0
1
5
a
(
4
.
8
3
)
0
.
0
1
5
a
(
4
.
7
8
)
0
.
0
1
5
a
(
4
.
8
1
)
-
0
.
0
0
6
a
(
3
.
1
2
)
-
0
.
0
0
6
a
(
3
.
1
6
)
-
0
.
0
0
6
a
(
3
.
1
4
)
S
i
z
e
t
0
.
0
0
0
(
0
.
8
5
)
0
.
0
0
0
(
0
.
9
3
)
0
.
0
0
0
(
0
.
8
4
)
-
0
.
0
0
3
a
(
1
0
.
5
8
)
-
0
.
0
0
3
a
(
1
0
.
0
1
)
-
0
.
0
0
3
a
(
1
0
.
0
3
)
F
Y
R
e
t
t
-
0
.
0
0
1
(
0
.
7
8
)
0
.
0
0
0
(
0
.
5
6
)
-
0
.
0
0
1
(
0
.
8
2
)
-
0
.
0
0
1
a
(
2
.
8
2
)
-
0
.
0
0
1
a
(
2
.
7
1
)
-
0
.
0
0
1
a
(
2
.
6
2
)
F
c
s
t
S
D
t
-
0
.
0
2
6
a
(
4
.
3
9
)
-
0
.
0
2
4
a
(
4
.
1
3
)
-
0
.
0
2
4
a
(
4
.
1
1
)
0
.
0
2
8
a
(
6
.
5
5
)
0
.
0
2
7
a
(
6
.
3
8
)
0
.
0
2
7
a
(
6
.
4
1
)
C
o
n
s
t
a
n
t
0
.
0
2
8
b
(
2
.
1
4
)
0
.
0
0
5
(
0
.
4
0
)
0
.
0
0
4
(
0
.
3
1
)
0
.
0
7
4
a
(
1
3
.
0
5
)
0
.
0
6
4
a
(
1
3
.
2
6
)
0
.
0
6
3
a
(
1
3
.
0
6
)
A
d
j
u
s
t
e
d
R
2
0
.
0
1
0
.
0
1
0
.
0
1
0
.
1
0
0
.
1
0
0
.
1
0
O
b
s
.
2
1
6
6
9
2
1
6
6
9
2
1
6
6
9
2
1
6
6
9
2
1
6
6
9
2
1
6
6
9
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
p
a
n
e
l
r
e
g
r
e
s
s
i
o
n
s
r
e
s
u
l
t
s
d
e
s
c
r
i
b
i
n
g
s
i
g
n
e
d
(
D
A
)
a
n
d
u
n
s
i
g
n
e
d
(
U
n
s
D
A
)
d
i
s
c
r
e
t
i
o
n
a
r
y
a
c
c
r
u
a
l
s
c
o
n
t
r
o
l
l
i
n
g
f
o
r
o
w
n
e
r
s
h
i
p
s
t
a
b
i
l
i
t
y
w
i
t
h
m
e
a
s
u
r
e
s
o
f
s
h
a
r
e
h
o
l
d
e
r
c
o
m
p
o
s
i
t
i
o
n
a
n
d
o
w
n
e
r
s
h
i
p
l
e
n
g
t
h
.
S
t
a
n
d
a
r
d
e
r
r
o
r
s
a
r
e
c
l
u
s
t
e
r
e
d
a
t
t
h
e
…
r
m
l
e
v
e
l
.
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
l
o
c
a
t
e
d
t
o
t
h
e
r
i
g
h
t
o
f
c
o
e
¢
c
i
e
n
t
e
s
t
i
m
a
t
e
s
i
n
p
a
r
e
n
t
h
e
s
e
s
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
i
s
e
q
u
a
l
t
o
e
i
t
h
e
r
s
i
g
n
e
d
o
r
a
b
s
o
l
u
t
e
d
i
s
c
r
e
t
i
o
n
a
r
y
a
c
c
r
u
a
l
s
c
a
l
c
u
l
a
t
e
d
u
s
i
n
g
a
m
o
d
i
…
e
d
J
o
n
e
s
(
1
9
9
1
)
m
o
d
e
l
.
I
n
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
i
n
c
l
u
d
e
…
r
m
-
l
e
v
e
l
m
e
a
s
u
r
e
s
o
f
f
u
n
d
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
c
o
m
p
o
s
i
t
i
o
n
,
m
a
r
k
e
t
-
t
o
-
b
o
o
k
,
s
t
a
n
d
a
r
d
d
e
v
i
a
t
i
o
n
o
f
a
n
a
l
y
s
t
f
o
r
e
c
a
s
t
s
,
d
e
b
t
,
s
i
z
e
,
a
n
n
u
a
l
r
e
t
u
r
n
,
y
e
a
r
…
x
e
d
-
e
¤
e
c
t
s
,
a
n
d
i
n
d
u
s
t
r
y
…
x
e
d
-
e
¤
e
c
t
s
b
a
s
e
d
o
n
t
h
e
F
a
m
a
-
F
r
e
n
c
h
4
8
I
n
d
u
s
t
r
y
C
l
a
s
s
i
…
c
a
t
i
o
n
.
O
w
n
e
r
s
h
i
p
m
e
a
s
u
r
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
4
.
B
o
t
h
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
a
n
d
a
l
l
o
t
h
e
r
e
x
p
l
a
n
a
t
o
r
y
v
a
r
i
a
b
l
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
3
.
S
i
g
n
i
…
c
a
n
c
e
a
t
t
h
e
1
%
l
e
v
e
l
i
s
d
e
n
o
t
e
d
w
i
t
h
a
,
t
h
e
5
%
l
e
v
e
l
w
i
t
h
b
,
a
n
d
t
h
e
1
0
%
l
e
v
e
l
w
i
t
h
c
.
121
T
a
b
l
e
1
.
1
2
:
D
i
¤
e
r
e
n
c
e
R
e
g
r
e
s
s
i
o
n
s
D
e
s
c
r
i
b
i
n
g
C
h
a
n
g
e
s
i
n
D
i
s
c
r
e
t
i
o
n
a
r
y
A
c
c
r
u
a
l
s

D
A
t
+
1

U
n
s
D
A
t
+
1
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)
(
7
)
(
8
)

S
I
H
t
+
1
-
0
.
0
0
8
a
(
3
.
7
5
)
0
.
0
0
0
(
0
.
3
1
)

A
R
O
L
t
+
1
0
.
0
0
3
(
0
.
8
4
)
-
0
.
0
0
2
(
1
.
0
7
)

O
w
n
%
S
t
+
1
0
.
0
6
9
a
(
2
.
7
3
)
0
.
0
6
3
b
(
2
.
4
9
)
0
.
0
0
8
(
0
.
5
1
)
0
.
0
0
5
(
0
.
3
1
)

O
w
n
%
M
t
+
1
0
.
0
3
9
b
(
2
.
2
6
)
0
.
0
3
4
c
(
1
.
9
3
)
-
0
.
0
2
1
c
(
1
.
8
8
)
-
0
.
0
1
9
c
(
1
.
6
5
)

O
w
n
%
L
t
+
1
-
0
.
0
1
2
(
0
.
8
7
)
-
0
.
0
1
2
(
0
.
8
6
)
-
0
.
0
0
1
(
0
.
0
9
)
0
.
0
0
0
(
0
.
0
5
)

A
R
O
L
S
t
+
1
0
.
0
0
4
b
(
1
.
9
8
)
0
.
0
0
3
(
1
.
3
9
)
0
.
0
0
1
(
0
.
8
8
)
0
.
0
0
1
(
0
.
8
2
)

A
R
O
L
M
t
+
1
0
.
0
0
3
(
1
.
6
0
)
0
.
0
0
3
(
1
.
2
7
)
-
0
.
0
0
2
(
1
.
6
0
)
-
0
.
0
0
2
(
1
.
2
3
)

A
R
O
L
L
t
+
1
0
.
0
0
1
(
0
.
3
0
)
0
.
0
0
1
(
0
.
4
4
)
-
0
.
0
0
2
(
1
.
4
8
)
-
0
.
0
0
3
(
1
.
5
5
)

M
B
t
+
1
0
.
0
0
1
b
(
2
.
0
2
)
0
.
0
0
1
c
(
1
.
8
3
)
0
.
0
0
1
b
(
2
.
1
4
)
0
.
0
0
1
c
(
1
.
7
0
)
0
.
0
0
1
a
(
3
.
2
2
)
0
.
0
0
1
a
(
3
.
1
7
)
0
.
0
0
1
a
(
3
.
2
3
)
0
.
0
0
1
a
(
3
.
2
4
)

D
e
b
t
t
+
1
0
.
0
0
2
(
0
.
4
6
)
0
.
0
0
2
(
0
.
5
6
)
0
.
0
0
1
(
0
.
3
5
)
0
.
0
0
2
(
0
.
6
1
)
0
.
0
0
4
c
(
1
.
9
1
)
0
.
0
0
4
c
(
1
.
8
8
)
0
.
0
0
5
b
(
1
.
9
6
)
0
.
0
0
4
c
(
1
.
9
2
)

S
i
z
e
t
+
1
0
.
0
1
9
a
(
7
.
9
2
)
0
.
0
1
9
a
(
7
.
5
6
)
0
.
0
2
0
a
(
8
.
0
5
)
0
.
0
1
9
a
(
7
.
4
9
)
-
0
.
0
0
5
a
(
3
.
0
9
)
-
0
.
0
0
5
a
(
2
.
9
3
)
-
0
.
0
0
5
a
(
3
.
1
6
)
-
0
.
0
0
5
a
(
2
.
9
8
)

F
Y
R
e
t
t
+
1
-
0
.
0
0
4
a
(
3
.
3
7
)
-
0
.
0
0
4
a
(
3
.
4
1
)
-
0
.
0
0
3
a
(
2
.
7
7
)
-
0
.
0
0
4
a
(
3
.
1
6
)
-
0
.
0
0
1
b
(
2
.
0
3
)
-
0
.
0
0
1
b
(
1
.
9
9
)
-
0
.
0
0
1
b
(
2
.
0
9
)
-
0
.
0
0
1
b
(
2
.
0
5
)

F
c
s
t
S
D
t
+
1
-
0
.
0
3
7
a
(
3
.
5
8
)
-
0
.
0
3
6
a
(
3
.
5
1
)
-
0
.
0
3
7
a
(
3
.
5
5
)
-
0
.
0
3
6
a
(
3
.
4
7
)
0
.
0
3
4
a
(
4
.
8
8
)
0
.
0
3
4
a
(
4
.
8
8
)
0
.
0
3
4
a
(
4
.
8
6
)
0
.
0
3
4
a
(
4
.
8
3
)
C
o
n
s
t
a
n
t
-
0
.
0
1
0
a
(
3
.
5
9
)
-
0
.
0
1
0
a
(
3
.
5
5
)
-
0
.
0
1
0
a
(
3
.
7
4
)
-
0
.
0
1
0
a
(
3
.
6
6
)
-
0
.
0
0
1
(
0
.
7
6
)
-
0
.
0
0
1
(
0
.
7
5
)
-
0
.
0
0
1
(
0
.
7
4
)
-
0
.
0
0
1
(
0
.
7
1
)
A
d
j
u
s
t
e
d
R
2
0
.
0
2
0
.
0
2
0
.
0
2
0
.
0
2
0
.
0
2
0
.
0
2
0
.
0
2
0
.
0
2
O
b
s
.
1
4
9
2
7
1
4
9
2
7
1
4
9
2
7
1
4
9
2
7
1
4
9
2
7
1
4
9
2
7
1
4
9
2
7
1
4
9
2
7
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
p
a
n
e
l
r
e
g
r
e
s
s
i
o
n
s
r
e
s
u
l
t
s
d
e
s
c
r
i
b
i
n
g
c
h
a
n
g
e
s
i
n
s
i
g
n
e
d
(
D
A
)
a
n
d
u
n
s
i
g
n
e
d
(
U
n
s
D
A
)
d
i
s
c
r
e
t
i
o
n
a
r
y
a
c
c
r
u
a
l
s
.
S
t
a
n
d
a
r
d
e
r
r
o
r
s
a
r
e
c
l
u
s
t
e
r
e
d
a
t
t
h
e
…
r
m
l
e
v
e
l
.
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
l
o
c
a
t
e
d
t
o
t
h
e
r
i
g
h
t
o
f
c
o
e
¢
c
i
e
n
t
e
s
t
i
m
a
t
e
s
i
n
p
a
r
e
n
t
h
e
s
e
s
.
T
h
e
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
i
s
e
q
u
a
l
t
o
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
e
i
t
h
e
r
s
i
g
n
e
d
o
r
u
n
s
i
g
n
e
d
d
i
s
c
r
e
t
i
o
n
a
r
y
a
c
c
r
u
a
l
s
f
r
o
m
y
e
a
r
t
÷
1
t
o
y
e
a
r
t
+
1
.
A
n
n
u
a
l
d
i
s
c
r
e
t
i
o
n
a
r
y
a
c
c
r
u
a
l
s
a
r
e
c
a
l
c
u
l
a
t
e
d
u
s
i
n
g
a
m
o
d
i
…
e
d
J
o
n
e
s
(
1
9
9
1
)
m
o
d
e
l
.
I
n
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
i
n
c
l
u
d
e
c
h
a
n
g
e
s
i
n
m
e
a
s
u
r
e
s
o
f
s
h
a
r
e
h
o
l
d
e
r
c
o
m
p
o
s
i
t
i
o
n
a
n
d
o
w
n
e
r
s
h
i
p
l
e
n
g
t
h
,
m
a
r
k
e
t
-
t
o
-
b
o
o
k
,
s
t
a
n
d
a
r
d
d
e
v
i
a
t
i
o
n
o
f
a
n
a
l
y
s
t
f
o
r
e
c
a
s
t
s
,
d
e
b
t
,
s
i
z
e
,
a
n
d
a
n
n
u
a
l
r
e
t
u
r
n
f
r
o
m
y
e
a
r
t
÷
1
t
o
y
e
a
r
t
+
1
.
I
a
l
s
o
i
n
c
l
u
d
e
y
e
a
r
…
x
e
d
-
e
¤
e
c
t
s
.
A
l
l
c
h
a
n
g
e
v
a
r
i
a
b
l
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
6
.
S
i
g
n
i
…
c
a
n
c
e
a
t
t
h
e
1
%
l
e
v
e
l
i
s
d
e
n
o
t
e
d
w
i
t
h
a
,
t
h
e
5
%
l
e
v
e
l
w
i
t
h
b
,
a
n
d
t
h
e
1
0
%
l
e
v
e
l
w
i
t
h
c
.
122
Table 2.1: Spin-o¤ Events By Announcement Year
Year N
1990 13
1991 7
1992 15
1993 17
1994 17
1995 26
1996 26
1997 19
1998 13
1999 20
2000 16
2001 7
2002 7
2003 13
2004 5
2005 7
2006 6
2007 12
Total 246
This table reports the number of spin-o¤ observations by announce-
ment year. Description of the methodology used to create dataset
is presented in Chapter 2.3.
123
T
a
b
l
e
2
.
2
:
P
e
r
c
e
n
t
a
g
e
o
f
S
h
a
r
e
s
H
e
l
d
B
e
f
o
r
e
&
A
f
t
e
r
S
p
i
n
-
o
¤
E
v
e
n
t
A
l
l
L
o
n
g
M
e
d
i
u
m
S
h
o
r
t
M
o
n
t
h
s
N
M
e
a
n
M
e
d
i
a
n
M
e
a
n
M
e
d
i
a
n
M
e
a
n
M
e
d
i
a
n
M
e
a
n
M
e
d
i
a
n
B
e
f
o
r
e
S
p
i
n
-
o
¤
3
6
2
3
0
0
.
0
5
6
0
.
0
3
7
0
.
0
2
3
0
.
0
1
2
0
.
0
2
1
0
.
0
1
0
0
.
0
1
0
0
.
0
0
5
3
0
2
3
5
0
.
0
6
1
0
.
0
4
3
0
.
0
2
7
0
.
0
1
4
0
.
0
2
1
0
.
0
1
3
0
.
0
1
2
0
.
0
0
5
2
4
2
4
0
0
.
0
6
3
0
.
0
5
0
0
.
0
2
7
0
.
0
1
4
0
.
0
2
1
0
.
0
1
4
0
.
0
1
4
0
.
0
0
5
1
8
2
4
5
0
.
0
6
5
0
.
0
4
4
0
.
0
2
8
0
.
0
1
5
0
.
0
2
0
0
.
0
1
2
0
.
0
1
5
0
.
0
0
6
1
2
2
4
7
0
.
0
6
6
0
.
0
4
9
0
.
0
3
0
0
.
0
1
6
0
.
0
2
0
0
.
0
1
3
0
.
0
1
3
0
.
0
0
5
6
2
4
8
0
.
0
6
8
0
.
0
5
2
0
.
0
3
1
0
.
0
1
7
0
.
0
2
3
0
.
0
1
5
0
.
0
1
2
0
.
0
0
5
A
f
t
e
r
S
p
i
n
-
o
¤
O
v
e
r
a
l
l
6
2
3
4
0
.
0
7
7
0
.
0
6
3
0
.
0
3
7
0
.
0
2
2
0
.
0
2
3
0
.
0
1
6
0
.
0
1
4
0
.
0
0
9
1
2
2
3
5
0
.
0
8
0
0
.
0
6
4
0
.
0
3
9
0
.
0
2
4
0
.
0
2
4
0
.
0
1
8
0
.
0
1
6
0
.
0
1
0
1
8
2
1
1
0
.
0
8
3
0
.
0
6
8
0
.
0
4
0
0
.
0
2
9
0
.
0
2
5
0
.
0
1
7
0
.
0
1
7
0
.
0
1
1
2
4
1
9
3
0
.
0
8
1
0
.
0
6
7
0
.
0
3
8
0
.
0
2
8
0
.
0
2
6
0
.
0
1
7
0
.
0
1
4
0
.
0
0
8
3
0
1
7
3
0
.
0
8
3
0
.
0
7
3
0
.
0
4
1
0
.
0
3
2
0
.
0
2
4
0
.
0
1
8
0
.
0
1
5
0
.
0
0
8
3
6
1
5
8
0
.
0
8
6
0
.
0
8
5
0
.
0
4
3
0
.
0
3
4
0
.
0
2
6
0
.
0
1
9
0
.
0
1
4
0
.
0
1
0
P
a
r
e
n
t
C
o
.
6
2
4
8
0
.
0
7
4
0
.
0
6
1
0
.
0
3
6
0
.
0
2
1
0
.
0
2
3
0
.
0
1
5
0
.
0
1
3
0
.
0
0
6
1
2
2
4
2
0
.
0
7
5
0
.
0
5
3
0
.
0
3
7
0
.
0
2
1
0
.
0
2
2
0
.
0
1
4
0
.
0
1
5
0
.
0
0
8
1
8
2
2
6
0
.
0
7
7
0
.
0
6
0
0
.
0
3
7
0
.
0
2
4
0
.
0
2
3
0
.
0
1
3
0
.
0
1
5
0
.
0
0
7
2
4
2
1
3
0
.
0
7
8
0
.
0
6
2
0
.
0
3
6
0
.
0
2
4
0
.
0
2
5
0
.
0
1
3
0
.
0
1
4
0
.
0
0
6
3
0
2
0
2
0
.
0
7
9
0
.
0
6
9
0
.
0
3
9
0
.
0
2
9
0
.
0
2
3
0
.
0
1
6
0
.
0
1
4
0
.
0
0
6
3
6
1
9
4
0
.
0
8
3
0
.
0
6
9
0
.
0
4
3
0
.
0
3
1
0
.
0
2
5
0
.
0
1
5
0
.
0
1
2
0
.
0
0
6
T
h
i
s
t
a
b
l
e
p
r
e
s
e
n
t
s
t
h
e
m
e
a
n
a
n
d
m
e
d
i
a
n
f
u
n
d
o
w
n
e
r
s
h
i
p
p
e
r
c
e
n
t
a
g
e
3
y
e
a
r
s
p
r
i
o
r
t
o
t
h
e
s
p
i
n
-
o
¤
a
n
n
o
u
n
c
e
m
e
n
t
d
a
t
e
a
n
d
3
y
e
a
r
s
f
o
l
l
o
w
i
n
g
t
h
e
s
p
i
n
-
o
¤
e
¤
e
c
t
i
v
e
d
a
t
e
.
F
u
n
d
h
o
l
d
i
n
g
s
a
r
e
t
a
k
e
n
a
t
6
m
o
n
t
h
s
i
n
t
e
r
v
a
l
s
u
s
i
n
g
t
h
e
m
o
s
t
r
e
c
e
n
t
S
E
C
…
l
i
n
g
s
,
a
g
g
r
e
g
a
t
e
d
u
s
i
n
g
e
i
t
h
e
r
a
l
l
f
u
n
d
s
h
a
r
e
h
o
l
d
e
r
s
o
r
f
u
n
d
s
h
a
r
e
h
o
l
d
e
r
s
b
y
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
(
F
I
H
)
t
e
r
c
i
l
e
.
O
w
n
e
r
s
h
i
p
p
e
r
c
e
n
t
a
g
e
o
f
t
h
e
p
a
r
e
n
t
c
o
m
p
a
n
y
p
r
i
o
r
t
o
a
n
d
a
f
t
e
r
t
h
e
a
n
n
o
u
n
c
e
m
e
n
t
d
a
t
e
i
s
e
q
u
a
l
t
o
t
h
e
t
o
t
a
l
n
u
m
b
e
r
o
f
s
h
a
r
e
s
h
e
l
d
d
i
v
i
d
e
d
b
y
s
h
a
r
e
s
o
u
t
s
t
a
n
d
i
n
g
.
O
v
e
r
a
l
l
s
h
a
r
e
h
o
l
d
i
n
g
s
a
f
t
e
r
t
h
e
e
¤
e
c
t
i
v
e
d
a
t
e
i
s
e
q
u
a
l
t
o
t
h
e
m
a
r
k
e
t
-
v
a
l
u
e
w
e
i
g
h
t
e
d
a
v
e
r
a
g
e
o
f
t
h
e
p
e
r
c
e
n
t
a
g
e
o
f
s
h
a
r
e
s
h
e
l
d
o
f
a
l
l
…
r
m
s
s
t
e
m
m
i
n
g
f
r
o
m
t
h
e
s
a
m
e
p
a
r
e
n
t
c
o
m
p
a
n
y
.
124
Table 2.3: Fund Ownership Changes
Panel A: Overall Fund Ownership - Unadjusted
Months N Mean t-stat Median z-stat
Total Ownership Percentage
12 218 0.013
a
(3.59) 0.009
a
(4.26)
24 181 0.017
a
(3.91) 0.019
a
(4.24)
36 155 0.028
a
(6.02) 0.023
a
(5.66)
Shareholder Investment Horizon
12 166 0.044 (1.54) 0.049 (1.49)
24 134 0.100
a
(3.15) 0.090
a
(2.91)
36 111 0.082
a
(2.70) 0.054
a
(2.63)
Panel B: Parent Company Fund Ownership - Unadjusted
Total Ownership Percentage
12 225 0.010
b
(2.56) 0.005
a
(3.06)
24 199 0.014
a
(3.11) 0.011
a
(3.67)
36 188 0.026
a
(5.37) 0.015
a
(5.39)
Shareholder Investment Horizon
12 162 0.017 (0.66) 0.038 (0.90)
24 141 0.059
b
(2.14) 0.074
b
(2.10)
36 131 0.097
a
(3.31) 0.056
a
(3.06)
Continued Next Page...
125
Panel C: Overall Fund Ownership - Adjusted
Months N Mean t-stat Median z-stat
Total Ownership Percentage
12 188 0.001 (0.20) 0.007 (0.32)
24 158 0.001 (0.08) 0.005 (0.31)
36 136 0.004 (0.57) 0.009 (0.76)
Shareholder Investment Horizon
12 145 0.005 (0.11) -0.040 (0.51)
24 118 0.035 (0.92) 0.046 (0.75)
36 98 0.004 (0.12) 0.042 (0.20)
Panel D: Parent Company Fund Ownership - Adjusted
Total Ownership Percentage
12 195 0.001 (0.11) 0.001 (0.09)
24 174 0.001 (0.13) 0.005 (0.57)
36 165 0.006 (0.87) 0.008 (0.85)
Shareholder Investment Horizon
12 144 -0.032 (0.84) -0.054 (0.84)
24 126 0.002 (0.05) 0.016 (0.11)
36 116 0.008 (0.21) 0.046 (0.25)
This table presents the mean and median unadjusted changes in
fund ownership from 6 months before the announcement date to
12 months, 24 months, and 36 months following the e¤ective date.
Ownership variables are equal to overall changes in total owner-
ship percentage, and shareholder investment horizon (SIH). Over-
all shareholdings after the e¤ective date is equal to the market-value
weighted average of the percentage of shares held of all …rms stem-
ming from the same parent company. Ownership percentage of the
parent company prior to and after the announcement date is equal
to the total number of shares held divided by shares outstanding.
Ownership variables are de…ned in Chapter 2.4. Adjusted own-
ership changes are equal to ownership changes in the event …rm
minus ownership changes in the match …rms. The algorithm to
choose match …rms is detailed in Chapter 2.4.2. For the di¤erence
in means, I use the two-tailed t-statistic to test sign…cance. For
the di¤erence in medians I use the two-tailed z-statistic from the
Wilcoxon rank-sum test. Signi…cance at the 1% level is denoted
with a, the 5% level with b, and the 10% level with c.
126
T
a
b
l
e
2
.
4
:
P
a
n
e
l
R
e
g
r
e
s
s
i
o
n
s
D
e
s
c
r
i
b
i
n
g
C
h
a
n
g
e
s
i
n
A
d
j
u
s
t
e
d
O
w
n
e
r
s
h
i
p
P
a
n
e
l
A
:
O
v
e
r
a
l
l
F
u
n
d
O
w
n
e
r
s
h
i
p
C
h
a
n
g
e
s
T
o
t
a
l
O
w
n
e
r
s
h
i
p
P
e
r
c
e
n
t
a
g
e
S
h
a
r
e
h
o
l
d
e
r
I
n
v
e
s
t
m
e
n
t
H
o
r
i
z
o
n
1
2
M
o
n
t
h
s
2
4
M
o
n
t
h
s
3
6
M
o
n
t
h
s
1
2
M
o
n
t
h
s
2
4
M
o
n
t
h
s
3
6
M
o
n
t
h
s
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)

A
n
N
u
m
0
.
0
5
7
(
0
.
5
2
)
0
.
2
4
6
(
1
.
5
0
)
0
.
2
0
1
(
1
.
3
6
)
0
.
7
0
2
(
0
.
7
7
)
0
.
3
0
8
(
0
.
3
7
)
0
.
1
7
6
(
0
.
2
5
)
I
C
D
-
0
.
0
0
2
(
0
.
2
7
)
-
0
.
0
1
9
c
(
1
.
8
1
)
-
0
.
0
2
1
c
(
1
.
8
5
)
0
.
0
6
3
(
0
.
8
8
)
0
.
1
1
7
c
(
1
.
9
2
)
0
.
0
9
6
(
1
.
3
5
)
I
C
D

S
D
S
i
z
e
0
.
0
0
4
(
0
.
8
5
)
0
.
0
0
3
(
0
.
4
5
)
0
.
0
1
2
(
1
.
5
4
)
-
0
.
0
0
6
(
0
.
1
3
)
-
0
.
0
8
3
c
(
1
.
8
3
)
-
0
.
0
7
3
(
1
.
5
8
)
S
D
M
B
0
.
0
0
2
(
1
.
0
2
)
-
0
.
0
0
1
(
0
.
1
8
)
0
.
0
0
1
c
(
1
.
7
8
)
-
0
.
0
0
4
(
0
.
1
9
)
0
.
0
1
7
(
0
.
9
9
)
-
0
.
0
0
1
(
0
.
0
6
)
S
D
S
i
z
e
-
0
.
0
0
4
(
0
.
4
1
)
0
.
0
0
5
(
0
.
3
1
)
-
0
.
0
1
5
(
0
.
9
3
)
0
.
0
0
4
(
0
.
0
4
)
-
0
.
0
0
2
(
0
.
0
2
)
0
.
0
7
9
(
0
.
9
9
)
S
D
R
O
A
0
.
1
1
9
b
(
2
.
1
8
)
0
.
0
4
3
(
0
.
4
0
)
0
.
1
2
4
b
(
2
.
2
4
)
0
.
0
9
3
(
0
.
2
0
)
-
0
.
7
2
4
(
1
.
2
7
)
-
0
.
5
8
3
(
0
.
8
0
)

R
O
A
0
.
0
9
4
c
(
1
.
8
7
)
0
.
0
2
4
(
0
.
2
2
)
0
.
1
8
3
b
(
2
.
6
2
)
0
.
3
3
9
(
0
.
9
3
)
-
0
.
8
2
5
(
1
.
3
9
)
-
0
.
2
0
5
(
0
.
2
3
)
O
b
s
.
1
7
5
1
4
1
1
2
1
1
1
9
9
7
8
0
A
d
j
.
R
2
0
.
1
0
0
.
1
5
0
.
2
4
0
.
1
1
0
.
2
1
0
.
2
0
C
o
n
t
i
n
u
e
d
N
e
x
t
P
a
g
e
.
.
.
127
P
a
n
e
l
B
:
P
a
r
e
n
t
C
o
m
p
a
n
y
F
u
n
d
O
w
n
e
r
s
h
i
p
C
h
a
n
g
e
s
T
o
t
a
l
O
w
n
e
r
s
h
i
p
P
e
r
c
e
n
t
a
g
e
S
h
a
r
e
h
o
l
d
e
r
I
n
v
e
s
t
m
e
n
t
H
o
r
i
z
o
n
1
2
M
o
n
t
h
s
2
4
M
o
n
t
h
s
3
6
M
o
n
t
h
s
1
2
M
o
n
t
h
s
2
4
M
o
n
t
h
s
3
6
M
o
n
t
h
s
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)

A
n
N
u
m
-
0
.
0
3
4
(
0
.
3
1
)
0
.
0
6
1
(
0
.
4
5
)
0
.
1
3
9
(
0
.
9
7
)
1
.
4
9
8
(
1
.
6
4
)
1
.
0
9
5
(
1
.
3
9
)
0
.
6
9
1
(
1
.
0
3
)
I
C
D
0
.
0
0
1
(
0
.
2
1
)
-
0
.
0
1
4
a
(
2
.
6
1
)
-
0
.
0
1
1
b
(
2
.
1
1
)
0
.
0
2
8
(
0
.
8
7
)
0
.
0
2
9
(
0
.
9
3
)
0
.
0
4
7
(
1
.
2
7
)

R
O
A
0
.
0
4
3
(
1
.
1
2
)
0
.
0
6
9
(
1
.
0
2
)
0
.
0
6
8
(
1
.
0
8
)
-
0
.
4
6
0
(
0
.
9
8
)
-
0
.
0
0
1
(
0
.
0
0
)
0
.
2
6
0
(
0
.
4
7
)

S
i
z
e
0
.
0
0
0
(
0
.
0
4
)
0
.
0
0
6
(
0
.
7
1
)
0
.
0
0
0
(
0
.
0
4
)
0
.
0
5
1
(
0
.
4
9
)
0
.
0
6
7
(
0
.
9
1
)
0
.
0
4
9
(
0
.
8
3
)

M
B
0
.
0
0
3
c
(
1
.
7
1
)
0
.
0
0
1
(
1
.
1
6
)
0
.
0
0
1
a
(
4
.
8
8
)
-
0
.
0
0
9
(
0
.
7
7
)
0
.
0
1
1
(
0
.
8
1
)
0
.
0
0
5
a
(
3
.
4
2
)

C
a
p
E
x
0
.
0
7
6
(
1
.
0
1
)
0
.
2
8
2
(
1
.
5
2
)
0
.
1
2
7
(
0
.
8
4
)
0
.
1
3
9
(
0
.
1
5
)
0
.
2
1
8
(
0
.
1
5
)
-
0
.
9
6
7
(
0
.
6
3
)

D
e
b
t
-
0
.
1
3
1
c
(
1
.
8
8
)
-
0
.
0
6
0
(
1
.
0
0
)
-
0
.
0
9
6
b
(
2
.
0
1
)
-
0
.
0
1
7
(
0
.
0
4
)
0
.
1
7
1
(
0
.
4
6
)
-
0
.
1
9
7
(
0
.
5
2
)

D
i
v
Y
l
d
0
.
0
6
0
(
0
.
4
0
)
0
.
7
5
2
a
(
4
.
3
7
)
0
.
6
2
7
(
1
.
4
6
)
8
.
5
9
8
c
(
1
.
7
1
)
-
0
.
7
1
9
(
0
.
5
3
)
4
.
3
1
7
(
1
.
0
9
)

R
e
p
Y
l
d
-
0
.
0
4
3
(
0
.
3
5
)
-
0
.
0
2
1
(
0
.
1
7
)
0
.
1
9
8
(
1
.
5
0
)
0
.
4
0
8
(
0
.
5
6
)
-
0
.
1
6
1
(
0
.
1
7
)
-
1
.
3
2
0
(
1
.
4
2
)
O
b
s
.
1
9
8
1
7
5
1
6
5
1
2
8
1
1
3
1
0
2
A
d
j
.
R
2
0
.
1
1
0
.
1
7
0
.
2
5
0
.
1
6
0
.
1
4
0
.
2
4
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
e
s
t
i
m
a
t
e
s
f
r
o
m
l
i
n
e
a
r
r
e
g
r
e
s
s
i
o
n
s
e
x
p
l
a
i
n
i
n
g
a
d
j
u
s
t
e
d
o
v
e
r
a
l
l
a
n
d
p
a
r
e
n
t
c
o
m
p
a
n
y
f
u
n
d
o
w
n
e
r
s
h
i
p
c
h
a
n
g
e
s
1
2
,
2
4
a
n
d
3
6
m
o
n
t
h
s
a
f
t
e
r
t
h
e
e
¤
e
c
t
i
v
e
d
a
t
e
.
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
i
n
p
a
r
e
n
t
h
e
s
e
s
.
S
t
a
n
d
a
r
d
e
r
r
o
r
s
a
r
e
h
e
t
e
r
o
s
c
e
d
a
s
t
i
c
r
o
b
u
s
t
.
D
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
a
r
e
e
q
u
a
l
t
o
o
v
e
r
a
l
l
c
h
a
n
g
e
s
i
n
t
o
t
a
l
o
w
n
e
r
s
h
i
p
p
e
r
c
e
n
t
a
g
e
a
n
d
s
h
a
r
e
h
o
l
d
e
r
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
(
S
I
H
)
.
O
w
n
e
r
s
h
i
p
c
h
a
n
g
e
s
o
c
c
u
r
f
r
o
m
6
m
o
n
t
h
s
p
r
i
o
r
t
o
t
h
e
a
n
n
o
u
n
c
e
m
e
n
t
d
a
t
e
t
o
3
6
m
o
n
t
h
s
f
o
l
l
o
w
i
n
g
t
h
e
e
¤
e
c
t
i
v
e
d
a
t
e
.
O
v
e
r
a
l
l
m
e
a
s
u
r
e
s
o
f
f
u
n
d
o
w
n
e
r
s
h
i
p
u
s
e
a
l
l
…
r
m
s
s
t
e
m
m
i
n
g
f
r
o
m
t
h
e
s
a
m
e
p
a
r
e
n
t
c
o
m
p
a
n
y
f
o
l
l
o
w
i
n
g
t
h
e
e
¤
e
c
t
i
v
e
d
a
t
e
,
a
n
d
t
h
e
p
a
r
e
n
t
c
o
m
p
a
n
y
p
r
i
o
r
t
o
t
h
e
a
n
n
o
u
n
c
e
m
e
n
t
d
a
t
e
.
A
d
j
u
s
t
e
d
o
w
n
e
r
s
h
i
p
c
h
a
n
g
e
s
a
r
e
e
q
u
a
l
t
o
o
w
n
e
r
s
h
i
p
c
h
a
n
g
e
s
i
n
t
h
e
e
v
e
n
t
…
r
m
m
i
n
u
s
o
w
n
e
r
s
h
i
p
c
h
a
n
g
e
s
i
n
t
h
e
m
a
t
c
h
…
r
m
s
.
T
h
e
a
l
g
o
r
i
t
h
m
t
o
c
h
o
o
s
e
m
a
t
c
h
…
r
m
s
i
s
d
e
t
a
i
l
e
d
i
n
C
h
a
p
t
e
r
2
.
4
.
2
.
O
v
e
r
a
l
l
f
u
n
d
o
w
n
e
r
s
h
i
p
c
h
a
n
g
e
s
a
r
e
e
x
p
l
a
i
n
e
d
w
i
t
h
v
a
r
i
a
b
l
e
s
d
e
s
c
r
i
b
i
n
g
d
i
¤
e
r
e
n
c
e
s
b
e
t
w
e
e
n
t
h
e
p
a
r
e
n
t
c
o
m
p
a
n
y
a
n
d
d
i
s
t
r
i
b
u
t
e
d
s
u
b
s
i
d
i
a
r
i
e
s
.
P
a
r
e
n
t
c
o
m
p
a
n
y
f
u
n
d
o
w
n
e
r
s
h
i
p
c
h
a
n
g
e
s
a
r
e
e
x
p
l
a
i
n
e
d
w
i
t
h
p
a
r
e
n
t
c
o
m
p
a
n
y
s
p
e
c
i
…
c
v
a
r
i
a
b
l
e
s
a
s
w
e
l
l
a
s
o
v
e
r
a
l
l
m
e
a
s
u
r
e
s
o
f
s
p
i
n
-
o
¤
i
m
p
o
r
t
a
n
c
e
.
B
o
t
h
s
e
t
s
o
f
r
e
g
r
e
s
s
i
o
n
s
a
l
s
o
i
n
c
l
u
d
e
a
c
o
n
s
t
a
n
t
a
n
d
y
e
a
r
…
x
e
d
-
e
¤
e
c
t
s
.
A
l
l
v
a
r
i
a
b
l
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
s
2
.
3
a
n
d
2
.
4
.
S
i
g
n
i
…
c
a
n
c
e
a
t
t
h
e
1
%
l
e
v
e
l
i
s
d
e
s
i
g
n
a
t
e
d
w
i
t
h
a
,
t
h
e
5
%
l
e
v
e
l
w
i
t
h
b
,
a
n
d
t
h
e
1
0
%
l
e
v
e
l
w
i
t
h
c
.
128
Table 2.5: Tests of Mean and Median Abnormal Returns
Months N Mean t-stat Median z-stat
Panel A: Overall Fund Ownership
Ownership Percentage Match
12 183 0.181
a
(3.02) 0.027 (1.55)
24 151 0.340
a
(2.65) 0.081
c
(1.94)
36 125 0.488
a
(3.60) 0.162
b
(2.49)
Shareholder Investment Horizon Match
12 141 0.102
c
(1.78) -0.018 (0.64)
24 114 0.186
b
(1.98) 0.057 (1.19)
36 94 0.161 (1.31) 0.051 (0.62)
Panel B: Parent Company
Ownership Percentage Match
12 182 0.203
a
(3.29) 0.002 (1.48)
24 150 0.349
a
(2.60) 0.043 (1.08)
36 124 0.499
a
(3.54) 0.137
c
(1.77)
Shareholder Investment Horizon Match
12 137 0.107
c
(1.95) -0.005 (0.71)
24 111 0.134 (1.35) -0.010 (0.57)
36 90 0.103 (0.78) 0.040 (0.14)
This table presents tests of mean and median abnormal returns 12
months, 24 months, and 36 months following the e¤ective date. Ab-
normal returns are calculated both overall and the parent company,
using both ownership percentage and shareholder investment horizon
(SIH) match …rms as a benchmark. Abnormal returns are de…ned
in Chapter 2.4. The algorithm to choose match …rms is detailed
in Chapter 2.4.2. For the di¤erence in means, I use the two-tailed
skewness adjusted t-statistic to test sign…cance. For the di¤erence in
medians I use the two-tailed z-statistic from the Wilcoxon rank-sum
test. Signi…cance at the 1% level is denoted with a, the 5% level with
b, and the 10% level with c.
129
T
a
b
l
e
2
.
6
:
P
a
n
e
l
R
e
g
r
e
s
s
i
o
n
s
D
e
s
c
r
i
b
i
n
g
C
h
a
n
g
e
s
i
n
A
d
j
u
s
t
e
d
R
e
t
u
r
n
s
P
a
n
e
l
A
:
O
v
e
r
a
l
l
A
d
j
u
s
t
e
d
R
e
t
u
r
n
s
1
2
M
o
n
t
h
s
2
4
M
o
n
t
h
s
3
6
M
o
n
t
h
s
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)

T
o
t
O
P
0
.
4
9
8
(
0
.
6
0
)
-
0
.
3
3
7
(
0
.
1
8
)
-
2
.
7
2
2
(
0
.
8
9
)

S
I
H
-
0
.
1
8
4
b
(
2
.
0
0
)
-
0
.
3
0
6
(
1
.
3
2
)
0
.
0
6
8
(
0
.
2
1
)

A
n
N
u
m
-
1
.
7
5
9
(
1
.
4
5
)
-
0
.
3
2
7
(
0
.
4
0
)
0
.
5
9
3
(
0
.
2
4
)
-
0
.
2
3
6
(
0
.
1
2
)
3
.
2
3
5
(
0
.
7
2
)
-
3
.
5
2
7
(
1
.
1
7
)
I
C
D
-
0
.
0
8
3
(
0
.
8
4
)
0
.
0
9
6
(
1
.
2
7
)
-
0
.
2
9
3
(
1
.
3
1
)
-
0
.
1
5
6
(
1
.
4
9
)
-
0
.
5
7
3
c
(
1
.
8
6
)
-
0
.
1
3
5
(
0
.
8
2
)
I
C
D

S
D
S
i
z
e
0
.
1
1
0
(
0
.
9
3
)
-
0
.
0
1
6
(
0
.
3
0
)
0
.
3
2
5
(
1
.
1
6
)
0
.
1
7
2
c
(
1
.
6
7
)
0
.
4
6
1
(
1
.
4
8
)
0
.
0
7
0
(
0
.
6
4
)
S
D
M
B
0
.
0
3
3
b
(
2
.
0
2
)
-
0
.
0
0
3
(
0
.
1
3
)
-
0
.
0
1
0
(
0
.
1
6
)
0
.
0
0
1
(
0
.
0
2
)
0
.
0
3
0
c
(
1
.
6
9
)
0
.
0
6
6
(
0
.
8
5
)
S
D
S
i
z
e
-
0
.
0
9
9
(
0
.
9
6
)
-
0
.
0
4
9
(
0
.
7
0
)
-
0
.
1
9
2
(
1
.
0
7
)
-
0
.
1
3
2
(
0
.
9
9
)
-
0
.
2
4
2
(
0
.
9
6
)
-
0
.
0
4
0
(
0
.
2
1
)
S
D
R
O
A
0
.
7
1
4
(
0
.
9
8
)
2
.
0
7
7
a
(
3
.
2
2
)
1
.
0
9
7
(
0
.
8
1
)
2
.
4
7
0
(
1
.
4
1
)
-
1
.
1
5
5
(
0
.
7
8
)
2
.
7
5
6
(
1
.
4
3
)

R
O
A
1
.
5
2
6
a
(
3
.
6
1
)
2
.
6
1
3
a
(
4
.
7
8
)
3
.
6
9
1
b
(
2
.
3
6
)
5
.
6
5
3
a
(
4
.
1
7
)
1
.
2
2
3
(
0
.
8
0
)
3
.
0
3
3
(
1
.
4
3
)
O
b
s
.
1
5
4
1
1
7
1
2
2
9
5
1
0
2
7
7
R
2
0
.
1
9
0
.
4
2
0
.
2
5
0
.
5
2
0
.
2
3
0
.
4
1
C
o
n
t
i
n
u
e
d
N
e
x
t
P
a
g
e
.
.
.
130
P
a
n
e
l
B
:
P
a
r
e
n
t
C
o
m
p
a
n
y
A
d
j
u
s
t
e
d
R
e
t
u
r
n
s
1
2
M
o
n
t
h
s
2
4
M
o
n
t
h
s
3
6
M
o
n
t
h
s
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)

T
o
t
O
P
2
.
1
0
5
(
1
.
3
5
)
0
.
0
4
3
(
0
.
0
2
)
0
.
4
7
6
(
0
.
2
4
)

S
I
H
-
0
.
2
6
5
a
(
2
.
6
8
)
-
0
.
7
0
9
a
(
3
.
3
7
)
-
0
.
1
5
9
(
0
.
4
7
)

A
n
N
u
m
0
.
5
7
8
c
(
1
.
6
6
)
3
.
0
4
2
a
(
4
.
3
4
)
1
.
2
4
1
(
0
.
8
6
)
3
.
9
6
3
a
(
3
.
0
8
)
-
0
.
3
9
4
(
0
.
1
8
)
2
.
3
5
2
c
(
1
.
7
3
)
I
C
D
-
1
.
7
1
1
(
1
.
4
4
)
-
0
.
1
1
6
(
0
.
1
3
)
-
0
.
0
1
2
(
0
.
0
1
)
0
.
6
3
2
(
0
.
3
8
)
-
0
.
0
5
4
(
0
.
0
2
)
-
3
.
2
0
4
(
1
.
2
7
)

R
O
A
-
0
.
0
1
2
(
0
.
2
3
)
0
.
1
4
0
b
(
2
.
1
6
)
-
0
.
0
1
2
(
0
.
1
4
)
0
.
1
4
3
(
1
.
3
4
)
-
0
.
0
5
1
(
0
.
3
9
)
-
0
.
0
9
1
(
0
.
8
6
)

S
i
z
e
0
.
3
8
3
a
(
2
.
8
0
)
0
.
0
6
0
(
0
.
6
2
)
1
.
1
1
2
c
(
1
.
6
7
)
0
.
2
5
0
c
(
1
.
9
3
)
1
.
0
7
3
c
(
1
.
7
1
)
0
.
1
0
7
(
0
.
6
8
)

M
B
-
0
.
0
0
3
(
0
.
1
9
)
-
0
.
0
1
3
(
0
.
5
6
)
-
0
.
0
0
3
(
0
.
1
0
)
0
.
0
3
7
(
0
.
9
8
)
0
.
0
4
9
(
0
.
7
3
)
0
.
1
5
5
c
(
1
.
9
9
)

C
a
p
E
x
6
.
1
1
9
b
(
2
.
4
5
)
-
2
.
4
9
7
(
1
.
1
0
)
-
2
.
7
4
4
(
0
.
7
4
)
-
5
.
0
1
6
(
1
.
1
8
)
-
5
.
6
7
7
(
1
.
0
0
)
0
.
9
5
8
(
0
.
2
6
)

D
e
b
t
-
0
.
7
9
8
(
1
.
1
8
)
-
1
.
9
1
1
b
(
2
.
3
5
)
-
0
.
5
3
7
(
0
.
2
5
)
-
2
.
6
9
2
(
1
.
4
8
)
-
1
.
3
3
9
(
0
.
9
5
)
-
1
.
4
3
9
(
1
.
1
3
)

D
i
v
Y
l
d
0
.
2
6
7
(
0
.
0
6
)
4
.
7
9
4
(
0
.
7
5
)
3
.
5
2
4
(
0
.
7
6
)
1
.
2
5
4
(
0
.
4
6
)
1
8
.
4
1
2
(
1
.
4
2
)
-
4
.
5
7
5
(
0
.
3
5
)

R
e
p
Y
l
d
0
.
1
9
8
(
0
.
1
3
)
-
3
.
7
9
5
c
(
1
.
9
4
)
1
.
1
7
3
(
0
.
5
8
)
3
.
6
7
0
(
1
.
5
9
)
-
1
.
1
4
6
(
0
.
3
7
)
1
.
2
7
9
(
0
.
4
6
)
O
b
s
.
1
6
4
1
2
3
1
3
6
1
0
2
1
1
3
8
1
R
2
0
.
4
3
0
.
4
0
0
.
3
3
0
.
4
3
0
.
3
8
0
.
5
6
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
e
s
t
i
m
a
t
e
s
f
r
o
m
l
i
n
e
a
r
r
e
g
r
e
s
s
i
o
n
s
e
x
p
l
a
i
n
i
n
g
o
v
e
r
a
l
l
a
n
d
p
a
r
e
n
t
c
o
m
p
a
n
y
a
b
n
o
r
m
a
l
r
e
t
u
r
n
s
1
2
,
2
4
a
n
d
3
6
m
o
n
t
h
s
a
f
t
e
r
t
h
e
e
¤
e
c
t
i
v
e
d
a
t
e
.
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
i
n
p
a
r
e
n
t
h
e
s
e
s
.
S
t
a
n
d
a
r
d
e
r
r
o
r
s
a
r
e
h
e
t
e
r
o
s
c
e
d
a
s
t
i
c
r
o
b
u
s
t
.
A
b
n
o
r
m
a
l
r
e
t
u
r
n
s
a
r
e
c
a
l
c
u
l
a
t
e
d
b
o
t
h
o
v
e
r
a
l
l
a
n
d
t
h
e
p
a
r
e
n
t
c
o
m
p
a
n
y
,
u
s
i
n
g
b
o
t
h
o
w
n
e
r
s
h
i
p
p
e
r
c
e
n
t
a
g
e
a
n
d
s
h
a
r
e
h
o
l
d
e
r
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
(
S
I
H
)
m
a
t
c
h
…
r
m
s
a
s
a
b
e
n
c
h
m
a
r
k
.
A
b
n
o
r
m
a
l
r
e
t
u
r
n
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
2
.
4
.
T
h
e
a
l
g
o
r
i
t
h
m
t
o
c
h
o
o
s
e
m
a
t
c
h
…
r
m
s
i
s
d
e
t
a
i
l
e
d
i
n
C
h
a
p
t
e
r
2
.
4
.
2
.
O
v
e
r
a
l
l
a
b
n
o
r
m
a
l
r
e
t
u
r
n
s
a
r
e
e
x
p
l
a
i
n
e
d
w
i
t
h
v
a
r
i
a
b
l
e
s
d
e
s
c
r
i
b
i
n
g
d
i
¤
e
r
e
n
c
e
s
b
e
t
w
e
e
n
t
h
e
p
a
r
e
n
t
c
o
m
p
a
n
y
a
n
d
d
i
s
t
r
i
b
u
t
e
d
s
u
b
s
i
d
i
a
r
i
e
s
.
F
u
n
d
o
w
n
e
r
s
h
i
p
i
s
e
q
u
a
l
t
o
e
i
t
h
e
r
t
h
e
c
h
a
n
g
e
i
n
t
h
e
t
o
t
a
l
p
e
r
c
e
n
t
a
g
e
o
f
s
h
a
r
e
s
h
e
l
d
o
r
t
h
e
c
h
a
n
g
e
i
n
S
I
H
,
d
e
p
e
n
d
i
n
g
o
n
t
h
e
m
a
t
c
h
a
l
g
o
r
i
t
h
m
.
P
a
r
e
n
t
c
o
m
p
a
n
y
a
b
n
o
r
m
a
l
r
e
t
u
r
n
s
a
r
e
e
x
p
l
a
i
n
e
d
w
i
t
h
p
a
r
e
n
t
c
o
m
p
a
n
y
s
p
e
c
i
…
c
v
a
r
i
a
b
l
e
s
a
s
w
e
l
l
a
s
o
v
e
r
a
l
l
m
e
a
s
u
r
e
s
o
f
s
p
i
n
-
o
¤
i
m
p
o
r
t
a
n
c
e
.
B
o
t
h
s
e
t
s
o
f
r
e
g
r
e
s
s
i
o
n
s
a
l
s
o
i
n
c
l
u
d
e
a
c
o
n
s
t
a
n
t
a
n
d
y
e
a
r
…
x
e
d
-
e
¤
e
c
t
s
.
A
l
l
v
a
r
i
a
b
l
e
s
a
r
e
d
e
…
n
e
d
i
n
S
e
c
t
i
o
n
s
2
.
3
a
n
d
2
.
4
.
S
i
g
n
i
…
c
a
n
c
e
a
t
t
h
e
1
%
l
e
v
e
l
i
s
d
e
s
i
g
n
a
t
e
d
w
i
t
h
a
,
t
h
e
5
%
l
e
v
e
l
w
i
t
h
b
,
a
n
d
t
h
e
1
0
%
l
e
v
e
l
w
i
t
h
c
.
131
T
a
b
l
e
2
.
7
:
P
r
e
-
E
e
x
i
s
t
i
n
g
F
u
n
d
S
h
a
r
e
h
o
l
d
e
r
O
w
n
e
r
s
h
i
p
P
a
t
t
e
r
n
s
P
a
n
e
l
A
:
P
r
o
p
o
r
t
i
o
n
o
f
F
i
r
m
s
H
e
l
d
(
P
r
o
p
)
1
2
M
o
n
t
h
s
2
4
M
o
n
t
h
s
3
6
M
o
n
t
h
s
0
1
2
0
1
2
0
1
2
L
o
n
g
F
I
H
T
e
r
c
i
l
e
F
r
e
q
u
e
n
c
y
1
,
5
8
8
1
,
1
8
7
7
0
2
1
,
4
6
3
8
3
0
4
6
4
1
,
1
7
6
4
7
3
4
0
1
P
e
r
c
e
n
t
4
5
.
7
%
3
4
.
1
%
2
0
.
2
%
5
3
.
1
%
3
0
.
1
%
1
6
.
8
%
5
7
.
4
%
2
3
.
1
%
1
9
.
6
%
M
e
d
.
F
I
H
T
e
r
c
i
l
e
F
r
e
q
u
e
n
c
y
1
,
7
6
1
1
,
0
0
1
3
1
2
1
,
5
1
9
7
2
6
2
0
0
1
,
1
9
5
3
6
5
2
0
4
P
e
r
c
e
n
t
5
7
.
3
%
3
2
.
6
%
1
0
.
2
%
6
2
.
1
%
2
9
.
7
%
8
.
2
%
6
7
.
7
%
2
0
.
7
%
1
1
.
6
%
S
h
o
r
t
F
I
H
T
e
r
c
i
l
e
F
r
e
q
u
e
n
c
y
1
,
4
6
9
6
4
4
1
6
3
1
,
2
5
3
4
4
4
1
1
1
9
5
6
2
5
2
1
0
8
P
e
r
c
e
n
t
6
4
.
5
%
2
8
.
3
%
7
.
2
%
6
9
.
3
%
2
4
.
6
%
6
.
1
%
7
2
.
6
%
1
9
.
2
%
8
.
2
%
P
a
n
e
l
B
:
C
h
a
n
g
e
i
n
O
w
n
e
r
s
h
i
p
P
e
r
c
e
n
t
a
g
e
(

O
P
)
1
2
M
o
n
t
h
s
2
4
M
o
n
t
h
s
3
6
M
o
n
t
h
s
M
e
a
n
%
_
0
%
<
0
M
e
a
n
%
_
0
%
<
0
M
e
a
n
%
_
0
%
<
0
L
o
n
g
F
I
H
T
e
r
c
i
l
e
0
.
0
0
0
3
6
2
.
1
%
3
7
.
9
%
0
.
0
0
0
5
6
5
.
2
%
3
4
.
8
%
0
.
0
0
0
7
6
6
.
2
%
3
3
.
8
%
M
e
d
.
F
I
H
T
e
r
c
i
l
e
0
.
0
0
0
7
5
4
.
8
%
4
5
.
2
%
0
.
0
0
0
5
5
5
.
7
%
4
4
.
3
%
0
.
0
0
0
4
5
4
.
5
%
4
5
.
5
%
S
h
o
r
t
F
I
H
T
e
r
c
i
l
e
0
.
0
0
0
3
5
4
.
3
%
4
5
.
7
%
0
.
0
0
0
4
5
4
.
2
%
4
5
.
8
%
0
.
0
0
0
4
5
3
.
0
%
4
7
.
0
%
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
s
u
m
m
a
r
y
s
t
a
t
i
s
t
i
c
s
f
o
r
t
h
e
t
w
o
v
a
r
i
a
b
l
e
s
d
e
s
c
r
i
b
i
n
g
s
h
a
r
e
h
o
l
d
i
n
g
s
o
f
p
r
e
-
e
x
i
s
t
i
n
g
f
u
n
d
s
h
a
r
e
h
o
l
d
i
n
g
s
f
o
l
l
o
w
i
n
g
t
h
e
e
¤
e
c
t
i
v
e
d
a
t
e
:
t
h
e
p
r
o
p
o
r
t
i
o
n
o
f
…
r
m
s
h
e
l
d
s
t
e
m
m
i
n
g
f
r
o
m
t
h
e
s
a
m
e
p
a
r
e
n
t
c
o
m
p
a
n
y
(
P
r
o
p
)
a
n
d
t
h
e
c
h
a
n
g
e
i
n
o
w
n
e
r
s
h
i
p
p
e
r
c
e
n
t
a
g
e
i
n
…
r
m
s
s
t
i
l
l
h
e
l
d
f
o
l
l
o
w
i
n
g
t
h
e
s
p
i
n
-
o
¤
d
a
t
e
(

O
P
)
.
F
u
n
d
o
w
n
e
r
s
h
i
p
i
s
m
e
a
s
u
r
e
d
1
2
m
o
n
t
h
s
,
2
4
m
o
n
t
h
s
,
a
n
d
3
6
m
o
n
t
h
s
f
o
l
l
o
w
i
n
g
t
h
e
e
¤
e
c
t
i
v
e
d
a
t
e
.
F
u
n
d
o
w
n
e
r
s
h
i
p
i
s
d
e
s
c
r
i
b
e
d
b
y
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
(
F
I
H
)
t
e
r
c
i
l
e
.
P
r
o
p
a
n
d

O
P
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
2
.
5
.
132
T
a
b
l
e
2
.
8
:
P
a
n
e
l
R
e
g
r
e
s
s
i
o
n
s
D
e
s
c
r
i
b
i
n
g
P
r
e
-
E
x
i
s
t
i
n
g
F
u
n
d
S
h
a
r
e
h
o
l
d
e
r
O
w
n
e
r
s
h
i
p
P
a
t
t
e
r
n
s
D
e
p
e
n
d
e
n
t
V
a
r
i
a
b
l
e
P
r
o
p

O
P
T
i
m
e
P
e
r
i
o
d
1
2
M
o
n
t
h
s
2
4
M
o
n
t
h
s
3
6
M
o
n
t
h
s
1
2
M
o
n
t
h
s
2
4
M
o
n
t
h
s
3
6
M
o
n
t
h
s
(
1
)
(
2
)
(
3
)
(
4
)
(
5
)
(
6
)

A
n
N
u
m
8
.
4
6
8
a
(
4
.
2
7
)
8
.
2
1
0
a
(
3
.
2
4
)
1
0
.
1
7
0
a
(
3
.
9
9
)
-
0
.
0
0
2
(
0
.
2
7
)
0
.
0
0
4
(
0
.
4
4
)
-
0
.
0
1
9
(
1
.
2
9
)
I
C
D
-
0
.
0
5
5
(
0
.
3
5
)
-
0
.
1
4
4
(
0
.
7
5
)
-
0
.
2
7
5
(
1
.
0
3
)
0
.
0
0
0
(
0
.
1
5
)
0
.
0
0
1
(
1
.
4
9
)
-
0
.
0
0
1
(
0
.
6
3
)
I
C
D

S
D
S
i
z
e
-
0
.
0
3
0
(
0
.
2
6
)
0
.
0
3
6
(
0
.
2
6
)
0
.
2
2
4
(
1
.
2
7
)
0
.
0
0
0
(
0
.
2
4
)
-
0
.
0
0
1
(
0
.
9
0
)
0
.
0
0
1
(
0
.
7
2
)
S
D
M
B
0
.
0
9
4
b
(
2
.
2
3
)
0
.
0
9
8
c
(
1
.
8
2
)
0
.
0
4
4
(
0
.
7
5
)
-
0
.
0
0
1
(
1
.
5
7
)
-
0
.
0
0
1
c
(
1
.
7
7
)
0
.
0
0
0
(
0
.
2
8
)
S
D
S
i
z
e
0
.
0
1
8
(
0
.
0
9
)
0
.
2
2
0
(
1
.
0
0
)
-
0
.
2
9
9
(
1
.
0
2
)
0
.
0
0
0
(
0
.
3
0
)
0
.
0
0
0
(
0
.
0
4
)
0
.
0
0
0
(
0
.
2
7
)
S
D
R
O
A
-
5
.
2
1
9
b
(
1
.
9
8
)
-
6
.
1
4
1
a
(
2
.
8
1
)
-
5
.
5
3
3
b
(
2
.
4
7
)
0
.
0
3
9
(
1
.
6
0
)
0
.
0
0
0
(
0
.
0
2
)
-
0
.
0
0
5
(
0
.
3
7
)

R
O
A
-
0
.
2
8
9
(
0
.
2
0
)
-
0
.
2
3
3
(
0
.
1
1
)
-
0
.
5
7
7
(
0
.
2
5
)
0
.
0
1
4
(
1
.
5
2
)
0
.
0
2
4
(
2
.
1
1
)
-
0
.
0
0
2
(
0
.
1
1
)
F
I
H
0
.
7
0
6
a
(
5
.
4
9
)
0
.
6
5
8
a
(
3
.
6
1
)
0
.
4
6
7
b
(
2
.
5
8
)
0
.
0
0
0
(
0
.
7
6
)
0
.
0
0
0
(
0
.
2
4
)
0
.
0
0
0
(
0
.
3
1
)

A
n
N
u
m

F
I
H
-
2
.
3
5
2
a
(
3
.
5
4
)
-
1
.
7
8
3
b
(
2
.
1
3
)
-
2
.
0
1
0
b
(
2
.
3
3
)
0
.
0
0
2
(
0
.
6
1
)
0
.
0
0
0
(
0
.
1
5
)
0
.
0
0
4
(
0
.
9
2
)
I
C
D

F
I
H
0
.
0
1
4
(
0
.
2
8
)
0
.
0
3
6
(
0
.
5
7
)
0
.
0
7
1
(
0
.
8
4
)
0
.
0
0
0
(
0
.
1
8
)
0
.
0
0
0
(
1
.
4
7
)
0
.
0
0
0
(
0
.
4
9
)
I
C
D

S
D
S
i
z
e

F
I
H
0
.
0
0
8
(
0
.
2
2
)
0
.
0
0
7
(
0
.
1
6
)
-
0
.
0
5
2
(
0
.
9
2
)
0
.
0
0
0
(
0
.
1
6
)
0
.
0
0
0
(
0
.
8
4
)
0
.
0
0
0
(
0
.
7
5
)
S
D
M
B

F
I
H
-
0
.
0
3
0
b
(
2
.
2
2
)
-
0
.
0
3
0
(
1
.
6
2
)
-
0
.
0
0
8
(
0
.
4
7
)
0
.
0
0
0
(
1
.
6
3
)
0
.
0
0
0
(
1
.
4
6
)
0
.
0
0
0
(
0
.
2
9
)
S
D
S
i
z
e

F
I
H
-
0
.
0
5
2
(
0
.
8
3
)
-
0
.
1
3
0
c
(
1
.
8
5
)
0
.
0
2
5
(
0
.
2
8
)
0
.
0
0
0
(
0
.
4
6
)
0
.
0
0
0
(
0
.
2
6
)
0
.
0
0
0
(
0
.
1
4
)
S
D
R
O
A

F
I
H
1
.
5
8
3
c
(
1
.
8
6
)
2
.
0
8
1
a
(
2
.
9
4
)
1
.
9
4
9
a
(
2
.
7
1
)
-
0
.
0
1
0
c
(
1
.
6
5
)
0
.
0
0
0
(
0
.
0
0
)
0
.
0
0
1
(
0
.
2
6
)

R
O
A

F
I
H
-
0
.
0
5
0
(
0
.
1
1
)
0
.
5
1
7
(
0
.
7
8
)
0
.
3
2
7
(
0
.
4
5
)
-
0
.
0
0
4
(
1
.
5
5
)
-
0
.
0
0
7
b
(
2
.
1
6
)
0
.
0
0
0
(
0
.
0
8
)
N
5
9
3
5
3
8
4
7
3
0
9
8
3
7
4
1
1
8
4
6
1
3
8
8
A
d
j
.
R
2
0
.
0
3
0
.
0
3
0
.
0
5
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
e
s
t
i
m
a
t
e
s
f
r
o
m
o
r
d
e
r
e
d
p
r
o
b
i
t
r
e
g
r
e
s
s
i
o
n
s
a
n
d
l
i
n
e
a
r
r
e
g
r
e
s
s
i
o
n
s
e
x
p
l
a
i
n
i
n
g
t
h
e
p
r
o
p
o
r
t
i
o
n
o
f
s
h
a
r
e
s
h
e
l
d
(
P
r
o
p
)
a
n
d
t
h
e
c
h
a
n
g
e
i
n
o
w
n
e
r
s
h
i
p
p
e
r
c
e
n
t
a
g
e
f
r
o
m
b
e
f
o
r
e
t
o
a
f
t
e
r
t
h
e
s
p
i
n
-
o
¤
e
v
e
n
t
b
y
p
r
e
-
e
x
i
s
t
i
n
g
s
h
a
r
e
h
o
l
d
e
r
s
a
t
t
h
e
f
u
n
d
l
e
v
e
l
.
F
o
r
e
a
c
h
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
o
n
e
r
e
g
r
e
s
s
i
o
n
i
s
e
s
t
i
m
a
t
e
d
u
s
i
n
g
s
h
a
r
e
h
o
l
d
i
n
g
s
a
t
t
h
e
f
u
n
d
-
l
e
v
e
l
1
2
m
o
n
t
h
s
,
2
4
m
o
n
t
h
s
,
a
n
d
3
6
m
o
n
t
h
s
a
f
t
e
r
t
h
e
e
¤
e
c
t
i
v
e
d
a
t
e
.
F
o
r
t
h
e
c
h
a
n
g
e
i
n
o
w
n
e
r
s
h
i
p
p
e
r
c
e
n
t
a
g
e
r
e
g
r
e
s
s
i
o
n
s
o
n
l
y
f
u
n
d
p
o
s
i
t
i
o
n
s
w
i
t
h
a
p
o
s
i
t
i
v
e
s
t
a
k
e
a
t
t
h
e
l
a
t
e
r
d
a
t
e
.
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
i
n
p
a
r
e
n
t
h
e
s
e
s
.
S
t
a
n
d
a
r
d
e
r
r
o
r
s
a
r
e
c
l
u
s
t
e
r
e
d
a
t
t
h
e
f
u
n
d
l
e
v
e
l
.
I
n
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
i
n
c
l
u
d
e
m
e
a
s
u
r
e
s
d
e
s
c
r
i
b
i
n
g
d
i
¤
e
r
e
n
c
e
s
b
e
t
w
e
e
n
t
h
e
p
a
r
e
n
t
c
o
m
p
a
n
y
a
n
d
d
i
s
t
r
i
b
u
t
e
d
s
u
b
s
i
d
i
a
r
i
e
s
,
a
c
o
n
s
t
a
n
t
,
y
e
a
r
…
x
e
d
-
e
¤
e
c
t
s
,
a
n
d
o
r
i
g
i
n
a
l
p
a
r
e
n
t
c
o
m
p
a
n
y
i
n
d
u
s
t
r
y
…
x
e
d
-
e
¤
e
c
t
s
b
a
s
e
d
o
n
t
h
e
F
a
m
a
-
F
r
e
n
c
h
4
8
I
n
d
u
s
t
r
y
C
l
a
s
s
i
…
c
a
t
i
o
n
.
F
u
n
d
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
(
F
I
H
)
a
n
d
i
n
t
e
r
a
c
t
i
o
n
t
e
r
m
s
b
e
t
w
e
e
n
F
I
H
a
n
d
a
l
l
i
n
d
e
p
e
n
d
e
n
t
v
a
r
i
a
b
l
e
s
a
r
e
a
l
s
o
i
n
c
l
u
d
e
d
.
A
l
l
v
a
r
i
a
b
l
e
s
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
s
2
.
3
t
h
r
o
u
g
h
2
.
5
.
S
i
g
n
i
…
c
a
n
c
e
a
t
t
h
e
1
%
l
e
v
e
l
i
s
d
e
s
i
g
n
a
t
e
d
w
i
t
h
a
,
t
h
e
5
%
l
e
v
e
l
w
i
t
h
b
,
a
n
d
t
h
e
1
0
%
l
e
v
e
l
w
i
t
h
c
.
133
Table 2.9: Panel Regressions Describing Changes in Fund Ownership Before &
After Spin-o¤ Events
Dependent Variable OP ROL Pos. Close
(1) (2) (3)
PSOInd -0.001
a
(4.22) -0.023 (0.79) 0.084
c
(1.68)
FIH 0.000
a
(2.59) 0.071
a
(9.19) -0.026
a
(11.60)
PSOIndFIH 0.000
a
(2.81) 0.036
a
(3.91) 0.002 (0.94)
ROA 0.000 (0.08) 1.089
a
(2.76) -0.165 (0.32)
ROAPSOInd -0.008 (1.35) -1.126
a
(2.67) 0.980
c
(1.70)
ROAFIH 0.000 (0.01) -0.159 (1.40) 0.031 (1.37)
ROAPSOIndFIH 0.002 (1.06) 0.174 (1.40) -0.049
c
(1.83)
AnNum 0.003 (1.04) 0.907
a
(2.79) -1.135
b
(2.55)
AnNumPSOInd 0.012
a
(3.85) -0.144 (0.40) 0.684 (1.47)
AnNumFIH 0.000 (0.12) -0.223
b
(2.25) 0.028 (1.37)
AnNumPSOIndFIH -0.003
a
(3.34) 0.026 (0.23) -0.034 (1.58)
Size 0.000 (0.99) 0.148
a
(3.65) -0.061 (1.39)
MB 0.000 (0.46) -0.001 (0.78) 0.000 (0.08)
CapEx 0.000 (0.06) 0.295 (1.36) -0.979
b
(2.33)
Debt -0.002 (0.82) 0.003 (0.02) -0.065 (0.41)
DivYld 0.003
c
(1.68) -0.515 (1.56) -0.422
c
(1.91)
RepYld 0.003 (1.23) -0.274 (1.38) -0.552
b
(2.07)
AnnRet 0.000 (0.28) 0.059
a
(4.57) -0.013 (0.95)
SizeFIH 0.000 (0.25) -0.016 (1.29) 0.001 (0.31)
MBFIH 0.000 (0.18) 0.000 (0.82) 0.000
a
(3.50)
CapExFIH 0.000 (0.00) -0.064 (1.06) 0.023 (1.24)
DebtFIH 0.001 (0.96) -0.016 (0.40) 0.014
b
(2.08)
DivYldFIH -0.001
c
(1.74) 0.125 (1.36) -0.008 (0.81)
RepYldFIH 0.000 (0.73) 0.086 (1.48) 0.019 (1.35)
AnnRetFIH 0.000 (0.30) -0.013
a
(3.35) 0.000 (0.67)
PrntCo 0.001
a
(4.37) -0.003 (0.45) 0.035 (1.44)
HldBef 0.000 (1.16) -0.160
a
(26.59) -0.014 (0.68)
Obs. 39057 39079 62890
Adj. R
2
0.02 0.05
This table reports estimates from linear and Cox panel regressions explaining changes in fund
ownership before the spin-o¤ announcement date and after the e¤ective date. t-statistics are in
parentheses. Standard errors are clustered at the fund level. The variables of interest include
those incorporating the post spin-o¤ indicator variable, the change in return-on-assets, and the
change in analyst coverage. Other independent variables include changes in …rm characteristics,
a pre-existing shareholder indicator variable, a parent company indicator variable, announcement
year …xed-e¤ects, and industry …xed-e¤ects based on the Fama-French 48 Industry Classi…cation.
Linear regressions also include a constant. Variables are de…ned in Chapters 1.4, 2.3, and 2.6.
Signi…cance at the 1% level is designated with a, the 5% level with b, and the 10% level with
c.
134
T
a
b
l
e
3
.
1
:
F
u
n
d
O
w
n
e
r
s
h
i
p
b
y
F
i
r
m
T
y
p
e
-
S
i
z
e
,
M
a
r
k
e
t
-
t
o
-
B
o
o
k
,
&
P
a
y
o
u
t
P
o
l
i
c
y
A
l
l
F
u
n
d
s
S
h
o
r
t
F
I
H
M
e
d
i
u
m
F
I
H
L
o
n
g
F
I
H
#
F
i
r
m
s
O
w
n
.
#
F
u
n
d
s
O
w
n
.
#
F
u
n
d
s
O
w
n
.
#
F
u
n
d
s
O
w
n
.
#
F
u
n
d
s
S
i
z
e
Q
u
i
n
t
i
l
e
S
m
a
l
l
1
3
2
5
1
0
.
0
1
0
1
.
0
0
.
0
0
1
0
.
1
0
.
0
0
2
0
.
2
0
.
0
0
6
0
.
7
2
1
3
2
5
1
0
.
0
4
1
4
.
2
0
.
0
0
8
0
.
8
0
.
0
0
9
1
.
0
0
.
0
2
4
2
.
3
3
1
3
2
5
1
0
.
0
8
0
1
2
.
1
0
.
0
2
0
3
.
6
0
.
0
2
2
3
.
5
0
.
0
3
8
5
.
1
4
1
3
2
5
1
0
.
1
1
1
2
9
.
7
0
.
0
3
1
9
.
9
0
.
0
3
4
9
.
2
0
.
0
4
6
1
0
.
7
L
a
r
g
e
1
3
2
5
1
0
.
1
3
2
1
1
0
.
8
0
.
0
3
0
2
9
.
9
0
.
0
3
8
3
5
.
7
0
.
0
6
4
4
5
.
2
M
B
Q
u
i
n
t
i
l
e
V
a
l
u
e
1
3
2
5
1
0
.
0
4
4
8
.
6
0
.
0
0
7
1
.
9
0
.
0
1
1
2
.
5
0
.
0
2
6
4
.
2
2
1
3
2
5
1
0
.
0
6
6
2
0
.
5
0
.
0
1
3
4
.
9
0
.
0
1
8
6
.
4
0
.
0
3
6
9
.
2
3
1
3
2
5
1
0
.
0
8
3
3
2
.
0
0
.
0
1
9
8
.
5
0
.
0
2
4
1
0
.
2
0
.
0
4
0
1
3
.
3
4
1
3
2
5
1
0
.
0
9
3
4
2
.
0
0
.
0
2
5
1
2
.
1
0
.
0
2
7
1
3
.
2
0
.
0
4
1
1
6
.
7
G
r
o
w
t
h
1
3
2
5
1
0
.
0
8
8
5
4
.
7
0
.
0
2
7
1
6
.
9
0
.
0
2
6
1
7
.
3
0
.
0
3
5
2
0
.
5
P
a
y
o
u
t
P
o
l
i
c
y
D
i
v
=
0
,
R
e
p
=
0
2
8
9
9
3
0
.
0
5
6
1
3
.
7
0
.
0
1
6
4
.
4
0
.
0
1
6
4
.
1
0
.
0
2
4
5
.
2
D
i
v
=
0
,
R
e
p
>
0
1
2
1
3
1
0
.
0
8
7
3
1
.
6
0
.
0
2
3
1
0
.
0
0
.
0
2
5
9
.
9
0
.
0
3
9
1
1
.
7
D
i
v
>
0
,
R
e
p
=
0
1
0
7
4
8
0
.
0
8
1
3
1
.
8
0
.
0
1
7
8
.
4
0
.
0
2
3
1
0
.
0
0
.
0
4
1
1
3
.
4
D
i
v
>
0
,
R
e
p
>
0
1
4
3
7
7
0
.
0
9
7
6
7
.
4
0
.
0
1
9
1
7
.
2
0
.
0
2
7
2
1
.
8
0
.
0
5
1
2
8
.
5
T
h
i
s
t
a
b
l
e
p
r
e
s
e
n
t
s
a
v
e
r
a
g
e
f
u
n
d
o
w
n
e
r
s
h
i
p
a
n
d
t
h
e
n
u
m
b
e
r
o
f
f
u
n
d
s
h
a
r
e
h
o
l
d
e
r
s
f
o
r
…
r
m
s
c
l
a
s
s
i
…
e
d
b
y
s
i
z
e
q
u
i
n
t
i
l
e
,
m
a
r
k
e
t
-
t
o
-
b
o
o
k
q
u
i
n
t
i
l
e
,
a
n
d
p
a
y
o
u
t
p
o
l
i
c
y
.
F
i
r
m
o
w
n
e
r
s
h
i
p
s
t
a
t
i
s
t
i
c
s
a
r
e
m
e
a
s
u
r
e
d
a
n
n
u
a
l
l
y
a
t
y
e
a
r
-
e
n
d
,
u
s
i
n
g
e
i
t
h
e
r
a
l
l
f
u
n
d
s
o
r
c
l
a
s
s
i
f
e
d
b
y
f
u
n
d
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
(
F
I
H
)
t
e
r
c
i
l
e
.
F
I
H
i
s
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
1
.
4
.
F
u
n
d
o
w
n
e
r
s
h
i
p
i
s
t
h
e
p
r
o
p
o
r
t
i
o
n
o
f
s
h
a
r
e
s
h
e
l
d
a
t
y
e
a
r
e
n
d
d
i
v
i
d
e
d
b
y
c
o
m
m
o
n
s
h
a
r
e
s
o
u
t
s
t
a
n
d
i
n
g
.
S
i
z
e
q
u
i
n
t
i
l
e
s
,
m
a
r
k
e
t
-
t
o
-
b
o
o
k
q
u
i
n
t
i
l
e
s
,
a
n
d
p
a
y
o
u
t
p
o
l
i
c
y
i
s
c
l
a
s
s
i
…
e
d
a
n
n
u
a
l
l
y
.
A
…
r
m
’
s
p
a
y
o
u
t
p
o
l
i
c
y
i
s
b
a
s
e
d
o
n
w
h
e
t
h
e
r
o
r
n
o
t
t
h
e
…
r
m
p
a
y
s
d
i
v
i
d
e
n
d
s
o
r
d
o
e
s
n
o
t
p
a
y
d
i
v
i
d
e
n
d
s
(
D
i
v
.
>
0
,
D
i
v
.
=
0
)
,
a
n
d
w
h
e
t
h
e
r
o
r
n
o
t
t
h
e
…
r
m
h
a
s
a
n
a
c
t
i
v
e
s
h
a
r
e
r
e
p
u
r
c
h
a
s
e
p
r
o
g
r
a
m
(
R
e
p
.
>
0
,
R
e
p
.
=
0
)
.
A
l
l
…
r
m
-
y
e
a
r
o
b
s
e
r
v
a
t
i
o
n
s
f
r
o
m
1
9
8
8
t
o
2
0
0
7
a
r
e
u
s
e
d
.
135
Table 3.2: Determinants of Ownership Proportion by Fund Investment
Horizon Tercile
Panel A: Dividend and Share Repurchase Control Variables
Fund Investment Horizon Tercile
Short Med. Long Short Med. Long
(1) (2) (3) (4) (5) (6)
DivYld
t
-0.054
a
-0.013
c
-0.001
(3.47) (1.92) (0.29)
RepYld
t
0.009 0.015
a
0.018
a
(1.54) (3.47) (3.39)
DivInd
t
-0.005
a
-0.001 0.002
(3.07) (0.57) (1.43)
RepInd
t
-0.001 0.000 0.003
a
(1.12) (0.01) (3.03)
ROA
t
0.018
a
0.016
a
0.004 0.018
a
0.017
a
0.003
(6.12) (3.57) (1.34) (6.30) (3.81) (1.41)
NonOp
t
0.009 0.005 -0.006 0.009 0.006 -0.006
(1.63) (0.56) (0.85) (1.63) (0.72) (0.84)
CapEx
t
0.025
a
0.018
a
0.008
c
0.024
a
0.018
a
0.009
c
(5.37) (3.85) (2.06) (5.48) (3.76) (2.04)
Debt
t
-0.010
a
-0.009
a
-0.009
a
-0.010
a
-0.009
a
-0.008
a
(5.98) (3.80) (6.41) (6.39) (3.71) (5.34)
Size
t
0.011
a
0.011
a
0.008
a
0.011
a
0.011
a
0.008
a
(10.24) (9.46) (7.09) (10.16) (9.29) (6.79)
MB
t
0.000
a
0.000
a
0.000 0.000
a
0.000
a
0.000
(3.10) (3.12) (0.05) (3.09) (3.19) (0.24)
AnnRet
t
0.005
a
-0.001 -0.003
b
0.005
a
-0.001 -0.003
b
(5.07) (1.59) (2.58) (5.22) (1.61) (2.63)
SDRet
t
-0.679
a
-0.750
a
-0.508
a
-0.702
a
-0.754
a
-0.492
a
(11.99) (10.38) (5.71) (11.15) (10.02) (5.35)
Beta
t
0.003
a
0.003
a
0.001 0.003
a
0.003
a
0.001
(5.82) (3.21) (1.11) (5.04) (3.02) (1.17)
Vol
t
0.108
a
0.070
a
0.015
a
0.107
a
0.071
a
0.018
a
(7.91) (9.14) (4.21) (7.87) (8.79) (4.01)
Log(Age
t
) -0.006
a
-0.005
a
0.002
b
-0.005
a
-0.005
a
0.002
b
(6.31) (4.43) (2.52) (6.96) (4.38) (2.41)
SP500
t
-0.017
a
-0.016
a
-0.001 -0.017
a
-0.015
a
-0.001
(4.64) (5.24) (0.16) (4.79) (5.17) (0.13)
Continued Next Page...
136
Panel B: Aggregate Payout Control Variables
Fund Investment Horizon Tercile
Short Med. Long Short Med. Long
(1) (2) (3) (4) (5) (6)
TotYld
t
-0.002 0.008
b
0.010
b
(0.40) (2.19) (2.15)
PayInd
t
-0.003
a
0.001 0.004
a
(5.21) (0.92) (4.55)
ROA
t
0.018
a
0.016
a
0.004 0.018
a
0.017
a
0.003
(6.35) (3.60) (1.39) (6.73) (3.84) (1.41)
NonOp
t
0.008 0.004 -0.006 0.009 0.005 -0.007
(1.46) (0.49) (0.85) (1.56) (0.61) (0.97)
CapEx
t
0.025
a
0.018
a
0.009
b
0.024
a
0.018
a
0.009
c
(5.35) (3.88) (2.09) (5.53) (3.84) (2.08)
Debt
t
-0.009
a
-0.009
a
-0.009
a
-0.010
a
-0.009
a
-0.008
a
(6.20) (3.80) (6.56) (6.38) (3.77) (6.42)
Size
t
0.011
a
0.011
a
0.008
a
0.011
a
0.011
a
0.008
a
(10.20) (9.46) (7.10) (10.29) (9.43) (7.05)
MB
t
0.000
a
0.000
a
0.000 0.000
a
0.000
a
0.000
(3.10) (3.12) (0.10) (3.07) (3.19) (0.31)
AnnRet
t
0.005
a
-0.001 -0.003
b
0.005
a
-0.001 -0.003
b
(5.15) (1.61) (2.58) (5.25) (1.61) (2.58)
SDRet
t
-0.676
a
-0.750
a
-0.510
a
-0.696
a
-0.751
a
-0.492
a
(11.69) (10.38) (5.65) (11.43) (10.23) (5.59)
Beta
t
0.003
a
0.003
a
0.001 0.003
a
0.003
a
0.001
(5.93) (3.23) (1.11) (5.58) (3.11) (1.24)
Vol
t
0.108
a
0.070
a
0.015
a
0.108
a
0.070
a
0.017
a
(7.97) (9.21) (4.23) (8.06) (9.09) (4.23)
Log(Age
t
) -0.006
a
-0.005
a
0.002
a
-0.006
a
-0.005
a
0.002
b
(6.16) (4.43) (2.41) (6.24) (4.42) (2.10)
SP500
t
-0.017
a
-0.016
a
-0.001 -0.017
a
-0.015
a
0.000
(4.74) (5.29) (0.13) (4.78) (5.25) (0.07)
This table reports Fama and MacBeth (1973) style estimates of tobit regressions.
Newey-West t-statistics (two-lags) are in parentheses. One cross-sectional regres-
sion for each fund investment horizon tercile is estimated per year from 1988 to
2007. The dependent variable is the proportion of common shares outstanding
held by mutual funds. Independent variables include total payout yield, a payout
indicator variable, operating income, non-operating income, capital expenditures,
debt, size, market-to-book, annual return, standard deviation of returns, beta,
volume, …rm age, and S&P 500 inclusion. Independent variables are de…ned in
Chapter 3.3. I also include industry …xed-e¤ects based on the Fama-French 48
Industry Classi…cation. Signi…cance at the 1% level is denoted with a, the 5%
level with b, and the 10% level with c.
137
Table 3.3: Determinants of Relative Ownership Length by Fund In-
vestment Horizon Tercile
Panel A: Dividend & Share Repurchase Control Variables
Fund Investment Horizon Tercile
Short Med. Long Short Med. Long
(1) (2) (3) (4) (5) (6)
DivYld
t
-0.239
b
-0.096 0.122
(2.72) (1.17) (1.51)
RepYld
t
0.020 0.065
c
0.089
b
(0.61) (1.86) (2.21)
DivInd
t
-0.010
b
-0.011
b
0.023
a
(2.74) (2.81) (5.59)
RepInd
t
0.000 0.007
b
0.024
a
(0.10) (2.37) (10.02)
ROA
t
0.061
b
0.042
c
-0.012 0.049
b
0.043
c
-0.007
(2.39) (1.80) (0.95) (2.56) (2.06) (0.42)
NonOp
t
0.391
a
0.375
a
0.100 0.381
a
0.367
a
0.115
(6.26) (3.65) (1.28) (6.42) (3.40) (1.31)
CapEx
t
0.006 -0.002 0.000 0.023 0.001 0.007
(0.39) (0.08) (0.01) (1.46) (0.05) (0.39)
Debt
t
-0.034
b
-0.057
a
-0.023 -0.035 -0.057
a
-0.017
a
(2.77) (5.79) (4.32) (2.76) (5.94) (3.15)
Size
t
0.012
a
0.016
a
0.007
a
0.013
a
0.016
a
0.005
b
(5.27) (5.39) (3.31) (6.05) (5.97) (2.28)
MB
t
0.001
a
0.001
b
0.000 0.001
a
0.001
b
0.000
(3.09) (2.03) (0.55) (2.98) (2.19) (0.02)
AnnRet
t
-0.019
a
-0.014
a
0.011
b
-0.019
a
-0.014
a
0.010
b
(4.76) (3.09) (2.28) (3.89) (2.89) (2.28)
SDRet
t
-0.592 -0.515 -1.024
b
-0.643 -0.598 -0.782
(1.54) (1.05) (2.24) (1.49) (1.35) (1.67)
Beta
t
-0.006
a
-0.013
a
-0.018
a
-0.006
a
-0.013
a
-0.017
a
(3.25) (5.49) (5.01) (3.19) (5.11) (4.87)
Vol
t
-0.205
a
-0.395
a
-0.501
a
-0.211
a
-0.404
a
-0.482
a
(5.19) (5.70) (4.99) (5.26) (5.60) (4.97)
Log(Age
t
) -0.006 0.004 0.069
a
-0.006
b
0.005 0.067
a
(2.57) (1.65) (25.45) (2.74) (1.70) (25.91)
SP500
t
-0.041
a
-0.050
a
-0.075
a
-0.042
a
-0.050
a
-0.076
a
(6.47) (7.08) (14.44) (6.73) (7.22) (14.53)
Continued Next Page...
138
Panel B: Aggregate Payout Control Variables
Fund Investment Horizon
Short Med. Long Short Med. Long
TotYld
t
-0.022 0.017 0.088
a
(0.58) (0.45) (2.90)
PayInd
t
-0.011
a
-0.007 0.024
a
(4.86) (1.48) (7.21)
ROA
t
0.053
b
0.038 -0.012 0.053
a
0.049
b
-0.001
(2.42) (1.65) (0.91) (2.88) (2.35) (0.05)
NonOp
t
0.385
a
0.374
a
0.102 0.386
a
0.376
a
0.119
(6.27) (3.63) (1.28) (6.32) (3.49) (1.34)
CapEx
t
0.009 0.000 0.001 0.017 -0.006 -0.002
(0.56) (0.02) (0.03) (1.12) (0.26) (0.11)
Debt
t
-0.031
b
-0.056
a
-0.023
a
-0.034
b
-0.056
a
-0.020
a
(2.62) (5.58) (4.60) (2.75) (5.52) (3.43)
Size
t
0.012
a
0.016
a
0.007
a
0.013
a
0.016
a
0.006
a
(5.42) (5.49) (3.30) (6.00) (5.86) (2.87)
MB
t
0.001
a
0.000
a
0.000 0.001
a
0.001
b
0.000
(3.01) (2.04) (0.52) (2.93) (2.14) (0.11)
AnnRet
t
-0.018
a
-0.013
a
0.010
b
-0.019
a
-0.015
a
0.010
b
(4.32) (2.94) (2.25) (3.79) (2.92) (2.17)
SDRet
t
-0.550 -0.511 -1.043
b
-0.648 -0.585 -0.893
c
(1.42) (1.05) (2.27) (1.57) (1.25) (1.91)
Beta
t
-0.005
a
-0.012
a
-0.018
a
-0.006
a
-0.013
a
-0.017
a
(2.90) (5.72) (5.21) (2.92) (5.50) (5.07)
Vol
t
-0.200
a
-0.391
a
-0.499
a
-0.207
a
-0.395
a
-0.494
a
(5.16) (5.74) (5.12) (5.14) (5.70) (5.14)
Log(Age
t
) -0.007
a
0.004 0.069
a
-0.007
a
0.004 0.069
a
(3.34) (1.42) (25.91) (3.74) (1.44) (24.89)
SP500
t
-0.041
a
-0.050
a
-0.075
a
-0.042
a
-0.049
a
-0.074
a
(6.57) (7.22) (13.94) (6.58) (7.33) (12.90)
This table reports Fama and MacBeth (1973) style estimates of truncated regres-
sions. Newey-West t-statistics (two-lags) are in parentheses. One cross-sectional
regression for each fund investment horizon tercile is estimated per year from 1988
to 2007. The dependent variable is the relative ownership length of a stock po-
sition within its mutual fund portfolio. Relative ownership length is de…ned in
Chapter 3.4. Independent variables include total payout yield, a payout indicator
variable, return-on-assets, non-operating income, capital expenditures, debt, size,
market-to-book, annual return, standard deviation of returns, beta, volume, …rm
age, and S&P 500 inclusion. Independent variables are de…ned in Chapter 3.3. I
also include industry …xed-e¤ects based on the Fama-French 48 Industry Classi-
…cation. Signi…cance at the 1% level is denoted with a, the 5% level with b, and
the 10% level with c.
139
Table 3.4: Shareholder Investment Horizon Changes Around Payout Events
Panel A: Dividend Events
t ÷1 to t + 1 t + 1 to t + 2 t ÷1 to t + 2
N Mean Med. N Mean Med. N Mean Med.
Unadjusted Changes
Increase 2150 0.045
a
0.027
a
1761 0.017
a
0.013
a
1860 0.055
a
0.048
a
(9.29) (8.54) (4.00) (3.67) (9.48) (9.47)
Decrease 492 -0.014 -0.011 382 0.005 -0.003 425 -0.028
b
-0.043
a
(1.17) (1.54) (0.57) (0.13) (2.09) (2.90)
Initiation 215 0.045
a
0.049
a
184 0.039
a
0.050
a
180 0.093
a
0.081
a
(2.62) (3.06) (3.51) (4.39) (4.77) (4.85)
Omission 152 -0.033
c
-0.038
c
117 0.013 0.005 132 -0.002 -0.023
(1.79) (1.92) (0.71) (0.46) (0.09) (0.44)
Adjusted Changes
Increase 2150 0.020
a
0.013
a
1761 0.027
a
0.018
a
1860 0.041
a
0.038
a
(2.95) (2.69) (4.45) (4.06) (4.94) (4.95)
Decrease 492 0.002 -0.010 382 0.005 0.004 425 -0.010 0.003
(0.12) (0.35) (0.38) (0.41) (0.55) (0.04)
Initiation 215 -0.010 -0.010 184 -0.004 0.013 180 0.031 0.043
(0.40) (0.17) (0.25) (0.24) (1.19) (1.46)
Omission 152 0.016 0.007 117 0.013 0.025 132 0.028 -0.003
(0.61) (0.85) (0.57) (0.80) (0.88) (0.61)
Continued Next Page...
140
Panel B: Share Repurchase Events
t ÷1 to t + 1 t + 1 to t + 2 t ÷1 to t + 2
N Mean Med. N Mean Med. N Mean Med.
Unadjusted Changes
Non-Dividend Paying Firms
All 2560 0.049
a
0.038
a
1940 0.032
a
0.020
a
2063 0.070
a
0.054
a
(11.27) (11.20) (7.71) (8.15) (12.05) (11.87)
Initiation 880 0.063
a
0.052
a
651 0.044
a
0.025
a
711 0.092
a
0.081
a
(7.73) (7.74) (5.78) (5.92) (8.65) (8.79)
Non-Init. 1680 0.042
a
0.033
a
1289 0.025
a
0.016
a
1352 0.058
a
0.038
a
(8.25) (8.17) (5.26) (5.75) (8.52) (8.13)
Dividend Paying Firms
All 2701 0.012
a
0.009
a
2171 0.003 0.002 2360 0.013
a
0.007
a
(3.13) (3.45) (1.01) (1.07) (2.67) (2.42)
Initiation 690 0.010 0.012
c
513 0.011 0.006 596 0.019
c
0.022
a
(1.22) (1.79) (1.56) (1.14) (1.82) (2.10)
Non-Init. 2011 0.013
a
0.008
a
1658 0.001 0.002 1764 0.011
b
0.004
(2.93) (2.94) (0.28) (0.57) (2.01) (1.55)
Adjusted Changes
Non-Dividend Paying Firms
All 2560 -0.001 0.002 1940 -0.006 0.002 2063 -0.006 -0.002
(0.12) (0.27) (1.00) (0.44) (0.70) (0.53)
Initiation 880 0.004 0.010 651 0.012 0.009 711 0.010 0.011
(0.34) (0.61) (1.11) (0.95) (0.66) (0.82)
Non-Init. 1680 -0.003 -0.002 1289 -0.015
b
-0.001 1352 -0.014 -0.007
(0.41) (0.10) (2.12) (1.22) (1.50) (1.29)
Dividend Paying Firms
All 2701 0.011
c
0.015
a
2171 0.000 -0.002 2360 0.005 0.007
(1.93) (2.60) (0.02) (0.47) (0.74) (0.65)
Initiation 690 0.003 0.015 513 -0.003 -0.006 596 0.002 0.008
(0.26) (1.10) (0.29) (0.40) (0.12) (0.20)
Non-Init. 2011 0.013
b
0.015
b
1658 0.001 -0.001 1764 0.006 0.006
(2.13) (2.37) (0.15) (0.30) (0.81) (0.64)
This table reports changes in average shareholder investment horizon (SIH) for payout event
…rms. I calculate both mean and median un-adjusted and adjusted changes. Test-statistics
are located below the reported change. SIH is de…ned in Chapter 3.5. Unadjusted change is
equal to the di¤erence in SIH between dates. Adjusted change is equal to the di¤erence in
SIH between dates minus a similar change in a control …rm. Control …rms are chosen based
on similar MB, ROA, ROA, and industry classi…cation. The algorithm to match event …rms
with control …rms is de…ned in Chapter 3.5. For the di¤erence in means, I use the two-tailed
t-statistic to test sign…cance. For the di¤erence in medians I use the two-tailed z-statistic from
the Wilcoxon rank test. Signi…cance at the 1% level is denoted with a, the 5% level with b, and
the 10% level with c.
141
Table 3.5: Current Ownership Length Changes Around Payout Events
Panel A: Dividend Events
t ÷1 to t + 1 t + 1 to t + 2 t ÷1 to t + 2
N Mean Median N Mean Median N Mean Median
Unadjusted Changes
Increase 2067 0.099
a
0.123
a
1670 0.051
a
0.092
a
1773 0.130
a
0.158
a
(9.91) (11.16) (5.00) (8.03) (10.91) (11.86)
Decrease 478 0.061
b
0.092
a
368 0.065
a
0.122
a
416 0.079
a
0.087
a
(2.43) (3.00) (2.72) (3.92) (2.66) (2.82)
Initiation 204 0.115
a
0.096
a
169 0.064
b
0.131
a
170 0.141
a
0.144
a
(3.05) (2.98) (2.04) (3.29) (3.71) (4.08)
Omission 140 0.078 0.091
c
106 0.052 0.101 122 0.114
c
0.133
b
(1.64) (1.88) (1.24) (1.53) (1.94) (2.21)
Adjusted Changes
Increase 2067 0.010 0.013 1670 -0.005 -0.018 1773 -0.002 -0.014
(0.71) (0.46) (0.39) (0.18) (0.11) (0.35)
Decrease 478 0.047 0.062 368 -0.023 0.007 416 0.016 0.007
(1.35) (1.49) (0.75) (0.23) (0.40) (0.39)
Initiation 204 0.004 0.020 169 -0.054 0.009 170 -0.051 0.000
(0.08) (0.27) (1.20) (0.64) (0.95) (0.77)
Omission 140 0.049 -0.008 106 0.005 0.069 122 0.096 0.119
(0.78) (0.48) (0.09) (0.06) (1.32) (1.42)
Continued Next Page...
142
Panel B: Share Repurchase Events
t ÷1 to t + 1 t + 1 to t + 2 t ÷1 to t + 2
N Mean Median N Mean Median N Mean Median
Unadjusted Changes
Non-Dividend Paying Firms
All 2390 0.148
a
0.173
a
1790 0.077
a
0.102
a
1905 0.183
a
0.194
a
(15.06) (16.25) (8.09) (9.80) (15.30) (15.43)
Initiation 832 0.172
a
0.178
a
606 0.058
a
0.113
a
664 0.189
a
0.209
a
(10.12) (10.09) (3.33) (4.83) (9.17) (9.31)
Non-Init. 1558 0.135
a
0.172
a
1184 0.087
a
0.100
a
1241 0.179
a
0.184
a
(11.24) (12.75) (7.67) (8.58) (12.25) (12.30)
Dividend Paying Firms
All 2600 0.121
a
0.147
a
2075 0.047
a
0.090
a
2265 0.148
a
0.173
a
(13.70) (15.09) (5.45) (9.04) (14.24) (14.64)
Initiation 654 0.124
a
0.153
a
487 0.057
a
0.107
a
560 0.148
a
0.189
a
(6.66) (7.67) (3.05) (4.83) (6.75) (7.12)
Non-Init. 1946 0.120
a
0.145
a
1588 0.044
a
0.087
a
1705 0.148
a
0.170
a
(11.99) (13.00) (4.53) (7.67) (12.55) (12.78)
Adjusted Changes
Non-Dividend Paying Firms
All 2390 -0.001 0.003 1790 -0.016 -0.014 1905 -0.051
a
-0.074
a
(0.10) (0.25) (1.10) (1.34) (3.00) (3.08)
Initiation 832 -0.004 -0.018 606 -0.038
c
-0.017 664 -0.060
b
-0.080
c
(0.16) (0.10) (1.65) (1.44) (1.99) (1.82)
Non-Init. 1558 0.000 0.008 1184 -0.002 -0.013 1241 -0.046
b
-0.071
b
(0.01) (0.24) (0.12) (0.60) (2.25) (2.50)
Dividend Paying Firms
All 2600 0.021 0.031
c
2075 0.034
a
0.033
a
2265 0.029
b
0.022
b
(1.64) (1.76) (2.68) (2.83) (1.99) (2.18)
Initiation 654 -0.009 -0.007 487 0.049
c
0.072
b
560 0.027 0.016
(0.35) (0.56) (1.88) (2.36) (0.85) (0.75)
Non-Init. 1946 0.031
b
0.046
b
1588 0.029
b
0.023
c
1705 0.030
c
0.023
b
(2.14) (2.38) (2.02) (1.91) (1.81) (2.11)
This table reports changes in average current ownership length (COL) for payout event …rms. I
calculate both mean and median un-adjusted and adjusted changes. Test-statistics are located
below the reported change. COL is de…ned in Chapter 3.5. Unadjusted change is equal to the
di¤erence in COL between dates. Adjusted change is equal to the di¤erence in COL between
dates minus a similar change in a control …rm. Control …rms are chosen based on similar MB,
ROA, ROA, and industry classi…cation. The algorithm to match event …rms with control …rms
is de…ned in Chapter 3.5. For the di¤erence in means, I use the two-tailed t-statistic to test
sign…cance. For the di¤erence in medians I use the two-tailed z-statistic from the Wilcoxon rank
test. Signi…cance at the 1% level is denoted with a, the 5% level with b, and the 10% level with
c.
143
Table 3.6: Adjusted Changes in Ownership Proportion Around Payout Events
Panel A: Dividend Events
t ÷1 to t + 1 t + 1 to t + 2 t ÷1 to t + 2
N Mean Median N Mean Median N Mean Median
Increase
Own%S 2150 -0.003
a
-0.001
a
1761 -0.002
a
-0.002
a
1860 -0.005
a
-0.004
a
(3.57) (3.39) (2.80) (3.53) (5.97) (6.19)
Own%M 0.001 0.000 0.000 0.000 0.000 -0.001
(0.72) (0.33) (0.52) (0.47) (0.30) (0.22)
Own%L 0.001 0.001 0.000 0.000 0.001 0.001
(0.61) (0.40) (0.20) (0.11) (1.06) (1.16)
Decrease
Own%S 492 -0.002 -0.002 382 0.000 0.000 425 -0.001 -0.002
(0.94) (1.43) (0.29) (0.28) (0.61) (1.16)
Own%M -0.005
a
-0.002
b
-0.001 -0.001 -0.003 -0.002
(2.66) (2.31) (0.44) (0.84) (1.57) (1.25)
Own%L -0.007
a
-0.003
b
0.000 0.001 -0.005
c
-0.004
c
(2.83) (2.34) (0.16) (0.38) (1.66) (1.81)
Initiation
Own%S 215 -0.003 -0.004 184 0.002 0.001 180 -0.007
c
-0.006
c
(0.85) (1.06) (0.87) (0.49) (1.83) (1.76)
Own%M 0.002 0.000 0.000 -0.001 -0.001 0.000
(0.54) (0.85) (0.13) (0.11) (0.26) (0.16)
Own%L -0.004 -0.003 0.002 0.003 -0.001 0.000
(1.13) (0.63) (0.69) (1.23) (0.13) (0.57)
Omission
Own%S 152 0.000 0.000 117 0.001 -0.002 132 0.000 -0.002
(0.06) (0.19) (0.38) (0.14) (0.04) (0.12)
Own%M -0.003 -0.002 0.004 0.004 0.000 0.003
(0.87) (0.75) (1.50) (1.29) (0.06) (0.02)
Own%L -0.002 -0.002 0.001 -0.001 -0.001 -0.001
(0.46) (0.34) (0.19) (0.34) (0.35) (0.20)
Continued Next Page...
144
Panel B: Share Repurchase Events
t ÷1 to t + 1 t + 1 to t + 2 t ÷1 to t + 2
N Mean Median N Mean Median N Mean Median
Non-Dividend Paying Firms
All
Own%S 2560 -0.003
a
-0.003
a
1940 0.002
a
0.001
a
2063 0.000 0.000
(2.60) (2.66) (2.71) (2.71) (0.06) (0.50)
Own%M -0.002
b
-0.001
b
0.001 0.000 -0.001 0.000
(2.31) (1.95) (0.84) (0.33) (1.03) (0.76)
Own%L -0.002 -0.001 0.000 0.000 0.001 0.002
(1.57) (1.30) (0.33) (0.09) (0.47) (0.90)
Initiation
Own%S 880 -0.003 -0.001 651 -0.002 -0.003 711 -0.003 -0.002
c
(1.36) (1.58) (1.13) (1.32) (1.46) (1.94)
Own%M -0.002 0.000 0.002 0.001 0.000 -0.001
(0.96) (0.61) (1.25) (1.18) (0.04) (0.05)
Own%L -0.001 0.000 0.003
b
0.002
b
0.002 0.002
(0.64) (0.31) (2.23) (2.55) (1.07) (1.28)
Non-Initiation
Own%S 1680 -0.003
b
-0.003
b
1289 0.005
a
0.003
a
1352 0.002 0.001
(2.24) (2.14) (4.39) (4.35) (1.10) (0.82)
Own%M -0.003
b
-0.002
b
0.000 -0.001 -0.002 0.000
(2.17) (1.99) (0.11) (0.46) (1.27) (0.97)
Own%L -0.002 -0.001 -0.001 -0.001 0.000 0.001
(1.49) (1.38) (1.08) (1.60) (0.18) (0.18)
Dividend Paying Firms
All
Own%S 2701 -0.002
a
-0.001
a
2171 -0.001 -0.001
c
2360 -0.002
b
-0.001
a
(2.65) (2.57) (1.54) (1.81) (2.34) (2.61)
Own%M -0.001 -0.001
c
0.000 0.000 0.000 0.000
(1.28) (1.92) (0.40) (0.58) (0.13) (0.00)
Own%L -0.001 -0.001
c
-0.001 0.000 -0.002
c
-0.002
b
(1.26) (1.65) (1.07) (1.09) (1.93) (2.13)
Initiation
Own%S 690 -0.001 -0.002 513 -0.001 0.000 596 0.000 -0.001
(0.72) (1.28) (0.50) (0.51) (0.06) (1.20)
Own%M 0.001 0.000 -0.001 0.000 -0.001 0.000
(0.52) (0.40) (0.94) (0.23) (0.56) (0.30)
Own%L -0.001 0.000 -0.001 -0.001 -0.001 -0.002
(0.40) (0.24) (0.55) (0.76) (0.36) (0.73)
Non-Initiation
Own%S 2011 -0.002
a
-0.001
b
1658 -0.001 -0.001
c
1764 -0.002
a
-0.001
b
(2.63) (2.22) (1.49) (1.75) (2.65) (2.33)
Own%M -0.001
c
-0.001
b
0.001 0.000 0.000 0.000
(1.82) (2.46) (0.99) (0.79) (0.19) (0.18)
Own%L -0.001 -0.001
c
-0.001 0.000 -0.002
b
-0.002
b
(1.21) (1.76) (0.91) (0.80) (2.01) (2.04)
This table reports changes in ownership proportion by fund investment horizon tercile (Own%S,
Own%M, Own%L) for payout event …rms. I calculate both mean and median adjusted changes.
Test-statistics are located below the reported change. Adjusted change is equal to the di¤erence
in ownership between dates minus a similar change in a control …rm. Control …rms are chosen
145
based on similar MB, ROA, ROA, and industry classi…cation. The algorithm to match event
…rms with control …rms is de…ned in Chapter 3.5. For the di¤erence in means, I use the two-tailed
t-statistic to test sign…cance. For the di¤erence in medians I use the two-tailed z-statistic from
the Wilcoxon rank test. Signi…cance at the 1% level is denoted with a, the 5% level with b, and
the 10% level with c.
146
Table 3.7: The E¤ect of the JGTRRA on Fund Ownership Characteristics
Ownership Percentage Current Own. Length
Short Medium Long Short Medium Long
(1) (2) (3) (4) (5) (6)
Panel A: Dividend and Repurchase Yields
TaxPd
t
-0.005
a
0.008
a
0.007
b
-0.025
a
-0.030
a
-0.014
b
(3.80) (7.55) (5.68) (3.74) (4.17) (2.15)
DivYld
t
-0.079 -0.024 -0.040 -0.486
b
-0.117 0.659
a
(1.57) (0.76) (0.80) (2.09) (0.59) (3.66)
RepYld
t
0.032
b
0.014 0.058
c
0.261
a
0.289
a
0.090
(2.27) (1.09) (1.77) (3.62) (3.25) (1.16)
TaxPd
t
DivYld
t
0.057 0.024 0.029 0.498
b
0.354
b
-0.485
a
(1.09) (0.74) (0.61) (2.22) (1.83) (2.75)
TaxPd
t
RepYld
t
-0.010 0.007 -0.032 -0.023 -0.055 0.093
(0.65) (0.51) (0.94) (0.28) (0.53) (1.04)
Panel B: Dividend and Repurchase Indicator Variables
TaxPd
t
-0.004
a
0.008
a
0.006
a
-0.045
a
-0.038
a
-0.006
(2.87) (6.39) (3.94) (5.12) (3.96) (0.61)
DivInd
t
-0.008
a
-0.002 -0.004
c
-0.036
a
-0.009 0.038
a
(4.98) (1.01) (1.69) (3.74) (0.87) (3.29)
RepInd
t
0.001 0.000 0.004
b
0.000 0.021
b
0.026
a
(0.59) (0.26) (2.11) (0.04) (2.53) (2.90)
TaxPd
t
DivInd
t
-0.002 -0.002 0.003 0.030
a
0.023
a
-0.012
(1.05) (1.51) (1.56) (3.42) (2.36) (1.32)
TaxPd
t
RepInd
t
0.002 0.003 0.002 0.015
c
0.001 0.000
(1.46) (1.66) (0.79) (1.86) (0.12) (0.01)
Continued Next Page...
147
Panel C: Total Payout Yield
TaxPd
t
-0.004
a
0.008
a
0.008
a
-0.020
a
-0.026
a
-0.019
a
(3.60) (7.86) (5.87) (3.24) (3.76) (3.05)
TotYld
t
0.016 0.008 0.044 0.157
b
0.229
a
0.183
a
(1.36) (0.73) (1.55) (2.48) (3.08) (2.76)
TaxPd
t
TotYld
t
-0.009 0.004 -0.032 0.044 0.010 -0.008
(0.74) (0.33) (1.12) (0.63) (0.12) (0.12)
Panel D: Payout Indicator Variable
TaxPd
t
-0.005
a
0.006
a
0.005
b
-0.043
a
-0.032
a
-0.011
(2.98) (4.18) (2.57) (4.68) (3.25) (1.07)
PayInd
t
-0.004
b
-0.003
c
0.001 -0.025
a
0.008 0.036
a
(2.35) (1.78) (0.39) (3.06) (0.96) (3.63)
TaxPd
t
PayInd
t
0.001 0.003 0.004
b
0.031
a
0.010 -0.005
(0.38) (1.56) (2.08) (3.48) (1.13) (0.52)
This table reports estimates from tobit regressions explaining changes in aggregate
fund ownership and truncated regressions explaining changes in relative ownership
length by fund investment horizon tercile before and after the JGTRRA. Dependent
variables are de…ned in Chapter 3.4. For each speci…cation, one regression with panel
data from 2002 and 2004 is estimated using one of four sets of payout variables:
dividend and repurchase yields, dividend and repurchase indicator variables, total
payout yield, and the payout indicator variable. t-statistics are in parentheses, with
standard errors clustered at the …rm level. Other independent variables include
a tax-period indicator variable, interaction terms between the tax-period indicator
variable and included payout variables, other …rm level variables, and industry …xed
e¤ects. Firm speci…c variables are de…ned in Chapter 3.4. Signi…cance at the 1%
level is denoted with a, the 5% level with b, and the 10% level with c. To conserve
space, only tax-period and payout related variables are presented.
148
Table 3.8: The Di¤erence in Shareholder Investment Horizon Changes
Before & After the JGTRRA (Before - After)
Di¤. Mean Di¤. Med.
# Obs. Change Change
Before After SIH t-stat. SIH z-stat.
Panel A: Dividend Events
Increase t ÷1 to t + 1 204 638 0.015 (0.76) -0.020 (0.22)
t + 1 to t + 2 118 396 -0.011 (0.62) -0.011 (0.83)
t ÷1 to t + 2 129 381 -0.023 (0.87) -0.016 (1.02)
Decrease t ÷1 to t + 1 88 98 -0.049 (1.10) -0.028 (1.43)
t + 1 to t + 2 37 62 -0.013 (0.34) 0.004 (0.11)
t ÷1 to t + 2 51 61 -0.056 (1.11) 0.015 (0.62)
Inititation t ÷1 to t + 1 11 89 0.020 (0.20) -0.027 (0.21)
t + 1 to t + 2 4 70 -0.055 (0.63) 0.037 (0.36)
t ÷1 to t + 2 4 64 0.048 (0.36) 0.007 (0.34)
Omission t ÷1 to t + 1 29 32 -0.113 (1.57) -0.111
c
(1.86)
t + 1 to t + 2 15 13 -0.209
b
(2.52) -0.200
b
(2.19)
t ÷1 to t + 2 16 14 -0.184 (1.38) -0.255 (1.58)
Continued Next Page...
149
Panel B: Share Repurchase Events
Di¤. Mean Di¤. Med.
# Obs. Change Change
Before After SIH t-stat. SIH z-stat.
Non-Dividend Paying Firms
All t ÷1 to t + 1 588 969 0.000 (0.01) -0.008 (0.09)
t + 1 to t + 2 362 572 -0.012 (0.81) 0.000 (0.32)
t ÷1 to t + 2 413 560 -0.034 (1.62) -0.068
b
(2.09)
Inititation t ÷1 to t + 1 169 315 -0.024 (0.82) -0.028 (0.75)
t + 1 to t + 2 111 193 0.014 (0.52) 0.007 (0.47)
t ÷1 to t + 2 131 184 0.006 (0.14) -0.032 (0.24)
Non-Inititation t ÷1 to t + 1 419 654 0.011 (0.59) -0.002 (0.64)
t + 1 to t + 2 251 379 -0.023 (1.30) -0.007 (0.57)
t ÷1 to t + 2 282 376 -0.052
b
(2.12) -0.076
b
(2.32)
Dividend Paying Firms
All t ÷1 to t + 1 647 541 0.001 (0.04) 0.000 (0.32)
t + 1 to t + 2 435 281 0.016 (1.13) 0.027 (1.57)
t ÷1 to t + 2 495 274 0.003 (0.13) -0.001 (0.60)
Inititation t ÷1 to t + 1 110 168 0.032 (1.11) 0.057
c
(1.78)
t + 1 to t + 2 69 99 0.034 (1.12) 0.045 (1.32)
t ÷1 to t + 2 93 96 0.026 (0.68) 0.022 (1.05)
Non-Inititation t ÷1 to t + 1 537 373 -0.012 (0.76) -0.015 (0.78)
t + 1 to t + 2 366 182 0.003 (0.20) 0.024 (0.88)
t ÷1 to t + 2 402 178 -0.016 (0.59) -0.009 (0.22)
This table reports the di¤erence in average shareholder investment horizon (SIH) changes as
the result of payout events before and after the JGTRRA. I calculate the di¤erence in mean
and median adjusted changes. Test-statistics are located below the reported change. SIH
is de…ned in Chapter 3.5. Adjusted change is equal to the di¤erence in SIH between dates
minus a similar change in a control …rm. Control …rms are chosen based on similar MB,
ROA, ROA, and industry classi…cation. The algorithm to match event …rms with control
…rms is de…ned in Chapter 3.5. For the di¤erence in means, I use the two-tailed t-statistic
to test sign…cance. For the di¤erence in medians I use the two-tailed z-statistic from the
Wilcoxon rank-sum test. Signi…cance at the 1% level is denoted with a, the 5% level with b,
and the 10% level with c.
150
Table 3.9: The Di¤erence in Current Ownership Length Changes Before
& After the JGTRRA (Before - After)
Di¤. Mean Di¤. Med.
# Obs. Change Change
Before After COL t-stat. COL z-stat.
Panel A: Dividend Events
Increase t ÷1 to t + 1 196 623 0.032 (0.61) 0.013 (0.56)
t + 1 to t + 2 112 393 0.130
b
(2.15) 0.103
b
(1.96)
t ÷1 to t + 2 122 371 0.001 (0.01) 0.024 (0.04)
Decrease t ÷1 to t + 1 89 96 0.017 (0.14) -0.114 (0.23)
t + 1 to t + 2 35 61 0.129 (1.21) 0.272 (1.36)
t ÷1 to t + 2 43 59 0.000 (0.00) 0.030 (0.11)
Inititation t ÷1 to t + 1 11 86 0.397
c
(1.75) 0.241 (1.18)
t + 1 to t + 2 4 67 0.096 (0.34) 0.078 (0.30)
t ÷1 to t + 2 4 61 0.320 (0.91) 0.568 (1.09)
Omission t ÷1 to t + 1 25 30 0.177 (0.99) 0.223 (0.96)
t + 1 to t + 2 14 13 -0.519
b
(2.33) -0.407
c
(1.84)
t ÷1 to t + 2 14 14 -0.159 (0.52) -0.312 (0.78)
Continued Next Page...
151
Panel B: Share Repurchase Events
Di¤. Mean Di¤. Med.
# Obs. Change Change
Before After COL t-stat. COL z-stat.
Non-Dividend Paying Firms
All t ÷1 to t + 1 547 919 -0.022 (0.62) -0.057 (1.31)
t + 1 to t + 2 331 374 -0.047 (1.12) -0.022 (1.17)
t ÷1 to t + 2 379 516 -0.051 (1.05) -0.013 (1.29)
Inititation t ÷1 to t + 1 160 312 0.001 (0.02) -0.031 (0.20)
t + 1 to t + 2 101 191 -0.089 (1.41) -0.055 (1.49)
t ÷1 to t + 2 123 182 -0.004 (0.04) -0.036 (0.26)
Non-Inititation t ÷1 to t + 1 387 607 -0.032 (0.77) -0.063 (1.43)
t + 1 to t + 2 230 183 -0.038 (0.66) 0.009 (0.57)
t ÷1 to t + 2 256 334 -0.075 (1.30) -0.004 (1.33)
Dividend Paying Firms
All t ÷1 to t + 1 626 537 -0.028 (0.78) 0.006 (0.52)
t + 1 to t + 2 413 280 0.031 (0.68) -0.004 (0.02)
t ÷1 to t + 2 476 271 -0.040 (0.77) 0.017 (0.95)
Inititation t ÷1 to t + 1 104 167 0.264
a
(3.40) 0.267
a
(3.49)
t + 1 to t + 2 63 98 -0.108 (1.21) -0.018 (0.89)
t ÷1 to t + 2 87 95 0.079 (0.67) 0.029 (0.55)
Non-Inititation t ÷1 to t + 1 522 370 -0.113
a
(2.73) -0.078
b
(2.40)
t + 1 to t + 2 350 182 0.067 (1.24) 0.006 (0.57)
t ÷1 to t + 2 389 176 -0.065 (1.08) 0.020 (1.11)
This table reports the di¤erence in average current ownership length (COL) changes as the
result of payout events before and after the JGTRRA. I calculate the di¤erence in mean
and median adjusted changes. Test-statistics are located below the reported change. COL
is de…ned in Chapter 3.5. Adjusted change is equal to the di¤erence in COL between dates
minus a similar change in a control …rm. Control …rms are chosen based on similar MB, ROA,
ROA, and industry classi…cation. The algorithm to match event …rms with control …rms
is de…ned in Chapter 3.5. For the di¤erence in means, I use the two-tailed t-statistic to test
sign…cance. For the di¤erence in medians I use the two-tailed z-statistic from the Wilcoxon
rank-sum test. Signi…cance at the 1% level is denoted with a, the 5% level with b, and the
10% level with c.
152
T
a
b
l
e
3
.
1
0
:
T
h
e
D
i
¤
e
r
e
n
c
e
i
n
O
w
n
e
r
s
h
i
p
P
r
o
p
o
r
t
i
o
n
C
h
a
n
g
e
s
B
e
f
o
r
e
&
A
f
t
e
r
t
h
e
J
G
T
R
R
A
(
B
e
f
o
r
e
-
A
f
t
e
r
)
b
y
F
u
n
d
I
n
v
e
s
t
m
e
n
t
H
o
r
i
z
o
n
T
e
r
c
i
l
e
P
a
n
e
l
A
:
D
i
v
i
d
e
n
d
E
v
e
n
t
s
t
÷
1
t
o
t
+
1
t
+
1
t
o
t
+
1
t
÷
1
t
o
t
+
2

C
h
a
n
g
e
M
e
a
n
M
e
d
i
a
n
M
e
a
n
M
e
d
i
a
n
M
e
a
n
M
e
d
i
a
n
I
n
c
r
e
a
s
e
O
w
n
%
S
-
0
.
0
0
5
b
(
2
.
2
9
)
-
0
.
0
0
5
b
(
2
.
1
5
)
0
.
0
0
4
c
(
1
.
7
1
)
-
0
.
0
0
2
(
0
.
0
1
)
-
0
.
0
0
1
(
0
.
1
5
)
-
0
.
0
0
3
(
0
.
4
3
)
O
w
n
%
M
-
0
.
0
1
0
(
1
.
4
8
)
0
.
0
0
7
b
(
2
.
0
6
)
0
.
0
0
3
(
1
.
0
0
)
0
.
0
0
3
(
1
.
2
1
)
0
.
0
1
5
a
(
3
.
8
2
)
0
.
0
1
8
a
(
3
.
9
6
)
O
w
n
%
L
-
0
.
0
0
5
(
0
.
0
7
)
-
0
.
0
0
4
(
0
.
3
7
)
-
0
.
0
0
3
(
0
.
6
5
)
-
0
.
0
0
4
(
1
.
3
9
)
0
.
0
0
0
(
0
.
0
4
)
0
.
0
0
0
(
0
.
1
7
)
D
e
c
r
e
a
s
e
O
w
n
%
S
0
.
0
0
1
(
0
.
1
4
)
0
.
0
0
1
(
0
.
8
2
)
0
.
0
0
3
(
0
.
4
6
)
-
0
.
0
0
1
(
0
.
2
1
)
0
.
0
0
1
(
0
.
2
3
)
0
.
0
0
3
(
0
.
6
3
)
O
w
n
%
M
-
0
.
0
1
0
(
1
.
6
3
)
-
0
.
0
1
0
(
1
.
2
7
)
0
.
0
0
0
(
0
.
0
6
)
0
.
0
0
1
(
0
.
2
8
)
-
0
.
0
0
9
(
1
.
1
2
)
-
0
.
0
0
2
(
0
.
5
5
)
O
w
n
%
L
-
0
.
0
2
4
b
(
2
.
4
4
)
-
0
.
0
1
4
b
(
2
.
1
4
)
-
0
.
0
0
2
(
0
.
2
7
)
-
0
.
0
0
1
(
0
.
5
5
)
-
0
.
0
2
8
b
(
2
.
2
0
)
-
0
.
0
2
3
b
(
2
.
1
8
)
I
n
i
t
i
a
t
i
o
n
O
w
n
%
S
0
.
0
0
7
(
0
.
3
8
)
-
0
.
0
0
9
(
0
.
2
7
)
0
.
0
0
8
(
0
.
5
2
)
0
.
0
0
4
(
0
.
3
3
)
-
0
.
0
0
1
(
0
.
0
2
)
-
0
.
0
2
1
(
0
.
3
6
)
O
w
n
%
M
-
0
.
0
0
3
(
0
.
2
1
)
-
0
.
0
1
0
(
0
.
0
1
)
-
0
.
0
2
7
c
(
1
.
6
7
)
-
0
.
0
0
9
(
0
.
9
3
)
-
0
.
0
3
7
(
1
.
4
5
)
-
0
.
0
3
7
(
0
.
9
6
)
O
w
n
%
L
-
0
.
0
4
9
b
(
2
.
2
8
)
-
0
.
0
1
3
(
0
.
8
8
)
0
.
0
0
6
(
0
.
2
4
)
0
.
0
0
1
(
0
.
3
6
)
-
0
.
1
0
6
a
(
2
.
7
3
)
-
0
.
0
0
1
(
0
.
6
3
)
O
m
i
s
s
i
o
n
O
w
n
%
S
-
0
.
0
1
2
(
1
.
3
6
)
-
0
.
0
0
1
(
0
.
8
5
)
0
.
0
0
5
(
0
.
4
4
)
-
0
.
0
0
2
(
0
.
1
2
)
0
.
0
0
2
(
0
.
1
4
)
0
.
0
1
7
(
0
.
6
2
)
O
w
n
%
M
0
.
0
0
4
(
0
.
3
5
)
-
0
.
0
0
6
(
0
.
3
6
)
-
0
.
0
0
2
(
0
.
1
1
)
-
0
.
0
0
3
(
0
.
2
5
)
-
0
.
0
2
3
(
1
.
0
3
)
-
0
.
0
2
7
(
1
.
4
1
)
O
w
n
%
L
-
0
.
0
1
9
c
(
1
.
7
1
)
-
0
.
0
1
1
(
1
.
1
6
)
-
0
.
0
0
9
(
0
.
8
2
)
-
0
.
0
0
5
(
1
.
2
2
)
-
0
.
0
2
2
(
1
.
2
3
)
-
0
.
0
2
6
(
1
.
3
3
)
C
o
n
t
i
n
u
e
d
N
e
x
t
P
a
g
e
.
.
.
153
P
a
n
e
l
B
:
S
h
a
r
e
R
e
p
u
r
c
h
a
s
e
E
v
e
n
t
s
t
÷
1
t
o
t
+
1
t
+
1
t
o
t
+
1
t
÷
1
t
o
t
+
2

C
h
a
n
g
e
M
e
a
n
M
e
d
i
a
n
M
e
a
n
M
e
d
i
a
n
M
e
a
n
M
e
d
i
a
n
N
o
n
-
D
i
v
i
d
e
n
d
P
a
y
i
n
g
F
i
r
m
s
A
l
l
O
w
n
%
S
0
.
0
0
7
b
(
2
.
3
9
)
0
.
0
0
3
c
(
1
.
7
4
)
0
.
0
0
4
(
1
.
6
4
)
0
.
0
0
1
(
1
.
1
8
)
0
.
0
0
6
c
(
1
.
7
8
)
0
.
0
0
8
b
(
2
.
3
8
)
O
w
n
%
M
0
.
0
0
0
(
0
.
0
1
)
-
0
.
0
0
2
(
0
.
6
7
)
0
.
0
0
9
a
(
4
.
1
3
)
0
.
0
0
5
a
(
3
.
7
3
)
0
.
0
0
7
c
(
1
.
8
9
)
0
.
0
0
6
(
1
.
4
3
)
O
w
n
%
L
0
.
0
0
6
b
(
2
.
1
2
)
0
.
0
0
5
b
(
2
.
1
1
)
-
0
.
0
0
3
(
1
.
2
8
)
-
0
.
0
0
2
(
1
.
5
4
)
0
.
0
0
2
(
0
.
4
9
)
-
0
.
0
0
4
(
0
.
2
3
)
I
n
i
t
i
a
t
i
o
n
O
w
n
%
S
0
.
0
1
4
a
(
2
.
6
5
)
0
.
0
0
2
(
1
.
6
5
)
0
.
0
1
2
b
(
2
.
4
9
)
0
.
0
0
7
b
(
2
.
5
5
)
0
.
0
1
6
b
(
2
.
4
6
)
0
.
0
1
4
a
(
2
.
6
8
)
O
w
n
%
M
0
.
0
0
0
(
0
.
0
1
)
-
0
.
0
0
2
(
0
.
0
9
)
0
.
0
1
2
a
(
2
.
9
1
)
0
.
0
0
8
a
(
2
.
6
8
)
0
.
0
0
4
(
0
.
7
1
)
0
.
0
0
5
(
0
.
5
4
)
O
w
n
%
L
-
0
.
0
0
4
(
0
.
8
1
)
-
0
.
0
0
4
(
0
.
5
6
)
-
0
.
0
0
1
(
0
.
1
4
)
-
0
.
0
0
2
(
0
.
2
2
)
-
0
.
0
0
2
(
0
.
2
5
)
-
0
.
0
0
4
(
0
.
1
2
)
N
o
n
-
I
n
i
t
i
a
t
i
o
n
O
w
n
%
S
0
.
0
0
4
(
1
.
1
5
)
0
.
0
0
3
(
1
.
0
0
)
0
.
0
0
1
(
0
.
1
9
)
-
0
.
0
0
2
(
0
.
3
3
)
0
.
0
0
2
(
0
.
4
1
)
0
.
0
0
6
(
1
.
0
5
)
O
w
n
%
M
0
.
0
0
0
(
0
.
0
7
)
-
0
.
0
0
3
(
0
.
6
8
)
0
.
0
0
8
a
(
3
.
0
3
)
0
.
0
0
4
a
(
2
.
6
9
)
0
.
0
0
8
c
(
1
.
8
2
)
0
.
0
0
8
(
1
.
4
1
)
O
w
n
%
L
0
.
0
1
1
a
(
3
.
0
1
)
0
.
0
0
8
a
(
2
.
8
7
)
-
0
.
0
0
5
(
1
.
3
5
)
-
0
.
0
0
4
(
1
.
5
7
)
0
.
0
0
4
(
0
.
7
8
)
-
0
.
0
0
4
(
0
.
1
5
)
D
i
v
i
d
e
n
d
-
P
a
y
i
n
g
F
i
r
m
s
A
l
l
O
w
n
%
S
0
.
0
0
0
(
0
.
2
1
)
0
.
0
0
0
(
0
.
4
0
)
0
.
0
0
0
(
0
.
0
2
)
-
0
.
0
0
1
(
0
.
2
8
)
0
.
0
0
2
(
0
.
7
4
)
0
.
0
0
1
(
0
.
6
8
)
O
w
n
%
M
-
0
.
0
0
1
(
0
.
2
7
)
0
.
0
0
1
(
0
.
1
2
)
-
0
.
0
0
2
(
1
.
2
0
)
-
0
.
0
0
1
(
0
.
8
4
)
-
0
.
0
0
2
(
0
.
5
1
)
-
0
.
0
0
3
(
0
.
9
4
)
O
w
n
%
L
-
0
.
0
0
5
c
(
1
.
7
3
)
0
.
0
0
1
(
0
.
3
5
)
-
0
.
0
0
4
(
1
.
3
9
)
-
0
.
0
0
3
c
(
1
.
9
3
)
0
.
0
0
0
(
0
.
0
6
)
0
.
0
0
3
(
0
.
6
8
)
I
n
i
t
i
a
t
i
o
n
O
w
n
%
S
-
0
.
0
0
6
c
(
1
.
6
6
)
-
0
.
0
0
4
b
(
2
.
2
5
)
-
0
.
0
0
1
(
0
.
3
0
)
-
0
.
0
0
1
(
0
.
6
9
)
-
0
.
0
0
5
(
1
.
1
9
)
-
0
.
0
0
3
(
1
.
0
9
)
O
w
n
%
M
-
0
.
0
0
1
(
0
.
1
3
)
0
.
0
0
3
(
0
.
5
2
)
-
0
.
0
0
5
(
1
.
1
6
)
-
0
.
0
0
1
(
0
.
8
1
)
-
0
.
0
0
5
(
0
.
8
4
)
0
.
0
0
0
(
0
.
4
9
)
O
w
n
%
L
0
.
0
0
3
(
0
.
5
1
)
0
.
0
0
2
(
0
.
4
1
)
-
0
.
0
0
2
(
0
.
3
9
)
-
0
.
0
0
8
(
1
.
0
4
)
0
.
0
0
3
(
0
.
3
6
)
0
.
0
0
6
(
1
.
2
5
)
N
o
n
-
I
n
i
t
i
a
t
i
o
n
O
w
n
%
S
0
.
0
0
2
(
1
.
2
7
)
0
.
0
0
1
c
(
1
.
7
4
)
0
.
0
0
0
(
0
.
1
2
)
0
.
0
0
0
(
0
.
0
5
)
0
.
0
0
5
c
(
1
.
6
8
)
0
.
0
0
4
(
1
.
4
9
)
O
w
n
%
M
0
.
0
0
0
(
0
.
0
3
)
0
.
0
0
1
(
0
.
0
3
)
-
0
.
0
0
1
(
0
.
5
9
)
-
0
.
0
0
1
(
0
.
3
6
)
0
.
0
0
0
(
0
.
0
8
)
-
0
.
0
0
3
(
0
.
7
0
)
O
w
n
%
L
-
0
.
0
0
8
b
(
2
.
2
9
)
0
.
0
0
1
(
0
.
7
4
)
-
0
.
0
0
5
(
1
.
5
8
)
-
0
.
0
0
3
c
(
1
.
9
2
)
-
0
.
0
0
2
(
0
.
4
1
)
-
0
.
0
0
2
(
0
.
1
2
)
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
o
w
n
e
r
s
h
i
p
p
r
o
p
o
r
t
i
o
n
b
y
f
u
n
d
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
t
e
r
c
i
l
e
(
O
w
n
%
S
,
O
w
n
%
M
,
O
w
n
%
L
)
c
h
a
n
g
e
s
a
s
t
h
e
r
e
s
u
l
t
o
f
p
a
y
o
u
t
e
v
e
n
t
s
b
e
f
o
r
e
a
n
d
a
f
t
e
r
t
h
e
J
G
T
R
R
A
.
I
c
a
l
c
u
l
a
t
e
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
m
e
a
n
a
n
d
m
e
d
i
a
n
a
d
j
u
s
t
e
d
c
h
a
n
g
e
s
.
T
e
s
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
l
o
c
a
t
e
d
b
e
l
o
w
t
h
e
r
e
p
o
r
t
e
d
c
h
a
n
g
e
.
T
h
e
p
r
o
p
o
r
t
i
o
n
o
f
f
u
n
d
o
w
n
e
r
s
h
i
p
i
s
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
3
.
4
.
A
d
j
u
s
t
e
d
c
h
a
n
g
e
i
s
e
q
u
a
l
t
o
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
o
w
n
e
r
s
h
i
p
p
r
o
p
o
r
t
i
o
n
b
e
t
w
e
e
n
d
a
t
e
s
m
i
n
u
s
a
s
i
m
i
l
a
r
c
h
a
n
g
e
i
n
a
c
o
n
t
r
o
l
…
r
m
.
C
o
n
t
r
o
l
…
r
m
s
a
r
e
c
h
o
s
e
n
b
a
s
e
d
o
n
s
i
m
i
l
a
r
M
B
,
R
O
A
,

R
O
A
,
a
n
d
i
n
d
u
s
t
r
y
c
l
a
s
s
i
…
c
a
t
i
o
n
(
s
e
e
C
h
a
p
t
e
r
3
.
5
)
.
F
o
r
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
m
e
a
n
s
,
I
u
s
e
t
h
e
t
w
o
-
t
a
i
l
e
d
t
-
s
t
a
t
i
s
t
i
c
t
o
t
e
s
t
s
i
g
n
…
c
a
n
c
e
.
F
o
r
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
m
e
d
i
a
n
s
I
u
s
e
t
h
e
t
w
o
-
t
a
i
l
e
d
z
-
s
t
a
t
i
s
t
i
c
f
r
o
m
t
h
e
W
i
l
c
o
x
o
n
r
a
n
k
-
s
u
m
t
e
s
t
.
S
i
g
n
i
…
c
a
n
c
e
a
t
t
h
e
1
%
l
e
v
e
l
i
s
d
e
n
o
t
e
d
w
i
t
h
a
,
t
h
e
5
%
l
e
v
e
l
w
i
t
h
b
,
a
n
d
t
h
e
1
0
%
l
e
v
e
l
w
i
t
h
c
.
154
T
a
b
l
e
3
.
1
1
:
F
u
n
d
O
w
n
e
r
s
h
i
p
C
o
m
p
a
r
i
s
o
n
s
A
r
o
u
n
d
D
i
v
i
d
e
n
d
I
n
c
r
e
a
s
e
s
&
S
h
a
r
e
R
e
p
u
r
c
h
a
s
e
s
D
i
v
i
d
e
n
d
I
n
c
r
e
a
s
e
S
h
a
r
e
R
e
p
u
r
c
h
a
s
e
M
e
a
n
D
i
¤
.
M
e
d
i
a
n
D
i
¤
.
S
t
a
t
.
T
i
m
e
P
e
r
i
o
d
N
M
e
a
n
M
e
d
i
a
n
N
M
e
a
n
M
e
d
i
a
n

t
-
s
t
a
t
.

z
-
v
a
l
.
P
a
n
e
l
A
:
S
h
a
r
e
h
o
l
d
e
r
I
n
v
e
s
t
m
e
n
t
H
o
r
i
z
o
n
M
e
a
n
t
÷
1
1
8
0
6
3
.
0
4
1
3
.
0
4
6
2
4
5
8
3
.
1
1
4
3
.
1
3
2
-
0
.
0
7
3
a
(
1
0
.
2
4
)
-
0
.
0
8
7
a
(
1
0
.
5
0
)
C
h
a
n
g
e
t
÷
1
t
o
t
+
1
1
8
0
6
0
.
0
4
3
0
.
0
2
5
2
4
5
8
0
.
0
1
3
0
.
0
0
9
0
.
0
3
1
a
(
4
.
6
3
)
0
.
0
1
7
a
(
3
.
6
6
)
C
h
a
n
g
e
t
+
1
t
o
t
+
2
1
4
4
9
0
.
0
1
8
0
.
0
1
3
1
9
5
6
0
.
0
0
3
0
.
0
0
2
0
.
0
1
5
a
(
2
.
6
2
)
0
.
0
1
0
b
(
2
.
2
4
)
C
h
a
n
g
e
t
÷
1
t
o
t
+
2
1
5
2
1
0
.
0
5
6
0
.
0
4
9
2
1
2
2
0
.
0
1
4
0
.
0
0
6
0
.
0
4
2
a
(
5
.
2
6
)
0
.
0
4
3
a
(
5
.
4
7
)
P
a
n
e
l
B
:
C
u
r
r
e
n
t
O
w
n
e
r
s
h
i
p
L
e
n
g
t
h
M
e
a
n
t
÷
1
1
7
5
8
2
.
8
6
8
2
.
8
6
9
2
3
8
1
2
.
9
3
2
2
.
9
4
1
-
0
.
0
6
5
a
(
4
.
5
8
)
-
0
.
0
7
2
a
(
5
.
1
3
)
C
h
a
n
g
e
t
÷
1
t
o
t
+
1
1
7
5
8
0
.
0
8
6
0
.
1
0
7
2
3
8
1
0
.
1
1
6
0
.
1
4
5
-
0
.
0
3
0
b
(
2
.
1
5
)
-
0
.
0
3
8
b
(
2
.
3
6
)
C
h
a
n
g
e
t
+
1
t
o
t
+
2
1
3
8
1
0
.
0
4
5
0
.
0
8
9
1
8
8
4
0
.
0
4
6
0
.
0
8
9
-
0
.
0
0
1
(
0
.
0
5
)
0
.
0
0
1
(
0
.
0
3
)
C
h
a
n
g
e
t
÷
1
t
o
t
+
2
1
4
6
8
0
.
1
1
5
0
.
1
4
5
2
0
5
1
0
.
1
4
5
0
.
1
6
8
-
0
.
0
3
0
c
(
1
.
7
4
)
-
0
.
0
2
3
c
(
1
.
7
4
)
C
o
n
t
i
n
u
e
d
N
e
x
t
P
a
g
e
.
.
.
155
P
a
n
e
l
C
:
O
w
n
e
r
s
h
i
p
P
e
r
c
e
n
t
a
g
e
D
i
v
i
d
e
n
d
I
n
c
r
e
a
s
e
S
h
a
r
e
R
e
p
u
r
c
h
a
s
e
M
e
a
n
D
i
¤
.
M
e
d
i
a
n
D
i
¤
.
S
t
a
t
.
T
i
m
e
P
e
r
i
o
d
N
M
e
a
n
M
e
d
i
a
n
N
M
e
a
n
M
e
d
i
a
n

t
-
s
t
a
t
.

z
-
v
a
l
.
O
w
n
%
S
M
e
a
n
t
÷
1
1
8
0
6
0
.
0
2
9
0
.
0
2
1
2
4
5
8
0
.
0
2
2
0
.
0
1
4
0
.
0
0
7
a
(
8
.
6
8
)
0
.
0
0
6
a
(
8
.
5
2
)
C
h
a
n
g
e
t
÷
1
t
o
t
+
1
1
8
0
6
-
0
.
0
0
6
-
0
.
0
0
3
2
4
5
8
-
0
.
0
0
2
-
0
.
0
0
1
-
0
.
0
0
4
a
(
5
.
5
7
)
-
0
.
0
0
2
a
(
5
.
5
9
)
C
h
a
n
g
e
t
+
1
t
o
t
+
2
1
4
4
9
-
0
.
0
0
3
-
0
.
0
0
1
1
9
5
6
-
0
.
0
0
1
0
.
0
0
0
-
0
.
0
0
2
b
(
2
.
4
3
)
-
0
.
0
0
1
a
(
3
.
1
6
)
C
h
a
n
g
e
t
÷
1
t
o
t
+
2
1
5
2
1
-
0
.
0
0
9
-
0
.
0
0
4
2
1
2
2
-
0
.
0
0
3
-
0
.
0
0
1
-
0
.
0
0
6
a
(
6
.
7
6
)
-
0
.
0
0
3
a
(
6
.
6
2
)
O
w
n
%
M
M
e
a
n
t
÷
1
1
8
0
6
0
.
0
3
4
0
.
0
2
6
2
4
5
8
0
.
0
3
6
0
.
0
2
8
-
0
.
0
0
2
b
(
2
.
1
5
)
-
0
.
0
0
1
c
(
1
.
9
4
)
C
h
a
n
g
e
t
÷
1
t
o
t
+
1
1
8
0
6
-
0
.
0
0
3
-
0
.
0
0
2
2
4
5
8
-
0
.
0
0
4
-
0
.
0
0
2
0
.
0
0
2
b
(
2
.
0
8
)
0
.
0
0
0
(
1
.
2
8
)
C
h
a
n
g
e
t
+
1
t
o
t
+
2
1
4
4
9
-
0
.
0
0
1
0
.
0
0
0
1
9
5
6
-
0
.
0
0
2
-
0
.
0
0
1
0
.
0
0
1
c
(
1
.
9
2
)
0
.
0
0
0
(
0
.
8
0
)
C
h
a
n
g
e
t
÷
1
t
o
t
+
2
1
5
2
1
-
0
.
0
0
4
-
0
.
0
0
3
2
1
2
2
-
0
.
0
0
6
-
0
.
0
0
4
0
.
0
0
2
b
(
1
.
9
6
)
0
.
0
0
1
c
(
1
.
8
0
)
O
w
n
%
L
M
e
a
n
t
÷
1
1
8
0
6
0
.
0
4
9
0
.
0
3
4
2
4
5
8
0
.
0
5
7
0
.
0
4
4
-
0
.
0
0
8
a
(
5
.
1
7
)
-
0
.
0
0
9
a
(
6
.
4
0
)
C
h
a
n
g
e
t
÷
1
t
o
t
+
1
1
8
0
6
0
.
0
0
3
0
.
0
0
2
2
4
5
8
-
0
.
0
0
2
0
.
0
0
0
0
.
0
0
5
a
(
5
.
0
8
)
0
.
0
0
2
a
(
4
.
8
5
)
C
h
a
n
g
e
t
+
1
t
o
t
+
2
1
4
4
9
0
.
0
0
1
0
.
0
0
0
1
9
5
6
0
.
0
0
0
0
.
0
0
0
0
.
0
0
0
(
0
.
4
4
)
0
.
0
0
0
(
0
.
1
6
)
C
h
a
n
g
e
t
÷
1
t
o
t
+
2
1
5
2
1
0
.
0
0
3
0
.
0
0
1
2
1
2
2
-
0
.
0
0
3
-
0
.
0
0
1
0
.
0
0
6
a
(
4
.
9
3
)
0
.
0
0
2
a
(
4
.
1
6
)
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
s
h
a
r
e
h
o
l
d
e
r
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
(
S
I
H
)
,
c
u
r
r
e
n
t
o
w
n
e
r
s
h
i
p
l
e
n
g
t
h
(
C
O
L
)
,
a
n
d
t
h
e
p
r
o
p
o
r
t
i
o
n
o
f
o
w
n
e
r
s
h
i
p
b
y
f
u
n
d
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
t
e
r
c
i
l
e
(
O
w
n
%
S
,
O
w
n
%
M
,
O
w
n
%
L
)
c
h
a
n
g
e
s
b
e
t
w
e
e
n
d
i
v
i
d
e
n
d
i
n
c
r
e
a
s
e
s
a
n
d
s
h
a
r
e
r
e
p
u
r
c
h
a
s
e
s
o
f
d
i
v
i
d
e
n
d
p
a
y
i
n
g
…
r
m
s
.
D
i
¤
e
r
e
n
c
e
s
a
r
e
t
a
k
e
n
a
r
o
u
n
d
e
v
e
n
t
y
e
a
r
t
f
r
o
m
t
÷
1
t
o
t
+
1
,
t
+
1
t
o
t
+
2
,
a
n
d
t
÷
1
t
o
t
+
2
.
I
c
a
l
c
u
l
a
t
e
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
m
e
a
n
a
n
d
m
e
d
i
a
n
u
n
a
d
j
u
s
t
e
d
c
h
a
n
g
e
s
.
T
e
s
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
l
o
c
a
t
e
d
b
e
l
o
w
t
h
e
r
e
p
o
r
t
e
d
c
h
a
n
g
e
.
T
h
e
p
r
o
p
o
r
t
i
o
n
o
f
o
w
n
e
r
s
h
i
p
b
y
f
u
n
d
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
t
e
r
c
i
l
e
i
s
d
e
…
n
e
d
i
n
S
e
c
t
i
o
n
3
.
4
,
a
n
d
S
I
H
a
n
d
C
O
L
a
r
e
d
e
…
n
e
d
i
n
S
e
c
t
i
o
n
3
.
5
.
F
o
r
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
m
e
a
n
s
,
I
u
s
e
t
h
e
t
w
o
-
t
a
i
l
e
d
t
-
s
t
a
t
i
s
t
i
c
t
o
t
e
s
t
s
i
g
n
…
c
a
n
c
e
.
F
o
r
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
m
e
d
i
a
n
s
I
u
s
e
t
h
e
t
w
o
-
t
a
i
l
e
d
z
-
s
t
a
t
i
s
t
i
c
f
r
o
m
t
h
e
W
i
l
c
o
x
o
n
r
a
n
k
-
s
u
m
t
e
s
t
.
S
i
g
n
i
…
c
a
n
c
e
a
t
t
h
e
1
%
l
e
v
e
l
i
s
d
e
n
o
t
e
d
w
i
t
h
a
,
t
h
e
5
%
l
e
v
e
l
w
i
t
h
b
,
a
n
d
t
h
e
1
0
%
l
e
v
e
l
w
i
t
h
c
.
156
Table 3.12: Bivariate Probit Models Describing Payout Choice
Panel A: Shareholder Investment Horizon & Current Ownership Length
Div. Increase Share Rep. Div. Increase Share Rep.
Est. MFX Est. MFX Est. MFX Est. MFX
(1) (2) (3) (4) (5) (6) (7) (8)
SIH
t1
-0.365
b
-0.129 0.109 0.028
(2.25) (1.49)
COL
t1
0.059 0.019 0.025 0.012
(1.41) (0.60)
AnnRet
t1
0.723
a
0.244 -0.537
a
-0.164 0.815
a
0.279 -0.539
a
-0.158
(5.85) (5.53) (6.32) (5.28)
Beta
t1
0.129
c
0.042 -0.055 -0.020 0.123
c
0.040 -0.091 -0.034
(1.79) (1.22) (2.01) (1.71)
SDRet
t1
38.704
a
13.322 -34.108
a
-10.885 44.205
a
15.210 -32.329
a
-9.730
(4.66) (5.47) (5.02) (5.25)
Vol
t1
-2.041
c
-0.790 1.442
b
0.400 -1.502 -0.628 1.370
a
0.374
(2.07) (2.83) (1.60) (2.92)
ROA
t1
0.634 0.253 2.246
a
0.725 0.318 0.179 2.640
a
0.833
(0.77) (4.14) (0.28) (4.24)
NonOp
t1
-0.343 -0.212 1.648 0.277 -0.476 -0.273 1.612 0.295
(0.15) (0.78) (0.21) (0.78)
AbROA
t1
1.717 0.551 0.066 -0.029 1.963 0.594 0.403 0.074
(1.28) (0.06) (1.37) (0.37)
CapEx
t1
2.182
b
0.706 -3.630
a
-1.183 2.308
b
0.708 -3.759
a
-1.178
(2.49) (6.36) (2.85) (6.04)
Debt
t1
-0.240
b
-0.075 -0.445
b
-0.155 -0.184
c
-0.055 -0.434
b
-0.148
(2.11) (2.31) (1.75) (2.53)
MB
t1
0.096
c
0.027 -0.038
c
-0.014 0.101
c
0.027 -0.050
b
-0.018
(2.09) (2.05) (1.99) (2.13)
Size
t1
-0.016 -0.004 0.157
a
0.052 -0.031 -0.008 0.165
a
0.054
(0.86) (8.25) (1.02) (7.12)
Continued Next Page...
157
Panel B: Ownership Proportion by Fund Investment Horizon Tercile
Div. Increase Share Rep.
Est. MFX Est. MFX
(1) (2) (3) (4)
Own%S
t1
3.754
a
1.232 -1.641 -0.482
(3.91) (1.44)
Own%M
t1
3.066
c
0.875 -1.071 -0.488
(1.79) (0.82)
Own%L
t1
-1.880
c
-0.587 1.026 0.241
(1.80) (1.26)
AnnRet
t1
0.739
a
0.248 -0.543
a
-0.166
(6.46) (5.37)
Beta
t1
0.096 0.032 -0.034 -0.013
(1.33) (0.69)
SDRet
t1
40.360
a
13.924 -34.012
a
-10.998
(4.94) (5.20)
Vol
t1
-2.421
b
-0.922 1.877
a
0.572
(2.60) (3.37)
ROA
t1
0.572 0.238 2.205
a
0.709
(0.71) (3.97)
NonOp
t1
0.537 0.089 1.777 0.315
(0.22) (0.86)
AbROA
t1
1.802 0.593 0.175 -0.001
(1.33) (0.16)
CapEx
t1
2.374
b
0.760 -3.618
a
-1.180
(2.75) (7.05)
Debt
t1
-0.291
b
-0.085 -0.451
b
-0.151
(2.54) (2.54)
MB
t1
0.095
c
0.026 -0.039
c
-0.014
(2.06) (2.03)
Size
t1
-0.012 -0.002 0.151
a
0.051
(0.60) (6.60)
This table reports Fama-MacBeth (1973) estimates from bivariate probit
regressions explaining the choice of dividend paying …rms to either increase
dividends or repurchase shares. Marginal E¤ects (MFX) are presented to
the right of the coe¢cient estimates. Newey-West t-statistics are in paren-
theses. One cross-section regression is estimated per year from 1988 to
2007. For the dividend equation, the dependent variable is equal to 1 if
the …rm increased dividends, 0 otherwise. For the repurchase equation, the
dependent variable is equal to 1 if the …rm repurchased shares, 0 otherwise.
Fund ownership is controlled with either shareholder investment horizon
(SIH), current ownership length (COL), or the proportion of ownership
by fund investment horizon tercile (Own%S, Own%M, and Own%L). The
proportion of ownership by fund investment horizon tercile is de…ned in
Chapter 3.4, and SIH and COL are de…ned in Chapter 3.5. Other explana-
tory variables are de…ned in Chapter 3.3. Signi…cance at the 1% level is
designated with a, the 5% level with b, and the 10% level with c.
158
T
a
b
l
e
3
.
1
3
:
F
u
n
d
O
w
n
e
r
s
h
i
p
C
o
m
p
a
r
i
s
o
n
s
P
r
i
o
r
t
o
D
i
v
i
d
e
n
d
I
n
c
r
e
a
s
e
s
&
S
h
a
r
e
R
e
p
u
r
c
h
a
s
e
s
D
i
v
i
d
e
n
d
I
n
c
r
e
a
s
e
S
h
a
r
e
R
e
p
u
r
c
h
a
s
e
M
e
a
n
D
i
¤
.
M
e
d
i
a
n
D
i
¤
.
N
M
e
a
n
M
e
d
i
a
n
N
M
e
a
n
M
e
d
i
a
n

t
-
s
t
a
t
.

z
-
v
a
l
.
P
a
n
e
l
A
:
S
h
a
r
e
h
o
l
d
e
r
I
n
v
e
s
t
m
e
n
t
H
o
r
i
z
o
n
M
e
a
n
t
÷
3
1
6
6
7
3
.
0
4
9
3
.
0
5
3
2
7
5
6
3
.
0
8
7
3
.
0
9
7
-
0
.
0
3
8
a
(
5
.
2
1
)
-
0
.
0
4
4
a
(
5
.
3
7
)
C
h
a
n
g
e
t
÷
3
t
o
t
÷
2
1
6
6
7
0
.
0
0
4
0
.
0
0
1
2
7
5
6
0
.
0
1
0
0
.
0
0
4
-
0
.
0
0
5
(
1
.
0
0
)
-
0
.
0
0
3
(
1
.
2
0
)
C
h
a
n
g
e
t
÷
2
t
o
t
÷
1
1
6
9
7
-
0
.
0
0
7
-
0
.
0
0
2
2
6
6
4
0
.
0
1
4
0
.
0
1
1
-
0
.
0
2
1
a
(
3
.
9
8
)
-
0
.
0
1
3
a
(
4
.
0
0
)
C
h
a
n
g
e
t
÷
3
t
o
t
÷
1
1
6
6
7
-
0
.
0
0
9
-
0
.
0
0
4
2
7
5
6
0
.
0
2
0
0
.
0
1
3
-
0
.
0
2
9
a
(
4
.
4
4
)
-
0
.
0
1
6
a
(
4
.
2
9
)
P
a
n
e
l
B
:
C
u
r
r
e
n
t
O
w
n
e
r
s
h
i
p
L
e
n
g
t
h
M
e
a
n
t
÷
3
1
5
4
8
2
.
8
7
6
2
.
8
8
8
2
5
6
7
2
.
8
8
3
2
.
8
9
2
-
0
.
0
0
7
(
0
.
4
7
)
-
0
.
0
0
4
(
0
.
7
1
)
C
h
a
n
g
e
t
÷
3
t
o
t
÷
2
1
5
4
8
0
.
0
5
6
0
.
0
9
7
2
5
6
7
0
.
0
6
4
0
.
1
0
4
-
0
.
0
0
8
(
0
.
6
8
)
-
0
.
0
0
7
(
0
.
8
3
)
C
h
a
n
g
e
t
÷
2
t
o
t
÷
1
1
5
8
1
0
.
0
4
7
0
.
0
8
8
2
4
7
8
0
.
0
5
5
0
.
0
9
0
-
0
.
0
0
8
(
0
.
6
6
)
-
0
.
0
0
2
(
0
.
5
7
)
C
h
a
n
g
e
t
÷
3
t
o
t
÷
1
1
5
4
8
0
.
0
7
7
0
.
1
2
0
2
5
6
6
0
.
1
1
0
0
.
1
4
1
-
0
.
0
3
3
b
(
2
.
2
8
)
-
0
.
0
2
1
b
(
2
.
2
5
)
C
o
n
t
i
n
u
e
d
N
e
x
t
P
a
g
e
.
.
.
159
P
a
n
e
l
C
:
O
w
n
e
r
s
h
i
p
P
e
r
c
e
n
t
a
g
e
D
i
v
i
d
e
n
d
I
n
c
r
e
a
s
e
S
h
a
r
e
R
e
p
u
r
c
h
a
s
e
M
e
a
n
D
i
¤
.
M
e
d
i
a
n
D
i
¤
.
N
M
e
a
n
M
e
d
i
a
n
N
M
e
a
n
M
e
d
i
a
n

t
-
s
t
a
t
.

z
-
v
a
l
.
O
w
n
%
S
M
e
a
n
t
÷
3
1
6
6
7
0
.
0
2
2
0
.
0
1
5
2
7
5
6
0
.
0
1
9
0
.
0
1
2
0
.
0
0
3
a
(
4
.
1
2
)
0
.
0
0
3
a
(
4
.
7
5
)
C
h
a
n
g
e
t
÷
3
t
o
t
÷
2
1
6
6
7
0
.
0
0
6
0
.
0
0
3
2
7
5
6
0
.
0
0
4
0
.
0
0
2
0
.
0
0
1
a
(
3
.
1
7
)
0
.
0
0
1
a
(
2
.
7
5
)
C
h
a
n
g
e
t
÷
2
t
o
t
÷
1
1
6
9
7
0
.
0
0
1
0
.
0
0
0
2
6
6
4
-
0
.
0
0
1
0
.
0
0
0
0
.
0
0
0
a
(
2
.
8
5
)
0
.
0
0
0
b
(
2
.
4
1
)
C
h
a
n
g
e
t
÷
3
t
o
t
÷
1
1
6
6
7
0
.
0
0
7
0
.
0
0
4
2
7
5
6
0
.
0
0
3
0
.
0
0
1
0
.
0
0
2
a
(
4
.
9
7
)
0
.
0
0
2
a
(
4
.
5
0
)
O
w
n
%
M
M
e
a
n
t
÷
3
1
6
6
7
0
.
0
2
8
0
.
0
2
0
2
7
5
6
0
.
0
3
1
0
.
0
2
3
-
0
.
0
0
3
a
(
3
.
7
1
)
-
0
.
0
0
3
a
(
3
.
7
5
)
C
h
a
n
g
e
t
÷
3
t
o
t
÷
2
1
6
6
7
0
.
0
0
3
0
.
0
0
2
2
7
5
6
0
.
0
0
5
0
.
0
0
3
-
0
.
0
0
1
a
(
3
.
1
1
)
-
0
.
0
0
1
a
(
3
.
3
6
)
C
h
a
n
g
e
t
÷
2
t
o
t
÷
1
1
6
9
7
0
.
0
0
0
0
.
0
0
0
2
6
6
4
-
0
.
0
0
2
-
0
.
0
0
1
0
.
0
0
2
a
(
3
.
1
6
)
0
.
0
0
0
b
(
2
.
5
2
)
C
h
a
n
g
e
t
÷
3
t
o
t
÷
1
1
6
6
7
0
.
0
0
2
0
.
0
0
1
2
7
5
6
0
.
0
0
3
0
.
0
0
2
-
0
.
0
0
1
(
0
.
6
3
)
-
0
.
0
0
1
(
1
.
2
7
)
O
w
n
%
L
M
e
a
n
t
÷
3
1
6
6
7
0
.
0
4
9
0
.
0
3
5
2
7
5
6
0
.
0
5
0
0
.
0
3
7
-
0
.
0
0
2
(
0
.
4
8
)
-
0
.
0
0
2
(
1
.
2
2
)
C
h
a
n
g
e
t
÷
3
t
o
t
÷
2
1
6
6
7
0
.
0
0
2
0
.
0
0
1
2
7
5
6
0
.
0
0
4
0
.
0
0
2
-
0
.
0
0
1
a
(
2
.
8
7
)
-
0
.
0
0
1
a
(
3
.
2
0
)
C
h
a
n
g
e
t
÷
2
t
o
t
÷
1
1
6
9
7
0
.
0
0
0
0
.
0
0
0
2
6
6
4
0
.
0
0
0
0
.
0
0
0
0
.
0
0
0
(
0
.
5
6
)
0
.
0
0
0
(
1
.
0
7
)
C
h
a
n
g
e
t
÷
3
t
o
t
÷
1
1
6
6
7
0
.
0
0
2
0
.
0
0
1
2
7
5
6
0
.
0
0
4
0
.
0
0
2
-
0
.
0
0
2
b
(
2
.
5
6
)
-
0
.
0
0
2
a
(
3
.
1
6
)
T
h
i
s
t
a
b
l
e
r
e
p
o
r
t
s
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
s
h
a
r
e
h
o
l
d
e
r
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
(
S
I
H
)
,
c
u
r
r
e
n
t
o
w
n
e
r
s
h
i
p
l
e
n
g
t
h
(
C
O
L
)
,
a
n
d
t
h
e
p
r
o
-
p
o
r
t
i
o
n
o
f
o
w
n
e
r
s
h
i
p
b
y
f
u
n
d
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
t
e
r
c
i
l
e
(
O
w
n
%
S
,
O
w
n
%
M
,
O
w
n
%
L
)
c
h
a
n
g
e
s
b
e
t
w
e
e
n
d
i
v
i
d
e
n
d
i
n
c
r
e
a
s
e
s
a
n
d
s
h
a
r
e
r
e
p
u
r
c
h
a
s
e
s
o
f
d
i
v
i
d
e
n
d
p
a
y
i
n
g
…
r
m
s
.
D
i
¤
e
r
e
n
c
e
s
a
r
e
t
a
k
e
n
p
r
i
o
r
t
o
e
v
e
n
t
y
e
a
r
t
f
r
o
m
t
÷
3
t
o
t
÷
2
,
t
÷
2
t
o
t
÷
1
,
a
n
d
t
÷
3
t
o
t
÷
1
.
I
c
a
l
c
u
l
a
t
e
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
m
e
a
n
a
n
d
m
e
d
i
a
n
u
n
a
d
j
u
s
t
e
d
c
h
a
n
g
e
s
.
T
e
s
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
l
o
c
a
t
e
d
b
e
l
o
w
t
h
e
r
e
p
o
r
t
e
d
c
h
a
n
g
e
.
T
h
e
p
r
o
p
o
r
t
i
o
n
o
f
o
w
n
e
r
s
h
i
p
b
y
f
u
n
d
i
n
v
e
s
t
m
e
n
t
h
o
r
i
z
o
n
t
e
r
c
i
l
e
i
s
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
3
.
4
,
a
n
d
S
I
H
a
n
d
C
O
L
a
r
e
d
e
…
n
e
d
i
n
C
h
a
p
t
e
r
3
.
5
.
F
o
r
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
m
e
a
n
s
,
I
u
s
e
t
h
e
t
w
o
-
t
a
i
l
e
d
t
-
s
t
a
t
i
s
t
i
c
t
o
t
e
s
t
s
i
g
n
…
c
a
n
c
e
.
F
o
r
t
h
e
d
i
¤
e
r
e
n
c
e
i
n
m
e
d
i
a
n
s
I
u
s
e
t
h
e
t
w
o
-
t
a
i
l
e
d
z
-
s
t
a
t
i
s
t
i
c
f
r
o
m
t
h
e
W
i
l
c
o
x
o
n
r
a
n
k
-
s
u
m
t
e
s
t
.
S
i
g
n
i
…
c
a
n
c
e
a
t
t
h
e
1
%
l
e
v
e
l
i
s
d
e
n
o
t
e
d
w
i
t
h
a
,
t
h
e
5
%
l
e
v
e
l
w
i
t
h
b
,
a
n
d
t
h
e
1
0
%
l
e
v
e
l
w
i
t
h
c
.
160
List of References
Abarbanell, Je¤ery S., Brian J. Bushee, and Jana Smith Raedy, 2003, Institutional
investor preferences and price pressure: The case of corporate spin-o¤s, Journal of
Business 76, 233-261.
Ajinkya, Bipin, Sanjeev Bhojraj, and Partha Sengupta, 2005, The association between
outside directors, institutional investors and the properties of management earnings
forecasts, Journal of Accounting Research 43, 343-376.
Allen, Franklin, Antonio E. Bernardo, and Ivo Welch, 2000, A theory of dividends
based on tax clienteles, The Journal of Finance 55, 2499-2536.
Anderson, T.W., and C. Hsiao, 1982, Formulation and estimation of dynamic models
using panel data, Journal of Econometrics 18, 570-606.
Barber, Brad M., and John D. Lyon, 1996, Detecting abnormal operating perfor-
mance: The empirical power and speci…cation of test statistics, Journal of Financial
Economics 41, 359-399.
Barber, Brad M., John D. Lyon, and Chih-Ling Tsai, 1999, Improved methods for
tests of long-run abnormal stock returns, The Journal of Finance 54, 165-201.
Barclay, Michael J., Cli¤ord G. Holderness, and Dennis P. Sheehan, 2009, Dividends
and corporate shareholders, The Review of Financial Studies 22, 2423-2455.
Barclay, Michael J., and Cli¤ord W. Smith, 1988, Corporate payout policy: Cash
dividends versus open-market repurchases, Journal of Financial Economics 22, 61-
82.
Bartov, Eli, Dan Givoly, and Carla Hayn, 2002, The rewards to meeting or beating
earnings expectations, Journal of Accounting and Economics 33, 173-204.
Bennett, James A., Richard W. Sias, and Laura T. Starks, 2003, Greener pastures
and the importance of dynamic institutional preferences, The Review of Financial
Studies 16, 1203-1238.
Bøhren, Øyvind, Richard Priestley, and Bernt Arne Ødegaard, 2005, The duration of
equity ownership, Working paper, Norwegian School of Management.
Bøhren, Øyvind, Richard Priestley, and Bernt Arne Ødegaard, 2008, Investor short-
horizonism and …rm value, Working paper, Norwegian School of Management, Uni-
versity of Stavenger.
Brennan, Michael J., and Anjan V. Thakor, 1990, Shareholder preferences and divi-
dend policy, The Journal of Finance 45, 993-1018.
Brown, Je¤rey R., Nellie Liang, and Scott Weisbenner, 2007, Executive …nancial
incentives and payout policy: Firmresponse to the 2003 dividend tax cut, The Journal
161
of Finance 62, 1935-1965.
Brown, Keith C., and Bryce A. Brooke, 1993, Institutional demand and security price
pressure: The case of corporate spin-o¤s, Financial Analysts Journal 49, 53-62.
Burch, Timothy R., and Vikram Nanda, 2003, Divisional diversity and the conglom-
erate discount: Evidence from spin-o¤s, Journal of Financial Economics 70, 69-98.
Burgstahler, David, and Ilia Dichev, 1997, Earnings Management to avoid earnings
decreases and losses, Journal of Accounting and Economics 24, 99-126.
Burgstahler, David, and Michael Eames, 2003, Earnings management to avoid losses
and earnings decreases: Are analysts fooled? Contemporary Accounting Research 20,
253-294.
Burgstahler, David, and Michael Eames, 2006, Management of earnings and analysts’
forecasts to achieve zero and small positive earnings surprises, Journal of Business
Finance and Accounting 33, 633-652.
Burns, Natasha, Simi Kedia, and Marc Lipson, 2006, The e¤ects of institutional
ownership on …nancial reporting practices, Working paper, University of Georgia,
Rutgers University, and University of Virginia.
Bushee, Brian J., 1998, The in‡uence of institutional investors on myopic r&d invest-
ment behavior, The Accounting Review 73, 305-333.
Bushee, Brian J., 2001, Do institutional investors prefer near-term earnings over long-
run value? Contemporary Accounting Research 18, 207-246.
Bushee, Brian J., and Christopher Noe, 2000, Corporate disclosure practices, insti-
tutional investors, and stock return volatility, Journal of Accounting Research 38,
171-202.
Cleves, Mario, William Gould, Roberto Gutierrez, and Yulia Marchenko, 2008, An
introduction to survival analysis using Stata, Stata Press, College Station, Texas.
Chemmanur, Thomas J., and Shan He, 2008, Institutional trading, information pro-
duction, and corporate spin-o¤s, Working paper, Boston College and Louisiana State
University.
Chemmanur, Thomas J., and An Yan, 2004, A theory of coporate spin-o¤s, Journal
of Financial Economics 72, 259-290.
Chen, Xia, Jarrad Harford, and Kai Li, 2007, Monitoring: Which institutions matter?
Journal of Financial Economics 86, 279-305.
Cheng, C.S. Agnes, and Austin Reitenga, 2001, Characteristics of Institutional In-
162
vestors and Discretionary Accruals, Working paper, University of Houston.
Chetty, Raj, and Emmanuel Saez, 2005, Dividend taxes and corporate behavior:
Evidence from the 2003 tax cut, The Quarterly Journal of Economics 120, 791-833.
Chiyachantana, Chiraphol, Christine Jiang, Nareerat Taechapiroontong, and Robert
Wood, 2004, The impact of Regulation Fair Disclosure on information asymmetry
and trading: An intraday analysis, The Financial Review 39, 549-577.
Chung, Richard, Michael Firth, and Joeng-Bon Kim, 2002, Institutional monitoring
and opportunistic earnings management, Journal of Corporate Finance 8, 29-48.
Çolak, Gönül, and Toni M. Whited, 2006, Spin-o¤s, divestitures, and conglomerate
investment, The Review of Financial Studies 20, 557-595.
Cusatis, Patrick J., James A. Miles, and J. Randall Woolridge, Restructuring through
spin-o¤s: The stock market evidence, Journal of Financial Economics 33, 293-311.
Daley, Lane, Vikas Mehrota, and Ranjini Sivakumar, 1997, Corporate focus and value
creation: Evidence from spin-o¤s, Journal of Financial Economics 45, 257-281.
Das, Somnath, and Huai Zhang, 2003, Rounding-up in reported EPS, behavioral
thresholds, and earnings management, Journal of Accounting and Economics 35, 31-
50.
Dechow, Patricia, Richard Sloan, and Amy Sweeney, 1995, Detecting earnings man-
agement, The Accounting Review 70, 193-225.
Degeorge, François, Jayendu Patel, and Richard Zeckhauser, 1999, Earnings manage-
ment to exceed thresholds, Journal of Business 72, 1-33.
Del Guercio, Diane, 1996, The distorting e¤ect of the prudent-man laws on institu-
tional equity investments, Journal of Financial Economics 40, 31-62.
Del Guercio, Diane, and Jennifer Hawkins, 1999, The motivation and impact of pen-
sion fund activism, Journal of Financial Economics 52, 293-340.
Denis, David J., Diane K. Denis, and Atulya Sarin, 1994, The information content
of dividend changes: Cash ‡ow signaling, overinvestment, and dividend clienteles,
Journal of Financial and Quantitative Analysis 29, 567-587.
Desai, Hemang, and Prem C. Jain, 1999, Firm performance and focus: Long-run
stock market performance following spin-o¤s, Journal of Financial Economics 54,
75-101.
Desai, Mihir A., and Dhammika Dharmapala, 2009, Dividend taxes and interna-
tional portfolio choice, Working paper, Harvard University and University of Illinois
163
at Urbana-Champaign.
Dittmar, Amy, 2004, Capital structure in corporate spin-o¤s, Journal of Business 77,
9-43.
Dittmar, Amy, and Anil Shivdasani, 2003, Divestitures and divisional investment
policies, The Journal of Finance 68, 2711-2743.
Falkenstein, Eric G., 1996, Preferences for stock characteristics as revealed by mutual
fund portfolio holdings, The Journal of Finance 51, 111-135.
Fama, Eugene, and Kenneth French, 2001, Disappearing dividends: Changing …rm
characteristics or lower propensity to pay?, Journal of Financial Economics 60, 3-44.
Fama, Eugene, and James MacBeth, 1973, Risk, return, and equilibrium: Empirical
tests, Journal of Political Economy 71, 607-636.
Ferreira, Miguel A., Massimo Massa, and Pedro Matos, 2009, Dividend clienteles
around the world: Evidence from institutional holdings, Working paper, Universidade
nova de Lisboa, INSEAD, and University of Southern California.
Gaspar, José-Miguel, Massimo Massa, and Pedro Matos, 2005, Shareholder invest-
ment horizons and the market for corporate control, Journal of Financial Economics
76, 136-165.
Gertner, Robert, Eric Powers, and David Scharfstein, 2002, Learning about internal
capital markets from corporate spin-o¤s, The Journal of Finance 57, 2479-2506.
Gilson, Stuart C., Paul M. Healy, Christopher F. Noe, and Krishna G. Palepu, 2001,
Analyst specialization and conglomerate stock breakups, Journal of Accounting Re-
search 39, 565-582.
Gompers, Paul A., and Andrew Metrick, 2001, Institutional investors and equity
prices, The Quarterly Journal of Economics 116, 229-259.
Graham, John R., Campbell R. Harvey, and Shiva Rajgopal, 2005, The economic
impact of corporate …nance reporting, Journal of Accounting and Economics 40, 3-
73.
Greenwood, Robin, 2006, Price pressure in corporate spin-o¤s, Working paper, Har-
vard University.
Grinblatt, Mark, and Sheridan Titman, 1989, Mutual fund performance: An analysis
of quarterly portfolio holdings, Journal of Business 62, 394-415.
Grinstein, Yaniv, and Roni Michaely, 2005, Institutional holdings and payout policy,
The Journal of Finance 60, 1389-1426.
164
Grullon, Gustavo, and Roni Michaely, 2002, Dividends, share repurchases, and the
substitution hypothesis, The Journal of Finance 57, 1649-1684.
Grullon, Gustavo, and Roni Michaely, 2005, The information content of share repur-
chase programs, The Journal of Finance 59, 651-680.
Guay, Wayne, and Jarrad Harford, 2000, The cash-‡ow permanence and information
content of dividend increases versus repurchases, Journal of Financial Economics 57,
385-415.
Habib, Michel A., D. Bruce Johnsen, and Narayanan Y. Naik, 1997, Spin-o¤s and
information, Journal of Financial Intermediation 6, 153-176.
Hall, Peter, 1992, On the removal of skewness by transformation, Journal of the Royal
Statistical Society, Series B 54, 221–228.
Hite, Gailen L., and James E. Owers, 1983, Security price reactions around coporate
spin-o¤ announcements, Journal of Financial Economics 12, 409-436.
Hotchkiss, Edith S., and Stephen Lawrence, 2007, Empirical evidence on the existence
of dividend clienteles, Working paper, Boston College.
Hotchkiss, Edith S., and Deon Strickland, 2003, Does shareholder composition mat-
ter? Evidence from the market reaction to corporate earnings announcements, The
Journal of Finance 58, 1469-1498.
Hribar, Paul, Nicole Jenkins, and Juan Wang, 2004, Institutional investors and ac-
counting restatements, Working paper, Cornell University and Washington University
in St. Louis.
Hsu, Grace C.-M., and Ping-Sheng Koh, 2005, Does the presence of institutional
investors in‡uence accruals management? Evidence from Australia, Corporate Gov-
ernance 13, 809-823.
Huson, Mark R., and Gregory MacKinnon, 2003, Corporate spin-o¤s and information
asymmetry between investors, Journal of Corporate Finance 9, 481-503.
Jagannathan, Murali, Cli¤ord P. Stephens, and Michael S. Weisbach, 2000, Finan-
cial ‡exibility and the choice between dividends and stock repurchases, Journal of
Financial Economics 57, 355-384.
Janakiraman, Surya, Suresh Radhakrishnan, and Rafal Szwejkowski, 2006, Regula-
tion Fair Disclosure and analysts’ …rst-forecast horizon, Working paper, University of
Texas at Dallas.
Jenkins, David, and Uma Velury, 2006, Institutional ownership and the quality of
earnings, Journal of Business Research 59, 1043-1051.
165
Jensen, Michael C., 1986, Agency costs of free cash ‡ow, corporate …nance, and
takeovers, American Economic Review 76, 323-329.
Jensen, Michael C., 2005, Agency costs of overvalued equity, Financial Management
34, 5-19.
Jones, Jennifer, 1991, Earnings Management during import relief investigations, Jour-
nal of Accounting Research 29, 193-228.
Kasznik, Ron, and Maureen McNichols, 2002, Does meeting earnings expectations
matter? Evidence from analyst forecast revisions and share prices, Journal of Ac-
counting Research 40, 727-759.
Kenney, William, David Burgstahler, and Roger Martin, 2002, Earnings surprise
"materiality" as measured by stock returns, Journal of Accounting Research 40, 1297-
1329.
Ke, Bin, Kathy Petroni, and Yong Yu, 2008, The e¤ect of Regulation FD on transient
institutional investors’ trading behavior, Journal of Accounting Research 46, 853-883.
Koh, Ping-Sheng, 2003, On the association between institutional ownership and ag-
gressive corporate earnings management in Australia, The British Accounting Review
35, 105-128.
Krishnaswami, Sudha, and Venjat Subramaniam, 1999, Information asymmetry, valu-
ation, and the corporate spin-o¤ decision, Journal of Financial Economics 53, 73-112.
Li, Xu, Suresh Radhakrishnan, Haeyoung Shin, and Jin Zhang, 2006, Regulation
FD, accounting restatements and transient institutional investors’ trading behavior,
Working Paper, University of Texas at Dallas.
Lie, Erik, 2001, Detecting abnormal operating performance: Revisited, Financial
Management 30, 77-91.
Matsumoto, Dawn, 2002, Management’s incentives to avoid negative earnings sur-
prise, The Accounting Review 77, 483-514.
Mehrota, Vikas, Wayne Mikkelson, and Megan Partch, 2003, The design of …nancial
policies in corporate spin-o¤s, Review of Financial Studies 16, 1359-1388.
Miles, James A., and James D. Rosenfeld, 1983, The e¤ect of voluntary spin-o¤
announcements on shareholder wealth, The Journal of Finance 38, 1597-1606.
Nanda, Vikram, and M.P. Narayanan, 1999, Disentangling value: Financing needs,
…rm scope, and divestitures, Journal of Financial Intermediation 8, 174-204.
Patro, Sukesh, 2008, The evolution of ownership structure of corporate spin-o¤s,
166
Journal of Corporate Finance 14, 596-613.
Perez-Gonzalez, Francisco, 2002, Large shareholders and dividends: Evidence from
U.S. tax reforms, Working paper, Columbia University.
Petersen, Mitchell A., 2009, Estimating standard errors in …nance panel data sets:
Comparing approaches, Review of Financial Studies 22, 435-480.
Rajan, Raghuram, Henri Servaes, and Luigi Zingales, 2000, The cost of diversity: The
diversi…cation discount and ine¢cient investment, The Journal of Finance 55, 35-80.
Rajgopal, Shivram, Mohan Venkatachalam, and James Jiambalvo, 1999, Is institu-
tional ownership associated with earnings management and the extent to which stock
prices re‡ect future earnings? Working paper, Stanford University and University of
Washington.
Roychowdhury, Sugata, 2006, Earnings manipulation through real activities manipu-
lation, Journal of Accounting and Economics 42, 335-370.
Skinner, Douglas, and Richard Sloan, 2002, Earnings surprises, growth expectations,
and stock returns or don’t let an earnings torpedo sink your portfolio, Review of
Accounting Studies 7, 289-312.
Scharfstein, David, and Jeremy Stein, 2000, The dark side of internal capital markets:
Divisional rent-seeking and ine¢cient investment, The Journal of Finance 55, 2537-
2564.
Schipper, Katherine, and Abbie Smith, 1983, E¤ects of recontracting on shareholder
wealth: The case of voluntary spin-o¤s, Journal of Financial Economics 12, 437-467.
Shleifer, Andrei, and Robert W. Vishny, 1986, Large shareholders and corporate
control, Journal of Political Economy 94, 461-488.
Thomas, Shawn, 2002, Firm diversi…cation and asymmetric information: Evidence
fromanalysts’ forecasts and earnings announcements, Journal of Financial Economics
64, 373-396.
Useem, Michael, 1996, Investor capitalism, BasicBooks, New York.
Vijh, Anand M., 1994, The spin-o¤ and merger ex-date e¤ects, The Journal of Finance
49, 581-609.
Wahal, Sunil, and John J. McConnell, 2000, Do institutional investors exacerbate
managerial myopia? Journal of Corporate Finance 6, 307-329.
Wermers, Russell, 2000, Mutual fund performance: An empirical decomposition into
stock-picking talent, style, transaction costs, and expenses, The Journal of Finance
167
55, 1655-1695.
Wruck, Karen H., 1994, Financial policy, internal control, and performance: Sealed
air corporation’s leveraged special dividend, Journal of Financial Economics, 36, 157-
192.
Yan, Xuemin, and Zhe Zhang, 2009, Institutional investors and equity returns: Are
short term institutions better informed? The Review of Financial Studies 22, 893-924.
Yoon, Pyung Sig, and Laura T. Starks, 1995, Signaling, investment opportunities,
and dividend announcements, The Review of Financial Studies 8, 995-1018.
Yu, Fang, 2008, Analyst coverage and earnings management, Journal of Financial
Economics 88, 245-271.
168

doc_466392813.pdf