SHARE REPURCHASES: HOW IMPORTANT IS MARKET TIMING?

By

Chao Zhuang

A Dissertation Presented to the

FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BUSINESS ADMINISTRATION)

August 2014

Copyright 2014 Chao Zhuang

 UMI Number: 3644731

All rights reserved

INFORMATION TO ALL USERS

The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.

UMI 3644731
Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author.
Microform Edition © ProQuest LLC.
All rights reserved. This work is protected against
unauthorized copying under Title 17, United States Code

ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106 - 1346
 Dedication

To my beloved parents,
Yuzhen Zhang and Chenzhao Zhuang,
and to my lovely wife,
Ye Fu,
whose support and encouragement made this dissertation possible.

ii
 Acknowledgements

I wish to acknowledge with deep gratitude the counsel and kindness of my dissertation chair,
Harry DeAngelo. I am especially grateful to my dissertation committee members, Kevin J.
Murphy and Oguzhan Ozbas, for long and fruitful discussions. I would like to express my
appreciation to Alice A. Bonaime, Daniel Carvalho, Mark DeFond, Ehud Kamar, John
Matsusaka, Lori Santikian, and to seminar participants at the University of Southern California,
and the FMA 2012 Doctoral Student Consortium, for helpful comments and suggestions. I also
thank Larry Dann for providing his tender-offer data.

iii
 Table of Contents

Dedication…………………………………………………………………………………..……. ii

Acknowledgements…………..……………………………………………………………..……iii

List of Tables…………………………………………………………………………….……...vi

List of Figures………………………………………………………………………………….....ix

Abstract………………………………………………………………………………….………...x

Chapter 1: Introduction…………………………………………………………………................1

Chapter 2: Theoretical Explanations for Share Repurchases…………………………………….12

2.1 The Market-Timing Hypothesis…………...……………………………………………12
2.1.1 Three Versions of the Market-Timing Theory of Share Repurchases……...……12
2.1.2 Prior Evidence Supporting the Market-Timing Theory………………………….15
2.2 The Free-Cash-Flow (FCF) Distribution Hypothesis….………………………………17
2.3 The Employee-Stock-Option (ESO) Hypothesis…..……………………………………19
2.4 The Leverage-Rebalancing Hypothesis…....……………………………………………20

Chapter 3: Sample Selection and Descriptive Statistics……………………………………….22

Chapter 4: Logit Analysis of the Decision to Conduct a Share Repurchase…….………...…….26

4.1 Basic Logit Tests……...…….…………………………………………………………..27
4.2 Relative Impact on the Repurchase Decision of Market Timing versus FCF……….31
4.3 The Interaction between Free-Cash-Flow and Market-Timing Variables...…...………..37

Chapter 5: Robustness Checks and Supplemental Analyses…………………………………….39

5.1 Alternative Measures of Market-Timing Opportunities………………………………...39
5.2 Logit Analysis with the Sample Period Expanded to 1971-2010…...…………………..40
5.3 Logit Analysis Comparing the Market-Timing and Employee-Stock-Option Effects..45
5.4 Logit Analysis Comparing the Market-Timing and Leverage-Rebalancing Effects……46
5.5 Logit Analysis of the Decision to Conduct a Tender-Offer Repurchase..………………48
5.5.1 Tender-Offer Repurchases, Market Timing, and Free Cash Flow……………….49
5.5.2 Tender-Offer Repurchases, Market Timing, and Employee Stock Options….….52
5.5.3 Tender-Offer Repurchases, Market Timing, and Leverage Rebalancing……...…53
5.6 Logit Analysis of the Decision to Pay Dividends……………………………………….57
5.7 Repurchase Frequency………………………………………………………………..…59

Chapter 6: Abnormal Stock Returns after Repurchases…………………………………………62

6.1 Median and Mean Post-repurchase Abnormal Stock Returns…………………………..62
6.2 Distribution Analyses of Post-repurchase Abnormal Stock Returns…………………....65
6.3 Characteristics of Large Gain and Large Loss Repurchases...………………………….70

iv
Chapter 7: Equity Payouts and Capital Structure………..…………………..…………………..74
7.1 Background………………………………………………………………..…………….74
7.2 Large and Debt-financed Payout Increases………………………………………..……75
7.3 Logit Analysis of Decisions to Make Large and Debt-financed Payout Increases….….82
7.4 Specially Designed Dividends (SDDs) in Late 2012……...…………………………….86

Chapter 8: Conclusion and Discussion…………………………………………………………..90

References……………………………………………………………………..…………………94

Bibliography……………………………………………………………………………………..97

v
 List of Tables

Table 2.1: Three versions of the market-timing theory of share repurchases…………………..105

Table 3.1: Aggregate tender-offer repurchases and all repurchases from 1985 to 2010…….…106

Table 3.2: Distribution of percent of shares repurchased………………………………………107

Table 3.3: Aggregate SEOs and all equity issues from 1985 to 2010….………………………108

Table 4.1: Logit analysis of share repurchase decisions………………………………………..109

Table 4.2: Estimated probability of a share repurchase……………………………………....110

Table 4.3: Logit analysis of share repurchase decisions as a function of free cash flow………112

Table 4.4: Estimated probability of a share repurchase as a function of free cash flow……….113

Table 4.5: Logit analysis of share repurchase decisions as a function of transitory cash flow...115

Table 4.6: Estimated probability of a share repurchase as a function of transitory cash flow…116

Table 4.7: Logit analysis of share repurchase decisions as a function of cash holdings……...118

Table 4.8: Estimated probability of a share repurchase as a function of cash holdings….…….119

Table 4.9: Logit analysis of share repurchase decisions as a function of excess cash holdings..121

Table 4.10: Estimated probability of a share repurchase as a function of excess cash

holdings………………………………………………………………………………………....122

Table 4.11: The interaction effect between operating cash flow and market timing on share
repurchase decisions …………………………………………………………………………...124

Table 5.1: Estimated probability of a share repurchase as a function of operating cash flow and
the RRV and PS mispricing indices………….…………………………………………………125

Table 5.2: Estimated probability of a share repurchase as a function of free cash flow and the
RRV and PS mispricing indices………………………………………………………………...126

Table 5.3: Logit analysis of share repurchase decisions as a function of operating cash flow over
1971 to 2010……………………………………………………………………………………127

Table 5.4: Estimated probability of a share repurchase as a function of operating cash flow over
1971 to 2010 ………………………………………………………………………………...…128

vi
Table 5.5: Logit analysis of share repurchase decisions as a function of free cash flow over 1971
to 2010……..…………………………………………………………………………………...130

Table 5.6: Estimated probability of a share repurchase as a function of free cash flow over 1971
to 2010………..………………………………………………………………………………...131

Table 5.7: Estimated probability of a share repurchase as a function of the change in outstanding
stock options………………………...…………………………………………...…………...133

Table 5.8: Estimated probability of a share repurchase as a function of the deviation from target
leverage…………………………………………………………………………………………135

Table 5.9: Logit analysis of tender-offer repurchase decisions as a function of operating cash
flow……………………………………………………………………………………………..137

Table 5.10: Estimated probability of a tender-offer repurchase as a function of operating cash

flow……………………………………………………………………………………………..138

Table 5.11: Logit analysis of tender-offer repurchase decisions as a function of free cash
flow……………………………………………………………………………………………..140

Table 5.12: Estimated probability of a tender-offer repurchase as a function of free cash

flow……………………………………………………………………………………………..141

Table 5.13: Estimated probability of a tender-offer repurchase as a function of the change in

outstanding stock options…………………………………………………………………….....143

Table 5.14: Estimated probability of a tender-offer repurchase as a function of the deviation from
target leverage………………………………………………………………………….……….145

Table 5.15: Leverage and traditional leverage determinants surrounding tender-offer

repurchases……………………………………………………………………………………...147

Table 5.16: Logit analysis of dividend decisions as a function of operating cash flow…….....148

Table 5.17: Estimated probability that a firm pays dividends as a function of operating cash
flow……………………………………………………………………………………………..149

Table 5.18: Frequencies of share repurchases………………………………………………….151

Table 5.19: Firm age and the number of years in which a firm repurchases shares……………152

Table 6.1: Histogram of abnormal stock rates of return following share repurchases…………153

Table 6.2: Histogram of abnormal dollar stock returns following share repurchases…….……154

vii
Table 6.3: Biggest winners and losers of share repurchases based on 12-month abnormal
returns…………………………………………………………………………………………..155

Table 6.4: Biggest winners and losers of share repurchases based on 36-month abnormal
returns…………………………………………………………………………………………..157

Table 6.5: Distribution of abnormal rates of return and abnormal dollar returns of share
repurchases……………………………………………………………………………………...159

Table 6.6: Distribution of future abnormal rates of return and dollar returns for large gain, loss,
and other repurchases………………………………………………………………………….160

Table 6.7: Distribution of future abnormal rates of return and dollar returns for middle
repurchases……………………………………………………………………………………...161

Table 6.8: Distribution of future abnormal rates of return and dollar returns for middle
repurchases for the tender-offer subsample…………………………………………………...162

Table 6.9: Characteristics of large gain, large loss and other repurchases based on future 12-
month abnormal returns………………………………………………………………………..163

Table 6.10: Characteristics of large gain, large loss and other repurchases based on future 36-
month abnormal returns………………………………………………………………………164

Table 7.1: The effect of a change in debt and payout increase on leverage and cash holdings...165

Table 7.2: Payout increases and excess leverage…………………………………………….…166

Table 7.3: Logit analysis of large payout and debt-financed payout increase decisions……….167

Table 7.4: Estimated probabilities of making large payout increases and debt-financed payout
increases: Part I………….……………………………………………………………………...168

Table 7.5: Estimated probabilities of making large payout increases and debt-financed payout
increases: Part II………………………………………………………………………………...169

Table 7.6: Firm characteristics and leverage around specially designated dividends (SDDs) in
late 2012……….……………………………………………………………………………..…170

Table 7.7: Logit analysis of the decision to pay a specially designated dividend (SDD) in late
2012……………………………………………………………………………………………..171

Table 7.8: Estimated probability of paying a specially designated dividend (SDD) in late
2012……………………..…………………………………………………………….............172

viii
 List of Figures

Figure 3.1: Aggregate share repurchases, dividends and level of S&P 500 index for each year
from 1971 to 2010…………………………………………………………………………..……98

Figure 3.2: Percent of firms that repurchase stock and pay dividends from 1971 to 2010...……99

Figure 4.1: Interaction effects of operating-cash-flow and timing variables on share repurchase
decisions…………………………………………………………………………...……………100

Figure 7.1: Evolution of leverage around large payout increases………...…………………… 101

Figure 7.2: Evolution of leverage around debt-financed payout increases……………...……103

Figure 7.3: Specially designated dividends (SDDs) in 2011 and 2012…….………………..…104

ix
 Abstract
Although market timing is empirically important, free-cash-flow (FCF) considerations have

much stronger effects on managers’ share-repurchase decisions. Managers either do not

systematically time the market or have poor timing ability as many firms fail to exploit good

timing opportunities through repurchases. Decisions to buy back shares often tend to be poor in a

market-timing sense as firms are more likely to experience negative than positive abnormal stock

returns after repurchases. Although the average post-repurchase abnormal returns are positive,

this finding is driven by the extreme abnormal stock returns following a modest number of

repurchases that are small in dollar magnitude and that account for a small percentage of the

aggregate dollar value repurchased. I also find that high levels of FCF greatly increase the

probability a share repurchase, and this effect strongly dominates market-timing considerations.

Firms with poor market-timing opportunities and high FCF are more than twelve times as likely

to buy back shares as firms with good market-timing opportunities and low FCF. Employee-

stock-option (ESO) and leverage-rebalancing motives are at least as important as market timing

in explaining managers’ share-repurchase decisions.

x
 Chapter 1

Introduction

The market-timing explanation for share repurchases holds that managers exploit stock

market undervaluation by buying back stock on the cheap. The literature views market timing as

the primary explanation for share repurchases (Vermaelen, 2005; Baker and Wurgler, 2002).

Evidence supporting this argument includes: (1) positive average abnormal announcement

returns for all types of repurchases except greenmail transactions (Dann, 1981; Vermaelen, 1981;

Masulis, 1980; Peyer and Vermaelen, 2005; Manconi, Peyer, and Vermaelen, 2013); (2) a

negative correlation between prior stock returns and open-market repurchases (Comment and

Jarrell, 1991; Stephens and Weisbach, 1998); (3) positive post-buyback long-term excess returns

for the average firm in open-market, tender-offer and private repurchases (Lakonishok and

Vermaelen, 1990; Ikenberry, Lakonishok, and Vermaelen, 1995 and 2000; Peyer and Vermaelen,

2005 and 2009; Manconi, Peyer, and Vermaelen, 2013); and (4) a negative correlation between

the market-to-book ratio (M/B) and open-market repurchases (Dittmar, 2000).

However, some researchers have concluded that market timing does not fully explain

aggregate repurchase activities. The market-timing hypothesis predicts that firms tend to

repurchase stock less often and in smaller dollar amounts when stock market valuations are high,

but the opposite tends to be true as indicated by Dittmar and Dittmar (2008). They find that the

aggregate dollar volume of stock repurchases is positively related to aggregate stock market

valuation. For example, aggregate stock repurchases (net of equity issuance) peaked at $452

billion (in 2010 dollars) in 2007 when stock markets were skyrocketing and not long before they

plummeted with the arrival of the financial crisis. An alternative possibility is that a subset of

firms may be undervalued and motivated to repurchase stock when aggregate stock market

1
valuations are high. In rejecting the latter hypothesis, Dittmar and Dittmar (2008) show that the

time-series pattern of aggregate repurchases is not explained by the relative valuation of

repurchasing firms. Their results suggest that, on the aggregate level, firms are more likely to

repurchase shares when their stock prices are high.

Moreover, in recent years, practitioners have criticized managers for buying back shares at

prices that seem too high. The Wall Street Journal made this point bluntly in an article titled

“Corporate buybacks test concept of value”:

“Warren Buffett knows a value stock when he sees it. Other executives can
struggle with the concept – particularly when it comes to their own company's
shares.
Take General Electric Co. From the start of 2005 until the end of this June, GE
bought back $29 billion dollars of stock, paying an average of $36 and change for
each share, according to regulatory filings. This week, it sold $12.2 billion worth
for $22.25 each (before fees) and put $3 billion worth of warrants, with the same
strike price, in Mr. Buffett's pocket.” (The Wall Street Journal, October 6, 2008).

The “CFO Journal” column raised a similar point in a story titled “Buying shares often brings

troubles,” which criticizes companies in the Dow Jones Industrial Average for poorly timed

buybacks:

“The 30 companies in the Dow Jones Industrial Average have spent a combined
total of about $70.6 billion this year to buy back their stock. Through Friday they
had lost 7.5% on their investment, far outstripping the industrial average's 3.9%
decline over the same period.” (The Wall Street Journal, October 12, 2011).

Bonaime, Hankins, and Jordan (2014) also raise a similar criticism in an academic study and

suggest that, conditional on the decision to repurchase shares, managers generally do not pick the

best time to do so. They examine a sample of 5,517 firms that bought back shares in at least one

quarter over 1984 and 2010, and compare stock prices and other valuation variables for the same

firm in buyback quarters and non-buyback quarters. They find that managers tend to repurchase

shares in quarters in which stock prices are high and other valuation variables are unfavorable.

2
Moreover, they show that the average annualized rate of return on each firm’s investment in its

own stock would have been almost 2% higher had managers just evenly spaced their repurchases

through all quarters during the firm’s life in the sample rather than repurchasing in some quarters

but not in others. The findings of Bonaime, Hankins, and Jordan (2014) and Dittmar and Dittmar

(2008), together with the practitioners’ criticisms, cast doubt on the market-timing explanation

for repurchases.

The purpose of this dissertation is to gauge whether or to what extent market timing is an

important explanation for managers’ repurchase decisions. Most prior studies focus on analyzing

the set of firms that have made the decision to repurchase shares and ignore firms that do not buy

back shares. This approach is problematic in that it is possible that many firms face favorable

market-timing opportunities, and yet do not exploit these opportunities by repurchasing shares.

To assess the scope of this problem, I examine possible market-timing opportunities for all

sample firms and investigate how many of them exploit stock market undervaluation through

repurchases. I also benchmark the explanatory power of market timing for the repurchase

decision against other theoretical motives for buying back stock. My approach differs from that

of Bonaime, Hankins, and Jordan (2014), who analyze firms that have made the decision to

repurchase stock and assess whether they pick the optimal time to do so. I seek to understand

how often managers take advantage of attractive market-timing opportunities via repurchases,

and whether the market-timing motive is more important than other motives in explaining

managers’ repurchase decisions.

To gauge the importance of market timing in explaining repurchases, I focus primarily on

comparing the explanatory power of market timing for the probability of a share repurchase

against a simple theory in which firms conduct repurchases to distribute their free cash flow

3
(FCF) from operations. 1 This simple hypothesis, which has its roots in basic agency theory,

predicts that firms are more likely to repurchase shares when they have high FCF.

I also benchmark the market-timing motive for share repurchases against employee-stock-

option (ESO) and leverage-rebalancing motives for buying back stock. The first motive holds

that, when firms have a larger number of ESOs outstanding, they are more likely to repurchase

shares either because managers want to substitute buybacks for dividends to avoid a value

reduction of their personally owned stock options, or because managers want to reduce the

number of shares outstanding to offset the EPS dilution impact of ESOs. The rebalancing motive

holds that, when firms have leverage ratios below their target levels, they use share buybacks to

increase their debt-to-equity ratios in the direction of their target leverage ratios.

My study presents a logit analysis of share repurchase decisions, as well as a detailed

descriptive analysis of the long-run abnormal stock return performance that follows actual

repurchases. In the stock return distribution analysis, I examine the post-repurchase abnormal

returns because prior studies argue that positive post-repurchase abnormal returns provide strong

evidence supporting the timing theory. My specific objective in the distribution analysis is to

gauge how often firms’ repurchase decisions turn out to be good or bad in a market-timing sense.

The logit regressions explain managers’ decision to repurchase stock as a function of

market-timing, FCF, ESO, and leverage variables. I use fitted logit models to predict the

probability that a firm repurchases stock in a given year, conditional on specific values of each

independent variable. The purpose is to evaluate how differences in market-timing, FCF, ESO,

and leverage variables translate to differences in the estimated probability of a share repurchase.

1
Firms could also distribute FCF by paying dividends. However, the focus in this study is not on explaining either
total payouts or the mix of dividends and repurchases. Rather, the focus is on why firms repurchase stock and, in
particular, the objective is to gauge the importance of market timing in explaining the repurchase decision.
Therefore, FCF serves as an appropriate alternative hypothesis to the market-timing timing hypothesis.

4
I use these comparisons to draw conclusions about the economic significance of the alternative

motives for buying back stock.

The main logit regressions use the market-to-book ratio (M/B), and pre- and post-repurchase

abnormal stock returns — proxies employed by prior studies — to measure market-timing

opportunities. I use the ratio of operating income before depreciation (OIBDP in Compustat) to

total assets as a proxy for FCF. To measure the ESO and leverage-rebalancing motives, I

respectively use the standardized increase in the number of outstanding ESOs and the deviation

of actual leverage from the estimated target leverage ratio.

I find that the probability that a firm repurchases stock in a given year is negatively related

to its (1) post-repurchase abnormal stock return, (2) pre-repurchase abnormal stock return, and (3)

M/B ratio. The first finding is inconsistent with the market-timing hypothesis, while the last two

findings are consistent with that hypothesis.

For three related reasons, negative coefficients on prior stock returns and M/B ratios are also

consistent with the FCF hypothesis. First, as a general matter, when firms’ investment

opportunities have dried up, they have less incentive to retain cash to fund potential investments

and are thus more likely to distribute cash to shareholders via share repurchases. Second, low-

growth firms typically have low M/B ratios, as is widely understood. And, of course, low-growth

firms have limited investment opportunities, which translate to stronger reasons to return cash to

stockholders. Third, low-growth firms might experience negative stock returns because of the

evaporation of growth opportunities. And the evaporation of growth opportunities provides more

incentive for managers to return cash to shareholders. Therefore, negative coefficients linking

repurchase decisions to prior stock returns and M/B ratios should not be viewed as conclusive

evidence supporting the market-timing hypothesis.

5
 The basic logit results also show that the probability that a firm repurchases stock is

positively related to its level of FCF, consistent with the FCF hypothesis. This finding is highly

significant, and I find that it obtains for a broad variety of repurchase measures and logit model

specifications.

The point estimates of the coefficients on the various explanatory variables are not

informative about the economic significance of the influence of market-timing versus FCF

variables on managerial decisions to repurchase stock. To assess the quantitative importance of

market timing and FCF on repurchase decisions, I investigate whether large variation in market-

timing and FCF variables translates to significant changes in the predicted probability that a firm

repurchases stock. I do so by using fitted coefficient values from logit regressions to estimate the

repurchase probability as a function of particular values of the timing and FCF variables, while

holding the other explanatory variables constant at their sample median values. I then examine

how the estimated probability of a repurchase changes as the market-timing and FCF variables

change from their sample 5th percentile values to their 95th percentile values.

In interpreting the logit results, I conservatively attribute the entire impact of variation in

M/B ratios on the repurchase decision to market timing. Although this approach biases the

analysis in favor of finding an economically more significant market-timing effect, I find that

FCF nonetheless has a stronger influence on the probability that a firm conducts a repurchase.

Variation in FCF induces both large relative and absolute changes in the repurchase probability,

while large differences in timing variables imply at most modest changes in the repurchase

probability. For example, I find that a firm with poor market-timing opportunities and high FCF

has a repurchase probability of 32.5%, while a firm with excellent timing opportunities and low

FCF has a repurchase probability of 2.6%. The former probability is more than 12 times the size

6
of the latter. And the nearly 30% absolute difference in the repurchase probability is clearly

economically material.

These estimates indicate that market timing has only a modest influence on share repurchase

decisions, and that its effect is dominated by FCF considerations. Without a high level of FCF,

firms facing excellent market-timing opportunities show only a weak tendency to exploit them

through share repurchases. For example, the data show that, more than 97% of firms facing

extremely attractive timing opportunities choose not to conduct a repurchase when they have low

FCF.

I thus conclude that the market-timing hypothesis does not provide a credible stand-alone

theory that can explain the majority of firms’ share-repurchase activities. This inference draws

further support from examination of employee-stock-option and leverage-rebalancing motives

for share repurchases.

My logit analysis indicates that ESO considerations have a material influence on repurchase

behavior, with large increases in the number of ESOs outstanding translating to a higher

repurchase probability. The data show no indication that the timing effect quantitatively

dominates the ESO effect.

I similarly find that leverage-rebalancing considerations are empirically important in

explaining repurchase decisions, and are not dominated by market-timing considerations. When

actual leverage falls below the estimated target leverage, there is a tendency to observe a higher

repurchase probability, as the rebalancing theory predicts. However, this finding is sensitive to

the model for estimating leverage targets. This is especially true when I consider tender-offer

repurchases, whose large size makes them the most plausible form of repurchases for leverage

rebalancing. In any case, the key point is that my findings on the ESO and leverage-rebalancing

7
hypotheses provide further support for the view that market timing is not the main motive

driving share repurchase decisions.

An important caveat here is that, while the above inferences are generally robust to a variety

of proxies for market-timing opportunities, the results are weaker for particular sample periods

and model specifications. In particular, when my logit analyses are extended to the period from

1971 to 2010 and exclude control variables for ESO and leverage-rebalancing considerations, the

quantitative advantage of FCF over market-timing motives becomes smaller but remains

economically material. When I focus on a subsample that consists of only tender-offer

repurchases, none of the FCF, market timing, ESO and leverage-rebalancing proxies is of first-

order importance in explaining managerial tender-offer repurchase decisions. And my logit

analyses yield no convincing evidence that any one of them is systematically more important

than the other one.

The distribution analyses of post-repurchase abnormal returns indicate that, even when

managers decide to repurchase shares, their decisions often tend to be poor in a market-timing

sense. Because the market-timing hypothesis concerns the behavior of managers, we should

focus on what most managers do with respect to repurchase decisions. The relevant research

question is therefore: “what is more likely to happen to the share prices if managers conduct a

repurchase.”

To answer this question, I examine the long-run abnormal returns following actual

repurchases in my sample (i.e., the full set of all repurchases). I find that firms are more likely to

experience abnormal stock price declines than abnormal stock increases after repurchases. In

particular, 54% of share repurchasing firms experience negative post-repurchase abnormal share

price performance. The median post-repurchase abnormal return is −3% over the 12 months

8
following the year in which a firm repurchases shares and it is −7% over the 36 months after the

year in which a firm buys back stock.

These findings are inconsistent with the market-timing hypothesis because, if managers do

time the market and buy back shares when their firms' stock is cheap, we should expect to see

firms more often experiencing positive rather than negative abnormal stock returns after

repurchases.

As with prior studies, I find that the mean post-repurchase abnormal returns are positive: 7%

over the 12 months following a repurchase and 18% over the 36 months after a repurchase.

However, it is not appropriate to view the positive mean post-repurchase abnormal stock returns

as conclusive evidence favoring the market-timing hypothesis. The reason is that the average

post-repurchase abnormal returns are not informative about what is more likely to happen after

repurchases: abnormal stock price depreciation or appreciation. This is because the average value

can be distorted by extreme values, so that we can have a positive average return even when a

majority of firms have negative share price performance. And that is exactly what I find when I

examine the full distributions of post-repurchase abnormal returns.

Specifically, the positive average post-repurchase abnormal stock returns are mostly driven

by extreme abnormal stock returns following small repurchases that account for only 6% of the

aggregate dollar value repurchased. Among all repurchases in the sample from 1985 to 2010, the

highest post-repurchase abnormal returns are 3,891% over the 12 months following the year in

which a firm repurchases shares and 4,457% over the 36 months after the year in which a firm

buys back stock. However, the lowest post-repurchase abnormal returns are limited to −100% by

the nature of the rate of return calculation. This asymmetry can and, in my sample, does generate

9
a positive average value even though more observations have negative rather than positive

returns.

I gauge the extent to which extreme values are responsible for the positive average post-

repurchase abnormal returns by examining the details of abnormal returns excluding the right

and left tails of the full distribution. When I focus on the middle 80% of the distribution, I find

that both 12-month and 36-month post-repurchase average abnormal returns are negative (−1%

in both cases). These share repurchases represent more than 94% of the aggregate dollar value

repurchased, indicating that positive average post-repurchase abnormal returns are mostly driven

by firms that repurchase small dollar amounts of stock and subsequently experience extremely

positive or negative abnormal rates of return. When I focus on the middle 90% of the distribution,

I find that the average 12-month future abnormal return is zero while the average 36-month

future abnormal return is 3.2%. This suggests that the inclusion in the sample of observations

with extreme values pushes the average future abnormal returns toward positive.

To summarize, the results from the logit analyses and distribution analyses indicate that

market-timing effects are at most weak, as they show either managers do not systematically time

the market or that their ability to time the market is poor. In supporting this main inference of my

investigation, I would emphasize three key pieces of evidence. First, the majority of firms that

face excellent market-timing opportunities fail to exploit them by conducting a share repurchase.

Second, firms that conduct share repurchases are more likely to experience abnormal stock return

decreases than increases. Third, in quantitative terms, FCF considerations show considerably

greater influence on the estimated probability of conducting a share repurchase than market

timing, while ESO and leverage-rebalancing considerations are at least as important as market-

timing considerations.

10
 The remainder of this dissertation is organized as follows. Chapter 2 discusses theoretical

explanations for repurchase decisions and reviews the related literature. Chapter 3 explains

sample procedure and presents descriptive statistics. Chapter 4 presents logit analyses that yield

my main results that market timing has only an economically modest influence on the repurchase

decision, and that its effect is markedly weaker than the FCF effect. Chapter 5 presents

robustness analyses that support the main results in Chapter 4 and additional analyses that

benchmark market-timing motives against ESO and leverage-rebalancing motives for repurchase

decisions. Chapter 6 reports the full distributions of post-repurchase abnormal stock returns and

analyzes the effect of extreme returns on the mean long-run abnormal stock returns after

repurchases. Chapter 7 reports results from supplemental analyses indicating that managers do

not use payout policy to rebalance firms’ capital structure toward a target ratio. Chapter 8

provides a short summary and discussion.

11
 Chapter 2

Theoretical Explanations for Share Repurchases

Share repurchases emerged as an economically significant corporate behavior in the early

1980s (Bagwell and Shoven, 1989). Over the last 30 or so years, the literature has viewed

managerial timing of the stock market as the primary motive driving repurchase decisions. In

addition to this market-timing hypothesis, three major motives for stock repurchases have

received prominent attention in the literature. These are the desires to (1) distribute free cash

flow, (2) offset the dilution impact of employee stock options, and (3) rebalance capital structure

toward a target leverage ratio. In this chapter, I discuss the market-timing hypothesis and each of

these three alternative motives for share repurchases. I also review the major studies on each of

them.

2.1 The Market-Timing Hypothesis

2.1.1 Three Versions of the Market-Timing Theory of Share Repurchases

Table 2.1 summarizes the main assumptions and implications of three distinct variants of the

market-timing theory that have been advanced in the literature and that potentially explain

managers’ share-repurchase decisions. These are the mispricing theory, the rational asymmetric

information theory, and the managerial perceptions theory.

The mispricing theory is the most popular market-timing theory in the repurchase literature.

Unless otherwise specified, the market-timing theory in this dissertation refers to the mispricing

variant. In this theory, investors are assumed to be irrational and sometimes make mistakes in

assessing firms’ fundamental values. Managers are assumed to be rational and to be better judges

of intrinsic values than outsiders are. Therefore when managers see their firms’ common stock is

undervalued by irrational investors, they seek to exploit investors’ pricing mistakes by

repurchasing shares. Managers gain personally from buying undervalued shares if they own

12
shares and do not sell them when the firm repurchases shares that are underpriced. So the

mispricing itself provides a motive for share repurchases. In this theory, a repurchasing firm need

not have a motive for distributing capital.

Under the mispricing theory, when repurchases are announced, stock prices are predicted to

increase immediately as investors know that managers have incentives to buy back underpriced

shares. However, because investors are irrational, they underreact on average to the news of

stock repurchases. Share prices are therefore expected to continue to increase after the buyback,

consistent with managers’ rational expectations of intrinsic share values. This mispricing version

of the market-timing theory implies that firms will repurchase shares when investors irrationally

underprice their shares and managers expect stock prices to increase in the long run.

In the rational asymmetric information variant of the market-timing theory, investors are

assumed to make unbiased forecasts about firms’ intrinsic values conditional on all public

information. As a result, no undervaluation conditional on public information should occur.

Managers are also rational about firms’ fundamental values and have more information about the

underlying economic fundamentals than outside investors have. An important assumption in this

theory is that firms have a fundamentals-based incentive for capital distribution, such as paying

out free cash flow or rebalancing the firm’s capital structure toward a target leverage ratio. Given

the incentive for distributing capital, managers time the stock market and conduct a share

repurchase when they believe current stock price is too low relative to the intrinsic value implied

by their information.

Since investors are rational in this theory, a firm’s stock price is expected to increase

immediately in response to a repurchase announcement to reflect investors’ unbiased estimates

of fair value conditional on all public information, including the newly disclosed information

13
about the repurchase itself. As a result, share prices are no longer undervalued on average once

the repurchase plan is announced. The implication is that managers cannot gain systematically

from buying undervalued shares. Thus, in order for share repurchases to occur, this theory

requires a motive for managers to buy back shares in at least some states of the world in which

the stock is overvalued. Otherwise, if managers only seek to buy back shares that are

undervalued, no rational investors would ever agree to sell shares back to the firm.

In the managerial perceptions variant of the market-timing theory, regardless of whether

investors have biased or unbiased expectation of firms’ fundamental values, managers are

overconfident and believe that they know the firm’s true value better than investors do and can

predict future stock returns. Therefore, when managers feel that their firms’ shares are

undervalued, they will seek to exploit this perceived undervaluation by buying back shares.

Similar to the mispricing theory, managers’ perceptions of undervaluation provide the motives

for share repurchases since overconfident managers believe they can exploit investors’ mistakes

and gain from buying undervalued shares. No other motives for share repurchases are needed in

the managerial perceptions version of market timing.

A key assumption of this theory is that managers on average have no real ability to predict

share price performance. This assumption implies that, although managers do buy back

undervalued shares in some cases, they are unable to do so systematically. Thus in the long run,

no positive abnormal stock returns should be observed on average after managers buy back stock.

However, in the short run, stock prices are expected to increase. The reason is that the repurchase

reveals managers’ perception that the current share price is too low. Even though managers have

no real ability to predict share prices, they are willing to buy back stock at a price that is higher

than the current price but does not exceed their expectation. Consequently, investors have a

14
temporary opportunity to sell the shares to the firm at a higher price and thus the stock market

reacts favorable in the short run. The implicit assumption here is that investors in the market

place simply take advantage of managers’ undue optimism, which leads to a market price that is

temporarily too high.

2.1.2 Prior Evidence Supporting the Market-Timing Theory

Four streams of research have reported evidence in support of the mispricing version of the

market-timing explanation.

The first stream focuses on the short-term stock returns around the announcement of

repurchases. Dann (1981) studies 143 fixed-price tender offers and Vermaelen (1981)

investigates 131 fixed-price tender offers that were conducted in 1960s and 1970s. They find that

the average cumulative abnormal return (CAR) over the period from one day prior to the

announcement to the expiration of offer is 13%. Vermaelen (1981) also examines 243 open-

market buyback authorizations in 1970s and finds a 3% average CAR during a 3-day window

centered on the day of announcement. Masulis (1980) examines 199 repurchase tender offers

announced between 1963 and 1978 and shows that the average CAR during the two-day period

around the announcement is 17%. Peyer and Vermaelen (2005) extend this stream of research by

examining a large sample of repurchase announcements from 1984 to 2001. They find a 3-day

average CAR of 2.4% for 6,470 open-market repurchase announcements, a 6% average CAR

over the period from one day prior to the announcement to the expiration of offer for 303 fixed-

price tender offers, and an 8.5% average CAR over the period from one day prior to the

announcement to the expiration of offer for 251 Dutch-auction tender offers. In addition,

Manconi, Peyer, and Vermaelen (2013) also find positive average CAR around open-market

repurchase announcements using a sample that includes firms from 31 foreign countries.

15
 The second stream of research examines the behavior of stock prices before open-market

repurchase announcements or actual open-market repurchases. Although the earlier studies in the

first stream of research also look at pre-repurchase stock price patterns, their focus is on the

abnormal returns around repurchase announcement days and they do not study the relationship

between repurchases and the change in share price in the period leading up to repurchase

announcements. In contrast, Comment and Jarrell (1991) investigate 1,197 open-market buyback

announcements from 1985 to 1988 and show that firms that announce open-market buyback

programs on average experience a 5% abnormal stock price decline over the 50 days before the

announcements. Stephens and Weisbach (1998) study a sample of 450 open-market repurchase

announcements from 1981 to 1990 as well as actual repurchases that follow those

announcements. They find that actual repurchases in one quarter are negatively related to the

performance of a firm’s stock price in the prior quarter, as well as to the cumulative stock returns

that follow the announcement of repurchase authorizations.

The third stream of empirical research investigates long-term post-repurchase abnormal

stock returns. Lakonishok and Vermaelen (1990) examine long-term price behavior after 258

fixed-price tender-offer repurchases over 1962 to 1986. They report an average abnormal excess

return of 8% (or approximately 4% per year) over the two years starting one month after offer

expiration.

Ikenberry, Lakonishok, and Vermaelen (1995) study 1,289 open-market repurchase

programs between 1980 and 1990. They find that buying shares when companies announce share

repurchase programs and holding them for four years on average generates excess returns of 12%

(or roughly 3% annually). In a subsequent study, Ikenberry, Lakonishok, and Vermaelen (2000)

examine 1,060 Canadian open-market repurchase programs. They report that Canadian

16
repurchasing firms earn an average monthly abnormal return of 0.59% (or roughly 7% per year)

over the 36-month period that follows the repurchase announcement.

Peyer and Vermaelen (2005) show that firms that conduct privately negotiated repurchases

on average experience significant positive excess returns of 6.7% in the 12-month period

following the announcement. Peyer and Vermaelen (2009) confirm the results of Ikenberry,

Lakonishok, and Vermaelen (1995) using a new sample of 3,465 open-market repurchase

announcements from 1991 to 2001. Manconi, Peyer, and Vermaelen (2013) further confirm these

results using an international sample that includes firms from 31 countries (excluding the United

States).

The last stream of research documents a negative correlation between market-to-book ratios

(M/B) and actual open-market repurchases, and interprets this finding as evidence supporting the

market-timing hypothesis. Dittmar (2002) runs Tobit regressions on repurchase decisions using

M/B ratio and several other variables for each year over 1977 to 1996. The coefficient on M/B

ratio is significantly negative in all years. She argues that this negative M/B coefficient indicates

that firms buy back shares to exploit share-price undervaluation because a low M/B ratio is an

indication that a firm is more likely to be undervalued.

2.2 The Free-Cash-Flow (FCF) Distribution Hypothesis

In basic corporate finance theory, shareholders want managers to distribute the full value of

FCF generated by firms’ investment policy over the life of the corporation (DeAngelo,

DeAngelo, and Skinner, 2008, Chapter 2). Moreover, Jensen (1986), Stulz (1990), and many

others argue that managers have incentives to retain excess cash because doing so enables them

to derive personal benefits from investing in projects that benefit themselves at the expense of

outside investors. The prospect of managerial abuse of retained resources leads outside investors

17
to pressure managers to pay out greater amounts of FCF. Firms that are in the late stage of their

lifecycles and have fewer growth opportunities are more likely to pay out FCF as they would

otherwise have too much cash, which can create an opportunity for managers to waste company

resources (DeAngelo, DeAngelo, and Stulz, 2006; Fama and French, 2001).

Investors interpret the announcement of managerial decisions to repurchase stock as good

news for firms that are likely to have high agency costs. Consistent with this prediction, Nohel

and Tarhan (1998) find that announcement of repurchase tender offers are associated with share

price changes that are higher among those firms that are estimated to have more excess cash or

few growth opportunities. Grullon and Michaely (2004) report similar findings for open-market

repurchases.

When managers decide to pay out cash, they can do so by paying dividends or buying back

shares. Unlike regular cash dividends, share repurchases are not implicit commitments by firms

to continue to pay out cash in future periods. Consequently, researchers such as Jagannathan,

Stephens, and Weisbach (2000) and Guay and Harford (2000) suggest that share repurchases can

be used to distribute cash flows that are unlikely to reoccur, i.e., transitory cash flows. These

authors also report evidence supporting the view that transitory cash flows are important

determinants of stock buybacks. For example, Jagannathan, Stephens, and Weisbach (2000) find

that repurchasing firms have higher transitory cash flows (as proxied by non-operating income)

than non-repurchasing firms. Guay and Harford (2000) also report evidence that firms tend to

use open-market repurchases to distribute transitory cash flows.

18
 2.3 The Employee-Stock-Option (ESO) Hypothesis

There was a surge in both share repurchases and the use of ESOs during the 1990s. The

latter trend is documented in Murphy (1999). Several studies have attributed the growth of share

repurchases to the use of ESOs.

Early studies on ESOs and share repurchases build on Lambert, Lanen, and Larcker (1989),

who suggest that employee-stock-option plans give managers incentives to substitute buybacks

for dividends. Specifically, paying dividends reduces the price of common stock and thus the

value of an option to buy that stock. In contrast, if a firm uses the same amount of cash as it

would have paid as dividends to buy back its own shares at a fair price, the market price of the

stock would not decline following the repurchase, nor will the stock option value. Therefore

managers who have more stock options are more likely to buy back shares rather than pay

dividends.

Consistent with this general view, Jolls (1996) reports that firms which rely more heavily on

stock-option compensations are more likely to buy back their shares, while Fenn and Liang

(2001) find that firms repurchase more shares when their managers own more ESOs.

ESOs can also affect repurchase decisions because of their impact on the “diluted EPS”

results that analysts use to evaluate the earnings disclosures that firms make in their annual

reports and related quarterly filings with the SEC. Diluted EPS is a measure of earnings per share

that is often used by investors and analysts for valuation purposes. Many studies find that the

level of diluted EPS is more highly associated with stock prices than the level of basic EPS is

(Core et al. 2002; Jennings et al., 1997). When a firm grants ESOs, there is an increase in the

number of shares that are used in computing fully diluted EPS, which implies a reduction in the

reported figure for diluted EPS.

19
 Managers who care about their reported diluted EPS figures and have a large number of

ESOs outstanding have incentives to repurchase shares to bolster diluted EPS. The simple reason

is that buying back shares reduces the denominator in the calculation of diluted EPS and thus

increases the diluted EPS figure that investors see. Weisbenner (2000) and Bens et al. (2003)

provide empirical evidence that firms repurchase their shares to reverse the impact of ESOs on

diluted EPS.

Another ESO-related motive for stock repurchases is that an increase in the number of ESOs

outstanding increases a firm’s need to have shares available to fund employee stock option

exercises. Kahle (2002) suggests that a share repurchase is one way to fund the exercise of ESOs.

The idea is that managers might choose to repurchase shares in the open market (over issuing

new shares) because a share repurchase does not increase the number of shares outstanding and

thus reduces or avoids the dilution of the basic earnings per share (EPS) figure that firms report.

Consistent with this view, Kahle finds that firms are more likely to buy back stock when

employees have a large number of options currently exercisable, and that the amount of stock

that is repurchased is positively related to the total number of options that are exercisable.

2.4 The Leverage-Rebalancing Hypothesis

Holding everything else equal, a share repurchase reduces the equity value of a firm and

thus results in a higher debt-to-equity ratio for firms that have debt outstanding. Bagwell and

Shoven (1988) and Hovakimian, Opler, and Titman (2001) suggest that if firms have an optimal

capital structure, they can buy back their shares to increase their leverage ratios toward the target

capital structure.

Direct evidence on the leverage-rebalancing hypothesis has been reported in Lie (2002),

who analyzes capital structures around 286 repurchase tender offers from 1980 to 1997 and uses

20
a linear regression model to estimate a firm’s target leverage ratio. He finds that firms generally

have debt-to-equity ratios below the estimated target level before repurchase tender offers. He

also reports that on average leverage ratios following tender-offer repurchases are close to or

above the predicted target level.

A few studies report indirect evidence that share repurchases are used to rebalance leverage

ratios toward targets. Dittmar (2000) shows that repurchasing firms usually have lower debt-to-

equity ratios compared to their industrial peers. Hence repurchases have a tendency to move a

firm’s leverage ratio toward that of firms in the same line of business, which is suggestive of

rebalancing if peer leverage is taken as an estimate of a firm’s target leverage ratio. Masulis

(1980) and Vermaelen (1981) find that announcement stock returns are higher when repurchase

tender offers are mostly financed by new debt instead of cash. Their findings are consistent with

the notion that under-leveraged firms are more likely to use new debt than cash to fund a tender

offer. The reason is that debt-financed tender offers imply a larger increase in leverage than

tender offers that are financed by paying out cash on hand2. Hence debt-financed tender offers

are more effective in increasing firms’ capital structures toward target leverage ratios. As a result,

debt-financed tender offers should, in theory, be associated with higher announcement returns

because they effectively move under-leveraged firms’ capital structures toward their targets as

implied by leverage-rebalancing theories.

2
Holding everything else equal and assuming that a firm has some debt outstanding, an equity payout increases a
firm’s leverage ratio. Assuming that the leverage ratio is always positive and less than one, a debt-financed equity
payout implies a larger increase in leverage than implied by an equity payout financed by cash on hand, given the
same amount of equity payout. Let TA and D respectively denote total assets and total debt before the equity payout,
and let C denote the amount of equity payout. In the case of a debt-financed payout, C also equals the incremental
debt taken on by the firm. We need to prove that (D+C)/TA is always larger than D/(TA-C). The assumption that the
leverage ratio is always less than one implies that TA > C + D > 0. This is obviously equivalent to TA-D-C > 0.
Multiplying both sides by C and then adding TA∙D to both sides, we obtain TA∙D+TA∙C-D∙C-C∙C > TA∙D. After
rearranging and simplifying, we have (TA-C)∙(D+C) > TA∙D. Dividing both sides by (TA-C)∙TA, we obtain
(D+C)/TA > D/(TA-C).

21
 Chapter 3

Sample Selection and Descriptive Statistics

I analyze common stock repurchases conducted by US industrial firms defined as those

firms from the Compustat and Center for Research in Security Prices (CRSP) merged file that (1)

have four-digit SIC codes outside the range 4900-4949 (utilities) and 6000-6999 (financial firms),

(2) have a CRSP share code of 10 or 11 (common shares), (3) are incorporated in the U.S., and

(4) have nonmissing values on Compustat for total assets.

Following Fama and French (2001) and Skinner (2008), I measure the magnitude of a firm’s

annual share repurchase volume as the net repurchase amount, which is defined as the increase in

common treasury stock amounts according to Compustat. If the firm does not use the treasury

stock method (i.e., treasury stock amount is zero in both the current and the prior year, or not

available), I follow these authors’ approach of using the difference between the stock purchase

and stock issuance amounts from Compustat to proxy for the net repurchase amount.

Figure 3.1 depicts the aggregate dollar volume of share repurchases and common share

dividends for 14,958 sample firms from 1971 to 2010. Share repurchases were rare before the

adoption of SEC rule 10b-18 in November 1982 that delineated a “safe harbor” for repurchasing

companies from being sued for price manipulation. After that regulation was adopted, net

repurchases surged in the 1980s and experienced rapid growth until reaching a peak of $452

billion (in 2010 dollars) in 2007 when the S&P 500 also reached its historical high. During this

period, the aggregate dollar volume of dividends also increased but at a much slower speed, and

was first surpassed by the aggregate dollar volume of repurchases in 1998.

Dittmar and Dittmar (2008) find that aggregate repurchase volume is positively correlated

with the overall level of stock market prices, i.e., managers spend a large amount on share

22
repurchases when aggregate stock market valuations are high and spend a small amount of cash

in repurchases when aggregate stock market valuations are low. Figure 3.1 confirms this finding.

This behavior seems difficult to reconcile with the idea that managers generally do a good

job buying back stock when it is inordinately cheap. At a minimum, the data suggest that, on the

aggregate level, managers often buy back stock after it has appreciated in value.

On the individual-firm level, Bonaime, Hankins, and Jordan (2014) examine firms that

repurchase shares and find that the timing of a firm’s repurchase tends to come in quarters when

its share price is high. They report that the average stock price and market-to-book ratios in

repurchasing quarters are significantly higher than those in non-repurchasing quarters.

The findings of Bonaime, Hankins, and Jordan (2014) and Dittmar and Dittmar (2008) cast

doubt on the market-timing hypothesis, as the market-timing explanation indicates that firms

should repurchase less frequently and in smaller amount when stock prices are high, i.e., when

timing opportunities are poor. The evidence here and in these two prior studies indicates the

opposite is true.

Figure 3.2 depicts the percent of firms that repurchase shares and the percent of firms that

pay dividends in a given year. The percent of firms that buy back stock also increased after 1982,

and eventually exceeds the percent of firms that pay dividends in 1997. It reaches its historical

high in 2008, the year after the aggregate dollar volume of repurchases peaked. Nearly 43% of

sample firms in 2008 spent some cash on share repurchases. This pattern largely reflects

repurchases undertaken before the onset of the financial crisis, as few firms bought back shares

after the stock market fell in October 2008. The fact that many firms buy back shares before a

financial crisis is inconsistent with the timing theory, since the latter theory indicates that firms

should not repurchase stock before a major stock market decline.

23
 Table 3.1 reports year-by-year repurchase amounts in two groups: tender-offer repurchases

and all repurchases. During the period between 1985 and 2010, there are 600 tender-offer

repurchases in my sample. These tender offers account for approximately 2% of the 30,379 share

repurchases in the whole sample. However, they account for about 5% of the aggregate dollar

volume repurchased. This indicates that tender offers are typically large in terms of the dollar

amount repurchased: the average (median) repurchased dollar amount in a tender-offer

repurchase is $269 ($44) million compared to $95 ($2) million for all repurchases during my

sample period.

Table 3.2 further documents that the percent of shares repurchased in tender offers is

considerably larger than the percent repurchased in most buybacks. I find in particular that 44%

of all share buybacks entail a repurchase of no more than 1% of total shares outstanding. The

majority of repurchases (81%) buy back no more than 5% of total shares. In contrast, the

majority of tender-offer repurchases (78%) buy back more than 5% of the total shares that are

outstanding. Moreover, more than half of tender-offer repurchases (59%) buy back at least 10%

of total shares outstanding. The mean (median) percent of shares repurchased is 18.7% (12.9%)

in tender offers and it is 3.4% (1.3%) in all repurchases.

Because of their large size, tender-offer repurchases would seem more likely to occur than

other repurchases when market-timing opportunities are highly attractive. The reason is that,

when extremely good timing opportunities arise, managers should take advantage of them by

repurchasing a large amount of shares as predicted by the market-timing theory. The implication

is that market-timing effects should be more prominent for tender-offer repurchases because

tender offers are generally used to buy back unusually large amounts of stock. Consistent with

24
this conjecture, Comment and Jarrell (1991) report that tender-offer repurchases are associated

with higher announcement-period stock market returns than open-market repurchases are.

This line of reasoning and prior evidence both suggest that, if managers do time the market

by buying back underpriced stock, the market-timing effect is more likely to be detected for the

tender-offer subsample than for full sample of repurchases. In structuring my analysis, I

conservatively emphasize results for the full sample because tender-offer repurchases occur

rarely and involve especially large buybacks and thus may not be fully representative of

repurchases generally. I use the tender-offer subsample to conduct robustness checks on the

inferences drawn from the full sample.

It is not unreasonable to use tender-offer repurchases, which constitute a small fraction of all

repurchases, to run a robustness check on my main tests. In support of this approach, I note that

the role of tender offers in repurchases is similar to the role of seasoned equity offerings (SEOs)

in the issuance of equity. SEOs are generally large equity-issuance events as a SEO on average

raises over $115 million. Even though SEOs only represent a small portion of total equity issues

(7% of total number of issuances and 10% of aggregate dollar amount of issuances) as shown in

Table 3.3, they have attracted wide attention and a large body of literature has been devoted to

explain why firms conduct SEOs.

25
 Chapter 4

Logit Analysis of the Decision to Conduct a Share Repurchase

In this chapter, I present my findings on the determinants of managers’ decisions to

repurchase common stock. In addition to market timing, researchers have proposed three major

motives for conducting stock repurchases: free-cash-flow (FCF) distribution, the offset of

employee-stock-option (ESO) dilution, and rebalancing the firm’s leverage ratio toward a target

debt-equity mix. The focus in this chapter is on a comparison of the market-timing and FCF

distribution hypotheses, while ESO and leverage-rebalancing considerations are included in the

logit models as control variables. I gauge and compare the economic effects of variation in

market-timing and FCF proxies on managerial share-repurchase decisions. Supplemental

analyses that gauge the magnitudes of ESO and leverage-rebalancing effects are presented in

Chapter 5.

The approach used in this chapter is similar to that employed by DeAngelo, DeAngelo, and

Stulz (2010) to examine the importance of the market-timing hypothesis in explaining firms’

decision to conduct seasoned equity offerings (SEOs), i.e., to sell large amounts of common

stock.

Unlike SEOs, most stock repurchases cannot be measured as discrete events that occur at a

precisely identified point in time. Consequently, standard logit regressions must be adapted so

that they apply to repurchase decisions. In doing so, I use the general approach of Huang and

Ritter (2009), who apply logit regressions to analyze general equity issuances. The equity

issuances that Huang and Ritter study are similar to most stock repurchases in that a researcher

cannot identify a precise time that they take place. In the Huang and Ritter study, a firm is

26
defined as an issuer of common stock in a given year if the change in book equity is at least 5%

of the prior year’s total assets.

In implementing the Huang and Ritter’s approach, I define a firm as one that repurchases

stock in a given year if it buys back at least 1% of the total number of outstanding shares in that

year. This approach enables me to conduct logit analyses on general repurchases to assess the

relative importance of market timing and FCF on the repurchase decision. Compustat does not

report the number of shares a firm buys back in a given year, and so I use the ratio of net

repurchases to the total market value of equity at the end of the prior year as my proxy for the

percent of shares a firm buys back. Since data that are needed to compute numbers of ESOs

outstanding are available on Compustat only after 1996, the sample period in this chapter is from

1996 to 2010.

4.1 Basic Logit Tests

I run logit regressions to assess whether the probability that a firm conducts a share

repurchase is negatively related to its market-to-book ratio (M/B) and recent abnormal stock

returns, and positively related to its future abnormal stock returns, as predicted by the market-

timing hypothesis. I use abnormal stock returns in these logit analyses as they are used as proxies

for timing opportunities in many prior studies that examine the timing hypothesis (e.g.,

Lakonishok and Vermaelen, 1990; Ikenberry, Lakonishok and Vermaelen, 1995 and 2000). I also

include the M/B ratio as an index of mispricing since it is the focus of the market-timing analysis

in Dittmar (2000). I use the realized future abnormal stock returns as a measure of managers’

expectations of firms’ future stock price performance. I do so because market-timing theories

assume that managers have the ability to predict future stock returns. If this hypothesis is

descriptive, then firms’ ex post (realized) returns should be positively correlated with managers’

27
ex ante expected returns. This approach is analogous to that used in Baker, Stein, and Wurgler

(2003) and Huang and Ritter (2009). Alternative proxies for market-timing opportunities are also

used to run robustness checks, with details reported in Chapter 5.

The logit regressions also assess whether the probability that a firm conducts a share

repurchase in a given year is positively correlated with the level of FCF and with increases in

number of ESOs outstanding, and is negatively correlated with deviations from the estimated

target leverage ratios.

For each firm, I compute the standardized M/B ratio as the raw M/B ratio divided by median

M/B ratio for all firms in the year in question. The prior 12-month (36-month) stock returns are

defined as the market-adjusted abnormal stock returns over the one-year (three-year) period

ending immediately before the year in question. The future 12-month (36-month) stock returns

are defined analogously for the one-year (three-year) period following the year in question.

Abnormal stock returns are calculated as the firm’s actual stock return minus the

contemporaneous return on the value-weighted market index. I use the standardized operating

cash flow (OCF), the ratio of operating income before depreciation (OIBDP in Compustat) to

total assets, as a proxy of FCF. The logit regressions use the lagged value of this FCF proxy, i.e.,

the value at the end of the fiscal year prior to the year in question. I also use the ratio of operating

income before depreciation (OIBDP in Compustat) to total assets minus the median capital

expenditure to assets ratio in the same industry to proxy for FCF. The results remain qualitatively

unchanged (see Section 4.2).

To control for ESOs motives to repurchase stock, I compute the standardized change in the

number of ESOs outstanding as the difference between the ratios of Compustat options

outstanding to common shares outstanding in the current and the prior year. To control for

28
leverage-rebalancing considerations, I compute the deviation from a firm’s estimated target

leverage ratio (hereafter, the deviation from target) as the difference between the firm’s debt-to-

assets ratio and an estimated target leverage ratio at the end of the fiscal year prior to the year in

question. The target leverage ratio is the fitted value from a linear regression of the debt-to-assets

ratio on variables often hypothesized to affect leverage decisions: log(sales), market-to-book

ratio, profitability and asset tangibility (Rajan and Zingales, 1995).

My logit regressions use data on firms’ decisions to conduct or not conduct a share

repurchase. I pool observations over 1996 to 2010 and compute standard errors clustered by both

firm and time (per Petersen, 2009). The dependent variable equal one if a firm repurchases at

least 1% of its shares in a given year, and zero otherwise. The explanatory variables include the

firm’s most recent standardized M/B ratio, its most recent and future 12-month (or 36-month in

some tests) abnormal stock returns, and its level of FCF. The standardized change in the number

of ESOs outstanding, and the deviation from the target leverage ratio are included as control

variables. To control for firm size, all logit regressions also include the natural log of total assets

at the end of the fiscal year prior to the year in question.

Table 4.1 reports basic logit results, with the rows of the table differing in (1) the measure of

market-timing variables, and (2) the inclusion of FCF, ESO, and leverage-rebalancing proxies as

explanatory variables. When only market-timing variables are included in the logit regressions,

Rows A through C show that the sign of coefficients on each timing variable is consistent with

the market-timing hypothesis, with the estimated repurchase probability negatively related to

M/B ratios and recent stock returns, and positively related to future stock returns. However, only

the coefficients on recent stock returns in Rows A and B are statistically significant (t-statistics =

−2.36 and −2.68).

29
 Importantly, I would note that negative coefficients on prior stock return and M/B ratios are

also consistent with the FCF hypothesis. There are three related reasons. First, as a general

matter, when firms face limited investment opportunities, they have less incentive to retain cash

to fund future investments. As a result, they are more likely to distribute cash to shareholders via

repurchases. Second, low-growth firms usually have low M/B ratios, since their poor investment

opportunities imply low market valuations. And, of course, their limited investment opportunities

also imply a stronger incentive to pay out cash to stockholders. Third, low-growth firms are

likely to experience negative stock returns because of the evaporation of growth opportunities,

which also provides a stronger incentive for managers to return cash to shareholders. Therefore,

negative coefficients linking repurchase decisions to prior stock returns and M/B ratios should

not be viewed as conclusive evidence supporting the market-timing hypothesis.

When FCF, ESO, and leverage-rebalancing proxies are included in logit regressions, Rows

D and E of Table 4.1 show that the coefficients on future abnormal returns are positive and

statistically indifferent from zero. The findings regarding M/B ratios and prior abnormal returns

are qualitatively unchanged.

More importantly, Rows D through F show that the level of FCF (as proxied by standardized

OCF) is positively and significantly correlated with the estimated probability that a firm

conducts a repurchase (t-statistic > 8.00 in all rows). The positive coefficients on FCF indicate

that firms with higher FCF are more likely to repurchase shares, consistent with the FCF motive

for repurchasing shares.

To provide a quantitative feel for how small changes in each explanatory variable translate

to changes in the estimated probabilities of a share repurchase, Table 4.1 also reports marginal

probabilities of a share repurchase with each explanatory variable at its sample median value.

30
These marginal probabilities are changes in repurchase probabilities that are induced by a small

(one-unit) change in the explanatory variable. I find that, when evaluated at the median value,

FCF has a much larger marginal probability of a share repurchase than timing variables. For

example, when standardized M/B ratios and 12-month abnormal returns are included in the

model, the marginal probability of a share repurchase is 1.00 when estimated at the median value

of FCF. This number is more than 50 times larger in absolute value than the marginal probability

of −0.02 when estimated at the median value of M/B ratio.

The basic logit regressions in Table 4.1 also yield findings that support the ESO and

leverage-rebalancing hypotheses. Coefficients on the standardized change in the number of ESOs

are positive and statistically significant, and coefficients on deviations from estimated target

leverage ratios are negative and statistically significant. These findings suggest that an increase

in ESOs outstanding and the deviation below the estimated target leverage ratio are all associated

with a higher probability of a share repurchase.

4.2 Relative Impact on the Repurchase Decision of Market Timing versus FCF

To assess the relative importance of market-timing versus FCF variables in explaining

managerial share-repurchase decisions, Table 4.2 reports the estimated probability of a share

repurchase conditional on specific hypothesized values of the timing and FCF variables, while

holding the other variables constant at their sample median values. Except for the estimated

probabilities in the far right column, the probabilities in Panel A of Table 4.2 are calculated using

the model in Row D of Table 4.1. The estimated probabilities in the far right column are based

on the model in Row A of Table 4.1, which only accounts for the timing effect. I begin by

discussing this last column to provide an assessment of the effect of market-timing variables.

31
 My full sample contains 11,058 repurchases that were conducted between 1996 and 2010.

This is about 21% of all firm-year observations during the sample period. Thus the probability

that a randomly selected firm conducts a share repurchase in a given year is around 21%. This

estimate is close to the estimated probability of 21.5% that a firm with neutral market-timing

opportunities (all timing variables equal their sample median values) conducts a share repurchase

(the last number in Row 1 in Panel A of Table 4.2). I find that the estimated repurchase

probability changes very little when the firm has recently experienced an extremely positive

abnormal return instead of an extremely negative abnormal return (compare the last numbers in

Rows 2 and 3). Moreover, when I consider a shift in future abnormal return from its 5th

percentile value to its 95th percentile value, the estimated probabilities virtually do not change

(Rows 4 and 5). When the standardized M/B ratio decreases from its 95th to 5th percentile values,

the estimated probability also changes very little (Rows 6 and 7).

Holding M/B ratio and past abnormal stock return constant at very high or very low levels, a

large swing in future abnormal stock return (from the 5th to the 95th percentiles) almost does not

change the probabilities of a share repurchase (Rows 8 through 13). However, the probability

that a firm repurchases stock is roughly 7% higher when a firm faces extremely favorable

market-timing opportunities (i.e., standardized M/B ratio and prior abnormal stock return both

equal their 5th percentile values, and future abnormal stock return equals its 95th percentile value)

than when it faces extremely unfavorable market-timing opportunities (standardized M/B ratio

and prior abnormal return equal their 95th percentile values, and future abnormal return equals its

5th percentile value). The estimated probability increases from 16.7% to 23.8%, implying of

every one hundred firms, seven additional firms decide to buy back stock when face extremely

32
favorable rather than extremely poor market-timing opportunities (repurchase probability

increase of 7.1% = 23.8% − 16.7%, Rows 14 and 15).

The latter estimates indicate that most firms that face extremely favorable market-timing

opportunities do not conduct share repurchases. Of every one hundred firms facing extremely

good timing opportunities, roughly 76 firms fail to exploit market undervaluation by conducting

a share repurchase (repurchase probability of 23.8%, Row 15 in Panel A of Table 4.2, far right

column).

The estimates above imply that market-timing opportunities only marginally influence

managerial repurchase decisions, and that the market-timing hypothesis is at best, an incomplete

theory that requires modification to explain why so many firms do not conduct share repurchases

when they face excellent market-timing opportunities.

These comparisons are incompatible with the view that market timing is the main motive

driving the decision to repurchase stock, as posited by Vermaelen (2005), and with the

mispricing version of Baker and Wurgler’s (2002) timing theory. I conclude that market timing

is not a first-order determinant of managerial repurchase decisions.

This inference is reinforced by the fact that my data also show that market timing is

dominated by FCF considerations in explaining managers’ share-repurchase decisions. When I

estimate the repurchase probability as a function of both market-timing and FCF variables, FCF

shows a much greater effect on the estimated repurchase probability than the proxies for timing

opportunities do.

The strength of the FCF effect on the estimated repurchase probabilities is evident in the

middle columns in Panel A of Table 4.2. Row 1 shows that, for firms with neutral market-timing

opportunities, the estimated repurchase probability increases from 2.3% to 46.4% when the level

33
of FCF changes from very low (5th percentile) to very high (95th percentile). The latter number is

more than 20 times the size of the former number. Moreover, the absolute change is also

economically significant (44.1% = 46.4% − 2.3%). Similar results hold for all other market-

timing scenarios in Panel A of Table 4.2 (middle columns, Rows 2 through 15).

The relative importance of market-timing opportunities and FCF in explaining repurchase

decisions is best assessed by comparing Rows 14 and 15 in Panel A of Table 4.2. This

comparison shows that a firm with high FCF and highly unfavorable timing opportunities has a

repurchase probability of 32.5% (Row 15), which is more than 12 times the size of the 2.6%

repurchase probability for a firm with low FCF and highly attractive timing opportunities (Row

14). The absolute difference in the repurchase probability is nearly 30% (29.9% = 32.5% −

2.6%). Hence the FCF effect quantitatively dominates the market-timing effect.

As discussed in Section 4.1 and as is well understood, although some studies (e.g., Dittmar,

2000) include the M/B ratio as an index for undervaluation, M/B ratio also measures other

factors such as investment opportunities. Consequently, it is difficult to provide an unambiguous

interpretation of the coefficient on M/B ratio. For example, the negative M/B coefficient is also

consistent with the FCF hypothesis because a lower M/B ratio implies poorer investment

opportunities and thus higher level of FCF. Therefore a firm with a lower M/B ratio may

repurchase its own shares to distribute FCF rather than to exploit market undervaluation.

In order to provide a better comparison between the FCF and the market-timing hypotheses,

I repeat the logit analyses as discussed above except that M/B ratio is not included in the logit

regressions. I highlight these tests that include only stock-return measures for undervaluation

because they are employed by Lakonishok and Vermaelen (1990) to support the market-timing

explanation for share repurchases. The results are reported in Panel B of Table 4.2.

34
 Qualitatively, the results do not change. Recent abnormal stock returns show modest

influences on the estimated repurchase probability, while future abnormal stock returns show

little impact. On the other hand, FCF has a much larger effect and continues to dominate the

market-timing effect. For example, for every one hundred firms with neutral future abnormal

stock returns, roughly five additional firms decide to conduct a share repurchase when they have

recently experienced an extremely negative rather than an extremely positive abnormal return

(Rows 2 and 3 in Panel B of Table 4.2).

On the other hand, when a firm faces extremely unfavorable market-timing opportunities, if

the level of FCF is very high, the repurchase probability is 32.4% higher (or more than 11 times

larger) than when it faces extremely attractive market-timing opportunities and has a low level of

FCF (35.5% versus 3.1%, Rows 12 and 13). Similar to the results in Panel A, both the relative

and the absolute changes in the repurchase probability are economically significant. My findings

here continue to indicate that the market-timing effect is dominated by the FCF effect.

The results in this section are robust to a variety of sensitivity checks. Some of them are

briefly reported here, and others are reported in Chapter 5. I would note in particular that the

logit regression results do not change qualitatively when I define a firm as one that repurchases

stock if it buys back at least 0.5% of the total number of outstanding shares in a given year

(details not tabulated). Also, the results are robust to adjusting operating cash flow by deducting

the industry median capital expenditure (Tables 4.3 and 4.4). The basic logit results are also

robust to including firm and year fixed effects.

To test whether firms use share repurchases to distribute transitory cash flow, I include both

the transitory cash flow and FCF proxies in logit regressions. I follow Jagannathan, Stephens,

and Weisbach (2000) and use the lagged values of non-operating income (scaled by the total

35
assets) to proxy for transitory cash flow. These logit regressions show that the transitory cash

flow has a statistically significantly positive coefficient (Table 4.5). Although this positive

coefficient indicates that firms with more transitory cash flow are more likely to conduct a share

repurchase, the economic magnitude is small. Large variation in non-operating income translates

to only minor changes in the estimated repurchase probabilities (Table 4.6). For example, for a

firm with neutral market-timing opportunities, a shift in non-operating income from its 5th to its

95th percentile values only induces an increase in the repurchase probability of 2.7% (2.7% =

31.7% − 29.0%, Row 1 in Panel A of Table 4.6).

Finally, my logit analyses explore whether the level of cash balances is an important

determinant of managers’ repurchase decisions by including both cash-holding and FCF

variables in logit regressions. The results show that both cash holdings (cash plus short-term

investments/assets) and excess cash holdings (defined as in Opler, Pinkowitz, Stulz, and

Williamson, 1999) exert statistically and economically significant positive effects on repurchase

decisions. And their effects are quantitatively similar to market-timing effects (Tables 4.7

through 4.10).

In particular, Rows E through F in Table 4.7 and Table 4.9 indicate that the probability that

a firm conducts a repurchase is significantly and positively related to its cash holdings and its

excess cash holdings (t-statistic > 5.75 in all rows), after controlling for FCF, ESO, and leverage-

rebalancing proxies. Panel A of Table 4.8 shows that a firm with poor market-timing

opportunities and high cash holdings has a repurchase probability of 28.7%, which is close to the

28.4% repurchase probability for a firm with good market-timing opportunities and low cash

holdings (Row 14 and 15). The comparable figures in the case of excess cash holdings are 29.5%

and 28.7% (Rows 14 and 15 in Panel A of Table 4.10).

36
 4.3 The Interaction between Free-Cash-Flow and Market-Timing Variables

The logit regression results in Table 4.1 could be consistent with a conditional market-

timing theory in which managers time the market to conduct repurchases when firms have a high

level of FCF as proxied by standardized operating cash flow (OCF). To test this hypothesis, I

include the interaction terms between FCF and market-timing variables in logit regressions

structured similarly to those analyzed above. Since only the M/B ratio and the prior stock return

exert economically significant effects on repurchase decisions, I focus on the interaction effects

of FCF with the M/B ratio and the recent abnormal stock returns. As emphasized by Ai and

Norton (2003), standard statistical software packages do not correctly compute the marginal

effects of interaction terms in Logit models. Ai and Norton have also made available a Stata

program, INTEFF, to estimate the interaction effect in Logit model. The interaction effects

computed by INTEFF are reported in Table 4.11.

According to the conditional market-timing theory, when firms have a high level of FCF,

firms that have a greater degree of undervaluation as measured by lower M/B ratios and recent

stock returns are more likely to conduct a share repurchase. Therefore, the interaction between

FCF and the M/B ratio, as well as the interaction between FCF and the prior stock return, should

exert negative effects on the probability of conducting a share repurchase.

However, Table 4.11 shows that both interaction terms have statistically insignificantly

negative effect on managers’ repurchase decisions. Moreover, the magnitudes of the estimated

interaction effects are small. For a firm with a median standardized FCF of 0.10, the interaction

effect between the M/B ratio and FCF is −0.004 (−0.044 × 0.10), significantly smaller than the

marginal effect of FCF. These interaction effects even turn positive when the estimated

repurchase probability is large (see Figure 4.1).

37
 These estimates show almost no support for the conditional market-timing theory. Therefore,

the results in this section further support my main conclusion that free cash flow is markedly

more important than market timing in explaining managerial repurchase decisions.

38
 Chapter 5

Robustness Checks and Supplemental Analyses

In this chapter, I present robustness checks and supplemental analyses that support the main

inferences drawn in Chapter 4. Specifically, the findings in this chapter confirm that market

timing has only an economically modest influence on managers’ repurchase decisions, and its

effect is quantitatively dominated by the desire to distribute free cash flow.

5.1 Alternative Measures of Market-Timing Opportunities

Empirical support for the market-timing theory of share repurchases is potentially sensitive

to measures of market-timing opportunities. To assess the robustness of my inferences to

alternative measures of market-timing opportunities, I re-run the logit regressions of Chapter 4

using (i) the mispricing index in Rhodes-Kropf, Robinson, and Viswanathan (RRV, 2005), (ii)

the mispricing index in Polk and Sapienza (PS, 2009), and (iii) RRV and PS indices with

standardized M/B ratio, and prior and future abnormal stock returns.

Table 5.1 reports repurchase probabilities based on logit models that include the RRV or PS

mispricing indices or both as explanatory variables. For comparison purposes, Panel A of Table

5.1 repeats the key results in Rows 14 and 15 in Panel A of Table 4.2. The estimated

probabilities in Panels B and C are based on models that only use the RRV or PS indices as a

proxy for timing opportunities. Panels D, E, and F report the estimated repurchase probabilities

based on logit models that include M/B ratio, prior and recent stock returns, together with the

RRV and PS indices. In these panels, I take the 95th percentile value of the RRV (PS) index as

indicative of highly unfavorable timing opportunities and the 5th percentile value of the RRV (PS)

index as indicative of highly favorable timing opportunities. All other variables are set as in

Table 4.2. For example, the last row in Table 5.1 (Panel F) reports the estimated probability of a

39
share repurchase as a function of operating cash flow for a firm that, by every measure, has

extremely attractive timing opportunities: M/B ratio, prior stock return, RRV and PS indices at

the 5th percentile values, and future stock return at the 95th percentile value.

Table 5.1 indicates that the estimated repurchase probabilities are close to those in Table 4.2

under all different model specifications with either the RRV or the PS index (or both) included.

For example, Panel F of Table 5.1 shows that the repurchase probability is 32.8% for a firm with

high FCF and highly unfavorable timing opportunities. This is more than 10 times the size of the

3.1% probability for a firm with low FCF and highly favorable timing opportunities. The

absolute differential of 29.7% is also large. The comparable figures in Panel A are 32.5% and

2.6%, with the former figure more than 12 times the size of the latter figure and an absolute

differential of 29.9%. Similar results are reported in Panels B through E.

Overall, the results in Table 5.1 indicate that market-timing considerations are quantitatively

dominated by FCF motives in explaining managerial repurchase decisions, with timing

opportunities measured by different mispricing proxies that have been used in the literature.

The same conclusion follows from similar logit analyses that adjust operating cash flow by

the median capital expenditure in the same industry, as documented in Table 5.2. For example,

Panel F of this table shows that the repurchase probability is 32.4% for a firm with high FCF and

highly unfavorable timing opportunities. This is more than 14 times the size of the 2.3%

probability for a firm with low FCF and highly favorable timing opportunities.

5.2 Logit Analysis with the Sample Period Expanded to 1971-2010

The sample period in Chapter 4 is from 1996 to 2010 because employee-stock-option (ESO)

data are available only after 1996. However, data that are needed to compute net repurchase and

free cash flow (FCF) are available on Compustat after 1971. Therefore, my logit analyses in

40
Chapter 4 can be extended to the period from 1971 to 2010 with only timing and FCF proxies

included.

As in Section 4.1, I run logit regressions to assess whether the probability that a firm

conducts a repurchase is negatively related to its standardized market-to-book ratio (M/B) and

recent abnormal stock return, and positively related to its future abnormal stock return and the

level of FCF (proxied by standardized operating cash flow). All these explanatory variables are

defined as in Chapter 4. All logit regressions also include the natural log of total assets as a

control for firm size and exclude proxies for ESO and leverage-rebalancing motives.

Table 5.3 reports the basic logit results that analyze managerial repurchase decisions for the

period from 1971 to 2010. The structure of Table 5.2 is the same as that in Table 4.1, with the

rows of the table differing in (1) the various measures of market-timing variables, and (2) the

inclusion of the FCF proxy. Rows A through F indicate that M/B ratios and prior abnormal stock

returns are significantly and negatively related to the probability that a firm conducts a

repurchases, with the exception that the coefficient on prior stock returns is insignificant in Row

C. Unlike in Table 4.1, the coefficients on future abnormal stock returns are positive and

significantly different from zero.

Overall, the signs of the coefficients on market-timing variables in Table 5.3 are consistent

with the predictions of the market-timing theory. These coefficients indicate that firms with more

favorable market-timing opportunities (lower M/B ratio and recent stock return, and higher

future stock return) are more likely to conduct a share repurchase. While these data offer

qualitative support for market timing, they do not speak to the issue of economic significance. I

evaluate the latter issue in detail below.

41
 When I use standardized operating cash flow (OCF) to proxy for FCF as an independent

variable in the logit regressions, the level of FCF is significantly (t-statistic = 3.28, 3.40, and

3.95) and positively related to the estimated repurchase probability, consistent with the FCF

hypothesis.

Table 5.3 reports the marginal probabilities of a share repurchase to provide a quantitative

feel for how a small change in each explanatory variable translates to changes in the repurchase

probability. I find that the marginal probability of a share repurchase evaluated at the median

value of FCF is much larger than the marginal probability evaluated at the median value of M/B

ratio. For instance, when standardized M/B ratio and 12-month abnormal return are used as

explanatory variables, the former figure is 0.38, more than 12 times the size (in absolute value)

of the latter figure of −0.03.

To assess the relative importance of the market-timing motive versus the FCF motive in

explaining managerial repurchase decisions, Table 5.4 reports the estimated repurchase

probabilities conditional on specific hypothesized values of the timing and FCF variables. The

probabilities in the far right column are estimated using the model in Row A of Table 5.3, which

does not account for the FCF effect. All other probabilities are estimated using the model in Row

D of Table 5.3.

The estimated repurchase probabilities in the last column of Panel A of Table 5.4 indicate

that (1) variation in abnormal stock returns has almost no effect on the estimated repurchase

probability, and (2) variation in the M/B ratio shows only a modest influence on the estimated

repurchase probability. A shift in future (recent) abnormal stock return from the 5th to the 95th

percentile values implies a small change in the estimated repurchase probability (the absolute

changes in the estimated probability are equal to or less than 2.0%, Rows 2 through 5 in Panel A

42
of Table 5.4). In contrast, when the standardized M/B ratio decreases from its 95th to its 5th

percentile value, the estimated repurchase probability increases by 6.3% (i.e., it equals 18.0%

instead of 11.7%, Rows 6 and 7). Holding M/B ratio and past abnormal stock return constant at

very high or very low levels, a huge swing in future abnormal stock returns (from the 5th to the

95th percentile) induces an absolute change in the repurchase probability that is less than 1.8%

(Rows 8 through 13).

The explanatory power of market timing can be best assessed from the last two rows in

Panel A of Table 5.4. Among firms facing excellent timing opportunities (standardized M/B ratio

and prior abnormal stock return at their 95th percentiles, and future abnormal stock return at its

5th percentile), 20.0% buy back stock. This is roughly twice as large as the probability that a firm

conducts a repurchase when it faces extremely unfavorable timing opportunities (standardized

M/B ratio and prior abnormal return equal their 5th percentiles, and future abnormal return equals

its 95th percentile). The absolute increase in the repurchase probability is 9.5%, which is

nontrivial and thus lends credence to the importance of market timing.

However, as before, when I estimate the repurchase probability as a function of both market-

timing and FCF variables, FCF shows a greater effect on the estimated repurchase probability

than the proxies for market-timing opportunities. The strength of the FCF effect on the estimated

repurchase probability can be seen in the middle columns in Panel A of Table 5.4. For example,

Row 1 shows that for firms with neutral market-timing opportunities, the estimated repurchase

probability increases from 6.1% to 24.8% when the level of FCF changes from very low (5th

percentile) to very high (the 95th percentile). The latter figure is more than four times the size of

the former figure. Meanwhile, the absolute change is also economically material (18.7% = 24.8%

− 6.1%). This absolute change of 18.7% is nearly twice the size of the 9.5% absolute change

43
induced by a large swing in market-timing opportunities. Similar results hold for all other

market-timing scenarios in Panel A of Table 5.4. This implies that variation in FCF induces a

large change in the estimated repurchase probability regardless of whether market-timing

opportunities are favorable or unfavorable.

The relative importance of market-timing opportunities and FCF in explaining repurchase

decisions can best be seen by comparing Rows 14 and 15 in Panel A of Table 5.4. This

comparison shows that a firm with high FCF and highly unfavorable timing opportunities has a

repurchase probability of 12.3% (Row 15), which is roughly one-and-a-half times the 8.0%

probability that a firm with low FCF and highly attractive timing opportunities conducts a share

repurchase (Row 14). As with prior comparisons, the FCF effect quantitatively dominates the

market-timing effect. The absolute differential (4.3%) is not as large as in some other

comparisons reported in this study, but it still points to a nontrivial advantage for FCF over

market-timing considerations.

As in Section 4.1, I also repeat the comparison between the FCF and market-timing effects

with M/B ratio excluded in the logit regressions. As shown in Panel B of Table 5.4, my results

do not change qualitatively. Abnormal stock returns show little influence on the estimated

repurchase probability while the FCF has a markedly stronger effect. A firm with extremely poor

timing opportunities and high FCF is more than twice likely to conduct a repurchase as a firm

with extremely good timing opportunities and low FCF (17.3% versus 8.6%, Rows 12 and 13).

The absolute differential (8.7%) is larger than that in the logit analysis with M/B ratio included.

When I adjust operating cash flow by the industry median capital expenditure, similar

results are obtained. The level of FCF continues to be significantly and positively correlated with

the repurchase profanity, and its effect quantitatively dominates the market-timing effect in

44
explaining managerial repurchase decisions (Tables 5.5 and 5.6). The estimated repurchase

probability for a firm with high FCF and highly unattractive timing opportunities is 12.2%, with

an excess of 64.8% over the 7.4% probability for a firm with low FCF and highly attractive

timing opportunities.

In sum, the results in Tables 5.3 through 5.6 are qualitatively comparable to those in

Sections 4.1 and 4.2. This provides further support for my main inference that the desire to

distribute FCF has a considerably stronger effect than market timing in explaining managers’

share-repurchase decisions.

5.3 Logit Analysis Comparing the Market-Timing and Employee-Stock-Option Effects

Recall that the logit models in Chapter 4 include the employ-stock-option (ESO) proxy as a

control variable. In this section, I use the estimated coefficients from Chapter 4 to compare the

market-timing and ESO effects on the decision to repurchase stock. Table 4.1 indicates that the

repurchase probability is significantly (t-statistic > 3.23) and positively correlated with the

standardized change in the number of ESOs outstanding. This finding is qualitatively consistent

with the predictions of the ESO hypothesis.

To assess the quantitative importance of the ESO effect on repurchase decisions and to

compare that with the magnitude of the market-timing effect, I estimate repurchase probabilities

as a function of the change in the number of ESOs outstanding (scaled by the number of

common shares outstanding) and variation in market-timing variables. The approach here is

analogous to that in Section 4.2. I use the coefficient estimates of Model D in Table 4.1 to

compute the repurchase probability conditional on specific hypothesized values of the timing and

ESO variables, while holding the other variables constant at their sample median values.

45
 The estimated repurchase probabilities in the middle columns of Table 5.7 indicate that: (1)

ESO considerations are empirically important in explaining managerial repurchase decisions,

and (2) the ESO effect is roughly comparable to the market-timing effect in magnitude.

For example, for a firm with neutral market-timing opportunities, the estimated repurchase

probability increases from 25.0% to 34.0% when the standardized change in the number of ESOs

outstanding increases from its 5th percentile value to its 95th percentile value (Row 1 in Panel A

of Table 5.7). The repurchase probability is 22.3% for a firm with poor timing opportunities and

a large change in number of ESOs, which is 5% in absolute terms or 18.3% in relative terms

below the 27.3% probability for a firm with excellent timing opportunities and a small change in

number of ESOs (Rows 14 and 15 in Panel A). The comparable numbers for the logit analysis

with M/B ratio excluded are 25.6% and 26.8% (Rows 12 and 13 in Panel B), with the two figures

quite close to each other.

The bottom line is that, although the market-timing effect is slightly greater than the ESO

effect in the logit analyses with M/B ratio included as a proxy for market-timing opportunities,

their impacts are comparable in magnitude in other logit analyses that exclude the M/B ratio.

Neither is as strong as FCF in explaining managerial repurchase decisions.

5.4 Logit Analysis Comparing the Market-Timing and Leverage-Rebalancing Effects

The logit models in Chapter 4 also include the leverage-rebalancing proxy as a control

variable. In this section, I use the estimated coefficients from Chapter 4 to compare the market-

timing and leverage-rebalancing effects on the decision to repurchase shares. The coefficient

estimates in Table 4.1 indicate that the probability of a share repurchase is significantly (t-

statistic < −5.56) and negatively correlated with the deviation from a firm’s estimated target

leverage ratio. Intuitively, this finding indicates that firms with lower debt-to-assets ratios below

46
their estimated target leverage ratios are more likely to buy back stock, consistent with the

leverage-rebalancing theory.

I follow the same approach as in Section 5.3 to assess the magnitude of the leverage-

rebalancing effect on repurchase decisions and to compare that with the magnitude of the

market-timing effect. In particular, I compute the repurchase probability as a function of the

deviation from the estimated leverage target ratios and the market-timing variables, using the

fitted coefficients from Model D in Table 4.1. Table 5.8 reports repurchase probabilities based

on specific hypothesized values of the market-timing variables and the deviation from the target

leverage ratios, while holding the other variables constant at their sample median values.

The middle columns of Table 5.8 show that (1) leverage-rebalancing considerations are

empirically important in explaining managerial repurchase decisions, and (2) the leverage-

rebalancing effect is stronger than the market-timing effect in explaining managerial repurchase

decisions.

For a firm with neutral market-timing opportunities, the estimated repurchase probability

decreases from 34.7% to 19.3% when its deviation from the estimated target leverage ratio

swings from the 5th percentile value to the 95th percentile value. The repurchase probability is

22.8% for a firm with poor timing opportunities and the deviation from target leverage at the 5th

percentile. This is almost the same as the 21.2% probability for a firm with good timing

opportunities and the deviation from target leverage at the 95th percentile (Rows 14 and 15 in

Panel A). The comparable figures for the logit analysis with M/B ratio excluded are 26.3% and

20.7%, with the former figure exceeds the latter figure by 5.6% in absolute terms and by 27.1%

in relative terms (Rows 12 and 13 in Panel B).

47
 When I use industry median leverage ratios (2-digit SIC or Fama-French 49 industries) as

proxies for firm-specific target ratios, the logit results in this section do not change qualitatively.

For example, when industries are defined as Fama-French 49 industries, the coefficient on the

deviation from the target leverage ratio is statistically significantly negative (t-statistic < −5.42).

For the logit analysis with M/B ratio excluded, the repurchase probability is 27.0% for a firm

with unfavorable timing opportunities and the deviation from target leverage at the 5th percentile

value. This is 5.7% in absolute terms and 26.8% in relative terms above the 21.3% repurchase

probability for a firm with excellent market-timing opportunities and the deviation from target

leverage at its 95th percentile value.

Overall then, my evidence offers some support for the view that firms use stock repurchases

to rebalance leverage toward a target ratio. The leverage-rebalancing effect is quantitatively

comparable to the market-timing effect in the logit analyses that use M/B ratio as a proxy for

market-timing opportunities, and is stronger than the timing effect in the logit analyses with M/B

ratio excluded. However, neither leverage rebalancing nor market timing is as strong as FCF in

explaining managers’ share-repurchase decisions.

5.5 Logit Analysis of the Decision to Conduct a Tender-Offer Repurchase

As discussed in Chapter 3, there is a reason to think that the market-timing effect might be

more prominent for tender-offer repurchases simply because they are much larger than most

repurchases. In this section, I conduct similar logit analyses to gauge and compare the

importance of the market-timing and free-cash-flow (FCF) considerations in explaining

managerial tender-offer repurchase decisions. I also gauge the magnitude of employee-stock-

option and leverage-rebalancing motives for tender-offer repurchases.

48
5.5.1 Tender-Offer Repurchases, Market Timing, and Free Cash Flow

I re-run the logit regressions of Chapter 4 to assess managers’ tender-offer repurchase

decisions as a function of a firm’s standardized market-to-book ratio (M/B), recent and future

abnormal stock returns, and its level of FCF. The standardized change in the number of

employee stock options (ESOs) outstanding, and the deviation from the target leverage ratio are

included as control variables. To control for firm size, all logit regressions also include the

natural log of total assets at the end of the fiscal year prior to the year in question.

Table 5.9 indicates that, when only proxies for market-timing opportunities are included in

the logit model, the signs of coefficients on timing variables are consistent with the market-

timing hypothesis. The coefficients on standardized M/B ratio and recent stock return are

negative, while the coefficient on future stock return is positive. However, only the coefficients

on M/B ratio are statistically significant (t-statistic = −4.71, −4.52, and −4.20 in Rows A through

C). As I discussed in Section 4.1, it is difficult to draw an unambiguous interpretation of the

coefficients on M/B ratio. Therefore, I conclude that the results in Rows A through C in Table

5.9 provide only weak support for the market-timing hypothesis.

When I include the proxies for FCF, ESO, and leverage-rebalancing considerations in the

logit regressions, the coefficients on future stock return become negative in sign and statistically

insignificant (Rows E through F), while the findings regarding M/B ratios and prior abnormal

returns are qualitatively unchanged. These results further undermine the market-timing theory

because, according to that theory, the probability of a tender-offer repurchase should be

positively related to future stock returns.

Importantly, Rows D through F of table 5.9 show that the level of FCF (proxied by

standardized operating cash flow) is positively and significantly correlated with the estimated

49
probability of a tender-offer repurchase (t-statistic = 2.66, 2.28, and 4.00). The positive

coefficients on FCF indicate that firms with higher FCF are more likely to conduct repurchase

tender offers, consistent with the FCF motive for repurchasing shares.

To gauge and compare the economic significance of the market-timing and FCF motives for

tender-offer repurchases, I estimate the probability of a tender-offer repurchase as a function of

the proxies for market-timing and FCF considerations. Table 5.10 reports the estimated

probability of a tender-offer repurchase conditional on specific hypothesized values of the timing

and FCF variables, while holding the other variables constant at their sample median values.

Except for the estimated probabilities in the far right column, the probabilities in Panel A of

Table 5.10 are calculated using the model in Row D of Table 5.9. The estimated probabilities in

the far right column are based on the model in Row A of Table 5.9, which only accounts for the

market-timing effect.

The results in Table 5.10 indicate that: (i) variation in abnormal stock returns has almost no

effect on the estimated tender-offer probability and (ii) both variations in M/B and FCF show a

modest influence on the probability of a tender offer.

In particular, the far right column in Panel A of Table 5.10 shows that, a shift in either past

or future abnormal stock returns (from the 5th to the 95th percentiles) implies a tiny absolute

change (less than 0.1%) in the estimated tender-offer probability. A swing in M/B ratio from its

5th to its 95th percentile values induces an increase in the tender-offer probability from 0.1% to

0.8%. The middle columns in Panel A indicate that for firms with neutral market-timing

opportunities, the estimated tender-offer probability increases from 0.2% to 1.2% when the level

of FCF increases from very low (5th percentile) to very high (95th percentile). Although the

relative change of 600% is large, the absolute differential (1.0%) is small.

50
 Table 5.10 reports mixed evidence regarding the relative importance of market timing versus

FCF in explaining tender-offer repurchases. When M/B ratio is included in the logit model, the

market-timing effect is stronger than the FCF effect. On the other hand, when M/B ratio is

excluded in the logit model, the previous result reverses.

Specifically, based on the logit model with M/B ratio included, Rows 14 and 15 in Panel A

of Table 5.10 show that for a firm with high FCF and poor timing opportunities, the tender-offer

repurchase probability is 0.1%, which is close to the 0.2% tender-offer probability for a firm with

low FCF and good timing opportunities. The comparable figures in the logit analysis with M/B

ratio excluded are 0.5% and 0.4%, with two figures close to each other in absolute terms (Rows

13 and 14 in Panel B of Table 5.10).

It is important to note that although the relative change in above estimates is nontrivial, the

absolute differential is quantitatively small. There are at least two possible explanations. First,

firms conduct tender-offer repurchases highly infrequently. A randomly selected firm in my

sample has a tender-offer probability in a given year of 0.55%, and hence the magnitude of the

estimated probability of a tender offer should not be too large.

Second, although FCF has a larger influence on the tender-offer repurchase decision than do

market-timing opportunities, it is not a complete theory that can fully explain all tender-offer

repurchases. This is because too many firms with high FCF fail to conduct a tender offer. An

alternative explanation is that firms with a high level of FCF may choose other methods to

distribute FCF, such as open-market repurchases or cash dividends. This is confirmed in the logit

analyses on general repurchase and dividend decisions. Panel B of Tables 4.2 and 5.17 indicates

that for firms with high FCF, 33% buy back shares and 30% pay cash dividends in the worst

hypothesized market-timing scenario.

51
 To provide a robustness check, I run similar logit regressions using standardized operating

cash flow adjusted by the industry median capital expenditure ratio. Almost identical results are

obtained as shown in Tables 5.11 and 5.12. FCF and M/B variables continue to show modest

effect (in relative terms) on the estimated tender-offer probability and the evidence on the

relative impact of the market-timing and FCF effects is still mixed, depending on whether or not

M/B ratio is used to proxy for market-timing opportunities. However, both the market-timing

and FCF effects are small in magnitude.

The bottom line is that, my data indicate that neither market timing nor FCF is of first-order

importance in explaining managers’ tender-offer repurchase decisions. And my logit analyses

show no convincing evidence that any one of them is systematically more important than the

other one.

5.5.2 Tender-Offer Repurchases, Market Timing, and Employee Stock Options

Table 5.9 indicates that the standardized change in the number of ESOs outstanding is

significantly positively correlated with the estimated tender-offer probability (t-statistic > 6.39 in

Rows D through F). These findings indicate that an increase in ESO outstanding is associated

with a higher probability of a tender offer.

Following the approach in Section 5.3, I estimate tender-offer repurchase probabilities as a

function of the standardized change in the number of ESOs outstanding and the market-timing

variables. Using the coefficient estimates from Model D in Table 5.10, I compute the tender-

offer probability conditional on specific hypothesized values of the timing and ESO variables,

while holding the other variables constant at their sample median values.

The estimated tender-offer probabilities in the middle columns of Table 5.13 indicate that:

(1) ESO considerations explain some managers’ tender-offer decisions, and (2) the ESO effect

52
on tender-offer repurchase decisions is slightly weaker than the market-timing effect when M/B

ratio is used to proxy for timing opportunities, and is slightly stronger than the market-timing

effect when M/B ratio is excluded in the logit models.

Specifically, when firms face neutral timing opportunities, a shift in the standardized change

in the number of ESOs from the 5th to the 95th percentile values implies an increase in the tender-

offer probability from 0.5% to 1.2% (Row 1 in Panel A of Table 5.13). When M/B ratio is

included in the logit model, a firm with poor timing opportunities and a large standardized

change in the number of ESOs has a tender-offer probability of 0.1%, which is close to the 0.6%

probability that a firm with excellent timing opportunities and a small standardized change in the

number of ESOs conducts a tender offer. The comparable numbers when M/B ratio is excluded

in the logit model are also close to each other – 0.7% and 0.5% (Rows 13 and 14 in Panel B of

Table 5.13).

Overall, the results in Tables 5.9 and 5.13 are qualitatively consistent with the ESO

explanation for tender-offer repurchases. Firms that experience an increase in the standardized

number of ESOs outstanding are more likely to conduct a tender-offer repurchase. However, the

magnitude of the ESO effect is modest. So too is the market-timing effect. There is no sign in

these data that the market-timing effect is systematically more important than the ESO effect in

explaining tender offers. The most reasonable interpretation is that neither the ESO effect nor the

timing effect is all that important.

5.5.3 Tender-Offer Repurchases, Market Timing, and Leverage Rebalancing

Table 5.9 also reports that, in the logits that seek to explain the decision to conduct a

repurchase tender offer, the coefficients on the deviation from a firm’s estimated target leverage

ratio are statistically significantly negative (t-statistic < −2.51). This finding indicates that firms

53
with lower debt-to-assets ratios below their estimated target leverage ratios are more likely to

conduct a tender offer, consistent with the leverage-rebalancing theory.

I follow the same approach as in Section 5.4 to assess the economic magnitude of the

leverage-rebalancing effect on tender-offer decisions and to compare that with the magnitude of

the market-timing effect. In particular, I use the estimated coefficients from Model D in Table

5.9 to compute the probability of a tender offer as a function of the deviation from the estimated

target leverage ratio and market-timing variables. Table 5.14 reports repurchase probabilities

based on specific hypothesized values of the timing variables and the deviation from the target

leverage ratio, while holding the other variables constant at their sample median values.

The middle columns of Table 5.14 show that (1) leverage-rebalancing considerations show

some modest degree of impact on managerial tender-offer decisions, and (2) the leverage-

rebalancing effect is slightly weaker than the market-timing effect in explaining tender-offer

decisions in logit analyses with M/B ratio included, and is slightly stronger than the market-

timing effect in logit analyses with M/B ratio excluded.

Specifically, for a firm facing neutral market-timing opportunities, the estimated tender-

offer probability increases from 0.5% to 1.0% when its deviation from the estimated target

leverage ratio drops from the 95th percentile value to the 5th percentile value. The estimated

tender-offer probability is 0.1% for a firm with poor timing opportunities and the deviation from

target leverage at the 5th percentile, which is close to the 0.6% probability for a firm with

excellent timing opportunities and the deviation from target leverage at the 95th percentile (Rows

14 and 15 in Panel A). The comparable figures for the logit analysis with M/B ratio excluded are

0.6% and 0.4%, close to each other again (Rows 12 and 13 in Panel B).

54
 To further investigate the effect of the deviation from target leverage on repurchase tender-

offer decisions, I examine the mean values of the deviation from the estimated target leverage

ratio and leverage determinants surrounding 539 tender-offer repurchases that are announced

between 1985 and 2010. Table 5.15 reports the mean values of the deviations from target

leverage in event years from -3 to 3, where event year 0 refers to the year in which a firm

conducts a tender-offer repurchase. The mean deviation from target leverage is −0.04 in event

year -1, and changes to 0.01 in event year 1 (Row 2 in Table 5.15). However, Rows 3 and 4

show that, when the industry median debt-to-assets ratio (2-digit SIC or Fama-French 49

industries) is used as a proxy for a firm’s target leverage ratio, the mean deviation from target is

positive in all event years. This indicates that firms on average are not under-leveraged prior to

tender-offer repurchases if we use industry median leverage as a proxy for the target leverage

ratio. And thus the finding on rebalancing explanation is not robust to the use of different

measures of target leverage ratio.

In addition, Table 5.15 suggests that the increase in debt-to-asset ratios after tender-offer

repurchases is due mainly to the change in total debt. The average debt-to-assets ratio increases

by 4.2% from event year −1 to event year 0, and this increase is statistically significant at

conventional levels (t-statistic = 3.51). During the same period, total debt (standardized by the

lagged total assets) increases by a statistically significant 5.0% (t-statistic = 3.98). Meanwhile,

total assets do not change materially, suggesting that the increase in leverage ratio is mostly

driven by the change in total debt.

One possible explanation for this phenomenon is that firms often use debt to finance tender-

offer repurchases. This makes sense because tender-offer repurchases are generally large in

dollar size, and so firms would have to have large cash balances to fund the buyback without

55
raising external capital. Moreover, total debt continues to increase in event years 1, 2 and 3 and

these increases drive the realized leverage ratios even higher above the target ratios. I also find

that, on average, target leverage ratios and traditional leverage determinants – asset growth,

capital expenditure, profitability, sales, M/B, and tangible assets – are not significantly different

one year before and one year after tender-offer repurchases.

These findings together suggest that firms move their leverage ratios toward and beyond the

estimated target leverage ratios after tender-offer repurchases. There is no sign that the increases

in leverage reflect a general increase in target leverage ratios, at least not when those ratios are

estimated using determinants commonly considered in the literature. Movements toward a target

ratio are, of course, consistent with the leverage-rebalancing theory, but movements beyond the

target are not. The latter point indicates that either there is something missing from the theory or

my proxies for leverage targets tend to be too low.

In addition, the average total dollar volumes of repurchases are higher in event years 1, 2,

and 3 than in event years −1, −2 and −3. Since firms usually do not conduct multiple tender

offers in a short period of time, this pattern indicates that firms that conduct a tender-offer

repurchase on average continue to buy back stock through other methods during the three-year

period following the tender offer. The above results are robust to using a “survivor sample,”

which is defined as a sample made up of firms that have all data available in all event years.

In sum, the results in this section show that more under-leveraged firms are more likely to

conduct a repurchase tender offer. However, when I use industry median leverage ratio as the

target leverage ratio, firms on average are not systematically under-leveraged before repurchase

tender offers. This is inconsistent with the leverage-rebalancing explanation for tender-offer

repurchases. Moreover, the magnitudes of both the market-timing and leverage-rebalancing

56
effects are small. Perhaps more importantly, I find no evidence that the market-timing effect has

a systematically stronger effect on the decision to conduct a repurchase tender offer than the

leverage-rebalancing effect has.

5.6 Logit Analysis of the Decision to Pay Dividends

The logit analyses in previous sections indicate that market timing has a statistically

significant effect on managers’ repurchase decisions. Among three proxies for market-timing

opportunities — M/B ratio and prior and future abnormal stock returns — M/B ratio exerts the

most significant impact. However, it is difficult to draw an unambiguous inference from the

impact of M/B ratio. The reason is that M/B ratio also measures investment opportunities.

Consequently, a firm with a low M/B ratio is more likely to repurchase shares either because it

has better timing opportunities, or because it has a stronger incentive to distribute free cash flow

(FCF) due to its lack of growth opportunities, or perhaps because both timing and FCF motives

are operative.

To better assess the impact of M/B ratio on repurchase decisions, I run similar logit analyses

as those in Chapter 4 on firms’ dividend decisions. If M/B ratio is also negatively correlated with

the probability that a firm pays dividends, the M/B impact on dividend decisions is likely to be

linked to investment opportunities rather than market-timing opportunities. This is because firms

with a low M/B ratio have limited investment opportunities and thus have strong incentive to pay

out cash (the FCF hypothesis for paying dividends). On the other hand, there is no obvious

reason why undervalued firms (as proxied by low M/B ratios) should return cash to shareholders.

The logit analyses use the same sample as in Chapter 4 to assess managers’ dividend

decisions. The dependent variable equals one in a given year if the firm pays cash dividends in

that year, and equals zero if the firm does not. Similar to the analysis in Section 4.1, the

57
independent variables include the firm’s most recent standardized M/B ratio, its most recent and

future 12-month (or 36-month in some tests) abnormal stock returns, and its level of FCF

(proxied by standardized operating cash flow). To control for firm size, all logit regressions also

include the natural log of total assets at the end of the fiscal year prior to the year in question.

In Table 5.16, coefficients on M/B ratio are negative and statistically significant regardless

of whether the proxy for FCF is included in the logit models (Rows A through F). This is

consistent with the FCF hypothesis for dividends. Firms with low M/B ratios are low-growth

firms with poor investment opportunities that have strong incentive to pay out cash through

dividends. This inference is further reinforced by the fact that the level of FCF is significantly

positively correlated with firms’ dividend decisions (Rows D through F). Intuitively, firms with

high FCF are more likely to distribute cash to shareholders through common dividends.

I also follow the approach in Section 4.2 to assess the economic significance of M/B and

FCF on managerial dividend decisions. Using the coefficient estimates from Table 5.16, I

compute the probability that a firm pays dividends as a function of M/B ratio, prior and future

stock returns, and the level of standardized FCF. Table 5.17 reports these estimated dividend

probabilities conditional on specific hypothesized values of the M/B ratio, the prior and future

stock returns, and the level of FCF.

I find that FCF has an economically important effect in explaining managers’ dividend

decisions, regardless of the values of M/B ratio and abnormal stock returns. For example, for

firms with M/B ratio and stock returns at their sample median values, a shift in FCF from its 5th

to its 95th percentiles markedly increases the dividend probability from 0.5% to 48.2% (Row 1 in

Panel A of Table 5.17). The latter figure is roughly 100 times the size of the former figure.

Clearly, FCF variation has a huge impact on dividend decisions.

58
 Overall, the findings in Tables 5.16 and 5.17 are consistent with the notion that FCF is the

primary driver of firms’ dividend decisions, and that the M/B ratio is negatively correlated with

the probability that a firm pays dividends because it measures growth opportunities. Although I

cannot directly generalize this inference to share repurchases, my evidence here is more

consistent with the notion that a firm with a low M/B ratio is more likely to repurchase stock

because it has poor growth opportunities rather than its share price is undervalued. This evidence

further supports the main results in Chapter 4 that FCF is more important than market timing in

explaining share repurchase decisions.

5.7 Repurchase Frequency

This section studies how often firms conduct a share repurchase and provides descriptive

statistics on the frequency of repurchases. If firms buy back stock frequently, it is difficult to

accept the idea that these frequent share repurchases are mainly driven by market-timing

opportunities. If managers buy back stock mainly because they want to time the market and

exploit market undervaluation, investors will eventually learn this ploy if managers do that with

great regularity. Consequently, investors would be able to implement a trading strategy that can

eliminate timing opportunities and make share repurchases unattractive in a market-timing sense.

As a result, share repurchases driven by the desire to exploit market undervaluation should not be

a frequently recurring phenomenon for a given firm.

However, the data in Table 5.18 show that frequent share repurchases are pervasive for my

sample of 13,186 industrial firms that are in the CRSP/Compustat merged files over 1983 to

2010. I focus on the period beginning in 1983 because share repurchases became popular after

the adoption of SEC rule 10b-18 in November 1982.

59
 I find that firms conduct share repurchases much more frequently in the last 15 years or so.

For example, Panel B of Table 5.18 shows that for 2,649 firms listed at least 10 years during the

period from 1983 to 1996, 23.1% repurchase stock at least once every two years, and 49.7% buy

back shares at least one out of every four years. The comparable figures for 2,627 firms listed at

least 10 years between 1997 and 2010 are 36.0% and 62.4% (Panel C of Table 5.18). Such

consistent and frequent share repurchases are difficult to reconcile with the notion that market

timing is the primary motive for share repurchases. The reason is that it seems unlikely that a

large portion of public firms are undervalued for extended periods of time.

I further investigate the relationship between the repurchase frequency and firm age (proxied

by the number of year listed in the sample). For each firm, I define the repurchasing year as the

year in which a firm buys back its own stock. I then count the total number of repurchasing years

for each firm, and divide the result by this firm’s number of years listed in the sample to compute

the percent of years with repurchase activity, which of course measures how often a firm buys

back shares.

Table 5.19 indicates that a firm’s repurchase frequency is positively related to its age. This

result holds for the full sample of 14,958 industrial firms in the CRSP/Compustat merged files

over 1971 to 2010. I use the full sample period because this approach is more likely to capture

firm age accurately. I find that the median (average) firm repurchases stock approximately once

every 10 (five) years. However, for the 2,060 firms that are listed at least 20 years, both the

median and the average firm buy back stock roughly once every three years. And for the

subsample of 240 firms that are listed every year between 1971 and 2010, the average firm

repurchases stock two out of every five years.

60
 The results in Table 5.19, and the other evidence in this dissertation, collectively support the

view that mature firms (as measured by firm age) are more likely to buy back stock. This view is

based on the FCF explanation for share repurchases. Mature firms typically have high FCF and

poor growth opportunities. As a result, they have stronger incentives to pay out FCF through

repurchases. On the other hand, mature firms are commonly large firms that have less

asymmetric information and are thus less likely to be undervalued. The findings in Table 5.19

are difficult to explain on the basis of market timing, as there is no reason to expect an inordinate

number of mature firms to be undervalued for extended periods of time.

61
 Chapter 6

Abnormal Stock Returns after Repurchases

This chapter examines the post-repurchase abnormal stock returns because prior studies

argue that there is a tendency toward positive abnormal stock returns following share

repurchases, which they interpret as providing strong evidence supporting the market-timing

theory. My specific objective in this chapter is to gauge how often firms’ repurchase decisions

turn out to be good or bad in a market-timing sense.

6.1 Median and Mean Post-repurchase Abnormal Stock Returns

The market-timing hypothesis argues that managers choose to buy back stock because their

shares are underpriced. If this is true, we should expect to see that more firms experience positive

than negative abnormal stock returns after share repurchases. Although prior studies have found

that the average abnormal stock return following repurchases is significantly positive, this does

not imply that the majority of repurchases are followed by positive abnormal stock returns.

In fact, the histogram of future abnormal stock returns in Table 6.1 indicates that more firms

experience negative than positive post-repurchase abnormal stock returns – a finding that is

inconsistent with the market-timing hypothesis. For all repurchases, 54.0% are followed by

negative abnormal stock returns over the 12-month period following repurchases. Moreover, the

median post-repurchase abnormal return is negative in the full sample and in all of the various

subsamples that I examine. For the full sample of repurchases, the median abnormal stock return

is −3% over the 12 months after repurchases. Similar results hold for one-shot repurchasers

(firms that only repurchase once during their time in the sample) and tender-offer repurchases.

The results are also robust to using 36-month abnormal stock returns. These findings in Table 6.1

62
cast doubt on the market-timing hypothesis as they show that most managers either do not

systematically time the market or that their ability to time the market is poor.

As with prior studies, Table 6.1 indicates that the average post-repurchase abnormal stock

return is positive in the full sample and in all other subsamples that I study. For the full sample of

repurchases, the mean post-repurchase abnormal return is 7% over the 12 months after

repurchases. The findings of positive mean abnormal stock returns are consistent with market

inefficiency as competition among investors to trade on the public information that a firm

repurchases stock should bid away the possibility of systematically positive future abnormal

returns. Prior studies claim this finding of positive average future abnormal returns as strong

evidence supporting the market-timing hypothesis. This inference is problematic for three

reasons.

First, the efficient market hypothesis (EMH) and the market-timing hypothesis deal with

different subjects. The focus of the EMH is the experience of investors. If all shares are priced

fairly based on public information as implied by the EMH, investors should not be able to earn

positive abnormal returns on average. Since the focus is on investors’ typical return experience,

the average abnormal returns should be examined.

On the other hand, the focus of the market-timing hypothesis is the behavior of managers, as

it suggests that managers undertake repurchases to take advantages of market undervaluation.

Thus we should focus on the typical managers’ behavior and examine whether most post-

repurchase abnormal stock returns are positive. The relevant research question is “what is most

likely to happen to the share prices in the future if managers conduct a repurchase today.”

63
 Second, market inefficiency is an underlying assumption of the market-timing hypothesis.3

The market-timing explanation suggests that managers exploit market undervaluation through

share repurchases. The possibility of exploitable market undervaluation is thus an assumption of

the timing theory. Because all shares should be priced fairly based on all public information in an

efficient market and because share repurchases are publicly disclosed, mispricing in the current

context is equivalent to stock market inefficiency. A finding that the stock market is inefficient

only validates the assumption of the market-timing hypothesis. It does not show that managers

systematically exploit mispricing opportunities through buying back stock.

Last but not least, the average post-repurchase abnormal returns are less informative than

median returns for the purpose of testing the market-timing hypothesis. This is because the focus

here is to investigate what is more likely to happen to the share prices after repurchases. As

documented later in this chapter, the average post-repurchase abnormal returns are distorted by

extreme values, and so they cannot tell us whether abnormal stock price declines or increases are

more likely to be observed. The median does, of course, provide direct information on this issue.

For example, if 70% of managers do not buy back shares based on timing opportunities and one

manager makes an extremely excellent timing decision to repurchase stock, the median post-

repurchase abnormal return will be zero but the average will tend to be positive. However, this

positive mean value does not indicate that managers systematically make repurchase decisions

that exploit favorable timing opportunities.

The stock returns reported in Panels A and B in Table 6.1 are percentage returns (or rates of

return). One drawback of a percentage stock return is that it doesn’t tell us the economic

3
Here I am referring to market timing theories in which mispricing of shares arises and is exploited systematically
by managers. The asymmetric information version of market timing in Baker and Wurgler (2002) does not assume
the market inefficiency. Their results do not discriminate the mispricing and asymmetric information versions of
market timing, but they do claim that the long-run return evidence points to the mispricing version of market timing.
(see Baker and Wurgler, 2002, p. 4)

64
significance (or the wealth effect) of share repurchases. Specifically, according to the market-

timing hypothesis, managers buy back shares when they believe doing so is a “good investment.”

The percentage return only tells us the rate of return, not the absolute wealth gains from this

“investment.” For example, if a firm buys back a very small amount of shares (e.g., close to zero)

and experiences a positive abnormal rate of return after that repurchase, the resultant wealth

effect is trivial.

To investigate the economic significance (the wealth effect) of repurchases, I also examine

the abnormal future dollar stock returns, defined as the product of the dollar amount of shares

repurchased and the future abnormal rate of return on the stock. Table 6.2 shows that the

distributions of future abnormal dollar returns are similar to those of future abnormal stock rate

of return. For the full sample, the mean 12-month future abnormal dollar return is $2.1 million

and the median is slightly negative and close to zero. This suggests that the wealth change for the

median repurchase is nearly zero. For one-shot repurchases, both mean and median abnormal

dollar returns are negative, suggesting firms on average experience wealth loss from these

repurchases. For tender-offer repurchases, the average 12-month future abnormal dollar return is

$3.2 million while the median is zero. These results are also robust to using 36-month future

abnormal dollar returns.

Overall, the abnormal dollar stock returns support my main inference that either managers

do not systematically time the market or their timing decisions are poor. The reason, in a nutshell,

is that most repurchases are not followed by abnormal wealth gains.

6.2 Distribution Analyses of Post-repurchase Abnormal Stock Returns

Table 6.1 also documents that, while there are both extremely positive and extremely

negative abnormal returns, the right tail of the distribution is longer than the left tail. For the full

65
sample of repurchases, I find that 7.5% of repurchases are followed by a 12-month abnormal rate

of return above 75%, while 1.8% of repurchases are followed by a 12-month abnormal rate of

return below −75%. The result is that the distribution is significantly skewed to the right. Thus

the mean value is significantly greater than the median value and does not represent the stock

price behavior of the majority of sample firms. Similar results are obtained when I examine one-

shot repurchases, tender-offer repurchases and for using future 36-month abnormal stock returns.

The evidence above suggests that the mean post-repurchase abnormal stock returns may be

distorted by extreme values. Since many studies draw inferences from the mean post-repurchase

abnormal returns, it is important to understand whether or to what extent the mean values are

distorted by extreme positive values.

To address this issue, Panels A and B of Table 6.3 document the 20 biggest winners and 20

biggest losers among all share repurchases based on future 12-month abnormal stock rates of

return. Some firms do exhibit remarkably large positive abnormal rates of return after

repurchases. The biggest winner on a rate-of-return basis is Mehl/Biophile, which spent a

modest $230,000 on share repurchases in 1994 and experienced an abnormal rate of return of

3,890% over the year following the repurchase. In addition, 12 other companies had a 12-month

future abnormal rate of return above 1,000%.

Strikingly, almost all of these 20 winners repurchased in very small dollar amount. More

than half of them repurchased less than $1 million. And the average percent of shares

repurchased is less than 5%. Moreover the majority of these firms have market capitalizations

below $100 million. The exception is Liberty Media Capital Group, which had a market

capitalization of $15 billion and repurchased $460 million of shares in 2008. It experienced a

66
1,026% abnormal rate of return in the next 12 months, with an associated abnormal wealth gain

of $4.6 billion.

On the other hand, large negative future stock rates of return are also observed. The worst

case of poor post-repurchase share price performance occurred with the 1990 repurchase by

Colorcos Corporation, which experienced a −99.8% abnormal rate of return over the year after it

bought back stock. Other top 20 worst repurchasing firms experienced similar negative abnormal

returns in absolute value terms, ranging from −94.9% to −99.6%. In these cases, the firms also

spent very little on their repurchases. Half of them repurchased less than $1 million. Similar

results hold for using 36-month abnormal rates of return. It is worth noting that, by the nature of

rate of return calculations, the magnitude of the largest negative abnormal returns is much

smaller than the magnitude of the largest positive abnormal returns, as the former is bounded

from below at −100% while the later has no upper limit.

To investigate the magnitude of extreme wealth gains and losses, I examine the largest 20

winners and 20 losers based on abnormal dollar stock returns (Panels C and D of Table 6.3). The

biggest winner on a dollar-return basis is Exxon in 2007, which experienced an abnormal future

dollar return of more than $10 billion in 12 months, or about 2.3% of its market value at the end

of 2006. The rest of top winners experienced an abnormal future dollar return in the range from

$1.8 to $4.9 billion during the one-year period after repurchases.

Exxon is also the biggest loser on a dollar-return basis. It experienced an abnormal dollar

loss of more than $12 billion in 12 months after its huge repurchase of $34 billion in 2008. The

other top 20 worst repurchasing firms experienced an abnormal dollar loss in the range between

$1.3 and $5.4 billion.

67
 All of the top 20 best and top 20 worst repurchasing firms based on a dollar-return basis

spent a huge amount of money on their repurchases. In these subsamples, the range in the dollar

value repurchased is from $400 million to $30 billion, with 24 firms buying back at least $5

billion in stock. For the full sample of repurchases from 1971 to 2010, the mean dollar value

repurchased is $95 million and the median dollar value repurchased is $2. All the results reported

above are qualitatively the same when I measure abnormal dollar returns over a 36-month period

instead of a 12-month period (Table 6.4).

The analysis detailed above indicates that the extreme values of positive abnormal returns

are much larger than the extreme values of negative abnormal returns in absolute value,

suggesting that the mean value is likely to be affected by extreme values. Table 6.5 further

confirms this by reporting the full distributions of abnormal stock returns, which shows that the

mean values are significantly greater than the median values. To investigate the extent to which

the mean values are distorted by extreme values, Table 6.6 reports the full distributions of

abnormal returns for three groups of repurchases: (1) large gain repurchases are those with future

12-month abnormal returns belonging to the top decile of the full sample, (2) large loss

repurchases are those in the bottom decile of the full sample, and (3) the middle 80% of

repurchases, which represents all of the remaining cases.

An important finding from Table 6.6 is that, for the middle 80% of repurchases, the mean

and median abnormal future rates of returns are −1% and −3% respectively. These negative

future stock returns are inconsistent with the market-timing hypothesis. These repurchases

represent 94% of the aggregate dollar value repurchased in my full sample. When I examine

future 36-month abnormal stock returns for the middle 80%, I find the mean and median

68
abnormal future returns are −1% and −8% respectively. These repurchases represent 95% of the

aggregate dollar value repurchased in my full sample.

The results above suggest that the positive average abnormal return following repurchases is

mostly driven by extreme values. However, these extreme cases only represent a small portion of

aggregate dollar value of shares that are repurchased. In the example of the 12-month future

abnormal stock return, the large gain and large loss repurchases (the outer 20%) together

represent 6% of the aggregate dollar value repurchased. The comparable number in the case of

36-month future abnormal return is 5%.

To further clarify the importance of extreme values, Table 6.7 reports the full distributions

of future abnormal returns for the middle 80%, 90%, 95%, and 99% of repurchases. These data

show that the mean future abnormal return increases as more extreme values are included in the

average. If we focus on the middle 80% of all repurchases, the positive mean future abnormal

return turn is negative.

Lakonishok and Vermaelen (1990) report a positive mean abnormal return of 8% over two

years starting from one month after the expiration date for 258 fixed-price tender-offer

repurchases. Based on the evidence above for my full sample, I conjecture that their conclusions

are likely to be driven by extreme returns of tender-offer repurchases that represents a small

amount of total dollar value repurchased in tender offers.

To test this prediction, Table 6.8 documents the full distributions of future abnormal returns

for the middle 80%, 90%, 95% and 99% tender-offer repurchases. When I focus on the middle

80% of the tender-offer distribution, both the average 12-month and 36-month abnormal returns

become negative, just as I found for the full sample. These middle 80% of all tender offers

account for 90% of the aggregate dollar value repurchased through tender offers. When I focus

69
on the middle 90% of tender-offer distribution, the average abnormal return is 0.6% for both the

12-month and 36-month periods following tender offers. And these middle 90% of all tender

offers account for more than 95% of the aggregate dollar value repurchased through tender offers.

On the other hand, the middle 99% of tender offer do exhibit a positive mean future abnormal

return of 3.9% in one year and 10.0% in three years. The numbers reported immediately above

indicate that the extreme values of future abnormal returns following tender-offer repurchases

also drive the average values to be positive.

If one is concerned about the behavior of the “typical” manager, it is more appropriate to use

the median rather than mean abnormal returns to test the market-timing hypothesis. Similar to

the finding for the sample of all repurchases, the median tender-offer repurchase is also followed

by negative abnormal returns. The median 12-month post-repurchase abnormal return is −2.4%

and the median 36-month post-repurchase abnormal return is −8.2%. The median firm

experiences a zero abnormal dollar stock return during both the 12 months and 36 months after

tender-off repurchases. These negative median abnormal returns suggest that tender-offer

repurchases are more likely to be followed by negative than positive post-repurchase abnormal

returns and wealth changes. This is not what one would expect if market timing were the main

motive for tender-offer repurchases.

6.3 Characteristics of Large Gain and Large Loss Repurchases

The evidence so far suggests that the extreme values of future abnormal returns drive the

average abnormal return to be positive. It is natural to examine whether those extreme

repurchases are different from the others in terms of firm and repurchase characteristics. Table

6.9 accordingly reports the firm and repurchase characteristics for the subsamples of large gain,

large loss, and middle repurchases which represent the middle 80% of repurchases.

70
 When I identify large gain and large loss repurchasing firms based on abnormal rates of

returns, large gain and loss repurchases are generally smaller in dollar amount compared to the

middle repurchases. The mean dollar repurchase is $36 million for large-gain repurchases and

the mean dollar repurchase is $24 million for large-loss repurchases, while the mean value for

the middle repurchases is $118 million. However, in terms of percent of shares repurchased, all

three groups are similar. The median percent of shares repurchased is 1% for all three groups.

The average percent of share purchased is 4% for both large-gain and large-loss repurchases, and

is 3% for the middle repurchases. Moreover, both large-gain and large-loss repurchases are

generally conducted by small firms. These findings further support the view that extremely

positive and negative post-repurchase abnormal returns generally arise from small repurchases in

dollar size that are conducted by small firms.

Large gain repurchases are conducted by firms with a mean market-to-book ratio (M/B) of

1.74, while large loss repurchases are conducted by firms with a higher mean M/B ratio of 2.01.

The middle repurchases are conducted by firms with a mean M/B of 1.86, which falls between

the means for the large-gain and large-loss groups.

These statistics are consistent with the market-timing theory. More undervalued firms

(proxied by lower M/B ratios) experience higher abnormal stock returns after repurchases.

When I identify large-gain and large-loss repurchasing firms based on dollar abnormal

returns, I find that these repurchase types are conducted by firms that have average M/B ratios of

2.23 and 2.36, respectively. These average M/B ratios are significantly higher than the mean

M/B ratio of 1.75 for the group of firms that fall in the middle of the dollar abnormal return

repurchase distribution.

71
 It is not surprising that large-gain and large-loss repurchases are generally large in terms of

dollar value repurchased. The mean dollar values repurchased are $441 million and $478 million,

and the corresponding median dollar values repurchased are $120 million and $130 million. The

comparable mean and median dollar values repurchased for middle repurchases are $11 million

and $1 million. The fact that large-gain and large-loss repurchases on average have significantly

higher M/B ratios seems to suggest that large dollar amount repurchases are associated with high

M/B ratios.

My data confirm the latter conjecture. During the sample period over 1985 to 2009, I find

that a firm’s dollar amount of repurchase is positively correlated with its M/B ratio. This

correlation is statistically significant at the 0.0001 level. This result is inconsistent with the

market-timing hypothesis, which predicts that firms should repurchase less if their shares are

overvalued as proxied by high M/B ratios.

An alternative explanation is that M/B ratio is not a good measure of market mispricing. If

this is true, it is difficult to interpret the results from some studies such as Dittmar (2000). These

studies show that M/B ratio has a negative effect on repurchases and use this observation to

conclude that undervalued firms as proxied by low M/B ratios are more likely to repurchase their

shares. But if M/B ratio measures undervaluation poorly, their findings are not conclusive

evidence supporting the market-timing hypothesis.

The results in section are robust to using 36-month abnormal returns. Table 6.10 shows that,

when large-gain and large-loss repurchases are identified based on 36-month abnormal rates of

return, the mean dollar values repurchased are $30 million and $24 million respectively. These

figures are significantly smaller than the mean dollar value repurchased of $125 million for firms

in the middle repurchase group.

72
 When large-gain and large-loss repurchases are identified based on 36-month abnormal

dollar returns, the average M/B ratios are 2.22 and 2.37, respectively. These average M/B ratios

are significantly higher than the mean M/B ratio of 1.75 for the set of firms that falls in the

middle of the repurchase distribution. The mean dollar values repurchased for large-gain and

large-loss repurchases are $455 million and $506 million, respectively. The comparable mean

dollar values repurchased for the set of middle repurchases is $11 million.

In sum, the findings of this section support the view that extreme abnormal rates of return

are generally experienced by small firms that buy back small dollar amounts of stock.

More importantly, the large-gain and large-loss statistics reported in this section provide

mixed evidence for market timing. On the one hand, extremely positive abnormal rates of return

following share repurchases are associated with undervalued firms (proxied by low M/B ratios),

consistent with the market-timing theory. On the other hand, large repurchases in dollar size are

associated with higher M/B. This indicates firms buy back stock in large dollar amounts when

timing opportunities are poor (proxied by high M/B ratios), which is the opposite of the relation

predicted by the market-timing theory.

73
 Chapter 7

Equity Payouts and Capital Structure

The evidence in Chapter 4 and Chapter 5 indicates that leverage rebalancing is an important

consideration in explaining managers’ share-repurchase decisions. This chapter conducts a more

thorough examination of the interplay between equity payout and capital structure.

7.1 Background

Tradeoff theories of capital structure typically view optimizing leverage as managers’ main

concern, with payout policy either ignored, or treated as a mechanism through which they adjust

leverage toward a target ratio. The latter view is difficult to reconcile with the CFO survey

evidence of Brav et al. (2005), which reveals no sign that managers use payout policy to manage

leverage. Moreover, firms sometimes take on large amounts of debt to fund payouts that result in

leverage ratios that are so high that they are not plausibly viewed as permanent capital structure

targets. For example, in leveraged recapitalizations, Denis and Denis (1993) find that large

equity payouts are funded by new debt that increases the median firm’s debt-to-assets ratio from

0.45 to 0.86. Also, in recent years, many firms have followed LBOs with dividend

recapitalization plans in which they take on even more debt to fund payouts to their private-

equity owners. These features of leveraged and dividend recapitalization plans suggest that

managers sometimes treat the objective of delivering payouts to stockholders as more important

than attaining a target leverage ratio.

Although dividend payments and share repurchases increase the leverage ratios of firms

with debt outstanding, empirical studies of leverage rebalancing have largely treated payout

policy and capital structure as separable issues. Lie (2002) and Denis and McKeon (2012) are

notable exceptions, with both papers analyzing the impact of increases in equity payouts on

74
leverage rebalancing. Lie (2002) finds evidence supporting the view that firms undertake tender-

offer repurchases to rebalance back to the estimated target leverage ratios. In contrast, the

evidence of Denis and McKeon (2012) suggests that increasing payouts sometimes takes priority

over rebalancing to a target leverage ratio. They find that firms continue to increase payouts after

they make large debt issues that move firms’ capital structures further above their estimated

targets. This behavior is the antithesis of rebalancing toward a target leverage ratio.

In this chapter, I build on the studies by Lie (2002) and Denis and McKeon (2012) and

analyze whether observed financial policies reflect empirically significant trade-offs between

equity payout increases and rebalancing leverage toward a target capital structure. My main

focus is on gauging the extent to which firms’ dividend increases and share repurchases represent

(i) desired adjustments back toward their target leverage ratios as opposed to (ii) temporary

increases in leverage ratios that facilitate delivery of large payouts to stockholders.

To shed light on the causal effect of equity payout increases on firms’ capital structures, I

focus on the surge in specially designated dividends (SDDs) in late 2012 that were paid by firms

looking to beat the impending federal dividend-tax rate increase. According to the Wall Street

Journal, between November and December 2012, at least 191 firms paid SDDs that distributed

more than $22 billion to stockholders. I empirically assess the extent to which firms paid SDDs

in late 2012 because they were under-leveraged and sought to move toward capital structure

targets at a time when the adjustment cost (specifically, the tax cost of paying dividends) was

especially low.

7.2 Large and Debt-financed Payout Increases

According to the dynamic rebalancing theory, firms should adjust back toward leverage

targets when they are under-leveraged, and they should be especially aggressive about doing so

75
when they are deeply under-leveraged. In order to increase leverage, firms need to take on more

debt, or distribute cash to shareholders if they have debt outstanding, or do both.

If firms issue more debt solely for the purpose of rebalancing toward leverage targets rather

than to meet a current financial deficit, the proceeds from debt issuances should either be held

inside the firm as cash balances or paid out to shareholders. However, holding too much cash in

the firm is costly for agency reasons, as Jensen (1986) indicates. As a result, there is

corresponding pressure on rebalancing firms to pay out the proceeds from debt issuances. This

implies that when firms are under-leveraged and issue debt to adjust back toward target leverage

ratios, they should generally increase payouts at the same time.

Alternatively, instead of increasing payouts to rebalance leverage, firms could increase

payouts simply because they want to distribute excess and/or transitory cash. To investigate

whether firms use payout policy to manage their capital structures toward a target leverage ratio,

I first examine the time path of leverage after large payout increases.

For each firm, I compute the dollar amount of a dividend increase as the difference between

its annual common dividends (DVC in Compustat) in the current year and the prior year. The

payout increase in a given year is then the sum of the dollar value of the dividend increase and

the amount of the net repurchase in that year. In this analysis, the measure of the payout increase

is identical to what Denis and McKeon (2012) call the firm’s discretionary payout. Intuitively,

managers have not made any commitments to increase the dividend or to repurchase stock, and

so it makes sense to view the sum of any dividend increase and the current level of repurchases

as the discretionary amount of cash paid out in a given year. I define a large payout increase as

one that equals or exceeds 10% of a firm’s total assets as of the end of the prior year.

76
 For each large payout increase, I track the leverage ratio (debt-to-assets ratio) in a period

from three years before the payout increase to seven years after the payout increase. For firms

that make additional large payout increases during the seven-year period after the first increase, I

exclude those increases from the current sample to avoid double counting. After the seven-year

period following the first payout increase, any subsequent large payout increase is treated as an

additional observation.

Using a sample of industrial firms listed in the Compustat/CRSP merged file between 1951

and 2005, I identify 2,088 payout increases that meet the definition of “large” that is given above.

Among these payout increases, 1,066 are associated with debt increases in the same year, 671 are

associated with debt reductions in the same year, and 347 are associated with no change in debt.

Figure 7.1 depicts the evolution of leverage around these three groups of large payout increases.

Panel A of Figure 7.1 examines the cases in which firms make large payout increases and

fund them with debt. The evolution of median leverage ratio around these large payout increases

is consistent with the leverage-rebalancing theory. The large payout increases and the newly

issued debt move the median firm’s leverage ratio from well below the estimated target ratio to

just above the estimated target ratio. The median leverage ratio subsequently remains roughly at

the same level over the following seven years.

In contrast, Panel B of Figure 7.1 indicates that large payout increases that are accompanied

by debt reductions temporarily reduce the median firm’s leverage ratio. In the year of the large

payout increase and debt reduction, the median leverage ratio drops significantly from 0.13 to

0.08 (p-value < 0.001). In the years that follow, the median leverage ratio gradually increases

and returns to the original level. It approaches 0.13 by the end of the third year after the large

payout increase, and reaches 0.14 by the end of the seventh year following the payout increase.

77
 Panel C of Figure 7.1 reveals that firms that make large payout increases without changes in

debt typically have no debt in their capital structures. For this type of large payout increases, the

median firm has no debt in its capital structure during the period from three years before the

payout increase to seven years after the payout increase. However, the median target leverage

ratios are nontrivial during this period, ranging from 0.14 to 0.16.

In sum, the latter two types of large payout increases are not attempts to adjust leverage back

toward the estimated target ratios. Although the data suggest that large payout increases funded

by debt might be used for rebalancing leverage toward target ratios, the data also leave open the

possibility that making large payouts to shareholders could be more important than moving

closer to a target leverage ratio.

I accordingly further examine the extent to which debt-financed payout increases are used to

adjust leverage ratios back to targets. I restrict attention to cases in which the firm increases its

debt by at least 1% of its total assets as of the end of the fiscal year prior to the year in question.

For each firm, I compute the annual total debt increase as the difference between the book values

of total debt in the current and prior years. I define a payout increase as a debt-financed payout

increase if that payout increase exceeds 50% of the firm’s total debt increase.

There are 2,878 debt-financed payout increases in my full sample. Of these 2,878 cases, 863

(or roughly 30%) are associated with positive excess leverage before the payout increase. Excess

leverage is the difference between a firm’s debt-to-assets ratio and its estimated target leverage

ratio. In other words, roughly 30% of firms that issue debt to fund payout increases have debt-to-

assets ratios above their estimated targets before the payout increases. So, these firms are moving

further away from their estimated target leverage ratios. Obviously, then, these actions are not

compatible with managers’ rebalancing leverage toward target ratios.

78
 An important caveat is that this pattern could be driven by the underestimation of the

relevant target leverage ratios. Although I cannot rule out this possibility, this pattern is also

observed when I use median industry leverage ratio (2-digit SIC and Fama-French 49 industry)

to proxy for firms’ target leverage ratios.

Consistent with firms deliberately moving away from target leverage, Panel A of Figure 7.2

suggests that my sample firms use debt as a transitory vehicle to fund payout increases. In the

year that a firm makes a debt-financed payout increase, there is a statistically significant jump in

the median leverage ratio from 0.34 to 0.37 (p-value < 0.001). The resultant median leverage

ratio stays flat in the two-year period after a debt-financed payout increases, and gradually

returns to the pre-payout-increase leverage ratio. Moreover, I find that firms typically continue to

increase payouts after debt-financed payout increases. These data show that continuing to

increase payouts to shareholders often takes priority over rebalancing leverage back to the target

leverage ratio. Firms could have returned to their pre-payout-increase leverage level within two

years instead of 6 years had they not continued to increase payouts.

For firms that make debt-financed payout increases and are under-leveraged before the

payout increases, their leverage time paths show some support for the leverage-rebalancing

theory, but are not completely consistent with that theory. Panel B of Figure 7.2 indicates that

these firms typically experience a drop in leverage before they use debt to fund a payout increase.

During the three-year period before the payout increase, the median leverage ratio decreases

from 0.16 to 0.11. After debt-financed payout increases, the median firm’s leverage ratio

increases significantly from 0.11 to 0.15 (p-value < 0.001). However, the resultant median

leverage ratio is still below the estimated target. This result is also robust to using industry

median leverage ratio to proxy for the target leverage.

79
 Another important pattern from Panel B of Figure 7.2 is that, for under-leveraged firms that

issue debt to fund payout increases, median leverage ratio continues to increase during the seven-

year period following the debt-financed payout increase. The median firm remains under-

leveraged during this period. There is no sign that the estimated target leverage ratios increase

during the same time. Rather, this pattern indicates a continuous movement toward the target, but

the net result is that the median firm still does not wind up close to the estimated target ratio. A

movement toward target is consistent with the leverage-rebalancing theory. However, this

movement lasts for seven years and yet does not reach the target ratio, indicating a slow average

speed of adjustment.

In sum, the evidence in Figures 7.1 and 7.2 suggests that payout increases and accompanied

changes in debt have an important effect on a firm’s leverage ratio. However, the resultant

changes in leverage ratios (as in Panels B and C of Figure 7.1 and Panel A of Figure 7.2) are not

consistent with attempts to adjust back toward target leverage ratios.

These changes in leverage reflect two factors: the change in debt, and the payout increase.

To measure the magnitude of each factor, Panel A of Table 7.1 reports hypothetical leverage

ratios for the firm as if (i) it had not changed the amount of outstanding debt but still increased

its payout to shareholders, or (ii) it had not increased the dollar amount of the payout but still

made the same change in debt.

This comparison reveals that leverage changes after payout increases are mostly driven by

changes in the amount of the firm’s outstanding debt. For example, for firms that make large

payout increases with debt increases, the median leverage ratio rises from 11.5% to 27.1%. Had

the firm not increased the debt but made the same amount of payout increase, the median

leverage ratio would be 11.3%. This figure might seem counterintuitive as the leverage ratio falls

80
slightly when total assets shrink. This is due to the fact that median values do not correspond to

the same firm. On the other hand, had the firm not increased its payout but still increased the

same dollar amount of debt, the median leverage ratio would be 22.3%. This comparison

indicates that payout changes per se tend to have a very small effect on altering the firm’s actual

leverage ratio.

In contrast, an increase in payouts tends to have an important effect on firms’ cash holdings.

For example, Panel B of Table 7.1 shows that, if firms increase payouts without taking on more

debt, their cash-to-assets ratios would be much higher had they not increased payouts (columns 2

and 3). For example, for the median firm that makes a large payout increase and reduces debt at

the same time, the cash-to-assets ratio would be 0.33 instead of 0.17 had it not increased payouts

and reduced debt.

On the other hand, those firms that increase payouts and debt simultaneously would not have

been able increase payouts without raising external funds, as they did when they issued new debt

(columns 1, 4 and 5 in Panel B of Table 7.1). For the median firm that increases payout

significantly and borrows at the same time, the cash-to-assets ratio would be −0.04 instead of

0.06 absent the debt infusion. This, of course, could never happen since cash holdings can never

be less than zero. In any case, these data are silent on whether the main objective of these firms

is to increase leverage ratios or to deliver payouts to shareholders.

To further examine whether firms use payout increases to move their capital structures

toward a leverage target, I investigate the histogram of excess leverage ratios before payout

increases. Panel A of Table 7.2 indicates that a nontrivial minority of firms (24.5%) that

simultaneously make large payout increases and issue new debt have debt-to-assets ratios above

81
their estimated target leverage ratios. Clearly, these firms do not seek to move their capital

structures toward their leverage targets.

Moreover, Panel B of Table 7.2 suggests that even when firms have leverage ratios that are

far below targets, their propensity to make a large (or debt-financed) payout increase is weak.

For example, among firms that are more than 0.2 below their estimated target leverage ratios,

only 1.1% make a large payout increase accompanied by a debt increase. This figure is close to

the 0.3% probability that a firm with a 0.2 excess leverage ratio conducts a large payout increase

with a debt increase (Row 1 in Panel B of Table 7.2).

7.3 Logit Analysis of the Decision to Make Large and Debt-financed Payout Increases

The descriptive evidence in the previous section suggests that the tendency to use payout

policy to manage capital structure is weak at best. In this section, I implement a multivariate logit

analysis to assess the importance of the motives to rebalance leverage and to distribute excess

(and/or transitory cash) in explaining payout-increase decisions.

Specifically, the logit analysis assesses whether the payout-increase decision is negatively

correlated with the magnitude by which excess leverage falls below the estimated targets, and is

positively related to excess cash holdings and transitory cash flow. The dependent variable

equals one if a firm makes a large (debt-financed) payout increase, and zero if it does not. The

independent variables include excess leverage, a dummy variable called below, the interaction

term between below and excess leverage, excess cash holdings, and transitory cash flow. The

dummy variable labeled below takes the value one if the excess leverage is negative, and

otherwise equals zero. The estimated excess cash holdings is a firm’s cash-to-assets ratio minus

the industry median cash-to-assets ratio (Fama-French 49 industries). Transitory cash flow is a

firm’s non-operating income (NOPI in Compustat) scaled by its total assets. The logit models

82
also include the natural log of sales, the market-to-book ratio, and profitability as control

variables. I use the ratio of operating income (OIBDP in Compustat) to total assets as a proxy for

the firm’s current-period profitability.

The leverage-rebalancing theory predicts that, when leverage is below target, the probability

of a large (debt-financed) repurchase is greater the more negative is the value of excess leverage,

i.e., the further below target is the value of actual leverage. The theory also predicts that, when

leverage is above target, the magnitude of excess leverage should have no effect on the decision

to increase payouts. My logit analyses yield results that support the rebalancing theory for large

payout increases but not for debt-financed payout increases.

Specifically, for large payout increases, Table 7.3 shows that the excess leverage ratio has an

insignificantly negative effect. The interaction effect between excess leverage and the dummy

variable labeled below is statically significantly negative (t-statistic = −2.34 and −2.48 in

Columns 1 and 2). These estimates indicate that firms with lower leverage ratios below the

estimated targets are more likely to make a large payout increase, and excess leverage ratios

above targets have no statistically significant effect. This finding is consistent with the

rebalancing theory.

For debt-financed payout increases, Table 7.3 shows that excess leverage has a significantly

negative effect on the decision to conduct a debt-financed payout increase (t-statistic = −5.76 and

−5.93 in Columns 5 and 6). However, the interaction effect between excess leverage and the

dummy variable labeled below is statically insignificant.

These estimates together imply that excess leverage has the same directional impact on the

probability of a debt-financed payout increase regardless of whether the firm is currently above

or below its estimated leverage target. One, of course, expects that lower excess leverage (i.e.,

83
actual leverage further below target) will lead to more debt-financed payout increases when

leverage is currently below target. It is tougher to explain why they would undertake such actions

when leverage is already above target. This cannot be explained by standard target-rebalancing

logic. It could be explained by a managerial willingness to increase debt and payouts when the

firm has a greater amount of untapped debt capacity.

Table 7.3 also reveals that both excess cash holdings and transitory cash flow have a

statistically significantly positive effect on the large-payout-increase decision (t-statistics = 7.45

and 2.80, Column 1 in Table 7.3). Colum 5 in Table 7.3 indicates that, although transitory cash

flow also has a significant positive effect (t-statistic = 5.35) on the decision to make a debt-

financed payout increase, the excess-cash-holding variable exerts a statistically significantly

negative effect on that decision (t-statistic = −2.53).

The results above collectively suggest that firms with higher excess cash holdings are more

likely to make a large payout increase, but are less likely to conduct a debt-financed payout

increase. On the other hand, higher transitory cash flows are associated with both a higher

probability of large payout increases (which could, but need not, be financed by debt) and a

higher probability of debt-financed payout increases.

These results together suggest that, while distributing excess cash is empirically relevant for

explaining large payout increases, it doesn’t explain the decision to conduct a debt-financed

payout increase. On the other hand, distributing transitory cash flow is empirically relevant for

explaining both large and debt-financed payout decisions.

To gauge the economic significance of the leverage-rebalancing motive in explaining

payout-increase decisions, I use the coefficient estimates from the model in column 1 of Table

7.3 to compute the estimated probability of a payout increase as a function of excess leverage

84
ratio while holding the other variables constant at their sample mean values. I find that the

impact of variation in excess leverage ratios is small in magnitude. Panel A of Table 7.4

indicates that when the excess leverage ratio increases from one standard deviation below the

mean value to one standard deviation above the mean value, the estimated probability of a large

payout increase declines from 1.7% to 0.9%. The comparable figures for debt-financed payout

increases are 2.4% and 1.3%. These small declines reflect the fact that the vast majority of firms

that are extremely under-leveraged do not increase equity payouts. These findings are

incompatible with the view that deeply under-leveraged firms move their capital structures

toward targets by increasing equity payouts.

The estimates above show that even though variation in excess leverage translates to a large

relative change in the probability of a payout increase, the absolute change is very small. This is

consistent with the view that, although managing capital structure toward target ratios is

empirically relevant to the decision to increase payouts, it is not of first-order importance.

Almost identical results are obtained when I estimate the payout-increase probability as a

function of excess leverage while holding the other variables constant at their sample median

values (Table 7.5).

I also use the similar approach to gauge the economic significance of excess cash holdings

and transitory cash flow. Both of these variables show only a tiny effect on the decision to make

a large or debt-financed payout increase. For example, when excess cash holdings swing from

one standard deviation below the mean value to one standard deviation above the mean value,

the absolute change in the probability of a large payout increase is roughly 0.7%. The

comparable figure for transitory cash flow is 0.2%.

85
 Overall, the evidence in this section suggests that, although large payout increases

sometimes move firms toward their estimated target leverage ratios, the tendency to do so is

weak even when firms are deeply under-leveraged. The motives to distribute excess and/or

transitory cash also exert weak influence on large and debt-financed payout increase decisions.

These findings together indicate that neither managing capital structure toward target leverage

ratios nor distributing excess cash (or transitory cash flow) is the primary motive driving the

decision to make large and/or debt-financed payout increases.

7.4 Specially Designed Dividends (SDDs) in Late 2012

An important feature of the leverage-rebalancing theory is that firms do not adjust their

leverage ratios toward targets frequently when it is costly to make such adjustments. If firms use

payout increases to move their leverage ratios toward targets, the cost of increasing payouts is

part of the adjustment costs. In particular, dividend taxes are effectively adjustment costs when

firms use payout increases to adjust their leverage toward targets. Consequently, a drop in

dividend tax rates should encourage under-leveraged firms to rebalance their capital structures

toward target leverage ratios.

The impending increase in the dividend tax rate in late 2012 provides a unique opportunity

to examine the causal relationship between payout increases and rebalancing toward target

leverage ratios. The political climate in late 2012 led to a consensus that federal dividend tax

rates would increase significantly in 2013. In other words, the cost of using dividend increases to

adjust firms’ leverage toward targets was perceived to be much higher after 2012. As a result, the

looming increase in dividend taxes should have encouraged under-leveraged firms to move

toward their targets with debt-financed special payouts.

86
 I first investigate whether managers treated dividend taxes as an important factor when

making decisions to pay out specially designed dividends (SDDs) near the end of 2012. Figure

7.3 depicts the number of firms that pay special dividends in each quarter over 2011 and 2012. I

identify SDDs from the CRSP distribution file. A cash distribution is classified as a specially

designated dividend if it carries distribution code 1262 or 1272, the code numbers that CRSP

uses to identify cash dividends labeled year-end, final, extra, or special.

Figure 7.3 shows that there is a clear surge in SDDs in late 2012. In the seven quarters prior

to the last quarter of 2012, there are less than 16 SDDs identified in each quarter using the CRSP

distribution file. However, in the last quarter of 2012, I identify 105 SDDs from the CRSP

distribution file. These figures support the view that taxes are an important cost of dividend

payouts and that, in late 2012, firms increased dividend payments to beat the impending

dividend-tax rate increase.

I note that CRSP sometimes does not code SDDs correctly. Some SDDs reported in the Wall

Street Journal (WSJ) are coded as regular dividends in the CRSP distribution file. In the

following analysis, I use the SDD sample obtained from the WSJ. I then obtained each firm’s

financial statement data and stock price information from Compustat and CRSP. The final SDD

sample consists of 126 firms that paid SDDs in November and December 2012 and that have the

required data available from Compustat and CRSP.

For firms that pay SDDs in late 2012, Table 7.6 documents firm characteristics and leverage

ratios around that time and compares them with other firms that did not announce SDDs, but that

are listed in the last quarter of 2012 in the CRSP/Compustat merged file. Row 1 of Table 7.6

indicates that SDDs payers typically have a debt-to-assets ratio below their estimated leverage

87
targets before they paid SDDs. Before paying out SDDs, the median SDD payer has a debt-to-

assets ratio of 0.06 and an excess leverage ratio of −0.12.

However, I find no evidence in Table 7.6 that SDD payers adjust their leverage ratios back

toward their estimated targets by paying SDDs. After paying SDDs, both the median change in

leverage and the median change in total debt are zero4. As a result, these SDD payers typically

remain under-leveraged after they paid SDDs (Columns 5 and 6).

I also conduct a logit analysis to assess the importance of leverage rebalancing and of

distributing excess (and/or transitory) cash in explaining SDD decisions. The dependent variable

takes the value one if a firm paid a SDD in November or December 2012, and otherwise equals

zero. As in the previous section, the independent variables include excess leverage, the dummy

variable labeled below, the interaction term between the dummy variable labeled below and

excess leverage, excess cash holdings, and transitory cash flow. The logit models also include

the same set of control variables as in Section 7.3.

Table 7.7 reports that both excess leverage and the interaction between excess leverage and

below have statistically insignificant effects on the decision to pay a SDD in late 2012 (Columns

1 and 2). I also find that excess cash holdings exert a statistically significantly positive effect (t-

statistic = 2.27 and 4.13 in Columns 1 and 3). On the other hand, transitory cash flow has a

statistically insignificant effect (t-statistic = 1.82).

I then use the coefficients estimated from the model in Column 1 of Table 7.7 to compute

the SDD probability as a function of excess leverage while holding the other variables constant

at their sample mean values. The estimated probabilities in Table 7.8 indicate that excess

4
For a firm that shrinks its total assets by paying a SDD, the leverage ratio should fall if it has debt outstanding and
the change in debt is zero. This seems to be contradicted with the descriptive statistics reported here as both the
median change in leverage and the median change in total debt are zero. It is important to note that median values
correspond to different firms. The firm that has a zero leverage change isn’t the firm that has a zero debt change.

88
leverage and excess cash holdings have modest influence on the decision to pay a SDD in late

2012. A change in excess leverage from one standard deviation above the mean value to one

standard deviation below the mean value induces an increase of the SDD probability from 2.4%

to 6.2%, or an absolute change of 3.8%. An important caveat here is that this change in the SDD

probability is not statistically significant. The t-statistic is 0.04 for the effect of excess leverage

and it is −0.79 for the interaction effect between below and excess leverage on the decision to

pay a SDD in late 2012.

Overall, the results in Tables 7.6, 7.7 and 7.8 together suggest that under-leveraged firms did

not generally use SDDs to move toward capital structure targets even when the leverage-

adjustment costs were perceived to be relatively low in late 2012. This finding provides further

support for the view that managing capital structure toward targets is not the primary motive

driving the decision to increase payouts to shareholders.

89
 Chapter 8

Conclusion and Discussion

Although market-timing motives are empirically important, free-cash-flow (FCF)

considerations have considerably stronger effects on managers’ share-repurchase decisions.

Large changes in market-timing opportunities translate to only modest changes in the estimated

probability of a share repurchase. Moreover, a large majority of firms with attractive market-

timing opportunities choose not to repurchase shares.

On the other hand, high levels of FCF greatly increase the probability of a share repurchase,

and this effect clearly dominates the market-timing consideration. Firms with bad market-timing

opportunities and high FCF are more than 12 times as likely to repurchase stock as firms with

good timing opportunities and low FCF. All these findings together suggest that the FCF effect

quantitatively dominates the market-timing effect in explaining managers’ repurchase decisions.

In addition, I find evidence that supports both employee stock options (ESOs) and leverage

rebalancing as important considerations in managers’ share-repurchase decisions. A large

increase in the standardized number of ESOs outstanding leads to a notably higher probability of

a share repurchase, regardless of whether market-timing opportunities are favorable or

unfavorable. A larger deviation below the estimated target leverage ratio is also associated with a

higher repurchase probability.

Moreover, in the logit analyses that exclude the M/B ratio, ESO considerations are

quantitatively comparable to market-timing considerations. The same logit model indicates that

leverage-rebalancing considerations are more important than market timing. In the logit analyses

that include the M/B ratio, the ESO effect is quantitatively weaker than the market-timing effect,

while the leverage-rebalancing effect is comparable to the market-timing effect. None of the

90
market-timing, ESO, and leverage-rebalancing effects is as strong as the FCF effect in explaining

managers' share-repurchase decisions.

An important caveat here is that, while these conclusions are generally robust to a variety of

proxies for market-timing opportunities, the results are weaker for particular sample periods and

model formulations. In particular, when my logit analyses are extended to the period from 1971

to 2010 and exclude control variables for ESO and leverage-rebalancing motives, the

quantitative advantage of FCF over market-timing considerations becomes smaller, but it is still

economically material.

Moreover, when I focus on a subsample that contains only tender-offer repurchases, none of

the FCF, market timing, ESO and leverage-rebalancing proxies is of first-order importance in

explaining managerial tender-offer repurchase decisions. And my logit analyses show no

convincing evidence that any one of them is systematically more important than the other one.

My evidence is strongly supportive of the view that, when managers choose to repurchase

stock, their decisions are typically poor in a market-timing sense. Firms are more likely to

experience negative than positive abnormal stock returns after share repurchases, as the median

post-repurchase abnormal returns are −3% over the 12 months following the year in which a firm

repurchases shares and −7% over the 36 months after the year in which a firm buys back stock.

The mean post-repurchase abnormal returns are positive as in prior studies. However, these

positive average post-repurchase abnormal returns are mostly driven by extreme abnormal

returns of firms that repurchase small amounts of stock in both percentage and dollar terms. As a

result, the mean values are not as informative as the median values regarding what is more likely

to happen to the stock price following repurchases.

91
 Overall, the evidence on the post-repurchase abnormal returns and on the relative

importance of market timing and other motives together suggest that market-timing

considerations are only marginally influential on the decision to conduct a repurchase. Managers

either do not systematically time the market to buy back shares or their timing ability is poor.

Why do managers so often fail to time the market and miss opportunities to repurchase

undervalued shares? One possibility is that managers simply have little ability to predict stock

returns. This explanation is consistent with the recent criticism from practitioners who blame

managers for buying back shares at prices that seemed too high. For example, financial

institutions such as AIG, Citigroup, JP Morgan, Wachovia, and WAMU all repurchased stock at

high prices not long before the financial crisis in 2008. Their share prices fell sharply as the crisis

developed and led to significant loses on these repurchases.

Another possible reason why managers may fail to exploit market-timing opportunities is

that investor rationality limits managers’ attempts to intentionally and regularly repurchase

undervalued shares. If managers knowingly and regularly buy back undervalued shares, investors

will eventually learn the ploy and should be able to implement trading strategy that can eliminate

timing opportunities and make share repurchases unattractive in a market-timing sense. As a

result, share repurchases would not become a pervasive economic phenomenon if market timing

were the primary motive.

The evidence, however, indicates that share repurchases have become quite popular in

recent years. For example, among firms that are listed at least 10 years during the period from

1997 to 2010, 36.0% repurchase stock at least one out of every two years, and 62.4% buy back

shares at least one out of every four years. It seems unlikely that such consistent and frequent

92
repurchase activities are driven by the market-timing motive as it seems implausible that a large

portion of public firms are undervalued for extended periods of time.

In sum, firms have a weak tendency to exploit market-timing opportunities through buying

back stock. And that market-timing effect is quantitatively dominated by the FCF effect.

Moreover, the market-timing effect is also not systematically more important than the ESO and

leverage-rebalancing effects in explaining managers’ stock repurchase decisions. These findings

suggest that a desire to time the market is not the primary motive driving the decision to

repurchase stock. Rather, the desire to distribute FCF is an empirically more important

determinant of managers’ share-repurchase decisions.

Although “market timing” in the discussion above refers to the mispricing version of the

timing hypothesis, similar analyses apply to the rational asymmetric information and the

managerial perceptions versions. In both of the latter versions, firms are not systematically

undervalued at a time when they buy back their own shares. This is consistent with the finding

that future abnormal stock returns have little influence on repurchase decisions, as well as the

finding that post-repurchase abnormal returns are more likely to be negative than positive.

On the other hand, the fact that FCF considerations are more important than market-timing

considerations in explaining repurchase decisions is inconsistent with the notion that managerial

perceptions alone drive share repurchase decisions. At the same time, however, the data do not

rule out the possibility that managerial beliefs, which are inherently difficult to observe, have

some effect on share repurchase decisions over and above the impact of FCF.

My main conclusion, then, is that the desire to exploit market undervaluation, either by

rational or irrational managers, is not the primary motive for share-repurchase decisions,

although there are some signs of market-timing behavior at work.

93
 References

Ai, C. R., & Norton, E. C. (2003). Interaction terms in logit and probit models. Economics
Letters, 80(1), 13-129.
Bagwell, L., & Shoven, J. (1988). Share repurchases and acquisition: an analysis of which firms
participate. In Auerbach, A. J. (Ed.), Corporate Takeovers: Causes and Consequence (pp.
191-220). Chicago, IL: University of Chicago Press.
Bagwell, L., & Shoven, J. (1989). Cash distributions to shareholders. Journal of Economic
Perspectives, 3, 129-149.
Baker, M., & Wurgler, J. (2002). Market timing and capital structure. Journal of Finance, 57, 1-
32.
Baker, M., Stein, J., & Wurgler, J. (2003). When does the stock market matter? Stock prices and
the investment of equity-dependent firms. Quarterly Journal of Economics, 118, 969-
1005.
Bens, D. A., Nagar, V., Skinner, D. J., & Wong, M.H. F. (2003). Employee stock options, EPS
dilution, and stock repurchase. Journal of Accounting and Economics, 36, 51-90.
Bonaime, A. A., Hankins, K. W., & Jordan, B. D. (2014). Wiser to wait: Do firms optimally
execute share repurchases? (Working paper), Retrieved March 19, 2014 from
http://ssrn.com/abstract=1977654.
Brav, A., Graham, J. R., Harvey, C., & Michaely, R. (2005). Payout policy in the 21st century.
Journal of Financial Economics, 77, 483-527.
Core, J., Guay, W., & Kothari, S. (2002). The economic dilution of employee stock options:
diluted EPS for valuation and financial reporting. The Accounting Review, 77, 627-652.
Comment, R., & Jarrell, G. A. (1991). The relative signaling power of Dutch-auction and fixed-
price self-tender offers and open-market share repurchases. Journal of Finance, 46(4),
1243-1271.
Dann. L. Y. (1981). Common stock repurchase: an analysis of returns to bondholders and
stockholders. Journal of Financial Economics, 9, 113-138.
DeAngelo, H., DeAngelo, L., & Skinner, D. J. (2008). Corporate payout policy. Foundations
and Trends in Finance, 3, 95-287.
DeAngelo, H., DeAngelo, L., & Stulz, R. M. (2006). Dividend policy and the earned/contributed
capital mix: a test of the lifecycle theory. Journal of Financial Economics, 81, 293-315.
DeAngelo, H., DeAngelo, L., & Stulz, R. M. (2010). Seasoned equity offerings, market timing,
and the corporate lifecycle. Journal of Financial Economics, 95, 275-295.
Dittmar, A. K. (2000). Why do firms repurchase stock? Journal of Business, 73, 331-355.
Dittmar, A. K, & Dittmar, R. F. (2008). The timing of financing decisions: an examination of the
correlation in financing waves. Journal of Financial Economics, 90, 59-83.
Denis, D. J., & Denis, D. K. (1993). Managerial discretion, organizational structure, and
corporate performance. Journal of Accounting and Economics, 16, 209-236.
Denis, D. J., & McKeon, S. B. (2012). Debt financing and financial flexibility: evidence from
proactive leverage increases. Review of Financial Studies, 25, 1897-1929.

94
Fama, E. F., & French, K. R. (2001). Disappearing dividends: changing firm characteristics or
lower propensity to pay? Journal of Financial Economics, 60, 3-43.
Fama, E. F., & French, K. R. (2005). Financial decisions: who issue stock? Journal of Financial
Economics, 76, 549-582.
Fenn, G. W., & Liang, N. (2001). Corporate payout policy and managerial stock incentives.
Journal of Financial Economics, 60, 45-72.
Grullon, G., & Michaely, R., (2004). The information content of share repurchase programs.
Journal of Finance, 59, 651-680.
Guay, W., & Harford, J., (2000). The cash-flow permanence and information content of dividend
increases versus repurchases. Journal of Financial Economics, 57, 385-415.
Hovakimian, A., Opler, T., & Titman, S. (2001). The debt-equity choice. Journal of Financial
and Quantitative Analysis, 36, 1-24.
Huang, R., & Ritter, J. R. (2009). Testing theories of capital structure and estimating the speed of
adjustment. Journal of Financial and Quantitative Analysis, 44, 237-271.
Ikenberry, D., Lakonishok, J., & Vermaelen, T. (1995). Market underreaction to open market
repurchases. Journal of Financial Economics, 39, 181-208.
Ikenberry, D., & Vermaelen, T. (2000). Stock repurchases in Canada: performance and strategic
trading. Journal of Finance, 55, 2373-2397.
Jagannathan, M., Stephens, C. P., & Weisbach, M. S. (2000). Financial flexibility and the choice
between dividends and stock repurchases. Journal of Financial Economics, 57, 355-384.
Jennings, R., LeClere, M., & Thompson, R. (1997). Evidence on the usefulness of alternative
earnings per share measures. Journal of Financial and Quantitative Analysis, 32, 24-33.
Jensen, M. C. (1986). Agency costs of free cash flow. American Economic Review, 76, 323-329.
Jolls, C. (1998). Stock repurchase and incentive compensation. (NBER working paper series
w6467), Retrieved March 19, 2014 from: http://ssrn.com/abstract=226212.
Kahle, K. M. (2002). When a buyback isn’t a buyback: open market repurchases and employee
options. Journal of Financial Economics, 63, 235-261.
Lakonishok, J., & Vermaelen, T. (1990). Anomalous price behavior around repurchase tender
offers. Journal of Finance, 45 (2), 455-477.
Lambert, R. A., Lanen, W. N., & Larcker, D. F. (1989). Executive stock option plans and
corporate dividend policy. Journal of Financial and Quantitative Analysis, 24, 409-425.
Lie, E. (2002). Do firms under take self-tender offers to optimize capital structure? Journal of
Business, 75 (4), 609-639.
Manconi, A., Peyer, U., & Vermaelen, T. (2013). Buybacks around the world. (INSEAD
working paper No. 2013/101/FIN), Retrieved March 19, 2014 from:
http://ssrn.com/abstract=2330807.
Masulis, R. W. (1980). Stock repurchase by tender offer: an analysis of the causes of common
stock price changes. Journal of Finance, 35, 305-319.
Murphy, K. J. (1999). Executive compensation. In Ashenfelter, O., Card, D. (Ed.), Handbook of
Labor Economics Vol. 3b (pp. 2485-2563). North Holland: Elsevier Science.

95
Nohel, T., & Tarhan, V. (1998). Share repurchases and firm performance: new evidence on the
agency costs of free cash flow. Journal of Financial Economics, 49, 187-222.
Opler, T., Pinkowitz, L., Stulz, R., & Williamson, R. (1999). The determinants and implications
of corporate cash holdings. Journal of Financial Economics, 52, 3-46.
Petersen, M. (2009). Estimating standard errors in finance panel date sets: comparing approaches.
Review of Financial Studies, 22, 435-480.
Peyer, U., & Vermaelen, T. (2005). The many facts of privately negotiated stock repurchases.
Journal of Financial Economics, 75, 361-395.
Peyer, U., & Vermaelen, T. (2009). The nature and persistence of buyback anomalies. Review of
Financial Studies, 22(4), 1693-1745.
Polk, C., & Sapienza, P. (2009). The stock market and corporate investment: a test of catering
theory. Review of Financial Studies, 22, 187-217.
Rajan, R.R., & Zingales, L. (1995). What do we know about capital structure: some evidence
from international data. Journal of Finance, 50, 1421-1460.
Rhodes-Kropf, M., Robinson, D., & Viswanathan, S. (2005). Valuation waves and merger
activity: the empirical evidence. Journal of Financial economics, 77, 561-603.
Skinner, D.J. (2008). The evolving relation between earnings, dividends, and stock repurchases.
Journal of Financial Economics, 87, 582-609.
Stephens, C.P., & Weisbach, M.J. (1998). Actual share reacquisitions in open-market repurchase
programs. Journal of Finance, 53, 313-334.
Stulz, R. (1990). Managerial discretion and optimal financial policies. Journal of Financial
Economics, 26, 3-27.
Vermaelen, T. (1981). Common stock repurchases and market signaling: an empirical study.
Journal of Financial Economics, 9, 139-183.
Vermaelen, T. (2005). Share repurchases. Foundations and Trends in Finance, 1, 171-268.
Weisbenner, S. J. (2000). Corporate share repurchases in the 1990s: what role do stock options
play? (Federal Reserve Board Working paper No. 2000-29), Retrieved March 19, 2014
from http://ssrn.com/abstract=234164.

96
 Bibliography

Denis, D. J. (1990). Defensive changes in corporate payout policy: share repurchases and special
dividends. Journal of Finance, 45(5), 1433-1456.

97
 Figure 3.1
Aggregate share repurchases, dividends and level of S&P 500 index for each year from 1971 to 2010

This figure depicts the total repurchases and dividends of all sample firms in each year from 1971 to 2010.
Repurchase is net repurchase as in Fama and French (2001), measured as the increase in common treasury stock
amounts. If the treasury stock amount is zero in the current and previous year or if the treasury stock amount is not
available, repurchase is measured as the difference between the stock purchase and stock issuance amounts from
Compustat. Dividends are the cash dividends paid by common shares reported by Compustat. The sample contains
14,958 industrial firms in the CRSP/Compustat merged files over 1971 to 2010 that (i) have SIC codes outside the
intervals 4900-4949 (utilities) and 6000-6999 (financials), (ii) have a CRSP share code of 10 or 11, (iii) are
incorporated in the US, and (iv) have non-missing total assets from Compustat. Level of S&P 500 index is the
average of daily S&P 500 index value in each year. All dollar amounts are inflation adjusted using the Consumer
Price Index (CPI) and are expressed in 2010 dollars.

700 1600

600 1400

1200
500

S&P 500 Index

1000
$ Billions

400
800
300
600
200
400

100 200

0 0
1971 1974 1977 1980 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010

Total Dividends Total Repurchases S&P 500 Index

98
 Figure 3.2
Percent of firms that repurchase stock and pay dividends in each year from 1971 to 2010

This figure depicts the percent of all sample firms that are repurchasers and dividend payers in in each year from
1971 to 2010. A firm is a repurchaser if it has a positive net repurchase as defined in Figure 1. A dividend payer is a
firm that pays cash dividends to common shareholders. A payers is a firm that is either a repurchaser or a dividend
payer. The sample contains 14,958 industrial firms in the CRSP/Compustat merged files over 1971 to 2010 that (i)
have SIC codes outside the intervals 4900-4949 (utilities) and 6000-6999 (financials), (ii) have a CRSP share code
of 10 or 11, (iii) are incorporated in the US, and (iv) have non-missing total assets from Compustat.

80%

70%

60%
Percent of firms

50%

40%

30%

20%

10%

0%
1971 1974 1977 1980 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010

Repurchasers Dividend payers Payers

99
 Figure 4.1
Interaction effects of operating-cash-flow and timing variables on share repurchase decisions

The estimated interaction effects are based on the model in Row D of Table 4.1 plus the interactions between
operating cash flow and each timing variable: the standardized market-to-book ratio, and the prior and future 12-
month stock returns. Since the method developed in Ai and Norton (2003) only computes the marginal effect of one
interaction term, I estimate the interaction effects separately for each timing variable. All variables are as defined in
Table 4.1.

Panel A. Interaction effects between operating cash flow and the standardized market-to-book ratio
.05

5
0
Interaction effect
0

T-statistic
-5
-. 05

-10
-15
-. 1

0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1
Estimated repurchase probability Estimated repurchase probability

Panel B. Interaction effects between operating cash flow and the prior 12-month stock return
5
.1

0
Interaction effect
0

T-statistic
-5
-.1

-10
-.2

-15

0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1
Estimated repurchase probability Estimated repurchase probability

Panel C. Interaction effects between operating cash flow and the future 12-month stock return
.1

2
.05

1
Interaction effect

T-statistic
0
0
-. 05

-1
-2
-. 1

-3
-.15

0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1
Estimated repurchase probability Estimated repurchase probability

100
 Figure 7.1
Evolution of leverage around large payout increases

The sample includes 1,066 (671) large payout increases with debt increases (reductions), 347 large payout increases
without debt changes between 1951 and 2005. Year 0 refers to the year in which a firm’s payout increase is at least
10% of its total assets as of the end of the prior fiscal year. Payout increase is the sum of the dollar value of the
annual dividend increase and repurchase amounts. Leverage is the ratio of total debt to total assets in book value
terms. Excess leverage is defined as the difference between the actual leverage and the estimated target. Target
leverage is the fitted value from a linear regression of debt-to-assets ratio on log(sales), market-to-book ratio,
profitability, asset tangibility, and industry median leverage. Median values are plotted here. All results are robust to
using the constant composition sample in which all firms have data available from year -3 through year 7 and to
using industry median leverage ratios as targets.

Panel A. Large payout increases with debt increases

0.30
0.25
0.20
0.15
Leverage

0.10
Leverage
0.05
Excess leverage
0.00
Target leverage
-0.05
-0.10
-0.15
-3 -2 -1 0 1 2 3 4 5 6 7
Year relative to payout increase

Panel B. Large payout increases with debt reductions

0.25
0.20
0.15
0.10
Leverage

0.05 Leverage

0.00 Excess leverage

Target leverage
-0.05
-0.10
-0.15
-3 -2 -1 0 1 2 3 4 5 6 7
Year relative to payout increase

101
Panel C. Large payout increases without changes in debt

0.20
0.15
0.10
0.05
Leverage

0.00 Leverage

-0.05 Excess leverage

Target leverage
-0.10
-0.15
-0.20
-3 -2 -1 0 1 2 3 4 5 6 7
Year relative to payout increase

102
 Figure 7.2
Evolution of leverage around debt-financed payout increases

The sample includes 863 (2,015) debt-financed payout increases with positive (negative) prior excess leverage
between 1951 and 2005. Year 0 refers to the year in which a firm’s payout increase exceeds 50% of its total debt
increase. In this analysis, the total debt increase must be at least 1% of total assets as of the end of the prior fiscal
year. Payout increase is the sum of the dollar value of the annual dividend increase and repurchase amounts.
Leverage is the ratio of total debt to total assets in book value terms. Excess leverage is the difference between the
actual leverage and estimated target. Target leverage is the fitted value from a linear regression of debt-to-assets ratio
on log(sales), market-to-book ratio, profitability, asset tangibility, and industry median leverage. Pro forma excess
leverage is computed as if all payout increases since year 1 had been used to reduce debt. If the payout increase in
any year exceeds the total debt outstanding, the excess cash saving is assumed to be used to reduce debt in future
years. Median values are plotted here. All results are robust to using the constant composition sample in which all
firms have data available from year -3 through year 7 and to using industry median leverage ratios as targets.

Panel A. Excess leverage > 0 at year -1

0.40
0.35
0.30
0.25
Leverage

0.20 Leverage
0.15 Excess leverage
0.10 Target leverage
0.05 Pro forma excess leverage
0.00
-0.05
-3 -2 -1 0 1 2 3 4 5 6 7
Year relative to payout increase

Panel B. Excess leverage < 0 at year -1

0.30
0.25
0.20
0.15
Leverage

0.10 Leverage
0.05 Excess leverage
0.00 Target leverage
-0.05 Pro forma excess leverage
-0.10
-0.15
-3 -2 -1 0 1 2 3 4 5 6 7
Year relative to payout increase

103
 Figure 7.3
Specially designated dividends (SDDs) in 2011 and 2012

The sample is obtained from CRSP distribution file. A cash distribution is classified as a specially designated if it
carries distribution code 1262 or 1272, the code numbers that CRSP uses to identify cash dividends labeled year-
end, final, extra, or special. Dollar amount of SDDs is defined as the product of the dividend cash amount per share
and the number of shares outstanding on ex-dividend date. Market capitalization is the product of share price and the
number of shares outstanding on ex-dividend date.

120 9.0%

$SDDs/Market Capitalization
8.0%
100
7.0%
Number of SDDs

80 6.0%
5.0%
60
4.0%
40 3.0%
2.0%
20
1.0%
0 0.0%
2011Q1 2011Q2 2011Q3 2011Q4 2012Q1 2012Q2 2012Q3 2012Q4
Number of SDDs Median value of $SDDs/Market Capitalization

104
 Table 2.1
Three versions of the market-timing theory of share repurchases

This table lists key features of different versions of the market-timing theory that explains share repurchase
decisions.

1. Mispricing 2. Rational Asymmetric 3. Managerial

Information Perceptions
Assumptions
Managers Rational Rational Overconfident
Investors Irrational Rational Either
Stock market implications
Announcement return Positive Positive Positive
Past excess return Negative Zero Zero/positive
Future excess return Positive Zero Zero/negative
Market-to-book ratio Low No prediction Moderate/high
Requiring other motives
for repurchases? No Yes No

105
 Table 3.1
Aggregate tender-offer repurchases and all repurchases from 1985 to 2010

Tender-offer repurchase announcements and transaction values are obtained from SDC. For all repurchases, the
value of repurchase is net repurchase as in Fama and French (2001), measured as the increase in common treasury
stock amounts. If the treasury stock amout is zero in the current and previous year or if the treasury stock amount is
not available, repurchase is measured as the difference between the stock purchase and stock issuance amounts from
Compustat. All firms in the sample (i) have SIC codes outside the intervals 4900-4949 (utilities) and 6000-6999
(financials), (ii) have a CRSP share code of 10 or 11, (iii) are incorporated within the US, and (iv) have non-missing
total assets from Compustat. All dollar amounts are inflation adjusted using the Consumer Price Index (CPI) and are
expressed in 2010 dollars.

Tender-offer Repurchases All Repurchases

Number Percent of all Value Percent of aggregate Number Aggregate value
Year of firms repurchasers ($billions) value repurchased of firms ($billions)
1985 15 1.6% $4.2 7.4% 918 $56.4
1986 14 1.5% $8.9 16.9% 952 $52.3
1987 35 2.7% $9.7 13.7% 1,312 $71.1
1988 35 2.9% $17.1 24.0% 1,225 $71.3
1989 37 3.5% $11.5 20.7% 1,050 $55.3
1990 23 1.9% $7.0 13.4% 1,188 $51.8
1991 19 2.2% $0.7 3.2% 868 $23.3
1992 27 3.2% $3.2 10.3% 838 $31.5
1993 26 2.9% $2.0 6.8% 891 $29.9
1994 30 2.9% $4.2 9.2% 1,045 $45.6
1995 22 1.9% $4.0 5.1% 1,156 $79.1
1996 34 2.7% $3.3 4.0% 1,246 $81.4
1997 34 2.3% $9.1 7.7% 1,449 $117.8
1998 26 1.5% $7.1 4.6% 1,763 $155.4
1999 23 1.4% $5.0 2.9% 1,692 $171.1
2000 22 1.4% $7.1 4.4% 1,523 $161.9
2001 13 1.0% $0.2 0.2% 1,330 $110.0
2002 16 1.3% $1.7 1.5% 1,208 $117.8
2003 23 2.2% $1.9 1.6% 1,043 $121.9
2004 11 1.2% $7.4 4.2% 945 $175.0
2005 24 2.3% $11.7 4.4% 1,049 $265.4
2006 26 2.2% $18.3 4.6% 1,177 $393.4
2007 29 2.3% $23.2 5.1% 1,265 $451.7
2008 23 1.6% $6.9 2.3% 1,402 $307.7
2009 6 0.6% $0.6 0.4% 964 $131.7
2010 7 0.8% $3.4 2.1% 880 $161.5
All Years 600 2.0% $179.4 5.1% 30,379 $3,491.4

106
 Table 3.2
Distribution of percent of shares repurchased

Dollar value of repurchases is the total amount repurchased reported by SDC for tender-offer repurchases. For other
repurchases, dollar value of repurchases is net repurchases as in the Fama and French (2001). Percent of shares
repurchased is the exact percent of shares repurchased reported by SDC for tender-offer repurchases. For other
repurchases, percent of shares repurchased is measured as the ratio between the dollar value of the repurchase and
the market capitalization as of the end of the prior year. For each group of firms that have a percent of shares
repurchased falls in a specified interval, I compute the percent of total dollar repurchases as the total dollar amount
of repurchases of every firm in that group divided by the total dollar amount of all repurchases in the sample. All
firms in the sample (i) have SIC codes outside the intervals 4900-4949 (utilities) and 6000-6999 (financials), (ii)
have a CRSP share code of 10 or 11, (iii) are incorporated within the US, and (iv) have non-missing total assets from
Compustat. The sample period is from 1985 to 2010.

Percent of shares repurchased in the interval:

A. All Repurchasers 0% to 1% 1% to 5% 5% to 10% above 10%
Percent of firms
that repurchase stock 43.8% 37.3% 11.7% 7.2%

Cumulative percent of firms

that repurchase stock 43.8% 81.1% 92.8% 100.0%

Percent of total
dollar repurchases 4.7% 46.4% 29.9% 19.1%

Cumulative percent of
total dollar repurchases 4.7% 51.1% 80.9% 100.0%

B. Repurchasers using tender offers

Percent of firms
that repurchase stock 8.8% 13.3% 19.4% 58.5%

Cumulative percent of firms

that repurchase stock 8.8% 22.1% 41.5% 100.0%

Percent of total
dollar repurchases 1.0% 5.8% 20.1% 73.1%

Cumulative percent of
total dollar repurchases 1.0% 6.7% 26.9% 100.0%

107
 Table 3.3
Aggregate SEOs and all equity issues from 1985 to 2010

SEO announcements and total proceeds are obtained from SDC. All equity issues are market measures of net equity
issued as in Fama and French (2005), i.e., the product of (1) the split-adjusted growth in shares and (2) the average
of the split-adjusted stock price at the beginning and end of the fiscal year. All firms in the sample (i) have SIC
codes outside the intervals 4900-4949 (utilities) and 6000-6999 (financials), (ii) have a CRSP share code of 10 or 11,
(iii) are incorporated within the US, and (iv) have non-missing total assets from Compustat. All dollar amounts are
inflation adjusted using the Consumer Price Index (CPI) and are expressed in 2010 dollars.

SEOs All Equity Issues

Number Percent of Total proceeds Percent of aggregate Number Aggregate amount
Year of firms all Issuers ($billion) amount issued of firms ($billion)
1985 177 7.6% $15.1 25.2% 2,314 $60.0
1986 186 7.9% $16.2 20.1% 2,346 $80.4
1987 136 6.1% $14.6 20.6% 2,238 $70.9
1988 68 3.1% $5.4 10.7% 2,187 $51.0
1989 99 4.3% $5.3 6.7% 2,318 $79.4
1990 78 3.7% $6.5 10.9% 2,088 $59.6
1991 223 9.2% $26.3 23.8% 2,431 $110.7
1992 192 7.1% $20.3 20.6% 2,713 $98.2
1993 251 8.3% $27.5 18.8% 3,006 $145.9
1994 198 6.1% $21.0 13.6% 3,252 $154.5
1995 261 7.7% $35.8 17.5% 3,368 $204.7
1996 316 8.9% $34.6 12.6% 3,570 $273.5
1997 264 6.9% $32.2 8.5% 3,837 $376.7
1998 186 5.6% $30.8 7.0% 3,349 $439.7
1999 181 6.0% $39.3 4.7% 3,022 $835.8
2000 174 5.7% $51.4 4.7% 3,027 $1,085.9
2001 150 4.8% $33.4 6.4% 3,100 $523.4
2002 146 5.1% $25.0 9.2% 2,845 $272.3
2003 175 6.3% $26.1 11.1% 2,763 $235.9
2004 198 7.2% $37.1 15.8% 2,756 $235.1
2005 145 5.6% $25.4 8.6% 2,610 $296.3
2006 141 5.8% $26.3 8.2% 2,414 $320.8
2007 114 5.1% $26.0 14.5% 2,226 $179.5
2008 66 3.4% $33.2 20.9% 1,933 $158.9
2009 231 10.3% $31.3 15.5% 2,252 $202.4
2010 167 9.0% $21.3 11.0% 1,862 $193.9
All Years 4,523 6.5% $667.6 9.9% 69,827 $6,745.5

108
 Table 4.1
Logit analysis of share repurchase decisions

The dependent variable equals one if a firm repurchases at least 1% of its shares during the year in question or zero
otherwise. The independent variables are (i) the standardized market-to-book (M/B) ratio at the end of the fiscal year
prior to the year in question, (ii) the market-adjusted abnormal return over the 12 months (or over the 36 months in
rows C and F) ending immediately before the year in question, (iii) the market-adjusted excess return over the 12
months (or over the 36 months in rows C and F) beginning immediately after the year in question, (iv) operating
cash flow at the end of the fiscal year immediately before the year in question, (v) the change in options outstanding
at the end of the year in question, and (vi) the deviation from the target leverage ratio at the end of the fiscal year
immediately before the year in question. The standardized M/B ratio in a given year is the firm’s M/B ratio divided
by the median M/B ratio in that year for all firms. The abnormal return is the firm’s actual stock return minus the
contemporaneous return on the value-weighted market index. Operating cash flow is operating income before
depreciation (OIBDP in Compustat) deflated by total assets. Change in options outstanding is computed as the
difference between options outstanding/common shares outstanding in the current and prior year. Target leverage
ratio is the fitted value from a liner regression of debt-to-assets ratio on variables often hypothesized to affect
leverage decisions: log(sales), market-to-book ratio, profitability and asset tangibility. All models include firm size
as proxied by the natural log of total assets. The sample period is from 1996 to 2010.

Market to book Prior stock Future stock Operating Option Deviation

Intercept ratio return return cash flow change from target
A. All firms
Coefficient -2.588 -0.041 -0.133 0.018
(Marginal probability) (-0.007) (-0.022) (0.003)
[t-statistic] [-11.70] [-1.01] [-2.36] [0.53]
B. All firms with raw M/B in lieu of standardized M/B
Coefficient -2.618 -0.014 -0.143 0.018
(Marginal probability) (-0.002) (-0.024) (0.003)
[t-statistic] [-11.72] [-0.53] [-2.68] [0.55]
C. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -2.484 -0.038 -0.038 0.032
(Marginal probability) (-0.0068) (-0.007) (0.006)
[t-statistic] [-10.16] [-0.91] [-1.78] [1.95]
D. All firms
Coefficient -2.936 -0.095 -0.248 -0.060 4.831 5.390 -1.336
(Marginal probability) (-0.020) (-0.052) (-0.012) (1.003) (1.119) (-0.277)
[t-statistic] [-10.58] [-1.57] [-4.62] [-2.04] [9.12] [3.61] [-6.89]
E. All firms with raw M/B in lieu of standardized M/B
Coefficient -2.989 -0.025 -0.279 -0.059 4.609 5.249 -1.335
(Marginal probability) (-0.005) (-0.058) (-0.012) (0.953) (1.085) (-0.276)
[t-statistic] [-10.54] [-0.52] [-5.59] [-2.07] [8.40] [3.49] [-7.00]
F. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -3.082 -0.160 -0.077 0.015 5.590 5.191 -1.165
(Marginal probability) (-0.034) (-0.016) (0.003) (1.194) (1.109) (-0.249)
[t-statistic] [-11.78] [-2.56] [-2.46] [1.22] [11.95] [3.23] [-5.56]

109
 Table 4.2
Estimated probability of a share repurchase

This table reports the probability of a repurchase conditional on specific hypothesized values of timing variables and operating cash flow. All other independent
variables are set equal to their sample median values. A firm is defined as a repurchaser if it buys back at least 1% of its shares in the year in question. In Panel A,
the estimated probabilities are based on the model in Row D of Table 4.1. Row 1 reports the probability of a repurchase for a firm that faces neutral market-
timing opportunities. Rows 2-7 show the impact of changing each timing variable by large amounts, while holding other timing variables constant at their sample
median values. Rows 8-13 describe the impact of large swings in future abnormal returns while holding M/B ratios and prior excess stock returns constant at
values that represent highly favorable or unfavorable timing opportunities. The probability of a repurchase for firms that face highly unfavorable versus highly
favorable timing opportunities is given in Rows 14-15 respectively. The estimated probabilities in the far right column are based on the model in Row A of Table
4.1, which includes only timing variables and firm size. Panel B is similar to Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
standardized prior excess future excess a function of percentile of operating cash flow/assets of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no operating-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 50th 2.33% 19.18% 27.29% 33.66% 46.38% 21.57%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 1.67% 14.41% 21.03% 26.47% 38.02% 18.63%
3. 50th 5th 50th 2.71% 21.67% 30.44% 37.17% 50.20% 22.99%
4. 50th 50th 95th 2.15% 17.94% 25.70% 31.87% 44.35% 21.99%
5. 50th 50th 5th 2.43% 19.81% 28.10% 34.57% 47.38% 21.36%
6. 95th 50th 50th 1.77% 15.18% 22.06% 27.67% 39.47% 19.57%
7. 5th 50th 50th 2.44% 19.91% 28.23% 34.71% 47.54% 21.91%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 2.95% 23.18% 32.31% 39.22% 52.38% 23.13%
9. 5th 5th 50th 2.83% 22.47% 31.43% 38.26% 51.37% 23.35%
10. 5th 5th 95th 2.62% 21.08% 29.70% 36.35% 49.33% 23.79%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 1.31% 11.67% 17.29% 22.03% 32.51% 16.67%
12. 95th 95th 50th 1.26% 11.26% 16.72% 21.35% 31.63% 16.84%
13. 95th 95th 95th 1.16% 10.47% 15.61% 20.01% 29.89% 17.18%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 1.31% 11.67% 17.29% 22.03% 32.51% 16.67%
15. 5th 5th 95th 2.62% 21.08% 29.70% 36.35% 49.33% 23.79%
110
Panel B.
Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
prior excess future excess a function of percentile of operating cash flow/assets of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no operating-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 2.84% 19.58% 27.10% 32.94% 44.57% 21.41%
Effect of large variation in each market-timing variable
2. 95th 50th 1.89% 13.87% 19.75% 24.53% 34.73% 18.02%
3. 5th 50th 3.39% 22.64% 30.89% 37.13% 49.15% 23.07%
4. 50th 95th 2.62% 18.34% 25.55% 31.19% 42.59% 21.84%
5. 50th 5th 2.94% 20.15% 27.82% 33.74% 45.45% 21.22%
Future returns effect, given very low prior returns
6. 5th 5th 3.51% 23.27% 31.66% 37.97% 50.05% 22.87%
7. 5th 50th 3.39% 22.64% 30.89% 37.13% 49.15% 23.07%
8. 5th 95th 3.14% 21.26% 29.20% 35.28% 47.14% 23.52%
Future returns effect, given very high prior returns
9. 95th 5th 1.96% 14.31% 20.32% 25.20% 35.54% 17.85%
10. 95th 50th 1.89% 13.87% 19.75% 24.53% 34.73% 18.02%
11. 95th 95th 1.75% 12.94% 18.50% 23.08% 32.93% 18.40%
Extremely unfavorable versus favorable timing opportunities
12. 95th 5th 1.96% 14.31% 20.32% 25.20% 35.54% 17.85%
13. 5th 95th 3.14% 21.26% 29.20% 35.28% 47.14% 23.52%

111
 Table 4.3
Logit analysis of share repurchase decisions as a function of free cash flow

The dependent variable equals one if a firm repurchases at least 1% of its shares during the year in question or zero
otherwise. The independent variables are (i) the standardized market-to-book (M/B) ratio at the end of the fiscal year
prior to the year in question, (ii) the market-adjusted abnormal return over the 12 months (or over the 36 months in
rows C and F) ending immediately before the year in question, (iii) the market-adjusted excess return over the 12
months (or over the 36 months in rows C and F) beginning immediately after the year in question, (iv) free cash
flow at the end of the fiscal year immediately before the year in question, (v) the change in options outstanding at
the end of the year in question, and (vi) the deviation from the target leverage ratio at the end of the fiscal year
immediately before the year in question. The standardized M/B ratio in a given year is the firm’s M/B ratio divided
by the median M/B ratio in that year for all firms. The abnormal return is the firm’s actual stock return minus the
contemporaneous return on the value-weighted market index. Free cash flow is the ratio of operating income before
depreciation (OIBDP in Compustat) to total assets minus the median industry capital expenditure ratio. To compute
the median industry capital expenditure ratio in each year, I first divide capital expenditures (CAPX in Compustat)
by total assets for each firm. Then I obtain the median value of all firms in the same 2-digit SIC code industry
excluding the firm in question. Change in options outstanding is computed as the difference between options
outstanding/common shares outstanding in the current and prior year. Target leverage ratio is the fitted value from a
liner regression of debt-to-assets ratio on variables often hypothesized to affect leverage decisions: log(sales),
market-to-book ratio, profitability and asset tangibility. All models include firm size as proxied by the natural log of
total assets. The sample period is from 1996 to 2010.

Market to book Prior stock Future stock Free cash Option Deviation
Intercept ratio return return flow change from target
A. All firms
Coefficient -2.588 -0.041 -0.133 0.018
(Marginal probability) (0.007) (-0.022) (0.003)
[t-statistic] [-11.70] [-1.01] [-2.36] [0.53]
B. All firms with raw M/B in lieu of standardized M/B
Coefficient -2.618 -0.014 -0.143 0.018
(Marginal probability) (0.0024) (-0.024) (0.003)
[t-statistic] [-11.72] [-0.53] [-2.68] [0.55]
C. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -2.484 -0.038 -0.038 0.032
(Marginal probability) (-0.0068) (-0.007) (0.006)
[t-statistic] [-10.16] [-0.91] [-1.78] [1.95]
D. All firms
Coefficient -2.810 -0.141 -0.243 -0.056 5.414 5.810 -1.432
(Marginal probability) (-0.029) (-0.051) (-0.012) (1.129) (1.212) (-0.299)
[t-statistic] [-10.14] [-2.40] [-4.74] [-1.96] [10.70] [3.86] [-7.09]
E. All firms with raw M/B in lieu of standardized M/B
Coefficient -2.863 -0.051 -0.273 -0.057 5.152 5.615 -1.427
(Marginal probability) (-0.011) (-0.057) (-0.012) (1.070) (1.166) (-0.296)
[t-statistic] [-10.09] [-1.07] [-5.96] [-2.02] [9.42] [3.69] [-7.22]
F. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -2.930 -0.209 -0.070 0.018 6.145 5.751 -1.270
(Marginal probability) (-0.045) (-0.015) (0.004) (1.318) (1.234) (-0.272)
[t-statistic] [-10.66] [-3.64] [-2.46] [1.59] [13.69] [3.50] [-5.77]

112
 Table 4.4
Estimated probability of a share repurchase as a function of free cash flow

This table reports the probability of a repurchase conditional on specific hypothesized values of timing variables and free cash flow. All other independent
variables are set equal to their sample median values. A firm is defined as a repurchaser if it buys back at least 1% of its shares in the year in question. In Panel A,
the estimated probabilities are based on the model in Row D of Table 4.3. Row 1 reports the probability of a repurchase for a firm that faces neutral market-
timing opportunities. Rows 2-7 show the impact of changing each timing variable by large amounts, while holding other timing variables constant at their sample
median values. Rows 8-13 describe the impact of large swings in future abnormal returns while holding M/B ratios and prior excess stock returns constant at
values that represent highly favorable or unfavorable timing opportunities. The probability of a repurchase for firms that face highly unfavorable versus highly
favorable timing opportunities is given in Rows 14-15 respectively. The estimated probabilities in the far right column are based on the model in Row A of Table
4.3, which includes only timing variables and firm size. Panel B is similar to Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
standardized prior excess future excess a function of percentile of free cash flow/assets of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no free-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 50th 1.75% 17.78% 26.95% 34.26% 48.59% 21.57%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 1.26% 13.40% 20.87% 27.15% 40.33% 18.63%
3. 50th 5th 50th 2.03% 20.08% 29.99% 37.70% 52.32% 22.99%
4. 50th 50th 95th 1.62% 16.68% 25.45% 32.54% 46.66% 21.99%
5. 50th 50th 5th 1.82% 18.35% 27.70% 35.12% 49.54% 21.36%
6. 95th 50th 50th 1.16% 12.46% 19.53% 25.54% 38.34% 19.57%
7. 5th 50th 50th 1.87% 18.82% 28.32% 35.83% 50.31% 21.91%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 2.25% 21.85% 32.29% 40.25% 54.99% 23.13%
9. 5th 5th 50th 2.17% 21.21% 31.46% 39.34% 54.04% 23.35%
10. 5th 5th 95th 2.01% 19.95% 29.82% 37.51% 52.12% 23.79%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 0.86% 9.56% 15.28% 20.30% 31.60% 16.67%
12. 95th 95th 50th 0.83% 9.24% 14.79% 19.69% 30.78% 16.84%
13. 95th 95th 95th 0.77% 8.61% 13.84% 18.50% 29.16% 17.18%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 0.86% 9.56% 15.28% 20.30% 31.60% 16.67%
15. 5th 5th 95th 2.01% 19.95% 29.82% 37.51% 52.12% 23.79%
113
Panel B.
Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
prior excess future excess a function of percentile of free cash flow/assets of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no free-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 2.39% 18.49% 26.74% 33.18% 45.77% 21.41%
Effect of large variation in each market-timing variable
2. 95th 50th 1.56% 12.82% 19.13% 24.35% 35.35% 18.02%
3. 5th 50th 2.89% 21.59% 30.70% 37.60% 50.60% 23.07%
4. 50th 95th 2.22% 17.38% 25.29% 31.53% 43.90% 21.84%
5. 50th 5th 2.47% 19.01% 27.40% 33.93% 46.60% 21.22%
Future returns effect, given very low prior returns
6. 5th 5th 2.99% 22.16% 31.42% 38.40% 51.44% 22.87%
7. 5th 50th 2.89% 21.59% 30.70% 37.60% 50.60% 23.07%
8. 5th 95th 2.68% 20.34% 29.11% 35.85% 48.71% 23.52%
Future returns effect, given very high prior returns
9. 95th 5th 1.62% 13.20% 19.66% 24.97% 36.13% 17.85%
10. 95th 50th 1.56% 12.82% 19.13% 24.35% 35.35% 18.02%
11. 95th 95th 1.45% 12.00% 17.99% 22.98% 33.65% 18.40%
Extremely unfavorable versus favorable timing opportunities
12. 95th 5th 1.62% 13.20% 19.66% 24.97% 36.13% 17.85%
13. 5th 95th 2.68% 20.34% 29.11% 35.85% 48.71% 23.52%

114
 Table 4.5
Logit analysis of share repurchase decisions as a function of transitory cash flow

The dependent variable equals one if a firm repurchases at least 1% of its shares during the year in question or zero
otherwise. The independent variables are (i) the standardized market-to-book (M/B) ratio at the end of the fiscal year
prior to the year in question, (ii) the market-adjusted abnormal return over the 12 months (or over the 36 months in
rows C and F) ending immediately before the year in question, (iii) the market-adjusted excess return over the 12
months (or over the 36 months in rows C and F) beginning immediately after the year in question, (iv) operating
cash flow at the end of the fiscal year immediately before the year in question, (v) transitory cash flow at the end of
the fiscal year immediately before the year in question (vi) the change in options outstanding at the end of the year
in question, and (vii) the deviation from the target leverage ratio at the end of the fiscal year immediately before the
year in question. The standardized M/B ratio in a given year is the firm’s M/B ratio divided by the median M/B ratio
in that year for all firms. The abnormal return is the firm’s actual stock return minus the contemporaneous return on
the value-weighted market index. Operating cash flow is operating income before depreciation (OIBDP in
Compustat) deflated by total assets. Transitory cash flow is non-operating income (NOPI in Compustat) deflated by
total assets. Change in options outstanding is computed as the difference between options outstanding/common
shares outstanding in the current and prior year. Target leverage ratio is the fitted value from a liner regression of
debt-to-assets ratio on variables often hypothesized to affect leverage decisions: log(sales), market-to-book ratio,
profitability and asset tangibility. All models include firm size as proxied by the natural log of total assets. The
sample period is from 1996 to 2010.

Market to book Prior stock Future stock Operating Transitory Option Deviation
Intercept ratio return return cash flow cash flow change from target
A. All firms
Coefficient -2.588 -0.041 -0.133 0.018
(Marginal probability) (0.007) (-0.022) (0.003)
[t-statistic] [-11.70] [-1.01] [-2.36] [0.53]
B. All firms with raw M/B in lieu of standardized M/B
Coefficient -2.618 -0.014 -0.143 0.018
(Marginal probability) (0.0024) (-0.024) (0.003)
[t-statistic] [-11.72] [-0.53] [-2.68] [0.55]
C. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -2.484 -0.038 -0.038 0.032
(Marginal probability) (-0.0068) (-0.007) (0.006)
[t-statistic] [-10.16] [-0.91] [-1.78] [1.95]
D. All firms
Coefficient -2.957 -0.104 -0.249 -0.060 4.904 2.496 5.326 -1.312
(Marginal probability) (-0.021) (-0.052) (-0.012) (1.016) (0.517) (1.103) (-0.272)
[t-statistic] [-10.82] [-1.71] [-4.66] [-2.00] [9.29] [3.40] [3.61] [-6.85]
E. All firms with raw M/B in lieu of standardized M/B
Coefficient -3.006 -0.030 -0.279 -0.059 4.674 2.245 5.186 -1.313
(Marginal probability) (-0.006) (-0.058) (-0.012) (0.964) (0.463) (1.070) (-0.271)
[t-statistic] [-10.75] [-0.62] [-5.63] [-2.04] [8.50] [2.95] [3.49] [-6.96]
F. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -3.106 -0.169 -0.077 0.015 5.676 2.624 5.110 -1.142
(Marginal probability) (-0.036) (-0.017) (0.003) (1.209) (0.559) (1.089) (-0.243)
[t-statistic] [-12.00] [-2.75] [-2.49] [1.22] [12.38] [3.63] [3.22] [-5.52]

115
 Table 4.6
Estimated probability of a share repurchase as a function of transitory cash flow

This table reports the probability of a repurchase conditional on specific hypothesized values of timing variables and transitory cash flow. All other independent
variables are set equal to their sample median values. A firm is defined as a repurchaser if it buys back at least 1% of its shares in the year in question. In Panel A,
the estimated probabilities are based on the model in Row D of Table 4.5. Row 1 reports the probability of a repurchase for a firm that faces neutral market-
timing opportunities. Rows 2-7 show the impact of changing each timing variable by large amounts, while holding other timing variables constant at their sample
median values. Rows 8-13 describe the impact of large swings in future abnormal returns while holding M/B and prior excess stock returns constant at values
that represent highly favorable or unfavorable timing opportunities. The probability of a repurchase for firms that face highly unfavorable versus highly favorable
timing opportunities is given in Rows 14-15 respectively. The estimated probabilities in the far right column are based on the model in Row A of Table 4.5,
which includes only timing variables and firm size. Panel B is similar to Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
standardized prior excess future excess a function of percentile of transitory cash flow/assets of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no transitory-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 50th 29.02% 29.41% 29.66% 30.17% 31.65% 21.57%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 22.47% 22.80% 23.01% 23.45% 24.71% 18.63%
3. 50th 5th 50th 32.28% 32.69% 32.95% 33.50% 35.06% 22.99%
4. 50th 50th 95th 27.36% 27.74% 27.98% 28.47% 29.90% 21.99%
5. 50th 50th 5th 29.86% 30.26% 30.51% 31.03% 32.53% 21.36%
6. 95th 50th 50th 23.11% 23.45% 23.66% 24.11% 25.39% 19.57%
7. 5th 50th 50th 30.07% 30.47% 30.72% 31.25% 32.75% 21.91%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 34.30% 34.73% 35.00% 35.56% 37.16% 23.13%
9. 5th 5th 50th 33.40% 33.82% 34.08% 34.64% 36.22% 23.35%
10. 5th 5th 95th 31.60% 32.01% 32.27% 32.81% 34.35% 23.79%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 18.16% 18.44% 18.62% 18.99% 20.08% 16.67%
12. 95th 95th 50th 17.56% 17.84% 18.02% 18.38% 19.44% 16.84%
13. 95th 95th 95th 16.41% 16.67% 16.84% 17.18% 18.19% 17.18%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 18.16% 18.44% 18.62% 18.99% 20.08% 16.67%
15. 5th 5th 95th 31.60% 32.01% 32.27% 32.81% 34.35% 23.79%
116
Panel B.
Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
prior excess future excess a function of percentile of transitory cash flow/assets of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no transitory-cash-flow effect)
Neutral market-timing opportunities
1. 0.0% 0.0% 28.74% 29.06% 29.25% 29.66% 30.84% 21.41%
Effect of large variation in each market-timing variable
2. 90.0% 0.0% 20.98% 21.24% 21.40% 21.73% 22.70% 18.02%
3. -90.0% 0.0% 32.71% 33.05% 33.26% 33.70% 34.96% 23.07%
4. 0.0% 90.0% 27.12% 27.43% 27.62% 28.01% 29.16% 21.84%
5. 0.0% -90.0% 29.48% 29.80% 30.00% 30.42% 31.61% 21.22%
Future returns effect, given very low prior returns
6. -90.0% -90.0% 33.50% 33.85% 34.06% 34.51% 35.78% 22.87%
7. -90.0% 0.0% 32.71% 33.05% 33.26% 33.70% 34.96% 23.07%
8. -90.0% 90.0% 30.96% 31.29% 31.50% 31.93% 33.15% 23.52%
Future returns effect, given very high prior returns
9. 90.0% -90.0% 21.58% 21.84% 22.01% 22.35% 23.33% 17.85%
10. 90.0% 0.0% 20.98% 21.24% 21.40% 21.73% 22.70% 18.02%
11. 90.0% 90.0% 19.68% 19.92% 20.07% 20.39% 21.32% 18.40%
Extremely unfavorable versus favorable timing opportunities
12. 90.0% -90.0% 21.58% 21.84% 22.01% 22.35% 23.33% 17.85%
13. -90.0% 90.0% 30.96% 31.29% 31.50% 31.93% 33.15% 23.52%

117
 Table 4.7
Logit analysis of share repurchase decisions as a function of cash holdings

The dependent variable equals one if a firm repurchases at least 1% of its shares during the year in question or zero
otherwise. The independent variables are (i) the standardized market-to-book (M/B) ratio at the end of the fiscal year
prior to the year in question, (ii) the market-adjusted abnormal return over the 12 months (or over the 36 months in
rows C and F) ending immediately before the year in question, (iii) the market-adjusted excess return over the 12
months (or over the 36 months in rows C and F) beginning immediately after the year in question, (iv) operating
cash flow at the end of the fiscal year immediately before the year in question, (v) cash holdings at the end of the
fiscal year immediately before the year in question (vi) the change in options outstanding at the end of the year in
question, and (vii) the deviation from the target leverage ratio at the end of the fiscal year immediately before the
year in question. The standardized M/B ratio in a given year is the firm’s M/B ratio divided by the median M/B ratio
in that year for all firms. The abnormal return is the firm’s actual stock return minus the contemporaneous return on
the value-weighted market index. Operating cash flow is operating income before depreciation (OIBDP in
Compustat) deflated by total assets. Cash holdings is cash and short-term investments (CHE in Compustat) deflated
by total assets. Change in options outstanding is computed as the difference between options outstanding/common
shares outstanding in the current and prior year. Target leverage ratio is the fitted value from a liner regression of
debt-to-assets ratio on variables often hypothesized to affect leverage decisions: log(sales), market-to-book ratio,
profitability and asset tangibility. All models include firm size as proxied by the natural log of total assets. The
sample period is from 1996 to 2010.

Market to book Prior stock Future stock Operating Cash Option Deviation
Intercept ratio return return cash flow holding change from target
A. All firms
Coefficient -2.588 -0.041 -0.133 0.018
(Marginal probability) (0.007) (-0.022) (0.003)
[t-statistic] [-11.70] [-1.01] [-2.36] [0.53]
B. All firms with raw M/B in lieu of standardized M/B
Coefficient -2.618 -0.014 -0.143 0.018
(Marginal probability) (0.0024) (-0.024) (0.003)
[t-statistic] [-11.72] [-0.53] [-2.68] [0.55]
C. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -2.484 -0.038 -0.038 0.032
(Marginal probability) (-0.0068) (-0.007) (0.006)
[t-statistic] [-10.16] [-0.91] [-1.78] [1.95]
D. All firms
Coefficient -3.439 -0.236 -0.224 -0.060 5.793 1.596 5.640 -1.058
(Marginal probability) (-0.047) (-0.044) (-0.012) (1.148) (0.316) (1.118) (-0.210)
[t-statistic] [-11.99] [-4.20] [-4.32] [-1.81] [10.19] [6.76] [3.94] [-5.86]
E. All firms with raw M/B in lieu of standardized M/B
Coefficient -3.433 -0.108 -0.255 -0.061 5.486 1.462 5.391 -1.088
(Marginal probability) (-0.021) (-0.051) (-0.012) (1.087) (0.290) (1.068) (-0.216)
[t-statistic] [-11.90] [-2.16] [-5.71] [-1.89] [8.56] [5.85] [3.68] [-5.95]
F. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -3.612 -0.302 -0.082 0.016 6.659 1.657 5.433 -0.884
(Marginal probability) (-0.062) (-0.017) (0.003) (1.362) (0.339) (1.111) (-0.181)
[t-statistic] [-12.21] [-5.91] [-2.61] [1.31] [14.48] [7.49] [3.50] [-4.42]

118
 Table 4.8
Estimated probability of a share repurchase as a function of cash holdings

This table reports the probability of a repurchase conditional on specific hypothesized values of timing variables and cash holdings. All other independent
variables are set equal to their sample median values. A firm is defined as a repurchaser if it buys back at least 1% of its shares in the year in question. In Panel A,
the estimated probabilities are based on the model in Row D of Table 4.7. Row 1 reports the probability of a repurchase for a firm that faces neutral market-
timing opportunities. Rows 2-7 show the impact of changing each timing variable by large amounts, while holding other timing variables constant at their sample
median values. Rows 8-13 describe the impact of large swings in future abnormal returns while holding M/B ratios and prior excess stock returns constant at
values that represent highly favorable or unfavorable timing opportunities. The probability of a repurchase for firms that face highly unfavorable versus highly
favorable timing opportunities is given in Rows 14-15 respectively. The estimated probabilities in the far right column are based on the model in Row A of Table
4.7, which includes only timing variables and firm size. Panel B is similar to Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
standardized prior excess future excess a function of percentile of cash/assets of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no cash-holding effect)
Neutral market-timing opportunities
1. 50th 50th 50th 25.07% 25.75% 28.14% 35.07% 51.43% 21.57%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 19.71% 20.28% 22.32% 28.38% 43.72% 18.63%
3. 50th 5th 50th 27.75% 28.48% 31.02% 38.28% 54.88% 22.99%
4. 50th 50th 95th 23.56% 24.22% 26.51% 33.23% 49.38% 21.99%
5. 50th 50th 5th 25.84% 26.53% 28.97% 36.00% 52.44% 21.36%
6. 95th 50th 50th 14.24% 14.68% 16.27% 21.14% 34.45% 19.57%
7. 5th 50th 50th 27.30% 28.02% 30.53% 37.75% 54.31% 21.91%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 30.99% 31.76% 34.45% 42.03% 58.70% 23.13%
9. 5th 5th 50th 30.13% 30.89% 33.54% 41.05% 57.72% 23.35%
10. 5th 5th 95th 28.43% 29.17% 31.74% 39.08% 55.70% 23.79%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 11.26% 11.62% 12.93% 17.00% 28.65% 16.67%
12. 95th 95th 50th 10.86% 11.21% 12.48% 16.43% 27.83% 16.84%
13. 95th 95th 95th 10.09% 10.42% 11.61% 15.34% 26.21% 17.18%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 11.26% 11.62% 12.93% 17.00% 28.65% 16.67%
15. 5th 5th 95th 28.43% 29.17% 31.74% 39.08% 55.70% 23.79%
119
Panel B.
Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
prior excess future excess a function of percentile of cash/assets of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no cash-holding effect)
Neutral market-timing opportunities
1. 0.0% 0.0% 25.72% 26.20% 27.86% 32.55% 43.48% 21.41%
Effect of large variation in each market-timing variable
2. 90.0% 0.0% 17.96% 18.33% 19.63% 23.38% 32.73% 18.02%
3. -90.0% 0.0% 29.81% 30.34% 32.15% 37.19% 48.56% 23.07%
4. 0.0% 90.0% 24.24% 24.71% 26.30% 30.85% 41.56% 21.84%
5. 0.0% -90.0% 26.39% 26.88% 28.57% 33.32% 44.35% 21.22%
Future returns effect, given very low prior returns
6. -90.0% -90.0% 30.55% 31.08% 32.92% 38.01% 49.44% 22.87%
7. -90.0% 0.0% 29.81% 30.34% 32.15% 37.19% 48.56% 23.07%
8. -90.0% 90.0% 28.19% 28.70% 30.45% 35.37% 46.60% 23.52%
Future returns effect, given very high prior returns
9. 90.0% -90.0% 18.49% 18.86% 20.19% 24.02% 33.51% 17.85%
10. 90.0% 0.0% 17.96% 18.33% 19.63% 23.38% 32.73% 18.02%
11. 90.0% 90.0% 16.83% 17.18% 18.42% 22.00% 31.02% 18.40%
Extremely unfavorable versus favorable timing opportunities
12. 90.0% -90.0% 18.49% 18.86% 20.19% 24.02% 33.51% 17.85%
13. -90.0% 90.0% 28.19% 28.70% 30.45% 35.37% 46.60% 23.52%

120
 Table 4.9
Logit analysis of share repurchase decisions as a function of excess cash holdings

The dependent variable equals one if a firm repurchases at least 1% of its shares during the year in question or zero
otherwise. The independent variables are (i) the standardized market-to-book (M/B) ratio at the end of the fiscal year
prior to the year in question, (ii) the market-adjusted abnormal return over the 12 months (or over the 36 months in
rows C and F) ending immediately before the year in question, (iii) the market-adjusted excess return over the 12
months (or over the 36 months in rows C and F) beginning immediately after the year in question, (iv) operating
cash flow at the end of the fiscal year immediately before the year in question, (v) excess cash holdings at the end of
the fiscal year immediately before the year in question (vi) the change in options outstanding at the end of the year
in question, and (vii) the deviation from the target leverage ratio at the end of the fiscal year immediately before the
year in question. The standardized M/B ratio in a given year is the firm’s M/B ratio divided by the median M/B ratio
in that year for all firms. The abnormal return is the firm’s actual stock return minus the contemporaneous return on
the value-weighted market index. Operating cash flow is operating income before depreciation (OIBDP in
Compustat) deflated by total assets. Excess cash holdings is the residual from a liner regression of cash/assets on
variables often hypothesized to affect cash holding decisions as in Opler, Pinkowitz, Stulz and Williamson (1999).
Change in options outstanding is computed as the difference between options outstanding/common shares
outstanding in the current and prior year. Target leverage ratio is the fitted value from a liner regression of debt-to-
assets ratio on variables often hypothesized to affect leverage decisions: log(sales), market-to-book ratio,
profitability and asset tangibility. All models include firm size as proxied by the natural log of total assets. The
sample period is from 1996 to 2010.

Market to book Prior stock Future stock Operating Excess cash Option Deviation
Intercept ratio return return cash flow holding change from target
A. All firms
Coefficient -2.588 -0.041 -0.133 0.018
(Marginal probability) (0.007) (-0.022) (0.003)
[t-statistic] [-11.70] [-1.01] [-2.36] [0.53]
B. All firms with raw M/B in lieu of standardized M/B
Coefficient -2.618 -0.014 -0.143 0.018
(Marginal probability) (0.0024) (-0.024) (0.003)
[t-statistic] [-11.72] [-0.53] [-2.68] [0.55]
C. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -2.484 -0.038 -0.038 0.032
(Marginal probability) (-0.0068) (-0.007) (0.006)
[t-statistic] [-10.16] [-0.91] [-1.78] [1.95]
D. All firms
Coefficient -3.414 -0.264 -0.205 -0.044 6.745 1.512 4.657 -2.018
(Marginal probability) (-0.054) (-0.042) (-0.009) (1.374) (0.308) (0.949) (-0.411)
[t-statistic] [-11.64] [-4.82] [-3.24] [-1.08] [11.06] [6.32] [2.50] [-8.00]
E. All firms with raw M/B in lieu of standardized M/B
Coefficient -3.420 -0.131 -0.240 -0.046 6.432 1.422 4.406 -2.032
(Marginal probability) (-0.027) (-0.049) (-0.009) (1.309) (0.289) (0.896) (-0.413)
[t-statistic] [-11.61] [-2.79] [-4.07] [-1.15] [9.59] [5.94] [2.30] [-8.16]
F. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -3.582 -0.326 -0.078 0.016 7.644 1.558 3.889 -1.824
(Marginal probability) (-0.069) (-0.016) (0.003) (1.613) (0.329) (0.821) (-0.385)
[t-statistic] [-10.96] [-6.07] [-1.59] [0.87] [12.66] [5.75] [1.91] [-6.61]

121
 Table 4.10
Estimated probability of a share repurchase as a function of excess cash holdings

This table reports the probability of a repurchase conditional on specific hypothesized values of timing variables and excess cash holdings. All other independent
variables are set equal to their sample median values. A firm is defined as a repurchaser if it buys back at least 1% of its shares in the year in question. In Panel A,
the estimated probabilities are based on the model in Row D of Table 4.9. Row 1 reports the probability of a repurchase for a firm that faces neutral market-
timing opportunities. Rows 2-7 show the impact of changing each timing variable by large amounts, while holding other timing variables constant at their sample
median values. Rows 8-13 describe the impact of large swings in future abnormal returns while holding M/B ratios and prior excess stock returns constant at
values that represent highly favorable or unfavorable timing opportunities. The probability of a repurchase for firms that face highly unfavorable versus highly
favorable timing opportunities is given in Rows 14-15 respectively. The estimated probabilities in the far right column are based on the model in Row A of Table
4.9, which includes only timing variables and firm size. Panel B is similar to Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
standardized prior excess future excess a function of percentile of excess cash/assets of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no excess-cash-holding effect)
Neutral market-timing opportunities
1. 50th 50th 50th 24.84% 27.51% 29.79% 37.34% 54.23% 21.57%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 19.93% 22.23% 24.22% 30.98% 47.16% 18.63%
3. 50th 5th 50th 27.28% 30.11% 32.50% 40.34% 57.35% 22.99%
4. 50th 50th 95th 23.74% 26.33% 28.55% 35.95% 52.74% 21.99%
5. 50th 50th 5th 25.40% 28.11% 30.42% 38.04% 54.96% 21.36%
6. 95th 50th 50th 13.10% 14.75% 16.21% 21.37% 35.08% 19.57%
7. 5th 50th 50th 27.34% 30.17% 32.57% 40.41% 57.42% 21.91%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 30.54% 33.55% 36.08% 44.22% 61.19% 23.13%
9. 5th 5th 50th 29.92% 32.90% 35.40% 43.49% 60.48% 23.35%
10. 5th 5th 95th 28.67% 31.58% 34.04% 42.02% 59.03% 23.79%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 10.47% 11.84% 13.05% 17.41% 29.54% 16.67%
12. 95th 95th 50th 10.19% 11.53% 12.72% 16.99% 28.92% 16.84%
13. 95th 95th 95th 9.66% 10.93% 12.07% 16.16% 27.70% 17.18%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 10.47% 11.84% 13.05% 17.41% 29.54% 16.67%
15. 5th 5th 95th 28.67% 31.58% 34.04% 42.02% 59.03% 23.79%
122
Panel B.
Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
prior excess future excess a function of percentile of excess cash/assets of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no excess-cash-holding effect)
Neutral market-timing opportunities
1. 0.0% 0.0% 25.25% 27.20% 28.84% 34.18% 46.16% 21.41%
Effect of large variation in each market-timing variable
2. 90.0% 0.0% 17.24% 18.73% 20.00% 24.26% 34.59% 18.02%
3. -90.0% 0.0% 29.53% 31.68% 33.46% 39.19% 51.55% 23.07%
4. 0.0% 90.0% 24.16% 26.06% 27.66% 32.88% 44.72% 21.84%
5. 0.0% -90.0% 25.74% 27.72% 29.38% 34.77% 46.81% 21.22%
Future returns effect, given very low prior returns
6. -90.0% -90.0% 30.08% 32.24% 34.05% 39.81% 52.20% 22.87%
7. -90.0% 0.0% 29.53% 31.68% 33.46% 39.19% 51.55% 23.07%
8. -90.0% 90.0% 28.33% 30.43% 32.17% 37.80% 50.09% 23.52%
Future returns effect, given very high prior returns
9. 90.0% -90.0% 17.61% 19.13% 20.42% 24.74% 35.18% 17.85%
10. 90.0% 0.0% 17.24% 18.73% 20.00% 24.26% 34.59% 18.02%
11. 90.0% 90.0% 16.42% 17.85% 19.08% 23.20% 33.28% 18.40%
Extremely unfavorable versus favorable timing opportunities
12. 90.0% -90.0% 17.61% 19.13% 20.42% 24.74% 35.18% 17.85%
13. -90.0% 90.0% 28.33% 30.43% 32.17% 37.80% 50.09% 23.52%

123
 Table 4.11
The interaction effect between operating cash flow and market timing on share repurchase decisions

The dependent variable equals one if a firm repurchases at least 1% of its shares in the year in question or zero otherwise. The independent variables are (i) the
standardized market-to-book (M/B) ratio at the end of the fiscal year prior to the year in question, (ii) the market-adjusted abnormal return over the 12 months (or
over the 36 months in rows C and F) ending immediately before the year in question, (iii) the market-adjusted excess return over the 12 months (or over the 36
months in rows C and F) beginning immediately after the year in question, (iv) operating cash flow at the end of the fiscal year immediately before the year in
question, (v) the change in options outstanding at the end of the year in question, and (vi) the deviation from the target leverage ratio at the end of the fiscal year
immediately before the year in question. The standardized M/B ratio in a given year is the firm’s M/B ratio for the fiscal year immediately before the year in
question, divided by the median M/B ratio in that year for all firms. The abnormal return is the firm’s actual stock return minus the contemporaneous return on
the value-weighted market index. Operating cash flow is operating income before depreciation (OIBDP in Compustat) deflated by total assets. Change in options
outstanding is computed as the difference between options outstanding/common shares outstanding in the current and prior year. Target leverage ratio is the fitted
value from a liner regression of debt-to-assets ratio on variables often hypothesized to affect leverage decisions: log(sales), market-to-book ratio, profitability and
asset tangibility. All models include firm size as proxied by the natural log of total assets. The sample period is from 1996 to 2010. The interaction effects are
estimated as in Ai and Norton (2003).

Market to book Prior stock Future stock

ratio × return × return ×
Market to book Prior stock Future stock Operating Option Deviation Operating Operating Operating
ratio return return cash flow change from target cash flow cash flow cash flow
Model 1
Marginal probability -0.020 -0.052 -0.012 1.003 1.119 -0.277
[t-statistic] [-1.57] [-4.62] [-2.04] [9.12] [3.61] [-6.89]

Model 2
Marginal probability -0.015 -0.052 -0.012 1.042 1.118 -0.276 -0.044
[t-statistic] [-1.31] [-4.69] [-2.04] [8.64] [3.60] [-6.76] [-1.81]

Model 3
Marginal probability -0.020 -0.049 -0.012 1.003 1.119 -0.278 -0.089
[t-statistic] [-1.54] [-4.20] [-2.05] [9.07] [3.60] [-6.88] [-1.82]

Model 4
Marginal probability -0.020 -0.051 -0.005 1.009 1.119 -0.277 -0.076
[t-statistic] [-1.58] [-4.59] [-0.73] [9.10] [3.59] [-6.88] [-1.57]

124
 Table 5.1
Estimated probability of a share repurchase as a function of operating cash flow and the RRV and PS mispricing indices

This table reports the probability of a repurchase conditional on specific hypothesized values of the independent variables. A firm is defined as a repurchaser if it
buys back at least 1% of its shares in the year in question. The base case findings in Panel A are those implied by the parameter estimates in Row D of Table 4.1,
which includes the base timing variables and operating cash flow. The base timing variables are the standardized M/B ratio, and the prior and future 12-month
market-adjusted excess returns. Panel B through Panel F report the estimated repurchase probabilities calculated analogously, but with the model in Row D of
Table 4.1 re-estimated using other mispricing measures. The mispricing index labeled RRV is the one employed by Rhodes-Kropf, Robinson, and Viswanathan
(2005, Table 4, Model 3). The mispricing index labeled PS is the one used by Polk and Sapienza (2009). I take the 95th percentile value of RRV (PS) as
indicative of highly unfavorable timing opportunities and the 5th percentile value of RRV (PS) as indicative of highly favorable timing opportunities. For each
pair of rows, the first row reports the estimated repurchase probabilities as a function of operating cash flow for firms with poor timing opportunities while the
second row reports the estimated repurchase probabilities for firms with excellent timing opportunities.

Percentile of Percentile of Percentile of Percentile of Estimated probability of a repurchase as

Market mispricing indices standardized prior excess future excess supplementary a function of percentile of operating cash flow/assets
included in logit model M/B ratio stock return stock return mispricing index 5th 25th 50th 75th 95th
A. Base case 95th 95th 5th - 1.31% 11.67% 17.29% 22.03% 32.51%
5th 5th 95th - 2.62% 21.08% 29.70% 36.35% 49.33%
B. RRV index alone - - - 95th 2.16% 15.47% 21.82% 26.93% 37.59%
- - - 5th 3.13% 21.14% 29.03% 35.07% 46.88%
C. PS index alone - - - 95th 2.90% 18.75% 25.77% 31.24% 42.21%
- - - 5th 3.13% 20.02% 27.36% 33.01% 44.21%
D. Base and RRV index 95th 95th 5th 95th 1.23% 11.06% 16.45% 21.03% 31.24%
5th 5th 95th 5th 2.99% 23.51% 32.74% 39.70% 52.91%
E. Base and PS index 95th 95th 5th 95th 1.35% 12.15% 18.00% 22.92% 33.72%
5th 5th 95th 5th 2.77% 22.38% 31.39% 38.27% 51.47%
F. Base and RRV and PS indices 95th 95th 5th 95th 1.30% 11.71% 17.38% 22.18% 32.77%
5th 5th 95th 5th 3.14% 24.65% 34.16% 41.28% 54.59%

125
 Table 5.2
Estimated probability of a share repurchase as a function of free cash flow and the RRV and PS mispricing indices

This table reports the probability of a repurchase conditional on specific hypothesized values of the independent variables. A firm is defined as a repurchaser if it
buys back at least 1% of its shares in the year in question. The base case findings in Panel A are those implied by the parameter estimates in Row D of Table 4.3,
which includes the base timing variables and free cash flow. The base timing variables are the standardized M/B ratio, and the prior and future 12-month market-
adjusted excess returns. Panel B through Panel F report the estimated repurchase probabilities calculated analogously, but with the model in Row D of Table 4.3
re-estimated using other mispricing measures. The mispricing index labeled RRV is the one employed by Rhodes-Kropf, Robinson, and Viswanathan (2005,
Table 4, Model 3). The mispricing index labeled PS is the one used by Polk and Sapienza (2009). I take the 95th percentile value of RRV (PS) as indicative of
highly unfavorable timing opportunities and the 5th percentile value of RRV (PS) as indicative of highly favorable timing opportunities. For each pair of rows,
the first row reports the estimated repurchase probabilities as a function of operating cash flow for firms with poor timing opportunities while the second row
reports the estimated repurchase probabilities for firms with excellent timing opportunities.

Percentile of Percentile of Percentile of Percentile of Estimated probability of a repurchase as

Market mispricing indices standardized prior excess future excess supplementary a function of percentile of free cash flow/assets
included in logit model M/B ratio stock return stock return mispricing index 5th 25th 50th 75th 95th
A. Base case 95th 95th 5th - 0.86% 9.56% 15.28% 20.30% 31.60%
5th 5th 95th - 2.01% 19.95% 29.82% 37.51% 52.12%
B. RRV index alone - - - 95th 1.74% 14.15% 20.97% 26.54% 38.06%
- - - 5th 2.64% 20.11% 28.84% 35.56% 48.42%
C. PS index alone - - - 95th 2.46% 17.87% 25.64% 31.72% 43.71%
- - - 5th 2.55% 18.40% 26.32% 32.49% 44.58%
D. Base and RRV index 95th 95th 5th 95th 0.83% 9.18% 14.70% 19.58% 30.63%
5th 5th 95th 5th 2.27% 22.01% 32.48% 40.47% 55.21%
E. Base and PS index 95th 95th 5th 95th 0.86% 9.92% 15.94% 21.23% 33.06%
5th 5th 95th 5th 1.99% 20.53% 30.78% 38.73% 53.67%
F. Base and RRV and PS indices 95th 95th 5th 95th 0.85% 9.71% 15.59% 20.76% 32.37%
5th 5th 95th 5th 2.26% 22.44% 33.18% 41.34% 56.28%

126
 Table 5.3
Logit analysis of share repurchase decisions as a function of operating cash flow over 1971 to 2010

The dependent variable equals one if a firm repurchases at least 1% of its shares during the year in question or zero
otherwise. The independent variables are (i) the standardized market-to-book (M/B) ratio at the end of the fiscal year
prior to the year in question, (ii) the market-adjusted abnormal return over the 12 months (or over the 36 months in
rows C and F) ending immediately before the year in question, (iii) the market-adjusted excess return over the 12
months (or over the 36 months in rows C and F) beginning immediately after the year in question, and (iv) operating
cash flow at the end of the fiscal year immediately before the year in question. The standardized M/B ratio in a
given year is the firm’s M/B ratio divided by the median M/B ratio in that year for all firms. The abnormal return is
the firm’s actual stock return minus the contemporaneous return on the value-weighted market index. Operating
cash flow is operating income before depreciation (OIBDP in Compustat) deflated by total assets. All models
include firm size as proxied by the natural log of total assets. The sample period is from 1971 to 2010.

Market to book Prior stock Future stock Operating

Intercept ratio return return cash flow
A. All firms
Coefficient -2.381 -0.159 -0.084 0.066
(Marginal probability) (-0.022) (-0.012) (0.009)
[t-statistic] [-18.13] [-3.63] [-2.11] [2.64]

B. All firms with raw M/B in lieu of standardized M/B

Coefficient -2.511 -0.053 -0.120 0.069
(Marginal probability) (-0.007) (-0.016) (0.009)
[t-statistic] [-20.87] [-2.22] [-2.93] [2.73]

C. All firms with 36-month market-adjusted return in lieu of 12-month return

Coefficient -2.324 -0.137 -0.020 0.059
(Marginal probability) (-0.020) (-0.003) (0.008)
[t-statistic] [-17.50] [-3.15] [-1.08] [3.68]

D. All firms
Coefficient -2.489 -0.220 -0.198 0.076 2.674
(Marginal probability) (-0.031) (-0.028) (0.011) (0.380)
[t-statistic] [-19.19] [-4.46] [-4.55] [3.28] [13.54]

E. All firms with raw M/B in lieu of standardized M/B

Coefficient -2.653 -0.055 -0.247 0.080 2.408
(Marginal probability) (-0.008) (-0.034) (0.011) (0.337)
[t-statistic] [21.55] [-2.08] [-5.28] [3.40] [11.52]

F. All firms with 36-month market-adjusted return in lieu of 12-month return

Coefficient -2.479 -0.216 -0.076 0.058 2.884
(Marginal probability) (-0.032) (-0.011) (0.008) (0.425)
[t-statistic] [-18.26] [-4.13] [-2.56] [3.95] [11.78]

127
 Table 5.4
Estimated probability of a share repurchase as a function of operating cash flow over 1971 to 2010

This table reports the probability of a repurchase conditional on specific hypothesized values of timing variables and operating cash flow. All other independent
variables are set equal to their sample median values. A firm is defined as a repurchaser if it buys back at least 1% of its shares in the year in question. In Panel A,
the estimated probabilities are based on the model in Row D of Table 5.3. Row 1 reports the probability of a repurchase for a firm that faces neutral market-
timing opportunities. Rows 2-7 show the impact of changing each timing variable by large amounts, while holding other timing variables constant at their sample
median values. Rows 8-13 describe the impact of large swings in future abnormal returns while holding M/B ratios and prior excess stock returns constant at
values that represent highly favorable or unfavorable timing opportunities. The probability of a repurchase for firms that face highly unfavorable versus highly
favorable timing opportunities is given in Rows 14-15 respectively. The estimated probabilities in the far right column are based on the model in Row A of Table
5.3, which includes only timing variables and firm size. Panel B is similar to Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
standardized prior excess future excess a function of percentile of operating cash flow/assets of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no operating-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 50th 6.05% 14.57% 17.26% 19.74% 24.77% 17.02%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 4.90% 12.00% 14.29% 16.43% 20.84% 15.72%
3. 50th 5th 50th 6.74% 16.07% 18.97% 21.63% 26.98% 17.73%
4. 50th 50th 95th 6.57% 15.69% 18.53% 21.15% 26.42% 18.11%
5. 50th 50th 5th 5.79% 13.99% 16.58% 18.99% 23.89% 16.45%
6. 95th 50th 50th 3.47% 8.70% 10.43% 12.07% 15.53% 11.86%
7. 5th 50th 50th 6.63% 15.82% 18.69% 21.32% 26.62% 18.04%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 7.06% 16.74% 19.73% 22.47% 27.96% 18.15%
9. 5th 5th 50th 7.38% 17.42% 20.50% 23.32% 28.93% 18.78%
10. 5th 5th 95th 8.00% 18.70% 21.95% 24.90% 30.75% 19.95%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 2.67% 6.77% 8.15% 9.47% 12.29% 10.50%
12. 95th 95th 50th 2.80% 7.08% 8.52% 9.89% 12.81% 10.90%
13. 95th 95th 95th 3.04% 7.67% 9.22% 10.69% 13.82% 11.65%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 2.67% 6.77% 8.15% 9.47% 12.29% 10.50%
15. 5th 5th 95th 8.00% 18.70% 21.95% 24.90% 30.75% 19.95%
128
Panel B.
Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
prior excess future excess a function of percentile of operating cash flow/assets of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no operating-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 6.75% 14.49% 16.81% 18.92% 23.15% 16.59%
Effect of large variation in each market-timing variable
2. 95th 50th 5.01% 10.99% 12.83% 14.53% 17.99% 14.30%
3. 5th 50th 7.85% 16.63% 19.21% 21.55% 26.18% 17.88%
4. 50th 95th 7.36% 15.69% 18.15% 20.39% 24.85% 17.74%
5. 50th 5th 6.45% 13.90% 16.14% 18.19% 22.30% 16.02%
Future returns effect, given very low prior returns
6. 5th 5th 7.51% 15.98% 18.48% 20.75% 25.26% 17.28%
7. 5th 50th 7.85% 16.63% 19.21% 21.55% 26.18% 17.88%
8. 5th 95th 8.55% 17.97% 20.71% 23.17% 28.02% 19.10%
Future returns effect, given very high prior returns
9. 95th 5th 4.78% 10.53% 12.30% 13.94% 17.29% 13.79%
10. 95th 50th 5.01% 10.99% 12.83% 14.53% 17.99% 14.30%
11. 95th 95th 5.47% 11.94% 13.91% 15.73% 19.41% 15.32%
Extremely unfavorable versus favorable timing opportunities
12. 95th 5th 4.78% 10.53% 12.30% 13.94% 17.29% 13.79%
13. 5th 95th 8.55% 17.97% 20.71% 23.17% 28.02% 19.10%

129
 Table 5.5
Logit analysis of share repurchase decisions as a function of free cash flow over 1971 to 2010

The dependent variable equals one if a firm repurchases at least 1% of its shares during the year in question or zero
otherwise. The independent variables are (i) the standardized market-to-book (M/B) ratio at the end of the fiscal year
prior to the year in question, (ii) the market-adjusted abnormal return over the 12 months (or over the 36 months in
rows C and F) ending immediately before the year in question, (iii) the market-adjusted excess return over the 12
months (or over the 36 months in rows C and F) beginning immediately after the year in question, and (iv) free cash
flow at the end of the fiscal year immediately before the year in question. The standardized M/B ratio in a given year
is the firm’s M/B ratio divided by the median M/B ratio in that year for all firms. The abnormal return is the firm’s
actual stock return minus the contemporaneous return on the value-weighted market index. Free cash flow is the
ratio of operating income before depreciation (OIBDP in Compustat) to total assets minus the median industry
capital expenditure ratio. To compute the median industry capital expenditure ratio in each year, I first divide capital
expenditures (CAPX in Compustat) by total assets for each firm. Then I obtain the median value of all firms in the
same 2-digit SIC code industry excluding the firm in question. All models include firm size as proxied by the natural
log of total assets. The sample period is from 1971 to 2010.

Market to book Prior stock Future stock Free cash

Intercept ratio return return flow
A. All firms
Coefficient -2.381 -0.159 -0.084 0.066
(Marginal probability) (-0.022) (-0.012) (0.009)
[t-statistic] [-18.13] [-3.63] [-2.11] [2.64]

B. All firms with raw M/B in lieu of standardized M/B

Coefficient -2.511 -0.053 -0.120 0.069
(Marginal probability) (-0.007) (-0.016) (0.009)
[t-statistic] [-20.87] [-2.22] [-2.93] [2.73]

C. All firms with 36-month market-adjusted return in lieu of 12-month return

Coefficient -2.324 -0.137 -0.020 0.059
(Marginal probability) (-0.020) (-0.003) (0.008)
[t-statistic] [-17.50] [-3.15] [-1.08] [3.68]

D. All firms
Coefficient -2.335 -0.237 -0.210 0.075 2.957
(Marginal probability) (-0.034) (-0.030) (0.011) (0.421)
[t-statistic] [-18.97] [-4.86] [-4.84] [3.28] [14.21]

E. All firms with raw M/B in lieu of standardized M/B

Coefficient -2.514 -0.066 -0.257 0.079 2.668
(Marginal probability) (-0.009) (-0.036) (0.011) (0.374)
[t-statistic] [-21.41] [-2.58] [-5.52] [3.40] [12.77]

F. All firms with 36-month market-adjusted return in lieu of 12-month return

Coefficient -2.323 -0.236 -0.079 0.058 3.203
(Marginal probability) (-0.035) (-0.012) (0.009) (0.473)
[t-statistic] [-17.89] [-4.52] [-2.73] [4.07] [12.64]

130
 Table 5.6
Estimated probability of a share repurchase as a function of free cash flow over 1971 to 2010

This table reports the probability of a repurchase conditional on specific hypothesized values of timing variables and free cash flow. All other independent
variables are set equal to their sample median values. A firm is defined as a repurchaser if it buys back at least 1% of its shares in the year in question. In Panel A,
the estimated probabilities are based on the model in Row D of Table 5.5. Row 1 reports the probability of a repurchase for a firm that faces neutral market-
timing opportunities. Rows 2-7 show the impact of changing each timing variable by large amounts, while holding other timing variables constant at their sample
median values. Rows 8-13 describe the impact of large swings in future abnormal returns while holding M/B ratios and prior excess stock returns constant at
values that represent highly favorable or unfavorable timing opportunities. The probability of a repurchase for firms that face highly unfavorable versus highly
favorable timing opportunities is given in Rows 14-15 respectively. The estimated probabilities in the far right column are based on the model in Row A of Table
5.5, which includes only timing variables and firm size. Panel B is similar to Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
standardized prior excess future excess a function of percentile of free cash flow/assets of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no free-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 50th 5.49% 14.15% 17.28% 20.05% 25.72% 17.02%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 4.38% 11.50% 14.14% 16.51% 21.45% 15.72%
3. 50th 5th 50th 6.17% 15.71% 19.10% 22.09% 28.13% 17.73%
4. 50th 50th 95th 5.95% 15.22% 18.53% 21.45% 27.39% 18.11%
5. 50th 50th 5th 5.25% 13.59% 16.61% 19.31% 24.83% 16.45%
6. 95th 50th 50th 3.01% 8.09% 10.04% 11.81% 15.61% 11.86%
7. 5th 50th 50th 6.06% 15.47% 18.82% 21.78% 27.77% 18.04%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 6.51% 16.48% 20.00% 23.09% 29.30% 18.15%
9. 5th 5th 50th 6.80% 17.14% 20.76% 23.93% 30.29% 18.78%
10. 5th 5th 95th 7.36% 18.38% 22.20% 25.52% 32.12% 19.95%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 2.28% 6.21% 7.74% 9.15% 12.21% 10.50%
12. 95th 95th 50th 2.39% 6.49% 8.08% 9.55% 12.73% 10.90%
13. 95th 95th 95th 2.60% 7.03% 8.74% 10.32% 13.71% 11.65%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 2.28% 6.21% 7.74% 9.15% 12.21% 10.50%
15. 5th 5th 95th 7.36% 18.38% 22.20% 25.52% 32.12% 19.95%
131
Panel B.
Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
prior excess future excess a function of percentile of free cash flow/assets of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no free-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 6.28% 14.15% 16.82% 19.15% 23.83% 16.59%
Effect of large variation in each market-timing variable
2. 95th 50th 4.58% 10.56% 12.65% 14.50% 18.30% 14.30%
3. 5th 50th 7.38% 16.38% 19.37% 21.95% 27.09% 17.88%
4. 50th 95th 6.85% 15.32% 18.15% 20.62% 25.55% 17.74%
5. 50th 5th 6.01% 13.58% 16.16% 18.42% 22.97% 16.02%
Future returns effect, given very low prior returns
6. 5th 5th 7.06% 15.73% 18.63% 21.15% 26.16% 17.28%
7. 5th 50th 7.38% 16.38% 19.37% 21.95% 27.09% 17.88%
8. 5th 95th 8.04% 17.68% 20.85% 23.58% 28.96% 19.10%
Future returns effect, given very high prior returns
9. 95th 5th 4.38% 10.12% 12.13% 13.92% 17.60% 13.79%
10. 95th 50th 4.58% 10.56% 12.65% 14.50% 18.30% 14.30%
11. 95th 95th 5.00% 11.47% 13.71% 15.68% 19.73% 15.32%
Extremely unfavorable versus favorable timing opportunities
12. 95th 5th 4.38% 10.12% 12.13% 13.92% 17.60% 13.79%
13. 5th 95th 8.04% 17.68% 20.85% 23.58% 28.96% 19.10%

132
 Table 5.7
Estimated probability of a share repurchase as a function of the change in outstanding stock options

This table reports the probability of a repurchase conditional on specific hypothesized values of timing variables and the standardized change in options
outstanding. All other independent variables are set equal to their sample median values. A firm is defined as a repurchaser if it buys back at least 1% of its
shares in the year in question. In Panel A, the estimated probabilities are based on the model in Row D of Table 4.1. Row 1 reports the probability of a
repurchase for firms with neutral timing opportunities. Rows 2-7 show the impact of changing each timing variable by large amounts, while holding other timing
variables constant at their sample median values. Rows 8-13 describe the impact of large swings in future abnormal returns while holding M/B ratios and prior
excess stock returns constant at values that represent highly favorable or unfavorable timing opportunities. The probability of a repurchase for firms that face
highly unfavorable versus highly favorable timing opportunities is given in Rows 14-15 respectively. The estimated probabilities in the far right column are
based on the model in Row A of Table 4.1. Panel B is similar to Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a repurchase as a function Repurchase probability as a function
standardized prior excess future excess of percentile of change in options /# of shares of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no option change effect)
Neutral market-timing opportunities
1. 50th 50th 50th 25.01% 28.48% 29.72% 30.81% 34.00% 21.57%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 19.13% 22.02% 23.08% 24.01% 26.76% 18.63%
3. 50th 5th 50th 27.99% 31.70% 33.02% 34.17% 37.52% 22.99%
4. 50th 50th 95th 23.51% 26.84% 28.04% 29.10% 32.19% 21.99%
5. 50th 50th 5th 25.78% 29.31% 30.57% 31.68% 34.91% 21.36%
6. 95th 50th 50th 20.09% 23.09% 24.18% 25.14% 27.97% 19.57%
7. 5th 50th 50th 25.90% 29.44% 30.70% 31.81% 35.05% 21.91%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 29.78% 33.61% 34.97% 36.15% 39.58% 23.13%
9. 5th 5th 50th 28.94% 32.72% 34.06% 35.23% 38.62% 23.35%
10. 5th 5th 95th 27.29% 30.94% 32.25% 33.38% 36.70% 23.79%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 15.66% 18.15% 19.06% 19.87% 22.29% 16.67%
12. 95th 95th 50th 15.14% 17.56% 18.44% 19.23% 21.60% 16.84%
13. 95th 95th 95th 14.12% 16.40% 17.25% 18.00% 20.25% 17.18%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 15.66% 18.15% 19.06% 19.87% 22.29% 16.67%
15. 5th 5th 95th 27.29% 30.94% 32.25% 33.38% 36.70% 23.79%
133
Panel B.
Percentile of Percentile of Estimated probability of a repurchase as a function Repurchase probability as a function
prior excess future excess of percentile of change in options /# of shares of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no option change effect)
Neutral market-timing opportunities
1. 50th 50th 24.81% 28.14% 29.33% 30.38% 33.43% 21.41%
Effect of large variation in each market-timing variable
2. 95th 50th 17.92% 20.58% 21.55% 22.40% 24.94% 18.02%
3. 5th 50th 28.40% 32.01% 33.29% 34.41% 37.65% 23.07%
4. 50th 95th 23.34% 26.54% 27.69% 28.70% 31.67% 21.84%
5. 50th 5th 25.49% 28.87% 30.08% 31.14% 34.23% 21.22%
Future returns effect, given very low prior returns
6. 5th 5th 29.14% 32.79% 34.09% 35.22% 38.49% 22.87%
7. 5th 50th 28.40% 32.01% 33.29% 34.41% 37.65% 23.07%
8. 5th 95th 26.80% 30.29% 31.53% 32.62% 35.78% 23.52%
Future returns effect, given very high prior returns
9. 95th 5th 18.46% 21.17% 22.16% 23.03% 25.62% 17.85%
10. 95th 50th 17.92% 20.58% 21.55% 22.40% 24.94% 18.02%
11. 95th 95th 16.77% 19.30% 20.22% 21.04% 23.47% 18.40%
Extremely unfavorable versus favorable timing opportunities
12. 95th 5th 18.46% 21.17% 22.16% 23.03% 25.62% 17.85%
13. 5th 95th 26.80% 30.29% 31.53% 32.62% 35.78% 23.52%

134
 Table 5.8
Estimated probability of a share repurchase as a function of the deviation from target leverage

This table reports the probability of a repurchase conditional on specific hypothesized values of timing variables and the deviation from the target leverage ratio.
All other independent variables are set equal to their sample median values. A firm is defined as a repurchaser if it buys back at least 1% of its shares in the year
in question. In Panel A, the estimated probabilities are based on the model in Row D of Table 4.1. Row 1 reports the probability of a repurchase for a firm with
neutral market-timing opportunities. Rows 2-7 show the impact of changing each timing variable by large amounts, while holding other timing variables constant
at their sample median values. Rows 8-13 describe the impact of large swings in future abnormal returns while holding M/B ratios and prior excess stock returns
constant at values that represent highly favorable or unfavorable timing opportunities. The probability of a repurchase for firms that face highly unfavorable
versus highly favorable timing opportunities is given in Rows 14-15 respectively. The estimated probabilities in the far right column are based on the model in
Row A of Table 4.1. Panel B is similar to Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
standardized prior excess future excess a function of percentile of deviation from target of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no deviation from target effect)
Neutral market-timing opportunities
1. 50th 50th 50th 34.70% 32.22% 29.62% 25.58% 19.26% 21.57%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 27.38% 25.22% 22.99% 19.60% 14.47% 18.63%
3. 50th 5th 50th 38.25% 35.66% 32.91% 28.60% 21.75% 22.99%
4. 50th 50th 95th 32.87% 30.47% 27.94% 24.05% 18.02% 21.99%
5. 50th 50th 5th 35.62% 33.11% 30.46% 26.35% 19.89% 21.36%
6. 95th 50th 50th 28.60% 26.39% 24.08% 20.58% 15.24% 19.57%
7. 5th 50th 50th 35.76% 33.25% 30.59% 26.47% 19.99% 21.91%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 40.32% 37.67% 34.85% 30.41% 23.27% 23.13%
9. 5th 5th 50th 39.36% 36.73% 33.94% 29.56% 22.55% 23.35%
10. 5th 5th 95th 37.42% 34.86% 32.13% 27.89% 21.16% 23.79%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 22.83% 20.93% 18.98% 16.06% 11.72% 16.67%
12. 95th 95th 50th 22.13% 20.27% 18.37% 15.53% 11.31% 16.84%
13. 95th 95th 95th 20.75% 18.98% 17.17% 14.48% 10.52% 17.18%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 22.83% 20.93% 18.98% 16.06% 11.72% 16.67%
15. 5th 5th 95th 37.42% 34.86% 32.13% 27.89% 21.16% 23.79%
135
Panel B.
Percentile of Percentile of Estimated probability of a repurchase as Repurchase probability as a function
prior excess future excess a function of percentile of deviation from target of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no deviation from target effect)
Neutral market-timing opportunities
1. 50th 50th 34.24% 31.79% 29.22% 25.23% 19.01% 21.41%
Effect of large variation in each market-timing variable
2. 95th 50th 25.63% 23.57% 21.45% 18.26% 13.44% 18.02%
3. 5th 50th 38.50% 35.91% 33.17% 28.87% 22.00% 23.07%
4. 50th 95th 32.45% 30.07% 27.58% 23.75% 17.80% 21.84%
5. 50th 5th 35.05% 32.57% 29.96% 25.92% 19.56% 21.22%
Future returns effect, given very low prior returns
6. 5th 5th 39.35% 36.74% 33.97% 29.61% 22.63% 22.87%
7. 5th 50th 38.50% 35.91% 33.17% 28.87% 22.00% 23.07%
8. 5th 95th 36.61% 34.08% 31.41% 27.24% 20.66% 23.52%
Future returns effect, given very high prior returns
9. 95th 5th 26.31% 24.23% 22.06% 18.80% 13.86% 17.85%
10. 95th 50th 25.63% 23.57% 21.45% 18.26% 13.44% 18.02%
11. 95th 95th 24.12% 22.16% 20.13% 17.09% 12.53% 18.40%
Extremely unfavorable versus favorable timing opportunities
12. 95th 5th 26.31% 24.23% 22.06% 18.80% 13.86% 17.85%
13. 5th 95th 36.61% 34.08% 31.41% 27.24% 20.66% 23.52%

136
 Table 5.9
Logit analysis of tender-offer repurchase decisions as a function of operating cash flow

Tender-offer repurchase announcements are obtained from SDC. The dependent variable equals one if the firm
conducts a tender-offer repurchase in the year in question or zero otherwise. The independent variables are (i) the
standardized market-to-book (M/B) ratio at the end of the fiscal year prior to the year in question, (ii) the market-
adjusted abnormal return over the 12 months (or over the 36 months in rows C and F) ending immediately before the
year in question, (iii) the market-adjusted excess return over the 12 months (or over the 36 months in rows C and F)
beginning immediately after the year in question, (iv) operating cash flow at the end of the fiscal year immediately
before the year in question, (v) the change in options outstanding at the end of the year in question, and (vi) the
deviation from the target leverage ratio at the end of the fiscal year immediately before the year in question. The
standardized M/B ratio in a given year is the firm’s M/B ratio divided by the median M/B ratio in that year for all
firms. The abnormal return is the firm’s actual stock return minus the contemporaneous return on the value-weighted
market index. Operating cash flow is operating income before depreciation (OIBDP in Compustat) deflated by total
assets. Change in options outstanding is computed as the difference between options outstanding/common shares
outstanding in the current and prior year. Target leverage ratio is the fitted value from a liner regression of debt-to-
assets ratio on variables often hypothesized to affect leverage decisions: log(sales), market-to-book ratio,
profitability and asset tangibility. All models include firm size as proxied by the natural log of total assets. The
sample period is from 1996 to 2010.

Market to book Prior stock Future stock Operating Option Deviation

Intercept ratio return return cash flow change from target
A. All firms
Coefficient -5.412 -0.696 -0.077 0.002
(Marginal probability) (-0.004) (-0.000) (0.000)
[t-statistic] [-16.03] [-4.71] [-0.77] [0.03]
B. All firms with raw M/B in lieu of standardized M/B
Coefficient -5.537 -0.368 -0.109 -0.001
(Marginal probability) (-0.002) (-0.001) (-0.000)
[t-statistic] [-17.10] [-4.52] [-1.19] [-0.02]
C. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -5.195 -0.747 -0.068 -0.011
(Marginal probability) (-0.005) (-0.000) (-0.000)
[t-statistic] [-15.02] [-4.20] [-0.79] [-0.33]
D. All firms
Coefficient -4.993 -0.856 -0.144 -0.217 2.496 10.863 -1.399
(Marginal probability) (-0.007) (-0.001) (-0.002) (0.019) (0.083) (-0.011)
[t-statistic] [-10.85] [-4.50] [-1.03] [-1.62] [2.66] [6.43] [-2.80]
E. All firms with raw M/B in lieu of standardized M/B
Coefficient -5.187 -0.391 -0.210 -0.230 2.114 10.728 -1.399
(Marginal probability) (-0.003) (-0.002) (-0.002) (0.016) (0.082) (-0.011)
[t-statistic] [-10.68] [-3.46] [-1.61] [-1.63] [2.28] [6.39] [-2.84]
F. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -4.759 -1.052 -0.167 -0.207 4.268 12.391 -1.646
(Marginal probability) (-0.009) (-0.001) (-0.002) (0.035) (0.102) (-0.014)
[t-statistic] [-8.72] [-3.92] [-1.00] [-1.45] [4.00] [6.59] [-2.51]

137
 Table 5.10
Estimated probability of a tender-offer repurchase as a function of operating cash flow

This table reports the probability of a tender-offer repurchase conditional on specific hypothesized values of timing variables and operating cash flow. All other
independent variables are set equal to their sample median values. In Panel A, the estimated probabilities are based on the model in Row D of Table 5.9. Row 1
reports the probability of a repurchase for a firm that faces neutral market-timing opportunities. Rows 2-7 show the impact of changing each timing variable by
large amounts, while holding other timing variables constant at their sample median values. Rows 8-13 describe the impact of large swings in future abnormal
returns while holding M/B ratios and prior excess stock returns constant at values that represent highly favorable or unfavorable timing opportunities. The
probability of a repurchase for firms that face highly unfavorable versus highly favorable timing opportunities is given in Rows 14-15 respectively. The estimated
probabilities in the far right column are based on the model in Row A of Table 5.9, which includes only timing variables and firm size. Panel B is similar to Panel
A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a tender-offer repurchase as Tender-offer probability as a function
standardized prior excess future excess a function of percentile of operating cash flow/assets of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no operating-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 50th 0.19% 0.61% 0.77% 0.89% 1.18% 0.58%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 0.15% 0.50% 0.63% 0.73% 0.96% 0.52%
3. 50th 5th 50th 0.20% 0.66% 0.84% 0.98% 1.28% 0.60%
4. 50th 50th 95th 0.14% 0.45% 0.57% 0.67% 0.88% 0.58%
5. 50th 50th 5th 0.22% 0.70% 0.89% 1.03% 1.36% 0.57%
6. 95th 50th 50th 0.01% 0.05% 0.06% 0.07% 0.09% 0.07%
7. 5th 50th 50th 0.28% 0.92% 1.16% 1.35% 1.78% 0.81%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 0.36% 1.16% 1.47% 1.71% 2.24% 0.85%
9. 5th 5th 50th 0.31% 1.00% 1.27% 1.48% 1.94% 0.85%
10. 5th 5th 95th 0.23% 0.75% 0.94% 1.10% 1.45% 0.85%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 0.01% 0.05% 0.06% 0.07% 0.09% 0.07%
12. 95th 95th 50th 0.01% 0.04% 0.05% 0.06% 0.08% 0.07%
13. 95th 95th 95th 0.01% 0.03% 0.04% 0.04% 0.06% 0.07%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 0.01% 0.05% 0.06% 0.07% 0.09% 0.07%
15. 5th 5th 95th 0.23% 0.75% 0.94% 1.10% 1.45% 0.85%
138
Panel B.
Percentile of Percentile of Estimated probability of a tender-offer repurchase as Tender-offer probability as a function
prior excess future excess a function of percentile of operating cash flow/assets of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no operating-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 0.38% 0.65% 0.72% 0.77% 0.86% 0.54%
Effect of large variation in each market-timing variable
2. 95th 50th 0.20% 0.34% 0.38% 0.41% 0.46% 0.35%
3. 5th 50th 0.51% 0.85% 0.95% 1.01% 1.14% 0.66%
4. 50th 95th 0.29% 0.49% 0.54% 0.58% 0.65% 0.55%
5. 50th 5th 0.44% 0.73% 0.81% 0.87% 0.98% 0.54%
Future returns effect, given very low prior returns
6. 5th 5th 0.58% 0.97% 1.08% 1.15% 1.30% 0.66%
7. 5th 50th 0.51% 0.85% 0.95% 1.01% 1.14% 0.66%
8. 5th 95th 0.38% 0.64% 0.71% 0.76% 0.86% 0.67%
Future returns effect, given very high prior returns
9. 95th 5th 0.23% 0.39% 0.43% 0.46% 0.52% 0.34%
10. 95th 50th 0.20% 0.34% 0.38% 0.41% 0.46% 0.35%
11. 95th 95th 0.15% 0.26% 0.29% 0.31% 0.35% 0.35%
Extremely unfavorable versus favorable timing opportunities
12. 95th 5th 0.23% 0.39% 0.43% 0.46% 0.52% 0.34%
13. 5th 95th 0.38% 0.64% 0.71% 0.76% 0.86% 0.67%

139
 Table 5.11
Logit analysis of tender-offer repurchase decisions as a function of free cash flow

Tender-offer repurchase announcements are obtained from SDC. The dependent variable equals one if the firm
conducts a tender-offer repurchase in the year in question or zero otherwise. The independent variables are (i) the
standardized market-to-book (M/B) ratio at the end of the fiscal year prior to the year in question, (ii) the market-
adjusted abnormal return over the 12 months (or over the 36 months in rows C and F) ending immediately before the
year in question, (iii) the market-adjusted excess return over the 12 months (or over the 36 months in rows C and F)
beginning immediately after the year in question, (iv) free cash flow at the end of the fiscal year immediately before
the year in question, (v) the change in options outstanding at the end of the year in question, and (vi) the deviation
from the target leverage ratio at the end of the fiscal year immediately before the year in question. The standardized
M/B ratio in a given year is the firm’s M/B ratio divided by the median M/B ratio in that year for all firms. The
abnormal return is the firm’s actual stock return minus the contemporaneous return on the value-weighted market
index. Free cash flow is the ratio of operating income before depreciation (OIBDP in Compustat) to total assets
minus the median industry capital expenditure ratio. To compute the median industry capital expenditure ratio in
each year, I first divide capital expenditures (CAPX in Compustat) by total assets for each firm. Then I obtain the
median value of all firms in the same 2-digit SIC code industry excluding the firm in question. Change in options
outstanding is computed as the difference between options outstanding/common shares outstanding in the current
and prior year. Target leverage ratio is the fitted value from a liner regression of debt-to-assets ratio on variables
often hypothesized to affect leverage decisions: log(sales), market-to-book ratio, profitability and asset tangibility.
All models include firm size as proxied by the natural log of total assets. The sample period is from 1996 to 2010.

Market to book Prior stock Future stock Free cash Option Deviation
Intercept ratio return return flow change from target
A. All firms
Coefficient -5.412 -0.696 -0.077 0.002
(Marginal probability) (-0.004) (-0.000) (0.000)
[t-statistic] [-16.03] [-4.71] [-0.77] [0.03]
B. All firms with raw M/B in lieu of standardized M/B
Coefficient -5.537 -0.368 -0.109 -0.001
(Marginal probability) (-0.002) (-0.001) (-0.000)
[t-statistic] [-17.10] [-4.52] [-1.19] [-0.02]
C. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -5.195 -0.747 -0.068 -0.011
(Marginal probability) (-0.005) (-0.000) (-0.000)
[t-statistic] [-15.02] [-4.20] [-0.79] [-0.33]
D. All firms
Coefficient -4.931 -0.870 -0.141 -0.216 2.504 10.909 -1.470
(Marginal probability) (-0.007) (-0.001) (-0.002) (0.019) (0.084) (-0.011)
[t-statistic] [-10.64] [-4.74] [-1.01] [-1.62] [3.14] [6.74] [-2.85]
E. All firms with raw M/B in lieu of standardized M/B
Coefficient -5.134 -0.397 -0.207 -0.229 2.091 10.757 -1.456
(Marginal probability) (-0.003) (-0.002) (-0.002) (0.016) (0.083) (-0.011)
[t-statistic] [-10.41] [-3.62] [-1.61] [-1.64] [2.60] [6.65] [-2.87]
F. All firms with 36-month market-adjusted return in lieu of 12-month return
Coefficient -4.644 -1.047 -0.153 -0.199 3.923 12.509 -1.783
(Marginal probability) (-0.009) (-0.001) (-0.002) (0.033) (0.104) (-0.015)
[t-statistic] [-8.42] [-4.13] [-0.95] [-1.37] [4.08] [7.06] [-2.65]

140
 Table 5.12
Estimated probability of a tender-offer repurchase as a function of free cash flow

This table reports the probability of a tender-offer repurchase conditional on specific hypothesized values of timing variables and free cash flow. All other
independent variables are set equal to their sample median values. In Panel A, the estimated probabilities are based on the model in Row D of Table 5.11. Row 1
reports the probability of a repurchase for a firm that faces neutral market-timing opportunities. Rows 2-7 show the impact of changing each timing variable by
large amounts, while holding other timing variables constant at their sample median values. Rows 8-13 describe the impact of large swings in future abnormal
returns while holding M/B ratios and prior excess stock returns constant at values that represent highly favorable or unfavorable timing opportunities. The
probability of a repurchase for firms that face highly unfavorable versus highly favorable timing opportunities is given in Rows 14-15 respectively. The estimated
probabilities in the far right column are based on the model in Row A of Table 5.11, which includes only timing variables and firm size. Panel B is similar to
Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a tender-offer repurchase as Tender-offer probability as a function
standardized prior excess future excess a function of percentile of free cash flow/assets of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no free-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 50th 0.19% 0.60% 0.77% 0.90% 1.18% 0.58%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 0.16% 0.49% 0.63% 0.74% 0.97% 0.52%
3. 50th 5th 50th 0.21% 0.65% 0.84% 0.98% 1.29% 0.60%
4. 50th 50th 95th 0.14% 0.45% 0.57% 0.67% 0.88% 0.58%
5. 50th 50th 5th 0.22% 0.69% 0.89% 1.04% 1.36% 0.57%
6. 95th 50th 50th 0.01% 0.05% 0.06% 0.07% 0.09% 0.07%
7. 5th 50th 50th 0.29% 0.92% 1.17% 1.37% 1.80% 0.81%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 0.37% 1.15% 1.47% 1.72% 2.26% 0.85%
9. 5th 5th 50th 0.32% 1.00% 1.28% 1.49% 1.96% 0.85%
10. 5th 5th 95th 0.24% 0.75% 0.95% 1.11% 1.46% 0.85%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 0.01% 0.04% 0.06% 0.07% 0.09% 0.07%
12. 95th 95th 50th 0.01% 0.04% 0.05% 0.06% 0.07% 0.07%
13. 95th 95th 95th 0.01% 0.03% 0.04% 0.04% 0.06% 0.07%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 0.01% 0.04% 0.06% 0.07% 0.09% 0.07%
15. 5th 5th 95th 0.24% 0.75% 0.95% 1.11% 1.46% 0.85%
141
Panel B.
Percentile of Percentile of Estimated probability of a tender-offer repurchase as Tender-offer probability as a function
prior excess future excess a function of percentile of free cash flow/assets of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no free-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 0.41% 0.65% 0.72% 0.76% 0.85% 0.54%
Effect of large variation in each market-timing variable
2. 95th 50th 0.22% 0.35% 0.38% 0.41% 0.46% 0.35%
3. 5th 50th 0.54% 0.86% 0.95% 1.01% 1.13% 0.66%
4. 50th 95th 0.31% 0.49% 0.54% 0.58% 0.64% 0.55%
5. 50th 5th 0.46% 0.74% 0.81% 0.87% 0.97% 0.54%
Future returns effect, given very low prior returns
6. 5th 5th 0.61% 0.97% 1.07% 1.14% 1.27% 0.66%
7. 5th 50th 0.54% 0.86% 0.95% 1.01% 1.13% 0.66%
8. 5th 95th 0.41% 0.65% 0.71% 0.76% 0.85% 0.67%
Future returns effect, given very high prior returns
9. 95th 5th 0.25% 0.39% 0.43% 0.46% 0.52% 0.34%
10. 95th 50th 0.22% 0.35% 0.38% 0.41% 0.46% 0.35%
11. 95th 95th 0.16% 0.26% 0.29% 0.31% 0.34% 0.35%
Extremely unfavorable versus favorable timing opportunities
12. 95th 5th 0.25% 0.39% 0.43% 0.46% 0.52% 0.34%
13. 5th 95th 0.41% 0.65% 0.71% 0.76% 0.85% 0.67%

142
 Table 5.13
Estimated probability of a tender-offer repurchase as a function of the change in outstanding stock options

This table reports the probability of a repurchase conditional on specific hypothesized values of timing variables and the standardized change in options
outstanding. All other independent variables are set equal to their sample median values. In Panel A, the estimated probabilities are based on the model in Row D
of Table 5.9. Row 1 reports the probability of a repurchase for a firm that faces neutral market-timing opportunities. Rows 2-7 show the impact of changing each
timing variable by large amounts, while holding other timing variables constant at their sample median values. Rows 8-13 describe the impact of large swings in
future abnormal returns while holding M/B ratios and prior excess stock returns highly favorable or unfavorable values. The probability of a repurchase for firms
that face highly unfavorable versus highly favorable timing opportunities is given in Rows 14-15 respectively. The estimated probabilities in the far right column
are based on the model in Row A of Table 5.9, which includes only timing variables and firm size. Panel B is similar to Panel A, except that standardized M/B
ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a tender-offer repurchase as a Tender-offer probability as a function
standardized prior excess future excess function of percentile of change in options /# of shares of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no option change effect)
Neutral market-timing opportunities
1. 50th 50th 50th 0.50% 0.72% 0.81% 0.90% 1.20% 0.58%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 0.41% 0.59% 0.66% 0.74% 0.99% 0.52%
3. 50th 5th 50th 0.55% 0.78% 0.88% 0.98% 1.31% 0.60%
4. 50th 50th 95th 0.37% 0.53% 0.60% 0.67% 0.89% 0.58%
5. 50th 50th 5th 0.58% 0.83% 0.94% 1.04% 1.39% 0.57%
6. 95th 50th 50th 0.04% 0.06% 0.06% 0.07% 0.10% 0.07%
7. 5th 50th 50th 0.76% 1.09% 1.23% 1.36% 1.81% 0.81%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 0.96% 1.37% 1.55% 1.71% 2.28% 0.85%
9. 5th 5th 50th 0.83% 1.19% 1.34% 1.48% 1.98% 0.85%
10. 5th 5th 95th 0.62% 0.88% 1.00% 1.11% 1.48% 0.85%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 0.04% 0.05% 0.06% 0.07% 0.09% 0.07%
12. 95th 95th 50th 0.03% 0.05% 0.05% 0.06% 0.08% 0.07%
13. 95th 95th 95th 0.02% 0.03% 0.04% 0.04% 0.06% 0.07%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 0.04% 0.05% 0.06% 0.07% 0.09% 0.07%
15. 5th 5th 95th 0.62% 0.88% 1.00% 1.11% 1.48% 0.85%
143
Panel B.
Percentile of Percentile of Estimated probability of a tender-offer repurchase as a Tender-offer probability as a function
prior excess future excess function of percentile of change in options /# of shares of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no option change effect)
Neutral market-timing opportunities
1. 50th 50th 0.46% 0.65% 0.73% 0.81% 1.08% 0.54%
Effect of large variation in each market-timing variable
2. 95th 50th 0.24% 0.35% 0.39% 0.43% 0.57% 0.35%
3. 5th 50th 0.61% 0.86% 0.97% 1.07% 1.42% 0.66%
4. 50th 95th 0.35% 0.49% 0.55% 0.61% 0.81% 0.55%
5. 50th 5th 0.52% 0.74% 0.83% 0.92% 1.22% 0.54%
Future returns effect, given very low prior returns
6. 5th 5th 0.69% 0.98% 1.10% 1.22% 1.61% 0.66%
7. 5th 50th 0.61% 0.86% 0.97% 1.07% 1.42% 0.66%
8. 5th 95th 0.46% 0.65% 0.73% 0.81% 1.07% 0.67%
Future returns effect, given very high prior returns
9. 95th 5th 0.28% 0.39% 0.44% 0.49% 0.65% 0.34%
10. 95th 50th 0.24% 0.35% 0.39% 0.43% 0.57% 0.35%
11. 95th 95th 0.18% 0.26% 0.29% 0.32% 0.43% 0.35%
Extremely unfavorable versus favorable timing opportunities
12. 95th 5th 0.28% 0.39% 0.44% 0.49% 0.65% 0.34%
13. 5th 95th 0.46% 0.65% 0.73% 0.81% 1.07% 0.67%

144
 Table 5.14
Estimated probability of a tender-offer repurchase as a function of the deviation from target leverage

This table reports the probability of a repurchase conditional on specific hypothesized values of timing variables and the deviation from the target leverage ratio.
All other independent variables are set equal to their sample median values. In Panel A, the estimated probabilities are based on the model in Row D of Table 5.9.
Row 1 reports the probability of a repurchase for a firm that faces neutral market-timing opportunities. Rows 2-7 show the impact of changing each timing
variable by large amounts, while holding other timing variables constant at their sample median values. Rows 8-13 describe the impact of large swings in future
abnormal returns while holding M/B ratios and prior excess stock returns constant at values that represent highly favorable or unfavorable timing opportunities.
The probability of a repurchase for firms that face highly unfavorable versus highly favorable timing opportunities is given in Rows 14-15 respectively. The
estimated probabilities in the far right column are based on the model in Row A of Table 5.9, which includes only timing variables and firm size. Panel B is
similar to Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of a tender-offer repurchase as Tender-offer probability as a function
standardized prior excess future excess a function of percentile of deviation from target of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no deviation from target effect)
Neutral market-timing opportunities
1. 50th 50th 50th 1.03% 0.92% 0.81% 0.65% 0.45% 0.58%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 0.84% 0.75% 0.66% 0.54% 0.37% 0.52%
3. 50th 5th 50th 1.12% 1.00% 0.88% 0.71% 0.49% 0.60%
4. 50th 50th 95th 0.76% 0.68% 0.60% 0.49% 0.33% 0.58%
5. 50th 50th 5th 1.19% 1.06% 0.93% 0.76% 0.52% 0.57%
6. 95th 50th 50th 0.08% 0.07% 0.06% 0.05% 0.04% 0.07%
7. 5th 50th 50th 1.55% 1.38% 1.22% 0.99% 0.68% 0.81%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 1.96% 1.75% 1.54% 1.25% 0.86% 0.85%
9. 5th 5th 50th 1.70% 1.51% 1.33% 1.08% 0.74% 0.85%
10. 5th 5th 95th 1.26% 1.13% 0.99% 0.81% 0.55% 0.85%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 0.08% 0.07% 0.06% 0.05% 0.03% 0.07%
12. 95th 95th 50th 0.07% 0.06% 0.05% 0.04% 0.03% 0.07%
13. 95th 95th 95th 0.05% 0.04% 0.04% 0.03% 0.02% 0.07%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 0.08% 0.07% 0.06% 0.05% 0.03% 0.07%
15. 5th 5th 95th 1.26% 1.13% 0.99% 0.81% 0.55% 0.85%
145
Panel B.
Percentile of Percentile of Estimated probability of a tender-offer repurchase as Tender-offer probability as a function
prior excess future excess a function of percentile of deviation from target of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no deviation from target effect)
Neutral market-timing opportunities
1. 50th 50th 0.91% 0.82% 0.73% 0.60% 0.42% 0.54%
Effect of large variation in each market-timing variable
2. 95th 50th 0.48% 0.44% 0.39% 0.32% 0.23% 0.35%
3. 5th 50th 1.21% 1.08% 0.97% 0.80% 0.56% 0.66%
4. 50th 95th 0.69% 0.62% 0.55% 0.45% 0.32% 0.55%
5. 50th 5th 1.03% 0.93% 0.83% 0.68% 0.48% 0.54%
Future returns effect, given very low prior returns
6. 5th 5th 1.37% 1.23% 1.10% 0.90% 0.64% 0.66%
7. 5th 50th 1.21% 1.08% 0.97% 0.80% 0.56% 0.66%
8. 5th 95th 0.91% 0.82% 0.73% 0.60% 0.42% 0.67%
Future returns effect, given very high prior returns
9. 95th 5th 0.55% 0.49% 0.44% 0.36% 0.26% 0.34%
10. 95th 50th 0.48% 0.44% 0.39% 0.32% 0.23% 0.35%
11. 95th 95th 0.36% 0.33% 0.29% 0.24% 0.17% 0.35%
Extremely unfavorable versus favorable timing opportunities
12. 95th 5th 0.55% 0.49% 0.44% 0.36% 0.26% 0.34%
13. 5th 95th 0.91% 0.82% 0.73% 0.60% 0.42% 0.67%

146
 Table 5.15
Leverage and traditional leverage determinants surrounding tender-offer repurchases

This table presents the mean values of leverage, target leverage, the deviation from target leverage, and various
other financial variables surrounding tender-offer repurchases for 539 tender-offer repurchases that are announced
between 1985 and 2010 and have required data before and after the year of tender-offer repurchases. The deviation
from target is the difference between the debt-to-assets ratio and an estimated target leverage ratio. Target leverage 1
is the fitted value from a liner regression of debt-to-assets ratio on variables often hypothesized to affect leverage
decisions: log(sales), market-to-book ratio, profitability and asset tangibility. Target leverage 2 (3) is the median
debt-to-assets ratio in 2-digit SIC (Fama-French 49) industries. Asset growth equals assets in year t minus assets in
year t-1, all divided by assets in year t-1. The same divisor is applied to the year t change in debt, capital
expenditures and EBITDA. For tangible assets in year t, I divide by assets in year t. Numbers in the parentheses in
Panel are cumulative percent. I use *** to identify a significant difference at the 0.01 level or better between the t=-
1 mean value of a variable and its t=1 value. The variables in row 6 to 14 show no significant differences at the 0.10
level.

Event year relative to tender-offer repurchases

Mean value of -3 -2 -1 0 1 2 3
1. Debt/Total Assets 0.238 0.226 0.215 0.257 0.266*** 0.271 0.276
2. Leverage deviation 1 -0.025 -0.034 -0.044 0.003 0.010*** 0.013 0.014
3. Leverage deviation 2 0.032 0.020 0.009 0.050 0.061*** 0.065 0.067
4. Leverage deviation 3 0.026 0.014 0.002 0.043 0.051*** 0.055 0.059
5. Change in debt 0.043 0.030 0.009 0.050 0.045*** 0.031 0.029

6. Target leverage 1 0.260 0.260 0.258 0.259 0.258 0.256 0.262

7. Target leverage 2 0.206 0.206 0.207 0.208 0.206 0.206 0.210
8. Target leverage 3 0.212 0.211 0.213 0.214 0.215 0.216 0.218
9. Asset growth 0.154 0.176 0.078 -0.001 0.068 0.079 0.083
10. Capital expenditures 0.083 0.073 0.068 0.067 0.071 0.067 0.071
11. EBITDA 0.148 0.152 0.141 0.136 0.140 0.118 0.157
12. Log(sales) 6.055 6.102 6.105 6.086 6.156 6.263 6.364
13. Market-to-book 1.578 1.518 1.425 1.485 1.866 1.713 1.590
14. Tangible assets 0.333 0.321 0.316 0.315 0.319 0.321 0.326

147
 Table 5.16
Logit analysis of dividend decisions as a function of operating cash flow

The dependent variable equals one if the firm pays cash dividends in the year in question or zero otherwise. The
independent variables are (i) the standardized market-to-book (M/B) ratio at the end of the fiscal year prior to the
year in question, (ii) the market-adjusted abnormal return over the 12 months (or over the 36 months in rows C and
F) ending immediately before the year in question, (iii) the market-adjusted excess return over the 12 months (or
over the 36 months in rows C and F) beginning immediately after the year in question, and (iv) operating cash flow
at the end of the fiscal year immediately before the year in question. The standardized M/B ratio in a given year is
the firm’s M/B ratio divided by the median M/B ratio in that year for all firms. The abnormal return is the firm’s
actual stock return minus the contemporaneous return on the value-weighted market index. Operating cash flow is
operating income before depreciation (OIBDP in Compustat) deflated by total assets. All models include firm size as
proxied by the natural log of total assets. The sample period is from 1996 to 2010.

Market to book Prior stock Future stock Operating

Intercept ratio return return cash flow
A. All firms
Coefficient -3.960 -0.159 0.003 -0.144
(Marginal probability) (-0.028) (0.001) (-0.025)
[t-statistic] [-29.20] [-4.23] [0.04] [-1.63]

B. All firms with raw M/B in lieu of standardized M/B

Coefficient -3.984 -0.090 -0.007 -0.145
(Marginal probability) (-0.016) (-0.001) (-0.025)
[t-statistic] [-29.39] [-3.74] [-0.11] [-1.64]

C. All firms with 36-month market-adjusted return in lieu of 12-month return

Coefficient -3.659 -0.159 -0.038 -0.083
(Marginal probability) (-0.032) (-0.008) (-0.017)
[t-statistic] [-24.53] [-3.89] [-1.18] [-1.73]

D. All firms
Coefficient -4.169 -0.443 -0.066 -0.157 7.165
(Marginal probability) (-0.076) (-0.011) (-0.027) (1.228)
[t-statistic] [-26.55] [-8.84] [-1.38] [-2.39] [19.77]

E. All firms with raw M/B in lieu of standardized M/B

Coefficient -4.174 -0.270 -0.080 -0.164 7.079
(Marginal probability) (-0.046) (-0.014) (-0.028) (1.214)
[t-statistic] [-27.09] [-8.77] [-1.76] [-2.46] [19.33]

F. All firms with 36-month market-adjusted return in lieu of 12-month return

Coefficient -3.974 -0.443 -0.096 -0.113 7.470
(Marginal probability) (-0.090) (-0.019) (-0.023) (1.510)
[t-statistic] [-24.17] [-7.87] [-2.85] [-2.74] [17.55]

148
 Table 5.17
Estimated probability that a firm pays dividends as a function of operating cash flow

This table reports the probability that a firm pays dividends conditional on specific hypothesized values of timing variables and operating cash flow. All other
independent variables are set equal to their sample median values. In Panel A, the estimated probabilities are based on the model in Row D of Table 5.16, which
includes timing variables, operating cash flow, and other control variables. Row 1 reports the probability of a repurchase for a firm that faces neutral market-
timing opportunities. Rows 2-7 show the impact of changing each timing variable by large amounts, while holding other timing variables constant at their sample
median values. Rows 8-13 describe the impact of large swings in future abnormal returns while holding M/B ratios and prior excess stock returns constant at
values that represent highly favorable or unfavorable timing opportunities. The probability of a repurchase for firms that face highly unfavorable versus highly
favorable timing opportunities is given in Rows 14-15 respectively. The estimated probabilities in the far right column are based on the model in Row A of Table
5.16, which includes only timing variables and firm size. Panel B is similar to Panel A, except that standardized M/B ratio is not used to estimate the probabilities.

Panel A.
Percentile of Percentile of Percentile of Estimated probability of dividends as Dividend probability as a function
standardized prior excess future excess a function of percentile of operating cash flow/assets of market-timing variables only
M/B ratio stock return stock return 5th 25th 50th 75th 95th (no operating-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 50th 0.45% 12.01% 21.23% 29.65% 48.17% 22.33%
Effect of large variation in each market-timing variable
2. 50th 95th 50th 0.41% 11.08% 19.74% 27.78% 45.90% 22.40%
3. 50th 5th 50th 0.47% 12.45% 21.92% 30.51% 49.19% 22.30%
4. 50th 50th 95th 0.36% 9.91% 17.84% 25.36% 42.83% 19.10%
5. 50th 50th 5th 0.50% 13.18% 23.06% 31.92% 50.83% 24.06%
6. 95th 50th 50th 0.12% 3.53% 6.74% 10.15% 19.95% 15.20%
7. 5th 50th 50th 0.56% 14.50% 25.08% 34.36% 53.58% 23.70%
Future returns effect, given very low M/B and prior returns
8. 5th 5th 5th 0.65% 16.42% 27.95% 37.76% 57.22% 25.47%
9. 5th 5th 50th 0.58% 15.01% 25.85% 35.29% 54.59% 23.67%
10. 5th 5th 95th 0.47% 12.46% 21.94% 30.53% 49.21% 20.30%
Future returns effect, given very high M/B and prior returns
11. 95th 95th 5th 0.12% 3.58% 6.84% 10.29% 20.19% 16.56%
12. 95th 95th 50th 0.11% 3.23% 6.19% 9.35% 18.53% 15.26%
13. 95th 95th 95th 0.09% 2.62% 5.05% 7.68% 15.49% 12.88%
Extremely unfavorable versus favorable timing opportunities
14. 95th 95th 5th 0.12% 3.58% 6.84% 10.29% 20.19% 16.56%
15. 5th 5th 95th 0.47% 12.46% 21.94% 30.53% 49.21% 20.30%

149
Panel B.
Percentile of Percentile of Estimated probability of dividends as Dividends probability as a function
prior excess future excess a function of percentile of operating cash flow/assets of market-timing variables only
stock return stock return 5th 25th 50th 75th 95th (no operating-cash-flow effect)
Neutral market-timing opportunities
1. 50th 50th 1.07% 13.54% 21.06% 27.48% 41.32% 21.67%
Effect of large variation in each market-timing variable
2. 95th 50th 0.80% 10.41% 16.53% 21.94% 34.32% 20.00%
3. 5th 50th 1.22% 15.17% 23.36% 30.21% 44.58% 22.44%
4. 50th 95th 0.88% 11.37% 17.94% 23.69% 36.59% 18.64%
5. 50th 5th 1.17% 14.61% 22.58% 29.28% 43.49% 23.13%
Future returns effect, given very low prior returns
6. 5th 5th 1.34% 16.35% 24.99% 32.11% 46.79% 23.94%
7. 5th 50th 1.22% 15.17% 23.36% 30.21% 44.58% 22.44%
8. 5th 95th 1.01% 12.78% 19.99% 26.18% 39.73% 19.33%
Future returns effect, given very high prior returns
9. 95th 5th 0.87% 11.27% 17.79% 23.50% 36.35% 21.39%
10. 95th 50th 0.80% 10.41% 16.53% 21.94% 34.32% 20.00%
11. 95th 95th 0.66% 8.69% 13.96% 18.72% 29.98% 17.15%
Extremely unfavorable versus favorable timing opportunities
12. 95th 5th 0.87% 11.27% 17.79% 23.50% 36.35% 21.39%
13. 5th 95th 1.01% 12.78% 19.99% 26.18% 39.73% 19.33%

150
 Table 5.18
Frequencies of share repurchases

This table documents the percent of firms with an annual frequency of repurchases that falls in the given intervals.
The sample contains 13,186 industrial firms in the CRSP/Compustat merged files over 1983 to 2010 that (i) have
SIC codes outside the intervals 4900-4949 (utilities) and 6000-6999 (financials), (ii) have a CRSP share code of 10
or 11, (iii) are incorporated in the US, and (iv) have non-missing total assets from Compustat. Repurchase is net
repurchase as in Fama and French (2001), measured as the increase in common treasury stock amounts. If the
treasury stock amount is zero in the current and previous year or if the treasury stock amount is not available,
repurchase is measured as the difference between the stock purchase and stock issuance amounts from Compustat.
For each firm, I compute the frequency of repurchases as the ratio of the number of repurchasing years to total
number of years listed in the sample. The repurchasing year is a year in which a firm’s repurchase amount is positive.
Panel A reports results for the whole sample from 1983 to 2010. Panel B deals with the first half of the sample
period while Panel C deals with the second half of the sample period.

Percent of firms with

frequency of repurchases in the interval:
Firms listed Firms listed
at least 10 years 5 to 9 years All firms
Panel A: 1983 -2010
(1) Always repurchase 0.1% 0.6% 1.3%
(2) Repurchase at least 3 out of 4 years but not always 4.3% 3.1% 2.8%
(3) Repurchase at least 2 but less than 3 out of 4 years 17.5% 10.0% 12.5%
(4) Repurchase at least 1 but less than 2 out of 4 years 31.7% 18.2% 20.1%
(5) Repurchase less than 1 out of 4 years 31.4% 23.8% 17.7%
(6) Never repurchase 15.0% 44.2% 45.6%
Number of firms 4,783 3,504 13,186

Panel B: 1983 -1996

(1) Always repurchase 0.5% 0.8% 1.6%
(2) Repurchase at least 3 out of 4 years but not always 4.7% 2.6% 2.3%
(3) Repurchase at least 2 but less than 3 out of 4 years 17.9% 9.0% 11.0%
(4) Repurchase at least 1 but less than 2 out of 4 years 26.6% 16.4% 14.9%
(5) Repurchase less than 1 out of 4 years 32.8% 25.6% 14.9%
(6) Never repurchase 17.6% 45.5% 55.2%
Number of firms 2,649 2,538 10,157

Panel C: 1997-2010
(1) Always repurchase 1.7% 1.5% 4.4%
(2) Repurchase at least 3 out of 4 years but not always 9.7% 5.3% 5.2%
(3) Repurchase at least 2 but less than 3 out of 4 years 24.6% 14.1% 16.3%
(4) Repurchase at least 1 but less than 2 out of 4 years 26.4% 19.4% 17.6%
(5) Repurchase less than 1 out of 4 years 23.8% 19.5% 12.3%
(6) Never repurchase 13.9% 40.2% 44.3%
Number of firms 2,627 1,978 8,255

151
 Table 5.19
Firm age and the number of years in which a firm repurchases shares

This table documents the relationship between the number of years listed in the sample (firm age) and the number of years in which a firm repurchases shares.
The sample contains 14,958 industrial firms in the CRSP/Compustat merged files over 1971 to 2010 that (i) have SIC codes outside the intervals 4900-4949
(utilities) and 6000-6999 (financials), (ii) have a CRSP share code of 10 or 11, (iii) are incorporated in the US, and (iv) have non-missing total assets from
Compustat. The repurchasing year is the year in which a firm’s repurchase amount is positive. Repurchase amount is net repurchase as in Fama and French
(2001), measured as the increase in common treasury stock amounts. If the treasury stock amount is zero in the current and previous year or if the treasury stock
amount is not available, repurchase amount is measured as the difference between the stock purchase and stock issuance amount from Compustat. Percent of
years with repurchases is defined as the number of repurchasing years divided by the number of years listed in the sample. Constant composition sample
contatins 240 firms that are inclued in the sample in 1971 and remain until 2010.

Median % of years Mean % of years Percent of firms with the number of repurchasing years that falls in the interval: Number
Years in Sample with repurchases with repurchases 0 1 2 to 4 5 to 9 10 to 14 15 and plus of firms
20 and plus 29.2% 31.3% 3.8% 5.2% 17.1% 30.6% 22.2% 21.0% 2,060
15 to 19 22.2% 26.4% 13.9% 11.3% 32.0% 32.7% 9.4% 0.6% 1,598
10 to 14 18.2% 23.1% 25.8% 16.2% 34.1% 21.7% 2.3% --- 2,489
5 to 9 12.5% 18.3% 46.2% 21.3% 27.9% 4.7% --- --- 4,091
2 to 4 0.0% 11.0% 75.5% 17.0% 7.5% --- --- --- 3,950
1 0.0% 3.1% 96.9% 3.1% --- --- --- --- 770

Constant comp sample 40.0% 40.8% 0.8% 0.4% 4.6% 11.7% 22.5% 60.0% 240

Full Sample 11.1% 19.0% 43.8% 15.1% 21.0% 12.6% 4.5% 2.9% 14,958

152
 Table 6.1
Histogram of abnormal stock rates of return following share repurchases

All firms in this table come from the same sample as in Table 3.1. The future 12-month (36-month) cumulative
abnormal return is the market-adjusted buy-and-hold abnormal return over the 12 months (36 months) beginning
immediately after the year in question. The abnormal return is the firm’s actual return minus the contemporaneous
return on the value-weighted market index. One-shot repurchasers are firms that only repurchase once during their
time in the sample. Tender-offer repurchases are reported by SDC. Numbers in the parentheses are cumulative
percent. The sample period in Panel A is 1985 to 2009 while the sample period in Panel B is 1985 to 2007.

Panel A. Future 12-month abnormal returns

Percent (cumulative percent) of repurchases with future
12-month abnormal stock return in the interval:
All One-shot Tender-offer
Below -0.750 1.8% (1.8%) 9.3% (9.3%) 1.1% (1.1%)
-0.750 to -0.500 6.8% (8.5%) 15.7% (25.0%) 4.3% (5.5%)
-0.500 to -0.250 17.4% (25.9%) 16.7% (41.7%) 18.5% (24.0%)
-0.250 to 0.000 28.1% (54.0%) 16.7% (58.4%) 28.9% (52.8%)
0.000 to 0.250 22.2% (76.2%) 13.9% (72.3%) 24.2% (77.0%)
0.250 to 0.500 11.1% (87.3%) 9.8% (82.2%) 11.3% (88.3%)
0.500 to 0.750 5.3% (92.5%) 4.9% (87.1%) 5.1% (93.4%)
Above 0.750 7.5% (100.0%) 12.9% (100.0%) 6.6% (100.0%)
Median abnormal stock return -0.03 -0.14 -0.02
Mean abnormal stock return 0.07 0.06 0.06

Panel B. Future 36-month abnormal returns

Percent (cumulative percent) of repurchases with future
36-month abnormal stock return in the interval:
All One-shot Tender-offer
Below -0.750 6.9% (6.9%) 25.3% (25.3%) 6.8% (6.8%)
-0.750 to -0.500 12.9% (19.7%) 19.2% (44.5%) 13.1% (20.0%)
-0.500 to -0.250 17.6% (37.3%) 13.2% (57.7%) 18.7% (38.7%)
-0.250 to 0.000 17.6% (54.9%) 10.7% (68.4%) 16.3% (55.0%)
0.000 to 0.250 14.1% (69.0%) 5.2% (73.6%) 16.8% (71.8%)
0.250 to 0.500 9.7% (78.7%) 5.5% (79.1%) 10.2% (82.0%)
0.500 to 0.750 6.0% (84.6%) 2.5% (81.6%) 4.1% (86.1%)
Above 0.750 15.4% (100.0%) 18.4% (100.0%) 13.9% (100.0%)
Median abnormal stock return -0.07 -0.41 -0.08
Mean abnormal stock return 0.18 0.00 0.13

153
 Table 6.2
Histogram of abnormal dollar stock returns following share repurchases

All firms in this table are from the same sample as in Table 3.1. The future 12-month (36-month) abnormal dollar
stock return is the product of the dollar amount of shares repurchased and the future 12-month (36-month)
cumulative abnormal return. Repurchase is net repurchase as in Fama and French (2001). The future 12-month (36-
month) cumulative abnormal return is the market-adjusted buy-and-hold abnormal return over the 12 months (36
months) beginning immediately after the year in question. The abnormal return is the firm’s actual return minus the
contemporaneous return on the value-weighted market index. One-shot repurchasers are firms that only repurchase
once during their time in the sample. Tender-offer repurchases and the corresponding dollar amounts of repurchase
(if available) are reported by SDC. Numbers in the parentheses are cumulative percent. Total abnormal dollar stock
return is the sum of the abnormal dollar stock return of every repurchase. Average abnormal dollar stock return is
total abnormal dollar stock return divided by the number of repurchases. Total dollar repurchase is the sum of net
repurchase of every repurchase. The sample period in Panel A is 1985 to 2009 while the sample period in Panel B is
1985 to 2007.

Panel A. Future 12-month abnormal dollar returns

Percent (cumulative percent) of repurchases with future
12-month abnormal dollar stock return in the interval:
All One-shot Tender-offer
Below -$200m 0.9% (0.9%) 0.5% (0.5%) 3.0% (3.0%)
-$200m to -$100m 0.8% (1.8%) 0.2% (0.7%) 3.2% (6.2%)
-$100m to -$50m 1.4% (3.2%) 0.5% (1.2%) 5.3% (11.5%)
-$50m to $0m 50.8% (53.9%) 57.2% (58.4%) 36.8% (48.3%)
$0m to $50m 42.3% (96.2%) 40.4% (98.8%) 41.7% (90.0%)
$50m to $100m 1.7% (97.9%) 0.7% (99.5%) 4.2% (94.2%)
$100m to $ 200m 1.0% (98.9%) 0.4% (99.9%) 2.5% (96.6%)
Above $200m 1.1% (100.0%) 0.1% (100.0%) 3.4% (100.0%)
Total abnormal dollar stock return ($b) 55.2 -1.7 1.7
Average abnormal dollar stock return ($m) 2.1 -1.7 3.2
Median abnormal dollar stock return ($m) -0.0 -0.0 0.0
Total dollar repurchases ($b) 2,687.3 13.9 156.7
Number of repurchases 26,775 976 530

Panel B. Future 36-month abnormal dollar returns

Percent (cumulative percent) of repurchases with future
36-month abnormal dollar stock return in the interval:
All One-shot Tender-offer
Below -$200m 1.5% (1.5%) 0.0% (0.0%) 4.1% (4.1%)
-$200m to -$100m 1.3% (2.7%) 0.5% (0.5%) 4.6% (8.8%)
-$100m to -$50m 1.7% (4.4%) 1.1% (1.6%) 4.1% (12.9%)
-$50m to $0m 50.4% (54.8%) 66.8% (68.4%) 38.0% (50.9%)
$0m to $50m 39.1% (93.9%) 30.2% (98.6%) 35.5% (86.4%)
$50m to $100m 2.2% (96.1%) 0.5% (99.2%) 4.6% (91.0%)
$100m to $ 200m 1.8% (98.0%) 0.5% (99.7%) 2.9% (93.9%)
Above $200m 2.0% (100.0%) 0.3% (100.0%) 6.1% (100.0%)
Total abnormal dollar stock return ($b) 118.5 -0.0 3.4
Average abnormal dollar stock return ($m) 5.8 -0.0 8.3
Median abnormal dollar stock return ($m) -0.0 -0.0 -0.0
Total dollar repurchases ($b) 2,136.1 4.4 126.9
Number of repurchases 20,327 364 411

154
 Table 6.3
Biggest winners and losers of share repurchases based on 12-month abnormal returns

All firms in this table are from the same sample as in Table 3.1. Dollar repurchase is net repurchase as in Fama and
French (2001). Abnormal return is the market-adjusted buy-and-hold abnormal return over the 12 months beginning
immediately after the year in question. Percent of dollar repurchase to market value is the dollar repurchase amount
divided by the market capitalization at the end of the prior year. Abnormal dollar return is the product of the dollar
repurchase amount and the abnormal return. Percent of dollar abnormal return to market value is the abnormal dollar
return divided by the market capitalization at the end of the prior year. The sample period is from 1985 to 2009.

Percent of $ repurchase
Panel A. Winners based on abnormal rates of return Year $ Repurchase Abnormal return to market value
1. MEHL/BIOPHILE INTL CORP 1994 $0.23 38.909 5.5%
2. EXX INC -CL A 2002 $0.20 15.334 3.9%
3. CATO CORP -CL A 1990 $0.05 14.419 0.1%
4. SILICON STORAGE TECHNOLOGY 1998 $1.01 13.571 1.4%
5. ELCOM INTERNATIONAL INC 1998 $0.63 13.468 0.3%
6. PERFECTDATA CORP 1998 $0.01 13.301 0.2%
7. VALUEVISION MEDIA INC -CL A 2008 $3.32 12.067 1.6%
8. AMERICA'S CAR-MART INC 1992 $0.08 11.697 4.7%
9. RC2 CORP 2000 $3.99 11.411 5.7%
10. USANA HEALTH SCIENCES INC 2001 $0.03 10.767 0.2%
11. MOVIE GALLERY INC 2000 $5.70 10.591 10.5%
12. ION NETWORKS INC -OLD 1998 $0.20 10.558 1.5%
13. LIBERTY MEDIA CAPITAL GROUP 2008 $462.00 10.266 3.1%
14. SCO GROUP INC 2002 $4.00 9.965 23.4%
15. NBTY INC 1991 $0.02 9.810 0.3%
16. BIOTIME INC 1995 $0.19 9.537 2.3%
17. ARRHYTHMIA RESEARCH TECH 2002 $0.73 8.890 10.1%
18. FORWARD INDUSTRIES INC 2004 $0.08 8.013 0.6%
19. ANAREN INC 1999 $1.47 7.867 1.8%
20. KEITHLEY INSTRUMENTS INC 1999 $7.64 7.703 19.2%
Percent of $ repurchase
Panel B. Losers based on abnormal rates of return Year $ Repurchase Abnormal return to market value
1. COLOROCS CORP 1990 $16.87 -0.998 13.8%
2. MOLTEN METAL TECHNOLOGY INC 1996 $0.48 -0.996 0.1%
3. PROGRESSIVE GAMING INTL CORP 2007 $0.13 -0.987 0.0%
4. PURCHASEPRO.COM 2001 $0.90 -0.987 0.1%
5. VESTRON INC 1989 $0.12 -0.980 0.1%
6. BRENTWOOD INSTRUMENTS INC 1988 $0.01 -0.979 0.7%
7. VITALSTREAM HOLDINGS INC 1999 $1.44 -0.978 32.1%
8. ANACOMP INC 1999 $4.16 -0.977 2.2%
9. DEX ONE CORP 2007 $94.86 -0.976 2.1%
10. ONCOR INC 1997 $0.00 -0.974 0.0%
11. CONSTAR INTERNATIONAL INC 2007 $0.24 -0.968 0.3%
12. CARMIKE CINEMAS INC 1999 $0.44 -0.967 0.2%
13. TECHNOLOGY DEVELOPMENT CORP 1990 $0.05 -0.964 1.0%
14. UNIROYAL TECHNOLOGY CORP 2001 $11.24 -0.963 2.9%
15. SABRATEK CORP 1998 $19.34 -0.961 6.5%
16. CAREMATRIX CORP 1999 $3.00 -0.960 0.5%
17. COMDISCO HOLDING CO INC 2000 $81.00 -0.958 2.7%
18. AMERICAN COMMUNITY NEWSPAPER 2007 $8.13 -0.951 8.9%
19. BLUE CHIP COMPUTERWARE INC 1994 $0.25 -0.949 2.0%
20. LEASING SOLUTIONS INC 1998 $1.24 -0.949 0.6%

155
 Percent of $ abnormal
Panel C. Winners based on abnormal $ returns Year $ Repurchase $ Abnormal return return to market value
1. EXXON MOBIL CORP 2007 $30,291 $10,163 2.3%
2. INTL BUSINESS MACHINES CORP 2007 $17,649 $4,932 3.4%
3. LIBERTY MEDIA CAPITAL GROUP 2008 $462 $4,743 31.5%
4. EXXON MOBIL CORP 2006 $28,040 $4,380 1.3%
5. AMGEN INC 2007 $4,823 $4,322 5.4%
6. ALTRIA GROUP INC 1999 $3,185 $3,973 3.1%
7. WAL-MART STORES INC 2007 $7,691 $3,825 1.9%
8. INTL BUSINESS MACHINES CORP 2000 $5,975 $3,819 2.0%
9. ORACLE CORP 1998 $781 $3,531 15.4%
10. EXXON MOBIL CORP 2005 $17,133 $3,458 1.1%
11. PFIZER INC 2007 $10,107 $2,921 1.6%
12. CISCO SYSTEMS INC 2006 $6,613 $2,492 2.1%
13. COMCAST CORP 2002 $7,517 $2,373 7.0%
14. BOEING CO 1999 $3,206 $2,338 7.6%
15. MCDONALD'S CORP 2007 $3,210 $2,192 4.1%
16. DELL INC 1997 $935 $2,078 18.2%
17. ANHEUSER-BUSCH COS INC 2007 $2,707 $2,040 5.4%
18. INTEL CORP 2002 $3,333 $1,904 0.9%
19. ORACLE CORP 2001 $2,460 $1,875 2.2%
20. CISCO SYSTEMS INC 2008 $7,324 $1,834 1.0%
Percent of $ abnormal
Panel D. Losers based on abnormal $ returns Year $ Repurchase $ Abnormal return return to market value
1. EXXON MOBIL CORP 2008 $34,420 -$12,492 -2.5%
2. TIME WARNER INC 2006 $19,140 -$5,481 -6.9%
3. PFIZER INC 2003 $13,011 -$4,157 -2.2%
4. GENERAL ELECTRIC CO 2007 $12,003 -$3,034 -0.8%
5. INTEL CORP 2005 $9,435 -$2,761 -1.9%
6. PROCTER & GAMBLE CO 2006 $34,235 -$2,601 -2.0%
7. HEWLETT-PACKARD CO 2000 $4,822 -$2,296 -3.1%
8. VALERO ENERGY CORP 2007 $4,701 -$2,146 -6.9%
9. ORACLE CORP 1999 $4,365 -$2,115 -6.0%
10. DELL INC 2005 $7,249 -$1,992 -1.9%
11. CONOCOPHILLIPS 2008 $8,213 -$1,812 -1.3%
12. CISCO SYSTEMS INC 2004 $7,823 -$1,786 -1.3%
13. DELL INC 2004 $4,219 -$1,641 -1.9%
14. AVIS BUDGET GROUP INC 1999 $2,821 -$1,626 -10.1%
15. ORACLE CORP 2000 $3,805 -$1,566 -0.8%
16. HOME DEPOT INC 2006 $6,671 -$1,564 -1.8%
17. CHEVRON CORP 2008 $7,484 -$1,490 -0.8%
18. INTL BUSINESS MACHINES CORP 2004 $7,038 -$1,486 -0.9%
19. MICROSOFT CORP 2009 $8,774 -$1,352 -0.5%
20. EXXON MOBIL CORP 2009 $18,312 -$1,346 -0.3%

156
 Table 6.4
Biggest winners and losers of share repurchases based on 36-month abnormal returns

All firms in this table are from the same sample as in Table 3.1. Dollar repurchase is net repurchase as in Fama and
French (2001). Abnormal return is the market-adjusted buy-and-hold abnormal return over the 36 months beginning
immediately after the year in question. Percent of dollar repurchase to market value is the dollar repurchase amount
divided by the market capitalization at the end of the prior year. Abnormal dollar return is the product of the dollar
repurchase amount and the abnormal return. Percent of dollar abnormal return to market value is the abnormal dollar
return divided by the market capitalization at the end of the prior year. The sample period is from 1985 to 2007.

Percent of $ repurchase
Panel A. Winners based on abnormal rates of return Year $ Repurchase Abnormal return to market value
1. USANA HEALTH SCIENCES INC 2001 $0.03 44.570 0.2%
2. USANA HEALTH SCIENCES INC 2000 $2.92 39.158 5.7%
3. IMMUCOR INC 2000 $1.49 24.990 2.5%
4. HANSEN NATURAL CORP 2002 $0.00 23.003 0.0%
5. INNODATA ISOGEN INC 1997 $0.04 22.585 0.6%
6. NBTY INC 1991 $0.02 21.542 0.3%
7. LCA VISION INC 2002 $2.45 19.864 6.0%
8. DYNACQ HEALTHCARE INC 1998 $0.57 19.828 5.8%
9. ARRHYTHMIA RESEARCH TECH 2000 $0.50 19.656 9.1%
10. RC2 CORP 2000 $3.99 18.547 5.7%
11. CATO CORP -CL A 1990 $0.05 18.437 0.1%
12. BRADLEY PHARMACEUTICL -CL A 1999 $0.55 18.041 5.6%
13. CHRISTOPHER & BANKS CORP 1998 $3.00 17.759 6.1%
14. STRATASYS INC 2000 $0.73 16.942 1.8%
15. BRADLEY PHARMACEUTICL -CL A 1998 $0.33 16.628 2.0%
16. EXX INC -CL A 2001 $0.50 16.478 7.2%
17. NBTY INC 1990 $0.37 16.275 7.3%
18. HECLA MINING CO 2000 $0.01 16.155 0.0%
19. SPINNAKER INDS INC -CL A 1992 $0.00 15.392 0.0%
20. HI TECH PHARMACAL CO INC 1999 $0.26 15.328 1.5%
Percent of $ repurchase
Panel B. Losers based on abnormal rates of return Year $ Repurchase Abnormal return to market value
1. PROGRESSIVE GAMING INTL CORP 2005 $0.43 -0.997 0.2%
2. BERKEY INC 1986 $0.64 -0.996 1.9%
3. IFLI ACQUISITION CORP-OLD 1997 $0.01 -0.996 0.1%
4. CORAUTUS GENETICS INC 2005 $0.16 -0.996 0.2%
5. ASSISTED LIVING CONCEPTS INC 1998 $6.32 -0.995 2.0%
6. IDT CORP 2007 $20.19 -0.994 1.6%
7. MSGI SECURITY SOLUTIONS INC 1999 $1.26 -0.993 2.8%
8. TIGERLOGIC CORP 1994 $1.56 -0.993 3.1%
9. DAMSON OIL 1985 $3.77 -0.993 8.0%
10. PARAGON TRADE BRANDS INC 1996 $8.64 -0.991 3.2%
11. YRC WORLDWIDE INC 2007 $35.00 -0.990 1.6%
12. HORIZON RESOURCES CORP 1990 $0.24 -0.989 0.8%
13. PRIME HOSPITALITY CORP 1989 $0.75 -0.989 0.1%
14. DOMINION HOMES INC 2004 $0.03 -0.989 0.0%
15. APPLIED MAGNETICS CORP 1997 $0.36 -0.989 0.1%
16. FLYI INC 2002 $0.28 -0.989 0.0%
17. ALTERNATIVE RESOURCES CORP 1997 $0.42 -0.988 0.2%
18. SYQUEST TECHNOLOGY INC 1995 $1.20 -0.988 1.1%
19. WORLDWIDE XCEED GROUP INC 1998 $0.02 -0.988 0.1%
20. FIRST MEDICAL GROUP INC 1987 $1.16 -0.987 1.7%

157
 Percent of $ abnormal
Panel C. Winners based on abnormal $ returns Year $ Repurchase $ Abnormal return return to market value
1. EXXON MOBIL CORP 2005 $17,133 $14,663 4.5%
2. INTL BUSINESS MACHINES CORP 2007 $17,649 $7,523 5.1%
3. ALTRIA GROUP INC 1999 $3,185 $6,627 5.1%
4. MICROSOFT CORP 2006 $17,106 $6,042 2.3%
5. PROCTER & GAMBLE CO 2006 $34,235 $5,800 4.4%
6. INTL BUSINESS MACHINES CORP 2006 $7,750 $4,418 3.4%
7. EXXON MOBIL CORP 2004 $8,853 $4,003 1.5%
8. CISCO SYSTEMS INC 2006 $6,613 $3,061 2.5%
9. INTL BUSINESS MACHINES CORP 2005 $7,474 $2,905 1.8%
10. AUTOZONE INC 2000 $608 $2,634 76.6%
11. HEWLETT-PACKARD CO 2006 $5,241 $2,602 3.3%
12. HEWLETT-PACKARD CO 2004 $2,739 $2,423 3.6%
13. EXXON MOBIL CORP 2003 $5,284 $2,196 0.9%
14. CHEVRON CORP 2005 $2,746 $2,148 1.9%
15. HOME DEPOT INC 2007 $10,705 $2,113 2.6%
16. MCDONALD'S CORP 2006 $3,179 $2,043 4.8%
17. GENERAL ELECTRIC CO 1996 $3,132 $2,042 1.7%
18. DELL INC 1996 $438 $2,002 78.3%
19. ALTRIA GROUP INC 2002 $5,688 $1,957 2.0%
20. GENERAL ELECTRIC CO 1997 $3,960 $1,945 1.2%
Percent of $ abnormal
Panel D. Losers based on abnormal $ returns Year $ Repurchase $ Abnormal return return to market value
1. EXXON MOBIL CORP 2007 $30,291 -$6,412 -1.5%
2. PFIZER INC 2003 $13,011 -$5,723 -3.0%
3. TIME WARNER INC 2006 $19,140 -$4,686 -5.9%
4. GENERAL ELECTRIC CO 2007 $12,003 -$4,193 -1.1%
5. DELL INC 2005 $7,249 -$3,649 -3.5%
6. GENERAL ELECTRIC CO 2006 $7,567 -$3,192 -0.9%
7. VALERO ENERGY CORP 2007 $4,701 -$2,916 -9.4%
8. DELL INC 2004 $4,219 -$2,621 -3.1%
9. PFIZER INC 2002 $4,963 -$2,515 -1.0%
10. CBS CORP 2007 $3,388 -$2,510 -10.5%
11. ORACLE CORP 1999 $4,365 -$2,032 -5.7%
12. DU PONT (E I) DE NEMOURS 1995 $10,434 -$1,975 -5.2%
13. PFIZER INC 2004 $6,640 -$1,961 -0.7%
14. CISCO SYSTEMS INC 2003 $5,406 -$1,942 -2.0%
15. WAL-MART STORES INC 2003 $5,046 -$1,811 -0.9%
16. MERCK & CO 2001 $3,529 -$1,717 -0.8%
17. AMGEN INC 2004 $3,619 -$1,664 -2.1%
18. INTEL CORP 2005 $9,435 -$1,648 -1.1%
19. INTEL CORP 2003 $3,045 -$1,625 -1.6%
20. GENERAL ELECTRIC CO 2005 $4,564 -$1,516 -0.4%

158
 Table 6.5
Distribution of abnormal rates of return and abnormal dollar returns of share repurchases

All firms in this table are from the same sample as in Table 3.1. The future 12-month (36-month) cumulative abnormal return is the market-adjusted buy-and-
hold abnormal return over the 12 months (36 months) beginning immediately after the year in question. The abnormal return is the firm’s actual return minus the
contemporaneous return on the value-weighted market index. The future 12-month (36-month) abnormal dollar stock returns is the product of the dollar amount
of share repurchased and the future 12-month (36-month) cumulative abnormal return. Repurchase is net repurchase as in Fama and French (2001). One-shot
repurchasers are firms that only repurchase once during their time in the sample. Tender-offer repurchases and the corresponding dollar amounts of repurchase (if
available) are reported by SDC. Percent negative is the percent of returns that are negative. The sample period in Panel A and C is from 1985 to 2009 while the
sample period in Panel B and D is from 1985 to 2007.

Mean Median Percent Min 1st 10th 25th 75th 90th 99th Max N
negative percentile percentile percentile percentile percentile percentile
Panel A. 12-month abnormal stock return
All 0.07 -0.03 54.0% -1.00 -0.81 -0.47 -0.26 0.23 0.61 2.36 38.91 26,775
One Shot 0.06 -0.14 58.4% -1.00 -0.94 -0.74 -0.50 0.30 0.92 4.27 10.56 976
Tender-offer 0.06 -0.02 52.8% -0.93 -0.78 -0.41 -0.24 0.23 0.56 2.33 7.70 530

Panel B. 36-month abnormal stock return

All 0.18 -0.07 54.9% -1.00 -0.95 -0.68 -0.42 0.39 1.14 4.58 44.57 20,327
One Shot 0.00 -0.41 68.4% -1.00 -0.99 -0.92 -0.75 0.31 1.51 5.48 7.99 364
Tender offer 0.13 -0.08 55.0% -0.95 -0.91 -0.65 -0.44 0.30 1.14 4.05 6.73 411

Panel C. 12-month abnormal dollar stock return ($m)

All 2.1 -0.0 53.9% -12,491.9 -183.4 -7.4 -0.6 0.6 9.1 221.8 10,163.5 26,775
One Shot -1.7 -0.0 58.4% -890.5 -78.8 -2.4 -0.2 0.1 2.1 58.2 246.1 976
Tender-offer 3.2 0.0 48.3% -890.5 -388.0 -56.5 -8.7 6.3 50.4 576.0 1,790.5 530

Panel D. 36-month abnormal dollar stock return ($m)

All 5.8 -0.0 54.8% -6,412.3 -301.2 -12.5 -1.0 1.3 20.0 399.3 14,662.8 20,327
One Shot -0.0 -0.0 68.4% -147.0 -72.0 -3.4 -0.5 0.0 1.2 83.2 209.6 364
Tender offer 8.3 -0.0 50.9% -1,302.3 -553.7 -77.9 -15.5 4.9 82.5 634.0 2,112.8 411

159
 Table 6.6
Distribution of future abnormal rates of return and dollar returns for large gain, loss, and other repurchases

All firms in this table are from the same sample as in Table 3.1. Large gain (loss) repurchases are those with future 12-month (36-month) abnormal returns (or
dollar return) belong to the top (bottom) decile of the full sample. Middle 80% repurchases are those in which the firm has future 12-month (36-month) abnormal
returns (or dollar return) in the middle 80% of the whole sample. The future 12-month (36-month) abnormal stock return is the market-adjusted buy-and-hold
abnormal return over the 12 months (36 months) beginning immediately after the year in question. The abnormal return is the firm’s actual return minus the
contemporaneous return on the value-weighted market index. The future 12-month (36-month) abnormal dollar stock return is the product of the dollar amount of
shares repurchased and the future 12-month (36-month) abnormal return. Dollar repurchase is net repurchase as in Fama and French (2001). Dollar amount
repurchased is the total dollar repurchase amount of all repurchases in the specified group. Percent of aggregate value repurchased is the dollar amount
repurchased divided by the sum of dollar amount repurchased in all three groups. The sample period in Panel A and C is from 1985 to 2009 while the sample
period in Panel B and D is from 1985 to 2007.

Mean Median Min 1st 10th 25th 75th 90th 99th Max N $ amount % of aggregate
percentile percentile percentile percentile percentile percentile repurchased value repurchased
Panel A. 12-month abnormal stock return
Large gain 1.38 0.97 0.61 0.61 0.66 0.75 1.44 2.36 7.16 38.91 2,677 92.4 3.5%
Middle 80% -0.01 -0.03 -0.47 -0.45 -0.34 -0.21 0.16 0.35 0.57 0.61 21,421 2484.4 94.2%
Large loss -0.63 -0.60 -1.00 -0.94 -0.81 -0.71 -0.52 -0.49 -0.47 -0.47 2,677 61.4 2.3%

Panel B. 36-month abnormal stock return

Large gain 2.65 1.84 1.14 1.15 1.23 1.40 2.90 4.58 15.25 44.57 2,033 57.7 2.8%
Middle 80% -0.01 -0.08 -0.68 -0.66 -0.53 -0.35 0.27 0.63 1.07 1.14 16,262 1991.5 94.9%
Large loss -0.81 -0.80 -1.00 -0.99 -0.95 -0.89 -0.73 -0.70 -0.68 -0.68 2,032 48.4 2.3%

Panel C. 12-month abnormal dollar stock return ($m)

Large gain 115.0 33.1 9.1 9.3 11.6 16.4 83.6 221.8 1,518.7 10,163.5 2,677 1181.7 44.0%
Middle 80% 0.1 0.0 -7.4 -6.2 -1.8 -0.3 0.2 2.1 7.7 9.1 21,421 226.0 8.4%
Large loss -95.0 -26.7 -12,491.9 -1,097.8 -183.4 -68.1 -13.0 -9.1 -7.5 -7.4 2,677 1279.6 47.6%

Panel D. 36-month abnormal dollar stock return ($m)

Large gain 196.7 65.6 20.0 20.3 24.7 35.2 168.0 399.3 1,890.0 14,662.8 2,033 924.5 43.3%
Middle 80% 0.4 0.0 -12.5 -10.7 -2.9 -0.5 0.5 4.5 16.9 20.0 16,262 183.8 8.6%
Large loss -141.9 -41.4 -6,412.3 -1,494.2 -301.2 -110.1 -21.4 -15.5 -12.8 -12.5 2,032 1027.8 48.1%

160
 Table 6.7
Distribution of future abnormal rates of return and dollar returns for middle repurchases

All firms in this table are from the same sample as in Table 3.1. Middle repurchases are those in which the firm has future 12-month (36-month) abnormal returns
(or dollar return) in the middle 80% (90%, 95%, and 99%) of the whole sample. The future 12-month (36-month) abnormal stock return is the market-adjusted
buy-and-hold abnormal return over the 12 months (36 months) beginning immediately after the year in question. The abnormal return is the firm’s actual return
minus the contemporaneous return on the value-weighted market index. The future 12-month (36-month) abnormal dollar return is the product of the dollar
amount of shares repurchased and the future 12-month (36-month) abnormal return. Dollar repurchase is net repurchase as in Fama and French (2001). Dollar
amount repurchased is the total dollar repurchase amount of all repurchases in the specified group. Percent of aggregate value repurchased is the dollar amount
repurchased divided by the sum of the dollar amount of shares repurchased by all firms. The sample period in Panel A and C is from 1985 to 2009 while the
sample period in Panel B and D is from 1985 to 2007.

Mean Median Min 1st 10th 25th 75th 90th 99th Max N $ amount % of aggregate
percentile percentile percentile percentile percentile percentile repurchased value repurchased
Panel A. 12-month abnormal stock return
Middle 80% -0.012 -0.035 -0.470 -0.454 -0.341 -0.209 0.165 0.355 0.572 0.607 21,421 2528.7 94.1%
Middle 90% 0.002 -0.035 -0.597 -0.568 -0.397 -0.234 0.198 0.459 0.877 0.975 24,099 2642.4 98.3%
Middle 95% 0.016 -0.035 -0.707 -0.656 -0.430 -0.247 0.214 0.526 1.219 1.442 25,437 2668.9 99.3%
Middle 99% 0.042 -0.035 -0.863 -0.769 -0.462 -0.256 0.228 0.590 1.963 3.473 26,509 2682.8 99.8%

Panel B. 36-month abnormal stock return

Middle 80% -0.008 -0.075 -0.678 -0.663 -0.526 -0.348 0.272 0.630 1.070 1.136 16,262 2026.3 94.9%
Middle 90% 0.032 -0.075 -0.802 -0.778 -0.600 -0.384 0.328 0.829 1.652 1.843 18,295 2096.9 98.2%
Middle 95% 0.068 -0.075 -0.890 -0.855 -0.638 -0.401 0.356 0.971 2.377 2.900 19,311 2121.0 99.3%
Middle 99% 0.128 -0.075 -0.966 -0.928 -0.670 -0.415 0.383 1.105 3.798 6.464 20,125 2134.2 99.9%

Panel C. 12-month abnormal dollar stock return ($m)

Middle 80% 0.1 0.0 -7.4 -6.2 -1.8 -0.3 0.2 2.1 7.7 9.1 21,421 226.0 8.4%
Middle 90% 0.3 0.0 -26.7 -20.4 -3.5 -0.4 0.4 4.2 25.0 33.1 24,099 516.8 19.2%
Middle 95% 0.5 0.0 -68.1 -44.7 -5.0 -0.5 0.5 6.2 58.0 83.6 25,437 862.7 32.1%
Middle 99% 1.0 0.0 -357.1 -127.3 -6.8 -0.6 0.6 8.4 141.6 436.8 26,509 1669.2 62.1%

Panel D. 36-month abnormal dollar stock return ($m)

Middle 80% 0.4 0.0 -12.5 -10.7 -2.9 -0.5 0.5 4.5 16.9 20.0 16,262 183.8 8.6%
Middle 90% 1.2 0.0 -41.4 -32.4 -5.8 -0.7 0.8 9.2 51.8 65.6 18,295 427.0 20.0%
Middle 95% 2.1 0.0 -109.7 -72.0 -8.4 -0.9 1.0 13.3 115.4 168.0 19,310 701.9 32.9%
Middle 99% 3.8 0.0 -589.5 -192.1 -11.6 -1.0 1.2 18.4 281.8 746.8 20,125 1380.8 64.6%

161
 Table 6.8
Distribution of future abnormal rates of return and dollar returns for middle repurchases for the tender-offer subsample

Tender-offer repurchases and the corresponding dollar amounts of repurchase (if available) are reported by SDC. Middle repurchases are those in which the firm
has future 12-month (36-month) abnormal returns (or dollar return) in the middle 80% (90%, 95%, and 99%) of the whole sample. The future 12-month (36-
month) abnormal stock return is the market-adjusted buy-and-hold abnormal return over the 12 months (36 months) beginning immediately after the year in
question. The abnormal return is the firm’s actual return minus the contemporaneous return on the value-weighted market index. The future 12-month (36-month)
abnormal dollar return is the product of the dollar amount of shares repurchased and the future 12-month (36-month) abnormal return. Dollar amount repurchased
is the total dollar repurchase amount of all repurchases in the specified group. Percent of aggregate value repurchased is the dollar amount repurchased divided
by the sum of the dollar amount of shares repurchased by all firms. The sample period in Panel A and C is from 1985 to 2009 while the sample period in Panel B
and D is from 1985 to 2007.

Mean Median Min 1st 10th 25th 75th 90th 99th Max N $ amount % of aggregate
percentile percentile percentile percentile percentile percentile repurchased value repurchased
Panel A. 12-month abnormal stock return
Middle 80% -0.008 -0.024 -0.412 -0.400 -0.305 -0.198 0.159 0.309 0.512 0.556 424 140.5 89.7%
Middle 90% 0.006 -0.024 -0.509 -0.499 -0.362 -0.226 0.188 0.428 0.789 0.833 478 149.4 95.3%
Middle 95% 0.017 -0.025 -0.614 -0.576 -0.386 -0.238 0.211 0.495 1.143 1.212 503 154.5 98.6%
Middle 99% 0.039 -0.025 -0.863 -0.705 -0.410 -0.243 0.221 0.524 1.733 2.494 525 155.7 99.3%

Panel B. 36-month abnormal stock return

Middle 80% -0.044 -0.083 -0.655 -0.647 -0.535 -0.359 0.196 0.498 1.016 1.134 328 115.2 90.8%
Middle 90% 0.006 -0.082 -0.822 -0.801 -0.604 -0.393 0.261 0.745 1.709 1.786 371 123.4 97.3%
Middle 95% 0.035 -0.083 -0.894 -0.879 -0.629 -0.417 0.277 0.968 2.338 2.991 390 124.7 98.3%
Middle 99% 0.100 -0.082 -0.926 -0.905 -0.653 -0.419 0.295 1.134 3.754 4.531 407 126.5 99.7%

Panel C. 12-month abnormal dollar stock return ($m)

Middle 80% -1.7 0.0 -56.3 -53.4 -21.9 -3.4 3.3 14.7 39.4 49.8 424 31.7 20.3%
Middle 90% -1.9 0.0 -124.3 -114.4 -40.3 -6.2 4.5 23.6 112.6 133.6 478 69.0 44.1%
Middle 95% -1.7 0.0 -215.7 -158.2 -46.4 -7.3 5.2 34.8 171.4 246.1 503 88.4 56.4%
Middle 99% 1.1 0.0 -648.7 -353.1 -54.7 -8.5 6.2 43.2 419.9 923.0 526 141.0 90.0%

Panel D. 36-month abnormal dollar stock return ($m)

Middle 80% -2.5 0.0 -54.7 -53.4 -22.9 -4.0 3.1 12.8 34.8 38.9 329 25.5 20.1%
Middle 90% -3.1 0.0 -114.4 -95.0 -40.2 -6.9 3.8 19.0 79.3 112.6 370 52.5 41.4%
Middle 95% -2.6 0.0 -205.3 -158.2 -46.0 -8.5 4.9 24.3 169.5 229.7 390 68.1 53.6%
Middle 99% 1.7 0.0 -559.7 -286.9 -54.2 -9.2 5.2 36.5 318.8 923.0 407 109.3 86.2%

162
 Table 6.9
Characteristics of large gain, large loss and other repurchases based on future 12-month abnormal returns

This table reports the characteristics of large gain, large loss and other repurchases from 1985 to 2009. Large gain (loss) repurchases are those with future 12-
month abnormal returns (or dollar returns) belong to the top (bottom) decile of the full sample. Other repurchases are those with future 12-month abnormal
returns (or dollar returns) belong to the middle 80% of the full sample. These three subsamples contain 2,677, 21,421, and 2,677 observations respectively.
Repurchase is net repurchase as in Fama and French (2001). The future (past) 12-month abnormal return is the market-adjusted buy-and-hold abnormal return
over the 12 months beginning immediately after (before) the year in question. The abnormal return is the firm’s actual return minus the contemporaneous return
on the value-weighted market index. The future 12-month abnormal dollar return is the product of the dollar amount of shares repurchased and the future 12-
month abnormal return. Assets are book value of total assets at the end of year in question. Market capitalization is the end of year stock price multiplied by the
number of common shares outstanding. Debt-to-assets ratio is the book value of total debt divided by the book value of total assets. Market to book ratio is the
book value of total assets less the book value of equity plus the market capitalization all divided by the book value of total assets.

(1) Large gain (2) Large loss (3) Other (1) - (2) (1) - (3) (2) - (3)
repurchases repurchases repurchases p-value p-value p-value
Variable Mean Median Mean Median Mean Median Mean Median Mean Median Mean Median
Panel A. Large gain and loss based on future 12-month abnormal return
$ repurchase ($m) 35.7 1.32 23.6 0.72 118.0 3.55 0.023 0.000 0.000 0.000 0.000 0.000
$ repurchase/market capitalization 0.04 0.01 0.04 0.01 0.03 0.01 0.136 0.000 0.009 0.204 0.700 0.000
Assets ($m) 998 142 848 107 3,501 357 0.109 0.000 0.000 0.000 0.000 0.000
Market capitalization ($m) 1,088 92 945 82 4,279 350 0.397 0.129 0.000 0.000 0.000 0.000
Prior year debt/assets 0.211 0.165 0.236 0.193 0.203 0.178 0.000 0.011 0.044 0.244 0.000 0.035
Prior year market-to-book 1.735 1.296 2.007 1.398 1.857 1.458 0.000 0.000 0.000 0.000 0.000 0.000
Past 12-month abnormal return -0.045 -0.161 -0.004 -0.148 0.036 -0.042 0.052 0.312 0.000 0.000 0.002 0.000
Future 12-month abnormal return 1.384 0.975 -0.626 -0.597 -0.012 -0.035 0.000 0.000 0.000 0.000 0.000 0.000
Future 12-month abnormal $ return ($m) 35.4 1.5 -13.8 -0.5 -0.1 0.0 0.000 0.000 0.000 0.000 0.000 0.000

Panel B. Large gain and loss based on future 12-month abnormal $ return
$ repurchase ($m) 441.4 119.36 478.0 130.00 10.6 1.20 0.347 0.133 0.000 0.000 0.000 0.000
$ repurchase/market capitalization 0.07 0.04 0.07 0.04 0.03 0.01 0.807 0.257 0.000 0.000 0.000 0.000
Assets ($m) 9,738 2,564 11,522 2,665 1,074 169 0.041 0.682 0.000 0.000 0.000 0.000
Market capitalization ($m) 12,697 2,859 15,136 3,349 1,053 139 0.015 0.006 0.000 0.000 0.000 0.000
Prior year debt/assets 0.215 0.199 0.213 0.195 0.205 0.172 0.623 0.339 0.011 0.000 0.054 0.000
Prior year market-to-book 2.234 1.806 2.360 1.831 1.750 1.355 0.003 0.250 0.000 0.000 0.000 0.000
Past 12-month abnormal return 0.047 -0.002 0.067 0.005 0.016 -0.080 0.102 0.642 0.017 0.000 0.000 0.000
Future 12-month abnormal return 0.567 0.334 -0.293 -0.254 0.049 -0.038 0.000 0.000 0.000 0.000 0.000 0.000
Future 12-month abnormal $ return ($m) 115.0 33.1 -95.0 -26.7 0.1 0.0 0.000 0.000 0.000 0.000 0.000 0.000

163
 Table 6.10
Characteristics of large gain, large loss and other repurchases based on future 36-month abnormal returns

This table reports the characteristics of large gain, large loss and other repurchases from 1985 to 2007. Large gain (loss) repurchases are those with future 36-
month abnormal returns (or dollar returns) belong to the top (bottom) decile of the full sample. Other repurchases are those with future 36-month abnormal
returns (or dollar returns) belong to the middle 80% of the full sample. These three subsamples contain 2,033, 16,262, and 2,032 observations respectively.
Repurchase is net repurchase as in Fama and French (2001). The future (past) 36-month abnormal return is the market-adjusted buy-and-hold abnormal return
over the 36 months beginning immediately after (before) the year in question. The abnormal return is the firm’s actual return minus the contemporaneous return
on the value-weighted market index. The future 36-month abnormal dollar return is the product of the dollar amount of shares repurchased and the future 36-
month abnormal return. Assets are book value of total assets at the end of year in question. Market capitalization is the end of year stock price multiplied by the
number of common shares outstanding. Debt-to-assets ratio is the book value of total debt divided by the book value of total assets. Market to book ratio is the
book value of total assets less the book value of equity plus the market capitalization all divided by the book value of total assets.

(1) Large gain (2) Large loss (3) Other (1) - (2) (1) - (3) (2) - (3)
repurchases repurchases repurchases p-value p-value p-value
Variable Mean Median Mean Median Mean Median Mean Median Mean Median Mean Median
Panel A. Large gain and loss based on future 36-month abnormal return
$ repurchase ($m) 29.6 1.87 24.4 0.79 124.6 4.14 0.196 0.000 0.000 0.000 0.000 0.000
$ repurchase/market capitalization 0.04 0.02 0.04 0.01 0.03 0.01 0.111 0.000 0.002 0.000 0.421 0.001
Assets ($m) 978 179 941 106 3,764 389 0.764 0.000 0.000 0.000 0.000 0.000
Market capitalization ($m) 1,017 105 916 87 4,788 405 0.531 0.006 0.000 0.000 0.000 0.000
Prior year debt/assets 0.226 0.197 0.233 0.197 0.200 0.177 0.241 0.913 0.000 0.003 0.000 0.014
Prior year market-to-book 1.623 1.230 1.919 1.336 1.879 1.478 0.000 0.000 0.000 0.000 0.298 0.000
Past 36-month abnormal return -0.001 -0.332 0.233 -0.219 0.223 -0.050 0.000 0.000 0.000 0.000 0.801 0.000
Future 36-month abnormal return 2.655 1.843 -0.814 -0.802 -0.008 -0.075 0.000 0.000 0.000 0.000 0.000 0.000
Future 36-month abnormal $ return ($m) 57.1 3.9 -19.3 -0.6 2.6 0.0 0.000 0.000 0.000 0.000 0.000 0.000

Panel B. Large gain and loss based on future 36-month abnormal $ return
$ repurchase ($m) 454.8 133.08 505.8 134.49 11.3 1.36 0.253 0.888 0.000 0.000 0.000 0.000
$ repurchase/market capitalization 0.07 0.04 0.07 0.04 0.03 0.01 0.612 0.541 0.000 0.000 0.000 0.000
Assets ($m) 10,189 2,826 12,658 2,951 1,148 186 0.023 0.406 0.000 0.000 0.000 0.000
Market capitalization ($m) 13,448 3,089 17,612 4,011 1,146 154 0.001 0.000 0.000 0.000 0.000 0.000
Prior year debt/assets 0.224 0.212 0.213 0.198 0.202 0.173 0.036 0.019 0.000 0.000 0.014 0.000
Prior year market-to-book 2.220 1.787 2.369 1.821 1.748 1.361 0.003 0.396 0.000 0.000 0.000 0.000
Past 36-month abnormal return 0.210 -0.004 0.398 0.100 0.176 -0.132 0.000 0.000 0.388 0.000 0.000 0.000
Future 36-month abnormal return 1.071 0.611 -0.416 -0.386 0.140 -0.082 0.000 0.000 0.000 0.000 0.000 0.000
Future 36-month abnormal $ return ($m) 196.7 65.6 -141.9 -41.4 0.4 0.0 0.000 0.000 0.000 0.000 0.000 0.000

164
 Table 7.1
The effect of a change in debt and payout increase on leverage and cash holdings

The sample includes 1,066 (671) large payout increases with debt increases (reductions), 347 large payout increases without debt change, and 863 (2,015)
debt-financed payout increases with positive (negative) prior excess leverage over 1951 and 2005. A large payout increase refers to a firm’s payout increase
that is at least 10% of its total assets as of the end of the prior fiscal year. A debt-financed payout increase is a firm’s payout increase that exceeds 50% of its
total debt increase. Payout increase is the sum of the dollar value of the annual dividend increase and repurchase amounts. Leverage is the ratio of total debt to
total assets in book value terms. Excess leverage is the difference between the actual leverage and the estimated target. Target leverage is the fitted value from
a linear regression of debt-to-assets ratio on log(sales), market-to-book ratio, profitability, asset tangibility, and industry median leverage. Cash/TA is the ratio
of cash to total assets. Hypothetical leverage and Cash/TA without debt changes is computed as if the firm had not changed total debt but still increased the
same amount of payouts. Hypothetical leverage and Cash/TA without payout increases is computed as if the firm had not increased payout but still kept the
same amount of debt change. Cash/TA without debt and payout changes is computed as if the firm had not changed both total debt and total payouts. Excess
cash/TA is the difference between a firm’s Cash/TA and the industry median Cash/TA using Fama-French 49 industries. In this analysis, all observations must
have all variables required to compute target leverage.

Panel A. Leverage
Leverage before Leverage after Hypothetical leverage:
Median value of payout increase payout increase no debt change no payout increase
1. Large payout increases with debt increases 0.115 0.271 0.113 0.223
2. Large payout increases with debt reductions 0.130 0.078 0.123 0.064
3. Debt-financed payout increases with positive prior excess leverage 0.341 0.372 0.333 0.351
4. Debt-financed payout increases with negative prior excess leverage 0.107 0.154 0.103 0.144

Panel B. Cash holdings

Large payout increases: Debt-financed payout increases:
Median value of change in debt > 0 change in debt < 0 change in debt = 0 prior excess leverage > 0 prior excess leverage < 0
1. Cash/TA before payout increases 0.103 0.176 0.410 0.038 0.094
2. Cash/TA after payout increases 0.060 0.171 0.373 0.036 0.064
3. Cash/TA without debt changes -0.043 0.215 --- -0.003 0.026
4. Cash/TA without payout increases 0.210 0.300 0.468 0.095 0.131
5. Cash/TA without debt and payout changes 0.123 0.333 --- 0.046 0.085
6. Excess cash/TA prior to payout increases 0.017 0.072 0.244 -0.021 0.018
7. Excess cash/TA after payout increases -0.006 0.063 0.210 -0.022 -0.002
8. Excess cash/TA without debt changes -0.123 0.119 --- -0.067 -0.041
9. Excess cash/TA without payout increases 0.127 0.197 0.313 0.027 0.058
10. Excess cash/TA without debt and payout changes 0.038 0.226 --- -0.018 0.015

165
 Table 7.2
Payout increases and excess leverage

A large payout increase refers to a firm’s payout increase that is at least 10% of its total assets as of the end of the prior fiscal year. A debt-financed payout
increase is a firm’s payout increase that exceeds 50% of its total debt increase. Payout increase is the sum of the dollar value of the annual dividend increase
and repurchase amounts. Leverage is the ratio of total debt to total assets in book value terms. Excess leverage is the difference between the actual leverage
and the estimated target. Target leverage is the fitted value from a linear regression of debt-to-assets ratio on log(sales), market-to-book ratio, profitability,
asset tangibility, and industry median leverage. In this analysis, all observations must have all variables required to compute target leverage and must have
lagged excess leverage available.

Panel A. Percent of payout increases with prior excess leverage that falls in specific intervals
-0.200 -0.200 -0.100 0.000 0.100 0.200 Number
or to to to to or of payout
less -0.100 0.000 0.100 0.200 higher increases
1. Large payout increases with debt increases 18.1% 31.7% 25.7% 12.7% 5.9% 5.8% 975
2. Large payout increases with debt reductions 11.3% 28.0% 28.6% 15.3% 7.8% 9.1% 629
3. Large payout increases without changes in debt 26.2% 43.5% 25.3% 3.1% 0.6% 1.2% 324
4. Debt-financed payout increases with positive prior excess leverage 0.0% 0.0% 0.0% 58.3% 24.0% 17.7% 863
5. Debt-financed payout increases with negative prior excess leverage 19.5% 40.5% 40.0% 0.0% 0.0% 0.0% 2,015

Panel B. Percent of firms that make payout increases with prior excess leverage that falls in specific intervals
-0.200 -0.200 -0.100 0.000 0.100 0.200
or to to to to or
less -0.100 0.000 0.100 0.200 higher
1. Large payout increases with debt increases 1.1% 1.0% 0.7% 0.5% 0.4% 0.3%
2. Large payout increases with debt reductions 0.5% 0.5% 0.5% 0.4% 0.3% 0.3%
3. Large payout increases without changes in debt 0.6% 0.4% 0.2% 0.0% 0.0% 0.0%
4. Debt-financed payout increases with positive prior excess leverage 0.0% 0.0% 0.0% 1.9% 1.3% 0.8%
5. Debt-financed payout increases with negative prior excess leverage 2.6% 2.5% 2.2% 0.0% 0.0% 0.0%

Number of total firm-year observations in each column 15,411 32,295 36,471 25,953 16,137 18,239

166
 Table 7.3
Logit analysis of large payout and debt-financed payout increase decisions

The sample includes 2,084 large payout increases and 2,878 debt-financed payout increases over 1951 and 2005. In
columns 1 to 4, the dependent variable takes the value of one if in the year in question a firm’s payout increase
exceeds 10% of its total assets as of the end of the prior fiscal year, and zero otherwise. In columns 5 to 8, the
dependent variable takes the value of one if in the year in question a firm conducts a debt-financed payout increase,
and zero otherwise. A debt-financed payout increase is a firm’s payout increase that exceeds 50% of its total debt
increase. Payout increase is the sum of the dollar value of the annual dividend increase and repurchase amounts.
Leverage is the ratio of total debt to total assets in book value terms. Excess leverage is the difference between the
actual leverage and the estimated target. Target leverage is the fitted value from a linear regression of debt-to-assets
ratio on log(sales), market-to-book ratio, profitability, asset tangibility, and industry median leverage. The variable
labeled “Below” is a dummy variable that takes the value of one if leverage is below the estimated target, and zero
otherwise. Excess cash is the difference between a firm’s Cash/TA and the industry median Cash/TA using Fama-
French 49 industries, where Cash/TA is the ratio of cash to total assets. Transitory cash flow is non-operating income
(NOPI in Compustat) scaled by contemporaneous total assets. Profitability is operating income before depreciation
(OIBDP in Compustat) divided by contemporaneous total assets. Lagged values of all explanatory variables are used
in the logit analysis. For each independent variable, the marginal effects are reported in the first row, and the t-
statistics are reported in the brackets. The interaction effects are estimated as in Ai and Norton (2003). I calculate t-
statistics using standard error clustered by both firm and year. The results are robust to using industry median
leverage as targets.

Large payout increases Debt-financed payout increases

(1) (2) (3) (4) (5) (6) (7) (8)
Below 0.00 0.00 0.00 0.00
[3.28] [3.79] [0.16] [0.18]
Excess leverage 0.00 0.00 -0.04 -0.04
[0.34] [0.01] [-5.76] [-5.93]
Below × Excess leverage -0.03 -0.05 0.00 0.01
[-2.34] [-2.48] [0.25] [0.39]
Excess cash 0.02 0.03 -0.01 0.00
[7.45] [12.87] [-2.53] [0.98]
Transitory cash flow 0.05 0.12 0.12 0.12
[2.80] [7.55] [5.35] [6.83]
Log of sales 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[3.42] [2.36] [4.07] [3.00] [7.29] [7.25] [8.89] [8.69]
Market-to-book 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[5.93] [7.15] [6.07] [7.68] [0.92] [0.96] [-0.54] [-0.78]
Profitability 0.02 0.03 0.02 0.03 0.06 0.06 0.05 0.05
[5.84] [5.80] [6.50] [6.76] [10.37] [10.41] [11.07] [11.26]
Pseudo R-squared 0.05 0.04 0.03 0.03 0.04 0.04 0.04 0.04

167
 Table 7.4
Estimated probabilities of making large payout increases and debt-financed payout increases: Part I

This table reports the estimated probability of making a large (debt-financed) payout increase in a given year
conditional on specific hypothesized values of explanatory variables. A large payout increase is one in which a
firm’s payout increase exceeds 10% of its total assets as of the end of the prior fiscal year. A debt-financed payout
increase is a firm’s payout increase that exceeds 50% of its total debt increase. Payout increase is the sum of the
dollar value of the annual dividend increase and repurchase amounts. The estimated probabilities are based on the
logit models (1) and (5) in Table 7.3, which includes excess leverage, excess cash, and transitory cash flow as
explanatory variables. Panel A reports the impact of varying one of the three explanatory variables while holding
other variables constant at their sample mean values. Each of these three variables are set at their sample mean
values, sample mean values plus one standard deviation (σ), and sample mean values minus one standard deviation.
Panel B reports the estimated probability as a function of excess leverage and excess cash while holding other
variables constant at their sample mean values. The results are robust to using industry median leverage as targets.

Panel A. Impact of varying one variable

(I) Estimated probability of a large
payout increase as a function of Mean value - σ Mean value Mean value + σ
1. Excess leverage 1.7% 1.2% 0.9%
2. Excess cash 0.9% 1.2% 1.6%
3. Transitory cash flow 1.1% 1.2% 1.3%

(II) Estimated probability of a debt-financed

payout increase as a function of Mean value - σ Mean value Mean value + σ
1. Excess leverage 2.4% 1.8% 1.3%
2. Excess cash 1.9% 1.8% 1.6%
3. Transitory cash flow 1.6% 1.8% 2.0%

Panel B. Impact of both excess leverage and excess cash

(I) Estimated probability of a large payout increase as a function of
Excess leverage Excess cash
Mean value - σ Mean value + σ
Mean value - σ 1.3% 2.4%
Mean value + σ 0.7% 1.2%

(II) Estimated probability of a debt-financed payout increase as a function of

Excess leverage Excess cash
Mean value - σ Mean value + σ
Mean value - σ 2.6% 2.2%
Mean value + σ 1.4% 1.2%

168
 Table 7.5
Estimated probabilities of making large payout increases and debt-financed payout increases: Part II

This table reports the estimated probability of making a large (debt-financed) payout increase in a given year
conditional on specific hypothesized values of explanatory variables. A large payout increase is one in which a
firm’s payout increase exceeds 10% of its total assets as of the end of the prior fiscal year. A debt-financed payout
increase is a firm’s payout increase that exceeds 50% of its total debt increase. Payout increase is the sum of the
dollar value of the annual dividend increase and repurchase amounts. The estimated probabilities are based on the
logit models (1) and (5) in Table 7.3, which includes excess leverage, excess cash, and transitory cash flow as
explanatory variables. Panel A reports the impact of varying one of the three explanatory variables while holding
other variables constant at their sample median values. Each of these three variables are set at their sample mean
values, sample mean values plus one standard deviation (σ), and sample mean values minus one standard deviation.
Panel B reports the estimated probability as a function of excess leverage and excess cash while holding other
variables constant at their sample median values. The results are robust to using industry median leverage as targets.

Panel A. Impact of varying one variable

(I) Estimated probability of a large
payout increase as a function of Mean value - σ Mean value Mean value + σ
1. Excess leverage 1.7% 1.2% 0.9%
2. Excess cash 0.9% 1.2% 1.6%
3. Transitory cash flow 1.1% 1.2% 1.3%

(II) Estimated probability of a debt-financed

payout increase as a function of Mean value - σ Mean value Mean value + σ
1. Excess leverage 2.4% 1.8% 1.3%
2. Excess cash 1.9% 1.8% 1.6%
3. Transitory cash flow 1.6% 1.8% 2.0%

Panel B. Impact of both excess leverage and excess cash

(I) Estimated probability of a large payout increase as a function of
Excess leverage Excess cash
Mean value - σ Mean value + σ
Mean value - σ 1.3% 2.4%
Mean value + σ 0.7% 1.2%

(II) Estimated probability of a debt-financed payout increase as a function of

Excess leverage Excess cash
Mean value - σ Mean value + σ
Mean value - σ 2.6% 2.2%
Mean value + σ 1.4% 1.2%

169
 Table 7.6
Firm characteristics and leverage around specially designated dividends (SDDs) in late 2012

The SDD sample includes 126 industrial firms that paid SDDs between November and December 2012. Firm
characteristics are calculated using the financial data at the end of the fiscal quarter in which a firm paid a SDD. The
comparison sample includes 2,557 other industrial firms that are listed in the Compustat quarterly file in December
2012, and their firm characteristics are computed using financial data at the end of the fiscal quarter that includes
December 2012. The lagged values refer to the values in the fiscal quarter prior to the fiscal quarter in question.
Change in debt is the change in total debt divided by total assets (TA) at the end of the prior fiscal quarter. Tangible
assets/TA is the ratio of the amount of property, plant and equipment (PPENT in Compustat) to total assets. Pro
forma Cash/TA ratios are computed as if the cash paid out as SDDs had been retained in total assets. For non-SDD
payers, pro forrna Cash/TA ratios are the same as the actual Cash/TA ratios. Excess Cash/TA is the difference
between a firm’s Cash/TA and the industry median Cash/TA using Fama-French 49 industries. Operating income is
operating income before depreciation (OIBDP in Compustat) divided by contemporaneous total assets. Non-
operating income is NOPI in Compustat divided by contemporaneous total assets. Leverage is the ratio of total debt
to total assets in book value terms. Excess leverage is defined as the difference between the actual leverage and the
target leverage ratio. Target leverage ratio is the fitted value from a linear regression of debt-to-assets ratio on
variables often hypothesized to affect leverage decisions: log(sales), market-to-book ratio, profitability, asset
tangibility, and industry median leverage ratio. Net debt ratio is the difference between leverage and Cash/TA.

Firms that paid All other firms Wilcoxon

Median value of SDDs in late 2012 listed in late 2012 p-value
1. Lagged leverage 0.055 0.159 0.000
2. Leverage 0.067 0.160 0.003
3. Lagged excess leverage -0.121 -0.051 0.000
4. Excess leverage -0.117 -0.051 0.000
5. Change in leverage 0.000 0.000 0.000
6. Change in debt 0.000 0.000 0.011
7. Operating income/TA 0.041 0.027 0.000
8. Non-operating income/TA 0.000 0.000 0.016
9. Market-to-book 1.706 1.451 0.003
10. Log(sales) 5.320 4.924 0.017
11. Tangible assets/TA 0.253 0.149 0.000
12. Median leverage over past 5 years 0.068 0.141 0.017
13. Median leverage over past 10 years 0.074 0.138 0.015
14. Median historical leverage on Compustat 0.132 0.146 0.953
15. Max historical leverage on Compustat 0.372 0.377 0.823
16. Min historical leverage on Compustat 0.001 0.001 0.763
17. Lagged cash/TA 0.151 0.138 0.804
18. Cash/TA 0.110 0.138 0.124
19. Pro forma cash/TA 0.188 0.138 0.010
20. Lagged excess cash/TA 0.026 0.000 0.006
21. Excess cash/TA 0.000 0.000 0.406
22. Pro forma excess cash/TA 0.046 0.000 0.000
23. Lagged net debt ratio -0.088 0.021 0.038
24. Net debt ratio -0.044 0.019 0.408
25. Total assets ($B) 0.589 0.546 0.202

170
 Table 7.7
Logit analysis of the decision to pay a specially designated dividend (SDD) in late 2012

The sample includes 2,683 industrial firms that are listed in the Compustat quarterly file in December 2012. The
dependent variable takes the value of one if a firm paid a SDD between November and December 2012, and zero
otherwise. Leverage is the ratio of total debt to total assets in book value terms. Excess leverage is the difference
between the actual leverage and the estimated target. Target leverage is the fitted value from a linear regression of
debt-to-assets ratio on log(sales), market-to-book ratio, profitability, asset tangibility, and industry median leverage.
The variable labeled “Below” is a dummy variable that takes the value of one if leverage is below the estimated
target, and zero otherwise. Excess cash is the difference between a firm’s Cash/TA and the industry median Cash/TA
using Fama-French 49 industries, where Cash/TA is the ratio of cash to total assets. Transitory cash flow is non-
operating income (NOPI in Compustat) scaled by contemporaneous total assets. Profitability is operating income
before depreciation (OIBDP in Compustat) divided by contemporaneous total assets. Lagged values of all
explanatory variables are used in the logit analysis. For each independent variable, marginal effects are reported in
the first row, and the t-statistics are reported in the brackets. The interaction effects are estimated as in Ai and Norton
(2003). I calculate t-statistics using standard error clustered by both firm and year. The results are robust to using
industry median leverage as targets.

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

Below -0.00 0.00
[-0.02] [0.13]
Excess leverage 0.00 0.01
[0.04] [0.15]
Below × Excess leverage -0.20 -0.24
[-0.79] [-0.86]
Excess cash 0.06 0.10
[2.27] [4.13]
Transitory cash flow 0.78 0.25
[1.82] [0.28]
Log of sales 0.00 0.00 0.00 0.00
[0.97] [0.45] [1.24] [0.29]
Market-to-book -0.00 -0.00 -0.01 -0.00
[-1.93] [-1.46] [-1.98] [-1.38]
Profitability 0.49 0.48 0.53 0.55
[3.87] [3.76] [3.57] [3.51]
Pseudo R-squared 0.08 0.08 0.05 0.04

171
 Table 7.8
Estimated probability of paying a specially designated dividend (SDD) in late 2012

This table reports the estimated probability of paying a SDD in late 2012 conditional on specific hypothesized
values of explanatory variables. The estimated probabilities are based on the logit model (1) in Table 7.7, which
includes excess leverage, excess cash, and transitory cash flow as explanatory variables. Panel A reports the impact
of varying one of the three explanatory variables while holding other variables constant at their sample mean values.
Each of these three variables are set at their sample mean values, sample mean values plus one standard deviation
(σ), and sample mean values minus one standard deviation. Panel B reports the estimated probability as a function of
excess leverage and excess cash while holding other variables constant at their sample mean values. The results are
robust to using industry median leverage as targets.

Panel A. Impact of varying one variable

Estimated probability of paying
a SDD in late 2012 as a function of Mean value - σ Mean value Mean value + σ
1. Excess leverage 6.2% 2.4% 2.4%

2. Excess cash flow 1.8% 2.4% 3.3%

3. Transitory cash flow 2.1% 2.4% 2.9%

Panel B. Impact of excess leverage and excess cash

Estimated probability of paying a SDD in late 2012 as a function of
Excess leverage Excess cash
Mean value - σ Mean value + σ
Mean value - σ 4.6% 8.2%

Mean value + σ 1.8% 3.2%

172