Scholarly article on Carbon dioxide 124-38-9 from Journal of the American Chemical Society p. 409

THE JOURNAL OF FINANCE • VOL. LV, NO. 5 • OCT. 2000

Stock Repurchases in Canada:

Performance and Strategic Trading

DAVID IKENBERRY, JOSEF LAKONISHOK, and THEO VERMAELEN*

ABSTRACT

During the 1980s, U.S. firms announcing stock repurchases earned favorable long-

run returns. Recently, concerns have been raised over the robustness of these find-

ings. This concern comes at a time of explosive growth in repurchase programs.

Thus, we study new evidence from the 1990s for 1,060 Canadian repurchase pro-

grams. Moreover, because of Canadian law, we can carefully track repurchase ac-

tivity monthly. Similarly to the situation in the United States, the Canadian stock

market discounts the information in repurchase announcements, particularly for

value stocks. Completion rates in Canada are sensitive to mispricing. Trades also

appear linked to price movements; managers buy more shares when prices fall.

In recent years, corporations have dramatically increased the amount of cap-

ital devoted to repurchasing their own shares. In the mid-1980s, repurchase

program announcements in the United States amounted to roughly $25 bil-

lion per year. Between 1996 and 1998 however, more than 4,000 open mar-

ket repurchase programs were announced, which, if fully completed, would

amount to roughly $550 billion.¹During the first six months of 2000 alone,

Securities Data Company reports over 375 program announcements total-

ling $85 billion. Interest in corporate repurchase programs is not limited to

the United States as repurchase activity worldwide has grown in recent years.

Countries such as Hong Kong and Japan recently implemented new regula-

tions allowing companies for the first time to repurchase their shares. A

recent Goldman Sachs study ~March 1999! foresees stock repurchases be-

coming more common in Europe and discusses the potential impact on Eu-

ropean equity values.

* Ikenberry is at Rice University and was a visiting scholar at the University of Washington

while working on this paper. Lakonishok is at the University of Illinois and the National Bu-

reau of Economic Research. Vermaelen is at INSEAD and Maastricht University. We appreciate

the research assistance of Pricha Sethapakdi. Timothy Baikie, Special Counsel Market Regu-

lation at the Toronto Stock Exchange, was helpful in providing data and background informa-

tion about Canadian repurchases. We also thank Dino Mastroianni and Vinod Chandrashekaran

from BARRA for their help and comments. We appreciate the comments of René Stulz ~the

editor!, an anonymous referee, Louis Chan, Ken French, Ravi Jagannathan, Narasimhan Jega-

deesh, Jason Karceski, Inmoo Lee, Raghu Rau, Cliff Stephens, Sheridan Titman, and seminar

participants at Catholic University Leuven, the University of Washington, and the University

of Texas.

1

For comparison, the NYSE reports that aggregate cash dividends paid by member firms

over this period were approximately $490 billion.

2373

2374

The Journal of Finance

The literature is rich with motives as to why firms might repurchase their

own stock. The list includes, for example, tax benefits, distributing excess

cash, and adjusting capital structure. Yet these reasons are not as popular

with managers, who often suggest when announcing buyback programs that

current market prices are too low and hence their stock is an attractive

investment ~Dann ~1983!, Wansley, Lane, and Sarkar ~1989!!. Several stud-

ies report evidence consistent with this managerial story of undervaluation.

In a comprehensive examination of U.S. open market repurchase programs

during the 1980s, Ikenberry, Lakonishok, and Vermaelen ~1995! find that

the market’s initial response to repurchase announcements is not complete.

They report annualized abnormal performance of 2.9 percent over a four-

year period following the announcement. For value stocks, where a stronger

case can be made for undervaluation, the annual abnormal return was 6.4 per-

cent over the same post-announcement period.²

Although previous studies have examined hundreds of repurchase pro-

gram announcements, the history of this transaction is rather short, only

becoming pervasive in the United States in the late 1980s. Moreover, inter-

national evidence regarding repurchases is limited, as many countries have

only recently allowed firms to repurchase stock. Thus, a degree of skepti-

cism regarding the existing evidence on share repurchases is warranted. As

Merton ~1985! suggests, we are biased toward publishing “exciting” results.

Indeed, recent papers have drawn into question the results of several stud-

ies that examine long-run stock performance because of the sensitivity of the

results to the methods used to calculate abnormal returns.³Studies on stock

repurchases have been included in these critiques.

Given the recent growth in repurchase activity worldwide, fresh evidence

on long-run performance of repurchasing firms is of interest. Although many

countries now allow repurchases, the only country outside the United States

that has a reasonable sample at this point is Canada. Thus, we examine

open market programs announced by firms trading on the Toronto Stock

Exchange ~TSE!. Moreover, we utilize programs announced between 1989

and 1997, a time period that generally follows the sample period examined

in earlier U.S. studies. Our sample is comprehensive and contains 1,060

program announcements authorizing ~Canadian! $35 billion in repurchases.

This analysis is supplemented with evidence of long-run stock performance

following the issuance of shares. Specifically, we consider Canadian sea-

soned equity offerings and acquisitions that were at least partially financed

with equity. A result consistent with repurchases would show a downward

drift in stock performance after an announcement of an increase in shares

outstanding.

2

In an earlier paper on fixed-price tender offer repurchases, Lakonishok and Vermaelen

~1990! also find evidence of positive abnormal returns lasting at least two years following the

announcement.

3

For example, Fama ~1998! and Mitchell and Stafford ~1997!.

Stock Repurchases in Canada

2375

Canadian data are of particular interest, however, because firms must

report each month the number of shares they actually repurchase. This in-

formation is summarized by the exchanges and published on a timely basis.

Precise, periodic information on repurchase activity does not exist for U.S.

firms. Moreover, U.S. reporting conventions even make it difficult to obtain

good estimates of overall completion rates ~Stephens and Weisbach ~1998!!.

Our unique data allow us to accurately measure program completion rates

and examine factors that affect these managerial decisions.

Access to monthly repurchase data allows us to study whether managers

execute share repurchases strategically in response to market movements.

Managers repurchase stock for a variety of reasons, many of which may not

relate to mispricing. For example, to the extent that firms repurchase shares

to avoid dilution in earnings from the exercise of options, or use repurchases

as a tax efficient alternative to paying cash dividends, one would not expect

to find a relation between price movements and the level of repurchase ac-

tivity in a given month. Also, if managers view the market as efficient, we

would not expect them to react to price movements. However, managers of-

ten mention undervaluation as an important motive for repurchasing shares.

To the extent that they believe prices deviate from their intrinsic value, we

expect to observe managers repurchasing shares more aggressively when

prices are falling. Likewise, we expect the incentive to repurchase shares to

diminish after a rise in prices. Data from Canada enable us to assess the

extent to which managers trade strategically and thus should help us better

understand what motivates repurchase activity.

We find that undervaluation appears to be just as important a consider-

ation for repurchasing shares in Canada as earlier findings suggest for the

United States. Using the Fama–French ~1993! three-factor model, Canadian

firms repurchasing shares show abnormal performance of 0.59 percent per

month over a three-year period following the announcement. In the year

prior to the announcement, abnormal returns are negative, Ϫ0.35 percent

per month. Such poor preannouncement performance and subsequent out-

performance are consistent with undervaluation being a motivating factor

for stock repurchases. A stronger case for undervaluation as a motive for

repurchasing shares can be made for value stocks. The results are consistent

with this hypothesis. Canadian value stocks that announce repurchases ex-

perience abnormal returns of 0.76 percent per month, whereas the compa-

rable return for growth stocks is only 0.28 percent per month.

We also evaluate performance subsequent to an increase in shares where

firms either issued seasoned equity or acquired other firms with stock. Here,

the results are again consistent with the notion that mispricing is a factor in

managerial decision making. Abnormal returns are high prior to SEOs or

equity-financed acquisitions made by TSE firms, yet subsequent abnormal

return performance is substantially negative.

The results suggest that managers trade strategically. Price changes seem

to have a major impact on repurchase activity. Consistent with the under-

valuation story, an increase in prices leads to a reduction in the number of

2376

The Journal of Finance

shares repurchased, whereas a drop in stock prices leads to a larger fraction

of shares repurchased. Completion rates are also higher for value stocks

than for growth stocks. Moreover, firms that actively repurchased shares

experience only modest returns while the program was in effect. Yet, these

firms enjoy sizable abnormal returns in subsequent years. Conversely, com-

panies that did not buy any shares show strikingly large excess returns

while the program is in effect, yet no evidence of positive abnormal returns

in subsequent years. Overall, the results suggest that managers’ repurchase

decisions are sensitive to prevailing market prices, a result consistent with

the undervaluation hypothesis.

The paper is organized as follows. Section I briefly describes the institu-

tional setting surrounding repurchase programs in Canada and also de-

scribes how we formed our sample. Section II reviews the methodology used

to evaluate long-run performance. Section III provides descriptive informa-

tion regarding sample programs and documents the short-run market reac-

tion to repurchase announcements. Section IV presents evidence on subsequent

long-run performance. Section V reports evidence on program repurchase

completion rates and the extent to which managers execute repurchase pro-

grams strategically. Section VI further examines the strategic nature of re-

purchase decisions by examining long-run performance conditional on various

levels of repurchase activity. Section VII provides results for the opposing

case where managers issue equity either through SEOs or stock-financed

acquisitions. Concluding remarks are provided in Section VIII.

I. Repurchases in Canada

A. Institutional Aspects of Repurchases in Canada

The regulatory environment affecting Canadian open market repurchase

programs differs from that in the United States. In Canada, these programs

are referred to as “Normal Course Issuer Bids.” The primary oversight mech-

anism for these programs is provided by the exchange where the company’s

stock trades. Unlike U.S. firms, Canadian firms must receive approval from

the exchange before initiating a repurchase program. The application pro-

cess to repurchase stock at the TSE is generally straightforward and is usu-

ally completed in a limited number of days. As in the United States, these

programs are the most common mechanism firms use to repurchase shares.⁴

Once authorized, programs last one year and are limited to the maximum of

either 10 percent of public float or five percent of shares outstanding. Thus,

in many cases, the number of authorized shares exceeds five percent of the

share base.

4

Fixed-price tender offers, referred to as “Substantial Issuer Bids,” are also permitted in

Canada and do not have limits on the amount of stock that can be repurchased. As in the

United States, this form of repurchase is not common. Dutch-auction tender offers, though

permitted by law, have only recently occurred in Canada. In this study, we only examine re-

purchase programs executed via open market transactions, or Normal Course Issuer Bids.

Stock Repurchases in Canada

2377

TSE rules also govern how listed firms execute their repurchase pro-

grams.⁵Briefly, these rules state that TSE firms may commence repur-

chasing stock two days after receiving approval from the TSE. Programs

must be executed through a single broker. Cumulative repurchases over the

trailing 30 calendar days must not exceed two percent of the outstanding

shares. Firms are also restricted from repurchasing shares on an up-tick.

Similar rules exist in the United States.⁶

B. The Sample

We obtained the complete record of common stock repurchase programs

announced by TSE-listed firms between 1989 and 1997. Information regard-

ing all authorized repurchase programs is published periodically in the Daily

Record, a TSE publication. Once a month, the TSE publishes a comprehen-

sive overview of the status of all authorized programs including each pro-

gram’s size, termination date, the number of shares purchased in the previous

month ~including those purchased on other exchanges!, and the total shares

repurchased to date in the program. During this time period, 1,159 program

authorizations were announced by TSE firms.

Monthly stock returns were obtained from the TSE0University of Western

Ontario database, a data set free of survivor bias. However, our returns

from this database end in December 1995. Thus, from January 1996 to

December 1998, monthly returns were obtained from COMPUSTAT and

Datastream. From these sources, we located returns for 1,060 repurchase

programs.

Some of our analysis requires fundamental information, specifically the

book-to-market ratio. COMPUSTAT was our primary source for this infor-

mation. However, COMPUSTAT’s coverage of Canadian firms is not com-

plete. In addition, Canadian firms that are delisted are not included in

COMPUSTAT’s research file. For these cases, we obtained book-equity

values from a special file prepared for us by Standard and Poor’s. As a

further step, we hand collected missing book-equity values from Moody’s

International Manuals. From these sources, book-to-market ratios were ob-

tained for 86 percent of our sample.

5

See Part 28 of the Policies of the Toronto Stock Exchange.

U.S. securities law governing stock repurchases has evolved gradually, adopting its present

6

form in 1982 when Congress adopted S.E.C. Rule 10b-18, a “safe-harbor” provision relating to

stock repurchases. This rule does not impose strict limits as is the case in Canada but rather

provides firms protection from legal liability for price manipulation when repurchasing stock if

firms execute repurchase trades within certain limitations relating to price, volume, and time

of day. These provisions are not statutory rules. Companies often voluntarily repurchase shares

outside of these limits ~Cook, Krigman, and Leach ~1997!!. Moreover, Rule 10b-18 does not

apply to all types of repurchase transactions, including, for example, pension-plan-related ~ERISA!

trades. In addition, at certain times ~e.g., following the crash of 1987!, some of the Rule 10b-18

limitations have been lifted.

2378

The Journal of Finance

II. Methodology

A. Portfolio Formation

Our examination of long-run performance focuses on portfolio returns

produced in calendar time. Several recent papers report evidence based on

this approach, including Ikenberry et al. ~1995!, Loughran and Ritter ~1995!,

Womack ~1996!, and Mitchell and Stafford ~1997!. To use this approach, we

calculate monthly returns in calendar time for a portfolio of repurchase firms.

Firms are added to the portfolio at the beginning of the month following

their repurchase announcement and retained for the next three years or

until the stock no longer trades. At the beginning of each month, the port-

folio is rebalanced so that each stock receives the same weight. Over time,

new companies enter the portfolio and old companies leave, causing the num-

ber of stocks in the portfolio to vary. In some months, the number of stocks

in the portfolio is small. To reduce the impact of idiosyncratic noise, partic-

ularly when we partition our data set into various subsamples, months where

a portfolio contains six stocks or less are dropped from our analysis.

The portfolio described above assumes that repurchase firms are held for

three years, beginning in the month following the announcement. In addi-

tion to this strategy, we calculate portfolio returns for other investment ho-

rizons including year 1, year 2, and year 3 following the announcement.

For example, for the year 1 portfolio, a monthly time series of returns is

calculated for firms that announced a program in the previous 12 months.

Similarly, we also report performance for year Ϫ1.

B. Benchmark Models and Index Returns

Using the calendar-time portfolio approach, we estimate abnormal perfor-

mance using two benchmarks. We first evaluate performance using a stan-

dard one-factor, CAPM approach. Here, we regress the time series of monthly

excess returns to the repurchase portfolio on excess market returns:

r

_{p, t}Ϫ r_{f, t}ϭ a ϩ b_m~r_{m, t}Ϫ r_{f, t}! ϩ e_t

~1!

where r_{p, t}is the monthly return to a given calendar-time portfolio. The risk-

free return, r_{f, t}, is the return to the 30-day Canadian government bill. The

return to the TSE 300 Index is our market return, r_{m, t}. Abnormal perfor-

mance is measured by a.

The focus of our performance analysis relies on the Fama–French ~1993!

three-factor model:

r

_{p, t}Ϫ r_{f, t}ϭ a ϩ b_m~r_{m, t}Ϫ r_{f, t}! ϩ b_vmg~r_{value, t}Ϫ r_{growth, t}

ϩ b_sml~r_{small, t}Ϫ r_{large, t}! ϩ e_t

!

~2!

The first term in equation ~2! controls for general market movements in a

given month t. The last two terms in equation ~2! are value0growth and size

factors. These factors are calculated using BARRA’s Canadian style returns

Stock Repurchases in Canada

2379

for small-cap-value stocks, small-cap-growth stocks, large-cap-value stocks,

and large-cap-growth stocks.⁷For its Canadian indices, BARRA defines its

large-cap universe as the largest 200 firms trading on the TSE. The next

largest 300 TSE firms define the small-cap universe. These two groups are

each subdivided into growth and value portfolios. BARRA accomplishes this

by calculating a measure that gives two-thirds weight to the market-to-

book ratio and one-third weight to dividend yield. The small-cap and large-

cap portfolios are sorted on this measure and split into two groups so that

the value and growth portfolios have the same total market capitalization.

For the three-factor model above, the book-to-market factor is defined as

the average return for BARRA’s two value portfolios ~large-cap-value and

small-cap-value! minus the average return to BARRA’s two growth port-

folios ~large-cap-growth and small-cap-growth!. This factor is labeled as the

value minus growth portfolio, or vmg. Similarly, the size factor is obtained

by taking the average return of BARRA’s two small-cap portfolios ~small-

cap-value and small-cap-growth! and subtracting the average return of

BARRA’s two large-cap portfolios ~large-cap-value and large-cap-growth!. This

factor is labeled as small minus large, or sml. BARRA updates these style

indices in January and July of each year. These series start in July 1990.

Thus whereas we include repurchase programs announced during 1989 and

the first half of 1990, we do not begin measuring excess performance until

July 1990.⁸

III. The Market Reaction to Open Market Share

Repurchase Announcements

Panel A of Table I reports summary statistics for our repurchase sample.

Looking at all years together, the 1,060 programs in our sample, if fully

completed, would amount to $35.6 billion. The mean program is for 5.22 per-

cent of shares outstanding, similar in magnitude to the typical open market

repurchase program in the United States. The mean market reaction to open

market programs in our sample is modest. The average abnormal return

7

BARRA is a U.S.-based firm that provides software and information services to investment

managers, pension fund consultants, plans sponsors, and others in the investment community.

The style indices used in this study were obtained from BARRA’s Canadian Internet site:

www.barra.ca. Although not reported in this paper, we examined the behavior of the three-

factor model using BARRA style indices. Overall, the model appears to work well in Canada.

8

To maximize the number of repurchase firms in our sample, we begin recording repurchase

announcements in January 1989, the first month for which copies of the Daily Record were

available from the TSE archives. We begin measuring performance in July 1990 when the

BARRA style factors are available. Of course, an alternative to using BARRA factors would be

to construct our own factors, thus allowing us to measure performance for 18 more months prior

to July 1990. Although possible, the portfolios would have less seasoning and, thus, would be

less populated and more noisy. Moreover, for much of the 1990 to 1998 period, BARRA calcu-

lates and reports its style factors in real time, thus reducing concerns over potential biases.

2380

The Journal of Finance

Table I

Summary Information for Open Market Stock Repurchase

Programs Announced by TSE-Listed Firms

This table reports summary information for sample firms listed on the Toronto Stock Exchange

that announced an open market stock repurchase program ~normal course issuer bid! during

the period 1989 to 1997. Reported are the number of firms, the mean percentage of outstanding

shares announced for repurchase, the cumulative dollar amount ~Canadian $ million!, the mean

abnormal announcement return, and its associated t-statistic. The mean abnormal announce-

ment return is the average total return for sample firms in the announcement month adjusted

for the respective TSE index total return. Panel A reports this summary information by year.

Panel B reports it by program size.

Mean %

Announced

for

Abnormal

Announcement

Return

Dollar

Year

n

Repurchase

Amount

~%!

t-Statistic

Panel A: By Year

1989

1990

1991

1992

1993

1994

1995

1996

1997

106

144

92

79

62

110

166

129

172

5.54

5.32

5.10

4.95

5.05

5.22

5.34

4.98

5.26

3,983.5

5,136.8

1,715.1

1,200.4

1,458.7

2,755.1

3,896.3

5,828.8

9,660.3

1.27

1.20

2.29

3.87

2.23

Ϫ1.71

0.14

0.18

0.96

~1.60!

~1.16!

~1.33!

~2.85!

~1.29!

~Ϫ1.41!

~0.15!

~0.22!

~0.78!

All

1,060

5.22

35,635.0

0.93

~2.36!

Panel B: By Program Size

Less than 4.5%

4.5 to 5.5%

More than 5.5%

194

603

263

2.58

4.98

7.72

8,864.5

15,662.1

11,348.4

0.90

1.05

0.69

~1.21!

~1.94!

~0.83!

All

1,060

5.22

35,635.0

0.93

~2.36!

relative to the TSE 300 Index in the month the program is announced is 0.93

percent.⁹This is somewhat less that the market reaction observed for U.S.

announcements.

Panel B reports the mean market reaction for various program sizes. A

substantial number of programs cluster right at five percent of shares out-

standing. Thus, we report evidence for all programs announced for between

4.5 percent and 5.5 percent of the share base. Those programs below 4.5 per-

cent or above 5.5 percent are reported in two separate groups. There is little

apparent difference in the market reaction between large and small pro-

9

As a check, we also calculated the market reaction adjusting for book-to-market and size

effects. Using Ibbotson’s Returns across Time and Securities RATS technique ~1975!, the

market reaction is nearly the same as obtained using the simple market-adjusted approach,

1.05 percent.

Stock Repurchases in Canada

2381

grams. The mean abnormal return for programs between zero and 4.5 per-

cent of shares outstanding is 0.90 percent. For the largest programs in our

sample, those for more than 5.5 percent of shares outstanding, the mean

market reaction is slightly lower, 0.69 percent.

In spite of managers’ frequent claims of undervaluation when announc-

ing a repurchase program, the announcement return is small. This is con-

sistent with managers either being overly optimistic relative to the market

about their firm’s value or, alternatively, the market being wrong and thus

underreacting to the repurchase signal. An investigation of the long-term

performance of repurchasing firms is a way to sort out the two competing

hypotheses.

IV. The Long-Term Performance of Canadian

Open Market Repurchase Programs

A. The General Case

In Table II, we report performance of the repurchase strategy using both

one-factor and three-factor models. Assuming a three-year holding period,

we observe abnormal performance using the one-factor model of 0.62 percent

per month ~t ϭ 2.70!. Using a two-year holding period, our estimate of ab-

normal performance is slightly higher, 0.70 percent per month ~t ϭ 2.90!.

Focusing on one-year holding periods, abnormal performance is positive in

each of the three years following the announcement.

After controlling for both size and book-to-market using the three-factor

model, the results are generally the same. The three-year holding period

strategy shows abnormal performance of 0.59 percent per month ~t ϭ 4.11!,

or roughly 7 percent per year. In this study, the similarity between the one-

and three-factor models is not unexpected given the relatively small premi-

ums associated with the additional two factors. The mean return on the vmg

factor during our sample period was less than 1.5 percent per year, whereas

the mean return to the sml factor was nearly zero, Ϫ0.25 percent per year.

Given these modest premiums, alternative approaches to adjust for size and

value growth should not substantially change the results.

A calendar-time approach weighs each month equally in the analysis even

though the number of companies in the portfolio is not the same over time.

Thus, in Table III, we report evidence using an event-time methodology where

each announcement receives the same weight. Here, we apply Ibbotson’s

Returns across Time and Securities ~RATS! procedure ~1975! and, as before,

measure abnormal performance using a three-factor model. Using this pro-

cedure, we continue to observe evidence of positive abnormal returns. After

three years, the total cumulative abnormal return is 21.40 percent ~t ϭ 6.87!.

This translates to an abnormal return of 0.59 percent per month, a result

that matches that obtained using the calendar-time approach.

To examine the robustness of our results, we performed additional tests in

the context of the calendar-time approach. We estimated abnormal perfor-

mance using weighted least squares to deal with the problem of an uneven

Table II

Long-Run Performance of TSE-Listed Firms Announcing Open Market Stock Repurchase Programs

Portfolios are formed in calendar time for TSE-listed firms announcing an open market stock repurchase program during the period 1989 to

1997. As time proceeds, firms are introduced to a portfolio at various points relative to the announcement month. For example, for the portfolio

labeled year 1, firms are included in the portfolio at the beginning of the month following the announcement ~month 1! and remain in the

portfolio until month 12. Portfolios are also formed for year 2 ~months 13 to 24!, year 3 ~months 25 to 36!, years 1 to 2 ~months 1 to 24!, years

1 to 3 ~months 1 to 36!, and year Ϫ1 ~months Ϫ12 to Ϫ1!. Excess performance ~in percentages! is measured using both a one-factor CAPM

approach and a three-factor model as suggested by Fama and French ~1993!:

r

_{p, t}Ϫ r_{f, t}ϭ a ϩ b_m~r_{m, t}Ϫ r_{f, t}! ϩ e_t

_{p, t}Ϫ r_{f, t}ϭ a ϩ b_m~r_{m, t}Ϫ r_{f, t}! ϩ b_vmg~r_{value, t}Ϫ r_{growth, t}! ϩ b_sml~r_{small, t}Ϫ r_{large, t}! ϩ e_t

where r_{p, t}Ϫ r_{f, t}is the excess portfolio return in month t and ~r_{m, t}Ϫ r_{f, t}! is the monthly excess return to the TSE. In the three-factor model,

~r_{small, t}Ϫ r_{large, t}! and ~r_{value, t}Ϫ r_{growth, t}! are size and book-to-market factors, respectively. Both factors are derived from large-cap-value, large-

cap-growth, small-cap-value, and small-cap-growth style indices defined by BARRA using 500 TSE-listed firms. The value minus growth factor

is calculated by taking the difference between the average small-cap-value and large-cap-value indices and the average small-cap-growth and

large-cap-growth indices. Likewise, the small minus large factor is calculated by differencing the average small-cap-value and small-cap-growth

indices with the average large-cap-value and large-cap-growth indices. Calendar-time portfolios are constructed beginning July 1990 and ending

December 1998. Months where the portfolios contained six or fewer stocks are ignored. t-statistics are provided in parentheses and test the

hypothesis that the parameter value is zero, except for b_m, where the hypothesis test value is 1.0.

Year Ϫ1

Year 1

Year 2

Year 3

Years 1 to 2

Years 1 to 3

All firms

~n ϭ 1,060!

a

b_m

R²

Ϫ0.100 ~Ϫ0.39!

0.82 ~Ϫ2.34!

0.57

0.700 ~2.98!

0.79 ~Ϫ3.57!

0.65

0.693 ~2.58!

0.77 ~Ϫ3.46!

0.58

0.575 ~2.03!

0.75 ~Ϫ3.65!

0.55

0.695 ~2.90!

0.78 ~Ϫ3.69!

0.64

0.616 ~2.70!

0.78 ~Ϫ3.98!

0.66

a

Ϫ0.348 ~Ϫ2.09!

0.93 ~Ϫ1.43!

0.671 ~4.25!

0.89 ~Ϫ2.65!

0.671 ~3.69!

0.87 ~Ϫ2.70!

Ϫ0.01 ~Ϫ0.12!

0.78 ~10.11!

0.81

0.498 ~2.02!

0.85 ~Ϫ2.27!

0.669 ~4.44!

0.88 ~Ϫ3.01!

0.587 ~4.11!

0.88 ~Ϫ3.35!

b_m

b_vmg

b_sml

R²

0.01

~0.12!

0.05

~0.68!

0.08

0.59

0.66

~0.65!

~5.33!

0.02

~0.30!

0.05

~0.72!

0.77 ~10.07!

0.83

0.70 ~10.53!

0.84

0.74 ~11.66!

0.86

0.72 ~11.94!

0.86

Stock Repurchases in Canada

2383

Table III

Abnormal Performance for TSE-Listed Repurchases

Estimated Using RATS

Monthly returns ~in percentages! for all firms in our sample are aligned in event time, where

month 0 represents the announcement month. Abnormal performance is measured using monthly

data over the period July 1990 to December 1998 using Ibbotson’s ~1975! Returns across Time

and Securities ~RATS! methodology and applying the Fama–French ~1993! three-factor model:

r

_{i, t}Ϫ r_{f, t}ϭ a_tϩ b_t~r_{m, t}Ϫ r_{f, t}! ϩ b_{vmg, t}~r_{value, t}Ϫ r_{growth, t}! ϩ b_{sml, t}~r_{small, t}Ϫ r_{large, t}! ϩ e_t.

Here, r_{i, t}Ϫ r_{f, t}is the excess return to firm i in month t. For that same month, ~r_{m, t}Ϫ r_{f, t}

!

is the excess return to the TSE, ~r_{small, t}Ϫ r_{large, t}! is the return for the size factor, and

~r_{value, t}Ϫ r_{growth, t}! is the return to the book-to-market factor. Both factors are derived from

large-cap-value, large-cap-growth, small-cap-value, and small-cap-growth indices defined by

BARRA using 500 TSE-listed firms. The value minus growth factor is calculated by taking the

difference between the average small-cap-value and large-cap-value indices and the average

small-cap-growth and large-cap-growth indices. Likewise, the small minus large factor is cal-

culated by differencing the average small-cap-value and small-cap-growth indices with the

average large-cap-value and large-cap-growth indices. Using this approach, a_tis our estimate

of abnormal return ~AR!, which is estimated each month and cumulated over time ~CAR!.

Month

AR

t-Statistic

CAR

t-Statistic

1

2

3

4

5

6

1.49

0.94

0.36

0.19

0.75

~2.32!

~1.80!

~0.76!

~0.47!

~1.85!

~Ϫ0.33!

1.49

2.43

2.79

2.98

3.73

3.59

~1.89!

~2.57!

~2.64!

~2.63!

~3.10!

~2.83!

Ϫ0.14

9

12

15

18

21

24

27

30

33

36

0.31

0.65

0.81

0.26

0.29

0.05

1.15

Ϫ0.49

0.58

Ϫ0.01

~0.65!

~1.59!

~1.78!

~0.55!

~0.63!

~0.11!

~1.26!

~Ϫ1.00!

~1.16!

~Ϫ0.01!

5.37

7.27

9.09

10.81

13.20

14.91

17.91

18.59

19.48

21.40

~3.63!

~4.39!

~4.97!

~5.45!

~6.14!

~6.50!

~6.80!

~6.71!

~6.72!

~6.87!

number of securities in the portfolio over time. We also calculated perfor-

mance using a value-weighted portfolio strategy. In both cases, the results

were similar to our basic approach.

B. Cross-Sectional Differences in Long-Run Performance:

Value versus Growth

One way to sort out the different motivations managers have for repur-

chasing shares is to consider the book-to-market ratio. Managers in growth

firms, with relatively low book-to-market ratios, may repurchase shares for

2384

The Journal of Finance

a variety of reasons, such as avoiding dilution of earnings. For value com-

panies where book-to-market ratios are relatively high, undervaluation may

be a more important motivating factor.

Thus, in June of each year, we obtain book-to-market ratios ~assuming a

three-month reporting delay following fiscal year-end for book-equity val-

ues! for all TSE-listed firms for which we have data. This universe is sorted

each year by book-to-market ratio. Those firms with above median book-to-

market ratios are called value stocks, whereas those with below average

ratios are labeled growth stocks. For each of these two groups, we estimate

long-run performance using the calendar-time approach, the results of which

are reported in Table IV.

Value companies announcing a repurchase program substantially outper-

form growth companies. Assuming a three-year holding period, the abnor-

mal return for value stocks using the three-factor model is 0.76 percent per

month, or 9 percent per year. For growth stocks, the comparable return is

less than half this amount, 0.28 percent per month. This spread between

value and growth firms in Canada announcing buybacks is consistent with

the pattern observed for U.S. firms ~Ikenberry et al. ~1995!!.

Prior to the announcement, we also see evidence consistent with the un-

dervaluation story.¹⁰Although, value firms perform well subsequent to a

program announcement, performance in the prior year is poor: Ϫ0.91 per-

cent per month ~t ϭ Ϫ3.80!. Growth firms, on the other hand, perform well

in the year prior to the announcement, with abnormal performance of 0.51

percent per month ~t ϭ 2.19!.

V. Actual Share Repurchase Activity in Canada

A. The Evolution of Repurchase Activity over Time

and Completion Rates

Canadian law differs from U.S. law in that the progress of all authorized

repurchase programs is reported each month regardless of whether firms

are buying shares or not. This information is summarized in Table V. Panel

A reports the distribution of repurchase completion rates. The fraction of

shares actually repurchased is surprisingly low. At termination of the pro-

gram, roughly a quarter of the firms did not repurchase any shares. Less

than five percent of the firms fully complete their repurchase program. Over-

all, the mean completion rate is 28.6 percent. For comparison, Stephens and

Weisbach ~1998! report one-year completion rates ranging between 46 per-

cent and 75 percent for a sample of U.S. programs announced between 1981

and 1995.¹¹

10

Similar evidence of negative abnormal returns prior to repurchase announcements is re-

ported by Vermaelen ~1981! and Comment and Jarrell ~1991!.

11

Because U.S. firms are not required to disclose repurchase activity directly, Stephens and

Weisbach ~1998! use a variety of approaches to estimate completion rates for U.S. stocks.

Table IV

Abnormal Long-Run Performance for Value versus Growth Repurchase Programs

Portfolios are formed in calendar time for TSE-listed firms that announce an open market stock repurchase program. In June of each year, value

and growth cutoff values are determined by sorting TSE-listed stocks into two groups on the basis of book-to-market. Those with relatively low

book-to-market ratios are labeled Growth Stocks, whereas the rest are labeled Value Stocks. Excess performance ~in percentages! is measured

using a three-factor model:

r

_{p, t}Ϫ r_{f, t}ϭ a ϩ b_m~r_{m, t}Ϫ r_{f, t}! ϩ b_vmg~r_{value, t}Ϫ r_{growth, t}! ϩ b_sml~r_{small, t}Ϫ r_{large, t}! ϩ e_t

where r_{p, t}Ϫ r_{f, t}is the excess portfolio return in month t and ~r_{m, t}Ϫ r_{f, t}! is the monthly excess return to the TSE. In the three-factor model,

~r_{small, t}Ϫ r_{large, t}! and ~r_{value, t}Ϫ r_{growth, t}! are size and book-to-market factors, respectively. Both factors are derived from large-cap-value, large-

cap-growth, small-cap-value, and small-cap-growth style indices defined by BARRA using 500 TSE-listed firms. The value minus growth factor

is calculated by taking the difference between the average small-cap-value and large-cap-value indices and the small-cap-growth and large-cap-

growth indices. Likewise, the small minus large factor is calculated by differencing the average small-cap-value and small-cap-growth indices

with the average large-cap-value and large-cap-growth indices. Calendar-time portfolios are constructed beginning July 1990 and ending De-

cember 1998. Months where the portfolios contained six or fewer stocks are ignored. t-statistics are provided in parentheses and test the

hypothesis that the parameter value is zero, except for b_m, where the hypothesis test value is 1.0.

Year Ϫ1

Year 1

Year 2

Year 3

Years 1 to 2

Years 1 to 3

Value stocks

~n ϭ 491!

a

Ϫ0.913 ~Ϫ3.80!

0.80 ~Ϫ2.66!

0.820 ~3.42!

0.81 ~Ϫ3.09!

0.894 ~3.17!

0.80 ~Ϫ2.70!

Ϫ0.10 ~Ϫ0.74!

0.661 ~2.31!

0.77 ~Ϫ2.93!

0.810 ~3.74!

0.82 ~Ϫ3.35!

0.762 ~3.77!

0.81 ~Ϫ3.64!

b_m

b_vmg

b_sml

R²

0.01

0.66

0.64

~0.02!

~6.06!

0.18

0.65

0.64

~1.65!

~6.44!

0.02

0.69

0.58

~0.15!

~5.37!

0.07

0.69

0.71

~0.65!

~7.58!

0.08

0.72

0.74

~0.83!

~8.44!

0.71

0.60

~5.92!

Growth stocks

~n ϭ 418!

a

0.509 ~2.19!

0.99 ~Ϫ0.09!

Ϫ0.04 ~Ϫ0.38!

0.349 ~1.95!

0.98 ~Ϫ0.53!

Ϫ0.10 ~Ϫ1.18!

0.315 ~1.69!

0.97 ~Ϫ0.64!

0.040 ~0.19!

0.90 ~Ϫ1.79!

0.374 ~2.59!

0.96 ~Ϫ1.07!

0.276 ~2.13!

0.95 ~Ϫ1.53!

b_m

b_vmg

b_sml

R²

0.10

0.69

0.82

~1.22!

~8.74!

0.05

0.42

0.76

~0.51!

~4.42!

0.00

~0.00!

0.03

~0.48!

0.63

0.73

~5.92!

0.56

0.83

~7.33!

0.64 ~10.42!

0.88

0.58 ~10.65!

0.90

2386

The Journal of Finance

Table V

Stock Repurchase Completion Rates in Canada

This table reports cumulative share purchase activity ~in percentages! by month for TSE-listed

firms announcing open market stock repurchases, or normal course issuer bids, during the

period 1989 to 1998. Normal course issuer bids are defined to last one year. Thus, months 0 and

12 are partial months. This table reports activity for all firms, including those that departed

the sample prior to one full year. Repurchase activity is reported overall, by year, by program

size, by market-capitalization, and by whether the firm is ranked as value or growth. Market-

cap is determined using June cutoff values each year for all TSE-stocks. Firms below median

market-cap in a given June are labeled as Small-cap, whereas those with above median market-

cap are labeled Large-cap. Similarly, each June, all TSE-listed firms are used to determine

book-to-market cutoff values. Those repurchase firms whose book-to-market ratio is below me-

dian are labeled Growth, whereas those above median are labeled Value. Extreme value and

Extreme growth are those firms whose book-to-market ratio ranks them in the highest or low-

est deciles respectively compared to all TSE-listed stocks in a given June.

Panel A: Completion Rate ~in %! after 12 Months

Portion of Program Completed

n

Mean

No shares repurchased

0 to 5%

236 ~22.3%!

149 ~14.1%!

0.0

2.1

5 to 10%

90

~8.5%!

7.4

10 to 25%

25 to 50%

50 to 75%

75 to 100%

161 ~15.2%!

171 ~16.1%!

16.7

35.4

63.0

93.4

99

~9.3%!

154 ~14.5%!

All firms

1,060

~100%!

28.6

Panel B: Cumulative Trading Activity

Program

Size

% of Program Completed by Month

~%!

0

1

2

3

6

9

12

All programs

5.22

2.3

5.8

9.0

11.9

19.1

24.8

28.6

Years 1989–1991

Years 1992–1993

Years 1994–1995

Years 1996–1997

5.33

4.99

5.29

5.14

1.8

1.0

3.4

2.4

4.3

3.8

7.4

7.0***

6.7

6.9

10.7

10.9***

8.9

9.5

14.3

14.2***

14.3

17.4

22.2

22.5***

18.7

22.8

28.0

29.7***

21.4

26.2

31.8

35.1***

Programs less

than 4.5%

Programs between

4.5 and 5.5%

Program greater

than 5.5%

2.58

4.98

7.72

2.7

2.1

2.5

7.0

5.3

6.2

10.3

8.2

14.6

10.8

12.6

25.0

32.5

37.9

17.4

22.4

25.8

9.6

18.7***

24.6***

28.3***

Small-cap

Large-cap

5.32

5.05

2.6

1.8**

6.3

5.0

9.6

7.8

12.7

10.5

19.9

17.8

25.9

22.9

29.6

26.9

Growth

Value

5.06

5.31

2.3

2.2

5.3

6.3

8.1

9.5

10.2

12.8**

17.0

20.1

21.8

26.3**

25.1

30.9***

Extreme growth

Extreme value

5.12

5.33

3.6

3.2

8.2

7.4

11.6

13.8

16.1

16.5

25.9**

18.2

32.9***

19.6

37.7***

** and *** denote significance at the 0.05 and 0.01 levels, respectively, for differences in com-

pletion rates between the top and bottom rows of a given subpanel.

Stock Repurchases in Canada

2387

Panel B reports cumulative repurchase activity at various points during

the program. This is reported for the overall sample and for various sub-

samples. Slightly less than half of all repurchase activity occurs during months

0 to 3. The completion rate in Canada has increased over time. In the first

two years, it was 21.4 percent, whereas in the last two years it increased

to 35.1 percent. Perhaps a more interesting result relates to the difference

in completion rates between value and growth stocks. Value stocks have a

higher completion rate, 30.9 percent, in comparison to growth stocks, 25.1 per-

cent. Extreme value stocks, defined as companies in the top book-to-market

decile, have an even higher completion rate of 37.7 percent. On the other

hand, extreme growth stocks have an average completion rate of only 19.6 per-

cent. The difference in completion rates between value and growth firms,

both overall and in extreme cases, is significant at traditional confidence

levels.

B. The Determinants of Completion Rates and Strategic Trading

Next, we consider in greater detail factors that influence repurchase com-

pletion rates. If undervaluation is an important reason leading managers to

repurchase stock, one would expect price movements during the program to

be an important consideration. A run-up in stock prices after a repurchase

announcement would seemingly diminish the attractiveness of buying shares.

Conversely, a drop in prices would provide an increased incentive to buy

stock. Thus, the first factor we consider is the excess stock return during the

year of the program relative to the TSE 300 Index, which we label as Year 1

Excess Return.

The next two variables considered are the firm’s Book-to-Market Decile

Rank and Size Decile Rank compared to all TSE firms as of the previous

June. Earlier, we observed that repurchasing firms with high book-to-

market ratios experienced higher postannouncement abnormal returns com-

pared to firms with lower ratios, a result consistent with undervaluation. If

undervaluation is an important motive for buybacks in value companies, we

expect them to have higher completion rates compared to growth stocks,

where the motives for buying stock are more complicated. We saw traces of

this behavior earlier in Table V. The variable Size Decile Rank also relates

to the undervaluation hypothesis. Evidence suggests that smaller firms are

less efficiently priced. Thus, one might expect to see higher completion rates

among smaller firms in comparison to larger, more efficiently priced firms.

Two additional control variables are included in the regressions reported

in Table VI. The dummy variable, Repeat, is assigned a value of one if the

company announced a repurchase within the past 15 months. Our sample

contains quite a few firms with consecutive program announcements, in-

cluding Seagrams, for example. It is possible that managers in these com-

panies are behaving opportunistically ~Ikenberry and Vermaelen ~1996!! and

will attempt to buy shares only if prices fall. Thus, lower completion rates

are expected from such bargain-hunter-like companies. A similar rationale

2388

The Journal of Finance

Table VI

Determinants of Program Completion Rates

This table reports coefficients for regressions of the program completion rate as a function of

firm characteristics and stock price performance for all TSE-listed firms in our sample an-

nouncing a share repurchase during the period 1989 to 1997. Year Ϫ1 Excess Return is the

firm’s total return during the program relative to the TSE index return. Book-to-Market Decile

and Size Decile Rank are determined in the prior June. Program size is the fraction of out-

standing common stock authorized for repurchase. Repeat is a dummy variable set to one if the

firm has announced a program within the past 15 months.

Year 1

Excess Return

B0M

Decile Rank

Size

Decile Rank

Program

Size

a

Repeat

~i! 0.2921

Ϫ0.0366

~Ϫ2.56!

~ii! 0.2126

~iii! 0.3174

~iv! 0.2927

~v! 0.3588

~vi! 0.3566

0.0121

~2.95!

Ϫ0.0067

~Ϫ2.03!

Ϫ0.0138

~Ϫ.68!

Ϫ1.6270

~Ϫ3.04!

Ϫ0.0332

~Ϫ2.12!

0.0123

~2.93!

Ϫ0.0055

~Ϫ1.50!

Ϫ0.0146

~Ϫ0.66!

Ϫ2.0236

~Ϫ3.53!

exists for the second control variable, Program Size, the percentage of out-

standing shares authorized for repurchase. Some firms may authorize pro-

gram amounts than are greater than they initially intend to buy, completing

the full program only if prices fall. Thus, we expect to find lower completion

rates for large authorizations.

The regression evidence reported in Table VI is generally consistent with

the undervaluation story. The coefficient for Year 1 Excess Return is nega-

tive, suggesting greater repurchase activity as prices fall. The impact of book-

to-market is positive as expected, even after controlling for other factors.

Moreover, given that the unconditional completion rate is 28.6 percent, the

impact of book-to-market is material. For example, a move from the lowest

to the highest book-to-market decile changes the estimated completion rate

by 10.9 percent ~0.0121 ϫ 9!. Similarly, size is also important: a move from

the highest to lowest size decile rank increases the expected completion rate

by 6.0 percent ~Ϫ0.0067 ϫ Ϫ9!. These results suggest that completion rates

depend on whether managers perceive their stock to be undervalued. The

results also suggest that companies with repeat announcements and those

that authorized large programs end up buying less.

To further investigate the possibility of strategic repurchase behavior by

managers, we focus on month to month repurchase activity. Such data allow

Stock Repurchases in Canada

2389

us to assess the extent to which managers trade in response to price move-

ments before, during, and subsequent to a given month. Thus, we estimate

equation ~3!:

Bought_{i, t}ϭ a ϩ b₁Repeat_iϩ b₂Program Size_iϩ b₃Size Rank_i

ϩ b₄B0M Rank_iϩ b₅Excess Return_{i; tϪ1, tϪk}

~3!

ϩ b₆Excess Return_{i; t}ϩ b₇Excess Return_{i; tϩ1, tϩk}

where Bought_{i, t}is the portion of the program firm i repurchased in month t.

Our key variables of interest are price changes in the current month t mea-

sured relative to the TSE 300 Index, Excess Return_{i; t}, and similarly, excess

returns preceding and following month t, Excess Return_{i; tϪ1, tϪk}and Excess

Return_{i; tϩ1, tϩk}, respectively. We consider three horizons for these leading

and trailing windows: one month, three months, and six months. As in Table VI,

we control for size, book-to-market, program size, and repeat.¹²

Evidence is reported in Table VII for all months in the program pooled

together and also for the first half and second half of the program. Overall,

managers trade strategically in response to market movements; as prices

fall, managers respond by buying more stock. This is true for price move-

ments both in the current month and in preceding months. In addition, these

regression coefficients are highly significant. Regression coefficients for each

of the forward excess return horizons are essentially zero, indicating the

difficulty in predicting future stock prices over relatively short horizons.

Managers respond differently to stock price movements in the first half

versus the second half of the program. Generally speaking, response coeffi-

cients for both concurrent and past excess returns in months 7 to 12 are

double that observed in months 0 to 6. Moreover, coefficient estimates from

this later period are quite material. For example, the expected increase in

the monthly completion rate for a company with an excess market return of

Ϫ50 percent over the previous three months and a concurrent excess monthly

return of Ϫ20 percent is roughly one percent. Given that the unconditional

repurchase rate in months 7 to 12 is roughly 1.6 percent per month, such a

move in stock prices translates to roughly a 65 percent increase in the rate

of repurchase activity. One reason for this increased sensitivity is because

early in the program, the possibility to postpone repurchases to later months

exists. Later in the program, this option begins to expire. As a result, re-

purchase activity becomes more sensitive to price movements.

Overall, these results point to underpricing as a motivating factor for man-

agers to repurchase shares. If nonvaluation reasons had a predominant ef-

fect on repurchase activity, one would not expect to see such behavior. For

example, if managers thought that the market efficiently priced their stock

12

Months where a firm had already purchased 90 percent or more of its authorized program

are eliminated from this analysis, as these programs are essentially completed.

Table VII

Determinants of Monthly Trading Activity during Stock Repurchases

This table reports regression results for the following model:

Bought_{i, t}ϭ a ϩ b₁Repeat_iϩ b₂Program Size_iϩ b₃Size Decile Rank_iϩ b₄B0M Decile Rank_i

ϩ b₅Excess Return_{i; tϪ1, tϪk}ϩ b₆Excess Return_{i; t}ϩ b₇Excess Return_{i; tϩ1, tϩk}

where Bought_{i, t}is the fraction ~ranging between 0 and 1! of firm i repurchased in month t relative to the total program authorization. The first four control variables

are defined the same as in Table VI. The last three terms relate to firm i’s excess total return over a given period preceding, coincident with, and following month

t. Excess return is measured relative to the TSE 300 Index over the same period. Months where firms have completed 90 percent or more of the program’s stated

amount are deleted. Panel A reports evidence pooling all months and all firms in our sample. Panel B reports evidence using only months 0 to 6 relative to the

announcement. Panel C reports evidence for months 7 through 12. t-statistics are in parentheses.

b₁

b₂

b₃

b₄

b₅

b₆

b₇

a

Repeat

Program Size

Size Decile

B0M Decile

Excess Return_{tϪ1, tϪk}

Excess Return_t

Excess Return_{tϩ1, tϩk}

Panel A: All Months

k ϭ 1 month

k ϭ 3 months

k ϭ 6 months

0.0265

0.0263

0.0266

Ϫ0.0016

~Ϫ1.51!

Ϫ0.0015

~Ϫ1.34!

Ϫ0.0015

~Ϫ1.38!

Ϫ0.1515

~Ϫ5.31!

Ϫ0.1508

~Ϫ5.31!

Ϫ0.1652

~Ϫ5.65!

Ϫ0.0003

0.0010

~4.72!

0.0010

~4.80!

0.0010

~4.76!

Ϫ0.0174

~Ϫ3.97!

Ϫ0.0095

~Ϫ4.00!

Ϫ0.0057

~Ϫ3.49!

Ϫ0.0155

~Ϫ3.56!

Ϫ0.0148

~Ϫ3.41!

Ϫ0.0168

~Ϫ3.70!

Ϫ0.0083

~Ϫ1.91!

0.0006

~1.26!

Ϫ0.0004

~Ϫ0.23!

~Ϫ1.19!

Ϫ0.0003

~Ϫ3.30!

Ϫ0.0002

~Ϫ0.94!

Panel B: Months 0 to 6

k ϭ 1 month

k ϭ 3 months

k ϭ 6 months

0.0344

0.0340

0.0346

Ϫ0.0043

~Ϫ2.63!

Ϫ0.0041

~Ϫ2.50!

Ϫ0.0031

~Ϫ1.31!

Ϫ0.1912

~Ϫ4.46!

Ϫ0.1937

~Ϫ4.53!

Ϫ0.1967

~Ϫ4.55!

Ϫ0.0003

0.0010

~3.32!

0.0011

~3.40!

0.0010

~3.12!

Ϫ0.0186

~Ϫ2.92!

Ϫ0.0077

~Ϫ2.32!

Ϫ0.0031

~Ϫ1.33!

Ϫ0.0118

~Ϫ1.85!

Ϫ0.0112

~Ϫ1.76!

Ϫ0.0110

~Ϫ1.71!

Ϫ0.0041

~Ϫ0.63!

Ϫ0.0008

~Ϫ0.74!

Ϫ0.0026

~Ϫ1.05!

Ϫ.0003

~Ϫ0.96!

Ϫ0.0003

~Ϫ1.10!

Panel C: Months 7 to 12

k ϭ 1 month

k ϭ 3 months

k ϭ 6 months

0.0165

0.0164

0.0019

~1.44!

0.0020

~1.53!

0.0018

~1.34!

Ϫ0.1018

~Ϫ2.93!

Ϫ0.0999

~Ϫ2.89!

Ϫ0.1260

~Ϫ3.52!

Ϫ0.0001

0.0009

~3.60!

0.0009

~3.59!

0.0010

~3.77!

Ϫ0.0167

~Ϫ2.95!

Ϫ0.0121

~Ϫ3.81!

Ϫ0.0078

~Ϫ3.80!

Ϫ0.0225

~Ϫ4.06!

Ϫ0.0209

~Ϫ3.80!

Ϫ0.0284

~Ϫ4.74!

Ϫ0.0145

~Ϫ2.73!

0.0004

~0.92!

0.0026

~1.23!

~Ϫ0.57!

Ϫ0.0001

~Ϫ0.64!

0.0000

~0.00!

Stock Repurchases in Canada

2391

and were using repurchases as means of returning capital to shareholders in

lieu of a dividend, we would not expect to see repurchase activity linked to

market movements. However, this is not the case. We see rather substantial

swings in repurchase levels associated with past and current market move-

ments. These results suggest that managers are indeed cognizant of market

prices and execute trades accordingly.

VI. Long-Term Performance Conditional on Program Completion

In this section, we further investigate the role of undervaluation and the

extent to which managers strategically repurchase stock by examining per-

formance using the three-factor model conditional on completion rates. If

managers are repurchasing strategically, we expect them to buy more shares

as prices decline. Similarly, we expect them to repurchases fewer shares as

prices rise and mispricing dissipates. Thus, in Table VIII, we report evi-

dence for three scenarios. First, we consider cases where firms repurchased

no shares at all after authorizing a program. The remaining programs are

sorted into two groups: those cases where managers bought less than 30 per-

cent versus those where they were more aggressive and bought more than

30 percent of the authorized amount.

Focusing on abnormal performance in year 1, noticeable differences are

observed across the three groups. For cases where no shares were repur-

chased, abnormal performance is 0.96 percent per month ~t ϭ 2.99!, or nearly

12 percent per year. Yet, abnormal performance in year 1 decreases as we

move to cases where firms were active buyers. For those completing more

than 30 percent of their program, the abnormal monthly return is 0.22 per-

cent per month ~t ϭ 1.07!.

Interestingly, after the programs have expired, there is no evidence of

abnormal performance in cases where no shares were repurchased. Yet, in

cases where companies did repurchase shares, substantial abnormal returns

are observed in years 2 and 3 after the program is over. This pattern in

returns is consistent with evidence reported earlier and suggests that man-

agers are sensitive to stock prices.

In Table IX, we provide one last piece of evidence of strategic trading

behavior by focusing on trading during the last half of the repurchase pro-

gram. We partition our sample into two groups: companies that bought shares

in the last six months of the program versus companies that had no repur-

chase activity during this period.¹³For cases where firms were active buyers

late in the program, we expect to see relatively low return performance dur-

ing months 7 to 12, followed by high abnormal returns in years 2 to 3. Con-

13

Here, only companies that bought less than 15 percent of their total program by month 6

are included. This allows us to focus on cases where the vast majority of the program remains

available for repurchase during the last half of the program. Sample sizes for the repurchase

and no repurchase groups are 259 and 399, respectively.

Table VIII

Long-run Performance Conditional on Completion Rate

Portfolios are formed in calendar time for TSE-listed firms that announced an open market stock repurchase program conditional on the extent

to which their programs were completed. Firms are sorted into three portfolios: those that purchased no shares, those that repurchased some but

less than 30 percent of the authorized amount, and those that repurchased more than 30 percent of the authorized amount. Excess performance

~in percentages! is measured using a three-factor model: