CONTROL PREMIUMS by Alexander Dyck, AND THE...

CONTROL PREMIUMSAND THE EEEECTIVENESSOE CORPORATEGOVERNANCE SYSTEMS

by Alexander Dyck,University of Toronto, andLuigi Zingales, University of Chicago*

he last decade has witnessed a tremendous increase of interest in

corporate governance, which in turn has triggered a search for the

key elements of an effective governance system. To evaluate the

effectiveness of different corporate governance systems, it is necessary to

develop an objective measure of effective corporate governance. The main

gauge used by most economists and policy makers—the size of a nation's equity

market—is generally considered too indirect to provide a useful guide for

analysis or reform. In this paper, we begin by summarizing the findings of our

recently published research that presents a more direct measure of the success

of corporate governance systems. Then we use this newly constructed measure

as a basis for identifying the most important elements of a well-functioning

national governance system.

One useful definition of corporate governance is "the ways in which

suppliers of finance to corporations assure themselves of getting a return on their

investment."^ If we adopt this definition, then one measure of the shortcomings

of a corporate governance system is the proportion of a company's value that

does not accrue to all shareholders on a per share basis, but is instead captured

by inside shareholders who control and sometimes manage the firm. Economists

refer to this extra value captured by insiders as the "private benefits of control."

This name reflects the reality that effective control of a corporation confers the

opportunity not only to improve performance and increase value, but also to

divert wealth away from shareholders and other groups in favor of the

controlling coalition.

• This article draws on and summarizes the findings of "l'ir.'L:te Benefits of Control: An International Comparison," whichwas published in the Journal of Finance earlier this year (i. '" 0

i. For a l^roader definition, see Zingales (1998). Full citations for all studies mentioned in the text are supplied in thebibliography at the end of the article.

51STERN STEWART • JOURNAL OF APPUED CORPORATE FINANCE

To illustrate the concept of private controlbenefits, take the case of company (or individual) Athat buys a controlling interest in company B. Oneway for A to realize private benefits is simply totransfer value—say, by getting B to sell its productto A at below-market prices or getting B to buy inputsfrom A at above-market prices (whether or not suchbehavior is legal and the likelihood that if illegal itwill be sanctioned depends on the domicile of bothcompanies A and B). Another possibility, which doesnot involve an actual transfer of wealth, is forcompany A to profit from an investment opportunitythat it learns about as a direct consequence of itsinvolvement with B (and that might have instead beenundertaken by B).

This kind of wealth diversion reduces whatminority shareholders are willing to pay for shares,lowering the value of all companies where suchbehavior represents a real possibility. And by raisingthe cost of finance, it limits the ability of such firms tofund attractive investment projects. But how can wetell how much of this activity goes on within a nationaleconomy? By their nature, such private control ben-efits are difficult to observe and even more difficult toquantify in a reliable way. (After all, if such valuetransfers could be observed and quantified, it wouldbe relatively easy, at least where such transfers areillegal, for non-controlling shareholders to stop them.)

Financial economists have come up with twomethods of estimating the size of private benefits. Oneis to look at differences across firms and among nationsin the size of premiums paid for voting stock relativeto non-voting stock.̂ If superior voting stock tradesat a premium while having equal or inferior dividendrights, this means that control is valuable. But suchcontrol should command additional value only ifcontrolling shareholders expect to receive somebenefits not available to minority shareholders. Andthus large premiums can be viewed as an indicationof inadequate protections for minority shareholdersand a weak corporate governance system.

Cross-country studies of voting premiums sug-gest widespread weakness in corporate governanceas well as significant variation among countries. Theearliest, single-country studies concluded that voting

rights tend to be worth 10% to 20%, on average, ofthe value of common stock, with countries such asthe United States (5.4%), Sweden (6.5%), and theUnited Kingdom (13.3%) reporting relatively lowpremiums, moderate premiums in Switzerland (20.0%)and Canada (23.3%), and much higher premiums incountries like Israel (45.5%) and Italy (82.0%).̂ Amore recent study of 18 countries by Nenova (2001),using a common methodology and for a commonyear, reported an average premium across all coun-tries of 13%, with low levels in countries like theUnited States (2%) and Sweden (1%) and muchhigher premiums in countries like Italy (29.4%) andMexico (36.4%). (These high average premiums, aswell as the variation among countries, were evidenteven when the studies controlled for possible inter-country differences in economic rights and liquidityacross voting and non-voting shares.)

But this method of evaluating corporate gover-nance systems clearly has limitations. One concemis the possibility that companies that choose to havemultiple classes of stock are fundamentally differentfrom other publicly traded companies in thosecountries. But perhaps even more limiting is the factthat the method cannot be applied to the governancesystems of the many nations that prohibit multipleclasses of stock.

The second method of estimating private con-trol benefits, which was pioneered in a 1989 studyby Michael Barclay and Cliff Holderness, focuses ondifferences in the price per share paid in a privatelynegotiated transfer of a controlling block and theprice that can be observed in the market onceinvestors have absorbed the fact that there will be anew controlling shareholder. The rationale for thismethod is that the price per share an acquirer paysfor the controlling block reflects both the cash flowbenefits it expects to receive as a shareholder (whichinclude the value of any improvements it expects tomake in the firm's performance) and any privatebenefits stemming from its controlling position. Bycontrast, the market price of a share just after thechange in control is announced should reflect onlythe cash flow benefits that all shareholders expect toreceive under the new management.

2. See, generally, L. Zingales, "The Value of the Voting Right: A Study of theMilan Stock Exchange Experience," Rev. Fin. Stud., Vol. 7 (Spring 1994).

3. See R. Lease, J. McConnell, and W. Mikkelson, "The Market Value ofControl in Publicly Traded Corporations," Journal of Financial Economics, Vol.7 (1983); K. Rydqvist, "Takeover Bids and the Relative Prices of Shares That Differin Their Voting Rights," Stockholm School of Economics Working Paper (1992);

W. Megginson, "Restricted Voting Stock, Acquisition Premiums, and the MarketValue of Corporate Control," The Fin. Rev., Vol. 25 (1990); Homer, "The Value ofthe Corporate Voting Right,"/ Banking&Fin., Vol. 12 (1988); Robinson and White,"The Value of a Vote in the Market for Corporate Control," York University WorkingPaper (1990); Levy, "Economic Evaluation of Voting Power in Common Stock,"Journal of Finance, Vol. 38 (1982); and Zingales (1995b).

52JOURNAL OF APPUED CORPORATE FINANCE

Hence, as Barclay and Holderness have argued,the difference between the price per share paid bythe acquiring party and the price per share prevailingon the market should reflect the differential payoffaccruing to the controlling shareholder. And aftersome adjustments (discussed below), this differencecan be used as a measure of the private benefits forthe controlling shareholder.

In a study that was published this year in theJournal of Finance, we used the Barclay and Holdemessmethod to infer the value of private benefits of controlin companies representing 39 different countries.Based on 393 control transactions that took placebetween 1990 and 2000, we found that controlpremiums averaged 14% of the equity value of a firm.As in other studies, we found considerable cross-country variation in the premiums, with 14 countrieshaving average premiums of less than 3% and 10countries with premiums of 25% or more (includingBrazil, with an average premium of 65%).

After performing a series of test to convinceourselves that these estimates were in fact likely tobe measures of private benefits, we then used thismeasure to explore the following question: Whatfactors within a national economy work to limit suchwealth transfers and thereby strengthen its corporategovernance system? Perhaps the most commonfactor associated with lower private benefits is thepresence of laws that protect minority shareholdersfrom abuses by insiders. In fact, the extent of suchlegal protections is often taken to be the measure ofthe strength of a governance system. But althoughour study investigated the explanatory power ofcorporate law, our investigation was not limited tolegal factors. Given significant gaps between thelaws on a country's books and what takes place inpractice, most systems develop less formal, non-contractual ways of limiting self-dealing by insiders.In exploring this possibility, our study considered anumber of extralegal mechanisms and institutionsthat have been proposed by scholars as controllingprivate benefits—notably product market competi-tion, labor pressures, and moral norms. To thesewell-known mechanisms, we added two of our own:public opinion and corporate tax enforcement. Wehypothesized that in countries with many indepen-dent communications media, particularly a thrivingbusiness press, the desire of insiders to maintain theirpublic reputation and of organizations to maintain a

reputation for fair treatment of minority stockholdersshould act to limit abuses by controlling sharehold-ers.'* We also reasoned that, for most companies,governments represent in effect the largest minorityshareholder as a result of their corporate income taxclaim; and that the government and minority share-holders have a common interest in preventing insidersfrom "tunneling" income out of the firm. In this sense,the ability of a government to ensure compliance withtax claims can be viewed as a measure of theeffectiveness of this additional monitor of corporateaffairs. And, as we discuss below, our study showedthat an independent and widely circulating press,high rates of tax compliance, and a high degree ofproduct market competition are all associated withlower private benefits of control.

HOW TO MEASURE PRIVATE BENEETTS

Consider a company whose stock trades at $8per share. One day a large block—say, 40%—changes hands (outside of the stock market) at aprice of $14 per share. The difference between $8and $14 can be broken down into two components:(1) the increase in the value of the company ex-pected under the new management team and (2) theprivate benefits the new shareholder expects from itscontrol of the company. By observing a market pricethe day after the transfer of control has taken place—which gives the market the opportunity to assess thestrength of the new management team—^we canseparate the expected increase in company valuefrom the expected private benefit of control.

To illustrate, let's say that after the transfer hasbeen announced, the stock jumps from $8 to $10.The $2 increase refiects outsiders' expectation ofimproved corporate performance, presumably re-sulting from more effective oversight or better man-agement by the buying firm (or individual). But whataccounts for the difference between $14 and $10? Insome cases, the $4 difference will refiect buyers'greater confidence in, or better information about,their own ability to improve the firm's performance.But on average—and provided the market does agood job of capturing expected gains in its near-termprice reactions—the new controlling shareholdersmust expect to receive the additional $4 in privatebenefits—that is, benefits that do not accrue tooutside minority shareholders (and may in fact come

4. Zingales (2000), Dyck and Zingales (2002).

53VOLUME 16 NUMBERS 2-3 • SPRING/SUMMER 2004

TABLE I • TRANSACTION CHARACTERISTICSThis taljle presents the summary statistics of the specific characteristics of all the 412 transactions analyzed in this paper.

VARIABLENumber of

Observations Mean Std. Dev. Min Max

The difference between the security value of the buyer(market price at t+2) and of the seller (market price at t-30)normalized by the market price at t+2

Buyers proportion of change in security value

Initial shareholding as a percent of total shares

Size of control block as a percent of total shares

Fixed assets as a percentage of total assets (U.S. 3-digit SIC)

Indicator Variables

412

412

412

412

412

0.06

0.02

1.31

37.71

0.29

0.24

0.09

3.72

17.51

0.23

-1.84

-0.43

0.00

6.00

0.00

0.90

0.40

20.00

90.00

0.77

Transactions with blocks greater than 20%

Transactions with anotlier large shareholder

Financial distress in target in year of transaction (EPS<=0)

Financial distress in target in year before transaction (EPS<=0)

Domestic acquirer

Control block created by issuing new shares

412

412

412

412

412

412

0.27

0.16

0.26

0.22

0.62

0.16

Agriculture, forestry, & fishing (SIC 01-09)

Mining (SIC 10-14)

Construction (SIC 15-17)

Manufacturing (SIC 20-39)

Transportation & pub. utilities (SIC 40-49)

Wholesale trade (SIC 50-51)

Retail trade (SIC 52-59)

Finance, insurance, & real estate (SIC 60-67)

Services (SIC 70-89)

Seller an individral (news stories)

Seller government (news stories)

Seller unidentified (news stories)

Seller public company (news stories)

Acquirer public company (SDC)

Acquirer subsidiary (SDC)

Acquirer private (SDC)

Acquirer government (SDC)

Acquirer unknown (SDC)

412

412

412

412

412

412

412

412

412

412

412

412

412

412

412

412

412

412

0.05

0.04

0.01

0.39

0.09

0.03

0.06

0.23

0.10

0.18

0.03

0.08

0.54

0.42

0.13

0.40

0.01

0.03

54JOURNAL OF APPLIED CORPORATE FINANCE

at their expense). Since this premium has been paidonly on 40% of the stock, we normalize it by the totalvalue of the company [0.4*((l4-10)-10) = 0.16].Hence, in this case we estimate the value of controlto be 16% of the value of the company. (Note that thecontrol premium can also be negative even if thenew buyer pays a premium over the existing stockprice; this would happen if the price on the exchangerises above the negotiated price of the block sale.)

DATA AND DESCRIPTIVE STATISTICS

One limitation of this method is that it can beapplied only to publicly traded companies thatexperience a transfer of a controlling block. Inperfortning our study of such transactions, westarted by compiling a list of all control blocktransactions reported by Securities Data Corporationfor the period 1990-2000. After imposing a set ofcriteria to ensure homogeneity, we were left with asample of 393 observations from 39 countries.^

Table I presents the characteristics of the dealsin our sample, while Table II reports descriptivestatistics of the block premiums from our sarnpleaccording to the country in which the acquired finnis located. The sample includes more than 40obsei^vations each from active equity markets such asthe United Kingdom and the United States. For somecountries, we have relatively few observations as aresult of the combination of weak coverage byDatastream, few reported prices for control sales,and limited observability of control premiums as aresult of laws regulating tender offers in the case ofcontrol sales. The rank ordering of countries bycontrol premium is very similar using mean andmedian values, suggesting that our results are notdriven by a few outliers.

The first column of Table II presents the averagecontrol premium by countiy, computed as thecoefficient of fixed coimtry effects in a regressionwhere the dependent variable is the control pre-mium as defined in section I. Overall, the averagecontrol premiutn was 14% when each country wasgiven an equal weighting, but only 10% when eachtransaction received equal weight (which means thatcountries with smaller control premiums tend tohave more transactions). In 10 of our 39 samplecountries, the average control premium exceeded

25% of equity value. The countries in our samplewith the highest private benefits (bad governance)were Argentina, Austria, Colombia, the Czech Re-public, Israel, Italy, Mexico, Turkey, Venezuela, andBrazil, which had the highest estimated value of 65%.At the other extreme were 14 countries for whichprivate benefits were estimated at 3% or less. Thesecountries were Australia, Canada, Finland, France,Hong Kong, Japan, Netherlands, New Zealand,Norway, Singapore, South Africa, Taiwan, the UnitedKingdotn, and the United States.

Of course, control premiums differ across firmsfor a variety of reasons related to specifics of the dealand characteristics of the firm. If such variation weremore pronounced in some countries than others, thiswould tnake country averages misleading indicatorsof the underlying strength or weakness of nationalgovernance systems. To address this concern wenext produced revised estimates of how controlpremiums differ across countries that were based ona regression that controlled for company and trans-action characteristics.

(1) Differences in the Extent to Which theBlock Conveys Control

First of all, in calculating our initial controlpremiums, we assumed that all transactions in oursample transfer absolute control. But this is probablyincorrect because the transfer of, say, a 20% blockdoes not cany the same amount of control as thetransfer of a 51% block. Similarly, the transfer of a30% block when there is another shareholder con-trolling 20% likely carries less control than thetransfer of the same block when the rest of the sharesare dispersed. Thus, control blocks above 50% arelikely to fetch a higher price. Similarly, the presenceof another large shareholder (a stake in excess of20%) should reduce the premium.^

As shown in Table III (column 2), for our sampleof 393 transactions, an absolute majority of votes, allother things equal, had the effect of increasing thevalue of a controlling block by 9.5% of the total valueof equity (with statistical significance at the 5% level).But contrary to our prediction, the presence ofanother large shareholder also had a small positive(though not statistically significant) effect on thecontrol premium.

5. For a complete descriptions of the criteria imposed and their justification,see Dyck and Zingales (200'i).

6. In Canada and Australia, we used 15% since exceeding 20% would triggera mandatory offer for remaining shares.


TABLE n a BLOCK

Country

Argentina

Australia

Austria

Brazil

Canada

Chile

Colombia

Czech Republic

Denmark

Egypt

' Finland

France

Germany

Hong Kong

Indonesia

Israel

• Italy

Japan

Malaysia

Mexico

Nethierlands

New Zealand

Norway

Peru

: Philippines

Poland

Portugal

Singapore

South Africa

South Korea

Spain

Sweden

Switzerland

Taiwan

Thaiiand

Turkey

; United Kingdom

United States

' Venezuela

Average/Number

PREMIUM AS

Mean

0.27

0.02

0.38

0.65

0.01

0.15

0.27

0.58

0.08

0.04

0.02

0.02

0.10

0.01

0.07

0.27

0.37

-0.04

0.07

0.34

0.02

0.03

0.01

0.14

"̂ 0.13

0.11

" " 0.20

0.03

~ 0.02

0.16

0.04

0.06

0.06

0.00

0.12

0.30

0.02

0.02

0.27

0.14

PERCENT OF

Median

0.12

0.01

0.38

0.49

0.01

0.12

0.15

0.35

0.04

0.04

0.01

0.01

0.10

0.03

0.07

0.21

0.16

-0.01

0.05

0.47

0.03

0.03

0.01

0.17

0.08

0.08

0.20

0.03

0.00

0.17

" 0.02

0.02

0.07

0.00

0.07

0.09

0.01

0.02

0.28

0.11

FIRM EQUITY

StandardDeviation

0.26

0.04

0.19

0.83

0.04

0.18

0.34

0.80

0.11

0.05

0.06

0.10

0.13

0.05

0.03

0.32

0.57

0.09

0.10

0.35

0.05

0.09

0.05

0.11

0.32

0.11

0.14

0.03

0^03

0.07

0.06

0.08

0.04

0.01

0.19

0.55

0.05

0.10

0.21

0.18

Minimum

0.05

-0.03

0.25

0.06

-0.02

-0.08

0.06

0.01

-0.01

0.01

-0.07

-0.10

-0.24

-0.12

0.05

-0.01

-0.09

-0.34

-0.08

-0.04

-0.07

-0.17

-0.05

0.03

-0.40

0.02

0.11

-0.01

0.00

0.04

-0.03

-0.01

0.01

-0.01

-0.08

-0.03

-0.06

-0.20

0.04

-0.04

Maximum

0.66

0.11

0.52

2.99

0.06

0.51

0.87

2.17

0.26

0.07

0.13

0.17

032

0.05

0.09

0.89

1.64

0.09

0.39

0.77

0.06

0.18

0.13

0.23

0.82

0.28

0.30

0.06

0.07

0.22

0.13

0.22

0.15

0.00

0.64

1.41

0.17

0.40

0.47

0.48

Number ofObservations

5"

13

2

11

4 "

9

y 56

52

14

518

92

98

21

41 ' "

5

5

19

14

3

15

5

2

44

6

5

138

312

6

^ 3

47

4

412

Number ofPositive

Observations

" 5 \

9

2

11

2

85 '

6

32

93

15

72

8

75

31 1

4

4

14

9

311

5

2

3

2

6

4

12

8

2

11

523

28

4

300

56JOURNAL OF APPUED CORPORATE RNANCE

(2) Differences in the Extent of the Seller'sBargaining Power

In estimating the private benefits of control, weassumed that the seller's bargaining power is con-stant across all deals. But in cases where sellers havegreater-than-average bargaining power, control pre-miums should be higher. And the converse is alsotrue—namely, that for a given amount of privatebenefits of control, the lower the seller's bargainingpower, the lower our estimates of control premiums.

We tried to control for these differences inbargaining power with three proxies:

First, if the selling company is in financialdistress, it is more likely to be forced to sell and henceto have less bargaining power. As a proxy forfinancial distress, our study created a dummy vari-able that takes the value 1 when earnings per shareare zero or negative in the year of the block trade orthe year preceding the block trade.^ As expected,companies in financial distress exhibited a controlpremium that is 5.4 percentage points lower thanotherwise identical companies that are not.

Similarly, when the acquisition of a controllingblock takes the form of an equity infusion this probablyindicates that a company needs to raise equity, a signof a weak bargaining position. We inserted a dummyif the block was formed by newly issued equity.Contrary to expectations, the creation of a blockthrough a new equity offering had a positive (thoughnot statistically significant) effect on the premium.

This method was particularly common in Japan,where in a majority of cases control was transferredby financially distressed companies through privateplacements of new equity. The concentration of suchtransactions in Japan underscores the importance ofcontrolling for company and deal characteristicsrather than focusing on the raw averages acrosscountries.

Finally, foreign acquirers generally face morecompetition, at the very least because their involve-ment implies that the transactions are open to foreignbuyers and thus there is a larger pool of potentialacquirers. Thus, the bargaining power of the sellerin these transactions is likely to be bigger. We findthat foreign buyers pay a premium of 6.9% (statisti-cally significant at the 5% level).

(3) Cross-Listing in the United States

Finance and legal scholars have argued thatforeign companies that list in the U.S. exchangesthereby submit themselves to tougher governancerules and limit their own ability to extract privatebenefits.^ Since we wanted to capture how "typical"firms across different countries compare in theirgovernance outcomes, and firms that cross list areunlikely to be "typical," we controlled for the pres-ence of cross-listing. For this purpose we inserted adummy variable equal to one for any selling com-pany that is cross-listed in the United States as wellas in its home market.^ As expected, cross-listedcompanies had lower private benefits (althoughgiven the dearth of dually listed companies in oursample (23), the statistical significance of this effectis below conventional levels).

(4) Adjusted Estimates of Private BenefitsControlling for Differences in Deal and FirmCharacteristics

After inserting all these deal and companycharacteristics in our basic regression, we re-estimated the country fixed effects. The results arereported at the bottom of column 2 in Table III. Sincemany of the control variables included capture partof the value of control, the country fixed effects canno longer be interpreted as the estimates of theaverage value of private benefits in that country, butonly as relative rankings. Including these controlsdramatically lowers the ranking for countries charac-terized by the higher-than-average presence of for-eign acquirers and sales of majority stakes likeGermany, Switzerland, Egypt, and Poland.

On the one hand, these estimates represent animprovement over our raw data because they adjustfor deal characteristics. But they suffer from aneconometric problem. To estimate the impact ofthese deal and firm characteristics, we had to assumethat this impact is constant across countries. In somecases this assumption might be untenable. Forexample, the difference between acquiring a 51%stake rather than a 30% one might be huge in acountry where private benefits of control are large,but it might be small or even irrelevant in a country

7. While other measures of cash flow are preferable, eamings per share is oneof the few data items consistently reported in Datastream for the companies in ourdatabase.

8. Coffee (1999), Reese and Weisbach (2001), and Doidge et al. (2001).9. We obtained the list of cross listing from Doidge et al. (2001). We thank

Andrew Karolyi for kindly providing us with the data.


TABLE m • ES'nMATED BLOCK PREMIA BY COUNTRYThe dependent variable is the Ijlock premium as a percent of firm equity. Each regression includes countiy fixed effects. In addition,in column (2) we introduce several deal characteristics: whether it is a majority block, whether there is another large shareholder,whether the firm is in financial distress, whether the block was created by issLiing new shares, whether the buyer is a foreigner,and if the finns' shares are cross-listed in the United States. In column (3) we introduce several industiy and seller/buyercharacteristics: identity ofthe buyer (individual, government, subsidiary, dispersed), identity ofthe seller (individual, government,unknown), 2-digit SIC code industry dummies, and the proportion of fixed to total assets. In column (4) we introduce in additioninteractions between foreign acquirer and cross-listing dummy and the difference in governance rules between acquiring coLintiy(or U.S. for cross listed-firms) and target country. All regressions are estimated by OLS. Robust standard errors are in parentheses.

INDEPENDENT VARIABLES DEPENDENT VARIABLE: BLOCK PREMIUM

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

Buyer's proportion of change in security value

Stake greater than 50%

Another large shareholder

Einancial distress in selling firm

Sold through new share issue

Buyer is foreign

Cross-listed in the U.S.Interaction of relative strength of anti-directorrights (home-target nation) and foreign acquirer

Interaction of relative strength of anti-director

rights (U.S.-target nation) and cross-listed in the U.S.

Buyer individual or private

Buyer government

Buyer subsidiary

Buyer dispersed or unknown

Seller individual

Seller government

Seller unknown

Fixed assets as percent of total

Industiy-Agriculture, forestry, fishing

Industry-Mining

Industry-Construction

Industry-Transportation & utilities

Industry-Wholesale tr-ade

Indirstry-Retail trade

Indirstry-Einance, insur-ance, r'eal est.

Industry-Services

0.323

0.095"

0.041

0.054"

0.041

0.069"

0.062

(0.211)

(0.039)

(0.043)

(0.028)

(0.057)

(0.034)

(0.040)

-0.319

0.095"

0.018

-0.043

0.034

0.065'

-0.067"'

-0.042

0.008

-0.001

-0.039

0.021

0.008

0.028

-0.097

-0.03

-0.071

-0.027

0.066''

0.046

-0.057

0.055

-0.024

(0.209)

(0.039)

(0.040)

(0.028)

(0.059)

(0.036)

(0.039)

(0.026)

(0.046)

(0.049)

(0.044)

(0.029)

(0.100)

(0.031)

(0.062)

(0.050)

(0.071)

(0.042)

(0.031)

(0.047)

(0.055)

(0.045)

(0.038)

-0.336

0.087"

0.02

-0.044

0.026

0.06"

0.113

-0.029"*

-0.07'"'

-0.045°

0.011

0.013

-0.031

0.024

-0.008

0.026

-0.09

-0.027

-0.068

-0.058

0.064'""

0.037

-0.046

0.054

-0.012

[0.211]

[0.039]

[0.039]

[0.028]

[0.058]

[0.036]

[0.083]

[0.011]

[0.034][0.026]

[0.049]

[0.049]

[0.043]

[0.029]

[0.097]

[0.031]

[0.062]

[0.050]

[0.072]

[0.071]

[0.032]

[0.046]

[0.055]

[0.044]

[0.038]

COUNTRY FIXED EFFECTS

Argentina

Australia

Airstria

Br-azil

0.268"

0.020

0.383'"

0.650""

(0.111)

(0.013)

(0.099)(0.252)

0.158

-0.001

0.318""

0.606"*

(0.131)

(0.034)

(0.054)

(0.229)

0.197

0.051

0.309""0.652""'

(0.123)

(0.052)

(0.050)

(0.245)

0.183

0.052

0.319""°

0.653""'

(0.113)

(0.051)

(0.051)

(0.245)(Continued on following page)

58JOURNAL Ol- AI'PUED CORPORATE FINANCE

(TABLE in • CONT'D.)

COUNTRY FIXED EFFECTS

Canada

Chile

Colombia

Czech Republic

Denmark

Egypt

Finland

France

Germany

Hong Kong

Indonesia

Israel

Italy

Japan

Malaysia

Mexico

Netherlands

New Zealand

Norway

Peat

Philippines

Poland

Portugal

Singapore

South Africa

South Korea

Spain

Sweden

Switzerland

Taiwan

Thailand

Turkey

United Kingdom

United States

Venezuela

Number of ObservationsR-Squared

• significant at 10% level; *' significant at

( 1 )

0.013

0.183*"

0.273*

0.578*

0.077

0.038

0.025

0.019

0.095*"

0.003

0.072***

0.270**

0.369*

-0.043"

0.072'"

0.345"

0.016

0.027

0.015

0.142***

0.129

0.133*"

0.203"'

0.030*

0.017

0.157*"

0.041

0.074*"

0.063***

-0.004

0.125"

0.371

0.014*

0.01

0.270*"

(0.017)

(0.069)

(0.142)

(0.312)

(0.048)

(0.024)

(0.016)

(0.052)

(0.034)

(0.019)

(0.017)

(0.107)

(0.199)

(0.021)

(0.017)

(0.146)

(0.020)

(0.024)

(0.014)

(0.053)

(0.083)

(0.052)

(0.073)

(0.016)

(0.015)

(0.027)

(0.027)

(0.027)

(0.015)

(0.004)

(0.054)

(0.246)

(0.007)

(0.013)

(0.094)

3930.389

5% level; ••• siignificant at

DEPENDENT VARIABLE:

( 2 )

-0.06

0.149"

0.197

0.462

0.039

-0.050

-0.016

0.040

-0.020

0.045

-0.034

0.238"

0.323*

-0.070

0.063'**0.296"

-0.054

-0.028

0.007

0.067

0.115

0.003

0.159"*

0.024

-0.045

0.086

0.021

0.033

-0.073

-0.047

0.073

0.276

0.000

0.002

0.256**

(0.056)

(0.065)

(0.137)

(0.297)

(0.050)

(0.061)

(0.027)

(0.059)

(0.052)

(0.033)

(0.040)

(0.108)

(0.191)

(0.044)

(0.018)

(0.143)

(0.068)

(0.042)

(0.026)

(0.080)

(0.081)

(0.081)

(0.052)

(0.035)

(0.061)

(0.066)

(0.042)

(0.047)

(0.056)

(0.039)

(0.080)

(0.232)

(0.019)

(0.031)

(0.105)

3930.4

1% level

31

BLOCK PREMIUM

( 3 )

-0.055

0.165"

0.242'

0.555*

0.036

0.025

-0.010

0.080

0.016

0.040

0.043

0.259"

0.311

-0.038

0.093'"

0.322**

-0.015

0.026

0.052

0.060

0.142*

0.041

0.197***

0.042

0.005

0.088

0.047

0.041

-0.067

-0.040

0.121

0.346

0.040

0.044

0.221"

(0.075)

(0.067)

(0.132)

(0.325)

(0.070)

(0.082)

(0.036)

(0.077)

(0.059)

(0.044)

(0.047)

(0.114)

(0.192)

(0.054)

(0.032)

(0.144)

(0.060)

(0.046)

(0.041)

(0.082)

(0.079)

(0.092)

(0.059)

(0.069)

(0.072)

(0.086)

(0.058)

(0.057)

(0.074)

(0.074)

(0.084)

(0.249)

(0.033)

(0.038)

(0.112) "

3930.459

( 4 )

-0.052

0.16**

0.325""

0.563*

0.027

0.112

0.002

0.084

0.041

0.008

0.043

0.252"

0.349'

-0.039

0.089"°

0.396*"

-0.015

0.028

0.061

0.08

0.148*

0.039

0.207°""

0.038

-0.014

0.137*

0.058

0.047

-0.051

-0.038

0.107

0.363

0.02

0.035

0.268*"

(0.083)

(0.065)

(0.128)

(0.330)

(0.065)

(0.093)

(0.037)

(0.078)

(0.058)

(0.048)

(0.045)

(0.116)

(0.199)

(0.051)

(0.033)

(0.133)

(0.062)

(0.045)

(0.043)

(0.075)

(0.080)

(0.092)

(0.059)

(0.062)

(0.074)

(0.081)

(0.052)

(0.056)

(0.073)

(0.073)

(0.080)

(0.246)

(0.033)

(0.038)

(0.119)

3930.470


where the private benefits of control are very tiThe regression, however, attributes the same effectto all the countries, underestimating differencesacross countries.

In the remainder of our study, where weexplored the effects and causes of these cross-country differences, we used a refined measure thatincorporates controls to keep constant deal (andother) characteristics. But recognizing that these dealcharacteristics may not be constant across countries,we also tested and reported results without controls.

(5) Differences in Industry and Buyer/SellerCharacteristics

Cross-country differences could also arise be-cause of other differences in industry and dealcharacteristics. Private benefits might differ acrossindustry. The media industry, for instance, is oftenmentioned as an industry where private benefits arelarger." Similarly, individuals might value opportu-nities for private benefits more highly than corporateblockholders).^^ We wanted to ensure that our cross-country comparison was not affected by any system-atic difference in the industry characteristics of thedeals or the nature of the seller and the buyer. Forthis reason we re-estimated the country averages,controlling for differences in industry characteristicsand identity of the controlling party.

To capture industry differences, we introducedan industry dummy based on the two-digit SIC codeof the acquired firm. About three-quarters of ourtransactions were accounted for by manufacturing(39%), finance insurance and real estate (24%), andservices (10%). In a crude way, these controls capturedifferences in private benefits linked to productmarket competition. Second, we constructed a mea-sure of tangibility of assets (percentage of total assetsthat are fixed) based on the three-digit SIC code oftheacquired firm. The rationale for this control was thatinsiders will have more difficulty diverting resourcesif assets are tied down and easily observable, as is thecase with tangible assets. ̂ ^

As shown in column 3 of Table III, companieswith more tangible assets were found to have lowerprivate benefits. Firms in wholesale trade, finance(financial, insurance, and real estate sector), andtransportation and utilities had higher level of privatebenefits than manufacturing firms (although theseresults are not significant)."

We also collected information on the identityof the acquirers and the sellers, attempting todetermine whether the seller was an individual,the company itself (through new share issues), ora corporate entity. We found the most commonseller is a corporation, followed next by individu-als (18%), new share issues (l6%), and govern-ments (3%). We used SDC data to identify whetherthe acquirer was a public company, subsidiary,the government, or a private company. The trans-actions in our sample were almost equally dividedbetween public acquirers (41%) and privateacquirers (also 41%). We provided a further clas-sification using news stories and the SDC synopsisfield. We identified 13% of our transactions asinvolving individual acquirers, and 4% as in-volving a financial intermediary that purchasesthe shares and then resells them to institutionalinvestors. We interpreted these latter acquisitionsas the dispersal of the controlling stake. None ofthese buyer or seller characteristics turned out tobe significant.

At the bottom of column 3 of Table III, we reportthe estimates of the country average level of privatebenefits after we control for the above differences inlevel of private benefits across industries. The rela-tive ranking, however, does not seem to be affectedvery much by these industry controls.

Alternative Interpretation of ControlPremiums

While conducting our study, we also consid-ered the possibility that the block premiums werenot really measuring private benefits, but somethingelse. The most plausible alternative interpretation is

10. Since we have enough observations for the United States (46), we canassess the realism of our assumption by estimating the same specification restrictedto U.S. data. While the other coefficients are very similar to the ones reported inTable IV, the coefficient of the majority block dummy is small and insignificant."Imposing" to the United States the same majority dummy effect as other countries,thus, will distort its average level of private benefits upward.

11. See, for example, Demsetz and Lehn (1985).12. See, for example, Barclay and Holderness (1989).

13. To avoid potential endogeneity problems, we used U.S. averages (seeRajan and Zingales (1998)). We derived U.S. measures in a two-step procedure.First, we computed the average ratio of fixed assets (property plant and equipment)to total assets for all companies in each three-digit SIC code for the period 1990-1999. Then we took the median value across all companies. We then impute thisvalue for all of the companies in our sample.

14. Including both types of industry controls in a sense "overcontrols" forindustry effects, with tangibility (coefficient of -0.11, p-value of 0.033) and financialindustry (coefficient of 0.076, p-value of 0.019) being significant in regressions thatinclude these industry variables separately (nol reported).

60JOURNAL OF APPLIED CORPORATE HNANCE

that such control premiums refiect a tendency ofbuyers to overpay. We tested for this possibility bylooking at the announcement effect on the stockprice of the acquiring company. If these premiumsrefiect overpayments, acquiring firms should expe-rience negative returns at the announcement of thetransaction.

Of the 393 transactions in our sample, therewere 115 in which the acquirer was a publicly tradedcompany and the stock price reaction was reportedin Datastream. But inconsistent with the overpay-ment hypothesis, the mean value of the announce-ment effect (not reported) was in fact slightly positive(0.5%) and not statistically diiferent from zero. Anotherimplication of the overpayment hypothesis is that thebuyer's announcement return should be negativelyrelated to the size ofthe control premium. But our studyfound that although the coefficient was negative, it wasneither economically nor statistically significant (coeffi-cient of-0.018, p-value of 0.64).

A second possible alternative interpretation—particularly for larger premiums in underdevelopedand therefore less efficient markets—is that buyershave better information (than the market) about theexpected payoff from a transaction and there istherefore a delay in incorporating new information.On average, such delays will infiate our estimates ofprivate benefits.

To test for this possibility, we re-estimated theprivate benefits using the market price 30 days afterthe announcement instead of just two days after. Butour results were virtually identical. If anything, theaverage premium in developing countries like Brazilwas higher rather than lower. We also examined thecumulative abnormal returns to shareholders intarget firms from two days to 30 days after theannouncement and tested whether the initial level ofprivate benefits was related to the subsequent cumu-lative abnormal returns. But we found no such effect(with an insignificant relationship between controlpremiums and post- announcement returns (coeffi-cient of 0.009, p-value of 0.80).

Finally, we were concerned about a possibledistortion stemming from selective nondisclosure ofthe terms of a transaction. In fact, one of the criteriawe had to impose to obtain our estimates was theobservability of the price paid for the controllingblock. A major concem was the possibility that, incountries with better protection of investors, control-ling parties were more reluctant to disclose largepremiums. If that were the case, then our study

would report lower private benefits in the UnitedStates not because they are indeed lower, butbecause large premiums are less likely to be dis-closed.

To check for this possibility, we computed thepercentage of deals we had to drop from our samplebecause the terms were not disclosed. On average,33% of the deals did not disclose the terms, rangingfrom 0% in Taiwan to 70% in Austria and 82% in theCzech Republic. Contrary to the selective nondisclo-sure argument, we found that countries with higherpremiums tend to have a higher percentage of dealswhose terms are not disclosed. Similarly, when weused as a proxy for protection of minority sharehold-ers the "antidirector rights" index constructed by LaPorta et al. (1997), we found (not surprisingly) thatin countries with greater shareholder protection alarger percentage of deals are disclosed. In sum, ifselective nondisclosure biased our results, it did soin a way that understated rather than exaggeratedcross-country differences.

ARE WE REALLY ESTEVIATBVG PRIVATEBENEFITS?

While we were able to reject these alternativeinterpretations, what evidence did we have thatour estimates were really reflecting private ben-efits of control?

At the anecdotal level, there are severalstudies documenting the pervasiveness of self-dealing transactions in countries like Italy (such asZingales (1994)) and the Czech Republic (Glaeseret al. (2000)). It is thus reassuring that our esti-mates of private benefits for these two countriesare both very high (37% and 58%, respectively). Itis particularly interesting to stress the differencesbetween our estimates for Poland and the CzechRepublic. Both of them are former socialist coun-tries, with a similar level of GDP per capita.Nevertheless, consistent with the analysis of Glaeseret al. (2000) who document differences in theprotections for minority shareholders, our esti-mates are very different (11% for Poland, 58% forthe Czech Republic).

A more subtle test of whether these estimatesreally refiect the ability to extract private benefits iswhether our estimated private benefits depend notonly upon the institutional variables of the countryof the company whose control has been acquired,but also on those of the country of the acquiring


company (when the two are different).'^ In otherterms, an acqtiirer from a countiy with less investorprotection is likely to be better able to siphon outcoiporate resources from a subsidiary than a companycoming from a countiy with very rigid ailes. Thisshotild be reflected in a higher willingness to pay andhigher control premiums. Thus, we should observehigher estimated private benefits in cases where aforeign acqtiirer comes from a country with poorprotection of investors.

To test this possibility, we re-estimated our basicspecification inserting as additional explanatory vari-ables two interaction temis: between the difference ingovernance ailes and die foreign acquii"er's dummy(eqtial to one if tlie acqtiii'er comes from a countrydifferent from the target), and between the difference ingovemance ailes and die cross-listing dummy. Ourproxy for die difference in govemance ailes was thedifference between countries of die anddirector rightsindex niendoned earlier. As shown in column 4 of TableIII, companies coming from more investor-friendlycoLintries paid control premiums diat were 2.8% less, onaverage (and diis effect was stadstically significant). Wealso found a statisdcally significant negative effect ofthesuperiority of govemance ailes on die control premiumsfor cross-listed firms. This result implies that die reduc-don in private benefits resulting from cross-listing isgreater for fimis from countiies with weaker investorprotecdons. (In the bottom of Table III we presentcountry fixed effects with diese controls—and suchcontrols were applied in all the tests reported in theremainder of this article.)

Besides providing some comfort that our esti-mates were indeed capturing private benefits ofcontrol for company insiders, these findings alsoprovided some new evidence that bears on theongoing debates about governance. On one hand, theresults suggest the possibility for firms in countrieswith weak investor protection to raise their value by"borrowing" another cotintry's insdtudons—either byselling a block to a firm in that country or, say, bycross-listing in the U.S. These results provide directsupport for the contention of studies (such as Coffee

(1999), Reese and Weisbach (2001), and Doidge etal. (2001) of a link between cross listing and privatebenefits. But perhaps more important, the resultscast some doubt on the prospects for the conver-gence of different corporate governance systems.For example. Coffee (1999) predicts that companiesfrom coimtries with better protection of investorswill end up buying companies from countries withweaker protection. Our result suggests that, in thecontext of controlling blocks, companies from coun-tries with better investor protection may be morelimited in their ability to extract private benefits andso less willing to bid high premiums for controllingblocks. This raises the possibility that controllingblocks may end up in the hands of companies fromthe countries with the worst rules, not the best.

An Analysis of Outliers

In Brazil we estimated private benefits to be 65%of the value of equity. Could private benefits reallybe this large, or was this finding the result of someproblem with the way we measured them? We didsome further analysis of this extreme outlier to helpascertain the reliability of our method.

As part of a study of the impact of legal reformon private benefits in Brazil, Nenova (2001b) indepen-dendy collected infonnation on control sales in Brazilbetween 1995 and 2000. In so doing, she identified 8transacdons that met our initial sample selecdoncriteria.'^ For those transacdons, she reported an aver-age value of private benefits of 42%, reasonably closeto our estimate. In addidon, we asked a Brazilianinvestment bank to give us all the privadzation data fortransactions in which the government sold a control-ling block of an already listed fimi.'^ Tlieir searchproduced 23 privadzadon transacdons with the requi-site data, including 21 transacdons that were notincluded in our original data set.'^ The average controlpremium in this sample was 129%.

In sum, independent estimates lead to a veiysimilar conclusion: Private benefits of control inBrazil are extremely high.

15. For example, Coffee (1999), Reese and Weisbach (2001), and Doidge etal. (2001) argue thai foreign companies list in the United States to precotnmitthemselves to extract fewer private benefits of control.

16. Her approach, albeit veiy similar, is not strictly comparable with our own.as she uses the price on the date of sale and compares the sale price with the priceof voting shares on the exchange.

J7. This samj")le only includes transactions where the sale price is cash. Thatis, we excluded privatizations where sale price could include so-called "privatizationcurrencies" that include government debt that was tr.iding at a discount.

18. They identified 12 transactions where the stake sold was 19.26% which weexcluded because this level was below our selection criteria, but in Brazilaccounted for 50.1% of the voting shares in the cotiipatiy. In addition, they wereable to identify stock market prices for a nutnber of firms that we were not ableto collect using Datastream or were not identified by SDC.


Within-Country Variation in Private Benefits

Another check to verify whether our methodcaptures private benefits was to see whether ourestimates changed when there are changes in exter-nal conditions that affect the ability to extract privatebenefits. While the fact that we had relatively fewtransactions from many countries limited our abilityto systematically explore time series variation, forthree events we had this possibility.

The first event we explored was the passage inItaly of a corporate governance reform in 1998known as the Draghi reform. Among other things,this reform made it easier for minority shareholdersto sue management appointed by the controllingshareholder. Such a reform should limit the ability toextract private benefits. When we segmented ourdata into obsei'vadons before and after July 1998, wefound that before the reform the average value ofprivate benefits was 47%, while after the reform theaverage value was only 6%.̂ '

The second event we explored focuses onBrazil in the 1990s where, as Nenova (2001)reported, there were two important changes incorporate law. The first change occurred on May 5,1997, when Law 9457 was adopted. This law,designed in part to enhance government revenueswhen they sold controlling stakes in privatizationtransactions, eliminated mandatory disclosure of theprice of sales of blocks, eliminated minority share-holders' right to withdraw from the firm in the caseof significant transactions such as mergers andspinoffs and to receive a price per share based on thebook value of the firm, and eliminated a requirementfor acquirers to make a mandatory offer to otherholders of voting shares at the same price and termsas that for the controlling block. The elimination ofwithdrawal rights gives controlling shareholders an-other avenue to extract private benefits, while the equalopportunity provision theoretically has more mixedeffects on the ability to extract private benefits (seeBebchtik (1994)). The second change, passage ofInstruction 299 by the Brazilian securities and ex-change commission (CVM), reinstated these disclo-sure, withdrawal, and equal opportunity provisions,and added even more disclosure requirements.

These legal changes suggest that private ben-efits will differ depending on which legal regime is

in effect, with private benefits expected to begreatest in the period when Law 9457 was ineffect, and lower both before and after. This is infact what we found in our sample of transactions;the average premium was highest in the 9457period, at 119%, with lower levels in the pre-9457period at 53% and 37% in the post-instruction 299period. A similar trend was revealed in ourprivatization sample where we had data only forthe first two periods (with values of 109% in thepre-9457 period increasing to 131% in the 9457period).

The third event we explored focuses on changesin the economic environment rather than changes inthe legal regime to protect investors. It has beensuggested that stealing will increase when the ex-pected return on investment declines (e.g., Yellen(1998)) and that the Asian crisis presents such anevent Qohnson, Boone, Breach, and Friedman(2001)). We tested for this possibility, examiningwhether the levels of private benefits were differentfor emerging markets in Asia during the Asian crisis(of 1997 and 1998).̂ ° Based on a regression of privatebenefits with countiy fixed effects, we found that theAsian crisis period was indeed associated withhigher private benefits (coefficient of 0.068), al-though this is not significant at conventional signifi-cance levels (p value=0.l62).

In sum, in all the three instances our estimatesof private benefits were consistent with our theory.Having established some degree of confidence inour estimates, our study next turned to an explo-ration of the key factors that lead to betterperformance.

WHAT WORKS TO LIMIT PRIVATE BENEFITS?

What sorts of institutions help to constrainabuses? And what types of corporate governancereforms—both by firms and governments—willultimately deliver the greatest returns?

The prevailing view is that the law and acountry's legal origin are the primaiy institutionalfactors that work to limit corporate abuses. This viewfocuses clearly on the interests of financial investorsand the specific contractual and legal mechanismsthat are designed to constrain managerial miscon-duct. Accordingly, we examined the importance of

19. 'l"he p-value for the equality ofthe two means is only 21%, but this is notsurprising given that we have only six observations before and two afterwatd.

20. Specifically, this test iticludes l-long Kong, Indonesia, Korea, Malaysia,Philippines, Singapore, Taiwan, and Thailand.

63VOLUME 16 NUMBERS 2-3 o SPRINCASUMMER 2004

legal institutions for governance outcomes. But wedid not limit our empirical investigation to just theseinstitutions. There are other stakeholders in firmsthat take an interest in governance and have thepotential to influence the extent of abuses. As aresult, we also explored whether explicit contractswith other stakeholders (for example, the govern-ment through its tax claim) and non-legal constraintson management behavior (such as concerns aboutreputation and avoiding criticism by the press) helplimit the consumption of private benefits.

The Role of Legal Institutions

The legal environment. The ability of a control-ling shareholder to appropriate some of the valuegenerated is limited by the possibility of being sued.Thus, a greater ability to sue should translate intosmaller private benefits of control (see Zingales(1995a)). The same reasoning applies to any legalright attributed to non-controlling shareholders (LaPorta et al., 1997). Accordingly, our study attemptedto examine the explanatory power of legal rights thatgive minority investors leverage over insiders. Morespecifically, we focused our attention on the level ofshareholder rights in the country of the target firm (asproxied for by the La Porta "anti-director rights"index for the transition countries in our sample).

Disclosure standards. Disclosure standards regu-late the information available to non-controllingshareholders. The more accurate is this information,the more difficult it is for a controlling shareholderto appropriate value without incurring legal penal-ties or at least reputational costs. Thus, we expectedour measures of the quality of disclosure to benegatively correlated with the size of private benefitsof control.

Enforcement. The strength of legal protectionsdepends on the expectation of speedy and predict-able enforcement. Thus, we included as one of ourcontractual variables a measure of the strength of acountry's law and order tradition as measured by thecountry risk rating agency International CountryRisk. This rule of law index is scaled from zero to ten.

Extralegal Institutions

The potential constraints imposed by extra-legal institutions have not been prominent in currentdebates, at least in part because of a lack of empiricalexamination. We focused our attention on five

institutional factors that, at least in theory, have thepotential to raise expectations of penalties foractivities that produce private benefits for control-ling shareholders. Some of these factors that canraise the costs to the controlling shareholder forwealth-diverting activities (such as the penaltiesproduced by product market competition and bypublic opinion pressure) are constraints that areexternal to the firm. Other factors (such as thesanctions that can be introduced by moral norms,labor, and the government as tax collector) aremore 'internal' to the firm.

Product market competition. The degree ofproduct market competition affects the opportunityto appropriate private benefits in two main ways.First, the more competitive markets are, the moreverifiable prices become. When prices are more"objective," it is more difficult for a controllingshareholder to tunnel out resources using manipu-lated transfer prices without incurring legal orreputational costs. Second, in a competitive market,the distortions produced by the extraction of privatebenefits are more likely to jeopardize the survival ofthe firm. Hence, competition represents a naturalconstraint on the extraction of private benefits. Inregressions in our study, we included controls forindustry characteristics and used as our cross-countryproxy for the extent of product market competitionthe general response to the survey question, "Docompetition laws prevent unfair competition in yourcountry?," as reported by the World CompetitivenessYearbook for 1996.

Public opinion pressure. Controlling share-holders might limit their efforts to divert firmresources not out of fear of legal sanction byfinancial investors but rather out of concern fortheir reputation. But for concern about reputationto act as a deterrent, it is necessary to have a "publicopinion"—that is, an independent press that pub-licizes the facts and a large set of educated investorswho read newspapers and have some (collective)power to punish improper behavior (Zingales,2000). As one example, shareholder activist Rob-ert Monks succeeded in initiating some majorchanges at Sears not by means of the norms of thecorporate code (his proxy fight failed miserably),but through the pressure of public opinion. Hepaid for a full-page announcement in the WallStreet Journal where he exposed the identities ofSears' directors and called them "non-perform-ing assets" (Monks and Minnow (1995)). The


embarrassment for the directors was so great thatthey implemented all the changes proposed byMonks. Likewise in Russia, one of the top-performing investment funds has achieved itsreturns in part by publicizing abuses and soembarrassing insiders and regulators to take cor-rective action (Dyck (2002)). In a study we pub-lished two years ago (Dyck and Zingales (2002)),we attempted a systematic examination of thispotential corporate governance role of the media.Specifically, we tested whether indicators of thepresence of the press are correlated with lowerlevels of abuse. As our measure of press presence,we used an indicator of newspapers' "diffusion,"measured as the circulation of daily newspapersnormalized by population.

Internal Policing through Moral Norms. Re-gardless of the reputational cost or the legal punish-ment associated with private benefits, a controllingshareholder might choose not to appropriate valueas a result of moral considerations. Coffee (2001)proposed the violent crime rate as a proxy for thesemoral norms, while Stulz and Williamson (2001)focused on culture as an indicator of norms. To testfor an impact of moral norms, we used both pro-posed measures: the number of violent crimes (asreported by the World Competitiveness Yearbookbased on Interpol data for 1993) and Stulz andWilliamson's classification of countries by their pri-mary religious orientation.

Labor as Monitor. Additional constraints oncontrolling shareholders might come from the pres-ence of economic entities with a direct interest in firmdecisions that could penalize efforts to extract pri-vate benefits directly without having to turn to thecourts. Labor is privy to inside information oncustomers and suppliers and can hold up the con-trolling shareholder by threatening to withholdservices and, in some cases, through its position onthe board of directors. On the other hand, becauselabor's interest are not the same as minority share-holders', it is not clear that labor can be counted onto constrain private benefits; in many cases, it maychoose to align itself with the controlling share-holder against outside investors. We tested for theeffect of labor on private benefits using as a cross-country measure of the extent of potential laborpower the degree of employee protection. Thismeasure is available for all OECD countries.

Government as Monitor through Tax Enforce-ment. Like labor, government has an economic

interest in corporate profitability and decision mak-ing, and it can take actions to reduce private benefitswithout having to turn to the courts. What hasreceived little recognition, though, is how the pres-ence of the state's claim (e.g., corporate income tax)and how the state enforces that claim can influencegovernance and have spillover effects on minorityshareholders. We make a distinction between theeffect of tax rates and tax enforcement. While thepresence of taxes can make governance problemsworse, by providing an additional incentive to stealfrom minority shareholders (see Desai, Dyck, andZingales (2003)), tax enforcement can improve thesituation for shareholders. This is because taxauthorities and non-controlling shareholders havea common objective: to ascertain the value pro-duced by a company and get a share of it. Forexample, a principle of corporate taxation fortransfer pricing is the use of an arms-length pricebased on what independent parties in a competitivemarket would charge. How tax authorities enforcetheir rules on transfer pricing affects the incentivesto reallocate returns through transfer pricing, withstrict enforcement reducing the likelihood that acontrolling shareholder will use transfer prices tosiphon out value at the expense of minority share-holders. Therefore, better tax enforcement shouldlead to smaller private benefits of control. Hence,to test for the effect of taxes on governance in ourstudy, we used as a measure of the effectiveness ofa taxation system an index developed by the WorldCompetitiveness Report that assesses the level oftax compliance.

The Tests

In our study, we then ran a series of testsexamining, first, the impact of each institution inisolation on private benefits (the results of which arereported in Table IV) and, second, their effects incombination with each other (see Table V). (For bothof these regressions, we included all of the controlvariables used in the test reported in Table III as wellas an indicator variable that identifies countries thathave any form of tender offer requirement.) In the"univariate" tests whose findings are reported inTable IV, we included log GNP per capita and ruleof law variables to capture important sources ofadditional variation across countries. In multivariatetests reported in Table V, we did not include thesecontrol variables.


TABLE IV • INSTITUTIONAL DETERMINANTS OF PRIVATE BENEFITS OF CONTROL—UNIVARIAlTi ANALYSISThe dependent variable is the block premia as a percent of firm equity. The explanatory variables include all of variables introducedin Table HI except the country fixed effects, but including a dummy to indicate the presence of a mandatory tender offer law. Inplace of the country fixed effects, we include as controls log gnp per capita and riile-of-law index, then introduce one at a timeseveral institutional variables: (1) accounting standards index; (2) antidirector rights index; (3) an index of the extent of competitionlaws; (4) diffusion of the press as measured by the newspaper circulation/population; (5) incidence of violent crimes; (6) extentof legal protections for labor; (7) a dummy variable if primaiy religion is Catholicism; and (8) tax compliance index. Standard errors,which are reported in parentheses, are robust and clustered by country.

INDEPENDENT VAKIABLES LEGAL INSTrrUTIONS EXTRA LEGAL INSTmjTION

Accounting standards

Antidirector rights

Competition laws

Newspaper circulation/pop.

Violent crime incidence

Labor protection

Catholicism is primary religion

Tax compliance

Variables Controlled for:Buyer bargaining powerOwnership variablesFinancial distress

Foreign acquirerCross-listed in the U.S.Buyer identitySeller identityIndustry groupTangibility of assets

Interaction of relative strength of anti-director rights(home-target nation) and foreign acquirer dummy

Interaction of relative strength of antidirector rights(U.S.-target nation) and cross-listed in the U.S. dummy

Presence of takeover lawLog gnp per capitaRule of lawConstant

Number of observationsCountries includedR-squared

• significant at 5% level; " .significant at 1% level

(1)

-0.004[0.002]'

yyyyyyyyy

y

yyyyy

38136

0.250

(2)

-0.026[0.011]"

yyyyyyy

•<:

y

y

yyyyy

39339

0.220

(3)

-0.066[0.034]*

yyyyyyyyy

y

yyyyy

39339

0.210

(4)

-0.021[0.012]*

yyyyyyyyy

y

yyyyy

39339

0.220

(5)

0.000(0.000)

yyyyyyyyy

y

yyyyy

37736

0.230

(6)

0.027[0.020]

yyyyyyyyy

y

yyyyy

23318

0.200

(7)

0.064[0.058]

yyyyyyyyy

y

yyyyy

39339

0.220

(8)

-0.069[0.025]"

yyyyyyyyy

y

yyyyy

39339

0.240

66JOURNAL OF AJ'FLIED CORPORATE FINANCE

We started by estimating the impact of "legal"factors—those factors that directly or indirectly relyon the court enforcement of certain rights. Becauseinformation disclosure is a prerequisite for any legalaction, we started with the quality ofthe accountingstandards, as measured by the CIFAR index. Asreported in column 1 of Table IV, companies incountries with better accounting standards havelower private benefits of control. This effect is bothstatistically and economically significant. A onestandard deviation increase in our measure of ac-counting standards reduced the value of control by9.0 percentage points. Viewed together with theother control variables, accounting standards ex-plained 25% of the variation in private benefits ofcontrol (the firm-specific control variables aloneexplained just 15%).

The second variable we examined was theextent of legal protections for minority investors,measured, again, using the La Porta et al. index ofantidirector rights. As reported in column 2, coun-tries with greater antidirector rights have lowerprivate benefits of control. A one standard deviationincrease in the La Porta index reduced the value ofcontrol by 4.4 percentage points. Together with thefirm-specific variables, antidirector rights explained22% of the variation in private benefits of control. Insum, we found that legal institutions are stronglyassociated with lower levels of private benefits.

We also tested the explanatory power providedby extra-legal institutions, which are suggested by afunctional rather than an institutional perspective.Here we focused on crude country-wide measuresof product market competition, size and extent ofreputational penalties, moral norms, employee pro-tections, and diligence of tax authorities.

As reported in column 3 of Table IV, we foundthat, after controlling for industry type, countrieswith more competitive product markets—at least asmeasured by this survey of the World Competitive-ness Report—have lower private benefits of control.A one standard deviation increase in our measure ofcompetition reduced the value of control by 6.0percentage points. Together with the firm-specific

variables, competition explained 21% of the varia-tion in private benefits of control.

As reported in column 4 of Table IV, consistentwith the idea that public opinion pressure might curbinsider abuse, countries where newspapers are morediffused (as measured by the number of copies soldper 100,000 inhabitants) have lower private benefitsof control. Diffusion captures both the importance ofpublic opinion and the credibility of newspapers(assuming that less credible newspapers sell fewercopies).^' A one standard deviation increase innewspapers' diffusion reduced the value of controlby 6.4 percentage points. Together with the firm-specific variables, newspapers' diffusion explained22% of the variation in private benefits of control.The findings reported in columns 3 and 4 bothsuggest that institutions external to the firm help tolimit private benefits.

Columns 5 and 7 report the findings of our testsof the idea that countries with higher moral normsmight have lower private benefits. Consistent withCoffee's prediction, countries with worse norms asproxied by higher violent crime rates have higherprivate benefits of control, but the effect is economi-cally and statistically insignificant. To investigatemoral norms as proxied by primary religion, weintroduced indicator variables for the four mainreligions (Buddhist, Catholic, Muslim, and Protes-tant). As a country religion, we used the dominantone (see Stulz and Williamson, 2001). We found thatCatholic countries have higher private benefits (al-though the result is not significant) while those inProtestant countries were significantly lower. Theeffects of the Muslim and Buddhist religion were notsignificant.

Reported in columns 6 and 8 are our tests of theextent to which the strength of other corporatestakeholders are associated with lower levels ofprivate benefits. In column 6, we report the esti-mated effect of labor as a monitor of private benefits.As an index of potential labor strength, we used bothan unweighted and a weighted (not reported) indexof employee protections based on average indictorson regular contracts and short-term contracts from

21. In Dyck and Zingales (2002b) we study the detenninants of newspapers'diffusion. We find that the type of dominant religion and the degree ofethnolinguistic fractionalization explain 41% of the variation in pre.ssdiffusion. When we use these as instrutnents for pre.ss diffusion, the resultsare unchanged.


TABLE V • INSTITUTIONAL DETERMINANTS OF PRIVATE BENEFITS OF CONTROL—MULTIVARIATE ANALYSISThe dependent variable is the block premia as a percent of firm equity. The explanatory variables include all of the variablesintroduced in Table III except the country fixed effects, but including a dummy to indicate the presence of a mandatory tenderoffer law. As institutional variables in specification (1) we use the antidirector rights index and the rule-of-law index. In specification(2) we use a dummy variable if the primary religion is Catholicism, a tax compliance index, the diffusion of the press as measuredby the newspaper circulation/population, and the index of the extent of competition laws. The independent variables inspecification (3) are the antidirector rights index, rule-of-law index, tax compliance index, and diffusion of the press as measuredby the newspaper circulation/population. Standard errors, which are reported in parentheses, are robust and clustered by country.

DEPENDENT VARIABLE: BLCKK PREMIUM

INDEPENDENT VARIABLES (1) (2 ) (3)

Antidirector rights

Rule of law

Catholicism

Tax compliance

Newspaper circulation/population

Competition laws

-0.026"(0.012)

-0.026*"(0.010)

y

y

y

y

y

y

y

y

y

0.019(0.056)

-0.064"*(0.021)

-0.020"(0.009)

-0.042(0.036)

y

y

y

y

y

y

y

y

y

-0.003(0.019)

-0.006(0.011)

-0.061*(0.033)

-0.018*(0.010)

y

y

y

y

y

y

y

y

y

Variables Controlled for:

Buyer bargaining power

Ownership variables

Financial distress

Buyer characteristics

Seller characteristics

Foreign acquirer

Cross-listed in the U.S.

Industry type

Tangibility of assets

Interaction of relative strength of anti-director rights(home-target nation) and foreign acquiror dummy

Interaction of relative strength of antidirector rights(U.S.-target nation) and cross-listed in the U.S. dummy

Presence of takeover law

Constant

y

y

y

393

39

0.213

y

y

y

393

390.245

y

y

y

393

39

0.243

Number of observations

Countries included

R-squared

• significant at 10% level; " significant at 5%; "'significant at 1% level


OECD data compiled in Pagano and Volpin (2000).The restriction to OECD countries admittedly limitsour number of countries and observations, but givesperhaps a purer test of the contention that labor canwork as monitors since this literature has focused onorganized labor in developed economies. Inconsis-tent with the hypothesis that labor is an effectivemonitor (but consistent with Pagano and Volpin'scounter-contention that entrepreneurs and workerswill align themselves against the interests of minorityinvestors), we found that increased labor power isassociated with higher private benefits, although thisresult is not statistically significant (p-value of 0.204for employee protections, 0.13 for weighted em-ployee protections).

Einally, we investigated the possibility that agovernment interested in enforcing tax rules canreduce private benefits. As reported in column 8,countries with a higher degree of tax compliance, asmeasured by the Wodd Competitiveness Report,have lower private benefits of control. A one stan-dard deviation increase in our measure of taxcompliance reduced the value of control by 8.6percentage points, a significant amount. Togetherwith the firm-specific variables, tax complianceexplained 24% of the variation in private benefits ofcontrol.

In interpreting these results, one should keep inmind that tax compliance is an "equilibrium out-come," one that is affected both by tax enforcementand by the attitude of citizens toward cheating ontheir taxes. To try to identify the impact of taxenforcement, we included (in an unreported regres-sion) a measure of willingness to cheat on taxes asmeasured in the World Value Survey. In this surveypeople are asked to rate from 1 to 10 the statement"cheating on taxes if you have a chance is ...",where 1 is never justifiable and 10 is alwaysjustifiable. We found this variable to be insignifi-cant, while the coefficient on tax compliance re-mained significant, suggesting that the effect of taxcompliance comes from tax enforcement and notfrom differences in moral values across countries.We also examined the robustness of this result tothe inclusion of the marginal tax rate and our resultswere unchanged.

Table V (column 2) reports the combinedeffect of the four extra-legal institutions that indi-vidually had a statistically significant effect. All fourvariables retained the predicted sign, but the magni-tudes of their coefficients fell and only tax compli-

ance and newspaper diffusion remained statisticallysignificant at the 5% level. Together these fourvariables were able to explain 24 % of the variationin private benefits.

Thus, all the evidence reported thus far isconsistent with both the legal and the extra-legalinstitutions playing a role in constraining privatebenefits. In fact, a crude R-squared test suggests thatboth sets of institutions have roughly the sameexplanatory power. Next we asked ourselves, canwe design a test that would enable us to distinguishwhich is more important?

There are two obstacles to doing so. Eirst, manyof these institutional variables are highly correlated.Shareholder's protection, however, showed no cor-related with newspaper circulation and had a corre-lation of only 0.4 with tax compliance. Second, andmost important, all these proxies are measured witherror. Hence, their statistical significance in a multi-variate analysis may well be more related to the levelof noise in these measures than to their actualimportance.

Nevertheless, we thought it would be interest-ing to try and put all these variables in one regres-sion. When we did so, as reported in column 3 ofTable V, of all the institutional variables we found tobe significant in the previous regressions, onlynewspapers' diffusion and tax compliance remainedsignificant. The paucity of observations and the highdegree of multi-collinearity caution us against draw-ing any strong conclusion from this comparison.What we can say, however, is that the results areinconsistent with an exclusive focus on legal factorsas institutional curbs to private benefits.

CONCLUSIONS

In a study recently published in the Journal ofFinance, we attempted to measure the quality of acorporate governance system by computing thelevel of private benefits enjoyed by insiders. We thenused these estimates to determine which institutionshelp shareholders limit the diversion of wealth byinsider or controlling shareholders.

We found that many institutional variables,taken in isolation, seem to be associated with a lowerlevel of private benefits of control: better accountingstandards, better legal protection of minority share-holders, better law enforcement, more intense prod-uct market competition, diffusion of an independentpress, and a high rate of tax compliance.


The possible role of tax enforcement in reduc-ing private benefits, and thus indirectly enhancingfinancial development, is probably the most impor-tant new insight that emerges from our analysis.Improving the corporate taxation system is well

within the range of feasible reforms. If this is indeeda primary mechanism by which private benefits ofcontrol can be curbed and financial markets fos-tered, the benefits of financial development might bewithin reach for many more countries.

• ALEXANDER DYCK

is Associate Professor of Finance at Joseph L. Rotman School ofManagement at the University of Toronto.

• LUIGI ZINGALES

is the Robert C. McCormack Professor of Entrepreneurship andFinance at the University of Chicago's Graduate School ofBusiness. He is also a faculty research associate at the NationalBureau of Economic Research and a research fellow at the Centerfor Economic Policy Research.

REFERENCES

Aghion, Philippe and Patrick Bolton, 1992, "An Incom-plete Contract Approach to Financial Contracting," Reviewof Economic Studies, Vol. 59, pp. 473-494.

Barclay, Michael and Clifford Holderness, 1989, "PrivateBenefits of Control of Public Corporations," Journal ofFinancial Economics, pp. 371-395.

Bebchuk, Lucien, 1999, "A Rent-ProtectionTheory of Corpo-rate Ownership and Control," NBER Working Paper #7203.

Bebchuk, Lucien and Christine JoUs, 1999,"ManagerialValue Diversion and Shareholder Wealth,"/oMm«/o/Zflz<;,Economics and 0)-ganization, Vol. 15, No. 2, pp. 487-502.

Black, Bernard, Reinier Kraakman, and Anna Tassarova, 2000,"Russian Privatization and Corporate Governance: What WentWrong?," Stanford Law Reinew, Vol. 52, pp. 1731-1808.

Burkart, M., D. Gromb, and F. Panunzi, 1998, "WhyHigher Takeover Premia Protect Minority Shareholders,"Journal of Political Economy, Vol. 106, pp. 172-204.

Coffee, John, 1999, "The Future as History: The Prospectsfor Global Convergence in Corporate Governance and ItsImplications," Northwestern University Law Review, Vol.93, No. 3, pp. 641-708.

Coffee, John, 2001, "Do Norms Matter? A Cross-CountryExamination of Private Benefits of Control," mimeo,Columbia University Law School.

DeAngelo, H. and L. DeAngelo, 1985, "Managerial Own-ership of Voting K\ghx.s," Journal of Financial Economics,Vol. 14, pp. 36-39.

Demsetz, H. and K. Lehn, 1985, "The Staicture of Corpo-rate Ownership: Causes and Consequences,"/owrwa/o/Political Economy, Vol. 93, pp. 1155-1175.

Mihir Desai, Alexander Dyck, and Luigi Zingales, 2003,"Corporate Govemance and Taxation," Harvard BusinessSchool Working Paper Series, No. 04-001.

Djankov , Simeon, Clara McLeish, Tatiana Nenova, andAndrei Shleifer, 2001, "Who Owns the Media?," NBERWorking Paper #8288.

Djankov, Simeon and Peter Murrell, 2000, "TlieDetenninantsof Enterprise Restaictudng in Transition: An Assessment ofthe Evidence," Working paper. University of Maryland.

Doidge, C, G. Andrew Karolyi, and R. Stub, 2001, "WhyAre Foreign Firms Listed in the U.S. Worth More?," OhioState University Working Paper.

70JOtJRNAL OF APPLIED CORPORATE FINANCE

Dyck, I.J. Alexander and Luigi Zingales, 2004, "PrivateBenefits of Control: An International Comparison/'Journal of Finance, Vol. LIX, No. 2, pp. 533-596.

Dyck, I.J. Alexander and Luigi Zingales, 2002, "TheCorporate Governance Role of the Media," in ne Right toTell: The Role of Mass Media in Development (Washington:The World Bank).

Dyck, I.J. Alexander, 2002, "The Hermitage Fund: Mediaand Corporate Governance in Russia," Harvard BusinessSchool Case 703-010.

Dyck, I.J. Alexander, 2000, "Ownership Structure,Legal Protections and Corporate Governance," inBoris Pleskovic and Nicholas Stern, eds., 2000 An-nual World Bank Conference on Development Eco-nomics, pp. 291-330.

Dyck, I.J. Alexander, 2001, "Privatization and CorporateGovernance: Principles, Evidence and Future Chal-lenges," World Bank Research Observer, Vol. l6. No. 1,pp. 59-84.

Federov, Oleg, 2000, "Case Studies on Abusive Self-Dealing," OECD/World Bank Corporate GovernanceRoundtable for Russia, February 24-25.

Glaeser, Edward, Simon Johnson, and Andrei Shleifer,2001, "Coase vs. the Coasians," Quarterly Journal ofEconomics, Vol. Il6, pp. 853-900.

Grossman, Sanford and Oliver Hart, 1980, "Takeover Bids,the Free Rider Problem, and the Theory of the Corporation,"Bell Journal of Economics, Vol. 11, pp. 42-69-

Grossman, Sanford and Oliver Hart, 1988, "One ShareOne Vote and the Market for Corporate Conivol," Joumalof Financial Economics, Vol. 20, pp. 175-202.

Harris, Milton and Artur Raviv, 1988, "Corporate Gover-nance: Voting Rights and Majority Rules," Journal ofFinancial Econo7nics, Vol. 20, pp. 203-235-

Holderness, Clifford and Dennis Sheehan, 1988,"The Role of Majority Shareholders in Publicly HeldCorporations: An Exploratory Analysis," Journal ofFinancial Economics, Vol. 20, pp- 317-346.

Holthausen, R.W., R.W. Leftwich, and D. Myers, 1990,"Large Block Transactions, the Speed of Response, andTemporary and Permanent Stock Price Effects," Journal ofFinancial Economics, Vol. 26, pp. 71-95-

Johnson, Simon, Rafael La Porta, Florencio Lopez-de-Salines, and Andrei Shleifer, 2000, "Tunneling," Ameri-can Economic Review, Vol. 90, pp. 22-27-

La Porta, Rafael, Florencio Lopez-de-Salines, and AndreiShleifer, 2000, "Investor Protection and Corporate Gover-nance,"/OMnza/q/"/'mflnc/fl/i;co7-zomiC5, Vol. 59, pp. 3-27-

La Porta, Rafael, Florencio Lopez-de-Salines, and AndreiShleifer, 1999, "Corporate Ownership Around the World,"Joumal of Finance, Vol- 54, pp- 471-517-

La Porta, Rafael, Florencio Lopez-de-Salines, AndreiShleifer, and Robert W. Vishny, 1997, "Legal Determi-nants of External Finance," Joumal of Finance, Vol. 53,pp. 1131-1150.

La Porta, Rafael, Florencio Lopez-de-Salines, AndreiShleifer, and Robert W. Vishny, 1998, "Law and Finance,"Joumal of Political Economy, Vol. 106, pp- 1113-155-

Lease, Ronald C, John J. McConnell, and Wayne H.Mikkelson, 1984, "The Market Value of Differential VotingRights in Closely Held Covpordtions," Joumal of Business,Vol. 57, pp. 443-67.

Lease, Ronald C, John J. McConnell, and Wayne H.Mikkelson, 1983, "The Market Value of Control in Pub-licly-Traded Corporations."/owrwfl/q/'/'''mflnctfl/£'conom-ics. Vol. 11, pp. 439-71-

Lopez-de-Silanes, Florencio, 1997, "Determinants ofPrivatization Prices," The Quarterly Joumal of EconomicsVol. 4, pp- 965-1025-

Megginson, William L., Robert C. Nash, Jeffrey M. Netter,and Annette B. Poulsen, 2000, "The Choice betweenPrivate and Public Markets: Evidence from Privatizations,"Working paper. University of Georgia.

MUgrom, P. and R. Weber, 1982, "A Theory of Auctions andCompetitive Bidding," Econometrica, Vol. 50, pp. 1089-1122.

Monks, Robert and Nell Minow, 1995, Watching theWatchers: Corporate Governance for the 21st Century(Maiden, MA: Blackwell Publishing).

Morck, Randall, Bernard Yeung and Wayne Yu, 2000,"Tlie Infonnation Content of Stock Markets: Why Do Emerg-ing Markets Have Synchronous Stock Price Movements?,"Joumal of Finattcial Economics, Vol. 58, pp- 215-60.

Nicodano, G. and A. Sembenelli, 2001, "Private Benefits,Block Transaction Premiums, and Ownership Structure,"Universita' di Torino Working Paper.

Nenova, Tatiana, 2001a, "The Value of Corporate Votesand Control Benefits: A Cross-Country Analysis," mimeo-graph. Harvard University-

Nenova, Tatiana, 2001b, "Control Values and Changes inCorporate Law in Brazil," World Bank Working Paper.


Pagano, Marco and Paolo Volpin, 2000, "The PoliticalEconomy of Corporate Governance," Working paper,London Business School.

Pistor, Katherina, Martin Raiser, and Satanislaw Gelfer,2000, "Law and Finance in Transition Economies," Eco-nomics of Transition, Vol. 8, pp. 325-368.

Rajan, Rajan and Luigi Zingales, 1998, "Financial Depen-dence and Growth," American Economic Review.

Reese, W. and M. Weisbach, 2001, "Protection of Minor-ity Shareholder Interests, Cross-Listings in the United States,and Subsequent Equity Offerings," NBER Working Paper.

Roe, Mark, 2001, "Rents and their Corporate Law Conse-quences," Discussion paper, Columbia Law school.

Rydqvist, K., 1987, "Empirical Investigation of theVoting Premium," Northwestern University WorkingPaper No. 35.

Shleifer, Andrei and Robert Vishny, 1997, "A Survey ofCorporate Govemance," fournal of Finance, Vol. 52, pp.737-783.

Stiglitz, Joseph, 1985, "Credit Markets and the Control ofCapital," foumal of Money, Credit and Banking, Vol. 17,No. 2, pp. 133-152.

Stulz, Rene and Rohan Williamson, 2001, "Culture, Open-ness and Finance," NBER Working Paper #8222.

Zingales, Luigi, 1994, "The Value of the Voting Right: AStudy ofthe Milan Stock Exchange Experience," Review ofFinancial Studies, Vol. 7, pp. 125-148.

Zingales, Luigi, 1995a, "What Determines the Value ofCorporate Votes?," Quarterly foumal of Economics, Vol.110, pp. 1047-1073.

Zingales, Luigi, 1995b, "Insider Ownership and the Deci-sion to Go Public," Review of Economic Studies, Vol. 62, pp.425-448.

Zingales, Luigi, 1998, "Why It's Worth Being In Control,"in George Bickerstaffe, Ed., The Complete Finance Com-panion (London: FT Pitman Publishing).

Zingales, Luigi, 2000, "In Search of New Foundations,"foumal of Finance, Vol. 55, pp. 1623-1653.

72JOURNAL OF APPLIED CORPORATE FINANCE

CONTROL PREMIUMS by Alexander Dyck, AND THE...

Documents

Transcript of CONTROL PREMIUMS by Alexander Dyck, AND THE...