Transcript Displacement loss from firm taxation in the Czech Republic - Cerge-ei

Rent Extraction by Large Shareholders:
Evidence Using Dividend Policy in the
Czech Republic
Jan Bena, Jan Hanousek
Round Table Seminar
CERGE-EI, Prague
May, 2006
Intro – Ownership, Corporate
Governance and Firm Performance
 Early surveys created a general presumption that the
effect of privatization on firm performance is positive
 Most recent studies used larger data sets and
controlled for endogeneity/selection of ownership
 Domestic private ownership has much less definite
impact on performance
 Foreign ownership (especially concentrated) has a
positive effect on many performance indicators
 For more details: Hanousek, Kočenda, Svejnar (2006),
also a survey Estrin, Hanousek, Kočenda, Svejnar
(invited to JEL, 2006).
Intro – Ownership, Corporate
Governance and Firm Performance
 New evidence:



Privatization to domestic owners has only
limited effects on performance
Only privatization to (certain types) of
foreign owners appears to have improved
efficiency of firms
effect of the state not only negative
Motivation: State of the Literature
 Empirical tests do not point out single main
driving force behind corporate payout policy in
cross-section
 Empirical tests of corporate dividend behavior
carried out almost exclusively using data from
the most developed countries
 Existing theoretical literature addresses
predominantly the U.S. experience
Motivation: Questions
 Does ownership structure affect dividend
policy?
 Concentration, Type, and Nationality
 How are benefits from firms distributed among
shareholders
 Do majority shareholders steal profits from
minority shareholders?
 Are dispersed shareholders able to extract
dividends from firms run by managers?
Contribution
 Evidence that ownership structure determines
corporate payout policy
 Dividend policy in the Czech Republic reflects
the conflict among shareholders how to
distribute benefits from firms
 The first empirical study of dividend behavior
in the Central and Eastern Europe
 Important robustness check to established
theories since we are using data from country
with unique recent economic history and
institutional setting
Outline










Stylized Facts about Corporate Dividend Policy
Survey of Dividend Theories
Dividend and Capital Gains Income Taxation
Closest Papers Review
Specific Features of the Czech Economy
Ownership Structures
Model & Data
Estimation Technique
Endogeneity of Ownership
Results
Stylized Facts about Corporate
Dividend Policy
 Long-term target dividend payout ratio
 Changes in dividends rather than absolute
levels: initiations, omissions, increase/decrease
announcements
 Dividend smoothing
 Changes in dividends reflect changes in long-term
ability of firms to generate earnings
 Managers increase dividends only when they are
confident that increased earnings are permanent
 Managers decrease dividends only in financial
distress
Survey of Dividend Theories
 Lintner (1956): partial adjustment process towards a
target payout ratio
 Free Cash Flow Theory
Divert free funds managers have power over
within corporations away from them
Theory: Easterbrook (1984); Jensen (1986);
Zwiebel (1996)
Empirics: Gugler (2003); DeAngelo, DeAngelo, and
Stulz (2004); Desai, Foley, and Hines (2002);
Dewenter and Warther (1998); Laporta, Lopez-deSilanes, Shleifer, and Vishny (2000)
Survey of Dividend Theories, cont.
 Signaling Theory
Communicate the level and growth of earnings
or future prospects of the company to investors
 Theory: Bhattacharya (1979); Miller and Rock
(1985)
 Empirics YES: Bernheim and Wantz (1995);
Amihud and Murgia (1997)
 Empirics NO: Benartzi, Michaely, and Thaler
(1997)
Survey of Dividend Theories, cont.
 (Tax) Clientele Theory


Some investors benefit from special treatment
in the tax law
Summary: Allen, Bernardo and Welch (2000)
Dividend and Capital Gains Income
Taxation
 Companies distribute dividends from after-tax-profits
 Same income dividend tax treatment applied to
individuals and corporations
 Flat tax rate 25 %; in 1999 lowered to 15 %
 Foreign owners: tax treaty between Czech Republic
and the country of the receiver
 Double taxation of dividends prevented
 Marginal tax rate on cash dividends is the same for
all types of shareholders
 Tax considerations or tax clientele effects cannot
drive cross-sectional differences in dividend policies
Closest paper: Gugler (2003)
 Estimates the effect of ownership on dividend
policy using data from Austria
 Ownership and control structure of a firm is a
significant determinant of its dividend policy
 State-controlled firms
 Engage in dividend smoothing, have the
highest target payout ratios, are the most
reluctant to cut dividends
 Family-controlled firms
 Do not smooth dividends, are the least reluctant
to cut dividends
Closest paper:
Gugler and Yurtoglu (2003)
 Analyze dividend announcements and pay-out
ratios in Germany -- Look at the conflict between
large controlling shareholder and minority
shareholders arising from private benefits of control
 Dividends are device for small shareholders to limit
rent extraction by controlling owners
 "Majority-controlled and unchecked" firms have
the smallest target pay-out ratio
 "Majority-controlled and checked" firms have the
largest target pay-out ratio
Major differences from previous
papers
 Czech economic environment and
institutional setting is very different from the
one in Austria (Germany)
 We benefit from a large sample
 We use a different estimation technique to
account for specifics of an emerging market
environment (including privatization)
Specific Features of the Czech
Economy
 Privatization
 Ownership structure of Czech companies was
primarily set (exogenously) by government
bureaucrats
 Economy was privatized and deregulated fairly
quickly
 Market forces drove majority of economic
activity very early during transition
 Ownership structure stabilized after 1996
Specific Features of the Czech
Economy, cont.
 Legal Uncertainty
 Evolution of institutional and legal framework
was considerably slower
 Lawmakers were well behind the economic
activity
 Corporate law incomplete and kept changing
literally every year
 Slow / weak law enforcement
31
.1
2.
1
31 99
.5 2
31 .19
.1 93
2.
1
31 99
.5 3
31 .19
.1 94
2.
1
31 99
.7 4
31 .19
.1 95
2.
1
30 99
.6 5
.1
15 99
.7 7
27 .19
.1 98
2.
1
14 99
.7 8
30 .19
.1 99
1.
19
6. 99
1.
2
30 00
.4 0
30 .20
.1 00
1.
2
30 00
.4 0
.2
31 00
.8 1
.2
29 00
.1 1
.2
30 00
.4 2
.2
11 00
.7 2
29 .20
.1 02
2.
2
28 00
.2 2
31 .20
.1 03
2.
2
30 00
.4 3
.2
30 00
.9 4
30 .20
.1 04
2.
20
04
Number of exceptions
230
210
190
60,000
170
150
50,000
130
40,000
110
30,000
90
20,000
70
50
10,000
30
0
Date when Came into Force (Total number of updates is 52)
Number of words
Number of Phrases "with exception of" and Total Number of words in
the Czech Income Tax Law (1993 - 2004)
90,000
250
80,000
70,000
Ownership Structures:
Concentration
 Large shareholding is the most important control
device in the Czech Republic
 Only highly concentrated owners are able to
control managers effectively
 Because of underdeveloped legal system and
financial market, dispersed ownership structures
cannot enjoy benefits from
 Greater market liquidity and
 Better risk diversification
Ownership Structures:
Variables Definition
 Domicile -- Czech, Foreign
 Type -- State, Private individual, Industrial firm,
Financial institution
 Concentration - Czech corporate law assigns
control rights to different ownership levels:
 Majority
> 50.0%
 Blocking minority
> 33.3% and ≤ 50.0%
 Legal minority
> 10.0% and ≤ 33.3%
 Minority
≤ 10.0%
 We define concentration of ownership variables:
majority, monitored majority, minority, dispersed
Starting Model
 Benartzi, Michaely, and Thaler (1997): "... conclusion
we draw from [our] analysis is that Lintner's model
of dividends remains the best description of the
dividend setting process available."
 Lintner's Model
Di,t i i 

1 i 
Di,t1 i,t
i,t 
 Ownership Structure Determines Dividend
Payments
Di,t  
j j j 

1 j 
Di,t1 
OWN
ji,t i,t
i,t 
j
Model: Control Variables
 Firm Size (Total assets)
 Leverage (Total liabilities / Total assets)
 Bank Power (Bank loans / Total liabilities)
 Cash Holdings (Cash / Total Assets)
 Growth Opportunities (Industry level sales
growth rate)
 Year dummies
Data
 Medium and large non-financial companies traded on
the Prague Stock Exchange
 Fix the population by choosing 1,664 companies
privatized in 1991-1995
 Financial and ownership data from database ASPEKT
 To estimate dividend equations we use data from 19962003 (post-privatization market economy period)
 Data from 1991-1994 (privatization period) are used
as instrumental variables that allow us to control for
the endogeneity of ownership
 Privatization period data come from the Ministry of
Privatization
Descriptive Statistics:
Dividend Payments
Number of
companies paying
dividends in a given
year
Year
1996
1997
1998
1999
2000
2001
2002
Total
NOB
71
86
75
61
63
58
54
468
Problems with direct application of
the Lintner's model
 Fewer firms paying dividends (<10% of sample) hence
OLS estimation leads to biased results
 Missing financial data (weak market supervision), more
pronounced for those not paying dividends (about 50%)
 Our study follows privatization when the ownership is
potentially endogenous with respect to corporate
performance (e.g., state versus private, domestic
versus foreign).
2 Stage Estimation Procedure
STAGE 1: Decision to pay dividends
Estimated as binary 0/1 regression, ownership
and control variables
Options: probit, logit, linear probability, tobit..

STAGE 2: Conditional decision about the
size of dividend paid.
Classical regression model on a subset of firms
decided to pay dividends
Options: OLS, IV, GMM, or use Heckman
selection procedure (combination Stage 1)

How to Fix for
Missing Financial Data?
STAGE 0: Heckman selection procedure (“Heckit”),
inverse Mills ratio included in the first stage.
variables related to not reporting and optimally not used later
Size -- Total number of Shares (TNS) and NS offered
under the voucher privatization (NSVP)

Previously having problems with reporting: set of 0/1
indicators of missing financial data (profit,sales, debt,
and the number of employees) in priv. projects (91-93)
VP characteristics -- average price, total holdings (in
%) of the investment privatization funds and individual
investors.

How to Fix for
Endogeneity of Ownership?
 Hanousek, Kočenda, and Švejnar (2004): ownership
is endogenous with respect to corporate
performance
 Expect bidirectional link between ownership
structure and the decision to pay dividends
 Therefore we should account for possible
endogeneity of ownership:
 A) Use a reduced form equation (PROBIT) to
predict the type of the single largest owner
 B) Use linear probability model in STAGE 1 and
use standard IV techniques here
LPM is Optimal Choice for the
First Stage
 See Angrist and Krueger (2001) for a deep
discussion..
 The linear probability model allows to instrument
ownership and provides consistent estimates
under standard assumptions, while
probit regression with plugged predicted values of
ownership "do not generate consistent estimates unless
the nonlinear model happens to be exactly right, a result
which makes the dangers of misspecification high" (ibid).
 Also, the linear probability model can be
corrected for sample selection. (We redo the
first stage using probit as a robustness check.)
Estimation Technique:
STAGE 0
 STAGE 0: Heckman regression for missing
financial data
 Regression using I[missing] as a binary response
I [MissF ]  f (const , TNS , NSVP, MissF (91 / 93), AP, IPF , II )  
Estimation Technique:
STAGE 1
 STAGE 1: Decision to pay dividends (0-1 variable)
 Regression using I[Di,t > 0] as a binary response
I [ Di,t ]  f (const , OWN ( j ), .T9698 , DIV (t  1),. MILLS )  
We also included a set of control (financial)
variables and efficiency measures..
Additional Variables in STAGE 1:
 Control variables: total assets, total liabilities
to total assets, bank loans to total liabilities,
cash holdings to total assets, and the growth
rate of average sales in the industry the firm is
part of excluding the firm itself.
 Efficiency measures: profit (or total sales) to
total assets and total sales to total labor costs.
Estimation Technique: STAGE 2
 STAGE 2: Conditional decision about the size of
dividend paid
 OLS regression on a sub-set of firms paying
dividends (Di,t > 0)
Di,t  
j j j 
1 j 
Di,t1 
OWN
ji,t 
i,t 
j
1 TA i,t 2 TL 3 BL 4 CH 5 grSA i, t1 
t
TL i,t
TA i,t
TA i,t
MILLS i,t i,t
Results Summary: Main Specification
Marginal
Contribution
to Probability
To Pay
Dividends
Target
Payout
Ratio
Weight
Placed on
Current
Earnings
Czech
Foreign
Industrial
%
11
35
14
%
49
60
47
%
13
46
56
(Cz) Financial
(F) Financial
24
100
99
99
54
54
Results Summary: Main Specification, cont.
CZ Majority
CZ Monitored
Majority
F Majority
F Monitored
Majority
Marginal
Contribution
to Probability
To Pay
Dividends
Target
Payout
Ratio
Weight
Placed on
Current
Earnings
10
47
13
16
45
82
26
72
61
58
85
86
Results (Domicile & Concentration)
 A dominant majority owner pay dividends less
often and their target payout ratio is small.
 Checked majority owners pay dividends more
often and the target payout ratio is large.
 Dominant owners extract rents from firms and
that strong minority shareholders can prevent
this behavior.
 Interference with institutional framework
Robustness Checks
 LOGIT/PROBIT in Decision to Pay Dividends Eq.
 Earnings Measures: Operating profit, Profit
after/before income tax, Including/excluding
extraordinary items
 Control Variables: Industry dummies for growth
opportunities
 Earned Equity



Retained earnings / Total equity
Dividends before they are paid out
Coefficients in front of ownership dummies remain
significant and almost unchanged
Descriptive Statistics: Ownership Concentration
Mean
Total Assets
Dividend / Profit
Profit / Total Assets
Liabilities / Total Assets
Sales / Staff Costs
Std. Dev.
NOB
Majority
1,0E+06
7,9E+06
1 775
Monitored
1,4E+06
8,2E+06
2 235
Dispersed
1,9E+06
9,0E+06
1 866
Majority
0,040
0,68
1 775
Monitored
0,026
0,29
2 235
Dispersed
0,032
0,16
1 866
Majority
0,02
0,16
1 719
Monitored
0,04
0,24
2 204
Dispersed
-0,01
0,12
1 853
Majority
0,40
0,28
1 719
Monitored
0,63
0,36
2 204
Dispersed
0,35
0,24
1 853
Majority
8,00
37,29
1 719
Monitored
15,91
38,51
2 204
Dispersed
6,31
7,72
1 853
Descriptive Statistics: Nationality of SLO
Mean
Total Assets
Dividend / Profit
Profit / Total Assets
Liabilities / Total Assets
Sales / Staff Costs
Std. Dev.
NOB
Czech
1,0E+06
7,4E+06
5 786
Foreign
1,8E+06
7,6E+06
844
Czech
0,012
1,25
5 786
Foreign
0,068
0,27
844
Czech
-0,01
0,29
5 688
Foreign
0,05
0,15
827
Czech
0,48
0,35
5 688
Foreign
0,43
0,34
827
Czech
11,42
46,78
5 688
Foreign
9,62
19,13
827