Who’s afraid of universal banks?
Bank affiliations and corporate dividend
policy in pre-World War I Belgium
Marc Deloof
University of Antwerp
Annelies Roggeman
University of Antwerp
Wouter Van Overfelt
University of Antwerp
Universal Bank Interlocks of Belgian non-Financial Companies
Listed on the Brussels Stock Exchange in 1905
All Belgian Non-Financial Companies
Number of
Companies
Percentage of
Bank Interlocked
Companies
Average Number of
Interlocks for Bank
Interlocked
Companies
797
22%
1.89
18
165
132
23
46
45
16
37
57
46
164
2
18
12
16
11%
25%
23%
13%
11%
18%
13%
11%
12%
46%
29%
0%
11%
25%
0%
2.50
2.27
1.42
1.00
1.00
1.25
1.00
1.00
1.00
1.71
2.40
505
26
266
23%
50%
18%
1.70
2.38
2.18
Industry
Agriculture, forestry and fishing
Mining
Primary metal industries
Chemicals and allied products
Stone, clay, glass and concrete products
Textile
Food
Other manufacturing
Electricity and gas
Railways
Tramways
Other transportation
Wholesale trade
Real estate
Other
2.50
1.67
Location of Main Activity
Belgium
Belgian Colonies
Other Countries
2
Universal banks
• Regarding the economic effects of universal banks,
there are basically two different views
- View 1: universal banks are good 
• They are efficient institutions that overcome problems of
asymmetric information
• Universal bank relations are characterized by a multitude of
links which allow the bank to reuse costly information and to
build up technical expertise
- View 2: universal banks are bad 
• Universal bank relations with affiliated companies give rise
to conflicts of interest
• Universal bank relations allow the bank to loot the company
at the expense of other investors, especially in an
environment characterized by weak institutions and poor
investor protection
3
Research question
• In this paper, we focus on corporate dividend
policy
- Dividend play a central role in the relationship
between insiders and outside investors
- Dividends limit the scope for insiders to expropriate
value from outside investors, because they reduce
the inside cash in the firm, and they guarantee a
pro-rata payout to all shareholders
• How did universal bank affiliation affect dividend
policy of 428 Belgian companies listed on the
Brussels Stock Exchange in 1905-1909?
4
The setting
• Why is Pre-World War I Belgium an interesting
setting to investigate this research question?
- By modern standards and even by contemporary
standards, investors in Belgium were poorly protected,
and they faced severe information problems (cf.
Théate, 1905)
- Nevertheless, Belgium combined an active stock
market with a strongly developed banking sector
• According to Rajan & Zingales (JFE, 2003), more than the U.S.,
the U.K., Germany and Japan
• A limited number of universal banks played a central role in
Belgian finance
5
The setting
• Moreover …
- There were no dividend taxes → tax based
dividend theories are irrelevant
- The late 19th and early 20th centuries are
widely regarded as the first great era of
globalization, and it has been argued that there
are close parallels with world finance today
6
Related research on the same sample
• Van Overfelt, Annaert, De Ceuster & Deloof (WP,
2007)
- Universal bank affiliation was positively related to firm
performance
• Deloof & Van Overfelt (JBFA, 2008)
- Universal bank affiliation was negatively related to
debt ratios
• Van Overfelt, Deloof & Vanstraelen (WP, 2008)
- Universal bank affiliation was positively related to
corporate financial disclosure
7
The certification hypothesis
• The role of dividends
- In the period considered, capital markets in the
U.S., the U.K., Germany and Belgium were
flourishing, even though investor protection was
weak
- When investor protection is weak, dividends may
be a substitute for weak legal protection
- By paying high dividends, companies can
establish a reputation for moderation in
expropriating shareholders
8
The certification hypothesis
• Certification by universal banks
- In the period considered, (universal) banks played
a central role in establishing a reputation of
honesty and reliability
• In the U.S.: e.g. Carosso (1970)), De Long (1991),
Ramirez (JF, 1995), Baskin and Mirante (1997)
• In Germany: e.g. Fohlin (2006), Franks, Mayer and
Wagner (2006)
• In Belgium: universal banks contributed most of the
money for the financing of new securities, either by
investing in the securities themselves, or by selling the
securitites to the public (Durviaux, 1947)
9
The certification hypothesis
• Certification by universal banks
- Universal banks pressured companies to pay
sufficiently high dividends in order to create
trust among investors
- Firms affiliated with a universal bank paid
higher dividends than stand-alone firms
10
The certification hypothesis
• We also expect a positive relation between
dividends and investment opportunities
- companies with better investment opportunities
have a stronger incentive to establish a good
reputation
- If universal banks played a central role in
establishing a reputation of honesty and
reliability, and dividends signaled good
investment opportunities, the positive
relationship between dividends and investment
opportunities may have been stronger for
affiliated companies than for stand-alone
companies.
11
Alternative hypotheses
• Monitoring by universal banks
- Monitoring by the bank reduces the likelihood
that managers will behave in opportunistic
behavior
- Affiliated companies have less need to pay high
dividends and are less reluctant to cut or omit
dividends
• Cf. Goergen et al. (JCF, 2005) for Germany and Dewenter
and Warther (JF, 1998) for Japan
12
Alternative hypotheses
• Conflicts of interest
- The bank may use its power to protect its interests
as a lender, or even to expropriate value from the
shareholders
• The universal bank may protect its interests as a
lender by limiting dividends of affiliated companies.
• The universal bank may abuse its power to expropriate
value from shareholders (cf. La Porta et al., JF, 2000;
Faccio et al., AER, 2001)
• If affiliated companies pay lower dividends, this
provides more room for expropriation by the bank
13
Sample
• Full sample (all sectors)
- 663 stocks issued by 428 Belgian listed companies on the
Brussels Stock Exchange:
- Stockmarket data from the SCOB database
• Subsample (coal mining, trams, railways and textiles)
- 151 stocks issued by 109 companies in the four largest
sectors:
- Handcollected financial statement data from the ‘Annexes
au Moniteur Belge:’
14
Dividend measures
• Dividends per stock
- Dividend Cut (dummy)
- Dividend Payer (dummy)
• Dividends per company
- Dividend Payout Ratio
- Dividend/Total Assets
15
Measures of bank affiliation
• Bank Interlocks
- An executive board member of a universal bank is on the
executive or supervisory board of a particular company
- Five universal banks considered: (i) Société Générale, (ii)
Banque de Bruxelles, (iii) Banque Internationale de
Bruxelles, (iv) Banque d’Outremer, (v) Crédit Général
Liègeois
• Equity Stakes
- Industrial portfolio of the universal banks
- Not necessarely a better indicator than bank interlocks;
universal banks provided all kind of services to the
companies (provision of loans,underwriting new
securities,…)
16
Panel A: Number of Bank Interlocks and Number of Banks Interlocked
Number of Banks Interlocked →
Number of Bank
Interlocks
↓
1
1
19
2
10
3
3
3
4
1
8
4
1
3
1
5
Total
33
10
2
45
2
3
Total
19
13
Panel B: Company Interlocks and Equity Stakes for each Universal Bank
Universal Bank
Number of
companies
interlocked
Average
number of
interlocks
Number
of
equity
stakes
(Average)
percentage
of shares
held
Société Générale
21
2.00
11
21.67% (*)
Crédit Général Liégeois
18
1.56
2
19.24%
Banque d’Outremer
10
1.00
0
Banque de Bruxelles
6
1.00
1
Banque Internationale de
Bruxelles
4
1.25
0
3.74%
(*) The maximum percentage of shares held by the Société Générale is 47.62%
17
Determinants of dividend policy
• Control variables
-
‘Market-to-book’: measure for investment opportunities
Cash Flow/Total Assets
Ln(Total Assets)
Firm Age
Debt/Total Assets
Industry dummies
Year dummies
18
Univariate results for the full sample
• Dividend policy affiliated companies significantly
different from stand-alone companies
• Affiliated companies: more likely to pay dividends,
less likely to cut or omit dividends, keep dividends
more constant
• Affiliated companies pay higher dividends
• These results are consistent with the ‘certification’
hypothesis
19
The Effect of Stock Return Performance on the
Decision to Cut Dividends (full sample)
Dependent Variable:
Dividend Cut
Estimation method:
Random Probit
Sample:
663 stocks
Constant
-0.470***
Bank Interlock
-0.108
Stock Market Return (t) x Bank Interlock
-0.898
Stock Market Return (t) x Stand-Alone
-0.770
Stock Market Return (t-1) x Bank Interlock
Stock Market Return (t-1) x Stand-Alone
Stock Market Return (t-2) x Bank Interlock
-2.382**
-3.212***
-1.056**
Stock Market Return (t-2) x Stand-Alone
-0.043
Number of Observations
1,312
20
(2)
(3)
(4)
Dividend
Payer
Dividend
Payout
Ratio
Dividends /
total assets
Estimation method:
Random
Probit
Random
Tobit
Random
Tobit
Sample:
151
stocks
104
companies
104
companies
0.957
(0.184)
-
0.041
(0.554)
-
-0.005
(0.689)
-
0.269
(0.588)
15.986***
(0.004)
1.596***
(0.000)
-0.016
(0.546)
-7.317***
(0.000)
0.064
(0.230)
-0.528
(0.173)
0.150***
(0.001)
0.002
(0.393)
-0.904***
(0.000)
0.018**
(0.024)
0.387***
(0.000)
0.014
(0.133)
0.000
(0.284)
-0.090**
(0.032)
Dependent Variable:
Bank Interlock
Bank Interlock x MtB
Market-to-Book
Cash Flow / Total Assets
Ln(Total Assets)
Age
Debt / Total Assets
21
(5)
(6)
(7)
Dividend
Payer
Dividend
Payout
Ratio
Dividends /
total assets
Estimation method:
Random
Probit
Random
Tobit
Random
Tobit
Sample:
151
stocks
104
companies
104
companies
-1.096
(0.613)
1.435
(0.475)
0.214
(0.505)
14.578**
(0.024)
1.630***
(0.000)
-0.033*
(0.075)
-7.213***
(0.000)
-0.162
(0.178)
0.095**
(0.039)
0.039
(0.385)
-0.482
(0.144)
0.141***
(0.005)
0.001
(0.692)
-0.953***
(0.000)
-0.022
(0.173)
0.016**
(0.039)
0.011
(0.225)
0.372***
(0.000)
0.010
(0.180)
0.001**
(0.027)
-0.111***
(0.002)
Dependent Variable:
Bank Interlock
Bank Interlock x MtB
Market-to-Book
Cash Flow / Total Assets
Ln(Total Assets)
Age
Debt / Total Assets
22
Alternative explanations
• Universal banks as outside investors
- Universal banks may basically have been
outside investors, demanding high dividends in
order to force companies to go to the capital
market for external financing and be subject to
monitoring by the external market
- Companies in which the bank has an equity
stake will pay higher dividends
- Companies which have a bank director but in
which the bank does not have an equity stake
will not pay higher dividends
23
(8)
(9)
(10)
Dividend
Payer
Dividend
Payout
Ratio
Dividends /
total assets
Estimation method:
Random
Probit
Random
Tobit
Random
Tobit
Sample:
151
stocks
104
companies
104
companies
1.253*
(0.090)
-0.931
(0.343)
-
0.054
(0.517)
-0.107
(0.305)
-
0.023*
(0.097)
-0.072***
(0.001)
-
0.274
(0.518)
16.410***
(0.006)
1.631***
(0.000)
-0.014
(0.496)
-7.485***
(0.000)
0.062
(0.240)
-0.501
(0.199)
0.148***
(0.002)
0.002
(0.327)
-0.911***
(0.000)
0.018**
(0.045)
0.374***
(0.000)
0.020*
(0.066)
0.001
(0.119)
-0.100***
(0.006)
Dependent Variable:
Bank Interlock
Bank Equity Stake
Number of Bank Interlocks
Market-to-Book
Cash Flow / Total Assets
Ln(Total Assets)
Age
Debt / Total Assets
24
Alternative explanations
• Fear of universal banks
- Affiliated companies may need to put more effort
in “seducing” investors, because investors fear
expropriation by universal banks.
- From this point of view, it can be expected that
dividends will be positively related to the degree
of bank control: the stronger the control of the
bank over the company, the more investors
should fear expropriation by the bank.
25
(11)
(12)
(13)
Dividend
Payer
Dividend
Payout
Ratio
Dividends /
total assets
Estimation method:
Random
Probit
Random
Tobit
Random
Tobit
Sample:
151
stocks
104
companies
104
companies
0.991
(0.333)
-
0.069
(0.501)
-
0.046**
(0.035)
-
-0.018
(0.968)
0.270
(0.562)
15.993**
(0.021)
1.600***
(0.000)
-0.016
(0.442)
-7.323***
(0.000)
-0.016
(0.741)
0.062
(0.167)
-0.543
(0.119)
0.153***
(0.005)
0.002
(0.291)
-0.940***
(0.000)
-0.014*
(0.065)
0.015
(0.125)
0.377***
(0.000)
0.015*
(0.079)
0.001
(0.129)
-0.103**
(0.021)
Dependent Variable:
Bank Interlock
Bank Equity Stake
Number of Bank Interlocks
Market-to-Book
Cash Flow / Total Assets
Ln(Total Assets)
Age
Debt / Total Assets
26
Conclusions
• Our results are generally consistent with the
hypothesis that companies affiliated to a universal
bank used dividends as a tool to convince investors
of their honesty and reliability
• Companies with several bank directors and
companies in which the bank had an equity stake
tended to pay lower dividends. We therefore
cannot reject the hypothesis that banks extracted
rents from companies they controlled.
27