Social-networks-corruption-and-collateral

Executive’s former banking experience,
entertainment expenditures and bank
lending decisions: Evidence from
China’s non-SOE firms
Gary Tian
Chinese Commerce Research Centre
School of Accounting and Finance
University of Wollongong
Coauthor Xiaofei Pan
1
2
Motivation


Heavily regulated credit market provides
increasing opportunities for exchanging rents with
bribes (Pei, 2008; Ngo, 2008; Cai et al., 2011)
thus firms seek rents by establishing connections
with government which can mitigate any financial
constraints and help them to access bank loans and
reduce the cost of borrowing (Cull and Xu, 2003;
Brandt and Li, 2003; Li et al., 2008; Faccio,
2010).
3
Motivation

but rent seeking literature provides mixed results:




Beck et al. (2006) argue that corrupt bank official is an
obstacle to firms raising external finance
Cai et al. (2011) find that bribery reduces firm
performance, while this effect is less pronounced in
regions with severe government intervention.
Chen et al. (2013) find corruption can improve the
efficiency of bank lending
using expenditure for entertainment and travelling
purposes (ETCs) as the measure of corruption and a
proxy for rent seeking activities is problematic.
4
Motivation


we propose a new measure to examine bank
lending decision making – firms’ social
network connections with banks serving to
reduce information asymmetry and
monitoring costs.
Adverse selection and moral hazard are the
main obstacle to access external capital
(Leland and Pyle, 1977; Sufi, 2007)
5
Social networks

Social network can reduce information asymmetry
and agency problem by lowering the costs of
acquiring necessary information about borrowers
and increase the availability of finances to them
(Peterson and Rajan, 1994; Beaver, 2002)

Recent evidence by Engelberg et al. (2012) find that
pre-existing personal relationship can alleviate
information asymmetry and reduce firms’ borrowing
costs
6
Social network in China


In the Asian context, social network can be
referred to as Guanxi network which is defined as
a special relationship between two parties for a
continued exchange of favours based on their
mutual benefits and interests (Alston, 1989; Chen
and Chen, 2004).
Guanxi requires the experience of interaction
through, for example, studying, working or living
together (Tsang, 1998)
7
Social network in this paper



We explore a specific type of social network with
banks through executives’ previous work
experience in banks.
Through these work experience, executives have
built personal relationship with bank managers
and could extend these relationships in other
banks through the networks they have built.
Trust established, then executives who posses
private information know how to communicate
more effectively and securely with banks.
8
Why China?




Chinese non-SOEs face discrimination of bank loans
and obstacle in accessing external finance
Emergence and vibrant growth of private sector in
China during the last decade.
How does private sector access external capital?
Cross-sectional variations of regional development
enables to examine the different roles and effects of
relationship
9
Why China ? (2)

Relationship-based business is prevalent in China.



About 23% of firms have bank connections, through
Chairman, CEO, CFO and other executives and
directors.
The mechanism through which social networks
help firms to obtain external finance and secure
bank loans.
Expenditure on entertainment through offering
gifts and dinner with bankers are essential
requirements (Fan 2002) to reinforce the
trust/bond relationship.
10
Why China? (3)

Collateral is used as the proxy for borrowing cost



First, collateral is a key ingredient used to enforce loan contracts as
a response to information asymmetry (the source of adverse
selection and moral hazard) (Besanko and Thakor, 1987; Boot et
al., 1991; Jimenez et al., 2006; Menkhoff et al., 2012).
Socend, since the recent global financial crisis of 2007, creditors
have expanded collateral requirements for their fund lending, and
this observed tendency has again attracted considerable attention
from academics and practitioners (Harrington, 2009).
Third, the interest rates charged on bank loans are relatively
regulated, so less endogeneity issue of collateral
11
Contribution

We propose explanations to the coexistence of
underdeveloped financial system and fast private
sector growth


Relationship-based economy (Allen et al., 2005). We
depart from political connection literature, by using
social network with banks which is more direct and
influential and suffer less endogeneity concern.
Entertainment as corruption (Cai et al., 2011; Chen et
al., 2013). Their measurement is problematic and
underestimate corruption. We depart from them using
entertainment
12
Contribution (Cont)

We add additional evidence to the literature of
borrower-lender relationship and its financial
implication (Engelberg et al., 2012)



Social networks can alleviate information asymmetry
and monitoring costs
Personal relationship between firms and banks can
facilitate bank loan access, reduce collateral
requirement and enhance bank lending efficiency
Non-SOEs use their existing social networks with
banks to be favoured by the institutional environment,
as well as justify their existence.
13
Hypothesis (1)




banks are more likely to obtain private
information and become less concerned about any
risk of defaulting and require less collateral.
Social network facilitates firm-specific
information flow to banks and reduce the
information asymmetry
H1a: Firms with social networks with banks have
better access to bank loans
H1b: Firms with social networks with banks have
lower collateral requirement
14
Hypothesis (2)



Banks become more informed and take
advantageous positions by obtaining inside
information;
facilitating banks more efficient scrutiny of loan
applications and help to better evaluate firm’s
future earnings.
H2: Bank lending is more efficient for firms with
social networks with banks
15
Hypothesis (3)



Social relationship can be reinforced by investing
in entertainment expenditures in eating, drinking,
gifts, Karaoke and sports club membership,
social network facilitates entertainment by
providing a channel through which entertainment
expenditure can be paid to maintain a better
relationship
H3: Social networks with banks facilitate
entertainment spending, which strengthens the
effect of social networks on firms’ access to bank
16
loans and reduces collateral requirements.
Hypothesis (4)


Cross sectional variations across administrative
regions in China. Different levels of government
intervention results in different levels of
information asymmetry (Rajan and Zingales,
1998)
H4: The effect of entertainment through social
networks on access to bank loans and reducing
collateral requirement is more pronounced in
regions with more government intervention.
17
Sample selection


All of non-SOEs listed firms in China from 2003
to 2010 from CSMAR.
Excludes:




Firms with ST and ST*
Financial industry
Firms with missing information
Final sample of 647 non-SOEs and 3302 firm-year
observations
18
Variable definition

Bank lending decision





Loan_size: Firm’s bank loan / Total debt
Long_size: Firm’s long-term loan / Total debt
Access_to_loan: dummy variable equal to 1 if
Loan_size is greater than 10%
Collateral: Collateralized loans / Total loans
Social network with banks

If at least one person in the top management team
(including the Chairman, CEO, CFO and other
executive and directors) was a former officer of a bank
19
Variable definition (Cont)

Entertainment spending


In the annual reports of China’s listed firms there is a
particular item recorded in the Statement of Cash Flow
called “Other Cash Payment for the Expenses Related
to Operating Activities”.
Two possible items related to entertainment that was
provided to cultivate and maintain social relationships.


Business entertainment expenses, business promotion expenses
In the empirical analysis: ETCs / Sales
20
Control variables
Return on sales (ROS)
Net income / Total sales
Return on assets (ROA)
Net income / Total assets
Firm size (Size)
Natural log of firm total assets
Cash-flow volatility
The volatility of cash flows for previous three years
Board size (Board)
Natural log of total number of directors on the board
Independent director (Indep)
Ratio of independent directors to total directors
Leverage (Lev)
Total debts / Total assets
Tangibility (Tang)
Net property, plant and equipment / Total assets
Sales growth (Sales)
Growth rate of sales for each year
Prime rate (Prime)
Prime lending rate set by the People’s Bank of China
Debt structure (Structure)
Long-term bank loans / Total bank loans
Cost of debt financing
(Interest expenses + capitalized interest) / Total debts
Guarantee
Guaranteed loans / Total debts
Age
Natural log of years since firm established
Employee
Natural log of number of employees
Duality
A dummy variable equal to1 if the CEO is also the
Chairman
21
Distribution of bank and political
connections
Table 2 Distribution of bank and political connections
Total
Firms with % of the Firms with
sample bank
total
political
number connection sample
connection
% of the
total
sample
Firms with both
bank connection
and
political
connection
% of the
total
sample
Panel A: By year
2003
2004
2005
2006
2007
2008
2009
2010
45.50%
43.90%
30.03%
26.13%
23.15%
25.29%
28.31%
20.72%
35
37
42
46
55
57
59
56
18.52%
15.04%
13.86%
12.27%
12.36%
11.18%
9.88%
8.79%
189
246
303
375
445
510
597
637
52
68
71
88
95
100
134
152
27.51%
27.64%
23.43%
23.47%
21.35%
19.61%
22.45%
23.86%
86
108
91
98
103
129
169
132
22
Difference tests between firms with and
without bank connections
With Networks
ETC
Loan_size
Long_size
Collateral
Firm size (million)
CF volatility
Return on assets
Return on sales
Board size
Independent director
Leverage
Tangibility (million)
Sales growth
Debt structure
Cost of debt
Guarantee
Age
Employee
1.55%
44.58%
9.56%
34.95%
2,470
8.15%
4.69%
8.53%
9.02
3.17
52.66%
1,440
121%
9.53%
8.35%
27.61%
7.18
2,477
Without
networks
1.32%
38.72%
8.41%
38.17%
4,430
8.12%
2.94%
5.85%
8.76
3.15
43.27%
1,280
121%
8.41%
8.36%
25.86%
7.08
2,670
Difference tests
0.23%(2.97)***
5.86%(5.77)***
1.15%(2.18)**
-3.22% (-2.02)**
-1,960(-2.64)**
0.03%(0.35)
1.75% (5.97)***
2.68%(4.09)***
0.26(3.26)***
0.02 (1.00)
9.39% (5.69)***
160(1.46)
0%(0.06)
1.12% (1.82)*
-0.01% (-0.61)
1.75% (0.91)
0.10(1.13)
-193(-0.83)
23
Regression of social network, entertainment
on loan access and bank lending efficiency
(ROS)
H1a
H3
H2
H2/3
Dependent variable
ROS
Network
ETC
Network*ETC
Network*ROS
ETC*ROS
Network*ETC*ROS
Sum tests
Adjusted R2
Observations
Access_to_loan
0.64***(6.20)
0.47***(6.44)
0.69**(2.19)
0.46**(2.26)
0.16(1.53)
0.15(0.25)
0.68**(2.48)
0.32**(2.27)
0.69***(2.86)
0.23(1.50)
0.70**(2.16)
2.23** a
2.02** b
0.18
0.23
3,302
3,302
Loan_size
0.48***(5.51)
0.28***(5.90)
0.07***(2.99)
0.05**(2.00)
0.03(0.66)
0.07(1.56)
0.11**(2.21)
0.06***(2.68)
0.33**(2.32)
0.75(1.47)
0.60**(2.27)
2.45** a
2.96*** b
0.25
0.27
3,302
3,302
24
Regression of effect of social network and
entertainment on collateral requirement
Dependent variable is collateral
-0.05***(-2.62)
H1b Network
ETC
H3
Network*ETC
Sum test a
Adjusted R2
0.33
Observations
1,815
-0.05**(-2.56)
-0.06**(-2.28)
0.34
1,815
-0.04**(-1.98)
-0.03(-1.09)
-0.09***(-2.92)
3.11***
0.34
1,815
25
Regression of effect of social network and
entertainment on long-term bank loan:
evidence of reduction of monitoring costs.
Dependent variable
ROS
Network
ETC
Network*ETC
Network*ROS
ETC*ROS
Network*ETC*ROS
Sum tests
Adjusted R2
Observations
Access_to_long
0.24**(2.57)
0.16*(1.80)
0.72(1.56)
0.52***(2.65)
2.42** a
0.18
3,302
0.09**(2.13)
0.10**(2.05)
0.72(1.62)
0.35**(2.45)
0.19**(2.45)
0.09(1.05)
0.63***(2.77)
2.59*** b
0.18
3,302
Long_loan
0.03**(1.97)
0.03**(2.46)
0.05(0.78)
0.10**(2.06)
2.26** a
0.19
3,302
0.02**(2.41)
0.02**(2.17)
0.05(0.81)
0.10**(2.05)
0.11**(2.01)
0.06(0.49)
0.06**(2.23)
2.04** b
0.19
3,302
26
Regression: effect of social network and
entertainment spending on bank loan and collateral
across regions (H4)
Dependent
Variable
Access_to_loan
Loan_size
Collateral
Good
Poor
Good
Poor
Good
Poor
Network
0.16
0.92**
0.03*
0.06**
-0.02*
-0.06**
(1.02)
(2.16)
(1.86)
(2.01)
(-1.83)
(-2.39)
ETC
0.27
0.11
0.16
0.01
-0.01
-0.08**
(1.10)
(1.55)
(1.04)
(0.02)
(-0.13)
(-2.46)
Network*ETC
0.20
0.35**
0.02
0.10***
-0.25
-0.13**
(0.65)
(1.98)
(0.12)
(2.79)
(-1.29)
(-2.17)
ROS
0.38***
0.57***
0.26***
0.33***
(4.50)
(4.01)
(5.23)
(3.44)
Network*ROS
0.67
0.70**
0.37
0.29**
(0.56)
(1.96)
(0.96)
(2.11)
ETC*ROS
0.23
0.16
0.50
0.77
(1.10)
(1.29)
(1.08)
(1.26)
Network*ETC*ROS
0.68
0.71**
0.23
0.76***
(0.23)
(2.23)
(0.31)
(2.85)
Each regression includes other control variables in our equation (1) and (2), such as firm age, firm size,
employee numbers, CEO duality, firm tangible assets, debt structure, cost of debt, cash flow volatility, ROA,
board size, independent director ratio, guaranteed loan, prime rate and year and firm fixed effects.
Sum test
1.38 a
2.98*** a
1.47 a
2.10*** a
0.99 b
2.71*** b
2
Pseudo R
0.17
0.19
0.08
0.15
0.37
0.33
Observations
1665
1637
1665
1637
919
896
27
Table 10. Effects of bank connection and political connection across industries
Full sample
Support industry
Panel A: Dependent variable is the access to bank loan
Network
0.51**(2.10)
0.45*(1.81)
Political
0.17***(3.19)
0.09(1.26)
Network*Political
0.59(0.70)
0.26(0.22)
a
Sum test
1.87*
1.55
2
Adjusted R
0.10
0.08
Observations
3,302
1,809
Panel B: Dependent variable is the bank loan ratio
Network
0.08***(4.60)
0.05(1.47)
Political
0.05**(2.37)
0.04(1.56)
Network*Political
0.02(0.45)
0.01(0.04)
b
Sum test
1.71*
1.31
2
Adjusted R
0.06
0.05
Observations
3,302
1,809
Panel C: Dependent variable is the collateral requirement
Network
-0.08**(-2.05)
-0.04**(-2.00)
Political
0.01(0.68)
0.05(0.67)
Network*Political
-0.02(-0.44)
-0.03(-0.45)
c
Sum test
1.03
1.00
2
Adjusted R
0.34
0.36
Observations
1,815
1,137
Non-support industry
0.57***(3.73)
0.23***(2.59)
0.12**(2.04)
2.48**
0.13
1,493
0.10***(4.58)
0.04**(2.55)
0.02**(2.33)
2.17**
0.09
1,493
-0.10**(-2.55)
-0.01(-0.50)
-0.02(-1.42)
1.60
0.34
678
28
Endogeneity issue

Endogeneity issue


Bank officials resigned their original positions and
acquired posts in better performing private firms for
their monetary and reputational concerns
Top executives in firms being discriminated against
with access to bank loans have more incentive to
appoint an executive with a social network with banks
in order to maintain a good relationship with banks and
help overcome the market barriers
29
Endogeneity issue

Change regression


Examine the effect of change in bank connection status
on change in bank loan finance, with a specific focus on
change accounts for time-invariant common
unobservable or omitted firm-specific characteristics
that might affect the social networks with banks and
firm’s bank loan finance.
We create two variables


NBCBC, equal to 1 if the firm’s status changes from
non-bank connection to bank connection
BCNBC, equal to 1 if the firm’s status changes from
bank connection to non-bank connection.
30
Change regression
Dependent variable
NBCBC
∆Loan_size
0.06***
(2.75)
BCNBC
∆ETC
NBCBC*∆ETC
0.03
(0.95)
0.30**
(2.05)
BCNBC*∆ETC
∆ROS
NBCBC*∆ROS
0.02
(1.63)
0.02**
(2.34)
∆Long_size
0.04*
(1.69)
-0.07***
(-2.75)
0.04
(1.12)
0.05
(0.70)
0.05
(1.36)
0.02
(0.73)
0.02
(1.33)
0.05*
(1.93)
0.03**
(2.73)
∆Collateral
-0.03**
(-2.27)
-0.02*
(-1.92)
0.03
(1.31)
0.08
(0.24)
-0.07
(-1.26)
0.06
(1.26)
0.03
(1.44)
0.13**
0.03
(2.15)
(0.82)
Each regression includes the change of other control variables in our equation (1) and (2)
Sum test
2.00** a
1.49 a
1.11 a
1.58 a
1.15 a
b
b
b
b
2.12**
2.64***
2.53**
1.67*
Pseudo R2
0.18
0.23
0.25
0.16
0.12
Observations
483
483
483
483
401
0.01
(1.48)
0.02
(0.44)
0.19
(1.32)
BCNBC*∆ROS
1.57 a
0.11
401
31
Endogeneity issue

We also apply the propensity-matching
method to address the endogeneity issue.
32
Further tests

Whether the effect of social network can be more
pronounced if bank connected executives come
from the lending banks


Redefining Network variable as equal to 2 for
Lender_network, 1 for Nonlender_network, and 0 for
no network. The results are broadly similar.
We divide top management into executives and
independent directors social networks

The coefficients on independent social network is
smaller, indicating incremental contribution.
33
Further tests (Cont)

We consider the number of bank connected
members


We repeat our analysis using political connection


We create one variable, Strength, equal to the number
of executives and directors who have the connections
with banks.
Results show the effect of political connection is less
significant than bank connection
We also repeat our regression by dividing total
sample into small firms and large firms
34
Findings

Social network with banks can:





lead to better access to bank loans;
Reduce collateral requirement;
Enhance bank lending efficiency.
Entertainment spending help to maintain social
network and further strengthen social network’s
effect
The above results are more pronounced after the
economic stimulus package and in the regions
with severe government intervetioin.
35
Findings


Social network and entertainment spending help to
explain the coexistence between a weak
institutional framework and vibrant private sector
growth in China.
Overall, social network with banks can be a
substitute for legal protection and help reduce
information asymmetry and creditor concerns and
monitoring costs, at least in the context of China.
36

Thank you and your comments
and suggestions are welcome.
37