Relationship Lending and Firm Innovation: The Case of GEM Firms

Relationship Lending and Firm Innovation: The Case of
GEM Firms in China 1
Xue Wang 2, Yanan Zhang 3, Han Luo 4
Southwestern University of Finance and Economics
May,2013
1
Supported by the Fundamental Research Funds for the Central Universities (JBK130209), and Ministry of
Education Humanities and Social Science Research Projects Funds (12YJC630218). I wish to thank Jimmy Ye,
Yongqiang Ma, Qiulan Han, Huifang Fu, Rong Quan, Fang Yang for useful discussions and suggestions. All errors
are my responsibility.
2
Xue Wang: Postal address:School of Accounting, Southwestern University of Finance and Economic, Chengdu,
Sichuan, China, 610074. E-mail address: [email protected]. Telephone number: 086-13699032599, fax
number 086-028-87092405
3
Yanan Zhang: Postal address:School of Accounting, Southwestern University of Finance and Economic, Chengdu,
Sichuan, China, 610074. E-mail address: [email protected]. Telephone number: 086-15828502715
4
Han Luo: Postal address:School of Accounting, Southwestern University of Finance and Economic, Chengdu,
Sichuan, China, 610074. E-mail address: [email protected]. Telephone number: 086-15882093491
1
Abstract
Using a sample of Chinese GEM listed companies in 2007-2011, we study the
determinants of company innovation through relationship lending. In bank choice
of loan type anticipate the inside bank’s strategic use of information and act
accordingly. Our findings suggest: Relationship Debt is an effective governance
mechanism for firm R&D expenditures; firms with higher R&D intensity will
tend to have higher relationship debt ratio, and higher relationship debt ratio will
increase the firm innovation efficiency; Firm size will provide significant
influence toward this relation, which is to say, relationship debt is more important
in small and medium firms than in large firms.
Keywords: Relationship financing, R&D intensity, Firm size
2
1 Introduction
The Chinese government encourages firm innovation for sustainable economic
growth. The importance of technological innovations for economic growth has been
well established (Solow,1957). However, there are many determinants of firm
innovation, including culture, IP protection, corporate governance, and capital market
and so on. For the majority of firms the main obstacle inhibiting innovation is
represented by financial factors. Lacking the visibility of more established firms,
young and small firms are likely to suffer even more for the financing of their
investments because of asymmetric information problems (Berger and Udell, 1998).
Despite theoretical works have already emphasized complex links between
finance and innovation (e.g., O’ Sullivan, 2004; de la Fuente and Marin,1996), and
notwithstanding current richness of enterprise-based innovation surveys, there is little
in the existing empirical micro-economic literature on the functions of the various
sources of funding across different innovation firms. Though researchers have argued
theoretically that there is a link between financing and innovation (e.g., Aghion and
Tirole, 1994), there is little in the extant literature about the relationship between
financing decisions and the creation of significant innovations.
We address this gap in the literature by investigating whether the innovative
activity of relatively established firms is related to their financial arrangements. Our
basic hypothesis is that certain external financing arrangements are more conducive to
innovative activity than others. Specifically, we hypothesize that for GEM firms,
relationship based financing, such as bank loans will be associated with significantly
more innovative activity than arm's length financing such as public debt and equity.
Our hypothesis is based on arguments about the possible impact of the two types of
financing on managerial incentives to produce innovations.
Traditionally, in relationship based financing, the investor or lender acquires
significant information about the firm, which reduces information asymmetry and
allows for closer monitoring of the firm's managers. The downside, however, is that
the decision to provide and sustain financing depends on the lender's ability to value
the firm's projects. The lender, represented by a bank loan officer, may lack the
necessary skills to evaluate investments in an innovative technology (Scherer,1984).
As a consequence, relationship based lenders will discourage managers from
investing in innovative projects and be more willing to shut down ones that are
ongoing (Rajan and Zingales, 2003). In comparison, arm's length financing such as
public debt and equity gives managers more discretion to invest in innovative
technologies. Managers have stronger incentives to pursue uncertain but potentially
breakthrough innovations since they are less concerned about being denied
refinancing and shut down, as they might with bank loans. Consequently, we would
expect a firm's innovation strategy and type of financing to be determined with
innovative firms choosing arm's length financing while firms with less innovative
projects choose bank borrowing.
However, this situation might be different in China. Since debt is a critical
source of funds for most firms, and strategic R&D investment is an importance use of
3
funds, the relationship between them is crucial. Unlike the USA and many other
countries, China has a long tradition of relational banking. The bank-firm relation in
China has been changed a lot during market economy transform, however, the nature
of bank loan is still complicated. The Chinese banks are trying to build a new
bank-firm relationship based on Japan practice (Yu,2003; Xu and Xiao,1999), and
thus build long-term relationship with firms. In recent years, many Chinese banks
have built strategic cooperation agreements with group corporations. Nevertheless,
many group corporations hold the shares of listed banks as well. According to Wen,
Feng and Liu(2011), 76% of sample listed companies have not changed their main
bank loan providers for 5 to 10 years. Banks become the major creditors and fund
providers for Chinese listed companies. According to (Yu, 2003), liabilities in China
have a strong similarity with equity, banks may be considered as “Sub-shareholders”.
These evidences indicate that relationship debt such as bank loan works more
efficiently than transactional debt in China.
The paper is structured in the following way. The next section gives an overview
of the literature, and a brief description of the empirical determinants of relationship
lending, explores its possible links with firm innovativeness. Section 3 presents the
dataset and the main descriptive statistics on the degree of firm innovativeness.
Section 4 presents the model, empirical results and robustness test. The final section
summarizes the paper.
2 Literature review
A lively macro-economic debate on the role of financial architecture in fostering
innovation and technology is the one on the bank-based versus market-based system
(i.e.,Carlin and Mayer, 2003; Levine,2002). Are bank-based systems at the advantage
in processing information particularly relevant for firms’ incentive to innovate? The
available evidence is rather mixed but findings suggest that market-based systems do
not dominate bank-based systems and vice-versa in all times. However,
knowledge-intensive industries, with soft, hard-to-monitor complex activities seem to
get on better in bank-based financial systems (Tadesse, 2007).
Even though researchers have argued theoretically, and tested empirically, that
there is a link between finance and innovation, there is still little in the existing
micro-economic literature about the functions of the various sources of funding in the
different phases of innovation (O’ Sullivan, 2004). The main contribution of this
paper is to make another step in this direction, enhancing the understanding of the role
played by bank ties in firm innovation.
O'Brien(2003) argue that consideration of firm strategy can help illuminate the
choices managers make between debt and equity financing. Within an industry, the
form of competition that each firm chooses will determine the strategic value to the
firm of maintaining financial slack. Empirical analysis yields strong support for the
proposition that financial slack should be a particularly critical strategic imperative
for firms pursuing a competitive strategy premised on innovation.
Boot(2000) reviews the contemporary literature on relationship banking. He
4
discussed how relationship banking fits into the core economic services provided by
banks, and its costs and benefits. This leads to an examination of the interrelationship
between the competitive environment and relationship banking as well as a discussion
of the empirical evidence.
There is a vast literature on bank-firm relationship literature based on U.S. and
Europe background. Using a large panel of US companies, Atanassov et al. (2007)
explore the relationship between arm’s length financing and innovation taking patents
as a measure of innovative output. They found that firms that relied more on arm’s
length financing are associated with a larger number of patents. They also conclude
that innovative firms choosing their capital structure mainly drive this correlation.
Relying on firm-level data from a survey conducted in Finland, but looking instead at
the role of public policy, capital-market imperfections delay innovation, and
government funding disproportionately helps firms in industries that are more
dependent on external finance. They used as a measure of firm innovativeness the
level of R&D expenditure. The first difference between relationship and arm's length
financing that may affect managerial incentives to pursue innovative investments is
the role of information acquired by the lender or investor. More specifically, in the
process of making loans, banks acquire significant non-public information about a
firm's financial condition and investment plans. This is done when the bank
investigates the firm at the time of lending and subsequently through its monitoring.
In contrast, arm's length investors do not usually acquire significant private
information about the borrower.
The advantage of relationship financing is that the bank's information acquisition
will reduce the usual information problems between the firm and its capital provider
(Diamond,1984). More specifically, banks compare the information they have
acquired about the firm's projects to their own criteria of what a good project should
be. These criteria are based on the bank's previous experience and knowledge. While
such criteria are helpful in assessing routine/incremental projects, banks are more
prone to errors when more novel technologies that lie outside of the previous
experience of the bank are involved (Scherer,1984). As a consequence, relationship
based lenders will discourage managers from investing in innovative projects.
Using similar arguments, Rajan and Zingales(2003) suggest that relationship
financing may be an optimal choice for firms with innovative projects. With arm's
length financing, managers have greater incentive to pursue uncertain but potentially
breakthrough innovations, as they might with relationship based financing in the form
of bank loans. In a similar spirit, Aghion and Tirole [1994] argue that if the lender
doesn't have much knowledge about the firm's projects, it is optimal to give more
discretion to the firm's manager to encourage her initiative.
The second difference between the two types of financing that might impact
managerial incentives to pursue innovative investments is that in relationship
financing, unlike arm's length financing, the lender may be more willing to
renegotiate the terms of a loan due to his private information about the firm's future
profitability. It may be hard to renegotiate the terms when there are a large number of
small investors, as is the case with public debt or equity. While renegotiation can have
5
benefits for the firm in certain situations (e.g., restructuring), Cremer(1995) argues
that the possibility of renegotiation with the relationship financier at a later date
weakens the ex-ante effort exerted by the manager. This is more of a problem when
the ex-ante uncertainty about success of the projects is high, as is the case with radical
R&D projects (Rajan and Zingales, 2003).
However, most studies are based on U.S. institutional background, and cannot
provide satisficing answers to firm innovation behavior in some countries where bank
has certain power, such as Japan and Germany. As Hoshi and Kashyap (2001)
explained, Japanese capital markets were historically highly regulated, and firms
relied almost exclusively on relational lenders for debt. David et al (2008) surveys a
sample of Japanese firms over 20 years of deregulation, and uses transaction cost
economics to develop a model relating debt structure to R&D intensity. Debt is
heterogeneous, with two distinct types: transactional debt (fixed time-periods and
simple attributes) and relational debt (extended duration and complex attributes). The
two types contrast in that transactional debt relies on market governance and cannot
give strong enough exchange safeguards for R&D investment, whereas the
hierarchical governance provided by relational debt aligns the interests of investors
and managers in R&D-intensive firms.
From the arguments above, it follows that a firm's innovation strategy and choice
of financing are determined in equilibrium. Accordingly, given Chinese background,
our main hypothesis is that firms with more innovations will optimally choose a
capital structure that relies more heavily on relationship lending in the form of bank
loan. Hence, our first empirical prediction is:
Prediction 1: firms with relatively more relationship financing such as bank loan
will have higher innovation intensity.
As transactional debt is purchased and traded in public markets, public disclosure
of information is needed. Relational debt, however, is a private transaction between a
firm and a lender and does not require public disclosure, thus helping safeguard
proprietary information about R&D from leakage.
Therefore, relational debt provides stronger appropriability safeguards for
investments in R&D. Misalignment between debt structure and R&D can impose
unnecessary costs, distort incentives, and impair firm performance. Firms with low
R&D intensity are appropriately governed by the high incentives and less invasive
administrative and monitoring mechanisms of transactional debt. These firms do not
need strong safeguards and would incur unnecessary transaction costs by utilizing
relational debt. Conversely, R&D-intensive firms incur higher exchange hazards and
require stronger safeguards. Although relational debt entails weaker performance
incentives and higher monitoring and administrative costs, it provides the stronger
governance safeguards necessary for R&D-intensive firms. Furthermore, the rigidity
of transactional debt can disrupt R&D programs or delay new-product launches when
cash shortfalls are experienced (O’Brien, 2003). Even during good times, anticipating
that transactional lenders will be unlikely to forbear in the event of a downturn can
undermine managerial commitment to risky long-term R&D projects. Additionally,
keeping transactional lenders abreast of strategic initiatives would raise the risks of
6
information leakage.
Prediction 2: firms’ relationship financing ratio and innovation intensity will
interact positively with respect to their impact on corporate performance.
The arguments used to develop our hypothesis suggest that the relationship
between type of financing and innovation should be stronger for firms for whom bank
loan is important. The reason is that many big firms, such as those large scale SOEs,
they have multiple ways of financing, including bank loans from several major banks.
Small and medium firms, on the contrary, they rely mostly on bank loans, and they
are expected to disclose more information to obtain the loan. The banks will have a
much stronger position in supervise the Small and medium firms operations, and thus
the agency problems caused by the information asymmetry will tend to be reduced.
Accordingly, our next prediction is:
Prediction 3: the relationship between relationship financing and innovative
activity will be stronger for firms with smaller size.
3 Research Design
(1) Sample Collection
According to the Chinese Accounting Standards, listed companies are obliged to
disclose related R&D information, and both the revenue expenditure and capital
expenditure amount should be disclosed at Notes to the financial statements.
However, the current situation in China is the detailed R&D information is still
not a mandatory disclosure requirement in China. The R&D information disclosure is
varied among listed companies. Mostly, we can find R&D information in the
following three ways:
(a) Balance sheet : “Development Expenditure”
(b) Notes to Cash Flow Statement: “Cash flow in other operational activities ”
(c) Notes to Income Statement: “Detailed Managerial Expense ”
Among the three ways of R&D disclosure, we do not use the data from the
“Development Expenditure” in Balance Sheet because this item reveals the
capitalized part of R&D expenditures but not yet become intangible assets at the end
of period. Also, we define R&D expenditure in a more traditional way by excluding
the equipment upgrading, production line upgrading and purchased intangible assets.
Thus, our data is mainly collected from notes. The R&D expenditure items vary
among listed companies, any related names are considered to be R&D expenditure,
such as “R&D expense”, “research expense”, “technology expense”, “scientific
expense”, “consulting and development expense ” and so on.
We collected R&D expenditure information by annual reports and director’s
reports of China GEM board (ChiNext 5 and SME Board 6)in 2007-2011, 310 listed
5
ChiNext market provides an important platform for implementing the national strategy of independent
innovation. It helps accelerate the transformation of economic development mode and galvanizes growth in
emerging industries of strategic importance. ChiNext was inaugurated in Shenzhen on 23 October, 2009. As of 30
December, 2011, there were 281 companies listed on the ChiNext, of which 93% are hi-tech firms. The total
market capitalization of ChiNext listed companies reached RMB 743.4 billion (USD 118 billion). IPO proceeds hit
RMB 196.1 billion (USD 31.1 billion). Total trading value of ChiNext was RMB 1.9 trillion (USD 301.6 billion) in
7
companies have R&D disclosure, and we get 1232 firm year observations, in which
833 firm year disclose the total R&D expenditure, 128 firm year disclose revenue
expenditure and 380 firm year disclose capital expenditure.
The financial data comes from CSMAR (China Stock Market Trading Database),
some data are collected from Cninfo Platform, and industry classification standards
are from CRSC. We use Stata10.0 and SPSS 12.0 for this paper.
(2) Variables
Dependent variables: (1)Relationship financing variable is R/B. Following
prior research (Anderson & Makhija, 1999; Hoshi, Kashyap, & Scharfstein, 1993; Wu,
Sercu, & Yao, 2001), we considered all bank loans to be relational debt and all bonds
to be transactional debt. As Aoki and Patrick (1994) noted, a main bank leads a de
facto lending syndicate and monitors firms on behalf of all other lenders, so it is
appropriate to treat all bank loans as relational debt. The variable R/B represents the
sum of all bank loans divided by total debt, where total debt is the sum of all bank
loans and all bonds and note payable outstanding. (2)Corporate performance variable
is M/B 7. To serve as a proxy for a firm’s performance, we used its market-to-book
ratio. This measure, which closely corresponds to Tobin’s Q (Chung & Pruitt, 1994),
was appropriate because it incorporated not just current performance but also
expected future performance. This measure was calculated as the sum of total debt
and the market value of equity divided by total assets. In China, M/B value is
calculated in two ways, because in China the market value of equity has two parts:
non-tradable shares and tradable shares. For M/B1, we use average market value for
tradable shares and bonds, and book value for non-tradable shares and other debts.
For M/B2, we use average market value for tradable shares, non-tradable shares and
bonds, and other debts are still calculated with book value.
Independent variables: (1) R&D intensity, was calculated as total research and
development expenditures divided by total sales. (2)Scale, is a dummy variable, we
use 1 for small and medium firms, and 0 for large firms. We define any firm with
assets size less than 1.5 billion yuan to be small and medium firm. This classification
is similar to Berger and Udell(1995), and Acs et al (1997), and China context is
considered. The scale classification ignores wether the listed companies are from
ChiNext or SME board of China.
Controlling variables: (1) Firm size. The variable lnsize is used the natural log of
firm total assets to measure firm size. (2) For profitability we use ROA, or return on
assets, was operating income divided by total assets. (3) For operation capacity, we
2011.ChiNext Market promoted allocation of social funds to innovative businesses and emerging industries.
6
The launch of the SME Board was a major step toward the establishment of a multi-tier capital market system
and paved the way for a second board market. The SME Board has witnessed continuous expansion in size and
gradual optimization of its industrial structure thanks to many high quality innovative issuers. The SME Board
highlights its role in supporting independent innovation. As of 30 December, 2011, there were 646 companies
listed on the SME Board with a total market capitalization of RMB 2.7 trillion (USD 428.6 billion). Total IPO
proceeds were RMB 558.8 billion (USD 88.7 billion). Total trading value stood at RMB 6.9 trillion (USD 1.1 trillion)
in 2011.
7
According to David et al(2007), the variable performance is taken the natural log of the market- to-book ratio.
Results were similar if performance was not logged.
8
use RAA to measure fixed assets turnover. (4) For solvency, we use COV to measure
times interest earned. (5) For growth, we use GMP for the growth rate of main
operations. (6) According to many literature with Chinese context, we do believe that
the nature of property rights do effect firm innovation (Feng and Wen, 2008; Li and
Song, 2010; Wang, 2011), thus we use SRS for the stated-owned shareholdings
divided by total top 10 shareholdings. (7) Although all types of debt have both costs
and benefits (see Harris & Raviv, 1991), our focus was on whether firm strategy
influences the type of debt a firm selects, controlling for the absolute amount of debt.
Hence, we controlled for overall firm leverage, which was defined as total debt
divided by firm market value (the sum of total debt and the market value of equity).
Furthermore, in our performance models, we also controlled for the interaction
between between R&D intensity and leverage (O’Brien, 2003).Finally, in addition to
the firm-level control variables described above, we also included a number of
industry-level control variables. For each industry, we used the median values for the
corresponding firm-level variable measured in all firms for which that industry was
primary to measure industry performance(In_M/B), industry leverage(In_leve), and
industry profitability(In_ ROA).
(3) Descriptive Statistics
The listed companies in GEM have an average 0.0508 R&D Intensity, which
reveals a large gap comparing U.S practice. This data is higher than the China
Enterprise Innovation Report 2010 published by China Enterprise Evaluation
Associsation (CEEA), and most of the results done by other researchers (e.g. Wen,
Feng and Liu, 2009; Li and Song, 2010) due to different sample selection. The
average R&D intensity of Chinese listed companies is around 0.03-0.04.
Table 1: Detailed R&D Intensity Disclosure Information
Industry Name
Industry
Sample
Average R&D Intensity
A0
17
0.0369
Mining
B5
7
0.0609
Food and beverage
C0
8
0.0355
Textiles, garments, leather
C1
22
0.0211
Furniture
C2
4
0.0175
Paper making,printing
C3
19
0.0191
Petroleum, Chemical, plastics
C4
104
0.0322
Electronics
C5
93
0.0494
Metals, non-metallic
C6
53
0.0339
Machine, equipment,instruments
C7
206
0.0477
Medicine,Living creature
C8
72
0.0531
Other manufacturing industries
C9
23
0.0302
Electrical power, coal gas and
D0
4
0.0089
Code
Agriculture, forestry, cattle
breeding, fishery
products
water production and supply
9
Building industry
E0
6
0.0267
F0
2
0.0053
Storage
F2
1
0.0062
Information technology
G8
145
0.0981
Retail
H1
4
0.0282
Business brokers and agents
H2
3
0.0045
Scientific Service
K2
19
0.0444
K3
1
0.1010
K9
9
0.0405
Information dissemination
L2
3
0.0045
Integrated industry
L9
3
0.0006
Total
M
833
0.0508
In general, the average R&D expenditure in Small and medium firms are
significant higher than in large firms. The average Relationship debt ratio in Small
and medium firms is 0.7832, significant higher than the 0.7704 in large firms and the
low bond issuing rate in GEM listed companies might provide an explanation for this
difference. In general, the descriptive statistics results provide primary support for our
predictions.
Table 2: Descriptive Statistics of Variables
Variable
Max
Min
Median
Average
L
0.3314
0.0000
0.0356
0.0401
0.0404
S&M
0.7561
0.0000
0.0406
0.0544
0.0281
Total
0.7561
0.0000
0.0397
0.0508
0.0394
L
1.0000
0.0000
0.7428
0.7704
0.2791
S&M
1.0000
0.0000
0.6458
0.7832
0.2979
Total
1.0000
0.0000
0.7569
0.7568
0.2859
M/B
Total
10.2441
0.5841
1.606
1.338
0.9515
Insize
Total
10.20
8.112
9.004
9.01
0.0158
ROA
Total
-0.150
0.612
0.106
0.123
0.0971
RAA
Total
5.2984
0.0653
0.5157
0.7061
0.9542
COV
Total
30.48
2.04
9.51
13.21
0.7364
GMP
Total
1.569
-0.436
0.175
0.212
0.8467
SRS
Total
0.811
0.000
0.263
0.249
0.2211
Leve
Total
0.732
0.173
0.483
0.439
0.1845
R&DInt
R/B
S.E.
T test
24.38***
39.41***
3.206**
6.888***
Notes: ***significant at 1% level.
4 Empirical Analysis
The empirical analysis of this paper have two parts: the first part is regression on
relationship debt ratio and R&D intensity, and the other part is regression on
corporate performance and multiple interactions between R&D intensity and
10
relationship debt ratio.
(1) Prediction 1 Model
We use Hausman test to choose between fix effects model and random effects
model, and the Hausman test results suggests that fix effects model is proper for
testing the regression on R&D intensity and Relationship financing ratio. Also,
dynamic models is suggested for this regression related with corporate governance
factors (Green, 2007), and we use a partial adjustment model based on (Wen, Feng
and Liu, 2009) models. Model (a) is a partial adjustment model, model (b) is a
transition model, and model (c) is the final regression model.
∗
𝑅/𝐵𝑖,𝑡 − 𝑅/𝐵𝑖,𝑡−1 = 𝛿(𝑅/𝐵𝑖,𝑡
− 𝑅/𝐵𝑖,𝑡−1 )
∗
𝑅/𝐵𝑖,𝑡
= 𝛽1 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 + 𝛽2 𝑆𝑐𝑎𝑙𝑒 × 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 + 𝛽 ′ 𝑋 + 𝜇𝑡 + 𝜇𝑖 + 𝜀𝑖𝑡
(a)
(b)
𝑅/𝐵𝑖,𝑡 = (1 − 𝛿)𝑅/𝐵𝑖,𝑡−1 + 𝑧1 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 + 𝑧2 𝑆𝑐𝑎𝑙𝑒 × 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 + 𝑍 ′ 𝑋 + 𝜇𝑡 + 𝜇𝑖 +
𝜔𝑖𝑡
(c)
In the models, i represents firm, t represents year, 𝜇𝑡 and 𝜇𝑖 represents period
and individual fix effects, 𝜔𝑖𝑡 and 𝜀𝑖𝑡 are random disturbance terms, and 𝛽 ′ 𝑋
represents the product of controlling variables vector and their regression coefficients.
Model (c) is a dynamic panel data regression model. We cannot eliminate the
correlation between lag dependent variables 𝑅/𝐵𝑖,𝑡−1 and error 𝜀𝑖𝑡 , since LSDV
estimate for fix effects and GLS estimate for random effects are biased, we take
𝑅/𝐵𝑖,𝑡−2 as the instrumental variable for𝑅/𝐵𝑖,𝑡−1. Instrumental variable is used here
for two reasons: first, this explanatory variable is not correlated to𝜀𝑖𝑡 , thus we don’t
have random explanatory variable problem; secondly, 𝑅/𝐵𝑖,𝑡−2is interpretative and
highly correlated to 𝑅/𝐵𝑖,𝑡−1.
(2) Prediction 2 Model
We use M/B to measure corporate governance variable, because M/B ratio may
respond quickly to the market reaction. We use the following regression model to
study the corporate performance and the cross production of R&D intensity and
Relationship debt ratio (David et al,2008; Wang,2003).
𝑀/𝐵𝑖,𝑡 = β1 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 + β2 𝑅/𝐵 × 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 + β3 𝑆𝑐𝑎𝑙𝑒 × 𝑅/𝐵 × 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 + β′ 𝑋 +
𝛼𝑡 + 𝛼𝑖 + 𝜀
(d)
𝛼𝑡 , 𝛼𝑖 are the fix effects of period and individual, and other variables have the
same applications with model (c). In model (d), 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 may have endogenous
problems, for higher 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 might cause higher 𝑀/𝐵𝑖,𝑡 and vice versa (Bhagat
and Welch, 1995). We run an endogenous test for model (d): first, we use 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡
to run regression with all the exogenous variables and instrumental variables to find
the predicted value of 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 and residual 𝜇𝑖 ; secondly, we run regression on the
original model using all exogenous variables, 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 and 𝜇𝑖 , and the t value of
𝜇𝑖 is not significant, thus we can reject the endogenous problem.
11
(3) Empirical results
The main results are shown in Table 2. Model (1) is regression on controlling
variables and relationship debt. Model (2) is regression on R&D intensity and
relationship debt, model (3) is to examine the scale impact on the correlation between
R&D intensity and relationship debt, model (4) and (5) is to examine prediction 2
with M/B1 and M/B2. The regression results of model (1) indicates that firm size,
ROA, plant assets turnover and industry ROA and industry debt structure have
significant impact on firm debt structure. The adjusted R square in model (1) is 0.739,
and in model (2) is 0.811, which indicates that the relationship debt has positive
reaction towards the R&D intensity. In model (2), the coefficient of R&D intensity is
1.263, and significant at 1% level. This also indicates that R&D intensity has positive
correlation to relationship debt. In model (3), we find a significant negative
coefficient of firm size, and a significant positive coefficient of the product of R&D
intensity and scale, which indicates that firm size has significant negative adjusting
effect on the correlation of R&D intensity and relationship debt. In model (4) and (5),
we find negative coefficient of R&D intensity and positive coefficient of the cross
product of R&D intensity and Relationship debt ratio, which shows that relationship
debt could be an efficient corporate governance mechanism, and impact on corporate
performance. The results show that R&D intensity has a positive impact on corporate
performance, and increasing relationship debt ratio can improve the efficiency of
R&D expenditures.
Table 3: Regression on Relationship Debt and R&D Intensity
Variable
Lnsize
ROA
RAA
COV
Model1
Model2
Model3
Model4
Model5
R/B
R/B
R/B
M/B1
M/B2
-0.133*
-0.182*
-0.135*
-0.278**
-0.252**
(0.094)
(0.123)
(0.048)
(0.132)
(0.129)
0.114***
0.098**
0.102***
1.124***
1.076***
(0.013)
(0.041)
(0.019)
(0.432)
(0.45)
-0. 039*
-0.005***
-0.193***
-0.026**
-0.021**
(0.006)
(0.002)
(0.037)
(0.092)
(0.011)
0.000
0.000
0.000
0.000***
0.000***
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
0.038*
0.057
(0.019)
(0.046)
GMP
Leve
In_ROA
In_Leve
In_M/B
In_R/B
0.222
0.201
0.214
0.251
0.169
(0.264)
(0.445)
(0.264)
(0.237)
(0.223)
1.925**
1.681*
1.999**
1.306
1. 032
(0.846)
(0.904)
(0.246)
(1.562)
(1.935)
0.182
-0.034
0.223
-0.267*
0.335*
(0.398)
(0.052)
(0.341)
(0.216)
(0.453)
0.012
0.053
0.022
-0.153
-0.157
(0.075)
(0.094)
(0.068)
(0.156)
(0.195)
0.038*
0.047*
0.043*
12
(0.027)
(0.031)
(0.03)
0.004
-0.004
-0.004
0.006*
0.006*
(0.004)
(0.003)
(0.004)
(0.004)
(0.004)
0.028
0.156
0.156
(0.087)
(0.238)
(0.229)
1.263***
0.874**
-5.431**
-5.202**
(0.487)
(0.523)
(2.278)
(2.127)
9.517***
7.992***
(2.098)
(2.474)
0.218
0.241
(3.134)
(2.005)
SPS
𝑅/𝐵𝑖,𝑡−1
𝑅&𝐷𝐼𝑛𝑡𝑖𝑡
𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 × 𝑆𝑐𝑎𝑙𝑒
4.178***
(1.296)
𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 × 𝑅/𝐵
𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 × 𝑅/𝐵
× 𝑆𝑐𝑎𝑙𝑒
𝑅�2
0.739
0.811
0.826
0.831
0.804
F
5.871***
5.934***
5.527***
7.943***
7.418***
Observations
310
310
310
310
310
Notes: ***significant at 1% level, **significant at 5% level, *significant at 10% level.
(4) Robustness Test
Certain concerns about robustness arise from the exogeneity setting of R&D
intensity. Some researchers (Balakrishnan and Fox,1993; Bhagat and Welch, 1995;
etc.) argue that corporate performance and the lagged value of R&D intensity will
impact on current period R&D intensity. Meanwhile, the TSLS estimation we use for
above models requires no heteroscedasticity, otherwise the estimation would have
bias. In order to solve the endogenous issues, we adopt another model for robustness
test.
𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 = 𝑧1 𝑅/𝐵𝑖,𝑡 + 𝑧2 𝐼𝑛𝑠𝑖𝑧𝑒𝑖𝑡 + 𝑧3 𝑅&𝐷𝐼𝑛𝑡𝑖𝑡−1 + 𝑍 ′ 𝑋 + 𝜇𝑡 + 𝜇𝑖 + 𝜔𝑖𝑡
(e)
Model (c), (d) and (e) can consist an equation system, and we can use 3SLS
method for robustness test. Table 3 shows the robustness test results. The first group
includes model (1),(2), (3) using M/B1 for corporate performance; while the second
group includes model (4),(5), (6) using M/B2 instead. There is little difference
between the two groups.
Table 4: Robustness Test Results
Variable
Lnsize
ROA
RAA
COV
Model1
Model2
Model3
Model4
Model5
Model6
R/B
R&DInt
M/B1
R/B
R&DInt
M/B2
0.009
-0.451
-0.256*
0.009
-0.451
-0.212*
(0.008)
(0.523)
(0.148)
(0.008)
(0.427)
(0.131)
1.158***
0.354**
0.148
1.156***
0.353**
0.098
(0.092)
(0.127)
(0.168)
(0.092)
(0.133)
(0.161)
-0. 021**
-0.715*
0.193***
-0. 021**
-0.704*
0.206***
(0.004)
(0.432)
(0.005)
(0.004)
(0.433)
(0.005)
0.000
0.099
-0.016
0.000***
0.094
-0.019
13
(0.000)
(0.081)
(0.053)
0.038*
0.019*
0.038*
(0.015)
(0.029)
(0.011)
(0.029)
0.387***
0.238*
0.224**
0.391***
0.268*
0.236**
(0.096)
(0.134)
(0.098)
(0.098)
(0.142)
(0.101)
-0.125
-1.131*
-0.359
-0.129
-1.134*
-0.361
(0.538)
(0.064)
(0.246)
(0.539)
(0.064)
(0.331)
-0.708***
-1.434*
-0.447***
-0.721***
-1.434*
-0.462***
(0.216)
(0.204)
(0.063)
(0.231)
(0.204)
(0.064)
0.004
0.154**
-0.102
0.004
0.149**
-0.099
(0.023)
(0.068)
(0.106)
(0.023)
(0.063)
(0.102)
0.003**
-0.014
-0.021**
0.003**
-0.018
-0.014**
(0.000)
(0.052)
(0.006)
(0.000)
(0.054)
(0.006)
1.728**
0.125*
(0.811)
(0.577)
1.731**
(0.806)
GMP
Leve
In_ROA
In_Leve
In_M/B
SPS
𝑅&𝐷𝐼𝑛𝑡𝑖𝑡
𝑅&𝐷𝐼𝑛𝑡𝑖𝑡−1
𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 × 𝑆𝑐𝑎𝑙𝑒
(0.082)
(0.048)
0.021*
(0.000)
0.128*
(0.569)
0.138*
(0.082)
3.214***
3.281***
(1.119)
(1.121)
𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 × 𝑅/𝐵
𝑅&𝐷𝐼𝑛𝑡𝑖𝑡 × 𝑅/𝐵
× 𝑆𝑐𝑎𝑙𝑒
M/B1
2.416**
3.154**
(1.295)
(1.498)
0.159
0.225
(3.231)
(3.001)
0.229
(0.201)
M/B2
0.208
(0.199)
R/B
𝑅/𝐵𝑖,𝑡−1
0.787***
Observations
310
-0.109*
-0.109*
(0.073)
(0.074)
0.779***
(0.049)
(0.051)
310
310
310
310
310
Notes: ***significant at 1% level, **significant at 5% level, *significant at 10% level.
In general, the regression result is robust. We conduct another test by removing
the R&D intensity top 5% and bottom 5% firms, the coefficient significance of R&D
intensity changed in model (3) and model (6), and no significant changes among other
variables.
5 Conclusions
This paper presents an in-depth analysis of the corporate governance role of
inside-debt transactions. Based on debt heterogeneity, we study the relationship
among bank loan, bond financing, firm size and R&D expenditure. Based on 310
14
GEM listed companies, we use the sample of 2007-2011 for empirical analysis. Our
results reveal that relationship financing such as banks loans is an efficient
governance mechanism to corporate R&D intensity. Firms with higher R&D
expenditure tend to have higher relationship debt proportion, and higher relationship
debt proportion will improve the efficiency of firm R&D expenditures. At the same
time, firm size have negative adjustment towards this correlation, which means
relationship financing is more important to small and medium companies than to large
companies.
Our findings may provide special policy suggestions in Chinese context. The
relationship between debt governance and firm innovation pattern in china is different
from U.S. practice, and much similar to the Japanese and Germany pattern. The
relationship financing may have positive governance role in motivate firm innovations.
In order to improve the efficiency of innovation by Chinese companies, bank lending
could be a possible way considering the information asymmetry in small and medium
innovative firms. By the same token, banks should allocate their loans to the more
innovative SMEs. However, we do not take bank size into consideration, and we leave
this question for future research.
15
References
1.
Ahmed, K., & Falk, H. (2006). The value relevance of management’s research and development
reporting choice: Evidence from Australia. Journal of Accounting and Public Policy, 25(3),
231-264.
2.
Aghion, Philippe and Tirole, Jean, (1994), “The Management of Innovations", Quarterly Journal
of Economics, 109, 1185-1209
3.
Baker, Malcolm; Stein, Jeremy and Wurgler, Jeffrey, (2003), “When Does The Market Matter?
Stock Prices And The Investment Of Equity-Dependent Firms", Quarterly Journal of Economics,
118, 969-1005
4.
Berlin, M., & Mester, L. J. 1992. Debt covenants and renegotiation. Journal of Financial
Intermediation,2: 95–133.
5.
Boot, A. W. A. 2000. Relationship banking: What do we know? Journal of Financial
Intermediation, 9:7–25.
6.
Bushee,B.,1998,“Institutional Investors,Long Term Investment and Earnings
management”,Accounting Review,(73):305—333.
7.
Chaganti,R.,and F.Damanpour,1991,“Institutional Investor,Capital Structure and Firm
Performance”,Strategic ManagementJournal,33,725—738.
8.
Choi,Suk.B.,H.L.Soo.,and C.Williams,2011,“Ownership and Firm Innovation In a Transition
Economy:Evidence from China”,Research Policy,40,441—452.
9.
Cremer, Jacques, (1995), “Arm's Length Relationships", Quarterly Journal of Economics, 110,
275-295
10. David, P., Hitt, M. A., & Gimeno, J. 2001. The influence of activism by institutional investors on
R&D. Academy of Management Journal, 44: 144–157.
11. David, P., O'Brien, J. P., & Yoshikawa, T. (2008). The implications of debt heterogeneity for R&D
investment and firm performance. The Academy of Management Journal ARCHIVE, 51(1),
165-181.
12. Feng G.F., and Wen J., 2008, “Empirical Studies on Corporate governance and Firm Innovation in
China”, Chinese Industrial Economy, (5)
13. Hill, C. W. L., & Snell, S. A. 1988. External control, corporate strategy, and firm performance in
research-intensive industries. Strategic Management Journal, 9: 577–590.
14. Huang, Haizhou and Chenggang, Xu, (1999), “Institutions, Innovations, and Growth",
AmericanEconomic Review, Papers and Proceedings, 89, 438-444
15. Kochhar,R.,and P.David,1996,“Institutional Investors and Firm Innovation:A Test of Competing
Hypothesis”,StrategicManagement Journal,17,73—84.
16. Lev, B., Sarath, B., & Sougiannis, T. (2005). R&D Reporting Biases and Their Consequences.
Contemporary Accounting Research, 22(4), 977-1026.
17. O’Brien, J. 2003. The capital structure implication of pursuing a strategy of innovation. Strategic
Management Journal, 24: 415–431.
18. Smith, C. W., & Warner, J. 1979. On financial contracting: An analysis of bond covenants. Journal
of Financial Economics, 7: 117–161.
19. Wen J., Feng G.F. and Liu ZY., 2011, “Debt Heterogeneity, Firm Size and R&D Expenditures”,
Financial Research, (1)
16