You are on page 1of 37

R&D Investments and Dividend Payout: Evidence concerning the Semi-

Mandatory Dividend Policy in China

Bao Yanga, Hsin-I Choub, and Jing Zhaoc

This version: 6 March, 2018

a
School of Accounting, Chongqing University of Technology, Chongqing, China. Email: baoy@cqut.edu.cn.
b
Department of Accountancy and Finance, University of Otago, New Zealand. Email: daisy.chou@otago.ac.nz.
c
Corresponding author, Department of Economics and Finance, La Trobe University, Bundoora, VIC 3086, Australia.
Tel: +61 3 9479 3120, Email: j.zhao@latrobe.edu.au. This project is supported by the National Social Science
Research Foundation of China (Grant No.: 15CJY006).
0

Electronic copy available at: https://ssrn.com/abstract=3108974


R&D Investments and Dividend Payout: Evidence concerning the Semi-

Mandatory Dividend Policy in China

ABSTRACT

Since 2001, the China Securities Regulatory Commission has implemented a series of policies to

ensure that dividend payments constitute a prerequisite of equity financing. This is known as the

semi-mandatory dividend policy. Using a sample of Chinese listed firms from 2007 to 2015, we

document that firms with more R&D investments tend to pay more dividends. While this is not

consistent with the classical dividend theorem, it can be explained by the semi-mandatory dividend

policy and the unique features of R&D investments. R&D firms are more likely to have equity

financing needs and they have strategic incentives to pay dividends in order to access external

capital markets. Such pay-for-financing incentives are stronger for firms with lower cash holdings

and less financial constraints. By examining future seasoned equity offerings, we further

demonstrate equity financing needs as the underlying mechanism for R&D investments to

positively affect dividend payout. In addition, we show that such semi-mandatory dividend

payments adversely affect firm value and sustainable growth and our findings hold after controlling

for endogeneity issues.

JEL Classification: G35, G38

Keywords: R&D investments, Dividend payout, Semi-mandatory dividend policy, China, Equity

financing

Electronic copy available at: https://ssrn.com/abstract=3108974


1. Introduction

Research and development (R&D) investments represent a main driver of long-term

economic growth, and in recent decades this has sharply increased. Firms’ dividend policy has

become an increasingly controversial topic in the finance literature. The essential question to ask

is: do R&D firms pay more or less dividends than non-R&D firms? It has been argued in the

classical dividend theorem that in an imperfect capital market internal funds offer a less costly

source of capital than the external market, and when investments and dividends compete for limited

internal funds, firms may be forced to choose between paying dividends and pursuing valuable

investment projects (Fama and French, 2002; DeAngelo et al., 2006). It ties dividend payments to

investments and predicts a negative effect of R&D investments on dividend payout. In the

literature, empirical studies provide supporting evidence for this prediction but are largely limited

to developed markets. For instance, Fama and French (2001) and Bates et al. (2009) show that

firms with more R&D investments pay less dividends in the U.S., while Gugler (2003) reports a

similar negative effect for a panel of Austrian firms. In this paper, we focus on understanding the

nexus between R&D investments and dividend payments in China which has a unique institutional

setting.

As a developing market, China has long been characterized as having weak investor

protection and corporate governance mechanisms (Allen et al., 2005). Chinese publicly listed firms

are eager to raise external funds yet reluctant to pay dividends. Since 2001, the China Securities

Regulatory Commission (CSRC)1 has implemented a series of policies to protect investors’ rights

by tying securities issuance to dividend payments. On May 6, 2006 and October 9, 2008, the CSRC

released two decrees to define minimum dividend payments as a prerequisite for equity financing.

These regulations cannot force all listed companies to pay dividends, but they impose constraints

1
CSRC is China’s counterpart of the US Securities and Exchange Commission (SEC).
2

Electronic copy available at: https://ssrn.com/abstract=3108974


of dividend payments on those listed firms with equity financing needs, which is called the semi-

mandatory dividend policy (hereafter, SMDP). Many studies have demonstrated the effectiveness

of the SMDP in enhancing dividend payout propensity and ratio of Chinese listed firms (Deng et

al., 2013; Wei et al., 2017). Its semi-mandatory nature, however, may weaken or even reverse the

explanatory power of classical corporate finance theories in the Chinese market.

Unlike capital expenditures, R&D investments are long-term and inflexible projects and

largely financed by external capital markets (Kim and Weisbach, 2008; Brown et al., 2012;

Guarigli and Liu, 2014). Firms with more R&D investments tend to have more equity financing

needs and therefore are more affected by the SMDP. In order to access external capital markets,

there are strategic incentives for R&D firms to pay dividends to meet the prerequisite of equity

financing, which we call pay-for-financing incentives. Using a sample of Chinese listed firms from

2007 to 2015, this paper documents that firms with more R&D investments tend to pay more

dividends, i.e. a positive effect of R&D investments on dividend payout. While this is not

consistent with the classical dividend theorem, it can still be explained by the pay-for-financing

incentives under the SMDP.

Firm with higher R&D needs are expected to face internal cash shortages and consequently

have stronger incentives for future refinancing. McLean (2011) shows that firms increasingly issue

shares for the purpose of cash savings. It suggests firms with less cash holdings having more equity

financing needs. Moreover, firms rely extensively on cash holdings to smooth their R&D

expenditures in response to transitory finance shocks (Brown and Peterson, 2011). For this reason,

we predict that firms with less cash holdings are more influenced by the SMDP due to their stronger

pay-for-financing incentives. Consistent with this prediction, this paper shows that the positive

effect of R&D investments on dividend payout is stronger for firms with less cash holdings.

3
We also examine whether firms’ financial constraints affect this positive relationship

between R&D investments and dividend payout. Prior research defines financially constrained

firms as those encountering severe agency and transaction costs when accessing external capital

markets (Korajczyk and Levy, 2003). The CSRC uses a merit-based system to regulate share

issuance of listed firms, which requires them to meet certain accounting performance measures.

Even if financially constrained firms have paid cash dividends and met the requirements of the

SMDP, it is less likely that they will qualify for equity financing due to their poorer accounting

performance, and therefore they have less pay-for-financing incentives. Consistent with this

argument, we demonstrate that the positive effect of R&D investments on dividend payout is less

pronounced for firms encumbered with more financial constraints.

In China, securities issuance mainly refers to seasonal equity offerings (SEOs). We show

that the propensity and intensity of future SEOs are positively associated with R&D investments.

It confirms equity financing needs as the underlying mechanism for R&D investments to positively

affect dividend payout. For firms with R&D investments, internal funds provide a less costly

source of capital compared to external equity markets, but under the SMDP they are forced to pay

dividends in order to meet the prerequisite of equity financing. We show that such semi-mandatory

dividend payments have adverse impacts on firm value and sustainable growth, and diminish the

value-creating capability of R&D investments. To address endogeneity concerns, we adopt two-

stage least squares and propensity score matching (PSM) approaches and our findings remain

robust.

This paper contributes to the broad literature on the investment-dividend relationship. To

the best of our knowledge, it is the first study documenting R&D investments as having a positive

effect on dividend payout. Existing studies are largely limited to developed markets, and they find

that firms with more R&D investments pay less dividends (Fama and French, 2001; Bah and

4
Dumontier, 2001; Bates, 2009), which is consistent with the classical dividend theorem. Our

findings can be explained by the unique institutional setting of China, in particular the SMDP, and

the unique features of R&D investments. It points out the importance of revisiting classical

corporate finance theories to establish their relevance and explanatory power in the context of

certain emerging markets.

Our findings also contribute to the literature analyzing corporate financial policies under

the SMDP. Many studies have demonstrated the effectiveness of the SMDP in enhancing dividend

payout propensity and ratio of Chinese publicly listed firms (Deng et al., 2013; Wei et al., 2017).

To promote dividend payout, the SMDP sets minimum dividend payments as a prerequisite of

equity financing, which generates extra costs for firms to access external capital markets. Although

it aims to protect investors’ rights, we demonstrate that such semi-mandatory dividend payments

adversely impact on firm value and sustainable growth, and such an impact is stronger for firms

with more R&D investments. Given that investments in R&D represent a key driver of long-term

economic growth, Chinese policy-makers may consider relaxing the requirements of SMDP on

R&D firms in order to stimulate innovation activities.

The remainder of this paper is organized as follows. Section 2 discusses the research

background and develops our hypotheses. Section 3 describes our research methodology and

sample selection criteria. Section 4 reports our empirical results and findings and finally Section

5 concludes the paper with a summary of the main themes covered here.

2. Background and Hypothesis Development

Dividend policy has long been a controversial topic in the finance literature. Miller and

Modigliani (1961) demonstrate that in a perfect capital market only investments affect firm value

but they may or may not affect dividend decisions. If the market is less than perfect, for instance,

5
in the presence of financing frictions, internal funds present a less costly source of capital than the

external market. When investments and dividends compete for limited internal funds, firms may

be forced to choose between paying dividends and pursuing valuable investment projects. It ties

dividend payments to investments and predicts a negative relationship between them (Fama and

French, 2002; DeAngelo et al., 2006). Although early empirical studies examining the effect of

investments on dividend payments provide mixed evidence (Fama, 1974; Partington, 1985), more

recent empirical results are largely consistent with investments having a negative effect. 2

Compared to capital expenditures, R&D investments are more relevant to long-term economic

growth. The sharp increase in R&D expenditures in recent decades may affect firms’ dividend

policies, but the empirical examinations are largely limited to developed markets. For instance,

Fama and French (2001) and Bates et al. (2009) show that firms with more R&D investments pay

less dividends in the U.S., and Gugler (2003) reports a similar negative effect for a panel of

Austrian firms. In this paper, we focus on understanding the nexus between R&D investments and

dividend payments in the unique institutional setting of China.3

2.1 Institutional Background in China

As an emerging market, China has long been plagued by weak investor protection

mechanisms (Allen et al., 2005). Chinese publicly listed firms are eager to raise external funds but

they are reluctant to pay dividends.4 Prior research argues that dividends can be used in an agency

2
For evidence concerning investments’ negative effect on dividend payment, see the following analyses: Adedeji
(1998), Fama and French (2001, 2002), Sarig (2004), DeAngelo et al. (2006), and Soo, Ramalingegowda and Yu
(2017).
3
China’s government announced its “Medium to Long Term Science and Technology Development Plan (2006-2020)”
in 2006, the aim being to: firstly, raise R&D intensity from 1.3% to 2.5% of GDP; secondly, sharply curtail the
country’s reliance on imported technology; and thirdly, obtain advanced core technologies in the manufacturing and
information industries by 2020. It highlights the importance of R&D investments in China. Detailed information
about the development plan can be found at: http://www.gov.cn/english/2006-02/09/content_183426.htm
4
See, for instance, Allen et al. (2005), Chen et al. (2009), and Huang et al. (2011) for evidence of Chinese firms paying
low dividends compared to similar firms operating in countries with stronger investor protection systems.
6
context to prevent insiders from diverting retained earnings for their own benefit (La Porta et al.,

2006). Since 2001, the China Securities Regulatory Commission (CSRC) has implemented a series

of policies to protect investors’ rights by tying equity financing to dividend payments. On March

28, 2001, the CSRC stipulated that listed firms that do not pay continuous cash dividends in the

previous three years do not qualify for securities issuance, which mainly refers to seasoned equity

offerings (SEOs). On May 6, 2006 and October 9, 2008, the CSRC further released two decrees

to define minimum dividend payments as a prerequisite for equity financing. The 2006 (2008)

regulation stipulated that “the cumulatively distributed profits in cash or stocks (cash) in the recent

3 years shall not be less than 20% (30%) of the average annual distributable profits realized in the

recent 3 years”.5 Although these regulations cannot force all listed companies to pay dividends,

they do impose constraints of dividend payments on those listed firms with equity financing needs,

which is referred to as the semi-mandatory dividend policy (hereafter, SMDP).

Many studies have demonstrated the effectiveness of the SMDP in enhancing dividend

payout propensity and ratio of Chinese listed firms, in particular those with financing needs (Deng

et al., 2013; Wei et al., 2017). Its semi-mandatory nature, however, may weaken or even reverse

the explanatory power of classical dividend theories. For instance, in contrast to the conventional

negative investment-dividend relationship evidenced in the literature, Deng et al. (2013) document

an “N-shaped” nonlinear relationship between dividends and investments in Chinese firms from

2000 to 2010. Furthermore the average coefficient from regressing investments on dividends is

insignificant.

R&D-driven innovation has long been regarded as a crucial generator of productivity and

economic growth. The Chinese government is committed to promoting the development of

innovation and urging firms to invest more in R&D activities (Guariglia and Liu, 2014). The

5
Detailed information about the regulations can be found at:
http://www.csrc.gov.cn/pub/csrc_en/laws/overRule/Decrees/200910/t20091028_166901.html
7
SMDP is especially pertinent to firms with R&D investments as an integral part of their business

strategies. On the one hand, equity financing is the predominant source of finance for R&D

investments, as widely documented in the literature. 6 Kim and Weisbach (2008) explore the

motivations for public equity offerings across 38 countries and find that investments (R&D in

particular) increase following equity offers. On the other hand, innovation activities are generally

considered as long-term projects (Fama and French, 2002). Unlike capital expenditures, R&D

investments are more inflexible and commonly used as a proxy for expected investment or

precautionary motives (McLean, 2011). Therefore, compared to firms without R&D investments,

R&D-intensive firms are more likely to have existing and future equity financing needs, and there

are strategic incentives for them to pay out dividends to meet the prerequisite of equity financing

under the SMDP, which we call pay-for-financing incentives. Although empirical literature

documents R&D firms paying less dividends in some developed markets (Fama and French, 2001;

Gugler, 2003), the effect of R&D investments on Chinese firms’ dividend policies is unclear given

the SMDP.

2.2 Hypothesis Development

As discussed above, firms with more R&D investments tend to have more existing and

future equity financing needs due to the nature of R&D investments. Under the SMDP, those firms

are more likely to pay dividends in order to be qualified for securities issuance. In China, a publicly

listed firm’s ability to make an SEO is highly restricted and must be approved by the CSRC. In

2013 the CEO of Shenzhen Stock Exchange pointed out that firms with a high payout ratio, say,

exceeding 50%, should be given priority for SEOs. Therefore, firms with more R&D investments

6
For international evidence regarding the connection between equity financing and R&D investments, see Brown et
al. (2009), Gatchev et al. (2009), Brown et al. (2012), and Guarigli and Liu (2014).
8
tend to pay more dividends in order to obtain access to external financing sources. Our first

hypothesis is as follows:

H1: The effect of R&D investments on dividend payout is positive.

As noted by Hall (2002), Brown et al. (2009) and many others, firms with a high R&D

emphasis use little debt, and R&D investments are financed mostly with external equity issuance

and internal funds. McLean (2011) contends that firms increasingly issue shares for the purpose

of cash savings, suggesting firms with less cash holdings have more equity financing needs. In

addition, Brown and Peterson (2011) demonstrate that firms rely extensively on cash holdings to

smooth their R&D expenditures in response to transitory finance shocks. Therefore, firms with

less cash holdings are more affected by the SMDP and their pay-for-financing incentives are

stronger. Our second hypothesis is proposed as follows:

H2: The positive effect of R&D investments on dividend payout is stronger for firms with

less cash holdings.

Prior research defines financially constrained firms as those that face severe agency and

transaction costs when accessing external capital markets (Korajczyk and Levy, 2003). Compared

to unconstrained firms, financially constrained firms experience higher costs of equity financing.

On the other hand, the CSRC uses a merit-based system to regulate share issuance of listed firms,

which requires them to meet certain accounting performance measures. Even if financially

constrained firms have paid cash dividends and met the requirements of the SMDP, it is less likely

for them to qualify for equity financing due to their poorer accounting performance when compared

to unconstrained firms. Therefore, firms with more (less) financial constraints have less (more)

pay-for-financing incentives due to their lower (higher) possibility of receiving external equity

financing. The above discussion leads to our third hypothesis:

9
H3: The positive effect of R&D investments on dividend payout is stronger for firms with

less financial constraints.

3. Research Methodology and Sample Selection

3.1 Regression Specification and Variable Measurement

To examine how R&D investments affect dividends, we consider the following regression

specification:
𝑁

𝑃𝑎𝑦𝑜𝑢𝑡𝑖,𝑡 = 𝛼1 + 𝛽1 𝑅&𝐷𝑖,𝑡 + ∑ 𝛾𝑛 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑛,𝑖,𝑡 + 𝜀𝑖,𝑡 , (1)


𝑛=1

where the dependent variable 𝑃𝑎𝑦𝑜𝑢𝑡𝑖,𝑡 is the dividend payout ratio of firm 𝑖 in year 𝑡, measured

by dividend per share divided by earnings per share,7 our main explanatory variable 𝑅&𝐷𝑖,𝑡 denotes

R&D investments of firm 𝑖 in year 𝑡, calculated as R&D expenditures scaled by total assets,8 and

𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑛,𝑖,𝑡 , 𝑛 = 1, … , 𝑁, represent a set of control varibales. Many publicly listed firms do not

pay dividends, which makes the dependent variable 𝑃𝑎𝑦𝑜𝑢𝑡𝑖,𝑡 censored at zero. Therefore we use

the Tobit model in our regressions.9 Our first hypothesis predicts R&D investments as having a

positive effect on dividend payout, i.e. 𝛽1 > 0 in (1).

To test the second hypothesis, we further allow the effect of 𝑅&𝐷 to change with firms’

cash holdings and adopt the following specification:

𝑃𝑎𝑦𝑜𝑢𝑡𝑖,𝑡 = 𝛼1 + 𝛽1 𝑅&𝐷𝑖,𝑡 + 𝛽2 𝑅&𝐷𝑖,𝑡 × 𝐶𝑎𝑠ℎ𝑖,𝑡 + 𝛽3 𝐶𝑎𝑠ℎ𝑖,𝑡 (2)

7
In unreported robustness tests, we use dividend payments to sales and dividend payments to net assets as alternative
measures of dividend payout ratio and obtain qualitatively similar results.
8
In unreported robustness tests, we use R&D expenditures scaled by total sales as an alternative measure of R&D
investments and obtain qualitatively similar results. When R&D investments are measured in the lagged year instead,
the results remain similar.
9
Tobit regression models are widely implemented in studies on payout policy (see Chay and Suh, 2009; Huang et al.,
2011).
10
𝑁

+ ∑ 𝛾𝑛 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑛,𝑖,𝑡 + 𝜀𝑖,𝑡 ,
𝑛=1

where 𝐶𝑎𝑠ℎ𝑖,𝑡 denotes the cash holdings of firm 𝑖 in year 𝑡, measured as cash and cash equivalent

divided by net assets measured at the beginning of year 𝑡.10 Our second hypothesis predicts 𝛽2 to

be negative in (2), indicating that the positive effect of R&D investments is stronger for firms with

smaller cash holdings due to their higher pay-for-financing incentives.

We further examine the role of financial constraints in affecting the relationship of R&D

investments and dividend payout. Hadlock and Pierce (2010) find that firm size and age are

particularly useful predictors of financial constraints levels, and propose a measure of financial

constraints based on these two firm characteristics called the SA index:11

𝑆𝐴 = 0.043 × 𝑆𝑖𝑧𝑒 2 − 0.737 × 𝑆𝑖𝑧𝑒 − 0.04 × 𝑎𝑔𝑒, (3)

where variable 𝑆𝑖𝑧𝑒 denotes firm size, calculated as the natural logarithm of book value of total

assets, and 𝑎𝑔𝑒 is firm age, i.e. the number of years since IPO. To ensure our variable of financial

constraints ranges between 0 and 1, we consider a transformed SA index defined as:

𝑆𝐴 − 𝑚𝑖𝑛(𝑆𝐴)
𝑇𝑆𝐴 = , (4)
max(𝑆𝐴) − min(𝑆𝐴)

where 𝑆𝐴 is the SA index, and 𝑚𝑎𝑥(𝑆𝐴) and 𝑚𝑖𝑛(𝑆𝐴) denote the maximum and minimum 𝑆𝐴

value in the whole sample, respectively.12 In particular, we have 𝑇𝑆𝐴 = 1 for 𝑆𝐴 = 𝑚𝑎𝑥(𝑆𝐴),

and 𝑇𝑆𝐴 = 0 for 𝑆𝐴 = 𝑚𝑖𝑛(𝑆𝐴), and the 𝑇𝑆𝐴 index represents a measure of financial constraints

ranging from 0 to 1. To test the third hypothesis, we allow the effect of 𝑅&𝐷 to vary with a firm’s

financial constraints through the following specification:

10
Cash holdings are measured at the beginning of the fiscal year to reflect the fact that investment and dividend
decisions are influenced by the cash reserve of the previous year.
11
The SA index has been adopted by a number of studies as a measure of financial constraints. Examples of recent
studies on this are He and Wintoki (2016) and Koo et al. (2017).
12
In unreported robustness tests, we use 𝑆𝐴 as a measure of financial constraints instead of 𝑇𝑆𝐴 and obtain
qualitatively similar results.
11
𝑃𝑎𝑦𝑜𝑢𝑡𝑖,𝑡 = 𝛼1 + 𝛽1 𝑅&𝐷𝑖,𝑡 + 𝛽4 𝑅&𝐷𝑖,𝑡 × 𝑇𝑆𝐴𝑖,𝑡 + 𝛽5 𝑇𝑆𝐴𝑖,𝑡

𝑁 (5)
+ ∑ 𝛾𝑛 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑛,𝑖,𝑡 + 𝜀𝑖,𝑡 ,
𝑛=1

where 𝑇𝑆𝐴𝑖,𝑡 denotes the 𝑇𝑆𝐴 index of firm 𝑖 in year 𝑡. Our third hypothesis predicts 𝛽4 to be

negative in (5), indicating that the positive effect of R&D investments is stronger for firms with

less financial constraints, since their pay-for-financing incentives are higher compared to firms

saddled by more financial constraints.

In our regression specifications, 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑛,𝑖,𝑡 , 𝑛 = 1, … , 𝑁, denote a set of control variables

of firm 𝑖 in year 𝑡 , representing other possible determinants of dividend payout ratio. Their

definitions are provided in Table I. The first control variable is firm size (Size). We expect its

coefficient to be positive, as empirical evidence shows that larger firms are more likely to pay

higher dividends (Denis and Osovov, 2008). Debt can be a substitute for dividends in mitigating

agency problems, as both debt and dividends reduce free cash flows (Brockman and Unlu, 2009;

Huang et al., 2011). We include leverage (𝐿𝑒𝑣) as a control variable and expect its coefficient to

be negative. Firms with higher profitability tend to pay more dividends (Brockman and Unlu,

2009). We use 𝑅𝑂𝐴 as a proxy for firms’ profitability, and include it as a control variable, of which

the coefficient is expected to be positive. Current literature documents an inverse relationship

between growth prospects and dividend payments (Cheng et al., 2009; Bradford et al., 2013). To

capture this effect, we consider Tobin’s Q ( 𝑇𝑜𝑏𝑖𝑛’𝑠 𝑄 ) as a control variable and expect its

coefficient to be negative. Firms with more free cash flows are expected to pay more dividends

(Gul and Tsui, 1998). We include free cash flow (𝐹𝐶𝐹) as a control variable and expect its

coefficient to be positive. Following Gugler (2003), Ramalingeowda et al. (2013) and many other

studies in dividend policy, we further include some other firm-specific financial variables as

controls, including firm age (𝐴𝑔𝑒), capital expenditures (𝐶𝑎𝑝𝑒𝑥), working capital (𝑊𝑜𝑟𝑘𝑐𝑎𝑝) and

12
firm risk (𝐵𝑒𝑡𝑎).13 As well, prior studies (Sharma, 2011; Chen et al., 2017) show that a firm’s

dividend policy is affected by its governance structure. To capture this effect, we further adopt

four firm-specific governance variables as controls, namely, controlling shareholder ownership

(Sh1), board size (Bsize), board independence (Bindpend), and CEO duality (Duality). All our

regression specifications include industry and year dummies.

INSERT TABLE I HERE

3.2 Sample Selection Criteria

In January 2007, all publicly listed Chinese firms had to adopt the New Chinese Accounting

Principle 2006, which requires them to disclose R&D expenditures. To ensure consistent

measurements of R&D expenditures and financial data, our sample stocks include all A-share firms

listed on the Shanghai and Shenzhen Stock Exchanges between 2007 and 2015. We exclude

financial and insurance firms as those firms have different accounting criteria. We further drop ST,

*ST and PT firms because those firms underperformed and are highly regulated by the

government.14 We also drop firms that have shareholders’ equity loss in certain years or missing

values in one or more explanatory variables. Our final sample consists of 11,840 firm-year

observations. We collect financial and corporate governance data from the China Stock Market

and Accounting Research Database (CSMAR). The information on R&D expenditures (𝑅&𝐷) is

sourced from the CSMAR and the Center of Research in Security Prices (CRSP). Given there are

many missing values in R&D expenditures in 2007 and 2008, we manually collect those data from

13
In unreported robustness tests, we adopt retained earnings ratio, return on equity, and sales growth rate as alternative
proxies for a firm’s lifecycle, profitability and growth, respectively, and obtain qualitatively similar results.
14
If a firm experiences abnormal financial or operating situations, for example losses in 2 consecutive years, it is
classified as a ST (Special Treatment) firm by the CSRC. *ST firms denote those firms warned about the risk of being
delisted due to their worsening situations, say experiencing losses over 3 consecutive years. Firms undergoing even
worse situations are classified as PT (Particular Transfer) firms by the CSRC, and they have been delisted and only
particular transfers of their stocks are permitted.
13
the information disclosure website designated by the CSRC (http://www.cninfo.com.cn/cninfo-

new/index). All the continuous variables are winsorized at 1% and 99% to reduce the impact of

extreme values.

4. Empirical Results

4.1 Descriptive Statistics

Panel A of Table II reports summary statistics of the main variables in the whole sample.

The average dividend payout ratio is 23.8%, which is higher than the median payout ratio. It

indicates that some firms pay very high dividends in certain periods, pushing up the average payout

ratio. A similar pattern can be found in dividend per share (DPS). The mean and median DPS is

¥0.103 and ¥0.05, respectively. The average R&D investments in our sample constitute 0.739%

of total assets and the median value is zero, suggesting that less than 50% of the firms have R&D

investments. The mean and median of cash holdings (𝐶𝑎𝑠ℎ) is 0.199 and 0.151, respectively. 𝑇𝑆𝐴

proxies for financial constraints and its mean and median values are 0.5595 and 0.5724,

respectively. Summary statistics of other firm characteristics are similar to those documented in

prior studies (Huang et al., 2011; Bradford et al., 2013). The average firm size (𝑆𝑖𝑧𝑒) is 21.84,

equivalent to book value of total assets of ¥3.055billions. The mean value of leverage ratio (𝐿𝑒𝑣),

return on assets (𝑅𝑂𝐴), Tobin’s Q (𝑇𝑜𝑏𝑖𝑛’𝑠 𝑄), free cash flow (𝐹𝐶𝐹), firm age (𝑎𝑔𝑒), capital

expenditures (𝐶𝑎𝑝𝑒𝑥), working capital (𝑊𝑜𝑟𝑘𝑐𝑎𝑝) and firm risk (𝐵𝑒𝑡𝑎) is 45.44%, 6.44%, 2.345,

¥0.347 per share, 13.952 years, 5.83%, 32.8%, and 1.009, respectively. The average proportion of

number of shares held by the controlling shareholders is 36.1%. On average, Chinese publicly

listed firms’ board size is 10.475 while the percentage of independent directors on the board is

33.7%. For around 21.5% of firm-year observations, CEO and chairman are in fact the same person.

INSERT TABLE II HERE

14
To examine the effect of R&D investments on dividend payout, we first compare the

characteristics of subsamples with and without R&D expenditures. Panel B of Table II reports:

firstly, the average statistics of the main variables for the two subsamples; and secondly, the

difference between them. The average payout ratio and dividend per share of R&D firms are higher

than those of non-R&D firms, and the differences between the two subsamples are statistically

significant at the 1% level (t-values are 11.435 and 8.386, respectively). This simple test indicates

that R&D firms have higher dividend payout ratio compared to non-R&D firms, which is consistent

with our first hypothesis. The average cash holdings (𝑇𝑆𝐴) for R&D firms is 0.228 (0.5534) while

that for non-R&D firms is 0.173 (0.5635); the difference is statistically significant at the 1% level.

It shows that firms with R&D investments tend to hold more cash and are less financially

constrained compared to non-R&D firms. This evidence aligns with prior studies arguing that cash

holdings and financial constraints play important roles in influencing firms’ investment and

dividend decisions (Bates et al., 2009; Chay and Suh, 2009; Brown et al., 2012; Hoberg et al.,

2014). Furthermore this panel shows that on average, R&D firms have a lower leverage ratio, more

growth opportunities, less free cash flow, more capital expenditures, lower controlling shareholder

ownership, larger board size, and less board independence. Added to these features, such firms are

more profitable, younger, and more likely to have dual CEOs.

Panel C of Table II illustrates the evolution of dividend policy in the subsamples with

varying 𝑅&𝐷. In particular, we sort observations into subsamples based on R&D investments and

report their average characteristics. As firms’ R&D investments increase, it is more likely for them

to pay dividends, and their average dividend per share and payout ratio also increase. The only

exception is the subsample with 𝑅&𝐷 ranging between 90% and 100% percentiles, i.e. the P100

subsample. Although their propensity to pay dividends is the highest, their average 𝐷𝑃𝑆 and

𝑃𝑎𝑦𝑜𝑢𝑡 is slightly lower than those of the P90 subsample. Overall, firms with more R&D

15
investments have higher propensity of dividends payments and higher payout ratio, consistent with

our first hypothesis.

4.2 Effect of R&D Investments on Dividend Payout and the Roles of Cash Holdings and Financial

Constraints

To rigorously test our first hypothesis and control for the effects of other firm characteristics,

we adopt regression specification (1) and report the results of estimating the Tobit model based on

the whole sample in column (1) of Table III. The regression coefficient of R&D is 0.973,

significantly positive at the 1% level. In contrast to the negative investment-dividend relationship

documented in the literature, we document a positive effect of R&D investments on dividend

payments in Chinese publicly listed firms, i.e. firms with more R&D investments pay more

dividends. It is consistent with our first hypothesis that under the SMDP there are strategic

incentives for R&D firms to pay dividends in order to meet the prerequisite of external equity

financing. The impact of R&D investments is also economically significant. For instance, a one-

standard-deviation rise in 𝑅&𝐷 increases the dividend payout ratio by 130.38 basis points.15 The

effects of control variables are largely consistent with prior studies (Brockman and Unlu, 2009;

Huang et al., 2011). In particular, firms tend to pay more dividends if they have the following

characteristics: larger size, lower leverage, higher profitability, less growth opportunities, more

free cash flows, more controlling shareholder ownership, less board independence, less CEO

duality, are younger and less prone to risk.

INSERT TABLE III HERE

To test our second hypothesis, we allow the effect of R&D investments to vary with firms’

cash holdings as specified in (2) and report the regression results of the whole sample in column

15
Because the sample standard deviation of 𝑅&𝐷 is 0.0134 and its regression coefficient is 0.973, we have
0.973 × 0.0134 = 1.3038%.
16
(2) of Table III. The regression coefficient of 𝑅&𝐷 × 𝐶𝑎𝑠ℎ is significantly negative at the 1%

level. It indicates that the positive effect of R&D investments on dividend payments is more (less)

pronounced for firms with lower (higher) cash holdings, and this is consistent with our second

hypothesis. Cash reserves provide a less costly source of capital and can be used to smooth R&D

expenditures in response to finance shocks, and therefore firms with higher cash holdings have less

pay-for-financing incentives and consequently R&D investments exert a weaker effect on dividend

payments. In other words, firms with higher (lower) cash holdings are more (less) flexible in their

dividend decisions and less (more) eager to meet the requirements of the SMDP to access the

external capital markets. The regression coefficient of 𝐶𝑎𝑠ℎ is significantly positive at the 1%

level, confirming that firms with more cash holdings tend to pay more dividends. To further

examine the role of cash holdings, we sort firm-year observations into two subsamples based on

cash holdings and report their regression results in column (3-4) of Table III, respectively. In the

subsample with low 𝐶𝑎𝑠ℎ, the regression coefficient of R&D is 1.833, significantly positive at the

1% level, indicating a strong positive effect of R&D investments on dividend payments. In the

subsample with high 𝐶𝑎𝑠ℎ, the regression coefficient of R&D is insignificant, which suggests that

dividend decisions are not affected by R&D investments for firms with high cash holdings.

To test our third hypothesis, we allow the effect of 𝑅&𝐷 to change with firms’ financial

constraints. Column (1) of Table IV reports the results of estimating the Tobit regression model

as specified in (5), where 𝑇𝑆𝐴 proxies for financial constraints. The regression coefficient of

𝑅&𝐷 × 𝑇𝑆𝐴 is significantly negative at the 1% level, suggesting that the positive effect of R&D

investments on dividend payments is more (less) pronounced for firms with less (more) financial

constraints. It is consistent with our third hypothesis. In addition, financially constrained firms

tend to pay less dividends, as evidenced by the regression coefficient of 𝑇𝑆𝐴 being significantly

negative. We further sort observations into two subsamples based on 𝑇𝑆𝐴 and report their

17
regression results in column (2-3), respectively. The positive effect of 𝑅&𝐷 on 𝑃𝑎𝑦𝑜𝑢𝑡 is

significantly positive in the subsample with low 𝑇𝑆𝐴 only. Financially constrained firms are less

likely to win approval from the CSRC to access external capital markets due to their poorer

accounting performance, even if they pay dividends and meet the requirements of the SMDP.

Therefore, their pay-for-financing incentives are lower compared to unconstrained firms. Our

results are consistent with those arguments that dividend decisions are less affected by R&D

investments for firms experiencing more financial constraints.

INSERT TABLE IV HERE

4.3 R&D Investments and Future SEOs

In our developed hypotheses, we argue there are strategic incentives for R&D firms to fulfill

the requirements of the SMDP in order to access external capital markets. Compared to capital

expenditures, R&D investments are more pertinent to future financing needs. Seasoned equity

offerings (SEOs) represent the source of external equity financing for Chinese publicly listed firms.

Our hypotheses conjecture equity financing needs as the underlying mechanism for R&D

investments to affect dividend payments. Therefore, firms with more R&D investments are

expected to conduct more SEOs in the future. Here we explore this underlying mechanism by

examining the effect of R&D investments on future SEO activities. Column (1) and (3) of Table

V report the results of logistic regression model with 𝑆𝐸𝑂𝑑𝑢𝑚𝑚𝑦 as the dependent variable, which

takes the value of 1 if the firm conducts SEO within the next three years and 0 otherwise. Column

(1) adopts R&D investments ( 𝑅&𝐷 ) as an explanatory variable, while column (3) uses

𝑅&𝐷𝑑𝑢𝑚𝑚𝑦 instead, which equals 1 if 𝑅&𝐷 > 0 and 0 otherwise. The regression coefficient of

𝑅&𝐷 and 𝑅&𝐷𝑑𝑢𝑚𝑚𝑦 is significantly positive at the 5% and 1% level, respectively. It indicates

that firms with more R&D investments are more likely to conduct SEOs in the future. Column (2)

18
and (4) of Table V report the results of ordinary least squares (OLS) model with 𝑆𝐸𝑂𝑠𝑖𝑧𝑒 as the

dependent variable, measured as the natural logarithm of total SEO amount within the next three

years. The regression coefficients of 𝑅&𝐷 and 𝑅&𝐷𝑑𝑢𝑚𝑚𝑦 are significantly positive, indicating

that firms with more R&D investments engage in more SEOs in the future. Overall, both

propensity and intensity of future SEOs are positively related to R&D investments. It confirms our

conjecture that equity financing needs serve as the underlying mechanism for R&D investments

having an effect on dividend payments.

INSERT TABLE V HERE

4.4 Financial Consequences of the Semi-Mandatory Dividend Policy

For firms with R&D investments, internal funds provide a less expensive source of capital

compared to external equity markets. Under the SMDP, they are forced to pay dividends in order

to meet the prerequisite of equity financing. Such semi-mandatory dividend payments represent

extra costs of equity financing and may have adverse impacts on firm value and sustainable growth.

We adopt market-to-book ratio, calculated as market value of total assets scaled by book value of

total assets, as a proxy for firm value. To understand the financial consequences of the SMDP, we

examine the effects of dividend payout and R&D investments on market-to-book ratio and report

the regression results in column (1-3) of Table VI. Prior studies document a positive association

between dividend payments and firm value (Fama and French, 1998). Dividends contribute more

to firm value in countries with weaker investor protection mechanisms (Pinkowitz et al., 2006).

However, the regression coefficient of 𝑃𝑎𝑦𝑜𝑢𝑡 is significantly negative at the 1% level in column

(1-3).

Firms making more dividend payments tend to have lower firm value, indicating a value-

diminishing effect of dividend payments under the SMDP. In addition, the regression coefficient

19
of 𝑃𝑎𝑦𝑜𝑢𝑡 × 𝑅&𝐷 is significiantly negative at the 1% level, meaning that the negative effect of

dividend payments on firm value is stronger for firms with more R&D investments. Under the

SMDP, firms with external financing needs are forced to pay dividends, and such payments

represent additional costs of accessing external capital markets and adversely impact on firm value.

Firms with greater levels of R&D investments tend to have more external financing needs, and

therefore their dividend policies are more constrained by the SMDP. In column (1-3), the

regression coefficient of 𝑅&𝐷 is significantly positive at the 1% level, consistent with the reported

evidence in literature that R&D expenditures have a positive impact on firm value (Fama and

French, 1998; Booth et al., 2006). However, the negative coefficient of 𝑃𝑎𝑦𝑜𝑢𝑡 × 𝑅&𝐷 indicates

a less positive impact of 𝑅&𝐷 for firms with more dividend payments. It implies that the value-

creating capability of R&D is diminished under the SMDP.

INSERT TABLE VI HERE

To proxy for firm sustainable growth (Klein and Belt, 1993; Chen et al., 2013), we use

sustainable growth ratio from the CSMAR database, which is calculated as (net profit margin on

sales × total asset turnover × retained earnings) divided by (1− (net profit margin on sales × total

asset turnover × retained earnings)). Column (4-6) of Table VI report the results of the regressions

with sustainable growth ratio as the dependent variable. The regression coefficient of 𝑃𝑎𝑦𝑜𝑢𝑡

( 𝑅&𝐷 ) is significantly negative (positive) at the 1% level, while that of 𝑃𝑎𝑦𝑜𝑢𝑡 × 𝑅&𝐷 is

significantly negative. Dividend payments are indicated as adversely affecting firms’ sustainable

growth under the SMDP; furthermore such an effect is stronger for firms with more R&D

investments. Although R&D continues to act as a driving force of sustainable growth, its impact

becomes weaker for firms with more dividend payments. Under the SMDP, equity financing is

tied to dividend payments. Firms with more R&D investments tend to have more equity financing

requirements and therefore are more likely to pay dividends in order to access external capital
20
markets. We demonstrate that such semi-mandatory dividend payments have adverse impacts on

firm value and sustainable growth.

4.5 Endogeneity Tests

Endogeneity is a possible concern in our analysis. On the one hand, firms with certain

patterns of R&D investments may have other unobserved firm-specific characteristics that affect

both R&D investments and dividend payout, which potentially bias our results. On the other hand,

dividend payout could potentially impact on R&D investments, indicating the reverse causality

issue. We use a two-stage least squares (2SLS) approach and a propensity score matching (PSM)

approach to mitigate the potential endogeneity of R&D investments. Following Wahal and

McConnell (2000), we adopt industry-adjusted 𝑅&𝐷 as an instrumental variable in our 2SLS

regressions, which is calculated by deducting industry average 𝑅&𝐷 of each year from firm-level

𝑅&𝐷. Column (1-2) of Panels A and B of Table VII report the results of estimating the first and

second stage models. In the first stage model, industry-adjusted 𝑅&𝐷 is shown to be positively

related to 𝑅&𝐷 at the 1% significance level. The second stage model reveals that the effect of 𝑅&𝐷

on 𝑃𝑎𝑦𝑜𝑢𝑡 is significantly positive, while the regression coefficients of 𝑅&𝐷 × 𝐶𝑎𝑠ℎ and

𝑅&𝐷 × 𝑇𝑆𝐴 are both significantly negative. After controlling for endogeneity issues, the results

continue to support our three hypotheses.

INSERT TABLE VII HERE

The PSM approach is widely used in the literature to address endogeneity issues (Attig et

al., 2015; He et al., 2017). To estimate the propensity score, we adopt a logistic model with

𝑅&𝐷𝑑𝑢𝑚𝑚𝑦 as the dependent variable, which is equal to 1 if the firm has R&D investments and

0 otherwise. To identify factors that significantly influence firms’ R&D investment decisions, we

perform a stepwise regression strategy and find several significant factors, including firm age, cash
21
holdings, industry dummy, 𝑅𝑂𝐴 and controlling shareholder ownership. The propensity score is

the probability that a firm has R&D investments predicted by a logistic model with those significant

factors as determinants. Before matching, the treatment (control) group includes observations of

firms with (without) R&D investment. The kernel density functions of the propensity score of the

two groups are plotted in the upper panel of Figure 1, which are quite different from each other.

INSERT FIGURE 1 HERE

For each observation with positive 𝑅&𝐷 (in the treatment group), we match it with an

observation with zero 𝑅&𝐷 to make their propensity scores as close as possible to form the control

group. We obtain 9,178 matching observations, including 4,589 observations with positive 𝑅&𝐷

(the treatment group) and 4,589 observations with zero 𝑅&𝐷 (the control group). As shown in the

lower panel of Figure 1, the two kernel density functions become a lot closer after the matching.

Column (3) of Panels A and B of Table VII report regression results estimated based on the

matching observations. The effect of 𝑅&𝐷 on 𝑃𝑎𝑦𝑜𝑢𝑡 is significantly positive and consistent with

our first hypothesis. The regression coefficient of 𝑅&𝐷 × 𝐶𝑎𝑠ℎ and 𝑅&𝐷 × 𝑇𝑆𝐴 are both

significantly negative, indicating that the positive effect of R&D investments on dividend payout

is stronger for firms with lower cash holdings and less financial constraints, as predicted by our

second and third hypotheses.

5. Concluding Remarks

This paper examines the effect of R&D investments on firms’ dividend policies in China

which has a unique institutional setting. In contrast to the negative investment-dividend

relationship widely documented in the literature, we show that firms with more R&D investments

tend to pay more dividends in China and this effect is more pronounced for firms with lower cash

holdings and less financial constraints. Unlike capital expenditures, R&D investments are long-

22
term projects and largely financed by external capital markets. Firms with more R&D investments

tend to have more future equity financing needs, and therefore under the semi-mandatory dividend

policy there are strategic incentives for them to pay dividends in order to meet the prerequisite of

external equity financing.

The SMDP promotes dividend payments by linking them to equity financing. It creates

extra costs of equity financing since dividend payments are mandatory for accessing external

capital markets. The SMDP aims to protect investors’ rights but we demonstrate that such semi-

mandatory dividend payments have adverse impacts on firm value and sustainable growth. These

effects are stronger for firms with more R&D investments. The CSRC uses a merit-based system

to regulate share issuance of listed firms, which means that in practice financially constrained firms

are less affected by the SMDP. Nonetheless, R&D firms are more disadvantaged by the SMDP

compared to non-R&D firms due to their greater need for equity financing. Given that investments

in R&D represent a key driver of long-term economic growth, Chinese policy-makers may have to

consider waiving the requirements of SMDP on R&D firms in order to stimulate business

innovation activities.

23
REFERENCES
Adedeji, A., 1998. Does the pecking order hypothesis explain the dividend payout ratios of firms in
the UK? Journal of Business Finance and Accounting 25, 1127−1155.
Allen, F., Qian, J., Qian, M., 2005. Law, finance, and economic growth in China. Journal of
Financial Economics 77, 57−116.
Attig, N., Boubakri, N., Ghoul, S. E., Guedhami, O., 2015. The global financial crisis, family control,
and dividend policy. Financial Management 45, 291−313.
Bah, R., Dumontier, P., 2001. R&D intensity and corporate financial policy: Some international
evidence. Journal of Business Finance and Accounting 28, 671−692.
Bates, T. W., Kahle, K. M., Stulz, R. M., 2009. Why do U.S. firms hold so much more cash than
they used to? Journal of Finance 64, 1985−2021.
Booth, G. G., Junttila, J., Kallunki, J., Rahiala, M., Sahlström, P., 2006. How does the financial
environment affect the stock market valuation of R&D spending? Journal of Financial
Intermediation 15, 197−214.
Bradford, W., Chen, C., Zhu, S., 2013. Cash dividend policy, corporate pyramids, and ownership
structure: Evidence from China. International Review of Economics and Finance 27, 445−464.
Brockman, P., Unlu, E., 2009. Dividend policy, creditor rights, and the agency costs of debt. Journal
of Financial Economics 92, 276–298.
Brown, J.R., Fazzari, S.M., Petersen, B.C., 2009. Financing innovation and growth: cash flow,
external equity, and the 1990s R&D boom. Journal of Finance, 64, 151–185.
Brown, J. R., Martinsson, G., Petersen, B. C., 2012. Do financing constraints matter for R&D?
European Economic Review 56, 1512–1529.
Brown, J. R., Petersen, B. C., 2011. Cash holdings and R&D smoothing. Journal of Corporate
Finance 17, 694–709.
Chay, J. B., Suh, J., 2009. Payout policy and cash-flow uncertainty. Journal of Financial Economics
93, 88–107.
Chen, D., Jian, M., Xu, M., 2009. Dividends for tunneling in a regulated economy: The case of China.
Pacific-Basin Finance Journal 17, 209−223.
Chen, H. Y., Gupta, M. C., Lee, A. C., 2013. Sustainable growth rate, optimal growth rate, and
optimal payout ratio: A joint optimization approach. Journal of Banking and Finance 37,
1205−1222.
Chen, J., Leung, W. S., Goergen, M., 2017. The impact of board gender composition on dividend
payouts. Journal of Corporate Finance 43, 86−105.
DeAngelo, H., DeAngelo, L., Stulz, R. M., 2006. Dividend policy and the earned/contributed capital
mix: A test of the life-cycle theory. Journal of Financial Economics 81, 227–54.
Deng, L., K, Li, S., Liao, M., Wu, X., 2013. Dividends, investment and cash flow uncertainty:
Evidence from China. International Review of Economics and Finance 27, 112–124.
Denis, D., Osobov, I., 2008. Why do firms pay dividends? International evidence on the determinants
of dividend policy. Journal of Financial Economics 89, 62–82.
Fama, E. F., 1974. The empirical relationships between the dividend and investment decisions of
firms. American Economic Review 64, 304–318.
Fama, E. F., French, K. R., 1998. Taxes, financing decisions, and firm value. Journal of Finance 53,
819–843.
Fama, E.F., French, K.R., 2001. Disappearing dividends: changing firm characteristics or lower
propensity to pay? Journal of Financial Economics 60, 3–43.
Fama, E., French, K., 2002. Testing trade-off and pecking order predictions about dividends and
debt. Review of Financial Studies 15, 1–33.

24
Gatchev, V. A., Spindt, P. A., Tarhan, V., 2009. How do firms finance their investments? The
relative importance of equity issuance and debt contracting costs. Journal of Corporate
Finance 15, 179−195.
Guariglia, A., Liu, P., 2014. To what extent do financing constraints affect Chinese firms’ innovation
activities? International Review of Financial Analysis 36, 223−240.
Gugler, K., 2003. Corporate governance, dividend payout policy, and the interrelation between
dividends, R&D, and capital investment. Journal of Banking and Finance 27, 1297−1321.
Gul, F., Tsui, J., 1998. A test of the free cash flow and debt monitoring hypotheses: Evidence from
audit pricing. Journal of Accounting and Economics 24, 219–237.
Hadlock, C. J., Pierce, J. R., 2010. New evidence on measuring financial constraints: Moving beyond
the KZ index. Review of Financial Studies 23, 1909−1940.
Hall, B. H., 2002. The financing of research and development. Oxford Review of Economic Policy
18, 35−51.
He, W., Ng, L., Zaiats, N., Zhang, B., 2017. Dividend policy and earnings management across
countries. Journal of Corporate Finance 42, 267−286.
He, Z., Wintoki, M. B., 2016. The cost of innovation: R&D and high cash holdings in US firms.
Journal of Corporate Finance 41, 280−303.
Hoberg, G., Phillips, G., Prabhala, N., 2014. Product market threats, payouts, and financial flexibility.
Journal of Finance 69, 293−324.
Huang, J. J., Shen, Y., Sun, Q., 2011. Nonnegotiable shares, controlling shareholders and dividend
payments in China. Journal of Corporate Finance 17, 122–133.
Kim, W., Weisbach, M. S., 2008. Motivations for public equity offers: An international perspective.
Journal of Financial Economics 87, 281–307.
Klein, D. P., Belt, B., 1993. Sustainable growth and choice of financing: A test of the pecking order
hypothesis. Review of Financial Economics 3, 143-154.
Koo, D.S., Ramalingegowda, S., Yu, Y., 2017. The effect of financial reporting quality on corporate
dividend policy. Review of Accounting Studies 22, 753–790.
Korajczyk, R. A., Levy, A., 2003. Capital structure choice: Macroeconomic conditions and financial
constraints. Journal of Financial Economics 68, 75−109.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 2006. Agency problems and dividend
policy around the world. Journal of Finance 55, 1–34.
McLean, R. D., 2011. Share issuance and cash savings. Journal of Financial Economics 99,693−715.
Miller, M.H., Modigliani, F., 1961. Dividend policy, growth, and the valuation of shares. Journal of
Business 34, 411–433.
Partington, G. H., 1985. Dividend policy and its relationship to investment and financing policies:
Empirical evidence. Journal of Business Finance and Accounting 12, 531−542.
Pinkowitz, L., Stulz, R., Williamson, R., 2006. Does the contribution of corporate cash holdings
and dividends to firm value depend on governance? A cross-country analysis. Journal of
Finance 61, 2725–2751.
Ramalingegowda, S., Wang, C., Yu, Y., 2013. The role of financial reporting quality in mitigating
the constraining effect of dividend policy on investment decisions. The Accounting Review 88,
1007–1039.
Sarig, O., 2004. A time-series analysis of corporate payout polices. Review of Finance 8, 515–536.
Sharma, V., 2011. Independent directors and the propensity to pay dividends. Journal of Corporate
Finance 17, 1001−1015.
Soo, D. S., Ramalingegowda, S., Yu, Y., 2017. The effect of financial reporting quality on corporate
dividend policy. Review of Accounting Studies 22, 753−790.

25
Wahal, S., McConnell, J., 2000. Do institutional investors exacerbate managerial myopia? Journal
of Corporate Finance 6, 307−329.
Wei, Z., Li, C., Wu, Y., 2017. The semi-mandatory dividend rules, refinancing motivation and
classical dividend theories––An empirical study based on agency theory and signaling theory.
Accounting Research 7, 55–61.
White, H., 1980. A heteroskedasticity-consistent covariance matrix estimator and a direct test for
heteroscedasticity. Econometrica 48, 817–838.

26
Figure 1. Kernel density function of the treatment and control groups before and after the propensity score
matching. We use logistic regression to model the probability that a firm has R&D investments, where firm age, cash
holdings, industry dummy, ROA and controlling shareholder ownership are adopted as determinants and the predicted
probability is obtained as the propensity score. Before matching, the treatment (control) group includes observations
of firms with (without) R&D investment. The kernel density functions of their propensity scores are plotted
respectively in the upper panel of this figure. We adopt propensity score matching (PSM) to negate observable
differences in firm characteristics between R&D firms and non-R&D firms. For each observation in the treatment
group, we match it with one observation of non-R&D firms to ensure their propensity scores are as close as possible
to form the control group. The plots in the lower panel of this figure illustrate the kernel density functions of the
propensity scores of the treatment and matched (control) groups, respectively.

Before Matching
8

6
Density

Treat
4
Control

0
0.0 0.2 0.4 0.6 0.8 1.0
Propensity Score

After Matching
8

6
Density

Treat
4
Control

0
0.0 0.2 0.4 0.6 0.8 1.0
Propensity Score

27
Table I
Variable Definitions
This table provides definitions of the variables.

Variables Definition
Dependent variable
Dividend payout ratio (𝑃𝑎𝑦𝑜𝑢𝑡) Dividend per share divided by earnings per share.
Independent variables
𝑅&𝐷 R&D expenditures divided by book value of total assets.
Cash holdings (𝐶𝑎𝑠ℎ) Cash and cash equivalents divided by net assets, measured at the
beginning of the year.
Transformed SA index (𝑇𝑆𝐴) 𝑆𝐴−𝑚𝑖𝑛(𝑆𝐴)
Transformed SA index, calculated as 𝑇𝑆𝐴 = ,
𝑚𝑎𝑥(𝑆𝐴)−𝑚𝑖𝑛(𝑆𝐴)
where 𝑆𝐴 denotes the size-age (SA) index, a measure of financial
constraints proposed by Hadlock and Pierce (2010) defined as
𝑆𝐴 = 0.043 × 𝑆𝑖𝑧𝑒 2 − 0.737 × 𝑆𝑖𝑧𝑒 − 0.04 × 𝐴𝑔𝑒, and 𝑚𝑎𝑥(𝑆𝐴)
and 𝑚𝑖𝑛(𝑆𝐴) are the maximum and minimum SA index value in
the whole sample, respectively.
Control variables
Firm size (𝑆𝑖𝑧𝑒) Natural logarithm of book value of total assets.
Leverage (𝐿𝑒𝑣) Total debts divided by book value of total assets.
Return on assets (𝑅𝑂𝐴) Ratio of earnings before interest and income tax to book value of
total assets.
Firm growth prospect (𝑇𝑜𝑏𝑖𝑛’𝑠 𝑄) Market value of equity plus book value of total liabilities then
divided by book value of total assets.
Free cash flow (𝐹𝐶𝐹) Total operating revenue-income tax expenses + depreciation &
amortization - changes in working capital - capital expenditure and
then divided by number of total outstanding shares.
Firm age (𝐴𝑔𝑒) Natural logarithm of “1 + 𝑎𝑔𝑒”, where 𝑎𝑔𝑒 is the number of years
since IPO.
Capital expenditures (𝐶𝑎𝑝𝑒𝑥) Capital expenditures divided by book value of total assets.
Working capital (𝑊𝑜𝑟𝑘𝑐𝑎𝑝) Working capital divided by book value of total assets.
Firm risk (𝐵𝑒𝑡𝑎) Firm stock’s beta coefficient, which is obtained by regressing
monthly excess stock return on excess market return over 3 years.
Controlling shareholder ownership Number of shares held by controlling shareholders divided by total
(𝑆ℎ1) number of outstanding shares.
Board size (𝐵𝑠𝑖𝑧𝑒) Natural logarithm of total number of board of directors.
Board independence (𝐵𝑖𝑛𝑑𝑒𝑝) Number of independent directors divided by total number of board
of directors.
CEO duality (𝐷𝑢𝑎𝑙𝑖𝑡𝑦) Dummy variable, which equals one if the CEO and chairman are the
same person and zero otherwise.

28
Table II
Descriptive Statistics
Panel A of this table presents the full sample summary statistics of the main variables, consisting of 11,840 firm-year
observations, while the average statistics for the subsamples of R&D and non-R&D firms and the difference between
the two subsamples are reported in Panel B. The variables are defined in Table I and dividend per share (DPS) is the
amount of cash dividends that a company’s shareholders receive on a per-share basis. In Panel C, we sort observations
into subsamples based on R&D investments and report their average characteristics of dividend policy. Propensity to
pay dividends is calculated as the percentage of observations with positive dividend payout ratio. For each subsample,
average R&D investments, propensity to pay dividends, average DPS, and average payout ratio are reported. Symbols
***, ** and * indicate significance at the 1%, 5%, and 10% levels, respectively.

Panel A: Summary statistics of the whole sample


Variable Obs. Mean Stdev. Median Max. Min.
𝑃𝑎𝑦𝑜𝑢𝑡 11,840 0.238 0.273 0.182 1.5 0
𝐷𝑃𝑆 11,840 0.103 0.181 0.05 6.419 0
𝑅&𝐷 11,840 0.0074 0.0134 0 0.0677 0
𝐶𝑎𝑠ℎ 11,840 0.199 0.163 0.151 1 0
𝑇𝑆𝐴 11,840 0.4112 0.0955 0.3975 1 0
𝑆𝑖𝑧𝑒 11,840 21.81 1.37 21.67 28.51 0
𝐿𝑒𝑣 11,840 0.454 0.229 0.450 1.140 0.045
𝑅𝑂𝐴 11,840 0.064 0.067 0.057 0.302 -0.172
𝑇𝑜𝑏𝑖𝑛’𝑠 𝑄 11,840 2.345 2.189 1.696 12.81 0.211
𝐹𝐶𝐹 11,840 0.347 0.778 0.270 3.326 -2.240
𝐴𝑔𝑒 11,840 13.952 0.552 15 26 2
𝐶𝑎𝑝𝑒𝑥 11,840 0.0583 0.0573 0.0407 0.264 0.0001
𝑊𝑜𝑟𝑘𝑐𝑎𝑝 11,840 0.328 4.399 0.23 610.1 -123.9
𝐵𝑒𝑡𝑎 11,840 1.009 0.297 1.008 8.812 -2.081
𝑆ℎ1 11,840 0.361 0.156 0.341 100 0.003
𝐵𝑠𝑖𝑧𝑒 11,840 10.475 1.973 11 26 4
𝐵𝑖𝑛𝑑𝑒𝑝 11,840 0.337 0.103 0.333 0.8 0
𝐷𝑢𝑎𝑙𝑖𝑡𝑦 11,840 0.215 0.411 0 1 0
Panel B: Summary statistics of R&D and non-R&D subsamples and their mean differences
Subsample with 𝑅&𝐷>0 Subsample with 𝑅&𝐷=0
Variable Mean Obs. Mean Obs. Difference T-statistics
𝑃𝑎𝑦𝑜𝑢𝑡 0.262 4,589 0.218 7,251 0.044 11.435***
𝐷𝑃𝑆 0.115 4,589 0.0934 7,251 0.0216 8.386***
𝐶𝑎𝑠ℎ 0.228 4,589 0.173 7,251 -0.055 -24.094***
𝑇𝑆𝐴 0.4172 4,589 0.4063 7,251 0.0109 7.3516***
𝑆𝑖𝑧𝑒 21.81 4,589 21.81 7,251 0.0000 0.2397
𝐿𝑒𝑣 0.400 4,589 0.498 7,251 -0.098 -7.112***
𝑅𝑂𝐴 0.0658 4,589 0.0631 7,251 0.027 2.834***
𝑇𝑜𝑏𝑖𝑛’𝑠 𝑄 2.571 4,589 2.155 7,251 0.416 13.236***
𝐹𝐶𝐹 0.319 4,589 0.37 7,251 -0.051 -4.588***
𝐴𝑔𝑒 2.329 4,589 2.706 7,251 -0.377 -51.049***
𝐶𝑎𝑝𝑒𝑥 0.061 4,589 0.0534 7,251 0.0076 11.198***
𝑊𝑜𝑟𝑘𝑐𝑎𝑝 0.298 4,589 0.353 7,251 -0.055 0.885
𝐵𝑒𝑡𝑎 1.009 4,589 1.009 7,251 0.0000 0.029
𝑆ℎ1 0.355 4,589 0.366 7251 -0.011 -5.257***
𝐵𝑠𝑖𝑧𝑒 2.346 4,589 2.352 7,251 -0.006 2.182**
𝐵𝑖𝑛𝑑𝑒𝑝 0.318 4,589 0.367 7,251 -0.049 -39.710***
𝐷𝑢𝑎𝑙𝑖𝑡𝑦 0.243 4,589 0.181 7,251 0.062 11.431***

29
Table II-Continued
Panel C: Statistics for dividend payout of subsamples sorted based on R&D percentiles
P65 P70 P80 P90 P100
Percentile of 𝑅&𝐷 (0%-65%) (65%-70%) (70%-80%) (80%-90%) (90%-100%)
Average 𝑅&𝐷 0.0028 0.0065 0.0148 0.0253 0.0677
Propensity to pay dividends 59.26% 62.16% 73.01% 80.55% 83.28%
Average 𝐷𝑃𝑆 0.0832 0.0880 0.1027 0.1298 0.1229
Average 𝑃𝑎𝑦𝑜𝑢𝑡 0.2086 0.2271 0.2625 0.2860 0.2599

30
Table III
Effect of R&D Investments on Dividend Payout and the Role of Cash Holdings
This table reports the results of estimating Tobit regression models with dividend payout ratio as the dependent
variable. Variable definition is provided in Table I. Column (1-2) report the regression results based on the whole
sample. Using the sample average of cash holdings as the cutoff, we further consider two subsamples, one with cash
holdings lower than the average value and the other with higher cash holdings. Column (3-4) report the regression
results of the two subsamples, respectively. Heteroscedasticity-robust White (1980) t-statistics are shown in
parentheses. Symbols ***, ** and * indicate significance at the 1%, 5%, and 10% levels, respectively.

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


Whole sample Whole sample Subsample with Subsample with
low 𝐶𝑎𝑠ℎ high 𝐶𝑎𝑠ℎ
Dependent variable 𝑃𝑎𝑦𝑜𝑢𝑡 𝑃𝑎𝑦𝑜𝑢𝑡 𝑃𝑎𝑦𝑜𝑢𝑡 𝑃𝑎𝑦𝑜𝑢𝑡
𝑅&𝐷 0.973*** 0.952*** 1.833*** 0.217
(3.42) (3.31) (3.68) (0.64)
𝑅&𝐷 × 𝐶𝑎𝑠ℎ -4.601**
(-2.51)
𝐶𝑎𝑠ℎ 0.148***
(4.53)
𝑆𝑖𝑧𝑒 0.0632*** 0.0628*** 0.0680*** 0.0532***
(17.04) (16.95) (12.90) (9.91)
𝐿𝑒𝑣 -0.573*** -0.550*** -0.617*** -0.490***
(-25.74) (-24.31) (-19.34) (-15.07)
𝑅𝑂𝐴 1.16*** 1.14*** 1.31*** 0.978***
(18.00) (17.75) (14.09) (10.93)
𝑇𝑜𝑏𝑖𝑛’𝑠 𝑄 -0.0412*** -0.0421*** -0.0461*** -0.0381***
(-14.57) (-14.83) (-8.94) (-11.24)
𝐹𝐶𝐹 0.0175*** 0.0184*** 0.0205*** 0.0148**
(4.25) (4.51) (3.48) (2.55)
𝐴𝑔𝑒 -0.132*** -0.131*** -0.123*** -0.135***
(-11.89) (-11.78) (-7.45) (-9.10)
𝐶𝑎𝑝𝑒𝑥 0.0343 0.0293 0.247** -0.219**
(0.48) (0.41) (2.45) (-2.10)
𝑊𝑜𝑟𝑘𝑐𝑎𝑝 0.0086 -0.00406 -0.0108 0.0195
(1.08) (-0.62) (-1.49) (1.33)
𝐵𝑒𝑡𝑎 -0.0770*** -0.0745*** -0.0890*** -0.0541***
(-5.90) (-5.73) (-4.42) (-3.17)
𝑆ℎ1 0.124*** 0.122*** 0.109*** 0.145***
(5.34) (5.23) (3.35) (4.34)
𝐵𝑠𝑖𝑧𝑒 0.0234 0.0217 0.0496** -0.0130
(1.35) (1.25) (2.03) (-0.53)
𝐵𝑖𝑛𝑑𝑒𝑝 -0.249*** -0.252*** -0.235** -0.273***
(-3.66) (-3.70) (-2.45) (-2.86)
𝐷𝑢𝑎𝑙𝑖𝑡𝑦 -0.0187** -0.0180* -0.0128 -0.0195
(-1.97) (-1.90) (-0.95) (-1.47)
Constant -0.597*** -0.624*** -0.742*** -0.341**
(-5.98) (-6) (-5.14) (-2.42)
Year dummy YES YES YES YES
Industry dummy YES YES YES YES
F statistics 83.63 80.48 52.41 46.85
Pseudo R2 0.212 0.214 0.222 0.208
Obs. 11,840 11840 6,764 5,076

31
Table IV
Effect of R&D Investments on Dividend Payout and the Role of Financial Constraints
This table reports the results of estimating Tobit regression models with dividend payout ratio as the dependent
variable. Variable definition is provided in Table I. TSA denotes the transformed SA index, which proxies for
financial constraints. Column 1 reports the regression results based on the whole sample. Using the sample average
TSA as the cutoff, we further consider two subsamples, one with TSA lower than the average value and the other with
higher TSA. Column (2-3) report the regression results of the two subsamples, respectively. Heteroscedasticity-
robust White (1980) t-statistics are shown in parentheses. Symbols ***, ** and * indicate significance at the 1%, 5%,
and 10% levels, respectively.

(1) (2) (3)


Whole sample Subsample with Subsample with
low 𝑇𝑆𝐴 high 𝑇𝑆𝐴
Dependent variable 𝑃𝑎𝑦𝑜𝑢𝑡 𝑃𝑎𝑦𝑜𝑢𝑡 𝑃𝑎𝑦𝑜𝑢𝑡
𝑅&𝐷 1.041*** 1.640*** -0.119
(3.56) (3.39) (-0.35)
𝑅&𝐷 × 𝑇𝑆𝐴 -18.68***
(-6.16)
𝑇𝑆𝐴 -4.989***
(-11.22)
𝑆𝑖𝑧𝑒 0.434*** 0.148*** 0.0288***
(12.61) (11.94) (6.27)
𝐿𝑒𝑣 -0.572*** -0.706*** -0.482***
(-25.36) (-20.08) (-14.95)
𝑅𝑂𝐴 1.04*** 1.34*** 0.637***
(15.75) (13.64) (6.36)
𝑇𝑜𝑏𝑖𝑛’𝑠 𝑄 -0.0352*** -0.0393*** -0.0231***
(-11.82) (-8.90) (-4.86)
𝐹𝐶𝐹 0.0197*** 0.0693*** 0.0107**
(4.82) (6.39) (2.55)
𝐴𝑔𝑒 -0.308*** -0.261*** -0.0355***
(-15.76) (-12.54) (-2.78)
𝐶𝑎𝑝𝑒𝑥 -0.0175 -0.0391 -0.0883
(-0.24) (-0.30) (-1.01)
𝑊𝑜𝑟𝑘𝑐𝑎𝑝 0.00776 0.00358 0.0040
(0.97) (0.33) (0.29)
𝐵𝑒𝑡𝑎 -0.0885*** -0.134*** -0.0548***
(-6.53) (-5.73) (-3.77)
𝑆ℎ1 0.128*** 0.193*** 0.0828***
(5.52) (4.33) (3.14)
𝐵𝑠𝑖𝑧𝑒 0.0255 -0.00459 0.0391**
(1.47) (-0.14) (2.00)
𝐵𝑖𝑛𝑑𝑒𝑝 -0.212*** -0.460*** -0.0889
(-3.12) (-3.53) (-1.15)
𝐷𝑢𝑎𝑙𝑖𝑡𝑦 -0.0201** -0.0367** -0.0078
(-2.12) (-2.39) (-0.64)
Constant -6.231*** -1.957*** -0.118
(-11.56) (-6.75) (-0.96)
Year dummy YES YES YES
Industry dummy YES YES YES
F statistics 80.27 53.72 21.98
Pseudo R2 0.221 0.242 0.138
Obs. 11,840 6,072 5,768

32
Table V
Effect of R&D Investments on Future Equity Financing
This table reports the regression results with propensity or intensity of future equity financing as the dependent variable.
𝑆𝐸𝑂𝑠𝑖𝑧𝑒 denotes the natural logarithm of total SEO amount within the next three years, which measures the intensity
of future equity financing. 𝑆𝐸𝑂𝑑𝑢𝑚𝑚𝑦 takes the value of 1 if the firm engages in SEO within the next three years
and 0 otherwise, which measures the propensity of future equity financing. 𝑅&𝐷𝑑𝑢𝑚𝑚𝑦 takes the value of 1 if the
firm has R&D investments and 0 otherwise. Other variables are defined in Table I. Column (1) and (3) and column
(2) and (4) report the results of logistic regression and OLS regression, respectively. Heteroscedasticity-robust White
(1980) t-statistics or z-statistics are shown in parentheses. Symbols ***, ** and * indicate significance at the 1%, 5%,
and 10% levels, respectively.

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


Dependent variable 𝑆𝐸𝑂𝑑𝑢𝑚𝑚𝑦 𝑆𝐸𝑂𝑠𝑖𝑧𝑒 𝑆𝐸𝑂𝑑𝑢𝑚𝑚𝑦 𝑆𝐸𝑂𝑠𝑖𝑧𝑒
𝑅&𝐷 5.486** 25.62**
(2.52) (2.44)
𝑅&𝐷𝑑𝑢𝑚𝑚𝑦 0.187*** 0.798***
(3.27) (3.29)
𝐶𝑎𝑠ℎ -0.452* -1.458* -0.437* -1.385
(-1.88) (-1.70) (-1.82) (-1.63)
𝑆𝑖𝑧𝑒 0.141*** 0.744*** 0.135*** 0.713***
(5.35) (6.60) (5.07) (6.30)
𝐿𝑒𝑣 1.06*** 3.89*** 1.06*** 3.87***
(7.53) (7.61) (7.52) (7.58)
𝑅𝑂𝐴 1.71*** 5.05*** 1.76*** 5.23***
(4.00) (3.28) (4.11) (3.41)
𝑇𝑜𝑏𝑖𝑛’𝑠 𝑄 -0.0160 -0.00177 -0.0148 0.00132
(-0.92) (-0.03) (-0.86) (0.02)
𝐹𝐶𝐹 -0.138*** -0.591*** -0.133*** -0.572***
(-4.17) (-4.39) (-4.00) (-4.24)
𝐴𝑔𝑒 -1.012*** -4.396*** -1.015*** -4.413***
(-12.50) (-12.20) (-12.61) (-12.33)
𝐶𝑎𝑝𝑒𝑥 4.996*** 22.40*** 5.021*** 22.49***
(10.16) (9.59) (10.22) (9.64)
𝑊𝑜𝑟𝑘𝑐𝑎𝑝 -0.00408 0.0138 -0.00314 0.00837
(-0.00) (0.09) (-0.07) (0.01)
𝐵𝑒𝑡𝑎 0.199** 0.812** 0.195** 0.802**
(2.41) (2.31) (2.35) (2.28)
𝑆ℎ1 -1.02*** -4.03*** -0.996*** -3.96***
(-5.93) (-6.03) (-5.81) (-5.91)
𝐵𝑠𝑖𝑧𝑒 -0.233* -0.838* -0.244** -0.883*
(-1.93) (-1.68) (-2.01) (-1.78)
𝐵𝑖𝑛𝑑𝑒𝑝 0.137 0.702 0.100 0.536
(0.29) (0.36) (0.21) (0.28)
𝐷𝑢𝑎𝑙𝑖𝑡𝑦 0.0912 0.475* 0.0921 0.479*
(1.41) (1.67) (1.42) (1.69)
Constant -1.427** 1.170 -1.284* 1.883
(-2.04) (0.39) (-1.83) (0.63)
Year dummy YES YES YES YES
Industry dummy YES YES YES YES
F statistics 599.25 16.51 604.74 16.67
Pseudo R2 0.0578 0.0641 0.0582 0.0645
Obs. 10,072 10,072 10,072 10,072

33
Table VI
Effect of R&D Investments and Dividend Payout on Firm Value and Sustainable Growth
This table reports the regression results with market-to-book ratio or sustainable growth ratio as the dependent variable.
Market-to-book ratio is calculated as market value of total assets scaled by book value of total assets, proxies for firm
value. Sustainable growth ratio is sourced from the CSMAR database, defined as (net profit margin on sales × total
asset turnover × retained earnings) divided by (1− (net profit margin on sales × total asset turnover × retained
earnings)), which proxies for firm financial sustainable growth ability. Sales growth (𝑆𝑎𝑙𝑒𝑠𝑔𝑟𝑜𝑤𝑡ℎ) is adopted as a
proxy for firm growth and included as a control variable. Dividend payout ratio (𝑃𝑎𝑦𝑜𝑢𝑡), R&D investments (𝑅&𝐷)
and other variables are defined in Table I. Heteroscedasticity-robust White (1980) t-statistics are shown in parentheses.
Symbols ***, ** and * indicate significance at the 1%, 5%, and 10% levels, respectively.

Dependent (1) (2) (3) (4) (5) (6)


variable Market-to-book ratio Sustainable growth ratio
𝑃𝑎𝑦𝑜𝑢𝑡 -0.440*** -0.466*** -0.0510*** -0.0527***
(-9.22) (-9.33) (-18.58) (-18.89)
𝑅&𝐷 9.859*** 10.08*** 0.366*** 0.383***
(6.59) (6.75) (4.29) (4.37)
𝑃𝑎𝑦𝑜𝑢𝑡 × 𝑅&𝐷 -15.62*** -1.002***
(-2.91) (-3.17)
𝐶𝑎𝑠ℎ 1.279*** 1.190*** 1.217*** 0.0282* 0.0223 0.0262*
(7.82) (7.29) (7.48) (1.94) (1.53) (1.79)
𝑆𝑖𝑧𝑒 -0.797*** -0.809*** -0.796*** 0.0125*** 0.0109*** 0.0125***
(-35.77) (-36.33) (-35.59) (6.53) (5.70) (6.53)
𝐿𝑒𝑣 -0.595*** -0.470*** -0.566*** -0.0343** -0.0217 -0.0332**
(-4.62) (-3.72) (-4.39) (-2.21) (-1.40) (-2.14)
𝑅𝑂𝐴 6.00*** 5.76*** 5.88*** 1.30*** 1.28*** 1.29***
(15.06) (14.43) (14.73) (26.07) (25.73) (25.85)
𝑆𝑎𝑙𝑒𝑠𝑔𝑟𝑜𝑤𝑡ℎ -0.105** -0.0941** -0.106** 0.0102* 0.0117** 0.0101*
(-2.31) (-2.08) (-2.34) (1.95) (2.25) (1.94)
𝐹𝐶𝐹 0.0106 0.00416 0.0109 0.00557*** 0.00486*** 0.00558***
(0.70) (0.27) (0.72) (3.39) (2.93) (3.40)
𝐴𝑔𝑒 0.148*** 0.232*** 0.197*** -0.000110 0.00553** 0.00167
(3.20) (4.99) (4.28) (-0.04) (2.02) (0.62)
𝐶𝑎𝑝𝑒𝑥 0.0583 0.0758 0.0236 0.0147 0.0208 0.0140
(0.20) (0.26) (0.08) (0.63) (0.88) (0.60)
𝑊𝑜𝑟𝑘𝑐𝑎𝑝 -0.112*** -0.106*** -0.106*** -0.00609 -0.00574 -0.00579
(-2.87) (-2.95) (-3.05) (-0.71) (-0.67) (-0.67)
𝐵𝑒𝑡𝑎 -0.134** -0.124** -0.141** 0.0102** 0.0122*** 0.0101**
(-2.33) (-2.16) (-2.44) (2.31) (2.77) (2.29)
𝑆ℎ1 0.316*** 0.294*** 0.319*** 0.00961 0.00631 0.00929
(3.67) (3.42) (3.71) (1.60) (1.04) (1.55)
𝐵𝑠𝑖𝑧𝑒 0.391*** 0.376*** 0.380*** -0.0172*** -0.0187*** -0.0177***
(5.30) (5.08) (5.14) (-3.55) (-3.85) (-3.65)
𝐵𝑖𝑛𝑑𝑒𝑝 2.256*** 2.318*** 2.241*** -0.0284 -0.0219 -0.0298*
(8.28) (8.47) (8.25) (-1.57) (-1.21) (-1.66)
𝐷𝑢𝑎𝑙𝑖𝑡𝑦 0.148*** 0.149*** 0.144*** 0.000133 0.000359 -0.000155
(3.63) (3.67) (3.56) (0.04) (0.12) (-0.05)
Constant 18.46*** 18.38*** 18.32*** -0.232*** -0.225*** -0.235***
(39.75) (39.32) (39.25) (-6.20) (-6.03) (-6.31)
Year dummy YES YES YES YES YES YES
Industry dummy YES YES YES YES YES YES
F statistics 141.9 141.6 137.2 96.50 92.29 95.03
Pseudo R2 0.489 0.489 0.492 0.473 0.466 0.474
Obs. 11,840 11,840 11,840 11,840 11,840 11,840

34
Table VII
Endogeneity Tests: Two-Stage Least Squares Regression and Propensity Score Matching
This table presents the endogeneity test results. Variables are defined in Table I. In column (1) and (2), we perform
two-stage-least squares (2SLS) regression. Industry-adjusted 𝑅&𝐷 is used as an instrumental variable for R&D
investment, which is the difference between 𝑅&𝐷 and the industry average 𝑅&𝐷 for that year. Column 1 reports the
results of the 1st stage OLS regression and column 2 reports the 2nd stage Tobit estimation results. In column (3), we
use propensity score matching (PSM) to match each observation with positive 𝑅&𝐷 (in treatment group) with an
observation with zero 𝑅&𝐷 to make their propensity scores as close as possible. The propensity score is the
probability that a firm has R&D investments predicted by a logistic regression with firm age, cash holdings, industry
dummy, 𝑅𝑂𝐴 and controlling shareholder ownership as determinants. We obtain 9,178 matching observations,
including 4,589 observations with positive 𝑅&𝐷 and 4,589 observations with zero 𝑅&𝐷 and report their regression
results in column (3). Heteroscedasticity-robust White (1980) t-statistics are shown in parentheses. Symbols ***, **
and * indicate significance at the 1%, 5%, and 10% levels, respectively.

Panel A:Effect of R&D investments on dividend payout and the role of cash holdings
(1) (2) (3)
2SLS: 1st stage 2SLS: 2nd stage PSM
Dependent variable 𝑅&𝐷 𝑃𝑎𝑦𝑜𝑢𝑡 𝑃𝑎𝑦𝑜𝑢𝑡
𝑅&𝐷 0.845*** 0.775***
(2.90) (2.62)
𝑅&𝐷 × 𝐶𝑎𝑠ℎ -4.184* -5.093***
(-1.91) (-2.64)
𝐶𝑎𝑠ℎ 0.0052*** 0.153*** 0.142***
(5.44) (4.70) (3.67)
Industry-adjusted 𝑅&𝐷 0.689***
(19.12)
𝑆𝑖𝑧𝑒 0.00016 0.0628*** 0.0653***
(1.63) (16.95) (14.44)
𝐿𝑒𝑣 -0.0020*** -0.553*** -0.565***
(-3.96) (-24.42) (-21.24)
𝑅𝑂𝐴 0.0086*** 1.15*** 1.25***
(5.17) (17.89) (17.05)
𝑇𝑜𝑏𝑖𝑛’𝑠 𝑄 0.0004*** -0.0419*** -0.0390***
(5.46) (-14.84) (-12.32)
𝐹𝐶𝐹 -0.00006 0.0184*** 0.0201***
(-0.54) (4.49) (3.91)
𝐴𝑔𝑒 -0.00486*** -0.136*** -0.141***
(-13.07) (-12.34) (-11.10)
𝐶𝑎𝑝𝑒𝑥 0.00209 0.0326 0.0263
(1.08) (0.45) (0.30)
𝑊𝑜𝑟𝑘𝑐𝑎𝑝 -0.00037 -0.00375 -0.0132*
(-1.08) (-0.56) (-1.92)
𝐵𝑒𝑡𝑎 0.00054* -0.0738*** -0.0627***
(1.74) (-5.68) (-4.19)
𝑆ℎ1 -0.109* 0.120*** 0.142***
(-1.75) (5.16) (5.06)
𝐵𝑠𝑖𝑧𝑒 0.00064 0.0224 0.0389*
(1.37) (1.30) (1.90)
𝐵𝑖𝑛𝑑𝑒𝑝 -0.00062 -0.252*** -0.247***
(-0.35) (-3.70) (-3.10)
𝐷𝑢𝑎𝑙𝑖𝑡𝑦 0.00024 -0.0178* -0.0189*
(0.84) (-1.88) (-1.80)
Constant 0.0126*** -0.455*** -0.535***
(4.51) (-4.39) (-4.20)

35
Table VII-Continued

Year dummy YES YES YES


Industry dummy YES YES YES
F statistics 78.07 80.40 62.95
Pseudo R2 0.2339 0.214 0.210
Obs. 11,840 11,840 9,178
Panel B:Effect of R&D investments on dividend payout and the role of financial constraints
(1) (2) (3)
2SLS: 1st stage 2SLS: 2nd stage PSM
Dependent variable 𝑅&𝐷 𝑃𝑎𝑦𝑜𝑢𝑡 𝑃𝑎𝑦𝑜𝑢𝑡
𝑅&𝐷 0.857*** 0.943***
(3.97) (4.43)
𝑅&𝐷 × 𝑇𝑆𝐴 -5.580** -18.62***
(-2.36) (-8.19)
𝑇𝑆𝐴 -0.0814*** -4.799*** -4.506***
(-10.53) (-15.32) (-11.91)
Industry-adjusted 𝑅&𝐷 0.702***
(27.43)
𝑆𝑖𝑧𝑒 0.0062*** 0.425*** 0.399***
(10.45) (17.49) (13.72)
𝐿𝑒𝑣 -0.0021*** -0.574*** -0.568***
(-5.92) (-36.03) (-30.18)
𝑅𝑂𝐴 0.0071*** 1.03*** 1.15***
(6.04) (21.94) (21.58)
𝑇𝑜𝑏𝑖𝑛’𝑠 𝑄 0.0006*** -0.0338*** -0.0339***
(10.44) (-16.16) (-14.40)
𝐹𝐶𝐹 -0.00004 0.0193*** 0.0218***
(-0.45) (6.66) (6.04)
𝐴𝑔𝑒 -0.0078*** -0.310*** -0.296***
(-19.12) (-22.45) (-18.38)
𝐶𝑎𝑝𝑒𝑥 0.0015 0.0024 -0.0180
(1.12) (0.05) (-0.30)
𝑊𝑜𝑟𝑘𝑐𝑎𝑝 -0.0002 0.0075 0.0022
(-0.80) (1.31) (0.34)
𝐵𝑒𝑡𝑎 0.0003 -0.0884*** -0.0891***
(1.29) (-9.19) (-8.23)
𝑆ℎ1 -0.0009* 0.127*** 0.147***
(-1.94) (7.74) (7.34)
𝐵𝑠𝑖𝑧𝑒 0.0007** 0.0258** 0.0318**
(2.18) (2.10) (2.20)
𝐵𝑖𝑛𝑑𝑒𝑝 0.0001 -0.219*** -0.164***
(0.06) (-4.54) (-2.96)
𝐷𝑢𝑎𝑙𝑖𝑡𝑦 0.0002 -0.0194*** -0.0191**
(1.14) (-2.87) (-2.54)
Constant -0.0839*** -6.100*** -5.696***
(-9.18) (-16.05) (-12.50)
Year dummy YES YES YES
Industry dummy YES YES YES
F statistics 158.2 159.1 124.3
Pseudo R2 0.237 0.218 0.210
Obs. 11,840 11,840 9,178

36

You might also like