Stock liquidity, financial constraints, and innovation in Chinese SMEs

Abstract

This study investigates the relationship between stock liquidity and firm innovation for publicly traded growing small and medium-sized enterprises (SMEs) in China using both innovation input and output. We collected samples of 785 SMEs from China’s Shenzhen Growth Enterprises Market without the financial industry from 2010 to 2020. The empirical findings demonstrate a significant positive relationship between stock liquidity and both innovation input, as measured by R&D investments, and innovation output, as proxied by patenting activities. A series of robustness tests demonstrate the reliability of our results. Increased liquidity enhances SMEs' innovation mainly by alleviating financial constraints, whereas the mediating effect of mergers and acquisitions (M&A) is not apparent at the firm level. Furthermore, the inhibitory effect of blockholder ownership on firm innovation is weak. Further analysis reveals that this favorable impact can last for at least four years, with manufacturing SMEs benefiting the most. Our study shows that the innovation abilities of SMEs can be enhanced by improving stock liquidity, which is mainly driven by tackling financial constraints.

Introduction

Small and medium-sized enterprises (SMEs) are critical to economic growth because of their contributions to job development, employment, and productivity, especially in developing countries (Ayyagari et al. 2007, 2011; Neagu 2016). SMEs constitute the majority of enterprises worldwide, representing approximately 90% of the world's enterprises and over 50% of employment. In emerging economies, SMEs contribute up to 40% of the national income and create seven out of ten jobs.¹ Cefis and Marsili (2006) argue that innovation is the engine for the survival of small and young firms. Moreover, innovation is one of the most important competitive strategies for both small and large firms, where the limited financing capacity of SMEs is a common problem limiting their innovative activities (Kaufmann and Tödtling 2002).

Berger and Udell (2006) point out that SMEs’ financing options change at all stages of their life cycles. Due to problems such as opaque information, SMEs rely heavily on internal financing during the start-up phase. As SMEs grow and enhance information transparency, they become better positioned to issue securitized loans and publicly listed shares to raise funds (Berger and Udell, 2006; Abdulsaleh and Worthington 2013; Ghak and Zarrouk 2022; Lim et al. 2023). Access to equity markets is a widespread option for national policymakers worldwide to solve financing problems for SMEs. According to the World Federation of Exchanges (WFE) (2018), 37 specialized SME exchanges operated throughout the WFE membership as of December 2019. These exchanges provide financially challenged SMEs with an alternative avenue to raise capital, both during the initial listing and following the initial public offering. Despite having only 13 SME exchanges, the Asia–Pacific region has the most listed SMEs, with the highest market capitalization and capital raised (WFE 2018). Similar to the NASDAQ market in the U.S., China offers a financing platform for growth-oriented SMEs: the Shenzhen Growth Enterprises Market (GEM). As of December 2020, 895 firms were listed on the Shenzhen GEM.²

Bulter et al. (2007) state that when market liquidity is low, issuance costs for securities dealers and transaction costs for investors are higher; thus, firms have higher financing costs. Therefore, the higher the liquidity, the more the market resource allocation can be utilized, and the better the access to financing for SMEs. Amihud and Noh (2021) find a positive and significant pricing of the illiquidity factor’s conditional risk, which increases during times of financial distress. Innovation requires investment in resources obtained from financing (Peneder 2008; Xiang et al. 2022). Thus, the relationship between stock liquidity and firm innovation has become a widespread concern. Fang et al. (2014) collect samples from the NYSE, AMEX, and NASDAQ markets,³ and their empirical results show that at the firm level, an increase in liquidity reduces future innovation through two possible mechanisms: (1) heightened risk of hostile takeovers and (2) a greater presence of institutional investors who do not actively obtain or monitor information. Based on the findings of Fang et al. (2014) at the market level, Vo (2014) investigates the relationship between asset liquidity and firm innovation from a macro perspective for all publicly traded firms in the U.S., whose findings contradict those of Fang et al. (2014). Vo (2014) suggests that management shortsightedness does not exist at the market level and that aggregate liquidity can promote firm innovation through two mechanisms. First, increased stock market liquidity reduces the cost of obtaining external financing, making it easier for small firms to issue shares and finance innovations. Second, higher stock market liquidity leads to higher firm value and reduced transaction costs, allowing large firms to acquire innovations from smaller firms through mergers and acquisitions (M&A). Meanwhile, at the firm level, Zhong (2018) also comes to the opposite conclusion of Fang et al. (2014). He examined all Chinese listed firms and found that stock liquidity improves innovation in terms of R&D expenditure by alleviating financial constraints and increasing agency costs.

Wen et al. (2018) and Tang et al. (2022) challenge the findings of Fang et al. (2014) by asserting that the Chinese market, following the Share Splitting Reform⁴ aimed to increase share liquidity and improve capital market efficiency, contradicting the conclusion that liquidity impedes firm innovation. Instead, they suggest that reform enhances sustainable innovation capacity and competitiveness. Specifically, Wen et al. (2018) compare the results for state-owned enterprises (SOEs) and private firms from the perspective of ownership of the firm's equity and show that increased liquidity can only promote innovation in SOEs through two mechanisms: (1) the long-term entry and (2) strategic institutional investors in the case of the implementation of Share Splitting Reform. Tang et al. (2022) also emphasize the importance of ownership concentration, arguing that after the Share Splitting Reform in 2005, stock liquidity and firm innovation are positively correlated because blockholders, particularly institutional investors, increased their involvement in regulating and balancing ownership in firms. From the above, the findings at the firm level of Fang et al. (2014), Wen et al. (2018), Zhong (2018), and Tang et al. (2022) are the opposite. The primary reasons for this discrepancy are the different market microstructures in the U.S. market and Chinese markets. Specifically, the proportion of individual investors in the Chinese market is higher than that in the U.S. market, whereas the number of institutional investors (blockholders) in the U.S. market is higher. This may make institutional investors’ regulatory effects less pronounced in China. Furthermore, distinct trading systems, as exemplified by up-and- down-price ranges, compel institutional investors in the Chinese market to employ different trading strategies. This study focuses on the Chinese stock market because of its distinct characteristics. First, the role of stock liquidity in fostering innovation in emerging markets such as China has received little attention. China, the world’s second largest economy after the U.S., holds significant global economic importance as a representative emerging economy. Second, Pissarides (1999) highlights that SMEs are potentially the most active businesses in emerging economies where access to capital is recognized as a primary barrier to their development, constraining their innovative capacities. Third, in a market driven mainly by retail investors who lack specialized investment knowledge, the impact of liquidity on firms' innovation capabilities is likely to differ substantially from that in developed countries, where institutional investors are dominant.

Driven by these concerns, this study examines the relationship between liquidity and innovation in Chinese listed SMEs. Our study employs all nonfinancial firms traded on the Chinese Shenzhen GEM, a financing platform for growth-oriented SMEs, from 2010 to 2020. The baseline models regress R&D expenditure or the number of granted patents against the Closing Percentage Quoted Spread (CPQS) and a set of standard control variables. The empirical results show evidence of a positive relationship between stock liquidity and firm innovation in Chinese-listed SMEs, with a long-term positive impact of at least four years. The positive relationships pass a series of robustness checks using (1) alternative stock liquidity measures, (2) alternative firm innovation measures, (3) alternative estimation methods, (4) industry-specific regressions, and (5) endogeneity tests. It is worth highlighting that the results of industry-specific regressions show that the role of higher liquidity in promoting firm innovation (innovation input or output) is strongest in SMEs in the manufacturing industry.

Based on existing studies, we propose two mechanisms without an exogenous policy impact that may lead to this positive relationship: (1) reducing financial constraints and (2) the willingness of large firms to buy innovation. Our subsequent tests on mechanisms further show that among SMEs, the positive relationship between liquidity and innovation is mainly caused by the financing difficulties they face. At the firm level, the mediating effect of M&A on stock liquidity and firm innovation, as proposed by Vo (2014), is not evident. Motivated by Fang et al. (2014), we also test the effect of blockholder ownership, showing that both external blockholders, represented by institutions, and internal blockholders can inhibit firm innovation under higher liquidity, but this effect is not strong in SMEs.

Our study contributes to the limited literature on the relationship between stock liquidity and firm innovation in SMEs, especially for SMEs in emerging markets. Browsing through the literature on SMEs innovation is confined to other determinants of firm innovation, which can be grouped as follows: (1) financial factors (Laforet 2011; Xie et al. 2013; Wonglimpiyarat 2015; Yao and Yang 2022), (2) institutional factors (Zhu et al. 2012; Volchek et al. 2013; Donbesuur et al. 2020), (3) management (Alegre et al. 2011; Chereau 2015; Adla et al. 2019; Madison et al. 2022; Timothy 2022), (4) economic factors (North and Smallbone 2000; Kumar and Subrahmanya 2010; Bala Subrahmanya 2013; Gherghina et al. 2020), and (5) culture (Aksoy 2017; Gonzalez-Loureiro et al. 2017).

Many scholars investigated the obstacles that affect the development of large firms. However, little research has been conducted on SMEs, which are critical for their survival. Notably, although Vo (2014) presents a greater role of higher liquidity in promoting innovation in small firms, he focuses mainly on the market and neglects the effects of blockholders at the firm level. Furthermore, it is worth noting that most studies on firm innovation are limited to mature markets, such as the U.S., which makes it doubtful that developing markets dominated by SMEs can draw the same conclusions. Often, SMEs lack external financing. Zhong (2018) emphasizes the problem of financial constraints, whereas Wen et al. (2018) and Tang et al. (2022) neglect it in favor of focusing on ownership concentration in Chinese SMEs because of the Chinese Share Splitting Reform’s policy effect. The Chinese Share Splitting Reform was established in 2005 and completed by the end of 2006.⁵ The GEM, a market with reduced listing standards for mainly SMEs, was launched on 30th, October 2009, to solve the difficult financing for SMEs, which is already after the implementation of the shareholding reform. It is reasonable to infer that the mechanisms proposed by Wen et al. (2018) and Tang et al. (2022) may not be applicable to SMEs facing financing difficulties without exogenous policy shocks.

Our study addresses these gaps and contributes to existing literature as follows. First, we provide empirical evidence from the Chinese stock market, adding to the growing body of research on the effect of stock liquidity on SME innovation in financial markets. Second, we comprehensively examine all possible mechanisms and propose two ways to enhance innovation in SMEs: (1) resolving financial constraints, and (2) lowering M&A activity. Third, we complement Wen et al. (2018) and Tang et al. (2022) by showing that liquidity can improve innovation in Chinese listed SMEs without exogenous shocks, primarily by reducing financial constraints. Furthermore, our empirical evidence contradicts those of Vo (2014) and Fang et al. (2014), who focus on the U.S. market. On the one hand, at the firm level, the mediating effect of M&A on stock liquidity and firm innovation is not evident. By contrast, blockholder ownership can inhibit firm innovation under higher liquidity, but this effect is relatively weak in SMEs.

The remainder of this paper is organized as follows. section "theoretical basis for the hypotheses" reviews the relevant literature and develops our hypotheses on stock liquidity and firm innovation based on them. Section "Measurement of variables and model specification" presents the variable measurements and model specifications. Section "The data" describes the data collection and discusses the results of the summary statistics. Section "Baseline results: Stock liquidity and firm innovation" presents the baseline results, and robustness checks are presented in Section "Robustness checks". Section "Potential mechanisms" addresses possible mechanisms that may affect the relationship between stock liquidity and firm innovation. Section "Conclusion and discussions" offers concluding remarks, policy recommendations, limitations, and future research.

Theoretical basis for the hypotheses

Previous studies on the relationship between stock liquidity and firm innovation are restricted to U.S. stock markets and all Chinese A-share stocks (Vo 2014; Fang et al. 2014; Wen et al. 2018; Zhong 2018). Their findings lead to the opposing conclusion that liquidity promotes or inhibits firm innovation. In addition to the Chinese and U.S. markets, recent studies have explored the impact of innovation on other markets using factors such as mediators. Amin et al. (2023) investigate the impacts of a firm's information asymmetry on corporate innovation in the Korean market and find that a firm's innovation activities are positively affected by the quality of its information. Arifin et al. (2022) examine the principal-agent relationship and financing constraints to explain the level of corporate innovation in Indonesia. Zhang (2023) finds that financial constraints can reduce Indian firms’ motivation to engage in product innovation. Hanelt et al. (2021) employ panel data regressions to analyze a longitudinal dataset of the top automakers in the world and discover that digital M&As have a positive impact on digital innovation, which is largely mediated through the generation of new digital patents filed by the acquiring firms. Extending the work of Vo (2014), Fang et al. (2014), Wen et al. (2018), and Zhong (2018) on the linear relationship between stock liquidity and firm innovation, we propose the following four perspectives to support our linear relationship hypothesis.

Information asymmetry

According to information asymmetry theory, persons with sufficient information are frequently in a more beneficial position, whereas those with poorer information are in a less favorable position. First, Stein (1989) finds that information asymmetry between managers and investors, coupled with market pressure, compels managers to forfeit long-term investments such as innovation to avert a near-term share price decline and the risk of a hostile takeover. Therefore, managers prefer short-term investments that stabilize stock prices, and forgo long-term innovative investments to avoid threats to their positions if the firm is acquired (Fang et al. 2014).

Second, as shown by Kyle (1984) and Holmström and Tirole (1993), traders with increased liquidity make it easier for well-informed parties to conceal their information and capitalize on it. In pursuit of short-term gains, investors tend to seek undisclosed information, which does not foster innovation. Information asymmetry prompts executives with inside information to engage in speculative behavior when a firm's share price is overvalued, leading to a high cash output of shares for short-term maximization. Third, Graham et al. (2005) argue that Chief Financial Officers (CFOs) are often willing to sacrifice long-term projects to achieve short-term profit goals. Kyle and Vila (1991) show that higher liquidity can exacerbate the myopia of firm management that fails to be aware of the entry of outsiders disguised as hostile takeovers. As innovation activities often do not yield short-term profitability, myopia can prompt management to curtail investment in innovation and overlook potentially disguised hostile actions. Fourth, Porter (1992) indicates that enhanced liquidity reduces transaction costs for institutional investors, thus enabling smoother entry and exit from transactions. However, their trading practices, driven by current earnings news, may result in misvaluation and underinvestment in innovation. Institutional investors tend to prefer investing in firms with higher expected short-term returns (Bushee 2001). Thus, researchers grounded given that information asymmetry argue that liquidity hinders firm innovation (Fang et al. 2014).

Principal-agent relationship

In corporate governance, ownership and operations are often separated, with the business owner retaining residual claims and ceding operating power. First, as a highly liquid market allows shareholders to sell their shares more easily, it is more likely to allow blockholders to continue holding more shares at a lower cost and attract more blockholders to the market (Maug 1998; Edmans 2009). Maug (1998) indicates that increased liquidity encourages blockholders to engage in corporate governance oversight because the benefits of informed trading offset the associated costs. As blockholders, institutional investors effectively discipline concentrated ownership and significantly influence innovation activities (Shleifer and Vishny 1997; Mahmood and Mitchell 2004; Choi et al. 2011; Belloc 2012). Blockholders can discipline a firm's management when the management's compensation is closely related to its share price (Admati and Pfleiderer 2009; Edmans and Manso 2011). Thus, blockholders’ monitoring role can inhibit managers' short-sighted behavior and encourage firms to invest in innovation. Consequently, more liquidity stimulates innovation.

Financial constraints

A higher stock liquidity leads to lower transaction costs. Higher liquidity is often accompanied by lower issuance costs because underwriters can maintain a net equity position at a lower cost with high liquidity. Higher stock liquidity lowers financing costs and increases the funds available to a company, thus mitigating financing constraints on innovation activities (Vo 2014; Zhong 2018). Based on these arguments, higher liquidity promotes innovation.

Merger and acquisition

Harford (2005) shows that higher liquidity can reduce transaction costs, creating a wave of M&A. Phillips and Zhdanov (2013) showed that large firms are more likely to buy innovation from small firms to obtain more innovation and avoid competing head-to-head with small firms in an active acquisition market. Higher aggregate liquidity enhances firm valuation and reduces transaction costs, facilitating large firms’ acquisition of innovation from small firms, thereby promoting increased innovation investment by small firms (Vo 2014). Zhao (2009) and Bena and Li (2014) document that less-innovative SMEs often acquire more innovative firms which can enhance innovation. Therefore, more liquidity stimulates innovation.

Measurement of variables and model specification

This section briefly introduces the dependent, independent, and control variables used in the analysis. We then specify baseline models to examine the relationships between stock liquidity and firm innovation in listed SMEs. All the variables and their respective data sources are shown in Appendix A.

Dependent variable of firm innovation

Prior studies propose two proxies from the perspectives of input and output to capture firm innovation: (1) R&D investments and (2) patenting activities (Fang et al. 2014). R&D investment represents capital allocated to innovation, whereas patent activities denote innovation output. Both proxies were employed in our models to better capture a firm's innovation activity. On the one hand, the number of granted patents (INNOV_PAT) is adopted as the innovation proxy to measure the output of innovation (Sun and Du 2010; Fang et al. 2014; Vo 2014; Zheng and Zhang 2021). On the other hand, Cohen and Klepper (1996) show that R&D efforts increase firm size, leading to product and process innovation. Therefore, we use the natural logarithm of R&D expenditure (INNOV_EXP) to measure firms’ innovation input (Czarnitzki and Lichi, 2006; Vo 2014; Liu et al. 2021).

Independent variable of stock liquidity

Liquidity, a key determinant of market quality, affects financial instrument pricing, portfolio allocation, and risk management (Amihud and Mendelson 2015). The liquidity-related literature divides liquidity measures into (1) high-frequency data measures (Jarnecic and Snape 2014; Easley et al. 2012) and (2) low-frequency data measures (Lesmond et al. 1999; Amihud 2002; Chung and Zhang 2014; Abdi and Ranaldo 2017). High-frequency liquidity measures were developed from intraday data, whereas low-frequency liquidity proxies were obtained from daily stock returns and volume data (Le and Gregoriou 2020). However, access to high-frequency data is often restricted, especially in emerging markets (Będowska-Sójka 2018). Due to the multifarious character of liquidity, Le and Gregoriou (2020) presented a set of proxies with low-frequency data; therefore, the bid/ask spread is the most popular liquidity estimator (Będowska-Sójka 2018). Some studies aim to determine the best proxy from a variety of low-frequency proxies in different markets (Lesmond 2005; Goyenko et al. 2009; Marshall et al. 2013; Fong et al. 2017; Będowska-Sójka and Echaust 2020). Fong et al. (2017) and Będowska-Sójka and Echaust (2020) show that CPQS introduced by Chung and Zhang (2014) outperforms other low-frequency percent-cost proxies. Therefore, CPQS is employed in this study to measure liquidity. The choice is based on Fong et al. (2017), who show that CPQS is the best daily percentage cost proxy for the Chinese stock market. We exclude an incomplete sample for years in which the firms are not listed for less than one full year.

Thus, CPQS is computed using the daily closing ask and bid prices from Thomson Reuters Datastream and is multiplied by 100 to make it easier to calculate (Chia et al. 2020):

$$CPQS_{i,d} = \frac{{Closing\, ask_{i,d} - Closing \,bid_{i,d} }}{{\left( {Closing \,ask_{i,d} + Closing\, bid_{i,d} } \right)/2}} {\text{x}}100,$$

where $CPQS_{i,d}$ is the Closing Percent Quoted Spread of stock i on day d, $Closing ask_{i,d}$ and $Closing bid_{i,d}$ are closing ask and bid prices of stock i on day d, respectively and the multiplication by 100 is for scaling purposes. Annual CPQS estimates were generated by averaging daily CPQS values over an entire year. Because higher values suggest larger spreads and higher transaction costs for investors, the value of CPQS is inversely associated with liquidity.

Control variables

We follow previous studies on firm innovation to control for a set of variables divided into ownership and firm characteristics. Appendix A provides the definitions of all variables. As shown in Table 1, following Choi et al. (2011) and the characteristics of the Chinese stock market, we employ three variables to control for ownership characteristics: TSHARE, SHRHFD5, and FREE. TSHARE is the number of shares of listed firms that can be traded on the exchange, whereas FREE is the proportion of all tradable shares that exclude blockholders holding more than 5% of the shares. SHRHFD5 is the sum of the squares of the firm's top five largest owners' shareholdings to measure ownership concentration. Firms with higher TSHARE or SHRFHD5 are expected to show higher firm innovation because blockholders, especially those with institutional and insider ownership, are positively related to firm performance by lowering the agency cost of management ownership (Jensen and Mecking, 1976; McConnell and Servaes 1990; Aghion et al. 2013). Turning to FREE, firms with higher FREE are expected to be associated with lower firm innovation owing to non-information trading by noise traders. This is due to the dominance of individual investors, who make up over 90% of the Chinese stock market (Yu et al. 2019). Management focuses on controlling public opinion and reducing R&D innovation efforts.

Table 1 The classification of control variables

Second, 12 variables were collected to control for firm characteristics and isolate the effect of stock liquidity on firm innovation (Choi et al. 2011; Chang et al. 2015; Fang et al. 2014; Brown et al. 2013): RET, TO, VOL, EARN, INTAN, LEV, ROE, Q, BTM, CAPITAL, SALES, AGE. Firms with higher RET, TO, VOL, EARN, INTAN, ROE, Q, BTM, SALES are associated with higher firm innovation (Hall 1999; Chan et al. 2001; Coad and Rao 2016; Luoma-aho and Halonen 2010; Piergiovanni and Santarelli 2013; Hirshleifer et al. 2013; Wang and Wang 2012; Setayesha and Daryae 2017; Mahmutaj and Krasniqi 2020).

On the contrary, LEV is negatively related to innovation input but positively associated with innovation output because the debt burden robs firms of R&D investment and pushes them to produce more innovative products (Iqbal et al. 2020). Hansen (1992) reported that firm age is inversely related to innovative output because young companies face special challenges in terms of innovation and engage in R&D with greater risks (Coad et al. 2016).

Model specifications

Motivated by Wen et al. (2018), we specify linear models (1) and (2) to assess whether stock liquidity enhances or impedes innovation input and output for Chinese growth-oriented SMEs.

$$\begin{gathered} INNOV\_EXP_{i,t} =\, \alpha_{0} + \alpha_{1} CPQS_{it} + \alpha_{2} TSHARE_{it} + \alpha_{3} SHRHFD5_{it} + \alpha_{4} TO_{it} + \alpha_{5} RET_{it}+ \alpha_{6} FREE_{it} + \alpha_{7} VOL_{it} + \alpha_{8} EARN_{it} + \alpha_{9} INTAN_{it} + \alpha_{10} LEV_{it} + \alpha_{11} ROE_{it} + \alpha_{12} Q_{it} + \alpha_{13} BTM_{it} + \alpha_{14} CAPITAL_{it} + \alpha_{15} SALES_{it} + \alpha_{16} lnAGE_{it} + \mathop \sum \limits_{j = 1}^{J - 1} \alpha_{17j} IND_{j} + \mathop \sum \limits_{t = 1}^{T - 1} \alpha_{18t} YR_{t} + \varepsilon_{it}\\ \end{gathered}$$

(1)

$$\begin{gathered} INNOV\_PAT_{i,t} =\, \alpha_{0} + \alpha_{1} CPQS_{it} + \alpha_{2} TSHARE_{it} + \alpha_{3} SHRHFD5_{it} + \alpha_{4} TO_{it} + \alpha_{5} RET_{it} + \alpha_{6} FREE_{it} + \alpha_{7} VOL_{it} + \alpha_{8} EARN_{it} + \alpha_{9} INTAN_{it} + \alpha_{10} LEV_{it} + \alpha_{11} ROE_{it} + \alpha_{12} Q_{it} + \alpha_{13} BTM_{it} + \alpha_{14} CAPITAL_{it} + \alpha_{15} SALES_{it} + \alpha_{16} lnAGE_{it} + \mathop \sum \limits_{j = 1}^{J - 1} \alpha_{17j} IND_{j} + \mathop \sum \limits_{t = 1}^{T - 1} \alpha_{18t} YR_{t} + \varepsilon_{it} , \end{gathered}$$

(2)

where $ln$ denotes the natural logarithm. The dependent variable is firm innovation, comprising innovation input, proxied by the natural logarithm of R&D expenditures, and innovation output, proxied by the number of patents. The key independent variable Closing Percent Quoted Spread (CPQS) is constructed by daily closing ask and bid prices to measure stock liquidity. The control variables are tradeable shares (TSHARE), the sum of the square of the top five shareholders' shareholdings (SHRHFD5), turnover (TO), stock returns (RET), free float (FREE), return volatility (VOL), earnings (EARN), intangibles (INTAN), leverage (LEV), return on equity (ROE), Tobin’s Q (Q), book to market (BTM), capital expenditure (CAPITAL), sales growth (SALES), firm age (AGE). Appendix A presents the definitions of all the variables in the models. We control for industry effects using industry dummies, where ${IND}_{j}$ =1 if firm i is in industry j and 0 otherwise, and J is the number of industries, following the classifications of the National Bureau of Statistics of China. We control for common shocks by including year dummies, where ${YR}_{t}$ =1 for year t and 0 otherwise and t is the number of years. Fixed effects (FE), random effects (RE), and pooled ordinary least squares (OLS) models are typically considered when selecting regression models for panel data. The assumption of the RF model requires that unobservable individual heterogeneity effects cannot be correlated with the explanatory variables. Many corporate finance and accounting studies avoid the RE model because of its challenging assumptions and inability to account for time-invariant omitted variables. Unlike the FE model, the pooled OLS model treats all individuals as homogeneous and ignores firm-specific and temporal effects. To estimate firm-level panel data, we typically use heteroscedasticity-robust standard errors or clustering adjustments across years and firms. Once we applied these controls, the estimation effects of the pooled OLS and FE models became nearly equivalent. The pooled OLS model, with constant coefficients for intercepts and slopes, combines all data for efficient OLS estimation, offering better analytical properties than other econometric models. Based on this, we follow Fang et al. (2014) by using pooled OLS and incorporating year-fixed effects to control for omitted firm characteristics that remain constant over time and intertemporal variation. To consider within-cluster correlations, we estimate liquidity models with standard error adjustments using double clustering (Petersen 2009).

The data

This section discusses how the sample data are constructed using firms listed on the Shenzhen GEM. Subsequently, descriptive statistics and a correlation matrix are presented.

Data collection

The Shenzhen GEM is a platform launched in October 2009 to satisfy the demands of growth-oriented innovative SMEs for funding. As of December 2020, 895 firms were listed on the Shenzhen GEM. After removing firms with insufficient data, delisted firms, and financial firms, our study comprises 785 growth-oriented innovative SMEs from 2010 to 2020. All variables are winsorized at the 1st and 99th percentiles, except for the dummy variables, to reduce the impact of outliers.

Summary statistics

We present summary statistics for all variables in Table 2, with our primary emphasis on stock liquidity and firm innovation. First, the mean CPQS for Chinese-listed growth-oriented SMEs is 0.110, significantly lower than the CPQS averages of 0.421 and 0.407 reported by Fong et al. (2017) for firms listed on the Shenzhen and Shanghai Stock Exchanges, respectively. CPQS, as an inverse measure of stock liquidity, the CPQS indicates that high-technology growth-oriented SMEs listed on the Shenzhen GEM have, on average, more liquid stocks than firms listed on the Chinese main board. The main reasons for this are summarized as follows. First, a significant number of GEM listed firms come from emerging industries with strong growth potential, attracting substantial investor interest. Second, the GEM market's trading rules are more permissive than those of the main board market. For instance, they have fewer restrictions on short stock release times and have no circuit breakers. These lenient trading rules provide investors with greater autonomy and flexibility, thereby bolstering their market liquidity. Furthermore, the CPQS average of 5.337 reported by Chia et al. (2020) for Malaysian listed firms significantly exceeds the CPQS of 0.110 for Chinese SMEs. This discrepancy suggests that the GEM market enjoys substantially higher liquidity than the Bursa Malaysian mainboard market. We conjecture that individual investors who exhibit low confidence and heightened sensitivity to market movements predominantly influence the GEM market. The prevalence of herd-like investment behavior and speculative actions affects stock market liquidity, resulting in market performance distinctions from the more mature mainboard market.

Table 2 Descriptive Statistics

Second, the mean number of patents is 5.969, whereas the mean R&D expenditure amounts to 17.641. Notably, the former is lower than 31.110 and the latter is higher than the 0.042 reported for all Chinese listed firms from 2006 to 2013 by Wen et al. (2018). This distinction suggests that high-risk growth-oriented SMEs grappling with underfunding intensify their investment efforts in pursuit of greater innovation outcomes, but still grapple with innovation shortfalls. Fong et al. (2017) report an average number of patents of 1.208 (e^0.792–1) for U.S. public firms, implying that despite the relatively short history of the GEM market, it demonstrates significant innovation output.

The median number of granted patents at zero suggests that a significant portion of the firms in our sample do not generate any innovative output. Figure 1 illustrates the distribution of zero- and non-zero-granted patents, with 1068 in the non-zero category and 7567 in the zero category.

Fig. 1

Number of patents granted. Notes: This figure presents the number of zero and non-zero-granted patents

Table 3 presents Pearson correlation matrices for the 18 variables included in the baseline models. The correlations between explanatory variables and firm innovation provide a preliminary view of their univariate relationships. First, CPQS is an inverse proxy for liquidity and is negatively associated with both innovation input and output, suggesting that higher liquidity promotes firm innovation. Second, only TSHARE, FREE, EARN, ROE, and BTM yield the expected positive relationships with firm innovation. Third, SALES is positively correlated only with innovation input, whereas SHRHFD5, RET, INTAN, and CAPITAL are only positively associated with innovation output. Fourth, both LEV and lnAGE positively correlate with firm innovation, whereas TO, VOL, and Q show the opposite signs. Next, we explore the explanatory power of CPQS and firm innovation in the pooled OLS estimation when controlling for other innovation-related variables. These univariate relationships may change or become insignificant when all competing factors are included in multivariate regression.

Table 3 Correlation Matrix

Baseline results: stock liquidity and firm innovation

This section discusses the baseline pooled OLS results and the statistical significance of the linear relationship between stock liquidity and firm innovation.

Key variable of firm innovation

In our baseline models, firm innovation is classified into innovation input and output. The former is measured as the natural logarithm of R&D expenditure (INNOV_EXP), whereas the latter is proxied by the number of patents granted (INNOV_PAT). Baseline linear models were estimated using white heteroscedastic-robust, firm-clustered, time-clustered, and double-clustered standard errors to control for heteroscedasticity and within-cluster autocorrelation (Petersen 2009). To save space, only the results of the linear relationships between stock liquidity and firm innovation for white heteroscedastic-robust and double-clustered adjustments are reported in Table 4; however, all four within-cluster estimations are considered in the statistical analysis. As Table 4 shows, the coefficient of CPQS is negative and significantly associated with both INNOV_EXP and INNOV_PAT at the 1% level. Because CPQS is an inverse proxy for liquidity, the negative and significant results suggest a positive linear relationship between stock liquidity and firm innovation. An increase in CPQS is associated with a decrease in INNOV_EXP by 4.062 points and INNOV_PAT by 28.561 points. These results suggest that stock liquidity is more sensitive to a firm’s innovation outputs.

Table 4 The Baseline Results between Stock Liquidity and Firm Innovation

First, the positive linear relationship between stock liquidity and innovation input is comparable to that reported in previous studies. Zhong (2018) attributes improvements in innovation inputs to reduced financing constraints through increased equity liquidity in listed Chinese firms. Wen et al. (2018) find that this positive relationship exists solely in state-owned enterprises (SOEs) and is negative in private firms, primarily because of the higher likelihood of takeovers compared to SOEs in China. Our findings support their results, confirming the existence of a positive relationship between stock liquidity and firm innovation in high-technology SMEs following the split-share structural reform.⁶

Second, considering innovation output as the dependent variable, the positive relationship between stock liquidity and the number of granted patents suggests that increased stock liquidity is associated with higher innovation output for SMEs. Our results contradict those reported by Fang et al. (2014) and Wen et al. (2018). Fang et al. (2014) suggest that increased liquidity reduces future innovation output through two mechanisms: (1) greater vulnerability to hostile takeovers and (2) increasing participation of institutional investors who do not actively obtain information or monitor. Moreover, Wen et al. (2018) show that higher liquidity impedes the innovation output of Chinese listed private firms because of the risk of a hostile takeover. Our main findings are consistent with those of Vo (2014) and Tang et al. (2022). Thus, we posit that growth-oriented SMEs face a reduced risk of hostile takeovers due to their inherently unstable investment risk.

Following Boubaker et al. (2019), we report the economic impacts of all explanatory variables on INNOV_EXP and INNOV_PAT in the last two columns of Table 4 (Economic Impact 1 and Economic Impact 2), computed by multiplying one standard deviation of the variable by its corresponding coefficient estimate from the results of the double-clustered regression. In terms of economic magnitude, a one-standard-deviation increase in CPQS leads to a 0.240 decrease in INNOV_EXP and a 1.083 decrease in INNOV_PAT. Given that the coefficient impacts of CPQS of INNOV_PAT and INNOV_EXP differ by a factor of 4.513, it is evident that CPQS has a greater influence on INNOV_PAT.

Control variables

Only five control variables are significant across both the innovation input and innovation output models within the double-clustered standard errors in Table 4: TSHARE, SHRHFD5, RET, BTM, and CAPITAL. In terms of INNOV_EXP, the coefficients of TSHARE, SHRHFD5, RET, BTM, and CAPITAL are 0.003, -0.627, 0.832, 0.785, 1.224, whereas their coefficients of 0.050, 17.405, 6.769, 11.700, 16.248 in INNOV_PAT. This finding suggests that an increase in TSHARE (RET, BTM, and CAPITAL) is associated with an increase of 0.003 (0.832, 0.785, and 1.224) points in INNOV_EXP and an increase of 0.050 (6.769, 11.700, and 16.248) points in INNOV_PAT. Moreover, the regression results of SHRHFD5 suggest that an increase in SHRHFD5 is associated with a decrease in INNOV_EXP by 0.627 points and an increase in INNOV_PAT by 17.405 points.

First, TSHARE is positively and significantly related to firm innovation, supporting the following empirical evidence: China's GEM has a significant share of institutional investors,⁷ particularly institutional blockholders in the Chinese market, who hold substantial tradeable shares. We posit that their active monitoring and discipline positively influence firm performance, and consequently, firm innovation (Choi et al. 2011; Aghion et al. 2013). Second, SHRHFD5 exhibits a negative and significant relationship with innovation input but a positive association with innovation output. However, the coefficient representing the impact of SHRHFD5 on innovation input is relatively small, indicating a weak influence. Typically, a firm's top five stockholders have substantial internal ownership, often comprising individuals closely related to the firm's management such as its founders, family members, affiliates, managers, and executive directors (Xu and Wang 1999; Chang et al. 2006). Our result is consistent with Fang et al. (2014), who report that internal owners prefer short-term investments to stabilize stock prices and potentially forgo long-term innovative investments to mitigate threats to their positions in the case of a firm acquisition. However, maintaining a firm's image and preventing abnormal share price fluctuations from damaging its profits may ensure that its existing innovative investments achieve good results. Third, RET has the expected positive coefficient, which is consistent with Hirshleifer et al. (2013). Fourth, the positive coefficient of BTM indicates that higher RET leads to improved firm innovation, supporting Hall's (1999) findings that apply to manufacturing firms. This suggests that the larger market value of growth-oriented SMEs can augment their innovation activities. Fifth, CAPITAL demonstrates a positive relationship with firm innovation, consistent with the findings of Piergiovanni and Santarelli (2013) who suggest that capital expenditure involving equipment inputs can enhance the efficiency of innovation efforts.

Turning to the economic impacts of the control variables in Table 4, the largest effect comes from BTM. More precisely, a one standard deviation increase in BTM was related to a increase of 0.139 and 2.071 in INNOV_EXP and INNOV_PAT. Closely followed by RET, a one-standard-deviation increase resulted in an increase of 0.161 in INNOV_EXP and 1.313 in INNOV_PAT.

Further analysis

Blockholders ownership and firm innovation

Fang et al. (2014) and Wen et al. (2018) point out that the regulatory role of blockholders has a disincentive effect on firm innovation. To identify whether this inhibiting effect exists in SMEs, we conduct the following analysis.

In China, since the 2005 equity share reform, all shares listed on the GEM have been tradable. However, some blockholders' shareholdings are required to be tradable under certain conditions. For example, the regulations require the shares held by the beneficial owner to be listed for three years before they can be transferred. All shares held by institutional holders are freely tradable. According to Xu and Wang (1999) and Chang et al. (2006), the top five stockholders of a firm are often those with a large percentage of internal ownership. These individuals are often closely associated with the firm’s management, including its founders and their families, affiliates, managers, and executive directors. Therefore, TSHARE includes external blockholders represented by institutional blockholders, whereas SHRHFD5 includes internal blockholders. Higher liquidity can allow blockholders to hold more shares at a lower cost, attracting larger shareholders to the market (Maug 1998; Edmans 2009). Hence, to analyze the effects of institutional and internal blockholders, we replace TSHARE with the shareholdings of institutions (INSTITUTION) and then exclude INSTITUTION and SHRHFD5 in our baseline models to regress the models again; the results are shown in Table 5.

Table 5 Blockholders Ownership and Firm Innovation

The results without SHRHFD5 are shown in columns 1 and 4; the results without INSTITUTION are shown in columns 2 and 5; and columns 3 and 6 present the results without both SHRHFD5 and INSTITUTION. Compared to the results in Table 4, the coefficient between CPQS and INNOV_EXP or INNOV_PAT increases. CPQS is an inverse proxy for liquidity, which means that the impact of liquidity on firm innovation diminishes when controlling for INSTITUTION or SHRHFD5 in our baseline models. This indicates that blockholders can weaken firm innovation, and that the weakening effect of internal blockholders (SHRHFD5) is stronger.

We can explain the results as follows. Under higher liquidity, blockholders can continue to hold more shares at a lower cost, facilitating the entry of more blockholders (Maug 1998; Edmans 2009). External blockholders, represented by institutional investors, tend to invest in projects with higher expected short-term returns rather than choosing innovative projects in the long term (Bushee 2001). Internal blockholders are also often related to a firm's management, including founders and their families, affiliates, managers, and executive directors (Xu and Wang 1999; Chang et al. 2006). Therefore, managers prefer short-term investments that can stabilize stock prices, and forgo long-term innovative investments to prevent challenges to their positions (Fang et al. 2014). Hence, blockholders have a negative effect on firm innovation as stock liquidity increases. However, this inhibition is not obvious in SMEs.

Long-term effect

We specify the innovation input and output in one-to eight-year lags and one five-year lag for the baseline models, respectively, and re-estimate the models with a double-clustered estimation. The results in Appendices B (a) and (b) show that the coefficient of CPQS is negative and significant for both seven-lag-year innovation input and four-lag-year innovation output. Specifically, the positive linear relationships between stock liquidity and innovation input or between stock liquidity and innovation output are highly significant at the 1% level in a one-to three-lag year, implying that higher stock liquidity promotes long-term firm innovation, especially in the first three years.

Robustness checks

This section presents the results of a series of robustness checks to ensure that the baseline models are reliable.

Alternative innovation measures

We replace INNOV_EXP in the baseline model (1) using the ratio of R&D expenditures to total revenue (EXP_REVENUE) and the ratio of R&D expenditures to total assets (EXP_REVENUE) to measure the input of firm innovation (Zhong 2018; Liu et al. 2021). The number of applied patents is used to replace INNOV_PAT as the output of firm innovation in the baseline model (2) (Zheng and Zhang 2021). Following Wen et al. (2018), we also adopt innovation efficiency (Efficiency) as an alternative innovation proxy by deflating the number of granted patents with the natural logarithm of R&D expenditures. Table 6 shows that the linear relationship between CPQS and these four alternative innovation measures are significantly positive at the 1% level, which reinforces our main findings.

Table 6 Robustness checks with alternative innovation measures

Alternative liquidity measures

The CPQS can capture the costs incurred by investors for immediate transactions. According to Fong et al. (2017), the Closing Percent Quoted Spread Impact (CPQSIM) and Amihud's (2002) illiquidity ratio (ILLIQ) have good performances in measuring liquidity in the Chinese market. CPQSIM is calculated as the ratio of CPQS scaled by trading volume, whereas ILLIQ is computed as the ratio of absolute stock returns scaled by trading volume. Both price impact measures are inverse liquidity indicators, with higher values indicating greater illiquidity. We reestimate the baseline models (1) and (2) and provide the regression results for both CPQSIM and ILLIQ in Table 7. The linear relationships among CPQSIM, ILLIQ, and INNOV_EXP are highly significant at the 1% level, whereas both the price impact measures and INNOV_PAT are significant at the 5% level.

Table 7 Robustness Checks with Alternative Liquidity Measures

Alternative estimation methods

Our baseline models are estimated using pooled OLS with standard errors adjusted for within-cluster correlations. To ensure the robustness of our findings, we re-estimate baseline model (1) using the Fama–MacBeth two-step regression and quantile regression, and baseline model (2) was re-estimated with the Fama-MachBeth two-step regression and Tobit regression. The Fama–MacBeth two-step regression averages the coefficients from yearly cross-sectional regressions across time (Fama and MachBeth 1973), whereas the quantile regression determines whether there is a linear relationship between stock liquidity and firm innovation over the entire range of the firm’s innovation conditional distribution, especially at the extreme upper and lower tails (Koenker and Bassett 1978; Koenker and Hallock 2001). We exclude year dummies in the two-step Fama–MacBeth regression because of its cross-sectional nature.

A Tobit regression is used to estimate the linear relationship between variables when there is a left or right subsumption in the dependent variable. Most patents as innovation output fall into the value of 0; thus, a Tobit regression is more appropriate than a quantile regression for our baseline model (2) (Zheng and Zhang 2021). However, the Tobit model can only fix the concentrated distribution of zero patents, and except for the 0 value, the other patent values belong to count samples (Wen et al. 2018; Tang et al. 2022). Therefore, we also conducted a Poisson regression and negative binomial regression (NBR) with robust standard error adjustments to regress the models with patents as the dependent variable. Because the Poisson regression requires that the expected and variance of the explanatory variables be equal, the results of NBR are more persuasive in Table 9. Tables 8 and 9 show that both R&D expenditure as the innovation input and the number of patents as the innovation output are highly positively significant with stock liquidity across different estimation methods, which is consistent with our main findings drawn from the pooled OLS estimator.

Table 8 R&D Expenditure as Innovation Proxy with Alternative Estimation Methods

Table 9 Patents as Innovation Proxy with Alternative Estimation Methods

Industry-specific regressions

Based on the classifications of the National Bureau of Statistics of China, we classify the listed growth-oriented SMEs into 16 industries: (1) agriculture; (2) construction; (3) culture, sports, and entertainment; (4) education; (5) environmental sports; (4) education; (5) environment protection; (6) finance; (7) IT; (8) leasing and business service; (9) manufacturing; (10) mining; (11) public health; (12) research & development; (13) residence service; (14) transportation; (15) utilities; and (16) wholesales & retail. Among them, only the manufacturing and IT industries have a sample of firms larger than 200, accounting for approximately 67% and 19% of the total sample size, respectively, as shown in Figs. 2(a) and (b). The remaining 14 industries, excluding the manufacturing and IT sectors, account for only 15% of the total sample.

Fig. 2

Notes: The figures present the number of the firms from each of the Manufacturing industry, IT industry, and other 13 industries as a percentage of our total sample, respectively.Footnote

Our samples are divided into 16 industries according to the industry classification of China Securities Regulatory Commission. Excluding the finance industry, we have 15 industries: (1) agriculture; (2) construction; (3) culture, sports and entertainment; (4) education; (5) environmental protection; (6) IT; (7) leasing and business service; (8) manufacturing; (9) mining; (10) public health; (11) research & development; (12) resident service; (13) transportation; (14) utilities; and (15) wholesale & retail.

R&D expenditures are the input of innovation as the dependent variable in Fig. 2(a), whereas the number of patents is the output of the innovation as the dependent variable in Fig. 2(b)

Percentage of firms by industry in the total sample.

We reestimate the baseline models, including those for the financial industry, to check whether the existence of a mutually offsetting relationship between different industries has a net effect. 14 other industries are combined into one category, namely "others,” together with manufacturing and IT as the other two categories for regression estimation. The regression results in Table 10 show that the positive relationship between stock liquidity and R&D expenditure remains intact across industries. However, a highly significant relationship between stock liquidity and the number of patents as the output of firm innovation appears only in the manufacturing industry.

Table 10 Innovation-liquidity relationships by industry

Endogeneity

In examining the endogeneity of stock liquidity and innovation in listed Chinese firms, Wen et al. (2018) and Tang et al. (2022) explore the impact of exogenous shocks on the Splitting Share Reform on this positive relationship. However, Chinese SMEs listed on the GEM board are not affected by this policy shock because it ended in 2006. Thus, the endogeneity in this study is mainly due to unobserved heteroscedasticity-omitted variables and reverse-causality problems. The extant literature on liquidity fails to locate a precise instrumental variable, and policy shocks are not the focus of this study.

For these reasons, as shown in Table 11 and 12, five robustness checks are conducted to address endogeneity; however, we are aware that this issue is not completely avoidable. First, we lagged all independent variables by one year to exclude the impact of the current period and re-estimated the baseline models using pooled OLS (Ali et al. 2016). The results in Columns 1 of both Table 11 and 12 strengthen our main finding that stock liquidity is highly positively correlated with firm innovation. Second, we conduct a one-year change regression for the dependent and independent variables to remove long-term effects (Chung et al. 2010). The results presented in Column 2(a) show that the relationship between CPQS and INNOV_EXP is significantly negative at the 1% level. The coefficients of CPQS and INNOV_PAT are statistically significant at the 10% level, and the explanatory power weakens. Third, Gormley and Matsa (2014) show that a fixed-effects estimator can address time-invariant unobserved heterogeneity. The presence of an unobservable factor correlating with both stock market liquidity and firm innovation may lead to biased coefficient estimates. For example, high-quality managers tend to manage highly liquid firms, which results in improved firm innovation. The significant positive relationship between stock liquidity and firm innovation remains intact, as shown in columns 3 (1) and (2), ruling out the possibility that unobserved firm factors simultaneously determine stock liquidity and firm innovation at the same time. Fourth, the system generalized method of moments (GMM) estimator is used to address the problems of reverse causation due to the difficulties in locating a strictly exogenous external instrument and correcting for unobserved heteroskedasticity problems, omitted variable bias, and measurement error (Wintoki et al. 2012). To capture the dynamic relationship between stock liquidity and firm innovation, we modify the baseline models by including the lagged dependent variable of Innovation as a regressor. The estimation results in columns 4 (a) and (b) provide further evidence of a significant and positive relationship between stock liquidity and firm innovation.

Table 11 Endogeneity Checks on Innovation Input (R&D Expenditures)

Table 12 Endogeneity Checks on Innovation Output (Patents)

Fifth, two-stage least squares are used to control for endogeneity, which deals with unobservable factors that do not have to be constant over time. According to Fang et al. (2009), one lag of CPQS (CPQS_t-1) and the mean CPQS of the two firms in firm i’s industry, which has the closest size (market value of equity) to firm i (MCPQS), are exogenous variables because CPQS_t-1 and MCPQS are correlated with liquidity but uncorrelated with the error term. The lag in CPQS can mitigate concerns that an unobservable in fiscal year t is correlated with both stock liquidity and firm innovation at time t. Compared to its liquidity, MCPQS is less likely to be connected with an unobservable that influences firm i's innovation. The outcomes in Table 13 for 2SLS are consistent with our baseline results.

Table 13 Two-stage Least Squares (2SLS) Regression Result

Potential mechanisms

This section analyzes the potential mechanisms that may account for the positive relationship between stock liquidity and firm innovation in Chinese SMEs.

Stock liquidity, firm size, and financing

SMEs tend to finance R&D using cash flows and external equity (Brown et al. 2009). SMEs face greater issues when raising capital than large firms, and Butler et al. (2005) indicate that increased stock liquidity can reduce flotation costs and investment bank fees, thereby increasing the possibility of raising more external capital. Mancusi and Vezzulli (2020) show that financial constraints have a markedly unfavorable impact on R&D operations. We infer that SMEs may face more serious financing difficulties than large firms. We further explore whether financing constraints can be relieved by improving the stock liquidity of SMEs, which in turn can enhance innovation inputs, and whether this facilitation mechanism is more evident in SMEs than in large firms. First, we collected data on 3,010 large firms from 2010 to 2020⁹ from the CSMAR database as the control group. Following Cotter (1996), net cash flows from financing activities (CFF) are used to measure the ability of external financing, which represents net cash receipts and disbursements resulting from a reduction or increase in issuing bonds and shares and repaying debts or dividends and cash. The following model is used to investigate the effects of financing:

$$\begin{gathered} CFF_{i,t} = \alpha_{0} + \alpha_{1} CPQS_{it} + \alpha_{2} TSHARE_{it} + \alpha_{3} SHRHFD5_{it} + \alpha_{4} TO_{it} + \alpha_{5} RET_{it} + \alpha_{6} FREE_{it} + \alpha_{7} VOL_{it} + \alpha_{8} EARN_{it} + \alpha_{9} INTAN_{it} + \alpha_{10} LEV_{it} + \alpha_{11} ROE_{it} + \alpha_{12} Q_{it} + \alpha_{13} BTM_{it} + \alpha_{14} CAPITAL_{it} + \alpha_{15} SALES_{it} + \alpha_{16} lnAGE_{it} + \mathop \sum \limits_{j = 1}^{J - 1} \alpha_{17j} IND_{j} + \mathop \sum \limits_{t = 1}^{T - 1} \alpha_{18t} YR_{t} + \varepsilon_{it} \end{gathered}$$

(3)

where ${CFF}_{i,t}$ is the net cash flow from the financing activities of firm i in year t. Appendix A presents the definitions of all the variables in the models. We controlled for industry and year effects.

The results for large firms and SMEs are listed in Table 14, respectively. The coefficient of CPQS for SMEs is much smaller than that for large firms, which means that higher liquidity is more sensitive to raising external financing for SMEs because CPQS is an inverse proxy for stock liquidity. Therefore, our conclusion complements and strengthens the findings of Vo (2014) finding that the firm-level effect of stock liquidity on external financing is stronger for innovative SMEs than for large firms. According to Hall et al. (2010), firms tend to finance their innovative projects through cash flow or equity; hence, our evidence suggests that SMEs rely on external financing to invest more in their R&D expenditures.

Table 14 Liquidity, firm size, and financing

Stock liquidity, M&As, and innovation

SMEs with more innovative abilities tend to be acquired by large firms, and large firms optimally choose to purchase innovation from small firms rather than invest more in R&D (Cremers et al. 2009; Phillips and Zhdanov 2013). Zhao (2009) and Bena and Li (2014) document that less innovative SMEs tend to acquire more innovative firms which can enhance large firms’ innovation. Motivated by these findings, we further explore whether acquisitions by large firms affect SMEs’ innovation. Dass et al. (2016) and Massa and Xu (2013) show that more liquid targets are more likely to be acquired, which can increase an acquirer’s stock value. Accordingly, our study examines whether a mediating mechanism from M&A facilitates the innovation output when stock liquidity increases.

In China, M&A for firms encompasses broad concepts, such as mergers, acquisitions, trusteeships, equity transfers, and asset swaps. The concept of M&A for Chinese listed firms is that a listed firm becomes the controlling shareholder of another listed firm by acquiring shares or becoming the actual controller through investment relationships, agreements, and so on.¹⁰ Based on this concept, SMEs can be considered acquired if a transfer of equity occurs. We extracted data from the CSMAR database related to asset restructuring events based on the date of the first announcement: (1) occurrence of equity transfer events (MA) and (2) number of equity transfer events per year (NMA). MA is a dummy variable that takes the value of 1 for M&As and 0 for not occurring for the entire year. Next, we use the procedures recommended by Baron and Kenny (1986) to test the mediation of M&A, as it is the most popular method to test for mediation effects. The three-step approach to testing for mediating effects has yielded good results in many studies (see the example of Alesina et al. 2011). Therefore, we choose a three-step approach to facilitate our understanding and perform manipulation and statistical analysis. Figure 3 and models (4)–(6) show the regression procedures, and the results are presented in Table 15, 16 and 17. From Table 15 and 16 and the last three columns in Table 17, we find that stock liquidity (CPQS, CPQSIM, ILLIQ), M&As (MA and NMA), patents (INNOV_PAT) are both significantly correlated.

$$Y = cX + e_{{{1} }}$$

(4)

$$M = aX + e{2}$$

(5)

$$Y = cX + bM + e{3},$$

(6)

where M&A(M) is measured by the MA and NMA of firm i in year t. Y is proxied by the number of granted patents, and X is proxied by CPQS. The definitions for all variables in the models are presented in Appendix A. Tobit is used for the regression when MA is the dependent variable, and OLS is used for the remaining models.

Fig. 3

Mediating effect test process for M&As

Table 15 M&As (MA, NMA) and Stock Liquidity (CPQS, ILLIQ, CPQSIM)

Table 16 M&As (MA, NMA) and Innovation Output (Patents)

Table 17 Stock Liquidity (CPQS, CPQSIM, ILLIQ) and Innovation Output (Patents) under Controlled and Uncontrolled M&As (MA, NMA)

We include M&As (MA and NMA) in our models to control for their effects, and the results are presented in Table 17. Compared to the outcomes without control variables, we observe little variation in the liquidity coefficient. The coefficients of MA and NMA are not significant except for the significant coefficient at 10% for NMA under the ILLIQ measure of liquidity. We can conclude that under firm stock liquidity, an improvement in liquidity promotes firm innovation, with little mediating effect from M&As.

To further confirm our conclusions, we introduce M&As (MA and NMA) as control variables in our baseline models. The regression results are presented in Table 18. The change in the coefficient of CPQS increases slightly compared with the results without control variables, but the overall magnitude of the change is not evident. We infer that the interaction of M&As with the other control variables has a modest negative impact on innovation. Our findings add to VO (2014) by indicating that the M&A effect does not significantly promote innovation by SMEs when stock liquidity is increased.

Table 18 Stock Liquidity (CPQS, CPQSIM, ILLIQ) and Innovation Output (Patents) under Controlled and Uncontrolled M&As (MA, NMA) in our Baseline Models

Conclusion and discussions

This section presents the study's findings and policy recommendations and discusses the limitations and potential future research.

Conclusion

This study explores the relationship between stock liquidity and firm innovation (innovation input and output) in Chinese publicly traded growth-oriented SMEs. Stock liquidity is proxied by the Closing Percentage Quoted Spread (CPQS), innovation input is captured by R&D expenditures, and innovation output is measured by the number of patents granted. Our results indicate that higher liquidity can promote both the innovation input and innovation output of SMEs in emerging markets at the firm level.

Based on existing studies, we propose two mechanisms without exogenous policy shocks for this positive relationship: (1) reducing financial constraints, and (2) the willingness of large firms to buy innovation. Our subsequent tests on mechanisms further show that among SMEs, the positive relationship between liquidity and innovation is mainly caused by the financing difficulties they face. Increased stock liquidity lowers the cost of acquiring external capital, encouraging SMEs to inject more money into R&D expenditures by issuing more equity and debt. The mediating role of M&A between stock liquidity and firm innovation at the firm level, as put forth by Vo (2014), is not readily apparent. Following Fang et al. (2014), we evaluate the impact of blockholder ownership. The findings indicate that, under the condition of increasing liquidity, both internal and external blockholders, represented by institutions, can impede firm innovation, but this effect is not strong in SMEs.

Our research also examines the long-term influence, and the results show that a positive relationship between stock liquidity and firm innovation has at least a four-year effect. The results of the industry-specific regressions show that higher liquidity can alleviate financing problems for SMEs in all sectors, prompting more investment in innovation. However, innovation output only achieves good results in the manufacturing industry.

Policy recommendations

Our study supplements the existing research on the relationship between stock liquidity and firm innovation in SMEs in emerging markets such as China. SMEs play a vital role in economic growth, and our findings have significant implications for SMEs, especially for managers, such as those in the manufacturing industry. First, our findings provide evidence that SME management can enhance firm innovation in terms of both input and output by improving stock liquidity. Management can improve disclosure for greater market transparency and investor confidence, while the government should tighten regulations to prevent manipulative behavior and boost trading activities. Second, addressing SMEs' financial constraints of SMEs is crucial for boosting investment in innovation, highlighting the importance of prioritizing financing solutions for SMEs. Therefore, SMEs must establish robust internal financial management systems and diversify their financing channels. Finally, the creation of a supportive innovation ecosystem is essential. To stimulate innovation in SMEs, the government should consider providing specific policy subsidies to facilitate talent recruitment and advanced equipment acquisition.

Limitations and future research

This study has two main limitations. First, while Tang et al. (2022) examine the impact of the exogenous shocks of the Share Splitting Reform on stock liquidity and firm innovation in Chinese listed firms, our study's samples from the GEM board remain unaffected, as this regulation ended in 2006. We do not investigate exogenous shocks in this study; instead, we construct a current period model for firm innovation based on Wen et al. (2018). The long-term effects on innovation are not the primary focus of this study, but we can explore them in the presence of other exogenous shocks in future research.

Second, further research should investigate SMEs in both developed and developing countries. As China represents an emerging economy, the findings may not be universally applicable to SMEs in other developing countries because of their varying economic structures. Moreover, although the sample of Vo (2014) comprises developed countries represented by the U.S. market, its concentration on aggregate stock liquidity and the evidence provided at the firm level for SMEs are not sufficient.

Appendix

See Table 

Table 19 Variable definitions and data sources

19,

Table 20 Long-term innovation effect on R&D expenditures

20 and

Table 21 Long-term Innovation Effect on the Patents

21.