The Effects of Green Innovation on Stock Liquidity: Evidence from US Companies

Abstract

This study investigates whether green innovation (GI) enhances stock liquidity by mitigating information asymmetry. Using a hand-collected panel of 4752 unique U.S. publicly listed firms from 2010 to 2024, we employ OLS regressions to show that GI is associated with significantly higher stock liquidity. Our empirical evidence indicates that the relation between GI and market liquidity is highly contingent on firms’ market and accounting environments. Specifically, the liquidity-enhancing effect of GI is stronger for firms facing higher exposure to climate change risk and environmental regulatory uncertainty, where green signaling is most valuable. Furthermore, we find that a strong stakeholder orientation both stimulates green innovation and amplifies its positive impact on stock liquidity. Finally, the liquidity benefits of GI are more pronounced among firms adopting conservative financial reporting practices, suggesting that reporting conservatism mitigates the endogenous risks associated with GI and enhances the credibility of innovation-related disclosures. Overall, our findings establish a robust link between product market sustainability and financial market efficiency, highlighting the role of green innovation in reducing information frictions.

1. Introduction

The global challenge of climate change and the limitation of natural resources have fundamentally shifted the operational landscape for modern corporations, making green innovation (GI), innovations specifically pertaining to green products and services (Chen et al., 2006), a crucial strategy for balancing economic development and environmental management (European Central Bank (ECB), 2020; Siegrist et al., 2020; Takalo et al., 2021). As firms increasingly commit capital to GI, market mechanisms must efficiently price the information contained in GI. A fundamental measure of market efficiency is stock liquidity, which reflects the extent to which investors can freely trade stocks with minimal holding costs and time delays.

Stock liquidity is crucial to the smooth and effective operation of the stock market (Naik & Reddy, 2021). High levels of liquidity signal the development of a high-quality securities market. For individual stocks, higher liquidity yields several benefits, such as improved richness of market information (Fang et al., 2009) and boosted corporate financing capacity (Brogaard et al., 2017), which, in turn, positively impacts corporate financial performance (W. Liu et al., 2021). Given the elevated global economic uncertainty observed in recent years (Zhang et al., 2020), exploring reliable strategies for enterprises to improve their stock liquidity remains critically essential. In contrast, high illiquidity, characterized by a wider bid–ask spread, is primarily associated with adverse selection risk stemming from information asymmetry (IA) between informed parties and the broader market (Amihud & Mendelson, 1986b).

This study investigates whether, and to what extent, GI, as an increasingly important corporate strategic behavior, may mitigate information asymmetry through its disclosure to enhance stock liquidity. Innovative firms, especially those investing heavily in GI, seek high stock liquidity for two critical reasons: financing and valuation. First, GI is a costly and long-term endeavor that favors equity financing over debt (Titman & Wessels, 1988; Brown et al., 2009). The management of innovative firms, therefore, has a strong incentive to increase stock liquidity to effectively raise adequate funds from the equity market.

Second, from a corporate evaluation perspective, intellectual properties (IPs) are intangible assets that are difficult to measure and reliably report under traditional accounting standards (Lev, 2001). Due to this insufficient IP accounting treatment, there is a substantial barrier to the transmission of true values of innovation on financial reports. To overcome this barrier and deliver positive information about their organizational performance, the management of innovative firms must utilize the stock market as an alternative signaling mechanism. Increasing stock liquidity enhances price discovery, as stock prices reflect hidden information and market beliefs, ultimately leading to higher stock prices and increased firm value (Grossman & Stiglitz, 1976; Hellwig, 1980).

Given that information is the main attribute in both GI and stock liquidity, we hypothesize that GI enhances stock liquidity by reducing information asymmetry. Specifically, GI, treated as a public, legally documented, and externally validated disclosure, offers high-quality non-financial information that increases investors’ financial forecast accuracy (Dhaliwal et al., 2012), thereby mitigating information asymmetry between insiders and outsiders. Cui et al. (2016) discovered that CSR activities reduce information asymmetry in the capital market (Cui et al., 2016). Therefore, GI enhances a more transparent information environment inside and outside of firms, as stakeholders can gather and exchange information freely.

However, the direct relation between GI and stock liquidity is still ambiguous due to mixed effects. First, the cost and interpretation barrier may weaken the positive relation. Investing in GI and disclosing the information is costly and often not favored uniformly across all stakeholders. Short-term investors, for instance, may actively discourage CSR activities by charging higher premiums (Meng & Wang, 2020). Furthermore, investors may not perceive green information properly. Short-term investors prioritize disclosures focused on feasibility features (i.e., quick financial returns) over desirability features (i.e., long-horizon societal value) (Puspitasari et al., 2025). Consequently, different interpretations of GI and its disclosure can prevent an effective reduction in information asymmetry.

Second, poor financial reporting quality increases the perceived holding costs for short-term investors, causing them to shift capital away from stocks with uncertain value (Bushee et al., 2019). When firms cannot ensure high reporting quality of GI, the missing information becomes a higher holding cost for investors, increasing information asymmetry and thus reducing stock liquidity. In conclusion, as firms increasingly dedicate capital to GI for contemporary environmental challenges, we are motivated to investigate the subsequent impact of this initiative on stock liquidity. Examining this relation provides valuable insights into sustainability efforts on information transparency, which is essential for investors, as it allows them to more accurately assess the financial quality and forecast the future performance of firms engaged in green innovation. In addition to the main research purpose, we further investigate the contingent effects of climate change exposures and the corresponding regulatory risks and sentiments of stakeholder orientation and of conservative financial reporting practice on the primary relation between GI and stock liquidity.

We constructed a panel of 33,660 observations (4752 unique firms) from 2010 to 2024. Following the existing literature on stock liquidity (Amihud & Mendelson, 1986a; Fang et al., 2014; Wang et al., 2020), we use the relative quoted bid–ask spread as our main dependent variable. Since the quoted spread is indeed a stock illiquidity measure, a narrower (wider) spread indicates a higher (lower) stock liquidity. Consequently, if there is a negative (positive) coefficient of GI, GI increases (decreases) stock liquidity. We proxy GI using the natural logarithm of a firm’s yearly count of green patents retrieved from the USPTO, as green patent data provides a more robust and objective means to assess a firm’s level of GI (Berrone et al., 2013). Our baseline regression result confirms our main hypothesis that GI increases stock liquidity, and it is robust after we used propensity score matching (PSM) to address potential selection bias and endogeneity concerns. We also perform other robustness tests by introducing other well-established stock illiquidity measures, such as Amihud’s illiquidity measure and RToTR illiquidity measure, to the main regression. Consistent with the results in prior studies, our contingent analysis of climate change exposure suggests that climate change risks reduce firms’ stock liquidity, and that GI serves as an effective mechanism to moderate such an effect. We also prove that stakeholder orientation, resulting from heightened public awareness, strengthens the main positive relation. Lastly, we document that conservative financial reporting practice reduces the endogenous uncertainties of GI and, therefore, reduces information asymmetry, resulting in a higher stock liquidity.

The remainder of this paper is structured as follows. Section 2 discusses the literature review on GI and stock liquidity. Section 3 describes the data, variables, and methodology. Section 4 presents the empirical results, including the baseline regression and the three contingent analyses. Section 5 concludes the paper and discusses policy implications.

2. Literature Review

2.1. Green Innovation

Green innovation (GI) is defined as innovations specifically pertaining to green products (Chen et al., 2006). Moreover, the Organization for Economic Cooperation and Development (OECD) defines GI from the environmental perspective as any new or significantly improved product, process, or organizational method that creates environmental benefits compared to alternatives (Urbaniec et al., 2021). To collect green patent statistics, OECD categorizes green patents into several categories, including (i) environmental management, (ii) water efficiency, (iii) climate change mitigation for energy, (iv) climate change mitigation for transport, (v) climate change mitigation for buildings, (vi) climate change mitigation for production, and (vii) capture and storage of CO₂.

Given the escalating public consciousness regarding environmental protection, GI has gained broad acceptance as a crucial strategy, enabling firms to achieve simultaneous economic development and effective environmental management (Yin et al., 2018). The literature generally separates the factors that drive GI into two broad categories: internal and external. Internal drivers originate within the company, often revolving around inherent resources and capacities for innovation. For example, prior studies have demonstrated that enterprise size and firm age influence GI adoption (Hao et al., 2021) and that major corporations deploy GI as a means to strengthen their market competitiveness (F. Li et al., 2021).

Conversely, external drivers arise from public demand and regulatory mandates. Key stakeholders significantly contribute to GI development. Environmentally conscious customers, for instance, eco-friendly customers, exhibit a greater propensity to purchase eco-friendly products (S. Li et al., 2016). Similarly, environmentally responsible suppliers support firms by providing materials and technologies that facilitate GI promotion (H. Lin et al., 2014). Furthermore, government incentives, such as tax breaks, motivate firms to increase their investment in GI, which, in turn, can help corporations secure financial support and alleviate financial constraints (Song et al., 2020; Holzner & Wagner, 2022).

While GI and CSR measures both aim to capture firms’ environmental orientation, CSR differs fundamentally from GI in terms of the nature, credibility, and informational content of the signal conveyed to financial markets.

From a signaling theory perspective, the key distinction lies in the costliness, verifiability, and irreversibility of the signal. CSR indicators are largely based on disclosures, ratings, or self-reported policies, which are often criticized as reputational or symbolic and susceptible to strategic behavior and “greenwashing.” Prior research documents substantial disagreement across ESG rating providers and weak links between ESG labels and real environmental outcomes, creating ambiguity rather than clarity for investors (Berg et al., 2022; S. Kim & Yoon, 2023).

In contrast, green innovation, measured by granted green patents, represents a costly and externally verified investment decision. Patent grants require substantial R&D expenditures, regulatory scrutiny, and technical validation, making them difficult to imitate and costly to reverse. As such, GI constitutes a “hard” signal that conveys credible information about a firm’s technological capabilities, regulatory preparedness, and long-term production strategy.

This distinction is particularly important in a market microstructure context. Liquidity providers are primarily concerned with uncertainty about future cash flows, regulatory exposure, and firm adaptability to environmental transitions. While CSR disclosures may signal ethical intent or reputational positioning, green innovation directly informs market participants about a firm’s capacity to internalize environmental constraints into its core operations. By reducing informational asymmetry along these dimensions, GI improves price discovery and market liquidity.

Finally, while much of the CSR literature emphasizes risk mitigation, such as reducing litigation, regulatory, or reputational risk, green innovation signals future growth potential and operational efficiency through technological advancement. For this reason, GI is not a subset of CSR/ESG activity but a distinct and economically meaningful market signal. Given that our study focuses on how firms’ operational investment decisions affect stock liquidity, green innovation is therefore the most theoretically appropriate construct for our analysis.

2.2. Stock Liquidity

One important characteristic of liquid stocks is that investors can trade them with little time delay and holding costs. According to market microstructure theory, the first and primary type of holding costs is associated with adverse selection due to information asymmetry between informed parties and noisy market participants (O’Hara, 1995). The second type of costs is attributed to inventory and/or transaction costs. Amihud and Mendelson (1986) argue that if the holding costs for stocks are high, they become illiquid and low-value, and firms are expected to incur higher required returns to compensate for the high costs (Amihud & Mendelson, 1986b).

The existing literature also divides the main determinants of stock liquidity into internal and external ones. Internal determinants include cash holdings (Charoenwong et al., 2014; Huang et al., 2018), financial transparency (Lang et al., 2012), and other corporate strategic behaviors like dividend policy (Sterenczak & Kubiak, 2022). On the other hand, external determinants are, for example, prevailing the information environment (Charoenwong et al., 2014), monetary policies (Chordia et al., 2005), and investor behaviors (Dunham & Garcia, 2021; S. Liu, 2015).

2.3. Green Innovation and Stock Liquidity

Innovative firms seek stock liquidity for several reasons. The first and most important reason is the financing motive, given that innovation is a costly project. Based on corporate structure theory, Titman and Wessels (1988) suggest that firms with more innovation and brand value have lower leverage in equilibrium because they want to avoid the greater ripple effects of bankruptcy for their stakeholders (Titman & Wessels, 1988). The management of innovative firms favors equity financing since innovation is a long-term project, as Brown et al. (2009) discovered that innovative firms depend less on leverage but more on equity financing (Brown et al., 2009).

The second reason is the corporation evaluation perspective. The existing literature often associates intangible intellectual properties (IP) with low book and collateral values. For instance, Lev (2001) highlights that the intangible nature of IP (such as patents and R&D) makes it inherently difficult to measure and report under traditional accounting standards. This lack of reliable measurement leads to low book value and increased information asymmetry because it is difficult for investors to seize, verify, and value (Lev, 2001). Due mainly to the insufficient IP accounting treatment, there is a barrier to the transmission of information on the true value of innovation on financial reports and statements. To deliver positive information on organizational performance in the stock market as an alternative signaling mechanism, the management of innovative firms has incentives to increase stock liquidity as stock prices reflect hidden information and market beliefs on firms (Grossman & Stiglitz, 1976; Hellwig, 1980). The higher the stock prices, the higher the valuation of the firms.

As mentioned several times before, information is the main attribute in both GI and stock liquidity; therefore, information asymmetry theory plays a role in explaining the relations between GI and stock liquidity. Even though GI disclosure is not mandatory in the U.S., the management of American green innovative firms is still willing to voluntarily disclose creditable non-financial CSR information (Gelb & Strawser, 2001; Ballou et al., 2018). With CSR disclosures, green innovative firms offer high-quality financial reporting information, increasing investors’ financial forecast accuracy (Dhaliwal et al., 2012) and thereby mitigating information asymmetry between insiders and outsiders. Cui et al. (2016) discovered that CSR activities reduce information asymmetry in the capital market (Cui et al., 2016). GI enhances a more transparent information environment inside and outside of firms, as stakeholders can gather and exchange information freely.

However, it costs to produce green innovation and disclose relevant information, which is not favored by certain groups of stakeholders. For example, Meng and Wang (2020) state that short-term investors discourage CSR activities by charging higher premiums on stock returns (Meng & Wang, 2020). Furthermore, even though the green information goes public through CSR disclosure, investors may not perceive it properly. Puspitasari et al. (2025) document that short-term investors are significantly more inclined to invest when a firm’s climate change disclosures prioritize feasibility features, the means to achieve goals, rather than desirability features, the goals themselves (Puspitasari et al., 2025). In other words, short-term investors are more willing to commit capital to firms that disclose high potential financial returns (i.e., profits realized quickly) rather than firms whose primary claims emphasize true societal value-added actions (i.e., long-horizon environmental or social impact). As a result, even though firms disclose their green innovations, the information asymmetry may not be reduced because different investors will have different interpretations of the information.

Lastly, a simple CSR disclosure is not sufficient, and reporting quality matters for reducing information asymmetry. Bushee et al. (2019) document that short-term investors avoid uncertain projects due to increased holding costs and prioritize clear financial signals (Bushee et al., 2019). Specifically, they find that short-term investors react to poor financial reporting quality by shifting their capital away from value stocks whose returns require a long period for investors to revise their fundamental beliefs. By contrast, these investors favor momentum stocks because their returns are realized sooner. This behavior implies that mis-valuations can become persistent when holding costs are perceived to be high due to low financial quality. To conclude, when firms cannot ensure their reporting qualities in their disclosure, the missing information incurs higher holding costs for investors, increasing information asymmetry and thus reducing stock liquidity.

The mixed effects of GI and its disclosure, which introduce ambiguity regarding the direct relationship with stock liquidity, must be acknowledged. However, due mainly to the management’s strong incentive to raise adequate funds from the equity market and to increase its reputation by improving firm value and social impression, we hypothesize that GI serves as a credible signal to overcome the existing information asymmetry barrier. Specifically, we hypothesize that GI strategically reduces the information asymmetry facing investors, which, in turn, increases stock liquidity. The strategic purpose of this signal is to raise adequate funds and increase the firm’s valuation, as stock prices reflect hidden information and market beliefs regarding firms.

3. Data, Variables, and Methodology

3.1. Data and Sample

We retrieved our samples of publicly traded companies in the period between 2010 and 2024 from Compustat, where firms are classified into different industries and sectors with their unique Standard Industrial Classification (SIC) codes. Following the existing finance and innovation literature, we exclude utility firms and finance firms, as these firms have different financial reporting requirements and standards (He & Ciccone, 2020; Wang et al., 2020). We also exclude firms located outside the mainland U.S., firms with missing or non-positive asset values, firms with non-positive book values of equity, firms with price per share less than US$1, and firms with non-negative sales revenues.

After cleaning the Compustat dataset, we merge our green patent dataset (discussed in Section 3.3) into Compustat and obtain control variables in the final merged dataset. The final dataset consists of 4752 unique firms and 33,660 firm-year observations across the selected sample period.

Table 1. Sampling procedure, industry distribution, and descriptive statistics.

Panel (A): Sampling Procedure
Research Steps				Unique Firm Count		Firm-Year Observation
Initial Sample				10,154		77,833
Drop Firms outside Mainland U.S.				−2096		−14,338
				8058		63,495
Drop Utility and Finance Firms				−2464		−22,862
				5594		40,633
Drop Firms with missing values or having non-positive asset values, non-positive book value of equity, less than $1 price per share, and non-negative sales revenues				−842		−6464
				4791		34,169
Drop Firms with missing variables needed for regression				−39		−509
Final Sample				4752		33,660
Panel (B): Industry Distribution of Unique Firms
Industry	SICs			Unique Firm Count		Percentage
Agriculture	0100 ≤ SIC < 1000			18		0.38%
Mining	1000 ≤ SIC < 1500			294		6.19%
Construction	1500 ≤ SIC < 1800			64		1.35%
Manufacturing	2000 ≤ SIC < 4000			2396		50.42%
Transport and Communication	4000 ≤ SIC < 5000			262		5.51%
Wholesale	5000 ≤ SIC < 5200			154		3.24%
Service	7000 ≤ SIC < 8000			950		19.99%
Other				614		12.92%
Total				4752		100.00%
Panel (C): Descriptive Statistics
Variable	Mean	Std Dev	p25	Median	p75	N	Min	Max
Quoted Spread	0.0397	0.0204	0.0253	0.0346	0.0488	33,660	0.0007	0.1781
LnGreen Patent	0.2291	0.7138	0.0000	0.0000	0.0000	33,660	0.0000	6.8046
Size	20.5515	2.0390	19.1534	20.5725	21.9158	33,660	14.1527	27.6987
Return on assets	−0.0250	0.4627	−0.0301	0.0256	0.0756	33,660	−28.5633	37.4005
Capital Expenditure	0.0428	0.0562	0.0122	0.0256	0.0501	33,660	−0.1518	1.4570
Book Leverage	0.2311	0.1975	0.0428	0.2086	0.3657	33,660	0.0000	0.9645
Research & Development	0.0636	0.2055	0.0000	0.0011	0.0649	33,660	0.0000	23.9378
Tobin’s Q	2.2252	1.9816	1.1830	1.6141	2.4937	33,660	0.1081	56.3527

Table 1 shows the sampling procedure, industry distribution, and descriptive statistics for 4752 firms from 2001 to 2024. The mean, standard deviation (Std Dev), median, and first (p25) and third (p75) quartiles are reported. Quoted Spread is the value of stock liquidity. LnGreen patent is the natural logarithm of the green patent count gathered from the USPTO. Size is the natural logarithm of total assets. Return on assets is the firm’s ROA. Capital expenditure is a firm’s capital expenditure scaled by total assets. Book leverage is the sum of short- and long-term liabilities scaled by total assets. Research & Development measured as total R&D expenditures scaled by total assets. Tobin’s Q is the sum of total assets and excess market capital scaled by total assets.

represents the variables used in our analyses.

3.2. Stock Liquidity

Following Wang et al. (2020) and Fang et al. (2014), we proxy stock liquidity with the relative quoted bid–ask spread (Amihud & Mendelson, 1986a; Fang et al., 2014; Wang et al., 2020). As already discussed by previous liquidity researchers, the spread is the best proxy for liquidity and also serves to be the benchmark for other liquidity measures (Hasbrouck, 2009; Goyenko et al., 2009; Fang et al., 2014).

To construct the relative quoted bid–ask spread, we follow the procedures carefully discussed in Chordia et al. (2001), Wang et al. (2020), and Fang et al. (2014). We first calculate the daily quoted bid–ask spread, Relative Quoted Spreadid, for firm i on day d using the highest price or closing ask price, Askid, and the lowest trading price or the closing bid price, Bidid, sourced from the CRSP daily stock file. To mitigate the effects of extreme quotes, we standardize each bid–ask quote by the midpoint of the prevailing quote, yielding the daily relative quoted bid–ask spread in Equation (1) (Chordia et al., 2001; Wang et al., 2020; Fang et al., 2014):

$$RelativeQuotedSprea{d}_{id}={\displaystyle \frac{As{k}_{id}-Bi{d}_{id}}{(As{k}_{id}+Bi{d}_{id})/2}}$$

(1)

Next, to obtain the final firm-year liquidity measure, we equally weigh each daily spread to calculate the yearly relative quoted spread of the stock over the calendar year. This procedure is similar to the methodology used for variables like Amihud’s illiquidity measure and RToTR illiquidity measure (Amihud, 2002; Florackis et al., 2011). Given that the quoted spread is indeed an illiquidity measure, a wider spread indicates a lower liquidity for the stock, and vice versa.

3.3. Green Innovation

Green innovation (GI) constitutes the central independent variable for this analysis. Given GI’s nature as a qualitative and highly intangible factor, the criteria for measuring GI frequently fluctuate across different research endeavors. Existing studies have often relied on survey data, metrics for green innovative productivity, detailed green-related R&D expenditures, annual reports, and various CSR scores to quantify GI (Anton et al., 2004; Kneller & Manderson, 2012; Mbanyele et al., 2022; Haščič & Migotto, 2015; Casciello et al., 2024). However, these approaches often face challenges, as voluntary questionnaire responses may introduce notable biases, and official R&D or annual reports may not fully capture green initiatives. As Berrone et al. (2013) argue, utilizing green patent data provides a more robust and objective means to assess a firm’s level of GI (Berrone et al., 2013).

We source patent data from the United States Patent and Trademark Office (USPTO). To accurately isolate green patents from the broader set of patent records, we utilize the Cooperative Patent Classification (CPC) Scheme and its official definitions for classification criteria and descriptions.1 This process involves detailed hand-matching of specific green-related patent codes to the appropriate classifications. For instance, the category D01F13/00 specifically addresses patents related to the recovery of starting materials, waste materials, and solvents during manufacturing, aligning with the objective of environmental protection. Crucially, Section Y of the CPC Scheme and Definition categorizes innovations pertinent to climate change protection, which directly matches our research purpose; consequently, we included all patents filed under any Y subclass in our final database of green patents.

Following the hand-matching, we establish the firm affiliation for each green patent. We adopt the linking methodology detailed by Stoffman et al. (2022) to associate the green patents in the USPTO database with their respective filing firms using PERMCO numbers, which are unique permanent identification numbers assigned by CRSP to all companies (Stoffman et al., 2022). Finally, this patent dataset was merged with our CRSP and Compustat datasets for the remaining empirical analyses. Due to the non-normality of green patent counts, we will use the natural logarithm of the green patent count in all regression analyses.

3.4. Control Variables

Following the innovation and finance literature, we control for firm and industry characteristics that may influence firms’ productivity and profitability (e.g., Fang et al., 2014; Pham et al., 2018; Wang et al., 2020). We retrieve all the control variables from Compustat and compute the variables for firm i over the corresponding fiscal year t. The first control variable is the firm size, Size, measured as the natural logarithm of total assets. To control for profitability, we include return on assets, ROA, in our control variable. We control for firms’ investment in fixed assets with capital expenditure measured as total capital expenditure scaled by total assets, as well as investments in innovation, and Research and Development measured as total R&D expenditures scaled by total assets. We control for firms’ book leverage, book leverage, measured as the sum of short- and long-term liabilities scaled by total assets. Finally, we control for firms’ growth opportunities, measured by Tobin’s Q. More detailed information on variables is discussed in Section 3.5.

3.5. Descriptive Statistics and Correlation Matrix

In Table 1, we present the descriptive statistics discussed previously. We winsorize all listed variables at 1% and 99% levels. We proxy stock liquidity with relative quoted spread, and the mean and median for the relative quoted spread are 0.0397 and 0.0346, respectively. Spreads are economically meaningful but not extreme. In other words, on average, the immediate cost for an investor to buy and then immediately sell a stock in our sample is nearly four cents per dollar of trade, which is a common measure of market friction. We measure GI as the natural logarithm of the firm’s green patent count, LnGreen Patent. We acknowledge that the green patent variable is highly zero-inflated, reflecting the early stage and uneven adoption of green innovation across firms. This pattern is consistent with prior studies of green and environmental innovation.

Consistent with the reported results in Fang et al. (2014), the median and the 75th percentile (p75) of our LnGreen Patent variable are both 0.0000. This indicates that at least 75% of the firm-year observations in our sample recorded zero green patents. The prevalence of zero values in both Fang et al. (2014) and our study across all these statistics underscores a common characteristic of innovation data, that green patent activity is highly concentrated. This distribution justifies the use of the natural logarithm transformation and highlights the difficulty in quantifying this intangible variable.

Therefore, to address both zero inflation and right skewness, we follow Fang et al. (2014)’s approach in the innovation literature and employ a natural logarithm transformation. This specification preserves the influence of extreme values. Also, the log-level specification yields an elasticity-type interpretation that the estimated coefficient captures how changes in the intensity of green innovation activity, conditional on firm participation, affect market liquidity measured in basis points. In this sense, the coefficient should be interpreted as an intensive-margin effect rather than a comparison between innovators and non-innovators.

The means of other control variables are as follows: Size, 20.5515; ROA, −0.0250; Capital Expenditure, 0.0428; Book Leverage, 0.2311; R&D, 0.0636; and Tobin’s Q, 2.2252. As discussed before, we include R&D expenditure to isolate the effect of green innovation from general innovation. Similar to the green patent count, the R&D expenditure variable is highly concentrated; the median is only 0.0011, but the mean is significantly higher at 0.0636. This indicates that R&D spending is concentrated among a small number of innovation-intensive firms.

We present Pearson’s correlation matrix for the variables in

Table 2. Pearson correlation matrix.

		1	2	3	4	5	6	7	8
1	Quoted Spread	1.0000
2	LnGreen Patent	−0.1596	1.0000
3	Size	−0.5176	0.3499	1.0000
4	Return on assets	−0.2714	0.0324	0.1907	1.0000
5	Capital Expenditure	0.0079	−0.0326	0.0886	0.0400	1.0000
6	Book Leverage	−0.0250	0.0113	0.3932	0.0288	0.1033	1.0000
7	Research & Development	0.2045	0.0481	−0.2093	−0.5050	−0.1033	−0.1474	1.0000
8	Tobin’s Q	0.0143	0.0566	−0.0926	−0.0737	−0.0594	−0.1422	0.2265	1.0000

Table 2 shows Pearson’s correlation matrix of key variables. Coefficients with less than 5% p-values are bolded.

. We cautiously checked the correlations to ensure that multicollinearity is not a concern. Moreover, we also perform additional variance inflation factor (VIF) tests, in which the test scores from the VIF tests are all below 5, further indicating that multicollinearity is not a concern in our sample.

3.6. Methodology

In this subsection, we develop the following equation to examine the relation between GI and stock liquidity:

Quoted Spread_it+n = β₀ + β₁LnGreen Patent_it + β₂Size_it + β₃ROA_it + β₄Capical Expenditure_it + β₅Book
Leverage_it + β₆R&D_it + β₇Tobin’s Q_it + Year Fixed Effect + Firm Fixed Effect + ϵ

(2)

Quoted Spread_it+n in Equation (2) is the measure of stock liquidity of firm i in year t + n, a narrower (wider) spread indicating a higher (lower) stock liquidity. The main independent variable is LnGreen Patent_it, which is the natural logarithm of the green patent count of firm i in year t. Discussion on the control variables in the model is in Section 3.4.

We cluster standard errors at the firm level. This approach is essential for accounting for potential serial correlation in error terms within a specific firm over the sample period. Given that both stock liquidity measures and corporate innovation activities exhibit high levels of persistence, firm-level clustering ensures that our t-statistics are not artificially inflated by within-firm dependence.

Furthermore, we include year fixed effects to account for intertemporal variation and firm fixed effects to control for omitted firm characteristics that are constant over time. The most important coefficient for identifying whether there is a relation between GI and stock liquidity is β1. A significant negative β1 indicates that one more unit of GI will reduce the level of stock illiquidity, thus increasing the stock liquidity level. Conversely, a significant positive coefficient points out that the more green innovations, the less liquid the stock. We further conduct OLS regressions for other contingent analyses in Section 4, Empirical Results, by adding interaction terms to further examine whether the baseline relation between GI and stock liquidity is strengthened or mitigated under different contingencies.

To ensure the interpretability of our interaction terms and provide a clearer scaling of the estimated effects, we employ a mean-centering approach for our independent variables involved in the interaction terms. Mean-centering can largely address the possible multicollinearity issue arising from the fact that the interaction terms share common sources with the variables involved in the interaction term. Subtracting the sample mean from the examined continuous variable, including LnGreen Patent, Climate Change Exposure, Regulatory Risk, and Accounting Conservatism, prior to generating interaction terms, ensures that the main effect of GI can be interpreted as its impact at the average level of the contingent variables. The interaction coefficients capture how the marginal effect of GI changes as the contingent variables deviate from their sample means. In addition, mean-centering helps reduce non-essential multicollinearity between interaction terms and their constituent variables.

4. Empirical Results

4.1. Baseline Regression

In

Table 3. Baseline regression results.

	Quoted Spread	Quoted Spreadt+1	Quoted Spread	Quoted Spreadt+1
	(1)	(2)	(3)	(4)
LnGreen Patent	−0.0006 ***	−0.0006 ***	−0.0004 **	−0.0004 **
	(0.0002)	(0.0002)	(0.0002)	(0.0002)
Size	−0.0046 ***	−0.0025 ***	−0.0046 ***	−0.0026 ***
	(0.0003)	(0.0003)	(0.0003)	(0.0003)
Return on assets	−0.0016 ***	−0.0023 ***	−0.0014 ***	−0.0023 ***
	(0.0006)	(0.0006)	(0.0005)	(0.0006)
Capital Expenditure	−0.0072 ***	0.0051	−0.0064 ***	−0.0003
	(0.0028)	(0.0036)	(0.0025)	(0.0032)
Book Leverage	0.0201 ***	0.0131 ***	0.0184 ***	0.0119 ***
	(0.0010)	(0.0011)	(0.0010)	(0.0010)
Research & Development	−0.0011	−0.0023 **	−0.0007	−0.0022 **
	(0.0009)	(0.0011)	(0.0010)	(0.0011)
Tobin’s Q	−0.0006 ***	−0.0003 ***	−0.0005 ***	−0.0002
	(0.0001)	(0.0001)	(0.0001)	(0.0000)
Constant	0.1312 ***	0.0885 ***	0.1314 ***	0.0883 ***
	(0.0059)	(0.0064)	(0.0062)	(0.0066)
Year FE	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	No	No
Industry FE	No	No	Yes	Yes
Observation	33,660	28,110	32,980	27,392
Adjusted R-squared	0.3984	0.3291	0.8114	0.7959

Table 3 shows the baseline results of green innovation and stock liquidity. Standard errors are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

, we present the results of the effects of green innovation on stock liquidity in the current year when the green patent is granted (column 1) and the following year (column 2). The economically significant coefficient of GI and stock liquidity, −0.0006, indicates that an increase in green innovation leads to a reduction in the spread, which means an increase in stock liquidity. Economically speaking, when there is a one-unit increase in green innovation, the spread is reduced by six basis points, yielding lower holding costs for investors. As a result of lower holding costs, the stock becomes more liquid (Amihud & Mendelson, 1986b). While the coefficient may appear numerically small, the reduction in spread is tremendous, as in market asset pricing research, basis points are the economically relevant unit, and changes in even one basis point are meaningful (Stoll, 2000).

To benchmark this effect, prior studies document that major liquidity shocks, such as the introduction of algorithmic trading or market automation, typically reduce quoted spreads by only a few basis points. For example, Hendershott et al. (2011) report liquidity improvements on the order of several basis points following the introduction of algorithmic trading activity (Hendershott et al., 2011), the most significant advance in finance. Against this backdrop, a 6bp reduction attributable to green innovation is comparable in magnitude to well-known structural liquidity improvements studied in the literature.

Relative to the mean quoted spread in our sample (0.0397 or 397 bps), this effect represents approximately a 1% improvement in liquidity for a 100% increase in our GI measure. Given that patent grants constitute a purely informational signal rather than a mechanical market design change, an effect of this size is economically non-trivial and indicates a meaningful reassessment of firm-specific risk by market participants.

Moreover, we also document that impact of a firm’s GI on its stock liquidity is not only immediate but also sustained into the subsequent period as the coefficient continues to be significant in the subsequent year. The fact that the magnitude of the effect (−0.0006) does not decay in the subsequent year indicates that the market views green patenting as a durable, non-transitory signal of firm quality or responsible corporate behavior. This signal consistently reduces the perceived information asymmetry and holding costs for investors, thus sustainably enhancing stock liquidity. In short, the coefficients in both years demonstrate that the market efficiently prices green innovation, and the liquidity benefit it provides is immediate and lasts at least one full year. Our findings are consistent with Huynh and Xia (2021)’s finding that GI pays off (Huynh & Xia, 2021).

To affirm that endogeneity is not a concern and that our baseline results are not influenced by the differences in the control variables, we conduct propensity score matching (PSM) for firms in our sample. This approach helps avoid selection bias arising from firm-specific characteristics (Rosenbaum & Rubin, 1983). We start PSM by identifying treatment and control groups in our sample. The treatment groups are the firms with at least one green patent count, and the control groups are the firms with no green patent count. Next, we match treated and controlled firms in the groups with a caliper of 0.05 and a neighbor of 3 based on the control variables used in the baseline regression (Table 3). The sample size is reduced because of the score-matching process. We then rerun the baseline regression using the matched sample in PSM and report the results in

Table 4. Propensity score matching.

Panel (A): Univariate tests of means between treatment and control group
Variable	Treatment Group Mean	Control Group	Diff
Size	21.9590	21.8770	−0.0820
Return on assets	−0.0061	−0.0337	−0.0276
Capital Expenditure	0.0343	0.0323	−0.0020
Book Leverage	0.2253	0.2178	−0.0075
Research & Development	0.1074	0.1155	0.0081
Tobin’s Q	2.6168	2.7438	0.1270
Panel (B): Propensity Score Regression
		Quoted Spreadt	Quoted Spreadt+1
		(1)	(2)
LnGreen Patent		−0.0004 ***	−0.0008 ***
		(0.0002)	(0.0002)
Size		−0.0027 ***	−0.0020 ***
		(0.0004)	(0.0004)
Return on assets		−0.0015	−0.0034 **
		(0.0012)	(0.0014)
Capital Expenditure		−0.0121	0.0041
		(0.0076)	(0.0081)
Book Leverage		0.0141 ***	0.0097 ***
		(0.0013)	(0.0014)
Research & Development		0.0012	−0.0016
		(0.0021)	(0.0023)
Tobin’s Q		−0.0002 *	0.0001
		(0.0001)	(0.0001)
Constant		0.0915 ***	0.0780 ***
		(0.0075)	(0.0088)
Year FE		Yes	Yes
Firm FE		Yes	Yes
Observation		12,723	11,052
Adjusted R-squared		0.4050	0.3642

Panel A of Table 4 examines mean differences in firm characteristics between no green patent firms (control group) and at least one green patent firm (treatment group) over the study period. Panel B shows the baseline regression of treated groups. Standard errors are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

.

Panel A of Table 4 shows the univariate tests of means between the treatment and control groups. The differences in means are trivial among treatment and control groups, indicating that selection bias is not a problem in our sample and that PSM removes the disparities between groups. For example, the difference in the mean of size between groups is 0.08, which is minimal compared to the overall value of size. Furthermore, the differences between the means of capital expenditure, book leverage, and R&D are all less than 0.01, supporting that the differences are trivial. We then regress the ex-post stock liquidity on GI using the matched sample and display the regression results in Panel B of Table 4. The statistically significant negative coefficient of the LnGreen patent variable shows that our baseline results remain unchanged. This outcome confirms that the baseline finding is robust and not driven by selection bias related to observable firm characteristics.

4.2. Baseline Regression: DiD Approach

In the previous sections, we have shown that GI increases stock liquidity and used the PSM approach to confirm that selection bias is not a concern in our sample construction. In this section, we employ the DiD approach to further determine the effects of GI on stock liquidity. This approach compares the ex-post firm performances on stock liquidity to ex-ante performances surrounding given exogenous events. Exogenous regulatory/policy shocks may alter the management’s judgment and decisions on GI, but they do not directly influence stock liquidity; therefore, the DiD approach helps determine the causality by ruling out the possibility of reverse causality that firms with high stock liquidity green innovate. Furthermore, with the inclusion of firm fixed effects, the DiD approach also controls for the unobserved, firm-specific, and time-invariant variable differences.

To ensure our results are robust, we utilize significant exogenous regulatory shocks as quasi-natural experiments in our DiD approach. In Oct 2018, America’s Water Infrastructure Act (AWIA) required community water systems to conduct “Risk and Resilience Assessments” and mandated new standards for water quality and infrastructure, forcing firms to seek green innovation to meet new resilience standards. The regulatory events listed before altered the management’s decision on GI, but did not introduce direct liquidity in the stock market.

In Column 1 of

Table 5. Baseline regression results: DiD approach.

	Quoted Spreadt	Quoted Spreadt	Quoted Spreadt	Quoted Spreadt
	(1)	(2)	(3)	(4)
LnGreen Patent	−0.0003 *	−0.0006 ***	−0.0001	−0.0004
	(0.0002)	(0.0002)	(0.0002)
PostAWIA	0.0000		0.0000
	(0.0004)		(0.0000)
LnGreen Patent × PostAWIA	−0.0006 ***		−0.0005 ***
	(0.0002)		(0.0002)
Placebo AWIA		-0.0001		0.0000
		(0.0003)		(0.0000)
LnGreen Patent × Placebo AWIA		−0.0001		0.0000
		(0.0001)		(0.0001)
Constant	0.1310 ***	0.1312 ***	0.1313 ***	0.1314 ***
	(0.0059)	(0.0059)	(0.0062)	(0.0062)
Control Variables	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	No	No
Industry FE	No	No	Yes	Yes
Observation	33,660	33,660	32,980	32,980
Adjusted R-squared	0.3991	0.3984	0.8114	0.8114

Table 5 shows the baseline results of green innovation and stock liquidity in the DiD approach. Standard errors are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

, we create a post-event dummy, PostAWIA, equaling 1 if observations start in 2019, to represent the effect of AWIA, and 0 otherwise. We then have the dummy variables interact with the LnGreen patent to derive the interaction variable for our analysis. Our finding supports our baseline regression that GI increases stock liquidity following the occurrences of those exogenous regulatory shocks. In Column 2 of Table 5, we add a temporal placebo test. We re-estimate the models by shifting the “Post” dummy to two years prior to the actual events, and create a placebo AWIA dummy for the placebo test. We find no significant interaction effects between LnGreen Patent and Placebo AWIA, supporting the validity of our original timing.

The negative and statistically significant coefficient on the interaction term between green innovation and the PostAWIA indicator suggests that the liquidity-enhancing effect of green innovation becomes stronger following the AWIA event. In Column 1 of Table 5, it is shown that before AWIA, a one unit increase in GI is associated with a 3bps reduction in spread; but, following AWIA, there is an extra 6bps reduction in the spread. This result indicates that green innovation is more strongly valued by market participants once environmental risks become more salient. Flammer and Kacperczyk (2015) also provide robust evidence that enacting policies that require directors to consider stakeholder interests leads to a significant increase in the number of patents and citations (Flammer & Kacperczyk, 2015). This positive effect of exogenous shocks on innovation is particularly pronounced in consumer-focused industries.

Given that GI is a publicly verifiable outcome of a firm’s environmental commitment, it serves as a response to this stakeholder pressure. Therefore, GI is not solely a random investment, but a strategic outcome. This strategic transparency allows GI to reduce information asymmetry concerning climate risk, thereby leading to increased stock liquidity. This mechanism aligns with CSR literature showing that socially responsible actions enhance market quality. For example, X.-Y. Lin et al. (2024) find a positive relation between corporate social responsibility and stock liquidity through the reduction in information asymmetry (X.-Y. Lin et al., 2024).

4.3. The Contingent Analysis of Climate Risk

Recent global environmental disasters underscore the fact that climate change presents a major operational risk for businesses (Ferdous et al., 2024). This risk manifests in various forms, including direct physical damage to firms’ assets and indirect disruptions throughout complex supply chain networks (Huang et al., 2018). Consequently, the existing academic literature has increasingly focused on how market participants interpret this exposure. Scholars have observed that investors specifically recognize climate change exposure as a significant source of business risk, leading to the belief that this exposure negatively influences both a firm’s financial results and its non-financial performance metrics (Krueger et al., 2020).

For example, Huynh and Xia (2021) document that corporate bonds that are more sensitive to negative climate change news subsequently yield lower future returns, which is consistent with the asset pricing theory that reflects investors’ demand for assets that offer a robust hedge against climate risk. Furthermore, when investors are concerned more about climate risks, they are more willing to turn to bonds issued by companies that maintain superior environmental performance (Huynh & Xia, 2021). Similarly, in the equity market, Ardia et al. (2023) reveal a divergence in the stock prices and returns, with the stock prices of green firms tending to appreciate during periods of heightened climate change concerns, while the prices of brown firms experience depreciation during these periods (Ardia et al., 2023).

Antoniuk and Leirvik (2021) point out that stock market investors quickly react to information on climate change and price stocks based on the information they receive (Antoniuk & Leirvik, 2021). Based on signaling theory, the key distinction lies in the costliness, verifiability, and irreversibility of the signal. GI, measured by granted green patents, represents a costly and externally verified investment decision. Patent grants require substantial R&D expenditures, regulatory scrutiny, and technical validation, making them difficult to imitate and costly to reverse. As such, GI constitutes a “hard” signal that conveys credible information about a firm’s technological capabilities, regulatory preparedness, and long-term production strategy. As liquidity providers are primarily concerned with uncertainty about future cash flows, regulatory exposure, and firm adaptability to environmental transitions, GI becomes an effective signaling mechanism to liquidity providers.

Therefore, we hypothesize that GI further increases stock liquidity as a hard signal, reducing information asymmetry among investors in the context of more climate change exposure and risk. To be more specific, as green patents are considered a public, legally documented, and externally validated disclosure of innovative activity, the uncertainty surrounding the firm’s true commitment to environmental strategy and future climate-related risk is rapidly minimized following a green patent grant. With lower information asymmetry, market makers face reduced adverse selection risk. Consequently, they are willing to quote tighter prices, leading to a narrower quoted spread and increased stock liquidity in the context of climate change exposure.

Sautner et al. (2023) create a metric to quantify firm-level exposure to climate change, specifically designed for predicting organizational performance outcomes during the transition to net zero. They utilize a machine learning-based algorithm for keyword discovery to pinpoint climate change exposures, encompassing regulatory impacts and new opportunities. The core of their measure involves tracking the frequency of terms such as ‘uncertainties’ and ‘risks,’ and associated synonyms, within textual contexts that discuss topics related to climate change (Sautner et al., 2023).

Specifically, for the climate change exposure variable that we refer to in the original article, it is the broad measure of how much a firm is affected by climate change. It counts the frequency of climate-related keywords (e.g., “global warming,” “greenhouse gas,” and “carbon”) mentioned during an earnings call, normalized by the total number of words in the transcript. A higher value indicates that climate change is a more salient topic for the firm’s management and analysts. It represents the overall “attention” or “exposure” the firm has to climate issues, regardless of whether that exposure is positive or negative.

Regulatory risk captures the specific exposure that a firm faces regarding government actions and climate policies. This sub-measure focuses on keywords related to “carbon tax,” “emissions trading,” “regulations,” or “cap and trade,” which reflects the firm’s sensitivity to transition risks. Negative sentiment variable captures the extent to which climate-related discussions are framed in negative terms. The algorithm of Sautner et al. (2023) identifies climate-related bigrams and checks if they are associated with negative language (e.g., “loss,” “damage,” “threat,” and “crisis”). A more negative sentiment score indicates that the management perceives climate change primarily as a source of financial or operational distress.

We consolidate firm-level values of climate change exposure, regulatory risks, and negative sentiments provided by Sautner et al. (2023) to our sample and display the regression results on GI and stock liquidity contingent on climate change in

Table 6. The contractual contingency of climate risk exposure.

	Quoted Spreadt	Quoted Spreadt	Quoted Spreadt	Quoted Spreadt	Quoted Spreadt	Quoted Spreadt
	(1)	(2)	(3)	(4)	(5)	(6)
LnGreen Patent	−0.0006 ***	−0.0007 ***	−0.0005 ***	−0.0002 *	−0.0004 **	−0.0003
	(0.0001)	(0.0002)	(0.0002)	(0.0002)	(0.0002)	(0.0002)
Climate Change Exposure	−0.0540			−0.0858
	(0.0918)			(0.0937)
LnGreen Patent × CC Exposure	−0.1296 **			−0.1201 **
	(0.0594)			(0.0611)
Regulatory Risk		−0.2702			−0.4045
		(0.5481)			(0.4726)
LnGreen Patent × Regulatory Risk		−0.9696 ***			−1.0383 ***
		(0.3719)			(0.3760)
High Negative Climate Change Sentiment			0.0003 **			0.0004 ***
			(0.0001)			(0.0001)
LnGreen Patent × Neg. CC Sentiment			−0.0003 **			−0.0003 **
			(0.0001)			(0.0001)
Constant	0.1176 ***	0.1177 ***	0.1177 ***	0.1175 ***	0.1176 ***	0.1177 ***
	(0.0058)	(0.0058)	(0.0058)	(0.0060)	(0.0060)	(0.0060)
Control Variables	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	No	No	No
Industry FE	No	No	No	Yes	Yes	Yes
Observation	28,420	28,420	28,420	27,871	27,871	27,871
Adjusted R-squared	0.4006	0.4035	0.4034	0.8265	0.8264	0.8264

Table 6 shows the baseline results of green innovation and stock liquidity under the contingent context of climate change exposure. Standard errors are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

. In column (1) of Table 6, the climate change exposure variable measures the extent to which firms’ businesses are exposed to climate change risk. When we have our main independent variable, GI, interact with climate change exposure, we spot a significant negative relation (−0.1296) between GI and quoted spread in the context of higher climate change exposures. In column (2), we have GI interact with the firm-specific climate change regulatory risk to examine whether GI benefits stock liquidity by mitigating firms’ regulatory risks due to climate change. We also notice a significant negative relation (−11.8748) between the interaction term and the relative quoted spread. Given that the quoted spread is a measure of stock illiquidity, we conclude that GI helps increase stock liquidity.

Most importantly, we specifically focus on GI and stock liquidity in the presence of firms’ negative news on climate change in column (3). The existing finance literature has documented that as uncertainty about the stock’s true value increases, the market maker must increase the spread to compensate for the higher probability of being exploited by an informed trader, thus increasing the adverse selection risk premium (Glosten & Milgrom, 1985; Kyle, 1985). Since negative news on climate change adds uncertainties to firms’ business risk, we should expect a wider spread as the compensation for the increased information asymmetry between informed and noisy traders. GI, acting as a means to reduce information asymmetry related to climate-related risk, should help reduce the spread, increasing stock liquidity.

The sentiment scores are all negative to represent negative news, and a more negative climate change sentiment score indicates more negative news for the specific firm. In column (3), we create a dummy, High Negative Climate Change Sentiment, that equals 1 for firms with a higher-than-median negative sentiment score, and 0 otherwise. We then have the dummy interact with the GI to test the contingency. The significant coefficient of the interaction term (−0.0003) strongly proves that GI helps increase stock liquidity for firms with more negative news on climate change risks.

Climate variables are measured on very small scales, as the means for climate change exposure and regulatory risk are 0.0009 and 1.64 × 10⁻⁶, respectively. Interaction coefficients capture the liquidity effect of GI that is contingent on the climate variables.

We find that the liquidity-enhancing effect of green innovation is highly contingent on firms’ climate-related risk exposure and related topics. The interaction between green innovation and climate change exposure is negative and statistically significant, indicating that the reduction in bid–ask spreads associated with green innovation is stronger for firms more exposed to climate change. Economically speaking, a one standard deviation (0.0022) increase in the climate change exposure variable is associated with an extra 3 bps reduction in spread.

Similarly, a one standard deviation (0.0003) increase in the regulatory risk variable is associated with an extra 3 bps reduction in spread. The coefficient of −0.0003 in Column (3) of Table 6 indicates that for firms in the high-negative sentiment group, having GI is associated with an additional 3 bps reduction in the bid–ask spread compared to firms with the same patenting level in the low-negative sentiment group. This suggests that green innovation serves as a more potent signal of future viability when management expresses greater concern regarding climate-related risks.

4.4. The Contingent Analysis of Conservative Financial Reporting Practice

Our prior analyses suggest that GI enhances stock liquidity by reducing information asymmetry. However, the initial investment in GI often constitutes a significantly uncertain project that introduces new business risks and potential volatility to a firm’s operations. In other words, although GI reduces current information asymmetry and environmental risks and, therefore, increases stock liquidity, the uncertain nature of GI introduces volatility and information asymmetry for future periods. To mitigate uncertainties and future information asymmetry, we encourage companies to adopt conservative reporting practices.

Conservative reporting practice, also known as accounting conservatism, is an established practice that mandates firms to prioritize the recognition of bad news (adverse losses) over favorable news (gains). The objective of this asymmetric reporting is to provide users with financial statements with a more reliable and realistic evaluation of a firm’s organizational performance, through deliberate understatements of potential gains and overstatements of potential losses (Basu, 1997). Researchers have shown the benefits of adopting conservative reporting practices, including a higher level of accounting disclosure (Iatridis, 2011), a better information environment (LaFond & Watts, 2008), and reduced stock price declines (J. B. Kim & Zhang, 2016).

Moreover, Y. Kim et al. (2013) document that accounting conservatism reduces holding costs and increases seasoned equity offerings (SEO) announcement returns by mitigating the negative impacts of information asymmetry (Y. Kim et al., 2013). Consistent with Y. Kim et al. (2013)’s finding, X. Li (2015)’s study also reveals that firms practicing accounting conservatism have a lower cost of equity and debt (X. Li, 2015). According to Amihud and Mendelson (1986), lower holding costs and less required returns of equity reduce the quoted spread, increasing stock liquidity (Amihud & Mendelson, 1986b). Therefore, we hypothesize that the firm’s choice of accounting conservatism can successfully mitigate this residual uncertainty, thereby strengthening the positive GI and liquidity relation.

To quantify the degree of accounting conservatism, we collect company financial statement data from Compustat and monthly stock return data from CRSP. Consistent with established methodology (Basu, 1997; Khan & Watts, 2009), we exclude observations with missing values for the variables utilized in the accounting conservatism estimation, we remove firm-year observations reporting negative assets, book value of equity, or sales, and we exclude observations where the fiscal year-end price per share is less than $1. To calculate the corporation’s annual return, we cumulate the CRSP monthly returns beginning from the fourth month after the fiscal year end. Utilizing the conditional accounting conservatism measure developed by Khan and Watts (2009), we have the following equation:

Earning_it = β₀ + β₁D_it + (µ₁ + µ₂Size_i + µ₃MTB_i + µ₄Leverage_i) R_it + (λ₁ + λ₂Size_i + λ₃MTB_i + λ₄Leverage_i) D_it × R_it + ϵ_it

(3)

Earning_it is net income before extraordinary items scaled by a lagged market value of equity for firm i in year t. D is a dummy equaling one if firm i’s return is negative in year t, and zero otherwise. R is the annual return for firm i in year t. Size is the natural log of market value of equity, MTB is the ratio of market value of equity to book value at the end of the year, and leverage is the sum of short- and long-term liabilities scaled by market value of equity.

We then regress Equation (3) to obtain cross-sectional estimates λ1 to λ4 for our accounting conservatism measure, C-Score:

C-Score = λ₁ + λ₂Size_i + λ₃MTB_i + λ₄Leverage_i

(4)

C-Score, developed by Khan and Watts (2009), indicates how conservative the management practices’ reporting standards are. A high (lower) C-Score represents that the firm has more conservative (aggressive) accounting numbers on its financial statements to shareholders. We have the C-Score interact with the LnGreen patent and report the regression results in

Table 7. The contractual contingency of conservative reporting practices.

	Quoted Spreadt	Quoted Spreadt
	(1)	(2)
LnGreen Patent	−0.0002	−0.0004 *
	(0.0002)	(0.0002)
Accounting Conservatism	0.0022 ***	0.0017 ***
	(0.0002)	(0.0002)
LnGreen Patent × Accounting Conservatism	−0.0007 ***	−0.0005 ***
	(0.0002)	(0.0001)
Size	−0.0053 ***	−0.0037 ***
	(0.0002)	(0.0003)
Return on assets	−0.0047 ***	−0.0031 ***
	(0.0016)	(0.0012)
Capital Expenditure	−0.0038 *	−0.0078 ***
	(0.0022)	(0.0024)
Book Leverage	0.0155 ***	0.0157 ***
	(0.0009)	(0.0009)
Research & Development	−0.0024	−0.0025
	(0.0027)	(0.0016)
Tobin’s Q	−0.0005 ***	−0.0004 ***
	(0.0001)	(0.0000)
Constant	0.1469 ***	0.1109 ***
	(0.0033)	(0.0066)
Year FE	Yes	Yes
Firm FE	Yes	No
Industry FE	No	Yes
Observation	26,236	25,555
Adjusted R-squared	0.4061	0.8286

Table 7 shows the baseline results of green innovation and stock liquidity under the contingent context of conservative financial reporting practices. Standard errors are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

. The significant negative coefficient of the interaction term, −0.0007, suggests that with conservative reporting practices in effect, firms mitigate the potential negative impacts on information asymmetry and uncertainties due to GI and further increase stock liquidity.

Accounting conservatism is measured on a continuous scale with a mean of 0.2161 and a standard deviation of 0.4224. Interaction coefficients capture the marginal change in the liquidity effect of GI that is contingent on the variation in reporting conservatism. We find that the liquidity-enhancing effect of green innovation becomes stronger as firms adopt more conservative financial reporting practices. In Column (1) of Table 7, the coefficient (−0.0007) of the interaction term indicates that a one standard deviation increase in accounting conservatism is associated with an extra 3bp reduction in spread, suggesting that conservative accounting practices effectively reduce the endogenous operational risks of GI through the reduction in information asymmetry.

4.5. Robustness Test

In this section, we substitute the relative quoted spread with two other well-established stock illiquidity measures, Amihud’s illiquidity measure and RToTR illiquidity measure, to ensure that our baseline regression result is still robust for different measures. The Amihud’s illiquidity measure, RtoVit, is computed as Equation (5):

$${\mathit{RtoV}}_{\mathit{it}}={\displaystyle \frac{1}{D_{it}}}\ast {\sum }_{d=1}^{D_{it}}{\displaystyle \frac{\left|R_{id}\right|}{V_{id}}}$$

(5)

where RtoV_it is the Amihud’s illiquidity measure for stock i in year t, D_it is the number of trading days for stock i in year t, |R_id| is the absolute return of stock i on day d of year t, and V_id is the monetary volume of stock i on day d of year t.

Arguing that Amihud’s illiquidity measure has limitations on the false representation that small-cap stocks are more illiquid and on negligence on the trading frequency dimension of stock liquidity, Florackis et al. (2011) propose a price impact ratio, RtoTRi, measured as the average ratio of daily absolute stock return to its turnover ratio (Florackis et al., 2011). The calculation of the RtoTR ratio is shown in Euqation (6):

$${\mathit{RtoTR}}_{\mathit{it}}={\displaystyle \frac{1}{D_{it}}}\ast {\sum }_{d=1}^{D_{it}}{\displaystyle \frac{\left|R_{id}\right|}{T{R}_{id}}}$$

(6)

where RtoTR_it is Florackis et al. (2011)’s illiquidity measure for stock i in year t, Dit is the number of trading days for stock i in year t, |R_id| is the absolute return of stock i on day d of year t, and TR_id is the turnover ratio of stock i on day d of year t.

We show regression results of the alternative illiquidity measure in

Table 8. Alternative measure of stock liquidity.

	R-to-V Ratiot	R-to-TR Ratiot	R-to-V Ratiot	R-to-TR Ratiot
	(1)	(2)	(3)	(4)
LnGreen Patent	−0.1719 ***	−4.1785 ***	−0.1799 **	−3.9070 **
	(0.0671)	(1.3065)	(0.0885)	(1.9796)
Size	−1.0809 ***	−11.8486 **	−1.0575 ***	−10.1749
	(0.3666)	(5.6452)	(0.4004)	(6.2810)
Return on assets	−0.0564	−1.6167	−0.0251	−1.1821
	(0.1901)	(3.0931)	(0.1902)	(3.0062)
Capital Expenditure	−5.0608 **	−56.0943	−3.4802	−22.9896
	(2.3821)	(43.3015)	(2.4718)	(45.3841)
Book Leverage	1.4634 **	29.0478 **	1.5335 *	30.6734 **
	(0.7331)	(13.1611)	(0.8269)	(14.6135)
Research & Development	−0.4329	−19.0756	−0.4406	−19.6571
	(0.6191)	(19.0184)	(0.6532)	(20.0682)
Tobin’s Q	−0.1835 ***	−2.74197 ***	−0.1535 ***	−2.0885 ***
	(0.0472)	(0.6488)	(0.0495)	(0.7775)
Constant	22.5183 ***	258.8643 **	23.5118 ***	246.9369 *
	(7.3802)	(113.2838)	(8.2952)	(131.5911)
Year FE	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	No	No
Industry FE	No	No	Yes	Yes
Observation	33,660	33,660	32,980	32,980
Adjusted R-squared	0.0221	0.0221	0.3892	0.4077

Table 8 shows the baseline results of green innovation and stock liquidity with two other well-established measures. Standard errors are reported in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

, where column (1) includes RtoV ratio as the main dependent variable, while column (2) includes RtoTR ratio as the main dependent variable. In column (1), where RtoV_it is the dependent variable, the coefficient on LnGreen Patent is a highly statistically significant negative value, −0.1719. Similarly, in column (2), where RtoTR_it is the dependent variable, the coefficient on LnGreen Patent is also highly significant and negative, −4.1785. Since both RtoVit and RtoTR_it are structured such that a negative coefficient signifies increased liquidity, the significant negative coefficients in column (1) and (2) prove that although different measures of stock liquidity are in use, the negative relation between GI and stock liquidity still exists, suggesting that the relation between green innovation and stock liquidity holds regardless of the liquidity proxy used.

5. Conclusions

In this study, we present a positive relation between GI and stock liquidity. We further suggest that climate change risks reduce firms’ stock liquidity, and that GI serves as an effective mechanism to moderate such effect. We also prove that stakeholder orientation, resulting from heightened public awareness, strengthens the main positive relation. Lastly, we document that conservative financial reporting practice reduces the endogenous uncertainties of GI and, therefore, reduces information asymmetry, resulting in a higher stock liquidity. Our findings continue to be consistent after robustness tests and the PSM approach that addresses potential endogeneity concerns and selection bias.

We contribute to the existing GI and stock liquidity literature in several ways. Firstly, we provide institutional investors with a critical and verifiable measure, GI, that helps reduce holding costs and refine stock valuation, particularly for firms operating under high climate risk. We also highlight the need for investors to align their investment horizon with the nature of GI disclosure. For example, short-term investors, who rely on clear reporting and disclosure signals, should focus on firms with high reporting quality to mitigate holding costs and avoid persistent mis-valuations.

Secondly, we connect different streams of the accounting, finance, and innovation literature by pointing out that a simple relation is not sufficient in the real, complex business environment. Specifically, we indicate that regulatory efforts should shift focus from simple CSR disclosure to a more comprehensive report with high financial reporting quality. Since low reporting quality increases holding costs and information asymmetry, standardizing and enforcing rigorous disclosure quality is essential for efficient pricing and market functioning. Moreover, we instruct the management that maximizing the financial benefit of GI requires an integrated strategy: CSR activities, such as GI, must be paired with conservative financial reporting practices to reassure information transparency and then reduce the cost of capital.

Lastly, we validate the decision to invest in GI by highlighting a direct and non-transitory financial payoff in terms of increased stock liquidity. This provides a strong business case for GI funding beyond regulatory compliance or public relations.

We recognize that there are still limitations in our research. Our study focuses exclusively on publicly traded companies within the U.S. mainland. This geographic focus limits the generalizability of our findings, particularly regarding the baseline GI-liquidity effect. The level of GI activity in the U.S. is considered moderate. By contrast, in regions like Europe, GI and related reporting are often more common due to stronger regulatory mandates, while in many developing economies, GI is rare. The market’s response to GI as an information asymmetry-reducing mechanism may vary significantly depending on the region’s overall GI prevalence and the resulting level of information asymmetry already present. The liquidity benefit of GI in a market where GI is rare might be dramatically larger than in a market where GI is common, or vice versa. Future research may conduct a comparative cross-country analysis, partitioning the sample into jurisdictions with different levels of GI activity to test the relation between GI, information asymmetry, and stock liquidity.

Second, although we identify GI as a tool to mitigate information asymmetry, the interaction between GI and other non-financial disclosures (such as carbon accounting or social impact reporting) remains underexplored. Future studies might investigate whether GI acts as a substitute for, or a complement to, broader ESG disclosure frameworks in enhancing market depth.